Pig model mimicking chronic hepatitis E virus infectionin immunocompromised patients to assess immunecorrelates during chronicityDianjun Caoa, Qian M. Caoa, Sakthivel Subramaniama, Danielle M. Yugoa, C. Lynn Heffrona, Adam J. Rogersa,Scott P. Kenneya, Debin Tiana, Shannon R. Matzingera, Christopher Overenda, Nicholas Catanzaroa, Tanya LeRoitha,Heng Wanga, Pablo Piñeyroa, Nicole Lindstromb, Sherrie Clark-Deenerb, Lijuan Yuana, and Xiang-Jin Menga,1

aDepartment of Biomedical Sciences and Pathobiology, Virginia-Maryland College of Veterinary Medicine, Virginia Polytechnic Institute and StateUniversity, Blacksburg, VA 24061; and bDepartment of Large Animal Clinical Sciences, Virginia-Maryland College of Veterinary Medicine, VirginiaPolytechnic Institute and State University, Blacksburg, VA 24061

This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2016.

Contributed by Xiang-Jin Meng, May 23, 2017 (sent for review April 3, 2017; reviewed by Alexander Ploss and Christopher M. Walker)

Chronic hepatitis E virus (HEV) infection is a significant clinicalproblem in immunocompromised individuals such as organ trans-plant recipients, although the mechanism remains unknown becauseof the lack of an animal model. We successfully developed a pigmodel of chronic HEV infection and examined immune correlatesleading to chronicity. The conditions of immunocompromised pa-tients were mimicked by treating pigs with an immunosuppressiveregimen including cyclosporine, azathioprine, and prednisolone. Im-munocompromised pigs infected with HEV progressed to chronicity,because 8/10 drug-treated HEV-infected pigs continued fecal virusshedding beyond the acute phase of infection, whereas the majority(7/10) of mock-treated HEV-infected pigs cleared fecal viral sheddingat 8 wk postinfection. During chronic infection, serum levels of theliver enzyme γ-glutamyl transferase and fecal virus shedding weresignificantly higher in immunocompromised HEV-infected pigs. Toidentify potential immune correlates of chronic infection, we deter-mined serum levels of cytokines and cell-mediated immune responsesin pigs. Results showed that HEV infection of immunocompromisedpigs reduced the serum levels of Th1 cytokines IL-2 and IL-12, andTh2 cytokines IL-4 and IL-10, particularly during the acute phase ofinfection. Furthermore IFN-γ–specific CD4+ T-cell responses were re-duced in immunocompromised pigs during the acute phase of infec-tion, but TNF-α–specific CD8+ T-cell responses increased during thechronic phase of infection. Thus, active suppression of cell-mediatedimmune responses under immunocompromised conditions may facil-itate the establishment of chronic HEV infection. This pig model willaid in delineating the mechanisms of chronic HEV infection andin developing effective therapeutics against chronic hepatitis E.

hepatitis E virus | chronic HEV infection | pig | immunosuppression |cell-mediated immune responses

Hepatitis E virus (HEV) infection causes an important globalpublic health disease burden with an estimated 20 million

individuals affected worldwide every year (1) resulting in 56,600hepatitis E-related deaths (2). HEV is a single-stranded, positive-sense, RNA virus (3) belonging to the familyHepeviridae (4). HEVhas been a major cause of acute viral hepatitis in many developingcountries, although sporadic and cluster cases of acute hepatitis Ehave been reported in many industrialized countries, including theUnited States (5–7). Recent reports suggest that the clinical casesand disease burden associated with HEV infection in industrial-ized countries have been underestimated (7).In general, HEV infection in immunocompetent individuals

develops a self-limiting acute viral hepatitis. However, the majorityof HEV infections in immunocompromised individuals, such assolid-organ transplant recipients and patients with HIV infection,lymphoma, or leukemia, are likely to progress to chronicity (8).Since the first report of chronic HEV infection in liver transplant

patients in 2008 (9), chronic hepatitis E has become recognized asan emerging and important clinical problem in immunocompro-mised individuals, especially in solid-organ transplant recipients(8, 10). Chronic hepatitis E can cause significant liver damage,which may eventually lead to cirrhosis with considerable mortality.Patients with chronic hepatitis E also shed HEV in feces for aprolonged period and can transmit the virus to immunocompetentindividuals (9). Broad-spectrum antivirals such as ribavirin andpegylated IFN have been used for the treatment of chronic hep-atitis E with some success (11, 12), although currently there is noestablished HEV-specific therapeutic protocol. Also, importantly,the fundamental mechanisms leading to the progression and es-tablishment of chronic hepatitis E in immunocompromised pa-tients are unknown because of the lack of an animal model forchronic hepatitis E. Therefore, an animal model that can mimicchronic HEV infection in immunocompromised individuals isurgently needed to study the underlying mechanisms of chronicinfection and to develop effective and specific therapeutics againstchronic hepatitis E in immunocompromised individuals.The Hepeviridae family has two genera (Orthohepevirus, and

Piscihepevirus) and five species. The species Orthohepevirus Aincludes HEV infecting humans and several other mammalianspecies and consists of at least seven distinct HEV genotypes (4):genotypes 1 and 2 infect humans exclusively; genotypes 3 and

Significance

An estimated 20 million hepatitis E virus (HEV) infections occuryearly worldwide, leading to 56,600 deaths. Chronic HEV in-fection has recently become a significant clinical problem in im-munocompromised individuals such as organ transplant patients.The lack of an animal model greatly hinders our ability to studychronic HEV infection and develop therapeutics. Here we reportthe successful development a pig model of chronic HEV infectionby mimicking the conditions of immunocompromised organtransplant patients. We demonstrate that active suppression ofHEV-specific cell-mediated immune responses under immuno-compromised conditions may facilitate the establishment ofchronic HEV infection. This unique model now affords the op-portunity to delineate the mechanism leading to chronicity andto test specific antivirals against chronic hepatitis E.

Author contributions: D.C., D.M.Y., C.L.H., D.T., S.C.-D., and X.-J.M. designed research;D.C., Q.M.C., S.S., D.M.Y., C.L.H., A.J.R., S.P.K., S.R.M., C.O., N.C., T.L., H.W., P.P., N.L.,S.C.-D., and L.Y. performed research; D.C., Q.M.C., S.S., T.L., L.Y., and X.-J.M. analyzeddata; and D.C., S.S., T.L., L.Y., and X.-J.M. wrote the paper.

The authors declare no conflict of interest.1To whom correspondence should be addressed. Email: xjmeng@vt.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1705446114/-/DCSupplemental.

6914–6923 | PNAS | July 3, 2017 | vol. 114 | no. 27 www.pnas.org/cgi/doi/10.1073/pnas.1705446114

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020

4 infect humans and several other animals such as pigs and rabbits(13); genotypes 5 and 6 infect wild boars; and genotype 7 infectscamels. The pig is a recognized major animal reservoir for zoonoticHEV transmission to humans (14). Strains of HEV genotypes3 and 4 are known to infect across species barriers (13, 15, 16). Infact, sporadic and cluster cases of acute hepatitis E in humans inindustrialized countries have been caused predominantly byzoonotic strains of HEV genotypes 3 and 4 (17). Similarly, theHEV strains isolated from chronically infected patients are almostexclusively the zoonotic genotype 3 (18, 19). Because pigs arenatural hosts for the HEV genotypes 3 and 4, a pig model hasbeen developed to study HEV biology, cross-species infection, andpathogenesis (17). However, the currently available animal modelsin pigs, chickens, rabbits, and nonhuman primates do not inducechronic HEV infection (20) and thus are suitable only for studiesof acute hepatitis E.In this study, we report the successful establishment of a unique

pig model for chronic HEV infection by treating pigs before andduring the course of infection with a genotype 3 human HEV withan immunosuppressive regimen similar to that used for humanorgan transplant recipients. In an attempt to identify the mecha-nism and immune correlates leading to chronic HEV infection,the magnitude and duration of viremia and fecal virus shedding,the types of immune responses developed against the virus, and theliver pathology associated with chronic HEV infection were alsodetermined and analyzed in chronically infected pigs.

