Encapsulating therapeutic cells in RGD-modified agarose microcapsules

Current cell-based strategies for repairing damaged tissue often show limited efficacy due to low cell retention at the site of injury. Encapsulation of cells within hydrogel microcapsules demonstrably increases cell retention but benefits can be limited due to premature cell escape from the hydrogel microcapsules and subsequent clearance from the targeted tissue. We propose a method of encapsulating cells in agarose microcapsules that have been modified to increase cell retention by providing cell attachment domains within the agarose hydrogel allowing cells to adhere to the microcapsules. We covalently modified agarose with the addition of the cell adhesion peptide, RGD (arginine, glycine, aspartic acid). We then used a microfluidic platform to encapsulate single cells within 50 μm agarose microcapsules. We tracked encapsulated cells for cell viability, egress from microcapsules and attachment to microcapsules at 2 h, 24 h, and 48 h after encapsulation. Many encapsulated cells eventually egress their microcapsule. Those that were encapsulated using RGD-modified agarose adhered to the outer surface of the microcapsule following egress. NIH 3T3 cells showed nearly 45% of egressed cells attached to the outside of RGD modified agarose microcapsules, while minimal cellular adhesion was observed when using unmodified agarose. Similarly, human umbilical vein endothelial cells had up to 33% of egressed cells attached and explant-derived cardiac cells showed up to 20% attachment with the presence of RGD binding domains within the agarose microcapsules.


Introduction
Cell-based therapeutic treatments are becoming an increasingly popular strategy for treating diseased and damaged tissue within the body. With tissue damage comes a loss of function that can lead to greater systemic issues. In fact, certain ailments require invasive treatments involving strong drugs with significant side-effects or potentially requiring full organ transplant. New cell-based therapies are often touted as less invasive alternatives for tissue regeneration.
Many cell-based therapeutic strategies involving the delivery of healthy donor cells to the site of injury are now being assessed for tissue regeneration purposes in preclinical and clinical trials [1,2]. Healthy cells are harvested either from an autologous or an allogeneic origin, expanded in vitro, then injected into the site of damage within the tissue [3,4]. Therapeutic effects from the injected cells are provided through two separate mechanisms. The first occurs through cell integration wherein therapeutic cells are injected into the tissue at the site of damage and gradually replace the damaged cells with healthy cells [5,6]. The second mechanism occurs through paracrine signaling. This involves retention of therapeutic cells at or very near the site of injury, where their presence triggers and enhances normal immune responses, leading to faster, more robust recovery that helps regenerate the damaged tissue [7][8][9][10][11]. Both pathways can result in regeneration of healthy tissues without the need for full transplant of an organ or tissues, or the long-term use of drugs [12]. While cell-based therapies are being employed in health care settings, there are still critical issues that need to be addressed before they can be widely used in patient care settings. Two common issues involve low engraftment into local tissue and low cell retention after introduction [6,13]. Negative immune responses against the injected cells can trigger their removal from the site of injury, therefore reducing retention and engraftment [6]. Commonly, vasculature and blood flow also inhibit the retention of injected cells, as free cells are very quickly cleared from the injection site from the constant flow of fluids throughout tissues [14,15]. Low engraftment and low retention both result in reduced therapeutic efficacy since the number of injected cells that remain localized and active are linked to positive therapeutic effects [14,15].
One method of overcoming the issues of low engraftment and poor retention is the microencapsulation of injected cells in hydrogel materials. The porous nature of most hydrogels allows small molecules, nutrients and waste products to passively diffuse through the material [16]. The hydrogel creates a barrier surrounding the encapsulated cells against undesired immune responses [16]. One previous study by Kanda et al showed that cell retention was enhanced with microencapsulated cells [15], presumably due to the larger volume which is more difficult for vasculature to clear. In vivo studies using encapsulated therapeutic cells have shown significant improvements in therapeutic outcomes over free floating cells. Injection of encapsulated mesenchymal stem cells (MSCs) within alginate microcapsules resulted in increased osteogenesis and mineralization compared to injection of non-encapsulated cells in a rat tibial marrow ablation model [17]. In other studies, improvements in the left ventricular ejection fraction of the heart following myocardial infarction, as well as reduction in scar size, were observed when using MSCs [18] and explant-derived cardiac cells (EDCs) [19] encapsulated in agarose microgels. These beneficial effects were not observed when using unencapsulated cells. Enhanced cardiomyocyte proliferation was also seen in one study when agarose encapsulated therapeutic cells were used [19]. Our previous studies investigating cell behavior inside agarose hydrogel microcapsules showed that cells quickly and efficiently egress from the microcapsules [20]. Once egressed, these cells are no longer associated with the microcapsules, and therefore will not benefit from the encapsulation process. We sought to enhance and prolong cellular association with the agarose microcapsules in an effort to promote cell retention at the site of injured tissue. We employed a strategy where we conjugated fragments of extracellular matrix (ECM) proteins to agarose, prior to encapsulation, providing cells with binding domains within and on the surface of the microcapsules. We used a peptide (CSGSGSGSRGDS) containing the RGD cell binding domain (Arg-Gly-Asp) found in ECM proteins and has been tested in other therapeutic models [21][22][23], along with a cysteine-terminated spacer sequence for conjugation to functionalized agarose. The goal was to promote cell adhesion to the agarose microcapsules after cellular egress. By encapsulating cells in this modified agarose (henceforth referred to as 'RGD-modified agarose'), we showed that while cells would still egress from the modified agarose microcapsules, the RGD peptides provided binding sites for the cells that promote cell survival and increased association with the microcapsules for at least 2 days post encapsulation.

