Basal and Reovirus-Induced Beta Interferon (IFN-β) and IFN-β-Stimulated Gene Expression Are Cell Type Specific in the Cardiac Protective Response

Abstract

Viral myocarditis is an important human disease, with a wide variety of viruses implicated. Cardiac myocytes are not replenished yet are critical for host survival and thus may have a unique response to infection. Previously, we determined that the extent of reovirus induction of beta interferon (IFN-β) and IFN-β-mediated protection in primary cardiac myocyte cultures was inversely correlated with the extent of reovirus-induced cardiac damage in a mouse model. Surprisingly, and in contrast, the IFN-β response did not determine reovirus replication in skeletal muscle cells. Here we compared the IFN-β response in cardiac myocytes to that in primary cardiac fibroblast cultures, a readily replenished cardiac cell type. We compared basal and reovirus-induced expression of IFN-β, IRF-7 (an interferon-stimulated gene [ISG] that further induces IFN-β), and another ISG (561) in the two cell types by using real-time reverse transcription-PCR. Basal IFN-β, IRF-7, and 561 expression was higher in cardiac myocytes than in cardiac fibroblasts. Reovirus T3D induced greater expression of IFN-β in cardiac myocytes than in cardiac fibroblasts but equivalent expression of IRF-7 and 561 in the two cell types (though fold induction for IRF-7 and 561 was higher in fibroblasts than in myocytes because of the differences in basal expression). Interestingly, while reovirus replicated to equivalent titers in cardiac myocytes and cardiac fibroblasts, removal of IFN-β resulted in 10-fold-greater reovirus replication in the fibroblasts than in the myocytes. Together the data suggest that the IFN-β response controls reovirus replication equivalently in the two cell types. In the absence of reovirus-induced IFN-β, however, reovirus replicates to higher titers in cardiac fibroblasts than in cardiac myocytes, suggesting that the higher basal IFN-β and ISG expression in myocytes may play an important protective role.

Viral myocarditis has occurred in an estimated 5 to 20% of the human population. It is often fatal in infants and, although usually resolved in adults, can progress to chronic myocarditis, dilated cardiomyopathy, and/or cardiac failure (62). In recent years, clinical trials of antiviral agents have shown promise in the treatment of the disease. Specifically, alpha interferon (IFN-α) (9, 17, 36, 37, 55) and IFN-β (26) have demonstrated an ability to decrease viral load, improve cardiac function, and increase the survival rate in patients with the disease. However, none of these therapies can completely restore cardiac function.

Many viruses have been implicated in human myocarditis (62), with enteroviruses and adenoviruses accounting for the majority of cases (33, 42, 43, 58). Enterovirus-induced myocarditis is both immune mediated (8, 42) and due to direct cytopathic effect (6, 20). In human myocarditis associated with adenovirus (33) or human immunodeficiency virus, the severity of disease correlates poorly with the presence of an inflammatory infiltrate, suggesting a direct viral effect. However, non-immune-mediated mechanisms of viral myocarditis remain largely unexplored.

Reovirus-induced myocarditis is not immune mediated (49, 50); indeed, the immune system is protective against cardiac damage (50). Instead, reovirus-induced myocarditis reflects virally induced apoptosis of cardiac cells (11). Thus, reovirus infection provides an excellent model for studying direct cytopathic effect in the heart. Previously, we demonstrated that in primary cardiac myocyte cultures (PCMCs), reovirus induction of, and sensitivity to, IFN-β is a determinant of viral myocarditic potential. Specifically, nonmyocarditic reoviruses induce more IFN-β and/or are more sensitive to the antiviral effects of IFN-β than myocarditic reoviruses (52).

Importantly, IFN-β mediates its antiviral effects though the induction of IFN-stimulated genes (ISGs). Following secretion from infected cells, IFN-β binds to the IFN-α/β receptor, resulting in receptor dimerization and activation of receptor-associated kinases of the Janus kinase family (JAKs). Specifically, Jak1 and Tyk2 are activated and phosphorylate tyrosine residues on STAT1 and STAT2 (18, 39, 46, 53). These newly phosphorylated STAT proteins associate with interferon regulatory factor 9 (IRF-9, or p48) to form the multimeric protein complex ISGF3, which migrates into the nucleus to bind gene promoter regions containing an IFN-stimulated response element (ISRE) (53).

To date, more than 300 genes have been identified as ISGs (5, 12, 13). Classical ISGs include those encoding PKR, 2′-5′ oligoadenylate synthetase, and Mx family GTPases. Notably, while PKR is critical for protection against reovirus-induced myocarditis (54), the role of other ISGs in protection against myocarditis is unclear. The ISG 561-encoded protein, p56, is the most strongly induced gene product following addition of IFN-α/β (12) to human cells, and like PKR, it inhibits translation in virally affected cells (15, 22). Specifically, p56 interacts with eukaryotic initiation factor 3 (eIF3), thereby limiting its interactions with eIF2 and the 40S ribosomal subunit and compromising the translation initiation complex (4, 22, 27).

