MHC class I pathway is not required for the development of crescentic glomerulonephritis in mice

Abstract

MHC II and CD4⁺ T cells are required for anti-glomerular basement membrane (GBM) globulin-initiated crescentic glomerulonephritis (GN) in mice, but the role of MHC I and CD8⁺ T cells is unclear. The cytolytic function of CD8⁺ T cells requires recognition of peptide antigens presented on MHC I. CD8⁺ T cells can also perform helper functions via cytokine production. The contribution of MHC I to crescentic GN was investigated using TAP-1 gene knock out (TAP-1−/−) mice, which have deficient MHC I antigen presentation. Heterozygous TAP-1 mice have normal MHC I expression and developed GN with crescents in 42 ± 4% of glomeruli (normal 0%), proteinuria (9·1 ± 1·6 mg/20 h, normal 1·5 ± 0·3 mg/20 h) and impaired renal function (creatinine clearance 110 ± 8 μl/min, normal 193 ± 10 μl/min) following administration of sheep anti-mouse GBM globulin. TAP-1−/− mice, which have extremely low MHC I expression and reduced CD8⁺ T cells, developed similar GN with 39 ± 3% crescents, proteinuria (12·7 ± 4·3 mg/20 h) and impaired renal function (creatinine clearance 123 ± 20 μl/min). In vivo antibody-induced CD8 depletion did not attenuate crescent formation or protect renal function in C57Bl/6 mice developing GN, although significant reduction in proteinuria (5·3 ± 1·2 mg/20 h, P = 0·012) and glomerular recruitment of CD4⁺ T cells and macrophages were observed compared with control treated mice with GN. These data demonstrate that MHC I is not required for development of crescentic GN in mice. The MHC I-independent contribution of CD8⁺ T cells to proteinuria and inflammatory cell recruitment suggests that they may serve a ‘helper’ rather than cytolytic role in this disease.

INTRODUCTION

Glomerulonephritis (GN) results from immune responses to endogenous or planted glomerular antigens, which may potentially be presented through either MHC class I or MHC II-dependent pathways. Crescentic GN induced by a planted nephritogenic antigen in mice is dependent on MHC class II expression by professional antigen-presenting cells (APC) and intrinsic renal cells [1]. Immune complex-initiated GN in ‘lupus prone’ MRL/lpr mice also requires systemic MHC II. However, selective deficiency of MHC II in the kidneys of MRL/lpr mice does not protect against the development of immune complex GN [2].

The contribution of MHC I and CD8⁺ cells to crescentic GN is less clear. β₂-microglobulin (β₂-M)-deficient mice have very low expression of MHC I and deficient MHC I pathway-dependent antigen presentation. Deficiency of β₂-M in ‘lupus prone’ MRL/lpr mice significantly attenuates all manifestations of their spontaneous autoimmune disease, including autoantibody formation and immune complex GN [3]. However, β₂-M-deficient kidneys transplanted into MRL/lpr mice are not protected from development of GN [2]. Antibody depletion of CD8⁺ cells abolishes crescentic GN induced by a planted antigen in Wistar Kyoto rats [4], whereas in genetically CD8-deficient mice, crescent formation was accelerated in a planted antigen model [5].

The requirement for MHC I-dependent antigen presentation was studied in crescentic GN initiated by a planted glomerular antigen. The transporter associated with antigen processing (TAP) delivers peptides to endoplasmic reticulum (ER) for presentation by class I molecules. In TAP-1-deficient mice, MHC I can only bind a very restricted repertoire of antigenic peptides which are generated inside the ER. This results in extremely low cell surface expression of MHC I and defective MHC I-dependent antigen presentation for the majority of intracytoplasmic antigens usually recognized through this pathway [6]. TAP-1-deficient mice also have deficient MHC I-dependent cellular cytotoxicity and markedly reduced numbers of CD8⁺ T cells. The numbers and distribution of CD4⁺ T cells is unaffected. Heterozygote TAP-1 littermates have normal MHC I expression and normal CD8⁺ T cell repertoire.

In the current studies, the development of crescentic GN, induced by administration of anti-glomerular basement membrane (GBM) globulin, was compared in mice with homozygous deletion of TAP-1 (TAP-1−/−) and their heterozygous litter mates (TAP-1+/−) which have normal MHC I expression. The contribution of CD8⁺ T cells during the effector phase of the disease was determined by depletion using a monoclonal anti-murine CD8 antibody.