Synthesis of modified agarose
Agarose was chemically modified via maleimide to include the amino acid sequence CSGSGSGSRGDS, as described in Luan et al [24] (supplemental figure 2). Briefly, low-melting point agarose was converted into maleimide-agarose using pmaleimidophenyl isocynanate (PMPI), where 100 mg of agarose was dissolved in 5 ml of DMSO then heated to 80 • C in an oil bath for 2 h. Once fully dissolved, the agarose solution was cooled to room temperature and 5 mg of PMPI was added and the mixture was stirred overnight at room temperature. The solution was subsequently dialyzed for 72 h in distilled, autoclaved water, with water exchanged twice daily. The now maleimide-agarose was lyophilized for storage until the addition of the amino acid chain. To accomplish this, maleimide-agarose was dissolved in PBS (pH 6.5-7.5) at 65 • C then degassed for 1 h to produce a 2.7% (w/v) solution. The peptide sequence was added to the agarose solution (1.5 mM in PBS) and allowed to react with the maleimide-agarose at 37 • C for 2 h, under argon. RGD agarose was melted then stored in 1 ml aliquots at 4 • C until used for encapsulation experiments. RGDS conjugation was measured using the Ellman assay to quantify the amount of free thiol after the reaction.

Production of microfluidic devices
Microfluidic devices were made from polydimethylsiloxane (PDMS; Sylgard 184 Silicone Elastomer Kit, Dow), using a silicon wafer mold as previously described [25,26]. Briefly, PDMS was prepared as per the manufacturer's directions and poured onto the wafer mold. Trapped air bubbles were removed by degassing using a vacuum chamber. PDMS was then cured in a 70 • C oven for 2-4 h. Cured PDMS was removed from the wafer using a sharp blade, and inlet and outlet holes were punched into the PDMS using a skin biopsy punch needle. PDMS was then plasma bonded to glass slides to form functional microfluidic devices using an air plasma chamber (Glow Research, AutoGlow System) exposing the glass slides and PDMS devices for 36 s at a power of 36 W. Bonded devices were then incubated in a 70 • C oven for 48 h to ensure complete bonding and to render the PDMS microchannel walls hydrophobic. Devices were stored at room temperature until use.
All cells were subcultured following standard general protocols for NIH 3T3 cells. Briefly, cells were grown in 10 cm culture plates to approximately 80% confluence. Cells were lifted using 2-3 ml of either TrypEDTA or TrypLE (Glibco) for therapeutic cells. Cells were plated between 600 000 and 1000 000 cells per plate, and maintained with the cell appropriate media at 37 • C 5% CO 2 .

