Koala and Wombat Gammaherpesviruses Encode the First Known Viral NTPDase Homologs and Are Phylogenetically Divergent from All Known Gammaherpesviruses

The genome sequences of the koala and wombat gammaherpesviruses show that the viruses form a distinct branch, indicative of a novel genus within the Gammaherpesvirinae. Their genomes contain several new ORFs, including ORFs encoding a β-galactoside α-2,6-sialyltransferase that is phylogenetically closest to poxvirus and insect homologs and the first reported viral NTPDase. NTPDases are ubiquitously expressed in mammals and are also present in several parasitic, fungal, and bacterial pathogens. In mammals, these cell surface-localized NTPDases play essential roles in thromboregulation, inflammation, and immune suppression. In this study, we demonstrate that the virus-encoded NTPDase is enzymatically active and is transcribed during natural infection of the host. Understanding how these enzymes benefit viruses can help to inform how they may cause disease or evade host immune defenses.

ABSTRACT

There is a large taxonomic gap in our understanding of mammalian herpesvirus genetics and evolution corresponding to those herpesviruses that infect marsupials, which diverged from eutherian mammals approximately 150 million years ago (mya). We compare the genomes of two marsupial gammaherpesviruses, Phascolarctid gammaherpesvirus 1 (PhaHV1) and Vombatid gammaherpesvirus 1 (VoHV1), which infect koalas (Phascolarctos cinereus) and wombats (Vombatus ursinus), respectively. The core viral genomes were approximately 117 kbp and 110 kbp in length, respectively, sharing 69% pairwise nucleotide sequence identity. Phylogenetic analyses showed that PhaHV1 and VoHV1 formed a separate branch, which may indicate a new gammaherpesvirus genus. The genomes contained 60 predicted open reading frames (ORFs) homologous to those in eutherian herpesviruses and 20 ORFs not yet found in any other herpesvirus. Seven of these ORFs were shared by the two viruses, indicating that they were probably acquired prespeciation, approximately 30 to 40 mya. One of these shared genes encodes a putative nucleoside triphosphate diphosphohydrolase (NTPDase). NTPDases are usually found in mammals and higher-order eukaryotes, with a very small number being found in bacteria. This is the first time that an NTPDase has been identified in any viral genome. Interrogation of public transcriptomic data sets from two koalas identified PhaHV1-specific transcripts in multiple host tissues, including transcripts for the novel NTPDase. PhaHV1 ATPase activity was also demonstrated in vitro, suggesting that the encoded NTPDase is functional during viral infection. In mammals, NTPDases are important in downregulation of the inflammatory and immune responses, but the role of the PhaHV1 NTPDase during viral infection remains to be determined.

IMPORTANCE The genome sequences of the koala and wombat gammaherpesviruses show that the viruses form a distinct branch, indicative of a novel genus within the Gammaherpesvirinae. Their genomes contain several new ORFs, including ORFs encoding a β-galactoside α-2,6-sialyltransferase that is phylogenetically closest to poxvirus and insect homologs and the first reported viral NTPDase. NTPDases are ubiquitously expressed in mammals and are also present in several parasitic, fungal, and bacterial pathogens. In mammals, these cell surface-localized NTPDases play essential roles in thromboregulation, inflammation, and immune suppression. In this study, we demonstrate that the virus-encoded NTPDase is enzymatically active and is transcribed during natural infection of the host. Understanding how these enzymes benefit viruses can help to inform how they may cause disease or evade host immune defenses.

INTRODUCTION

Marsupial divergence from eutherian mammals occurred approximately 150 million years ago (mya) (1, 2). Rapid eutherian diversification occurred about 20 million years before that of marsupials (at 82 mya), and this explosive radiation led to the rise of nearly all extant placental orders by 85 mya (1). This period of taxonomic mammalian diversity coincided with, or even drove with it, herpesvirus evolution (3). Herpesviruses are widespread throughout the animal kingdom and can cause disease and death in a wide range of species (4, 5). They are among the largest and most complex of viruses and are superbly well adapted to their hosts, with many examples of their ability to manipulate and bypass host immune and cellular defenses (6).

Herpesvirus-induced disease and mortality in marsupials were first recognized in the mid-1970s, in several macropod species (7–9). Currently, there are 15 distinct marsupial herpesviruses reported, consisting of 4 alphaherpesviruses and 11 gammaherpesviruses, infecting diverse marsupial species (10–16). The clinical significance of many of these viruses remains largely unexplored, but macropodid alphaherpesvirus 1 (MaHV1) and MaHV2 have caused fatal systemic disease and severe clinical signs, including reproductive failure, conjunctivitis, respiratory disease, pyrexia, vesicular anogenital lesions, and hepatic disease, in wild and captive populations of threatened macropods (7–9, 17). The gammaherpesvirus MaHV3 has been associated with ulcerative cloacitis and mammary gland tumors in a captive population of eastern gray kangaroos (EGKs; Macropus giganteus) (14) and respiratory disease and mortality in wild EGKs (18). The alphaherpesvirus MaHV4 has been associated with respiratory and possible neurological disease, also in wild EGKs (12).

While a great deal is known about many herpesviruses that infect placental animals, particularly humans and animals of agricultural importance, the first marsupial herpesvirus genome, that of an alphaherpesvirus, was only published in 2016 (19). Studying the genetics of herpesviruses from noneutherian mammals contributes to our understanding of mammalian herpesvirus genetics and evolution. In this study, we focused on the genetics of two metatherian gammaherpesviruses, Phascolarctid gammaherpesvirus 1 (PhaHV1) from koalas (Phascolarctos cinereus) and Vombatid gammaherpesvirus 1 (VoHV1) from wombats (Vombatus ursinus) (11, 13). PhaHV1 was first described in 2011 in three geographically distinct wild koalas that were suffering from other concurrent infections and were showing signs of severe sarcoptic mange, pulmonary congestion, bilateral conjunctivitis, and chronic cystitis, most likely due to Chlamydia infection. VoHV1 was first identified in 2015 in a large molecular survey of herpesviruses in marsupial species that included 33 wild wombats, as well as 99 koalas (13). In that study, VoHV1 was detected in 15% of wombats and PhaHV1 was detected in 5% of koalas. The two viruses sequenced in this study both originated from samples collected from wild-caught animals in the 2015 study, specifically, the oropharyngeal swab of a koala and the nasal swab of a bare-nosed wombat.

We compared the genomes of PhaHV1 and VoHV1, with particular attention to their relationship to each other and to the Gammaherpesvirinae more generally, and used these data to trace viral dissemination and transcription within the host. We also investigated novel enzymes encoded by the viruses, including their activity in vitro.

RESULTS

PhaHV1 and VoHV1 whole-genome sequence analysis.

