• Open Access

Concise Review: Two negative feedback loops place mesenchymal stem/stromal cells at the center of early regulators of inflammation


  • Darwin J. Prockop

    Corresponding author
    1. Texas A&M Health Science Center College of Medicine Institute for Regenerative Medicine, Scott & White, Temple, Texas, USA
    • M.D., Ph.D., Texas A&M Health Science Center College of Medicine Institute for Regenerative Medicine, Scott & White, Module C, 5701 Airport Road, Temple, Texas 76502, USA. E-mail: Prockop@medicine.tamhsc.edu Telephone: 254-771-6800, Fax: 254-771-6839

    Search for more papers by this author

  • Author contributions: D.J.P.: conceived the manuscript and wrote it with no input or support from other sources.


Recent data demonstrated that MSCs can be activated by proinflammatory signals to introduce two negative feedback loops into the generic pathway of inflammation. In one loop, the activated MSCs secrete PGE2 that drives resident macrophages with an M1 proinflammatory phenotype toward an M2 anti-inflammatory phenotype. In the second loop, the activated MSCs secrete TSG-6 that interacts with CD44 on resident macrophages to decrease TLR2/NFκ-B signaling and thereby decrease the secretion of proinflammatory mediators of inflammation. The PGE2 and TSG-6 negative feedback loops allow MSCs to serve as regulators of the very early phases of inflammation. These and many related observations suggest that the MSC-like cells found in most tissues may be part of the pantheon of cells that protect us from foreign invaders, tissue injury, and aging. STEM Cells 2013;31:2042–2046

Mesenchymal stem/stromal cells (MSCs) produced beneficial effects in a series of models for human diseases by multiple actions that include modification of excessive inflammatory and immune responses, enhancement of tissue regeneration, and transfer of microvesicles that can contain microRNAs and mitochondria [[1-4]]. The role of MSCs in modifying inflammation has attracted attention in part because inflammation has been recognized to be an essential but very complex defense mechanism [[5-7]]. But inflammatory responses can frequently “go awry” [8]. Accordingly, “sterile inflammation” [6] that occurs when there is no threat from invading micro-organisms is now recognized to be an excessive response that contributes to the tissue destruction seen in diseases such as myocardial infarction and stroke. Also, “unresolved inflammation” is now known to contribute to the tissue destruction seen in a long list of chronic diseases that include obesity, diabetes, degenerative diseases of the central nervous system, atherosclerosis, and some cancers [[6-9]]. MSCs have entered the story of inflammation because recent reports suggest that modulation of inflammation accounts for some of the beneficial effects that have been observed with administration of cells in animal models for several diseases. Explaining the effects of MSCs on inflammation is not simple, since research on inflammation now spans multiple disciplines. Also, MSCs are proving to be more and more complex as the evidence mounts that they drastically change their properties with different culture conditions and in response to different tissue microenvironments [1]. But there is good news. Some of recent observations with MSCs can be nicely incorporated into the detailed road map we now have for inflammatory responses.

As suggested by Medzhitov [[7, 9]], inflammation has four basic components (Fig. 1; Table 1): (a) Inducers that are either exogenous or endogenous stimuli for inflammation such as the products of micro-organisms or injured cells; (b) Sensors that consist primarily of resident macrophages and mast cells that express receptors for Inducers and respond by producing a variety of proinflammatory Mediators; (c) Mediators in the form of cytokines, chemokines, amines, and eicosanoids released by the Sensors; (d) Effectors consisting of adjacent cells in tissues that respond to Mediators by amplifying their signals to usher in the exudate of plasma and leukocytes. The result is the four cardinal signs of inflammation that have been traced back to the first century Roman physician Cornelius Celsus [9], that is, redness, swelling, heat, and pain. Among the many recent reports linking MSCs to inflammation, several demonstrate that MSCs can modulate the generic pathway of inflammation by introducing two negative feedback loops (Fig. 2).

Figure 1.

