Bernhard Ehlers, Division 12 ‘Measles, Mumps, Rubella, and Viruses Affecting Immunocompromised Patients’, Robert Koch Institute, Nordufer 20, 13353 Berlin. e-mail: email@example.com
Since the discovery of Merkel cell polyomavirus and its causative association with Merkel cell carcinoma (MCC), six human polyomaviruses (HPyVs) have been identified that, so far, lack any disease association, which include the human polyomaviruses (HPyV) 6, 7, 9, 10 and 12 as well as the Saint Louis polyomavirus (STLPyV). PCR studies revealed that HPyV6 and HPyV7 are shed from the skin of healthy subjects and of patients suffering from various skin tumours. HPyV6, 7 and 9 were sporadically detected in body fluids and excretions of immunocompromised patients and healthy subjects. HPyV10 was identified in papillomavirus-induced anal condylomas, and variants of HPyV10, named MWPyV and MX polyomavirus (human) (MXPyV), as well as STLPyV were detected in faeces of diarrheal and healthy children. HPyV12 was discovered in organs of the digestive tract of patients suffering from various malignant diseases. Serological studies using capsomer-based or virus-like particle (VLP)-based enzyme-linked immunosorbent assay (ELISA) revealed that HPyV6, 7, 9 and 12 are circulating in the human population. As all other HPyVs, the novel polyomaviruses encode small and large T antigens and thus are potentially oncogenic. However, several studies have revealed a lack of association of HPyV6, 7 and 9 with numerous human tumours. In the future, it will be important to unravel the cell types and body compartments of the novel HPyVs′ reservoir and to search for possible associations with cancer and non-malignant diseases.
warts, hypogammaglobulinaemia, infections and myelokathexis
WU polyomavirus (human)
Discovery, Genomic Organization and Phylogenetic Analysis
Since the discovery of Merkel cell polyomavirus (MCPyV), the fifth human polyomavirus, in 2008 and its causative association with MCC, the interest of researchers in identifying formerly unknown human polyomaviruses (HPyVs) experienced a renaissance. Seven novel human polyomaviruses (HPyVs) have been discovered since then, namely the human polyomaviruses HPyV6, HPyV7, Trichodysplasia spinulosa-associated polyomavirus (TSPyV), HPyV9, HPyV10, Saint Louis polyomavirus (STLPyV) and HPyV12 [1-6]. Here, we describe and discuss what is known about the molecular and biological properties, the epidemiology and possible disease associations of these novel human polyomaviruses (except TSPyV).
The novel PyVs share a genome organization that is typical among polyomaviruses. The early region encodes the regulatory proteins small (sTag) and large Tumour antigen (LTag) and the late region encodes the structural proteins VP1, VP2 and VP3. These regions are separated by a non-coding control region (NCCR). An open reading frame encoding an agnoprotein, present in the human PyVs BK polyomavirus (human) (BKPyV) and JC polyomavirus (human) (JCPyV) , was not identified in the novel PyVs. In general, the genomes and proteins of HPyV6, 7, 9, 10, 12 and STLPyV display only limited sequence homology to those of the other human PyVs. However, their LTags have the domains known to be associated with oncogenic potential. A detailed comparison of the structural and functional properties predicted for the viral proteins of HPyV6, 7 and 9, was published recently . In combined phylogenetic analysis of LTag, VP1 and VP2 proteins, HPyV6 forms a clade with HPyV7, HPyV10 with MWPyV (and MXPyV), and HPyV9 with lymphotropic polyomavirus (LPyV; a polyomavirus of African green monkeys) and other non-human primate PyVs (Fig. 1). More distantly, they are all related to MCPyV and TSPyV (Fig. 1). The recently discovered STLPyV and HPyV12 are not included in the phylogenetic analysis shown in Fig. 1. The reader is referred to the phylogenetic analyses in the original reports, which uncovered a relationship between STLPyV and MWPyV/HPyV10  and revealed a discrete position in the phylogenetic tree for HPyV12 .