De novo assemblies of PhaHV1 and VoHV1 resulted in core genomes (excluding large reiterative repeat regions and genomic termini) of approximately 117 kbp and 110 kbp in length, respectively (Fig. 1; see also Fig. S1 in the supplemental material; GenBank accession numbers MG452721 and MG452722, respectively) and genome G+C values of 45% and 43%, respectively. The sizes of the large tandem repeat regions were not resolved. The final genome assembly of PhaHV1 had a mean depth of 176 reads per bp (0.2 million mapped reads), and approximately 95% of reads had a quality score of at least Phred₂₀. VoHV1 had a mean depth of 333 reads per bp (0.3 million mapped reads), and approximately 95% of reads had a quality score of at least Phred₂₀.

FIG 1.

Whole-genome sequence alignment of PhaHV1 and VoHV1. Conserved genes are annotated in gray using the herpesvirus saimiri (HVS) ORF nomenclature. ORFs colored orange are homologs of genes found in Epstein-Barr virus (BZLF2), Kaposi’s sarcoma-associated herpesvirus (E3), or equine gammaherpesvirus 2 (E4). Novel herpesvirus ORFs shared by PhaHV1 and VoHV1 (magenta) are given a V prefix, while virus unique ORFs (green or black) were given a Vp (PhaHV1) or Vv (VoHV1) prefix, and regions of low confidence are pale blue. Increasing nucleotide sequence identity between viral genome sequences is indicated by an increasing darker gray scale: light gray, no identity; medium gray, similar; black, identity; dashes, gap sites. A high-resolution image is provided in Fig. S1 in the supplemental material.

Both genomes had gene arrangements consistent with those of the genomes of other gammaherpesviruses and encoded 60 genes, or open reading frames (ORFs), common to eutherian herpesviruses more generally (Fig. 1; Table 1). The predicted protein sequences of these core genes shared between 44.2% (ORF45; host unknown protein) and 80.2% (ORF60; ribonucleotide reductase small subunit 2) amino acid sequence identity with each other. We also detected homologs of genes identified less commonly in the genomes of herpesviruses, including several E3-like (ubiquitin ligases) (20) and E4-like (viral Bcl-2) (21) ORFs and BZLF2 (22). Phylogenetic relationships with representative members of the Herpesviridae were examined using the conserved glycoprotein B (gB) and DNA polymerase (DPOL) herpesvirus genes. PhaHV1 and VoHV1 branched separately from existing genera (Fig. 2 and Table S1).

TABLE 1.

Predicted ORFs identified within the assembled core genomes of PhaHV1 and VoHV1 and their amino acid pairwise sequence identities^a

Gene	Description (putative)	Length (nt)		Amino acid sequence identity (%) for PhaHV1 and VoHV1
		PhaHV1	VoHV1
ORF2	Dihydrofolate reductase (DHFR)	588
ORF6	Single-stranded DNA-binding protein	3,381	3,384	72.4
ORF7	Terminase subunit; capsid associated	2,040	2,046	63.7
ORF8	gB	2,430	2,415	75.2
ORF9	Catalytic subunit of the DNA polymerase complex	3,036	3,027	77.1
ORF16	Bcl-2 homolog; antiapoptotic activity	507	507	42.0
ORF17	Serine protease; maturation of scaffold proteins	1,467	1,479	55.2
ORF18	Unknown	825	831	57.6
ORF19	Portal capping protein	1,593	1,635	66.5
ORF20	Nuclear protein	876	927	44.3
ORF21	Thymidine kinase	1,533	1,485	61.3
ORF22	gH; complexed with gN	2,196	2,223	74.7
ORF24	Unknown	2,172	2,202	64.4
ORF25	Major capsid protein	4,140	4,140	76.5
ORF26	Capsid triplex subunit	912	912	78.9
ORF29a	Terminase ATPase subunit	1,053	1,056	80.1
ORF29b	Terminase ATPase subunit	963	948	67.0
ORF30	Interacts with ORF18; role in late gene expression	309	345	51.6
ORF31	BDLF4; unknown	699	681	65.5
ORF32	Tegument protein; capsid localization in the nucleus	1,371	1,359	56.5
ORF33	Interacts with ORF38; regulates ORF36	1,020	1,020	67.8
ORF34	Interacts with ORF38; regulates ORF36	978	972	64.5
ORF35	BGLF3.5; unknown	459	519	55.7
ORF36	Tegument protein; serine-threonine protein kinase	1,341	1,326	69.5
ORF37	DNase; DNA maturation and recombination	1,434	1,434	73.0
ORF39	gM: complexed with gN	1,167	1,164	74.1
ORF40	Helicase-primase complex subunit	1,893	1,875	47.0
ORF42	Tegument protein; nuclear egress	798	798	63.5
ORF43	Capsid portal protein	1,677	1,752	74.1
ORF44	Helicase-primase complex subunit	2,358	2,403	79.6
ORF45	Nuclear phosphoprotein	1,050	1,002	44.2
ORF46	Uracil-DNA glycosylase	753	753	74.4
ORF47	gL; complexed with gH	864	528	45.8
ORF48	BRRF2; unknown	1,320	1,332	42.1
ORF49	Transcontrol protein	960	927	31.3
ORF50	R transactivator (RTA)	1,887	1,818	44.5
ORF52	BLRF2; tegument protein, p23 capsid antigen	411	402	69.9
ORF53	gN; complexed with gM	750	919	33.1
ORF54	Deoxyuridine triphosphatase	498	498	65.5
ORF55	Cytoplasmic egress facilitator	1,098	1,215	61.8
ORF56	Helicase-primase complex subunit; primase	2,553	2,520	63.9
ORF57	Transcriptional control factor	1,470	978	63.9
ORF58	Putative integral membrane glycoprotein	1,098	1,092	55.3
ORF59	Processivity subunit of the DNA polymerase complex	1,062	933	46.7
ORF60	Ribonucleotide reductase; small subunit	932	927	80.2
ORF61	Ribonucleotide reductase; large subunit	2,415	2,373	68.0
ORF62	Capsid triplex subunit	1,032	1,017	58.8
ORF63	Tegument protein; interacts with ORF64	2,805	2,856	52.8
ORF64	Large tegument protein; interacts with ORF63	6,741	6,723	49.3
ORF65	Small capsid protein; interacts with dynein	540	546	49.5
ORF66	Nuclear egress membrane protein	1,350	1,344	64.3
ORF67	Inner nuclear membrane protein; capsid docking	834	849	63.2
ORF67A	Viral DNA genome packaging; viral release	267	324	57.5
ORF68	Capsid transport nuclear protein	1,433	1,362	55.8
ORF69	Nuclear matrix protein; capsid docking	954	951	63.6
ORF70	Thymidylate synthase	867
ORF75	Large tegument protein/viral FGAM synthase	3,975	3,969	49.9
E3-M1	MARCH1-like homolog; ubiquitin ligase		576
E3-M1/8	MARCH1/8-like homolog; ubiquitin ligase	528
E3-M3	MARCH3-like homolog; ubiquitin ligase		552
E3-M4	MARCH4-like homolog; ubiquitin ligase		570
E3-M8	MARCH8-like homolog; ubiquitin ligase	618
E4	Viral Bcl-2; BALF1 homolog	579
BZLF2	B cell lymphocyte infection factor; envelope glycoprotein	1,059	933	35.3
Conserved ORFs within PhaHV1 and VoHV1
V1	β-Galactoside α-2,6-sialyltransferase 1 (ST6Gal1)	1,209	1,215	56.9
V2	Hypothetical protein	756	768	48.9
V3	Hypothetical protein	360	360	55.8
V4	Nucleoside triphosphate diphosphohydrolase (NTPDase)	1,548	1,524	39.0
V5	Hypothetical protein	2,847	3,336	31.4
V6	Hypothetical protein; ORF59-like paralog	1,626	1,164	47.7
V7	Hypothetical protein	627	768	30.9

