Autoantibodies against C-Reactive Protein Influence Complement Activation and Clinical Course in Lupus Nephritis

Abstract

Autoantibodies against the major acute-phase reactant C-reactive protein (CRP) are frequently found in patients with lupus nephritis. Further defining the autoimmune epitopes on CRP may not only improve patient stratification but also, hint at mechanisms of CRP action. Herein, we show that amino acids 35–47 constitute the major epitope recognized by anti-CRP autoantibodies in patients with lupus nephritis. Notably, the presence of autoantibodies against amino acids 35–47 associated with more severe renal damage and predicted worse outcome. This epitope is exposed on CRP only after irreversible structure changes, yielding a conformationally altered form termed modified or monomeric CRP (mCRP). ELISA and surface plasmon resonance assays showed that amino acids 35–47 mediate the interaction of mCRP with complement factor H, an inhibitor of alternative pathway activation, and this interaction greatly enhanced the in vitro cofactor activity of complement factor H. In contrast, autoantibodies against amino acids 35–47 inhibited these actions of mCRP. Our results thus provide evidence for the in vivo generation of mCRP in a human disease and suggest that mCRP actively controls the pathogenesis of lupus nephritis by regulating complement activation. Therefore, amino acids 35–47 constitute a functional autoimmune epitope on CRP that can be targeted therapeutically and diagnostically.

SLE is an autoimmune disease characterized by massive production of autoantibodies, excessive activation of complement, and defective clearance of cell debris and immune complexes, leading to extensive tissue damage in multiple organs.¹ C-reactive protein (CRP) is a serum marker of inflammation and a putative soluble pattern recognition receptor that contributes to efficient removal of dead cells or invading pathogens by activating complement.^2,3 It is, therefore, interesting to note the blunted CRP response in active SLE⁴ and the presence of anti-CRP autoantibodies in a subset of patients.^5–8 These might point to the potential of CRP and its autoantibodies to be therapeutic targets and/or diagnostic indices.

However, anti-CRP autoantibodies add little to the current diagnosis,⁹ despite their prevalence in patients with SLE.^6–8 This would instead argue that the generation of CRP autoantibodies may merely be an epiphenomenon resulting from systemic inflammation. Alternatively, the actual pathogenetic and diagnostic significance could be obscured by the polyclonal nature of CRP autoantibodies, which comprise both disease-specific and passive clones. Identification of the immune epitopes on CRP and their respective disease associations is, therefore, mandatory to address this issue.

In addition, accumulating evidence suggests that CRP autoantibodies do not react with the native conformation but instead, recognize a conformationally altered variant (i.e., monomeric or modified C-reactive protein [mCRP]).^6–8 Whereas CRP is secreted by hepatocytes as a cyclic pentamer,^2,3 it has been proposed that pentameric CRP is induced to undergo irreversible conformation changes at sites of tissue injury, giving rise to mCRP with greatly enhanced bioactivities.^10,11 Although in vivo support for this proposal has been lacking, the discovery of autoantibodies against mCRP in SLE^6–8 strongly argues for the generation and actions of mCRP in the disease. However, modest and reversible conformation changes in CRP could also occur as implicated by the structural features¹² and exemplified experimentally.¹³ Those changes may expose cryptic epitopes that are shared by mCRP but nevertheless, are not the result of its generation. This uncertainty could also be addressed by the identification of autoimmune epitopes that are unique to mCRP.

The majority of patients with SLE are diagnosed with renal involvement (i.e., lupus nephritis [LN]).¹ Herein, we identify amino acids (a.a.) 35–47, a sequence exposed only in mCRP, as the predominant epitope recognized by CRP autoantibodies in patients with LN. Patients with anti-a.a. 35–47 autoantibodies showed more severe renal damages and worse outcomes. Complement factor H (CFH) is an inhibitor of complement alternative pathway activation.^14,15 Genetic polymorphism of CFH affects the susceptibility of SLE,¹⁶ and its deficiency accelerates the development of LN in mice.¹⁷ We show that mCRP binds CFH and enhances its cofactor activity via a.a. 35–47, whereas autoantibodies against this epitope inhibit these actions. Because complement overactivation is a key driver of renal damage,¹ our findings suggest that anti-a.a. 35–47 autoantibodies may adversely affect LN through interfering with the function of mCRP as a complement regulator.

