Correction of the TNF-LTα-Deficient Phenotype by Bone Marrow Transplantation

Mice with inactivated TNF-LTα genes have profound abnormalities of the immune system with hypoimmunoglobulinaemia, lack of lymph nodes, undifferentiated spleen, and defective Ig class switch. Transplantation of bone marrow cells from wild-type mice restored the synthesis of TNF, corrected the splenic microarchitecture, repopulated the lamina propria with IgA-producing plasma cells, and normalized the serum immunoglobulin levels of TNF-LTα deficient mice. Furthermore, the formation of germinal centers in the spleen and the defective Ig class switch in response to a T-cell-dependent antigen is corrected. These data demonstrate that most TNFproducing cells are bone-marrow-derived, and that the immunodeficiency due to TNF-LTα deletion can be corrected to a large extent by normal bone marrow, cell transplantation.

We have generated a TNF-LTc-deficient mouse, which has a complete disruption of the TNF-LTc system, since it has neither of the two ligands signaling through the TNFR-1 and -2 nor through the LT/3-R . These TNF-LTo double-deficient mice develop normally with the exception of a profound immunodeficiency. The mutant mice have a sixfold increase of circulating B lymphocytes, no peripheral lymphoid organs, an undifferentiated spleen, a hypoimmunoglobulinemia, and very few IgA-producing plasma cells in the lamina propria of the small intestine.
Here, we asked to what extent the complex phenotype of TNF-LTc deficiency is due to incompetent lymphohemopoietic progenitors cells or rather due to a defective environment. For this, we attempted to restore the mutant phenotype by transplantation of wild-type bone marrow cells.

RESULTS
The Cells Producing TNF Systemically On Endotoxin Stimulation Are of Bone Marrow Origin TNF-LTc double-deficient mice fail to produce TNF after endotoxin injection . On transplantation with wild-type bone marrow cells, TNF-LTc-deficient mice respond to endotoxin with TNF plasma levels at 1 h/z in the ng/ml range similar to wild-type mice (Fig. 1A). TNF-LTc-deficient mice lethally irradiated and reconstituted with TNF-LTcdeficient bone marrow cells served as controls and behaved essentially as nontransplanted TNF-LTodeficient mice. Conversely, lethally irradiated wildtype mice produced high levels of TNF when reconstituted with wild-type, but not with mutant bone marrow cells. These results demonstrate that the majority of cells producing TNF after endotoxin stimulation are radiosensitive and bone-marrow- FIGURE Correction of TNF production in TNF-LTc (-/-)-deficient mice on bone marrow transplantation from wild-type mice: (A) TNF plasma levels hr after LPS injection. (B) In vitro TNF production by bone marrow-derived macrophages stimulated with LPS (3hr, 10 ng/ml) in the presence of IFNy (100 U). Bone-marrow cells (2 106) from wild-type mice (+/+) or from TNF-LTc knockout (-/-) donor mice were transplanted into lethally irradiated wild-type (+/+) or TNF-LTc knockout recipient mice (-/-). Analysis of the in vivo TNF production in response to LPS (10 /xg i.p.) was performed 14 weeks posttransplantation, at a time point where the hematological parameters were stabilized (A). derived. Ex vivo, bone-marrow-derived macrophages from mutant mice reconstituted with wild-type donor cells produced similar amounts of TNF on LPS stimulation as those from wild-type controls ( Figure  1B). The spleen of TNF-LTc-deficient mice are normal sized, but show no organization of the white pulp . On transplantation of normal bone marrow cells, the architecture of the follicles of TNF-LTc-deficient mice is corrected with a distinct separation of Tand B-cell areas (not shown) with appearance of MAdCAM-1 expression in the marginal sinus endothelium (not shown). As expected, mutant bone marrow was unable to correct the splenic structure of TNF-LTo-deficient mice (data not shown). Whereas the correction of the spleen was successful, bone marrow transplantation did not reconstitute peripheral lymph nodes in TNF-LTcdeficient mice, indicating a developmental defect in these mice.
TNF-LTc-deficient mice have a profound reduction of serum immunoglobulin level, especially of IgA with an almost complete absence of IgAproducing plasma cells in the intestine. Upon wildtype marrow-cell transplantation, TNF-LTc-deficient mice displayed a population of IgA-producing plasma cells in the lamina propria that was indistinguishable from that of wild-type controls (not shown). Since the splenic structure and the production of IgA in the lamina propria of the intestine were corrected by bone marrow transplantation, we asked whether the hypoimunoglobinaemia was also normalized. In agreement with previous studies, a profound reduction of Ig isotypes in TNF-LTc-deficient mice was observed . Figure  Furthermore, we tested whether the Ig class switch defect described previously in TNF-LTce-deficient mice

