|Home | About | Journals | Submit | Contact Us | Français|
Herpes B virus (BV) naturally infects macaque monkeys and is genetically similar to herpes simplex virus (HSV). Zoonotic infection of humans can cause encephalitis and if untreated has a fatality rate of ~80%. The frequent use of macaques in biomedical research emphasizes the need to understand the molecular basis of BV pathogenesis with a view toward improving safety for those working with macaques. MicroRNAs (miRNAs) are small noncoding RNAs that regulate the expression of mRNAs bearing complementary target sequences and are employed by viruses to control viral and host gene expression. Using deep sequencing and validation by expression in transfected cells, we identified 12 novel BV-encoded miRNAs expressed in lytically infected cells and 4 in latently infected trigeminal ganglia (TG). Using quantitative reverse transcription-PCR (RT-qPCR), we found that most of the miRNAs exhibited a high level of abundance throughout infection. Further analyses showed that some miRNAs could be generated from multiple transcripts with different kinetic classes, possibly explaining detection throughout infection. Interestingly, miRNAs were detected at early times in the absence of viral gene expression and were present in purified virions. In TG, despite similar amounts of viral DNA per ganglion, it was notable that the relative amount of each miRNA varied between ganglia. The majority of the miRNAs are encoded by the regions that exhibit the most sequence differences between BV and HSV. Additionally, there is no sequence conservation between BV- and HSV-encoded miRNAs, which may be important for the differences in the human diseases caused by BV and HSV.
Viruses have evolved numerous mechanisms to regulate the expression of their genome in an optimal manner. These include transcriptional and posttranscriptional regulatory mechanisms. Of interest is the capacity of viruses to encode microRNAs (miRNAs), which are small noncoding RNA molecules that modulate gene expression posttranscriptionally. Currently, most reports of virus-encoded miRNAs are associated with herpesviruses. The family of these viruses includes large enveloped viruses with double-stranded DNA genomes that cause diseases of great medical and veterinary importance, for which there are limited treatments and no cures. A hallmark of herpesviruses is their ability to establish and maintain latent infections, during which there is limited virus gene expression. Small RNA molecules are nonantigenic and thus appear ideally suited to be exploited during latency. Herpes simplex viruses 1 and 2 (HSV-1 and -2) are members of the alphaherpesvirus subfamily and are characterized by their ability to establish a latent infection in neurons, typically in the trigeminal ganglia (TG) for HSV-1 or dorsal root ganglia for HSV-2 (28). To date, 16 HSV-1-encoded miRNAs and 17 HSV-2-encoded miRNAs have been identified and shown to be expressed during the lytic and latent phases of the virus life cycle (14, 22, 35, 36, 38).
A close relative of HSV, herpes B virus (BV; Macacine herpesvirus 1; herpes B) is an alphaherpesvirus that is enzootic in macaque monkeys (genus Macaca) (44). Infection in its natural host usually results in only mild localized self-limiting or asymptomatic infections (44). However, BV can zoonotically infect humans and is associated with an extremely high mortality rate (20). Left untreated, BV typically causes encephalitis and paralysis, resulting in ~80% lethality (44). Even with timely antiviral therapy, BV infection results in ~20% lethality. Many survivors suffer severe neurological complications and need to be maintained on high doses of antiviral therapy for life.
Several macaque species, particularly rhesus (Macaca mulatta) and cynomolgus (Macaca fascicularis) monkeys, are commonly used in biomedical research. According to the USDA Animal Care Annual Report of Activities for 2007 (41), approximately 70,000 nonhuman primates were used in research and the majority of those are likely to have been rhesus macaques. BV is present in most macaque colonies. Thus, animal care staff are at high risk for BV infection. Moreover, despite great financial investment to generate BV-free colonies, all macaque monkeys need to be treated as potentially infected. Therefore, from a safety and financial viewpoint, BV is a great impediment for those working with macaques in the research setting. Due to the highly pathogenic nature of BV, the Centers for Disease Control and Prevention (CDC) has classified BV as a Risk Group 4 agent that may only be propagated in a maximum containment laboratory (Biosafety Level 4 [BSL-4]). Additionally, the U.S. Department for Health and Human Services has designated the virus and viral DNA as select agents.
There is mounting evidence supporting the notion that herpesvirus-encoded miRNAs are important for virus pathogenesis. Some important regulatory functions of herpesvirus-encoded miRNAs are thought to include control of latent/lytic cycle entry and exit and immune evasion, in addition to cell survival and proliferation (7). We have reported that BV encodes three miRNAs using computational predictions and validation by Northern blot hybridization (6). Based on studies of HSV and other alphaherpesviruses, we hypothesized that BV encodes more miRNAs than originally detected and chose to use ultra-deep sequencing technology to identify additional BV-encoded miRNAs. We examined RNA from both naturally infected TG harvested from macaque monkeys and lytically infected cells. We reasoned that looking throughout the course of BV infection would provide the maximum opportunity for complete identification of BV-encoded miRNAs.
African green monkey (Vero) cells and HeLa cells were obtained from the American Type Culture Collection and maintained in Dulbecco's modified Eagle's medium, supplemented with 10% fetal bovine serum, at 37°C and 5% CO2. Cynomolgus macaque TG were obtained at necropsy from the Southwest National Primate Research Center (SNPRC), immediately frozen in liquid nitrogen, and stored at −80°C in BSL-3 containment until processed. This occurs as rapidly as 20 min following euthanasia. BV strain E2490 (NCBI reference sequence no. NC_004812) was used for the in vitro experiments in this study.
BV is classified as a Risk Group 4 agent (CDC) and may only be propagated in a maximum containment laboratory. All experiments using infectious virus were performed in the Animal Biosafety Level 4 (ABSL-4) laboratory at the Texas Biomedical Research Institute (certified by the CDC) by trained personnel wearing protective biosafety suits.
cDNA libraries for deep sequencing were prepared using the small RNA sample prep kit v1.5.0 protocol from Illumina. Between 100 ng and 500 ng small enriched RNA was ligated to adapters, followed by reverse transcription (RT) and PCR amplification. The amplified cDNA libraries were resolved using a 6% polyacrylamide Tris-borate-EDTA (TBE) gel. cDNAs that corresponded to approximately 18 to 30 nucleotides (nt) were identified by comparison to an SRA ladder (Illumina) and excised from the gel. The cDNA was eluted from the gel for 4 h with 100 μl 1× gel elution buffer (Illumina). Gel debris was removed using Spin-X cellulose acetate filters (Illumina), and the cDNA was precipitated with 1 μl glycogen (Ambion), 3 M sodium acetate (Ambion), and 325 μl chilled 100% ethanol. After being washed with 70% ethanol, the cDNA pellet was resuspended in resuspension buffer (Illumina).
