A Role for Endogenous Transforming Growth Factor β1 in Langerhans Cell Biology:  The Skin of   Transforming Growth Factor β1 Null Mice Is Devoid of  Epidermal Langerhans Cells

TGF-β1 −/− and littermate control (+/+) mice were derived from matings of mice that were heterozygous for a null mutation of the TGF-β1 gene (generated [4] and supplied by A. Kulkarni and S. Karlsson [National Institute of Neurological Disorders and Stroke, Bethesda, MD]). Since their derivation, heterozygous males have been backcrossed to C57BL/6 females four times and the colony has been maintained by interbreeding. Thus, although mice in this colony express only H-2^b MHC antigens, each animal has a mixture of 129Sv and C57BL/6 background genes and is unique. TGF-β1 −/− and littermate control mice used in these studies were reared in a pathogen-free facility and were used at 8–20 d of age. All animals were housed and used in experiments in accordance with institutional guidelines.

Epidermal cell suspensions were prepared from trunk skin by limited trypsinization (0.25% trypsin [USB, Cleveland, OH] in HBSS without calcium or magnesium at 4°C for 18 h) and trituration in HBSS, 0.05% DNase, 30% FCS. Epidermal sheets were prepared from ear skin by incubation in 0.5 M ammonium thiocyanate (37°C for 20 min), fixed in acetone (−20°C for 30 min), and rehydrated in PBS (14).

Hybridomas secreting mAb Y3-P (anti-I-A^b), MK-D6 (anti-I-A^d), and N418 (anti-CD11c) were obtained from American Type Culture Collection (Rockville, MD). G8.8 (anti-gp40) was produced and characterized as described (15). mAbs Y3-P, MK-D6, and N418 were purified from hybridoma supernatants by protein A affinity chromatography (Pierce Chemical Co., Rockford, IL); G8.8 was purified using protein G (Pierce Chemical). These mAb were modified with FITC (Sigma) or NHS–LC–biotin (Bio; Pierce Chemical) as described (16). The following mAbs were purchased from PharMingen (San Diego, CA): FITC–anti-TCR-αβ (H57-597), FITC–anti-TCR-γδ (GL3), FITC–Thy-1.2 (30-H12), Bio–anti-CD45 (30F11.1), and relevant isotype controls. Phycoerythrin– streptavidin (PE–SA) was purchased from TAGO, Inc. (Burlingame, CA).

For flow cytometry, cells were suspended in cold PBS containing 5% FCS and 0.01% NaN₃ and preincubated with saturating concentrations of 2.4G2 (anti-FcRγII [17]) followed by FITC–mAb, Bio–mAb, and PE–SA. Surface antigen expression was analyzed using a FACScan^® flow cytometer equipped with Research Software (Becton Dickinson, Mountain View, CA). Propidium iodide permeable cells were excluded by live gating. Epidermal sheets were stained for LC and DETC with FITC–anti-I-A or FITC–anti-TCR mAb diluted in PBS/FCS/NaN₃, washed, and analyzed by epifluorescence microscopy.

Epidermal cells derived from trunk skin of TGF-β1 −/− and control mice were cocultured in flatbottomed 96-well plates with 2 × 10⁵ accessory cell–depleted T cells prepared from the skin-associated lymph nodes of female BALB/c mice (18) for 120 h at 37°C. [³H]TdR (1 μCi/well) was added for the final 12 h of the culture period. Cell-associated radioactivity was determined by direct β counting.

Cells reactive with various lineageselective mAbs were localized in lymph nodes using a three-step immunohistochemical procedure (19). In brief, acetone-fixed frozen sections of tissues were washed with PBS and incubated with hybridoma supernatants. After washing, mAbs were detected using digoxigenin-modified anti-rat (or anti-hamster) IgG Ab (Pierce Chemical), peroxidase-conjugated sheep anti-digoxigenin Fab, and 3,3′-diaminobenzidine. Digoxigenin-3-O-methylcarbonyl-ε-aminocaproic acid, N-hydroxy-succinimidyl ester, and anti-digoxigenin Ab were purchased from Boehringer Mannheim (Indianapolis, IN) and used as suggested by the manufacturer.

