Stromal cells in tertiary lymphoid structures: Architects of autoimmunity

The molecular mediators present within the inflammatory microenvironment are able, in certain conditions, to favor the initiation of tertiary lymphoid structure (TLS) development. TLS is organized lymphocyte clusters able to support antigen‐specific immune response in non‐immune organs. Importantly, chronic inflammation does not always result in TLS formation; instead, TLS has been observed to develop specifically in permissive organs, suggesting the presence of tissue‐specific cues that are able to imprint the immune responses and form TLS hubs. Fibroblasts are tissue‐resident cells that define the anatomy and function of a specific tissue. Fibroblast plasticity and specialization in inflammatory conditions have recently been unraveled in both immune and non‐immune organs revealing a critical role for these structural cells in human physiology. Here, we describe the role of fibroblasts in the context of TLS formation and its functional maintenance in the tissue, highlighting their potential role as therapeutic disease targets in TLS‐associated diseases.


| INTRODUC TI ON
Originally described by Louis Picker and Eugene Butcher in 1992 as the third lymphoid compartment of the immune system, 1 ectopic or tertiary lymphoid structures (ELS or TLS) are currently defined as non-capsulated anatomical entities comprised of organized aggregates of immune cells that harbor within non-immunological organs. 2,3 TLS formation is classically induced in an inflammatory microenvironment, either in response to exogenous stimuli or as a reaction of abundant expression of local antigens in the context of autoimmunity, cancer or tissue transplant. 2,[4][5][6][7][8][9] TLS forming at specific sites has been assigned individual identities, and those include the following: the mucosa-associated lymphoid tissues (MALT); lymphocyte-rich clusters forming in the gastrointestinal tracts (GI); the fat-associated lymphoid clusters or FALCs, present in the adipose tissue of the mesenteries; and the nasal and inducible bronchial-associated lymphoid tissues (NALT and iBALT) that encompass the aggregates detected in the respiratory tract from the nasal cavity to the lung parenchyma. 2,10-12 Reliance on antigen exposure has been reported and believed to support the functional role of TLS in disease, as hub for affinity maturation of a humoral response against the specific antigen and, often, autoantigen. 13 While the link between antigen, antigen-specific B cell maturation and autoantibody production has not always been proved in the different conditions, 13 there are enough evidence to support the formation of TLS as response to local antigen displayed on both professional and non-professional antigen-presenting cells, in the context of an environment rich in inflammatory mediators, such as TNF, IL-6, IL-17 and IL-22. 10,[13][14][15] The molecular mediators present within a chronically inflamed microenvironment appear to be necessary but not sufficient to support TLS assembly. In chronic disease settings, TLS formation has been described in "permissive organs" such as the thyroid of patients with Hashimoto thyroiditis, the synovium of patients with Rheumatoid arthritis or the salivary glands of patients with Sjögren's syndrome. 2,13,[15][16][17][18] Intriguingly all of those, except the synovium, are tissues characterized by an epithelial component that appears to play a critical role in particular in the initial release of inflammatory mediators and in the process of local antigen presentation that underpins TLS formation. [19][20][21] Importantly, TLS does not form in all patients and, even within the same tissue, and have been characterized by the detection of different organizational features, suggesting a potential gradient, either in the quantity/quality of the antigen presented or in the intensity of the inflammatory response shaping TLS assembly 9,13,22 that supports the establishment of structures characterized by different degrees of organization. The mechanistic switch that drives, in certain patients, the formation of TLS has not been elucidated, and the presence of biomarkers predictive of TLS development in individuals is still being explored.
It is currently believed that TLS acts as amplifiers of the immune response and acts as hubs for survival and maintenance of pathogenic effector cells within the tissue. 3,4,10,13,14,23 In this context, the presence of TLS has been associated with poor prognosis when detected in chronic autoimmune conditions and with scarce response to lymphocyte-depleting agents. 24,25 At present, the prognostic role of TLS in cancer is debated and appears variable, based on the type of cancer within which they are found and the nature of the clinical stage of the cancer itself. 4,6,26 More recently, TLS formation in solid tumors has been associated in patients with a positive response to treatment with immune checkpoint inhibitors, opening a potential avenue of research aimed at favoring, rather than blocking, TLS development in the context of cancer. 27 There are no indications of whether TLS has a supporting role in the catalysis of disease initiation as well as perpetuation itself.
Early publications reported the formation of TLS in association with the inflammatory process, but not necessarily with the pathogenic process underpinning disease. Indeed, the presence of TLS was described by Kratz et al in the pancreas and kidneys of rat insulin promoter-lymphotoxin (RIP-LT) transgenic mice in the absence of pathology. 28 TLS in these mice was characterized by delineated areas of T and B cells, presence of plasma cells, primary and secondary follicles as well as high endothelial venules (HEVs). These structures possessed the ability to respond to antigen, supporting the process of local B cell affinity maturation but not tissue damage.
Interestingly, the remodeling of the vascular bed observed within the TLS was also shown to be dependent on lymphotoxin (LT), as the ectopic expression of LT in Rag2 −/− mice leads to vascular changes even in the absence of lymphocytes. 28 These findings, using RIP-LT transgenic mice, led to the conclusion that LT, a molecule well known for its role in secondary lymphoid organ development, was able to imprint the stromal compartment of a non-lymphoid organ to acquire morphologic and functional features of a lymph node.
Fibroblasts are the most predominant non-hematopoietic stromal cells, primarily functioning as producers of the extracellular matrix (ECM) that shapes tissue anatomy. Traditionally considered only for their plastic and architectural properties, fibroblasts have been most recently understood to play a functional role in homeostasis and disease, including supporting some of the acquired immunological functions observed in TLS. 2,15,20,29,30 Originally grouped coarsely, fibroblasts have been more recently recognized as a largely heterogeneous population defined by uniquely assigned phenotypes and functions. The use of multi-omics, followed by in vivo validation studies, has enabled several groups with the ability to demonstrate fibroblasts diversity and plasticity in different organs and in response of different conditions, unveiling the key role of fibroblasts in the process of immune surveillance, inflammation and repair. 15,19,29 In order to capture the granular landscape of the role of this compartment in organ disease, the shared and/or exclusive features of fibroblasts in different organs are currently under investigation.
Our group and others have attempted to establish the specific role of fibroblasts in the context of TLS, with the aim of dissecting the contribution of this population, not only in the establishment and maintenance of TLS in the tissue, but also in supporting TLS pathogenic functions. We are going to review some of those data in the current review, mapping our work in the broader effort made by the community in trying to understand the role of fibroblasts in health and disease.

