Tertiary lymphoid structures (TLS), also known as tertiary lymphoid organs, ectopic lymphoid neogenesis, or ectopic lymphoid tissues, are organized clusters of immune cells that form in non-lymphoid tissues after birth. TLS are neither stable structures nor present in embryonic life, but instead can be induced by chronic inflammatory stimuli and tissue injury (1, 2) in autoimmune diseases such as Sjogren syndrome, rheumatoid arthritis, systemic lupus erythematosus (SLE), and myositis (3–5), allograft rejection (2, 6), chronic infections (5), and cancers (7–9).

Impaired tolerance to self-antigens via various mechanisms marks the early step in the pathogenesis of autoimmune diseases (10, 11), resulting in the initial activation of autoreactive cells. These activated autoimmune cells may then infiltrate the target organs, where the persistent exposure to antigens in chronically inflamed environments keeps them activated. Such chronically activated immune cells can serve as a substitute for lymphoid tissue inducer (LTi) cells and initiate the formation of TLS in tissues. TLS may execute tissue-specific immune responses, thereby providing communications between immune cells and local resident cells (12). This process typically aids in the clearance or neutralization of pathogens through the local induction of plasma cells that generate specific antibodies (13). TLS often resolve after successful antigen clearance or upon resolution of inflammation. However, TLS resolution might not happen in the context of persistent antigen presentation in autoimmune-prone backgrounds, thereby enabling the local induction and expansion of autoreactive T and B cells, which may cause increased autoantibody production and contribute to local pathology as in SLE.

SLE is a highly heterogeneous disease, with different patients exhibiting different manifestations. Kidneys may be involved in up to 50% of adults and up to 80% of children with SLE (14, 15). Kidney disease in SLE, called lupus nephritis (LN), continues to be a major contributor of morbidity and mortality in SLE (16, 17). Current treatments for LN may cause systemic immune suppression with many adverse effects. Hence, understanding local mechanisms of disease in kidneys may open new avenues to develop organ-targeted treatments for LN, thereby avoiding systemic toxicity. In this article, we will review the current understanding of TLS as it pertains to LN.

TLS share similarities with secondary lymphoid organs (SLO) with regard to their structure and function (18–21). Understanding the development of SLO may provide clues to mechanisms underlying TLS formation. SLO consist of lymph nodes, spleen, tonsils, Peyer’s patches, and mucosa-associated lymphoid tissues distributed throughout the body. These structures play a role in initiating and organizing adaptive immune responses by facilitating the interaction between antigens and immune cells. The development and formation of SLO, particularly lymph nodes and Peyer’s patches, involve CD34⁺ hematopoietic stem cells and TNF-related activation-induced cytokine (TRANCE⁺) stromal cells, as well as cytokines, chemokines, adhesion molecules, and specialized vasculature (22). SLO formation, which occurs during embryogenesis or in the first few weeks after birth, involves hematopoietic lymphoid tissue inducer (LTi) cells, a subtype of innate lymphoid cells (CD45⁺CD4⁺CD3⁻) that express RORγt and Id2. The process also involves the interactions of LTi cells with TRANCE and CXC-chemokine ligand 13 (CXCL13) through their receptors, TRANCE-R and CXCR5, respectively (23). Once LTi cells aggregate, their function is sustained by IL-7. This induces the expression of lymphotoxin α1β2 which binds to the lymphotoxin β receptor expressed on mesenchymal lymphoid tissue organizer (LTo) cells. This binding promotes LTo cells to express adhesion molecules, including vascular cell adhesion molecule 1 (VCAM1), intercellular adhesion molecule 1 (ICAM1), mucosal addressin cell-adhesion molecule 1 (MAdCAM1), and peripheral node addressin (PNAd), and to generate lymphoid chemokines, such as CC-chemokine ligand 19 (CCL19), CCL21, and CXCL13. Then, the arrangement of lymphocytes expressing CCR7 and CXCR5 receptors, which contributes to the formation of T cell and B cell zones within SLO, is facilitated by follicular reticular cells that express CCL19 and/or CCL21 and follicular dendritic cells (FDCs) that express CXCL13 (24).

