The Genome Sequence of Herpes Simplex Virus Type 2

Abstract

The genomic DNA sequence of herpes simplex virus type 2 (HSV-2) strain HG52 was determined as 154,746 bp with a G+C content of 70.4%. A total of 74 genes encoding distinct proteins was identified; three of these were each present in two copies, within major repeat elements of the genome. The HSV-2 gene set corresponds closely with that of HSV-1, and the HSV-2 sequence prompted several local revisions to the published HSV-1 sequence (D. J. McGeoch, M. A. Dalrymple, A. J. Davison, A. Dolan, M. C. Frame, D. McNab, L. J. Perry, J. E. Scott, and P. Taylor, J. Gen. Virol. 69:1531–1574, 1988). No compelling evidence for the existence of any additional protein-coding genes in HSV-2 was identified.

The complete 152-kbp genomic DNA sequence of herpes simplex virus type 1 (HSV-1) was published in 1988 (56) and since then has been very widely employed in a great range of research on HSV-1. Additionally, results from this most studied member of the family Herpesviridae have fed powerfully into research on other herpesviruses. In contrast, although a substantial number of individual gene sequences have been determined for the other HSV serotype, HSV-2, the complete genome sequence for this virus has not been available hitherto. In this paper we report the sequence of the genome of HSV-2, strain HG52.

At a gross level the 155-kbp genome of HSV-2 is viewed as consisting of two extended regions of unique sequence (U_L and U_S), each of which is bounded by a pair of inverted repeat elements (TR_L-IR_L and IR_S-TR_S) (17, 66) (Fig. 1). There is a directly repeated sequence of some 254 bp at the genome termini (the a sequence), with one or more copies in the opposing orientation (the a′ sequence) at the internal joint between IR_L and IR_S (21). U_L plus its flanking repeats is termed the long (L) region, and U_S with its flanking repeats is termed the short (S) region. In individual molecules of HSV-2 DNA, the L and S components may be linked with each in either orientation, so that DNA preparations contain four sequence-orientation isomers, one of which is defined as the prototype (66). The sequences of the terminal and internal copies of R_L and of R_S are considered to be indistinguishable.

FIG. 1.

Overall organization of the genome of HSV-2. The linear double-stranded DNA is represented, with the scale at the top. The unique portions of the genome (U_L and U_S) are shown as heavy solid lines, and the major repeat elements (TR_L, IR_L, IR_S, and TR_S) are shown as open boxes. For each pair of repeats the two copies are in opposing orientations. As indicated, TR_L, U_L, and IR_L are regarded as comprising the L region, and IR_S, U_S, and TR_S are regarded as comprising the S region. Plasmid-cloned fragments used for sequence determination are indicated at the bottom: BamHI and HindIII fragments are indicated by B and H, respectively, followed by individual fragment designations in lowercase; KH and HK indicate KpnI/HindIII fragments as described in the text.

This paper presents properties of the HSV-2 DNA sequence and our present understanding of its content of protein-coding genes and other elements. We are also interested in comparative analysis of the HSV-1 and HSV-2 genomes to examine processes of molecular evolution which have occurred since the two species diverged, and we intend to pursue this topic in a separate paper.

MATERIALS AND METHODS

Sources of virus and cloned DNA fragments.

HSV-2 strain HG52 was obtained from stocks in the Institute of Virology, Glasgow, United Kingdom (74). HindIII and BamHI fragments of HSV-2 strain HG52 cloned into pAT153 were provided by our colleague A. J. Davison. HindIII/KpnI fragments representing the two ends of U_S with adjacent parts of R_S were cloned into plasmids during the course of the sequence determination project. A plasmid clone of HSV-2 strain 25766 BamHI d was from A. C. Minson.

Sequence determination.

Large plasmid-cloned fragments of HSV-2 DNA were fragmented by sonication and subcloned into M13mp series vectors to give random libraries, and sequences from M13 clones were then determined by standard methods (3, 55). The programs of Staden (67) were used to overlap the shotgun sequence data from sets of M13 clones, to assemble databases for plasmid clones, and to edit the assembled sequences. Sequencing problems associated with G+C-rich DNA were resolved as previously described (55, 56). The complete genome sequence was assembled by utilizing overlaps between sequences represented in neighboring plasmid clones or, where sequences from adjacent plasmid clones abutted but did not overlap, by employing PCR amplification across these junctions with whole virus DNA as the template followed by sequence determination of the PCR product.

Sequence interpretation.

Evaluations of the gene content and other aspects of the HSV-2 sequence were carried out with the Genetics Computer Group program set (30). Database searches used FastA, TfastA, and Blast. The program Diverge (version 9; Genetics Computer Group) was used to compute nonsynonymous (i.e., causing amino acid substitution) and synonymous (silent) substitutional divergences (K_a and K_s, respectively) between pairs of aligned HSV-1 and HSV-2 coding sequences. It should be noted that K_a and K_s are not simply scores of differences: they estimate numbers of mutations that have occurred for each class as a fraction of the total number of sites of the same class and make allowance for multiple hits and differing transition and transversion rates (44).

Nucleotide sequence accession number.

The genome sequence of HSV-2 strain HG52 has been submitted to the EMBL Sequence Library (accession no. Z86099).

RESULTS

Determination of the DNA sequence of HSV-2 strain HG52.

The DNA sequence of HSV-2 strain HG52 was determined by using plasmid-cloned fragments of the genome, as indicated in Fig. 1. The sequences of the central part of U_S (59), the whole of R_L (55), and small parts of U_L (7, 55) were reported previously. The U_S region was completed by using two cloned KpnI/HindIII fragments running from KpnI sites in the flanking R_S elements to HindIII sites proximal to each extremity of U_S; these fragments, together with data from BamHI g (55) for the part of R_S adjacent to R_L, also provided the R_S sequence. The major part of U_L was sequenced with clones of HindIII fragments b, n, h, e, and a (17). Locations of genomic termini were obtained from the results of Davison and Wilkie (21).

