Tertiary Lymphoid Structures: Diversity in Their Development, Composition, and Role

Lymph nodes (LNs) are complex, strategically positioned organs that promote the immune response against pathogens by ensuring the encounter of naive B and T lymphocytes with APCs and with each other. In LNs, stromal cells can be broadly subdivided into fibroblastic reticular cells (FRCs)—which include T cell area reticular cells, marginal reticular cells (MRCs), and B cell area follicular dendritic cells (FDCs)—lymphatic endothelial cells (LECs), and blood endothelial cells (1). These populations make up the LN matrix and are organized in specific niches, expressing characteristic cytokines, chemokines, and adhesion molecules that permit survival, migration, and optimal positioning of immune cells (1, 2). Under conditions of prolonged inflammation, tissue stromal cells acquire LN-like properties and drive the formation of hematopoietic aggregates, named tertiary lymphoid structures (TLSs). TLSs are continuously dependent on inflammatory stimuli, disappearing when inflammation is resolved (3–5). TLSs constitute immune functional sites, present in different tissues, with variable degrees of organization. The process of their formation is termed lymphoid neogenesis (6, 7). Consequently, TLSs can be observed at inflammatory sites that can develop in response to infection, autoimmune responses, but also within tumor microenvironments (8, 9). Although these structures can have therapeutic and biomarker potential in these different contexts, the signals that regulate TLSs formation and the cells that promote and sustain the immune response within these structures remain fairly uncharacterized. A better understanding of what TLSs are and how they are formed will be absolutely crucial to modulate their presence or role in clinical settings. In this article, we briefly review what is currently known about TLSs development and function, with a focus on stromal cells in relation to recent findings regarding the role of LN stromal cells (LNSCs) in LN ontogeny and particularly in LN tolerogenic functions.

During chronic inflammation but also within tumors, the presence of (adaptive) immune cells within clusters have been noted. Often, these structures are organized, exhibiting similarities with secondary lymphoid organs (SLOs), hence their name tertiary lymphoid organs or structures. But when do we call grouped immune cells TLS? TLSs are defined as aggregates of lymphocytes and myeloid cells within inflamed tissues that resemble SLOs (10). They constitute nonencapsulated structures with a variable degree of organization that can range from simple B/T cells clusters with rudimentary segregation to more complex, mature structures that possess high endothelial venules (HEVs), B cell follicles with FDCs, and germinal centers (GC) and occasionally lymphatic vessels (6, 11–13). On top of this broad definition, TLSs structure, function, and formation are likely modulated by both location and specific inflammatory stimuli and will therefore differ between different tissues and among various disorders. Nonetheless, tissue fibroblasts acquire common LN-like characteristics in TLS niches, expressing both chemokines and adhesion molecules that permit the recruitment and retention, respectively, of hematopoietic cells (13–18).

Recently, several single-cell RNA sequencing studies revealed several new LN subsets within the already described FRC, LEC, and blood endothelial cell subsets. These newly defined subsets are present within the LNs at specific locations, thereby forming specific niches, and are likely to have distinct functions (19–24). Importantly, these niches are at least partially established by anatomic-specific cues and influenced by immunization (21–24). Because stromal cells in TLSs harbor LN-like characteristics, it becomes important to address which specific lymphoid subsets, if any, are present within TLSs, and whether inflamed fibroblasts from different origins have similar characteristics. In this respect, it was recently reported that LN-like fibroblasts in TLSs have some degree of heterogeneity, with local CD34⁺podoplanin (Pdpn)⁺ and CD34⁻Pdpn⁺ stromal subsets, each producing their own set of cytokines (16). Moreover, both local fibroblasts and epithelial cell can express different chemokines (17). This indicates that niche specificity can also occur in TLSs.

Another important question derived from single-cell RNA sequencing studies is the contribution of steady-state tissue fibroblasts to lymphoid neogenesis. For instance, synovial fibroblasts encompass several cellular subsets, displaying functional heterogeneity (25–27) and some degree of resemblance to LN populations, as recently reviewed (28). Simultaneously, various studies in other organs revealed specific fibroblast populations that are present as a consequence of the ongoing inflammation or the presence of a tumor (29–36). In line with this, it would be interesting to address in future studies how specific tissue fibroblast populations respond to inflammatory stimuli and contribute to TLSs formation and maturity.