Microcapsule production
Cells were encapsulated using a microfluidic device (figure 1) to produce a high throughput, monodisperse sample of cells within spherical hydrogel microdroplets as previously described [20].
Briefly, agarose was melted at 70 • C for 45 min, then cooled to 37 • C prior to addition of cells and ECM proteins. Cells and ECM proteins were combined with the molten agarose before being introduced to the device (2% (w/v) agarose, 8.5 million cells ml −1 , 0.1 mg ml −1 fibrinogen and 0.1 mg ml −1 fibronectin). Agarose and oil are made to flow into the device using pressurized inlets, forcing the fluids through the device. A 35 µm nozzle, located downstream of the inputs facilitates the formation of agarose microdroplets via flow focusing. In this process, the agarose stream becomes 'pinched off ' by the intersecting oil streams creating an oil agarose emulsion. The microdroplets were stabilized by the addition of 1.5% (v/v) Span 80 surfactant. The emulsion was made to flow through a cooling serpentine to gel the agarose microdroplets into microcapsules. Molten samples were maintained at 37 • C prior to and during the process of encapsulation, while microcapsules were collected in sample vials maintained in ice water to further facilitate gelling of the agarose. The capsules were isolated by centrifugation for 3 min at 300 × g. The pellet containing microcapsules was collected and washed with 500 µl clean cold media. Capsules were again isolated by centrifugation as described and prepared for microscopy and Cell Counting Kit-8 (CCK-8) experiments.

Microscopy
Time-point microscopy samples were prepared in 35 mm cell culture dishes, containing 70-100 µl of microcapsules in 1000 µl cell appropriate media. Replicate samples were prepared in dishes coated in poly(2-hydroxyethyl methacrylate) (pHEMA). Samples were treated with fluorescence viability stain (LIVE/DEAD Viability/ Cytotoxicity kit, Invitrogen) as per the manufacturer's directions and incubated at room temperature for 30 min. Images were collected using an Olympus IX51 fluorescence microscope, point grey camera and spinview image software. Images were compiled using ImageJ software. Capsules and cells were counted as either and identified as either occupied/unoccupied, live/dead or adhered/not. Cell viability was calculated for all time points of 2 h, 24 h, and 48 h, while cumulative egress, and cell attachment were calculated for 24 h, 48 h.

CCK-8
All samples analyzed with the CCK-8 assay were done so as per the manufacturer's directions. Briefly, samples were prepared in 96 well plates coated with pHEMA, by the addition of 10 µl of microcapsule suspension (corresponding to roughly 12 000 encapsulated cells) with 100 µl of media in each well. Triplicate wells were used for each sample condition. CCK-8 reagent was added to each well (10 µl of stock solution) and incubated for 2 h before reading the absorbance at 460 nm using a plate reader. A calibration curve using a range of different cell concentrations (between 0-6250 cells per well of 96-well plate) was also prepared for accurate quantification of cells.

Confocal microscopy of focal adhesions
The presence of focal adhesions on encapsulated and egressed cells was investigated using a staining protocol for actin cytoskeleton and vinculin. Cell samples were stained according to the protocol described for the Actin Cytoskeleton and Focal Adhesion kit (Millipore) that was used. Briefly, cells encapsulated in agarose microcapsules were incubated at 37 • C for 24 h before staining. Encapsulated cells were collected by centrifugation, then fixed with 4% (w/v) paraformaldehyde and permeabilized with 1% (v/v) triton X. Cells were rinsed three times with clean PBS then stained for vinculin, actin fibers, and nuclei using vinculin specific antibodies, phalloidin, and DAPI respectively. Cells were finally washed and sealed under coverslips for microscopy. Confocal microscopy was completed using a Zeiss LSM 880 confocal microscope, and Carl Zeiss ZEN 2.3 (black; release version 13.0.0) was used for image editing and manipulation.

