Mesenchymal stem cell-derived neural progenitors attenuate proinflammatory microglial activation via paracrine mechanisms

Background

Multiple sclerosis (MS) is an autoimmune-mediated focal demyelinating disease of the CNS. In many patients, the initial relapsing–remitting phase of the disease (relapsing–remitting MS) transitions into a progressive form (secondary progressive MS) with a steady worsening of neurological function independent of relapses. A subgroup of patients with primary progressive MS are diagnosed with progressive disease from the onset of symptoms. The underlying mechanisms that drive MS disease progression include chronic demyelination, compartmentalized neuroinflammation, microglial activation, reactive astrogliosis and axonal loss, all of which contribute to neurodegeneration and accumulating neurological disability, with limited therapeutic options [1]. Therapies that target these underlying mechanisms to promote regeneration and repair of the brain and spinal cord are necessary to address this unmet medical need.

A novel cellular therapy utilizing autologous bone marrow mesenchymal stem cell (MSC)-derived neural progenitors (MSC-NPs) is being investigated as a potential treatment for progressive forms of MS [2]. MSC-NPs are a neural derivative of MSCs which exhibit neurosphere morphology, increased expression of many of the immunomodulatory and trophic factors associated with MSC-mediated tissue repair, and reduced capacity to differentiate into mesodermal cell types [3]. In preclinical experiments in mice with chronic experimental autoimmune encephalomyelitis (EAE), the administration of multiple intrathecal injections of MSC-NPs was associated with reduced EAE scores and improved pathology including increased spinal cord myelination, decreased immune infiltration in the CNS and increased recruitment of endogenous neural progenitor cells [4]. A phase I clinical trial in 20 patients with progressive MS demonstrated safety and improved neurological function in some patients, and a larger placebo-controlled phase II clinical trial is near completion [2].

A better understanding of the mechanisms underlying the therapeutic efficacy of MSC-NPs is required in order to optimize this strategy as a treatment for MS. Similar to MSCs, the therapeutic action of MSC-NPs is associated with secreted immunomodulatory and trophic factors that act in a paracrine manner to influence tissue repair [3,4]. Although the expression of many candidate paracrine factors is upregulated in MSC-NPs compared with MSCs, the CNS target(s) of MSC-NP-mediated paracrine actions has not been identified [3]. Interestingly, cerebrospinal fluid (CSF) biomarker analysis demonstrated that intrathecal MSC-NP treatment in patients with MS correlated with reduced levels of the chemokine CCL2 [5]. CCL2 (MCP-1) is a proinflammatory chemoattractant that regulates neuroinflammation and has been implicated in neuroinflammatory diseases including MS [6]. In MS brain autopsy tissue, CCL2 expression colocalizes with microglia/macrophages and with hypertrophic astrocytes in demyelinated lesions, identifying two possible CNS sources of CCL2 that may be responding to MSC-NP injections [7,8].

Microglia are the resident innate immune cells of the CNS that play a key role in MS pathogenesis [9]. Microglia, along with infiltrating macrophages, are highly abundant in MS lesions, where they have diverse functions including myelin phagocytosis, antigen presentation, immunomodulation, repair and remyelination [10,11]. In response to the microenvironment, microglia display various activation states ranging from a proinflammatory phenotype (referred to as M1) to a regenerative phenotype (referred to as M2), along with intermediate phenotypes that translate into multiple distinct clusters of microglial populations in vivo [9,10,12]. Chronically activated microglia are associated with neurodegeneration and damage of oligodendrocytes through the release of proinflammatory cytokines and glutamate, as well as the production of reactive oxygen species and nitric oxide radicals, all mechanisms that contribute to MS progression [10,11,13]. Current therapeutic strategies are aimed at inhibiting proinflammatory microglia and/or promoting proregenerative microglia to slow disease progression and reduce the accumulation of disability in MS [13].

The aim of the current study was to investigate whether MSC-NPs modulate microglial polarization. MSCs, the parent population of MSC-NPs, have an immunomodulatory effect on microglia both in vitro and in animal models [14–21]. Utilizing multiple in vitro models of microglia, including human induced pluripotent stem cell (iPSC)-derived microglia, we show that MSC-NPs inhibit the proinflammatory phenotype of microglia, including the reduction of microglia-derived CCL2. MSC-NPs derived from multiple donors act via paracrine mechanisms, potentially including the expression of TGF-β, to inhibit M1 microglia markers as well as to enhance proregenerative M2 markers.