From a functional perspective, TLSs can be seen as the evolutionary predecessors of SLOs, providing specific sites to mount an immune response against invading pathogens. Accordingly, mice that lack SLOs are able to generate efficient immune responses against influenza in induced BALT (iBALT), albeit these adaptive immune responses are delayed compared with wild-type controls (3). Furthermore, these TLSs can coordinate memory formation and Ag-specific B cell responses (37). In the context of human disease, TLSs are generally perceived as immunogenic structures that can be present in transplant allografts, sites of chronic inflammation in the context of autoimmunity and within tumors (8, 9, 38–40) Consequently, the beneficial or harmful role of the presence of TLSs is largely dependent on disease context (40).

Lymphoid structures (LSs) are specified by stromal cells that drive accumulation of lymphocytes and myeloid cells to appropriate regions because of their synthesis of chemokines, cytokines, and adhesion molecules and ultimately function as sites of Ag encounter. LSs can be divided in three different categories based on their developmental requirements: 1) SLOs are embryonically programmed structures whose development is independent of exogenous signals. 2) Solitary intestinal lymphoid tissues are programmed structures that form after birth as cryptopatches (CPs). CPs are clusters of lymphoid tissue inducer (LTi) cells at designated areas, that—unlike LNs—require ongoing microbiota stimuli for their further development into isolated lymphoid follicles (ILFs) within the small intestine, although this occurs independently of microbiota in the colon (41–43). These developmental specifics classify them as partially programmed and partially inducible. And 3) TLSs are formed in tissues in response to inflammation or infection, requiring a continuous inflamed environment for their maintenance. Yet, the presence of inflammatory stimuli does not necessarily drive TLSs formation, and hence, they can be considered inducible but not programmed. Of note, other LSs exist throughout the body, for example, fat-associated lymphoid clusters and nasal associated lymphoid tissue, which fall into these categories as previously reviewed (44) and are not described in this review.

LSs can be induced ectopically by the intradermal injection of newborn LN–derived cells into naive adult mice, forming clusters of naive lymphocytes that lacked defined B and T cell areas (45). When injected into neonatal mice, these structures acquired an organization that resembled that of adult LNs. This study suggested that precursors to LNSCs subsets are present at birth within developing LNs and their differentiation toward the diverse subsets requires signals that are underrepresented in adult mice. Nevertheless, an increased degree of organization could be achieved in adult mice upon inflammatory stimuli, leading to the formation of structures containing functional HEVs and B cell follicles with FDCs, indicating that shared or analogous developmental clues take place. In this review, SLOs and ILFs can be considered learning platforms to understand TLSs development.

The specific requirements for the formation of the different SLOs, being LNs at different locations and Peyer’s patches, differ slightly and are reviewed elsewhere (44, 46). In general, for all SLOs, interactions of LTi cells with mesenchymal cells and/or LECs are a requirement for the formation of the first structure, which will then develop into an organized structure upon the influx of B and T lymphocytes. The very first clustering depends on the attraction of LTi cells via chemokines, in most occasions CXCL13 (47), allowing the definitive formation of the structures when signaling via lymphotoxin (LT), expressed on LTi cells, and LT-β receptor (LTβR) on stromal cells. This signaling results in the retention of these first LTi cells, more LTβR signaling followed by the attraction and retention of more LTi cells. Subsequent development of lymphatic vessels allows LTi cell drainage to the developing LNs and the formation of an LN capsule and collecting lymphatic vessels leaving the LN (48). This definitive structure will become organized when B and T lymphocytes start to enter, which occurs in mice after birth and in human during fetal life (49, 50). This larger influx of cells from the bloodstream coincides with the appearance of HEVs. Formation of both HEVs as well as FDCs within the developing B cell follicles is also dependent on signaling via LTβR and reviewed elsewhere (51–53).

