Morphological Dynamics of Leukemia Cells on TiO₂ Nanoparticle Coatings Studied by AFM

Abstract

Coatings of modified TiO₂ nanoparticles (TiO₂-m) have been shown to effectively and selectively trap non-adherent cancer cells, with an enormous potential for applications in photodynamic therapy (PDT). Leukemia cells have a remarkable affinity for TiO₂-m coatings, adhering to the surface by membrane structures and exhibiting morphologic characteristics of amoeboid locomotion. However, the details of the cell–substrate interaction induced by the TiO₂-m coating remain elusive. With the aim to obtain a better understanding of this phenomenon, leukemia cell adhesion to such coatings was characterized by atomic force microscopy (AFM) for short contact times up to 60 min. The cell and membrane morphological parameters mean cell height, contact area, cell volume, and membrane roughness were determined at different contact times. These results reveal cell expansion and contraction phases occurring during the initial stage of adhesion. Subsequently, the leukemic cells reach what appears to be a new resting state, characterized by pinning of the cell membrane by TiO₂-m nanoparticle aggregates protruding from the coating surface.

1. Introduction

Photodynamic therapy (PDT) is an approach to treating infections and cancer by exploiting the production of reactive oxygen species (ROS) by illumination of an administered photosensitizer (PS). Different types of PSs with different advantages and disadvantages have been developed, including organic PSs such as porphyrins and porphyrin derivatives [1], noble metal complexes containing for instance Ru(II) and Ir(III) [2,3], protein-based PSs such as KillerRed [4], as well as various nanomaterials. With regard to the latter, TiO₂ nanoparticles have attracted particular attention because they present numerous advantages over other established PSs, such as low production cost and straightforward synthesis, chemical stability, biological inertness, and high light conversion efficiency [5,6]. Consequently, TiO₂ nanoparticles have already been employed in the PDT treatment of different types of adherent cancer cells such as prostate, cervical, breast, and colon cancer, among others [6,7,8,9]. Nevertheless, there are still many obstacles to be overcome, in particular with regard to tumor targeting, accumulation in the reticuloendothelial system, hypoxic tumor microenvironments, and the poor tissue-penetration capability of light [10]. Additionally, it is rather challenging to treat advanced disseminated disease or non-adherent cancers such as leukemia with PDT. This is because the PS would have to freely travel through the patient’s bloodstream and thus require whole body irradiation, which is often impractical and sometimes even impossible [11,12]. Thus, there are only a few published in vivo or ex vivo studies on the PDT treatment of leukemia [13].

As a possible solution to address these issues, we have synthesized a family of TiO₂ nanoparticles (TiO₂-m) by a modified sol-gel method [14]. As photosensitizers in PDT, these nanoparticles have been found to cause a selective cytotoxic effect to cancer cells over healthy cells [14]. The mechanisms of interaction between leukemic cells and TiO₂-m coatings was recently described [15]. Briefly, the adhesion of leukemic cells on TiO₂-m coatings is a fast dynamic process involving expansion and contraction of the membrane induced by surface functional groups, particularly hydroxyl groups, which act as chemotactic factors. Additionally, the development of membrane structures such as blebs and pseudopods, characteristic of amoeboid locomotion, was identified by scanning electron microscopy (SEM). These structures were observed to envelop TiO₂-m nanoparticle aggregates that protrude from the coating surface to internalize them into the cytoplasm, indicating that nanoparticle internalization was triggered by TiO₂-m. Amoeboid-like locomotion thus appears to be an integral part of the interaction of leukemic cells with TiO₂-m coatings. In sum, these observations show that TiO₂-m nanoparticle coatings have an enormous potential for ex vivo PDT of non-adherent cancer cells since these coatings have been shown to efficiently and selectively trap acute lymphoblastic leukemia cells. However, the detailed morphological changes and the variation of cellular properties during the adhesion of leukemia cells to these TiO₂-m coatings are still largely unknown.

