Heliothis zea nudivirus 1 (HzNV-1) is a baculiform insect virus with a circular double-stranded DNA genome. This viral genome has been sequenced and was found to encode approximately 154 open reading frames^1,2,3,4,5. This virus was originally regarded as a member of the Baculoviridae family, but due to its lack of an occlusion body and low sequence homology to baculoviruses⁶, it is now temporarily re-classified with other non-occluded viruses as a newNudivirus genus^7,8. TheHzNV-1 virus has a relatively broad host range and has been reported to infect many insect cell lines^1,9,10,11.

HzNV-1 virus can establish distinct latent and productive infection cycles in insect cell cultures^3,12,13,14. It thus provides a convenient model system for the study of molecular switches between these two viral infection cycles. During productive infection, the virus generates more than 100 transcripts and high titers of virus progeny are produced. In this infection stage, the majority of Tn-368 and Sf-21 insect host cells are killed leaving a small percentage of the cells become latently infected^3,13. During viral latency, the persistency-associated gene 1 (pag1), is the only viral gene transcript detected⁴. In this infection stage, virus genomes exist either as episomes or are inserted into the host genome and persist through many cell passages without the releasing of viral particles¹⁴. Occasionally, viruses can be released from very small proportions of latently-infected cells (usually less than 0.2%), resulting in the death of these cells. This small quantity of continuously-released virions results in the presence of low viral titers (around 10³ pfu/ml) in the culture medium of the latently-infected cells^4,14.

Althoughpag1 is the only transcript detectable during latent viral infection, it is also expressed during productive viral infection. Uniquely, the transcript ofpag1 is a non-coding RNA, previously also referred to as the persistency-associated transcript (PAT1)^3,4 and was found to be involved in the establishment of latentHzNV-1 virus infection^3,4. Thepag1 has several interesting features. Sequence analysis ofpag1 revealed abundant direct and inverted repeats. The transcript ofpag1 is not translated, as it does not associate with polysome^4,5 and was later shown to be a nuclear RNA^4,5. Although we can not rule out the possibility thatpag1 transcript is an intron of a larger transcript, the fact that the direct upstream sequence ofpag1 coding region is a promoter suggesting that this is more likely to be an independent transcript^3,5.

During productive viral infection, an abundant 6.2 kb transcript is expressed from a gene located in the fragment ofHzNV-1Hind III-I (hhi1) of the viral genome^3,4,15,16.hhi1 is a viral gene expressed very early during viral productive infection (0.5 h post viral infection, hpi) and it was shown to involve in viral re-activation from latency^3,16. It was also shown that the suppression ofhhi1 expression can switch viral infection from productive to latent infection^3,16. In contrast with the expression of most immediate-early genes of baculovirus, the proper expression ofhhi1 requires viral factors¹⁵.

Recently the herpes simplex virus (HSV-1) was shown to use the miRNAs derived from a latent transcript to block the function of early genes, without the proper function of early transcripts, it is speculated that virus may enter latency¹⁷, however, a proof is still lacking. In this study, we found thatHzNV-1 can serve as a convenient system to investigate such possibility. We provided evidences showing that thepag1 is capable of down-regulatinghhi1 transcript in this virus. Subsequent experiments showed thatpag1 functions through the expression of two microRNAs (miRNAs), which target to and degradehhi1 transcript. These miRNAs could be detected in viral infected cells and our data indicated that the miRNA derived frompag1 can establish latent viral infection in the cells.