Inflammatory pathway components. The inflammatory pathway consists of Inducers, Sensors, Mediators, and target tissues. Inducers initiate the inflammatory response and are detected by sensors. Sensors, such as TLRs are expressed on specialized sentinel cells, such as tissue-resident macrophages, dendritic cells, and mast cells. They induce the production of mediators, including cytokines, chemokines, bioactive amines, elcosandoids, and products of proteolytic cascades, such as bradykinin. These inflammatory Mediators act on various target tissues to elicit changes in their functional states that optimize adaptation to the inflammatory response. The specific components shown represent only a small sample of the myriad different sensors, mediators, and target tissues involved in the inflammatory response. Reported with permission from Medzhitov [9]. Abbreviations: CCL2, chemokine (C-C motif) ligand; CXCL8, (IL-8); IL, IL-1, IL-6; TNFα, tissue necrosis factor alpha; TLR, Toll-like receptor.

Figure 2.

Schematic of TSG-6 and PGE2 negative feedback loops introduced into inflammatory responses by MSCs. Abbreviations: MSCs, mesenchymal stem/stromal cells; IL, IL-10, IL-1; PGE2, prostaglandin E2; TSG-6, TNFα stimulated gene/protein 6; TNFα, tissue necrosis factor alpha.

Table 1. Examples of inflammatory pathways
  1. Reproduced with permission from Medzhitov [7].
  2. Abbreviations: IL, IL-1, IL-6; NALP3, NOD-like receptor protein; PGE2, prostaglandin E2; TLR, Toll-like receptor.
LipopolysaccharideTLR4TNF-α, IL–6, and PGE2Endothelial cells, hepatocytes, leukocytes, the hypothalalus, and others
AllergensIgEVasoactive aminesEndothelial cells and smooth muscle cells
Monosodium urate crystals and calcium pyrophosphate dihydrate crystalsNALP3IL–1βEndothelial cells, hepatocytes, leukocytes, the hypothalalus, and others
CollagenHagema factorBradykininEndothelial cells and smooth muscle cells

One negative feedback loop is initiated by proinflammatory Mediators from Sensors that activate MSCs to upregulate expression of COX2 and other components of the arachidonic acid pathway. As a result, the MSCs increase secretion of prostaglandin E2 (PGE2). The PGE2 then drives transition of resident macrophages from the classic M1 phenotype that is proinflammatory toward a more poorly defined M2 phenotype in which the cells secrete anti-inflammatory mediators such as interleukin-10 (IL-10) and interleukin-1 (IL-1) receptor antagonist.

The second negative feedback loop is also introduced by MSCs being activated by Mediators from Sensors. The activated MSCs increase expression of a number of genes, including the anti-inflammatory protein TNFα stimulated gene/protein 6 (TSG-6). TSG-6 has multiple anti-inflammatory actions [[10, 11]]. Among its many effects, TSG-6 was recently shown [12] to interact with CD44 on resident macrophages, either directly or in a complex with hyaluronan, to dissociate CD44 from Toll-like receptor (TLR2) and thereby limit TLR2 driven NFκ-B signaling. The result is a decrease in the secretion of TNFα and other Mediators that target Effectors (Fig. 2).

The PGE2 negative feedback loop introduced by MSCs was first observed in a model of sepsis in mice induced by cecal ligation and puncture to release intestinal bacteria into the peritoneum [13]. Intravenously administered mouse MSCs were activated by Mediators to secrete PGE2, and the PGE2 drove resident macrophages toward what is referred to as an M2 anti-inflammatory phenotype [[5, 14]]. The M2 macrophages, in turn, secreted IL-10 that was released into the circulation to reduce the systemic effects of sepsis and decrease mortality of the mice. Numerous recent reports have described the propensity of MSCs in secreting PGE2 to suppress inflammatory and immune responses in reactions that may involve generation of M2 macrophages or more complex effects of PGE2 [[15-20]].

