A host non-coding RNA, nc886, plays a pro-viral role by promoting virus trafficking to the nucleus

Abstract

Elucidation of the interplay between viruses and host cells is crucial for taming viruses to benefit human health. Cancer therapy using adenovirus, called oncolytic virotherapy, is a promising treatment option but is not robust in all patients. In addition, inefficient replication of human adenovirus in mouse hampered the development of an in vivo model for preclinical evaluation of therapeutically engineered adenovirus. nc886 is a human non-coding RNA that suppresses Protein Kinase R (PKR), an antiviral protein. In this study, we have found that nc886 greatly promotes adenoviral gene expression and replication. Remarkably, the stimulatory effect of nc886 is not dependent on its function to inhibit PKR. Rather, nc886 facilitates the nuclear entry of adenovirus via modulating the kinesin pathway. nc886 is not conserved in mouse and, when xenogeneically expressed in mouse cells, promotes adenovirus replication. Our investigation has discovered a novel mechanism of how a host ncRNA plays a pro-adenoviral role. Given that nc886 expression is silenced in a subset of cancer cells, our study highlights that oncolytic virotherapy might be inefficient in those cells. Furthermore, our findings open future possibilities of harnessing nc886 to improve the efficacy of oncolytic adenovirus and to construct nc886-expressing transgenic mice as an animal model.

Graphical Abstract

In this study, we discovered that a human non-coding RNA, nc886, promotes propagation of adenovirus by modulating the kinesin pathway and thereby facilitating viral nuclear entry. Our findings provide important information when harnessing AdV in clinic and research, such as oncolytic virotherapy and amelioration of an animal model.

Introduction

When viruses infect host cells, the fate of a virus as well as the clinical outcome of hosts is mainly determined by genetic and epigenetic factors of infected cells. Recently, a growing number of studies have reported evidence for the role of non-coding RNAs (ncRNAs) therein. Most of them are about microRNAs and long ncRNAs. Unraveling the interplay between viruses and host cells is crucial in preventing or treating virus infection itself and related diseases. Furthermore, we could reprogram a virus for our benefit in clinics. A prominent example is adenovirus (AdV), which has been widely used in oncolytic virotherapy (reviewed in Jounaidi et al¹) and gene therapy (reviewed in Lee et al²).

Human AdV is a non-enveloped virus with a double-stranded linear DNA genome (reviewed in Jounaidi et al¹ and Mennechet et al³). Upon entry into cells via interaction with cell surface receptors and attachment factors, AdV particles traverse through the cytoplasm along microtubules, to inject AdV DNA into the nucleus (reviewed in Wolfrum and Greber⁴ and Pied and Wodrich⁵). Therein, the transcription of AdV occurs temporally to express early and late genes by host RNA polymerases. Among a total of 30 to 40 AdV genes, most of them are protein-coding genes and are transcribed by RNA polymerase II (Pol II), while two ncRNAs are transcribed by RNA polymerase III (Pol III). These are called virus-associated (VA) RNAs, which are expressed in all human AdV serotypes⁶ and essential for efficient AdV replication (reviewed in Vachon and Conn⁷). The viral gene expression is followed by assembly of progeny virions and release of them accompanying cell lysis.

A clinical utility of AdV is based on cell lysis at the final stage of its life cycle. AdV prefers proliferating cells for its propagation.⁸ In addition, the interferon signaling is usually compromised in cancer cells.^9,10 These facts provided a rationale for AdV to kill cancer cells preferentially (called “oncolytic virotherapy”; reviewed in Baker et al¹¹). Owing to extensive knowledge on AdV biology, AdV has been subjected to genetic modification for development of more selective oncolytic agents to tumor cells. Nonetheless, a basic setback is that the AdV-mediated oncolytic virotherapy is not effective in all cancer cells. It has been noted that approximately one-third of patients treated with AdV exhibited no response.^1,12 In addition, human AdVs replicate inefficiently in mice,13, 14, 15, 16 which makes it challenging to construct an appropriate in vivo model to assess effectiveness of oncolytic AdV for therapeutic purpose.

nc886 is a human ncRNA that is transcribed by Pol III and modulates the activity of target proteins.17, 18, 19 The established function of nc886 is to repress the activity of an antiviral protein called Protein Kinase R (PKR).^17,20, 21, 22 VA RNAs, Pol III transcripts of AdV, also suppress PKR.⁷ It is intriguing that nc886, a host-encoded ncRNA, shares similar properties with AdV ncRNAs. In addition, the interplay between nc886 and virus has been reported in a few studies. nc886 expression is increased upon infection of Epstein-Barr virus and influenza A virus.23, 24, 25 nc886 plays a stimulating role in propagation of influenza A virus.²⁵ All these facts prompted our curiosity as to whether nc886 plays a role during AdV infection. Here our study has uncovered the pro-AdV role of nc886 and its action mechanism.