DISCUSSION
The present experiments demonstrate that bone marrow cells are able to correct the TNF-LTce deficiency. Furthermore, the phenotype can be transferred into normal mice by bone transplantation from TNF-LTcedeficient mice . Therefore, the genotype of the donor cells determines largely the phenotype of the irradiated host. Furthermore, the data suggest that the majority of cells producing TNF and LTce are radiosensitive. Several previous studies indicated that not only monocytes and T cells, but many other cell types, for example, Kupffer cells and hepatocytes, produce TNF and/or LTce (Aggarwal and Vilcek, 1992;Hunt et al., 1992;Vassalli, 1992  Kupffer cells are in part bone-marrow-derived, but are relatively radioresistant (Freudenberg et al., 1986). Our results suggest that the nonbone-marrow-derived cells apparently contribute little to the systemic release of TNF in response to endotoxin. Together with the reconstitution of TNF and LT production, the splenic follicular structure was corrected, the lamina propria of the intestines repopulated by IgA-producing plasma cells, and the hypoimmunoglobulinemia restored.
Repeated immunization with antigen did not induce splenic germinal centers in TNF-LTa-deficient mice. However, germinal centers were induced in TNF-LTce-deficient mice reconstituted with normal bone marrow cells. A similar reconstitution of germinal centers by bone marrow cell transplantation was reported in LTce-deficient mice (Matsumoto et al., 1996a). Therefore, TNF or LTce are necessary for the formation of germinal centers. Interestingly, TNF-R1deficient mice lack also a germinal center response, suggesting that signaling through the type 1 TNF receptor is necessary for this effect (Le Matsumoto et al., 1996a). This finding is supported by our previous observation that follicular dendritic cells express high levels of TNF-R1 (Ryffel et al., 1991). The reconstituted mice were immunized with SRBC (2 108 i.p.) and the SRBC-specific (A) IgM, (B) IgG1, (C) IgG2b, and (D) IgG3 titers were determined by ELISA prior to (solid bar) and 6 days (shaded bar) and 15 days (white bar) after immunization. Results are pooled from (two) independent experiments and expressed as mean +_ SEM (n 7 to 8 mice per group).
Although bone marrow transplantation from wildtype mice was able to restore the immunodeficiency of adult TNF-LTc-deficient mice to a large extent, no lymph nodes were formed. Therefore, LTc is required during the embryonic development of lymph nodes. However, in the absence of TNF and LTc, the spleen is formed although the microarchitecture is abnomal. These findings indicate that the spleen and lymph nodes have different requirements in embryonic development. Since LTc-deficient mice (De Togni et al., 1994;Banks et al., 1995) also lack lymph nodes, but TNF-R-1 and -2 have normal lymph nodes (Pfeffer et al., 1993;Rothe et al., 1993;Erickson et al., 1994), the developmental pathways for lymph nodes require LT/3-receptor signaling.
Several reports showed a possible involvement of soluble and membrane-bound TNF in immunoglobulin production. In particular, soluble TNF is able to enhance the immune response to SRBC (Ghiara et al., 1987). In addition it has been shown that TNF expressed on the membrane of CD4 T cells, together with the TNF-R1, participate in T-B cell interactions, resulting in enhanced immunoglobulin production (Aversa et al., 1993;Del-Prete et al., 1994). Our data unequivocally demonstrate that TNF and/ or LTc play a role in Ig class switching. Since no defect in the Ig class switch has been observed in TNF-R-1 and -2 deficient mice (Pfeffer et al., 1993;Rothe et al., 1993;Erickson et al., 1994), we propose that it is the signal from the LTo//3 heteromer through the LT/3-R that is necessary for isotype switching. Of note is also the finding that the defective Ig class switch was corrected by bone-marrow transplantation. Finally, germinal centers might not be necessary since a normal Ig class switch and affinity maturation was reported in LTc-deficient mice on immunization at a high protein dose (Matsumoto et al., 1996a).
In conclusion, transplantation of normal bone marrow into TNF-LTo mutant mice allowed functional and to some extent morphological correction of the immune defect seen in the TNF-LTo-deficient mice. Conversely, the mutant phenotype could be transferred by bone marrow cells transplanted into wild-type recipient mice.