MATERIALS AND METHODS

Mice

TAP-1 gene knockout mice (TAP-1−/−), generated and characterized as previously described [6], were obtained from a colony at Monash University Central Animal Services. These mice have a mixed C57Bl/6 and 129/SV background. Heterozygous (TAP-1+/−) mice were used as controls. The phenotype of these mice was confirmed by flow cytometric analysis of MHC I expression on peripheral blood leucocytes using an FITC-conjugated anti-H-2K^b antibody (AF6-88.5; PharMingen, San Diego, CA). CD8 depletion studies were performed in C57Bl/6 mice. All experiments were performed on male mice between 12 and 14 weeks of age.

Induction of anti-GBM GN

Mice were immunized with 100 μg of sheep globulin in 50 μl Freund's complete adjuvant (FCA; Sigma, St Louis, MO) given subcutaneously into each flank. Fourteen days later, mice were given a single i.v. injection of 10 mg sheep anti-mouse GBM globulin, prepared as previously described [7,8]. Disease was assessed 14 days after administration of anti-GBM globulin.

Histological assessment of glomerular injury

Kidney tissue was fixed in Bouin's fixative, embedded in paraffin and 2-μm sections were stained using periodic acid-Schiff (PAS) reagent. Crescent formation was assessed on a minimum of 30 glomeruli, as previously described [8,9]. Glomerular T cell and macrophage infiltration was assessed on cryostat cut kidney tissue sections (6 μm) using two-layer peroxidase technique as described [1]. The primary MoAbs were H129 (anti-mouse CD4; American Type Culture Collection (ATCC), Rockville, MD), 53.6 (anti-mouse CD8), M1/70 (anti-mouse Mac-1; ATCC). The secondary antibody was goat anti-rat Ig–horseradish peroxidase (HRP) (Chemicon Int., Temecula, CA). Natural killer (NK) cells were detected using rabbit anti-mouse/rat asilo GM1 (Cedarlane, Hornby, Ontario, Canada), followed by swine anti-rabbit Ig–HRP (Dako, Glostrup, Denmark). Sections of spleen provided a positive control and protein G-purified rat immunoglobulin was substituted for primary MoAbs to provide a negative control. A minimum of 30 equatorially sectioned glomeruli were assessed per animal and the results were expressed as cells per glomerular cross section (c/gcs).

Functional assessment of renal injury

Proteinuria was quantified on urine collected over the final 20 h of each experiment. Urinary protein concentrations were determined by the Bradford method [10]. Creatinine clearance was calculated from the serum and urine creatinine concentrations, which were measured by the alkaline picric acid method using a Cobas Bio autoanalyser (Roche Diagnostic, Basel, Switzerland).

Assessment of systemic immune responses to sheep globulin

Humoral immune responses were assessed by mouse anti-sheep globulin antibody titres in serum collected at the end of each experiment and by immunofluorescence detection of binding of mouse immunoglobulin to the GBM. Circulating antibody titres were measured by ELISA as previously described [1]. Microtitre plates were coated with 20 μg/ml of sheep globulin. Dilutions of mouse serum (1:50–1:12 800) of mouse serum were added and binding was detected using a HRP-conjugated goat anti-mouse–Ig antibody (Silenus, Victoria, Australia). Colour was developed with ABTS and optical density (OD) was read at 405 nm. Glomerular binding of mouse immunoglobulin was assessed by immunofluorescence staining of frozen sections with FITC-labelled sheep anti-mouse immunoglobulin antibody (Silenus) as previously described [1]. Cell-mediated immunity was assessed by cutaneous DTH responses to sheep globulin. Twenty-four hours prior to the end of the experiment, mice were injected with 0·5 mg/20 μl of sheep globulin in the left footpad and 0·5 mg/20 μl irrelevant antigen (ovalbumin; Sigma) into the right footpad. Swelling was measured 24 h later using a micrometer and the DTH response was taken as the difference in swelling between the left and right footpad in each animal.

In vivo CD8 depletion

Anti-mouse CD8 MoAb YTS169.4 [11] was prepared from hybridoma supernatant by protein G purification. In vivo CD8 depletion was induced during the effector phase of the disease by i.v. injection of 1 mg of immunoglobulin, 2 h prior to administration of anti-GBM globulin and again 7 days later. The control group received similar doses of normal rat IgG. The extent of depletion of CD8⁺ T cells in blood was monitored by flow cytometry using a PE-conjugated anti-mouse CD8 MoAb (CT.CD8b; Exalpha, Boston, MA). Effects on other circulating lymphocytes were monitored using PE-conjugated KT.174 (anti-CD4; Exalpha), and FITC-conjugated KT3 (anti-CD3; Exalpha).

