**Formulation of Tioconazole and** *Melaleuca alternifolia* **Essential Oil Pickering Emulsions for Onychomycosis Topical Treatment**

**Barbara Vörös-Horváth 1, Sourav Das 2,3, Ala' Salem 1, Sándor Nagy 1, Andrea Böszörményi 4, Tamás K˝oszegi 2,3, Szilárd Pál 1,\*,**† **and Aleksandar Széchenyi 1,\*,**†


Academic Editors: Marina A. Dobrovolskaia and Francesca Mancianti Received: 13 October 2020; Accepted: 23 November 2020; Published: 26 November 2020

**Abstract:** Onychomycosis is a disease that affects many adults, whose treatment includes both oral and topical therapies with low cure rates. The topical therapy is less effective but causes fewer side effects. This is why the development of an effective, easy to apply formulation for topical treatment is of high importance. We have used a nanotechnological approach to formulate Pickering emulsions (PEs) with well-defined properties to achieve site-specific delivery for antifungal drug combination of tioconazole and *Melaleuca alternifolia* essential oil. Silica nanoparticles with tailored size and partially hydrophobic surface have been synthesized and used for the stabilization of PEs. In vitro diffusion studies have been performed to evaluate the drug delivery properties of PEs. Ethanolic solution (ES) and conventional emulsions (CE) have been used as reference drug formulations. The examination of the antifungal effect of PEs has been performed on *Candida albicans* and *Trichophyton rubrum* as main pathogens. In vitro microbiological experimental results suggest that PEs are better candidates for onychomycosis topical treatment than CE or ES of the examined drugs. The used drugs have shown a significant synergistic effect, and the combination with an effective drug delivery system can result in a promising drug form for the topical treatment of onychomycosis.

**Keywords:** pickering emulsions; onychomycosis topical treatment; tioconazole; tea tree essential oil; antifungal activity

#### **1. Introduction**

Onychomycosis is a fungal infection of nails and nail bed and occurs on both finger and toenails. This fungal infection affects about 11.4% [1] of the adult population and is responsible for more than 50% of nail diseases [2]. For the treatment of onychomycosis, oral, topical, mechanical, and chemical therapies or a combination of these methods are used in practice [3]. The therapy is a long process (10–12 months) and has a poor cure rate [4]. Oral therapy is the most effective, but in the case of prolonged use, orally administered drugs can cause severe side effects because of their high toxicity [5]. Drug interactions can also occur, which is the main reason for the contraindication of oral therapy.

In such cases, topical therapy is recommended, and it is a more attractive alternative for patients. Unfortunately, the topical treatment of onychomycosis is limited because the nail plate acts as a barrier to drug diffusion. Its hydrophilic nature and keratinized structure reduce the diffusion of high molecular weight or lipophilic antifungal drugs [6]. In order to enhance the penetration of drugs, diffusion enhancers (e.g., mechanical pretreatment [7], phosphoric acid [8], keratolytes [9]) or an appropriate formulation should be used [3].

The azole antifungal agents have been used since 1980 for topical and oral therapy of fungal infections, among others for onychomycosis; in clinical treatment, they are used for superficial and systemic fungal infections with safety [10]. The water solubility of generally applied antifungal azole derivatives in the treatment of fungal nail infections is very low (<0.01 mg/mL) [10]; therefore, their formulation contains organic solvents in most cases. Commercially available nail lacquers contain organic solvents to increase the solubility of antifungal drugs, but these solvents act unfavorably to the drug permeability [11]. Their restricted drug delivery ability is caused by the rapid evaporation of organic solvents, so some drugs remain on the surface of the nail [12]. Several researches have proved that drugs with aqueous-based formulations have higher nail permeability than non-aqueous ones [13,14]. Another problem with azole derivatives is that fungi can become resistant to the drug in a long-lasting treatment [15]. A combination of azoles with antifungal essential oil (EO) could solve this problem because fungi cannot easily acquire resistance to multiple antifungal components of EOs [16]. Moreover, the azole derivatives could show synergistic antifungal activity with some essential oils (EOs) [17], presumably because of their different mode of action. The azole derivatives inhibit the action of cytochrome P-450 enzyme, lanosterol demethylase of fungi [18], thereby preventing the synthesis of ergosterol, while the EOs damage the cell membranes and organelles of fungi [19]. Because of the lipophilic character of azole derivatives [10], it is likely that it can be dissolved in lipophilic EO, and their solution can be used for drug formulation.

The oil in water type emulsions as water-based drug formulations can provide a possible way to overcome the water solubility problems. Conventionally, the emulsions are stabilized with surfactants. In a long-lasting topical treatment, the use of surfactants should be avoided because they can cause irritation and side effects, or in some cases, they can get into the blood circulation [20]. Using particle-stabilized emulsions, i.e., Pickering emulsions (PEs) [21] instead of a conventional emulsion, has several advantages. The solid particles spontaneously adsorb on the oil-water interface and form a shell-like structure on the PE droplet surface [22]. This adsorption can be considered as irreversible adsorption because the solid particles have higher adsorption energy than surfactants on oil-water interfaces [23], so the stability of PEs can be the same or better than conventional emulsions. Another important parameter for emulsion-based drug formulation and therapy is the size of emulsion droplets. The droplet size of PE can be influenced by several parameters, such as emulsification time and energy, oil to solid particle volume ratio, and concentration of oil phase [24]. The fungal hypha damages the nail structure, creating pores in μm size range [25], with porosity in the 5–20% range depending on pretreatment of the nail [26]. An emulsion droplet with the appropriate size could penetrate into the porous nail structure and retain the antifungal drug on the nail bed for a longer period, which is the main site of reinfection [27], and thereby a targeted drug delivery can be achieved. Inert and biocompatible particles should be chosen as stabilizing particles in PEs drug formulation.

Silica nanoparticles (SNPs) are widespread in pharmaceutical technology in topical treatments because of their favorable chemical and surface properties, thermal stability, mechanical resistance, and biocompatibility [28,29]. The effects of topically applied SNPs have been examined in detail [30], and it has been found that they have no toxic effect even after prolonged usage. Because of the above-mentioned advantageous properties of SNPs, they can be suitable for PEs stabilization. The PEs are most stable when the partial wetting conditions of stabilizing particles are the same for the oil and water phase [31]. The native SNPs are hydrophilic because of the high number of free silanol groups at their surface [32], which has to be modified with organic ligands to achieve appropriate wettability and strong stabilizing effect.

In the present study, our aim was to formulate PE of an azole derivative and antifungal essential oil as an alternative formulation for onychomycosis topical treatment. Tioconazole (TIO) has a broad antifungal activity for common dermatophytes, which has proved to be efficient for the topical treatment of fungal infections [33]. Nenoff et al. determined that *Melaleuca alternifolia* (MA) EO (tea tree EO) inhibited the growth of several clinical fungal isolates, so they suggested its use in the topical treatment of fungal infections [34]. For PE formulation, we used the solution of TIO in MA EO. Synthesis and surface modification of SNPs with different sizes were performed, and they were used as stabilizing agents of PEs. We characterized the stability, droplet size, and emulsion type of PEs. In vitro diffusion studies were also performed through artificial membranes. The aim of the in vitro diffusion test was to compare drug delivery characteristics of different formulations in the membranes that have similar porous structure and surface properties as the nail plate and nail bed. The antifungal activity against *Candida albicans* and *Trichophyton rubrum*—the species mainly responsible for fungal nail infections—has been investigated [3].

#### **2. Results and Discussion**

#### *2.1. Characterization and In Vitro Di*ff*usion Study of PEs*

#### 2.1.1. Characterization of SNPs

The size distribution and PDI values for synthesized and surface-modified SNPs were determined by DLS and TEM. Data for mean diameter and PDI values are presented in Table 1. The TEM images showed that SNPs were monodispersed, spherical, and had a smooth surface. It can be clearly seen that the size and morphology did not change significantly during the surface modification (Table 1 and Figure 1). The surface modification of SNPs was confirmed by FTIR spectroscopy; the results were previously published [35].


**Table 1.** Properties of SNPs. HS: hydrophilic particle. ET: with ethyl functional group modified particle. Size represented as mean ± SD of three parallel syntheses. The numbers refer to the particle sizes.

**Figure 1.** TEM images of silica nanoparticles (SNPs). (**A**): HS100, d = 103.0 nm. (**B**): ET100, d = 110.7 nm. (**C**): HS50, d = 53.0 nm. (**D**): ET50, d = 55.0 nm. (**E**): HS20, d = 20.0 nm. (**F**): ET20, d = 20.0 nm.

#### 2.1.2. GC Analysis of *Melaleuca Alternifolia* EO

The composition of MA was determined by gas chromatography. The components were identified by comparing their retention times and relative retention factors with standards and oils of known composition. Two parallel measurements were performed. The main compounds were *p*-cymene 35.2% and terpinene-4-ol 32.5%. A detailed composition is presented in Table 2.


**Table 2.** Composition of *Melaleuca alternifolia* EO. The results of GC analysis showed the average of the two parallel measurements in the percentage of volatile compounds. The main components of MA have been highlighted.

#### 2.1.3. Characterization of PEs

Properties of PEs are influenced by many parameters, like the interfacial surface tension of the phases, size, wettability, and concentration of stabilizing particles, o/w phase ratio, and emulsification energy. In this study, we examined the influence of the o/w phase ratio and size of the stabilizing SNPs on the droplet size and stability of formulated PEs, while other parameters were kept constant.

The results can be seen in Table 3, including data for composition, droplet size, and appearance of PEs. For the microbiological experiments and in vitro diffusion studies, we used emulsions that were stable for at least one week, which means that their droplet size did not change within this period, and creaming, phase separation, aggregation, or sedimentation of SNPs did not occur.

As shown by Binks and Horozov [31], the size of stabilizing particles influences the emulsion droplet size at the same o/w phase ratio. We found that the increment of oil phase concentration caused an increment of emulsion droplet size at all examined oil phase concentrations. When 20 nm SNPs were used, the droplet size increased until reaching a droplet size of 1.8 μm. Further increment of oil phase concentration did not cause significant droplet size increment, and the stability of emulsions was much higher in the 11.16–16.12 mg/mL concentration range (for oil phase) (see Table 3). We observed a similar effect when 50 nm SNPs were used in the 4.48–11.19 mg/mL concentration range and 1.6 μm droplet size (Table 3). We could not observe such behavior for PE 100ET. In this case, the droplet size continuously increased as the oil phase concentration increased, and the stability of emulsions was much lower (less than one week).

The stability of PE was also influenced by the size of the stabilizing SNPs. We found that with the increasing SNPs size, the stability of PE decreased, at the same o/w ratio. The PE stability was 20 weeks using 20ET, 8 weeks for 50ET, and 1 week for 100ET SNPs, at 11.19 mg/mL oil phase concentration. The zeta potential values of PEs could also provide their colloidal stability, whose values did not differ significantly from the SNPs suspensions. The zeta potential of 20ET, 50ET, and 100ET SNPs was −28.3, −25.2, and −25.0 mV, while the values of PE 20ET, PE 50ET, and PE100Et in the 0.90–17.91 mg/mL oil concentration range were from −28.0 to −19.3 mV, −24.9 to −19.0 mV, and −24.3 to −17.6 mV, respectively.

The type of emulsions was determined by conductivity measurements. The conductivity values of stabilizing SNPs suspended in distilled water were 215.0, 211.4, and 268.3 <sup>μ</sup>S·cm−<sup>1</sup> for 20ET, 50ET, and 100ET, respectively, while conductivity for the oil phase was 0.058 <sup>μ</sup>S·cm−1. The conductivity values of PEs were in the 157.33–257.50 <sup>μ</sup>S·cm−<sup>1</sup> range, which means that all the PEs were o/w type emulsions.


**Table 3.** Parameters of Pickering emulsions of *Melaleuca alternifolia* EO and tioconazole stabilized with 20ET, 50ET, and 100ET SNPs. \* These emulsions were creaming, but after 30 s shaking, their droplet size recovered to the original value, and they retained again their stability for 1 week.

#### 2.1.4. In Vitro Diffusion Studies through Artificial Membranes

Our goal was to formulate an emulsion that is capable of delivering the antifungal drugs through the nail plate and retain the drugs in the site of the infection (nail bed) for a prolonged time to provide a sustained drug release. The diffusion studies of PEs on the artificial membranes were performed in Franz diffusion vertical cells in order to examine the drug delivery ability of the formulated PEs. For diffusion studies, we applied PEs, CE, and ES of the same concentration, 17.91 mg/mL, as an antifungal drug combination. Because of the droplet size similarity between the CE and PEs, we can assume that only the dosage form determined the diffusion properties of the drug.

We found that PEs possessed better drug delivery properties through agar gel membrane compared to CE and ES (Table 4 and Figure 2). We examined the diffusion properties of PEs with different droplet sizes and found that the PEs with smaller droplet size (1.85 μm) could deliver as much as 89.9% of TIO through the agar membrane. In the experiment where the composite membrane was used, we found that the ES had diffused through the composite membrane structure, and only a small portion (2.4%) of the drug remained in the composite membrane (Table 4 and Figure 3). The PE 20ET delivered 89.9% of TIO through the agar gel membrane, and only 5.7% had diffused through composite membranes, suggesting that 84.2% of the applied drug remained in the targeted area. This amount was 61.1% at PE 50ET and 45.13% at PE 100ET. These in vitro experimental results suggested that PEs had better on-site drug delivery properties.


**Table 4.** Results of in vitro diffusion studies. ES: ethanolic solution, CE: conventional emulsion, PE: Pickering emulsion, CA: cumulative TIO amount after 2 h. The concentration of TIO was 3.58 mg/mL in each formulation.

**Figure 2.** In vitro diffusion studies through agar gel membrane. ES: ethanolic solution, CE: conventional emulsion, PE: Pickering emulsion, CA: cumulative TIO amount after 2 h. CTIO = 3.58 mg/mL.

**Figure 3.** In vitro diffusion studies through the composite membrane. ES: ethanolic solution, CE: conventional emulsion, PE: Pickering emulsion, CA: cumulative TIO amount after 2 h. CTIO = 3.58 mg/mL.

#### *2.2. Microbiological Tests Using Candida albicans and Trichophyton rubrum*

Data obtained for the minimum inhibitory concentration (MIC) and minimum fungicidal concentration (MFC) on *T. rubrum* and on *C. albicans* are shown in Table 5 and Figures 4 and 5 for the ethanolic solution of TIO (ES-TIO) and ethanolic solution of MA (ES-MA) and their combinations. The TIO and MA combination showed a significant synergistic effect. When *T. rubrum* and *C. albicans* were treated with the combination of TIO and MA, both the MIC and MFC values decreased significantly compared to the separately used drugs.

Analyzing the antimicrobial data of the different formulations of TIO and MA clearly showed that the PEs were more effective than CE or ES against the two pathogens. The PE 100ET showed the most effective growth inhibition against both *T. rubrum* and *C. albicans*, and this formulation had the highest fungicidal activity.


**Table 5.** Minimum inhibitory concentration (MIC) and minimum fungicidal concentration (MFC) of the test samples in combinations on *T. rubrum* and on *C. albicans.*

**Figure 4.** Minimum inhibitory concentration (MIC90) of ES-TIO, ES-MA, ES-TIO-MA, CE-TIO-MA, PE 20ET-TIO-MA, PE 50ET-TIO-MA, and PE 100ET-TIO-MA in μg/mL on *T. rubrum* (**A**) and *C. albicans* (**B**). Six independent experiments with three technical replicates were performed. The green (**\***) and red (**\***) asterisks represent a significance value of *p* < 0.01 for the MIC90, respectively.

