Lymph node formation and B cell homeostasis require IKK-α in distinct endothelial cell–derived compartments

Significance

Noncanonical NF-κB controls lymph node (LN) formation and B cell homeostasis. We previously demonstrated that IKK-α–dependent noncanonical NF-κB signaling is activated in vascular endothelial cells (ECs). Here, we find that ablation of IKK-α in ECs leads to complete loss of LNs and markedly reduced B cell numbers, recapitulating the phenotype of global IKK-α inactivation. Using cell type–specific conditional knockout models, we find that loss of IKK-α in hematopoietic cells underlies the B cell defect, whereas deletion in lymphatic ECs (LECs) results in the absence of LNs. Thus, our findings reveal that IKK-α in distinct EC-derived compartments is uniquely required to promote B cell homeostasis and LN development and demonstrate that LEC-intrinsic IKK-α is essential for LN formation.

Abstract

Global inactivation of IκB kinase (IKK)-α results in defective lymph node (LN) formation and B cell maturation, and loss of IKK-α–dependent noncanonical NF-κB signaling in stromal organizer and hematopoietic cells is thought to underlie these distinct defects. We previously demonstrated that this pathway is also activated in vascular endothelial cells (ECs). To determine the physiologic function of EC-intrinsic IKK-α, we crossed Ikkα^F/F mice with Tie2-cre or Cdh5-cre mice to ablate IKK-α in ECs. Notably, the compound defects of global IKK-α inactivation were recapitulated in Ikkα^Tie2 and Ikkα^Cdh5 mice, as both lacked all LNs and mature follicular and marginal zone B cell numbers were markedly reduced. However, as Tie2-cre and Cdh5-cre are expressed in all ECs, including blood forming hemogenic ECs, IKK-α was also absent in hematopoietic cells (HC). To determine if loss of HC-intrinsic IKK-α affected LN development, we generated Ikkα^Vav mice lacking IKK-α in only the hematopoietic compartment. While mature B cell numbers were significantly reduced in Ikkα^Vav mice, LN formation was intact. As lymphatic vessels also arise during development from blood ECs, we generated Ikkα^Lyve1 mice lacking IKK-α in lymphatic ECs (LECs) to determine if IKK-α in lymphatic vessels impacts LN development. Strikingly, while mature B cell numbers were normal, LNs were completely absent in Ikkα^Lyve1 mice. Thus, our findings reveal that IKK-α in distinct EC-derived compartments is uniquely required to promote B cell homeostasis and LN development, and we establish that LEC-intrinsic IKK-α is absolutely essential for LN formation.

Vascular endothelial cells (ECs) play critical roles in normal immune homeostasis and in the initiation and development of inflammation (1). Activation of the NF-κB family of transcription factors is a major mechanism controlling the homeostatic and inflammatory functions of ECs (1). NF-κB activation occurs via classical and noncanonical pathways, each of which are induced by distinct stimuli and involve separate signaling components (2, 3). Classical NF-κB requires IκB-kinase (IKK)-β−dependent IκB-α degradation, driving the release of p50:p65 NF-κB heterodimers that are crucial for proinflammatory EC activation (1, 4–7). In contrast, noncanonical NF-κB signaling does not require IKK-β but instead involves NF-κB–inducing kinase (NIK) and IKK-α–dependent processing of NF-κB2/p100 to p52 and the nuclear translocation of p52:RelB heterodimers (3). In comparison to classical NF-κB signaling, the physiological role of IKK-α and noncanonical NF-κB in ECs in vivo remains relatively less defined.

The noncanonical NF-κB pathway is activated by a subset of stimuli including lymphotoxin-β receptor (LTβR) ligation by LTα1β2 or LIGHT (TNFSF14) and BAFF-R and CD40 ligation by BAFF and CD40L, respectively. In turn, noncanonical NF-κB drives expression of anti-apoptotic genes and the lymphoid and homeostatic chemokines Ccl19, Ccl21, and Cxcl13 (3). Consistent with the roles of these inducers and targets in lymphoid organogenesis and B cell maturation, loss of function of NIK or IKK-α results in defective lymph node (LN) formation and disrupted B cell homeostasis (8–21). Studies of the immunoregulatory function of noncanonical NF-κB have largely focused on lymphocytes or stromal cells (3, 22); however, we demonstrated that LTβR ligation activates this pathway in blood ECs (BECs) (23, 24), and subsequent studies support homeostatic and pathophysiological functions in ECs (25–31). Recent reports have also described lymphorganogenic functions for NIK and LTβR in lymphatic ECs (LECs) (30); however, the specific role of IKK-α in LECs has not been investigated.

To elucidate the physiologic role of EC-intrinsic IKK-α in vivo, we crossed Ikk-α^F/F mice with Tie2-cre (32) or Cdh5-cre (33) mice. Remarkably, these separate approaches resulted in progeny with reduced numbers of mature B cells and a complete absence of LNs. Tie2 and Cdh5 are expressed in all ECs during development, including the subset of blood-forming hemogenic ECs (HECs) that generate multilineage hematopoietic stem cells (33–37). Consistent with an important role for IKK-α in the hematopoietic cell (HC) compartment for B cell homeostasis, we found that HC-intrinsic loss of IKK-α in Ikk-α^Vav mice significantly impairs B cell homeostasis. In contrast, LN formation was independent of IKK-α in HCs but instead required IKK-α in LECs, as mice lacking LEC-intrinsic IKK-α had no LNs. Our findings reveal that deleting IKK-α in ECs recapitulates the overall compound phenotype of global disruption of noncanonical NF-κB signaling and that IKK-α in distinct EC-derived compartments is required for LN development and B cell homeostasis, with LEC-intrinsic IKK-α being essential for normal LN formation.

Results

Lymph Nodes Are Absent and B Cell Numbers Are Reduced in Mice Lacking IKK-α in ECs.

