Background IL-7 and IL-15 are produced by hepatocytes and are critical for the expansion and function of CD8 T cells. IL-15 needs to be presented by IL-15Rα for efficient stimulation of CD8 T cells.
Methods We analysed the hepatic levels of IL-7, IL-15, IL-15Rα and interferon regulatory factors (IRF) in patients with chronic hepatitis C (CHC) (78% genotype 1) and the role of IRF1 and IRF2 on IL-7 and IL-15Rα expression in Huh7 cells with or without hepatitis C virus (HCV) replicon.
Results Hepatic expression of both IL-7 and IL-15Rα, but not of IL-15, was reduced in CHC. These patients exhibited decreased hepatic IRF2 messenger RNA levels and diminished IRF2 staining in hepatocyte nuclei. We found that IRF2 controls basal expression of both IL-7 and IL-15Rα in Huh7 cells. IRF2, but not IRF1, is downregulated in cells with HCV genotype 1b replicon and this was accompanied by decreased expression of IL-7 and IL-15Rα, a defect reversed by overexpressing IRF2. Treating Huh7 cells with IFNα plus oncostatin M increased IL-7 and IL-15Rα mRNA more intensely than either cytokine alone. This effect was mediated by strong upregulation of IRF1 triggered by the combined treatment. Induction of IRF1, IL-7 and IL-15Rα by IFNα plus oncostatin M was dampened in replicon cells but the combination was more effective than either cytokine alone.
Conclusions HCV genotype 1 infection downregulates IRF2 in hepatocytes attenuating hepatocellular expression of IL-7 and IL-15Rα. Our data reveal a new mechanism by which HCV abrogates specific T-cell responses and point to a novel therapeutic approach to stimulate anti-HCV immunity.
- Chronic Viral Hepatitis
- Gene Expression
- Immune Response
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Chronic infection caused by HCV is characterised by poor HCV-specific CD8 T-cell responses, which display an exhausted phenotype and are unable to control viral replication.
IL-7 is a cytokine produced by hepatocytes and stromal cells essential for the differentiation and maintenance of memory CD8 T cells, which can prevent their functional exhaustion.
IL-15, a cytokine presented by the IL-15Rα expressed on the membrane of dendritic and epithelial cells, is crucial for the activation, survival and differentiation of both memory and naive CD8 T cells.
What are the new findings?
The expression of both IL-7 and IL-15Rα is significantly diminished in livers from patients with chronic hepatitis C and in cells harbouring HCV genotype 1 replicon.
Basal expression of both IL-7 and IL-15Rα is controlled by IRF2, a molecule that is downregulated in hepatocytes from patients with CHC and in replicon cells.
The combination of IFNα plus oncostatin M stimulates the expression of IL-7 and IL-15Rα in hepatocytes, an effect mediated by a vigorous upregulation of IRF1.
How might it impact on clinical practice in the foreseeable future?
These data reveal a new mechanism by which HCV could attenuate antiviral T-cell responses.
Our findings point to a novel therapeutic approach to stimulate anti-HCV immunity.
A large body of evidence indicates that CD8 T cells are major players in the elimination of hepatitis C virus (HCV) infected hepatocytes.1–3 CD8 T cells exert antiviral effects by cytolytic and non-cytolytic effector functions.4 ,5 While robust CD8 T-cell responses are induced in acute infection, in patients with chronic hepatitis C (CHC) virus-specific CD8 T cells are inefficient at controlling viral replication as they exhibit various degrees of exhaustion, with a progressive decline in their cytolytic activity and in their ability to produce cytokines mainly interferon (IFN) γ, interleukin (IL)-2 and tumour necrosis factor (TNF) α.1–3 ,6 ,7
IL-7 and IL-15 are key cytokines controlling the homeostasis of naive and memory T cells.8 ,9 IL-7 is produced constitutively by hepatocytes and by stromal cells,10 and has a non-redundant role in immune development and homeostasis. It is essential for the differentiation of memory CD8 T cells following the effector stage and for the maintenance of memory CD8 T cells partly by inducing bcl-2 expression.9 Recent data show that IL-7 can prevent the functional exhaustion of CD8 T cells.11
IL-15, on the other hand, is a critical cytokine for the activation, survival and differentiation of both memory and naive CD8 T cells, in addition to being essential for the development and function of natural killer (NK) and NK T cells.12–15 IL-15 also confers CD4 and CD8 T cells resistance to the influence of regulatory T cells.16 IL-15 receptor shares the β and the γc subunits with the IL-2 receptor, but the interaction of IL-15 with this heterodimer requires binding of IL-15 to IL-15Rα, a unique subunit that ligates IL-15 with high affinity.17 ,18 IL-15Rα is expressed on the membrane of cells including dendritic and epithelial cells.19 ,20 The formation within the cell of the complex IL-15/IL-15Rα, and its display at the cell membrane, is necessary for IL-15 to activate NK and CD8 T cells. These responder cells possess the heterodimeric receptor IL-2/15Rβγc, which engages IL-15 bound to IL-15Rα.20 This mechanism, called IL-15 transpresentation, is used by dendritic cells and also by epithelial cells, including hepatocytes, to stimulate the expansion and effector functions of naive and memory CD8 T cells.13 ,17 ,19
From the above data it seems likely that the production of both IL-7 and IL-15Rα by liver parenchymal cells may critically influence the homeostasis and activation of protective CD8 T-cell responses in the setting of viral infection. In the present work, we have analysed the levels of IL-7 and IL-15/IL-15Rα in the livers of patients with CHC and in cells harbouring HCV replicon as well as the mechanisms responsible for the disturbed expression of these molecules.
