Adipose-Derived Stem Cells Integrate into Trabecular Meshwork with Glaucoma Treatment Potential

Abstract

The trabecular meshwork (TM) is an ocular tissue that maintains intraocular pressure (IOP) within a physiologic range. Glaucoma patients have reduced TM cellularity and, frequently, elevated IOP. To establish a stem cell-based approach to restoring TM function and normalizing IOP, human adipose-derived stem cells (ADSCs) were induced to differentiate to TM cells in vitro. These ADSC-TM cells displayed a TM cell-like genotypic profile, became phagocytic, and responded to dexamethasone stimulation, characteristic of TM cells. After transplantation into naive mouse eyes, ADSCs and ADSC-TM cells integrated into the TM tissue, expressed TM cell markers, and maintained normal IOP, outflow facility and extracellular matrix. Cell migration and affinity results indicated that the chemokine pair CXCR4/SDF1 may play an important role in ADC-TM cell homing. Our study demonstrates the possibility of applying autologous or allogeneic ADSCs and ADSC-TM cells as a potential treatment to restore TM structure and function in glaucoma.

INTRODUCTION

Glaucoma, the leading cause of irreversible blindness worldwide (1), is a progressive optic neuropathy with loss of retinal ganglion cells and axons, resulting in visual field loss. In the United States, the most common type of glaucoma is primary open angle glaucoma (POAG). Age and elevated intraocular pressure (IOP) are the primary risk factors for POAG (2). Multiple randomized clinical trials have shown that IOP lowering is effective in delaying or preventing the onset of vision loss in individuals with glaucoma (3, 4).

The trabecular meshwork (TM) consists of three elements, with the juxtacanalicular connective tissue (JCT) thought to be the major site of resistance to aqueous outflow (5-7). TM cells help regulate IOP/aqueous outflow, through functions such as phagocytosis of debris and foreign materials (8), modulating permeability of Schlemm’s canal endothelium (9) and producing extracellular matrix (ECM) (10, 11).

In glaucomatous eyes, the TM displays several pathologic features in addition to elevated flow resistance. Notably, TM cellularity is reduced in glaucomatous eyes compared to age-matched control eyes (12). This likely leads to adhesion of trabecular lamellae, thickening of trabecular beams and possibly accumulation of fibrillary plaque material, all of which disturb TM microstructure (13-15). The mechanical properties of the TM itself are altered: glaucomatous TM has increased stiffness that may be associated with increased outflow resistance (16-18).

Current pharmacological and surgical therapies seek to lower IOP by facilitating aqueous outflow and suppressing aqueous humor production (19); recently, two classes of compounds (Rho kinase inhibitors (20-22) and nitric oxide agonists (23, 24) have been developed to specifically target conventional outflow pathway tissues to reduce IOP. However, these strategies fail to directly target TM cellularity loss, hypothesized to be an important factor in the pathophysiology of the IOP elevation. Theoretically, repopulation of the TM by stem cells could compensate for decreased cellularity in glaucomatous eyes and restore TM function, thus normalizing IOP. Such an outcome would allow the TM to dynamically respond to IOP changes and avoid the continued dosing needed with current medical therapies. Further, since stem cells are self-renewing with an asymmetric division property (25), another advantage of stem cell-based therapy would be to maintain a stem cell pool to ensure the treatment is efficacious in the long term.

It has been reported that there are tissue-specific stem cells in the TM (26-30). Several groups have successfully isolated and characterized human TM stem cells (31-33). After injection in the normal mouse anterior chamber, these stem cells can home to TM tissue and maintain IOP in the normal range (34). After injection into the anterior chamber of mice with partially damaged TM tissue due to laser photocoagulation, TM stem cells can specifically home to laser-damaged TM regions and repair the tissue (35).

As alternatives to harvesting TM stem cells, which are likely depleted/absent in glaucoma patients, other stem cell types have been explored for TM regeneration, such as mesenchymal stem cells (MSCs) and induced pluripotent stem cells (iPSCs). After bone marrow-derived MSCs were injected into laser-treated rat anterior chambers, IOP rapidly declined to normal levels and TM structure was restored at one month (36). iPSCs can be induced to differentiate into TM cells in vitro (37), and can restore TM function ex vivo (38) and in vivo (39, 40). Thus, both MSCs and iPSCs are potential candidates as cell therapy resources.

Another potential cell therapy candidate is adipose-derived stem cells (ADSCs). ADSCs can be obtained in large quantities with minimally-invasive procedures (41-43) and undergo multilineage differentiation (44). We have successfully induced ADSCs to differentiate into corneal keratocytes (45), a cell type derived from neural crest and thus sharing the same embryonic origin as TM cells. Snider et al (46) reported that adipose-derived mesenchymal stem cells can be specifically delivered to the TM tissue using magnetic nanoparticles, which suggests the feasibility of using ADSCs for TM regeneration. Here, we demonstrate that ADSCs can be induced to differentiate into TM-like cells (ADSC-TM) with characteristics of TM cells, namely gene expression patterns, response to dexamethasone (DEX) stimulation, phagocytic capacity, and ability to maintain aqueous humor dynamics in vivo.

It has been reported that chemokine stromal cell–derived factor 1 (SDF1) and its receptor C-X-C chemokine receptor type 4 (CXCR4) play important roles in bone marrow hematopoietic stem cell homing (47) and in TM stem cell homing to the TM (35). Here, we compared the specific integration of primary ADSC and ADSC-TM cell grafts into mouse TM tissue and explored possible mechanisms of cell homing. This study motivates further research into a potential regenerative therapeutic strategy to restore glaucomatous TM tissue function and thus protect vision loss.

MATERIALS and METHODS

Cultivation of Human ADSCs and TM Cells

Human TM cells were isolated and cultured as previously described (31). TM cells isolated from three donors of 46, 58 and 62 years of age were used in the experiments. Human ADSCs were obtained from three individuals of 34, 36 and 38 years of age undergoing elective lipoaspiration surgeries with informed consent under a protocol approved by the Institutional Review Board (IRB) of the University of Pittsburgh, consistent with the principles of the Declaration of Helsinki. ADSCs were isolated by collagenase digestion and differential centrifugation (48) (45). Both TM cells and ADSCs in this study were used between passages 3 and 5.

