Excretion of infectious hepatitis E virus into milk in cows imposes high risks of zoonosis

SAMPLE COLLECTION

Fresh stool, serum, and milk samples of Holstein cows were collected from Dali, Yunnan, China, during September to December 2015. The region where samples were collected has traditional mixed farming of different types of domestic animals. Typically, each household owns one to three cows, whereas a few families own larger numbers of cows. The samples were stored at -80°C until use. Details are described in Supporting Table S1.

DETECTION OF HEV RNA

Stool specimens were suspended at 10% w/v in phosphate-buffered saline (pH 7.4), containing 0.01% diethyl pyrocarbonate, and centrifuged at 12,000g for 10 minutes. Total RNA was extracted from the supernatant of stool, milk, or serum using the QIAamp Viral RNA mini-kit (Qiagen, Germany) according to the manufacturer's instructions. Reverse-transcription was performed using a reverse transcriptase kit (AMV XL for real-time polymerase chain reaction [RT-PCR]; Takara, Japan) according to the manufacturer's directions. A 348-nucleotide amplicon from HEV open reading frame 2 (ORF2) was amplified by nested RT-PCR as described (Supporting Table S2).23

GENOME SEQUENCING

Three full-length nucleotide sequences of HEV were obtained by amplification of five overlapping fragments covering the complete genome (Supporting Table S2). The extreme 5′ and 3′ ends of the viral genome were amplified using the rapid amplification of complementary DNA (cDNA) ends technique as in our previous study.24 The complete nucleotide sequence was assembled and analyzed using the MegAlign software. All sequences have been submitted to the GenBank database.

VIRAL TITER QUANTIFICATION BY QUANTITATIVE RT-PCR

The viral titer of HEV in serum, feces, or milk was quantified using SYBR green-based quantitative RT-PCR (qRT-PCR) with HEV-specific primers as described (Supporting Table S2).23 In brief, 200 μL of milk or serum or 0.4 g of fecal samples was subjected to RNA isolation. Isolated RNA was used to synthesize the first-strand cDNA, and cDNA was added as a template for qRT-PCR. qRT-PCR was performed under the following conditions: 95°C for 30 seconds, followed by 39 cycles of 95°C for 5 seconds and 60°C for 31 seconds. It was performed using an ABI PRISM 7300 Real-Time PCR System.

VIRAL TITER CALCULATION

A standard plasmid was constructed by cloning the RT-PCR-amplified ORF2 partial gene (348 bp) into the pMD 18-T clone vector. The copy number of the recombinant plasmid was calculated using the following formula: (DNA concentration [ng/μL] × 10⁻⁹ × 6.0233 × 10²³ copies/mol)/(DNA size [bp] × 660). The standard curve was drawn based on the copy number and the cycle threshold value of qRT-PCR. The accuracy of HEV RNA quantification by qRT-PCR was validated with a 10-fold serial dilution of this standard plasmid. We first performed dilutions from 1 × 10¹ copies/mL to 1 × 10⁷ copies/mL to generate a linear range. However, there was no clear distinction of cycle threshold values between 1 × 10¹ copies/mL and 1 × 10² copies/mL, suggesting that this assay is not sensitive to quantify at a low copy number of 1 × 10¹ copies/mL. Therefore, a linear range was generated from 1 × 10² copies/mL to 1 × 10⁷ copies/mL with R² = 0.9966 (Supporting Fig. S1).

PHYLOGENETIC ANALYSIS

The nucleotide sequences of the amplified PCR products and of prototypes of different genotypes of HEV strains were aligned using MEGA 4.1 software (version 4.10, http://www.megasoftware.net). The reference sequences of prototype HEV strains were obtained from the GenBank database. The standard classification of HEV genotypes and subtypes was according to previous studies.25, 26

Phylogenetic trees were generated by the minimum evolution and interior branch methods. Bootstrapping with 1,000 resamplings of the data was performed to calculate branch percentages. The identity between nucleotide sequences was calculated using the MegAlign program (DNAstar package, version 5.03).

