Activated CD4+ T Cells Target Mesangial Antigens and Initiate Glomerulonephritis

CD4+ T cells are important mediators of systemic lupus erythematosus [1]. In addition to helping autoreactive B cells to produce autoantibodies, CD4+ T cells are directly involved in the renal pathology of lupus. Our studies in New Zealand Mixed 2328 mice, a spontaneous mouse model of systemic lupus erythematosus, showed that mesangial immune complex deposition was accompanied by CD4+ T cell activation in the renal lymph nodes and glomerular T cell infiltration [2]. T cell receptor analyses indicated a local antigen-driven T cell response. A direct role for T cells in pathogenesis of lupus nephritis is also supported by the finding that inhibiting T cell costimulation increases the life span of nephritic mice. [3]. Yet, whether CD4+ T cells mediate the initiation of renal injury is unclear.

In this study, we tested the hypothesis that T cells responding to antigens in the mesangium can induce glomerulonephritis. Since the identity of antigens recognized by pathogenic T cells in lupus is not known, we used ovalbumin (OVA) as a surrogate mesangial antigen. We have previously described anti-α8-integrin immunoliposomes (α8ILs), a unique methodology for targeted delivery to the glomerular mesangium [4, 5]. Liposomes sized to approximately 100 nm in diameter enter the mesangial space through the fenestrated glomerular endothelium following intravenous injection. The anti-α8-integrin antibody conjugated to the liposomes reacts with α8-integrin on the mesangial cell surface. This allows accumulation of liposomes in the glomerular mesangium and leads to delivery of the liposomal contents at the site. Thus, in the present study, OVA was deposited in the mesangium using α8ILs and the ability of anti-OVA CD4+ T cells responses to initiate glomerulonephritis was investigated.

Anti-α8ILs were prepared as previously described [4]. To incorporate OVA, the lipid mixture was hydrated either with chicken OVA (Sigma Aldrich Corporation, St. Louis, Mo., USA) or fluorescein isothiocyanate-conjugated OVA (FITC-OVA) (Invitrogen, Carlsbad, Calif., USA) in water, lyophilized and rehydrated in 0.025 m Tris, pH 8.0. After repeated washing, the liposomal OVA content was estimated by modified BCA assay (Peirce Biotechnology, Rockford, Ill., USA). Liposomes were then conjugated to either a rabbit anti-α8 integrin antibody (OVA-α8ILs) or to rabbit IgG (control OVA-ILs).

Mice were maintained in a specific pathogen-free mouse facility at the University of Virginia. All experimental procedures were approved by the Institutional Animal Care and Use Committee. Breeders for B6.Cg-Tg(TcrαTcrβ) 425Cbn/J (OT2) and B6.129S4-Ccr2^tm1Ifc/J (B6.ccr2^–/–) mice were obtained from Jackson Laboratories (Bar Harbor, Me., USA). C57BL/6 (B6) mice were purchased from National Cancer Institute (Frederick, Md., USA). OT2 transgenic mice express CD4+ T cell receptor transgene (Vα2, Vβ5) reactive with OVA_323–339 peptide and were used as donors for antigen-specific T cells [6]. B6 and B6.ccr2^–/– female mice were used as recipients.

OT2 T cells were activated with OVA_323–339 peptide (Ana Spec, Fremont, Calif., USA) for 72 h using standard protocols [7]; live cells were harvested over a Ficoll gradient and injected intravenously into recipient mice. For some experiments, CD4+ T cells were purified by negative selection (Miltenyi Biotec, Auburn, Calif., USA), activated with OVA_323–339 peptide and irradiated syngenic spleen cells. This method of stimulation yielded Th1-deviated OT2 cells, and this was confirmed by detection of IFNγ in the culture supernatants at 24 and 48 h after stimulation (data not shown). To investigate activation of OT2 cells in vivo, naïve OT2 T cells were isolated and injected intravenously into recipient mice. In all experiments, mice received 12–14 million OT2 T cells.