**Figure 5.** Minimum fungicidal concentration (MFC) of ES-TIO, ES-MA, ES-TIO-MA, CE-TIO-MA, PE 20ET-TIO-MA, PE 50ET-TIO-MA, and PE 100ET-TIO-MA in μg/mL on *T. rubrum* (**A**) and *C. albicans* (**B**). Six independent experiments with three technical replicates were performed. The green (**\***) and red (**\***) asterisks represent a significance value of *p* < 0.01 for the MFC, respectively.

#### **3. Materials and Methods**

#### *3.1. Preparation and Characterization of Silica Nanoparticle-Stabilized Pickering Emulsions*

#### 3.1.1. Synthesis, Surface Modification, and Characterization of Silica Nanoparticles

Synthesis of hydrophilic SNPs (HS) was performed based on the work of Stöber et al. [36]. Size-selective synthesis parameters were set based on our previous work [37], as well as the reaction circumstances for surface modification. The synthesis route and details can be found in the Supplementary Materials. We previously reported that SNPs that have theoretical surface coverage of 20% with ethyl groups could stabilize the MA droplets to give a stable PE [24]. For the synthesis and surface modification of silica nanoparticles, the following chemicals were used: tetraethoxysilane ([TEOS] (Alfa Aesar GmbH, Karlsruhe Germany, purity 98%), ethyltriethoxysilane ([ETES] Alfa Aesar Karlsruhe Germany, purity 96%), absolute ethanol (VWR Chemicals Ltd., Debrecen Hungary, AnalaR NORMAPUR® <sup>≥</sup>99.8%), and 28 m/m% ammonium solution (VWR Chemicals Ltd., Debrecen Hungary, AnalaR NORMAPUR®, analytical reagent).

The size distribution was determined by dynamic light scattering (DLS) (Malvern Zetasizer Nano S, Malvern Panalytical Ltd., Great Malvern, Worcestershire, UK). The size distribution was confirmed, and the morphology of silica nanoparticles was studied by transmission electron microscopy (TEM) (JEOL-1400 electron microscopy, JEOL Ltd., Tokyo, Japan). For TEM experiments, 400 mesh copper grids coated with carbon were used (Micro to Nano Ltd., Haarlem, Netherlands).

#### 3.1.2. Gas Chromatography Analysis of *Melaleuca Alternifolia* Essential Oil

#### Solid-Phase Microextraction (SPME) Conditions

Samples were loaded into vials (20 mL headspace) sealed with a silicon/PTFE septum prior to SPME-GC/MS analysis. Sample preparation using the static headspace solid-phase microextraction (sHS-SPME) technique was carried out with a CTC Combi PAL (CTC Analytics AG, Zwingen, Switzerland) automatic multipurpose sampler using a 65 μM StableFlex polydimethyl siloxane/ carboxene/divinyl benzene (CAR/PDMS/DVB) SPME fiber (Supelco, Bellefonte, PA, USA). After an incubation period of 5 min at 100 ◦C, extraction was performed by exposing the fiber to the headspace of a 20 mL vial containing the sample for 10 min at 100 ◦C. The fiber was then immediately transferred to the injector port of the GC/MS and desorbed for 1 min at 250 ◦C, in split mode, and the split ratio was 1:90. The SPME fiber was cleaned and conditioned in a Fiber Bakeout Station in a pure nitrogen atmosphere at 250 ◦C for 15 min.

#### GC-MS Conditions

The analyses were carried out with an Agilent 6890N/5973N GC-MSD (Agilent Technologies, Santa Clara, CA, USA) system equipped with Supelco (Sigma-Aldrich Ltd., Budapest, Hungary) SLB-5MS capillary column (30 m × 250 μm × 0.25 μm). The GC oven temperature was programmed to increase from 60 ◦C (3 min isothermal) to 250 ◦C at 8◦C/min (1 min isothermal). High purity helium (6.0) was used as carrier gas at 1.0 mL/min (37 cm/s) in constant flow mode. The mass selective detector (MSD) was equipped with a quadrupole mass analyzer and was operated in electron ionization mode at 70 eV in full scan mode (41–500 amu at 3.2 scan/s). The data were evaluated using MSD ChemStation D.02.00.275 software (Agilent Technologies, Santa Clara, CA, USA). The identification of the compounds was carried out by comparing retention data and the recorded spectra with the data of the NIST 2.0 library. The percentage evaluation was carried out by area normalization.

3.1.3. Determination of Solubility of Tioconazole in *Melaleuca Alternifolia* Essential Oil

Solubility Calculations by Hansen Solubility Parameters (HSPs)

As a preliminary study, the calculations of solubility parameters were performed using the Hansen Solubility Parameters in Practice (HSPiP) software version 5.0.11 using the simplified molecular-input line-entry system (SMILES), obtained from PubChem. HSPs (Equation (1)) use group contribution to split the total cohesion energy of a solvent into contributions from atomic dispersion (δ*d*), polar interactions (δ*p*), and hydrogen bonding (δ*h*) [38].

$$\delta = \left(\delta\_d^2 + \delta\_p^2 + \delta\_h^2\right)^{0.5} \tag{1}$$

Differences in solubility parameters were calculated with the HSP difference (Equation (2)). A value below that of the reported cut-off value 7 Mpa0.5 indicates miscibility [39].

$$
\Delta \delta = |\delta\_{\text{solvent}} - \delta\_{\text{tiocouazole}}| \tag{2}
$$

For the calculation, the three main components of MA were used (*p*-cymene, terpinene-4-ol, γ-terpinene), and it could be established that TIO can be dissolved in the EO. The results of the calculation can be found in Supplementary Materials Tables S2–S4. In order to determine the exact solubility of TIO in MA, the solvent addition method was performed (Section 3.1.3 Determination of Kinetic Solubility).

#### Determination of Kinetic Solubility

The kinetic solubility of TIO (tioconazole, purity ≥98%, Alfa Aesar, Karlsruhe, Germany) in water-saturated MA (*Melaleuca alternifolia* essential oil, Tebamol®, BIO-DIÄT-BERLIN GmbH, Berlin, Germany) was determined by the solvent addition method [40]. The examination was performed at ambient temperature (25 ◦C). The initial suspension was prepared by weighing the exact amount of 1.0 mg TIO and the addition of 500 μL of MA. The volume of MA was increased until the suspension turned into a clear solution. The light scattering of suspension was determined visually and with

instrumental measurement of scattered light intensity (DLS, Malvern Zetasizer Nano S). The kinetic solubility of TIO in MA was found to be 0.213 mg/mL (23.8 *m*/*m*%). The concentration of TIO in MA was set to 20 *m*/*m*% for the PE preparations.

#### 3.1.4. Preparation and Characterization of Pickering Emulsions

The concentration of emulsifiers in the water phase was set to 1 mg/mL and was kept constant for all experiments. Three different sizes of SNPs were used for PE formulation (20ET, 50ET, 100ET) and Tween80® surfactant (Tw80) (Polysorbate80 Acros Organics, Thermo Fisher Scientific, Waltham, MA, USA) for conventional emulsions (CE). The concentration of oil phase varied between 0.90 and 17.91 mg/mL, and the ratio of TIO and MA was always constant (20 *m*/*m*%).

The emulsification was performed in two steps. The coarse emulsions were prepared by sonication for 2 min (Bandelin Sonorex RK 52H, BANDELIN electronic GmbH & Co. KG, Berlin, Germany). The final emulsification was performed with UltraTurrax (IKA Werke T-25 basic, IKA Werke GmbH, Staufen im Breisgau, Germany) for 2 min at 13,500 rpm. The emulsions' droplet size was determined with DLS using a Malvern Zetaziser Nano S instrument (Malvern Panalytical Ltd., Great Malvern, Worcestershire, UK). The stability of the emulsions was determined from periodical droplet size determination. The emulsions were stored at room temperature (25 ◦C).

The type of emulsions was determined with conductivity test using Mettler Toledo Seven2Go S3 conductivity meter (Mettler Toledo GmbH, Giessen, Germany) and InLab® 738-ISM sensor (Mettler Toledo GmbH, Giessen, Germany).

All experiments, measurements, and standard deviation calculations were performed from 3 parallel sample preparations.

#### 3.1.5. In Vitro Diffusion Studies—Static Franz Diffusion Cell Method

Accepted models for testing drugs and their formulations for onychomycosis treatments include penetration tests through cadaver nails [41], nail clippings, bovine hoof slices, or keratin films [42] made from human keratin source. The non-uniformity of natural membranes causes huge inhomogeneity in the results [43–45], which makes the comparison of different formulations impossible. The aim of our study was to examine the diffusion properties of applied drugs in complex colloidal systems; therefore, in our opinion, the similarity in hydrophilicity and surface charge between the nail plate or nail bed and artificial membranes was of the highest importance for testing and comparison of the formulations. The nail plate acts as a negatively charged aqueous hydrogel, as it is described in the literature [46], and it has properties similar to that of the agar gel [47]. Based on the literature data obtained from independent researches, we compared the diffusion coefficient and flux of well-studied antibiotic chloramphenicol (5 mg/mL in phosphate-buffered saline) with different membranes, namely agar gel [48], bovine hoof slice, and cadaver nail plate [49]. We found that the diffusion coefficients and flux values were very similar for agar gel and bovine hoof slice membranes. Flux for bovine hoof was 4.07 <sup>±</sup> 1.18·10−<sup>6</sup> mg/cm2·s, for agar gel 1.96 <sup>±</sup> 0.47 10−<sup>6</sup> mg/cm2·s, and 8.21 <sup>±</sup> 2.11 10−<sup>7</sup> mg/cm2·s for the cadaver nail plate. The flux values for agar gel were closer to the value for the cadaver nail plate, which might suggest that agar gel is a good model membrane for water-based formulations.

The agar gel membrane was used to model the nail plate. The composite membrane, consisting of the agar gel layer on top of the cellulose acetate membrane, was used to simulate the complex structure of nail plate and nail bed since nail bed has similar properties as skin [50,51], and the cellulose membrane has been commonly used as a model for skin permeability [52]. The main aim of the study on two types of membranes was to examine whether the examined formulations could deliver the applied lipophilic drugs through agar gel as a model for nail plate, and the composite membrane was used to examine if the formulation could retain the drugs on the main site of the infection, namely nail bed. The amount of the drug transported through the membrane was calculated based on the amount introduced to the membrane. In the case of agar membrane, the goal was to prepare the drug delivery system that could deliver the highest drug amount through that membrane. The composite membrane was used to test the on-site retention of drugs in different formulations. The amount of retained drug was calculated as a difference between the drug amount passed through the agar gel membrane and the amount passed through the composite membrane

Forin vitro testing, the 2.1 mm thick 6 *m*/*m*% agar gel membrane (Agar powder, purity >95%, VWR Chemicals Ltd., Debrecen, Hungary) and the same agar gel membrane combined with 0.8 mm thick cellulose acetate with effective penetration area of 2.54 cm<sup>2</sup> (Membranfilter Porafil, Macherey-Nagel GmbH&Co. KG, Düren, Germany, pore size 0.2 μm) were used. Before each measurement, the agar gel was always freshly prepared. The agar powder was dispersed in demineralized water, and the mixture was boiled in a closed vial for 3 min until all agar was completely dissolved. Exactly 10 mL of agar gel was poured into a plastic vessel (i.d. 70.8 mm), then left to cool (25 ◦C) and gelate. After the gelation, the agar gel was soaked in PBS buffer for 12 h. Finally, the agar membrane was cut out with a sharp home-made tool and placed on the Franz cell. The cellulose acetate membranes were also freshly soaked in PBS buffer before the experiments.

The examination of diffusion properties was performed at 32 ◦C in static vertical Franz diffusion cells (Hanson Microette Plus, Hanson Research 60-301-106, Hanson Research Corporation, Chatsworth, CA, USA); six parallel cells with effective penetration area 2.54 cm<sup>2</sup> were used, and each experiment was made in triplicates. The volume of the receiver chamber was 7 mL; the receiver solution was PBS buffer. For PBS preparation, the following salts were used: NaCl (high purity, VWR Chemicals Ltd., Debrecen Hungary), KCl (purity 99–100.5%, VWR Chemicals Ltd., Debrecen Hungary), Na2HPO4·2H2O (AnalaR NORMAPUR®, purity <sup>≥</sup>99.0%, VWR Chemicals Ltd., Debrecen Hungary), and KH2PO4 (purity ≥99.0%, VWR Chemicals Ltd., Debrecen Hungary). The 600 μL volume of emulsion or solution sample was placed into the donor chamber, and the diffusion was examined for 2 h; the stirring rate was 700 min<sup>−</sup>1, and the samples were collected after 5, 10, 15, 30, 60, 90, and 120 min. The withdrawn sample volume was replaced with a fresh PBS buffer.

The TIO content was determined with HPLC measurements using UV-Vis detector (SPD 10-A, Shimadzu Europa GmbH, Duisburg, Germany); the method is based on Bagary et al. [53]. Separations were carried out using a monolithic silica type column (ODS-AM302, S-5μm, 120A, YMC Co., Kyoto, Japan). The mobile phase consisted of methanol/0.02 M K2HPO4 = 85/15 *V*/*V*% and 0.2 *V*/*V*% trimethylamine (methanol dehydrated, ultrapure ≥99.8%, VWR Chemicals Ltd., Debrecen Hungary; trimethylamine: HiPerSolv CHROMANORM®, VWR Chemicals Ltd., Debrecen Hungary), pH = 7.0. The mobile phase was freshly filtered through Millipore Nylon membrane (pore size: 0.2 μm, Merck KGaA, Darmstadt, Germany) before the analysis. Isocratic elution was programmed with a 1.5 mL/min flow rate; the temperature of measurement was 32 ◦C. The detection wavelength of tioconazole was 254 nm, and its retention time was 3.5 min.

#### *3.2. Microbiological Tests against Candida albicans and Trichophyton rubrum*

#### 3.2.1. Instruments Used in the Microbiological Experiments

UV-Vis spectrophotometer (Hitachi U-3900, Hitachi High-Tech Corporation, Japan), microbiological incubator (Thermo Scientific Heraeus B12, Thermo Fischer Scientific, Waltham, MA, USA), Bürker cell counting chamber (Hirschmann Laborgeräte GmbH & Co., Germany), Multiskan EX 355 (Thermo Fischer Scientific, Waltham, MA, USA) spectrophotometer were used throughout the experiments.

#### 3.2.2. Materials Used in the Microbiological Experiments

For the microbiological experiments, the following materials were used: sterile 96-well microtiter plates (Greiner Bio-One, Kremsmunster, Austria), potato dextrose agar (PDA) (BioLab, Budapest, Hungary), sterile filter inserts (pore size 10 μm) from PluriSelect (pluriSelect Life Science, Leipzig, Germany), dextrose, adenine, bacteriological peptone and agar-agar (Reanal Labor, Budapest, Hungary), sterile centrifuge tubes (TPP Techno Plastic Products, Trasadingen, Switzerland), homemade Sabouraud dextrose agar or SDA (containing 4% dextrose, 1% bacteriological peptone, and 1.5% agar-agar in double-distilled water), yeast extract peptone dextrose agar (containing 2% bacteriological peptone, 1% yeast extract, 2% dextrose, and 1.5% agar-agar in double-distilled water), 3-(*N*-Morpholino)-propanesulfonic acid (MOPS) from Serva Electrophoresis GmBH (Heidelberg, Germany), and RPMI 1640 medium (containing 3.4% MOPS, 1.8% dextrose, and 0.002% adenine) from Sigma-Aldrich Chemie GmBH (Steinheim, Germany). Highly purified water (<1.0 μS) was applied throughout the experiments.

