Article Text
PDF
Abstract
Background Russell Silver syndrome (RSS) leads to prenatal and postnatal growth retardation. About 55% of RSS patients present a loss-of-methylation of the paternal ICR1 domain on chromosome 11p15. CDKN1C is a cell proliferation inhibitor encoded by an imprinted gene in the 11p15 ICR2 domain. CDKN1C mutations lead to Beckwith Wiedemann syndrome (BWS, overgrowth syndrome) and in IMAGe syndrome which associates growth retardation and adrenal insufficiency. We searched for CDKN1C mutations in a cohort of clinically diagnosed RSS patients with no molecular anomaly.
Method The coding sequence and intron–exon boundaries of CDKN1C were analysed in 97 RSS patients. The impact of CDKN1C variants on the cell cycle in vitro were determined by flow cytometry. Stability of CDKN1C was studied by western immunoblotting after inhibition of translation with cycloheximide.
Results We identified the novel c.836G>[G;T] (p.Arg279Leu) mutation in a familial case of intrauterine growth retardation (IUGR) with RSS phenotype and no evidence of IMAGe. All the RSS patients inherited this mutation from their mothers (consistent with monoallelic expression from the maternal allele of the gene). A mutation of this amino acid (p.Arg279Pro) has been reported in cases of IMAGe. Functional analysis showed that Arg279Leu (RSS) did not affect the cell cycle, whereas the Arg279Pro mutation (IMAGe) led to a gain of function. Arg279Leu (RSS) led to an increased stability which could explain an increased activity of CDKN1C.
Conclusions CDKN1C mutations cause dominant maternally transmitted RSS, completing the molecular mirror with BWS. CDKN1C should be investigated in cases with family history of RSS.
- Clinical genetics
- Fetal growth
- Imprinting
- Russell Silver syndrome
- CDKN1C
Statistics from Altmetric.com
Introduction
Russell Silver syndrome (RSS) (OMIM #180 860) is a growth disorder defined by severe intrauterine growth retardation (IUGR)/children born small for gestational age (SGA), associated with at least three of the following five criteria: postnatal growth retardation, relative macrocephaly, a distinctive facial features (ie, a prominent forehead), feeding difficulties and/or low body mass index (BMI) and body asymmetry (hemihypotrophy).1
A methylation defect is identified at the 11p15 locus in about 55% of RSS patients. The 11p15 locus is organised into two domains, both of which are imprinted2: the telomeric domain contains the insulin growth-like factor 2 (IGF2) and H19 genes. The expression of these two genes is controlled by imprinting centre region 1 (ICR1), which is methylated on the paternal allele (leading to a monoallelic expression of IGF2 from the paternal allele). The centromeric domain contains the cyclin-dependent kinase inhibitor 1C (CDKN1C) and potassium voltage-gated channel, KQT-like subfamily, member 1 (KCNQ1) genes. Both genes are expressed from the maternal allele, and their expression is controlled by imprinting centre region 2 (ICR2), which is methylated on the maternal allele. A loss of methylation of the inherited paternal 11p15 ICR1 allele leads to a loss of expression of IGF2 and the biallelic expression of H19.3 About 5–10% of RSS patients present maternal uniparental disomy (UPD) of chromosome 7.1 For the remaining 40% of patients, no molecular cause is identified despite a suggestive phenotype.1 ,3 ,4 Most RSS cases are sporadic, probably occurring as a postzygotic event, with cell mosaicism. A few rare cytogenetic events, such as maternal uniparental disomy of 11p15,5 11p15 duplication,6–9 mutation/deletion in the ICR1 domain9–11 have been implicated in RSS. All but one of the reported maternal duplications of 11p15 include the centromeric domain (including CDKN1C), the remaining case being a RSS patient with a maternally inherited ICR1 duplication.11 A deletion in the maternal centromeric domain leading to severe recurrent IUGR has recently been reported, leading to a gain of expression for CDKN1C.12 No gain of methylation on the maternal allele at the ICR2 locus (which might theoretically lead to a gain of expression of the CDKN1C gene) has yet been described in this population. Obermann et al looked for CDKN1C mutations in RSS patients, but found no evidence for such a mechanism in their cohort.13
Beckwith–Wiedemann syndrome (BWS) is the clinical and molecular mirror of RSS, with children presenting overgrowth, hemihyperplasia and a high risk of paediatric tumours.2 ,4 Several groups have reported the presence of CDKN1C mutations in familial or sporadic cases of BWS.14–24 Functional studies of some of these mutations suggested that they result in a loss of protein function.25 Furthermore, CDKN1C inactivation in mouse models leads to a phenotype closely resembling BWS.26 ,27
