The TGF-β family comprises a group of pleiotropic cytokines exerting a broad range of responses on both immune and non-immune cells. The key role of TGF-β in immunity has been established in genetically-engineered mice. This cytokine orchestrates T lymphocyte regulation by limiting effector T cell functions and promoting regulatory T cell development and functions. In humans, genome wide association studies (GWAS) have linked the SMAD3 locus - a key component of the TGF-β pathway - to predisposition to inflammatory bowel disease (IBD) and asthma (1). In addition, observations in patients carrying single gene defects in individual components of the TGF-β family (such as TGF-β1) (2) and their downstream signaling components (such as TGFβ-R1/2 or SMAD2/3) have further underscored the role of TGF-β in the pathogenesis of intestinal inflammation and allergic diseases. In addition to gene defects that directly affect the TGF-β cytokine family signaling, others affect TGF-β responses indirectly by impacting the nuclear transport of signaling components (such as IPO8) (3), or by interfering with cross-regulatory non-canonical pathways (such as ERBIN or JNK1) (4, 5). Here, we provide an overview of the TGF-β signaling pathway ( Figure 1 ) and review Mendelian disorders affecting components of the TGF-β signaling machinery, also known as TGF-β signalopathies (6). These defects commonly result in a complex syndrome combining craniofacial, cardiovascular and skeletal defects, features that have been extensively reviewed elsewhere (6, 7). Here, we intend to focus on features of immune dysregulation which are also associated with these defects and often overlooked and to discuss the possible mechanisms underlying abnormal TGF-β signaling and disease pathology.

The human TGF-β family includes thirty-three genes that encode for homodimeric or heterodimeric secreted cytokines, including TGF-β1, TGF-β2 and TGF-β3, activins, nodals, bone morphogenetic proteins (BMPs), and growth and differentiation factors (GDFs). Identified in 1983 and best characterized (8), TGF-β1 remains the most emblematic member of the family and, as discussed below, the key actor of the family in the immune system. TGF-β1 is synthesized in the rough endoplasmic reticulum in precursor form featuring a large ~250 residue amino-signal peptide, a central pro-domain, called latency associated protein (LAP), and a short ~110 residue carboxy-terminal mature peptide. After initial biogenesis and signal peptide cleavage, two pro-peptides dimerize via intermolecular disulfide bonds and translocate into the Golgi complex, where TGF-β1 is cleaved from LAP by FURIN. The two products remain non-covalently bound and form the small latent complex (SLC) (9). In this conformation, the SLC is inactive as mature TGF-β is hidden and protected deep inside the protein structure (10). The SLC typically associates with either a family of proteins called latent TGF-β binding proteins (LTBPs), forming the large latent complex (LLC) that in turn associates with the extra-cellular matrix (ECM) upon secretion (11, 12), or with membrane proteins GARP (glycoprotein-A repetitions predominant) (13). Several mechanisms regulate the release of active TGF-β, including αV integrins expressed at the cell surface of many cell types and extracellular proteases (9, 14). Mechanistically, it has been suggested that physical traction resulting from interaction between ECM-associated latent TGF-β and cell-surface αV integrins ultimately leads to conformational changes of the latent complexes and to active TGF-β release (15). In addition, this conformational change is thought to allow higher accessibility of the latent complex to extracellular proteases (9). Overall physiological release of active TGF-β likely requires synergic actions of both integrins and proteases (16).