In addition to upregulating ISGs that are directly antiviral, ISGF3 also binds to the IRF-7 promoter, which contains two tandem ISRE sequences that confer ISGF3 binding (31, 56). Importantly, IRF-7 is a transcription factor responsible for an amplification of the IFN response. While initial IFN-β induction is predominantly IRF-3 dependent (25, 29, 30, 60), robust and sustained IFN-β production is IRF-7 dependent (32). Unlike many other ISGs, IRF-7 is regulated solely via ISGF3: fibroblasts null for IRF-9 and thus are unable to form ISGF3, do not generate IRF-7 following IFN treatment (45).

Cardiac myocytes are exposed to a variety of viruses, are not replenished, and yet are critical to survival, suggesting that this highly specialized cell type may have evolved a unique response to viral infection. We have previously shown that reovirus induction of IFN-β and IFN-β-mediated protection in PCMCs correlate inversely with viral spread in PCMCs and the degree of cardiac damage in a mouse model (52). Surprisingly, and in contrast, the IFN response in skeletal muscle cells is not a determinant of viral replication (52), suggesting a cell type-specific role for IFN-β in antiviral protection. Here, we compare the IFN-β response in PCMCs to that in primary cardiac fibroblast cultures (PCFCs), a readily replenished cardiac cell type. We use the nonmyocarditic reovirus strain T3D, because it both induces IFN-β well and is highly sensitive to the antiviral effects of IFN-β, a characteristic not common to myocarditic reovirus strains (52). Results demonstrate that IFN-β and the ISGs IRF-7 and 561 are each basally expressed at higher levels in PCMCs than in PCFCs. Moreover, while reovirus T3D induced expression of IFN-β to significantly higher levels in PCMCs than in PCFCs, it induced similar expression of IRF-7 and 561 in the two cell types. Interestingly, we also show that while reovirus replicated to equivalent titers in PCMCs and PCFCs, removal of IFN-β resulted in significantly greater reovirus replication in PCFCs than in PCMCs. Together, these data indicate that the IFN-β response is more critical for protection in PCFCs than in PCMCs. Additionally, these data suggest that in the absence of IFN-β-stimulated gene expression, higher basal IFN-β and ISG expression in PCMCs may play a protective role.

MATERIALS AND METHODS

Mice and inoculations.

Timed pregnant Cr:NIH(S) mice were obtained from the National Cancer Institute. IFN-α/β receptor^−/− (38) and wild-type parental 129Sv/Ev mice were maintained in breeding colonies for timed matings to generate fetuses for primary cultures. Mice were housed according to the recommendations of the Association for Assessment and Accreditation of Laboratory Animal Care, and all procedures were approved by the North Carolina State University institutional animal care and use committee.

Cell cultures.

To generate primary cardiac myocyte and fibroblast cultures from Cr:NIH(S), IFN-α/β receptor^−/− (38), or 129Sv/Ev mice, term fetuses or 1- to 2-day-old neonates were euthanized and the apical two-thirds of the hearts were removed, minced, and trypsinized (3). Cells were plated at a density of 1.25 × 10⁶ per well in six-well clusters (Costar, Cambridge, Mass.) and incubated for 1.5 to 2 h in order to isolate rapidly adhering cardiac fibroblasts. Fibroblasts were trypsinized from six-well clusters and resuspended in Dulbecco's modified Eagle medium (DMEM) (Gibco BRL, Gaithersburg, Md.) supplemented with 7% fetal calf serum (HyClone, Logan, Utah) and 50 μg of Mezlin (Miles Labs) per ml. Fibroblasts were then plated as indicated for each procedure. The remaining cells (predominantly myocytes) were resuspended in DMEM (Gibco BRL) supplemented with 7% fetal calf serum (HyClone), 0.06% thymidine (Sigma Co., St. Louis, Mo.), and 50 μg of Mezlin (Miles Labs) per ml and plated as indicated for each procedure. Myocyte cultures contained 5 to 20% fibroblasts (3), consistent with levels reported by others (16, 21, 34), and consistent with cell heterogeneity in the heart. Mouse L929 cells were maintained in suspension culture in minimal essential medium (Gibco BRL) supplemented with 5% fetal calf serum (HyClone) and 2 mM l-glutamine (Gibco BRL).

Viruses.

All reovirus stocks (triply plaqued and passaged twice in mouse L929 cells) were characterized previously for their myocarditic phenotypes (47, 49, 51). Virus 8B is a reassortant virus derived from a mouse infected with serotype 1 Lang (T1L) and serotype 3 Dearing (T3D) (51). All other reassortant viruses were derived from mouse L929 cells infected with 8B-, T1L-, or T3D-derived viruses (49). All EW-series reassortant viruses were derived from 8B and EW121. Viral myocarditic potentials were determined by injecting 2 × 10⁵ to 5 × 10⁷ PFU into two 2-day-old Cr:NIH(S) mice and examining their hearts for macroscopic lesions (gross myocarditis) at death or 14 days postinjection (47, 49).

Infections for RNA harvests.

