Increased association between Epstein-Barr virus EBNA2 from type 2 strains and the transcriptional repressor BS69 restricts B cell growth

Natural variation separates Epstein-Barr virus (EBV) into type 1 and type 2 strains. Type 2 EBV is less transforming in vitro due to sequence differences in the EBV transcription factor EBNA2. This correlates with reduced activation of the EBV oncogene LMP1 and some cell genes. Transcriptional activation by type 1 EBNA2 can be suppressed through the binding of two PXLXP motifs in its transactivation domain (TAD) to the dimeric coiled-coil MYND domain (CC-MYND) of the BS69 repressor protein (ZMYND11). We identified a third conserved PXLXP motif in type 2 EBNA2. We found that type 2 EBNA2 peptides containing this motif bound BS69CC-MYND efficiently and that the type 2 EBNA2TAD bound an additional BS69CC-MYND molecule. Full-length type 2 EBNA2 also bound BS69 more efficiently in pull-down assays. Molecular weight analysis and low-resolution structures obtained using small-angle X-ray scattering showed that three BS69CC-MYND dimers bound two molecules of type 2 EBNA2TAD, in line with the dimeric state of full-length EBNA2 in vivo. Importantly, mutation of the third BS69 binding motif in type 2 EBNA2 improved B-cell growth maintenance. Our data indicate that increased association with BS69 restricts growth promotion by EBNA2 and may contribute to reduced B-cell transformation by type 2 EBV. Author summary Epstein-Barr virus (EBV) drives the development of many human cancers worldwide including specific types of lymphoma and carcinoma. EBV infects B lymphocytes and immortalises them, thus contributing to lymphoma development. The virus promotes B lymphocyte growth and survival by altering the level at which hundreds of genes are expressed. The EBV protein EBNA2 is known to activate many growth-promoting genes. Natural variation in the sequence of EBNA2 defines the two main EBV strains: type 1 and type 2. Type 2 strains immortalise B lymphocytes less efficiency and activate some growth genes poorly, although the mechanism of this difference is unclear. We now show that sequence variation in type 2 EBNA2 creates a third site of interaction for the repressor protein (BS69, ZMYND11). We have characterised the complex formed between type 2 EBNA2 and BS69 and show that three dimers of BS69 form a bridged complex with two molecules of type 2 EBNA2. We demonstrate that mutation of the additional BS69 interaction site in type 2 EBNA2 improves its growth-promoting function. Our results therefore provide a molecular explanation for the different B lymphocyte growth promoting activities of type 1 and type 2 EBV. This aids our understanding of immortalisation by EBV.


4
Epstein-Barr virus (EBV) is a ubiquitous γ -herpesvirus that immortalises human B 5 5 lymphocytes to establish a lifelong persistent infection that is usually harmless. Delayed 5 6 primary EBV infection can however give rise to infectious mononucleosis. EBV is also 5 7 associated with the development of malignancies that include Burkitt's (BL), Hodgkin's, infected with type 2 EBV that have the required levels of expression of these genes to support 1 0 4 their long term proliferation. In vivo other factors may create an environment that helps 1 0 5 support B cell immortalisation by type 2 EBV. New research also suggests that type 2 EBV may use alternative approaches to persist in vivo. approximately 2-fold for EBNA2 348-422 SD + m3 mt compared to the EBNA2 348-422 SD mutant 3 7 1 ( Figure 5C and D). These data indicate that motif 3 contributes to the second binding event.