Results

nc886 promotes AdV gene expression

Since nc886 suppresses antiviral responses,^17,26 we hypothesized that nc886 would act in favor of AdV. To assess the role of nc886, we needed a pair of nc886-expressing and -deficient cells (designated nc886⁺ or nc886^– cells, respectively). Since nc886 plays roles in cell proliferation and apoptosis,^27,28 construction of nc886-expressing or knockout (KO) cells was not successful in all cell lines. Nonetheless, we collected a pair of nc886⁺ and nc886^– cells from four cell lines: Huh7, Nthy-ori 3-1, UM-SCC-19, and RWPE-1 (Figure 1A). A hepatoma cell line Huh7 expressed nc886 abundantly. While culturing Huh7 cells, we fortuitously obtained nc886^– cells and named them Huh7i. When gene expression profiles were analyzed, correlation between Huh7 and Huh7i was higher than that between any pair of unrelated cell lines (Figure S1). This ensured that Huh7i was a genuine derivative from Huh7 but a contaminant. Nthy-ori 3-1 is an immortalized thyroid cell line that expresses nc886. We had constructed an nc886^– cell line by sequential KO of PKR and nc886 in our previous study²⁹ (to be used in Figure 2 and later). UM-SCC-19 and RWPE-1 are cancer cell lines derived from tongue and prostate, respectively. Both of them were nc886-silenced. We constructed nc886-expressing derivative cell lines by transfecting a lentiviral plasmid harboring the nc886 gene. Corresponding control cell lines were also constructed via a parallel procedure with an empty vector. The scheme for cell line establishment is depicted in Figure 1A. The nc886 status was validated by Northern or RT-qPCR measurement (Figures 1B and 1C).

Figure 1.

nc886 promotes gene expression from infected AdV

(A) A diagram depicting cell lines used in (B–D). Dotted and solid arrows indicate fortuitous derivation and experimental production, respectively. Relative nc886 expression levels are designated by a superscripted hyphen (silenced expression) and number of + (with +++ being the highest). nc886⁺ (red letters) and nc886⁻ (blue letters) pairs that were actually used in (B–D) are bold-highlighted. (B) Northern hybridization of nc886 and 5S rRNA to ensure equal loading. (C) RT-qPCR of nc886. The normalized value (to 18S rRNA) of RWPE-1:vector was set as 1. (D) GFP signal upon Ad5/35CMV-GFP infection, with DAPI staining (for Huh7, Nthy-ori 3-1, and UM-SCC-19) or cell photo (for RWPE-1). GFP positive cells were counted and their percentage out of total cells (from DAPI or cell images) are plotted on the right. Infection dose and detection time are 10 MOI at 24 h post-infection (Huh7), 20 MOI at 24 h (Nthy-ori 3-1), and 20 MOI at 48 h (UM-SCC-19 and RWPE-1).

Figure 2.

The impact of nc886 on AdV is independent of PKR

(A) AdV infected and uninfected control (designated as + and –, respectively) to Nthy-ori 3-1 and derivatives that were constructed sequentially as shown in Figure 1A. Infection was at 10 MOI in (B–F). Western blot of indicated proteins, with molecular size markers indicated on the right (top panels). Northern blot of nc886 and 5S rRNA (bottom panels). (B) qPCR of released progeny AdV DNA amounts at 24 h post-infection. An average and the standard deviation from triplicate samples are shown with a p value. (C) Cell photos at 72 h post-infection. (D) Measurement of cell numbers by BCA assay at 72 h post-infection. An absorbance value at 570 nm was converted to a cell density as described in materials and methods. The value of uninfected control was set 100 and relative values were plotted. Each value is an average of triplicates. Standard deviations and a p value are indicated. (E and F) qPCR of AdV from culture supernatant (E) upon infection of AdV (for 24 h) and transfection of indicated anti-oligonucleotides (for 22 h), as described in (B). Cells were harvested and RNA was isolated for RT-qPCR of nc886 (F).

“Ad5/35CMV-GFP” is a modified AdV that is devoid of its early gene E1 but expresses GFP under the cytomegalovirus (CMV) promoter.³⁰ For GFP to be expressed, Ad5/35CMV-GFP should enter into cells upon attachment to the cell surface receptor CD46, the AdV genome should translocate to the nucleus, and AdV genes should be transcribed. So, GFP signal is indicative of how efficiently the AdV propagation cycle operates. The higher GFP signal in nc886⁺ cell lines than in nc886^– counterparts (Figure 1D) indicated that nc886 stimulates a step(s) of the AdV propagation cycle.

nc886 promotes AdV replication in a PKR-independent manner

As stated earlier, an established role of nc886 is to inhibit the activity of PKR, an antiviral protein.³¹ So, it is plausible that the inhibition of PKR accounts for the stimulatory effect of nc886 on AdV. To test this possibility, we employed a set of cell lines that we constructed in our previous study.²⁹ Nthy-ori 3-1, an immortalized thyroid cell line, expresses PKR and nc886 (PKR^wt:nc886^wt in Figure 2A). Since nc886 KO in the PKR wild-type (WT) background was deleterious due to PKR activation and the resultant apoptosis, we had sequentially made PKR and nc886 KO cell lines (PKR^KO:nc886^WT and PKR^KO:nc886^KO) (Figure 2A). While validating the PKR and nc886 status, we found that nc886 expression declined at 24 h post-infection of AdV. Since AdV DNA was replicated and a huge quantity of VAI RNA was expressed at this time point (shown in Figures 3A–3C), the decrease of nc886 was most likely due to an ample copy number of VA promoters that titrated available Pol III enzymes away from the nc886 promoter.

