Cynomolgus monkeys (Macaca fascicularis) experimentally and naturally infected with hepatitis E virus: The bone marrow as a possible new viral target

Hepatitis E virus (HEV) transmission through infected blood and blood products has already been described. However, little is known about the bone marrow (BM) as source of HEV infection. Our study aimed to investigate the presence of HEV antigen (Ag) and histological changes in BM of cynomolgus monkeys (Macaca fascicularis) experimentally and naturally infected with HEV. Four cynomolgus monkeys with acute, and two with chronic hepatitis E ─ after immunosuppressive therapy with tacrolimus ─ were compared with one colony-bred animal naturally infected. Both, natural and experimental infections were characterized by anti-HEV IgG seroconversion detected by ELISA, and viral RNA isolation confirmed by RT-qPCR and qualitative nested RT-PCR. BM biopsies were collected from all animals, submitted to histology and indirect immunofluorescence techniques and observed, respectively, by light and confocal microscopy. The HEV Ag-fluorescent-labeled cells were detected from BM biopsies obtained from three monkeys with acute and one with chronic hepatitis E, and also from the naturally infected monkey. In the experimentally infected animals with acute hepatitis, HEV Ag detection occurred at 160 days post-infection, even after viral clearance in serum, feces, and liver. Double-stranded RNA, a replicative marker, was detected in BM cells from both acute and chronically infected animals. Major histological findings included vacuolization in mononuclear and endosteal cells, an absence of organized inflammatory infiltrates, and also some fields suggesting displasic focal BM disease. These findings support the hypothesis of BM cells as secondary target sites of HEV persistence. Further experimental studies should be carried out to confirm the assumption of HEV transmission through BM transplantation.


Introduction
Hepatitis E virus (HEV) infection is a major cause of acute viral hepatitis worldwide [1]. Every year, there are an estimated 20 million new human cases, more than three million symptomatic cases and 44,000 disease-related deaths [1]. HEV is classified in family Hepeviridae, genera Orthohepevirus A, lately grouped in five genotypes infecting humans (1, 2, 3, 4 and 7) and three genotypes infecting wild boars (5 and 6) and Bactrian camels (8) [2][3][4][5]. HEV genotypes 1 and 2 are endemic in developing countries where transmission occurs mainly via the fecal-oral route and contaminated water is the main source of infection [6]. Genotypes 3 (HEV-3) and 4 (HEV-4) are prevalent in developed countries where autochthonous cases of acute HEV infection are attributed mainly to zoonotic transmission to humans, associated with consumption of raw or undercooked meat from pigs, wild boars and other mammals [7,8].
Further, human cases of persistent HEV-3 infection evolving to chronic hepatitis were described in solid organ transplant (SOT) patients under immunosuppressive therapy with, e.g. tacrolimus, considered a potent macrolide immunosuppressant derived from Streptomyces tsukubaensis (calcineurin pathway inhibitor) and a first-line medication employed to reduce the rate of rejection, especially in parenchymal organ transplantation [9]. High doses of tacrolimus showed to promote infection of liver cells with HEV in cell culture models [10]. It is considered a risk factor for virus persistence in the host [11,12]. Also, occurrence of HEV-3 related chronic hepatitis was reported in recipients of allogeneic bone marrow transplantation [13][14][15]. Besides, Hepatitis-Associated Aplastic Anemia (HAAA) and Diamond-Blackfan Anemia (DBA), as well as some unspecific hematological changes, such as lymphopenia and leukopenia, were reported in patients with hepatitis E [16][17][18][19].
Cynomolgus (Macaca fascicularis) is the best model to mirror human organ transplantation [25,26]. Besides, it is considered the primary model for studying acute and chronic clinical course of HEV infection, as confirmed by our group [12].
In this context, the aims of our study were (i) to investigate the presence of HEV Ag in bone marrow cells from two groups of cynomolgus monkeys infected experimentally, with acute and chronic hepatitis E; (ii) to determine the anti-HEV seroconversion profile using different point-in-time samples from a colony-bred cynomolgus monkey infected naturally with HEV and (iii) to compare detection of HEV Ag and histological findings of these three groups of HEV-infected animals.