^aORF naming and descriptions follow the convention for gammaherpesvirus, as described previously (21, 67–69). nt, number of nucleotides.

FIG 2.

Phylogenetic relationships between PhaHV1, VoHV1, and other representative herpesviruses using the translated amino acid sequences of two highly conserved genes, glycoprotein B (A) and DNA polymerase (B). The sequences were aligned using the MAFFT plug-in and refined using the ClustalW program. Maximum likelihood trees were built using the Jones-Taylor-Thornton substitution model with 100 bootstrap replicates. Refer to Table S1 in the supplemental material for the list of sequences used for each analysis.

Shared novel hypothetical ORFs in PhaHV1 and VoHV1.

Seven ORFs shared by both viruses (annotated V1 to V7) were either novel or not previously found in herpesviruses. No putative functional domains or motifs were identified in ORF V2, V3, V5, or V7. GenBank and Conserved Domain Database (CDD) searches identified putative functions for ORFs V1 (Pfam00777, β-galactoside α-2,6-sialyltransferase 1 [ST6Gal1]), V4 (Pfam01150, nucleoside triphosphate diphosphohydrolase [NTPDase]), and V6 (Pfam04929, DNA replication accessory factor, which may be a distant ORF59 paralog), and these predicted putative functions were supported by structural prediction analysis with the I-TASSER server. The V4 homologs in both viruses shared 39.3% pairwise amino acid sequence identity (69% nucleotide sequence identity) and shared up to 36% amino acid sequence identity with NTPDases of vertebrates. NTPDases catalyze the conversion of nucleoside triphosphates to di- and monophosphates, most commonly converting ATP to ADP and AMP. In PhaHV1, V4 contained all five apyrase conserved regions (ACRs), including conserved residues necessary for enzymatic function (Fig. 3A), as well as conserved cysteines important for tertiary structure (23). The identification of two putative transmembrane domains (TMDs) suggests that the enzyme was membrane bound. Localization predictions indicate that it is likely to be cell surface located, with the catalytic site facing the extracellular milieu and the N and C termini being found within the cytoplasm, although localization to the membrane of intracellular vesicles or organelles is also possible. Similar analysis of the VoHV1 V4 homolog identified well-conserved ACRs 1, 2, 3, and 5 and a second putative ACR3 (ACR3b), but ACR4 contained a number of amino acid changes, importantly, at conserved residues implicated in enzymatic function (D221H, G223D, G224T) (Fig. 3A). The V4 catalytic sites from three additional VoHV1-positive and two PhaHV1-positive animals were also PCR amplified and sequenced. Results showed that all 5 ACRs were identical to those of the isolates whose full genomes were sequenced. Six single nucleotide polymorphisms (SNPs; four of which were nonsynonymous) were identified across two of the three amplified VoHV1 V4 sites, and these were located between ACRs 2 and 3 and between ACRs 4 and 5. No SNPs were detected within PhaHV1 V4. VoHV1 V4 also contained the two putative TMDs identified in PhaHV1 V4. Phylogenetic analysis of the viral NTPDase homologs (Fig. 3B) found that they clustered most closely with the mammalian cell surface-located NTPDases (NTPDases 1 to 3 and 8), despite the low similarity across the whole gene.

FIG 3.

Phylogenetic comparisons of V4 ORFs from PhaHV1 and VoHV1 with the NTPDases of representative mammalian, bacterial, parasitic, and fungal species. (A) ClustalW alignment of NTPDase apyrase conserved regions 1 to 5 (ACR1 to ACR5, pink underline). Predicted viral ACRs are indicated with black boxes. Asterisks indicate conserved residues important for enzymatic function. (B) Neighbor-joining distance tree built using the Jukes-Cantor genetic distance model of amino acid substitution with 1,000 bootstrap replicates. The PhaHV1 and VoHV1 cluster is circled. Mammalian NTPDase clusters are underlined and colored as follows: red, NTPD1 to NTPD3 and NTPD8; blue, NTPD4 and NTPD7; green, NTPD5 and NTPD6. Md, Monodelphis domestica (gray short-tailed opossum); Sh, Sarcophilus harrisii (Tasmanian devil); Pc, Phascolarctos cinereus (koala); H, Homo sapiens (human). Refer to Table S4 in the supplemental material for the list of sequences used.

The V1 (ST6Gal1) homologs encoded in the genomes of both viruses shared 57% pairwise amino acid sequence identity with each other and shared only up to 37% amino acid sequence identity with the ST6Gal1s of vertebrates. Both homologs contained similar amino acid sequence structural features (Fig. 4A): a putative N-terminal TMD and signal sequence cleavage site and an ST6Gal1 catalytic site, identified by both PFAM and I-TASSER analyses. Additionally, a conserved substrate-binding motif (SSG) was identified. These features suggest that both these ST6Gal1 homologs are likely to be functional. Phylogenetic analysis of the viral ST6Gal1 homologs (Fig. 4B) found that they clustered with the only two other known virally encoded ST6Gal homologs, found in Yoka poxvirus (PV) and squirrel poxvirus (24, 25), although they shared low amino acid sequence similarity (30% pairwise amino acid sequence identity). As a group, the viral homologs clustered with the insect genome-encoded ST6Gal rather than the ST6Gals found in vertebrates (Fig. 4B).

FIG 4.

Sequence structure and phylogenetic comparisons of V1 ORFs from PhaHV1 and VoHV1 with the α-2,6-sialyltransferases (ST6Gals) of representative vertebrate, invertebrate, insect, and viral species. (A) ClustalW alignment of virus-encoded ST6Gal polypeptide sequences. Predicted PhaHV1 and VoHV1 transmembrane and catalytic domains are outlined in orange and blue boxes, respectively, and substrate-binding sites (SSG) are indicated with red asterisks. The identity of aligned proteins sequences is indicated by an increasingly darker gray scale: light gray, no identity; medium gray, similar; black, identity; dashes, gapped sites. (B) Neighbor-joining distance tree built using the Jukes-Cantor genetic distance model of amino acid substitution with 1,000 bootstrap replicates. The viral cluster is indicated in blue, with the PhaHV1 and VoHV1 branches marked, and the ST6Gal1 and ST6Gal2 clusters are indicated. Refer to Table S5 in the supplemental material for the list of sequences used. PV, poxvirus.

