Twenty years of human bocavirus research: from an unculturable virus of unclear pathogenicity to a culturable human pathogen and gene therapy vector candidate

SUMMARY

Twenty years after the first description of human bocavirus 1 (HBoV1) as a respiratory pathogen, significant progress has been made in both clinical and basic research; however, important clinical, diagnostic, and molecular challenges remain before bocavirus pathobiology is fully understood. The discovery of HBoV1 and its notorious prolonged shedding have challenged the new sensitive multiplex PCR panel-based diagnostic testing that replaced the old antigen assays, leading to erroneous classification of HBoV1 as an innocent bystander. Both sophisticated diagnostics and cytopathic effects in cell culture have now confirmed HBoV1 to be a common cause of upper and lower respiratory tract infections, mainly in children. While many questions have been answered, new questions have emerged as our understanding of parvoviruses has significantly expanded over the past two decades. In this review, key findings from 20 years of clinical, basic, and applied research on human bocaviruses are summarized and open questions highlighted to guide future investigations.

INTRODUCTION

Several new respiratory viruses were discovered in the early years of the millennium, such as the human metapneumovirus in 2001 (1), SARS-coronavirus-1 in 2002/2003 (2), and human coronavirus NL63 in 2004 (3). The application of molecular techniques contributed to the rapid and exact characterization of these viruses, but initial detection was still enabled by traditional virus culture. Among those new pathogens was the human bocavirus (HBoV) 1, described in 2005, which is the first example of a virus that was identified by a shotgun DNA-amplification and sequencing approach without preceding cell culture (4). The virus was identified in respiratory tract samples of infants and young children seeking hospital care for symptoms of lower respiratory tract infection (RTI) and has since been associated with both upper and lower RTI in children. Three related human bocaviruses (HBoV2-4) and a gorilla bocavirus were later discovered in stool and seem to be enteric (5-8).

Despite substantial evidence for the causative role of HBoV1 in mild to severe respiratory diseases (9-12), the general awareness of the virus is still insufficient, and therefore, its clinical significance may often be called into question due to its standing as an innocent bystander. The reason for this misinterpretation and negligence is the long-term persistence of HBoV1 DNA in the respiratory tract after the acute primary infection, which leads to clinically false PCR results and thereby “co-infections” with other respiratory pathogens (13-16). Moreover, the majority of published studies on HBoV1 were performed with inadequate diagnostic methods, such as qualitative PCR, which cannot distinguish between an acute infection and persistent shedding. To circumvent this issue, an endonuclease treatment prior to PCR can be conducted to reveal if the DNA originates from an acute infection or from persistent shedding due to the fact that the HBoV1 genome is protected by a capsid in the acute phase (17). Like many other newly discovered viruses, HBoV1 lacks a proper animal-host model (12, 18). Besides being a relevant pathogen in children, HBoV1 attracted great interest regarding molecular biology and gene therapy research. First and foremost, it is the only autonomous human parvovirus that can be productively cultured in cell culture. Second, these culture experiments have enabled a number of new insights into the viral DNA replication mechanisms, virus-host interactions, and pathogenicity. Third, the in-depth characterization of the virus has paved the way for the utilization of HBoV1 as a novel vector for respiratory tract-specific human gene therapy. These issues and developments, as well as controversial aspects and future challenges, will be critically discussed in this review.

CLASSIFICATION

Parvoviridae is a virus family of over 240 species, 30 genera, and 3 subfamilies (Table 1), comprising viruses with non-enveloped icosahedral capsids of 20–25 nm, surrounding single-stranded (ss) linear DNA genomes of ~5,000 bases with short double-stranded hairpin termini (19). The taxonomy of parvoviruses follows the amino acid (aa) sequence of the largest nonstructural (NS) protein, NS1. Viruses belonging to the same species share over 85% and those in the same genus at least 35%–40% aa sequence within NS1 (19). The human bocaviruses (HBoV1-4) belong, together with many animal bocaviruses, to the genus Bocaparvovirus in the subfamily Parvovirinae of the family Parvoviridae (Table 2). The genus Bocaparvovirus is named after the bovine parvovirus (BPV1) and the minute virus of canines (MVCs), whose hosts defined the genus name. HBoV1 and HBoV3, as well as the gorilla bocavirus (GBoV), belong to the species Bocaparvovirus primate1, whereas HBoV2 and HBoV4 belong to the species Bocaparvovirus primate2; all species are named according to the binomial nomenclature introduced by the International Committee on Taxonomy of Viruses in 2020 (19, 20).

TABLE 1.

Taxonomy of the family Parvoviridae with current subfamilies and genera^a

Densovirinae	64	Hamaparvovirinae	42	Parvovirinae	136
Aquambi *	24	Brevi #	3	Amdo ¶	11
Blattambi *	6	Chap #	36	Arti ¶	1
Diciambi *	3	Hepan #	1	Ave ¶	11
Hemiambi *	3	Ichtachap #	1	Boca ¶	36
Itera *	7	Penstyl #	1	Copi ¶	10
Miniambi *	1			Dependo ¶	27
Musco *	2			Erytho ¶	7
Pefuambi *	2			Lori ¶	1
Protoambi *	7			Proto ¶	23
Scindoambi *	6			Sande	1
Tetuambi *	3			Tetra ¶	8

^aGenus names continue with *densovirus, ^#hamaparvovirus, and ^¶parvovirus. Numbers refer to the number of species, in total 243, including the floating genus Metalloincertoparvovirus with one species. For a complete list of species names and future updates, see https://ictv.global/taxonomy.

TABLE 2.

Classification of human and the most known other vertebrate parvoviruses in the 11 genera of the subfamily Parvovirinae^a

Bocaparvovirus	Protoparvovirus	Erythroparvovirus	Tetraparvovirus	Dependoparvovirus
Human bocavirus 1	Human bufavirus	Human parvovirus B19	Human parvovirus 4	Adeno-associated viruses
Human bocavirus 2-4	Human cutavirus	Simian parvovirus	Porcine hokovirus	Duck parvovirus
Gorilla bocavirus	Minute virus of mice	Bovine parvovirus 3	Bovine hokovirus	Goose parvovirus
Bovine parvovirus 1	Canine parvovirus		Ovine hokovirus
Canine minute virus	Feline panleukemia virus
Porcine bocavirus	H-1 parvovirus
	Kilham rat virus
	Porcine parvovirus 1
Amdoparvovirus	Aveparvovirus	Copiparvovirus	Sandeparvovirus	Loriparvovirus
Aleutian mink disease virus	Chicken parvovirus	Bovine parvovirus 2	Zander parvovirus	Slow loris parvovirus
Skunk amdoparvovirus	Turkey parvovirus	Porcine parvovirus 4
		Equine parvovirus-hepatitis
Artiparvovirus
Bat artiparvovirus

^aThe five genera in the first row include human viruses. Some viruses belong to the same species. For the complete list, official binomial species names, and future changes in the taxonomy, see https://ictv.global/taxonomy.

Typical for bocaviruses are heterotelomeric hairpins, >90% negative polarity of the ssDNA genome (21, 22), and the unusual 25 kDa nuclear protein NP1 encoded in the middle of the genome (23). HBoV1-4 and gorilla bocavirus were predicted to express the bocaviral small non-coding RNA called BocaSR, encoded in the far right-hand end (23, 24). Other members in the subfamily Parvovirinae include, for example, the human parvovirus B19 (B19V) of the Erythroparvovirus genus, the adeno-associated viruses (AAVs) of the Dependoparvovirus genus, the Aleutian mink parvovirus of the Amdoparvovirus genus, as well as the human cutavirus and bufavirus, canine parvovirus (CPV), and minute virus of mice (MVM) of the Protoparvovirus genus (Table 2) (19). In the prioritization for pandemic preparedness by the World Health Organization in 2024, the carnivore protoparvoviruses (CPV as model virus) were listed as both the parvovirus “Prototype Pathogen” and “Pathogen X,” and the carnivore amdoparvoviruses and primate erythroparvoviruses as “Viruses of Concern” (25). Currently, there are 11 genera in the subfamily Parvovirinae, comprising vertebrate viruses of 136 species (Tables 1 and 2). Furthermore, there are two other subfamilies in this family, called Densovirinae and Hamaparvovirinae, including non-vertebrate and a mix of vertebrate and non-vertebrate viruses, respectively. The Parvoviridae family belongs to the order Piccovirales, class Quintoviricetes, phylum Cossaviricota, kingdom Shotokuvirae, and realm Monodnaviria (the highest taxonomic rank corresponding to the three domains of life).

EPIDEMIOLOGY

Most studies on HBoV1 have been performed in hospital settings, in particular, in pediatric emergency wards (26), but a few cohort studies were also published (15, 27-29). In all geographic locations investigated, HBoV1 can be found in infants and young children with symptoms of respiratory tract infection (26, 30). Thus, the virus is considered endemic worldwide. The virus is rarely detected in older children or adults, suggesting that natural infection induces long-lasting protective immunity. Seroprevalence studies show a rapid increase in seropositivity rates in early childhood, reaching 80% in children by the age of 6 when applying competitive IgG enzyme-linked immunoassays (EIAs) (blocking cross-reactive HBoV2-4 antibodies), while the seroprevalences of HBoV2-4 were 50%, 10%, and 0% (31-33). Notably, there is an incidence peak around 18 months of age, but the incidence is lower in children younger than 18 months (34, 35), suggesting protection from maternal antibodies, followed by a rapid acquisition of the virus after maternal antibody loss (31-33). In adults from China, the specific seroprevalence of HBoV1 was 67% (36).

No studies proving routes of infection are available yet, but it is reasonable to assume that HBoV1 can be transmitted via respiratory droplets or aerosols like other respiratory viruses. The high incidence and rapidly increasing seropositivity in the first years of life indicate that transmission is highly efficient.

Clearance of HBoV1 DNA from the upper respiratory tract after primary infection is often slow, and prolonged detection of HBoV1 DNA in respiratory tract samples for several weeks or months is common (14, 15, 27, 28). In combination with the narrow age span of primary infection, this makes HBoV1 DNA a prevalent finding in healthy young children or children with other infections, in particular, during the second year of life (15, 35). Qualitative detection of HBoV1 DNA is therefore a poor diagnostic test for diagnosing acute HBoV1 infection, and most published studies on the epidemiology of HBoV1 suffer from this limitation (13). For the same reason, there is very limited reliable information on the seasonality of HBoV1 infections. In the pediatric hospital context, detection of HBoV1 DNA in infants and young children is mainly driven by the number of samples tested, which in turn is driven by the seasonality of other respiratory viruses. There appears to be limited seasonal variability of HBoV1 infections when taking the sampling bias into account (15, 35). However, a recent study indicates that primary HBoV1 infection nevertheless could have an incidence peak in early winter (37). Moreover, HBoV1 re-emerged rapidly immediately after the COVID-19 pandemic peak despite the fact that optimized protection measures were already implemented, as shown for Yucheng, which is located in the central Chinese province Henan (38).

CLINICAL FEATURES AND RARE EVENTS

It is generally accepted that HBoV1 causes respiratory tract infections, mainly in children aged less than 5 years (39, 40). Despite a vast number of cross-sectional clinical studies using HBoV PCR or quantitative (q)PCR, only a few have used more sophisticated approaches for the diagnosis of HBoV infections such as serology, mRNA or antigen detection, viremia, or repeated/advanced sampling of broncho-alveolar lavage fluid, middle ear fluid, induced sputum, or tonsil tissue samples (13, 15, 27, 29, 41-52). Collectively, the data show that HBoV1 infection causes similar signs and symptoms as many other respiratory viruses in children. The most frequent of these are cough, fever, rhinitis/nasal congestion, and dyspnea/wheezing, and less often pharyngitis, rashes, red eyes, and diarrhea. In addition, adults have commonly experienced sore throat, fatigue, and headache (39, 40). It is remarkable that a need for oxygen therapy or hypoxia may occur in up to 30%–40% of hospitalized cases (53, 54). The most common diagnoses include common cold, acute otitis media, pneumonia, bronchiolitis, acute wheezing, and asthma exacerbations. In a recent study, including 1,465 hospitalized children (median 3 years of age) with respiratory tract infections, nasopharyngeal swab samples were analyzed with multiplex PCR for 22 respiratory viruses and bacteria, including HBoV1, SARS-CoV-2, influenza virus, and respiratory syncytial virus (10). HBoV1 single infections were identified in 28.3% of cases in the aforementioned pediatric cohort, in which the need for intensive care and mechanical ventilation was most frequently observed. HBoV1 infections also exhibited the highest mortality among all the analyzed pathogens. As expected, given the case severity in the respective settings, lower RTI dominates in hospital materials and upper RTI in community-based materials (15, 27, 42, 55). In controlled hospital-based studies, statistical associations were predominantly found between symptoms of lower RTI and the presence of HBoV1 DNA in airway samples (55-57) and acute HBoV1 infection (high viral load, viremia, or mRNA) (13, 35, 56, 57). Population-based studies additionally showed associations with symptoms of upper RTI such as cough, rhinorrhea, and acute otitis media (15, 33). Furthermore, out of 258 children with RTI (wheezing), 48 (19%) had an acute HBoV1 infection by serology (and PCR in airways or serum); of all children with bronchiolitis, 25% had acute HBoV1 infection (13, 35, 56, 57). The severity of HBoV1 infections can be comparable to that of respiratory syncytial virus (RSV) infections. Bronchiolitis seems to be more common among RSV-infected children, whereas pneumonia is more frequent among HBoV1-infected children (13, 16, 41, 58, 59), and even life-threatening HBoV1 infections were reported (9, 34, 60-67). Furthermore, HBoV1 is capable of exacerbating chronic pulmonary diseases like asthma, chronic obstructive pulmonary disease (COPD), and cystic fibrosis (CF) (68, 69).