Because of the accepted equivalence of terminal and internal copies for each pair of major repeats, we did not determine separate sequences for the whole of both copies of each repeat (TR_L-IR_L and IR_S-TR_S). Complete sequences were obtained for R_L and for R_S, and these were supplemented by sequences running from each extremity of U_L and U_S into the adjacent repeat and across the joint between IR_L and IR_S. For the purpose of assembling the complete genome sequence, internal parts of the determined R_L and R_S sequences were then used for both the terminal and internal copies. The complete genomic sequence was assembled in the prototypic orientation, with a single copy of the a sequence at each terminus and one in the opposing orientation (a′) at the joint between the L and S regions.

In the genomes of both HSV-1 and HSV-2, there is an origin of DNA replication located near the center of U_L (termed Ori_L) and also a distinct origin (Ori_S) with copies in both IR_S and TR_S (see Table 3). Studies on Ori_L of both HSV-1 and HSV-2 have defined this element as a palindrome of some 136 bp overall, and general experience has been that it is highly prone to deletion from plasmid clones carried in Escherichia coli (47, 64, 80). We therefore expected to obtain a version of the sequence lacking Ori_L from the HSV-2 HindIII e plasmid used for sequence determination across the Ori_L locus. In the event, sequences were obtained that represented both a majority population of molecules with Ori_L deleted and a minority population with Ori_L intact, as judged by comparison with the analysis by Lockshon and Galloway (47) of Ori_L in HSV-2 strain 333; i.e., the plasmid preparation contained both intact and deleted versions of the Ori_L sequence. We incorporated the intact version into our genomic sequence.

TABLE 3.

Locations of protein-coding regions and other features in the HSV-2 genomic sequence

ORF or feature	Sense	Location		Poly(A) signal position	Size (codons)	Divergencesa		Comment(s)b
		Start	End			Ka	Ks
a sequence		1	254					Terminal direct repeat
RL1	+			1738	261	0.272	0.660	Neurovirulence factor
Exon 1	+	440	934		165
Exon 2	+	1089	1376		96
RL2	+			5618	825	0.316	0.464	Immediate-early protein; modulator of cell state and gene expression
Exon 1	+	2303	2377		25
Exon 2	+	2785	3463		226
Exon 3	+	3645	5365		574
LAT	−	7732		153827				LAT initiation site; poly(A) site in circularized genome
Start of UL		9298
UL1	+	9427	10098	11019	224	0.247	0.554	Virion surface glycoprotein L
UL2	+	10211	10975	11019	255	0.095	0.577	Uracil-DNA glycosylase
UL3	+	11033	11731	11800	233	0.159	0.646	Nuclear phosphoprotein
UL4	−	11935	12537	11846	201	0.156	0.678
UL5	−	12607	15249	11846	881	0.065	0.480	Component of DNA helicase-primase
UL6	+	15248	17281	18158	678	0.075	0.361	Minor capsid protein
UL7	+	17259	18146	18158	296	0.125	0.379
UL8	−	18410	20665	18393	752	0.120	0.417	Component of DNA helicase-primase
UL9	−	20718	23318	18393	867	0.064	0.369	Ori binding protein
UL10	+	23209	24609	24629	467	0.116	0.470	Virion membrane glycoprotein M
UL11	−	24813	25100	24809	96	0.156	0.367	Myristylated tegument protein
UL12	−	25019	26878	24809	620	0.116	0.407	DNase
UL13	−	26922	28475	24809	518	0.079	0.369	Protein kinase; tegument protein
UL14	−	28232	28888	24809	219	0.080	0.278
UL15	+			34824	734	0.030	0.411	Role in DNA packaging
Exon 1	+	28969	29997		343
Exon 2	+	33597	34769		391
UL16	−	30146	31261	30142	372	0.147	0.328	Proposed initiator CTG codon
UL17	−	31366	33471	30142	702	0.103	0.454
UL18	−	35118	36071	35049	318	0.047	0.511	Capsid protein
UL19	−	36451	40572	36275	1,374	0.033	0.358	Major capsid protein (start ATG quoted is second possible)
UL20	−	40887	41552	36275	222	0.092	0.448	Virion membrane protein
UL21	+	42201	43796	43808	532	0.091	0.424	Tegument protein
UL22	−	44019	46532	44015	838	0.129	0.521	Virion membrane glycoprotein H
UL23	−	46873	48000	46819, 46806	376	0.158	0.413	Thymidine kinase (2 possible poly(A) sites)
UL24	+	47902	48744	52994	281	0.132	0.511
UL25	+	49037	50791	52994	585	0.072	0.433	Virion protein; roles in penetration and virus assembly
UL26	+	51029	52939	52994	637	0.156	0.485	Capsid maturation protease
UL26.5	+	51953	52939	52994	329	0.202	0.573	Capsid assembly protein
UL27	−	53406	56117	53367	904	0.072	0.343	Virion membrane glycoprotein B
UL28	−	56128	58482	53367	785	0.058	0.317	Role in DNA packaging
UL29	−	58860	62447	58805	1,196	0.043	0.340	Single-stranded DNA binding protein
OriL		62862	62997					Origin of DNA replication; location of palindrome given
UL30	+	63265	66984	67021	1,240	0.050	0.329	DNA polymerase catalytic subunit
UL31	−	66935	67849	66850	305	0.065	0.330
UL32	−	67845	69638	66850	598	0.069	0.416
UL33	+	69637	70026	71464	130	0.046	0.207	Role in DNA packaging
UL34	+	70119	70946	71464	276	0.154	0.576	Membrane-associated phosphoprotein
UL35	+	71061	71396	71464	112	0.090	0.514	Capsid protein
UL36	−	71569	80934	71503	3,122	0.129	0.410	Very large tegument protein (reiterations omitted for calculation of Ka and Ks)
UL37	−	81237	84578	81206	1,114	0.090	0.350	Tegument protein
UL38	+	85061	86458	86578	466	0.115	0.464	Capsid protein
UL39	+	87024	90449	91553	1,142	0.110	0.381	Ribonucleotide reductase large subunit
UL40	+	90505	91515	91553	337	0.065	0.383	Ribonucleotide reductase small subunit
UL41	−	91800	93275	91728	492	0.082	0.498	Tegument protein; host shutoff factor; defective in HSV-2 (HG52) (see text)
UL42	+	93769	95178	95250	470	0.171	0.518	DNA polymerase subunit
UL43	+	95433	96674	96689	414	0.233	0.476	Probable membrane protein
UL44	+	96979	98418	99294	480	0.209	0.447	Virion membrane glycoprotein C
UL45	+	98651	99166	99294	172	0.138	0.644	Tegument/envelope protein
UL46	−	99432	101597	99361	722	0.139	0.404	Tegument protein
UL47	−	101685	103772	99361	696	0.106	0.377	Tegument protein
UL48	−	104297	105766	104201	490	0.078	0.525	Tegument protein; transactivator of immediate-early genes
UL49	−	106250	107149	106213	300	0.198	0.522	Tegument protein
UL49A	−	107491	107751	106213	87	0.330	0.394	Probable virion membrane protein
UL50	+	107759	108865	108905	369	0.151	0.529	Deoxyuridine triphosphatase
UL51	−	109076	109807	109048	244	0.170	0.547
UL52	+	109875	113072	114309	1,066	0.099	0.462	Component of DNA helicase-primase; ATG initiator codon quoted corresponds to HSV-1 (see text)
UL53	+	113027	114040	114309	338	0.084	0.484	Membrane glycoprotein K
UL54	+	114589	116124	116133	512	0.129	0.482	Immediate-early protein; posttranslational regulator of gene expression
UL55	+	116342	116899	116941	186	0.077	0.526
UL56	−	117078	117782	117044	235	0.262	0.563
Start of IRL		117987
LAT	+	119518		127915				LAT initiation and poly(A) sites
RL2	−			121627	825	0.316	0.464	Immediate-early protein; modulator of cell state and gene expression
Exon 3	−	121885	123605		574
Exon 2	−	123787	124465		226
Exon 1	−	124873	124947		25
RL1	−			125507	261	0.272	0.660	Neurovirulence factor
Exon 2	−	125874	126161		96
Exon 1	−	126316	126810		165
a′ sequence		126996	127249					Opposite-sense copy of sequence directly repeated at genomic termini
RS1	−	128079	132032	127956	1,318	0.146	0.296	Immediate-early protein; transcriptional regulator
OriS		132623	132898					Origin of DNA replication; limits given are for directly repeated 138 nucleotides
Start of US		133707
US1	+	133739	134977	135019	413	0.241	0.706	Immediate-early protein; intron in 5′ noncoding region
US2	−	135167	136039	135144	291	0.163	0.703
US3	+	136325	137767	140050	481	0.150	0.474	Protein kinase
US4	+	137878	139974	140050	699			Virion membrane glycoprotein G
US5	+	140287	140562	143587	92	0.507	0.577	Putative membrane glycoprotein J
US6	+	141016	142194	143587	393	0.105	0.471	Virion membrane glycoprotein D
US7	+	142399	143514	143587	372	0.225	0.637	Virion membrane glycoprotein I
US8	+	143843	145477	146208	545	0.173	0.460	Virion membrane glycoprotein E
US8A	+	145329	145766	146208	146	0.215	0.473	Nucleolar protein
US9	+	145867	146133	146208	89	0.106	0.559	Tegument protein
US10	−	146628	147533	146618	302	0.238	0.546	Virion protein
US11	−	147247	147699	146618	151	0.261	0.464	Nucleolar, RNA binding protein
US12	−	147778	148035	146618	86	0.303	0.647	Immediate-early protein; inhibitor of antigen presentation; intron in 5′ noncoding region
Start of TRS		148036
OriS		148844	149119					Origin of DNA replication; limits given are for directly repeated 138 nucleotides
RS1	+	149710	153663	153781	1,318	0.146	0.296	Immediate-early protein; transcriptional regulator
a sequence		154493	154746					Terminal direct repeat

