Copy number of FCGR3B, which is associated with systemic lupus erythematosus, correlates with protein expression and immune complex uptake

We sought to replicate the previously reported association between low FCGR3B CN and SLE (26). We assessed FCGR3B CN using a quantitative PCR method modified after that described by Fanciulli et al. (26), which compares the number of cycles required for amplification of FCGR3B to reach threshold, compared with the number of cycles required for amplification of a control gene that does not show CNV; in this case, CD36. FCGR3B/CD36 ratios as determined by this assay are not integers, as might have been expected, but form a near continuous distribution, which is a well recognized feature of CN assays (7, 8). We therefore compared cases and controls on each plate using a Student's t test, and then performed a Student's t test stratified by the two plates of samples. In total, 171 samples from Caucasian patients with SLE were compared with 176 ethnically matched control samples. We found lower CN was significantly associated with SLE in Caucasians (P = 0.027; Fig. 1 A and Fig. S1, available), but such an association was not seen in a Chinese SLE cohort from Hong Kong, comprising 159 cases and 150 controls (Fig. 1 B and Fig. S1). This lack of association held true when the patients with lupus nephritis (n = 57) were analyzed independently (Fig. S2). FCGR3B CN distributions did not differ significantly between the control Caucasian and Chinese populations (Fig. S3). SLE is more common, more severe, of earlier onset, and has some distinct clinical features in East Asians compared with Europeans. These clinical differences are independent of environment and likely to be genetically determined, thus it is to be expected that SLE will have different genetic influences in different ethnic groups (29); e.g., a transmembrane polymorphism in neighboring FCGR2B is associated with SLE in Asian (30), but not European or African-American, populations (31).

To define the function of FcγRIIIb, and whether this function was influenced by CNV, we first studied CN, expression, and function using a family with FCGR3B deficiency. As part of a comprehensive program of expression profiling of purified cell subsets in autoimmune disease, we identified an individual with SLE who had no FCGR3 mRNA in neutrophils, but normal levels of FCGR3 mRNA in monocytes (Supplemental clinical information, available, and Fig. 2 A). The probes used on our array do not differentiate between FCGR3A and B, but the expression pattern was consistent with absent FCGR3B. Flow cytometric analysis of the proband confirmed the lack of neutrophil cell surface FcγRIIIb. Surface expression of FcγRIIIb showed a clear trimodal pattern, consistent with Mendelian inheritance of an FCGR3B-null allele (Fig. 2, B and C). Fig. 2 C shows a flow cytometry plot for six family members labeled A–F showing the clear differences in protein expression in the proband (A), her son, daughter, and brother with a single FCGR3B-null allele (B–D), and her husband and other brother (E and F) with two copies of FCGR3B. FCGR3B CN, as measured by quantitative real-time PCR (qPCR), was lower in individuals B, C, and D than in individuals E and F (Fig. 2 D). FCGR3B CN, as determined by qPCR, correlated with CN inferred from protein expression and heredity in family members (Fig. 2 E). Further PCR analysis demonstrated that the genetic defect spared FCGR3A and FCGR2B and included HSPA7 and the FCGR2C (Fig. 2 F), consistent with previous observations in FCGR3B-null individuals (25), although the CN variable region at this locus may show more complexity and variability in the general population. The genomic defect seen in this family was confirmed at the protein level; FcγRI, FcγRIIa, and FcγRIIb were expressed on the cell surface of neutrophils, and FcγRIIIa was expressed on monocytes (Fig. S4). CD59 was also present on neutrophils (excluding a general defect in GPI linkage as an explanation for reduced FcγRIIIb; Fig. 2 G).

We next analyzed neutrophils from family members with known FCGR3B CN to define its function. Superoxide anion production in response to several stimuli thought to be FcR independent, such as formyl-Met-Leu-Phe (fMLP), granulocyte-macrophage colony-stimulating factor (GM-CSF), and PMA, was not affected by FCGR3B CN (Fig. 3 A). To assess the ability of neutrophils to localize to immune complexes, we analyzed the adherence of neutrophils flowed over surfaces coated with IgG, and demonstrated that adherence was proportional to FCGR3B CN (Fig. 3 B). Immune complex adherence to, and uptake by, neutrophils also increased with increasing CN (Fig. 3, C and D). Reduced neutrophil function was not associated with reduced expression of other FcγRs; indeed, a slight increase in expression of FcγRI and FcγRIIa was observed in the FcγRIIIb-deficient proband (Fig. S4). There was also a correlation between CN, as measured by qPCR, and soluble circulating FcγRIIIb (Fig. 3 E), which is released upon neutrophil activation (15).

