HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Submit a Letter to Editor Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Vaxillaire, M. Articles by Pontoglio, M. Search for Related Content PubMed PubMed Citation Articles by Vaxillaire, M. Articles by Pontoglio, M. Social Bookmarking                      What's this?

J Biol Chem, Vol. 274, Issue 50, 35639-35646, December 10, 1999

Anatomy of a Homeoprotein Revealed by the Analysis of Human MODY3 Mutations^*

Martine Vaxillaire^§¶, Amar Abderrahmani^§, Philippe Boutin^§, Bernard Bailleul^§, Philippe Froguel^§, Moshe Yaniv, and Marco Pontoglio

From the  Unité des Virus Oncogènes, Unité de Recherche Associée 1644, Centre National de la Recherche Scientifique, Département des Biotechnologies, Institut Pasteur, 25 rue du Dr Roux, 75724 Paris Cedex 15, France and ^§ Institut de Biologie de Lille, Unité Propre de Recherche et d'Enseignement Supérieur A 8090, Centre National de la Recherche Scientifique, Institut Pasteur de Lille, 1 rue du Professeur Calmette, BP 245, 59019 Lille, France

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Hepatocyte nuclear factor 1 (HNF1) is an atypical dimeric homeodomain-containing protein that is expressed in liver, intestine, stomach, kidney, and pancreas. Mutations in the HNF1 gene are associated with an autosomal dominant form of non-insulin-dependent diabetes mellitus called maturity-onset diabetes of the young (MODY3). More than 80 different mutations have been identified so far, many of which involve highly conserved amino acid residues among vertebrate HNF1. In the present work, we investigated the molecular mechanisms by which MODY3 mutations could affect HNF1 function. For this purpose, we analyzed the properties of 10 mutants resulting in amino acid substitutions or protein truncation. Some mutants have a reduced protein stability, whereas others are either defective in the DNA binding or impaired in their intrinsic trans-activation potential. Three mutants, characterized by a complete loss of trans-activation, behave as dominant negatives when transfected with the wild-type protein. These data define a clear causative relationship between MODY3 mutations and functional defects in HNF1 trans-activation. In addition, our analysis sheds new light on the structure of a homeoprotein playing a key role in pancreatic  cell function.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Hepatocyte nuclear factor 1 (HNF1 or LF-B1)¹ was initially characterized as a transcriptional regulator of a large set of hepatic genes, including albumin,  fibrinogen, and ₁-antitrypsin (1-3). It is highly conserved among vertebrates; e.g. there is 94% amino acid identity between rat and human proteins (4). HNF1 is an atypical homeoprotein that binds as a dimer to palindromic cis-acting DNA elements related to the GTTAATnATTAAC consensus sequence. HNF1 and its close homologue HNF1 (vHNF1 or LF-B3) constitute a unique subfamily of homeoproteins. The two proteins share strong homologies in the dimerization and DNA binding domains and can form heterodimers that bind to the same DNA sequences (5, 6). All the residues essential for DNA recognition are located in the 281 amino-terminal amino acids of these proteins. The dimerization domain was mapped to the first 33 amino-terminal amino acids (2, 7). This region can be bound by dimerization cofactor of HNF1, a small protein of 11 kDa that reduces the exchange rate of HNF1 molecules in both homodimer and heterodimer configurations (8). Dimerization cofactor of HNF1 was shown to be identical to 4-carbinolamine dehydratase, an enzyme that accelerates the regeneration of tetrahydrobiopterin (9, 10). HNF1 contacts its cognate sites via an extra large homeodomain (81 amino acids), consisting of four  helices with a loop of 16 amino acids, instead of the usual three, between the helices 2 and 3 (11-13). A conserved sequence of 85 amino acids located between the dimerization domain and the homeodomain, previously defined as the B domain, is necessary for high-specificity DNA recognition (7, 14). It was suggested to share a weak homology with the POU-A box; however this homology does not comprise the POU domain recognition helix, and it is not spaced correctly relative to the homeodomain. This region is highly conserved in both HNF1 and HNF1 of different species, and we will refer to it as the HNF domain. Finally, the carboxyl-terminal trans-activation domain extends from amino acid residue 282 to 631, including two proline-rich (29%) regions and a serine-rich (26%) region. Although initially assumed to be liver-specific, further studies have shown that HNF1 is not restricted to hepatocytes but is expressed in epithelial cells of the intestine, stomach, kidney proximal tubules, and pancreas (15, 16). The expression of HNF1 in the pancreas raised the possibility that it may be involved in the transcriptional regulation of pancreatic genes.

