TYPE Original Research PUBLISHED 15 November 2022 DOI 10.3389/fmicb.2022.1016201 Agreement and differential use OPEN ACCESS of laboratory methods for the detection and quantification of EDITED BY Tiong Gim Aw, Tulane University, United States REVIEWED BY Connor Bamford, SARS-CoV-2 in experimentally infected animals Queen's University Belfast, United Kingdom Akatsuki Saito, University of Miyazaki, Japan *CORRESPONDENCE Joaquim Segalés Carla Usai 1,2‡, Lola Pailler-García 1,2‡, Cristina Lorca-Oró 1,2, joaquim.segales@irta.cat Leira Fernández-Bastit 1,2, Núria Roca 1,2, Marco Brustolin 1,2†, PRESENT ADDRESS † Marco Brustolin, Jordi Rodon 1,2, Mónica Pérez 1,2, Guillermo Cantero1,2, Unit of Entomology, Department of Jorge Carrillo 3,4,5, Nuria Izquierdo-Useros 3,4,5, Julià Biomedical Sciences, Institute of Tropical Medicine, Antwerp, Belgium Blanco 3,4,5,6, Bonaventura Clotet 3,6, Sebastián Napp 1,2, ‡ These authors have contributed equally to Joaquim Segalés 1,7*§ and Júlia Vergara-Alert 1,2§ this work and share first authorship 1 Unitat Mixta d'Investigació IRTA-UAB en Sanitat Animal, Centre de Recerca en Sanitat Animal § These authors have contributed equally to (CReSA), Campus de la Universitat Autònoma de Barcelona (UAB), Bellaterra, Catalonia, Spain, 2 IRTA this work and share last authorship Programa de Sanitat Animal, Centre de Recerca en Sanitat Animal (CReSA), Campus de la UAB, SPECIALTY SECTION Bellaterra, Catalonia, Spain, 3 IrsiCaixa AIDS Research Institute, Badalona, Spain, 4 Germans Trias This article was submitted i Pujol Research Institute (IGTP), Badalona, Spain, 5 CIBERINFEC, ISCIII, Madrid, Spain, 6 Infectious to Virology, Diseases and Immunity, Faculty of Medicine, University of Vic-Central University of Catalonia (UVic- a section of the journal UCC), Barcelona, Spain, 7 Department de Sanitat i Anatomia Animals, Facultat de Veterinària, Frontiers in Microbiology Campus de la UAB, Bellaterra, Catalonia, Spain RECEIVED 11August 2022 ACCEPTED 10 October 2022 Rodents are widely used for the development of COVID-19-like animal PUBLISHED 15 November 2022 models, the virological outcome being determined through several laboratory CITATION Usai C, Pailler-García L, Lorca-Oró C, methods reported in the literature. Our objective was to assess the agreement Fernández-Bastit L, Roca N, Brustolin M, between methods performed on different sample types from 342 rodents Rodon J, Pérez M, Cantero G, Carrillo J, experimentally infected with SARS-CoV-2 (289 golden Syrian hamsters and Izquierdo-Useros N, Blanco J, Clotet B, Napp S, Segalés J and 53 K18-hACE2 mice). Our results showed moderate agreement between Vergara-Alert J (2022) Agreement and methods detecting active viral replication, and that increasing viral loads differential use of laboratory methods for the detection and quantification of SARS- determined by either RT-qPCR or infectious viral titration corresponded to CoV-2 in experimentally infected animals. increasing immunohistochemical scores. The percentage of agreement Front. Microbiol. 13:1016201. between methods decreased over experimental time points, and we observed doi: 10.3389/fmicb.2022.1016201 poor agreement between RT-qPCR results and viral titration from COPYRIGHT © 2022 Usai, Pailler-García, Lorca-Oró, oropharyngeal swabs. In conclusion, RT-qPCR and viral titration on tissue Fernández-Bastit, Roca, Brustolin, Rodon, homogenates are the most reliable techniques to determine the presence Pérez, Cantero, Carrillo, Izquierdo-Useros, and replication of SARS-CoV-2 in the early and peak phases of infection, and Blanco, Clotet, Napp, Segalés and Vergara- Alert. This is an open-access article immunohistochemistry is valuable to evaluate viral distribution patterns in the distributed under the terms of the Creative infected tissues. Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original KEYWORDS author(s) and the copyright owner(s) are severe acute respiratory syndrome coronavirus 2, RT-qPCR, virus titration, credited and that the original publication in this journal is cited, in accordance with immunohistochemistry, comparison, agreement, tissues, oropharyngeal swab accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Frontiers in Microbiology 01 frontiersin.org Usai et al. 10.3389/fmicb.2022.1016201 Introduction multinucleated syncytia formation and apoptosis both in vitro and in vivo, this technique, although time-consuming, is a Since the identification of the severe acute respiratory valuable tool for the diagnosis of active infections and the syndrome coronavirus 2 (SARS-CoV-2) as the causative agent of quantification of viral load (Fenner et al., 1987; Buchrieser et al., Coronavirus disease 19 (COVID-19), there has been a global effort 2020; Zhu et al., 2020). to develop animal models for the preclinical evaluation of vaccines This work aimed to determine the degree