**Vertically-Aligned Functionalized Silicon Micropillars for 3D Culture of Human Pluripotent Stem Cell-Derived Cortical Progenitors**

**Alessandro Cutarelli 1,**†**, Simone Ghio 2,**†**, Jacopo Zasso 1, Alessandra Speccher 3, Giorgina Scarduelli 4, Michela Roccuzzo 4, Michele Crivellari 2, Nicola Maria Pugno 5,6,7, Simona Casarosa 3, Maurizio Boscardin 2,\*,**‡ **and Luciano Conti 1,\*,**‡


Received: 25 November 2019; Accepted: 23 December 2019; Published: 30 December 2019

**Abstract:** Silicon is a promising material for tissue engineering since it allows to produce micropatterned scaffolding structures resembling biological tissues. Using specific fabrication methods, it is possible to build aligned 3D network-like structures. In the present study, we exploited vertically-aligned silicon micropillar arrays as culture systems for human iPSC-derived cortical progenitors. In particular, our aim was to mimic the radially-oriented cortical radial glia fibres that during embryonic development play key roles in controlling the expansion, radial migration and differentiation of cortical progenitors, which are, in turn, pivotal to the establishment of the correct multilayered cerebral cortex structure. Here we show that silicon vertical micropillar arrays efficiently promote expansion and stemness preservation of human cortical progenitors when compared to standard monolayer growth conditions. Furthermore, the vertically-oriented micropillars allow the radial migration distinctive of cortical progenitors in vivo. These results indicate that vertical silicon micropillar arrays can offer an optimal system for human cortical progenitors' growth and migration. Furthermore, similar structures present an attractive platform for cortical tissue engineering.

**Keywords:** human cortical progenitors; 3D culture; silicon pillars; cell growth; hiPSC-derived neural progenitors; cerebral cortex

#### **1. Introduction**

In biology, developing tissues' microarchitecture is fundamental to allow correct cell differentiation and organization into appropriate structures that relate to specific physiological functions [1,2]. Standard in vitro cell culture models are mainly set in reductionist monolayer settings, conditions that intrinsically lack any structural architecture. This condition often represents a poor proxy to extrapolate cell growth in vivo, thus substantially affecting cell performance and biological assays outcomes [3,4]. Indeed, with respect to whole tissues, monolayer cultured cells are usually more responsive to toxic or therapeutic agents [5,6]. Additionally, cell culture on rigid surfaces can enhance cell proliferation, but might impact cell differentiation due to the partial cell interactions [7].

In this view, more appropriate three-dimensional (3D) cell culture environments can allow more physiological cell-to-cell contact, assisting cell growth and allowing better modelling of developmental processes [8–10]. A 3D environment also facilitates cells to organize into tissue-like structures that better mimic the in vivo function of cells, thus enhancing physiological relevance and predictive accuracy [11–14]. In the last few years, there has been a gradual development and adoption of technologies that enable tissue-like 3D cultures. Both scaffold-free organoid-based technologies and natural or synthetic scaffold-based culture systems have been developed [7]. In particular, since different tissue types show definite assemblies associated with their functional organization, scaffold-based methods allow assisted mimicking of complex tissue geometrical topographies, such as cerebral cortex, thus facilitating effective biofabrication of in vitro 3D tissue-like models [15–17]. With this purpose, silicon-based micro-fabricated culture substrates with well-defined continuous and discontinuous topographies, including the development of surfaces patterned with grooves, nanopillars or nanowires for the study of neural guidance and polarity, have been extensively developed in order to create scaffolds for a variety of applications [18–24].

In the present study, we exploited vertically-aligned silicon micropillar arrays to reproduce the developing cerebral cortex 3D architecture, where effective control over neural columns width, resembling the mammalian neocortex, is required for a large spectrum of applications. In the developing cerebral cortex, cortical progenitors are oriented from the ventricular to the pial surface with an apical-basal polarity. They divide to form radial glia and neuroblasts, the latter can migrate using the apico-basal oriented radial glia fascicles of the subventricular zone (SVZ) as a scaffold, thus forming the different cortical layers [25,26]. Here we focused on mimicking the radially-oriented cortical radial glia fibers, as they are the key players in controlling the expansion, radial migration and differentiation of cortical progenitors, thus allowing for the establishment of the correct multilayered cerebral cortex structure. We show that the scaffold material and structures are compatible with human cortical progenitors' maintenance. Immunofluorescence imaging analysis and RT-PCR results reveal that silicon vertical micropillar arrays efficiently promote the expansion and stemness preservation of human cortical progenitors, when compared to monolayer growth conditions. Furthermore, the precise orientation of the micropillars allows the radial migration, movement that is distinctive of cortical progenitors in vivo. These results indicate that vertically-aligned silicon micropillar arrays can offer an optimal system for human cortical progenitor growth and migration, and a potentially interesting platform for cortical tissue engineering.

#### **2. Materials and Methods**

#### *2.1. Vertically-Aligned Silicon Micropillar Array Fabrication*

Silicon slides containing vertically-aligned micropillar arrays have been realized on silicon surface through a CMOS like process at the Micro-Nano Facility of Fondazione Bruno Kessler (FBK). The silicon slides are composed of a central zone containing a matrix of vertically-aligned micropillars (height: 250–600 μm). Slides with different micropillar diameters have been obtained, with a diameter of 10 or 15 μm that leads to an aspect ratio that varies from 1:25 to 1:17. These significant proportions are particularly hard to achieve for pillar-like structures. The classical MEMS devices usually have aspect ratio of 1:12, for this reason the Bosch process used to realize the surfaces has been tuned in order to be able to reach higher aspect ratio.

The slide surface used for cellular seeding and culturing is 14 mm long and 5 mm wide these sizes were specifically designed to allow each slide to be accommodated in a well of a 12-well cell culture plate. The design foresees a range of distances between the micropillars, from 20 to 50 μm to evaluate the advantages of smaller and larger spaces between the micropillars for cellular deposition and growth. To realize the structures, the silicon wafer was first oxidized, patterned through soft lithography, then the silicon dioxide (SiO2) was patterned through dry etching, and finally a deep reactive ion etching (DRIE) process was used to realize the micropillar-like structures. The DRIE process consists of a two-step process: first an isotropic plasma etch, second the deposition of a passivation layer that protects the lateral part of the structure. This two-step process is called Bosch process and it is repeated several times to realize nearly vertical wall. The time lapse of each of the two steps defines the roughness of the vertical wall as the process feed rate while the maximal depth that can be reached is defined by the power of the radio frequency bias that accelerates the ions toward the surface.

Since the fabrication process does not damage the mask, the same stamp can be employed for many subsequent cycles in a very reproducible manner.

During slides production, the micropillars on the open edge of the structures are systematically damaged during the DRIE process. This effect is connected with the presence of a blank space between slides, this void space increments the effective power of the ions impacting on the surfaces and then destroy or modify the shape of the first micropillars row. These defects have been eliminated in the cutting step. The cutting diamond disk was placed over the border, in order to cut away the defected micropillars and leaving a sharp edge.

#### *2.2. Cell Cultures*

Human cortical progenitors used in this study were differentiated from a commercial control hiPSC line (Gibco, Thermo Fisher Scientific, Monza, Italy) as previously described [27,28]. Briefly, hiPSCs commitment to a neural lineage and subsequently to the dorsal telencephalic lineage was performed by using N2B27 supplemented with human recombinant Noggin (500 ng/mL, Peprotech, London, UK), SB431542 (20 mM, Santa Cruz Biotechnologies, Heidelberg, Germany) and Fibroblast Growth Factor-2 (4 ng/mL, Peprotech). Cells were then detached and seeded on poly-ornithine and laminin-coated plastic dishes (Sigma-Aldrich, Milan, Italy) in medium supplemented with 10 μM Rock inhibitor Y27632 (Sigma-Aldrich). At day 10, neural rosettes containing cortical progenitor cells were manually collected and plated on poly-ornithine/laminin-treated culture dishes in N2B27 medium containing epidermal growth factor (10 ng/mL), fibroblast growth factor-2 (10 ng/mL) and brain-derived neurothrophic factor (20 ng/mL). Confluent cultures were passed as small multicellular clumps at a ratio of 1:3 using trypsin and amplified until passages 8–10. Cryopreserved stocks of human cortical progenitors were prepared from confluent cultures by trypsinization and resuspension in freezing medium (10% DMSO and 90% foetal calf serum).

Neuronal differentiation of hCPs was induced as previously described. Briefly, hCPs were plated at high density (105 cells per cm2) on laminin-treated silicon vertical micropillar arrays. Cells were allowed to grow for four days and then switched in N2B27 medium without growth factors and cultured for 35 days. The medium was changed every four days.

To visualize the cells for fluorescent imaging analyses and time-lapse experiments, hiPSC-derived cortical progenitors were transduced with a lentiviral vector carrying eGFP cDNA under Cytomegalovirus (CMV) promoter and containing a puromycin selection cassette [29].

Procedure for generation and expansion of mouse NS cells (i.e., radial glia-like neural progenitors) were previously described [30–32]. In this study we used the LC1 NS cell line grown in standard conditions in expansion medium composed of Euromed-N medium (Euroclone) supplemented with 1% N2 (Thermo Fisher Scientific) and 20 ng/mL human recombinant epidermal growth factor (20 ng/mL) and fibroblast growth factor-2 (20 ng/mL).

Human cortical progenitors and mouse NS cells were seeded on silicon slides pre-coated with mouse laminin (Thermo Fisher Scientific).

#### *2.3. Scanning Electron Microscopy (SEM)*

Cultures were fixed for 30 min at 4 ◦C in 25% Glutaraldehyde and 0.1 M cacodylic acid in distilled water (pH 7.2). Then, samples were washed three times with 0.1 M cacodylic acid and dehydrated by the exposure to 50%, 70%, 90% and (2×) 100% *v*/*v* ethanol concentration, 10 min each. After being air-dried under an air flow, samples were gold coated by evaporation of a thin gold layer on top of the sample surface (thickness 6 nm, 1.5 nm Cr adhesion layer). Silicon micropillar-based devices deprived of cells did not require any treatment prior to SEM image acquisition. SEM micrographs were acquired by using a TESCAN VEGA III scanning electron microscope (Tescan Analytics, Fuveau, France) (operating voltage 4 kV, working distance 18 mm, stage tilting angle 45◦).

#### *2.4. Cell Growth*/*Viability Assay*

Analysis of cell growth/viability of all cell types employed in this work (hCPs and mouse NS cells) was performed by MTT assay (Sigma-Aldrich). Briefly, MTT powder was dissolved into culture medium at a final concentration of 1.5 mg/mL. For culture incubation with MTT solution, cell medium was removed, cultures rinsed twice with PBS (Thermo Fisher Scientific) and incubated with MTT solution for 1h at 37 ◦C. Following incubation, MTT solution was removed, cells were air-dried and violet MTT precipitates dissolved with isopropanol. The absorbance was read at 570 nm wavelength with a microplate reader (Tecan Infinite M200PRO, Tecan Italia, Milan, Italy).

#### *2.5. Immunocytochemistry*

To process samples for immunofluorescence analyses, cultures were fixed in 4% paraformaldehyde for 30 min at room temperature (RT), permeabilized in PBS containing 0.5% Triton X-100 for 15 min at RT and then blocked in blocking solution (PBS containing 0.3% Triton X-100 and 5% FCS) for 1 h at RT. Samples were next incubated overnight at 4 ◦C with primary antibody diluted in antibody solution (PBS containing 0.2% Triton X-100 and 2% FCS), then washed three times with PBS and incubated for 2 h at RT with secondary antibodies. Samples were then counterstained with 1 μg/mL Hoechst 33,258 (Thermo Fisher Scientific) and further rinsed with PBS before proceeding with visualization. Fluorescent signals and Z-Stack of eGFP+ve human cortical progenitors (12 slices of 7.7 μm each, shown at 7 fps) were detected using a Leica DMi8 microscope equipped with an Andor Zyla 4.2 PLUS, monochromatic, sCMOS sensor, 4.2 megapixel camera. Acquired images were processed with the open-source Fiji software (v2.0.0, open source under the GNU General Public License, Madison, WI, USA) [33].

Antibodies used in this study: primary mouse monoclonal anti-NESTIN antibody (R&D Systems, Minneapolis, MN, USA, 1:300), primary mouse monoclonal β3-TUBULIN antibody (Promega, Milan, Italy, 1:1000), primary rabbit polyclonal anti-SOX2 antibody (Millipore, Milan, Italy, 1:200), primary rabbit polyclonal anti-MAP2 antibody (Santa Cruz, Heidelberg, Germany, 1:200), primary mouse monoclonal anti-TBR2 antibody (ABCAM, Cambridge, UK, 1:500), primary mouse monoclonal anti CUX1 (ABCAM, 1:200), AlexaFluor-488 or -568 conjugated secondary antibodies (Thermo Fisher Scientific, 1:500).

#### *2.6. Time Lapse Analysis*

Time-lapse movies of live GFP-expressing cells migrating along micropillars were acquired with a Zeiss Axio Observer Z1 inverted microscope equipped with the Apotome 2 module for structured illumination and a 2.83 Megapixel AxioCam 503 mono D (all from Zeiss Italia, Castiglione Olona, Italy). Time-lapses were acquired as z-stacks (10 μm z-step) using a plan-apochromatic 10×/0.3 objective, with a frame interval of 30 min for 12.5 h. The movies shown are maximum intensity projections. Optimal focus selection was performed by manual extraction of each focus z-slices from original z-stack time-lapses to select the best focused z position for each time point, then adjusted for brightness and contrast and saved as 7 fps AVI files using Fiji software [33].

#### *2.7. RNA Isolation and Quantitative RT-PCR (qRT-PCR)*

Total RNA was isolated by using TRIzol Reagent (Thermo Fisher Scientific) following the manufacturer's protocol, then retro-transcribed with iScript cDNA Synthesis Kit (BioRad, Segrate, Italy). cDNA was used to verify the expression of specific target genes by qRT-PCR (quantitative RT PCR), using the SsoAdvanced Universal SYBR Green Supermix Kit. Specific primers sets were used (RT-qPCR data were analyzed according to the comparative ΔΔCt method and normalized by using β-Actin housekeeping gene. Sequence of primers used in this study:

Nestin forward 5 -GGAGAAGGACCAAGAACTG-3 , reverse 5 -ACCTCCTCTGTGGCATTC-3 ; β3-tubulin forward 5 -TCAGCGTCTACTACAACGAGGC-3 , reverse 5 -GCCTGAAGAGATGTCCAAAGGC-3 ; β-Actin forward 5 -GACAGGATGCAGAAGGAGATTACTG-3 , reverse 5 -CTCAGGAGGAGCAATGATCTTGAT-3 .

#### *2.8. Statistical Analysis*

Analyses were performed using either a two-sided unpaired Student's *t*-test or a one-way analysis of variance with a Dunnett's post hoc test. Experiments were repeated three times in triplicate and values were considered statistically significant for *p* < 0.05 (\*), *p* < 0.01 (\*\*), *p* < 0.001 (\*\*\*), *p* < 0.0001 (\*\*\*\*).

#### **3. Results and Discussion**

#### *3.1. Generation of Vertically-Aligned Silicon Micropillar Array Structures*

Biomimetic cortical-like 3D platforms have been created based on different approaches, such as cell spheroids, organoids and engineered constructs based on hydrogels [34–36]. These structures present advantages based on cell self-assembly or enabling spatial self-organization. Nonetheless fine control over the 3D micro-architecture, phenotype and reproducibility have been reported to be challenging [37]. To overcome these limitations, we sought to develop a novel scaffold-based approach mimicking the structural organization of the developing cerebral cortex by ideally controlling network topography.

Mammalian cerebral cortex is a complex layered structure organized in columns generated during developmental stages by means of cortical progenitors (CPs) and neuroblasts that move radially along radial glia fibers serving as scaffolds for directed columnar migration [25,26]. Our aim here is to fabricate a 3D culture platform that mimics in vitro these cerebral radial structures, being able to optimally support the outer growth (from the bottom to the top) of CPs such as that occurring in vivo during cortical neurogenesis. With this aim we designed micropatterned structures containing topographically ordered micropillar arrays. Micropatterned substrates with different geometries have been reported in the form of discontinuous micro-grooved configurations and discontinuous geometries like nanopillar arrays made from a several materials comprising polymers, such as PS, PLGA and PDMS together with hard materials, such as silicon and quartz [18,38–41]. In particular, in the neural field pillars and cone geometries at micron scale have been shown to control the outgrowth of neuronal processes, guiding neurite outgrowth alignment and cell growth [20,38,42,43].

Here, we produced a novel 3D neural cell culture platform based on silicon substrates displaying arrays of micropillars fabricated by lithographic patterning (see Methods) of crystalline silicon (Si) wafers. Such technique enables the fabrication of discontinuous micropillars exhibiting at the same time anisotropic geometry. The silicon structures consist of a rectangular seeding surface of 0.7 cm2 containing vertically arranged micropillars (Figure 1A,B).

**Figure 1.** Design, size and morphology of 3D silicon micropillar array slides. (**A**) Schematic structure of a silicon slide containing vertically-aligned micropillars (micropillars are not in scale). (**B**) The image shows a 3D silicon slide. Scale bar: 1 mm.

Silicon-based nanostructures exhibit conceivable applications in several fields. As examples, plasmonic nanostructures based on array of silicon nanopillars could be used for surface enhanced Raman spectroscopy (SERS) or to control the wettability of a silicon surface [44–47]. Silicon micropillar arrays can be assembled by lithographic techniques allowing tight control over the size and density of the micropillars, differently from randomly generated rough surfaces as those presented in several other works [17,43,48,49]. Accordingly, they allow greatest control over the topographic structure of the system, thus warranting high reproducibility and robustness of the experiments. Additionally, differently from other materials, silicon surfaces hold particular technological impacts and potential.

The size of the slide was specifically designated to easily accommodate in a well of a tissue culture 12-multiwell plate (Figure S1A,B). By varying the reaction conditions of growth, we can produce pillars that have 10–15 μm diameter and height that can be regulated in the range of 200–600 μm. For this study we used silicon slides with 250 μm tall micropillars. Different micropillars topographies were successfully fabricated by using different masks. We produced 12 distinctive layouts of silicon micropillar arrays, coded as A1 to S6, with variable micropillar density and topographies (square grid-aligned and hexa grid-staggered micropillar arrays and with variable distances between the micropillars; see Table 1).


**Table 1.** List of different silicon slide arrays fabricated and tested in this work.

Additionally, to set up the best conditions for cell growth, we fabricated both oxidized silicon (SiO2) and silicon nitride (Si3N4) vertically-aligned silicon micropillar arrays.

The fabrication accuracy and morphology of the structures at the nanoscale were evaluated by measuring the distance between the micropillars and their diameter by scanning electron microscopy (SEM). Cross-sectional SEM images were produced to derive the internal structure of the samples (Figure 2A,B) and top-view SEM images to derive the density of the silicon micropillar arrays.

**Figure 2.** Silicon micropillars can be arranged with different topographies. Silicon micropillar arrays are perfectly vertically-aligned with a height that can be tuned in a 200–600 μm range. Specific spacing, density and morphology of the silicon pillars can be arranged by changing the mask of lithographic process. Low and high magnification of cross-sectional SEM images of aligned (**A**) and staggered (**B**) micropillar arrays. Scale bar: 50 μm.

We found that while the distance between micropillars was quite accurate, the diameter of the micropillars varied depending on the height of the structure. Indeed, the diameters at the bottom of the micropillars were reduced of about 40% (~5–7 μm) with respect to the top of the same micropillar.

This effect is particularly noticeable on the first row of micropillars, and we assumed that this is mainly due to the fact that the chemical-physical processes on which the Bosch process is based depending on the area exposed to the engraving process. Therefore, the etching rate is different in the areas between the micropillars than in the areas between the slides. Therefore, we assume that the first row of micropillars is etched much faster than the others and that this is the cause of the reduction of the diameter of the micropillar. Up to now, we were able to reach structures with a diameter of 15 μm and a height of 600 μm, nevertheless further tests to improve this aspect ratio are ongoing.