^aDivergence between HSV-1 and HSV-2 coding regions, as described in the text. 

^bProperties and functions of genes and proteins depend mostly on analyses of HSV-1, as described in recent reviews (19, 51, 60). For reasons of space, references are not supplied here. 

In a study of HSV-2 gene UL41 (which encodes a protein involved in shutting down host gene expression and is nonessential for growth in tissue culture), Everett and Fenwick (26) showed that in strain HG52 this gene is defective by reason of a frameshift mutation in its coding region (with this conclusion based on sequence data for four independent molecular clones). Our data concurred: relative to HSV-2 strain G, a single nucleotide was missing within a homopolymeric run of G residues at nucleotides 92249 to 92255. This posed for us the question of how best to represent the complete genomic sequence, in particular for lodging in sequence libraries. We decided that it would be most generally useful to have a library entry with a complete reading frame for gene UL41, and we have therefore inserted an extra G at position 92256 in our HSV-2 HG52 sequence.

The complete genome sequence of HSV-2 strain HG52 thus obtained comprised 154,746 bp with a G+C content of 70.4%. The sizes and base compositions of the major regions of the HSV-2 genome are shown in Table 1. The elevated G+C compositions of the major repeat elements, in particular of R_S, are striking. As with other herpesviruses, the genome of HSV-2 contains families of short reiterated sequences. The copy numbers of these elements are represented in the assembled sequence as found in the plasmid clones sequenced. Details of families of reiterations in HSV-2 R_L have been published previously (55). In HSV-2 U_L there are two reiterated sets: at nucleotides 72098 to 72266 within the coding region of gene UL36 and at nucleotides 106045 to 106165 between genes UL48 and UL49 (18, 32). There are two regions of reiterated sequences in HSV-2 R_S: one lies between the a sequence and the RS1 gene at nucleotides 127672 to 127914 in IR_S and 153828 to 154070 in TR_S, and the other, a complex assemblage of reiterations previously described by Whitton and Clements (81), is near the R_S-U_S boundaries at nucleotides 133227 to 133645 in IR_S and 148097 to 148515 in TR_S. In addition, Ori_S in HSV-2 HG52 is in effect duplicated relative to its HSV-1 strain 17 counterpart and comprises two adjacent similarly oriented copies of a 138-bp sequence (each copy containing an imperfect palindrome) (82). There are no reiterated families in HSV-2 U_S.

TABLE 1.

