SEARCH

SEARCH BY CITATION

Keywords:

  • hyaline fibromatosis syndrome;
  • ANTXR2;
  • NMD;
  • degradation

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Hyaline fibromatosis syndrome is an autosomal recessive disease caused by mutations in ANTXR2, a gene involved in extracellular matrix homeostasis. Sixty percent of patients carry frameshift mutations at a mutational hotspot in exon 13. We show in patient cells that these mutations lead to low ANTXR2 mRNA and undetectable protein levels. Ectopic expression of the proteins encoded by the mutated genes reveals that a two base insertion leads to the synthesis of a protein that is rapidly targeted to the ER-associated degradation pathway due to the modified structure of the cytosolic tail, which instead of being hydrophilic and highly disordered as in wild type ANTXR2, is folded and exposes hydrophobic patches. In contrast, one base insertion leads to a truncated protein that properly localizes to the plasma membrane and retains partial function. We next show that targeting the nonsense mediated mRNA decay pathway in patient cells leads to a rescue of ANTXR2 protein in patients carrying one base insertion but not in those carrying two base insertions. This study highlights the importance of in-depth analysis of the molecular consequences of specific patient mutations, which even when they occur at the same site can have drastically different consequences.


Introduction

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Hyaline fibromatosis syndrome (HFS) is a rare autosomal recessive disease caused by mutations in ANTXR2 (MIM #608041) (also called Capillary Morphogenesis Gene 2, CMG2) [Dowling et al., 2003; Hanks et al., 2003]. The disease is progressive, disfiguring, and disabling. As suggested by the name, it is characterized by the accumulation of amorphous, unidentified hyaline material in the skin and other organs of patients [Shieh et al., 2006; Urbina et al., 2004], which could be either a direct effect or a secondary consequence due to the alterations in the basal membrane of capillary vessels [Stucki et al., 2001]. These noncancerous tissue proliferations are the most outstanding external hallmarks of the patients [Deuquet et al., 2011b]. In addition, patients usually present papular skin lesions, gingival hyperplasia, and joint contractures [Urbina et al., 2004]. Two allelic forms are now subsumed in HFS: infantile systemic hyalinosis (ISH; MIM #236490), the more severe form, whose patients have very early onset even at birth and suffer from severe diarrhea, recurrent infection and malnutrition that lead to death; and the milder form, juvenile hyaline fibromatosis (JHF; MIM #228600), for which afflicted individuals reach adulthood even though highly incapacitated by the cutaneous tumors [Deuquet et al., 2009]. Better pediatric care in severely affected infants allows some of them to survive infancy; these infants grow to develop the full-blown picture of HFS. Molecular results have confirmed that these disorders form a continuous phenotypic spectrum determined at least partially by the combination of specific ANTXR2 gene mutations [Deuquet et al., 2011a; 2011b].

The responsible gene ANTXR2 is expressed in all tissues except the brain, consistent with development of normal intelligence in patients [Deuquet et al., 2011b; Stucki et al., 2001]. ANTXR2 encodes a 55 kDa type I transmembrane (TM) protein, consisting of an extracellular N-terminal von Willebrand factor type A domain (vWA) followed by an Ig-like domain, a single TM helix and a cytosolic tail [Deuquet et al., 2011a; 2011b]. vWA domains, also called I-domain of integrins, contain a metal ion-dependent adhesion site (MIDAS) motif with which they interact with extracellular matrix proteins such as collagens [Engel and Chiquet, 2011]. The ANTXR2 vWA domain has been proposed to bind collagen IV and laminin [Bell et al., 2001]. The exact function of the ANTXR2 Ig-like domain has not been determined. It was however shown to contain two disulfide bonds essential for proper ANTXR2 folding in the endoplasmic reticulum (ER) [Deuquet et al., 2011a]. The ANTXR2 cytoplasmic tail is composed of 148 residues and contains multiple posttranslational modification sites such as palmitoylation [Abrami et al., 2006b], phosphorylation [Abrami et al., 2010], and ubiquitination [Abrami et al., 2010], modifications that are important for the surface dynamics and plasma membrane turn over of the protein [Deuquet et al., 2011b].

So far, more than 150 cases of HFS have been reported and some 34 different mutations, mostly in exons, have been identified [Deuquet et al., 2011b]. Among them, exon 13 is a hot spot for frameshift mutations, which include insertion of one or two bases—c.1073_1074insC (hereafter referred to as insC), c.1073_1074insCC (insCC)—and deletion of one base—c.1074delT (delT). These three frameshift mutations account for approximately 60% of all the pathogenic alleles. The frequency of insertions and deletions at positions 1,073–1,074 is likely due to its proximity to a low complexity, GC-rich region encoding a stretch of proline residues (Fig. 1A) that could constitute a vulnerable site for errors during DNA replication.

image

Figure 1. Analysis of frameshift mutations and the ectopic expression of ANTXR2insC, ANTXR2insCC, and ANTXR2delT. A: Amino-acid sequence alignment of ANTXR2WT, ANTXR2insCC, ANTXR2delT, and ANTXR2insC generated with Jalview. Sequences were highlighted according to the percentage of identity. B: HeLa cells were transfected or not (ctrl) for 48 hr with cDNA expressing either WT or mutant ANTXR2. Twenty micrograms of total cell extracts were analyzed by SDS-PAGE under nonreducing condition and Western blotting using the 2F6 anti-hANTXR2 and anticalnexin antibodies. C: TEM8 knocked-down HeLa cells grown on coverslips were transfected or not (ctrl) for 48 hr with cDNA expressing either WT or ANTXR2insC. TEM8 knockdown was performed using an shRNAi anti-TEM8 lentivector (Supp. Fig. S4). TEM8 silencing was performed in order to monitor PA binding to CMG2 only. Cells were incubated with biotin-labeled anthrax PA at a final concentration of 1 μg/ml on ice for 1 hr and then shifted to 37°C for 5 min before fixation and labeling with anti-hANTXR2 and streptavidin antibodies and Hoechst. The scale bars represent 10 μm. D: HeLa cells were transfected for 48 hr with cDNA expressing ANTXR2insC. Cells were then treated with or without (control) 500 ng/ml anthrax PA at 4°C for 1 hr followed by at 37°C for different time as indicated. Cell lysates were immunoprecipitated with an anti-hANTXR2 antibody and were subsequently analyzed by SDS-PAGE and Western blotting using anti-ubiquitin antibody, anti-PA antibody, and anti-hANTXR2 antibody.