Chest radiographies frequently show peribronchial or interstitial infiltrates, hyperinflation, or atelectasis (41, 42, 56, 70). In one study, interstitial infiltrates were seen in 75% of children hospitalized with lower RTIs associated with serologically confirmed HBoV1 infections (71).

In adults, HBoV1 is, in principle, detectable in nasopharyngeal aspirate (NPA) and/or blood samples most often with RTI and less commonly without RTI, compared with children (57, 72-80). One study found 1 viremic patient out of 111 (1%) HBoV1 IgG-positive and IgM-negative children (13, 35, 56, 57). HBoV1 infections generally seem to be rare in adults, with detection rates around 1% in most larger studies (11, 57, 72-76). In contrast, one study reported a rate of 7% (81), demonstrating that the frequency of symptomatic versus asymptomatic infections in adults is still unknown.

The other human bocaviruses, HBoV2-4, have also been detected—almost exclusively—in stool samples, but their causal role in gastrointestinal disease is uncertain, with HBoV2 being the most common finding. This goes ahead with the fact that gastroenteritis and diarrhea in children are the most commonly reported conditions in context with HBoV2-4 detections, but equal detection rates were found in asymptomatic children (82-85). Studies on viral load and serology failed to show associations with gastrointestinal disease (82, 86). HBoV1 was also detected in feces, but this probably represents intestinal passage of viruses shed from the respiratory tract (82). Therefore, further studies on serology, viral load, and parameters for viral gene expression are necessary.

A few case reports were published describing encephalitis, hepatitis, and myocarditis in patients with HBoV1, 2, or 3 as the only potential trigger (87-93). Especially the encephalitis cases in children and adults are intriguing since HBoV1, and in one case HBoV2, were the only infectious agents detected in cerebrospinal fluid (87, 89, 90). Gastroenteritis, meningitis, and encephalitis have also been described in immunosuppressed subjects and transplant recipients (94, 95). Thus, HBoV1 and HBoV2 could be among the viruses triggering encephalitis, thereby reducing the number of cases without diagnosed causative agents. However, these cases are sparse, and the causal relations between the viruses and the respective diseases are highly uncertain. Probably, due to high HBoV antibodies in adulthood, primary infections are extremely rare in pregnancy. HBoV infection was not detected in stillborn children or hydrops fetalis (96, 97). HBoV1 DNA was detected in tonsillar tissue in a few studies, without a clear association with symptoms, as discussed below (98-101).

While a causal link between HBoV1 and respiratory disease was reported in a number of studies (13, 35, 56, 71, 102, 103), the exact clinical picture awaits determination. A vast number of papers reported HBoV1 in the context of acute respiratory illnesses, including common cold, asthma, acute wheezing, bronchiolitis, pneumonia, acute otitis media, and even plastic bronchitis (16, 34, 56-58, 66, 70, 103-129). It is not possible to clinically differentiate between respiratory tract infections caused by different viruses such as rhinovirus, RSV, human metapneumovirus, influenza virus, and HBoV1, or even bacteria (130). Nevertheless, hypoxia and neutrophilia were more severe in HBoV1-positive than in RSV-positive children with lower RTI (58). In a recent review, the prevalences of respiratory manifestations in NPA of HBoV1 PCR-positive children included cough, 79%; fever, 67%; rhinorrhea, 66%; oxygen therapy or hypoxia, 40%; tachypnea, 35%; wheeze, 27%; pharyngitis, 13%; and other respiratory symptoms (includes respiratory distress, increased work of breathing, cyanosis, apnea, rales, and shortness of breath), 48% (53). Since most studies are based on samples drawn for diagnostic purposes from children seeking hospital care for acute RTI, these numbers, to a large extent, reflect the symptoms of the entire study population. In addition, most studies behind these numbers lack stringent criteria for HBoV1 laboratory diagnosis. In a study combining serum PCR and IgG seroresponse in children with various infectious and non-infectious diseases, only lower respiratory tract symptoms correlated with HBoV1 (77). Among children with acute otitis media, HBoV1 DNA was found in the NPAs in 6% and in the middle ear fluids in 3%–4% of cases (52, 131, 132).

Signs of pneumonia, i.e., patchy or interstitial infiltrates in chest radiography, lung hyperinflation, peribronchial cuffing, or atelectasis, appear to occur up to 43%–83% in pediatric HBoV infection, whereas lobar infiltrates or pleural effusion occur rarely (4, 70, 71, 110, 111, 121, 127, 129, 133-135). No radiographic sign, however, is pathognomonic for HBoV1. In one study, 18% of induced sputum samples from children with pneumonia contained HBoV1 DNA (135). The problems discussed above associated with diagnosing HBoV infections by PCR in respiratory tract secretions must be taken into account when interpreting such data. Don et al. (102) described serologically confirmed acute HBoV1 infection in 12% of children with pneumonia. In conclusion, the only symptoms that were statistically associated with HBoV1 confirmed by serodiagnosis are lower respiratory tract symptoms, in particular, wheezing and pneumonia (13, 35, 56, 57, 102). No study has so far provided statistical evidence for an association between HBoV1 and gastrointestinal disease, whereas one study indicated that HBoV2 is associated with gastroenteritis (5). Other symptoms such as rash or exanthema, thrombocytopenia, clinical sepsis, or life-threatening events are rare in HBoV1-positive patients (34, 70, 136). Blood counts and C-reactive protein levels are usually normal (70, 105, 108, 121), but increased neutrophilia was noted (58).

In one study, 31% of 16 children with Kawasaki disease exhibited HBoV1 DNA in serum, NPA, or stool compared to 5.5% (NPA or stool) of 579 controls with RTI. One child with Kawasaki disease even had HBoV1 DNA in NPA, serum, and liquor (137). However, the latter study has not included in-depth statistical analyses; thus, the statistical significance is yet unclear. The association between Kawasaki disease and HBoV was not confirmed in a more recent study in which HBoV1 infections were diagnosed serologically; comparable seroprevalences of bocavirus-specific IgG and IgA were found between the Kawasaki disease and the control groups (138). Similarly, another study using HBoV PCR on serum samples showed no positive findings in 12 Kawasaki patients (139).

HBoVs IN SOLID TISSUES

Because of the lack of animal models for HBoV infections, virus-tissue interactions, and sparse data from clinical studies and cases, there is still very limited information on the in vivo course of HBoV infections. While there is consensus on selected aspects such as the airway pathobiology of HBoV1 infections, there are conflicting data from studies on the role of HBoVs in fetal abortions.

Beyond the detection in human tissues, HBoV1 is found in several solid tumors and is able to trigger molecular pathways related to cancer development in cell culture, thus there is a putative role in cancer, which is not yet clear and discussed below (96, 140).

The first study that investigated the distribution of HBoV1 in human tissue specimens was published in 2007, 2 years after the virus was initially discovered (141), with negative results in rather small patient cohorts of 23 HIV-infected patients and 8 HIV-negative controls. In 2008, one study found a remarkably high rate of HBoV1 DNA positivity of 43%–71% of nasopharyngeal aspirates from an asymptomatic control group of patients who were negative for respiratory infection symptoms but hospitalized for elective surgery of ear, nose, or throat surgeries (16). A possible explanation for this phenomenon was residual DNA from previously cured infections or an uncharacterized persistence of HBoV1.

Later in 2008, Lu and coworkers (142) picked up the hypothesis that HBoV1 may persist in tonsillar tissues. When these authors tested DNA extracts of Ficoll-gradient purified lymphocytes from 164 routine adenoidectomies and tonsillectomies, they found HBoV1 DNA in 32.3% of specimens. For the first time, this indicated that these tissues, or at least the lymphocytes residing in tonsillar tissue, may be a natural target and source for HBoV1 replication and persistence.

A year later, another group identified HBoV in 5% of tonsillar and adenoid tissues of children, more than expected for a seasonal distribution, thus supporting the hypothesis that HBoV1 is able to persist in those tissues (143). In the same year, it was reported that HBoV1 from patient specimens could be productively replicated in a differentiated human airway epithelium air-liquid interface (HAE-ALI) culture from primary respiratory tissues. This was not only a major breakthrough for human parvovirus research but also provided indirect proof that HBoV1 has a tropism for respiratory epithelia (144). Two further breakthroughs were that HBoV1 caused pathological effects in HAE-ALI cultures, such as disruption of the tight junction barrier, loss of cilia, and epithelial cell hypertrophy mimicking the in vivo respiratory injuries and that this virus was able to productively replicate in non-dividing cells by utilizing the cellular DNA damage response (DDR) (145, 146).

These new pieces in the bocavirus puzzle revealed new directions in HBoV research regarding development of optimized cell cultures, tissue tropism of the virus, its presence in solid tumors, and its usage as a gene therapy vector.

Making use of the cell culture DNA extracts from the nasopharyngeal washing aspirates from children, head-to-tail DNA intermediates of the bocaviral genome were identified (147). As these intermediates contradict an exclusive rolling hairpin replication model that would produce head-to-head and tail-to-tail intermediates, they may represent parts of alternative replication models, or because they lack the full hairpins, they could represent a genomic form associated with intracellular persistence or be a sidetrack endpoint product of so far unknown origin. These findings are supported by novel preliminary data on AAV, suggesting that this parvovirus is able to switch between rolling hairpin and rolling circle, which appears to depend on the helper virus (148) (Fig. 1). The DNA circle that is used in this alternative replication mode is most likely a double-stranded circular DNA form that is better known as episomal or covalently closed circular (ccc) DNA (147-149), which is a persistent DNA form in herpes- and hepadnaviruses in tissues (150, 151). While herpes- and hepadnaviruses have a proven oncogenic role in both humans and other mammals (152-157), HBoV1 may only be carcinogenic at this point. In general, parvoviruses are considered oncolytic rather than oncogenic, even though it has been speculated that a persistent parvovirus—such as the human cutavirus associated with cutaneous T-cell lymphoma and its precursor parapsoriasis—possibly could lead to chronic inflammation and transformation (158-161). As proof for an oncogenic or an oncolytic role of HBoV is still missing, more efforts are required to address this phenomenon and elucidate the purpose and function of HBoV cccDNA.

FIG 1.

A rolling-hairpin DNA replication model of HBoV1 DNA replication (162). HBoV1 contains a >95% (−)ssDNA genome with hetero-telomeres (A). Replication begins when the 3′-OH of the left-end hairpin (LEH) primes complementary-strand synthesis by cellular DNA polymerase, producing the monomer turnaround replicative form (mtRF) DNA (B). The newly synthesized strand (blue) is ligated to the right-end hairpin (REH), likely by DNA-PK and its associated ligase, yielding covalently closed RF (cRF) DNA (C). Replication of cRF DNA or a transfected HBoV1 dsDNA genome (e.g., the infectious clone pIHBoV1) requires NS1 to unwind the OriR and nick the top strand (D) at the terminal resolution site (TRS), followed by unwinding of the REH (potentially aided by cellular helicases). This generates the extended/5′-opened RF (5′oRF, E) through a process known as hairpin transfer. Hairpin refolding, dependent on NS1 and possibly a cellular helicase (F), creates a new 3′-OH end that primes complementary-strand synthesis (G) via strand displacement. The resulting double RF (dRF) DNA serves as the primary template for junction resolution at the putative junction resolution site (jrs) in the LEH (H), ultimately generating ssDNA genomes for packaging into preassembled capsids (not shown).

