Introduction

Coxsackievirus A16 (CV-A16) is a positive-sense, single-stranded RNA virus classified under the genus of Enterovirus in the family Picornaviridae. The virion is about 30 nm in diameter and is composed of a non-enveloped capsid surrounding a naked RNA genome of about 7500 nucleotides. CV-A16 is transmitted predominantly via the fecal–oral route, but can also spread through contact with virus-contaminated oral secretions, vesicular fluid, surfaces or fomites, and in respiratory droplets. CV-A16, together with Enterovirus A71 (EV-A71), is one of the most common enteroviruses that cause epidemics of hand-foot-and-mouth disease (HFMD) involving thousands of children worldwide. Most prevalent in children younger than 5 years old, HFMD usually presents with typical clinical manifestations such as fever, painful sores in the mouth, and a rash with blisters on hands, feet and also buttocks. Although often mild and self-limiting, HFMD can be associated with neurological complications. In some CV-A16 infections, severe and often fatal neurological complications including aseptic meningitis, encephalitis, acute flaccid paralysis [1,2,3,4,5], and cardiopulmonary complications such as pneumonitis, pulmonary edema, pulmonary hemorrhage, or myocarditis [6,7,8,9,10] have been reported. The central nervous system (CNS) pathology of CV-A16 neurological complications is unknown since no human autopsy studies has ever been done [11]. A brain magnetic resonance imaging (MRI) of a probable case of CV-A16 rhombencephalitis showed distribution of hyperintense lesions in pons (tegmentum) and cerebellum [1].

Previously, a few CV-A16 animal models, including murine, gerbil, and tree shrew models have been developed for pathogenesis studies and vaccine efficacy evaluation [12,13,14,15,16,17,18], but none of these models could reproduce the important features observed in human HFMD, such as typical palmar and plantar skin vesicles, and buccal mucosa and tongue ulcers. Although some of these animal models, including a mouse model previously described by our group [19], had demonstrated viral antigens in the cerebral cortex, cerebellum, brainstem, and spinal cord, confirming CV-A16 neurotropism [14, 15, 17, 18], CNS neuropathology has not been described in sufficient detail.

Based on an earlier work by our group on an orally infected EV-A71 hamster model [20] that developed squamous lesions strikingly reminiscent of HFMD, and encephalomyelitis, we used the same approach and developed a hamster model using a CV-A16 mouse-adapted virus strain (MAVS). In this study, we report the successful development of a novel orally infected, CV-A16 hamster model that manifests both HFMD-like lesions and encephalomyelitis to investigate tissue tropism and pathology, viral replication kinetics, and viral spread within the host. The data obtained further extend our knowledge of CV-A16 infectious disease pathology in general and the CNS pathology in particular. Moreover, evidence of significant oral and fecal viral shedding and animal-to-animal transmission that recapitulated human fecal–oral/oral–oral routes of CV-A16 transmission, establishes this hamster model as a useful model to investigate person-to person transmission.

Materials and methods

Virus adaptation, stock preparation, and titration

The CV-A16 MAVS used in this study was prepared by serial passages, by subcutaneously inoculating 100 µl of a clinical strain (CV-A16/N132; viral stock titer: 3.56 × 106 CCID50/ml) into each 7-day-old ICR mouse. After each round of infection when animals (n = 3–5 animals per group) developed severe signs of disease (e.g., paralysis), skeletal muscles were harvested and pooled for the next passage in another group of animals using a homogenate suspension (10% wt/vl) prepared with Dulbecco’s Modified Eagle’s Medium (DMEM) (Sigma-Aldrich, USA) maintenance medium (MM) supplemented with 2% fetal bovine serum (FBS) (JR Scientific, USA).

The MAVS derived from the fifth passage in mice was used for all infection experiments, after increasing the viral loads by culturing in confluent Vero cells maintained in DMEM MM supplemented with 2% FBS, and kept until > 70% of cells showed cytopathic effect (CPE). After three freezing-thawing cycles, the culture was clarified by centrifugation at 4000 revolutions per minute (rpm) at 4 °C for 10 min. The supernatant was filtered through a 0.22 µm pore-size filter (Minisart; Sartorius, Germany), aliquoted and stored at −80 °C before use. The virus titer expressed as CCID50 was determined by a standard microtitration assay in Vero cells using the Spearman–Karber method as described previously [21, 22].

Animal infection experiments

All the animal experiments were approved by the Institutional Animal Care and Use Committee, Faculty of Medicine, University of Malaya (Ethics number: 2013-12-03/PATHO/R/WKT and 201617-12-31/PATHO/R/WKT) and performed according to its guidelines.

