Epstein-Barr virus-encoded RNAs (EBERs) complement the loss of Herpesvirus telomerase RNA (vTR) in virus-induced tumor formation

Marek’s disease virus (MDV) is an alphaherpesvirus that causes fatal lymphomas in chickens and is used as a natural virus-host model for herpesvirus-induced tumorigenesis. MDV encodes a telomerase RNA subunit (vTR) that is crucial for efficient MDV-induced lymphoma formation; however, the mechanism is not completely understood. Similarly, Epstein Barr-virus (EBV) encodes two RNAs (EBER-1 and EBER-2) that are highly expressed in EBV-induced tumor cells, however their role in tumorigenesis remains unclear. Intriguingly, vTR and EBER-1 have interaction partners in common that are highly conserved in humans and chickens. Therefore, we investigated if EBER-1 and/or EBER-2 can complement the loss of vTR in MDV-induced tumor formation. We first deleted vTR (v∆vTR) and replaced it by either EBER-1 or EBER-2 in the very virulent RB-1B strain. Insertion of either EBER-1 or EBER-2 did not affect MDV replication and their expression levels were comparable to vTR in wild type virus. Intriguingly, EBER-2 restored tumor formation of MDV that lacks vTR. EBER-1 partially restored MDV oncogenicity, while tumor formation was severely impaired in chickens infected with v∆vTR. Our data provides the first evidence that EBERs possess tumor-promoting properties in vivo using this natural model for herpesvirus-tumorigenesis.

antigen (La) 22 and the double-stranded-RNA-activated protein kinase (PKR) 23,24 . EBER-2 also interacts with La 22 as well as the transcription factor paired box protein 5 (PAX-5) 25 . Intriguingly, these factors are all conserved between humans and chickens.
In the current study, we investigated if EBER-1 and/or EBER-2 can complement the loss of vTR in MDV-induced tumor formation. We generated recombinant MDV-viruses that lack vTR and encode either EBER-1 or EBER-2 instead. Analysis of their replication properties in vitro and in vivo revealed that neither deletion of the entire vTR nor insertion of EBERs affects MDV replication. Deletion of the vTR severely impaired tumor formation. Intriguingly, expression of EBER-2 efficiently restored tumor formation, while EBER-1 only partially complemented the loss of vTR. Our study provides the first evidence that EBERs possess tumor-promoting effects in vivo using this natural animal model for herpesvirus-induced tumor formation.

Generation and characterization of the recombinant viruses in vitro.
To determine if EBERs can complement the loss of vTR, we generated recombinant viruses that encode either EBER-1 (vEBER-1) or EBER-2 (vEBER-2) instead of vTR (Fig. 1A). EBER-1 and EBER-2 were sequentially introduced into the RB-1B MDV strain lacking the entire vTR (vΔvTR) (Fig. 1A) using en passant mutagenesis. In addition, a revertant virus (vRev) was generated in which vTR was restored in the original locus. Mutants were analyzed by RFLP, Sanger and Illumina MiSeq sequencing (coverage > 1000-fold) to confirm that the entire virus genome is correct. To determine if deletion of vTR or insertion of EBERs affects viral replication, we assessed the replication of the recombinant viruses. Plaque size assays revealed that the recombinant viruses replicated comparable to wild type and revertant virus (Fig. 1B). We confirmed this observation using multi-step growth kinetics (Fig. 1C), highlighting that neither deletion of vTR nor insertion of EBERs alters MDV replication.

Recombinant viruses efficiently express EBERs.
To determine if the EBERs are efficiently expressed during MDV replication, we infected CECs with the wild type or recombinant viruses and performed qRT-PCR. As expected, vTR was only expressed in wild type and revertant virus, while no vTR expression was detected upon deletion of the vTR gene ( Fig. 2A). EBER-1 and EBER-2 were highly expressed in the corresponding recombinant viruses at copy numbers comparable to vTR in the wild type virus and revertant virus ( Fig. 2B and C). No significant difference was observed for the expression of the viral ICP4 or the cellular GAPDH genes ( Fig. 2D and E). Taken together, vEBER-1 and vEBER-2 efficiently expressed the expected EBER gene, while no vTR expression was detectable in both viruses.