Statistical analysis methods
All statistical analysis was completed using Excel and GraphPad Prism. All tests between different timepoints (same agarose types and plate coatings) were conducted using paired t-tests, and all others comparing agarose types and plate coating conditions for the same time point and were conducted using Student t-tests.

Results
We designed a microfluidic device to encapsulate individual cells in agarose (figure 1). The use of unmodified agarose for microencapsulation presented few difficulties. However, microcapsule formation using RGD-modified agarose was more challenging compared to when using unmodified agarose. Microcapsules were often asymmetrical and inconsistently sized ( figure 2(A)). Additionally, the use of RGD-agarose often clogged the microfluidic device due to premature partial gelling in the inlet. The conjugation of RGD to agarose also altered the temperature-dependent gelation profile of the hydrogel. To solve these handling issues, we used a blend of 2:1 (RGD-modified to unmodified) agarose, henceforth referred to as RGD-agarose. This provided uniform encapsulation of cells at physiological temperatures using our microfluidic devices. While we did not explore the molecular origins of this lack of uniformity, we suspect it could be due to interactions between the peptide chains which alter the gelling temperature. The 2:1 dilution of RGD-modified agarose results in a concentration of ∼3 × 10 8 RGD sites pl −1 . This corresponds to ∼2 × 10 10 sites/microcapsule, or a surface concentration of ∼4.5 × 10 3 RGD sites µm −2 . Previous work by Le Saux et al has shown that silicate surfaces required a minimum 6 RGD sites µm −2 to promote focal adhesion [27]. Based on simple conversions, this is roughly 3 orders of magnitude less than we used in our system. We opted to use an RGDmodified to unmodified blend ratio of 2:1 to maximizing the number of available adhesion sites, but cell binding does occur at lower RGD site densities.
Previous work done by Collins et al showed that the number of cells within each microcapsule is influenced by microcapsule diameter and followed a Poisson distribution [28]. For this reason, microcapsule diameter was kept consistent at a diameter of about 52 ± 4 µm between different experiments as size can greatly affect cell behavior, including viability and egress [15,20]. Various cell types, including NIH 3T3 cells, HUVECs and explant-derived cardiac cells (EDCs), were used to assay the feasibility of the RGD modification to improve cell health and adhesion using this cell encapsulation system. As seen in figure 2(D), all visible cells are contained within an agarose microcapsule 2 h after encapsulation. Cell egress from the agarose microcapsules was observed at significant levels as early as 24 h post encapsulation (data not shown), however is much more obvious 48 h post encapsulation (figure 2(E)). When freefloating cells are mixed and incubated with unoccupied RGD-agarose microcapsules, we observed cells binding to and in some cases partially encompassing the empty capsules (figure 2(F)), however lower levels than observed when cells were encapsulated. At 48 h post encapsulation, cells that have egressed from their capsules can be seen attached to both the plate surface and to the exterior surfaces of the microcapsules when RGD is present (figures 2(G) and (H)). This is not observed in cells which are encapsulated in unmodified agarose (figures 2(I) and (J)).
We assessed cell viability and adhesion using live/dead staining. Cells were encapsulated using different agarose formulations and counted at intervals of 2 h, 24 h and 48 h post encapsulation. Using NIH 3T3 cells, we observed only a modest loss in cell viability at up to 48 h post encapsulation in RGS agarose ( figure 3(A)). Using maleimide modified agarose, an intermediate form before the addition of RGD, showed no effect on cell viability ( figure 3(A)). Preliminary evidence had indicated possible benefits of the addition of ECM protein to encapsulated cells, therefore we assessed the effects of the addition of ECM (fibrinogen and fibronectin) to the hydrogel solutions. Our results indicate that in unmodified agarose, the presence or absence of ECM has little effect on cell viability, while adding ECM to RGD-agarose promotes cell viability at 2 h post encapsulation, 78 ± 1% as compared to 57 ± 18% respectively. These results were consistent even when cells and microcapsules were incubated in non-polyhema coated plates (supplementary figure 1).
