Herpes simplex virus type 1 strain KOS carries a defective US9 and a mutated US8A gene

The DNA genomes of members of the subfamily Alphaherpesvirinae of the family Herpesviridae contain long and short unique regions (U_L and U_S), which are separated by inverted repeat sequences (Pellett & Roizman, 2007). Whereas many of the genes encoded within the U_L regions are conserved throughout the entire virus family, the U_S regions exclusively contain alphaherpesvirus-specific genes lacking detectable homologues in other subfamilies. Moreover, the gene content of the U_S region varies considerably even between different alphaherpesviruses.

In the genomes of herpes simplex viruses types 1 and 2 (HSV-1 and HSV-2), herpes B virus, as well as canine, equine, feline and several avian alphaherpesviruses, the ORF US8, which encodes envelope glycoprotein gE, is partly overlapped or immediately followed by an additional protein-encoding ORF of unknown function, which has been previously designated US8A or US8.5 (Fig. 1a⇓) (Davison, 2010; Georgopoulou et al., 1993, 1995; McGeoch et al., 1985; Perelygina et al., 2003; Telford et al., 1998; Tyack et al., 1997; Willemse et al., 1995). Also the adjacent ORF US9 of HSV-1 (Fig. 1a⇓) is not conserved throughout the subfamily Alphaherpesvirinae, but absent e.g. from the lymphotropic avian pathogens of the genus Mardivirus (Davison, 2010; Tulman et al., 2000).

All US8A and US9 homologues investigated to date are dispensable for virus replication in epithelial cell cultures. However, in the neuroinvasive mammalian alphaherpesviruses pseudorabies virus (PrV), bovine herpesviruses 1 and 5 (BoHV-1 and BoHV-5) and HSV-1, US9 has been reported to be critical for neuronal spread and neurovirulence (Brideau et al., 2000; Butchi et al., 2007; Chowdhury et al., 2002; Polcicova et al., 2005). Both US8A and US9 are predicted to encode type II membrane proteins (Davison, 2010; Sonnhammer et al., 1998), and the US9 gene products (pUS9) of HSV-1, BoHV-5 and PrV are phosphorylated and incorporated into virions (Brideau et al., 1998; Chowdhury et al., 2002; Frame et al., 1986). Detailed studies of PrV revealed that pUS9 localizes to the trans-Golgi network, and is specifically required for anterograde transport of nascent virus particles in neurons by targeting other viral envelope proteins, and, possibly indirectly, also nucleocapsids to axons (Brideau et al., 2000; Lyman et al., 2007; Tomishima & Enquist, 2001). Similar results have been described for the HSV-1 and BoHV-1 homologues (Butchi et al., 2007; Snyder et al., 2008). Conservation of pUS9 function was proven by complementation of US9-deleted PrV by the homologous proteins of equine herpesvirus 1 and BoHV-1, as well as of US9-deleted BoHV-5 by the BoHV-1 protein (Chowdhury et al., 2006; Lyman et al., 2009). However, pUS9 of HSV-1 or varicella-zoster virus (VZV) failed to compensate functionally for the PrV protein (Lyman et al., 2009), and recent studies indicate that pUS9 is dispensable for anterograde axonal transport of HSV-1 particles, whereas the product of the US8 gene, gE, is required (McGraw et al., 2009).

Since different results regarding the relevance of pUS9 for axonal transport of HSV-1 have been obtained using strains F (Snyder et al., 2008) and NS (McGraw et al., 2009), we attempted to investigate the role of this protein for HSV-1 strains KOS (Smith, 1964) and 17 (Brown et al. 1973). The latter two HSV-1 strains were originally isolated from human labial lesions, and are frequently used as parental wild-type strains for the generation of virus mutants and investigation of viral gene functions (Brown et al., 1973; Schaffer et al., 1973). Moreover, bacterial artificial chromosomes (BAC) containing the infectious full-length genomes of both strains have been created to permit convenient mutagenesis in Escherichia coli (Gierasch et al., 2006; Leege et al., 2009), and strain 17 has been used for determination of the first complete genome sequence of HSV-1 (McGeoch et al., 1985, 1988; GenBank accession no. X14112).

