Laminin-521 is a Novel Target of Autoantibodies Associated with Lung Hemorrhage in Anti-GBM Disease

Abstract

Visual Abstract

Background 

Antiglomerular basement membrane (anti-GBM) disease is characterized by GN and often pulmonary hemorrhage, mediated by autoantibodies that typically recognize cryptic epitopes within α345(IV) collagen—a major component of the glomerular and alveolar basement membranes. Laminin-521 is another major GBM component and a proven target of pathogenic antibodies mediating GN in animal models. Whether laminin-521 is a target of autoimmunity in human anti-GBM disease is not yet known.

Methods 

A retrospective study of circulating autoantibodies from 101 patients with anti-GBM/Goodpasture’s disease and 85 controls used a solid-phase immunoassay to measure IgG binding to human recombinant laminin-521 with native-like structure and activity.

Results 

Circulating IgG autoantibodies binding to laminin-521 were found in about one third of patients with anti-GBM antibody GN, but were not detected in healthy controls or in patients with other glomerular diseases. Autoreactivity toward laminin-521 was significantly more common in patients with anti-GBM GN and lung hemorrhage, compared with those with kidney-limited disease (51.5% versus 23.5%, P=0.005). Antilaminin-521 autoantibodies were predominantly of IgG1 and IgG4 subclasses and significantly associated with lung hemorrhage (P=0.005), hemoptysis (P=0.008), and smoking (P=0.01), although not with proteinuria or serum creatinine at diagnosis.

Conclusions 

Besides α345(IV) collagen, laminin-521 is another major autoantigen targeted in anti-GBM disease. Autoantibodies to laminin-521 may have the potential to promote lung injury in anti-GBM disease by increasing the total amount of IgG bound to the alveolar basement membranes.

GN is a significant cause of ESKD, often caused by autoantibody (autoAb) deposition in the glomerular capillary wall.¹ In antiglomerular basement membrane (GBM) disease, also known as Goodpasture’s (GP) disease, autoAbs that bind to α345(IV) collagen mediate rapidly progressive GN, sometimes accompanied by pulmonary hemorrhage.² Most patients with anti-GBM/GP disease produce IgG autoAbs recognizing two conformational epitopes within the noncollagenous (NC1) domain of α3(IV) collagen (α3NC1).^3–5 These immunodominant epitopes are hidden (“cryptic”) in the native GBM at the interface between NC1 subunits of α345NC1 hexamers, but become accessible to anti-GBM autoAbs on hexamer dissociation.^6,7 Additional autoAbs, possibly induced by intramolecular epitope spreading, may variably target other sites within α345(IV) collagen, including the α4NC1 or α5NC1 subunits or epitopes accessible in NC1 hexamers.^8–11 Yet, the repertoire of GBM autoantigens is not limited to α345(IV) collagen, as shown by the recent identification of autoAbs against peroxidasin (an ubiquitous basement membrane component) in patients with GP disease, ANCA-associated vasculitis, and lupus nephritis.^12,13

A prominent potential GBM autoantigen is laminin, a family of αβγ heterotrimeric glycoproteins that are abundant components of all basement membranes.¹⁴ At least 16 laminin isoforms with tissue-specific distribution are assembled from a repertoire of five α chains, four β chains, and three γ chains.¹⁵ Several laminin isoforms are well-established autoantigens implicated in various human autoimmune diseases: laminin-332 in mucous membrane pemphigoid, laminin γ1 in anti-p200 pemphigoid and cutaneous lupus erythematosus, and laminin-511 in autoimmune pancreatitis.^16,17 However, it is not yet known whether laminin-521 (LM521; α5β2γ1), the major isoform in the mature GBM, is a target of autoimmunity in human anti-GBM GN.

The nephritogenicity of anti-GBM antibodies targeting laminin has been convincingly demonstrated in animal models. Female wild-type mice mated with human LAMA5 transgenic males become alloimmunized against the fetal human laminin α5 expressed in transgenic pups. In subsequent pregnancies, transfer of maternal IgG alloantibodies results in IgG binding to the GBM laminin in LAMA5-transgenic pups (but not in wild-type littermates), causing perinatal anti-GBM GN.¹⁸ In addition, rats develop GN mediated by GBM-bound antilaminin antibodies¹⁹ or induced by immunization with peptides derived from laminin α1, α5, and β1 chains.²⁰

Whether true antilaminin autoAbs occur in human glomerulonephritides is controversial. Although antibodies binding to mouse laminin-111 have been described in patients with GP disease, poststreptococcal GN, IgA nephropathy, SLE, or ANCA-associated vasculitis, proof of their binding to human laminins is lacking.^21–25 One caveat about immunoassays using mouse laminin-111 is the presence of galactosyl (α1–3)-galactose (αGal) determinants, which humans have lost the ability to synthesize.²⁶ Most individuals have high titers of natural anti-αGal antibodies, which can crossreact with the αGal structure present on mouse laminin-111.²⁷ Therefore, immunoreactivity of human sera toward mouse laminin-111 may simply reflect the presence of natural anti-αGal antibodies.²⁸

In this study, we tested the hypothesis that LM521 is a target autoantigen in human anti-GBM/GP disease. To this end, we developed an immunoassay using recombinant full-length human LM521, which has been shown to have native-like structure and biologic activities.²⁹ We discovered that autoAbs to native human LM521 occur in a large proportion of patients with anti-GBM/GP disease and investigated their properties and clinical associations.

