Characterization of the Second Apoptosis Inhibitor Encoded by Guinea Pig Cytomegalovirus

ABSTRACT

Despite the usefulness of guinea pig cytomegalovirus (GPCMV) for studies on congenital CMV infection, its viral mechanisms for the evasion of host defense strategies have not been fully elucidated. We reported previously that GPCMV gp38.1 functions as a viral mitochondria-localized inhibitor of apoptosis-like function, and its weak activity suggested the presence of an additional inhibitory molecule(s). Here, we identified gp38.3-2, a 42-amino-acid (aa) reading frame embedded within the gp38.3 gene that encodes a positional homolog of murine CMV (MCMV) m41. Characterization of gp38.3-2 resulted in the following findings: (i) the aa sequence of gp38.3-2 shows some similarity to that of MCMV m41.1, a viral inhibitor of oligomerization of a member of Bcl-2 family protein BAK, but there is no correspondence in their predicted secondary structures; (ii) gp38.3-2, but not gp38.3, showed inhibitory activities against staurosporine-induced apoptosis; (iii) three-dimensional protein complex prediction suggests that the N-terminal α-helix of gp38.3-2 interacts with residues in the BH3 and BH1 motifs of BAK, and analysis of gp38.3-2 and BAK mutants supported this model; (iv) guinea pig fibroblast cells infected with gp38.3-2-deficient GPCMV strain Δ38.3-2 died earlier than cells infected with rescued strain r38.3-2, resulting in lower yields of Δ38.3-2; (v) Δ38.3-2 exhibited a partial but significant decrease in monocyte and macrophage infection in comparison with r38.3-2; and, however, (vi) little difference in the viral infection of guinea pigs was observed between these two strains. Therefore, we hypothesize that gp38.3-2 contributes little to the evasion of host defense mechanisms under the experimental conditions used.

IMPORTANCE Although GPCMV provides a useful animal model for studies on the pathogenesis of congenital CMV infection and the development of CMV vaccine strategies, our understanding of the viral mechanisms by which it evades apoptosis of infected cells has been limited in comparison with those of murine and human CMVs. Here, we report a second GPCMV apoptosis inhibitor (42 amino acids in length) that interacts with BAK, a Bcl-2 family proapoptotic protein. Three-dimensional structural prediction indicated a unique BAK recognition by gp38.3-2 via the BH3 and BH1 motif sequences. Our findings suggest the potential development of BH3 mimetics that can regulate inhibition or induction of apoptosis based on short ~40-amino-acid peptide molecules as with GPCMV.

INTRODUCTION

As congenital infection with human cytomegalovirus (HCMV) often causes birth defects and developmental abnormalities, the development of effective vaccines is important for its prevention (1). Guinea pig CMV (GPCMV) provides a useful model for understanding the pathogenesis of congenital CMV infection and for the development of vaccine strategies, as GPCMV, but not other small animal CMVs, infection occurs in utero, and disease outcomes with GPCMV develop in a manner similar to those observed in newborns with congenital HCMV infection (2–4).

CMVs employ diverse strategies to escape host antiviral responses. Some of the evasion mechanisms employed by HCMV are also reported for GPCMV (5–8). In addition to the activation of both innate and adaptive immune pathways, upon viral infection, the host cells often induce cell death to limit viral spread (9, 10). For example, the intrinsic cell death pathway, which is activated not only by cellular stresses but also by viral infection, is regulated by members of the B-cell lymphoma 2 (Bcl-2) family of proteins (11). These Bcl-2 family proteins have one to four α-helical regions with certain degrees of sequence homology, which interact with their own or other Bcl-2 members. The Bcl-2 family proteins consist of antiapoptotic proteins, such as Bcl-2, as well as the proapoptotic proteins BAK, BAX, and BH3-only proteins. To ensure their survival, many viruses encode a variety of proteins to prevent cell death by modulating proapoptotic Bcl-2 proteins (12–14). In the case of CMVs, HCMV viral mitochondria-localized inhibitor of apoptosis-like function (vMIA) encoded by UL37ex1 directly inactivates both BAX and BAK, while murine CMV (MCMV) utilizes vMIA encoded by m38.5 and a viral inhibitor of BAK oligomerization (vIBO) encoded by m41.1 for the inhibition of BAX and BAK, respectively (15–22). MCMV vIBO, together with vMIA, is required for optimal viral replication and has a modest impact on dissemination in infected mice (23–26). In our previous study, we characterized the GPCMV GP38-gp38.4 locus and found that the gp38.1 product functioned as a vMIA homolog both in vitro and in vivo (27). The gp38.1 product was found to bind to BAX but not to BAK. Although the lack of gp38.1 expression reduced viral growth in cultured cell lines and viral dissemination in infected guinea pigs, these activities were not robust, suggesting the presence of an additional inhibitory molecule(s), such as a BAK-specific inhibitor similar to that in MCMV (27). In this study, we identified the gp38.3-2 gene as the second apoptosis inhibitor of GPCMV.

RESULTS

Prediction of the second guinea pig apoptosis inhibitor locus.

Based on comparison of gene alignment in the GPCMV GP38-gp38.4 region with those of the HCMV UL38-UL42 region and the MCMV M38-m42 region (Fig. 1A), we speculated that gene gp38.2 or gp38.3 would contain an open reading frame (ORF), corresponding to MCMV m41.1, that interacts with BAK and is required for optimal replication in vitro and in vivo. The gp38.2 amino acid (aa) predicted sequence shows weak similarity to those of the HHV-6 and HHV-7 U21 genes that encode immunoevasins downregulating MHC class I antigen presentation (28). Therefore, we examined the similarity of the primary aa sequence of the 42-aa ORF gp38.3-2, which was embedded in the gp38.3 gene (Fig. 1B), with that of m41.1. Needleman-Wunsch pairwise sequence alignment (EMBL-EBI EMBOSS Needle with the following parameters: datafile EBLOSUM62, gapopen 10, gapextend 1, endoweight, endopen 10, endextend 1) showed a weak sequence identity (22%) and similarity (31%) between gp38.3-2 and m41.1 (Fig. 1C). However, the secondary structures predicted by the JPred server showed differences in terms of both the number and length of their α-helices (Fig. 1C, highlighted in pink). There are two known GPCMV strains, 22122 (used in this study) and CIDMTR (29). Between the two strains, DNA and aa sequences of gp38.3-2 ORF differ in 1 out of 126 bases (0.8%) and 1 out of 42 aa residues (2.4%), respectively, suggesting conservativeness of the ORF. The difference in the gp38.3-2 aa sequence is Ser versus Phe at position 27.