Download figure to PowerPoint

Here, we have investigated the molecular consequences of the frameshift mutations in exon 13, which encodes the beginning of the cytosolic tail of ANTXR2. We first show that in fibroblasts derived from patients carrying such mutations, the mRNA level of ANTXR2 was drastically lower than in fibroblasts from healthy donors. This is most likely due to the generation of a premature stop codon that would be recognized by the nonsense-mediated mRNA decay (NMD) pathway [Rebbapragada and Lykke-Andersen, 2009]. This observation would point to the NMD pathway as a potential therapeutic target. We however first studied the effects of the insC, insCC, and delT mutations at the protein level by ectopic expression of the corresponding cDNAs in HeLa cells. In this system, we found the insCC and delT mutations lead to mutant ANTXR2 proteins—p.A359LfsX51 (hereafter referred to as ANTXR2insCC) and p.A359HfsX50 (ANTXR2delT)—that are rapidly targeted to ER-associated degradation (ERAD). Patients with these mutations would therefore not benefit from drugs targeting the NMD pathway and consistently silencing UPF1 (MIM #601430), a central factor of this mRNA decay pathway, did not rescue the ANTXR2insCC and ANTXR2delT protein expression. In contrast, mutant protein p.A359CfsX13 (ANTXR2insC) caused by insC, a mutation that occurs at the same hotspot but causes a different shift in the reading frame, was well expressed in HeLa cells and properly targeted to the cell surface, where it was competent for ligand binding. Importantly, in patient fibroblasts, silencing UPF1 led to a partial restoration of ANTXR2insC protein, to its proper targeting to the cell surface where it exhibited the ability to bind a surrogate ligand.

This study shows that certain HFS patients could benefit from therapies that would target the NMD pathway. They however also highlight the importance of an in-depth analysis of the consequences of specific mutations to determine the appropriate potential treatment.

Materials and Methods

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Cell Culture, Plasmids, Transfections, and Real Time q-PCR

Primary patient-derived fibroblasts were grown in DMEM medium (Gibco, Carlsbad, CA, USA) at 37°C complemented with fetal bovine serum (FBS), penicillin and streptomycin. HeLa cells (ATCC) were grown in complete Modified eagles medium (Sigma, St. Louis, MO, USA) at 37°C supplemented with FBS, l-glutamine, nonessential amino acids, penicillin, and streptomycin.

The human ANTXR2 (isoform 4, Swiss-Port P58335–4, GenBank AK091721.1) gene, tagged with a V5 epitope at the C-terminus, was cloned in the pcDNA3.1/V5-HIS-TOPO expression vector (Invitrogen, Carlsbad, CA, USA). Mutations were generated with the QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). The full-length cDNAs encoding EGFP and human CD4 (Gene: CD4, Swiss-Port P01730, GenBank M35160, MIM #186940) was cloned from pEGFP-C2 (Invitrogen, Carlsbad, CA, USA) and pCD4-CMX (a gift from D. Trono, Lausanne, VD, Switzerland), respectively, into the pcDNA3.1/V5-HIS-TOPO expression vector (Invitrogen, Carlsbad, CA, USA). To construct the plasmids expressing chimeric protein EGFP-tailmut or CD4-tailmut, the full-length EGFP, the 1,254 bp from 5′ end of CD4 and the 156 bp from 3′ end of delT were cloned. The EGFP or CD4 fragment was then ligated with the fragment corresponding to the mutant ANTXR2delT cytosolic tail with an NdeI cutting site designed at one end of them. The final cDNAs expressing the chimeric protein was inserted into the pcDNA3.1/V5-HIS-TOPO expression vector (Invitrogen, Carlsbad, CA, USA). The 49 amino acids corresponding to the ANTXR2delT cytosolic tail were synthesized by EZBiolab (Carmel, IN, USA).

For the real-time PCR, RNA was extracted from a 5 cm dish using the RNeasy kit (Qiagen, Venlo, Netherlands). One milligram of the total RNA extracted was used for the reverse transcription using random hexamers and superscript II (Invitrogen, Carlsbad, CA, USA). A 1:40 dilution of the cDNA was used to perform the real-time PCR using SyBr green reagent (Roche, Penzberg, Upper Bavaria, Germany). mRNA levels were normalized using three housekeeping genes: TATA-binding protein, ß-microglobulin, and ß-glucoronidase.

Toxin, Antibodies, and Reagents

The toxin subunit protective antigen (PA) and biotin-labeled PA were a gift from S. Leppla. The anti-PA antibody was purchased from List Biological Laboratories, Inc (Campbell, CA, USA). The antihuman ANTXR2 monoclonal antibodies 2F6, generated in our lab, was previously described [Deuquet et al., 2011a]. Polyclonal antibody against calnexin was from Eurogenetec (Seraing, Liege, Belgium), polyclonal goat anti human ANTXR2 and polyclonal goat anti human CD4 antibodies were from R&D Systems (Minneapolis, MN, USA), antiubiquitin antibody from Santa Cruz (Dallas, TX, USA), anti-EGFP-HRP antibody from the D. Trono laboratory (Lausanne, VD, Switzerland). Protein G-agarose-conjugated beads and Protein A-agarose-conjugated beads were from GE Healthcare (Little Chalfont, UK). MG132, from Sigma (St. Louis, MO, USA) and Bortezomib, from Santa Cruz (Dallas, TX, USA), were dissolved in DMSO and used at a final concentration of 10 μM in complete medium. Bafilomycin was from Sigma (St. Louis, MO, USA) and was dissolved in absolute ethanol and used at a final concentration of 0.1 μM. Leupeptin was purchased from Roche (Penzberg, Upper Bravaria, Germany), dissolved in water and used at a final concentration of 250 μg/ml. Bis-ANS (4,4′-dianilino-1,1′-binaphthyl-5,5′-disulfonic acid, dipotassium salt) was from Invitrogen (Carlsbad, CA, USA), dissolved in water and used at a final concentration of 50 μM. Treatments with EndoH were performed as previously described [Abrami et al., 2006b]. The fixation reagent BD CellFix and permeabilization reagent FACS Permeabilizing Solution 2 were from BD BioScience (Franklin Lakes, NJ, USA).