The establishment of latent viral infection through miRNA-producing non-coding viral RNA represents a simple yet highly effective way forHzNV-1 to achieve latent viral infection in the host cells. Although different fromHzNV-1, a further prove for the miRNA to establish viral latency in HSV-1 may be still needed, the similarity between HSV andHzNV-1 in using miRNA derived from latent-specific transcripts to control early transcript and subsequently result viral latency represents an interesting viral convergent evolution. Furthermore, miRNA as a switch for the transition of productive to latent viral infections may be a widely-spreaded phenomenon of virus/host interactions in nature.

hhi1 expression is negatively regulated bypag1

In order to study the interactions ofpag1 andhhi1, two expression plasmids, pKSpP1 and pKShHI, were constructed. The former usedpag1 and the latter usedhsp 70 promoters to expresspag1 andhhi1 transcripts, respectively. Northern analysis showed thatpag1 was not only expressed in the productive and latent viral infected cells, it was also properly expressed in pKSpP1 transfected cells (Fig. 1A). Both plasmids were then co-transfected into SF-21 cells to identify any possible interactions. Thehhi1-expressing vector pKShH1 was also transfected into SF-21 cells with an empty vector (pBluescript II KS−, Stratagene) as a control. Northern blot analysis showed that the level ofhhi1 expression decreased dramatically in the presence ofpag1 (Fig. 1B). To confirm thatpag1 can negatively down-regulatehhi1 expression duringHzNV-1 infection, cells were first transfected withpag1-expression vector pKSpP1 and the transfected cells were subsequently infected withHzNV-1. Similarly,hhi1 expression from viral infection was suppressed by the transfection of pKSpP1 (Fig. 1C). In this experiment, the level ofhhi1 transcript was detected at 2 hours postHzNV-1 infection (hpi), because, according to our previous study, this is the timing at whichhhi1 expression reaches to the highest level¹⁶.

Interactions ofhhi1 andpag1 by northern analysis.

(A) The expression ofpag1 transcripts in the productive infected, latent infected and pKSpP1 transfected cells were probed. (B, C) pKShHI was either co-transfected with pKSpP1 (B), or co-infected withHzNV-1 (C), then subjected to northern analysis usinghhi1 probe. Actin signal were used as a loading control.

Full size image

Prediction of possiblepag1 miRNA precursors

Previously, sequence analysis ofpag1 revealed the presence of abundant direct and inverted repeats with no significant open reading frames (ORFs) in its coding region^4,5. In this experiment, our analysis showed thatpag1 transcript contains abundant stem-loop structures. Therefore, it is possible thatpag1 transcript may produce miRNAs to suppress the expression ofhhi1. In order to explore this possibility, 23 possible miRNAs were first predicted frompag1 coding region and five of them were predicted to target tohhi1 coding region (Fig. 2).

Mapping of the predicted miRNAs inpag1 transcript.

Schematic drawing of thepag1 transcript to show its 5 predicted miRNAs againsthhi1 and 2 miRNAs, hv-miR-246 and hv-miR-2959, proved to be expressed by virus in the infected Sf21 cells.

Full size image

Proper expression of miRNAs frompag1 is confirmed in productively and latently infected cells and also in cells transfected withpag1 plasmid

To confirm whether the above predicted miRNAs truly come frompag1 and are expressed in infected Sf-21 cells, a stem-loop PCR^18,19 was performed to analyze the miRNA expression inpag1-transfected cells. Thepag1-transfected cells were harvested at 12 hpi and total RNA was extracted. The cDNAs of miRNAs were obtained using miRNA stem-loop RT primers (Table 1). Stem-loop PCR analysis was performed using these cDNAs as the templates with specific miRNA primers designed based on our predictions. In order to investigate whether if any of the predicted miRNAs are expressed in viral infected cells, total RNAs were extracted from productive and latent viral infected cells and subsequently analyzed by stem-loop PCR. Real-time PCR products were also separated by gel electrophoresis and visualized by EtBr staining as a further control to show proper expression of different PCR products (Fig. 3 andSupplemental Fig.1). The result showed that two miRNAs, hv-miR-246 5P and hv-miR-2959 5p, were expressed in theHzNV-1 productively infected cells (Fig. 3A),HzNV-1 latently infected cells (Fig. 3B) andpag1-transfected cells (Fig. 3C). These two miRNAs were cloned and the predicted secondary structures of hv-miR-246 5P and hv-miR-2959 5p precursors are shown inFig. 3D and Fig. 3E, respectively. Mature miRNA are shadowed in red and the numbers on nucleotides indicating their corresponding positions topag1 transcript.