Material and methods
Analysis of messenger RNA expression was carried out in percutaneous liver biopsies from: (1) 34 patients with CHC (24 untreated patients, 10 non-responders to pegylated IFNα2 plus ribavirin); (2) 14 HCV patients with sustained virological response (SVR) after antiviral therapy; (3) 20 healthy livers (11 from living liver donors and nine from subjects studied because of non-specific minor elevation of serum transaminases; in all cases, histological examination of liver biopsies showed normal hepatic architecture); (4) 26 patients with miscellaneous liver disorders unrelated to HCV (nine chronic hepatitis B, 17 steatohepatitis). The diagnosis of CHC was based on elevation of serum transaminases for more than 6 months, positivity for anti-HCV antibodies, the presence of HCV-RNA by reverse transcription (RT) PCR and histological evidence of chronic hepatitis. Alcohol consumption and other causes of liver disease were excluded. Liver biopsies from an additional cohort of patients were used for immunohistochemical determinations of interferon regulatory factors (IRF) (11 patients with CHC, five HCV patients with SVR after antiviral therapy; five healthy living liver donors and six patients with miscellaneous liver disorders unrelated to HCV). The clinical and virological characteristics of the patients are indicated in supplementary table S1 (available online only). Samples and data from patients included in the study were provided by the Biobank of the University of Navarra and were processed following standard operating procedures approved by the ethical and scientific committees and conform to the ethical guidelines of the 1975 Declaration of Helsinki. Written consent was obtained from each patient included in the study.
Cell culture and treatments
Huh7 cells expressing full-length HCV replicon genotype 1b (Huh7 Core-3′) or genotype 2a and JFH1 cell culture system supporting complete replication and production of infectious virus particles were established as described previously.21–23 Cells were seeded and 24 h later, cells were left untreated or treated with 15 ng/mL of oncostatin M (R&D Systems, Minnesota, USA), 50 U/mL of IFNα2 (Sicor Biotech, Lithuania) or 15 ng/mL of oncostatin M plus 50 U/mL of IFNα2 during the different time points indicated in each experiment.
RNA extraction and real time RT–PCR
Total RNA extraction from liver tissue and Huh7 cell cultures was performed as previously described.21 Human IL-7, IL-15Rα, IL-15, IRF1, IRF2 and β-actin expression was measured by quantitative real time PCR using specific primers for each gene (see supplementary table S2, available online only) as described.24 The amount of each transcript was expressed by the formula: 2ct(β-actin)−ct(gene), ct being the point at which the fluorescence rises appreciably above the background fluorescence.
Immunohistochemical analysis of IRF2 was performed in paraffin liver sections using anti-IRF2 (Ab55331; Abcam, UK) antibody. Image analysis was performed using FIJI software.
Protein extracts from Huh7 cell cultures were obtained as described.21 Western blot was performed using the following antibodies: anti-IRF1 (sc-497; Santa Cruz Biotechnology, Germany), anti-actin (Sigma-Aldrich, Missouri, USA), anti-IRF2 (Ab55331), anti-rabbit IgG horseradish peroxidase-linked and anti-mouse IgG horseradish peroxidase-linked (Cell Signalling Technology, Massachusetts, USA).
Quantitation of IL-7 protein levels
IL-7 protein levels were quantified in supernatants of Huh7 or Huh7 Core-3′ replicon cells by ELISA (R&D Systems) following the manufacturer's instructions.
