Endothelial cell–specific lymphotoxin-β receptor signaling is critical for lymph node and high endothelial venule formation

LTβR is broadly expressed in various tissues of the developing mouse embryo (Browning and French, 2002), and LTβR deficiency severely impairs the development of peripheral lymphoid tissues (Fütterer et al., 1998). To assess the LTβR expression pattern in major LN stromal cell populations, we separated CD45^– stromal cells by FACS sorting using the well-established markers podoplanin and CD31 (Malhotra et al., 2012) into Pdpn⁻CD31⁺ blood ECs (BECs), Pdpn⁺CD31⁺ lymphatic ECs (LECs), and Pdpn⁻CD31⁺ FRCs (Fig. 1 a). Mesenchymal versus endothelial lineage identity of the sorted cell populations was confirmed by exclusive Wnt1 mRNA expression in the FRC fraction (Fig. 1 b). Ltβr mRNA expression was comparable in the three major LN stromal cell populations (Fig. 1 c), whereas LTβR expression on the surface of both BECs and LECs was significantly higher compared with FRCs (Fig. 1, d and e) suggesting that ligation of the LTβR on ECs may precipitate important functional changes in ECs.

To assess the developmental and functional consequences of LTβR ablation in ECs, we used transgenic VE-cadherin-Cre mice. In these mice, VE-Cadherin-Cre activity starts in the embryonic endothelium at E8, i.e., before the initial steps of LN development (Alva et al., 2006). At E16.5, transgene-expressing ECs co-localized with CD4⁺ LTi cells in the anlage of the inguinal LN (Fig. 2 a). Importantly, transgene activity in developing inguinal LNs (E18.5) of VE-Cadherin-CrexR26-RFP mice was not associated with CD45⁺ cells (Fig. 2 b). Likewise, in neonatal LNs, transgene expression was confined to CD31⁺ ECs (Fig. 2 c). Importantly, ECs in the neonatal LN coexpressed LTβR and the VE-cadherin-Cre transgene, with the highest abundance in the region of the subcapsular sinus (Fig. 2 c, i–iv). Confocal microscopic (Fig. 2 d) and flow cytometric (Fig. 2 e) analysis of adult LNs from VE-cadherin-CrexR26-RFP mice confirmed that ∼70% of LN LECs and BECs express the RFP reporter, indicating that the VE-cadherin-Cre transgene targets a major proportion of LTβR-expressing LN ECs.

To ablate the LTβR on ECs, we crossed the VE-cadherin-Cre transgene onto the Ltβr^fl/fl background (Wimmer et al., 2012). As shown in Fig. 3 a, VE-cadherin-CrexLtβr^fl/fl mice failed to develop 20–45% of their peripheral LNs. Injection of 1% Chicago Sky Blue solution one week before the analysis revealed that the loss of peripheral LNs in VE-cadherin-CrexLtβr^fl/fl mice occurred in a random fashion with the failure to develop even rudimentary structures as shown in Fig. 3, b and c for the brachial or inguinal LN, respectively. Interestingly, formation of the mesenteric LN was not affected by the EC-specific LTβR deficiency (Fig. 3, a and d). Those peripheral LNs that had succeeded to develop in VE-cadherin-CrexLtβr^fl/fl mice appeared smaller, as shown in Fig. 3 e for an inguinal LN. Optical projection tomographical (OPT) analysis of inguinal LNs confirmed the significant size differences between conditionally LTβR-deficient and control LNs (Fig. 3, f and g). As a consequence, the size of B cell follicles (Fig. 3 h), as well as overall cellularity (Fig. 3 i), were reduced. Furthermore, the OPT analysis revealed a profound structural deficiency in the vasculature, i.e., a severely impaired MECA-79⁺ HEV network (Fig. 3 f). Quantification of HEV parameters revealed that the overall length of the MECA-79⁺ vascular structures (Fig. 3 j) and the mean number of branching points (Fig. 3 k) were reduced by >60% and >70%, respectively. The exclusive impact of the conditional LTβR deficiency on ECs is shown by the preserved LTβR expression on FRCs, whereas both BECs and LECs in LNs of VE-cadherin-CrexLtβr^fl/fl mice exhibited a significantly reduced LTβR expression (Fig. 3, l and m). In contrast to the profound effects of the EC-specific LTβR ablation on the LN vasculature, the formation of intranodal lymphatic structures was less severely affected in the absence of EC-specific LTβR signaling (Fig. 3, n and o). Likewise, lymphatic skin drainage toward the popliteal LN and internodal lymph flow from the popliteal to the inguinal LN was not affected in VE-cadherin-CrexLtβr^fl/fl mice (Fig. 3 p). Collectively, these data suggest that the VE-cadherin-Cre transgene targets an early EC population that is LTβR dependent and is critical for LN development. Furthermore, in those adult LNs that succeeded in their development, likely due to the incomplete targeting of ECs in VE-cadherin-Cre mice (i.e., ∼70% of all LN BECs and LECs), a pronounced impact of direct LTβR signaling in ECs was evident.

