Role of the vaccinia virus O3 protein in cell entry can be fulfilled by its Sequence flexible transmembrane domain

Abstract

The vaccinia virus O3 protein, a component of the entry–fusion complex, is encoded by all chordopox-viruses. We constructed truncation mutants and demonstrated that the transmembrane domain, which comprises two-thirds of this 35 amino acid protein, is necessary and sufficient for interaction with the entry–fusion complex and function in cell entry. Nevertheless, neither single amino acid substitutions nor alanine scanning mutagenesis revealed essential amino acids within the transmembrane domain. Moreover, replication-competent mutant viruses were generated by randomization of 10 amino acids of the transmembrane domain. Of eight unique viruses, two contained only two amino acids in common with wild type and the remainder contained one or none within the randomized sequence. Although these mutant viruses formed normal size plaques, the entry–fusion complex did not co-purify with the mutant O3 proteins suggesting a less stable interaction. Thus, despite low specific sequence requirements, the transmembrane domain is sufficient for function in entry.

Introduction

Vaccinia virus (VACV) enters cells by fusion either at the plasma membrane or via a low pH-dependent endosomal mechanism, depending to some extent on the virus strain (Bengali et al., 2009, 2012; Chang et al., 2010; Huang et al., 2008; Mercer and Helenius, 2008; Mercer et al., 2010; Townsley et al., 2006) and cell type (Bengali et al., 2009, 2011; Whitbeck et al., 2009). Four proteins participate in the attachment of mature virions to cell surface glycosaminoglycans and laminin (Chiu et al., 2007; Chung et al., 1998; Hsiao et al., 1999; Lin et al., 2000). Twelve proteins that are conserved in all poxviruses mediate the fusion step, of which eleven have been shown to associate in a complex (Bisht et al., 2008; Brown et al., 2006; Izmailyan et al., 2006; Nichols et al., 2008; Ojeda et al., 2006a, 2006b; Satheshkumar and Moss, 2009; Senkevich and Moss, 2005; Senkevich et al., 2004, 2005; Townsley et al., 2005). O3, the subject of the present study, has only 35 amino acids and is the smallest protein with a known function encoded by VACV (Satheshkumar and Moss, 2009). O3 is a component of the entry–fusion complex (EFC) and deletion or repression of O3 results in the production of virus particles that exhibit a greatly reduced ability to enter cells (Satheshkumar and Moss, 2009). Although the absence of O3 does not prevent other components of the EFC from inserting into the MV membrane, the complex does not form or is unstable under this condition.

The precise roles of the individual proteins comprising the EFC have not yet been elucidated. However, nine of the eleven EFC proteins including O3 are required for the initial hemi-fusion step in entry (Laliberte et al., 2011). Although the amino acid sequence of O3 is highly conserved among orthopoxviruses, the sequences are extremely divergent between chordopoxvirus genera. Common features of all O3 homologs are their location in the genome, small size and presence of a N-terminal hydrophobic domain. Despite sequence diversity, O3 homologs from other genera are able to complement a VACV null mutant (Satheshkumar and Moss, 2012). In VACV, the predicted α-helical TM domain comprises two-thirds of the 35-amino acid O3 protein. Preliminary evidence that the TM domain determines O3 function was obtained by swapping the N-terminal TM and C-terminal domains of O3 with the corresponding domains of an unrelated VACV EFC protein (Satheshkumar and Moss, 2012). The chimeric protein, which retained the O3 TM domain, complemented an O3 null mutant and interacted with other EFC proteins, whereas the chimeric protein that retained the C-terminal domain of O3 was nonfunctional. These data suggested that the O3 TM domain interacts with other EFC proteins possibly through their membrane-spanning domains. The purpose of the present study was to directly determine whether the TM domain is sufficient for O3 function and to interrogate the sequence requirements of this domain.

