Essential Role of Canonical NF-κB Activity in the Development of Stromal Cell Subsets in Secondary Lymphoid Organs

Secondary lymphoid organs (SLOs) develop through complex processes of intimate interactions between lymphoid cells and nonhematopoietic stromal cells (1, 2). The interplay of lymphoid tissue inducer cells, an innate lymphoid cell population, with mesenchymal lymphoid tissue organizer (LTo) cells in fetal anlagen is necessary for the formation of lymph nodes (LNs) and Peyer’s patches (PPs) (3–6), whereas the organization of white pulps in the spleen involves postnatal collaboration of mature lymphocytes and perivascular stromal cells (7, 8). In these cellular processes, lymphotoxin (LT) α1β2 and LTβR, which are displayed on lymphoid cells and mesenchymal cells, respectively, and the activation of NF-κB transcription factors via the canonical (RelA/p50 complex) and noncanonical (RelB/p52 complex) pathways are believed to play the key roles; gene deficiencies of these signaling mediators lead to the absence or malformation of SLOs (1, 9–12). In particular, NF-κB is required for the expression of chemokines such as CXCL13, CCL19, and CCL21, which recruit lymphoid cells to SLOs (11–16). However, in a strict sense, it has not been proven directly that NF-κB activity in LTo cells or mesenchymal progenitors itself is essential for the development and tissue organization of SLOs in vivo. Recent studies have suggested the presence of other complicated cellular interactions involving endothelial compartments in LN formation and distinct organ-specific development programs (8, 17).

In adult SLOs, B and T cells are clearly segregated in the follicles and T cell area. Different types of fibroblastic stromal cells (FSCs) that form intricate networks support each of the regions by providing a structural backbone (18). Follicular dendritic cells (FDCs) and marginal reticular cells (MRCs) in the center and outer part of the follicle, respectively, produce CXCL13 that regulates B cell localization and motility (19–21). MRCs, which display characteristics similar to those of LTo cells, such as the expression of RANKL and MAdCAM-1, have been shown to be FDC precursors (21–23). In T cell area, T zone reticular cells (TRCs), also known as fibroblastic reticular cells, produce CCL19 and CCL21, affecting the localization and migration of T cells and dendritic cells (24–27). These three stromal cell subsets are common to all types of SLOs and are considered to originate from a common mesenchymal progenitor (17, 28, 29).

The factors required for the development of FDCs are well studied, as this stromal cell type is absent in mice engineered to lack some immune system–related genes (10, 30). In particular, signals through LTβR and TNFR together with subsequent NF-κB activation play critical roles in the differentiation and maintenance of FDC network in the follicles (16, 31, 32). In contrast, the developmental processes of MRCs and TRCs remain poorly understood. Moreover, the mechanisms that regulate the generation of FSC subsets may possibly differ between SLO types, although it could not be verified in animals that completely lack LNs and PPs (e.g., in the cases of deficiencies in NF-κB components).

Ccl19-Cre BAC transgenic mouse strain is a useful tool for conditional gene targeting and fate mapping in FSC lineages during SLO development because LTo cells in fetal anlagen or postnatal progenitors express Cre recombinase driven by the Ccl19 gene (8, 33). A mutant form of IκBα, IκBSR, is resistant to the degradation upon signaling; it strongly inhibits the activation of the canonical NF-κB pathway (34), which in turn suppresses the expression of noncanonical pathway components. Thus, IκBSR potentially inhibits both pathways (15). In this study, to assess the consequences of in vivo inactivation of the canonical NF-κB pathway specific to FSCs in SLOs, we generated a mouse in which both IκBSR and EGFP are conditionally induced under the control of the Ccl19-Cre locus. Characteristic structural alterations of SLOs in these animals prompted us to address the cause of these abnormalities. Fate mapping using bicistronic EGFP expression permitted the identification of IκBSR-expressing cells in situ, which allowed us to elucidate the indispensable role of the canonical NF-κB pathway activity in the development of FSC subsets and suggested the presence of unique differentiation programs.