HEV INFECTIVITY IN RHESUS MACAQUES

Healthy male rhesus macaques (n = 7), 2-3 years old, negative for HEV RNA and anti-HEV immunoglobulin G (IgG) and immunoglobulin M (IgM) antibodies, were used. The protocol of animal experimentation was approved by the committee of Laboratory Animal Welfare and Ethics of the Kunming University of Science and Technology. Monkeys were numbered randomly and housed in individual cages. HEV-contaminated raw or pasteurized milk and a swine HEV strain (KM01, 2 × 10⁴ copies/mL) that we have demonstrated has robust infectivity in rhesus macaques27 were used. The first monkey was inoculated with gavage of 5 mL supernatant of KM01 (viral titer 2 × 10⁴) pretreated at 62°C for 30 minutes. The second monkey was inoculated with gavage of 5 mL supernatant of KM01 pretreated at 72°C for 30 seconds. The third monkey was inoculated with gavage of 5 mL supernatant of KM01 pretreated at 100°C for 3 minutes. The fourth monkey was gavaged with 5 mL of pasteurized (the exact pasteurization condition was unclear) commercial milk positive for HEV RNA (shelf life 24 hours, viral titer 3 × 10⁴ copies/mL). The fifth monkey was gavaged with 5 mL of raw milk from the cow that was HEV-positive in both feces and milk (viral titer in milk 2 × 10⁴ copies/mL). The sixth monkey was gavaged with 5 mL of phosphate-buffered saline. The seventh monkey was gavaged with 5 mL of milk negative for HEV RNA. Heating treatment of the indicated samples was performed using a PCR machine. The effectiveness of our pasteurization protocols was validated on Escherichia coli (Supporting Fig. S2). Fecal samples were collected from each monkey twice per week for HEV RNA detection/quantification. Sera were collected every week for anti-HEV IgG and IgM testing, HEV RNA quantification, or liver enzymes analysis. All samples were tested immediately or stored at -80°C until use.

DETECTION OF ANTI-HEV IgG AND IgM ANTIBODIES

Serum samples were tested for the presence of anti-HEV IgG and IgM antibodies using commercial enzyme-linked immunosorbent assay kits (KHB, China) containing recombinant ORF2 peptides from the HEV genome as well as both positive and negative controls. Samples were tested in duplicate according to the manufacturer's instructions, with cutoff values for IgG and IgM assays set at 0.22 and 0.26, respectively, which were determined based on the mean optical density 450 values from the negative controls (± standard deviation).

SERUM LIVER CHEMISTRY PROFILE

The activities of alanine aminotransferase and aspartate aminotransferase in serum were measured with an automated biochemistry analyzer (Olympus 2700, Japan).

HIGH PREVALENCE OF ACTIVE HEV INFECTION IN COWS

To investigate the prevalence of active HEV infection in cows, we collected stool samples from 140 individual Holstein cows from Dali, Yunnan Province, southwest China, from September to December 2015. Livestock are important components of the agricultural economy in that region, and many of the households typically own one to three cows, as well as other types of domestic animals (Supporting Table S1). The presence of HEV genomic RNA (a hallmark of active infection) in stool samples was examined by nested RT-PCR as described.23 We found that 52 out of 140 samples were positive (37.14%), in line with previous reports of HEV RNA positivity in cows (8.79%, 8/91) on dairy farms in Xinjiang Province (northwest China)28 and in swine (7.8%, 20/256) in Kunming, the capital of Yunnan Province.29

Complete genomic nucleotides of three strains of cow HEV were sequenced by amplification of five overlapping fragments covering the complete genome from fecal samples (Supporting Table S2). The extreme 5′ and 3′ ends of the viral genome were amplified using the rapid amplification of cDNA ends technique, similar to our previous study.24 The complete genomic sequence was assembled and analyzed using the MegAlign software. Sequences were submitted to the GenBank database (KU356187, KU356188, and KU356189 for HEV isolated from fecal samples of cows and KU974927, KU974928, KU974934, KU974936, KU974938, KU974946, KU974947, KU974948, KU974949, KU974950, and KU974952 for partial HEV ORF2 sequences isolated from raw milk in Dali, Yunnan Province, China) (Supporting Table S3).

EXCRETION OF HEV INTO COW MILK

A recent case report showed the presence of HEV in breast milk of a woman with acute hepatitis E.30 We thus investigated whether HEV is excreted in the milk of infected cows. Paired serum, feces, and milk samples were collected from 6 out of the 52 HEV-infected cows that we have identified. As shown by qRT-PCR analysis, all samples were positive, and the titers in milk were considerably high (Fig. 1A; Supporting Table S4).