Results

Effect of O3 truncations on trans-complementation of an O3 null mutant

The N-terminal 22 amino acids of O3 are predicted to include the α-helical TM domain, which is followed by a 13 amino acid external domain that starts with a conserved positively charged amino acid followed by proline. We constructed a series of truncation mutants in which the C-terminus of O3 was replaced by a 9-amino acid HA tag (Fig. 1A). The abilities of these constructs to trans-complement the O3 null mutant vO3-HAi-LUC, which contains an inducible O3 with an HA epitope tag at the C-terminus as well as the luciferase (LUC) gene regulated by a VACV promoter, were tested using a two-step protocol (Satheshkumar and Moss, 2012). First, BS-C-1 cells were infected with vO3-HAi-LUC in the absence of inducer to prevent O3 synthesis and transfected with a control plasmid or plasmid encoding a truncated O3 under the natural O3 promoter. Use of the natural promoter ensured that transcription would take place in the virus factory. The cells were harvested after 24 h and the lysates were used to infect fresh cells. After 1 h, LUC activity was determined as a measure of entry-competent virus in the lysate from the first step. With lysates from untransfected cells that were infected with vO3-HAi-LUC in the absence and presence of inducer, there was about a 2-log difference in LUC activity (Fig. 1B). The LUC activity declined further in cells infected with vO3-HAi-LUC in the absence of inducer and transfected with the control TOPO plasmid, probably due to some reagent toxicity. Transfection of the WT O3 plasmid (N-35) restored more than a log of LUC activity relative to untransfected cells infected with vO3-HAi-LUC in the absence of inducer (Fig. 1B), indicating the formation of infectious virus. About 60% of the activity, relative to N-35, was obtained by transfection of the N-24 truncation mutant, which retained the TM as well as the conserved positively charged amino acid followed by proline (Fig. 1B). Activity decreased to about 40% and 20% of the WT value with deletions of the proline (N-23) and arginine (N-22), respectively. With removal of additional amino acids, the LUC activities were in the range of 10–20% of WT until N-17, at which no activity above the control was detected (Fig. 1B). Western blotting indicated that the amounts of N-24 to N-21 proteins were similar to WT but that the amounts decreased with greater truncations (Fig. 1C).

Fig. 1.

Trans-complementation by O3-truncation mutants. (A) The predicted sequence of the WT O3 ORF (N-35) and progressive C-terminal truncation mutants (N-24 to N-17). Tag refers to the nine amino HA sequence appended to the C-terminus of each protein. The predicted TM domain from position 4 to 22 is in bold. (B) Complementation activity. BS-C-1 cells were infected with vO3-HAi-LUC in the presence (+) or absence (−) of IPTG and in the latter case transfected with a control topoisomerase plasmid (TOPO) or with a plasmid encoding full length (N-35) or a truncated (N-24 to N-17) O3 protein. Relative virus yields were determined by a LUC-based entry assay. RLU/s (relative light unit per second). (C) Western blots of cells infected and transfected as in panel B. The blots were probed with antibodies to the HA tag on O3 mutants, the VACV A11 protein as a viral loading control, and cellular glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a cellular loading control.

Construction and analysis of viruses with C-terminal truncations of O3

The above transfection experiments demonstrated that the TM domain with the two additional conserved amino acids (N-24) was sufficient for significant O3 function but that N-17 had little or no activity. To confirm these results in the context of a virus, recombinant VACV were constructed with WT N-35 and the truncation mutants N-24 and N-17 by replacing GFP in vO3Δ (Fig. 2A). DsRED under a separate promoter was co-inserted in order to facilitate virus isolation. The plaque size of the N-35 virus was similar to that of the parental VACV strain WR suggesting that the DsRED and the HA epitope tag had no appreciable effect on virus replication and spread (Fig. 2B). The N-24 virus also formed WT-size plaques, confirming the sufficiency of the shortened ORF. However, the N-17 virus, with a truncated TM, formed tiny plaques similar to those of the O3 deletion virus.

Fig. 2.