Detection in Clinical Samples with Nucleic Acid-Based Methods
HPyV6 and HPyV7 were first identified in skin swabs of healthy individuals. Initially, HPyV6 was detected by sequencing of a restriction fragment from rolling circle amplification (RCA) analysis. HPyV7 was then amplified with degenerate PCR targeting regions conserved between HPyV6 and the WU polyomavirus (human) (WUPyV). Complete HPyV6 and HPyV7 genomes were cloned from 5/35 (14%) and 4/35 (11%) individuals respectively  (Table 1). HPyV9 was discovered in 2011 in the serum of a kidney transplant patient using degenerate PyV PCR targeting the VP1 gene of the majority of known polyomaviruses. Re-screening of the analysed panel of clinical specimens with specific nested PCR revealed the presence of HPyV9 in four additional specimens (plasma and urine from kidney transplant patients, serum from a patient with leukoencephalopathy and whole blood from a patient suffering from acute myeloid leukaemia)  (Table 1). Quantitative real-time PCR (qPCR) revealed <20 copies/μL extracted DNA in all HPyV9-positive samples (N. Scuda and B. Ehlers, unpublished data). An HPyV almost identical to HPyV9 was later identified by high-throughput sequencing and subsequent specific PCR in skin swabs of 2/8 patients with MCC and 1/111 healthy volunteers. It was named IPPyV strain of HPyV9  (Table 1). The tenth human polyomavirus, HPyV10, was identified in 2012 with RCA and PCR in condylomas of a patient suffering from warts, hypogammaglobulinaemia, infections, and myelokathexis (WHIM) syndrome . Two very similar PyVs of likely human origin were discovered in the same year. MW polyomavirus (human) (MWPyV) was initially identified by shotgun pyrosequencing of DNA purified from virus-like particles (VLP) in the faeces of a healthy child from Malawi (MW). With PCR methods, a PyV with 95% identity was then detected in stool samples from children with diarrhoea in the US . With deep sequencing and PCR, MXPyV, being more than 99% identical to the US strain of MWPyV and to HPyV10, was detected in stool samples of children with and without diarrhoea from Mexico (MX), Chile and California  (Table 1). Most recently, STLPyV and HPyV12 were discovered. Two variants of STLPyV were identified with RCA and shotgun 454 pyrosequencing in stool samples of children from Malawi and the US respectively . HPyV12 was detected in organs of the gastrointestinal tract of German individuals suffering from various malignant diseases .
Table 1. HPyV DNA prevalence in body fluids, excretions and skin swabs
834 stool samples, mainly from children, with or without diarrhoea; 136 nasal washes from children with pneumonia; 480 plasma and urine samples from IS pat. (patients' age not reported)
qPCR and qRT-PCR for HPyV10 isolate MXPyV; Less frequent in 96 stool samples of children with diarrhoea (0%) than in 96 age- and sex-matched controls without diarrhoea (4.2%); persistent stool shedding (91 days) reported in 1 child
56 stool samples from patients with diarrhoea; 99 plasma/serum samples from TX recipients; 152 urine samples from IS patients with various diseases; 30 oral fluids from healthy individuals; 22 BAL from patients with pneumonia; 25 CSF from patients with leukoencephalopathy (patients' age not reported)
Since the first reports on these novel PyVs, a number of studies were conducted to evaluate their occurrence in healthy subjects and in immunocompetent and immunosuppressed patients suffering from a variety of diseases (Table 1). By high-throughput metagenomic sequencing, HPyV6 was found in skin swabs (normal-appearing forehead skin) from three of six individuals, and HPyV7 and HPyV9 were detected in one skin swab  (Table 1). We have recently analysed over 300 skin swabs from different groups of healthy volunteers (who were free of skin diseases) by type-specific qPCRs and detected HPyV6, 7 and 9 DNA in 28.7%, 13.5% and 0.3% of the swabs respectively. Consecutively collected swabs were available from over 100 healthy individuals and persistent cutaneous infections could be shown in 24% of the volunteers for HPyV6 and in 10% for HPyV7. So far, persistent HPyV9 infections have not occurred (U. Wieland, A. Kreuter, and S. Silling, unpublished data). To analyse the possible reactivation of HPyV9 (and three other HPyVs) due to immunosuppression during pregnancy, prevalence was determined with PCR in plasma, urine and respiratory samples (n = 100 each) from