Maturity-onset diabetes of the young (MODY) is a subtype of non-insulin-dependent diabetes mellitus inherited as an autosomal dominant genetic disease (17). Affected individuals are characterized by a primary insulin secretion defect that generally becomes clinically apparent before 25 years of age, without insulin resistance (18, 19). MODY is a genetically heterogeneous disease that can be accounted for by at least three major loci called MODY1, MODY2, and MODY3. They were initially mapped by linkage analysis to chromosomes 20q (20), 7p (21), and 12q (22), respectively. Positional cloning and search for candidate genes have shown that they correspond to HNF4 (23), glucokinase (24), and HNF1 (25), respectively. HNF4 belongs to the steroid-thyroid hormone receptor superfamily of transcription factors and was shown to be a transcriptional regulator of HNF1 in hepatoma cell lines (26). Glucokinase is a key enzyme of the pancreatic glycolytic pathway, known to be a glucose sensor for regulating insulin secretion in response to glucose (27). The molecular basis of the  cell insulin secretion defect in MODY1 and MODY3 is not known. The clinical course of MODY3 subjects is characterized by a more severe diabetes than MODY2, requiring oral hypoglycemic agents or insulin treatment (18, 19). MODY3 represents the most common cause of early-onset non-insulin-dependent diabetes mellitus. In fact, it has been estimated to account for 30% of all MODY cases in France (22). Furthermore, the prevalence of the disease may be substantially higher than previously thought, given the wide range of the age of onset and the possible misdiagnosis of patients as insulin-dependent or non-insulin-dependent diabetes mellitus. Recent genetic studies of MODY3 kindreds have revealed a large spectrum of mutations in the HNF1 gene. So far, >80 different mutations (missense, nonsense, or frameshift) have been identified in the HNF1 sequence in association with the MODY3 phenotype (28-33). These mutations are localized along the different functional domains of the protein and involve highly conserved amino acid residues, suggesting that MODY3 is accounted for by impaired HNF1 function. However, the exact nature of the molecular defects and the mechanisms that explain the dominant transmission of the mutations are unknown. These mutations could generate proteins that might interfere with the function of the protein encoded by the wild-type allele from the other homologous chromosome. To investigate this issue, we analyzed the transcriptional activity of a number of representative MODY3 mutations. This was done by transient transfection of native and/or mutant human HNF1 with a target reporter gene. Our results revealed that most of the mutations strongly affected protein stability. Some of them were defective in their DNA binding capacity or in trans-activation potential. Finally, several but not all of the mutants behaved as dominant negative effectors on the wild-type protein.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Plasmids and Site-directed Mutagenesis-- In vitro mutagenesis was carried out on full-length human HNF1 expression vector (34) from double-stranded 5981-kilobase plasmid DNA using high-fidelity thermostable Pfu DNA polymerase (QuikChange; Stratagene, La jolla, CA). The full-length and truncated mutants in the Rous sarcoma virus (RSV)-driven human HNF1 expression plasmid were generated by polymerase chain reaction using two oligonucleotides primers each complementary to opposite strands of the vector. The oligonucleotides used for mutagenesis were 5'-CGAAGATGGTCAAGTCCTGCCTGCAGCAGCACAATATTCCACAGCGGGAGG-3' (mutant Y122C); 5'-GCAGCAGCACAATATTCCACAGCAGGAGGTGGTCGATACCAC-3' (mutant R131Q); 5'-GGCCTCAACCAGTTCCACCTGTCCCAACACC-3' (mutant S142F); 5'-TCCCATGAAGACGCAGAAGCAGGCCGCCCTGTATACCTGGTACGTC-3' (mutant R159Q); 5'-CCTGGTATGTCCGCAAGCAGTGAGAGGTGGCGCAGCAGTTC-3' (mutant R171ter); 5'-GGAGGAACCGTTTCCAGTGGGGCCCAGCATCCCAGCAAATATTGTTCC-3' (mutant K205Q); 5'-GGTTTGCCAACCGGCACAAAGAAGAAGCTTTCCGGCAC-3' (mutant R272H); and 5'-CAGGCACAGAGTGTGCTGGTGATCAACAGCATGGGCAG-3' (mutant P447L). The mutated nucleotides are underlined. To check the identity of each mutant plasmid by enzymatic digestion, we introduced in the vicinity of the point mutation some conservative nucleotide changes, when that was possible, to create or eliminate a restriction endonuclease site. We checked that nucleotide changes did not generate rarely used codons.

For the construction of the two frameshift and deletion mutants, sense (5'-CGTAAGCTTGCGGTAGAGGA-3') and antisense (5'-AAGTCTAGACTCACTGGAGGCCTGTGTCTGAGGT-3') oligonucleotides modified to create a HindIII restriction site at position 20 and an XbaI site at position +1630, respectively, based on position +1 for the ATG of human HNF1 cDNA, were used as the primer set for the amplification of a TG-deleted fragment at codons 547-548 and a fragment containing one C insertion at codon 289. The TG deletion and the in frame stop codon are in the antisense primer. A second set of oligonucleotides with the C insertion at position +739 (sense, 5'-TCCCCATCACAGGCAC-3') and position +858 (antisense, 5'-CTGGGGGGGGGCCCG-CTG-3') were used as internal primers with the first set to yield the P291fsinsC mutated fragment. The HindIII-XbaI-digested fragments were inserted downstream of the cytomegalovirus promoter in a pCDNA3 expression vector. Each of the 10 mutant plasmids was verified by DNA sequencing of proximal promoter and full-length HNF1 coding sequence to confirm the introduction of the desired mutations and to exclude other sequence changes.

Cell line, Transient Transfection, and Chloramphenicol Acetyltransferase (CAT) Assay-- The C33 human epithelial cervical carcinoma cells (35), which lack endogeneous HNF1 expression, were cultured and transfected as described previously (34).

For each transfection, a total of 5 µg of DNA were used, including 1 µg of the p28₃ reporter plasmid (36), varying amounts of HNF1 expression vectors, and 1 µg of pRSV--galactosidase construct for normalization of transfection efficiencies. Amounts from 10 ng to 1 µg of wild-type or mutant HNF1 expression vectors were tested, and the total quantity of plasmids was maintained constant by adding suitable amounts of a pGEM3 plasmid. 40-48 h after transfection, the cells were harvested, and total cell extracts were prepared. CAT and -galactosidase activities were measured as described previously (37). Each experiment was repeated at least three to five times with at least two different plasmid preparations.

Electrophoretic Mobility Shift Assays with Nuclear Extracts from Transfected Cells-- C33 cells (2.5 × 10⁶ cells) were seeded into 10-cm Petri dishes and transfected with a similar amount (2.5 µg) of native or mutant HNF1 plasmids. For electrophoretic mobility shift assay, a cytomegalovirus enhancer-driven E2TR expression plasmid (38) was co-transfected with HNF1 plasmid in a similar amount (2.5 µg) as an internal marker for transfection efficiency as well as reproductibility of nuclear extract preparation. Nuclear extracts from the transfected cells were prepared from two 10-cm Petri dishes essentially as described previously (34, 39). The protein concentration was measured by Bradford assay. A single-stranded oligonucleotide (5'-CTTTAGTTAATATTTGACAGTTT-3'), corresponding to the HNF1 DNA binding element of the rat -fibrinogen promoter, was end-labeled by using [-³²P]dATP and T4 polynucleotide kinase and then annealed with an excess (1.5-fold) of cold complementary oligonucleotide. For detection of E2TR DNA binding activity, the oligonucleotides were as follows: 36Y1 (5'-CTAGACAACCGATTTCGGTTGT-3') and 36Y2 (5'-CTAGACAACCGAAATCGGTTGT-3'); 36Y1 was end-labeled as described above. Binding reactions were performed as described previously (34).