of concordance and therapeutic agents. For emerging infectious diseases with a between the diagnostic methods applied in previous studies and potentially lethal outcome, as for COVID-19, animal models are to identify the most robust tools for the characterisation of animal crucial prior to the development of any medical counteraction, to models of SARS-CoV-2 infection. Here we show a moderate understand the pathogenesis, the molecular interactions between agreement between techniques detecting active viral replication, the infectious agent and the host, and the dynamics of disease partly influenced by the time post-infection and the sample type progression. The characterisation and validation of such models analysed. We also found that increasing viral loads determined by require the study of viral replication, clinical signs, pathological viral titration correlated with IHC scores, confirming the features, immune response, transmission patterns and the effect of relevance of IHC in the evaluation of viral distribution patterns, demographic characteristics using different laboratory methods and in assessing the association between tissue lesions and viral (Muñoz-Fontela et al., 2020, 2022). Often animal species are not presence. Since all experimental studies were performed with the naturally susceptible to human pathogens, and the development same variant, our results are not affected by the variability of humanised or transgenic strains is necessary (Swearengen, 2018). associated with different variants of concern (VOC). In February 2020, the World Health Organization established To our knowledge, this is the first side-to-side comparison of an ad hoc Expert Group focused on COVID-19 disease modelling the performance of such methods under experimental settings, (WHO-COM) to accelerate progress and reduce duplication of and it has the potential to contribute to accelerating the progress effort. Amongst the small animal models, golden Syrian hamsters in the field as well as suggesting a more efficient allocation of (GSH), as well as mice expressing the human Angiotensin- laboratory resources. Converting Enzyme 2 (hACE2, the main receptor used by the virus to bind and enter cells) under the control of the human cytokeratin 18 promoter (K18-hACE2 mice), have proved Materials and methods particularly relevant, despite presenting some significant differences with the human disease. In particular, the hamster Sample selection model shows extensive apoptosis in the lungs which is not reported in human pulmonary lesions, whilst the K18-hACE2 This integrated analysis includes data from GSH and mice model is characterised by lethal brain lesions (Chan et al., K18-hACE2 mice experimentally inoculated with SARS-CoV-2. 2020; Moreau et al., 2020; Vidal et al., 2022). RT-qPCR and viral titration were performed from Here we report a pooled analysis of data produced by our oropharyngeal swabs (OS), nasal turbinate (NT), and lung (L) group from a total of 342 animals (including GSH and tissues from both species, and brain (B) from mice. K18-hACE2 mice) inoculated with the D614G variant, Nucleoprotein (NP)-specific IHC was performed on all tissue characterised by an aspartic acid to glycine shift at the amino samples. The samples were generated in eight different studies, acid position 614 of the Spike protein, and dominant worldwide six performed in GSH and two in K18-hACE2 mice. All animals during the early phase of the pandemic (Korber et al., 2020). The (n = 342) were intranasally inoculated with 103, 104, or 105.8 methods considered in the present study are routinely applied TCID50/ml of the hCoV-19/Spain/CT-2020030095/2020 for laboratory diagnosis of viral diseases. Quantitative real-time (GISAID ID EPI_ISL_510689) isolate (D614G variant), each polymerase chain reaction (qPCR) is often considered the gold animal receiving a total volume of 100 μl (50 μl/nostril). standard for pathogen detection; several PCR protocols have Depending on the study, some animals received treatments such been validated and widely used during the first 2 years of the as prophylactic vaccination or post-inoculation antiviral SARS-CoV-2 pandemic to confirm positive results of point-of- treatments. Samples were collected 2-, 4-, and 6 or 7-days post- care tests in patients (Liu et al., 2020; Lieberman et al., 2020; inoculation (dpi) as a final point. Table 1 summarises the Zhen et al., 2020). Moreover, immunohistochemistry (IHC) distribution of sex and treatment status of the animals from relies on specific monoclonal or polyclonal antibodies to which samples were collected at each time point. Animal determine the presence of a specific microorganism in tissue experiments were approved by the Institutional Animal Welfare samples (Rockx et al., 2020; Oumarou Hama et al., 2022). On the Committee of the Institut de Recerca i