Eight additional hypothetical ORFs (annotated Vp1 to Vp8) were identified only in the PhaHV1 genome, and five unique ORFs (tentatively annotated Vv1 to Vv5) were identified in the VoHV1 genome. Bioinformatic analyses revealed no putative functional domains or motifs of interest in these ORFs.

Increased SNVs identified in VoHV1 sequence read data.

Tenfold more SNVs were detected in VoHV1 (100 single nucleotide variants [SNVs]) than in PhaHV1 (9 SNVs). All SNVs in PhaHV1 occurred within coding sequences and resulted in nonsynonymous changes. In VoHV1, 86 of the SNVs occurred within a putative coding DNA sequence, and 35 of these were nonsynonymous (Fig. S2). Variant frequencies ranged from 10.2% to 48.0% (mean = 27.0% ± 8.13%) in VoHV1 and 17% to 35% (mean = 29.7% ± 3.6%) in PhaHV1. In VoHV1, the SNVs occurred within 29 annotated ORFs, mostly in core genes, but also in 7 auxiliary (BZLF2 and E3-M4) or novel putative ORFs (V4 to V6, Vv3, and Vv5). In PhaHV1, most SNVs were within the central region of the genome and occurred within five putative ORFs: four core genes and one novel ORF (V3).

Transcripts mapping to PhaHV1 were identified in koala SRA data sets.

Interrogation of the NCBI Sequence Read Archive (SRA) database was performed to investigate host tissue tropism and to gather viral gene expression information. No VoHV1-specific reads were identified, but PhaHV1-specific transcripts were detected within the koala Transcriptome Sequencing Project (SRA accession number SRP033633) (26), which included transcript data sets from tissues of a female koala with severe chlamydiosis and a male koala attacked by a dog. We detected virus transcripts mostly in the liver and lung, but also in the spleen, of the female koala and mostly in the spleen, bone marrow, and lymph node, but also the liver, from the male koala (Table S2 and Fig. S3). Of these reads, 81% mapped to 35 predicted viral ORFs, with 93.3 to 100% pairwise nucleotide sequence identity with the PhaHV1 reference sequence. Of the 35 ORFs, 4 were shared viral ORFs (V2, V4, V6, V7) and 4 were PhaHV1-specific ORFs (Vp1, Vp3, Vp5, Vp8). The remaining transcripts mapped to five apparently noncoding regions and were mostly centered around ORF49 and ORF50, which are important for the viral lytic cycle (27). Some variants from the PhaHV1 sequence used as a reference were identified, but SNV analysis was not performed because of the very low depth of mapped reads per SNV site (1 to 4 reads), and therefore, the possibility of read sequencing errors could not be excluded. Searches of transcript data sets from other marsupial species (e.g., Tasmanian devil [Sarcophilus harrisii], tammar wallaby [Macropus eugenii]) did not identify transcripts mapping to the viral genomes (Table S3). Similarly, use of VoHV1 as a query with similar search parameters did not extract virus-specific transcripts from the koala transcriptome data sets, nor did use of the available sequences from other marsupial herpesviruses.

The putative NTPDase from PhaHV1 hydrolyzes nucleoside tri- and diphosphates in vitro.

Recombinant viral NTPDases (vNTPDases) from both PhaHV1 and VoHV1 lacking the predicted TMDs were expressed and purified (Fig. 5) and then tested for NTPDase activity using a capillary electrophoresis (CE)-based enzymatic assay. PhaHV1 NTPDase activity was detected with 6 of the 21 tested substrates: ATP, ADP, GTP, CTP, UTP, and, at a low level, TTP (Table 2). Incubation for 2 h at 37°C was optimal for detection of PhaHV1 NTPDase activity against nucleoside triphosphates. Recombinant VoHV1 V4 was tested against nine of the substrates but showed minimal activity (<3% conversion) (Table 2). Other nucleotides, including cyclic mono- and dinucleotides, dinucleotide polyphosphates, nucleotide sugars, and p-nitrophenylated nucleotide monophosphates and p-nitrophenyl phenylphosphate (p-NPPP), were not hydrolyzed by either of the recombinant viral proteins.

FIG 5.

(A) Design of recombinant expression constructs for viral NTPDase genes from PhaHV1 and VoHV1. The identity of the aligned protein sequences is indicated by an increasingly darker gray scale: light gray, no identity; medium gray, similar; black, identity; dashes, gapped sites. Annotations indicate the extracellular domain (ECD; gray), the IL-3 signal peptide (pink; MVLASSTTSIHTMLLLLLMLFHLGLQASIS), and the C-terminal dual Flag-6×His tag (FLAG-HIS; red and purple, respectively; DYKDDDDKGGGSHHHHHH, where the first and second underlined sequences represent the Flag and His sequences, respectively). (B and C) Purified PhaHV1 and VoHV1 NTPDases stained by Coomassie brilliant blue staining (B) and Western blotting using anti-Flag antibodies (C). HEK293T cell lysate controls are included. The predicted protein sizes are approximately 50 kDa without glycosylation, so the increased sizes of 70 kDa in the reduced fraction may be due to posttranslational modifications. The reduced binding of M2-HRP antibodies to VoHV1 NTPDase may be due to reduced exposure of the Flag epitope region.

TABLE 2.

Substrates used to test PhaHV1 and VoHV1 recombinant proteins for enzymatic activity and percentage of total nucleotide converted to NMP or NDP products after 2 h of incubation at 37°C, quantified using CE^a

Substrate	Concn (μM)	Assay	PhaHV1 NTPDase			VoHV1 NTPDase
			Activity detected	% NMP (avg)	% NDP (avg)	Activity detected	% NMP (avg)	% NDP (avg)
ATP	1,000	CE	Yes	10.5	1.0	Yes	0.3	2.4
ADP	1,000	CE	Yes	21.5		Yes	1.9
AMP	1,000	MG	No			No
GTP	1,000	CE	Yes	26.5	−2.0	Yes	0.6	−0.9
CTP	1,000	CE	Yes	15.5	1.5	Yes	0.6	2.9
TTP	1,000	CE	Yes	2.0	0.0	Yes	0.4	1.6
UTP	1,000	CE	Yes	11.0	0.5	Yes	0.3	1.0
2′,3′-cAMP	1,000	CE	No			—
3′,5′-cAMP	1,000	CE	No			—
3′,3′-c-di-IMP	200	CE	No			—
3′,3′-c-di-GMP	200	CE	No			—
2′,5′-c-di-GMP	200	CE	No			—
2′,3′-cGAMP	200	CE	No			—
3′,3′-cGAMP	200	CE	No			—
AP3A	200	CE	No			—
AP4A	200	CE	No			—
UDP-glucose	1,000	CE	No			—
NAD	1,000	CE	No			—
p-Nph-5′-TMP	1,000	p-NP	No			No
p-Nph-5′-AMP	1,000	p-NP	No			No
p-NPPP	1,000	p-NP	No			No

