Dissection of the organizational phases of lymphoid neogenesis in RA synovitis
The aggregational phases of the synovial lymphoid tissue were defined using a grading score as we have previously described 23. Briefly, the small perivascular clusters (total number of cells 43.2±17.3, mean ± standard deviation), reasonably representing the first aggregational phases post-diapedesis, were defined as Grade 1 and were present in 100% of the specimens. A variable number of Grade 2 (total number 176.1±64 cells) and Grade 3 (808.1±575.3 cells), representative of progressive enlargement stages of organizing aggregates, coexisted in 17 out of 20 tissues together with a variable degree of diffuse/interaggregate infiltration (Fig. 1 and Table 1).
Figure 1. Histomorphological grading. H&E staining shows representative examples of Grade 1 (A), Grade 2 (B) and Grade 3 (C) aggregates in RA synovium. (D) A low magnification view of a synovial specimen demonstrates the coexistence of Grade 1, 2 and 3 aggregates with a variable level of diffuse/interaggregate infiltration.
Download figure to PowerPoint
Table 1. Histomorphological grading
Using this score, we determined the relationship between the aggregational level of the inflammatory infiltrate and lymphoid organization, quantifying the prevalence of specific lymphoid features (T-B cell segregation, CD21+ FDC networks, PNAd+ HEV). Grade 1 aggregates showed rare presence of PNAd+ HEV, poor B cell representation, no defined T-B compartmentalized areas and the absence of CD21+ FDC networks in all tissues analyzed (Tables 2–4 and Fig. 2). The progressive enlargement of the aggregates was associated with a global increase of PNAd+ HEV containing aggregates, significant B cell enrichment [Grade 1: median 3.6%, interquartile range 0–11.4; Grade 2: median 14.3%, interquartile range 6.3–21.1; and Grade 3: median 27.2%, interquartile range 22.6–40.9; p<0.01] and the detection, in a proportion of tissues and in selected Grade 3 aggregates, of T-B compartmentalized areas and defined CD21+ FDC networks (Tables 2–4 and Fig. 2). CD35 immunostaining, performed to further evaluate the phenotype of synovial FDC, showed aggregate colocalization with CD21 for the presence of dendritiform networks (Fig. 2).
Table 2. Histomorphological grading and T-B cell segregation
Table 3. Histomorphological grading and CD21+ FDC networks
Table 4. Histomorphological grading and PNAd+ HEV
Figure 2. Histomorphological grading and lymphoid organization in RA synovium. Consecutive sections of RA synovium (A–T) and human tonsil (U–X) were immunostained for CD3, CD20, CD21 and PNAd antigens (in red). Representative examples of Grade 1, Grade 2 and Grade 3 aggregate areas are shown. Grade 3 aggregates were characterized by increasing levels of structural organization: intermixed T and B cells (I–K), T-B cell compartmentalization (M–O) and T-B cell compartmentalization with centrally located CD21+ FDC networks (follicular pattern) (Q–S). PNAd+ HEV can be recognized in aggregates of different dimensional grade and structural organization (H, L, P, T). (U–X) Lymphoid organization in human tonsil. The distribution of PNAd+ HEV in synovial (T) and tonsillar (X) follicular areas are shown (arrows indicate the localization of the CD21+ FDC network and the GC). SyA-SyB and ToA-ToB show parallel sections stained for CD21 and CD35 (in red) in synovial aggregate and tonsil, respectively.
Download figure to PowerPoint
These data demonstrate a positive relationship between cellular clusterization and the acquisition of lymphoid organizational features. However, as demonstrated by serial section analysis (Fig. 2), this trend was not uniformly recognized, and heterogeneous levels of organization were observed within different Grade 3 and among different samples. In some tissues, together with Grades 1 and 2 (Fig. 2A–H), we detected Grade 3 aggregates with only T-B intermixed distribution and no FDC (Fig. 2I–L); in other tissues, T-B segregated clusters, still in the absence of FDC, were also detected (Fig. 2M–P), while in 7 out of 20 specimens, we found a variable number of T-B segregated clusters with centrally located FDC networks (“follicular” pattern) (Fig. 2Q–T) coexisting with the previously described grades. In some follicular aggregates, the subset distribution seemed to recapitulate the architecture of secondary lymphoid organs with the definition of T cell-rich microdomains positioned at the periphery of CD21+ FDC network-rich B cell areas (Fig. 2Q, R for synovium and U, V for tonsil). In keeping with this organizational program, we demonstrated PNAd+ HEV in 66.7% of follicular structures, mainly distributed, as in tonsil secondary follicles, outside CD21+ FDC network-rich areas and in T cell-rich regions (Fig. 2Q–T for synovium and U–X for tonsil).
