- β 2m:
Natural killer T cell
Mechanisms responsible for the development of autoimmune skin disease in humans and animal models with lupus remain poorly understood. In this study, we have investigated the role of CD1d, an antigen-presenting molecule known to activate natural killer T cells, in the development of inflammatory dermatitis in lupus-susceptible MRL-lpr/lpr mice. In particular, we have establishedMRL-lpr/lpr mice carrying a germ-line deletion of the CD1d genes. We demonstrate that CD1d-deficient MRL-lpr/lpr mice, as compared with wild-type littermates, have more frequent and more severe skin disease, with increased local infiltration with mast cells, lymphocytes and dendritic cells, including Langerhans cells. CD1d-deficient MRL-lpr/lpr mice had increased prevalence of CD4+ T cells in the spleen and liver and of TCRα β +B220+ cells in lymph nodes. Furthermore, CD1d deficiency was associated with decreased T cell production of type 2 cytokines and increased or unchanged type 1 cytokines. These findings indicate a regulatory role of CD1d in inflammatory dermatitis. Understanding the mechanisms by which CD1d deficiency results in splenic T cell expansion and cytokine alterations, with increased dermal infiltration of dendritic cells and lymphocytes in MRL-lpr/lpr mice, will have implications for the pathogenesis of inflammatory skin diseases.
MRL-lpr/lpr (MRL-lpr) mice spontaneously develop inflammatory lesions in skin, kidneys and blood vessels, along with marked lymphoproliferation, autoantibody production and hypergammaglobulinemia 1–6. These mice have been widely used as a model to study the pathogenesis of SLE, a multi-organ autoimmune disease 7. Skin lesions that resemble human discoid lupus erythematosus typically develop on the dorsum of the neck, forehead and ears of MRL-lpr mice. Histology of these lesions reveals epidermal hyperplasia, vacuolar cell changes, cellular infiltration and epidermal ulcerations 3, 4. These lesions markedly worsen in MRL-lpr mice that have been rendered deficient in β 2-microglobulin (β 2m) 8. β 2m-deficient MRL-lpr mice, however, experience mild nephritis and have reduced levels of autoantibodies 8–10. While serum IgG1 levels are reduced, IgM, IgG2a and IgG3 levels remain elevated in β 2m-deficient MRL-lpr mice 8, 9. The mechanism(s) by which β 2m deficiency differentially regulates the expression of various manifestations of lupus is unclear.
Since β 2m is required for the optimal expression of MHC class I and class I-related proteins with which it noncovalently associates, β 2m deficiency can affect various molecules such as classical MHC class I, CD1d, Qa-1 and neonatal Fc receptor 11. One possibility is that different β 2m-dependent molecules have different effects on the development of various manifestations of lupus. For example, amelioration of kidney disease in β 2m-deficient MRL-lpr mice may be due to deficiency of neonatal Fc receptor 10, which plays a role in the regulation of serum Ig levels 10, 12. However, decreased Ig levels would not account for the exacerbation of dermatitis in β 2m-deficient MRL-lpr mice. Another possibility is that the lack of conventional MHC class I-restricted CD8+ T cells is responsible for the increased skin disease in β 2m-deficient MRL-lpr mice. However, CD8-deficient MRL-lpr mice have no increase in skin disease, at least until 16 weeks of age 13. A third possibility is that a deficiency in regulatory CD1d-dependent natural killer T (NKT) cells is responsible for the increased skin disease in β 2m-deficient MRL-lpr mice. In fact, MRL-lpr mice exhibit a selective reduction in the numbers and functions of invariant (Vα 14Jα 18) NKT cells before the onset of clinical disease 14. Other studies have also found a specific decrease in the expression of invariant Vα 14 TCR mRNA by NKT cells before the onset of disease in MRL-lpr mice 15 and in the numbers of NK1.1-expressing cells in C57BL/6-lpr mice 16. Consistent with a protective role of CD1d-reactive T cells, patients with SLE also have a selective reduction of NKT cells 17–19. Finally, activation of these cells can reduce autoantibody production and protect against various immune-mediated diseases including type 1 diabetes and experimental autoimmune encephalomyelitis 20–26.
