In Hong Kong, when a person is diagnosed with COVID-19, they are hospitalized and household contacts regarded as ‘close contacts’ are quarantined in designated centres. Affected pet owners are given the option of having their dogs and cats looked after and isolated by the Hong Kong Agriculture, Fisheries and Conservation Department (AFCD). Specimens are collected from these animals to assess whether they are infected with SARS-CoV-2 and to assist in determining the best methods for managing animals in quarantine, including timing of release back to the owner. Fifteen dogs and seven cats from households with known COVID-19 cases had been quarantined and tested as of 27 March 2020. During this period, two dogs returned virological test results demonstrating that they were infected.


Dog 1 is a 17-year-old neutered male Pomeranian that had a number of pre-existing diseases, including a grade II heart murmur, systemic and pulmonary hypertension, chronic renal disease, hypothyroidism and a previous history of hyperadrenocorticism (F. Chan, personal communication). The owner of dog 1 was a 60-year-old woman who developed symptoms on 12 February 2020 and was diagnosed with COVID-19 on 24 February 2020. A female domestic helper in the household developed a fever on 16 February 2020 and was subsequently confirmed to be infected with SARS-CoV-2 (secondary case A). The remaining three members of the household were sent to a quarantine centre on 26 February 2020, and one of them was confirmed to be infected on 7 March 2020 (secondary case B). Dog 1 was transferred to a holding facility managed by AFCD on 26 February 2020 and nasal, oral and rectal swabs and a faecal sample were collected. Additional specimens for virus detection were collected from the dog on six further occasions. A blood sample was collected on 3 March 2020 for serological testing (see Fig. 1). Throughout the period in quarantine, the dog remained bright and alert with no obvious change in clinical condition.

Fig. 1: Timeline.
figure 1

A timeline of clinical events in the human and dog SARS-CoV-2 infection cases that were analysed in this study.

SARS-CoV-2 RNA was detected from nasal swabs collected from dog 1 by quantitative PCR with reverse transcription (RT–qPCR)4,5 in five consecutive specimens collected on and between 26 February and 9 March 2020 (Table 1). Rectal and faecal specimens tested negative. Attempts to culture the virus from the dog were unsuccessful, probably owing to the low viral load (range 7.5 × 102 to 2.6 × 104 RNA copies per ml of specimen); in human patients with COVID-19, virus isolation had a low probability of success when viral load in the specimen was less than 106 per ml (ref. 6).

Table 1 RT–qPCR testing results on nasal and oral swabs of the dogs and serology

Dog 2 was a 2.5-year-old male German shepherd in good health from a household in which the owner developed symptoms on 10 March 2020 and was diagnosed with COVID-19 on 17 March 2020. Specimens from this dog were collected six times between 18 and 30 March 2020. Oral and nasal swabs tested positive for SARS-CoV-2 RNA on the first two occasions (Table 1). Rectal swabs collected on 18 March 2020 tested positive in four of the six assays, all with higher Ct values (lower viral load) than those obtained from oral and nasal swabs. A second dog kept in the household was sampled on four occasions between 18 and 30 March and tested negative for SARS-CoV-2 RNA in all tests.

Serum samples collected from dog 1 on 3 March 2020, and from dog 2 on 19, 23 and 30 March 2020 were tested for SARS-CoV-2 antibody using 90% plaque-reduction neutralization tests (PRNT90)7. Serum from dog 1 had a PRNT90 titre of 1:80; serum from dog 2 had PRNT90 titres of 1:10 (19 March), 1:40 (23 March) and 1:160 (30 March). The second dog in the household of dog 2 remained antibody-negative on 30 March 2020. Twenty control dog sera tested negative for PRNT90-neutralizing antibody.