The coexistence of aggregates of progressively increasing size and increasing level of structural organization supports the concept that the synovial inflammatory infiltrate is not stable and that the dynamic events of cellular infiltration, progressive aggregation and organization can be histologically captured by our grading score. We thus defined the expression patterns of CXCL13 and CCL21 in relationship with the structural organization of fully formed Grade 3 as well as their relationship with the progressive stages of aggregate formation (Grades 1 and 2).
CXCL13 in situ production and microstructural lymphoid organization
We demonstrated CXCL13 protein and mRNA in 100% of Grade 3 aggregates with a follicular pattern. In this context, extending previous immunohistochemical analysis indicating the colocalization between CXCL13 protein and FDC in ectopic GC 17, 18, we found the CK produced both within CD21+ FDC network-rich area and in the external B cell distribution zone (Fig. 3O–Q for structure and R, S for CXCL13). This localization is in keeping with previous studies showing CXCL13 production both in GC as well as in the B cell mantle zone in secondary lymphoid organs 24–26 and provide evidence of a similarity in the expression pattern of CXCL13 between secondary lymphoid organs and synovial lymphoid tissue.
Figure 3. CXCL13 and CCL21 production and lymphoid organization. IHC for CD3, CD20, CD21, CXCL13 and CCL21 (red) and ISH for CXCL13 and CCL21 (black) were performed on consecutive sections of RA synovium. Representative aggregate areas featuring increasing levels of structural organization, with intermixed T and B cells (A–C), T-B cell compartmentalization (H–J) and T-B cell compartmentalization with centrally located CD21+ FDC networks (O–Q), are illustrated. CXCL13 and CCL21 protein and mRNA are present inside a follicular Grade 3 aggregate (R, S and T, U) and in less organized aggregate areas (K, L and D, E for CXCL13 and M, N for CCL21). Inside the follicular aggregate, CXCL13-producing cells (R, S) are found inside and outside the CD21+ FDC network-rich area (Q), with a predominant localization in the B cell distribution zone (P). Note the colocalization between CCL21 (T, U) and the T cell-rich compartment (O).
Download figure to PowerPoint
However, in RA synovium, CXCL13 was also found in aggregate areas characterized by the absence of organized CD21+ FDC networks with (Fig. 3H–J for structure and K, L for CXCL13) or without (Fig. 3A–C for structure and D, E for CXCL13) distinct T-B cell segregation, with a trend toward co-localization with B cell distribution areas in compartmentalized aggregates (Fig. 3H–J for structure and K, L for CXCL13). These observations were supported by serial section analysis in some samples where the whole volume of the aggregates was examined. The same experiments also demonstrated CXCL13 production in small aggregates and in some diffusely infiltrated areas (Fig. 4A–F), confirming that neither the follicular structure nor the environment of a Grade 3 cluster is necessary for CXCL13 production within the synovial lymphoid tissue. This phenomenon, which we previously reported in a published abstract 27, was recently confirmed independently by Carlsen et al. 19. They also extended this observation by demonstrating, in agreement with the work of Perrier et al. 28, that CXCL13 can be produced by non-stromal hemopoietic cells of the monocyte lineage.
Figure 4. CXCL13 production and histomorphological grading. Protein and mRNA expression of CXCL13 were detected by IHC and ISH, respectively, on consecutive sections of RA synovium (positive cells in black). CXCL13-expressing cells were localized in sublining areas characterized by a high density of mononuclear inflammatory infiltrate, with a strong association with large cellular aggregates (A, B). Few CXCL13-producing cells were present in areas featuring a high density of diffuse infiltrate and small-sized aggregates (C, D; higher magnification E, F). Representative images of Grade 1 (G, H), Grade 2 (I, J) and Grade 3 aggregates (K, L) are shown.
