Abstract
Hemolytic uremic syndrome (HUS) is characterized by the triad of thrombocytopenia, microangiopathic hemolytic anemia, and acute renal failure. The non–Shiga toxin–associated HUS (atypical HUS [aHUS]) has been shown to be a disease of complement dysregulation. Mutations in the plasma complement regulators factor H and factor I and the widely expressed membrane cofactor protein (MCP; CD46) have been described recently. This study looked for MCP mutations in a panel of 120 patients with aHUS. In this cohort, approximately 10% of patients with aHUS (11 patients; nine pedigrees) have mutations in MCP. The onset typically was in early childhood. Unlike patients with factor I or factor H mutations, most of the patients do not develop end-stage renal failure after aHUS. The majority of patients have a mutation that causes reduced MCP surface expression. A small proportion expressed normal levels of a dysfunctional protein. As in other studies, incomplete penetrance is shown, suggesting that MCP is a predisposing factor rather than a direct causal factor. The low level of recurrence of aHUS in transplantation in patients with MCP mutations is confirmed, and the first MCP null individuals are described. This study confirms the association between MCP deficiency and aHUS and further establishes that a deficiency in complement regulation, specifically cofactor activity, predisposes to severe thrombotic microangiopathy in the renal vasculature.
Hemolytic uremic syndrome (HUS) is a microvascular thrombotic disorder that is characterized by microangiopathic hemolytic anemia, thrombocytopenia, and acute renal failure (1,2). Typical HUS occurs mostly in children and is preceded by a diarrheal illness, usually an infection with Escherichia coli O157:H7 (3). Less frequently, HUS is not preceded by diarrhea. This form, known as idiopathic or atypical HUS (aHUS), often has relapses and a poorer outcome. aHUS may be either sporadic or familial (4).
Recently, a predisposition to aHUS was associated with inherited deficiencies in complement regulatory proteins; namely, the plasma protein factor H (CFH), the plasma serine protease factor I (IF), and the widely expressed membrane cofactor protein (MCP; CD46) (5–12). Several series have reported CFH mutations in up to 30% of patients with aHUS (13,14). A few mutations also have been reported in MCP (8,11,15) and IF (6,7,15). These inhibitors regulate the complement cascade at the steps of C3 and C5 activation, and mutations in these three proteins impair the cell’s ability to protect itself against complement activation (16,17). These cases implicate inadequate control of the complement system in the pathophysiology of aHUS.
MCP is a single-chain transmembrane glycoprotein that is expressed on most human cells (18). MCP binds C3b and C4b that are deposited on the cell surface and serves as a cofactor for IF, thereby promoting C3b/C4b inactivation through limited cleavage. IF, a serine protease, cleaves the α chain of C3b and C4b but only in the presence of a cofactor protein. The resulting fragments cannot form a C3 convertase (19).
In this study, we describe a cohort of patients with aHUS and mutations in MCP. In most cases, these were heterozygous mutations that resulted in reduced levels of MCP. This study of aHUS cases also describes, for the first time, complete MCP deficiency in two patients. These data further strengthen the association between aHUS and deficient complement regulation.
Materials and Methods
Clinical Screening
We studied 120 patients with aHUS. Informed consent was obtained. Twelve of these patients had mutations in factor H (20), three had antibodies to factor H (21), and three had mutations in factor I (6). Patients were recruited from pediatric and adult nephrology departments from 2000 to 2005. We describe from this series 11 patients with MCP mutations.
Assays for Complement Components and Regulators
Investigations were performed at the Laboratoire d’Immunologie Biologique of Hôpital Européen Georges Pompidou, Paris, a reference laboratory for investigating the complement system. EDTA plasma samples were obtained. Plasma concentrations of CFH and IF were measured by ELISA, and C4, C3, and factor B were measured by nephelometry (Dade Behring, Paris La Defense, France) (6,20).