7 2
We conclude that the SD mutation previously shown to enhance the growth maintenance 3 7 3 properties of type 2 EBNA2 (21) does not affect BS69 binding but likely alters the 3 7 4 conformation of the type 2 EBNA2 TAD. This may therefore impact on the binding of other 3 7 5 transcriptional regulators that influence type 2 EBNA2 function. Full-length type 2 EBNA2 binds BS69 CC-MYND more efficiently in pull-down assays 3 7 8 To confirm our in vitro observations that a type 2 EBNA2 polypeptide binds an additional 3 7 9 BS69 dimer, we examined the interaction of BS69 CC-MYND with full-length EBNA2 proteins in type 2 EBNA2, we found that GST-BS69 CC-MYND pulled down type 2 EBNA2 more 3 8 5 efficiently than type 1 EBNA2 at short incubation times ( Figure 6). In agreement with our in 3 8 6 vitro observations using the type 2 EBNA2 SD mutant, we found that this protein interacted were no longer evident. A control GST fusion protein (GST-Rab11) did not precipitate 3 9 0 EBNA2, confirming the specificity of the interactions. These data therefore confirm the 3 9 1 increased association of BS69 CC-MYND with type 2 EBNA2.

8
To determine whether the presence of the additional BS69 binding motif in type 2 EBNA2 3 9 5 (motif 3) had functional consequences for the activity of type 2 EBNA2, we examined the 3 9 6 ability of a type 2 EBNA2 motif 3 mutant to maintain B cell growth. We utilised a previously 3 9 7 described assay using an EBV-infected LCL (EREB2.5) in which the activity of a type 1 3 9 8 estrogen receptor-EBNA2 fusion protein can be switched off by estrogen withdrawal (37).

9 9
Loss of EBNA2 activity leads to growth arrest, but transfection of a stably-maintained 4 0 0 plasmid expressing type 1 EBNA2 into these cells supports their survival (20). In contrast, 4 0 1 the expression of type 2 EBNA2 cannot maintain the growth of these cells (20). We found 4 0 2 that mutation of BS69 binding motif 3 produced a type 2 EBNA2 protein that was able to 4 0 3 support the recovery of these cells from the loss of type 1 EBNA2 activity, with cells 4 0 4 recovering well 2-4 weeks following estrogen withdrawal ( Figure 7A). The type 2 EBNA2 4 0 5 motif 3 mutant behaved similarly to the type 2 EBNA2 SD mutant that was previously shown 4 0 6 to support B cell growth in this assay (21). We also examined the ability of the SD and motif 4 0 7 3 double mutant in this assay and found that it showed a slightly increased ability to support 4 0 8 B cells growth ( Figure 7A). In our hands, expression of type 1 EBNA2 supported initial 4 0 9 growth in this assay better than any type 2 mutants, with the mutants supporting growth 4 1 0 recovery from 2 weeks ( Figure 7A). We confirmed that all EBNA2 proteins were expressed 4 1 1 at similar levels ( Figure 7B). We conclude that the presence of the additional BS69 binding 4 1 2 motif in type 2 EBNA2 impairs the ability of type 2 EBNA2 to maintain B cell growth. infection of resting B cells by EBV and is low in the resulting immortalised LCLs (35).