Download figure to PowerPoint
Having established that CXCL13 can be produced in the absence of an advanced organizational stage and a critical cell mass (Grade 3), we next examined the relationship between CXCL13 production and lymphoid tissue formation as it progresses through enlargement phases analyzed according to the grading score. We demonstrated a rare but definite expression of CXCL13 in Grade 1, with a significant increase in the global percentage of CXCL13+ aggregates alongside their progressive dimensional enlargement (Grade 1 vs. Grade 2, p<0.001; Grade 2 vs. Grade 3, p<0.001) (Table 5). Accordingly, while one or few CXCL13+ cells were observed in positive Grade 1 and Grade 2, many more positive cells were present in the majority of Grade 3 (Fig. 4G–L). This trend was in contrast with the rare or absent CXCL13 production recognized in the lining layer and sublining stroma devoid of mononuclear infiltrate, indicating the existence of a specific link between lymphoid tissue formation and CXCL13 production. Importantly, despite the consistent variability in the number of CXCL13-producing cells observed among different aggregates and tissues, a positive relationship between lymphoid aggregate formation and CXCL13 expression was recognized in all 17 out of 17 samples featuring Grade 3 aggregates (Table 5), providing evidence of a correlation between CXCL13 production and lymphoid cluster formation recognizable across the spectrum of RA synovitis.
Table 5. Histomorphological grading and CXCL13 expression
In summary, our data indicate that CXCL13-producing cell differentiation/local migration is associated with cellular aggregation and lymphoid tissue formation. Our results also suggest that CXCL13 is inducible independently and possibly upstream of FDC networks, GC reactions and a defined T-B compartmentalization, with the acquisition in organized aggregates of a secondary lymphoid organ-like distribution. In keeping with this analogy, similarly to lymph node HEV (Fig. 5G), we demonstrated the absence of detectable levels of CXCL13 production in synovial CD31+/PNAd+ vessels (Fig. 5A–F), although we cannot exclude vascular transcytosis or passive absorption.
Figure 5. CXCL13 production and the synovial vascular system. Expression of CXCL13 (A, C, E), CD31 (D) and PNAd (F) was detected in sections of RA synovium by IHC (positive cells in black). mRNA expression of CXCL13 was determined by ISH (B) (black). Synovial vascular structures (arrowheads) were not recognized as CXCL13 protein-producing sites as determined by the absence of detectable levels of endothelial CXCL13 protein production (A for CXCL13 protein and B for CXCL13 mRNA). This observation was confirmed by the analysis of CXCL13 protein expression in CD31+ vessels and PNAd+ HEV (arrowheads in C, D and E, F). CXCL13 mRNA was not detected in human lymph node HEV (arrowheads in G).
Download figure to PowerPoint
CCL21 in situ production and microstructural lymphoid organization
Although the cooperative function and reciprocal distribution of CXCR5 and CCR7 ligands in secondary lymphoid organ embryogenesis and post-natal life have been documented 1, 9, no data are currently available about the mutual distribution of CCL21- and CXCL13-producing cells under inflammatory conditions and in the developing synovial lymphoid tissue. To address this point, we performed immunohistochemistry (IHC) and in situ hybridization (ISH) for CCL21 on sections sequential to those stained for CXCL13.
CCL21 production was detected in 42.9% (compared to 100% for CXCL13) of follicular aggregates. In line with their secondary lymphoid organ distribution, while CXCL13 was associated mainly with B cell distribution areas, CCL21 was localized outside the CD21+ FDC network-rich area and was predominantly found in T cell-rich zones containing HEV (Fig. 3O–Q for structure and T, U for CCL21). To our knowledge, this is the first study in RA that demonstrates the in situ production of both CXCL13 and CCL21 in a compartmentalized fashion within organized follicular microstructures (CD21+ FDC network-rich). Our data also demonstrate that within the T cell-rich domains of FDC-rich T-B compartmentalized aggregates, both CCL21-producing cells (constitutively produced in the T cell area of secondary lymphoid organs) and PNAd+ HEV (mainly localized in the same lymphoid microenvironment) can be found (see Fig. 2). These observations provide evidence that synovial lymphoid aggregates, together with an FDC-rich B cell area, can also acquire a vestige of the T cell compartment of secondary lymphoid organs.