Membrane expression of complement regulators was analyzed on granulocytes from patients and control subjects using a FACSCalibur flow cytometer (Becton-Dickinson, Heidelberg, Germany). Cells were stained with saturating concentrations of biotinylated antibodies against MCP (clones GB24 and TRA-2-10) (22) and phycoerythrin-conjugated antibodies against the following: MCP (clone MEM 258; Serotec, Oxford, UK), CD55, and CD59 (clones IA 10 and P282; BD Pharmingen, San Diego, CA). Isotypic mouse IgG1 phycoerythrin (Serotec) or biotinylated MOPC-21 (IgG1; Cappel, West Chester, PA) were used as controls.
We analyzed the MCP expression on granulocytes for 21 healthy donors, 20 patients who were on renal dialysis, and 19 patients with chronic renal failure. MCP expression in those with renal abnormalities was within the range established for the normal donors. The expression of MCP also was assessed by Western blotting of NP-40 solubilized peripheral blood mononuclear cells (PBMC) (11,22).
Genetic Screening
Genomic DNA was isolated from PBMC using the proteinase K/phenol method (23). Coding sequences of the MCP gene were amplified with primers that flanked all 14 exons (Table 1), purified using the Multiscreen plates (Millipore, Molsheim, France), and sequenced with 96 capillary Sequencer 3700 using the dye terminator method (Applied Biosystems, Courtaboeuf, France). For genomic DNA and cDNA sequences, the first nucleotide A of the initiator ATG codon is denoted nucleotide +1. The amino acid number begins at the first amino acid of the mature protein.
Primers for direct sequencing of each exon of the MCP genea
Total RNA was extracted from the leukocytes of patients 3 and 5 and isolated using the QIAamp RNA Blood Mini kit (Qiagen, Hilden, Germany). A mixture of oligo (dT) primers and random hexamers were used to synthesize the first strand of cDNA from leukocyte-derived total RNA, according to the manufacturer’s (Promega, Madison, WI) protocols. Oligonucleotide primers for amplification and sequencing of the part of the hMCP cDNA that encompassed exons 1 through 4 were as follows: MCP-1f (5′-TGTTGCTGCTGTACTCCTTCT-3′) and MCP-4r (5′-aggatcagtagcaatttggag-3′). With these primers, amplification of cDNA that contained the wild-type allele is expected to result in a 379-bp fragment.
Mutagenesis, Expression, and Functional Assessment
Substitutions were generated using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). MCP isoform BC1 (GenBank accession no. X59405) (18) was cloned into the EcoRI site of plasmid pcDNA3.1+ (Invitrogen, Carlsbad, CA) and sequenced for confirmation (24). CHO K1 cells were transfected using Lipofectamine (Invitrogen). MCP expression was evaluated by ELISA, Western blotting and FACS (25). Ligand binding ELISA and cofactor assays were performed as described previously (22,25). In short, purified C3b and C4b were coated on microtiter wells, and diluted lysates were added in low-ionic-strength buffer. MCP detection was by antibody. For cofactor activity, biotinylated ligands were treated with samples that contained MCP and factor I. SDS-PAGE was used to monitor the generation of cleavage fragments.
Case Reports
All patients were white. Patients 1 through 7 represented sporadic cases, whereas patients 8 through 11 were from two families. Clinical features are summarized in Table 2.
Clinical features associated with MCP mutations in aHUSa
Patient 1.
This patient first presented with aHUS at age 27. She has had three episodes of aHUS that required dialysis and plasma exchange and on each occasion regained renal function. Her creatinine clearance currently is 34 ml/min, and she has hypertension. She was born with the classic features of Pierre-Robin syndrome and also has common variable immunodeficiency. There is no family history of renal disease.
Patient 2.
At age 40, this patient developed ESRD secondary to recurrent aHUS. He had 10 documented episodes during 35 yr of follow-up.
Patient 3.
Disease onset occurred at age 2. She initially had complete recovery of renal function. Thereafter, she had two recurrences at ages 3 and 14. She now is 25 yr old with normal renal function (creatinine clearance 80 ml/min) but persistent proteinuria (2.66 g/24 h).
Patient 4.
This patient presented at age 6 with proteinuria and hypertension associated with a “membranous glomerulopathy.” He developed aHUS at 10 yr of age with recovery of renal function. He had another aHUS episode 2 yr later, which led to impairment of renal function that was severe enough to require hemodialysis. At age 15, he received a cadaveric renal transplant, which failed after 4 yr as a result of chronic rejection.