The role of NK1.1+ cells, which include NK cells, classical NKT cells and small subsets of other cell types that express this marker, in the development of autoantibodies in C57BL/6-lpr mice has been investigated 16. These studies showed that: (a) NK1.1+ cells inhibit anti-DNA Ab-secreting cells in vitro; (b) injection of these animals with anti-NK1.1 Ab results in increased numbers of anti-DNA Ab-forming spleen cells ex vivo; and (c) adoptive transfer of C57BL/6-lpr mice with B cell-depleted spleen cells, which contained 30–40% NK1.1+CD3+ cells, 20–30% NK1.1+CD3– cells and 30–50% non-NK/ NKT cells, results in delayed appearance of anti-DNA Ab-secreting spleen cells ex vivo16. Thus, this report suggests a regulatory role for NK1.1-expressing cells on autoantibody-producing B cells. However, the effect on in vitro anti-DNA Ab-forming cells observed in this study could have been due to either NKT (NK1.1+CD3+) or classical NK (NK1.1+CD3–) cells. Moreover, some NKT cells, e.g. most invariant NKT (CD1d-α GalCer tetramer+) cells in MRL-lpr mice, do not express the NK1.1 marker 14. The latter CD1d-reactive T cells would not have been depleted by treatment with anti-NK1.1 Ab. Finally, this study did not report on any clinical manifestations of lupus disease.
NKT cells recognize glycolipid antigens in the context of the non-classical MHC class I molecule CD1d 27. Two closely related genes, CD1d1 and CD1d2, share 95% sequence homology and encode mouse CD1 28. CD1d1 is widely expressed on B and T cells, macrophages and dendritic cells 29, 30, whereas CD1d2 is detected mainly on thymocytes and is considered nonfunctional 30.
We generated CD1d-deficient MRL-lpr mice by crossing the CD1d1- or CD1d1d2-null genotypes 31, 32 onto the MRL-lpr background and evaluated the effects of CD1d deficiency on the development of inflammatory dermatitis. Our results demonstrate that CD1d deficiency exacerbates the frequency and severity of skin lesions in MRL-lpr mice.
2.1 CD1d deficiency exacerbates inflammatory skin lesions in MRL-lpr mice
To investigate the role of CD1d in the development of lupus-like disease, we generated CD1d knockout (KO) MRL-lpr mice by crossing the CD1d1-null genotype from the stock C57BL6/Sv129 mice 31 onto the MRL-lpr background. The final CD1d1 KO MRL-lpr mice and their wild-type (WT) littermates were monitored for the development of skin, kidney and lymphoproliferative disease, and their sera were tested for autoantibodies.
MRL-lpr mice develop inflammatory skin lesions that typically manifest as hair loss and scab formation on the dorsal neck region as well as cellular infiltration and epidermal hyperplasia visible on histology 3, 4. We monitored the development of such skin lesions and scored the skin biopsy sections from dorsal regions of the neck for cellular infiltration, epidermal hyperplasia and epidermal ulcerations. As shown in Fig. 1A and 1B, the cumulative frequency of skin lesions and composite skin biopsy scores were increased in CD1d1 KO mice as compared to WT mice. Five of 11 (45%) WT mice had a minimal (0.5) to mild (1+) inflammation, and epidermal hyperplasia or focal ulcerations were detectable in two separate mice, whereas 7 of 9 (77%) CD1d1 KO mice had a moderate (2+) to mild (1+) cellular infiltration and epidermal hyperplasia, with multifocal ulcerations in 2 of these mice (Fig. 1B). Whereas 16-week-old WT MRL-lpr mice had a minimal infiltration with neutrophils, mast cells and lymphocytes, the CD1d1 KO littermates had an intense infiltration with mast cells, neutrophils and lymphocytes (Fig. 1C). Most infiltrating lymphocytes in both WT and KO mice were CD3+ T cells, as determined by immunohistochemistry (data not shown). Skin sections from the KO animals also exhibited other features of chronic cutaneous lupus, such as vacuolar cells in the basal layers of epidermis and fibroplasia in the dermis (Fig. 1C).