Raw sequence data were collapsed into a single file of nonredundant sequences with an associated abundance. Adapter sequences were removed, and the file was converted to FASTA format. The analysis was performed with Formatdb, Megablast, filter_alignments, excise_candidate, auto_blast, and miRDeep scripts from the miRDeep package (15), together with the BV genomic sequence (NC_004812) and miRBase database (release 14). These algorithms consider pre-miRNA structure, the relative frequency at which a mature miRNA sequence and star sequence were observed, and likelihood that the mature miRNA was generated through a Dicer-dependent mechanism. However, for the very rare reads that are sometimes observed for alphaherpesvirus miRNAs, considering the relative frequency of mature sequences to star sequences can be too stringent. Therefore, reads that failed based on this criterion were further analyzed manually: a candidate sequence was considered a potential miRNA when the sequence was present in a hairpin that was predicted as the lowest free-energy structure using mfold.
To generate plasmid pAMBVgB, the BV virion membrane glycoprotein B (gB; UL27) coding sequence was amplified from BV strain E2490 DNA by PCR and cloned into pCR2.1 TOPO using the TOPO TA cloning system (Invitrogen). The PCR primers were as follows: BV glycoprotein B forward primer (5′-TCCGGGCCCGGAATGCGGCCCCGCGCC-3′) (IDT) and reverse primer (5′-GCGGCCGCCTATAGCTCCTCTTCGTCTGTGTCGCCG-3′) (IDT).
A portion of the second exon of immediate early protein ICP0 coding sequence (between bases 122943 and 123375) was synthesized as a “minigene” by IDT using the published BV sequence, yielding pIDTSMART-KAN:ICP0. The minigene was digested with HindIII and XhoI (NEB) to release the ICP0 sequence insert, which was then inserted into pCR2.1 TOPO using the TOPO TA cloning system (Invitrogen) to yield pMAICP0.
To generate miRNA expression plasmids, each miRNA sequence together with ~200 nt of flanking sequence was amplified from BV strain E2490 DNA by PCR and first subcloned into pCR2.1TOPO or pCR-Blunt II TOPO using the TOPO cloning system (Invitrogen). The PCR primers, along with the TOPO subcloning vectors and enzymes, are listed in Table S1 in the supplemental material. To generate a cassette-type vector that can accommodate cloned inserts from pCR2.1 TOPO or pCR-Blunt II TOPO, regardless of initial insert orientation, a multiple cloning site was cloned into pSIREN-Retro-Q-ZsGreen1 (Clontech), using a duplex of synthetic oligonucleotides. The oligonucleotides used were as follows: 5′-GATCCGTTAACGCGGCCGCGGGCCCCCAGTGTGCTGGTCTAGAT-3′ and 5′-AATTATCTAGACCAGCACACTGGGGGCCCGCGGCCGCGTTAACG-3′ (BamHI, HpaI, NotI, ApaI, BstXI, XbaI). The recognition sequence for EcoRI was destroyed to serve as a diagnostic tool. Oligonucleotides were annealed, and the resulting duplex was ligated to BamHI- and EcoRI-digested pSIREN-Retro-Q-ZsGreen1 vector to yield pSIREN-MCS. miRNA sequences were excised from the TOPO vectors and cloned into pSIREN-MCS using the appropriate restriction sites for correct orientation of the insert.
All constructs generated by oligonucleotide cloning or PCR were confirmed to be correct by sequencing at the Nucleic Acids Core Facility at the University of Texas Health Science Center, San Antonio (UTHSCSA).
Candidate miRNA sequence-containing plasmids were transfected into HeLa cells using the Neon transfection system (Invitrogen) according to the manufacturer's recommendations. To confirm that the candidate miRNAs were synthesized in a Drosha-dependent manner, 44 nM Drosha siRNA (RNASEN siRNA identification no. s26492; Ambion) was also cotransfected; this amount of siRNA had been determined to reduce Drosha expression by approximately 90%. A second siRNA (Silencer negative control 2 siRNA; Ambion) was used as a negative control. Briefly, 2 × 107 HeLa cells were resuspended in 100 μl of buffer R (Invitrogen), containing 8 μg plasmid and 44 nM Drosha siRNA or negative control siRNA. The electroporation was performed using 2 pulses at 1,400 V for 20 ms. After electroporation, cells were plated on 60-mm dishes in 10% fetal calf serum (FCS) without antibiotics. Forty-eight hours later, total RNA was harvested using the mirVana miRNA isolation kit (Ambion) according to the manufacturer's instructions.
To harvest RNA from infected cells, 60-mm dishes of 80% confluent Vero cells were infected with BV at a multiplicity of infection (MOI) of 10. At various times postinfection, cells were scraped from the dishes and pelleted in a low-speed centrifuge; the supernatant was removed and the pellet was flash-frozen in liquid nitrogen. Dry pellets were stored at −80°C until processed. Both large (>200 nt) and small (<200 nt) RNA were harvested from infected cells using the mirVana miRNA isolation kit (Ambion) according to the manufacturer's instructions.
Nucleic acid isolation from macaque TG began in the BSL-3 laboratory. Lysis/binding solution (600 μl; from the mirVana kit) was added to each ganglion, and the tissues were homogenized on ice using an RNase-free plastic pestle and tube. The tissues were processed immediately without allowing the tissue to thaw, thereby preventing the rupture of cells by ice crystals and the subsequent release of RNases. miRNA homogenate additive (60 μl; from the mirVana kit) was added to the homogenized tissue, and the samples were incubated on ice for 10 min. Fifty microliters of the homogenized tissue was then set aside for DNA isolation, and the remainder was used for RNA isolation. An equal volume of acid phenol-chloroform (Ambion) was added to the homogenate set aside for RNA isolation, and the samples were vortexed thoroughly. Likewise, an equal volume of phenol-chloroform (Ambion) was added to the homogenate set aside for DNA isolation, and the samples were vortexed thoroughly. Samples were then transferred to the BSL-2 laboratory, where the remainder of the RNA isolation was performed using the mirVana miRNA isolation kit according to the manufacturer's instructions. RNA concentration was determined by spectrophotometry.