To further determine the consequences of EC-specific LTβR deficiency, we performed high-resolution confocal microscopy and assessed the phenotypical changes in HEV ECs. Despite the absence of LTβR signaling, ECs had maintained some MECA-79 expression (Fig. 4 a). Furthermore, the ERTR-7⁺ FRC network surrounding the vascular structures was maintained (Fig. 4 a, arrowheads). However, the endothelium had lost its cuboidal shape and polarization, with decreased ICAM-1 staining on the luminal side (Fig. 4 a, arrow) resulting in a significantly reduced overall ICAM-1 expression (Fig. 4 b). Likewise, MECA-79⁺ ECs in the conditionally LTβR-deficient mice had lost expression of CCL21 (Fig. 4 c, arrows, top left), whereas stromal cells surrounding the blood vessel still produced CCL21 as visualized in three-dimensional reconstructions of the paracortical stromal network (Fig. 4 c, right). Consistent with this finding, we found that the selective ablation of LTβR on ECs resulted in a reduction of Ccl21 and Ccl19 mRNA expression in total LN tissue (Fig. 4 d). Furthermore, expression of the HEV addressin Glycam1 was almost completely lost in LNs of VE-cadherin-CrexLtβr^fl/fl mice, whereas the expression of the follicular dendritic cell marker Mfge8 was not affected (Fig. 4 e). Collectively, the MECA-79⁺ blood vessels in VE-cadherin-CrexLtβr^fl/fl mice had lost typical properties of HEVs.

To assess whether the structural and phenotypical alterations in MECA-79⁺ blood vessels in VE-cadherin-CrexLtβr^fl/fl mice would impact on lymphocyte migration, we adoptively transferred dye-labeled lymphocytes and assessed their LN-homing behavior in VE-cadherin-CrexLtβr^fl/fl and control mice. As shown in Fig. 5 a, lymphocytes were still able to home to the LN parenchyma in VE-cadherin-CrexLtβr^fl/fl mice. However, flow cytometry–based quantification revealed that homing of both T and B cells was strongly affected by the EC-specific LTβR deficiency (Fig. 5 b). High-resolution microscopic analysis showed that adoptively transferred lymphocytes could be found within the HEVs of LTβR-competent mice (Fig. 5 c, bottom, arrows), whereas those lymphocytes that had successfully entered LNs of VE-cadherin-CrexLtβr^fl/fl mice were found exclusively in the LN parenchyma (Fig. 5 c, top, arrowheads). A recent study has shown that retention of lymphocytes in HEV pockets is an important regulatory step for lymphocyte trafficking (Mionnet et al., 2011). Indeed, in all LNs of LTβR-proficient mice, we found aggregates of CD4⁺ T cells surrounded by the MECA-79⁺ endothelium of HEVs (Fig. 5 d, bottom, arrows). However, this retention function was missing in the flat MECA-79⁺ endothelium of LNs from VE-cadherin-CrexLtβr^fl/fl mice (Fig. 5 d, top). The CCR7 ligands CCL19 and CCL21 are important for the regulation of general T cell motility (Förster et al., 2008). Because we observed a general down-regulation of Ccl19 and Ccl21 in LNs of VE-cadherin-CrexLtβr^fl/fl mice, we used intravital two-photon microscopy to determine whether general T cell motility would be affected by the EC-specific LTβR deficiency. As shown in Fig. 5 e, the specific LTβR defect in ECs and the resulting structural and functional alteration in the HEV network did not significantly impact on general T cell motility, neither in the T cell zone nor in the direct vicinity of MECA-79⁺ vessels. Collectively, constitutive LTβR signaling in LN vascular ECs appears to be necessary for the differentiation of vascular ECs into the typical HEV EC and, hence, the appropriate formation of the HEV network.

