Dependence of Glomerulonephritis Induction on Novel Intraglomerular Alternatively Activated Bone Marrow-Derived Macrophages and Mac-1 and PD-L1 in Lupus-Prone NZM2328 Mice

Abstract

Glomerular damage mediated by glomerulus-infiltrating myeloid-derived cells is a key pathogenic event in lupus nephritis (LN)^c but the process is poorly understood. Confocal microscopy of kidney sections and flow cytometry analysis of glomerular cells from magnetic-bead purified glomeruli have identified glomerulus-infiltrating leukocyte populations in NZM2328 (NZM) lupus-prone mice with spontaneous chronic glomerulonephritis (cGN) and anti-glomerular basement membrane (GBM)-induced nephritis. The occurrence of a major glomerulus-infiltrating CD11b⁺F4/80⁻I-A⁻ macrophage population exhibiting the markers PD-L1, Mac-2, and macrophage mannose receptor (MMR, CD206) and producing Klf4, Il10, Retnla, Tnf, and Il6 mRNA which are known to be expressed by alternatively activated (M2b) macrophages correlated with proteinuria status. In NZM mice with spontaneous LN, glomerular macrophage infiltration is predominant. CD11b⁺F4/80⁻I-A⁻ intraglomerular macrophages and PMN are important in inducing GN as anti- CD11b and ICAM-1 mAb inhibited both proteinuria and macrophage and PMN infiltration. The predominant and high expression of PD-L1 by CD11b⁺F4/80⁻I-A⁻ glomerular macrophages in kidneys of mice with GN and the inhibition of proteinuria by anti-PD-L1 mAb supported the pathogenic role of these macrophages but not the PD-L1⁻ PMN in GN development and in inducing podocyte damage. In NZM mice with spontaneous cGN and severe proteinuria, few glomerulus-infiltrating PMN were found, leaving macrophages and to a lesser extent DC as the major infiltrating leukocytes. These data taken together support the important pathogenic effect of CD11b⁺F4/80⁻I-A⁻ M2b-like glomerulus-infiltrating macrophages in LN and reinforce macrophages as a promising target for GN treatment.

Introduction

Systemic lupus erythematosus (SLE) is an autoimmune disease initiated by Ab against a wide variety of self-antigens (Reviewed in (1–3)). The etiology of SLE is composed of both genetic and environmental components. Two mouse strains utilizing the autoimmune (NZB × NZW)F1-derived sublines NZM2328 (NZM) and NZM2410 have been most useful in identifying lupus-associated genes in SLE (1, 4). The roles of many susceptibility genes in SLE-related immune response pathways have been studied (2, 4–6) while novel roles for other genes continue to be discovered (7). Despite this extensive knowledge base for the immune effector molecules and cells, the pathogenesis of SLE remains unclear due to the uncertainties of the functional defects in susceptible gene alleles, the relative contribution of each of the susceptible genes, the interactions among inflammatory and regulatory immune pathways, and the functional roles of various cell types including non-immune stromal cells in SLE.

Lupus nephritis occurs in 30–60% of SLE patients (8). The most important pathological condition occurs in the glomerulus which exhibits at various disease stages glomerular immune complex deposition, hypercellularity, increased matrix protein accumulation and fibrosis, endothelial cell damage and loss of cell surface glycocalyx, and podocyte cellular damage and foot process effacement (9, 10). It is the loss of endothelial glycocalyx and podocyte foot process effacement that cause the loss of glomerular filtration function and proteinuria (11, 12). The initial inciting factors are immune complexes comprising auto-Ab and complement components deposited in the glomerular basement membrane (GBM) that cause the activation of glomerular parenchymal cells and the infiltration of leukocytes. The resulting cascades of growth factor release, upregulation and interactions of receptor-ligand pairs and adhesion molecules, and activation of glomerular cells lead to the damaging cellular events in GN.

Glomerular leukocyte infiltration is a hallmark of lupus nephritis (10). Proliferating macrophages have been identified in the glomeruli of patients with various GN types (13–15). Their reduction following treatment correlates with the degree of disease response (16, 17). Increases in intraglomerular T cells and PMN in GN patients have also been documented (18, 19). In animal models, PMN, macrophage, and dendritic cell (DC) influxes into the glomerulus after TNF-α treatment have been enumerated by flow cytometry analysis of isolated glomeruli (20). Anti-GBM-mediated PMN and macrophage influx into the glomeruli has recently been shown to be CD11b-dependent by multiphoton imaging (21). Recent studies on yolk sac-derived kidney F4/80⁺ resident macrophage functions showed that these macrophages do not infiltrate the glomerulus in immune-complex-mediated inflammation, thus implicating that glomerulus-infiltrating macrophages are bone-marrow derived (22). Despite these observations of leukocyte influx, there have been few studies on the kinetics of leukocyte influx and the function of these cells in GN pathogenesis due to the technical difficulties associated with isolating and analyzing intraglomerular cells. In this study, we have used the autoimmune mouse NZM model to study intraglomerular leukocyte infiltration and function in both spontaneous cGN and anti-GBM-mediated accelerated nephritis models. Flow cytometry analysis of single cell suspensions of highly purified glomeruli identified the C11b⁺ F4/80⁻I-A⁻ macrophages as the largest infiltrating population in mice with proteinuria in both spontaneous and anti-GBM-induced nephritis models. This population expresses the alternatively activated macrophage markers Mac-2, PD-L1, and MMR and Klf4, Il10, Retnla, Tnf, and Il6 mRNA known to be present in M2b macrophages (23). In cGN, significant percentages of I-A⁺CD11b⁺ DC-like cells but few PMN were found to be infiltrating the glomerulus. In contrast, in glomeruli of mice with anti-GBM-induced nephritis, large percentages of PMN but few I-A⁺ DC were found. Both proteinuria and macrophage and PMN infiltration were blocked by anti-CD11b mAb and anti-ICAM-1 mAb. Furthermore, blocking of PD-L1 which is expressed predoimantly by glomerular macrophages in GN inhibited proteinuria, thus supporting the pathogenic role of infiltrating CD11b⁺F4/80⁻I-A⁻ macrophages in LN and other GN-related diseases.

