• Identificar os genótipos de amostras de astrovírus detectadas a partir de amostras fecais provenientes de crianças da região Centro-Oeste do Brasil, através do seqüênciamento genômico e/ou RT-PCR;
• Correlacionar os diferentes genótipos encontrados com a faixa etária da população estudada;
• Analisar seqüências de nucleotídeos provenientes dos três ORFs do genoma do astrovírus;
• Identificar os genogrupos das amostras virais analisadas;
• Obter a seqüência total do genoma de duas amostras de astrovírus pertencentes a genótipos que ainda não foram publicados na literatura.
4. ARTIGO I – Artigo científico aceito para publicação na revista Archives of Virology.
Molecular characterization of human astroviruses isolated in Brazil, including
the complete sequences of astrovirus genotypes 4 and 5
P.A. Silva 1, 2*, D.D.P. Cardoso 1 and E. Schreier 2
1 Laboratory of Virology, Institute of Pathology and Public Health, Federal University of Goiás, Goiânia, Brazil
2 Robert Koch-Institute, Berlin, Germany
Running title: Characterization of human astroviruses isolated in Brazil
*Author’s address: Paula Andreia Silva
Department of Molecular Epidemiology of Viral Pathogens Robert Koch-Institute
Nordufer 20 13353 Berlin GERMANY
e-mail: SilvaP@rki.de; SchreierE@rki.de
Phone: +49 1888 754 2379 Fax: +49 1888 754 2617
Summary
Human astroviruses (HAstV) are recognized as an important cause of gastroenteritis in young children worldwide. This study describes the molecular characteristics of astroviruses isolated in Brazil, using RT-PCR and molecular sequencing of segments of all three viral ORFs. Genetic analysis of a 348-nucleotide segment from ORF 2 demonstrated that the Brazilian isolates belong to HAstV genotypes 1 to 5 and 8. ORF 1b sequences displayed a high degree of nucleotide identity even between different genotypes, which disfavours HAstV genotyping in this region. ORF 1a sequence analysis classified all Brazilian samples as genogroup A. The complete sequences of HAstV genotype 4 (putative serotype 4) and genotype 5 (putative serotype 5) were determined for the first time.
Introduction
Gastrointestinal diseases belong to the most important causes of childhood morbidity and mortality, especially in developing countries where they are the second most common cause of death in children younger than 5 years [13, 25]. Enteric viruses, such as rota-, noro-, astro- and adenoviruses have been recognized as important causes of acute gastroenteritis. Corona-, picobirna-, and picornaviruses (enterovirus, Aichi virus) have also been shown to be an etiologic agent of gastroenteritis [2, 9, 12, 38].
Human astroviruses (HAstV) were first described as a human pathogen in 1975 during an outbreak of acute gastroenteritis among infants in the United Kingdom [18]. Since then, HAstV infections have been shown to be a significant cause of gastrointestinal diseases [9, 34]. Epidemiological studies worldwide have reported astrovirus prevalence rates of 2 to 26 % among children with acute gastroenteritis [5, 19, 23]
Astroviruses are spherical, non-enveloped 28 nm particles [18] belonging to the family
Astroviridae. Two genera have been described, Mamastrovirus and Avastrovirus [34]. The genus Mamastrovirus includes all human astrovirus strains, as well as the feline, porcine, and ovine
astrovirus. The genus Avastrovirus includes turkey astrovirus and avian nephritis virus [34]. The virus has a positive-sense, single-strand, polyadenylated RNA genome of approximately 6.8 kb and consists of three sequential open reading frames (ORFs) designated ORF 1a, ORF 1b, and ORF 2. ORF 1a and 1b, which are at the 5’ end of the genome, encode the non-structural viral proteins, including a serine protease and a RNA-dependent RNA polymerase (RdRp), respectively. ORF 1b is expressed as a fusion protein generated by a ribosomal frameshift [17, 20]. ORF 2 encodes a capsid precursor protein [16, 35]. The complete RNA sequences of the serotypes/genotypes 1, 2, 3, and 8 have been characterized [14, 21, 27, 37].
Based on a 348-nucleotide (nt) region within ORF 2, HAstV has been divided into eight genotypes (HAstV1 to HAstV8), correlating with the HAstV serotypes [22, 26, 33]. A similar categorization has been observed for nucleotide sequences in the ORF 1b region, although the distances between different genotypes are smaller. Phylogenetic analysis of a region in ORF 1a demonstrates two distinct astrovirus genogroups, A and B. Genogroup A is composed of serotypes 1 to 5 and 8, and genogroup B includes serotypes 6 and 7 [3].
