MSCs are ideal cellular candidates for therapy. They expand well in culture, have relatively low tumorigenic and alloreactive risk, and preferentially home to areas of injury [1, 2]. MSCs ameliorate disease in various animal models of neuroinflammation [12, 13, 24, 33, 34], and these outcomes are attributed mainly to the immunomodulatory properties of MSCs. To ensure their therapeutic efficacy, knowledge of the number of MSCs required to control inflammation, the need for MSCs to be in close vicinity to the site of tissue injury, and the effects of the inflammatory environment on MSCs should be deduced. We demonstrate here that interactions between MSCs and microglia are distinct dependent on cell number and their proximity, and that the inflammatory milieu dictates the modulatory properties of MSCs.
Bone marrow-derived MSCs inhibit proliferation of lipopolysaccharide (LPS)-activated BV2 microglia, signifying the suppressive effects of MSCs on microglia that have previously been reported [32, 35]. LPS serves as a trigger for microglia activation to an inflammatory phenotype [23, 32, 36, 37], increasing their proliferation, upregulating expression of activation markers, and inducing production of inflammatory mediators. One of the mediators produced is nitric oxide (NO), generated by inducible NO synthase (iNOS). NO mediates inflammatory responses within the CNS [38–40], and excessive amounts can cause neurotoxicity [19, 39].
We demonstrated MSCs to increase NO levels in LPS-treated microglia cocultures. As these cells are grown within the same culture well, we are unable to tag the increase in NO levels in cocultures to BV2 microglia or MSCs; however, we have found both cell types capable of producing their own large amounts of NO. For MSCs, a direct inflammatory stimulus (LPS) does not induce NO production, nor do soluble factors from resting, inactivated microglia. Only when exposed to soluble factors from LPS-activated microglia did we observe MSCs producing substantial amounts of NO. Similarly, MSCs produce NO only when cocultured with stimulated, and not unstimulated, T lymphocytes . Ren and colleagues  also showed mouse bone marrow MSCs to secrete high concentrations of NO in a specific paradigm; LPS stimulation fails to induce NO production in MSCs, but exposure to inflammatory cytokines IFN-γ and TNF-α results in an NO surge. Therefore, the biologic cue for MSCs to produce NO seems not to be directly from an inflammatory agent, but rather from the ensuing cellular/tissue reaction. We do, however, show that a preexposure of MSCs to LPS augments NO production by MSCs cultured subsequently with microglial soluble factors. This demonstrates the priming effect of MSCs, appearing as if the inflamed microenvironment prepares and licenses subsequent MSC interactions with microglia within this paradigm. This is corroborated by the fact that both mouse and human MSCs can recognize LPS as they express TLR4 , the Toll-like receptor that binds LPS. The pleiotropic functions of NO make it difficult to deduce the implications of the increased NO in MSC/microglia cocultures. T-lymphocyte immunosuppression by MSCs is mediated by NO; MSCs secrete high levels of NO that suppress T-cell proliferation, and inhibition of iNOS restores proliferation of splenocytes in MSC cocultures . In the MSC/microglia model, reducing NO levels by 40% to 50% with an iNOS inhibitor did not affect microglia proliferation , and a role for NO in the MSC/microglia paradigm remains undefined.
We also showed distinct responses of MSCs and microglia in terms of inflammatory cytokine secretion. LPS triggers both BV2 microglia and MSCs to produce IL-6. Similarly, ligation of TLR4 has been shown to induce MSCs to secrete IL-6 [42–44]. MSCs also secrete large amounts of IL-6 when exposed to macrophages [45, 46] and astrocytes ; however, the ensuing effects of IL-6 are unclear. In MSC/microglia cocultures, IL-6 increases in coculture, whereas TNF-α levels decrease in a dose-dependent manner. For T-lymphocyte immunosuppression, IL-6-dependent production of prostaglandin E2 (PGE2) was shown to be important for antiproliferative effects of MSCs . Kim and Hematti  suggested that an IL-10-high, IL-12-low, IL-6-high, and TNF-α-low expression pattern defines a subtype of M2 macrophages that may have a role in tissue repair, but cytokine expression for such a tissue-repair subtype is undefined for microglia. The surge in IL-6 was augmented in the presence of both soluble factors and cell-to-cell contact, whereas soluble factors alone were sufficient to reduce TNF-α levels in cocultures. Therefore, if the combinatory effects of an IL-6 surge and a TNF-α reduction are essential to confer microglia modulation, MSCs may require being in the immediate vicinity of activated microglia. Although IFN-γ is strongly implicated in modulating T-cell inhibition by MSCs , its levels in MSC/microglia cocultures were negligible.
An interesting question to ask is whether MSCs are required to be in close vicinity to microglia to dampen microglia inflammatory responses. We have already discussed that the reduction in TNF-α does not require cell-to-cell contact between MSCs and microglia. Increased NO in MSC/microglia cocultures also occurs without cell-to-cell contact. These effects appear to be conferred without MSCs having to be in direct contact with microglia. We showed that MSCs do not home toward LPS, but are attracted instead to resting and, more so, to activated BV2 microglia. Similarly, MSCs migrate toward human macrophages, with the authors identifying CCL2, CCL5, and IL-8 as chemotactic signals for MSCs . It is often shown that MSCs home toward areas of tissue damage [49–52], and our results indicate that, within the brain, it is not the inflammatory agent that serves as a homing signal, but mediators secreted by microglia. It would be interesting to determine the chemotactic signals that microglia produce to attract MSCs. Although CCL2 is chemoattractant for MSCs [45, 53], we speculate that it does not have an autocrine effect on MSCs, as MSCs do not migrate significantly toward LPS, although it produces large amounts of CCL2.
Microglia themselves actively (and equally) migrate toward both an LPS inflammatory stimuli and MSCs, regardless of the primed status of the stem cells. This is different from that for T lymphocytes that only migrate toward MSCs that have been exposed to proinflammatory cytokines . It appears that microglia are compelled to be in close proximity to MSCs. Although LPS-primed MSCs produced high levels of CCL2, a macrophage chemokine, unprimed MSCs conversely do not secrete CCL2; therefore, the chemoattractant responsible for the pronounced migration of microglia toward unprimed MSCs must be other than CCL2. It is possible that MSCs may be diverting the migration of microglia toward them and away from an inflamed site. We showed that, even without interacting with microglia, MSCs produce soluble factors that increase microglia migration. Perhaps the chemotactic impetus from MSCs encourages microglia to form closer contact with MSCs within the inflamed area for subsequent immunosuppression. These are attractive possibilities, best answered with in vivo or ex vivo approaches. Interestingly when cultured with T lymphocytes, MSCs are inclined to produce large amounts of T cell-associated chemokines, such as CXCL9, CXCL10, and CXCL11 , indicating that MSCs may have distinct mechanisms for modulating different immune cells, and their effects are dictated by the cells with which they come in contact.