Cell suspensions from spleens, lymph nodes and kidneys were studied by flow cytometry using standard methods [2]. Antibodies conjugated to fluorochromes were obtained from eBiosciences (San Diego, Calif., USA) or BD Biosciences (San Jose, Calif., USA) except as indicated. Cells were incubated with anti-CD16/32 (Clone 93) and OT2 T cell frequencies were analyzed by 4-color staining for CD45 (30-F11), CD4 (L3T4), T cell receptor Vα2 (B20.1) and Vβ5 (MR9-4). Kidney-infiltrating macrophages and dendritic cells were studied by 6-color flow cytometry using antibodies to CD45 (30-F11), F4/80 (BM8), CD11b (M1/70), CD11c (HL3) and Ly6C (HK1.4, Abcam). Bromodeoxyuridine (BrdU) incorporation was measured by intracellular staining using anti-BrdU antibody (BU20A) followed by FITC-conjugated F(ab)₂ anti-mouse IgG. Stained cells were acquired on 8-color BD FACS Calibur using Cytek dxP8 software (Cytek Technologies, Fremont, Calif., USA). Data were analyzed using FlowJo Software 9.1 (Tree Star Inc., Ashland, Oreg., USA).

For detection of FITC-OVA, glomeruli were isolated and stained as previously described [4]. Nonspecific staining was blocked with normal rabbit serum and incubated with goat anti-FITC antibody (Vector Laboratories, Burlingame, Calif., USA). Bound antibody was detected with fluoresceinated rabbit anti-goat IgG (Vector Laboratories).

In the kidney, FITC-OVA was detected in frozen sections, fixed in cold acetone and observed under a fluorescence microscope. Infiltrating cells were characterized by immunostaining with antibodies to CD4 (GK1.5), CD11c (HL3), CD68 (AbD serotec, Raleigh, N.C., USA) and MHC II (M5/114.15.2) as previously described [2, 8]. MCP1 was detected using rabbit anti-MCP1 antibody (Peprotech, Rocky Hill, N.J., USA) followed by PE-conjugated anti-rabbit IgG (Jackson ImmunoResearch Laboratories Inc., West Grove, Pa., USA). Glomerular fibronectin was detected with rabbit anti-fibronectin antibody (Millipore Corporation, Sunnyvale, Calif., USA) [8]. All fluorescence images were captured on a Zeiss Confocal Microscope or Zeiss Apotome fluorescence microscope, and the images were analyzed using LSM5 and Axiovision 4.7 software (Carl Zeiss MicroImaging LLC, Thornwood, N.Y., USA), respectively.

Kidneys were collected in formalin and processed for paraffin embedding. Sections (3 µm thickness) were stained with hematoxylin and eosin and evaluated for pathologic changes as previously described [2].

Statistical analyses were done by Student’s t test or ANOVA with Bonferroni posttest using GraphPad Prism 5.0 (GraphPad Software Inc., La Jolla, Calif., USA).

To study the kinetics of glomerular delivery, FITC-OVA-loaded α8ILs were injected intravenously into B6 mice. FITC-OVA was rapidly deposited in the glomeruli and could be detected by indirect immunofluorescence 6 h after injection (fig. 1a). The amount of FITC-OVA in the glomeruli increased by day 5 (fig. 1b). In the kidney, FITC-OVA was restricted to glomeruli and showed a typical mesangial pattern of distribution (fig. 1c). FITC-OVA loaded liposomes conjugated to rabbit IgG (control OVA-ILs) did not traffic to the renal glomeruli.

Mice were given two intravenous injections of OVA-α8ILs or control liposomes, 12 h apart followed by a single injection of activated OT2 splenocytes as shown in figure 2. Kidneys studied on day 8 after transfer showed inflammation in the renal cortex predominantly in glomeruli and periglomerular regions (fig. 3a). These results were further confirmed by adoptive transfer of purified and activated OT2 T cells (94% purity, 99% Vα2 Vβ5-positive). Kidneys were harvested on days 3, 8 and 14 after cell transfer, and infiltrating cells analyzed by flow cytometry showed an increased frequency of OT2 cells on days 8 and 14 (fig. 3b). This was accompanied by an increase in the CD4+ T cells infiltrating the kidney. Interestingly, OT2 cells were only 3–8% of the infiltrating CD4+ cells. Thus, trafficking of OT2 cells into the kidney led to the recruitment of endogenous, non-OT2, CD4+ T cells.

