Inactivation of genes encoding members of TNF and TNF receptor families reveal their divergent roles in the formation and function of secondary lymphoid organs. Most lymphotoxin α (ltα)- and all lymphotoxin β receptor (ltβr)-deficient mice are completely devoid of lymph nodes (LNs); however, most lymphotoxin β (ltβ)-deficient mice develop mesenteric LNs. Tnf- and tnfrp55-deficient mice develop a complete set of LNs, while ltβ/tnfrp55 double-deficient mice lack all LNs, demonstrating cooperation between LTβ and TNFRp55 in LN development. Now we report that ltβ/tnf double-deficient mice develop the same set of mucosal LNs as do ltβ-deficient mice, suggesting that ligands other than TNF signal through TNFRp55 during LN development. These LNs retain distinct T and B cells areas; however, they lack follicular dendritic cell networks. Structures resembling germinal centers can be found in the LNs from immunized ltβ-deficient mice but not in ltβ/tnf double-deficient mice. Additionally, stromal components of the spleen and LNs appear to be more severely disturbed in ltβ/tnf double-deficient mice as compared with ltβ-deficient mice. We conclude that LTβ and TNF cooperate in the establishment of the correct microarchitecture of lymphoid organs.
In recent years, a significant effort has been made to dissect lymphotoxin (LT)4 and TNF signaling using mice-deficient in TNF, LT, or their receptors as well as employing biochemical inactivation of LT/TNF signaling pathways (see Refs. 1 and 2 ; reviewed in Ref. 3). From in vitro studies, it is known that TNF3 and LTα3 can bind to two TNF receptors p55 and p75, while heterotrimeric LTα1/LTβ2 is the main ligand for the LTβR (4). Specific contributions of LT/LTβR and TNF/TNFR signaling pathways to the development and the maintenance of microarchitecture of peripheral lymphoid organs are starting to emerge. Mice deficient in LTα (5, 6) lack proper T and B cell compartmentalization, marginal zone and FDC networks, and do not develop germinal centers (GC) in the spleen upon immunization. LTβ-deficient mice have similar defects in spleen; however, T and B cell zones are not as mixed as in LTα-deficient mice (7, 8). TNF-deficient mice have even milder disturbances—marginal zone, albeit altered, is present in the spleen of these mice—but FDC networks are absent and GCs are not formed (9, 10, 11). Additionally, recent data by Alexopoulou et al. (12) suggested that TNF expression may be disturbed in the widely used neo-containing LTα-deficient mice and thus could have contributed to the severity of splenic white pulp disorganization.
TNF-deficient mice develop lymph nodes (LNs) normally, whereas LTα-deficient mice lack all LNs, except for abnormal lymphoid structures found in the mesenteric fat in 2–30% of mice (6, 13). LTβ-deficient mice show an intermediate phenotype in this regard, they usually develop mesenteric LNs (MLNs), and, more rarely, sacral LNs (SLNs) and cervical LN. In these LNs, T and B cells segregate well and peanut agglutinin (PNA)-binding clusters are formed, but associated mature folicular dendritic cell (FDC) networks do not develop (7, 8, 14).
Defects in the spleen of LTβ-deficient mice (7, 8) are more severe than in TNF−/− mice (9). In contrast, several reports using biochemical rather than genetic inactivation of TNF (15, 16) suggested that TNF may play a distinct role in lymphoid organogenesis. Therefore, it was of considerable interest to determine whether a combined neo-free genetic excision of ltβ and tnf genes may result in a more severe lymphoid deficiency as compared with single knockouts of the same genes.
In the present study, we generated and characterized a new mouse line with a combined TNF/LT deficiency, (LTβ/TNF)Δ/Δ double-deficient mice. We further dissected the contributions to the development and organization of peripheral lymphoid tissues of the two main signaling pathways involving LT and TNF. We show that under conditions where signaling both by LTα1β2 via the LTβR and by TNF via the TNFRp55 are abrogated, two previously unrecognized effects become apparent: 1) redundant contributions of alternative ligands (i.e., LTα3 and/or LIGHT) to the organogenesis of LNs, and 2) cooperative contributions of LTβR and TNFRp55 signaling to the microarchitecture of lymphoid organs.
