Marek's Disease Virus-Encoded MicroRNA 155 Ortholog Critical for the Induction of Lymphomas Is Not Essential for the Proliferation of Transformed Cell Lines

Marek’s disease virus (MDV) is an alphaherpesvirus associated with Marek’s disease (MD), a highly contagious neoplastic disease of chickens. MD serves as an excellent model for studying virus-induced T-cell lymphomas in the natural chicken hosts. Among the limited set of genes associated with MD oncogenicity, MDV-miR-M4, a highly expressed viral ortholog of the oncogenic miR-155, has received extensive attention due to its direct role in the induction of lymphomas. Using a targeted CRISPR-Cas9-based gene editing approach in MDV-transformed lymphoblastoid cell lines, we show that MDV-miR-M4, despite its critical role in the induction of tumors, is not essential for maintaining the transformed phenotype and continuous proliferation. As far as we know, this was the first study in which precise editing of an oncogenic miRNA was carried out in situ in MD lymphoma-derived cell lines to demonstrate that it is not essential in maintaining the transformed phenotype.

phoblastoid cell lines, we show that MDV-miR-M4, despite its critical role in the induction of tumors, is not essential for maintaining the transformed phenotype and continuous proliferation. As far as we know, this was the first study in which precise editing of an oncogenic miRNA was carried out in situ in MD lymphoma-derived cell lines to demonstrate that it is not essential in maintaining the transformed phenotype.
KEYWORDS CRISPR/Cas9 editing, Marek's disease virus, microRNA, oncogenesis, transformation M icroRNAs (miRNAs) are ϳ22-nucleotide (nt) small RNA molecules that function as master regulators of gene expression in many species, including plants, worms, flies, and animals, as well as in a number of viruses. Most of the virus-encoded miRNAs are seen in DNA viruses, with members of the family Herpesviridae accounting for the vast majority demonstrating the significance of miRNA-mediated gene regulation in the biology of herpesvirus infection (1)(2)(3). Identification of miRNAs encoded by human oncogenic gammaherpesviruses such as Kaposi's sarcoma-associated herpesvirus (KSHV) and Epstein-Barr virus (EBV) as well as avian oncogenic alphaherpesvirus Marek's disease virus (MDV) has highlighted the potential contribution of the virus-encoded miRNAs to the oncogenicity of these viruses. Among the several roles of the herpesvirus-encoded miRNAs, such as immune evasion and control of viral latency/lytic replication and oncogenic potential (4)(5)(6), the role of viral orthologs of host microRNA 155 (miR-155) encoded by KSHV and MDV in oncogenesis has been the most extensively studied (5,7). As a multifunctional miRNA expressed primarily in the hematopoietic cells and in cells of the immune systems, microRNA 155 (miR-155) is highly conserved in most species, including humans and chickens, and is associated with different lymphomas (8)(9)(10)(11). In EBV-induced B-cell transformation as well as in a number of EBV-associated B-cell lymphomas, including Hodgkin's lymphoma, diffuse large B-cell lymphoma (DLBCL), and Burkitt's lymphoma in humans, upregulation of miR-155 resulting in escalated cell proliferation and neoplastic transformation has been reported (12,13). KSHV, a human gammaherpesvirus associated with lymphoproliferative disorders such as primary effusion lymphoma (PEL), multicentric Castleman disease (MCD), and B lymphomagenesis in AIDS patients, encodes 25 miRNAs. Among these miRNAs, KSHV-K12-11, which plays critical role in pathogenesis, is a functional ortholog of hsa-miR-155 sharing identical seed sequences (14)(15)(16). MDV encodes MDV-miR-M4-5p (miR-M4), a functional ortholog with seed sequences identical to those of miR-155 and KSHV-K12-11 that has been shown to play a critical role in the induction of lymphomas (6).