In this work, we report the molecular characteristics of human astroviruses over segments of all three ORFs, examining virus isolates from Brazil. In addition, the complete sequences of HAstV genotype 4 and genotype 5 were determined for the first time.
Materials and Methods
Samples
Twenty fecal specimens, positive for astrovirus by nested-PCR, were included in this study. The samples were obtained from children with acute gastroenteritis in three cities of the West Central region of Brazil (Brasília-Distrito Federal, Goiânia-Goiás, and Campo Grande-Mato Grosso do Sul) between December 1994 and July 2003. The dates of collection and geographic information of the astrovirus-positive samples are given in Table 1.
Primers
Primers used to obtain partial sequences of the three astrovirus ORFs and the three overlapping amplicons of the isolates, Goiania/GO/12/95/Brazil (BrG4) and Goiania/GO/12/94/Brazil (BrG5), were either deduced from an alignment of published HAstV sequences or they were based on oligonucleotides Mon 340, Mon 348 (ORF 1a), Mon 343, Mon 344 (ORF 1b), Mon 269, and Mon 270 (ORF2) [3, 26]. All primers are listed in Table 2.
RNA Extraction and RT-PCR
Total viral RNA was extracted from 140 µl of a 20 % fecal suspension using the QIAamp viral RNA Kit (Qiagen). The RT-PCR, first and second amplification conditions were the same as described previously for astrovirus [28]. In order to avoid false-positive results, quality control measures were taken as recommended by Kwok and Higuchu [15]. Negative sample controls were included in each set of amplifications.
To determine the complete sequence of two isolates, BrG4 and BrG5, three fragments of the full genome of each isolate were reverse transcribed using M-MLV, Thermo-X, or PowerScriptTM Reverse Transcriptases (Invitrogen, BD Biosciences) according to the
manufacturer’s recommendations. PCRs were carried out using either Platinum (94°C, 2.5 min; 35 cycles: 94°C, 30 sec; 55°C, 30 sec; 72°C, 1.5 min; final extension of 3 min at 72°C), AccuPrimeTM Pfx (95°C, 1 min 45 sec; 35 cycles: 95°C, 15 sec; 58°C, 30 sec; 68°C, 3 min; final extension of 3’ at 68°C) or Phusion DNA polymerases (98°C, 30 sec; 35 cycles: 98°C, 10 sec; 60°C, 10 sec; 72°C, 1 min 40 sec; final extension of 3 min at 72°C). The enzymes were manufactured by Invitrogen and Finnzymes.
Molecular Sequencing
PCR products and clones were sequenced using an ABI Prism 3100 Genetic Analyzer and Big Dye Terminator Cycle Sequencing Mix (Applied Biosystems). All PCR products of three ORF fragments (Fig. 1A) were sequenced directly in both directions, using the same primer pair as in the nested PCR. The complete sequences of BrG4 and BrG5 were generated by amplifying 3 overlapping fragments of about 1.4, 3, and 2.4 kb (from 5’ to 3’ term) covering the complete viral genome (Fig.1B and 1C). These fragments were cloned into pGEM®-T Vector (Promega) according to manufacturer’s instructions, and sequenced using either primers of PCR or primers derived from sequence data obtained during the characterization.
Phylogenetic Analysis
Phylogenetic analysis was performed over regions of 348 nt in ORF 2, 267 nt in ORF 1b, and 198 nt in ORF 1a. For the complete sequences, additional analyses were performed over ORF2 (2,353 nt), ORF 1b (600 nt), and over the full-length genome. Sequence alignments and phylogenetic analysis were carried out using the CLUSTAL W program and the Phylogeny Interference Package (PHYLIP), version 3.57c. Genetic distances were estimated using the DNADIST program, and unrooted phylogenetic trees were constructed by the neighbour-joining
method. Bootstrap analysis of 100 replicate data sets was carried out using the SEQBOOT and CONSENSE programs [7, 8].
Results
Phylogenetic Analysis
Nucleotide sequences of three genomic regions localized in ORF1a, ORF1b, and ORF2 were obtained from 20 stool samples from Brazilian children with acute gastroenteritis. Phylogenetic analysis was performed for all samples and eight reference strains published in GenBank (HAstV-1 to HAstV-8; accession numbers are shown in Fig. 2), representing the eight genotypes.