We wished to determine if this association of FCGR3B CN with expression and function was seen in normal populations. We used qPCR to determine CN in 33 healthy individuals. Although we have shown in the family, and by comparison with the HapMap data, that FCGR3B CN is proportional to results from the qPCR assay, definitive assignment of CN in the general population is difficult (7, 8). We therefore selected individuals whose FCGR3B/CD36 ratios were clearly low (lowest five values), intermediate, or high (highest five values), and using neutrophils from these individuals showed that CNV was associated with FcγRIIIb expression (Fig. 4, A and B, and Fig. S5, available). As in the index family, there was a clear correlation between CN and neutrophil adherence to immunoglobulin-coated surfaces under flow conditions, with high CN individuals having four times the rate of adherence of those with low FCGR3B CN (Fig. 4, C and D, and Supplemental videos). Once adherence had occurred, spreading of the neutrophils on the surface appeared independent of FCGR3B CN (Table S1). Immune complex uptake was also reduced in those with low CN (Fig. 4 E), whether or not complement was present (not depicted). As in the family, levels of circulating soluble FcγRIIIb correlated with FCGR3B CN as determined by qPCR (Fig. S6 A). Thus, the association between FCGR3B CN, protein expression, and function seen in the index family could be extrapolated to the population at large.

SLE has long been known to be associated with defects in immune complex clearance, and SLE lesions are invariably associated with immune complex deposition (32). The functional data presented here, in both an FcγRIIIb-deficient individual and her family and in normal volunteers with different CN, show clearly that reduced expression of FcγRIIIb is associated with reduced immune complex uptake and with reduced neutrophil adhesion to immune complex–bearing surfaces. We have also confirmed a genetic association between SLE and low CN. Our data would suggest that the mechanism underlying this association is a failure of neutrophil trafficking to inflammatory lesions and reduced ability to ingest immune complexes once there, thereby reducing immune complex clearance and predisposing to SLE. Hence “normal” levels of FcγRIIIb (both neutrophil-surface and soluble) may be important in protection against SLE. Our results thus provide a mechanism explaining the genetic association of low FCGR3B CN and SLE.

Unlike SLE, AASV is not associated with immune complex deposition. Rather, whereas neutrophils are not thought to play a major role in inflammatory damage in SLE, it is clear that Fc receptor-mediated neutrophil activation is important in AASV pathogenesis (33). Thus, it might be expected that increased FCGR3B CN would be associated with AASV. We therefore analyzed FCGR3B CN in 286 ethnically matched controls and 556 patients derived from the following 3 independent AASV cohorts: a UK cohort with biopsy-proven, ANCA-associated renal vasculitis (cohort 1, n = 347), an independent ANCA-associated vasculitis cohort (cohort 2, n = 136), and a vasculitis cohort from Birmingham, UK (cohort 3, n = 73). Further details of these cohorts are described in the online supplemental material. Cases and controls were plated together and compared using a Student's t test as described above and in the Materials and methods. No association between disease and low CN was seen in any of the three cohorts; in fact, AASV was associated with high CN when cases from all cohorts were compared with controls (P = 10⁻⁸ calculated using a Student's t test stratified by plate of samples; Fig. 5, A–C, and Fig. S1). To substantiate the observation that FCGR3B CN distribution differed between SLE and AASV, we compared samples from both groups directly on a single plate (Fig. 5 D). Patients with SLE showed a significantly lower mean FCGR3B/CD36 ratio than patients with AASV (P = 0.0028; Fig. 5 D). We then measured levels of soluble circulating FcγRIIIb in patients with SLE and AASV at both disease onset and after therapy (patients described in the Supplemental clinical information and Table S2, available). In the 43 patients with AASV, soluble FcγRIIIb levels at diagnosis correlated with FCGR3B/CD36 ratio, and levels were significantly higher than both controls and the 15 lupus patients (Fig. S6, B and C).