Cloning and analysis of the predicted miRNA by stem-loop PCR and northern blot.

Stem-loop PCR was performed to clone and analyze the proper expression of the predicted miRNAs in (A)HzNV-1 productively infected cells, (B)pag1-transfected cells and (C)HzNV-1 latently infected cells. (D, E) Predicted secondary structures of hv-miR-246 5P (D) and hv-miR-2959 5p (E), precursors. (F) Small RNAs harvested fromHzNV-1 productively infected cells at various time points were analysis by northern blots with probes against predictedHzNV-1 miRNAs (top panels) or let-7a miRNA as a positive control (bottom panels).

Full size image

The melting curve for hv-miR-246 5p had two peaks in theHzNV-1 productively infected cells (Fig. 3A). We cloned these two fragments into cloning vector and analyzed their sequences. The sequence of the upper band did not match to the predicted miRNA and therefore is a non-specific product. The molecule size of the upper band is also lager than the predicted miRNA. In these experiments, proper expression of let-7a miRNA gene was performed as a positive control²⁰. To further confirm the results from stem-loop PCR, northern blot analysis showed that hv-miR-246 and hv-miR-2959 became detectable in productively infected cells at 4 hpi and 8 hpi, respectively (Fig. 3F). However, they were not detected in mock infected cells (data not shown). These results indicated that thepag1 transcript was indeed processed into miRNAs during the infection ofHzNV-1 (Fig. 3F).

hhi1 expression can be down-regulated bypag1 miRNAs

To further test whether these two miRNAs targethhi1 and suppress its transcripts, these miRNAs were separately transfected into Sf-21 cells along with ahhi1-expressing vector and the levels ofhhi1 transcript were analyzed by northern blot analysis at 12 hpi. Both miRNAs were able to reduce the amount ofhhi1 transcript within the cells (Fig. 4). The sequences of hv-miR-246 and hv-miR-2959 are mapped to nucleotides 226325 to 226366 and 226527 to 226555 of thehhi1 coding region, respectively (Fig. 2). To confirm if the matching of miRNA sequences tohhi1 was important,i.e., acting as targeting sites, we constructed mutant miRNAs with three base-pair mutations in the matching regions for these two miRNAs (Fig. 4A, 4C). The mutational substituted bases are indicated by arrows. Northern analysis (Fig. 4B, 4D; left panels) and RT-PCR (Fig. 4B, 4D; right panels) showed that these two mutant miRNAs did not affect the levels ofhhi1 transcript (Fig 4B, 4D), suggesting that our predicted target positions are crucial for functioning. In these experiments, the levels of actin transcript were used as a loading control.

Down-regulation ofhhi1 expression bypag1, hv-miR-246 and hv-miR-2959.

(A, C) sequences of hv-miR-2959 and hv-miR-246 were shown and miRNAs with mutations were denoted as hv-miR-246m and hv-miR-2959m, separately. (B, D) levels ofhhi1 transcript in various treatments were analyzed by northern hybridization (left panel) and RT-PCR (right panel).

Full size image

pag1 miRNA functions in establishing latent viral infection

Previously, we showed thatpag1 gene can promote the establishment of latent viral infection⁴. To test whether these miRNAs alone are enough to promote latent viral infection, Sf-21 cells were first transfected with miRNA followed byHzNV-1 infection. The number of colonies was then recorded at 12 dpi. Most of the Sf-21 cells died when they were infected withHzNV-1 virus alone and only a small percentage of cells became latently infected (Fig. 5A). However, the number of latently-infected cell colonies increased dramatically upon transfection of individual hv-miR-246 or hv-miR-2959 (Fig. 5B). We also analyzed the gene expression ofpag1 andhhi1 inHzNV-1 infected cells with or without miRNA treatments at 12 dpi (Fig. 5C).hhi1 expression could only be detected inHzNV-1 productively infected cells (Fig. 5C, lane 1) whereaspag1 expression could be detected in productive and latent cells SFP4 (Fig. 5C, lanes 1 and 3). Cells transfected with these two miRNAs, hv-miR-246 and hv-miR-2959, produced onlypag1 transcript but nothhi1 transcript, this is an evidence of viral latency (Fig. 5C, lanes 4 and 5). These results indicate that miRNAs alone can function to establish or at least to strongly promote latent viral infection. Furthermore,pag1 transcript, but not that ofhhi1, was expressed in the latent cells induced by the transfection of miRNA. Besides, these latent cells can be cultured for long-term passages. All these evidences indicate strongly that the establishment of latent viral infection by miRNA is not a transient effect; it is a long term latent viral infection in the cells.