The CDKN1C protein strongly inhibits the cyclin/CDK complexes of the G1 phase, thereby downregulating cell proliferation.28 ,29 The C-terminal part of CDKN1C contains a QT domain able to bind PCNA (figure 1A). PCNA is essential for DNA replication processes and can bind CDKN1C and other cell-cycle regulators.30 CDKN1C gain-of-function mutations have recently been identified in a very rare syndrome known as IMAGe.31 Patients with IMAGe syndrome present severe fetal and postnatal growth retardation, metaphyseal dysplasia, neonatal adrenal insufficiency and genitourinary tract malformations.32 The analysis of several familial cases of IMAGe led to the identification of five different missense mutations, all located in the PCNA-binding domain of the protein. Functional analyses of two of these missense mutations suggested a gain of function in in vivo experiments, but did not affect the cell cycle in vitro.31 From a molecular point of view, the IMAGe missense mutations impaired the binding of CDKN1C to the PCNA protein and disturbed the ubiquitination of CDKN1C, potentially enhancing CDKN1C activity.31
(A) CDKN1C structure and mutations identified in IMAGe (bottom) and RSS (top). (B) Evolutionary conservation of Arginine 279 (black arrow) in: Hs: Homo sapiens; Ma mu: Macaca mulatta; Pa: Pongo abelii, Og: Otolemur garnettii; Ss: Sus scrofa; Oa: Ovis aries; Bt: Bos taurus; Rn: Rattus norvegicus; Mu mu: Mus musculus; Dn: Danio rerio.
Patients with IMAGe have several features in common with RSS patients: IUGR and postnatal growth retardation, with a prominent forehead. However, IMAGe patients systematically display neonatal adrenal insufficiency and bone structure dysplasia31–33 which is not observed in RSS.
As RSS patients have some clinical traits similar to those observed in patients with IMAGe syndrome, and as loss-of-function mutations of CDKN1C have been described in BWS (a clinical and molecular mirror of RSS), we searched for similar gain-of-function mutations of CDKN1C in our cohort of RSS patients in whom no molecular anomalies had yet been identified. We report a novel mutation of CDKN1C affecting the PCNA-binding domain in a family with a history of RSS with no evidence of IMAGe syndrome.
Materials and methods
Ethical considerations
All patients or parents gave written informed consent for RSS analysis, in accordance with national ethics rules (authorisation no. 682, Assistance Publique—Hôpitaux de Paris). Informed consent was also obtained from the patients or guardians for the publication of the images and the clinical and molecular data.
Patients
Patients were either followed at our clinic or were referred by other clinical centres for molecular analysis due to suspected RSS. Each patient was examined by a geneticist and/or a paediatric endocrinologist. Whether the patient was referred to our Paediatric Endocrinology Unit or had been seen at another clinic, a clinical form, including extensive clinical data, was completed and clinical RSS was retained as a diagnosis on the basis of a clinical scoring system1: prenatal growth retardation (low birth weight and/or length ≤−2 SDS for gestational age) was mandatory for the diagnosis and had to be associated with at least three of the five following criteria: (1) postnatal growth retardation (height ≤−2 SDS according on French reference growth charts at 2 years of age, or the measurement for the closest time point available), (2) relative macrocephaly at birth (ie, arbitrarily defined as a head circumference at birth at least 1.5 SDS above that predicted from birth weight and/or length according to Usher and McLean charts), (3) prominent forehead during early childhood, (4) body asymmetry and (5) feeding difficulties during early childhood and/or BMI of <–2SDS at the age of about 2 years. DNA from all the selected patients was sent to the same molecular laboratory for RSS investigation. DNA was extracted from leukocytes obtained from a blood sample. For all patients, the known causes of RSS (ie, 11p15 ICR1 loss of methylation, 11p15 ICR2 duplication, maternal UPD of chromosome (7) had been ruled out, as previously described,1 ,34 and patients previously diagnosed with ICR1 loss of methylation or chromosome 7 maternal UPD were excluded from CDKN1C sequencing. Inclusion criteria for CDKN1C analysis were: clinical diagnosis of RSS based on the scoring system described above1 including cases with a familial history of RSS/growth retardation (29 out of the 97 patients had a familial history of small stature ≤−2 SDS in a parent and/or sister/brother).