ILFs are small LSs within the lamina propria of the intestine, where B cells aggregate, surrounded by Rorγ⁺IL7R⁺ and CD11c⁺ cells but lack defined T cell areas (41, 54). For development of these structures, LTβR⁺ stromal cells are initially activated by LTα₁β₂ expressing Rorγ⁺ LTi cells, promoting synthesis of CCL19, CXCL13, and VCAM-1 for recruitment of dendritic cells (DCs) and B cells (41). Cellular aggregates are first established at the base of the crypts in the intestine as CPs, where microbiota stimuli further promote stromal activation (41). In parallel, LTα₁β₂ expression on B cells promotes ILF maturation (55, 56), and mature ILFs possess FDCs, increased T cell numbers, and support GC reactions because of constant antigenic challenge—likely promoted by their inclusion of follicle-associated epithelium M cells (54, 55). Here, monocyte-derived DCs and gut stromal cells produce TGF-β1, APRIL, and BAFF, sustaining IgA production that is important for maintenance of gut homeostasis (41). In contrast to SLOs, ILFs formation starts in the early postnatal period in mice. Establishment of small intestine ILFs is dependent on gut microbiota because they are absent in germ-free mice and reduced upon antibiotics treatment (42, 55, 57). In agreement with the need for microbial encounter, epithelial cells and pattern recognition receptors have a prominent role in ILFs development, with continued TLR/MyD88 signaling being required for maturation and maintenance of the initial structures (42). In fact, recognition of peptidoglycans by NOD1 in epithelial cells prompts ILF formation by driving synthesis of the CCR6 ligand, CCL20 (42). In line with this, CCR6^−/− mice exhibit deficient ILF development (42). Full transition of CPs into ILFs is dependent on B cell migration requiring CXCL13 production and CXCR5⁺ expression (41, 58, 59). CCL20 also contributes to recruitment of CCR6⁺ B cells (60), suggesting that microbiota stimuli trigger specific pathways that complement the organizing function of CXCL13. In addition, in contrast to SLO ontogenesis, in ILFs not only stromal cells can be a source of this important chemokine but also recruited CD11c⁺ cells (58). Colonic ILFs have, at least to some extent, different developmental requirements compared with their small intestine counterparts (61). Most strikingly, colonic ILFs develop independent of microbiota stimuli and CXCL13 production (43, 61), which indicates that tissue-specific requirements modulate ILFs ontogeny. It is likely that this is also the case for TLS formation and function.

These data indicate that LTβR signaling and expression of CXCL13 are crucial for development of programmed LSs, both prenatal and inducible. Yet, specific cells and chemokines are required to complement this axis for ILF formation, wherein both epithelial cells and recruited myeloid cells can assume chemoattractant functions.

In the context of lymphoid tissue development, inflammation appears to not only promote lymphoid neogenesis, but additionally change its baseline requirements. In dextran sulfate sodium–induced colitis, TLSs are likely formed de novo, independently of existing ILFs (62). Overexpression of TNF-α in Rorγ-deficient mice generated TLSs in the ileum and compensated for the absence of LTi cells (63). In the colon, LTα, but not IL-22, is needed for ILF development in steady-state. However, IL-22 was shown to act downstream of LTα, to support ILFs maintenance during infection with Citrobacter rodentium and treatment of infected LT-deficient mice with IL-22 could rescue ILFs organization (64). As a consequence, TLSs can be generated by signals distinct from the inductive factors involved in the formation of other LSs. For instance, TLSs can be formed independently of LTi cells because they are present in Rorγ- and Id2-deficient mice (65–67). Based on a model previously established by Buckley et al. (4), we have subdivided TLSs formation in three main steps: 1) local fibroblast activation, 2) recruitment of immune cells, and 3) maturation (Fig. 1).

Priming of tissue fibroblasts has been recognized as a crucial step in TLSs genesis, and the signals that drive fibroblast activation and further TLS maturation have been extensively reviewed elsewhere (5). Under inflammation, tissue fibroblasts acquire lymphoid-stroma characteristics, such as expression of Pdpn, CCL19, IL-7, CXCL13, and adhesion molecules ICAM-1 and VCAM-1, creating a chemokine niche that attracts and organizes hematopoietic cells with the potential to retain them in place as well (14, 15, 17, 18, 62, 68). Recently, Nayar et al. (16) used a mouse model in which diphtheria toxin receptor expression under the FAP promoter was used to delete Pdpn⁺ fibroblasts. Elimination of this LN-like population impaired TLSs establishment and organization upon Adv5 infection in salivary glands, confirming a crucial role for reprogramming of local fibroblasts in TLSs formation. Priming of tissue fibroblasts can occur via different routes, with, for instance, IL-13 or IL-17, indicated as cytokines that can remodel local fibroblasts in distinct infectious and autoimmune disease mouse models (15, 16, 66, 68). Henceforth, these initially required signals are likely to vary per location as well as per inflammatory stimuli. For example, IL-17 is essential for lymphoid neogenesis of iBALT in response to LPS induced inflammation but seems dispensable for iBALT formation in response to modified vaccinia virus Ankara or Pseudomonas aeruginosa (66, 69, 70). Consequently, the cells that provide activation signals for local fibroblasts also differ with context, as this function has been attributed to myeloid cells, CD4⁺ T cells as well as ILCs (14–16, 68). The signals responsible for fibroblast priming are not necessarily mediating the subsequent expansion of primed stromal cells or maintenance of TLSs structure (15, 16). In this context, the inflammatory environment needs to provide priming and expansion signals, with crucial receptors being upregulated on primed fibroblasts that permit their expansion (16). Notably, IL-22 has been shown to lead to fibroblast expansion and to accompany TLSs formation in different models, which indicates a prominent role for this cytokine. However, it is likely that IL-22 can be provided by different cells (15–17, 71). The expansion of an LN-like stroma matrix is then likely to create a sufficiently strong chemokine niche to attract and cluster hematopoietic cells. Importantly, cytokines can also compensate for each other, and it is possible that to some extent, inflammation creates a redundant environment that promotes and sustains TLSs formation. For instance, both IL-17 and IL-22 can drive fibroblast priming in experimental autoimmune encephalomyelitis mice (15). Moreover, tissue fibroblasts can also maintain an LN-like profile upon deletion of different inflammatory receptors (14). Overall, this suggests that priming of local fibroblasts by different hematopoietic cell types and cytokines can result in the production of chemokines necessary to start TLS formation.

