The family Picornaviridae is one of the largest RNA virus families [reviewed in ref. 2], containing 8 established and 6 proposed genera and at almost 33 species. Four genera contain clinically important viruses infecting humans, i.e. Enterovirus (which since recently also includes the human rhinoviruses), Parechovirus, Hepatovirus, and Kobuvirus. This family also contains economically important animal pathogens such as Foot-and-mouth-disease virus, a species belonging to the Aphthovirus genus. Other well-known animal pathogens are Theilovirus and Encephalomyocarditis virus (EMCV), two species that belong to the Cardiovirus genus and are associated with disease in rodents and swine. The Theilovirus species is represented by Theiler's murine encephalomyelitis virus (TMEV) and rat encephalomyelitis virus (also called Theiler's rat virus, TRV). TMEV is an enteric pathogen that primarily causes asymptomatic infections of the alimentary tract. However, extra-intestinal infection can occur and produce acute fatal encephalomyelitis or a chronic demyelinating disease relevant for multiple sclerosis, depending on the TMEV strain involved [3]. TMEV can furthermore cause serious foetal pathology and placental damage, depending on the gestational phase of infection [4].

The question of whether authentic human cardioviruses exist has remained unclear for a long time. A Theiler's-like cardiovirus, named Vilyuisk human encephalomyelitis virus (VHEV), has been implicated in an outbreak of a neurodegenerative disease among the Yakuts people in Vilyuisk, Siberia, in the 1950s. However, this virus, which was isolated upon multiple passages in mice and cell cultures, shows close relationship with TRV and TMEV, raising the possibility that it represents a contaminating animal cardiovirus [5]. Recently, however, strong evidence has been obtained for the existence of human cardioviruses. In 2007, the genomic sequence was reported of a cardiovirus, provisionally named Saffold virus (SAFV), amplified in cell culture from the stool of an infant presenting with fever of unknown origin in 1981 [6]. Following this discovery, SAFVs were detected by molecular techniques in nasopharyngeal aspirates of 3 children with respiratory symptoms in Canada [7], in stool samples from 6 paediatric patients suffering from gastroenteritis from Germany and Brazil [8], and in 1 respiratory secretion from a patient presenting with an influenza-like illness and 6 stool specimens from patients in a gastroenteritis cohort in California [9]. Recently, SAFVs were detected in stool samples from 6 South Asian children with non-polio acute flaccid paralysis (AFP), but also in stool samples from 5 children without overt neurological symptoms [10]. All SAFVs identified to date are significantly divergent from the documented animal cardioviruses, especially in the region encoding the structural capsid proteins that are important for receptor binding. Phylogenetic analysis suggests the existence of 8 distinct genetic SAFV lineages [10],[11]. Together, these findings suggest that SAFVs are genuine human viruses that can be sporadically found in fecal and respiratory specimens of children, with or without symptomatology, but that until recently have gone largely undetected most likely because of their fastidious growth [6],[7]. Thus, their relationship with clinical disease remains to be established.

The serological survey was performed with a SAFV-3 isolate that we isolated and characterized. Our SAFV-3 isolate is the first reported SAFV that efficiently replicates in cell lines and produces a clear cytopathic effect, thereby allowing to conduct a large-scale survey by means of a specific virus-neutralization assay. Using the assay, we showed that SAFV-3 is a highly common and wide-spread human virus causing infection in early childhood: the seroprevalence of SAFV-3 reaches high levels (>90%) in young children and adults. Nine months after birth, maternal antibodies have almost disappeared. In accordance, at nine months of age, many infants that we tested were seronegative and those that had antibody showed low titers which may still have been of maternal origin. A steep rise was found in 2-year-old children, both in seroprevalence and in titers, which indicates that infection occurs frequently and early in life. It is too early to state that SAFV infections occur before the age of 2 years because age of infection can be influenced by periodicity of outbreaks with a particular genotype (e.g. SAFV-3). The latter could explain why the Finnish children in 1997/98 showed a somewhat lower seroprevalence. Alternatively, the low population density in Finland compared to the Netherlands may have limited the speed of circulation of the virus. High rates of SAFV-3 antibody were furthermore found in African countries and in Indonesia, strongly supporting that SAFVs are ubiquitous human viruses with a global prevalence: SAFVs have now been reported from countries in North- and South-America, Europe, South-Asia and Africa [6]–[10, this study].