both pregnant and non-pregnant women. HPyV9 was detected in 2–3% of the samples of pregnant and 2–6% of the samples of non-pregnant women, thus giving no evidence of pronounced reactivation of HPyV9 during pregnancy . The presence of all known human polyomaviruses (with the exception of HPyV10, MWPyV, MXPyV, STLPyV and HPyV12 that were discovered later) was investigated with qPCR in 716 clinical specimens of 32 children receiving a transplant (2 lung, 11 liver, 5 heart, 2 kidney, 1 liver/lung and 11 bone marrow transplants). HPyV6 was detected in the faeces of a lung and a bone marrow transplant recipient and in a nasopharyngeal swab of a heart transplant patient; HPyV7 was found in a nasopharyngeal swab of a liver transplant patient and HPyV7 and HPyV9 in a urine sample of 2 liver transplant patients respectively . MWPyV and STLPyV were detected with PCR in 2.2% and 1.1%, respectively, of stool samples collected from children with diarrhoea in Saint Louis, USA . In the same study, 0.3% (1/337) of urine samples collected from adult renal transplant recipients were PCR-positive for STLPyV; MWPyV was not detected. Stool, plasma and nasopharyngeal swabs of the transplant recipients were negative for both viruses  (Table 1). In a case–control study of children with diarrhoea, stool samples were analysed with PCR targeting large T antigen sequences of both MWPyV and STLPyV. MWPyV was detected in 1.5% (5/327) and in 1.3% (5/389) of cases and controls respectively. STLPyV was only detected in 0.3% (1/389) of controls  (Table 1). The prevalence of HPyV12 was analysed with nested PCR in 636 samples of organs, body fluids and excretions of transplantation donors and patients suffering from different diseases. HPyV12 was detected in 11% (14/124) of liver samples and in one sample each of colon, rectum and faeces .
In summary, HPyV6 and HPyV7 are shed from normal skin of healthy subjects and of patients suffering from various skin tumours (see below), and DNA of both viruses is sporadically detected in body fluids and excretions. HPyV9 was identified both in immunodeficient patients and in healthy subjects. Various body fluids and excretions, and rarely also skin samples, were HPyV9-positive. HPyV12 was detected in organs of the digestive tract and also in stool (Table 1). So far, a tropism for other body compartments was not reported for HPyV6, 7, 9 and 12. Since the discoveries of HPyV10 in anal condylomas and MWPyV, MXPyV and STLPyV in faecal samples, additional studies on the occurrence of these novel PyVs in healthy subjects or in diseased individuals have, to our knowledge, not been conducted. Sensu stricto, the classification of MWPyV, MXPyV and STLPyV as human viruses awaits further confirmatory studies on sample materials other than stool as well as serological surveys.
Seroepidemiology of the Novel Hpyvs
Data on seroepidemiology of the novel HPyVs are summarized in Table 2. The most common method for studying seroprevalence of polyomaviruses is enzyme-linked immunosorbent assay (ELISA) using VP1 capsomers or VP1-based virus-like particles (VLPs) as the capture antigen [e.g. [15-23]]. Of these two possibilities, VLP-based ELISA is likely more sensitive as VP1 VLPs morphologically resemble mature infectious particles and thus present conformational epitopes . Therefore, the use of capsomers or VLPs is noted for the studies specified in Table 2.
Judging from seroprevalence studies available for HPyV6, 7, 9 and 12, exposure to these viruses seems to occur frequently in humans (Table 2). Data on HPyV10, MXPyV, MWPyV and STLPyV serology have not been published so far. HPyV6 and HPyV7 seroprevalence was investigated in 2 studies (Table 2): Schowalter and colleagues found antibodies against VP1 of HPyV6 and HPyV7 in 69% and 35% of blood donors (n = 95), respectively, compared with 69% for MCPyV. Twenty-eight percent of the blood donors had antibodies against both HPyV6 and HPyV7, and 16% against all three cutaneous PyVs . In a very recent study on the seroprevalence of HPyV6, HPyV7, HPyV9, TSPyV and MCPyV, HPyV6 antibodies were detected in 37.5% of 1- to 4-year-old children. Their prevalence rose to 67% in 30- to 39-year-old adults and to 98% in people 80 years and older. Seroprevalence for HPyV7 was lower than for HPyV6. Only 10% of the 1- to 4-year-old children were seropositive. In adults 20 years or older, HPyV7 seroprevalence increased with age from 45 to 86% .