The TSG-6 feedback loop introduced by MSCs was first observed in a mouse model for myocardial infarction [21]. Intravenously infused human MSCs were trapped in the lungs where they were activated to express TSG-6, and the TSG-6 decreased the inflammatory responses of the infarcted heart. The result was decreased scarring and improved function of the left ventricle. The effects of TSG-6 on TLR2/nuclear factor kappa-light-chain-enhancer of activated B cells signaling in resident macrophages was elucidated in a model for zymosan-induced peritonitis [12] and then in a model for chemical injury of the cornea [22]. However, earlier studies demonstrated that TSG-6 has multiple additional anti-inflammatory actions that may predominate in some situations: It crosslinks proinflammatory fragments of hyaluronan [[10, 11, 23]]; it catalytically transfers a heavy chain from inter-α-inhibitor to hyaluronan and other acceptors to increase inhibition of the cascade of proteases released by inflammation [[10, 11, 24]]; it inhibits transport of leukocytes through endothelial cells [25]; it inhibits osteoclasts by preventing the formation of lacunae on bone [26]; it is involved in matrix remodeling during ovulation [[10, 11]]. These earlier observations prompted the initial suggestion that TSG-6 might serve as a negative feedback regulator of inflammation [[10, 11]].

The two feedback loops operate on different anatomical scales. The PGE2 loop requires intimate association or perhaps even cell-to-cell contact of MSCs with resident macrophages, since PGE2 is a potent reagent that produces multiple unwarranted effects when administered systemically [27]. In contrast, MSCs can “act at a distance” through the TSG-6 negative feedback loop, since intravenously infused MSCs suppressed inflammation in the heart or cornea without significant numbers of the cells reaching these tissues [[21, 28]]. Also, intravenously infused recombinant TSG-6 reproduced most of the effects of intravenously infused MSCs in both the model of myocardial infarction [21] and a model for chemical injury to the cornea (Fig. 3) [29].

Figure 3.

TSG-6 reduced corneal opacity following injury. Representative photographs demonstrated the corneal surface on days 3, 7, and 21 after injury to the cornea of rats produced by 15-second exposure to ethanol and then scraping to remove the epithelium and the stem cells located in the limbus. Marked corneal opacity was present by day 3 postinjury. Note the clear pupillary margin and well-demarcated light reflex in corneas following injection of 5 μg into the anterior chamber immediately following the injury. Reproduced from Oh et al. [29]. Abbreviations: PBS, physiological buffered saline; TSG-6, TNFα stimulated gene/protein 6.

The TSG-6 loop was defined in some detail in a model of inflammation of cornea [[22, 29]]. As in most models for inflammation, two phases were apparent. There was an initial mild inflammatory phase I that lasted 6–8 hours (Fig. 4), a time during which Inducers were released from injured cells and Sensors were activated (Fig. 1) [[7, 9]]. This was followed by a larger inflammatory phase II that peaked at 24–48 hours (Fig. 4) during which Mediators released by Sensors activated Effectors. In Phase I, the primary Inducers were secretoneurin released from sensory nerves and a small heat shock protein (HSPB4) released from necrotic and injured keratocytes that form the stroma of the cornea (Fig. 5). The primary Sensor was TLR2 on resident macrophages. HSP4 interacted with TLR2 to increase NFκ-B signaling and release of a series of Mediators. The negative feedback loop introduced by TSG-6 largely aborted phase II (Fig. 4). TSG-6 was effective only if applied within the first 6 hours after injury to the cornea [29], an observation consistent with its interaction with resident macrophages during Phase I.

Figure 4.

Time course of neutrophil infiltration in the cornea as measured by MPO. Corneas were injured as in Figure 3. Neutrophils infiltrated the cornea in two phases: phase I, a small initial phase that began within 15 minutes and reached a plateau level at 4–8 hours; phase II, a larger infiltration of neutrophils that followed and peaked at 24–48 hours. Reproduced from Oh et al. [29]. Abbreviations: MPO, myeloperoxidase; PBS, physiological buffered saline; TSG-6, TNFα stimulated gene/protein 6.

Figure 5.

Graphic summary of sterile inflammation in the cornea. Immediately after injury, secretoneurin is released from nerve endings in the cornea to recruit circulating neutrophils and induce an initial inflammatory response (phase I). The necrotic or injured keratocytes release HSPB4 in response to injury and oxidative stress. The HSPB4 activates resident macrophages in the cornea via TLR2/NF-KB signaling pathway to produce proinflammatory cytokines including IL-1 and IL-6. The proinflammatory signals released by resident macrophages are amplified by keratocytes that produce chemokines to recruit further, large amount of neutrophils (phase II). TSG-6 inhibits the initial activation of resident macrophages by modulating TLR2/CD44/NF-KB signaling and thereby decreases the phase II inflammatory response. Reproduced from Oh et al. [29]. Abbreviations: CXCL, chemokine (C-C motif) ligand; HSPB4, heat shock protein; IL, IL-1, IL-6; MIP, macrophage inflammatory protein; ROS, reactive oxygen species; TLR, Toll-like receptor; TSG-6, TNFα stimulated gene/protein 6.

