The key molecular events that trigger progression from self-limited viral illness to severe COVID-19, remain unknown. Non-invasive profiling of cells from human patients and analysis of tissue specimens collected post-mortem has been valuable in characterising inflammatory pathology associated with severe COVID-19 1,2. However, defining the key early events that trigger progression towards severe COVID-19 requires longitudinal studies and invasive sampling of tissues.
Animal models are important tools for testing hypotheses about mechanisms of disease in COVID-19 and complement clinical studies from patients, enabling direct analysis of target tissues following controlled infections 3. However, at present, there is a lack of efficient, tractable model systems that replicate the primary feature of severe COVID-19: pulmonary inflammation. Rodent models such as human angiotensin-converting enzyme 2 (hACE2) transgenic mice and golden Syrian hamsters are susceptible to SARS-CoV-2 infection. Non-transgenic mice can be infected with mouse-adapted strains of SARS-CoV-2. Ferrets are also susceptible to infection and are effective transmission models, while non-human primates (NHPs) are the most comparable model to human infections 3. While each of these models has advantages, there are important drawbacks. Rodent models do not accurately replicate the pattern of disease seen in humans, and tissue morphology and immunological responses to SARS-CoV2 are significantly different from humans 3. There is a lack of molecular tools for ferret models, while non-human primates are expensive and restricted to a few institutions around the world. The lack of large animal models that accurately reflect the pathology associated with severe COVID-19 in humans hampers our ability to understand the mechanisms that drive disease and the development of effective interventions. Furthermore, the development of therapeutic interventions in a model with similar physiology to humans is more likely to successfully translate into effective therapies in humans.
There is an increasing appreciation of livestock as biomedical models, with pigs being one of the most important 4. The short gestation period, large litter size and a suite of appropriate tools and methodologies allow relatively rapid development of transgenic large animal models. 5. Pigs are genetically, anatomically, physiologically and immunologically closer to humans than rodents or ferrets, are relatively inexpensive, accessible and are more ethical acceptable than NHPs 4. However, pigs are not susceptible to SARS-CoV-2 infection which limits their utility as a model 6,7.
To address this, we generated hACE2 transgenic pigs. hACE2 is the cellular receptor for SARS-CoV-2 and the main determinant of species tropism. In brief a custom lentivirus expressing hACE2 under the Keratin 18 promoter was microinjected into oocytes which were then surgically implanted into five surrogate gilts (Fig. 1A). Three were confirmed pregnant and a total of 32 piglets were born. Genomic DNA was generated from ear biopsies and subjected to PCR for the transgene sequence. Twenty-eight piglets were confirmed as transgenic (data not shown). Relative lentivirus copy number was determined using a provirus-specific qPCR on the same genomic DNA with cycle threshold (Ct) values ranging from 20.8 to 30.7 (Supplemental table 1). Based on Ct values, three females and two males were selected for breeding, resulting in generation of an F1 cohort of 30 piglets. Total RNA was extracted from ear biopsies and levels of hACE2 transcription determined by RT-qPCR. Piglets were ranked based on hACE2 transcription levels, and the 9 highest expressing piglets (7 females and 2 males) were selected for challenge with SARS-CoV-2 (Supplemental table 2). Prior to the challenge study, primary fibroblast cells were generated from ear biopsies taken from all nine of the selected pigs and an additional two transgenic pigs that showed low or undetectable levels of transgene expression (P35 and P38). We were unable to recover cells from one of the biopsy samples (P57) due to bacterial contamination. However, analysis of the cells that were established from the other animals showed those expressing higher levels of hACE2 mRNA were more susceptible to SARS-CoV-2 infection in vitro (Supplemental Fig. 1).
To determine in vivo susceptibility, the nine transgenic pigs, and three genetically similar non-transgenic controls were challenged at biosafety level 3 on a single occasion with 1 x 106 TCID50 of an early pandemic isolate of SARS-CoV-2 (EDB2) 8. The inoculum was delivered intranasally in a single 2 ml dose using a mucosal atomiser attached to a syringe. The dose and route of infection were based on previous challenge studies in pigs using SARS-CoV-2 or the porcine adapted coronavirus PRCV 9. Rectal temperature and clinical status of the pigs were monitored twice daily and SARS-CoV-2 lateral flow tests (LFT) performed on nasal swabs collected 2, 4 and 7 days post infection (DPI). All nine transgenic pigs showed clinical signs consistent with mild to moderate SARS-CoV-2 infection, including fever, sneezing, coughing and respiratory distress (Fig. 1B, Supplemental Fig. 2). Furthermore, all transgenic pigs tested positive by LFT of nasal swabs, as early as 2 DPI (Fig. 1C and Supplemental Fig. 3). In contrast, control pigs showed no clinical signs of infection and were negative by LFT. Three transgenic pigs and one wild-type pig were euthanised at 2, 4 and 7 DPI (cohort 1, 2 and 3 respectively) with tissues, including the nasal turbinates, tracheal epithelium and lung, collected for virus detection and histological analysis.
