Bioinformatics Advance Access originally published online on October 6, 2005
Bioinformatics 2005 21(23):4216-4222; doi:10.1093/bioinformatics/bti706
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Discovering hidden viral piracy
1Compugen Ltd Tel Aviv 69512, Israel
2Faculty of Life Sciences, Bar-Ilan University Ramat Gan 52900, Israel
*To whom correspondence should be addressed.
 |
ABSTRACT |
|---|
 TOP ABSTRACT 1 INTRODUCTION2 METHODS3 RESULTS4 DISCUSSIONREFERENCES |
|---|
Motivation: Viruses and developers of anti-inflammatory therapies share a common interest in proteins that manipulate the immune response. Large double-stranded DNA viruses acquire host proteins to evade host defense mechanisms. Hence, viral pirated proteins may have a therapeutic potential. Although dozens of viral piracy events have already been identified, we hypothesized that sequence divergence impedes the discovery of many others.
Results: We developed a method to assess the number of viral/human homologs and discovered that at least 917 highly diverged homologs are hidden in low-similarity alignment hits that are usually ignored. However, these low-similarity homologs are masked by many false alignment hits. We therefore applied a filtering method to increase the proportion of viral/human homologous proteins. The homologous proteins we found may facilitate functional annotation of viral and human proteins. Furthermore, some of these proteins play a key role in immune modulation and are therefore therapeutic protein candidates.
Contact: kliger{at}compugen.co.il
|
1 INTRODUCTION |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTS4 DISCUSSIONREFERENCES |
|---|
Viruses must develop means to evade the host immune system. One evasion strategy is the modulation of the host immune response by encoding different immunomodulatory proteins. Often, one can find sequence similarity between these viral immunomodulatory proteins and host proteins, suggesting that these viral proteins were ‘pirated’ from the host. This phenomenon, named viral piracy, is a strategy predominantly taken by large dsDNA viruses (Seet et al., 2003). Members of the Herpesviridae and the Poxviridae families are known to encode proteins that modulate the host immune response, including homologs of cytokines, interferon regulatory factors, chemokines and their receptors.
Epstein–Barr virus (EBV) (Moore et al., 1990) and Human Cytomegalovirus (HCMV) (Kotenko et al., 2000) harbor their own homolog of IL-10—a powerful immunosuppressor that downregulates the synthesis of pro-inflammatory cytokines and prevents macrophages from acting as antigen presenting cells (Bogdan et al., 1991; D'Andrea et al., 1993; de Waal Malefyt et al., 1991). EBV IL-10 has apparently retained the immune-inhibitory properties associated with the cellular ligand, but not the immune-stimulatory features associated with Human IL-10 (reviewed in Lucas and McFadden, 2004). Kaposi's sarcoma-associated herpesvirus (KSHV) encodes a homolog of IL-6, a multi-functional cytokine with a wide range of activities. Moreover, while normally cells must express both the IL-6 receptor and the gp130 co-receptor to respond to IL-6, the viral IL-6 can induce signaling through gp130 alone and thus stimulate peripheral blood B cells that are unresponsive to the human IL-6 (Breen et al., 2001; Mullberg et al., 2000). Hence, not only do viruses encode homologs of immune-related proteins, often these homologs have evolved and gained new capabilities not present in the original host protein.
Myxoma virus, a member of the Poxviridae family, takes a different path to suppress the immune response. It encodes a protein homologous to the human TNF receptor. This viral protein lacks the transmembrane domain, and is therefore a soluble receptor, which acts as an inhibitor of the human TNF signaling pathway (Upton et al., 1991). In addition, Myxoma virus encodes several serine protease inhibitors, named Serp-1, Serp-2 and Serp-3, that exert anti-inflammatory as well as anti-apoptotic activities (Guerin et al., 2001; Macen et al., 1993; Petit et al., 1996). The Serp-1 protein is currently in clinical trials for the treatment of various inflammatory-related diseases.