A high seroprevalence contrasts with the low numbers (<1%) of respiratory and fecal samples that tested positive by PCR, at least in developed countries [6],[7]. We have no explanation for this discrepancy. A possible explanation can be that virus is produced and excreted only for a very brief period, limiting the chance of its detection. However, the latter would also shorten the period of contagiousness and virus transmission which is not in line with ubiquity and infection early in life. The majority of SAFVs has been detected in feces suggesting transmission by the fecal-oral route. Fecal-oral transmission is generally associated with a prolonged fecal excretion typically found for enteroviruses, parechoviruses as well as rodent cardioviruses. Enteroviruses and parechoviruses are detected 5–10 times more frequently in feces compared to SAFVs in developed countries [20, our unpublished data] although their epidemiology is quite similar. A higher percentage of SAFVs (10% positive feces samples) was recently reported in children from Pakistan and Afghanistan [10]. This may be due to differences in sanitation, but this remains to be established.

Another aspect of consideration is the specificity of serology. Virus neutralization tests are highly specific and can discriminate serotypes within a single species [21]. From the phylogenetic analysis in this study it is unlikely that the neutralization assay with SAFV-3(NL2007) measured antibodies that cross-react with other genotypes, let it be with other species: The phylogenetic distinction between SAFV genotypes is largely based on differences within the capsid region. Known SAFV-3 strains show a high level of conservation (cave drift) in their VP1 protein (97–98% aa identity) but a significant level of divergence with that of SAFV-1 and -2 (68–73% aa identity), which is most probably due to immune selective pressure. The reported divergence is analogous to that observed in VP1 between serotypes of human enteroviruses, where nt differences in VP1 of more than 25% or aa differences of more than 12% are indicative of different serotypes [22]–[24]. The highest variability among the different SAFVs was observed in the VP1 CD loop and the VP2 EF loop which are located at the outer surface of the virion and implicated as important sites for binding of neutralizing antibodies. Given the large differences between their VP1 proteins, it is reasonable to assume that the 8 SAFV genotypes identified to date are serologically distinct. Accepting that the serological data in this study regard largely, if not entirely, one single type (SAFV-3), the contrast with sparse detection of SAFVs by PCR is even greater and additional studies are required to solve this discrepancy.

Our serological data are consistent with infection early in life which is corroborated by PCR data being mainly positive in children below 6 years of age [6]–[9]. In this age group, SAFVs may be a cause of respiratory disease [7],[9] and/or gastro-enteritis [8],[9], but the evidence is thin and the possibility of it being an innocent bystander in fecal samples is not excluded. Similarly, the relevance of SAFVs in the feces of non-polio AFP cases in South Asia remains to be established since SAFVs were also found in asymptomatic children [10]. Apparently, severe pathology due to SAFV infection is rare, at least in the young, and whether it occurs remains to be established. Importantly, not all adults that were tested in our serological study have passed SAFV-3 infection and a small proportion of them may thus still be susceptible (as shown in Fig. 5 for pregnant women). Moreover, a high seroprevalence early in life causes a severe constraint on further circulation of the virus, which can be overcome by subtle antigenic drift, as known for influenza viruses, resulting in re-infection at older age. Variation in the loop sequences - as observed in the SAFV-3(NL2007) isolate, which differs a single aa from other SAFV-3 strains in one of its VP1 loops (Fig. 3) - is in agreement with such a drift, but whether re-infections occur remains to be proven: Significantly higher antibody levels in adults, compared to children (Fig. 5), may suggest so but an increase in antibody levels may also be the consequence of infection by other members of the species (phenomenon of the original antigenic sin, OAS) [25]. Thus, one has to consider that adults may still be susceptible and may acquire infection under unfavorable conditions, e.g. during pregnancy when infection may be a threat for the fetus or newborn [4],[26],[27].