Characterization of recombinant VACV encoding mutant O3 proteins. (A) Diagram of recombinant viruses. vO3 mutant viruses were constructed by replacing GFP of vO3Δ with full length (N-35) or truncated (N-24 or N-17) O3 ORF and a DsRed ORF regulated by a VACV promoter. Arrows indicate the direction of transcription of indicated ORFs. vWR, parental VACV strain WR. (B) Plaque formation. Revertant viruses were isolated by picking red fluorescent plaques and clonally purified. Appropriate dilutions of indicated viruses were applied to BS-C-1 monolayers, covered with methylcellulose and incubated at 37 °C for 2 days. Plaques were visualized by staining with crystal violet. (C) Western blot of proteins from purified virus particles. Virus particles were purified by sedimentation through a sucrose cushion and banding on a sucrose gradient. Equal amounts of virus, determined by optical density, were solubilized with SDS and subjected to polyacrylamide gel electrophoresis. The proteins were transferred to a nitrocellulose membrane and the O3 proteins were detected by staining with antibody to the HA epitope tag. The A3 protein was detected with specific antibody as a viral loading control. (D) Association of O3 proteins with the EFC. Infected BS-C-1 cells were lysed with triton X-100 and the O3 proteins were bound to beads coupled with anti-HA antibody. After washing, the proteins were eluted with SDS and analyzed by Western blotting using antibodies to HA to detect O3 proteins and to specific antibodies to two representative EFC proteins L5 and A16. WR represents the parental VACV strain, which has no HA-epitope tag.

Next we checked whether the mutated O3 proteins were incorporated into MVs. The virus particles were purified by sucrose gradient sedimentation and analyzed by Western blotting using antibody to the HA tag. Both mutant proteins were detected in virions, although the N-17 was present in a low amount (Fig. 2C). We also determined the ability of the mutated O3 proteins to interact with other VACV EFC proteins. Our previous affinity purification studies showed stable association of O3 with the EFC. Here, we used the HA tag attached to the O3 proteins to co-purify associated proteins from infected cell lysates and performed Western blotting with antibodies to two representative EFC proteins L5 and A16. As a negative control, we incubated lysates from cells infected with the parental VACV WR, which lacks the HA tag. L5 and A16 were associated with both the N-35 and N-24 O3 proteins but were not detected above the WR background with the N-17 O3 protein (Fig. 2D). The result obtained with N-24 confirmed our idea that the interaction of O3 with other EFC proteins occurs through the TM domain.

Site-directed mutagenesis

The effects of truncation mutants could be due to a combination of removal of specific amino acids and progressive shortening of the expressed protein. To evaluate the sequence effects specifically, we made amino acid substitution mutations within the TM domain of the full length O3 protein. The selection of amino acids to mutate was made by identifying the most conserved amino acids in O3 homologs of other chordopoxviruses. As depicted in Fig. 3A, the N-terminal 24 amino acids of the O3 homologs of representatives of chordopoxvirus genera are diverse in sequence, although there is an enrichment of hydrophobic amino acids. Indeed, there are no completely conserved amino acids within the TM domain, which is predicted to be between amino acids 4–22. Highly or moderately conserved amino acids included the bulky hydrophobic amino acid phenylalanine at positions 7 and 11, cysteine at position 14, hydroxyl side chain amino acids serine and tyrosine at positions 18 and 19, respectively, and a leucine at position 22. The following substitutions in the above amino acids were made: F7A, F11A, C14A, S18A and the SYSY to AFAF in a N-35 O3 ORF containing a C-terminal HA tag that was regulated by the natural O3 promoter (Fig. 3B). The effects of these mutations were evaluated by transfecting plasmids expressing the mutated ORFs into cells infected with vO3-HAi-LUC in the absence of inducer. Remarkably, there was little or no effect on the virus yields as measured by LUC activity (Fig. 3C) or protein expression determined by Western blotting (Fig. 3D).

Fig. 3.

Site-specific mutations within the TM domain. (A) Sequences of O3 homologs. The N-terminal 24 amino acids of representative chordopoxvirus O3 homologs are shown. Relatively conserved amino acids are shaded and shown at the bottom. Abbreviations: YMTV, Yaba monkey tumor virus; SWPV, swinepox virus; SPPV, sheeppox virus; ORV, orf virus; MYXV, myxoma virus; FWPV, fowlpox virus. (B) Site-specific mutations within the full-length N-35 O3 protein. Alanine substitutions and addition shown in bold. C-terminal HA tag indicated. (C) Effects of site-specific mutations on complementation of O3 null mutant. BS-C-1 cells were infected with vO3-HAi-LUC in the presence (+) or absence (−) of IPTG and the latter were transfected with TOPO control plasmid or plasmid containing mutated O3 ORFs indicated in panel B. Complementation was assayed as in Fig. 1B. (D) Western blots of transfected cells probed with antibody to HA to detect O3 proteins and with specific antibodies to the VACV control A11 and the cell control GAPDH.