Hamster susceptibility to CV-A16 MAVS infection

Groups of 7-day, 14-day and 21-day-old hamsters (n = 7–10 animals per group) were orally infected with 100 µl of CV-A16 MAVS (viral stock titer: 1.12 × 105 CCID50/ml) per animal to identify the most susceptible age group to infection. Two hamster controls for each group, housed in separate cages, were mock-infected with phosphate-buffered saline (PBS). All hamsters were frequently monitored several times daily for signs of infection up to 14 days post infection (dpi). Hamsters were euthanized by isoflurane inhalation as soon as they developed severe signs of disease (e.g., paralysis). The group of 7-day-old hamsters was found to be the most susceptible to infection and therefore used for subsequent infection experiments in this study.

Kinetic study of CV-A16 MAVS infection in 7-day-old hamsters

Four groups of hamsters (n = 8 per group), designated as 1–4 dpi groups, were orally infected with 100 µl of CV-A16 MAVS per animal (viral stock titer: 1.12 × 105 CCID50/ml), with two hamster controls per group mock-infected with PBS as before. Animals were closely observed for signs of infection, and sequentially sacrificed following isoflurane inhalation euthanasia over 1–4 dpi.

Whole carcasses of four hamsters from each designated group were randomly chosen to be adequately fixed in 10% neutral buffered formalin for 3–5 days. The tissues were prepared as standard 8–9 cross-sectional blocks, routinely processed and paraffin embedded. In this way, most CNS and non-CNS organs were available for study. Each animal was individually numbered as: A1–A4 (1 dpi group); A5–A8 (2 dpi group); A9–A12 (3 dpi group) and A13–A16 (4 dpi group). The other four hamsters from each designated group were sacrificed to obtain tissues for viral titration.

Histopathological analysis

Four micrometer thick tissue sections were cut from each paraffinized block and mounted on 3-aminopropyltrietoxysilane coated slides, dried overnight, and stained with hematoxylin and eosin (H&E) for light microscopy. Further sections were cut for immunohistochemistry (IHC) to detect viral antigens and in situ hybridization (ISH) to detect viral RNA.

IHC

The IHC was performed as previously described [23]. Briefly, the tissue sections were deparaffinized with xylene, washed and rehydrated through graded concentrations of ethanol in water. Endogenous peroxidase was blocked, followed by antigen retrieval (30 min, Tris-EDTA buffer, pH 9.0) and normal goat serum blocking (Dako, Denmark). Enterovirus Blend mouse monoclonal antibody (3321 Light Diagnostics™, Merck Millipore, USA) diluted 1:100, was applied on the tissue sections overnight at 4 °C. This was followed by horseradish peroxidase (HRP)-conjugated secondary antibody (ENVISION; Dako, Denmark) incubation at room temperature for 30 min, and staining with 3,3′-diaminobenzidinetetrahydrochloride (DAB) chromogen (Dako, Denmark). The sections were counterstained with hematoxylin and mounted in DPX (BDH, England).

ISH

A modified ISH technique was applied as previously described using a digoxigenin (DIG)-labeled, CV-A16-specific DNA probe [24]. Briefly, deparaffinized and rehydrated tissue sections were pretreated with HCl for 20 min followed by proteinase K digestion. The probes were diluted in freshly prepared hybridization solution, then applied onto sections to incubate for 16–20 h at 42 °C in a moist chamber. Hybridization was detected using an anti-DIG antibody conjugated to alkaline phosphate that could react with nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) substrate (Roche, Germany). The sections were counterstained with Mayer’s hematoxylin and mounted in aqueous mounting medium.

Viral titration

From the four hamsters in each dpi group sacrificed for viral titration, blood were collected by cardiac puncture and centrifuged at 2500 rpm for 10 min to separate sera. Other organs/tissues including brain, limb skeletal muscles, brown adipose tissue (interscapular), heart, lung, liver, kidney, intestine, and spleen were harvested using different sets of sterile forceps to minimize cross-contamination. Tissues were weighed and immediately frozen at −80 °C for later use. Serum and selected frozen tissues (brain, skeletal muscle, brown adipose tissue, and heart) shown to be IHC-positive for viral antigens in paraffinized tissues, were homogenized in PBS to 10% (wt/vol) and viral titrated with the microtitration assay as described above.

Viral transmission study of CV-A16 MAVS infection in 7-day-old hamsters

In a separate experiment, a group of hamsters (n = 10) from one single mother hamster were used to investigate intra-family CV-A16 transmission in the ratio of 1:1. Five “index” hamsters (animals numbered as B1–B5) were each orally infected with 100 µl of CV-A16 MAVS (viral stock titer: 1.12 × 105 CCID50/ml) and returned to the same cage together with their mother and non-infected, “contact” littermates (animals numbered as B6–B10). The index and contact hamsters were observed daily for signs of infection up to 14 dpi and days post exposure (dpe), respectively. Euthanized sick or dead hamsters were fixed in 10% neutral buffered formalin, sectioned, processed and paraffin embedded for histopathological analysis as before.