^aNMP, nucleoside monophosphate; NDP, nucleoside diphosphate; MG, malachite green; CE, capillary electrophoresis; —, not tested; UDP-glucose, uracil diphosphate glucose; cAMP, cyclic AMP; c-di-IMP, cyclic di-inosine monophosphate; c-di-GMP, cyclic diguanosine monophosphate; cGAMP, cyclic guanosine AMP; AP3A, diadenosine triphosphate; AP4A, diadenosine tetraphosphate; p-Nph-5′-TMP, p-nitrophenyl 5′- dTMP; p-Nph-5′-AMP, p-nitrophenyl 5′-AMP; p-NPPP, p-nitrophenyl phenylphosphate.

Enzyme kinetic analyses were performed on the four nucleoside triphosphates that PhaHV1 was most active against: ATP, GTP, CTP, and UTP. ATP and GTP showed the lowest affinity for the active site (Michaelis-Menten constant [K_m] values, 409 μM and 381 μM, respectively) compared to that of CTP and UTP (K_m values, 205 μM and 206 μM, respectively). Though conversion to products was faster for ATP and GTP than for CTP and UTP, the substrate specificity parameter (k_cat/K_m) values were similar for all 4 substrates (224 to 300 M⁻¹·s⁻¹), suggesting that the relative rate of hydrolysis for each substrate was similar (Fig. 6). The k_cat/K_m values for the four substrates were relatively low compared to those for the other ectonucleotidases (28).

FIG 6.

Michaelis-Menten plots for PhaHV1 NTPDase activity with GTP, ATP, UTP, or CTP as the substrate and the corresponding kinetic values for each substrate. Nonlinear regression was performed using GraphPad Prism software.

Inhibitor screening, IC₅₀ determination, and mechanism of inhibition.

Inhibitor screening utilized a small library of 20 compounds (Table 3) with diverse structures known to be potent inhibitors of NTPDases from eutherian mammals (specifically, human and rat) (29). The ability of compounds to inhibit the hydrolysis of ATP to AMP by the PhaHV1 enzyme was tested in a CE-based assay and a malachite green (MG) phosphate assay (Fig. 7). Both assays yielded comparable results. Nine compounds inhibited the PhaHV1 enzyme activity by ≥70%, and two compounds, compound 4 (PV4, PSB-POM141, K₆H₂[TiW₁₁CoO₄₀]·13H₂O) and compound 9 (Na₁₄[P₅W₃₀O₁₁₀]·30H₂O), completely inhibited enzymatic activity at 10 μM. A concentration-inhibition 50% inhibitory concentration (IC₅₀) assay demonstrated that compound 9 was more potent than compound 4 (IC₅₀s, 0.169 ± 0.005 μM and 0.784 ± 0.176 μM, respectively) (Fig. 8). As both compounds typically acted via a similar mechanism on other NTPDases, the mechanism of inhibition was determined experimentally only for compound 9. A Hanes-Woolf plot analysis showed that it acted as a noncompetitive inhibitor of the PhaHV1 enzyme (Fig. 8) by crossing all lines at the x axis, which suggested that the K_i values of compound 9 (and compound 4) are approximately equivalent to their IC₅₀ values.

TABLE 3.

NTPDase inhibitor compounds tested in PhaHV1 NTPDase inhibition studies

Inhibitor no.	Inhibitor	Formula	Mol wt (Da)
1	PV1	Na6(H2W12O40)·21H2O	3,366
2	PV2	H3(PW12O40)·H2O	2,880
3	PV3	K7(Ti2W10PO40)·8H2O	3,023
4	PV4	K6H2(TiW11CoO40)·13H2O	3,240
5	PV5	K10[Co4(H2O)2(PW9O34)2]·22H2O	5,518
6	PV6	(NH4)18(NaSb9W21O86)·14H2O	6,932
7	NaPW12	Na3(PW12O40)·7H2O	3,074
8	NaP6W18	Na20(P6W18O79)·37H2O	5,888
9	NaP5W30	Na1(NaP5W30O110)·30H2O	8,320
10	Reactive blue 2	C29H20ClN7Na3O11S3	774
11	KB1	K4[(Re6S8)(OH)6]·8H2O	1,776
12	KB2	K4[(Re6Se8)(OH)6]·8H2O	2,151
13	KB3	K4[(Re6S8)(CH3COO)6]·8H2O	2,028
14	KB4	K4[(Re6S8)(HCOO)6]·3H2O	1,854
15	Le135	C29H30N2O2	439
16	Resveratrol	C14H12O3	228
17	Quercetin	C15H10O7	302
18	Suramin	C51H40N6O23S6	1,297
19	Methotrexate	C20H22N8O5	454
20	IBMXa	C10H14N4O2	222

^aIBMX, 3-isobutyl-1-methylxanthine.

FIG 7.

Inhibitor screening results of PhaHV1 NTPDase activity using a panel of 20 known inhibitors of NTPDases from placental mammals. (A) The ability of these compounds to inhibit hydrolysis of ATP to AMP by the PhaHV1 enzyme was tested using a CE-based assay. (B) Individual members of the compound library were also tested with the malachite green method, used to measure P_i production. Compounds were screened at 10 μM, and the percent inhibition of enzymatic activity was calculated. The detection of NTPDase inhibitors using this method was comparable to that using CE-based assays. Refer to Table 3 for the list of compounds used.

FIG 8.

Concentration-inhibition curves for polyoxometalate compounds 4 [(TiW₁₁CoO₄₀)⁸⁻, PV4, or PSB-POM141] (A) and 9 [(NaP₅W₃₀O₁₁₀)¹⁴⁻ or NaP₅W₃₀] (B) against PhaHV1 NTPDase. (C) Hanes-Woolf plot of vNTPDase inhibition by compound 9, indicating that it acts as a noncompetitive inhibitor of the viral enzyme. [S], concentration of ATP substrate (in micromolar); v, velocity of reaction (in nanomoles per minute per milligram of protein); IC₅₀, the concentration of an inhibitor showing 50% residual enzyme activity. A schematic representation of the structure of each compound, adapted from reference 29, is shown in red.

DISCUSSION

The PhaHV1 and VoHV1 genomes are the first marsupial gammaherpesvirus genomes to be sequenced. Although a macropod alphaherpesvirus (MaHV1) genome was published in 2016 (19), gammaherpesviruses are of particular interest because of their typically restricted host and tissue range and apparently greater variation between viral species (30, 31). It is consistent with the evolutionary history of marsupials and their early divergence from eutherian mammals (1, 2) to find that gammaherpesviruses that infect and that have probably coevolved with marsupials formed an isolated branch outside the existing gammaherpesvirus genera.

