Kidney dendritic cell activation is required for progression of renal disease in a mouse model of glomerular injury (original) (raw)
Generation and characterization of mice expressing model autoantigens in glomeruli. To straightforwardly study the role of glomerular antigen–specific T cells in nephritis, we generated mice expressing the model autoantigens OVA and HEL in podocytes. These mice allowed use of the well-characterized transgenic OT-I and OT-II mice, which produce OVA-specific CTLs (OT-I cells) and Th cells (OT-II cells), respectively, as donors for autoreactive T cells specific for a glomerular antigen. A construct containing the podocyte-specific human nephrin promoter (48) fused to cDNA encoding the transmembrane domain of the transferrin receptor, OVA, and HEL (49) (Figure 1A) was used to generate nephrin-OVA-HEL mice (NOH mice) on the C57BL/6 background. Immunohistochemistry revealed OVA expression in glomerular podocytes of NOH mice (Figure 1B) without expression in other organs, in particular pancreatic islets or the brain, where expression of a reporter driven by the murine nephrin promoter had been reported (50), or in non-Tg controls (Figure 1B and Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI38399DS1). Quantitative RT-PCR detected OVA message in the kidney after 40 cycles, while other tissues remained negative for 50 cycles, indicating that renal expression was at least 1,000-fold (10 cycles) higher than in other tissues (Supplemental Table 1).
Generation and characterization of mice expressing model autoantigens in glomeruli. (A) Plasmid used to generate NOH mice. (B) Kidney sections of a NOH (left) and non-Tg control (right) mouse stained for OVA expression. Red, OVA; blue, nuclei. Original magnification, ×1,000. (C) Division indices of CFSE-labeled OT-I cells in various LNs and the spleen of NOH and non-Tg mice on day 3 after adoptive transfer. (D) CFSE-labeled OT-I cells were injected into NOH × CD11c-DTR, NOH, or non-Tg mice, and DCs were depleted by injection of DT on the same day. Bars indicate division indices of OT-I cells in the renal LN on day 3. nOT-I, naive OT-I. (E) CD69 expression of CFSE-labeled OT-II cells in the renal LN of NOH and non-Tg mice on day 3 after transfer. In vitro–activated OT-II cells were used as positive staining control. Representative flow cytometry data for C and D appear in Supplemental Figure 2. Results are representative of 3 experiments in groups of 3 mice. Data are presented as mean ± SD. aOT-II, activated OT-II cells.
CFSE-labeled OT-I cells have previously been used as sensitive probes for in vivo OVA presentation (33). When these cells were injected i.v. into NOH mice, they divided only in the renal LN but not in any other LNs nor in the spleen nor in non-Tg control mice (Figure 1C; Supplemental Figure 2A), indicating kidney restriction of T cell activation and thus of antigen expression. Furthermore, we noted neither neurological symptoms nor diabetes mellitus in NOH mice injected with OT-I cells, which was the case in mice expressing OVA in pancreatic islets or in neurons, respectively (40, 47). These findings demonstrated that transgene expression in NOH mice was restricted to kidney podocytes.
To determine whether OT-I activation in the renal LN was a function of DCs, we crossed NOH mice to CD11c-DTR/eGFP (in which DTR indicates diphtheria toxin receptor) mice (here simply termed CD11c-DTR mice), which allow DC deletion by injection of diphtheria toxin (DT) (51). A single DT injection into NOH × CD11c-DTR mice reduced CD11c+ cells in the LNs by more than 90% (Supplemental Figure 2B) and abrogated proliferation of injected OT-I cells in the renal LN (Figure 1D; Supplemental Figure 2B), demonstrating that CD11c+ DCs activated OT-I cells by cross-presentation of glomerular autoantigen. This was not due to differences in the activation state of DCs from NOH mice, as this was identical to the activation state in non-Tg controls (Supplemental Figure 3).