Materials & methods

MSC & MSC-NP cell culture

Human MSCs were isolated from bone marrow as described previously with institutional review board (Western IRB) approval and informed consent [2]. MSCs were cultured in growth media (Lonza, MD, USA) supplemented with 5% Plastate™ human platelet lysate (New York Blood Center), 2 U/ml heparin and 2 mM GlutaMAX™I CTS™ (Life Technologies, NY, USA). MSC-NPs were generated by culturing MSCs in neural progenitor maintenance medium (NPMM) (Lonza) supplemented with 20 ng/ml each of epidermal growth factor and basic fibroblast growth factor for 2 weeks. Cells were incubated in a humidified 37°C incubator at 5% CO₂ and 5% O₂. All MSC-NPs were derived from donors with MS participating in clinical trials (cliniclatrials.gov ID NCT03355365 and NCT03822858), and were quality-tested in compliance with cGMP manufacturing [2]. Conditioned medium (NPCM) was collected from MSC-NPs after 2–3 days in culture, centrifuged for 10 min at 800 × g and 4°C to remove cell debris, and stored at -20°C until use. Twenty-five separate NPCM samples were collected (NPCM1 through NPCM25), and each milliliter represented a cell density range from 30,000 to 170,000 cells/ml.

BV-2 mouse microglia cell culture

The immortalized BV-2 murine microglial cell line was purchased from Banca Biologica e Cell Factory (Genova, Italy). Cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum and 2 mM GlutaMAX. BV-2 cells were plated at a density of 850 cells/cm² for all experiments. Cells were polarized and activated as previously described [22]. Briefly, a proinflammatory ‘M1’ phenotype was induced by culturing in growth medium supplemented with granulocyte–macrophage colony-stimulating factor (5 ng/ml) for 2 days, followed by a half media change with the same supplemented medium for an additional 3 days. Cells were then stimulated with IFN-γ (20 ng/ml) for 1 h, followed by stimulation with lipopolysaccharide (LPS; 100 ng/ml) for 24 h. A proregenerative ‘M2’ phenotype was induced by culturing cells in growth medium supplemented with macrophage colony-stimulating factor (M-CSF; 25 ng/ml) for 2 days, followed by a half media change with M-CSF medium for an additional 3 days. Cells were then stimulated with IL-4 (20 ng/ml) and IL-13 (20 ng/ml) for 24 h. Additional IL-4 and IL-13 was added to the medium without a medium change for another 24 h before harvesting cells. All cytokines were purchased from PeproTech (Cranbury, NJ, USA). For coculture experiments, MSCs or MSC-NPs were plated at increasing densities (850, 2550 and 8500 cells/cm²) 1 day before plating BV-2 cells to allow time to attach. For conditioned media experiments, BV-2 cells were cultured in 50% NPCM or 50% control media (NPMM) during the entire stimulation protocol.

Microglial differentiation of human iPSC lines

iPSC lines were generated from peripheral blood-derived CD34⁺ hematopoietic stem cells (EasySep RosetteSep kit, STEMCELL Technologies Inc., BC, Canada) with institutional review board approval and informed consent. Cells were reprogrammed using Epi5™ Episomal iPSC Reprogramming Kit (Thermo Fisher Scientific, CA, USA) following the manufacturer’s instructions. iPSCs were cultured and expanded onto Matrigel-coated dishes (100 μg/ml, Corning, MA, USA) in mTeSR1 medium (STEMCELL Technologies, Inc.). Cells were passaged every 3–4 days using a 1:1000 dilution of 0.5 M EDTA, pH 8.0 (Thermo Fisher) for 5 min and replated in mTeSR1 medium with 10 μM Thiazovivin (Biogems, CA, USA) for 24 h. Two iPSC lines were generated, one from one healthy donor and one from a donor with MS (Supplementary Figure 1A). Normal karyotype was verified (Cell Line Genetics, Inc., WI, USA) (Supplementary Figure 1B) and pluripotency was validated by immunohistochemical staining for pluripotency markers (Supplementary Figure 1C).

Floating microglia progenitor cells were induced from iPSCs as described previously [23]. After day 25, a sample of floating cells was collected for flow cytometry analysis to confirm that the percentage of CD45⁺/CD14⁺/CX3CR1⁺ myeloid progenitors was >60% (Supplementary Figure 2A). Microglia progenitors continued to be released into suspension up to day 100.