Immunoprecipitation and RNAi Experiments

siRNA of human UPF1 (isoform 1, Swiss-Port Q92900–1, GenBank CB039817.1) was purchased from Qiagen (Venlo, Netherlands). As control siRNA we used the following target sequence of the viral glycoprotein VSV-G: 5′attgaacaaacgaaacaagga 3′. For gene silencing, transfections of 100 nM siRNAs were carried out with Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) and the cells were analyzed 72 hr after transfection.

For immunoprecipitation, cells were washed three times with PBS and then were lysed by incubation for 30 min at 4°C with 0.5% NP-40, 500 mM Tris–HCl (pH 7.4), 20 mM EDTA, 10 mM NaF, 30 mM sodium pyrophosphate decahydrate, 2 mM benzamidine, 1 mM PMSF, 1 mM NEM, and a cocktail of protease inhibitors (Roche, Penzberg, Upper Bravaria, Germany). Cell lysates were centrifuged for 3 min at 2,000g and the supernatants were precleared with protein G-agarose conjugated beads and were incubated for 16 hr at 4°C with antibodies and beads. The beads were washed three times with the same lysis buffer and were resuspended in sample buffer (4×). The samples were heated at 95°C for 5 min and migrated on SDS-PAGE. Western blotting was performed with the iBlot (Invitrogen, Carlsbad, CA, USA) according to the manufacturers instructions. Quantification of the blots was done using the Typhoon Imager (Image QuantTool; GE Healthcare, Little Chalfont, UK).

Metabolic Labeling

For metabolic labeling, HeLa cells were transfected for 24 hr with respective cDNA and cells were starved with methionine/cysteine-free medium, incubated for a 5 min or 25 min pulse at 37°C with 50 μCi/ml [35S] methionine (Hartman Analytics, Braunschweig, Germany). For chasing experiments, already pulsed cells were incubated for different chase time period at 37°C in complete medium with a 10-fold excess of non-radioactive methionine and cysteine. Cells were then lysed with lysis buffer described above and immunoprecipitations were performed with cell lysates using protein A-agarose conjugated beads and respective antibodies. Samples were then migrated on SDS-PAGE after which the gel was incubated with a fixative solution (25% isopropanol, 65% H2O, 10% acetic acid), followed by a 30 min incubation with signal enhancer Amplify NAMP100 (Amersham, Little Chalfont, UK). The dried gels were exposed to a Hyperfilm MP (Amersham, Little Chalfont, UK) or quantified with the Typhoon Imager (Image QuantTool; GE Healthcare, Little Chalfont, UK).

Molecular Dynamics Simulations

Molecular dynamics (MD) simulations were used to further characterize the ANTXR2 cytoplasmic domain. As a first step, ANTXR2delT sequence and the corresponding section of the wild type (i.e., P342—S401) were modeled as a random coil. To explore efficiently the conformational space of these peptides, the MD simulations were carried out in the generalized Born implicit solvent. This solvent approximation, where the water molecules are described as a dielectric continuum, considerably reduces the computational time, allowing to reach hundreds of ns long simulations.

In a second step, the equilibrated cytoplasmic tails were linked to ANTXR2 helical TM domain. The resulting proteins were inserted and equilibrated in a 80 × 80 Å2 Palmitoyl Oleoly Phosphatidyl Choline membrane patch [Humphrey et al., 1996] to characterize its structure in a phospholipid bilayer.

All simulations were performed using NAMD [Phillips et al., 2005] engine, with the CHARMM27 force field [Brooks et al., 2009], including CMAP corrections. For the explicit solvent simulations, TIP3P water [Jorgensen et al., 1983] parameterization was used to describe the water molecules. The spatial overlapping of lipid molecules and protein were removed and the resulting protein-membrane system was solvated in variable-size water box, neutralized through the addition of NaCl at a concentration of 150 mM. The periodic electrostatic interactions were computed using the particle-mesh Ewald summation with a grid spacing smaller than 1 Å. All systems were first minimized by 2,000 conjugate gradient steps. All systems were subsequently gradually headed from 0 to 300 K in 800 psec with a constraint on the protein backbone scaffold. Finally, the systems were equilibrated for 10 nsec at 300 K. Free MD of all equilibrated system were performed with a 2 fsec integration time step using the RATTLE algorithm applied to all bonds. Constant temperature (300 K) was imposed by using Langevin dynamics [Brunger and Brooks, 1984], with damping coefficient of 1.0 psec. In explicit solvent simulations, constant pressure of 1 atm was maintained with a Langevin piston dynamics [Feller et al., 1995], 200 fsec decay period, and 50 fsec time constant.

First, we performed 300 nsec-long simulations for the characterization of each of the cytoplasmic tails folding in implicit solvent. Two simulations, lasting 100 nsec each, were then carried out on the model of ANTXR2 TM and cytoplasmic tail.

Hydrophobicity Plot, Disorder Tendency, and Secondary Structure Prediction

The hydrophobicity of sequences corresponding to the cytosolic tail of WT ANTXR2 or delT mutant (starting from A359 and H359 for ANTXR2WT and ANTXR2delT, respectively) was analyzed using the online ExPASy tool ProtScale (http://web.expasy.org/protscale/). The disorder tendency was predicted with IUPred (http://iupred.enzim.hu/). All data were extracted and plotted with Prism graphpad 5. No structural data are currently available for ANTXR2 wild type and delT mutant cytoplasmic domain. We used secondary structure prediction tools to estimate the structural content of these domains: HNN, Jpred, NetSurfP, PSIpred, ProteinPredict. The final estimated structure was obtained by averaging the output of these different algorithms.

Circular Dichroism

Circular dichroism (CD) experiments were carried out at room temperature on a Jasco J-815 spectrometer. Cells with 0.1 cm path length were used. The synthetic peptide of ANTXR2delT was diluted to a concentration of 20 μM in a buffer containing 20 mM Na2HPO4 (pH 7.4) and 40% of acentonitrile. The buffer spectrum was subtracted for each sample.

