Human C-reactive protein (CRP) binds apoptotic cells and alters blood clearance of injected chromatin in mice. To test whether CRP participates in the pathogenesis of systemic lupus erythematosus (SLE), we examined disease development in lupus-prone (NZB × NZW)F1 (NZB/NZW) mice expressing a human CRP transgene (hCRPtg/BW).
Mortality was monitored, proteinuria was determined by dipstick, and serum levels of human CRP and anti–double-stranded DNA (anti-dsDNA) were determined by enzyme-linked immunosorbent assay in NZB/NZW and hCRPtg/BW mice. Thin sections of kidneys were analyzed by immunofluorescence microscopy to compare deposition of IgG, IgM, C3, and human CRP, and electron microscopy was used to reveal differences in ultrastructure. In situ hybridization was performed to detect human CRP messenger RNA expression.
The hCRPtg/BW mice had less proteinuria and longer survival than NZB/NZW mice. They also had lower IgM and higher IgG anti-dsDNA titers than NZB/NZW mice, although the differences were transient and small. In hCRPtg/BW mice, accumulation of IgM and IgG in the renal glomeruli was delayed, reduced, and more mesangial than in NZB/NZW mice, while end-stage accumulation of IgG, IgM, and C3 in the renal cortex was prevented. There was less glomerular podocyte fusion, basement membrane thickening, mesangial cell proliferation, and occlusion of capillary lumens in hCRPtg/BW mice, but dense deposits in the mesangium were increased. With disease progression in hCRPtg/BW mice, there was little rise in the plasma CRP level, but CRP in the kidneys became increasingly apparent due to local, disease-independent, age-related expression of the transgene.
In hCRPtg/BW mice, CRP protects against SLE by increasing blood and mesangial clearance of immune complexes and by preventing their accumulation in the renal cortex.
Systemic lupus erythematosus (SLE) is an autoimmune disease that can damage the skin, joints, lungs, blood, blood vessels, heart, liver, brain, and especially, the kidneys (1). Even with contemporary medical care, renal involvement is a major cause of morbidity and mortality in SLE patients (2). Much of our current knowledge of the pathogenesis of SLE is based on inference from murine models, the (NZB × NZW)F1 (NZB/NZW) hybrid strain being the oldest and best characterized of these (3). The disease process in NZB/NZW mice accurately reflects the process of SLE in humans. Like humans with SLE, NZB/NZW mice display a pathognomonic antinuclear autoantibody response that includes anti–double-stranded DNA (anti-dsDNA), and they spontaneously develop a fatal glomerulonephritis (4). Also similar to the characteristics in humans, the disease is most frequent in female mice, susceptibility is influenced by the major histocompatibility complex (MHC; mouse H-2), and non–H-2 loci are linked to the organ-specific symptoms (4, 5).
In most diseases, the blood level of C-reactive protein (CRP) accurately reflects the degree of underlying inflammation and tissue necrosis (6). In SLE patients, however, there is often little or no increase in blood levels of CRP, even during disease flares (7, 8). Indeed, when CRP elevation is detected in an SLE patient, it is indicative of intercurrent infection rather than active lupus (8, 9). There is additional clinical and experimental evidence that together suggests a possible participatory role of CRP in the pathogenesis of lupus. For example, CRP binds to nuclear autoantigens of relevance in SLE, such as chromatin, histones H1, H2A, and H2B, and the 70-kd protein associated with U1 small nuclear RNP (10–13). Also, ligand-complexed human CRP can interact with the complement system (14) and Fcγ receptors (FcγR) in mice as well as humans (15–17). Furthermore, in SLE, and probably in other diseases, CRP can potentially mediate an immunoregulatory effect by binding and opsonizing apoptotic cells (18). Importantly in mice, where CRP is not an acute-phase protein and is only a trace serum component (19), injected human CRP alters the clearance of administered chromatin (20) and offers transient protection against murine lupus in the NZB/NZW strain (21).
In the present study, we show that the development of autoimmune glomerulonephritis and death are delayed in NZB/NZW mice carrying a human CRP transgene (hCRPtg). Our data associate this protection with the ability of CRP to limit renal damage, an effect it achieves by preventing glomerular and extraglomerular deposition of immune complexes and/or enhancing phagocytosis of immune complexes by mesangial cells.