Patient 5.
This 23-yr-old woman was referred with aHUS 3 mo after a second childbirth. She received plasma exchange therapy but did not recover. She is on hemodialysis.
Patient 6.
At 8 mo of age, this girl presented with aHUS, and there was no recovery of renal function. Subsequently, she received three cadaveric kidney transplants. The first graft, at 4 yr of age, was lost a few days after transplantation with a renal biopsy suggesting either vascular rejection or recurrence of aHUS. Microangiopathic hemolytic anemia and thrombocytopenia suggested that this loss of renal function was due to recurrent aHUS. A second transplant at 6 yr of age initially functioned normally, but after several months, graft function deteriorated and the graft was lost. Transplant biopsy did not show evidence of recurrent aHUS. A third graft, at 10 yr of age, was lost in the first week as a result of acute humoral rejection and an ensuing intrarenal hemorrhage.
Patient 7.
This 13-yr-old girl presented with four recurrences of aHUS. The first episode was at age 5. She has regained normal renal function after each episode.
Patients 8 and 9.
In this family, three children presented with aHUS at an early age. The son presented with severe aHUS that led to death at 9 mo of age. His two sisters had relapsing aHUS (four to six relapses each) beginning at ages 3 and 5. Kidney transplantation was performed at age 17 in patient 8. The kidney transplant failed 4 yr later. There was no evidence of recurrent aHUS. She remains on hemodialysis at age 21. The other child has chronic renal failure (creatinine clearance and level of proteinuria are not available).
Patients 10 and 11.
In this family, two children are affected. The son (patient 10) presented at 18 mo of age with HUS after an E. coli O157:H7 infection. The child had a complete recovery but then had three further episodes of aHUS during the next 2 yr. He was treated successfully with plasma exchange therapy. His sister presented at 8 yr of age with aHUS that resolved spontaneously.
Results
Complement Component Assessment
Complement component profiles are presented in Table 3. In eight (72%) of 11 patients, C3 and factor B levels were in the normal range at the time of investigation. Patient 6 exhibited low C3 and factor B levels, consistent with alternative pathway activation. Her healthy parents had normal plasma levels of C3 and factor B. Patients 5 and 9 had borderline low C3 and factor B levels. Concentration of CFH and IF was in the normal range for all patients.
Complement profile of the 11 patients with aHUS and MCP mutationsa
Molecular Characterization of MCP Mutations
Direct sequencing of the MCP gene included all exons and their flanking regions (Table 4, Figure 1). Five patients (four pedigrees) had a single mutation, and six patients (five pedigrees) had two mutations. Of the latter group, three were homozygous for the same mutation, one had two mutations on the same allele, and two siblings were compound heterozygotes for two mutations. None of the detected mutations was found in 100 normal individuals. In the patients with MCP mutations, mutations were not detected in the 23 exons of the CFH gene and the 13 exons of the IF gene.
Genetic studies of membrane cofactor protein (MCP). The approximately 45-kb gene encoding MCP is in the regulators of complement activation gene cluster at 1q32. The 14 exons of the MCP gene are shown as boxes, and the introns are shown as a solid line. The position of each mutation is shown. MCP possesses four complement control protein (CCP) domains, which are encoded by five exons (two exons encode CCP 2). Exons 7, 8, and 9 encode short peptides that contain O-linked sugars that are alternatively spliced. Exon 10 encodes a 13–amino acid juxtamembranous peptide of unknown function. Exons 11 and 12 encode the transmembrane domain and cytoplasmic anchor. Exons 13 and 14 encode the alternatively spliced cytoplasmic tails of 16 and 23 amino acids, respectively.
Genetic analysis of patients with aHUSa
Several types of mutations were identified. Three patients had nonsense mutations, and three patients had a mutation in the invariant dinucleotide of a splice site. The two compound heterozygous siblings (patients 10 and 11) had both a splice site mutation and a missense mutation in exon 5 (Y155D). Patient 7 as well as her healthy father and brother were heterozygous for an allele that carries the substitutions D151N and Y155D. Other missense mutations in complement control protein (CCP) domain 3 (E145Q and G162R) were identified in two pedigrees (patient 6 and patients 8 and 9).
