MSC (also known as MyoR (42), ABF-1 and bHLHa22) is a basic helix-loop-helix transcription repressor that is involved in the regulation of muscle development (42, 43). MSC can bind to an E-box motif (CANNTG) either as a homodimer or as a heterodimer with E2A gene products, where the dimers bind the same E-box sequence as myogenic bHLH/E protein heterodimers, but MSC functions as a potent transcriptional repressor that inhibits myogenesis and E-box-dependent muscle gene activation (44). MSC shows two isoforms (45): a 201 aa protein (MSC 1a) and an equally effective 180 aa short form (MSC 1b), with a repressor domain between aa 158-173. Although early studies indicate that MSC has a muscle-specific, developmental stage-restricted expression profile, more recent studies support the notion that “Musculin” could very well be a misnomer, as its expression is also found in several adult tissues of non-muscle lineage. For instance, MSC is expressed in kidney side population (SP) cells, which are enriched with stem cells, and can regulate their function (46).

In lymphoid compartment, human MSC is known as activated B cell factor-1 (ABF-1). MSC facilitates the formation of germinal center B cells and memory B cells, but inhibits the differentiation of antibody-secreting plasma cells (47). In Hodgkin’s lymphoma, B cell-specific transcription program is interrupted by overexpression of ABF-1 (a.k.a. MSC) and Id2, which antagonize the function of the transcription factor E2A in B cell lineage determination. As a result, expression of B cell specific genes is lost, whereas genes normally not associated with the B cell lineage are upregulated (48).

The expression and regulatory roles of MSC in T cells, especially CD4+ T helper cells, await further clarifications. One group reported that MSC is induced in murine Tfh but not Th17 subset, yet MSC is not required for the differentiation of Tfh cells, nor plays a significant role in antibody production in response to T-dependent antigens (49). This is in contradiction to the finding that MSC is selectively expressed in human Th17, but not Tfh cells, and inhibits Th17 proliferative response to IL-2 (50). Furthermore, MSC is found to be preferentially upregulated in peripherally-induced Tregs (iTregs) than in other Th subsets, and it promotes unidirectional development of iTregs by inhibiting the Th2 differentiation pathway (51). It is likely that confusion in the expression profiles of MSC stems from the various conditions of in vitro Th cell differentiation.

To avoid the potentially non-physiological anti-CD3/CD28-stimulated differentiation, in our own work, we generated alloantigen-activated Th subsets whose TCR recognizes α-chain of I-E in the context of I-A^b. MSC is dramatically upregulated in alloantigen-stimulated murine Th17, marginally in Th1, but not in Th0, Th2, or iTreg subsets (52). APC-derived IL-23 has a salutary role in MSC induction. In the in vitro activation system with anti-CD3/CD28 in the absence of antigen-presenting cells (APC), Th17 cells also express more MSC than Th0 and Th1 cells, but only at later time points (6-8 days) (52). Such kinetics suggests that most researchers could have missed MSC in fully developed Th17 cells if they performed Th subset differentiation in the absence of APC in the first 2–4 days (49, 51). In addition, we found that when naïve CD4+CD62L^hiGFP- cells from IL-17A^GFP knockin mice were differentiated and FACS-sorted based on GFP intensities, MSC correlates with stronger Th17 phenotypes (higher IL-17A/IL-17F/RORγt/RORα/Id2) (52). Clearly, MSC is a Th17-specific TF in both human (50) and mouse (52) systems.

Recently, AhR has been implicated as a master regulator involved in the postnatal maintenance of intestinal RORγt+ ILCs and intraepithelial lymphocytes (69). Various populations of IL-22-producing ILCs all express RORγt and AhR, and all have a shared requirement for IL-23 to produce IL-22 (24, 70). As all the ILC progenitors express Id2 (71) and MSC interacts with Id2 to antagonize the function of E2A (44, 48), MSC could directly or indirectly participate in ILC development through modulating E2A activities on RORγt and AhR transcription, as discussed above. More studies are needed to reveal the role of MSC in ILCs.

