Molecular characterization and epidemiological aspects of non-polio enteroviruses isolated from acute flaccid paralysis in Brazil: a historical series (2005–2017)

ABSTRACT Due to the advanced stage of polio eradication, the possible role of non-polio enteroviruses (NPEVs) associated to acute flaccid paralysis (AFP) cases has been highlighted. In this study, we described epidemiological aspects of NPEVs infections associated to AFP and explore the viral genetic diversity, information still scarce in Brazil. From 2005 to 2017, 6707 stool samples were collected in the scope of the Brazilian Poliomyelitis Surveillance Program. NPEVs were isolated in 359 samples (5.3%) and 341 (94.9%) were genotyped. About 46 different NPEV types were identified with the following detection pattern EV-B > EV-A > EV-C. The major EV-types were CVA2, CV4, EV-A71, CVB3, CVB5, E6, E7, E11, CVA13 and EV-C99, which corresponds to 51.6% of the total. Uncommon types, such as CVA12, EV-90 and CVA11, were also identified. Different E6 genogroups were observed, prevailing the GenIII, despite periods of co-circulation, and replacement of genogroups along time. CVA2 sequences were classified as genotype C and data suggested its dispersion in South-American countries. CVA13 viruses belonged to cluster B and Venezuelan viruses composed a new putative cluster. This study provides extensive information on enterovirus diversity associated with AFP, reinforcing the need of tailoring current surveillance strategies to timely monitor emergence/re-emergence of NPEVs.


Introduction
Worldwide, enteroviruses (EVs) are one of the major causes of human viral infections. EVs (genus Enterovirus, family Picornaviridae) are small, non-enveloped, positivesense, single-stranded RNA viruses. The genomic RNA (∼7.5 kb), encodes a large polyprotein, further processed into mature structural (VP1 to VP4) and nonstructural proteins [1]. EVs are classified into fifteen species: EV A-L and human rhinoviruses (HRV A-C) [2]. Seven species (EV A-D and HRV A-C) are able to cause human infection [3]. These viruses present a high genetic diversity, due to cumulative mutations and/or recombinations [4,5]. These events are frequently observed between polioviruses and other species C enteroviruses [4,6], favouring the emergence of novel viruses of unpredictable pathogenicity and clinically unknown outcomes [7].
EVs are responsible for a wide variety of clinical disease, ranging from common cold to cutaneous syndromes, visceral and neurological disorders, as acute flaccid paralysis (AFP) [8]. AFP is a clinical disorder with a broad array of etiologies, including infectious and non-infectious causes [9,10]. Polioviruses are the main cause of AFP, however, a considerable proportion of non-polio enteroviruses (NPEVs) has been increasingly associated with this neurological disorder [11].
AFP surveillance is part of a global programme coordinated by the World Health Organization (WHO) and conducted by the Brazilian Ministry of Health (MoH) at the national level. These activities have been kept at appropriate levels in Brazil [12]. In the context of global eradication of polioviruses, information on NPEVs circulation is key to understand their role in AFP context. Surveillance data shows that in Europe, USA and Africa, most cases have been associated with EV-B species, whereas species A are normally more prevalent in Asia [13][14][15][16][17]. The etiology of AFP cases in Brazil is systematically determined, although information on molecular diversity and NPEV circulation patterns is still scarce [18].
In this report, we assessed the prevalence of different EVs among AFP cases and explored the molecular diversity and putative circulation patterns of CVA2, E6 and CVA13 viruses in different Brazilian geographical regions, along 13 years of surveillance (2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017). Our findings highlight a high genomic diversity of EV strains, reinforcing the need to improve the current surveillance programmes, to regularly detect and characterize at molecular level the circulating viral strains.

Ethics statements
Samples were collected in the scope of the Global Enterovirus Surveillance Program, as part of the WHO and National Public Health responses, dispensing a formal review by IRB.