The earliest steps in LN development involve the PROX1-dependent metamorphosis of venous ECs to lymphatic ECs at the anterior cardinal vein (Srinivasan et al., 2007). These early LECs form the lymph sac which serves as the primordial tissue for both the LN anlagen and the sprouting lymphatic system (Blum and Pabst, 2006). After these initial steps, mesenchymal LTo and hematopoietic LTi cells are recruited to the LN anlage and the subsequent developmental steps are dependent on the constitutive chemokine receptor ligands CXCL13, CCL19, and CCL21 (Luther et al., 2003). Furthermore, triggering of the CD127/IL-7Rα on LTi cells in concert with constitutive chemokine signals is critical for the formation of LNs and Peyer’s patches. It has been suggested that LTβR signaling in mesenchymal LTo cells is the critical event that steers the cascading induction of constitutive chemokines and IL-7 (van de Pavert and Mebius, 2010). This particular view has been supported by the finding that the absence of lymphatic endothelium in PROX1-deficient mice did not block LN anlagen formation (Vondenhoff et al., 2009). The present study challenges the concept that there is only one dedicated LTβR-dependent stromal LTo cell population. We found that the ablation of LTβR expression in ECs in VE-cadherin-CrexLtβr^fl/fl mice led to impaired LN development. Because the VE-cadherin-Cre transgene is active in ECs but not in mesenchymal stromal cells (Alva et al., 2006; this study), our study provides conclusive evidence that LTβR signaling in endothelial LTo cells is mandatory for LN formation. At this point, we can only speculate at which stage of LN development LTi cells interact with early endothelial LTo cells. It is possible that lymphotoxin-expressing LTi cells stimulate LTβR⁺ LECs in the early LN anlage to produce more CCL21, and thereby contribute to the CCR7 dependency of LN development (Luther et al., 2003). Moreover, the development of the intranodal blood vasculature may be dependent on LTβR signals provided to BECs. Insufficiently developed LN blood vessels may restrict LN formation because growing tissues are particularly vulnerable to impaired blood supply. Future studies of conditional and inducible LEC- versus BEC-specific ablation of the LTβR will reveal which of the two EC populations function as an endothelial LTo cell.

ECs and hematopoietic cells share the hemangioblast as a common precursor cell (Domigan and Iruela-Arispe, 2012) and therefore, some molecular traits are shared between the lineages. For example, VE-cadherin is involved in endothelial homotypic cell adhesion, a function that is critical for vascular development and function (Carmeliet et al., 1999). Consequently, the VE-cadherin-Cre transgene exhibits uniform expression in the endothelium of developing and quiescent vessels (Alva et al., 2006). However, a small proportion of hematopietic cells express the VE-cadherin-Cre transgene (Alva et al., 2006). Among the different hematopietic cell types, macrophages (Wimmer et al., 2012) express the LTβR and exhibit some functional changes after LTβR triggering. Nevertheless, because macrophage-specific ablation of the LTβR did not affect LN formation (Wimmer et al., 2012), we conclude that the defect in LN development in VE-cadherin-CrexLtβr^fl/fl mice is a direct result of LTβR deficiency in ECs.