Auxology
Birth length, weight and head circumference are expressed in SDS with respect to Usher and Mac Lean standard curves. Postnatal growth parameters are expressed in SDS with respect to French reference charts.
CDKN1C sequencing
The first two coding exons, and the 5′ part of the non-coding third exon of genomic CDKN1C (RefSeq NG_008022.1; NP_000067.1) was amplified by PCR with the AmpliTaq Gold 360 polymerase (Invitrogen, Life Technologies, CA) in the presence of 12% (v/v) of the 360 GC enhancer, in a Biometra T3 thermocycler. The primers used for PCR amplification are described in table 1.35 Sequencing was performed with Big Dye Terminator v3.1 technology, in a final volume of 20 µl on a Biometra T3 thermocycler, and sequences were acquired on a 96-capillary AB 3730xl DNA Analyser (Life Technologies, California, USA). Variants are described in accordance with the Human Genome Variation Society recommendations. We recorded the NM_000076.2:c.836G>T variation in the dbSNP database (http://www.ncbi.nlm.nih.gov/snp) with the subSNP ID ss748770442.
Primers for genomic CDKN1C sequencing (modified from35)
Site-directed mutagenesis
The coding sequence (CDS) of CDKN1C isoform a (NM_000076.2) was inserted into the bicistronic vector pIRES2 containing the coding sequence of the green fluorescent protein AcGFP1 fused to three nuclear localisation signal (NLS) sequences (pIRES-AcGFP1-Nuc, Clontech, Takara Bio, Japan). c.836G>T (CDKN1CR279L leading to RSS), c.836G>C (CDKN1CR279P leading to IMAGe31) and c.158T>C (CDKN1CL53P leading to BWS20) mutations were introduced into the initial construct with the Site-Directed Mutagenesis XL kit (Stratagene, California, USA), according to the manufacturer's protocol, but with the addition of 8% (v/v) DMSO because of the GC-rich content of the gene.
Cell-cycle analysis
HeLa cells were grown in a Dulbecco's modified Eagle medium (DMEM) with 10% FCS (v/v), supplemented with 1% penicillin/streptomycin (P/S). We plated 3×105 cells on a 6-wells plate and transfected them with 2 µg of plasmid in the presence of 5 µl of Lipofectamine2000 (Invitrogen, Life Technologies, California, USA) in an antibiotics-free medium, according to the manufacturer's protocol. Thirty-six hours after transfection, the cells were treated with trypsin, fixed by incubation in 1% paraformaldehyde (1 h, 4°C) and permeabilised by incubation in 70% ethanol (overnight, 4°C). Cells were stained with 10 µg/mL propidium iodide in PBS supplemented with 100 µg/mL RNase A, and were incubated for 15 min at 37°C. For each sample, we counted at least 10 000 GFP-positive cells on a BD LSRII cytometer (BD Bioscience, California, USA). We gated out doublets and debris, and then analysed the cycle distribution, including the G1 phase in particular, with the ModFit LT software (Verity Software House, Maine, USA).
CDKN1C stability
An inhibitor of the translation (cycloheximide, CHX) and an inhibitor of the proteasome (MG-132), were used for the determination of CDKN1C stability. HEK293 cells were grown in DMEM—10% FCS—1% P/S; 4.105 cells were plated on a 6-wells plate and transfected with 1 µg of plasmid in the presence of 2.5 µl of Lipofectamine 2000 for 24 h. Cells were then incubated in a medium containing CHX (100 µg/mL) during 30 min, 1 and 3 h or CHX (100 µg/mL)+MG-132 (20 µM) during 3 h. Cells were then lysed, and lysates were sonicated. An amount of 30 µg of total proteins was separated in a Bis-Tris polyacrylamide gel (NuPage Bis-Tris precast gels from Invitrogen, Life Technologies, CA) and transferred on a nitrocellulose membrane. The membrane was then incubated in a TBS-Tween-BSA (5% w/v) solution containing a monoclonal anti-CDKN1C antibody (anti p57Kip2, ab75 974, Abcam) 1/1000 overnight. For the normalisation, the membrane was then incubated in a TBS-Tween-BSA (5% w/v) solution containing the anti-GAPDH antibody (clone 6C5, sc-32233, Santa Cruz Biotechnology, California, USA) 1/5000 during 1 h at room temperature. CDKN1C and GAPDH were revealed with the Clarity Western ECL substrate (BioRad, CA). Quantification of CDKN1C and GAPDH signals was performed with the Image J software. Quantity of CDKN1C was then normalised by calculating the CDKN1C/GAPDH ratio. Results were expressed as % of the H0 quantity.
