Control Susceptibility to Viral Myocarditis H28 , Or Fpgt , Tnni3k

Quantitative Trait Locus Analysis, Pathway Analysis, and Consomic Mapping Show Genetic Variants of Tnni3k, Fpgt, or H28 Control Susceptibility to Viral Myocarditis This information is current as of September 25, 2021. Sean A. Wiltshire, Gabriel André Leiva-Torres and Silvia M. Vidal J Immunol published online 27 April 2011 http://www.jimmunol.org/content/early/2011/04/27/jimmun ol.1100159 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2011/04/27/jimmunol.110015

Material 9.DC1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 25, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2011 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published April 27, 2011, doi:10.4049/jimmunol.1100159 The Journal of Immunology

Quantitative Trait Locus Analysis, Pathway Analysis, and Consomic Mapping Show Genetic Variants of Tnni3k, Fpgt, or H28 Control Susceptibility to Viral Myocarditis

Sean A. Wiltshire, Gabriel Andre´ Leiva-Torres, and Silvia M. Vidal

Coxsackievirus B3 (CVB3) infection is the most common cause of viral myocarditis. The pathogenesis of viral myocarditis is strongly controlled by host genetic factors. Although certain indispensable components of immunity have been identified, the genes and pathways underlying natural variation between individuals remain unclear. Previously, we isolated the viral myocarditis susceptibility 1 (Vms1) locus on chromosome 3, which influences pathogenesis. We hypothesized that confirmation and further study of Vms1 controlling CVB3-mediated pathology, combined with pathway analysis and consomic mapping approaches, would elucidate both pathological and protective mechanisms accounting for natural variation in response to CVB3 infection. Vms1 was originally Downloaded from mapped to chromosome 3 using a segregating cross between susceptible A/J and resistant B10.A mice. To validate Vms1, C57BL/ 6J-Chr 3A/NaJ (a chromosome substitution strain that carries a diploid A/J chromosome 3) were used to replicate susceptibility compared with resistant C57BL/6J (B6). A second segregating F2 cross was generated between these, confirming both the localization and effects of Vms1. Microarray analysis of the four strains (A/J, B10.A, C57BL/6J, and C57BL/6J-Chr 3A/NaJ) illuminated a core program of response to CVB3 in all strains that is comprised mainly of IFN-stimulated genes. Microarray analysis also revealed strain-specific differential expression programs and genes that may be prognostic or diagnostic of suscep- http://www.jimmunol.org/ tibility to CVB3 infection. A combination of analyses revealed very strong evidence for the existence and location of Vms1. Differentially expressed pathways were identified by microarray, and candidate gene analysis revealed Fpgt, H28, and Tnni3k as likely candidates for Vms1. The Journal of Immunology, 2011, 186: 000–000.

iral myocarditis and its long-term sequela, dilated car- mentally in the absence of virus by administration of cardiac diomyopathy (DC), are a common cause of morbidity myosin with adjuvant. As with viral myocarditis, inbred strains of V and mortality (1). Adenoviruses, influenza, and herpes- mice also differ in their inherent susceptibility to experimental viruses have been implicated; however, enteroviruses, particularly autoimmune myocarditis, implicating genetic controls of myo- coxsackievirus B3 (CVB3) strains, are the most prominent etiol- cardial inflammation (5). Therefore, two outstanding questions in by guest on September 25, 2021 ogy (2). In mice, experimental infection with CVB3 has served the field are: 1) does uncontrolled viral replication drive inflam- to elucidate three distinct clinical aspects of human disease pro- mation during myocarditis, or is virus merely an equivalent trig- gression: 1) an initial phase of viral replication and cytolytic ger to unequal inflammation; and 2) which host genetic variants damage of cardiomyocytes; 2) a phase of intense myocardial in- are associated with the outcome of virus-induced disease? filtration by immune effectors, which may resolve infection or Phenotype characterization of CVB3 infection in mutant mouse persist indefinitely in susceptible animals (3); and 3) a delayed stocks has so far delimited three major systems essential in pre- phase in which initial infiltration and response may transition into dicting disease severity: integrity of myocardial structure, pro- autoimmune or inflammatory cardiomyopathy (4). Furthermore, tection by innate immunity, and the balance between curative different individuals or inbred mouse strains display variable immunity and chronic inflammation or autoimmunity. The cyto- degrees of cardiomyopathy, likely influenced by their genetic skeletal protein dystrophin is cleaved by CVB3 protease 2A, background. Myocarditis and DC can also be induced experi- a protein essential for viral life cycle; this cleavage affects cardiac function and morphology (6). Type I IFN is produced by immune and nonimmune cells in response to pathogen sensing through Department of Human Genetics, McGill University, Montreal, Quebec H3A 1B1, innate receptors. High levels of type I IFN activate antiviral re- Canada sponse systems (7) and enhance early cytotoxic effectors such as Received for publication January 20, 2011. Accepted for publication March 23, 2011. NK cells (8). Thus, mice deficient in TLR3 suffer from severe This work was supported by the Canadian Institutes of Health Research. S.A.W. is myocarditis and a high viral load in the heart and liver (9). Sim- supported by a doctoral award from the Fonds de la Recherche en Sante´ du Que´bec. S.M.V. is a Canada Research Chair. ilarly, type I IFNR (10) and IFN-b knockout mice (11) show in- Address correspondence and reprint requests to Dr. Silvia M. Vidal, McGill Univer- creased viral burden in the liver and heart, respectively. These data sity, McGill Life Sciences Complex, Room 367, 3649 Promenade Sir William Osler, strongly suggest a protective role for type I IFN, particularly IFN- Montreal, QC H3G 0B1, Canada. E-mail address: [email protected] b, during coxsackieviral infection. Recently, the constitutive ex- The online version of this article contains supplemental material. pression of CXCL10 in the heart has been shown to attract NK Abbreviations used in this article: cnSNP, coding nonsynonymous single nucleotide cells to the heart, reducing overall viral replication (12). Though polymorphism; CSS3, chromosome substitution strain that carries a diploid A/J chromosome 3; CVB3, coxsackievirus B3; DC, dilated cardiomyopathy; FC, fold much progress has been made, there is still a need to account for change; ISG, IFN-stimulated genes; LOD, logarithm of odds; PI, postinfection; differential susceptibility to viral myocarditis among otherwise qPCR, quantitative PCR; QTL, quantitative trait locus; SNP, single nucleotide poly- normal individuals due to natural variation. morphism; Vms1, viral myocarditis susceptibility 1. The MHC genes have been linked to susceptibility to myocar- Copyright Ó 2011 by The American Association of Immunologists, Inc. 0022-1767/11/$16.00 ditis in both humans and experimental mouse models (13, 14).