Establishment of latent viral infection by miRNAs.

(A) Sf21 cells were transfected with or without miRNA followed byHzNV-1 infection. (B) Column representation of the results of panel (A). (C) Confirmation ofhhi1 andpag1 expressions in various cells by RT-PCR.

Full size image

In these and previous experiments, we have observed that the number of latently-infected cell colonies increased upon miRNA orpag1 transfection. However, whether these miRNA orpag1 can only promote viral latency or are essential for the establishment of latent viral infection remain unknown. In order to confirm the function of the miRNA andpag1, we further constructed apag1-nullHzNV-1 virus for infection experiment. Sf-21 cells were infected withpag1-null virus or transfected with a siRNA targeting topag1 transcript followed byHzNV-1 infection. RT-PCR confirmed that the suppression ofpag1 gene expression by siRNA was successful (Fig. 6A) and also, nopag1 expression was detected frompag1-nullHzNV-1 virus (Fig. 6B). Further experiments showed that upon knocking down (Fig. 6C, lane 5) or deletion ofpag1 transcripts (Fig. 6C, lane 1), the number of latent colony dropped significantly and none of the cell colony can survive two weeks after viral infection. However, the number of latently-infected cell colonies increased dramatically upon transfection of individual hv-miR-246 or hv-miR-2959 (Fig. 6C, line 3, 4, 7 and 8). These cells keep on viable for long time (at least 2 months up to now). These experiments demonstrated thatpag1 and its derived miRNAs play a crucial role in the establishment ofHzNV-1 latent viral infection.

Suppression ofHzNV-1 viral latency by knocking downpag1 expression.

(A) RT-PCR showed that artificial siRNA can efficiently suppresspag1 expression inHzNV-1 infected cells. (B)pag1 expression is not detectable by RT-PCR in thepag1-nullHzNV-1-infected cells. (C) Formation of latent colony is not observed by the infection ofpag1-nullHzNV-1.

Full size image

Previously, we found thathhi1 can activate the expression of many earlyHzNV-1 genes and reactivateHzNV-1 virus from latency¹⁶. We had also discovered that the transcript ofpag1, a non-coding RNA, plays a critical role to blockhhi1-induced apoptosis²¹ and to the establishment of latent viral infection⁴. In this study we found thatpag1 transcript could suppresshhi1 expression remarkably through the targeting ofhhi1 transcript by miRNAs. Interestingly, these two miRNAs, hv-miR-246 or hv-miR-2959, were individually sufficient for strong promotion and/or establishment of latent viral infection. The observation thatpag1-nullHzNV-1 virus was unable to establish latent infection further suggested thatpag1 plays a key role in the establishment of latent infection.

Most miRNA target sites have perfect pairing to the seed region located near the miRNA 5′ end. Nevertheless, recent reports have also demonstrated that perfect pairing between target sequence and the miRNA can also take place near the 3′ end of miRNA^22,23. Our results also showed that the 3′ ends of miRNAs derived frompag1 have perfect base pairings to the target sequences and mutagenesis experiments showed that these pairing are critical for functioning. To further confirm that inhibition ofhhi1 expression was mediated through miRNA pathway, the effects of silencingdicer1 andago1 expressions (Supplemental Fig. 2A) on the production ofpag1-derived miRNA and controllet-7a were assessed inDrosophila S2 cells (Supplemental Fig. 2B). It is known thatdicer1 is an important component of the miRNA progenesis; whereas,ago1 is not involved in the generation of mature RNAs, rather, is involved in the interaction between miRNA and target transcript. Thus, it is reasonable that levels of miRNAs hv-miR-246, hv-miR-2959 andlet-7a were not affected byago1 knockdown using stem-loop RT-PCR analysis (Supplemental Fig. 2B). We also found thathhi1 expression was not down-regulated bypag1 in eitherdicer1 orago1 knocked down cells (Supplemental Fig. 2C), suggesting that miRNA pathway is essential for the production ofpag1-derived miRNAs.