The SNV analysis identified a relatively high number of SNVs in the VoHV1 sequence data set, with read frequencies of up to 48%, indicating that there was probably a mixed population of VoHV1s within the original sample. Similarly, SNV analysis of the PhaHV1 read data indicated that the source animal was also likely the host to multiple virus variants, although the lower frequency of SNVs and their nonsynonymous nature may suggest that culture of this koala virus on wombat cells placed it under a higher selective pressure, resulting in a restricted SNV profile, or bottlenecking (32, 33). Infection with multiple strains of the same gammaherpesvirus species has been documented in other hosts (34–38). Interrogation of the koala transcriptomic data sets detected PhaHV1 in two animals and in multiple tissues. Some of these sites of viral replication (spleen and liver) were already known (11), but others (bone marrow, lung, lymph node) had not been reported previously.

The genomes of both viruses encoded putative NTPDase and ST6Gal1-like proteins. The presence of genes encoding these enzymes has not been reported previously in a herpesvirus, and an NTPDase gene has never been reported in any virus to date. Their positional and sequence homology, as well as phylogenetic comparisons to homologs in vertebrates and pathogens, indicated that they were likely acquired in a single event, prior to their viral evolutionary divergence and possibly prior to the speciation of koalas and wombats, which is estimated to have occurred 30 million to 40 million years ago (39). While it is likely that these genes were acquired from a common ancestor of koalas and wombats, the vNTPDases and viral ST6Gal1 shared only 35% to 37% pairwise amino acid sequence identity with those of contemporary mammals.

Members of the ST6Gal family of enzymes are broadly expressed across most taxonomic lineages, including vertebrates, invertebrates, and insects. They are typically localized to the Golgi apparatus, where they sialylate galactose-containing glycosylated products. STGal1s are essential for regulating the production of multiple T and B lymphocyte cell surface differentiation antigens, CD75, HB-6, and CD76 (40), as well as regulating macrophage apoptosis (41). To our knowledge, no ST6Gal homologs have been reported in prokaryotes or parasites to date. Although unusual in viruses, the ST6Gal homolog encoded by V1 is among a small number of viral sialyltransferases that have previously been identified in only three poxviruses. These are an ST6Gal in Yoka poxvirus (a 1972 isolate from a mosquito pool) and squirrel poxvirus (a cause of lethal outbreaks of disease in Eurasian red squirrels [Sciurus vulgaris]), with which V1 appears to form a clade, and an ST3Gal (α-2,3-sialyltransferase) encoded by the genome of myxoma virus (the cause of lethal disease in European rabbits [Oryctolagus cuniculus]) (24, 42, 43). In myxoma virus, the ST3Gal homolog has been identified to be a virulence determinant, and its absence enhances both inflammatory cell migration to the site of infection and the neutralizing antibody response to infection (42, 44). The role of virus-encoded ST6Gal homologs has not been investigated, so it is unknown whether it would have a similar effect and function during viral infection.

The NTPDase family of enzymes is commonly distributed across animal species and is present in many pathogenic protists, although the role of these enzymes in these organisms is poorly defined (45). Many NTPDases are membrane bound, and four of the eight mammalian subtypes are cell surface localized with an extracellular active site (46). In mammals, these cell surface-localized NTPDases play key roles in thromboregulation, inflammation, and immune suppression by affecting the extracellular concentration of nucleoside tri- and diphosphates, with subsequent effects on the purinergic receptors involved in these processes (46, 47). In humans, NTPDases are a therapeutic target for both genetic and infectious diseases (29, 47–50) and so may also have potential as an antiviral target.

Initial sequence analysis of the PhaHV1 NTPDase homolog revealed the presence of all gene features that are reportedly important for enzyme activity (51). Purification and enzymatic characterization of recombinant PhaHV1 NTPDase demonstrated hydrolysis of nucleoside tri- and diphosphates in the presence of divalent cations, but no other type of nucleotidase activity was detected, as is typical for the NTPDase family. The affinity of PhaHV1 NTPDase for the nucleoside triphosphates (NTPs) tested was lower than that of mammalian (human and rat) NTPDases 1 to 3 (K_m for vNTPDase = 200 to 400 μM; K_m for human NTPDases 1 to 3 = 10 to 200 μM) (51) but was still within the same order of magnitude, suggesting that the enzyme is likely active at the physiological concentrations at which NTPs can be found in mammalian hosts.

The role of V4 during PhaHV1 infection is unknown, though herpesviruses are most active (and cause disease) when hosts are stressed or immunosuppressed (5). Extracellular ATP is a danger signal for the host, and activation of purinergic receptors by ATP is a key component of the inflammatory process (52). Expression of the enzyme on the surface of the host cell may reduce the inflammatory response by lowering ATP (and ADP) concentrations and reducing activation of proinflammatory P2 receptors, leading to increased concentrations of AMP. This is further converted to anti-inflammatory and immunosuppressive adenosine by another mammalian ectonucleotidase, CD73 (or ecto-5′-nucleotidase), with the effect of redirecting the immune response toward the P1 (adenosine) receptor pathways, resulting in local immunosuppression through altered cytokine expression profiles or inhibition of apoptotic cellular pathways (51). Our results show that PhaHV1 NTPDase readily hydrolyzes ADP. Hydrolysis of ATP to only ADP promotes platelet aggregation, which is one role of NTPDase2 (53), whereas ongoing hydrolysis to AMP results in inhibition of platelet aggregation (54), which may be advantageous during viral infection.

Further biochemical analysis is needed to determine if the VoHV1 V4 encodes a functional NTPDase or another biologically relevant active enzyme. An incomplete ACR4 in VoHV1 V4 may explain the negligible activity detected against the NTP and nucleoside diphosphate (NDP) substrates tested, though other contributing factors could be the removal of the TMDs (which can affect the activity of mammalian NTPDases) or a lack of the preferred divalent cation (only Mg²⁺ and Ca²⁺, two commonly preferred cations for NTPDases, were tested) (55). The presence of a second ACR3 could also have resulted in an altered functional state. For example, this protein may play a different (nonnucleotidase-like) role during VoHV1 infection. Other NTPDases have been identified in the genomes of bacteria (e.g., Legionella), fungi (e.g., Candida, Cryptococcus), and parasites (e.g., Plasmodium, Leishmania, and Schistosoma). While in some cases they have been identified as a virulence factor, e.g., in Legionella pneumophila (50), for most of these pathogens the precise role or importance of their encoded NTPDases has not been described. Although this is the first NTPDase encoded by a virus, it is not the first example of the viral appropriation of NTPDases. In human immunodeficiency virus infection there is an increase in NTPDase1 (CD39) expression, and functional host NTPDase1 can be found embedded in the viral envelope (56, 57). Further localization and functional enzymatic studies, as well as targeted transcriptomic or proteomic studies, may help identify what purpose these viral NTPDases have during infection.