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1100159 2 DISTAL CHROMOSOME 3 CONTROLS VIRAL MYOCARDITIS IN MICE

Non-MHC genes may also have an important impact on disease 3.1 3 105 cells per well. Homogenized organs were serially diluted 10-fold susceptibility. Recently, functional TLR3 variants (homozygous in nonsupplemented DMEM, and the dilutions were used to infect HeLa 412F and heterozygous P554S) were found in excess among pa- cells in triplicate for 60 min. After initial infection by homogenate, HeLa cells are covered in a layer of media with 0.25% agarose and incubated for tients presenting enteroviral myocarditis or DC (15). To identify 3 d. Cells were fixed in formaldehyde and stained with crystal violet to novel genetic factors associated with disease, we have applied count plaques. quantitative trait locus (QTL) mapping in an experimental mouse Genotype analysis model. Previously, using an F2 cross between susceptible A/J and a resistant B10.A-H2 (B10.A) mice, we identified three QTL on Genomic DNA was prepared from tail biopsies of individual F2 mice by chromosomes 1 (viral myocarditis susceptibility 2 [Vms2]), 3 overnight incubation at 55˚C in 700 ml buffer (100 mM Tris-HCl [pH 8], (Vms1), and 4 (Vms3). Only Vms1 was linked to the three phe- 0.0.5 mM EDTA, 200 mM NaCl, and 0.2% SDS) containing 0.5 mg/ml proteinase K and followed by RNAse treatment (0.3 mg/ml; 2 h at 37˚C). notypes under study [i.e., myocarditis score (p = 0.005), extent DNA was purified by serial phenol-chloroform extractions and ethanol of sarcolemmal damage (p = 0.001) (16), and heart viral load precipitation. In this genome scan, nine MIT markers spaced ∼15 cM apart (p = 0.010)]. This suggested that Vms1 is a central determinant of from each other, with increased density toward distal chromosome 3 to disease severity; in the current study, we seek to determine the distinguish between A/J and B6 alleles (D3Mit164, D3Mit224, D3Mit346, precise contribution of Vms1 in viral myocarditis. To achieve this, D3Mit291, D3Mit147, D3Mit128, D3Mit116, D3Mit19, and D3Mcg1). Markers in the peak of Vms1 were chosen to match those previously used we validated the existence of Vms1 through phenotyping of a to facilitate direct comparison between the two datasets. Reactions per- chromosome substitution strain that carries a diploid A/J chro- formed manually used a 10 ml total reaction volume, 200 mM 2’-deoxy- mosome 3 (CSS3) on a C57BL/6 (B6) background (17). We then nucleoside 5’-triphosphate, 1.5 mM MgCl2, 2 pmol of each primer, and 0.5 used a second F cross, this time between B6 and CSS3 mice, to U Taq polymerase (Invitrogen, Burlington, ON, Canada). Reactions were Downloaded from 2 performed as follows: 96˚C for 2 min; 30 cycles of 94˚C for 45 s, 56˚C for refine the Vms1 genetic interval. Investigation of the mechanistic 45 s, 72˚C for 60 s; and a final extension step at 72˚C for 7 min. PCR basis of Vms1 function using genomewide expression and pathway products were then separated on 2 to 3% high-resolution agarose (USB, analysis in the four parental strains (susceptible A/J and CSS3 Cleveland, OH) gels containing ethidium bromide and visualized under mice, as well as resistant B10.A and B6 mice) indicated that Vms1 UV light. Identification of likely genotyping errors was performed using R/ . susceptibility alleles impair antiviral mechanisms downstream qtl; all genotypes with error logarithm of odds (LOD) scores 3 were repeated and verified (19). of viral sensing or type I IFN pathways. This combined effort http://www.jimmunol.org/ demonstrated that viral load and myocarditis severity are likely Statistical and bioinformatics analysis controlled by a more resolved locus, Vms1.1, for which the critical Statistical analyses were conducted with the freely available program R, and interval contains three candidate genes: Fpgt, H28, and Tnni3k. linkage was performed with package “R/qtl.” The scanone function of the “R/qtl” library was used to perform maximum likelihood interval mapping Materials and Methods of phenotype on genetic markers. Significance values were evaluated with 10,000 permutations. In joint analysis between [A/J 3 B10.A]F and Mice and virus 2 [CSS3 3 B6]F2, combined cross-analysis was performed with experi- Inbred A/J (000469), B10.A-H2a H2-T18a/SgSnJ (B10.A, 000646), mental cross used as a covariate term in all calculations as previously C57BL/6J-Chr 3A/J/NaJ (CSS3, 004381), and C57BL/6J (B6, 000664) described (20). LOD support interval was calculated using a 1.5 LOD drop by guest on September 25, 2021 mice were purchased from The Jackson Laboratory. The mice were by “R/qtl.” Single nucleotide polymorphisms (SNPs) were identified using maintained in the McGill University animal facility in compliance with the the mouse phenome database (http://phenome.jax.org/SNP), were also Canada Council on Animal Care as approved by the McGill University provided by the mouse genome project group at the Wellcome Trust Animal Care Committee. Coxsackievirus B3 used for time-course and Sanger Institute, and can be obtained at http://www.sanger.ac.uk/resources/ microarray experiments was the CVB3-CG strain, as used previously (16). mouse/genomes/. SNPs were analyzed using the sorts intolerant from tolerant algorithm (21) for potentially deleterious effects. IFN-stimulated All other experiments including CSS3/B6/F1 as well as [CSS3 3 B6]F2 were infected with the plaque purified CVB3-H3 (18), generously provided genes (ISG) were annotated using the Interferon Stimulated Gene and as a plasmid by the laboratory of Dr. Kirk Knowlton (University of Cal- Interferome databases (22, 23), the Database for Annotation, Visualization ifornia, San Diego, San Diego, CA). Like CVB3-CG, CVB3-H3 was also and Integrated Discovery (24), and by Ingenuity Pathways Analysis (In- derived from the highly myocarditic CVB3 Woodruff lineage (18). In this genuity Systems, http://ingenuity.com). Pathways identified through In- model, CVB3-H3 behaves comparably to CVB3-CG and has the added genuity are characterized with a Network Score, which is defined as the benefit of being a known and widely used genotype allowing for more negative exponent of the right-tailed Fisher’s exact test. The p values , , direct comparisons by other researchers. shown in figures are a result of a two-tailed t test, with *p 0.05, **p 0.01, and ***p , 0.001. Mouse infection and phenotype determination Microarray and quantitative PCR Mice were inoculated i.p. with 400 PFU CVB3-CG/g of body weight (or 10 PFU/g CVB3-H3) diluted in sterile PBS at 7 to 8 wk of age. Experiments Hearts used for microarrays were first perfused with cold sterile buffer (110 shown in Fig. 1 were replicated with both viruses with equivalent results. mM NaCl, 10 mM NaHCO3, 16 mM KCl, 16 mM MgCl2, 1.2 mM CaCl2, Uninfected controls in microarray experiments were mock infected with and 5 U/l heparin [pH 7.9]), and tissues were preserved in RNA-later 2 PBS in parallel. Survival, weight, and signs of infection were evaluated on stabilization reagent (Ambion) before being frozen at 80˚C. Total ∼ a daily basis, and the mice were humanely sacrificed 1, 2, 4, or 8 d post- RNA was extracted from 30 mg heart tissue from a transverse section infection (PI). Moribund mice were humanely euthanized at an acceptable using the Qiagen RNeasy extraction kit (Qiagen). Quality control by clinical endpoint. At necropsy, mouse hearts were removed aseptically, cut Agilent bioanalyzer (Agilent Technologies), cDNA synthesis, and micro- in half sagitally, and fixed in buffered formalin; the remainder was stored array hybridization was performed at the McGill University and Genome at 280˚C. The apex of hearts were subsequently homogenized and sub- Quebec Innovation Centre. Three infected and three uninfected female mitted to three freeze-thaw cycles to release infectious virus. After cen- mice of each genotype were independently prepared and analyzed on an trifugation at 1000 3 g for 5 min to remove debris, the supernatant Illumina Mouse-WG6 version 2.0 array (Illumina) (one genotype and six was isolated for quantification by plaque assay. Samples to be used for mice per array, four arrays total). A/J versus B10.A and CSS3 versus B6 microarray were taken from a transverse section above the apex of experiments were performed and analyzed separately. Raw data are avail- ∼30 mg. able through the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE19496. Normalization Plaque assay analysis was performed using FlexArray (25), and expression values were normalized using the log2 and RMA configuration of Lumi. Principal HeLa cells (American Type Culture Collection: CCL-2) were grown and component analysis figures (Supplemental Fig. 3) were also generated maintained in DMEM supplemented with 10% FBS and 100 mg/ml within FlexArray. Genes were filtered for consideration as follows: the penicillin/streptomycin (13 media). One day prior to infection, the cells 40th percentile was determined across all normalized probes as an arbitrary were harvested at 70–80% confluence and plated on 12-well plates at cutoff value. If no probes from any experimental group were greater than The Journal of Immunology 3 the 40th percentile, the gene was considered absent in all samples and not considered further. A starting list was used in which at least one genotype either infected or uninfected met this criterion for 20,261 targets. Paired experimental groups (e.g., A/J infected versus uninfected) were compared by CyberT (26) with default parameters and considered to be differentially expressed if the fold change (FC) was ,0.5 or .2 and unadjusted p value was ,0.05. Interaction between strain and genotype was defined as: at least one strain is differentially expressed (0.5 $ FC $ 2, p , 0.05), ANOVA (strain/infection) significant (p , 0.05); as well as the absolute difference of difference between strain means: [|((mAJ 2 mAJ infected) 2 (mB10.A 2 mB10.A infected))| . 1.25]. This final step was determined as an effective filter in removing parallel changes between genotypes following infection. This much-abbreviated list was then annotated to identify which strain was significantly different (e.g., AJ responds to infection by increasing expression of a gene, and B10.A does not; this is an AJ-specific gene). See Supplemental Fig. 1 for example and distribution of this pa- rameter. Highly differentially expressed genes were identified with an arbitrary cutoff of 0.2 $ FC $ 5(p , 0.05). Quantitative PCR (qPCR) primers were designed to span exon junctions with the help of primer3plus (27). List of primers used