Materials and methods

Materials

DNase I, and Liberase TM were from Roche Diagnostic Corp. (Indianapolis, IN). Tosyl-activated Dynabead M450 (Invitrogen, Grand Island, NY) were treated with triethylamine before use for glomeruli isolation. Rat or hamster mAb against the following Ag with or without fluorochrome conjugation were from Biolegend (San Diego, CA): B220 (RA3–6B2), CD3 (145-2C11), CD4 (RM4–5), CD8 (53.6.7), CD11b (M1/70), CD11c (N418), CD25 (PC61), CD26 (H194-112), CD31 (MEC13.3), CD44 (IM7), CD45 (30-F11), CD62L (MEL-14), CD69 (H1.2F3), CD73 (Ty/11.8), CD102 (3C4(MIC2/4)), CD105 (MJ7/18), CD169 (3D6.112), F4/80 (CI:A3-1), I-A/I-E (M5/114.15.2), Ly6C (HK1.4), Ly6G (1A8), Ly6C/G (GR1), MMR (CD206; C068C2), NK1.1 (PK136), TNP (Rat isotype controls: RTK2071 IgG1; RTK2758 IgG2a; RTK4530 IgG2b; Armenian hamster IgG, HTK888), Thy1.2 (53.2.1), VCAM-1 (429 (MVCAM.A)), and VE-cadherin (BV13). Blocking mAb against PD-L1 (10F.9G2), CD11b (M1/70), ICAM-1 (YN1), and Ly6C/G (Gr-1) and rat isotype control mAb (LTF-2) were from Bio X Cell (West Lebanon, NH). The following goat affinity-purified polyclonal Ab against recombinant mouse proteins were from R&D (Minneapolis, MN, USA): CD103, Itgα8, MMR, nephrin, Mgl1/2, NKp46, PD-L1, and SLAMf9 and they were conjugated with Pacific Blue (PacBlue) or Alexa Fluors (A) with mAb labeling kits (Invitrogen). Affinity purified rabbit Ab against CD8 and CD14 were from Sinobiological (Beijing, China). Purified goat and sheep IgG were from Rockland Immunochemicals (Limerick, PA) and Pel-Freez (Rogers, AR) respectively. CFA (Difco) was from Fisher Scientific (Suwanee, GA).

Mouse

C57BL/6NCr mice were from NCI (Frederick, MD, USA). The lupus-prone NZM mouse strain was originally from Jackson Laboratory (Bar Harbor, ME, USA). The congenic non-lupus-prone NZM2328.Lc1R27 (R27) strain has been described (24). Only female NZM and R27 mice were used unless otherwise noted. Proteinuria was monitored by Multistix 10 SG (Bayer Diagnostics Division, Elkhart, IN) and mice with ≥ 300 mg/dL urinary protein (3+) were sacrificed for immunological and histological evaluations. Animal use and manipulation followed protocols approved by the University of Virginia Institutional Animal Use and Care Committee.

Anti-GBM-induced nephritis

Female 10 – 12 week old NZM mice were primed with 0.5 mg sheep IgG in CFA 3 days prior to anti-GBM injection. Sheep anti-GBM antiserum was prepared as described (24), absorbed with mouse erythrocytes (1:1) and titered for maximal dose without immediate toxicity before use in i.v. injections. Severe proteinuria were observed by day 4 and mice die by day 14. For blocking experiments, mice were injected with mAb on alternate days beginning on the day of antiserum injection (day 0). Mice were harvested either on day 3, day 5, or day 7 for assessment of histology, cellular infiltration by confocal microscopy, and glomerular leukocyte infiltration by flow cytometry.

Confocal Microscopy

Kidney sections were fixed in 2% p-formaldehyde-lysine-periodate (PLP) for 3 h and equilibrated in sucrose as described (25, 26). Tissue sections (5 µm) were stained by fluorochrome-conjugated Ab. Only Ab directly conjugated with Alexa Fluors were used in tissue staining because secondary staining reagents gave high non-specific background. Fluorochrome-labeled isotype control Ab showed no background staining above tubule autofluorescence in confocal microscopy. Four-color confocal images were captured on a Zeiss LSM-700 confocal microscope equipped with 405, 488, 561, and 633 nm laser lines at the University of Virginia Advanced Microscopy Center and analyzed by the program ZEN (Zeiss, Thornwood, NY). For the enumeration of macrophage infiltration, glomeruli in 4 low power fields were captured by confocal microscopy and scored for CD11b⁺F4/80⁻I-A⁻ cells, with 20 – 30 glomeruli scored per mouse.

Glomerulus and Glomerular Cell Isolation

Magnetic bead isolation of glomeruli was performed as described (20, 27). In brief, anesthetized mice were perfused with 40 ml of cold PBS containing 8 × 10⁷ magnetic beads through the right ventricle. The excised kidneys were minced and incubated for 25 min at 37°C with 30 µg/ml Liberase and 50 µg/ml DNase I in culture medium (RPMI1640 containing 20% heat-inactivated FBS (hiFBS), 25 mM HEPES, pH 7.4, 2 mM glutamine, 50 U/ml penicillin and 50 µg/ml streptomycin (Invitrogen); 2.5 ml/mouse). The suspension was pipetted a few times, and passed through a 106 µm Endecotts sieve (3 in. diameter, Thomas, Swedesboro, NJ). The filtrate containing the glomeruli were pipetted into 15 ml conical tubes and the glomeruli were retrieved on a DynaMag 15 magnet (Invitrogen). After 4 to 5 washes with 10 ml each of cold PBS, the glomeruli were 98% pure. An isolation efficiency of 24,000 glomeruli per kidney comparable to published results was obtained (20, 27). The glomeruli were digested with 0.75 mg/ml Liberase TM and 50 µg/ml DNase I in 20% hiFBS for 30 min initially and 20 min periods subsequently with repeated pipetting between periods. Cellular dispersal was monitored under the microscopy. The total glomerular cell yields were 1.5 – 3 × 10⁶ cells per mouse.

Flow Cytometry Analysis and Sorting

Single cell suspensions from lymph node, kidney, and spleen were prepared as previously described (28), stained with fluorochrome-labeled Ab, and analyzed on a Cytek (Fremont, CA, USA)-modified FACSCalibur (BD Bioscience, Palo Alto, CA, USA) equipped with 3 lasers for 8-color analysis. For myeloid cell analysis, single cell suspensions were stained with the following fluorochrome-conjugated Ab: Brilliant Violet (BV) 421-anti-Ly6C, BV-510-anti-Ly6G, A488-anti-I-A, PE-anti-B220, PE-Cy7-anti-CD11b, A647-anti-F4/80, and allophycocyanin-eFluor780-anti-CD45. For glomerular parenchymal cells and other leukocytes, glomerular cells were stained with PacBlue-anti-CD73, BV-510-anti-CD31, A488-anti-CD102, PE-anti-CD3, PE-Cy7-anti-CD105, A647-anti-NK1.1, and allophycocyanin-eFluor780-anti-CD45. For T cell analysis, cell preparations were stained with BV421-anti-CD4, BV510-anti-CD11b, PE-Cy7-anti-CD8, and allophycocyanin-eFluor780-anti-CD45 plus A488-, PE-, and A647- conjugated mAb against activation markers. 7-AAD monitored in the PE-Cy5.5 channel was used to discriminate between live and dead cells. Flow cytometry data were analyzed by the program FlowJo (Ashland, OR).