Summary of regions in the genomes of HSV-1 and HSV-2

Region	Virus	Length (bp)	% G+C	No. of genesa
RL (as TRL)	HSV-1	9,212	71.6	2
	HSV-2	9,297	75.4	2
UL	HSV-1	107,947	66.9	58
	HSV-2	108,689	68.9	58
US	HSV-1	12,980	64.3	13
	HSV-2	14,329	66.2	13
RS (as TRS)	HSV-1	6,677	79.5	1
	HSV-2	6,711	80.1	1
Whole genome	HSV-1	152,261	68.3	74
	HSV-2	154,746	70.4	74

^aEstimates of the number of genes encoding distinct proteins, as discussed in the text. 

Evaluation of the gene content of HSV-2.

We have previously published sequences for a number of genes of HSV-2 strain HG52, and other authors have reported sequences for various genes from several HSV-2 strains, as summarized in Table 2.

TABLE 2.

Published sequences of HSV-2 genes

Gene(s) or elementa	Strain(s)	Length (bp)	Reference
UL1–UL5 (part)	HG52	5,829	55
UL11–UL13 (part)	HG52	2,469	24
UL23, UL24 (part)	333	1,601	70
UL26, UL26.5	G	2,151	68
UL27, UL28 (part)	HG52	3,325	11
UL27, UL28 (part)	333	3,473	69
UL28 (part), UL29	8620K	5,105	75
OriL	333	326	47
UL30, UL31 (part)	186	3,769	76
UL38	G	2,158	84
UL39, UL40 (part)	333	3,998	71
UL39 (part)–UL41 (part)	333	1,890	28
UL40 (part)–UL42 (part)	HG52, G	3,653	26
UL43 (part)–UL46 (part)	333	3,410	72
UL47 (part), UL48	HG52	2,211	18
UL48	333	2,228	32
UL49A	HG52	2,851	7
UL52 (part), UL53	HG52	1,384	22
UL53 (part)–RL1	HG52	13,680	55
UL54 (part)	HG52	773	83
LAT, RL2 (part)	333	3,314	41
a sequence	HG52	348	21
OriS	HG52	1,100	82
US1 (part)	HG52	1,560	81
US2–US8 (part)	HG52	9,629	59
US6	G	1,635	79
US7	333	1,649	36
US8		1,880	14
US12 (part)	HG52	1,072	81

^aPreviously published sequences that contributed directly to the complete sequence for HSV-2 (strain HG52) are in boldface. 

Protein-coding genes of HSV-2 were identified primarily by comparison with corresponding sequences in the genome of HSV-1, since the DNA sequences for HSV-1 and HSV-2 are closely similar by the standards of the herpesvirus family (see the following section). Corresponding coding regions were in general readily identified by alignment of sections of HSV-1 and HSV-2 DNA sequences and by similarity between encoded amino acid sequences. Translational start and stop signals identified for equivalent HSV-1 and HSV-2 genes mostly correspond closely. In both genomes most genes show a strongly marked pattern of codon usage characterized by a very high proportion of G and C residues in the third positions of the set of codons employed (as expected in a G+C-rich genome, given the properties of the genetic code). As outlined in the following section, the patterns of divergence between the HSV-1 and HSV-2 DNA sequences also support our interpretation of gene content. In addition, we searched the HSV-2 genome sequence for possible additional protein-coding regions by using criteria of codon usage plus similarity and divergence between HSV-1 and HSV-2, and gene US8A was discovered by this route (see below). Finally, most of the proposed HSV proteins have recognizable homologs inferred from DNA sequences for other herpesviruses (for instance, varicella-zoster virus [20] and equine herpesvirus 1 [73]), and this too helps to ensure the validity of coding assignments.

We use for HSV-2 genes the nomenclature previously employed for HSV-1 (56–58, 62) and parts of HSV-2 (7, 55, 59). Table 3 and Fig. 2 present the set of HSV-2 protein-coding genes that we consider, from a critical and conservative viewpoint, can presently be identified with confidence. These number 58 in U_L, 2 in each copy of R_L, 1 wholly contained within each copy of R_S, and 13 in U_S (of which 2 extend their 5′ noncoding regions into the adjacent R_S elements), that is, 74 distinct genes, three of which are present in two copies each. Table 3 also lists the latency-associated transcript (LAT), which we presently regard as probably not protein coding (55). Each HSV-2 gene listed has an HSV-1 homolog, and all occur in locations and orientations corresponding to the HSV-1 versions. On the basis of limited available transcript mapping data for HSV-2 and the occurrence of AATAAA sequences associated with polyadenylation, HSV-1 and HSV-2 appear to have very closely equivalent sets of transcripts, with identical groupings of adjacent, similarly oriented genes into 3′-coterminal transcript families.

FIG. 2.

Layout of genes and other elements in the genome of HSV-2. The genome is shown expanded from the representation in Fig. 1, with unique regions as narrow open boxes and major repeats as wider boxes, on four successive lines; sequence numbering (in kilobases) is indicated above each section. Positions of the terminal repeat (a) sequences and the internal inverted copy (a′) are marked. Protein-coding regions and orientations for recognized genes as listed in Table 3 are shown as grey-filled arrow shapes (non-3′ exons are shown as box shapes). Locations of proposed transcript polyadenylation signals (AATAAA and variants thereof) are marked by small arrows (for rightward transcripts these are above the genome, and for leftward transcripts they are below). The positions of origins of replication are marked.

When we published the complete genome sequence of HSV-1 in 1988, we identified 70 genes encoding distinct proteins (56). As of the middle of 1997, four further HSV-1 protein-coding genes that we consider should be definitely added to the list have been identified, namely, UL26.5 (45, 63), UL49A (5, 7), RL1 (15, 23), and US8A; all of these have HSV-2 homologs. Given the conservation of gene content between HSV-1 and HSV-2 and the various facets of evidence supporting assignments, we have confidence in the reality of all the gene assignments listed in Table 3. The status of possible additional protein-coding genes is discussed below.

Comparisons of the HSV-1 and HSV-2 genomes.

As noted above, the genomic sequences of HSV-1 and HSV-2 are closely related. Comparison of HSV-2 sequence data with the previously determined sequence of HSV-1 strain 17 was therefore included as a late-stage check during the sequencing process. This revealed occasional problems in the HSV-2 data that required further attention and also led to the uncovering of several errors in the published HSV-1 sequence. Changes to UL56 and RL1 of HSV-1 were reported previously (23, 55), and we now add corrections in the UL14, UL46, and US8A coding sequences; these are all described in Table 4 and in the next section. To the best of our understanding, these changes do not affect interpretation of HSV-1 genome organization outside the reading frames named. The revisions to the HSV-1 sequence have been communicated to the EMBL Sequence Library (accession no. X14112).