IRF inhibition was assessed in Huh7 cells by transfection with Silencer Select small interfering RNA (Ambion, USA) specific for IRF1 and IRF2. To this end, 5×104 Huh7 cells were seeded in 24-well plates and 24 h later, cells were transfected with 10 nM of siRNA and 1.5 μL of lipofectamine RNAiMax reagent (Invitrogen) in a final volume of 500 μL of Opti-MEM I reduced serum medium (Invitrogen). The effect of IRF overexpression was studied in Huh7 cells transiently transfected with plasmids that express IRF1 (pCMV6-XL5-IRF1; OriGene, Rockville, MD) or IRF2 (pCMV6-XL5-IRF2; OriGene). Briefly, 5×104 Huh7 cells were seeded in 24-well plates and 24 h later, cells were transfected with 0.5 μg of vector and 2 μL of lipofectamine 2000 reagent (Invitrogen) in a final volume of 500 μL of Opti-MEM I reduced serum medium. In some experiments, after 48 h post-transfection with Silencer Select siRNA Huh7 cells were treated with IFNα plus oncostatin M as described above for another 24 h.
Chromatin immunoprecipitation assay
Chromatin immunoprecipitation assay was performed in Huh7 or Huh7 Core-3′ cells essentially as described,25 using anti-IRF1 antibody (sc-497×) or a control rabbit immunoglobulin G (Santa Cruz Biotechnology). The promoter regions −384 to −164 of the human IL-7 gene and −518 to −334 and −237 to −67 of the human IL-15Rα gene,25 ,26 which contain the IRF1 binding sequences, were analysed from the immunoprecipitated DNA samples by quantitative real time PCR with specific primers (see supplementary table S3, available online only). The amount of immunoprecipitated DNA in each sample was indicated as a percentage of total input DNA. IL-15Rα promoter was studied for transcription factor binding sequences with the bioinformatic tool MotifScanner using the motif databases TRANSFACT and JASPAR.
CD8 T cell response
Huh7 cells were treated for 12 h with IFNα (50 U/mL) plus oncostatin M (15 ng/mL) or left untreated. Then, cells were washed twice to remove the cytokines and left for an additional 48 h and supernatants were harvested. On the other hand, 3×104 Huh7 cells stably transduced with the HLA-A2 gene (Huh7-A2, kindly provided by Dr R Thimme) were seeded in 48-well plates. After 18 h, cells were pulsed with 1 µg/mL of the HLA-A2-restricted FLU M1-derived peptide (GILGFVFTL) for 2 h at 37°C, washed three times and irradiated at 20.000 cGy in a gamma cell 3000 Elan apparatus. CD8 T cells (5×105) obtained from a HLA-A2+ healthy donor (provided by Navarre Blood and Tissue Bank, Navarrabiomed Biobank, Navarre Health Department) were isolated from peripheral blood mononuclear cells using a CD8 selection kit (Miltenyi Biotech, Madrid, Spain), and co-cultured with peptide-pulsed Huh7-A2 cells in the presence of 500 µL of supernatant from untreated or IFNα plus oncostatin M-treated Huh7 cells. In some cases, before the addition of the supernatant of IFNα plus oncostatin M-treated cells, it was preincubated for 2 h at 37°C with 0.1 µg/mL neutralising anti-human IL-7 antibody (R&D Systems) or IgG isotype control (R&D Systems). Human IL-2 (10 U/mL) was added to co-cultures every 2–3 days starting on day 2. On days 7–8, antigen-specific CD8 cells were counted by staining with PE-conjugated HLA-A2/FLU M1 Dextramer (Immudex, Copenhagen, Denmark) and FITC-conjugated anti-CD8 antibody (BD Biosciences, San Jose, California, USA) using a FACSCanto II flow cytometer (BD Biosciences).
Normality was assessed with the Shapiro–Wilk W test. Statistical analyses were performed using parametric (Student’s t test and two-way analysis of variance) and non-parametric (Kruskal–Wallis and Mann–Whitney U) tests. Correlation was assessed by Spearman’s or Pearson’s correlation coefficients. All p values were two-tailed and were considered significant if the associated value was less than 0.05. Descriptive data for continuous variables are reported as means±SD. SPSS V.15.0 for Windows was used for statistical analysis.