HPyV9 seroprevalence has been analysed in three studies that delivered similar results [15, 25, 26] (Table 2). Exposure to HPyV9 probably occurs early in life, as immunoglobulin G (IgG) antibodies against VP1 of HPyV9 have been detected in 1- to 4-year-old children [15, 25, 26]. In all three studies, HPyV9 seroprevalence increased with age. Trusch et al.  found a peak HPyV9 seroprevalence of 53% in the age group between 21 and 30 years, followed by a slight decline in older individuals (to 35% in persons above 60 years of age), while Nicol et al. found in their earlier study the highest HPyV9 seroprevalence (42%) in the oldest age group (51–85 years) . Later Nicol et al. reported 41% for 60- to 69-year-old individuals and a peak HPyV9 seroprevalence of 70% in ≥80 year-old adults . Nevertheless, prevalence rates described in the three studies for the different age groups were in the same range respectively [e.g. 13% in 2- to 5-year-old  compared with 10% in 1- to 5-year-old children [25, 26]]. Peak HPyV9 seroprevalences were below those reported for BKPyV, JCPyV, KIPyV, WUPyV, MCPyV, HPyV6 and TSPyV in most studies, indicating a less frequent distribution of HPyV9 in the human population compared with other HPyVs, with the possible exception of HPyV7 [lower than HPyV9 in one study , higher than HPyV9 in another study ]. Nicol et al.  described a more than twice as high HPyV9 seroprevalence in men compared with women. This is in contrast to Trusch et al.  who found no difference in HPyV9 seroprevalence of adult males and females (Table 2).
Trusch et al.  have additionally analysed sera of immunosuppressed patients, and could detect significantly higher HPyV9 seroprevalences and IgG titres in renal and in haematopoietic stem cell transplant recipients compared with age-matched healthy controls. This effect could not be observed in liver transplant recipients and in patients with neurological symptoms (Table 2). The severity of immunosuppression is probably associated with an increased risk for HPyV9 primary infection, reinfection or reactivation. Most of the haematopoietic stem cell transplant recipients did receive blood transfusions or immunoglobulins, however. Therefore, the passive transfer of HPyV9 antibodies could be an additional explanation for the increased HPyV9 seroprevalence observed in this patient group .
Two studies on HPyV9 serology [15, 26] further showed that the previously reported reactivity of human sera against LPyV [18, 27] can probably be explained by cross-reactivity of HPyV9 antibodies with LPyV VP1 (HPyV9 and LPyV VP1 share 87% amino acid identity) [15, 26]. In one immunosuppressed patient with documented HPyV9 infection 837 days after transplantation, the presence of LPyV (and of human PyVs other than HPyV9) could be excluded by PCR analyses . This patient also served to show that IgM antibodies against HPyV9 are detectable in the course of primary HPyV9 infection, followed by IgG antibodies shortly thereafter. HPyV9-DNA was only detectable for a few days in serum samples. IgM antibody levels declined as PCR became negative, while IgG levels and IgG avidity increased within a period of over 200 days to reach a plateau that was constant for up to 1.5 years .
Seroprevalence of the most recently detected human polyomavirus, HPyV12, was 12 % in small children and 27 % in young adults. In older adults, prevalence rates ranged between 15% and 33%. This indicates a lower overall seroprevalence in comparison to the other human polyomaviruses. A difference in HPyV12 seroprevalence between male and female adults was not observed  (Table 2).
Only two studies examined both seroprevalence and DNA detection rates of HPyVs, but not in the same individuals respectively [2, 4].