TABLE 4.

Corrections to the genome sequence of HSV-1 strain 17

Gene	Locationa	Alteration	Reference
RL1 (TRL copy)	780	Change T to C	23
	818	Change C to G
	823	Delete CG
UL14	28866	Change C to GCG	This paper
UL46	100407	Insert G	This paper
	100446	Delete G
UL56	116344	Insert CG	55
RL1 (IRL copy)	125547	Delete CG	23
	125553	Change G to C
	125591	Change A to G
US8A	143173	Insert G	This paper

^aNumbers refer to the DNA sequence of HSV-1 (strain 17) as listed in reference 56. 

HSV-1 and HSV-2 comprise the most closely related pair of herpesviruses for which complete genome sequences are presently known, but their genomes are nonetheless substantially diverged; in an analysis of herpesvirus phylogeny, we estimated the divergence date of the two HSV lineages as 8 million years ago (53, 54). Comparisons of aspects of their genome structures should help in interpreting the contents of the genomes and in assessing differences in functional capabilities of the viruses and will also be of interest in illuminating a relatively early stage of evolutionary divergence. In this paper our attention is primarily on aspects of gene content and on a comparative description of the two genomes.

The determined lengths of the HSV-1 and HSV-2 sequences are 152,261 and 154,746 bp, respectively. Neither of these values is unique, since high-frequency variation in lengths of homopolymer runs and in copy numbers of families of short reiterated sequences will act to distribute lengths of genome molecules over a range of perhaps several hundred base pairs. As shown in Table 1, the major difference in length is located in the U_S region, which in HSV-2 is 1,349 bp longer. This is almost all attributable to the US4 gene, which in HSV-1 appears to have suffered a large internal deletion (50, 59). U_L is 742 bp longer in HSV-2, the net result of many small differences in lengths (in both directions) between corresponding coding and noncoding constituents of the two U_L sequences. The sizes determined for each of the major repeats are very close for the two genomes. As was previously well known from buoyant density analyses of the virus DNAs, HSV-2 has a slightly more G+C-rich genome than HSV-1 (2.1% higher overall from the sequences). The difference is distributed across the genome, being greatest in R_L and least in R_S (Table 1). Within coding regions, the effect is most pronounced in the sets of nucleotides constituting the third positions of codons (data not shown).

Characteristics of the aligned pairs of protein-coding regions were examined for all genes (except US4, which in HSV-1 is grossly truncated). This set of alignments required introduction of gapping characters representing 2% of the total alignment length. Excluding such gapped regions, the overall incidence of identical aligned nucleotides was 83%. Noncoding regions are typically much less conserved; this topic was previously discussed in some detail for the major part of U_S (59) and for R_L, which contains the most diverged regions of the genomes (55). The families of short reiterations found in both genomes are conserved only partially in their locations and are distinct in their sequences.

Nucleotide substitutional differences between aligned pairs of coding regions were examined for all genes, except US4, in terms of whether they correspond to a substitution at the amino acid level. The general expectation is that across the gene set the synonymous divergence, K_s, should be approximately constant and that K_s should be greater than the nonsynonymous divergence, K_a. Table 3 lists K_a and K_s for pairs of aligned HSV-1 and HSV-2 gene sequences. In every case K_s is higher than K_a. There is substantial variation within each of the K_s and K_a datasets; in both sets differences from the mean diminish sharply with increasing size of open reading frames (ORFs), suggesting pronounced stochastic effects, especially with the values for the smaller ORFs. Number averages taken over the whole set are 0.47 for K_s and 0.14 for K_a. Plotting the K_a/K_s ratio against ORF length gives a good view of the relative magnitudes of the two measures of divergence for single genes and the spread of values, as shown in Fig. 3. There are four outliers in this plot that have notably high K_a/K_s ratios for their ORF lengths, namely, UL49A, US5, RL2, and RS1. We consider that these can all be adequately rationalized in terms of specific peculiarities and do not have any wider significance for our present purposes. The first two are very small genes considered to encode membrane proteins (7, 58), and the high relative incidence of nonsilent mutations probably reflects both stochastic effects and nonstringent requirements for amino acid sequence conservation. For RL2 and RS1 the atypical substitution ratios may be associated with the genes’ locations in G+C-rich major repeat elements, where recombinatory processes are thought to be particularly active (57), and also with the sensitivity of procedures estimating K_a and K_s toward inaccuracy with sequences of highly biased base composition (39). It is clear from this analysis that there are no genes showing strong positive selection effects overall (i.e., with K_a greater than K_s). For the great majority of identified genes in HSV-1 and HSV-2, the pattern of nucleotide substitutions accumulated during the divergent evolution of the two genomes is thus thoroughly consistent with coding assignment.

FIG. 3.

Plot of K_a/K_s ratio versus length of coding region for each aligned pair of HSV-1 and HSV-2 genes. The ratio of nonsynonymous to synonymous divergences (K_a/K_s; see Table 3) for pairs of HSV-1 and HSV-2 coding sequences (excluding UL26.5 and US4) is plotted against the corresponding ORF length (average of HSV-1 and HSV-2 lengths). The median value for K_a/K_s is indicated with a dashed line. Plotted points for four outliers with high K_a/K_s values are annotated with gene names.

Notes on individual genes and proteins of HSV-1 and HSV-2.

In this section we present background, evaluation, and comments on certain genes of HSV-1 and HSV-2 where sequence determination and interpretation raised particular points of difficulty or interest not previously discussed. Table 3 includes a summary of gene functions, mostly as understood from studies on HSV-1 (19, 52, 60).

(i) Gene UL14.

The sequences of HSV-1 and HSV-2 showed a relative frameshift near the 5′ end of the UL14 ORF, which was resolved by the finding of an error in the HSV-1 sequence on reexamination of the region (Table 4). The UL14 ORF of HSV-1 now starts at an ATG upstream of the previously proposed start, so that 14 codons at the 5′ end of the old version are replaced by 18 new codons.