The effectiveness of MSCs and TSG-6 in the cornea model was surprising in the light of the previous history of tests of anti-inflammatory agents. Chemical and thermal injuries to the cornea are a common medical problem with over 600,000 patients with such injuries appearing in emergency rooms in the US each year. Anti-inflammatory agents have been tested extensively in the patients. All have failed. In fact, the broad class of nonsteroidal anti-inflammatory drugs are a danger to the patients because they produce melting of the cornea and a risk of rupture [[30, 31]], apparently because the tissue expresses high levels of prostaglandins and lipoxigenases [32]. The effectiveness of TSG-6 raises the possibility that it may represent a new class of anti-inflammatory agents.

The two negative feedback loops may in part explain the ability of MSCs to modulate immune responses [[33-36]]. Inflammation, also referred to as innate immunity, is an essential first step for triggering adaptive immunity [9]. By modulating inflammation MSCs may therefore reduce the acquired immune response. In addition, a series of observations indicate MSCs can directly decrease the acquired immune response through their effects on antigen-presenting cells, T lymphocytes, and B lymphocytes [[33-36]].

The evidence for the two negative feedback loops was developed primarily through experiments in which the MSCs were administered shortly after tissue injury and the release of Inducers. Further experiments will be required to determine whether they also can be invoked during the mild and unresolved inflammation that characterizes many chronic diseases, in part because it is still not known how the generic pathway for inflammation (Fig. 1) applies to these conditions.

At the same time, it is apparent that the PGE2 and TSG-6 feedback loops cannot account for all the beneficial effects that have been observed with administration of MSCs in disease models. The cells express a large number of cytokines, chemokines, and other regulatory factors in culture under normal conditions and an even larger number in response to different microenvironments. The list of such additional factors produced by MSCs is very long [[1, 2, 37]] and includes factors that enhance vascularization (vascular endothelial growth factor and IL-6), enhance cell proliferation (tissue growth factor beta, keratinocyte growth factor), modulate immune responses (indoleamine 2,3-dioxygenase in human MSCs; induced nitric oxide synthase with mouse MSCs, CCL2, and matrix metalloproteinase 9), reduce reactive oxygen species and apoptosis (stanniocalcin-1), and are antibacterial (peptide LL-37). Any of these alone or in combination may predominate in some situations.

Most of the data for the PGE2 and TSG-6 negative feedback was developed from experiments in which large numbers of MSCs were administered to mice, usually 40 million/kg. Therefore, there is no clear demonstration that there are sufficient numbers of MSC-like cells in most tissues to play the same role in modulating inflammation. The most suggestive evidence comes from transgenic mice with null alleles for TSG-6: they demonstrated increased inflammatory responses when challenged by an inducer of arthritis [38], perhaps because of the absence of the TSG-6 negative feedback loop. There is still uncertainty about the precise definition of MSCs, and multiple roles have been assigned to tissue endogenous MSCs. For example, there are observations that they serve as niche cells for hematopoietic stem cells [39] and some cancers [[17, 40]], that they are perivascular adventitial cells [41], and that they may be widely dispersed in most tissues as cells with many of the properties of induced pluripotent cells [42]. The PGE2 and TSG-6 negative feedback loops together with a large number of related observations indicate that MSCs are multifaceted chameleons much like T lymphocytes, and that they probably belong to the pantheon of cells that protect us from foreign invaders, injury, and aging.

Disclosure of Potential Conflicts of Interest

Consultant/advisory role: Advisory Board, Temple Therapeutics LLLC; research funding/contracted research (including funds paid to the institution) and ownership interest (stocks, stock options, or other ownership interest excluding diversified mutual funds), Temple Therapeutics LLC < 5%.