#### *3.2. Culturing Human Cortical Progenitors on Vertically-Aligned Silicon Micropillar Arrays*

To establish which human cortical progenitors (hCPs) density was the most comparable to the standard 2D monolayer method used as control and to detect potential toxic effects of silicon material, we first performed a cell viability assay. hCPs were seeded on laminin-coated samples A6–S6 with cell density ranging from 2 <sup>×</sup> 104 to 8 <sup>×</sup> 104 cells per 3D device (i.e., 2.8 <sup>×</sup> 104 and 1.1 <sup>×</sup> 10<sup>5</sup> cells per cm2) and cultured for 48 h before being processed in an MTT assay (see Section 2). We found that cell viability was reduced for the lowest cell density (2 <sup>×</sup> 10<sup>4</sup> cells) and increased for the highest (8 <sup>×</sup> 104 cells), compared to the 2D culture. Indeed, 4 <sup>×</sup> <sup>10</sup><sup>4</sup> cells per 3D device showed no differences compared to the 2D monolayer culture (Figure 3).

**Figure 3.** hCPs seeded on silicon micropillar arrays maintain their viability. MTT assay performed on hCPs plated on 3D silicon slide and in standard 2D monolayer shows that hCPs efficiently maintain their viability. Different cell densities were assessed 48 h after seeding. *p* < 0.01 (\*\*), not significant (ns).

We found no significant difference between oxidized silicon or silicon nitride material (not shown), thus indicating that these 3D structures are fully compatible with living cells without affecting cell growth. Moreover, our silicon micropillar arrays might also allow for increased hCPs proliferation activity in long-term cultures by offering a 3D environment in which the cells can also exploit the third (vertical) dimension for growth. Additionally, we did not detect any significant difference between different topographic layouts and silicon types (oxidized silicon or silicon nitride) tested in terms of cell adhesion, viability and growth (not shown).

As further confirmation of these results, an analogous MTT analysis was performed on mouse radial glia-like NS cells plated at the same aforesaid cell densities (Figure S2A). NS cell analysis gave similar results to the hCPs, indicating that 4 <sup>×</sup> 104 cells per 3D device represents the preferred cell density. This was then selected as standard seeding density in the subsequent experiments in this work. The compatibility of oxidized silicon or silicon nitride for in vitro functional studies on neurons has been already reported, nevertheless, this is the first study to report compatibility with hCPs [50]. Furthermore, contrary to other studies employing silicon uncoated surfaces, we used surface topographies coated with laminin to optimize cell adhesion as also been reported by others [51–53]. In addition, these results demonstrate that our silicon micropillar arrays can be functionalized according to the cells' needs. We then performed SEM analysis in order to monitor the behaviour of the hCPs seeded on the arrays in terms of interaction with the vertical silicon micropillars. We found that cells at high density interact with each other and with the micropillars establishing a uniform network (Figure 4A).

Interestingly, hCPs plated at low density show an increased propensity to adhere to the micropillars rather than to the bottom flat surface of the device, with cells that can be found at different heights of the micropillars (Figure 4B,C). Cells formed 3D neural networks and suspended bridges throughout the micropillar height and along and between the micropillar walls. A similar formation of suspended neural process bridges has also been reported with microtowers, microfibres and 2PP-DLW fabricated microstructures [54–57]. In order to visualize the cells seeded on the silicon micropillar arrays with an inverted microscope, the device has to be set upside down, facing the microscope objectives. To facilitate this operation, we fabricated, by 3D-printing technique, a slide holder support to be placed on a glass bottom dish that allows to visualize the slide with a tilt of 45 degrees, so as to have a 3D visual prospect of the cells (Figure S3). Also, a glass bottom dish was used to place the holder (Figure S2B). To visualize live cultures, we generated eGFP+ve hCPs by infection with lentiviral particles allowing constitutive expression of eGFP cDNA (see Methods). eGFP+ve hCPs were used to live monitor the interactions of hCPs with the device and the single silicon micropillars by time-lapse analysis, and to

capture the dynamics of cells movements among the micropillars. We found that hCPs move within the chip and interact with each other and with micropillars. In particular, hCPs can move along the whole micropillar from the bottom to the top (Movie S1) in a manner that resembles the in vivo migration along the radial glia during cortical neurogenesis, and extend processes embracing the micropillars (Movie S2). We also noticed the progressive neurite extension surrounding micropillars at the same height, which might represent the way for the cells to preferentially form a layered architecture.

**Figure 4.** hCPs seeded on silicon slide establish close interactions with micropillars. (**A**) Top-view SEM image showing hCPs seeded at standard density on silicon micropillars. (**B**,**C**) Cross-sectional SEM images showing hCPs seeded at low density on silicon structures to visualize single cells-micropillars interactions. Scale bar: 20 μm (**A**,**B**) and 10 μm (**C**).

#### *3.3. hCPs Seeded on Silicon Micropillar Arrays Proliferate and form 3D Layered Structures*

We next assessed the proliferation of hCPs seeded on different silicon 3D devices in order to find the silicon type better supporting hCPs stable long-term growth. To this aim, hCPs were plated and maintained for several days on silicon oxide (SiO2) and on silicon nitride (Si3N4) micropillar arrays, as well as on plastic dishes as standard monolayer growth conditions. Cultures were analysed by MTT-based growth assays at different time points. Resulting growth curves from the three experimental groups show a modest, yet statistically significant, reduction in cell growth within the first four days for both types of 3D silicon devices. After seven days cultures grown on Si3N4 devices still exhibit a significant reduction, whereas cultures on SiO2 device show no difference with respect to the standard monolayer conditions. At the later time point considered, 14 days, a statistically significant difference was detected for the cells seeded on SiO2 3D silicon device, which resulted in increased growth when compared to the other experimental groups (Figure 5).

**Figure 5.** hCPs seeded on silicon slides maintain their proliferation capability. Cell growth analysis (MTT assays) performed on hCPs seeded on oxidized silicon (OxSi) or nitride silicon (NiSi) micropillar arrays. Standard 2D monolayer cultures were used as control. Cultures were assessed at different days in vitro (DIV) after seeding. *p* < 0.01 (\*\*), *p* < 0.0001 (\*\*\*\*), not significant (ns).

These results indicate that growth on silicon 3D silicon devices improves hCPs long-term maintenance. Additionally, since oxide silicon 3D arrays resulted to be more compatible and efficient in sustaining hCPs long-term survival and growth with respect with the nitride silicon devices, for the next experiments we employed the former type. Immunofluorescence imaging analysis of hCPs grown on silicon 3D micropillar arrays for two weeks show that cells are well distributed inside the device, filling the whole space among the micropillars (Figure 6A).

Additionally, SEM imaging shows that the cells form a complex three-dimensional multilayered-like structure by growing among the micropillars and interacting with each other, finally establishing regular horizontal layers (Figure 6B,C).

**Figure 6.** hCPs cultured on silicon micropillar arrays establish layered structures. (**A**) Picture of eGFP+ve hCPs cultured for 14 days on silicon slides showing the establishment of a high-density 3D culture. Inset shows the same culture stained with Hoechst. (**B**,**C**) SEM images of hCPs maintained on silicon pillars device for 14 days. Cultures exhibit the generation of multiple cell layers. Scale bar: 50 μm (**A**) and 20 μm (**B**,**C**).

#### *3.4. hCPs Grown on Silicon Micropillar Arrays Retain Their Multipotency and Regional Identity*

We then assessed if growth on silicon 3D arrays might interfere with hCPs (i) multipotency and (ii) preservation of their cortical regional identity. To assess these issues, we first performed an immunostaining analysis for SOX2 and NESTIN, two key neural multipotent markers, on hCP cultures expanded for 14 days on silicon 3D arrays or in standard monolayer conditions. To this respect, we found that silicon 3D cultures show the great majority of cells to co-express SOX2 and NESTIN, comparably to the cultures maintained in standard monolayer conditions previously characterized to efficiently preserve hCPs multipotency (Figure 7A,B and Figure S4).

**Figure 7.** hCPs seeded on silicon slide maintain their neural immature identity. Immunostaining of hCPs for NESTIN (green), SOX2 (red) and nuclear staining with Hoechst (blue) in 3D culture (**A**) and 2D culture (**B**) cultured for 14 days. (**C**) Quantitative RT-PCR assay showing the expression levels of Nestin and β3-Tubulin transcripts in hCPs grown in 2D or 3D cultures. Scale bar: 100 μm (**A**) and 50 μm (**B**). *p* < 0.01 (\*\*), not significant (ns).

To extend this result, we measured the transcript levels of Nestin and of the neuronal marker β3-Tubulin by quantitative RT-PCR (Figure 7C). This assay showed a 30% increase and a 60% decrease on the expression levels of Nestin and β3-Tubulin, respectively, in cultures maintained on silicon 3D arrays with respect to standard monolayer conditions (Figure 7C). These results further confirmed that growth on our silicon 3D devices does not alter hCP's multipotency. Furthermore, growth on silicon 3D arrays further lowers the occurrence of spontaneous neuronal differentiation in the cultures.

We then investigated the capability of hCPs cultured in silicon 3D arrays to retain their original cortical identity. To assess this issue, we first performed an immunostaining analysis for TBR2, a key maker of cortical progenitors, on hCP cultures expanded for 14 days on silicon 3D arrays or in standard monolayer conditions (Figure S5). We found that TBR2 immunoreactivity was maintained in nearly all of the cells in the silicon 3D cultures (Figure 8A; Movie S3). Finally, we tested the ability of hCPs plated in silicon 3D arrays to undergo neuronal maturation giving rise cortical glutamatergic neurons. To this end, hCPs were cultured for 5 days on 3D silicon devices and then exposed for 35 days to differentiative conditions and processed for immunofluorescent analysis for β3-TUBULIN, the mature neuronal marker MAP2 and the cortical neuronal marker CUX1. We found that differentiated cultures exhibited the presence of β3-TUBULIN+ve neuronal cells (Figure 8B, left panel) and the appearance of neurons co-expressing MAP2 and CUX1 (Figure 8B).

**Figure 8.** hCPs seeded on silicon micropillar arrays preserve their cortical regional identity and upon exposure to differentiative conditions, generate cortical glutamatergic neurons. (**A**) eGFP+ve hCPs cultured for 14 days on 3D silicon devices preserve the expression of the cortical progenitor marker TBR2 (red). Nuclei are stained with Hoechst (blue). Scale bar: 50 μm. (**B**) hCPs cultured for 5 days on 3D silicon devices and then exposed for 35 days to differentiative conditions maturate into cortical glutamatergic neurons. Left: cultures stained for the pan-neuronal neuronal marker β3-TUBULIN (red). Nuclei are stained with Hoechst (blue). Scale bar: 10 μm. Right: cultures stained for the mature neuronal marker MAP2 (red) and for the cortical neuronal marker CUX1 (green). Nuclei are stained with Hoechst (blue). Scale bar: 10 μm.

These results demonstrate that hCPs grown on silicon micropillar arrays keep their multipotency and cortical identity and are able to generate glutamatergic cortical neurons. However, further experiments are required to further investigate whether the cortical neurons obtained in the 3D environment are able to organize themselves into defined upper and deeper cortical neuronal layers.

#### **4. Conclusions**

We have described a novel culture platform for hCPs growth based on 3D vertically-aligned silicon micropillar arrays. This structure mimics the radially-oriented cortical radial glia fibres that during embryonic development are essential to control the expansion, radial migration and differentiation of hCPs.

The silicon micropillar arrays can be arranged using different topographic organizations and report micropillar heights not tested with neural progenitors so far. The structures conceived practically combine the advantages of microscale topographies, without the need of complex fabrication techniques. Importantly, our fabrication process allows the production of biocompatible, 3D devices in a highly versatile and reproducible manner. In fact, the micrometre-sized base confers to micropillars good mechanical stability, while establishing a tight interface with the living hCPs.

Other scaffold-based methods have been reported for the biofabrication of in vitro 3D tissue-like models [15–17]. In particular, some of these systems, especially the ones based on 3D polymeric materials, i.e., microfibres and hydrogel scaffolds, offer a highly useful and strong method for creating large-scale 3D tissue cultures. Additionally, these systems have been shown to be extremely flexible in terms of production, biocompatibility, biodegradability, mechanical properties and functionalization with chemicals or oligopeptides in order to implement their adhesion and/or ECM mimicking properties [16]. The 3D vertically-aligned silicon micropillar arrays here described exhibited a lower flexibility with respect to these microfibers and hydrogel scaffolds, nevertheless they represent highly reproducible scaffolds in which both the height and the topography of the pillars can be finely regulated. Similar pillar-like structures have been previously reported by Limongi and colleagues and characterized for their ability to allow culturing of functional neuronal and glial cells in a 3D manner, allowing the formation of viable and functional neuronal networks [17]. In particular, the pillars there reported were different from ours since they were designed to reach limited height (cylindrical pillars of 10 μm in height and 10 μm in diameter) and to include a patterning in the nanoscale on their sidewall, leading to a spatial modulation in the z direction. Similarly, culturing platform containing arrays of microchannels arranged into ordered 2D arrays (with maximum height of ~3 μm and amplitudes ranging from ~10 to ~1.5 μm) have been reported [15]. These structures were shown to host neuronal cells strongly guiding their axons' growth direction and with additional advantage to enable coupling to devices for active sensing and stimulation at the local scale.

Our future efforts should reveal the actual potential of our 3D vertically-aligned silicon micropillar arrays to extensively support hCPs neuronal maturation in order to generate cortical-like tissue comparable to the self-assembled brain organoid technology but with the advantage of an increased reproducibility intrinsic to the scaffold-assisted process.

Additionally, to further exploit the potential of our 3D vertically-aligned silicon micropillar arrays, in the future we aim at coupling this system with a compartmentalized microfluidic device to reach a complete control of culture environment and reduce media volumes and related costs. The acquired knowledge will certainly pave the path towards the generation of valuable tools to study cortical development in humans and for cortical tissue engineering.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/9/1/88/s1, Figure S1: Silicon devices were specifically designed to fit into 12 multiwell plates; Figure S2: Analysis of mouse NS cells viability seeded on 3D silicon micropillars arrays; Figure S3: Silicon slide holder device's picture. The slide lodges on a home-made holder inserted in a bottom glass dish; Figure S4: hCPs seeded on 3D silicon slide preserve their multipotency and create a cell network; Figure S5: hCPs grown in standard 2D conditions exhibit expression of cortical progenitor markers; Video S1: 12.5 hours time-lapse movie performed on eGFP+ve hCPs seeded on silicon micropillars array. A single cell migrating from the bottom to the top of the micropillar is visible; Video S2: 12.5 hours time-lapse movie of eGFP+ve hCPs seeded on 3D silicon slide. Examples of cells extending neurites that interact with pillars are shown; Video S3: Z-stack of eGFP+ve hCPs stained for TBR2 (red) and Hoechst (blue). Cells have been cultured for 14 days on silicon micropillars arrays before the time lapse analysis.

**Author Contributions:** S.G., M.B. and L.C. designed the pillar architecture. S.G. and M.B. developed the fabrication technology, realized the PDMS molds for pillar imprinting, and with help from M.C. produced the silicon micropillar array slides. A.C. prepared hCPs cultures and performed viability assay, growth curve analyses, morphological and immunohistochemical analyses on hCPs. A.C., J.Z. and A.S. prepared samples for the SEM analyses. J.Z. and A.S. prepared mNS cultures and performed viability assay on mouse NS cells. A.C. and J.Z. performed the time lapse experiments with the help of G.S. and M.R. G.S. designed and produced the slide holder devices. L.C., A.C., S.C., S.G., and M.B. planned the experiments, analysed the data and wrote the main manuscript, with contribution from all authors. N.M.P. and M.B. supervised S.G. activity. L.C. and M.B. conceived and supervised the work. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by intramural funding from the University of Trento to L.C. and S.C. N.M.P. is supported by the European Commission under the Graphene Flagship Core 2 grant no. 785219 (WP14 'Composites') and FET Proactive 'Neurofibres' grant no. 732344, as well as by the Italian Ministry of Education, University and Research (MIUR) under the 'Departments of Excellence' grant L. 232/2016 and AR 901–01384—PROSCAN and PRIN-20177TTP3S.

**Acknowledgments:** The authors wish to thank Francesca Agostinacchio for helpful discussions and Mirco D'Incau for technical assistance for SEM imaging.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Data Availability:** The raw data required to reproduce these findings are available upon request to the corresponding authors.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **Tendon and Cytokine Marker Expression by Human Bone Marrow Mesenchymal Stem Cells in a Hyaluronate**/**Poly-Lactic-Co-Glycolic Acid (PLGA)**/**Fibrin Three-Dimensional (3D) Sca**ff**old**

#### **Maria C. Ciardulli 1, Luigi Marino 1, Joseph Lovecchio 2, Emanuele Giordano 2, Nicholas R. Forsyth 3, Carmine Selleri 1, Nicola Ma**ff**ulli 1,3,4 and Giovanna Della Porta 1,5,\***


Received: 30 March 2020; Accepted: 18 May 2020; Published: 20 May 2020

**Abstract:** We developed a (three-dimensional) 3D scaffold, we named HY-FIB, incorporating a force-transmission band of braided hyaluronate embedded in a cell localizing fibrin hydrogel and poly-lactic-co-glycolic acid (PLGA) nanocarriers as transient components for growth factor controlled delivery. The tenogenic supporting capacity of HY-FIB on human-Bone Marrow Mesenchymal Stem Cells (hBM-MSCs) was explored under static conditions and under bioreactor-induced cyclic strain conditions. HY-FIB elasticity enabled to deliver a mean shear stress of 0.09 Pa for 4 h/day. Tendon and cytokine marker expression by hBM-MSCs were studied. Results: hBM-MSCs embedded in HY-FIB and subjected to mechanical stimulation, resulted in a typical tenogenic phenotype, as indicated by type 1 Collagen fiber immunofluorescence. RT-qPCR showed an increase of type 1 Collagen, scleraxis, and decorin gene expression (3-fold, 1600-fold, and 3-fold, respectively, at day 11) in dynamic conditions. Cells also showed pro-inflammatory (IL-6, TNF, IL-12A, IL-1β) and anti-inflammatory (IL-10, TGF-β1) cytokine gene expressions, with a significant increase of anti-inflammatory cytokines in dynamic conditions (IL-10 and TGF-β1 300-fold and 4-fold, respectively, at day 11). Mechanical signaling, conveyed by HY-FIB to hBM-MSCs, promoted tenogenic gene markers expression and a pro-repair cytokine balance. The results provide strong evidence in support of the HY-FIB system and its interaction with cells and its potential for use as a predictive in vitro model.

**Keywords:** hBM-MSCs; cytokines; tenogenic markers; cyclic strain; 3D microenvironment; PLGA carriers; bioreactor

#### **1. Introduction**

Tissue engineering strategies for tendon healing and regeneration are designed to improve existing therapies or provide new treatment possibilities. Three-dimensional (3D) bioengineered systems have the potential to promote our understanding of the physiopathology of tendinopathy and the role of stem cells in tendon regeneration. In this sense, 3D scaffold design and fabrication coupled to specific bioreactor arrangements could develop highly predictive 3D in vitro culture and differentiation systems to explore cell behaviors in response to defined external biochemical and mechanical stimuli [1]. The 3D scaffold provides a model of fidelity via its provision of a microenvironment with defined stiffness and elastic modulus as well as the necessary surfaces for cell attachment [2–4].

Detailed understanding of cell behavior when incorporated into specific biomaterials allows to develop designs with specific functionalization. These functionalization may, for instance, stimulate local stem cells, attract specific circulating nucleated blood cells, such as macrophages, and induce their polarization into M2 phenotype to accelerate tissue regeneration and healing following the biomaterials in vivo implantation [5]. For example, human Mesenchymal Stem Cells (hMSCs) are largely used in tissue engineering strategies, and their immune-modulatory activity in the development of tendon pathologies have been explored, but the precise mechanisms involved remain undetermined [6–10]. Neutrophils and macrophages infiltrate injured tendons, potentially interacting with MSCs and stimulating cytokine release at the site of repair and promoting degradation of the extracellular matrix (ECM), inflammation, apoptosis, and, in the later stages of acute tendon healing, they release anti-inflammatory cytokines to alleviate inflammation and promote tendon remodeling [11–13].

