A Role for Mac-1 (CDIIb/CD18) in Immune Complex–stimulated Neutrophil Function In Vivo: Mac-1 Deficiency Abrogates Sustained Fcγ Receptor–dependent Neutrophil Adhesion and Complement-dependent Proteinuria in Acute Glomerulonephritis

Mac-1–deficient mice were recently generated by standard gene targeting techniques (2). These mice and their wild-type mates are a mixed strain of 129SV and C57Bl/6. They are bred and maintained in a virus-antibody free facility at the Longwood Medical Research Center (Boston, MA). Experimental animals were 8–14 wk of age and were sex-matched. Both males and females were used. 25% of the experimental mice used at each time point were littermates from heterozygous matings. The remainder were the progeny of wild-type and Mac-1–null homozygous matings. Previously generated mice deficient in ICAM-1 (18) were of a pure C57Bl/6 background strain, were 8 wk old, and were age- and sex-matched with their wild-type mates. Mice deficient in complement C3 and C4 were 7–8-wk-old females of a mixed 129Sv and C57Bl/6 strain. ICAM-1 and complement-deficient mice were bred and maintained in a virus-antibody free facility at Harvard Medical School.

Anti-GBM nephritis was induced essentially as previously described by our group (13). In brief, preimmune urine was collected ∼1 h before injection of 0.25 ml anti-GBM antiserum per mouse (14). At 1, 2, 4, 8, and 18 h after the injection, urine and blood samples (by retroorbital bleed) were collected. Animals were then killed and both kidneys were harvested.

Kidneys were fixed in 4% paraformaldehyde, paraffin-embedded or frozen in OCT compound embedding medium (Miles, Inc., Elkhart, IN), and sectioned. 5-μM paraffin-embedded sections were stained for dichloroacetate esterase to identify PMNs and tissue was counterstained with nuclear-fast red (19). The number of PMNs in 100 glomeruli (50/each kidney) in cross-sections were counted per animal without prior knowledge of the genotype of the animal. For electron microscopy, glomeruli from one wild-type and one Mac-1–null nephritic animal, 1 h after anti-GBM antibody injection, were examined by standard techniques. 15 wild-type and Mac-1–null intraglomerular capillary PMNs were examined.

Rabbit anti-GBM antiserum deposition in mouse glomeruli was quantitated by immunofluorescence. 4 μM frozen sections prepared from kidneys harvested from wild-type and mutant mice 1 h after anti-GBM injections were incubated with a 1:200 and 1:350 dilution of FITC-tagged secondary antibody (goat anti–rabbit, Vector Laboratories Inc., Burlingame, CA). These dilutions were saturating as determined by the following method: secondary antibody dilutions ranging from 1:100 to 1:5,000 were applied to wild-type kidney sections and the fluorescence images from stained sections were acquired for 0.1 s using Oncor Image Analysis software (Oncor, Inc., Gaithersburg, MD), in combination with a Nikon microphot-FXA microscope and a 12-bit cooled CCD SenSys camera (Photometrics, Tucson, AZ), to obtain digital images. These digital images were then processed using a Optimas Image analysis program. An index of fluorescence specifically associated with the glomeruli was calculated using an Optimas program (designed by Edward Marcus, Ed Marcus Laboratories, Boston, MA) which calculates the following: average luminance (filtered) which is defined as the sum of gray levels/ number of pixels in the selected region both within selected glomeruli (Fg), and in the area within the tubules that is designated as the background (Bg). The filter function of the program is applied to smooth small luminance variations, before detection of the primary luminance peak. This was necessary because measurement of the staining intensity within the glomeruli might be affected by FITC-stained glomerular capillaries present within an unstained background. The ratio of Fg over Bg (Fg/Bg) was used as an index of fluorescence specifically associated with the glomeruli. Plotting of these indices obtained for the serial dilutions of secondary antibody assured that secondary antibody dilutions of 1:200 and 1:350 were saturating. Glomerular cross-sections of wild-type and mutant mice were processed with secondary antibody dilutions of both these dilutions and analyzed as described above. These dilutions gave similar results, therefore only data obtained from the 1:200 dilution is shown.

