Connexin 43 Communication Channels in Follicular Dendritic Cell Development and in Follicular Lymphomas

Follicular dendritic cells (FDC) show homo- and heterocellular metabolic coupling through connexin 43 (Cx43) gap junctions and support B cell selection and maturation in germinal centers. In follicular lymphomas B cells escape apoptosis while FDC develop abnormally. Here we tested Cx43 channels in reactive FDC development and follicular lymphomas. In culture, the treatment of FDC-B cell clusters (resembling to “ex vivo” germinal centers) with Gap27 peptide, mimicking the 2nd extracellular loop of Cx43 protein, significantly impaired FDC-B cell cluster formation and cell survival. In untreated cultures of intact clusters, cell proliferation showed a moderate reduction. In tissues, Cx43 protein levels run parallel with the density of FDC both in reactive germinal centers and in malformed follicles of follicular lymphomas and showed strong upregulation in newly generated and/or degrading bi-/multinuclear FDC of rudimentary processes. However, the inverse correlation between Cx43 expression and B cell proliferation seen in reactive germinal centers was not detected in follicular lymphomas. Furthermore, Cx43 levels were not associated with either lymphoma grade or bone marrow involvement. Our results suggest that Cx43 channels are critical in FDC and “ex vivo” germinal center development and in the persistence of FDC in follicular lymphomas but do not affect tumor progression.


Introduction
Follicular dendritic cell (FDC) meshwork serves as a scaffolding and antigen trap in germinal centers for boosting the generation of high affinity antibodies and B cell memory [1]. High affinity B cell clones are selected and rescued for survival through their binding antigens presented by FDC [2]. We earlier showed that, besides cell-cell adhesion and secreted cytokines, FDC can also mediate homo-and heterocellular direct interactions through cell membrane channels formed by connexin 43 (Cx43) protein [3]. Gap junction direct cellcell communication couples members of the FDC meshwork for a functional syncytium, which can also be directly linked through Cx43 channels to germinal center B cells [3,4].
In this study, we tested FDC-B cell cultures and follicular lymphomas (FL) of atypical FDC to see how critical Cx43 channels are in FDC development and maintenance and if Cx43 levels influence FL progression.
Gap junction channels are formed by aligning of connexin hemichannels of adjacent cell membranes [5]. The hemichannels, which may also function independently as a secretory pathway for releasing ATP, are made of six tetraspan transmembrane connexin molecules of 21 isotypes in human [6,7]. Gap junctions allow the rapid exchange of <1.8 kDa molecules including morphogenes, second messengers (e.g., Ca 2+ , IP 3 , and c-AMP), metabolites (e.g., nucleic acids and amino acids), and linear antigenic peptides of <1.2-1.8 kDa between cells at selective permeability depending on the isotype(s) involved [8]. Connexins and their channels can mediate signals concerned with cell differentiation, apoptosis, cell cycle control, and the formation of functional compartments within tissues [9][10][11]. Connexins are expressed in most immune cells and have been linked to the regulation of early hematopoiesis and immunoglobulin production and to the promotion of cellular immune response through antigen cross-presentation [5,12].

Journal of Immunology Research
FDC is also an important component of FL, an indolent neoplasia of follicular B cells, which frequently show t(14;18)(q32;q21) chromosomal translocation resulting in the overexpression of the antiapoptotic Bcl-2 protein [13,14]. FL tumor cells, characterized by CD10, bcl-2 and bcl-6 immunoreactions, are arranged in follicular structures of irregular sizes that may resemble reactive follicles but with a lost polarization [15,16]. Follicular areas can be highlighted by CD21, CD23, and CD35 positive distorted FDC meshwork which shows alternating focal hyperplasia and fragmentation [17]. Diminishing and lost follicular pattern in grade 3 FL correlates with the progression of FL into diffuse large B cell lymphoma (DLBCL) of poor outcome in 25-35% of the patients [14]. Bone marrow involvement of FL affecting 40-70% of the cases, where FDC and follicular structures may also form, indicates an adverse FL prognosis too [18][19][20]. On the other hand, stromal cells including FDC can support lymphoma B-cell survival and may contribute to the resistance both to chemotherapy and biological (trastuzumab) therapy targeting CD20 protein in tumor cells with a humanized monoclonal antibody [21][22][23][24].
In cultured low density cell fractions of reactive human tonsils, FDC-B cell clusters can form resembling to ex vivo germinal centers [4,25]. In this study, Cx43 docking and channel functions were perturbed by treating of these cultures with Gap27 connexin mimetic peptide which corresponds to the second extracellular loop (E2) of Cx43 protein [26]. Gap27 peptide significantly impaired the formation of FDC-B cell clusters and cell survival. Cx43 expression showed strong statistical correlation with the FDC meshwork both in reactive germinal centers and in FL, and it was upregulated in bi-or multinuclear FDC of underdeveloped processes. However, Cx43 levels had no significant association with either neoplastic B cell proliferation, tumor grade, or bone marrow involvement in FL. Our results suggest that Cx43 channels are critical in FDC and in "ex vivo" germinal center development and in the persistence of FDC in FL but may not play a significant role in FL progression. Since FDCs nurse FL B cells, inhibition of Cx43 in FL would likely to hinder FDC growth and eliminate its trophic factors for tumor cell survival and support chemotherapy and/or biological therapy.