The phylogenetic analysis based on VP1 sequences was also performed for the other highly frequent NPEVs, such as EV-A71, E7 and CVB5 (supplementary Figure S1). Although EV-A71 was the most prevalent within EV-A species viruses, only a small fragment (300 bp) was available for the majority of Brazilian sequences. The most recent Brazilian sequences could be classified as genotype C2 (2015-2016), whereas those obtained in former years belonged to genotype B2 (1999-2014). EV-A71, E7 and CVB5 revealed a genetic diversity. CVB5 sequences were classified as genogroup C, together with European and USA representatives. E7 Brazilian sequences were grouped into different clusters, including global representatives from distinct countries (supplementary Figure S1).

Amino acid substitutions in the BC loop
The VP1 region comprises major neutralizing antigenic sites, including the BC loopan important immunogenic site located between the β-sheets B and C [26,27]. Thus, we explored the deduced amino acid sequences from the BC loop and surrounding regions of prototype strains and some viral types. Because length differences in VP1 were described in the amino terminal domain, the BC loop and carboxyl-terminal domain [28], we defined the VP1 BC-loop positions, as previously reported [29,30]. Among CVA2 isolates, the substitution A102S was found in almost all analysed sequences. Furthermore, three other sequences presented different substitutions in the same position Ala→Gly, Ala→Arg and Ala→Asn (Figure 3(A)). Among E6 isolates, five different substitutions were found. R82K and T84D were shared by Brazilian and representative sequences, whereas R82N, D83G, T84A were only observed in single Brazilian isolates (Figure 3(B)).
To evaluate the probable effects of the aminoacid substitutions in the context of IgG-EV interaction, we carried out a molecular modelling analysis using our E6 isolates (supplementary Figure S2). Our findings suggested a small change in the net charge and volume with a limited effect of these substitutions in the BC loop region.

Discussion
Since global wild poliovirus eradication is near to completion, the vast majority of the existing AFP cases have been associated to other etiologies where NPEVs play an important role [9,11]. The Monitoring NPEVs circulation, such as EV-D68 and EV-A71, which also causes severe neurological disease [31] must be a key component of AFP surveillance, designed to control emergence/re-emergence of WPV and/or vaccine derived polioviruses (VDPV).
In the present study, the NPEV isolation efficiency in different cell culture systems revealed a higher susceptibility to viral infection of RD (59.3%) compared to HEp-2C cells (15.1%). EV-A and EV-B species were more efficiently isolated in RD cells, in contrast to EV-C species, more promptly isolated in HEp-2C cells. Almost 15% of all NPEVs isolates and 60% of all NPEVs typed from HEp-2C cells were EV-C (EV-A:1.5% and EV-B:38.5%). In accordance to previous studies [37][38][39], these findings reinforce the need to include the HEp-2C cell culture in the current NPEVs surveillance protocols, in order to avoid EV-C underestimation.
Interestingly, it has been suggested that climatic factors, as humidity and elevated temperature, seem to favour EV transmission and isolation rates of NPEVs [40]. However, we did not observe clear seasonal distribution patterns, in line with previous observations. In fact, due to complexity of the diseases caused by enterovirus mainly in neurological manifestations it may be hard to explain the influence of climatic factors on enterovirus infections [41]. The present study revealed that EV-A, EV-B and EV-C species accounted for 29.3%, 57.5% and 13.2% of all isolates, respectively. The high prevalence of EV-B species and the detection pattern EV-B > EV-A > EV-C is aligned to other AFP surveillance reports [16,33,42], even though the prevalence of EV-C can be eventually higher than EV-A infections in some contexts [32,37,38]. Our results on the higher frequency of EV-B infections reinforce previous data on the increase of aseptic meningitis cases associated to EV-B infection in Brazil [43]. This work also identified a number of EV-A isolates, which was higher than EV-B/EV-C during the years 2011-2012. Furthermore, it is worth to mentioning that EV-C species was not more prominent than EV-A species in this study. This is the opposite of what was suggested for tropical countries [39], although in 2008 and 2017 the number of EV-C species isolates had been slightly higher. Moreover, we observed a slight fluctuation in the number of isolates that were typed as EV-B species, which decreased from 2010 to 2012 while the EV-A species increased in frequency. This pattern was reversed between 2012 and 2014 and the cycle looks to repeat itself in 2015. To the best of our knowledge, this is the first report on a putative fluctuation between EV-A and EV-B species associated with enteroviruses surveillance. Further studies are required to understand these temporal fluctuations.