Microglial differentiation was performed by collecting and centrifuging floating myeloid progenitors at 350 × g for 5 min at 4°C and plating in NeuroBasal serum-free medium (Life Technologies) with 1% Gem21 Neuroplex, 0.5% N2 Neuroplex (both from Gemini Bio, CA, USA), 50 mM NaCl (Sigma-Aldrich, MO, USA), 0.2% Albumax II, 1 mM sodium pyruvate and 2 mM GlutaMAX (all from Life Technologies) containing TGF-β1 (25 ng/ml), IL-34 (100 ng/ml) and M-CSF (12.5 ng/ml) (PeproTech) as previously described [24]. Differentiated microglia were confirmed to be IBA1⁺ and TMEM119⁺, displayed ramified microglial morphology with extended processes and were functionally capable of phagocytosis (Supplementary Figure 2B–D). We found no differences in differentiation efficiency, marker expression, phagocytic capacity, morphology or microglial activation between the healthy control or MS-derived iPSC microglia lines (Supplementary Figure 3), which were used interchangeably throughout the study.

iPSC microglia activation

Resting microglia consisted of differentiated microglia cultured in neurobasal medium with all added supplements with differentiation factors (IL-34, TGF-β1 and M-CSF) omitted. Proinflammatory ‘M1’ microglia were stimulated with IL-1β (50 ng/ml), IFN-γ (20 ng/ml) and TNF-α (10 ng/ml) for 24 h. Proregenerative ‘M2’ microglia were stimulated with IL-4 (40 ng/ml) and IL-13 (40 ng/ml) for 24 h. All stimulatory factors were obtained from PeproTech. For NPCM treatment, microglia were treated with 50% neurobasal medium without differentiation factors and 50% NPCM or NPMM for 24 h unless otherwise indicated. iPSC microglia were treated with 5 μM of the TGF-β receptor type I inhibitor SB-505124 (MedChemExpress, NJ, USA) or 15 μg/ml rabbit anti-human CX3CR1 neutralizing antibody TP-502 (Torrey Pines Biolabs, NJ, USA) for 24 h to inhibit TGF-β and CX3CL1 signaling, respectively. For MSC-NP coculture experiments, MSC-NP cells were added to iPSC microglia in a 1:1 ratio.

Gene expression analysis

Cells were lysed in RLT buffer with 0.01% 2-mercaptoethanol (Qiagen, MD, USA). Total RNA was isolated using the RNeasy Mini Kit (Qiagen) and reverse-transcribed using Superscript VILO Master Mix (Life Technologies). qRT-PCR was performed using TaqMan^® Universal PCR Master Mix with prevalidated TaqMan gene expression assays (Thermo Fisher). TaqMan assays for mouse cells were Ccl2 (Mm00441242_m1), Ccl8 (Mm01297183_m1), Nos2 (Mm00440502_m1), Arg1 (Mm00475988_m1) and Hprt (endogenous control) (Mm03024075_m1). Human TaqMan assays were CCL2 (Hs00234140_m1), CCL5 (Hs00982282_m1), CCL8 (Hs04187715_m1), CD206 (Hs00267207_m1), TGFB1 (Hs00998133_m1), TGFB2 (Hs00234244_m1), TGFB3 (Hs01086000_m1), IL6 (Hs00174131_m1), CX3CL1 (Hs00171086_m1) and endogenous control IPO8 (Hs00183533_m1). PCR was performed using QuantStudio™ 7 Flex real time qPCR (Thermo Fisher). Relative changes in gene expression were determined by the ΔΔCt method with QuantStudio 7 Flex software (Thermo Fisher).

Secreted protein analysis

Supernatant was collected, centrifuged for 10 min at 800 × g and 4°C to remove cell debris, and stored at -80°C before use. The concentrations of mouse CCL2 and CCL8 proteins secreted by BV-2 cells were measured by ELISA (R&D Systems, MN, USA) after 1:1000 and 1:10 dilution, respectively. The concentrations of human CCL2, CCL5, CCL8 and IL-6 were determined by Luminex^® assay (R&D Systems) after 1:50 dilution.

Quantitation of cells with activated microglial morphology

Microglia were treated as indicated for 48 h and a representative photo was taken in the center of each well (at least two wells per condition). For each photo, a blinded observer manually counted the total number of cells, the number of cells with ameboid cell bodies representative of activated microglia and the number of cells with longer ramified processes characteristic of resting microglia. Cells with ambiguous morphology were excluded.