Activated dendritic cells, such as Langerhans cells, infiltrate the skin during the active stages of spontaneous skin lesions in humans with SLE and in MRL-lpr mice 3, 33. To investigate if exacerbation of skin disease in CD1d KO mice is associated with an expansion of these cell types, we performed immunohistochemistry to detect resident skin dendritic cells (Fig. 1D, E). Although occasional cells in the epidermis of WT mice expressed S100 antigen 34 and displayed the morphology of Langerhans cells, these cells were clearly increased in the dermis of CD1d KO mice (Fig. 1D). However, not all S100-expressing cells had the morphology of Langerhans cells, and the S100 antigen is known to be expressed by melanocytes, adnexal glands and Schwann cells of the skin 34. We therefore stained skin sections with anti-CD11c; a marked increase in CD11c-expressing cells was found in KO animals as compared to WT littermates (Fig. 1E). Thus, CD1d KO MRL-lpr mice exhibit a significant expansion of skin dendritic cells.
MRL-lpr mice also develop nephritis 6 and autoantibodies to nuclear antigens 1, 2. We found that renal histological changes, severe proteinuria, serum anti-DNA Ab levels and total serum IgM, IgE, IgG1, IgG2a, IgG2b and IgG3 levels were not significantly different between CD1d KO and WT littermates (Fig. 2 and data not shown).
In the absence of the CD1d1 gene, the second murine CD1d gene, CD1d2, might partially or completely compensate for the genetic deficiency of CD1d1. To address this possibility, we crossed the CD1d1/d2-null 32 genotype onto the MRL-lpr background and examined the resulting KO mice for the above-mentioned parameters of lupus at 3–4 and 5–7 months of age. The results in the double KO mice essentially corroborated the findings obtained with the single KO MRL-lpr mice.
2.2 Increased prevalence of CD4+ and TCRα β +B220+ cells in CD1d KO MRL-lpr mice
MRL-lpr mice develop enlargement of lymphoid organs, with accumulation of conventional T cells and an abnormal cell population, double-negative TCRα β +B220+ cells 1, 2. To address the effect of CD1d deficiency on these manifestations, we examined the extent of organomegaly and cellularity as well as the surface phenotypes of lymphocytes in various lymphoid tissues by flow cytometry (Fig. 3, 4). While the overall cellularity and organ weights in WT and KO mice were similar (Fig. 3), the TCRα β +B220+ cell population was significantly increased (Fig 4A), whereas the percentage of B cells was decreased (Fig. 4B), in the lymph nodes of 3-month-old CD1d KO MRL-lpr mice. In the spleens and livers of CD1d KO mice, CD4+ T cells were significantly increased as compared to WT littermates (Fig. 4B). The absolute cell numbers also showed a similar trend, although the differences did not reach statistical significance. No significant differences in lymphocyte phenotypes were observed in older (6 months) KO mice vs. WT littermates (data not shown).
2.3 Decreased type 2 cytokines (IL-4 and IL-10) in CD1d KO MRL-lpr mice
Abnormalities in cytokine production have been suggested to contribute to the development of disease in MRL-lpr mice 35. Since repeated activation of CD1d-reactive NKT cells can alter the overall cytokine balance towards a type 2 cytokine profile in MRL-lpr mice 14, the lack of such cells in CD1d KO mice may result in decreased type 2 and/or increased type 1 immune responses. Indeed, while the production of the type 1 cytokine IFN-γ by Con A-stimulated splenic T cells was slightly increased, the type 2 cytokines IL-4 and IL-10 were significantly decreased in CD1d KO mice as compared to WT animals (Fig. 5).
Our results indicate that CD1d deficiency exacerbates inflammatory skin disease in MRL-lpr mice. To explore the mechanisms by which CD1d deficiency may affect the development and course of inflammatory skin disease, we investigated its effect on lymphocyte populations and their cytokine responses. CD4+ T cells were consistently increased in the spleens of CD1d KO MRL-lpr mice, and CD3+ T cells infiltrated their skin (Fig. 4B and data not shown). Such T cell expansion may contribute to disease exacerbation, as TCRα β + cells are essential for the development of disease in MRL-lpr mice 36. Furthermore, upon T cell stimulation, spleen cells from CD1d KO MRL-lpr mice produced lower levels of the immunoregulatory cytokines IL-4 and IL-10 than the WT littermates (Fig. 5), which could be responsible for skin disease exacerbation in CD1d KO mice. In fact, a recent study described significant acceleration of skin disease in IL-10 KO MRL-lpr mice 37. IL-10 also plays a critical role in the development of immune tolerance 38. Thus, IL-10 produced by CD1d-reactive NKT cells may play a role in the development of immune tolerance that can be facilitated by CD1d-reactive NKT cells 39. It remains to be determined whether loss of this mechanism results in the activation of autoreactive T cells and exacerbation of autoimmunity in CD1d-deficient MRL-lpr mice. Future experiments are also needed to determine whether imbalances in type 1 versus type 2 cytokines or a reduction in regulatory cytokines or the regulatory T cells that secrete such cytokines result in the expansion of effector CD4+ T cells that infiltrate target organs such as skin.