Atomic force microscopy (AFM) is a powerful tool for the imaging and characterizing of various biological entities such as biomolecules, cells, and tissues [16,17,18,19,20]. It provides three-dimensional surface topographic information and thus enables the quantification of morphological cell properties such as height, volume, and membrane roughness [21,22,23,24,25]. Because of that, AFM has become an important technique in investigating the complex behaviors of cells and cellular components, most importantly the cell membrane. The cell membrane plays a key role in numerous physiological processes such as cell adhesion, cell-surface recognition, and migration [26,27,28,29]. Therefore, the aim of this work was to characterize the initial stages of leukemia cell adhesion to the TiO₂-m coatings by AFM. For that, leukemia cells were cultured on TiO₂-m-coated glass substrates for times ranging from 10 to 60 min. Resulting changes in the morphology of the trapped cells were quantified by extracting various morphological parameters from the recorded AFM images, i.e., mean cell height, cell volume, area of contact with the substrate surface, and membrane roughness. We found that leukemia cell adhesion to TiO₂-m nanoparticle coatings starts with rapid movements of expansion and contraction, after which the cells seem to enter an immobile state with their cell membranes being pinned by TiO₂-m nanoparticle aggregates protruding from the coating surface.

2. Materials and Methods

2.1. Materials

Acute lymphoblastic leukemia cells (Molt-4 cell line, CRL-1582) were purchased from American Type Culture Collection (Manassas, VA, USA). The culture medium was created from RPMI-1640 (Gibco, Waltham, MA, USA), fetal bovine serum (Gibco, Waltham, MA, USA), and antibiotic (10,000 U/mil, penicillin-streptomycin, Gibco, Waltham, MA, USA). Standard glass slides (CITOTEST, Labware Manufacturing Co., Haimen, China) were used as substrates. Titanium (IV) butoxide (TBT, Sigma-Aldrich, Burlington, MA, USA), citric acid (99.5%, Sigma-Aldrich, Burlington, MA, USA), nitric acid (65%, Merck, Darmstadt, Germany), acetone (99.5%, Sigma-Aldrich, Burlington, MA, USA), and ethanol (99.8%, Sigma-Aldrich, Burlington, MA, USA) were used to clean the slides and prepare the TiO₂-m sol-gel suspension. Cells not adhering to the surfaces were removed with Hank’s Balanced Salt Solution (HBSS, Gibco, Waltham, MA, USA). Glutaraldehyde (3%, Sigma-Aldrich, Burlington, MA, USA) was used to fix the adhering cells on the samples before AFM imaging.

2.2. TiO₂-m Nanoparticle Coating

The glass slides were cleaned by successive immersion in 30% nitric acid, acetone, and absolute ethanol for 10 min each using an ultrasonic bath. After each cleaning step, the slides were rinsed with an abundance of distilled water. A convection oven was used to dry the cleaned slides for 120 min at 90 °C. Prior to deposition, the slides were treated with a low temperature corona discharge plasma using a corona treatment equipment 2KVA (INDELEC Ltd.a., Cali, Colombia) at 15 kV, 0.3 A, and a band speed of 1 m/min.

TiO₂-m sol-gel suspension was used to coat the treated glass slides by dip coating, as described previously [14]. Briefly, a suspension consisting of 3% functionalized multi-walled carbon nanotubes and absolute ethanol was prepared and sonicated for 5 min. Then, deionized distilled water was added at a rate of 10 mL/min. The volume fractions of the reagents were 76.34% TBT, 21.96% ethanol, and 1.7% water. Finally, the condensation reaction was carried out in a desiccator for 72 h at ambient temperature.

Dip coating was carried out by deposition of 5 coating layers at a withdrawal speed of 180 mm/min with intermediate drying steps between deposition steps at 60 °C for 30 min. Subsequently, the TiO₂-m coated slides were heat-treated to calcine and immobilize the nanoparticles on the surface using a muffle oven. The thermal treatment had 3 stages: (1) initial temperature (Ti) = 28 °C, heating rate (Hr) = 3 °C/min, final temperature (Tf) = 100 °C, heat stage time (Ht) = 30 min; (2) Ti = 100 °C, Hr = 3 °C/min, Tf = 500 °C, Ht = 60 min; (3) Ti = 500 °C, Hr = 3 °C/min, Tf = 600 °C, Ht = 180 min. A simple nanomaterial detachment test under agitation was carried out to determine the TiO₂-m coating stability (see Supplementary Materials), revealing that the coatings are sufficiently stable to survive the cell culture largely intact.