Materials and Methods
Generation of LTβ/TNF double-deficient mice
Construction of the targeting vector, embryonic stem (ES) cell transfection, and screening of homologous recombinants have been previously described (7). Of the 43 ES cell clones with a targeted insertion of the neo cassette, 15 clones revealed a cointegration of the second and the third loxP sites without cointegration of the fourth loxP motif of pTV2-TK. Two of these ES cell clones were used for injection into C57BL/6 blastocysts (17). Chimeric male mice were crossed to C57BL/6 mice to obtain germline transmission of the (LTβ/TNF)+/T mutation. (LTβ/TNF)+/T mice were crossed to Cre-transgenic “deleter” mice with ubiquitous expression of Cre recombinase (18). As a result, the neo-cassette and a region of the TNF/LT locus containing exon 3 of the ltβ gene and exons 3 and 4 of the tnf gene were deleted in the mutant allele. The progeny was genotyped by PCR using primers gtype1, 5′-CGG GTC TCC GAC CTA GAG ATC and gtype5a, 5′-CAG ACC CTC ACA CTC AGT AAG. Correct excision of the portion of the TNF/LT locus was further confirmed by Southern analysis using BamHI digestion of genomic DNA and the SphI-PstI portion of the LTβ promoter as a probe. The Cre-transgene was crossed out during further backcrossing to C57BL/6 mice. Heterozygous mice were bred to obtain (LTβ/TNF)Δ/Δ double-deficient mice, which were then embryo rederived and maintained under specific pathogen-free conditions.
All mice were generated on a mixed C57/BL6/129cv background and then backcrossed to C57/BL6. LTα−/− mice (5) were purchased from The Jackson Laboratory (Bar Harbor, ME). LTβ-deficient mice (LTβΔ/Δ mice) have been described previously (7). All mice were maintained under specific pathogen-free conditions and used for experiments between 8 and 12 wk of age. Animal care was provided in accordance with the procedures outlined in the “Guide for the Care and Use of Laboratory Animals” (National Institutes of Health Publication No. 86-23, 1985).
Total cellular RNA was extracted with Trizol reagent and used as described by the manufacturer in RNAase protection assays with the mCK-3 set of probes, including ltα, ltβ, and tnf genes (PharMingen, San Diego, CA).
Where indicated, mice received a single i.p. injection of 50 μg of alum-precipitated nitrophenol-haptenated chicken γ-globulin or 108 SRBC in PBS and were analyzed 8 days after immunization.
Immunohistochemical analysis was performed as described (2, 7). Hamster Abs to CD3α and rat Abs to B220, IgD, IgM, CR1, and mucosal addressin cell adhesion molecule-1 were obtained from PharMingen. Biotinylated PNA and the alkaline phosphatase staining kit were purchased from Vector Laboratories (Burlingame, CA). ER-TR7 mAb was purchased from Biogenesis (Poole, U.K.). MOMA-1 mAb was purchased from Research Diagnostics (Flanders, NJ). FDC-M1 Abs were generously provided by Dr. M. Kosco-Vilbois (Serono Pharmaceutical Research Institute, Geneva, Switzerland). Peroxidase-conjugated mouse anti-rat IgG Abs (F(ab′)2) were obtained from Jackson ImmunoResearch (West Grove, PA). Alkaline phosphatase-conjugated streptavidin and other reagents were obtained from Sigma (St. Louis).
Flow cytometric analysis was performed on single-cell suspensions as described (2). Biotin-, FITC-, or PE-conjugated rat Abs to the following mouse surface markers were used: Thy1.2 (clone 30H12), CD4 (clone RM4-5), and CD8 (clone 53-6.7) to detect T cells; B220 (clone RA3-6B2) for B cells; and Gr1 (clone RB6-8C5) and Mac1α (clone M1/70) for granulocytes and macrophages. All primary Abs were obtained from PharMingen.