Results

a.a. 35–47 Are the Major Autoimmune Epitope on CRP in LN

We first screened for autoantibodies against immobilized mCRP in sera of 80 patients with LN (Table 1), and 24 of them were identified as mCRP autoantibody positive (Figure 1A and Table 2). Six strong positive samples were then used to conduct epitope mapping against a panel of synthesized CRP peptides (Figure 1B); a.a. 35–47 and 199–206 emerged as the predominant epitopes recognized by autoantibodies. Interestingly, only a.a. 35–47 efficiently inhibited autoantibody binding to immobilized mCRP (Figure 1C), although both epitopes are prevalent in patients with LN (Figure 1, D and E). The lack of inhibitory effects of a.a. 199–206 might be due to their low affinity to autoantibodies. We thus concluded that a.a. 35–47 were the major autoimmune epitope on CRP.

Table 1.

Clinical characteristics of patients with LN

Characteristic	Value
Clinical evaluation
No. of patients	80
Sex, men/women	7/73
Age, mean±SD, yr	30.0±12.1
Follow-up time median (range), mo	48 (34–58)
Fever (noninfectious), no. (%)	27 (33.8)
Malar rash, no. (%)	33 (41.3)
Photosensitivity, no. (%)	16 (20.0)
Oral ulcer, no. (%)	72 (90.0)
Alopecia, no. (%)	16 (20.0)
Arthralgia, no. (%)	32 (40.0)
Serositis, no. (%)	9 (11.3)
Neurologic disorder, no. (%)	4 (5.0)
Anemia, no. (%)	43 (53.8)
ARF, no. (%)	13 (16.3)
Nephrotic syndrome, no. (%)	50 (71.4)
Laboratory assessment
No. of patients	80
Leukocytopenia, no. (%)	17 (21.3)
Anemia, no. (%)	49 (61.3)
Thrombocytopenia, no. (%)	16 (20.0)
Hematuria, no. (%)	64 (80.0)
Leukocyturia (noninfection), no. (%)	43 (53.8)
Hemoglobin, mean±SD, g/L	104.8±22.2
Urine protein (mean±SD), g/24 h	3.8 (1.7–6.9)
Serum creatinine median (range), μmol/L	70.8 (54.6–137.8)
C3 (mean±SD), g/L	0.4 (0.3–0.6)
Antinuclear antibody (+), no. (%)	77 (96.3)
Antidouble-stranded DNA antibody (+), no. (%)	63 (78.8)
Anti-SSA antibody (+), no. (%)	44 (55.0)
Anti-SSB antibody (+), no. (%)	8 (10.0)
Anti-Smith antibody(Sm) (+), no. (%)	18 (22.5)
Renal histopathology indices
No. of patients	80
Activity indices score	6.5±2.8
Chronicity indices score	1 (1–2)

SSA, Sjögren's syndrome A antigen; SSB, Sjögren's syndrome B antigen.

Figure 1.

The a.a. 35–47 are the major autoimmune epitope on CRP in LN. (A) Urea-denatured mCRP was immobilized onto microtiter wells and tested for autoantibody binding of sera from 24 healthy controls and 80 patients with LN. Twenty-four patient sera were assigned as mCRP autoantibody positive using a cutoff given by the mean OD value obtained with the control sera plus two times the SD. (B) Synthesized CRP peptides were immobilized and tested for autoantibody binding of six sera from patients with LN with strong signals of mCRP autoantibody (patients 1–6). Apparent binding to a.a. 35–47 and 199–206 was consistently observed. The binding to other peptides, however, were comparable with that of CRP signal peptide, which is cleaved before secretion and considered as the background control. The cutoff was set as the interquartile mean of the OD values obtained with all peptides plus two times the SD. (C) The serum of patient 1 was mixed with the indicated peptides and added to immobilized mCRP; a.a. 35–47 were the only peptide that significantly reduced the binding of mCRP autoantibodies. Similar results were obtained with the serum of the other patient (not shown). Sera of 24 patients with LN with mCRP autoantibodies and 24 healthy controls were tested for autoantibody binding to immobilized a.a. (D) 35–47 or (E) 199–206. Binding was normalized to that of patient 1. Autoantibodies against a.a. 35–47 and 199–206 were detected in 11 and 17 patients, respectively. (F) a.a. 35–47 mutants with a single residue replaced with alanine were tested for their inhibitory effects on the binding of autoantibodies in the serum of patient 1 to immobilized mCRP. Leu37, Phe39, Tyr40, and Leu43 were found to be critical residues. Similar results were obtained with the serum of the other patient (not shown). (G) The crystal structure of pentameric CRP (1B09)¹²; a.a. 35–47 are shown in blue, with key residues colored red. Most parts of a.a. 35–47, including the three critical residues, were buried inside the native structure. In B, C, and F, each data point or bar represented the mean of three technical replicates. Dashed lines represent cutoff values.