#### 3.2.3. Fungal Cultures and Inoculum Preparation

*Trichophyton rubrum* (*T. rubrum*) DSM 21146 and *Candida albicans* (*C. albicans*) ATCC 001 were obtained from Leibniz Institute DSMZ GmbH (Braunschweig, Germany) and from Department of General and Environmental Microbiology (Institute of Biology, University of Pécs, Hungary), respectively.

We followed the methods described previously [54–57] for *T. rubrum* and *C. albicans* culture preparation. In brief, *T. rubrum* stock inoculum suspensions were prepared from 7-day old cultures grown on PDA at 28 ◦C for sporulation. Ten days later, the observed fungal colonies were flooded with 10 mL distilled water, followed by scraping the surface using a sterile loop. Conidia and hyphal mixed suspensions were withdrawn and were transferred to a sterile centrifuge tube through sterile filter inserts (10 μm, pluriSelect) to remove hyphae, leaving a filtered inoculum containing spores only. The inoculum cell population was adjusted to 0.5 to 5 <sup>×</sup> 106 spores/mL visually using a Bürker cell counting chamber, followed by further turbidity calibration with a UV-Vis spectrophotometer (Hitachi U-3900) at 520 nm. The spores were further diluted to the desired population according to the experimental requirements.

*C. albicans* stock inoculum was prepared from 48 h old culture grown on YEPD agar plates at 30 ◦C. After 18 h of incubation at 30 ◦C in a microbiological incubator, on YEPD agar slant, the cells were looped out, diluted with 0.9% sterile saline, and were counted by a Bürker cell counting chamber, followed by turbidity calibration with a UV-Vis spectrophotometer (Hitachi U-3900) at 595 nm. The fungal cell population was set to ~1 <sup>×</sup> <sup>10</sup><sup>6</sup> cells/mL and was diluted later according to the experimental designs.

#### 3.2.4. Determination of Antifungal Activities

For the evaluation of the minimum inhibitory concentration (MIC) of *T. rubrum* and *C. albicans*, we followed previously published methods [56–59]. The ethanolic solutions of TIO and MA in a wide concentration range (0.5–300 μg/mL) were used for the assay. CE and PEs formulations were also tested; an initial concentration of the oil phase was 160 μg/mL for *T. rubrum*, whereas 180 μg/mL for *C. albicans* treatment was applied. The treating mixtures were further diluted up to 256 times in a serial half-dilution format.

One hundred microliters of fungal cell suspensions (see Sections 3.2.5 and 3.2.6) with equal fungal contents were applied thereafter to the microplate wells containing 100 μL of the different samples. Detailed information on the assay conditions can be found in Sections 3.2.5 and 3.2.6. As a blank, suspensions of 20ET, 50ET, 100ET SNPs, pure ethanol, Tw80 solution were used.

#### 3.2.5. Determination of Minimum Inhibitory Concentration of *T. rubrum*

The *T. rubrum,* inoculum size of ~2.5 <sup>×</sup> 104 spores/mL, containing the test drugs in half-dilution format, was incubated in RPMI media for 7 days in a microbiological incubator at 28 ◦C. The microplates containing *T. rubrum* incubated for 7 days with the test drugs were evaluated following the protocol as recommended by the Clinical and Laboratory Standards Institute (CLSI) M38-A2. The untreated cell samples and the medium without cells were considered as the growth control and blank, respectively. The endpoint determination readings for the minimum inhibitory concentrations (MIC) were performed visually based on the comparison of the growth in the wells containing the test drugs with that of the growth control [60]. All evaluations were performed in triplicates in six independent experiments.

#### 3.2.6. Determination of Minimum Inhibitory Concentration of *C. albicans*

A population size of ~2 <sup>×</sup> <sup>10</sup><sup>3</sup> CFU/mL was incubated in RPMI media with the above-mentioned test drug concentration range at 30 ◦C for 48 h in the case of *C. albicans*. A Multiskan EX 355 spectrophotometer was used to measure the absorbance (at 595 nm) of the samples in the microtiter plate in the case of *C. albicans*. The absorbance values of the respective treatments were converted to a percentage and were compared to growth control (100%). The untreated fungal samples and the medium without cells were considered as the growth control and blank, respectively. All evaluations were performed in triplicates in six independent experiments.

#### 3.2.7. Determination of the Minimum Fungicidal Concentration (MFC)

Determination of MFC was performed using the methods as described earlier with modifications [56]. After performing the MIC, 10 μL of the content from each well (not visibly turbid) was inoculated onto sterile SDA plates. The plates were incubated at 30 ◦C for 48 h. MFC was evaluated as the lowest drug concentration, resulting in no growth (≥99.9% growth inhibition). Measurements were performed by applying three technical replicates in six independent experiments.

#### 3.2.8. Statistical Analyses

The statistical analyses were conducted using a one-way ANOVA test (Origin 2016, OriginLab Corp., Northampton, MA, USA), and the significance was set at *p* ≤ 0.05.

#### **4. Conclusions**

The choice of drugs used in this research was based on careful consideration. The TIO was chosen as a drug with high antifungal activity but low water solubility and permeability through the nail plate. MA EO was selected because of the antifungal activity and because it is a liquid and can be used as a solvent for TIO. The combination of the drugs applied in this study showed a significant synergistic effect. The solution of TIO in MA EO was successfully formulated into stable Pickering emulsions. In vitro studies have demonstrated that PEs are effective drug formulations that can provide site-specific and effective drug delivery through artificial membranes. 20ET PE achieved the highest drug delivery efficiency as it could deliver 40% of the drug introduced to the artificial membrane within 10 min. The amount delivered at this time was 572 μg of TIO through the agar model membrane, while the MFC of the TIO in this formulation was 4.69 μg/mL. To prove the real applicability of the suggested drug combination and PE formulation, we have to perform experiments on the natural nail model. Still, from the presented data, we can conclude that the application of both site-specific drug delivery and synergistic antifungal drug combinations is a promising route for the development of effective onychomycosis topical treatment formulation.

**Supplementary Materials:** The following are available online: Table S1. Parameters of hydrophilic and surfacemodified silica nanoparticle synthesis. Table S2. HSPs of tioconazole and the three main components of *Melaleuca alternifolia* essential oil. Table S3. Calculated solubility parameters of tioconazole compared to the three main components of *Melaleuca alternifolia* essential oil. Table S4. Calculated solubility parameters.

**Author Contributions:** Conceptualization A.S. (Aleksandar Széchenyi), B.V.-H., S.P.; methodology B.V.-H., S.N., S.D., T.K.; formal analysis, B.V.-H.; investigation, B.V.-H., S.D., S.N., A.S. (Ala' Salem), A.B.; resources A.S. (Ala' Salem), T.K.; data curation A.S. (Ala' Salem); writing—original draft preparation, B.V.-H., A.S. (Aleksandar Széchenyi); writing—review and editing A.S. (Aleksandar Széchenyi); visualization, B.V.-H.; supervision, A.S. (Ala' Salem); funding acquisition A.S. (Ala' Salem), S.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** The project has been supported by the European Union, co-financed by the European Social Fund grant No. EFOP-3.6.1.-16-2016-00004 (Comprehensive Development for Implementing Smart Specialization Strategies at the University of Pécs). Transmission electron microscopy studies were performed using the JEOL-1400 TEM electron microscope, which was funded by the grant GINOP-2.3.3-15-2016-0002 (New generation electron microscope: 3D ultrastructure). The work was also funded by the grant KA-2018-17, University of Pécs, Medical School.

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

#### **References**


**Sample Availability:** Samples of the Pickering emulsions are available from the authors.

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 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/).

**Sakkarapalayam M. Mahalingam 1, Karson S. Putt 2, Madduri Srinivasarao <sup>3</sup> and Philip S. Low 2,3,\***


**\*** Correspondence: plow@purdue.edu; Tel.: +1-765-494-5272

**Abstract:** The inadvertent severing of a ureter during surgery occurs in as many as 4.5% of colorectal surgeries. To help prevent this issue, several near-infrared (NIR) dyes have been developed to assist surgeons with identifying ureter location. However, the majority of these dyes exhibit at least some issue that precludes their widespread usage such as high levels of uptake in other tissues, overlapping emission wavelengths with other NIR dyes used for other fluorescence-guided surgeries, and/or rapid excretion times through the ureters. To overcome these limitations, we have synthesized and characterized the spectral properties and biodistribution of a new series of PEGylated UreterGlow derivatives. The most promising dye, UreterGlow-11 was shown to almost exclusively excrete through the kidneys/ureters with detectable fluorescence observed for at least 12 h. Additionally, while the excitation wavelength is similar to that of other NIR dyes used for cancer resections, the emission is shifted by ~30 nm allowing for discrimination between the different fluorescenceguided surgery probes. In conclusion, these new UreterGlow dyes show promising optical and biodistribution characteristics and are good candidates for translation into the clinic.

**Keywords:** ureter imaging; fluorescence-guided surgery; near-infrared dye; PEG pharmacokinetics

#### **1. Introduction**

Because ureters are not commonly visible on visceral surfaces, their localization during abdominal surgery can be problematic, leading to accidental severance of the ureter in as many as 4.5% of colorectal surgeries [1] and 0.3% of all gynecological procedures [2,3]. To prevent the resulting leakage of urine into the peritoneum and the ensuing long-term complications [1,4,5], any cleaved ureter must be immediately religated by a time-consuming, complicated, and expensive procedure, thereby dramatically increasing the cost and complexity of the surgery. Not surprisingly, considerable effort has been focused on the development of methods to prevent ureter injuries during surgery.

One of the earliest approaches to avoid accidental ureter cleavage was to insert a stent into the ureter that would rigidify the duct and render it detectable by palpation [6,7]. However, because the process of stent insertion was found to cause occasional injury [1,8,9] and since physical palpation was not possible during robotic or endoscopic surgeries, the stent insertion strategy never attracted significant usage. Systemically administered near-infrared (NIR) fluorescent dyes such as indocyanine green (ICG) [10] and Ureter-Glow [11] were then explored for similar intraoperative ureter visualization, but these initial fluorescent dyes were found to clear primarily through the liver, bile duct, and intestines [10,11], creating high background fluorescence that could mask the location of proximal ureters. While much brighter fluorescent signals have been achieved by intraureter dye injection [12–14], the injection process has been considered by many surgeons to be too involved for routine ureter localization, leading to similar problems with widespread adoption [15]. Finally, although a zwitterionic near-infrared (NIR) fluorescent dye has

**Citation:** Mahalingam, S.M.;

Putt, K.S.; Srinivasarao, M.; Low, P.S. Design of a Near Infrared Fluorescent Ureter Imaging Agent for Prevention of Ureter Damage during Abdominal Surgeries. *Molecules* **2021**, *26*, 3739. https://doi.org/10.3390/molecules 26123739

Academic Editors: Kirill A. Afonin

Received: 29 May 2021 Accepted: 18 June 2021 Published: 19 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

been recently designed to excrete predominantly through the ureter [16], its transit time in ureters has been found to be very brief, and its fluorescence excitation and emission unnecessarily overlap with many tumor-targeted NIR dyes (e.g., IR800CW [17,18], LS288 [19,20], ICG [10,12–15], and OTL38 [11], Table 1), creating a potential discrimination problem when malignant lesions reside near a ureter.

**Table 1.** Photo-physical properties of NIR dyes used for fluorescence-guided surgeries and new PEGylated probes.


In an effort to create a ureter imaging agent with (1) reduced fluorescence in healthy tissues (except ureters), (2) prolonged transit through the ureters, and (3) an emission spectrum distinct from that of commonly used tumor-targeted fluorescent dyes, we have conjugated a longer wavelength NIR dye to a series of polyethylene glycol (PEG) oligomers of different lengths and examined their clearance following intravenous injection into mice. We report that conjugation of the NIR dye S0456 via a thioether bridge to PEG oligomers of 11–45 oxyethylene units yields ureter imaging agents with little uptake in healthy tissues, prolonged excretion almost exclusively through the ureters, and facile ureter visualization at an emission maximum that is easily distinguished from the common tumor-targeted NIR dyes.

#### **2. Methods**

#### *2.1. Materials*

S0456 was purchased from Few Chemicals (Bitterfeld-Wolfen, Germany). 2-(1*H*-7- Azabenzotriazol-1-yl)-1,1,3,3-tetramethyl uronium hexafluorophosphate methanaminium (HATU) was obtained from Genscript Inc. (Piscataway, NJ, USA). The PEG3, PEG11, and PEG2000 oligomers were purchased from TCI America (Portland, OR, USA), BroadPharm (San Diego, CA, USA), and Laysan Bio (Arab, AL, USA), respectively. Diisopropylethylamine (DIPEA), dimethyl sulfoxide (DMSO), and all other chemical reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA). Tubes, pipette tips, microtiter plates, and all other consumables were purchased from Fisher Scientific (Waltham, MA, USA).

#### *2.2. Synthesis of PEGylated S0456 NIR Dyes*

The S0456 near-infrared dye (1 equiv) was reacted with 2-(4-mercaptophenyl) acetic acid (1.1 equiv) in 2 mL DMSO for 12 h (90% yield, 95% purity). The resulting compound (1 equiv) was coupled to one of three different PEG linkers (1.1 equiv) of sizes 3, 11, and ~45 (PEG2000) in the presence of HATU (1.1 equiv) and DIPEA (5 equiv) in 5 mL DMSO for 6 h to yield the final PEGylated S0456 dyes.

The crude product was purified by preparative reverse-phase high-performance liquid chromatography using a mobile phase of 20 mM ammonium acetate buffer and gradient of 5 to 80% acetonitrile over 30 min (xTerra C18; Waters; 10 μm; 19 × 250 mm). Elution of the conjugate was monitored at 280 nm, and identities of eluted compounds were analyzed by liquid chromatography-mass spectrometry (LC-MS) (see Supplementary Materials). The molecular weight of UreterGlow-0 was calculated (calcd) for [M + H]+

(C46H54N2O14S5): 1019.2, found 1019.3. UreterGlow-3 calcd. (C54H72N4O16S5): 1193.4, found 1193.6. UreterGlow-11 calcd (C54H72N4O16S5): 1545.9, found 1545.9. UreterGlow-45 was synthesized using commercially available PEG2000 which is comprised of PEG oligomers of various lengths that should be centered around a PEG chain length of 45. Calcd. for PEG45 length (C138H240N4O58S5): 3043.7, found 3001.3, which corresponds to the most prominent PEG chain length centered around 44.

#### *2.3. Characterization of Spectral Properties*

The excitation and emission wavelengths of the various dyes (1 μM in PBS) were scanned using a Cary Eclipse fluorimeter (Agilent, Santa Clara, CA, USA). The wavelength which resulted in the maximum excitation and emission values for each dye was determined.

#### *2.4. Animal Husbandry*

ND4 Swiss Webster mice (Harlan Laboratories, Indianapolis, IN, USA) were maintained on normal rodent chow and housed in a sterile environment on a standard 12 h light and dark cycle for the duration of the study. All animal procedures were approved by the Purdue Animal Care and Use Committee in accordance with NIH guidelines (protocol #1111000316 approved 2 February 2021).

#### *2.5. In Vivo Biodistribution*

Mice were injected via tail vein with 10 nmol of a fluorescent dye conjugate. Mice were sacrificed at 2, 4, and 6 h post-injection, and organs were removed (*n* = 1 per timepoint per conjugate for the initial UreterGlow-0, -3, -11, and -45 biodistribution studies and *n* = 3 per timepoint per conjugate for the UreterGlow-11, IR800CW, IR800BK, and ZW800-1 biodistribution studies). For urine analysis, mice were administered 10 nmol UreterGlow-11 via tail vein injection, sacrificed at various timepoints, and urine was removed from the bladder via syringe. The organs and urine were imaged using a Caliper IVIS Lumina II Imaging Station (PerkinElmer, Waltham, MA, USA) coupled with an ISOON5160 Andor Nikon camera equipped with Living Image Software Version 4.0 (PerkinElmer, Waltham, MA, USA). The settings were as follows: lamp level, high; excitation, 745 nm; emission, ICG; epi illumination; binning (M) 4; FOV, 12.5; f-stop, 4; acquisition time, 1 s.