In our own work, we examined Rorc(γt)^+/GFP mice whose RORγt^GFP+ cells are highly enriched in early thymic CD4+ cells and small groups of LTi/Th17/ILCs found in the gut-associated lymphoid tissues (GALT) (72). In Peyer’s patches (PP), memory CD62L^loRORγt^GFP+ cells are either CD4+ or CD4- (52). Compared to CD4+RORγt^GFP−, both CD4+RORγt^GFP+ and CD4-RORγt^GFP+ cells express high levels of MSC, as well as RORγt, RORα and Id2 (52). While CD4+RORγt^GFP+ LTi/Th17 cells express IL-22 and IL-17A, CD4-RORγt^GFP+ cells express more IL-22, RORγt and RORα, but no IL-17A (52), suggesting they contain IL-22-secreting ILC3s (a.k.a. NK22 cells or ILC22s). Thus, MSC is uniquely co-expressed with transcription factors RORγt/RORα/Id2 in CD4+IL-17A+ (LTi/Th17) cells, as well as in CD4-IL-17A^−/lo populations, the latter of which contains IL-22-producing ILC3s (52). Taken together, MSC is expressed not only in freshly induced Th17 cells, with MSC levels correlated to more committed Th17 phenotypes, but also in RORγt+ IL-22-producing ILC3s in the Peyer’s patches of naïve mice.

Indeed, MSC has a more profound effect on cells from the GALT (such as PP and mesenteric lymph nodes, MLN) than from the spleen and peripheral lymph nodes, which explains why meaningful differences between MSC+/+ and MSC-/- Th subsets are somewhat elusive when naïve cells of spleen or lymph node origin were used for differentiation. For instance, upon Th17 induction by TGF-β1 and IL-6, MACS-enriched CD4+ cells from pooled MLN of MSC-/- mice secreted more IL-22 than the wild type counterparts. Under the same condition though, there was no difference between CD4+ T cells from the spleens of MSC-/- and MSC+/+ mice (52). Moreover, the selective production is more dramatic for IL-22 than for IL-17. In line with this, PP cells from MSC-/- mice express much higher levels of IL-22 than MSC+/+ PP cells, and this difference is greatly amplified during colitis (52). All these results suggest that the elusive effect of MSC is on: 1) a special population of IL-22-secreting cells in the GALT, most likely ILC3s, and/or 2) synthesis and secretion of IL-22 cytokine per se. More experiments are needed to distinguish whether MSC is a negative gatekeeper for ILC3 developmental pathway, and/or a checkpoint TF controlling IL-22 synthesis.

In order to maintain a functioning barrier, all the intestinal epithelial cells are replenished within days by continuously proliferating stem cells residing in the crypt region. Given the role of IL-22 in supporting intestinal LGR5+ stem cell mediated epithelial regeneration (73), as well as its stimulation on the synthesis of antimicrobial peptides (AMPs), it is no surprise that the prevailing view is for IL-22 promoting gastrointestinal health (73–80). For instance, IL-22 plays a beneficial role in acute intestinal infections where rapid repair of colonic epithelium is required, such as infection caused by Citrobacter rodentium (74, 81). Likewise, IL-22 facilitates the restitution of epithelium upon inducing acute colonic injury in dextran sulfate sodium (DSS)-induced murine colitis (75). In cancer chemotherapy, dose-limiting side effects are mainly caused by toxicity in the mucosal surfaces as a result of their high mitotic index. For example, chemotherapy drug methotrexate can induce acute mucositis featured by damage in the epithelium of small intestines, in which IL-22 plays an important restorative role in this self-resolving condition (76). Taken together, the primary insult in these examples is epithelial disruption, where IL-22 drives epithelial cell proliferation and restitution, restoring epithelial barrier. These data have promoted the paradigm of IL-22 being clinically beneficial to repair epithelial damage in human inflammatory bowel disease (IBD), and has led to a clinical trial (NCT02749630), evaluating the role of recombinant IL-22-Fc (Efmarodocokin alfa) in patients with active IBD.