Experimental design and statistical analysis

Experimental protocols were in accordance with the National Health and Medical Research Council of Australia ‘Australian code of practice for care and use of animals for scientific purposes’ and were approved by the Monash University Animal Experimentation Ethics Committee. Anti-GBM GN was induced in the following groups of mice: TAP-1−/− (n = 8), TAP-1+/− (n = 6), C57Bl/6 treated with anti-CD8 antibody (n = 6) and C57Bl/6 treated with control antibody (n = 6). Results are expressed as the mean ± s.e.m., statistical analysis was performed by anova and Fisher's protected least significant differences (PLSD) test.

RESULTS

Effect of MHC I deficiency on development of crescentic GN

Sensitized TAP-1−/− mice given anti-GBM globulin developed severe crescentic GN with similar histological appearances to that induced in TAP-1+/− mice (Fig. 1). The percentage of glomeruli exhibiting crescent formation was 38·9 ± 2·8% in TAP-1−/− mice and 41·9 ± 3·7% in TAP-1+/− mice (normal 0%). TAP-1−/− mice with GN developed proteinuria 12·7 ± 4·3 mg/20 h (baseline 3·2 ± 0·2 mg/20 h, P = 0·0003) which was not significantly different from that induced in TAP-1+/− mice (9·1 ± 1·6 mg/20 h, baseline 1·5 ± 0·3 mg/20 h). Creatinine clearance was also reduced in TAP-1−/− mice with GN (123 ± 20 μl/min, baseline 223 ± 45 µl/min, P = 0·017) to a similar extent to that observed in TAP-1+/− mice with GN (110 ± 8 μl/min, baseline 175 ± 22 μl/min) (Fig. 2). Significant glomerular infiltration of CD4⁺ T cells and macrophages (compared with normal mice) was observed in both TAP-1−/− and TAP-1+/− mice with GN. Glomerular infiltration of CD8⁺ cells was observed in TAP-1+/− mice (P = 0·0002 compared with normal), but this was significantly reduced in TAP-1−/− mice (P < 0·0001 compared with TAP-1+/− mice with GN). NK cells were rarely detected in glomeruli in any group (Table 1).

Fig. 1.

Photomicrographs illustrating the histological appearances of the glomerulus in TAP-1−/− mice (A), TAP-1+/− mice (B), anti-CD8 antibody-treated mice (C) and control antibody-treated mice (D). All groups developed proliferative glomerulonephritis with prominent crescent formation. (PAS stain, mag. × 400.)

Fig. 2.

Crescent score, proteinuria and creatinine clearance in TAP-1−/− mice and TAP-1+/− mice before (□) and 14 days after (▪) initiation of anti-glomerular basement membrane glomerulonephritis. *P < 0·05 compared with normal (pre-disease) values.

Table 1.

Glomerular accumulation of macrophages, CD4⁺ cells, CD8⁺ cells and natural killer (NK) cells in normal mice and mice with crescentic glomerulonephritis (GN)

	Macrophages	CD4+ cells	CD8+ cells	NK cells
Normal mice	0·37 ± 0·09*	0·21 ± 0·4	0·29 ± 0·04	N/A
TAP-1−/− mice with GN	1·42 ± 0·15	1·58 ± 0·12	0·27 ± 0·04	0·10 ± 0·02
TAP-1+/− mice with GN	1·28 ± 0·11	1·20 ± 0·05	0·87 ± 0·1	0·07 ± 0·07
CD8-depleted mice with GN	1·02 ± 0·10†	0·99 ± 0·05‡	0·01 ± 0·01§	0·12 ± 0·02
Control-treated mice with GN	1·54 ± 0·08	1·47 ± 0·18	0·89 ± 0·1	0·11 ± 0·002

^*Cells per glomerular cross-section, mean ±s.e.m.

^†P < 0·0024 compared with control-treated.

^‡P = 0·013 compared with control-treated.

^§P < 0·0001 compared with control-treated, P = 0·038 compared with normal.

N/A, Data not available.

Effect of MHC I deficiency on systemic immune responses to sheep globulin

Circulating anti-sheep antibody titres were similar in TAP-1−/− and TAP-1+/− mice (Fig. 3) and no differences were observed in glomerular deposition of mouse immunoglobulin between the two groups. Skin swelling in response to sheep globulin challenge (TAP-1−/−0·19 ± 0·10 mm; TAP-1+/− 0·17 ± 0·10 mm) was similar in both groups, indicating equivalent cutaneous DTH responses to the nephritogenic antigen.

Fig. 3.

Serum titres of mouse anti-sheep globulin antibody in TAP-1−/− mice (○), TAP-1+/− mice (•), anti-CD8 antibody-treated mice (□) and control antibody-treated mice with glomerulonephritis (▪). The titration of serum from non-immunized mice is shown with black crosses.