The first evidence that HBoV1 may trigger carcinogenesis was found during the analyses of bronchoalveolar lavage fluids. Here, it was observed that adult patients without a co-infection or otherwise underlying disease had an HBoV1-specific cytokine profile compared to HBoV1-negative patients, including an elevated IP10/TARC ratio, which indicates a profibrotic or fibrotic change in the affected tissue (163). Such a fibrotic intermediate stage is also well known for hepatitis B virus infections, where the fibrotic stage shifts to cirrhosis and finally hepatocellular carcinoma, while cccDNA is formed during viral replication and the persistence phase. This is congruent with earlier clinical observations in which HBoV DNA formed a cccDNA, followed by a supercoiled structure in replication-free episodes and relaxed genomes during disease flares or exacerbations (147, 164-166).

Following the first study that described head-to-tail fragments congruent with a rolling-circle replication of HBoV1 (147), Kapoor and coworkers (149) confirmed the occurrence of HBoV cccDNA, including HBoV3, from solid tissue specimens. Based on these data, it may be carefully concluded that the gastrointestinal tract may be a target organ for HBoV3 or that at least some HBoV3 isolates show a gastrointestinal organ tropism. In concert with the fact that the respiratory and gastrointestinal tracts both originate from the endoderm, these data led to the methodological approach to test solid tumors of these tissues for the presence of HBoV cccDNA (167). In fact, HBoV1 DNA was present in 18.3% of lung tumors and 20.5% of colorectal tumors. Additionally, the presence of HBoV1 in colorectal cancers was supported by later studies from Abdel-Moneim and colleagues (168). In contrast, Karbalaie Niya and coworkers (169) identified HBoV1 DNA in only one of each 66 malignant and 91 non-malignant colorectal tissues. Similarly, Xu and coworkers (170) found only one of HBoV1, 2, and 3, respectively, in healthy intestinal tissues, whereas all 16 malignant tissues were negative. This is in sharp contrast to parvovirus B19, whose DNA was detected in 100 out of 427 (23.4%) intestinal mucosal specimens. For this reason, the issue of HBoV occurrence in tumors and tissues requires further investigation in larger study cohorts in order to elucidate the HBoV genoprevalences and identify factors that lead to persistence. The tropism of HBoV for immortalized colorectal cells is supported by an Argentinian study by Ghietto and coworkers (171), who demonstrated that CaCo-2 cells could be infected using Dextran, although the infections were not productive.

Proenca-Modena and coworkers (99) confirmed and extended previous studies and reported high rates of HBoV1 in tonsillar tissues of pediatric patients with a chronic adenotonsillar disease. This supported the hypothesis that HBoV1 is indeed able to persist. In concert with previous studies on HBoV1 in tumors, it raised the presumption that also tonsillar tumors could be positive for HBoV1. Combined with cccDNA detection and the already shown tumor tropism (167, 168, 172), another study gave evidence that HBoV1 could also be a causative agent (172), although more evidence is needed to validate this hypothesis.

Additionally, Xu et al. (173) found persisting HBoV1 in tonsillar germinal centers in patients from Finland, and HBoV1, 2, and 3 in intestinal tissues, respectively (170). A further study from Finland (174) confirmed the persistence of HBoV. The authors found that HBoV persisted in tonsillar tissues or in the nasopharyngeal area in about 17% of the study patients.

Moreover, it was concluded that HBoV1 suppressed the transcription factors RORC2 and FOXP3, is negatively correlated with the expression of interferon-λ and IL-13, and induces immunosuppression; this immunosuppression, in turn, supports tumorigenesis (175-177).

By analogy with the fact that hepatitis B virus-induced tumorigenesis starts with organ fibrosis, HBoV1 forms cccDNA, increases the release of pro-fibrotic cytokines in vivo and in cell culture, and was observed in cases with idiopathic lung fibrosis (163, 164, 166). Therefore, it was hypothesized that HBoV could be a trigger of tumorigenesis (178). In fact, a controlled cell culture transcriptome analysis showed that, besides others, pathways involved in the development of fibrosis were upregulated (Fig. 2), while apoptosis and cell death pathways were downregulated or inhibited (179).

FIG 2.

Core analysis network predictions for HBoV-infected CuFi-8 cells. This analysis shows that the 54 HBoV-specific genes are involved in apoptosis, necrosis, and (re-)organization of the extracellular matrix. (A) Interaction of different genes with phosphorylation processes (1), apoptosis (2), cell death in general (3), cell death of pancreatic cancer cells (4), tumor cell lines (5) in particular, as well as necrosis (6). (B) Influence of HBoV-specific genes on the organization of connective tissues, including growth (1), proliferation (2), fibroblasts (3), and the quantity of cells (4). (C) Prediction legend. Orange indicates upregulation, and blue indicates downregulation of the respective pathway. Gray indicates that no pathway alterations based on single transcripts could be predicted. Brighter color indicates a weaker alteration, whereas darker color indicates a strong regulation. Reused in agreement with a Creative Commons Attribution (CC BY) license from reference 179.

The latest support for this hypothesis results from the fact that (172) a 60% HBoV1 positivity rate of tonsillar tumors excised from adult patients was much higher than that for lung and colorectal tumors. This finding was independent of the simultaneous presence of papillomaviruses, which are commonly known as causative agents of oropharyngeal cancer. Additionally, further observations support the concept of HBoV1 being an important noxious agent contributing to cancer development during patients’ lifespan (167, 168, 179-182). In this context, it is worth mentioning that tonsillar tumors occur rarely in childhood, and that HBoV1 was undetectable in nasopharyngeal washes obtained 4 years post-adenotonsillectomy in a pediatric cohort (183), indicating that tonsillar tissue is a major target mostly in childhood. Xu et al. (173) later analyzed the HBoV1 tonsillar persistence by in situ hybridization with both sense and anti-sense probes to B cells and monocytes of lymphoid germinal centers, with infrequent expression of mRNA. Furthermore, in B-cell and monocyte cultures and ex vivo tonsillar B cells, the cellular uptake of HBoV1 occurred via the Fc receptor through antibody-dependent enhancement, which resulted in spliced viral mRNA transcription in the nucleus; however, with no detectable productive replication. Furthermore, it is known that HBoV1 DNA is shed for a long time after acute infection, as it was detectable for up to 6 months in young children even after complete resolution of clinical symptoms (184, 185). In addition to this, it was possible to prove chronic persistence of more than 6 months in respiratory tissues of an adult patient based on the occurrence of relaxed and supercoiled cccDNA isoforms, depending on the active replication phase of the virus (165). As children are the largest cohort of patients suffering from acute HBoV infections while having nearly no risk to suffer from tonsillar carcinoma (180), aging is most likely one among several factors on the way to tumor development.

It has to be taken into account that in the bronchoalveolar lavage fluid of HBoV1-positive patients, a profibrotic cytokine response was observed, which was confirmed in vivo by the innate immune cytokine profile from cell cultures infected with HBoV1 (163). This indicates initial evidence for a pathomechanism similar to chronic HBV infection and subsequent development of hepatocellular carcinoma. Mechanistically, it supports the above-mentioned hypothesis that, under yet unknown circumstances, HBoV1 is able to trigger pathways that are strongly and statistically significantly associated with tumorigenesis (Fig. 3). This was shown in a cell culture study in which HBoV1 infection had a significant influence on the expression of 54 genes, of which a remarkable portion is involved in carcinogenesis. This includes increased expression of FN1 as well as decreased expression of FOXO1, which led to the downregulation of apoptotic mechanisms and increased cell proliferation (179). However, these studies need further confirmation, and especially in vivo data are of foremost importance to address this topic.

FIG 3.

Re-analyses of interaction prediction of gene transcripts regulated in CuFi-8 cells infected with HBoV. The regulation of genes was HBoV-specific (179). Part A shows that growth of connective tissue, neoplasia of epithelial tissue, and tumorigenesis are predicted to be activated. Part B uses a different anchor and shows additional activation of phosphorylation, a typical feature of cells undergoing tumorigenesis. Part C adds that fibroblast proliferation is also predicted to be upregulated.

Colorectal tumors were among the first tumor entities identified as tissues associated with HBoV-DNA detection. Concurrently, colorectal carcinoma, especially hereditary cancers such as Lynch syndrome, is frequently associated with microsatellite instability (MSI), which is a marker for impaired DNA repair mechanisms. As HBoV DNA replication is facilitated through pathways taking part in the DDR system that activates PI3KK enzymes (145, 146), the question arose whether there is a correlation between microsatellite instability and HBoV-DNA persistence/occurrence in tumor tissues (172). Although the postulated link between HBoV and MSI could neither be excluded nor confirmed by the aforementioned study, it was shown that MSI patterns vary among tumor tissues and are most likely specific for distinct tumor tissues. While the MSI patterns neither enable a stratification of HBoV-positive tumors nor offer a detailed insight into the usage of the DNA damage response systems for viral replication in situ, there are further studies that confirmed the likely role of HBoV in tumorigenesis. Recently, Nozarian and coworkers (186) found HBoV DNA in 18% of gastric cancer specimens and in 20% of cases with chronic gastritis of Iranian patients. While overexpression of RORC2 is typically a marker for tumorigenesis (187), its downregulation by HBoV1 would support a tumor tropism of HBoV1 rather than a role of the virus in tumorigenesis. The situation is different with FOXP3, as both its up- and downregulation are associated with tumorigenesis (188).

In summary, HBoV1 DNA has already been found in lung tumors (167), colorectal tumors (167, 168, 170, 181), tonsillar carcinoma (172, 189), and gastric cancer (186). While these findings may have serious long-term impact on clinical outcomes, HBoV DNA was also present in tissues other than the aforementioned tumors, such as non-malignant tonsils (101, 142, 143, 170, 173, 174) and intestinal mucosa (149, 170). An observational study identified HBoV in 3.8% of benign cholesteatoma likely associated with persisting HBoV infections (190). In this study, the immune cells invaded the tissues due to chronic HBoV-associated inflammation, resulting in cholesteatoma.

Beyond tumor tissues, HBoV1-4 DNA has also been studied in fetal and placental tissues, since the HBoV-related bovine and canine parvoviruses are known to cause serious complications in cattle and dog gestations, respectively, which can culminate in fetal death and abortions (191-196). Nevertheless, a causative link between HBoV DNA and the clinical course in humans has not been established. Among the samples tested for HBoV are amniotic fluid, fetal tissues, and placenta from abortions (96, 97, 140). In a study from Cologne, 25% of the investigated tissues were positive for HBoV DNA (6 placental tissues and aborted fetal tissue from 37 deceased fetuses) (140). In contrast, a study from Finland investigated by PCR archived formalin-fixed paraffin-embedded autopsied fetal tissue specimens from 289 miscarriages or fetal deaths and 246 induced abortions as controls, with no HBoV1-DNA positives found (96). Almost all mothers (97%) were HBoV-IgG positive (at the time without blocking of cross-reactive HBoV2-4 antibodies) already in the first trimester of pregnancy, and 0.9% (4/462) exhibited HBoV-IgM antibodies without viremia. Another study analyzed 87 amniotic fluid samples from fetuses with hydrops or isolated effusions (97). In this study, the mothers (19/19) had HBoV-IgG antibodies, whereas none exhibited HBoV IgM, implying that in those specimens, HBoV was neither the causative agent nor persisting. In contrast, of the fetuses with hydrops or anemia, 5 of 289 (1.7%) in the first study and 12 of 60 (20%) in the second study were positive for parvovirus B19 DNA (97, 197). As the studies did not differentiate between HBoV1 and HBoV2-4 IgG, further studies addressing this issue are required.

Rarely, HBoV DNA is also detected in heart tissues of patients with inflammatory heart disease, but in contrast to B19V, HBoV is likely not persisting in the heart (198). In 2022, Benitez Fuentes and coworkers (199) identified the virus in the respiratory tract of a prostate cancer patient with symptoms radiologically resembling congestive heart failure. This could, however, also be because of long-term shedding of free HBoV1 DNA from tissues damaged by cancer or heart condition (17).