| PL A S TI CIT Y AND S PECIALIZ ATI ON: CRITIC AL FIB ROB L A S T FUN C TI ON S IN SUPP ORT OF S LOS AND TL S FORMATI ON
The molecular mechanisms underpinning the formation and maintenance of TLS within a specific tissue are not completely understood.
Many efforts over the years have aimed to overlay TLS and SLO developmental factors. However, it now appears that TLS aggregation is the result of a very different process from SLO assembly. While some similarities are shared, TLS formation involves a sequence of events and signaling cascades critically different from those regulating the development of lymph node (LN), spleen, or Peyer's Patches (PP). 3,15,[31][32][33][34][35][36] Mature SLOs play numerous functions in homeostatic and disease conditions, mainly acting as immunological filters, providing an adequate microenvironment to facilitate interaction between naïve cells in search for their cognate receptor antigen. 32,33,37 In this context, lymph node (LN) provides surveillance to the lymphatic system, while spleen filters the blood for bloodborne antigens. 38 While providing a hub for maturation and proliferation of autologous, antigen-experienced immune cells, SLOs support the screening of autoreactive clones escaped from central tolerance. Key structural anatomical differences among different SLOs support different functions. LN are tightly organized with a system of lymphatic vessels and canaliculi for antigen movement from the outer to the inner part of the organ. The spleen presents a less organized structure, with lack of a capsule defining the boundary between the red and white pulp (respectively, inhabited by macrophages and lymphocytes) and the absence of antigen-delivering, afferent lymphatic vessels. 38 to support the process of B cell affinity maturation. 43,44 Similar occurrence has been described in the spleen. 37,[45][46][47][48] Even during development, SLO stroma presents a large degree of plasticity. Early anlagen mesenchyme has the ability to differentiate, upon specific stimuli into diverse and highly specialized stroma, which then create functional micro domains within the lymphoid organs. The efficacy of the immune responses is highly dependent on this specialization of the SLO resident mesenchyme into these micro domains, which support the formation of diverse anatomical and functional areas that provide different lymphocyte survival and developmental needs. 32,[49][50][51][52] Loss of SLO compartmentalization has been demonstrated detrimental for the immune response. Similarly, the adaptability of the stroma during the immune response has been deemed critical to enable expansion of the B follicle required to accommodate the germinal center reaction upon immunization.
Specialization and plasticity of the fibroblasts in SLOs can therefore be defined as the most critical properties required to shape an efficient immune response.
It has been shown in both humans and mice that TLS lack some of the critical architectural features that secondary lymphoid organs, including the presence of a capsule and of an organized lymphatic and blood vasculature. 2,18 Furthermore, while the development of SLO is genetically programmed during development at fixed anatomical locations, TLS development occurs postnatally, in response to chronic inflammatory cues in non-immunological organs. Importantly, TLS might resolve and disappear upon antigen clearance or persist, in pathophysiological settings augmenting the process of tissue damage and aberrant antibody production. 9,53 Similar to SLOs, TLS functions are largely supported by a specialized network of stromal cells that share some of the specialization and plastic features ascribed to the SLO network. 2 However, differently from SLOs, where the general anatomy of the organ is maintained over time, the anatomy of TLS in its cellular composition is highly variable, 2,8,13,14,54 likely reflecting the different maturation stages of these structures.
We and other attempted to identify the presence of a TLS stromal cell precursor, present in non-immune organs but "preprimed" to support the development of a stromal cell network able to sustain lymphocyte migration, survival and proliferation. It is possible to speculate that the presence of this "preprimed "dormant fibroblast precursor could define the permissiveness of certain tissues to harbor TLS. 15,55,56 The signals regulating the ability to differentiate a TLS stroma precursor in response to microenvironmental cues will be later discussed in this review.
In animal models of TLS, early TLS is characterized by small T cell aggregates either surrounding epithelial or endothelial structures.