MATERIALS AND METHODS
Mice TNF-LTc knockout mice were generated as described on a mixed genetic 129/Ev/Sv C57BL/6 background . Sevento ten-week-old wild-type and knockout mice bred in our animal facility and maintained under specific pathogen-free conditions were utilized.

Bone-Marrow Transplantation
Recipient mice received a lethal total-body irradiation (determined as 7.0 Gy in the knockout and wild-type mice), prior to the intravenous injection of 2 10 6 fresh, unseparated bone marrow cells.

Injection of Endotoxin
Mice were injected intraperitoneally with 10 #g of LPS (E. coli, serotype O111 :B4, Sigma, St. Louis) in saline solution (0.9%). Blood was taken at h/z after LPS injection and plasma was analyzed for TNF levels. Blood samples were obtained from the tail vein, the retroorbital sinus, or cardiac puncture at euthanasia. Plasma concentrations of TNF were determined by a cytotoxicity assay using the WEHI-164 clone 13 cells (Espevik and Nissen-Meyer, 1986).

Stimulation of Macrophages
Bone-marrow-derived macrophages were isolated from femurs and cultivated (106/ml) for 7 days in Dulbecco's minimal essential medium (DMEM) supplemented with 20% horse serum and 30% L929 cellconditioned medium (as source of M-CSF). After 7 days culture, the cell preparation contains 99% macrophages (Kelso et al., 1982). The bone-marrowderived macrophages were stimulated with LPS (10 ng/ml) for 3 h/z and the supernatant was assesssed for TNF activity as described.
Anti-SRBC Serum Isotype and Ig Levels Mice were immunized with 2 l0 SRBC i.v. and serum was collected before and at 3, 6, 9, and 15 days postimmunization. Serum levels of anti-SRBC-specifi Ig levels for IgM, IgG1, Ig2a, Ig2b, IgG3, and IgA were determined by a sandwich ELISA. Maxisorp microtiter plates were coated overnight with 50 /1 of a solubilized extract (3 /g/ml) from SRBC prepared according to (Kelly et al. 1979). Thereafter, plates were blocked with 2% BSA containing PBS for 2 h/z at 37C, and serial dilutions of the immune sera were incubated overnight at room temperature. Bound antibodies were detected with biotinylated goat antimouse Ig-isotype-specific antibodies (4 h/z at room temperature). Subsequently, streptavidin-conjugated to alkaline phosphatase and substrate were added for 45 min each, and the reaction was stopped with 1.5 M NaOH. Absorbance was read at 405 nm in a Titertek Multiskan spectrophotometer (Flow Laboratories). Serum titers are expressed as the reciprocal value of the dilution showing an optical density of 0.1 over background.
lmmunofluorescence Staining of Tissues Tissues were snap frozen in supercooled (liquid nitrogen) isopentane and stored at -80C, as previously described (Ryffel et al., 1991). Routinely prepared frozen sections were cut at 5/zm and fixed in acetone for 10 min, blocked in wash solution (50 mM Tris/HC1, 150 mM NaC1) for 30 min, and then incubated overnught at 4C with rat anti-mouse mAb specific for CD3, B220, and FDC-M1. Controls, including incubation with secondary antibody alone, were uniformly negative. After two washes, sections were incubated with goat--anti-rat antibody labeled with Cy-3 at room temperature for h/z. After a final wash, the slides were mounted in a Shandon Immunomount and coverslipped. Immunostained sections were examined with a Laser Scan Microscope 320 (Carl Zeiss, Zirich) in a confocal mode with a excitation wavelength of 543