A human-ACE2 knock-in mouse model for SARS-CoV-2 infection recapitulates respiratory disorders but avoids neurological disease associated with the transgenic K18-hACE2 model

Abstract

Animal models have been instrumental in elucidating the pathogenesis of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and in testing coronavirus disease 2019 (COVID-19) vaccines and therapeutics. Wild-type (WT) mice are not susceptible to many SARS-CoV-2 variants, and therefore, transgenic K18-hACE2 mice have emerged as a standard model system. However, this model is characterized by a severe disease, particularly associated with neuroinfection, which leads to early humane endpoint euthanasia. Here, we established a novel knock-in (KI) mouse model by inserting the original K18-hACE2 transgene into the collagen type I alpha chain (COL1A1) locus using a recombinase-mediated cassette exchange (RMCE) system. Once the Col1a1-K18-hACE2 mouse colony was established, animals were challenged with a B.1 SARS-CoV-2 (D614G) isolate and were monitored for up to 14 days. Col1a1-K18-hACE2 mice exhibited an initial weight loss similar to the K18-hACE2 transgenic model but did not develop evident neurologic clinical signs. The majority of Col1a1-K18-hACE2 mice did not reach the pre-established humane endpoint, showing a progressive weight gain 9 days postinfection (dpi). Importantly, despite this apparent milder pathogenicity of the virus in this mouse model compared to the K18-hACE2 transgenic model, high levels of viral RNA were detected in the lungs, oropharyngeal swab, and nasal turbinates. Moreover, the remaining lesions and inflammation in the lungs were still observed 14 dpi. In contrast, although low-level viral RNA could be detected in a minority of Col1a1-K18-hACE2 animals, no brain lesions were observed at any timepoint. Overall, Col1a1-K18-hACE2 mice constitute a new model for investigating SARS-CoV-2 pathogenesis and treatments, with potential implications for studying long-term COVID-19 sequelae.IMPORTANCEK18-hACE2 mice express high levels of the human protein angiotensin-converting enzyme 2 (ACE2), the receptor for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), and are therefore susceptible to infection by this virus. These animals have been crucial to understanding viral pathogenesis and to testing coronavirus disease 2019 (COVID-19) vaccines and antiviral drugs. However, K18-hACE2 often dies after infection with initial SARS-CoV-2 variants, likely due to a massive brain infection that does not occur in humans. Here, we used a technology known as knock-in (KI) that allows for the targeted insertion of a gene into a mouse, and we have generated a new human ACE2 (hACE2) mouse. We have characterized this new animal model demonstrating that, upon challenge with SARS-CoV-2, the virus replicates in the respiratory tract, damaging lung tissue and causing inflammation. In contrast to K18-hACE2 mice, only limited or no brain infection could be detected in this new model. After 14 days, most animals recovered from the infection, although lung tissue lesions were still observed. This new model could be instrumental for the study of specific disease aspects such as post-COVID-19 condition, sequelae, and susceptibility to reinfection.

Keywords: animal model; lung damage; lung inflammation; neuroinvasion; post-COVID condition.

Fig 1

Expression of hACE2 in the Col1a1-K18-hACE2 KI model. (A) Schematic representation of the insertion strategy. The original K18-hACE2 transgene was inserted into the collagen COL1A1 locus using a recombinase-mediated cassette exchange (RMCE) FLP-FRT system in KH2 cells via blastocyst injection. pPGK-ATG-frt plasmid: vector backbone with ampicillin-resistant gene (Amp), transcription start site (ATG), and Flippase recognition target (Frt). Hygro: hygromicin resistance gene. pCAGGS-FlpE: expression plasmid for FLPe recombinase expression. (B) Relative quantification of hACE2 receptor expression to GAPDH expression in uninfected Col1a1-K18-hACE2 (blue empty dot, n = 3) and K18-hACE2 males (black empty square, n = 3). Delta Ct values are inversely shown to facilitate interpretation. The lower the absolute number, the higher the relative expression. Solid line and bars represent the mean and SEM. (C) Relative comparison of hACE2 receptor expression in Col1a1-K18-hACE2 versus K18-hACE2 mice (n = 3 each, males). The higher receptor expression in Col1a1-K18-hACE2 (negative values in y-axis) is marked by blue bars, and the higher expression in K18-hACE2 (positive values in y-axis) is marked in black bars. (D) Western blot analysis of hACE2, mACE2, and GAPDH in representative tissues. hACE2 signal was obtained using a specific monoclonal antibody (top panels); mACE2 and hACE2 signal were obtained with a cross-reactive polyclonal antibody (middle panel), and anti-GAPDH was analyzed as a reference (bottom panels). Molecular weight markers are shown on the right.