Western Blot Analysis-- C33 nuclear extracts were prepared from transfected cells as described above. 20 µg of protein extract were resolved by 7 or 10% SDS-polyacrylamide gel electrophoresis and electrotransferred to a nitrocellulose membrane (Bio-Rad). After control of equal loading by Ponceau staining, the membranes were incubated in phosphate-buffered saline (PBS) containing 10% dried milk for 30 min, and subsequently washed twice for 10 min in PBS supplemented with 0.1% Tween 20 and once for 10 min in PBS. The membranes were then incubated with a 1:1000 dilution of polyclonal rabbit anti-HNF1 antibody, rHNt-283, as described previously (40).

Metabolic Labeling of Transiently Transfected Cells and Immunoprecipitation-- C33 cells were transfected as described above with 10 µg of native or mutant (Y122C and R159Q) HNF1 plasmids. To analyze protein stability, 40-48 h after transfection the cells were pulse-labeled for 15 min with ³⁵S-containing amino acids (cysteine and methionine; Tran³⁵S-label, ICN) and chased for various lengths of time in the presence of excess nonradioactive cysteine and methionine added to the Dulbecco's modified Eagle's medium plus 3% dialyzed fetal calf serum culture medium. The cells were washed twice with ice-cold PBS, and then all subsequent steps were performed at 4 °C. The cells were lysed in HNB buffer and 1% Nonidet P-40 as described above for nuclear extract preparation (39). The nuclei were resuspended in 30 µl of cold RB buffer (60 mM KCl, 15 mM NaCl, 15 mM HEPES, pH 7.5, 2 mM EDTA, 0.5 mM EGTA, 0.15 mM spermine, 0.5 mM spermidine, 1 mM -mercaptoethanol, 0.2 mM phenylmethylsulfonyl fluoride, 2 µg/ml bastatin, 2 µg/ml aprotinin, 2 µg/ml pepstatin) and then treated with 3 µl of DNaseI (10 units/µl) to reduce the viscosity by incubating for 5 min on ice with the enzyme in the presence of 4.5 mM MgCl₂. Then the digestion of DNA was stopped by adding 6 µl of 166 mM EDTA, and the nuclear extracts were resuspended in 120 µl of radioimmunoprecipitation assay buffer (150 mM NaCl, 1.0% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0) and subjected to immunoprecipitation with saturating amounts (2.5 µl) of rHNt-283 antibody according to standard procedures (41).

Immunodetection in Transfected Cells-- C33 cells were grown and transfected onto 18-mm glass coverslips as described above. To assess the cellular localization of each mutant and wild-type HNF1 protein, we performed indirect immunofluorescence on the fixed cells. The cells on the coverslip were fixed for 20 min in 4% paraformaldehyde at room temperature, rinsed in PBS, and then permeabilized with 1% Triton X-100, rinsed again in PBS, and incubated with the primary anti-HNF1 antibody, rHNt-283, in the same conditions as those described for the immunoblot analysis. The coverslips were then incubated with a fluorescein-coupled anti-rabbit antibody (Amersham Pharmacia Biotech) in PBS supplemented with 0.1% Tween 20 and 10% FCS, rinsed twice in PBS supplemented with 0.1% Tween 20 and then in PBS, and mounted in antifade medium for viewing on a Zeiss Axiophot epifluorescence microscope.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Construction of HNF1 Mutants-- Based on previous structural and functional studies of HNF1 and HNF1, predictions could be made about the activity of some of the MODY3 mutations. It has been shown that deletions of the activation domain generate proteins that can still dimerize, bind DNA, and behave as dominant negative mutants on the activity of the normal protein (34, 42). However, the properties of point mutations were difficult to predict, because no single amino acid substitution has been previously analyzed. For this reason, we chose to evaluate the properties of 10 mutants in different domains of HNF1 (Fig. 1). Seven of them were point mutations resulting in amino acid substitution (Y122C, R131Q, S142F, R159Q, K205Q, R272H, and P447L); one was a nonsense mutation (R171ter), and two were frameshift mutations (P291fsinsC/ter and T547E548fsdelTG/ter) resulting in protein truncation. These mutations are responsible for the MODY3 phenotype in different families. The seven missense mutants and the amino-terminal truncated protein (R171ter) were introduced into the human HNF1 cDNA sequence and cloned into an expression vector under the control of the Rous sarcoma virus long terminal repeat. The two carboxyl-terminal frameshift mutations of the same cDNA were introduced in a pCDNA3 expression vector under the control of a cytomegalovirus promoter. The constructs were sequenced in their promoter and coding portions to ensure that additional mutations were not introduced during the mutagenesis step.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Schematic representation of the HNF1 functional domains. The positions of the 10 HNF1 mutations analyzed are indicated at the bottom. ter, stop codon; fs, frameshift mutation, both resulting in protein truncation. The filled arrows show missense mutations, and open arrows show truncation mutations.

Subcellular Localization of HNF1 Mutants-- To investigate whether HNF1 mutations affected nuclear translocation, we transfected the different mutants in C33 cells and monitored the subcellular localization by indirect immunofluorescence. Almost all of the mutant proteins tested gave a nuclear signal with antibodies prepared against the amino-terminal half of HNF1, showing that they accumulate in the nucleus (Fig. 2). For some of them (Y122C, S142F, R272H, and P291fsinsC(ter)), in a certain percentage of transfected cells, a faint signal was also detected in the cytoplasm. R171ter was, on the contrary, the only mutant that was abundantly localized both in the cytoplasm and in the nucleus. These results suggest that the nuclear import is normally operating for most of the mutants tested.

View larger version (40K): [in this window] [in a new window]   Fig. 2.   Subcellular localization. Subcellular localization of HNF1 mutants was monitored by indirect immunofluorescence on fixed, transiently transfected C33 cells. A, polyclonal anti-HNF1 immunoreactive proteins were detected with fluorescein isothiocyanate-labeled goat anti-rabbit immunoglobulin secondary reagent. B, blue nuclear fluorescence from the 4', 6-diamidino-2-phenylindole complexed with DNA. C, superposition of fluorescein isothiocyanate (green) and 4', 6-diamidino-2-phenylindole (electronically converted to red) stainings. D, phase-contrast images of the immunostained cells.