Urinary albumin was determined by a double-sandwich ELISA as previously detailed (13). Urinary creatinine was quantitated spectrophotometrically using a commercially available kit (Sigma Chemical Co., St. Louis, MO). To standardize urine albumin excretion for glomerular filtration rate, proteinuria was expressed as microgram of urinary albumin per milligram of urinary creatinine (20).

Peripheral blood was obtained by retroorbital sampling and was collected in 5 mM EDTA. Blood samples from 6 to 7 mice, typically yielding 6–7 ml of blood, were pooled and layered over a Nim.2 neutrophil isolation medium according to manufacturer's protocols (Cardinal Associates, Santa Fe, NM), and referred to as one experimental sample. Red cells contaminating the neutrophil preparation were lysed with ice-cold 0.2% NaCl. Neutrophils were resuspended in RPMI medium.

BSA–anti-BSA ICs and BSA were immobilized on glass coverslips. According to the manufacturer's technical information, the BSA (a 35% solution, ICN Pharmaceuticals, Inc., Costa Mesa, CA) was heat-shocked, and contained no vitronectin or fibrinogen. Glass coverslips (12 × 12 cm) were treated with 0.1 mg/ml poly-d-lysine (mol wt ⩾ 70,000, Sigma Chemical Co.) for 1 h, treated with 2.5% glutaraldehyde for 15 min, and washed extensively in ddH₂O followed by two PBS washes. Coverslips were then coated with 1 mg/ml BSA for 30 min, washed twice in PBS, incubated with 0.1 M glycine for 2 h to quench aldehyde groups, and then incubated with 20 μg of rabbit anti-BSA IgG (Sigma Chem. Co., St. Louis, MO) in 0.5 ml of PBS for 1 h. This concentration of anti-BSA IgG has been previously shown to yield maximum adhesion of human PMNs to ICs and LTB₄ release from these cells (21). All these procedures were done at room temperature.

Isolated murine PMNs (3–5 × 10⁵cells/well, in 0.6 ml RPMI) were placed in 12-well dishes containing coverslips and incubated in a 5% CO₂ incubator at 37°C. At indicated time points, coverslips were washed twice in PBS, fixed with 1.25% glutaraldehyde, stained with Giemsa, and mounted on slides with permount. Adherent PMNs were quantitated as follows: the total number of adherent PMNs in 10 high (×40) powered fields (HPF) was determined and expressed as number of adherent cells per HPF. This number was then converted to the percentage of control binding (set at 100%), which was that observed for wild-type PMNs incubated with IC-coated coverslips for 5 min. Therefore, adherent cells (% control) = (total bound cells per HPF)/(total bound cells per HPF at 5 min for wild-type) × 100%. Spread PMNs were identified as cells with abundant cytoplasm and a flattened nucleus and the percentage of spread was determined as explained above for determining percentage of adherent PMNs. This conversion to percentage of control was necessary since the number of murine PMNs isolated from wild-type and mutant mice varied between individual experiments, although within the same experiment equal numbers of wild-type and mutant PMNs were used. Adherence of mouse PMNs to IC and BSA was very efficient. For example, for wild-type PMNs, after a 5 and 40 min incubation the number of PMNs adherent to IC-coated coverslips (144 mm²) was 1.9 ± 0.4 × 10⁵ and 3.2 ± 0.2 × 10⁵, respectively. This represents ∼44 ± 4 and 82 ± 21% of the total number of PMNs applied to the coverslip. The number of wild-type PMNs adherent to coverslips coated with BSA alone after 5 and 40 min incubations was 2.1 ± 0.5 × 10⁵ and 2.8 ± 0.5 × 10⁵, respectively, and represented 40 ± 4 and 59 ± 9% of the total number of PMNs applied to the coverslip.

PMNs that adhered to IC-coated glass coverslips for 5 or 40 minutes were stained with rhodamine phalloidin (Molecular Probes, Inc., Eugene, OR) according to manufacturer's protocols. In brief, coverslips were washed in PBS, fixed in 3.7% formaldehyde/PBS for 10 min, and extracted for 3–5 min on ice with 0.1% Triton X-100 buffer. Coverslips were then washed in PBS and incubated with 160 nM rhodamine phalloidin in PBS for 20 min at room temperature. Coverslips were mounted with Cytoseal (Stephens Scientific, Riverdale, NJ) and stored at 4°C.