Marek's disease (MD) is a lymphoproliferative disease of chickens characterized by rapid-onset lymphomas in multiple organs and by infiltration into peripheral nerves, causing paralysis. MD serves as an excellent model for studying virus-induced T-cell lymphomas. Among the more than 100 genes carried by the MDV (17,18), the gene encoding the basic leucine zipper protein Meq (MDV EcoRI Q), which is undisputedly expressed in both lytic and latent infections, is the most important viral gene associated with MD oncogenicity (19,20). Deletion of the Meq gene or inhibition of its important interactions with host proteins such as c-Jun, c-Fos, and C-terminal binding protein (CtBP) can affect the oncogenicity of the virus (21)(22)(23). Although the viral telomerase RNA (vTR) also has been shown to promote MDV-induced oncogenesis (24), the role of MDV-encoded miRNAs in oncogenesis has drawn extensive attention (25)(26)(27). MDV encodes 14 miRNA precursors producing 26 mature miRNAs which are clustered into three separate genomic loci within the repeat regions of the viral genome. MDV-miR-M4, located in cluster 1, was shown previously to be the viral ortholog of miR-155 (28). The oncogenic properties of miR-155, together with the observation of a high level of miR-M4 expression in tumor cells and the identification of several cancer pathwayrelated target genes, suggested the important role of this miRNA in MDV-induced oncogenesis. Indeed, we and others have previously demonstrated the direct role of miR-M4 in the induction of tumors using recombinant MDV engineered to have deletion mutations or seed region mutations in miR-M4 by the use of in vivo experiments in chickens (6,29). Furthermore, we showed previously that the loss of the oncogenic phenotype of a miR-M4 deletion mutant of MDV can be partially rescued by MDV expressing gga-miR-155, demonstrating the similarities in the functions of the two orthologs (6). While the role of miR-M4 in the induction of MD lymphomas has been clearly demonstrated in these studies, it remains unclear whether continued high-level expression of miR-M4 is essential for maintaining the transformed phenotype of MDV-transformed tumor cells. As clonal populations of transformed tumor cells with latent MDV genome and limited gene expression (30)(31)(32), lymphoblastoid cell lines (LCL) derived from MD lymphomas have served as valuable resources to improve understanding of distinct aspects of virus-host interactions in transformed cells. However, detailed investigations into the role of different viral and host determinants in these cells have been difficult due to the lack of tools for manipulation of viral/host genomes of these cells in situ. Following our recent success in efficient editing of the MDV genome in cell culture systems that support lytic virus replication in vitro (33), we explored the use of a gene editing approach in an MDCC-HP8 cell line that is latently infected with the GA strain of MDV. Using MDCC-HP8 cells that stably expressed Cas9 and synthetic guide RNAs (gRNAs) with a two-part guide RNA system, we examined the effect of deletion of miR-M4 to gain insights into its functional role. Continued proliferation of the miR-M4 knockout cell lines suggested that expression of miR-M4 gene is not essential for maintenance of the transformed state of the MDCC-HP8 tumor cell line, despite its known critical role in the induction of MD lymphomas.

RESULTS
Knockout of MDV-miR-M4 in HP8 cells. On the basis of our success in efficient editing of the MDV genome during lytic replication in infected chicken embryo fibroblast (CEF) cultures in our previous studies (33), we attempted editing of the latent MDV genome in virus-transformed cell lines. The initial attempt performed with a transfected gRNA-Cas9 expression plasmid showed low editing efficiency, thought to be largely due to the relatively low transfection efficiency of the hard-to-transfect MDV-transformed cell lines (data not shown). A new gene editing strategy involving the transfection of synthetic gRNAs with a two-part guide RNA system into a MDVtransformed cell line stably expressing Cas9 (HP8-Cas9) showed great success. For the targeted editing of MDV-miR-M4 in the latent viral genome in this cell line, two gRNAs, M4-gN and M4-gC, were designed using CRISPR guide RNA designing software (http:// crispr.mit.edu/). M4-gN targeted the upstream sequence of the mature miR-M4 sequence, and M4-gC targeted the sequence spanning the mature miR-M4 sequence and the loop region of the pre-miRNA hairpin structure, resulting in the predicted cleavage site lying exactly at the end of the miR-M4 mature sequence ( Fig. 1a and b). Successful miR-M4 deletion would release a 54-nt fragment following the successful cleavage of the sequence by the two gRNAs. Considering the presence of several MDV genomes integrated in