Phagocytosis assay

Microglia were pretreated with 50% NPMM or 50% NPCM for 24 h. Fluoresbrite^® carboxylate microspheres 1.00 μM (Polysciences, PA, USA) were added at a ratio of 100 beads/cell for 2 h at 37°C. Cells were washed with phosphate-buffered saline, detached using TrypLE™ (Thermo Fisher) and resuspended in 200 μl of cold phosphate-buffered saline with 1% bovine serum albumin. Fluorescein isothiocyanate fluorescence was measured by flow cytometry using BD FACSAria™ II (BD Biosciences, CA, USA).

Statistical analysis

Data with multiple groups were analyzed by one-way analysis of variance followed by Tukey’s multiple comparison post-test. Correlations were determined by simple linear regression analysis. Statistical significance was set to p < 0.05. GraphPad Prism 9 (GraphPad, CA, USA) was used to calculate significance.

Results

MSCs & MSC-NPs attenuate the proinflammatory phenotype of BV-2 microglia

To investigate the effect of MSC-NPs compared with MSCs on microglial activation, we utilized the mouse microglial cell line BV-2 as a model of microglial polarization. BV-2 cells upregulate proinflammatory cytokine expression including Ccl2 in response to M1 stimulation with IFN-γ and LPS (Figure 1A) and conversely upregulate proregenerative markers including Arg1 in response to M2 stimulation with IL-4 and IL-13 (Figure 1B). We found that as previously described with mouse MSC coculture, increasing numbers of human MSC or MSC-NP cells cocultured with IFN-γ/LPS-stimulated BV-2 cells resulted in reduced microglial gene expression of proinflammatory cytokine Ccl2 (Figure 1A) and inducible nitric oxide synthase (Nos2, data not shown) [18]. In addition, coculture with MSCs or MSC-NPs significantly upregulated Arg1 expression in IL-4/IL-3-stimulated BV-2 cells in a dose-dependent manner (Figure 1B). MSC-NPs inhibited Ccl2 expression and induced Arg1 expression in BV-2 cells at lower cell ratios compared with MSCs, suggesting that MSC-NPs may be more potent in a coculture setting.

To determine whether attenuation of BV-2 activation could be mediated by secreted factors from MSC-NPs, conditioned media were collected from multiple individual batches of human MSC-NP cells and added to BV-2 cells during IFN /LPS stimulation. Similar to coculture, cell-free conditioned medium from MSC-NPs (NPCM) significantly inhibited gene expression of both Ccl2 (Figure 1C) and Ccl8 (Supplementary Figure 4A). Furthermore, NPCM samples reduced the levels of mouse CCL2 and CCL8 proteins (Supplementary Figure 4B & C, respectively) secreted by BV-2 cells compared with the unconditioned control medium (NPMM). The inhibition of CCL2 and CCL8 by NPCM samples from different donor batches of MSC-NPs suggests that paracrine suppression of the M1 response is a consistent characteristic of MSC-NPs. NPCM samples from different donor batches of MSC-NPs demonstrated variability in the degree of paracrine suppression of the M1 response in BV-2 cells (e.g., sample NPCM6). The degree of M1 suppression did not appear to correlate with the cell density used to condition the media (data not shown).

MSC-NP paracrine factors suppress proinflammatory markers & increase proregenerative markers in human iPSC-derived microglia

In order to better model the effects of human MSC-NPs on human microglia, we utilized human iPSC-derived microglia cells [23,24]. We initially confirmed that the iPSC microglia model was capable of M1 and M2 polarization and found that M1 stimulation with proinflammatory cytokines IFN-γ, TNF-α and IL-1β significantly induced gene expression of M1 markers including CCL2, CCL5, CCL8 and IL6 and downregulated the M2 marker CD206 (Figure 2 & Supplementary Figure 3B). Furthermore, M2 stimulation with IL-4 and IL-13 resulted in upregulation of M2 markers including CD206 (data not shown), demonstrating the capacity of iPSC microglia to express proregenerative markers in response to specific stimuli.