EBERs complement the loss of vTR in MDV-induced tumor formation.
To determine if EBERs can complement the loss of vTR in MDV-induced tumor formation, we infected one-day old chickens subcutaneously with 2,000 PFU of vRB-1B, v∆vTR, vEBER-1, vEBER-2 or vRev and monitored the onset of clinical symptoms and tumor formation. To investigate if the recombinant viruses replicated efficiently in infected animals, we quantified viral genome copies in the blood by qPCR. Replication of vΔvTR, vEBER-1 and EBER-2 was not significantly altered compared to the wild type and the revertant virus (Fig. 3A), indicating that expression of the EBERs did not affect MDV replication in vivo.
During the course of infection, animals were monitored for the development of clinical symptoms and tumors. In the absence of vTR, tumor incidence was significantly reduced (p < 0.0125) (Fig. 3B) as described previously 11 . Intriguingly, the tumor incidence of vEBER-2 was comparable to the wild type and revertant virus indicating that the expression of EBER-2 completely restored tumor formation. Expression of EBER-1 only partially restored MDV-induced tumor formation, as vEBER-1 (40%) only showed a mild increase in tumor incidence compared to vΔvTR (28%; Fig. 3B). To confirm that the EBERs are efficiently expressed in MDV-induced tumor cells, we quantified EBER-1 and EBER-2 expression in tumor tissue by qRT-PCR. Both EBER-1 and EBER-2 were highly expressed and at comparable levels ( Supplementary Fig. 1). To elucidate the effect of EBERs expression in tumor dissemination, we determined the numbers of organs with gross tumors during necropsy. The average number of the tumors per animals was significantly reduced in the absence of vTR compared to wild type and revertant virus (Fig. 3C). Most importantly, EBER-2 expression also efficiently restored tumor dissemination. No significant difference in the average number of tumors was observed between the animals infected with vEBER-2 compared to the wild type or revertant virus. A partial restoration was observed for vEBER-1 when compared to v∆vTR. Taken together, our data demonstrates that the EBV-encoded-EBERs can either fully (EBER-2) or partially (EBER-1) complement the loss of vTR in MDV-induced tumor formation using this small animal model for herpesvirus-induced oncogenesis. Our study provides thereby the first evidence that EBERs possess tumor promoting function in vivo.

Discussion
vTR plays an important role in MDV-induced tumor formation, however the mechanism remains poorly understood. We recently demonstrated that vTR possesses tumor-promoting functions that are independent of its role in the telomerase activity 13 . The telomerase activity mediated by vTR only contributed to the rapid onset of tumors; however, tumor incidence and dissemination was not affected when incorporation of vTR into the telomerase complex was abrogated 13 . Therefore, vTR likely drives virus-induced transformation via the interaction with the ribosomal protein RpL22 and/or other cellular factors. Intriguingly, both vTR and EBER-1 interact and re-localize RpL22 13 , which is almost completely conserved between humans and chickens. Therefore, we set to determine if EBER-1 and/or EBER-2 can complement the loss of vTR in MDV-induced tumor formation.
We generated recombinant MDVs that express either EBER-1 or EBER-2 instead of vTR. Virus replication was not affected in vitro and in vivo, revealing that neither deletion of the entire vTR nor insertion of the EBERs affects MDV replication. Our data on the complete deletion of vTR is therefore consistent with the previously published partial deletions of the conserved regions (CR1-CR4) of vTR 11 . To confirm the efficient expression of EBER-1 and EBER-2, we performed qRT-PCR and we could demonstrate that EBERs were highly overexpressed. The observed expression levels of EBERs in MDV infected cells were also comparable to latently infected cells and EBV-induced cancers (>10 6 per cell) 16,26 . EBER expression levels were also similar to vTR in wild type virus and revertant virus due to the strong nature of the vTR promoter 12 . Expression levels of vTR, and likely also the EBERs, play a crucial role in the transformation process as viruses that expressed vTR at lower levels were severely impaired in tumor formation in vivo 27 .
To determine the effect of the complete deletion of vTR and if the EBERs can complement the loss of vTR, we infected SPF chickens with the recombinant viruses. As expected, deletion of the entire vTR severely attenuated MDV and is consistent with the partial deletion of the gene published previously 11 . Intriguingly, EBER-1 that also interacts and re-localizes RpL22 only partially restored MDV-induced tumor formation, suggesting that this interaction could indeed play a minor role in the cellular transformation. However, certainly also other interaction partners or mechanisms are responsible for vTR mediated tumor formation. Alternatively, the dysregulation of RpL22 could differ between EBER-1 and vTR, possibly due to differences in the binding affinity to the ribosomal protein as observed previously 11 . Surprisingly, EBER-2 expression efficiently restored MDV-induced tumor formation and metastasis of a virus that lacks vTR. Intriguingly, EBER-2 has been previously shown to inhibit apoptosis 28,29 and increase cell-proliferation 30 , which could contribute to the increased tumor incidence of the EBER-2 expressing virus. It remains unknown which interaction partners of EBER-2 mediate these effects and if they are conserved between humans and chickens as La and PAX-5. We will address these aspects and if conserved stem loop structures in EBER-1, EBER-2 and vTR ( Supplementary Fig. 2) mediate these functions in future studies.
Taken together, our data demonstrate that EBER-1 and EBER-2 possess tumor promoting activity that can complement the activity of vTR in MDV-induced transformation. Future studies will focus on the conserved interaction partners and possible mechanism(s) for EBER mediated transformation using this natural virus-host animal model for herpesvirus induced tumor formation.