Our findings also highlight that Echoviruses (n=141) and CV-A (n=105) were the major detected EVs among AFP patients. A high frequency of Echoviruses and CV-A has been described in many countries [16,32,33,37,42,44]. Although CV-A seems to be responsible for less than 1% of all EV-types associated with aseptic meningitis cases in Brazil [43], it was frequently found in association with AFP cases in this study. Altogether, these results suggest different circulation patterns of these viruses associated with AFP cases and other neurological disease in the country. Further studies must be conducted to ascertain regional features during EV infection.
According to previous studies, E6 and E11 have been the most commonly detected EV types, especially among patients showing severe central nervous system syndromes [16,30,33,[42][43][44]. In this investigation, EV-A71 was the main pathogen associated with AFP (7.6%), followed by E6 (7.3%) and E11 (6.7%) suggesting a high occurrence of these types from 2005 to 2017. Indeed, the detection rate of these viruses observed here corroborates other studies conducted within AFP surveillance [33]. Although EV-A71 has been regularly detected in many studies involving AFP surveillance [16,33,37,42,[44][45][46], it is not common to be the main EV type associated with AFP as demonstrated in the current study. Surprisingly, while EV-A71 point out as major EV type associated with AFP, a recent report suggested that it represented less than 1% of all types associated with aseptic meningitis in Brazil [43], whereas E6 played a relevant role in both neurological disorders, AFP and aseptic meningitis [43]. These findings may suggest a higher tropism to neuronal cells of E6 in comparison to EV-A71. In fact, EV-types present different cell/tissue tropisms, as a consequence of host and viral factors [47]. Further studies are needed to better understand the mechanisms and main factors associated to viral neurotropism.
Although an increasing number of HFMD and aseptic meningitis cases associated with CVA6/CVA16 and E30 infections, respectively, has been reported in Brazil [43,48], an increase isolation rate of these EV-types was not observed among AFP patients. Consistently, even in countries with a high prevalence of these EVs, such as China and Brazil, these viral types seem to be poorly involved in AFP disease [43,48,49].
Another remarkable finding was the first description of some EV-types in Brazil, such as CVA2, CVA10 and EV-90. However, the identification of these EV-types does not mean a new or recent introduction into Brazil, but that they have not been previously identified. Similarly, this work also reports the detection of uncommon EV-types associated with AFP, such as CVA11, EV-90 and CVA12 [11].
Although E6, E11 and EV-A71 were the most detected EVs, it is worthwhile to highlight two other important types belonging to EV-C species, CVA13 and EV-C99. These types represented almost 10% of all EV identified in this study. These findings add valuable data on the role of these EV-types in neurological syndromes [11]. Interestingly, previous work have demonstrated that the co-circulation of a high number of different NPEV types, especially those from EV-C strains, favours recombination events [4][5] as shown by the recent isolation of type 2 VDPV polioviruses from the environment and the identification of an uncommon recombinant type3/type2 Sabin-related poliovirus in Brazil [34,35]. Moreover, multiple infections by different NPEVs have also been associated with atypical clinical presentations [7,50].
EV surveillance alongside with regular viral sequencing is relevant to (i) understand temporal and geographical patterns of viral dispersion; (ii) investigate putative associations between EV genotypes and clinical outcomes and (iii) timely identify the emergence of novel recombinant viruses of public health relevance [4,5,34,35,43,[51][52][53]. This study adds valuable information on EV genetic diversity in Brazil, contributing to the knowledge on EV infections in South America, where data on viral sequencing is still scarce.