MATERIALS AND METHODS
Our C57BL/6 congenic (H-2b/b) hCRPtg mouse strain has been described in detail elsewhere (22). These animals carry a 31-kb Cla I fragment of human genomic DNA comprising of the CRP gene, 17 kb of the 5′-flanking sequence, and 11.3 kb of the 3′-flanking sequence, and they express high levels of human CRP in response to injection of lipopolysaccharide (LPS) (23) or bacterial infection (22). We crossed C57BL/6 hCRPtg mice with NZW mice (H-2z/z) obtained from The Jackson Laboratory (Bar Harbor, ME) to generate hybrid pups (H-2b/z). Pups were screened for the presence of the human CRP transgene by polymerase chain reaction (PCR) (24), and hCRPtg F1 animals were backcrossed with NZW mice. PCR amplification of an H-2z–linked polymorphism in the promoter of the tumor necrosis factor α gene (25) allowed us to identify F2 progeny that were H-2z/z, and hCRPtg/H-2z/z mice were then backcrossed to the NZW strain to fix the H-2 complex. Transgenic mice were backcrossed at least 9 times prior to mating hCRPtg/NZW congenic males with NZB females (obtained from The Jackson Laboratory) to produce lupus-prone hybrids.
Disease development in female hCRPtg/BW and female NZB/NZW littermates was then monitored. All mice eventually developed detectable levels of anti-dsDNA autoantibodies and exhibited severe anasarca and concomitant proteinuria, indicating that backcrossing transferred the NZW loci responsible for lupus. Mice were housed in groups of 4, given food and water ad libitum, and maintained according to protocols approved by the Animal Resources Program of the The University of Alabama at Birmingham.
Monitoring mortality, proteinuria, and serum proteins.
The course of disease in hCRPtg/BW mice was compared with that in age-matched NZB/NZW mice. Urine and blood were collected at biweekly intervals. Proteinuria was assessed with Albustix test strips (Bayer, Elkhart, IN) and scored as trace (<0.3 mg/ml), 1+ (0.3–0.9 mg/ml), 2+ (1–2.9 mg/ml), 3+ (3–20 mg/ml), or 4+ (>20 mg/ml), according to the manufacturer's scoring system. Blood was diluted immediately in ice-cold GVB–EDTA buffer (0.1% gelatin, 141 mM NaCl, 1.8 mM sodium barbital, 3.1 mM barbituric acid, 10 mM EDTA, pH 7.4), centrifuged at 1,000g, and the plasma was collected and frozen at −70°C until used. Deaths of mice were noted 4 times daily.
Plasma levels of human CRP were measured by enzyme-linked immunosorbent assay (ELISA) (22). The assay has a lower limit of detection of 20 ng of human CRP/ml and does not detect mouse CRP. To induce acute-phase expression of human CRP, mice received an intraperitoneal injection of 25 μg of LPS (Sigma-Aldrich, St. Louis, MO).
Anti-dsDNA antibody titers were measured by ELISA (26) using proteinase/ribonuclease-treated and sonicated calf thymus DNA (Sigma-Aldrich) as immobilized antigen and horseradish peroxidase–conjugated goat anti-mouse immunoglobulins specific for IgM, IgG, IgG1, IgG2a, IgG2b, and IgG3 (Southern Biotechnology, Birmingham, AL) as reporters. Plasma was diluted serially in GVB–EDTA, and ABTS (Sigma-Aldrich) was used as chromogenic substrate. The anti-dsDNA end-point titer for each sample corresponds to the highest dilution of plasma that resulted in an absorbance at 405 nm that was more than twice the level achieved using undiluted plasma from a 6-week-old C57BL/6 mouse. As positive controls for IgG and IgM anti-dsDNA, we used DNA6 and 25-12 monoclonal antibodies, respectively (26).