Host IL-1 receptor binds IL-1
, IL-1ß and the natural competitor IL-1 receptor antagonist. Vaccinia virus, another member of the Poxviridae family, encodes an IL-1 receptor homolog that can only bind IL-1ß. Deletion experiments reveal that the Vaccinia virus-encoded IL-1 receptor homolog diminishes the systemic response to infection. This illustrates how the study of the mechanisms, devised by viruses to modulate the host defenses, promotes the understanding of the different roles of IL-1 and IL-1ß (Alcami and Smith, 1992).
In addition, Vaccinia virus encodes the Vaccinia virus complement-control protein (VCP), a soluble protein that shares homology with C4-binding protein. VCP inhibits both the classical and the alternative complement pathways and contributes to virulence (Isaacs et al., 1992). Variola virus, the most virulent member of the Orthopoxvirus genus, encodes a complement-control protein termed SPICE, which is significantly more potent than VCP at inhibiting the formation of C3/C5 convertases necessary for complement-mediated viral clearance (Rosengard et al., 2002).
These examples illustrate that viruses pirate host proteins that modulate the effect of both the adaptive and innate immune systems. The importance of viral pirated proteins as immune modulators is emphasized by the finding that a TNF receptor homolog, and homologs of chemokines were found in primary virus isolates, but not in laboratory-adopted strains (Benedict et al., 1999; Cha et al., 1996). This finding suggests that these viral proteins are not essential for viral replication, but are important for the survival of the virus within the host.
Computational efforts have been successful in revealing previously unknown viral/human homologs (Holzerlandt et al., 2002). Yet, we hypothesized that many viral/human homologs have not yet been identified, due to low sequence similarity. For an alignment hit, the expectation value (E-value) reflects the expected number of random alignments having equal or better scores. Hence, the E-value is a conventional way for estimating the authenticity of an alignment hit. In general, homologous proteins gain significant E-values, while false hits gain low-significance E-values. Our primary objective is to test whether diverged viral/human homologous proteins exist in very-weak alignment hits that are usually ignored. However, weak alignment hits often have low-significance E-values, making it difficult to distinguish homologous proteins from false hits. Hence, we developed a method that estimates the number of homologous proteins in a group of alignment hits. This analysis confirmed that homologous proteins exist even in very-low-similarity alignment hits. Obviously, these homologs are masked by many false hits. Thus, we developed a scoring function that can be used to filter out many of the false hits. The detection of novel viral/human homologs facilitates functional annotation of viral and human proteins and may provide therapeutic protein candidates.
|
2 METHODS |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTS4 DISCUSSIONREFERENCES |
|---|
2.1 Data preparation
Human RefSeq NPs were retrieved from the RefSeq ftp site on May 5, 2004 (21 196 sequences) (Pruitt and Maglott, 2001). Sequences of viral proteins belonging to the Herpesviridae and the Poxviridae families were downloaded from the NCBI server on June 27, 2004 (35 893 sequences) and were subject to a non-redundant algorithm (Holm and Sander, 1998), with a cutoff of 90% (resulting in 8185 sequences). Mammalian protein sequences were retrieved from SwissProt (Boeckmann et al., 2003) on June 27, 2004 (30 399 sequences).
2.2 Signal peptide prediction
Predicting whether a protein possesses a signal peptide was performed using the neural-network method of the SignalP 3.0 prediction tool with default parameters (Bendtsen et al., 2004).
2.3 Identifying sequence similarity between viral and human proteins
Viral proteins were queried against the human RefSeq NPs database with Blastp using default parameters (Altschul et al., 1990), except for the substitution matrix, where we used the BLOSUM45 matrix in order to enhance detection of distant homologs. All signal sequences of the human proteins were removed prior to the Blastp run, so the alignments are not dependent on the signal peptide sequences. Alignment hits with length of less than 30 amino acid residues were discarded.