1 9
Transcriptional repression of BS69 by EBNA2 was implicated in BS69 downregulation 4 2 0 indicating that EBNA2 may act to restrict expression of its own negative regulator (35). The 4 2 1 cell lines examined in this previous study all harboured type 1 EBV or type 1 EBNA2, so we 4 2 2 next addressed whether BS69 was expressed at similar levels in cells infected with type 1 and 4 2 3 type 2 EBV. We examined BS69 protein levels in type 1 and type 2 LCLs using an anti-BS69 4 2 4 antibody raised against a region within the MYND domain of BS69. We found that BS69 was 4 2 5 expressed at similar levels in type 1 and type 2 LCLs, but surprisingly levels in LCLs were  BS69 did not therefore appear to be downregulated as a result of EBNA2 expression. We also was not downregulated as previously reported ( Figure 8C). We therefore explored the possibility that we were detecting a different isoform of BS69. Alternative splicing has been reported to give rise to different BS69 isoforms and four have  protein detected in Figure 8A and B has a molecular weight of approximately 64 kD 4 4 8 consistent with that expected for isoform 2. This was the only protein detected by this could not exclude the possibility that one or more of these isoforms was also expressed and 4 5 2 that an alternative BS69 isoform was detected previously (35). In line with this possibility,  No detail was provided on the anti-BS69 antibody used previously (35) and we were not able 4 5 7 to find another antibody that detected isoform 3 and 4 in Western blotting. We therefore took 4 5 8 a non-quantitative PCR approach to screen for different BS69 isoforms using cDNA prepared 4 5 9 from LCLs and from B cells during a primary EBV infection. PCR using a forward primer in 4 6 0 exon 3 and a reverse primer in exon 13 amplified two products indicating the presence of at 4 6 1 least two different isoforms, one containing exon 4 (1139 bps) and one lacking exon 4 (977 BS69 CC-MYND dimers. It is possible that in solution in vitro there is a mixed population of 6 0 1 dimeric type 1 EBNA2 and monomeric type 1 EBNA2 complexes (where a single EBNA2 6 0 2 polypeptide is bound by one BS69 CC-MYND dimer as previously proposed). We were not able 6 0 3 to investigate this further using SAXS as this 'shoulder' was not clearly defined, so SAXS 6 0 4 analysis for both type 1 and type 2 EBNA2-BS69 complexes focused on the major elution 6 0 5 peak of the large complex. Given that full length EBNA2 expressed in EBV-infected cells is 6 0 6 a dimer, complexes involving two EBNA2 molecules are more likely to be physiologically Surprisingly, in our GAL4-EBNA2 fusion protein assays we did not see weaker 6 1 0 transactivation by the type 2 EBNA2 fusion protein compared to the type 1 EBNA2 fusion 6 1 1 protein as reported previously (21). We used a longer region of EBNA2 compared to this 6 1 2 previous study that encompassed all three BS69 binding sites for type 2 EBNA2 and the 6 1 3 corresponding region of type 2 EBNA2 (with only two functional BS69 binding sites).

1 4
Previously GAL4-EBNA2 fusion protein constructs were used that expressed a type 1 6 1 5 EBNA2 protein containing only BS69 binding motif 2 or the corresponding region of type 2 6 1 6 EBNA2 that contained BS69 binding motif 2 and 3 (21). It is not completely clear why the increased association of BS69 with type 2 EBNA2 is not associated with weaker 6 1 8 transactivation in our assays in the context of a longer region of EBNA2, but it could point to 6 1 9 the importance of the dimerisation that occurs in the context of the full-length protein in the 6 2 0 assembly of larger BS69-EBNA2 complexes. When considering the nature of assembly of BS69-EBNA2 complexes, it is likely that 6 2 3 binding to motif 1 (which in type 1 EBNA2 has the highest affinity for BS69 CC-MYND ) would 6 2 4 drive the initial interaction between EBNA2 and BS69 and binding to motif 1 probably 6 2 5 constitutes the first binding event that can be distinguished in our ITC analysis using an 6 2 6 increased number of injections. For type 2 EBNA2, since both motif 2 and 3 bind BS69 with 6 2 7 similar affinity, binding to both of these motifs probably occurs with similar kinetics and is 6 2 8 detectable as a single second binding event by ITC. Given the fact that BS69 CC-MYND dimers 6 2 9 are predicted in the solution structure of the BS69-EBNA2 complex to be located side by side 6 3 0 along a dimeric EBNA2 molecule, it is possible that interactions between BS69 coiled-coil 6 3 1 dimers play a role in stabilising the oligomeric complex. Our initial interest in examining type-specific binding of EBNA2 to BS69 centred around the 6 3 4 influence of a serine residue in the TAD of type 2 EBNA2 that plays a key role in restricting 6 3 5 B cell growth maintenance by type 2 EBNA2 (21). Although this residue is located 6 3 6 immediately adjacent to BS69 binding motif 2 in type 2 EBNA2, we found that it did not 6 3 7 increase BS69 binding (as might have been expected) when binding was compared to the 6 3 8 corresponding region of type 1 EBNA2 where there is an aspartate residue in its place. It does 6 3 9 not appear therefore that the influence of serine 409 on growth maintenance is mediated 6 4 0 through alterations in BS69 binding affinity. Our ITC analysis however did find that a serine repressor or co-repressor to the type 2 EBNA2 TAD or decreased binding of an activator or 6 4 6 co-activator. An S200 10/300 GL gel filtration column (GE Healthcare) was equilibrated with buffer Coomassie staining. EBNA2-BS69 complexes were prepared by pre-incubating proteins in a 1:3 molar ratio for at  Technology). For normalization of the light scattering and data quality, BSA was used as a 8 1 5 calibration standard.