Among the biomaterials described for tendon tissue engineering [14], a promising emerging strategy is the use of a complex biomimetic matrix with a hydrogel component and extracellular matrix mimicking properties [15,16]. Tenocyte precursors can be harvested from different sources, including periosteum [17,18], bone marrow [19–21], tendon [21,22], and adipose tissue [21,23]. To overcome the intrinsic poor mechanical properties of the hydrogel, they can be merged with more force resistant biopolymers. Cells and biomaterials alone are not sufficient to achieve optimal levels of differentiation and matrix organization. Mechanical stimulation has a key role in tenogenic differentiation induction [19]. Scaffolds are therefore required to display an appropriate elastic behavior to deliver strain [24] or compression [25] inputs. Strain is a tenogenic differentiation signal [26–29], and several bioreactors have been used to impart tenogenic mechanical stimuli to cells in culture [19,30–35]. For example, Rinoldi et al. designed and fabricated 3D multilayered composite scaffolds, where an electrospun nanofibrous substrate was coated with a thin layer of GelMA-alginate composite hydrogel carrying MSCs. MSCs were subsequently differentiated by the addition of bone morphogenetic protein 12 (BMP-12) and, to mimic the natural function of tendons, the scaffolds were mechanically stimulated using a custom-built bioreactor [34]. Grier et al. described an aligned collagen-glycosaminoglycan scaffold able to enhance tenogenic differentiation of MSCs via cyclic tensile strain within a bioreactor, in the absence of growth factor supplementation [36]. Another protocol, proposed by Youngstrom et al., promoted tenogenic differentiation of MSCs cultivated on decellularized tendon scaffolds with the application of 3% cyclic strain for one hour per day for 11 days [31]. Additionally, several growth factors and other small molecules can stimulate transcriptional activation of genes involved in tenogenic differentiation [15,37,38]. Growth Differentiation Factor 5 (GDF-5), for instance, induces the expression of genes linked to the neo-tendon phenotype [39–41].

Tendinopathies associated with physical activity and age-related degeneration are a major medical issue [23], and recent healing and regeneration studies include the use of human Bone Marrow Mesenchymal Stem Cells (hBM-MSCs) [42–44]. hBM-MSCs are a multipotent population present in bone marrow that can be readily differentiated in vitro [45,46] into cells of three mesodermal lineages, namely adipocytes, chondrocytes and osteoblasts under appropriate conditions [47–50]. MSCs-based therapies include direct transplantation of MSCs populations, growth factor-loaded scaffolds for local MSCs recruitment or implantation of scaffolds containing in vitro culture-expanded MSCs populations [51,52].

We previously described an engineered multiphase three-dimensional (3D) scaffold as an in vitro model for tendon regeneration studies. The multiphase 3D construct was totally absorbable and consisted of a braided hyaluronate elastic band merged with a fibrin hydrogel containing hBM-MSCs and poly-lactic-co-glycolic acid nano-carriers (PLGA-NCs) themselves loaded with human Growth Differentiation Factor 5 (hGDF-5) [53]. In that work, the PLGA nano-carriers were transient scaffold components to ensure sustained and controlled delivery of hGDF-5 with benefits beyond those associated with standard culture medium supplementation [53,54]. The study reported an early tenogenic commitment of hBM-MSCs after three days of cultivation under dynamic conditions.

In the present study, we describe the use of the same scaffold (named HY-FIB here) to investigate the effect of the 3D environment on hBM-MSCs for 11 days with or without mechanical stimulation and in the absence of any specific biochemical differentiation signal. HY-FIB was assembled with hBM-MSCs as previously described [53] including PLGA-NCs stratified within the 3D fibrin structure. Importantly, the PLGA-NCs carried an inactive form of human Growth Differentiation Factor 5 (ihGDF-5) enabling the overall 3D scaffold structure to be safely evaluated with or without mechanical input. Gene expression of type 1 Collagen, decorin, scleraxis, and tenomodulin were considered; type 3 collagen was also monitored, as negative control. Histology and quantitative immunofluorescence were used to monitor cell behavior and their interaction with the synthetic extracellular matrix. Moreover, to understand whether HY-FIB microenvironment configuration would stimulate any cell inflammation responses, the cells expression of cytokine markers was also monitored, including pro-inflammatory cytokines and anti-inflammatory ones. The results provide strong evidence that HY-FIB environment plus mechanical signaling, promoted tenogenic markers expression, collagen production and better pro-repair cytokine balance by hBM-MSCs.

#### **2. Materials and Methods**

#### *2.1. hBM-MSCs Isolation and Harvesting*

Human bone marrow mesenchymal stem cells (hBM-MSCs) were obtained from the bone marrow of three independent healthy donors (age 36, 38, 44 years). The donors gave written informed consent in accordance with the Declaration of Helsinki to the use of their filter residual bone marrow aspirate for research purposes, with approval from the University Hospital of San Giovanni di Dio e Ruggi d'Aragona (Salerno, IT). Review Board authorization number: (24988 achieved on April 9, 2015).

Briefly, total bone marrow aspirate was directly seeded at a concentration of 50,000 total nucleated cells/cm2 in T75 plastic flask in Minimum Essential Medium Alpha (α-MEM) supplemented with 1% GlutagroTM, 10% Fetal Bovine Serum (FBS), and 1% Pen/Strep and incubated at 37 ◦C in 5% CO2 atmosphere and 95% relative humidity [55]. After 72h, non-adherent cells were removed by medium change, and the adherent cells were further fed twice a week with new medium. On day 14, colonies of adherent hBM-MSCs were detached and re-seeded at 4000 cells/cm2 in the same culture conditions. Once the cell cultures reached 70–80% confluence, cells were detached using 0.05% trypsin-0.53mM EDTA and washed with PBS 1× (Corning Cellgro, Manassas, VA, USA), counted using Trypan Blue (Sigma-Aldrich, Milan, IT) and subcultured at a concentration of 4 <sup>×</sup> 103 cells/cm2. Flow Cytometry analysis was performed on hBM-MSCs obtained at Passage 1 examining levels of CD90, CD105, CD73 CD14, CD34, CD45, and HLA-DR expression (Miltenyi Biotec, DE).

#### *2.2. ihGDF-5 E*ff*ect on hBM-MSCs*

These sets of experiments were performed to assure the absence of any effect of inactivated human GDF-5 (ihGDF-5, Cloud-Clone Corp., USA) on both tenogenic markers stimulation and cytokines expression by hBM-MSCs. Cells were seeded on coverslips in 12 well plates at a concentration of <sup>4</sup> <sup>×</sup> <sup>10</sup><sup>3</sup> cells/cm2. Once the cultures reached 60% confluence, cells were treated with either 1.6 ng/mL or 100 ng/mL of ihGDF-5. Cells were fed twice a week with new medium and fresh ihGDF-5 supplementation for up to 16 days. Untreated cells for matched time-points studied were used for control purposes. Passage 3 cells were seeded in the 3D environment (~8 <sup>×</sup> <sup>10</sup><sup>5</sup> cells/mL) and were fed twice a week with new medium, without any growth factor added.

PLGA carriers were not tested because they cannot be supplemented to the cells planar monolayer culture, indeed, in static conditions their sedimentation on cells reduced oxygen exchange and prevented cells survival (data not shown).

#### *2.3. Immunofluorescence and Immunohistochemical Assays*

Cells were fixed with 3.7% formaldehyde for 30 min at room temperature (RT) followed by permeabilization with 0.1% Triton X-100 for 5 min and blocking with 1% Bovine Serum Albumine (BSA) for 1h. For type 1 and type 3 Collagen staining, cells were incubated overnight at 4 ◦C with a mouse monoclonal anti-type 1 Collagen antibody (1:100, Sigma-Aldrich) and a rabbit polyclonal anti-type 3 Collagen antibody (1:100, Santa Cruz Biotechnology). Following incubation with the primary antibody, cells were incubated for 1h at RT with the DyLight 649 anti-mouse IgG (1:500, BioLegend, CA) and the Alexa FluorTM 488 goat anti-rabbit IgG (1:500; Thermo Fisher Scientific, USA). Cell nuclei were stained with DAPI solution (1:1000) for 5 min. Images acquisition was at 20× magnification on a fluorescent microscope (Eclipse Ti-E Inverted Microscope; NIKON Instruments Inc., USA).

For 3D scaffold immunohistochemical analysis, slices were permeabilized with 0.1% Triton X-100 for 5 min, and non-specific staining blocked with 1% BSA for 1h at RT. For type 1 Collagen staining, slices were incubated overnight at 4 ◦C with a rabbit polyclonal anti-type 1 Collagen antibody (1:200, AbCam). Following incubation with the primary antibody, slices were incubated for 1h at RT with Alexa FluorTM 488 goat anti-rabbit IgG (1:400, Thermo Fisher Scientific, USA) antibody. Subsequently, cell nuclei were stained with DAPI solution (1:1000) and incubated for 5 min. Images were acquired as described above. Image quantification was performed using image analysis software (ImageJ, National Institutes of Health, USA) [56,57] by measuring the red and green areas where type 1 and type 3 collagen, respectively, are expressed. A minimum of 10 image fields was used for the image analysis at each time point. Signal intensity at each time point was normalized by the cell number (e.g., by amount of cell nuclei revealed by DAPI staining).

Sirius red staining was performed using the Picrosirius Red Stain Kit (Polysciences, Inc., USA). Sections of 15 μm of thickness were stained in hematoxylin for 8 min, then washed in water for 2 min. The sections were dipped into phosphomolybdic acid for 2 min, then washed in water for 2 m. Then they were dipped into Picrosirius Red F3BA Stain for 60 min and dipped into HCl 0.1M solution for 2 min. The sections were dehydrated in increasing ethanol gradient solutions (70–75–95–100%) and finally dipped into xylene for 5 min. Eukitt medium was used to mount the samples.

#### *2.4. RNA Isolation and Gene Expression Profile by Quantitative Reverse Transcription PCR (RT-qPCR)*

Total RNA was extracted from hBM-MSCs seeded into the 3D construct of each experimental group using QIAzol® Lysis Reagent (Qiagen, DE), chloroform (Sigma-Aldrich, Milan, IT) and the RNeasy Mini Kit (Qiagen, DE). For each sample, 300 ng of total RNA was reverse-transcribed using the iScriptTM cDNA synthesis kit (Bio-Rad, Milan, IT). Relative gene expression analysis was performed in a LightCycler® 480 Instrument (Roche, IT), using the SsoAdvancedTM Universal SYBR® Green Supermix (Bio-Rad) with the validated primers for COL1A1, COL3A1, DCN, IL-1β, IL-6, IL-10, IL-12A, SCX-A, TGF-β1, TNF, and TNMD (Bio-Rad), and following MIQE guidelines [58]. Amplification was performed in a 10 μL final volume, including 2 ng of complementary DNA (cDNA) as template. Specificity of the formed products was addressed via melting curve analysis. Triplicate experiments were performed for each condition explored, and data were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression (reference gene), applying the geNorm method [59] to calculate reference gene stability between the different conditions (calculated with CFX Manager software; M <0.5). Fold changes in gene expression were determined by the 2−ΔΔCp method, and are presented as relative levels versus hBM-MSCs just loaded within the HY-FIB system.

#### *2.5. PLGA-NCs Fabrication, Size, Morphology, and ihGDF-5 Release Profile*

PLGA nano-carriers (PLGA-NCs) were obtained using Supercritical Emulsion Extraction (SEE) technology enabling rapid polymer NCs production from multiple emulsions via dense gas extraction of the oily phase organic solvent utilizing a countercurrent packed tower operating in continuous mode [60]. In detail, ihGDF-5 (Cloud-Clone Corp., USA) was dissolved into 1% (*w*/*v*) human serum albumin (hSA; Sigma-Aldrich, Milan, IT) containing 0.06% polyvinyl alcohol (PVA). Human serum albumin (hSA) was included as an ihGDF-5 stabilizer. This solution was added to the oily phase formed in an Ethyl Acetate (EA, purity 99.9%) and PLGA (RG 504H, 38,000–54,000 kDa, Evonik, DE) at 5% (*w*/*w*) solution. All emulsions were processed immediately after their preparation.

SEE technology operative pressure and temperature conditions in the high-pressure column were set at 8 MPa and 38 ◦C, respectively, with a dense gas flow of Carbon Dioxide (CO2) set at 1.4 kg/h with Liquid/Gas ratio of 0.1 (*w*/*w*) [61]. Carrier suspensions were collected at the bottom of the extraction column, washed, and lyophilized. Each run allowed the recovery of 98% of the loaded biopolymer. Empty and loaded NCs were produced using the same process conditions.

Carrier particle size distributions (PSDs) were measured using a laser granulometer (mod. Mastersizer S; Malvern Instruments Ltd., Worcestershire, UK), based on dynamic light scattering (DLS). Sizes are expressed as volume mean size (MS) with standard deviation (SD) in nanometers (nm). The shape and morphology of the PLGA-NCs were investigated by field emission-scanning electron microscopy (FE-SEM; mod. LEO 1525; Carl Zeiss, Oberkochen, D). Samples were placed on a double-sided adhesive carbon tape previously glued to an aluminum stub and coated with a gold film (250 A thickness) using a sputter coater (mod.108 A; Agar Scientific, Stansted, UK).

ihGDF-5 release profile was monitored in vitro from 20(±0.3) mg of carriers suspended in 1mL of α-MEM, placed in an incubator at 37 ◦C, and stirred continuously at 1× *g*. Every 24 h, samples were centrifuged at 160× *g* for 10 min and the supernatant completely removed and replaced with fresh media to maintain sink conditions. Released ihGDF-5 concentrations from collected samples were then measured with an Enzyme Linked Immunosorbent Assay (ELISA, Cloud-Clone Corp., USA). Release experiments were performed in triplicate (n = 3), and the curve describing the mean profile calculated as ng/g (protein released/PLGA-NCs) versus time.

#### *2.6. HY-FIB Preparation and Characterization*

For each sample, a mixture of 50 mg/mL fibrinogen from human plasma (Sigma-Aldrich, Milan, IT), 15,600 U/mL aprotinin (Sigma-Aldrich, Milan, IT), and α-MEM (Corning, NY, USA) supplemented with 10% FBS (referred to as growing media, GM) was added at a 1:1:1 ratio to 100 mg of PLGA-NCs (ihGDF-5 loading: 350 ng/g) and, then, to an average of 8 <sup>×</sup> <sup>10</sup><sup>5</sup> cells. A homogeneous cells/PLGA-NCs/fibrinogen suspension was then embedded into a mold (30 × 20 × 4.5 mm) where the braided band had been previously positioned. Free ends were left to enable HY-FIB fixing into the bioreactor. Upon addition of 100 U/mL thrombin (Sigma-Aldrich, Milan, IT), the mold was placed in a 37 ◦C humidified incubator for 30 min to allow fibrin polymerization. When the hydrogel was formed, the band was entrapped inside a uniformly distributed hydrogel. The construct was then transferred from the mold to either a standard polystyrene culture plate or to the bioreactor culture chamber, each containing 30 mL of the culture media, and placed in an incubator at 37 ◦C in a 5% CO2 atmosphere and 95% relative humidity.

HY-FIB morphology was observed by field emission-scanning electron microscopy (FE-SEM; mod. LEO 1525; Carl Zeiss, Oberkochen, Germany). Samples were fixed in 4% PFA (4 ◦C, overnight) and then dehydrated by multiple passages across ethanol:water solutions (10 min each) with increasing concentrations of ethanol (10%, 20%, 30%, 50%, 70%, 90%), ending in a 100% dehydrating liquid (3 changes, 10 min each).

Samples were then lyophilized in a Critical Point Dryer (mod. K850 Emitech, Assing, Rome IT), placed on a double-sided adhesive carbon tape previously glued to an aluminum stub and coated with a gold film (250 A thickness) using a sputter coater (mod. 108 A; Agar Scientific, Stansted, United Kingdom) before observation.

HY-FIB mechanical characterization was performed according to the ASTM 1708 by a dynamometer (CMT 6000 SANS, Shenzen, China) equipped with a 1 kN load cell. The sample was conditioned in Dulbecco's Modified Essential Medium (DMEM) for 1 h, and then shaped to obtain a specimen with gauge length (Lo) of 22 mm and width (W) of 5 mm. Sample thickness (S) was measured with a thickness gauge brand at three different averaged points. Monoaxial deformation was applied to the sample at a speed of 10 mm/min, and force (F) and elongation (L) during traction were recorded. The elastic modulus and ultimate tensile strength (both expressed in MPa) were calculated from the stress/strain plot. For the immuno-histochemical analysis, at different time points, a portion of HY-FIB was fixed in 4% PFA (4 ◦C, overnight), cryo-protected in 30% sucrose overnight, mounted in OCT embedding compound, frozen at −20◦C and then cut in slices of 10 μm of thickness using a cryostat. The remaining portion of HY-FIB was placed in QIAzol® Lysis Reagent for total RNA extraction.

#### *2.7. Dynamic Culture*

HY-FIB was clamped at both free ends, one motionless and one sliding (operated by a linear motor actuator) arm, into the bioreactor system culture chamber, described in detail elsewhere [23]. A maximal load, set by pre-tensioning, was relaxed to a minimum value cycling at a pre-determined frequency. In addition, continuous feedback signals provided by strain gauges located onto the fixed arm, allowed the maintenance of a defined load on the scaffold in response to physical system modifications, by automatic adjustment of the pre-tensioning position.

#### *2.8. Finite Element Modeling*

Finite Element Modeling (FEM) was implemented by using COMSOL Multiphysics Software (rel. 5.3a®) to assess the stress distribution into 3D constructs when a 10% mechanical cyclic strain stimulus is applied. All components were obtained using primitive geometries and Boolean operations. Linear elasticity equations were set as boundary conditions for the scaffold. A sensitivity study of the mesh obtained the most computationally efficient solution. The specific parameters used in the model are listed in Table 1. The simulation considered only the stress within the fibrin 3D environment neglecting any further contribution of the band, cells and PLGA-NCs.


**Table 1.** Finite Element Modeling (FEM) parameters used.

<sup>1</sup> HY-FIB Young Modulus was measured experimentally [53] <sup>2</sup> Poisson ratio value was taken in the literature [62].

#### *2.9. Statistical Analysis*

Statistical analysis was performed using GraphPad Prism software (6.0 for Windows). Data obtained from multiple experiments are expressed as mean+/−SD and analyzed for statistical significance using ANOVA test, for independent groups. Differences were considered statistically significant when *p* ≤ 0.05 [63].

#### **3. Results**

#### *3.1. hBM-MSCs Cultivation in 2D Environment*

hBM-MSCs were cultivated in a two-dimensional (2D) monolayer environment with medium supplemented with either 1.6 ng/mL or 100 ng/mL of inactive human Growth Differentiation Factor 5 (ihGDF-5) for up to 16 days. These two concentration conditions (two order of magnitude of difference) were chosen to ascertain absence of any effect of the inactive hGDF-5 form on cells expression of

tenogenic markers (COL3A1, COL1A1, DCN, SCX-A and TNMD) and of cytokines (pro-inflammatory: IL-6, TNF, IL-12A, IL-1β; anti-inflammatory: IL-10, TGF-β1) by RT-qPCR (see Figure 1a–d).

**Figure 1.** Gene expression profiles for tenogenic markers and pro-inflammatory and anti-inflammatory cytokines by hBM-MSCs treated with 1.6 ng/mL (**a**,**c**) and 100 ng/mL (**b**,**d**) of ihGDF-5 in monolayer 2D culture up to 16 days. mRNA levels of COL1A1, DCN, SCX-A, and TNMD were considered as tenogenic markers and COL3A1 selected as negative ones; pro-inflammatory (IL-6, TNF, IL-12A and IL-1β) and anti-inflammatory (IL-10 and TGF-β1) cytokines were monitored. ihGDF-5 not had significant impact on genes expression, especially at the lower concentration tested. Untreated cells for matched time-points were used as control. \* ≤ 0.05; \*\* <0.01; \*\*\* <0.005; \*\*\*\* <0.001 N = 3 (biological replicates); n = 3 (technical replicates).

Transient and slight, though significant, upregulation of COL3A1 (0.4-fold), DCN (0.2-fold), and COL1A1 (0.5-fold) was observed at Day 1 in cultures supplemented with 100 ng/mL ihGDF-5 (Figure 1b). The reduced dose of 1.6 ng/mL induced low-level transient expression at Day 1 for only COL3A1 (0.3-fold) and DCN (0.2-fold) only (Figure 1a). No significant upregulation was noted for TNF, IL-12A, IL-1β, IL-10, or TGFβ at any time point or ihGDF-5 concentration tested (Figure 1c,d). Compared to controls, IL-6 displayed significant levels of elevation at Days 8 (0.8-fold) and 16 (0.5-fold) with 100 ng/mL ihGDF-5 (Figure 1d). Types 1 and 3 Collagen expression levels were monitored by immunofluorescence during the 16 day culture period, as illustrated in Figure 2a,b. Immunofluorescence quantitative data by image analysis were congruent with RT-qPCR outputs when 1.6 ng/mL ihGDF-5 was supplemented, in this case, both proteins signals were not significantly elevated, compared to untreated cells (see Figure 2c).