FIG 1.

Sequence analyses of gp38.3-2. (A) Schematic comparison of the GPCMV GP38-gp43 region with the corresponding regions of HCMV and MCMV. The ORFs of gp38.3-2 and its positional homolog m41.1 are colored in red, and vMIAs are in blue. Adapted from reference 27. (B) Nucleotide and aa sequences of gp38.3 and those of gp38.3-2 (shown in red). Numbers indicate the nucleotide positions in the GPCMV genome (44). (C) Pairwise homology search result between gp38.3-2 and m41.1. “|” indicates an identical base, and “:” and “.” indicate >0 and >−2 in the Blosum62 matrix, respectively. Positions of α-helices predicted by JPred (Jnet version 2.3.1; UniRef90 release 2014) (45) are shown by pink highlights.

gp38.3-2 in the gp38.3 locus has an antiapoptotic function.

To determine whether the gp38.3 locus encodes an antiapoptotic protein(s) and whether the antiapoptotic activity of the gp38.3 locus can be ascribed to the gp38.3-2 ORF, we generated pcDNA-Δgp38.3-2, in which a nucleotide sequence alteration from T to C at position 66340 (based on GPCMV sequence) destroyed the start codon ATG of gp38.3-2 but kept the aa sequence of gp38.3 unchanged since the GAT-to-GAC alteration is a silent mutation (Fig. 1B). The induction of apoptosis and necrosis by staurosporine (STS) in the 293T cells transfected with pcDNA-gp38.3, which encodes both the gp38.3 and gp38.3-2 products, pcDNA-Δgp38.3-2, which encodes only the gp38.3 product, or pcDNA3 vector were evaluated by flow cytometry using Cell Event caspase-3/7 green detection reagent. Examples of the assay outcomes and the mean percentages of apoptotic cells or necrotic cells obtained in triplicate experiments are shown in Fig. 2A and B, respectively. The results clarified that the gp38.3-2 ORF contributed to the antiapoptotic activity.

FIG 2.

gp38.3-2 has an antiapoptotic function. (A, B) 293T cells were transfected with pcDNA-gp38.3 (38.3), pcDNA-Δgp38.3-2 (Δ38.3-2), or pcDNA3 empty vector (vec). Forty-eight hours later, the cells were treated with 5 μM staurosporine (STS) and cultured for an additional 12 h. The cells were then stained with Cell Event Caspase-3/7 green reagent along with SYTOX AADvanced dead cell staining reagent. Examples of the assay outcomes (A) and means ± SDs of the percentages of apoptotic cells or necrotic cells obtained in triplicate experiments (B) are shown. (C) 293T cells were transfected with pcDNA-gp38.3-2F (38.3-2F), -Δgp38.3-2F (Δ38.3-2F), or pcDNA3 empty vector (vector). Forty-eight hours later, the cells were harvested. The cell lysates were analyzed in immunoblotting with anti-FLAG monoclonal antibody (clone M2; Sigma-Aldrich). (D) 293T cells were transfected with pcDNA-gp38.3 (gp38.3), pcDNA-Δgp38.3-2 (Δ38.3-2), pcDNA-gp38.3-2F (38.3-2F), pcDNA-Δgp38.3-2F (Δ38.3-2F), or an empty vector pcDNA3 (vec). Forty-eight hours later, the cells were untreated or treated with 5 μM STS for 16 h. The cells were then stained with Apo-15 and analyzed by flow cytometry. Means ± SDs of the percentage of Apo-15-positive cells in the singlet cell population obtained in triplicate experiments are shown. ****, P < 0.0001; ***, P < 0.001; *, P < 0.05; ns, not significant. The ATG-to-ACG mutation in Δ38.3-2 and FLAG tag in 38.3-2F and Δ38.3-2F are shown in Fig. 1B.

As no antibodies were available to gp38.3-2, a FLAG tag sequence was introduced in front of the stop codon of gp38.3-2 in pcDNA-gp38.3 and pcDNA-Δgp38.3-2 (Fig. 1B), resulting in pcDNA-gp38.3-2F and pcDNA-Δgp38.3-2F, respectively. Immunoblotting analysis of cell lysates, which were prepared by transfection with pcDNA-gp38.3-2F or pcDNA-Δgp38.3-2F, demonstrated that anti-FLAG tag monoclonal antibody detected a product in cells transfected with pcDNA-gp38.3-2F, but not with pcDNA-Δgp38.3-2F (Fig. 2C). To evaluate the antiapoptotic activity of FLAG-tagged gp38.3-2, 293T cells transfected with a pcDNA3 vector, pcDNA-gp38.3, pcDNA-Δgp38.3-2, pcDNA-gp38.3-2F, or pcDNA-Δgp38.3-2F were treated with STS, and the population of apoptotic cells was analyzed by flow cytometry using Apotracker Green. The mean percentages of Apo-15-positive apoptotic cells obtained from triplicated experiments are shown in Fig. 2D. Addition of the FLAG tag at the carboxyl end of gp38.3-2 did not affect the inhibition of STS-induced apoptosis.

gp38.3-2 exhibits antiapoptotic activities.

To confirm that there is no involvement of gp38.3 ORF in the antiapoptotic activities of gp38.3-2, we generated plasmids encoding gp38.3-enhanced green fluorescent protein (EGFP) or gp38.3-2-EGFP. Guinea pig lung fibroblasts (GPLs) were transfected with these plasmids or a pEGFP-N1 vector and untreated or treated with STS, and then the inhibitory activities of the gene products on apoptosis were observed by the fluorescent dUTP-based terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) method. Images captured by a cell imager demonstrated that transfection of a plasmid encoding gp38.3 or gp38.3-2 led to an obvious decrease in the TUNEL signals (Fig. 3), indicating that gp38.3-2 alone is enough for antiapoptotic activities.