Reovirus infected similar percentages of cells in PCMCs and PCFCs by immunocytochemistry (data not shown). PCMCs were plated at confluency (10⁶ cells), and PCFCs were plated at half-confluency (5 × 10⁵ cells), in a 24-well cluster; PCFCs were confluent by the day of infection (10⁶ cells). Two days postplating, PCMCs or PCFCs were washed twice with supplemented DMEM and then immediately infected with reovirus T3D (Dearing) at 10 PFU per cell in 300 μl of supplemented medium. After incubation for 1 h at 37°C under 5% CO₂, 700 μl of a DMEM overlay was added, and these cultures were harvested at the indicated times. “Mock infected” refers to uninfected cultures washed with supplemented DMEM and harvested either immediately or 5 to 6 h after being overlaid with 1 ml of supplemented DMEM. For harvests performed at 0 h, virus was added and immediately removed (with no incubation time), and cells were harvested.

Infections for viral replication studies.

PCMCs and PCFCs were plated at densities of 4.0 × 10⁵ and 2.0 × 10⁵ per well, respectively, in 300 μl in 96-well tissue culture plates (Costar). At 2 days postplating, cultures were infected with a panel of parental and reassortant reoviruses at 0.1 PFU per cell. Cultures were treated with either 3 μl of an antibody containing 165 National Institutes of Health neutralizing units of rabbit-anti-mouse IFN-α/β (catalog no. 21032; Lee Biomolecular Research, Inc., San Diego, Calif.) or a control antibody. Seven days postinfection, cell monolayers were frozen at −70°C and subjected to additional freeze-thaw cycles. Cultures were then lysed in 0.5% Nonidet P-40, and titers were determined by plaque assay on mouse L-cell monolayers as previously described (48).

RNA harvest and reverse transcription.

At the indicated times postinfection, supernatants were aspirated. Cells were lysed directly from the culture plates and homogenized by using Qiashredders (QIAGEN), and total RNA was isolated by using an RNeasy kit (QIAGEN). Contaminating genomic DNA was removed by DNase treatment using an RNase-free DNase set (QIAGEN). Total RNA was extracted and quantified in initial experiments by fluorometry using the RiboGreen total RNA detection kit (Molecular Probes; Eugene, Oreg.). RNA from each well was then subjected to reverse transcription using the following components (final concentrations) for a total reaction volume of 50 μl: 5 μM oligo(dT) (Promega), 1× Taq buffer, 7.5 mM MgCl₂, 1 mM dithiothreitol, 1 mM each deoxynucleoside triphosphate, 0.5 U of RNA Guard (Pharmacia)/μl, and 0.22 U of avian myeloblastosis virus reverse transcriptase (Promega)/μl.

Real-time PCR.

Approximately 10% of the reverse transcription reaction product was then amplified by real-time PCR. Experiments were performed in duplicate 25-μl reaction mixtures in 96 well plates (Bio-Rad) using Quantitech master mix (QIAGEN) spiked with 10 nM fluorescein (Molecular Probes) to optimize fluorescent data quality and analysis. Duplicate PCRs were carried out in 96-well plates with optical sealing tape (Bio-Rad). Amplification, quantification, and melting curve analysis (detection of specific products) were performed on an iCycler iQ fluorescence thermocycler (Bio-Rad) with the following cycle profile: 95°C (PCR initial activation step, 13.5 min), followed by 50 cycles of 95°C for 10 s (denaturation) and 59°C for 60 s (annealing, synthesis, and fluorescent data collection step), followed by a melting curve protocol designed for decrement temperatures of 0.5°C with a starting temperature of 95°C and an ending temperature of 50°C.

The sequences for the primers were selected by using online software designed by Steve Rosen and Helen J. Skaletsky (1998) (available at http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi). Primers were checked for base complementarity by using http://www.operon.com/oligos/toolkit.php (online software provided by QIAGEN, 1996 to 2003). Primers were then designed as follows: IRF-7 forward primer, 5′ CCCATCTTCGACTTCAGCAC 3′; IRF-7 reverse primer, 5′ TGTAGTGTGGTGACCCTTGC 3′; 561 forward primer, 5′ TGGCCGTTTCCTACAGTTT 3′; 561 reverse primer, 5′ TCCTCCAAGCAAAGGACTTC 3′; IFN-β forward primer, 5′ GGAGATGACGGAGAAGATGC 3′; IFN-β reverse primer, 5′ CCCAGTGCTGGAGAAATTGT 3′; glyceraldehyde-3-phosphate dehydrogenase (GAPDH) forward primer, 5′ CAACTACATGGTCTACATGTTC 3′; GAPDH reverse primer, 5′ CTCGCTCCTGGAAGATG 3′.