An early report suggested that low, rather than high, FCGR3B CN is associated with AASV (26). The most likely reason for this is that that study compared groups after CN assignation, despite the continuous nature of the data. This is known to risk producing misleading results (7, 8), is likely to explain the variation in CN distribution observed between ethnically similar control groups (26), and may be part of the reason why low FCGR3B CN was found to be associated with lupus nephritis alone in one study (5), but not others ([26] and this study). To ensure the accuracy of our data, we used three independent cohorts (including one overlapping with [26]) and larger numbers of patients, and we compared raw values using a stratified Student's t test rather than arbitrary “binning” (as discussed in the Materials and methods). In addition, we correlated results with a family of known FCGR3B CN and with DNA from individuals with CN determined independently from the HapMap data. Finally, we correlated our assay with FcγRIIIb surface expression and function (Fig. 4, A and B, and Fig. S5), and with soluble FcγRIIIb titers (Fig. S6). We are therefore confident AASV is associated with increased FCGR3B CN. Nonetheless, all of the results in this field will need to be reassessed as more sophisticated CN assays are developed in the future.

The associations of high FCGR3B CN with AASV and low CN with SLE is consistent with what is known of the pathophysiology of these diseases. In AASV, unlike SLE, inflammatory lesions are characterized by neutrophil infiltration, but not immune complex deposition (33). ANCA is thought to mediate its pathogenic effect by activating neutrophils in an Fcγ receptor–dependent fashion (34) and should therefore be exacerbated by increased FCGR3B CN. In addition, AASV is associated with FCGR3B*1, the isoform of FcγRIIIb that confers increased phagocytosis, (16, 18), whereas SLE is associated with FCGR3B*2, which is associated with reduced phagocytosis (16, 30, 35), providing further evidence that the receptor plays a different role in each disease.

CNV is an important cause of genetic variation in populations, and has recently been implicated in complex disease causation. The known associations with disease will increase as whole-genome association studies are analyzed for CNV. These associations will need to be investigated to determine the mechanism by which CNV contributes to disease. In this study, we use a family with FCGR3B deficiency to demonstrate not only the physiological function of the receptor, but also the effect of FCGR3B CNV on protein expression and function. By extending these studies into the general population, we show for the first time the association between CNV, expression, and function of a disease-related gene, providing a functional explanation for the genetic association between low FCGR3B CN and SLE.

FCGR3B CN was measured using qPCR modified from that previously described (5), with the exception that FCGR3B values were normalized to CD36 rather than FOXP2. In brief, 10 μl PCR reactions, using 2.5 ng of genomic DNA as template, were performed using SYBR green (Quantifast; QIAGEN) and an ABI 7900HT real-time PCR system with a 384-well module (ABI). Cycling conditions were 95°C for 5 min, and then 40 cycles of 95°C for 10 s, followed by 60°C for 30 s. All samples were measured in duplicate, and for each sample the difference between replicate Ct values was determined. The mean difference across all samples was then determined, and any samples with a difference in values exceeding the mean plus two standard deviations were removed from the dataset.

The number of cycles required for amplification of FCGR3B to reach threshold (the Ct value) was then compared with the number of cycles required for amplification of CD36, a gene that does not show CNV. The ratio of FCGR3B Ct/CD36 Ct is therefore a measure of FCGR3B CN. Two difficulties posed by the assay had to be circumvented as follows. First, we found that the FCGR3B/CD36 ratios were not integers, as might be expected from the principle of CNV. This is a well recognized problem with CN assays (7). Indeed, when we downloaded array comparative hybridization data for probes encompassing the FCGR3B locus (from 270 HapMap individuals), we saw a spread of log intensities similar to the pattern seen using the qPCR assay (Fig. S7, available). We obtained DNA from 14 of these individuals from the Coriell Institute (Camden, NJ), and found a close correlation between aCGH data and qPCR ratios of FCGR3B/CD36 (Fig. S8 B). The second potential problem we found was that, although the distribution of FCGR3B/CD36 ratios for a given set of samples is remarkably constant upon repeat measurement, the absolute values may vary on repetition (Fig. S8 B). We therefore plated control and diseased samples together. We compared cases and controls on each plate using a Student's t test, and then performed a Student's t test stratified by both plates of samples. We were able to do this because the means of the FCGR3B/CD36 ratios for SLE cases were lower than control means on both plates of samples used, and the means of the AASV samples were greater than controls on all seven plates assayed. Fig. S1 shows the raw data for all plates of SLE and AASV cases and controls.