Tecnologia other hand, the observation of virus-induced cytopathic effect Agroalimentàries and by the Ethical Commission of Animal (CPE) in permissive cell lines is one of the gold standards for Experimentation of the Autonomous Government of Catalonia titrating, isolating and identifying viruses, and it allows the and conducted by certified staff. Experiments with SARS-CoV-2 quantification of infectious viral particles (viral titration, VT). were performed at the Biosafety Level-3 facilities of the Since SARS-CoV-2 is known to cause CPE characterised by Biocontainment Unit of IRTA-CReSA (Barcelona, Spain). Frontiers in Microbiology 02 frontiersin.org Usai et al. 10.3389/fmicb.2022.1016201 TABLE 1 Distribution of animals included in the study per sex, time of sacrifice, and treatment status. 2 dpi 4 dpi 6–7 dpi Total Treatment Males Females Males Females Males Females Males Females Hamsters Infection 24 27 25 30 30 33 79 90 Vaccine candidates 1 dose 9 8 14 15 14 15 37 38 Vaccine candidates 2 doses 8 7 7 8 8 7 23 22 Total N = 289 41 42 46 53 52 55 139 150 Mice Infection 3 3 8 4 9 10 20 17 Antiviral 0 0 0 8 0 8 0 16 Total N = 53 3 3 8 12 9 18 20 33 Total N = 342 159 183 dpi, day post inoculation. Laboratory detection of SARS-CoV-2 environment (RStudio Team, 2022), with stats, DescTools, epiR, ggplot2, and tidyverse packages. RT-qPCR, IHC and viral titration procedures are described elsewhere (Brustolin et al., 2021; Vidal et al., 2022). Briefly, RT-qPCR was performed to detect viral genomic RNA (gRNA) and subgenomic Dichotomous comparison RNA (sgRNA) in two separate reactions, using in both cases Envelope (E)-specific primers and probes targeting regions upstream of the E The numerical results were converted into dichotomous gene (UpE; Corman et al., 2020; Wölfel et al., 2020). IHC was variables (positive/negative) following the criteria described performed on formalin-fixed tissue slides using a Nucleoprotein above. To determine the agreement between the dichotomised (NP)-specific monoclonal primary antibody (Sino Biological, ref. results of different methods, contingency tables were created, 40143-R019; Buchrieser et al., 2020; Zhu et al., 2020). Viral titration and Cohen’s kappa coefficient was computed considering all by CPE assay was performed in vitro on Vero-E6 cells (ATCC possible pairwise comparisons (Watson and Petrie, 2010; CRL-1587) and the median tissue culture infectious dose (TCID50/ml) Kwiecien et al., 2011). was calculated with the Reed-Muench method (On the Calculation of TCID50 for Quantitation of Virus Infectivity | SpringerLink, 2022). Assessment of the correlation between positive results Definition of variables and interpretation of the results In order to evaluate the association between the diagnostic test’s measures on a continuous scale (i.e., gRNA RT-qPCR, The results were expressed as cycle threshold (Ct) values for sgRNA RT-PqPCR and VT), the correlation between tests was both gRNA and sgRNA RT-qPCR, TCID50/ml for the CPE assay, assessed. This allows us to measure the strength and direction and a semi-quantitative score ranging from 0 to 3 for IHC. Samples of the association between the positive results of those with undetermined Ct values or Ct > 36.8 were considered RT-qPCR methods. Since gRNA RT-qPCR, sgRNA RT-qPCR and VT are negative. Samples below the limit of quantification (LOQ, <101.8 measured on different scales, the results had to be standardised TCID50/ml) of the CPE assay were considered negative, as well as to be comparable, for which the mean was substracted and the those with no CPE in any of the wells. Increasing IHC scores result was divided by the standard deviation. The Spearman’s indicate increasing amounts of detectable viral antigen (0: absence correlation test was then performed, because of the lack of of viral antigen; 1: low amount, multifocal localisation; 2: moderate normality of the data. Also, as recommended by Watson and amount, multifocal localisation; 3: high amount, diffuse Petrie (2010) the Lin’s concordance correlation coefficient localisation); score assignment was performed in blind. (CCC) was calculated. Comparison between different technique Distribution of positive results according results to their IHC score All statistical analyses were performed using R programming The distribution of standardised positive results of RT-qPCR language (version 4.2.0, R Core Team, 2022) and the RStudio and VT was compared across positive IHC semi-quantitative Frontiers in Microbiology 03 frontiersin.org Usai et al. 10.3389/fmicb.2022.1016201 scores. Kruskal–Wallis test and Dunn’s test for multiple TABLE 2 Frequency of the dichotomised results (positive/negative) by laboratory methods. comparisons with Bonferroni correction were performed. gRNA sgRNA VT IHC Total RT-CR RT-PCR Relationship of experimental results with Negative 86 304 439 123 952 other