ResultsSuccessful Establishment of a Pig Model for Chronic HEV Infection. Tomimic the immunosuppressive conditions in human solid-organtransplant recipients, pigs in the immunocompromised groupwere orally administered a drug mixture compounded with threeimmunosuppressive drugs (see Materials and Methods for details)routinely used to prevent rejection in human organ transplantrecipients. The immunosuppressive drug regimen was given 1 wkbefore infection with a genotype 3 human HEV as well as dur-ing the course of HEV infection to induce an effective non-specific immunosuppression in pigs similar to that seen in organtransplant recipients.

Fecal virus shedding was initially monitored by HEV-specificregular RT-PCR. Viral RNAs were detected in the fecal samplesof both drug-treated (immunocompromised) and mock-treated(immunocompetent) HEV-infected pigs up to 8 wk postinfection(wpi). At 12 wpi, 7 of the 10 pigs in the drug-treated HEV-infected group still shed HEV in feces, whereas only 1 of the10 pigs in the mock-treated HEV-infected group still had de-tectable viral RNA in feces by the HEV-specific RT-PCR. Ingeneral, during acute HEV infection in pigs, fecal virus sheddingtypically is cleared by 7–8 wpi (16); thus, for the purpose of datainterpretation, we arbitrarily set 8 wpi as the time point sepa-rating acute and chronic infections in this study. The resultsshowed that we successfully established chronic HEV infectionin immunocompromised pigs, because virus shedding in thesepigs continued beyond 8 wpi, for at least five additional 5 weeks.All animals were killed at 13 wpi after we determined thatchronic HEV infection had been established, with the exceptionof two immunocompromised HEV-infected pigs that were keptalive for nine more weeks. These two pigs continued to receivedaily immunosuppressive drugs, and viral RNAs in their feceswere detected by regular RT-PCR during the additional 9 wk. Atthe end of the 9-wk period these two animals were killed also.HEV RNAs were quantified in weekly fecal samples and in bi-

weekly serum samples by qRT-PCR. The viral RNA loads in fecespeaked at 3 wpi in immunocompetent HEV-infected pigs butcontinued to rise and reached peak level at 5 wpi in immunocom-promised HEV-infected pigs (Fig. 1A). Between 5 and 13 wpi, theviral RNA loads in feces were significantly higher in immunocom-promised HEV-infected pigs than in mock-treated HEV-infectedpigs (P < 0.05), except at 10 and 11 wpi (Fig. 1A). At 8 wpi, 8 ofthe10 mock-treated HEV-infected pigs but only 3 of the 10 of thedrug-treated HEV-infected pigs stopped fecal virus shedding (Fig.1A). At 13 wpi, 9 of the 10 drug-treated HEV-infected pigs but only2 of the 10 mock-treated HEV-infected pigs still shed HEV in feces(Fig. 1A). Therefore, the result of fecal virus shedding tested withqRT-PCR is similar to that tested with HEV-specific regularRT-PCR. During the chronic phase of infection, from 8 to 13 wpi,the average levels of viral RNA loads were consistently higher infeces of immunocompromised HEV-infected pigs than in the fecesof immunocompetent HEV-infected pigs (Fig. 1A).

Fig. 1. Quantification of HEV RNA loads by qRT-PCR in fecal and serum samples from HEV-infected pigs. (A) HEV RNA load in fecal samples collected weeklypostinfection. (B) HEV RNA loads in serum samples collected biweekly. The amounts of HEV RNAs were measured by qRT-PCR at the indicated time points,converted to log10 of copies per milliliter of 10% fecal suspension or serum, and expressed as mean ± SEM in a scatter plot. Each symbol in the scatter plotrepresents the value of an individual pig; symbols on the x axis are below the assay detection limit (400 copies per 1-mL sample) and were defined as negative.Circled numbers below the x axis indicate the number of negative animals at each time point. HEV, HEV infected (n = 10); IC HEV, immunocompromised andHEV infected (n = 10). *P < 0.05.

Cao et al. PNAS | July 3, 2017 | vol. 114 | no. 27 | 6915

MICRO

BIOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020

Interestingly, at 1 wpi, the viral RNA levels were significantlyhigher (about 0.5 log difference) in the sera of immunocompe-tent HEV-infected pigs than in the sera of immunocompromisedHEV-infected pigs (P < 0.05) (Fig. 1B). From 1 wpi to 13 wpi theviral RNA levels in sera were comparable, at a low level, inimmunocompetent and immunocompromised HEV-infectedpigs, with the exception of 9 wpi, when the serum viral RNA levelwas significantly higher in immunocompromised HEV-infectedpigs than in immunocompetent HEV-infected pigs (P < 0.05)(Fig. 1B), which is due to further clearance of HEV in immu-nocompetent HEV-infected pigs.During the chronic phase of HEV infection the serum levels of

liver enzyme γ-glutamyl transferase (GGT) were significantlyhigher at 9 and 11 wpi in immunocompromised HEV-infectedpigs than in immunocompetent HEV-infected pigs (P < 0.05)(Fig. 2A). However, there was no statistically significant differ-ence in the serum levels of sorbitol dehydrogenase (SDH) inimmunocompromised HEV-infected and immunocompetentHEV-infected pigs (Fig. 2B). Similarly, we did not observe a sig-nificant increase in serum levels of other liver enzymes, includingaspartate aminotransferase (AST), total bilirubin, and alkalinephosphatase, in immunocompromised HEV-infected pigs com-pared with immunocompetent HEV-infected pigs. At the time ofnecropsy (13 wpi), no gross lesions were observed in livers from allpigs, although mild microscopic lesions were found in the liversfrom 5 of the 10 immunocompromised HEV-infected pigs and4 four the 10 immunocompetent HEV-infected pigs. There wereno significant differences in microscopic liver lesion scores be-tween the two groups of HEV-infected pigs, even though themajority of immunocompromised pigs continued to shed virus infeces during the chronic phase of infection.

Reduction of Baseline Serum Levels of Th1 Cytokines (IL-2 and IL-12)and Th2 Cytokines (IL-4 and IL-10) in Immunocompromised HEV-Infected Pigs. To understand the host immune response to HEVinfection under immunosuppressive conditions, we simultaneouslymeasured the levels of IFN-γ, IL-2, IL-4, IL-10, and IL-12 in se-rum samples by Luminex multiplex assays. The average levels ofserum IL-2 were significantly (about three- to sevenfold) lower inimmunocompromised HEV-infected pigs than in immunocom-petent HEV-infected pigs at 1 and 4 wpi (P < 0.05) (Fig. 3A).Similarly, the average levels of serum IL-12 were significantly(about 0.5- to 0.75-fold) lower in immunocompromised HEV-infected pigs than in immunocompetent HEV-infected pigs at1 and 4 wpi (P < 0.01) (Fig. 3B). Although serum IFN-γ levelswere lower in immunocompromised HEV-infected pigs than inimmunocompetent HEV-infected pigs at 1 and 4 wpi, the ob-served differences were not significant (Fig. 3C).

The average serum IL-4 level in immunocompromised HEV-infected pigs at 1 wpi was the lowest observed in all treatmentgroups and was significantly lower than that in immunocompe-tent HEV-infected pigs (P < 0.05) (Fig. 3D). Similar to IL-2 andIL-12 responses in serum, the average levels of serum IL-10 weresignificantly (four- to eightfold) lower in immunocompromisedHEV-infected pigs than in the HEV-infected pigs at 1 and 4 wpi(P < 0.05) (Fig. 3E). However, the serum levels of all afore-mentioned cytokines were not significantly different among im-munocompromised mock-infected, immunocompetent mock-infected, and immunocompetent HEV-infected pig groups (Fig. 3).Overall, the serum levels of Th1 cytokines IL-2 and IL-12 andTh2 cytokines IL-4 and IL-10 were modestly reduced from base-line level of these cytokines in immunocompromised HEV-infected pigs, particularly during the early phase of HEV infection.