multiple chromosomes of the chicken genome determined on the basis of fluorescence in situ hybridization (FISH) analysis (unpublished data) and the location of miR-M4 in the terminal repeat region, which doubles the number of copies of miR-M4, two distinct bands are expected to be identified by the use of PCR tests on the genomic DNA from cells harvested 48 h after transfection and the use of specific primers located at the flanking region of Cas9 targeting sites. The top band of around 205 bp represented the unedited sequence or edited target site(s), with small indels present if the two sites were not cleaved simultaneously. The bottom, smaller band product of around 151 bp corresponded to the edited region with a 54-bp deletion between the two Cas9 cleavage sites. Interestingly, only the bottom band was detected by PCR analysis, indicating the highly efficient cleavage resulting from the use of the two gRNAs, with the majority of the cells transfected and edited efficiently. Despite the observation of only a single band, single-cell sorting was carried out to obtain a pure population of miR-M4-deleted cell. Although only the bottom band was obtained by PCR before sorting was performed in the mixed population, clones with the top band were predominant after single-cell cloning (Fig. 1c). Sequence analysis of four bottom bands confirmed that the results represented the direct end joining (EJ) product of two predicted Cas9 target sites (Fig. 1a). Interestingly, the sequences of all four clones were identical, suggesting that further screening of several additional clones might be required to identify variations within the edited sequences. The successful knockout of the miR-M4 sequence was further confirmed by reverse transcriptionquantitative PCR (qRT-PCR) analysis, using uninfected CEFs as a negative control. As expected, miR-M4 was absent from all four miR-M4-deleted HP8 clones and control CEF, contrasting with the high-level expression detected in the parental HP8-Cas9 cells (Fig. 1d). These experiments demonstrated that miR-M4 had been deleted successfully by the use of the two-part guide RNA system in the HP8 cell line stably expressing Cas9. miR-M4 is not essential in maintaining the transformed phenotype of MDVtransformed cell line. miR-M4 has been shown to be essential for the MDV in inducing tumors (6,29). To explore the role of miR-M4 in maintaining the transformed state, we examined the effect of deletion of miR-M4 on the proliferation of HP8 cells. For this, we carried out kinetic monitoring of proliferation of the wild-type HP8-Cas9 strain and the miR-M4-deleted clones using an IncuCyte S3 live-cell imaging system (Fig. 2). The cell proliferation data obtained in real time from the images collected at 4-h intervals showed that the miR-M4-deleted clones proliferated at a significantly higher rate within the first 3 days than the parental HP8-Cas9 cells, although different clones showed different levels of significance at various time points. These results suggested that expression of miR-M4 was not essential for the proliferation phenotype of these transformed cells.
Pu.1 is upregulated in HP8-⌬miR-M4 cells. Having shown that miR-M4 can be deleted from HP8 cell line and that it is not essential for the continued proliferation of the transformed cells, we wanted to examine the effect of miR-M4 deletion on expression of its target proteins. For this, we chose to analyze the expression levels of Pu.1, one of the very well characterized and validated miR-M4 targets (28). The expression levels were first assessed using a luciferase reporter assay by transfection of the reporter construct containing the wild-type predicted miR-M4 response element (MRE) or the mutant MRE region of the 3= untranslated region (UTR) of Pu.1 into the miR-M4-deleted and the parental HP8-Cas9 cells. This assay showed that the relative Renilla luciferase levels of reporter constructs with wild-type MRE sequences were reduced by nearly 40% compared with the levels seen with the mutant MRE construct in the parental HP8-Cas9 cells. Compared to this, such a reduction of luciferase levels was absent in all of the miR-M4-deleted clones (Fig. 3a), demonstrating the functional effect of miR-M4 deletion on the Pu.1 target. Next, we determined the miR-M4-mediated silencing by directly measuring the level of Pu.1 expression in one of the selected mutant clones, i.e., clone C48, along with the parental cells. Immunoprecipitation-Western blot analysis showed that the Pu.1 expression level was much higher in the miR-M4-deleted cells than in the parental cells (Fig. 3b). Results from the reporter assay and direct expression analysis of the Pu.1 target thus confirmed the deletion of miR-M4 and the functional consequences in the mutant C48 clone.

Effect of miR-M4 deletion on expression of other viral miRNAs and Meq protein.