Secondary structure predictions suggested the presence of an α-helix within the TM domain of O3. An O3 mutant was constructed by adding two alanines, which have high helix propensity, within the α-helix (Fig. 3B). Since looping out of residues should be disfavored in a lipid environment, the consequence could be to displace amino acids and alter geometry and hydrogen bonding. The mutated O3 protein was expressed at WT level (Fig. 3D) but complementation ability was reduced by about 80% (Fig. 3C), consistent with a structural perturbation. Single proline insertions at positions 8 or 13 or a double mutation at both positions were made in order to break the predicted α-helix. Although these mutations almost completely abrogated complementation, the proteins could not be detected by Western blotting even when the proteasome inhibitor MIG132 was added (data not shown). Therefore, we could not interpret the effects of these proline insertions on O3 function.

In addition to making mutations within the predicted TM domain, we also made alanine substitutions of the conserved positively charged amino acid and proline, at positions 23 and 24, respectively (Fig. 4A). These mutations were made in the N-24 truncation mutant. The mutated protein was stably expressed and the amino acid substitutions had only a small effect relative to the N-24 control (Fig. 4B, C). Thus, none of the mutations of conserved amino acids had a major effect on the function of O3.

Fig. 4.

Mutations of the two amino acids following the TM domain. (A) Sequences of full length N-35 O3 ORF, truncated N-24 O3 ORF and truncated N-24 ORF with alanine substitutions. The putative TM domain is in bold letters. (B) BS-C-1 cells were infected with vO3-HAi-LUC in the presence (+) or absence (−) of IPTG and the latter were transfected with control TOPO plasmid or plasmids indicated in panel A. Complementation assays were carried out as in Fig. 1B. (C) Western blot of lysates of cells infected with vO3-HAi-LUC and transfected as in panel B. The Western blot was probed with antibody to the HA tag to detect O3 proteins and to A11 as a virus loading control.

Alanine-scanning mutation of TM domain

Since changes of individual conserved amino acids did not affect O3 function, we made more drastic alanine-scanning mutations. Groups of four consecutive amino acid residues, from positions 2 to 17, were mutated to alanine (Fig. 5A). Surprisingly, none of these mutations had a significant impact on O3 complementation ability (Fig. 5B). We also made a plasmid containing the O3 ORF in which each amino acid from position 7 to 16 was changed to alanine. Although this mutant was unable to complement vO3-HAi-LUC, the protein could not be detected by Western blotting precluding as assessment of O3 function (data not shown).

Fig. 5.

Alanine scanning mutations. (A) Diagram of sites of AAAA substitutions in O3. The substitutions are indicated in bold. (B) Complementation. Infection, transfection and complementation were determined as in Fig. 1B.

Randomization of the O3 TM domain sequence

Since neither alanine substitutions of individual conserved amino acids nor four-amino acid alanine scanning mutations of O3 had much effect on the ability to complement an O3 null mutant, we took another approach to determine the limits of sequence flexibility. As we knew that substitution of the amino acids 7–16 with alanines resulted in loss of stability and function, we made a panel of recombinant viruses with those 10 amino acids randomized and screened for functional O3 by determination of plaque size. A chemically-synthesized degenerate codon library of 30 nucleotides, in which there was an enrichment of hydrophobic amino acid codons (Marlatt et al., 2011), was flanked by the remaining N- and C-terminal O3 and neighboring sequences (Fig. 6A). Homologous recombination was achieved by transfecting the library into BS-C-1 cells that had been infected with vO3Δ, a VACV deletion mutant in which GFP had replaced O3. After five serial passages, to enrich for recombinant viruses that replicated better than the parental vO3Δ, a plaque assay was performed. Viruses were isolated from 39 large, non-fluorescent plaques and clonally purified. There were eight unique O3 sequences in the 39 isolates. The plaque sizes of the eight were similar to those of WT virus and larger than the deletion mutant (Fig. 6B). Although hydrophobic amino acids were enriched, three of the mutants (d23, d19, d7) shared no amino acids with WT, three of the mutants (d25, d10, d8) shared one amino acid and two (d22, d20) shared two amino acids (Fig. 5C). The most common amino acids shared with WT were the phenylalanines at positions 7 and 13. Interestingly, four of the mutants (d22, d20, d19. d10) had potential helix breakers: a glycine in three cases and a proline in one. In three of these mutants, the glycine or proline was at the position of a bulky hydrophobic amino acid in WT O3.