We identified a heterozygous A to G mutation at position −2 of exon 3 (IVS2 −2A>G). Reverse transcription–PCR using exons 1 and 4 primers of the MCP transcript revealed a 276-bp fragment (Figure 2B). Direct sequencing of this fragment after gel purification revealed exon 3 skipping. This resulted in the deletion of amino acids 62 to 95 with three amino acid changes (G96I, Y97I, and Y98T) and a premature stop codon at amino acid 99. Using a similar approach, we demonstrated that the IVS2 +2T>G mutation also results in abnormal splicing. A 235-bp fragment corresponded to the first 45 bp from exon 2 spliced onto exon 3 (Figure 2A). This resulted in the deletion of 144 bp and 48 amino acids in phase with the wild-type sequence.
Analysis of the splice mutations for patients 3 and 5. Reverse transcription–PCR (RT-PCR) products were run on a 2% agarose gel and ethidium bromide stained. (A) Effect of IVS2 +2T>G on exon splicing. RT-PCR from wild-type control (lane 1) generated a single band of 379 bp. Sequencing of this product confirmed this to be exons 2, 3, and 4. RT-PCR from patient 3 with a homozygous IVS2 +2T>G mutation (lane 2) generated a 235-bp band that was sequencing and revealed a 144-bp deletion of the 3′ end of exon 2. (B) Effect of IVS2 −2A>G mutations on exon splicing. Lane 1 again contains a wild-type control RT-PCR product of 379 bp that contains exons 2, 3, and 4. Lane 2 from patient 5 with a heterozygous IVS2 −2A>G mutation gave two RT-PCR products of 379 and 276 bp. These products were sequenced. The 379-bp product was normally spliced exons 2, 3, and 4. The 276-bp product lacked exon 3, confirming exon splicing.
The IVS7 −2A>G splice site mutation was observed only in the setting of the Y155D mutation. Because the latter is not expressed, the protein observed is derived from the other allele bearing the IVS7 −2A>G mutation. On Western blotting of cells from patients 10 and 11, the protein is expressed with an Mr consistent with the wild-type isoforms MCP-C1 or MCP-C2 (26). In these isoforms, exon 8 is spliced out, and this exon also would be predicted to be spliced out in this mutation.
Membrane Expression of MCP in Patients with Mutations
Flow cytometry of PBMC was used to assess the expression of the three membrane complement regulators: Decay accelerating factor (CD55), MCP (CD46), and the inhibitor of the membrane attack complex (CD59) (Table 3). Granulocytes were chosen because this is an easily purified and relatively homogenous cell population. Expression of CD55 and CD59 did not differ between the patients and normal donors.
Figure 3 shows the MCP FACS profile for the 11 patients. No staining was observed on the granulocytes of patients 1 and 2, who were homozygous for the R25X and Y214X mutations, respectively. The binding of two other mAb to these patients’ granulocytes also was tested: GB24, which blocks C3b and C4b binding and interacts with an epitope that requires CCP 3 and 4, and TRA-2-10, which binds CCP 1 and does not block function (22). Neither TRA-2-10 nor GB24 stained these patients’ granulocytes. Patient 3 with homozygous splice site substitution exhibited reduced expression of MCP. Except for patient 6, the heterozygous patients expressed approximately 50% of the normal level of MCP. Patient 6 and her homozygous father expressed high levels of MCP on both their granulocytes (Figure 3) and lymphocytes (data not shown). Western blotting of PBMC confirmed the lack of MCP expression in patients 1 and 2 and markedly reduced expression in patient 3 (data not shown).
Analysis of MCP expression on granulocytes of patients with atypical hemolytic uremic syndrome (aHUS). Background binding is indicated by the dotted line. Mean fluorescence intensity (MFI) is given in the upper right hand corner. FACS analysis was performed the day after venipuncture except in the case of patients 6, 8, and 9, whose samples arrived 3 d after venipuncture. To control for this, patient 5 was studied on four consecutive days after venipuncture, and there was no modulation of MCP expression during this trial period (MFI on day 0, 371; day 1, 428; day 2, 450; day 3, 409; day 4, 371).