2.4 Calculating evolutionary rate per residue of human proteins
Human RefSeq proteins were searched for mammalian homologs with Blastp using default parameters. For each human protein, alignment hits having Blastp E-value < 0.001 were selected (at least 5 sequences, maximum 50). Each group, consisting of a human protein and its mammalian homologs, was subject to multiple alignment using ClustalW (Thompson et al., 1994). The multiple alignment was used as input to the Rate4Site empirical Bayesian method (Mayrose et al., 2004), using default parameters, but with no branch length optimization to reduce computing time. This method calculates the evolutionary rate for each of the residues of a protein, based on the multiple alignment. The use of multiple alignments may introduce artifacts related to the non-equal representation of proteins in databases (del Sol Mesa et al., 2003). The Rate4site method addresses this difficulty by the construction of an evolutionary tree. The use of an evolutionary tree diminishes the effect of redundant sequences by weighting clusters of closely related sequences differently (Pupko et al., 2002).
|
3 RESULTS |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTS4 DISCUSSIONREFERENCES |
|---|
We hypothesized that high sequence divergence impedes the discovery of many viral pirated proteins, because these are hidden in low sequence similarity alignments, where they are masked by many false hits. In order to examine this hypothesis, we developed a method to estimate the amount of viral/human homologs in a set of alignment hits.
3.1 A method to assess the amount of homologous proteins in a set of alignment hits
Our primary goal was to estimate the number of homologous proteins in a set of alignment hits. To this end, a protein feature, which is expected to be conserved among homologous proteins, was needed. Ideally such a protein feature should be easily detected and encompass 50% of the proteins to minimize the fraction of random alignment hits in which both proteins either have or lack the feature. The signal sequence is an abundant protein feature, encompassing 
20% of proteins, which can be reliably detected using the SignalP tool (Bendtsen et al., 2004; Zhang and Henzel, 2004). Thus, we tested whether the presence of a signal sequence is a feature conserved among homologous proteins.
Viral proteins were queried against human proteins with Blastp. For each viral protein, only the best alignment hit was analyzed. We divided the alignment hits into five partitions according to their E-values. The fraction of the alignment hits in which both the human and the viral proteins either have or lack a signal sequence was calculated (Fig. 1). For comparison, the expected fraction of random alignment hits, in which both the human and viral proteins either have or lack a signal sequence, is plotted as well (Fig. 1, ‘Random’ column). As expected, the stronger the sequence similarity (low E-values), the higher the chance that both proteins will either have or lack a signal sequence. The results are statistically significant for all partitions where E-value < 5 (P < 1e–4, 
2 test). Thus, the signal sequence is, in general, a conserved feature of homologous proteins. Therefore, we used this finding to estimate the amount of homologous proteins.
|
Alignment hits could be classified into one of four subgroups according to the presence (or absence) of a signal sequence in the human and viral proteins. Alignment hits, where both human and viral proteins have (or lack) a signal sequence, correspond to the YY (or NN) subgroup. Alignment hits, where the human protein has (or lacks) a signal sequence but the viral protein lacks (or has) one, correspond to the YN (or NY) subgroup.
A set of alignment hits is comprised of homologous proteins and random hits. It is possible to calculate the theoretical maximum number of random hits in a given alignment hits set from the observed distribution of the alignment hits in the four subgroups (YY, YN, NY and NN) and the expected distribution of random alignment hits. In order to calculate the distribution of random alignment hits in the four subgroups, we generated a random alignment hits set comprising viral/human insignificant alignment hits (E-value > 5). As expected, the null hypothesis, that the observed distribution of the insignificant alignment hits in the four subgroups is similar to the one that would be expected if all hits were random ones, could not be rejected (P-value = 0.99, 2 test). Since this group of random alignment hits is large (590 alignment hits), the distribution of any set of random alignment hits in the four subgroups should be similar to the observed one.
Hence, in a given set of alignment hits, we defined the subset of random alignment hits as the largest that could still maintain the distribution we calculated for random data. This way, we can calculate the number of homologous proteins (Table 1, ‘entire groups’). The described method will always result in an underestimation of the number of homologous proteins, because we consider the theoretical maximum number of random hits.
|
3.2 Filtering method
We examined alignment hits within and below the twilight zone (10% < Identity < 35%) (Blake and Cohen, 2001; Rost, 1999), in order to discover unknown distant homologs. Of course, in this range, many false alignment hits mask the homologous proteins. In order to enrich the proportion of viral/human homologs, we looked for a feature that differentiates between homologs and false alignment hits. Next, the quality of this feature was determined according to its ability to enrich the data with homologous proteins, estimated as described in Section 3.1.