However, the key difference between animal models of colitis and human IBD is that, the former usually results from acute insults on the epithelial barrier with self-limiting tissue injury that can quickly resolve upon withdrawal of the insult, whereas the latter is characterized by chronic immune-mediated inflammatory reactions. While IL-22 definitely has a protective role on gut epithelium (82, 83), it also promotes colitis in a context-dependent manner (30, 84), especially when immune components are involved. For instance, in an adoptive transfer colitis model when Treg-depleted memory/effector CD4+CD45RB^lo T cells were transferred into RAG-1-/- mice, IL-22 derived from these memory/effector T cells is pathogenic, whereas IL-22 is protective when CD4+CD45RB^hi naïve T cells were transferred (84). Note that in this chronic colitis model, transfer of Treg-depleted CD4+CD45RB^lo T cells into RAG-1-/- mice was characterized by a Th1<Th17 cytokine signature in the colon tissues, whereas transfer of CD4+CD45RB^hi naïve T cells demonstrated a Th1>Th17 cytokine signature (84). In an innate colitis model with anti-CD40 administration in RAG-1-/- mice, IL-22-neutralizing antibody significantly reduced the colon pathology, colitis scores and weight loss triggered by anti-CD40 injection, whereas IL-22 plasmid administration exacerbated colitis with infiltration of inflammatory cells and mucosal hyperplasia (30). Interestingly, in that model, the only source of IL-22 in the gut of RAG-1-/- mice is RORγt+ ILC3s, which are required for colitogenesis (30). Thus, it seems that the pathogenic role of IL-22 could be associated with ILC3s and RORγt+ Th17 cells (e.g., through its soluble factor, IL17A, for chemotaxis of immune cells).

Our own work also supports the notion that heightened IL-22 expression by ILC3s when MSC was deficient has an aggravating effect in DSS-induced murine colitis (52, 85). While the wild type mice survived DSS-induced colitis, MSC-/- mice showed increased production of pro-inflammatory cytokines (e.g., IL-22, IL-6, TNF-α) in the gut, with more infiltrating immune cells in colon tissues, reduced body weight, and earlier onset of death. The epithelial injury in the colon tissues was much severer in MSC-/- mice than in the wild type counterpart during colitis, with the counterintuitive observation of higher concentrations of IL-22 in the supernatants of ex vivo cultured colon tissues from the MSC-/- group than from the wild type animals (52, 85). Surprisingly, administration of IL-22BP-Fc, an Fc fusion protein containing the antagonizing IL-22 binding protein (IL-22BP), ameliorated the colitis symptoms in MSC-/- mice (52). This is the first report demonstrating a beneficial effect of IL-22BP in colitis model, where other studies showed the opposite (86, 87). Similar to our finding, in an acute polymicrobial sepsis model, IL-22 blockade by IL-22BP-Fc was also shown to be protective against bacteria spread and organ failure (88). Clearly, a balanced production of IL-22BP is context dependent for intestinal homeostasis.

In addition to the animal data, recent human studies also challenge the dogma of IL-22 being beneficial for colitis, especially during chronic inflammation, instead of in acute, self-limiting mucosal injury (89, 90). First, IBD is not an acute inflammatory disease. Environmental and genetic factors, as well as gut microbiome, cause stable changes of epithelium functions, which have been considered as the major etiology of IBD (91, 92). Second, a humanized monoclonal antibody (Mab) Risankizumab that specifically blocks IL-23, which triggers IL-22 production, demonstrated promising results in a Phase II clinical study in IBD (89). In patients with moderate to severe Crohn’s disease (CD) who had failed prior treatment with TNF-α blockade, another human IL-23-blocking Mab (MEDI2070) also showed clinical improvement in a Phase 2A study (93). Of note, in that study, patients with higher baseline serum IL-22 levels were more responsive to MEDI2070 (93). More detailed mechanistic studies using colonic epithelial organoids and tissue transcriptome analysis in a large datasets of CD patients revealed that elevated IL-22 orchestrates a pathological endoplasmic reticulum (ER) stress response, amplified by IL-17A (90) which is a cytokine normally co-produced with IL-22 by ILC3s (94).