Viral RNA from the nasal swab specimen collected from dog 1 on 26 and 28 February 2020 was sequenced directly from the clinical specimen and compared with the virus found in clinical specimens from the owner and secondary cases A and B. The full virus genome sequence (29,764 nucleotides) was obtained from the index case and from secondary cases A and B. Viral sequences of length 27,871 nucleotides (nt) (94% of the genome) and 26,025 nt (93% of the genome) were obtained from the nasal swabs of dog 1 collected on 26 and 28 February 2020, respectively. The viral sequences from the index case and two secondary cases were identical across the full genome. Viral RNA from the nasal swabs of dog 2 collected on 18 and 19 March 2020 and the human index case from the same household were sequenced and found to be identical across the full genome (29,764 nucleotides). The viruses from the two households, however, were clearly distinguishable (Fig. 2).

Fig. 2: A phylogenetic tree of SARS-CoV-2 showing viruses from infected dogs and humans in Hong Kong.
figure 2

Virus sequences from humans and dogs from the two affected households are shown in red. Other virus sequences from human patients in Hong Kong are shown in blue. Selected full and partial (longer that 23,000 nt) virus genomes from the GISAID database are included in this analysis. The tree is unrooted and was constructed using the maximum-likelihood method with PhyML.


Our results demonstrate infection of two dogs by SARS-CoV-2. Angiotensin-converting enzyme 2 (ACE2) is known to be the human receptor for SARS-CoV-2, and canine ACE2 is similar to that of humans (Extended Data Fig. 1). Of the 18 amino acids known to be involved in the interaction between ACE2 and the spike receptor binding domain (RBD) of SARS-CoV-2, five differ between humans and dogs, but none of these are in regions known to disrupt the interaction between the RBD of SARS-CoV and ACE28.

Our evidence suggests that human-to-animal transmission of SARS-CoV-2 is possible. We do not have information on whether the virus can cause illness in dogs, but there were no specific symptoms in either of the infected dogs while they were shedding virus. The Pomeranian died two days after release from isolation, probably owing to the pre-existing underlying diseases; we were unable to perform a post mortem examination. Whether infected dogs could transmit the virus to other animals or back to humans remains unknown. The owner of dog 2 had a second, crossbreed dog in which neither viral RNA nor antibody responses were detected, suggesting that transmission had not occurred between the two dogs sharing the household.

These two cases in Hong Kong demonstrate that dogs can acquire infection in households with SARS-CoV-2-infected humans. A survey of 4,000 specimens from dogs, cats and horses from places where community transmission of SARS-CoV-2 was occurring in humans did not detect any positive results, suggesting that the virus is not widely circulating in pet animals9. Unlike our study, this previous study did not specifically investigate dogs from households of patients with COVID-19. A challenge study in five six-week-old beagles demonstrated seroconversion in two dogs and detection of viral RNA (up to 106.5 copies) in rectal swabs two days after challenge, and one dog had viral RNA in a rectal swab six days after challenge. No virus was detected in oropharyngeal swabs, but nasal swabs were not collected10. Our results suggest higher viral load and increased duration of viral shedding in nasal swabs compared with oral swabs. The experimental challenge study reported that cats had large quantities of virus in nasal mucosa and other tissues, and that they shed sufficient virus to allow cat-to-cat transmission10. A cat that was in contact with a human patient with COVID-19 tested positive for SARS-CoV-2 in Belgium11. SARS-CoV-2 RNA was detected in a cat in Hong Kong after the cut-off date for the present study; the cat was from a household with a confirmed case of COVID-19.

These findings and the results from animal testing during the SARS outbreak in 20033 have potential implications for the management of mammalian pets owned by people who develop SARS-CoV-2 infection. There is no evidence that domestic animals had any role in onward transmission of the SARS outbreak3. However, from a precautionary point of view, pets belonging to patients with COVID-19 could be isolated and tested for SARS-CoV-2, as is being done in Hong Kong.

