A rapid and sensitive method to measure numbers of live cells in alginate capsules following depolymerization with ethylenediaminetetraacetic acid

Cells are encapsulated after extrusion of a cell suspension in a monomer solution of alginate through a needle to generate microspheres that are cross-linked in a bath containing divalent cations [1]. Microspheres containing islets have enabled long-term glycemic control in diabetic rodent models and in nonhuman primates [2]. Encapsulated mesenchymal stem/stromal cells (MSCs) have immunomodulatory effects in inflammatory animal models that promote recovery of function in spinal cord injury, myocardial infarction and hindlimb ischemia [3–6]. MSCs have been approved for clinical treatment but large doses are required for efficacy and the fate of the cells is uncertain, with the vast majority of them disappearing soon after injection.

Hundreds of clinical trials are testing the safety and efficacy of MSCs and very large doses of these cells may be needed for efficacy, at a considerable cost [7]. Encapsulation in alginate is advantageous for transplantation by restricting cells locally to sites of injection and protecting the cells from the host for long-term survival. This enables extended and sustained delivery of secreted factors, which may be effective using lower doses of MSCs when they are encapsulated by comparison with free (unencapsulated) MSCs [6].

To evaluate efficacy in preclinical studies, one needs to measure the number of live cells/capsule during the production process, their recovery after storage and their efficacy in vitro and in vivo. This requires accurate assays that measure the number of cells/capsule during optimization to scale up production for clinical development [8]. Rapid assays are also needed to measure live cells before transplantation and after recovery of encapsulated cells from transplanted subjects to count live cells/capsule ex vivo.

Large capsules are advantageous for long-term cell survival in vivo because they accommodate many cells, including islets, and allow more cells to be encapsulated/volume of alginate than in smaller capsules. In addition, adhesion of host cells due to foreign body reactions, for example, by macrophages, can occlude the surface of alginate capsules. Foreign body reactions are reduced dramatically when capsules contain MSCs compared with other cells because the MSCs are anti-inflammatory and suppress activation and adhesion of macrophages to surfaces of alginate capsules [9].

Most current methods to count the number of cells/capsule use confocal imaging to acquire multiple optical sections from the bottom to the top of a capsule and then count the number of cells after the 3D images are collapsed into a single optical plane [3,4]. However, overlapping fluorescent cells in different optical planes make it difficult to resolve individual cells even at higher magnifications (Figure 1). When capsules exceed ∼0.4 mm in diameter, confocal images of extreme planes cannot be collected causing some cells to be excluded. Moreover, the confocal method requires long acquisition times to cover all the X–Y planes along the Z-axis, and confocal time is expensive. Thus, confocal-based methods undercount overlapping cells, are time-consuming and are only applicable to capsules <0.4 mm to yield accurate cell counts/capsule.

The current authors have developed an alternative method to measure the number of cells/capsule. Given that alginate polymers are cross-linked by divalent cations to form capsules, the cross-links can be dissociated by the chelation of divalent cations with ethylenediaminetetraacetic acid (EDTA). Treatment with 50 mM EDTA of 2.25% (w/v) alginate capsules formed with 50 mM BaCl₂ depolymerizes cross-linked alginate, thereby collapsing the capsules and releasing the cells onto the bottom of a tissue culture dish. The cells become distributed in an area that is approximately nine times larger than the maximal cross-section of the capsule, making it much easier to count individual cells. A program was also developed to count the cells much faster than current methods using confocal microscopy.

Materials & methods/experimental

Alginate preparation

Alginate (Pronova UP LVG) was purchased from FMC Biopolymers (Muiden, The Netherlands) and dissolved in Ca²⁺ free Dulbecco’s modified Eagle medium (DMEM; Life Technologies, NY, USA) to create 2.5% (weight/volume) alginate on a hot plate/stirrer at 50°C with a stir bar for 3–4 h. The sterile viscous solution was then filtered at 37°C for ∼2 h using a 0.22 μm vacuum filter (Millipore) and stored at room temperature.