CVB3-H3 forward: 59-ATGGCAGAAAACCTCACCAG-39, CVB3-H3 re- Downloaded from verse: 59-ATGTTCTGCCCAAACAGTCC-39; Gbp1 forward: 59-ACTG-

AGAAGATGGAGCAGGAAC-39, Gbp1 reverse: 59-GTTGCAAGCTCT- A CATTCTGG-39; H28 forward: 59-ACAGGACTGGATTTGCCTTG-39, FIGURE 1. Replication of Vms1 phenotypes in C57BL/6j-Chr3 /NaJ. A, 3 H28 reverse: 59-GCATGCTTTCAACCAGAGAC-39; Ifnb1 forward: 59- Viral titer in CSS3, B6, and [CSS3 B6]F1 mice 8 d postinfection, as TGACGGAGAAGATGCAGAAG-39, Ifnb1 reverse: 59-ACCCAGTGCT- measured by plaque assay. CSS3 mice of both sexes have signiﬁcantly GGAGAAATTG-39; Tnni3k (as described in Ref. 28) forward: 59-AGA- more virus in the heart than B6 or [CSS3 3 B6]F1 (p , 0.001). B,

TTTCTGCAGTCCCTGGAT-39, Tnni3k reverse: 59-AAGACATCAGCCT- Myocarditis is scored with 4 being most severe and 0 being no inflam- http://www.jimmunol.org/ TGATGGTG-39; Eif4e forward: 59-TACCACTAATCCCCCACCTG-39, mation evident. CSS3 mice were more susceptible than B6 or [CSS3 3 Eif4e reverse: 59-GGTTTGCTTGCCAAGTTTTG-39; Lrrc40 forward: 59- B6]F (p , 0.05). n $ 11 per group. Both A and B represent a composite CCACAAAAACGACCCTCATC-39, Lrrc40 reverse: 59-CCAACACCAT- 1 of multiple CVB3-H3 infections ([CSS3 3 B6]F controls) and were TTCCTTCAGC-39; Gapdh forward: 59-AAGGGCTCATGACCACAGTC- 2 , , C-39, Gapdh reverse: 59-GATGCAGGGATGTTCTGG-39; Adh1 forward: replicated with CVB3-CG. *p 0.05, ***p 0.001. 59-TGTGGCTGACTTCATGGCTA-39, Adh1 reverse: 59-TGGTGCTTG- AATCAGTCTGG-39; and D3Mcg1 forward: 59-GGAGTTCTGAGGAG- noted between viral load and myocarditis severity (Spearman, r = GTCGAG-39, D3Mcg1 reverse: 59-CATGGCGTAACGAAACAAAA-39. 0.54; p , 0.0001). Unlike the previous [A/J 3 B10.A]F2, sex dif- Cytokine measurement ferences in susceptibility within the F2 mice were not evident in by guest on September 25, 2021 IFN-b concentration was measured using an ELISA kit from PBL Inter- either myocarditis or viral replication (p = 0.470, p = 0.305, re- feronSource. spectively). Results Chromosome 3 controls viral myocarditis severity In mice, severity of viral myocarditis is controlled by genetic background (29). Strong linkage of chromosome 3 to myocarditis, sarcolemmal disruption (16), and viral titer (Supplemental Fig. 2) encouraged us to isolate the role of chromosome 3 in viral myocarditis. The CSS3 mice are ideally suited to assess the contribution of chromosome 3, as CSS3 mice carry the susceptible A/J allele of Vms1 (Vms1s) while controlling for background compared with B6. Differential control of viral replication was observed at day 8, with CSS3 mice being significantly more susceptible than B6 mice (2.61 6 0.13 versus 4.04 6 0.08 log [PFU/mg]; p , 0.001) (Fig. 1A). Furthermore, CSS3 mice have significantly higher levels of inflammation in the heart than B6 mice (2.85 6 0.26 versus 1.4 6 0.23; p , 0.001) (Fig. 1B). No significant differences were observed between the sexes, and [CSS3 3 B6]F1 mice were not significantly different from B6 mice in terms of viral replication or inflammation. These results replicate Vms1s susceptibility to viral replication and intensity of peak inflammation; they also suggest Vms1s is recessive to Vms1r.