FIG 3.

Antiapoptotic activities of gp38.3-2-EGFP fusion protein. GPL cells were transfected with a plasmid expressing EGFP-tagged gp38.3, EGFP-tagged gp38.3-2, or EGFP. Forty-eight hours later, apoptotic cells were labeled with fluorescence-dUTP using a TUNEL assay kit and counterstained with DAPI. Images with fluorescent signals, including DAPI (blue) for the nuclei, EGFP (green) for fused proteins, and TUNEL (red) for apoptosis, were captured using a ZOE fluorescent cell imager. Bars, 100 μm.

Structural predictions of gp38.3-2.

Although the lack of any homologs of gp38.3-2 limited the reliability of its three-dimensional (3D) structure prediction, we performed Colab Alphafold2 analysis of the monomer of gp38.3-2 and Alphafold-Multimer analysis of heterodimer formation between gp38.3-2 and gpBAKΔTM (gpBAK lacking the transmembrane domain) on the assumption that gp38.3-2 interacts with guinea pig BAK (gpBAK). The generated rank 1 model predicted the interaction of the α-helix 1 of gp38.3-2 with α-helices 3, 4, and 5 of gpBAKΔTM (Fig. 4A). Predicted aligned error was reasonably low in the complex covering gpBAKΔTM and the first α-helix of gp38.3-2 (Fig. 4B). To visualize the interacting residues between gp38.3-2 and gpBAKΔTM, we conducted Edmundson wheel analysis of the predicted model in combination with measurement of the distances between two aa residues by the Zone and Distance functions of Chimera. The α-helix 1 of gp38.3-2 was predicted to interact with the BH3 and BH1 motifs located on α-helices 3, 4, and 5 of gpBAK (Fig. 4C to F).

FIG 4.

Prediction of the 3D structure of the complex of gp38.3-2 and gpBAK. (A) Structures of a heterodimer between gp38.3-2 (aa 1 to 42) and gpBAK without the transmembrane domain (aa 1 to 182) predicted by Colab Multimer are shown from two directions. Rank 1 model among 5 models was used for identification of the gpBAK residues positioning within 5 Å from the α-helix 1 (aa 2 to 8) of gp38.3-2. The α-helix 1 and the other part of gp38.3-2 are colored in red and yellow, respectively. The gpBAK residues of α-helices 3, 4, and 5 that interact with gp38.3-2 are highlighted in green, blue, and cyan, respectively. (B) Plot of predicted aligned error (PAE) of the model shown in panel A. (C) Pairs of particular residues on the α-helix 1 (α1) of gp38.3-2 and those on α-helix 3, 4, or 5 (α3, α4, or α5) of gpBAK with distances less than 4 Å in the predicted model are indicated with line-connected circles in the Edmondson wheel drawings of these four α-helices. (D) Residues of gp38.3-2 within a 4-Å distance from gpBAK are colored in red. Alterations introduced as m1 and m2 in gp38.3-2 are shown. (E) Structure predicted as the gp38.3-2-m1-gpBAK complex (light blue) was compared with the wild-type (wt) complex (yellow) using the MatchMaker function of Chimera. (F) Residues of gpBAK within a 4-Å distance from gp38.3-2 are colored as panel A. Mouse (mo) and human (hu) BAK sequences are aligned with the gpBAK sequence, and the BH3 and BH1 motifs are indicated.

To examine whether this prediction, i.e., that gp38.3-2 interacts with BAK and inhibits its apoptotic function, is experimentally supported, two kinds of mutated gp38.3-2 were designed (Fig. 4D). One of them, gp38.3-2-m1 (m1), contained an alteration from KRR to AAA at positions 15 to 17, in which the structure of α-helix 1 was predicted to be conserved, and the structure of the complex of gpBAK and the KRR15-17AAA mutant was, indeed, predicted to be almost identical to that with the parental gp38.3-2 except for an extension of the α-helix 1 of gp38.3-2 (Fig. 4E). In contrast, the software could not generate a reasonable prediction of the complex of gpBAKΔTM and gp38.3-2-m2 that contained an alteration from L to P at position 4 (L4P), as L4 is essential for the interactions with Y82, D85, and M89 nearby the BH3 motif of gpBAKΔTM (Fig. 4C).

Effects of the gp38.3-2 mutations on antiapoptosis activity and cellular localization.

To examine the effects of alterations in gp38.3-2 on antiapoptosis activity, 293T cells were transiently transfected with a plasmid expressing EGFP, wild-type (wt) gp38.3-2-EGFP, or mutated gp38.3-2-EGFP and treated with STS, and their degree of apoptosis was analyzed by TUNEL assay. Results revealed that gp38.3-2-m2 (m2), but not m1, showed a decrease in inhibitory activity against STS-induced apoptosis (Fig. 5A), supporting the hypothesis that the interaction of gp38.3-2 with BAK is required for its antiapoptosis activity.

FIG 5.

Effects of the gp38.3-2 mutations on antiapoptosis activity and cellular localization. (A) 293T cells in 96-well plates were transfected with pEGFP-N1(vec), pEGFP-gp38.3-2-wt, gp38.3-2-m1, or gp38.3-2-m2 and, 48 h later, untreated or treated with STS at a final concentration of 1 μM for 1 h, and apoptotic cells were detected by the TUNEL assay using fluorescence dUTP. Means ± SEMs of the percentage of apoptotic cells obtained in triplicate wells are shown. *, P < 0.05. (B, C) pEGFP-N1 or a plasmid expressing gp38.3-2-wt, gp38.3-2-m1, or gp38.3-2-m2 fused with EGFP was transfected into GPL cells in two wells per set of conditions of 8-well chamber slides using Fugene and, 48 h later, untreated or treated with STS at a final concentration of 0.5 μM for 1 h. The cells were then fixed with 4% paraformaldehyde, counterstained with DAPI, and observed under a confocal microscope. Images of five fields were captured for each well. (B) An example of the image for each condition is shown. (C) GFP-positive cells were classified blindly into cells with fibrous signals in the cytoplasm and those with nuclear signal alone, and percentages of cells with fibrous signals in the cytoplasm among GFP-positive cells obtained from the duplicated wells are shown as two closed circles per set of conditions. The average percentages are shown with horizontal bars. (D) 293T cells were transfected and treated with STS as described above for panel B. Then, the cells were harvested, and the cell lysates, cytoplasmic fraction. and nuclear extracts were prepared as described in Materials and Methods. The same volume of samples was analyzed by immunoblotting with anti-GFP antibody. The cytoplasmic fraction and nuclear extracts were concentrated 0.82 and 9.4 times the cell lysates, respectively.