3.3 Relating evolutionary rate per residue with viral piracy
The feature examined was the tendency of viruses to conserve the same amino acid residues that are conserved among mammalian orthologs and paralogs. Viral proteins are subject to major sequence changes, due to their constant need to evade the host immune system. Nevertheless, in order for a viral pirated protein to maintain its function, amino acid residues that are crucial for the structure or function of the protein are expected to be conserved. We assigned a mammalian evolutionary rate for all residues of the human proteins, for all viral/human alignment hits sharing significant sequence similarity (E-value < 1e–5). For each of the 20 natural amino acid residues, separately, we gathered the evolutionary rates at all positions, where the amino acid residue is identical in the viral and human proteins. Similarly, we gathered the evolutionary rates at all positions, where this amino acid residue is not identical in the viral and human proteins. The two lists of evolutionary rates—for the identical and not identical positions—were compared to determine whether viral pirated proteins tend to conserve the same residues that are conserved among mammals. The same was performed for each of the 20 amino acid residues.
We expect that residues that are not important will often gain high evolutionary rates, while important residues will often gain low evolutionary rates. Indeed, we found that residues that are identical in the viral/human homologs have lower mammalian evolutionary rates than residues that are not identical. This is true for each of the 20 natural amino acids. In order to estimate the statistical significance of the results, the null hypothesis that the two lists of mammalian evolutionary rates—for the identical and non-identical positions—are of the same distribution was rejected (P-value < 1e–7 for each of the 20 natural amino acids, Mann–Whitney test).
This procedure was performed on alignment hits with E-value < 1e–5, where most hits are of viral/human homologous proteins. For comparison, the same procedure was performed on insignificant alignment hits. The results revealed that the difference, between the mammalian evolutionary rates of residues that are identical in the viral and human aligned proteins and the residues that are not, was insignificant.
The identification of the important residues in proteins depends on the quality of the pairwise and multiple alignments (Casari et al., 1995). However, there is no reason to suspect that alignment errors will bias the results in a specific manner. Indeed, our results indicate that, in general, the accuracy of the alignments is sufficient for this procedure.
In conclusion, the tendency of viral proteins to maintain the residues that are conserved among mammals is a feature that distinguishes between homologous proteins and false alignment hits. Thus, we used this feature to enrich the proportion of homologous proteins.
3.4 Calculating a ‘conservation tendency score’ for each viral/human alignment hit
For each viral/human alignment hit the mammalian evolutionary rate scores, of all the residues of the human proteins, were computed by the Rate4Site program. We gathered all the evolutionary rate scores of residues that were identical between the human and the viral proteins. Similarly, we gathered all the evolutionary rate scores of residues that were not identical in the alignment. In order to test the hypothesis that these two score groups are similar, a Mann–Whitney test was applied. Next, we assigned the score of the corresponding Mann–Whitney test [conservation tendency score (CTS)] to each of the alignment hits.
In order to confirm that this method enriches the proportion of alignment hits comprising viral/human homologous proteins, we divided the list of alignment hits into two equally sized sub-partitions according to their CTSs. The enriched sub-partition comprised the alignment hits with better CTSs, and the depleted sub-partition comprised the alignment hits with worse CTSs. We then divided each of the sub-partitions to the four subgroups (YY, YN, NY and NN) and calculated the fraction of homologs, as well as the chance that the distribution is the same as would be expected for random alignment hits. The results revealed that for significant alignment hits (>35% sequence identity) the enriched sub-partition has a higher fraction of homologs than the depleted sub-partition (Table 1, Fig. 2A).