Virus isolation from oral washes and feces

Oral washes and feces were collected daily from index and contact hamsters up to 14 dpi/dpe or earlier, if they developed severe infection and had to be euthanized. The oral cavity was rinsed a total of six times consecutively, each time with 50 µl of PBS to obtain pooled oral washes from each animal. Whenever available, fecal pellets expelled spontaneously from the anus during handling were also collected from each hamster using sterile forceps and weighed. The oral washes and fecal pellets collected were stored at −80 °C for processing later.

Oral washes were treated with chloroform (1:10), while fecal samples were homogenized with PBS (10% wt/vol) before chloroform treatment. Treated samples were vortexed for 15 min, and centrifuged at 3500 rpm for 20 min in 4 °C. For virus isolation, the supernatant from each sample was diluted in DMEM MM (dilution 1:10), and propagated on confluent Vero cells plated in 96-well plates and observed for CPE after 7 dpi.

Viral RNA detection in oral washes and feces

Reverse transcription polymerase chain reaction (RT-PCR) was performed to confirm the CPE results obtained from viral isolation assay. After three freezing-thawing cycles, the viral culture supernatants were collected from the 96-well plates for total RNA extraction using High Pure Viral RNA Isolation Kit (Roche, Germany) following the manufacturer’s protocols. Reverse transcription was done using RevertAid™ First Strand cDNA Synthesis Kit (#K1622, Thermo Scientific, USA) (Supplementary Table 1), followed by PCR in a Veriti™ 96-well thermal cycler (Applied Biosystems, USA) using CV-A16-specific forward primer (CV-A16-F: 5′-GAGGACATTGAGCAGACAGCT-3′) and reverse primer (CV-A16-R: 5′-CACTTGAGCTGGTGGGTCCGT-3′), which amplifies 600 nt of the genome (Supplementary Table 2). The PCR procedure was performed for 3 min at 95 °C, followed by 35 cycles, with each consisting of 15 s at 95 °C, 30 s at 59 °C, and 1 min at 72 °C; with a final extension step of 5 min at 72 °C. The presence of amplicons was confirmed by 1.5% agarose gel electrophoresis stained with EtBr in TAE buffer and visualized on a UV transilluminator (Alpha Imager HP, USA).

Statistics

Mann–Whitney U test was used to evaluate the significance of the differences in the mean viral titers obtained in the viral titration assays. If P value is < 0.05, statistical significance is achieved.

Results

Susceptibility of hamster to CV-A16 MAVS

All ten infected 7-day-old hamsters (100%) consistently developed signs of infection such as closure of eyes, hunchback posture, reduced mobility and paralysis, progressing to a moribund stage between 3 and 4 dpi. However, only 4 out of 7 (about 57%) of the 14-day-old hamsters exhibited signs of infection between 5 and 11 dpi, and 5 out of 10 (50%) of the 21-day-old hamsters did so between 5 and 13 dpi. Mock-infected hamsters remained healthy throughout the experiments. Thus, the 7-day-old hamster model was chosen for further studies.

Kinetic study

No signs of infection were observed in 1, 2, and 3 dpi groups (Animals A1–A12), but in the 4 dpi group, all 4 hamsters showed signs of infection: 2 hamsters (Animals A13 and A15) were found dead, 1 hamster (Animal A14) had limb paralysis and 1 hamster (Animal A16) showed reduced mobility. All mock-infected hamsters remained healthy.

Pathological findings

No viral antigens were detected in hamster tissues sacrificed at 1 dpi (Animals A1–A4). At 2 dpi, focal viral antigens were only detected in skeletal muscles in the tongue, face, neck, shoulder, limbs, paraspinal and pelvic areas (Table 1, Animals A5–A8). In the 3 dpi group, viral antigens were found in squamous epithelial cells lining the oral cavity (Fig. 1a) and tongue in 75% of the animals, while viral antigens in paw (Fig. 1b) and skin epidermal squamous cells were observed in 100% of animals (Table 1). There was only minimal necrosis and inflammation. A few viral antigen-positive neurons (Fig. 1d) were detected in the cervical spinal cord anterior horns in Animals A11 and A12 (50% of animals), but not in other parts of the CNS (Table 1 and Fig. 2). Focal viral antigens were detected in the myocardium (67% of animals), salivary gland acinar cells (Fig. 1c) (25% of animals), and brown adipose tissue (25% of animals). Extensive and dense viral antigens were detected in all skeletal muscles (100% of animals) (Table 1), which correlated with widespread and severe fiber necrosis and myositis. Thus, despite apparent absence of signs of infection, animals in the 3 dpi group had fairly extensive tissue evidence of infection.