C57BL/6JJcl mice were purchased from CLEA Japan. Ccl19-Cre [Tg(Ccl19-cre)^489Biat] and R26-stop^flox-EYFP[B6.129 × 1-Gt(ROSA)26Sor^tm1(EYFP)Cos/J] mice were described previously (33). Mice were maintained under specific pathogen-free conditions in the animal facility of the Niigata University. All animal procedures were approved by the Committee on Animal Research at the Niigata University.

Murine IκBα cDNA was amplified from LN total RNA by RT-PCR using the following primers: 5′-TTAGATCTATGTTTCAGCCAGCTGGGCA-3′ and 5′-TTGCGGCCGCTTATAATGTCAGACGCTGGC-3′. To generate IκBSR, phosphorylation-resistant mutations (S32A/S36A) were introduced into the cloned cDNA by stepwise PCR using the above primers and following primers: 5′-CCTTCATGGCGTCCAGGCCGGCGTCGTGGC-3′ and 5′-GCCACGACGCCGGCCTGGACGCCATGAAGG-3′. R26-stop^flox-IκBSR mice were generated by knocking a targeting construct that contained CAG promoter, a reverse-oriented Neo^r cassette, transcription termination signal held between two loxP sites (stop^flox), and a bicistronic gene cassette encoding IκBSR-ires-EGFP, into the Rosa26 locus, as described previously (35). The linearized construct was electroporated into C57BL/6J mouse primary embryonic stem cells, and G418-resistant colonies were screened by Southern blotting of genomic DNA using a Neo^r-flanking probe. R26-stop^flox-IκBSR mice were crossed with Ccl19-Cre mice to obtain Ccl19-IκBSR animals that were heterozygous for the Ccl19-Cre allele and homozygous for the R26-stop^flox-IκBSR allele. Mice of 8–12 wk of age were used for experiments. Homozygous R26-stop^flox-IκBSR mice were used as controls.

Primary Abs against the following proteins were used: CD3e (145-2C11), B220 (RA3-6B2), PNAd (MECA-79), MAdCAM-1 (MECA-367), CD45 (30-F11), CD31 (390), and podoplanin (8.1.1) (eBioscience); desmin (Abcam); CD35 (8C12) (BD Biosciences); VCAM-1 (BAF643), RANKL (BAF462), CCL21 (BAF457), CXCL13 (BAF470), and LYVE-1 (BAF2125) (R&D Systems); GFP (598) (MBL); Cre (D7L7L) (Cell Signaling). For secondary reagents, PE-, allophycocyanin-, AlexaFluor 488–, 546–, 555–, 594–, or 633–conjugated streptavidin, anti-rabbit IgG, and anti-rat IgG were purchased from Molecular Probes.

Single-cell suspensions were prepared from excised SLOs and stained with the Abs. To analyze stromal cells, LNs were digested with 1 mg/ml collagenase D and 0.1 mg/ml DNase I (Roche) as described (36) and stained with Abs against CD45, CD31, and podoplanin. Data were acquired using a FACSCalibur (BD Biosciences) flow cytometer and analyzed with CellQuest (BD Biosciences) or FlowJo software.

SLOs were fixed with 0.05% phosphate buffer containing 0.075 M l-lysine (pH 7.4), 0.01 M NaIO₄, and 1% paraformaldehyde at 4°C for 16–24 h. After fixation, LNs were equilibrated gradually with 10, 20, and 30% sucrose in PBS at 4°C, embedded in OCT compound (Sakura), and frozen at −80°C. Frozen sections (10 μm) were stained with Abs and mounted with Permafluor (Thermo Fisher Scientific). The specimens were examined using an LSM710 (Carl Zeiss) or an FV1200 (Olympus) confocal microscope. Digital images were prepared using Adobe Photoshop CS6 software.