The findings also have implications for future zoonotic transmission events by the precursor virus of SARS-CoV-2. Rhinolophid bats are considered a probable reservoir of the precursor of SARS-CoV-212. However, on the basis of experiences with SARS virus, intermediate hosts probably serve to bridge transmission from bats to humans. Dogs, other canids and felids can be sold in or present in the vicinity of wild-game animal markets, the presumed source for the initial zoonotic spillover of SARS-CoV-2. Studies into the origin of SARS-CoV-2 should investigate these species to determine whether they have any role in spillover events.


Data reporting

No statistical methods were used to predetermine sample size. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment.

Specimen collection

Specimens from dogs and cats were collected by veterinarians from animals sent to the AFCD isolation centre and included deep oropharyngeal and nasal swabs and a sample of fresh faeces and/or a rectal swab, placed in virus transport medium and kept on cool-packs until arrival in the laboratory. Virus transport medium comprised Medium 199 (Sigma M0393) as basal medium, 0.5% bovine serum albumin, antibiotics (penicillin g, streptomycin sulfate, polymyxin B sulfate, sulfamethoxazole, nystatin, gentamicin sulfate, ofloxacin). Specimens were collected on at least 3 occasions (on arrival in the isolation centre and in the two days before release). Any animal that had a positive test was retested until no positive results were obtained. Owners provided written consent at the time their pets were moved to isolation to allow specimens to be collected and tested.

Control specimens including nasal, oral, rectal swabs and faeces were collected from 21 stray dogs soon after euthanasia. Stored residual sera from 20 dogs collected for diagnostic purposes from veterinary clinics during 2017–2018 were used as controls for serology.

Specimens from humans were collected and tested by RT–qPCR as part of routine clinical care and the viruses genetically sequenced as part of the routine public health response (Institutional Review Board approval UW20-168).

Quantitative RT–PCR

At the AFCD laboratory, RNA from 200 μl specimen in virus transport medium was extracted using NucliSENS easyMag extraction kit (BioMerieux) following instructions provided by the manufacturer and eluted into 60 μl. The RNA was tested for SARS-CoV-2 RNA in a commercial assay RT–qPCR assay for the E and RdRp gene sequences (TIB Molbiol Lightmix Modular Assays) based on published RT–qPCR assay for SARS-CoV-25. Positive, negative and inhibitor controls were included in each RT–qPCR run and work-flow precautions were in place to minimise PCR contamination. Positive samples were sent to the HKU as an independent reference laboratory for confirmation.

Viral RNA from the original swabs referred by the AFCD laboratory were independently extracted at the HKU using the QIAamp viral RNA minikit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. Swab supernatant (160 μl) was used for RNA extraction with the final elution volume being 60 μl. One-step RT–qPCR assays were run for previously published nsp14 and N genes, which detect SARS-CoV, SARS-CoV-2 and bat SARS-CoV4. In addition, RT–qPCR assays for nsp16 and M that are specific for SARS-CoV-2 with no cross-reaction with SARS-CoV were also used. The forward primer (5′-GGWCAAATCAATGATATGATTTT), reverse prime (5′-GTTGTTAACAAGAACATCACTAGA) and probe (5′-FAM-AAGTCTRCCTTTACTAAGAAGAGA-TAMRA-3′) were used for the ORF1b-nsp16 assay and forward primer (5′-GGYTCTAARTCACCCATTCA-3′), reverse prime (5′-TGATACTCTARAAAGTCTTCATA-3′) and probe (5′-FAM-AATTTAGGTTCCTGGCAATTAATT-TAMRA-3′) were used for the M gene assay. The thermal cycling conditions were identical to those published for the nsp14 and N gene assays4. Positive, negative and inhibitor controls were included in each RT–qPCR run and work-flow precautions were in place to minimise PCR contamination13.

Nasal, oral, rectal swabs and faecal samples from 21 control dogs were run by all six RT–qPCR assays with negative results. No evidence of PCR inhibition was seen in any of these RNA extracts.