(ii) Gene UL16.

The HSV-2 UL16 ORF does not possess any candidate ATG for translational initiation, and we have assigned translational initiation to the codon CTG (Leu) aligned with the ATG of HSV-1 UL16. Sequence analysis of this region for the distinct HSV-2 strain 25766 gave a sequence identical to that of HSV-2 HG52. We note that CTG (CUG) is a known start codon for eukaryotic genes and indeed is the most frequently described among non-AUG start codons (8).

(iii) Gene UL19.

The position of the start codon for the HSV-2 UL19 ORF as listed in Table 3 is equivalent to that proposed for HSV-1. There is another possible ATG start codon in the HSV-2 sequence, 10 codons upstream and in frame, which lacks an HSV-1 counterpart.

(iv) Gene UL36.

In both HSV-1 and HSV-2 the very large UL36 ORF (encoding a tegument protein) contains a set of reiterated sequences, which are presumed to be translated. The HSV-1 set encodes the amino acid sequence (PQ)₃₅, while the distinct set in HSV-2 encodes (PQPPL)₁₁. This region in the protein thus has the appearance of a flexible linker, whose requirement for conservation of sequence and length is not stringent.

(v) Genes UL46 and UL47.

There are published sequences for genes UL46 and UL47 for both HSV-1 strain 17 and HSV-1 strain F, and for both genes the pairs of aligned sequences show relative frameshifting differences (56, 61). We have corrected a local frameshifted region affecting codons 170 to 182 in UL46 of HSV-1 strain 17 after redetermining the sequence for the region (Table 4). For both genes, comparisons with the HSV-2 versions suggest that differences remaining between the sequences of the two HSV-1 strains probably represent frameshifts in the HSV-1 strain F data.

(vi) Gene UL52.

The HSV-2 UL52 ORF of 1,066 codons as listed in Table 3 is the distal portion of an ATG-initiated, 1,240-codon ORF that extends over most of the adjacent, oppositely oriented UL51 ORF. The upstream part was discounted because it does not have an equivalent in HSV-1 or other herpesviruses, and the start codon given in Table 3 corresponds closely to that in HSV-1.

(vii) Gene US8A.

US8A is an additional reading frame proposed for both HSV-2 and HSV-1, overlapping the 3′ portion of the US8 ORF and extending into the region between US8 and US9. The presence of a HSV-1 gene (US8.5) in this location and detection of its product in infected cells were described by Georgopoulou et al. (31). Our assignment differs in that the sequence correction shown in Table 4 changes the reading frame from codon 144 of the 159-codon HSV-1 US8A ORF. Among members of the Simplexvirus genus, DNA sequences corresponding to this genomic region are also available for herpesvirus B and for simian agent 8 (25, 40). We have found counterparts of US8A (not detected by the original investigators) in both these sequences. In both cases frameshift corrections have to be proposed, but the conservation of encoded protein sequences is compelling for the C-terminal region, as shown in Fig. 4. In the Varicellovirus genus (the other genus in the Alphaherpesvirinae subfamily), no convincing homologs were found, although there is a possible counterpart in equine herpesvirus 1 (gene 75) (73) with no sequence similarity but in a corresponding genomic location. Our current evaluation is that the US8A gene family is probably specific to the Simplexvirus genus. Its function remains obscure.

FIG. 4.

US8A amino acid sequences for HSV-1, HSV-2, and two simian alphaherpesviruses. Proposed amino acid sequences for US8A in HSV-1 and HSV-2 are aligned together with sequences of counterparts for herpesvirus B (BV) and simian agent 8 (SA8) as interpreted from the DNA sequence data of references 40 and 25, respectively. In the consensus (Cons) line, completely conserved residues are indicated in uppercase and residues conserved in three of the sequences are indicated in lowercase. The position corresponding to the end of the US8 ORF in the DNA sequences (−1 frame relative to US8A) is marked (– – – |). In BV and SA8 a region of uncertainty representing the limits for a proposed shift of reading frame in the reported DNA sequence is shown as bounded by //// and \\\\, with inserted gapping characters indicated by dots. The position corresponding to a presumptive frameshift in SA8 only is marked by •. The location corresponding to the identically placed end of the ORF in all four sequences is marked ∗.

(viii) Gene US12.

The products of US12 (ICP47 or Vmw12) have been shown for both HSV-1 and HSV-2 to interdict antigen presentation in infected cells by binding to the TAP transporter (27, 29, 34, 35, 85). Optimization of alignment of the two coding sequences (by using Bestfit with default and also other values for gapping parameters) indicated a relative deletion of 13 nucleotides in the HSV-1 sequence around codon 59, which was judged to be convincing in terms of the incidence of identical aligned residues that it gave (62% in the 23 following codons, versus 72% in the 58 upstream codons, in heteropolymeric sequences and without any other gapping). This implies that the two sequences would be read in different frames for the distal 24 and 23 codons of the HSV-1 and HSV-2 ORFs, respectively (29). No error was apparent in either DNA sequence, and both US12 proteins have been sufficiently characterized to give good confidence in the interpretation of the coding arrangements. Thus, the shift of reading frame is taken as genuine, being the result of a mutation that has occurred in one of the HSV lineages, and this is consistent with experimental data showing that the C-terminal parts of the proteins are not required for TAP binding (27, 29, 35, 85). This is the only unambiguous instance of a frameshift within still-functional coding regions that emerged from comparisons of the two HSV genomic sequences.

US12 gene homologs are known only for HSV-1 and HSV-2 and appear to be specific to the Simplexvirus genus. It thus seems possible that this small gene evolved de novo within this virus group. Nuclear magnetic resonance and circular dichroism analyses of the HSV-2 US12 protein indicate an absence of stable secondary structure (6), so the protein may be better regarded as an oligopeptide with a TAP binding epitope than as a typical globular protein. This view is also thoroughly compatible with the small size of the peptide chain and the lack of conservation in the C-terminal part and is particularly attractive in considering evolution of the gene de novo.

Evaluation of possible additional genes of HSV-1 and HSV-2.