Flow Cytometric Analysis

To permeabilize the cells, cells were fixed with BD CellFix reagent for 15 min and permeabilized by FACS Permeabilizing Solution 2 for 20 min at room temperature. Either permeabilized or nonpermeabilized cells were incubated with anti-hANTXR2 antibody for 30 min on ice. And after two times of wash with PBS + 20 mM EDTA + 2% FCS, the cells were then incubated with a fluorescent secondary antibody for another 30 min on ice. Cells were washed for three times before they were fixed in the end. The stained cells were evaluated with Accuri C6 (BD BioScience, Franklin Lakes, NJ, USA). FACS data were analyzed using FlowJo software (FlowJo, Tree Star Inc., Ashland, OR, USA).

ANS Binding Assay

Binding of ANS was measured in 96-well plate with the spectrofluorimeter SpectraMax plate reader (Molecular Device, Sunnyvale, CA, USA). An excitation wavelength of 360 nm and an emission wavelength of 470 nm were applied. Synthetic peptides were dissolved and diluted to a final concentration of 0.1 mg/ml with 20 mM sodium phosphate, pH 7.4, and 2% acetonitrile. The final concentration of ANS was 50 μM.

Immunofluorescence Microscopy

HeLa cells were fixed with 3% paraformaldehyde for 20 min and labeled with an anti-hANTXR2 antibody followed by Alexa-conjugated anti-rat (555 nm) and Alexa-conjugated streptavidin antibody (647 nm). The nuclei were stained by Hoeschst dye. Images were acquired using a 63× lens on LSM-710 Laser scanning microscope (Carl Zeiss Microimaging, Inc., Oberkochen, Germany). The Fiji software was used for the processing of the images.

DNA Numbering System and Variants Description

For cDNA numbering, the nucleotide numbering reflects cDNA numbering with +1 corresponding to the A of the ATG translation initiation codon in the reference sequence, according to journal guideline (www.hgvs.org/mutnomen). The initiation codon is codon 1.

All the variants of ANTXR2 reported in this article can be found in the LSDB Leiden Open Variation Database at www.lovd.nl/ANTXR2, with the exception of the p.I198T variant, which can be found in the LSDB Human Gene Mutation Database at http://www.hgmd.cf.ac.uk/ac/gene.php?gene=ANTXR2&accession=CM033743.

Results

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Analysis of Frameshift Mutations in HFS Patients

The three frameshift mutations in exon 13 of ANTXR2 that have been identified in HFS patients all lead to premature stop codons. The mutated insC, insCC, and delT genes encode for ANTXR2 proteins that have modified and shorter cytosolic tails of 29, 67, 66 amino acids, respectively, compared with the 148 residue wild type ANTXR2 tail (Fig. 1A.) Note that the insCC and delT mutations lead to the same altered reading frame and thus to the same cytosolic tail, with the exception of one additional residue at position 359 of ANTXR2insCC.

To understand the consequences of these frameshift mutations, we analyzed fibroblasts from five previously described patients (Table 1) [Deuquet et al., 2011a; Dowling et al., 2003]. Patient 1 is homozygous for the insC single nucleotide insertion [Deuquet et al., 2011a]. All other patients are compound heterozygotes, carrying one of the three frameshift mutations on one allele and one missense mutation on the other. Patient 2 [Dowling et al., 2003] carries the insC mutation and a c.566T>C transversion, which causes a p.I198T amino-acid substitution (Table 1). Patient 3 [Deuquet et al., 2011a] carries the insC mutation and a c.928G>T transversion resulting in a p.V310F missense mutation (Table 1). Patient 4 [Hanks et al., 2003] carries the insCC mutation and a missense mutation c.566T>C as patient 2. Patient 5 [Deuquet et al., 2011a] carries the delT mutation and a c.116G>T transversion leading to the p.C39F amino-acid substitution (Table 1). All patients suffer from the severe infantile form of HFS, with the exception of patient 3, who suffers from a somewhat milder form possibly due to better pediatric care [Deuquet et al., 2011a; Dowling et al., 2003].

Table 1. HFS Mutations Analyzed in This Article
PatientHFSZygocityVariantDNADNA AliasProteinProtein AliasExonReferences
  1. Nucleotide numbering reflects cDNA numbering with +1 corresponding to the A of the ATG translation initiation codon in the reference sequence, according to the nomenclature proposed by the Human Genome Variation Society (www.hgvs.org/mutnomen). The initiation codon is codon 1. The GenBank number of human ANTXR2 is AK091721.1.

1InfantileHomozygotec.1073_1074insCc.1073dupinsCp.A359CfsX13ANTXR2insC13Deuquet et al. (2011)
2InfantileCompoundc.566T>Cc.566T>C p.I189T 7Dowling et al. (2003)
  Heterozygotec.1073_1074insCc.1073dupinsCp.A359CfsX13ANTXR2insC13 
3JuvenileCompoundc.928G>Tc.928G>T p.V310F 11Deuquet et al. (2011a,b)
  Heterozygotec.1073_1074insCc.1073dupinsCp.A359CfsX13ANTXR2insC13 
4InfantileCompoundc.566T>Cc.566T>C p.I189T 7Hanks et al. (2003)
  Heterozygotec.1073_1074insCCc.1073_1074dupinsCCp.A359LfsX51ANTXR2insCC13 
5InfantileCompoundc.116G>Tc.116G>T p.C39F 1Deuquet et al. (2011a,b)
  Heterozygotec.1074delTc.1074deldelTp.A359HfsX50ANTXR2delT13 

We performed a real-time quantitative PCR analysis of mRNA extracted from the patient fibroblasts [Deuquet et al., 2011a]. The ANTXR2 mRNA levels were significantly lower in patient cells than the healthy control fibroblasts (Supp. Fig. S1), consistent with our previous findings [Deuquet et al., 2011a]. Patient 1, the only one carrying a frameshift mutation on both alleles, showed the lowest mRNA level. The compound heterozygous patients had ANTXR2 mRNA levels ranging from 46% to 72% of that of the control fibroblasts. The decrease of mRNA level observed in patient-derived fibroblasts is presumably due to the NMD pathway, which can recognize the presence of a premature stop codon in mRNA caused by frameshift mutations [Rebbapragada and Lykke-Andersen, 2009]. The NMD pathway would thus target both alleles in patient 1 and only one allele in all other patients.