FDC-B Lymphocyte
Cultures. Low density cell fractions of human tonsils, known to be enriched in activated germinal center B lymphocytes and FDC, were isolated with the consent of the patients as described before [4,27]. All reagents specified were from Sigma-Aldrich (St Louis, MO). Briefly, surgically excised tonsils were minced and then digested twice for 30 min each in 200 U/mL collagenase IV (C5138) and 10 U/mL DNase I (D4263) containing 90 g/mL gentamycin (G1272) in 30 mL Iscove's modified Dulbecco's medium (IMDM; I3390), at 37 ∘ C. The supernatants kept on ice were centrifuged at 600 g for 10 min, resuspended in IMDM, and layered (at 1 ×g) on discontinuous BSA gradients consisting of layers of 1.5%, 2.5%, and 5% BSA (A9706) in Hanks' balanced salt solution (H6648). Then the low density cells between the interfaces of 2.5% and 5% BSA were layered on Percoll gradients (P4937) and centrifuged at 1200 ×g at 4 ∘ C for 20 min to allow collection of <1060 mg/mL density cells. Isolated cells were either cytospinned or cultured on 24× 24 mm microscope coverslips (2-4 × l0 5 cells) for 2-24 h in IMDM containing 0.1% gentamycin and 10% fetal calf serum under 5% O 2 , 5% CO 2 , and 90% N 2 . For immunostaining, cells were fixed in acetone for 10 min and dried at room temperature for 30 min. Cell phenotype, all cell numbers, and average cell numbers within clusters were monitored and compared between Gap27 treated cultures and cultures treated with either 100 g BSA (untreated) or a scrambled peptide sequence (TFEPDRISITK) at 2 , 4, 6, 12, 16, and 24 h in at least 3 parallel coverslip cultures each time by counting 300 cells on each coverslip. Coverslip cultures were either immunostained (see below) or analyzed after nuclear staining with 7aminoactinomycin D (7-AAD; 1 : 1000; Invitrogen, Carlsbad, CA) for 1 min, postfixed in 4% neutral buffered formaldehyde for 10 min, and mounted using Faramount (Dako, Glostrup, Denmark). Samples were tested with Leica TCS4D confocal laser scanning microscope (Leica Lasertechnique, Heidelberg, Germany) using single and multichannel fluorescence combined with differential interference (Nomarski) optics.

Biopsy Samples and Tissue
Microarray. Formaldehydefixed, paraffin embedded lymph node biopsies of 35 untreated FL patients and the corresponding bone marrow core biopsies from 20 patients were selected for this study from the archives of the 1st Department of Pathology and Experimental Cancer Research (Budapest). The diagnoses were based on standard criteria using the classification of the World Health Organization (WHO) by considering histopathology, immunophenotype, and molecular features [14]. Clinicopathological data of the studied patients are summarized in Table 1. Bone marrow sampling took place within 2 weeks of the primary diagnosis of FL. Archived, reactive lymph node samples from 3 patients were used as normal controls. Representative areas were selected from each paraffin block based on routine hematoxylin and eosin staining and at least duplicate cores of 2 mm diameter were collected into tissue microarray (TMA) blocks using the computer-assisted TMA Master (3DHISTECH Ltd, Budapest, Hungary) [28]. The use of all human tissues including native tonsils for research purposes in this study has been approved by the Ethical Board of Semmelweis University (TUKEB 7/2006).

Immunohistochemistry.
Four m thick TMA sections were cut, mounted on charged adhesive slides (Super-Frost Ultra Plus, Braunschweig, Germany), and immunostained following routine dewaxing. For heat induced epitope

Cx43 Expression in Reactive Lymphoid
Follicles. In reactive lymphoid follicles particulate Cx43 immunoreaction was primarily associated to the FDC meshwork identified with CD21 (C3d, complement receptor) immunostaining (Figures 1(a)-1(b)). Significantly higher Cx43 levels were seen in the light zone where FDC was more developed and proliferating B cells were fewer than in the dark zone (see also Figures 4(l) and 4(m)). Cx43 reaction colocalized with CD21 and the desmosomal adhesion protein desmoplakin, both produced by FDC (Figure 1(c)). It was also detected in B cells (Figure 1(d)) and less frequently in CD4 positive T cells (Figure 1(e)) within germinal centers.