LTB₄ levels in media collected from PMNs adherent to IC-coated surfaces were quantitated using an ELISA assay for murine LTB₄ (PerSeptive Diagnostics, Cambridge, MA). According to manufacturer's data, the ELISA assay has a detection limit of 8.9 pg/ml. Its cross-reactivity with other major arachidonic acid products, e.g., LTC₄, LTD₄, LTE₄, 5-HETE (hydroxy eicosatetraenoic acid), 12-HETE, and 15-HETE, is less than 1%, as is its cross-activity with LTB₄ degradation products 20-carboxy-LTB₄ and 20-hydroxy-LTB₄.

At the indicated time points, medium was collected into ice-cold eppendorf tubes and spun at 750 g (4°C) for 10 min to remove cells. The supernatants were transferred to fresh ice-cold tubes and the ELISA assay was performed in duplicate following the manufacturer's protocol.

Data are expressed as mean ± SEM. Unpaired student's t tests were performed for data and P <0.05 was considered statistically significant.

We injected anti-GBM antibody into Mac-1–deficient mice and their wild-type counterparts and measured glomerular neutrophil accumulation and proteinuria. 1 h after anti-serum injection, neutrophil glomerular accumulation was the same in null and wild-type mice, suggesting that initial PMN recruitment is unaffected by Mac-1 deficiency. In contrast, at 2 h, the peak of PMN accumulation in wild-type animals, there was a significant reduction in the number of glomerular PMNs present in null mice compared with wild-type. The difference in glomerular PMN counts between wild-type and mutant mice persisted at 4 h but by 18 h the counts declined to baseline levels in mice of both genotypes (Fig. 1 A). The peripheral blood PMN counts were similar in wild-type and null mice at each experimental time point (data not shown).

The PMN accumulation in the glomeruli of wild-type mice was associated with significant proteinuria. Proteinuria appeared at 4 h in wild-type animals and peaked at 8 h. In contrast, Mac-1–deficient mice failed to develop proteinuria at any of the time points examined (Fig. 1 B). Proteinuria is transient in this model and declined in the wild-type mice by 18 h. At this time point, proteinuria in the mutant mice was still negligible and was significantly reduced compared to wild-type.

The decreased PMN accumulation in null mice cannot be attributed to reduced binding of anti-GBM antibody in the glomeruli of these mice since a quantitative immuno-fluorescence analysis of anti-GBM deposition in the glomeruli revealed no differences between wild-type and null mice (Table 1). This conclusion is also supported by the fact that neutrophil recruitment at 1 h after anti-GBM injection (Fig. 1 A) was similar in wild-type and mutant mice.

Glomeruli of kidneys were examined by electron microscopy 1 h after anti-GBM injection. In mice of both genotypes, PMNs were observed either unattached in the capillary lumen or making contacts of varying extents with the GBM. The latter was associated with denudation of the fenestrated endothelium (Fig. 2), suggesting that Mac-1 deficiency does not protect against endothelial denudation by PMNs and that such damage is not responsible for the proteinuria in this model. However, the lack of proteinuria in Mac-1–deficient mice observed in Fig. 1 B suggests that mutant neutrophils are incapable of inducing functional GBM damage leading to proteinuria, which is characteristic of wild-type mice.

Divalent cation-dependent Mac-1 adhesion to ICAM-1 and iC3b occurs through the A domain of Mac-1 (1). To determine whether the engagement of Mac-1 with complement or ICAM-1 contributes to the reduction in PMN accumulation in Mac-1–null mice 2 h after anti-GBM antiserum injection, groups of mice genetically deficient in C3 (the central component in complement activation), C4 (the classical pathway of complement activation triggered by antigen-antibody complexes), or ICAM-1 were examined 2 h after injection of anti-GBM antiserum. We observed no significant differences in glomerular PMN counts between ICAM-1–deficient mice and their wild-type mates, or C3- and C4-deficient mice and wild-type mice (Fig. 3). These data confirm that neutrophil accumulation is complement-independent (14, 15) and suggest that known Mac-1 ligands that are most likely to support neutrophil adhesion to the vessel wall do not play a role in this process.

