Molecular Characterization of Avian Rotaviruses F and G Detected in Brazilian Poultry Flocks

Avian rotaviruses (RVs) are important etiologic agents of gastroenteritis in birds. In general, avian RVs are understudied; consequently, there is a paucity of information regarding these viruses. Therefore, the characterization of these viral species is highly relevant because more robust information on genetic, epidemiologic, and evolutionary characteristics can clarify the importance of these infections, and inform efficient prevention and control measures. In this study, we describe partial genome characterizations of two avian RV species, RVF and RVG, detected in asymptomatic poultry flocks in Brazil. Complete or partial sequences of at least one of the genomic segments encoding VP1, VP2, VP4, VP6, VP7, NSP1, NSP4, NSP4, or NSP5 of 23 RVF and 3 RVG strains were obtained, and demonstrated that multiple variants of both RVF and RVG circulate among Brazilian poultry. In this study, new and important information regarding the genomic characteristics of RVF and RVG is described. In addition, the circulation of these viruses in the study region and the genetic variability of the strains detected are demonstrated. Thus, the data generated in this work should help in understanding the genetics and ecology of these viruses. Nonetheless, the availability of a greater number of sequences is necessary to advance the understanding of the evolution and zoonotic potential of these viruses.


Introduction
Rotaviruses (RVs) are members of the Rotavirus genus in the Reoviridae family and are divided into 12 species A-L [1][2][3][4][5]. Species A, B, C, and H (RVA, RVB, RVC, and RVH) infect both humans and animals; species E (RVE) has only been detected in pigs, while the other species have only been detected in other animal hosts. In addition, species D, F, and G (RVD, RVF, and RVG) have been detected in birds; species I (RVI) has been detected in canines and felines; species J (RVJ) in bats; and species L (RVL) in shrews [2][3][4][5].
The RV genome is composed of 11 segments of double-stranded RNA (dsRNA) encoding six structural proteins, VP1-VP4, VP6 and VP7, and five or six nonstructural proteins, NSP1-NSP5 and NSP6, depending on the viral strain [6,7]. The virion lacks a lipid envelope and measures approximately 75 nm in diameter. The genome is surrounded by three layers of proteins. The innermost layer consists of VP2, which surrounds the genome, and two other structural proteins, VP1 (RNA-dependent RNA polymerase, RdRp) and VP3 (responsible for capping RV ss(+)RNAs acting as mRNAs), comprising the core of the particle. VP6 forms the intermediate layer (intermediate capsid), which surrounds the core, thereby assembling into a double-layer particle (DLP). VP6 is the main structural component of the virion, and plays a fundamental role in the structure of the particle

RNA Extraction, Amplification, and Sequencing
Stool suspensions were prepared in 10% (w/v) phosphate-buffered saline (pH 7.2) and then centrifuged at 2500× g for 5 min; viral dsRNA was extracted from 300 µL of the supernatant using the guanidine isothiocyanate-phenol-chloroform method [44]. RV cDNA was synthesized and amplified using specific primers (Appendix Table A1) designed according to RVF and RVG sequences available in GenBank. The amplified genomic segments were sent to Macrogen Inc. (Seoul, Republic of Korea) for sequencing. Overlapping sequences were assembled and edited using SeqMan, EditSeq, and MegAlign in the Lasergene software package (DNASTAR, Madison, WI, USA). Phylogenetic analysis was performed with MEGA software (version 11.0.11) [45] and

RNA Extraction, Amplification, and Sequencing
Stool suspensions were prepared in 10% (w/v) phosphate-buffered saline (pH 7.2) and then centrifuged at 2500× g for 5 min; viral dsRNA was extracted from 300 µL of the supernatant using the guanidine isothiocyanate-phenol-chloroform method [44]. RV cDNA was synthesized and amplified using specific primers (Appendix A Table A1) designed according to RVF and RVG sequences available in GenBank. The amplified genomic segments were sent to Macrogen Inc. (Seoul, Republic of Korea) for sequencing. Overlapping sequences were assembled and edited using SeqMan, EditSeq, and MegAlign in the Lasergene software package (DNASTAR, Madison, WI, USA). Phylogenetic analysis was performed with MEGA software (version 11.0.11) [45] and aligned with corresponding sequences obtained from GenBank (https://www.ncbi.nlm.nih.gov/nucleotide/accessed on 21 March 2023). Access codes of the reference strains are presented in Appendix A Table A2. Dendrograms were constructed using the maximum likelihood method and the Kimura 2-parameter model. Statistical significance was estimated by bootstrap analysis with 1000 pseudoreplicates. Sequences generated in this study were deposited in Gen-Bank under accession numbers OL688642, OL688643, OL688646, OL688659-OL688661, and OQ627756-OQ627800.