CFSE-labeled OVA-specific Th cells (OT-II cells) neither proliferated nor expressed the activation marker CD69 after transfer into NOH mice (Figure 1E). This was consistent with previous studies showing failure of OT-II cells to proliferate in response to OVA expressed as transgenic self antigen and has been explained by their limited affinity in particular for autoantigen (35, 40, 46). In summary, glomerular autoantigen was constitutively presented by DCs in the renal LNs, but only OT-I cells proliferated in response.
Glomerular antigen–specific CTL release antigen for cross-presentation in the renal LN. To investigate whether activated CTLs caused glomerular damage in our experimental setting, we injected in vitro–activated OT-I cells into NOH mice to induce CTL-mediated glomerular damage. Subsequently, we injected CFSE-labeled naive OT-I cells as in vivo probes for cross-presentation. Indeed, proliferation of CFSE-labeled OT-I cells (Figure 2A; Supplemental Figure 4) and their expression of the activation marker CD69 (Figure 2B; Supplemental Figure 4) in the renal LNs were increased after injection of activated OT-I cells. This was not due to OT-I cell–induced increased expression of costimulatory molecules by renal LN DCs (Figure 2C). Thus, activated CTLs caused the release of podocyte antigen, which was subsequently cross-presented in the renal LNs.
Glomerular antigen–specific CTLs release antigen for cross-presentation in the renal LN. 5 × 106 activated OT-I cells were injected into NOH or WT mice. After 2 days, 2 × 106 CFSE-labeled OT-I cells were injected, and their proliferative response (A) and CD69 expression (B) were determined in the renal LNs. (C) Maturation state of renal LN DCs of NOH (black bars) and WT mice (white bars) injected with activated OT-I cells was determined by measuring CD86 and CD40 on CD11c+ cells. Representative flow cytometry data appear in Supplemental Figure 4. Results are representative of 3 experiments in groups of 3 mice. *P < 0.05; **P < 0.01. MFI, mean fluorescence intensity.
Glomerular antigen–specific CTLs and Th cells jointly induce periglomerular mononuclear infiltration. To determine what type of immunopathology results from glomerular antigen–specific T cells, we injected OT-I and OT-II cells into NOH mice. Neither cell type alone nor a combination of naive OT-I and naive OT-II cells caused any immunopathology. We speculated that this may be due to the lack of constitutive OT-II cell activation in NOH mice (Figure 1E). To circumvent this problem, we activated OT-II cells in vitro prior to injection, which has been shown to enable potent in vivo helper functions (35). Activated OT-II cells produced IL-2, TNF-α, and IFN-γ (data not shown), indicating Th1 differentiation, consistent with previous work using these cells (52). While transfer of activated OT-II cells alone failed to cause immunopathology, their coinjection with naive OT-I cells caused a significant focal periglomerular mononuclear infiltrate in all NOH recipient mice after 7 days (Figure 3, A and B), as determined by semiquantitative (Figure 3, J and K) and quantitative analysis (Figure 3L). Non-Tg recipients did not develop infiltration, indicating antigen specificity (data not shown). Interestingly, intraglomerular infiltration was not detected (Figure 3B). Electron microscopy showed that podocytes maintained regular foot processes, but in contrast to normal podocyte physiology, these cells were in direct contact with parietal cells (Figure 3C). Parietal cells featured large cytoplasm with increased numbers of organelles and were separated by a very thin membrane from the periglomerular infiltrate. Bowman capsule was surrounded by small lymphocytes and monocyte/macrophage-like cells (Figure 3C). Immunohistochemistry revealed expression of CD8α, CD4, CD11c, CD86, MHC II (I-Ab allele), and CD11b within periglomerular infiltrates (Figure 3, D–I), indicating the presence of T cells and antigen-presenting cells, such as DCs and macrophages.