covariates Positive 901 513 456 329 2,199 Total 987 817 895 452 Dichotomous comparison RT-qPCR, reverse transcription followed by quantitative polymerase chain reaction; The numerical results were converted into dichotomous gRNA, genomic RNA; sgRNA, subgenomic RNA; VT, infectious viral titration; IHC, variables (positive/negative) as described above. For each species, immune histochemistry. sex, time-point (days post-inoculation), treatment, and tissue type, a Chi-squared test with Bonferroni correction was performed to compare the percentage of agreement between TABLE 3 Qualitative assessment of the techniques used in the methods (pairwise). This analysis was performed by categorising experimental studies calculated on (A) the entire dataset or (B) the results into “Agree” or “Disagree.” selecting the samples analysed by all methods. (A) Distribution of IHC scores according to each covariate Techniques N Proportion of Kappa 95% CI For each species, sex, time-point, treatment status and tissue agreement type, a Chi-squared test with Bonferroni correction was gRNA vs sgRNA 816 71.94% 0.29 [0.24, 0.34] performed to compare the percentage of samples assigned to each gRNA vs VT 894 59.17% 0.17 [0.13, 0.21] IHC score. The same analysis was performed on the entire dataset gRNA vs IHC 445 83.60% 0.49 [0.40, 0.57] and the hamster and mice datasets separately. sgRNA vs VT 739 78.08% 0.56 [0.49, 0.63] sgRNA vs IHC 440 78.18% 0.52 [0.43, 0.61] VT vs IHC 424 79.01% 0.52 [0.43, 0.61] Results (B) Techniques N Proportion of Kappa 95% CI Composition of the pooled data agreement gRNA vs sgRNA 424 71.46% 0.30 [0.24, 0.37] Results obtained with each method were dichotomised into gRNA vs VT 424 73.11% 0.32 [0.25, 0.39] positive and negative as shown in Table 2. gRNA RT-qPCR data were gRNA vs IHC 424 84.20% 0.47 [0.39, 0.56] available for all samples; other methods were performed only on a sgRNA vs VT 424 86.08% 0.70 [0.61, 0.80] fraction of the samples, depending on the objective of each study. sgRNA vs IHC 424 78.30% 0.51 [0.42, 0.60] VT vs IHC 424 79.01% 0.52 [0.43, 0.61] Cohen’s kappa coefficient was computed; 0 ≤ kappa ≤ 0.2: slight agreement; Moderate agreement between methods 0.2 < kappa <=0.4: fair agreement; 0.4 < kappa ≤ 0.6: moderate agreement; detecting actively replicating virus 0.6 < kappa ≤ 0.8: substantial agreement; kappa > 0.8 almost perfect agreement. CI, confidence interval; gRNA, genomic RNA; sgRNA, subgenomic RNA; VT, infectious viral titration; IHC, immune histochemistry. A moderate concordance was found between methods indicative of active viral replication (sgRNA RT-qPCR, IHC and VT), whilst Cohen’s kappa coefficient was lower between gRNA and the other methods, reaching the “moderate” range only when Therefore, the concordance correlation coefficient (CCC) for considered in combination with IHC (Table 3A). Reducing the each pair of numerical methods was calculated. Results dataset to the samples that were analysed using all available indicated a positive, highly correlated relationship between methods (Table 3B), the concordance between sgRNA and VT gRNA and sgRNA (CCC = 0.76 with 95% CI = [0.73, 0.78]), a was substantial; similarly, the kappa coefficient between gRNA negative, moderate correlation between gRNA and Viral and VT increased, indicating a fair agreement; all other pairwise titration (CCC = −0.48 with 95% CI = [−0.54, −0.41]), and comparisons maintained the same concordance ranges. negative, moderate correlation between sgRNA and Viral titration (CCC = −0.50 with 95% CI = [−0.58, −0.41]). Negative relationships are expected due to the nature of the Significant correlation between techniques. Similar results were found calculating the standardised results Spearman’s correlation coefficient, where all techniques showed statistically significant correlations; such results Next, we wanted to determine whether the samples with seemed independent of the (positive) IHC scores assigned to positive results by the different methods were correlated. samples (Figure 1). Frontiers in Microbiology 04 frontiersin.org Usai et al. 10.3389/fmicb.2022.1016201 A B C FIGURE 1 Correlation between the standardised results of positive samples by (A) sgRNA and gRNA RT-qPCR, (B) VT and gRNA RT-qPCR, and (C) VT and sgRNA RT-qPCR. Different colours represent semiquantitative IHC scores assigned to positive samples. p-Values lower than 0.05 are considered significant. (A) r=0.93, p-value<0.0001. (B) r=−0.78, p-value<0.0001. (C) r=−0.70, p-value<0.0001. r, Spearman’s correlation coefficient; RT-qPCR, reverse transcription followed by quantitative polymerase chain reaction; gRNA, genomic RNA; sgRNA, subgenomic RNA; IHC, immune histochemistry; VT, infectious viral titration. NA, non-available. Differences in viral load according to the except for sgRNA versus VT. The percentage of agreement IHC score between