HNF1 Mutants Are Expressed at Different Levels-- To evaluate a possible defect in protein stability, we first evaluated the expression levels of the mutant proteins in the transfected cells by Western blot analysis. As shown in Fig. 3A, protein levels of mutants Y122C, R159Q, K205Q, and P447L were reduced compared with the wild type. Conversely, mutants R131Q, S142F, and R272H retained a relatively normal expression level. In agreement with previous studies, HNF1 protein migrates as a slightly diffuse band with an apparent molecular mass ranging approximately between 92 and 98 kDa (40). Strikingly, P447L was the only mutant that reproducibly appeared as a sharper band. A specific HNF1 fragment was detected with an amino-terminal anti-HNF1 antibody for the mutant R171ter as a weak band migrating with an apparent molecular mass <30 kDa. Finally, mutants P291fsinsC and T547E548fsdelTG were overproduced relative to the wild-type protein.

View larger version (58K): [in this window] [in a new window]   Fig. 3.   Protein expression and DNA binding of the different mutants. A, Western blot analysis and immunodetection of HNF1 wild-type and mutant proteins in nuclear extracts from transfected C33 cells. An equal quantity of nuclear proteins was analyzed as detailed under "Experimental Procedures." B, gel shift analysis of HNF1 mutants with the -fibrinogen HNF1 binding site (see "Experimental Procedures" for details). C, gel shift analysis with an E2 DNA binding site for normalization of transfection efficiency and quality of nuclear extract preparation. Note that two retarded bands are usually seen with this papillomavirus-derived E2TR transcriptional regulator. Left and right panels derive from two independent experiments using RSV long terminal repeat and cytomegalovirus expression vectors, respectively. The data shown represent a typical experiment that was reproduced several times.

To test whether the reduced protein levels obtained for Y122C and R159Q resulted from decreased stability of these mutant proteins, we performed pulse-chase experiments in transfected C33 cells followed by immunoprecipitation of mutant proteins. Upon very short labeling times (15 min) and before the chase, the protein level of both mutants was only partially reduced compared with the wild-type protein, indicating that the dramatically reduced protein levels observed at steady state could not be explained by a possible defective protein synthesis. Indeed, after the chase, the decline of both mutant proteins was much more pronounced when compared with that of the wild-type protein (Fig. 4 and results not shown). From these data, we have estimated that the wild-type HNF1 half-life in C33 cells is ~12 h, whereas for the mutant proteins it is only 2.5 h.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Stability of Y122C mutant compared with wild-type HNF1 protein. C33 cells were labeled with [35S]methionine and cysteine for 15 min (0) and pulse-chased for various lengths of time (45, 90, and 180 min) in the presence of excess nonradioactive methionine and cysteine. Samples were immunoprecipitated under normal stringency conditions using polyclonal anti-HNF1 antibody, rHNt283, and subjected to SDS-polyacrylamide gel electrophoresis. Nuclear extracts prepared from untransfected C33 cells were immunoprecipitated in the same conditions than for those prepared from cells transiently transfected with wild-type or mutant HNF1-encoding plasmid and used as negative control (Mock).

Several HNF1 Mutants Are Defective in DNA Binding-- Next, we performed electrophoretic mobility shift assays to assess the specific DNA binding activity of the different mutants. As shown in Fig. 3B, specific DNA binding activity was detected for mutants Y122C, R131Q, R159Q, K205Q, R272H, and P447L. The first five proteins showed the same mobility as that obtained with the wild-type HNF1 protein. Conversely, mutant P447L showed a slightly different migration pattern characterized by a band that migrated reproducibly slightly faster than the wild-type HNF1 band. The two carboxyl-terminal deletion mutants clearly showed a strong increase in the DNA binding activity that reflected their increased expression level. These proteins showed a faster mobility of the retarded complexes as previously reported for carboxyl-terminal truncation mutants and were able to heterodimerize with a cotransfected full-length wild-type protein (Ref. 43 and data not shown). As expected, no binding activity was detected for mutant R171ter. Surprisingly, S142F lacked any DNA binding activity. To monitor the transfection efficiency and the quality of the nuclear extract preparation, we included in the transfection the papillomavirus-encoded E2TR protein expression vector (38). Fig. 3C shows that E2TR DNA binding activity was equally detected in each transfection. Comparison of the intensity of the retarded bands relative to the level of expressed proteins demonstrated that the two carboxyl-terminal truncated mutants bound DNA at least as well as normal HNF1. On the contrary, mutant R272H had reduced DNA binding activity when compared with the steady-state protein levels. Finally, Y122C, R159Q, K205Q, and P447L had very low binding activities compatible with their low level of expression, whereas R131Q showed only a small decrease in DNA binding and expression level.

Trans-activation Potential of HNF1 Mutants-- To assess the trans-activation potential of the HNF1 mutants compared with that of the native protein, we transfected increasing amounts of the HNF1 expression vector together with a CAT reporter gene under the control of the chimeric 28 promoter containing three HNF1 binding sites upstream of the rat -fibrinogen TATA box sequence (construct p28₃; Ref. 36). Transfections were performed in the human epithelial C33 cells, which lack any endogeneous HNF1 expression. In agreement with previous studies, HNF1 activated this reporter plasmid up to 50-70-fold. No appreciable effect (<2-fold) was observed by further addition of dimerization cofactor of HNF1-encoding plasmid (data not shown). The capacity of increasing amounts of mutant HNF1 expression vectors to activate the reporter was compared with the wild-type protein in the same experiment. Fig. 5 shows the different patterns of trans-activation we obtained for the 10 mutants analyzed. The profiles shown are representative of three to five experiments and reveal either markedly decreased activity, particularly at low input of transfected HNF1 encoding plasmid, or no transcriptional activity. The 10 mutants can be grouped as follows.