Differentiation of ADSC-TM Cells

Three methods were used and compared to generate ADSC-TM cells. (1) Co-culture: A 3-Dimensional co-culture environment was established by using cell culture inserts (0.4 μm pores; Corning) placed in 6-well plates at 2 x 10³ cells/cm². TM cells were seeded into the insert with the same seeding density without contacting the ADSCs that were on the bottom of the culture wells. (2) ECM+CM: ECM generated from TM cells at passages 3-5 was prepared by completely lysing confluent TM cells using 0.02N ammonium hydroxide + 0.05% Triton X-100 for about 5 minutes and washing away the lysed cell debris using phosphate buffered saline (PBS). TM-conditioned medium (CM) was collected 3 days after TM cells were passaged at 5 x 10³ cells/cm² at p3-p5 to about 80% confluence and centrifuged at 10,000 RPM for 30 min to remove any possible remaining cells in the medium; CM was stored at 4°C for future use within 1 month. ADSCs were then seeded at 2 x 10³ cells/cm² and cultured on TM-secreted ECM in medium composed of 50% DMEM/F12-10% FBS and 50% CM. (3) ECM+AdvM: ADSCs were seeded at 2 x 10³ cells/cm² and cultured on the ECM in Advanced MEM (AdvM, ThermoFisher) with 0.1M ascorbic acid-2-phosphate (Sigma-Aldrich) without serum. AdvM has previously been used for differentiation of neural crest derived keratocytes (49, 50). Culture medium was changed every 3 days and induction efficiency in the above three approaches was assayed at day 10.

Phagocytosis Assay

ADSC-TM cells, primary ADSCs and TM cells at the same passage number (passage 3-5 were used and compared at the same passage) were cultured on coverslips in 6-well plates until they were 70% confluent. pHrodo Red S. aureus Bioparticles conjugate (ThermoFisher) were diluted in DMEM/F12 to make a 1mg/ml dispersed suspension and incubated with cells for 4-hr at 37°C. To completely remove the unphagocytozed bioparticles, cells were then trypsinized and reseeded onto coverslips in 6-well plates. After incubation at 37°C for 3 hours, cells were fixed in 4% paraformaldehyde (PFA) and stained with DAPI at 1 μg/mL and phalloidin conjugated with Alex-488 at 1:500 for 1-hr. Using a confocal microscope (Olympus), the internalization of bioparticles was observed and imaged, and the quantification of phagocytosis was established by calculating the proportion of bioparticle-ingesting cells as a fraction of total cell number in randomly selected fields.

Cross-Linked Actin Network (CLAN) Formation by DEX Stimulation

Responsiveness to DEX stimulation is one of the characteristics of TM cells. Differentiated ADSC-TM cells were exposed to 200 nM DEX for 11 days. The 200 nM concentration we used differs from other studies that used 100nM (51-53). Our primary ADSCs were cultured initially with 100nM DEX to reduce possible fibroblast contamination (54) and DEX was removed from passage 2 to 3. Differentiated ADSC-TM cells without DEX but cultured in the same medium as DEX-treated cells (containing the DMSO which Dex was dissolved in) were used as controls. Cells were stained with Phalloidin-488 and DAPI. Images were taken under a confocal microscope. The minimum requirement for an actin structure to be counted as a CLAN was the presence of three intensely fluorescent vertices connected by three actin spokes (55). The number of CLAN-forming cells and total cells per field on at least 10 fluorescent fields from each treatment group was counted and calculated. Data were pooled from three experiments and represent mean ± SD. CLAN counting was done in a masked manner.

Transplantation of Cells into Mouse Anterior Chamber

All experimental procedures were reviewed and approved by the University of Pittsburgh Institutional Animal Care and Use Committee and carried out according to guidelines of the Association for Research in Vision and Ophthalmology Resolution on the Use of Animals in Ophthalmic and Vision Research. Intracameral injection followed the procedures previously described, with transplanted cells being prelabeled with DiO (34, 35). 10,000 cells (ADSC-TM cells, primary ADSCs, fibroblasts, n=25 in each group) in 2-μL DMEM/F12 were injected. An equal volume of DMEM/F12 was injected as sham control. Mice were anesthetized for IOP measurement as previously described (34) and sacrificed at day 30 after transplantation. The selection of 30 days as the timepoint to examine the transplantation effect was based on our previous work (56), which indicated that TM stem cells at week 4 were able to integrate into the TM region with some maintaining stem cell status and some differentiating into TM cells expressing AQP1 and CHI3L1. Eyes were enucleated and either perfused to measure outflow facility followed by dissection for RNA extraction and qPCR or fixed to create cryosections for staining.

Another set of cell injection experiments was conducted to obtain tissue for cryosections and qPCR to detect cell distribution and inflammatory response at 3 days after injection, and for wholemount staining to detect injected cell distribution at 1 week and 2 weeks after injection. There were at least 3 eyes for each condition at each time point.

Measurement of Aqueous Humor Outflow Facility

Outflow facility measurement followed existing procedures (35, 57) with minor modifications. The perfusion system consisted of a computer-controlled syringe pump (Harvard Apparatus) that delivered a variable flow rate (Q) to the anterior chamber to maintain a desired IOP, as monitored by a pressure transducer (Honeywell) connected to a computer control system (Labview). The anterior chamber was cannulated with a microinjection glass pipette connected to the pressure transducer. A 25-μl Hamilton syringe was loaded onto the syringe pump. Eyes were perfused with PBS at constant pressures of 4mmHg, 8mmHg, 15mmHg and 25mmHg sequentially for about 20~30 minutes at each pressure level. The average flow rate at each set pressure was calculated. We used the Goldmann equation as described (57, 58), namely F = (Po-Pv)C + Fu, in which Po is the IOP (mmHg), F is the rate of aqueous formation (equivalent to pump flow rate at steady state in enucleated eyes), C is the conventional outflow facility, Pv is the episcleral venous pressure (equal to zero in enucleated eyes), and Fu is the pressure-independent (unconventional) outflow rate. Linear regression was used to fit the pressure-flow data and hence estimate outflow facility: C = (F-Fu)/Po. Facility is the slope of the regression line. Data from a given eye were only accepted when the regression demonstrated R² > 0.95.