Figure 3.

Effect of nc88 on AdV DNA replication and gene expression

(A) AdV infection timeline. To identify which step nc886 stimulates, we infected AdV (at 10 MOI unless specified otherwise) and conducted various assays at time points that are indicated in each figure. (B) qPCR of intracellular AdV DNA. qPCR Ct values were normalized to qPCR Ct values of 18S rDNA. (C) Northern hybridization of AdV VAI RNA. EtBr staining is shown for equal loading and also for highlighting the abundance of VAI RNA. (D and E) RT-qPCR and Western blot of hexon mRNA and protein. The GAPDH image is the same as Figure 2A, since this and Figure 2A experiments were done in a batch of infection and measurement. (F) RT-qPCR, using primers (indicated in upper panel), of AdV E1A mature mRNA (bottom left) and pre-mRNA (bottom right).

We measured AdV DNA that was released into culture supernatant after completing AdV DNA replication and virion assembly. Importantly, AdV propagation was barely affected by PKR KO but was remarkably diminished upon nc886 KO (Figure 2B). Accordingly, the AdV-mediated cytotoxicity was much higher in nc886-expressing cells than in nc886 KO cells (Figures 2C and 2D).

Since nc886 KO cells were a selected clone, we had a concern about whether the dramatic decrease of AdV release and cytotoxicity might be attributed to a genetic or epigenetic alteration(s) other than nc886. To ensure that it is genuinely the effect pf nc886, we performed a short-term knockdown (KD) experiment by transfecting an anti-oligonucleotides targeting nc886 (“anti-nc886”). The released AdV was significantly diminished by “anti-nc886” as compared with “anti-control,” a non-targeting anti-oligonucleotides (Figure 2E). Efficient KD was verified by RT-qPCR of nc886 (Figure 2F).

There were two points that should be underlined. First, PKR WT and KO cells yielded almost the same results. Second, more importantly, nc886 KO and KD experiments were done in the PKR KO cells. If PKR inhibition were critical for the effect of nc886 on AdV, there must have been no difference regardless of nc886 in the PKR-null situation because there was no PKR to inhibit. These data led to a conclusion that PKR itself did not obstruct AdV life cycle nor could account for the stimulatory effect of nc886 on AdV propagation.

The pro-AdV effect of nc886 operates at or prior to AdV early gene expression

As the first step toward elucidating a mechanism, we attempted to determine a stage(s) at which nc886 exerted its pro-AdV role. Briefly, sequential events in the AdV life cycle are (depicted in Figure 3A) as follows: attachment onto a cell surface receptor, entry into cells, traverse across the cytosol, release of AdV DNA into the nucleus, expression of the immediate-early gene E1A and early genes, AdV DNA replication, expression of late genes, and viral assembly and release (reviewed in Sohn and Hearing³²).

First, we assessed the effect of nc886 on AdV DNA replication by measuring the intracellular copy number of AdV DNA. AdV DNA was barely detectable at 8 h but clearly seen at 24 h (Figure 3B). At 24 h in the absence of nc886, we observed severely impaired AdV DNA replication (Figure 3B) and accordingly very low expression of VAI RNA (Figure 3C). Although VAI is regarded to be an early gene, it is known to be transcribed also from ample copies of replicated AdV DNA. Actually, we observed its expression at 24 h, as huge as could be detectable even by ethidium bromide staining (Figure 3C, bottom panel). Second, we examined the expression of a late gene, hexon. The expression levels of hexon mRNA and protein were markedly decreased in nc886 KO cells as compared with nc886-expressing cells (Figures 3D and 3E). All these results indicated that nc886 acted at or prior to AdV replication.

We measured expression of the AdV immediate-early gene, E1A, at 8 h and earlier. The mature, spliced mRNA was substantially decreased upon nc886 KO (Figure 3F). mRNA maturation involves several post-transcriptional events, which might have been altered by nc886. To exclude these events, we conducted RT-qPCR with a primer pair that measured the precursor mRNA (pre-mRNA) (Figure 3F). The E1A pre-mRNA level was significantly lower in nc886 KO cells than in WT cells (Figure 3F). This lower expression was evidently seen as early as 2 h post-infection. Collectively from our data in Figures 2 and 3, we determined that nc886 promoted AdV at or prior to transcription of the immediate-early gene.