**Figure 2.** IF and quantitative-IF assays of type 1 Collagen (COL1A1) and type 3 Collagen (COL3A1) monitored along hBM-MSCs treatment with 1.6 ng/ and 100 ng/mL of ihGDF-5 for 16 days. Type 1 collagen was stained in red; type 3 collagen was stained in green; cell nuclei highlighted with DAPI in blue (**a**,**b**). Quantitative signal detection was performed via ImageJ software (**c**). A slight up-regulation of COL1A1 was observed when hBM-MSC were in routine culture for 16 days. Color intensity in each time point was normalized by the cell number. \* ≤0.05; \* \*<0.01; \*\*\* <0.005; \*\*\*\* <0.001. n = 10 (image fields for each time point).

Quantitative image analysis displayed COL1A1 signal increase (1 fold) at Day 16 only with 100 ng/mL ihGDF-5 supplementation. This last data is in contrast with gene expression ones. COL1A1 and COL3A1 are the major components of the extracellular matrix in connective tissues, and their slight up-regulations was reported when hBM-MSC were in routine culture for 16 days [64]. However, in our case, ihGDF-5 seemed not to impact on their production, especially at the lower concentration tested. This preliminary information is important to confirm the inactivity of the biochemical input in regards to the gene expression and proteins that will be monitored in the 3D experiments.

#### *3.2. PLGA-NCs Fabrication and ihGDF-5 Controlled Deliver*

Poly-lactic-co-glycolic-acid nano-carriers (PLGA-NCs) displayed a spherical morphology with a mean size of 230 ± 80 nm (Figure 3a). PLGA-NCs had an ihGDF-5 loading of 350 ng/g and provided a daily released peptide mean concentration of 1.6 ng/mL/day (Figure 3b), when an amount of 100 mg were inserted within HY-FIB over 11 days of culture. As highlighted above, these ihGDF-5 concentration levels did not stimulate sustained impacts on hBM-MSCs gene expression in 2D monolayer culture (see Figure 1a,c).

**Figure 3.** Poly-lactic-co-glycolic-acid (PLGA) transient carriers field emission-scanning electron microscopy (FE-SEM) image, particle size distribution, and ihGDF-5 release profile. FE-SEM images indicated spherical morphology of carriers (**a**); the size distribution set at 230 ± 80 nm their mean size (**b**). Release profiles performed in vitro at 37 ◦C indicated a ihGDF-5 mean concentration of 1.6 ng/mL/day released from the 100 mg of PLGA-NCs loaded within HY-FIB over 11 days of culture (**c**).

Therefore, by excluding any non-specific ihGDF-5 induction (released within the 3D scaffold by the NCs), we could now observe cell behaviors arising from the HY-FIB microenvironment in both static and dynamic conditions.

#### *3.3. hBM-MSCs Cultivation in HY-FIB 3D Microenvironment*

The HY-FIB assembly featured a braided band (3 × 10cm) joined to a fibrin hydrogel (on a band surface of 6 cm2) containing 8 <sup>×</sup> 105 hBM-MSCs and 100 mg of ihGDF-5/PLGA-NCs. The picture and schematic representation of the 3D system is shown (Figure 4a,b). Field Emission Scanning Electron Microscopy (FE-SEM) images of the scaffold illustrate hyaluronate fibers, embedded within a fibrin hydrogel (Figure 4c), which provided an entrapment surface for both NCs and hBM-MSCs (Figure 4d).

HY-FIB was exposed to 10% deformation over a 1 Hz frequency for 4 h a day during the dynamic culture experiments via a cyclic strain bioreactor, illustrated in Figure 4e [24]. In greater detail, a HY-FIB braided band was held at one end by a motionless arm and at the other end by a sliding one. Motion was driven by a linear motor and transmitted through the braided band to cells embedded within the fibrin hydrogel. The motionless arm comprises a base, attached to the side wall of the culture chamber, housing the electronic components for load monitoring, and from which extended a cantilevered shelf whose deformation is measured by four strain gauges. The whole system was housed within an incubator to ensure the appropriate CO2 gaseous environment to control the pH of the cell culture media and 37 ◦C operational temperature.

**Figure 4.** HY-FIB three-dimensional (3D) scaffold features and cyclic strain bioreactor. Schematic HY-FIB representation (**a**) and image of 3D scaffold (**b**)**.** SEM images of hyaluronate braided fibers (10 μm mean diameter size) (**c**) joined to a fibrin web which entrapped both NCs and hBM-MSCs (**d**). Cyclic strain bioreactor (**e**) and in-silico study of stress distribution over HY-FIB upon mechanical strain of 10% (**f**). The simulation involved only the stress of the fibrin 3D environment, neglecting any further contribution.

The stress delivered to the cells immobilized within the system was explored via computational analysis that estimated a mean shear stress value estimated at 9 <sup>×</sup> 10−<sup>2</sup> Pa within the fibrin 3D environment (Figure 4f). This order of magnitude of stress value was reported for tenogenic induction [30]; larger deformation for longer times were excluded to focus the study on 3D environment assembled.

HY-FIB samples were collected at Day 1, 2, 5, and 11 to monitor tenogenic and cytokine marker expression. Time points at Day 1 and Day 2 were added for 3D culture to monitor the effect of HY-FIB on cells behavior alone or in combination with cyclic strain culture. Indeed, in static conditions COL1A1 and DCN both displayed significant upregulation of 3.8 fold (COL1A1) and 2.6-fold (DCN) at Days 1 and 2 before dropping progressively to elevated but non-significant levels (Figure 5a and Figure S1 in Supplementary Materials), confirming an HY-FIB effect on this gene expression in the first days of culture. In dynamic conditions, COL1A1 levels displayed responses similar to

the static culture in the first two days but progressively rising thereafter to significant levels (2.9 fold) at Day 11, probably due to strain input. DCN expression levels in response to dynamic culture were to be elevated throughout, achieving significance at Day 11 (3-fold) (Figure 5b and Figure S1 of Supplementary Materials).

**Figure 5.** Gene expression profiles of tenogenic markers and pro-inflammatory and anti-inflammatory cytokines from hBM-MSCs within HY-FIB environment in static (**a**,**c**) and dynamic culture (**b**,**d**) up to 11 days. Days 1, 2, 5, and 11 were selected as time points to study the mRNA levels of positive tenogenic markers (COL1A1, DCN, SCX-A, and TNMD), negative ones (COL3A1) and pro-inflammatory (IL-6, TNF, IL-12A and IL-1β) and anti-inflammatory (IL-10 and TGF-β1) cytokines. Effect of HY-FIB environment on cells behavior was visible along the first two days of culture; a better over expression of tenogenic markers and anti-inflammatory cytokines was observed in dynamic culture at Day 11. hBM-MSCs within HY-FIB at time zero were used as control. \* <0.05; \*\* <0.01; \*\*\* <0.005; \*\*\*\* <0.001. N = 3 (biological replicates); n = 3 (technical replicates).

SCX-A displayed significant upregulation (~340-fold) in both static and dynamic conditions at Day 1, suggesting an effect of HY-FIB system, on this gene expression. SCX-A levels were substantially elevated in both static and dynamic condition at all following time points studied, even if a larger and significant increase was observed in dynamic condition; an increase of 800-fold in static and of 1600-fold in dynamic culture conditions was monitored at Day 11 (Figure 5a,b and Figure S1 of Supplementary Materials). Tenomodulin gene expression was also tested by RT-qPCR, but no expression was detected, probably because it is an event occurring during late differentiation [65]. Sustained, significant, downregulation of COL3A1 was observed in either static or dynamic conditions which instead either decreased progressively (static) or decreased through to Day 5 before reestablishing Day 0 levels at Day 11 (dynamic).

The data suggested an overall effect of the 3D environment on cells behavior clearly visible along the first two days of culture; furthermore, a statistically significant COL1A1, DCN, and SCX-A over-expression was observed after 11 days when mechanical strain was provided (Figure S1 in Supplementary Materials).

Cytokine transcript expression data is illustrated in Figure 5c,d, for static and dynamic culture, respectively. HY-FIB system has an effect also on cytokines gene expression, as observed in all time points monitored with respect to Day 0, within static culture. Indeed, pro-inflammatory cytokines IL-6 (~6-fold), TNF (~10-fold), IL-12A (≤600-fold), and IL-1β (~200-fold) displayed rapid and significant upregulation that was maintained for the entirety of the experimental duration. Anti-inflammatory TGF-β1 on the other hand displayed either no change (Day 11) or significant down-regulation (other time points) while IL-10 exhibited an overall similar profile to Il-1β culminating in marked upregulation at day 11 (~300-fold) (Figure 5c).

Dynamic culture conditions had a distinct and significant effect on IL-6 with expression levels achieving a peak upregulation of 1.5-fold at Day 1 and decreasing to undetectable levels by Day 11 (Figure S2 in Supplementary Materials). TNF and IL-1β were both gradually upregulated before achieving ~200-fold and ~300-fold, respectively, upregulations at day 11 (compared to 10-fold and 100-fold in static conditions). IL-12 displayed a similar profile of upregulation in dynamic vs. static culture conditions while achieving maximal levels that were 3X less in dynamic. Anti-inflammatory IL-10 expression levels were consistent across both dynamic and static conditions. In contrast to static culture, TGF-β1 was significantly downregulated until day 5, and it underwent a 5-fold increase at Day 11 in dynamic culture conditions (Figure 5d and Figure S2 in Supplementary Materials).

Histological characterization of HY-FIB scaffold in both static and dynamic culture at Days 5 and 11 are reported in Figure 6; the overall scaffold structure was stained with Sirius Red for collagen highlighting. Despite fibrin hydrogel matrix, collected at Day 0, was only light pink stained, the same matrix was clearly stained in red at Day 5 and 11 in both samples taken from static and dynamic culture. However, a less homogeneous matrix organization and staining was observed in the samples taken from static culture. This data is in agreement with gene expression indications and confirmed that both HY-FIB alone and HY-FIB plus cyclic strain had an effect on cells phenotype commitment.

**Figure 6.** Histology characterization of the overall HY-FIB scaffold structure with Sirius Red staining. HY-FIB scaffolds in both static and dynamic culture at Days 5 and 11 are reported; the overall scaffold

structure was stained with Sirius Red for collagen highlighting. Fibrin hydrogel was light pink stained in the sample collected at Day 0. Fibrin matrix was clearly stained in red at Day 5 and 11 in both samples from static and dynamic culture. Less homogeneous scaffold matrix structure and staining was observed in samples taken from static culture.

The expression of type 1 Collagen, a tenogenic matrix-associated marker, was monitored by immunofluorescence over the culture time (see Figure 7). At day 1, empty areas surrounding the cells are present, probably due to absence of uniform fibrin hydrogel. These spaces were then progressively filled with the protein, presumably via secretion into the extracellular environment. The level of staining observed under static conditions decreased after Day 1 and was maintained at ~50% of original levels thereafter while levels were maintained consistent to Day 0 in dynamic culture. Moreover, in the dynamic condition a more uniform cells distribution was noted throughout the hydrogel matrix, especially at day 11.

**Figure 7.** IF and quantitative-IF assays of type 1 Collagen (COL1A1) in 3D static and dynamic2 culture of hBM-MSCs for 11 days. At day 1, holes within the hydrogel structure are evident nearby cells

surroundings. These areas were progressively filled by COL1A1 protein (stained in green) and it happened more uniformly during the dynamic cultivation (see day 11). Fluorescence quantification by ImageJ software, reported in the plot, confirmed the presence of more abundant signal (0.2–0.3 fold changes) in dynamic condition; signal intensity in each time point was normalized by cell number (e.g., by amount of cell nuclei revealed by DAPI staining). \* ≤0.05; \*\* <0.01. N = 3 (biological replicates); n = 3 (technical replicates).

#### **4. Discussion**

The HY-FIB system is engineered to support delivery of PLGA nanocarriers (PLGA-NCs) within the hydrogel matrix, enabling controlled delivery of specific molecules within 3D environment, e.g., drugs or other biological signals. The active form of hGDF-5 loaded into PLGA-NCs for controlled delivery within HY-FIB environment was investigated in a previous study [53]. Here, we investigated the effect of the HY-FIB 3D environment (hyaluronate band + PLGA carriers + fibrin gel) on hBM-MSCs tenogenic and cytokine marker gene expression in both static and dynamic, mechanical, input scenarios. We adopted the previous HY-FIB configuration including PLGA-NCs, but on this occasion, we delivered an inactive form of hGDF5. In this manner, the biochemical input provided by the growth factor was excluded, but the complete HY-FIB configuration was maintained, and we were thus able to investigate the impact of mechanical input alone.

HY-FIB braided fibers enabled a defined mechanical stimulation of 9 <sup>×</sup> 10−<sup>2</sup> Pa provided to hBM-MSCs during the 4h/day dynamic culture regime. The mean shear stress was calculated by FEM modeling [62], assuming a system homogeneous behavior at a density of 1050 kg/m3 and Young's modulus of 4.56 Mpa [53]. A Poisson ratio value of 0.25 was adopted as described elsewhere [62]. Further mechanical inputs with different intensities and durations were not investigated, not being the aim of the present work. Stress values resembling reduced physiological activity, similar that the ones used here, have been reported to direct stem cell commitment to a tenogenic phenotype [31,36,66].

COL1A1 is the major component of tendon tissue (75–85% of the dry mass of tendon), and is responsible for its mechanical strength [64]. In the static group, COL1A1 showed a ≥3-fold upregulation during the first and second day of cultivation. These data seem to suggest an overall effect of the 3D environment on cells behavior. COL1A1 expression was progressively reduced to a 2-fold upregulation at Day 11, in static environment. In dynamic conditions, its mRNA levels showed similar behavior during the first two days of cultivation (an increase up to 3-fold-changes, then reduced at Day 2). However, its expression was subsequently increased again to 2.8-fold at Day 11. Decorin (DCN), a small leucine-rich proteoglycan implicated in the regulation of fibrillogenesis, is a fundamental component of the tendon extracellular matrix (ECM) [67]. Compared to the static condition, a significant enhancement, up to 2.5-fold, of the mRNA level of DCN was shown when hBM-MSCs were cultured for 11 days with mechanical stimulation.

Scleraxis-A (SCX-A) is a neotendon marker, expressed in pro-tendon sites in the developing embryo. Specifically, SCX-A is a tendon-specific basic helix-loop-helix transcription factor responsible for the transition of MSCs into tendon progenitors [68]. We observed substantial increases in SCX-A expression, up to 800-fold in static and 1600-fold in dynamic conditions after 11 days, demonstrating a stimulatory effect via the 3D system organization and consistent with previous observations [69–71].

COL3A1 mRNA level seems to be downregulated after 2 days in the static group and after 5 days in the dynamic group. Its downregulation can be considered a positive indication of proper cell differentiation; indeed, it seems that COL3A1 is the main responsible of fibrotic and scarred tissue arrangement and has been consistently reported at the rupture site of human tendons. [64].

From both histology and immunofluorescence assays, we noted that the area surrounding the cells was progressively filled by type 1 Collagen and, at Day 11, the extracellular matrix seemed to undergo remodeling (Figures 6 and 7). Moreover, in dynamic conditions a more homogeneous cell distribution within the hydrogel matrix was observed in the IF images. These findings support the concept that 3D cultivation provides cues to the hBM-MSCs, and that dynamic signaling enables the adoption of a more uniform behavior including type 1 Collagen protein deposition in the externally available space of the fibrin hydrogel. A near total absence of type 3 Collagen was found, except for a very small fluorescence signal at day 1, in both static and dynamic conditions (data not shown). These data suggest that tenogenic commitment of hBM-MSCs cultured within HY-FIB environment may be enhanced when dynamic stretching is applied.

MSCs secrete a variety of cytokines and growth factors that promote cell recruitment, migration, proliferation, and differentiation. MSCs are also immunomodulatory, which may allow them to exert beneficial effects on the local immune cell population at the site of muscle injury [72]. To better understand the hBM-MSCs inflammatory response when cultured within HY-FIB, cytokine expression was monitored along the 11 days of culture. The balance between pro- and anti-inflammatory soluble factors in the tendon healing process exerts a major impact on successful resolution of inflammation [73]. Recent analysis of tendinopathy biopsies showed a distinct inflammatory infiltrate in the initial phase of tendinopathy with a high content of pro-inflammatory factors such as IL-6, TNF-α and IL-17 [74].

To exclude a role for ihGDF5 in cytokine expression induction we evaluated their expression in hBM-MSCs undergoing 2D planar cultivation as a negative control. Indeed, in disc degeneration models using in vitro three-dimensional cultures, human annulus cells display increased expression of pro-inflammatory cytokines, such as IL-1β and TNF-α, while exposure to TNF-α and IL-1β resulted in significant downregulation of GDF-5 [75]. Therefore, it is plausible that the GDF-5 may upregulate the expression of pro-inflammatory genes in hBM-MSCs leading to the maintenance of an autocrine feedback. However, when the ihGDF-5 was added, no statistically significant expression of pro-inflammatory cytokines was observed; therefore, ihGDF-5 did not exert any effect on cytokines expression.

The addition of PLGA-NCs enabled an informed analysis of the inductive role of the HY-FIB overall structure. Previous studies have noted that cytotoxicity of SEE-fabricated PLGA-NCs on blood mononucleate viability, monitored with MTT assay [76], was not affected after either 24 or 48 h. Here, the overall HY-FIB system (loaded with SEE fabricated NCs) does not evidence any toxic effect on hBM-MSC cultivated within it for 11 days, providing an indirect indication about SEE technology as suitable process for biomedical carriers production.

In general, we observed that pro-inflammatory gene expression was higher in static than in dynamic conditions at all-time points. On the contrary, the anti-inflammatory cytokines IL-10 was consistently upregulated in both static and dynamic conditions; while TGF-β1 was downregulated at all the time points tested except day 11, when, it showed a marked increase (4-fold) only in dynamic environment. The described behavior confirmed that MSCs respond to a variety of biophysical cues; indeed, as suggested by Qazi et al., 3D culture of MSCs on biomaterials can promote cell-cell interactions and enhance the paracrine effects of MSCs [77]. Moreover, as concluded by Ogle et al., historically, biomaterial-based therapies to promote tissue regeneration were designed to minimize the host inflammatory response. Recently, the roles that monocytes and macrophages can play in tissue repair have been highlighted. In this context, material properties and their possibility of specific biomolecule controlled delivery has been engineered to achieve a given biological response that can be tuned not only to a better integration with biological systems but also in regulating the inflammatory response [5].

The overall and statistically significant balance of pro- vs. anti-inflammatory cytokines expressed by cells provided indications regarding the importance of dynamic culture for 3D in vitro model systems. For instance, IL-6, a well-known pleiotropic cytokine delivered by tissues in response to physio-pathological changes such as physical exercise, infection, and injury, was reported to deeply alter skeletal muscle milieu, by affecting the activity and quality of cellular interactors during tissue regeneration and leading to the fibrotic response [78]. In our 3D model system, IL-6 gene expression was considerably reduced in hBM-MSCs that underwent dynamic 3D HY-FIB cultivation when compared to the same cells cultivated in static condition.

It is worth of note that there is no specific literature on cytokines response by hBM-MSCs cultivated within 3D scaffold. Almost all published studies described cell-specific differentiation toward a given phenotype, without considering how cytokines expression may be related to a 3D in vitro scaffold system. In this sense, the present investigation is the first study, which suggests cytokines expression as a further variable to monitor cell behavior and reaction when loaded into a 3D in-vitro model. Moreover, improved balance in anti-inflammatory cytokines observed for HY-FIB plus cyclic strain may be considered an indication of better cells response to the 3D in vitro system designed and proposed.

#### **5. Conclusions**

The 3D cell culture yielded evidence of type 1 Collagen expression observed by both immunohistology and gene assay. When the same 3D system was cultivated under cyclic strain stimulation, the mechanical input stimulated a statistically significant increase in tenogenic markers expression when compared to the same cells assembled into the 3D system, but cultivated in a static culture. Further studies may involve a deeper understanding of the relation between collagen type I production, cell commitment and mechanical input strain percentage or duration; in this sense, HY-FIB system can be considered a good instrument for this study. The 3D culture system activated also the expression of pro-inflammatory cytokines, and, when cyclic strain was applied, pro-inflammatory cytokine gene over-expression by hBM-MSCs was better balanced against over-expression of anti-inflammatory cytokines. It remains to be determined what the involvement and the immunomodulatory activity of hBM-MSCs are, and the role of implantable biomaterials in the stimulation of inflammatory reactions. For instance, the stimulation of local inflammation is reported as an important event in triggering repair in avascular tissues, such as cartilage and tendons [5].