Blood samples were obtained from 15 SLE patients with active disease and 10 age- and sex-matched normal controls. Neutrophils and monocytes were isolated by magnetic cell sorting, total RNA was extracted, and microarray hybridizations were performed as previously described (36), except that pooled targets were resuspended in 60 μl hybridization buffer and hybridized on a SlideBooster hybridization station (Advalytix). All samples were hybridized to custom spotted oligonucleotide microarrays in duplicate, using a dye-swap strategy, against a common reference RNA prepared from pooled PBMC RNA obtained from seven normal controls. Raw image data were extracted using Koadarray v2.4 (Koada Technology); probes were listed as present if they had a spot confidence value >0.3 in at least 1 channel. Background subtracted intensity values for all probes considered present were imported into GeneSpring v7.2 and Lowess normalized before further analysis. Microarray data have been deposited in ArrayExpress with the accession no. E-TABM-463.

DNA was extracted from peripheral whole blood using the QIAamp Mini blood kit (QIAGEN) as per the manufacturers' protocol. For gene-specific PCRs, optimum specificity was obtained by performing annealing temperature gradients on each primer pair. All PCRs were performed in 1× Promega PCR buffer (including 1.5 mM MgCl₂) with Taq polymerase. Gene specificity was confirmed by sequencing of all amplicons. Primer sequences are shown in Table S3.

Whole blood was diluted in PBS with Ca²⁺/Mg²⁺ to give a neutrophil concentration of 10⁶ cells/ml. To make the immune complexes, Alexafluor-labeled ovalbumin (Invitrogen) was opsonized with rabbit polyclonal antiovalbumin antibody (Sigma-Aldrich) at 37°C for 1 h. Neutrophils were incubated with ovalbumin, in opsonized or nonopsonized form, for 10 min, at either 37°C or 4°C. Erythrocytes were then lysed with BD FACS Lysing Solution. The percentage of neutrophils positive for Alexafluor-labeled ovalbumin was assessed by flow cytometry, as described previously.

Neutrophils were isolated from whole blood by two-step gradient centrifugation as previously described (37). The adhesion of flowing neutrophils to IgG-coated microslides and the subsequent behavior of the adherent cells was measured as previously described (37).

Respiratory burst activity was determined by means of the superoxide dismutase–inhibitable reduction of cytochrome c, as previously described (38).

The antibodies used in this study were: anti-FcγRIIIb (clone 3G8; BD Biosciences), anti-FcγRIIa (clone FLI8.26; BD Biosciences), anti-FcγRIIb (MacroGenics) (39), anti-FcγRI (clone 10.1; Invitrogen), anti-CD56 (clone MEM88; Invitrogen), and anti-CD59 (clone Mem-43; Invitrogen). Isotype controls were IgG2a (clone G155-178), IgG1κ (clones MOPC-21), and IgG2bκ (clone MPC-11). 100-μl aliquots of whole blood were incubated with 10 μl Fc block (Miltenyi Biotec) and 10 μl of appropriate labeled antibodies for 20 min at room temperature. Erythrocytes were lysed using BD FACS Lysing Solution, and surface expression assessed by flow cytometry on a FACSCalibur (BD Biosciences). Data were analyzed using FlowJo software (Tree Star, Inc.).

Microtitre plates (MaxiSorp; Nunc) were coated with 0.3 μg/ml anti-CD16 mAb (clone DJ130c; Dako) overnight at 4°C. Plates were washed three times with PBS/0.05% Tween-20 and blocked for 2 h at room temperature with PBS/10% FCS. Serum samples were diluted in PBS, added to the plate, and incubated for 4 h at room temperature. Plates were washed and incubated for 2 h at room temperature with biotinylated anti-CD16 mAb (clone 3G8). After a further wash, plates were revealed with streptavidin-HRP followed by TMB (Sigma-Aldrich) according to manufacturer's instructions. Absorbance at 450 nm was measured after 15 min on an OPTImax tunable microplate reader.