techniques on samples collected at 6–7 dpi was significantly lower than the percentage of agreement between the We compared the distribution of standardised RT-qPCR and same methods on samples collected at 2 and 4 dpi (Figure 3B). VT results across the different IHC scores. We found statistically Also, the proportion of IHC scores at 6–7 dpi was significantly significant differences in gRNA and sgRNA values between different from the distributions recorded at earlier time points samples with different IHC scores, as well as in the viral load (Figure 3C); a strong increase in the percentage of samples with determined by VT (Figure 2). no or lower amount of viral antigen was observed at the expenses of the proportion of samples with moderate and high amounts. A statistically significant difference in IHC scores was also found Influence of other covariates on the between samples from treated and non-treated animals, and agreement between laboratory methods between GSH and K18-hACE2 mice (treated animals and K18-hACE2 mice had a higher percentage of negative and slightly The results of pairwise comparisons were categorised positive samples). into “Agree” and “Disagree”, and the proportion of agreement between each pair of techniques was compared across sexes, species, sample types, dpi, and treatment status. No differences Independent analyses of hamster and were found across sexes, species, or treatment status mice datasets (Supplementary Table 1). When we considered the sample type as a covariate, the percentage of agreement between the VT and The same analyses were performed separately for hamsters both gRNA and sgRNA results was lower for OS than for other and mice (Supplementary Tables 2, 3; Supplementary Figure 1). sample types (NT, L; Figure 3A). The comparison of the results GSH results (Supplementary Table 2) showed that the proportion obtained at different time points showed that the percentage of of IHC scores by other covariates was consistent with that agreement decreased over time in all pairwise comparisons, observed in the entire dataset, with significant differences between Frontiers in Microbiology 05 frontiersin.org Usai et al. 10.3389/fmicb.2022.1016201 A B C FIGURE 2 Comparison of the distribution of standardised (A) gRNA RT-PCR, (B) sgRNA and (C) VT results by IHC score. Kruskal-Wallis test and Dunn’s test for multiple comparisons with Bonferroni correction; p-values lower than 0.05 are considered significant. RT-qPCR, reverse transcription followed by quantitative polymerase chain reaction; gRNA, genomic RNA; sgRNA, subgenomic RNA; IHC, immune histochemistry; VT, infectious viral titration. 6 and 7 dpi and earlier time points (Supplementary Figure 1A). RT-qPCR detection of gRNA and sgRNA, IHC, and VT by the In the case of results obtained in K18-hACE2-mice observation of CPE in vitro, performed on OS, NT, L, and B (Supplementary Table 3), sgRNA and VT showed a higher Cohen’s samples collected at 2, 4, and 6–7 dpi. kappa coefficient than the entire dataset, indicating substantial gRNA RT-qPCR is used to determine the presence of the virus agreement. We also found that more covariates influenced the through the detection of its genetic material, without any agreement between techniques performed in K18-hACE2 mice; information on its replicative potential. On the other hand, sgRNA in particular, statistically significant differences between sexes (in molecules are produced only during the coronavirus replication the comparison between sgRNA and IHC), time-points (the cycle and have been used as a proxy of infectivity (Wölfel et al., agreement between gRNA and sgRNA, and sgRNA and VT 2020; Bruce et al., 2022). Similarly, IHC detects the presence of increased over time, whilst decreased between the VT and IHC) viral proteins inside the cells, produced during active infection and sample types (B recorded the highest percentage of agreement, (Lieberman et al., 2020). We dichotomised the results into positive OS the lowest) were found. The distribution of IHC scores and negative to determine the agreement between methods by was significantly different in all comparisons performed computing Cohen’s kappa coefficient, which gives a quantitative (Supplementary Figure 1C). assessment of how well two tests agree by comparing the observed rate of agreement with the rate expected by chance (Watson and Petrie, 2010; Kwiecien et al., 2011). Methods determining active Discussion replication showed a higher kappa coefficient between each other than with the detection of gRNA. RT-qPCR and VT results, both The development of animal models has been instrumental in indicating viral load either indirectly or directly, respectively, are the study of SARS-CoV-2 infection and the evaluation of medical expressed in different units and scales; therefore, we standardised interventions. The characterisation