Quantitative Reverse Transcription–Polymerase Chain Reaction (qPCR)

Cells and mouse TM tissues were lysed with RLT buffer (Qiagen) and RNA was isolated following the manufacturer’s instructions. cDNAs were transcribed from the RNAs using cDNA MasterMix (WorldWide). qPCR was performed by direct dye binding (SYBR Green, Applied Biosystems) and analyzed as previously described (34, 59). Primers were designed using online software (Primer3) and blasted via NIH website (https://www.ncbi.nlm.nih.gov/tools/primer-blast/)) with the sequences shown in Table S1. Sequences of primers of mouse CD45, CD11b, CD11c and CD3 were used from published data by Wu et al (60) and sequences of primers of mouse F4/80 were from published data by Shaul et al (61). Primers were purchased from Invitrogen (Carlsbad, CA).

Western Blotting

Cell lysates were collected using RIPA (Santa Cruz) and sonicated until solubilized. Samples were mixed with 2x Laemmli loading buffer (BIO-RAD) and loaded onto 8-16% Tris-Glycine Gels, and electrophoresis was performed for 1 hour at 200 V. Protein was transferred to a polyvinylidene difluoride membrane (Millipore) and blocked for 1 hour followed by incubation with primary antibodies overnight and incubation with fluorescent secondary antibodies. The fluorescent signal was captured using an infrared imager (LI-COR).

Immunohistochemistry

Cultured cells were rinsed briefly in PBS and fixed in 4% PFA before staining. Enucleated mouse eyes were fixed in 1% PFA and embedded for 8 μm cryosections. Cells or sections were incubated overnight at 4°C with primary antibodies (Table S2) followed by fluorescent secondary antibodies and nuclear dye (DAPI). Samples were imaged using a confocal microscope (Olympus). Image analysis was carried out with FV10 Viewer software (Olympus).

Transmission Electron Microscopy (TEM)

The ultrastructure of mouse TM was examined by TEM, following previously described methods (62). Sections were viewed on a JEOL JEM 1011 transmission electron microscope (JEOL).

Wholemount Staining

The anterior eye, including the cornea and TM, was dissected and incubated with DAPI for 30 minutes. After thoroughly washing with PBS, the tissues were flat-mounted and imaged on a confocal microscope (Olympus) with a movable stage for stitched stacks and analyzed using FV10 Viewer.

Assessment of Apoptosis by TUNEL Staining

The TUNEL assay was performed using a cell death detection kit (Roche) following the manufacturer’s protocol for cryopreserved tissue. At least 3 eyes and 9 cryosections of each condition were stained, imaged and counted.

Cell Migration Assay

Chemotaxis of TM cells to ADSCs and ADSC-TM cells was evaluated by a cell migration assay on a Transwell cell culture system. Primary TM cells; TM cells treated with SDF1α and SDF1β (Millipore) at 80ng/ml each (TM-SDF1αβ) to increase the TM cell expression of SDF1; or TM cells treated with SDF1 antibody (Millipore) at 1μg/ml (TM-SDF1Ab) for 72 hrs to block the TM cell surface SDF1 expression were seeded at the bottom of 24-well culture plates as chemoattractant. ADSCs, ADSC-TM cells, or the cells treated with the CXCR4 antagonist II IT1t (35, 63) at 44nM for 72 hrs were labeled with DiO and seeded in the cell culture inserts with 8-μm pores. After 24-hr incubation, the Transwell membranes of the culture inserts were fixed, stained with DAPI and mounted. Cells migrating to the other side of the membrane and nonmigrating cells were counted in five randomly selected microscopic fields at 20x magnification. The ratio of migrated cell number to the sum of migrated and nonmigrated cell numbers was taken as the percentage of migrating cells. The experiment was repeated twice with different ADSC strains from different donors.

Cell Affinity Assay

We tested the affinity between TM cells with ADSCs and ADSC-TM cells. Primary or pretreated TM cells as described above were seeded as feeders. After attachment and growth for 24-hr, ADSCs, ADSC-TM, or cells treated with IT1t were labeled with DiO and added on top of the feeder cells for 60-min. Then the cultures were washed thoroughly to remove any unattached cells, fixed and stained with DAPI, and imaged. The experiment was repeated twice with different ADSC strains from different donors. The attached DiO+ cells were counted and averaged and compared between groups.

Statistical Evaluation

Except as noted, at least three biologically independent experiments were performed for both in vitro and in vivo data. All statistical analyses were performed with one-way ANOVA followed by different post-tests as stated in the figure legends. Values were considered statistically significant if p was less than 0.05.

RESULTS

ADSC Differentiation into TM Cells

ADSCs were induced to differentiate towards a TM cell phenotype using three different conditions: (1) co-Culture with TM cells (Co-Culture), (2) exposure of ADSC cells to extracellular matrix and conditioned media from TM cells (ECM+CM), and (3) exposure of ADSC cells to extracellular matrix from TM cells and culture in Advanced MEM (ECM+AdvM). After 10 days of induction, expression of the stem cell marker OCT4 in induced TM cells was reduced and expression of the TM cell markers CHI3L1 and aquaporin 1 (AQP1) was increased, approaching levels in primary TM cells (Fig. 1A). Western blotting (Fig. 1B) showed increased protein expression of CHI3L1 and AQP1 in induced cells, displaying a pattern similar to human TM cells, but different from that seen in ADSCs. Immunofluorescent staining indicated that ADSC-TM cells expressed CHI3L1 and AQP1, similar to primary TM cells (Fig. 1C). In addition, staining demonstrated reduced expression of OCT4 in induced cells compared to ADSCs. The nestin expression was not dramatically reduced overall after induction. TM cells, originating from neural crest, expressed nestin as well (Fig. 1C). The number of cells staining positive with each antibody was counted from 7-16 fields of view in 2-3 cell strains per condition, from which the percentage of positive staining cells was computed (Fig. 1D).

Figure 1. ADSCs differentiated into TM-like cells (ADSC-TM).