Germinal center (GC) formation in SLO is regulated by antigen-driven interactions between B cells and follicular T helper (Tfh) or follicular T regulatory cells (Tfr). Tfh cells support the GC reaction through their expression of CD40L that engages CD40 on GC B cells, activates the NF-κB pathway, and triggers B cell proliferation. Tfh cells can also upregulate the Tfr–associated transcription factor Foxp3, which dampens Tfh cells’ expression of CD40L and IL-21 and contributes to GC extinction (25). The transcription factor BCL6 is essential to the differentiation of Tfh cells and the formation of GCs. In the absence of T cell–expressed BCL6, B cell responses to protein antigens do not occur and GCs fail to form (26). GCs are crucial for generating high-affinity antibodies against pathogens. GC B cells are positively selected for clonal expansion based on affinity of their cell surface receptors for an antigen – high affinity for an antigen favors differentiation of GC B cells into proliferative plasmablast cells. Some of these plasmablasts subsequently enter the quiescent plasma cells compartment, ultimately resulting in the generation of high-affinity, long-lived plasma cells and memory B cells within the GC (27). Somatic hypermutation has the potential to generate self-reactive immunoglobulins and humoral autoimmunity, whereas negative selection in GC proper entails elimination by apoptosis of GC B cells that acquire self-reactivity through somatic hypermutation (25).

The distinct microenvironmental niches within SLO that harbor different sets of innate and adaptive immune cell populations are formed by the structural scaffold of follicular reticular cells, which also generate conduit networks for cell communication (28). In addition, specialized blood vessels known as high endothelial venules (HEVs) play a role in guiding lymphocytes to SLO. They facilitate the extravasation of lymphocytes from the bloodstream, a necessary step for the initiation of adaptive immune responses in lymphoid tissues (29–31). HEVs are critical components of the immune system that facilitate lymphocyte recirculation and immune surveillance, playing a role in recognizing and responding to pathogens as well as in chronic inflammatory diseases (32).

Patients with SLE exhibit many abnormalities of SLO, including follicular hyperplasia, coagulative necrosis, diverse histological lesions, and increased 18-F-fluoro-2-deoxy-D-glucose (¹⁸F-FDG) uptake (33, 34). A higher ¹⁸F-FDG accumulation was also found in organs with higher immune cell compositions such as the lymph nodes, spleen, and thymus as compared to organs like the brain, liver, and lungs in NZB/NZW F1 mouse model of lupus (35). These observations suggest that immune cells in SLO in humans and animals with lupus are in an active state. Although the exact pathogenic contributions of these abnormalities are not fully understood, SLO abnormalities may contribute to autoimmune pathogenesis in multiple ways including initial breakdown of tolerance and activation of autoreactive T and B cells (10, 11, 36). For example, tissue-resident dendritic cells that carry local antigens to tissue-draining lymph nodes to induce tolerance do not migrate well to lymph nodes in lupus-prone mice, thus contributing to loss of tolerance and inflammation (37, 38).

After the initial activation in SLO, activated immune cells may migrate to the target organs where the persistent exposure to antigens in chronically inflamed environments may keep immune cells activated. Such activated cells in tissues may serve as primary regulators of local TLS formation, analogous to the role of classical LTi cells in SLO. Activated immune cells produce CXCL13 and IL-7, and recruit LTi cells to the site of inflammation (39) in a positive feedback loop. T helper 17 (Th17) cells (40), B cells (41), and M1-polarized macrophages (42) have been shown to possess the ability to replace LTi cells and initiate TLS formation in various pathological conditions ( Figure 1 ).

On the other hand, stromal cells play the same role in TLS formation as classical LTo in SLO. At sites of inflammation, LTi cells aggregate and release lymphotoxin αβ, which binds to the lymphotoxin β receptor expressed on stromal cells residing in the area (43). The resident stromal cells express the classical LTo cell phenotype and respond to the accumulation of LTi cells by inducing the production of homeostatic chemokines such as CXCL13, CXCL12, CCL21, and CCL19. These chemokines recruit lymphocytes to the site of inflammation and initiate TLS formation (18–20, 44). Upon activation of the lymphotoxin β–lymphotoxin β receptor signaling pathway in stromal cells and surrogate LTi cells, the secretion of several growth factors such as vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF), and hepatocyte growth factor are induced (45). Additionally, adhesion molecules such as VCAM1 and intercellular adhesion molecule 1 (ICAM1) are also secreted (46). VEGF, FGF, and hepatocyte growth factor are critical for promoting the development of HEVs. Additionally, the presence of ICAM and VCAM on stromal cells in the vicinity aids in the recruitment of T and B cells from nearby HEVs and promotes the organization of these cells into TLS (47).