Methods

Patients and Clinical Data

We enrolled 101 patients diagnosed with anti-GBM disease in Beijing University First Hospital from January 2000 to September 2018. The diagnostic criteria of anti-GBM disease were the detection of circulating antibodies against human α3NC1 and/or the observation of bright linear deposit of IgG along the GBM, with or without necrotizing crescentic GN. Pulmonary hemorrhage (lung involvement) was defined as the finding of hemoptysis, and/or signs of hemorrhage in computerized tomography of the chest, and/or hemosiderin in sputum. Patients without complete clinical data were excluded. The clinical data were collected from the time of diagnosis. The history of exposure to hydrocarbons and smoking status, self-reported by the patients, were retrieved from the medical records. All 101 patients were monitored in follow-up visits. The renal end point was set as maintenance dialysis lasting for 3 months. Death by all causes was set as the observation end point. The composite end point referred to the presentation of renal end point or patient death. Complete histologic data, available for 43 patients who underwent renal biopsy, were assessed.

Plasma from 30 age- and sex-matched healthy individuals was used as normal control. Disease controls included plasmapheresis effluents from patients with ANCA-associated vasculitis (20 patients), thrombotic microangiopathy (15 patients), and crescentic IgA nephropathy (20 patients), and sera from patients with membranous nephropathy (20 patients), diabetic nephropathy (20 patients), minimal change disease (20 patients), and membranoproliferative GN (20 patients), collected on the day of the kidney biopsy (i.e., during active disease). The study was performed with informed consent and approved by the Ethics Committee of Peking University First Hospital.

Immunoassays for Circulating IgG AutoAbs Against Human LM521

An ELISA immunoassay originally developed to assay murine antilaminin antibodies was modified to measure human IgG autoAbs binding to LM521.¹⁸ Full-length recombinant human LM521 was purchased from BioLamina (Sundbyberg, Sweden). Polystyrene microtiter plates (Nunc Maxisorb; ThermoFisher Scientific, Waltham, MA) were coated overnight at 4°C with LM521 at 200 ng/well, whereas antigen-free wells were coated with 300 ng/well BSA, both in bicarbonate buffer, pH 9.6. After blocking with 1% BSA for 1 hour at 37°C, wells were incubated for 1 hour at 37°C with sera from patients or controls diluted 1:100. After washing, horseradish peroxidase (HRP)-labeled goat anti-human IgG (Sigma, St Louis, MO) diluted 1:5000 was added for 1 hour at 37°C. Finally, HRP substrate 3,3′,5,5′-tetramethylbenzidine was added. The plates were read at 450 nm and OD values recorded. For each serum, the background OD value in BSA-coated wells was subtracted from the sample OD value in LM521-coated wells. The threshold for positivity was set at 3 SDs above the mean of control sera. For readings in different plates, OD values were normalized using a common positive control sample. The same assay was used at both sites with similar results. A subset of 31 anti-GBM samples were initially analyzed at one site. All anti-GBM samples and disease controls were then assayed at the second site. Experiments analyzing IgG subclasses were performed as above, except that HRP-conjugated sheep antihuman IgG1, IgG2, IgG3 and IgG4 (The Binding Site, San Diego, CA), diluted 1:4000, were used as secondary antibodies. In other experiments, LM521 (2 μg/ml) was coated in PBS or in 3 mol/L guanidinium hydrochloride, in the presence and the absence of 10 mM tris(2-carboxyethyl)phosphine as a reducing agent.

For immunoblot analyses, LM521 (4 μg/lane) was separated by electrophoresis in 6% SDS-polyacrylamide gel under both nonreducing and reducing conditions, along with 300 kDa prestained marker (HiMark; Life Technologies) and 1200 kDa nonstained marker (NativeMark; Life Technologies). One part of the gel was stained by Coomassie Brilliant Blue G250, and the other was transferred to a 0.45 μm polyvinylidene fluoride membrane (Merck KGaA, Darmstadt, Germany) using semidry blotting. The membrane was blocked for 60 minutes with 2% skimmed milk in Tris-buffered saline, pH 7.2, containing 0.1% Tween 20, and then incubated overnight at 4°C with anti-GBM effluent or normal plasma, diluted 1:50 in blocking buffer. After washing and incubation with alkaline phosphatase-conjugated anti-human IgG (1:5000; Sigma Aldrich, St. Louis, MO) for 60 minutes at room temperature, IgG bound to the membrane was detected using the nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate as a substrate.