For whole-mount immunohistochemistry, excised organs were fixed with 1% paraformaldehyde in PBS overnight at 4°C, and endogenous peroxidase activity was blocked with 0.3% H₂O₂ and 40% methanol in PBS overnight at 4°C. The specimens were permeabilized with 0.3% Triton X-100 in PBS at 4°C for 1.5 h followed by blocking with 1% BSA in PBST (0.05% Tween-20 in PBS) at 4°C for 2 h. The specimens were then incubated with 2 μg/ml biotin–anti-B220 Ab (eBioscience) in 1% BSA in PBST overnight at 4°C. Biotinylated Ab was detected using Vectastain ABC Kit and DAB Substrate Kit (Vector).

Total RNA was extracted from SLOs using an RNeasy Mini kit (QIAGEN) or TRIzol reagent (Invitrogen), and cDNA was synthesized using the Superscript III First-Strand Synthesis System (Invitrogen). Quantitative PCR was performed using the LightCycler FastStart DNA Master^PLUS SYBR Green I (Roche) on a LightCycler (Roche). Primers used were as follows: Cxcl13, 5ʹ-TGGCTGCCCCAAAACTGA-3ʹ and 5ʹ-TGGCACGAGGATTCACACAT-3ʹ; Ccl19, 5ʹ-CTGCCTCAGATTATCTGCCAT-3ʹ and 5ʹ-AGGTAGCGGAAGGCTTTCAC-3ʹ; Ccl21a, 5ʹ-ATCCCGGCAATCCTGTTCTC-3ʹ and 5ʹ-GGTTCTGCACCCAGCCTTC-3ʹ; Des, 5ʹ-GTGGATGCAGCCACTCTAGC-3ʹ and 5ʹ-TTAGCCGCGATGGTCTCATAC-3ʹ; Gapdh, 5ʹ-GCCAAGGTCATCCATGACAACT-3ʹ and 5ʹ-GAGGGGCCATCCACAGTCTT-3ʹ. The value was normalized to expression level of the Gapdh gene.

The spleen was digested with 1 mg/ml collagenase D, 0.02 mg/ml Liberase TM, and 0.05 mg/ml DNase I (Roche). The CD45⁻ stromal cell fraction was enriched from the cell suspension by magnetic cell sorting using microbeads-conjugated anti-CD45 Ab (Miltenyi) and biotin–anti-CD31 Ab (eBioscience) followed by microbeads-conjugated anti-biotin Ab (Miltenyi). Cells were suspended in 10% FCS DMEM medium supplemented with penicillin and streptomycin and cultured in 24-well plates coated with gelatin (0.3% in PBS, overnight at 37°C) for 4–6 d. Semiconfluent cells were stimulated without or with 20 ng/ml TNF-α (PeproTech) for 30 min. Cells were fixed with 4% paraformaldehyde-PBS at 4°C for 10 min and permeabilized with 0.5% Triton X-100 and 5 mM EDTA in 50 mM Tris-HCl/150 mM NaCl buffer. After blocking with 3% BSA-PBS, cells were stained with Abs against anti-GFP (598 or RQ2; MBL) and anti-p65 (A or F-6; Santa Cruz). Nuclear DNA was stained with DAPI (DAPI; Sigma). The specimens were examined using an FV1200 (Olympus) confocal microscope.

GraphPad Prism 6 was used for statistical analyses. The means of two groups were compared by the unpaired Student t test. The Mann–Whitney U test was used to compare nonparametric datasets. The p values <0.05 were considered statistically significant.