In the BSL-2 laboratory, the samples set aside for DNA isolation were centrifuged and the aqueous layer was removed. DNA was precipitated by adding one volume of isopropanol (Sigma-Aldrich) and centrifuged for 1 h at 4°C and 14,000 × g. The supernatant was removed, and the DNA pellet was washed with 70% ethanol, followed by resuspension in sterile water and spectrophotometry to determine concentration.
Various pharmacological agents were used to investigate whether miRNA accumulation was affected by blocking expression of genes from particular kinetic classes. To inhibit BV DNA synthesis, which prevents expression of true late-class genes, cells were pretreated with 300 μg/ml phosphonoacetic acid (PAA) (Sigma-Aldrich) for 30 min before cells were infected with BV at an MOI of 10 in the presence of the same concentration of the drug. At 18 h postinfection (hpi), cells were processed as described above. To confirm PAA activity, RNA was harvested from infected cells and subjected to Northern blot hybridization probed for the true late-class viral gene UL44 (see Fig. 7B).
To inhibit expression of all but immediate early class genes, Vero cells were pretreated with 100 μg/ml of the protein synthesis inhibitor cycloheximide (CHX) (Sigma-Aldrich) for 30 min before cells were infected with BV at an MOI of 10 in the presence of the same concentration of drug. At 1 hpi, cells were scraped from the dishes and pelleted in a low-speed centrifuge, the supernatant was removed, and the pellet was flash-frozen in liquid nitrogen and stored at −80°C until processed. To confirm CHX activity, uninfected Vero cells were treated with and without CHX in the presence of 0.1 mCi/ml of [35S]methionine (PerkinElmer) in parallel to the infected cells. After 1 h, the cell lysates were resolved by SDS-PAGE and processed for autoradiography (see Fig. 7C).
To inhibit both viral and cellular transcription, actinomycin D (act D; MP Biomedical) was used similarly to CHX, but at a concentration of 2 μg/ml. To confirm act D activity, RNA was harvested from infected cells and subjected to Northern blot hybridization probed for the immediate early class viral gene ICP4 (see Fig. 7D).
Virions were purified by centrifugation through a 20% sorbitol cushion as described by Stinski (34). Briefly, Vero cells were grown to 80% confluence in 150-cm2 flasks and infected at an MOI of 10. The cells were harvested 24 hpi and centrifuged at 2,000 rpm for 5 min. The cell pellet was lysed in reticulocyte standard buffer (RSB) (10 mM Tris, pH 7.5, 10 mM KCl, 1.5 mM MgCl2) containing 0.5% (vol/vol) NP-40. The lysate was incubated on ice for 10 min, and the nuclei was pelleted. Viral particles in the supernatant were recovered by centrifugation at 50,000 rpm for 1 h at room temperature in a SW 55 Ti rotor through 20% (wt/vol) D-sorbitol (Sigma-Aldrich) in 50 mM Tris-HCl, pH 7.5, 1 mM MgCl2, and 100 μg/ml bacitracin (Sigma-Aldrich). The pellet was resuspended in RNase ONE buffer (Promega) containing 9.15 × 107 molecules of synthetic ath-miR-166 miRNA standard (IDT) and 50 U RNase ONE (Promega). ath-miR-166 is a plant miRNA that is absent from mammalian cells (26), including Vero cells (M. A. Amen and A. Griffiths, unpublished observations) and was used to determine the efficiency of RNase digestion of miRNAs. The reaction mixture was incubated at 37°C for 30 min. To boost digestion, an additional 50 U RNase ONE (Promega) was added and the reaction mixture was incubated for a further 30 min at 37°C. To dissociate aggregates, 5 mM urea (Fisher) was added and the mixture was sonicated for 10 to 15 s. An equal volume of phenol-chloroform (Ambion) was added and vortexed thoroughly before the sample was transferred to the BSL-2 laboratory. Next, the aqueous phase was harvested and half was devoted to RNA isolation and half to DNA isolation. RNA isolation was performed using the mirVana miRNA isolation kit (Ambion) according to the manufacturer's instructions, and the final RNA concentration was determined by spectrophotometry. DNA was precipitated by adding one volume of isopropanol (Sigma-Aldrich) and centrifuged for 1 h at 4°C and 14,000 × g. The supernatant was removed, and the DNA pellet was washed with 70% ethanol, followed by resuspension in sterile water and spectrophotometry to determine concentration.
To determine if macaque TG contained BV DNA, we used a previously reported quantitative PCR (qPCR) assay (21). The primers and probe were designed to detect the conserved glycoprotein B (gB) sequence of BV and differentiate between BV and other alphaherpesviruses such as HSV. A standard curve was generated using DNA isolated from uninfected Vero cells spiked with a known amount of pAMBVgB.
To account for the possibility that an infected TG was supporting lytic cycle replication at the time of necropsy, a reverse transcription-qPCR (RT-qPCR) assay was developed to detect expression of the immediate early ICP0 gene. Large RNA (5 ng/μl) was subjected to reverse transcription using 500 nM ICP0 reverse transcription primer (5′-AACTGGTGCCCCTACCAC-3′) (IDT), 5× iScript select reaction mix (Bio-Rad), GSP enhancer solution (Bio-Rad), and iScript reverse transcriptase (Bio-Rad) in a total volume of 20 μl. RTs were performed at 42°C for 60 min and 85°C for 5 min. One microliter of the resulting cDNA was added to 1× iQ SYBR green Supermix (Bio-Rad), 500 nM ICP0 forward primer (5′-CAGACGTGCCTCGCGTA-3′) (IDT), and 500 nM ICP0 reverse primer (5′-TCGACAACGCGTACCCG-3′) (IDT) in a total volume of 25 μl. A standard curve was generated using large RNA isolated from uninfected Vero cells spiked with a known amount of ICP0 RNA in vitro transcribed using an mMESSAGE mMACHINE kit (Ambion) from pMAICP0. Reverse transcription and qPCR were performed using the iScript select cDNA synthesis kit (Bio-Rad) and iQ SYBR green Supermix (Bio-Rad), respectively, according to the manufacturers' instructions. To determine if there was equivalent RNA recovery between samples of large RNA, a previously reported RT-qPCR assay was utilized to detect RPL13a, a cellular marker shown to be effective for use with rhesus macaque tissues (1). The sequence of RPL13a for cynomolgus macaque has several base differences from the rhesus macaque sequence. To account for these differences, the following forward primer was used: 5′-CTTGGAGGAGAAGAGGAAGGAGA-3′ (IDT).