The chemokines that are necessary to attract lymphocytes can be produced by stromal cells, but also by hematopoietic cells, thereby forming the organizing center of the TLS. A similar functional overlap has also been described for ILFs, in which CD11c⁺ DCs acquire a chemoattractant function, producing CXCL13, CXCL12, and CCL19 (58). Moreover, antigenic stimulation plays an important role in TLSs immune function, in which both myeloid and B cells can act as Ag presenting cells in these structures (4, 72, 73). The importance of Ag presentation for sustaining TLSs formation was recently demonstrated in mice infected with Salmonella enterica, in which Ag presentation by CXCL13⁺CX3CR1⁺ resident macrophages recruited and activated B and T cells at the sites of infection. This contributed to TLS genesis and local GC response (74). These macrophages were detected within TLSs, suggesting that these cells are instructed by the TLS microenvironment to recruit other hematopoietic cells. Nevertheless, deletion of CXCL13⁺ CD11c⁺ DCs impaired GC reactions in iBALT, although these cells were not acting as APCs (75, 76). Therefore, both acquisition of chemoattractant properties and Ag presentation can be essential to drive TLSs formation, but those functions do not necessarily overlap. Interestingly, CD8⁺ T cells and other hematopoietic cells have also been described to express CXCL13 in TLSs, which is instrumental for TLS maturation (74, 77–80). These data may indicate that production of chemokines by lymphoid tissue–like fibroblasts is not always sufficient to drive TLS neogenesis, and more than one hematopoietic population can be recruited to boost this process, probably depending on specific inflammatory stimuli. Furthermore, these studies highlight the importance of CXCL13 for the development of LSs (81). Herewith, fibroblast priming is followed by a more apt chemokine niche where Ag presentation can take place.

In SLOs, LTβR signaling is required for the formation of primordial structures but also for their further maturation, as HEV and FDC formation require continued LTβR mediated signaling (51, 52). In TLSs, LTβR stimulation does not seem to be always required for TLSs neogenesis, although LTβR expression is readily upregulated in LN-like cells in inflamed tissues and downstream signaling directly induces lymphoid neogenesis in Apolipoprotein E^−/− mice (3, 13, 14, 18, 62). Similarly, in dextran sulfate sodium–induced colitis, LTα₁β₂ signaling has been shown to be crucial for TLS formation (65). However, more recent studies indicate that fibroblast priming, and initial recruitment of hematopoietic cells can also occur independent of LTβR signaling, but these structures became disorganized in its absence and the LN-like stroma profile was lost with time (15, 16). These data indicate that LTβR signaling may be required for maintenance of LN-like fibroblasts in the tissue, if dispensable for the very initial activation, although LTβR signaling can also have a priming function. It is possible that LTβR signaling also induces or sustains CXCL13 expression as described for the primordial LN (47, 82).