| FUN C TIONAL IMPRINTING OF ME S EN CHYME IN S LO VER SUS TL S
The development of the TLS stromal network that ultimately supports lymphocyte organization in the tissue presents key similarities but also critical differences with the development of SLO stroma. The spatial development of TLS has been observed in proximity to vasculature or epithelial structures, often associated with pericytes, smooth muscle cells, and myofibroblast, thus suggesting that the resident vasculature facilitates the extravasation of lymphocytes from the circulation into the inflamed or infected organ and that pericytes play a role in TLS establishment. 2,13 Our group has identified a series of sequential events underpinning the maturation of resident stromal cells of non-immune organs to acquire an immunofibroblast phenotype: Those include priming, expansion, and maturation of the harboring immunofibroblast network.

| Immunofibroblast priming
We and others have demonstrated that fibroblast priming occurs independently from the LTβR signaling cascade during TLS formation.
The secretory cues involved in TLS stroma development encompass a series of cytokines belonging to the tumor necrosis factor (TNF) and interferon (IFN) family, but also IL-13, IL-1 family cytokine, IL-17, and IL-22. 15,[68][69][70][71][72][73] The cellular source of these cytokines and the predominant role in immunofibroblast remodeling varies, depending on the tissue within which the TLS is developing and the stimuli in response to which it forms, revealing a diversity of drivers in different diseases.
The upregulation of intercellular adhesion molecule 1 (ICAM-1) and vascular cell adhesion molecule (VCAM-1) and PDPN defines the key priming phenotypical changes, supporting the acquisition of an "adhesive" phenotype that facilitates the physical interaction of primed fibroblasts with incoming lymphocytes which express complementary integrins. 15,72 In some studies, the transient expression of lymphoid chemokines, likely responsible for the initial wave of lymphocytes infiltration has also been defined within the priming step. 10,69 The cellular source for priming cytokines in context of TLS formation has been actively studied. The role of CD4 + T cell-derived IL-17 in immunofibroblast priming has been described in several models including experimental autoimmune encephalomyelitis (EAE) and iBALT formation. 10,72,73 Additionally, epithelial-derived type 1 IFN has been reported to induce immuno-remodeling of lung fibroblasts upon influenza infection. 69 Our group has recently demonstrated the ability of IL-13, to induce the initial priming of fibroblast in salivary glands during TLS formation. 15 Interestingly, this cytokine, already implicated in tumor immunosurveillance, is produced in response to pathogenic stimuli, at site of TLS formation by epithelial cells and resident ILC2, suggesting the presence of an innate process, conserved across species (data were confirmed in human and mouse setting), whereby sentinel cytokines prime the resident fibroblast to support the immune reaction required to deal locally with an immune (or autoimmune) process.