Table 2.

Comparisons of clinical data between patients with LN with gross mCRP autoantibodies and patients with LN without gross mCRP autoantibodies

Clinical Feature	Patients with LN, n=80
	Anti-mCRP Positive	Anti-mCRP Negative	P Value
No. of patients	24	56
Age mean±SD, yr	30.42±13.68	31.86±11.57	0.46
Sex, men/women	11/13	6/50	0.60
No. of leukocyturia (%)	14 (58.3)	29 (51.8)	0.77
No. of hematuria (%)	19 (79.1)	45 (80.3)	0.80
No. of ARF (%)	5 (20.8)	8 (14.3)	0.42
Urine protein median (interquartile range), g/24 h	2.72 (0.62–4.88)	3.96 (2.15–7.66)	0.03
Serum creatinine median (interquartile range), μmol/L	69 (52.28–153)	71.7 (54.6–125.25)	0.59
Creatinine clearance rate mean±SD, ml/min	76.30±43.77	81.84±45.54	0.67
C3 median (interquartile range), mg/ml	0.31 (0.23–0.48)	0.44 (0.37–0.64)	<0.01
No. of anti–ds-DNA positive (%)	21 (87.5)	42 (75)	0.34
Pathologic AI score median (interquartile range)	7 (4–9)	6 (5–8)	0.94
Pathologic CI score median (interquartile range)	1 (1–2.5)	1.5 (1–2)	0.54
SLEDAI median (interquartile range)	19 (17–22)	17 (12–21)	0.13

ds-DNS, anti-double stranded DNA antibody; AI, activity index; CI, chronicity index; SLEDAI, systemic lupus erythematosus disease activity index.

By alanine scanning and competition assays, Leu37, Phe39, Tyr40, and Leu43 were determined as the key residues critical to the antigenicity of a.a. 35–47 epitope (Figure 1F). Mutating each of these residues reversed the inhibition of the corresponding peptide mutant on autoantibody binding to immobilized mCRP. Of note, most of a.a. 35–47, including Leu37, Phe39, and Tyr40, were buried in the native subunit structure of CRP (Figure 1G) and would be exposed only after the generation of mCRP.^18,19 The exposure of a.a. 199–206, the sequence lining the subunit contact interface, however, may occur both in mCRP^18,20 and on reversible conformation changes in CRP.^12,13 Taken together, identification of a.a. 35–47 as the predominant epitope recognized by autoantibodies strongly supports the in vivo generation of mCRP in LN.

Anti-a.a. 35–47 Autoantibodies Are Associated with Prognosis of LN

Further analysis indicated that patients with LN with anti-a.a. 35–47 autoantibodies had more serve renal damage than those without as evidenced by the higher activity index (P=0.04) and chronicity index (P=0.004) scores (Table 3). The average follow-up time of our patients was 48 months (34–58 months), during which eight patients reached the secondary end point of doubling of serum creatinine, two patients reached ESRD, and no one died. Importantly, the presence of anti-a.a. 35–47 autoantibodies was associated with poor outcome by log rank test (P=0.02) (Figure 2) and univariate Cox hazard analysis (hazard ratio, 5.17; 95% confidence interval, 1.44 to 18.62; P=0.01) (Table 4). Anti-a.a. 35–47 autoantibodies remained as an independent risk factor even after multivariate adjustment (hazard ratio, 6.77; 95% confidence interval, 1.14 to 39.92; P=0.04). By contrast, anti-a.a. 199–206 autoantibodies were not associated with either clinical characteristics or prognosis of LN. The inverse association with prognosis thus suggests a pathogenic role of anti-a.a. 35–47 autoantibodies in the disease.

Table 3.