(A) Relative transcript levels of stem cell marker (OCT4) and TM cell markers (CHI3L1, AQP1) in ADSCs after 10 days of induction. The asterisks (*) denote values different from primary ADSCs. ***p < 0.001 by two-way ANOVA with Dunnett’s multiple comparisons test, n = 6 (2 cell strains, 3 replicates each). Pooled data are represented as mean ± SD. (B) Protein levels of TM markers (CHI3L1, AQP1) after 10-day induction assayed by Western blotting. β-actin served as internal control. ADSCs did not express AQP1 or CHI3L1. (C) Marker expression after 10 days of induction, assayed by immunofluorescent staining. TM markers (top): CHI3L1(green)/ AQP1(red); Stem cell markers (bottom): Nestin (green)/OCT4 (red); Nuclei stained blue with DAPI. Scale bars, 50 μm. (D) Percentage of cells labeling positive by individual antibodies (n=7-16 fields of view, 2-3 cell strains/antibody). *p<0.05, **p<0.005, ***p<0.001, ****p<0.0001, one-way ANOVA followed by Tukey’s multiple comparisons test.

TM cells are phagocytic, removing debris from the aqueous humor under normal circumstances (8, 31). We thus examined whether induced ADSC-TM cells were also phagocytic. After incubating with fluorescent S. aureus bioparticles, most ADSC-TM cells and primary TM cells ingested the bioparticles with a circular peri-nuclear distribution indicating a cytoplasmic location (Fig. 2A). In contrast, most ADSCs did not ingest the bioparticles. Quantitative assessment (Fig. 2B) demonstrated phagocytosis in 86 ± 12 % of ADSC-TM cells in the Co-culture group, comparable to that of human TM cells (93 ± 3 %, p > 0.05). 65 ± 13 % of the cells in the ECM+CM group and 67 ± 5 % of cells in the ECM+AdvM group cells contained phagocytosed bioparticles, yet both groups had significant higher phagocytic rates than primary ADSC cells (11 ± 3 %, ***p < 0.001).

Figure 2. Differentiated ADSC-TM cells gained phagocytic function.

(A) Phagocytosis after 10-day induction. Ingested S. aureus bioparticles (red) were present in induced ADSC-TM cells and TM cells but not in ADSCs. Phalloidin and DAPI were used to stain cellular F-actin (green) and nuclei (blue). Bottom: Higher-magnification images. Scale bars, top: 100 μm; bottom: 20 μm.

(B) Quantitation of phagocytosis. Numbers of cells ingesting S. aureus bioparticles (red) divided by numbers of total cells (blue) were calculated. ***p < 0.001 by two-way ANOVA with Tukey’s multiple comparisons test; n = 10. Pooled data are represented as mean ± SD.

Response of ADSC-TM Cells to DEX Stimulation

One important characteristic of TM cells is their responsiveness to DEX exposure (64), including the formation of cross-linked actin networks (CLANs) (51). We thus exposed cells to 200 nM DEX for 11 days and visualized the actin component of the cytoskeleton. A profound morphological change in the organization of actin microfilaments due to DEX exposure was detected in TM and ADSC-TM cells, but not in ADSCs (Fig. 3A). The reorganized actin fibers resembled geodesic-dome-like polygonal lattices. The percentage of CLAN-forming cells was comparable between ADSC-TM cells in ECM+CM and primary TM cells (20.1 ± 6.0 % vs 21.4 ± 7.1 %, p > 0.05) (Fig. 3B). The percentage of CLAN-forming cells in the Co-culture group (12.4 ± 8.2 %) was lower, but still significantly higher than that in ADSCs, where no CLANs were observed. ECM+AdvM induction did not show a significant difference in CLAN-forming cells (5 ± 4 %) compared to ADSCs (0%).

Figure 3. ADSC-TM Cells Responded to DEX treatment.

(A) CLAN formation and MYOC expression visualized by staining. Cells were treated with 200nM DEX for 11 days. Bottom figures show magnified views of the boxed regions in the top row, showing CLANs (dotted area), polygonal structures formed by actin crosslinking. MYOC (red) expression was detected in ADSC-TM and TM cells. Scale bars, top: 50 μm; bottom: 20 μm.

(B) Quantitation of CLANs. Numbers of CLAN-forming cells and total cells per field were counted and the ratio was calculated. (**p < 0.01, ***p < 0.001, NS = no statistical significance by one-way ANOVA followed by Tukey’s multiple comparisons test; n=10)

(C) Quantified MYOC gene expression by qPCR comparing same cell type before and after DEX treatment. The results are average of two independent experiments with ADSCs and TM cells from two different donors. *p < 0.05, ***p < 0.0001 by one-way ANOVA with Tukey’s multiple comparisons test; n=6. All pooled data are represented as mean ± SD.

Another characteristic response of TM cells to DEX is upregulation of myocilin (MYOC) (64, 65). MYOC mRNA levels were significantly elevated in primary TM cells and in ADSC-TM cells induced by Co-Culture and ECM+CM after DEX treatment compared to untreated cells (Fig. 3C). DEX-treated ADSCs and ADSC-TM cells with ECM+AdvM induction did not show increased MYOC expression compared to untreated cells (Fig. 3C).

Integration of Transplanted Human ADSC-TM Cells into Mouse TM Tissue

We next investigated the fate of ADSC-TM cells in vivo, selecting the ECM+CM approach for cell induction, since it efficiently induced a TM-like phenotype without potential contamination from primary TM cells as in co-culture. Human ADSCs and induced ADSC-TM cells at passage 4 were labeled with DiO dye and injected into the anterior chamber of adult C57BL/6 wildtype mice. DiO-labeled human fibroblasts (Fibro) and DMEM/F12 only (sham) injections and age-matched naïve mice were used as controls. 10,000 cells were transplanted in a 2-μl volume of DMEM/F12.