Fig 2

Experimental setting and progression of SARS-CoV-2 infection in Col1a1-K18-hACE2 and K18-hACE2 mouse models. (A) Schematic representation of the experimental setting. Knock-in Col1a1-K18-hACE2 (n = 27) and transgenic K18-hACE2 mice (n = 4) were intranasally challenged with a 1,000 TCID₅₀ dose of a B.1 SARS-CoV-2 isolate. A Col1a1-K18-hACE2-uninfected control group (n = 5) was challenged with PBS. Mice were monitored for weight loss and clinical signs for 14 dpi. Euthanasia was performed at 3, 7, and 14 dpi or upon fulfilment of humane endpoint criteria, for sample and tissue collection (n = 8 per timepoint). Infections were performed in two separate experiments between January and June 2022. Created with Biorender.com. (B) Relative body weight follow-up referred to day 0. Col1a1-K18-hACE2 uninfected (blue empty dot), Col1a1-K18-hACE2 infected (blue dot), and K18-hACE2 infected (black square). Solid lines and bars represent the mean ± SD. (C) Survival (Kaplan-Meier). All K18-hACE2-infected animals (n = 4) had to be euthanized due to endpoint criteria by 7 dpi, and only three were infected with Col1a1-K18-hACE2 at 8, 9, and 10 dpi. No uninfected Col1a1-K18-hACE2 had to be euthanized under this criterion. Col1a1-K18-hACE2 uninfected (blue dashed line), Col1a1-K18-hACE2 infected (blue line), and K18-hACE2 infected (black line). Statistical differences were identified using a log-rank (Mantel-Cox) test (<0.0001), followed by individual comparisons (**P < 0.005, ****P < 0.0001).

Fig 3

Progression of SARS-CoV-2 infection in Col1a1-K18-hACE2 KI mice. Col1a1-K18-hACE2 (blue circles, n = 27) and K18-hACE2+ mice (black squares, n = 4) were inoculated with 1,000 TCID₅₀ of a SARS-CoV-2 B.1 isolate (full shapes) or uninfected (empty shapes of each color, n = 5). Animals were euthanized at 3 dpi (n = 8), 7 dpi (n = 8), 14 dpi (n = 8), or upon fulfilment of humane endpoint criteria (n = 3, euthanized at 8, 9, and 10 dpi). (A) SARS-CoV-2 viral RNA loads (copies per milliliter) of oropharyngeal swab, lung, and nasal turbinate samples. Dashed lines represent the limit of detection, established by 2SD of uninfected animals. Statistical differences were identified using a Peto-Peto left-censored test (*P < 0.05, **P < 0.005). (B) Viral titration of replicative virus (TCID₅₀/mL) in lung samples from B.1-infected mice at different endpoints in Vero E6 cells on day 5 of culture. Titers were compared using an independence asymptotic generalized Pearson chi-squared test for ordinal data (*P < 0.05, **P < 0.005). (C) Lung histopathological scoring of bronchointerstitial pneumonia (left) and SARS-CoV-2 NP immunohistochemical (IHC) scoring (right) in both models at 3, 7, and 14 dpi and endpoint. Statistical differences were identified using an independence asymptotic generalized Pearson chi-squared test for ordinal data (*P < 0.05). (D) Representative lung histology and IHC pictures of both models at 3, 7, and 14 dpi and endpoint. Images show low-power magnification bars (200 µm).

Fig 4

Inflammatory response in the lung. The concentration of inflammatory cytokines in lung extracts in the knock-in and transgenic models at 3, 7, and 14 dpi or endpoint is shown. Col1a1-K18-hACE2-infected (blue full circles, n = 23), uninfected (blue empty circles, n = 3), and K18-hACE2-infected mice (black full squares, n = 4). The bar shows the median with an interquartile range. The limit of detection for each cytokine is indicated by dotted lines. Statistical differences were identified using a Kruskal-Wallis test and a Conover’s nonparametric all-pairs comparison test (*P < 0.05, **P < 0.005).

Fig 5

Tissue tropism of SARS-CoV-2 in Col1a1-K18-hACE2 and K18-hACE2 mice. SARS-CoV-2 viral RNA loads (copies per milliliter) were analyzed by RT-qPCR in Col1a1-K18-hACE2 mice tissues collected at 3, 7, and 14 dpi (n = 3 per timepoint) or upon fulfilment of humane endpoint criteria (n = 1 at 8 dpi, n = 1 at 9 dpi). K18-hACE2 tissues were collected at endpoint at 6 (n = 1) and 7 (n = 1) dpi. Solid lines show the mean ± SD. Dashed lines represent the limit of detection, established by 2SD of uninfected animals.

Fig 6

Progression of SARS-CoV-2 infection in the brain of Col1a1-K18-hACE2 KI mice. Col1a1-K18-hACE2 (blue circles, n = 27) and K18-hACE2+ mice (black squares, n = 4) were inoculated with 1,000 TCID₅₀ of a B.1 SARS-CoV-2 isolate (full shapes) or were uninfected (empty shapes, n = 5) and were monitored until 14 dpi. Samples were collected at 3, 7, and 14 dpi and at endpoint. (A) SARS-CoV-2 viral RNA loads (copies per milliliter) in brain extracts. Dashed lines represent the limit of detection, established by 2SD of uninfected animals. Statistical differences were identified using a Peto-Peto left-censored samples test with correction for multiple comparisons. Solid lines and bars represent the mean and SD. (B) Viral titration of replicative virus (TCID₅₀/mL) in brain samples of B.1-infected mice at different endpoints in Vero E6 cells on day 5 of culture. Solid lines and bars represent the mean and SD. (C) Brain histopathological scoring of multifocal lymphoplasmacytic meningo-encephalitis in the brain (left) and SARS-CoV-2 NP IHC scoring (right) in the brain of both models. Statistical differences were identified using an independence asymptotic generalized Pearson chi-squared test for ordinal data (*P < 0.05, **P < 0.01). (D) Representative images of brain histology and IHC images of both hACE2+ mouse models. Images show low-power magnification (top; bars: 600 µm) and medium-power magnification (bottom; bars: 300 µm). Col1a1-K18-hACE2 KI mice samples shown were collected at 7 dpi, while K18-hACE2 samples were collected at humane endpoint (6–7 dpi).