We further extended the collection of milk samples from all of the identified HEV-positive cows. The presence of HEV RNA (100%, 52/52) was first confirmed by RT-PCR, and the genomic copy number was subsequently quantified by qRT-PCR (Fig. 1B; Supporting Table S4). With respect to the milk samples derived from cows for which the complete HEV genome was sequenced, the amplified ORF1 and ORF2 (partial) sequences were also determined. As expected, these ORF1 and ORF2 sequences were perfectly matched with their parental genome (Supporting Fig. S3), providing further confirmation that HEV in milk was excreted by the infected cows.

HIGH HOMOLOGY OF COW/MILK HEV TO HUMAN AND SWINE STRAINS

To find the homology relationship, phylogenetic analysis was performed based on the complete genomic sequences of our identified cow HEV strains and ORF2 (partial) sequences from milk. A phylogenetic tree clearly illustrated that all HEV isolates from cow/milk belong to genotype 4 (Fig. 2A), subtype 4h (Fig. 2B). More importantly, these HEV strains were closely clustered to human and swine HEV isolated in Kunming City, the capital of Yunnan Province.

Homology analysis based on the complete sequence of HEV indicated that these cow HEVs shared 99.2%-99.4% similarity to human HEV isolated from a pregnant woman in Kunming City, in early 2015,31 and shared 99.5%-99.8% identity with swine HEV isolated in Kunming City in 2010.32 These results strongly suggest that these HEV isolates prevalent in human, swine, and cows in Yunnan Province probably originated from the same source.

HEV FROM RAW OR PASTEURIZED MILK IS INFECTIOUS IN RHESUS MACAQUES

Cow milk produced in this area is mainly for household consumption and local market supplies. To address the key issue of whether HEV-contaminated milk can be a mediator of zoonosis, we next evaluated whether it is infectious in rhesus macaques. We employed this sophisticated animal model for HEV infection that had been previously used by our group27 and others.33, 34 We selected a representative milk sample with intermediate HEV viral titers (approximately 2 × 10⁴ copies/mL of genomic RNA) (Fig. 1B).

Experimental rhesus macaques were prescreened and confirmed as being HEV-negative. Oral gavage of raw milk containing HEV resulted in active infection in a monkey showing an elevated viral titer from 7-10 days postinoculation in feces (Fig. 3A) and a high titer at 4 weeks postinoculation in serum (Fig. 3B). In contrast, HEV was not detectable in the monkey gavaged with HEV-negative raw milk.

Fresh pasteurized milk is readily available at local markets supplied by the farmers. As expected, we identified HEV-contaminated packages but were curious to know whether they remained infectious. Surprisingly, oral gavage of pasteurized milk in a rhesus macaque resulted in a substantially high level of HEV viral load (viral genomic RNA approximately 1.1 × 10⁵ copies/mL) in serum at 4 weeks postinoculation (Fig. 3B), although viral RNA was hardly quantifiable in feces (Fig. 3A). Thus, pasteurization seems insufficient to completely inactivate HEV.

COMPLETE INACTIVATION OF HEV BY BOILING

To more clearly demonstrate the effects of pasteurization on HEV infectivity, we mimicked the commonly used protocols of pasteurizing raw milk by preheating at 62°C for 30 minutes or at 72°C for 30 seconds using a PCR machine. We used a swine genotype 4 strain that we have previously demonstrated to produce robust infection in monkeys.27 We found that pretreatment with our pasteurization protocols is not sufficient to completely inactivate the HEV infectivity in monkeys as shown by quantified viral load in both feces and serum (Fig. 4).

We next explored whether boiling of HEV at 100°C for 3 minutes is sufficient for complete inactivation. Indeed, similar to the phosphate-buffered saline-treated negative control monkey, HEV viral RNA was not detectable in both feces and serum of the monkey inoculated with the boiled HEV sample (Fig. 4). Thus, these results alert us to the potential zoonotic risk of both raw and pasteurized cow milk. Fortunately, a short period of boiling is already sufficient to completely inactivate the virus, providing a cost-effective approach that can be easily implemented even in resource-limited regions.