#### *2.6. Effect of pH on UreterGlow-11 Emission Spectra*

UreterGlow-11 (1 μM) was added to PBS (pH 7.4), freshly collected human urine (pH 5.5) or sodium carbonate buffered saline with pH values ranging from 2.5 to 10. The dye was excited using 800 nm light and the emission spectrum was obtained using a Cary Eclipse fluorimeter (Agilent, Santa Clara, CA, USA).

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

#### *3.1. Design and Synthesis of the Ureter Probes*

In an effort to remedy the deficiencies of current ureter imaging agents, we undertook to design a water-soluble NIR dye that would (i) excrete for several hours primarily through the ureters, (ii) avoid uptake by normal tissues, (iii) excite with the same light source used for visualization of tumor-targeted fluorescent dyes, and (iv) emit at a longer wavelength than the tumor-targeted fluorescent dyes; i.e., to allow discrimination of the tumor from ureter fluorescence. Because PEGylation can prolong circulation times and reduce nonspecific uptake by healthy tissues [21–23], we synthesized a series of optical probes comprised of PEG oligomers of different lengths linked to the cyanine dye, S0456, via a thioether bond to 4-mercaptophenylacetic acid (see Methods). As summarized in Scheme 1, PEG oligomers containing 0, 3, 11, and ~45 oxyethylene units were conjugated to the modified S0456 dye and designated as UreterGlow-0, -3, -11, -45. All four conjugates were purified using preparative reverse-phase HPLC and then characterized by LC-MS (see Supplementary Materials). With sufficient quantities synthesized and purities of >95%

achieved for all conjugates, characterization of their physical and biological properties could commence.

**Scheme 1.** Synthesis of PEGylated NIR dyes for improving renal clearance. (**a**) DMSO, RT, 12 h. (**b**) HATU (1.1 equiv), DIPEA (5 equiv), DMSO, RT, 6 h.

#### *3.2. Characterization of Physical Properties*

Following synthesis of the desired conjugates, molecular weights were confirmed by mass spectrometry, and excitation and emission spectra were obtained using a fluorescence spectrophotometer. As shown in Table 1 and Figure 1, use of a thioether in these UreterGlow conjugates instead of an oxoether bridge connecting the S0456 dye to a phenyl ring shifted the excitation maximum of the conjugate from 776 nm (i.e., similar to OTL38 and many other tumor-targeted fluorescent dyes) to 800 nm. The corresponding emission maxima also shifted from 796 nm to 830 nm, respectively. Because the excitation spectra of both the oxo- and thioether bridged S0456 dyes, as well as the other major NIR dyes used for tumor imaging (IR800CW, LS288, ZW800-1, ICG, and OTL38) overlap over most of their excitation spectra (Figure 1), all of the above NIR dyes should be excitable with the same light source; i.e., avoiding the need to change light sources or cameras to image cancer tissues and ureters simultaneously. Moreover, because the emission spectra of the thioether dyes are shifted ~30–40 nm to longer wavelengths from the major tumor-imaging dyes (Table 1 and Figure 1), it should be possible to display tumor tissue and ureters in different colors on any imaging monitor [24]. Based on these considerations, we expect our ureter probes to function well in combination with most tumor-targeted fluorescent dyes to help prevent accidental ureter damage during abdominal surgeries.

**Figure 1.** Excitation and emission spectra of various NIR dyes for fluorescence-guided surgery applications. Dyes (1 μM in PBS, pH 7.4) were excited at 680 nm and their emissions were scanned from 700 to 900 nm. Alternatively, excitation wavelengths were scanned from 600 nm to 850 or 900 nm while the emission wavelength was set at 900 nm followed by normalization of their intensities.

#### *3.3. In Vivo Imaging and Biodistribution*

To test the hypothesis that a PEG linker of the appropriate length can reduce healthy tissue uptake while prolonging passage of the conjugate through the ureter, the aforementioned PEGylated probes were injected via tail vein into live mice and allowed to circulate for different lengths of time (i.e., 2, 4, and 6 h) before euthanasia and analysis of tissue fluorescence (*n* = 1 for each timepoint and each conjugate). As shown in Figures 2 and 3, UreterGlow-0 showed significant uptake in all major organs except the heart, spleen, and lungs, demonstrating that UreterGlow-0 would not function well for ureter imaging. However, as the length of the appended PEG chain was increased, healthy tissue retention decreased at all time points, with minimal if any healthy tissue fluorescence of UreterGlow-11 and -45 detected at the 2 h time point and no significant healthy tissue fluorescence observed at any subsequent time points. These data demonstrate that the longer PEG chains suppress uptake of the UreterGlow conjugates by healthy tissues, and that their capture by the liver and subsequent excretion via the bile duct into the intestines is also suppressed by longer PEGylation. Because compounds not excreted via the liver/bile duct must excrete through the kidneys, this redirection of UreterGlow-11 and -45 to clearance through the kidneys should enhance and prolong their flow through the ureters. However, because both UreterGlow-11 and -45 performed similarly, UreterGlow-11 was employed in all further studies because it could be synthesized as a homogeneous molecular species.

To compare the properties of UreterGlow-11 with other NIR dyes previously examined for ureter imaging, mice (*n* = 3 per time point per conjugate) were intravenously injected with UreterGlow-11, IR800CW, IR800BK, or ZW800-1) [16,25] and sacrificed 2, 4, or 6 h after injection prior to analysis of tissue-retained fluorescence. As shown in Figure 4, UreterGlow-11 showed little or no uptake in any tissues except the kidneys at all time points examined, suggesting its signal to background contrast along the urinary tract should be very high. In contrast, all other dyes investigated displayed significant accumulation in healthy organs, likely due to their partial excretion through the liver, bile duct and intestines and/or nonspecific retention by an unknown process in these tissues. These nonspecific uptake properties could be troublesome during fluorescence-guided surgeries of metastatic cancers since the latter dyes are also commonly used in fluorescent probes for imaging malignant lesions.

**Figure 2.** In vivo biodistribution of UreterGlow conjugates. Mice were administered 10 nmol of various conjugates (*n* = 1 per time point per conjugate) via tail vein injection. After varying times, mice were euthanized and their organs were removed. Organs were imaged and the fluorescence intensity recorded.

Although the small size of murine ureters rendered them difficult to image, because any dye that appears in the urine will have recently passed through the ureters, we collected urine samples at different times, post-intravenous injection, and measured their fluorescence intensities in order to confirm that UreterGlow-11 could provide strong ureter fluorescence for prolonged periods following administration. As shown in Figure 5, urine fluorescence remained high for at least 12 h after UreterGlow-11 infusion and then only gradually declined over the subsequent 12 h. These data suggest that UreterGlow-11 should illuminate ureters well, even during protracted abdominal surgeries.

**Figure 3.** Quantitation of in vivo biodistribution of UreterGlow conjugates. Mice were administered 10 nmol of various conjugates (*n* = 1 per time point per conjugate) via tail vein injection. After varying times, mice were euthanized and their organs were removed. Organs were imaged, fluorescence intensity was recorded, and relative fluorescence was plotted.

**Figure 4.** Comparison of in vivo biodistribution of UreterGlow-11 and other NIR ureter imaging dyes/conjugates. Mice were administered 10 nmol of various conjugates (*n* = 3 per time point per conjugate) via tail vein injection. After varying times, mice were euthanized, their organs were removed, fluorescence intensity was imaged.

**Figure 5.** Quantitation of urine fluorescence after administration of UreterGlow-11. Mice were administered 10 nmol of various conjugates (*n* = 1 per time point) via tail vein injection. After varying times, mice were euthanized, and urine was collected from their bladders. The fluorescence of the isolated urine was quantitated.

Finally, because urine pH can vary from pH 4.5 to pH 8 [26], it was important to ensure that the UreterGlow-45 fluorescence did not vary with urine pH. As shown in Figure 6A, the emission spectrum of UreterGlow-45 was independent of pH between 2.5 and 10 and also showed no impact when dissolved in urine (Figure 6B). Taken together, these data collectively suggest that UreterGlow-11 should perform well as a ureter imaging agent during abdominal surgeries for cancer.

**Figure 6.** Sensitivity of UreterGlow-45 emission spectra in different pH buffers. (**A**) UreterGlow-11 (1 μM) was excited at 800 nm while dissolved in sodium acetate buffered saline at various pH levels, and the emission spectra were characterized. (**B**) UreterGlow-11 (1 μM) was excited at 800 nm while dissolved in PBS (pH 7.4) or human urine (pH 5.5), and the emission spectra were characterized.

#### **4. Conclusions**

Although this brief report described only the impact of two compositional variables on the properties of a NIR dye for intra-operative ureter imaging, many other modifications could also have been explored for further optimization. Thus, NIR dyes with other excitation and emission wavelengths could have been generated by the insertion of other heteroatoms at other locations in the UreterGlow conjugate. PEGs of intermediate lengths between 11 and 45 oxyethylene units could also have been examined for improved biodistribution and pharmacokinetic properties. And finally, a targeting ligand could have been designed that would enable sustained binding of the fluorescent conjugate to the epithelial cells lining the ureters. Thus, while the above improvements in ureter specificity, emission wavelength, and transit time through the ureters now renders UreterGlow-11 a good candidate for translation into the clinic, opportunities may remain for further optimization with an eventual goal of totally eliminating accidental damage to ureters during abdominal surgeries.

**Supplementary Materials:** The following are available online, Figure S1: Chemical Structures of Selected Dyes Used in Fluorescence-Guided Surgeries; Figure S2: Structure and LC-MS Characterization of UreterGlow-0; Figure S3: Structure and LC-MS Characterization of UreterGlow-3; Figure S4: Structure and LC-MS Characterization of UreterGlow-11; Figure S5: Structure and LC-MS Characterization of UreterGlow-45.

**Author Contributions:** Conceptualization, M.S. and P.S.L.; methodology, S.M.M., M.S. and P.S.L.; formal analysis S.M.M. and K.S.P.; data curation, K.S.P., writing-original draft preparation K.S.P. and P.S.L.; writing review and editing; K.S.P., P.S.L., M.S., S.M.M.; supervision P.S.L.; project administration M.S. and P.S.L.; funding acquisition P.S.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors gratefully acknowledge support from the Purdue University Center for Cancer Research, P30CA023168, the Purdue Institute for Drug Discovery, and a grant from On Target Laboratories LLC.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Purdue Animal Care and Use Committee of Purdue University (protocol #1111000316 approved 2 February 2021).

**Data Availability Statement:** The data presented in this study are available in Putt KS. 2021. Supplemental Information Design of a near infrared fluorescent ureter imaging agent for prevention of ureter damage during abdominal surgeries; Zenodo http://doi.org/10.5281/zenodo.4987325.

**Conflicts of Interest:** The authors declare the following competing financial interest(s): This work was supported in part by a grant from On Target Laboratories. Philip Low is a cofounder and a member of the Board of Directors of on Target Laboratories.

**Sample Availability:** Samples of the compounds are not available from the authors.

#### **References**


## *Article* **Crosslinking of CD38 Receptors Triggers Apoptosis of Malignant B Cells**

**M. Tommy Gambles 1,2, Jiahui Li 1,2, Jiawei Wang 1,2, Douglas Sborov 3, Jiyuan Yang 1,2,\* and Jindˇrich Kopeˇcek 1,2,4,\***


**Abstract:** Recently, we designed an inventive paradigm in nanomedicine—drug-free macromolecular therapeutics (DFMT). The ability of DFMT to induce apoptosis is based on biorecognition at cell surface, and crosslinking of receptors without the participation of low molecular weight drugs. The system is composed of two nanoconjugates: a bispecific engager, antibody or Fab' fragment morpholino oligonucleotide (MORF1) conjugate; the second nanoconjugate is a multivalent effector, human serum albumin (HSA) decorated with multiple copies of complementary MORF2. Here, we intend to demonstrate that DFMT is a platform that will be effective on other receptors than previously validated CD20. We appraised the impact of daratumumab (DARA)- and isatuximab (ISA)-based DFMT to crosslink CD38 receptors on CD38+ lymphoma (Raji, Daudi) and multiple myeloma cells (RPMI 8226, ANBL-6). The biological properties of DFMTs were determined by flow cytometry, confocal fluorescence microscopy, reactive oxygen species determination, lysosomal enlargement, homotypic cell adhesion, and the hybridization of nanoconjugates. The data revealed that the level of apoptosis induction correlated with CD38 expression, the nanoconjugates meet at the cell surface, mitochondrial signaling pathway is strongly involved, insertion of a flexible spacer in the structure of the macromolecular effector enhances apoptosis, and simultaneous crosslinking of CD38 and CD20 receptors increases apoptosis.

**Keywords:** CD38; drug-free macromolecular therapeutics; human serum albumin conjugates; morpholino oligonucleotides; daratumumab; isatuximab; multiple myeloma; lymphoma

#### **1. Introduction**

The use of monoclonal antibodies (mAb) in the treatment of hematological malignancies has become an essential part of immunotherapy regimens [1]. Often mAb's are used in combination with small molecule chemotherapeutics to improve patient prognoses. Immunotherapy offers highly specific targeting to overexpressed cancer cell surface antigens. Once engaged with their target cell surface receptor, various mechanisms of action may occur to initiate cancer cell death. Immune effector cells can interact with the Fc domains leading to a variety of cell death events including complement-dependent cytotoxicity (CDC), antibody-dependent cellular cytotoxicity (ADCC), and antibody-dependent cellular phagocytosis (ADCP) [2]. Additionally, crosslinking of some receptor-bound antibodies leads to a clustering effect of surface receptors that triggers apoptotic mechanisms within the cell.

Crosslinking of cell-surface receptors has important biological consequences, including enhancing internalization of receptor-ligand complexes [3,4], changing the subcellular

**Citation:** Gambles, M.T.; Li, J.; Wang, J.; Sborov, D.; Yang, J.; Kopeˇcek, J. Crosslinking of CD38 Receptors Triggers Apoptosis of Malignant B Cells. *Molecules* **2021**, *26*, 4658. https://doi.org/10.3390/ molecules26154658

Academic Editors: Marina A. Dobrovolskaia and Kirill A. Afonin

Received: 4 June 2021 Accepted: 28 July 2021 Published: 31 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

fate of the receptor-ligand complex from recycling to the lysosomal route [5], and phosphatidylserine translocation and apoptosis initiation [6].

The CD38 receptor is substantially expressed on multiple myeloma cells and at low levels on normal lymphoid and myeloid cells. It is also expressed on lymphoma cells. In addition to its receptor function, CD38 is an ectoenzyme that cleaves nicotinamide dinucleotide phosphate (NADP) and nicotinamide adenine dinucleotide (NAD+) [7]. DARA (fully human IgG1-κ) [8] and ISA (chimeric IgG1-κ) [9] are two FDA approved antibodies for multiple myeloma (MM) treatment [7]. DARA's mechanisms of action include CDC, ADCC, and ADCP. In addition, FcγR-mediated crosslinking of tumor-bound DARA initiates cell death (Scheme 1) [6]. ISA's mechanism of action includes CDC, ADCC, and ADCP. Importantly, ISA has a strong apoptosis-inducing activity that is independent of crosslinking and inhibits the enzymatic activity of CD38 [10].