For analyses of relative expression of miRNAs, stem-loop RT-qPCR using a single RT reaction mixture containing multiple RT primers was used as previously described (13). Stem-loop RT-qPCR of miRNAs has been shown to be highly specific for a particular miRNA, even able to differentiate between miRNAs of different sizes and miRNAs that have single-base substitutions in an miRNA sequence (11). Briefly, 250 ng small RNA was reverse transcribed with a final concentration of 12.5 nM of each stem-loop RT primer (IDT) (see Table S2 in the supplemental material), 1× RT buffer (Applied Biosystems), 0.25 mM (each) deoxynucleoside triphosphates (dNTPs) (Applied Biosystems), 20 U/μl RNase inhibitor (Applied Biosystems), and 3.33 U/μl MultiScribe reverse transcriptase (Applied Biosystems). The volume of the RT reaction mixture was adjusted according to the number of PCRs to be performed. The RT reactions were performed at 16°C for 30 min, followed by 42°C for 30 min and 85°C for 5 min. The efficiencies of two reactions (hbv-mir-B3RC-3P and hbv-mir-B14RC-3P) were improved by increasing the RT temperature. These RTs were performed at 19°C for 30 min, followed by 48°C for 30 min and 85°C for 5 min. For real-time PCR, a 20-μl reaction mixture was made containing 1.33 μl reverse transcription reaction, 1× TaqMan universal PCR master mix (Applied Biosystems), 0.2 μM TaqMan probe, 1.5 μM forward primer, and 0.7 μM reverse primer (5′-AGTGCAGGGTCCGAGGTA-3′) (IDT). The reactions were performed in triplicate and incubated in a 96-well plate at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Detection of the most commonly sequenced isoform of each miRNA was performed using forward primers (IDT) and TaqMan probes (Applied Biosystems). The primer and probe sequences are listed in Table S2 in the supplemental material.
To confirm that Drosha mRNA was knocked down in the presence of the Drosha siRNA, a Drosha qPCR assay was used (Applied Biosystems, Hs00294820_m1). An assay to detect 18S rRNA (Hs03928990_g1; Applied Biosystems) was used as a control. Briefly, 100 ng/μl total RNA was DNase treated using a Turbo DNA-free kit (Ambion) according to manufacturer's instructions. The DNase-treated samples were reverse transcribed with oligo(dT)20 primers using the SuperScript III first-strand synthesis system (Invitrogen) according to the manufacturer's instructions. For the real-time PCR, a 20-μl reaction mixture was made containing 1 μl Drosha or 18S TaqMan assay mix, 10 μl TaqMan universal PCR master mix, no AmpErase uracil N-glycosylase (UNG) (Applied Biosystems), and 1 μl cDNA. The reactions were performed in triplicate and incubated in a 96-well plate at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min.
To estimate the number of molecules of each miRNA packaged in virions, standard curves were generated using small RNA isolated from uninfected Vero cells spiked with a known amount of synthetic RNA oligonucleotides (IDT) representing each miRNA sequence. Eighty-five nanograms of small RNA isolated from purified virions treated with and without RNase was used for RT-qPCR of the miRNAs as described above. To estimate the number of viral genomes in each sample, 293 ng DNA isolated from purified virions treated with and without RNase was used for the qPCR of gB as described above.
To determine the changes in levels of individual miRNAs between multiple samples (time course and multiple TG), we used a relative quantification strategy. To determine if the cellular miRNA Let-7a was an appropriate normalization factor for these assays, we measured its expression during infection. Using absolute quantification we determined that it was expressed at constant levels between samples harvested at various times postinfection using RT-qPCR (see Fig. S1 in the supplemental material). This confirmed our previous observation, using Northern blot hybridization that Let-7a expression was unaffected by virus infection (6). Briefly, a standard curve was generated using a synthetic RNA oligonucleotide (IDT) that represents the 22-nt form of Let-7a. Reverse transcription and qPCR was performed as described above (see Table S2 in the supplemental material).
RNA species of >200 nt isolated from infected cell lysates (1 hpi and 18 hpi) were resolved using a glyoxal-agarose gel and transferred to a nylon membrane using the NorthernMax-Gly kit (Ambion) according to the manufacturer's instructions. The RNA was cross-linked to the membrane using a Stratalinker (Stratagene). The membranes were probed with DNA oligonucleotides (IDT) end-labeled with [32P]ATP (MP Biomedicals, Irvine, CA). Unincorporated nucleotides were removed using micro Bio-Spin columns, (Bio-Rad) according to the manufacturer's instructions. Only probes with specific activities exceeding 5 × 106 dpm/pmol, as measured using a liquid scintillation counter, were used. The sequences of the probes are detailed in Table S3 in the supplemental material. Hybridizations were performed in ULTRAhyb-Oligo buffer (Ambion) according to the manufacturer's instructions. Probed membranes were exposed to phosphor storage screens (GE Life Sciences) and detected on a Storm phosphorimager (GE Life Sciences).
BV establishes a latent infection in the TG of macaques (reviewed in reference 44). To identify BV-encoded miRNAs expressed during latent infection, we harvested macaque TG at necropsy. To determine if a particular TG to be tested contained BV DNA, we employed a real-time PCR assay previously developed to measure BV DNA in macaque TG (20). The levels of viral DNA varied only slightly between all four TG (Fig. 1), estimated to range between 1.4 × 104 and 2.4 × 105 molecules per TG. It was crucial to discount the possibility that the virus had reactivated in these TG, as we wished to identify miRNAs expressed during latency. To address this, we developed an RT-qPCR assay to detect ICP0 mRNA, an immediate early transcript encoding a protein required for HSV reactivation from latency (10). The limit of detection for these assays was determined to be 1.9 × 102 molecules per TG. No ICP0 transcripts were detected in any tested TG. Furthermore, using RNA isolated from lytically infected Vero cells, we determined that 1.7 × 103 molecules of ICP0 were present per infected cell at 3 hpi. Together, these data are consistent with absence of lytic replication in all reported TG, thereby supporting the notion that the virus was present in the latent state.
cDNAs corresponding to the approximate size of miRNAs were subjected to deep sequencing. At 3 hpi, 12,595,747 total reads and 3,414,075 unique reads were returned; at 9, 15, and 21 hpi and in the TG the numbers were 7,289,829 and 1,209,685, 8,303,409 and 1,409,962, 6,679,109 and 1,246,758, and 7,904,359 and 1,907,690, respectively. Similarly to results in a report describing HSV miRNAs expressed in human TG (39), the most frequently observed miRNAs in monkey TG were members of the Let-7 family. The most frequently observed miRNA in all lytically infected Vero cells was miR-21.