Competition ELISA

Recombinant human α3NC1 protein was expressed and purified as described before.^5,30 Plates were coated with α3NC1 (200 ng/well) or LM521 (200 ng/well) in 50 mM carbonate buffer, pH 9.6, at 4°C overnight. Plasmapheresis effluents from three patients with anti-GBM that recognized LM521 with high reactivity were diluted at 1:100 and incubated for 1 hour at 37°C with varying concentrations of α3NC1 or LM521. After washing, the preincubated solutions were applied to antigen-coated wells for 30 minutes at 37°C. The wells were then incubated with alkaline phosphatase-conjugated goat anti-human IgG (γ chain specific; Sigma, St Louis, MO) diluted 1:5000, for 30 minutes at 37°C. After color development with p-nitrophenyl phosphate substrate (1 mg/ml in 1 M diethanolamine, 0.5 mM MgCl₂, pH 9.8), the absorbance was measured at 405 nm (Bio-Rad, Tokyo, Japan). For each sample, the OD value in antigen-free wells was subtracted from the OD value in the antigen-coated wells.

Affinity Purification of Anti-LM521 IgG

Circulating total IgG from plasmapheresis of five patients with anti-GBM disease and plasma of five health donors was affinity purified on a protein G column (GE Healthcare, Piscataway). Commercial His-tag protein purification beads (Beaver, Jiangsu, China) were incubated for 1 hour at room temperature with His-tagged recombinant human α3(IV)NC1 protein in 20 mM phosphate buffer, 500 mM NaCl, 20 mM imidazole, pH 7.4. The α3NC1-coated beads were washed with PBS, gently mixed for 1 hour at 37°C with IgG (0.1 mg/ml) purified from patients and healthy donors, then the supernatants were analyzed for binding to microtiter plate wells coated with α3NC1 or LM521.

In Vivo Binding of a Rat IgG mAb to α345(IV) Collagen in Mouse Kidneys and Lungs

Mice (wild-type C57Bl/6, either sex) were injected in the tail vein with anti-α5NC1 mAb b14 (250 μg rat IgG in PBS per 20 g body weight), which binds to native α345(IV) collagen in tissues and causes dose-dependent GN and pulmonary hemorrhage in WKY rats.³¹ Then 16 hours later, mice were sacrificed and perfused from the heart with PBS. Lungs were inflated with 50% optimal cutting temperature compound and collected, along with the kidneys. The tissues were embedded in optimal cutting temperature compound and snap frozen. Cryostat sections (5 μm thick) fixed with acetone for 10 minutes were blocked with 10% BSA for 1 hour and stained with Alexa Fluor 488–conjugated goat anti–rat IgG (Invitrogen, Carlsbad, CA) diluted 1/500. After mounting with antifade reagent (Invitrogen), the sections were examined under a Nikon Eclipse E800 epifluorescence microscope using the same exposure settings. Animal experiments were approved by the Institutional Animal Care and Use Committee at Vanderbilt University.

Analyses of Collagen IV in Kidneys and Lungs of Nonhuman Primates

Frozen kidneys and lungs from adult squirrel monkeys (Saimiri sp.) were obtained from the Squirrel Monkey Breeding and Research Resource (University of Texas MD Anderson Cancer Center, Houston, TX). For indirect immunofluorescence staining, acetone-fixed cryosections (5 μm thick) were treated for 5 minutes with 6 M urea in 0.1 M glycine, pH 2.2, on ice. After washing and blocking, sections were stained for 60 minutes with rat mAbs H11 or H31, specific for the NC1 domains of α1(IV) and α3(IV) collagen chains, respectively.³² Subsequent steps were as described above. For immunoblot analyses, basement membranes isolated from squirrel monkey kidneys and lungs were digested with bacterial collagenase (Worthington, Lakewood, NJ), as previously described.⁶ The supernatants containing NC1 domains (20 μg protein per lane) were separated by SDS-PAGE in a 4%–12% gradient gel. Proteins were transferred to 0.2 μM nitrocellulose membrane (Bio-Rad), which were blocked with 5% whole milk in PBS for 60 minutes, and then incubated overnight with rat mAb H11 or mAb H31. After incubation with HRP-labeled goat anti-rat IgG, bands were visualized with the Pierce ECL Plus chemifluorescent detection system (ThermoFisher Scientific).