Analysis of RVF Strains
Complete or partial sequences of the open reading frame (ORF) of gene segment 1, encoding VP1, were obtained from three strains detected in Bom Jardim, Rio de Janeiro. A comparison of the complete sequence of the BJ12 strain, and reference strains from Germany and the Republic of Korea showed nucleotide identities of 88.7% and 89.7%, respectively, and amino acid identities of 96% and 97%, respectively. Partial sequences were obtained for strains BJ1 and BJ7 from Bom Jardim. To facilitate the analysis, all sequences were edited to the length of 1115 bp, which permitted the analysis of the three strains from Rio de Janeiro and reference strains from Germany and South Korea, and other RVF strains detected previously in Brazil ( Figure 2). The comparison of the partial sequences of the three strains detected in Bom Jardim showed 100% identity and that they were closer to the Brazilian RS-15-4S-2 strain (90.5% identity), detected in the state of Rio Grande do Sul, southern Brazil, than to the strains from Germany (03V0568; 88.4%) and Republic of Korea (D62; 89.7%).

Genomic Segment Encoding VP1
Complete or partial sequences of the open reading frame (ORF) of gene segment 1, encoding VP1, were obtained from three strains detected in Bom Jardim, Rio de Janeiro. A comparison of the complete sequence of the BJ12 strain, and reference strains from Germany and the Republic of Korea showed nucleotide identities of 88.7% and 89.7%, respectively, and amino acid identities of 96% and 97%, respectively. Partial sequences were obtained for strains BJ1 and BJ7 from Bom Jardim. To facilitate the analysis, all sequences were edited to the length of 1115 bp, which permitted the analysis of the three strains from Rio de Janeiro and reference strains from Germany and South Korea, and other RVF strains detected previously in Brazil ( Figure 2). The comparison of the partial sequences of the three strains detected in Bom Jardim showed 100% identity and that they were closer to the Brazilian RS-15-4S-2 strain (90.5% identity), detected in the state of Rio Grande do Sul, southern Brazil, than to the strains from Germany (03V0568; 88.4%) and Republic of Korea (D62; 89.7%). Dendrograms constructed from complete or partial sequences of the VP1, VP2, VP4, VP6, VP7, NSP1, NSP2, NSP4, and NSP5-encoding genome segments of RVF strains. Distances were corrected using the Kimura 2-parameter model. The dendrogram was constructed using the maximum likelihood method. Statistical support was provided by bootstrapping 1000 pseudoreplicates. Bootstrap values above 70% are given as branch nodes. GenBank reference strain accession numbers are shown next to the strain identification. RVF strains detected in this study are indicated by black triangles.

Genomic Segment Encoding VP2
Partial sequences of gene segment 2, encoding VP2 of three strains, BJ1, BJ7, and BJ12, were obtained. For phylogenetic analysis, the sequences were edited to the length of 1200 bp ( Figure 2). The strains detected in this study had 99.4 to 100% identity among themselves, and 88.7% to 90.7% nucleotide identity with German and Republic of Korean reference strains, respectively.