Genomewide linkage replicates Vms1 FIGURE 2. Phenotype distributions within the [CSS3 3 B6]F2 pop- ulation. A, Histogram of the distribution of viral titers in [CSS3 3 B6]F To characterize the inheritance of susceptibility to CVB3, we de- 2 mice, with a mean of 3.22 6 0.09. B, Histogram of the distribution of termined heart viral titer and myocarditis in 126 [CSS3 3 B6]F2 myocarditis scores in [CSS3 3 B6]F2 mice, with a mean of 1.73 6 0.11. mice. The mean of F2 viral titers (3.22 6 0.09) and myocarditis C, The distribution of [CSS3 3 B6]F2 mice, showing signiﬁcant correla- scores (1.73 6 0.11) were similar to that of the B6 parental strain, tion (r2 = 0.29; p , 0.0001) between myocarditis score and viral titer. This s which is consistent with a major gene effect and recessive Vms1 reveals a trend in which higher myocarditis score is correlated to higher susceptibility (Fig. 2A,2B). As shown in Fig. 2C, correlation was viral titer. 4 DISTAL CHROMOSOME 3 CONTROLS VIRAL MYOCARDITIS IN MICE

Linkage analysis revealed a locus on distal chromosome 3, exerts some effect before the peak of myocarditis at day 8. As a overlapping the previously described Vms1 locus, controlling heart prerequisite to transcriptional profiling by cDNA microarray, we viral titer (LOD = 3.74; p = 0.018) (Fig. 3A). Vms1r mice were first ascertained at which time point expressed genes might play protected from high viral load in the heart (2.87 6 0.29 log [PFU/ a role. We assessed viral titer at multiple time points in A/J and mg]) when compared with Vms1s (3.75 6 0.13 log [PFU/mg]) B10.A mice (Fig. 4A), as well as in CSS3 and B6 mice (Fig. 4B), (Fig. 3B). Vms1 linkage to inflammation was also significant prior to the development of inflammation. After 96 h PI, A/J and (LOD = 3.57; p = 0.0032); Vms1r (1.07 6 0.23) mice were pro- CSS3 mice have significantly higher levels of viral replication tected compared with Vms1s mice (2.28 6 0.24). In a single QTL than B10.A or B6 mice. Higher type I IFN levels have been model, Vms1 accounted for 11.89% of the variance observed in reported in the B6 strain (30), suggesting a possible mechanism viral titer and for 11.02% in myocarditis score, a notable increase for resistance. We observed that susceptible A/J mice produce from, respectively, 5.87 and 3.25% in the [A/J 3 B10.A]F2 cross. significantly more IFN-b than resistant B10.A mice during in- The Vms1 interval in both crosses is defined between D3Mit291 fection (Fig. 4C). CSS3 and B6 mice did not produce significantly and D3Mcg1, spanning 16 cM and 16 Mb. different amounts of IFN-b at the times measured. These observations are consistent with a genetically determined lack of ability Differential control of viral replication occurs between 2 and to control viral replication by 96 h PI, which is not accounted for 4 d postinfection by higher IFN-b in resistant mice. Differential gene expression in response to infection may be causally related to the observed phenotypes or may be a reaction Microarray analysis of resistant and susceptible hearts reveals a core program of response to viral myocarditis secondary to the primary causal locus. Understanding differential Downloaded from gene expression globally and within Vms1 will aid in identifying Global gene expression 96 h PI was compared by microarray in A/J, potential candidates for Vms1. Presumably, the gene mediating B10.A, CSS3, and B6 hearts. Although it has not yet been dem- Vms1r resistance or Vms1s susceptibility to viral myocarditis onstrated that Vms1 is a heart-specific effect, the lack of difference in viral control at 48 h PI suggests an equal seeding of the myocardium followed by an unequal response to infection toward 96 h

PI (Fig. 4B). Therefore, the heart seemed the most likely tissue to http://www.jimmunol.org/ interrogate to identify mechanisms or cells identiﬁable by gene expression. Principal component analysis indicates that infection accounted for the majority of variation, as expected (Supplemental Fig. 3), followed by variation due to genotype in A/J versus B10.A and CSS3 versus B6 mice. Genes highly expressed or repressed PI (at least a 5-fold change in all strains) were considered to be part of a common pathway. by guest on September 25, 2021

FIGURE 3. Linkage of chromosome 3 to CVB3 replication in the heart and to myocarditis. Overlapping LOD plots of [A/J 3 B10.A]F2 (black) and [CSS3 3 B6]F2 (gray) mice shows significant linkage on distal chromosome 3 to viral replication in the heart (A) and to myocarditis (C). Threshold values corresponding to a = 0.05 are represented as horizontal lines of corresponding shade. Peak linkage to viral replication and myocarditis in [A/J 3 B10.A]F2 occurred at D3Mit19 (16). Peak linkage to FIGURE 4. Viral replication is higher in A/J and CSS3 than in B10.A viral replication and myocarditis in [CSS3 3 B6]F2 occurs at D3Mit19 and B6 96 h PI. Level of infectious CVB3 within the heart measured by (LOD = 3.74; p = 0.018) and D3Mit116 (LOD = 3.57; p = 0.0032). B, Viral plaque assay at 24, 48, and 96 h in A and B. Differential control is sig- titer versus genotype at D3Mit19 shows that the AA genotype leads to nificant at 96 h postinfection in both comparisons (p , 0.05). C and D, a higher viral titer than BB in both crosses: 3.5 6 0.11 versus 2.84 6 0.12 Concentrations of IFN-b in the serum were measured by ELISA in the log [PFU/mg] in [A/J 3 B10.A]F2 and 3.75 6 0.13 versus 2.87 6 0.29 log same animals as in A and B. Significantly higher amounts of IFN-b were [PFU/mg] in [CSS3 3 B6]F2. D, D3Mit19 genotype of AA leads to higher detected in A/J mice compared with B10.A at 48 and 96 h PI (p , 0.05). myocarditis scores than BB in both crosses: 2.83 6 0.11 versus 2.06 6 No significant differences were observed between CSS3 and B6 mice (A

0.12 in [A/J 3 B10.A]F2 and 2.28 6 0.24 versus 1.07 6 0.23 in [CSS3 3 and C, n = 4 per time point; B and D, n = 6 per time point, replicated with B6]F2. n = 8 per strain at 48 h). The Journal of Immunology 5

This set of 129 genes was strongly indicative of an IFN response: for this strain (Fig. 6B). The CSS3-specific pathway (Fig. 6C) 46 of the 129 (35%) genes were annotated as being IFN induced shared several component genes with A/J, most of which are not (23). The interferome database used for comparison contains 1925 located on chromosome 3. CSS3-specific pathways were identified mouse genes annotated as being IFN induced in various cell types, to be tissue and organ development related. Notably, cell-mediated corresponding to 6.3% of the arrays used in this study. Of the 46 immune response pathways not present in A/J were expressed in genes identified, 14 suggested a type I IFN response, 1 a type II B6 and CSS3 hearts. No B6-specific pathways were identified, IFN response, and the remaining 31 were inducible by either. and only five genes are particular to a B6 response (Chad, Dkk3, Functional clustering of these by Database for Annotation, Visu- Mybphl, Prss35, and Sln); of these, only Mybphl is expressed on alization and Integrated Discovery revealed that the strongest chromosome 3, over 30 cM away from Vms1. enrichment terms were responses to virus and RIG-I–like receptor Differentially expressed genes following infection within Vms1 signaling (p , 0.00005). Highly expressed genes included effector (Adh1, Gbp1, and H28) were identified between A/J and B10.A ISG (Oas1g, Mx1), transcription factors (Irf7), and MHC class I and confirmed in CSS3 and B6; no genes were found within Vms1 (H2-d1, H2-k1). These results are summarized in Fig. 5 and de- that were differentially expressed between CSS3 and B6 but not tailed in Supplemental Table I. The observation that ISG pathways between A/J and B10.A (Fig. 7). themselves are strongly upregulated in resistant (B6, B10.A) and Similarly, genes not differentially expressed following infection, susceptible (CSS3, A/J) strains is consistent with the hypothesis but with basal expression differences (e.g., B10.A express Tnni3k that although IFN is required to survive systemic CVB3 infection before and PI at equivalent levels, whereas A/J do not) represent (10, 11), IFN may not be sufficient to control cardiac viral repli- possible candidate genes, as the presence or absence of one or cation (31, 32). more key genes may provide a mechanism of resistance (or sus- Downloaded from ceptibility) for which activity does not require induction. Between Differential microarray analysis reveals myocarditis CSS3 and B6, a total of 28 genes with basal expression differences susceptibility expression programs were identified. Of these, 14 were located on chromosome 3, and 4 As a robust IFN response seems approximately equal between of those were located within Vms1: Gbp1, Eif4e, Tnni3k, and strains, differential expression of sets of genes within a pathway Lrrc40.