TGF-βs signal via a pair of transmembrane serine/threonine protein kinases known as the type I (TGF-βR1 or ALK5) and type II receptors (TGF-βR2). Binding of one of these cytokines to a TGF-βR2 dimer triggers the recruitment and subsequent transphosphorylation of TGF-βR1 which, in turn, activates the downstream canonical signaling cascade via the receptor-regulated SMADs (R-SMADs). BMP subfamily receptors phosphorylate SMAD1, SMAD5 and SMAD8, while TGF-β subfamily receptors mainly phosphorylate SMAD2 and SMAD3. Upon phosphorylation, R-SMADs dissociate from the receptor and form a heterotrimeric transcriptional complex with SMAD4. Typically, heterotrimers of SMAD2, SMAD3 and SMAD4 are formed, but complexes that contain homodimers of SMAD2 or SMAD3 may form as well and target distinct genes (17). Localization and concentration of SMADs are regulated by anchor for receptor activation (SARA) proteins which recruit non-phosphorylated SMADs to the activated receptors for phosphorylation. Upon receptor activation, phosphorylation of SMADs causes their dissociation from SARA and formation of SMAD2/3–SMAD4 complexes (18). Heteromeric SMAD complexes translocate and accumulate in the nucleus where they bind DNA directly but with low affinity and specificity or indirectly by interacting with other DNA-binding proteins to target specific genes for transcriptional regulation. Regulatory mechanisms are in place to avoid spontaneous transcriptional activation of TGF-β target genes. Thus, in absence of TGF-β stimuli, SKI inhibits the expression of TGF-β target genes by binding Smad Binding Elements (SBEs) complexed with SMAD4 and by recruiting other repressors and histone deacetylases (HDA) (19). SMADs turnover in the nucleus is tightly controlled. Nuclear SMAD levels are kept in check either by regulating protein turnover, before and after translocation into the nucleus, or by regulating the subcellular localization of SMADs. SMAD6 and SMAD7 are inhibitory molecules that suppress receptor and SMAD signaling functions. SMAD7 recruits SMURF2 ubiquitin ligases to TGF-β receptors to promote their ubiquitin-mediated degradation (20, 21). Inhibitory SMADs also associate with R-SMADs, preventing the formation of R-SMAD complexes with SMAD4. In addition, the expression of SMAD6 and SMAD7 induced by TGF-β family members also triggers negative feedback (22). SMAD interactions with distinct karyopherin proteins might also result in differential regulation of nuclear import depending on the SMAD, which may represent a complementary layer of regulation of the dynamics of SMAD-mediated transcription responses. Once in the nucleus, SMAD trimers interact with specific transcriptional factors as well as with regulatory proteins, to initiate transcription of specific target genes. Upon dephosphorylation by the nuclear phosphatase PPM1A, SMAD trimeric complexes disassemble and SMAD proteins are exported out of the nucleus by different exportins, such as RanBP3, thus facilitating termination of TGF-β signals (23, 24).

TGF-β receptors stimulation can also result in the activation of SMAD-independent pathways that are collectively referred to as non-canonical TGF-β signaling pathways (25). They include branches of ERK, JNK, p38 MAPK, NF-κB, PI3K-AKT, JAK-STAT and small GTPase pathways (10, 26, 27). Upon activation, direct phosphorylation of ShcA by TGF-βRI induces its association with Grb2/Sos and sequential Ras, MEK1/2 and ERK1/2 activation (28). Thus, initiation of the ERK pathway also depends on TGF-βRI mediated phosphorylation. In contrast, TGF-βRI kinase activity seems to be dispensable for the activation of other non-SMAD signaling pathways. Recruitment of the E3 ubiquitin ligases TRAF4 and TRAF6 onto the TGF-β receptors leads to intramolecular polyubiquitination of TRAF4 or TRAF6, and consequent recruitment and ubiquitination of the MAPKK kinase TAK1. This post-translational modification does not lead to TAK1 degradation, but instead to MAPK kinases (MKKs) phosphorylation by TAK1 and finally to activation of either JNK or p38 (29, 30). Eventually, TRAF6 can also interact with IKK proteins and trigger NF-κB signaling (31). In a similar way, TRAF6-mediated ubiquitination seems to be central for TGF-β-dependent activation of PI3K and AKT in the mTOR pathway (32). Finally, TGF-β can also regulate other non-SMAD pathways, including Rho-like GTPases. Yet, more work is needed to better elucidate the details of their TGF-β-dependent activation (10, 33, 34). When and how TGF-β stimulation can lead to SMAD-dependent or -independent signals is clearly cell- and context- dependent, but the intricate regulatory mechanisms that orchestrate these events remain to be fully deciphered. It has been shown that SMAD-dependent signaling is associated with clathrin-bound receptors, while non-SMAD signaling pathways originate from lipid-raft caveolar compartments (35), suggesting that the variability in signaling output might depend on functionally different receptor complexes (10). Several kinases, including some members of the MAP kinase family, can also phosphorylate R-SMADs, and thereby regulate their effector functions in a TGF-β-independent manner. As a consequence, R-SMADs activation can result from stimulation by cytokines distinct from TGF-β family members. Moreover, translocation of R-SMADs to the nucleus can induce a delayed activation of non-SMAD pathways via gene-expression regulation, further complicating the study of downstream molecular events (10). Overall, these multiple signaling cross-talks enable a broad diversity of responses and underlie the versatility in terms of dosage, spatiotemporal activity and cell context of this family of cytokines in regulating cellular processes. A range of inherited human diseases results from single-gene variants in several components of this tightly controlled signaling pathway. On one end of the range, these gene defects lead to exclusively developmental defects, while on the other end, they are restricted almost purely to the immune system. Therefore, comparing the phenotypic consequences of pathogenic variants in genes encoding TGF-β receptors or downstream components of the signaling machinery, and in genes encoding members of the family (TGF-β1, TGF-β2, and TGF-β3) enable to differentiate and appreciate the role of those individual cytokines in immune and non-immune cells in humans.