Statistical Analyses

Statistical analyses were conducted on the software of GraphPad Prism (version 6.01) and SPSS statistical software (version 22.0; IBM). Quantitative data obeying to normal distribution were presented as mean±SD, or as median (interquartile range; IQR) when they were not normally distributed. Significance of differences between two groups was analyzed by t test for normally distributed data or by Mann–Whitney test otherwise. Qualitative data were described as counts and proportions (%), and the significance of differences was tested using chi-squared test or Fisher’s exact test. Additional statistical tests are described in the figure legends. Differences were considered to be statistically significant when the multiplicity-adjusted P<0.05.

Results

Clinical Data of Enrolled Patients

Demographic and clinical characteristics of patients with anti-GBM disease are shown in Table 1. This study enrolled 48 males and 53 females with a median age of 54.0 (IQR, 42.5–65.5) years. In total, 34.7% of patients were smokers and 14.9% were exposed to hydrocarbon before disease onset. Hemoptysis was observed in 25.7% of patients and pulmonary hemorrhage was identified in 32.7% of patients. Overall, 37.6% of patients presented gross hematuria and 43.6% presented oliguria or anuria. The levels of serum creatinine were 777.0 (IQR, 502.3–1064.5) μmol/L on diagnosis. The amounts of urine protein excretion ranged from a normal to nephrotic level, with a median level of 1.90 (IQR, 0.78–4.11) g/24 hour in 80 patients with available data, of which 13 (16.3%) reached nephrotic syndrome. Circulating ANCA, typically antimyeloperoxidase (MPO), were positive in 24.8% of patients. In 43 patients with available histologic data, 90.7% presented crescentic GN and the median proportion of crescents was 89.3% (Table 2). The majority of patients (71.3%, 72 out of 101) received combined immunosuppressive therapy comprising methyl prednisone pulse therapy (7–15 mg/kg per day for 3 days) sequenced by oral prednisone administration (1 mg/kg per day), and together with cyclophosphamide and plasmapheresis. During follow-up, 69.3% of patients reached the renal end point and 15.8% of patients died. A significantly higher proportion of patients with lung involvement reached the composite end point (90.9%, 30 of 33 versus 66.2%, 45 of 68; P=0.008).

Table 1. - Clinical characteristics of patients with anti-GBM disease stratified on the basis of the presence of IgG autoAbs targeting LM521

Characteristics a	Anti-LM521 Negative Patients (n=68)	Anti-LM521 Positive Patients (n=33)	Total patients (n=101)	P value
Demography
Gender (male/female) (n/N)	28/40	20/13	48/53	0.07
Age (yr)	56.8±16.8	45.4±17.6	54.0 (42.5–65.5)	0.003
Risk factors
Smoker n (%)	18 (26.5)	17 (51.5)	35 (34.7)	0.01
Hydrocarbon exposure n (%)	11 (16.2)	4 (12.1)	15 (14.9)	0.81
Prodromal infection n (%)	37 (54.4)	18 (54.5)	55 (54.5)	0.99
Clinical feature
Hemoptysis n (%)	12 (17.6)	14 (42.4)	26 (25.7)	0.008
Pulmonary hemorrhage n (%)	16 (23.5)	17 (51.5)	33 (32.7)	0.005
Gross hematuria n (%)	21 (30.9)	17 (51.5)	38 (37.6)	0.045
Oliguria/anuria n (%)	30 (44.1)	14 (42.4)	44 (43.6)	0.87
SCr at diagnosis (μmol/L)	799.4 (502.1–1030.0)	777.0 (500.8–1304.4)	777.0 (502.3–1064.5)	0.73
Urine total protein (g/24 h)	1.92 (0.78–3.52); n=57	1.83 (0.62–4.64); n=23	1.90 (0.78–4.11); n=80	0.93
Serum albumin (g/L)	29.8±5.1	29.1±4.9	29.6±5.0	0.52
Nephrotic syndrome n (%)	7 (12.3%); n=57	6 (26.1%); n=23	13 (16.3%); n=80	0.24
Hemoglobin (g/L)	85.5±18.3	81.9±17.0	84.3±17.9	0.35
OD of anti-α3NC1 Abs	1.25 (0.80–1.69)	1.34 (1.06–1.59)	1.32 (0.86–1.65)	0.63
ANCA n (%)	21 (30.9)	4 (12.1)	25 (24.8)	0.04
MPO/PR3/double n/n/n	18/2/1	4/0/0	22/2/1
Dialysis on diagnosis n (%)	37 (54.4)	16 (48.5)	53 (52.5)	0.58
Treatment
MP pulse therapy n (%)	50 (73.5)	27 (81.8)	77 (76.2)	0.36
Prednisone n (%)	66 (97.1)	33 (100)	99 (98.0)	0.81
Cyclophosphamide n,(%)	57 (83.8)	30 (90.9)	87 (86.1)	0.51
Plasmapheresis n (%)	68 (100)	33 (100)	101 (100)	1.00
On discharge
SCr (μmol/L) b	544.1±273.0	584.5±252.3	557.3±265.9	0.48
Free from dialysis n (%)	21 (30.9)	9 (27.3)	30 (29.7)	0.71
Outcome
Composite end point n (%)	46 (67.6)	29 (87.9)	75 (74.3)	0.03
Renal end point n (%)	44 (64.7)	26 (78.8)	70 (69.3)	0.15
Death n (%)	8 (11.8)	8 (24.2)	16 (15.8)	0.11

PR3, proteinase 3; MP, methyl prednisone; SCr, serum creatinine.