Fig. 6.

Isolation of large plaque-forming mutants with randomized TM sequences. (A) Diagram of region of O3 subjected to randomization. Upper: Diagrams of WT O3 and randomized O3 sequences. The randomized amino acids are indicated by Xs. Lower: Diagrams of region of vO3Δ genome showing O3 ORF replaced by GFP and revertant viruses containing randomized O3. (B) Plaque formation. BS-C-1 cells were infected with vO3Δ and transfected with a library of randomized O3 sequences containing HA tag. After five consecutive passages, 39 large non-fluorescent plaques were clonally purified and the O3 ORF was sequenced. Eight of the large plaque-forming viruses (d7, d8, d10, d19, d20, d22, d23 and d25) had unique sequences. Plaques formed by the parent vO3Δ, vO3-HA WT, and mutants. BS-C-1 monolayers were infected, covered with methylcellulose and incubated at 37 °C for 2 days. Plaques were visualized by staining with crystal violet. (C) Comparison of O3 sequences from amino acid 7 to 16 of WT and mutants. Amino acids repeated in multiple isolates are shaded.

We also clonally isolated and characterized virus from one small, white plaque (S10) that was similar in size to the deletion mutant (Fig. 7A). The sequence differed from WT in lacking any bulky hydrophobic residues (Fig. 7C). However, one of the large plaque isolates (d19) also lacked bulky hydrophobic amino acids. Western blot of the lysate indicated that the S10 O3 protein was present in amounts similar to that of WT O3 and the large plaque mutant d19 (Fig. 7B). Therefore the small plaque was not due to instability of the O3 protein.

Fig. 7.

Isolation of small plaque-forming mutant with randomized TM sequence. (A) Plaque formation by mutant virus. Procedures as in Fig. 6B except virus was isolated from a small non-fluorescent plaque and clonally purified. Image shows crystal violet stained plaques that formed after 3 days to better view the small S10 plaques. (B) Cell lysates from cells infected with vO3-HA WT, small plaque mutant S10 or large plaque mutant d19 were analyzed by Western blotting and probed with antibody to the HA tag on O3 or specific antibody to A11 as a viral protein loading control. (C) Comparison of O3 sequences from amino acid 7 to 16 of WT and mutant. Amino acid repeated in mutant and WT is shaded. (D) Association of O3 mutant protein with purified virus particles. Virus particles were purified from infected cell lysates by sedimentation through a sucrose cushion and banding on a sucrose gradient. Equal amounts of purified virus determined by optical density were incubated with buffer containing NP-40 and centrifuged to separate the solubilized membrane proteins (S) and the pelleted core proteins (P).

Properties of viruses with randomized O3 sequences

The viruses from the randomized sequence screen were purified from infected cells by sucrose-density gradient centrifugation to determine whether the mutated O3 proteins were incorporated into the MV membrane like WT O3. Western blot analysis demonstrated that the modified O3 proteins of the small plaque isolate (Fig. 7D) as well as the large plaque isolate (Fig. 8A) were indeed present in purified virions and could be extracted with non-ionic detergent, indicating membrane association. Next we investigated whether the mutated O3 proteins, like WT O3, formed a stable interaction with proteins in the EFC. When the WT O3 protein was affinity purified, three representative components of the EFC were detected by Western blotting (Fig. 8B). However, the EFC proteins did not detectably co-purify with the O3 proteins with randomized sequences (Fig. 8B), despite the ability of the latter to functionally replace O3. These results were reminiscent of a previous finding that the Orf virus sequence divergent O3 homolog could complement O3 for VACV replication but did not pull down EFC proteins in a similar assay (Satheshkumar and Moss, 2012).