The statistical significance of data characterizing the function of FcγRIIIb was determined using either a one- or two-way analysis of variance test. When comparing FCGR3B/CD36 ratios between cases and controls on an individual plate, an unpaired t test with Welch's correction was performed. All statistical tests were performed using either the Prism software (GraphPad) or the R statistical system www.r-project.org). P values <0.05 were considered significant. Further details regarding individual statistical tests performed can be found in the appropriate figure legend.

This study was approved by the Cambridge Local Research Ethics Committee, the Leeds East Research Ethics Committee, the Oxford Multi-center Research Ethics Committee, and the Institutional Review Board of the University of Hong Kong/Hospital Authority.

The Supplemental materials and methods contain additional methodology. The Supplemental clinical information contains clinical details for the FCGR3B-deficient individual and her family, as well as clinical information on the SLE patients whose microarray data are shown in Fig. 1, and the AASV patients whose soluble FcγRIIIb levels are shown in Fig. S6. Table S1 shows the percentage of cells that undergo spreading after adhesion is unrelated to FCGR3B CN. Table S2 is a summary of clinical features of SLE and AASV patients whose soluble FcγRIIIb levels are shown in Fig. S6. Table S3 shows gene-specific primer sequences used for Fig. 1 E. Fig. S1 shows qPCR raw data for all plates of samples analyzed. Fig. S2 shows that, in a Chinese population from Hong Kong, FCGR3B CN does not differ significantly in patients with SLE nephritis compared with healthy controls. Fig. S3 shows that FCGR3B CN does not differ significantly in healthy Chinese individuals from Hong Kong compared with UK Caucasians. Fig. S4 Cell surface expression by flow cytometry of FcγRIIa (A), FcγRI (B) and FcγRIIb (C) on neutrophils and FcγRIIIa (D) on monocytes of patient A (in red) and two individuals known to have FCGR3B. Fig. S5 shows that the finding that FCGR3B CN is proportional to gene expression is reproducible. Fig. S6 shows correlation of FCGR3B CN and soluble FcγRIIIb. Fig. S7 shows array comparative genomic hybridization data across the FCGR3B locus on chromosome 1q22-23. Fig. S8 shows the FCGR3B/CD36 ratio detected by qPCR correlates with aCGH data and is reproducible. Videos 1–4 show neutrophil adhesion and spreading to IgG-coated glass slides during 4 min of flow at a shear stress of 0.1 Pa. Videos are shown for representative individuals with low, intermediate, and high FCGR3B CN. The online version of this article is available.

We thank Dr. Tom Freeman for contribution to array development; Drs. Kevin Harris, David Jayne, and Afzal Chaudhry for clinical input; and Dr. Chaudhry for statistical advice. We are grateful to all the patients involved in the study, but in particular for the generous contribution to this research made by our FcγRIIIb-deficient patient and her family. We also acknowledge Dr. Dawn L. Cooper's contribution to the flow cytometry data.

K.G.C. Smith is supported by a Wellcome Trust Clinician Research Leave Award and is a Lister Prize Fellow, L.C. Willcocks is a Medical Research Council Clinical Training Fellow, M.R. Clatworthy is a Wellcome Trust Intermediate Fellow, and additional support was provided by the Medical Research Council (MRC), Kidney Research UK (KRUK), the Genzyme Renal Innovations Programme, and the national Institute for Health Research Cambridge Biomedical Research Centre. A.W. Morgan and J.I. Robinson are supported by the Arthritis Research Campaign. We acknowledge the MRC/KRUK National DNA Bank for Glomerulonephritis and Caroline Savage, Lavanya Kamesh, and Lorraine Harper, for contributions to DNA cohorts. MRC/KRUK DNA bank sample management was undertaken by the UK DNA Banking Archive Network, funded by the MRC at the Centre for Integrated Genomic Medical Research, University of Manchester. HapMap DNA was supplied by the Coriell Institute for Medical Research.

The authors have no conflicting financial interests.