We next analyzed protein extracts from the patient cells by Western blotting using an anti-human ANTXR2 monoclonal antibody [Deuquet et al., 2011a]. As expected, a band corresponding to full-length ANTXR2 (55 kDa) was readily detected for control fibroblasts. A weaker band of similar molecular weight was detected for cells from the compound heterozygous patients as previously observed [Deuquet et al., 2011a]. However, none of the truncated ANTXR2 forms (ANTXR2insC: 42 kDa, ANTXR2insCC and ANTXR2delT: 45 kDa) could be detected, even at high exposure (Supp. Fig. S2) or following immunoprecipitation using an antibody against the N-terminal ectodomain of ANTXR2 (data not shown), consistent with the low mRNA levels.

Ectopic Expression of ANTXR2insC, ANTXR2insCC, and ANTXR2delT

The above observations show that the mRNA levels of the frame-shifted ANTXR2 variants are very low, which could account for the absence of ANTXR2, and suggest that the NMD pathway might be a potential therapeutic target. Before exploring this option, we however first investigated whether the ANTXR2insC, ANTXR2insCC, and ANTXR2delT proteins can be properly expressed in tissue culture cells upon transfection with the appropriate cDNAs.

Unexpectedly, ANTXR2insCC and ANTXR2delT were essentially undetectable by Western blotting whereas ANTXR2insC, which migrated at an expected molecular weight of 42 kDa, was expressed at a similar level as ANTXR2WT (Fig. 1B). The low expression of ANTXR2insCC and ANTXR2delT was not due to reduced synthesis as shown by performing a 5-min metabolic pulse (Supp. Fig. S3).

Immunofluorescence on nonpermeabilized cells using a monoclonal antibody directed toward the ectodomain of ANTXR2 showed that ectopically expressed ANTXR2insC was properly targeted to the cell surface (Fig. 1C). Because ANTXR2insC lacks most of the WT tail, it is unlikely to retain the full ANTXR2 function. It might however retain some of the WT signaling capacity. To test this possibility, we analyzed whether ANTXR2insC was able to binding a ligand and respond to this event. HeLa cells were transfected with ANTXR2insC and we monitored binding of the anthrax PA, used as a surrogate ligand [Deuquet et al., 2011b; van der Goot and Young, 2009b]. PA is initially an 83 kDa (PA83) protein, which can be processed by cell surface proteases to a 63 kDa form (PA63) [van der Goot and Young, 2009a]. This mature form can undergo heptamerization, endocytosis, and conversion to an SDS-resistant heptameric form (PA7mer) [van der Goot and Young, 2009a]. We have previously shown that extracellular addition of PA leads to ubiquitination of ANTXR2 in its cytosolic tail [Abrami et al., 2010; 2006a]. As shown both by Western blotting (Fig. 1D) and by immunofluorescence (Fig. 1C), ANTXR2insC was PA-binding competent, in agreement with a previous report [Liu and Leppla, 2003]. Moreover, PA underwent conversion from the 83 kDa form to the SDS-resistant heptameric form (Fig. 1D) indicating that it is capable of undergoing toxin-induced endocytosis [van der Goot and Young, 2009a]. Importantly, as observed for the WT ANTXR2 [Abrami et al., 2010; 2006a], the toxin triggered ANTXR2insC ubiquitination (Fig. 1D). These observations indicate that ANTXR2insC is ligand-binding competent and has at least partial WT signaling competence. ANTXR2insC must however lack some functional properties since point mutations in the cytosolic tail have been reported in HFS patients [Deuquet et al., 2011b]. The precise consequences of the later mutations have not yet been addressed. Patients homozygous for such mutations however suffer from the mildest form of the disease as opposed to patients carrying the c.1073_1074insC mutation [Deuquet et al., 2011b]. Altogether these observations suggest that if the ANTXR2 mRNA could be stabilized in patients carrying the c.1073_1074insC mutation, symptoms might be attenuated.

ANTXR2insCC and ANTXR2delT are Powerful ERAD Substrates

The absence of detectable ANTXR2insCC/delT upon ectopic expression, despite normal synthesis, indicates that these two ANTXR2 mutants undergo accelerated degradation. A metabolic pulse-chase analysis revealed that indeed the two mutant ANTXR2 proteins had a half-life of about 20 min (Fig. 2A). Degradation of ANTXR2insCC and ANTXR2delT could be inhibited by treating cells with the proteasome inhibitor MG132 (Fig. 2B and C) [Kisselev and Goldberg, 2001]. Not only were the levels of metabolically labeled proteins after a 25 min pulse similar to that of ANTXR2WT (Fig. 2B and C), but the proteins could also be detected by Western blotting following 4 or 8 hr MG132 treatment (Fig. 2D). MG132 did not affect the level of ANTXR2insC (Fig. 2D). No rescue of ANTXR2insCC and ANTXR2delT was observed following treatment with either Bafilomycin A, an inhibitor of the vacuolar ATPase [Drose and Altendorf, 1997] or leupeptin, an inhibitor of lysosomal proteases [Aoyagi et al., 1969]. These observations indicate that ANTXR2insCC and ANTXR2delT are rapidly degraded by the proteasome and not in lysosomes.