Genomic Segment Encoding VP4
Complete sequences of the ORFs of gene segment 3, which encodes VP4, of two strains, BJ1 and BJ7, were obtained. Phylogenetic analysis showed high nucleotide (99.9%) and amino acid (100%) identities. Comparison with reference strains showed that the strains from Bom Jardim were closer to the German strain 03V0568, with percentage identities of nucleotides and amino acids of 96.1% and 96.8%, respectively. Identities with the Republic of Korean strain (D62) were 78.1% for nucleotides and 83.7% for amino acids. Dendrograms constructed from complete or partial sequences of the VP1, VP2, VP4, VP6, VP7, NSP1, NSP2, NSP4, and NSP5-encoding genome segments of RVF strains. Distances were corrected using the Kimura 2-parameter model. The dendrogram was constructed using the maximum likelihood method. Statistical support was provided by bootstrapping 1000 pseudoreplicates. Bootstrap values above 70% are given as branch nodes. GenBank reference strain accession numbers are shown next to the strain identification. RVF strains detected in this study are indicated by black triangles.

Genomic Segment Encoding VP2
Partial sequences of gene segment 2, encoding VP2 of three strains, BJ1, BJ7, and BJ12, were obtained. For phylogenetic analysis, the sequences were edited to the length of 1200 bp ( Figure 2). The strains detected in this study had 99.4 to 100% identity among themselves, and 88.7% to 90.7% nucleotide identity with German and Republic of Korean reference strains, respectively.

Genomic Segment Encoding VP4
Complete sequences of the ORFs of gene segment 3, which encodes VP4, of two strains, BJ1 and BJ7, were obtained. Phylogenetic analysis showed high nucleotide (99.9%) and amino acid (100%) identities. Comparison with reference strains showed that the strains from Bom Jardim were closer to the German strain 03V0568, with percentage identities of nucleotides and amino acids of 96.1% and 96.8%, respectively. Identities with the Republic of Korean strain (D62) were 78.1% for nucleotides and 83.7% for amino acids. Interestingly, another Brazilian strain from Rio Grande do Sul, RS-BR-14-4S-3, was closer to the German and Republic of Korea strains than to the strains from Bom Jardim.
The partial sequences of BJ1, BJ7, and BJ12 strains were analyzed by adjusting all sequence lengths to 1130 bp ( Figure 2). Phylogenetic analyses showed that the strains detected in this study had 99.9 to 100% identity with each other, and were more closely related to the reference strains from Germany than to those from South Korea and Brazil.

Genomic Segment Encoding VP6
Complete sequences of the OFR of gene segment 6, encoding VP6, were obtained for the BJ1, BJ7, and BJ12 strains. Phylogenetic analysis showed that the strains detected in this study were identical. Comparison with German (03V0568) and Swiss (PB56-SII36) reference strains showed nucleotide identities of 88.1-85.8% and amino acid identities of 95.7-92.1%, respectively, with RVF from Bom Jardim. On the other hand, the comparison with a Brazilian strain detected in the state of Rio Grande do Sul showed 91.9% nucleotide and 99.7% amino acid identities.
In addition, we obtained the partial VP6 sequences of 18 RVF strains from both Bom Jardim in the state of Rio de Janeiro and Marechal Floriano in the state of Espírito Santo. Sequences were adjusted to the size of 280 bp ( Figure 2). Phylogenetic analysis showed that our strains presented greater genetic proximity to RVFs previously detected in Brazil, although they form a separate clade than those detected in Germany (03V0568), the Republic of Korea (D62), and Switzerland (PB56-SI36). Interestingly, the BRA57 strain, detected in the state of Pará, northern Brazil, grouped separately with a partridge strain (956_1) detected in Italy.

Genomic Segment Encoding VP7
Partial sequences of gene segment 9, which encodes VP7, were obtained for strains BJ1, BJ7, and BJ12. Sequences were adjusted to 398 bp, which covered the same ORF regions of all samples ( Figure 2). Phylogenetic analysis showed that the strains detected in this study had 99.2 to 100% identity. Comparison with German (03V0568), Chinese (361BR-K141_94516), and Republic of Korean (D62) strains showed identities of 75.9% to 80.4%.