Effect of CD8 depletion on development of crescentic GN and systemic immune responses

Administration of anti-CD8 MoAb resulted in 94% depletion of circulating CD8⁺ T cells. By flow cytometry, CD8⁺ cells comprised 19·2 ± 1·6% of circulating lymphocytes in control treated mice and only 1·1 ± 0·2% of lymphocytes in anti-CD8 antibody-treated mice, at the end of the experimental protocol. CD8 depletion did not alter circulating titres of anti-sheep globulin antibody (Fig. 3), glomerular deposition of mouse immunoglobulin or cutaneous DTH responses to sheep globulin (antigen-specific skin swelling: CD8-depleted 0·26 ± 0·10 mm, control-treated 0·22 ± 0·09 mm).

Both anti-CD8 antibody and control antibody-treated groups of mice developed severe crescentic GN (Fig. 1). The incidence of crescents was not different between the two groups (CD8-depleted 38 ± 2%, control 38 ± 3%). The reduction of renal function associated with development of GN was not affected by CD8 depletion (creatinine clearance, CD8-depleted 99 ± 8 μl/min; control-treated 107 ± 5 μl/min), although both groups had significantly impaired creatinine clearance compared with baseline (193 ± 10 μl/min, P = 0·0003 and P < 0·0001, respectively). Significant reduction in proteinuria was observed in CD8-depleted mice (5·3 ± 1·2 mg/20 h) compared with control-treated mice (12·3 ± 2·0 mg/20 h, P = 0·012) (Fig. 4). The glomerular infiltration of CD8⁺ cells was virtually abolished in anti-CD8 antibody-treated mice. In these mice, numbers of CD8⁺ cells were even less observed than in normal mice (P = 0·038). Glomerular accumulation of CD4⁺ cells (P = 0·013) and macrophages (P = 0·002) was reduced in anti-CD8 antibody-treated mice compared with control-treated mice (Table 1).

Fig. 4.

Crescent score, proteinuria and creatinine clearance in normal C57Bl/6 mice (□) and anti-CD8 antibody-treated mice (hatched bars) and control antibody-treated mice with glomerulonephritis (▪). *P < 0·05 compared with pre-disease (baseline) values; †P < 0·05 compared with control treatment.

DISCUSSION

In the current study, TAP-1-deficient mice were used to demonstrate that antigen presentation via the MHC I pathway is not required for the full expression of crescentic GN, initiated by a planted exogenous antigen. TAP-1-deficient mice lack one of two TAP proteins. These proteins are located within the ER membrane and transport processed cytoplasmic peptides into the ER lumen, where these peptides bind to MHC I proteins prior to transport to the cell surface. Mice deficient in TAP-1 have very low levels of MHC I on the cell surface and cannot process and present intracytoplasmic antigens via the class I pathway. They have limited expression of MHC I, which are either occupied by TAP-independent peptides generated within the endoplasmic reticulum or empty [12]. They have very few CD8⁺ T cells and deficient cytotoxic function (due to the absent of MHC I peptide) but normal CD4⁺ T cell compartment [6]. They appear to have normal NK cell numbers, but their cells have altered sensitivity to NK cell lysis [13].

In anti-GBM induced crescentic GN, the nephritogenic antigen is exogenous antigen (sheep immunoglobulin) planted on the glomerular basement by virtue of Fab specificity. Such extracellular protein antigens would be expected to be presented by the MHC II pathway rather than the MHC I pathway. In some situations, bone marrow-derived APC can present extracellular derived peptide via class I pathway to CD8⁺ T cells. This process, termed cross-presentation, is TAP-dependent [14–16]. The current studies in TAP-1-deficient mice suggest that cross-presentation does not play a major role in development of crescentic GN induced by a planted nephritogenic antigen.

The current studies are consistent with the traditional view of the pathways for presentation of extra-cellular antigens. They add to previous studies which demonstrated the requirement for MHC II in primary immune responses to anti-GBM globulin [1] and the role of MHC II expression by intrinsic renal cells for subsequent antigen recognition in the glomerulus and development of crescentic GN. They also indicate that MHC I-restricted CD8⁺ T cell-dependent cytotoxicity does not play a major role in the generation of renal injury in this model. Few NK cells were observed in glomeruli in this model, as in a rat model of anti-GBM GN [17], and their numbers were unaffected in TAP-1-deficient mice.