nc886 promotes nuclear entry of AdV DNA

There were two possibilities for the lower E1A pre-mRNA level in nc886 KO cells: a slower transcription rate or fewer AdV DNA molecules that are transcription templates in the nucleus. In the former possibility, we would search for a candidate transcription factor. In the latter possibility, we would further track back earlier steps in the AdV life cycle (see Figure 3A). To determine which was the case, we measured the nuclear amount of AdV DNA by immunofluorescence (IF) of AdV protein VII, a protein associated with the AdV genome. When AdV particles traverse the cytosol and reach the peri-nuclear region, the AdV capsid disintegrates and only the genomic DNA and associated proteins are released into the nucleus. During this course of events, the protein VII epitope is obscured by the capsid in the cytosol but is exposed only after release into the nucleus. Thus, the intensity of protein VII provides an alternative but apt indicator for the AdV DNA quantity in the nucleus.⁵ Also in our IF experiment, protein VII signal was seen within DAPI staining (Figure 4A). Importantly, protein VII signal intensity was significantly weaker in nc886 KO cells than in WT cells (Figures 4A and 4B). The 2-fold difference in IF signal, a proxy of AdV DNA quantity, was concordant with our measurement of E1A pre-mRNA (Figure 3F). Figure 4A images left a slim possibility that the signal might have come from protein VII attached outside of the nuclear surface. To exclude this possibility, we processed the IF images for three-dimensional visualization to confirm intranuclear localization of protein VII signal (Figure 4C). Furthermore, we conducted biochemical fractionation to isolate nuclei and measured AdV DNA therein. This assay also showed an approximately 2-fold decrease of nuclear AdV DNA amount upon nc886 KO (Figure 4D). All these data suggested that nuclear import of the AdV genome was attenuated without nc886, which resulted in fewer DNA templates for E1A transcription.

Figure 4.

nc886 promotes nuclear entry of AdV DNA but not cellular entry

(A and B) IF of AdV protein VII (green), together with nuclei staining with DAPI (blue) at indicated times after infection of AdV at 50 MOI. Representative images are in (A) and quantitation is in (B). Signals were counted by using the Zen program (Carl Zeiss; Oberkochen, Germany). Each value is an average of three independent experiments (n = 15–18 cells examined for each experiment). Standard deviations and p values are indicated. (C) IF images of (A) were processed through the MATrix LABoratory (MATLAB) software for three-dimensional visualization. (D) qPCR of AdV DNA in isolated nuclei, at 4 h after AdV infection at 50 MOI. qPCR Ct values of nuclear AdV were normalized to those of corresponding cytoplasmic AdV. (E and F) qPCR of DNA isolated from attached and internalized AdV. qPCR Ct values were normalized to RT-qPCR Ct values of 18S rDNA.

nc886 does not affect AdV entry into cells

nc886 most likely took effect at a step before AdV nuclear import. Until AdV particles reach the nucleus, they undergo multiple interactions with host factors. The AdV journey to the nucleus can be briefly described as follows. The initial attachment to cell surface receptors leads to clathrin-mediated endocytosis. After escape from endosome, AdV particles associate with dynein motor proteins to travel along microtubules to the nuclear pore complex (reviewed in Greger and Flatt³³). Among host factors related to above processes, we interrogated whether any of them was differentially expressed between nc886-expressing and -deficient cells, by performing nCounter analysis, which provided expression data on a set of mRNAs and proteins from a nanoString panel (Figure S2). In addition, we looked into our previous Illumina microarray data.²⁹ These analyses identified several candidate genes and pathways including epidermal growth factor receptor (EGFR), Rac Family Small GTPase 1 (RAC1), and the kinesin pathway (shown later in Figure 5). EGFR and RAC1 are implicated in uptake of several viruses^4,34; kinesin proteins implicated in AdV cytoplasmic trafficking.³⁵

Figure 5.

The elevated expression of kinesins is the reason for inefficient AdV replication in nc886 KO cells

(A) Nomenclature of comparison pairs for analysis of the Reactome pathway (B) and gene expression (C). (B) A scatterplot of Z-scores for 674 Reactome gene sets (see also Table S1) showing the correlation between two pairs. The pathway “KINESINS” is marked by a larger red square. (C) A heatmap depicting relative expression of 30 kinesin family genes selected from a total of 61 kinesin family members (see Table S2 and legend for the selection criterion). (D) KD of kinesin proteins. A mixture of siRNAs against KIF20A and KIF22 (40 nM each) was transfected into nc886 KO cells for 12 h, and then followed by AdV infection for 12 h for replicated AdV DNA measurement and 4 h for AdV DNA that entered to nuclei. Graphs are as follows: qPCR to measure replicated AdV DNA (top panel, the same experiment as Figure 3B), qPCR to measure AdV DNA that entered nuclei (middle panel, the same experiment as Figure 4D), and RT-qPCR for KIF20A and KIF22 expression to assess KD efficiency (bottom two panels).