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Transcriptional activation by HNF1 mutants. The trans-activation profiles obtained for each of the 10 mutants analyzed are shown as a mean ± S.E. from three to five independent experiments. Expression vectors for the wild-type and mutant proteins were cotransfected with the reporter construct p(28)3 CAT containing three HNF1 binding sites upstream of the -fibrinogen minimal promoter. RSV--galactosidase was included as a reference plasmid. The -galactosidase-normalized CAT activity is expressed as -fold activation (y axis). Trans-activations with expression vector encoding for wild type (WT, diamonds) or mutant (squares) are plotted against the amount (ng) of the vector transfected into C33 cells (x axis).

Class I: Trans-activation Defect Brought about by Reduced Mutant Protein Stability Coupled with a Possible Decrease in DNA Binding-- Y122C activated transcription of the CAT reporter gene at significantly lower levels than those obtained with the wild-type HNF1. This was particularly true in the lower range of the transfected plasmid (10-200 ng). Activation levels similar to those of the wild-type protein (65-70-fold) were obtained when 1 µg of mutant DNA was transfected. However, although the trans-activation by the wild-type protein approached a plateau at 200 ng of plasmid in most experiments, a 5-fold excess of mutant protein-encoding plasmid was required to reach this level of activity. Comparable profiles were obtained for mutants R131Q, R159Q, K205Q, and R272H. For mutants Y122C, R159Q, and K205Q, the impaired trans-activation potential can be correlated with decreased mutant protein accumulation. In contrast, mutant R272H showed a decreased intrinsic trans-activation brought about by a decrease in specific DNA binding, whereas the protein stability was only marginally affected.

Class II: Very Low or No Activity-- S142F mutant displayed no transcriptional activity below 200 ng of plasmid transfected and very weak activation levels when 1 µg of plasmid was transfected compared with the wild-type activation level (11 ± 3.5- versus 65 ± 19.7-fold, mean ± S.E.; n = 4). A similar profile was obtained for R171ter mutant, as could be expected. Surprisingly, P447L, a point mutation in the trans-activation domain, had almost no trans-activation function, even when 1 µg of DNA was transfected. As discussed below, mutants P291fsinsC(ter) and T547E548fsdelTG(ter) also belong to this class.

The residual transcriptional activity of the different mutants can be compared by considering the trans-activation efficiency obtained in the linear range of the dose-response curve, namely when 50 ng of plasmid were transfected. As shown in Fig. 5, mutants S142F, R171ter, P447L, P291fsinsC(ter), and T547E548fsdelTG(ter) exhibited very weak activation capacities. On the other hand, only two mutants showed a trans-activation capacity >20% of that of the wild-type protein. These mutants were Y122C and R131Q with a trans-activation capacity of 22 ± 2% and 52 ± 6%, respectively. Three others, K205Q, R272H, and R159Q exhibited ~10% of the wild-type activity.

Several Mutant Proteins Can Affect the Activity of the Wild-type Protein-- We next examined which of the mutations may have generated dominant negative effectors. To this end, we cotransfected a low constant amount (50 ng) of the wild-type-encoding plasmid with increasing amounts of mutant expression vector (50 ng to 1 µg). Class I mutants, i.e. Y122C, R131Q, R159Q, K205Q, and R272H, did not interfere with the wild-type HNF1 activity but rather increased slightly the reporter activity (see Fig. 6A for an example). The mutants grouped in class II could be further subdivided into two subgroups. In the first, which we called IIa, increasing amounts of plasmid mutants, including S142F and R171ter, only marginally changed the activity of the cotransfected wild-type protein (Fig. 6B and results not shown). On the contrary, mutants belonging to subgroup IIb decreased the activity of the coexpressed normal protein almost to baseline. These included P447L and the two carboxyl-terminal frameshift mutants, P291fsinsC(ter) and T547E548fsdelTG(ter) (see Fig. 6C for an example).

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Effect of mutant proteins on wild-type protein activity. The three different profiles of interference with wild-type activity are illustrated for mutants Y122C (class I), S142F (class IIa), and P447L (class IIb). Expression vectors were cotransfected with the reporter construct p(28)3 containing three HNF1 binding sites. The -galactosidase-normalized CAT activity is expressed as -fold activation (y axis). Trans-activations with expression vector encoding for wild type (WT, diamonds) or mutant (squares) are plotted against the amount (ng) of the vector transfected into C33 cells (x axis). Trans-activations with a constant amount (50 ng) of wild-type and increasing amounts of mutant vector are indicated by triangles. The arrows indicate the values for equimolar amounts of wild-type and mutant plasmids when cotransfected. The profiles shown are representative of three to five experiments. Mutants R131Q, R159Q, K205Q, and R272H behaved similarly to Y122C (A); mutant R171(ter) behaved similarly to S142F (B); and mutants P291fsinsC(ter) and T547E548fsdelTG(ter) behaved similarly to P447L (C).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Transcriptional activation by MODY3 mutants-- MODY3 mutations in HNF1 are probably one of the largest set of known mutations affecting a homeoprotein. In fact, >80 different mutations have been identified so far in the human HNF1 gene. Most of the mutations affect amino acid residues that have been conserved during evolution in all HNF1 and HNF1 genes. Missense mutations concern all the functional domains of HNF1; however, the highest mutation density occurs in the HNF and homeodomains (Table I). This suggests that protein domains that are involved in DNA binding are more sensitive to single amino acid changes than the trans-activation domain. The present work aimed to investigate how MODY3 mutations could affect HNF1 function. Ten mutants located in different functional domains of HNF1 have been monitored for nuclear localization, protein stability, DNA binding, and transcriptional activation.

View this table: [in this window] [in a new window]   Table I Density of MODY3/HNF1 mutations in the different functional domains of the protein A total of 51 single amino acid substitutions were considered. For each domain, the mutation density is expressed as the number of amino acid changes for 100 amino acid residues.

View this table: [in this window] [in a new window]   Table II Biological properties of the HNF1 mutants Class distribution, transcriptional activation, expression level, and estimated specific activity for the 10 MODY3/HNF1 mutants. The expression levels were estimated from the Western blot shown in Fig. 3, and the specific activity represents the fold of transactivation with limiting plasmid concentration divided by the protein expression level.