Kidneys harvested from randomly selected age-matched hCRPtg/BW and NZB/NZW mice were embedded in OCT compound (Sakura, Torrance, CA) and stored (−70°C) until they were processed for light and immunofluorescence microscopy, or they were fixed in glutaraldehyde and processed for electron microscopy. For light and immunofluorescence microscopy, serial sections (4 μm thick) were cut on a cryostat microtome at −20°C, mounted on glass slides, air-dried, acetone-fixed, and stained with hematoxylin and eosin to allow comparison of gross pathology.
To detect immunoglobulins and C3, sections were blocked with normal horse serum and stained with tetramethylrhodamine isothiocyanate (TRITC)–labeled goat F(ab′)2 anti-mouse IgG or anti-mouse IgM (μ chain) and fluorescein isothiocyanate (FITC)–labeled goat F(ab′)2 anti-mouse C3 (ICN Pharmaceuticals, Aurora, OH). To detect human CRP, we used biotinylated HD2-4 monoclonal antibody (27) in conjunction with R-phycoerythrin (PE)–streptavidin (Southern Biotechnology). Sections were washed and mounted in Fluormount G (Southern Biotechnology) and viewed with a Leica DMRB fluorescence microscope (Bannockburn, IL) equipped with appropriate filter cubes (Chromatechnology, Battleboro, VT). Images were acquired with a Hamamatsu Photonic System digital color camera (Hamamatsu, Bridgewater, NJ) and processed with Adobe Photo Shop (Adobe Systems, San Jose, CA) and IP LAB Spectrum software (Signal Analytics Software, Vienna, VA). Images of TRITC and/or R-PE fluorescence (red) superimposed on images of FITC fluorescence (green) produced a yellow-orange color when the 2 fluorochromes coincided.
Glutaraldehyde-fixed tissue was processed for electron microscopy according to a standard procedure that included postfixation in osmium tetroxide, dehydration with acetone, and embedding in Spurr's resin. Ultrathin sections were cut on an LKB Ultratome III (Stockholm, Sweden), poststained with aqueous uranyl acetate and Reynold's lead citrate, and examined with a Phillips model TEM 301 transmission electron microscope at an accelerating voltage of 60 kV.
In situ hybridization.
Plasmids (pcDNA3.1; Promega, Madison, WI) carrying CRP complementary DNA inserts derived from clone HLCRP-23 (28) were used as templates for the synthesis of digoxigenin-labeled riboprobes. One plasmid carrying a 1.5-kb fragment that spans the entire CRP gene, including 0.76 kb of 3′-untranslated sequence, was used to make the sense probe; and a second plasmid carrying the protein coding sequence only (0.74 kb) was used to make the antisense probe. Plasmids were linearized by digestion with Hind III (Promega), and in vitro transcription was performed using an RNA transcription kit containing T7 RNA polymerase (Promega).
Kidneys were harvested from mice, frozen, and sectioned as described above for light microscopy. Serial sections were mounted on glass slides and fixed in 3% paraformaldehyde, pH 7.4. These were hybridized for 16 hours at 50°C with the riboprobes diluted 1:10 in hybridization buffer (Fisher Scientific, Fair Lawn, NJ). After hybridization, 3 washes were performed at 50°C. To reveal the location of RNA/RNA duplexes, slides were incubated with alkaline phosphatase–conjugated anti-digoxigenin F(ab′)2 (Boehringer Manheim, Indianapolis, IN), followed by reaction with BCIP/nitroblue tetrazolium substrate (Sigma-Aldrich).
Survival curves were generated using 25 hCRPtg/BW mice and 25 NZB/NZW mice, and kidneys were obtained from 2 contemporaneous but separate groups of 18 hCRPtg/BW mice and 20 NZB/NZW mice housed in the same facility. The deaths of the mice had a normal distribution, so an unpaired Student's t-test was used to evaluate differences in lifespan. The chi-square test with Fisher's correction was used to evaluate differences in the incidence of severe (≥2+) proteinuria. Anti-dsDNA titers were log-transformed to achieve normality and then compared using Student's t-tests; in some cases, this analysis could not be applied due to zero variance. P values less than 0.05 were considered significant in all analyses. Data are reported as the mean ± SEM.