|
Even though most of these alignment hits are easy to detect by sequence alignment tools, some alignment hits are not trivial to explain. One such alignment hit is a VEGF homolog encoded by Bovine popular stomatitis virus (GI:41057442, E-value = 2e–25, Identity = 54%, CTS = –3.86), a member of the Poxviridae family. Human VEGF is a well-studied protein that participates in angiogenesis, induces endothelial cell proliferation, promotes cell migration and is a vascular permeability factor (Ferrara, 2001). However, it was not obvious what advantage the virus achieves by encoding a VEGF homolog. A recent publication suggests that human VEGF has also a regulatory effect on T cells and that human VEGF could promote the Th1-Th2 shift (Lee et al., 2004). This finding suggests that by encoding the VEGF homolog, Bovine popular stomatitis virus promotes the Th1-Th2 shift, thereby inhibiting the anti-viral cellular immune response.
Next, we examined weak alignment hits—with Identity < 35%—and found that a significant amount of homologs exists in this group as well. We then confirmed that the CTS-based method is able to decrease the fraction of false alignment hits in this set of alignments, thereby enriching the fraction of homologous proteins (Table 1, Fig. 2B).
The aim of this study is to find novel viral piracy events, therefore we focused on weak alignment hits having sequence identity <35% and E-value > 1e–5. Remarkably, the results revealed that 29% of these alignment hits, which are usually ignored, are of homologous proteins (Table 2, ‘entire group’). Furthermore, Figure 3 reveals that the CTS-based scoring function could also differentiate between weak hits of homologous proteins and false ones. Therefore, we applied the CTS-based scoring function and focused on the 100 alignment hits having the best CTS, where we estimated that 64% are of homologous proteins (Table 2).
|
|
|
4 DISCUSSION |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTS4 DISCUSSIONREFERENCES |
|---|
Few viral proteins, which were pirated from their mammalian host, are currently in clinical and pre-clinical studies for the treatment of various diseases associated with excessive inflammatory or immune responses (reviewed in Lucas and McFadden, 2004). Most annotated viral/human homologs, such as homologs of soluble TNF receptor, IL-6 and others share high sequence similarity. In order to find novel viral/human homologs, we examined low-to-moderate sequence similarity alignment hits having 10–35% identity and E-value > 1e–5. Our results indicated that
29% of the alignment hits are of homologous proteins (Table 2). We assigned each of the alignment hits a CTS, and examined the 100 proteins having the best CTS among this group. The results revealed that 64% of these alignment hits are of homologous proteins. Of special interest are the secreted viral proteins, as they may be used as therapeutics. Thus, we focused on the YY subgroup of the top 100 proteins of the 10–35% identity and E-value > 1e–5 alignment hits set. The tendency of homologous proteins to maintain a functional signal sequence is the feature that was used to assess the amount of homologous proteins in a list of alignment hits (Fig. 1). As this feature distinguishes between homologs and random hits, focusing on the YY subgroup results in a higher fraction of homologous proteins than the one calculated. The following are several examples of secreted and membranal viral/human homologs that have been detected.
Human IL-6 (NP_000591 [GenBank] ) shares very low sequence similarity (E-value = 1.4; Identity = 20%) with Cercopithecine herpesvirus 17 R2 protein (GI:18653819). This alignment hit has a good CTS score of –3.36, supporting the reports, which were based on genomic location and followed by functional analysis, that this viral protein is an IL-6 homolog (Desrosiers et al., 1997; Kaleeba et al., 1999).
Human IL-10 (NP_000563 [GenBank] ) shares very high sequence similarity (E-value = 3e–53; Identity = 92%) with Human EBV IL-10 homolog (GI:114886). In addition, human IL-10 shares moderate sequence similarity (E-value = 7e–9; Identity = 27%) with Human herpesvirus 5 UL111A (GI:39841977). Support for the authenticity of the latter alignment hit comes from the CTS scoring method, which gives a score of –2.54. Indeed, this moderate sequence similarity alignment, which was not detected using a computational analysis that used a PSSM representing herpesvirus protein motifs (Holzerlandt et al., 2002), was shown to bind to the human IL-10 receptor and compete with human IL-10 for the receptor binding sites (Kotenko et al., 2000).