may represent unequal mechanisms of adaptation to infection or http://www.jimmunol.org/ pathological processes. To determine expression programs differentially activated by infection, sets of genes were created that identify qualitatively different expression patterns (e.g., A/J increase expression of H28 PI, but B10.A do not). Using these priority ranked lists (Supplemental Table I), pathway analysis and functional grouping was performed. A/J-specific programs (Fig. 6A) were involved in metabolism, cell structure, and cardiovascular development pathways. B10.A differed in that cell-mediated immunity markers, Ag presentation, by guest on September 25, 2021 and immune cell trafficking were identified as primary pathways

FIGURE 5. Core program of highly and not differentially expressed genes responding to CVB3 infection. Microarray analysis of A/J, B10.A, FIGURE 6. Differentially expressed genes in A/J, B10.A, CSS3, and B6 CSS3, and B6 hearts at 4 d PI. Genes highly expressed or repressed PI (at mice identify common and distinct pathways. Mice strains (A/J, B10.A, least 5-fold change in all strains) were considered to be part of a common CSS3, and B6) are represented in columns stratified by infection status, pathway. Mice strains (A/J, B10.A, CSS3, and B6) are represented in where + indicates infected tissue and 2 indicates mock-infected control rows stratified by infection status, where + indicates infected tissue and tissue. A, A/J-specific genes (increased or decreased expression in A/J 2 indicates mock-infected control tissue. This set of 129 genes constitutes mice PI, but not in B10) cluster into pathways involving cardiovascular a shared program of response to infection and damage in both susceptible development, survival, inflammatory disease, and metabolism. B, B10.A- and resistant mice. Pathway analysis of these genes identifies inflammation specific genes differentially expressed compared with A/J were primarily and response to infection. Functional clustering suggested RIG-I–like involved in cell-mediated immunity and strongly indicative of a cytotoxic signaling was involved, and Interferome annotation identified 46 out of response. C, CSS3-specific genes (increased or decreased expression in 129 genes as known ISG. Network score represents a statistical likelihood CSS3 mice PI, but not in B6) were less numerous than A/J, but fell into that the pathway identified is involved in the tissue analyzed given the similar functional categories of development. There were no B6-specific genes involved. pathways identified. 6 DISTAL CHROMOSOME 3 CONTROLS VIRAL MYOCARDITIS IN MICE

Candidate gene analysis for Vms1 (as defined by either F2 cross) identified seven candidate genes for Vms1 on the basis of allele- specific expression, differential response to infection, or coding sequence variation. Of these, only Tnni3k, Lrrc40, H28, and Fpgt are contained within Vms1.1. Confirmation of significantly higher basal expression levels was found in Tnni3k by qPCR (Fig. 7D), but not in Lrrc40 (data not shown). Tnni3k was found to be 2.7- fold less expressed in Vms1s than Vms1r mice. Increased expression of H28 following infection in Vms1s and not Vms1r mice was also confirmed by qPCR (Fig. 7C). Finally, five coding nonsynonymous SNP (cnSNPs) within five genes (Tyw3, Tnni3k, Fpgt, Sfrs11, and Rpe65) have been identified within Vms1.1,but only two are predicted to be nonconservative and damaging to function: Fpgtrs30203847-C (C517R) and Tnni3krs30712233-T (T659I). After alignment, the Fpgtrs30203847-C allele possessed by B6, AKR, and 129S1 mice is predicted to be damaging (score 0.03) because the Fpgtrs30203847-R allele (A/J, BALB/c, C3H) is shared with most vertebrates. In fact, the cnSNP in Tnni3k predicted as damaging is

secondary to a more recently identiﬁed SNP, rs49812611, which Downloaded from accounts for the lack of basal expression of Tnni3k in A/J or CSS3 mice due to aberrant splicing (28). This interval shortens the list of likely candidate genes for Vms1.1 to H28, Tnni3k,orFpgt.

Discussion http://www.jimmunol.org/ The myocarditic potential of the coxsackieviruses was recognized FIGURE 7. Enhanced resolution of linkage of Vms1 to chromosome 3 shortly after their discovery (33). Despite considerable progress in by combined analysis and integration of candidate gene analysis. A, the understanding of CVB3 pathogenesis, coxsackieviral infection Linkage of viral replication to distal chromosome 3 is shown as LOD is a persistent cause of human disease (34, 35). There remains, proﬁles of A/J 3 B10.A]F2 and [CSS3 3 B6]F2 (gray and black lines, however, a need to account for variation of clinical presentation respectively) for the 16 cM support interval of Vms1. Combining both between any two otherwise healthy, normal, and immunocompe- analyses leads to signiﬁcantly improved resolution (dashed line), with tent individuals. Surveys of inbred mouse strains have long in- a new support interval of 4.5 cM linked to viral titer referred to as dicated a genetic component in susceptibility to CVB3-induced