To delete IKK-α in ECs in vivo, we bred Ikk-α^F/F (38) with Tie2-cre mice (32), and as shown in Fig. 1A, IKK-α was absent in ECs from the Ikk-α^Tie2 mice. As IKK-α and noncanonical NF-κB are activated by stabilization of NIK, we incubated ECs with the Smac mimetic compound GT13072 that stabilizes NIK through inhibition of cIAP1/2 (39). GT13072 induced p100 processing to p52 in Ikk-α^F/F ECs but not in Ikk-α^Tie2 cells, confirming inhibition of noncanonical NF-κB signaling (Fig. 1A). In contrast, loss of IKK-α did not affect classical NF-κB signaling, as TNF-induced IκB-α degradation and resynthesis were intact in both Ikk-α^F/F and Ikk-α^Tie2 ECs (Fig. 1B).

Fig. 1.

Ikk-α^Tie2 mice lack LNs and have decreased numbers of B cells. (A and B) Lung microvascular ECs from Ikk-α^F/F and Ikk-α^Tie2 mice were incubated for the times indicated with either (A) the smac mimetic compound GT13072 (GT; 1 μM) or (B) TNF (10 ng/mL), then cell lysates were immunoblotted using the antibodies indicated. (C) Ikk-α^Tie2 and Ikk-α^F/F mice were injected with Evans blue dye into rear foot pads to identify LNs. The normal positions of axillary (ALN), inguinal (ILN), iliac, and mesenteric (MLN) LNs are indicated by the circles or arrows. (D) PPs are shown along with a plot of the number of PPs counted per small intestine from groups of Ikk-α^F/F and Ikk-α^Tie2 mice (n = 10). (E) Representative flow cytometry plots of splenocytes showing the percentage of B cells (CD19⁺) and T cells (CD3⁺) within the singlet live lymphocyte gate. (F) Total cell numbers of T and B cells calculated using the percentages (E) from spleens of Ikk-α^F/F and Ikk-α^Tie2 mice (n ≥ 5). **P < 0.01, ****P < 0.0001. ns, not significant.

Gross macroscopic examination of secondary lymphoid organs revealed that Ikk-α^Tie2 mice lack all detectable LNs (Fig. 1C). Accumulation of Evan’s blue dye following footpad injection readily identified LNs in control mice; however, no peripheral or mesenteric LNs were detected in any Ikk-α^Tie2 mice examined. Although Peyer’s patches (PPs) were present in Ikk-α^Tie2 mice, they were smaller and reduced in frequency compared with Ikk-α^F/F littermates (Fig. 1D). In contrast to the lack of LNs, the spleens of Ikk-α^Tie2 mice were normal in size, weight, and overall splenocyte numbers (SI Appendix, Fig. S1 A–C). Cellularity analysis showed increased frequency of CD3⁺ T cells in Ikk-α^Tie2 mice but no significant effect on total T cell numbers (Fig. 1 E and F). However, total CD19⁺ B cell percentages and numbers were markedly reduced in Ikk-α^Tie2 mice. Thus Tie2-cre–driven deletion of IKK-α results in the complete absence of LNs and a significant reduction in the number and frequency of splenic B cells.

To verify that these defects arise from loss of IKK-α specifically in the EC lineage, we utilized Cdh5-cre mice, a separate well-established model for conditional deletion in ECs (33). IKK-α was absent in ECs from Ikk-α^Cdh5 (Cdh5-cre; Ikk-α^F/F) mice (Fig. 2A), and as we found in the Ikk-α^Tie2 mice, Ikk-α^Cdh5 spleens were normal in size, weight, and cell numbers (SI Appendix, Fig. S1 D–F), whereas all LNs were missing and PPs were smaller and significantly reduced in frequency (Fig. 2 B and C). The frequency and numbers of splenic T cells in Ikk-α^Cdh5 mice were increased; however, B cell numbers were significantly reduced compared with Ikk-α^F/F littermates (Fig. 2 D and E). Thus, the overall LN and B cell phenotype of Ikk-α^Tie2 mice is recapitulated in Ikk-α^Cdh5 mice. Moreover, these overt defects are remarkably similar to the well-described compound phenotype of mice in which key components of the noncanonical NF-κB pathway, including NIK and IKK-α, are globally deleted or inactive (8–20).

Fig. 2.

LN formation is defective and B cell numbers are reduced in Ikk-α^Cdh5 mice. (A) Lysates of lung microvascular ECs from Ikk-α^F/F and Ikk-α^Cdh5 mice were immunoblotted using the antibodies indicated. (B) Ikk-α^F/F and Ikk-α^Cdh5 mice were injected in the rear foot pad with Evans blue dye to identify LNs. The normal positions of axillary (ALN) and brachial (BLN) LNs are indicated by the circles or arrows. (C) Representative PPs are shown together with the number of PPs counted per small intestine in groups of Ikk-α^F/F and Ikk-α^Cdh5 mice (n = 5). (D) Flow cytometry plots of splenocytes stained with the indicated antibodies. The percentages of B cells (CD19⁺) and T cells (CD3⁺) within the singlet live lymphocyte gate are shown. (E) Total numbers of T and B cells from groups of Ikk-α^F/F and Ikk-α^Cdh5 mice (n = 6). (F and G) Splenocytes and thymocytes from Ikk-α^F/F and Ikk-α^Cdh5 (F) or and Ikk-α^Tie2 (G) mice were immunoblotted using the antibodies indicated. (H and I) Splenocytes from Ikk-α^F/F and Ikk-α^Tie2 mice were incubated for the times indicated with either (H) GT13072 (GT; 1 μM) or (I) PMA (500 ng/mL) and ionomycin (100 ng/mL) (P/I), and then cell lysates were immunoblotted using the antibodies indicated. *P < 0.05, ***P < 0.001, ****P < 0.0001.

IKK-α Is Absent in HCs in Ikk-α^Tie2 and Ikk-α^Cdh5 Mice.