The characterization of the developmental pathways of LN organogenesis is most helpful for the understanding of the primary functions of these immune compartments in the adult, namely concentration of pathogens and immune cells to secure swift activation of innate and adaptive immunity (Junt et al., 2008). During immune activation, LNs undergo profound morphological changes to optimally accommodate or restore particular interaction compartments. For example, the LTβR-dependent interaction of adult LTi cells with T cell zone FRCs is critical for the restoration of immunocompetence after viral infection (Scandella et al., 2008). Furthermore, B cell–derived lymphotoxin significantly contributes to the remodeling and adaptation of the HEV network during the acute phase of a noncytopathic viral infection (Kumar et al., 2010). It has also been suggested that restructuring of the vascular LN microenvironment can be mediated by dendritic cells (Chyou et al., 2011). Likewise, the LTβR-dependent, homeostatic control of the adult LN HEV network (Browning et al., 2005) may be a result of multicellular processes (Chyou et al., 2008). The results of this study indicate that direct triggering of the LTβR on vascular ECs is important for the development of the cuboidal HEV appearance and the acquisition of lymphocyte traffic–regulating properties such as polarized ICAM1 expression, production of CCL21, and formation of pouches for lymphocyte retention. These findings bear interesting implications for the formation of tertiary lymphoid tissues in the course of infections and autoimmune diseases: chronically activated, lymphotoxin-expressing lymphocytes may directly induce the differentiation of ECs into the HEV phenotype, hence recapitulating the generation of the vascular LN microenvironment for optimal lymphocyte trafficking (Hayasaka et al., 2010). Further molecular dissection of this pathway may help to identify targets that attenuate the formation of tertiary lymphoid tissues at sites of chronic inflammation.

B6.Cg-Tg(Cdh5-cre)7Mlia/J (VE-Cadherin-Cre; Alva et al., 2006) and B6.129X1-Gt(ROSA)26Sor^tm1Hjf (R26-RFP) mice were purchased from The Jackson Laboratory. Ltbr^fl/fl mice were described previously (Wimmer et al., 2012). C57BL/6 mice were obtained from Charles River. All animals were kept under conventional conditions in individually ventilated cages. Experiments were performed in accordance with federal and cantonal guidelines (Tierschutzgesetz) under the permission numbers SG11/05 and SG11/04 granted by the Veterinary Office of the Canton of St. Gallen.

Single-cell suspensions from spleens of C57BL/6 mice were subjected to hypotonic red blood cell lysis and stained with CFSE (Molecular Probes). A maximum concentration of 2.5 × 10⁷ cells/ml were incubated in 5 µM CFSE in PBS for 10 min at 37°C. Cells were washed twice with ice-cold balanced salt solution (BSS) and resuspended in BSS. VE-Cadherin-Crex Ltbr^fl/fl or control mice were injected intravenously with 10⁷ C57BL/6 splenocytes in 200 µl BSS. Homing of transferred cells was analyzed 2 h after transfer by flow cytometry and histology.

LNs were dissected into small pieces and transferred into a 24-well dish filled with RPMI 1640 medium containing 2% FCS, 20 mM Hepes (all from Lonza), 1 mg/ml Collagenase Type IV (Sigma-Aldrich), and 25 µg/ml DNaseI (AppliChem). Dissociated LNs were incubated at 37°C for 30 min. After enzymatic digestion, cell suspensions were washed with PBS containing 0.5% FCS and 10 mM EDTA (MACS buffer). To further enrich the stromal cell fraction, hematopoietic cells were depleted by incubating the cell suspension with MACS anti-CD45 Microbeads and passing over a MACS LS column (Miltenyi Biotec).

Single-cell suspensions were incubated for 20 min at 4°C in PBS containing 1% FCS and 10 mM EDTA with the following fluorescently labeled antibodies: anti-CD45, anti-gp38/podoplanin (BD), anti-CD31, anti-LTβR (eBioscience), anti-CD4, anti-CD8, anti-CD11c, anti-CD11b, and anti-F4/80 (BioLegend). Cells were acquired with a FACS Canto (BD) and analyzed using FlowJo software (Tree Star). Cell sorting was performed using a FACS Aria cell sorter (BD).