Note that the sequence of the IRF-7 amplicon is present in all three IRF-7 spliceoforms (α, β, and γ); therefore, all spliced variants are detected. To determine the copy number of experimental samples, standard curves of known concentrations of DNA were generated for each gene of interest on each plate, using four 10-fold dilutions of PCR-generated DNA fragments which included the amplicon sequence. To generate these DNA fragments, the following primers were used to amplify sequences from mouse DNA: IRF-7 forward primer, 5′ TGGGTTCCTGGATGTGAC 3′; IRF-7 reverse primer, 5′ TTCACCAGGATCAGGGTC 3′; 561 forward primer, 5′ CTGAGGCCCACATTTGAGAT 3′; 561 reverse primer, 5′ GGAGCATTGGAACACTTGGT; IFN-β forward primer, 5′ GCGTTCCTGCTGTGCTTC 3′; IFN-β reverse primer, 5′ CCATCCAGGCGTAGCTG 3′; GAPDH forward primer, 5′ GTGAAGGTCGGTGTGAACGG 3′; GAPDH reverse primer, 5′ GTGGCAGTGATGGCATGGAC 3′.

These DNA fragments were then purified by gel extraction and quantified by fluorometry. iCycler software was used to analyze data (confirming single peak melt curve) and generate standard curves on each plate, and then copy numbers for the gene of interest were determined relative to GAPDH expression. Although the average total yields of RNA were similar for PCMCs and PCFCs (1.3-fold difference as determined by fluorometry [data not shown]), GAPDH basal expression was 3.0-fold greater in PCMCs than in PCFCs (data not shown). Thus, data were normalized to compensate for this difference. GAPDH expression remained constant, regardless of reovirus infection or addition of IFN (data not shown). Controls using water or reverse-transcribed water were included for each primer set and were always negative.

Statistical analysis.

A Student one-tailed t test and pooled variance were used for statistical analysis. Data were subjected to outlier analysis (Systat 9.0). For analysis of viral replication, the nonparametric Kruskal-Wallis test (Systat 9.0) was used. In all cases, results were considered significant at a P value of ≤0.05.

RESULTS

Reovirus T3D induces greater IFN-β expression in PCMCs than in PCFCs.

The possibility that the innate cardiac response to viral infection is cell type specific had not previously been explored in any viral system. Here, we compared viral induction of IFN-β gene expression in PCMCs to that in PCFCs. PCMCs and PCFCs were derived from Cr:NIH(S) (wild-type) mice. Two or 3 days postplating, cultures were mock treated or virally infected, total RNAs were extracted and reverse transcribed, and cDNAs were then analyzed by real-time PCR to determine IFN-β absolute expression (mRNA copy number). Basal IFN-β expression was 3.0-fold higher in PCMCs than in PCFCs (Fig. 1A) (P < 0.001). In addition, at multiple time points postinfection, T3D induced IFN-β expression (copy number) to significantly higher levels in PCMCs than in PCFCs (Fig. 1B) (3.8-fold at 4 h [P = 0.03]; 12.8-fold at 8 h [P = 0.01]). When results from Fig. 1A and B were combined to express data as fold induction, T3D was again found to induce IFN-β expression more in PCMCs than in PCFCs (Fig. 1C).

FIG. 1.

Basal and induced IFN-β gene expression is greater in PCMCs than in PCFCs. PCMCs and PCFCs were derived from Cr:NIH(S) mice. Cultures were mock treated or infected with reovirus T3D at a multiplicity of infection of 10 PFU/cell. Total RNA was extracted and quantified via fluorometry. Equivalent RNA was reverse transcribed, cDNAs were analyzed via real-time PCR, and the IFN-β copy number was normalized to the GAPDH copy number for each well. Asterisk indicates a significant difference in IFN-β expression between PCMCs and PCFCs. (A) Basal expression of IFN-β in PCMCs and PCFCs. Bars are averages of six replicates from three independently derived myocyte and fibroblast cultures ± standard errors of the means (SEM). (B) Reovirus T3D induction of IFN-β in PCMCs and PCFCs. Each time point is the average of five or six replicates from three independently derived myocyte and fibroblast cultures (except for PCFC values for 2 and 4 h, which are averages of four replicates from two cultures) ± SEM (present, but not discernible for all symbols). (C) Data from panels A and B are combined to express results as fold induction over mock expression. Note that, since these data are ratios of data from the two previous panels, there are no SEM.

Basal IRF-7 expression is higher in PCMCs than in PCFCs, but reovirus induction of IRF-7 is similar.

Basal and reovirus-induced expression of IFN-β was greater in PCMCs than in PCFCs; thus, we inquired whether basal and/or induced ISG expression was similarly higher in PCMCs. IRF-7 was specifically chosen for this experiment because it is an ISG solely regulated by ISGF3 (45) and furthermore because it is critical to the robust amplification of the IFN-β response. Reverse transcription and real-time PCR were performed on RNAs from mock- and virus-infected PCMCs and PCFCs as described above. Basal expression of IRF-7 was dramatically higher in PCMCs than in PCFCs (12.7-fold; P < 0.001 [Fig. 2A ]). In contrast, reovirus T3D induced similar IRF-7 expression (copy numbers) in PCMCs and PCFCs (Fig. 2B) except at 8 h postinfection, when the difference was only 1.9-fold (with no increase in difference at later times [data not shown]). When the data in Fig. 2A and B were combined, fold induction of IRF-7 was found to be dramatically higher in PCFCs (3.5- to 17.5-fold) than in PCMCs (0.63- to 2.3-fold) (Fig. 2C). Thus, dramatically different basal levels of IRF-7 expression in the two cell types resulted in a large difference in fold induction, despite a minimal difference in final copy number. Interestingly, when reovirus T3D was added to PCMCs and PCFCs and then immediately washed off, IRF-7 was induced in PCFCs but not in PCMCs relative to basal expression (3.8-fold in PCFCs [P < 0.01], similar to fold induction at 2 h in Fig. 2C; 0.84-fold in PCMCs [P > 0.1] [data not shown]). Future studies will address whether this finding reflects cell type-specific signaling upon virus binding.