Our results suggest that the developmental requirement for IKK-α in normal LN formation and B cell homeostasis is limited to Tie2⁺Cdh5⁺ cells. Lineage-tracing studies have established that multilineage hematopoietic stem cell (HSC)/progenitor cells emerge during development from a small subset of ECs named HECs (33–37). Notably, Tie2 and Cdh5 are expressed by all ECs, including HECs, and this is maintained throughout development of both the EC and HC compartments (33, 34, 40). To begin to address how conditional Tie2-cre– and Cdh5-cre–driven IKK-α deletion recapitulates the composite defects in mice lacking functional IKK-α or NIK (8–20), we asked whether IKK-α expression was compromised in HCs in addition to ECs. IKK-α was absent in splenocytes and thymocytes from Ikk-α^Cdh5 and Ikk-α^Tie2 mice (Fig. 2 F and G), demonstrating that Tie2-cre– and Cdh5-cre–driven deletion occurs in both ECs and HCs. Incubation with GT13072 induced p100 processing to p52 in Ikk-α^F/F but not Ikk-α^Tie2 splenocytes, confirming inhibition of noncanonical NF-κB activation (Fig. 2H). In contrast, phorbol 12-myristate 13-acetate (PMA) and ionomycin (P/I)-induced IκB-α degradation was normal in Ikk-α^Tie2 splenocytes showing that classical NF-κB signaling was intact (Fig. 2I).

Loss of HC-Intrinsic IKK-α Does Not Affect LN Formation but Impairs B Cell Development.

As IKK-α is absent from both HCs and ECs in Ikk-α^Tie2 and Ikk-α^Cdh5 mice, we asked whether defective LN and B cell development results from this compound loss of IKK-α or whether these defects arise due to separate requirements for IKK-α in distinct cellular compartments. To address this, we crossed Ikk-α^F/F and HSC-specific Vav-cre mice (41) to generate Ikk-α^Vav mice. In these mice, IKK-α was absent in HCs, including splenocytes and thymocytes (Fig. 3A), but was intact in ECs (SI Appendix, Fig. S2A), confirming that deletion was restricted to HCs. Consistent with loss of IKK-α, basal and GT13072-induced p100 processing to p52 was defective in Ikk-α^Vav splenocytes, while P/I-induced IκB-α degradation was normal (SI Appendix, Fig. S2 B and C).

Fig. 3.

LNs are intact but B cell numbers are reduced in Ikk-α^Vav mice. (A) Lysates of splenocytes and thymocytes from Ikk-α^F/F and Ikk-α^Vav mice were immunoblotted using the antibodies indicated. (B) Ikk-α^F/F and Ikk-α^Vav mice were injected in rear foot pads with Evans blue dye to identify LNs. The normal positions of popliteal (PLN), axillary (ALN), and mesenteric (MLN) LNs are indicated by the circles or arrows. The two right panels show representative PPs. (C) LN index of Ikk-α^F/F, Ikk-α^Tie2, Ikk-α^Cdh5, and Ikk-α^Vav mice. The values indicate the presence or absence of brachial, axillary, superficial cervical, deep cervical, MLN, inguinal, and PLN LNs. (D) The number of PPs was determined in groups of Ikk-α^F/F and Ikk-α^Vav mice (n = 8). (E) Representative cytometry plots of splenocytes showing the percentage of B cells (CD19⁺) and T cells (CD3⁺) within the singlet live lymphocyte gate. (F) Total numbers of T and B cells from groups of Ikk-α^F/F and Ikk-α^Vav mice (n ≥ 5). (G and H) Sections of brachial LNs from Ikk-α^F/F and Ikk-α^Vav mice were stained with antibodies against CD3, B220, CD31, LYVE-1, and VEGFR3 (Scale bar, 50 μm.) (G) or PNAd (H). In H, PNAd+ HEV are shown at 20× (Top) (Scale bar, 30 μm) and 63× (Bottom) (Scale bar, 10 μm) magnification. *P < 0.05, **P < 0.01. ns, not significant.

Unlike the Ikk-α^Cdh5 and Ikk-α^Tie2 mice, all LNs were present in Ikk-α^Vav mice, and these were normal in size, number, and location (Fig. 3 B and C). Ikk-α^Vav spleens were also of normal size and weight, and total splenocyte numbers were not significantly different from littermate controls (SI Appendix, Fig. S2 D and E). PPs in Ikk-α^Vav mice were smaller and less distinct than the Ikk-α^F/F controls (Fig. 3B) and resembled the diminutive PPs in Ikk-α^Tie2 and Ikk-α^Cdh5 mice (Figs. 1D and 2C). However, unlike the Ikk-α^Tie2 and Ikk-α^Cdh5 mice, the frequency of PPs in Ikk-α^Vav mice was not markedly reduced (Fig. 3D). LN formation requires the activity of CD3⁻CD4⁺IL-7Rα⁺ lymphoid tissue inducer (LTi) cells (42), and we found that LTi numbers in the spleens of the LN-deficient Ikk-α^Tie2 and Ikk-α^Cdh5 mice, and in the Ikk-α^Vav mice that have intact LNs (SI Appendix, Fig. S3 A and B), were consistently increased. Thus, inactivation of HC-intrinsic IKK-α does not compromise the frequency of LTi cells required for LN development. Importantly, these data establish that IKK-α in the hematopoietic compartment is entirely dispensable for LN development.