| TLS maturation
The final step identified in the acquisition of the complete immunofibroblast phenotype and function consists in the maturation of the primed and expanded PDPN + population of fibroblasts.
Immunofibroblast maturation is characterized by the stable expression of lymphoid chemokines CXCL13, CCL19, CCL21, and lymphocyte survival factors such as IL-7 and BAFF. [13][14][15]67 Differently from the other two phases, this last step in fibroblast maturation appears to be fully dependent on both lymphocytes and LTβR signaling. As

| S TROMAL CELL S PECIALIZ ATI ON THROUG H THE LEN S OF NOVEL TECHNOLOG IE S
The emergence of novel transcriptomic technologies has enabled a granular characterization of the stromal cellular compartment in both murine and human organs. lymphoma. [87][88][89] The reactive area of TLS is observed in these cases in close proximity to malignant areas of centrocytes like B cell infiltration. The presence of TLS in MALT formation has been identified as essential pathogenetic step, suggesting that the antigen-driven B cell proliferation process harboring within the salivary gland GC represents the key event in lymphoma development. Intriguingly, those GCs anatomically closely recapitulate the GC observed in tonsil, another highly inflamed microenvironment. Nonetheless, the frequency of malignancy development associated with Sjögren's is much higher than the incidence of tonsil lymphoma, suggesting that the inflammatory and autoimmune process that shapes formation and function of salivary glands TLS is intrinsically different from that supporting GC formation in the tonsils. The identifications of these differences could pave the way to the design of novel therapeutics able to prevent the process of lymphomagenesis occurring during Sjögren's syndrome.