Rapamycin (Wyeth-Ayerst; Princeton, NJ) was dissolved in 0.2% high viscosity carboxymethyl cellulose (Sigma)/0.25% polysorbate 80 in dH₂0 and administered to all progeny resulting from the mating of several breeding pairs of TGF-β1+/− mice by i.p. injection (4 mg/kg) on postnatal day 10 and 3×/wk thereafter.

To begin to characterize epidermal leukocytes in TGF-β1 −/− mice, cells were prepared from the trunk skin of −/− and +/+ littermates and examined for the simultaneous expression of CD45 and I-A antigens, and CD45 and TCR-γδ (or Thy1.2) via multicolor flow cytometry. Although TGF-β1 +/+ mice contained a normal contingent of LC (CD45⁺ I-A^b+ cells), LC could not be identified in TGF-β1 −/− epidermis (see Fig. 1 for results with 17-d-old mice). However, the frequency of DETC (CD45⁺ TCR-γδ⁺ cells) among TGF-β1 −/− epidermal cells was normal. In addition, essentially all of the leukocytes (CD45⁺ cells) present in TGF-β1 −/− epidermis were DETC. This latter result excludes the existence of a significant population of immature LC (CD45⁺ I-A^b− TCR-γδ⁻ cells) in TGF-β1 −/− epidermis. The CD45⁻ I-A⁺ cells detected in TGF-β1 −/− epidermis (see Fig. 1,a) represent keratinocytes that inappropriately express class II MHC antigens; this finding is consistent with the previous report of disordered MHC Ag expression in TGF-β1 −/− mice (8). The LC deficiency in TGF-β1 −/− mice was confirmed in six additional experiments with mice ranging in age from 7–18 d (Table 1). Younger animals were not studied because normal numbers of epidermal LC would not be expected before postnatal day 7. Older animals were not studied because the health status of TGF-β1 −/− mice deteriorates rapidly after 18 d.

LC were also quantitated in TGF-β1 −/− and wildtype epidermis in situ. Immunofluorescence microscopy of epidermal sheets prepared from the ear skin of TGF-β1 −/− mice confirmed that LC were absent; DETC were normal in number and morphology (data not shown). In addition, epidermal cell suspensions from TGF-β1 −/− mice failed to initiate a proliferative response in naive BALB/c T cells (Fig. 2). Thus, LC (or LC precursors) could not be detected in the epidermis of TGF-β1 −/− mice, or in epidermal cell suspensions prepared from these animals, using assays of LC surface phenotype or function.

LC represent the epidermal contingent of the dendritic cell (DC) lineage (20). To determine whether lymphoid DC were present in TGF-β1 −/− mice, we examined skin-draining lymph nodes for several leukocyte subpopulations. Axillary lymph nodes were sectioned and stained with several lineage-selective mAbs (see Fig. 3). CD11c⁺ (N418-reactive) DC were readily identified in T cell–dependent regions of lymph nodes from TGF-β1 −/− as well as control mice. In contrast, cells reactive with anti-gp40 (G8.8), another DC-selective mAb (21), were absent from TGF-β1-deficient lymph nodes. F4/80⁺ macrophages were similarly represented and distributed in both TGF-β1 −/− and TGF-β1 +/+ tissue. These results suggest that although lymphoid DC are present in TGF-β1 null mice, all subpopulations may not be normally represented. Alternatively, TGF-β1 may be required for expression of certain lymphoid DC differentiation antigens.

The TGF-β1 −/− phenotype is characterized by marked inflammation of several major organs, wasting, and death at an early age (3, 4). The skin is not prominently involved in this inflammatory syndrome. Histopathologic examination of 17 d TGF-β1 −/− skin revealed only a sparse interstitial inflammatory infiltrate that was difficult to distinguish from the minimal infiltrate present in controls (data not shown). Nonetheless, to exclude the possibility that aberrant cytokine secretion associated with inflammation (in skin or elsewhere) influenced the ability of LC (or LC precursors) to develop or localize in TGF-β1 −/− epidermis, we assessed the LC content of the skin of TGF-β1 −/− mice that were treated with the immunosuppressive rapamycin (22). Rapamycin (4 mg/kg) was administered i.p. to littermates born to TGF-β1 heterozygotes beginning at 10 d of age and 3×/wk thereafter. This regimen obviated the wasting that is invariably present in untreated TGF-β1 −/− mice older than 14 d (Fig. 4), and almost completely inhibited the inflammation in lungs, heart, and liver that is characteristic of TGF-β1 −/− mice (data not shown).