Cell culture

GFP-expressing cell line (C6-GFP) [10] were grown in DMEM supplemented with 10% fetal calf serum (Atlanta Biologicals, USA), and penicillin-streptomycin-glutamine (100X; Gibco, MA: 10378016). C6 cells were grown in tissue culture dishes in an incubator at 37°C and 5% CO₂ (complete media). C6 cells were allowed to proliferate to 70–90% confluency. For encapsulation in alginate, the cells were detached using trypsin-EDTA (Gibco, MA, USA), washed in media DMEM (Gibco, MA, USA) without serum, counted and resuspended in DMEM without serum at a cell density of 5 × 10⁷ cells/ml; 100 μl of this cell suspension was then diluted into 0.9 ml of 2.5% (w/v) alginate to yield a final cell density of 5 × 10⁶ cells/ml as a standard condition. Human bone marrow-derived MSCs were purchased from Texas A&M at passage 1 and cultured in minimum essential medium (MEM)-a (Gibco Life Technologies, MA, USA) without deoxy- or ribonucleosides, and supplemented with 10% fetal bovine serum (FBS; Atlanta Biologicals, GA, USA), 1 ng/ml basic fibroblast growth factor (ßFGF), 2 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin (Gibco, MA, USA). MSCs were plated at 5000 cells/cm² and allowed to proliferate to ∼70% confluence. Only low passages (4–5) were used for experiments.

Alginate microencapsulation

Cells were mixed into alginate solutions by gentle swirling and pipetting using a 1-ml pipette tip to yield a final cell suspension of 2.25% alginate. Alginate droplets were generated using a Spraybase electrostatic encapsulator (Profector, Spraybase, MA) at an applied voltage of 8.00 kV, constant positive gauge pressure of 0.750 bar, a 24-gauge needle at a height of 5 cm and at room temperature [8]. The extruded microdroplets were subsequently cross-linked into a 50-ml collecting vessel containing 50 mM BaCl₂, 145 mM NaCl, 10 mM MOPS and 14 mM glucose where they were incubated for 10 min to allow additional time for cross-linking at room temperature Chemicals were purchased from Sigma Aldrich (St. Louis, MO, USA). BaCl₂ was used for cross-linking because it binds with higher affinity than CaCl₂ to the Pronova UP LVG with a high (∼60%) G content. Spherical microcapsules were washed with HEPES-buffered cell media (Gibco, MA), allowed to settle under gravity and were resuspended in 20 ml of complete media and incubated upright in 25 cm² tissue culture flask.

Depolymerization, image acquisition & counting

To develop this assay, C6-GFP was used, which allowed the identification of cells by fluorescence without additional staining [10]. Cell viability in the capsules was assessed using two different methods. In the first, called the confocal method, the number of live cells was counted from projections of confocal images using a LSM 510 as described in [6]. In the second method, capsules were collapsed from a 3D sphere to a 2D circle. The capsules were first washed with 1 × PBS without calcium chloride or magnesium chloride (ThermoFisher Scientific, Catalog # 14190) to ensure most free divalent cations are not present. The capsules were transferred into a 96-well plate with 2–3 capsules in each well. The PBS solution was aspirated and subsequently replaced with 200 μl PBS containing EDTA (Fisher, Catalog # S-311) or sodium citrate (Thermo Scientific), chelating solutions of varying concentrations and incubated for varying times. The chelating solutions were prepared using varying concentrations of EDTA or sodium citrate, ranging from 25 to 100 mM, dissolved in PBS without calcium or magnesium (Gibco). The collapsed capsules were imagined using a Zeiss AxioVision Inverted Fluorescent Microscope IX81 (Olympus, Tokyo, Japan). For both methods, the images were uploaded into a MATLAB algorithm and counted. Further details of the collapsed method and MATLAB algorithm can be found in the appendix. Individuals doing the counting were blinded to the identity of the samples.

Results & discussion

Current methods to visualize and count cells in capsules use confocal microscopy. This is challenging because overlapping fluorescent cells in different optical planes makes it difficult to resolve individual cells, even at higher magnifications (Figure 1). We developed a fast, sensitive and inexpensive method to count the number of live cells in individual capsules. A brightfield image of an intact capsule shows the cells and the outer surface of the capsule (Figure 2A).