The term “TGF-β signalopathies” defines a number of Mendelian disorders associated with inappropriate activation of TGF-β signaling and ultimately impacting developmental processes, immune responses, or both. A large group of these gene defects result in complex syndromes of variable expressivity encompassing a broad spectrum of cardiovascular, craniofacial, and skeletal features associated or not with impaired immune regulation (6) ( Table 1 ). Those, include, among others, Loeys–Dietz syndrome (LDS) which is caused by heterozygous loss-of-function (LOF) variants in TGFBR1 (MIM: 190181), TGFBR2 (MIM: 190182), TGFB2 (MIM: 190220) or TGFB3 (MIM: 190230), SMAD2 (MIM: 601366) or SMAD3 (MIM: 603109) and Marfan syndrome (MFS) (MIM: 154700) caused by variants in FBN1, encoding fibrillin 1, the main component of extracellular matrix microfibrils that scaffolds latent TGF-β ( Figure 1 ). Clinically, MFS and LDS show overlapping cardiovascular and skeletal manifestations. Yet, in contrast to LDS patients, immune regulation is not affected in Marfan patients, likely because fibrillin 1 is mainly expressed in connective and soft tissues, and muscles, while it is almost absent in sites that are exposed to environmental immune triggers such as lungs and intestine, or in sites of adaptive immune priming such as lymph nodes and spleen. Similarly, Shprintzen-Goldberg syndrome (MIM: 182212), which is caused by heterozygous variants in SKI features all the craniofacial, skeletal and cardiovascular manifestations of LDS but no immune dysregulation, likely due to the expression pattern of SKI. Indeed, expression of this negative regulator of the pathway, at least in vessels, seems to be regionally confined and temporally limited at early stages of development in mice (41). Contrasting with this group of diseases, the more recently identified TGF-β1 deficiency results in a prominent immune phenotype underscoring the central immunoregulatory role of this cytokine.

Patients carrying single gene defects in individual components of the TGF-β signaling machinery are invaluable resources to dissect the relative contribution, signaling cross-talks, specificity and redundancy of each of those components in the orchestration of the immune response. Human TGF-β1 deficiency results in very early onset severe IBD, demonstrating the non-redundant role of TGF-β1 in suppressing intestinal inflammation and supporting the hypothesis that the aberrant TGF-β1/SMAD signaling observed in active lesions of IBD patients participate in disease pathogenesis (88). Increased Smad7 associated with downregulation of the TGF-β signaling, shown to amplify gut inflammatory responses in those patients (89), could be targeted by specific antisense oligonucleotides (mongersen), which however provided conflicting results in phase II/III studies (89–91). Pathogenic monoallelic variants in TGFBR1/2 as well as in SMAD3 predispose to IBD, strengthening the key role of the SMAD3 canonical pathway for maintaining intestinal homeostasis, a role previously supported by GWAS studies in common multifactorial forms of IBD (1) as well as by Smad3 inactivation in mice (53, 54). Strikingly, these gene defects also predispose to allergy, overall suggesting that TGF-β is instrumental in humans to control immune responses to environmental triggers. While TGF-β1 deficiency may benefit from supplementation by this cytokine, the complexity of TGF-β signaling remains a major challenge to define appropriate therapeutic strategies in other defects impairing either chains of the receptor or downstream elements of the signaling cascade.

MR, NC-B and MP wrote the manuscript. MR prepared the tables and the figure with BioRender. All authors contributed to the article and approved the submitted version.