Our results demonstrate that all the expression vectors coding for MODY3 mutant proteins present a defective HNF1 transcriptional function in transient transfection studies. In this respect, we have shown that some mutants have reduced protein stability, whereas others are either defective in the DNA binding or impaired in their intrinsic trans-activation potential. Three mutants, characterized by a complete loss of trans-activation, have also acquired a dominant negative potential on the wild-type protein in the trans-activation assay (Table II).

Involvement of the HNF Domain and Homeobox Domain in DNA Binding-- Among the different point mutants in the HNF domain, the S142F mutant is the only one that totally failed to activate transcription, presumably because it failed to bind DNA. The HNF domain was shown to enhance sequence-specific DNA binding and to mediate the formation of DNA-stabilized protein dimers (14). In fact, the HNF domain seems to impose an oriented 2-fold symmetry on the dimer molecule, thus defining a reciprocal orientation of the two homeodomains in the bound dimer. In this way, the HNF region might increase the binding site specificity by permitting the recognition of only those DNA sequences containing properly oriented half-sites. In addition, the HNF domain could interact with the natural cognate DNA sequence, for example, with the strictly conserved GT residues immediately 5' to the TAAT recognition sequence. Indeed, UV cross-linking experiments suggest that the HNF domain may be in close proximity to the target DNA.² From this, we speculate that the amino acid change in the S142F mutant could have disturbed a cooperative interaction between the HNF domain and the homeodomain of HNF1. The present data further reinforce the role of the 85 amino acids containing the HNF domain in high-affinity specific DNA binding. Because the homology of this domain with the POU-specific domain is not significant, we suggest that it be considered as a new conserved motif outside the homeodomain. It will be interesting to discover new atypical homeogenes containing this HNF motif and to search for its possible occurrence in less developed organisms.

Mutant R272H activates only weakly, compared with its expression level. It seems that R272 lies at a critical position, in helix 3 of the homeodomain, which is part of the WFXNXR motif found to be strictly conserved among all known homeodomain sequences. In conclusion, both the homeodomain and the HNF domain contribute to the strength of DNA binding by HNF1 and are crucial in vivo. Such an interdependence could influence the selection of target sites by homeobox proteins.

Negative Dominance of Some Mutant Proteins-- It is worth nothing that the three mutants that have lost the trans-activation potential but that still were able to bind DNA efficiently and dimerize had a strong dominant negative effect in the co-transfection and trans-activation assay. The dominant negative effect, in these cases, is very likely attributable to the formation of a heterodimeric inactive complex. Particularly interesting is the case of the mutant T547E548fsdelTG, which results in a carboxyl-terminal deletion of the last 85 amino acid residues. Previous studies have shown that progressive unidirectional deletions of the carboxyl-terminal domain resulted in a gradual decrease in the trans-activation potential of the protein. In fact, deletion mutants of the last 62 and 114 amino acid residues retained a specific activity of 75 and 55% relative to the intact protein, respectively (43). However, the mutant T547E548fsdelTG is characterized by a much more drastic effect; the mutation not only abolished trans-activation but also gave rise to a dominant negative protein. The carboxyl-terminal domain of this mutant contains four unrelated additional amino acids introduced by the frameshift, which may cause the protein to adopt an inactive conformation. Alternatively, HNF1 residual activity may be more sensitive to the position of protein truncation than previously thought.

A surprising result came from mutant P447L, which fails to activate transcription, acts as a dominant negative effector, and shows a particular migration pattern in both the electrophoretic mobility shift assay and Western blot analysis. It seems that the amino acid substitution from proline to leucine leads to an important conformational change of the protein. Proline has a major effect on protein structure, because the nitrogen atom of the amino group is incorporated into a ring. The peptidylprolyl amide bond can exist in either cis or trans stereochemical configurations, so that a proline residue disrupts the usual organization of the backbone of a polypeptide chain by introducing a constraint in the direction of the chain. It is interesting to note that another proline residue changed into leucine at position 519 in the trans-activation domain of HNF1 has been reported as a mutation responsible for MODY3 (28). Somewhat similar trans-dominant negative effects have been reported for some mutants in the ligand-dependent trans-activation domain, AF2, of various steroid hormone receptors. A point mutant in the AF2 progesterone receptor region, PRBE911A, has been shown to inhibit trans-activation by a cotransfected wild-type receptor without influencing hormone and DNA binding affinities (44). A carboxyl-terminal deleted form of glucocorticoid receptor, called Gr, has also been claimed to exhibit a dominant negative effect (45). It is possible that the P447L mutant, via the heterodimerization with the wild-type protein, gives rise to a complex that is unable to activate transcription. The aberrant conformation of such a complex could prevent protein contacts essential for transcriptional activation. All these observations suggest that the function of the trans-activation domain of HNF1 could rely on its dimeric structure.

HNF1 Mutations and MODY3 Phenotype-- The first two classes of mutants (I and IIa) showed decreased and no trans-activation, respectively, whereas class IIb mutants showed strongly decreased trans-activation and behaved as dominant negative mutants. These different behaviors should predict important consequences on the clinical course of the pancreatic dysfunction of individuals carrying distinct types of mutations. MODY3 patients heterozygous for mutations in HNF1 belonging to the last class of mutants (IIb) should have a more pronounced reduction of HNF1 activity compared with patients carrying mutations of class I or IIa. This should be particularly true for carriers of the two carboxyl-terminal truncated proteins that accumulated to levels much higher than that of the wild-type protein and readily inhibited the activity of the wild-type protein. However, the analysis of the clinical data did not show any clear correlation between the severity of the phenotype of MODY3 patients and the nature of their mutations.³ Such an apparent paradox is further complicated by the observation that MODY3 is inherited as a dominant trait in humans, whereas in mice, the same type of pancreatic dysfunction is inherited as a recessive trait. In fact, a null mutation in the murine HNF1 gene (46) gives rise to a severe insulin secretion defect only in homozygous animals, whereas the heterozygous mice are normoglycemic and have a normal glucose tolerance (47). We should bear in mind that the affected mice are nullizygous, because the mutation consists of a deletion that prevents the synthesis of any trace of HNF1 protein. How can we explain these discrepancies? The lack of any correlation between the clinical data and the dominant negative nature of the mutations would suggest that a unique mechanism is at the basis of the MODY3 phenotype.

