We determined the complete 33,763 base pair genome of the HAdV-A31 prototype strain and identified 34 putative genes. HAdV-A31 is a highly significant pathogen, which has been frequently isolated from severely affected hSCT recipients, and frequently presents itself as a disseminated disease, which is usually caused by species HAdV-C [10
]. Experimental data strongly suggested that species HAdV-C types have the ability to establish latent infections in mucosal lymphocytes and that stimulation of those cells can cause viral reactivation in cases of immunosuppression [21
]. Especially the early gene products (e.g. E3), which counteract host anti-viral defence mechanisms, might play a key role in the process of persistence and reactivation [39
]. A similar mechanism of persistence and reactivation can be suspected in case of HAdV-A31, which would explain its high incidence in immunosuppressed patients. Therefore, the early coding regions E1A, E1B and E3 of the newly generated HAdV-A31 prototype sequence were analyzed in detail for functional motifs. Moreover, these genome regions of seven HAdV-A31 wild type strains isolated from immunosuppressed patients were also sequenced in order to clarify whether a highly pathogenic subtype of HAdV-A31 was circulating in recent years. However, analysis of nucleic acid and predicted amino acid sequences of seven HAdV-A31 clinical isolates revealed high identity to the prototype strain (Table ). Non-synonymous mutations in clinical isolates clustered in the E3 region, but did not affect previously described and predicted functional sites and motifs (Figure ) [37
]. An S/N ratio of 2.36 within the species HAdV-A specific CR1 beta protein suggested selection of a potentially highly functional E3 protein, which is assumed to interact with the immunosurveillance of adenovirus infected cells [39
]. Unfortunately, experimental data about the functions of the E3 gene products CR1 alpha and beta were not available. Therefore, in silico
protein analysis of the predicted E3 CR1 alpha and beta proteins of HAdV-A31 were performed. Interestingly, the E3 CR1 beta protein of the HAdV-A31 prototype and all analyzed clinical isolates were predicted to exhibit an immunoglobulin-like (V-set) domain, which was predicted neither for the closely related HAdV-A12 nor for the corresponding E3 gene products of the related enteric species F adenoviruses. Immunoglobulin-like domains are described to be involved in cell-cell interaction of the immune system [52
]. A similar domain was previously described as involved in a novel feature of the soluble 49 K E3 gene product of HAdV-D19a adenoviruses [37
]. Functional studies of the 49 K protein of HAdV-D19a demonstrated proteolytic processing and secretion of the type Ia transmembrane protein [37
]. Furthermore, a NK cell binding activity was detected and the immunoglobulin-like domain of the HAdV-D19a 49 K protein was assumed to interact directly with NK cells, protecting infected cells against lyses [37
]. For comparison, in silico
protein analysis of HAdV-A31 CR1 beta also predicted a type Ia transmembrane domain, a (signalpeptide-)cleavage probability of ~83%, a C-terminal sorting motif and various glycosylation sites, all of which are analogous to the confirmed predictions for HAdV-D19a E3 49 K protein.
In contrast to these predictions for CR1 beta, potential functionality of the predicted CR1 alpha protein of HAdV-A31 have remained obscure, since all performed analyses and predictions did not reveal similarities with functional sites or motifs of any E3 counterparts of other HAdV species. Overall, CR1 alpha and beta amino acid sequence comparison between HAdV-A31 and -A12 showed a particularly low identity of only 76.1% and 72.1%, respectively. This is in considerable contrast to corresponding E3 gene products of HAdV-F40 and -41, which had a high intraspecies homology of 98.8% and 99.2%, respectively. Significant differences between both species A adenoviruses were also identified in the theoretical molecular weight and isoelectric point (pI) for CR1 alpha and beta gene products computed and visualized in virtual 2D gel analysis (data not shown). These considerable differences between HAdV-A31 and -A12 in the primary structure, protein size, pI and predicted functional domains consequently indicated differences in protein function and might be an important feature in explaining the described higher virulence of HAdV-A31.