| FIB ROB L A S T ROLE IN THE CONTE X T OF AUTOIMMUNIT Y AND C AN CER
The availability of advanced high-throughput techniques has unveiled novel functions and phenotypes of stromal cells also in non-lymphoid organs, broadening from the spatial arrangement of neighboring structural cells and immune cells to their unknown immunological and nursing functions. Many of the novel functions and subpopulations of fibroblasts have been investigated at a single cell level in murine models and human patient biopsies of inflammation and cancer, with many differences and conserved similarities being brought to light which present potential novel therapeutic avenue s. 2,14,15,20,27,[90][91][92][93] Use of computational biology approaches applied to transcriptomic has provided in the past few years, novel cellular atlases of healthy and disease tissue, and useful gene cassettes able to identify potentially pathogenic stromal cells in the tissue of interest. Given the presence of pathogenic cellular signatures in different diseases, much interest has been given to the definition of "gene cassettes," which are able to identify fibroblasts presenting similar pathogenic features in different diseases. 94 Muhl et al reported the presence of a shared fibroblast cassette in murine organs identified by Pdgfra , Cd34, Col1a1, Col1a2, Col5a1, and Lox1 and a mural cell cassette that included Des, Mcam (Cd146), Tagln, and Notch3. 94 Those markers have been largely used to define distribution and expansion of fibroblast and mural cells in the context of cancer and inflammation both in humans and mice. 29,95 Diseases explored to date include rheumatoid arthritis, and inflammatory bowel disease, in particular ulcerative colitis (UC). 15,29,[96][97][98][99] The stromal compartment in rheumatoid arthritis (RA) is uniquely characterized by individually assigned phenotype and function according to their spatial location within the RA joint. 29 In a study conducted by Croft et al, 2 distinct populations of fibroblasts, FAPα + Thy1 − or Fapα + Thy-1 + , were identified in both species, using single cell analysis. The differing functional effects of the two populations were characterized as either pro-inflammatory or supportive of joint damage. 29 Elegantly, Croft demonstrated that depletion of these different populations defines topographic and functional specialization of the synovium stromal compartment previously not appreciated. 29 The developmental signals driving fibroblast specialization in RA have been elsewhere explored and identified in the Neurogenic locus notch homolog protein 3 (NOTCH3). 100 Brenner and colleagues demonstrated that endothelial cells are able to establish a NOTCH3 gradient, which governs differentiation of the diverse fibroblast subpopulations, suggesting that fibroblast identity within a given microenvironment is governed by its spatial location and temporal interaction with different cell compartments. In contrast to RA, the pSS microenvironment has yet to be characterized at cellular level  response within the context of the cancer has been described. 90 The use of a curated receptor-ligand scRNA database enabled this group to identify the receptor ligand axis of complement component C3 as key molecule involved in the crosstalk between CAFs and cancer macrophages. Pharmacologically interference with this signaling axis in an animal model was able to disrupt the immunosuppressive microenvironment and slow tumor growth in vivo.
The engagement of the C3 and C3aR complement cascade has also been described in the context of the contribution to the priming of immunomodulatory synovial fibroblasts. 106 The dependency of the inflammatory synovial fibroblasts on this signaling cascade was made evident with the observation of reduced tissue priming in the bone marrow chimera models in C3 −/− mice. 106 This suggests that fibroblasts may be similar in phenotype, but unique in their function depending on the disease context from which they are derived. From these studies, it is made evident that the use of computational approaches, matched to single cell transcriptomic, has vastly enabled scientists to achieve a better understanding of the heterogeneous stromal compartment within disease and can be further exploited to resolve the complexity of the fibroblast landscape ( Table 1).
The ability to locate conserved signatures of fibroblast populations across species (human and mouse tissues) has been also sought in order to interrogate fibroblast specification and function in the as synovium, intestine, lung, and salivary glands. 55 Two fibroblast populations were identified as CXCL10 + CCL19 + immune-interacting and SPARC + COL3A1 + vascular-interacting fibroblasts, which were expanded in all inflamed tissues across the four diseases at different anatomical sites. 55 The origin of this population in the tissue has not been fully elucidated. An extensive study by Beuchler et al investigated the presence of conserved fibroblasts across species and multiple organs evaluating commonalities between human and mice in different organs in steady and perturbed states, and two universally distributed fibroblast types were differentiated by the expression of Pi16 or Col15a1. 56 The developmental link between steady state and "inflammation induced" fibroblasts in the various mouse organs was demonstrating that the inflammatory activated subsets could be traced back to arise from the steady state fibroblasts using transcriptomic trajectory analysis. 56  To this end, coupled with in vitro and in vivo assays, deciphering the cellular landscape of TLS-driven disease would allow us to

Characterization of fibroblast subsets in cancer Function
PDPN − CD34 − CD146 + myoCAFs in melanoma • Acta2 high contractile stromal had pericyte associated markers Ng2, CD146, Rgs5 • Also shared the expression of Col1a1, Cola1a2 with the neighboring PDPN+ fibroblasts 91 PDGFRα + IL-6 + CAFs in bladder carcinoma • PDGFRα + express various cytokines and chemokines including CXCL12, IL6, CXCL14, CXCL1, and CXCL2 and were termed inflammatory CAFs • iCAFs had enrichment of cytokine-cytokine receptor interaction pathway • Also had increased proliferation ability as compared to non-inflammatory CAFs 95 PDGFRα − RGS5 + myoCAFs in bladder carcinoma • RGS5 + stromal cells have been characterized to be similar myoCAFs in melanoma 95 Sox9 + developmental CAFs in breast cancer • dCAFs distinguished by the expression of genes related to stem cells Scrg1, Sox9, and Sox10 • Found during early developmental stages of cancer involved in the development and morphogenesis of TME 94

TA B L E 1 (Continued)
characterize the specific stromal subsets that may drive the overall functional heterogeneity seen in different TLS-associated diseases.
Resolving the nature of this heterogeneity would allow us to therapeutically target-specific cellular components of mechanisms enabling manipulation of TLS formation to become advantageous tools in cancer therapy or dissolved in the context of transplant rejection and autoimmunity.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflicts of interest in the production and completion of this manuscript.