Following the organization of TLS by cytokines/chemokines, some local stromal cells differentiate into FDCs upon stimulation with lymphotoxin α. FDCs play a role in supporting the development and maintenance of TLS by producing chemokines and providing a cellular network for B cell migration (13). A subset of CD4⁺ T cells polarizes into T follicular helper (Tfh, CXCR5^hiPD-1^hiICOS^hi) cells that promote B cell activation and antibody production (48). In addition to CD4⁺ T follicular helper (Tfh) cells, T cell compartments also contain CD8⁺ cytotoxic T cells, CD4⁺ T helper 1 (Th1) cells, and regulatory T cells (Tregs) (49). TLS display similar organization to SLO, where B cells aggregate around FDCs expressing the CD21 antigen, and T cells areas contain dendritic cells expressing the marker MIDC-8 (50). Subsequently, TLS facilitate the process of affinity maturation and differentiation of naïve B cells into GC B cells and plasma cells that produce antibodies (51). As in SLO, follicular reticular cells form a dense network to anchor TLS at the site of chronic inflammation (1). At the same time, vasculature comprised of HEVs and lymphatic vessels forms the surrounding compartments to facilitate ingress and egress of immune cells.

Taken together, activated immune cells can function as primary regulators of TLS formation, substituting for LTi cells. Resident stromal cells function as LTo cells to stimulate the production of cytokines and chemokines, creating a positive feedback loop that induces the formation of GCs by lymphoid cells. The development of TLS can be divided into three sequential stages. The first stage, known as the ‘immature’ or ‘early’ TLS, is composed of dense lymphocytic aggregates with mostly T cells and fibroblasts, but no FDCs and without segregated T and B cell zones. The second stage, referred to as the ‘premature’ or ‘primary follicle-like’ TLS, contains FDCs and small B cell follicle but lacks GC reaction. The third stage, considered as the ‘mature’ or ‘secondary follicle-like’ TLS, is characterized by a GC that appears as well-organized aggregates of T and B cells with segregated T and B cell zones, presence of FDC networks, development of HEVs, and specialized lymphoid fibroblasts (49, 52).

TLS formation relies on inducible inflammatory triggers, such as CXCL13 and CCL19, as well as lymphorganogenic cytokines like lymphotoxin αβ. Transcription factor BCL-6, which is essential for Tfh differentiation (52), was overexpressed in TLS in skin lesions (53). Importantly, reducing B cell activating factor (BAFF) decreased TLS formation in an inducible model of LN (54), suggesting a role of BAFF in TLS formation. The processes of B cell selection, activation, affinity maturation, isotype switching, and plasma cell formation can occur in GCs of TLS in a manner similar to that in GCs of SLO (55). Studies in patients and animal models of autoimmune diseases, such as autoimmune diabetes, myasthenia gravis and Wegener’s granulomatosis, have demonstrated the process of selection and affinity maturation of B cells in TLS. For example, granulomas of upper respiratory tract in patients with Wegener’s granulomatosis contain clusters of B lymphocytes, dendritic cells and plasma cells with evidence of initial selection and affinity maturation of B cells, which is potentially driven by self-antigen proteinase-3 (56). B cell selection, activation, affinity maturation, isotypic switching and plasma cell formation have also been demonstrated in TLS associated with tumors, resulting in an “in situ” anti-tumor antibody production (57). Thus, mature TLS with GCs are active sites of generation of autoantibodies and antibodies against tumor antigens. Additionally, molecular and cellular interaction in the TLS promotes the local inflammatory response in the organs targeted by autoimmune diseases, as described below for kidneys in LN.

TLS are found in autoimmune diseases’ target tissues such as kidneys (lupus), salivary glands (Sjögren’s syndrome), and synovial joints (rheumatoid arthritis) (58). TLS have been reported in aged human kidneys as well as in kidneys from patients and animals with various kidney diseases including IgA nephropathy, renal allograft rejection, and pyelonephritis (59, 60). In many conditions, the extent and stage of TLS appear to correlate with disease severity (59, 60). These observations suggest that TLS formation may be a common pathological process, which may progress with disease such as LN.