Figure 2 illustrates the principle of the collapse assay. Capsules tend to adhere to the bottom of plastic tissue culture wells. After the gentle addition of 50 mM EDTA solution, capsules swell to ∼2.5–3-times their original size within 5 min. With total incubation for 15 min, each capsule collapses in place and the cells within them settle into a single circular plane at the bottom of the well. The schematic shows that, after depolymerization, the cells spread over a much larger area, approximately ninefold (Figure 2C) larger than that of the maximal intact capsule cross-sectional area (Figure 2B). This reduces the density of cells in a 2D plane and essentially eliminates overlapping cells by comparison to the optical projection (collapse) after confocal microscopy, which is restricted to a much smaller area defined by the diameter of the intact capsule. Cells derived from individual capsules are then imaged in a single 2D plane with epifluorescence microscopy and counted to yield numbers of cells in individual capsules.

During assay development, we compared 25–100 mM EDTA with another calcium chelator, sodium citrate, but the latter was much less effective in depolymerizing capsules (Table 1). Given that the more widely used EDTA was promising, we continued to optimize the methods using EDTA. However, depolymerization for 30 min was less effective for a given concentration of EDTA in wells containing larger numbers compared with fewer capsules. We used a semiquantitative assay to score response to EDTA as - swollen capsules (+), the disappearance of capsules without settlement of all cells (++) and settlement of all cells into a single plane (+++). We found that 100 mM EDTA was most effective for depolymerization using 5–15 capsules/well. The extent of depolymerization decreased progressively at lower EDTA concentrations and with higher numbers of capsules/well (Table 1). Depolymerization is time-dependent and was incomplete with 25 mM EDTA even with only 5 capsules/well when incubated for 20 min. The results suggest that an appropriate ratio of cross-linked alginate to EDTA is critical for optimal depolymerization. Complete depolymerization occurred within 5 min with 100 mM EDTA (Table 2), but this is a harsh treatment for cells. Given that incubation for 15 min in 50 mM EDTA yielded complete collapse, we chose it as the standard condition.

We then compared cell counts on the same capsule preparations (diameter of 396 ± 36 μm) using the confocal method (Figure 3A) and the collapse method (Figure 3B). The results show that 13% more cells were counted with the collapse method (Figure 3C), consistent with the prediction that overlapping cells would be underestimated in intact capsules by confocal microscopy (Figure 3A). When we used the collapse method with larger capsules of 595 ± 27 μm in diameter containing more cells, 27% more cells were counted with the collapse versus the confocal method (Figure 4). The relatively larger numbers of cells detected by the collapse method in the ∼600 versus the ∼400 μm capsules is probably due to limitations with the depth to which the confocal microscope can penetrate a capsule, which we estimate is ∼400 μm and the overlap of the cells after individual confocal images are merged. Undercounting in a confocal image may also be due to the overlap of cells in the confocal images (Figure 1) Thus, the confocal method is significantly less accurate than the collapse method for capsules of at least 600 μm and this discrepancy should increase with even larger capsules and higher density of cells. An example of depolymerization of larger capsules (801 ± 53 μm, containing 1388 ± 206 cells/capsules) is shown in Supplementary Figure 1A. The confocal method for visualization of these capsules showed extensive overlapping making it difficult to count. Moreover, the depth of view only penetrates about half the ∼800 μm capsule. Interestingly, there is an exponential increase in the number of cells/capsule with increasing diameters (Supplementary Figure 1B).

The counting methods used here for the confocal and collapse assays are the same insofar as the area of the capsule is encircled, the cells are marked manually and a software program is then used to count the marked cells. We purposely used the same counting method to ensure that any differences in the counts were not due to different counting methods. Whereas confocal imaging packages typically have sophisticated counting programs, they still are not accurate for counting cells in capsules >400 μm. Therefore, we designed a simple counting method that does not require purchasing an imaging package. Dispersal of cells after capsule collapse reduces cell density dramatically, making cell identification and counting easier and more accurate when overlapping cells are not present.

We developed an objective MATLAB algorithm that identifies and counts fluorescent cells even in large capsules after depolymerization. After encircling the region of the depolymerized capsule (Supplementary Figure 1A), the first step was to increase the number of pixels in the image (Supplementary Figure 2A). We used algorithms to increase image resolution (Supplementary Figure 2B) and increase the signal-to-noise ratio (Supplementary Figure 3B). The circumference of the cells is used to define the external watershed marker (Supplementary Figure 3C) and the cell centroid of each cell is used to define the internal watershed marker (3D). These images are subsequently merged (Supplementary Figure 3E) and a watershed algorithm is implemented to segment the cells (Supplementary Figure 3F). The number of segmented areas is counted. Further details about the algorithm are found in the Supplementary Materials.