Table 1 IHC findings in CV-A16 MAVS-infected hamsters (kinetic study).
Full size table
Fig. 1: IHC detection of viral antigens in CV-A16-infected hamsters (kinetic study).
figure 1

At 3 dpi, focal viral antigens were found in squamous cells in the oral mucosa (a, arrows), paw epidermis (b, arrows), and in salivary gland acinar cells (c, arrows). Viral antigens were detected in cervical spinal cord anterior horn (d, arrow). At 4 dpi, focal viral antigens were found in squamous cells in the tongue mucosa (e, arrow) and hair follicle epithelium (f, arrow). Viral antigens were detected in some neurons in brainstem (medulla/pons) (g, arrow) and the cervical spinal cord anterior horn (h, arrow). Viral antigens were also focally detected in myocardium (i, arrow), intestinal smooth muscle (j, arrows), and brown adipose tissue (k, arrow). The skeletal muscle (l) showed the most intense viral antigens. All are IHC with DAB chromogen and hematoxylin counterstaining (a–l). Scale bars: 10 μm (b, c, g, i, l); 5 μm (a, d–f, h, j, k). Positive neurons in (d) correspond to row V of Animal A11, and (g, h) corresponds to row IV, V of Animal A15 in Fig. 2.

Full size image
Fig. 2: Approximate CNS distribution of viral antigen-positive neurons in CV-A16-infected hamsters.
figure 2

The animals from the kinetic study (A1–A16) and transmission study (B1–B5) are shown. For each hamster, seven standard CNS cross sections are displayed and labeled (left most column), representing from top to bottom, rows I: cerebral cortex/striatum; II: hippocampus/thalamus; III: cerebral cortex/midbrain; IV: cerebellum/medulla/pons; V: cervical spinal cord; VI: thoracic spinal cord; and VII: lumbar spinal cord. Each red dot represents one-positive neuron. No positive neurons were detected at 1–2 dpi. At 3 dpi, one-positive neuron was found in the cervical spinal cord anterior horn of Animals A11 (row V) and two-positive neurons were found in Animal A12 (row V). At 4 dpi, about 9–14-positive neurons were detected in the medulla/pons of all hamsters (Animals A13–A16, row IV), and about 3–15-positive anterior horn neurons in total were detected at different levels of spinal cords (Animals A13–A16, rows V–VII). From 5 to 6 dpi, all dead index hamsters showed positive neurons in the CNS. Similar to 4 dpi, in the brainstem (medulla/pons, row IV), Animals B1, B2, and B5 showed about 6–14-positive neurons while Animals B3 and B4 appeared to show more positive neurons (about 26–29 neurons). In the spinal cord (rows IV–VII), Animals B1–B4 showed a similar number of total positive anterior horn neurons (about 6–15 neurons), but Animal B5 appeared to show much higher number (about 30) of positive neurons. No positive neurons were detected in the cerebral cortex, hippocampus, thalamus, midbrain and cerebellum in all animals (rows I–III). Positive neurons in Animal A11 (row V) correspond to Fig. 1d; Animal A15 (rows IV and V) correspond to Fig. 1g, h, respectively; and Animal B3 (rows IV and V) correspond to Fig. 4a, g, respectively.

Full size image

In the 4 dpi group, viral antigens were also detected in oral cavity and tongue squamous epithelia (Fig. 1e) (75% of animals), paw and skin epidermis (75% of animals), and occasionally in hair follicle epithelium (Fig. 1f and Table 1). Viral antigens were detected bilaterally in brainstem (medulla/pons) neurons in the reticular formation and trigeminal motor nucleus (Fig. 1g), and in the multiple levels of spinal cord anterior horns (Fig. 1h) in all the infected hamsters (Table 1 and Fig. 2). No viral antigens were detected in the cerebrum, cerebellum, hippocampus, thalamus, hypothalamus, midbrain, and other CNS areas. Viral antigens were also detected in non-CNS tissues in more animals compared to 3 dpi (Table 1), including myocardium (Fig. 1i), brown adipose tissue (Fig. 1k) and salivary gland, and more densely in all skeletal muscles (Fig. 1l). Intestinal smooth muscle (Fig. 1j) (50% of animals) and lacrimal gland (33% of animals) were found to be viral antigen-positive only at 4 dpi. All other organs including lung, liver, kidney, pancreas, thymus, spleen, lymph node, blood vessel, and peripheral nerve (both afferent and efferent) were negative for viral antigens in all infected animal groups. In selected tissues, IHC findings were confirmed by ISH which generally showed comparable viral positivity (Supplementary Fig. 1).