**Scheme 1.** CD38-targeting DFMT system depicted using DARA-MORF1 as the bispecific engager and HSA-(MORF2)y as the crosslinking effector molecule. DFMT is a two-step process involving step (1): specific antigen engagement by the bispecific engager followed by step (2): receptor crosslinking via the effector nanoconjugate. Crosslinking occurs via hybridization of complementary morpholino oligonucleotides conjugated to the bispecific engager and effector molecules.

Drug-free macromolecular therapeutics (DFMT) is a new paradigm for the treatment of malignancies [11–13]. Induction of apoptosis is mediated by receptor crosslinking facilitated by biorecognition of complementary peptide or oligonucleotide motifs; no low molecular weight drug is needed [11,12]. DFMT is comprised of two complementary nanoconjugates: (a) the bispecific engager: an antibody or Fab' fragment conjugate with morpholino oligonucleotide MORF1; and (b) the crosslinking effector: *N*- (2-hydroxypropyl)methacrylamide (HPMA) copolymer or human serum albumin (HSA) modified with multiple copies of complementary oligonucleotide MORF2. Hybridization of MORF1/MORF2—mediated receptor crosslinking initiates apoptosis (Scheme 1). We have demonstrated the efficacy of DMFT on CD20 positive (CD20+) Raji B cells in vitro [14,15], in vivo on a disseminated non-Hodgkin lymphoma (NHL) model in SCID mice [16,17], and on patient cells diagnosed with various blood borne malignancies [18]. Apoptosis induction by DMFT is triggered by relocation of crosslinked CD20 complexes to lipid rafts resulting in calcium influx, mitochondrial depolarization, and caspase 3 activation [19].

The linkers used in the DFMT nanoconjugations are based on a bifunctional PEG. Intact nanoconjugates, therefore, have inert linkers consisting of short PEG dimers with amide and thioether bonds on either terminus. All antibodies used in this work are FDA approved products which have overcome much scrutiny in terms of toxicity, pharmacokinetics and biocompatibility [20–22]. Human serum albumin is a ubiquitous protein in human plasma and is gaining more and more interest in drug delivery systems because of its long circulation half-life and non-immunogenicity [23]. The morpholino oligomer strands are DNA analogues that have had their backbone chemistry altered to allow for protease resistance in vivo. Numerous oligonucleotides have been evaluated in clinical trials and proven their biocompatibility [24]. A detailed study of the immune properties of the crosslinking effector HSA-(MORF2)x is planned using protocols we developed when evaluating the peptide containing crosslinking effector P-(CCK)x [25].

Rituximab (RTX) and other Type I antibodies dramatically improved treatment of CD20+ B-cell hematological malignancies [26]. However, a number of patients develop resistance and due to polymorphism of Fcγ receptors on immunocompetent cells the hypercrosslinking of RTX bound to CD20 is not efficient. This catalyzed the advance of Type II antibodies, such as Obinutuzumab (OBN), that do not need crosslinking. They induce apoptosis by actin rearrangement, lysosomal disruption, and homotypic cell adhesion [21].

Interestingly, when DFMT was applied to OBN, classified as a Type II antibody, a system that combines Type I and Type II mechanisms was developed [27]. The first nanoconjugate was OBN-MORF1 (OBN conjugated to one morpholino oligonucleotide MORF1); and the second nanoconjugate was HSA-(MORF2)y (HSA grafted with multiple copies of complementary morpholino oligonucleotide 2). Modification of OBN with one MORF1 does not impact the binding of OBN-MORF1 to CD20 and following binding to CD20 Type II effects occur. Further exposure to multivalent effector HSA-(MORF2)y results in crosslinking of CD20-OBN-MORF1 complexes, their clustering into lipid rafts and initiation of Type I effects. This new approach, called "clustered OBN (cOBN)" combines effects of both cell death-inducing mechanisms resulting in very high apoptotic levels [27].

Aiming to improve the efficacy of treatment of B cell malignancies in general and of multiple myeloma in particular, we evaluated the impact of crosslinking CD38 receptors on the mechanism and extent of apoptotic induction in four CD38 positive malignant B cells (Daudi, Raji, RPMI 8226, and ANBL-6). DFMT based on Fab'DARA-MORF1, Fab'ISA-MORF1, DARA-MORF1, and ISA-MORF1 as bispecific engagers and HSA-PEGx-(MORF2)y as multivalent crosslinking effector were evaluated. The ultimate goal of our studies is to establish the cell and nanoconjugate structure-dependent participation of Type I and Type II apoptotic mechanisms in CD38 crosslinking-mediated apoptosis induction.

#### **2. Results and Discussion**

#### *2.1. Nanoconjugates Synthesis and Cell Lines*

DARA and ISA nanoconjugates were synthesized using procedures previously reported by our group (Figure 1A) [27–29]. First, whole antibody was selectively reduced targeting the interchain disulfide bonds located in the hinge region. In parallel to the reduction reaction, the 3 -amine-functionalized MORF1 was reacted with maleimide-PEG2-NHS by aminolysis, resulting in maleimide-MORF1 intermediate. The latter was then coupled with freshly reduced whole antibody via thiol–ene click chemistry yielding the desired antibody-MORF1 nanoconjugate. The Fab' fragment MORF1 analogues were generated by first digesting whole antibody with pepsin resulting in dual chain cleavage below the disulfide hinge region generating the divalent F(ab')2 intermediate. F(ab')2 was then reduced to generate two equivalents of Fab' fragments which were further coupled via a thiol–ene click reaction with the maleimide-MORF1 intermediate yielding the desired Fab'-MORF1 nanoconjugate. The HSA-(PEG)x-(MORF2)y nanoconjugate was synthesized in a similar fashion utilizing the bifunctional reactivity of maleimide-(PEG)x-NHS (SM-PEG). Free lysine amine groups on the periphery of the HSA molecule were coupled with maleimide-PEGx-NHS (x = 2, 8, or 24) yielding the multivalent maleimide functionalized

HSA-PEGx-maleimide intermediate. The complementary morpholino, MORF2, was customized with an easily reducible disulfide bond on its 3 terminus. HSA-PEGx-maleimide was decorated with freshly reduced MORF2 molecules via thiol–ene maleimide click reactions. A valency greater than about 5 in HSA-(MORF2)y did not show significant increase in efficacy of CD20 receptor crosslinking [27]; so minor variations in valency (Figure 1C) should not have an impact on efficacy. Reaction intermediates and final nanoconjugates were characterized by size exclusion chromatography (SEC) (Figure 1B). Each nanoconjugate morpholino valency was characterized by UV-Vis spectrophotometry (Figure 1C) and BCA assay. The valence of MORF per macromolecule was calculated by (i) attaining the concentration of MORF in solution by UV-Vis spectrophotometry and (ii) determining concentration of protein in solution by BCA assay, then (iii) dividing the two concentrations yielding ratio of MORF per molecule. Hybridization between complementary nanoconjugates was confirmed by UV-Vis spectrophotometry by observing absorbance changes at λ = 260 nm of varying molar ratio solutions of MORF1:MORF2 (Figures S2 and S5); and size exclusion chromatography (Figures S1 and S4).

**Figure 1.** Nanoconjugate synthesis and characterization. (**A**) Synthetic route to produce antibody-MORF1, Fab'-MORF1 and HSA-(MORF2)y conjugates. Note: If not otherwise stated all conjugates contain a diethyleneglycol unit in the spacer inserted by SM-PEG2. Conjugates containing longer spacers are denoted as, e.g., Fab'-PEG8-MORF1. (**B**) Nanoconjugate size exclusion chromatography profiles detected on a Superdex 200 10/300 GL column, PBS (pH 7.4) as eluant at 0.4 mL/min flow rate. (**C**) Characterization (MORF valence) of nanoconjugates determined using UV-Vis absorbance at λ = 260 nm and BCA assay.

Four CD38 positive cell lines were used: Raji (Burkitt's lymphoma), Daudi (Burkitt's lymphoma), RPMI 8226 (multiple myeloma), and ANBL-6 (multiple myeloma). CD38 negative cell line U266 (multiple myeloma) was used as control. Level of CD38 expression was estimated by DARA binding to each cell line at 4 ◦C followed by exposure to a fluorescently labeled anti-human secondary antibody. The level of CD38 expression was Daudi > RPMI 8226 > Raji > ANBL-6 >> U266 (Figure 2A).

**Figure 2.** (**A**) DARA binding on various B cell lines as determined by secondary fluorescence of a Fluor 488-labeled antihuman goat antibody. (**B**) Apoptosis induced by DARA-MORF1 and Fab'Dara-MORF1 DFMT treatments on various CD38+ cell lines and one CD38- cell line: U266. Cell viability was measured by Annexin V and Propidium Iodide staining and analyzed by flow cytometry. (**C**) Confocal microscopy of fluorescently labeled nanoconjugates: Cy5-DARA-MORF1 (left, green); Cy3-HSA-(MORF2)10 (middle, red); overlay (right, yellow) of Cy5 and Cy3 channels on Daudi cells. (**D**) Effect of time lag (15, 30, and 60 min) between the administration of the bifunctional engager, DARA-MORF1 and the multivalent effector, HSA-(MORF2)10 on apoptosis initiation. (**E**,**F**) Effect of HSA PEG linker length on DFMT apoptosis efficacy on Daudi cells. (**E**) DARA-MORF1 + HSA-PEG2,8,24-(MORF2)y. (**F**) Fab'DARA-MORF1 + HSA-PEG2,8,24-(MORF2)y. Flow cytometry cell population distribution data for the HSA PEG linker length studies are also shown. All experiments were performed in triplicate. \*\*\*\* *p* < 0.0001, \*\*\* *p* < 0.001, \*\* *p* < 0.01, \* *p* < 0.05, n.s., not significant by One-Way ANOVA and Tukey test.

#### *2.2. DFMT Triggers Apoptosis in CD38+ Lymphoma and Myeloma Cell Lines by Consecutive Binding of Nanoconjugates*

To validate the hypothesis that crosslinking of CD38 directly initiates apoptosis, we evaluated the levels of apoptosis initiation in Daudi, Raji, RPMI 8226, ANBL-6, and U266 cell lines by exposing them to DARA-MORF1 or Fab'DARA-MORF1 (0.5 μM MORF1) for 1 h, followed (after washing and resuspending) to HSA-(MORF2)10 (0.5 μM MORF2) for 24 h. High levels of apoptosis were achieved in the three CD38+ cell lines (Daudi cells exhibited the highest levels) as well as in controls, premix and Daratumumab + sec. antibody. As expected, CD38- U266 cells exhibited negligible levels of apoptosis. Interestingly, percentage of apoptotic cells for the various cell types correlated with the level of CD38 expression observed in the binding studies (Figure 2A,B).

We next investigated the biorecognition of nanoconjugates at the cell surface employing confocal fluorescence microscopy. Consecutive exposure of Raji cells to Cy5-DARA-MORF1 resulted in cell surface green signal; exposure of decorated cells to HSA-(MORF2)10 showed red surface signal. Both signals were colocalized (yellow color) indicating successful biorecognition (hybridization) of MORF1/MORF2 at cell surface (Figure 2C).

DFMT is a two-step process: The first nanoconjugate a bispecific engager, DARA-MORF1 or Fab'DARA-MORF1, binds to CD38 and decorates the cell surface with MORF1 moieties. After a time lag, the second nanoconjugate, a multivalent macromolecular effector, HSA-PEGx-(MORF2)y, hybridizes and crosslinks multiple CD38 receptors resulting in apoptotic response. One important factor related to the efficacy of the process is the potential internalization of CD38 following binding with the bispecific engager. It is known that surface CD38 is internalized after receptor binding [30,31]. The internalization is gradual with time and crosslinking enhances the rate of internalization on the Jurkat cell line [30]. To validate the two-step pretargeting approach, we compared apoptosis induction for different time lags between cells' (Raji, Daudi, and RPMI 8226) exposure to the two nanoconjugates; the second nanoconjugate HSA-(MORF2)10 was administered after 15 min, 30 min, and 1 h after the administration of the bispecific engager (Figure 2D and Figure S7). Additionally, we exposed cells to a multivalent premix of both conjugates (control). In all three cell lines the length of the time lag had no impact on the level of apoptosis. Premixing nanoconjugates before cell exposure enhanced apoptotic levels when compared to two-step administration. The difference was largest in Raji cells and minor in Daudi and RPMI 8226 cells. This may be the effect of crosslinking enhanced internalization of the loaded CD38 receptor. The difference in apoptosis induction between premixed nanoconjugates and consecutive administration was minimal for the CD20 receptor [28], reflecting different internalization kinetics of CD20 vs. CD38 following receptor binding and crosslinking.

We described the advantages of the two-step administration previously, e.g., [32]. Importantly, a two-step approach permits pretargeting in vivo, a strategy commonly used in cancer radioimmunotherapy [33,34]. The experiments in this work were performed in vitro which makes the nanoconjugate premixture a meaningful control treatment group because hybridization is allowed to occur in an idealized setting and no washing step between treatments is needed. This provides a theoretical "maximum efficacy" for the in vitro experiments. For in vivo applications, one must consider important factors such as immune response, effector cell interactions and clearance and how each of these factors influence both the targeting of the system and the hybridization.

Pretargeting strategy (two-step treatment) permits the amplification of therapeutic efficacies and reduces adverse side reactions [35]. For example, in our previous work with DFMT targeted to CD20 we determined the time lag when the pretargeting agent (Fab'-MORF1) was mostly cleared from the blood and reached a steady plasma concentration, and, second, by determining the tumor targeting efficiency when using this time interval [14]. Results indicated a suitable timing for P-(MORF2)x administration at 5 h (in female SCID mice); at this time, Fab'-MORF1 was efficiently distributed to the tumors. Based on this result, we further performed therapy experiments in a disseminated B-NHL

mouse model. When the optimized pretargeting time lag (5 h) was used, the therapeutic efficacy was significantly better than that of identical experimental conditions but with a 1 h interval. A low dose (58 μg × 3) of Fab'-MORF1 with a 5× excess P-(MORF2)x resulted in significantly delayed tumor growth and substantially improved animal survival [14]. The optimized therapeutic system surpassed rituximab in anticancer efficacy and completely eradicated lymphoma B-cells in 83% of the animals. This pretargeting approach may constitute a novel personalized nanotherapy to enable more efficient treatment and limit potential side effects associated with off-target binding.

The decreased adverse (off-target) effects of two-step administration in vivo are based on these phenomena: CD20 is a very slowly internalizing receptor. When Fab'-MORF1 is administered, the part bound to CD20 remains at the surface, whereas the off-target bound part is internalized (and degraded in the lysosomes) during the time-lag before the administration of the crosslinking effector. Thus, the crosslinking effector administered following a time-lag, finds the bispecific engager bound just to the target (CD20).