HzNV1 virus is an insect-specific virus. In our studies, we found it shares striking similarity in the gene regulation networks during viral latency with the vertebrate virus, HSV-1²⁴. Despite the high numbers of viral transcripts produced during productive viral infections, both viruses silence the expression of all genes except the expression of latency-associated transcript (LAT) in HSV-1²⁵ andpag1 transcript inHzNV-1^3,4, during latent viral infection. It is interesting to note thatHzNV-1 is a baculiform insect virus and HSV, on the other hand, is an icosahedra virus. They are quite distinct in shape and genomic sequences and are obviously not evolved from the same origin. Nevertheless, LAT is a latency-associated transcript, which contains a non-coding intron²⁵ andpag1 transcript is similarly a non-coding transcript predominately expressed during viral latency⁴. Such striking similarity between these transcripts strongly suggests that the strategy for producing a non-coding RNA during latency is not an isolated viral infection phenomenon restricted to the HSV of vertebrates.

LAT was previously identified as an 8.3 kb transcript expressed from the genome of HSV²². This RNA is spliced and yields a stable intronic non-coding region of 2 kb and an unstable exonic RNA. This unstable exonic RNA was further processed to produce several miRNAs, which are found to suppress protein expression of early HSV genes, ICP0 and ICP4^17,26. More recently, two sRNAs of 62 nt and 32 nt, but not miRNA, were predicted to be located on the 2 kb stable intronic region. Shenet al. (2009) found that these sRNAs play a role through an unidentified mechanism to inhibit apoptosis and productive infection²⁷. Although these authors showed that these miRNAs or sRNAs were likely responsible for latent HSV infection due to the suppression of two important early genes, direct proof of latent establishment due to these miRNAs or sRNAs is still lacking.

In our results, due toHzNV-1 can establish latent viral infection in insect cells, we were able to show thatHzNV-1 virus establishes latent viral infection through the suppression of thehhi1 transcript using non-coding transcriptpag1. We also showed that even miRNAs frompag1 transcript alone are sufficient to establish latent viral infection. Since the transiently transfected miRNA is sufficient for the triggering of latent viral infection usingpag1-nullHzNV-1, suggesting that once entering viral latency, viral gene expression is some how restricted without the need of a continuous supply ofpag1 transcript forhhi1 suppression.

Our experiments showed that, althoughHzNV-1 virus is not a herpes virus andpag1 transcript has no sequence homology with LAT, shutting-off of all genes except the expression of one non-coding transcript during latency of these two viruses is a striking convergent evolution of viral-host interactions. In addition, the use of miRNAs derived from the non-coding RNA for similar functions by two unrelated viruses suggests that miRNA can be an important and effective mechanism for switching between productive and latent infections in a wide-range of host cells. Furthermore,pag1 transcript is a stable non-coding RNA consistently observed during productive and latent viral infections^3,4, which resembles the intronic stable non-coding RNA of LAT. Although in the study of herpes virus, miRNA was not found from the stable non-coding RNA of LAT, we showed thatpag1 transcript can generate functional miRNAs.