Inactivation of HC-intrinsic IKK-α did not affect total splenic CD3⁺ T cell numbers (Fig. 3 E and F), and we found no overt defects in CD4⁺ or CD8⁺ numbers in the spleen or thymus of Ikk-α^Tie2, Ikk-α^Cdh5, or Ikk-α^Vav mice (SI Appendix, Fig. S4). In contrast, Ikk-α^Vav mice had markedly fewer CD19⁺ B cells than did control littermates (Fig. 3 E and F), demonstrating that IKK-α deletion in HCs replicates the reduced B cell numbers in Ikk-α^Tie2 and Ikk-α^Cdh5 mice (Figs. 1 and 2). Consistent with reduced numbers of splenic B cells, the LNs in Ikk-α^Vav mice contained fewer B cells, and the number of B cell follicles was diminished (Fig. 3G). Despite this loss of B cells, the overall LN structure in Ikk-α^Vav mice was intact with no overall changes in the extent or distribution of CD31⁺ vessels or the LYVE-1⁺ and VEGFR3⁺ lymphatic network. Furthermore, PNAd⁺ high endothelial venules (HEV) were intact and displayed the normal plump, cuboidal morphology in Ikk-α^Vav mice, indicating that loss of HC-intrinsic IKK-α does not prevent development of the LN HEV network (Fig. 3H). These collective findings, therefore, establish that loss of IKK-α in the HC compartment underlies the reduced B cell numbers in Ikk-α^Tie2 and Ikk-α^Cdh5 mice, whereas defective LN development is a specific consequence of IKK-α deletion in ECs.

B Cell Homeostasis Is Defective in Ikk-α^Tie2 and Ikk-α^Cdh5 Mice.

To define the nature of the B cell defect in Ikk-α^Tie2 and Ikk-α^Cdh5 mice, we sought to determine the stage of B cell development impacted by deletion of EC-intrinsic IKK-α. Cell-intrinsic BAFF-R–induced noncanonical NF-κB is essential for mature B cell survival (13–21). However, defective early B cell development has been reported in mice in which IKK-α is globally inactivated (8). We therefore asked whether loss of IKK-α in ECs contributes to the reduced B cell numbers in Ikk-α^Tie2 and Ikk-α^Cdh5 mice or whether these defects reflect the previously defined role for noncanonical NF-κB signaling in mature B cell homeostasis. Cytometric analysis of bone marrow (BM) revealed no significant differences in the total number of BM cells or the B cell progenitor subsets in Ikk-α^Tie2, Ikk-α^Cdh5, or Ikk-α^Vav mice compared with littermate controls (SI Appendix, Fig. S5 A–C). In contrast, mature recirculating B cells in the BM were significantly reduced in all three mice (SI Appendix, Fig. S5 B and C). We conclude from these data that EC-intrinsic IKK-α does not control early B cell development and that the reduced numbers of recirculating and splenic B cells in Ikk-α^Tie2, Ikk-α^Cdh5, and Ikk-α^Vav mice are due to a defect at a later stage.

Immature B cells egress from the BM and migrate to the spleen, where they differentiate through transitional (T) stages into mature follicular (FO) or marginal zone (MZ) B cells (43). As loss of IKK-α did not affect immature subpopulations in the BM, we assessed the spleens of Ikk-α^Tie2, Ikk-α^Cdh5, and Ikk-α^Vav mice to determine the stage at which IKK-α deletion impacts peripheral B cell development (SI Appendix, Fig. S6). The numbers of peripheral cells in the immature/transitional (T1/T2) stages (CD19⁺CD43⁻CD21/35⁻IgM^hi) were not significantly different from controls in the Ikk-α^Tie2, Ikk-α^Cdh5, and Ikk-α^Vav mice (Fig. 4A). In contrast, the numbers of FO (CD19⁺CD43⁻B220⁺AA4⁻CD23⁺CD21/35⁻IgM^lo) and MZ (CD19⁺CD43⁻B220⁺AA4⁻CD23⁻CD21/35⁺IgM^hi) B cells were markedly reduced in all three mice (Fig. 4A). As this indicates that the blockade in development occurs during the transitional stages, we analyzed IgM and CD23 cell expression on B220⁺AA4⁺ immature splenic B cells. Examination of the transitional subsets revealed a significant increase in the percentage of cells within the B220⁺AA4⁺CD23⁻IgM^hi T1 stage, resulting in an almost twofold drop in the frequency of cells within the B220⁺AA4⁺CD23⁺IgM^lo transitional 3 (T3) subset in all three mice (Fig. 4B). Thus, mice lacking IKK-α in all ECs, including HECs, or in HCs alone exhibit the same developmental block in transitional stages within the immature peripheral B cell compartment that has been described for mice lacking B cell–intrinsic IKK-α or NIK or functional BAFF-R signaling (13–21).

Fig. 4.

B cell maturation and MZ cellularity is defective in Ikk-α^Tie2 and Ikk-α^Cdh5 mice. (A) Total numbers of splenic B cell populations in groups of Ikk-α^Tie2, Ikk-α^Cdh5, Ikk-α^Vav, and Ikk-α^F/F littermate control mice (n ≥ 5) were obtained by flow cytometry following pregating on live singlet cells. Immature/transitional T1/T2 (CD21/35⁻IgM^hi) B cells together with mature B cells separated into FO (CD19⁺CD43⁻B220⁺AA4⁻CD23⁺CD21/35⁻) and MZ (CD19⁺CD43⁻B220⁺AA4⁻CD23⁻CD21/35⁺) populations are shown. (B) Mean percentage of the separate transitional subpopulations of B220⁺AA4⁺ immature B cells subdivided into T1 (CD23⁻IgM^hi), T2 (CD23⁺IgM^hi), and T3 (CD23⁺IgM^lo) for the Ikk-α^Tie2, Ikk-α^Cdh5, Ikk-α^Vav, and Ikk-α^F/F littermate control mice (n ≥ 5). (C–E) Sections of spleens from the mice indicated were stained using antibodies against CD3 and B220 (C), IgM and IgD (D), MOMA-1, SIGN-R1, and MAdCAM-1 (E), and representative areas of white pulp are shown. (Scale bar, 50 μm.). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant.

Loss of MZ Cellularity in Ikk-α^Tie2 and Ikk-α^Cdh5 Mice.