A sequence was considered an miRNA candidate when (i) the sequence was within a hairpin-like RNA secondary structure and (ii) the predicted miRNA precursor structure and the position of the small RNA sequence met the requirements of a genuine pre-miRNA as described by Ambros et al. (2). Twenty-eight sequences fulfilled these criteria and were further tested using stem-loop RT-qPCR assays to detect the candidate miRNAs in infected cells. Of those 28 candidate sequences, 10 were not amplified, and we concluded that they were not miRNAs. To determine if the remaining candidate sequences were miRNAs or products of RNA degradation, the remaining 18 candidate sequences were then cloned into expression vectors and transfected into Vero cells. As an additional method to test if a candidate sequence is an miRNA, we asked if the transfected sequence was processed in a Drosha-dependent manner. Drosha is an RNase III enzyme that is responsible for cleaving the primary miRNA (pri-miRNA) into the characteristic stem-loop precursor miRNA in the nucleus and is necessary for miRNA biogenesis (23). To this end, an siRNA targeting Drosha (or a control siRNA) was cotransfected with each miRNA expression plasmid. Small RNAs that are not miRNAs will be unaffected by the reduction of Drosha. Only genuine miRNAs will be affected by the reduction of Drosha. In the presence of the Drosha siRNA, Drosha mRNA was shown to be present at approximately 15% of the amount compared to when a control siRNA is transfected (Fig. 2). Passing all criteria were 13 miRNA sequences resulting from 10 hairpin precursors (Fig. 3). All recovered sequence reads of the BV-encoded miRNAs, including the number of hits generated in cells harvested at each time point or in TG, are listed in Table 1. Several of the miRNA sequences exhibited variable 5′ and 3′ ends, which are also shown in Table 1. The majority of the novel miRNAs were located in the repeat region of the BV genome (Fig. 4). A few miRNAs were located in the unique long (UL) region, and none were found in the unique short (US) region (Fig. 4). Similar observations have been made for HSV-1- and HSV-2-encoded miRNAs (14, 22, 40).
It is interesting to note that the majority of the miRNAs are encoded within the long repeat (RL) region, in which there is no sequence homology to HSV or in the short repeat (RS) region where there is an additional ~1.5 kb of nonhomologous sequence relative to HSV (Fig. 4) (17). The majority of the miRNAs detected in the repeat regions map to a locus that in HSV would encode the latency-associated transcript (LAT), the only abundantly expressed viral RNA during latency. Three miRNAs were recovered in the UL region. The position and location of all the miRNAs detected are shown in Table 1.
Since the deep sequencing technique was not quantitative and is a poor predictor of the amount of a viral miRNA in an infected cell (16), we chose to analyze expression of virus-encoded miRNAs by stem-loop RT-qPCR in latently infected TG and throughout productive infection. This technique has been shown to have a high level of specificity for the precise miRNA sequence for which the probes are designed. Notably, miRNAs with only single-base differences in either length or sequence are readily differentiated using this technique (11). Expression levels were normalized to levels of a cellular miRNA, Let-7a, and values are expressed relative to a negative control sample of small RNA isolated from a BV DNA-negative TG or noninfected Vero cells (Fig. 5 and and6).6). The expression of Let-7a was shown to be unaltered during virus infection (see Fig. S1 in the supplemental material) (6). The use of relative expression permits the comparisons between the relative abundance of an individual miRNA under different conditions (e.g., at different times during infection), but not between the different miRNAs, and has been used in other studies of virus-encoded miRNAs (24, 38–40). We analyzed RNA recovered from cells harvested at 0, 3, 6, 9, 15, 18, 21, and 24 hpi, one BV-negative TG, and four BV-positive TG. We reasoned that several of the miRNAs not detected in the TG during deep sequencing and located in the region that encodes the LAT transcript in HSV may still be expressed during latency; thus, ganglia were tested for all miRNAs in this region.
The following four BV-encoded miRNAs were detected in latently infected TG by RT-qPCR: hbv-mir-B8-3P, -B20-5P, -B22-3P, and -B26-5P (Fig. 5). hbv-mir-B20-5P was readily detectable in all positive ganglia, but at variable levels (Fig. 5). The other miRNAs also exhibited inconsistent expression between TG (Fig. 5). Although identified from TG during deep sequencing, hbv-mir-B7-5P was not detectable in any TG by RT-qPCR. This observation suggests that a combination of techniques may be required to discover all miRNAs encoded in a given system.
RT-qPCR of BV-infected cells at various times postinfection was used to analyze the expression of 13 BV miRNAs during lytic infection (Fig. 6). Surprisingly, most of the miRNAs did not exhibit the expected regulated cascade-type pattern of alphaherpesvirus gene expression. Rather, most of the miRNAs were abundantly detected at early times, and either continued to increase in abundance or remained at constant levels as infection progressed (Fig. 6). An exception is hbv-miR-B19-5P, which was first detected at 15 hpi.
The absolute amount of each miRNA was determined at the time of maximal expression using RT-qPCR assays calibrated with synthetic standards. The measurements of each miRNA are in the following format: miRNA, time postinfection assayed, number of molecules per cell. The measurements are as follows: hbv-mir-B2-3P, 21 hpi, 94; hbv-mir-B3RC-5P, 21 hpi, 20; hbv-mir-B3RC-3P, 15, 94; hbv-mir-B7-5P, 21, 3,994; hbv-mir-B8-5P, 21, 2,897; hbv-mir-B8-3P, 21, 32,757; hbv-mir-B14RC-3P, 18, 303; hbv-miR-B19-5P, 21, 88; hbv-mir-B20-5P, 9, 2,048; hbv-mir-B20-3P, 21, 2,127; hbv-mir-B21RC-5P, 21, 57; hbv-mir-B22-3P, 21, 2,702; hbv-mir-B26-5P, 9, 17,313. Further work will be required to determine if these levels of expression are biologically meaningful, particularly the lower values. It has been proposed by others that expression of >100 molecules of an miRNA per cell suggests an active miRNA (9). However, it should be noted that the measurements described above reflect a steady-state level of miRNA abundance, which is dependent on both expression and stability. Recent evidence suggests that miRNA stability is influenced by target abundance; thus, an miRNA may be present at low levels because of a large amount of target in a cell (3). Further work is also required to determine the effect of stability on the abundance of the miRNAs.