Human interferon gamma receptor 1 (NP_000407 [GenBank] ) shares moderate sequence similarity (E-value = 1e–5; Identity = 20%) with lumpy skin disease virus LSDV008 putative soluble interferon gamma receptor (GI:22595541). This alignment hit has a good CTS score of –2.51. Our finding is consistent with the prediction by Kara et al. (2003) that LSDV008 is an interferon gamma receptor. Several members of the Poxviridae family encode a soluble IFN-gamma receptor that efficiently blocks the binding of IFN-gamma to cellular receptors, thus inhibiting the functions of IFN-gamma (reviewed in Alcami and Smith, 1996).
Human endothelial lipase precursor (NP_006024 [GenBank] ) shares low sequence similarity (E-value = 0.3; Identity = 24%) with Gallid herpesvirus 2 lipase (GI:10180702). This alignment hit has a good CTS score of –3.82. Despite the low sequence similarity, this viral protein retained several conserved lipase features (Tulman et al., 2000).
Human OX-2 membrane glycoprotein (CD200) (NP_005935 [GenBank] ) shares low sequence similarity with Yaba-like disease virus 141R protein (GI:12085124; E-value = 0.0002; Identity = 30%), Yaba-monkey tumor virus 141R protein (GI:38229299; E-value = 0.012; Identity = 28%) and Myxoma m141R protein (GI:4097179; E-value = 0.025; Identity = 33%), all are members of the Poxviridae. Despite the low sequence similarity, these alignment hits have good CTS scores of –3.39, –3.06 and –2.53, respectively. OX-2 homolog of Myxoma virus has been shown to have an immunomodulatory effect and to diminish the activation level of circulating T lymphocytes. Moreover, while this OX-2 homolog is necessary for the full development of a lethal infection in rabbits, it is not required for efficient virus replication in susceptible cell lines, in vitro (Cameron et al., 2005). Herpesviruses have also adopted OX-2 homologs. KSHV, a gamma-herpesvirus, encodes the K14 protein, an OX-2 homolog that shares 40% sequence similarity with the human protein. This OX-2 homolog binds to CD200 receptor with identical affinity and kinetics and inhibits TNF-alpha production in a similar manner to the human protein (Foster-Cuevas et al., 2004). Viruses of distinct families adopted OX-2, probably by independent events, indicating that it is a good point for the viruses to intervene in the regulation of the immune system. Thus, we suggest OX-2 as a target for drug intervention in inflammatory disorders. Callitrichine herpesvirus 3 C2 protein (GI:24943099) also shares low sequence similarity (E-value = 0.038; Identity = 22%) with Human OX-2 membrane glycoprotein. Even though the sequence similarity is weak, this alignment hit has a good CTS score of –1.89. Callitrichine herpesvirus 3 is a gamma-herpesvirus, a subfamily whose members were shown to encode OX-2 homologs, consistent with this finding.
Homologous proteins are of the same evolutionary origin. Various applications motivate the search for homologous proteins. An ideal homology detection method would identify all homologs for every target sequence in a test set, while also having no false predictions. However, some homologous proteins share only low sequence similarity and therefore are difficult to distinguish from false alignment hits. Herein, we presented a method to assess the amount of homologous proteins in a set of alignment hits. In addition, we presented a scoring function that can be used to enrich the proportion of homologous proteins in that set of alignment hits. This technique is applicable for the identification of homologous proteins sharing low-to-moderate sequence similarity. We implemented this technique to detect diverged homologs between human and large dsDNA viruses. Our results are enriched with viral/human homologous proteins, enabling manual reviewing aimed at identifying the alignment hits that are a result of events of viral piracy.
|
Acknowledgments |
|---|
We are thankful to T. Pupko and I. Mayrose for the Unix version of Rate4Site. We are also grateful to P. Akiva, I. Borukhov, E. Eisenberg, E. Gofer, M. Havilio, I. Hecht, D. Gerber, Y. Kinar, G. Kojekaro, G. Kol, T. Lapidot, E. Levanon, A. Novik, R. Sarid, R. Sorek, A. Toporik, N. Tsabar, L. Tsirulnikov and A. Wool for helpful suggestions and fruitful discussions.