Splenic organization is defective in mice lacking components of the TNF and LTβR signaling pathways, including LTβR, TNFR-1, NIK, RelB, and p50 and p52 (8, 12–14, 44–48). To determine if deletion of EC- or HC-intrinsic IKK-α affects splenic architecture and cellularity, we analyzed spleens from Ikk-α^Tie2, Ikk-α^Cdh5, and Ikk-α^Vav mice. The structure of white pulp was largely intact in all three mice, including separate B- and T-cell zones (Fig. 4C), and the presence of a normal fibroblastic reticular network and compartmentalized CXCL13 and CCL21 expression in the B cell follicles and T cell–rich periarteriolar lymphoid sheath (PALS), respectively (SI Appendix, Fig. S7). However, consistent with our cytometric analyses (Figs. 1–3), the B cell follicles in Ikk-α^Tie2, Ikk-α^Cdh5, and Ikk-α^Vav mice contained fewer B220⁺ cells (Fig. 4C and SI Appendix, Fig. S7). Moreover, IgD^hiIgM^lo FO B cells were less abundant in Ikk-α^Tie2, Ikk-α^Cdh5, and Ikk-α^Vav mice, and the IgD^loIgM^hi MZ B cell compartment was severely diminished compared with controls (Fig. 4D). As loss of MZ integrity is a major defect associated with global deletion of noncanonical NF-κB signaling (8, 12–14, 44–48), we examined the MZ cellularity and structure in all of the mice. As shown in Fig. 4E, MOMA-1⁺ marginal metallophilic macrophages (MMMs) and SIGN-R1⁺ marginal zone macrophages (MZMs) were completely absent in Ikk-α^Tie2 and Ikk-α^Vav mice. Notably, MMMs and MZMs remained detectable in Ikk-α^Cdh5 mice, although both populations were markedly reduced. Contrasting the effects of IKK-α loss on these distinct HC populations, MAdCAM-1 expression by marginal sinus lining cells was intact in Ikk-α^Tie2, Ikk-α^Cdh5, and Ikk-α^Vav mice (Fig. 4E). Consequently, EC-intrinsic IKK-α does not regulate MAdCAM-1 expression in the marginal sinus; however, normal MZ cellularity requires IKK-α in the HC compartment.

Lymphatic EC-Intrinsic IKK-α Controls LN Formation.

Our data establish that Tie2-cre– or Cdh5-cre–driven deletion of Ikk-α results in B cell and LN defects similar to mice lacking global noncanonical NF-κB signaling (8–20). The B cell deficiency is caused by HC-intrinsic loss of IKK-α, whereas defective LN formation is due to IKK-α deletion in the Tie2⁺Cdh5⁺ EC compartment. In addition to early hematopoiesis, lymphatic vessels also form by budding from the blood vascular endothelium during development (49). To determine if LEC-intrinsic IKK-α plays a role in LN formation, we crossed Ikk-α^F/F and Lyve1-cre mice (50) to generate Ikk-α^Lyve1 mice. As shown in Fig. 5A, qRT-PCR analysis confirmed that Ikk-α expression was absent in LECs in Ikk-α^Lyve1 mice. Notably, however, none of the Ikk-α^Lyve1 mice or any of the other lines we generated exhibited lymphatic defects such as edema or disrupted developmental lymphangiogenesis. In this regard, the overall frequency of CD31⁺podoplanin⁺ LECs and CD31⁺podoplanin⁻ BECs in the skin of Ikk-α^Lyve1 mice (Fig. 5 B and C and SI Appendix, Fig. S8) was normal. Similarly, Ikk-α^Tie2 mice displayed normal LEC and BEC frequency in the skin (SI Appendix, Fig. S8).

Fig. 5.

LN formation is defective but B cell numbers and splenic cellularity are normal in Ikk-α^Lyve1 mice. (A) CD31⁺Podoplanin⁺ LECs (Left) and CD31⁻Podoplanin⁻ stromal cells (Right) were sorted from single-cell lung homogenates, and qRT-PCR analysis was performed to determine the relative expression of Ikk-α. (B) Sections of ears from Ikk-α^F/F and Ikk-α^Lyve1 mice were stained using antibodies against CD31 and LYVE1. Autofluorescence from cartilage is evident (SI Appendix, Fig. S8). (C) The percentage of BECs (CD31⁺Podoplanin⁻) and LECs (CD31⁺Podoplanin⁺) in total singlet live cells in homogenates of skin from Ikk-α^F/F and Ikk-α^Lyve1 mice determined by flow cytometry (n = 4). (D) Ikk-α^F/F and Ikk-α^Lyve1 mice were injected in the rear foot pad with Evans blue dye to identify LNs. The normal positions of axillary (ALN), cervical (CLN) and inguinal (ILN) LNs are indicated by the circles or arrows. (E) LN index of Ikk-α^F/F and Ikk-α^Lyve1 mice. The values indicate the presence or absence of brachial, ALN, superficial CLN, deep CLN, mesenteric, ILN, and popliteal LNs. (F) PPs are shown together with a plot of the number of PPs counted per small intestine from groups of Ikk-α^F/F and Ikk-α^Lyve1 mice (n = 8). (G) Total cell numbers of B cells (CD19⁺) and T cells (CD3⁺) within the singlet live lymphocyte gate determined by flow cytometry of splenocytes from Ikk-α^F/F and Ikk-α^Lyve1 mice (n ≥ 6). (H) Total numbers of splenic B cell populations in groups of Ikk-α^F/F and Ikk-α^Lyve1 mice (n ≥ 6) were obtained by flow cytometry and identified as described in Fig. 4A. (I) Sections of spleens from Ikk-α^F/F and Ikk-α^Lyve1 mice were stained using antibodies against CD3 and B220 (Left), IgM and IgD (Middle), and MOMA-1 and SIGN-R1 (Right), and representative areas of white pulp are shown. (Scale bar, 50 μm.). ***P < 0.001. ns, not significant.