It should be brought to attention that, in IBD patients, persistently elevated IL-22 levels could alter the baseline of the inflammatory responses in the colonic epithelium. IL-22 induces TNF-α, inducible nitric oxide synthetase (iNos2) and apoptosis-inducing caspase 12 and STING (90), which can trigger ER stress. ER stress is caused by overwhelmed protein synthesis under adverse stimuli such as infection and non-physiological cellular insults, resulting in the accumulation of aggregated mis-folded proteins that are potentially toxic to the cells (95). To mitigate these harmful effects, cells develop a highly conserved protective process called unfolded protein response (UPR). IL-22 significantly stimulates proteins involved in regulating UPR in colonic epithelium (90). For example, IL-22-induced iNos2 is a trigger as well as a downstream effector of the ER stress response (96). In addition, iNos2 causes DNA damage associated colon cancer in chronic colitis (97). IL-22 also significantly upregulates TLR-4 and its signaling adapter MyD88 in colonoids (90). Engagement of this pattern recognition pathway likewise induces ER stress in colonic epithelial stem cells (98). IL-22-induced pro-inflammatory molecule STING can increase small intestinal epithelial apoptosis by inducing Type I Interferon production (99). Thus, the role of IL-22 in the context of chronic inflammation during IBD needs to be carefully re-evaluated, and the supplementation of exogenous IL-22, especially in an Fc fusion with long serum half-life, to patients with active colitis should be re-considered with extreme caution.

Nature always has checks and balances. A classical example of a naturally-existing cytokine antagonist is interleukin-1 receptor antagonist (IL-1RA). IL-1RA is not an alternatively spliced form of the extracellular domain of any receptor, but is encoded by a separate gene. It binds non-productively to interleukin-1 receptor (IL-1R) and functions as a molecular sink to quench the activities of IL-1α and IL-1β, modulating a variety of IL-1-related immune and inflammatory responses (100). In a very similar fashion, IL-22 has its own nemesis: IL-22 binding protein (IL-22BP, a.k.a. IL22RA2) (101, 102). IL-22BP is a soluble receptor, 231 amino acids in length with 34% sequence homology to IL-22R1 (101, 103), but binds IL-22 with more than 1,000-fold higher affinity than the membrane form IL-22R1 (104) and inhibits IL-22 biological functions by sequestration from its targets (105).

IL-22BP is produced by epithelial cells, eosinophils, macrophages and dendritic cells (106). While IL-22BP expression is significantly increased during IBD, because of the elevated numbers of IL-22BP-producing eosinophils in human gut (86), it should be kept in mind that such increase could be cause or consequence of colonic inflammation, in relation to the surge of IL-22. More relevant to the context of this review, although many reports suggest a detrimental effect of IL-22BP at blocking the protective effects of IL-22 in acute colitis (75, 86, 87), the role of IL-22BP in chronic IBD, in addition to its beneficial effect on blocking IL-22-mediated tumorigenesis (87) and in acute polymicrobial sepsis (88), warrants further investigation. First, species and model differences have produced some contradictory observations. IL-22BP is highly expressed by immature DC in the normal colon but significantly reduced during acute inflammation in DSS-induced murine colitis (75, 107, 108). In contrast to findings in mice, in patients with IBD increased expression of intestinal IL-22BP was rather observed (86, 87). Second, unlike in mice, human IL-22BP has three functionally different isoforms, which are generated by alternative splicing (109). Isoform 1 is very poorly secreted and fails to antagonize IL-22 signaling. More recently, isoform 1 was found to be largely retained in ER and functions as an intracellular activator of the UPR response (110). Isoform 2 is the only isoform that is secreted in large amounts when myeloid cells are activated by retinoic acid or Toll-like receptor 2. Isoform 3 is the most widely distributed, but its affinity to IL-22 is 27-fold lower than that of isoform 2 (109). Thus, human IL-22BP isoforms play distinct temporal and spatial roles fine-tuning the functions of IL-22.

Regardless of its expression profiles and rheostatic regulation on IL-22, IL-22BP is considered a potential therapeutic target. For example, IL-22BP-Fc reduces the severity of cutaneous pathology and inflammation in the mouse model of psoriasis (111). Anti-IL-22BP may have therapeutic value for influenza infection, as a pro-IL-22 environment was shown to alleviate pulmonary inflammation in IL-22BP-knockout mice (IL-22RA2−/−) during H1N1 (PR8/34 H1N1) infection and protected the lung by promoting tight junction formation (112). It needs to be actively pursued whether IL-22BP has therapeutic value for human IBD, in lieu of IL-22-induced colonic epithelial ER stress and completely attenuated chronic colitis in IL-22-/- mice or by anti-IL-22 neutralization (90). Testing IL-22BP-Fc as a therapeutic venue in this model and in clinical trials should be a logical next step.