In silico analysis
We determined the likelihood of each mutation being deleterious with Polyphen (http://genetics.bwh.harvard.edu/pph2/) Align GVGD (http://agvgd.iarc.fr/) and SIFT (http://sift.jcvi.org/) web software. Alignments for conservation studies were generated with the Jalview V.2.8 software.
Statistical analysis
Quantitative data are expressed as means±SEM, and were compared in one-way ANOVA tests. Conditions were compared in pairs, in Bonferroni multiple comparison tests. A p value <0.05 was considered significant.
Results
Case report for the familial case identified (figure 2): The clinical and biological characteristics of the patients are described in table 2. The proband (IV.1) is a girl who was 5 years old at the time of CDKN1C analysis. She was born severely SGA. RSS was first suspected at the age of 4 years, because of SGA, postnatal growth failure, relative macrocephaly, low BMI and a prominent forehead during early childhood (figure 2D). Growth hormone secretion was normal. Growth hormone therapy was started at the age of 4 years, in accordance with European indications for IUGR/SGA. CDKN1C sequencing showed a heterozygous mutation at c.836G>[G;T] leading to the replacement of the arginine 279 residue with a leucine residue (p.Arg279Leu) (figure 2B). Adrenal function, assessed in an ACTH stimulation test, was normal. The proband's younger sister (IV.2) was born with normal height, weight and head circumference, is still growing normally, and did not carry the mutation. Their mother (III.6) was born with severe SGA and has a short adult height despite growth hormone treatment between the ages of 10 and 15 years. The proband's maternal aunt (III.7) was born with severe IUGR and was treated with growth hormone between the ages of 5 years and 16 years resulting in an adult height of 157 cm. Growth charts are shown in figure 2C. Patients III.6 (mother) and III.7 (aunt) were both diagnosed as RSS, and both carried the c.836G>[G;T] mutation, which was inherited from their mother (II.2). Two other members of this family (a mother and her daughter, II.1 and III.4) presented severe SGA and poor postnatal growth with a RSS diagnosis and carried the c.836G>[G;T] mutation. Subject III.10 (presenting IUGR and a low final height, but no other RSS criteria) and her mother (II.4, normal phenotype) did not carry the mutation. Thus, the mutation and the RSS phenotype were always transmitted via the maternal line. Standard X-ray examination of the proband and her mother showed no evidence of bone dysplasia (figure 2E).
Clinical presentation of the studied subjects
(A) Genealogy of the CDKN1C Arg279Leu mutation. Closed circles: RSS. Open circles/squares: not RSS. Asterisk (*): growth hormone therapy during childhood. Numbers: final height (when available). G/G or G/T: genotype of nucleotide 836. (B) Electropherogram of the CDKN1C genomic sequences of a control (top panel) and the proband (bottom panel) showing the c.836G>[G;T] variant. (C) Growth charts of the proband (IV.1) (left), her mother (III.6) (centre) and her maternal aunt (III.7) (right). (D) Images of the proband (IV.1) (left), her mother (III.6) (centre) and the proband's maternal aunt (III.7) (right). Note that prominent forehead is usually evident in early childhood and usually become less obvious in adult RSS patients. (E) X-rays of the proband (IV.1) (left) and her mother (III.6) (right).
We identified no other variants of the CDKN1C sequence in the 97 patients with clinically suspected RSS. The c.836G>T (p.Arg279Leu) variant was not described in polymorphism databases and was not present in the other 192 alleles tested.
In silico analysis
The arginine 279 residue has been strongly conserved throughout evolution as shown in figure 1B. The Arg279Leu mutation was predicted to be ‘probably damaging’ by Polyphen-2 software with a score of 0.978, and Align GVGD predicted that it would be ‘most likely to interfere’ with the protein function. SIFT predicted Arg279Leu to be ‘damaging’.