image

Figure 2. ANTXR2insCC and ANTXR2delT are powerful ERAD substrates. A: HeLa cells were transfected or not (ctrl) for 24 hr with cDNA expressing either WT or mutant ANTXR2. Cells were pulsed for 5 min with [35S] cysteine/methionine and then chased for different time periods. Immunoprecipitation was performed using anti-hANTXR2 antibody and samples were analyzed by SDS-PAGE and autoradiography. Quantification was performed using the Typhoon Imager. For each experiment, the intensities of signals were normalized to that of the same cDNA at time t = 0 min. Error bars represent standard errors (n = 3). The curve was smoothed by nonlinear fit according to the one phase decay model using the software Prism graphpad 5. B: HeLa cells were transfected or not (ctrl) for 24 hr with cDNA expressing either WT or mutant ANTXR2. Cells were treated with 10 μM MG132 or not for 4 hr before being pulsed for 25 min with [35S] cysteine/methionine. Immunoprecipitation was performed using anti-hANTXR2 antibody and samples were analyzed by SDS-PAGE and autoradiography. C: Quantification for WT or mutant ANTXR2 with or without MG132 treatment in the 25 min radiolabeling experiments was performed using the Typhoon Imager. Error bars represent standard errors (n = 3). D: HeLa cells were transfected or not (ctrl) for 48 hr with cDNA expressing either WT or mutant ANTXR2. Cells were treated or not (ctrl) with 10 μM MG132 for 4 or 8 hr, or with 100 μM Bafilomycin for 4 hr, or with 250 μg/ml leupeptin for 16 hr. Twenty micrograms of total cell extracts were analyzed by SDS-PAGE under non-reducing condition and Western blotting using the 2F6 anti-hANTXR2 antibody and anti-calnexin antibody. E: HeLa cells were transfected or not (ctrl) for 48 hr with cDNA expressing either WT or mutant ANTXR2 and were treated or not with 10 μM MG132 for 4 hr. Cell lysates were subjected to immunoprecipitation with an anti-hANTXR2 antibody. The products from immunoprecipitation were treated with or without EndoH and were analyzed by SDS-PAGE and Western blotting using anti-hANTXR2 antibody. F: HeLa cells were transfected or not (ctrl) for 48 hr with cDNA expressing either WT or mutant ANTXR2. Cells were either treated with or without 10 μM MG132 for 4 hr or treated with 500 ng/ml anthrax PA at 4°C for 1 hr followed by at 37°C for 10 min. Cell lysates were immunoprecipitated with an anti-hANTXR2 antibody and were subsequently analyzed by SDS-PAGE and Western blotting using anti-ubiquitin antibody and anti-hANTXR2 antibody.

Download figure to PowerPoint

We next investigated whether the MG132 rescued proteins were properly trafficked through the biosynthetic pathway. We have previously shown that ANTXR2 is glycosylated [Abrami et al., 2006b; Deuquet et al., 2009]. Because of the modifications of the N-linked glycans in the Golgi, ANTXR2 migrates as a smear that is insensitive to treatment with Endoglycosidase H (Endo H), an enzyme that only removes the noncomplex oligosaccharides. Although the majority of expressed ANTXR2WT and ANTXR2insC were insensitive to EndoH treatment, MG132 rescued ANTXR2insCC and ANTXR2delT, which migrated as a well-defined band, were EndoH sensitive (Fig. 2E), indicating that these mutant ANTXR2 molecules were unable to reach the Golgi. Altogether, the above observations show that ANTXR2insCC and ANTXR2delT are rapidly degraded by the ERAD pathway and that even MG132-rescued ANTXR2insCC and ANTXR2delT cannot exit the compartment. This is in marked contrast to previously characterized HFS ANTXR2 mutants for which surface expression could be rescued with proteasome inhibitors [Deuquet et al., 2011a]. Consistent with the notion that ANTXR2insCC and ANTXR2delT are efficient ERAD substrates, we found that the MG132 rescued proteins were polyubiquitinated (Fig. 2F), a hallmark of proteins destined to degradation. This ubiquitination signature was markedly different from the monoubiquitination of ANTXR2WT observed during anthrax toxin endocytosis (Fig. 2F, cells treated with PA) [Abrami et al., 2010], and far more pronounced than for WT ANTXR2 in the presence of MG132, especially when taking into account the difference in protein levels (Fig. 2F).

Structural Predictions of the ANTXR2insCC/delT Tail

In order to understand why ANTXR2insCC/delT are powerful ERAD substrates, we analyzed the primary sequence of their cytosolic tail (Fig. 1A) and compared it to that of the WT ANTXR2 as well as ANTXR2insC. Little is known about the structure of the cytosolic tail of ANTXR2WT. We tested five different secondary structure prediction algorithms, which all indicated that the ANTXR2WT tail is mostly composed of random coil as is the much shorter ANTXR2insC tail (Fig. 3A). We therefore performed disorder predictions [Dosztanyi et al., 2005], which indeed predicted that residues 342 to ≈450 are highly disordered (Fig. 3B, left panel). Disordered domains generally are rich in charged residues and poor in hydrophobic residues [Dyson and Wright, 2005; Tompa, 2012] and consistently, the calculated hydrophobic index (Gasteiger et al., 2005) was negative along the entire length of the cytoplasmic tail of ANTXR2WT as it was for ANTXR2insC (Fig. 3C).

image

Figure 3. Structural predictions of the ANTXR2insCC/delT tail. A: Sequence and structural content for WT ANTXR2, ANTXR2delT, and ANTXR2insC cytoplasmic tail. The sequences were color coded with polar residues in black, nonpolar in green, positively and negatively charged residues in red and blue, respectively. The predicted secondary structure was the average output of five secondary structure prediction tools. B: Disorder prediction for the cytosolic tails of WT ANTXR2, ANTXR2delT, and ANTXR2insC. The curves’ surfaces above the threshold of 0.5 were colored in red, representing the predicted disordered portion. C: Hydrophobicity plot of the cytosolic tails of WT ANTXR2 and ANTXR2delT (starting from A359, H359, and C359 for WT ANTXR2, ANTXR2delT, and ANTXR2insC, respectively). The curves’ surfaces below the value of zero were colored in blue, representing the hydrophilic portion whereas the surfaces above zero in red, representing the hydrophobic portion. D: Molecular dynamics based model of ANTXR2delT cytosolic tail. The modeled structure was linked to ANTXR2 TM domain. The protein was color coded as follows: polar residues in white, nonpolar in green, acidic and basic residues in red and blue, respectively. The cytoplasmic hydrophobic core interacts with the membrane surface. Side chains of critical residues are shown in ball and sticks.

Download figure to PowerPoint

Similar analyses on the ANTXR2delT tail led to markedly different results. The ANTXR2delT tail was predicted to be largely structured, based both on secondary structure (Fig. 3A) and disorder (Fig. 3B, middle panel) predictions. Moreover, most of the tail showed a positive hydrophobic index (Fig. 3C, middle panel).

To further strengthen these predictions, we investigated the structure of the ANTXR2delT tail using MD under physiological conditions. The ANTXR2 helical TM domain was inserted into a lipid membrane bilayer and linked to the ANTXR2delT cytosolic tail initially modeled as random coil. During the MD simulation, the tail spontaneously folded into a domain showing three structured α-helices, in good agreement with the secondary structure prediction for the ANTXR2delT tail (Fig. 3A). The hydrophobic core produced by the helical arrangement stabilized the cytosolic domain in a globular fold (Fig. 3C and D), which remained in close contact with the membrane surface due to the presence of several lysine residues (K361, K365, K399, and K404) interacting with the lipid head-groups.