In another case report from France that found HBoV2 DNA in the heart tissue of a 13-month-old patient who died from a subacute myocarditis, HBoV2 was the only pathogen identified (93). In general, HBoV myocarditis appears to be a rare event, as shown by a South African study that also reported only a single case (1/82) of HBoV1-DNA-positive myocarditis (200), and the study by Kuethe and colleagues (198), who found 5% of heart tissues were HBoV1 positive in their cohort. Nevertheless, a causative association cannot be excluded, especially when taking into account that H9C2 cardiomyocytes are permissive for HBoV1, which triggers inflammation factors in these cells (201). However, HBoV1 DNA was not found in the hearts or any other tissues of 31 adults who recently deceased of non-viral causes (202); thus, there remains the possibility for a mechanistic involvement of HBoV.

DIAGNOSTICS

After the discovery of HBoV1 in 2005 and the first prevalence studies (4, 17, 57, 70, 106, 203, 204), the virus was for a long time considered a bystander by many clinicians and virologists. This was because the sensitive PCRs often detected HBoV1 in airway samples of children with RTI together with other RTI pathogens as “co-infections,” and it was also found in respiratory samples of some healthy controls. However, accumulating evidence showed that these “co-infections” are in fact only “co-detections” of long-term HBoV1 shedding after the acute phase of infection, which could continue for weeks, months, and even up to a year, sometimes intermittently (52, 184, 185). It was recently shown, by endonuclease pretreatment prior to PCR (ePCR), that this notorious persistent shedding is actually due to free viral DNA, leaking out from damaged tissues, as opposed to capsid-protected genomes in acute phase (17). Intermittent PCR positivity could possibly also occur upon new tissue damage due to subsequent RTI episodes caused by other viruses. The commonly used qualitative PCR-panel tests cannot distinguish between a true acute infection and prolonged shedding of viral DNA. Hence, while HBoV1 RTIs are very common in children and sometimes even life-threatening, their diagnosis is inadequate (39, 205, 206).

A more accurate diagnosis would instead necessitate qPCR, mRNA reverse transcription (RT) PCR, antigen detection, ePCR, or serology with enzyme-linked immunoassays, which all have been successfully set up in research laboratories (13, 17, 32, 55, 205, 207-213). A limitation is that many of them are too complex to be directly applicable in routine hospital settings and that few, if any, such commercial tests are currently commercially available. Viral mRNAs are not very stable since HBoV RNA is not protected inside capsids as in RNA viruses but remains exposed to RNases. Nonetheless, it was possible to diagnose HBoV1 acute infection by measuring viral mRNA, endonuclease-resistant encapsidated DNA, and also antigen, although they are detectable only for a few days during the acute phase (17, 55, 205, 210-212). Due to the general decrease of viral load in airway samples with time after acute infection, a high viral load (≥10⁴ copies/mL) measured by qPCR is more specific in determining an acute infection than mere qualitative PCR (35, 205, 214, 215). Still, not even qPCR is totally accurate. In a Finnish study, a third of children with low viral loads in NPA exhibited, according to serology, a true acute HBoV1 RTI (13).

On the other hand, serology was also shown to be challenging. One obstacle for serodiagnosis is that the four HBoVs induce cross-reactive antibodies that hamper the interpretation of the test findings. To circumvent this, competition assays were applied to block the interfering HBoV2-4 antibodies with free virus-like particles of VP3 (earlier called VP2) from the enteric HBoVs in heterotypic blocking, and, for confirmation, VLPs of HBoV1 itself in homotypic or “self”-blocking, prior to the EIA (32, 36).

Serologic studies may be hampered not only by HBoV1-4 cross-reactivity but also by another immunologic phenomenon referred to as original antigenic sin (OAS) or imprinting (33, 216, 217). OAS has been known for over 60 years and has been studied particularly in influenza, dengue, and SARS-CoV infections, but also for other viruses that exist in multiple serotypes (216, 218-220). If two highly related viruses sequentially infect the same host, the second virus infection may not induce a proper specific antibody response; instead, the immune response to the original virus infection will be enhanced. This may have an impact on the severity of disease, vaccination efficacy, serodiagnosis, and seroepidemiology. In HBoV2-4, the VP3 protein sequences are approximately 20% divergent from those of HBoV1, and they contain common epitopes, which are prerequisites for OAS to occur. In children with prior immunity to HBoV2 or 3, a PCR-verified HBoV1 infection was shown to result in low or absent HBoV1-specific antibody responses (33). The same was shown in a more controlled setting with sequentially HBoV1-4 VLP-inoculated rabbits (217). This may imply that some children lack HBoV1-specific antibody responses even if they are HBoV1 infected. It is not known whether this affects the severity of HBoV-caused disease.

Conclusive diagnosis of an acute HBoV1 infection is, therefore, not easy and takes a wider repertoire of methodologies. Nevertheless, it is possible to stage the acute HBoV1 infection according to the presence of high loads of virus-specific DNA in NPA and serum or by endonuclease treatment of DNA (ePCR), combined with the detection of HBoV1 IgM and diagnostic IgG (13, 17, 32, 48, 71, 208, 209, 221, 222). The latter comprises either seroconversion or a ≥4-fold increase in IgG titer in paired serum samples or IgG of low avidity.

Of 258 wheezing children from Finland, 48/258 (19%) had an acute primary HBoV1 infection exhibiting IgM; 27 with IgG conversion and 5 with a ≥4-fold IgG titer increase in paired samples, both representing acute infection, 5 with no IgG as very acute, and 10 with stable IgG as subacute infection, as shown in Fig. 4 (13). Interestingly, among 49 children with viremia, 45 (92%) exhibited an HBoV1 serodiagnosis, while among 49 children with HBoV1 PCR-positive NPA, only 35 (71%) had a serodiagnosis, 33 of whom were also viremic. This showed that, due to the long-term viral shedding in the airways, viremia is a better marker of acute HBoV1 infection than PCR-positive NPA. Moreover, while 96% of those with high viral loads in NPA had serodiagnoses, the percentage was only 38% for those with low virus loads, further confirming the inadequacy of qualitative PCR.

FIG 4.

Scatter plots of individual absorbance values at 492 nm (A₄₉₂) of immunoglobulin (Ig) G (×) and IgM (red dots) against human bocavirus in enzyme immunoassays for acute phase (I), convalescence phase (II), and 5-year follow-up (III) serum samples from wheezing children and single serum samples from young healthy adults, Finland. The 45 children with confirmed acute HBoV1 infections (by viremia and serodiagnosis) were divided into three groups according to the degree of acuteness (very acute, acute, and subacute) on the basis of findings in I and II serum samples. Very acute, I sample seronegative but II sample IgM positive (n = 12); acute, I sample IgM positive but IgG showed seroconversion (n = 20); subacute, IgG positive with a diagnostic increase or constant level in I and II samples, IgM positive (n = 13). Also shown are results for children without viremia or serodiagnosis (nonacute [for only the first 45 children with seropositive samples]) and young healthy adults (n = 115). EIA cutoffs are indicated by a blue line (IgG; 0.188) or a red line (IgM; 0.167). Dots below the cutoff lines indicate samples with absorbance values less than the negative cutoff values (i.e., IgM and IgG-negative results) (modified from reference 13).

CELL CULTURE MODELS

The longest known human-pathogenic parvovirus, B19V, was initially detected by electron microscopy 50 years ago (223) and has been associated with human diseases since 1981 (224). Despite major efforts, B19V cannot be propagated in cell culture to date. Culturing is only possible in primary erythroid progenitor cells or the UT7/Epo-S1 cell line, but the virus cannot be passaged in these cells (225). HBoV1, detected exactly 30 years later (4), was initially also considered unculturable (226). However, in 2009, two out of three clinical isolates containing HBoV1 were successfully inoculated and propagated on primary human airway epithelium cultured at an air-liquid interface (144). Despite this breakthrough, the full-length HBoV1 genome was not sequenced at that time and the terminal repeats, postulated to be hairpin DNA structures similar to all other known mammal parvoviruses and relevant for the postulated replication mechanism of rolling hairpins, remained unknown.

This changed in 2011 when two studies published the finding that HBoV1 (147) and HBoV3 (149) persisted as episomes. Both identified intermediate linker sequences, which were assumed to comprise parts of the terminal hairpin-forming sequences of the genome. Sequencing of the full-length hairpin termini reveals “rabbit ear”-like structures similar to those of other bocaviruses (227). The head-to-tail cccDNAs could theoretically result from a rolling-circle mechanism of parvovirus DNA replication in addition to the classical rolling-hairpin model of parvoviruses (228). Besides that, there is increasing evidence for an alternative replication of human bocaviruses based on a broad deepsequencing approach published recently by Zhang and coworkers (229), who identified replication intermediates compatible with a rolling circle replication in 42 (12%) out of 351 clinical specimens positive for HBoV. There is also accumulating evidence for a rolling-circle replication mechanism in addition/alternatively to the rolling-hairpin replication of AAV2, which belongs to the Dependoparvovirus genus. In this case, the mode of replication is determined by HSV-1 as a helper virus (148).

A deeper analysis of the linking sequence of the HBoV1 cccDNA showed that these sequences may be evolutionarily related to the terminal sequences of the bovine and canine parvoviruses (227, 230), suggesting that HBoV may have a common zoonotic origin lagging behind (178), rather than coevolution.

After the identification of the HBoV1 terminal structures and construction of an infectious HBoV1 clone, a reverse genetic system was established by transfection of the clone in HEK293 cells (227). Once transfected, the clone was able to produce HBoV1 particles, which were infectious in HAE-ALI cultures prepared from either primary bronchial airway epithelial cells isolated from lung donors or from CuFi-8 cells. CuFi-8 (cystic fibrosis, University of Iowa) cells are immortalized human airway cells isolated from bronchial epithelium of a CF patient (Fig. 5) (227, 231). While some other permissive cell lines exist, CuFi-8 cells appear to be the only permissive CuFi-cell line supporting productive HBoV1 infection. It is worth noting that some CuFi cell lines are available at ATCC but are not permissive for HBoV1, while the CuFi-8 cell line can be purchased elsewhere (https://cancertools.org/cell-lines/cufi-8-154454/).

FIG 5.

HAE-ALI model. Primary human airway epithelial cells taken from bronchi from the lungs of healthy donors, or immortalized HAE cell line (CuFi-8) (231) cells, are plated onto Transwell inserts (Costar, Corning, NY, USA) at an air-liquid interface for 4 weeks. Four major types of epithelial cells in the well-differentiated or polarized HAE cultures, i.e., basal, ciliated, goblet, and club cells, are illustrated in the Transwell insert. Numbers (%) shown are the numbers of various cell types determined in HAE-ALI made of CuFi-8 cells (232).

Since then, CuFi-8 and primary airway cells have been repeatedly and successfully used for the study of HBoV1 replication (146, 227, 233), tumorigenesis (179), and innate immunity (163), and a detailed technical protocol is available for the setup of these ALI culture systems (233).

As the limited availability of CuFi-8 cells remains an obstacle for HBoV1, alternative research approaches are required. To improve the toolbox of HBoV cultures, T84 cells, originating from a lung metastasis of a colorectal tumor, support a productive replication of HBoV1 and have been used as an alternative to CuFi-8 cells (234). They grow easily in an ALI epithelium, even if they exhibit a rather pleomorphic shape compared to that of CuFi-8 (234). Most recently, it has been shown that even monolayers of T84 could be productively infected with HBoV, albeit at much lower efficiency (235). Moreover, the MA104 cell line (ATCC #CRL-2378.1), based on epithelial cells from the kidney of an African green monkey, has been reported to be also permissive to HBoV1 infection (236). An additional but not yet broadly used model is a matrix-based three-dimensional cell culture model originally used for human rhinovirus culturing, which is also permissive for HBoV1 (237). Therefore, a comparative study between these approaches will be useful to identify an appropriate cell culture for productive HBoV1 infection.

To date, three-dimensional ALI cultures remain the best available tool for studying the molecular mechanisms of HBoV1 infection, as discussed in detail in the upcoming section of this review. Among others, the relevant findings included the detection of pyroptotic cell death caused by HBoV1 (238) and the involvement of the DNA damage repair system in viral DNA replication (145). Moreover, HBoV1 caused cytopathic effects in all three-dimensional cell culture systems while being successfully passaged (238-240). Chiu and coworkers (241) have shown that VP1 of HBoV1 is involved in disrupting the tight junctions between A549 cells, which could mechanistically explain the observed cytopathic effects. The pathologic effect of VP1 was additionally proven by Lin and colleagues (242), who described lung injury in mice after receiving a VP1 peptide derived from the unique regions of HBoV1 and B19, respectively. The group showed that a serious cellular and cytokine-mediated inflammatory response in the animals’ lungs was induced by subcutaneous infection of VP1, which was caused by altered p65 (NF-κB) and MAPK signaling. Taking into account that NP1 also reduced viability and fitness of A549 cells (241), the cytopathogenic effects of HBoV1 provide further support for HBoV1 being a true pathogen.