Ethics statement
All monkeys obtained for this study originated from a breeding colony from the Institute of Science and Technology in Biomodels (ICTB), of the Oswaldo Cruz Foundation (Fiocruz), Rio de Janeiro, Brazil. Animals submitted to the experimental infection were kept in Biohazard Level 2 facilities in a single house in stainless steel squeeze-back cages (0.77 m height x 0.60 m width x 0.68 m depth) in a climate-controlled room (temperature of 22 ± 1˚C and humidity 55 ± 5%) with a 12h light / dark cycle. Those animals were euthanized by exsanguination (cardiac puncture) under deep barbiturate anesthesia with sodium thiopental 2.5% at 25 mg/kg (Thiopentax, Cristalia, São Paulo, Brazil), which was delivered intravenously. The colony-bred animal naturally HEV infected was housed outdoor in a big cage (6 m x 6.4 m x 4.2 m), together with other ten cynomolgus monkeys.
All animals were fed daily with a commercial primate diet supplemented with vegetables and fresh fruits. Water was provided ad libitum. Environmental enrichment programs, such as: alimentary (popcorn and nuts), audio-visual (audios with forest themes and movies), and tactile enrichment (toys such as hanging balls) were offered throughout the study.
Clinical procedures and samples collection were performed under anesthesia with ketamine hydrochloride at 20 mg/kg (Vetanarcol, Konig, Argentina) combined with midazolam at 0.1 mg/kg (Cristalia, Rio de Janeiro, Brazil). To reduce post-operatory pain, 0.2 ml of 2% lidocaine was injected subcutaneously at bone marrow (BM) puncture site.
The housing standard adopted in our study attended to space recommendations for individually NHP with a maximum weight of 7 kg, in accordance with the Brazilian Normative Resolution CONCEA n. 28

Animals and study design
In order to evaluate the hypothesis of BM cells as reservoirs of HEV, biopsies were performed in iliac crest from six cynomolgus monkeys HEV infected experimentally: four with acute and two with chronic hepatitis E, as previously described by our group (Table 1) [12]. HEV acute hepatitis was defined by absence of inflammatory infiltrates before 160 days post-infection (dpi), and sustained virological response before 69 dpi. At the end of the experiment (160 dpi), chronic hepatitis was characterized by histology as a limited area of interface hepatitis (piecemeal necrosis), associated with the detection of HEV RNA in serum, feces and liver. All animals that developed chronic hepatitis E were previously treated with tacrolimus. Histological analysis and immunostaining for detection of HEV Ag were performed in BM biopsies.
Aiming to compare natural and experimental infections, histological and HEV Ag profiles were also evaluated using BM samples from an adult cynomolgus monkey infected naturally. This animal was noted as strongly reactive for anti-HEV IgG under routine screening of three cynomolgus monkeys from a NHP research facility of the ICTB / Fiocruz (Table 2). Aiming to confirm natural HEV infection, three earlier and two subsequent serum samples were tested for detection of anti-HEV IgG and HEV RNA (Fig 1).

Bone marrow biopsy and blood collection
BM biopsies were obtained aseptically from the iliac crest using a 14G needle (Ítaca Laboratórios Ltda, Brazil). The interpretable biopsy lengths were 3 to 5 mm with a diameter of 2.1 mm. When the fragments were too small, two to three perforations were performed. Samples were obtained at 80 days pre-infection (T0) and 160 dpi (T1) from both groups of animals, with acute and chronic hepatitis E. With the naturally infected cynomolgus monkey (AC10) similar procedures were adopted for BM biopsy. Blood samples were collected from femoral vein. Serum samples were stored at -20˚C for serological tests and -70˚C for virological analysis.

Serological assays
Serological detection of anti-HEV IgG antibodies was performed using the commercially available kit recomWell HEV IgG (Mikrogen Diagnostik, Germany) according to the manufacturer's instructions.

RNA extraction, nested RT-PCR and RT-qPCR
HEV RNA was extracted from serum samples and 10% w/v fecal suspensions using QIAamp viral RNA mini kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. Reverse transcription (RT) and PCR reactions were performed in a single step using the SuperScript III One-Step RT-PCR System with Platinum Taq DNA Polymerase (Invitrogen Life Technology, USA). RT-PCR and nested RT-PCR were performed using a set of primers targeting ORF2 region, as previously described [28]. RT-qPCR was performed using AgPath-ID one-step RT-PCR kit (Applied Biosystems, USA) using primers and probe previously described [29].

Sequencing reactions and phylogenetic analysis
The amplification product of ORF2 was purified using reagents and protocols of the commercial Wizard SV Gel kit and PCR Cleaning System (Promega, USA). Sequencing reactions were performed using reagents and protocols of Big Dye Terminator 3.1 kit (Applied Biosystems, EUA). Phylogenetic analyses were conducted with Bayesian inference using Markov Chain Monte Carlo (MCMC) statistical framework implemented in the program BEAST v1.8.1 [30] under TRN+G nucleotide substitution model. A phylogenetic tree, based on the HEV ORF2 region (302bp), was constructed with sequences retrieved from GenBank, including prototype sequences from HEV genotypes 3 and 4.