We continued to trace stages of AdV infection retrospectively, by performing two experiments: attachment assay and internalization assay.³⁶ In the attachment assay, the mixture of AdV particles was kept for a short time at 4°C, in which condition AdV internalization was mostly blocked and thus we could measure the degree of AdV binding to the cell surface. The internalization assay measures only AdV particles inside cells, as there were multiple thorough washing steps to eliminate AdV on the cell surface. The amount of AdV particles on the cell surface was similar between nc886 WT and KO cells (Figure 4E). In line with this result, cell surface receptors were expressed both in nc886-expressing and KO cells to a comparable degree (Figure S3).

Since EGFR and RAC1 were one of the top suppressed genes upon nc886 KO (Figure S2), we expected that AdV entry to cells was defective in the absence of nc886. Against our expectation, our internalization assay showed no significant difference in the amount of internalized AdV particles (Figure 4F). This result was corroborated by IF of hexon, the major capsid protein, whose intracellular signal was comparable between nc886 WT and KO cells (Figure S4). Furthermore, inhibition of EGFR and RAC1 by small interfering RNA (siRNA) or chemical inhibitors did not decrease AdV replication in Nthy-ori 3-1 nc886 WT cells (E.S. and Y.S.L., unpublished data). Collectively, we concluded that nc886 did not affect AdV cellular entry.

The suppression of the kinesin pathway is the reason for the pro-AdV effect of nc886

We performed Reactome pathway analysis in the Molecular Signature Database (MSigDB; https://www.gsea-msigdb.org/gsea/msigdb/). Reactome is composed of 674 pathway sets and each set contains several tens to hundreds of genes whose overall change of their relative expression levels is expressed in Z-scores. Positive or negative Z-scores indicated whether a pathway is activated or suppressed, respectively. From pairwise comparisons of gene expression data, we obtained 674 Z-scores from “change upon nc886^KO only” and “change upon both PKR/nc886^KO” and plotted them (Figures 5A and 5B, Table S1). Most notably, the kinesin pathway (“KINESINS”) was one of the most up-regulated ones in both the two pairwise comparisons (Figure 5B). It is also worth mentioning that Reactome Z-scores between “change upon nc886^KO only” and “change upon both PKR/nc886^KO” were positively correlated (Pearson’s R value = +0.843, Figure 5B). Such a high correlation indicated that nc886 posed similar impact on gene expression regardless of whether the PKR status was WT or KO and that sole PKR KO had very modest impact on gene expression. Actually in our previous report, PKR KO per se did not result in any recognizable phenotypic change in an unstressed growing condition.²⁹

Kinesins are a family of motor proteins moving along microtubules.³⁷ They move from the center of a cell to its periphery (called “plus end transport”), as opposed to dynein that is responsible for “minus end transport” of AdV. A recent work reported that KD of a handful of kinesin proteins leads to increase of AdV in the perinuclear region, implying antagonistic action of kinesins in AdV nuclear trafficking.³⁵ Thus, we hypothesized that the increase of a kinesin protein(s) was the reason why AdV replication was inefficient in nc886 KO cells. There are 61 genes in the kinesin family (Table S2). After filtering out 31 kinesin genes whose expression levels are negligible, the expression change of 30 kinesin genes were displayed in a heatmap (Figure 5C and Table S2). Based on their expression levels depending on nc886 as well as KD phenotypes in previous literature,³⁵ we selected six kinesin genes for RT-qPCR validation. The expression of four kinesin genes (KIF5C, KIF20A, KIF22, and KIF23) was significantly increased in PKR^KO:nc886^KO cells as compared with PKR^WT:nc886^WT cells, whereas KIF4A and KIF4B did not change significantly (Figure S5). Among the four, we chose KIF20A and KIF22 for the next KD experiments, because they were expressed more abundantly or increased more remarkably than the other two. Transfection of siRNAs against KIF20A and KIF22 (in one mixture) efficiently decreased their expression and, importantly, rescued the AdV nuclear entry and replication in nc886 KO cells to a significant degree (Figure 5D). Conclusively, suppression of kinesin motor proteins, particularly KIF20A and KIF22, and consequent facilitation of AdV trafficking to the nucleus were the mechanism how nc886 played a stimulating role in the AdV life cycle.

nc886, which is conserved in most primates and some other mammals, promotes AdV when expressed in mouse cells

There is a notion that replication of human AdV is restricted in human cells.⁸ However, several reports showed replication of human AdV in non-primate species such as pig, tree shrew, and dog (Jogler et al¹⁶ Li et al³⁸ Ternovoi et al³⁹; see Figure 6A and Table S3). In addition, human AdV appears to replicate in non-human primates. A number of AdV isolates from several primates are highly homologous in sequence, indicating the zoonotic transmission of AdV among primate species such as human, chimpanzee, gorilla, baboon, monkey, and macaque (Wevers et al⁴⁰ Roy et al⁴¹ Medkour et al⁴²; see Figure 6A).