Comparisons of clinical data between patients with LN with epitope-specific mCRP autoantibodies and patients with LN without epitope-specific mCRP autoantibodies

Clinical Feature	Patients with LN and Anti-mCRP Positive, n=24
	Anti-a.a. 35–47 Positive	Anti-a.a. 35–47 Negative	P Value	Anti-a.a.199–206 Positive	Anti-a.a. 199–206 Negative	P Value
No. of patients	11	13		17	7
Age mean±SD, yr	36.36±16.97	25.38±7.69	0.08	31.76±14.48	27.14±11.88	0.66
Sex, men/women, %	1/10	0/13	0.93	0/17	1/6	0.64
No. of leukocyturia (%)	5 (45.4)	9 (69.2)	0.45	9 (52.9)	5 (71.4)	0.70
No. of hematuria (%)	3 (27.2)	8 (61.5)	0.83	12 (70.6)	7 (100)	0.29
No. of ARF (%)	2 (18.2)	3 (23.1)	0.66	3 (17.6)	2 (28.6)	>0.99
Urine protein median (interquartile range), g/24 h	1.46 (0.61–3.41)	3.94 (1.21–5.86)	0.25	3.41 (0.48–6.63)	2.51 (0.90–4.56)	0.90
Serum creatinine median (interquartile range), mmol/L	94 (64–214)	63 (44–109)	0.11	66 (53–139)	85 (43–544)	0.46
Creatinine clearance rate mean±SD, ml/min	54.71±36.18	93.97±42.79	0.04	90 (43–115.6)	71 (35.4–94.75)	0.40
C3 median (interquartile range), mg/ml	0.39 (0.19–0.5)	0.31 (0.24–0.42)	0.96	0.40±0.19	0.38±0.22	0.46
No. of anti–ds-DNA positive (%)	8 (72.7)	13 (100)	0.16	16(94.1)	5 (71.4)	0.40
Pathologic AI score median (interquartile range)	8.5 (6–9)	6 (2–8)	0.04	7.5 (6.75–8.25)	6 (3–9)	0.47
Pathologic CI score median (interquartile range)	2.5 (1.75–2.45)	1 (0–1)	0.004	1 (1–2)	1.5 (0.75–3.25)	0.73
SLEDAI median (interquartile range)	18 (16–22)	20 (17–23)	0.42	19.43±5.32	19.43±5.31	0.90

ds-DNS, anti-double stranded DNA antibody; AI, activity index; CI, chronicity index; SLEDAI, systemic lupus erythematosus disease activity index.

Figure 2.

Anti-a.a. 35–47 autoantibodies predict worse prognosis of patients with LN. Kaplan–Meier plots of composite end point events in 80 patients with LN with or without (A) gross anti-mCRP, (B) anti-a.a. 35–47 autoantibodies, or (C) anti-a.a. 199–206 autoantibodies.

Table 4.

Univariate analysis of renal survival of patients with LN

Clinical Feature	HR	95% Confidence Interval	P Value
Age	0.11	0.02 to 0.79	0.003
Sex	2.64	1.34 to 5.72	0.01
Leukocyturia	1.33	0.38 to 4.72	0.66
Hematuria	4.28	0.76 to 24.10	0.10
ARF	7.18	2.03 to 25.33	0.002
Urine protein	1.25	1.09 to 1.43	0.001
Serum creatinine	1.00	1.00 to 1.01	0.003
Creatinine clearance rate	0.96	0.93 to 0.99	0.01
C3	0.44	0.03 to 7.06	0.56
Anti–ds-DNA positive	0.57	0.15 to 2.22	0.42
Pathologic AI score	1.39	1.06 to 1.83	0.02
Pathologic CI score	1.6	1.07 to 2.40	0.02
SLEDAI	1.01	0.97 to 1.05	0.79
Antigross mCRP autoantibodies	1.71	0.48 to 6.07	0.41
Anti-a.a. 35–47 mCRP autoantibody	5.17	1.44 to 18.62	0.01
Anti-a.a. 199–206 mCRP autoantibody	0.98	0.21 to 4.64	0.98

HR, hazard ratio; ds-DNS, anti-double stranded DNA antibody; AI, activity index; CI, chronicity index; SLEDAI, systemic lupus erythematosus disease activity index.

