Communication

During the herpesvirus replication cycle, viral transcription, DNA replication, formation of capsids and DNA packaging occur in the nucleus. The subsequent nuclear egress of newly synthesized nucleocapsids is performed by budding of the inner leaflet of the nuclear membrane, which creates the primary envelope. Although products of two genes conserved throughout the Herpesviridae family (HSV-1 UL34 and UL31) have previously been shown to be involved in the execution of this process, the molecular basis of their activity is not clear. Here we present results of protein structure prediction for the conserved domain of UL34. The applied methodology suggests that this protein adopts a pleckstrin homology (PH) fold to perform its function. A detailed inspection of the ligand binding site strongly supports the hypothesis that UL34 orthologs can recognize phosphoinositides. Since previous works suggest that alterations of UL34 gene product result in a drastic impairment of primary envelopment of HSV-1 and trapping of capsids in the nucleus, the presented data may lead to the development of novel anti-herpetic therapeutic strategies where analogs of phosphoinositides are administered.


INTroDUcTIoN
The herpesviruses are important infectious agents of various acute conditions, including severe encephalitis induced by HSV-1, congenital disorders (CMV) and neoplasia (EBV,HHV8).With eight different viruses infecting human, the extremely high seroprevalence and the ability to induce latent infections -Herpesviridae remains a challenging pathogen for both basic and applied research.
During the herpesvirus replication cycle, viral transcription, DNA replication, formation of nucleocapsids and DNA packaging is performed in the cell nucleus.To form the mature virions the nucleocapsids directs to cytoplasm by budding of the inner leaflet of the nuclear membrane, which creates the primary viral envelope (Mettenleiter et al., 2006).
Products of two viral genes homologous to HSV-1 UL34 and UL31 -has been previously identified as critical for the execution of this process (Mettenleiter, 2002).
The UL34 gene encodes a C-terminally anchored membrane protein (pUL34) located in the nuclear membrane leaving the majority of the protein exposed on the nuclear side (Reynolds et al., 2001).The product of UL31 (pUL31) is a phosphoprotein targeted to the nuclear membrane in the presence of pUL34 (Reynolds et al., 2002).Both proteins were shown to interact in both GST pull down and two-hybrid yeast system (Fuchs et al., 2002) and mutations of the UL34 gene product resulted in a decreased primary envelopment of HSV-1 with trapping of newly synthesized nucleocapsids in the nucleus (Mettenleiter 2002).
L. S. Wyrwicz and others Previously reported studies of point and deletion mutants of pUL34 positioned the region involved in a direct interaction with pUL31 as spanning a 45 amino acid interval located between codons 137 and 181 of the HSV-1 UL34 open reading frame (Liang & Baines, 2005).Further characterization of pUL34 by Bjerke et al. (2003) by means of non-reconstituting pUL34 mutants in a UL34-deficient HSV-1 model strain, defined six short patches of charged amino acids essential for the activity of this viral protein.
Since the both mentioned proteins involved in the execution of viral envelopment possess no detectable homology to any proteins of known function, the overall picture of their molecular activity remains unclear.Ogg et al. (2004) hypothesized that there are at least three non-mutually exclusive ways by which pUL34 could facilitate nuclear egress.The authors pointed out the possibility that pUL34 might either directly mediate the envelopment, through bridging interactions between the nucleocapsid and the inner nuclear membrane, or, alternatively, mediate the traffic of other important nuclear egress factors.Finally, they postulated that pUL34 might affect the architecture of the nuclear lamina (lining the interior face of the nuclear envelope) in order to allow nucleocapsids access to the envelopment machinery (Scott & O'Hare, 2001).