Results and Discussion
Generation of LTβ/TNF double-deficient mice
To address the possible redundancy of TNF/LT ligands in the development of peripheral lymphoid organs we have generated a mouse strain deficient in both LTβ and TNF. Because the tnf and lt genes are closely linked (19, 20, 21) it would be virtually impossible to generate such mice by crossing single TNF and LTβ knockout mice. Therefore, (LTβ/TNF)Δ/Δ mice were generated using Cre-loxP technology, similarly to the previously described LTβΔ/Δ mice (7). In particular, we generated ES cells with a partial homologous recombination of the targeting vector pTV2-TK (7) and used them for injection into C57BL/6 blastocysts. After germline transmission, Cre-mediated recombination in vivo using “deleter” mice (18) was employed to remove the selectable marker (neo gene) and to create a 5.3-kb deletion (Fig. 1,A). This deletion (confirmed by Southern analysis; Fig. 1 B) included the third exon of the ltβ gene (encoding most of the extracellular portion of this cytokine) and the third and the fourth exons of the tnf gene (encoding the entire secreted portion of the TNF protein) and did not allow the expression of functional LTβ and TNF mRNAs.
Mice with the double LTβ/TNF deficiency were born at the expected Mendelian ratios, appeared healthy, and were fertile. Thymi had normal weight and appearance, and the major cell populations in thymus and bone marrow were normal as measured by FACS analysis (data not shown). Lymphoid infiltrates were found around blood vessels in lung and liver of (LTβ/TNF)Δ/Δ mice (data not shown), similarly to the infiltrates described for LTβΔ/Δ mice (7), LTα−/− mice (6), and LTβR−/− mice (2). Transcription of the ltα gene was not disturbed by the deletion, as confirmed by RNase protection assay performed with total RNA isolated from Con A-activated splenocytes (Fig. 1 C).
TNF and LTβ do not cooperate in LN organogenesis
Peripheral lymphoid organs of (LTβ/TNF)Δ/Δ mice were examined. No visible Peyer’s patches were found and no inguinal, brachial, axillary, iliac, or popliteal LN could be detected by histopathological examination (data not shown). However, MLNs and sometimes SLNs could be readily detected (in 15 of 17 and 6 of 17 mice, respectively). Thus, lymphoid organs in (LTβ/TNF)Δ/Δ mice develop similarly to the previously described LTβΔ/Δ mice (MLNs, 15 of 23; SLNs, 6 of 23). The anatomical shape of MLNs was unusual in both single ltβ- and double ltβ/tnf-deficient animals. In most cases, only one large, spherical MLN was found in mutant mice at a point where the root of the mesentery merges with the parietal peritoneal mesothelium, as opposed to a chain of three elongated LN segments normally present along the mesenteric artery in wild-type mice. The specific anatomical location of the MLN found in LTβΔ/Δ and (LTβ/TNF)Δ/Δ mice suggests that some of the segments of the MLN draining the lymph from the gut (small bowel, cecum, and sigmoid colon) may be missing in these animals.
MLNs from (LTβ/TNF)Δ/Δ mice were examined by immunohistology and compared with MLNs from wild-type mice and from LTβΔ/Δ mice (Fig. 2). As expected, FDC networks were absent in all mutant mice as assessed by staining for CR1 (Fig. 2) or FDC-M1 (data not shown).
Additional abnormalities of the MLN microarchitecture were observed in the (LTβ/TNF)Δ/Δ mice as compared with LTβΔ/Δ mice. Both types of mice retained a distinct segregation of lymphocytes into a superficial cortical B cell zone and a deep cortical T cell zone. Furthermore, LTβΔ/Δ mice were able to form structures in the superficial cortex that resembled GCs in MLN upon immunization (Fig. 2), in agreement with the recent report by Koni and Flavell (14). Because CR1+ FDC networks appeared to be absent in LTβΔ/Δ mice, it is not possible to state with certainty if these GC-like structures were segregated into light zone or dark zone compartments. However, small IgD+/IgM+ lymphocyte coronas and IgM+/PNA+ GCs were found both in unstimulated and in Ag-stimulated mice (data not shown).
In contrast to LTβΔ/Δ mice, PNA+ B cells found in the LNs from (LTβ/TNF)Δ/Δ mice were scattered and did not form foci resembling GCs. There was segregation of B220+ B cells to the superficial cortex and of CD3+ T cells to the deep cortex of MLN, but in areas where one would expect to find B cell follicles, only scattered individual PNA+ B cells were found (Fig. 2).