To assess the impact of FSC-specific inactivation of the canonical NF-κB pathway in vivo, we first generated a mouse strain bearing the transgene, CAG promoter, and a bicistronic expression cassette encoding IκBSR-ires-EGFP interrupted with a transcription termination signal held between two loxP sites, which was knocked in to the Rosa26 locus (R26-stop^flox-IκBSR) (Fig. 1A). This strain was further crossed with Ccl19-Cre BAC transgenic mouse to obtain animals that had both transgenic alleles (Ccl19-IκBSR) in which constitutive IκBSR expression was expectedly induced in SLO FSCs that were expressing or had once expressed the Ccl19 gene. We hypothesized that SLOs may disappear in such mice if the activity of the canonical NF-κB pathway in mesenchymal stromal cells, especially in fetal LTo cells, was essential for SLO development. However, virtually all LNs developed in Ccl19-IκBSR mice, although they were apparently hypoplastic with smaller organ sizes and morphological abnormalities compared with their appearance in control animals (R26-stop^flox-IκBSR). In fact, the total cell number was markedly lower in LNs from Ccl19-IκBSR mice (Fig. 1B, 1E). The size and morphology of the spleen were comparable to those in control mice; however, the total spleen cell number was significantly reduced in Ccl19-IκBSR animals (Fig. 1C, 1F). It was also evident that the development of PPs was incomplete, although the numbers of PPs in the small intestine and of the follicles in a single PP were normal (Fig. 1D, 1G, 1H).

To determine the cause of these morphological abnormalities of SLOs in Ccl19-IκBSR mice, we next examined tissue structures in greater detail by fluorescence immunohistochemistry. The most remarkable alteration in each SLO was the defective formation of the T cell area (Fig. 2). In the paracortex of LNs, the accumulation of T cells was dramatically reduced in accordance with the diminished T cell proportion in total LN cells compared with that in controls (Fig. 2A, 2D). In contrast, the number of B cells was proportionally higher, and follicular B cell accumulations were clearly formed in the outer cortex, albeit with somewhat disordered morphology. PNAd⁺ high endothelial venules (HEVs) were present, and the expansion of the medullary sinus was evident in Ccl19-IκBSR LNs. In the spleen, T cell localization, especially in the periarteriolar lymphoid sheath, the T cell area in the white pulp, was severely affected in Ccl19-IκBSR mice in sharp contrast to the follicle-like B cell accumulations (Fig. 2B). The T cell–B cell ratio in the whole spleen was unaltered likely due to the mislocalization of T cells mainly in the red pulp instead of the white pulp (Fig. 2B, 2E). The interfollicular T cell area was nearly absent in PPs of Ccl19-IκBSR mice, whereas formed follicles were smaller but relatively substantial compared with those in controls (Fig. 2C). Follicular B cell area and CD35⁺ FDC networks were clearly formed in all SLOs of Ccl19-IκBSR mice (Supplemental Fig. 1A). RANKL⁺MAdCAM-1⁺ MRCs were also detected in the outer margin of follicles beneath the subcapsular sinus in LNs (Supplemental Fig. 1B). Thus, in contrast to T cell area, follicular structures and associated FSC subsets developed appropriately in the Ccl19-IκBSR environment.

Given the defective formation of the T cell area in SLOs from Ccl19-IκBSR mice, the conditional inhibition of NF-κB activation was assumed to affect the subset of FSCs in this area. Thus, we next checked the expression of CCL21, a typical hallmark of TRCs. In normal LN sections, an anti-CCL21 Ab strongly stained reticular TRCs as well as endothelial cells in HEVs. In contrast, the CCL21 signal was dramatically reduced in Ccl19-IκBSR LNs; only faint expression was detected in HEVs (Fig. 3A). CCL21 expression was nearly absent in the white pulp of the spleen as well (Fig. 3B). CXCL13 expression in Ccl19-IκBSR LNs was comparable to that in control, whereas follicular stromal cells in the spleen of Ccl19-IκBSR animals showed substantial but somewhat disarranged pattern of CXCL13 expression (Fig. 3A, 3B). Quantitative PCR analysis demonstrated dramatically diminished expression of the Ccl19 and Ccl21a genes in Ccl19-IκBSR LNs, whereas Cxcl13 expression showed an inversely correlated increase, probably due to the relatively augmented proportion of CXCL13-expressing cells in the total population of LN cells (Fig. 3C). Likewise, expression of CCL19 and CCL21a was nearly abolished in Ccl19-IκBSR spleen, whereas CXCL13 expression was only slightly decreased compared with that in control (Fig. 3D), which was consistent with immunohistochemical observations. Accordingly, these findings indicate that conditional inactivation of the canonical NF-κB pathway under the control of Ccl19 expression led to a selective loss of TRCs in SLOs.