Expression of IL-7, IL-15Rα and IRF in the livers of patients with CHC
We observed a significant reduction in the hepatic levels of IL-7 mRNA in subjects with chronic HCV infection compared to either healthy controls, patients with SVR or patients with other forms of chronic inflammatory liver damage. Hepatic IL-15 mRNA expression was similar in all groups, but IL-15Rα mRNA values were significantly diminished in CHC patients compared to normal livers or patients with SVR (figure 1A–C). We observed a significant positive correlation (r=0.51, p<0.05) between IL-7 and IL-15Rα mRNA values in CHC patients (figure 1D). We did not found any relationship between IL-7 or IL-15Rα mRNA levels and biochemical or histological parameters of liver damage (see supplementary figure S1, available online only).
Both IRF1 and IRF2 have been reported to stimulate IL-7 in human epithelial intestinal cells.25 Also, our in-silico studies showed that IL-15Rα promoter has two putative sequences able to bind IRF. Therefore, we hypothesised that dysregulation of IRF could explain the low expression of both IL-7 and IL-15Rα present in HCV-infected livers. We found, however, that in CHC the percentage of IRF1-positive hepatocyte nuclei was not significantly different from that detected in normal livers or in other forms of liver diseases (see supplementary figure S2A,B, available online only). In contrast, we observed a marked reduction in the percentage of IRF2-positive hepatocyte nuclei in CHC with respect to healthy controls or patients with SVR (figure 2A,B). This was associated with an intense decrease of IRF2 mRNA in CHC compared to normal controls, patients with SVR and patients with other liver conditions (figure 2C).
IRF2 controls IL-7 and IL-15Rα basal expression in hepatic cells
We then investigated the influence of IRF1 and IRF2 on IL-7 and IL-15Rα expression in Huh7 cells by silencing each of the two transcription factors with two independent siRNA. We found that IRF1 silencing with either siIRF1-1 or siIRF1-2 (see supplementary figure S3A, available online only) impaired IL-7 but not IL-15Rα expression (figure 3A,B). In contrast, targeting IRF2 with either siIRF2-1 or siIRF2-2 (see supplementary figure S3B, available online only) induced a marked decrease of both IL-7 and IL-15Rα mRNA levels (figure 3C,D). These data indicate that IRF2 plays a critical role in the control of the basal expression of both IL-7 and IL-15Rα in hepatic cells.
HCV replication attenuates IRF2, IL-7 and IL-15Rα expression
We then analysed the effect of HCV replication on the expression of IRF, IL-7 and IL-15Rα. To examine this point, we carried out in-vitro studies using control Huh7 cells and two different clones containing the HCV replicon (Huh7 Core-3′ cells) corresponding to genotype 1b, which is the commonest genotype in our patients. In the two replicon cell clones, IRF1 expression remained unchanged (data not shown) but mRNA and protein levels of IRF2 decreased markedly (figure 4A). Consonant with the fall in IRF2 expression we observed an abrupt drop in the values of IL-7 and IL-15Rα mRNA in Huh7 Core-3′ (figure 4B,C). These data were further confirmed using a pool of Huh7 Core-3′ clones (see supplementary figure S4A–C, available online only). On the other hand, forced IRF2 overexpression in Huh7 Core-3′ cells induced a significant elevation of IL-7 and IL-15Rα transcripts (figure 4D,E and see supplementary figure S3C, available online only).
Interestingly, the effects on IRF2, IL-7 and IL-15Rα expression was not observed in genotype 2a replicon cells (transfected with plasmid pFGR-JFH1) nor in cells infected with JFH1 genotype 2a virus (data not shown), suggesting that the impact of HCV on these immunostimulatory molecules is genotype dependent.