The anterior chamber and TM structure were examined in hematoxylin and eosin stained sections (Fig. S1A-E). At 30 days post-injection, there was no obvious morphological difference among the groups. Immunofluorescent staining demonstrated that both injected ADSC-TM cells and ADSCs were present in the TM region at 30 days post injection (Fig. 4A), distributed throughout all cell layers of the TM, suggesting that these cells integrated into the TM. Few ADSCs were attached to the iris (Fig. 4A2, 4A4, arrowheads). Although fibroblasts were also seen in the TM (green label), these cells displayed significant off-target attachment into other tissues in the anterior chamber, such as corneal endothelium and iris (Fig. 4A2, arrowheads). Tissue wholemounts at 1-week (Fig. S2A-E) and at 2-weeks (Fig. S2F-J) post-injection showed that the majority of injected ADSC-TM cells homed to the TM region, with few cells attached to the corneal endothelium, similar to injected TMSCs. Injected ADSCs and fibroblasts were present in the TM region as well as on the corneal endothelium.

Figure 4. Human ADSCs and ADSC-TM cells integrated into mouse TM tissue 30 days after intracameral injection.

(A) Detection of injected cells in mouse TM tissue. Phase images (A1) of anterior segment cryosections show structural landmarks. (A2) Immunofluorescent images at the same magnification as phase images show expression of AQP1 (red) and injected DiO+ cells (green). Arrowheads point to DiO+ fibroblasts on the corneal endothelium and iris. Higher magnification images (A3, from boxes in A2) show AQP1+ TM cells and merged images (A4, from boxes in A2) show AQP1 expression in DiO+ injected cells and endogenous TM cells. Scale bars, A1-A2, 50μm; A3-A4, 10μm.

(B) Human CHI3L1 (n=9) and AQP1 (n=6) mRNA expression in mouse TM tissue by qPCR using human-specific primers. Pooled data represent mean ± SD. *p < 0.05, ***p < 0.001 by one-way ANOVA with Dunnett’s multiple comparisons test.

(C) Ultrastructure of mouse TM tissue as visualized by TEM. The extracellular matrix (ECM) material appears generally normal, with no evidence of significant ECM accumulation or loss. Scale bars, 2μm.

TM = Trabecular meshwork, I = Iris, CS = Cornea/Sclera region, SC = Schlemm’s canal, JCT = Juxtacanalicular connective tissue.

Importantly, integrated ADSCs and ADSC-TM cells expressed the water channel aquaporin 1 (AQP1), which plays a role in modulation of aqueous outflow and can serve as a functional marker for TM cells (31, 66). AQP1 was detected in both injected DiO+ cells and endogenous mouse TM cells (Fig. 4A2-A4). AQP1 was also expressed in corneal stromal and endothelial cells. In contrast, the majority of injected fibroblasts were AQP1-negative, although a few fibroblasts had weak AQP1 labeling (Fig. 4A2-A4). To further confirm that DiO+ cells expressed AQP1, we acquired orthogonal confocal images showing that the green DiO+ ADSCs and ADSC-TM cells colocalized with AQP1 (red), but fibroblasts did not (Fig. S3). qPCR using human-specific primers (Fig. 4B, Table S1) indicated dramatically increased human CHI3L1 and AQP1 mRNA levels in the ADSC-TM- and ADSC-injected mouse TM. The elevated expression of CHI3L1 and AQP1 in both ADSC-TM and ADSC groups was statistically significant compared to the naive control.

TEM (Fig. 4C) shows intact TM microstructure, with thin and well-defined extracellular beams covered by cells in ADSC-TM and ADSC groups, similar to normal and sham controls. In the fibroblast group, no obvious difference was observed.

Cell Apoptosis after Transplantation

Although DiO+ cells remained visible in the tissue one month after transplantation, it was important to determine whether the injected human cells survived in the mouse TM after xenotransplantation. The TUNEL assay was used to identify any apoptotic cells after transplantation. We observed that a few of the DiO+ ADSCs and fibroblasts were TUNEL-positive, and that many of the DiO- mouse TM cells in the fibroblast-injected TM tissue were positive (Fig. 5A). TUNEL-stained cells and DiO+ cells in the TM region from 7-10 cryosections in each experimental group were counted. Statistical analysis indicated that the number of apoptotic cells in the fibroblast-injected group was significantly greater than that in other groups (Fig. 5B). Further, the number of apoptotic cells in the ADSC group was greater than those in the naïve control and ADSC-TM groups. There was no statistically significant difference in apoptotic cell numbers between naive control and ADSC-TM groups. The average DiO+ cell numbers per TM region are shown in Fig. 5C. We did not observe a statistically significant difference between the numbers of attached ADSCs, ADSC-TM or fibroblasts prelabeled with DiO in these sections.

Figure 5. Apoptotic cells detected in the TM tissue 30 days after fibroblast injection.

(A) Apoptotic cells in the mouse TM tissue visualized by TUNEL staining. Apoptotic cells stain red by TUNEL; DiO+ injected cells are green. DAPI stains nuclei blue. Arrows indicate apoptotic cells. Scale bars, 10 μm. (B) Quantitation of apoptotic cells in the TM. TUNEL (+) cells within the TM region were counted and averaged (n=7 sections per experimental group). (C) Quantitation of attached cells in the TM. DiO+ cells within the TM region were counted and averaged (n=10 sections per experimental group). Attached cell numbers were not statistically different between the 3 groups. ns = not significant; *p < 0.05; ***p < 0.001 by one-way ANOVA with Tukey’s multiple comparisons test; n = 7. Pooled data represent mean ± SD.

Inflammatory and Immune Response after Cell Transplantation

Human ADSCs and ADSC-TM cells were transplanted into normal mouse anterior chambers. To determine whether xenotransplantation would induce inflammatory response or immunorejection, cryosections were cut and stained for the inflammatory markers CD45 and GR1, and for the T-cell markers CD4 and CD3. As shown in Fig. 6A-6D, at day 3 after cell transplantation, few cells in the TM tissue receiving ADSC- and ADSC-TM-cell injections were positive for the markers, while more positive cells could be seen on the iris and the cornea. A few ADSCs were located off-target on the iris as observed in Figure 4A2 and 4A4. Fig. 6E shows mRNA expression levels of the inflammatory markers CD45, CD11b, CD11c, F4/80 and the T-cell marker CD3 at day 3 after cell transplantation. ADSC-TM cell-injected tissues including the TM and surrounding corneal tissue had increased expression of all of those detected markers except CD11b. ADSC injected tissues had mildly increased expression of CD11b and CD3.