The kidney infiltrates in patients with LN may range from scattered B and T cells to the fully organized TLS with GCs containing FDCs (67, 68, 77) ( Table 2 ). Among LN kidney biopsies with organized infiltrates, about 10% exhibited TLS with typical features of GCs containing FDCs with B cell clonal expansion and somatic hypermutation (68), about 50% of patients had well-circumscribed T:B cell aggregates, and the remaining patients had dense lymphocytic aggregates without segregated T and B cell zone and without FDCs (68–70). The latter have been suggested to be ‘immature’ or ‘early’ TLS or class I ectopic lymphoid tissue, while others have argued that these organized kidney infiltrates are distinct from TLS (92). Chronic autoimmune responses in these organized lympho-myeloid infiltrates have been suggested to be extrafollicular (92, 93). Animal studies have also suggested that spontaneous or induced responses in TLS can, but do not have to, include GCs (62). In fact, extensive intraclonal somatic V region diversification was observed in a persistent non-canonical extrafollicular response in lupus-prone mice (94). The same autoreactive B cell clonotype proliferating and mutating extrafollicularly and in GCs was found in lupus-prone NZM.2410 mice (95). Furthermore, the mutation rate in autoreactive extrafollicular responses in MRL/lpr mice was similar to that seen in peak GCs (96). Somatic hypermutation and isotype switching were also observed in synovial lymphoid infiltrates without a GC-like structure of rheumatoid arthritis patients (97). In these and other studies in autoimmune diseases’ target tissues, at least some of the microdissected B cell clusters that yielded mutated V region sequences did not resemble GC structures and were most likely an extrafollicular-like response (92).

There is some evidence to suggest that certain infections and immune inflammatory states favor the GC or the extrafollicular response (92). It has been suggested that the extrafollicular response may occur in the context of severe infections with high pathogen burden (92). Corollary to this, the abundant antigenic target of the immune response, such as DNA or DNA-containing immune complexes as in SLE, may trigger an extrafollicular response, whereas the localized antigen, such as insulin as in type I diabetes, may trigger a GC response. The degree of inflammation and cytokine response may also dictate the GC vs. extrafollicular response. It is also likely that autoimmune responses exist along a spectrum from GC- to extrafollicular-dominant, and may even evolve over time with the severity of disease in an individual.

In SLE, B cells in the GC are influenced by the expression of Absent in Melanoma 2 (AIM2). Reduced AIM2 expression can decrease B cells in the GC and alleviate lupus symptoms (98), highlighting the role of GC in lupus disease. However, blocking the GC pathway by knocking out the transcription factor Bcl6 in GC B cells did not protect animals from anti-dsDNA antibody production, plasma cell output, and immune complex deposition in glomeruli in the R848 model (99). Understanding these different autoimmune scenarios might help to stratify patients within a given autoimmune disease (100), which might help guide the selection of targeted therapies.

Persistence of TLS in autoimmune conditions is associated with a poor prognosis and with a failure to respond to B cell-depleting therapies (3, 101). Therefore, identifying mechanisms underlying the formation and persistence of TLS could present new therapeutic options for autoimmunity. For example, treatment of mice with BR3-Fc (soluble BAFF receptor fused to the Fc portion of mouse IgG1) reduced BAFF levels and prevented the formation of TLS in kidneys and development of LN in an inducible model of SLE (54). In recent clinical trials, adding belimumab that blocks BAFF to standard therapy in people with active LN improved renal function, decreased flare rate and allowed glucocorticoid withdrawal (102, 103). It remains to be determined whether belimumab reduced TLS in the kidneys of patients with LN.

In animal studies, several treatments have been shown to reduce TLS and autoimmune disease ( Table 7 ). For example, treating MRL-lpr mice with a blocking antibody against CXCL13 that plays a role in TLS formation improves neuropsychiatric lupus (108). Treatment with a classical pathway complement inhibitor C4BP(β-) or with the selective S1P1 receptor modulator, Cenerimod, also reduced TLS and autoimmune disease in lupus-prone mice (105, 106). Additionally, dexamethasone was highly effective in preventing TLS formation in MR/lpr mice (63). These observations further emphasize the role of TLS in the development and progression of lupus and open avenues for modulating TLS as a treatment option.

TLS can also serve as a prognostic marker, as the presence of TLS in the kidneys of lupus-prone mice and patients with LN (68, 111) is often associated with poor prognosis and increased autoantibody production (3). Furthermore, analysis of TLS densities in LN biopsies, their cell composition and cytokine/chemokine secretions, and the antibody produced by plasma cells educated in TLS might help in predicting therapeutic responses to immune modulating drugs. Further work is needed to determine if the persistence of TLS correlates with a poor response to treatment in LN.