Virus titration

Viral titers in the serum, skeletal muscle, brown adipose tissue, brain and heart harvested from 1 to 4 dpi groups are shown in Fig. 3; all values being expressed as mean viral titers.

Fig. 3: Mean viral titers in various tissues derived from CV-A16 infected hamsters from 1 to 4 dpi groups (Kinetic study).
figure 3

In the 1 dpi group, virus was only detected in serum; all other tissues were negative (*). Generally, viral titers from all tissues increased in 2, 3, and 4 dpi groups, but significantly (P < 0.05) so in the 2 and 3 dpi groups. Only brown adipose tissue and skeletal muscle titers significantly increased in the 4 dpi group. Viral titers were expressed as the mean CCID50/ml ± standard error of the mean.

Full size image

In the 1 dpi group, a viral titer of 5.38 × 102 CCID50/ml was only obtained in serum; other tissues were all negative. Serum viral titers significantly increased from 2.17 × 103 CCID50/ml (P = 0.03) to 2.34 × 105 CCID50/ml (P = 0.02) in the 2 dpi and 3 dpi groups, respectively, before stabilizing at 1.78 × 105 CCID50/ml (P > 0.05) in the 4 dpi group. Likewise, in the 2 and 3 dpi groups, brain and heart viral titers significantly increased from 9.98 × 102 CCID50/ml (P = 0.01) to 1.10 × 104 CCID50/ml (P = 0.04), and from 4.94 × 103 CCID50/ml (P = 0.01) to 2.22 × 104 CCID50/ml (P = 0.02), respectively. No significant differences (P > 0.05) in the 4 dpi group were observed in the brain (3.64 × 104 CCID50/ml) and heart (1.94 × 105 CCID50/ml). In the 2, 3, and 4 dpi groups, brown adipose tissue viral titers significantly increased from 2.05 × 103 CCID50/ml (P = 0.01), 5.63 × 104 CCID50/ml (P = 0.02) to 9.19 × 105 CCID50/ml (P = 0.02), and in skeletal muscle titers increased from 1.78 × 104 CCID50/ml (P = 0.01), 4.33 × 105 CCID50/ml (P = 0.02) to 1.56 × 106 CCID50/ml (P = 0.02), respectively. Skeletal muscles showed the highest titers from 2 to 4 dpi.

Transmission study

In the separate viral transmission study, all the index hamsters showed typical signs of infection from 3 dpi onwards (Table 2). Three index hamsters (Animals B2–B4) were found dead at 5 dpi and another 2 index hamsters (Animals B1 and B5) died at 6 dpi. All the contact hamsters developed signs of infection 3–5 days later than index hamsters, i.e., at 8–9 dpe. Two contact hamsters (Animals B6 and B10) were sacrificed at 9 dpe, while another 3 (Animals B7–B9) were sacrificed at 10 dpe (Table 2).

Table 2 Virus isolation and IHC findings in CV-A16 MAVS-infected hamsters (transmission study).
Full size table

Pathological findings

CNS viral antigen distribution in index hamsters at 5–6 dpi (Fig. 2, Animals B1–B5) was similar to the hamsters at 4 dpi in the kinetic study (Fig. 2, Animals A13–A16), involving brainstem (medulla/pons) reticular formation and motor trigeminal nucleus (Fig. 4a, c, d, f), and multiple levels of spinal anterior horns (Fig. 4g–i). However, the number of positive neurons may be relatively more in index hamsters. Similarly, viral antigens/RNA were absent in the cerebral and cerebellar cortices, hippocampus, thalamus, hypothalamus, midbrain, and other CNS areas. Interestingly, IHC and ISH positive signals were detected in the cerebellar external granular layer, possibly in neuroblasts, in two index hamsters (Animals B2 and B3) (Fig. 4a, b, d, e). Similar to the kinetic study, viral antigens were detected in oral cavity (Fig. 5a) and tongue (Fig. 5b) squamous mucosa, skin and paw epidermis (Fig. 5c), and lacrimal gland (Fig. 5d), where viral RNA localization was confirmed by ISH (Fig. 5e–h). Viral antigens/RNA were also found in salivary gland, brown adipose tissue and all muscle types. Other organs were negative for viral antigens/RNA.