RNA isolation and cDNA Real-Time PCR

The methods for the quantitation of gene expression of flow-sorted macrophages by real-time PCR has been described (26). Total RNA was isolated from sorted cells by TRIzol (Invitrogen) and cDNA obtained by the Advantage RT-for-PCR kit (BD Clontech). Jumpstart Taq polymerase (Sigma, St. Louis, MO) was used in real-time PCR reactions in a Bio-rad SFX thermocycler (Bio-rad, Richmond, CA) with the conditions: 94°C for 1 min, 39 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 1 min and 1 cycle of 94°C for 1 min, 58°C for 30 s, and 72°C for 5 min. Primer pairs designed by Primer 3 with 100 – 300 bp PCR products are shown in Supplemental Table I. Satisfactory amplification were confirmed by discrete melt curves with single peaks above 78°C.

Albuminuria Measurements

Mice were kept in metabolic cages (Techniplast, Buguggiate, Italy) and urine was collected daily. Urine albumin was measured by ELISA using goat anti-mouse albumin Ab (Bethyl Laboratories, Montgomery, TX) for capture and HRP-conjugated goat-anti-mouse albumin (Bethyl) as the detection Ab. O-Phenylenediamine was used as the HRP substrate and after reaction termination by 3N HCl, absorbance was measured at 492 nm. Mouse albumin (EMD Chemicals, San Diego, CA) was used to generate standard curves. The detection limit of the assay was 100 pg/ml. Albumin excretion was normalized against urine creatinine measured by the modified Jaffe assay (29) as described by the manufacturer (Pointe Scientific, Canton, MI). Proteinuria in each experiment was normalized to the highest value in % and averaged to adjust for variations between experiments.

Statistics

Statistical analysis was performed using Prism (Graphpad Software Inc., La Jolla, CA). Statistical significance was assessed by 2-way ANOVA with proteinuria, days after anti-GBM injection, and Ab treatment, or cell number, leukocyte population, and treatment as variables and followed by Tukey’s multiple comparison test. The following are used to designate p values: *, <0.05; **, <0.01; ***, <0.001; and ****, <0.0001.

Results

Intraglomerular CD11b⁺F4/80⁻I-A⁻ macrophages accumulation in NZM mice with severe proteinuria

The macrophage markers F4/80 and CD11b were used to detect glomerulus-infiltrating macrophages in lupus-prone NZM mice. In young NZM mice with no disease, few macrophages were detected in the glomerulus (Fig. 1A–a). However, large numbers of infiltrating CD11b⁺ macrophages were found in NZM mice with severe proteinuria (Fig. 1A–c). They are F4/80⁻, which suggest that they are blood-derived and distinct from the F4/80⁺ resident macrophages found in the interstial and periglomerular regions in sick NZM mice (22, 24). They are also I-A⁻ and thus are not dendritic cells (DC). Scoring of NZM intraglomerular macrophages showed that a mean of 13 macrophages/glomerulus were found in NZM mice with severe proteinuria (Fig. 2A). Few were found in R27 mice which do not develop cGN with severe proteinuria regardless of age ((24); Fig. 1A–b, 1A–d). Glomeruli of male NZM mice which do not exhibit cGN also contained few macrophages (Fig. 2). Thus blood-derived CD11b⁺F4/80⁻I-A⁻ macrophages are the major glomerulus-infiltrating leukocyte population in NZM mice and this infiltration only occurs in NZM mice with severe proteinuria.

FIGURE 1.

Infiltration of CD11b⁺F4/80⁻I-A⁻ macrophages in lupus nephritis. Kidneys from NZM mice with varying ages (3 Mo – 15 Mo) and proteinuria status (Trace (Tr) – 3+) were stained by Ab against the indicated Ag and images were captured by confocal microscopy according to Materials and Methods. (A) Kidneys of NZM mice with and without GN and the congenic R27 mice of varying ages with cGN resistance (24) were stained with mAb against macrophages. (B, C) Staining of tissue from NZM mice with cGN (3+ proteinuria by dipstick) by Ab against DC (B), and against T, B, and NK cells (C). In B-a - B-f and D-a - D-f, pseudocolors of 4-color staining are shown. In B, arrows show CD11b⁺I-A⁺CD11c⁺ DC and arrowheads show extraglomerular CD103⁺I-A⁺CD11c⁺CD103⁺ DC. In C, arrows indicate CD4⁺ T cells. Arrowheads are CD8⁺ T cells (a) and NK cells (b). (D) In mice with cGN, CD11b staining (b) colocalized with CD45 (arrows, a, b, e, f) but not with mesangial cell (c, e), and endothelial cell (d, f) markers. (E) CD11b staining (a) did not colocalized with podocyte marker (b). Arrows are CD11b⁺ cells. Glomeruli are circled. g, glomerulus. Bars are 10 µm.

FIGURE 2.

Extensive infiltration of CD11b⁺ macrophages were found only in NZM mice with severe proteinuria. (a) Urine from young 3 month old NZM mice, male NZM mice, R27 mice, and old NZM mice with no proteinuria and NZM mice with severe proteinuria (3+ and 4+ with dipsticks) were collected and assayed for urine albumin and creatinine. (b) For determining glomerular macrophage numbers, the mice were sacrificed, and kidneys fixed, sectioned, and stained, and CD11b⁺F4/80⁻I-A⁻ glomerular infiltrating macrophages were scored by fluorescence microscopy. The numbers in parenthesis are mouse numbers used in each group.

CD11b⁺ macrophages are distinct from DC and glomerular parenchymal cells

Glomulus-infiltrating CD11b⁺ macrophages are distinct from DC because they do not express the DC markers I-A and CD11c (Fig. 1B–a – 1B–e; asterisks). This immunofluorescence method stains all expressed I-A and thus rules out the possibility that these macrophage express intracellular I-A , as is the case in immature DC {Pierre, 1997 #139}. In sick NZM mice, there were some intraglomerular CD11b⁺ DC which are CD11b⁺I-A⁺CD11c⁺ (Fig. 1B–a, b, e; arrows). However, few CD103⁺ DC were found inside the glomerulus and they occur only in the interstitium (Fig. 1B–d, –f, arrowheads). Because glomerular mesangial cells may express macrophage markers (30), confocal microscopy were used to differentiate mesangial cells and infiltrating macrophages (Fig. 1D). Staining with Ab against CD31 for endothelial cells (31) and nephrin for podocytes (32) were also performed to differentiate macrophages from other glomerular parenchymal cells. Mesangial cells identified by integrin α8 (Itgα8; (33)) did not colocalize with CD11b, thus confirming that the CD11b⁺ cells were indeed infiltrating cells (Fig. 1D–e). Furthermore, the CD11b⁺ cells were CD45⁺ (Fig. 1D–e, arrows) and thus were hematopoietic cells. CD11b⁺CD45⁺ macrophages were also distinct from the CD31⁺ endothelial cells (Fig. 1D–e,–f) and the nephrin⁺ podocytes (Fig. 1E–b). In the glomeruli of sick NZM mice, there were a small number of infiltrating T cells (Fig. 1C–a, 1C–b) but few NKp46⁺ NK cells (Fig. 1C–b) and B220⁺ B cells (Fig. 1C–a).