Next, prior to an analysis of the cellular colocalization of gp38.3-2 with BAK, we examined the effects of the gp38.3-2 mutations on localization. GPL cells were transfected with a plasmid expressing wt or mutated gp38.3-2-EGFP (Fig. 5B and C). Wild-type (wt) gp38.3-2 proteins were detected in the nuclei and partly around the nuclei as fibrous signals in the cells without STS treatment, suggesting its nuclear and mitochondrial-like localization. Further, m1 proteins were found to be localized in a pattern like that of the wt, although m2 was present mainly in the nuclei. In addition to the nuclear signals, STS treatment of the transfected GPL cells increased the fibrous signals in the cytoplasm of the cells expressing wt and m1 modestly, while m2 was mostly in the nuclei.

To examine the localization of gp38.3-2 more objectively, 293T cells were transfected with pEGFP-N1 (EGFP), pEGFP-gp38.3-2 wt, m1, or m2, and their lysates and cellular fractions were analyzed in immunoblotting with anti-GFP antibody (Fig. 5D). Difference in the amounts of gp38.3-2 in cell lysates between wt and mutants was within 4-fold. In the absence of STS treatment, ratios of the amount in the nuclei per that in the cytoplasm among wt, m1, and m2 were in a range of 0.05 to 0.11, indicating that almost 90% of gp38.3-2 was in the cytoplasmic fraction. EGFP was detected mainly in the cytoplasm and slightly in the nuclei. Even after STS treatment, wt and m1 were localized mainly in the cytoplasm, although m2 behaved differently from wt and m1. In the case of m2, STS treatment increased the nuclear signals more than 6 times. The immunoblotting results suggest that detection of gp38.3-2 mainly in the nuclei in the fluorescent protein imaging was probably due to overlooking weak signals of gp38.3-2 localized in the cytosol.

Interaction of gp38.3-2 with BAK.

To examine whether gp38.3-2 colocalizes with BAK, GPL cells were transfected with the plasmids for EGFP-tagged gp38.3-2 and red fluorescent protein (RFP)-tagged gpBAK. Confocal microscopic analysis using the z-stack function demonstrated that all gpBAK signals appeared to be localized on the fibrous gp38.3-2 signals (Fig. 6A).

FIG 6.

Interaction of gp38.3-2 with BAK. (A) A plasmid expressing gp38.3-2 fused with EGFP and/or a plasmid expressing BAK fused with RFP were transfected into GPL cells. Forty-eight hours later, the cells were fixed with 4% paraformaldehyde, counterstained with DAPI, and observed under a confocal microscope. Images of the cotransfected cells were captured using the z-stack function. (B) 293T cells were transfected with pEGFP-N1 (control), pEGFP-gp38.3-2-wt (wt), gp38.3-2-m1 (m1), or gp38.3-2-m2 (m2) along with pRFP-gpBAKΔTM and, 48 h later, untreated or treated with STS at a final concentration of 0.5 μM for 1 h. One milliliter of cell extracts (pre-IP) was immunoprecipitated with anti-GFP antibodies, and the proteins bound to protein G resin were eluted into 30 μL of gel loading buffer (post-IP). Pre- and post-IP samples were analyzed by immunoblotting with anti-huBAK antibody and with anti-GFP antibody. The ratio of the normalized intensity for each set of conditions against that of wt was calculated as described in Materials and Methods. (C) 293T cells were transfected with pEGFP-gp38.3-2-wt (wt) along with pRFP-gpBAKΔTM or -gpBAKΔTM I86P and, 48 h later, untreated or treated with STS at a final concentration of 0.5 μM for 1 h. Cell extracts (pre-IP) were immunoprecipitated with anti-GFP antibodies, and the pellets (post-IP) were analyzed by immunoblotting with anti-huBAK antibody. **, P < 0.01; *, P < 0.05; ns, not significant.

To clarify whether gp38.3-2 and gpBAK form a complex, 293T cells were transfected with a pEGFP-N1 vector encoding EGFP-tagged gp38.3-2 and the plasmid encoding RFP-tagged gpBAKΔTM, which lacks the transmembrane domain (aa 183 to 214) of gpBAK to increase solubility. The cells were untreated or treated with STS for 1 h at 48 h posttransfection. The cell extracts (pre-IP) were reacted with anti-GFP antibody to immunoprecipitate gp38.3-2 and its mutant forms. The precipitated products (post-IP) were analyzed by immunoblotting using anti-human BAK (huBAK) antibody. As shown in Fig. 6B, gpBAK formed a complex with wt gp38.3-2 and mutant m1 (from KRR to AAA), but not with GFP control (vec) or mutant m2 (L4P). In addition, STS treatment increased the amounts of coprecipitated BAK products (Fig. 6B). Immunoblotting of the same post-IP products using anti-GFP antibody ensured that the amounts of GFP fusion proteins in the post-IP products were in parallel with those in the cell extracts (pre-IP) and that the lack of BAK products in the post-IP products of m2 (L4P) was not due to the less amount of gp38.3-2 m2 fusion protein in the pre-IP sample.

To confirm that the interaction of gp38.3-2 with gpBAK is indeed specific, the complex formation of gp38.3-2 with gpBAK containing an alteration of Ile at position 86 to Pro (I86P) was evaluated since I86 of gpBAK was predicted to interact with Phe at position 8 (F8) of gp38.3-2 (Fig. 4C and F). Extracts of 293T cells transfected with pEGFP-gp38.3-2, along with the plasmid encoding RFP-tagged gpBAKΔTM or gpBAKΔTM I86P, were prepared, and post-IP products of the extracts using anti-GFP antibody were analyzed in immunoblotting using anti-huBAK antibody. The complex formation between gp38.3-2 and gpBAK was lost by the I86P alteration (Fig. 6C). Although the results strengthened the predicted three-dimensional (3D) model that gp38.3-2 interacts with gpBAK physically through the BH3-BH1 motif, the results cannot clarify whether the interaction is exclusive for BAK and how the interaction affects biological outcomes.