On the other hand, the presence of PLGA-NCs within the fibrin hydrogel would allow the delivery of specific biomolecules that may be studied for the ability to further modulate inflammation reactions or promote regeneration/repair events. In this sense, HY-FIB provides a potential strategic approach to address a range of issues via the provision of a tightly controlled in vitro protocol. The 3D scaffold is a potential system to organize the sustained release of different biochemical signals and opens concrete perspectives for developing 3D bioengineered models to understand specific molecular and cellular composition of damaged systems.

**Supplementary Materials:** The following materials are available online at http://www.mdpi.com/2073-4409/9/ 5/1268/s1, Figure S1: Profiles of positive (SCX-A, COL1A1, DCN) and negative (COL3A1) tenogenic markers between HY-FIB static vs. dynamic culture. \* < 0.05; \*\* < 0.01; \*\*\* < 0.005; \*\*\*\* < 0.001, Figure S2: Profiles of pro-inflammatory (IL-6, TNF, IL-12A, IL-1β) and anti-inflammatory (IL-10, TGFβ1) cytokines markers between HY-FIB static vs. dynamic culture. \* < 0.05; \*\* < 0.01; \*\*\* < 0.005; \*\*\*\* < 0.001.

**Author Contributions:** M.C.C. developed the experimental activity and optimized the protocols and methodology; she was also responsible for the paper draft preparation; L.M. isolated the stem cells and characterized them with formal analysis and validated methodology; J.L. followed the cyclic strain bioreactor protocols and provided the FEM data; E.G. supervised the bioreactor protocols and data acquisitions; N.R.F. provided contribution in data supervision and paper writing; C.S. provided the bone marrow aspirate and the methodology for hBM-MSCs cultivation; N.M. helped in the interpretation of the data, reviewed the manuscript, and was responsible for funding acquisition and management; G.D.P. was responsible for all experimental data production, curation and supervision, paper writing and editing, and research project administration. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by: (i) University of Salerno: FARB-Della Porta Year: 2019-22; (ii) American Orthopedic Foot & Ankle Society (AOFAS) Research Committee, Grant ID#: 2019-133-S, Grant Project Title: Nano-FT3C+: An Innovative Liposome-Based Formulation for Thyroid Hormone Controlled Delivery. An In Vitro Study on Tendinopathic Human Achilles Tendon Tenocytes.

**Acknowledgments:** The authors acknowledge MiUR within the framework of PON-RI 2014/2020. Action I.1–"Innovative PhDs with industrial characterization" Cicle XXXIII (D.D. n 0001377 June 5th, 2017) for the PhD fellowship entitled: "Scaffold innovation for the cure of tendon disorders: new generation of poly-hyaluronate functionalized biocomposites. The authors acknowledge Devis Galesso and R&D team at Fidia Farmaceutici S.p.A (Abano Terme, PD, Italy) for their participation into the PhD training program. We also thanks Annamaria Giordano for help in cells culture during her laboratory internship.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

**Patents:** The SEE technology for nanocarriers fabrication was described in the US Patent US/8628802 B2 Jan 2014. Inventors: Reverchon E., Della Porta G., Continuous process for microspheres production by using expanded fluids. Applicant: University of Salerno.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **Self-Assembling Sca**ff**olds Supported Long-Term Growth of Human Primed Embryonic Stem Cells and Upregulated Core and Naïve Pluripotent Markers**

#### **Christina McKee 1,2, Christina Brown 1,2 and G. Rasul Chaudhry 1,2,\***


Received: 5 October 2019; Accepted: 14 December 2019; Published: 16 December 2019

**Abstract:** The maintenance and expansion of human embryonic stem cells (ESCs) in two-dimensional (2-D) culture is technically challenging, requiring routine manipulation and passaging. We developed three-dimensional (3-D) scaffolds to mimic the in vivo microenvironment for stem cell proliferation. The scaffolds were made of two 8-arm polyethylene glycol (PEG) polymers functionalized with thiol (PEG-8-SH) and acrylate (PEG-8-Acr) end groups, which self-assembled via a Michael addition reaction. When primed ESCs (H9 cells) were mixed with PEG polymers, they were encapsulated and grew for an extended period, while maintaining their viability, self-renewal, and differentiation potential both in vitro and in vivo. Three-dimensional (3-D) self-assembling scaffold-grown cells displayed an upregulation of core pluripotency genes, *OCT4*, *NANOG*, and *SOX2*. In addition, the expression of primed markers decreased, while the expression of naïve markers substantially increased. Interestingly, the expression of mechanosensitive genes, *YAP* and *TAZ,* was also upregulated. YAP inhibition by Verteporfin abrogated the increased expression of *YAP*/*TAZ* as well as core and naïve pluripotent markers. Evidently, the 3-D culture conditions induced the upregulation of makers associated with a naïve state of pluripotency in the primed cells. Overall, our 3-D culture system supported the expansion of a homogenous population of ESCs and should be helpful in advancing their use for cell therapy and regenerative medicine.

**Keywords:** embryonic stem cells; three-dimensional; self-assembling scaffold; pluripotency; culture conditions; expansion; growth; niche

#### **1. Introduction**

Pluripotency is defined by the potential of a cell to differentiate into germline cells as well as the cells of all three germ layers [1,2]. During development, pluripotency ranges from the formation of the epiblast until gastrulation and lineage commitment, resulting in a population of cells representing a range of pluripotent states [3,4]. Pluripotency in vitro is determined by the developmental stage from which the cell line is derived and by culture conditions [5–8] with distinct naïve and primed states of pluripotency corresponding to in vivo pre-implantation and post-implantation epiblast cells, respectively [4].

Embryonic stem cells (ESCs) display unlimited self-renewal and differentiation potential in vitro [9], making them an ideal source for the development of cell therapies and regenerative medicine applications. However, the clinical use of ESCs requires a high quality and quantity of cells, which is limited by currently used culturing techniques [10]. ESCs are typically grown as a monolayer in two-dimensional (2-D) plastic culture plates coated with extracellular matrix (ECM) components (such as gelatin from porcine skin [11], matrigel [12], laminin [13], fibronectin [14], vitronectin [15], or collagen [16]) or a mouse embryonic fibroblast (MEF) feeder layer to aid in attachment [17]. Monolayer culture also necessitates routine passaging and the removal of spontaneously differentiated colonies to maintain the self-renewal and potency of cells [18], which pose a major impediment to the large-scale expansion of cells. Generally, 2-D culturing methods often lead to heterogeneous cell populations as well as batch-to-batch variation.

Moreover, 2-D culture conditions lack the intricacies necessary to mimic the ESC niche, dynamic, and specialized three-dimensional (3-D) microenvironments, which are critical for regulating cell fate in vivo. Furthermore, native 3-D niches allow for complex spatial interactions between cells ECM components as well as gradients of nutrients, oxygen, and metabolic waste [19–24]. The microenvironment is important for the self-renewal of ESCs, since cell fate and function is affected by the composition and organization of the ECM [25–27] as well as mechanical forces generated between cells and attachment substrates [25,28–31]. Furthermore, mechanical forces generated by the expansion of the blastocoel have been shown to play an important role in blastocyst lineage formation, stimulating the generation of pluripotent cells [32]. These early morphogenic events in the mammalian embryo indicate a significant interaction between mechanical forces, cell polarity, and the regulation of gene expression in cell fate determination [33]. We hypothesized that 3-D culture would better mimic the in vivo microenvironment, promoting the proliferation and maintenance of human ESCs.

We have previously demonstrated that 3-D self-assembling scaffolds composed of thiolated dextran and 4-arm polyethylene glycol (PEG) functionalized with acrylate groups (Dex-SH/PEG-4-Acr) supported the growth and maintenance of naïve mouse ESCs [34]. However, these 3-D culture conditions failed to support the growth of human primed ESCs. Unlike mouse ESCs, human cells display poor viability and clonogenicity following single cell dissociation [9,10].

In this study, we describe 3-D hydrogel scaffolds that support the long-term growth and maintenance of human primed ESCs (H9). Chemically cross-linked hydrogels were formed by a Michael-type addition reaction by combining two 8-arm PEG polymers functionalized with either thiol (PEG-8-SH) or acrylate end groups (PEG-8-Acr). The PEG-8-SH/PEG-8-Acr scaffold provided a microenvironment that maintained self-renewal and pluripotency for an extended time period. Interestingly, H9 cells displayed an upregulation of core pluripotency markers during 3-D culture. However, the expression of core markers reverted to normal levels when 3-D grown cells were subcultured under 2-D culture conditions. Interestingly, 3-D cultured H9 cells also exhibited a significantly higher expression of naïve pluripotent markers when compared to human naïve ESCs (Elf1) cultured under 2-D conditions. Our results suggest the importance of the 3-D scaffold microenvironment in maintaining the stemness of ESCs.

#### **2. Materials and Methods**

#### *2.1. Maintenance of Human ESCs in 2-D culture*

H9 cells, derived from a human blastocyst [35], obtained from WiCell (Madison, WI, USA) were maintained in the culture medium containing Knockout/F12 Dulbecco's Modified Eagle Media (DMEM; Life Technologies, Carlsbad, CA, USA) with 20% KnockOut serum replacement (Life Technologies), 0.1 mM 2-mercaptoehtanol (Life Technologies), 1% GlutaMax (Life Technologies), and 1% non-essential amino acid solution (Life Technologies), supplemented with 20 ng/mL of basic fibroblast growth factor (FGF2, Prospec, Ness Ziona, Israel) and 10 μM ROCK inhibitor, Y-27632 (Cayman Chemical, Ann Arbor, MI, USA) and subcultured by manual passaging. For dissociation into single cells, H9 were treated with Accutase (Thermo Fisher Scientific, Waltham, MA, USA) for 5 min, and then cells were centrifuged at 200 g for 5 min. The supernatant was aspirated, and the cell pellet was resuspended in H9 culture media for plating.

Elf1 cells, isolated from a cryopreserved 8-cell human embryo [36], were maintained in the culture medium containing KnockOut/F12 DMEM with GlutaMax (Life Technologies) with 20% KnockOut serum replacement (Life Technologies), 1 mM sodium pyruvate (Life Technologies), 0.1mM

2-mercaptoehtanol (Life Technologies), 1% non-essential amino acids (Life Technologies), and 0.2% penicillin–streptomycin solution (Life Technologies), supplemented with 12 ng/mL of FGF2 (Prospec), 1.5 μM CHIR99021 (Caymen Chemical), 0.4 μM PD03296501 (Caymen Chemical), and 0.01 μg/mL human LIF (Prospec) and cultured according to the published protocol [36].

#### *2.2. Selection and Composition of Sca*ff*olds for 3-D Culture of Human ESCs*

Several polymers end functionalized with thiol and acrylate end groups were tested for their ability to self-assemble and form hydrogels scaffolds via a thiol–Michael addition reaction, allowing for the formation of covalent bonds by the addition of a nucleophile to a nucleophile acceptor containing an α,β-unsaturated carbonyl compound [37]. Then, the scaffolds were tested in various compositions to determine the optimal concentration, molar ratio, and degree of modification that would best support the growth of human ESCs. Our preliminary results showed that scaffolds made using 8-arm PEG polymers yielded optimal growth. These scaffolds were prepared using 8-arm PEG-thiol (PEG-8-SH, 20 kDa) and 8-arm PEG-acrylate (PEG-8-Acr, 20 kDa) purchased from JenKem Technology USA (Plano, TX, USA). Functionalized PEG polymers were stored at −20 ◦C and protected from light.

The preparation of scaffolds for the 3-D culture of ESCs is depicted in Figure 1. Briefly, PEG-8-SH and PEG-8-Acr polymers were dissolved at a concentration of 2.5 *w*/*v* % (dry weight of polymer per volume of culture medium), combined at a 1:1 molar ratio and mixed with cells. Then, the resulting mixture was transferred to a 1 cc syringe mold for polymerization. After self-assembly, scaffolds were placed in a 24-well culture plate (Fisher Scientific, Pittsburgh, PA, USA), supplemented with culture medium, and maintained in a 5% CO2 incubator at 37 ◦C. The medium was changed daily or as needed. Cell growth in the scaffolds was monitored by phase-contrast microscopy.

**Figure 1.** Schematic of self-assembling scaffolds. (**A**) Self-assembly of functionalized polymers, 8-arm polyethylene glycol functionalized with thiol (PEG-8-SH) and acrylate (PEG-8-Acr) via a thiol–Michael addition reaction. (**B**) The encapsulation of H9 cells human embryonic stem cells (ESCs), was achieved upon mixing with the self-assembling polymers in a syringe mold. Following polymerization, the scaffolds were then incubated in culture plates containing medium.

#### *2.3. Cell Proliferation and Viability Assays*

The growth rate of cells grown under 2-D and 3-D culture conditions were analyzed at various time intervals using a proliferation assay. Briefly, triplicate samples were treated with 5 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reagent (Sigma, St. Louis, MO, USA), protected from light, and incubated at 37 ◦C for 4 h to obtain insoluble formazan, which was then solubilized using 15:1 isopropanol/hydrochloride. Then, the absorbance of the solubilized formazan was measured at 570 nm using an Epoch microplate reader (BioTek, Winooski, VT), and the background absorbance of the cells was subtracted from all measured values. The viability of encapsulated cells was determined by direct microscopic counts and trypan blue exclusion assay. Briefly, cells were counted using a hemocytometer and cells stained blue were considered non-viable.

#### *2.4. Di*ff*erentiation of Human ESCs*

Germ layer differentiation was achieved by the spontaneous formation of embryoid bodies (EBs). ESCs were allowed to spontaneously aggregate for 3 days in non-adherent flat-bottomed 96-well plates in their respective ESC culture medium containing growth factors. Then, the resultant EBs were transferred to 0.1% gelatin-coated wells for adherent growth and grown in high-glucose DMEM supplemented with 10% fetal bovine serum (FBS). Spontaneous differentiation into all three germ layers was assessed by germ layer marker expression by quantitative real time-polymerase chain reaction (qRT-PCR) and immunocytochemistry.

#### *2.5. Teratoma Assay*

For teratoma formation, ESCs were harvested following accutase treatment, washed and resuspended in PBS, and mixed with an equal volume of matrigel (BD Biosciences, San Jose, CA, USA). Cells (1 <sup>×</sup> 106) were subcutaneously injected (20 <sup>μ</sup>L) using a Hamilton syringe into 4-week-old male immune-compromised SCID (severe combined immunodeficient) Beige mice (Fox Chase SCID Beige, Charles River, Wilmington, MA, USA). Animals were monitored daily and humanely euthanized by CO2 overdose after teratoma formation at 10–12 weeks. Teratomas were explanted, and teratoma tissue was either fixed for histological analysis or flash frozen in liquid nitrogen for RNA isolation. Teratoma assays were performed in triplicate. All the procedures involving animals were reviewed and approved by the Institutional Animal Care and Use Committee of Oakland University (IACUC protocol number: 17031).

#### *2.6. Gene Expression Analysis*

Transcriptional analysis was performed by qRT-PCR. Briefly, cells, scaffolds, and teratoma tissue (100–250 mg) were harvested and total cellular mRNA was isolated following the manufacturer's instructions using the GeneJET RNA purification kit (Thermo Fisher Scientific) and RNeasy Midi kit (Qiagen, Germantown, MD, USA), respectively. cDNA was synthesized with the iScript kit (BioRad, Hercules, CA, USA). qRT-PCR was performed using SsoAdvanced SYBR Green Supermix (Bio-Rad) and the CFX90 Real-Time PCR system. The primers (IDT Technologies, Coralville, IA, USA) used in this study are in Table 1. All reactions were prepared in triplicate and normalized to reference genes, *HMBS*, *GAPDH*, and β*-ACTIN*.

#### *2.7. Immunocytochemical Analysis*

Protein expression was determined by immunocytochemical staining using selected antibodies. H9 cells grown under 2-D culture conditions on coverslips, H9 cells encapsulated in the self-assembling scaffolds, and harvested teratoma tissue were fixed in 4% paraformaldehyde for 10 min, 30 min, and overnight, respectively. Subsequently, teratoma tissue and scaffolds were embedded and frozen in optimal cutting temperature (O.C.T) compound and cryosectioned into 10 μm sections. For immunochemical analysis, fixed cells and cryosections were permeabilized with 0.5% Triton X-100 (Sigma) for 10 min and blocked with 2% bovine serum albumin (BSA; Sigma) for 1 h at room temperature. Next, samples were incubated with primary antibodies (1:100 dilution) overnight at 4 ◦C. Primary antibody-treated samples were washed three times with phosphate buffer saline (PBS), stained with secondary antibody at 1:200 dilutions for 2 h at 37 ◦C, and counterstained with 1 mg/mL 4 ,6-diamidino-2-phenylindole (DAPI; Life Technologies). The stained samples were visualized by using confocal microscopy. The antibodies used are listed in Table 2.


**Table 1.** List of human primer sequences used in qRT-PCR.


**Table 2.** List of primary and secondary antibodies used in immunocytochemical staining.

#### *2.8. E*ff*ect of YAP Inhibitor on Human ESCs Grown in 3-D Self-Assembling Sca*ff*olds*

For analysis of yes-associated protein (YAP) signaling in 3-D culture, human ESCs were grown in self-assembling scaffolds for 7 days, and then treated with 2 μM YAP inhibitor (YAPi), Verteporfin (VP, R&D Systems, Minneapolis, MN), for an additional 7 days. Cell growth in the scaffolds was monitored by light microscopy. ESCs grown in 3-D self-assembling scaffolds were harvested after 14 days of culture with and without YAPi treatment as a control. Cells were harvested for RNA and immunocytochemical analysis.

#### *2.9. Statistical analysis*

Data are presented as mean ± standard error of the mean (SE). One-way ANOVA analysis was performed and analyzed for unequal variances using post hoc tests for multiple comparisons. Results with a p-value less than 0.05 were considered to be significant (\* *p* < 0.05 and \*\* *p* < 0.01). All analyses were performed using SPSS version 26 (SPSS Inc., Chicago, IL., USA).

#### **3. Results**

#### *3.1. Growth and Characterization of H9 Cells Grown under 3-D Culture Conditions*

H9 cells encapsulated in self-assembling scaffolds composed of PEG-8-SH and PEG-8-Acr polymers grew for extended periods without requiring routine passaging or manipulation. The optimal growth of ESCs was achieved by using a concentration of 2.5 *w*/*v* % (dry weight of polymer per volume of culture medium) at a 1:1 molar ratio of PEG-8-SH and PEG-8-Acr. The results depicted in Figure 2 show steady cell growth up to 21 days as observed by light microscopy (Figure 2A–D). Cells grew clonally in a time-dependent manner. Visual observations were consistent with the results obtained by MTT assay. A significant and continuous increase in the proliferation of cells was observed during day 7 to 21. However, cells grew more rapidly between day 14 and 21 than between day 1 and 14 (Figure 2E), suggesting that the growth of ESCs required a longer acclimation period in 3-D self-assembling scaffolds compared to 2-D culture conditions. The viability of 3-D cultured ESCs was further validated by direct cell counts, as depicted in Figure 2F. The results showed that while cell proliferation significantly and consistently increased from day 1 to day 21, the number of dead cells remained low.

To confirm whether H9 cells remained pluripotent during 3-D culture, the transcriptional and translational analysis of selected ESC-specific markers was performed using qRT-PCR and immunocytochemical analysis, respectively. The results indicated a successive increase in the expression of *OCT4*, *NANOG*, and *SOX2* throughout the duration of the 3-D culture, which increased 2.85, 2.23, and 3.69-fold, respectively in ESCs grown in 3-D scaffolds for 21 days as compared to cells grown under 2-D culture conditions (Figure 2G). The protein expression of these markers was also increased in cells cultured in the 3-D scaffolds (Figure 2H), which was consistent with the transcriptional upregulation.

**Figure 2.** Growth of pluripotent human ESCs in 3-D self-assembling scaffolds. (**A**–**D**) Clonal growth of ESCs (H9 cells) encapsulated in PEG-8-SH/PEG-8-Acr scaffolds and incubated in culture medium was observed by light microscopy at 0, 7, 14, and 21 days. (**E**) Quantitative determination of cell proliferation by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay using microplate reader. Results were expressed as the absorbance ± standard error (SE) with a significant increase in cell number. (**F**) Growth of ESCs encapsulated in 3-D scaffolds was assayed by direct counts using a hemocytometer, and cell viability was determined by trypan blue exclusion assay at various time intervals. Data presented as cell number (×10<sup>6</sup> cells/mL) <sup>±</sup> SE. (**G**) Expression of selected pluripotency markers, *OCT4*, *NANOG*, and *SOX2*, in ESCs cultured in self-assembling scaffolds for 0, 7, 14, and 21 days as determined by qRT-PCR. The expression of genes at day 0 was set to 1 and results were expressed as the fold expression ± SE normalized to reference genes *HMBS*, *GAPDH,* and β*-ACTIN* (\* *p* < 0.05 and \*\* *p* < 0.01). (**H**) Confocal images (20X) of 2-D and 3-D grown ESCs displaying the expression of pluripotent proteins, OCT4, NANOG, and SOX2. All scale bars represent 100 μm.