^aSummary statistics are shown as counts and proportion (%) for qualitative data, as mean±SD for normally distributed quantitative data, and as median (IQR) for non-normally distributed data.

^bFor patients on dialysis, Scr was the or the last value before dialysis began.

Table 2. - Comparison of pathologic characteristics of patients with anti-GBM disease with or without autoAbs to LM521

Pathology Findings in The Renal Biopsy a	Anti-LM521 Negative (n=30)	Anti-LM521 Positive (n=13)	Total Patients (n=43)	P value
IgG	28 (93.3)	12 (92.3)	40 (93.0)	1.00
IgA	13 (43.3)	3 (23.1)	16 (37.2)	0.36
IgM	17 (56.7)	7 (53.8)	24 (55.8)	0.86
C3	24 (80.0)	10 (76.9)	34 (79.1)	1.00
Percentage of glomeruli with any crescents (%)	88.4 (63.5–94.7)	93.8 (67.5–98.6)	89.3 (65.2–96.3)	0.54
Cellular crescents (%)	23.1 (12.0–50.7)	38.1 (16.1–86.9)	29.4 (12.0–63.6)	0.15
Fibrocellular crescents (%)	44.1 (22.8–60.3)	21.2 (0.0–56.3)	41.9 (10.0–59.3)	0.10
Fibrous crescents (%)	0.0 (0.0–7.5)	0.0 (0.0–1.8)	0.0 (0.0–6.9)	0.34

^aSummary statistics are shown as counts and proportion (%) for qualitative data, and as median (with the interquartile range IQR1–IQR3) for quantitative data.

Recognition of LM521 by Circulating IgG from Patients with Anti-GBM Disease

Plasmapheresis effluents from patients with anti-GBM/GP disease, along with sera from healthy control and patients with other kidney disease, were analyzed for the presence of IgG binding recombinant human LM521 with native structure as antigen. A total of 33 out of 101 (32.7%) patients with anti-GBM disease had circulating IgG binding to LM521. In contrast, serum samples from 30 healthy controls and sera or effluents from 155 disease controls (including patients with ANCA-associated vasculitis, thrombotic microangiopathy, membranous nephropathy, crescentic IgA nephropathy, diabetic nephropathy, FSGS, minimal change disease, and membranoproliferative GN) failed to recognize LM521 (Figure 1A). IgG immunoreactivities toward LM521 and α3NC1 were not correlated (Spearman r=0.084; P=0.40). The existence of IgG autoAbs binding to LM521 was corroborated by immunoblot (Figure 1B). IgG from anti-GBM effluents, but not from normal sera, showed binding to intact LN521 heterotrimers. IgG also bound one of approximately 200 kDa subunits separated under reducing conditions (β2 or γ1), which suggests at least some LM521 epitopes are not disulfide dependent.

Figure 1.: AutoAbs recognizing LM521 are detectable in patients with anti-GBM/GP disease. (A) Plasmapheresis effluents from patients with GP, ANCA-associated vasculitis (AAV), thrombotic microangiopathy (TMA) and crescentic IgA nephropathy (cIgAN), sera from patients with membranous nephropathy (MN), diabetic nephropathy (DN), FSGS, minimal change disease (MCD), and membranoproliferative GN (MPGN), and sera from normal controls (NC) were assayed for antibodies binding to native recombinant human LM521 (200 ng/well) by indirect ELISA. All samples were diluted 1:100. The scatterplot shows human IgG binding, expressed as OD. The dotted line shows the positivity threshold, defined as 3 SD above the mean OD of normal control sera (n=30). The significance of differences between groups was analyzed by Kruskal-Wallis test (P<0.0001), followed by Dunn’s test for pairwise comparisons to the normal control group (multiplicity-adjusted P values are shown in the figure). (B) After electrophoresis in 6% SDS-polyacrylamide gel, nonreduced LM521 (lane 1) is visualized as a single band at approximately 800 kDa by Coomassie Brilliant Blue (CBB) staining. Under reducing conditions (lane 2), LM521 subunits are separated producing one band at approximately 400 kDa (the α5 chain) and two bands at approximately 200 kDa (β2 and γ1 chains). After transferred to nitrocellulose and immunoblotting with an anti-GBM effluent (IB-GP), IgG binding is observed to the intact LM521 heterotrimer and to the approximately 200 kDa subunits. No IgG staining is observed after immunoblotting with a healthy control serum (IB-NC). Stained (M₁) and unstained (M₂) mol wt markers.