MATerIALS AND MeTHoDS
The identification of UL34 orthologs and assembly of UL34 family.The conserved domain of HSV-1 pUL34 was retrieved from the PfamA database (Finn et al., 2006) entry under the accession code PF03582.It was subsequently scanned for distant homologues in form of other PfamA entries using MetaBasic (Ginalski et al., 2004).The sequences of HSV-1 UL34 (PF03582; gi|9629415) and its orthologs -CMV UL50 (PfamA entry: PF04541; GenBank: gi|157780067) and EBV BFRF1 (PF05900; gi|82503207) -were retrieved from National Cent-er for Biotechnology Information protein database.A set of distant homologues was obtained by PSI-BLAST searches against the NR (non-redundant) NCBI database (as accessed on 01/12/2007) (Altschul et al., 1997).The search was executed until profile convergence, with the E-value cutoff for inclusion in the profile set equal to 0.001.The obtained collection of homologues was merged into a single set and clustered at 90% of sequence identity using the CD-HIT tool to remove redundancy (Li & Godzik, 2006).The resulting sequences were aligned with ClustalW (Thompson et al., 1994) with subsequent minor manual corrections.Potential active site residues were annotated by an overlap of two datasets consisting of: (a) data on point mutants which did not reconstitute native activity of pUL34 in the UL34-null HSV-1 model (Bjerke et al., 2003), (b) residues conserved throughout the assembled alignment of pUL34 orthologs from three subfamilies of Herpesviridae (compare Fig. 1).
Fold recognition of the UL34 protein.
The annotation of potential globular regions was performed with GlobPlot (Linding et al., 2003).The sequences of the conserved globular domain of pUL34 from three subfamilies of Herpesviridae were subjected to the Structure Prediction Meta Server (http://bioinfo.pl/meta) (Bujnicki et al., 2001).Secondary structure prediction was performed with PsiPred (McGuffin et al., 2000) and ProfSec (Rost & Sander, 1993) via Meta Server.Collected models were sorted according to the results of a consensus fold recognition prediction method -3D-Jury (Ginalski et al., 2003).Since no predictions achieved a reliable score in 3D-Jury, manual inspection of conserved active site residues within query-hits alignments collected from MetaServer results was performed.

reSULTS AND DIScUSSIoN
The MetaBasic distant homology mapping performed for HSV-1 UL34 confirmed the existence of HSV-1 UL34 orthologs throughout the whole family of Herpesviridae.These included the previously reported sequences of CMV UL50 of Betaherpesvirinae deposited in the PfamA database entry PF04541 and EBV BFRF1 of Gammaherpesvirinae (PfamA: PF05900; Table 1).Multiple iterations of PSI-BLAST searches (see Methods) resulted in a set of 52 homologous sequences.Filtering at 90% identity resulted in a reduced, non-reduntant set of 39 sequences (18, 8 and 13 sequences from Alpha-, Beta-and Gammaherpesvirinae, respectively).
The above-mentioned sequences of conserved globular domains of pUL34 homologues (representing all subfamilies of Herpesviridae) were submitted to the Protein Structure Prediction MetaServer and the 3D-Jury prediction assessment system.Since no reliable predictions were identified by the standard procedure of protein homology modeling (3D-Jury score >50 (von Grotthuss et al., 2003;Ginalski et al., 2005)), additional screening of the potential active site was performed (Koczyk et al., 2007).
To increase the confidence of predicting the likely structure of UL34 homologues, we identified invariant residues within the alignment of the three subfamilies from Fig. 1 and compounded them with the positions of six non-complementing chargedcluster mutants (reported by Bjerke et al., 2003).Of the about 150 residues forming a potential globular domain (according to a compilation of GlobPlot and secondary structure prediction methods, as reported previously (Wyrwicz et al., 2007;Wyrwicz & Rychlewski, 2007a)) only three residues were preserved in all analyzed sequences.These included: the glutamate E67, aspargine N89 and glycine G91 of HSV-1 pUL34 (GenBank gi|9629415), which corresponded to CMV (Betaherpesvirinae; GenBank gi|157780067) residues 56, 76 and 78, and EBV (Gammaherpesvirinae; GenBank gi|82503207) residues 59, 84 and 86, respectively.One of these three residues (E67) was localized in the region of a non-complementing mutation -CL06 of the Bjerke et al. (2003) study.The two remaining conserved amino acids (N89, G91) were not assessed in the mentioned report.