Sequence analysis over an internal 348 nucleotides from the ORF 2 region showed that the Brazilian isolates clustered into six different genotypes. Twelve isolates (60 %) clustered into genotype 1, four (20 %) into genotype 2, and one (5 %) into each genotype 3, 4, 5, and 8 (Fig. 2A). Phylogenetic analysis exhibited intragenotypic identities of 89.4 – 97.4 % within genotype 1, 92 % within genotype 4, and > 96.6 % within the other genotypes. On the basis of nucleotide sequence variation within the genotypes, we propose that the BrG1-1 to BrG1-5 (89.4 – 89.7 % identity, 0.11 distance) and BrG4 isolates (92 % identity, 0.08 distance) represent subtypes of HAstV genotype 1 and 4, respectively. An isolate is considered to be a subtype when it displays < 95 % identity and > 0.05 distance from the reference strain [11, 33].
Analysis of the reference sequences over the same region exhibited 74.7 – 84.5 % identity among the genotypes, with the exception of genotypes 8 and 4, which presented 87.9 % identity between them. The BrG4 isolate, which clustered with genotype 4, showed 92 and 89.7 % identity compared to the reference sequences HAstV-4 and -8, respectively. The isolate that clustered into genotype 8 (BrG8) showed an identity of 87.9 % with HAstV-4 and 98.6 % with HAstV-8. These isolates had 89.9 % nucleotide identity among them and were identical on the
amino acid level (96.6 % identity with HAstV-4 and 100 % with HAstV-8). Based on the identity to the reference sequence and the topology of the phylogenetic tree, we considered the BrG4 isolate to be genotype 4 and the BrG8 isolate to be genotype 8.
Predicted amino acid sequences (aa) of the ORF 2 fragment showed that there is no divergence among the Brazilian genotype 1 isolates, with the exception of the BrG1-10 isolate, for which a divergence of 0.9 % was observed. The four genotype 2 isolates and the genotype 3 isolate presented amino acid sequences identical to the HAstV-2 and HAstV-3 reference strains. The genotype 5 isolate from Brazil showed 0.9 % amino acid divergence to HAstV-5.
Phylogenetic analysis of ORF 1b sequences showed that the identities among the genotypes (80.6 – 96.3 %) were higher compared to ORF 2. Smaller divergences were observed between genotypes 1, 2, 4, and 8 (3.8 – 9.7 %). 19 isolates clustered into the same genotypes as in ORF 2 and one isolate (BrG4), grouped into genotype 4 in ORF 2, could not be clearly assigned to any genotype in ORF 1b (Fig. 2B). This isolate showed a high nucleotide and amino acid sequence identity with HAstV-1 (91.8 % nt, 96 % aa), HAstV-2 (90.3 % nt, 97 % aa), HAstV-4 (89.9 % nt, 94.9 % aa), and HAstV-8 (89.6 % nt, 94.9 % aa).
In order to eliminate the possibility of coinfection with other HAstV genotypes, a segment spanning the ORF 1b / ORF 2 transition region (position 3742-4892 nt) of the BrG4 sample was amplified and sequenced (data not shown). The nucleotide sequence of this segment was identical with fragments obtained separately, and analysis of the 1150 nts showed that the BrG4 isolate presented a higher identity (93 %) with HAstV-4 (AF292075) than with the other reference sequences.
Analysis of the ORF 1a sequences demonstrated that all samples were classified as genogroup A viruses (Fig. 2C). The nucleotide identity between viruses of the same genogroup
varied from 88.4 to 98.5 %, whereas the identity among isolates of different genogroups varied from 78.4 to 82.9 %.
The accession numbers (acc. no.) of the Brazilian sequences deposited in GenBank are shown in Table 3.
Molecular Characterization of the Complete Genome Sequences
Based on the phylogenetic analysis of 348 nt of the ORF 2 region, the BrG4 (acc. no. DQ070852) and BrG5 (acc. no. DQ028633) isolates were classified into genotype 4 and genotype 5, which exhibited 92 % and 96.7 % identity with the respective reference strain. Pairwise sequence comparison over two additional genome regions between the Brazilian isolates and published sequences were performed.1
Over the complete capsid protein precursor gene localized in ORF2 (approximately 2352 nt), pairwise sequence comparison demonstrated that the BrG4 and BrG5 isolates showed 91.7 and 95.8 % identity with HAstV4 and HAstV5, respectively. The maximum intergenotypic identity (71 %) for BrG4 was observed with HAstV-8, and for BrG5 with HAstV-7. Considering the predicted amino acid sequences, the isolate identities within the respective genotypes were higher than 95 % but ranged between 65 and 77 % compared with different genotypes. The nucleotide and amino acid data obtained from the region analysed suggested that the BrG4 and BrG5 isolates belong to genotype/serotype 4 and 5, respectively.