Strikingly, as we found in the Ikk-α^Tie2 and Ikk-α^Cdh5 mice (Figs. 1 and 2), all LNs, including the brachial, axillary, superficial cervical, deep cervical, mesenteric, inguinal, and popliteal nodes, were absent in Ikk-α^Lyve1 mice (Fig. 5 D and E). The spleens of Ikk-α^Lyve1 mice were normal in size, weight, and overall cell numbers (SI Appendix, Fig. S9 A–C); however, unlike the Ikk-α^Tie2 and Ikk-α^Cdh5 mice, PPs were normal in size and frequency (Fig. 5F). The numbers and percentages of splenic B cells in Ikk-α^Lyve1 mice were normal (Fig. 5G and SI Appendix, Fig. S9D), and there were no defects in the transitional T1/T2 or mature FO and MZ B cell populations (Fig. 5H). Analysis of splenic cellularity revealed no defects in the B220⁺ B cell and CD3⁺ T-cell zones, the localization of IgD^hiIgM^lo FO and IgD^loIgM^hi MZ B cells, the presence of SIGN-R1⁺ MZMs and MOMA-1⁺ MMMs (Fig. 5I), the integrity of the MAdCAM-1⁺ MZ lining cells (SI Appendix, Fig. S9E), the compartmented CXCL13 and CCL21 expression in the B- and T-cell areas (SI Appendix, Fig. S9F), or the presence of an intact fibroblastic reticular network (SI Appendix, Fig. S9G). These data, therefore, reveal that LEC-intrinsic IKK-α is absolutely required for normal LN formation but has no role in B cell maturation or splenic cellularity that are each defective in Ikk-α^Tie2 and Ikk-α^Cdh5 mice.

Discussion

The major recurring defects in mice lacking components of the noncanonical NF-κB signaling pathway are disrupted LN formation and defective B cell homeostasis (8–21). The peripheral B cell deficiency results from loss of cell-intrinsic BAFF-R ligation-induced noncanonical NF-κB activity that controls B cell survival and maturation (13–21). The precise role of noncanonical NF-κB signaling in LN formation remains elusive; however, the established model is that LTβR-induced noncanonical NF-κB in stromal organizer cells initiates formation of the early LN anlage (51, 52). Remarkably, using independent genetic models to delete IKK-α in ECs during development, we generated mice that replicate the global “noncanonical NF-κB phenotype.” Thus, both Ikk-α^Tie2 and Ikk-α^Cdh5 mice lack all LNs and exhibit defective B cell homeostasis. Consequently, our data reveal that the well-described compound defects in B cells and LNs can be largely accounted for by loss of IKK-α in distinct cellular compartments that originate from early ECs.

Lineage-tracing studies have established that HSCs emerge from blood-forming HECs during development (33–37). In comparing B cell populations in Ikk-α^Tie2 and Ikk-α^Cdh5 mice that lack IKK-α in both ECs and HCs with Ikk-α^Vav mice lacking IKK-α only in HCs, we found identical defects in all three mice. Based on these data and earlier reports of similar defects in mice specifically lacking IKK-α, NIK, or BAFF signaling in B cells (13–21), we conclude that the reduced numbers of mature B cells in Ikk-α^Tie2 and Ikkα−^Cdh5 mice result from loss of HC-intrinsic IKK-α due to its deletion during development in HECs.

Splenic cellularity and organization are also abnormal in mice lacking functional IKK-α or NIK (8, 12–14). Defects include disruption of B cell zones, failure to form germinal centers, and loss of MZ structure and cellularity including defective MAdCAM-1 expression by marginal sinus lining cells. Similar defects also occur in LTβR-, LTα-, LTβ-, RelB-, and NF-κB2–deficient mice (12, 44–48), underscoring the crucial role of LTβR signaling and noncanonical NF-κB in normal splenic organization. Notably, overall splenic architecture was normal in Ikk-α^Tie2 and Ikk-α^Cdh5 mice, including an intact reticular fibroblastic network and normal MAdCAM-1 expression by marginal sinus lining cells; and although the numbers of mature follicular B cells were reduced, B cell zones were present. MZ B cell numbers were significantly reduced in Ikk-α^Vav, Ikk-α^Tie2, and Ikk-α^Cdh5 mice, and this was accompanied by a loss of metallophilic and MZ macrophages. This may reflect B cell–dependent effects on metallophilic and MZ macrophage localization; however, reduction of MZ B cells in mice conditionally lacking NIK in B cells did not affect the frequency of metallophilic or MZ macrophages (14), suggesting a specific macrophage-intrinsic function for IKK-α. Whether IKK-α is required for development of these cells or plays a fundamental role in controlling their migration and residence in the MZ, therefore, remains to be determined.

Our findings demonstrate that HC-intrinsic IKK-α is dispensable for LN formation. Thus, the lack of LNs in Ikk-α^Tie2 and Ikk-α^Cdh5 mice is specifically due to loss of IKK-α in the Tie2⁺Cdh5⁺ EC compartment. LTβR-induced NIK stabilization drives noncanonical NF-κB activation, and an earlier study showed that LNs were present in Cdh5-cre;Ltßr^F/F mice, although these were reduced in size and number and their architecture was disrupted due to defective formation and function of the HEV network (53). Similar defects were also reported in Tie2-cre;Ltßr^F/F mice (29). Recently, Onder et al. described a partial loss of LNs in Cdh5-cre;Nik^F/F mice, whereas LNs were almost entirely absent in mice with Cdh5-cre–driven deletion of both NIK and LTβR (30). We found that neither Ikk-α^Tie2 nor Ikk-α^Cdh5 mice have any LNs, revealing that IKK-α loss in the EC lineage more profoundly impacts LN development than targeting EC-intrinsic LTβR or NIK alone. Thus, our data suggest that IKK-α in Tie2⁺Cdh5⁺ lineage cells may function, at least in part, independently of LTβR and NIK to provide signals necessary for LN formation.