CD11b⁺ macrophages were the most numerous glomerulus-infiltrating cell population in LN

To corroborate the immunofluorescence results on glomerulus-infiltrating cell types and numbers in lupus nephritis, flow cytometry analysis of single cell suspensions of magnetic bead-isolated glomeruli was performed (Fig. 3). Glomeruli preparation obtained by this method was highly purified and decapsulated (Fig. 3A), and thus do not contain periglomerular resident F4/80⁺ macrophages (22, 24). In NZM mice with severe proteinuria, the CD45⁺ leukocytes were consisted primarily of myeloids cells (>80%) with the great majority being the CD11b⁺I-A⁻ macrophages (Fig. 3B–f) which are also F4/80⁻ (not shown). Most of these CD11b⁺ macrophages were Ly6C⁻ (85%, Fig. 3B–g). A significant number of CD11b⁺I-A⁺ DC-like cells, most of which were Ly6C⁻, were also found (Fig. 3B–f, 3B–h). PMN constituted only a small proportion of the infiltrating leukocytes (Fig. 3B–d). There was a small population of CD11b⁻I-A⁺ population (Fig. 3B–f), most of which were B cells with negligible numbers of CD11b^Low DC (not shown). T cell were consistently found as a small population among the infiltrating cells (Fig. 3B–e). To control for poor glomeruli recovery in diseased mice by the magnetic trapping method because of structural destruction, glomerular parenchymal cells were quantified. Glomerular endothelial cell numbers remained constant in diseased and nondiseased mice, thus supporting the good recovery of glomeruli by the isolation procedure (Fig. 3E) and suggest that glomeruli capillaries remain intact for magnetic bead trapping despite podocyte damage in NZM mice with cGN.

FIGURE 3.

CD11b⁺ macrophages, DC, and T cell comprise the major glomerular infiltrating leukocyte population in NZM mice with cGN. (A) Glomeruli were isolated from mice with severe proteinuria (3+ and 4+ by dipstick) by magnetic beads as described in Materials and Methods. Bar in C is 200 µm, 100 µm and 50 µm in panels a, b, and c respectively. (B) A representative analysis of glomerular CD45⁺ cell from NZM with severe proteinuria is shown. After gating for total cells (a), CD45⁺ 7-AAD⁻ live cells (b), and singlets (c), the cells were analyzed for Ly6G⁺Ly6C⁺ PMN (d) and other myeloid cells. Ly6G⁻ cells contained mostly I-A⁻CD11b⁺ macrophages (f). The macrophages are mostly Ly6C^Lo (g). Similarly, DC in f are also mostly Ly6C^Lo (h). The CD11b⁻I-A⁺ population in f are about 80% B220⁺ B cells. T cells are shown as CD3⁺Thy1.2⁺ cells among the CD45⁺ population (e). (C) Similar cell populations with much smaller numbers are found in the glomeruli of young NZM mice with trace (Tr) urinary protein by dipstick. (D) The number of glomerular leukocyte population in 3 month old NZM mice and NZM mice with cGN and severe proteinuria are shown. The number of glomerular leukocyte populations from 4 month old C57BL/6 (B6) mice was similar to that in young NZM mice. (E) The numbers of glomerular endothelial cells enumerated by flow cytometry (see Fig. 5D) were similar between 3 month old NZM mice and sick NZM mice with cGN, thus validating the yield of glomeruli in NZM mice with cGN. In D and E, mean ± SEM of 5 young NZM mice, 5 NZM mice with severe proteinuria, and 4 C57BL/6 mice are shown. In D, sick NZM mice with cGN have significantly higher CD11b⁺ macrophage, I-A⁺ DC, and CD11b⁻I-A⁺ cells (D, p<0.0001).

An average of 2.7 leukocytes/glomerulus was found in glomeruli of young nondiseased NZM mice by flow cytometry (Fig. 3C, 3D). This number translates to about 0.4 leukocyte per 5 µm glomerulus sections based on the calculation that 15 sections of 5 µm sections would be needed to include the whole glomerulus (73 µm diameter) and only 1/7 of the sections will contain a cell 10 µm in diameter. This number approximates the results obtained by microscopic scoring (Fig. 2). The glomerulus-infiltrating leukocytes populations in young NZM mice were similar to those in older NZM with proteinuria (Fig 3C). There were higher percentages of PMN and T cells but lower percentages of CD11b⁺ macrophages in these infiltrates (Fig. 3C–b, 4). In absolute numbers, there were marked increases of CD11b⁺I-A⁻ macrophages and I-A⁺CD11b⁺ DC and a moderae increase of CD11b⁻I-A⁺ cells in NZM mice with cGN compared to those without overt disease (Fig. 3D). These results implicated that intraglomerular CD11b⁺F4/80⁻I-A⁻macrophages and/or I-A⁺CD11b⁺ DC but not PMN were responsible for the cellular injury in NZM mice with cGN.

FIGURE 4.

Blocking Mac-1 and ICAM-1 binding inhibits anti-GBM-induced proteinuria and myeloid infiltration into the glomerulus in NZM mice. (A) NZM mice (10 – 12 weeks old) were injected on day 0 with anti-GBM to induce nephritis. Anti-CD11b (M1/70), anti-ICAM-1 (YN1), and rat isotype control (LTF-2) mAb were injected (100 µg/injection) on alternate days beginning on day 0. Mice were sacrificed on days 3 and 7. Urine was collected daily and assayed for albumin and creatinine. The albumin/creatinine ratio were normalized to the highest value in each experiment which is set at 100% with the range for individual experiments at 3.7 – 4.9 mg/mg. The % maximum for replicates were averaged and displayed as mean ± SEM. The experiments were performed 4 times and a total of 10, 6, and 8 mice were used for LTF-2-, M1/70-, and YN1- injected groups respectively. (B) Confocal microscopy of kidneys from control LTF-2-injected mice at day 7 after anti-GBM injection. The Ag visualized are as shown in the labeling and the micrographs show low (a) and high power (b, c) magnifications. Glomeruli (a,b) are shown by mesangial cell labeling (blue). CD11b⁺F4/80 macrophages are shown by arrows (b, c, red). F4/80⁺ macrophages marked by asterisk (c, blue or magenta) showed varying degrees of CD11b expression (red). Circles demarcates glomeruli (b, c). g, glomeruli. Bars, 50 µm in a and 10 µm in b, c. (C) Single cell suspensions of purified glomeruli from anti-GBM-injected mice on day 3 immediately preceding proteinuria onset were analyzed by flow cytometry as described in Fig. 3. Representative analyses of mice from LTF-2-, M1/70-, and YN1- injected group are shown in a, b, and c respectively and the mean ± SEM of various cell types with 4 mice in each group and havested on day 3 is plotted and shown in (E). (D) The direct comparison of intraglomerular myeloid cell population numbers for young NZM mice, NZM with cGN (3+ proteinuria, Fig. 3D) and NZM mice injected with anti-GBM plus control LTF-2 mAb and harvested on day 3 (from Fig. 4E) and on day 7 (10 mice) are plotted. *, p<0.05; ***, p<0.001; ****, p<0.0001.