STRUCTURAL AND MOLECULAR INSIGHTS

HBoV1 life cycle

HBoV1 infects both upper and lower airways (16, 34, 41, 51, 52, 56-58, 64, 70, 71, 103, 105-107, 116, 119, 122, 123, 126, 205, 222, 243-248) in vivo and in vitro. HBoV1 efficiently and productively infects polarized/well-differentiated HAE-ALI (144, 232) (Fig. 5). At a multiplicity of infection of 0.001 viral genome copies (vgc)/cell or ~1,000 viral particles per 1 cm² of HAE, HBoV1 infection was productive with release of virions from the apical side and caused epithelial barrier dysfunction (239). Considering physical particles, HBoV1 is much more infective than SARS-CoV-2, which requires a high viral load of 2.5 × 10⁵ virions per cm² of HAE (249). HAE is composed of four major epithelial cell types. These are ciliated, goblet, club, and basal cells, making up ~50%, 4%, 22%, and 24% of the lower large airway epithelia (232) (Fig. 5). It has not yet been carefully studied which of those are the primary HBoV1 target cells so far, but during in vitro infections of HAE-ALI cultures, most of the apical cells died in the epithelial layer, while the basal cells remained alive. This finding suggests that these are not the primary target cells for HBoV1 (239). Furthermore, the cilia on the top of the HAE-ALI/CuFi-8 organoids are destroyed by the HBoV1 infection as previously shown (179). Utilizing rAAV2/HBoV1 vectors to transduce HAE-ALI, most of the tubulin IV- and K18-expressing cells expressed the reporter mCherry, suggesting the ciliated cells are susceptible to HBoV1. Goblet and club cells are also permissive to rAAV2/HBoV1 vector (250).

In vitro, HBoV1 is able to infect HAE-ALI cultures through both the apical and the basolateral surfaces (Fig. 6), but the receptors necessary for HBoV1 cell entry have not yet been identified. In general, parvoviruses enter the cells through receptor-mediated endocytosis (251-253). After its entry, HBoV1 likely traffics through the early to late endosomes (Fig. 6, Step 3) (254, 255). Within the nucleus, the virion releases its ssDNA genome that is recognized by the cellular DNA damage and repair machinery (145, 146). Subsequently, the complementary strand of the viral ssDNA genome is synthesized, transcribed, and the viral NS proteins, as well as the sncRNA BocaSR, are produced (Fig. 6, Steps 6–11). These proteins initiate and support further viral DNA amplification (Fig. 6, Steps 12–16). Then, capsid proteins are synthesized in the cytoplasm, where they assemble into oligomers before translocating into the nucleus. There, the empty capsids incorporate the replicated ssDNA genome (Fig. 6, Steps 17–19) likely conducted by the NS proteins. Finally, the mature virions are released from the nucleus into the cytoplasm and are then transported outside of the infected cell (Fig. 6, Steps 19–20) or are released due to the death of the cell undergoing pyroptosis (238).

FIG 6.

The infection life cycle of HBoV1 in human airway epithelial cells. A ciliated epithelial cell is depicted with cilia and tight/adherens junction molecules. HBoV1 enters the cells through binding to a primary proteinaceous receptor, which is expressed on both the apical and the basal sides of the ciliated cells, followed by endocytosis and intracellular trafficking (Steps 1–3). The virion escapes from the late endosome and enters the nucleus (Step 4). In the nucleus, the single-stranded DNA viral genome is uncoated, converted to its replicative form. This transcriptional-capable dsDNA is provided for the expression of viral NS proteins and the sncRNA BocaSR (Steps 5–8). During ongoing replication of viral DNA (Steps 12–16), both viral NS and capsid proteins are expressed in the cytoplasm (Steps 9–11), followed by formation of empty capsids in the nucleus that capture replicated ssDNA genomes (Steps 16–18). In the last step, matured progeny virions leave the nucleus and the plasma membrane (Steps 19–20). Not in scale.

HBoV1 genome

The full-length genome of HBoV1 was only sequenced once from a nasopharyngeal aspirate that contained a high virus load (1.2 × 10⁸ vgc/mL; [HBoV1 Salvador 1 isolate]) (222). The genome of the Salvador 1 isolate (GenBank accession no. JQ923422) contains 5,543 nucleotides (nt) with distinct hairpin structures at the left and right ends (227). The left-end hairpin (LEH) is 140 nt in length and is predicted to be “Y”-shaped with short axial ears and mismatches causing an unpaired bubble. The right-end hairpin (REH) is 200 nt in length and is predicted to form a perfect palindromic structure. The REH contains a minimal DNA replication origin (Ori) from nt 5,357 to nt 5,402 (256). It is composed of a cleavage site and four repeats of the triple nucleotides TGT, representing NS1-binding elements (NSBE). HBoV1 NS1 binds to the NSBE and nicks the OriR in front of a T nucleotide, which is located 12 bp upstream of the NSBE (256). The HBoV1 LEH does not contain any sequences similar to NSBE.

The HBoV1 genome is extensively methylated at CHG and CHH sites (H represents A, C, or T) in transfected HEK193 cells. HBoV1 genome methylation facilitated viral DNA replication but decreased NS1 and NP expression, which is mediated by repression of viral RNA splicing and polyadenylation at (pA)p (257).

Transcriptional profile of HBoV1

The HBoV1 genome has an RNA Pol II promoter with a proximal (pA)p and a distal (pA)d polyadenylation site. Therefore, HBoV1 transcribes only a single pre-mRNA (Fig. 7), which undergoes alternative splicing and alternative polyadenylation to generate multiple viral mRNA transcripts. The left half (3′ end) of the viral genome encodes the NS proteins, and the right half encodes the structural proteins VP1-3 (258-261). The middle of the genome encodes the phosphorylated protein NP1 (227, 258, 259). Due to the small genome capacity, the NP1 ORF largely overlaps with the 3′ end of the NS PRF.

FIG 7.

Genetic map of HBoV1. The viral genome is depicted with terminal repeats (LEH and REH). The transcription units, the promoter (P), splice donors (D) and acceptors (A), and (pA)p and (pA)d sites are depicted. The major RNA Pol II-mRNA transcripts, designated R1–R6, are listed below the dsDNA genome, with their respective sizes shown in kilobases on the left and the detected sizes of the expressed proteins shown in kilodalton to the right. The three ORFs are illustrated in blue, red, or green. The RNA Pol III-transcribed BocaSR is illustrated at the bottom. NCR, noncoding region (24).

Unspliced mRNA that contains the D2-A2 intron encodes the NS1-70 protein (260). The mRNA splicing at the D2-A2 sites results in the expression of NS1 due to a shift of the NS1 ORF at the C-terminus. Among the NS1-coding region, the mRNA transcripts alternatively spliced from intron D1 to A1′, D1′-A1, and both generate R2, R3, and R4 mRNAs, which encode NS2, NS3, and NS4, respectively. NP1-encoding mRNA (R5) is spliced at both the D1-A1 and D2-A2 introns. VP-encoding R6 mRNA is consecutively spliced at all the three introns, D1-A1, D2-A2, and D3-A3. VP-encoding transcripts only use the (pA)d sites, whereas all NS-encoding mRNA transcripts have short (R_S) and long (R_L) forms, which are terminated at the proximal [(pA)p1 and (pA)p2] or the distal polyadenylation sites [(pA)d1 and (pA)d_REH] (Fig. 7) (262, 263).

The most striking feature of HBoV1 transcription is the expression of the sncRNA BocaSR (24). It is transcribed from nt 5,199 to nt 5,338 of the double-stranded DNA genome through an RNA Pol III promoter that lies entirely within the VP/Cap gene and composes A- and B-boxes related to those in adenovirus viral-associated (VA) I RNA (264). Its transcription level is much higher than that of all other HBoV1 mRNA transcripts and in accordance with the level of the adenovirus VAI RNA expressed in cells.

HBoV1 proteins

HBoV1 expresses five NS proteins: NS1, NS1-70, NS2, NS3, and NS4, as well as NP1 during infection. HBoV1 NS1 has a size of ~100 kDa. It is composed of three functional domains, including the N-terminal DNA-binding/endonuclease domain, the central helicase domain, and the C-terminal transcription activation domain (265). The N-terminal domain binds to the Ori of viral replicative form dsDNA, also referred to as the origin or DNA-binding domain (OBD). It exhibits a strand- and sequence-specific endonuclease activity (266). The HBoV1 OBD comprises aa 1–275. Superimposition of the OBD structures of AAV5 Rep68 (267), MVM NS1 (268), and HBoV1 NS1 revealed a conserved beta-sheet core flanked by several alpha helices on both the left and right sides (266). They belong to the HUH-nuclease superfamily, members of which share recognizable conserved motifs. The HBoV1 OBD/Ori complex has not yet been structurally resolved. The central helicase domain of the HBoV1 NS1 contains four conserved Walker motifs, which belong to the SF3 helicase family and carry out the 3′–5′ helicase function (269, 270). The HBoV1 C-terminal domain (aa 638-781) is predicted to have transcription transactivation capability, but its function has not yet been studied (271, 272). NS1 specifically interacts with Ku70 with a K_D of 160 nM and processes the strongest interaction at the C-terminal domain (273). Interestingly, NS1 promoted the degradation of methyltransferase 1 (DNMT1), the major host methyltransferase, through the ubiquitin-proteasome pathway. Whether a balanced DNMT1 expression is required is arguable (257).

NS1-70 is a shorter version of NS1 lacking the C-terminus, which, in principle, is able to induce a DDR like the full-length NS1 and supports viral DNA replication, but it is expressed at a lower level than NS1 during virus infection (146, 274).

HBoV1 NS2 is composed of the OBD and the putative transcription activation domain. Knockout of NS2 in the HBoV1 dsDNA (an infectious clone) demonstrated that it is not essential in replication in HEK293 cells, but NS2 is required during the infection of HAE-ALI (274). It is also required to support a productive AAV2 infection in both HEK293 and HeLa cells, together with the helper functions of HBoV1 NP1 and BocaSR (275). Moreover, it was used in a better adenovirus (Ad) helper plasmid (pPLUS AAV-Helper; Sartorius) for rAAV production.

The NS3 protein is not essential for HBoV1 DNA replication in HEK293 cells and virus infection of HAE-ALI (276). It overlaps completely with the NS1 helicase domain and, thus, likely has a similar function as the AAV Rep52/40 in viral genome packaging (277).

The NS4 protein is also not essential for HBoV1 DNA replication in HEK293 cells and virus infection of HAE-ALI (276). It is composed of only the putative transcription transactivation domain, which is also encompassed by NS1, NS2, and NS3. In AAV2, it can substitute the function of NS2 in supporting the replication of the AAV2 dsDNA genome (an infectious clone) in HEK293 and HeLa cells (275). Although NS4 of HBoV1 has only 199 aa with a predicted size of 22 kDa, it displays a ~34 kDa protein in SDS-polyacrylamide gel electrophoresis (274). Therefore, it is likely that the NS proteins are post-translationally modified at the C-terminal domain.

The nonstructural protein NP1 has a size of ~200 aa with a non-canonical nuclear localization signal located at aa 7–50 (278). Although NP1 shares only ~48% identity among various bocaparvoviruses in amino acid sequence, its function in viral DNA replication is conserved (227, 256, 258, 261). NP1 proteins of BPV1 MVC are exchangeable with each other, and HBoV1 NP1 is also able to supplement the function of MVC NP1 (258). Moreover, NP1 plays an important role during the processing of viral pre-mRNA (238, 261, 276, 279). Both MVC and HBoV1 NP1 facilitate transcription of viral pre-mRNA through the (pA)p site to generate full-length VP-encoding transcripts (238, 279). Cellular cleavage and polyadenylation specificity factor 6 (CPSF6) interacts with both MVC and HBoV1 NP1 (280, 281). CPSF6 is one of the cellular factors in the polyadenylation complex that associates with the AAUAAA motif of the polyadenylation signal (282). It diminishes MVC NP1’s suppression of the internal polyadenylation at (pA)p, enhances the splicing of the third intron, and further modulates the export of MVC mRNA (280). Notably, CPSF6 facilitates splicing of VP-encoding mRNA and the nuclear import of HBoV1 NP1 (281). Additionally, NP1 plays an important role in viral DNA replication (256) by directly interacting with Ku70 and RPA70 with a high binding affinity in vitro (K_D = 95 and 122 nM, respectively) through the Ku70 β-barrel and RPA70 AB domains, respectively (273). The interactions of Ku70 and RPA70 with NP1 play a significant role in HBoV1 DNA replication, as determined in an in vitro viral DNA replication assay but also in HBoV1-infected HAE-ALI cultures. NP1 mutants bearing mutations in the AlphaFold2-predicted helices 1, 2, 3, and 4 prevented their binding with Ku70 and RPA70 and nearly abolished viral DNA replication. These NP1 mutants that lost the in vitro binding with Ku70 and RPA70 also exhibit a markedly reduced capability to support HBoV1 DNA replication in vivo, demonstrating that the interaction of NP1 with Ku70 and RPA70 is critical for HBoV1 DNA replication (273).