#### *3.2. Maintenance of Pluripotency in H9 Cells Grown under 3-D Culture Conditions*

To investigate if the 3-D grown H9 cells maintained their pluripotency after 21 days in 3-D culture, cells were subcultured under 2-D culture conditions and analyzed for cell morphology and the expression of pluripotent markers. The results showed that there were no morphological differences between cells that were passaged from 3-D to 2-D culture conditions and the initial ESCs used for encapsulation in the 3-D self-assembling scaffolds (Figure 3A–C). When H9 cells grown in 3-D self-assembling scaffolds for 21 days were subsequently subcultured under 2-D culture conditions, the expression of core pluripotent markers, *OCT4* and *NANOG*, was statistically similar to that of initial 2-D grown cells, while the expression of *SOX2* (1.55 fold) was slightly upregulated (Figure 3D). Apparently, repeated the subculturing of 3-D grown cells in 2-D culture conditions reverted the expression of these markers back to the normal level of expression. Despite the high expression of key

pluripotent markers, it was prudent to further investigate the 3-D grown cells for the maintenance of their differentiation potential both in in vitro and in vivo.

**Figure 3.** Self-renewal and pluripotency of ESCs were maintained following extended culture in 3-D scaffolds. (**A**–**C**) Cell morphology of ESCs (H9 cells) grown in 2-D cultures prior to encapsulation, in self-assembling scaffolds for 3-D culture for 21 days, and then subsequently subcultured back to 2-D culture conditions, respectively, as determined by light microscopy. All scale bars represent 100 μm. (**D**) Comparison of expression of *OCT4*, *NANOG*, and *SOX2* in ESCs grown in 2-D conditions and first in 3-D self-assembling scaffolds, and then passaged in 2-D culture conditions as determined by qRT-PCR. Results were expressed as the fold expression ± SE normalized to reference genes *HMBS*, *GAPDH,* and β*-ACTIN* (\* *p* < 0.05 and \*\* *p* < 0.01).

#### *3.3. Di*ff*erentiation of H9 Cells Grown in 3-D Self-Assembling Sca*ff*olds*

The maintenance of the pluripotency of 3-D grown H9 cells was further investigated by the induction of differentiation into three germ layers following EB formation. The results presented in Figure 4 show that EBs from 3-D grown cells spontaneously differentiated into endoderm, mesoderm, and ectoderm germ layers, as evident by the protein expression of specific markers, GATA4, BRACHYURY, and TUJ1, respectively (Figure 4A). Transcriptional analysis also confirmed that differentiated derivatives of 3-D grown cells expressed markers of all three germ layers, including *SOX7* and *GATA6, BRACHYURY* and *MIXL1*, and *PAX6* and *NCAM,* representing the endoderm, mesoderm, and ectoderm, respectively (Figure 4B).

Next, the pluripotency of 3-D grown H9 cells was validated in vivo by an analysis of teratoma formation. The results showed that the 3-D grown cells injected into SCID Beige mice formed teratomas (Figure 5). An analysis of teratoma tissue showed the expression of GATA4, BRACHYURY, and TUJ1 proteins, indicating that teratoma tissue had cells representing the endoderm, mesoderm, and ectoderm, respectively (Figure 5C). The transcriptional analysis of teratoma tissues also showed the expression of genes of all three germ layers: the endoderm (*SOX7*, and *SOX17*), mesoderm (*BRACHYURY*, and *MIXL1*), and ectoderm (*PAX6*, and *TUJ1*) (Figure 5D). Taken together, these results indicated that H9 cells grown in 3-D self-assembling scaffolds maintained their pluripotency and differentiation potential.

**Figure 4.** Differentiation potential of human ESCs grown in 3-D self-assembling scaffolds. Embryoid bodies (EBs) derived from 3-D grown ESCs (H9 cells) spontaneously differentiated into three germ layers and expressed specific proteins analyzed by immunocytochemistry. (**A**) Differentiated derivatives of 3-D grown ESCs expressed GATA4, BRACHYURY, and TUJ1 proteins representing the endoderm, mesoderm, and ectoderm, respectively, as shown by confocal images (20X). All scale bars represent 100 μm. (**B**) Differentiated derivatives of 3-D grown ESCs expressed germ layer-specific genes *SOX7* and *GATA6* (endoderm), *BRACHYURY* and *MIXL1* (mesoderm), and *PAX6* and *NCAM* (ectoderm) as determined by qRT-PCR. Results are expressed as the fold expression ± SE normalized to reference genes *HMBS*, *GAPDH,* and β*-ACTIN* (\* *p* < 0.05 and \*\* *p* < 0.01).

**Figure 5.** Teratoma formation by 3-D grown human ESCs in severe combined immunodeficient (SCID) Beige mice. (**A**) Tumor growth was observed in all mice (*n* = 3) injected with 3-D grown ESCs (H9 cells). (**B**) Explanted tumor at 4 weeks showed encapsulated, lobular, and well-circumscribed gross morphology consistent with teratoma growth. (**C**) Expression of GATA4, BRACHYURY, and TUJ1 proteins representing the endoderm, mesoderm, and ectoderm, respectively in excised teratomas, as shown by confocal images (20X). All scale bars represent 100 μm. (**D**) Expression of germ layer-specific genes, *SOX7* and *GATA6* (endoderm), *BRACHYURY* and *MIXL1* (mesoderm), and *PAX6* and *NCAM* (ectoderm) in excised teratomas, as determined by qRT-PCR. Results are expressed as the fold expression ± SE normalized to reference genes *HMBS*, *GAPDH,* and β*-ACTIN* (\* *p* < 0.05 and \*\* *p* < 0.01).

#### *3.4. Expression of Naïve Pluripotent Markers in H9 Cells Grown under 3-D Culture Conditions and the E*ff*ect of YAP Inhibition*

Since a significant upregulation of core pluripotent markers, OCT4, NANOG, and SOX2, was observed in ESCs cultured in the 3-D scaffolds, we also assessed the expression of both primed and naïve pluripotent markers. A comparative transcriptional analysis of H9 cells grown in 3-D scaffolds and 2-D cultured H9 and Elf1 cells (representing primed and naïve ESC lines, respectively) is depicted in Figure 6. Expected primed pluripotent markers (*FOXA2*, *ZIC2*, *SALL2*, and *SOX11*) were expressed at higher levels in H9 than Elf1 cells cultured under 2-D conditions. Interestingly, the expression of primed pluripotent markers was significantly decreased in H9 cells grown in 3-D scaffolds to levels comparable to Elf1 cells grown under 2-D conditions. More strikingly, 3-D grown H9 cells expressed significantly higher levels of naïve markers (*KLF2*, *ESRRB*, *DNMT3L*, *KLF17*, *STAT3*, *DPPA3*, *TBX3*, *PRDM14*, *KLF5*, *ZFP42*, *TFCP2L1*, *FGF4*, and *GDF3*) in comparison to 2-D cultured Elf1 cells, suggesting that the 3-D scaffold microenvironment modulated gene expression.

**Figure 6.** Effect of 3-D culture on the expression of naïve pluripotent markers in primed ESCs. (**A**) Expression of select core (*OCT4*, *NANOG*, and *SOX2*), primed (*FOXA2*, *ZIC2*, *SALL2*, and *SOX11*), and naïve (*KLF2*, *ESRRB*, *DNMT3L*, *KLF17*, *STAT3*, *DPPA3*, *TBX3*, *PRDM14*, *KLF5*, *ZFP42*, *TFCP2L1*, *FGF4*, and *GDF3*) pluripotent markers in ESCs cultured in 3-D scaffolds for 21 days and 2-D grown primed ESCs (H9 cells) and naïve ESCs (Elf1 cells, set to control) was analyzed by qRT-PCR. Results were expressed as the fold expression ± SE normalized to reference genes *HMBS*, *GAPDH*, and β*-ACTIN* (\* *p* < 0.05 and \*\* *p* < 0.01).

Many reports have stated that 3-D scaffolds induce differential gene expression due to mechanical and biological stimuli [10,38,39]. Since the overexpression of YAP has been shown to induce the naïve state of pluripotency in primed ESCs [40], we investigated the effect of YAP in 3-D cultured H9 cells using VP, which is a YAP inhibitor (YAPi). Our results in Figure 7A showed that mechanosensitive genes, *YAP* and *TAZ,* were upregulated in H9 cells grown in 3-D scaffolds in comparison to 2-D cultured cells. In contrast, *YAP* and *TAZ* expression significantly decreased when 3-D grown H9 cells were subjected to YAPi. A similar trend was observed at a translational level, where it appears that the immunofluorescence signal for YAP in 3-D grown H9 cells was brighter than in 2-D cultured cells. Moreover, an increased signal for YAP was observed in the cytoplasm of 3-D grown cells treated with YAPi (Figure 7B).

The incubation of YAP also affected the clonal growth of H9 cells encapsulated in 3-D scaffolds, with YAPi-treated H9 cells exhibiting a significantly smaller colony size in comparison to untreated cells (Figure 7C). Interestingly, YAPi treatment also abrogated the upregulation of core (*OCT4*, *NANOG*, and *SOX2*), and naïve (*ESRRB*, *KLF4*, *DNMT3L*, *KLF17*, *DPPA3*, *KLF5*, *ZFP42*, *TFCP2L1*, and *FGF4*) pluripotent markers in 3-D cultured cells (Figure 7D). In addition, a concurrent decrease in the expression of core pluripotent proteins, OCT4, NANOG, and SOX2 was observed when compared with the untreated 3-D grown cells (Figure 7E).

#### *3.5. Mechanism of Regulation of Pluripotent Genes in H9 Cells Grown under 3-D Culture Conditions*

To determine the basis for the upregulation of pluripotent markers observed in H9 cells grown in 3-D self-assembling scaffolds, we screened multiple signaling pathways with YAP-associated mechanotransduction using transcriptional analysis (Figure 8). The results depicted in Figure 8A show that 3-D cultured cells expressed higher levels of genes encoding integrin subunits, *ITGA5*, *ITGA6*, *ITGAV*, *ITGB1*, and *ITGB3* as well as G-coupled protein receptors (GCPRs), *LPAR1*, *LPAR2*, *S1PR1*, and *S1PR3.* In addition, these cells displayed the upregulation of *RHOA* and *RAC1* (Rho signaling), *YAP*, *TAZ*, and *TEAD4* (Hippo signaling), *LIFR*, *GP130*, *SOCS3*, and *TBX3* (LIF signaling). However, the expression of *ITGA2, ROCK1, LATS1*, and *LATS2,* which is associated with integrin, Rho, and Hippo signaling, respectively, decreased significantly, while *ITGB5* remained unchanged in 3-D cultured cells.

**Figure 8.** Molecular mechanism of upregulation of pluripotent markers in human ESCs grown in 3-D scaffolds. (**A**) The effect of 3-D culture of ESCs (H9 cells) on the expression of markers associated with integrin (*ITGA2*, *ITGA5*, *ITGA6*, *ITGAV*, *ITGB1*, *ITGB3*, and *ITGB5*) and G-coupled protein receptors (GCPRs; *LPAR1*, *LPAR2*, *S1PR1*, and *S1PR3*), Hippo (*YAP*, *TAZ*, *TEAD4*, *LATS1*, and *LATS2*), Rho (*RHOA*, *RAC1*, and *ROCK1*), and LIF (*LIFR*, *GP130*, *SOCS3*, and *TBX3*) signaling pathways as determined by qRT-PCR. Results were expressed as the fold expression ± SE normalized to reference genes *HMBS*, *GAPDH*, and β*-ACTIN* (\* *p* < 0.05 and \*\* *p* < 0.01). (**B**) Proposed pathway involved in the induction of naïve pluripotency in primed human ESCs (H9 cells) encapsulated in 3-D self-assembling scaffolds.

These results further indicate that 3-D self-assembling scaffolds stimulated mechanosensitive signaling resulting in the upregulation of integrin receptors and GPCRs, thus promoting the activation of Rho signaling, which is associated with actin cytoskeleton remodeling; in turn, this led to the upregulation of mechanosensitive YAP/TAZ signaling. Activated YAP/TAZ act as transcription factors in the nucleus assisted by transcription co-factor, TEAD4, which binds to the DNA, to stimulate the expression of pluripotent genes. The simultaneous upregulation of LIF signaling also contributed to the upregulation of naïve pluripotent markers in H9 cells grown in 3-D self-assembling scaffolds, as proposed in Figure 8B.

#### **4. Discussion**

The expansion of human ESCs using traditional 2-D culture techniques is technically challenging and requires routine manipulation and passaging by dissecting colonies via enzymatic digestion or non-enzymatic methods [41–43]. These manipulations can result in poor viability, large batch-to-batch variation, and spontaneous differentiation. To address these problems, we hypothesized that 3-D culture may better mimic the in vivo environment from which ESCs are derived, which would improve both the long-term growth and maintenance of these cells.

Our study investigated the effect of the microenvironment in stemness by developing 3-D scaffolds made of two functionalized PEG polymers that self-assembled via a Michael addition reaction. When ESCs were included in the polymer mixtures, they were encapsulated upon self-assembly of the scaffolds. It has been previously reported that fully hydrated hydrogels mimic the 3-D native microenvironment, which allow nutrient diffusion and promote the growth of cells [44]. We have previously shown that soft 3-D scaffolds composed of Dex-SH and PEG-4-Acr self-assembling polymers were capable of supporting mouse ESC pluripotency for over 6 weeks [34]. Several other studies have also reported that mouse ESC self-renewal could be maintained on soft and low stiffness substrates in 2-D culture [45–47]. In addition, mechanically stiffer prefabricated scaffolds have been shown to promote the differentiation of mouse ESCs [48], while softer 3-D scaffolds supported the differentiation of human adult stem cells [49]. The culture of H9 cells in soft scaffolds did not support the viability of encapsulated cells, and growth was severely curtailed. This is consistent with previous studies, which showed that stiffer substrates promoted the maintenance of human ESCs [22,50,51]. Taken together, this led us to develop stiffer scaffolds made of functionalized PEG polymers.

To optimize scaffold polymerization as well as the encapsulation and growth of H9 cells, we tested several self-assembling polymers at various polymer concentrations in our preliminary studies (unpublished data). These studies showed that scaffolds made of PEG-8-SH and PEG-8-Acr prepared at 2.5% *w*/*v* improved the clonal growth of H9 cells in comparison to scaffolds with lower cross-linking densities and higher swelling ratios (i.e., Dex-SH/PEG-4-Acr and PEG-4-SH/PEG-4-Acr). These results were in line with other the reported studies performed using scaffolds made of multi-arm PEG functionalized with vinyl sulfone (VS) [52]. In these studies, H9 cells grew upon encapsulation in scaffolds made of VS functionalized 4-arm and 8-arm but not 3-arm PEG hydrogels [52]. However, this study was performed using clumps of H9 cells and not single cells for encapsulation. Whereas, in our study, PEG-8-SH/PEG-8-Acr scaffolds supported cell viability, allowing for even dispersal and clonal growth of H9 cells encapsulated as single cells. This has important implications because the growth of single cell inoculations and the generation of homologous populations of pluripotent cells is necessary for cell-based therapeutics [53]. Furthermore, the maintenance and growth of H9 cells was achieved using PEG-8-SH/PEG-8-Acr self-assembling scaffolds for long-term 3-D culture without passaging or manipulation. The pluripotency of the 3-D grown H9 cells was further demonstrated by their ability to differentiate into three germ layers and teratoma formation in vitro and in vivo, respectively. Additionally, core pluripotent markers, *OCT4*, *NANOG*, and *SOX2*, were upregulated during growth in the self-assembling scaffolds, showing significantly higher expression on day 21 of 3-D culture as compared to cells grown in 2-D culture. When 3-D grown H9 cells were passaged back to 2-D culture conditions, they exhibited undifferentiated morphology, and the expression of pluripotent

markers decreased to levels similar to the 2-D cultured cells, suggesting that the encapsulated cells cultured under 3-D conditions were not altered.

Changes in the expression of core and naïve pluripotent markers in H9 cells cultured in 3-D self-assembling scaffolds can be attributed to multiple factors, including matrix dimensionality, stiffness, and/or bioinductive signaling [10]. While the incorporation of natural biomaterials has been shown to increase biological signaling, synthetic biomaterials lack biological activities minimizing batch-to-batch variation, but still allow for biophysical modifications, including pore size and mechanical stiffness [54]. Scaffolds composed of natural polymers including hyaluronic acid [55], chitosan, and alginate [56] have been shown to support human ESC self-renewal without a significant change in pluripotent marker expression. Whereas thermoresponsive synthetic hydrogels composed of PEG functionalized with poly-N-isopropylacrylamide allowed for the continuous 3-D culture of cells for 60 passages but only yielded cells 95% positive for *OCT4* [57]. In contrast, we observed the upregulation of not only *OCT4* but also *NANOG* and *SOX2* during the maintenance of human ESCs in 3-D scaffolds.

Previous studies with mouse ESCs reported that the differential upregulation of pluripotent markers in 3-D culture was dependent on scaffold composition and stiffness [34,58]. A recent study using single cell inoculation and the expansion of human ESCs in large-scale bioreactors resulted in the maintenance of *OCT4* levels but the upregulation of *SOX2* in 3-D grown ESC aggregates [59]. In another study, a 3-D culture of H9 clumps in VS functionalized PEG scaffolds resulted in upregulated gene expression of *SOX2* and *KLF4* but not *OCT4* and *NANOG* when compared to 2-D cultured cells [52]. Interestingly, we observed that primed ESCs, H9 cells, encapsulated in the 3-D self-assembling scaffolds exhibited a decreased expression of primed pluripotency markers, *FOXA2*, *ZIC2*, *SALL2*, and *SOX11*, which is characteristic of post-implantation epiblast cells [60,61], to levels comparable to naïve Elf1 cells. Furthermore, a 3-D culture of H9 cells resulted in an increase in the expression of naïve pluripotent markers, (*KLF2*, *ESRRB*, *DNMT3L*, *KLF17*, *STAT3*, *DPPA3*, *TBX3*, *PRDM14*, *KLF5*, *ZFP42*, *TFCP2L1*, *FGF4*, and *GDF3*), which was associated with naïve pre-implantation epiblast cells [7,60–66]. Recent studies have shown that traditionally derived primed human ESC lines can be reprogrammed to naïve pluripotency using the ectopic expression of select genes and specific media conditions containing inhibitor cocktails [67,68]. In one study, a transcriptomic comparison of primed and reprogrammed naïve human demonstrated the differential expression of certain genes such as *KLF2*, *KLF4*, *GDF3*, *SOCS3*, *STAT3,* and *TBX3* expressed at higher levels in naïve than primed cells [63]. However, the expression of core pluripotent markers *OCT4* and *NANOG* remained unchanged, while *SOX2* levels decreased upon reversion to naïve pluripotency [63]. Here, we observed an increase in core and naïve pluripotent markers during 3-D culture, which was presumably influenced by the 3-D scaffold microenvironment.

It has been well established that the transduction of biophysical signals, including substrate stiffness, cell–cell interactions, and mechanical forces can influence human ESC fate and self-renewal in 2-D culture conditions [39]. The interplay between external and internal mechanical stresses of cells and their ECM play important roles in tensional homeostasis of tissues in vivo [69]. However, the effect of 3-D culture on the regulation of the pluripotency of ESCs has not been fully explored.

Physical interactions with cells or the ECM can be transduced into biological signals and influence actin dynamics via mechanosensitive receptors, such as integrin receptors and GPCRs [70]. Evidence has shown that integrin heterodimers, α5β1, α6β1, and αvβ3, mediate interactions between ESCs and various ECM substrates [13,71–73]; they also play an important role in the maintenance of pluripotency [74], and matrix stiffness regulates integrin binding [75]. Whereas GPCRs, including lysophosphatidic acid receptors (LPARs) and sphingosine-1-phosphate receptors (S1PRs), play a role in the YAP signaling axis. In response to LPA and S1P ligand binding, the dephosphorylation of YAP and TAZ allows them to enter the nucleus to activate transcription [76]. Treatment with exogenous LPA has been shown to aid in the reversion of primed pluripotency to naïve pluripotency in ESCs due to the role that both YAP and LPA serve in suppressing differentiation associated with GSK3 inhibition [40]. However, the transduction of mechanical signals has also been shown to activate G-protein signaling

independent of ligand binding [77]. We observed the upregulation of numerous integrin subunits as well as LPARs and S1PRs in 3-D grown H9 cells. This suggests that the 3-D scaffolds promoted mechanical signaling pathways in H9 cells; however, future studies are required to determine the mechanism by which these receptors are activated.