Fig. 4: IHC and ISH detection in a CV-A16-infected index hamster at 5 dpi.
figure 4

Viral antigens (a–c) were detected in Animal B3 cerebellar granular layer and brainstem (medulla/pons), shown in higher magnification in b corresponds to the area indicated in (a) (full circle), and c corresponds to the area indicated in (a) (dashed circle), respectively. In the adjacent section, viral RNA (d–f) were demonstrated in the cerebellar granular layer and brainstem (medulla/pons), shown in higher magnification in (e) corresponds to the area indicated in (d) (full circle) and (f) corresponds to the area indicated in (d) (dashed circle), respectively. Viral antigens (g, h) and RNA (i, arrows) were detected in the anterior horn of the cervical cord. The higher magnification of viral antigens in neurons (h, arrows) corresponds to the area shown in (g) (dashed circle). IHC with DAB chromogen and hematoxylin counterstaining (a–c, g, h). ISH with NBT/BCIP substrate and Mayer’s hematoxylin counterstaining (d–f, i). Scale bars: 20 μm (b, c, e, f); 10 μm (h, i). Microscopic images were taken at 4× (a, d) and 10× objective (g) before digitally combined using the AutoStitch v2.2 software. a–c Correspond to row IV, and g, h correspond to row V of Animal B3 in Fig. 2, respectively.

Full size image
Fig. 5: IHC and ISH detection in index and contact hamsters.
figure 5

In index hamsters (5–6 dpi), focal viral antigens were detected in squamous epithelial cells of oral cavity (a, arrows), tongue (b, arrow) and paw epidermis (c), and viral RNA (e–g, arrows) were detected in these same locations, respectively. Viral antigens (d, arrow) and RNA (h, arrow) were also demonstrated in lacrimal glands. In contact hamsters (9–10 dpe), viral antigens were also found in squamous cells of oral cavity (i, arrows), tongue (j, arrows), paw epidermis (k, arrow), and hair follicle epithelium (l, arrows). In the CNS, viral antigens were detected in neurons of the brainstem (medulla/pons) (m, arrows) and spinal cord anterior horn (n, arrow). As in index hamsters, viral antigens were also observed in the myocardium (o, arrow) and salivary gland (p, arrows). IHC with DAB chromogen and hematoxylin counterstaining (a–d, i–p). ISH with NBT/BCIP substrate and Mayer’s hematoxylin counterstaining (e–h). Scale bars: 10 μm (a, c–e, g–k, m–o); 5 μm (b, f, l, p).

Full size image

The CNS findings in contact hamsters (Animals B6–B10) were generally similar to index hamsters, with positive neurons in the brainstem (medulla/pons) (Fig. 5m) and spinal anterior horns (Fig. 5n), except that the number of positive anterior horn neurons appear to be less in contact hamsters. Similar to the index hamsters, viral antigens and/or viral RNA were consistently exhibited in the oral cavity (Fig. 5i) and tongue (Fig. 5j) squamous mucosa, paw and skin epidermis (Fig. 5k) and hair follicle (Fig. 5l), myocardium (Fig. 5o), lacrimal and salivary glands (Fig. 5p), intestinal smooth muscle, brown adipose tissue, and skeletal muscle. Except for myositis, inflammation was absent or minimal in most tissues examined.

Virus isolation and viral RNA detection in oral washes and feces

The presence of viruses as shown by CPE in the viral culture was confirmed in some but not all feces and oral washes from index and contact hamsters. Among index hamsters, only 1 oral wash out of 5 (20%) showed CPE at 5 dpi or after 1 day of illness (Table 2, Animal B5). All fecal samples were negative.

Oral washes from 2 out of 5 (40%) contact hamsters showed CPE on 6 and 8 dpe, 2 days before and on the day of onset of signs of infection, respectively (Table 2, Animals B6 and B7). Feces from 3 out of 5 (60%) contact hamsters (Table 2, Animals B6, B9 and B10) showed CPE on 5 and 6 dpe, or 2 to 3 days before onset of signs of infection.

The results of virus isolation from feces and oral washes from index and contact hamsters were confirmed with CV-A16-specific primers using RT-PCR, where amplification was only detected in samples that had showed cell culture CPE. The PCR amplicons were observed in agarose gel electrophoresis as shown in Supplementary Fig. 2.

Discussion

Although there are increasing reports of neurological complications associated with HFMD caused by CV-A16 infection, very little is known about human CNS infection and pathology since no autopsy findings are available so far. In this study, we have developed a consistent, orally infected hamster model of CV-A16 HFMD with encephalomyelitis. Apart from our hamster model, in none of the other limited animal model studies, including our previous mouse model, was CNS neurotropism described in sufficient detail, nor was the preceding squamous lesions reminiscent of HFMD ever demonstrated [14, 15, 17, 19]. Moreover, the routes of infection in these models were parenteral, and so could not mimic the natural oral route of CV-A16 infection and transmission in humans. Thus, we believe our current hamster model will be able to provide valuable information to further extend our knowledge regarding CV-A16 infection and transmission in general, and neuropathogenesis in particular.