In addition to the 74 distinct genes identified in this paper, other HSV-1 genes have been proposed in recent years. For some of the latter there are readily visible equivalents in the HSV-2 sequence, and for others there are not. The possible extra genes of HSV-1 and HSV-2 fall into three classes with respect to their relationship to genes in the canonical list of Table 3. In the first class are those with coding regions based on a translational start internal to and in frame with an established coding ORF, and in the second class are genes that occupy the same DNA sequence as a recognized coding ORF but in the opposite sense. Finally, some proposed ORFs are in regions distinct from established ORFs.

The use of an internal start in an ORF is now well established for the UL26.5 gene in HSV-1 (46, 65) and HSV-2 (68) and for homologs in other herpesviruses (37). In summary, the 5′ portion of the UL26 ORF encodes a proteinase activity, and the 3′ portion (UL26.5) encodes a scaffolding component for capsid assembly. Two transcripts are produced, which initiate translation, respectively, at the 5′ end of the ORF (UL26) and internally (UL26.5). Processing of the resulting translation products is carried out by the proteinase. Evidently the proteinase and scaffolding functions are each active both in the product representing the whole UL26 ORF and in the relevant separate domain.

The next best characterized case in HSV-1 concerns gene UL12, encoding DNase. It has long been known, for both HSV-1 and HSV-2, that there is a transcript initiated within the UL12 ORF with an ATG codon appropriately located in the UL12 reading frame to act as a translational initiator in this RNA (24). It has recently been shown for HSV-1 that this 3′ portion of the UL12 ORF (termed UL12.5) is translated as expected from the transcript described and that the protein product is present in virions (9, 49). However, no distinct phenotype has yet been shown to depend on expression from the UL12.5 start, and we have therefore omitted UL12.5 from our gene list, at least for now. A less well characterized but formally similar case concerns the so-called UL8.5 ORF of HSV-1 (actually the 3′ portion of UL9) (4), for which a late transcript, an initiator ATG in the UL9 ORF, and a protein product have been identified. We note that an equivalent ATG exists in the HSV-2 sequence (nucleotides 22178 to 22176). The same phenomenon has been described for HSV-1 US1 (12), but in this case there is not an equivalently placed ATG in HSV-2. The final example concerns HSV-1 gene UL15, which has two exons, both protein coding. A protein species smaller than the product of the complete gene that apparently is translated from all or part of the downstream exon only and is probably the product of a separate translational initiation event and separate mRNA has been observed; no distinct phenotype associated with this UL15.5 protein has been detected (1, 2, 86). There are several ATG codons in the HSV-1 exon 2 ORF that might act as translational initiators for the protein, and all are conserved in HSV-2.

Turning to the class of proposed genes with ORFs antisense to already recognized functional ORFs, we address two instances where the HSV-1 loci have been investigated experimentally, namely, UL43.5 and ORF P. We note first that ORFs other than authenticated protein-coding ORFs are rather common in the HSV genomes, as a consequence of the high G+C contents and the resulting low incidence of adventitious stop codons. This topic is developed further in Discussion.

Ward and colleagues (78) have described expression in HSV-1-infected cells of a protein from a 311-codon ORF antisense to UL43, termed UL43.5. The protein was produced late in infection and was associated with nuclear structures. The UL43 region is not essential for growth of virus in tissue culture (48). There is no counterpart in the HSV-2 sequence to the ATG that in HSV-1 opens the UL43.5 ORF, and the equivalent HSV-2 reading frame has stop codons at positions corresponding to codons 16 and 68 in the HSV-1 ORF. A resemblance to a motif of aminoacyl-tRNA synthetase noted in the HSV-1 UL43.5 amino acid sequence (78) is absent from the HSV-2 counterpart, and its locus lies between the two stop codons mentioned. After the second stop codon, the HSV-2 reading frame is open until a location equivalent to five codons past the end of the HSV-1 ORF, but the first potential initiating ATG is well downstream in this HSV-2 ORF at a position equivalent to codon 163 of HSV-1 UL43.5. The HSV-2 sequence thus does not appear to be a likely candidate for a protein-coding region.

ORF P lies in the R_L element of HSV-1 within the region traversed by LATs and in the LAT orientation. It is antisense to and largely overlaps the ORF of gene RL1 (43). Mutation of an ICP4 binding motif in the proposed promoter for ORF P allowed expression of the ORF in infected cells, giving a protein of 248 residues in HSV-1 strain F and 233 residues in HSV-1 strain 17 (with the difference resulting from distinct copy numbers of a reiterated element). The ORF P protein was observed to associate with splicing factors, and it was proposed that it could have a gene-regulatory function, perhaps in latently infected cells (10, 42). In HSV-2 the 5′ portion of the ORF P region is preserved, with a putative ATG initiator codon corresponding to that in HSV-1. This 5′ part of ORF P corresponds on the opposite strand to a part of the RL1 coding sequence specifying a 63-amino-acid sequence that is highly conserved among HSV-1, HSV-2, and cellular homologs (16, 51). Thus, by the criteria of sequence interpretation, the similarity of the 5′ parts of the ORF P frames in HSV-1 and HSV-2 appears to be primarily as a consequence of coding requirements of the RL1 gene. The HSV-2 RL1 gene contains an intron (33, 55), unlike its HSV-1 equivalent, and in the ORF P phase the intron sequence causes divergence from ORF P of HSV-1 after a position corresponding to codon 48 of HSV-1 ORF P. After a 50-codon stretch traversing the intron sequence (which consists mostly of reiterations of a 19-bp sequence), the HSV-2 ORF reenters sequence with an HSV-1 counterpart but is now out of phase with HSV-1 ORF P, and it then terminates after 130 codons in all. These pronounced differences render it rather unlikely that HSV-1 ORF P and the HSV-2 ORF encode equivalent functional proteins.

In the third class of potential additional genes, a number of coding regions in HSV-1 that are unrelated to other ORFs have been proposed. These include ORFs in the LAT region of R_L (77) and an ORF across Ori_S (38). These are not conserved convincingly in HSV-2, as previously discussed for LAT (55).

This section has outlined the background for certain HSV-1 candidate genes and a comparative interpretation of corresponding loci in the HSV-2 genome. We found that this analysis fell short of yielding firm conclusions on protein coding and functionality, in that there was no strong support for extra HSV-2 genes but negative evaluations were not definitive; discussion of this topic is continued below.