Anti-a.a. 35–47 Autoantibodies Inhibit the Interaction of mCRP with CFH

We and others have shown that mCRP can bind CFH^21,22 (Figure 3A), which may play a beneficial role in LN.^16,17 However, it is unclear which sequence motif in mCRP mediates the binding. By competition with peptides derived from CRP sequences, only a.a. 35–47 were found to abrogate the interaction of fluid-phase mCRP with immobilized CFH (Figure 3, B and C). We further showed that a.a. 35–47 could directly bind CFH (Figure 3D), whereas mCRP mutant without this motif (mCRP Δ35–47) showed negligible binding (Figure 3E). The above results were independently replicated by surface plasmon resonance (SPR) assays (Figure 4, A–E). These together show that mCRP binds CFH via a.a. 35–47.

Figure 3.

a.a. 35–47 mediate the binding of mCRP to CFH in ELISAs. (A) Urea-denatured mCRP, recombinant Cys-mutated mCRP, or native CRP at the indicated concentrations was added to CFH immobilized onto microtiter wells. Cys-mutated mCRP represents an mCRP conformation with enhanced activities.¹⁹ The binding was detected with mCRP-specific mAb 3H12 or native CRP-specific mAb 1D6.^18,20 mCRP bound strongly to CFH, whereas native CRP did not. (B) Urea-denatured mCRP or (C) Cys-mutated mCRP was added to immobilized CFH together with the indicated CRP peptides (n=3). Prominent inhibition of mCRP binding was only observed with a.a. 35–47. (D) a.a. 35–47 with a C-terminal biotin tag were added to immobilized CFH (n=3). Peptide binding was determined with HRP-labeled streptavidin. This confirmed the direct binding of a.a. 35–47 to CFH. (E) Wild-type and mutant mCRP lacking a.a. 35–47 were expressed in Escherichia coli. These proteins were purified, and their binding to immobilized CFH was examined (n=3). Wild-type mCRP showed strong binding as expected, whereas the binding capacity was lost on deletion of a.a. 35–47. (F) Purified anti-a.a. 35–47 but not anti-a.a. 199–206 autoantibodies inhibited the binding of mCRP to CFH. *P<0.05.

Figure 4.

a.a. 35–47 mediate the binding of mCRP to CFH in SPR assays. (A) Urea-denatured mCRP, (B) Cys-mutated mCRP, (C) a.a. 35–47 peptide, (D) mCRP Δ35–47, or (E) native CRP was injected in fluid phase to CFH conjugated to CM5 chips. CM5 chips without CFH conjugation served as controls. Dose-dependent binding was observed for mCRP and a.a. 35–47 but was not detected with mCRP Δ35–47 and native CRP. (F) Binding of 6.25 (blue), 12.5 (pink), 25 (green), or 50 nM (purple) mCRP to CFH in the absence or presence of 26.5 nM anti-a.a. 35–47 autoantibodies or human IgG. mCRP binding was markedly reduced by the coinjection of anti-a.a. 35–47 autoantibodies. Human IgG showed little inhibition.

One major function of CFH is to act as the cofactor for factor I–mediated cleavage and inactivation of C3b, thus inhibiting the formation of C5 convertase and the overactivation of complement.^14,15 We thus next examined whether a.a. 35–47-mediated mCRP binding affected the cofactor activity of CFH. Consistent with the previous report,²¹ mCRP but not CRP significantly enhanced the cofactor activity of CFH, resulting in increased cleavage of C3b as detected by immunoblotting (Figure 5, A–C). Of note, such enhancement was also observed with the peptide of a.a. 35–47 (Figure 5D). These suggest that mCRP might be able to limit excessive complement activation through a.a. 35–47-mediated interaction with CFH. Consistent with such speculation, we found that CFH was enriched, whereas the terminal complement complex C5b-9 was frequently depleted at sites stained positive with mCRP in kidney tissues of patients with LN (Figure 5E). To exclude the possibility that the colocalization of mCRP and CFH in Figure 5E was not simply the result of the secondary anti-IgG identifying the human IgG in the immune deposits, we further performed the additional control experiments omitting primary antibodies or using irrelevant primary mAb (i.e., 8D8 recognizing native CRP but not mCRP), yielding a faint background or no staining (Supplemental Figure 1).