Results
HEV Ag-labelled bone marrow (BM) cells were detected in three out of four cynomolgus with acute hepatitis E (AE3, AC11, and AD8) and in one out of two with chronic hepatitis E (AE6) at 160 dpi (Fig 2A-2D;2F) ( Table 3). These cells presented a dotted-shape green labelling (HEV Ag-positive), sometimes spread through the cytoplasm (Fig 2A;2D), sometimes concentrated in a few inclusions (Fig 2B;2C). BM samples collected before HEV infection (T0) did not show specific labeled-cells (Table 3). Double-stranded RNA (dsRNA) was detected at 160 dpi, from both acute and chronically infected animals, by immunostaining (Fig 3A;3B) ( Table 3). The pattern of dsRNA labeledcells was similar to those observed in the liver of the chronically infected monkey witch was found to be positive for negative strand HEV RNA by RT-PCR (Fig 3C). The frequency of labeled cells was highlighted in chronically HE-infected monkeys. Negative controls (omitting the J2 primary antibody) did not show specific labeled-cells (Fig 3D-3F) Histological analysis of BM, at 160 dpi, from all HEV Ag-positive monkeys (acute and chronic hepatitis E) revealed vacuolated mononuclear cells (Fig 4A;4B). Also, mononuclear cells spread within BM parenchyma did not show organized inflammatory infiltrates. Clusters of lymphocyte proliferation and activation (Fig 4C), as well as megakaryocytosis (>5 megakaryocytes / field) ( Fig 4D) were observed in a single animal with acute hepatitis E (AE3). The other animals showed megakaryocytes counts similar to those observed in pre-inoculation step (0-2 megakaryocytes / field) ( Fig 4E). Vacuolization in endosteal cells was observed in both groups, acute and chronic (Fig 4F) ( Table 4).
The naturally infected monkey (AC10) showed typical serological and virological changes, as previously described by our group [31]. Results are summarized in Fig 1. Anti-HEV IgG seroconversion was confirmed by ELISA, and HEV RNA was detected by RT-PCR and RT-qPCR in serum samples (Fig 5). Serum levels of ALT and AST were within the normal range for the species, 27 and 59 IU/L, respectively. Phylogenetic reconstruction using a partial nucleotide (nt) sequence (ORF2, 302 nt) revealed that AC10 HEV isolate belongs to HEV genotype 3 (Fig 6). The nucleotide sequence shared 99% identity with chronically infected cynomolgus isolates from our experimental infection study (accession numbers: KX578268, KX578269, and KX578270) [12]. The partial genomic sequence reported in this study was deposited in the GenBank under the accession number MG573667.
HEV Ag was also detected by immunofluorescence in AC10 BM biopsy ( Fig 2E). However, this animal did not show any apparent BM histological change (Table 4). Comparatively, frequency of HEV Ag-labelled cells was higher in the animal infected chronically (! 8 cells / image field) than in the animals infected acutely and naturally (1-2 cells / image field) ( Fig  7A-7D).

Fig 2. Bone marrow cells of cynomolgus monkeys with hepatitis E virus infection at 160 dpi. HEV antigen detection in: (A) one monkey with chronic hepatitis E; (B-D) three monkeys with acute hepatitis E; (E) one monkey naturally HEV infected. (F) Negative results in a monkey chronically infected. HEV antigen detection (!) in green, nuclei stained with DAPI in blue and erythroid cells and background in red.
https://doi.org/10.1371/journal.pone.0205039.g002 Table 3. HEV antigen detection in bone marrow biopsies from animals infected experimentally.