InFig. 7, we summarize the interactions betweenhhi1 andpag1 to better illustrate how they function as molecular switches to determine should the viruses enter productive or latent infections in the host cells. The initial infection ofHzNV-1 results in high levels ofhhi1 expression (Fig. 5C, lane 1)¹⁶ and a moderate level ofpag1 expressions (Fig. 5C, lane 1)^3,4. The newly produced abundanthhi1 transcript can tolerate some RNA degradation caused by the miRNAs derived frompag1. Then, thehhi1 transcript, before degradation or survived from initial degradation, can quickly transactivate many viral early genes for productive viral infection to occur in most of the infected Sf21 cells (Fig. 7A)¹⁶. A low percentage of cells, by whichhhi1 transcript is successfully suppressed bypag1 miRNA during initial viral infection, enter latency¹⁴, where the level ofhhi1 transcript becomes undetectable while still maintain a moderate level ofpag1 transcript (Fig. 7B;Fig. 5C, lanes 3–5)³. Viral re-activation of latent cells can be induced by introducinghhi1 transcript, which then stimulates HzNV-1 early genes expression, resulting in virus production (Fig. 7C)¹⁶. Ifpag1 gene⁴ or its derived miRNAs (Fig. 6C lanes 3,4) were transfected into the cells prior to the initial viral infection, the chance of cells entering latent viral infection becomes strongly boosted (Fig. 7D). Different from the function of miRNAs produced from LAT of the herpes virus, which blocks protein expression, miRNAs derived frompag1 transcript degrade the transcript ofhhi1. More significantly, our evidences proved thatpag1 transcript and its derived miRNAs can establish latent viral infection in the cells.

A comprehensive model for the establishment of productive and latentHzNV-1 viral infections throughhhi1 andpag1 interactions.

Full size image

Cells and virus

Spodoptera frugiperda IPLB-Sf-21 was incubated in TC-100 insect cell culture medium, which contained 10% of FBS at 26°C (Gibco BRL)^28,29. StandardHzNV-l virus was derived by a serial dilution of the stock viral solution and isolated by plaque purification. Latently infected colonies were calculated as described in Wuet al.¹⁶. The titers of the virus clones were estimated by both Q-PCR³⁰ and TCID₅₀^28,29.

Plasmid construction

We obtainedpag1 andhhi1 coding regions by PCR and inserted them down-stream to thehsp 70 promoter (p-hsp)³¹ andpag1 promoter¹⁶ of plasmids pKSh and pKSp, respectively to result plasmids pKShHI and pKSpP1. The nucleotide sequence of thehhi1 gene was submitted to GenBANK, with an accession number of AF264019.

Computational prediction of viral miRNAs

Thepag1 miRNA prediction was carried out using the complete gene sequence ofpag1 (NC_004156) obtained from the National Center for Biotechnology Information. Thepag1 miRNA prediction was performed using miRNA prediction database (Vir-Mir) (http://alk.ibms.sinica.edu.tw)³².

RNA isolation and northern blot analysis

2 × 10⁵ cells/well were seeded in a 24-well culture plate (Corning) and then transfected with two plasmids, pKShHI and pKSpP1, 0.5 µg each, separately or collectively, using Cellfectin according to the manufacturer's protocol (Invitrogen). Sf-21 cells were inoculated at a multiplicity of infection (MOI) of 1 withHzNV-l virus and incubated at 26°C with gentle rocking. After absorption, total cellular RNA samples were extracted from productively-infected cells at the indicated time and from latently-infected SFP4 cells^3,16 using RNeasy Mini Kit (Qiagen). Briefly, total RNA were extracted from virus-infected Sf-21 or SFP4 cells (Qiagen). Then, equal amounts of RNA samples (15 µg each) were loaded, blotted onto Hybond-N+ nylon transfer membrane (GE Healthcare) and hybridized withhhi1 orpag1 specific probes, which were labeled with DIG by PCR amplification of templateHzNV-1 viral DNA using the PCR DIG Probe Synthesis Kit (Roche Applied Sciences) according to the manufacturer's instructions.