One possibility is that all MODY3 mutant proteins behave in reality as dominant negative effectors in MODY3 patients. In this case, however, we have to assume that transient transfection experiments cannot reproduce the exact conditions occurring in the pancreatic  cells. We have to postulate that even in the presence of minimal amount of transfecting plasmid, a saturating intranuclear level of normal HNF1 protein will be produced in the successfully transfected cells. Mutant proteins that are either unstable or unable to bind DNA will not prevent the wild-type protein from binding to the test promoter. Only mutants that generate stable proteins that bind DNA without activating transcription will behave as dominant negative in ex vivo assays if this hypothesis is correct.

Alternatively, one could suggest that the negative dominance is not the mechanism at the basis of the dominant inheritance of MODY3. In this case we should postulate that HNF1 protein levels are in excess in mice but not in humans. If this is true, in a MODY3 patient, when the increased demand for  cell function occurs with age, the decrease HNF1 activity might become insufficient and thus elicit the onset of diabetes. If this reasoning is correct, all the MODY3 mutations studied might in fact be merely loss of function mutations resulting in haploinsufficiency. Such a hypothesis is compatible with the identification of promoter mutations that decrease HNF1 mRNA synthesis and with a truncation mutation at the seventh codon (33).

The discrepancy in the inheritance of the traits between mouse and human is reminiscent of what has been observed for HNF4/MODY1, HoxD13 (48), and Brn-3.1/Pou4f3 (49). In these cases, the traits are transmitted in a dominant fashion in humans, whereas they are in a recessive manner in mice. The analysis of the HNF4 (Q268X) mutation showed a loss of function effect (50), whereas HNF4 heterozygous mice showed no signs of diabetes and exhibited normal glucose tolerance. Conversely, the analysis of HoxD13 and Brn-3.1/Pou revealed a dominant negative effect of the mutated proteins.

Further experiments that can distinguish between the two alternative explanations, loss of function resulting in haploinsufficiency or formation of a dominant negative protein, should await the reconstruction of knock-in mice with different types of MODY3 mutations. In this case, one will be able to follow the level of HNF1 activity and hopefully the transcription of target genes with age.

Altogether, our data define a clear causative association between MODY3 mutations and functional defects in HNF1-dependent trans-activation. This work also sheds new light on the role played by specific amino acid residues in the function of a homeoprotein that is involved in the maintenance of normal pancreatic  cell function.

	ACKNOWLEDGEMENTS

We are indebted to Tanguy Chouard for the rHNt-283 antibody. We thank Antonia Doyen for advice on cell culture and technical assistance, Armelle Lengronne for her help in transfection experiments, and Françoise Thierry for the gift of E2TR-encoding plasmid. We also thank all the other members of the Virus Oncogenes Unit and are very grateful to Jonathan Weitzman for critically reading the manuscript.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^¶ Present address: Institut de Biologie de Lille, Unité Propre de Recherche et d'Enseignement Supérieur A 8090, Centre National de la Recherche Scientifique, Institut Pasteur de Lille, 1 rue du Professeur Calmette, BP 245, 59019 Lille, France.

To whom correspondence should be addressed: Unité des Virus Oncogènes, Unité de Recherche Associée 1644, Centre National de la Recherche Scientifique, Département des Biotechnologies, Institut Pasteur, 25 rue du Dr Roux, 75724 Paris Cedex 15, France. Tel.: 33-1-45-68-85-12; Fax: 33-1-40-61-30-33; E-mail: yaniv@pasteur.fr.

² T. Chouard, unpublished observations.

³ M. Vaxillaire, unpublished observations.

	ABBREVIATIONS

The abbreviations used are: HNF1, hepatocyte nuclear factor 1; MODY, maturity-onset diabetes of the young; RSV, Rous sarcoma virus; CAT, chloramphenicol acetyltransferase; PBS, phosphate-buffered saline.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				L. W. Harries, M. J. Sloman, E. A.C. Sellers, A. T. Hattersley, and S. Ellard Diabetes Susceptibility in the Canadian Oji-Cree Population Is Moderated by Abnormal mRNA Processing of HNF1A G319S Transcripts Diabetes, July 1, 2008; 57(7): 1978 - 1982. [Abstract] [Full Text] [PDF]

				M. Vaxillaire and P. Froguel Monogenic Diabetes in the Young, Pharmacogenetics and Relevance to Multifactorial Forms of Type 2 Diabetes Endocr. Rev., May 1, 2008; 29(3): 254 - 264. [Abstract] [Full Text] [PDF]

				C. Bellanne-Chantelot, C. Carette, J.-P. Riveline, R. Valero, J.-F. Gautier, E. Larger, Y. Reznik, P.-H. Ducluzeau, A. Sola, A. Hartemann-Heurtier, et al. The Type and the Position of HNF1A Mutation Modulate Age at Diagnosis of Diabetes in Patients with Maturity-Onset Diabetes of the Young (MODY)-3 Diabetes, February 1, 2008; 57(2): 503 - 508. [Abstract] [Full Text] [PDF]

				A. D. Tward, K. D. Jones, S. Yant, S. T. Cheung, S. T. Fan, X. Chen, M. A. Kay, R. Wang, and J. M. Bishop Distinct pathways of genomic progression to benign and malignant tumors of the liver PNAS, September 11, 2007; 104(37): 14771 - 14776. [Abstract] [Full Text] [PDF]

				W. Winckler, M. N. Weedon, R. R. Graham, S. A. McCarroll, S. Purcell, P. Almgren, T. Tuomi, D. Gaudet, K. B. Bostrom, M. Walker, et al. Evaluation of Common Variants in the Six Known Maturity-Onset Diabetes of the Young (MODY) Genes for Association With Type 2 Diabetes Diabetes, March 1, 2007; 56(3): 685 - 693. [Abstract] [Full Text] [PDF]