Figure 5.

a.a. 35–47 promote the cofactor activity of CFH. (A) Urea-denatured mCRP, (B) Cys-mutated mCRP (0–45 μg/ml), (C) native CRP (100–300 μg/ml), or (D) a.a. 35–47 peptide (1–3 mg/ml) was added to the indicated reaction mixtures for 1 hour (n=3). C3b cleavage was determined by immunoblotting; α′, β, and α″ represent uncleaved C3b α′-chain, C3b β-chain, and cleaved fragment of C3b α′ chain, respectively. The intensity ratio between bands α′ and α″ was calculated to quantify the cofactor activity of CFH. This index is more reliable than the absolute intensity of α″, because variations in loading and/or developing across samples and experiments are largely eliminated by the calculation of the ratio. mCRP and a.a. 35–47 significantly promoted the cofactor activity of CFH, whereas CRP did not. (E) Immunostaining of mCRP, CFH, and C5b-9 in kidney tissues of patients with LN. The signals of mCRP colocalize with that of CFH but show little colocalization with that of C5b-9. The Pearson coefficients evaluating the extent of colocalization are also shown in the right panel. (F) The cofactor activity of CFH was measured by iC3b generation using a sandwich ELISA after coincubation of the indicated components for 1 hour (n=3): C3b, 7.5 μg/ml; CFH, 5 μg/ml; complement factor I, 0.25 μg/ml; mCRP, 20 μg/ml; and autoantibodies, 15 μg/ml. The enhancement of mCRP on CFH’s cofactor activity was reversed by anti-a.a. 35–47 but not anti-a.a. 199–206 autoantibodies. *P<0.05; **P<0.01.

The above findings would predict an inhibition of anti-a.a. 35–47 autoantibodies on mCRP-CFH interaction. We thus isolated anti-a.a. 35–47 autoantibodies from a patient with LN undergoing plasma exchanging. The purified anti-a.a. 35–47 autoantibodies bound specifically to a.a. 35–47 but did not bind to a control CRP peptide (not shown). As expected, anti-a.a. 35–47 autoantibodies markedly inhibited mCRP binding to immobilized CFH in both ELISA (Figure 3F) and SPR assays (Figure 4F) and reversed the enhancement of mCRP on CFH’s cofactor activity (Figure 5F). By contrast, human IgG and anti-a.a. 199–206 autoantibodies showed little effect. Similar results were also obtained for the interaction of surface-bound mCRP with fluid-phase CFH (not shown). These results further corroborate the functional importance of a.a. 35–47 and imply a pathogenic role of autoantibodies against this epitope in LN.

Discussion

This study has identified a.a. 35–47 as the predominant autoimmune epitope on CRP associated with renal injury and prognosis of LN. Importantly, this epitope is unique to mCRP conformation and constitutes the major ligand binding site. Because the presentation of a.a. 35–47 epitope reflects both the generation and the function of mCRP, it is plausible that mCRP plays a direct role in renal injury. CRP transported from the circulation and produced locally by tubular epithelial cells²³ likely are major sources for mCRP generation in inflamed kidneys. Accordingly, mCRP has been found in renal tissues of diabetic kidney disease²⁴ and LN,²⁵ although sample preparation procedure (e.g., fixation or antigen retrieval) might affect the detected conformation of CRP. On generation, mCRP could then promote silent clearance of cell debris via activating the classic pathway of complement by interacting with C1q,^19,22 while preventing complement overactivation by recruiting CFH.^21,22

However, in situations where clearance mechanisms are overwhelmed, mCRP accumulated on cell debris^26,27 or immune complexes²⁸ in kidney tissues would instead be transported to lymph nodes and recognized by B cells,²⁹ leading to the production of autoantibodies. Of the two autoimmune epitopes identified herein, the exposure of a.a. 35–47 is a better proxy for mCRP generation, whereas a.a. 199–206 could be exposed on both mCRP and CRP with reversible^12,13 or moderate conformation changes.³⁰ Therefore, the levels of anti-a.a. 35–47 autoantibodies are more specific to the underlying mechanisms of tissue damage, but a large portion of anti-a.a. 199–206 autoantibodies is presumably derived from CRP conformations other than mCRP, likely induced by systemic autoimmune activation. This might account for the lack of association of a.a. 199–206-specific autoantibodies with renal prognosis.