Among the collection of predictions obtained with the Protein Structure Prediction MetaServer, we observed a single sequence matching all three critical residues described above -the pleckstrin homology (PH) domain of protein kinase B (Akt; Protein Data Bank PDB accession: 1h10, chain A), as predicted by the 3D-PSSM method (Kelley et al., 2000) for CMV pUL34 ortholog.Further circumstantial evidence was provided by the fact that three out of ten topscoring predictions by this method also exhibited a consistent PH domain-like fold.Also, this fold was predicted for CMV protein by INUB (Fischer, 2003), for HSV-1 pUL34 by FFAS3 (Jaroszewski et al., 2005), and for EBV by 3D-PSSM (the top-scoring hit).
Pleckstrin homology (PH) domains are small, 100-residue long domains which occur in a large variety of proteins associated with intracellular signaling or the cytoskeleton (Haslam et al., 1993;Saraste & Hyvonen, 1995).The domain is involved in a wide range of biological processes, including binding to the γ and β subunits of heterotrimeric G proteins (Inglese et al., 1995), binding to lipids (Yan et al., 2005), attachment to membranes (Cozier et al., 2004), and cytoskeleton organization (Lemmon et al., 2002).Although the binding properties of various PH domains are highly divergent, crystallographic studies of several PH domain proteins revealed a common core structure of two perpendicular anti-parallel β sheets, followed by a C-terminal amphipathic helix (Yan et al., 2005).With an extremely low sequence conservation throughout the family, no totally invariant residues seem to be critical for retention of the fold, and the length of the loops connecting the β-strands is highly variable.Thus, PH domains are relatively difficult to detect in the sequence (Saraste & Hyvonen, 1995).This fact is augmented by the presence of split PH domains, resulting from insertions of other structurally and functionally distinct domains (Yan et al., 2005).
The initial identification of the fold was followed by manual curation of the sequence-to-structure alignment according to both the predicted (PSIPRED (McGuffin et al., 2000)) and observed (PDB 1h10A) secondary structure elements (Fig. 1).Although a low sequence identity between the Akt kinase PH domain and the viral proteins was observed (ranging between 10-16%, depending on virus subfamily), several additional tests supported this fold assignment.In particular, highly consistent predictions of secondary structure elements were observed.After refining the alignment with 3D-Jury (Ginalski et al., 2003), we proceeded to map the strands creating the perpendicular β-sheets (Fig. 1).Also, the amphipatic helix was present in all pUL34 orthologs.Notably, the local sequence conservation within the subfamilies of herpesvirus (Alpha-, Betaand Gammaherpesviridae) seems strongly reduced outside the region mapped onto the PH fold (not shown).
Preservation of the physicochemical properties of aligned residues within the conserved secondary structure elements can be noticed, especially within the following β-strands: beta-2 (β2), beta-4 (β4) and beta-6 (β6).The most conserved residues constitute a potential binding site which resembles a known phosphoinositide-binding module (Thomas et al., 2002).Four clusters of residues involved in the contact with the inositol 1,3,4,5-tetrakisphosphate molecule create a corresponding phosphoinositidebinding site in the already determined structure of Akt kinase (PDB entry 1h10A).These clusters are located in the loop between strands β1 and β2, as well as within strands β2, β4 and β7.Of these, we observed partial preservation of the first site, with two residues totally conserved: the E67 residue and the asparagine (N) residue located proximally at β4 strand.Additionally, in Alphaherpesvirinae, the central arginine (R) of β7 strand, involved in binding of phosphoinositides, is also preserved.Again, according to the alignment, the most divergent residues are observed in the proximal part of β2 strand (where in the native structure two arginines stabilize the phosphate at position 3 of the inositol ring).