From this sequence, primers were deduced and used for PCR amplification (Pfx DNA Polymerase; Invitrogen) of the US8A and US9 genes of HSV-1 strains KOS and 17 (Fig. 1b⇑). Using artificially introduced EcoRI and XhoI restriction sites, the PCR products obtained with primers HUS8A-F and HUS9-R, or HUS9-F and HUS9-R, were cloned into appropriately digested eukaryotic and prokaryotic expression vectors (Fig. 1c⇑).

DNA sequence analyses of the obtained plasmid clones of HSV-1 strain 17 revealed 100 % identity with the published genome sequence (accession no. X14112). However, the newly determined DNA sequence of HSV-1 KOS (GenBank accession no. HM188403) exhibited three base substitutions within the investigated 971 bp fragment (Fig. 2a⇓). Identical changes were found in two independently obtained isolates of HSV-1 KOS from different laboratories, as well as in two available BAC clones of this virus (Gierasch et al., 2006; Leege et al., 2009).

All three ‘mutations’ of HSV-1 strain KOS are predicted to affect viral gene expression. The first exchange at nucleotide position 142802 (numbering according to the HSV-1 strain 17 genome sequence) is a T to C transition, which leads to the substitution of a methionine by a threonine codon at US8A position 20. The second base substitution (nt 143221; T to C) abrogates the stop codon of US8A present in HSV-1 strain 17 (Fig. 2a⇑). This leads to an extension of the deduced protein from 159 to 190 aa in strain KOS (Fig. 2b⇑).

The latter mutation also affects the predicted TATA-box of US9 between nt 143219 and 143224 (McGeoch et al., 1985), and reduces similarity to the consensus sequence TATA^T/_AA^T/_A (Breathnach & Chambon, 1981). Furthermore, another base substitution (nt 143384; C to T) not only deletes a SalI restriction site present in HSV-1 strain 17, but also introduces a stop codon at position 58 of US9 in HSV-1 strain KOS (Fig. 2a⇑). This mutation shortens the deduced US9 protein from 90 aa in HSV-1 strain 17 to 57 aa in strain KOS, and prevents expression of the predicted hydrophobic C-terminal transmembrane domain (Fig. 2b⇑). Thus, if expressed at all, pUS9 of HSV-1 strain KOS may not actually be functional as a membrane protein that localizes and targets other viral envelope proteins in neurons.

To test for differences in gene expression, RNA was prepared from cells infected with HSV-1 strain KOS or HSV-1 strain 17, and analysed by Northern blot hybridization (Fuchs & Mettenleiter, 1996) by using a labelled antisense cRNA of pcDNA3/US9 (Fig. 1c⇑) specific for the 3′-co-terminal mRNAs of US8, US8A and US9 (Fig. 1a⇑). This probe detected three viral transcripts of the expected sizes of approximately 2.5, 1.1 and 0.6 kb in cells harvested 6 h after infection with either virus (Fig. 3a⇓). However, densitometric analyses revealed that the intensity of the US9-specific 0.6 kb transcript was reduced by approximately 65 % in HSV-1 strain KOS- compared with HSV-1 strain 17-infected cells, whereas the amounts of the two other transcripts were similar. This was also found at earlier and later times after infection (data not shown), indicating that the TATA-box mutation decreases US9 transcription.

Fusion proteins containing pUS9 or the N-terminal 110 aa of pUS8A of HSV-1 strain 17 were isolated from bacteria that had been transformed with prokaryotic expression plasmids (Fig. 1c⇑), and used for the preparation of monospecific rabbit antisera (Fuchs et al., 2002). In Western blot analyses of cells infected with HSV-1 strain 17, the anti-US9 serum specifically detected several proteins with apparent molecular masses of 14–20 kDa, which were neither found in uninfected cells (Fig. 3b⇑, upper panel), nor detected by the pre-immune serum (data not shown). Although the calculated mass of wild-type pUS9 is only 10.0 kDa, the observed masses are in agreement with previous findings (Frame et al., 1986), and might be due to differential phosphorylation. Similar proteins were also detected in cells infected with several other HSV-1 strains (data not shown). However, no US9 gene products were detectable by Western blot in lysates of cells infected with HSV-1 strain KOS (Fig. 3b⇑, upper panel), although a truncated US9 protein of the calculated mass of 6.4 kDa should be visible on the blot, and although sera against other viral gene products such as pUL11 (Leege et al., 2009) showed similar reactivity with HSV-1 strain KOS- or 17-infected cells (Fig. 3b⇑, lower panel). The pUS9-specific rabbit serum showed strong indirect immunofluorescence reactions with HSV-1 strain 17-infected cells, whereas only very weak signals were found in strain KOS-infected cells (Fig. 3c⇑, left panels), which might have been caused by unspecific binding of antibodies to HSV-1 gE via their Fc-part (Para et al., 1982).