Figure 6. Inflammatory and immune response 3 days after xenotransplantation of cells.

(A-D) Immunofluorescent staining of mouse eye cryosections from normal control, human ADSC-injected and human ADSC-TM cell-injected eyes at day 3. Injected cells were DiO stained (green). CD45 (A), GR1(B), CD4 (C), CD3 (D) stained as magenta, while nuclei stained with DAPI (blue). Scale bars, 50μm. (E) mRNA levels in mouse TM and adjacent angle tissues determined by qPCR at day 3 after cell injection. *p < 0.05, **p <0.01, ***p < 0.001 by two-way ANOVA with Tukey’s multiple comparisons test; n = 3. Pooled data represent mean ± SD.

At day 30 after transplantation, CD45+ inflammatory cells were detected in the TM tissue and corneal stroma in the fibroblast-injected group (Fig. S4D) while no CD45 staining was detected in other groups (Fig. S4A-C).

Homeostasis of Aqueous Humor Dynamics After Cell Transplantation

To assess whether transplanted cells could maintain IOP within the normal range, we measured mouse IOP before and at 3, 5, 10, 14, 21 and 30 days after cell transplantation (Fig. 7A). IOPs of eyes with ADSC-, ADSC-TM cell- or sham-injections were similar to IOPs in naive controls, with no statistically significant difference at each time point (p > 0.05). The IOP of eyes with fibroblast injection was elevated at each time point measured and the elevation at days 3, 10, 14, 21 and 30 was statistically different (*p < 0.05, **p < 0.005) compared with the naive control group.

Figure 7. IOP and aqueous outflow facility remained normal in ADSC and ADSC-TM injected eyes.

(A) IOP before cell injection (Day 0) and at 3, 5, 7, 10, 14, 21 and 30 days post-injection. IOPs of eyes with cell injection or sham injection were compared to normal control animals. *p < 0.05, **p < 0.01, by two-way ANOVA with Sidak’s multiple comparisons test.

(B) Outflow facility at day 30 post injection. The slope (Y) of the linear regression line to the flow rate-IOP data represents outflow facility. *p = 0.01 by SAS software using generalized linear mixed model for multiple comparisons after Bonferroni correction. (n = 7, ADSC-TM; n = 8, ADSC; n = 6, Fibro; n=5, Sham Ctrl; n = 6, Naive Ctrl).

Aqueous humor outflow facility is inversely proportional to the fluid flow resistance of the conventional outflow pathway, primarily the TM. Ex vivo mouse eye perfusions were conducted to measure facility, following existing procedures (57) (35) with typical data traces shown in Fig. S5. Outflow facility (Fig. 7B1-B5) in ADSC-injected eyes (0.021 ± 0.025 μl·min⁻¹·mmHg⁻¹) was not different from that of the naïve control group (0.020 ± 0.006, p = 0.763). Similarly, there was no significant difference in outflow facility between ADSC-TM cell-injection (0.020 ± 0.025) and naïve control (p = 0.963). In contrast, the outflow facility of the fibroblast-injected group (0.012 ± 0.040) was significantly lower than that of controls (p = 0.01). There was no significant difference in outflow facility between Sham control (0.018 ± 0.030) and naïve control (p = 0.790). The outflow facilities of all conditions are shown together in Fig. 7B.

Chemotaxis and Affinity Between ADSC-TM Cells and TM Cells

The process whereby injected cells attach to the TM can be assumed to occur in two ways: either injected cells are carried by flowing aqueous humor to reach the TM and reside there in a passive way, or there is specific recognition between injected cells and local TM cells so that the integration is an active paring of two types of cells. The first option is possible for every kind of cell that is injected into the anterior chamber. Fibroblasts attached to not only the TM, but also other reachable tissues in the anterior chamber (Fig. 4A) and the corneal endothelium (Fig. S2). The injected ADSCs and ADSC-TM cells were seen more specifically homing in the TM tissue although some ADSCs did attach to the iris and corneal endothelium.

In order to investigate whether CXCR4/SDF1 plays a role in ADSC homing, we first measured cell migration rates to determine if CXCR4/SDF1 has a chemotactic effect between ADSCs or ADSC-TM cells and TM cells. CXCR4 expression in ADSC-TM cells was more than 100-fold higher than that in ADSC cells (Fig. 8A). After treatment with the CXCR4 inhibitor IT1t for 72 hrs, the CXCR4 expression in ADSC-TM cells dramatically reduced to the same level as ADSCs, while the CXCR4 expression level in ADSCs remained unchanged after IT1t treatment (Fig. 8A). SDF1 expression in TM cells was increased after recombinant SDF1α+β treatment (SDF1αβ) and reduced after SDF1 antibody treatment to block the cell surface SDF1 (SDF1Ab) (Fig. 8B).

Figure 8. Chemokine axis CXCR4/SDF1 affects chemotaxis and affinity between ADSC-TM and TM cells in vitro.

(A) CXCR4 gene expression in ADSCs and ADSC-TM cells by qPCR before and after treatment with the CXCR4 inhibitor IT1t for 72 hrs. ***p < 0.001 by two-way ANOVA with Tukey’s multiple comparisons test; NS = no statistical significance, n = 6.

(B) SDF1 gene expression changes of TM cells by qPCR after recombinant SDF1α and SDF1β treatment or SDF1 antibody treatment. Cells were treated for 72 hrs. *p < 0.05 by two-way ANOVA with Tukey’s multiple comparisons test; n = 9.

(C-D) Chemotaxis of TM cells to ADSC-TM cells. X-axis: chemoattractant: medium only; TM cells; TM cells treated with SDF1α and β factors (TM-SDF1αβ); TM cells treated with SDF1 antibody (TM-SDF1Ab). C: ADSC-TM cells seeded on a Transwell membrane insert for migration. D: ADSC-TM cells treated with the CXCR4 inhibitor IT1t (ADSC-TM-IT1t) seeded on the insert for migration. Y-axis: percentage of migrated cells (migrated cell number/total cell number x 100%).