Additional experiments were performed to rule out crossreactivity between autoAbs against α3NC1 and LM521. Competitive inhibition assays showed that soluble LM521 protein did not inhibit IgG binding to immobilized α3NC1 (Figure 2A), whereas soluble α3NC1 did not inhibit IgG binding to LM521 (Figure 2B). After IgG samples purified from five anti-GBM effluents with dual reactivity were immunoabsorbed with α3NC1 antigen immobilized on to beads; the reactivity toward LM521 remained whereas binding to α3NC1 was abolished (Figure 2C). This implies distinct subsets of IgG autoantibodies bind each antigen.

Figure 2.: Verification of the specificity of anti-LM521 autoAbs. (A and B) Competitive inhibition of IgG binding to α3NC1 (A) or LM521 (B). Plasmapheresis effluents from three patients with anti-GBM disease, which recognized LM521 with high reactivity, were diluted 1:100, preincubated for 1 hour at 37°C with varying concentrations of soluble α3NC1 (circles) or LM521 (squares), and then IgG binding to plate-bound α3NC1 (200 ng/well) or LM521 (200 ng/well) was measured. (C) Affinity purification of anti-LM521 antibodies. IgG purified from effluents from five patients with anti-GBM with dual reactivity was absorbed with beads coated with α3NC1. Immunoabsorbed IgG reacted with LM521 but no longer bound α3NC1.

Association of Anti-LM521 AutoAbs with Clinico-pathologic Features and Prognosis

Patients who tested positive for anti-LM521 autoAbs, comprising 20 males and 13 females, were younger than those who tested negative (45.4±17.6 versus 56.8±16.8 years, P=0.003). They had significantly higher prevalence of smoking (51.5% versus 26.5%, P=0.01), hemoptysis (42.4% versus 17.6%, P=0.008), and pulmonary hemorrhage (51.5% versus 23.5%, P=0.005). The prevalence of gross hematuria was also higher in patients with anti-LM 521 autoAbs than those without (51.5% versus 30.9%, P=0.05). However, there were no significant differences between the two subgroups in the levels of serum creatinine at diagnosis, proteinuria, the percentages of glomerular crescents, or the therapeutic regimens (P>0.05). ANCA positivity was significantly less common among patients with anti-LM521 autoAbs, compared with those without (12.1% versus 30.9%, P=0.04).

Among the 33 patients with detectable anti-LM521 autoAbs, 26 (78.8%) progressed to ESKD and eight (24.2%) died. Meanwhile, among 68 patients with negative anti-LM521 Abs, 44 (64.7%) progressed to ESKD and eight (11.8%) died. A higher proportion of patients with anti-LM521 autoAbs reached the composite end point (defined as the occurrence of renal end point or the patient’s death) than those without anti-LM521 (67.6% versus 87.9%, P=0.03).

Properties of Anti-LM521 autoAbs and Their Epitopes

We further analyzed the profile of IgG subclasses of anti-LM521 autoAb, a determinant of IgG effector functions. The predominant subclasses of anti-LM521 were IgG1, which is complement fixing, and IgG4, which is not (Figure 3A). A predominance of IgG1 and IgG4 was also found for anti-α3NC1 autoAbs, consistent with previous reports.³³ There were no significant differences in anti-α3NC1 autoAbs between samples with and without anti-LM521 autoAbs, with regard to IgG subclasses (Figure 3B) or α3NC1 epitope specificity (not shown).

Figure 3.: The predominant subclasses of anti-LM521 autoAbs are IgG1 and IgG4. Scatterplots show the binding of human IgG1, IgG2, IgG3, and IgG4 subclasses to coated LM521 200 ng/well, (A) or α3NC1, 100 ng/well, (B). For each IgG subclass, the significance of differences between anti-GBM samples containing (filled symbols; n=12) or not containing (open symbols; n=8) anti-LM521 autoAbs was analyzed by unpaired t test, with Welch’s correction for unequal variance in (A). The Bonferroni-adjusted threshold of significance was P<0.01.

To determine whether the LM521 epitopes are linear or conformational, we compared IgG binding to native versus misfolded antigen. IgG autoAbs that reacted with native LM521 no longer bound to LM521 treated with reducing agent under denaturing conditions (Figure 4). This indicates anti-LM521 autoAbs recognize conformation-dependent epitopes, similar to anti-α3NC1 autoAbs. In the absence of a reducing agent, LM521 treatment with 3 mol/L guanidinium hydrochloride did not significantly affect IgG binding, suggesting the LM521 epitopes are accessible. This contrasts with anti-α3NC1 autoAbs, which show increased immunoreactivity toward guanidinium-treated α345NC1 hexamers because they recognize cryptic epitopes.³

Figure 4.: Properties of the epitopes targeted by anti-LM521 IgG. Before-after plot shows human IgG binding to LM521 (200 ng/well) coated in PBS (circles) or in 3 M guanidinium chloride (Gdm) in the presence (triangles) or absence of 10 mM tris(2-carboxyethyl)phosphine (squares). Anti-GBM sera containing (n=8) or not containing (n=4) anti-LM521 autoAbs were diluted 1/100. The significance of differences between groups was analyzed by repeated-measures ANOVA, followed by Tukey’s multiple comparison test. Multiplicity-adjusted P values are shown in the figure.