On average, hCRPtg/BW mice lived 8 weeks longer than NZB/NZW mice (mean ± SEM 290 ± 12 days versus 235 ± 9 days; P = 0.004, by Student's t-test) (Figure 1A), and 85% of the hCRPtg/BW mice were still alive when 50% of the NZB/NZW mice had already succumbed to lupus (33 weeks). The onset of severe proteinuria was also delayed by at least 4 weeks in hCRPtg/BW mice compared with NZB/NZW mice, and the incidence of severe proteinuria in hCRPtg/BW mice was 2–5 times lower (Figure 1B). The difference in the incidence of severe proteinuria between NZB/NZW and hCRPtg/BW mice achieved statistical significance at 30 weeks (χ2 = 5.788, P = 0.016).
To decipher the basis for CRP protection, plasma was collected from mice and assayed for the presence of circulating anti-dsDNA antibodies. There were marked age-related variations in the titers of the different classes and subclasses of anti-dsDNA in NZB/NZW mice (Figure 2). This finding confirms those of previous investigators who studied the characteristics of anti-dsDNA responses in NZB/NZW mice (29). In general, in young NZB/NZW mice, the predominant anti-dsDNA class was IgM. As the animals aged, there was increased production of IgG. Anti-dsDNA of all isotypes peaked at ≥30 weeks.
The anti-dsDNA response of the hCRPtg/BW mice was equally variable and also peaked at ≥30 weeks, but had a slightly different tempo (Figure 2). For example, in NZB/NZW mice, there was early and transient rise in IgM anti-dsDNA, but this did not occur in hCRPtg/BW mice. Consequently, in 16- and 20-week-old hCRPtg/BW mice the IgM anti-dsDNA titer was 6–7-fold lower than that in age-matched NZB/NZW mice (P < 0.05, by Student's t-test). In contrast to IgM, the IgG anti-dsDNA titer was 3–5-fold higher in 18- and 24-week-old hCRPtg/BW mice than in age-matched NZB/NZW mice (P < 0.05, by Student's t-test). The elevation in IgG antibody was a consequence of the increased production of IgG1, IgG2a, and IgG2b subclasses.
Kidneys harvested at different time points and stained with hematoxylin and eosin revealed progression of glomerular disease typical of murine lupus in both the hCRPtg/BW and the NZB/NZW groups, with mesangial thickening and hypercellularity evolving into crescent formation and end-stage sclerosis (data not shown). Kidneys were then analyzed more extensively by immunofluorescence (Figures 3 and 4) and electron microscopy (Figure 5).
In the kidneys of healthy 5-week-old mice, no IgG or IgM was detected in either strain, but C3 was detected in both (Figures 3 and 4). Importantly, in young mice, immunoreactive C3 was detected mainly in the epithelial layer, with the strongest signal in the Bowman's capsule, most likely because of local production of C3 rather than the accumulation of blood-borne C3. By 26 weeks, heavy accumulations of immunoglobulin and C3 were seen inside the glomeruli of NZB/NZW mice (Figures 3 and 4), with a fluorescence pattern suggesting the presence of immune complexes in both the capillary loops and the mesangium. In comparison, the glomeruli of age-matched hCRPtg/BW mice had less accumulation of IgG and IgM, with a more mesangial pattern but a similar distribution of C3. By 34 weeks, accumulation of immunoglobulins and C3 in the kidneys of NZB/NZW mice had expanded beyond the glomeruli to the cortex (Figures 3 and 4). In contrast, in age-matched hCRPtg/BW mice, there was no accumulation of IgM, IgG, or C3 in the cortex. Again, there was no apparent difference in the amount of C3 present within glomeruli in the two strains.
Electron microscopy (Figure 5) confirmed the presence of dense deposits in the glomeruli of both NZB/NZW and hCRPtg/BW mice in a pattern consistent with the presence of immune complexes. In the glomeruli of NZB/NZW mice, there was extensive fusion of podocyte foot processes, thickening of the glomerular basement membrane, and proliferation of mesangial cells with invasion and occlusion of the capillary lumens (Figure 5, top panels). In contrast, in the glomeruli of hCRPtg/BW mice, neither obvious foot process fusion nor any obvious basement membrane thickening was seen, and the capillary lumens largely remained more open (Figure 5; bottom panels). Importantly, in accordance with the increased mesangial pattern of fluorescence seen when hCRPtg/BW glomeruli were stained with anti-Ig (Figures 3 and 4), the number and size of dense deposits seen in the mesangium of hCRPtg/BW mice (Figure 5F) were increased compared with that in NZB/NZW mice.