To assess cell adhesion to RGD-agarose it was important to monitor cellular egress. However, since no cells were observed to egress at 2 h post encapsulation, we assessed cumulative egress at 24 h and 48 h post encapsulation. We found that in all agarose formulations tested, significant egress was observed 24 h post encapsulation with only slight increases beyond that at 48 h ( figure 3(B)). The highest levels of egress was observed in cells encapsulated in unmodified agarose in the presence of ECM. The presence of RGD and ECM proteins had a modest effect of reducing the percentage of egress from 30 ± 10% to 23 ± 7% at 48 h post encapsulation. In the context of egress, the lack or presence of ECM only impacted cellular egress from unmodified agarose, where the addition of ECM increased egress from 11 ± 3% to the previous stated 30 ± 10%. As with cell viability, the maleimide modification alone had no impact on cell egress from their microcapsules. We assessed attachment of NIH 3T3 cells to capsules following egress by counting. We found that significant cell attachment could only be observed in cases when RGD-agarose was used. Furthermore, addition of ECM proteins to the gel mix appears to potentiate the ability of cells to adhere to the capsule surface enhancing the number of attached cells from 23 ± 10% with no ECM to 32 ± 6% with the addition of ECM ( figure 3(C)).
We proceeded to evaluate two therapeutically relevant cell lines: HUVEC and EDC similarly to NIH 3T3 cells, both HUVEC and EDC showed high viability of 84 ± 3% viable, 76 ± 3% viable in RGDmodified agarose respectively at 2 h post encapsulation ( figure 4(A)). As with NIH 3T3 cells, viability was assessed 24 h and 48 h after encapsulation. HUVECs initially encapsulated in unmodified agarose showed a trend of progressively reduced cell viability from 2 h to 48 h ranging from 96 ± 1% and 55 ± 15%. EDCs encapsulated in unmodified agarose showed a statistically significant reduction in cell viability ranging from 95 ± 1% and 54 ± 7%, for 2 h and 48 h respectively ( figure 4(A)). On the other hand, results for all cell types encapsulated in RGD-modified agarose showed no statistically significant changes in viability between 2 h and 48 h time points. HUVECs ranging from 84 ± 3% to 76 ± 7% and EDCs from 77 ± 3% to 72 ± 3%, for 2 h and 48 h respectively.
For the therapeutic cell types tested, cell egress from unmodified agarose microcapsules was slightly reduced, 24 ± 8% and 25 ± 4% at 48 h for HUVEC and EDCs respectively (figure 4). With HUVECs, we observed a maximal 30 ± 10% attachment at 24 h ( figure 4(B)), while EDCs exhibited 16 ± 1% attachment at 24 h (figure 4(C)). All data collected with cells encapsulated in unmodified agarose showed insignificant cell attachment for all cell types tested, in both environmental conditions, confirming negligible attachment of cells to unmodified agarose microcapsule surfaces. As an added measure, we assessed viability of encapsulated EDC cells using a CCK-8 assay. We observed no statistically significant change in live cell number up to 48 h post encapsulation in either agarose type. We measured number of live cells ranging from 12 600 ± 400 cells to 12 500 ± 500 cells between 2 h and 48 h in unmodified agarose, and from 11 300 ± 400 and 12 500 ± 300 cells between 2 h and 48 h respectively for RGD-agarose ( figure 5).
Since RGD is a functional group originally identified in fibronectin, we sought to illustrate that the cells were in fact using our synthetic RGD peptides in a similar fashion to their in vivo counterparts. Using a focal adhesion staining kit we were able to identify focal adhesion points which can been seen between the egressed cells and the capsules which are present. Focal adhesions were clearly seen along the perimeter of the cells, shown by the anti-vinculin stain (green)