Histological analysis of periglomerular infiltrates in NOH mice. (A–C) PAS staining (A and B) or electron microscopy analysis (C) of kidney sections of NOH mice injected with 5 × 106 OT-I cells and 5 × 106 activated OT-II cells 7 days before analysis. Note in C multiple contacts of podocytes with parietal epithelia separated by a very thin membrane from the periglomerular infiltrate. C, capillary; E, erythrocyte; M, mesangium; MnC, mononuclear cell; P, podocyte; PE, parietal epithelium; pgs, periglomerular space; asterisk, capsule membrane. Original magnification, ×3,000. Scale bar: 10 μm. (D–I) Representative immunohistochemistry for expression of CD8 (D), CD4 (E), CD11c (F), CD11b (G), MHC II (I-Ab) (H), and CD86 (I). (J) The frequency of glomeruli surrounded by mononuclear infiltrates was determined in HE-stained kidney sections of NOH or non-Tg mice injected with OT-I and/or activated OT-II cells as indicated. Shown are data from a group of mice that repetitively received T cell injections on days 7, 14, and 21. That group was analyzed on day 28 (histology in Figure 8). (K) Affected glomeruli were scored for the severity of periglomerular infiltrates. Results are representative of 4 experiments in groups of 3–5 mice. (L) Quantitative analysis of 2 of these experiments, an example of which was given in Supplemental Figure 7. In J–L, symbols indicate results from individual mice and the bars their mean. *P < 0.05; **P < 0.01; ***P < 0.001.
DC subset changes in infiltrated kidneys. The expression of CD11c, MHC II, and CD86 within the infiltrate (Figure 3, F–H) suggested the presence of DCs. Indeed, flow cytometry revealed CD11c+MHC II+ cells that expressed CD11b either at high or intermediate levels (Figure 4A) but lacked CD8 and B220 (Supplemental Figure 5), identifying them as conventional CD8– DCs. DC numbers were higher in infiltrated kidneys, and this was mostly due to an increase in CD11c+CD11bhi cells, whereas CD11c+CD11bint cells hardly changed in number (Figure 4B), suggesting that the latter might represent resident kidney DCs and that the former may have been derived from circulating precursors, e.g., from monocytes, which express similarly high levels of CD11b (53). In support of this interpretation, the CD11bhi cells expressed high levels of Gr1 (Figure 4C), a marker of proinflammatory monocytes, which can give rise to both DCs and macrophages (38, 53, 54). Further analysis of CD11bhi cells confirmed Gr1 expression on both CD11bhiCD11c+ and CD11bhiCD11c– cells (Figure 4C), consistent with recruitment of proinflammatory DCs and macrophages. Quantitative analysis verified that CD11bhiCD11c– macrophages had increased in number and that nearly all of them expressed Gr1 (Figure 4B). Also, the increase of CD11c+ DCs in infiltrated kidneys was mostly due to Gr1-expressing cells (Figure 4B).
DC subsets in infiltrated kidneys. (A) MHC II+ cells in kidney single-cell suspensions from NOH (lower dot plot) or non-Tg (upper dot plot) mice injected with 5 × 106 OT-I cells and 5 × 106 activated OT-II cells were analyzed for expression of CD11b versus CD11c. (B) Absolute numbers of CD11c+CD11bint DCs, CD11c+CD11bhi DCs, CD11c–CD11blo macrophages, CD11c+CD11bhi Gr1+ DCs, and CD11c–CD11blo Gr1+ macrophages in groups of 4 OT-I/II-injected NOH (black bars) or non-Tg (white bars) mice. (C) Dot plots shown as in A except that Gr1 was given instead of CD11c (left 2 dot plots). The right 2 dot plots show expression of CD11c versus Gr1 on the CD11bhi cells represented in A. Numbers in quadrants of dot plots indicate the proportion of cells. Results are representative of 3 experiments. **P < 0.01; ***P < 0.001. Data are presented as mean ± SD.