Having demonstrated successful knockout of miR-M4 from the MDV genome in the HP8 cell line, we next analyzed the effect of miR-M4 deletion on expression of other MDV-encoded miRNAs and the Meq major viral oncoprotein. The 14 MDVencoded miRNA precursors are clustered into three separate genomic loci. Cluster 1 (the Meq cluster), containing miR-M2, miR-M3, miR-M4, miR-M5, miR-M9, and miR-M12, is located upstream of the Meq gene. The midcluster, containing three miRNA precursors (miR-M11, miR-M31, and miR-M1) is located downstream of Meq. The third cluster, referred to as the LAT cluster, lies within the first intron of the latency-associated transcript (LAT). To assess the potential effect of miR-M4 deletion on other miRNAs, we first amplified the cluster 1 miRNAs by PCR with the primers at the flanking region of the cluster. The sequence of the PCR product was determined to confirm the absence of any changes (data not shown) except for the edited region as shown in Fig. 1a. Next, we analyzed the level of expression of each miRNA in cluster 1, miR-M31 from cluster 2, and miR-M6 and miR-M8 from cluster 3 by the use of the RNA extracted from miR-M4-deleted clone 48 and the parental HP8-Cas9. The level of expression of the host miRNA gga-let-7a was also measured, with total RNA from uninfected CEF used as the control. As shown in Fig. 4a, all viral miRNAs were absent and only let-7a was detectable in the CEF sample. Except for the absence of miR-M4 from miR-M4-deleted clone 48, both the viral and the host miRNAs were detected in HP8 before or after miR-M4 deletion. Quantitation of selected viral miRNAs by qRT-PCR indicated that they were still expressed in miR-M4-deleted clone 48, although their expression levels differed from those seen with the parental HP8 cells (Fig. 4a). We also examined Meq expression in the miR-M4-deleted cells by Western blot analysis. Avian leukosis virus (ALV)transformed B-cell line HP45 and uninfected CEFs which do not express Meq were used as negative controls. Results of the Western blot analysis confirmed the expression of Meq in the miR-M4-deleted cells, demonstrating that miR-M4 was not required for Meq expression in these cells (Fig. 4b).
v-rel relieves the inhibition of miR-155 expression in HP8-⌬miR-M4. We have previously shown that miR-155 is consistently downregulated in MDV-transformed tumors and cell lines (34) and that this downregulation can be rescued by expressing v-rel, which also results in activation of the expression of miR-M4 in these cells (35). We  (Fig. 5a). Expression of v-rel increased the level of miR-155 expression by approximately 6,026-fold in HP8-ΔmiR-M4 cells but only 25-fold in HP8-Cas8 cells (Fig. 5b), demonstrating that deletion of miR-M4 increased miR-155 expression induced by v-rel.

DISCUSSION
Virus-host interactions in herpesviruses are characterized by long-term survival as latent infections in different cell types. With total dependence on the host cell, several viruses have adopted strategies to modulate the host cellular environment, including the modulation of miRNAs. A number of studies have demonstrated the role of miRNAs in replication, pathogenesis, and oncogenesis of herpesviruses (3,4,7,(36)(37)(38). These included our own studies demonstrating the critical role of miR-M4 in the induction of lymphomas by MDV (6). While these observations have also been confirmed by other studies (29), the role of viral miRNAs in maintaining the transformed state, as well as in  other functions such as the switch of latency/lytic replication in tumor cells, has not been examined. In particular, the role of miR-M4, the viral ortholog of oncogenic miR-155 encoded by oncogenic MDV, in maintaining the transformed phenotype of the tumor cell line is unknown. MDV-transformed LCLs derived from MD lymphomas which contain multiple copies of the MDV genome integrated in different chromosomes are valuable to study latency, transformation, and reactivation in situ. Having established the CRISPR/Cas9-based editing of the viral genome at relatively high efficiency in MDV-transformed cell lines, we report here the precise knockout of miR-M4 from the MDV genome in LCL HP8. Results from these studies show that miR-M4, despite its critical role in the induction of lymphomas by oncogenic MDV strains, is not required for the continued proliferation of MDV-transformed HP8 LCL. As far as we know, this was the first study to have made use of the CRISPR/Cas9-based gene editing technology in situ to demonstrate that a critical virus-encoded miRNA is not essential to maintain the transformed phenotype of a virus-induced cancer cell line.