Our understanding of the role of herpesviruses in marsupial health is still in its infancy. A 2015 study of marsupials in the Australian state of Victoria detected herpesviruses in 33.3% of koalas and 45.5% of wombats. Herpesvirus infections were significantly associated with both poor body condition in wombats and coinfection with Chlamydia pecorum in koalas, but the direct clinical relevance of these herpesviruses remains undetermined (13). This is consistent with studies of gammaherpesviruses in other species, which indicate that their clinical effects can be insidious and difficult to evaluate, particularly in nonlaboratory animal populations (6, 58–60). Currently, there are no treatments for herpesvirus-induced diseases in any marsupial species, but the therapeutic potential of inhibitors that prevent the degradation of extracellular nucleotides by host- and/or pathogen-derived enzymes is of interest in treating human diseases (29, 47–50). This study identified 9 compounds capable of inhibiting viral NTPDase activity; however, their effect on an active viral infection remains to be determined. As the inhibitors selected for this study are known to be potent inhibitors of mammalian (eutherian) NTPDases, it is likely that they inhibit host enzymes and virus-expressed homologs. Although this may have possible benefits in alleviation of virus-induced immunosuppression, evidence of a direct interaction with viral NTPDase would be difficult to demonstrate in vivo. The future identification of inhibitors that are specifically active against the viral (rather than host) enzymes offers a potential target for controlling herpesvirus infection in koalas and wombats.

MATERIALS AND METHODS

Virus propagation and genome sequencing and assembly.

The PhaHV1 and VoHV1 isolates originated from an oropharyngeal swab specimen from a koala and a nasal swab specimen from a bare-nosed wombat, respectively, and were cultured in wombat joey primary kidney cells (13). Approval for swab collection was granted by the Animal Ethics Committee for the Faculty of Veterinary Science, The University of Melbourne (Animal Ethics ID 1112058.1). Herpesvirus nucleocapsid DNA was sequenced, using Illumina MiSeq paired-end chemistry, and genomes were assembled as described previously (19), with minor modifications. Specifically, conserved gammaherpesvirus ORF arrangements (61) were used as a guide to construct the final genomic structures. Regions of unresolved ambiguity, primarily due to tandemly repeating sequences, were annotated as such.

Genome annotation and single nucleotide variant analysis.

Prediction of ORFs using a Glimmer3 system was restricted to core genomic regions of low ambiguity and to ORFs larger than 250 bp. ORF annotation was performed by searching against the NCBI nonredundant protein and nucleotide databases with the BLASTX and BLASTN programs, respectively (62, 63). Predicted ORFs of unknown function were annotated as hypothetical proteins, and their sequences were analyzed using the PFAM database and the Conserved Domain Database (CDD) to predict their putative function. Other sequence analysis algorithms used were SignalP to identify putative signal peptides, TMHMM (64, 65) to identify transmembrane domains, and Phobius, a combined transmembrane and signal peptide prediction tool for cellular localization predictions. Similarity searches at the level of secondary and tertiary protein structure were performed using the I-TASSER server (66). Threshold cutoff values of >1 for the normalized Z-score, a root mean square deviation (RMSD) of <3.0, and a template modeling (TM) score of >0.7 for these were considered significant and used to identify structural homologs. The nomenclature of common gene annotations followed the gammaherpesvirus naming convention (21, 67–69), while PhaHV1/VoHV1 shared hypothetical/novel ORFs were prefixed with V (Vombatiformes) and virus-specific unique ORFs were prefixed with Vp (PhaHV1) or Vv (VoHV1).

Single nucleotide variant analysis was performed with a minimum variant frequency threshold of 0.10, a maximum variant P value of 10⁻⁶ (determined using the approximate P value calculation method), and the plug-in default setting of a minimum-strand-bias P value of 10⁻⁵ with a threshold bias of 65%. Regions of low sequence confidence, specifically, unresolved sites with complex sequence repeats, were excluded from variant analysis to reduce the incidence of false-positive results. Variant analyses were performed using the software package Geneious (version 9.1.5).

In vivo ORF transcript detection.

Interrogation of public transcript data sets was performed using the Sequence Read Archive nucleotide BLAST (SRA-BLAST) function in the NCBI website (https://www.ncbi.nlm.nih.gov/sra). We used the assembled viral genomes as the query to search all marsupial SRA experimental gene expression data sets (see Table S3 in the supplemental material). Searches were restricted to highly similar sequence settings, using MegaBLAST conditions, in order to exclude non-virus-specific reads.

Phylogenetic analysis of Vombatiformes herpesviruses.

Alignments of the assembled genome sequences of PhaHV1 and VoHV1 (Fig. 1) were prepared using the Multiple Alignment with Fast Fourier Transformation (MAFFT; version 7) plug-in in Geneious (version 6.1.8) software (70). The sequences of regions of high diversity, specifically, the sequence termini, were manually adjusted to obtain a better alignment. Phylogenetic analyses with representative sequences of the Alpha-, Beta-, and Gammaherpesvirinae were performed using the most conserved herpesvirus genes (Fig. 2): DPOL (ORF9) and gB (ORF8). The sequences of the encoded polypeptides were aligned using the MAFFT plug-in, and the alignments were refined using the ClustalW program. Alignment columns containing gaps were removed from all sequences prior to further analysis. Trees were built using maximum likelihood with the Jones-Taylor-Thornton substitution model with 100 bootstrap replicates. Phylogenetic analyses were performed using selected viral ORFs (V1 and V4) that included homologs from representative eukaryotic and prokaryotic taxa. Neighbor-joining distance trees were built using the Jukes-Cantor distance model with 1,000 bootstrap replicates. These polypeptide sequences were aligned as described above for the core ORFs, and phylogenetic alignments were restricted to concatenated conserved ORF regions. The sequences used in all analyses can be found listed in Tables S1, S4, and S5.

Generation of recombinant viral NTPDases.