The small viral noncoding RNA BocaSR

HBoV1 expresses the small viral noncoding RNA BocaSR, which is unique among small DNA viruses, including polyomaviruses, papillomaviruses, parvoviruses, and circoviruses. BocaSR folds into a secondary structure composed of terminal stem, central domain, and apical stem and loops (Fig. 8A) (283). This sncRNA is essential for HBoV1 replication not only in HEK293 cells but also in HAE-ALI (24). BocaSR, localized exclusively in the nucleus, shares a similarity of 46.1%–51.2% with the other four known RNA Pol III-transcribed viral small RNAs: Ad- VA-I, VA-II, Epstein-Barr virus-encoded small RNA (EBER1), and EBER2. Although BocaSR does not inhibit the phosphorylation of protein kinase R and the eukaryotic initiation factor 2 (24), it plays a role in the regulation of the expression of NS1, NS2, NS3, and NP1, but not NS4 (24).

FIG 8.

Structure and function of BocaSR. (A) BocaSR secondary structure in HAE. The secondary structure of BocaSR in HAE-ALI cultures was determined by dimethyl sulfate-mutational profiling with sequencing (DMS-MaPseq). Determined stems, central domain, and loop structures are indicated. High normalized DMS activity score (ΔnDMS) indicates high DMS accessibility and low base pair probability. Nucleotides of the BocaSR sequence are shown. (B) Model of the assembly of the HBoV1 replication initiation complex (RIC). BocaSR is methylated by the m6A-methyltransferase, METTL3, which is complexed with other m6A processing proteins. BocaSR interacts with the negative strand of the HBoV1 genome located in the OriR. The dsDNA OriR is recognized by NS1 that executes its helicase activity to unwind it. Then, the BocaSR-m6A-METTL3 complex recruits Y-family DNA Pol η and Pol κ to the OriR. The DNA damage/repair proteins, e.g., KU complex and phosphorylated replication protein A (RPA), recognize the DNA damage foci. p-RPA recruits NP1 that further recruits the KU complex. All these factors assemble the RIC for nicking the OriR and the initiation of viral DNA replication (60).

In addition, BocaSR plays a direct role in viral DNA replication, which cannot be fully complemented by VA-I RNA (24). BocaSR interacts with the REH of the HBoV1 genome. BocaSR binds to its target sequence in the OriR with high affinity in vitro (K_D = 26.7 nM), demonstrating a strong interaction between BocaSR and the OriR and indicating that BocaSR may bind to the unwound ssDNA at OriR during HBoV1 DNA replication (283).

HBoV1 BocaSR is m6A-methylated at multiple sites and is associated with the METTL3-YTHDC1 complex (283). The m6A-RNA-METTL3 complex recruits DNA Pol κ to repair UV-damaged DNA (284, 285). HBoV1 utilizes Y-family DNA repair DNA Pol η and Pol κ for viral DNA replication (145, 146), and BocaSR localizes to the HBoV1 DNA replication centers (24). Once associated with the m6A processing complex, BocaSR interacts with the viral DNA replication origin and recruits the Y-family DNA repair DNA Pol η and Pol κ to the HBoV1 replication initiation complex (273) in order to facilitate viral DNA replication (Fig. 8B) (283).

DNA repair-based HBoV1 DNA replication

In contrast to the DNA Pol-dependent replication model of other parvoviruses (286-292), HBoV1 infects terminally differentiated, non-proliferating human airway epithelia. Thereby, HBoV1 DNA replication follows a model of DNA repair (145, 146). HBoV1 infection activates Ataxia-Telangiectasia Mutated protein, Ataxia-Telangiectasia and Rad3-related protein, and DNA-dependent protein kinase catalytic subunit. The three DDR pathways are activated during HBoV1 infection of HAE and dsDNA genome transfection of HEK293 cells and play significant roles in viral DNA replication (145, 146). The DNA repair polymerases Pol κ and Pol η also play a significant role in HBoV1 genome replication in both HAE-ALI and HEK293 cells (145, 146). Upon entering the nucleus, the viral genome is recognized by Pol κ and/or Pol η, which synthesize the complementary strand primed by the 3′-OH at the LEH terminus. NS1 is essential for HBoV1 DNA replication and induces DDR signaling but no cellular DNA damage.

In addition, NP1 and BocaSR are required for viral replication in HAE. They are localized within the viral DNA replication centers (293) and significantly enhance viral DNA replication in an in vitro DNA replication assay (283). NP1 directly interacts with Ku70 and RPA70 and functions as a mediator to recruit these replication-necessary proteins during HBoV1 replication (273). Ku70 directly interacts with NS1 through its C-terminal domain. Knockdown of Ku70 and overexpression of the C-terminal domain of Ku70 significantly decreased HBoV1 replication in HAE-ALI. Thus, Ku70 facilitates the recruitment of the NS1 to the OriR. The Ku complex was reported to exhibit helicase activity (294), which may facilitate the NS1 to nick and unwind the OriR. NP1 and NS1 did not interact with each other. Thus, HBoV1 NP1 enhances viral DNA replication through its direct interactions with Ku70 and RPA70 (Fig. 8B). HBoV1 DNA replication was shown to be hairpin transfer-independent (162).

STRUCTURE

The structural proteins—VP1, VP2, and VP3

HBoV1 expresses three overlapping capsid proteins from a single mRNA through alternative usage of the start codons. This results in VP3, the smallest structural protein, being entirely contained in VP1 and VP2, as well as VP2 within VP1 (Fig. 9). VP1, VP2, and VP3 are expressed and incorporated into a capsid at a stoichiometry of ~1:1:10, which is similar to that of AAVs (295-297).

FIG 9.

HBoV1 VPs. A schematic depiction of the overlapping VPs is shown with their size in amino acids. The first and last aa of each VP and selected sequence motifs are indicated. Ca²⁺BS, calcium binding site; PLA₂, phospholipase A2; BR, basic region; VP1u, VP1 unique; and VP1/2, VP1/2-common region.

The largest structural protein VP1 has a size of ~75 kDa or 671 aa. Its 90 aa unique N-terminus, the VP1 unique (VP1u) region, codes for a phospholipase A2 (PLA₂) domain (aa 11–66) (298). The HBoV1 PLA₂ domain contains a conserved calcium-binding site (Ca²⁺BS) (aa 18–22) and the core catalytic HDXXY motif (aa 41–45) as known for AAVs and other parvoviruses (299, 300). The PLA₂ domain was also confirmed in MVC and is likely present in most bocaviruses (258). The C-terminal end of the VP1u region contains a basic region (BR) (aa 86–89) (Fig. 9), which is equivalent to BR1 of AAV and BR2 of MVM (301). This region may act as a nuclear localization signal (NLS), but this has not been verified yet.

The second structural protein VP2 has a size of ~70 kDa or 581 aa. VP2 is translated from a non-canonical start codon (GUG) located between VP1 and VP3, which is uncommon for parvoviruses (302, 303). The N-terminal region shared between VP1 and VP2 is often referred to as the VP1/2-common region and consists of 39 aa. Similar to AAV, this region contains two BRs (aa 98–105 and aa 113–120) (Fig. 9), which may represent NLSs to target the capsid to the nucleus (80, 81). However, as with the BR in VP1u, it is unknown whether these regions act as true NLSs (301).

The smallest and major capsid protein VP3 has a size of ~61 kDa or 542 aa and is expressed at a ~10-fold higher abundance than VP1 and VP2. VP3 is the sole protein required to form VLPs or capsids (304). For the determination of HBoV amino acid position, both VP1 and VP3 numbering systems were used in the past. In this review, the VP1 numbering system comparable to AAV is used, and we propose to use this as standard in the future for HBoVs. The VP3 N-terminus is characterized by a high number of glycine residues similar to that in most parvoviruses (Fig. 9) (305). This accumulation of glycines was suggested to confer flexibility to the N-terminal VP1u or VP1/2 common region, which is needed to support their functions during the viral life cycle. At the same time, this contributes to the inability to obtain structural information on the N-terminal side of the glycine-rich motif (306).

The HBoV1 capsid structure

The virion of HBoV1 is composed of a non-enveloped, T = 1 icosahedral capsid with a diameter of ~25 nm that packages its ssDNA genome (306). The capsid is assembled from 60 subunits of the overlapping VP proteins, VP1:VP2:VP3 in an approximate ratio of 1:1:10 (Fig. 10). Although the capsid assembly process has not been studied for HBoV1 or other bocaviruses, its mechanism is likely comparable to other parvoviruses such as CPV or the AAVs (307-310). Following the translation in the cytoplasm, the VPs might form trimers before being translocated into the nucleus and potentially to the nucleolus, where capsid assembly occurs (307-310). For capsid formation, 20 trimers come together, generating the icosahedral particle that consists of 60 capsid proteins. The VPs interact with each other via specific 2-, 3-, and 5-fold amino acid interactions to assemble an empty capsid, enclosing an interior volume of ~2,500 nm³, which is used in a subsequent step to package the HBoV1 genome (311); however, the exact amino acids involved are not yet identified. The arrangement of the VPs in the capsid results in the formation of pores at the fivefold axes of ~9 Å in diameter (Fig. 10), which connect the interior of the capsid to the exterior environment and likely represents the route of genome packaging for the capsid (312, 313). Furthermore, the HBoV1 capsid displays protrusions surrounding the threefold axes that are smaller in comparison to other viruses of the Parvovirinae and even other bocaviruses (314, 315). The most depressed regions of the capsids are found at the twofold axes and around the fivefold axes, which are separated by a raised region termed the 2/5-fold wall (Fig. 10).

FIG 10.

The HBoV1 capsid structure. Top: a side view of the HBoV1 VP structure is shown as a ribbon/coil diagram with the β-strands in blue and α-helices in red. The N- and C-termini, the surface loops, and secondary structures are labeled. The positions of the icosahedral symmetry axes are indicated. Bottom: the arrangement of three VPs around the threefold axis and five VPs around the fivefold axis, viewed from the exterior side, forms the protrusions and pores of the capsid, respectively. Their approximate location is depicted by a triangle and pentagon on the surface of the HBoV1 capsid, which is assembled from a total of 60 VPs. The capsid surfaces are colored blue to red, according to the distance to the capsid’s center as indicated by the scale bar. VP, viral protein.

The HBoV1 capsid structure of VLPs, composed of only VP3, was determined by cryo-electron microscopy (cryo-EM) at 2.8 Å resolution (306). Capsid structures with and without VP1 of other parvoviruses have been shown to be indistinguishable, and only the shared VP region of the major capsid protein was observed (316, 317). This is likely also the case for HBoV1 and other bocaviruses. For HBoV1, structural ordering starts with glycine 162, which is located in the center of the glycine-rich motif and situated under the fivefold pore in the interior of the capsid (Fig. 10). Thus, the VP1u region and the VP1/2 common region are also believed to be placed in the interior of the HBoV1 capsid. Following glycine 162, the VP structure is ordered to its C-terminus and contains an eight-stranded β-barrel, which is formed by the two β-sheets βBIDG and βCHEF (306). In each sheet, the β-strands run antiparallel to each other. This fold is also referred to as the jelly-roll motif and is observed in virtually all icosahedral capsids (311). In addition to the eight β-strands, the HBoV1 VP, as most parvoviruses, possesses an additional β-strand A, which is positioned antiparallel to β-strand B (Fig. 10). Another conserved feature of the parvoviruses is the α-helix A, located between the β-strands C and D. A unique feature of the bocavirus capsids, including HBoV1, is an additional α-helix B between β-strands E and F (306, 314, 318-320). While the β-strands line up the interior surface of the capsid, the loops between them form the surface of the capsid. These are named after the strands that connect them (e.g., the DE-loop is located between the β-strands D and E). The longest loop is the GH-loop with a length of 203 aa, followed by the EF-loop with 71 aa. The GH-loop contains multiple sub-loops and is located primarily in the threefold region of the capsid. At the fivefold axis, the DE-loops from five VPs shape the cylindrical channel (Fig. 10). The HI-loop and partially the BC-loop contribute to the depressed area surrounding the fivefold pore. Finally, the 2/5-fold wall and the 2-fold depression are formed by the combination of the BC and EF-loops, as well as the C-terminal residues following the βI-strand.