However, proteinuria in the C3-deficient mice was significantly reduced compared to their wild-type counterparts at the 8 h time point (Fig. 3). We have previously shown that Mac-1 is the predominant receptor for C3bi since phagocytosis of complement opsonized particles was completely absent in neutrophils isolated from Mac-1–deficient mice (2). This suggests that interaction of Mac-1 on PMNs with complement fragment ic3b leads to proteinuria.

The next series of in vitro experiments were undertaken to address the following hypothesis: equal glomerular PMN accumulation in wild-type and mutant mice at 1 h is driven by neutrophil Fc receptors engaging immobilized IC, since this model is Fc-dependent, but Mac-1 may be required for subsequent sustained adhesion and elaboration of chemo-kines that promote further PMN recruitment in wild-type animals at the 2 h time point (Fig. 1 A).

We examined adhesion and spreading of wild-type and Mac-1–null PMNs on IC-coated surfaces. To mimic the GBM–anti-GBM interactions in vitro, slides were coated with BSA–anti-BSA IgG complexes. The IC-coated surfaces were incubated with resting peripheral blood PMNs isolated from wild-type and null mice (under static conditions) in the absence of serum, thus eliminating potential effects of complement in the assay. The total number of adherent cells was measured at 5, 12, 25, and 40 min. A similar analysis was conducted on PMNs incubated with BSA-coated surfaces alone to discern effects due to BSA from those that were IC-specific. A time course of total binding of PMNs to IC-coated surfaces revealed that initial adhesion to IC, within 5 and 12 min after addition of PMNs, was similar in mutant and wild-type mice (Fig. 4). At 25 and 40 min, wild-type PMNs continued to accumulate on the IC, leading to a >50% increase over that seen at 12 min. In Mac-1–null PMNs, the number of IC-adherent PMNs remained the same over a similar time course. Incubation times >40 min did not lead to further enhancement in numbers of adherent PMNs in either wild-type or mutant samples (data not shown). The number of wild-type and mutant PMNs adherent to BSA at 5 or 40 min was similar in wild-type and null cells, suggesting that the observed Mac-1–mediated effects are specific for IC-coated surfaces (Fig. 4).

To determine whether spreading of mutant PMNs on IC was compromised, we  determined the percentage of spread cells in the samples generated in Fig. 4. Initial spreading on ICs was the same for wild-type and Mac-1–null cells at 5 min. The percentage of spread cells increased over time in wild-type cells (Fig. 5,A), consistent with the increased numbers of PMNs adherent to the IC at these later time points as seen in Fig. 4. On the other hand, the percentage of mutant PMNs spread on IC declined rapidly over time and was <10% of wild-type levels at 40 min (Fig. 5,A) and <0.03% at 120 min (data not shown). On BSA-coated surfaces, a difference in spreading was observed between wild-type and null cells at 5 min and remained constant up to 40 min (Fig. 5 B). This is consistent with previous studies that indicate a role for Mac-1 in spreading on BSA (1). The equivalent spreading on ICs at 5 min versus the difference in spreading seen on BSA-coated surfaces at this time point again distinguishes Mac-1–mediated effects on these two substrates.

To pursue the underlying cause for the decrease in PMN spreading in mutant neutrophils, we examined actin distribution by staining the IC-adherent PMNs for filamentous actin (F-actin) (Fig. 5 C). At 5 min, F-actin staining was observed in the periphery of both wild-type and mutant neutrophils. By 40 min, wild-type PMNs had extended protrusions, increased cell size (spreading), and had a polarized as well as central punctate F-actin staining. In contrast, Mac-1–null PMNs at this time point were contracted and were much smaller than wild-type PMNs (which indicates a lack of spreading). Several of the rounded Mac-1–null PMNs detached from the dish during processing of the sample for rhodamine phalloidin staining. The few that remained at 40 min were the small percentage of spread cells that had an actin distribution identical to PMNs adherent to IC only for 5 min. We conclude that the actin redistribution associated with sustained adhesion is compromised in Mac-1–deficient PMNs.