To determine if the deficiency of TRC and decrement of lymphocytes of Ccl19-IκBSR mice were also observed in developing LNs, we examined tissues from neonates (1 wk) and young (2 and 4 wk of age) mice. We found that the total LN number, especially in 2 and 4 wk after birth, was markedly reduced in Ccl19-IκBSR mice compared with that in control animals, whereas no significant difference was observed in the numbers of neonatal LNs in mutant and control animals (Supplemental Fig. 2A). CCL21⁺ FSCs were already virtually absent in neonatal as well as young Ccl19-IκBSR mice, whereas CCL21-producing stromal network was clearly formed in T cell area of neonatal LNs in control mice (Supplemental Fig. 2B). Therefore, these observations suggest that the differentiation of TRCs is inhibited in the early stages of development, most likely immediately after the expression of Ccl19-Cre, which induces IκBSR and blocks NF-κB signaling.

Although the expression of Ccl19 is a typical feature of TRCs, LTo cells in fetal anlagen, from which all adult type FSC subsets are supposed to differentiate, also expressed this gene, and, in fact, Ccl19-Cre has shown functional expression from this stage (33). Therefore, it was puzzling that the conditional inactivation of NF-κB under the control of Ccl19-Cre showed a selective effect on TRCs. In particular, in a clear contradiction to this finding, the development of FDC network has been suggested to require NF-κB pathways (9, 31). Most FSCs, including follicular subsets, actually displayed EYFP expression in Ccl19-Cre/R26-stop^flox-EYFP (Ccl19-EYFP) simple fate mapping system (Fig. 4A–C) (30), demonstrating that MRCs and FDCs, in addition to TRCs, had once expressed Ccl19 gene during development. In adult Ccl19-Cre mice, Cre-expressing cells were in fact detected in the paracortex but not in the follicles (Supplemental Fig. 3A). These observations support the notion that the Ccl19 gene is transiently expressed in the precursor of follicular FSCs and subsequently becomes downregulated during the course of their differentiation.

To identify IκBSR-expressing cells in SLOs of Ccl19-IκBSR mice, we took advantage of detecting bicistronic EGFP expression. Interestingly, EGFP⁺ cells were detected only in a small fraction of mesenchymal cells in the perifollicular and medullary regions, especially in the perivascular area, of LNs (Fig. 4A, 4B arrows), whereas the majority of FSCs were devoid of EGFP expression. Flow cytometry analysis of enzymatically digested LNs also supported that observation, as only a limited cell population (<10%) showed EGFP signal in CD45⁻CD31⁻podoplanin⁺ stromal fraction (Fig. 4C). Notably, EGFP was almost undetectable in follicular stromal cells, showing that these cells did not express (and, in fact, had never expressed) Ccl19-Cre (i.e., IκBSR). Likewise, fluorescence signal was absent in follicle-like B cell accumulations in the spleen, whereas a fraction of stromal cells in perivascular regions outside the follicles and in the red pulp exhibited EGFP expression (Supplemental Fig. 3B). We confirmed that cultured primary FSCs expressing EGFP from Ccl19-IκBSR mice did not exhibit the nuclear translocation of RelA(p65) upon TNF-α stimulation, showing that IκBSR in fact blocked the activation of canonical NF-κB pathway in the target cells (Supplemental Fig. 4). Taken together, TRCs were essentially absent, whereas no MRCs and FDCs expressing IκBSR were present in Ccl19-IκBSR mice.