(E-F) Chemotaxis of TM cells to ADSCs. Same conditions as ADSC-TM cells (C-D). E: ADSCs; F: ADSCs treated with It1t (ADSCs-IT1t).

(G-H) Affinity of TM cells to ADSC-TM cells. Same conditions as Chemotaxis (C-D). Y-axis: attached cell numbers/field. G: ADSC-TM cells; H: ADSC-TM-IT1t.

(I-J) Affinity of TM cells to ADSC cells. Same conditions as Chemotaxis of ADSCs (E-F). I: ADSCs; J: ADSCs-IT1t.

C-J: *p<0.05, ** p < 0.01, ***p<0.001, ns: no statistical significance. Two-way ANOVA with Tukey’s multiple comparisons test; n = 6-8)

To assay chemotaxis, untreated, SDF1αβ or SDF1Ab TM cells were seeded on the bottom of cell culture wells as chemoattractant. Treated or non-treated ADSCs or ADSC-TM cells were seeded on Transwell membranes for 24 hrs. Migrating cells (those moving to the other side of the Transwell membrane through the pores) and non-migrating cells (those remaining on the seeded side; Fig. S6A-B) were counted and averaged (Fig. 8C-F). The migration rate of ADSC-TM cells with SDF1αβ TM cells on bottom as chemoattractant was significantly greater than the case of no chemoattractant (medium) (71.2 ± 7.0% vs. 34.9 ± 7.3%, **p < 0.01). With SDF1Ab TM cells as chemoattractant, the migration rate was reduced to 20.0 ± 16.5 % (*p < 0.05, Fig. 8C). As shown in Fig. 8D, the cell migration rates of ADSC-TM cells treated with the CXCR4 inhibitor IT1t did not differ across all conditions, demonstrating that inhibition of CXCR4 abolishes its ability to modulate SDF1 mediated chemotaxis. In ADSCs (Fig. 8E) and ADSCs treated with IT1t (Fig. 8F), however, SDF1 manipulation did not influence the chemotaxis between ADSCs and TM cells, similar to that of ADSC-TM cells treated with IT1t (Fig. 8D). These three conditions had similar low CXCR4 expression levels (Fig. 8A).

Consistent with the chemotaxis data, SDF1 also mediated cell affinity between ADSC-TM cells and TM cells. Increased expression of SDF1 in TM cells (TM-SDF1αβ), as a feeder layer, significantly enhanced the number of attached ADSC-TM cells compared to no feeder layer control (21.2 ± 4.1/field vs 12.0 ± 0.8/field, ***p < 0.001, Fig. 8G). When SDF1 antibody blocked TM cell surface SDF1 (TM SDF1Ab), the attached cell number decreased to 6.2 ± 3.8/field (*p < 0.05, Fig. 8G). However, after ADSC-TM cells were treated with the CXCR4 inhibitor IT1t, the observed enhancement by SDF1 disappeared (Fig. 8H), consistent with the chemotaxis results. In ADSCs (Fig. 8I) and ADSCs treated with SDF1 antibody (Fig. 8J), SDF1-mediated enhancement of cell affinity was not observed, similar to the chemotaxis results. Fig. S6C-D show that DiO labeled green ADSCs or ADSC-TM cells attached to the cell culture chamber slides with or without pre-seeded TM cells as feeder layer. The feeder cells were not labeled with DiO but can be seen with DAPI staining.

When ADSC-TM cells were blocked with the CXCR4 inhibitor IT1t, the cells mainly attached to the corneal endothelium with very few attaching to the TM region (Fig. S7B), similar to ADSCs treated with IT1t (Fig. S7A) but different from ADSC-TM cells (Fig. S2C). There was no obvious difference between ADSCs (Fig. S2B) and ADSC treated with IT1t (Fig. S7A).

DISCUSSION

In this study, we report that human ADSCs can be induced in vitro to differentiate into TM-like cells expressing the TM cell markers CHI3L1 and AQP1. These ADSC-TM cells showed phenotypic similarities to native TM cells, including phagocytic capacity and responsiveness to dexamethasone stimulation, with dramatically increased expression of MYOC and formation of CLANs. After intracameral injection into wildtype mice, ADSC-TM cells specifically homed and integrated into the TM tissue and expressed CHI3L1 and AQP1. Integrated ADSCs also expressed differentiated TM cell markers CHI3L1 and AQP1. The xenotransplantation of ADSCs and ADSC-TM cells did not adversely affect IOP or outflow facility in naïve mouse eyes for up to 30 days after transplantation. The chemokine CXCR4 and its ligand SDF1 may play an important role in ADSC-TM cell homing to the TM tissue. In contrast, injected fibroblasts attached to the TM tissue, corneal endothelium and iris, suggesting that ADSC-TM cells actively home to the TM tissue while fibroblasts passively attach to tissues.

This study is novel and shows that it is feasible to consider ADSCs as an autologous or allogeneic cell candidate for TM regeneration. Our previous work showed that human TM stem cells can home and integrate into normal mouse TM tissue (34) and laser-damaged TM region (35) and maintain the mouse IOP within the normal range, motivating further exploration of stem cell-based therapy for glaucoma. On the other hand, it is not easy to use autologous TM stem cells for glaucoma treatment, since cell harvesting would require ocular microsurgery and TM stem cells may be depleted or dysfunctional in POAG patients. ADSCs have advantages over other types of stem cell candidates, in that they can be harvested in large amounts with minimally-invasive approaches. In this study, we show that ADSCs can be induced to differentiate into TM-like cells by co-culturing with TM cells or by culturing on TM-secreted ECM with TM-conditioned medium for only 10 days. Both approaches yielded differentiated TM-like cells as evaluated by phagocytosis and response to DEX stimulation. Specifically, previous studies reported that about 7% (55) to 34% (53) of human TM cells formed CLANs after DEX exposure. Our results are generally consistent with these rates, showing 12 to 21% of cells forming CLANs, including primary TM cells and ADSC-TM cells induced by ECM+CM or by co-culture. On the other hand, ADSC-TM cells induced by ECM+AdvM had fewer (5%) CLAN-forming cells, and we therefore eliminated this approach as an induction protocol. Although the induced ADSC-TM cells had reduced expression of OCT4 and increased expression of CHI3L1 and AQP1, the expression levels of CHI3L1 and AQP1 were still less than those of primary TM cells, especially AQP1 expression. AQP1 is expressed in vivo (56, 67, 68), suggesting that AQP1 has an important role in the TM. AQP1 is also detected in cultured TM cells (31, 66) and its expression is dramatically increased after static mechanical stretch indicating its protective role in response to mechanical stimulation (69). We hypothesize that induced ADSCs had less AQP1 expression than primary TM cells because induced ADSCs were not exposed to physiologic stressors such as IOP and ciliary muscle contraction.