FIG. 2.

IRF-7 gene expression in PCMCs and PCFCs. Reverse transcription and real-time PCR were performed on RNAs from mock- and T3D-infected PCMCs and PCFCs as for Fig. 1. Asterisk indicates a significant difference in IRF-7 expression between PCMCs and PCFCs. (A) Basal expression of IRF-7 in PCMCs and PCFCs. Bars are averages of 17 or 18 replicates from at least eight independently derived primary cell cultures ± standard errors of the means (SEM). (B) Reovirus T3D induction of IRF-7 (copy number) in PCMCs and PCFCs. Each time point is the average of six to eight replicates from three to four myocyte and fibroblast cultures (except for PCFC values for 2 and 4 h, which are averages of four replicates from two cultures) ± SEM. (C) Data from panels A and B are combined to express results as fold induction over mock expression. Again, because these data are expressed as ratios of data from the two previous panels, there are no SEM.

Basal expression of ISG 561 is higher in PCMCs than in PCFCs, but reovirus induction of 561 is similar.

The ISG 561 is the most strongly induced ISG following treatment of human cells with IFN-α/β (12), but unlike induction of IRF-7, virus has been shown to directly induce expression from the 561 promoter in the absence of IFN (15, 59). This direct induction of 561 is most likely mediated by IRF-3 or IRF-1 binding to the 561 ISRE (2, 7, 10, 14, 15, 40). Importantly, direct induction of 561 is cell type and/or stimulus specific (2, 19). As with IRF-7, basal expression of 561 was higher in PCMCs than in PCFCs (3.6-fold; P < 0.001 [Fig. 3A ]). However, in contrast to IRF-7, reovirus T3D induced 561 with cell type-specific kinetics (Fig. 3B). After infection, the copy number of 561 increased and then decreased after 4 h in PCFCs but continued to rise in PCMCs, resulting in a significant difference in 561 expression at 8 h postinfection in PCMCs (fourfold; P < 0.001). Expression of 561 remained elevated in PCMCs at 10 and 12 h postinfection (data not shown). As with IRF-7, when the data in Fig. 3A and B were combined, fold induction provided a different curve because of the differences in basal expression. At 4 h postinfection, despite little difference in absolute copy numbers, fold induction differed dramatically: 12.1-fold in PCMCs compared to 52.7-fold in PCFCs (Fig. 3C). Moreover, while at 8 h postinfection PCMCs expressed fourfold more 561 than did PCFCs (Fig. 3B), there was a negligible difference in relative induction of 561 between the cell types (Fig. 3C). Thus, as with IRF-7, differences in basal 561 expression between PCMCs and PCFCs resulted in a curve for fold induction that did not reflect the actual final copy numbers.

FIG. 3.

561 gene expression in PCMCs and PCFCs. Reverse transcription and real-time PCR were performed on RNAs from mock- and T3D-infected PCMCs and PCFCs as for Fig. 1. Asterisk indicates a significant difference in 561 expression between PCMCs and PCFCs. (A) Basal expression of 561 in PCMCs and PCFCs. Bars are averages of 20 or 16 replicates (respectively) from at least eight independently derived primary cell cultures ± standard errors of the means (SEM). (B) Reovirus T3D induction of 561 (copy number) in PCMCs and PCFCs. Each time point is the average of 6 to 10 replicates from three to five myocyte and fibroblast cultures ± SEM. (C) Data from panels A and B are combined to express results as fold induction over mock expression. Again, because these data are expressed as ratios of data from the two previous panels, there are no SEM.

IFN-α/β treatment is similar to viral infection.

To determine whether viral induction of IRF-7 and 561 in PCMCs and PCFCs reflected additional virus-specific events other than induction of IFN-β, PCMCs and PCFCs were overlaid with a medium containing IFN-α/β and then harvested for analysis by real-time PCR. While IFN-β induced the expression of IRF-7 (Fig. 4A) more rapidly than T3D did (Fig. 2B), the overall pattern of induction was similar, in that copy numbers of IRF-7 were comparable in PCMCs and PCFCs but fold induction (Fig. 4B) was dramatically different due to differences in basal expression. In contrast, while IFN-α/β induction of the ISG 561 in PCFCs (Fig. 5A) followed kinetics similar to those of reovirus induction of 561 in that cell type (Fig. 3B), IFN-α/β (Fig. 5A) and reovirus (Fig. 3B) induction of 561 followed different kinetics in PCMCs. Specifically, while 561 expression remained elevated following reovirus infection, 561 expression rapidly decreased in PCMCs following IFN-α/β stimulation, kinetics similar to those of 561 in PCFCs. Again, differences in the fold induction of 561 expression (Fig. 5B) between PCMCs and PCFCs were largely determined by differences in basal expression.