As in female hCRPtg mice with non-autoimmune genetic backgrounds (22, 24), blood levels of CRP in the hCRPtg/BW mice were low. However, relative to baseline values determined at age 6 weeks, there was a 50–100-fold increase observed with aging (Figure 6). Since hCRPtg/BW mice did not secrete human CRP into their urine (as determined by ELISA; data not shown) and the transgene maintained full acute-phase inducibility, as judged by LPS responsiveness (Figure 6), we examined whether human CRP might be present in the kidneys. Human CRP–specific immunofluorescence was detected in the kidneys of older hCRPtg/BW mice (Figure 7). Parallel analysis of kidneys obtained from older, non–lupus-prone hCRPtg/C57BL/6 mice (22) also revealed human CRP in the kidneys (data not shown), suggesting that the appearance of CRP in the kidneys of hCRPtg/BW mice is age-rather than disease-related.
In situ hybridization experiments showed that in the kidneys of hCRPtg/BW mice, human CRP messenger RNA was produced locally and was up-regulated with age (Figure 8), and thus appears there coincident with the progression of glomerular disease. In concordance with the staining pattern seen with immunofluorescence (Figure 7), the staining pattern seen with in situ hybridization (Figure 8) suggests that cells in the cortical tubular area are the main source of the renal human CRP.
We show here that endogenously expressed human CRP is protective in a mouse model of SLE, that is, the spontaneous development of lethal glomerulonephritis was delayed and the lifespan was extended in NZB/NZW mice carrying a human CRP transgene. This finding confirms and extends earlier observations made by Du Clos et al (21), who showed that 26-week-old NZB/NZW mice injected repeatedly with chromatin-coated latex beads pretreated with human CRP lived longer than age-matched animals injected with untreated chromatin-coated beads. In those experiments, CRP did not alter the IgM and IgG anti-DNA response, the beneficial effect of CRP was short-lived, and protection was not observed in 19-week-old mice. The transient nature of the CRP protective effect was attributed to development of anti-CRP responses in mice given human CRP (21). In comparison, we found that in our transgenic NZB/NZW mice, expression of human CRP was associated with slight and transient, but significant, lowering of IgM anti-dsDNA autoantibody titers, and slight and transient elevation of IgG1, IgG2a, and IgG2b anti-dsDNA titers. Precisely how human CRP changes the tempo and repertoire of the anti-dsDNA response in NZB/NZW mice is not known with certainty, but the data from both our study and that by Du Clos et al are consistent with the proposal by Du Clos and colleagues (21) that this is a consequence of altered clearance and/or processing of autoantigens by circulating CRP. Such a process could hasten the isotype switch and might also explain the apparent discordance in the IgM and IgG results. Presumably, renal CRP could have similar contributions.
As shown in Figure 6, although CRP was elevated up to 100-fold during disease development, its absolute level remained low compared with that in the LPS-induced acute phase. We found also that human CRP was locally produced in the kidneys of hCRPtg/BW mice, and like the blood-borne (hepatically synthesized) protein, its expression there increased with age. The finding that CRP is expressed in the kidney is not surprising, since in a previous study of mice carrying the same CRP transgene, Klein et al (30) detected (by nested PCR) the expression of human CRP in the kidneys. The levels they detected, however, were scant, and they failed then to recognize that renal expression of the CRP transgene increased with age; this presumably occurred because only the kidneys of young adult mice were examined (30). Recent clinical evidence suggests that our observation of CRP expression in the kidneys is an accurate reflection of the situation found in SLE in humans. For example, in SLE patients, CRP is often not elevated in the blood even during disease flares (5–8) and is rarely lost into the urine (31), yet it is frequently found in the kidneys of patients with glomerular disease (32). These clinical observations together with our observations of hCRPtg/BW mice suggest that locally expressed CRP, and perhaps its regulation, might be more important in the pathogenesis of SLE than is blood-borne CRP and its hepatic expression.