IFN-γ–Specific CD4+ T-Cell Responses Decrease During the AcutePhase of HEV Infection, but TNF-α–Specific CD8+ T-Cell ResponsesIncrease During the Chronic Phase of HEV Infection in Immuno-compromised Pigs. Peripheral blood mononuclear cells (PBMCs)were prepared, stimulated in vitro with purified recombinantHEV ORF2 antigen, and subsequently stained and analyzed byflow cytometry. The frequencies of IFN-γ– and/or TNF-α–producing T cells and IL-4–producing T cells were analyzedin three subpopulations, CD4+CD8− T cells (hereafter, “CD4+

T cells”), CD4−CD8+ T cells (hereafter, “CD8+ T cells”), andCD4+CD8+ T cells, and were expressed as the percentage ofT cells (CD3gated).HEV-specific T-cell responses were not significantly induced

in immunocompetent HEV-infected pigs compared with theimmunocompetent control pigs (Fig. 4 A–C). Within the threeaforementioned T-cell subpopulations in the peripheral blood,we observed no statistically significant differences in the meanfrequencies of IFN-γ+–secreting T-cells in immunocompromisedHEV-infected pigs and in immunocompetent HEV-infected pigsat 5, 8, and 13 wpi (Fig. 4 A–C). These results are consistent withthe absence of a significant increase in serum IFN-γ levels duringHEV infection in immunocompromised pigs compared withimmunocompetent control pigs (Fig. 3C).Similar to the IFN-γ responses in the peripheral blood, the

TNF-α–producing T-cell responses were not significantly in-creased within all three T-cell subpopulations during the acutephase of HEV infection (i.e., up to 8 wpi) (Fig. 4 D and E).However, at 13 wpi the TNF-α–producing CD8+ T-cell frequen-cies were increased significantly (two- to threefold) in both im-munocompromised and immunocompetent HEV-infected pigscompared with the frequencies in the immunocompromisedand immunocompetent control groups (P < 0.05) (Fig. 4F). Incontrast, the TNF-α responses at 13 wpi were not significantly

Fig. 2. Serum levels of the liver enzymes GGT and SDH. The serum levels of GGT (A) and SDH (B) were measured by established protocols and expressed asmean ± SEM. Control, immunocompetent noninfected (n = 9); HEV, HEV infected (n = 10); IC, immunocompromised (n = 10); IC HEV, immunocompromisedand HEV infected (n = 10). *P < 0.05.

6916 | www.pnas.org/cgi/doi/10.1073/pnas.1705446114 Cao et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020

augmented in the CD4+ T-cell and CD4+CD8+ T-cell subpop-ulations in response to HEV in both immunocompromised andimmunocompetent HEV-infected pigs compared with mock-infected pigs (Fig. 4F). When we analyzed the frequencies ofsingle- and double-cytokine T-cell producers specific to HEVwithin CD3gated T-cell populations, the frequencies of IFN-γ+TNF-α− CD4+CD8+ T cells were significantly higher in immuno-competent HEV-infected pigs than in immunocompromised HEV-infected pigs (P < 0.01) (Fig. 5A) and reached similar levels to those

in other groups at 8 wpi and 13 wpi (Fig. 5 B and C). At all timepoints, however, the mean frequencies of all three subpopulations ofIFN-γ+TNF-α+ T cells did not differ significantly in HEV-infectedpigs (Fig. 5 D–F). Moreover, the frequencies of IFN-γ−TNF-α+CD8+ T cells were significantly lower in immunocompromised pigsthan in immunocompetent pigs, regardless of HEV infection statusat 8 wpi, and at 13 wpi the levels in immunocompromised HEV-infected pigs were increased significantly (about twofold) above thebackground levels observed in pigs in other treatment groups (Fig. 5I).

Fig. 3. Levels of cytokines in pig serum samples collected at 1, 4, 7, and 13 wpi. The serum levels of IL-2 (A), IL-12 (B), IFN-γ (C), IL-4 (D), and IL-10 (E) weremeasured by Luminex multiplex technology and are expressed as mean ± SEM. Control, noninfected immunocompetent (n = 9); HEV, HEV infected (n = 10);IC, immunocompromised (n = 10); IC HEV, immunocompromised and HEV infected (n = 10). *P < 0.05; **P < 0.01.

Fig. 4. Frequencies of IFN-γ– and TNF-α–secreting CD4+, CD8+, and CD4+CD8+ T cells at 5, 8, and 13 wpi. The frequency IFN-γ–secreting T cells at 5 wpi(A), 8 wpi (B), and 13 wpi (C) and the frequency of TNF-α–secreting T cells at 5 wpi (D), 8 wpi (E), and 13 wpi (F) are expressed as mean ± SEM. PBMCs wereisolated from pigs at each time point and were stimulated with recombinant HEV ORF2 protein in vitro. The frequencies of intracellular IFN-γ+ or TNF-α+ T cellswere determined by flow cytometry. At each time point, n = 10 pigs in each group, except for the control group (n = 9 pigs). **P < 0.01; ***P < 0.001.

Cao et al. PNAS | July 3, 2017 | vol. 114 | no. 27 | 6917

MICRO

BIOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020

In summary, in immunocompromised pigs IFN-γ–specific CD4+

T-cell responses are reduced during the acute phase of HEVinfection, but TNF-α–specific CD8+ T-cell responses increasemoderately during the chronic phase of HEV infection.

Increased IL-4–Specific CD4+ T-Cell Responses Against HEV in Immunocompromised HEV-Infected Pigs. The mean frequencies ofIL-4–secreting T cells were not increased as a result of HEV infectionin all three T-cell subpopulations examined in immunocompetent

HEV-infected pigs as compared with the immunocompetent controlgroup (Fig. 6). However, in both HEV-infected and noninfectedimmunocompromised pigs, the mean IL-4–specific CD4+ T-cell fre-quencies were significantly higher than in the immunocompetentcontrol group at 8 wpi (P < 0.05) (Fig. 6B). On the other hand, themean IL-4–specific CD4+CD8+ T-cell frequencies were signifi-cantly higher in immunocompromised HEV-infected pigs than inimmunocompetent HEV-infected pigs at 13 wpi (P < 0.001) (Fig.6C). In contrast, the mean frequencies of IL-4+ CD8+ T cells

Fig. 5. Frequency of TNF-α+ and IFN-γ+ double-positive CD4+, CD8+, and CD4+CD8+ T cells at 5, 8, and 13 wpi. The frequencies of TNF-α−IFN-γ+–secreting T cells at5 wpi (A), 8 wpi (B), 13 wpi (C), the frequency of TNF-α+IFN-γ+–secreting T cells at 5 wpi (D), 8 wpi (E), and 13 wpi (F), and the frequency of TNF-α+IFN-γ−–secretingT cells at 5 wpi (G), 8 wpi (H), and 13 wpi (I) are expressed as mean ± SEM. PBMCs were isolated from pigs at each time point and were stimulatedwith recombinant HEV ORF2 protein in vitro. The frequencies of intracellular IFN-γ+ and/or TNF-α+ T cells were determined by flow cytometry. At each time point,n = 10 pigs in each group, except for the control group (n = 9 pigs). *P < 0.05; **P < 0.01; and ***P < 0.001.

Fig. 6. Frequency of IL-4–secreting CD4+, CD8+, and CD4+CD8+ T cells at 5 (A), 8 (B), and 13 (C) wpi. The frequency of IL-4–secreting T cells is expressed asmean ± SEM. PBMCs were isolated from pigs at each time point and were stimulated with recombinant HEV ORF2 protein in vitro. The frequencies of in-tracellular IL-4+ T cells were determined by flow cytometry. At each time point, n = 10 pigs in each group, except for the control group (n = 9 pigs). *P < 0.05;***P < 0.001.