Examination of the underlying fibroblastic reticular cell (FRC) architecture with ER-TR7 Ab (22) provided useful correlates. In wild-type mice, there was a delicate FRC network in the deep cortex (paracortex) that formed the narrow perivenular channel, FRC corridors, and cortical columns that are characteristic of this compartment (23). In the superficial cortex, this delicate pattern undergoes an abrupt transition where there are interfaces with B cell follicles (Fig. 2, bottom row). The corona usually retains the delicate pattern of the paracortex but GCs are separated from the corona by an FRC “barrier” and the GC areas contain broad zones that are relatively free of fibroblastic reticulum. This may presumably be to accommodate the FDC network. Regardless of the reason, the FRC pattern usually reflects the presence of structures like GCs.
In the LTβΔ/Δ mice, the reticulum of the deep cortex was coarser than in the wild-type mice, with thicker fibers and wider spaced corridors. In these mice, the transition of FRC at GC-like structures was less distinct than that seen in the wild-type mice, but some alteration in the reticulum pattern in the areas of PNA+ B cell foci was evident, with a ring of more dense stromal components surrounding PNA+ area (Fig. 2, bottom row). In contrast, in (LTβ/TNF)Δ/Δ mice the coarse pattern of the fibroblastic reticulum was uniform across the T-dependent and B-dependent cortex and no GC-like architectural transitions were seen. Apparently, the loosely scattered PNA+ B cells were of insufficient aggregate size to have an impact on the underlying reticulum, so the cortical pattern extended to the subcapsular sinus.
We concluded that PNA+ B cell clusters resembling GCs can be formed in the absence of FDC network (as assessed by CR1 and FDC-M1 staining) in LTβΔ/Δ but not in (LTβ/TNF)Δ/Δ mice, suggesting independent contribution of TNF into underlying signaling mechanism. We also concluded that stromal components of the MLN were more severely disturbed in the double knockout.
In an attempt to characterize cervical LNs in (LTβ/TNF)Δ/Δ mice, we performed immunohistological examination of LN-like structures found in the normal location of cervical LNs (data not shown). However, we were unable to identify any true cervical LNs either in (LTβ/TNF)Δ/Δ mice (n = 5) or in LTβΔ/Δ mice (n = 7). This result, which is somewhat contradictory to the published observations (7, 8), may be due to the fact that the mice used in this study have been rederived by embryo transfer and maintained under specific pathogen-free conditions thereafter or due to a low incidence of cervical LNs in this mouse model.
The double ltb and tnfrp55 knockout mice (24) lack all LNs, including MLN and SLN (see Table I). Because inactivation of the TNFRp55 gene disrupts signaling by both TNF and LTα3, the comparison with the (LTβ/TNF)Δ/Δ model suggests that it is binding of LTα3, and not of TNF, to TNFRp55 that is important for cooperation with LTβR signaling for the development of the mucosal subset of LNs.
|Genotype .||Splenic Architecture .||.||.||Peyer’s Patch .||LN .||References .|
|.||Marginal zone .||T/B cell areas .||GC/FDC .||.||.||.|
|ltβ−/− tnf−/−||Absent||Variable||No||Absent||Mucosal||This work|
|Genotype .||Splenic Architecture .||.||.||Peyer’s Patch .||LN .||References .|
|.||Marginal zone .||T/B cell areas .||GC/FDC .||.||.||.|
|ltβ−/− tnf−/−||Absent||Variable||No||Absent||Mucosal||This work|
The features of the secondary lymphoid organs of (LTβ/TNF)Δ/Δ mice are highlighted by comparing them to several selected knockout mice with deficiencies in TNF/LT and their receptors.
NR, Not reported.
Additionally, Rennert et al. (1) used injections of LTβR-Ig and TNFRp55-Ig or LTβR-Ig and anti-TNF mAb into pregnant mice to inactivate simultaneously both signaling pathways. Results with the first combination of antagonists are in agreement with genetic data (24).
However, the absence of LNs from wild-type mice simultaneously injected in utero with LTβR-Ig and anti-TNF mAb (1) is in apparent contradiction with our results. Several explanations can be offered. First, if effects of TNF are mediated by cell-surface-bound ligands, then TNF-bearing cells may be lysed in the Ab-injection experiment by a complement-mediated mechanism. Second, LN genesis may involve an additional signaling through LTβR by LIGHT (25) or by another as yet unidentified ligand, and such signaling would be disrupted in the model of Rennert et al. but not in (LTβ/TNF)Δ/Δ mice. This hypothesis is further supported by the report on the mice with genetic inactivation of LTβR (2), which do not develop any LNs.