Analysis of 600 nt from ORF 1b (position 3526-4126 of BrG4 and 3525-4125 of BrG5) showed a high identity (91.2 – 94.2 %) among reference sequences of genotypes 1, 2, 4, and 8, as already observed for the 267 nt ORF 1b segment . The BrG5 isolate clustered with the HAstV-5 reference sequence, presenting 95.2 % identity. The highest intergenotypic identity of this isolate
1The GenBank accession numbers of the sequences compared to the Brazilian isolates are as follow: L23513, L13745, AF141381, L23510, and L23512, AF260508 over ORF 2, and L23513, L13745, AF117209, Z33883, U15136, Z46658, AF248738, Z66541 over the ORF 1b fragment.
was 85.3 % with HAstV-3. The BrG4 isolate did not cluster with any reference strain in the phylogenetic tree, and presented 91.8, 90.7, 90.5, and 90 % nucleotide identities with HAstV-1, HAstV-2, HAstV-4, and HAstV-8, respectively.
The complete sequences were obtained by molecular sequencing of three cloned overlapping amplicons. The full-length genomic RNA of BrG4 and BrG5 isolates consisted of 6723 and 6762 nt, in each case, followed by a putative poly (A) tract. The isolates have a 5’ and 3’ non-translated region of 84 and 81 nt, and of 83 and 85 nt, respectively. The Brazilian isolates displayed typical human astrovirus genomic features, comprising three sequential open reading frames, which corresponded to ORFs 1a, 1b, and 2 [14]. ORF 1b overlaps ORF 1a by 64 nucleotides, containing a potential ribosomal frameshift signal.
Amino acid sequence analysis of the complete genome demonstrated that ORF 1a and 1b encode the putative non-structural proteins, including a serine protease and a RdRp. ORF 2 of the BrG4 isolate encodes a structural protein of 771 aa. The structural protein of the BrG5 isolate consists of 783 amino acids. Comparison of the non-structural protein regions of BrG4 and BrG5 with reference sequences of genotypes 1, 2, 3, and 8 showed a high level of conservation (higher than 93.5 % identity on the amino acid level). ORF 1a of both isolates contained a deletion of 21 and 45 nucleotides (position 2385 and 2456 of BrG4, and 2384 and 2457 of BrG5), compared with HAstV-3 and HAstV-1, respectively.
The sequence identity between the full-length genome of the Brazilian isolates and available complete sequences (HAstV-1 to -3 and -8) was less than 84 %, which implied the involvement of different genotypes.
Phylogenetic Analysis
Molecular characteristics of HAstV have been described for samples from different regions of the world [1, 6, 11, 23, 24, 27, 33]. This study is the first report of the molecular characteristics of astroviruses, encompassing three genome regions of Brazilian samples. Twenty astrovirus isolates from stool samples of children with acute gastroenteritis were included in the study.
All astrovirus genotypes except HAstV-6 and HAstV-7 were detected in the central west of Brazil, and HAstV-1 (60 %) was the most frequent genotype identified. These findings reinforce data of other surveys carried out in several regions of the world [10, 11, 23, 26, 27, 30], including Brazil that reported the prevalence of only HAstV-1 [5, 31]. The less common HAstV-2 to 5 and the rare HAstV-8 genotypes were also detected, demonstrating the simultaneous circulation of different astrovirus genotypes in the same geographical area, which corroborates the results of preceding studies [23, 24, 33].
Although strains and subtypes are not yet clearly defined, Walter et al. [33] considered an isolate to be a subtype when it presents < 95 % sequence identity and > 0.05 distance to the reference strain. Based on this definition, we have identified subtypes of genotypes 1 and 4 among the Brazilian isolates. Some authors have observed this variability within the genotypes and related the occurrence of new HAstV-1 subtypes to time progression, suggesting a genetic shift or introduction of a new strain [11, 29]. The majority of Brazilian HAstV-1 positive samples were collected between 2000 and 2003, and we considered this period to be insufficient to analyse the chronological order of the occurrence of different subtypes. We observed that among 5 viruses of the same subtype, 4 had been isolated in the same city (Campo Grande), suggesting a regional subtype predominance.
Over the 348 nts of the ORF 2 fragment, the reference sequences HAstV-4 and HAstV-8 as well as the Brazilian isolates BrG4 and BrG8, classified as astrovirus genotype 4 and 8,
presented a high degree of nucleotide identity between them. Moreover, the predicted amino acid sequences of both Brazilian isolates were identical, emphasizing a closer relationship between these genotypes. This similarity between HAstV-4 and -8 has been reported by other authors [27, 32], including Espul et al. [6], who also described identical amino acid sequences in the same genomic region. Analysis of the hypervariable 3’ end of the capsid region may be suited for differentiating these genotypes [22, 32].