Genomic Segment Encoding NSP2
Sequence of the complete ORF of gene segment 8, encoding the NSP2 of the BJ1 strain, and partial sequences of the BJ7 and BJ12 strains were obtained. The sequence analysis of the BJ1 ORF disclosed an identity of 88.6% and 92.8% of nucleotides and amino acids, respectively, with the German reference strain (03V0568), and 90.3% and 95% with two Brazilian strains (RS-BR-15-4S-7 and RS-BR-15-5R) detected in Rio Grande do Sul. When the length of the sequences was adjusted to 594 bp (Figure 2), phylogenetic analysis showed that they grouped in a single clade close to, but distinct from, other strains detected in southern Brazil. This could suggest the independent evolution of the RVF strains circulating in the country. However, the analysis of the complete ORFs of a larger number of samples is necessary to confirm this hypothesis. Complete sequences of the ORF of gene segment 11, which encodes NSP4, were obtained for six strains, BJ1, BJ7, BJ12, and BJ37 from Rio de Janeiro, and MF104 and MF110 from Espírito Santo (Figure 2). Phylogenetic analyses showed that the strains detected in this study presented high nucleotide and amino acid identity. Comparison with reference strains from Germany (03V0568), China (361bR-k141_94516), and the Republic of Korea (D62) showed that the Brazilian strains had nucleotide identities of 88.4-88.6% and amino acid identities of 88.9-91.8%. Unfortunately, genetic information regarding NSP4 of other Brazilian RVF strains is not available.

Genomic Segment Encoding NSP5
The complete ORF of gene segment 10, encoding NSP5, was obtained for five RVF strains: BJ1, BJ7, BJ10, BJ11, and BJ12. Phylogenetic analyses showed that the BJ strains had 91.2-100% nucleotide identity and 94.5-100% amino acid identity among themselves. Comparison with reference strains from Germany, China, and the Republic of Korea showed that the strains from Rio de Janeiro showed nucleotide identities of 86.7% to 89.5%, respectively, and amino acid identities of 89.8% and 94.5%, respectively. Adjusting the length of the sequences to 400 bp enabled the analysis of eight more strains: BJ14, BJ30, BJ31, BJ32, BJ37, BJ39, BJ41, and BJ51 detected in Bom Jardim, and MF06 detected in Marechal Floriano. The dendrogram shows that the Brazilian strains were closely related but not identical (Figure 2).

Genomic Segment Encoding VP6
Complete or partial sequences of the sixth segment encoding VP6 of three RVG strains detected in Marechal Floriano were obtained: MF48 (complete ORF), MF40, and MF41 (partial ORFs). When comparing the complete ORF sequence of the MF48 strain with the reference strains that are available on GenBank, we observed that the strain from Brazil is genetically closer to chicken strains from South Africa (MRC-DPRU1679) and Germany (03V0568), with 89% and 89.3% nucleotide identity, and 96.9 and 97.7% amino acid identity, respectively. It has 80% and 89% nucleotide and amino acid identity, respectively, with the US turkey strains (Minnesota-1 and Minnesota-2) and is phylogenetically distant from the Hong Kong pigeon strain (HK18) (Figure 3). When we adjusted the sequence length to 642 bp, we were able to include the MF40 and MF41 strains in the analysis (Figure 3). The dendrogram showed that the strains detected in Brazil were genetically similar (94.5 to 96.9% identity) yet grouped differently to form three sub-clades: the strains from Espírito Santo (MF40, MF41 and MF48), detected in 2018, grouped in one clade and the strains from Rio Grande do Sul, detected in 2009, grouped in two different clades. These data suggest that the Brazilian strains have a common ancestor and possibly evolved independently of each other, leading to the observation of different variants circulating in our country. Partial sequences from a partridge strain (956-2) from Italy and a gull strain (H01-10385) from the Netherlands were also included in this analysis. The phylogenetic analysis showed a separation between strains of different host species, except for the partridge strain that grouped with chicken samples. These findings suggest (i) that different VP6 genotypes circulate among RVG strains and (ii) that these genotypes may be associated with different hosts (chickens, turkeys, and pigeons).

Genomic Segment Encoding NSP2
A partial sequence of the eighth gene segment that encodes NSP2 was obtained for the MF48 RVG strain ( Figure 2) and showed 89.6-90.1% identity with the RVG strains detected in chickens, 84.7% with the pigeon strain, and 80.4% with turkey strains.