Discussion

Our results identify LM521 as a novel autoantigen targeted by IgG autoAbs in approximately 33% of patients with anti-GBM/GP disease. Anti-LM521 autoAbs were specific to anti-GBM/GP disease, being undetectable in other glomerular diseases. The prevalence of anti-LM521 autoAbs in patients with anti-GBM was comparable to that of anti-MPO autoAbs (10%–38%).³⁴ However, anti-LM521 autoAbs were rarely found in patients with GP and ANCA. Hence, serologic profiles afford classification of patients with anti-GBM/GP into three major subgroups. In this study, almost half (47 of 101) have isolated immunoreactivity toward α3NC1 alone, almost one third (29 of 101) have dual positivity toward α3NC1 and LM521, and about one fifth (21 of 101) have dual positivity against α3NC1 and MPO-ANCA. These groups appear to differ with regard to immunologic features, clinical profiles, and outcomes. For instance, a recent study has shown that patients with dual anti-α3NC1 autoAbs and ANCA had a greater tendency to recover from dialysis.³⁵ This study showed that lung involvement was significantly more common in patients with dual anti-α3NC1 and anti-LM521, who also had worse composite outcome.

Anti-LM521 autoAbs were associated with lung involvement, presumably because the alveolar basement membranes (ABM) contain LM521 subunits (α5, β2, and γ1 chains).^36–38 Anti-LM521 autoAbs were twice as common in patients with GP and both kidney and lung involvement compared with patients with kidney-restricted anti-GBM disease (51.5% versus 23.5%, P=0.005). Conversely, more than half of patients with anti-GBM/GP with anti-LM521 autoAbs presented pulmonary hemorrhage, compared with less than a quarter of patients negative for anti-LM521 (risk ratio, 2.19; 95% confidence interval, 1.27 to 3.76; P=0.005). Patients with anti-LM521 autoAbs also had higher prevalence of macroscopic hematuria (P=0.05), although proteinuria, serum creatinine, and the proportion of crescents did not differ significantly between the two groups.

A prominent role of anti-LM521 autoAbs in lung injury may be the result of differences in the collagen IV composition of the ABM and the GBM. In the human ABM, the relative amount of “Goodpasture antigen” per mg protein was measured to be about half (48%) compared with the GBM.³⁹ In bovine tissues, the proportion of α3NC1 to total NC1 is about three-fold lower in the ABM than in the GBM (5% versus 16%).⁴⁰ Likewise, our analyses of lungs and kidney from squirrel monkeys by immunofluorescence staining and immunoblot showed that α3NC1 is abundant in the GBM but scarce in the ABM, whereas α1(IV) collagen was abundant in the ABM and almost absent from the GBM (Figure 5, A and B). In mice injected with a model rat “anti-GBM antibody,” which binds to native α345(IV) collagen, IgG deposition along the ABM was conspicuously less intense than in the GBM (Figure 5C). The lower proportion of α3NC1 autoantigen in the ABM implies that less anti-α3NC1 IgG can bind to the ABM, which may be insufficient to cause lung injury. Additional autoAbs binding to laminin would increase the total amount of IgG bound to the ABM, possibly above the critical amount needed to incite antibody-mediated injury (Figure 6). This may explain why anti-LM521 are critical for lung injury but not for GN.

Figure 5.: ABMs contain less α345(IV) collagen than the GBM. (A) Indirect immunofluorescence staining of α1(IV) and α3(IV) collagen in the kidneys and lungs from squirrel monkeys. In kidneys, the GBM stains strongly for α3(IV) collagen whereas α1(IV) collagen mostly occurs in the mesangial matrix, Bowman’s capsule, and tubular basement membranes. Compared with the GBM, the lung ABM contains more α1(IV) collagen and less α3(IV) collagen. (B) Immunoblot analysis of collagenase-solubilized proteins from the basement membranes of squirrel monkey kidneys (K) and lungs (L). NC1 monomers (M) and dimers (D) were visualized after staining with mAbs specific for α1NC1 (top) and α3NC1 (bottom). (C) Direct immunofluorescence staining of tissues from mice injected with mAb b14 shows in vivo binding of rat IgG to α345(IV) collagen to kidney and lung basement membranes. There is bright linear staining for rat IgG in the GBM and moderately intense staining in the renal tubular basement membranes, but only faint staining in the lung ABM.