To assess whether the decrease in PMN accumulation at 2 h is caused by inhibition of secretion of a chemoattractant by mutant PMNs, we examined the release of an arachidonic acid metabolite LTB₄, since a previous study had shown that β₂ integrins are required for IC-stimulated LTB₄ production (21). In addition, LTB₄ has been linked to the activation and morphologic polarization of PMNs and therefore may be required for sustaining PMN spreading. We measured LTB₄ secreted by wild-type and mutant PMNs bound to IC-coated surfaces by an ELISA assay for murine LTB₄. In wild-type PMNs, LTB₄ was detected in the media at 5 min after neutrophil interaction with IC-coated surfaces and had increased 25-fold by 40 min (Fig. 6). We detected no significant differences in the amount of LTB₄ released by wild-type and mutant PMNs at all time points tested, suggesting that Mac-1 is not required for LTB₄ production.

This study demonstrates an important role for Mac-1 in glomerular PMN accumulation and proteinuria following the formation of ICs in the GBM. This is consistent with conclusions of previous studies using functional blocking antibodies to Mac-1 in acute glomerulonephritis models in rats (22, 23). Importantly, in our work, a study of the kinetics of PMN accumulation revealed that Mac-1 is not required for the initial PMN influx at 1 h but is essential for further PMN accumulation and PMN-dependent proteinuria. Furthermore, since this model is Fc-dependent (14), our results suggest a role for Mac-1 in events downstream of the initial Fc-dependent neutrophil recruitment. These downstream events are probably not related to the ability of Mac-1 to function as an adhesion receptor for obvious ligands: mice deficient in Mac-1 ligands, ICAM-1, or complement had glomerular PMN accumulation comparable to wild-type mice. This is consistent with other studies which show that PMN adhesion to the glomerular endothelium under nonstatic conditions does not require Mac-1 (24). It is not possible to make analogies of the FcγR-Mac-1–mediated PMN recruitment model in glomerular capillaries with the multistep paradigm of selectin-induced PMN rolling, which is most relevant at high shear stress (25, 26), followed by firm integrin–dependent adhesion seen in venules. This is because the hemodynamic forces in the glomerular capillaries are different than in venules. Additionally, these forces are known to be significantly altered during glomerulonephritis, with no measurements of shear stress available under these conditions. Finally, leukocyte rolling has not been directly investigated in glomerular capillaries.

The in vitro study suggests that the decrease in neutrophil accumulation in glomeruli of mutant mice at 2 h is due to an inability of mutant PMNs to sustain Fc-mediated adhesion to the immobilized ICs. In vivo this presumably leads to their detachment into the bloodstream. Since complement C3 is not involved in PMN accumulation in vivo, the in vitro adhesion assays to ICs were performed in the absence of serum, thus eliminating potential effects of complement in the assay. In vitro, we have shown that Mac-1– deficient PMNs can initially accumulate and spread on immobilized ICs at wild-type levels, but have a defect in subsequent adhesion and spreading compared to wild-type PMNs. Therefore, Mac-1 is not essential for initial FcγR attachment to ICs but is required for stabilizing PMN interactions with these complexes. We also observed a lack of F-actin redistribution in Mac-1–deficient PMNs contacting ICs. This indicates a role for Mac-1 in actin regulation after engagement of ICs, which may be necessary for PMN spreading. In fact, Mac-1 has previously been implicated in regulating F-actin assembly; Mac-1 is required for tyrosine phosphorylation of the cytoskeletal protein paxillin after neutrophil adhesion to ICs (27). Phosphorylation of paxillin, known to provide a binding site for the SH3 domains of src family tyrosine kinases, may play a key role in IgG-induced signaling (8).

The defect in sustained spreading of mutant PMNs on ICs (<10% of wild-type at 40 min) had no effect on release of LTB₄ from these cells, suggesting that spreading is not required for this process. Release of LTB₄, a potent neutrophil chemoattractant produced during PMN interaction with ICs, was previously shown to require the β₂ integrins (21). Together, these studies suggest that one of the other β₂ integrins present on neutrophils, LFA-1 or p150,95, may be required for LTB₄ release. The release of other chemoattractants may be affected in the absence of FcγR–Mac-1 interactions, thus accounting for the increase in PMN accumulation in wild-type mice, but not in the mutant mice, at 2 h. The potential chemoattractants that could be responsible for this effect are numerous and were not investigated.