Total cellular RNA was extracted from homogenized tissues and sorted cells using TRIzol reagent (Invitrogen) following the manufacturer’s protocol. cDNA was prepared using cDNA archive kit (Applied Biosystems), and quantitative RT-PCR was performed using the Light Cycler-FastStart DNA Master SYBR Green I kit (Roche) on a LightCycler machine (Roche). Expression levels were measured using the following primers: Wnt1; Mfge8; QT00098917; Glycam1; QT00141526 (QuantiTect primer assay; QT00103985; QIAGEN); Ccl19, forward 5′-CTGCCTCAGATTATCTGCCAT-3′, reverse 5′-AGGTAGCGGAAGGCTTTCAC-3′; Ccl21, forward 5′-AAGGCAGTGATGGAGGGG-3′, reverse 5′-CGGGGTAAGAACAGGATTG-3′; tata-box binding protein (tbp), forward 5′-CCTTCACCAATGACTCCTATGAC-3′, reverse 5′-CAAGTTTACAGCCAAGATTCAC-3′.

LNs were fixed over night in freshly prepared 4% paraformaldehyde (Merck) at 4°C under agitation. Fixed LNs were embedded in 4% low melting agarose (Invitrogen) in PBS and sectioned with a vibratome (VT-1200; Leica). 20–30-µm-thick sections were blocked in PBS containing 10% FCS, 1 mg/ml anti-FcRγ (BD) and 0.1% Triton X-100 (Sigma-Aldrich). Sections were incubated over night at 4°C with the following antibodies: anti-RFP, anti-CCL21 (Abcam), anti-gp38/podoplanin, anti-CD45, anti-B220, conjugated anti-CD31, anti-LyveI (eBioscience), anti-CD4, anti–ICAM-1 (BioLegend), and anti MECA-79. Unconjugated antibodies were detected with the following secondary antibodies: Dylight649-conjugated anti–rat-IgG, Alexa Fluor 488–conjugated anti–rabbit-IgG, Dylight549-conjugated anti–Syrian hamster-IgG, and Dylight549-conjugated Streptavidin (all purchased from Jackson ImmunoResearch Laboratories). Fluorescence signal of LTβR staining was amplified by using a tyramide-amplification kit (Molecular Probes) according to the manufacturer’s protocol. Microscopic analysis was performed using a confocal microscope (LSM-710; Carl Zeiss) and images were processed with ZEN 2010 software (Carl Zeiss). For tracing of lymphatic flow, mice were injected with 50 µg of 40-kD FITC-dextran into the hind footpad. 10 min after injection, mice were sacrificed and draining LNs were carefully excised under the stereomicroscope, fixed, and cut with a vibratome (VT-1200; Leica). FITC fluorescence per LN section was quantified using the LSM-710 confocal microscope.

Mice received an intravenous injection of fluorescently labeled MECA-79 (12–15 µg) to label the HEV network. After 15 min, mice were sacrificed, LNs were carefully excised, and surrounding tissue was removed under a stereomicroscope. Sample preparation and OPT were performed as previously described (Kumar et al., 2010).

Purified C57BL/6 T cells were fluorescently labeled with 2.5 µM chloromethyl-benzoyl amino-tetramethylrhodamine or CFSE for 15 min at 37°C. After washing, labeled T cells were injected intravenously into sex-matched mice, which were anesthetized and surgically prepared to expose the right popliteal LN. Immediately after injection, two-photon microscopy was performed as previously described (Soriano et al., 2011).

Two-photon microscopy and OPT data were analyzed with Volocity (Perkin Elmer) and Imaris (Bitplane). All statistical analyses were performed with Prism 5.0 (GraphPad Software Inc.). Data were analyzed with the nonpaired Student’s t test. A p-value of <0.05 was considered significant.

We would like to thank Rita DeGiuli for technical support.

This study received financial support from the Swiss National Science Foundation (Sinergia 125447/1 to J.V. Stein and B. Ludewig and 130823/1 to B. Ludewig).

The authors declare no potential conflict of interest.

Author contributions: B. Ludewig designed the study and wrote the paper; L. Onder performed research and wrote the paper; E. Scandella, J.V. Stein, and T. Hehlgans analyzed data; R. Danuser, S. Firner, and Qian Chai performed research.