DCs are essential for periglomerular infiltration. To determine whether DCs were necessary for the periglomerular infiltrate, we injected OT cells into NOH × CD11c-DTR double-transgenic mice and depleted CD11c+ cells by DT injection after 5 days, when infiltrates were established. This reduced kidney DC numbers by 85%–90% (Figure 5A). Histological analysis demonstrated that the periglomerular infiltrate had completely disappeared after 40 hours (Figure 5, B, F, and G; Supplemental Figure 6) but remained detectable in double-transgenic mice that had not received DT (Figure 5, D, F, and G; Supplemental Figure 6), and in NOH single-transgenic mice, regardless of DT injection (Figure 5, C, E, F, and G; Supplemental Figure 6). We reasoned that such rapid resolution of the infiltrate was unlikely due to depletion of DCs in the renal LN because cessation of T cell priming in this node would not affect previously primed T cells in the kidney. Therefore we concluded that kidney DCs were necessary for maintenance of the periglomerular infiltrate.
DCs are essential for periglomerular infiltration. (A) 4 ng/g body weight DT was injected into WT, NOH, or NOH × CD11c-DTR mice that had received 5 × 106 OT-I cells and 5 × 106 activated OT-II cells 5 days before. After an additional 40 hours, kidney single-cell suspensions were examined for surviving DCs by flow cytometry for CD11c+ and eGFP expression within the transgene. Dot plots show representative results, and a quantitative analysis is given in the same panel to the right. Numbers in quadrants of dot plots indicate the proportion of cells. (B–G) 5 × 106 OT-I cells and 5 × 106 activated OT-II cells were injected into NOH × CD11c-DTR (B and C) or NOH mice (D and E). After 5 days, DT was injected (B and C). Kidney sections were scored after 40 hours for frequency (F) and severity (G) of infiltrates. Representative H&E stainings are shown in B–E. Scale bars: 400 μm. Semiquantitative analyses show 2 further controls: non-Tg mice injected with DT and NOH mice not injected with DT. Symbols indicate sections from individual mice and the bars their mean. Results are representative of 2 experiments. *P < 0.05; **P < 0.01; ***P < 0.001. Data are presented as mean ± SD.
Intrarenal DCs present CTL-released glomerular antigen only to Th cells. When we examined activation of DCs in infiltrated kidneys of NOH mice, expression of CD86 and CD40 was higher compared to that of DCs in non-Tg controls (Figure 6, A and B), and a DC subset produced IL-12 (Figure 6C). Notably, these activation signs were also detectable albeit less pronounced when activated OT-II cells were injected alone but not when only activated OT-I cells were transferred (Figure 6, A–C), suggesting that interaction with Th cells alone was sufficient for DC maturation and that CTLs could further increase it.
Kidney DCs present CTL-released glomerular antigen to OT-II cells. (A–C) DC maturation markers CD86 (A) and CD40 (B) and intracellular IL-12 production (C) were determined on CD11c+MHC II+ kidney DCs on day 3 after injection of 5 × 106 activated OT-I cells and/or 5 × 106 activated OT-II cells as indicated. OT-I cells were used here also in an activated state to synchronize their effector phase with that of activated OT-II cells. Numbers in quadrants of dot plots indicate the proportion of cells. (D) NOH and non-Tg mice were injected with activated OT-I cells (+aOT-I) or not (–aOT-I). After 3 days, CD11c+ DCs were isolated from the kidney and spleen. 5 × 106 DCs were cultured with 5 × 106 OT-II cells. After 2 days, IFN-γ concentrations in the supernatant were determined by ELISA. Results are representative of 2 experiments. *P < 0.05; **P < 0.01; ***P < 0.001. Data are presented as mean ± SD.
To verify this interpretation, we injected NOH mice with activated OT-I cells in order to release glomerular antigen, isolated kidney DCs on the following day, and cocultured them with OT-II cells. Indeed, such DCs could stimulate IFN-γ production from OT-II cells (Figure 6D). IL-2 production was only marginally increased, and this was not significant (data not shown). Cytokine production was induced neither by DCs from OT-I cell–injected non-Tg mice nor by DCs from NOH mice that had not been injected with OT-I cells (Figure 6D), confirming that indeed CTL-released antigen was presented. DCs from the spleens of OT-I cell–injected NOH mice did not stimulate OT-II cells (Figure 6D), demonstrating that released glomerular antigen was not systemically available for presentation. These findings confirmed that intrarenal presentation of glomerular antigen by DCs elicited cytokine production by specific Th cells.