For further TLS maturation, HEV have to form to allow the entrance of more lymphocytes from the bloodstream. Indirectly, this concept implies that initial lymphocyte cluster formation is dependent on the grouping of activated B and T cells, potentially together with myeloid cells, whereas naive lymphocytes will only enter TLSs in later stages when HEVs are formed. HEVs within TLSs express peripheral LN vascular addressin and CCL21, thereby allowing the entry of naive T cells as they express the peripheral LN vascular addressin ligand CD62L, as well as CCR7, the chemokine receptor for CCL21 and CCL19. Similar to LNs, HEVs formation seems to be dependent on LTα₁β₂ signaling, in which LTβR is to some extent required (12, 13, 53, 83, 84). However, in mouse models of melanoma and lung carcinoma, HEV formation was independent of LTβR signaling and required LTα and TNFR interactions, likely promoted by infiltrating CD8⁺ T cells and NK cells (85). Intriguingly, in a therapeutic tumor model, in which T-bet–expressing DCs drive TLSs neogenesis, HEV formation was independent of both LTα and LTBR, and seemed to partially rely on IL-36γ (86). Because depletion of T regulatory cells (Tregs) and consequent activation of T cells can lead to formation of HEVs within tumors with no association to activated stroma (87), it is possible that these specific requirements for TLS formation are a consequence of the tumor microenvironment and not of stromal driven lymphoid neogenesis.

A final marker of TLSs maturation is the formation of FDCs within the B cell follicles, which requires LTβR and TNFR1 signaling, in SLOs. In the spleen, MRCs are believed to serve as precursors for FDCs (52). Similarly, within LNs, MRCs have also been reported to act as precursors for FDCs, but other progenitors may exist (88). LTα₁β₂ is important for FDC generation within TLSs, allowing GC reaction and Ag presentation (73, 89, 90). Although TNFR1^−/− mice do not develop FDCs, these cells could be found in TLS-like structures after intradermal injection of newborn LNs from wild-type mice, indicating the presence of FDC precursors within LNs at day of birth (45). Therefore, although FDCs progenitors in TLSs remain unknown and MRCs are not present within these structures, these data indicate that activated local stromal cells differentiate into FDCs upon encounter with migrating immune cells (45).

Overall, similar to the formation of programmed SLOs and ILFs, TLSs development is a multistep process in which immune cells derived proinflammatory signals can be perceived as inducers and activated fibroblasts establish and sustain a lymphoid niche, thereby acting as organizers. However, in TLSs, recruited immune cells can also acquire organizing roles by producing crucial chemokines, which can be seen as re-enforcement of primed fibroblasts functions. This initial cluster of immune cells then permits further maturation of TLSs by also enabling local Ag presentation. In this regard, TLS developmental requirements are rather diverse as a consequence of their heterogeneous nature and this is likely influenced by restricted activating stimuli, tissue permissiveness and Ag specificity. Perhaps unsurprisingly, in contrast to fibroblast priming, the signals that drive TLSs maturation show an increased overlap with SLO ontogeny, particularly in the need for CXCL13 and LTα₁β₂, suggesting the presence of evolutionarily conserved pathways upon initial activation.

TLSs are present at several inflammatory sites, including autoimmune diseases such as Hashimoto thyroiditis, diabetes, and multiple sclerosis (8). In Hashimoto thyroiditis, mature TLSs are detected in the majority of the patients, and these structures seem to be active centers of autoantibody production and autoimmune response (6, 91). In rheumatoid arthritis (RA) TLSs are seen at varying degrees of maturation and only present in a small percentage of patients (6, 11, 92, 93). As pointedly reviewed before (38, 72), TLSs in rheumatic diseases, including RA and Sjogren syndrome, can display GC characteristics, including expression of activation-induced cytidine deaminase (AID), and promote B cell maturation in the inflamed tissues. Hereto, TLSs are likely to initiate or support production of self-reactive Abs and B cell survival (94, 95). Yet, the clinical relevance of GCs and TLSs detected in patients remains controversial, and TLSs do not seem to provide a stable biomarker for disease progression. In RA patients, the presence of lymphocyte aggregates in the synovium of untreated patients has been associated with clinical responsiveness to infliximab treatment, suggesting that TLSs may associate with therapeutic responsiveness (96). More recently, the presence of lymphoid aggregates associated genes was also associated to therapy responsiveness (97). However, this information requires further confirmation as the presence of B cell clusters has also been described to be increased in patients that were resistant to anti-TNF blockade compared with early, untreated RA patients and may in these patients thus be an indicator of a failure to respond to therapy (98) Notably, two large studies in RA patients point out that the presence of B cell rich clusters associated with higher histologic scores and disease activity (97, 98). In contrast, TLSs are usually considered a favorable prognostic marker in several cancers, contributing to antitumor immune responses and are associated with therapeutic response and lower recurrence (9, 99–104). These TLSs within the tumors are sites of Ag presentation and GC reaction, promoting effector T cell responses, clonal expansion, and Ab production, highlighting an important role for B cells in driving antitumor responses (9, 99–101). Henceforth, promoting TLSs formation in the cancer microenvironment seems to have potential as a therapeutic target, for instance by delivery of activated stromal cells to the tumor microenvironment (9, 105). Considering the prominent role of IL-13, IL-22, and IL-17 in lymphoid neogenesis, it would be interesting to know whether these cytokines can also have a role in TLS formation in tumors or whether tumor TLSs have different developmental requirements.