High levels of SARS-CoV-2 RNA were detected in the nasal turbinates and trachea of the transgenic pigs (Fig. 1D). In contrast, viral RNA was only detected in lung samples from three of the transgenic pigs (P53, P60 and P61) and only at very low levels. Infectious virus was recovered from the nasal turbinates with titres peaking at 4 DPI, but was undetectable by 7 DPI (Fig. 1E). No infectious virus was recovered from the trachea or lung samples from any animals. No viral RNA or infectious virus was detected from any tissues taken from the non-transgenic control pigs, consistent with previous reports that non-transgenic pigs are not susceptible to SARS-CoV-2 6,7. Immunohistochemical staining of fixed tissue samples showed extensive expression of hACE2 protein in transgenic pigs (Fig. 1F) with high levels of hACE2 RNA also detected (Supplemental Fig. 4). Focal staining for SARS-CoV-2 in the nasal turbinates, trachea and lung was also detected (Fig. 1F). No hACE2 or SARS-CoV-2 staining was detected in any tissue in the non-transgenic pigs (Supplemental Fig. 5). Crucially, histological examination of tissues revealed clear signs of significant neutrophil and macrophage-rich inflammation within the lungs of infected animals from day four post infection (Fig. 2 and Supplemental table 3) involving both bronchi and alveolar spaces with evidence of diffuse alveolar damage (DAD), oedema, and focal, fibrin-rich intravascular thrombi– all consistent with histological changes observed in fatal COVID-19 1.
The lack of large animal models that faithfully reproduce the pathology of severe COVID-19 has impeded progress in understanding the underlying mechanisms that drive the inflammatory processes causing disease. A previous study reported the generation of hACE2 transgenic pigs by inserting the hACE2 cDNA downstream of the porcine ACE2 promoter 10. Although cells generated from those pigs displayed increased susceptibility to SARS-CoV-2 infection, no challenge studies were reported.
Here, we describe the generation of a transgenic porcine model of COVID-19 that is highly susceptible to infection with SARS-CoV-2 and demonstrates clinical signs and histopathology consistent with moderate to severe disease. Unlike current animal models, infected hACE2 pigs displayed the full range of common clinical signs of COVID-19 including fever, coughing, sneezing, respiratory distress and key pathological signatures in the lung, including DAD, making the hACE2 pigs a unique model for COVID-19.
Substantial inflammation, despite low levels of virus, suggests that pathology in the lungs of infected pigs is driven by a dysfunctional host response, rather than damage caused by virus replication. This is also a key hallmark of severe COVID-19 in patients 1 and further reflects the similarity in the anatomy, physiology and immune responses of humans and pigs.
While infection with SARS-CoV-2 did not result in fatal outcomes, for animal welfare, practical and safety reasons, the study was designed to specifically avoid this outcome. Pigs displaying moderate to severe clinical signs, such as respiratory distress, were included in the next time point for culling, reducing the potential for fatal outcomes.
High levels of infectious virus in the upper airways, along with observed coughing and sneezing, suggests that airborne transmission between transgenic pigs would be likely to occur. Rapid diagnosis with existing LFT tests and onset of clearly observable clinical signs provide a potentially powerful model of airborne transmission. Such studies will be critical for evaluation of vaccines and their ability to block transmission, a goal that is yet to be achieved.
Co-circulation of new variants of SARS-CoV-2 and seasonal Influenza virus A (IAV), as well as the threat of avian IAV, has raised significant concerns on the potential impacts of co-infection on vulnerable individuals and the population as a whole. As pigs are naturally susceptible to IAV, this new model will be hugely valuable for investigating potential consequences of co-infection on disease progression, clinical outcomes, airborne transmission and vaccine and antiviral efficacy. Finally, cross-breeding hACE2 pigs with established porcine biomedical models of underlying co-morbidities, such as obesity and diabetes11 would leverage additional impact, while cross-breeding with Cas9 pigs12 would generate a powerful COVID-19 disease model for in vivo and ex vivo precision gene editing.