#### *2.3. Impact of Spacer Length on Apoptosis Initiation*

Spacer length is an important factor in the efficiency of nanoconjugates. PEG spacers enhanced the efficacy of DMFT systems based on flexible HPMA copolymer molecules [28] as well as multivalent liposomes [36]. Since HSA has a relatively rigid structure, introduction of a flexible spacer between HSA and MORF2 should enhance biorecognition and apoptosis induction. We used succinimidyl-PEGx-maleimides, hetero-bifunctional crosslinkers with different numbers of repeating ethyleneglycol (EG) units to synthesize HSA-PEGx-MORF2 conjugates with variable spacer length. In particular, NHS-PEG2-maleimide (succinimidyl-[(*N*-maleimidopropionamido)-diethyleneglycol]ester), NHS-PEG8-maleimide (succinimidyl-[(*N*-maleimidopropionamido)-octaethyleneglycol] ester), and NHS-PEG24 maleimide (succinimidyl-[(*N*-maleimidopropionamido)-tetracosaethylene glycol] ester) (Thermo Scientific) were used for the synthesis of HSA-PEG2-(MORF2)10, HSA-PEG8- (MORF2)9, and HSA-PEG24-(MORF2)13. The characterization of conjugates is shown in Figure 1b (*right panel*). Apoptosis was determined on Raji cells by Annexin V/PI assay. Two bispecific engagers, DARA-MORF1 and Fab'DARA-MORF1 were employed. Data in Figure 2E,F show no statistically significant difference between spacers containing PEG2 (17.6 Å) and PEG8 (39.2 Å). Increasing the length of the spacer to PEG24 (95.2 Å) resulted in statistically significant enhancement of apoptosis induction. This is valid for both bispecific engagers used. It appears that a relatively rigid carrier, such as HSA, needs a longer spacer to enhance efficacy when compared to a flexible HPMA copolymer carrier. In the latter DFMT system (Fab'-MORF1 + HPMA copolymer-MORF2) a statistically significant enhancement of apoptosis was observed when increasing the spacer length form PEG2 to PEG8 [28].

#### *2.4. Prevention of Calcium Influx and Cholesterol Depletion from Lipid Rafts Lessen Apoptosis*

Two important features were observed when initiating CD20 mediated apoptosis by DFMT. Crosslinking of CD20 receptors in Raji cells resulted in rapid rise in the Ca++ intracellular concentration [19]. Additionally, extracting cholesterol from cell membranes by β-cyclodextrin (β-CD) impacted receptor clustering as detected by STORM (stochastic optical reconstruction microscopy) [15]. Cholesterol is an important part of lipid rafts and contributes to mechanisms of anti-CD20 antibodies action [37,38]. Both phenomena seem to be correlated; transfer of loaded CD20 into lipid rafts promotes calcium influx [39].

We hypothesized that crosslinking of CD38 receptors by DFMT will have similar impact on the apoptosis initiation as observed with CD20 receptors. To this end, we preincubated Raji cells either with 0.02% β-CD (to extract cholesterol) or with 1 mM EGTA (to chelate extracellular calcium) before exposing them to DFMT. Following pretreatment, decrease in apoptotic levels was observed for both, DARA-based DFMT (DARA-MORF1 followed 1 h later by HSA-(MORF2)10) and Fab'DARA-based DFMT (Fab'DARA-MORF1 followed 1 h later by HSA-(MORF2)10). Data seem to suggest a higher impact of pretreatment on Fab'DARA-based DFMT. When normalized to untreated, the percent apoptotic cells pretreated with β-CD decreased by 0.7-folds for DARA-based DFMT and by about 1.0-folds for Fab'DARA-based DFMT. The percent apoptotic cells pretreated with EGTA decreased by 0.5-folds for DARA-based DFMT and by 1.2-folds for Fab'DARA-based DFMT (Figure 3A).

**Figure 3.** (**A**) Investigation of Raji cell apoptosis induced by DFMT with or without pretreatment with 0.02 wt% β-CD or 1 mM EGTA. (**B**) Flow cytometry time lapse fluorescence of calcium chelator Fluo-3AM after addition of nanoconjugates in DARA-based DFMT (DARA-MORF1 + HSA-(MORF2)10 (left panel) and Fab'DARA-based DFMT (Fab'DARA-MORF1 + HSA-(MORF2)10) (right panel). Red arrows indicate time of nanoconjugate addition to cell sample. (**C**) Confocal microscopy images of β-CD or EGTA treated Daudi cells compared to normal cells undergoing Fab'DARA-based DFMT. \*\*\* *p* < 0.001, \*\* *p* < 0.01, n.s. nonsignificant, by One-Way ANOVA and Tukey test.

Calcium influx was also measured directly by using a fluorescent calcium chelator, Fluo-3AM, to observe calcium influx into Raji cells following DARA-based DFMT and Fab'DARA-based DFMT (Figure 3B). Raji cells were used because the cell line is both CD20 and CD38 positive. Therefore, we could compare CD38-induced calcium influx with previously reported CD20-induced calcium influx data [19,27]. Red arrows indicate the time at which the nanoconjugates were added to the cell samples (Figure 3B). The DARA-MORF1 treatment had a distinct calcium signal spike immediately upon addition to the sample; however, the Fab'DARA-MORF1 had a much less prominent Ca2+ influx event upon addition of the Fab' nanoconjugate. Conversely, upon addition of HSA-(MORF2)10 to both samples, the calcium spike was more pronounced for the Fab'DARA-based DFMT than whole antibody-based DFMT corresponding with the calcium inhibition data (Figure 3B). As mentioned above, partial inhibition of calcium influx impacted the efficacy of Fab'DARAbased DFMT more than it impacted DARA-based DFMT. The calcium influx observed in Fab'DARA-based DFMT treated cells corresponds with the results of the calcium inhibition experiments where calcium inhibition hampered the Fab'DARA-based DFMT efficacy over DARA-based DFMT. The inhibition of calcium influx was also confirmed by confocal fluorescence microscopy (Figure 3C). Fluorescence was markedly lower in the EGTA and β-CD pretreated cells. Raji cells are CD38+/CD20+ so comparison of effects resulting from crosslinking of both receptors seem to indicate a stronger response following crosslinking of CD20 (comparing data of this manuscript with [19,27]).

#### *2.5. DARA- and Fab'DARA-based DFMT Induce Apoptosis via Mitochondrial Signaling Pathway*

We next investigated the possible activation of the mitochondrial signaling pathway following crosslinking of decorated CD38 receptors on Daudi cells by multivalent macromolecular effector. The major features of the mitochondrial pathway include mitochondrial depolarization, cytochrome C release, caspase 3 activation, and bcl-2 downregulation [19,40].

Mitochondrial depolarization was assayed using the JC-1 mitochondrial membrane polarization sensor. In healthy mitochondria, membrane polarization remains intact and JC-1 aggregation occurs resulting in red fluorescence emission. As membrane potential diminishes, JC-1 can diffuse out of the mitochondria, thereby losing its red fluorescent signature as aggregates disperse into monomers. This solubilization event is observed by a change in fluorescent signature from red to green fluorescence. Therefore, red fluorescence indicates healthy mitochondria while green fluorescence indicates depolarized mitochondrial membranes. Mitochondrial membrane potential for DARA-based DFMT and Fab'DARA-based DFMT was investigated using flow cytometry (Figure 4A) and confocal microscopy (Figure 4D) with and without the presence of EGTA and β-CD. The amount of observed mitochondrial depolarization was larger for DARA-based DFMT than Fab'DARA-based DFMT, but both had higher mitochondrial membrane depolarization than DARA alone. This higher mitochondrial membrane depolarization observed in DFMTtreated cells compared to naked mAb is consistent with the higher apoptosis observed in the cell viability experiments.

B-cell lymphoma 2 (Bcl-2) and Bcl-2-associated X protein (Bax) expression levels were assayed by fluorescent immunostaining (Figure 4B). DFMT treated or untreated Daudi cells were incubated with fluorescently labeled antibodies specific to these two proteins: Bcl-2 mAb (100) Alexa Fluor® 488 and Bax mAb (2D2) Alexa Fluor® 647. Bcl-2 is located in the outer mitochondrial membrane and inhibits actions of pro-apoptotic proteins such as Bax. The expression level ratio of Bax to Bcl-2 is often used to indicate apoptotic states of cells [41,42]. DARA-based DFMT and Fab'DARA-based DFMT treated cells were tested against one another and against untreated cells for Bcl-2/Bax expression. The enhanced Bax/Bcl-2 ratio, especially in DARA-based DMFT (Figure 4B) is the indication of mitochondrial signaling pathway involvement, as supported by data on cytochrome C release (Figure 4D) and caspase 3 activity (Figure 4C). Both, DARA-based DFMT and Fab'DARAbased DFMT treated cells demonstrated 150–200% of caspase 3 activity when compared to untreated cells. The enhancement of activity was similar for both DFMT approaches. Apparently, the release of cytochrome C initiates the formation of the apoptosome with procaspase 9 and Apaf-1, followed by activation of procaspase 3 and the activation of the caspase cascade results in cell death [40,43].

**Figure 4.** Investigation of mitochondrial pathway involvement in apoptosis induction in Daudi cells. (**A**) Compiled mitochondrial membrane depolarization of DARA DFMT and Fab'DARA DFMT treated Daudi cells with or without the presence of inhibitors (EGTA or β-CD). Controls include untreated cells, a DARA-based DFMT "Premix" group, DARA crosslinked with a secondary antibody (goat antihuman: GAH), and a positive control treatment with CCCP. Depolarization was measured by JC-1 red/green fluorescence by flow cytometry. (**B**) Bax/Bcl-2 expression ratio of DARA-based DFMT and Fab'DARA-based DFMT treated Daudi cells. Expression was measured using immunostaining and flow cytometry. (**C**) Caspase 3 activity of DARA DFMT and Fab'DARA DFMT treated Daudi cells was measured using PhiPhiLux® assay kit and flow cytometry. (**D**) Cytochrome C release of DARA and Fab'DARA DFMT on Daudi cells. Increase in Cytochrome C release shown as fold increase over untreated cells. Determined by ELISA. (**E**) Confocal fluorescence microscopy visualization of mitochondrial depolarization of DARA DFMT with or without the presence of EGTA or β-CD. More red fluorescence indicates healthier mitochondria while more green fluorescence indicates the occurrence of depolarization. \*\*\*\* *p* < 0.0001, \*\*\* *p* < 0.001, \*\* *p* < 0.01, n.s. nonsignificant, by One-Way ANOVA and Tukey test.

#### *2.6. Reactive Oxygen Species (ROS) Generation, Lysosomal Enlargement, Translocation to Lipid Rafts*

ROS are generated by both Type I and Type II antibodies. In the mechanism of CD20 apoptosis initiation there is a clear distinction of mechanisms between Type I Abs, such as RTX [19] and Type II Abs, such as OBN [44]. Following receptor crosslinking, Type I antibodies induce apoptosis by receptor crosslinking followed by calcium influx, mitochondrial depolarization, and caspase activation. In contrast, Type II antibodies do not need crosslinking, they initiate apoptosis by actin remodeling, homotypic cell adhesion and lysosome disruption. Both types produce ROS.

ISA binds to an epitope independent to that of DARA and provides more enzymatic inhibition of the CD38 function [22]. ISA can induce apoptosis without crosslinking, however, lysosomal breakage and cathepsin B leakage was observed [45]. Initiation of direct apoptosis by DARA upon crosslinking and by ISA without crosslinking was reported in refs. [6,22,45]. In contrast, Moreno et al. did not observe it [46]. The discrepancy was explained by the level of CD38 expression—use of cells transduced with CD38 vs. cells with lower CD38 expression levels close to those in MM patients.

To compare mechanisms of DFMT based on CD38-targeting antibodies, DARA and ISA, with CD20-targeting OBN, Raji cells positive for both receptors were used. Three bispecific engagers, DARA-MORF1, ISA-MORF1, and OBN-MORF1, were employed. Following binding to the corresponding receptor, HSA-(MORF2)10 was used for crosslinking.

The production of ROS by DARA, ISA, DARA-based and ISA-based DFMT was at the same level without statistically significant differences (Figure 5A). However, OBN and OBN-based DFMT produced considerably higher amounts of ROS. The highest ROS production was observed in OBN-based DFMT, an indication that both mechanisms (Type I and II) of apoptosis induction are operative.

OBN DFMT showed 1.5-fold increase in lysosome size compared to untreated cells (Figure 5B). Lysosomal enlargement and increased reactive oxygen species are indicative of Type II and coincides with data our group published previously [27].

Additionally, confocal microscopy was employed to investigate redistribution of receptor-bound nanoconjugates on the cell surface into lipid rafts (Figure 5C). Fluorescently labeled antibody-MORF1 conjugates and immunostaining of lipid compartments on the cell surface with cholera toxin subunit B (CTxB) were used to visualize clustering or aggregation of antibody-bound receptors on the cell membrane. Only OBN-based DFMT exhibited typical Type II behavior of intercell homotypic adhesion and lipid raft distribution at cell–cell adhesion sites (Figure 5C *top row*).

DARA-based and ISA-based DFMT treated cells showed no significant increase in lysosome size compared to untreated cells and about only half as much ROS production as OBN-based DFMT treated cells. The confocal imaging showed minimal homotypic cell–cell adhesion and even when some cell adhesion was observed, no pronounced lipid raft accumulation at cell adhesion sites was distinguishable. All of these results are consistent with Type I antibody characteristics for DARA- and ISA-based DFMT. Only OBN-based DFMT demonstrated Type II mechanism characteristics.

The purpose of the ROS assay, the lysosomal enlargement study and the homotypic cell adhesion experiments was to elucidate any Type II antibody mechanisms of action of ISA and ISA-based DFMT. DARA (Type I antibody) and DARA-based DFMT, OBN (Type II antibody) and OBN-based DFMT were used to compare and contrast ISA's apoptosis induction mechanisms. Type II antibodies are characterized by lysosomal disruption, ROS production and homotypic cell adhesion. Type II antibodies, like OBN, induce apoptosis upon binding to their receptor epitope on the cell surface without the need for crosslinking. It is currently unknown if ISA behaves as a Type II antibody; however, ISA has been shown to induce apoptosis without the need for crosslinking [10]. Therefore, we hypothesized ISA could induce similar lysosomal disruption, ROS production and homotypic cell adhesion as OBN. Our investigation proved otherwise. ISA and ISA-based DFMT did not increase

ROS, did not enlarge the lysosomes, nor did we observe any homotypic cell adhesion under confocal microscopy.

**Figure 5.** (**A**) Generation of ROS in DARA-, ISA-, and OBN-based DFMT treated Raji cells. The ROS production was measured by oxidation of 2 ,7 -dichlorodihydrofluorescein diacetate (H2DCFDA) and quantified with flow cytometry. (**B**) Lysosomal enlargement of DARA-, ISA-, and OBN-based DFMT was quantified by LysoTracker Green DND-26 and flow cytometry. (**C**) Confocal microscopy was employed to observe any redistribution of CD38 receptors into cholesterol-rich lipid rafts. Ab-MORF1 (red); lipid raft marker: cholera toxin subunit B (CTxB) Alexa Fluor 555 (green). Nuclei were stained with Hoechst (blue). \*\*\*\* *p* < 0.0001, \*\*\* *p* < 0.001, \*\* *p* < 0.01, \* *p* < 0.05, n.s. nonsignificant by One-Way ANOVA and Tukey test.

To investigate ISA-based DFMT further, cellular apoptosis experiments were conducted on Daudi cells (Figure 6) and Raji cells (Figure S18) to examine CD38 receptor crosslinking effects when using ISA nanoconjugates. Under the same DFMT cell apoptosis treatment conditions as used for DARA DFMT (Figure 1B), ISA DFMT showed several key differences. First, the same concentration of ISA antibody induced roughly 20 percent more apoptosis than DARA antibody. This could indicate the ISA-CD38 binding epitope to inhibit CD38 enzymatic activity to a larger extent over DARA-CD38 binding, and/or binding of ISA stimulates apoptotic pathways within the cell to a larger extent. Second, Fab'ISA-MORF1 control samples showed levels of apoptosis comparable with whole antibody. This is drastically different from Fab'DARA-MORF1 controls that showed little to no apoptotic efficacy (Figure S6). Binding of either whole ISA antibody or Fab' fragment at the ISA binding site induces the therapeutic effect. Lastly, ISA DFMT did not enhance the overall apoptotic efficacy on either Daudi or Raji cells over antibody alone. However, Fab'ISA DFMT did produce significantly increased apoptosis over Fab'ISA-MORF1 control and whole antibody.