Cx43 Expression in Follicular
Lymphomas. CD10 positive FL follicles lost polarization compared to reactive germinal centers and frequently enclosed a fragmented FDC meshwork detected with CD21 immunoreaction (Figures 4(a)-4(b)). Similar to that in reactive follicles, Cx43 expression was significantly associated with the FDC meshwork (Figures 4(c)-4(d)), while Cx43 levels were very low in CD10 positive but CD21 negative/low tumor areas. Cx43 protein was highly upregulated in bi-or multinuclear FDCs with rudimentary processes and was partly colocalized with CD10 protein on FL B cells. In tumor regions without obvious vessels, reactive centroblasts, or extracellular matrix, cell proliferation was not significantly different between areas of compact or fragmented CD21/Cx43 double positive FDC (Figures 4(e)-4(f)). In bone marrow FL infiltrates, the highest Cx43 levels were detected in CD21/NGFR positive follicles (Figures 4(g)-4(h)), where Cx43 levels were usually lower and FDC was more fragmented (Figure 4(i)) than in lymph node FL follicles. Furthermore, Cx43 levels were higher in FDCfree diffuse bone marrow FL infiltrates than either in the preexisting bone marrow (Figure 4(j)) or in the diffuse FDCfree FL areas in lymph nodes. On the contrary, proliferating cell fractions were significantly higher in the preexisting marrow than in bone marrow FL infiltrates. Both FDC and stromal cell processes were positive for NGFR and occasionally showed continuity with each other in FL infiltrates at both localizations. Few Cx43 signals were also seen in NGFR positive stromal cell processes. Graphs summarize the significant association between CD21 and Cx43 levels in FL samples (Figure 4(k)) and the correlations of Cx43 or Ki67 levels with CD10 + /CD21 + (high) and CD10 + /CD21 − (low) FL areas (Figures 4(l)-4(m)). Spearman rank correlation was also highly significant between Cx43 and CD21 expression (rho = 0.981, < 0.001), supporting the colocalization of these proteins. This test showed only a weak negative trend either between Cx43 and Ki67 (rho = −0.154) or between CD21 and Ki67 (rho = −0.128) expression. In lymph node FL, there was no significant correlation between Cx43 expression and bone marrow involvement or tumor grade (Table 2) and between the proportion of FDC, detected using CD21 or CD23 immunoreactions, and tumor grade ( CD21 = 0.449; and CD23 = 0.112).