Apparently -as reported recently (Carpten et al., 2007) -a mutation of the glutamic acid in AKT1 (E17K), corresponding to the critical E67 in HSV-1 pUL34, is a common somatic mutation in breast, colorectal and ovarian cancers.The mutation was identified as being responsible for altered membrane localization and aberrant activation of the related kinase domain.
The phosphorylation of HSV-1 UL34 by viral kinase US3 was previously indicated as a major mechanism regulating its activity (Klupp et al., 2001;Ryckman & Roller, 2004).Such phosphorylation site occurs outside of the conserved domain (C-terminally, not shown in the alignment).This regulation of activity of PH domains by post-translational modifications, especially phosphorylation, has previously been reported (Nawaratne et al., 2006), and the PH domain represents one of top five phosphorylated domains in the PhosphoELM resources, with 17 phosphorylation sites within various proteins (Diella et al., 2007).
In conclusion of our study, the described bioinformatics analysis strongly supports the hypothesis that nuclear egress factor pUL34 and its orthologs from all herpesviruses retain the fold of a pleckstrin homology domain (PH fold).To our knowledge this fold has not been previously observed in any viral protein.Although the PH domain realizes a wide range of functions, its general property focuses on binding of either small ligands or proteins.Thus it mainly acts as a regulatory interface for protein-protein or protein-membrane interactions of modular proteins.With such a low sequence conservation (10-16%), we are unable to provide an undisputable evidence for the postulated recognition of phosphoinositols using bioinformatics and in this respect additional biochemical assays are needed.Nevertheless, the suggested mechanism is supported by previously published data.The disruption of one of the most conserved (essential) amino acids in pUL34 results in alteration of both the nuclear egress of virions and the correct internal nuclear membrane (INM) localization (CL06 mutant in Bjerke et al. (2003) study).The fact that a similar phenotype is only observed when the C-terminal transmembrane-spanning region is deleted from native pUL34 further corroborates our hypothesis.Therefore we venture to speculate that the binding module of the PH-fold domain of pUL34 is indeed essential for interaction with INM, in correspondence with the first of the mentioned hypotheses put forth by Ogg et al.L. S. Wyrwicz and others (2004) (i.e., direct mediation of envelopment through bridging interactions between the nucleocapsid and the inner nuclear membrane).
Other corroborating evidence comes from the observation that the commonly found sorting of modular proteins to membranes, mediated by PH domains, is guided by recognition of phosphoinositides (Cozier et al., 2004;Takenawa & Itoh, 2006).Additionally, certain other phosphoinositide-binding domains can promote bilayer curvature by inserting into the membrane upon binding phosphoinositides (reviewed by Lemmon, 2003), in a process that resembles the budding of the viral primary envelopment pictured in electron microscopy (Mettenleiter, 2002).
Finally, the pUL34 molecule can preferably interact with nuclear membranes of cells, and the herpesvirus primary envelope derived from the nuclear membrane does differ from the secondary envelope obtained in the cytoplasm (Mettenleiter, 2004;Mettenleiter et al., 2006) (in terms of phospholipid content (Dygas & Baranska, 2001)).This seems likely to result in the lack of pUL34 localization to other membranes than INM.
Although the proof of the direct mechanism used by pUL34 to perform its action is lacking, our finding is supported by the above-mentioned previous studies and is highly important in terms of possible clinical applications.Since the previous works suggested that alterations in the function of pUL34 result in a drastic impairment of primary envelopment of HSV-1 and trapping of nucleocapsids in the nucleus (Mettenleiter, 2002), the presented data may lead to the development of novel anti-herpetic therapeutic strategies -ones where analogs of phosphoinositides are applied (Powis et al., 1995;Weglarz et al., 2006) similarly to the proposed targeting of mutated Akt kinases (Carpten et al., 2007).