Positive viral antigens/RNA mainly in brainstem (medulla/pons) and spinal cord (anterior horn) neuronal cell bodies and processes in infected hamsters confirmed CV-A16 neuronotropism or predilection for neurons, and could explain the acute flaccid paralysis and encephalitis occasionally reported in human CV-A16 infection [1, 2, 4, 5, 9, 25, 26]. The distinct and unique distribution of the viral antigens/RNA in the hamster CNS appears to be consistent, at least in part, to the only available brain MRI findings in a nonfatal case of human CV-A16 rhombencephalitis, where hyperintense lesions were restricted to the pons and cerebellum around the fourth ventricle with no apparent involvement of spinal cord, medulla, midbrain or other parts of the CNS [1]. There were two other case reports of clinical CV-A16 encephalitis in immunocompromised hosts in which brain MRI findings were available. One patient on Obinutuzumab treatment showed bilateral, multiple hyperintense white matter lesions in the periventricular region and cerebral hemispheres [2], while another patient on Rituximab treatment had normal MRI findings [25]. However, these findings may not represent the true nature or distribution of CNS lesions in immunocompetent hosts. Further studies, including brain imaging and autopsies, are needed to investigate this.

Nonetheless, recently published brain MRI findings in other neurotropic enteroviral infections, appear to be consistent with the distribution of viral antigens/RNA in our hamster model. EV-A71, echovirus 7, CV-B1, CV-B3, and CV-B4 CNS infections showed MRI lesions mainly involving the brainstem or parts thereof [27,28,29,30,31]. Hyperintense MRI signals in the anterior horns at multiple levels of the spinal cord were also reported in EV-A71 and echovirus 7 encephalomyelitis [29, 30, 32,33,34]. MRI findings and human autopsy studies of EV-A71 encephalomyelitis showed a distinct and stereotyped distribution of viral antigens/RNA and inflammation in the midbrain, pons tegmentum, medulla, cerebellar dentate nucleus, hypothalamus and spinal cord anterior horns [35,36,37]. Thus, overall, CV-A16 encephalomyelitis in our hamster model is generally consistent with other neurotropic enteroviruses, demonstrating the ability of enteroviruses to directly infect the CNS to cause neuronal cell injury, and that the brainstem and spinal cord are predominantly involved early. This has not been established in other animal models [14, 15, 17, 19].

Human CV-A16 neuroinvasion still remains poorly understood but our hamster model suggests a possible mechanism. Viremia at 1 dpi (Fig. 3), initially spreads infection to skeletal muscles, which were the only tissues positive for viral antigens at 2 dpi (Table 1). Other tissues were negative for viral antigens despite positive viral titration (Fig. 3) possibly because of inadvertent contamination of these tissues by adjacent infected skeletal muscles or even contamination by viruses within blood vessels since vascular perfusion was not performed before tissue harvest. We speculate that with increasing skeletal muscle infection, viruses crossed motor neuron junctions to travel up bilateral somatic motor and fifth cranial nerves to infect neuronal bodies in spinal anterior horns and brainstem motor trigeminal nuclei, respectively. Although we were unable to demonstrate viral antigens/RNA in ventral roots in our hamster model, these have been demonstrated in cranial and peripheral nerves, efferent motor axons adjacent to spinal anterior horns, and the motor cortex in EV-A71 encephalomyelitis animal models, together with motor trigeminal nucleus and spinal anterior horn infection [23, 38]. Moreover, abnormal MRI signal intensity and enhancement in spinal anterior nerve roots had been reported in EV-A71, CV-A16 and echovirus CNS infections [30]. Further investigations are needed to confirm retrograde motor nerve CV-A16 transmission.

Viral antigens/RNA in squamous epithelia in the oral cavity, tongue, skin and paw epidermis could explain the oropharyngeal lesions, maculopapular and vesicular rashes present in human CV-A16 HFMD. In a case of CV-A6 HFMD, a skin biopsy showed viral antigens predominantly in epidermal keratinocyte cytoplasm associated with necrosis [39]. Viral antigens/RNA in squamous cells were also described in an EV-A71 hamster model and infected human organotypic skin cultures [20, 40]. Thus, CV-A16 squamous epitheliotropism in our hamster model is consistent with other enteroviral infections. Squamous cells as active replication sites may play an important role in virus shedding and person-to-person transmission. Likewise, acinar cell infection in salivary and lacrimal glands could shed more virus into saliva and tears, respectively. Positive CV-A16 cultures from human skin vesicles have been described [41, 42] but so far, there are no reports of live virus in saliva or tears in human or animal model studies.