Vms1.1a B by guest on September 25, 2021 . , Linkage of myocarditis score to distal chromosome 3 shown disease (14, 29). Our previous analysis identified a locus on as in A with a new support interval of 7.5 cM myocarditis score referred to chromosome 3, Vms1, which controlled several aspects of disease. as Vms1.1b. The overlapping region of Vms1.1a and Vms1.1b is Vms1.1. Candidate genes identified by differential expression (Adh1, Gbp1, H28), This led us to isolate the role of chromosome 3 from other po- by basal expression difference (Tnni3k), and containing nonsynonymous tentially segregating or confounding loci. mutations (Fpgt) are shown in approximate position. B, Box plots of CSS represent a valuable model in which to study complex traits. microarray data for candidate genes with expression differences. Hori- As a prelude to further refinement and positional cloning of Vms1, zontal lines represent the 40th and 60th percentile of expression of all it was necessary for us to at least replicate our initially observed genes. Adh1, Gbp1, and H28 are expressed at significantly higher levels in phenotypes in the CSS3 mice. CSS3 mice control for other pre- A/J and CSS3 mice following infection, but not in B10.A and B6 mice. viously identified loci respective to B6 (Vms2, Vms3 and H2); Gbp1, Eif4e, and Tnni3k were expressed at different levels even in the however, it was not obvious that Vms1s susceptibility would be absence of infection. C and D, Quantitative RT-PCR confirmation that H28 3 a s recaptured. Unlike our initial [B10.A A/J]F2 cross (all H2 ), is induced preferentially in Vms1 following infection and that Tnni3k b s H2 H2 transcript is less present in Vms1 at basal levels. CSS3 and B6 mice are ; is already known to be a strong modifier of disease (14, 29). CSS3 are also on a C57BL/6 and not C57BL/10 background; though B6 and B10 are closely related, we have found at least 7500 SNPs [S.A. Wiltshire, G.A. Leiva-Torres, Refined QTL and candidate gene prioritization and S.M. Vidal, unpublished observations and previous data (36)], A common set of genetic markers, as well as the high degree of some of which may cause interstrain differences in response to similarity between B10.A and B6 on chromosome 3, permitted infection. Nevertheless, phenotypes of both myocardial inflamma- a joint analysis with the aim of increasing linkage resolution using tion and viral replication were recapitulated. Variability in mean both the previous [A/J 3 B10.A]F2 and the present [CSS3 3 B6] viral titer was similar in [CSS3 3 B6]F2 (3.22 6 0.09 log [PFU/ F2 crosses. An increased number of meioses in the combined cross mg]) compared with the larger [A/J 3 B10.A]F2 cross (3.01 6 has considerably shortened the interval under consideration. The 0.06 log [PFU/mg]); however, Vms1 explained more variance in interval controlling viral replication was narrowed between the present cross (11 to 12% versus 3–6%). Even in this geneti- D3Mit116 and D3Mcg1 (78.7–83.29 cM), which we will refer to cally simplified context, a considerable amount of variation exists as Vms1.1a. Using the same technique, a region between between mice, even of identical genotype (Fig. 1). As an entero- D3Mit128 and D3Mcg1 (76.7–84 cM) was found to control in- virus, CVB3 exists as a quasispecies with a high mutation rate flammation, which we refer to as Vms1.1b. The Vms1 interval is (∼1 aa substitution/genome) (37, 38), providing ample opportu- thus reduced from 16 cM or 16 Mb as defined previously (16) to nity for every possible mutation within an infection in which 6.3 cM or 9 Mb containing Vms1.1a and Vms1.1b. The region infected tissue can contain as many as 105 PFU/mg. This provides containing both Vms1.1a and Vms1.1b will be referred to as an example of increased resolution and power by isolating genetic Vms1.1 for the sake of simplicity (Fig. 7A,7B). variation to a single chromosome despite considerable environ- The Journal of Immunology 7 mental variation. Replication of linkage between distal chromo- previously described as being presented during allograft rejection some 3 and both inflammation and viral titer phenotypes strongly between BALB.B (H2b) and B6 mice. Of the five cnSNPs in supports the strong role of Vms1 in infection. Vms1.1 (Tyw3, Tnni3k, Fpgt, Sfrs11, and Rpe65), only Tnni3k and Previous studies have successfully used gene expression to better Fpgt were both expressed in the heart and predicted to be dele- understand myocarditis and dilated cardiomyopathy (5, 39–41). In terious. Fpgt catalyzes the formation of the fucosyl donor GDP–b- addition to identifying likely candidate genes, global expression fucose as part of the fucose salvage pathway (46). L-fucose is analyses give us key mechanistic insights that aid in the identifi- a sugar present in glycoproteins involved in inflammation and cation of Vms1. Because viral replication in the heart is controlled immunity (e.g., sialyl-lewis X, the inflammatory ligand of selec- r in Vms1 mice as early as 4 d PI, this time point was chosen for tins, contains L-fucose). A possible link between the ability of expression analysis. It has previously been reported that A/J and infected cells to effect immediate inflammatory signals and Fpgt B6 mice differ substantially in their production of type I IFN may make this gene an attractive candidate. following infection (30, 42). Another QTL that has never been Tnni3k was identified as having a basal expression difference: cloned but influences peak IFN levels called If1 colocalizes with Vms1r mice express significantly higher basal levels of the gene Vms1 (43). These together suggested an initial hypothesis that than Vms1s mice. Indeed, the SNP rs49812611 was recently a sensing or signaling defect in IFN might be present in suscep- identified as leading to nonsense mediated decay of Tnni3k tran- tible mice. However, in this study, IFN response between sus- scripts (28). Tnni3k was identified as a cardiac specific tropinin- ceptible and resistant mice was roughly equivalent, or in the case interacting kinase in 2003 (47). Although the normal function of of A/J and B10.A mice, higher viral load was accompanied by Tnni3k is still obscure, some recent results have somewhat clari- higher IFN. Susceptible Vms1s mice sense virus and produce an- fied the situation. Transgenic overexpression of Tnni3k in differ- Downloaded from tiviral IFN-b with similar kinetics and magnitude as compared entiated P19cl6 cells led to increased contractile force and fre- with Vms1r mice. This makes it so that a defect in viral sensing, quency, as well as increased adrenergic response and protection signaling, or response to IFN is unlikely to underlie Vms1s. from ischemic injury (48). Tnni3k has also been implicated in Strain-specific expression analysis revealed similarities to pre- dilated cardiomyopathy (39) and heart failure. Mice not express- vious work and several novel results. Comparison between A/J and ing Tnni3k were resistant to CSQ heart failure (DBA/2, BALB,