Discussion
Our study reports for the first time detection of HEV Ag in bone marrow (BM) cells from non-human primates (NHP). Experimentally infected animals, with chronic and acute hepatitis E, presented incipient histological signs, suggesting BM hyperactivation and dysfunction, characterized by vacuolization in endosteal cells, with some fields suggesting displasic focal disease. Medical research studies described severe aplastic anemia in association with parvovirus B19 (B19V), hepatitis A virus (HAV) and hepatitis C virus (HCV) infections [32][33][34]. Besides, recent studies described association between endosteal niche and loss of hematopoiesis homeostasis [35,36]. According to our results, BM can be considered a site of viral persistence since detection of HEV Ag and dsRNA, a replicative marker, occurred at 160 dpi in acutely HEV infected monkeys, even though viral clearance in serum, feces, and liver had occurred within 56, 42, and 69 dpi, respectively.
HEV Ag was also detected in BM cells from the naturally infected animal, however, in absence of histological changes. Viral replication was confirmed, in both acute and chronic animals, by detection of dsRNA by immunostaining [37]. Unfortunately, HEV RNA detection by RT-PCR could not be performed in BM biopsies due to the limited amount of material available.
The monkey infected naturally, as well as the three NHP with acute hepatitis E, showed rare HEV Ag-labelled cells. On the other hand, in the chronically infected animal, a high number of those target cells was observed. A possible deactivation of immune system cells caused by immunosuppressive treatment with tacrolimus could explain such increased number of HEV Ag-positive cells persisting in BM [38,39]. Adverse effects induced by tacrolimus long-term therapy associated to immune deactivation, such as: viral persistence and chronic hepatitis has been described [12]. The hemolytic-uremic syndrome may occur after administration of tacrolimus [40]. Moreover, microangiopathic hemolytic anemia and thrombocytopenia ─ rare, but potentially severe complications due to the use of immunosuppressive therapy ─ were not observed in our tacrolimus treated monkeys [41].
In order to characterize HEV target cells, it would be necessary to carry out phenotypic characterization. In our study, viral inclusions were not observed in hematoxylin and eosin staining. The use of immunostaining allowed the observation that detection of HEV Ag labeled Table 4. Summary of bone marrow histological changes and HEV antigen detection in naturally and experimentally HEV infected cynomolgus monkeys. cells were compatible with immature myeloid cells or even stem cells, both topographically and morphologically (large cells with a large nucleus and protruding nucleolus, with variable amounts of cytoplasm). Our findings corroborate with detection of HEV infection in post-alloHSCT in anti-HEV seronegative patients [42]. Detection of HEV Ag in BM cells also corroborates with evidence of transfusional HEV transmission and alerts to the possibility of transmission through BM transplantation. However, pathways of medullary infection are not yet confirmed [43,44]. It is important to emphasize that in immunocompromised patients transfusional transmission of HEV, similar to other hepatotropic viruses, can worsen patient's clinical conditions [45,46].

Chronically experimentally HEV infected monkeys
Despite low incidence of HEV infection as cause of acute hepatitis in post-alloHSCT, a high risk of developing chronic hepatitis with subsequent progression to liver failure, cirrhosis or even death has been associated with HEV infection in this risk group [23,43]. Our results reinforce the need for systematic HEV screening of HSC donors with either a risk profile for HEV or abnormal liver tests, as suggested by other authors [20,22,23]. Besides, pre and post transplantation screening of patients should be considered once HEV target cells may present a late expression potential.
Besides, in July 2017, a naive colony-bred cynomolgus monkey (AC10) was found to be positive for HEV RNA. Serum samples obtained within a four years-monitoring period (November 2013; April 2015; April 2016; and July, August, and September 2017) were tested, with anti-HEV IgG seroconversion detected between April 2015 and April 2016. Although anti-HEV IgM was not tested, increasing levels of anti-HEV IgG in serial samples from April 2015 to July 2017 suggest recent infection (Fig 4). HEV RNA was detected in serum samples obtained in July and August 2017, concomitantly with decrease of anti-HEV IgG levels.
In general, the viremia period in cynomolgus monkeys is around 20 to 30 days [12,47]. In our study, anti-HEV IgG seroconversion occurred between 2015 and 2016. Hence, detection of HEV RNA in serum samples from 2017 may suggest persistence of HEV infection, similar to that observed in a naturally infected Japanese monkey (Macaca fuscata) in a non-human primate colony in Japan, showing no evidences of immunosuppression [48]. Failure to detect HEV RNA in the sample retrospectively investigated in our study may be correlated with low viral load or long storage period, and successive freeze / thaw cycles. Another plausible assumption would be related to reactivation of HEV infection that has, so far, only described for alloHSCT recipients or antiviral-treated hepatitis C patients [11,49]. Nevertheless, a viral reactivation was not observed in cynomolgus monkeys with acute hepatitis E, during the monitoring period of our NHP experiment, even after immunosuppression challenge with tacrolimus [12].
A high nucleotide identity was observed between HEV isolate obtained from the monkey (AC10) infected naturally and isolates from monkeys infected experimentally, as well as with inoculum obtained from a pig from a commercial farm located in Paraná state [12]. However, it is not possible to state the source of HEV transmission to the colony-bred animal as a full genome analysis would be required.
According to serological results, AC10 might have been infected at least three months after the end of the experimental study (February 2015). Thus, it is possible that contamination occurred due to protocol failures in waste management in the experimental area since the principal mode of HEV transmission is via the fecal-oral route. Moreover, as reported in a NHP colony in Japan, handlers can act as source of HEV transmission [48]. Unfortunately, serum samples of employees were not available in our study.
To the best of our knowledge, this is the first report of natural HEV infection among monkeys breed in South America. Together with the study conducted in the Japanese colony [48], our study highlights the need of continuing improvements and monitoring of preventive measures in non-human primate's captive facilities.
Importantly, our results reinforce the hypothesis that HEV can be retained in BM cells even in resolving HEV infections with sustained viral response achieved spontaneously. Long-lasting NHP studies are necessary to follow up BM changes after the recovery phase of hepatitis E and its late hematological effects.