Glomerular leukocyte infiltration was not restricted to the autoimmune prone mice. Similar cellular populations and numbers were found in the non-lupus prone C57BL/6 mice when compared to nondiseased NZM mice (Fig. 3D).

Anti-GBM Ab induced proteinuria and glomerular PMN and CD11b⁺F4/80⁻I-A⁻ macrophage infiltration

Anti-GBM induction of proteinuria and GN in NZM mice was used as a more consistent and predictable method for studying the functional role of intraglomerular macrophages in GN (24). Severe albuminuria in 10 −12 week old NZM mice was induced by day 4 with anti-GBM (Fig. 4A) whereas no increase in urine albumin was found in control mice injected with sheep serum (not shown). Extensive infiltration of Ly6G⁺CD11b⁺ PMN and CD11b⁺F4/80⁻ blood-derived macrophages were found in the kidney (Fig. 4B–a). This cellular infiltration was concentrated in the glomerulus as expected due to the deposition of immune complexes and complement activation at the GBM (Fig. 4B–a – c). Few F4/80⁺ macrophages were found in the glomerulus (Fig. 4B–c). Flow cytometry analysis showed that there were marked increases in PMN and CD11b⁺ macrophages in the glomerulus on days 3 and day 7 after anti-GBM administration (Fig. 4C, 4D). Although the glomerulus-infiltrating cell populations were similar based on surface markers, there were clear differences in the glomerular infiltrating leukocyte population numbers between NZM mice with cGN and mice with anti-GBM-induced nephritis (Fig. 4D). PMN numbers up to 12 times higher than that in unimmunized mice and in NZM mice with cGN were found in mice with anti-GBM-induced nephritis (Fig. 4D). On the other hand, only about half the number of CD11b⁺F4/80⁻ macrophages and 25% of I-A⁺CD11b⁺ DC were found in anti-GBM-treated mice compared to sick NZM mice with cGN (Fig. 4D). The numbers of glomerulus-infitrating T cells were similar in both disease models. These data showed that similar populations of myeloid cells infiltrated the glomerulus in acute and chronic GN although the numbers of individual populations varied greatly.

Anti-CD11b and anti-ICAM-1 mAb block anti-GBM-induced proteinuria and CD11b⁺ PMN and macrophage infiltration in NZM mice

In short term assays, glomerular leukocyte infiltration is particularly dependent on Mac-1 (CD11b/CD29) interaction with ICAM-1(21). The effects of blocking CD11b and ICAM-1 on leukocyte infiltration on days 3 and 7 and proteinuria from days 3 to 7 after anti-GBM injection was examined. Both Ab blocked proteinuria almost completely. However, anti-CD11b only reduced PMN by 50–55% on both days (Fig. 4C–b, 4E; day 7 not shown). Anti-ICAM-1 inhibited this infiltration better at 75% (Fig. 4C–c, 4E). This moderate reduction in PMN infiltration seemed to be inadequate to account for the inhibition of proteinuria. Inhibition of I-A⁻ macrophages and I-A⁺ DC infiltration by anti-CD11b and anti-ICAM-1 was more pronounced and at 65% and 75% - 80% respectively (Fig. 4C, 4E). This blocking showed that the reduction in CD11b macrophage numbers correlated better than PMN numbers in proteinuria inhibition in anti-GBM-injected mice. Only small reduction in CD11b⁻I-A⁺ cell and T cell infiltration were found in anti-CD11b and anti-ICAM-1 injected mice (Fig. 4C, 4E).

Glomerular endothelial cells express high levels of ICAM-2 constitutively and upregulates ICAM-1 but not VCAM-1 in inflammation

Because Mac-1 on blood leukocytes play a major role in glomerular surveillance and emigration, the expression of CD11b ligands on glomerular endothelial cells were examined. Little ICAM-1 was detected on glomerular endothelial cells under homeostatic conditions (Fig. 5A) but ICAM-1 upregulated after anti-GBM injection (Fig. 5B). Very high levels of ICAM-1 was expressed by glomerular endothelial cells in NZM mice with cGN (Fig. 5C). ICAM-2, on the other hand, was expressed constitutively at high levels in the glomerulus in both non-diseased and diseased mice (Fig. 5). Flow cytometry confirmed the expression of ICAM-2 by glomerular endothelial cells (Fig. 5D–c). Thus ICAM-2 may play a significant role in vascular CD11b⁺ myeloid cell surveillance in the glomerulus. Interestingly, ICAM-2 was also expressed on mesangial cells, thus suggesting that CD11b⁺ macrophage infiltration in the mesangium may also involve this Mac-1-ICAM interaction (Fig. 5D–e). VCAM-1, on the other hand, was expressed at low levels in the glomerulus even in nephritic kidneys (Fig. 5A–5C). Some VCAM-1 was expressed at the Bowman’s capsule (Fig. 5, arrowheads) and at high levels on kidney tubules in NZM mice with cGN (Fig. 5C, asterisks).

FIGURE 5.

Glomerular endothelial cells express high constitutive levels of ICAM-2, high ICAM-1 in NZM mice with cGN, and low levels of VCAM-1. Kidney sections from young healthy NZM mice (A), NZM mice on day 7 after injection with anti-GBM (B), and sick NZM mice with cGN (C) were analyzed by 4-color confocal microscopy. ICAM-2 and VCAM-1 were represented separately in green and merged with ICAM-1 (red) and CD31 (blue). White circles outline the glomeruli. Arrowheads indicates VCAM-1 staining of the Bowman’s capsule (A, B). * in C, left bottom panel, VCAM-1 staining of tubules. g, glomerulus. Bars, 50 µm in left column, 20 µm in right column. (D) ICAM-2 (CD102) expression by glomerular endothelial cells and mesangial cells are shown by flow cytometry. Single cell suspensions of isolated glomeruli from 3 month old NZM mice were stained for glomerular endothelial cell (b, c; CD31) and mesangial cell (d, e; CD73) markers and ICAM-2. Anti-CD105 stained both endothelial cells and mesangial cells (28).