Although enteroviruses and parechoviruses mostly cause asymptomatic infections, they can cause severe pathology depending on the conditions of infection, which involve both host determinants (e.g. age of infection, genetic background) and virus determinants (e.g., serotype, virulence). Some good examples are provided by the human enterovirus C (HEV-C) species which includes coxsackie A viruses causing mild respiratory symptoms, as well as the polioviruses which can cause severe neurological disease and paralysis [2]. Virulence of a virus variant may depend on a single or a few amino acid substitutions as is illustrated by poliovirus type 3 Sabin, the vaccine strain, which differs from the parental strain in not more than 3 aa [28]. Similarly, SAFV as species may comprise a whole spectrum of variants, including variants that cause pathology dependent on the conditions of infection and the virus type and/or strain involved. For the two animal species of Cardiovirus there is robust evidence that such variations exists. A single point mutation, resulting in a change in amino acid within the VP1 capsid region, is sufficient to transform EMCV from wild type into a diabetogenic phenotype in mice [29]. TMEV strains are divided into two sublineages according to the pathology they cause in the central nervous system of susceptible mice. Highly neurovirulent strains (e.g. GDVII) cause an acute and lethal encephalomyelitis. Low neurovirulent strains (e.g. DA) cause a mild, transient encephalitis that is followed by viral persistence in the spinal cord white matter, resulting in a chronic inflammatory response and lesions of primary demyelination, highly similar to the signs and symptoms observed in Multiple Sclerosis [2],[3]. The difference in pathology caused by these TMEV strains reflects most likely a difference in neural tropism in the brains due to a difference in coreceptor usage [30]. Different SAFVs may also exhibit a difference in tropism. Consistent with this, we found a difference in tropism between our SAFV-3 isolate and a SAFV-2 that was isolated in our lab. This latter virus could be propagated on LLC-MK2 cells, albeit poorly, consistent with the growth pattern of SAFV-2 Can112051 on this cell type [7], but failed to replicate on HeLa or HELF cells (data not shown). The intriguing possibility of involvement of SAFV in MS awaits further investigation. The search for implications of SAFV infections is just starting and as depicted before, it may require testing for systemic infection, e.g. in blood, cerebrospinal fluid or biopsies of affected organs (e.g. brains). Results for cerebrospinal fluid specimens have been reported, but so far with a negative outcome [9]. Hence, the human cardioviruses are currently orphan viruses as were ECHO viruses in the past.

New molecular approaches have rapidly expanded the list of enteric viruses (e.g. SAFV, Cosa virus, Boca virus) for which pathology is currently less evident [31],[32]. Moreover, the increased sensitivity of nucleic acid-based techniques confronts the physician with a relatively new phenomenon of co-infection with 2, 3, or even more viruses at a time, making it hard to predict which of these is the culprit, if causing pathology at all. The latter holds not only for these new viruses but also for well known pathogens as enterovirus, parechovirus and adenovirus. Herewith, a picture emerges of a ‘viral flora’ quite similar to the microbiome of intestinal resident bacteria which are largely beneficial to the host, e.g. by competing out pathogenic invaders and playing an active role in shaping an intestinal immune barrier [33],[34]. A beneficial effect (cross-protection against bacterial invaders) has also been demonstrated for beta- and gamma-herpesviruses in mice [35]. Human cytomegalovirus has probably a similar effect as it changes the T-cell system dramatically, inducing a unique population of effector-memory CD8 T cells with innate-response features [36],[37]. Viruses, however, are -by nature- host-cell invasive microbes (which gut-resident bacteria are not) and a beneficial role of enteric viruses has not been investigated. Coxsackie B viruses, and probably also other enteric viruses, affect the tight junctions of the intestinal epithelial cells, which are the gatekeepers of the intestine, thereby increasing gut leakiness [38]. The latter is a direct cause of local inflammation, it may alter mucosal immunity in a beneficial way but leakiness has also been associated with chronic inflammatory disease, as type 1 diabetes and celiac disease [39]. Several lines of evidence indicate a role for coxsackie B viruses and other enteric viruses in type 1 diabetes in humans ,[reviewed in 40 and 41]. A possible role of SAFV in the development of this disease should be considered in future studies. In such studies, a role of different genotypes and, by analogy with EMCV, different strains belonging to one genotype should be taken into account.

Figure S1

Figure S2

Table S1