Function of gp38.3-2 in viral infection in cell cultures.

To examine whether the antiapoptotic activity of gp38.3-2 is required for efficient infection in cell cultures, GPCMVd9kΔ38.3-2, which lacks the initiation codon of gp38.3-2, and GPCMVd9k r38.3-2, in which the wild-type gp38.3-2 sequence was rescued, were prepared.

First, the viabilities of GPL cells infected with GPCMVd9k WT, Δ38.3-2, r38.3-2, or Δ38.1 at a multiplicity of infection (MOI) of 2 were compared. Percentages of cell survival based on intrinsic ATP amounts of the cells infected with Δ38.3-2 decreased earlier than those of the cells infected with r38.3-2 or WT (Fig. 7A), suggesting that gp38.3-2 plays a role in the survival of infected cells. There was no significant difference in cell survival percentages between cells infected with Δ38.3-2 and those with Δ38.1 (Fig. 7A).

FIG 7.

Function of gp38.3-2 for viral infection in cell cultures. (A–C) Effects of gp38.3-2 on cell survival, apoptosis, and viral yields in fibroblasts. (A) GPL cells were infected with GPCMVd9k WT, Δ38.1, Δ38.3-2, or r38.3-2 at an MOI of 2. Survival of the infected cells at the indicated time points after infection was determined by a commercial kit based on intrinsic ATP amounts (CellTiter-Glo luminescent cell viability assay; Promega). The measurements at 0 h were defined as 100% survival. Means ± SEMs of the percentages of cell survival obtained from triplicated wells are shown. (B) At 72 h after the infection of GPL cells with GPCMVd9k Δ38.3-2 or r38.3-2 at an MOI of 0.5, apoptosis in the infected cells was monitored using a commercial TUNEL assay kit. Means ± SEMs of the ratios between TUNEL- and GFP-positive cell numbers obtained in triplicate experiments are shown. (C) Culture supernatants of GPL cells infected with GPCMVd9k-Δ38.3-2 (Δ38.3-2) or GPCMVd9k-r38.3-2 (r38.3-2) at an MOI of 0.01 were collected at the indicated time points. Means and SDs of the viral titers in the supernatants from triplicated wells are shown. (D) SV40 T antigen-immortalized guinea pig epithelial cells (GPE-7) in the triplicated wells were infected with Δ38.3-2 or r38.3-2 at an MOI of 3. Two days later, the numbers of GFP-expressing cells were counted as viruses encoding a GFP-expressing cassette. Individual counts and means ± SDs are shown. (E) Monocytes prepared from the guinea pig spleen and macrophages prepared by treatment of the monocytes with phorbol myristate acetate (PMA) for 2 days were infected at an MOI of 3. Two days later, the numbers of GFP-positive cells were measured. **, P < 0.01; *, P <0.05; ns, not significant.

To further clarify whether the antiapoptotic function of gp38.3-2 results in differences in such cell survival, GPL cells were infected with r38.3-2 or Δ38.3-2 at an MOI of 0.5, and nuclear DNA fragmentation at 72 h after infection was visualized by TUNEL assay. The ratio of TUNEL-positive cells among GFP-positive cells infected with r38.3-2 was 10% that of cells infected with Δ38.3-2 (Fig. 7B).

Next, the effects of the defect in gp38.3-2 on viral replication kinetics were analyzed by titration of viruses in culture supernatants of infected GPL cells. Yields of infection with GPCMV r38.3-2 were about 5-fold higher than those with Δ38.3-2, and the differences were statistically significant (P < 0.05 or 0.01) at most time points (days 3 to 9), suggesting that gp38.3-2 plays a role in viral growth in cell cultures (Fig. 7C). There was a tendency, although not statistically significant, for the number of infected epithelial GPE7 cells to differ (Fig. 7D). To evaluate the effects of the defect in gp38.3-2 on the efficiency of infection of splenic monocytes and macrophages, the cells were plated into the wells of 96-well plates and infected with the recombinant GPCMVs at an MOI of 3, and 48 h later, the number of GFP-positive cells was measured (Fig. 7E). The results showed that there were significant differences (P < 0.05 and P < 0.01) in the numbers of infected monocytes and macrophages, respectively.

gp38.3-2 was not required for pathogenic outcomes in animals.

To investigate the effect of gp38.3-2 on pathogenic outcomes, normal guinea pigs were infected intraperitoneally with 5 × 10⁶ IUs/animal of GPCMVΔ38.3-2 or r38.3-2 and sacrificed on day 6 after infection. Viral loads in specimens from visceral organs and blood were determined by quantitative PCR (Fig. 8). The results showed that gp38.3-2 did not affect viral loads in a significant manner.

FIG 8.

Female 3-week-old guinea pigs were infected intraperitoneally with 5 × 10⁶ IU of GPCMVd9k-Δ38.3-2 (Δ38.3-2) or GPCMVd9k-r38.3-2 (r38.3-2). At 6 days after infection, the animals were sacrificed, and the copy numbers of viral DNA per 10⁶ cells in the visceral organs and the whole-blood specimens were measured by quantitative PCR (qPCR) assays for cellular actin and GPCMV GP83 genes. Means ± SEMs are shown.

DISCUSSION

This study demonstrated the following: (i) despite marginal aa sequence homology between GPCMV gp38.3-2 and MCMV m41.1, gp38.3-2 showed antiapoptotic activities; (ii) the 3D structural prediction suggested complex formation of gp38.3-2 and BAK in a manner in which the N-terminal α-helix 1 of gp38.3-2 interacts with the residues in the BH3 and BH1 motifs of BAK; (iii) colocalization as observed by confocal microscopic analysis, as well as immunoprecipitation followed by immunoblotting, demonstrated an interaction between gp38.3-2 and BAK proteins, and the loss of interaction by the I86P alteration in BAK supported the specificity of interaction; (iv) gp38.3-2-m2 (L4P), but not gp38.3-2-m1 (KRR to AAA at positions 15 to 17), lost antiapoptotic activities, fibrous signals in the cytoplasm, and complex formation with BAK, supporting the 3D structural prediction; and finally, (v) the effects of antiapoptotic activities of gp38.3-2 in the context of viral infection in cell cultures and animals were limited.