Similar to mouse BV-2 cells, we found that M1-stimulated iPSC microglia in the presence of either cocultured MSC-NP cells or NPCM showed a significantly attenuated response in terms of CCL2 and CCL5 protein secretion (Supplementary Figure 5). To better define the paracrine effects of MSC-NPs on microglial gene expression, we performed all additional iPSC microglia activation experiments in the presence of cell-free NPCM to avoid human RNA contamination from MSC-NPs. Unstimulated resting microglia expressed low levels of proinflammatory markers that were unchanged in the presence of NPCM (data not shown). However, microglia stimulated under M1 conditions demonstrated a significant reduction in CCL2, CCL5, CCL8 and IL6 gene expression when cultured in the presence of NPCM from seven individual MSC-NP cell lines compared with stimulation in NPMM (Figure 2). CCL2 (Figure 2A), CCL5 (Figure 2B) and IL6 (Figure 2D) expression was more uniformly suppressed by NPCM than was CCL8 expression (Figure 2C), which showed variable suppression in the presence of NPCM. The changes in gene expression correlated strongly with reduced secretion of CCL2, CCL5, CCL8 and IL-6 proteins from NPCM-treated microglia (Figure 3). Although MSC-NPs also express CCL2 and IL-6, protein levels were very low in uncultured NPCM (NPCM only) compared with cultures with NPCM and microglia together (Figure 3).

To further explore the effects of MSC-NP on microglial polarization, we examined the effect of NPCM on the proregenerative, or M2, phenotype of microglia. In the presence of NPCM, resting unstimulated microglia as well as M1-stimulated microglia demonstrated significant upregulation of M2 markers CD206 (Figure 4A) and TGFB1 (Figure 4B), suggesting that MSC-NPs shift microglial polarization toward a proregenerative phenotype. Furthermore, M2 stimulation of microglia with IL-4 and IL-13 demonstrated upregulation of CD206 and TGFB1, which was even further upregulated in the presence of NPCM (Figure 4A & B). Overall, these results demonstrate that MSC-NPs from multiple donors suppress the proinflammatory phenotype and promote the proregenerative phenotype of human microglia via paracrine mechanisms.

Morphological & functional changes in activated microglia in the presence of NPCM

In addition to changes in gene expression, we examined morphological and functional changes in iPSC microglia when exposed to NPCM. In the presence of proinflammatory cytokines, microglia cultures exhibited more cells with ameboid cell bodies representative of activated microglia (Figure 5B & Supplementary Figure 6B) compared with unstimulated (Figure 5A & Supplementary Figure 6A) or M2-stimulated (Supplementary Figure 6C) microglia. In the presence of NPCM, the morphology of activated microglia exhibited longer ramified processes, appearing similar to resting microglia (Figure 5C). The effect of NPCM on microglial morphology was confirmed by quantitation (Figure 5D & E). In addition, NPCM significantly reduced the phagocytic function of microglia (Figure 5F & G). These data show that NPCM impacts microglial morphology and function in addition to suppression of proinflammatory markers in microglia.

Effects of MSC-NPs on microglia are mediated through TGF-β signaling

The TGF-β and CX3CL1 signaling pathways have previously been implicated in the polarization of microglia by MSCs [15,17]. To investigate the role that NPCM-derived TGF-β might play in inhibiting microglial activation, we tested whether inhibition of TGF-β signaling in microglia could block the effect of NPCM. We found that NPCM-mediated suppression of CCL2 (Figure 6A) and IL6 (Figure 6B) expression by microglia was significantly reversed in the presence of a TGF-β receptor inhibitor. In contrast, inhibition of CX3CR1 signaling had no impact on NPCM-mediated suppression of CCL2 (Figure 6A) and minimally reversed the NPCM-mediated suppression of IL6 (Figure 6B) expression by microglia. Overall, these data suggest that TGF-β but not CX3CL1/CX3CR1 signaling may play a role in mediating the effect of NPCM on microglial polarization.

We next tested whether the suppressive effects of NPCM on microglial activation correlated with TGF-β expression in MSC-NPs. We found that MSC-NPs express all three isoforms of TGF-β (TGFB1, TGFB2 and TGFB3) and that both TGFB1 and TGFB3 are significantly upregulated in MSC-NPs compared with MSCs, whereas TGFB2 was expressed at similar levels in the two cell types (Supplementary Figure 7A–C). CX3CL1 was also expressed at higher levels in MSC-NPs compared with MSCs (Supplementary Figure 7D). Interestingly, the relative level of TGFB3 expression in the MSC-NPs used to generate NPCM significantly correlated with the degree of both CCL2 (Figure 6C) and IL6 (Figure 6D) inhibition in microglia. In contrast, there was no correlation between TGFB1 (data not shown) or CX3CL1 (Figure 6E & G) expression in MSC-NPs with the degree of inhibition of microglia. Despite the correlations in gene expression of TGFB isoforms in MSCs and MSC-NPs, we were only able to detect TGF-β1 protein in NPCM, whereas secreted TGF-β2 and TGF-β3 were below the level of detection by ELISA (data not shown). Nevertheless, these data suggest that the gene expression of specific TGF-β isoforms in different batches of MSC-NPs may relate to their potency in terms of suppression of a proinflammatory phenotype in microglial cells in vitro.