Detailed observations of tissue structures in Ccl19-IκBSR mice SLOs in this study suggest that the differentiation or maintenance of all FSC subsets, namely TRCs, MRCs, and FDCs, requires canonical NF-κB activation. To explain the outcome of conditional NF-κB inactivation in FSCs driven by Ccl19-Cre, we propose a speculative model of the developmental program of FSC subsets in normal and IκBSR-expressing condition (Fig. 4D). As Ccl19 expression is a hallmark of TRCs, it is a reasonable consequence that Ccl19-driven induction of IκBSR straightforwardly suppressed TRC differentiation through the inhibition of the canonical NF-κB pathway. In normal conditions, MRCs and FDCs are likely to exhibit transient expression of Ccl19 at progenitor level in fetal anlagen or postnatal period, which would be the reason why most follicular FSCs were marked with EYFP expression in Ccl19-EYFP fate mapping system. In Ccl19-IκBSR mice, in contrast, transient Ccl19-Cre expression and subsequent NF-κB inactivation in FSC progenitors likely suppressed the development of MRCs (and following FDCs), as virtually none of these cells expressed EGFP (and IκBSR). IκBSR⁻ follicular stromal cells present in adult Ccl19-IκBSR SLOs likely derived from a small number of CCL19⁻ cells that had circumvented transient CCL19 expression but had a potential to substitute the follicular area. Such a compensatory development of follicles in Ccl19-IκBSR SLOs resulted in the similarity of the whole tissue structure to that of plt mice that lack T cell area because of the absence of Ccl19 and Ccl21a genes (37). The alternative differentiation pathway could possibly take place in normal setting, as EYFP⁻ FSCs were constantly detected in the LNs of adult Ccl19-EYFP mice. EGFP⁺ cells still present in Ccl19-IκBSR SLOs were possibly 1) unusual cells that could not differentiate to mature FSCs, 2) FSC subsets that did not require NF-κB activity for their differentiation, or 3) cells preserving undifferentiated state (equivalent to that of progenitors) in adult SLOs. Transient expression of Ccl19 gene and the existence of EYFP⁻ FSCs suggest the presence of a stochastic state in terms of chemokine expression in FSC progenitor cells. During that period, direct contact with lymphoid cells and/or other hematopoietic cells might fix the characteristics of FSCs to a particular subset and facilitate the localization of hematopoietic cells to construct tissue subcompartments in a manner of a positive feedback loop or self-organization.

The conclusion that all three FSC subsets require the activation of the NF-κB canonical pathway in their differentiation was drawn from the conditional induction of NF-κB suppressor coupled with a fate mapping system that visualized targeted cells. However, because the expression of noncanonical NF-κB components such as Relb and Nfkb2 depend on canonical NF-κB activity to some extent and IκBSR is in fact known to inhibit both pathways (13–15), the phenotypes observed in Ccl19-IκBSR mice are considered to be a consequence of whole NF-κB suppression in FSCs. In addition, potential limitations or uncertainties are inevitable in a conditional gene induction system in vivo. The expression levels of fluorescent reporter proteins induced by cell type–specific Cre recombinases do not always correlate with the corresponding deletion/recombination of another gene of interest. In our system, IκBSR may not express efficiently in any of the CCL19⁺ FSCs or not block NF-κB activation completely. These are the future issues that should be verified by using complementary approaches. Furthermore, the conditional ablation of genes that are essential for the differentiation and maintenance of cells that use Ccl19-Cre system need a careful consideration because the compensatory substitution of FSCs could derive from CCL19⁻ progenitors. Consequently, we cannot yet exclude a possibility that NF-κB activity in LTo cells or FSC progenitors is essential for SLO development, as alternatively differentiated FSCs with intact NF-κB signaling could promote early organogenesis in Ccl19-IκBSR mice.

We thank Changchun Xiao for CTV targeting vector.

The authors have no financial conflicts of interest.