By transfection of two forms of synthetic gRNA into HP8 cells stably expressing Cas9, we have shown here that miR-M4 can be deleted at a relatively high level of efficiency (Fig. 1c). Considering the presence of the multiple copies of the target loci in these cell lines, the high editing efficiency highlighted that efficient gRNA, rather than the copy numbers of the target genes, is the key to achieving the desired editing even in hard-to-transfect cell lines such as MDV-transformed LCL. Although the editing efficiency seen with the transfected cell lines appeared to be very high on the basis of the PCR test results, sorting of the single-cell populations did identify a number of unedited clones, further highlighting the importance of single-cell sorting in gene editing pipelines. These findings are also consistent with our observation that the rate of recovery of edited cells is probably much lower than that of the unedited cell populations, suggesting that single-cell cloning is a required step to get the pure populations of the edited cells regardless of the efficiency of gene editing. The successful knockout of miR-M4 demonstrated the value of this approach in identifying other molecular determinants associated with different phenotypes, including the latency/ lytic switch in LCLs. While the growth of the miR-M4-deleted cells confirmed that miR-M4 expression is not essential for maintenance of the transformation and proliferation of LCL, the ΔmiR-M4 cell line that we have generated will also be a valuable research tool in the future for addressing significant biological issues concerning the functional role of this important miRNA homolog. For example, it will be interesting to learn if the populations of shared target genes of MDV-miR-M4, miR-155, and KSHV-miR-K12-11 (5) are upregulated in miR-M4-deleted cells and downregulated after the v-rel transduction which activates miR-155 expression (Fig. 5b). Similarly, future studies employing global analysis of the changes in the transcriptome and proteomes of the edited cell populations, together with analysis of the changes in the viral and host epigenomes, will provide more insights into the fine tuning of the molecular regulatory network around the members of these families of miRNAs in these virus-transformed cell lines. Finally, these cells also provide the opportunity to investigate the role of miR-M4 in induction of lymphomas (transplantable tumors) in vivo in experimentally infected target chicken hosts.
Repair by nonhomologous end joining (NHEJ) is usually accompanied by random nucleotide insertions/deletions at the cleavage site. As a result, the edited sequence is likely to represent a mixed population. However, sequencing results have shown that virtually all of the edited sequences are end joining products of the two predicted Cas9 cleavage sites. Although additional variations may be discovered when more clones are analyzed, the edited loci often contained only the predominant mutant sequences, as we have shown previously (33,39). The reasons for the clonal nature of the appearance of the single population are not fully clear. Whether this is related to the stable expression of Cas9 in these cells or to other factors requires further investigation.
The oncogene v-rel activates miR-155 expression by binding to NF-B site in a Bic promoter. We have shown previously that the downregulation of miR-155 in MDVtransformed cell lines could be rescued by expressing v-rel in these cells (35). Using the same approach, we have shown here that the downregulation of miR-155 can also be rescued in the context of miR-M4-deleted HP8 by transduction of v-rel with RCAS(A)v-rel-GFP virus in HP8-ΔmiR-M4 clone C48. Interestingly, only a 25-fold increase in the miR-155 level could be induced in the unedited HP8-Cas9 cells compared to a 6,026fold increase in miR-M4-deleted clone C48, suggesting that the absence of miR-M4 significantly enhances the ability of v-rel to induce miR-155 expression in MDV tumor cell lines. As has been demonstrated previously, miR-M4 is highly expressed in MDV tumor cell lines compared to miR-155, which is actively downregulated, although the precise mechanisms of the differential downregulation have not been identified. On the basis of the findings from the present study, it appears that the downregulation of miR-155 may be directly linked to the high level of miR-M4 expression, as the activation of miR-155 by v-rel was more robust in the miR-M4-deleted cells. However, further studies are required to delineate the associated mechanisms involved in such regulation.
The precise editing of the miR-M4 locus to abolish the expression of mature miR-M4 in MDV-induced T-lymphoma-derived cell line HP8 clearly demonstrated that the proliferative capacity of the transformed cell line is not dependent on continued high-level expression of miR-M4. The continued proliferation of cells is unlikely to be due to the inability to express other viral miRNAs such as all other miRNAs in cluster 1 and selected miRNAs from both the midcluster and LAT cluster detected by miRNA qRT-PCR (Fig. 4a). MDV-miR-M4 is very important for the oncogenicity of MDV, but other miRNAs in the cluster also contribute, since the mutant virus expressing miR-M4 alone in cluster 1 remained nononcogenic (6). Whether or not other versions of miRNA contribute to the maintenance of transformed phenotype remains to be elucidated. The continued proliferation of cells is also not due to the lack of expression of adjacent viral gene such as Meq, as we were able to demonstrate expression of the protein by Western blot analysis (Fig. 4b). The significantly increased proliferation capacity of miR-M4 knockout clones suggests that miR-M4 in these contexts may have a proliferation suppressor function. Additional studies on the detailed analysis of the gene expression profiles of these clones will be required to gain further insights into the biology of miR-M4 in these cells. Although it is possible that LCLs may have acquired other mutations that may have made them no longer dependent on miR-M4 for proliferation, the failure of attempts to rescue the Meq-deleted cell line after repeated attempts indicated that this is unlikely to be the case. Whether or not other genes or miRNAs are involved in maintaining the transformed phenotype of MD tumor cell lines remains to be investigated.