Functional analysis
The proportion of cells in the G1 phase was similar for cells transfected with the CDKN1CR279L(RSS) plasmid and those transfected with CDKN1CWT (48.16±0.25% and 48.64±1.73%, respectively). However, the proportion of cells in the G1 phase was significantly lower for cells transfected with CDKN1CL53P(BWS) than for those transfected with CDKN1CWT (39.54±0.60% and 48.64±1.73%, respectively). Conversely, the proportion of cells in G1 phase was higher for cells transfected with CDKN1CR279P(IMAGe) than for cells transfected with CDKN1CWT (56.87±1.17% and 48.64±1.73% respectively) (figure 3A).
(A) Graphical representation of the fraction of cells in G1 phase (% of total cells) for the wild type (WT), CDKN1CL53P(BWS), CDKN1CR79P(IMAGe) and CDKN1CR279L(RSS); (B) Stability of CDKN1C after inhibiton of the translation with cycloheximide. Top: representative western immunoblotting after treatment with cycloheximide±MG-132 (0 to 180: time in minutes). Bottom: quantification of three independent experiments (CHX 0 to 180) and two independent experiments (CHX+MG-132) of CDKN1CWT (white bars) and CDKN1CR279L(RSS) (black bars). Results are expressed as mean±SEM.
CDKN1C stability (figure 3B): Inhibition of the translation led to a decrease of the CDKN1CWT, whereas the CDKN1CR279L(RSS) was not affected, suggesting an increased stability of the CDKN1CR279L(RSS). Inhibition of the proteasome partially blocked the degradation of the CDKN1CWT, suggesting that CDKN1C degradation involves, at least in part, the proteasome pathway.
Discussion
We report here the first CDKN1C mutation identified in a familial case of RSS. Diagnosis of RSS in the proband and her family was based on the combination of severe IUGR/SGA with three or four positive criteria, according to a validated clinical score.1 None of them displayed body asymmetry (one of the clinical signs used for RSS diagnosis), consistent with a genetic mechanism rather than a postzygotic epigenetic defect expressed as a mosaic. In this family, the c.836G>[G;T] mutation was identified in the proband (IV.1), her mother (III.6), her maternal aunt (III.7), her grandmother (II.2) and two other female relatives (II.1 and II.4), both presenting severe growth failure and meeting the clinical criteria for RSS. Conversely, two unaffected subjects (IV.2 and II.4) with normal height and weight at birth, and a girl with SGA but not RSS (III.10), did not carry the mutation. The mutation segregated fully with the RSS phenotype, strongly implicating the Arg279Leu mutation as the cause of RSS in this family.
Beside the fact that Arginine 279 residue has been strongly conserved during evolution, further evidence for a role of this mutation in the RSS phenotype is provided by the observation that all affected patients inherited the mutation from their mothers. This finding is consistent with the imprinting of CDKN1C; methylation of the 11p15 ICR2 site on the maternally inherited allele leads to mono-allelic expression of the gene from the maternal allele, leading to a dominant, maternally transmitted, mode of inheritance for CDKN1C mutations. Women carrying CDKN1C mutations thus transmit BWS, RSS or IMAGe in 50% of their offspring in cases of loss-of-function (BWS) or gain-of-function (RSS, IMAGe) mutations. Conversely, male patients carrying CDKN1C mutations would not transmit these growth disorders (as CDKN1C is not expressed from the paternally inherited allele), but 50% of their children would carry the mutation.
BWS and RSS are clinical and molecular mirrors of each other, both involving molecular anomalies (in most patients) of the 11p15 locus. CDKN1C loss-of-function mutations have been reported in BWS patients, but no mutation has yet been identified in patients with RSS.13 Mutations of this gene have recently been reported in a rare syndrome (IMAGe) causing severe growth retardation.31 Patients with IMAGe have some traits in common with RSS patients, but they also display some specific traits, such as early adrenal insufficiency or bone structure abnormalities, not described in patients with RSS. The family affected by RSS described in this report displayed no adrenal insufficiency, and the affected individuals could not be considered to have IMAGe but were diagnosed with RSS according to a validated clinical score.1
Loss-of-function CDKN1C mutations identified in BWS are distributed along the entire length of the coding sequence of the gene.23 Conversely, the CDKN1C mutations identified in IMAGe patients affect the domain binding to the PCNA protein31 and a substitution of the arginine 279 residue by a proline residue has been identified in this syndrome. Thus, the mutation identified in RSS patients affects the same functional domain of the protein as the mutations identified in IMAGe patients, highlighting the essential nature of this region for CDKN1C function, and implicating this conserved domain in the control of fetal growth.