In seeking to identify the Tie2⁺Cdh5⁺ ECs in which IKK-α signaling controls LN formation, we found that LEC-intrinsic IKK-α is essential for LN development. The role of IKK-α in lymphorganogenesis has been implied previously from the well-established two-cell paradigm involving the interaction of LTi cells with stromal lymphoid tissue organizer (LTo) cells (51). In this model, LTis drive early LN anlage formation via ligation of LTβR on LTo cells driving noncanonical NF-κB–dependent expression of lymphorganogenic genes (52). Intriguingly, although LECs were originally considered dispensable for anlage formation (54), recent elegant studies have expanded this model and revealed important roles for LECs and lymphatic interstitial flow in early LN development (30, 55). One new model proposes that LECs themselves function as LTo cells, as conditional deletion of both NIK and LTβR in LECs led to partial loss of LNs (30). However, as deletion of LTβR or NIK alone in LECs did not prevent LN formation (30, 31), our findings reveal a central role for LEC-intrinsic IKK-α that may be separate from LTβR-induced noncanonical NF-κB signaling via NIK. Intriguingly, Onder et al. also demonstrated that conditional deletion of receptor activator of NF-κB (RANK) in LECs led to reduced LN numbers (30). As IKK-α is activated by RANK signaling (56), it is possible that the LEC-intrinsic lymphorganogenic function of IKK-α lies in this pathway or is a function of its role in both LTβR and RANK signaling in LECs during development.

In summary, our findings reveal that normal LN development and B cell homeostasis require IKK-α in separate EC-derived cellular compartments. Importantly, we have identified an absolutely essential role for LEC-intrinsic IKK-α in LN development.

Materials and Methods

Animals.

Tie2-Cre [B6.Cg-Tg(Tek-cre)12Flv/J], Vav-Cre [B6.Cg-Tg(Vav1-icre)A2Kio/J] and Lyve1-cre [B6;129P2-Lyve1^{tm1.1(EGFP/cre)Cys}/J] transgene mice were purchased from The Jackson Laboratory. Cdh5-Cre transgene mice [B6;129-Tg(Cdh5-cre)1Spe/J] were a kind gift from Nancy Speck, University of Pennsylvania, Philadelphia, PA. Ikk-α^F/F mice (129/B6) were previously described (38). Male and female mice between 8 to 14 wk old were used in equal ratios in all experiments. All animal studies were carried out in compliance with the guidelines of the Institutional Animal Care and Use Committee (IACUC) of the University of Pennsylvania and in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH (57). The animal protocol was approved by the IACUC of the University of Pennsylvania.

Materials and Reagents.

Recombinant human TNF and goat anti-cIAP1 antibody (1:1,000; AF8181) were from R&D Systems, anti-p100/52 (1:1,000; 4882) and anti-IκB-α (1:1,000; 4814) were from Cell Signaling Technology, anti–IKK-α (1:500; NB100-56704), anti–IKK-β (1:500; NB100-56509), and anti–β-actin (1:1,000; NB600-501) were from Novus Biologicals, and anti-NEMO (1:5,000; K0159) was from MBL International. Mouse anti-tubulin (1:10000; T5168), phorbal-12-myristate 13-acetate, and ionomycin were from Sigma-Aldrich. The Smac mimetic GT13072 was generously provided by TetraLogic Pharmaceuticals. GT13072 was dissolved in dimethylsulfoxide (DMSO) to a 1-mM stock and used at a final concentration of 1 μM for all experiments. Cells were treated with 0.1% DMSO as a control as we have described previously (39).

Isolation and Culture of Murine Pulmonary ECs.

ECs were isolated by magnetic bead sorting from the lungs of neonatal pups (10 to 14 days postnatal), modified from a previously described procedure (58). Pups were euthanized by ketamine/xylazine, and lungs were filled with collagenase A solution via tracheal injection and further digested in collagenase solution at 37 °C for 30 min. ECs were isolated from single-cell suspension by Dynabeads conjugated with a rat anti-mouse CD31 (Invitrogen) antibody and a Magnetic Particle Concentrator (Invitrogen). Cells were grown in VascuLife VEGF-Mv medium (Lifeline Cell Technology) containing 100 units/mL penicillin and 100 μg/mL streptomycin and cultured on gelatin-coated dishes. ECs were sorted two additional times using Dynabeads conjugated with an anti-CD102 (Invitrogen) antibody. EC purity was determined by flow cytometry using antibodies against the following cell surface antigens: CD45 (1:200; 17-0451-82) and CD105 (1:200; 12-1051-82) (eBioscience) and CD31 (1:400; 562861) and CD102 (1:200; 557444) (BD Biosciences). Cells were run on a BD FACSCalibur cytometer and analyzed using FlowJo software. All stimulations were performed using EC cultures that had a minimum purity of 95% after the third magnetic bead sort.

LEC Isolation.

Single-cell suspensions of mouse lungs were prepared by enzymatic digestion and mechanical dissociation. The digestion buffer contained Collagenase IV (200 U/mL), Dispase (0.6 U/mL), and DNase I (200 U/mL). Erythrocytes were lysed with ACK Lysing Buffer. CD45⁺ cells were depleted using anti-CD45 MicroBeads and LD columns (Miltenyi Biotec), and then single-cell suspensions were stained with LIVE/DEAD Fixable Aqua (Invitrogen) and fluorophore-conjugated antibodies against mouse CD45 (1:200; 17-0451-82) and podoplanin (PDPN) (1:200; 12-5381-82) from eBioscience and CD31 (1:400; 562861) from BD Biosciences. LECs (Aqua⁻, CD45⁻, CD31⁺, PDPN⁺) and stromal cells (Aqua⁻, CD45⁻, CD31⁻, PDPN⁻) were sorted on an FACS Aria II (BD Biosciences).

Real-Time qRT-PCR.