gRNAs. A two-part guide RNA system containing a crRNA:tracrRNA guide complex was used for editing. The sequences of gRNA miR-M4-gN and miR-M4-gC (listed in Table 1 Table 1. Sorting. For single-cell cloning, cells were washed twice with phosphate-buffered saline (PBS) containing 5% fetal bovine serum (FBS) and centrifuged at 450 ϫ g for 5 min at room temperature. The cell pellets were resuspended in cold PBS-5% FBS and sorted into 96-well U-bottom plates (Corning) containing growth medium by fluorescence-activated cell sorter (FACS) analysis using a FACSAria II system (BD Bioscience).
qRT-PCR analysis of miRNA expression. The expression levels of miRNAs were analyzed using a TaqMan MicroRNA assay system (Life Technologies) and 10 ng total RNA as a template for reverse transcription. Each reverse transcription reaction was tested by PCR in triplicate and performed twice independently. For relative quantification of miRNA-M4 in HP8-ΔmiR-M4 cells (Fig. 1d) and of miR-155 in v-rel-transduced cells (Fig. 5b), all values were normalized to the expression level of endogenous let-7a, and levels were calculated as fold expression change relative to those from HP8-Cas9 cells (miR-155) and CEFs (miR-M4). For relative quantification of viral miRNAs and host gga-let-7a in HP8-ΔmiR-M4 clone 48 and controls HP8-Cas9 and CEF (Fig. 4a), all values were normalized to the level of expression of the endogenous GAPDH (glyceraldehyde-3-phosphate dehydrogenase) gene, and levels were calculated as fold expression change relative to those from CEF.
Dual luciferase reporter assay. A previously constructed reporter construct for the validated miR-M4 and miR-155 target Pu.1 in psiCHECK vector was used to measure the miR-M4 activity in HP8 (28). The reporter constructs contain a 110-bp fragment of the chicken Pu.1 3= untranslated region (UTR) sequence with MRE wild-type sequence (Pu.1-3=UTR-wt) or MRE mutant sequence (Pu.1-3=UTR-mu) inserted downstream of Renilla luciferase in the psiCHECK-2 vector (Promega) (28). HP8-ΔmiR-M4 cells and HP8-Cas9 cells (5 ϫ 10 5 ) were transfected with 4 g of either Pu.1-3=UTR-wt or Pu.1-3=UTR-mu using an NEPA21 electroporator as described above. The luciferase expression was assayed 48 h later using a Dual-Glo luciferase assay system (Promega) and following the manufacturer's instructions. The relative levels of expression of Renilla luciferase were determined with the normalized levels of firefly luciferase. For each sample, values from four replicates representative of results from at least two independent experiments were used in the analysis.
Analysis of HP8-Cas9-⌬miR-M4 cell growth. The growth of HP8-Cas9-ΔmiR-M4 clones along with unedited HP8-Cas9 was monitored by IncuCyte S3 live-cell imaging (Essen Bioscience Ltd., Hertfordshire, United Kingdom). Briefly, 8,000 cells were seeded in a 96-well plate (Corning) and images were captured every 4 h for 132 h from four separate regions per well using a 10ϫ objective. By recording the phase object confluence, the levels of growth of the HP8-Cas9-ΔmiR-M4 clones were compared with that of parental HP8-Cas9. IncuCyte data were analyzed by two-way analysis of variance (ANOVA) with Tukey's multiple-comparison test using GraphPad Prism version 7.01 (GraphPad Software, Inc., San Diego, CA). The results are shown as means Ϯ standard errors (SE) of results from four replicates each with 4 separate regions per well representative of three independent experiments. P values of Ͻ0.05 were considered to be significant.

ACKNOWLEDGMENTS
We thank Radmila Hrdlickova, Henry Bose, Jr. (University of Texas at Austin), and Tom Gilmore (Boston University) for kindly providing v-rel reagents.
This project was supported by the Biotechnology and Biological Sciences Research