and validation of new animal their numerical results to compare the correlation between the models entail the study of multiple parameters, using a variety of positive results obtained for the same samples with the different laboratory methods. Here we report the side-by-side comparison methods. We found a statistically significant correlation between of four different analytical methods for the detection of SARS- the results of all techniques (analysed pairwise). This suggests that CoV-2 in experimentally inoculated GSH and K18-hACE2 mice: RT-qPCR (for both gRNA and sgRNA) and VT have similar Frontiers in Microbiology 06 frontiersin.org Usai et al. 10.3389/fmicb.2022.1016201 A B C FIGURE 3 Influence of (A) sample type and (B) day post-infection on the percentage of agreement between techniques. (C) The proportion of IHC scores is different across time points, species, treatment status and sexes. Chi-squared test followed by Bonferroni correction; p-values lower than 0.05 are considered significant. L, lung; NT, nasal turbinate; OS, oropharyngeal swab; gRNA, genomic RNA; sgRNA, subgenomic RNA; IHC, immune histochemistry; dpi, day post-inoculation; F, females; M, males. abilities to discriminate amongst different levels of positivity. Also, IHC scores, indicating that, as expected, higher viral loads statistically significant differences between groups were found in corresponded to higher amounts of viral proteins in the the distribution of standardised viral loads across the different infected tissues. Frontiers in Microbiology 07 frontiersin.org Usai et al. 10.3389/fmicb.2022.1016201 We next analysed the effect of covariates on the concordance of agreement between techniques, reflecting a high proportion between techniques and found that species, sex, and treatment of infected samples, and usually with moderate to high viral status did not influence the results, suggesting that the methods load. We also found that the percentage of agreement between we used were equally reliable irrespective of the mentioned sgRNA and IHC was lower in male mice than in female mice; covariates. On the contrary, the percentage of agreement was this might be explained by actual differences in the infection influenced by the experimental time point and the sample type. dynamic and lesions between male and female subjects, as The percentage of agreement amongst SARS-CoV-2 detection already shown in natural infection in humans where a gender- methods decreased over time for all pairwise comparisons, being biased risk factor was observed (Dorjee et al., 2020). However, significantly lower at 6–7 dpi than at earlier time points; the only our results might have been skewed by the fact that within our exception being the comparison between sgRNA and VT. mice subgroup, only females were treated with antivirals: a The disagreement recorded between gRNA and the results bigger and more balanced sample is needed for a correct indicative of viral replication is reflected by the lower kappa interpretation. The increase in the percentage of agreement coefficient as previously discussed. In particular, the lower over time in K18-hACE2 mice can also be explained by the percentage of agreement between sgRNA and VT versus IHC different natural history of SARS-CoV-2 infection in this host: might be due to the low amount and multifocal distribution of unlike hamsters, K18-hACE2 mice are not able to clear the antigen at later time points (Supplementary Figure 2). Accordingly, virus within a week, therefore there is still a high proportion the distribution of IHC scores across time points shows that at of positive samples detected 6–7 dpi (Buchrieser et al., 2020). 6–7 dpi the proportion of negative IHC samples increases at the The differences observed between hamsters and mice in our expense of the proportion of samples assigned 2 and 3 scores, analyses stress the importance of choosing the correct consistent with ongoing viral clearance. This result suggests that experimental setting for each study. It is crucial, therefore, to IHC should be preferentially used not as a method to determine have a clear understanding of the differences between the viral infection, but rather as a follow-up technique to monitor experimental animal model and the human disease, and, as antigen production and clearance/retention. Additionally, its more VOCs are characterised, of the natural history of every combination with histology represents a powerful tool to study the specific VOC. association of the viral antigen with inflammation or other According to the obtained results, gRNA RT-qPCR and VT pathological features. on tissue homogenates were the most reliable techniques to Considering the sample type as a covariate, the percentage of determine the presence and the replication of SARS-CoV-2, agreement between VT and both gRNA and sgRNA obtained