TLSs can also harbor Tregs in both tumor and autoimmune microenvironments, thereby providing potential tolerogenic properties to TLS (13, 106, 107). Similarly, in the context of solid organ transplantation, the presence of TLSs has been associated with both rejection in some studies, whereas in other studies, a correlation with acceptance of allografts was reported (40, 108). More studies are required to understand what distinguishes an immunogenic TLSs from more protective structures. In particular, it will be important to determine whether such difference is controlled by infiltrating cells or established by the stromal microenvironment. Suggesting a potential role for stromal cells, FRCs in the cancer microenvironment can directly present Ags and modulate T cell activation (35). In SLOs, LNSCs possess several tolerogenic functions in steady-state and inflammatory conditions that account for both synthesis of specific factors (e.g., NO) and expression of surface receptors, such as PDL-1 and MHC class II (109–113). In a mechanism analogous to T cell selection in the thymus, LNSCs synthesize several tissue-restricted Ags (TRAs) and can present them via MHC class I or MHC class II (114–120). Consequently, autoreactive CD8⁺ T cells that escaped negative selection in the thymus can be eliminated in the LNs, whereas Ag-specific naive CD4⁺ T cells can be converted into Tregs (114–116, 119). Also, in the thymus, fibroblasts have recently been described as a source of self-antigens, expression of which was dependent LTβR signaling (121). However, whether stromal cells in TLSs possess similar mechanisms has yet to be determined. It remains to be seen how self-antigen expression by stromal cells in secondary and potentially in TLSs is controlled. It has been shown that LNSCs stimulation with TLR3 ligand Poly:IC shifted the patterns of TRA expression, thereby reducing the capacity to induce CD8 tolerance to a transgenically expressed self-antigen (120). Therefore, presentation of TRAs may not occur under inflammatory stimuli, and it is possible that other tolerogenic functions are also affected. Accordingly, TLSs stroma may not be intrinsically devoid of tolerogenic functions but, instead, suppressed by the ongoing inflammation as described for the LN (120). It is tempting to speculate that a balance between the strength of tolerogenic inducing signals, potentially LTβR signaling, and inflammation inducing signals (i.e., via TLR) modulates a tolerogenic versus immunogenic profile of fibroblasts within TLSs. Transgenic expression of CCL21 under Ins2 promoter in NOD mice resulted in pancreatic TLSs with a higher proportion of Tregs compared with nontransgenic NOD mice and prevented autoimmune diabetes. These observations indicated that tolerogenic LSs can ectopically develop. Hereto, LN-like fibroblasts displayed increased expression of pancreatic self-antigens and LT signaling–associated genes, whereas the production of proinflammatory cytokines such as IL-6 and TNF was decreased (122). Therefore, addressing the specific properties of TLS stromal component and understanding their involvement in ongoing diseases can elucidate more on TLSs function and heterogeneity and potentially also contribute to therapeutic improvement.

TLSs are transient structures whose development is likely dependent on both site-specific and inflammatory context. Importantly, TLSs have potential as both biomarkers and therapeutic targets in autoimmune disease and cancer. In the future, it is possible that this potential will be further reinforced or clarified by newer analysis technologies and better sampling methods. Nevertheless, more studies are required to characterize the LN-like population that form TLSs matrix; particularly, in light of all recent stromal populations described for LNs, the specific characteristics of the stromal populations that make these structures across different tissues remains fairly undescribed. In this regard, lymphoid neogenesis can be induced by diverse initial signals, and it will therefore be important to pinpoint the signals that drive fibroblast priming in different diseases. It will be particularly relevant to know whether inflammatory cytokines, described to promote TLSs formation in infectious and autoimmune models, are also capable of having a similar function in cancer microenvironments. Finally, understanding how tolerogenic properties are induced and maintained in LNs may lead to direct modulation of TLSs-associated stroma instead of inducing or blocking the structure completely.

The authors have no financial conflicts of interest.