Genomic Segment Encoding NSP5
Three complete ORFs of gene segment 11, encoding the NSP5, from strains MF40, MF41, and MF48 were obtained. Phylogenetic analysis showed that, similar to the findings of VP6, the Brazilian strains were genetically closer to chicken strains and once again separated into different clades according to host species (Figure 2). . Dendrograms constructed from complete or partial sequences of the VP6, NSP2, and NSP5encoding genome segments of RVG strains. Distances were corrected using the Kimura 2-parameter model, and the dendrogram was constructed using the maximum likelihood method. Statistical support was provided by bootstrapping 1000 pseudoreplicates. Bootstrap values above 70% are given as branch nodes. GenBank reference strain accession numbers are shown next to the strain identification. RVG strains detected in this study are indicated by black triangles.

Genomic Segment Encoding NSP2
A partial sequence of the eighth gene segment that encodes NSP2 was obtained for the MF48 RVG strain ( Figure 2) and showed 89.6-90.1% identity with the RVG strains detected in chickens, 84.7% with the pigeon strain, and 80.4% with turkey strains.

Genomic Segment Encoding NSP5
Three complete ORFs of gene segment 11, encoding the NSP5, from strains MF40, MF41, and MF48 were obtained. Phylogenetic analysis showed that, similar to the findings Viruses 2023, 15, 1089 9 of 18 of VP6, the Brazilian strains were genetically closer to chicken strains and once again separated into different clades according to host species (Figure 2).