**Figure 6.** ISA-based DFMT performed on Daudi cells. Apoptosis was measured after ISA DFMT or Fab'ISA DFMT using Annexin V and Propidium Iodide staining. Flow cytometry was performed to assess percentage of apoptotic cells compared to an untreated sample group. Experiment was performed in triplicate and statistically analyzed by One-Way ANOVA and Tukey test. \*\* *p* < 0.01.

The findings with ISA suggest antibodies capable of triggering apoptosis directly upon binding to their target antigen will do so regardless of secondary crosslinking. However, the Fab' fragment crosslinking of this variety of antibody could lead to enhanced efficacy over antibody alone.

#### *2.7. Simultaneous Crosslinking of CD38 and CD20 Receptors Enhances Apoptosis*

Raji cells are CD38+/CD20+ and crosslinking of both receptors initiates apoptosis. We investigated if dual crosslinking with combination of DFMT based on Fab' fragments from DARA and RTX would enhance the efficacy of apoptosis induction. To this end we exposed Raji cells to three DFMT treatments (concentrations relate to MORF): (a) 0.5 μM Fab'DARA-MORF1 + (1 h later) HSA-(MORF2)10; (b) 0.5 μM Fab'RTX-MORF1 + (1 h later) HSA-(MORF2)10; (c) combination of 0.25 μM Fab'DARA-MORF1 + (1 h later) HSA-(MORF2)10 and 0.25 μM Fab'RTX-MORF1 + (1 h later) HSA-(MORF2)10.

Combination treatment resulted in substantially enhanced apoptotic level (Figure 7). In addition to more effective crosslinking of receptors, probably, more signaling pathways will be involved in apoptosis initiation via two receptors when compared with one. This will be evaluated in our future research.

**Figure 7.** Combinational DFMT treatment of Raji cells with CD38-targeting Fab'DARA DFMT coupled with CD20-targeting Fab'RTX DFMT. The combined treatment was compared to individual therapies at the same MORF1 equivalence. Percent apoptotic cells were determined after DFMT treatment and analyzed with Annexin V/Propidium Iodide staining via flow cytometry. All experiments were performed in triplicate and statistically analyzed by One-Way ANOVA and Tukey test. \*\*\* *p* < 0.001, n.s. nonsignificant.

Notably, in relapsed non-Hodgkin lymphoma [47] and diffuse large B-cell lymphoma [48] successful treatment of patient derived xenografts was achieved by replacing anti-CD20 RTX with anti-CD38 DARA or DARA-drug conjugates. Our data combined with these results suggests that treatment of lymphoma patients with therapies involving both antibodies might be beneficial.

#### **3. Methods and Materials**

#### *3.1. Materials*

A pair of 25-mer phosphorodiamidate morpholino oligonucleotides, MORF1 and MORF2, were customized from Gene Tools (Philomath, OR, USA). In particular, MORF1 (5 -GAGTAAGCCAAGGAGAATCAATATA-3 ) with 3 primary amine modification, and its complementary MORF2 (5 -TATATTGATTCTCCTTGGCTTACTC-3 ) with 3 -disulfide amide modification were used. Human serum albumin (HSA, chromatographically and fractionation purified with purity > 95%) was purchased from Innovative Research (Peary CourtNovi, MI, USA). DARA (Darzalex® 20 mg/mL, Janssen Biotech (Harsham, PA, USA), ISA, OBN, and RTX were obtained from Huntsman Cancer Hospital, University of Utah. Cy3-/Cy5-NHS (*N*-hydroxysuccinimide) were purchased from Lumiprobe (Hallandale Beach, FL, USA). Tris(2-carboxyethyl) phosphine (TCEP) and heterobifunctional crosslinkers NHS-PEGx-maleimide (SM(PEG)x, x = 2, 8, and 24) were purchased from Thermo Fisher Scientific (Rockford, IL, USA). Pepsin (from porcine gastric mucosa) was from Sigma-Aldrich (St. Louis, MO, USA). Anti-Bcl-2 (100) Alexa Fluor® 488 mAb and anti-Bax (2D2) Alexa Fluor® 647 mAb were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). Fluo-3 AM, JC-1 (5,5 ,6,6 -tetrachloro-1,1 ,3,3 -tetraethylbenzimidazoylcarbocyanine iodide), and carbonyl cyanide 3-chlorophenylhydrazone (CCCP) were purchased from Invitrogen (Carlsbad, CA, USA). PhiPhiLux® kit was purchased from OncoImmunin (Gaithersburg, MD, USA). Lysosome tracker Green DND-26, H2DCFDA (2 ,7 -dichlorofluorescein), Cytochrome C ELISA kit (human) and cholera toxin subunit B Alexa Fluor 555 were purchased from Thermo Fisher. All solvents were purchased from Fisher Scientific as the highest purity available.

#### *3.2. Synthesis and Characterization of Nanoconjugates*

#### 3.2.1. Antibody-MORF1 Nanoconjugates

Antibody-MORF1 nanoconjugates are synthesized in two steps as previously described [20]: first, a monoclonal antibody is reduced with TCEP to generate a sulfhydryl group, which is then conjugated with maleimide-modified MORF1. Here, is an example of synthesizing DARA-MORF1. DARA (225 μL, 4.5 mg) was buffer exchanged into 100 mM citrate buffer (pH 5.5) using an Amicon® 4 mL ultra-centrifugal filter unit (MWCO 30,000 Da). The DARA solution was then added to 20 mM TCEP (7 mg/mL, pH rebalanced to pH 5.5), and the reaction was kept for 2 h at 37 ◦C water bath with gentle shaking. The reduced DARA was obtained after removal of TCEP by washing with 10 mM PBS (pH 6.5) over 6 times. In parallel, maleimide-modified MORF1 (MORF1-MAL) was prepared by reaction of MORF1-NH2 with 50 molar excess SM(PEG)2 (succinimidyl-[(*N*maleimidopropionamido)-diethyleneglycol]ester). In brief, SM(PEG)2 (4.25 mg, 10 μmol) was dissolved in 50 μL DMSO, then added into 100 μL MORF1-NH2 solution (1.8 mg, 200 nmol) in 10 mM PBS pH 7.4. The reaction was performed for 2 h at room temperature. Then, MORF1-MAL was isolated by removal of excess SM(PEG)2 with 6 times washing using 20 mM PBS pH 6.5 buffer via Amicon® 0.5 mL Ultra Centrifugal filter unit (MWCO 3000 Da). Finally, freshly reduced DARA was conjugated with MORF1-MAL in 500 μL 20 mM PBS pH 6.5 for 2.5 h at room temperature, with 1:1.5 molar ratio of [Ab]:[MORF1- MAL]. The resultant DARA-MORF1 was purified by ultracentrifugation using the Amicon® 4 mL Ultra Centrifugal filter unit (MWCO 30,000 Da). The purity of conjugate DARA-MORF1 was confirmed with SEC on AKTApure with a SuperdexTM 200 10/300 GL column using PBS pH 7.4 as eluent. There was no detectable peak of either free MORF1 or free DARA. Furthermore, MORF1 content in the conjugate was quantified using NanoDrop (ND-1000 Spectrophotometer) at 260 nm (ε = 278,000 M<sup>−</sup>1cm−<sup>1</sup> in 0.1 N HCl aq.), whereas the concentration of DARA was determined using a bicinchoninic acid (BCA) assay.

To prepare Cy5-labeled DARA-MORF1 (Cy5-DARA-MORF1), DARA was first labeled with Cy5-NHS. The reaction was performed with molar ratio of [Ab]:[Cy5] = 1:2 by adding Cy5-NHS solution in DMSO to DARA in 1 mL PBS pH 7.4. After 2 h reaction at room temperature, free dye was removed using a PD10 column (GE Healthcare). The collected product was then buffer exchanged using ultracentrifugation 4 times with 100 mM citric acid buffer pH 5.5. Then, procedures described above were followed.

#### 3.2.2. Fab'-MORF1 Nanoconjugates

Fab'-MORF1 nanoconjugates are synthesized in multiple steps as previously described [28]. First, a monoclonal antibody is digested into F(ab')2 in the presence of pepsin, followed by reduction with TCEP to generate Fab' with a sulfhydryl group (Fab'-SH), which is then conjugated with maleimide-modified MORF1. Here, is an example of synthesizing Fab'DARA-MORF1. Briefly, DARA was buffer exchanged into citrate buffer pH 4.0. Pepsin (10 *w*/*w*%) was added to DARA solution, and the reaction was kept at 37 ◦C. The digestion process was monitored on AKTApure till complete disappearance of DARA peak, while a new peak with lower molecular weight (at right side) showed up, which was related to F(ab')2. Complete digestion occurred in 80 min. The F(ab')2 was purified using ultracentrifugation with a 30,000 MWCO tube and stored at 4 ◦C in PBS 7.4. F(ab')2 (4 mg, 4 mg/mL) was reduced with TCEP (4.6 mg, 20 mM) in 100 mM citric acid buffer pH 5.5 for 2 h at 37 ◦C and purified using ultracentrifugation with a 10,000 MWCO tube. The MORF1-MAL was synthesized in parallel as described above. MORF1-MAL (1.2 equiv.) was reacted with freshly reduced Fab' (1 equiv.) for 3 h at room temperature in PBS pH 6.5 buffer. The final product was purified by ultracentrifugation using a 30,000 MWCO tube by washing 6–8 times with PBS pH 7.4 buffer. MORF1 content in Fab'DARA conjugates was determined using UV-Vis absorbance at 260 nm; the concentration of antibody fragment was determined using BCA assay.

ISA-MORF1 and Fab'ISA-MORF1 were synthesized as described above for DA-RA conjugates.

#### 3.2.3. Multiple MORF2 Modified Human Serum Albumin (HSA-(MORF2)y)

Two steps were conducted to conjugate complementary MORF2 to HSA as previously described [29]. First, amino groups from accessible lysine residues in HSA were converted to maleimide groups by using heterobifunctional crosslinker NHS-PEGx-maleimide, then freshly reduced MORF2-SH was attached to HSA in multiple copies via thiol-ene reaction. In this study, SM(PEG)x (x = 2, 8, and 24) that differ in length were used in order to investigate spacer effect on MORF1-MORF2 biorecognition and induction of apoptosis. Here, is an example in which SM(PEG)2 was used to synthesize nanoconjugate HSA-PEG2- (MORF2)10. Briefly, HSA (5 mg, 3.9 μmol NH2 equiv.) was dissolved in 400 μL PBS pH 7.4 buffer. SM(PEG)2 (18.3 mg, 10 eq) in 150 μL DMSO was added into HSA solution. The reaction was kept stirring for 2 h at room temperature. The maleimide-modified HSA was then purified by ultracentrifugation using an Amicon® 4 mL ultra centrifugal filter unit (MWCO 30,000 Da). The number of maleimide groups per HSA molecule was determined by a modified Ellman's assay (for maleimide group) and BCA assay (for quantification of HSA).

In a parallel reaction, 3 -disulfide MORF2 (2.89 mg) was reduced with 3.5 mg/mL TCEP (10 mM) in 250 μL PBS pH 7.4 at 37 ◦C for 45 min, followed by purification via ultrafiltration using a Amicon® 0.5 mL ultracentrifugal filter unit (MWCO 3000 Da). Freshly reduced MORF2-SH was then reacted with HSA-PEG2-MALx with molar ratio of [SH]:[MAL] = 2:1 in 500 μL PBS (pH 6.5) for 3 h at room temperature. HSA-PEG2-MORF2 was purified by ultracentrifugation using a 30,000 MWCO ultra centrifugal unit washing 6–8 times with PBS pH 7.4 buffer. Purity was confirmed by AKTApure. The ratio of MORF2 per HSA molecule was determined by UV-Vis spectrophotometry (ε = 252,120 M−<sup>1</sup> cm−<sup>1</sup> in 0.1 N HCl) and BCA assay.

For synthesis of Cy3-labeled HSA-PEG2-(MORF2)y, HSA was first reacted with Cy3- NHS with molar ratio of [HSA]:[Cy3] = 2:1. For example, a solution of Cy3-NHS (70 μg, 100 nmol) in DMSO was added to a solution of HSA (3.35 mg, 50 nmol) in 1 mL PBS pH 7.4 buffer and reacted for 2 h at room temperature. Cy3-labeled HSA was purified using a PD-10 column using PBS pH 7.4 as eluent. Then, the synthesis proceeded as described above.

#### 3.2.4. MORF1-MORF2 Hybridization

MORF1-MORF2 hybridization upon mixing Ab-MORF1 (or Fab'-MORF1) with HSA- (MORF2)y was determined by the changes of optical density at 260 nm (ND-1000 spectrophotometer) that was a reflection of the hypochromic effect. For example, DARA-MORF1 (or ISA-MORF1) and HSA-(MORF2)10 solutions in PBS, pH 7.4 were mixed in different ratios with a constant total MORF (MORF1 + MORF2) concentration of 2.5 μM at room temperature. Then, 10 min post mixture, the optical density at 260 nm was recorded. All measurements were performed in triplicate.

In addition, MORF1-MORF2 hybridization among Ab-MORF1 (or Fab'-MORF1) and HSA-(MORF2)10 was also confirmed by SEC by comparison with individual conjugates, Ab-MORF1 and HSA-(MORF2)10.

#### *3.3. Cell Culture*

Human lymphoma cell lines (Daudi and Raji) and human MM cell lines (RPMI 8226 and U266) were purchased from the American Type Culture Collection (ATCC). All cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS), penicillin (100 units mL−1) and streptomycin (0.1 mg/mL−1) at 37 ◦C in a 5% CO2 humidified atmosphere. Human MM cell line ANBL-6 was obtained from Dr. Diane Jelinek of Mayo Clinic (Rochester, MN, USA). The cells were cultured in IMDM with 10% FBS and interleukin-6 (1 ng/mL).

#### *3.4. Cell Surface CD38 Expression and Binding Assay*

DARA binding experiments were conducted by incubating Daudi, RPMI 8226, Raji, ANBL-6, and U266 cells with a range of DARA (primary antibody) concentrations, and subsequently exposing them to a fluorescently labeled, anti-human secondary antibody. Specifically, 2 × 105 cells were treated with DARA (1 <sup>μ</sup>M, 0.5 <sup>μ</sup>M, 0.25 <sup>μ</sup>M, 0.1 <sup>μ</sup>M or 50 nM; in PBS pH 7.4) in a 24-well plate in 400 μL media for 1 h at 4 ◦C. Then, the cells were washed with PBS and resuspended in PBS containing 1 *w*/*v*% BSA. An Alexa Fluor 488-labeled goat anti-human antibody (3 μL, 2 mg/mL) was added to each well and further incubated for 1 h at 4 ◦C. The cells were washed and resuspended in PBS and analyzed by flow cytometry. The fluorescence was normalized to an untreated control for each cell type and reported as a "fold-increase over untreated" value. All experiments were performed in triplicate.

#### *3.5. MORF1-MORF2 Hybridization on Cell Surface—CD38 Crosslinking*

Daudi cells (2 × 105) were treated in a 24-well plate (400 <sup>μ</sup>L RPMI-1640 medium per well) with Cy5-labeled DARA-MORF1 (0.5 μM MORF1) for 30 min, followed by a Cy3-labeled HSA-(MORF2)10 (0.5 μM [MORF2]) treatment for 1 h. Then, the cells were washed with PBS and cell nuclei were stained with Hoechst 33342 (5 μg/mL) for 5 min. The cells were washed and resuspended in PBS for imaging. Cy3 fluorescence was measured by 488 nm excitation with 530/30 nm band-pass filter. Cy5 fluorescence was measured by 633 nm excitation with 695/40 nm band-pass filter. An overlay of the respective fluorescent signals was used to visualize MORF1-MORF2 hybridization and co-localization of the nanoconjugates on the cell surfaces.