Rapamycin-treated TGF-β1 −/− mice not only survived indefinitely, but gained weight for the duration of the experiments (39 and 42 d). Although the most dramatic components of the TGF-β1 −/− phenotype were eliminated by rapamycin treatment, epidermal sheets from ears of rapamycin-treated TGF-β1 −/− mice were devoid of LC (data not shown), as were epidermal cell suspensions prepared from trunk skin of these mice (Table 2). Rapamycin treatment had no effect on LC in TGF-β1 +/+ or +/− mice and, as in the case of untreated mice, DETC were the only leukocytes present in appreciable numbers in TGF-β1 −/− epidermis. Thus, the LC deficiency in TGF-β1 −/− mice does not result from the inflammatory syndrome that causes the demise of these animals, and cannot be attributed simply to delayed development.

TGF-β1 −/− mice represent the first animal model in which an absolute deficiency of LC (or other nonlymphoid DC) has been identified. Decreased numbers of LC were previously documented in the op/op mouse (23), a null mutant of the M-CSF (CSF-1) gene, but the LC deficiency in the TGF-β1 −/− mouse is much more profound. Although the LC deficiency is the most obvious alteration in the DC lineage in TGF-β1 −/− mice, lymphoid DC may also be abnormal. DC expressing gp40 (15, 21) were absent from skin-draining lymph nodes of TGF-β1 −/− mice, whereas CD11c⁺ DC were readily detected. Perhaps gp40-bearing DC in lymph nodes represent a distinct subpopulation of DC that in this site are derived exclusively from epidermal LC. Alternatively, TGF-β1 may modulate cell surface expression of certain DC antigens. The deficiency of epidermal LC in TGF-β1 −/− mice reported here also constitutes one of the most striking abnormalities in hematopoetic cells that has been documented in these animals. Previous studies focused primarily on abnormalities in the representation of mononuclear cell subpopulations in lymphoid tissues, and the immune dysregulation that underlies the inflammatory wasting syndrome that is associated with this genotype (6, 7, 24). The relationship of the LC/DC abnormalities in TGF-β1 −/− mice to other features of the phenotype is unknown, although abnormalities in DC may contribute to the decreased T cell mitogenic responses that are seen in TGF-β1 −/− mononuclear cells (24).

The nature of the requirement for TGF-β1 as a regulator of LC has not been determined. TGF-β1 may be required for proliferation or differentiation of LC precursors. This suggestion is compatible with the recent report that TGF-β1 promotes the in vitro expansion of DC in serumfree cultures of human CD34⁺ hematopoetic progenitors (25). Alternatively, TGF-β1 may influence the ability of LC precursors (or LC) to localize in epidermis or facilitate LC survival. Candidate mechanisms for TGF-β1 action include modulation of cytokine production (3), induction of growth factors or growth factor receptors (26), and regulation of adhesion molecule expression or function (27). Regulation could occur at transcriptional or posttranscriptional levels. Interestingly, there are a number of similarities between TGF-β1 null mice and mice that are deficient in the transcription factor relB (28, 29). Inflammatory syndromes and DC abnormalites are present in both models. RelB −/− mice have LC, but lack lymphoid DC. In contrast, TGF-β1 −/− mice have lymphoid DC, but lack LC. These results, in light of the recent determination that TGF-β1 influences the activity of NF-κB/Rel proteins in murine B cells by regulating levels of the inhibitor IκBα (30), suggest that TGF-β1 may regulate DC via a similar pathway. Resolution of these questions should result in elucidation of mechanisms by which murine LC/DC are regulated in vivo.

The authors thank Vivian McFarland for technical assistance, Harry Schaefer for help preparing the figures, Dr. Suren Sehgal (Wyeth-Ayerst) for supplying rapamycin, and Drs. George L. Barnes, Thilo Jakob, Stephen I. Katz, and Kim B. Yancey for reviewing the manuscript.

This work was supported in part by grants from the National Institutes of Health (AI 24137 and AG 04350 to A.G. Farr).