				L. W. Harries, S. Ellard, A. Stride, The European MODY consortium, N. G. Morgan, and A. T. Hattersley Isomers of the TCF1 gene encoding hepatocyte nuclear factor-1 alpha show differential expression in the pancreas and define the relationship between mutation position and clinical phenotype in monogenic diabetes Hum. Mol. Genet., July 15, 2006; 15(14): 2216 - 2224. [Abstract] [Full Text] [PDF]

				C. W. Rowley, L. J. Staloch, J. K. Divine, S. P. McCaul, and T. C. Simon Mechanisms of mutual functional interactions between HNF-4{alpha} and HNF-1{alpha} revealed by mutations that cause maturity onset diabetes of the young Am J Physiol Gastrointest Liver Physiol, March 1, 2006; 290(3): G466 - G475. [Abstract] [Full Text] [PDF]

				E. Barbacci, A. Chalkiadaki, C. Masdeu, C. Haumaitre, L. Lokmane, C. Loirat, S. Cloarec, I. Talianidis, C. Bellanne-Chantelot, and S. Cereghini HNF1{beta}/TCF2 mutations impair transactivation potential through altered co-regulator recruitment Hum. Mol. Genet., December 15, 2004; 13(24): 3139 - 3149. [Abstract] [Full Text] [PDF]

				C. Bellanne-Chantelot, D. Chauveau, J.-F. Gautier, D. Dubois-Laforgue, S. Clauin, S. Beaufils, J.-M. Wilhelm, C. Boitard, L.-H. Noel, G. Velho, et al. Clinical Spectrum Associated with Hepatocyte Nuclear Factor-1{beta} Mutations Ann Intern Med, April 6, 2004; 140(7): 510 - 517. [Abstract] [Full Text] [PDF]

				L. W. Harries, A. T. Hattersley, and S. Ellard Messenger RNA Transcripts of the Hepatocyte Nuclear Factor-1{alpha} Gene Containing Premature Termination Codons Are Subject to Nonsense-Mediated Decay Diabetes, February 1, 2004; 53(2): 500 - 504. [Abstract] [Full Text]

				S. B. Smith, R. Gasa, H. Watada, J. Wang, S. C. Griffen, and M. S. German Neurogenin3 and Hepatic Nuclear Factor 1 Cooperate in Activating Pancreatic Expression of Pax4 J. Biol. Chem., October 3, 2003; 278(40): 38254 - 38259. [Abstract] [Full Text] [PDF]

				J. K. Divine, S. P. McCaul, and T. C. Simon HNF-1{alpha} and endodermal transcription factors cooperatively activate Fabpl: MODY3 mutations abrogate cooperativity Am J Physiol Gastrointest Liver Physiol, June 9, 2003; 285(1): G62 - G72. [Abstract] [Full Text] [PDF]

				L. Bjorkhaug, J. V. Sagen, P. Thorsby, O. Sovik, A. Molven, and P. R. Njolstad Hepatocyte Nuclear Factor-1{alpha} Gene Mutations and Diabetes in Norway J. Clin. Endocrinol. Metab., February 1, 2003; 88(2): 920 - 931. [Abstract] [Full Text] [PDF]

				A. Stride, M. Shepherd, T. M. Frayling, M. P. Bulman, S. Ellard, and A. T. Hattersley Intrauterine Hyperglycemia Is Associated With an Earlier Diagnosis of Diabetes in HNF-1{alpha} Gene Mutation Carriers Diabetes Care, December 1, 2002; 25(12): 2287 - 2291. [Abstract] [Full Text] [PDF]

				T. Klupa, J. H. Warram, A. Antonellis, M. Pezzolesi, M. Nam, M. T. Malecki, A. Doria, S. S. Rich, and A. S. Krolewski Determinants of the Development of Diabetes (Maturity-Onset Diabetes of the Young-3) in Carriers of HNF-1{alpha} Mutations: Evidence for parent-of-origin effect Diabetes Care, December 1, 2002; 25(12): 2292 - 2301. [Abstract] [Full Text] [PDF]

				J. Ferrer A Genetic Switch in Pancreatic {beta}-Cells: Implications for Differentiation and Haploinsufficiency Diabetes, August 1, 2002; 51(8): 2355 - 2362. [Abstract] [Full Text] [PDF]

				B. L. Triggs-Raine, R. D. Kirkpatrick, S. L. Kelly, L. D. Norquay, P. A. Cattini, K. Yamagata, A. J. G. Hanley, B. Zinman, S. B. Harris, P. H. Barrett, et al. HNF-1alpha G319S, a transactivation-deficient mutant, is associated with altered dynamics of diabetes onset in an Oji-Cree community PNAS, April 2, 2002; 99(7): 4614 - 4619. [Abstract] [Full Text] [PDF]

				J. Eeckhoute, P. Formstecher, and B. Laine Maturity-Onset Diabetes of the Young Type 1 (MODY1)-Associated Mutations R154X and E276Q in Hepatocyte Nuclear Factor 4{{alpha}} (HNF4{{alpha}}) Gene Impair Recruitment of p300, a Key Transcriptional Coactivator Mol. Endocrinol., July 1, 2001; 15(7): 1200 - 1210. [Abstract] [Full Text] [PDF]

				M. PONTOGLIO Hepatocyte Nuclear Factor 1, a Transcription Factor at the Crossroads of Glucose Homeostasis J. Am. Soc. Nephrol., November 1, 2000; 11(90002): S140 - S143. [Abstract] [Full Text] [PDF]

				B. L. Triggs-Raine, R. D. Kirkpatrick, S. L. Kelly, L. D. Norquay, P. A. Cattini, K. Yamagata, A. J. G. Hanley, B. Zinman, S. B. Harris, P. H. Barrett, et al. HNF-1alpha G319S, a transactivation-deficient mutant, is associated with altered dynamics of diabetes onset in an Oji-Cree community PNAS, April 2, 2002; 99(7): 4614 - 4619. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Submit a Letter to Editor Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Vaxillaire, M. Articles by Pontoglio, M. Search for Related Content PubMed PubMed Citation Articles by Vaxillaire, M. Articles by Pontoglio, M. Social Bookmarking                      What's this?