GPCMV is phylogenetically separated both from HCMV and from MCMV (30). Some functions, such as the pentamer (gH/gL/UL128s)-dependent entry in epithelial and endothelial cells, are conserved between GPCMV and HCMV, but not in MCMV (31). In the case of the inhibitory mechanisms for Bcl-2 family-dependent apoptosis, GPCMV resembles MCMV, as MCMV uses vMIA for the inhibition of BAX and m41.1 (vIBO) for that of BAK, while HCMV vMIA inhibits both BAX and BAK (20). Interestingly, the primary sequences of vMIA and m41.1 of MCMV have extremely limited homologies with those of GPCMV, and the 2D and 3D structures predicted for m41.1 and gp38.3-2 showed little similarity. Therefore, it may not be surprising that the cellular localizations and mechanisms for BAK recognition differ.

Most antiapoptotic proteins encoded by several viruses, including adenovirus and human herpesviruses, show conservation of sequence homologies with Bcl-2 family proteins (32). Although some viral antiapoptotic proteins, including myxoma virus M11L and vaccinia virus F1L, do not share obvious sequence similarity with any known cell death modulators, they have a protein fold virtually identical to Bcl-2, which allows their interaction with the BH3 motif (33). Although it is merely a computational prediction, this study proposed a third way by which viral apoptosis modulators access Bcl-2 family proteins; that is, a short α-helix in a small viral protein that interacts with the sequences overlapping the BH3 and BH1 motifs. The introduction of an alteration to break the α-helix for the interaction with BAK resulted in nuclear localization, a lack of BAK binding, and STS-induced apoptosis inhibitor activities. These experimental outcomes are consistent with the computational prediction. Targeting the BH3 motif-binding domain of antiapoptotic Bcl-2 family proteins with BH3 mimetics has proven useful for the induction of apoptosis, which makes BH3 mimetics important among apoptosis-dependent therapeutics (34, 35). Screening for drugs that specifically bind to BAK suggested that drug binding to BH3 is critical to increasing drug specificity (36). Further studies on the structural characteristics of the α-helix 1 of gp38.3-2 are expected to shed light on the development of BH3 mimetics.

It was reported that the lack of m41.1 expression reduced the viability of fibroblasts infected with a recombinant MCMV at a high MOI followed by STS treatment (24, 25). In addition, the lack of m41.1 expression led to a slight delay in viral growth in fibroblasts but not in endothelial cells (23), which is consistent with our observation regarding recombinant GPCMV lacking gp38.3-2 expression. Although the lack of m41.1 decreased viral growth significantly in macrophage cell lines (23–25), we demonstrated a significant decrease in the efficiency of infection of primary splenic monocytes and macrophages due to the lack of gp38.3-2 in this study, as guinea pig macrophage cell lines were not available. Although the lack of gp38.3-2 made a slight but significant difference in viral growth in vitro, no significant differences in viral loads in vivo were observed between GPCMVd9k r38.3-2 and Δ38.3-2. In the case of MCMV lacking m41.1 expression, viral loads in the lungs and salivary glands at 10 or 14 days after infection were decreased to 1/10 of those for MCMV expressing m41.1 (23–25). It is unclear whether the negative results for gp38.3-2 in guinea pigs were due to the small number of animals that we were allowed to use or whether the antiapoptotic activities of gp38.3-2 were weaker than those of m41.1. Taking account of the findings that a recombinant GPCMV lacking the expression of BAX inhibitory protein gp38.1 also showed significant differences in viral loads only in the spleen (27) and that GPCMV does not encode a homolog of viral inhibitor of caspase-8-induced apoptosis (vICA) encoded by HCMV UL36 or MCMV M36, it seems that antiapoptotic proteins of GPCMV have limited roles in viral propagation and pathogenicity in vivo compared with those of MCMV. However, it would be important to note the limitations of our study. The activities of gp38.3-2 may depend on genetic characteristics of GPCMV strain used, cellular types and expression level of cellular factors, route of inoculation, or immunological factors. GPCMVd9K WT and its derivatives, including GPCMVd9k Δ38.3-2 and r38.3-2, prepared from pBAC-GPCMVd9K lacks a 9-kb region of the GPCMV genome (31). The GP1 gene encoding a macrophage inflammatory protein 1 (MIP-1) homolog is encoded in the 9-kb region, and the lack of MIP-1 attenuated in vivo phenotypes of GPCMV (37). We confirmed that GPCMV SG, a strain that contains the 9-kb region and has the full capacity to grow in animals, showed a more virulent viral growth phenotype than GPCMVd9K WT (5). Therefore, we cannot exclude the possibility that viruses with a more virulent background exhibit more robust effects of gp38.1 and gp38.3-2 on apoptosis. As gp38.1 may be functionally redundant with gp38.3-2, gp38.1 may mask or diminish the measured effects of gp38.3-2. It would be desirable to evaluate the phenotypes of gp38.3-2 and gp38.1 in vitro or in vivo in the context of gp38.1-null and gp38.3-2-null viruses, respectively. In addition, the lack of key materials, such as guinea pig cell lines deficient in BAK or BAX, left room for argument on the functional specificity of interaction of gp38.1 or gp38.3-2 with the Bcl-2 family proteins, including BAX and BAK.

In conclusion, we identified gp38.3-2 as the second antiapoptotic protein of GPCMV, and it is likely that it functions through binding to the BH3 and BH1 motifs of BAK. Further studies are required for elucidation of the pathogenic role of gp38.3-2 in viral infection both in vitro and in vivo, although the antiapoptotic activities of gp38.3-2 played a limited role under the experimental conditions used in this study.

MATERIALS AND METHODS

Plasmids.