From 2006 to 2017, we observed the spread of three different E6 genogroups in Brazil, prevailing the GenIII in recent years. Despite periods of co-circulation, the replacement of genogroups along time was evident, reinforcing previous observations on the global [22] and Brazilian contexts [43]. Our data also document the genetic variability of E6 viruses, evidenced by the co-circulation of different clades from each genogroup. Furthermore, no clear temporal or geographical patterns were perceived, in line with previous reports [22].
CVA2 is detected worldwide and has been responsible for outbreaks [21,54]. Based on nucleotide sequences of the entire VP1 gene, CVA2 could be classified into four different genotypes and subgenotypes [20]. Brazilian CVA2 sequences belonged to genotype C (subgenotype C2) and closely clustered with a representative from Venezuela. Given the currently available global sequences, we did not find further close genetic relationships between our Brazilian sequences and those sampled outside of South American (supplementary Figure S1). Interestingly, among E30 Brazilian viruses, a close relationship between viruses circulating in Brazil, USA and South-American countries had been reported [43]. The USA representative from our study belonged to subgenotype C1, composed of sequences from different continents and assessed along a decade, proposing the global distribution of this subgenotype. Our study also corroborates findings of Yang et al. on the geographical restriction of genotype D infections within China [20]. CVA2 has played a major role in herpangina [54], which is not a notifiable disease in Brazil. Thus, CVA2 disease can be underestimated, as observed in other countries [20].
CVA13 VP1 sequences present a high genetic diversity and different viral clusters have been described (A-F) [17,55]. As little was known about the genetic variability of CVA13 viruses circulating in Brazil, we analysed sequences obtained from different geographical regions, assessed from 2005-2012. All Brazilian viruses belonged to cluster B and were distributed into two distinct clades. In contrast to CVA2 and E6, no relatedness between Brazilian viruses and those circulating in other South American countries was perceived. Noteworthy, Venezuelan viruses composed a new putative cluster, showing 72.5% to 75.8% of nucleotide similarity with former CVA13 clusters A to D [17]. Altogether, these figures suggest the circulation of multiple viral clusters in South America.
Our phylogenetic results suggest a diversity of E6, A71, CVB5 and E7 in Brazil, with circulation of distinct genotypes, genogroups or clades, in line with previous reports in the country [43] and elsewhere [22,38,[56][57][58][59][60]. In addition to the lack of a clear geographical pattern of viral dissemination among the Brazilian regions, these figures suggest the occurrence of multiple introduction events. The apparent lower variability of CVA2 and CVA13 viruses in this sample when compared to those other explored EVsshould be interpreted in the light of the small number of assessed sequences (n=8, each). For those presented as supplementary material, the same is true due to the small sequenced fragment (about 300 bp). Further analyses, comprising a more representative sample are necessary to compose a better picture of viral variability and circulation in the country. Besides the identification and characterization of novel viral types, the increasing number of deposited complete VP1 or full genome sequences will allow improvement of NPEV genotyping and a better understanding on the patterns of viral dissemination within and between countries and continentsa critical information to tailor global surveillance initiatives.
In this study, we also analysed the BC loop region, which is associated with viral antigenicity and can play a relevant role in host-virus interactions [26]. At least, three aminoacid substitutions within this loop and in adjacent regions were found in CVA2, E6 and CVA13 sequences. A higher variation was observed among CVA13 sequences, with potential impact on the three-dimensional protein structure. The putative impact of these mutations on viral antigenicity, fitness and virus-host cell interaction must be further ascertained.
In conclusion, the present study reveals the circulation of many different NPEVs types and genetic variants of these viruses in Brazil. Many of these EVtypes have not been frequently associated with AFP, highlighting the need of studies on their role in the context of the central nervous system infections. Considering that EV infections are a public health concern, EV surveillance policies and strategies should be continuously tailored to allow the timely identification of novel emergent viruses and to improve monitoring of those already in circulation.