Fig. 8.

Properties of O3 large plaque randomization mutants. (A) Association of O3 proteins with purified virus particles. Virus particles were purified from infected cell lysates by sedimentation through a sucrose cushion and banding on a sucrose gradient. Equal amounts of purified virus determined by optical density were incubated with buffer containing NP-40 and centrifuged to separate the solubilized membrane proteins (S) and the pelleted core proteins (P). The proteins from both fractions were analyzed by SDS polyacrylamide gel electrophoresis and Western blotting with antibodies to HA to detect O3. Antibodies to L1 and A3 were used to visualize a membrane and core protein, respectively. (B) Affinity purification of EFC proteins. Infected BS-C-1 cells were lysed with triton X-100 and the O3 proteins were bound to beads coupled with anti-HA antibody. After washing, the proteins were eluted with SDS and analyzed by Western blotting using antibodies to HA to detect O3 proteins and to specific antibodies to three representative EFC proteins L5, A16 and A28. WR represents the parental VACV strain with no HA-epitope tag; WT represents virus with the WT O3 containing an HA epitope tag. Western blots of the extract prior to affinity purification (Input) and after HA affinity purification (HA affinity precipitation) are shown.

We also checked the ability of the viruses with randomized O3 sequences to mediate cell–cell fusion at low pH, which had previously been shown to require O3 (Satheshkumar and Moss, 2009). Like WT, none of the mutants acquired the ability to mediate syncytium formation at neutral pH and most triggered fusion at low pH (Fig. 9). Two mutants that induced few syncytia were vO3-d22 and vO3-d20, although their plaques were similar in size to WT O3.

Fig. 9.

Abilities of O3 randomization mutants to mediate cell–cell fusion. BS-C-1 cells were infected with vO3-WT or randomized O3 mutant viruses with HA epitope tag. After 16 h, the cells were briefly incubated with medium at pH 5.5 or 7.4 at 37 °C. Cells were incubated in regular medium for 2 additional hours and viewed by light microscopy.

Discussion

The O3 protein, despite its small size, is a bona fide component of the EFC: it associates with the MV membrane, is required for entry and low pH triggered cell–cell fusion, stably interacts with other components of the EFC, and is required for EFC assembly and stability (Satheshkumar and Moss, 2009). Several previous observations had suggested a specific role of the TM domain of O3. First, the TM domain was predicted to comprise two-thirds of this 35 amino acid protein (Satheshkumar and Moss, 2009). Second, the external domains of O3 homologs varied in length, with some even shorter than the VACV O3, and had no significant sequence homology. Third, in a domain swapping experiment between O3 and the A21 EFC protein, a chimera containing the O3 TM and a shortened A21 C-terminal domain could complement an O3 null mutant, whereas the opposite, i.e. a chimera with the A21 TM and the O3 C-terminal domain, could not (Satheshkumar and Moss, 2012). Here, we have shown directly that the first 22–24 amino acids of O3 are sufficient for complementation of an O3 null mutant. Like WT O3, the 24 amino acid protein was inserted into the MV membrane and interacted with other protein components of the EFC.

In view of the evidence for the importance of the TM domain for O3 function, the abilities of sequence divergent TM domains from different chordopoxvirus genera to functionally complement a VACV O3 null mutant is surprising (Satheshkumar and Moss, 2012). Multiple sequence analysis of TM domains of O3 homologs did not reveal highly conserved features, although there was modest conservation of large bulky hydrophobic amino acids. Substitutions of the individual most conserved amino acids in the TM domain of VACV O3 with alanines or replacement of the unusual four consecutive hydroxyl-group containing amino acids (SYSY) with AFAF failed to significantly reduce the complementation efficiency of O3. There was also little effect of substituting the conserved positively charged amino acid and proline following the TM domain with alanines. Even replacing four amino acids at a time by alanine scanning did not appreciably affect O3 function. However, adding rather than replacing two alanines within the TM domain, which should alter the geometry and hydrogen-bonding, reduced complementation appreciably.