DISCUSSION

As presented in this paper, HSV-1 and HSV-2 possess corresponding sets of 74 genes that encode distinct proteins. From the characteristics of the genomic sequences, from the comparative analyses described above, and from published experimental work on many genes, particularly of HSV-1, we have good confidence that these gene assignments are valid. The next question that arises about the gene sets of the viruses is whether the present listings are complete. Several extra genes have been proposed for HSV-1, and in some cases apparently substantive experimental evidence is published. We have found evaluation of the limits of the protein-coding capacities to be a complex and vexed matter, as outlined below.

Dealing first with proposed genes in which the coding sequence comprises a distal, in-frame subsection of the coding sequence of another gene, we noted above that most of the HSV-1 cases have equivalents in the HSV-2 sequence (“UL8.5,” UL12.5, and UL15.5, as well as the thoroughly characterized UL26.5), while US1 does not. However, all that is required for an equivalent downstream ORF to be registered for HSV-2 is conservation of a Met codon at the appropriate location. We are concerned that designating such sections of ORFs as distinct genes should be a matter of some caution. The UL26.5 gene is presently exceptional in this class in that an explicit functional basis is known, which clearly justifies the separate name. It appears that in the other examples, the novel transcript and/or protein species are typically observed late in the infectious cycle, that is, in a situation where control of virus gene expression may be collapsing, so it could be argued that functional significance is dubious. It may be that experimental observation of distinct, explicable phenotypes associated with a complete ORF (or, in the case of UL15, set of exons) and a sub-ORF will provide the only definitive evidence for these cases, and we consider that this state has not been attained for any of the proposed genes in this class, in either HSV-1 or HSV-2.

Turning to additional HSV-2 genes based on distinct ORFs (including antisense ORFs), the primary examples from HSV-1 in this class are UL43.5 and ORF P. As outlined above, neither of these has a compellingly conserved counterpart in HSV-2, so our evaluation is that the existence of functional HSV-2 homologs is unlikely in these cases. The topic of possible extra coding sequences in both genomes can be broadened: as already noted, both HSV-1 and HSV-2 sequences possess many ORFs in addition to those in the canonical gene set. To illustrate this point, Table 5 lists by length class the numbers of complete ORFs and of ATG-initiated ORFs for the HSV-2 sequence and compares these with the numbers of identified genes: the incidence of unassigned ORFs, up to quite large sizes, is striking. These data can be viewed in two complementary ways. First, they provide context, demonstrating that there is nothing out of the ordinary about any given moderately sized ORF that lacks a gene assignment. Second, they suggest, at least superficially, that here is a potential resource which evolution might turn to a protein-coding role in at least some cases.

TABLE 5.

ORFs in the HSV-2 sequence

Length range (codons)	No. of:
	ORFsa		Genes as M-ORF equivalentsb	Unassigned M-ORFs
	C-ORFs	M-ORFs
51–100	752	200	6	194
101–200	416	143	7	136
201–400	212	79	25	54
401–600	52	24	16	8
601–800	23	12	10	2
801–1,000	7	7	4	3
1,001–1,200	5	3	3	0
1,201–1,400	7	4	4	0
>1,400	1	1	1	0

^aORFs lying wholly within R_L or R_S were counted once only. C-ORF denotes a complete ORF, and M-ORF denotes an ATG-initiated ORF. 

^bData for genes are based on Table 3, but with reading frames scored for length as their M-ORF equivalents. 

Our evaluation of the presence of protein-coding regions primarily involved criteria based on characteristics of DNA and protein sequences, as distinct from experimental observation of transcripts and proteins or from functional studies. We were thus concerned with evolutionary conservation and divergence of coding DNA sequences, with similarities in predicted protein sequences, and with patterns of codon usage and amino acid composition (these last two are not detailed in this paper). This approach clearly distinguished properties of the standard protein-coding ORFs from those of other frames and regions. However, a potential weakness is that it might fail to discern functional coding regions that are far from the norm for these criteria. For instance, there might exist genuinely novel protein-coding sequences that have arisen only recently on an evolutionary time scale and are therefore present in only one of the two HSV genomes and are as yet unmarked by evolutionary change, and these could present as ORFs with atypical characteristics. We thus consider that analysis of sequence characteristics gives a clear positive evaluation of coding potential, but a negative indication must in principle be less than definitive. We have to conclude that if candidate genes of atypical structure are to be validated, then discriminating experimental analysis has to represent the final criterion. Again, we consider it appropriate to adopt a conservative outlook on adding candidate genes to the canonical list. While our main concern in this paper is with HSV-2’s genetic content, we regard the most-studied novel genes of HSV-1, UL43.5 and ORF P, as also still being of tentative status. For these and other antisense transcripts, the possibility of a regulatory role through annealing to complementary, sense transcripts remains a possibility, separate from any direct protein-coding function (13).

Concerning the possible correlation of genome contents with biological differences between HSV-1 and HSV-2, we consider it quite likely that, notwithstanding these unresolved aspects concerning the standing of extra genes, it might well turn out to be the case that there are no HSV-1 or HSV-2 genes that lack a homolog in the other virus. Nor is there any substantive indication in nucleotide substitution patterns (i.e., K_a high relative to K_s) of genes under strong differential selection between the two lineages, which might reflect evolutionary divergence into particular niches in the host organism. There is one major difference between genes that does command attention in this context: relative to HSV-2, there is a large deletion in the HSV-1 US4 gene, encoding virion glycoprotein G (59), and it remains plausible that this could correspond with some key difference (for instance, in cell tropisms). Nonetheless, our judgment is that differences in the behaviors of the viruses are likely for the most part to arise from multiple local differences in sequences of many proteins.

The HSV-2 genome sequence should provide a powerful aid to work on HSV-2 per se, integrating the previous quite extensive but incomplete and scattered information on sequences of HSV-2 genes. The analyses described in this paper were in essence comparative, with interpretation of the HSV-2 sequence depending heavily on the wider knowledge of HSV-1 gene content, and with availability of the two genomic sequences, significantly diverged but still quite close in evolutionary terms, allowing some interesting insights. Finally, the sequence for HSV-2 completes (belatedly) the set of determined genomic sequences for all of the eight known human herpesviruses.