6918 | www.pnas.org/cgi/doi/10.1073/pnas.1705446114 Cao et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020

did not differ significantly between treatment groups at any ofthe time points observed (Fig. 6). Our results suggest that theIL-4–specific CD4+ T-cell responses against HEV were greatlyincreased in immunocompromised HEV-infected pigs; theseresponses were found mainly within CD4+CD8− T cells at8 wpi and shifted to the CD4+CD8+ T-cell compartment duringthe chronic phase of HEV infection.

The Immunosuppressive Regimen, Not HEV Infection, Up-Regulatedthe Regulatory T-Cell Immune Responses in Pigs. Using multicolorflow cytometry, we also analyzed the frequencies of Treg cells(CD4+CD25+Foxp3+ and CD4+CD25−Foxp3+ Treg cells) andintracellular regulatory cytokine production within these cellsafter in vitro antigen stimulation of PBMCs. No significantchanges in the frequencies of either Treg cell subpopulation wereobserved in the peripheral blood of pigs from any of the treat-ment groups (Fig. 7 A–C), except that at the time of necropsy(13 wpi) the frequencies of CD4+CD25+Foxp3+ Treg cells wereslightly lower in immunocompromised pigs (both mock- andHEV-infected) than in immunocompetent pigs (both mock- andHEV-infected) (P < 0.05) (Fig. 7C).When we analyzed the regulatory cytokine production in these

Treg cell subpopulations at 5 wpi, the frequencies of IL-10–secretingCD4+CD25+Foxp3+ Treg cells were highest in immunocompe-tent HEV-infected pigs (Fig. 7D) and reached insignificant low

levels at 8 wpi and 13 wpi (Fig. 7 E and F); however, the differencewas not statistically significant. In contrast, HEV infection exertedno apparent influence on the frequencies of IL-10–secreting CD4+

CD25−Foxp3+ Treg cells in immunocompromised pigs (Fig. 7 D–F). A moderate but significant increase in the frequencies ofIL-10–secreting CD4+CD25−Foxp3+ Treg cells was observed in im-munocompromised mock-infected pigs at 13 wpi compared withnegative control pigs (P < 0.05) (Fig. 7F).There was no significant increase in the frequencies of TGF-

β–secreting CD4+CD25+Foxp3+ Treg cells in the HEV-infectedpigs compared with other groups at 5 wpi (Fig. 7G). We also foundthat the HEV infection reduced the frequencies of TGF-β–secretingCD4+CD25−Foxp3+ Treg cells in both immunocompromised andimmunocompetent pigs at 8 wpi (Fig. 7H); however, the reductionwas significant only in immunocompetent pigs (P < 0.05). Themean frequencies of TGF-β–secreting CD4+CD25−FoxP3+ andCD4+CD25+FoxP3+ Treg cells were at very low levels by the endof study (13 wpi) (Fig. 7I).

DiscussionChronic hepatitis E has been increasingly recognized as anemerging and significant clinical problem in immunocompro-mised individuals, particularly in solid-organ transplant recipi-ents (8, 10). The lack of an animal model for chronic HEVinfection greatly hinders the development of specific antiviral

Fig. 7. Frequency of CD25+ and CD25− Treg cells at 5, 8, and 13 wpi. The frequency of Treg cells at 5 wpi (A), 8 wpi (B), and 13 wpi (C); of IL-10–secreting Tregcells at 5 wpi (D), 8 wpi (E), and 13 wpi (F); and of TGF-β–secreting Treg cells at 5 wpi (G), 8 wpi (H), and 13 wpi (I) were measured by flow cytometry andexpressed as means ± SEM. At each time point, n = 10 pigs, except for the control group (n = 9 pigs). *P < 0.05.

Cao et al. PNAS | July 3, 2017 | vol. 114 | no. 27 | 6919

MICRO

BIOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020

therapies to tackle the public heath burden caused by chronichepatitis E. In this study, we successfully established a pig modelfor chronic hepatitis E through oral treatment of pigs with aclassical immunosuppressive drug regimen similar to that usedfor human organ transplant recipients and experimental infectionof pigs with a genotype 3 human HEV. This unique chronic HEVpig model will aid in the future study of the mechanism of HEVpathogenesis and immune responses during chronic HEV infectionas well as in the development of specific antivirals for chronichepatitis E.The pig model presented here is ideally suited for the study of

chronic hepatitis E for a number of reasons. Recently, humanliver-grafted chimeric mice were proved to be a valuable smallanimal model for HEV study, especially for studying the intrinsictype I IFN signaling pathways in human hepatocytes againstHEV infection (21–23). Although the chimeric mouse model canreproduce chronic HEV infection, it is not a natural host forHEV, and interaction of innate pathways of human cells with theadaptive immune pathways of mouse cells cannot be fullyreproduced in chimeric mouse model. Nevertheless, this modelcould be improved by restoring the adaptive immune response inthe chimeric mouse to provide a broader application for HEVstudy in the future. As a natural host of genotypes 3 and 4 HEV,the pig has been consistently proven to be an excellent animalmodel for HEV infection, pathogenesis, and immunity studies(24–26). Because the chronic hepatitis E cases in humans havebeen associated almost exclusively with genotype 3 HEV (7),and, more importantly, because the physiological characteristicsand immune system of pigs closely resemble those of humans,the pig is a very attractive model system to study chronic hepa-titis E by mimicking the status of immunocompromised humanpatients, especially solid-organ transplant recipients.During the course of HEV infection, fecal virus shedding

typically ceases at ∼3–4 wk in immunocompetent humans, al-though in some patients, it may last up to 7–8 wk (27). In thecases of chronic hepatitis E in humans, the fecal virus sheddingcontinues beyond 4 wk for more than 3 mo after infection (28).In our study, fecal virus shedding ceased after 8 wpi in the greatmajority of pigs not treated with immunosuppressive drugs, as isconsistent with the previous reports on the duration of fecal virusshedding in pigs infected with genotype 3 HEV (16). In theimmunocompromised pigs, however, fecal virus shedding lastedat least 5–14 wk longer than in immunocompetent pigs, in-dicating that the disease course of chronic HEV infection in pigsmimics that of human organ transplant recipients undergoing theimmunosuppressive regimen treatment and meets the definitionof chronic HEV infection suggested by Kamar et al. (28). Ourresults also suggest that the serum viral RNA levels are not re-liable predictors for the establishment of chronic hepatitis E,because the viral RNA levels were transiently increased in seraof immunocompromised pigs at the start of the chronic phase(9 wpi), and thereafter the levels were similar to those in immu-nocompetent HEV-infected pigs, remaining at a low level until theend of the study.Because HEV infection normally causes only mild microscopic

liver lesions with no gross lesions between 5 wpi and 8 wpi inpigs, it is not surprising that no pathological lesions were ob-served during necropsy at 13 wpi in the immunocompetentHEV-infected pigs. Moreover, the immunocompromised condi-tion in pigs apparently did not aggravate liver damage in thechronic phase of HEV infection, as evidenced by the absence ofsignificant differences in gross and microscopic pathological le-sions in the immunocompromised HEV-infected and immuno-competent HEV-infected groups. Indeed, the higher serumlevels of GGT in the immunocompromised HEV-infected pigssuggested that unapparent liver damage occurred during thechronic phase of HEV infection but did not proceed to signifi-cant pathological lesions in liver as previously reported with