In addition to such expression within follicular structures, CCL21 protein and mRNA were also detected, by serial section analysis, independently of CD21+ FDC network-positive aggregates, with a trend to associate with T cell enrichment areas (Fig. 3H–J for structure and M, N for CCL21). In some cases, CCL21 was also detected in the absence of defined T-B compartmentalization and in minor aggregational stages, indicating that as for CXCL13, the mechanisms involved in CCL21 production do not require the stable microenvironment of a follicular structure or a Grade 3 aggregational level.
We thus investigated the relationship between CCL21 expression and ectopic lymphoid tissue formation. We observed, in keeping with previous reports 20, 21, that CCL21 expression was associated with scattered or clustered cells (Fig. 6A, B) as well as with vascular structures (Fig. 6C, D). Some of these scattered non-vascular CCL21+ cells, whose phenotype is currently under investigation, appeared irregular in shape, with prolonged cytoplasmic processes (not shown). Although it has been suggested that CCL21 may play a role in ectopic lymphoid neogenesis in RA synovium 20, 21, no comparative analysis between the vascular and non-vascular pattern was carried out in those studies. Our systematic in situ analysis demonstrated that non-vascular CCL21-producing cells were specifically associated with lymphoid aggregates; they were only rarely or were not detected in the sublining stroma devoid of a consistent mononuclear infiltrate. In addition, the grading analysis demonstrated a significant positive relationship between the expression of CCL21 and the progressive enlargement of cell clusters (Grade 1 vs. Grade 2, p<0.002; Grade 2 vs. Grade 3, p<0.02) (Table 6), with rare expression in Grade 1 (Fig. 6E, J). Despite quantitative differences in the number of CCL21+ cells among different aggregates and tissues, the positive relationship between lymphoid aggregate formation and CCL21 production was observed in 16 out of 17 patients featuring Grade 3 aggregates (Table 6). These results indicate a similar trend between the developmental expression patterns of non-vascular CCL21 and CXCL13. However, despite this similarity, the comparative grading analysis demonstrated more restricted distribution of CCL21 compared to CXCL13 both in terms of the percentage of positive aggregates (Tables 5 and 6) and the number of positive cells within the aggregates (Fig. 3). This observation indicates a quantitative difference in the relative balance of these factors in the ectopic T-B lymphoid aggregates compared to adult secondary lymphoid organs where, in contrast, T cell-rich CCL21+ areas predominate.
Figure 6. CCL21 production and histomorphological grading. Protein and mRNA expression of CCL21 were detected by IHC and ISH on consecutive sections of RA synovium (positive cells in black). CCL21 production was detected in scattered cells strongly associated with cellular aggregates (A, B) and in vascular-like structures recognizable both inside and outside cellular clusters (C, D). Representative images of a Grade 1 (E, F), Grade 2 (G, H) and Grade 3 aggregates (I, J) are shown.
Download figure to PowerPoint
Table 6. Histomorphological grading and CCL21 expression
Another important aspect that emerged from the analysis of non-vascular CCL21 production was the fact that CCL21+ cells were often found associated perivascularly with CD31+ vessels (Fig. 7A, B), characterized by PNAd expression and the HEV morphology (Fig. 7C, D). This resembles the CK-addressin molecular system involved in CCR7+/L-selectin+ T and B cell recruitment to secondary lymphoid organs 29. However, of considerable importance, PNAd+ HEV in RA synovium did not reveal CCL21 endothelial production, as shown by IHC and ISH (Fig. 7E, F). This is in line with an independent study that reported no CCL21 production by endothelial cells in human lymph nodes 30 but contrasts with established data in murine secondary lymphoid organs demonstrating strong CCL21 production by HEV 1. To obtain direct proof of species-specific differential expression of CCL21 in lymphoid HEV, we performed a comparative CCL21 mRNA analysis in human and mouse lymph nodes. We obtained clear evidence of strong CCL21 expression in mouse but not human HEV (Fig. 7G, H). In addition, as in RA synovium, in human lymph nodes CCL21 was produced in T cell-rich areas and in close proximity of HEV (Fig. 7H). In summary, in keeping with a lymphoid neogenetic program, synovial non-vascular CCL21 is related to aggregate development and to the acquisition of the molecular and morphological features of human but not mouse secondary lymphoid organ vascular system. These data, as for CXCL13, do not exclude that transcytosis of CCL21 onto the lumen of human synovial or lymph node HEV may take place.