Many questions remain regarding TLS as well as relative contributions of TLS among the cascade of pathogenic events that lead to LN. TLS studies thus far have used variable evaluation strategies. A comprehensive analysis of TLS phenotypes, density and spatial distribution using a standardized, reproducible, quantitative approach is needed. For example, a study in human melanoma used 7-color multiplex immunostaining of whole tissue sections from 103 human melanoma samples to characterize TLS phenotypes along the expression of established TLS-defining molecular and cellular components (112). Such integrated qualitative, quantitative, and spatial analysis of TLS at different stages of LN using a large human sample size can inform the relative contributions of TLS in LN.

The lack of non-invasive methods to visualize/quantify TLS in kidneys is a major hurdle in using TLS in the clinical assessment. However, Dorraji and colleagues were able to visualize TLS in the pancreas of lupus-prone, anti-dsDNA antibody positive, NZB/NZW F1 mice by single photon emission computed tomography (SPECT) using ^99mTC labeled Albumin Nanocoll (35). Further advancement in SPECT and in vivo imaging technologies may make it possible to detect and longitudinally monitor TLS in LN.

Further research is needed to understand the composition and anatomy of TLS and roles of different cell types in TLS function. TLS develop in near proximity to large veins and arteries and contain lymph vessels and micro capillaries (50, 72). The larger renal drainage lymphatics in mice can be found in the same hilar area (50, 113). However, it is unknown if the blood flow in TLS is connected to the larger vessels. Little is known about the lymphatic flow within TLS (114). There is also limited research on the role of certain immune cell types, such as Treg cells, within TLS.

The presence of TLS may also pose challenges to treatment in LN. For example, local B cells in TLS may not be easily inhibited by systemic therapy, such as rituximab. In a study of chronic renal allograft rejection, systemic rituximab failed to cause TLS regression in kidney despite the depletion of peripheral B cells (115). This finding suggests that the local TLS may facilitate B cell survival and allow the evasion of systemic rituximab-mediated depletion. A case series reported three cases with recurrent and refractory oral pemphigus vulgaris. Two of the three had negative circulating desmoglein antibodies, whereas intralesional rituximab was effective in the treatment of the two (116). Intralesional rituximab likely killed the lesional B cells in the TLS and enhanced the lesion healing. The presence of TLS may also explain the relapse and treatment resistance of local lesions in the absence of circulating autoantibodies.

TLS recapitulate the functions of SLO at local chronic inflammatory sites. After the initial loss of tolerance, the activated autoimmune cells may infiltrate the target organs, where they may induce the formation of TLS that, in turn, may maintain the autoimmune response by facilitating interactions among different immune cell types and between various macromolecules and immune cells. TLS may facilitate a local activation of T and B cells in tissues resulting in a rapid response to local antigens, which may resemble the systemic adaptive immune response that occurs in SLO. Additionally, TLS may provide a platform for encounters between local autoantigens, antigen-presenting cells and lymphocytes, thus promoting an efficient antigen processing and presentation leading to the spreading of autoimmune response to additional autoantigens. In fact, animal studies have demonstrated the induction of high affinity antibodies in TLS. Furthermore, locally activated immune cells as well as tissue-resident cells may become a source of inflammatory mediators, thus exacerbating kidney injury. Both animal model and human studies support the role of TLS in the kidney to amplify autoimmune responses as well as cause the progression of renal damage in LN. These observations provide a strong rationale for developing new therapies that target events that initiate TLS formation, molecules that promote the maturation and organization of TLS, and by depleting the locally activated immune cells in the TLS of kidneys in LN. However, there are still many hurdles to resolve before realizing the potential of anti-TLS therapies. For example, the lack of non-invasive methods to visualize/quantify TLS in kidneys is a major hurdle in using TLS for clinical assessment, but recent success in visualizing TLS in the pancreas of lupus-prone mice by SPECT provides hope for early detection of TLS in LN using in vivo imaging.

MW and RS developed the outline and objective of the article, MW and YL wrote the draft of the article, TI and DN assisted with literature review, preparation of figures, and writing, and MW and RS revised it critically for important intellectual content. XL, SR, and JH contributed to the review. All authors contributed to the article and approved the submitted version.