Discussion
Using cultured FDC-B cell clusters, we earlier revealed that Cx43 gap junctions functionally couple cells of the FDC meshwork and FDC to germinal center B cells [4]. Here we show in the same model that interfering with Cx43 docking and channel functions, by using Gap27 connexin mimetic peptide, can significantly impair the development of FDC-B cell clusters and cell survival. In tissues, Cx43 protein expression runs parallel to the density of FDC both in reactive germinal centers and in disorganized FL follicles. It is highly upregulated in large FDC bodies of rudimentary processes, consistent with either newly developing or degrading FDC. In reactive germinal centers, local Cx43 levels show an inverse correlation with B cell proliferation, which is lost in FL. Still Cx43 expression in FL lymph nodes is not linked statistically to either FL grade or bone marrow involvement. These data suggest that Cx43 protein and its channels are important in FDC meshwork and in "ex vivo" germinal center development and in the persistence of FDC in FL but do not influence significantly FL progression. Cx43 is the most ancient and prevalent connexin isotype, in which involvement in immune functions has been reported by several groups [12]. In the humoral arm, blocking of Cx channels was shown to inhibit the immunoglobulin production of B cells [29]. Cx43 gap junctions were also implicated in the LFA1-induced Rap1 GTPase activation, which can mediate CXCL12-directed B cell spreading, migration, and adhesion [30,31], which are important in germinal center reaction. Cx43 channels have also been involved in the regulation of T cell maturation and proliferation, in particular of CD4/Foxp3 double positive regulatory T cells [32][33][34] and in the cross-presentation of tumor or viral peptides between cells [35][36][37].
The contribution of Cx43 in FDC development and germinal center reaction was raised earlier by our detection of progressive accumulation of Cx43 plaques during the formation of FDC and secondary lymphoid follicles after repeated antigen challenge in mice [4]. In line with this, CXCL13, an essential chemokine for FDC and lymph node generation, can be induced by retinoic acid derived from vitamin A, a known promoter of Cx43 channels [38][39][40]. Here we provide further evidence on the impact of Cx43 in FDC growth and potential involvement in germinal center functions. Binding of Gap27 to the 2nd extracellular loop of Cx43 hemichannels is known to result in early channel closure and diminishing cell coupling [41]. Since FDCs normally grow long processes and establish strong interactions with each other, which look abortive and disrupted in Gap27 treated cultures, FDC adherence may also be affected through Gap27  directly preventing anchorage of the opposing connexin loops.
In germinal centers, FDC provide a supportive niche for B cell survival involving direct FDC-B cell interactions including CXCL13/CXCR5, CXCL12/CXCR4, ICAM-1/LFA-1, and VCAM-/VLA-4 antigen trapped by CD21 and CD35 complement receptors or by FC R and FC R (CD23) immunoglobulin Fc receptors for reaching out by B cell receptors as well as by releasing cytokines such as BAFF, IL-6, and IL-15 (16) [1,[42][43][44][45]. In addition, Cx43 channels are formed by both FDC and B lymphocytes, which permit their homo-and heterocellular (FDC-B cell) metabolic coupling [3,4,46,47]. In culture, activated B cells are also prone to apoptotic cell death without FDC help and/or T cell interactions [48]. By interfering with FDC development using Gap27 peptide, FDC support of B cell survival is also abused, resulting in concomitant damage of B cells. Besides reducing cell numbers within clusters, Gap27 treatment also led to a significant decrease in the absolute cell numbers, while proliferating cell fractions were moderately elevated compared to untreated controls. This is consistent with a significantly reduced cell survival and increased detachment and loss of damaged cells after Gap27 treatment, while numerous cells which are dominantly out of clusters can still proliferate. It is likely that both gap junctions and hemichannels are affected and both homo-and heterocellular Cx43 channels which are involved [41]. However, further studies are required to clarify the mechanisms of action of Gap27 peptide on FDC-B cell interactions, B cell maturation, and the potential involvement of hemichannels.
In FL B cells, forced Bcl-2 expression protects tumor cells from apoptosis, but B cell survival factors released by FDC and T cells are still important for FL B cells persistence. In line with this, FL B cells cannot be grown in culture without factors produced by FDC and CD4 + T helper cells including CD40L, IL-2, and IL-4 [49][50][51]. Since Cx43 channels may support FDC development and survival even in FL, blocking of Cx43 would most likely impair FDC and its production of B cell survival factors, which have been implicated in the FDC mediated therapy resistance of FL B cells [22][23][24].
The FDC meshwork is frequently distorted in FL, showing local hyperplastic cell bodies with rudimentary processes and complete loss of FDC in CD10 + diffuse tumor infiltrates [14]. Since normal germinal center structure is based on trophic and regulatory factors shared between T helper cells, B cells, and FDC, deformed FDC reflects inappropriate trophic signaling by malignant B cells [52]. The importance of Cx43 channels in syncytial FDC is reflected by their following of the pattern and density of FDC detected with CD21 (and CD23) immunoreactions both in reactive germinal centers and in FL follicles. In addition, Cx43 protein is particularly upregulated in bi-or multinuclear FDCs of large cell bodies but rudimentary processes, which are likely to be either newly forming or degrading FDC in FL. These findings are in line with our results in Gap27 treated cultures by both supporting the role for Cx43 channels in FDC development and survival.
Cx43 channels have been implicated in cell cycle control and cell differentiation [53]. In reactive germinal centers the inverse correlation between Cx43 expression and proliferating B cell fraction suggested Cx43 a role in the control of B cell proliferation. A similar but nonsignificant trend was observed in untreated FDC-B cell cultures of intact cell clusters. Also, in the diffuse bone marrow infiltrates of FL, LNGFR positive stromal cell hyperplasia was accompanied by significantly higher Cx43 expression, lower B cell proliferation, and tumor grade. It is of note, however, that tumor microenvironment including elevated CD8 + T cells, forkhead box protein 3 (FoxP3) + T cells, and CD68 + macrophages can also contribute to the downgrading of bone marrow FL [28]. Nevertheless, Cx43 levels in lymph node FL showed no significant link with bone marrow involvement or tumor grade, two known indicators of FL progression.
In conclusion, the exact role of Cx43 channels and cell coupling in germinal center is still needed to be clarified. Our earlier study showed that the FDCs are coupled through Cx43 gap junctions and Cx43 protein is upregulated in FDC during repeated immune challenge [4]. Here we revealed that Cx43 channels are crucial for FDC to develop into a syncytial network which support the survival of FDC-B cell clusters and for the persistence of FDC in FL but do not influence significantly the progression of FL. Since both B cells and T cells can produce Cx43 channels and are in direct contact with the functionally coupled FDC meshwork, it is likely that Cx43 channels are involved in the integration of germinal center functions. Details of the potential metabolic cooperation between FDC, B cells, and T cells need further in vivo studies, which have so far been prevented by the complexity germinal center reaction.