Induction of IRF1 upregulates IL-7 and IL-15Rα: the effect of HCV replication
Upregulation of IL-7 and IL-15Rα is a desirable therapeutic goal. We observed that forced expression of IRF1 in Huh7 cells (see supplementary figure S3D, available online only) resulted in IL-7 and IL-15Rα upregulation without altering IRF2 expression (figure 5A–C and data not shown). Interestingly, IRF1 overexpression upregulates IL-7 and IL-15Rα even when IRF2 was silenced (see supplementary figure S3E (available online only) and figure 5D,E), indicating a direct role of IRF1 in the control of these two immunostimulatory molecules. IRF1, but not IRF2, is induced by IFNα and, more intensely, by the combination of IFNα plus oncostatin M (see supplementary figure S5, available online only).24 In experiments involving IRF1 silencing, we found that IFNα plus oncostatin M treatment upregulated IL-7 and IL-15Rα in an IRF1-dependent manner (figure 5F,G). Therefore, we tested whether IFNα or IFNα plus oncostatin M could influence IRF1, IL-7 and IL-15Rα in HCV replicon cells. We observed that while basal IRF1 expression was similar in Huh7 Core-3′ and control cells, its induction with IFNα or IFNα plus oncostatin M was significantly reduced in the replicon cells both at the transcriptional and protein level, the values being higher with the combined treatment (figure 6A). In Huh7 Core-3′ cells we found that, in parallel to IRF1, IL-7 and IL-15Rα were induced by IFNα and more intensely by IFNα plus oncostatin M, albeit to values lower than in Huh7 cells (figure 6B,C and see supplementary figure S4D,E, available online only). In agreement with these data, chromatin immunoprecipitation assays showed that IRF1 was recruited to the promoter regions of IL-7 and IL-15Rα on stimulation with IFNα plus oncostatin M, although this effect was attenuated in cells sustaining HCV replication (figure 6D,E).
IL-7 produced by IFNα plus oncostatin M-treated Huh7 cells expands CD8 T cells
Finally, we asked whether the levels of IL-7 induced by the combined action of IFNα plus oncostatin M could be sufficient to stimulate the expansion of CD8 T cells recognising an antigen at the hepatocellular membrane. To this end, Huh7-A2 were pulsed with a flu peptide containing a CD8 epitope recognised by HLA-A2+ individuals, and the cells were then incubated with isolated CD8 T cells from an HLA-A2+ donor in the presence of medium from Huh7 cells treated or not with IFNα plus oncostatin M. At this time point, all CD8+FLU+ cells were CD127high, allowing these cells to respond to IL-7. We found that CD8 T cells proliferated more intensely in the presence of supernatant from IFNα plus oncostatin M-treated Huh7 cells, and that this effect was attenuated in the presence of neutralising anti-IL-7 antibody (figure 7A,B). These findings indicate that the release of IL-7 by hepatocytes treated with IFNα plus oncostatin M favours the expansion of antigen-specific CD8 T cells.
Hepatocytes express both IL-7 and IL-15Rα.9 ,10 ,17 In the liver, parenchymal cells are the main producers of IL-7, a cytokine that is not synthesised by T, B or NK lymphocytes.9 ,10 In contrast, IL-15Rα is also expressed by a variety of immune cells.17 In this work, we show that CHC livers exhibit a significant reduction of both IL-7 and IL-15Rα mRNA levels compared to normal hepatic tissue, probably reflecting the reduced expression of these molecules in HCV-infected hepatocytes. This view is supported by data in Huh7 Core-3′ cells showing that HCV genotype 1b has the ability to block the expression of these molecules. This effect may contribute to the chronicity of infection, because by interfering with the expression of IL-7 and IL-15Rα HCV can dampen the induction of antiviral T-cell immunity through various mechanisms. On one hand, impaired IL-7 production would reduce the capacity of HCV-infected hepatocytes to maintain the pool of functionally active anti-HCV memory T cells.9 ,11 On the other hand, a defective display of IL-15Rα at the hepatocellular membrane would hamper the aptitude of hepatocytes to transpresent the complex IL-15/IL-15Rα to neighbouring NK and T cells expressing the intermediate-affinity heterodimeric IL-2/15Rβγc receptor.17 ,18 The ensuing defect in IL-15 signalling would compromise the activation and expansion of anti-HCV cytotoxic T lymphocytes and the stimulation of NK activity.
As the promoters of both IL-7 and IL-15Rα possess IRF-binding sites, we analysed the expression of IRF1 and IRF2 in CHC to investigate whether altered function of these transcriptional regulators could account for the reduced expression of immunostimulatory molecules. It is now becoming clear that IRF exert wide effects on host defence mechanisms beyond their function in the IFN system.27 IRF1 is upregulated by both type I and type II interferons and mediates the expression of diverse antiviral and immunostimulatory genes.28 ,29 IRF2 was originally described as an IRF1 antagonist, but accumulating evidence indicates that it is a functional agonist of IRF1 for diverse IFN-stimulated regulatory element-responsive genes.30 Notably, it has been reported that IRF2 is essential to drive T helper type 1 responses and NK development in vivo.30 ,31 In consonance with this concept, our present data have revealed that IRF2 controls basal IL-7 and IL-15Rα expression in liver cells.