The MHC I independence of this model contrasts with the findings in ‘lupus prone’ MRL/lpr mice, in which crossbreeding to induce β₂-M deficiency substantially reduced their autoantibody formation and immune complex GN. This suggests an important role for MHC I in allowing autoantibody production [3]. Induction of autoantibodies to DNA and a lupus-like syndrome by immunization with human DNA is also impaired in MHC I-deficient mice [18]. Antigens responsible for this form of GN include intracellular antigens, and loss of tolerance to these antigens is important in the pathogenesis of this syndrome. MHC I may therefore play a complex role in this autoimmune process. β₂-M-deficient kidneys transplanted into MRL/lpr mice are not protected from development of GN [2]. Thus, unlike MHC II, MHC I would not appear to have a local role in targeting the immune effector response to the kidney.

Although TAP-1-deficient mice have substantially reduced numbers of CD8⁺ cells, the remaining CD8⁺ cells are largely functionally competent [12]. Thus, studies in TAP-1−/− mice do not entirely exclude a role for CD8⁺ cells in crescentic anti-GBM GN. In rats, CD8⁺ cells play an important role in glomerular macrophage recruitment in anti-GBM GN [4,19], suggesting that they may have important helper cell functions in this species. CD8 expression has also been detected on rat macrophages and CD8 ligation can regulate their function [20]. Similar findings have not been reported for mouse macrophages. Studies in CD8-deficient mice show that development of the immune system in the absence of CD8 accelerates crescentic GN induced by a planted nephritogenic antigen [5]. This could be attributed to an expansion of the CD4⁺ compartment during immune development in the absence of CD8. However, mice with combined CD4 and CD8 deficiency do not develop anti-GBM GN, demonstrating that CD4⁺ T helper cells are essential effectors of injury in crescentic GN [5]. This is supported by studies in which immunological depletion of CD4⁺ cells in mice [8] and rats [17] demonstrated that CD4⁺ T cells direct glomerular macrophage recruitment [17,19]. It is interesting that despite the dependence of the lupus syndrome in MRL/lpr mice on MHC I, this disease is also independent of CD8⁺ T cells [21].

Immunological depletion of CD8⁺ cells was used to investigate the contribution of CD8⁺ cells during the effector phase of anti-GBM GN. Depletion of circulating CD8⁺ cells to <5% of normal, abrogated their accumulation in glomeruli during the development of GN but did not attenuate crescent formation or impairment of creatinine clearance. However, CD8 depletion reduced proteinuria and there was a decrease in glomerular accumulation of macrophages. This suggests a ‘helper cell’ function for CD8⁺ T cells in mice which is consistent with (although less important in terms of overall injury) their role in crescentic GN in WKY rats [4]. CD8⁺ T cells have been previously demonstrated to produce cytokine/chemokine profiles consistent with T helper 1 subset functions [22–24], which may play a role in local macrophage recruitment and activation. Direct (macrophage-independent) effects of CD8⁺ T cells on GBM permeability (via perforin, granzyme, or other mediators) also cannot be excluded.

It is also possible that TAP-1-deficient mice have an expanded CD4⁺ T cell compartment and/or enhanced CD4⁺ responses, to compensate the loss of CD8⁺ T cells. However, TAP−/− mice did not develop more severe crescentic GN than wild-type controls, as was observed with CD8−/− [5]. The lack of any effect of TAP-1 deficiency on development of crescentic GN does not exclude a role for activation of CD8⁺ T cells by MHC I-independent antigens. MHC II-dependent selection and activation of cytotoxic CD8⁺ T cells has been previously demonstrated for viral antigens [25,26]. Some non-classical antigen-presenting molecules, for instance Qa-1b molecules encoded within the MHC I region by non-H-2 genes, have been demonstrated to efficiently present exogenous antigens to cytotoxic lymphocytes in a manner similar to class II MHC and do not require TAP-1 [27]. The involvement of TAP-1-independent antigens as the result of ‘epitope spreading’ during the development of this disease, or ‘cross-presentation’ of antigens via MHC II, could provide an alternative explanation for the reduction of proteinuria following CD8 depletion, in the absence of a demonstrable role for TAP-1.

In conclusion, this study demonstrates that MHC I-restricted antigen presentation and recognition is not required for development of crescentic GN induced by a planted exogenous glomerular antigen. It also demonstrates that CD8⁺ effector T cells do not play a critical role in the development of crescents, although they may contribute to inflammatory cell recruitment and proteinuria. This would be consistent with a MHC I-independent ‘helper’ function for CD8⁺ T cells in this situation. It supports the hypothesis that, although a variety of humoral and cellular mechanisms may induce glomerular injury, crescentic GN represents a DTH-like lesion which is MHC I-independent in mice.