There is evidence of significant cross-talk between CD1d-reactive T cells and dendritic cells 40, 41, resulting in alterations in the differentiation/maturation of dendritic cell subsets, which may contribute to modulation of autoimmune responses 22. CD40/CD40L interactions may play an important role in this cross-talk 41. In this regard, we found that skin dendritic cells including Langerhans cells, which are known to infiltrate the dermis during the active stages of spontaneous skin lesions in SLE 3, 33, were significantly increased in CD1d KO MRL-lpr mice (Fig. 1D, E). Our preliminary studies suggest that the expression of CD40 as well as CD40L, as detected by immunohistochemistry, is also increased in skin sections from CD1d KO MRL-lpr mice as compared to WT littermates (data not shown). Ongoing studies will examine whether deficiency in CD1d and CD1d-reactive T cells affects dendritic cell infiltration, which then contributes to local infiltration with activated T cells and inflammation in target organs such as skin. Alternatively, dendritic cell infiltration may be an indirect effect of inflammation.
An intriguing finding of our study is that CD1d deficiency exacerbates skin disease but does not seem to affect circulating levels of autoantibodies or renal disease. These data suggest that different regulatory mechanisms might account for the various manifestations of lupus. The regulatory effect of CD1d-restricted events may be critical in controlling the local inflammation in the skin but dispensable for the regulation of nephritis. It is possible, however, that the time point that we chose for the detection of skin disease may not be optimal for the detection of renal pathology, as MRL-lpr mice develop renal disease before the onset of skin disease. Nevertheless, proteinuria and renal histological changes were not significantly different in the KO and WT littermates in two separate cohorts. A similar differential regulation of lupus skin vs. renal disease was also observed in β 2m-deficient MRL-lpr mice, which had worsened skin lesions, whereas nephritis was ameliorated 8. In that study, however, CD1d-deficient and CD1d-sufficient MRL-lpr mice had similar incidences of skin and renal disease. The discrepancy between that study and ours could arise from variability in the expression of skin disease in various animal colonies and cohorts. For example, in the previous study 8, the age at which 50% of the WT MRL-lpr mice developed skin disease was 18 weeks in the CD1d cohort and 31 weeks in the β 2m cohort, whereas in our animal colony, skin disease appears at approximately 28 weeks (data not shown). Different strategies used to generate the KO mice could also affect the CD1d KO phenotype. In the previous study 8, data obtained from N6 and N10 CD1d KO MRL-lpr mice were pooled, whereas we analyzed data in separate cohorts of N8 or N7 CD1d KO MRL-lpr mice. The N6 to N10 backcrossed mice are expected to carry approximately 0.1–2% founder 129/B6 genes, which may affect the phenotype of CD1d KO MRL-lpr mice. Although our genotype analyses of congenic strains using simple sequence repeat markers centered on mostknown lupus-suppressor or -susceptibility loci does not suggest a replacement with donor genes at these loci (data not shown), we cannot rule out the possibility that observed effects of the KO reflect the introduction of an adjacent gene cotransferred during the backcross of the mutated 129 locus onto the MRL-lpr genetic background.