Because the interactions of mCRP with C1q and CFH are both mediated by a.a. 35–47 on the basis of our previous work³¹ and this study, autoantibodies recognizing this epitope could then interfere with these interactions as shown herein. Such a mistaken targeting by autoantibodies may thus represent a vicious feedback loop that further aggravates renal damage. In this regard, other than being a potential marker of diagnosis, the epitope of a.a. 35–47 might also be exploited therapeutically to offer additional benefit when combined with conventional therapy. Administration of a synthesized peptide of a.a. 35–47 will (1) in principle, neutralize the corresponding autoantibodies to release their inhibition on the interactions of mCRP with CFH; (2) suppress mCRP-induced proinflammatory cell responses^31,32; and (3) directly enhance the cofactor activity of CFH to limit excessive complement activation. Given the central role of CFH in complement regulation,^14,15 the third action of a.a. 35–47 peptide allows its application to other complement-mediated diseases independent of mCRP involvement.

Genetic association between noncoding polymorphism of CRP and SLE susceptibility in humans has been inconclusive.⁹ Previous studies have also reported inconsistent results on the effects of overexpressed^33,34 or administrated^35,36 human CRP in SLE and diabetic kidney disease in animal models. Here, we report the discovery of anti-a.a. 35–47 autoantibodies and their close association with renal prognosis, suggesting a critical role of mCRP in LN. However, the clinical prognosis findings of anti-a.a. 35–47 autoantibodies are preliminary due to the small number of patients, and larger validation cohorts are necessary to substantiate our findings. Furthermore, the localized generation of mCRP in inflamed tissues and the conformation-dependent actions may underlie the current controversies and require a refined animal model, in which the actions of different CRP conformations could be faithfully recapitulated and differentiated.

Concise Methods

Reagents

Human native pentameric CRP (purity >99%; purified from ascites) was purchased from the BindingSite (Birmingham, United Kingdom; catalog no. BP300.X). Urea-denatured mCRP and recombinant mCRP mutants were prepared as described.³¹ Proteins were dialyzed to remove NaN₃ and passed through Detoxi-Gel Columns (catalog no. 20344; Thermo Fisher Scientific, Rockford, IL) to remove the endotoxin where necessary. CRP peptides (purity >98%) were synthesized by Scilight Biotechnology (Beijing, China) and Science Peptide Biologic Technology (Shanghai, China). CFH was obtained from Calbiochem (Darmstadt, Germany).

3H12 mAb recognizing mCRP and 1D6 and 8D8 mAbs recognizing CRP were prepared as described.^18,20 Autoantibodies were purified according to the protocol used in our previous work.³⁷ Briefly, gross IgG proteins were first purified from the serum of a patient with LN undergoing plasma exchange through a protein G column. Epitope-specific autoantibodies were then isolated from gross IgG proteins by a.a. 35–47- or 199–206-conjugated resins.

Patient Selection

Eighty patients were enrolled with renal biopsy–proven LN diagnosed between January of 2003 and July of 2013 at Peking University First Hospital. These patients fulfilled the 1997 American College of Rheumatology revised criteria for SLE,³⁸ and their clinical disease activities were assessed using the SLEDAI.³⁹ They were followed up by outpatient clinic. Death was defined as the primary end point, and ESRD or doubling of serum creatinine was defined as the secondary end point. The demographic and clinical data of the patients are summarized in Table 1.

The plasma samples of patients were obtained on the same day of renal biopsy before initiation of immunosuppressive treatment. Plasma samples of 24 healthy donors were collected as normal controls with matched sex and age distribution. All of the samples were stored at −80°C in aliquots. Informed consents for blood sampling and renal biopsy were signed by all of the participants. The research was in compliance of the Declaration of Helsinki and approved by the local ethical committee.

The renal biopsy specimens were examined by light, immunofluorescence, and electron microscopy. LN was reclassified according to the International Society of Nephrology/Renal Pathology Society 2003 classification system.⁴⁰ Pathologic parameters, including activity indices and chronicity indices, were determined by renal pathologists using a semiquantitative scoring system.^41,42

Characterization of CRP Autoimmune Epitopes

Autoantibody binding assays were performed according to our previous work.⁸ Briefly, urea-denatured mCRP (2 μg/ml) or synthesized CRP peptides (10 μM) were immobilized onto microtiter wells (Nunc; Roskild, Denmark) in 100 μl 0.05 M bicarbonate buffer (pH 9.6) for 1 hour at 37°C. Wells with buffer alone served as antigen-free controls. Test sera were diluted at 1:100 in PBS containing 0.05% Tween-20 and 0.2% BSA and added in duplicate to both antigen-coated wells and antigen-free wells at 37°C for 1 hour followed by washing. In competition assays, test sera were added to immobilized mCRP together with synthesized CRP peptides. Autoantibody binding was determined with alkaline phosphatase–conjugated goat anti-human IgG (1:5000; Fc specific; Sigma Aldrich, St. Louis, MO). Wells were then developed, and absorbance at 405 nm was measured. Twenty-four sera from healthy donors were examined following the same protocol, and the resultant mean plus 2 SDs of the absorbance values were set as the autoantibody-positive cutoff.