2.3. Leukemia Cell Culture

Leukemia cell culture (RPMI medium, 10% fetal bovine serum, and 1% antibiotic) with a cell concentration of 800,000 cells/mL was used for seeding. The coated and bare glass slides (control) were sterilized in an autoclave. Then, the slides were placed in rectangular well plates and a volume of 3 mL cell culture was slowly added to each well, covering the samples completely. The cells were incubated on the samples for different times ranging from 10 to 60 min at 37 °C in a humid atmosphere (60% relative humidity) and 5% CO₂. Because of the fast dynamics of cell adhesions, different samples were prepared and analyzed for each incubation time, instead of following the dynamics over time on a single sample surface. After incubation, any cells not adhering to the TiO₂-m coatings were removed by immersing the samples three times in HBSS. The cells cultured on the control slides were not washed since leukemia cells do not adhere to bare glass. To fixate the cells for AFM imaging, we used the well-established glutaraldehyde method [30,31,32,33,34,35]. The samples were immersed in 3% glutaraldehyde solution for 24 h at 4 °C. Afterwards, excess glutaraldehyde was removed by washing the samples with HBSS. Finally, the cells were dehydrated by successive immersion in a set of ethanol dilutions (50%, 60%, 70%, 80%, 90%, and 100%) for 20 min each. The samples were left to dry at room temperature.

2.4. Atomic Force Microscopy

The morphology of the cells was characterized by ex situ AFM using an MFP-3D-SA (Asylum Research, Santa Barbara, CA, USA) system operated in intermittent-contact mode in air. 50 × 50 µm² images were recorded at a resolution and scan rate of 512 px × 512 px and 0.3 Hz, respectively, using HQ:NSC15/AIBS cantilevers (325 kHz, 40 N/m, nominal tip radius 8 nm, Mikromasch, Wetzlar, Germany). For each incubation time, six to ten images recorded at different position on the sample surfaces were analyzed. The AFM images were processed and analyzed in Gwyddion 2.52 open source software [36] (see Figures S2–S6 for representative examples of overview images). The root-mean-square (RMS) roughness (S_q) of the cell surfaces was calculated from cropped regions of the images that did not include any of the cells’ surroundings. The mean height of each cell (z_m), the projected contact area between cell and substrate (A), and the minimum basis cell volume (V) were determined using the Grain Analysis tool after manually masking each cell individually using the Edit Mask tool (see Figures S7–S11 for representative examples).

3. Results and Discussion

3.1. Morphology of Leukemia Cells on Glass

The morphology of acute lymphoblastic leukemia cells on bare glass slides was characterized by AFM to provide a reference for further experiments with leukemia cells adhering to the TiO₂-m nanoparticle coatings. A representative AFM image of a resting cell is shown in Figure 1a. The inactivated cell has a typical elliptical shape with a maximum height of about 2 µm (Figure 1a,c). In the 3D representation shown in Figure 1b, pseudopods extending from the membrane and microvilli on the membrane can easily be identified. Figure 1d shows a 3D representation of a zoom of the membrane region indicated in Figure 1a. The membrane has a corrugated morphology with pore-like features, consistent with previous observations of leukemic cells [22,37]. In general, the cell membrane components appear to be uniformly distributed and granular. The height distribution function of this membrane area is comparatively narrow, with the majority of cell surface features having heights between 200 and 300 nm (Figure 1e).

3.2. Time-Dependent Morphology of Leukemia Cells on TiO₂-m Coatings

Cells seeded on the TiO₂-m coatings were characterized by AFM after different incubation times ranging from 10 to 60 min. Figure 2 evaluates the morphology of a representative cell after 10 min incubation. Within the first 10 min of contact with the coating, the cell membrane completely loses its original ellipsoidal shape (Figure 1a) and expands over the coating (Figure 2a,b), maximizing the contact area with the coating surface. This behavior is consistent with amoeboid locomotion on the TiO₂-m coatings as previously observed [15]. In the absence of signals from extracellular matrix components and other cells, amoeboid locomotion is activated by chemotactic agents. Therefore, the spreading of the leukemia cell on the TiO₂-m coating likely reflects the cell trying to find a gradient chemotactic signal and establish a direction of movement or anchor points [38]. Figure 2b shows that the expanded cell (highlighted with a green mask) adopts the topography of the coating, demonstrating the flexibility of the cell membrane to accommodate the rough topography of the coating surface, which exhibits numerous nanoparticle aggregates with heights up to about 1 µm (Figure 2c). This results in a comparatively broad height distribution of the zoomed membrane region shown in Figure 2d, which now has become highly asymmetric and extends to heights well beyond 500 nm (Figure 2e).