Expression of alphaherpesvirus gene can be grouped into three kinetic classes, immediate early, early, or late, based on their requirement for the expression of viral proteins and viral DNA replication (19). To determine the specific kinetic class of each of the viral miRNAs, viral infections were performed in the presence of various pharmacological agents.
PAA inhibits DNA replication. Therefore, BV transcripts that are expressed in a DNA replication-dependent manner—true late-class transcripts—are not expressed in the presence of PAA. The absence of the BV true late gene UL44 (gC) mRNA in cells infected with BV shows that PAA effectively inhibits BV-late gene expression (Fig. 7B). In the presence of PAA, hbv-mir-B2-3P, -B14RC-3P, and -B21RC-5P were not affected, suggesting they could be generated from immediate early, early, or leaky late transcripts (Fig. 7A). The remaining miRNAs appear to fall into two categories: (i) hbv-mir-B7-5P and hbv-mir-B19-5P were undetectable in the presence of PAA, suggesting they are generated with true-late kinetics; (ii) hbv-mir-B3RC-5P, -B3RC-3P, -B8-5P, -B8-3P, -B20-5P, -B20-3P, -B22-3P, and -B26-5P exhibit intermediate levels of expression in the presence of PAA, which suggests they may be generated with leaky late, early, or immediate early kinetics, or some combination of multiple kinetic classes.
We considered that miRNAs readily detectable at 3 hpi (Fig. 6) were potentially expressed with immediate early kinetics (hbv-mir-B8-5P, -B8-3P, -B20-5P, -B20-3P, -B22-3P, and -B26-5P). Only immediate early class genes are synthesized in the absence of viral protein synthesis. Cells were infected in the presence of the protein synthesis inhibitor CHX, and RNA was harvested at 60 min postinfection. Expression of candidate immediate early miRNAs was quantified by RT-qPCR and compared to RNA harvested from infected cells not treated with CHX (Fig. 7E). miRNAs hbv-mir-B8-5P, -B8-3P, -B20-5P, -B20-3P, and -B26-5P were expressed at similar levels in the presence or absence of CHX consistent with immediate early class expression of their pri-miRNAs (Fig. 7E). hbv-mir-B22-3P was slightly affected by CHX (Fig. 7E), suggesting that most of the miRNA detected at 60 min postinfection arose from an immediate early pri-miRNA, in addition to miRNA that arose from pri-miRNAs from another kinetic class. As a control, effective inhibition of de novo protein synthesis was demonstrated by using a pulse of [35S]methionine (Fig. 7C); protein from these cells was resolved by SDS-PAGE, and the gel was stained with Coomassie blue and then exposed to a phosphor storage screen. Equal loading of both lanes on the gel was demonstrated by similar Coomassie blue intensities, and inhibition of protein synthesis was demonstrated by the absence of incorporated radiolabel in the CHX-treated cells (Fig. 7C).
Several herpesviruses have been reported to package RNA molecules into virions (5, 8, 12, 18, 30). To investigate the possibility that BV-encoded miRNAs were detectable in the absence of viral gene expression—suggestive of incorporation into the virion—cells were infected in the presence of the inhibitor of transcription act D, and RNA was harvested at 60 min postinfection. Expression of candidate virion-packaged miRNAs was quantified by RT-qPCR and compared to RNA harvested from infected cells not treated with act D. miRNAs hbv-mir-B8-5P, -B8-3P, -B20-5P, -B20-3P, and -B26-5P were all detected from cells infected in the presence of act D, thereby suggesting that miRNAs may be incorporated into virions (Fig. 7E). As a control, effective inhibition of de novo viral transcript synthesis was demonstrated by showing that ICP4, an immediate early class viral gene, is not expressed in the presence of act D (Fig. 7D). For each miRNA, the amounts detected were lower than those for RNA harvested from infected cells treated with CHX. The amount of each miRNA detected in the presence of CHX likely represents newly synthesized miRNAs plus miRNAs brought into the cell with the virion. Thus, the difference between the amount observed with CHX and that with act D likely represents only newly synthesized miRNA.
To further investigate the possibility that miRNAs detected in the presence of act D were packaged into BV virions, we harvested RNA from purified virions. To reduce the possibility that cytoplasmic miRNAs copurified with virions, the virions were RNase treated. Prior to RNase treatment, the virions were spiked with a known amount of synthetic ath-mir-166, an miRNA from Arabidopsis thaliana that is absent from mammalian cells (26), including Vero cells (M. Amen and A. Griffiths, data not shown). The absolute amount of ath-mir-166 was determined in RNase-treated samples and non-RNase treated samples by calibrating the assay using a standard curve. Accounting for 28% recovery of input ath-mir-166 (determined from knowing the input and mass recovered in the absence of RNase), the RNase treatment procedure was shown to be effective and resulted in ~94% degradation of the input ath-miR-166.
Using standard curves generated for each miRNA by using synthetic RNA oligonucleotides, the number of molecules of miRNAs hbv-mir-B8-5P, -B8-3P, -B20-5P, -B20-3P, and -B26-5P harvested from virions were quantified. These values were normalized to the number of viral genomes harvested from the virions and ranged from 1 to 28 miRNA molecules per viral genomic DNA molecule (Table 2). Additionally, the number of infectious units per DNA genome was determined to be 51 and 113, from two different viral stocks (Amen and Griffiths, not shown). Thus, the number of miRNAs per infectious virion is likely to be greater than the estimate per genomic DNA molecule.
Recent evidence of a new class of small RNA species, known as miRNA offset RNAs (moRNAs), in Kaposi's sarcoma-associated herpesvirus (KSHV) (25) and HSV (22, 37) prompted us to search our sequencing data for BV-encoded moRNAs. Manual analysis of the longer precursor stem-loops revealed four BV-encoded moRNAs originating from three loci (Table 3). The precursor stem-loops are shown in Fig. 3. As previously reported, these moRNAs were detected at much lower levels than BV-encoded miRNA levels (32). The moRNAs were detected later in infection, at 15 and 21 hpi only, and were located within the nonhomologous regions relative to HSV. We assume that only when the precursor stem-loops reach high abundance can the moRNAs be detected by deep sequencing. We also detected 5′ sequence heterogeneity of hbv-mor-B20-3P, as has been observed for other moRNAs (32).