Generation of recombinant viruses. Recombinant viruses encoding EBER-1 or EBER-2 instead of vTR
were generated using a bacterial artificial chromosome (BAC) of the very virulent MDV strain RB-1B that lacks most of the internal repeat long region (IRL; pRB-1B∆IRL) 7 , which is rapidly restored upon virus reconstitution. Therefore, only one copy of vTR region had to be manipulated by two-step Red-mediated mutagenesis as described previously 32,33 , while the resulting recombinant virus contained the substitution/deletion in both loci 7 . First, we deleted the entire vTR, then sequentially introduced either EBER-1 (vEBER-1) or EBER-2 (vEBER-2) of the B95-8 EBV-strain (RefSeq M80517.1), allowing EBER expression under control of the native vTR promoter. In addition, a revertant virus (vRev) was generated in which the original vTR locus restored. Primers used for mutagenesis are listed in Table 1. Recombinant BAC clones were confirmed by RFLP, PCR and Sanger sequencing of the target area ( Supplementary Fig. 3). In addition, we performed Illumina MiSeq sequencing to ensure that the entire nucleotide sequence of the constructs is correct. Recombinant viruses were reconstituted by transfection of CECs with BAC DNA as described previously 7,34 . instruction. Samples were treated with DNase I (Promega) and cDNA generated using the high Capacity cDNA Reverse Transcription Kit (Applied Biosystems). vTR and EBER expression levels in the corresponding viruses were normalized to the expression levels of viral ICP4 and cellular GAPDH genes. Primers and probes used for qRT-PCR are shown in Table 1.

Quantification of vTR and EBERs expression. vTR and EBER
Plaque size assays and growth kinetics. Virus replication and spread was determined by plaque size assays and multi-step growth kinetics as described previously 35 . For plaque size assays, at least 50 randomly selected plaques were captured and plaque areas were determined using Image J software (NIH). Significant difference in plaque diameters was evaluated by One-way analysis of variance (ANOVA).

Ethics statement and in vivo experiments.
All animal work was conducted according to relevant national and international guidelines for humane use of animals. Animal experiments were approved by the Landesamt für Gesundheit und Soziales (LAGeSo) in Berlin (approval number G0218/12). One-day old specific pathogen free (SPF) chickens (ValoBioMedia) were randomly assigned into four groups. Animals were infected subcutaneously with 2000 PFU of either wild type vRB-1B (n = 9), v∆vTR (n = 25), vEBER-1 (n = 25), vEBER-2 (n = 23) or the revertant virus vRev (n = 24). Peripheral blood samples were collected from the infected chickens at 4, 7, 10, 14, 21 and 28 dpi to determine MDV genome copy numbers in the blood, as described previously 34,36 . Chickens were monitored for clinical symptoms of MD on a daily basis throughout the 91 days of the experiment. To eliminate bias, the animal experiment was performed in a blinded manner until all data was collected and evaluated to avoid subjectivity. Animals were euthanized and examined for tumor lesions either once clinical symptoms were evident or after termination of the experiment. To confirm the presence of the introduced mutations in the virus genome, DNA was extracted from tumor tissue and the target region analyzed by Sanger sequencing.
Quantification of MDV genome copies. DNA was extracted from the blood of the infected chickens using the E-Z96 blood DNA kit (OMEGA biotek, USA) following the manufacturer's instructions. MDV genome copies were determined by quantitative PCR (qPCR) using specific primers and a probe for the MDV ICP4 gene (Table 1) 37,38 . ICP4 copy numbers were normalized to cellular inducible nitric oxide synthase (iNOS) gene as described previously 39 . Statistical analyses. The statistical analyses were performed using Graph-Pad Prism v7 and the SPSS software (SPSS, Inc). Plaque size assays were analyzed using one-way analysis of variance (ANOVA). MDV genome qPCR data were analyzed using the Kruskal-Wallis test. Data sets were first tested for normal distribution and results were considered significant when p < 0.05. Animal experiment data was analyzed by Fisher's exact test, with a Bonferroni correction for multiple comparisons and results were considered significant when p < 0.0125.