Cells were also transfected with eukaryotic expression plasmids containing the US9 ORFs of HSV-1 strains KOS or 17 under human cytomegalovirus immediate-early promoter control (Fig. 1c⇑). Again, only pUS9 of HSV-1 strain 17 was detectable in Western blot (Supplementary Fig. S1, available in JGV Online) and indirect immunofluorescence analyses (Supplementary Fig. S2, available in JGV Online), whereas no reaction was obtained with the US9 constructs of HSV-1 strain KOS despite similar transfection efficiencies (Supplementary Fig. S2).

To exclude the possibility that our polyclonal anti-US9 serum lacks antibodies against the N-terminal part of pUS9, which is preserved in the deduced gene product of HSV-1 strain KOS (Fig. 2b⇑), cells were transfected with an expression plasmid encoding pUS9 aa 1–58 fused to EGFP. This fusion protein was stably expressed and detected by the anti-US9 serum in Western blot and immunofluorescence analyses (Supplementary Fig. S1). Thus, our findings indicate that stable expression of the mutated pUS9 of HSV-1 strain KOS is impaired. The protein might be either rapidly degraded or eliminated from the cells.

Comparable amounts of the US8A proteins of HSV-1 strains 17 and KOS (pUS8A') were expressed in infected cells as shown by Western blot (Fig. 3b⇑, middle panel), and immunofluorescence reactions (Fig. 3c⇑, right panels) of the US8A-specific antiserum. In contrast to earlier data (Georgopoulou et al., 1995), pUS8A was found in the cytoplasm and no nucleolar localization was observed. As expected, the sizes of the US8A proteins differed and were in agreement with the calculated molecular masses 20.2 or 16.8 kDa for strains KOS and 17, respectively. In both cases, at least two protein bands were detected (Fig. 3b⇑, middle panel), which might be due to differential phosphorylation (Georgopoulou et al., 1995).

In summary, all tested cell culture samples and BAC clones of HSV-1 strain KOS contain mutations affecting the US8A and US9 genes. Both genes also exhibit some variation in other HSV-1 strains. Compared to strain 17, the deduced US8A proteins of the clinical isolate H129 and of strain HF10 exhibit one and four amino acid substitutions, respectively (McGeoch et al., 1985; Szpara et al., 2010; Ushijima et al., 2007). Likewise, the US9 proteins of HSV-1 strains 17 and HF10, and of isolate H129 differ at one or two positions.

However, two of the mutations in HSV-1 strain KOS are located at nucleotide positions crucial for the expression or size and overall structure of the US8A and US9 gene products. A functional interdependence of these changes appears conceivable, since the two structurally related type II membrane proteins might contribute to related functions during virus replication. Like the preceding glycoprotein gene cluster (McGeoch, 1990), US8A and US9 may have evolved from a common ancestor, although no sequence homologies are detectable (Davison, 2010), Furthermore, computer predictions (e.g. TMpred; ) suggest opposite membrane topologies for pUS8A and pUS9 of HSV-1 (Fig. 2b⇑). Our antibodies might help to determine localization and orientation of pUS8A.

Although HSV-1 strain KOS is actually a phenotypic US9 null mutant, it replicates efficiently in cell culture and is neurovirulent in mice (Dix et al., 1983). The described attenuation of certain KOS isolates has been mapped to genome regions other than the US9 gene (Thompson et al., 1986). This might indicate a limited functional relevance of pUS9, which would be in line with the inconspicuous phenotypes of engineered US9 deletion mutants of HSV-1 and other alphaherpesviruses in vitro (Brideau et al., 2000; Chowdhury et al., 2002; Polcicova et al., 2005), and partly also in vivo (Klopfleisch et al., 2006; McGraw et al., 2009; Nishiyama et al., 1993).

Parallel investigations of different HSV-1 strains including KOS, targeted US9 and US8A mutants as well as rescuants in rodents and cultured neurons should finally elucidate the function of the US8 to US9 gene region in neurotropism and neurovirulence of alphaherpesviruses.