Functional and Structural Studies in Cell Lines
The four missense mutations are in CCP 3 (Figure 1). CCP 2, 3, and 4 are required for cofactor activity (22). These mutations in CCP 3 were analyzed using transfected CHO cells. The Western blot of these cell lysates demonstrates that MCP that carry mutations E145Q, D151N, and G162R are expressed with the expected Mr (Figure 4). In this process, cell lines are selected on the basis of FACS expression; consequently, they are not necessarily representative of the surface expression profile in vivo. In the case of Y155D, the ratio of precursor to mature protein is increased to approximately 1:1. Normally, the precursor of MCP is approximately 1% of the mature form; therefore, unless the blot is overexposed, it usually is not observed (Figure 4, lane 6). On the basis of structural modeling, the Y155D substitution places a charged amino acid in the hydrophobic core of the protein (22). Retention of the precursor form in the endoplasmic reticulum is indicative of a misfolded protein. Further evidence for a structural abnormality is the apparent aggregates at approximately 180 kD. Precursor forms also are increased in the case of mutant G162R, so this also likely is an aberrantly folded protein.
Western blot of CHO cell lysates that were probed with a rabbit polyclonal antibody to MCP. Lane 1 contains the size markers. Lane 7 is CHO cell line not expressing MCP. Lane 6 shows the phenotype of wild-type MCP (BC1 isoform) as expressed by transfected CHO cells. Lanes 2 to 5 are the mutations in wild-type MCP identified in the patients with HUS. MCP that carried mutations E145Q (lane 2), D151N (lane 3), and G162R (lane 5) are expressed with the expected electrophoretic mobility. In the case of Y155D (lane 4) and G162R (lane 5), the ratio of precursor to mature protein is increased, likely indicating a folding problem. These two mutants had decreased surface expression (Table 3, Figure 2) on granulocytes. These cell lines were selected by FACS, so their surface expression, especially in the cases of Y155D and G162R, are not necessarily representative of their in vivo expression pattern. M, mature protein; Pro, precursor protein.
Cell lysates also were used to analyze the functional profile of the mutated proteins (Table 5). Both E145Q and G162R mutations have similar profiles. They are comparable to wild-type MCP relative to their binding to C3b and C4b and cofactor activity for C3b. E145Q, however, has approximately 50% of expected C4b cofactor activity, whereas G162R has no detectable C4b cofactor activity. Because G162R binds C4b normally, this is a mutation that selectively alters protease cleavage of C4b by factor I. Mutations with this profile have been described in CR1 and CD46 (22,27). Although these in vitro data implicate the classical pathway in aHUS pathogenesis, this may not reflect the in vivo setting. Patients 8 and 9, bearing the G162R mutation, have low expression levels (Table 3), consistent with minimal or no surface expression of this protein. This lack of expression will result in a defect in both classical and alternative pathway regulation. E145Q is highly expressed, which could compensate for its reduced function. Further studies on expression, function, and membrane stability of mutants E145Q and G162R are in progress.
Expression and functional assessment of MCP mutants in expressed CHO cell lines
The Y155D mutant has no detectable C3b or C4b binding activity and negligible cofactor activity. D151N lacks C4b binding capability and, as would be expected, has no C4b cofactor activity. However, because it was observed only in the setting of the Y155D mutation, its contribution in vivo to deficient complement regulation is not possible to ascertain.
Discussion
We describe nine novel MCP mutations in 11 patients (nine pedigrees) who presented with aHUS. Our findings, together with three recent reports (8,11,15), confirm mutations in MCP as a predisposing factor for aHUS.
In this cohort, the onset of aHUS typically was in early childhood, although two patients presented in their mid-20s. Unlike patients with IF or CFH mutations, most of the patients whom we describe did not develop end-stage renal failure after aHUS. Only 18% of the patients developed end-stage renal failure after a single episode of aHUS, and 55% remain dialysis free despite, in many cases, numerous recurrences. In the report by Richards et al. (11), four of seven patients with aHUS and MCP mutations remain dialysis independent.