Structural comparison of other bocaviruses and the variable regions

To date, the capsid structures of nine bocaviruses have been determined either by X-ray crystallography in the case of bovine parvovirus (320) or by cryo-EM for all the remaining viruses (306, 314, 318, 319). Despite aa sequence identity as low as ~40% for the VP of different members of this genus, the overall VP structure is highly conserved with ≥~80% of the VP structure aligned (Cα positions ≤ 2 Å apart) (Fig. 11). The highest structural and aa-sequence diversity was observed in the surface loops. Here, 10 variable regions (VRs) were defined where at least two consecutive amino acid Cα positions diverge ≥2 Å. The VR-I is located in the BC-loop, VR-II in the DE-loop, VR-III in the EF-loop, VR-IV to -VIII in the GH-loop, VR-VIIIB in the HI-loop, and VR-IX in the C-terminal stretch following the βI-strand (Fig. 10 and 11).

FIG 11.

Variable regions of bocaviruses. This figure shows a superposition of the bocavirus VP structures determined to date. The regions of structural variability are indicated, and the residue ranges for HBoV1 are provided. PBoV, porcine bocavirus; RBoV, rat bocavirus.

The lowest level of structural variability is observed for the fivefold region, formed by VR-II, VR-VIIIB, and partially VR-I. For the parvoviruses, this region has been suggested as the location of genome packaging mediated by the NS proteins interacting with the fivefold channel (312, 321). These proteins show less variability compared to the parvovirus capsid proteins, potentially limiting structural variability of the fivefold region.

During the viral life cycle, the capsid of HBoV1 is exposed to mild acidic conditions in the respiratory tract ranging from pH 5.5 to pH 6.7, while trafficking through the endo-lysosomal pathway following host cell entry, ranging from pH 4.0 to pH 6.0 (322, 323). In these environments, the HBoV capsid remains intact but undergoes conformational changes, particularly in VR-I and VR-V (319). These rearrangements may facilitate interactions with host cell factors during the infection process, albeit to date, no host cell receptors have been described for HBoV1. In contrast, binding assays performed for HBoV1 excluded sialic acid, galactose, and heparan sulfate proteoglycan as potential glycan receptors, which are commonly utilized within the Parvovirinae, including other bocaparvoviruses (324, 325). Additionally, structural changes at the VP N-termini at the fivefold axis were observed at low pH conditions analogous to the continuous acidification of the endosome following receptor-mediated endocytosis (Fig. 6, Step 4), which in wild-type capsids may result in the externalization of the VP1u region from the interior of the capsid through the fivefold channel (319). This externalization enables the HBoV1 virion to utilize its enzymatic PLA₂ domain within VP1u to cleave phospholipids of the endosomal membrane and escape from it to traffick to the nucleus (300, 326).

Due to its high prevalence, HBoV1 DNA has been isolated several times from human samples all over the world (327-332). The capsid proteins of these variants differ in one or more amino acids from the Salvador 1 isolate that has been fully sequenced (215, 223). Compared to the Salvador 1 reference sequence, 88 unique amino acid substitutions were identified among the 100 most similar HBoV1 capsid variants in the NCBI database. While 16 of these amino acid substitutions are located in the VP1u region (Fig. 12), none of these affect the critical Ca²⁺BS, the PLA₂ domain, or the BR1 (Fig. 9). This is also true for the six amino acid substitutions in the VP1/2 common region of BR2 and BR3. For this reason, these exchanges may only have a minor impact on HBoV1 infection but require further experimental investigation.

FIG 12.

Natural HBoV1 capsid variants. Amino acid substitutions of natural HBoV1 capsids deposited in the NCBI database relative to the HBoV1 Salvador 1 isolate are indicated on HBoV1 VP. The location of the VP1u, the VP1/2-common, and variable regions (yellow) and the β-strands (blue) is shown on the VP diagram. The residues within the determined VP structure above the schematic are located on the exterior surface, whereas the residues below are either buried or located at the interior surface of the capsid.

The remaining 66 amino acid changes are situated in VP3, with 28 of these positioned at the exterior surface of the capsid and primarily within the VRs. These substitutions have the potential to alter the receptor binding and antigenic properties of the HBoV1 capsids. This is not true for one of the most frequently observed S590T substitution (333), which is located within the HI loop or VR-VIIIB. HBoV1 variants with either serine or threonine in this position did not change their transduction ability or immunoreactivity, but capsids with a serine in position 590 instead of threonine were found to be associated with lower viral loads in infected patients (328). However, when these HBoV1 variants were used in vector manufacturing settings, the produced vector genome titers of the S590 variant exceeded those of the T590 variant by ~2.4-fold (328). These opposing results may be due to the utilization of the AAV2 Rep proteins during recombinant HBoV1 vector production.

Interestingly, a significant number of amino acid substitutions are located at the interior surface or are embedded within the capsid shell (Fig. 12). Among the variants compared, the Salvador 1 isolate was the only capsid having a lysine in position 613, whereas every other HBoV1 isolate in the NCBI database carries a glutamic acid in the same position, which is near the twofold axis on the interior side of the capsid. Residues in these locations have the potential to affect capsid assembly or the interaction with the packaged genome. However, capsid assembly and genome packaging remain insufficiently studied for the bocaviruses to date.

Mapping the antigenic epitopes of HBoV1

During an infection with HBoVs, the adaptive immune system generates antibodies against the viral proteins, particularly targeting the capsid, to eliminate the virus and to prevent future re-infections. Antibodies against HBoV1 are observed in a high percentage of the human population (32, 33, 36) and may partially be responsible for numerous HBoV1 capsid escape variants due to selection pressure (Fig. 12). While these antibodies are favorable to combat pathogenic viruses, they are detrimental to the viral vectors being used in gene delivery applications. Thus, studies to characterize the antigenic epitopes of the HBoV capsids were conducted to develop engineered HBoV capsids that are capable of circumventing pre-existing neutralizing antibodies.

To date, the epitopes of four mouse monoclonal antibodies (mAbs) have been mapped to the HBoV1 capsid (Fig. 13) at resolutions ranging from ~9 to 16 Å (334). Three of these antibodies, 4C2, 9G12, and 12C1, bind very similarly to the threefold protrusions, contacting the capsids’ VR-IV, V, VIII, and partially VR-I and VI. While no cross-reactivity to HBoV2-4 nor GBoV1 was reported for mAbs 4C2 and 9G12, mAb 12C1 also bound to the GBoV1 capsid (318, 334). Unlike the other antibodies, mAb 15C6 binds to the fivefold region of the HBoV1 capsid involving VR-I, -II, and -VIIIB and cross-reacts with HBoV2, HBoV4, and GBoV1 (318, 334). This cross-reactivity of 15C6 is due to the higher conservation of the fivefold region mentioned above. Furthermore, these fivefold binding antibodies are likely responsible for the challenge in serodiagnosis to specifically identify HBoV1 infections. Whether the antigenic epitopes determined with the mouse mAbs faithfully simulate the human antibody response needs to be investigated. Similar to HBoV1, mouse mAbs appeared to primarily bind the threefold region of the AAV9 capsid (335). However, a recent study showed that the twofold region of the AAV9 capsid is the main target for human mAbs, which could also be the case for HBoV1 (336).

FIG 13.

Mapping the antigenic epitopes of HBoV1. Cryo-EM maps of HBoV1 in complex with the Fabs of four mouse monoclonal antibodies (4C2, 9G12, 12C1, and 15C6) are shown. The capsid portion is gradiently colored according to the scale bar, and the Fabs are colored brown. Below the overall binding region, contacted VRs and cross-reactivity of the antibodies are provided.

HBoV1 IN GENE THERAPY

HBoV1 is a helper virus for AAV2 replication and infection

HBoV1 can provide helper functions for productive AAV2 infection in polarized HAE (275). Upon transfection of both HEK293 and HeLa cells, the transfected HBoV1 dsDNA genome rescues the AAV2 dsDNA genome replication at a level similar to the Ad helper genes, E2a, E4orf6, L4-22K, and VA RNA (encoded in the pHelper plasmid) (337, 338). The minimal HBoV1 genes essential to facilitate AAV2 DNA replication after transfection and virus production are NP1, BocaSR, and NS4, whereas during AAV2 infection, NS2 is required for AAV2 DNA replication and progeny production (275, 337).

HBoV1 helper for recombinant AAV production

Expression of the HBoV1 genes NP1, NS2, and BocaSR in HeLa cells can fully rescue replication of a full-length AAV2 clone, whereas expression of Ad helper genes, E2, E4orf6, and VA, does not (275). The presence of the AAV2 inverted terminal repeat (ITR) in front of the p5 promoter likely facilitates the expression of Rep78/68, which transactivates the downstream p19 and p40 promoters (339). Importantly, when HBoV1 NP1 and NS2 genes are combined with Ad helper genes in the pABHelper plasmid, their expression increases rAAV2 genome replication and significantly enhances rAAV2 vector production in HEK293 cells by more than twofold (337). Thus, pABHelper is a novel synergistic helper plasmid for rAAV vector production.

HBoV1 capsid-pseudotyped recombinant rAAV vector

The rAAV2 genome can also be pseudopackaged into the HBoV1 capsid to assemble a chimeric parvoviral vector, rAAV2/HBoV1. The HBoV1 capsid can package an over-sized rAAV genome of up to 5.8 kb without sacrificing packaging efficiency (297). Inherited from the airway epithelia tropism of HBoV1, the rAAV2/HBoV1 vector can efficiently transduce human airway epithelia. By carrying a full-length CFTR (cystic fibrosis transmembrane conductance regulator) cDNA of 4.5 kb, this vector could rescue approximately one-third of the CFTR function in CF phenotype exhibiting HAE (297). The rAAV2/HBoV1 vector also demonstrated efficient transduction of the distal airways in the lungs of both newborn and juvenile ferrets (340). Currently, this pseudotyped vector is undergoing preclinical development for lung gene therapy of cystic fibrosis in ferret models. Additionally, the HBoV1 capsid can package a recombinant HBoV1 genome as an rHBoV1 vector (297). These properties of HBoV1-based vectors provide a new tool for gene delivery in airway applications.

An NS protein-free rAAV2/HBoV1 vector production system using co-transfection of three plasmids was established in HEK293 cells (341), which next to the vector plasmid includes one trans helper plasmid encoding both HBoV1 VP1 and AAV2 Rep proteins, and another encoding VP2, VP3, and Ad5 helper genes. The system yielded rAAV2/HBoV1 vectors at a level similar to that of conventional recombinant AAV2 vectors, while retaining a high transduction activity in HAE-ALI. In addition, HBoV1 capsids produced in insect Sf9 cells can cross-package an rAAV2 genome (342), with NP1 playing an enhancing role as it increases total particle yields. The insect Sf9-based vector production system (343) generates more empty particles (~50%) than the mammalian cell-based system, although it produces an overall high yield. In this context, it should be considered that the Sf9 cell culture can be easily scaled up, and empty capsids can be separated by anion chromatography. Thus, the Sf9-based rAAV2/HBoV1 vector system is very promising regarding large-scale vector production.

Alternative natural or synthetic BoVs as vectors for gene delivery

The ground-breaking 2013 study by Yan and colleagues (297) that reported proof-of-principle for the feasibility to pseudotype AAV2 genomes with capsids of HBoV1 raised the question whether alternative bocavirus variants could also be used to encapsidate and deliver AAV genomes (Fig. 14).

FIG 14.

NS-free rAAV2/HBoV1 vector production system in HEK293 cells (341). The rAAV2 transgene plasmid (pAAV2.CMVmCherry-F5tg83:uc) carries the gene of interest flanked by AAV2 ITRs, expressing mCherry under the CMV promoter linked to the F5tg83 synthetic promoter (344) driving firefly luciferase (Luc). The pCMVHBoVP1-AAV2Rep plasmid is a dual-expression construct encoding codon-optimized HBoV1 VP1 (optVP1) and AAV2 Rep78/52 (Rep2) under separate CMV-driven cassettes. The pAd4.1-CMVoptVP2/3 plasmid expresses codon-optimized HBoV1 VP2/3 (optVP2/3) and adenoviral helper genes (Ad5 VA RNA, E2A, and E4) required for AAV production. CMV, cytomegalovirus immediate-early promoter; mCherry, red fluorescent protein; F5tg83, synthetic promoter; Luc, firefly luciferase; polyA, polyadenylation signal; and bGHpA, bovine growth hormone polyadenylation signal.