In vivo, a striking absence of PMN-dependent proteinuria in Mac-1–null mice was observed despite the wild-type levels of PMN accumulation 1 h after induction of nephritis. This is even more significant given that, in electron micrographs, the glomeruli of mutant mice at this time point showed similar endothelial denudation as that of wild-types, and PMNs were in intimate contact with the GBMs. This would suggest that Mac-1–null PMNs are able to activate IC-stimulated functions that lead to endothelial injury. Indeed, IC-triggered superoxide production (21, 28) and rise in intracytoplasmic calcium in PMNs (21) do not require the β₂ integrins. However, events leading to GBM damage and subsequent increased permeability to albumin are lacking in mutant PMNs. Our results in the Mac-1–deficient mice are strikingly similar to those obtained in beige mice subjected to the same model of anti-GBM nephritis (17). Beige mice, which lack cathepsin G and elastase in their azurophilic granules, exhibit PMN accumulation and glomerular endothelial denudation, as assessed by electron microscopy, but develop no proteinuria. Similarly, in Mac-1–deficient mice, release of azurophilic granules containing elastase and cathepsin G may not occur in the mutant PMNs and may account for the absence of proteinuria in these animals. We also demonstrate that complement C3– deficient mice have decreased proteinuria despite accumulating wild-type levels of glomerular PMNs. Importantly, PMN phagocytosis of complement C3b–opsonized zymosan leads to the release of elastase which is Mac-1–dependent (29), and we have previously shown that Mac-1–deficient PMNs are unable to phagocytose complement-opsonized particles (2). Therefore, the interaction of Mac-1 on PMNs with complement C3 is probably necessary for the release of azurophilic granules in our anti-GBM nephritis model and explains the lack of proteinuria in both Mac-1–deficient and complement C3–deficient mice. In addition, another group has shown that PMNs extravasated into the peritoneum of Mac-1–deficient mice after thioglycollate-induced peritonitis have increased β-glucuronidase content, suggesting a deficit in release of azurophilic granules by activated Mac-1–null PMNs (30). Although we propose that an interaction of Mac-1 with complement is responsible for proteinuria in this model, a separate role for C3 in generation of a membrane attack complex leading to proteinuria cannot be ruled out and was not explored further in this work.

These results suggest the following model for the role of Mac-1 in IC-triggered neutrophil accumulation in this form of acute nephritis (Fig. 7). The initial attachment of neutrophils to ICs in the glomerular capillary wall occurs via Fc interactions with FcγR on the surface of neutrophils, and does not require Mac-1. Thus the early neutrophil accumulation is equal in both mutant and wild-type mice. However, Mac-1, through interactions with, or signals from, FcγR, is required for the F-actin reorganization necessary for sustained neutrophil spreading on ICs as shown in our in vitro studies. The functional interaction may be required for elaboration of mediators that signal further neutrophil influx, accounting for the increase in neutrophil accumulation in wild-type mice at 2 h, which does not occur in mutant mice. We have not identified the specific chemoattractants that are effected by Mac-1 deficiency. Production of LTB₄, a powerful neutrophil chemoattractant, is not affected by Mac-1 deficiency or PMN spreading. Interaction of Mac-1 on PMNs with complement iC3b deposited on the vessel wall leads to the release of azurophilic granules, leading to protease-induced damage to the GBM and proteinuria, which is abrogated by Mac-1 or C3 deficiency. The glomerular PMN accumulation in this model is transient. There is evidence to suggest that the PMNs detach from the vessel wall, return to the bloodstream (16), and are probably cleared in the liver or spleen.

In conclusion, this study is the first to suggest a role for Mac-1 in FcγR-dependent neutrophil responses in vivo, thus providing pathophysiological relevance for Mac-1 interactions with FcγR. It also demonstrates a crucial role for Mac-1 in complement-dependent increased permeability, which leads to proteinuria.

We would like to thank Chris Ridolfi and George Stavrakis (Department of Pathology, Brigham and Women's Hospital) for excellent assistance in electron microsopy. We thank Ed Marcus for developing a computer program for quantitative analysis of immunofluorescence-associated glomeruli.