After 20 min of contact, the cell shown in Figure 3a,b can be seen to contract from all directions without a defined anchor point. This can be explained by the TiO₂-m coating having a homogeneous chemical composition [15], so that the adhering cell could not detect any gradient in the chemotactic signals originating from the TiO₂-m coating to define a migration direction during the initial expansion phase. Additionally, some TiO₂-m nanoparticle aggregates up to about 2 µm in height can be identified in the boundary region of the cell membrane (red arrows in Figure 3b,c). The cell membrane appears to be pinned by these large nanoparticle aggregates, which indicates that they are either surface features that hinder further membrane contraction or were dislodged from the coating and entrained in the previous contraction phase. However, since the performed detachment test showed only little nanoparticle release from the substrate over much longer time scales (see Supplementary Materials), we assume that the former is a more likely scenario. Regarding the morphology of the cell membrane, the pronounced topographic features seen before in the center of the cell (Figure 2d) have mostly disappeared during contraction (Figure 3d). This is also reflected in the height distribution function of the membrane in Figure 3e, which has become narrower again. Together with the observed increase in overall cell height from few hundred nanometers to about 1 µm (Figure 3c), this further supports the above interpretation that the larger topographic features seen in Figure 2d were actually caused by the rough topography of the underlying coating surface that was closely followed by the highly spread and thus very thin cell.

Figure 4a,b shows a representative leukemia cell after 40 min of contact with the TiO₂-m coating. Here, some pseudopodia (white arrows in Figure 4a) and protuberances (red arrow in Figure 4b) in the cell membrane can be identified. The latter is probably the result of the endocytosis of immobile TiO₂-m nanoparticle aggregates protruding from the coating surface. The red arrow in Figure 4b points to one of the membrane protuberances, which is located inside the cell close to the membrane boundary and has a height of close to 1.5 µm (red arrow in Figure 4c). This observation is a further indication that the cell internalizes immobile TiO₂-m particle aggregates, consistent with previously reported results [15]. Aside from these protrusions, the membrane morphology recorded in the center of the cell has become rather homogeneous but still exhibits some corrugations and granular features, as can be seen in Figure 4d. Consequently, the height distribution in Figure 4e has further decreased in width with all height values now lying below 600 nm.

Finally, after 60 min of contact, the cell shown in Figure 5a,b has reverted to a mostly ellipsoidal shape but still exhibits membrane protrusions (indicated by the red arrow in Figure 5b). Notably, there are almost no more TiO₂-m nanoparticle aggregates visible right at the membrane boundary but either clearly outside or inside the cell, with the latter being another indicator of TiO₂-m getting internalized into the cytoplasm by membrane structures, such as pseudopods, which envelop the surrounding aggregates completely. Regarding the membrane morphology in the center of the cell, a slightly corrugated topography with a homogeneous distribution of surface features can be seen in Figure 5d. This is also reflected in the height distribution function shown in Figure 5e, which exhibits a small width but has a visible shoulder at higher z values, corresponding to the few yet pronounced membrane protrusions.

As was described previously, the response of the leukemic cells to contact with the TiO₂-m coatings does not only include amoeboid locomotion and membrane structure formation, but also the internalization of agglomerated TiO₂-m nanoparticles on the coating surface [15]. Therefore, the membrane protrusions identified close to the membrane boundary (red arrows in Figure 3) or inside the cytoplasm (red arrows in Figure 4 and Figure 5) most likely result from the internalization of TiO₂-m nanoparticle aggregates that protrude from the coating surface, as previously observed by SEM. Since these nanoparticle aggregates are immobile on the coating surface, this internalization results in the pinning of the cell.