#### *3.6. Apoptosis Assay*

Apoptosis of DFMT-treated cells was quantified using Annexin V/Propidium Iodide staining and measured by flow cytometry. Each cell type's (Daudi, Raji, RPMI 8226, ANBL-6, and U266) apoptotic response to DARA, DARA-based DFMT, Fab'DARA-based DFMT and DARA plus a secondary antibody were evaluated. Briefly, 2 × 105 cells were treated in a 24-well plate in 400 μL appropriate media. Cells were first incubated with DARA-MORF1 or Fab'DARA-MORF1 (0.5 μM MORF1) for 1 h, followed by a PBS wash and subsequent treatment with HSA-MORF2)10 (0.5 μM MORF2) in fresh media for 24 h. Then, the cells were washed with Binding (HEPES saline) Buffer and stained with Annexin V/Propidium Iodide for 20 min at 4 ◦C. Cells were washed and resuspended in Binding Buffer for flow cytometry analysis. Untreated cells were used to gate as Annexin V/PI -/ population. Each treated sample was then compared to untreated. Values reported are the percentage of Annexin V positive cells. All treatments were performed in triplicate with appropriate controls. One-way ANOVA (α = 0.05) followed by Tukey Test analysis was used to determine significant differences in the reported data.

#### Details and Nomenclature of DFMT Apoptosis Assays

DARA DFMT: DARA-MORF1 (0.5 μM MORF1) 1 h followed by HSA-MORF2)10 (0.5 <sup>μ</sup>M MORF2) in fresh media for 24 h; 2 × 105 cells/well; 24-well plate.

Fab'DARA DFMT: Fab'DARA-MORF1 (0.5 μM MORF1) 1 h followed by HSA-MORF2)10 (0.5 <sup>μ</sup>M MORF2) in fresh media for 24 h; 2 × 105 cells/well; 24-well plate.

ISA DFMT: ISA-MORF1 (0.5 μM MORF1) 1 h followed by HSA-MORF2)10 (0.5 μM MORF2) in fresh media for 24 h; 2 × <sup>10</sup><sup>5</sup> cells/well; 24-well plate.

Fab'ISA DFMT: Fab'ISA-MORF1 (0.5 μM MORF1) 1 h followed by HSA-MORF2)10 (0.5 <sup>μ</sup>M MORF2) in fresh media for 24 h; 2 × 105 cells/well; 24-well plate.

#### *3.7. Apoptosis Inhibition*

To validate the participation of calcium influx in apoptosis initiation, Daudi cells were pretreated with lipid raft inhibitor β-cyclodextrin (β-CD; inhibiting CD20 crosslinking) or Ca2+ chelating agent EGTA (ethyleneglycol-bis(β-aminoethyl ether)-*N*,*N*,*N* ,*N* -tetraacetic

acid) which had significantly reduced the calcium influx after DFMT treatment. Daudi cells were pretreated with either 1 mM Ca++ doped RPMI-1640 medium or 0.02% βcyclodextrin containing RPMI-1640 medium for 1 h. The cells were washed with fresh medium and exposed to DARA-MORF1 or Fab'DARA-MORF1 for 1 h at 37 ◦C, followed by exposure to HSA-(MORF2)10. Cell viability was quantitated using Annexin V/PI labeling and flow cytometry.

#### *3.8. Caspase 3*

Caspase 3 activity was evaluated using a PhiPhiLux®-G1D2 kit (OncoImmunin, Gaithersburg, MD). The manufacturer's protocol was followed, including the final stage Propidium Iodide staining to assess cell membrane integrity. The reported values are "fold-increase over control" fluorescence measurements of the PhiPhiLux indicator on treated Daudi cells over untreated cells.

#### *3.9. Cytochrome C*

Levels of cytochrome C were evaluated in DARA and Fab'DARA DFMT-treated Daudi cells by ELISA. After treatment, Daudi cells (2 × 105 cells) were washed with cold PBS and cell pellets were subsequently lysed with 100 μL cell extraction buffer (1 mM phenylmethanesulfonyl fluoride (PMSF) and 1:100 protease inhibitor cocktail) for 30 min on ice. The extract was transferred to a microcentrifuge tube and centrifuged at 13,000 rpm for 10 min at 4 ◦C. The amount of cytochrome C in the lysate was measured using an ELISA kit (R&D Systems) according to the manufacturer's instructions. All samples were conducted in triplicate.

#### *3.10. Mitochondrial Depolarization*

JC-1 mitochondrial membrane potential sensor (Thermo Scientific) was used to evaluate the extent of mitochondrial depolarization of Daudi cells. After indicated treatments, the cells (2 × <sup>10</sup>5) were washed with PBS two times and resuspended in 100 <sup>μ</sup>L PBS. JC-1 (4 μM) was added to each sample and incubated at 37 ◦C for 30 min. For the positive control group, CCCP (0.5 μM) was added and incubated simultaneously with JC-1 for 30 min. After washing by PBS, cells were resuspended in PBS and analyzed by flow cytometry using 488 nm excitation with 530/30 nm and 585/42 nm band-pass filters, or observed under confocal microscopy. All experiments were carried out in triplicate.

#### *3.11. Calcium Influx by Confocal Microscopy*

Daudi cells (2 × 105) were incubated with Fluo-3AM (5 <sup>μ</sup>M) in 100 <sup>μ</sup>L RPMI-1640 medium containing 2.5 mM Ca2+ for 30 min at 37 ◦C and compared with cells pretreated with 0.02 wt.% β-cyclodextrin or 1 mM EGTA. Following treatment, cells were washed with PBS and resuspended in RPMI-1640 medium containing 2.5 mM Ca2+ and observed under confocal microscopy.

#### *3.12. Calcium Influx by Flow Cytometry*

Raji cells (4 × 105 cells/well) were counted and stained with Fluo-3 AM (5 <sup>μ</sup>M) for 30 min at 37 ◦C. After staining, the cells were washed with PBS and resuspended in 400 μL cell culture medium containing 2.5 mM Ca2+ and immediately taken for flow cytometry analysis (excitation at 488 nm and emission at 530 nm). Baseline fluorescence was measured for 100 s. Then, 1 μM DARA-MORF1 or Fab'DARA-MORF1 was added to the sample. The fluorescence was measured continuously for another 200 s, followed by addition of HSA- (MORF2)10 (1 μM). Fluorescence was monitored for another 600 s or until all cells were counted. Rituximab-based DFMT calcium influx was employed as control (Figure S6).

#### *3.13. Bcl-2/Bax Detection*

Following treatment, levels of expression of Bcl-2 and Bax were quantified by fluorescent immunostaining. The cells were sequentially fixed by 4% paraformaldehyde

for 15 min at room temperature, permeabilized by 90% methanol for 30 min on ice, and immunostained by Alexa Fluor 488 conjugated anti-Bcl-2 mAb (1:50, Santa Cruz Biotechnology) and AF647 conjugated anti-Bax mAb (1:50, Santa Cruz Biotechnology) in 1% BSA buffer for 1 h at room temperature. After washing by cold PBS twice, the fluorescence was quantified by flow analysis. All experiments were carried out in triplicate.

#### *3.14. Lysosomal Enlargement*

Type II antibody-induced cell death involves lysosomal enlargement/breakage. Lysosome activity was tested by LysoTracker Green staining followed by flow cytometry. Raji cells (2 × <sup>10</sup>5) were exposed to naked antibody or antibody-MORF1 conjugate for 1 h at 37 ◦C, followed by exposure to HSA-(MORF2)10. After 24 h treatment, cells were stained with LysoTracker Green DND-26 (200 nM) for 20 min at 37 ◦C. Fluorescence was quantified using flow cytometry. Each sample was prepared in triplicate.

#### *3.15. Reactive Oxygen Species Production*

Quantification of reactive oxygen production was performed on Raji cells (2 × 105) by oxidation of 2 ,7 -dichlorodihydrofluorescein diacetate (H2DCFDA). Cells (2 × 105 cell/well, 24-well plate) were treated with antibody (DAR, ISA, or OBN) (0.5 μM) or antibody (DARA, ISA, or OBN)-MORF1 conjugate followed by HSA-(MORF2)10 (MORF1 = MORF2 = 0.5 μM). After 24 h treatment, cells were incubated with H2DCFDA (5 μM) for 30 min at 37 ◦C. Cells were washed with PBS and analyzed with flow cytometry. Each sample was prepared in triplicate.

#### *3.16. Translocation of CD38 into Lipid Rafts*

The motility and translocation of CD38 receptors under the influence of DFMT was evaluated by cholera toxin B staining and observed under confocal microscopy. Briefly, Raji cells (2 × <sup>10</sup><sup>5</sup> cells/well) were loaded into a 24-well plate. The cells were treated with either 0.5 μM antibody alone or 0.5 μM Cy5-labeled antibody-MORF1 for 1 h at 37 ◦C followed by PBS wash and 0.5 μM HSA-(MORF2)10 exposure for 2 h at 37 ◦C. Then, the cells were washed to remove unbound antibodies, and stained with Alexa Fluor-555 conjugated cholera toxin B subunit (10 μg/mL) for 1 h at 4 ◦C. The samples were immediately imaged by confocal microscopy.

#### *3.17. Statistical Analysis*

All statistical analysis was performed on Microsoft Excel. No samples were left out of any analysis calculations. Sample groups were compared using one-way ANOVA followed by Tukey test. *p* < 0.05 was considered statistically significant. All experiments were performed with at least *n=3* samples per group. Percent apoptotic cells was calculated by summation of quadrants one, two and four in the Annexin V/PI gated flow cytometry runs. Magnitude of fluorescence in immunostaining procedures was quantified by flow cytometry's geometric mean average function. The geometric mean average was either normalized to untreated control cells or presented as fold increase over untreated, as specified above.

#### **4. Conclusions**

The experiments presented herein illustrate the versatility of the DFMT system and demonstrate how it can be applied towards the treatment of several B cell malignancies including multiple myeloma, lymphoma and leukemia. Four novel antibody nanoconjugates as bispecific engagers (Fab'DARA-MORF1, Fab'ISA-MORF1, DARA-MORF1, and ISA-MORF1) and adaptations to the HSA-based multivalent crosslinking effector molecule (HSA-PEGx-(MORF2)y) were synthesized. Biorecognition of complementary engagers and effectors at cell surface mediated by MORF1/MORF2 hybridization resulted in crosslinking of CD38 receptors and apoptosis initiation in CD38+ cells: Daudi, Raji RPMI 8226 and ANBL-6. The level of apoptosis induction, Daudi > RPMI 8226 > Raji > ANBL-6 >> U266,

correlated with CD38 expression. Additionally, insertion of a flexible PEG24 (95.2 Å) spacer into the effector conjugate HSA-PEG24-(MORF)13 significantly increased apoptosis of Raji cells when compared to effectors containing PEG8 (39.2 Å) or PEG2 (17.6 Å) moieties.

Beyond the synthesis and apoptosis efficacy studies with the new conjugates, a thorough investigation into the mechanisms of action of DARA, DARA DFMT, and Fab'DARA DFMT on CD38+ cell lines were conducted. Preincubation of cells with β-cyclodextrin (to extract cholesterol) or EGTA (to complex extracellular Ca++) decreased levels of apoptosis. DARA-based and Fab'DARA-based DMFT induced, following crosslinking of CD38 receptors, apoptosis via the mitochondrial signaling pathway as indicated by enhanced Bax/Bcl-2 expression ratio, ROS generation, cytochrome C release, and caspase 3 activation.

A comparison of ISA and ISA-based DFMT to DARA and a known Type II antibody, OBN was conducted. ISA induced apoptosis in Daudi and Raji cells. ISA DFMT did not enhance apoptosis when compared to ISA; however, crosslinking of the CD38-Fab'ISA-MORF1 complex with HSA-(MORF2)10 resulted in enhanced apoptotic levels. Additionally, Fab'ISA-MORF1 induced apoptosis in Daudi and Raji cells on its own. Comparison with DFMT based on anti-CD20 Type II antibody OBN revealed that both DARA and ISA did not exhibit features related to Type II apoptotic mechanisms (lysosomal enlargement, homotypic cell adhesion).

Finally, simultaneous crosslinking of CD38 and CD20 receptors on Raji cells increases the level of apoptosis when compared to crosslinking of individual receptors. This finding suggests a therapeutic potential of lymphoma treatment with a mixture of antibodies.

**Supplementary Materials:** The following are available online. Figure S1. Size exclusion chromatography (SEC) of DARA-MORF1 and HSA-(MORF2)10 hybridization. Figure S2. UV-Vis spectroscopy to determine DARA-MORF1 and HSA-(MORF2)10 hybridization. Figure S3. SEC of Fab'ISA-MORF1 and its intermediates. Figure S4. SEC of ISA-MORF1 and HSA-(MORF2)10 hybridization. Figure S5. UV-Vis spectroscopy to determine ISA-MORF1 and HSA-(MORF2)10 hybridization. Figure S6. Flow cytometry cell population shifts for DARA DFMT experiments. Figure S7. Flow cytometry cell population shifts for Fab'DARA DFMT experiments. Figure S8. DARA DFMT on ANBL-6 cells. Figure S9. Flow cytometry readout of Rituximab DFMT to monitor calcium influx. Figure S10. Confocal microscopy of Fab'DARA DFMT apoptosis inhibition by β-CD and EGTA. Figure S11. Bax/Bcl-2 expression of DFMT-treated Daudi cells. Figure S12. Cytochrome C calibration curve from ELISA assay. Figure S13. Flow cytometry of Bax and Bcl-2 expression in Daudi cells. Figure S14. Flow cytometry of caspase 3 population gating. Figure S15. Flow cytometry of caspase 3 post-measurement propidium iodide staining. Figure S16. Confocal microscopy of assessment of lipid raft redistribution using Cy5-labeled antibodies. Figure S17. Additional confocal microscopy of lipid raft redistribution induced by DFMT systems. Figure S18. ISA DFMT & Fab'ISA DFMT apoptosis assessment on Raji cells.

**Author Contributions:** Conceptualization, J.K., J.Y., M.T.G.; methodology, M.T.G., J.L., J.W., J.Y.; validation J.K., J.Y., M.T.G., D.S.; investigation, M.T.G., J.L., J.W.; data analysis, M.T.G., J.K., J.Y., J.L.; writing—original draft preparation, M.T.G., J.L.; writing—review and editing, J.K., J.Y., D.S., M.T.G.; funding acquisition, J.K., J.Y., D.S.; supervision, J.K., J.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** Research was supported by NIH grant RO1 CA246716 from the National Cancer Institute (to JK) and by Huntsman Cancer Institute ET grant 39024 (to DS/JY).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** All data generated or analyzed during this study are included in the article.

**Acknowledgments:** We acknowledge support of funds in conjunction with grant P30 CS042014 awarded to Huntsman Cancer Institute. We thank core facilities, flow cytometry, confocal fluorescence microscopy, and mass spectrometry for support. We are indebted to Diane Jelinek of Mayo Clinic for kind gift of ANBL-6 cells.

**Conflicts of Interest:** The authors declare the following competing financial interest(s): J.Y. and J.K. are co-inventors on a pending US patent application (PCT/US2014/023784; assigned to the University of Utah) related to this work. J.K. is Chief Scientific Advisor and J.Y. Scientific Advisor for Bastion Biologics. Otherwise, the authors declare no competing financial interests.

**Sample Availability:** Samples of the compounds are not available from the authors.

#### **References**


*Article*