FIG. 4.

(A) IFN-α/β induction of IRF-7 in PCMCs and PCFCs. Reverse transcription and real-time PCR were performed on RNAs from mock- and IFN-treated PCMCs and PCFCs as for Fig. 1. Asterisk indicates a significant difference in IRF-7 expression between PCMCs and PCFCs. Each time point is the average of four to eight replicates from two or four myocyte and fibroblast cultures ± standard errors of the means (SEM). (B) Data from Fig. 2A and 4A are combined to express results as fold induction over mock expression. Again, because these data are expressed as ratios of data from the two previous panels, there are no SEM.

FIG. 5.

(A) IFN-α/β induction of 561 in PCMCs and PCFCs. Reverse transcription and real-time PCR were performed on RNAs from mock- and IFN-treated PCMCs and PCFCs as for Fig. 1. Each time point is the average of four to six replicates from two to three myocyte and fibroblast cultures (except at 4 h, where values are averages of two replicates from a single experiment) ± standard errors of the means (SEM). (B) Data from Fig. 3A and 5A are combined to express results as fold induction over mock expression. Again, because these data are expressed as ratios of data from the two previous panels, there are no SEM.

PCFCs are more dependent on the IFN response than are PCMCs.

The data presented in Fig. 1 to 5 suggest that relative (fold) induction of ISGs is determined, at least in part, by basal ISG expression. Specifically, PCMCs may rely on high basal ISG levels, thereby eliminating the need for extensive induction of ISGs, while in contrast, PCFCs express low basal ISG levels but achieve protective ISG levels through rapid and substantial IFN-mediated induction. These data suggest that PCFCs may rely more than PCMCs on the IFN-mediated response for antiviral protection.

To address this possibility, PCMCs and PCFCs were infected with a large panel of reoviruses in the presence of control or anti-IFN antibodies. After incubation to allow viral replication and spread, cultures were harvested and viral titers were determined. In the presence of the control antibody, reovirus replicated to similar titers in PCMCs and PCFCs (Fig. 6A) and, as previously observed, myocarditic viruses replicated to higher titers than nonmyocarditic viruses in cardiac cells (52). However, in the presence of an anti-IFN antibody, reovirus replicated to significantly higher titers in PCFCs than in PCMCs. Specifically, the anti-IFN antibody increased viral replication an average of 50-fold in PCMCs but 550-fold in PCFCs (Fig. 6B) (P < 0.001). These results demonstrate that induced IFN provides greater antiviral protection in PCFCs than in PCMCs. The viral genes that determined the degree of IFN-mediated protection in PCFCs (data not shown) were the same as those previously identified for PCMCs (52). Given that viruses replicated equivalently in PCMCs and PCFCs in control cultures but that in the absence of IFN, viruses replicated to titers 10-fold higher in PCFCs than in PCMCs, the data suggest that constitutively higher ISG levels in PCMCs may provide a degree of protection even in the absence of IFN.

FIG. 6.

PCFCs are more dependent on the IFN response than PCMCs. Replicate wells of PCMCs and PCFCs were infected at a multiplicity of infection of 0.1 and overlaid with an anti-IFN-α/β antibody or a control antibody. At 7 days postinfection, cultures were lysed and viral titers were determined by a plaque assay. (A) Results expressed as mean viral yield of duplicate wells ± standard deviation. (B) Results from PCMCs and PCFCs expressed as ratio of viral yield in anti-IFN-α/β-treated wells relative to that in control-treated wells.

Viruses can directly induce ISGs such as 561 (15, 59), and the mechanisms may be cell type specific. That is, it is possible that the residual protection in the absence of IFN in PCMCs relative to PCFCs reflected greater direct induction of ISGs in PCMCs than in PCFCs. Accordingly, we tested whether reovirus can directly induce expression of 561 in PCMCs and PCFCs derived from IFN-α/β receptor-null (IFN-R-null) mice. PCMCs and PCFCs were derived from IFN-R-null mice and were infected with reovirus T3D. Cultures were harvested at 0 and 8 h postinfection. By real-time PCR analysis, reovirus directly induced 561 expression in both PCMCs and PCFCs, with no significant difference between cell types (Fig. 7) (P = 0.11). Additionally, reovirus T3D did not induce IRF-7 expression in PCMCs derived from IFN-R-null mice (data not shown). Therefore, differences in viral replication between PCFCs and PCMCs in the absence of IFN-α/β are not associated with direct viral induction of 561 or IRF-7.

FIG. 7.

Reovirus directly induces 561 expression to similar extents in PCMCs and PCFCs. PCMCs and PCFCs were derived from IFN-R-null mice. Cultures were infected with reovirus T3D at a multiplicity of infection of 10 and were harvested at 0 and 8 h postinfection. Total RNA was extracted and reverse transcribed. Resultant cDNAs were analyzed by real-time PCR. Each bar is the average of 10 replicates from five experiments with PCMCs, or of 6 replicates from three experiments with PCFCs, ± the standard error of the mean. Asterisk indicates a significant difference in 561 expression between 0 and 8 h postinfection in a given cell type.