Conflict of Interest: none declared.
Received on July 28, 2005; revised on September 15, 2005; accepted on October 5, 2005
|
REFERENCES |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTS4 DISCUSSIONREFERENCES |
|---|
Alcami, A. and Smith, G.L. (1992) A soluble receptor for interleukin-1 beta encoded by vaccinia virus: a novel mechanism of virus modulation of the host response to infection. Cell, 71, 153–167[CrossRef][ISI][Medline].
Alcami, A. and Smith, G.L. (1996) Soluble interferon-gamma receptors encoded by poxviruses. Comp. Immunol. Microbiol. Infect. Dis, . 19, 305–317[CrossRef][ISI][Medline].
Altschul, S.F., et al. (1990) Basic local alignment search tool. J. Mol. Biol, . 215, 403–410[CrossRef][ISI][Medline].
Bendtsen, J.D., et al. (2004) Improved prediction of signal peptides: SignalP 3.0. J. Mol. Biol, . 340, 783–795[CrossRef][ISI][Medline].
Benedict, C.A., et al. (1999) Cutting edge: a novel viral TNF receptor superfamily member in virulent strains of human cytomegalovirus. J. Immunol, . 162, 6967–6970
Blake, J.D. and Cohen, F.E. (2001) Pairwise sequence alignment below the twilight zone. J. Mol. Biol, . 307, 721–735[CrossRef][ISI][Medline].
Boeckmann, B., et al. (2003) The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003. Nucleic Acids Res, . 31, 365–370
Bogdan, C., et al. (1991) Macrophage deactivation by interleukin 10. J. Exp. Med, . 174, 1549–1555
Breen, E.C., et al. (2001) Viral interleukin 6 stimulates human peripheral blood B cells that are unresponsive to human interleukin 6. Cell Immunol, . 212, 118–125[CrossRef][ISI][Medline].
Cameron, C.M., et al. (2005) Myxoma virus M141R expresses a viral CD200 (vOX-2) that is responsible for down-regulation of macrophage and T-cell activation in vivo. J. Virol, . 79, 6052–6067
Casari, G., et al. (1995) A method to predict functional residues in proteins. Nat. Struct. Biol, . 2, 171–178[CrossRef][ISI][Medline].
Cha, T.A., et al. (1996) Human cytomegalovirus clinical isolates carry at least 19 genes not found in laboratory strains. J. Virol, . 70, 78–83[Abstract].
D'Andrea, A., et al. (1993) Interleukin 10 (IL-10) inhibits human lymphocyte interferon gamma-production by suppressing natural killer cell stimulatory factor/IL-12 synthesis in accessory cells. J. Exp. Med, . 178, 1041–1048
de Waal Malefyt, R., et al. (1991) Interleukin 10 (IL-10) and viral IL-10 strongly reduce antigen-specific human T cell proliferation by diminishing the antigen-presenting capacity of monocytes via downregulation of class II major histocompatibility complex expression. J. Exp. Med, . 174, 915–924
del Sol Mesa, A., et al. (2003) Automatic methods for predicting functionally important residues. J. Mol. Biol, . 326, 1289–1302[CrossRef][ISI][Medline].
Desrosiers, R.C., et al. (1997) A herpesvirus of rhesus monkeys related to the human Kaposi's sarcoma-associated herpesvirus. J. Virol, . 71, 9764–9769[Abstract].