The gp38.3 ORF fragment was prepared from the cDNA of GPCMV-infected GPL cells by PCR amplification with primers P1 and P2 (Table 1) and cloned between the EcoRI and NotI sites of pcDNA3, resulting in pcDNA-gp38.3 that encodes gp38.3 tagged with a FLAG epitope at the carboxyl terminus. The gp38.3 ORF fragment prepared by PCR amplification with primers P1 and P3 was cloned between the EcoRI and SalI sites of pEGFP-N1 (Clontech) to obtain pEGFP-gp38.3 encoding gp38.3 fused with EGFP at the carboxyl terminus. pcDNA-Δgp38.3-2, in which the start codon of gp38.3-2 was knocked out without any change to the gp38.3 aa sequence, was generated by site-directed mutagenesis of pcDNA-gp38.3 using a commercial kit (QuikChange mutagenesis kit; Stratagene) with primers P4 and P5. The FLAG tag sequence was introduced in front of the stop codon of gp38.3-2 in pcDNA-gp38.3 and pcDNA-Δgp38.3-2 using the same mutagenesis kit described above with primers P6 and P7, resulting in pcDNA-gp38.3-2F and pcDNA-Δgp38.3-2F, respectively. The gp38.3-2 ORF fragment was prepared from the gp38.3 ORF fragment by PCR amplification with primers P8 and P9 and cloned between the EcoRI and SalI sites of pEGFP-N1 (Clontech) to obtain pEGFP-gp38.3-2 that encodes gp38.3-2 tagged with EGFP at the carboxyl terminus. The KRR sequence (aa position 15 to 17) in the gp38.3-2 gene was replaced with the AAA sequence using the commercial kit described above with primers P10 and P11, resulting in pEGFP-gp38.3-2-m1. pEGFP-gp38.3-2-m2, containing an alteration of Leu to Pro at aa position 4 (L4P), was generated similarly with primers P12 and P13. pRFP-gpBAK that encodes guinea pig BAK (gpBAK) fused with TurboRFP was described previously (27). The transmembrane domain (aa 183 to 208) of gpBAK was truncated by mutagenesis using the commercial kit described above with primers P18 and P19, resulting in pRFP-gpBAKΔTM. pRFP-gpBAKΔTM containing an alteration of Ile to Pro at aa position 86 (I86P) was generated similarly with primers P20 and P21.

TABLE 1.

Primers used in this study^a

Code	Sequence (5′ to 3′)	Usage
P1	gcggaattcgccgccATGCGCGCGATTTCGGTCGT	Cloning of pcDNA-gp38.3 and pEGFP-gp38.3
P2	cgcgcggcggccgcTCActtgtcatcgtcgtccttgtagtcATTGACGGAGCTGCTGT	Cloning of pcDNA-gp38.3
P3	cgcgtcgacttATTGACGGAGCTGCTGTATT	Cloning of pEGFP-gp38.3
P4	TCTGTCCCTACTTCCGAc GACGTTTCTACTTCACC	Mutagenesis of the initiation codon (Δgp38.3-2)
P5	GGTGAAGTAGAAACGTCg TCGGAAGTAGGGACAGA	Mutagenesis of the initiation codon (Δgp38.3-2)
P6	GGCACGGATTTTCCGATCAgactacaaggacgacgatgacaagTAGATCTTTCACCGATAGA	Insertion of FLAG tag at the C-terminal end (gp38.3-2F and Δgp38.3-2F)
P7	TCTATCGGTGAAAGATCTActtgtcatcgtcgtccttgtagtcTGATCGGAAAATCCGTGCC	Insertion of FLAG tag at the C-terminal end (gp38.3-2F and Δgp38.3-2F)
P8	gcggaattcgccgccATGACGTTTCTACTTCACCGCTT	Cloning of pEGFP-gp38.3-2
P9	gcggtcgacaaTGATCGGAAAATCCGTGCCG	Cloning of pEGFP-gp38.3-2
P10	CCGAAGTGATCGCACCACCGGGgctgctgc CGTAACTCGAGGATCCGTAAAG	Mutagenesis to generate gp38.3-2 m1
P11	CTTTACGGATCCTCGAGTTACGgcagcagc CCCGGTGGTGCGATCACTTCGG	Mutagenesis to generate gp38.3-2 m1
P12	GCCGCCATGACGTTTcca CTTCACCGCTTTACG	Mutagenesis to generate gp38.3-2 m2
P13	CGTAAAGCGGTGAAGtgg AAACGTCATGGCGGC	Mutagenesis to generate gp38.3-2 m2
P14	CGACCGACGTGTCTACGGCGAGCTCTGTCCCTACTTCCGAc GACGTTTCTACTTCACC GCTtagggataacagggtaatcgattt	Markerless Red-mediated recombination to prepare recombinant GPCMVs
P15	CGTAACTCGAGGATCCGTAAAGCGGTGAAGTAGAAACGTCg TCGGAAGTAGGGACA GAGCTgccagtgttacaaccaattaacc	Markerless Red-mediated recombination to prepare recombinant GPCMVs
P16	CGACCGACGTGTCTACGGCGAGCTCTGTCCCTACTTCCGAt GACGTTTCTACTTCACC GCTtagggataacagggtaatcgattt	Markerless Red-mediated recombination to prepare recombinant GPCMVs
P17	CGTAACTCGAGGATCCGTAAAGCGGTGAAGTAGAAACGTCa TCGGAAGTAGGGACA GAGCTgccagtgttacaaccaattaacc	Markerless Red-mediated recombination to prepare recombinant GPCMVs
P18	ggtaccgtcgaggtGCGGAAGGTTCCGG	Mutagenesis to generate gpBAKΔTM-RFP
P19	CCGGAACCTTCCGCacctcgacggtacc	Mutagenesis to generate gpBAKΔTM-RFP
P20	GCGCTATGGCAGTGACcc TGAAGCCATGCTCCAG	Mutagenesis to generate gpBAKΔTM-RFP I86P
P21	CTGGAGCATGGCTTCAgg GTCACTGCCATAGCGC	Mutagenesis to generate gpBAKΔTM-RFP I86P

^aSequences of GPCMV, alterations introduced, and others are shown in uppercase, italic lowercase, and lowercase, respectively. Restriction enzyme sites are underlined.

Cells and recombinant GPCMVs.