Brazilian and reference strains over 267 (for all isolates) and 600 (for BrG4 and BrG5 isolates) nucleotide segments of the ORF 1b region demonstrated a high degree of identity on the nucleotide and amino acid levels. The similarity of the sequences in this region was high between genotypes 1, 2, 4, and 8, which has also been reported in previous surveys [3, 16]. Although the genotyping results over the ORF 1b and ORF 2 regions had coincided among them, with the exception of one isolate (BrG4), we suggest that the ORF 2 region, which displays a higher degree of variability, should be used for astrovirus genotyping.
Analysis of the BrG4 isolate in the ORF 2 region demonstrated its classification as genotype 4, and sequence comparison over the ORF 1b region showed that it was highly similar to reference sequences of other genotypes (1, 2, 4, and 8). This isolate, as already suggested by the analysis of the 5’ end capsid region, appears to be a different HAstV-4 subtype, which would explain the different characteristics over the ORF 1b region.
Of all of the sequences obtained in this study, 21 were submitted to GenBank. The criterion adopted was a representative sequence from a group of sequences that were identical or had higher than 99 % identity. This small variability means the exchange of less than 5 nucleotides, which may also be attributed to reverse transcriptase and polymerase error or to the quasispecies population. In this context, we considered one sequence to represent a very similar group and the publication of all sequences to be unnecessary.
Molecular Characterization of the Complete Genome Sequences
Complete genome sequences have so far been determined for 4 of the 8 human astrovirus genotypes (1, 2, 3, and 8) [14, 21, 27, 37]. In this study, two more genotypes (4 and 5) were characterized over the full genome, adding molecular information for each HAstV genotype.
The classification of the Brazilian isolates into genotypes was suggested by phylogenetic analysis over a region at the 5’ end of the capsid gene, which has been confirmed to correlate with the serotypes [22, 26, 33]. Considering this correlation, BrG4 and BrG5 isolates are the first serotype 4 and 5 astroviruses to be molecularly characterized over the complete genome.
Since the 5’ end of the complete sequences was obtained using a primer based on the reference sequence (L23513), the first 23 nucleotides resemble the reference sequence. The 3’ end was amplified using an oligo (dT) primer, which shows that HAstV contains a poly A tract, as reported by other authors [4, 21].
Analysis of both complete genomes confirmed the organization of the astrovirus genomic organization in three sequential ORFs (1a, 1b, and 2), with the two first ORFs being linked by a conserved ribosomal frameshift signal [14, 16]. Predicted amino acid sequences of each ORF suggested the expression of a serine protease, a RdRp, and a capsid protein, confirming the observation of other groups [16, 37].
Sequences of ORF 1a demonstrated that BrG4 and BrG5 isolates contain deletions of 21 and 45 nucleotides, as compared to HAstV-3 and HAstV-1, respectively. The absence of 15 amino acids at position 790 has been associated with an astrovirus adaptation to HEK and LLCMK2 cells [36, 37]. Controversially, some authors observed that a strain characterized directly from fecal samples [27] or grown in CaCo-2 cells [21] also presented the deletion in the same region. The Brazilian isolates, which were extracted from fecal samples, corroborate the
data observed for the German isolate [27], demonstrating that the occurrence of the deletion strain is not limited to astrovirus adapted to cells and that the selection of the deletion strains can occur. The deletion of 7 amino acids at position 768 could be observed only in comparison with HAstV-3. To our knowledge, there is no report that addresses the cause or importance of this deletion, and more detailed studies are needed to clarify these aspects.
The results of this study provide further molecular information on the astroviruses genome. This information can contribute to the development of vaccines and other preventive strategies for astrovirus infection.
Acknowledgements
P.A. Silva has been supported by a grant from DAAD (German Academic Exchange Service) and CNPq (Brazilian National Council for Scientific and Technological Development).
We would like to thank our Brazilian collaborators, José M. S. Teixeira (Institute of Health of Brasília-DF), Loreny G. Giugliano (Foundation of University of Brasília-DF) and Márcia S. A. Andreasi (Federal University of Mato Grosso do Sul-MS), for providing the samples. We also thank Djin-Ye Oh, Marina Höhne and Stefan Taube for the critical reading of the manuscript and Guilherme Rangel for the support on graphics. We also thank colleagues from FG15 in Robert Koch Institute-Berlin and from Laboratory of Virology in Goiânia-Brazil for the