Functional analysis
Arboleda et al showed, in an elegant model of transgene overexpression in Drosophila, how p.Phe276Val and p.Lys278Glu could lead to a gain-of-function of the mutated protein with respect to the wild type (WT). However, they did not observe the same phenomenon in flow cytometry studies of the cell cycle.31 In our conditions, the p.Arg279Leu (RSS) mutation did not affect the cell cycle. By contrast, we tested the Arg279Pro (IMAGe) which was not tested by Arboleda et al,31 and this variant was found to affect cell cycle distribution, with a significantly higher proportion of cells in the G1 phase than for the WT. This finding is consistent with the function of the protein which acts mostly during the G1 to S transition.28 ,29 Arg279Leu may have a weaker effect on the cell cycle than Arg279Pro, and it could be argued that this kind of in vitro test is not sensitive enough to show an effect of the RSS mutation on the cell cycle. However, we demonstrated that the Arg279Leu mutation led to an increased stability, probably by preventing its degradation by the proteasome processes as suggested by Arboleda et al.31
Adrenal function
One of the key traits differentiating our patients with the CDKN1C mutation from IMAGe patients was the absence of adrenal failure. Arboleda et al showed that CDKN1C was strongly expressed during the embryonic development of the adrenal glands in mice.31 Furthermore, adrenal cytomegaly is one of the most common sign of BWS fetuses.36 This suggests that CDKN1C is required for the embryonic development of the adrenal glands. Medical history and adrenal function evaluations in the children and adult patients described here clearly excluded adrenal insufficiency. Again, it could be suggested that the Arg279Leu mutation is less deleterious, yielding a less severe phenotype than Arg279Pro, in accordance with the in vitro results. As a consequence, Arg279Leu would have no effect on adrenocortex development, unlike the mutations identified in IMAGe syndrome. However, further studies of the impact of all these gain-of-function mutations (IMAGe and RSS) on adrenal development are required.
In conclusion, CDKN1C mutations cause dominant, maternally transmitted, fetal growth disorders, and have been implicated in RSS, IMAGe and BWS. This highlights the major role of CDKN1C in controlling fetal growth. Gain-of-function mutations can lead to severe growth retardation (prenatal and postnatal), and CDKN1C should be investigated in patients with features of IMAGe and in patients with a family history of RSS and an absence of body asymmetry. Finally, the identification of such mutations in patients with RSS provides further support for the hypothesis that RSS and BWS are clinical and molecular mirror images of each other.
Acknowledgments
We thank the patients and their families and the physicians for referrals of their patients and providing precious clinical information. We thank Claire Calmel and Simon Crequit (Centre de Recherche Saint-Antoine, Paris) for their technical assistance for plasmid preparation and western blot experiments. We thank Arnaud Besson (Inserm UMR1037, Toulouse) for his valuable help to design the experiments about CDKN1C stability. IN is a member of the COST Action BM1208.
References
Footnotes
-
Contributors FB designed the study (genetic and functional studies), realised the functional experiments, analysed the results (molecular and functional experiments) and wrote the paper. IN and YLB designed the study, analysed the results (molecular and functional experiments) and revised the paper. SR designed the molecular experiments. AB, MLJ and NT realised the molecular experiments and analysed the results (genetic experiments). FP and AMF designed the functional experiments and analysed the results (functional experiments). IO and MT provided DNA and clinical data from the studied subjects and revised the paper.
-
Funding This work was supported by recurrent Institut National de la Santé et de la Recherche Médicale (INSERM) and Université Pierre et Marie Curie (UPMC) funding; Agence Nationale de la Recherche (ANR) [EPIFEGRO 2010 to I.N., F.B., Y.L.B.]; and People Programme (Marie Curie Actions) of the European Union's Seventh Framework Programme FP7/2007–2013/under REA grant agreement n° 290123. UPMC: YLB, IN, FB, AB: salaries; INSERM, ANR and FP7: lab equipments and reagents.
-
Competing interests None.
-
Patient consent Obtained.
-
Ethics approval Assistance Publique—Hôpitaux de Paris, authorisation no. 682.
-
Provenance and peer review Not commissioned; externally peer reviewed.