Thus the insCC and delT hotspot mutations lead to the synthesis of a cytosolic tail that has drastically different properties from that of the original ANTXR2 tail. Instead of being intrinsically disordered and hydrophilic, the ANTXR2insCC/delT tail is predicted to form a compact well-folded domain. This domain however exposes hydrophobic patches that are likely to be recognized by the ER quality control machinery and thus targeted to ERAD [Hegde and Ploegh, 2010]. These findings are rather counterintuitive in the sense that the native ANTXR2 tail despite its lack of predicted structure is not seen by the ER quality control system as problematic. In contrast, the ANTXR2insCC/delT tail, although well structured, is immediately spotted.

The ANTXR2insCC/delT Tail Sequence is Sufficient for Targeting to Degradation

In order to experimentally validate some of the above predictions, we performed CD on a synthetic peptide corresponding to the sequence of the ANTXR2delT tail. The far UV CD spectrum of the peptide was typical of that of polypeptide rich in α-helices, with two negative peaks at 222 and 208 nm (Fig. 4A). We next analyzed the capacity of the peptide to bind ANS acid, a fluorescent dye that increases its quantum yield when binding hydrophobic patches [Slavik, 1982]. As a control peptide, we used a 40 residues peptide of the human ß-amyloid [Iacovache et al., 2011]. Unfortunately, we were unable to produce the WT ANTXR2 cytosolic tail recombinantly, presumably due to its unstructured nature. As shown in Figure 4B, the ANTXR2delT tail peptide showed strong ANS binding capacity, confirming the predicted presence of surface-exposed hydrophobic patches.

image

Figure 4. The ANTXR2insCC/delT tail sequence is sufficient for targeting to degradation. A: Far UV CD spectra of the cytoplasmic tails of ANTXR2delT. A synthetic peptide corresponding to the cytoplasmic tail of ANTXR2delT (starting from H359) was dissolved in 20 mM sodium phosphate (pH 7.4) with 40% acetonitrile and its far UV CD spectrum was analyzed with a Jasco J-815 spectrometer. B: Binding of the hydrophobic dye ANS to control (ctrl) and ANTXR2delT cytoplasmic tail. Synthesized peptides of ctrl and delT cytoplasmic tail were dissolved in 20 mM sodium phosphate (pH 7.4) with 2% acetonitrile to a final concentration of 0.1 mg/ml, respectively. Fifty micromolars of ANS was added to each sample and the relative fluorescence unit (RLU) was measured with a SpectraMax plate reader (Molecular Device). The graph showed was one representative result. C: HeLa cells were transfected or not (ctrl) for 48 hr with cDNA expressing either WT EGFP or mutated tail (cytoplasmic tail of ANTXR2delT starting from H359) fused chimeric EGFP. Cells were treated or not with 10 μM MG132 for 4 or 8 hr. Twenty micrograms of total cell extracts were analyzed by SDS-PAGE and Western blotting using anti-EGFP antibody and anti-calnexin antibody. To ensure that proteins were well synthesized, 24 hr-transfected or not HeLa cells with or without 1 hr 10 μM MG132 treatment were pulsed for 5 min with [35S] cysteine/methionine in parallel. Immunoprecipitation was performed using anti-EGFP antibody and the samples were analyzed by SDS-PAGE and autoradiography. D: HeLa cells were transfected or not (ctrl) for 48 hr with cDNA expressing either WT CD4 or chimeric protein composed of the ectodomain and TM part of CD4 and the mutated cytoplasmic tail of ANTXR2delT (starting from H359). Cells were treated or not with 10 μM MG132 for 4 or 8 hr. Twenty micrograms of total cell extracts were analyzed by SDS-PAGE and Western blotting using anti-CD4 antibody and anti-calnexin antibody. To ensure that proteins were well synthesized, 24 hr-transfected or not HeLa cells with or without 1 hr 10 μM MG132 treatment were pulsed for 5 min with [35S] cysteine/methionine in parallel. Immunoprecipitation was performed using anti-CD4 antibody and the samples were analyzed by SDS-PAGE and autoradiography.

Download figure to PowerPoint

We next analyzed whether the ANTXR2delT tail sequence is sufficient to target a TM protein to ERAD. We therefore generated a chimera between CD4 and the ANTXR2delT tail. CD4 is a type I TM protein widely used in the construction of chimeric proteins to study membrane trafficking and protein targeting [Jackson et al., 1990; Nilsson et al., 1989; Shin et al., 1991]. It has a very stable ectodomain containing three disulfide bonds, a single TM helix, and a 40 residues cytoplasmic tail. This tail was swapped with that of ANTXR2delT generating CD4-tailmut. Although CD4 expression was readily detectable upon ectopic expression in HeLa cells, CD4-tailmut expression was very low (Fig. 4C, top panel), even though synthesis was normal (Fig. 4C, bottom panel). This was due to post-translational degradation via the proteasome, as revealed by the ability of MG132 to rescue expression (Fig. 4C, top panel). We also fused the ANTXR2delT tail sequence to a soluble cytosolic protein, namely EGFP (EGFP-tailmut), and found that it was also rapidly targeted to degradation via the proteasome (Fig. 4D).

Thus, the ANTXR2delT tail sequence is a well-folded domain, which exposes hydrophobic patches. Most likely due to this exposed hydrophobicity, the domain is efficiently recognized by quality control systems both at the surface of the ER and in the cytosol, targeting the protein to proteasomal degradation.

These observations show that targeting the NMD pathway in patients carrying the insCC or delT frameshift mutations is pointless, since the mRNA that would be rescued encodes for a protein that is immediately recognized by the folding quality control machinery and targeted to degradation at the ER level.

Inhibiting NMD Pathway Can Rescue ANTXR2insC

Figure 1 illustrates that, in contrast to ANTXR2insCC and ANTXR2delT, ectopically expressed ANTXR2insC is normally trafficked through the biosynthetic pathway, and reaches the cell surface where it can bind a surrogate ligand and signal to the interior of the cell, altogether indicating that the protein is at least partially functional.