RNA from sorted lung LECs and stromal cells was isolated using the ReliaPrep RNA Miniprep kit (Promega). Total RNA was reverse-transcribed using a High-Capacity complementary (c)DNA Reverse Transcription Kit (Applied Biosystems). qRT-PCR was performed using PerfeCTa SYBR Green SuperMix (QuantaBio) on a QuantStudio 6 Flex System (Applied Biosystems). Expression of Ikk-α was normalized to the expression of β-actin (Actb) or GAPDH (Gapdh). The primers used for qRT-PCR analysis were Actb forward, 5′-GGCTGTATTCCCCTCCATCG-3′; and reverse, 5′- CCAGTTGGTAACAATGCCATGT-3′: Gapdh forward, 5′- CTCCCACTCTTCCACCTTCG-3′; and reverse, 5′- GCCTCTCTTGCTCAGTGTCC-3′: Ikk-α forward, 5′-CAGTATTTGGCCCCAGAGCT-3′; and reverse, 5′- CAGGTAAATGGCTGCAGATGA-3′.

LN Visualization.

Peripheral and mesenteric LNs were visualized by injecting each rear foot pad subcutaneously (s.c.) with 30 μL 1% solution of Evans blue dye in sterile saline 16 h prior to euthanization. The LN index was calculated by dividing the total number of brachial, axillary, superficial cervical, deep cervical, caudal mesenteric, inguinal, and popliteal LNs in the experimental mice by 13, which is the total number of these nodes in a normal mouse (two of each node per mouse except the caudal mesenteric, which is a single node). To examine PPs, the small intestine was excised, cleared of mucus and digesta, and soaked for 15 min in 10% glacial acetic acid. All images were captured on a Nikon SMZ1000 stereomicroscope.

Flow Cytometry.

Lymphocytes were collected from the BM (femur and tibia of one leg), spleen, and thymi of mice in flow buffer (phosphate buffered saline [PBS] containing 0.5% bovine serum albumin [BSA] and 0.2% ethylenediaminetetraacetic acid [EDTA]). Cell counts were performed following red blood cell lysis. Cells were stained using the antibodies to the following antigens for flow cytometry: CD45 (17-0451-82), CD3e (12-0031-82), CD8 (12-0081-82), B220 (11-0452-82), CD93 (AA4.1) (17-5892-82), CD21/CD35 (48-0212-82) and CD127 (12-1271-82) (eBioscience), CD19 (552854), CD4 (553052), IgM (550676), CD23 (561773), and Fc Block (BD Biosciences). All antibodies were used at a dilution of 1:200. Exclusion of DAPI (Invitrogen) was used to identify live cells. Analysis was performed on a BD FACSCANTO II cytometer, and results were analyzed using FlowJo software. The representative plots were obtained following gating for singlets, then further gating for DAPI-negative live cells. For staining endothelial subsets in the skin, ears were digested using Collagenase IV and DNase I and stained with anti-CD31 (1:400; 562861) from BD Biosciences and anti-podoplanin (1:200; 12-5381-82) from eBioscience.

Cell Stimulation and Immunoblotting.

For stimulation, single-cell suspensions of splenocytes were suspended in complete Roswell Park Memorial Institute (RPMI) (containing 10% fetal bovine serum [FBS], 0.1% 2-mercaptoethanol, 1% nonessential amino acids, 1% sodium pyruvate, 100 units/mL penicillin, and 100 μg/mL streptomycin) and were stimulated following a 1-h rest period following isolation. Following stimulation, cells were washed in PBS and lysed in lysis buffer (50 mM Tris-Cl, pH 6.8, 150 mM NaCl and 1% Triton X-100) containing complete protease inhibitors (Roche Applied Science). Protein was separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) and electrophoretically transferred onto a polyvinylidene fluoride (PVDF) membrane (Millipore). Membranes were incubated with the identified primary and species-specific horseradish peroxidase (HRP)-conjugated secondary antibodies and detected by chemiluminescence.

Immunofluorescence Microscopy.

Spleens, LNs, or ears were freshly harvested from young adult mice, embedded in optimal cutting temperature (OCT) compound, and snap frozen in liquid nitrogen. The frozen blocks were cut using a cryostat to 8-μm-thick sections and stored at −80 °C. The sections were fixed using ice-cold acetone for 5 min, then blocked for 1 h at room temperature (RT) using 5% normal horse serum from Abcam. For staining, fluorescent labeled anti-mouse antibodies against B220 (Alexa 647) (1:500; 557669), IgM allophycocyanin (APC) (1:500; 550676) and IgD fluorescein isothiocyanate ( FITC) (1:500; 553439) were purchased from BD Biosciences; anti-CD3 (FITC) (1:500; 11-0031-82), anti-SIGN-R1 (APC) (1:100; 17-2093-82), anti-PNAd (MECA-79; Alexa Flour 488) (1:250; 53-6036-80), and anti-LYVE1 (Alexa Flour 488) (1:100; 53-0443-82) were from eBioscience; and anti-MOMA-1 (FITC) (1:100: MCA947) was from Bio-Rad. All antibodies were diluted in PBS containing 0.1% BSA and incubated for 1 h at RT over the tissue section. Unconjugated anti-CCL21(1:200; AF457) and anti-CD31 (1:200; AF3628) from R&D Systems, anti-CXCL13 (1:200; PA5-47018) from Thermo Fisher Scientific, anti-ER-TR7 (1:250: NB100-64932) from Novus Biologicals, or anti-MAdCAM-1 (MECA-89) (1:100; 553805) from BD Biosciences were incubated overnight at 4 °C followed by the appropriate secondary antibody for 30 min at RT. Stained slides were washed in PBST (0.01% Tween 20 in PBS) and then mounted using ProLong gold anti-fade reagent from Invitrogen. Imaging was performed using a Leica DM6000 wide-field microscope at magnifications of 10×, 20×, or 63× (oil immersion), and images were processed using Volocity or Leica software.

Statistical Analyses.

All experiments were performed at least three times, and statistical analyses were performed using GraphPad Prism 8 software. Data were analyzed with the unpaired Student’s t test. A P value <0.05 was considered significant. (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns = not significant).

Data Availability

All study data are included in the article and/or SI Appendix.