Glomerular CD11b⁺ macrophage express alternatively activated macrophage markers

To further confirm the identity of these glomerular macrophages and their functional role in GN, a large Ab panel against macrophages and their subsets was used to stain the kidneys of NZM mice with cGN (Fig. 6). The M2 macrophage markers Mac-2, PD-L1, and MMR colocalized with CD11b, thus indicating that CD11b⁺F4/80⁻ macrophages bear the characteristics of alternatively activated macrophages (Fig. 6–a, b, d). These macrophages are also positive for the pan macrophage marker CD14 (Fig. 6–c) but negative for Mgl1/2 and Slamf9 (Fig. 6–e, f). Anti-Mgl1/2 and Slamf9 stained extraglomerular macrophages (Fig. 6–e, f; arrowheads). In other staining studies, the M2 markers Mac-2, PD-L1, and MMR staining did not coincide with I-A, thus confirming that the M2 markers were staining the CD11b⁺F4/80⁻I-A⁻ macrophages.

FIGURE 6.

CD11b⁺ glomerular macrophages exhibit M2-like phenotype and are F4/80⁻, CD11c⁻, and CD169⁻. (A) Kidney sections from NZM mice with cGN and severe proteinuria (3+ to 4+ by dipstick) were stained by mAb against CD11b and F4/80, and Ab against the indicated markers in each row. Confocal microscopy was performed as described in Fig. 1. Arrows show the CD11b⁺ macrophages colocalizing with the macrophage subset markers. Arrowheads show marker⁺ cells that are not CD11b⁺ macrophages (e, f). (B) Lack of F4/80, CD11c, and CD169 expression by glomerular CD11b⁺I-A⁻F4/80⁻ macrophages. Anti-GBM injected NZM mice were harvest on day 7. Glomerular and interstitial leukocytes were analyzed by flow cytometry and one representative among triplicates is shown. Interstital CD11b⁺ macrophages were gated on F/80^Lo and F4/80⁺ populations (a) and used as positive controls. The expression of F4/80 (b), CD11c (c), and CD169 (d) by intraglomerular CD11b⁺I-A⁻ macrophages with or without Ly6C expression were compared against that by interstitial (Int) F4/80⁺ and F4/80^Lo macrophages. The expression by Ly6C⁻ glomerular macrophages are shown and the expression by Ly6C⁺ macrophages were similarly negative. (C) Confocal microscopy analysis of kidneys in (B) confirmed that CD169-expressing cells (green, arrows) is restricted to the extraglomerular macrophages and the intraglomerular macrophages, which are CD11b⁺ (red, arrowheads) and Ly6G^Lo (not shown) are CD169. In all panels, glomeruli are outlined and bars are 20 µm.

CD11b⁺F4/80⁻I-A⁻ macrophages expressed macrophage-specific and M2-specific mRNA

The intraglomerular CD11b⁺F4/80⁻I-A⁻ macrophages were further characterized by their expression of macrophage-specific genes (Fig. 7). Gene expression by the interstitial resident F4/80⁺I-A⁺CD11c⁺ macrophages (22) was used as a reference. These latter macrophages expressed high mRNA levels of macrophage markers Csf1r, Cx3cr1, and Cd169 and lower levels of the markers Fcgr1, and Mertk. Furthermore, because of their resident nature and their specialized capability to capture immune complexes (22), they exhibit an M2 phenotype with high expression of M2-specific genes Mrc1, Lgals3, and Arg1 and low expression of M1 markers Nos2, Ccr7, Il6, Il12a, Il12b, and Tnf (Fig. 4; (34, 35)). By comparison, the predominant Ly6c⁺ and the minor Ly6C⁺ glomerular macrophage populations expressed high levels of the macrophage marker Csf1r and moderate levels of Cx3cr1 when compared to F4/80⁺ resident macrophages (Fig. 4A) but their expression of the macrophage markers Fcgr1 and Mertk (36) were low. However, their macrophage lineage were further strengthened by their high expression of M2-specific markers Cd274, Lgals3, Il10, Klf4, and Retnla (Fig. 4D, 4E). Although the glomerular macrophage populations expressed low levels of M1 marker, they expressed high levels of inflammatory cytokines Tnf, Il6, and Ccl2, a pattern characteristic of immune complex-activated M2b macrophages (Fig. 4B, 4C; (23, 37, 38)). These results further confirm the microscopy results and indicate that the intraglomerular macrophages exhibit a M2b gene expression pattern.

FIGURE 7.

Intraglomerular CD11b⁺I-A⁻ macrophage M2b-like gene expression pattern. (A) Intraglomerular CD45⁺CD11b⁺I-A⁻Ly6G^Lo macrophages with and without Ly6C expression were FACS-purified from mice injected with anti-GBM and harvested on day 7. Their mRNA expression was analyzed by real-time PCR. Interstitial CD45⁺CD11b⁺I-A⁺F4/80⁺ macrophages (A) were sorted and their mRNA were used for comparison. Genes expressed in different macrophage subsets were grouped in the same graphs (B–G). Gene expression values were normalized against that of β-actin and shown as mean of 3 mice (2 independent experiments) ± SD. Primers used for each gene are shown in Supplemental Table I.

Intraglomerular CD11b⁺F4/80⁻ macrophages are distinguishable from kidney interstitial F4/80⁺ and F4/80^Lo macrophages and myeloid-derived suppressor cells (MDSC)

Renal interstitial F4/80⁺ macrophages have been shown to be CD11b^IntermCD11c^IntermI-A⁺CD64⁺ (22, 34, 35). The intraglomerular CD11b⁺F4/80⁻ macrophages were found to be very different in surface marker and gene expression (Fig. 6, 7). Confocal microscopy showed that glomerular macrophages were F4/80⁻, CD11c⁻, and F4/80⁻ (Fig. 1A, 1B, 6A). Flow cytometry analysis confirmed that these glomerular macrophages were F4/80⁻ and CD11c⁻ (Fig. 6B). Furthermore, they expressed much lower levels of Fcgr1 (Cd64) mRNA (Fig. 7B) and their H2-Ab1 (IAb) gene expression is 0.2% to 0.5% of that in F4/80⁺ interstitial macrophages (not shown). Glomerular CD11b⁺F4/80⁻ macrophages were similarly different in phenotype from the interstitial F4/80^Lo macrophages. These latter macrophages are CD11b^HiCD11c^LoI-A⁺CD64⁺(35) and direct comparison with the F4/80⁻CD11c⁻I-A⁻ glomerular macrophages for F4/80 and CD11c expression showed that these populations are distinct (Fig. 1B, 6A, 6B–b, –c). Glomerular macrophages are also Cd64⁻ (Fig. 7B). Monocytic MDSC has been reported to be IL-4Ra⁺F4/80^LoMgl1/2⁺ (39, 40). A separate inflammation-suppressing MDSC-like population expresses CD169 (41). Besides being F4/80⁻ (Fig. 6A, 6B–b), the intraglomerular macrophages are Mgl1/2⁻ as shown by immunofluorescence (Fig. 6A–e), CD169⁻ as determined by flow cytometry (Fig. 6B–d), immunofluorescence (Fig. 6C), and real-time PCR (Fig. 7C), and very low in Il4ra mRNA (Fig. 7C) and thus are not likely to be identical to the conventional or the CD169⁺ MDSC populations.