What determines the greater basal ISG expression in PCMCs relative to PCFCs? PCMCs express higher basal IFN-β levels than do PCFCs (Fig. 1A), providing a possible explanation for differences between PCMCs and PCFCs in IRF-7 expression (Fig. 2A) and 561 expression (Fig. 3A). Indeed, IFN-R-null PCMCs and PCFCs express dramatically lower basal levels of 561 (Fig. 8A) and IRF-7 (Fig. 8B) than wild-type Cr:NIH(S) cultures (similar results were obtained for mock-treated parental 129Sv/Ev PCMCs and PCFCs [data not shown]). Together, these data strongly suggest that basal expression of IFN is a primary determinant of basal ISG expression and that basal ISG expression is an important determinant of antiviral protection in PCMCs but not PCFCs.

FIG. 8.

Basal IFN expression is a determinant of basal ISG expression. (A) Basal 561 expression from Fig. 3A (wild-type cultures) compared to 561 expression at 0 h from Fig. 7 (IFN-R-null cultures). (B) Basal IRF-7 expression from Fig. 2A (wild-type cultures) compared to IRF-7 expression from IFN-R-null-derived cultures at 0 h. Asterisk indicates a significant difference in 561 (A) or IRF-7 (B) expression between wild-type and IFN-R-null cultures.

DISCUSSION

Cardiac myocytes are not replenished yet are critical for survival. Given that many viruses gain access to the heart, these highly specialized cells may have a unique response to viral infection. Specifically, cardiac myocytes may be more dependent on the innate immune response than cell types that are easily replenished. Indeed, we have previously shown that the innate IFN response is critical in determining viral spread in PCMCs but not in differentiated skeletal muscle cells (52).

In this report, we demonstrate that PCMCs express greater levels of basal IFN-β than PCFCs (Fig. 1A). In addition, reovirus induces greater IFN-β expression in PCMCs than in PCFCs, as indicated by both copy number and relative fold induction (Fig. 1A and B). Cells “primed” with small amounts of IFN-α/β and then virally infected express more IFN-α/β than nonprimed cells (44). This priming is IRF-7 dependent: following viral infection, spleen cells lacking IRF-3, but primed by IFN to express IRF-7, induce IFN-α/β mRNA to levels similar to those of wild-type spleen cells (57). Given that basal IRF-7 expression was significantly higher in PCMCs than in PCFCs (12-fold), our data support a mechanism whereby relatively high basal IFN-β expression in PCMCs results in similarly high basal IRF-7 expression, which “primes” PCMCs to induce greater IFN-β expression in response to viral infection. This would also be in agreement with previous work suggesting that higher basal IRF-7 expression in plasmacytoid dendritic cells (PDCs) compared to monocyte-derived dendritic cells (MDDCs) results in Sendai virus inducing greater expression and a broader set of IFN-α genes in PDCs than in MDDCs (24). Interestingly, basal expression of IRF-7 is higher in lymphoid than in nonlymphoid cells (1, 63). While this likely reflects the prominent role of IRF-7 in the induction of distinct IFN-α genes (28), the higher basal expression of IRF-7 in PCMCs than in PCFCs could reflect a “prearming” of PCMCs, thereby allowing these cells to respond more quickly to viral infection. This could be critical for effective protection of these differentiated, nondividing, essential cells. Surprisingly, even though reovirus induced more IFN-β in PCMCs than PCFCs (Fig. 1B), reovirus induction of IRF-7 expression (copy number) was similar at all time points (Fig. 2B). These data suggest that there may be an upper limit to IRF-7 expression in cardiac cells and that cell types with higher basal expression (PCMCs) are less responsive to viral infection and/or IFN-α/β stimulation.

As with IRF-7, peak expression of the ISG 561 was comparable in PCMCs and PCFCs (Fig. 3B), but unlike IRF-7, the kinetics of viral induction of 561 were cell type specific: induction of 561 was prolonged in PCMCs but transient in PCFCs. Interestingly, following IFN treatment, the kinetics and levels of 561 expression were identical for the two cell types (Fig. 5A), resembling that in PCFCs following reovirus infection and resembling transient induction of 561 mRNA by double-stranded RNA in human glioblastoma GRE cells (2). One possibility could be that cell type-specific differences in the kinetics of induction of 561 reflect differences in virally activated IRF-3. That is, IRF-3 could remain activated longer, or could be more efficient at induction of 561 expression, in PCMCs than PCFCs. However, our data demonstrating similar reovirus-induced direct induction of 561 in PCMCs and PCFCs argue against a role for IRF-3 (see below) (Fig. 8A). Alternatively, it is possible that differences in the kinetics of 561 expression reflect cell type-specific differences in the kinetics of reovirus-induced IRF-1 expression. Indeed, in human fibroblasts, the pattern of peak expression and then decline of the ISG 6-16 coincides with the kinetics of IFN-induced expression of IRF-1 but not ISGF3 (23, 53). Finally, it is possible that differences in accumulated levels of 561 mRNA, as measured by real-time reverse transcription-PCR, reflect stimulus-specific or cell type-specific differences in mRNA degradation rates (61). While full gene sequences are not available for comparisons between the 561 and IRF-7 untranslated regions to identify possible differences in AU-rich elements, future studies can address possible differences in cis-regulated mRNA degradation.