The risk for recurrent aHUS after renal transplantation is high (28,29). However, aHUS comprises a heterogeneous mix of predisposing factors. When these factors are examined separately, two distinct groups emerge. To date, approximately 75% of patients who have CFH mutations and have received a transplant have had recurrent disease (30). Similarly, all four renal transplants in patients with IF mutations have had recurrent disease (6,7). In contrast, in four patients with MCP mutations, there has been no aHUS recurrence in the allograft (8,11). This would be expected because MCP is a transmembrane regulator and allografts therefore will be protected by wild-type MCP.
In our study, only one renal transplant of five had recurrent aHUS. Functional studies of this patient’s mutation (E145Q) in vitro demonstrated a defect in only C4b cofactor activity. However, there was a two- to three-fold increase in surface expression of MCP, which may negate any deleterious effect of the mutation. Furthermore, despite the finding of a defect in only the classical pathway cofactor activity in vitro, the patient had low C3 levels, suggesting alternate pathway activation. It is possible that this patient has a mutation in another complement regulator, and further studies are under way. Even allowing for this complex case, in renal transplantation in aHUS with MCP mutations alone, the recurrence rate is only 11% compared with much higher rates for the serum regulators.
To be able to give an accurate assessment of the risk for recurrence in transplantation, it is necessary to screen individuals for mutations. Normal C3 levels, as in most of our patients, does not exclude the presence of mutations in complement regulatory proteins. Genetic screening is expensive and time consuming. In our study, 91% of patients had reduced MCP expression on granulocytes. Reduced expression also is commonly reported in other series (8,11,15). This suggests that FACS analysis of PBMC may provide a useful screening tool to identify MCP mutations.
An interesting corollary of this research has been the first discovery of MCP null individuals. We report two individuals with no MCP on the surface of their granulocytes. Homozygous deficiency of the most analogous molecule in mice, Crry, is lethal in utero, demonstrating the importance of controlling complement activation in the developing placenta (31). Therefore, other complement regulatory proteins may be able to compensate in humans. We have been able to examine only PBMC for DNA analysis and surface expression, so it is possible that there is expression of MCP on other cell types, although we believe that this is unlikely.
In our study, as in those of two other groups (8,11), there is incomplete penetrance. This also has been described in aHUS associated with IF (6,7) and CFH mutations (13,15,32). One explanation for this is that other genetic factors act as modifiers of disease. Caprioli et al. (32) first showed an association between CFH single-nucleotide polymorphisms and aHUS. More recently, this association was extended to include MCP single-nucleotide polymorphisms in three separate cohorts of patients (13,15). It is likely, therefore, that many complement regulatory genes in the regulators of complement activation cluster work in harmony to prevent cell damage. Only when an unfavorable network of complement regulatory genes is present in the face of endothelial injury will aHUS result. This is most elegantly illustrated by the independent segregation of three complement risk factors in aHUS (33). Only when all three risk factors were present did aHUS result.
Conclusion
We confirm the role of MCP mutations in aHUS. In our panel, approximately 10% of patients with sporadic or familial aHUS have mutations in MCP. Most of these mutations led to reduced MCP surface expression, whereas a small proportion resulted in only a functional defect. As in other studies, we show incomplete penetrance, suggesting that MCP is a predisposing factor rather than a direct causal factor. We also confirm the low level of recurrence of aHUS in transplantation in patients with MCP mutations and describe the first MCP null individuals.
Acknowledgments
This work was supported in part by grants from Assistance Publique-Hôpitaux de Paris (Progres Medical 2004 and Programme Hospitalier de Recherche Clinique [AOM05130/P051065] 2005; V.F.B.), National Kidney Research Foundation, Peel Medical Research Trust (D.K.), and National Institutes of Health grant AI37618 (J.P.A.).
Nelly Poulain, Florence Marliot, and Stephane Roncelin provided expert technical support. We are particularly grateful to the clinicians who referred their patients for complement investigations. We are extremely grateful to Prof. Jean-Pierre Grunfeld for his many contribution and careful revision of the manuscript.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
- © 2006 American Society of Nephrology