Discussion

Based on early-phase clinical trials, the intrathecal injection of autologous MSC-NPs appears to be a promising new therapeutic strategy to delay or reverse disability worsening in progressive MS [2,5]. Thus a better understanding of the mechanisms underlying the therapeutic effects of MSC-NPs is important to optimize this approach. MSC-NPs are a subpopulation of ex vivo-expanded MSCs that were named based on their similarities to neural progenitor cells with respect to neural gene upregulation and neurosphere morphology [3]. Despite these similarities, MSC-NPs do not appear to differentiate into neural lineages either in vitro or in vivo, suggesting that their ability to improve paralysis scores in EAE mice is due to an indirect trophic and immunomodulatory effect [4]. In a previous study, in vitro cocultured MSC-NPs promoted T-regulatory cell expansion as well as oligodendrocyte differentiation, thus supporting an indirect, or bystander, mechanism of action [3]. The aim of the current study was to better understand the impact of MSC-NPs on microglial cells, which play an important role in MS pathology [11]. MSC-NPs were found to shift microglial cells from a proinflammatory M1 phenotype to a proregenerative M2 phenotype utilizing two separate in vitro models of microglia. The effect of MSC-NPs on microglia was mediated by factors secreted by MSC-NPs that act in a paracrine manner on microglial cells. Importantly, we showed unequivocally that clinically relevant human MSC-NPs from multiple different MS donors can mediate this effect, highlighting batch-to-batch consistency and therapeutic potential of autologous MSC-NPs.

A distinct advantage of the current study is the use of human iPSC-derived microglia as a model to investigate mechanisms of microglial polarization in response to human MSC-NPs. Although mouse BV-2 cells are widely used as an in vitro model of microglia, BV-2 cells are immortalized and do not completely recapitulate the morphology, phenotype and function of activated primary microglia [25]. In our own study, stimulated BV-2 cells upregulated nitric oxide synthase whereas activated human iPSC-microglia did not, consistent with previous reports [26]. Although previous studies with MSC cocultures utilized primary microglia as a model, both mouse and human primary microglia differ substantially from resting microglia that reside in the CNS with regard to their phenotype and gene expression signatures, including genes associated with microglial activation [15,20,21,27–29]. During efforts to develop a more appropriate model of human microglia, recent reports have shown that iPSC-derived microglia recapitulate the microglial morphology, phenotype, gene expression signature and function of mature primary microglia both in vitro and when transplanted into mice [23,24]. There have been no reported differences in iPSC-derived microglia generated from MS patients, despite the finding from MS genome-wide association studies that MS risk genes are enriched in microglia [30,31]. As such, the current study utilized iPSC-derived microglia from one MS patient and one healthy control interchangeably.

The microglial response to the bacterial cell-wall component LPS has been well characterized; it results in a robust proinflammatory M1 response with increased microglial expression of a wide range of both pro- and anti-inflammatory markers [32–34]. However, the microglial response to LPS may lack physiological relevance in the context of CNS diseases such as MS that occur in the absence of a bacterial infection. Therefore the cytokines IFN-γ, TNF-α and IL-1β, which are known to be elevated in the CNS in MS, are likely to better represent the consequences of the neuroinflammatory milieu in MS [35]. Indeed, previous studies have shown that stimulation of microglia with IFN-γ, TNF-α and IL-1β compared with LPS results in a more selective proinflammatory response in microglia, which was more conducive to polarization into a proregenerative phenotype [32,33]. Similarly, we demonstrated evidence of M1-to-M2 polarization when human iPS-derived microglia were stimulated with IFN-γ, TNF-α and IL-1β in the presence of NPCM.