Immunostaining showed aggregates of CD4+ T cells and MHCII antigen presenting cells in the renal cortex on day 3; with significant intraglomerular and periglomerular infiltrates on days 8 and 14. The CD11c+ dendritic cells (fig. 4a) and CD68 and F4/80 positive macrophages (fig. 4b) in the infiltrates were the predominant MHCII+ cells, suggesting that they are the antigen-presenting cells (insets to fig. 4a, b, respectively). Inflammatory foci were not seen in control mice.

To evaluate the pathogenic effect of glomerular inflammation, kidneys were stained for fibronectin, a component of the extracellular matrix and an early marker for mesangial cell activation [9]. On day 8, glomerular fibronectin was increased in mice given OVA-α8ILs and OT2 cells (fig. 4c). At these time points, the kidney function was normal. Thus, activated CD4+ T cells reacting to mesangial antigens can initiate mesangial injury and glomerulonephritis.

Lymph nodes and spleens were harvested from mice to study trafficking of OT2 T cells (fig. 5). On day 8 after transfer, mice with mesangial OVA consistently showed an increased frequency of OT2 cells in the renal lymph nodes (p < 0.01) compared to controls (fig. 5b). Furthermore, the enrichment of OT2 cells was specific to renal lymph nodes compared to the nonrenal lymph nodes (p < 0.001). This enrichment was not seen in control mice given activated OT2 cells alone. In the spleen, frequencies of OT2 cells in mice injected with OVA-α8ILs and activated OT2 cells were 11.6 ± 1.7 (absolute numbers 1.06 ± 0.15 × 10⁶; mean ± SEM) and were not significantly higher than mice injected with activated OT2 cells alone with an average frequency of 8.3 ± 0.5 (absolute numbers 0.85 ± 0.15 × 10⁶). These data suggest that the presentation of OVA was occurring predominantly in the kidney and local lymph nodes and provides additional evidence that delivery by OVA-α8ILs was specific to the kidney.

Monocyte chemoattractant protein 1 (MCP1) in the glomerulus is an early indicator of inflammation [10]. MCP1 mediates macrophage recruitment through its receptor, chemokine CC motif receptor 2 (CCR2) [11]. To study the role of macrophages and the MCP1-CCR2 axis, OVA-α8ILs were injected into ccr2^+/+-sufficient (wild-type) and ccr2^–/– mice, and followed by adoptive transfer of activated OT2 cells. MCP1 expression measured by qPCR was increased in the kidneys from both ccr2^+/+ (2.8 ± 1.0; mean ± SD, n = 5) and ccr2^–/– mice (6.0- ± 2.5; mean ± SD, n = 5) fold over controls injected with activated OT2 cells alone. These results were confirmed by immunostaining for MCP1 protein in the glomeruli (fig. 6). Control mice did not show glomerular MCP1 expression. Flow cytometry analyses showed that CD4+ T cell infiltration was comparable between wild-type and ccr2^–/– mice (fig. 7a, left panel) and there was no significant difference in the recruitment of antigen-specific OT2 cells (data not shown) into the kidney. Further, the frequencies of CD11b macrophages and CD11c dendritic cells were comparable (data not shown). However, there was a small though consistent, statistically significant reduction in the frequency of Ly6C inflammatory monocytes in ccr2^–/– mice (n = 5/group; p = 0.0096; fig. 7a, right panel). This was reproduced in an additional experiment; p = 0.019; n = 4/group, data not shown). Representative gates used for analyses of Ly6C cells are shown (fig. 7b). Surprisingly, despite comparable OT2 and endogenous CD4+ T cell recruitment, ccr2^–/– mouse kidney showed a complete lack of inflammatory foci (fig. 7c). Thus, trafficking of antigen-specific and endogenous T cells into the kidney was not affected by CCR2 deficiency. However, recruitment of Ly6C cells was important for the development of glomerulonephritis.