Figure 6.

nc886 promotes AdV gene expression and replication when ectopically expressed in a mouse cell line

(A) A phylogenetic tree of a collection of mammalian species whose information on AdV permissiveness and nc886 status are available. AdV-permissive species and non-permissive species are distinguished by blue and red letters respectively. Information on the nc886 gene in nucleotides is indicated in square bracket in this way [length of the nc886 gene in indicated species, the number of matched nucleotides to the human nc886]. No conservation of the nc886 gene is designated as [-, -]. (B) The mouse genomic locus spanning Tfgbi and Smad5 genes. The cartoon has been edited from the UCSC genome browser view and the IGV sequence view, with addition of informative captions. The sequence view portion is blank, indicative of no sequence read captured (see Figure S6 for comparison). (C–E) qPCR of intracellular replicated AdV DNA (C), northern of VAI RNA and 5S rRNA for equal loading (D), and qPCR of released AdV DNA (E), at 24 h post-infection of AdV at indicated MOIs. All other descriptions are the same as Figures 3B, 3C, and 2B.

The nc886 gene is evolutionarily conserved in most primates and several non-primate species in the class Mammalia (Stadler et al⁴³; Figure 6A and Table S3). We surveyed existence of the nc886 gene in the aforementioned animal species and compared with their permissiveness to human AdV. Albeit with few exceptions, such as pigs and guinea pigs, those possessing the nc886 gene mostly tended to be permissive to human AdV (Figure 6A).

The mouse genome does not seem to have the nc886 gene, according to a previous report.⁴³ In that report, the authors searched the nc886 gene in various animal species based on sequence homology to vault RNAs (vtRNAs) and classified nc886 (also known as vtRNA2-1) as one of them. Nonetheless, we clearly demonstrated that nc886 is functionally distinct from canonical vtRNAs (vtRNA1-1, 1–2, and 1–3 in humans).¹⁷ In all animal species where the nc886 gene exists, it is located between TGFBI and SMAD5. Stadler et al failed to find a sequence homologous to vtRNAs in the mouse genomic region spanning Tgfbi and Smad5.⁴³ Since the sequence-based in silico search was so far the only evidence for the absence of nc886 in mice, we wanted to ascertain this by looking into high-throughput RNA-sequencing (RNA-seq) data. No RNA-seq read was captured in this region (Figures 6B and S6A). For comparison, we could detect RNA-seq reads for Rpph1 and Vaultrc5 (the murine ortholog of canonical vtRNAs), which are both transcribed by Pol III (Figures S6B and S6C). Furthermore, we rummaged the Tgfbi-Smad5 region but could not see any vestige of a Pol III gene, such as an A/B box and an oligo-T stretch, which are the promoter and termination elements for Pol III transcription. Collectively, we reassured that nc886 was not conserved in mice.

Several studies showed that mouse cells do not support the complete life cycle of human AdV.13, 14, 15, 16 This impeded the development of a mouse model to evaluate the utility of human AdV in vivo. We questioned whether introduction of nc886 into mouse cells promotes the propagation of human AdV. We had generated an nc886-expressing derivative from RAW 264.7 mouse macrophage cell line (“RAW:nc886” and a control cell line “RAW:vector”²⁶). We infected AdV into these cell lines and found AdV DNA replication, gene expression, and released AdV amount to be elevated when nc886 was expressed (Figures 6C–6E). These data indicated that nc886 promotes human AdV replication and gene expression, when xenogeneically expressed in mice. Among many differences between humans and mice, nc886 could be one reason why mouse cells do not support human AdV replication.

Discussion

In this study we identified nc886 to be a crucial factor in the AdV infection cycle. Albeit a host ncRNA, nc886 helped AdV replication and gene expression, by facilitating the nuclear entry of AdV (Figure 7). In addition, we demonstrated that PKR was not essential for the stimulatory role of nc886 on AdV. Our data also showed that PKR deficiency per se barely took effect on AdV.

Figure 7.

Cartoon depicting the role of nc886 in the AdV life cycle

Our study has discovered a novel mechanism for a host ncRNA to promote a virus. A host ncRNA confers a milieu for favorable trafficking to the nucleus upon AdV. When we looked into the magnitude of the impact of nc886, the beginning was weak but the end was prosperous. When nc886 was absent, the nuclear import of AdV and its direct consequence, transcription of the immediate-early gene, was only about 2-fold lower. In comparison, AdV DNA replication, and subsequent events accordingly, were almost abrogated in nc886-deficient cells. One possibility is that the initial modest difference has been amplified, similar to the butterfly effect. Alternatively, nc886 might provide AdV with additional favor at a step(s) from transcription of early genes until AdV DNA replication.

Our finding that PKR KO had barely an effect on AdV propagation was against initial anticipation, given the established role of PKR in antiviral responses. During the AdV infection cycle, PKR activity displays two waves of activation and repression, one at a very early time point (∼90 min) and the other at the time point of DNA replication (∼18 h).⁴⁴ It has been shown that PKR activity is suppressed by VA RNAs^45,46 as well as by AdV E1B-55K and E4orf6 proteins.⁴⁴ During in vitro infection at high AdV dose, those PKR-inhibitory genes will be expressed at a sufficient level to well suppress PKR activity. This might be a reason why AdV infection was equally efficient regardless of PKR WT or KO. We also speculate that PKR suppression by nc886 might be needed in natural situations when a limited amount of AdV infects a cell.