Deep sequencing was used to identify a total of 13 BV-encoded miRNAs. All miRNAs were detected during lytic infection, and four were detected in naturally infected cynomolgus macaque TG.
Unlike the high degree of sequence conservation seen between cellular miRNAs, herpesvirus-encoded miRNAs are generally not well conserved (17, 33, 42). Until recently, Epstein-Barr virus (EBV) and rhesus lymphocryptovirus (rLCV) were the only herpesviruses known to share evolutionarily conserved miRNAs. New evidence has revealed that HSV-1 and HSV-2 share 11 positional and/or sequence homologs (22). Three of them share perfectly homologous seed sequences, while five are similar in 6 of the 7 bases of the seed region (22). These observations suggest conserved targets for the homologous miRNAs. Since BV is a close relative to HSV-1 and HSV-2, it is notable that only one of the BV-encoded miRNAs reported here shared any sequence similarity to the HSV-encoded miRNAs. hbv-mir-B14RC exhibits 68% identity to hsv1-miR-H16, which does not have a homolog in HSV-2 (22). Interestingly, hbv-mir-B14RC is located in the RL region that is nonhomologous to HSV-1 (Fig. 4), while hsv1-mir-H16 is located in the 5′ untranslated region (UTR) of UL33 in the UL region (22). Both hbv-mir-B14RC and hsv1-mir-H16 were expressed during the lytic-cycle (Fig. 6) (22). The miRNAs are not likely to have the same target sequence, as only 3 of 7 nucleotides of the seed sequence are identical. There were no sequence similarities between BV-encoded miRNAs and other virus-encoded miRNAs.
Most BV-encoded miRNAs are located in genomic locations similar to those of the HSV-encoded miRNAs (Fig. 4) (22). This is similar to other related herpesviruses, such as KSHV and rhesus rhadinovirus (RRV) (29, 42). The significance of positional homology and sequence diversity is unclear, although it may reflect similarities in the expression of the miRNAs (e.g., during latency), rather than the targets of the miRNAs. However, we cannot rule out the possibility that the BV-encoded miRNAs share homologous viral targets or cellular targets with HSV-encoded miRNAs. Interestingly, 10 BV-encoded miRNAs are located in the region of the BV genome that lacks sequence similarity to HSV-1 (Fig. 4). While the majority of the BV genome is similar to that of HSV, three areas in the repeat regions have notably low sequence similarity to the corresponding regions of HSV. These regions are the flanking sequences to ICP0 (which are also shorter than in HSV) and a region of an additional ~1.5-kb sequence in the short repeat (Fig. 4) (25). Given the sequence differences, we are investigating the possibility that the miRNAs encoded in these regions play important roles in the differing severity of human disease seen between BV and HSV.
The majority of miRNAs identified in HSV-1 and −2 are generated from regions in or proximal to LAT, and many are thought to be important for latency (12). Like HSV, most of the BV-encoded miRNAs are located in or proximal to LAT (Fig. 4). The four miRNAs identified in the TG by deep sequencing are likely to be generated from LAT (Fig. 4). Although the BV LATs have not been formally mapped, based on the latent expression of BV-encoded miRNAs, we infer that LAT begins upstream of hbv-mir-B22-3P and extends downstream of hbv-mir-B8-3P (Fig. 4). Thus, it appears that the LAT region is at least 8.3 kb and includes ~1.5 kb of sequence in the RS region that is considered absent in HSV. The two miRNAs in this region generated from the opposite strand (hbv-mir-B14RC and -B21RC) were detected only during lytic replication. There were variable expression levels of the miRNAs in the TG (Fig. 5), which was surprising given they are likely to be generated from the same pri-miRNA (LAT). While the significance of this is unclear, it is interesting to speculate that it relates to the function of the miRNAs in the different ganglia. It has been recently shown that miRNA stability is decreased when the concentration of its target sequence is increased (3). Thus, it is possible that the BV-encoded miRNAs are acting upon target mRNAs that have variable expression between the ganglia, resulting in variable miRNA stability between ganglia.
We investigated the expression of the miRNAs during lytic infection by examining their changes in expression over time and their sensitivity to various agents that inhibit expression of genes in particular kinetic classes. Interestingly, expression of many of the miRNAs could not be easily categorized into the traditional immediate early, early, and late (true late or leaky) classes. It should be remembered that rather than directly inhibiting the expression of the miRNAs, these inhibitors are affecting expression of the progenitor RNAs, the pri-miRNAs. While it is possible that the pri-miRNAs are expressed in a noncanonical manner, we believe that it is more likely that several of the miRNAs are expressed from multiple pri-miRNAs from different temporal classes. This would result in a highly complex BV transcriptome, although it is recently becoming clear that herpesviruses have a much more complex pattern of transcription than previously recognized (46). The reasons for expression from multiple pri-miRNAs is unclear, but it is interesting to speculate that this represents a strategy that allows the virus to express a particular miRNA throughout infection to maximize the time an miRNA has to perform its function.
Analyses of miRNA expression during the course of infection led us to examine their expression in the absence of de novo gene transcription. Both viral and cellular RNAs have been reported to be packaged in virions of herpesviruses (8, 18, 30). Packaging of RNAs in the virion allows for immediate expression upon virus entry. After the first round of infection, the infected cell releases new virus particles as well as cell debris and cellular products, such as cytokines, that may affect the surrounding cells. RNAs packaged in the virion may be influential in overcoming the activated host response immediately upon entry into these uninfected cells. For example, the human cytomegalovirus (CMV) has been shown to package UL21.5 mRNA, which binds the chemokine receptor RANTES and modulates the host cell response by blocking its ability to bind to its receptor (43). This CMV-encoded protein can therefore potentially modify the antiviral response before viral transcription occurs. Additionally, viruses that have short replication cycles may take advantage of RNAs packaged in the virion to establish productive and persistent infection in a timely manner. While it has been well established that RNAs can be packaged in virions, it is not widely recognized that viruses package noncoding RNA molecules, including miRNAs. Cliffe et al. demonstrated that viral tRNA-like molecules are packaged in murine gammaherpesvirus 68 (MHV-68) virions at levels detectable by Northern blot hybridization (12), which suggests that they are present at higher levels in virions than the BV-encoded miRNAs. While their role in virus pathogenesis is unknown, their presence in the virion suggests they have a role early in infection. Similarly, several BV-encoded miRNAs were packaged into virions. The number of molecules of hbv-mir-B8-5P, -B8-3P, -B20-5P, -B20-3P, and -B26-5P ranged from 1 to 28 molecules per viral genome. While these numbers are fairly low, we recognize that not every genome is packaged within an infectious particle. The genome infectivity ratio was determined to be 51 and 113; therefore, these numbers are likely an underestimation of the number molecules per infectious virion. This is the first report showing that miRNAs are packaged into virions, and further work is required to understand the biological significance of miRNAs packaged into virions.