miRNA assay by stem-loop real-time PCR and northern blot

The total RNA samples frompag1 transfected cells were isolated using RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. The miRNA cDNAs were synthesized from total RNAs using miRNA specific stem-loop primers (Table 1) according to criteria described previously¹⁸. Analysis of mature miRNA expression by RT–PCR was also carried out using a method described previously¹⁸. The cDNAs were diluted 10 times to perform PCR for expression confirmation and expression pattern analysis. Real-time PCRs were performed, respectively, in 20 µl mixture containing 1 µl cDNA, 0.5 µM forward and reverse primers and 2 × SYBR Green qPCR master mix (Fermentas) with the following parameters: 94°C for 15 s, 60°C for 30 s and 72°C for 30 s. The real-time PCR products were detected by electrophoresis with 3% agarose gel containing ethidium and photographed under UV light. Expression of let-7a, a miRNA normally expressed in most of the cells²⁰, was used as a positive control.

Thepag1-associated small RNA samples frompag1 transfected cells at different time points were isolated using a mirVanaTM miRNA isolation kit (Ambion) according to the manufacturer's instructions. For northern blot analysis of small RNAs, 1 µg aliquots of each small RNA fraction, as well as radio-labeled Decade Markers (Ambion), were fractionated in 15% denaturing polyacrylamide gel electrophoresis (PAGE) gels (acrylamide∶bis ratio, 19∶1) containing 8 M urea in 0.5 × TBE buffer. The gels were soaked briefly in 0.2 µg/ml ethidium bromide in TBE to allow visualization of the RNA using a UV transilluminator (Bio-Rad, Hercules, CA). RNAs were transferred by electroblotting to a Hybond-N+ nylon transfer membrane (GE Healthcare) and UV cross-linked to the membrane. PAGE-purified DNA oligonucleotides (Integrated DNA Technologies) with the reverse complementary sequence to either candidate miRNAs or let-7a, were end labeled with [³²P]ATP (MP Biomedicals, Irvine, CA) to high specific activity. The labeled probes were purified using themirVana Probe & Marker kit (Ambion) according to the manufacturers' protocols. Hybridizations and washes were carried out using the ULTRAhyb-Oligo hybridization buffer according to the manufacturer's directions (Ambion).

Mutant miRNA experiments

2 × 10⁵ cells/well were seeded in a 24-well culture plate and then transfected the plasmid pKShHI (0.5µg) with hv-miR-246, hv-miR-246m, hv-miR-2959 and hv-miR-2959m (50nM each), separately, using Lipofectamine RNAiMAX (Invitrogen). Transfected cells were harvested at 12 hours post-transfection (hpt) for northern analysis using ahhi1 specific probe. hv-miR-246m and hv-miR-2959m are three base mutants created in the perfect pairing regions of hv-miR-246 and hv-miR-2959, separately, to abolish the function of these miRNAs. All miRNAs duplex were synthesized by MDbio Inc. The miRNAs or mutant miRNAs used in this study are as follows:

hv-miR-246 (5′-AGGCUAAGCCCAGCUAAUGAGGCGAG-3′)

hv-miR-246m (5′-AGACAAACCCCAGCUAAUGAGGCGAG-3′)

hv-miR-2959 (5′-CGAGAACGGUUAAUUGCAAUUGCAUC-3′)

hv-miR-2959m (5′-CGAGAAUGAUCAAUUGCAAUUGCAUC-3′)

miRNA transfection uponHzNV-1 infection

Forhhi1 andpag1 miRNA transfected experiments, all miRNAs were synthesized by MDbio Inc. Sf-21 cells (4 × 10⁴/well) in 96-well plates were transfected with 50 nM of the miRNA using Lipofectam RNAiMAX (Invitrogen). Cells were transfected withpag1 miRNAs; then, at 4 hpt, the cells were infected withHzNV-1 virus (MOI = 1) and the number of latently infected colonies was again calculated at 12 dpi⁴. Several primers were designed for the RT-PCR amplification ofhhi1,pag1 andactin as follows:

hhi1-1F: 5′-CGATATGAACATTAACGATGACGATC-3′;

hhi1-1R: 5′-AAACGGATGCAAAATGGACTCAA-3′;

pag1-F: 5′-ACGGGAATTCAGTGTCGAGGACTT-3′

pag1-R:5′-CATGTCTAGAACCCTACCTACCT-3′

actin-F: 5′-CGTGATGGTGGGCATGGGTCAG-3′;

actin-R:5′-CTAATGTCACGCACGTATTCC-3′.