Disease Association of the Novel Hpyvs
While the human polyomaviruses JCPyV, BKPyV, TSPyV and MCPyV are associated with disease in immunosuppressed and MCPyV also in immunocompetent MCC patients [28, 29], no disease association could be established for HPyV6, 7, 9, 10, 12 and STLPyV so far. In contrast to HPyV9, HPyV6 as well as HPyV7 seem to be a regular part of the human skin microbiome (Table 1) [4, 9, 12]. Therefore, their association with different skin tumours and some non-tumorous skin lesions has been investigated in several studies. MCC was analysed with PCR for the presence of HPyV6 and HPyV7. In none of 28 MCC cases (22 MCPyV-positive and 6 MCPyV-negative) was DNA of either virus detectable, arguing against a role of these viruses in the genesis of MCPyV-negative (and MCPyV-positive) MCC . Skin cancers other than MCPyV-positive MCC were also analysed for the presence of HPyV6 and HPyV7. A total of 108 samples consisting of 21 squamous cell carcinomas (SCC), 18 basal cell carcinomas (BCC), 20 melanomas, 20 MCPyV-negative MCC, 12 cutaneous T-cell lymphomas (CTCL), and 17 cutaneous B-cell lymphomas (CBCL) were tested with qPCR. Very low copy numbers (<1 copy per cell) of both viruses were found in 14% (HPyV6) and 2% (HPyV7) of the specimens, thus giving no evidence for a role of these viruses in the analysed skin cancers . The presence of HPyV6, HPyV7, TSPyV and HPyV9 was analysed with qPCR in 193 non-melanoma skin cancers (41 BCC, 31 actinic keratoses, 8 SCC in situ, 52 SCC, 42 keratoacanthomas, 5 microcystic adnexal carcinomas, 14 atypical fibroxanthomas). HPyV6 DNA was found in 7.3% of BCC, 3.2% of actinic keratoses, 12.5% of SCC in situ, 3.9% of SCC, 4.8% of keratoacanthomas, and 0% of microcystic adnexal carcinomas and atypical fibroxanthomas. HPyV6-DNA loads were consistently very low. None of the 193 samples was positive for HPyV7 or HPyV9 DNA. These findings argue against a pathogenic role of HPyV6, HPyV7 and HPyV9 in the analysed types of non-melanoma skin cancer .
It has been speculated by some authors that HPyVs, more precisely MCPyV, could play a co-pathogenic role – together with human papillomaviruses (HPV) – in the development of cutaneous SCC or warts . However, viral loads are usually very low in these tumours, MCPyV LTag is not expressed, and detection rates of MCPyV DNA are not higher than in healthy skin [32, 34-36]. Furthermore, in contrast to HPV DNA, MCPyV DNA was less common in premalignant and malignant anogenital HPV-induced lesions of HIV-positive men than in normal samples . Thus, a co-tumorigenic role of HPyV and HPV seems unlikely at the moment. Specimens of 43 neuroendocrine tumours (NETs) were analysed with multiplex PCR based on a fluorescent magnetic suspension assay for the presence of 8 human PyVs including HPyV6 and HPyV7, and a variety of other human viruses. While 2/3 MCCs were found to be MCPyV-positive, no other human PyVs were detected in the complete NET panel, thus negating their causative role in the analysed NETs that included malignancies of the brain, thymus, gastrointestinal tract, lung, skin (MCC) and thyroid gland . Archival biopsy specimens (n = 130) from 83 patients with cutaneous B-cell lymphomas and CTCL were analysed with qPCR for the presence of MCPyV, HPyV6, HPyV7 and TSPyV. The investigated cutaneous lymphomas comprised classic mycosis fungoides (MF), MF variants as folliculotropic MF and pagetoid reticulosis, Sezary syndrome, primary cutaneous CD30+ lymphoproliferative disorders, several rare CTCLs, and different types of CBCLs . Only low MCPyV loads were found in 30 samples (23%) and even lower viral loads were detected for HPyV6 and HPyV7 in six (4.6%) and one (0.8%) sample, respectively. Therefore, an aetiological role of HPyV6 and HPyV7 (as well as of MCPyV and TSPyV) in CBCL and CTCL seems to be very unlikely . A recent analysis of CTCL by high-throughput sequencing of total extracted RNA confirmed the absence HPyVs in these malignancies. No known viral transcripts could be detected in the investigated six CTCL that comprised 3 MF and 3 Sezary syndrome lesions . Furthermore, a causal role of HPyV6, 7 and 9 (as well as of MCPyV and