The results presented above show that mesangial antigens are presented to activated CD4+ T cells in the kidney leading to recruitment of bystander T cells and inflammatory macrophages inducing glomerular inflammation. This is associated with activation of antigen-specific T cells in the renal lymph nodes. To investigate whether mesangial antigens drain to the renal lymph nodes, naïve OT2 cells were transferred into mice injected with OVA-α8ILs. Control mice received only naïve OT2 cells. The mice were given a single injection of BrdU intraperitoneally (2 mg/mouse), 12 h prior to sacrifice on day 3 after transfer (fig. 8a). Accumulation of naïve OT2 cells in the renal lymph nodes was accompanied by increased proliferation in mice injected with OVA-α8ILs compared to controls (fig. 8b). At this point, few OT2 cells were detected in the kidneys of either group and none of them showed BrdU uptake (data not shown). These results suggest the glomerular antigens drain into the renal lymph nodes or are carried there by antigen-presenting cells and are presented to antigen-specific CD4 T cells.

Patient and experimental mouse studies suggest that pathology in lupus glomerulonephritis initiates in the mesangium [12, 13] and CD4+ T cells participate in the pathogenesis of lupus nephritis [2, 3]. However, the lack of information on target antigens in lupus nephritis is a major barrier in investigating mechanisms of T cells in initiating disease. In this study, we demonstrate that activated antigen-specific CD4+ T cells, directed against a surrogate mesangial antigen, are sufficient to initiate the typical inflammatory lesions consisting of periglomerular macrophage, dendritic cell and T cell infiltrates. This induces mesangial cell activation and extracellular matrix production. These changes represent early events and at this time point, are not associated with loss of kidney function. This pattern of glomerulonephritis mimics the patterns reported in some lupus patients with CD4+ T cell infiltrates forming focal or circumferential caps around glomeruli in the renal cortex [14]. The novelty of this study lies in being able to target antigens that are in the mesangium with antigen-specific T cells in immune sufficient mice, mimicking the location and pathology of spontaneous glomerulonephritis. This will allow us to dissect the individual contributions of inflammatory mediators in mesangioproliferative glomerulonephritis.

Use of antigen-specific T cells in the induction of glomerulonephritis have been previously described. Mesangial deposition of OVA polymers followed by injection of OVA-reactive T cell lines in SCID mice showed that Th1 T cells could traffic into the kidney leading to glomerulonephritis [15]. Transgenic mice expressing OVA and hen egg lysozyme under the nephrin promoter in podocytes (NOH) mice have been recently used to investigate glomerulonephritis [16]. Repeated transfers of CD8+ OT1 T cells coinjected with activated CD4+ OT2 T cells in NOH mice resulted in glomerulonephritis. However, injection of either cell type alone failed to cause disease. These results are different than our study, where activated CD4+ OT2 cells were sufficient for glomerulonephritis. Other studies have proposed glomerular basement membrane proteins like collagen [17] as potential targets for autoreactive T cells using rats primed with pertussis toxin to facilitate trafficking of T cells into nonlymphoid tissues, or antigens trapped on the glomerular basement membrane using Rag-deficient recipient mice facilitating homeostatic expansion of autoreactive T cells [18]. In our experiments, glomerular inflammation was comparable in mice with or without priming with pertussis toxin (data not shown) and was seen in normal B6 recipient mice. Although the role of podocyte antigens in spontaneous glomerulonephritis is unclear [19], together these studies show that the glomerular location of the target antigen clearly influences CD4 T cell-mediated pathology.

This study focuses on the early events in initiation of glomerulonephritis. Whether this sequence of events occurs with each renal flare would be of clinical relevance and would provide the rationale for preventing local T cell responses and macrophage trafficking into the kidney as important strategies for therapeutic intervention.

The authors would like to thank Mr. Saleh Mohammad for outstanding technical assistance. This work was supported by grants from National Institutes of Health R01DK069769 (H.B.), ARRA supplement to R01DK069769 (H.B.), R01AI079621 (U.D.), DK76095 (W.K.B.), and from Alliance Lupus for Research No. 113300 (H.B.).

The authors have no financial conflicts of interest to disclose.