B10.A mice showed A/J preferentially express genes and pathways and C3H), whereas normal expressers (B6, 129/X1, and AKR) or http://www.jimmunol.org/ involved in cardiovascular development and metabolism. This was mice transgenic for the human allele were highly susceptible to consistent with previous findings (also A/J mice), which postulated heart failure (28). This raises the possibility that Tnni3k may that these changes reflect undamaged cardiomyocytes compen- prevent myocardial injury during acute injury (48), but also be- sating for damaged neighbors (41). A notable addition to the A/J- come pathogenic and increasingly disregulated during chronic specific response was Amica1 (also known as JAML), which was disease (28, 39). highly induced (19.6-fold) in A/J compared with B6, B10.A, or This work was begun with the understanding that the outcome of CSS3 mice (1.3–3-fold). Amica1 is a costimulatory molecule that CVB3 infection is largely influenced by three factors: integrity of becomes strongly induced on g∂ T cells upon ligation to the CVB3 the myocardium, IFN response, and progression to adaptive im- receptor (Cxadr) on injured cells (44). g∂ T cells are known to munity. In this study, we present evidence that differential viral by guest on September 25, 2021 play a pathogenic role during viral myocarditis (45). Conversely, replication and myocardial inflammation at early time points may all mice of the B6/B10 background, including susceptible CSS3 be controlled by Tnni3k (implicating myocardial integrity), Fpgt mice, expressed several markers (Cd8b1, Nkg7, Cd244/2B4, (implicating cellular immune response), or H28 (by unknown Klra7) of antiviral CD8/NK immune responses. Lack of CD8/NK mechanisms). Future work will include the generation of congenic cells and overabundance of inflammatory g∂ T cells in A/J mice and subcongenic mice, as well as reverse genetic approaches, to suggests that this is another possible (non-Vms1s) mechanism assess the contribution of individual genes before progressing accounting for A/J-specific susceptibility to CVB3 infection. In further. contrast to A/J versus B10.A mice, expression profiling did not s r reveal any Vms1 or Vms1 -specific cell type, suggesting that Vms1 Acknowledgments does not control the recruitment of a cellular immune response. We thank Francois Pepin for continuing support with R, Tony Kwan for Three candidate genes were identified within the refined Vms1.1 advice on expression analysis, Mahmoud Aly, Emeric Bojarski, and Han locus via expression and publicly available SNP data (150–159.6 Yao for [A/J 3 B10.A]F2 data acquisition, and Eve-Marie Gendron- Mb on chromosome 3): H28, Tnni3k, and Fpgt. This method of Pontbriand for editing services. identifying candidate genes via associative data is limited in that if a modest (FC , 2-fold) change in expression underlies our phenotype, it would be missed. An alteration of protein function, but Disclosures not expression level, in genes containing nonsynonymous SNPs The authors have no financial conflicts of interest. (Tyw3, Sfrs11,orRpe65) is also possible even in the absence of predicted negative impact. We have also treated Vms1.1a and Vms1.1b controlling viral replication and myocarditis score, re- References 1. Cooper, L. T., Jr. 2009. Myocarditis. N. Engl. J. Med. 360: 1526–1538. spectively, as a single entity Vms1.1, which controls both. In the 2. Gauntt, C., and S. Huber. 2003. Coxsackievirus experimental heart diseases. absence of congenic and subcongenic mice that parse individual Front. Biosci. 8: e23–e35. gene contributions, it is impossible at this stage to identify 3. Klingel, K., C. Hohenadl, A. Canu, M. Albrecht, M. Seemann, G. Mall, and R. Kandolf. 1992. Ongoing enterovirus-induced myocarditis is associated with whether changes in expression are causal or consequential and of persistent heart muscle infection: quantitative analysis of virus replication, tissue which phenotype. However, differential control of viral replication damage, and inflammation. Proc. Natl. Acad. Sci. USA 89: 314–318. before myocarditis suggests protective mechanisms acting at day 4 4. Maisch, B., A. D. Ristic, I. Portig, and S. Pankuweit. 2003. Human viral car- r diomyopathy. Front. Biosci. 8: s39–s67. or earlier in Vms1 mice at which time H28, Tnni3k, and Fpgt 5. Ligons, D. L., M. L. Guler, H. S. Li, and N. R. Rose. 2009. A locus on chro- remain the most attractive candidates. mosome 1 promotes susceptibility of experimental autoimmune myocarditis s and lymphocyte cell death. Clin. Immunol. 130: 74–82. H28 is expressed in the Vms1 genotype following infection but 6. Xiong, D., T. Yajima, B. K. Lim, A. Stenbit, A. Dublin, N. D. Dalton, r not in Vms1 . H28 has no known function and has only been D. Summers-Torres, J. D. Molkentin, H. Duplain, R. Wessely, et al. 2007. 8 DISTAL CHROMOSOME 3 CONTROLS VIRAL MYOCARDITIS IN MICE

Inducible cardiac-restricted expression of enteroviral protease 2A is sufficient to 28. Wheeler, F. C., H. Tang, O. A. Marks, T. N. Hadnott, P. L. Chu, L. Mao, induce dilated cardiomyopathy. Circulation 115: 94–102. H. A. Rockman, and D. A. Marchuk. 2009. Tnni3k modifies disease progression 7. Garcıá-Sastre, A., and C. A. Biron. 2006. Type 1 interferons and the virus-host in murine models of cardiomyopathy. PLoS Genet. 5: e1000647. relationship: a lesson in de´tente. Science 312: 879–882. 29. Chow, L. H., C. J. Gauntt, and B. M. McManus. 1991. Differential effects of 8. Biron, C. A., K. B. Nguyen, G. C. Pien, L. P. Cousens, and T. P. Salazar-Mather. myocarditic variants of Coxsackievirus B3 in inbred mice. A pathologic char- 1999. Natural killer cells in antiviral defense: function and regulation by innate acterization of heart tissue damage. Lab. Invest. 64: 55–64. cytokines. Annu. Rev. Immunol. 17: 189–220. 30. Baig, E., and E. N. Fish. 2008. Distinct signature type I interferon responses are 9. Negishi, H., T. Osawa, K. Ogami, X. Ouyang, S. Sakaguchi, R. Koshiba, determined by the infecting virus and the target cell. Antivir. Ther. (Lond.) 13: H. Yanai, Y. Seko, H. Shitara, K. Bishop, et al. 2008. A critical link between 409–422. Toll-like receptor 3 and type II interferon signaling pathways in antiviral innate 31. Yajima, T., and K. U. Knowlton. 2009. Viral myocarditis: from the perspective of immunity. Proc. Natl. Acad. Sci. USA 105: 20446–20451. the virus. Circulation 119: 2615–2624. 10. Wessely, R., K. Klingel, K. U. Knowlton, and R. Kandolf. 2001. Cardioselective 32. Yasukawa, H., T. Yajima, H. Duplain, M. Iwatate, M. Kido, M. Hoshijima, infection with coxsackievirus B3 requires intact type I interferon signaling: M. D. Weitzman, T. Nakamura, S. Woodard, D. Xiong, et al. 2003. The sup- implications for mortality and early viral replication. Circulation 103: 756–761. pressor of cytokine signaling-1 (SOCS1) is a novel therapeutic target for 11. Deonarain, R., D. Cerullo, K. Fuse, P. P. Liu, and E. N. Fish. 2004. Protective enterovirus-induced cardiac injury. J. Clin. Invest. 111: 469–478. role for interferon-beta in coxsackievirus B3 infection. Circulation 110: 3540– 33. Melnick, J. L. 1950. The poliomyelitis, encephalomyocarditis, and coxsackie 3543. groups of viruses. Bacteriol. Rev. 14: 233–244. 12. Yuan, J., Z. Liu, T. Lim, H. Zhang, J. He, E. Walker, C. Shier, Y. Wang, Y. Su, 34. Khetsuriani, N., A. Lamonte-Fowlkes, S. Oberst, M. A. Pallansch; Centers for A. Sall, et al. 2009. CXCL10 inhibits viral replication through recruitment of Disease Control and Prevention. 2006. Enterovirus surveillance—United States, natural killer cells in coxsackievirus B3-induced myocarditis. Circ. Res. 104: 1970-2005. MMWR Surveill. Summ. 55: 1–20. 628–638. 35. Spanakis, N., E. N. Manolis, A. Tsakris, S. Tsiodras, T. Panagiotopoulos, 13. Poffenberger, M. C., I. Shanina, C. Aw, N. El Wharry, N. Straka, D. Fang, G. Saroglou, and N. J. Legakis. 2005. Coxsackievirus B3 sequences in the A. E. Baskin-Hill, S. H. Spiezio, J. H. Nadeau, and M. S. Horwitz. 2010. Novel myocardium of fatal cases in a cluster of acute myocarditis in Greece. J. Clin. nonmajor histocompatibility complex-linked loci from mouse chromosome 17 Pathol. 58: 357–360. confer susceptibility to viral-mediated chronic autoimmune myocarditis. Circ. 36. Xia, Y., S. Won, X. Du, P. Lin, C. Ross, D. La Vine, S. Wiltshire, G. Leiva, Cardiovasc. Genet. 3: 399–408. S. M. Vidal, B. Whittle, et al. 2010. Bulk segregation mapping of mutations in Downloaded from 14. Wolfgram, L. J., K. W. Beisel, A. Herskowitz, and N. R. Rose. 1986. Variations closely related strains of mice. Genetics 186: 1139–1146. in the susceptibility to Coxsackievirus B3-induced myocarditis among different 37. Drake, J. W. 1999. The distribution of rates of spontaneous mutation over strains of mice. J. Immunol. 136: 1846–1852. viruses, prokaryotes, and eukaryotes. Ann. N. Y. Acad. Sci. 870: 100–107. 15. Gorbea, C., K. A. Makar, M. Pauschinger, G. Pratt, J. L. Bersola, J. Varela, 38. Duffy, S., L. A. Shackelton, and E. C. Holmes. 2008. Rates of evolutionary R. M. David, L. Banks, C. H. Huang, H. Li, et al. 2010. A role for Toll-like change in viruses: patterns and determinants. Nat. Rev. Genet. 9: 267–276. receptor 3 variants in host susceptibility to enteroviral myocarditis and dilated 39. Barth, A. S., R. Kuner, A. Buness, M. Ruschhaupt, S. Merk, L. Zwermann, cardiomyopathy. J. Biol. Chem. 285: 23208–23223. S. Kaä¨b, E. Kreuzer, G. Steinbeck, U. Mansmann, et al. 2006. Identification of