An interesting observation was deprivation of nc886 by AdV, whose significance deserves further discussion in the point of PKR regulation, although the stimulatory effect of nc886 on AdV was independent of PKR. nc886 deprivation might be the genuine trigger of the late activation of PKR at ∼18 h. Despite the frustration of PKR activation by AdV genes, nc886 deprivation might still be another device for cells to evoke innate immune responses. Besides PKR, nc886 inhibits Interferon Regulatory Factor 3 (IRF3) via Mitochondrial AntiViral-Signaling protein (MAVS) to suppress the interferon response.²⁶

Our finding is of clinical importance in several aspects. nc886 is transcribed by Pol III. During tumorigenesis, Pol III activity is generally elevated and accordingly nc886 expression is usually higher in cancer cells than normal quiescent cells.¹⁸ However, nc886 expression is epigenetically silenced in a subset of cancer cells⁴⁷ or declines abruptly during chemotherapy.⁴⁸ Our key finding here, that AdV replication and the cytotoxicity were inefficient in the absence of nc886, provides an explanation of why oncolytic virotherapy is not effective for all cancer patients.¹² Our study warrants a need to determine the nc886 status, when designing oncolytic virotherapy for a patient. In case of nc886-silenced tumor cells, co-administration of nc886 might improve the oncolytic efficacy of AdV. In-depth understanding of the interplay between AdV genes and nc886 (and related pathways) should precede harnessing nc886 or engineering AdV for therapeutic purposes.

Another significance of our study lies in the mouse cell data (Figure 6). Inefficient replication of human AdV in mice and other animal species hampered the development of an animal model for preclinical studies in which safety and efficacy of a therapeutically engineered AdV are to be evaluated. Since we found that the pro-AdV role of nc886 operates when xenogeneically expressed, construction of an nc886-expressing transgenic animal would be a choice when we are to ameliorate an in vivo model system for AdV.

Although AdV is renowned for its clinical utility, the natural pathology of AdV cannot be ignored. Although infection of naturally occurring AdV usually causes mild symptoms and is self-limiting, it can be fatal to immune-deficient individuals. Yet, there is no effective drug for the treatment of AdV infection until now (reviewed in Hendrickx et al⁴⁹). Our data suggest that inhibition of nc886 by administration of an oligonucleotide targeting nc886 might provide a treatment option.

Materials and methods

Cell lines, AdVs, antibodies, and other reagents

Cell lines were purchased from American Type Culture Collection (ATCC; Manassas, VA) or our laboratory stock. Cells were tested for mycoplasma contamination at the Genomics Core in the National Cancer Center, Korea, and were confirmed to be devoid of it. The plasmid “pLL3.7.Puro.U6” was modified from a lentiviral vector pLL3.7 (Addgene; Watertown, MA) by replacing GFP with the puromycin-resistance gene. We constructed an nc886-expressing plasmid by inserting a 102-nucleotide long DNA fragment corresponding to the nc886 RNA region¹⁷ into lentiviral vector pLL3.7.puro.U6. This construct and pLL3.7.puro.U6 were used to construct nc886-expressing and control cell lines from original UM-SCC-19 and RWPE-1. Preparation of lentiviruses, infection onto cells, and subsequent steps to isolate cell clones were according to standard laboratory protocols. Nthy-ori 3-1 WT and KO cell lines are described in Lee et al²⁹ nc886-expressing and control cell lines derived from RAW264.7 (designated “RAW:nc886” and “RAW:vector”) are described in Lee et al.²⁶

AdV5 is a WT human AdV (species C serotype 5, accession number AY339865.1)⁵⁰ and was obtained from ATCC (VR-1516). Ad5/35CMV-GFP was our laboratory stock and constructed as described in Do et al.³⁰ AdV virions were propagated in HEK293 cells and purified by using Adeno-X Maxi Purification Kit (Takara Bio USA, Inc., Mountain View, CA) or by the CsCl method as previously described.⁵⁰ Our standard procedure for AdV infection was as follows: addition of AdV at indicated multiplicity of infection (MOI) in serum-free medium, incubation for 2 h, and then replacement with medium containing 10% fetal bovine serum.