solid-organ transplant recipients with chronic hepatitis E (29,30). Many factors, such as the duration of HEV infection anddifferences in immunological and physiological systems in pigsand humans may contribute to the observed milder pathologicallesions in immunosuppressed pigs. For example, as shown by theabsence of obvious adverse effects or opportunistic infections indrug-treated pigs during the course of infection, the level ofimmunosuppression induced in pigs is milder than that observedin human patients, and this milder immunosuppression mightcontribute to the observed milder pathological lesions in pigs ascompared with immunosuppressed human patients. The durationof chronic HEV infection also likely played a role in the severity ofimmunopathogenicity. Were the chronically infected pigs kept fora longer period, it is possible that significant pathological liverlesions might develop gradually because of repeated viral-inducedinjury in the liver. In fact, a recent study reported more pro-nounced hepatic lesions in HEV-infected cynomolgus monkeysthat received long-term (160-d) immunosuppressive treatmentwith tacrolimus (31).Because the deficiency in immune response contributes to the

establishment of chronic HEV infection in humans (32, 33), it iscritical to examine the nature of immune responses developedagainst chronic HEV infection in immunocompromised pigs toidentify potential immune correlates leading to chronic HEVinfection. Cytokines play important roles in the regulation of theinnate, humoral, and cellular immune responses against virusinfection (34). We found that HEV infection of immunocom-petent pigs did not up-regulate serum levels of the cytokinesparticipating in cell-mediated immunity. However, the baselineserum levels of Th1 cytokines (IL-2 and IL-12) were reducedwhen immunocompromised pigs were infected with HEV,whereas, importantly, the immunosuppressive drug treatmentalone without HEV infection did not suppress these cytokines.This finding indicates that HEV infection actively suppressesTh1 immune responses in immunocompromised pigs, as hasbeen observed in human patients chronically infected with HEV(32). Interestingly, HEV infection did not significantly increaseor reduce IFN-γ cytokine levels in sera of either immunocom-petent or immunocompromised pigs. However, HEV infectionspecifically increased the activation of CD4+CD8+ T cells toproduce IFN-γ in immunocompetent pigs, and the activation wasreduced significantly in immunocompromised pigs. IFN-γ+CD4+CD8+ T cells are the critical effector memory Th cells in pigsnecessary to mount an effective cell-mediated immune responseagainst virus infections (35, 36), and these findings are consistentwith previous observations in human hepatitis E patients (37, 38).Therefore, the active suppression of HEV-specific Th1 immuneresponses under immunocompromised conditions may be aplausible explanation for the prolonged fecal virus shedding andestablishment of chronic infection observed in the immunocom-promised HEV-infected pigs. Furthermore, this HEV-specificimmunosuppression may be a potentiating factor for the effectsof the immunosuppressive drugs in pigs, because one of the drugs,cyclosporine A, is known to down-regulate serum IL-2 and IL-12levels (39). However, the effect of drugs alone was not strongenough to reduce these cytokines levels significantly, whereas thepossible synergistic effect between drugs and HEV infection re-duced these cytokine levels significantly in immunocompromisedHEV-infected pigs compared with the levels in immunocompetentHEV-infected pigs.T cells producing multiple cytokines are functionally more

potent than those producing a single cytokine (40) and are re-liable correlates of protective T-cell immunity. In contrast tomice and humans, there are significant CD4+CD8+ T cell sub-populations consisting of activated Th cells, memory Th cells,and effector T cells (41) in pigs (Fig. S1). The porcine double-positive CD4+CD8+ cells are effector memory CD4+ T cells whichrepeatedly experienced a wide variety of antigens (pathogenic or

6920 | www.pnas.org/cgi/doi/10.1073/pnas.1705446114 Cao et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020

derived from gut-residing microbes). They produce both Th1 andTh2 cytokines, are major producers of these cytokines whenstimulated with exogenous antigen in vitro, and also have a sub-population of cytotoxic T cells that are important for antiviralimmunity in pigs. Therefore, it is important to analyze the functionof this T-cell subpopulation in HEV-infected pigs. When welooked into the IFN-γ and TNF-α production in CD4+CD8+

T cells during the early phase of HEV infection, we found bothsingle (IFN-γ only) and double cytokine producers in infectedpigs, but, interestingly, the single (IFN-γ only) but not the doublecytokine-producing CD4+CD8+T cells were reduced in immuno-compromised HEV-infected pigs. As the HEV infection pro-gressed into the chronic phase, there was no induction of doublecytokine-producing CD4+CD8+T cells, suggesting that the pro-tective T-cell immune responses against HEV were not inducedduring chronic infection. Moreover, HEV failed to induce theproduction of IFN-γ in the activated CD8+ T cells while main-taining or even increasing the TNF-α production in those cells, asevident from the increased levels of the single cytokine (TNF-αonly)-producing CD8+ T cells in the chronic phase of HEV infection.The contribution of these cells in the progression of chronic hep-atitis E warrants further investigation.Contrary to the previous report on the study of hepatitis E

patients (42), HEV infection in pigs did not increase IL-4 andIL-10 serum levels in either normal or immunocompromised pigsat 1 wpi. Even though HEV infection reduced the baseline levelsof Th2 cytokines in the serum of immunocompromised pigs, itincreased the levels of IL-4–producing CD4+ T cells in the pe-ripheral blood, particularly during the chronic phase of infection.These observations suggest that, although Th1 immune responsesagainst HEV were reduced by the treatment with immunosup-pressive drugs, the Th2 immune responses against the virus wereincreased in immunocompromised pigs. Specifically, the Th2 im-mune responses were shifted from a CD4+CD8− T-cell populationto a CD4+CD8+ T-cell population as the HEV infection pro-gressed into the chronic phase, further confirming the sustainedstimulation of Th2 immunity. Because HEV in sera exists in aquasi-enveloped form, which is resistant to neutralization by anti-bodies (43), the enhanced levels of antibodies against HEV inimmunocompromised pigs may not neutralize the virus efficiently.Overall, the imbalance in T-cell immunity (Th1 vs. Th2) againstHEV in immunocompromised pigs might play a key role in theprogression of chronic HEV infection and needs further in-depthinvestigation.In addition to Th1 and Th2 cells, Treg cells play important

roles in regulating the immune response to maintain a balancebetween the recognition and clearance of infectious agents andminimizing immune-mediated pathology (44). Treg cells gener-ally produce immunosuppressive cytokines such as IL-10 andTGF-β, which reflect the effector functions of the Treg cells (44–46). Our study showed that the immunosuppressive treatment ofpigs increased IL-10–producing Treg cell (CD25−) levels non-specifically. In general, HEV did not modulate the immunesystem by inducing Treg cells (CD25+ and CD25− cells producingTGF-β or IL-10), as observed in this study. However, HEV-specific Treg responses in the CD25− Treg compartment mightbe affected, and TGF-β production in these cells was reducedtransiently at the start of the chronic phase of HEV infection.The relatively weak T-cell response in HEV-infected non-

immunosuppressed pigs observed in this study is not surprising.The strength of antiviral immune responses is determined partlyby the virulence of the virus and the level of pathogenicity. Al-though pigs are the natural host for genotype 3 HEV, the virusinfection in pigs is typically subclinical and thus may have in-duced a dampened immune response. For example, at 5 wpi (thefirst time point tested) HEV replication and fecal virus sheddingwere significantly reduced, and by 8 wpi the majority of animalscleared virus infection. Thus, the weak immune response seen in

nonimmunosuppressed HEV-infected pigs at these time pointsmay be caused by a low level of virus replication or by virusclearance. Furthermore, pig blood is only a transit point for theantigen-activated T cells leaving the draining lymph nodes andspleen, and these cells may infiltrate quickly into the infected liverand reside there for a long period, until viral clearance has takenplace. Therefore, it is likely that we may have missed the right timepoint(s) for analyzing the peak T-cell responses in peripheralblood and that a more robust immune response could have oc-curred earlier during the course of infection. These factors couldhave contributed to the observation of lower HEV-specific T-cellfrequencies in pig peripheral blood. Additionally, although themajor subpopulations of porcine lymphocytes are very similar tothose in humans, two important differences between human andpig—the abundant surface expression of CD8a and the expressionof MHC-II DR molecules on resting T lymphocytes in pigs—alsomight contribute to the observed differences in the course of HEVinfection and immune responses in pig and human.In conclusion, we successfully established a pig model of

chronic HEV infection by oral treatment with immunosuppres-sive drugs routinely used for human organ transplant patientsand infection with a genotype 3 human HEV. We demonstratedthat this unique pig model mimicked the course of chronic HEVinfection and the immune response status observed in humansolid-organ transplant recipients chronically infected with HEV.We found that HEV-specific Th1 immune responses were sig-nificantly reduced by treatment with immunosuppressive drugsand that the immune responses against HEV were skewed to-ward Th2 immunity through the chronic phase of HEV infection.Although we observed transient effects of CD4+CD25− Tregimmune responses against HEV infection in pigs, these effectsmight not have influenced the progression of chronic HEV in-fection. With this unique chronic HEV model, future in-depthstudies can be conducted to delineate the precise roles of thedifferent arms of immune system in the progression of chronichepatitis E in immunosuppressed individuals and to developeffective and specific antiviral therapies for chronic hepatitis E.