16. Aly, M., S. Wiltshire, G. Chahrour, J. C. Osti, and S. M. Vidal. 2007. Complex a common gene expression signature in dilated cardiomyopathy across in- http://www.jimmunol.org/ genetic control of host susceptibility to coxsackievirus B3-induced myocarditis. dependent microarray studies. J. Am. Coll. Cardiol. 48: 1610–1617. Genes Immun. 8: 193–204. 40. Szalay, G., S. Meiners, A. Voigt, J. Lauber, C. Spieth, N. Speer, M. Sauter, 17. Nadeau, J. H., J. B. Singer, A. Matin, and E. S. Lander. 2000. Analysing complex U. Kuckelkorn, A. Zell, K. Klingel, et al. 2006. Ongoing coxsackievirus myo- genetic traits with chromosome substitution strains. Nat. Genet. 24: 221–225. carditis is associated with increased formation and activity of myocardial 18. Knowlton, K. U., E. S. Jeon, N. Berkley, R. Wessely, and S. Huber. 1996. A immunoproteasomes. Am. J. Pathol. 168: 1542–1552. mutation in the puff region of VP2 attenuates the myocarditic phenotype of an 41. Taylor, L. A., C. M. Carthy, D. Yang, K. Saad, D. Wong, G. Schreiner, infectious cDNA of the Woodruff variant of coxsackievirus B3. J. Virol. 70: L. W. Stanton, and B. M. McManus. 2000. Host gene regulation during cox- 7811–7818. sackievirus B3 infection in mice: assessment by microarrays. Circ. Res. 87: 328– 19. Broman, K. W., H. Wu, S. Sen, and G. A. Churchill. 2003. R/qtl: QTL mapping 334. in experimental crosses. Bioinformatics 19: 889–890. 42. Ja¨kel, S., U. Kuckelkorn, G. Szalay, M. Plo¨tz, K. Textoris-Taube, E. Opitz, 20. Li, R., M. A. Lyons, H. Wittenburg, B. Paigen, and G. A. Churchill. 2005. K. Klingel, S. Stevanovic, R. Kandolf, K. Kotsch, et al. 2009. Differential in-

Combining data from multiple inbred line crosses improves the power and terferon responses enhance viral epitope generation by myocardial immuno- by guest on September 25, 2021 resolution of quantitative trait loci mapping. Genetics 169: 1699–1709. proteasomes in murine enterovirus myocarditis. Am. J. Pathol. 175: 510–518. 21. Ng, P. C., and S. Henikoff. 2002. Accounting for human polymorphisms pre- 43. Kozak, C. A., Y. Su, N. B. Raj, and P. M. Pitha. 1999. Identification and genetic dicted to affect protein function. Genome Res. 12: 436–446. mapping of differentially expressed genes in mice differing at the If1 interferon 22. de Veer, M. J., M. Holko, M. Frevel, E. Walker, S. Der, J. M. Paranjape, regulatory locus. Mamm. Genome 10: 853–857. R. H. Silverman, and B. R. Williams. 2001. Functional classification of 44. Verdino, P., D. A. Witherden, W. L. Havran, and I. A. Wilson. 2010. The mo- interferon-stimulated genes identified using microarrays. J. Leukoc. Biol. 69: lecular interaction of CAR and JAML recruits the central cell signal transducer 912–920. PI3K. Science 329: 1210–1214. 23. Samarajiwa, S. A., S. Forster, K. Auchettl, and P. J. Hertzog. 2009. INTER- 45. Huber, S. A., A. Moraska, and M. Choate. 1992. T cells expressing the gamma FEROME: the database of interferon regulated genes. Nucleic Acids Res. 37 delta T-cell receptor potentiate coxsackievirus B3-induced myocarditis. J. Virol. (Database issue): D852–D857. 66: 6541–6546. 24. Huang, da W., B. T. Sherman, and R. A. Lempicki. 2009. Systematic and in- 46. Pastuszak, I., C. Ketchum, G. Hermanson, E. J. Sjoberg, R. Drake, and tegrative analysis of large gene lists using DAVID bioinformatics resources. Nat. A. D. Elbein. 1998. GDP-L-fucose pyrophosphorylase. Purification, cDNA Protoc. 4: 44–57. cloning, and properties of the enzyme. J. Biol. Chem. 273: 30165–30174. 25. Blazejczyk, M. M., and R. Nadon. 2007. FlexArray: A Statistical Data Analysis 47. Zhao, Y., X. M. Meng, Y. J. Wei, X. W. Zhao, D. Q. Liu, H. Q. Cao, C. C. Liew, Software for Gene Expression Microarrays. Genome Quebec, Montreal, Quebec, and J. F. Ding. 2003. Cloning and characterization of a novel cardiac-specific Canada. kinase that interacts specifically with cardiac troponin I. J. Mol. Med. 81: 297–304. 26. Baldi, P., and A. D. Long. 2001. A Bayesian framework for the analysis of 48. Lai, Z. F., Y. Z. Chen, L. P. Feng, X. M. Meng, J. F. Ding, L. Y. Wang, J. Ye, microarray expression data: regularized t -test and statistical inferences of gene P. Li, X. S. Cheng, Y. Kitamoto, et al. 2008. Overexpression of TNNI3K, changes. Bioinformatics 17: 509–519. a cardiac-specific MAP kinase, promotes P19CL6-derived cardiac myogenesis 27. Rozen, S., and H. Skaletsky. 2000. Primer3 on the WWW for general users and and prevents myocardial infarction-induced injury. Am. J. Physiol. Heart Circ. for biologist programmers. Methods Mol. Biol. 132: 365–386. Physiol. 295: H708–H716.