Comparison of the other E3 gene products, 12.5 K, RID alpha, RID beta and 14.7 K, revealed sequence identities between 80 and 96% with HAdV-A12. With the exception of predicted phosphorylation sites of RID beta (Figure ), previously described functional sites and motifs are conserved in HAdV-A31 and HAdV-A12, suggesting a comparable functionality. As determined for HAdV-A12 and species F adenoviruses, the E3 transcription unit of HAdV-A31 also lacked the extensively studied GP 19 K protein, which down regulates the expression of MHC I molecules and NK activation receptors [53
]. As a substitute for this important immune escape mechanism, the E1A gene product of HAdV-A12 was identified to down regulate the expression of MHC class I molecules by interfering with the transcription of MHC I gene products [56
]. This unique feature can be assumed for the E1A protein of HAdV-A31 as well, because amino acid stretches of HAdV-A12, which have been identified as essential for this mechanism, were identified in the predicted E1A gene product of HAdV-A31 at corresponding positions. While the E1A protein is known to have pro-apoptotic features, the E1B 19 K gene product of HAdV-C5 shares homology with the cellular anti-apoptotic Bcl-2 protein and interferes with a set of different cellular pro-apoptotic proteins (Bak, Bax and Nbk/Bik), thus protecting infected cells against apoptosis [32
]. Comparison of the E1B 19 K small t
- antigen homologue of HAdV-A31 with HAdV-A12 revealed high divergences of the C-terminus. Since the E1B 19 K C-terminus of HAdV-C5 has been identified as exhibiting a functional domain that influences the lateral viral spread of HAdV-C5 by interfering with cellular apoptotic pathways [57
], the observed divergence between HAdV-A31 and -A12 might have functional relevance.
In addition to the capability of persistence and of reactivation by interfering with the host immune response, the capacity for dissemination seems to be essential for a highly pathogenic HAdV subtype. For example, an outbreak of the strictly enterotropic HAdV-F41 did not cause any fatalities in paediatric hSCT recipients [17
], which might be due to the missing integrin binding RGD motif within the penton protein of species F adenoviruses. Since fatal dissemination of adenoviruses affects various organic systems, a more promiscuous behaviour in cell attachment and entry can be assumed to be favourable [58
]. Interestingly, the protein IX of HAdV-A31 revealed an RGD motif at amino acid position 102 - 106, which is conserved within all seven analyzed clinical isolates. This RGD motif is only 40 residues distant from the C - terminus, and it is unique among all sequenced human adenoviruses. As described previously, the C-terminus of HAdV protein IX is exposed on the outer surface of the virion [42
]. It was shown for protein IX of BAdV-3, which is 125 amino acids in length that the N-terminus (13 - 32) and the central region (61 - 80) have immunogenic sites but are not exposed on the outer surface of the virion [60
]. So far, no structural data about the exposed region of the C-terminus of protein IX of HAdV-A31 is available, but the observations for BAdV-3 indicate that the amino acid stretch of the protein IX of HAdV-A31 that contains the RGD motif might be present on the outer surface of the virion. As the HAdV-A31 fiber shaft is shorter and probably less flexible than the fiber shaft of HAdV-A12, a functional binding of the secondary cellular receptor to the RGD motif of protein IX instead of the RGD motif in the penton base may be possible or even preferred. Moreover, the construction of adenoviral vectors with an incorporated RGD motif within the C - terminus of protein IX has demonstrated that the additional RGD motif resulted in a significant augmentation of fiber independent infection of CAR-deficient cell types [60
]. Therefore, the additional RGD motif within the protein IX HAdV-A31 might lead to a more effective targeting and internalization, and could be a factor in increased transmission and infectivity of the virus.
In addition to protein IX, sequence comparison of the cellular receptor binding sites of the penton protein of the clinical isolates with the HAdV-A31 prototype revealed an amino acid substitution (F305Y) in two clinical isolates. This substitution is close to the functional RGD motif and may influence integrin binding. Moreover, a low S/N ratio of 2.78 (with exception of clinical isolate number: 95/8866) for the fiber knob indicated selective pressure on a major structural protein of the clinical isolates.
Overall, the sequence divergence of the isolated clinical strains in comparison to the prototype sequence was determined to range between 99.2 and 100%. These results suggested that all isolated strains were closely related to the prototype; a single outbreak subtype associated with severe disease in stem cell transplant recipients was not identified (Table ). This is in congruence with previous results of RFLP analysis of 79 HAdV-A31 wild type isolates from immunocompromised and immunocompetent hosts, where a wide variety of slightly genetically different subtypes of HAdV-A31 was described [22