Mechanical signals such as substrate stiffness have been shown to lead to the activation of Rho signaling, which regulates actin cytoskeleton organization, leading to increased F-actin and nuclear localization of YAP and TAZ [78]. YAP is a mechanosensitive transcription factor in the Hippo signaling pathway, which plays a crucial role in cancer, regeneration, and the regulation of organ size [79]. In addition, Hippo signaling regulates the first cell fate specification to the trophoectoderm and the inner cell mass (ICM) of early blastocysts via mechanically sensitive pathways [80]. YAP is mostly retained in the cytoplasm in the ICM of early blastocysts, but becomes active during epiblast specification with a strong nuclear signal [64]. YAP and its transcription cofactor, TAZ, act as major downstream effectors in the Rho signaling pathway and have been shown to control the pluripotent state, allowing for the long-term survival and expansion of human ESCs in vitro [81].

Furthermore, cell culture on stiffer substrates has also been shown to increase the nuclear function of YAP/TAZ in human ESCs [82]. TAZ itself is required for the maintenance of self-renewal marker expression in ESCs, and the loss of TAZ leads to the inhibition of transforming growth factor beta (TGFβ) signaling and differentiation into a neuroectoderm lineage [83]. Likewise, a knockdown of YAP results in a loss of pluripotency in mouse ESCs, whereas the ectopic expression of YAP prevents differentiation [84].

In the nucleus, YAP mediates transcriptional enhanced associate domain (TEAD) transcription, and YAP/TAZ complexes associate with TEAD to regulate pluripotency by activating *OCT4* expression [85]. Pluripotency is determined by an autoregulatory core transcriptional circuitry comprised of *OCT4*, *NANOG*, and *SOX2,* which inhibits lineage specification genes [86]. The phosphorylation of YAP by LATS kinases prevents interaction with TEAD and results in cytoplasmic retention and the inactivation of YAP [87]. In contrast, YAP overexpression has been shown to promote the reversion of primed pluripotency into naïve pluripotency [40].

When 3-D grown H9 cells were cultured with a YAPi (VP), the upregulation of core and naïve pluripotent markers was abrogated. VP acts to inhibit the interaction of YAP and TEAD, disrupting downstream transcriptional activation, and sequestering YAP in the cytoplasm for inactivation [88]. Therefore, we postulate that mechanical forces in 3-D scaffolds stimulated the upregulation of mechanosensitive receptors, including integrins and GPCRs, leading to enhanced Rho signaling and higher levels of YAP/TAZ. In addition, the mechanical forces generated by the expansion of cells may also have contributed to the inhibition of phosphorylated LATS kinases [87], which in turn allowed YAP/TAZ to enter the nucleus, activating the transcription of pluripotent genes. This was concurrent with the observed increase in LIF signaling receptors and downstream naïve pluripotent target genes including *KLF4*, *KLF5,* and *TFCP2L1* [89]. While LIF/STAT3 signaling fails to maintain the self-renewal of primed human ESCs [90], naïve ESCs are dependent on LIF signaling [63]. Although these results are interesting and suggest a role of mechanical signaling in the regulation of cell fate in vitro, further analysis is needed to confirm the mechanism of upregulation of pluripotency markers and the genetic stability of ESCs grown in 3-D culture conditions.

Overall, the 3-D scaffolds made of PEG-8-SH/PEG-8-Acr support the clonal growth of primed ESCs as well as the enhanced expression of both core and naïve pluripotency markers, suggesting that the scaffold provided a permissive microenvironment for the induction of a naïve-like state of pluripotency. Our 3-D culture method is robust, simple, and less labor-intensive for the long-term amplification of homogenous populations of ESCs, which could promote their use in basic science and therapeutic applications.

**Author Contributions:** Conceptualization, G.R.C.; methodology, C.M.; validation, C.M., and C.B.; data curation, C.M., and C.B.; formal analysis, G.R.C.; writing—original draft preparation, C.M.; writing—review and editing, C.B, and G.R.C.; supervision, G.R.C.

**Funding:** This research received no external funding.

**Acknowledgments:** The study was supported by the OU-WB Institute for Stem Cell and Regenerative Medicine (ISCRM), Oakland University, and Michigan Head and Spine Institute. C. McKee received the Provost Graduate Research Award from Oakland University for this project. We are thankful for C. Govind and S. Dinda for reviewing the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **Generation of Di**ff**erentiating and Long-Living Intestinal Organoids Reflecting the Cellular Diversity of Canine Intestine**

**Nina Kramer 1,\*, Barbara Pratscher 1,**‡**, Andre M. C. Meneses 1,2,**†**,**‡**, Waltraud Tschulenk 3, Ingrid Walter 3, Alexander Swoboda 1, Hedwig S. Kruitwagen 2, Kerstin Schneeberger 2, Louis C. Penning 2, Bart Spee 2, Matthias Kieslinger 1, Sabine Brandt <sup>4</sup> and Iwan A. Burgener <sup>1</sup>**


Received: 17 February 2020; Accepted: 26 March 2020; Published: 28 March 2020

**Abstract:** Functional intestinal disorders constitute major, potentially lethal health problems in humans. Consequently, research focuses on elucidating the underlying pathobiological mechanisms and establishing therapeutic strategies. In this context, intestinal organoids have emerged as a potent in vitro model as they faithfully recapitulate the structure and function of the intestinal segment they represent. Interestingly, human-like intestinal diseases also affect dogs, making canine intestinal organoids a promising tool for canine and comparative research. Therefore, we generated organoids from canine duodenum, jejunum and colon, and focused on simultaneous long-term expansion and cell differentiation to maximize applicability. Following their establishment, canine intestinal organoids were grown under various culture conditions and then analyzed with respect to cell viability/apoptosis and multi-lineage differentiation by transcription profiling, proliferation assay, cell staining, and transmission electron microscopy. Standard expansion medium supported long-term expansion of organoids irrespective of their origin, but inhibited cell differentiation. Conversely, transfer of organoids to differentiation medium promoted goblet cell and enteroendocrine cell development, but simultaneously induced apoptosis. Unimpeded stem cell renewal and concurrent differentiation was achieved by culturing organoids in the presence of tyrosine kinase ligands. Our findings unambiguously highlight the characteristic cellular diversity of canine duodenum, jejunum and colon as fundamental prerequisite for accurate in vitro modelling.

**Keywords:** intestinal organoids; canine intestine; differentiation; organoid culture

#### **1. Introduction**

Gastrointestinal (GI) disorders such as inflammatory bowel disease (IBD), infection-induced GI disorders and cancer have a major negative impact on human health and impose a high financial burden on healthcare systems. Infectious diarrhea is the second leading cause of death among young children [1] and GI cancers are the third most common cancer type worldwide, with one million newly diagnosed cases per year [2]. Potentially lethal GI diseases also affect livestock and companion animals, with enterotoxic bacteria and enteropathogenic viruses being frequently involved in disease onset and progression [3]. Although considerable efforts are made to establish new therapies for human and veterinary GI diseases, mortality rates remain high, because translation of biomedical research into clinical practice is hampered by the lack of epithelial models that reliably mimic the organ and recapitulate disease in patients [4–6].

Early attempts to bring the intestinal epithelium to in vitro culture were mostly short-lived, as exemplified by explanted biopsies and epithelial cells, which disintegrated after 72 h and two weeks, respectively [7–9]. To establish culture systems for long-lived homogenous cell populations tissue explants were either cultivated in collagen gels or on 3T3 feeder layers on an air-liquid interface [10,11]. However, growth factor-providing (myo)fibroblasts, which are prerequisites for prolonged cultivation in these systems, made them susceptible to inconsistencies from one experiment to another. A major step to overcome these limitations was the establishment of three-dimensional (3D) murine intestinal organoids derived from adult intestinal stem cells [12]. These organoids contained long-lived stem cells that differentiated into the main cell types of murine small intestine such as Paneth cells, enterocytes, enteroendocrine cells and goblet cells. Establishment of this innovative in vitro model was achieved by using a laminin-rich extracellular matrix (Matrigel) and growth media supplemented with epithelial growth factor (EGF), Noggin, and R-spondin as reviewed in detail by Date and Sato [13]. For the cultivation of human intestinal organoids, growth medium had to be supplemented with Wnt3a, gastrin, p38-MAPK inhibitor, nicotinamide and ALK4/5/7 inhibitor, thereby preventing the differentiation into goblet and enteroendocrine cells and retaining enterocytes in a premature state [14]. Withdrawing nicotinamide and p38-MAPK inhibitor induced de novo differentiation, thus shortening the lifespan of the intestinal stem cell pool and rendering this system into an endpoint assay. Further improvement of human 3D intestinal organoid models was achieved by microscaffolds that mimic the size and distance between crypts in transwell assays [15]. Differentiation was promoted by using a growth factor gradient system based on (i) intestinal stem cell-supporting expansion medium in lower wells, and (ii) addition of differentiation medium to upper wells, thereby inducing differentiation of migrating cells to goblet cells, enteroendocrine cells and enterocytes along the modeled crypt/villus axis. However, this technique is unsuitable for, e.g., high-throughput screenings because it is very labor-intensive, time-consuming and hampers down-stream analysis. An approach characterizing receptor tyrosine kinase signaling between the crypt base and its niche managed to omit nicotinamide and p38-MAPK inhibitor usage by a combination of two ligands, i.e., insulin-like growth factor 1 (IGF1) and basic fibroblast growth factor (FGF2). This combination sustains stem cell growth and allows simultaneous differentiation into enteroendocrine and goblet cells, mimicking corresponding natural epithelia more authentically with minimal effort.

In recent years, organoid technology has also been introduced into veterinary research, albeit to a lesser extent [16–22]. Animal organoid systems that are already available today mainly consist of stem and undifferentiated cells requiring tailor-made species-specific media to allow for differentiation [17]. Interestingly, dogs and their owners share similar environments, food and carcinogenic load, and develop similar GI diseases including GI cancer, infectious disease and IBD [23]. Consequently, canine patients represent an interesting natural model for human GI disorders, even more so since dogs show reduced genetic variation within the majority of breeds [24]. To exploit the full potential of canine intestinal organoids to mimic the corresponding organ accurately, a balance between self-renewal and differentiation of stem cells must be achieved to design a physiologically relevant system.

Herein, we report the establishment of a novel culture system for canine intestinal organoids that is based on recent findings by Fujii et al. [25] and reliably supports the sustained proliferation of duodenum-, jejunum- and colon-derived stem cells, while concomitantly allowing their differentiation into secretory lineage cells.

#### **2. Materials and Methods**

#### *2.1. Isolation and Cultivation of Canine Intestinal Organoids*

Duodenal, jejunal and colonic samples were obtained from three dogs (two males and one female) that were euthanized due to non-intestinal disease. In addition, duodenal biopsies were taken from one dog undergoing routine gastroduodenal endoscopy. Tissue sampling was approved by the institutional ethics committee in accordance with Good Scientific Practice guidelines and Austrian legislation. Based on the guidelines of the institutional ethics committee, use of tissue material collected during therapeutic excision or post mortem is included in the University's "owner's consent for treatment", which was signed by all patient owners. Intestinal crypts were isolated from tissue samples and biopsies according to established protocols [12]. In short, tissue was incubated with 5 mM EDTA (Sigma-Aldrich, St. Louis, MO, USA) in order to dissociate crypts for 30 min to one hour depending on bowel segment, followed by vigorous shaking until crypts were released. After two washing steps with PBS and Advanced Dulbecco's modified Eagle's medium/F12 (DMEM/F12, Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) 500 crypts were resuspended in 50 μL Matrigel (BD Biosciences, Franklin Lakes, NJ, USA) and seeded per well of a 24-well plate. Following Matrigel polymerization expansion medium was added. Canine intestinal growth media were prepared as indicated in Table 1. Penicillin/streptomycin, HEPES, Glutamax, B27 (with Vitamin A) and N2 supplement were obtained from Invitrogen, Thermo Fisher Scientific. N-acetylcysteine, gastrin and nicotinamide were purchased from Sigma-Aldrich. EGF was provided by Thermo Fisher Scientific. ALK5 kinase inhibitor (A83-01) was obtained from Tocris Bioscience, Bristol, UK. p38 MAPK inhibitor (SB202190) was purchased from Selleck Chemicals, Houston, TX, USA. Human hepatocyte growth factor (HGF), human Noggin, human IGF1 and human FGF2 were provided by PeproTech, Rocky Hill, NJ, USA. Cultrex R-spondin1 cells were obtained from Trevigen, Gaithersburg, MD, USA. Expansion medium was supplemented with 10 μM Rock inhibitor (Y-27632, Selleck Chemicals) for the first two days after isolation, then medium was changed to not-supplemented expansion medium. For isolations using refined medium, EGF and 10 μM Y-27632 were additionally added only for the first two days. Growth medium was changed every two to three days. For weekly passaging at 1:4 to 1:8 split ratios, organoids were harvested, mechanically disrupted using a flame-polished Pasteur pipette. Depending on the splitting ratio, the corresponding quantity of organoid fragments were then embedded in 50 μL fresh Matrigel, seeded per well of a 24-well plate and cultured with either expansion or refined growth medium. To induce differentiation, organoids cultivated in expansion medium were transferred to differentiation medium and analyzed after 4 days. For assessing derivation efficiency in expansion and refined medium, organoids were counted 10 days after initial isolation and projected area was calculated using ImageJ64 (NIH).


**Table 1.** List of media compositions used in this study. Basal medium consists of Advanced DMEM/F12, 2 mM GlutaMAX, 10 mM HEPES, 1x B27, 1 mM N-acetylcystein. Final concentrations are given.

<sup>a</sup> in refined medium 50 ng/mL EGF is only added for the first two days after isolation or medium adaption.

#### *2.2. Gene Expression Analysis*

Total RNA was purified from duodenal, jejunal and colonic organoids in expansion, differentiation and refined medium and from corresponding intestinal epithelium that was harvested during crypt isolation using ReliaPrep™ RNA Tissue Miniprep System (Promega, Madison, WI, USA) according to instructions of the manufacturer. Individual RNA concentrations were determined spectrophotometrically (Nanodrop One C, Thermo Fisher Scientific) and RNA integrity was checked using Agilent Tape Station 4200 (Agilent, Santa Clara, CA, USA). 500 ng RNA was reverse-transcribed to cDNA with oligo-dT and random hexamer primers according to recommendations of the manufacturer (GoScript™ Reverse Transcription System, Promega). Subsequently, qPCR reactions were carried out using GoTaq® qPCR Master Mix (Promega). Primer sequences are given in Table S1. Amplification conditions were as follows: 2 min of initial denaturation at 95 ◦C, 40 cycles of 15 s of denaturation at 95 ◦C, 60 s of annealing/extension at 60 ◦C and a read step, followed by 10 s of dissociation at 95 ◦C and a melting curve from 65 ◦C to 95 ◦C in 5 s per 0.5 ◦C increments. Data analysis was performed according to Pfaffl et al. [26] taking PCR efficiency into account. Relative quantification was achieved by normalization of values to the stably expressed canine housekeeping gene hypoxanthine phosphoribosyl transferase (HPRT). Heat maps were generated using Excel conditional formatting of log2 fold changes.

#### *2.3. Periodic Acid-Schi*ff *Reaction of Organoid and Tissue Sections*

For the generation of formalin-fixed paraffin-embedded (FFPE) samples, organoids were fixed using 2% *v*/*v* paraformaldehyde (PFA, Merck, Darmstadt, Germany) for 15 min at room temperature, cast in 1.5% *w*/*v* agarose and dehydrated prior to embedding in paraffin (Sigma-Aldrich). 2.5 μm FFPE sections of organoids derived from duodenum, jejunum and colon and corresponding tissue were stained using PAS-Reaction staining kit and counterstained with Haematoxylin acidic after Mayer according to manufacturer's instructions (Morphisto, Frankfurt am Main, Germany). Images were acquired with a DMi8 microscope and LASX software (Leica, Wetzlar, Germany).

#### *2.4. Transmission Electron Microscopy of Organoids*

Samples were fixed in 3% *v*/*v* buffered glutaraldehyde (pH 7.4, Merck, Darmstadt, Germany), pre-embedded in 1.5% *w*/*v* agarose (Invitrogen, Thermo Fisher Scientific), washed three times in 0.1 M phosphate buffer (Soerensen, pH 7.4) afterwards and post-fixed in 1% *v*/*v* osmium tetroxide (Electron Microscopy Sciences, Hatfield, USA) for 2 h at room temperature. Dehydration was performed in a series of graded ethanol solutions (70%, 80%, 96% and 100%), subsequently infiltrated with propylene oxide (Sigma-Aldrich), followed by increasing ratios of epoxy resin-propylene oxide and finally pure resin (Serva, Mannheim, Germany). After an additional change, the resin was polymerized at 60 ◦C for 48 h. Semi-thin sections were cut at 0.8 μm and stained with toluidine blue, ultra-thin sections were cut at 70 nm, mounted on copper grids (Gröpl, Tulln, Austria) and stained with uranyl acetate (Fluka Chemie AG, Buchs, CH) and lead citrate (Merck, Darmstadt, Germany). Transmission electron micrographs were made with an EM900 (Zeiss, Oberkochen, Germany).

#### *2.5. Proliferation Assay*

In order to assess cell proliferation, the Click-iT® EdU Imaging Kit (Invitrogen, Thermo Fisher Scientific) was used. Canine intestinal organoids cultivated in expansion and refined medium were incubated with 5-ethynyl-2'deoxyuridine (EdU) at a final concentration of 10 μM for one hour at 37 ◦C and were then fixed with 2% *v*/*v* PFA for 15 min at room temperature. Staining procedure was carried out according to manufacturer's instructions. DNA was counterstained using 4 ,6-Diamidine-2 -phenylindole dihydrochloride (DAPI, Sigma-Aldrich). Confocal images were taken using an LSM 880 (Zeiss). For 3D reconstruction of z-stack images, the Arivis 3D plugin of Zen 2.3 lite (Zeiss) was used.

#### *2.6. Viability and Apoptosis Assays*

Viability and apoptosis of organoids in the three media compositions were assessed using the RealTime-Glo MT Cell Viability Assay (Promega) and RealTime-Glo Annexin V Apoptosis Assay (Promega). Organoids were trypsinized for 2 min and seeded in 24-well plates with expansion or refined medium, both supplemented with Y-27632 inhibitor. After 2 days in culture, organoids were harvested, counted and 50 organoids were seeded in six wells of white 96-well plate for each condition and assay. Detection reagents were prepared according to manufacturer and added to the different culture media before overlaying the organoids with either expansion, differentiation or refined medium. Luminescence was measured 0 h, 24 h, 48 h and 72 h after adding the substrates using a GloMax plate reader (Promega). Six wells without organoids, but with medium, served as background controls.

#### *2.7. Statistical Evaluation*

Data were analyzed and plotted in Prism5 (GraphPad, San Diego, CA, USA). Using a t-test for independent samples (two tailed, unpaired) *p*-values ≤ 0.05 were considered statistically significant. qPCR data were analyzed by two-way ANOVA (factors are growth medium and individual dogs) and Tukey multiple comparison test using R (version 3.3.1). Statistical results are shown in Tables S2–S7. \*\*\* *p* < 0.001 \*\* *p* < 0.01 \* *p* < 0.05. *p* < 0.1.

#### **3. Results**

#### *3.1. Expansion Medium Does Not Support Expression of Secretory Lineage Di*ff*erentiation Markers*

Canine intestinal organoids derived from duodenal, jejunal and colonic tissue were generated and cultivated in expansion medium. Initial cell isolates representing the intestinal epithelium after separation from the underlying tissue, as well as low- (≤5) and high-passage (≥10) organoids derived therefrom, were quantitatively assessed for expression of selected cell markers using RT-qPCR.

The stem cell marker *LGR5* displayed similar expression levels in low- and high-passage organoids, except high-passage duodenal organoids, which exhibited significant up-regulation of *LGR5* in comparison to their low-passage and initial cell isolate counterparts (Figure 1A). The enterocyte marker villin 1 (*VIL1*) exhibited significantly reduced expression in duodenal, jejunal and colonic organoids in low- and high-passage organoids compared to the initial cell isolates. Enteroendocrine cell marker chromogranin A (*CHGA*) [27] and goblet cell marker mucin 2 (*MUC2*) [28,29] transcripts were barely detectable in organoids when kept in expansion medium throughout passages (Figure 1A).