Calcein-ethidium staining was used to identify live MSCs (Supplementary Figure 1A), which do not express an intrinsic marker such as GFP. Cells stained with this method must be sampled in a timely manner before the signal decays, which is more problematic with the longer acquisition times for confocal microscopy versus imaging a single plane using the collapse assay. The major advantage of this new method to measure the number of cells/capsule is based on a simple technique to collapse a complex 3D alginate capsule into a 2D plane. The collapse distributes cells in a nonoverlapping manner in a single layer at a much lower density over a larger area that eliminates cells from overlapping along the Z-axis. A major advantage of this method is its application to measure number of cells/capsule accurately in capsules as large as 0.8 mm in diameter (Supplementary Figure 1A). Such larger capsules are advantageous because they are less susceptible to capsule coating by reactive cells in the host, which can compromise cell viability and secretion of bioactive factors [11]. With a fixed input concentration of cells, the number of cells in larger capsules increases exponentially as does the volume of polymerized alginate (Supplementary Figure 1B).

Most current methods use confocal microscopy [12], which underestimates cell counts because of overlapping cells in optical planes that are not captured in capsules >0.4 mm in diameter. Reducing 3D capsules into larger 2D planes eliminated this serious drawback. Cell counting is much easier and more accurate by spreading the cells over a ∼ninefold larger 2D area by comparison to the diameter of the original capsule.

We posit that the limitations of the confocal methods regarding capsule size and relatively long times for analysis make it impractical to support the clinical translation of encapsulated cells, especially if they are larger than 0.4 mm in diameter. Rapid and high throughput assays will be needed for large numbers of samples during preclinical development to measure secretion of factors (e.g., cytokines) from cells in vitro or ex vivo [8], which will require normalization to the number of live cells/capsule to measure cytokine yield/cell. Rapid assays may also be advantageous during optimization for scale-up and quality control at various stages up to patient delivery. Given that the collapse method does not require a confocal microscope, it is feasible to perform such assays in clinical labs, many of which already perform cell imaging in wells.

The capsules used here contain 2.25% alginate, which is a commonly used concentration for cell encapsulation. More dense capsules are rarely used and would require higher concentrations of EDTA for collapse. Less dense capsules would be more susceptible to EDTA collapse and will need shorter incubation times and/or lower EDTA concentrations. Optimization for these different situations should be straightforward using the methods described here.

C6-GFP cells were used to develop this assay because they are intrinsically fluorescent and easy to obtain in large quantities [10]. We have also used this method for hybridomas, which secrete antibodies, larger cells including Chinese hamster ovarian cells, which have been used extensively to express genes encoding bioactive proteins and MSCs that secrete anti-inflammatory and growth factors. We have used calcein-ethidium assays to stain for live cells, which require additional staining time for this assay, but it would require twice as much confocal microscopy time to collect a second set of 3D images to measure dead cells in capsules using a second fluorescence channel.

Conclusion

The collapse technique for encapsulated cells into a 2D plane combined with the segmentation counting program overcomes the overlap problem with the optical collapse of 3D confocal images. Moreover, it enables counting in capsules >0.4 mm in diameter that cannot be measured accurately by confocal methods. It provides rapid and accurate measurements of number of cells/capsule, which will enable the normalization of cell activity (e.g., secretion of factors/cell) in vitro and ex vivo after transplantation. Additional optimization of depolymerization conditions may be required with different types of alginates and crosslinkers but that should be straightforward as we have done here by optimizing EDTA concentrations and times of exposure.

Future perspective

It is important to prolong the survival of encapsulated cells that act by secreting factors including islets and MSCs so that they can have long-term effects in vivo. The techniques described here will allow one to determine the number of live cells/capsule, for example, for scale-up, which is necessary for clinical development. One can also determine whether decreased activity over time in vivo is due to the survival of fewer cells or decreased activity/cell. These and future improvements in capsule technologies will make the analysis of transplantation of various types of cells feasible as a method for long-term cell therapy.