Analysis of the mature miRNA sequences revealed sequence variation at both the 5′ and 3′ ends (Table 1). This sequence heterogeneity has major implications for miRNA function. The 5′ end is important for miRNA target selection; full sequence complementarity of nt 2 to 8 of the mature miRNA, referred to as the seed region, is important for mRNA translational inhibition (4). Therefore, even differences of 1 nt at the 5′ end potentially alter the target mRNAs. miRNA sequence heterogeneity has been observed in humans (45), Drosophila melanogaster (31), Caenorhabditis elegans (27), and herpesviruses, including KSHV (37), HSV (39), and rLCV (27). In all three of these herpesviruses, it was reported that 3′-end variation occurred more frequently than 5′-end variation (27, 37, 39). Importantly, it has been extensively documented that sequence heterogeneity observed in mature miRNAs is not due to RNA degradation or any other artifact (27, 39). The 5′ sequence heterogeneity observed in BV-encoded miRNAs may enable BV to expand the number of regulated genes, thereby optimizing the limited capacity within the viral genome. While the 3′ end of miRNAs may have a minor role in target discrimination, there is recent evidence that they are important for stability (3). 3′-sequence heterogeneity may result in differing degrees of stability between the various species. In the context of BV-encoded miRNAs, the significance of the observed 3′-end heterogeneity is unclear; however, it is interesting to consider that it may afford the virus a mechanism to regulate miRNA activity at a postexpression level.
A newly recognized class of small RNA molecules, called miRNA offset RNAs (moRNAs), has recently been described. They were first discovered in 2009 by using deep sequencing of small RNAs expressed in the simple chordate Ciona intestinalis (32) and recently in KSHV (37) and HSV-1 and -2 (22). While the significance and biogenesis of this new class of RNA species are unclear, it is understood that they arise from sequences located adjacent to the predicted pre-miRNA stem-loop, and Drosha is responsible for cleavage at one end. It may be possible that these molecules are still loaded into the RNA-induced silencing complex (RISC) and could therefore be potential regulatory molecules. Further investigation of this new class of small-RNA species will be required to understand their significance in BV biology.
There are advantages and disadvantages to all of the methods we have used to identify miRNAs. For example, it has been noted previously that there are drawbacks in the use of deep sequencing for miRNA identification, including its ability to report certain miRNAs more efficiently than others (22, 39). Furthermore, the number of sequence reads for a single miRNA is not necessarily proportional to its relative abundance (22). The order that we addressed each criterion for candidate miRNAs was dictated by practicality. However, there remains an anomaly: hbv-mir-B19-3P did not pass the final criterion to be classified as an miRNA. Deep sequencing detected this molecule in abundance; the most prevalent species from deep sequencing had 86 hits at 3 hpi, 198 at 9 hpi, 454 at 15 hpi, and 489 at 21 hpi. RT-qPCR revealed latent expression of hbv-mir-B19-3P; however, it was not detected by deep sequencing RNA from TG. RT-qPCR also revealed it was highly expressed throughout infection; at 3 hpi, it was expressed at 103-fold over the background and increased to 104 by 24 hpi. We anticipate that hbv-mir-B19-3P is most likely a real miRNA but was not detected in transfected cells, perhaps because there is a difference in the production of the star and mature miRNA as in infected cells. Further analysis will be required to understand whether hbv-mir-B19-3P is indeed an miRNA. Taken together, it appears that a combination of techniques and conditions is desirable for comprehensive virus-encoded miRNA discovery.
Although the list of virus-encoded miRNAs continues to grow, especially within the herpesvirus family, understanding their function remains a challenging task. Herpesvirus-encoded miRNAs have been shown to function during both latency and productive infection (7). Given the rapid replication cycle of simplexviruses, it is perhaps easier to imagine a productively expressed miRNA repressing only viral genes, while a latently expressed miRNA has sufficient time to repress both viral and/or cellular genes. However, the observation that miRNAs are packaged into virions allows the virus to regulate gene transcription before de novo synthesis occurs, and thus we anticipate that all herpesvirus-encoded miRNAs may regulate both viral and cellular genes in both latent and lytic infections. Latent infections of herpesviruses are characterized by very low levels of viral protein expression, effectively avoiding detection by the immune system. Thus, viral miRNAs may provide herpesviruses with a nonimmunogenic method to modify the cellular environment. During productive infection, virus-encoded miRNAs may be important for replication and the production of infectious progeny.
Herein we describe the discovery and expression analysis of BV-encoded miRNAs using a combination of deep sequencing, RT-qPCR, and expression in transfected cells. Using these techniques, a total of 13 miRNAs were detected during lytic infection, and 4 were detected in latently infected macaque TG. This is the first report that miRNAs are incorporated into virions. These data provide the framework for future analyses of BV-encoded miRNA functions.
We thank Mallory Harden for excellent technical support and critical reading of the manuscript, S.-J. Gao and Robert Lanford for advice, Adriana Mejia for generating plasmid pAMBVgB, Jean Patterson and Ricardo Carrion, Jr., for help in the ABSL-4 laboratory, and Edward Dick for help with necropsies. We also thank Roy Garcia, Sarah Williams-Blangero, and John Blangero for help with deep sequencing. In particular, we are deeply grateful to Tim Anderson and Shalini Nair for help with the initial stages of deep sequencing.
This work was supported by Texas Biomedical Research Institute V&I startup funds and a grant from the Southwest Foundation Forum, P51 RR013986, and was conducted in facilities constructed with support from the Research Facilities Improvement Program (grant number C06 RR012087) from the NCRR.
†Supplemental material for this article may be found at http://jvi.asm.org/.
Published ahead of print on 4 May 2011.