#### *3.2. Canine Intestinal Organoids Grow E*ffi*ciently in Refined Medium*

To best reflect the in vivo situation and thus allow for downstream applications including drug and toxicity assessment, microbiome studies and modelling of infectious disease, intestinal organoids should harbor stem cells, undifferentiated transit-amplifying cells and differentiated cells. Three different media were compared to identify the ideal medium composition supporting such a scenario: (i) an expansion medium for the culture of canine organoids; (ii) a differentiation medium characterized by omission of nicotinamide and p38-MAPK inhibitor [14]; and (iii) a novel refined medium based on canine expansion medium without nicotinamide, p38-MAPK inhibitor, N2 and EGF, but supplemented with IGF1 and FGF2 as suggested for human organoid culture [25]. After several passages, duodenal, jejunal and colonic canine organoids grown in expansion medium were partly transferred to refined medium for long-term cultivation. For experiments, expansion medium-grown organoids where partly transferred to differentiation medium and after four days, organoids in all three media were harvested (Figure S1A). Growth characteristics and gene expression were assessed.

**Figure 1.** Establishment of new culture conditions for canine intestinal organoids. (**A**) RT-qPCR analysis of initial cell isolates (triangles), organoids with up to five passages (dark circles) and organoids with 10 or more passages (light circles) isolated from duodenum, jejunum or colon using *LGR5*, *VIL1*, *CHGA* and *MUC2* primer. log2 fold changes normalized to expression of initial cell isolates are shown with scatter dot plots; mean is shown; whiskers present SEM; \* *p* < 0.05, \*\* *p* < 0.01 and \*\*\* *p* < 0.001, statistical analysis given in detail in Supplementary Tables S2–S4; n = 3 individual dogs. (**B**) Light microscopic images of organoids derived from duodenum, jejunum and colon cultivated in expansion, differentiation and refined medium four days after seeding; scale bars represent 250 μm. (**C**) Mean activity of Annexin V-induced luciferase is shown for 0, 24, 48 and 72 h after addition of substrate to duodenal, jejunal and colonic organoids in expansion (circle), differentiation (square) and refined medium (triangle); whiskers represent SEM; \*\*\*\* *p* < 0.0001; 50 organoids seeded per replicate; n = 6 replicates.

Bright field images of duodenal, jejunal and colonic organoids grown in expansion and refined medium displayed intact, budding structures (Figure 1B). In contrast, all organoid types grown in differentiation medium displayed dark and necrotic areas (Figure 1B, Figure S1B). In accordance with these observations, luciferase-based Annexin V assay revealed low and comparable apoptosis in organoids grown in expansion and refined medium, but pronounced apoptosis in organoids grown in differentiation medium (Figure 1C). Organoids in expansion and refined medium could be cultivated for prolonged period of 25 passages after adaptation of growth medium (Figure S1C). Taken together, only expansion and refined medium supported organoid growth.

#### *3.3. Refined Medium Promotes Expression of Cell Di*ff*erentiation Markers in Canine Intestinal Organoids*

We next aimed at defining the cellular composition of organoids with respect to the different culture conditions. To this end, total RNA isolated from initial cell isolates and from the three different organoid types cultured in the three different media were subjected to gene expression analysis of stem cell and differentiation markers using RT-qPCR.

In general, a heat map representing log2 fold changes of organoids to respective initial cell isolates showed that transcription of stem cell markers was up-regulated in intestinal organoids grown in all three media. Conversely, transcription of differentiation markers was exclusively up-regulated in intestinal organoids grown in differentiation or refined medium (Figure 2A).

In more detail, duodenal organoids displayed significant up-regulation of stem cell marker *LGR5* [30,31] gene expression irrespective of culture conditions (Figure 2B). Culture of duodenal organoids in differentiation and refined medium enhanced mRNA expression of the differentiation markers *NEUROG3* (enteroendocrine progenitor cells) [32,33], *CHGA* (enteroendocrine cells) and *MUC2* (Goblet cells). Transcription of stem cell markers *ASCL2*, *EPHB2*, *BMI1*, *NOTCH1* and *PROM1* [30,31] did not vary significantly throughout duodenal organoids analyzed, while enterocyte marker *VIL1* mRNA was significantly down-regulated in duodenal organoids irrespective of culture conditions (Figure S2A).

Jejunal organoids grown in the three different media exhibited no significant difference in *LGR5*, *ASCL2*, *EPHB2*, *BMI1*, *NOTCH1* or *PROM1* transcription compared to corresponding initial cell isolates (Figure 2C, Figure S2B). *NEUROG3*, *CHGA* and *MUC2* gene expression was significantly down-regulated in jejunal organoids grown in expansion medium. Transcription of differentiation markers *NEUROG3*, *CHGA*, and *MUC2* mRNA was up-regulated when culturing jejunal organoids in refined medium. Expression of *VIL1* was slightly, but significantly, reduced throughout jejunal organoids (Figure S2B).

In colonic organoids, medium compositions had no significant impact on *LGR5* expression, while *NEUROG3*, *CHGA* and *MUC2* transcription was significantly down-regulated in organoids kept in expansion medium (Figure 2D). Culturing colonic organoids in differentiation and refined medium induced an increase of *CHGA* and *MUC2* expression almost to levels displayed by initial cell isolates. In contrast to duodenal and jejunal organoids, colonic organoids showed aberrant *ASCL2*, *EPHB2*, *BMI1* and *PROM1* transcription in comparison to initial cell isolates, whereas *NOTCH1* expression remained grossly unchanged (Figure S2C). *VIL1* transcription was generally slightly lower expressed in colonic organoids irrespective of culture conditions.

Marker transcription levels in organoids varied to some extent with respect to the individual dog they were derived from. These variations were less pronounced when cultivating organoids—notably jejunal and colonic organoids—in refined medium (Figure 2, Figure S2). Despite these individual variations the presented data indicate that differentiation and refined medium supported cell differentiation to secretory lineage cells within organoids.

**Figure 2.** Refined medium and differentiation medium show elevated marker expression of enteroendocrine and goblet cells. (**A**) Heat map of log2 fold changes derived from RT-qPCR data of duodenal, jejunal and colonic organoids cultivated four days in expansion, differentiation and refined medium normalized to initial cell isolates (i.c.i. = 0); data from three individuals are shown in side-by-side columns. (**B**–**D**) Individual scatter dot plots of gene expression data from (**A**) shown for stem cell marker *LGR5*, secretory lineage precursor marker *NEUROG3*, enteroendocrine cell marker *CHGA* and goblet cell marker *MUC2* for organoids derived from duodenum (**B**), jejunum (**C**) and colon (**D**) four days after seeding; mean is shown, whiskers are SEM; \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, statistical analysis given in detail in Supplementary Tables S5–S7; n = 3 dogs; i.c.i., initial cell isolates; exp., expansion medium; diff., differentiation medium; ref., refined medium.

#### *3.4. Refined and Di*ff*erentiation Medium Both Support Organoid Di*ff*erentiation*

The detection of markers for goblet and enteroendocrine cells by RT-qPCR prompted us to confirm and localize these cells within organoids. Therefore, periodic-acid Schiff (PAS) reaction and transmission electron microscopy (TEM) were used. PAS reaction of mucin as revealed by a purple signal demonstrated the presence of goblet cells that were evenly distributed in duodenal and jejunal tissue sections along the crypt-villus axis, and at higher frequency around the crypt base in colon tissue sections (Figure 3A). In organoid culture, expansion medium failed to induce differentiation into goblet cells, while growth in differentiation and refined medium gave rise to goblet cells irrespective of the intestinal segment. These observations were in accordance with *MUC2* transcription data (see Figure 2) and further substantiated by TEM, which revealed the presence of goblet cells with apical large mucin granules and enteroendocrine cells characterized by small dark basal granules in organoids grown in differentiation and refined medium (Figure 3B and Figure S3B). In contrast, TEM of organoids cultivated in expansion medium revealed uniformly appearing cells harboring microvilli (Figure S3A). These cells were also present in all other conditions suggesting the presence of enterocytes, which is correlating with *VIL1* expression data (see Figure 2 and Figure S2). Interestingly, cell-to-cell interactions such as tight junctions, adherens junctions and desmosomes could be visualized by organoid TEM throughout culture conditions (Figure S3C). These data demonstrate the presence of secretory lineage cells in canine organoids upon cultivation in refined and differentiation medium.

**Figure 3.** *Cont*.

**Figure 3.** Goblet and enteroendocrine cells are present in organoids upon cultivation in differentiation and refined medium, but not in expansion medium. (**A**) PAS reaction in FFPE sections of canine intestinal tissue or organoids in expansion, differentiation and refined medium derived from duodenum, jejunum and colon; counterstained with Hematoxylin, images of duodenal and jejunal tissue were cropped in order to visualize crypts; scale bars 200 μm, inset depicts higher magnification; arrow heads indicate PAS positive goblet cells; one representative image is shown. (**B**) TEM images of duodenal, jejunal and colonic organoids in expansion, differentiation and refined medium; scale bars are indicated; L, lumen; B, basal lamina; G, goblet cell; E, enteroendocrine cell.

#### *3.5. Refined Medium Supports Continuous Growth and Derivation of Organoids*

So far, we have been able to verify that organoids in refined medium were differentiating similarly or even better than organoids in differentiation medium (see Figures 2 and 3) without concomitant induction of apoptosis (see Figure 1B,C). Therefore, we conclude that a refined medium composition is beneficial over the use of a separate differentiation medium. In a next step we assessed the effectiveness of refined versus expansion medium on the continuous growth of organoids.

Importantly, all organoid types showed a similar proliferation pattern, as revealed by a scattered distribution of single, EdU-positive cells within organoids when exposed to the two different medium conditions (Figure 4A). However, viability of duodenal and jejunal organoids in expansion medium was reduced compared to colonic organoids, which could be significantly enhanced by using refined medium (Figure 4B).

To assess the suitability of refined medium for the establishment of new organoid lines, duodenal biopsies were taken during routine gastroduodenoscopy. Following isolation, starting cell material was cultured either with expansion medium containing Rock inhibitor, or with EGF- and Rock inhibitor-supplemented refined medium for the first two days, then changed to unsupplemented expansion and refined medium. After 10 days in culture, an almost 4-fold higher number of organoids had grown in refined medium compared to in expansion medium (Figure 4C). Organoids exposed to refined medium also displayed a higher projected area compared to organoids formed in expansion medium, indicating that refined medium allows for easier adaption of canine intestinal cells to in vitro culture conditions.

**Figure 4.** Derivation of organoids is more efficient using refined medium conditions. (**A**) 3D reconstruction of confocal z-stack images from duodenal, jejunal and colonic organoids in expansion and refined medium; staining for EdU after one hour labeling (magenta); organoids counterstained using DAPI (blue); scale bar represents 100 μm. (**B**) Relative luciferase activity of cell viability assay is shown for serial measurements at 0, 24, 48 and 72 h after addition of substrate to duodenal, jejunal and colonic organoids in expansion (circle) and refined medium (triangle); mean is shown; whiskers represent SEM; \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 and \*\*\*\* *p* < 0.0001; 50 organoids seeded per replicate; n = 6 replicates. (**C**) Organoid number is shown for freshly isolated duodenal biopsies seeded in expansion and refined medium, in scatter dot plots with mean and range indicated by whiskers, n = 2 wells; projected area of organoids is depicted with mean and SEM defined by whiskers, \*\*\*\* *p* < 0.0001, n = 7 organoids; bright field images of representative organoids are shown, scale bar represents 250 μm.

#### **4. Discussion**

Since the first reports on the derivation of organoids from murine and subsequently from human intestine, the establishment of canine intestinal organoids was a matter of time. Particularly as dogs develop GI diseases like cancer, infectious disease and IBD naturally [23], thereby representing a faithful model system for the corresponding human diseases. In early 2017 our group [17], Kingsbury et al. [34], and Powel and Behnke [16] reported on the successful derivation of canine intestinal organoids. We then focused on advancing organoid cultivation to bring them a step closer to the in vivo situation.

In the presented study, we found that expansion medium does not support stem cell commitment in the secretory lineage using transcription profiling. This finding is supported by reports on human intestinal organoids, which usually harbor only a limited number of differentiated cell types, because they are grown in expansion media that allow for the maintenance of stem cells but do not support secretory lineage differentiation [14,17]. This absence of differentiated cells in expansion medium could not be overcome by prolonged cultivation: low- (≤5) and high-passage (≥10) canine intestinal organoids all tested negative for differentiation marker expression. Recently, Chandra et al. published a well-written paper on the derivation of canine intestinal organoids and the presence of goblet and enteroendocrine cells [35]. The formulation of the expansion medium differs in a 5-fold higher concentration of p38-MAPK inhibitor and the addition of 8% fetal bovine serum FBS (Chandra) compared to the medium formulation used herein. p38-MAPK inhibitor in human organoid culture is used to suppress secretory lineage commitment of intestinal stem cells to prevent their depletion and maintain long-term culture [14,36,37]. The addition of 8% FBS is a likely explanation for the presence of secretory cells, potentially via subsequent activation of p38-MAPK signaling. To create a culture

system with defined components, the culture medium used in the study presented herein corresponds to a widely used human organoid medium without FBS [14] that was only supplemented with HGF to allow for the cultivation of over 50 passages. These variations in medium additives may explain the differences between both studies.

To further develop canine intestinal organoid culture, we focused on optimizing the growth medium to obtain a more in vivo-like cell composition with stem cells, enterocytes, goblet and enteroendocrine cells. Recently, a newly composed medium was shown to enable stem cells of human intestinal organoids to undergo differentiation, while concomitantly preserving the stem cell pool [25]. To overcome the intrinsic limitations of classical expansion medium, we adapted our canine intestinal expansion medium based on the suggestions of Fujii et al. [25] by (i) withdrawal of nicotinamide, p38-MAPK inhibitor, N2 and EGF; and (ii) addition of IGF1 and FGF2. This medium termed "refined medium" promoted stem cell growth and differentiation of canine intestinal organoids simultaneously for over 6 months of cultivation (data not shown). To assess the benefits and limitations of this refined medium in regard to organoid growth and lineage commitment, canine intestinal organoids grown in expansion, differentiation or refined medium were comparatively characterized.

While duodenal, jejunal and colonic organoids displayed a viable morphology upon culture in expansion and refined medium, differentiation medium induced apoptosis in all three intestinal segments. Interestingly, apoptosis peaked as early as 24 h after transfer to differentiation medium, indicating that programmed cell death was initiated by differentiation-unrelated factors. This assumption is supported by previous reports on the promoting effect of nicotinamide on cell survival in human pluripotent stem cells [38], on proliferation of aged murine organoids [39] and on inducing apoptosis via reduced intracellular NAD concentrations upon its withdrawal [40]. Taken together, differentiation medium induced apoptosis, thus not supporting long-term culture of canine intestinal organoids. This induction of cell death by the medium formulation is unfavorable when differentiation, proliferation or even apoptosis are investigated upon compound treatment for drug development. In addition, differentiation of organoids in expansion medium must be repeated for every individual experiment making them susceptible to fluctuations between media batches, illustrating the need for a better culture system.

Refined medium induced similar levels of differentiation or even outperformed differentiation medium in this regard. Duodenal, jejunal and colonic organoids grown in refined medium revealed enhanced expression of markers indicating the presence of secretory lineage precursors, enteroendocrine and goblet cells. Interestingly, organoids in refined medium expressed comparable or even higher levels of stem cell marker compared to organoids in expansion medium. Organoids irrespective of the intestinal segment or media composition expressed the enterocyte marker *VIL1*. These data confirm findings reported for human intestinal organoids grown in niche-inspired culture media that supported expression of enteroendocrine, goblet and stem cell markers [25] and data reported for villin1 positive enterocytes in murine intestinal organoids [12]. However, expression of the differentiation markers *CHGA*, *MUC2* and *VIL1* in differentiation and refined medium reached the levels of the i.c.i. only to a certain extent. While an in vitro culture model cannot fully reproduce the complex in vivo situation, our data show that growth conditions in refined medium are more similar compared to those in expansion medium.

Since only a limited number of specific antibodies were available for reliable characterization of canine cells, we used two well-established alternative methods, PAS reaction and TEM, for the detection of goblet cells and enteroendocrine cells. These results supported our RT-qPCR data, substantiating that only organoids grown in differentiation and refined medium harbored enteroendocrine and goblet cells. These findings are in accordance with data obtained for human intestinal organoids grown in differentiation [14] or niche-inspired medium [25]. Furthermore, the presence of enterocytes in organoids cultivated in expansion, differentiation and refined medium of all segments was supported by microvilli bearing cells as reported for murine intestinal organoids [12], indicating that enterocytes are present irrespective of culture condition.

So far, our data provided evidence that cultivation of canine intestinal organoids in refined medium is preferable to the use of differentiation medium, since it promotes stem cell growth and differentiation simultaneously, allows long-term growth without fluctuations between media batches and does not induce apoptosis. Therefore, further analysis focused on comparison of organoid growth in expansion and refined medium.

The comparison of proliferative behavior differences between canine intestinal organoids grown in expansion versus refined medium by EdU incorporation assays revealed similar proliferation levels for all organoid types irrespective of the medium used. Yet, viability data of duodenal and jejunal organoids indicate that expansion medium does not support small intestinal organoid cultivation as effective as refined medium. The colonic epithelium is highly populated with various bacteria and has less digestive and absorptive function than small intestinal epithelium possibly resulting in more robustness with respect to cultivation. Therefore, colonic organoids can be expanded in both expansion and refined medium efficiently.

Importantly, generation of organoids from duodenal biopsies was more successful in refined than expansion medium. Not only the number of organoids per well was enhanced, but also their projected area. This feature has an important impact on canine intestinal disease modelling, especially when donor tissue stems from GI enteropathies with severely affected epithelium. The high suitability of refined medium for canine organoid establishment also matches the higher plating efficacy shown for human single-cell dissociated organoids grown in niche-inspired media [25].

#### **5. Conclusions**

Taken together, we strongly recommend the use of refined medium for establishment and long-term culture of canine intestinal organoids. We could clearly show that this medium sustains stem cell growth, while simultaneously promoting differentiation of stem cells/immature cells into enterocytes, enteroendocrine and goblet cells. Canine intestinal organoids cultivated in refined medium bear the advantage of an easy to handle, reproducible and stable culture system, thereby representing a physiologically superior in vitro system for disease modelling, drug development, toxicity studies and personalized medicine. Compared to conventional expansion and differentiation media, the refined medium presented herein allows for more accurate assessment of genetic and epigenetic impacts on canine intestinal cell differentiation, bringing organoids a step closer to the in vivo situation.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/9/4/822/s1, Figure S1: Cultivation scheme of organoids and images of organoids in differentiation medium and after prolonged cultivation in expansion and refined medium; Figure S2: Gene expression of stem cell marker is increased in colonic organoids; Figure S3: TEM images of microvilli and cell-cell junctions in all conditions and goblet cells in organoids cultivated in differentiation and refined medium; Table S1: qPCR primer specifications used in this study; Table S2: Statistical analysis of duodenal tissue vs. duodenal organoids; Table S3: Statistical analysis of jejunal tissue vs. jejunal organoids; Table S4: Statistical analysis of colonic tissue vs. colonic organoids; Table S5: Statistical analysis of duodenal organoids in different media; Table S6: Statistical analysis of jejunal organoids in different media; Table S7: Statistical analysis of colonic organoids in different media.

**Author Contributions:** Conceptualization, N.K., B.P., A.M.C.M., A.S., H.S.K., K.S., L.C.P., B.S., M.K., S.B. and I.A.B.; Formal analysis, N.K.; Funding acquisition, L.C.P., B.S., S.B. and I.A.B.; Investigation, N.K., B.P., A.M.C.M., W.T. and I.W.; Methodology, N.K., B.P., A.M.C.M., H.S.K. and K.S.; Project administration, B.P. and I.A.B.; Resources, I.A.B.; Supervision, M.K., S.B. and I.A.B.; Validation, N.K., W.T. and I.W.; Visualization, N.K.; Writing—original draft, N.K. and S.B.; Writing—review & editing, B.P., A.M.C.M., W.T., I.W., A.S., H.S.K., K.S., L.C.P., B.S., M.K. and I.A.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** The Austrian Research Association FFG, project № 13610974, funded this work. The Dutch Research Council NWO ZON/MW (116004121) funded parts of this work.

**Acknowledgments:** This research was supported using resources of the VetCore Facility (Imaging, VetBioBank and Genomics) of the University of Veterinary Medicine Vienna. Open Access Funding by the University of Veterinary Medicine Vienna.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