The exploration of this issue was pushed through several publications from the AAV field. These collectively show that AAV2 genome cross-packaging into capsids from other AAV serotypes can drastically alter gene delivery efficiency, organ or cell-type specificity, reactivity with neutralizing antibodies, manufacturability, physical particle properties, and/or in vivo safety of the ensuing vectors (345).

For this reason, Fakhiri and colleagues (250) created bocaviral helper plasmids encoding the capsid genes of HBoV2, -3,-4, and GBoV to cross-package AAV vectors into the respective capsids. To this end, the authors streamlined the protocol for rAAV/BoV vector production by replacing the two separate helper plasmids expressing AAV2 rep or adenoviral genes with the plasmid pDGΔVP, which combines both functions in a single backbone. Furthermore, using AAV vector genomes that would be too large (i.e., over 5 kb) for packaging into AAV capsids, Fakhiri et al. demonstrated that the HBoV1 capsid can accommodate conventional single-stranded AAV genomes with a size of up to 6.1 kb, which is congruent with extending earlier data by Yan and colleagues. The extended packaging capacity of the HBoV1 capsid was also confirmed for self-complementary (sc)AAV genomes, as efficient packaging of up to 3.2 kb scAAV DNA was measured (250).

Not only the HBoV1 capsid, but also those of HBoV4 or GBoV, mediated robust transduction of HAE and CuFi-8 cells when delivered via the apical side of the HAE cultures. However, these three vectors differed in their preference for specific lung cell types. While the HBoV1 and GBoV capsids mostly target basal, ciliated, and club cells, the HBoV4 capsid preferred basal cells and was less effective in ciliated cells. Taking into account additionally HBoV2 and -3, the five vectors also differed in their ability to transduce human lung organoids, as HBoV2 and -3 only functioned upon direct microinjection into the lumen of the organoids (250). In contrast, HBoV1 and HBoV4, as well as GBoV, also enter the cells when the organoids were broken mechanically. Although HBoV2 and HBoV3 were least effective in the lung cell culture systems (CuFi-8 and HAE), all five BoV variants transduced primary human hepatocytes comparably well, whereas HBoV4 and GBoV outperformed the other capsids in human CD4+ T cells.

Another important finding with respect to clinical translation of BoV variants in humans was that HBoV4 and GBoV largely evaded neutralization with pooled human antibodies (IVIg) (250), which enables the promising application of vectors comprising these two capsids in human patient cohorts.

The authors of this 2019 study also provided the first report of the ability to create synthetic BoV capsids not found in nature. Therefore, they explored the technology of “DNA family shuffling,” in which related viral capsid genes (here, HBoV1- to -4 and GBoV) are first fragmented enzymatically and then reassembled in novel patterns via series of PCR reactions. Previously, another study used this technique in the AAV context and documented its power to yield novel synthetic parvoviruses that exhibit unprecedented cell-targeting specificities and efficiencies of transgene delivery (346), as exemplified with the AAV-DJ capsid that merges portions of AAV2, -8, and -9 and that outperforms all parental viruses while largely evading neutralization (347).

The successful adaptation of this powerful methodology to BoV was an important step in the emerging field of BoV vector technology (250), as it paves the way for the future creation and high-throughput in vivo screening of complex libraries composed of engineered BoV capsid variants. Additional optimism is provided by recent, as-of-yet unpublished work showing the possibility to also adapt peptide-display technology to the BoV vector field and to concurrently dissect BoV biology as well as optimize BoV capsids through rational capsid engineering based on structural data (348, 349).

A follow-up study in 2020 highlighted the manifold benefits that can be gathered for the BoV vector field from an expanded understanding of natural BoV biology (328). Briefly, in this work, the authors cloned, sequenced, and characterized 29 distinct HBoV1 capsid variants derived from a cohort of HBoV1-positive patient samples. This identified a hot spot at HBoV1 capsid position 590, of which mutation from threonine to serine significantly boosted titers of AAV/HBoV1 vectors carrying this alternative residue. This finding may be influenced by the utilization of the AAV2 Rep proteins during HBoV1 vector production as already mentioned above. In addition, other critical amino acids were identified that govern capsid assembly, transduction, recombinant DNA packaging, or antibody neutralization. When 16 of the newly discovered HBoV1 variants were tested and compared in HAE, this revealed not only differences between the donors but also between the HBoV1 variants within the individual donors. This is again pivotal for the future translation of BoV vector technology into human patients, as it implies both the necessity as well as the ability to tailor BoV capsids to target cells and patients.

OPEN QUESTIONS AND CHALLENGES

The role of HBoVs in cancer needs more investigations and further studies. While there are indications of an active contribution of HBoV to cancerogenesis, reliable proofs are still missing. This is because limited access to clinical specimens and matching control tissues, as well as the lack of an adequate in vivo model to study HBoV as a putative carcinogen, impedes processing of this issue. Until the final clarification, a tumor tropism of HBoV as the opposite effect may still be relevant or even more important. A possible tumor tropism could be addressed with regard to the development of viral vectors for cancer treatments with high specificity for lung and colorectal tumors. In this context, already published findings on rAAV/BoV vectors should be taken into account, although future studies will need to show that bocavirus gene therapy vectors are indeed viable alternatives to treat chronic lung diseases such as lung fibrosis, COPD, or cancer.

One major task in the future is the identification of the respective specific HBoV receptor(s). This molecular task is not yet completed but would enable more simplified cell culture models based on monolayers rather than on expensive air-liquid interface cultures boosting HBoV research. Moreover, the lack of in vivo models hinders the elucidation of the role of HBoV in heart diseases and idiopathic lung fibrosis, which have been clinically described in some cases to be possibly linked with HBoV infections.

Biographies

Professor Dr. Oliver Schildgen is a molecular biologist, certified clinical virologist, and professor of Virology at the Private University of Witten/Herdecke. He completed his Diploma thesis in Cologne (Lab of Hans-Joachim Eggers) and Ph.D. thesis in Essen (Michael Roggendorf). Afterward, he worked at the University of Bonn (Bertfried Matz, DNA-replication of the Herpes simplex Virus). He was the coordinator of the FP6 RespViruses project and received the Habilitation and the venia legend. He was Editor in Chief of Respiratory Research, is still Editor in Chief of Reviews in Medical Microbiology, Section Editor of Medicine, Editor of PLOS One, Cancers, and several other journals. He received Wolfgang-Stille-Award in 2014, the Hygiene-Award Rudolf-Schülke-Foundation in 2013, ESCV-Abbott-Diagnostic-Award in 2011, ÖGHMP-Meteka-Award in 2010, DGKH-Science-Award in 2010, GfV-Clinical-Virology-Award in 2009, DGPI-Science-Award in 2009, Commendation of the Lord Mayor of the City of Cologne for Voluntary Commitment in 2014, and teaching award of the University of Witten in 2024.

Jianming Qiu earned his B.S. in Veterinary Medicine and M.S. in Biochemistry from Zhejiang University, China, and completed his Ph.D. in Virology at the National Institute for Viral Disease Control and Prevention, China CDC, in Beijing. He received postdoctoral training in parvovirology at the Hematology Branch of the National Institutes of Health (NIH) in Bethesda, MD, and at the Department of Molecular Microbiology and Immunology at the University of Missouri—Columbia. Dr. Qiu is currently a Professor in the Department of Microbiology, Molecular Genetics, and Immunology at the University of Kansas Medical Center in Kansas City, KS. His research focuses on the molecular biology and pathogenesis of human parvoviruses HBoV1 and B19, as well as the development of adeno-associated viruses (AAVs) as vectors for gene therapy. He is a member of the ICTV Parvoviridae Study Group and a Fellow of the American Academy of Microbiology.

Dr. Mario Mietzsch is a structural and molecular virologist. He studied biology at the Freie Universität Berlin and received his PhD in the lab of Prof. Dr. Regine Heilbronn at the Charité - Universitätsmedizin Berlin, Germany, where he worked on the development of a scalable production system for AAV vectors and the molecular characterization of the AAVs including their glycan interactions. After his PhD, he joined the lab of Dr. Mavis Agbandje-McKenna at the University of Florida as a postdoc being trained in structural virology, using cryo-electron microscopy and 3D-image reconstruction also expanding to other parvoviruses, including the bocaviruses. Currently, he is a Research Assistant Professor in the Department of Biochemistry and Molecular Biology at the University of Florida.

Tobias Allander, M.D., Ph.D., is a consultant clinical virologist at Karolinska University Hospital and associate professor of virology at Karolinska Institute in Stockholm, Sweden. He studied medicine and completed his Ph.D. at Karolinska Institutet and did postdoctoral training in the Laboratory of Infectious Diseases at the National Institutes of Health, Bethesda, MD, before receiving specialist training in clinical virology at Karolinska University Hospital. His research is since more than 20 years focused on virus discovery and viral metagenomics. He was first to describe human bocavirus in 2005 and has conducted several studies on diagnostic and clinical aspects of the virus. Dr. Allander has served as head of Clinical Microbiology at Karolinska and is currently chairman of the Swedish Society for Clinical Microbiology.

Tuomas Jartti, MD, is a professor of pediatrics in the University of Turku, and pediatrician and pediatrics allergist in the Turku University Hospital. His research has focused on preventing childhood asthma. Molecular virology has played an important role, which has led to discoveries of rhinovirus induced wheezing being an important and early risk factor for childhood atopic asthma. Also, the search for new viruses lead to the discovery of human bocavirus as a third most important virus triggering wheezing, and improved bocavirus diagnostics in collaboration with Dr. Maria Söderlund-Venermo in the University of Helsinki. The investigation of inflammatory mechanisms has revealed unique mechanisms by bocavirus. Virus-virus interaction studies have shown that bocavirus may have anti-inflammatory properties, which could have clinical significance.

Priv.-Doz. Dr. Verena Schildgen is a molecular biologist with focus on molecular pathology. She studied biology at the University of Cologne and performed her Diploma thesis on recombinant T-cell receptors in cancer therapies. After her PhD in myelodysplastic syndromes, she started working on acute and chronic respiratory infections during the FP6 RespViruses project as a project manager. The scientific work of Verena Schildgen was honored with the International Rudolf-Schülke-Award (2013), the Wolfgang-Stille-Award of the Paul-Ehrlich-Society (2014), and the Publication Award of the German-Austrian-Swiss Society for Medical Mycology (2014). She received her habilitation and venia legendi for Molecular Pathology at the University of Witten/Herdecke and is currently working on molecular diagnostics in the Institute of Pathology at the Kliniken der Stadt Köln.

Dirk Grimm is a professor at Heidelberg University where he heads the section “Viral Vector Technologies” whose members focus on parvovirus biology and vector engineering. In his own 30 years of research in the parvovirus community, Dr. Grimm made numerous pivotal contributions to the field of human gene therapy using vectors derived from AAV or Bocavirus. These include the first report of a helpervirus-free AAV production system, the introduction of DNA family shuffling technology into the AAV space, or the use of DNA/RNA barcoding for the identification of highly muscle-specific AAV capsids. Recently, Dr. Grimm’s team has generated vectors based on different Bocavirus isolates and demonstrated their usefulness for gene delivery to a variety of transformed or primary cells. The current focus of his laboratory is on the high-throughput screening of Bocavirus libraries in various species with the aim to identify novel and superior viral vectors for human gene therapy.

Maria Söderlund-Venermo is adjunct professor, specialist in clinical microbiology, researcher and lecturer at the University of Helsinki Medical Faculty, Department of Virology. She obtained her PhD in 1996 at the University of Helsinki, Helsinki, Finland. She was a visiting scientist at the University College London, London, UK, and did her postdoctoral training at the Departments of Molecular Microbiology and Immunology and Veterinary Pathobiology, University of Missouri—Columbia, Columbia, MO, USA. She is the current President of the World Society for Virology (WSV). Her other positions of trust include the Finnish Society for the Study of Infectious Diseases (secretary 2007-10), the Finnish-Norwegian Medical Foundation (secretary since 2016, and chair since 2025) and member of the International Committee on Taxonomy of Viruses (ICTV) Parvoviridae Study Group. Her research interests include both clinical and molecular virology, with the main focus on B19V, HBoVs, and the human protoparvoviruses.