Materials and MethodsVirus and Immunosuppressive Drugs. The genotype 3 HEV infectious stock (US-2 strain) (47) was prepared from the feces of a pig experimentally infectedwith the US-2 strain of human HEV and was used to infect pigs in this study.The viral genomic equivalent (GE) titer of this virus stock was determined byHEV-specific real-time RT-PCR. Drugs used in the immunosuppressive regi-men of the study are cyclosporine (Cyclosporine oral solution USP modified,100 mg/mL; Teva Pharmaceutical Industries Ltd.), azathioprine (50-mg tab-lets; Roxane Laboratories, Boehringer Ingelheim), and prednisolone (50-mgtablets; Roxane Laboratories, Boehringer Ingelheim) (48). All animal exper-iments were approved by the Virginia Tech Institutional Animal Care andUse Committee (IACUC no. 12-165-CVM).

Treatment of Pigs with an Immunosuppressive Regimen and HEV Infection. Atotal of 39 HEV-negative 6-wk-old pigs (purchased without consideration ofsex from the Virginia Tech Tidewater Agricultural Research and ExtensionCenter) were divided randomly into four treatment groups with 10 pigs ineach group, except for the control group, which had nine pigs. In general,pigs become fully immunocompetent at the age of 7 wk (49); therefore wepurchased 6-wk-old immune-maturing piglets and infected them with HEVat age 7 wk. Pigs in groups 1 and 2 were treated daily for 7 d with an oralimmunosuppressive regimen similar to that used for human organ transplantrecipients (cyclosporine at 10 mg·kg−1·d−1, azathioprine at 2 mg·kg−1·d−1,and methylprednisolone at 4 mg·kg−1·d−1) (48). Briefly, azathioprine tabletswere crushed into powder, mixed with 1 mL sterile water together with onedose of methylprednisolone and cyclosporine solutions to produce a sus-pension. The three-drug suspension then was mixed thoroughly with twovolumes of maple syrup and was administered to each pig orally with afeeding syringe. Pigs in groups 3 and 4 were treated with PBS with the sameprotocol. After 7 d of treatment, pigs in groups 1 and 3 were experimentallyinfected with 1 mL (∼3.78 × 109 GE titer) of a genotype 3 human HEV (US-2strain) through i.v. injection in the ear vein. Pigs in groups 2 and 4 weremock-infected and served as immunocompromised and immunocompetent

Cao et al. PNAS | July 3, 2017 | vol. 114 | no. 27 | 6921

MICRO

BIOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020

negative controls, respectively. After HEV inoculation, the pigs in groups1 and 2 were treated with the same daily immunosuppressive regimen untilthe end of the study. Blood, serum, and fecal samples were collected weeklyuntil 13 wpi, when the pigs were humanely killed and necropsied. Tissuesamples of liver, small intestines, and mesenteric and superficial inguinallymph nodes were collected during necropsy for histological examination ofpathological lesions. Serum, fecal, and tissue samples were stored at −80 °Cuntil further processing and analysis.

Two of the immunocompromised HEV-infected pigs were not necropsiedat 13 wpi and were kept on the daily immunosuppressive regimen for anadditional 9wk.Weekly fecal samples from these two pigs were collected andtested by HEV-specific RT-PCR for fecal virus shedding to verify further thechronic HEV infection status.

Detection and Quantification of HEV RNA by qRT-PCR. Viral RNAs in weekly pigserum and fecal samples were quantified by qRT-PCR using HEV-specific primers.Briefly, the fecal sampleswere suspended in sterile PBS at 10% (wt/vol). The serumsampleswere diluted in sterile PBS at 10% (vol/vol). Total RNAwas extracted from250 μL of 10% fecal suspensions or diluted serum samples with TRIzol LS Reagent(Invitrogen). HEV genomic RNAs were quantified using the SensiFAST ProbeNo-ROX One-Step kit (Bioline USA, Inc.), using the forward primer (JVHEVF,5′-GGTGGTTTCTGGGGTGAC-3′), reverse primer (JVHEVR, 5′-AGGGGTTGGTTG-GATGAA-3′), and a hybridization probe (JVHEVP, 5′-TGATTCTCAGCCCTTCGC-3′)following a protocol described previously (50). The qRT-PCR assays were per-formed in a CFX96 real-time (RT) PCR system (Bio-Rad Laboratories). In vitro-transcribed and -purified HEV genomic RNAs were used to produce a standardcurve in qRT-PCR assays. The thermal cycling conditions in qRT-PCR assays were asfollows: 45 °C for 10 min (reverse transcription); 95 °C for 2 min (initial de-naturation); and 95 °C for 5 s followed by 60 °C for 20 s (PCR amplification) for40 cycles. The detection limit of the qRT-PCR assay is set at 10 viral genomiccopies, as reported previously (50, 51), which is equivalent to 400 copies per 1-mLsample. The titer below the detection limit was considered as negative.

Determination of Serum Levels of Liver Enzymes in Pigs. A panel of liver en-zymes including GGT, SDH, AST, total bilirubin, and alkaline phosphatase weremeasured in serum samples by established protocols at the Clinical PathologyLaboratory at Iowa State University College of Veterinary Medicine (Ames, IA).

Detection of Cytokines in Serum Samples from Pigs. The levels of IL-2, IL-4, IL-10, IL-12, and IFN-γ in pig serum samples were assayed using the MILLIPLEXMAP Porcine Cytokine kit (Millipore Corporation) according to the manu-facturer’s instructions. The cytokine analyses were performed using the Bio-Plex system (Bio-Rad Laboratories).

Expression and Purification of HEV ORF2 Capsid Protein. A truncated ORF2(amino acid residues 112–660) of the genotype 3 human HEV (US-2 strain)was amplified by RT-PCR using viral RNAs extracted from a HEV-infected pigfecal sample. The primers used for RT-PCR were 430F22 (5′-TGGGATCCA-TATGGCCGTGTCACCGGCTCCTGACA-3′) and 2064R25 (5′-CCAAGCTTAGGAAT-TAATTAAGACTCCCGGGTT-3′). Restriction enzyme sites for NdeI and Hind III(underlined in primer sequences) were appended in the primers and used toinsert the amplified cDNA into a pET28a expression plasmid. The HEV ORF2cDNA insert in the recombinant plasmid pET28a-US-2-ORF2 was confirmed bySanger DNA sequencing. Escherichia coli BL21(DE3)pLysS cells were transformedwith the recombinant plasmid, and the truncated ORF2 protein expression wasinduced by isopropyl β-d-1-thiogalactopyranoside (IPTG) induction. The re-combinant truncated HEV ORF2 protein was purified from E. coli soluble frac-tions using a His-tag by nickel-NTA agarose affinity chromatography (ProBondkit; Invitrogen). For the in vitro T-cell stimulation experiment, endotoxin-freetruncated HEV ORF2 protein was subsequently prepared by treating the puri-fied recombinant protein with Triton X-114 as previously described (52).