Ferrara, N. (2001) Role of vascular endothelial growth factor in regulation of physiological angiogenesis. Am. J. Physiol. Cell Physiol, . 280, C1358–1366
Foster-Cuevas, M., et al. (2004) Human herpesvirus 8 K14 protein mimics CD200 in down-regulating macrophage activation through CD200 receptor. J. Virol, . 78, 7667–7676
Guerin, J.L., et al. (2001) Characterization and functional analysis of Serp3: a novel myxoma virus-encoded serpin involved in virulence. J. Gen. Virol, . 82, 1407–1417
Holm, L. and Sander, C. (1998) Removing near-neighbour redundancy from large protein sequence collections. Bioinformatics, 14, 423–429
Holzerlandt, R., et al. (2002) Identification of new herpesvirus gene homologs in the human genome. Genome Res, . 12, 1739–1748
Isaacs, S.N., et al. (1992) Vaccinia virus complement-control protein prevents antibody-dependent complement-enhanced neutralization of infectivity and contributes to virulence. Proc. Natl Acad. Sci. USA, 89, 628–632
Kaleeba, J.A., et al. (1999) A rhesus macaque rhadinovirus related to Kaposi's sarcoma-associated herpesvirus/human herpesvirus 8 encodes a functional homologue of interleukin-6. J. Virol, . 73, 6177–6181
Kara, P.D., et al. (2003) Comparative sequence analysis of the South African vaccine strain and two virulent field isolates of Lumpy skin disease virus. Arch. Virol, . 148, 1335–1356[ISI][Medline].
Kotenko, S.V., et al. (2000) Human cytomegalovirus harbors its own unique IL-10 homolog (cmvIL-10). Proc. Natl Acad. Sci. USA, 97, 1695–1700
Lee, C.G., et al. (2004) Vascular endothelial growth factor (VEGF) induces remodeling and enhances TH2-mediated sensitization and inflammation in the lung. Nat. Med, . 10, 1095–1103[CrossRef][ISI][Medline].
Lucas, A. and McFadden, G. (2004) Secreted immunomodulatory viral proteins as novel biotherapeutics. J. Immunol, . 173, 4765–4774
Macen, J.L., et al. (1993) SERP1, a serine proteinase inhibitor encoded by myxoma virus, is a secreted glycoprotein that interferes with inflammation. Virology, 195, 348–363[CrossRef][ISI][Medline].
Mayrose, I., et al. (2004) Comparison of site-specific rate-inference methods for protein sequences: empirical Bayesian methods are superior. Mol. Biol. Evol, . 21, 1781–1791
Moore, K.W., et al. (1990) Homology of cytokine synthesis inhibitory factor (IL-10) to the Epstein–Barr virus gene BCRFI. Science, 248, 1230–1234
Mullberg, J., et al. (2000) IL-6 receptor independent stimulation of human gp130 by viral IL-6. J. Immunol, . 164, 4672–4677
Petit, F., et al. (1996) Characterization of a myxoma virus-encoded serpin-like protein with activity against interleukin-1 beta-converting enzyme. J. Virol, . 70, 5860–5866[Abstract].
Pruitt, K.D. and Maglott, D.R. (2001) RefSeq and LocusLink: NCBI gene-centered resources. Nucleic Acids Res, . 29, 137–140
Pupko, T., et al. (2002) Rate4Site: an algorithmic tool for the identification of functional regions in proteins by surface mapping of evolutionary determinants within their homologues. Bioinformatics, 18, Suppl. 1, S71–S77[Abstract].
Rosengard, A.M., et al. (2002) Variola virus immune evasion design: expression of a highly efficient inhibitor of human complement. Proc. Natl Acad. Sci. USA, 99, 8808–8813
Rost, B. (1999) Twilight zone of protein sequence alignments. Protein Eng, . 12, 85–94
Seet, B.T., et al. (2003) Poxviruses and immune evasion. Annu. Rev. Immunol, . 21, 377–423[CrossRef][ISI][Medline].
Thompson, J.D., et al. (1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res, . 22, 4673–4680
Tulman, E.R., et al. (2000) The genome of a very virulent Marek's disease virus. J. Virol, . 74, 7980–7988
Upton, C., et al. (1991) Myxoma virus expresses a secreted protein with homology to the tumor necrosis factor receptor gene family that contributes to viral virulence. Virology, 184, 370–382[ISI][Medline].
Zhang, Z. and Henzel, W.J. (2004) Signal peptide prediction based on analysis of experimentally verified cleavage sites. Protein Sci, . 13, 2819–2824
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||