Figure 6.: Model depicting how anti-LM521 IgG may contribute to lung injury. Because the GBM contains mostly α345(IV) collagen, anti-α3NC1 IgG (solid “Y”) can accumulate in sufficient amounts to cause GN, whether anti-LM521 IgG (open “Y”) are also present (left). In contrast, in the ABM, α121(IV) collagen is more abundant, “diluting” α345(IV) collagen and thus limiting the amount of anti-α3NC1 IgG that can bind, possibly below the pathogenic threshold (right). Additional autoAbs targeting laminins within the ABM would increase the total amount of IgG bound at this site, which may then trigger antibody-mediated alveolar injury.

Anti-LM521 autoAbs mainly comprised IgG1 and IgG4 subclasses, which may contribute to tissue injury by different effector mechanisms. Tissue binding of anti-LM521 IgG1 can cause inflammation via complement activation and activating IgG Fc receptors. This occurs, for example, in perinatal anti-GBM GN induced by allo-antibodies against human laminin α5—a mouse model featuring complement activation, an influx of neutrophils, and proteinuria.¹⁸ In contrast, anti-LM521 IgG4 autoAbs are probably noninflammatory, yet they may be also pathogenic by interfering with LM521 biologic activities. Indeed, certain antilaminin antibodies have the ability to disturb laminin assembly into trimers, block interactions with type IV collagen, and/or prevent binding to integrin receptors on cells.^41,42

The finding of anti-LM521 autoAbs raises questions about their origin and relationship to “classical” anti-α3NC1 autoAbs. Several etiologic mechanisms are possible. First, anti-LM521 autoAbs may be elicited secondary to an ongoing autoimmune response toward α3NC1, possibly through intermolecular epitope spreading secondary to GBM structural damage or a bystander effect. This seems to occur in rats immunized with a nephritogenic α3NC1-derived peptide, which exhibit epitope spreading to peptides from laminins α1 and β1.²⁰ In a minority of patients, anti-LM521 autoAbs may be the primary cause of disease, eventually followed by epitope spreading to α3(IV) collagen. This may occur in rare cases of patients initially presenting with seronegative hemoptysis and normal kidney function, who later develop GN and become seropositive for anti-GBM autoAbs.⁴³ Finally, autoimmunity to α3NC1 and LM521 may be simultaneously induced by exposure to certain environmental risk factors altering both antigens. This mechanism is supported by our observation that anti-LM521 autoAbs were specifically associated with smoking, but not with exposure to hydrocarbons or prodromal infections. We postulate that smoking-induced alterations of the ABM are among the triggers initiating anti-LM521 autoimmunity.

Limitations of this study include its retrospective nature. Our experimental design cannot distinguish whether anti-LM521 autoAbs are a primary cause of disease or a secondary consequence of the disease process. Longitudinal observations will be essential to clarify this point. The pathogenicity of human anti-LM521 is not yet formally proven by demonstrating transfer of disease after passive immunization. Unsolved questions remain about the number and the location of LM521 epitopes, and whether the repertoires of anti-LM521 autoAbs are similar or different among patients.

In conclusion, this study identifies LM521 as a novel autoantigen specifically targeted in anti-GBM/GP disease. Anti-LM521 autoAbs occur in about one third of these patients and are significantly associated with lung involvement. Future studies should address whether and how anti-LM521 contribute to lung and kidney injury and their value as potential biomarkers.

Disclosures

D.-B. Borza reports receiving a Paul Teschan Research Fund grant from Dialysis Clinic Inc.; reports being a scientific advisor or member of the Editorial Board of BMC Nephrology (Associate Editor). M.-H. Zhao reports consultancy agreements with AstraZeneca, Norartis, and Roche; reports receiving honoraria from Chinese Medical Association, Chinese Society of Nephrology; reports being a scientific advisor or member as the Vice President of Chinese Society of Nephrology, Vice President of the Chinese Society of Internal Medicine, and Executive Member of the Asian-Pacific Society of Nephrology. All remaining authors have nothing to disclose.

Funding

This work was supported by a pilot project funded by the National Institute of Minority Health and Health Disparities grant U54 MD0007586 (to D.-B. Borza) and the National Key Research and Development Program (2016YFC1305400 to M.-H. Zhao), National Natural Science Foundation of China (81870482 to X.-Y. Jia; 81870486 to Z. Cui) and Chinese Academy of Medical Sciences Innovation Fund for Medical Sciences (2019-I2M-5-046). D.-B. Borza gratefully acknowledges support from the US Department of Defense Congressionally Directed Medical Research Programs award LR190150 and from a Dialysis Clinic Inc. Paul Teschan Research Fund grant (2018-04).