In contrast to the lack of an effect of CD1d deficiency on the development of nephritis in MRL-lpr mice, pristane-injected CD1d KO BALB/c mice develop more severe lupus-like nephritisthan their WT littermates 26. Lupus-prone NZB/NZW F1 mice, when rendered deficient in CD1d, also experience an exacerbation of lupus nephritis (manuscript in preparation). In addition, old (>1.5 years) CD1d-deficient BALB/c mice have increased serum anti-DNA Ab levels as compared to age-matched controls (S. Porcelli, personal communication). More direct evidence for the involvement of CD1d-reactive NKT cells in regulation of lupus-like autoimmunity comes from aging B6 Jα 18 KO mice, which have elevated anti-DNA and anti-cardiolipin Ab and renal IgG and complement deposition as compared to age-matched B6 controls (S. Porcelli, F. Dieli and G. Sireci, personal communication). It is therefore possible that the regulatory effect of CD1d-restricted events on nephritis require intact Fas signaling, which is absent in MRL-lpr mice, or that the anti-apoptotic effects of mutant Fas ligand are capable of bypassing the role of CD1d-reactive T cells.
In a previous study, adoptive transfer of NK1.1-enriched (NK1.1+CD3+ and NK1.1+CD3–) cells reduced the number of CD3+B220+ cells in C57BL/6-lpr mice 16. This regulatory effect of NK1.1-expressing cells is probably conferred by CD1d-reactive NKT cells, since CD1d KO MRL-lpr mice have an increased TCRα β +B220+ cell population (Fig. 4A). However, the CD1d deficiency per se does not appear to directly cause the expansion of this cell population, asthe numbers of TCRα β +B220+ cells were comparable in the peripheral lymphoid tissues of normal B6 and CD1d–/–B6 mice (our unpublished data). Furthermore, whereasup-regulation of CD1d on dendritic cells has been found to lead to excessive negative selection of NKT cells 42, it appears to have no effect on the conventional CD8 and CD4 compartments or on TCRα β +B220+ cells 43. The levels of CD1d expression on dendritic cells and thymocytes in MRL-lpr mice are also comparable to those of B6mice (our unpublished data). Thus, it is unlikely that the abnormal accumulation of CD4–CD8–TCRα β +B220+ cells seen in MRL-lpr mice occurs as aconsequence of abnormal CD1d-mediated positive and/or negative selection in the thymus. However, we cannot completely exclude the possibility that increased accumulation of this unusual population in CD1d KO MRL-lpr mice occurs as a result of CD1d deficiency influencing central vs. peripheral tolerance.
In conclusion, CD1d appears to play a protective role in the development of lupus dermatitis. Indeed, treatment with α -galactosylceramide, which results in the expansion and activation ofCD1d-reactive NKT cells, alleviates dermatitis in MRL-lpr mice 14. Humans with SLE and other systemic autoimmune diseases also have reduced numbers of NKT cells, and disease activity in patients with SLE appears to inversely correlate with circulating NKT cell numbers 17–19. Understanding the mechanisms by which CD1d-restricted events regulate the development of lupus dermatitis in animal models and subsets of lupus patients should prove invaluable.
4 Materials and methods
Stock MRL-lpr mice were purchased from the Jackson Laboratory (Bar Harbor, ME). To generate CD1d-deficient MRL-lpr mice, CD1d1-null 31 or CD1d1/d2-null 32 stock mice were crossed onto the MRL-lpr background for eight and seven generations, respectively. The CD1d-null genotypes were identified by Southern blotting 31 or PCR 26. The CD1d KO phenotype was further confirmed by flow cytometry of peripheral blood cells stained with a conjugated anti-CD1d mAb (1B1; PharMingen, San Diego, CA). Mice at the final backcross were screened to confirm that they were indeed congenic for the disease-associated simple sequence repeat markers (including D2Mit324, D3Mit42, D3Mit137, D4Mit12, D4Mit89, D4Mit125, D4Mit147, D4Mit338, D5Mit356, D7Mit211, D10Mit11 and H2-Aa1 44; www.informatics.jax.org). All but one of the polymorphic markers discriminated the congenic MRL-lpr mice from the B6 or 129 strains. One of the 12 primer sets discriminated the 129 strain from only 70% of the congenic MRL-lpr mice; however, the 129 marker wasequally distributed among the WT and KO littermates in the congenic mice.