If, as we suspect, the presence of human CRP in the kidneys of hCRPtg/BW mice is the key to understanding its protective effect, then exactly how does CRP retard the development of lupus nephritis? The ultimate fate of CRP produced locally in the kidney remains elusive. We could not detect (by ELISA) the protein in the urine, and it is still not known if the presence of CRP in the cortical tubular area has any impact on renal physiology per se. However, we do know that in hCRPtg/BW mice, there is a virtual absence of immune complex deposition in the renal cortex and an increased accumulation of dense deposits in the glomerular mesangium, and we think this is informative. As we demonstrated here for C3, other complement proteins are also expressed by the renal epithelium in murine lupus nephritis (33–35). CRP is known to activate complement via the classical pathway (14), and the classical pathway plays an important physiologic role by preventing the formation of immune complexes (36). Thus, one possibility is that by activating the classical pathway of complement, CRP inhibits the formation of immune complexes in the cortex, thereby reducing the local inflammatory response and delaying the onset of glomerulonephritis. This scenario requires that immune complex formation occur in the cortex, but what if immune complexes are preformed distally and then deposited locally?
In contrast to its effect on the classical pathway, CRP inhibits the alternative pathway (37), and the alternative pathway affects the solubilization of large insoluble preformed immune complexes (38). Recent evidence shows that activation of the alternative pathway has a prominent detrimental effect in the lupus kidney (39). In a second scenario, by inhibiting the alternative pathway, CRP could still reduce local inflammation and support the improved clearance of deposited immune complexes by resident phagocytes.
In a third scenario, since CRP is known to bind FcγR (15–17) and since occupation of FcγR is known to modulate the complement-mediated inflammatory response in the kidney (40), the protective effect might involve the interaction of CRP with cells bearing FcγR. Importantly, mesangial cells do express FcγR and have a phagocytic clearance function (ref. 41 and for review, see ref. 42). Perhaps CRP enhancement of the mesangial cell clearance of immune complexes is responsible for the protection observed and the increase in dense deposits in the mesangium of hCRPtg/BW mice. It is important to point out that these 3 scenarios are not mutually exclusive, and that in all instances, the ability of blood CRP to clear apoptotic and necrotic cells (18, 43) would further enhance the antiinflammatory effect.
SAP is very closely related to CRP and supports many of the same effector functions (44). In a recent report (45), Bickerstaff et al showed that in an otherwise healthy strain of mice, targeted deletion of the SAP gene led to the development of antinuclear autoimmunity and severe glomerulonephritis. Like CRP, SAP binds and can solubilize chromatin (46), and it binds to apoptotic and necrotic cells (47, 48). Thus, the SLE-like phenotype of SAP-deficient mice was linked to their impaired handling and degradation of chromatin, which allowed for increased anti-DNA responses (45). These results from SAP-deficient mice and our findings from hCRPtg/BW mice strongly support the notion that CRP and SAP provide an important physiologic function. CRP-deficient NZB/NZW mice are not available to confirm this, but presumably, mouse CRP, which is also expressed in the kidney (30), is protective as well. Although the additional local expression of human CRP in hCRPtg/BW mice does not prevent glomerular injury, it clearly delays disease development above and beyond that achieved by endogenous mouse CRP and SAP.
Recognizing that generalizing the results of mouse studies to humans must be done with caution, our data suggest that the beneficial role of CRP likely is 2-fold in SLE and in other diseases in which the kidney is a major target organ. On the one hand, local and systemic production of CRP delays the onset of renal involvement, and on the other hand, circulating CRP has significant antimicrobial properties (for review, see ref. 49) and reduces the likelihood of intercurrent infection.
The authors thank Nuzhat Iqbal (Department of Medicine, The University of Alabama at Birmingham) for assistance and Dr. A. Agrawal (Department of Pharmacology, East Tennessee State University, Johnson City, TN) for helpful advice.