RNA interference

Forpag1 knockdown experiments, all siRNAs were predicted and synthesized by MDbio Inc. The siRNAs used in this study werepag1 siRNA (5′- GGAGUAGGUAGCAUAGAUG-3′) andegfp siRNA (5′- GGCGAUGCCACCUACGGCAAG-3′ ). Sf-21 cells (4 × 10⁴) in 96-well plates were transfected with 50 nM of the siRNA using Silencer siRNA transfection kit II (Applied Biosystems). Cells were transfected with designed siRNAs and, at 4 hpt, the cells were infected withHzNV-1 virus (MOI = 1). The number of latently infected colonies was calculated at 12 dpi⁴. Primers designed for the detection ofhhi1 andpag1 by RT-PCR are listed as following:

hhi1-1F: 5′-CGATATGAACATTAACGATGACGATC-3′

hhi1-1R: 5′-AAACGGATGCAAAATGGACTCAA-3′

pag1-F: 5′-ACGGGAATTCAGTGTCGAGGACTT-3′

pag1-R:5′-CATGTCTAGAACCCTACCTACCT-3′

Construction ofpag1-nullHzNV-1

A fragment containingegfp gene flanked byEcoRV sites was inserted down-stream of thehsp 70 promoter (p-hsp) of plasmid pBShsp70, generating plasmid pBShE. Fragments containing −50 to −300 nt and +300 to +550 nt ofpag1, respectively, were obtained by PCR and subsequently cloned into the upstream and downstream, respectively, ofhsp-egfp of pBShE, yielding pBSphEp. pBSphEp andHzNV-1 viral DNA were co-transfected into Sf-21 cells to generatepag1-nullHzNV-1 virus. Several primers were designed for the PCR amplification ofpag1 upstream and downstream fragments and they are listed as following:

pag1UF: 5′-ATTGAGCTCCCGAGACTGCACTTTTTAAA-3′

pag1UR: 5′-ATTCCGCCGTAGTAGGTGTGCACGTTTT 3′

pag1DF: 5′-ATTATCGATGTGTATTTATAGATGCCCGA-3′

pag1DR: 5′-ATTATCGATAATACGCAGGGTGGACTCTT-3′

We thank Drs. Chang-Chi Lin, Wen Chang, Bryan R. Cullen and May-Ru Chen for valuable discussions and suggestions; and Ms. Miranda Loney and Dr. Harry Wilson for revision. This research is funded by grants 098-2811-B-001-025, 098-2811-B-001-036 and 98-2321-B-001-031-MY3 from the National Science Council and 94S-1303 from Academia Sinica, Taiwan.

Authors and Affiliations

1. Institute of Molecular Biology, Academia Sinica, Nankang, Taipei, 105, Taiwan

   Yueh-Lung Wu, Carol P. Wu, Catherine Y. Y. Liu, Paul Wei-Che Hsu, Eric C. Wu & Yu-Chan Chao

2. Department of Life Sciences, National Chung Hsing University, Taichung, 400, Taiwan

   Yu-Chan Chao

3. Department of Plant Pathology and Microbiology, National Taiwan University, Taipei, 106, Taiwan

   Yu-Chan Chao

Contributions

Guarantors of integrity of entire study, study concepts and manuscript preparation, Y.C.C.; study design, data acquisition/analysis, literature research and manuscript preparation, Y.L.W.; computational prediction, P.W.C.H.; data acquisition/analysis, manuscript editing and revision, C.P.W., C.Y.Y.L. and E.C.W. All authors reviewed the manuscript.

Competing interests

The authors declare no competing financial interests.

This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 3.0 Unported License. To view a copy of this license, visithttp://creativecommons.org/licenses/by-nc-nd/3.0/

Reprints and permissions