Basal levels of IRF-7 and 561 expression were higher in PCMCs than in PCFCs (Fig. 2A and 3A). Importantly, our data indicate that basal IFN expression is a primary determinant of basal ISG expression (Fig. 8). The greater decrease in IRF-7 expression (426- and 20-fold in PCMCs and PCFCs, respectively [Fig. 8B]) than in 561 expression (58- and 16-fold in PCMCs and PCFCs, respectively [Fig. 8A]) in IFN-R-null cultures likely reflects IRF-7's greater dependence on ISGF3-mediated transcription (data not shown) (45); in contrast, 561 can also be induced by IRFs (2, 7, 10, 14, 15, 40) and thus is likely less affected by loss of ISGF3. The greater decrease in ISG expression in PCMCs than in PCFCs likely reflects the higher basal expression levels of ISGs (IRF-7 and 561) in PCMCs than in PCFCs.

Interestingly, our results in cultures derived from IFN-R-null mice demonstrate that like vesicular stomatitis virus, encephalomyocarditis virus, and Sendai virus (15), reovirus T3D can directly induce expression of 561 (Fig. 7). Moreover, these cultures display no statistical difference in 561 expression levels at 8 h postinfection, indicating that in the absence of JAK/STAT signaling, reovirus induces 561 expression to similar extents in PCMCs and PCFCs. These results contrast with a report indicating that reovirus does not directly induce the expression of p56 (encoded by 561) in GRE cells (15). Both laboratories used the Dearing strain of T3 reovirus, but this does not exclude the possibility of substrain-specific differences. Alternatively, direct induction of ISGs by reovirus may be cell type specific, occurring in cardiac cells but not in human glioblastoma cells.

Viruses replicated to higher titers in PCFCs than in PCMCs in the absence of induced IFN-β (Fig. 6). Together with data demonstrating higher basal ISG expression in PCMCs (Fig. 2 to 5), this finding suggests that PCMCs benefit from a “prearming” while PCFCs are more dependent on induced expression of IFN-β and ISGs. Others have shown that mutant cell lines expressing high basal levels of IFN also express high basal levels of the ISGs 6-16, 9-27, and 2′-5′ oligoadenylate synthetase (35). Our results are also in agreement with observations that in the absence of IFN-mediated protection, cells stably transfected with an IRF-1 transgene are significantly more resistant to vesicular stomatitis virus, encephalomyocarditis virus, and Newcastle disease virus than wild-type cells (41). Importantly, because prearming requires higher levels of basal IFN, and IFN is antiproliferative, it is likely that this mechanism of protection is restricted to nondividing cell types such as cardiac myocytes.

Finally, our results demonstrate that relative induction of gene expression can be misleading, in that final absolute expression (copy number) may differ minimally while relative induction differs drastically due to differences in basal expression. That is, reovirus or IFN-β always induced equivalent or more copies of 561 and IRF-7 in PCMCs than in PCFCs (Fig. 2B, 3B, 4A, and 5A), but relative induction was always higher in PCFCs than in PCMCs (Fig. 2C, 3C, 4B, and 5B). These differences were often dramatic: for example, while IFN induced equivalent IRF-7 copies in PCMCs and PCFCs at 5 h posttreatment (Fig. 4A), there was a 10-fold-greater relative induction in PCFCs than in PCMCs (47-fold induction in PCFCs versus 4-fold induction in PCMCs [Fig. 4B]). This discrepancy was also apparent when 561 expression in wild-type and IFN-R-null-derived cultures was compared. At 8 h postinfection in PCMCs derived from IFN-R-null mice, there was greater relative induction of 561 (26-fold [Fig. 7]) than at 8 h in wild-type cultures (22-fold [Fig. 3C]), yet wild-type cultures had dramatically higher (47-fold) absolute expression of 561 (0.47 versus 0.01 GAPDH-normalized copies). Even at 8 h postinfection, IFN-R-null-derived PCMCs still expressed less 561 (0.01 GAPDH-normalized copies) than wild-type cultures expressed basally (0.02 GAPDH-normalized copies). This indicates that relative induction data such as those obtained from reporter gene and microarray analyses should be interpreted carefully when one is comparing multiple cell types (including null and wild type). Moreover, these data further demonstrate that basal expression can be a principal determinant of induced expression.

Future experiments will continue to identify the molecular differences between PCMCs and PCFCs following viral infection and/or IFN-β treatment. Of particular interest will be the STAT-signaling cascade. Given that relative induction of ISGs is higher in PCFCs than in PCMCs, one could imagine a more robust and/or sustained STAT activation in PCFCs.