Interestingly, in CHC we observed a marked reduction of hepatic IRF2 mRNA levels and a striking drop in the percentage of hepatocytes with positive nuclear IRF2 staining. These in-vivo alterations are in close agreement with findings in replicon cells showing reduced basal IRF2 expression. Importantly, the downregulation of IL-7 and IL-15Rα observed in Huh7 Core-3′ cells can be reverted by forcing IRF2 expression. It thus seems likely that HCV would impair IL-7 and IL-15Rα synthesis by inhibiting IRF2 expression. It should be noted that the great majority of the patients who entered this study were genotype 1b. Importantly, repression of the immunostimulatory molecules was observed in Huh7 cells with genotype 1b replicon but not in cells sustaining genotype 2a replication (JFH1 strain), suggesting that this effect is genotype dependent. It is possible that JFH1 strain may yield atypical results in the field of IRF functions. Moreover, genotypes 1 and 2a also differ in their response to IFNα, the former being more resistant to this therapy than the latter.32 It is tempting to speculate that there may be a relationship between the ability of the different genotypes to downregulate immunostimulatory molecules and their sensitivity to IFNα treatment. This is a point deserving further studies.
Therapies aimed at increasing IL-7 and IL-15Rα expression in HCV-infected livers should be sought as they would be likely to result in an enhancement of antiviral T-cell immunity. Our data showed that, in liver cells, IL-7 and IL-15Rα can be upregulated by overexpressing IRF1, which we found is capable of binding to the promoter of both molecules. IRF1 is induced by IFNα but more efficiently by the combination of IFNα plus oncostatin M,24 a treatment that promotes the upregulation of IL-7 and IL-15Rα to levels higher than those attained with either cytokine alone. Interestingly, we found that treatment with IFNα plus oncostatin M induced Huh7 cells to produce amounts of IL-7 capable of promoting the expansion of antigen-specific CD8 T cells. In the setting of HCV infection, we observed that cytokine-mediated IRF1 induction and its recruitment to the promoters of IL-7 and IL-15Rα were inhibited. This observation was not unexpected as IRF1 is a STAT1-responsive gene, and it has been reported that HCV can block IFNα-mediated STAT1 activation.33 ,34 Despite the interference of HCV in IRF1 induction, we noticed that the combination of IFNα plus oncostatin M can still achieve a substantial increase in the levels of IRF1, IL-7 and IL-15Rα in the infected cells, thus representing a potential therapeutic approach for patients with CHC. Strategies aimed at enhancing the antiviral effect of IFNα are still a desired goal as the newly introduced IFNα-based tritherapeutic approaches show limited efficacy in null responders to previous IFNα therapy.35
To summarise, our data indicate that HCV genotype 1 downregulates IRF2 in hepatocytes attenuating the expression of IL-7 and IL-15Rα in the infected cells. These two molecules can be upregulated by IFNα but more intensely by the combination of IFNα plus oncostatin M, an effect that is mediated by IRF1, which is strongly upregulated by the combination of the two cytokines. In replicon cells the cytokine-mediated induction of IRF1, IL-7 and IL-15Rα is dampened, but still IFNα plus oncostatin M upregulates these molecules at higher levels than either cytokine in isolation. As IL-7 and IL-15Rα are critical for the activation and expansion of anti-HCV cytotoxic T cells, our present findings reveal a new mechanism by which HCV may escape immunity. Our data also point to a new strategy to activate anti-HCV immune response.
The authors would like to thank B Carte and S Jusue for their excellent technical assistance.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online figure 1
- Data supplement 2 - Online figure 2
- Data supplement 3 - Online figure 3
- Data supplement 4 - Online figure 4
- Data supplement 5 - Online figure 5
- Data supplement 6 - Online legends
- Data supplement 7 - Online table 1
- Data supplement 8 - Online table 2
- Data supplement 9 - Online table 3
Contributors Study concept and design: EL, JP, PS and MPC; acquisition of data: EL, JIR, RA, IE, AB, LG and PG; analysis and interpretation of data: EL, JP, PS, MPC, RA and JIR; drafting of the manuscript: EL, JP and PS, technical support: LG, IE and AB.
Funding This work was funded by grants from Fondo Investigación Sanitaria (PI10/00149 to EL), Ministerio de Ciencia e Innovación (SAF2010-15074 to PS), Fundación Mútua madrileña, ‘UTE Project CIMA’, Instituto de Salud Carlos III, Fundacion Pedro Barrié de la Maza and Condesa de Fenosa.
Competing interests None.
Ethics approval This study received ethics approval from the ethics committee of the University Clinic of Navarra.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.