Intracellular Cytokine Staining and Flow Cytometry Analysis. PBMCs wereisolated by density-gradient centrifugation with Ficoll-Paque PREMIUM (GEHealthcare) and were resuspended in RPMI medium 1640 supplemented with10% FBS and antibiotics (1% penicillin/streptomycin). The PBMCs wereseeded in 96-well tissue-culture plates (1 × 106 cells per well) andwere stimulated with purified recombinant HEV ORF2 protein at a final

concentration of 12 μg/mL for 12 h. A cell stimulation mixture (eBioscience,Inc.) was used as the positive control in T-cell stimulations. An anti-humanCD49d (0.5 μg/mL) mAb (Thermo Fisher Scientific) was added to each well forcostimulation. Brefeldin A (0.2 μL per well) (GolgiPlug; BD Biosciences) wasadded to each well and incubated for five additional hours before pro-ceeding to antibody staining.

PBMCs were washed once with PBS and were stained with LIVE/DEADFixable Aqua Dead Cell Stain Kit (catalog no. L34966; Thermo Fisher Scientific)at 4 °C for 30 min. The stained cells were washed twice with PBS containing2% FBS. LIVE/DEAD-stained PBMCs were divided into two groups, and eachgroup was stained again with one of the following two panels offluorochrome-conjugated antibodies according to the procedures previouslydescribed (53, 54), with minor modifications. Briefly, in panel one, PBMCs(1 × 106 cells per tube) were sequentially stained with the following mAbsets diluted in staining buffer (PBS containing 2% FBS) at 4 °C and an in-cubation time of 15 min for each mAb set: FITC-conjugated mouse anti-pigCD4 (IgG2b, clone 74-12-4; BD Pharmingen), Spectral Red-conjugated mouseanti-pig CD8α (IgG2aκ, clone 76-2-11; Southern Biotech), and mouse anti-pigCD3e (IgG1κ, clone PPT3; Southern Biotech), followed by phycoerythrin (PE)-Cy7–conjugated rat anti-mouse IgG1 (eBioscience). After staining of cell-surfacemarkers, the PBMCs were permeabilized with BD Cytofix/Cytoperm buffer (BDPharmingen) at 4 °C for 30 min. PBMCs were washed three times with BD Perm/Wash buffer (BD Pharmingen) and stained with R-PE–conjugated mouse anti-pigIFN-γ (IgG1, clone P2G10; BD Pharmingen), Brilliant Violet 421 rat anti-human IL-4 antibody (IgG1κ clone MP4-25D2; BioLegend), and allophycocyanin (APC) anti-human TNF-α antibody (IgG1κ, clone MAB11; BioLegend) at 4 °C for 30 min.PBMCs stained with appropriate isotype-matched irrelevant control antibodies(BD Pharmingen, VMRD, BioLegend, or Southern Biotech) served as backgroundstaining controls. The frequencies of IFN-γ+ CD4+CD8−, IFN-γ+ CD4−CD8+, andIFN-γ+ CD4+CD8+ T cells were expressed as percentages of parental CD3gated

T cells.In panel two, in vitro antigen-stimulated PBMCs were stained with relevant

antibodies to determine the frequencies of CD4+ FoxP3+ Treg cells, the acti-vation status (CD25), and the expression of regulatory cytokines (IL-10 andTGF-β) in these cells (54). Briefly, PBMCs (1 × 106 cells per well) were firststained at 4 °C for 15 min with FITC-conjugated mouse anti-porcine CD4,Spectral Red-conjugated mouse anti-porcine, and mouse anti-porcine CD25(IgG1, clone K231.3B2; AbD Serotec), followed by APC-conjugated rat anti-mouse IgG1 (IgG1, clone X56; BD Pharmingen). The PBMCs were sub-sequently permeabilized with the FoxP3-staining buffer set (eBioscience) at4 °C for 30 min. The permeabilized cells were stained with the following an-tibodies: PE-Cy7–conjugated rat anti-mouse/rat FoxP3 (IgG2a, clone FJK-16s;eBioscience), Brilliant Violet 421 anti-human IL-10 antibody (IgG1, JES3-9D7;BioLegend), and PE-conjugated mouse anti-human TGF-β1 (IgG1, clone 27232;R&D Systems) at 4 °C for 30 min. The appropriate isotype-matched irrelevantcontrol antibodies and PE-conjugated mouse IgG1 isotype control (IgG1, cloneP3.6.2.8.2; eBioscience) were used as the background staining controls. At least100,000 events were acquired on a FACSCalibur flow cytometer (BD Biosci-ences), and the data were analyzed using FlowJo V10 software (Tree Star).

The absolute number of Treg cells in each sample was calculated basedon the frequencies of Treg cells. The frequencies of CD4+CD25+FoxP3+ andCD4+CD25−FoxP3+ cell subpopulations were expressed as the percentages ofgated CD4+ PBMCs. The frequencies of IL-10+ or TGF-β+ cells were expressedas the percentages of parental Treg populations. In all cases, the T-cell fre-quencies were calculated by subtracting the values obtained with unstimu-lated cells from the values obtained with stimulated cells.

Statistical Analysis. All data were processed with GraphPad Prism 6.01(GraphPad Software Inc.). The differences between the mean values of twotreatment groups were analyzed by two-tailed unpaired student’s t test ortwo-way ANOVA followed by Tukey multiple comparisons test. A P value ofless than 0.05 was considered as statistically significant.

ACKNOWLEDGMENTS. We thank Ms. Melissa Makris for her technicalassistance in flow cytometry; Dr. Ke Wen for his helpful discussion on thestaining protocol of intracellular cytokines; and Karen Hall, Kimberly Allen,and Peter Jobst for their timely support in the animal study. This study wassupported by NIH Grants R01AI050611 and R01AI074667.

1. Rein DB, Stevens GA, Theaker J, Wittenborn JS, Wiersma ST (2012) The global burden

of hepatitis E virus genotypes 1 and 2 in 2005. Hepatology 55:988–997.2. Lozano R, et al. (2012) Global and regional mortality from 235 causes of death for

20 age groups in 1990 and 2010: A systematic analysis for the Global Burden of Dis-

ease Study 2010. Lancet 380:2095–2128.

3. Cao D, Meng XJ (2012) Molecular biology and replication of hepatitis E virus. Emerg

Microbes Infect 1:e17.4. Smith DB, et al.; International Committee on Taxonomy of Viruses Hepeviridae Study

Group (2014) Consensus proposals for classification of the family Hepeviridae. J Gen

Virol 95:2223–2232.

6922 | www.pnas.org/cgi/doi/10.1073/pnas.1705446114 Cao et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020

5. Emerson SU, Purcell RH (2007) Hepatitis E virus. Fields Virology, eds Knipe DM,Howley PM (Lippincott-Raven Publishers, Philadelphia), 5th Ed, pp 3048–3059.

6. Pérez-Gracia MT, García M, Suay B, Mateos-Lindemann ML (2015) Current knowledgeon hepatitis E. J Clin Transl Hepatol 3:117–126.

7. Dalton HR, Webb GW, Norton BC, Woolson KL (2016) Hepatitis E virus: Time tochange the textbooks. Dig Dis 34:308–316.

8. Kamar N, Dalton HR, Abravanel F, Izopet J (2014) Hepatitis E virus infection. ClinMicrobiol Rev 27:116–138.

9. Haagsma EB, et al. (2008) Chronic hepatitis E virus infection in liver transplant re-cipients. Liver Transpl 14:547–553.

10. Sue PK, et al. (2016) Hepatitis E virus infection among solid organ transplant recipi-ents at a North American Transplant Center. Open Forum Infect Dis 3:ofw006.

11. Lee GY, et al. (2015) Hepatitis E virus infection: Epidemiology and treatment impli-cations. World J Virol 4:343–355.

12. Ding Q, et al. (2017) Hepatitis E virus ORF3 is a functional ion channel required forrelease of infectious particles. Proc Natl Acad Sci USA 114:1147–1152.