To explore permissible amino acids within the TM domain, we chemically synthesized degenerative codon libraries to replace all or part of the WT TM domain of O3 and screened for viruses that made WT-size plaques. Because the fully randomized sequence space derived from 10 positions with codons for all 20 amino acids would require impossible screening of 10¹⁵ members (Mena and Daugherty, 2005), we made a library that was biased for hydrophobic amino acids (Marlatt et al., 2011). This library replaced the 10 amino acids from residues 7 to 16 of O3. DNA containing the mutated O3 ORF was transfected into cells infected with an O3 deletion mutant and the progeny were passaged multiple times to select viruses with enhanced replication. The latter were then screened by plaque formation and viruses from 39 “WT-like” plaques were isolated. Eight of these isolates had unique sequences. Remarkably, three of the mutants shared no randomized amino acids with the WT O3 sequence, three shared one amino acid and two shared two amino acids. Moreover, the mutants had few randomized amino acids in common with each other, suggesting a very large degree of plasticity of the O3 TM domain. Interestingly, a potential helix-breaking glycine at positions 8, 11 or 15 or a proline at position 11 was found in four WT-like O3 mutants. In contrast, the specific insertion of a proline at position 8 or 13 into the WT O3 sequence resulted in instability of the O3 protein indicating the importance of sequence context. In an additional, admittedly high-risk experiment, we made a 21-codon hydrophobic amino acid biased library encompassing the entire TM domain of O3 hoping that the permitted sequence diversity would allow isolation of some totally random O3 sequence mutants with wild type size plaques. However, the sequence space of a degenerative library increases exponentially with each additional residue and it was not too surprising that we failed to isolate WT-like virus (data not shown).

Evidence that the O3 TM domain interacts with the EFC complex was obtained by affinity isolation of representative EFC proteins by both the wild type O3 and the N-terminal 24-amino acid truncation mutant. We considered that mutations within the TM domain might have a subtle effect on the O3 interaction with the EFC that was not revealed by plaque size. With this in mind, we performed affinity isolations with the TM randomization mutants. Although the mutant O3 proteins were incorporated into the virion membrane, EFC proteins did not co-purify with them. In view of the ability of these mutant viruses to mediate virus infection, large plaque formation and cell–cell fusion, we suggest that the interaction of the O3 TM domain with the EFC is functional but too weak to withstand the rigors of detergent extraction, binding to an antibody affinity column and stringent washing. However, an alternative explanation is that the association of O3 with other EFC components is not necessary for function.

Virus entry and syncytium formation require EFC proteins and likely employ a similar mechanism of membrane fusion (Wagenaar and Moss, 2007). Nevertheless, there are significant differences. While a single virus particle can fuse with the cell membrane, hundreds of particles are necessary to demonstrate syncytium formation. Furthermore, virus particles can fuse with the plasma membrane at neutral pH whereas a low pH pulse is needed for MVs to mediate cell–cell fusion. Cell–cell fusion may require multiple fusion events to occur nearly simultaneously (Moss, 2006). We found that most of the mutant O3 proteins enabled both cell entry and syncytium formation. However, two mutants, d20 and d22, were positive in the cell entry assay but not for syncytium formation. In theory, an inability to mediate cell–cell fusion could result if d20 and d22 reacted differently to low pH or had slower fusion kinetics than other mutants. Although d22 is unusual in having four methionine substitutions and both mutants have a glycine substitution within the TM, we have no evidence to show that these sequence changes are consistent with the above explanations.

We embarked on this study with optimism that we would be able to define a sequence or structure that accounted for the specific requirement of O3 for VACV entry. Although we can definitively conclude that the TM domain is necessary and sufficient, the extremely high degree of sequence flexibility within the TM domain has prevented us from making a structural model. Like many studies, the present one opens up a multitude of new questions such as: how is the 35 amino acid full length O3 protein and the 24 amino acid truncated version, containing a 9 amino acid HA tag, transported to the viral membrane; does the TM domain penetrate through the viral membrane or is part exposed; does the TM domain interact with the TM domain of other EFC proteins; and which EFC proteins directly interact with O3?