We therefore investigated whether blocking the NMD pathway in patient fibroblasts, to inhibit mRNA degradation, would lead to a partial rescue of ANTXR2insC. We chose to silence the expression of UPF1 (MIM #601430), a member of the group I RNA helicase and ATPase family [Bhattacharya et al., 2000], which is crucial for the initiation of the NMD pathway [Isken et al., 2008]. Knockdown of UPF1 was recently shown to rescue the phenotypes of Ulrich Disease and long-QT Syndrome in fibroblasts [Gong et al., 2011; Usuki et al., 2006]. UPF1 expression could be efficiently silenced by siRNA in fibroblasts from patients 1, 4 and 5, which harbor the insC, insCC, and delT, respectively, as well as in fibroblasts from a healthy control (Fig. 5A). Remarkably, ANTXR2insC could be detected by Western blotting (Fig. 5B) and by FACS analysis (Supp. Fig. S5 and S6) upon UPF1 silencing in patient 1 cells. Because of the homozygosity of patient 1 and the low level of ANTXR2 mRNA expressed in the fibroblasts, we could detect a significant increase in ANTXR2 mRNA upon UPF1 silencing (Supp. Fig. S7). Consistent with the ectopic expression studies (Fig. 1), ANTXR2insC rescued by UPF1 silencing was properly targeted to the cell surface of fibroblasts (Fig. 5C and D). No rescue of ANTXR2insCC and ANTXR2delT could be observed upon UPF1 silencing in cells from patients 4 and 5 (Fig. 5B), as expected from their efficient targeting to ERAD.

image

Figure 5. Inhibiting NMD pathway can rescue ANTXR2insC in patient-derived fibroblasts. A: mRNA levels of UPF1 in patient-derived fibroblasts after 72 hr knocking down with or without siRNA targeting UPF1 were analyzed by quantitative RT-PCR. Under the condition that cells were without siRNA targeting UPF1, a control siRNA against VSVG was used. The values representing each mRNA level were plotted as the percentage of fibroblast without siRNA targeting UPF1. Error bars represent standard errors (n > 4). Two-tailed paired t-test analysis between the same patient-derived fibroblasts with or without siRNA targeting UPF1 was performed. ***P < 0.001. B: Ctrl1 and patient-derived fibroblasts were transfected with or without siRNA targeting UPF1 for 72 hr. Under the condition that cells were without siRNA targeting UPF1, a control siRNA against VSVG was used. Ten micrograms of total cell extracts were analyzed by SDS-PAGE under non-reducing condition and Western blotting using anti-hANTXR2 antibody and anti-calnexin antibody. C: Ctrl1 and Patient1 (P1) derived fibroblasts were transfected with or without siRNA targeting UPF1 for 72 hr. Under the condition that cells were without siRNA targeting UPF1, a control siRNA against VSVG was used. Nonpermeabilized cells were stained with anti-hANTXR2 antibody for 30 min on ice followed by a secondary fluorescent antibody staining for 30 min. The cells were then washed, fixed and analyzed on a FACS analyzer Accuri C6 (BD BioScience). D: The mean fluorescence intensity of each sample from the FACS data shown above was extracted with the software FlowJo 7.6. Mean fluorescence intensities representing the surface ANTXR2 were plotted as the percentage of that of Ctrl1 without siRNA targeting UPF1. Error bars represent standard errors (n = 4). Two-tailed paired t-test analysis was performed. *: P < 0.05.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

HFS is a disease that is due to the loss of ANTXR2 functions [Deuquet et al., 2011b]. Mutations can lead to loss of function through various scenarios: frameshift mutations that lead to premature stop codons can trigger mRNA degradation, point mutations can lead to defects in protein folding of in loss of function of a correctly folded protein. We have previously shown that missense mutations reported in patients that map to the extracellular domain of ANTXR2 mostly affect folding in the ER and thus lead to degradation [Deuquet et al., 2009, 2011a) and that proteasome inhibitors can lead to rescued functional ANTXR2. Here, we have analyzed the most frequent mutational site, which is in exon 13. Mutations however include single or double base insertion or single base deletion.

The present analysis allows us to draw several interesting conclusions. Firstly, it highlights the importance of in-depth analyses of the molecular consequences of the specific mutations found in a given patient, be it HFS or another monogenic disease, in order to evaluate the potential therapeutic targets. Indeed not only the site of a mutation is crucial but the exact nature of the insertion/deletion, that is, the change in reading frame. Secondly, our study underlines that the genotype–phenotype analyses should be performed at the mRNA, the protein and the functional level, before significant conclusions can be drawn in terms of therapeutic strategy and targets. And finally, more specific to HFS, our study shows that the HFS c.1073_1074insCC and c.1074delT mutations lead to defects not only at the mRNA level but also at the protein folding level. This leaves little hope for a chemical based therapy to rescue ANTXR2 expression levels and thereby function recovery, leaving gene replacement as the only possibility in these patients. In contrast, a potential drug based therapy is not to be excluded for patient carrying the c.1073_1074insC mutation, fortunately the most frequent of the three frameshift mutations [Deuquet et al., 2011b]. Indeed our study indicates that rescuing the ANTXR2 mRNA in these patients allows the synthesis of a viable protein and points to the NMD pathway as a potential therapeutic target. Consistent with the differences, we observe in terms of protein properties between ANTXR2insC and ANTXR2insCC/delT, a survey of the literature indicates that patients harboring the c.1073_1074insCC or c.1074delT mutations always suffer from the severe form of the disease, whereas more moderate symptoms are reported for patients with the c.1073_1074insC mutations, further raising the hope that even limited mRNA rescue could have an impact on the severity of the symptoms.

Acknowledgments

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We are very grateful to the patients and their parents for accepting to be part of this study and for their collaborative attitude. We thank the Associazione I.S.I (www.associazioneisi.it) for linking patients, and their families, suffering for HFS. We thank Béatrice Kunz for performing the qPCR analysis, Ioan Iacovache and Bruno Fauvet for the CD, Miguel Garcia and his platform for helping with the FACS analysis, Asvin Lakkaraju and Sarah Friebe for the microscopy.

Disclosure statement: The authors declare no conflict of interest.

References

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
humu22324-sup-0001-S1.pdf290KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.