GPL, simian virus 40 (SV40) T antigen-immortalized guinea pig epithelial clone GPE7, and guinea pig monocytes/macrophages prepared from the spleens were described previously (27, 31, 38). The markerless Red-mediated recombination method (39) was used to generate pBAC-GPCMVd9k-Δ38.3-2, pBAC-GPCMVd9k (31) containing an ATG-to-ACG mutation that knocks out the first initiation codon of gp38.3-2 without alteration to the aa sequence of gp38.3, and pBAC-GPCMVd9k-r38.3-2, in which the wild-type gp38.3-2 sequence was rescued into pBAC-GPCMVd9k-Δ38.3-2. Primers P14 to P17 (Table 1) were used for the procedures. The BAC DNA samples were purified using a commercially available kit (NucleoBond Xtra BAC; TaKaRa Bio, Japan) and transfected into GPL cells using Fugene HD (Promega) to generate GFP-expressing recombinant GPCMVs and GPCMVd9k-Δ38.3-2 and GPCMVd9k-r38.3-2 viruses, respectively. The integrity of the viruses was confirmed by restriction fragment length polymorphisms (RFLP) patterns and gp38.3 sequencing. Viruses were propagated and partially purified, and the titers of the recombinant GPCMV stocks were determined as described previously (27).

Assays for evaluation of apoptosis.

GPL cells in 96-well plates were transfected with the indicated plasmids using polyethylenimine (PEI) Max transfection reagent (Polysciences). Forty-eight hours later, the cells were left untreated or treated with 1 μM STS (Adipogen Life Science, San Diego, USA) for 3 h, and apoptotic cells were detected by the TUNEL assay using fluorescence-dUTP (TF3; Polysciences). Fluorescent images of the cells were captured using a ZOE fluorescent cell imager (Bio-Rad, Hercules, CA, USA).

293T cells (ATCC CRL-3216) were transfected with the indicated plasmid by PEI Max and, 48 h later, treated with 5 μM STS for an additional 16 h. Then, to detect the apoptosis-induced cell surface localization of phosphatidylserine, the cells were stained with Apotracker Green (Apo-15) (Thermo Fisher) (40). At least 10⁴ singlets of the cells were analyzed to identify Apo-15-positive cells by flow cytometry (FACSverse; BD) using the fluorescein isothiocyanate (FITC) channel as described in the manufacturer’s instructions.

293T cells were transfected with the indicated plasmid by PEI Max and, 48 h later, treated with 5 μM STS for an additional 12 h. The cells were then stained with Cell Event Caspase-3/7 green reagent along with SYTOX AADvanced dead cell staining reagent (Thermo Fisher). At least 3,000 singlets of the cells were analyzed by flow cytometry using the FITC and PerCP-Cyanine5.5 channels as described in the manufacturer’s instructions.

Fluorescent protein imaging.

GPL cells in 8-well chamber slides were transfected with the indicated plasmid(s) using Fugene transfection reagent, cultured for 48 h, fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS), counterstained with DAPI (4′,6-diamidino-2-phenylindole), and observed under a confocal microscope (LSM 900; Zeiss, Germany).

Cellular fractionation and immunoprecipitation followed by immunoblotting.

293T cells were transfected with the indicated plasmid by PEI Max and, 48 h later, untreated or treated with 0.5 μM STS for 1 h. For cellular fractionation, the cells were harvested and resuspended in 500 μL of PBS. Thirty microliters of the cells was separated and frozen for usage as cell lysates. The remaining cells were pelleted, and the cytoplasmic and nuclear fractions were prepared by a commercial kit (NE-PER nuclear and cytoplasmic extraction reagents; Thermo Fisher Scientific).

For immunoprecipitation, the harvested cells were suspended in PBS containing 0.5% NP-40 and lysed by sonication. Cell extracts were prepared as supernatants after centrifugation of the lysates at 20,000 × g for 20 min at 4°C. The cell extracts were reacted with monoclonal anti-GFP antibody (clone 1E4; catalog no. M048-03; MBL, Japan) followed by a reaction with protein G resin (Merck Millipore). The crude extracts used for the immunoprecipitation (pre-IP) and the proteins bound to the resin (post-IP) were analyzed by immunoblotting with rabbit polyclonal anti-BAK antibody (catalog no. ALX-210-002; Enzo Life Sciences, USA) or anti-GFP antibody (MBL). The anti-BAK antibody was against a synthetic peptide corresponding to the aa sequence positions 14 to 36 of huBAK. This 23-aa sequence has 91% similarity with the corresponding parts of gpBAK sequences. Intensities of bands were measured by Volume Tools of Image Lab software equipped in the ChemiDoc imaging system (Bio-Rad). Intensities of the BAK proteins in post-IP samples were normalized against those in the pre-IB samples, and the normalized intensities were compared with that obtained from wt gp38.3-2-EGFP.

Analysis of GPCMV growth in animals.

Three-week-old female guinea pigs (strain Hartley; SLC, Hamamatsu, Japan) were inoculated intraperitoneally with 5 × 10⁶ infectious units (IU) of GPCMVd9k-Δ38.3-2 or -r38.3-2. Their blood specimens and visceral tissues, including the livers, spleen, pancreas, and lungs, were collected on day 6 after virus inoculation. Viral loads in each specimen were determined by quantitative PCR assays for cellular β-actin and GPCMV GP83 genes as described previously (27). All animal experiments were approved by the Animal Care and Use Committee of Gifu Pharmaceutical University.

Prediction of protein structures.

Prediction of 3D protein structures was done using Colab Alphafold2 (41) and Alphafold-Multimer (42) with the default parameters. Models ranked at the top of the selected prediction by these software programs were analyzed by Chimera version 1.15 to select the sequences of gpBAK that interacted with the α-helix 1 of gp38.3-2 within a distance of 4 Å. Numbering of the α-helices is based on the structure of human BAK (PDB accession no. 2IMS) (43).

Statistical analyses.

Statistical analyses were performed using GraphPad Prism version 7 (GraphPad Software, CA, USA). Statistical significance between two groups was calculated using the Student's t test, and a P value of <0.05 was considered significant; statistical significance among three groups was calculated using the one-way analysis of variance (ANOVA) and Dunnett’s post hoc test, and a P value of <0.05 was considered significant.