Materials and methods

Cells and viruses

BS-C-1 cells were maintained in minimum essential medium with Earle's salts supplemented with 10% fetal bovine serum, 100 units of penicillin and 100 mg of streptomycin per ml (Quality Biologicals, Gaithersburg, MD). The recombinant virus vO3-HAi-LUC (Satheshkumar and Moss, 2012) and vO3Δ (Satheshkumar and Moss, 2009) were described previously. Virus particles were purified by sedimentation through a 36% sucrose cushion and banding on 25–40% sucrose density gradient (Earl and Moss, 1998). Extraction of membrane proteins was carried out as previously described (Satheshkumar and Moss, 2012).

Trans-complementation assay

Complementation assays were carried out by infecting BS-C-1 cells in four wells of a 12-well plate with vO3-HAi-LUC, transfecting with plasmids encoding O3 mutants, and determining virus yields at 24 h after infection in three of the wells by a LUC-based entry assay (Satheshkumar and Moss, 2012). The fourth well was used for Western blotting to ensure expression of the O3 protein. For the LUC assay, 0.1 ml of lysate was added to BS-C-1 cells in a 24-well plate, which was incubated for 1 h at room temperature to allow virus adsorption, washed with medium and incubated in 0.5 ml of medium for 1 h at 37 °C to allow LUC expression. The cells were lysed with 150 μl of Cell Culture Lysis Reagent (Promega, Madison, WI) for 15 min at room temperature on an orbital shaker. Following addition of 20 μl of cell lysate to 100 μl LUC substrate, LUC activity was measured with a luminometer (Berthold Sirius, Bad Wilbad, Germany). Preliminary experiments verified that the LUC activity was proportional to virus concentration as determined by plaque assay.

Construction of recombinant viruses with mutated O3 genes

The GFP gene in an O3 deletion virus (vO3Δ) was replaced by homologous recombination with a mutated O3 ORF regulated by the O3 promoter and an adjacent DsRED ORF controlled by VACV P11 late promoter. Red fluorescent plaques were picked repeatedly and characterized by PCR and sequencing.

Construction of recombinant viruses with randomized O3 TM sequences

DNA primers with degeneracy at 30 nucleotide positions were used to randomize the 10 amino acids corresponding to positions 7–16 of the O3 amino acid sequence. To enrich hydrophobic amino acids, the specific nucleotide composition was N=28–28–28–16 (A=28%; C=28%; G=28%; T=16%), N1=07–17–07–69 (A=07%; C=17%; G=07%; T=69%) and N2=00–90–10–00 (A=0%; C=90%; G=10%; T=0%) at each position of the coding triplet sequence, N, N1, N2 as described (Marlatt et al., 2011). The degenerate PCR product was fused with left and right flanking sequences and transfected into BS-C-1 cells infected with vO3Δ to replace the GFP ORF with the randomized O3 ORF sequence. After five consecutive passages, large plaque-forming viruses without green fluorescence were identified and clonally purified.

Affinity purification

Interactions of O3 with other proteins were determined in infected cells by affinity purification with anti-HA antibody affinity beads (Roche, Indianapolis, IN). Briefly, infected cells were lysed in 50 mM Tris pH 8.0, 200 mM NaCl, 1% triton X-100 buffer containing complete protease inhibitor tablet (Roche) for 30 min at 4 °C. Lysates were clarified by centrifugation at 10,000g for 10 min at 4 °C and the supernatant was incubated with anti-HA antibody beads. After binding and washing, the beads were boiled with 1 × lithium dodecyl sulfate loading solution containing dye and proteins were separated in a 4–12% Bis–Tris gel, transferred to a nitrocellulose membrane, and probed with specific antibodies.

Cell–cell fusion assay

BS-C-1 cells were infected with WT virus or O3 mutants at a multiplicity of 3 PFU/cell. After 16 h, the cells were treated with pH 5.0 buffer for 3 min, washed and incubated in fresh medium for another 2 h. Syncytia were visualized by light microscopy.