Discussion
The high level of cell viability following encapsulation indicates that the microfluidic encapsulation process and sample preparation were not detrimental to cell viability, consistent with our previous findings [15,20]. Additionally, HUVEC cells showed similar trends in viability and their ability to adhere to the capsules following egress. There were statistically significant differences between agarose types at 2 h for both NIH 3T3s and EDCs. While HUVEC cells did show similar trends to NIH3T3 and EDC cells, the differences were not statistically significant. The initial loss of viability for EDC cells however, was not observed during CCK-8 analysis suggesting that it may not be a biologically relevant change. Additionally, the CCK-8 assay results differ from those observed by fluorescence viability assay, the latter showing a significant decrease in EDC viability over 48 h when encapsulated in unmodified agarose compared to RGD agarose ( figure 4). This suggests that while RGD may have an impact on initial viability of the cells, it preserves cellular viability over time compared to unmodified agarose as we expected. Cells encapsulated in RGD-modified agarose did not demonstrate significant differences in cell egress compared to cells encapsulated in unmodified agarose microcapsules, suggesting the presence of RGD-binding sites within the microcapsules did not have a major impact on the cells' ability to egress the microcapsules. This is consistent with past findings, which showed that 25% of NIH 3T3 cells and 30% HUVECs egressed from 50 µm unmodified agarose microcapsules within 48 h [20].
In vivo studies suggest that free cells injected into damaged tissue have low rates of engraftment [29,30], and are cleared quickly from the site of damage [15,29,31]. A previous study suggested that agarose microcapsules tend to remain in the targeted tissue for days, whereas cells that egress these microcapsules are quickly cleared via vasculature [15]. Here we show that cells encapsulated in RGD agarose also egressed their microcapsules (figure 2), but these cells were observed to subsequently adhere to the outside of these microcapsules (figures 2(G) and (H)). Cellular protrusions can be observed in these images, as egressed cells have partially enveloped RGD agarose microcapsules. Small clumps of attached cells were commonly observed, as well as cells linking multiple microcapsules ( figure 2(H)). In contrast, cells encapsulated in unmodified agarose showed little to no evidence of cell attachment to the microcapsules,  nor was any interaction with microcapsules observed (figure 2(I)). This lack of cellular adhesion was observed for all cell types investigated.
In all experiments, the hydrogel solutions were supplemented with ECM proteins (fibrinogen and fibronectin). When using unmodified agarose, no perceivable effect on viability, egress or cellular adhesion to the outer surface of the microcapsules was observed with or without the addition of ECMs, consistent with previous observations [20]. When using RGD agarose, cellular adhesion to the outer surface of the microcapsule was observed regardless of whether the ECM proteins were added. However, adhesion was enhanced by the addition of the ECM proteins ( figure 3(C)). This may due to the manner by which cells form focal adhesions. Briefly, the RGD motif interacts with integrin receptors on the surface of the cells [32,33]. This in turn stimulates the recruitment of additional integrin receptors which bind to soluble fibronectin fibers. As this process continues and more fibronectin fibers accumulate, the fibronectin subunits interact with each other producing an insoluble mass providing the cells with a stable adhesion substrate [34]. In this context, the RGD-agarose likely provides a nucleation point to promote the formation of focal adhesion sites and facilitates adhesion to the agarose capsules.

Conclusion
Ensuring the retention of therapeutic cells in the targeted tissue is key to optimizing the efficacy of the treatment. In many cell-based therapies, the therapeutic benefit is often ephemeral as cells introduced to the site of injury are cleared out of the targeted tissue via vasculature. Encapsulating therapeutic cells in hydrogel microcapsules has been shown to lengthen the retention time of cells within the targeted tissue.
Unfortunately, cells have been observed to egress from these capsules in shorter timeframes than the expected clearance of the capsules which again allows them to be cleared from the target site, limiting their long term potency [15]. In the present study, we demonstrated through the introduction of cell binding moieties to the hydrogel material, specifically the RGD sub-group, that therapeutic cells adhere to the surface of the hydrogel microcapsules upon egress. This allows for an enhanced long term cell survival advantage.

Data availability statements
All data that support the findings of this study are included within the article (and any supplementary files).