Viral NTPDase-like amino acid sequences were evaluated for protein features required for enzymatic activity, including ACRs and conserved cysteine residues (23). A codon-optimized sequence of the predicted extracellular domains (ECD) containing the five putative ACRs was cloned into pCAGGS_MCS, a mammalian cell expression vector (Fig. 5A; GenScript Biotech, China). The predicted native signal sequence was replaced by an interleukin-3 (IL-3) signal sequence to promote protein secretion and C-terminal Flag (DYKDDDDK) and His purification tags. Suspension-adapted cultures of FreeStyle-293 cells (Life Technologies) were grown in Freestyle 293 expression medium (Life Technologies) supplemented with GlutaMAX-I supplement. Scale-up transient transfection was performed on 1-liter cultures of FreeStyle-293 cells (2 × 10⁶ per ml) in a shake flask using linear polyethyleneimine (PEI; Polysciences Inc.) according to a published protocol (71). The cultures were harvested after cell viability reached ∼50% (7 to 11 days), clarified by centrifugation, and filtered. Soluble, recombinant viral NTPDases were purified by anti-Flag immunoaffinity chromatography followed by size exclusion chromatography through a Superdex S200 column (GE Healthcare) as previously described (72). Final protein elutions were performed using Tris-buffered saline (5 mM Tris-Cl, pH 7.5, 150 mM NaCl). Eluates were concentrated using Amicon Ultra-15 centrifugal filter units (Merck Millipore), aliquoted, and stored at −70°C until use. Fractions of each protein and an HEK293T cell lysate control were separated on a 4 to 15% mini-Protean TGX precast protein gel (Bio-Rad) as previously described (73) under denaturing conditions. The proteins were transferred to a polyvinylidene difluoride membrane and blocked in phosphate-buffered saline (PBS) containing 0.05% Tween 20 (PBS-T) and 5% (wt/vol) skim milk powder for 60 min at 37°C, the membrane was then incubated with M2-horseradish peroxidase (HRP) antibodies detecting the Flag tag, and binding was detected using a ClarityWestern enhanced chemiluminescence blotting substrate (Bio-Rad),

Substrate screening.

PhaHV1 recombinant V4 was screened against the 21 substrates in Table 2 (in duplicate) by CE, an MG phosphate assay, or a p-nitrophenolate (p-NP)-based assay (28). The assay buffer for the CE system contained Ca²⁺ (without Mg²⁺), as Ca²⁺ yielded the highest activity in this system. The concentrations of the substrates were 0.2 or 1.0 mM with 2.4 μg of each vNTPDase in reaction buffer (10 mM HEPES, 2 mM CaCl₂, pH 7.4) and 10% (vol/vol) dimethyl sulfoxide (DMSO). The reaction mixtures were incubated for 1 and 2 h at 37°C and then for 3 min at 90°C to stop the reactions. No-enzyme (substrate-only) controls were included for each substrate, and 20 μM monophosphates (AMP, GMP, UMP, CMP, TMP) were used as standards. The reaction mixtures were diluted 1 in 3 (0.2 mM substrate reaction mixtures) or 1 in 10 (1.0 mM substrate reaction mixtures) in reaction buffer prior to CE to prevent overloading the instrument. CE analyses were carried out using a P/ACE MDQ capillary electrophoresis system (Beckman Instruments, USA), and data collection and peak area analysis were performed using the P/ACE MDQ software 32 KARAT (Beckman Coulter, USA). The electrophoretic separations were performed in a 40-cm polyacrylamide-coated capillary (CS-Chromatography, Langerwehe, Germany) with electrokinetic injections at a voltage of −6 kV for 30 s and separations at a voltage of −15 kV using 100 mM phosphate buffer (pH 6.5). Analytes were detected using direct UV absorbance at 260 nm with a diode array detection system. The hydrolysis of AMP was investigated using the MG assay, and enzymatic assay conditions were as described above for the CE-based analysis, but the detection of released inorganic phosphate was performed using a PHERAstar plate reader (BMG Labtech GmbH, Ortenberg, Germany) (74).

A series of artificial substrates, including p-nitrophenyl-5′-dTMP (p-Nph-5′-TMP), p-nitrophenyl-5′-AMP (p-Nph-5′-AMP), and p-nitrophenyl phenylphosphate (p-NPPP), was investigated using the p-nitrophenolate (p-NP)-based assay (28). Briefly, 2.4 μg of each vNTPDase was incubated at 37°C for 2 h with each substrate (1 mM) in reaction buffer and 10% (vol/vol) DMSO in triplicate. The reactions were stopped using 20 μl of 1.0 M NaOH, and the release of p-NP was measured at 400 nm using a PHERAstar plate reader, with human NPP1 included as a positive reaction control and 20 μM p-NP included as a standard.

Enzyme kinetic analysis to determine substrate preference.

Detailed enzyme kinetic analysis was performed for PhaHV1 NTPDase. The Michaelis-Menten constant (K_m), maximal velocity (V_max), turnover number (k_cat), and substrate specificity parameter (k_cat/K_m) values were determined for the nucleoside triphosphate substrates ATP, GTP, CTP, and UTP. Briefly, 2.4 μg of vNTPDase was incubated for 2 h at 37°C with substrates at concentrations of 0.1, 0.2, 0.4, 0.8, 1.0, 2.0, 4.0, or 8.0 mM in reaction buffer, in triplicate. CE conditions were as described above with 8 to 10 min of separation at −15 kV. Data were analyzed using GraphPad Prism (version 6.0) software.

Inhibitor screening, IC₅₀ determination, and mechanism of inhibition.

Twenty compounds known to act as potent inhibitors of mammalian NTPDases were tested for their ability to inhibit the PhaHV1 NTPDase-catalyzed conversion of ATP to ADP and AMP. Briefly, 2.4 μg of vNTPDase and 400 μM ATP were incubated for 2 h at 37°C with 10 μM each compound (dissolved in 10% DMSO). Inhibitor-free controls (enzyme plus substrate only) were included in triplicate as blanks, and inhibition values were determined relative to blank conversion values. Reaction samples were diluted 1 in 4 in reaction buffer and analyzed using CE and MG assays. CE conditions were as described above, with 14 to 16 min of separation at −10 kV, and the results of the MG phosphate assays were measured on a PHERAstar plate reader (BMG Labtech) at 623 nm (74).

Concentration-inhibition curves were performed by incubating 2.4 μg of vNTPDase with 400 μM ATP for 2 h at 37°C with inhibitor at a concentration of 0.01, 0.03, 0.1, 0.3, 1.0, 3.0, or 10.0 mM in reaction buffer with 10% DMSO in triplicate. No-enzyme (0.01 μM inhibitor and 400 μM ATP only), no-inhibitor, and buffer-only samples were included as reaction controls. Phosphate concentrations were measured using MG phosphate assays, as described above (74). Percent residual protein activity values were calculated relative to the activity for the inhibitor-free controls.

The mechanism of inhibition was determined by testing the inhibitors at 0, 0.1, and 0.2 μM (in 10% DMSO) with substrate (ATP) at 0.1, 0.2, 0.4, 0.8, or 1 mM and 2.4 μg of vNTPDase protein (37°C, 2 h, in reaction buffer). The phosphate released was measured by the MG assay, and the type of inhibition was evaluated graphically from a Hanes-Woolf plot using GraphPad Prism (version 6.0) software (74).

Data availability.

The sequences were deposited in GenBank under accession numbers MG452721 and MG452722. Raw read data sets were submitted to the NCBI Sequence Read Archive (SRA) under BioProject number PRJNA477676.