Blocking PD-L1 on glomerular macrophages inhibited anti-GBM-induced proteinuria but not myeloid cell glomerular infiltration

In both spontaneous cGN and anti-GBM-induced GN in NZM mice, PD-L1 expression in kidneys was found predominantly in the glomeruli (Fig. 8A–a – A–d, A–i –A–l, arrows). CD11b⁺ macrophages and not other myeloid cell types were the major PD-L1 expressing cells (Fig. 8–e – 8–h, 8–m – 8–p). PD-L1 expression also coincided with the M2 marker Mac-2 (Fig. 8A–h, arrowheads) , but not with the interstitial resident macrophage marker F4/80 (Fig. 8A–o, 8A–p). Intraglomerular PMN with high CD11b expression also showed no detectable PD-L1 expression (Fig. 8A–n, 8A–p, asterisk). Blocking PD-L1 inhibited anti-GBM-induced proteinuria (Fig. 8B) although no significant differences in glomerular myeloid cell infiltration was found (Fig. 8C). These data support the hypothesis that the predominant PD-L1-expressing cells, i.e, the glomerular infiltrating CD11b⁺F4/80⁻I-A⁻ macrophages, play significant roles in causing glomerular damage and proteinuria.

FIGURE 8.

PD-L1 expression by intraglomerular CD11b⁺F4/80⁻I-A⁻ macrophages in NZM mice with cGN and with anti-GBM-induced nephritis and anti-PD-L1 mAb blocking of proteinuria. (A) Glomerular CD11b⁺ macrophages are the predominant PD-L1 expressing cell population in NZM with severe proteinuria (4+ proteinuria by dipstick; a – h) and anti-GBM-induced nephritis (i – p). Kidney sections were stained with mAb against PD-L1, CD11b, and Mac-2 or F4/80 as labeled and analyzed by confocal microscopy as described in Fig. 1. Low (a – d, i – l) and high magnification (e – h, m – p) micrographs in which glomeruli are shown by arrows and circles respectively are presented. *, Ly6G^hi PMN (m – p). Bars, 50 µm in low magnification and 10 µm in high magnification micrographs. Mac-2 intensity in A-c and A-d were highly attenuated because of the bright Mac-2 staining of tubular cells. (B) Inhibition of anti-GBM-induced proteinuria by anti-PD-L1. mAb (200 µg/injection) were administered on alternate days beginning on day 0 and proteinuria were measured as described in Fig. 4. Proteinuria results are expressed as mean ± SEM of values normalized to the maximal albumin/creatinine value in each experiment. Eight mice were used in each group and the experiment has been performed 4 times. (C) Intraglomerular infiltrating leukocytes were analyzed on day 3 as described in Fig. 3 by flow cytometry and the cell numbers in myeloid populations and T cells are presented as mean ± SEM for 3 mice in each group. The experiment has been performed twice.

Anti-PD-L1 did not induce statistically significant differences in T cell and myeloid cell subpopulation numbers in kidneys and lymphoid organs

Whether anti-PD-L1 inhibition of proteinuria was caused by changes in cell numbers in T cell subpopulations, renal lymph node, kidney, and spleen T cell subsets in mice treated with anti-GBM plus anti-PD-L1 were analyzed. The total numbers of CD4⁺ T cells in mice treated with anti-GBM plus anti-PD-L1 was significantly higher than that in mice treated with control sheep serum (Fig. 9A–a) in renal lymph nodes but not in other organs (Fig. 9A–b, 9–A–c). Anti-PD-L1 did not induce significantly higher CD4⁺ T cells in all 3 sites compared to anti-GBM plus isotype control mAb-treated mice. There was no significant difference in CD8⁺ T cell numbers among the 3 treatement groups in all sites. When CD4⁺ and CD8⁺ T cells were further subdivided into subsets based on activation markers CD25 and CD69, there were no difference in cell number among all treatment groups for the subsets and in all sites. Similarly, cell numbers in CD44⁺ and CD62L⁻ memory or effector CD4⁺ or CD8⁺ T cell subsets were similar among all treatment groups in all sites. Thus anti-PD-L1 induced no clear difference in T cell subset numbers compared to control LTF-2 mAb in the injury site and lymphoid organs despite the difference in proteinuria induced by anti-GBM. No difference in myeloid cell populations in kidneys or spleen were found between anti-PD-L1 and control LTF-2 mAb treatment (Fig. 9B). Thus it is likely that anti-PD-L1 mAb inhibits anti-GBM-induced proteinuria by either blocking cell-cell interaction through PD-L1 or through anti-PD-L1-induced cell signaling.

FIGURE 9.

T cell and myeloid cell population numbers in anti-GBM-treatment mice with and without anti-PD-L1 blockade. NZM mice primed with sheep IgG were injected with control sheep serum (serum-Veh) or nephrotoxic anti-GBM antiserum followed by control (αGBM-LTF2) or anti-PD-L1 (αGBM-αPD-L1) mAb administered and harvested on day 5. Leukocytes isolated from renal lymph nodes, kidney, and spleen were analyzed for T cell (A) and myeloid cell numbers (B) as described in Materials and Methods. Cell numbers are mean ± SD from 3 experiments with a total of 4, 6, and 6 mice from serum control, LTF2 control, and anti-PD-L1-blocked groups respectively. *, p<0.05.