Human ADSC-TM cells transplanted into mouse anterior chambers survived for up to 1 month and maintained IOP and outflow facility within normal ranges. The integrated ADSC-TM cells expressed CHI3L1 and AQP1. Although some ADSCs were seen off-target (e.g. on the iris and corneal endothelium), some ADSCs integrated into the TM tissue and expressed TM cell markers CHI3L1 and AQP1. These results suggest that the local environment of mouse TM can induce human ADSCs to differentiate into TM-like cells in vivo. Even though fibroblasts could also be detected in the TM region, only a few fibroblasts had weak expression of AQP1 which further indicates the effect of local environment. The injected fibroblasts led to IOP elevation which might be related to an inflammatory response as shown previously (56) and in Fig. S4, and might also be related to fibroblasts blocking (occluding) the TM. Therefore, we suggest that the fibroblasts were not truly homed to the TM but passively followed the aqueous outflow to reside in the TM.

On day 3 after transplantation, ADSC-TM cells induced a transient inflammatory response in the tissues surrounding the anterior chamber angle, while ADSCs induced a less inflammatory response with slightly increased expression of CD11b and CD3. Since the response was not severe, the IOP elevation was not significant in both ADSC and ADSC-TM cell injected eyes. In contrast, fibroblast-injected eyes showed elevated IOP, persisting for up to 1 month. We previously reported (70) that after transplantation of human cells into mouse corneal stromal, transient inflammatory cells migrated from mouse bone marrow to the corneal stromal at 24 hours and then reduced dramatically at 72 hours. There were no inflammatory cells at 2 weeks after stem cell transplantation, but inflammatory cells persisted in the corneas with fibroblast injection. ADSCs have effective immunomodulation characteristics (71), which may have been partially lost in differentiated ADSC-TM cells and which might explain why ADSC-TM cells had higher inflammatory and immune responses than ADSCs. Further studies are needed to explore this phenomenon. At 30 days after transplantation, a few CD45+ cells were only found in fibroblast-injected TM tissue. This observation may be related to the persistent IOP elevation in eyes receiving fibroblast injection, even though we did not observe microstructural changes by TEM.

In contrast, ADSC-TM and ADSC cell transplantation maintained aqueous humor homeostasis, including normal IOP and outflow facility. Although a few ADSCs were attached to tissues other than the TM, they did not induce obvious side effects compared to fibroblasts. Moreover, TEM indicated that both ADSC-TM cells and ADSCs transplantation did not affect the microarchitecture of the TM, with generally normal cell and ECM morphology. These findings suggest that both ADSCs and ADSC-TM cells may be safe cell therapy resources for POAG. Although some ADSCs were attached to the iris and corneal endothelium, they didn’t induce obvious ocular abnormalities. Using magnetic nanoparticles to improve stem cell homing (46) might be a good option to further increase the efficiency of ADSC homing and integration.

The natural flow patterns of aqueous humor facilitate stem cell delivery from the anterior chamber into the TM. However, it is interesting that fibroblast transplantation did not exclusively follow this route; instead, fibroblasts also attached into anterior segment structures such as iris and corneal endothelium. Such non-specific delivery has also been observed previously. Although the integrated ADSCs could differentiate into TM cells, some ADSCs also attached to the corneal endothelium. This observation led us to further investigate the mechanism of integration of ADSC-TM and ADSCs into TM tissue. In vitro experiments showed that the CXCR4/SDF1 axis – an essential pathway for controlling the navigation of progenitors between the bone marrow and blood (72) or ischemic myocardium (73)– plays a role in chemotaxis and affinity between ADSC-TM and TM cells in vitro, similar to TMSC and TM cells (35). After expression of CXCR4 in ADSC-TM cells was blocked with IT1t, the cells lost the ability to home to the TM region. No such phenomenon was observed between ADSCs and TM cells, consistent with a previous report (74) that CXCR4 and its ligand SDF1 are likely not major homing factors for ADSCs. Admittedly, more pathways ought to be examined to thoroughly understand the mechanism(s) of specific homing of ADSC-TM cells, including in vivo experiments. For example, a conditional SDF1 knockdown mouse model could be generated to confirm the effect of the CXCR4/SDF1 axis on ADSC-TM cell homing in TM regeneration. Further, CXCR7 or other receptors may be involved in ADSC homing to the TM tissue and other factors may be involved in those non-homed ADSCs, and this area requires further exploration. Understanding such homing mechanism(s) is a promising tool to manipulate these pathways for enhanced homing and integration of transplanted stem cells and consequently better restoration of TM function.

In conclusion, our results provide novel evidence that ADSCs can be induced to form functional TM-like cells and may be an attractive autologous stem cell resource for TM restoration as a stem cell-based therapy for glaucoma. Further studies to discover the effectiveness of stem cell transplantation in an animal model of ocular hypertension are needed.

Abbreviations

ADSCs

adipose-derived stem cells

AQP1

aquaporin 1

CM

conditioned medium

CLANs

cross-linked actin networks

CXCR4

C-X-C chemokine receptor type 4

DEX

dexamethasone

ECM

extracellular matrix

Fibro

Fibroblasts

iPSCs

induced pluripotent stem cells

IOP

intraocular pressure

JCT

juxtacanalicular connective tissue

MSCs

mesenchymal stem cells

MYOC

myocilin

POAG

primary open angle glaucoma

qPCR

Quantitative Reverse Transcription–Polymerase Chain Reaction

SDF1

stromal cell–derived factor 1

TM

trabecular meshwork

TMSCs

trabecular meshwork stem cells