The current study gives important insight into the mechanism of action of MSC-NPs on microglia and suggests that at least some of the paracrine action of MSC-NPs can be attributed to TGF-β signaling. TGF-β plays a well-established role in microglial homeostasis and in the shift from an M1 to an M2 phenotype [36,37]. We found that the effects of NPCM on microglial activation were reversed when TGF-β signaling through the TGF-β type I receptor was inhibited. These findings were consistent with a previous study demonstrating that TGF-β secretion from rat MSCs similarly modulated rat microglial activation in response to LPS [17]. On the other hand, we did not find that CX3CL1 contributed to this mechanism, because inhibition of the CX3CR1 receptor had minimal effect. Our finding is in contrast to a previous study showing that MSC-derived CX3CL1 inhibited LPS-stimulated microglial function [15]. The discrepancy may be due to the use of mouse cell models, as previously discussed. Interestingly, we found a correlation between TGFB3 (but not TGFB1 or CX3CL1) gene expression by MSC-NPs and their capacity to inhibit M1 marker expression in microglia. Although these data suggest that TGF-β3 secretion by MSC-NPs may be playing a key role, we were only able detect TGF-β1 protein in NPCM, possibly due to differences in latent protein activation of different TGF-β isoforms [38]. Further studies will be required to determine how TGF-β isoforms play a role in mediating this effect and whether additional MSC-NP factors also contribute to the paracrine action on microglia. Interestingly, both TGF-β1 and TGF-β3 are highly upregulated in MSC-NPs compared with MSCs, possibly correlating with the increased potency of MSC-NPs observed in coculture experiments.

The definition of potency is critical to the clinical translation of cell-based products and depends on an understanding of their mechanism of action in vivo. Potency assays for cell therapies should be based on the route of administration, disease state and applicable mechanism of action. In the context of progressive MS – where innate immune-cell activation predominates in the CNS – potency evaluation based on suppression of T-cell activation lacks relevance, especially when MSC-based products are administered intrathecally [39]. The current study provides evidence supporting a link between CCL2 expression by microglia and exposure to the paracrine action of MSC-NPs, suggesting that potency assays involving in vitro microglial activation may be more appropriate in this context. The observation that CSF levels of CCL2 may serve as a clinical correlate of in vivo function of intrathecally injected MSC-NPs further supports the link between MSC-NPs and microglial activation [5]. The MSC-NP-associated changes in CSF CCL2 may also reflect an impact on reactive astrocytes, which have been shown to express CCL2 in MS lesions [6]; therefore the impact of MSC-NP treatment on MS-associated astrogliosis requires further investigation. Interestingly, two previous clinical trials testing MSC-based cell therapies administered intrathecally in patients with amyotrophic lateral sclerosis also reported reduced CCL2 levels in CSF following treatment, suggesting a common therapeutic target in both diseases [40,41]. Whether or not CSF CCL2 will serve as a clinical biomarker for therapeutic response in MS remains to be determined. Because the MSC-NPs used in this study were manufactured under cGMP conditions for the purposes of administration into clinical trial subjects (clinicaltrials.gov ID NCT03355365 and NCT03822858), all cell samples maintained a high degree of batch-to-batch consistency and were coupled to complete donor information. Future investigations will allow correlation of MSC-NP cellular attributes, including the expression level of TGF-β and suppression of microglial activation in vitro, with clinical outcomes and biomarker analysis in MSC-NP-treated MS patients.

There are several limitations to the current study. Although the iPSC-derived microglia model has certain advantages over other in vitro models, this simplified model does not fully recapitulate the diversity of phenotypes seen in vivo, including disease-associated microglial phenotypes that may be specific to MS [42]. Further studies in the EAE animal model will be necessary to understand how specific microglial populations are altered in response to MSC-NPs. In addition, exciting new neuroimaging approaches to detect microglial activity in MS patients may help corroborate the effect of MSC-NPs on microglial activation in future clinical trials [9]. Finally, the mechanism of action of MSC-NPs on microglia is likely to involve a cocktail of paracrine factors, in addition to TGF-β. More detailed characterization of the MSC-NP secretome will be required to elucidate the specific factors that participate in suppressing microglial activation.

Conclusion

In conclusion, this study is the first to show that MSC-NPs promote microglial polarization from a proinflammatory to a proregenerative phenotype based on microglial marker expression, morphology and phagocytic function. The paracrine mechanisms underlying this effect may involve TGF-β signaling in microglia. This study is the first step in developing a robust in vitro potency assay for MSC-NPs that correlates to their predicted mechanism of action, with the potential to help predict clinical efficacy in patients with MS.