Collectively, our data support the cooperation of TNFRp55 and LTβR signaling cascades in organogenesis of LNs and suggest a distinct role for LTα3-TNFRp55 signaling. Together with observations by Koni and Flavell (24), this represents a novel biological function of LTα3 detected in vivo in a nontransgenic model. Previously evidence for distinct function of LTα3-TNFRp55 interaction in lymphoid organogenesis and inflammation was provided by a series of elegant studies using LTα transgene under the control of the rat insulin promoter (26, 27, 28, 29). The hypothesis concerning the role of an additional LTα-specific receptor in LN development (30) is not supported by the phenotype of tnfrp55−/−ltβ−/− (24) or (LTβ/TNF)Δ/Δ mice (this report).
The development of LNs is a complex process involving several cell lineages and multiple signaling pathways. Nevertheless, the role of LT-LTβR has been reinforced by the findings on aly/aly mice (31) in which case the defect is caused by mutation in carboxyl-terminal end of NF-κB-inducing kinase (NIK) (32). This result also places NIK (33) in the signaling cascade downstream of LTβR (and probably of some other receptors of the TNFR superfamily) and not of TNFRp55 (32). Additionally, opgl−/− mice (34) also completely lack LNs. Osteoprotegerin (OPG) ligand (OPGL) (also known as TNF-related activation-induced cytokine (TRANCE), receptor activator of NF-κB ligand (RANKL), or osteoclast differentiation factor (ODF); Refs. 35, 36, 37) is a ligand for OPG and RANK that is also a member of the TNFR superfamily. It is possible that the OPGL/(OPG/RANK) signaling pathway shares some downstream molecules with LTβR signaling pathways (see Ref. 35). Whether NIK is involved in this signaling remains to be demonstrated. Alternatively, OPG/RANK signaling may be required at a different step in LN development.
TNF and LTβ cooperate in the maintenance of splenic microarchitecture
Spleens of (LTβ/TNF)Δ/Δ mice were studied by immunohistochemistry after i.p. immunization with T cell-dependent Ags SRBC (Fig. 3) or nitrophenol-haptenated chicken γ-globulin (data not shown). Wild-type littermate controls, LTβΔ/Δ (7), and LTα−/− (5) mice were examined in parallel (Table I). As a common feature, FDC networks could not be detected in any of the mutant mice using staining for CR1 (Fig. 3) or FDC-M1 (data not shown), and there were no clustered PNA-positive cells in B cell areas, although some aberrant PNA staining with a variable pattern could be detected around central arterioles and blood vessels (Fig. 3). In agreement with the published data, the staining of the marginal zone of the spleen by mucosal addressin cell adhesion molecule-1 and MOMA-1 markers was absent in (LTβ/TNF)Δ/Δ, LTβΔ/Δ, and LTα−/− mice (data not shown).
Importantly, the overall splenic structure was more severely disturbed in the double-deficient mice, as compared with single LTβ knockouts (Fig. 3 and Table I). Spleens of LTβΔ/Δ mice still retained relatively distinct B cell zones, while in the double-deficient mice B cell areas were smaller and much less defined (Fig. 3), even though the numbers of B and T cells in the spleen of the double (LTβ/TNF)Δ/Δ mice were not changed as assessed by flow cytometry analysis (data not shown) The relative numbers of T and B cells in spleen of (LTβ/TNF)Δ/Δ mice (15–20% and 30–35%, respectively) as well as the absolute numbers of lymphocytes in spleen (0.8–1.5 × 108) and in thymus (1.3–1.7 × 108) were not significantly different from the wild-type mice, while the white cell counts in blood and peritoneum were two to three times higher in (LTβ/TNF)Δ/Δ mice as compared with the wild-type mice, the phenomenon previously described for LTα−/− (5, 6) and LTβΔ/Δ mice (7, 8).