It was reported that intrafamilial transmission may be one of the important factor in HFMD outbreaks, particularly within households with seropositive adult/siblings [43,44,45]. Thus, our hamster model, showing successful viral transmission among intra-family littermates, could further our understanding of person-to-person transmission. Consistent with findings in the kinetic study, index hamsters developed typical signs of infection from 3 to 5 dpi and death occurred from 5 to 6 dpi (Table 2). Although oral wash was positive for virus in only one index hamster at 5 dpi, we assume that oral and cutaneous virus shedding probably began from 3 dpi onwards, since extensive oral mucosa and epidermal infections were observed in the kinetic study from 3 dpi onwards (Fig. 1 and Table 1). Unsuccessful virus detection from fecal samples, and only one successful virus isolation from oral washes in index hamsters may be the result of the relatively short duration of illness of 2 days in index hamsters, and varying amounts of fecal samples available.

The longer incubation period of 5 days in contact animals, rather than the 3 days in index animals, also suggests that the viral doses contact hamsters were exposed to either directly from index hamsters and/or the contaminated environment, were relatively low, at least at the beginning. However, the interval between initial virus exposure and death in contact hamsters of 6–7 days was only slightly longer than the 5–6 days in index hamsters (Table 2). Interestingly, viruses could be isolated from feces and oral washes of contact hamsters even before the onset of signs of infection, suggesting that virus shedding could occur soon after the virus exposure.

Infection of all contact hamsters in the transmission study arguably, partially fulfils the original Koch postulates in that following viral exposure to index hamsters and/or environment, contact hamsters developed signs of infection, and the same virus could be isolated from oral washes and fecal samples in 4 out of 5 animals (Table 2). However, according to revised versions of Koch postulates, confirmation of infection by tissue localization of viral antigens/RNA is sufficient evidence for its fulfillment in all the index and contact hamsters [46]. Recovery of live virus from oral washes and fecal samples relatively late from 5 to 9 dpi/dpe suggests that they probably represent viral progeny rather than the original inoculums. Since viral antigens/RNA were absent in gastrointestinal epithelia, pancreas and liver, the source of fecal CV-A16 was likely to be virus progeny mainly originating from the oral cavity or tongue mucosa. Based on studies in human and hamster infections, it was hypothesized that fecal EV-A71 probably originates from the upper orodigestive tracts, since viral replication in the lower gastrointestinal tract has never been demonstrated [11]. To our knowledge, a reliable orally infected model for CV-A16 person-to-person transmission and/or viral shedding has so far not been available, but had been described in EV-A71 animal models [47,48,49,50].

There appear to be relatively more viral-positive brainstem and spinal cord neurons in the index hamsters (transmission study) than the infected hamsters in the kinetic study at 4 dpi (Fig. 2), probably because the endpoint for index hamsters was 5–6 dpi. However, the number of positive neurons in the brainstem of index and contact hamsters appears to be the same, but more positive neurons were found in the spinal cord of index hamsters, possibly because contact hamsters may be exposed to lower total viral doses.

CV-A16 myocarditis had been reported in rare patients who presented with cardiomegaly, and autopsy findings had shown severe inflammatory cell infiltration in one case and positive viral isolation in another [7,8,9]. Demonstration of myocardial infection in our hamster and other mouse models [12, 19] showed that myocarditis was caused by direct cytopathic effect rather than by pathologic immune response or autoimmunity more commonly associated with CV-B infections [51, 52]. Extensive skeletal muscle, and more limited gastrointestinal smooth muscle myotropism were observed in our hamster and other animal models of enteroviral infections [12, 18,19,20]. However, human skeletal and smooth muscle infection has so far not been demonstrated, possibly because there have been very few studies. From our findings, it appears that CV-A16 could potentially infect all types of muscles, including cardiac, skeletal and smooth muscles. Viral antigens/RNA were found in brown adipose tissue and confirmed by viral titration, similar to a few other enteroviruses animal models [19, 38, 53, 54], but so far no human data are available. Pulmonary edema, pneumonitis, or IHC-positive lung tissues were reported in several human cases and CV-A16 animal models [6, 10, 14, 16, 18, 55], but were not observed in our hamster model.

In conclusion, this unique hamster model could further our understanding of neurotropism, neurovirulence, as well as neuropathology of CV-A16 infection. Our novel CV-A16 hamster model is also useful to model HFMD and person-to-person transmission of CV-A16 in human populations. Moreover, this model could be used for the development of vaccine or antiviral agents to reduce viral replication and shedding, and to prevent and treat CNS infections.