1 6
Small-angle X-ray scattering 8 1 7 Synchrotron radiation X-ray scattering data from solutions of individual proteins or 8 1 8 complexes prepared as for SEC-MALS were collected on beamline B21 at Diamond Light Source (Didcot, United Kingdom), with an inline HPLC system. X-ray scattering patterns and 1mM TCEP, pH 7.5. Samples were analysed at 20°C using a flow-rate of 0.25 ml/min. Initial data processing (background subtraction and radius of gyration Rg calculation) was model for the complex were prepared using DAMMIF (50). 23 independent dummy atom 8 2 6 models were obtained by running the program in 'slow' mode. DAMAVER was then used to 8 2 7 align and average the models (51). The ab initio generated beads models were refined using   Lysates containing GST-tagged BS69 CC-MYND or GST-RAB11B were prepared from 100 ml an OD 600nm of 0.6 and protein expression was induced at 25ºC with 0.5 mM IPTG for 3-4 h.

4 9
Cells were pelleted at 2,800g for 20 min at 4ºC and then resuspended in 10 ml of Lysis buffer Systems) and cell debris pelleted at 17,900g for 30 min at 4ºC. Lysates were stored at -80ºC 8 5 4 until required. with nuclear extracts containing EBNA2 at 4ºC for different times (5, 10 and 30 minutes).

6 4
Loaded GST-RAB11B beads were incubated with lysates for 30 minutes. Beads were then bromophenol blue) and incubated at 95ºC for 5 min and analysed for EBNA2 levels by SDS-  Sigma) and anti-BS69 1/1000 (ab190890, Abcam). Western blot visualisation and signal 8 7 5 quantification was carried out using a Li-COR Imager. Gels were stained using Quick Coomassie stain (Generon Ltd). RNA was extracted from cells using Trireagent (Sigma), further purified using the RNeasy   of all primers was determined prior to use and in each experiment and all had amplification 8 8 7 efficiencies within the recommended range (90-105%). Agarose gel analysis of PCR products generated using primers located in specific BS69  Scattering intensity I (q), scattering vector (q), radius of gyration (Rg). refined with DAMMIN to produce refined dummy atom models (magenta mesh). The 1 0 7 3 maximum dimension (D max ) and volume were calculated using the ScÅtter programme. In  This gave a similar χ 2 (2.59) to that shown in Figure 4A indicating a much better fit to the The hydration parameter (C 2 ) was fixed to 0 to prevent the hydration shell increasing to high levels to 1 0 9 6 attempt to fit the structure into the envelope. (E) The structural model shown in Figure 4D (three 1 0 9 7 BS69 CC-MYND dimers and two type 2 EBNA2 348-422 polypeptides) was refitted to the SAXS envelope 1 0 9 8 using FoXS with the C 2 value set to 0 for comparison. This gave a similar χ 2 (1.47) to that shown in 1 0 9 9 Figure 4B indicating a much better fit to the scattering data.  show the mean ± standard deviation for three independent experiments. For peptides n values 1 1 0 6 were fixed to 1. Data from type 2 EBNA2 peptides and polypeptides are in shaded columns. virus exploits intrinsic B-lymphocyte transcription programs to achieve immortal cell growth.     T1  T2  T2 SD  T1  T2  T2 SD  T1  T2  T2 SD  T1  T2  T2 SD  T1  T2  T2 T1 E2  T2 E2  T2 E2 SD  T2 E2 m3 mt  T2 E2 SD m3     Fold activation