Figure 7. CCL21 production and the synovial vascular system. Two CCL21 production patterns in RA are illustrated: perivascular (upper panel) and vascular (lower panel). In the upper panel, consecutive sections (A and B, IHC) show the anatomical association of CCL21+ cells with CD31+ vessels (red). IHC (positive cells in red) also revealed perivascular CCL21-expressing cells (C) associated with strongly PNAd+ HEV-like vessels (D). Insets show high magnification views of the aggregate areas indicated by the long arrows. High magnification views of the same vessel on consecutive sections show CCL21 protein (E) and mRNA (F) detected by IHC and ISH, respectively. Note that CCL21 is produced not by the endothelium itself but at the abluminal side of the vessel with clear HEV morphology (protein in red, mRNA in black). In the lateral panel, ISH for CCL21 in murine (G) and human (H) lymph node is shown. It can be clearly seen that mouse HEV strongly produce CCL21 (in black, white arrows), while no detection could be observed on human HEV (black arrows). In the lower panel, IHC for CCL21 (I) and CD31 (J) on consecutive sections show a double-positive vessel (in red, arrows). The arrows in K and L show the absence of PNAd immunoreactivity in three CCL21+ vessels (red). (M) The selective immunoreactivity for CCL21 (red) in one of two adjacent vessels is illustrated at a higher magnification; while no red blood cells are observed in the CCL21+ vascular structure, erythrocytes can be clearly seen in the lumen of the CCL21– vessel (arrows). (N) The luminal presence of mononuclear cells inside CCL21+ vessels (red) is shown.
Download figure to PowerPoint
In contrast to the non-vascular pattern, vascular CCL21 could be recognized in selected vessels both inside and outside cellular aggregates, even in sublining areas lacking diffuse mononuclear infiltrate. Furthermore, CCL21 vascular expression was found in only 7 out of 17 samples featuring Grade 3 aggregates, indicating a less specific link and lower global correlation between this pattern and lymphoid aggregate development in RA. Notably, even within lymphoid aggregates, CD31+ CCL21-producing vessels (Fig. 7I, J) were PNAd– (Fig. 7K, L) and frequently characterized by a flat endothelial wall (Fig. 7M). This, together with the absence of red blood cells and the presence of mononuclear cells (CD3+, CD20+, not shown) intraluminally (Fig. 7N), suggests that some of these vessels may be lymphatic in origin. Thus, vascular CCL21 production appears to be partially independent from lymphoid tissue formation and organization, suggesting the existence of a different regulatory mechanism controlling the two patterns of CCL21 production in human inflamed tissues. A dual regulatory control of CCL21, similar to the one proposed here, has been reported for mouse lung, where CCL21 was shown to be constitutively expressed in a lymphotoxin (LT)- and lymphocyte-independent manner, but up-regulated following airway inflammation and cellular infiltration by membrane-bound LTβ- and cell-dependent pathways 31.
Summary and conclusions
Taken together, our data show that the in situ production of CXCL13 and CCL21 (non-vascular pattern) is associated with aberrant lymphoid tissue formation and is significantly correlated with cluster enlargement with a predominant expression of CXCL13. We demonstrated the possible coexistence of these CK in the ectopic environment of RA synovial aggregates, showing that CXCL13 and CCL21, when expressed in organized clusters, tend to associate with B and T cell distribution areas. Also, in keeping with human lymph nodes, we showed that CXCL13 and CCL21 are not produced by synovial PNAd+ HEV. In addition, we provide evidence that CCL21 frequently associates with the abluminal side of such vessels, re-creating the CK-addressin expression pattern that characterizes the human lymphoid venular system. In conclusion, this study provides direct in situ evidence that both CXCL13 and CCL21 are linked with lymphoid tissue development in the context of a lymphoid neogenetic organizational program. In this context, we also showed that CXCL13 and CCL21 production is inducible in the absence of fully formed follicular organization and in minor aggregational stages, demonstrating the presence of protein and mRNA in the putative upstream phases of the process of lymphoid organization. These data, combined with previous studies indicating the marked increase of CXCL13 and CCL21 mRNA expression levels in samples characterized by GC development 18, support the concept of a causative involvement of these factors in lymphoid neogenesis, acting upstream or in parallel with the formation of T-B cell-rich areas, FDC and GC. Better understanding of the molecular mechanisms of induction and regulation of CXCL13 and CCL21 under inflammatory conditions could open new therapeutic avenues for the treatment of RA.