13. Meng XJ (2016) Expanding host range and cross-species infection of hepatitis E virus.PLoS Pathog 12:e1005695.

14. Meng XJ (2011) From barnyard to food table: The omnipresence of hepatitis E virusand risk for zoonotic infection and food safety. Virus Res 161:23–30.

15. Feagins AR, Opriessnig T, Huang YW, Halbur PG, Meng XJ (2008) Cross-species in-fection of specific-pathogen-free pigs by a genotype 4 strain of human hepatitis Evirus. J Med Virol 80:1379–1386.

16. Halbur PG, et al. (2001) Comparative pathogenesis of infection of pigs with hepatitis Eviruses recovered from a pig and a human. J Clin Microbiol 39:918–923.

17. Meng XJ (2010) Hepatitis E virus: Animal reservoirs and zoonotic risk. Vet Microbiol140:256–265.

18. Murali AR, Kotwal V, Chawla S (2015) Chronic hepatitis E: A brief review. World JHepatol 7:2194–2201.

19. Singh A, et al. (2016) Chronic hepatitis E–An emerging disease in an immunocom-promised host. Gastroenterol Rep (Oxf), 10.1093/gastro/gow024.

20. Yugo DM, Cossaboom CM, Meng XJ (2014) Naturally occurring animal models ofhuman hepatitis E virus infection. ILAR J 55:187–199.

21. van de Garde MD, et al. (2016) Hepatitis E virus (HEV) genotype 3 infection of humanliver chimeric mice as a model for chronic HEV infection. J Virol 90:4394–4401.

22. Allweiss L, et al. (2016) Human liver chimeric mice as a new model of chronic hepatitisE virus infection and preclinical drug evaluation. J Hepatol 64:1033–1040.

23. Sayed IM, et al. (2017) Study of hepatitis E virus infection of genotype 1 and 3 in micewith humanised liver. Gut 66:920–929.

24. Meng XJ, et al. (1998) Experimental infection of pigs with the newly identified swinehepatitis E virus (swine HEV), but not with human strains of HEV. Arch Virol 143:1405–1415.

25. Feagins AR, et al. (2011) Intergenotypic chimeric hepatitis E viruses (HEVs) with thegenotype 4 human HEV capsid gene in the backbone of genotype 3 swine HEV areinfectious in pigs. Virus Res 156:141–146.

26. Sanford BJ, et al. (2011) Prior infection of pigs with a genotype 3 swine hepatitis Evirus (HEV) protects against subsequent challenges with homologous and heterolo-gous genotypes 3 and 4 human HEV. Virus Res 159:17–22.

27. Chauhan A, et al. (1993) Hepatitis E virus transmission to a volunteer. Lancet 341:149–150.

28. Kamar N, Rostaing L, Legrand-Abravanel F, Izopet J (2013) How should hepatitis Evirus infection be defined in organ-transplant recipients? Am J Transplant 13:1935–1936.

29. Marion O, Abravanel F, Lhomme S, Izopet J, Kamar N (2016) Hepatitis E in trans-plantation. Curr Infect Dis Rep 18:8.

30. Gérolami R, Moal V, Picard C, Colson P (2009) Hepatitis E virus as an emerging cause ofchronic liver disease in organ transplant recipients. J Hepatol 50:622–624.

31. Gardinali NR, et al. (2017) Cynomolgus monkeys are successfully and persistently in-fected with hepatitis E virus genotype 3 (HEV-3) after long-term immunosuppressivetherapy. PLoS One 12:e0174070.

32. Lhomme S, et al. (2012) Hepatitis E virus quasispecies and the outcome of acutehepatitis E in solid-organ transplant patients. J Virol 86:10006–10014.

33. Kamar N, et al. (2008) Hepatitis E virus and chronic hepatitis in organ-transplant re-cipients. N Engl J Med 358:811–817.

34. Wack A, Openshaw P, O’Garra A (2011) Contribution of cytokines to pathology andprotection in virus infection. Curr Opin Virol 1:184–195.

35. Ober BT, et al. (1998) Vaccine-induced, pseudorabies virus-specific, extrathymicCD4+CD8+ memory T-helper cells in swine. J Virol 72:4866–4873.

36. Zuckermann FA, Husmann RJ (1996) Functional and phenotypic analysis of porcineperipheral blood CD4/CD8 double-positive T cells. Immunology 87:500–512.

37. Tripathy AS, Das R, Rathod SB, Arankalle VA (2012) Cytokine profiles, CTL responseand T cell frequencies in the peripheral blood of acute patients and individuals re-covered from hepatitis E infection. PLoS One 7:e31822.

38. Brown A, et al. (2016) Characterization of the specificity, functionality, and durabilityof host T-cell responses against the full-length hepatitis E virus. Hepatology 64:1934–1950.

39. Dean SK, Horley DC, Tuch BE (2003) The immunosuppression of pigs: A review.Transplant Rev 17:45–59.

40. Kannanganat S, Ibegbu C, Chennareddi L, Robinson HL, Amara RR (2007) Multiple-cytokine-producing antiviral CD4 T cells are functionally superior to single-cytokine-producing cells. J Virol 81:8468–8476.

41. Gerner W, Käser T, Saalmüller A (2009) Porcine T lymphocytes and NK cells–An up-date. Dev Comp Immunol 33:310–320.

42. Saravanabalaji S, et al. (2009) Viral load, antibody titers and recombinant openreading frame 2 protein-induced TH1/TH2 cytokines and cellular immune responses inself-limiting and fulminant hepatitis e. Intervirology 52:78–85.

43. Yin X, Ambardekar C, Lu Y, Feng Z (2016) Distinct entry mechanisms for non-enveloped and Quasi-enveloped hepatitis E viruses. J Virol 90:4232–4242.

44. Ndure J, Flanagan KL (2014) Targeting regulatory T cells to improve vaccine immu-nogenicity in early life. Front Microbiol 5:477.

45. Boer MC, Joosten SA, Ottenhoff TH (2015) Regulatory T-cells at the interface betweenhuman host and pathogens in infectious diseases and vaccination. Front Immunol 6:217.

46. Hoeppli RE, Wu D, Cook L, Levings MK (2015) The environment of regulatory T cellbiology: Cytokines, metabolites, and the microbiome. Front Immunol 6:61.

47. Meng XJ, et al. (1998) Genetic and experimental evidence for cross-species infectionby swine hepatitis E virus. J Virol 72:9714–9721.

48. Adam SJ, Counter CM (2008) A method to generate genetically defined tumors inpigs. Methods Enzymol 439:39–51.

49. Stokes CR (2017) The development and role of microbial-host interactions in gutmucosal immune development. J Anim Sci Biotechnol 8:12.

50. Jothikumar N, Cromeans TL, Robertson BH, Meng XJ, Hill VR (2006) A broadly reactiveone-step real-time RT-PCR assay for rapid and sensitive detection of hepatitis E virus.J Virol Methods 131:65–71.

51. Mokhtari C, et al. (2013) Comparison of real-time RT-PCR assays for hepatitis E virusRNA detection. J Clin Virol 58:36–40.

52. Liu S, et al. (1997) Removal of endotoxin from recombinant protein preparations. ClinBiochem 30:455–463.

53. Yuan L, et al. (2008) Virus-specific intestinal IFN-gamma producing T cell responsesinduced by human rotavirus infection and vaccines are correlated with protectionagainst rotavirus diarrhea in gnotobiotic pigs. Vaccine 26:3322–3331.

54. Wen K, et al. (2012) CD4+ CD25− FoxP3+ regulatory cells are the predominant re-sponding regulatory T cells after human rotavirus infection or vaccination ingnotobiotic pigs. Immunology 137:160–171.

Cao et al. PNAS | July 3, 2017 | vol. 114 | no. 27 | 6923

MICRO

BIOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Aug

ust 2

8, 2

020