Interaction between kidney DCs and Th cells causes intrarenal CTL accumulation. We next studied the consequences of cytokine production for OT-I cells. Their numbers were increased in the infiltrated kidneys of NOH mice to more than 10 times the amount in the kidneys of non-Tg mice (Figure 7A), and they produced more IFN-γ, which indicated their activation (Figure 7B). Such an increase was not observed in the spleen (Figure 7C), demonstrating that OT-II cells caused local expansion of OT-I cells in the kidney. OT-I cell numbers in both kidney and spleen of NOH mice injected with OT-I cells alone were lower than in the respective organs of non-Tg mice (Figure 7, A and C), presumably indicating the beginning of their systemic deletion by cross-tolerance (40). Also, this reduction was prevented by OT-II cells (Figure 7C), demonstrating that these cells also exerted a distinct helper function, impairment of cross-tolerance (40). The numbers of intrarenal OT-II cell numbers were not affected by coinjection of OT-I cells (Figure 7D), confirming that the former acted upstream of the latter. These findings indicated that interaction between kidney DCs and OT-II cells caused intrarenal accumulation of OT-I cells.
Interaction between kidney DCs and Th cells causes intrarenal CTL accumulation. NOH (black bars) or non-Tg (white bars) mice were injected with 5 × 106 OT-I cells alone or together with 5 × 106 activated OT-II cells. (A–C) After 7 days, single-cell suspensions of the kidney (A and B) or the spleen (C) were analyzed by flow cytometry to determine numbers of CD8+Vα2+Vβ5+ OT-I (A and C) or proportions of (B) IFN-γ–producing OT-I cells. (D) Numbers of CD4+ Vα2+Vβ5+ OT-II cells were determined after injection of 5 × 106 OT-II cells alone or together with 5 × 106 activated OT-II cells. (E) Experiments depicted in A–C show determination of Ki-67+ OT-I cells. In vitro–activated OT-II cells served as positive control, naive OT-II cells as negative control for proliferating cells. (F and G) Vα2+Vβ5+CD4+ OT-II (F) or Vα2+Vβ5+CD8+ OT-I cells (G) were sorted from kidney cell suspensions of experiments depicted in A–C. mRNA encoding CCL3, CCL4, CCL5, and CCR5 was determined, and the ratio between cells from NOH and non-Tg controls was displayed. (H) Mononuclear cell infiltrates and noninfiltrated tubulointerstitial control areas were excised from kidney cryosections as shown in Supplemental Figure 8. mRNA encoding CCL3, CCL4, CCL5, and CCR5 was determined, and the ratio between infiltrates from injected NOH and non-Tg controls was displayed. Results are representative of 2 experiments. *P < 0.05; **P < 0.01; ***P < 0.001. Data are presented as mean ± SD.
This increase of OT-I cells in the kidney did not result from their local proliferation because these cells did not express the Ki-67 proliferation marker when analyzed by intracellular flow cytometry (Figure 7E). To study the possibility of recruitment of OT-I cells, we determined expression of CCR5 ligands, which are produced by Th cells to attract CTLs (55). Indeed, we noted that OT-II cells isolated from infiltrated kidneys of NOH mice expressed mRNA levels of the CCR5 ligands CCL3, CCL4, and CCL5 that were 2.5, 6, or 2 times higher, respectively, than those of non-Tg mice. Of these, the increase in CCL4 was statistically significant. (Figure 7F). Intrarenal OT-I cells did not show such an increase but instead showed 6.5 times greater levels of CCR5 itself (Figure 7G). Also, OT-II cells showed a moderate 2-fold CCR5 increase in infiltrated kidneys, but this was not statistically significant (Figure 7F). When periglomerular infiltrates were excised by laser dissection microscopy and their mRNA was compared to that from corresponding tubulointerstitial areas of non-Tg mice, an even more striking increase of CCR5 ligands was observed, which reached 200-fold for CCL4 (Figure 7H). Here CCL3 and CCL5 were also significantly increased, implicating cells other than OT-I or OT-II cells as the source. These findings suggested that OT-II cells stimulated by kidney DCs in periglomerular infiltrates recruited OT-I cells to the kidney via CCR5 ligands.