4.2 Assessment of skin disease
Inflammatory skin lesions on the forehead, ears and dorsum of the neck 3, 4, 14 were scored on a scale of 0–3, where 0= no visible skin changes, 1= minimal hair loss with redness and a few scattered lesions, 2= redness, scabbing and hair loss with a small area of involvement and 3= ulcerations with an extensive area of involvement. Forhistology, skin biopsies from the back of neck were stored in 4% paraformaldehyde and sectioned. The hematoxylin and eosin (H & E)-stained sections were independently scored by a pathologist (Greg Boivin) and two of the authors (R.R.S. and J.Y.) for cellular infiltration (score 0–3), epidermal hyperplasia (0–2) and epidermal ulcerations (0–2). Results are expressed as an average of total scores from all readers.
4.3 Skin immunohistochemistry
Frozen sections of skin biopsies were fixed in acetone for 5 min, followed by washing, blocking and incubation with hamster anti-mouse CD11c and CD40L (e-Bioscience, San Diego, CA), hamster anti-CD3 (BD PharMingen), rabbit anti-mouse CD40 (Santa Cruz Biotechnology, Santa Cruz, CA) or isotype control IgG. After washing, sections were incubated with biotinylated goat anti-hamster IgG (e-Bioscience) or anti-rabbit IgG (Vector Lab, Burlingame, CA) and then stained using the Vectastain® ABC-AP kit and Vector® red alkaline phosphatase substrate kit I (Vector Lab) following the manufacturer's instructions. Staining for S-100 antigen was performed on paraffin sections using a rabbit polyclonal anti-S100 Ab with predominantly beta subunit specificity and the CONFIRMTM detection kit (Ventana, Tucson, AZ) following the manufacturer's instructions.
4.4 Assessment of renal disease
Proteinuria was measured using a colorimetric assay strip for albumin (Albustix, Bayer Co., Elkhart, IN), as described 45. For renal histology, paraffin sections were stained with H & E, periodic acid-Schiff and Masson's trichrome. Stained sections were scored for the glomerular, tubulointerstitial and vascular lesions, as described 6. The mean scores were summed to determine a composite kidney biopsy score.
4.5 Flow cytometry
Single cell suspensions from spleen and mesenteric and axillary lymph nodes were prepared by mechanical disruption in complete RPMI medium. Lymphocytes from perfused liver were isolated according to the method described by Goossens et al. 46. Red blood cells were removed by hypotonic lysis. Cells were stained with FITC-conjugated mAb specific for CD4 (RM4–5) and TCRβ (H57–597); PE-conjugated mAb specific for CD8α (53–6.7) and B220 (RA3–6B2); and CyChrome-conjugated mAb specific for CD3 (145–2C11) and TCRβ (H57–597). All mAb were purchased from PharMingen. Stained cells (106) were analyzed with a Becton Dickinson (Mountain View, CA) FACSCalibur flow cytometer using CellQuest software. Absolute cell numbers of each cell subset were calculated by the percentage of staining cells × total cell number.
4.6 Detection of autoantibodies and Ig isotypes
IgG anti-dsDNA Ab was measured by ELISA, as described 45. Rheumatoid factor (Ig κ chain anti-IgG2aa) was determined by ELISA, as described previously 47. Sera from normal BALB/c mice were used as negative controls. Total serum Ig and its isotypes were measured by a standard sandwich ELISA using the appropriate Ab pairs (Southern Biotechnology).
4.7 Detection of cytokines by ELISA
Splenocytes (1×106–2×106/ml) were stimulated with Con A (1–5 μ g/ml) or plate-bound anti-CD3 mAb (1–10 μ g/ml) for 48 h. The supernatants were tested for cytokines by ELISA using mAb pairs and recombinant cytokine standards from PharMingen, as described 38.
4.8 Statistical analysis
Levels of Ab and cytokines, lymphocyte percentages and numbers, and renal scores were compared using Graphpad InState software (San Diego, CA). The Student's t-test was used if the data followed a normal distribution; otherwise, a Mann-Whitney U-test was used. Frequencies of Ab and proteinuria were compared using the two-sided Fisher's exact test.
We thank Mark Shlomchik for providing reagents for the rheumatoid factor assay, Dwight Kono for suggestions on genotyping the congenic strains, Gregory Boivin and Diya Mutasim for suggestions on skin histology and immunohistochemistry, and Jonathan Jacinto and Tam Bui for the animal care and genotyping. This work was supported in part by grants from the National Institutes of Health (AR47322, AR50797 and DK69282 to R. R. S., AI43407 to C.-R. W. and HL68744 to L. V. K.).