CFH Binding Assays

ELISAs were performed as described³¹ with minor modification. Briefly, 4 μg/ml CFH was immobilized onto microtiter wells in 100 μl TBS-Ca (10 mM Tris, 140 mM NaCl, and 2 mM CaCl₂, pH 7.4) for 1 hour at 37°C. Wells were washed with TBS-Ca containing 0.02% Tween 20 and blocked with 1% BSA/TBS-Ca for 1 hour. CRP, mCRP, or mCRP mutants at the indicated concentrations were added for 1 hour. In competition assays, mCRP was added together with the indicated CRP peptides. The binding was detected with CRP or mCRP mAbs and horseradish peroxidase (HRP)–conjugated goat anti-mouse IgG antibodies (1:5000; Gibco BRL). Wells were developed, and absorbance at 450 nm was measured.

The interactions of CFH with CRP, mCRP, and CRP peptides were also examined by SPR using a Biacore T200 (Biacore AB, Uppsala, Sweden). CFH was immobilized onto CM5 chips to reach an response unit of 1000. Test ligands were then injected at a flow rate of 30 μl/min in TBS-Ca containing 0.005% P-20. Injection of buffer alone served as negative control.

CFH Activity Assay

C3b cleavage assays were performed as described.²¹ Briefly, 3 μg C3b was added to 1 μg factor I (Sigma Aldrich), 1 μg CFH, and tested ligands (CRP, mCRP, or CRP peptides) at the indicated concentration in 20 μl buffer and incubated for 1 hour at 37°C in TBS-Ca. The reaction mixtures were then separated by 12% SDS-PAGE and transferred to nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany) using an electrophoretic semidry blotting system (Amersham Pharmacia). Membranes were blocked for 1 hour with TBS, 0.1% Tween-20, and 20 g/L skimmed milk and then incubated with rabbit anti-human C3c antibody (1:10,000) overnight at 4°C. Bands were visualized with an HRP-conjugated goat anti-rabbit IgG (1:5000) and chemiluminescent HRP substrate (Millipore Corporation).

Alternatively, 7.5 μg/ml C3b, 5 μg/ml CFH, 0.25 μg/ml factor I, and 20 μg/ml mCRP were incubated for 1 hour at 37°C in 1% BSA and TBS-Ca with or without 15 μg/ml purified autoantibodies. The reaction mixtures were diluted and added to microtiter wells coated with rabbit anti-human C3b antibody (Dako). The levels of iC3b were detected with mouse anti-human iC3b mAb (Quidel Corporation) and HRP-conjugated goat anti-mouse IgG antibodies (Bioworld Technology).

Immunofluorescence

Renal tissues of patients with LN were obtained at biopsy and embedded in optimal cutting temperature compound. Cryostat-cut tissue sections (10-μm thick) were fixed in cold acetone for 15 minutes followed by washing with PBS and blocking with 1% BSA and 3% FBS for 1 hour at room temperature. The sections were then incubated with primary antibodies (mouse anti-human mCRP mAb 3H12, rabbit anti-human CFH mAb, or rabbit anti-human C5b-9 polyclonal antibody; Abcam) overnight at 4°C. After extensive washing, sections were stained with secondary antibodies (Alexa-647 goat anti-mouse IgG H&L and Dylight-550 goat anti-rabbit IgG H&L; Abcam) for 1 hour at room temperature. After DAPI staining, the sections were examined with Zeiss 710 confocal microscopy.

Statistical Analyses

Data were expressed as means±SD or medians with range (minimum to maximum). Statistical analysis was performed with t test, one-way ANOVA, or chi-squared test as appropriate. Kaplan–Meier curves were used to analyze patients’ prognosis. Univariate survival analysis was carried out using the log rank test. Multivariate survival analysis was performed using the Cox regression model. Values of P<0.05 were considered significant.

Disclosures

None.