Chronic T cell–mediated damage causes structural and functional kidney damage. We finally examined the consequences of periglomerular infiltration at later time points. To this end, we repetitively injected NOH mice at weekly intervals with OT-I and activated OT-II cells because we speculated that the short life span of activated T cells might limit renal damage after a single injection. No pathology was seen when OT-I or OT-II cells were injected separately (data not shown). Coinjection resulted in pronounced tubulointerstitial infiltration by mononuclear cells, intratubular protein casts, focal tubular atrophy, and focal segmental glomerular sclerosis with retraction of the glomerular tuft to the vascular pole on day 28 (Figure 8, A and B). Type IV collagen staining on day 28 but not on day 7 revealed increased interstitial matrix deposition in the outer areas of the infiltrates, denoting induction of fibrosis at later time points (Figure 8, C and D). Together, these findings indicated chronic tubulointerstitial damage.
Repetitive OT cell injection causes functional and structural kidney damage in NOH mice. NOH mice were injected with 5 × 106 OT-I cells and with 5 × 106 activated OT-II cells on days 0, 7, 14, and 21. On day 7 (C and E) or day 28 (A, B, D, and F–I), kidneys were taken for PAS staining (A and B), type IV collagen staining for fibrotic areas (C and D), OVA staining (E and F; black arrows indicate OVA+ cells in Bowman capsule wall), and electron microscopy (G). White arrow, contact between podocytes and parietal cells; black arrow, podocyte foot processes. Original magnification, ×3,000. Scale bar: 10 μm. (H) Daily excretion of albumin (g/l) per creatinine (g/l) was determined in overnight urine of groups of NOH or non-Tg mice injected with nOT-I and/or activated OT-II cells. (I) 20 μl urine from mice in groups denoted by diamonds and open squares on day 7 were separated by gel electrophoresis and stained with Coomassie blue. Results are representative of 2 experiments.*P < 0.05; **P < 0.01; ***P < 0.001.
Notably, we regularly observed OVA-expressing cells in the Bowman capsule wall of infiltrated kidneys on day 28 (Figure 8F). The glomerular damage at this time point did not permit unequivocal classification of these cells. On day 7, such cells were rare and their OVA-staining intensity seemed lower (Figure 8E).
By electron microscopy, we determined that the parietal cells of the Bowman capsule appeared activated, with an increased number of organelles and edematous cytoplasm (Figure 8G). Glomerular capillaries showed regular endothelial cells, basement membranes, and podocytes. The Bowman capsule membrane was thickened. The composition of the periglomerular mononuclear infiltrate was similar to that seen on day 7 (Figure 3C), but the infiltrate was more pronounced and embedded in increased extracellular matrix (Figure 8G).
Consistent with glomerular sclerosis, daily excretion of albumin per creatinine, an indicator of glomerular damage, had increased substantially from day 7 to day 28 (Figure 8H). No albuminuria was detected when OT-I or OT-II cells were injected alone or when non-Tg recipients were used (Figure 8H and data not shown). Renal insufficiency as determined by creatinine clearance did not develop until day 28 (data not shown). Gel electrophoresis revealed that excreted protein was primarily of albumin and not of Ig size (Figure 8I), indicative of selective glomerular proteinuria. Very low albuminuria was found when activated OT-I cells were injected alone (Figure 8I), supporting the conclusion that CD8+ T cells could cause kidney damage but that the presence of Th cells was required for progression to manifest immunopathology and functional damage.