Immunomodulation and tissue regeneration are fundamental mechanisms in MSC-based cellular therapy. These biological features are closely regulated in response to local environmental cues including the oxygen level, presence of growth factors and inflammatory cytokines, and other factors. Changes in the microenvironment during disease progression could provoke protective “pre-homeostatic” mechanisms in MSCs and further enhance their therapeutic efficiency [29–32]. Indeed, our current findings show that MSCs preconditioned with LPS plus TNF-α, macrophage-associated inflammatory mediators, synergistically enhanced M2 macrophage polarization and osteogenic differentiation.
Previous studies of MSC preconditioning have mainly focused on studying the effects of a single inflammatory mediator on MSCs [19, 29–33]. These approaches have clarified valuable information regarding the biological features of MSCs. However, as the complex tissue microenvironment contains multiple inflammatory factors, we speculated that it might be possible to further enhance the beneficial osteogenic and immunomodulatory properties of MSCs by exposing the cells to multiple inflammatory mediators. IFN-γ is mainly produced by activated Th1 lymphocytes and is a crucial regulator of both innate and adaptive immune response . Ren et al.  demonstrated that MSC-mediated suppression of T-lymphocyte activation required the presence of IFN-γ and one of the proinflammatory cytokines TNF-α, IL-1α, or IL-1β. LPS is a potent ligand for macrophages that activates the proinflammatory response including TNF-α production via TLR4 signaling. TNF-α is a strong proinflammatory cytokine, produced in high amounts during inflammatory and immune reaction by various cells including M1 macrophages and Th1 cells. Our preconditioning strategy using high-dose LPS and TNF-α could incite physiological protective mechanisms against an overt inflammatory reaction and enhance the immunomodulatory properties of MSCs . However, the proinflammatory stimulation-induced immunomodulation could also manipulate the infiltrated MSCs into tumor-derived MSCs and enhance tumor progression . Indeed, as both TLR ligands and macrophage-derived TNF-α are likely to be present at the site of injury, they might work synergistically to induce the physiological transition from acute inflammation to tissue regeneration. Previous studies have shown that the application of MSCs preconditioned by a single factor improved tissue repair using in-vivo models [32, 36], demonstrating that ex-vivo conditioning with factors such as LPS/TNF-α can stably increase their therapeutic efficiency. The underlying mechanisms of synergistic effect by LPS plus TNF-α or other proinflammatory cytokines remain unclear.
Crosstalk between MSCs and macrophages including paracrine regulation and direct cell contact is crucial for effective immunomodulation and tissue regeneration [7, 19, 33]. Our current transwell coculture and conditioned medium treatment models demonstrated that paracrine factors secreted by the preconditioned MSCs were effective in modulating macrophage polarization from the inflammatory M1 phenotype toward the anti-inflammatory and tissue-regenerative M2 phenotype. Previous studies have shown that direct PGE2 treatment of macrophages increased M2 markers including Arg1, IL-10, and MMP9, and that this effect was blocked by PGE2 inhibitors [37–39]. In the current study, inhibition of the COX2/PGE2 pathway blocked the upregulation of M2-related Arg1 expression in macrophages, which is correlated with tissue repair. However, blocking of the COX2/PGE2 pathway had limited effects on CD206 expression and TNF-α/IL-1Ra ratio, suggesting that other secreted factors are also involved in immunomodulation. A previous proteomic analysis has identified 118 TNF-induced paracrine factors in human adipose-derived MSCs including various cytokines, chemokines, proteinases, and immune modulators such as long pentraxin 3 . Analysis of paracrine factors secreted by the preconditioned MSCs is required to further clarify the underlying mechanisms of immunomodulation. Nemeth et al.  have shown that MSCs induced anti-inflammatory cytokine production in macrophages in the coculture system of direct contact or transwell (paracrine regulation only) models. The induction is higher with direct cell-to-cell contact compared to paracrine regulation alone, indicating that both physical contact and paracrine factors are involved in the MSC-mediated immunomodulation. Our study focused on the transwell system to clarify the MSC-mediated paracrine factors including nitric oxide and PGE2. Currently, the molecular candidates involved in the immunomodulation via physical contact between MSCs and macrophages remain unclear. Our data showed that MSC preconditioning alone increased osteogenic differentiation in the absence of macrophages in vitro. Sequential modulation of macrophage polarization status by the preconditioned MSCs could further enhance the osteogenic potential during this crosstalk .
MSCs can modulate adaptive and innate immunity by distinct mechanisms. Ren et al.  showed that MSC-mediated suppression of T-cell proliferation required iNOS-dependent nitric oxide production. Exposure of LPS plus TNF-α synergistically induced iNOS expression in MSCs (Additional file 1: Figure S4); however, selective inhibition of iNOS activity showed a limited effect on M2 macrophage polarization (data not shown). In addition, mechanistic studies demonstrated that the IFN-γ and TNF-α-induced immunomodulation in human MSCs is mediated by indoleamine 2,3-dioxygenase (IDO) instead of the iNOS pathway . The different potential mechanisms to modulate macrophage polarization between human and mouse MSCs exposed to LPS plus TNF-α remain to be clarified.
The critical roles of COX2 in bone formation and repair have been demonstrated in COX2 knockout transgenic mouse [28, 41, 42], and by rat and rabbit models using nonselective nonsteroidal anti-inflammatory drugs (NSAIDs) [43–45]. However, clinical findings regarding the effects of COX-2 inhibitors and other NSAIDs on bone formation remain controversial [46–48]. Our data showed that transient inhibition of COX2 in MSCs by the selective COX-2 inhibitor Celecoxib increased osteogenesis, regardless of the treatment time points (Additional file 1: Figure S3). Interestingly, continuous treatment with Celecoxib had no significant effects on osteogenic differentiation. Other in-vitro studies showed that the suppressive effect of Celecoxib on osteogenesis was only observed in the presence of inflammatory environments, such as proinflammatory cytokines or macrophages [49, 50]. Therefore, the increased osteogenesis in MSCs treated with Celecoxib could be limited to our in-vitro model in the absence of other proinflammatory environmental factors.
Preconditioned MSCs could display proinflammatory or anti-inflammatory phenotypes in response to different ligands . Transient exposure (< 1 hour) of MSCs to 10 ng/ml LPS led to proinflammatory “MSC1” phenotypes and increased T-cell activation in the coculture experiment. Comparatively, the “MSC2” phenotype exposed to 1 μg/ml poly I:C was able to suppress T-cell activation, enhance IDO expression, and increase PGE2 production . Our data and other studies show that MSCs exposed to LPS (1–20 μg/ml) induced a protective anti-inflammatory effect on macrophages [19, 52]. The difference in the dose, exposure time, and affected cell types (T lymphocytes or macrophages) could determine the biological roles of LPS-preconditioned MSC.