Discussion
RVs are the most prevalent etiologic agents of diarrhea among diverse hosts that include mammals and birds [6]. Avian RVs have great epizootic potential and can thereby bring economic losses to poultry farming. Despite their importance, they are poorly studied, and consequently, genomic and epidemiologic information regarding these viruses, especially RVF and RVG, is limited. Genomic characterization of circulating strains is necessary for accurate identification and classification, and to provide a better understanding of viral evolution [30,[46][47][48]. Moreover, knowledge of genetic diversity is important to inform the prevention and control of viral infections. For these reasons, genomic sequencing studies have been conducted to demonstrate the wide diversity of RVs. However, most data in the literature on RV are related to RVA, underscoring the need to study other species.
In this study, we obtained the complete ORF sequences of the genome segments that encode VP6, NSP2, NSP4, and NSP5 of RVF strains circulating in two southeastern states of Brazil: Rio de Janeiro and Espírito Santos. When compared to the sequences of reference strains, they presented nucleotide identity values for VP6, NSP2, and NSP4 greater than 85%, the suggested cutoff value for RVA. The application of the same criteria to RVF would suggest that these belong to the same genotype. However, it is important to point out that the Brazilian strains grouped in a distinct clade in the dendrograms, suggesting their independent evolution.
On the other hand, the percentages of nucleotide identity observed for NSP5 could suggest the circulation of different genotypes, as they are below 91%, which is the cutoff value for this gene in the case of RVA. Furthermore, analysis of partial NSP5 sequences showed that although the strains in this study clustered closely, they are not all in the same clade. This difference may have arisen through the typical mechanisms of RV evolution, primarily through point mutations. The MF06 strain originating from Espírito Santo presented similarity with the strains from Rio de Janeiro; however, its distribution in the dendrogram shows segregation in a specific clade, suggesting the circulation of separate variants in different Brazilian states.
When analyzing partial sequences, we observed that for VP1, using a fragment that corresponded to 54% of the ORF, the Brazilian strains and the reference strains shared >83% identity, a cutoff value stipulated to differentiate the VP1 genotypes from RVA. For genes encoding VP2, VP4, VP7, and NSP1, the sequences corresponded to <50% of the ORF (20%, 20%, 45%, and 40%, respectively). In the case of genes encoding VP2 and NSP1, the identity between all strains was well above a cutoff value of 84%, suggesting a single genotype. However, for VP4, the nucleotide identity values and the topology of the dendrogram suggest two distinct genotypes: one represented by the strain from Germany and the strains from Rio de Janeiro, and a second represented by the strain from South Korea and the strain from Rio Grande do Sul. Interestingly, these data suggest a distinct origin of the strains from Rio de Janeiro and Rio Grande do Sul. In the case of VP7, the results suggest three different genotypes represented by the Brazilian strains, the strains from Germany and China, and the strain from South Korea.
VP4 and VP7 are responsible for cell entry and the induction of neutralizing antibodies and, therefore, represent the main targets for vaccine development [49]. Consequently, an understanding of the genetic characteristics of these proteins is important. Unfortunately, sequences of VP7 from Brazilian strains have not been obtained, and only a partial sequence of VP4 is available. The possibility of varying genotypes of VP4 and VP7 carries extremely important implications for vaccine development, as changes in the antigenicity of these proteins can impact the generation of neutralizing antibodies. No vaccines are available to prevent avian rotavirus disease; consequently, an understanding of the different circulating variants is essential to the future development of an effective vaccine. At present, the need to develop specific vaccines against infections caused by RVF and RVG has not yet been established. Although this virus has been detected in different species of birds, with and without disease symptoms, knowledge of its epidemiology and pathogenic importance is still incipient to allow such an evaluation. On the other hand, RVA and RVD species are already established as important disease agents in birds [18].
Overall, phylogenetic analyzes of all sequenced RVF genome segments demonstrated that Brazilian strains, although possibly having a common ancestor, evolved independently, leading to different variants circulating in the country.
As for RVG strains, three genes, VP6, NSP2, and NSP5, were sequenced. Analysis of the complete and partial ORFs of VP6 suggests the existence of different genotypes and possibly the segregation of these genotypes according to host species. Furthermore, the circulation of different variants in Rio de Janeiro and Rio Grande do Sul was also demonstrated. This evidence was reinforced by the analysis of NSP2 and NSP5.
The RCWG proposal for RVA defines the cutoff values for the genetic classification of individual genes. However, for the application of these criteria, the RCWG imposes conditions for the identification of genotypes. Therefore, no genotype should be assigned based on less than 500 nucleotides or less than 50% of the ORF sequence. As the percentage cutoff values for each of the 11 genomic segments were calculated based on complete ORFs, applying these cutoff percentages to a partial ORF sequence can lead to erroneous conclusions. Only under certain circumstances, when all three of the following restrictions are met, can a partial gene sequence be used to assign an already established genotype to a given RVA gene: (1) at least 50% of the ORF gene sequence must be determined; (2) at least 500 nucleotides of ORF must be determined; and (3) the identity between the analyzed strain and the reference strains belonging to an established genotype must be at least 2% above the cutoff value. New genotypes can be identified only by analyzing the complete ORF of the gene in question [31]. The extremely small number of available RVF and RVG genomic sequences, particularly complete ORF sequences, impedes the establishment of reliable cutoff values for the determination of genotypes of these species. Thus, our analyses have theoretical value, intended only to demonstrate variants within these viral species, which, in the future, may or may not be defined as different genotypes.
Birds are recognized as important reservoirs of viruses that can threaten human health, such as the influenza A virus, which infects both birds and mammals, enabling the reassortment of genomic segments and changes in tropism and transmission efficiency [50]. Influenza viruses and RVs are both zoonotic pathogens with segmented RNA genomes, suggesting that the same phenomena could occur in RVs. The zoonotic potential of RVs that infect mammals, particularly RVA and RVC, has already been demonstrated unequivocally [30,51]. On the other hand, the zoonotic potential of avian RVs has yet to be explored. Thus, further studies will be essential to understand better the molecular characteristics and evolution of RV species that infect birds.
In this study, new and important information regarding the genomic characteristics of RVF and RVG was described, despite the small number of characterized strains. These difficulties can be partially explained by the samples used for amplification of the viral genomes, which came from asymptomatic chickens and therefore contained low viral loads, as well as the limited number of available RVF and RVG sequences, which confounded the selection of more efficient primers. However, despite the limitations of this study, the circulation of RVF and RVG variants among poultry was clearly demonstrated. Nonetheless, the availability of a greater number of sequences is necessary for a better understanding of the evolution and zoonotic potential of these viruses.