Sequencing the viral genomes

To amplify the virus genome, reverse transcription reactions were set up using superscript IV reverse transcriptase (Thermo Fisher Scientific) with multiple gene specific primers targeting different regions of the viral genome (Supplementary Table). The synthesized cDNA was then subjected to multiple overlapping PCRs using Platimum Taq DNA polymerase (Thermo Fisher Scientific) using the protocol provided by the manufacturer. The PCRs performed were in sizes of around 2,000 bp designed to cover the whole virus genome. PCR amplicons were visualized by agarose gel electrophoresis. Nested PCRs were performed when necessary for genome amplification. Aliquots of 5 μl PCR products and DNA ladder were loaded into wells in 2% agarose gel. Electrophoresis was run at 120 V for 20 min in TAE buffer and the DNA band was visualized with SYBR safe DNA gel stain.

PCR amplicons obtained from the same specimens were pooled and sequenced using MiSeq sequencing platform (Illumina). Sequencing library was prepared by Nextera XT DNA library prep Kit (Illumina) following standard protocols. Generated sequencing reads were mapped to a reference virus genome by BWA14 and genome consensus was generated by Geneious version 11.1.4 ( with a minimal coverage depth of 20. Percentage of nucleotides at each position of the genome was calculated by bam-readcount ( with minimal base quality score of 20 and minimum mapping quality score of 20.

Plaque reduction neutralization tests

BetaCoV/Hong Kong/VM20001061/2020 isolated from the nasopharynx aspirate and throat swab of a COVID-19 patient in Hong Kong was grown in Vero E6 cells (ATCC CRL-1586). Cells were regularly tested to exclude mycoplasma contamination. Stock virus was prepared and aliquoted and stored at −80 °C until use. The virus stock was titrated in quadruplicate in Vero-E6 cells in 24-well tissue culture plates (TPP Techno Plastic Products) in a biosafety level 3 facility. After one hour incubation in 5% CO2 incubator, the plates were overlaid with 1% agarose in cell culture medium and incubated for 3 days when the plates were fixed and stained and plaque forming units per ml of the virus stock was determined. Serial dilutions of serum samples were then incubated with 30–40 plaque-forming units of virus for 1 h at 37 °C. The virus–serum mixtures were added on to Vero cell monolayers, incubated, overlaid and stained as above. Antibody titres were defined as the highest serum dilution that resulted in >90% (PRNT90) reduction in the number of plaques7.

Virus isolation

Fresh nasal and oral swab fluid collected from SARS-CoV-2 PCR confirmed dogs in viral transport media were used as the inoculum for virus isolation. In brief, Vero E6 (ATCC CRL-1586) cells were cultured for 24 h in a 24-well plate format (TPP Techno Plastic Products) before inoculation. Culture medium was minimal essential medium containing 2% fetal bovine serum, 100 units ml−1 penicillin and 100 μg ml−1 streptomycin. The swab fluids were centrifuged at 5,000 rpm for 10 min at 4 °C in a benchtop centrifuge and the supernatant was separated and inoculated on to Vero E6 cells in alternative wells of the 24-well plate. After two hours incubation for adsorption in a 37 °C incubator containing 5% CO2, fresh virus growth medium was added to a final volume of 1 ml and then incubated in a 37 °C incubator containing 5% CO2 for six days. The presence of cytopathic effect (CPE) was looked for daily. Additionally, the aliquots of culture supernatant samples was collected into AVL buffer at 0 h, 24 h, 48 h and 72 h post inoculation for PCR. The culture medium was replaced as required with fresh culture medium. Cell cultures that were negative for virus growth were blind-passaged again after six days. The cultures that were positive for virus growth as judged by cytopathic effect and increasing viral load by RT–qPCR were collected and passed on to new cull culture wells in 24-well plates and then progressively onto cells in T25 culture flasks (Greiner Bio-one). Mock inoculated Vero E6 cells were used as negative control for each isolation experiment.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.