The profound defect of the splenic T/B cell segregation was further characterized using labels specific for the stromal components of the spleen. Labeling with ER-TR7 Ab, which detects reticular fibroblasts and blood vessel walls (22), revealed that stromal components of the spleen that support spatial organization of the lymphoid tissues (38) were more disturbed in (LTβ/TNF)Δ/Δ mice than in LTβΔ/Δ mice (Fig. 3). ER-TR7-positive stromal elements in the spleen of LTβΔ/Δ mice, even though markedly disorganized in comparison with the wild type, still formed a distinct ellipsoid-like structure confining the white pulp area. In contrast, ER-TR7 labeling of the spleen of (LTβ/TNF)Δ/Δ mice did not reveal a clear boundary between the red and white pulp (Fig. 3). Routine chemical stains for elastic or collagen fibers (Orcein and Masson’s TriChrome) (39) revealed an increased density of these connective tissue elements around the central arterioles in the spleen of (LTβ/TNF)Δ/Δ mice as compared with LTβΔ/Δ and wild-type mice (Fig. 4). This accumulation of the connective tissue around central arterioles was not due to insufficient number of T cells (see above) and may be responsible for the deficient compartmentalization by affecting lymphocyte migration or by providing dislocated chemotactic or adhesion signals (40).
Expression levels of certain chemokines also play an important role in forming functional compartments of the spleen. Mice deficient in one of the chemokine receptors, BLR-1/CXCR5 (41), have defects in their peripheral lymphoid tissues that are reminiscent, although not as severe, as those reported for LTα- or LTβ-deficient mice. A substantial decrease with regard to expression levels of the B lymphocyte chemoattractant and of the secondary lymphoid tissue chemokine reported in association with LTα deficiency (42) is also observed in both (LTβ)Δ/Δ and (LTβ/TNF)Δ/Δ mice (without substantial difference between these two models, data not shown). In the future, it will be interesting to determine whether any of these chemokine genes may be under direct control of LTβR or TNFR55 signaling.
In summary, our genetic data demonstrate that there is an independent contribution of TNF to the LT-dependent maintenance of the microarchitecture of the spleen and LN. Our data also suggest a distinct role of LTα3-TNFRp55 interaction in the development of mucosal LNs.
We are greatly indebted to K. Rajewsky for making this project possible, to A. Tarakhovsky for his help and advise on generating ES clones, and to R. Endres for his thoughtful comments and suggestions. We are grateful to A. Shakhov for chemokine analysis and helpful discussion. We also thank U. Huffstadt and L. Drutskaya for assistance, M. Kosco-Vilbois for FDC-M1 Abs, R.Turetskaya for helpful discussions, and H. Wagner and J.J. Oppenheim for continuing support and critical reading of the manuscript. We acknowledge the personnel of the Animal Facility and of the Central Pathology and Histotechnology Laboratory, National Cancer Institute-Fredrick Cancer Research and Development Center, for expert technical assistance.
This project has been funded in whole or in part with U.S. Federal funds from the National Cancer Institute, National Institutes of Health (Contract NO1-CO-56000), by the Deutsche Forschungsgemeinschaft (Grant Pf259/2-4), the Sonderforschungsbereich 391, Klinische Forschergruppe Postoperative Immunparalyse und Sepsis (Grant Si208/5-1), by Grant 98-04-49029 from the Russian Foundation for Basic Research, and by a grant from the Russian State Program “Frontiers in Genetics.” M.B.A. was supported by a fellowship from the European Molecular Biology Organization. S.A.N. is an International Research Scholar of the Howard Hughes Medical Institute. The contents of this publication do not necessarily reflect the view or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. The publisher or recipient acknowledges right of the U.S. Government to retain a nonexclusive, royalty-free license in and to any copyright covering the article.
D.V.K., M.B.A., and A.V.T. contributed equally to this work.
Address correspondence and reprint requests to Dr. Dmitry V. Kuprash, Engelhardt Institute of Molecular Biology, Russian Academy of Sciences, 32 Vavilov Street, 117984 Moscow, Russia. E-mail address: email@example.com
Abbreviations used in this paper: LT, lymphotoxin; GC, germinal center; FDC, follicular dendritic cell; FRC, fibroblastic reticular cell; LN, lymph node; MLN, mesenteric LN; SLN, sacral LN; OPG, osteoprotegerin; PNA, peanut agglutinin; ES cell, embryonic stem cell; NIK, NF-κB-inducing kinase.