3.3. Statistical Analysis of Morphological Parameters

To better support the above qualitative observations, we have statistically analyzed a number of parameters to quantify cell morphology, i.e., the mean height of the cells z_m, the contact area between the cell and the substrate A, the cell volume V, and the RMS roughness of the cell surface S_q. As can be seen in Figure 6a, the mean height of the cells decreases significantly upon initial contact with the TiO₂-m coating, compared to the mean height of the cells on the bare glass substrate. This reduced height is accompanied by an almost threefold increase in the area of contact with the substrate (see Figure 6b), which is evidence for the spreading of the cells on the TiO₂-m coating. After 20 min of contact, however, the mean height has increased and is now comparable to the glass control (Figure 6a), while the contact area is now right between the glass control and 10 min contact with the TiO₂-m coating (Figure 6b). Notably, at this contact time of 20 min, the volume of the cells is about 50% larger than for all other contact times as well as for the glass control (see Figure 6c). This change in cell volume may be related to some cytoskeletal rearrangement by polymerization and depolymerization of actin filaments and microtubules [39]. This interpretation is further supported by the fact that after 20 min contact with the TiO₂-m coatings, the membranes of the cells exhibit a drastically increased surface roughness (see Figure 6d), as it was shown that membrane roughness is controlled mostly by actin dynamics [40]. However, the increase in cell volume and membrane roughness may also be caused to some extent by TiO₂-m internalization.

For longer incubation times exceeding 20 min, the cells on the TiO₂-m coatings appear to approach a new resting state. This state is characterized by a reduction of the mean height of the cells, while their area of contact with the TiO₂-m coating surface remains constant at the intermediate value observed already at 20 min incubation (see Figure 6a,b). Most interestingly, in this state, both the cell volume (Figure 6c) and the membrane roughness (Figure 6d) are back at their initial values, indicating that the processes of cytoskeleton remodeling and/or TiO₂-m internalization are coming to an end. This steady state might be related to the pinning of the cell membranes by larger TiO₂-m nanoparticle aggregates as observed in the AFM images in Figure 3, Figure 4 and Figure 5 and quantified by the intermediate value of the contact area in Figure 6b. Such a pinning would hinder further membrane retraction and thereby interrupt or at least delay the completion of the first cycle of amoeboid locomotion.

It was previously shown that the number of trapped leukemic cells on the TiO₂-m coatings increases with contact time until it reaches a maximum value at 40 min, after which it decreases again [15]. Therefore, this time point should be chosen for efficient cell trapping. For PDT, however, not just efficient cell trapping but also efficient cell damage by the generated ROS is of utmost importance. Based on the data presented in Figure 6b, the latter should be accomplished already at a shorter contact time of 10 min, where the contact area between the cell and coating surface has reached its maximum, so that the cell membrane should be exposed to the maximum concentration of ROS upon light irradiation. On the other hand, the current AFM as well as previous SEM data [15] suggest the internalization of large TiO₂-m nanoparticle aggregates protruding from the coating surface for longer contact times. This would present the opportunity to generate cytotoxic ROS not only close to the outer surface of the cell but even within the cytoplasm. Further in vitro and ex vivo studies are necessary to determine whether such effects indeed result in contact time-dependent cell killing efficiency.

4. Conclusions

In this work, the dynamics of cell and membrane morphology during the adhesion of leukemia cells to TiO₂-m nanoparticle coatings was characterized by AFM. During the first 20 min of contact, fast expansion and contraction movements occurred. In the former, the cells reached the maximum contact area with the substrate and minimum height values. This is attributed to the cells trying to find a chemotactic signal gradient and establish a migration direction. In the following contraction phase, the cells contracted from all directions because they did not find a signal gradient in the previous expansion phase. This led to maximum height, volume, and membrane roughness values. These changes in the cells’ morphological parameters may be related to cytoskeletal rearrangements at each phase and to some extent also to TiO₂-m internalization. For longer contact times exceeding 20 min, there was no evidence of a new fast expansion phase. Instead, the cells seemed to approach to a new steady state that is different from the control and the initial expansion phase. In this new state, which may be related to the pinning of the cell membrane by large, immobile TiO₂-m nanoparticle aggregates, the values of the cell volume and membrane roughness slowly go back to their control values, indicating that the initial stimulation triggered by the TiO₂-m coating is slowly coming to an end. These results also suggest a possible internalization of TiO₂-m nanoparticle aggregates into the cytoplasm, consistent with previous reports [15]. In summary, our results provide some deeper insights into the unique type of interaction between non-adherent cancer cells and TiO₂-m nanoparticle coatings, which lead to an interrupted amoeboid locomotion of the cells in the initial stages of contact with the coatings.