Discussion

In this study, we have characterized glomerulus-infiltrating leukocyte populations in lupus nephritis extensively. The most significant population in both spontaneous LN and anti-GBM-induced GN is the blood-derived CD11b⁺F4/80⁻I-A⁻ macrophage population which expresses surface markers and genes that identify them as M2b macrophages (23). Besides expressing the M2 markers Mac-2, PD-L1, MMR, Klf4, and IL-10, they also express inflammatory cytokines TNF-α and IL-6 but do not express M1-specific genes. The marker expression for these macrophages is summarized in Table I. They are distinct from the resident interstitial F4/80^hiCD11c^IntI-A⁺ macrophages whose increases in numbers correlate with the disease state of SLE in lupus-prone mice (34). These interstitial resident macrophages are also primarily responsible for immune complex uptake in the interstitium by traversing the endothelial barrier (22). However, they do not migrate into the glomerulus. The glomerular macrophages are also phenotypically distinct from the interstitial F4/80^Lo macrophages which are CD11c^LoI-A⁺CD64⁺CX₃CR1^Hi (35) and distinct from the F4/80^Lo M1-like macrophage population found in mice with ischemia-reperfusion-mediated early-onset lupus nephritis (42). Glomerular macrophages are F4/80⁻Mgl1/2⁻CD169⁻ (Fig. 6) and express very low levels of IL-4Rα mRNA (Fig. 7C) and thus are unlikely to be the F4/80⁺Mgl1/2⁺IL-4Rα⁺ MDSC from tumor-bearing mice (39, 40) or the CD169⁺ inflammation-suppressing MDSC-like macrophages (41). The results showed that intraglomerular CD11b⁺F4/80⁻I-A⁻ are M2b-like and phenotypically distinct from other characterized macrophage populations in the kidney. The unique phenotype and gene expression profile of these macrophages may reflect the difference between the isolated environmental conditions in the glomerulus compared to the kidney interstitium.

Table I.

Marker expression by glomerular CD11b⁺F4/80⁻I-A⁻ macrophages

Markers	Glomerular Mac	Interstitial Resident Mac
Mac
CD11b	+++	+++
Ly6G	+	+
Ly6C	+ - +++	+
F4/80	−	++++
I-A	−	++++
CD11c	−	++
CD14	+++	++++
CD169	−	+++
CX3CR1	+	+++
Csf1r	++	++++
Mertk	+	++
Fcgr1	±	+++
M1
Nos2	+	−
CCR7	−	−
Il6	+ - +++	±
Il12a	±	±
Il12b	+	±
M1/M2b
Tnf	++++	−
Ccl2	+	++
M2
PD-L1	+++	±
Mac-2	+++	+++
MMR	± - ++	++++
Il10	++++	−
Klf4	++++	+
Arg1	++	+++
Retnla	+++	++

Marker expression were determined by fluorescence microscopy, flow cytometry or real-time PCR in this paper or earlier publication (24). The expression intensities were assessed by the comparison with the highest expression level by kidney cells. ±, 0 – 10%; +, 10% - 30%; ++, 30% - 60%; +++, 60% - 80%, ++++, 80% - 100%. Genes with expression levels determined by mRNA expression are shown in italics.

The Ly6C⁺ and Ly6c⁻ subpopulations in glomerular macrophages exhibit similar gene expression profiles (Fig. 7). Given the similar glomerular environment between chronic and acute GN, glomerular macrophages in NZM mice with cGN and with anti-GBM immunization may be equivalent. In cGN, intraglomerular macrophages are predominantly Ly6C⁻ (85% of I-A⁻ cells, 70% of total myeloid cells; Fig. 3B–g) and they are 6 – 7 times higher in number than the Ly6C⁺ macrophages. Thus immunofluorescence results of glomerular macrophages in NZM with cGN largely reflect the phenotype of Ly6C⁻ macrophages. It is unclear whether the Ly6C⁺ and Ly6C⁻ macrophage populations are related and can interconvert, or if they are derived separately from the blood GR-1^Hi and GR-1^Lo monocyte populations (43). Functionally the Ly6C⁻ intraglomerular macrophages are likely to be the most important intraglomerular leukocyte in LN because of their dominance in numbers, especially in view of the low number of other infiltrating cell types including PMN. The occurrence of a significant CD11b⁺CD11c⁺I-A⁺ DC-like population in cGN may be relevant in LN pathogenesis. Marked increases in intraglomerular DC have been found in TNF-α-injected mice (20) but not in anti-GBM-induced nephritis (Fig. 4C, 4D). DC has been shown to facilitate Th-17 T cell accumulation in unilateral ureteral obstruction and in LN (44)(45). It is possible that the proinflammatory Th17 cells are present among glomerular T cells and they induce glomerular damage upon DC stimulation. On the other hand, intraglomerular DC can be protective. In rat anti-GBM-induced GN, an intraglomerular CD8⁺ DC-like population provided protection by inducing apoptosis of antigen-specific CD4⁺ T cells (46).

The results showing that the interruption of CD11b–ICAM interactions abolished proteinuria in anti-GBM-induced nephritis and concomitantly reduced glomerulus-infiltrating PMN and macrophages support the important role of intraglomerular PMN and/or macrophages in inducing GN. The subsequent studies of anti-PD-L1 inhibition of proteinuria further focuses the nephritis-inducing function on glomerular macrophages since it is the major PD-L1-expressing population and these cells are in close proximity to the site of glomerular injury in GN. Thus we propose that intraglomerular macrophages are the primary candidate for inducing GN in nephritis models.

The mechanism of PD-L1-induced pathogenesis in autoimmune disease and GN is complex. PD-L1 interacts with both PD-1 and B7-1 and mediates both proinflammatory and anti-inflammatory responses (47). B7-1 cross-linking activates CD4⁺ T cells and increases T cell proinflammatory cytokine production (48). Disrupting PD-L1-PD-1 interaction by anti-PD-1 Ab protects lupus-like nephritis in NZB/W F1 mice by either reducing CD4⁺PD-1⁺ T cells (49) or promoting suppressor CD8⁺ T cell activity (50). However, these anti-PD-1 blocking experiments are long term studies and the results do not necessarily apply to the current short-term anti-GBM model. An additional possibility for PD-L1 function in LN is that macrophage PD-L1 stimulates the newly expressed anti-GBM-induced podocyte B7-1 which inactivates α₃β₁, the integrin which is required for podocyte cytoskeleton integrity. This PD-L1-induced loss of cytoskeletal integrity increases podocyte motility and leads to proteinuria (51, 52). The last possibility was explored by anti-B7-1 staining of podocytes without success, perhaps because of the low or transient expression of B7-1 on podocytes. Thus the mechanism for PD-L1 expressed on intraglomerular macrophages in inducing GN remain unclear. The possibility that anti-PD-L1 blockade in the anti-GBM model reduces inflammation by changing the cell numbers in T cell and myeloid cell subpopulations was examined (Fig. 9). However, no statistically significant differences were found. Thus more methodical studies on PD-L1 functions in GN by different inflammatory cell populations is needed. It is also uncertain why PD-L1 blockade in anti-GBM-induced nephritis failed to yield the opposite effect of accelerated inflammation found in alloimmune responses (53). Similarly, in anti-GBM-injected PD-L1^−/− C57BL/6 mice, more kidney inflammation and interstitial leukocyte infiltration were induced (54). The differences in mouse strain used or the mechanism of action between PD-L1 deficiency and blockade can account for the disparity between this earlier study and the current one.