Hexon and GAPDH antibodies were purchased from Merck Millipore (Burlington, MA) and Cell Signaling Technology, Inc. (Danvers, MA). Total PKR antibody was from Abcam (Cambridge, MA). AdV protein VII antibody was a mouse monoclonal antibody generated and kindly provided by Wodrich laboratory.⁵¹

AdV DNA and RNA measurement

Total RNA from cells was isolated by Trizol reagent (Life Technologies, Carlsbad, CA). Northern hybridization was done as previously described.¹⁷ cDNA was synthesized by amfiRivert kit (GenDEPOT; Barker, TX) and real-time PCR was done with LightCycler 480 SYBR Green I MasterMix (Roche; Penzberg, Germany) and LightCycler 480 Instrument II (Roche). When AdV mRNAs and pre-mRNAs were measured, two tactics were employed to ensure that the PCR amplification was from RNA but not from AdV DNA. First, prepared RNA was treated with DNase I (New England Biolabs; Ipswich, MA). Second, a reaction without reverse transcriptase (“no RT reaction”) was done in parallel with a cDNA synthesis reaction (“+ RT reaction”). A PCR value from “no RT reaction” was used as a baseline for the corresponding value of “+ RT reaction.” AdV DNA within infected cells was measured by qPCR directly on isolated nucleic acids by Trizol without cDNA synthesis. AdV DNA in AdV virions released from cells was measured by qPCR of 3 μL of culture supernatant, which had been 8-fold diluted in Tris-EDTA buffer. In all graphs for AdV DNA PCR data, we displayed 2^−Ct values relative to uninfected samples, for which sampling and qPCR were performed in parallel. Primer and probe sequences are summarized in Table S4.

Cell viability assay

We assessed the cytopathic effect of AdV on infected cells as performed in O-Carroll et al⁵²: 1 × 10⁴ cells were seeded in a 96-well plate and incubated overnight. Cells were infected with AdV at indicated MOIs for 72 h and then floating cells were removed by aspiration. The protein content of attached cells was measured by adding 100 μL of Pierce BCA protein assay reagent (Thermo Fisher Scientific, Waltham, MA) to each well. The subsequent steps were according to the manufacturer's instructions. The protein content was measured by the absorbance at 570 nm. We performed the same BCA assay with 96 wells containing serially diluted uninfected cells, to plot a standard curve from which we converted an absorbance value to a cell density.

IF for AdV protein VII

Cells were cultured on a 12-well plate containing coverslips (18-mm diameter) coated with poly-L-lysine. Cells were infected with AdV at 50 MOI for the indicated hours and were fixed for 10 min with 4% formaldehyde in PBS. Fixed cells were exposed for 30 min at room temperature to PBS containing 0.1% Triton X-100 and 5% horse serum (Gibco-BRL, Thermo Fisher Scientific), and were incubated overnight at 4°C with primary antibody in the same solution, washed three times with PBS, and were incubated for 30 min at room temperature with Alexa Fluor 488–conjugated secondary antibody (Molecular Probes, Thermo Fisher Scientific). Cells were also treated with DAPI (0.2 μg/mL) to stain nuclei.

Assays for AdV attachment to cells, internalization, and nuclear entry

AdV particles that were attached on the cell surface and had entered within cells were measured by “attachment assay” and “internalization assay,” respectively; 2 × 10⁵ cells were seeded in a 6-well plate and incubated overnight. Cells were incubated at 4°C for 30 min prior to infection, then infected with AdV at 10 MOI, and further incubated at 4°C for 1 h. After washing three times with ice-cold PBS, cells in a well were harvested by scrapping for attachment assay. Cells in another well were further proceeded for internalization assay, by incubation at 37°C in fresh medium for 1 h. Cells were washed with PBS three times and harvested by trypsinization. Harvested cell pellets were again washed three times with PBS. These extensive wash steps were to remove AdV particles attached on the cell surface. DNA was isolated from harvested cells, by DNeasy Blood & Tissue Kit (Qiagen, Germantown, MD). AdV DNA amount was quantified by qPCR and normalized to qPCR values of cellular 18S rDNA.

Subcellular fractionation was performed to isolate nuclei from cytosolic fractions, as described in Lee et al,¹⁷ with minor modifications at a washing step and buffer composition. After cell lysis, nuclear pellets were washed twice (once in the original protocol) to ensure complete elimination of contaminating cytoplasmic fraction and AdV particles on the nuclear surface. We also omitted RNase inhibitors from all buffers, since they were dispensable for AdV DNA measurement.

Antisense oligonucleotides, siRNA, and transfection

Modified antisense oligonucleotides (anti-oligos) for nc886 KD experiments (“anti-nc886” and “anti-control”) in this study were the same as our previous study.^17,26 siRNAs targeting kinesins were Stealth RNAi siRNA from Invitrogen (Carlsbad, CA). Their targeting sequences are 5′-cagugcaaagcagagcuaaacucua-3′ (for KIF20A) and 5′-cgacgcagcagaggcgacgcgagau-3′ (for KIF22). Small RNAs, at 100 nM (for anti-oligos) or 40 nM (for siRNAs), were transfected using Lipofectamine RNAiMAX Reagent (Invitrogen) per the manufacturer's instruction.

Statistical analysis

Unless specified otherwise in the figure legends, statistical significance in most experiments (PCR, GFP, and IF quantification, etc.) was indicated by p values that were calculated from triplicate samples by using unpaired Student's t test.