TSPyV) could be excluded for dermatofibroma, a common benign skin neoplasm. None of the 20 dermatofibroma biopsies analysed by qPCR carried HPyV7, TSPyV or HPyV9. HPyV6 and MCPyV DNA were found in two samples (10%), respectively, but again viral loads were very low (<0.003 HPyV6 DNA copies per betaglobin gene copy) (A. Kreuter, B. Danby, and U. Wieland, unpublished data). Similarly, neither HPyV6, HPyV7 and HPyV9, nor MCPyV or TSPyV could be detected in several lesional cutaneous biopsies of a patient with paraneoplastic ancanthosis nigricans . Finally, HPyV6 and HPyV7 were both detected with qPCR in low copy numbers in one (1.9%) of 54 breast cancer biopsies . HPyV9 and MCPyV were studied with qPCR in Japanese cases of chronic lymphocytic leukaemia (CLL) and healthy blood donors. HPyV9 was not detected in any of the blood samples (Table 1), and extremely low levels of MCPyV were obtained in 9 CLL cases. Thus, both viruses did not seem to be involved in the leukemogenesis of CLL .
In summary, HPyV6 and HPyV7 DNAs have been detected in a small fraction of cutaneous and non-cutaneous benign and malignant tumours. In the few positive samples, viral loads were consistently far below one viral copy per cell equivalent, making an aetiologic role of HPyV6 and HPyV7 in the investigated diseases very unlikely. HPyV9 DNA was not found in any of the tumours analysed.
As mentioned above, HPyV10 was isolated from HPV6-positive perianal warts (condylomas) of an immunocompromised patient with WHIM syndrome . Human papillomavirus 6 (and other low-risk HPVs) typically causes condylomas and HPV6 was more abundant than HPyV10 in the investigated lesions. Therefore, HPyV10 probably did not cause the perianal warts from which it was isolated, but presumably can be considered another HPyV with cutaneous tropism. HPyV10 isolates have also been detected in stool samples from healthy persons and from patients with diarrhoea (Table 1) [3, 10, 11]. In one study, detection rates in stool samples of patients with diarrhoea (0%) were lower than those of sex- and age-matched controls without diarrhoea (4.2%) and no association between the presence of HPyV10 isolate MXPyV and diarrhoea and/or vomiting was found . In another case–control study that included over 700 small children, the HPyV10 isolate MWPyV was similarly infrequently found in cases (1.5%) and controls (1.3%) . Therefore, HPyV10 probably does not constitute a novel gastroenteritis-causing pathogen. The same seems to be true for STLPyV, which has only occasionally (0.3%) been found in stool samples of controls, and in none of the cases with diarrhoea  (Table 1).
Finally, HPyV12 has been detected in organs of the digestive tract of patients suffering from various malignant diseases. Due to the relatively small number of positive samples and the lack of a case–control study, HPyV12 can presently not be associated with any malignant disease. Furthermore, HPyV12 may have a reduced transforming potential compared with other human polyomaviruses, because it is presently the only HPyV lacking a binding site for the retinoblastoma protein on its LTag .
Further studies are needed to clarify or exclude a pathogenic role of these potentially oncogenic novel HPyVs, to determine their cellular receptors, their mode of transmission, their distribution in healthy and immunocompromised individuals, and their potential sites of persistence. Mainly missing are serological studies on HPyV10 (and its variants) and STLPyV as well as further studies on HPyV6, HPyV7, HPyV9 and HPyV12 serology that comprise children, chronically ill patients and especially immunocompromised individuals.
In situ detection of the novel HPyVs by hybridization or immunohistochemistry of pathological organ biopsies compared with normal control biopsies could help to elucidate if they potentially play a pathogenic role in humans or if they are just innocent bystanders.
The authors thank Sébastien Calvignac for phylogenetic analysis, Ugo Moens for critical reading of the manuscript and Zebulon Tolman for proofreading our manuscript (as a native speaker). UW was supported by the German Federal Ministry of Health, grant no. 1369-401.