in respectively, especially in the early and peak phases of the OS was significantly lower than in the tissue homogenates. This infection. sgRNA detection by RT-qPCR has been used as a proxy difference might be explained by the different nature of the for viral replication, but the results reported so far are sample, since the OS collects what is released by the tissue, being, controversial. Some authors suggested that the absence of sgRNA therefore, an indirect approximation of the molecular events could be used as a test to rule-out active infection, but it should taking place inside the tissue. OS were performed instead of nasal not be considered as proof of infectiousness. In fact, others have swabs for technical reasons (nasal cavities in GSH and mice are shown that sgRNA can be detected even after seroconversion very small and therefore of difficult access for sampling), but our when the shedding of the infectious virus has stopped (Wölfel results suggest that OS are not the most adequate sample type for et al., 2020; van Kampen et al., 2021; Bruce et al., 2022). We, the diagnosis of SARS-CoV-2 infection in these animal models. therefore, consider that VT is to be preferred to sgRNA RT-qPCR On the other hand, the percentage of agreement between all whenever the infectious capacity of the samples at any time point pairs of methods was not statistically different between L and NT, is of interest, since in some cases sgRNA positive samples were not reflecting that both tissues have a similar capability of supporting able to induce any CPE on permissive cells. On the other hand, to replication of the SARS-CoV-2 variant used in the studies. maximise the sensitivity of SARS-CoV-2 genetic material Interestingly, we observed different effects of covariates detection, the use of gRNA RT-qPCR is suggested, since it has when analysing GSH and K18-hACE2 mice separately. proven able to detect the highest number of positive samples in K18-hACE2 mice showed a significant effect of sexes and an our studies. We also consider the use of IHC as a valuable tool for increase in the percentage of agreement over time for some the evaluation of viral distribution patterns, also in light of the comparisons. Unlike GSH, K18-hACE2 mice constitutively good agreement with both sgRNA RT-qPCR and VT at the express the hACE2 receptor on virtually every cell type, beginning and peak of the infection. Despite being used mainly as therefore allowing viral entry and potential replication in a qualitative or semi-quantitative assay, IHC has the advantage of tissues not infected in natural hosts for SARS-CoV-2 infection. being easily performed on different tissue portions at the same It was shown that both SARS-CoV and SARS-CoV-2 cause time, allowing a more complete study of viral tropism and lethal infection in K18-hACE2 mice, invading the brain and distribution, which might be not uniform throughout the infected causing acute meningoencephalitis (McCray et al., 2007; Vidal organ. In conclusion, all analytical methods considered in this et al., 2022). For these reasons, we included brain tissue study allowed the detection of SARS-CoV-2 presence and/or samples in this analysis, which showed the highest percentage replication in different samples of these two animal models of Frontiers in Microbiology 08 frontiersin.org Usai et al. 10.3389/fmicb.2022.1016201 COVID-19-like disease; nonetheless, the choice of the laboratory Science and Innovation (grant no. PID2020-117145RB-I00), EU method must be accurately guided case by case, driven by the HORIZON-HLTH-2021-CORONA-01 (grant no. 101046118) and objective of each study. by institutional funding from Grifols, Pharma Mar, HIPRA, Amassence and Palobiofarma. The funders had no role in the design of this study and did not have any role during its execution, Data availability statement analyses, interpretation of the data, or decision to submit results. The original contributions presented in the study are included in the article/Supplementary material, further inquiries can Acknowledgments be directed to the corresponding author. The authors thank the IRTA-CReSA staff for the care and handling of the animals in the BSL3 facility. Ethics statement The animal study was reviewed and approved by IRTA and Conflict of interest Generalitat de Catalunya. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could Author contributions be construed as a potential conflict of interest. CU, LP-G, JS, and JV-A conceived and designed the experiments. CL-O, LF-B, MB, GC, JS, and JV-A performed the Publisher’s note animal procedures. CU, LP-G, CL-O, LF-B, NR, MB, JR, MP, JS, and JV-A performed the analytical experiments. CU, LP-G, JC, All claims expressed in this article are solely those of the NI-U, JB, BC, SN, JS, and JV-A analysed and interpreted the data. authors and do not necessarily represent those of their affiliated CU, LP-G, JS, and JV-A wrote the paper. All authors contributed organizations, or those of the publisher, the editors and the to the article and approved the submitted version. reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Funding The research of the CBIG consortium (constituted by IRTA- Supplementary material CReSA, BSC & IrsiCaixa) is supported by Grifols. This work was also supported by CERCA Programme/Generalitat de Catalunya, The Supplementary material for this article can be found and the crowdfunding initiative #joemcorono (https://www. online at: https://www.frontiersin.org/articles/10.3389/fmicb. yomecorono.com). NI-U is supported by the Spanish Ministry of 2022.1016201/full#supplementary-material References Bruce, E. A., Mills, M. G., Sampoleo, R., Perchetti, G. A., Huang, M. L., Dorjee, K., Kim, H., Bonomo, E., and Dolma, R. (2020). Prevalence and predictors Despres, H. W., et al. (2022). Predicting infectivity: comparing four PCR-based of death and severe disease in patients hospitalized due to COVID-19: a assays to detect culturable SARS-CoV-2 in clinical samples. EMBO Mol. Med. comprehensive systematic review and meta-analysis of 77 studies and 38,000 14:e15290. doi: 10.15252/emmm.202115290 patients. PloS One 15:e0243191. doi: 10.1371/journal.pone.0243191 Brustolin, M., Rodon, J., Rodríguez de la Concepción, M. L., Ávila-Nieto, C., Fenner, F., Bachmann, P. A., Gibbs, E. P. J., Murphy, F. A., Studdert, M. J., and Cantero, G., Pérez, M., et al. (2021). Protection against reinfection with D614- or White, D. O.. (1987). Chapter 13 – Laboratory diagnosis of viral diseases. In: F. G614-SARS-CoV-2 isolates in golden Syrian hamster. Emerg. Microbes. Infect. 10, Fenner, P. A. Bachmann, E. P. J. Gibbs, F. A. Murphy, M. J. Studdert and D. O. 797–809. doi: 10.1080/22221751.2021.1913974 White, editors. Veterinary Virology. Academic Press; 237–264. Available at: https://www.sciencedirect.com/science/article/pii/B9780122530555500177 Buchrieser, J., Dufloo, J., Hubert, M., Monel, B., Planas, D., Rajah, M. M., et al. (Accessed April 5, 2022). (2020). Syncytia formation by SARS-CoV-2-infected cells. EMBO J. 39:e106267. doi: 10.15252/embj.2020106267 Korber, B., Fischer, W. M., Gnanakaran, S., Yoon, H., Theiler, J., Abfalterer, W., et al. (2020). Tracking changes in SARS-CoV-2 spike: evidence that D614G increases Chan, J. F. W., Zhang, A. J., Yuan, S., Poon, V. K. M., Chan, C. C. S., Lee, A. C. Y., infectivity of the COVID-19 virus. Cell 182, 812–827.e19. doi: 10.1016/j. et al. (2020). Simulation of the clinical and pathological manifestations of cell.2020.06.043 coronavirus disease 2019 (COVID-19) in a golden Syrian hamster model: implications for disease pathogenesis and transmissibility. Clin. Infect. Dis. Off. Publ. Kwiecien, R., Kopp-Schneider, A., and Blettner, M. (2011). Concordance analysis: Infect. Dis. Soc. Am. 71, 2428–2446. doi: 10.1093/cid/ciaa325 part 16 of a series on evaluation of scientific publications. Dtsch. Arzteblatt. Int. 108, 515–521. doi: 10.3238/arztebl.2011.0515 Corman, V. M., Landt, O., Kaiser, M., Molenkamp, R., Meijer, A., Chu, D. K. W., et al. (2020). Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT- Lieberman, J. A., Pepper, G., Naccache, S. N., Huang, M. L., Jerome, K. R., and PCR. Eurosurveillance 25:2000045. doi: 10.2807/1560-7917.ES.2020.25.3.2000045 Greninger, A. L. (2020). Comparison of commercially available and laboratory- Frontiers in Microbiology 09 frontiersin.org Usai et al. 10.3389/fmicb.2022.1016201 developed assays for in vitro detection of sars-cov-2 in clinical laboratories. J. Clin. Rockx, B., Kuiken, T., Herfst, S., Bestebroer, T., Lamers, M. M., Oude Munnink, B. B., Microbiol. 58:e00821-20. doi: 10.1128/JCM.00821-20 et al. (2020). Comparative pathogenesis of COVID-19, MERS, and SARS in a nonhuman primate model. Science 368, 1012–1015. doi: 10.1126/science.abb7314 Liu, X., Feng, J., Zhang, Q., Guo, D., Zhang, L., Suo, T., et al. (2020). Analytical comparisons of SARS-COV-2 detection by qRT-PCR and ddPCR with multiple RStudio Team (2022). RStudio: Integrated Development for R. RStudio, PBC, primer/probe sets. Emerg. Microbes Infect. 9, 1175–1179. doi: 10.1080/22221751. Boston, MA. Available at: http://www.rstudio.com/ 2020.1772679 Swearengen, J. R. (2018). Choosing the right animal model for infectious disease McCray, P. B., Pewe, L., Wohlford-Lenane, C., Hickey, M., Manzel, L., Shi, L., et al. research. Anim. Models Exp. Med. 1, 100–108. doi: 10.1002/ame2.12020 (2007). Lethal infection of K18-hACE2 mice infected with severe acute respiratory van Kampen, J. J. A., van de Vijver, D. A. M. C., Fraaij, P. L. A., Haagmans, B. L., syndrome coronavirus. J. Virol. 81, 813–821. doi: 10.1128/JVI.02012-06 Lamers, M. M., Okba, N., et al. (2021). Duration and key determinants of infectious Moreau, G. B., Burgess, S. L., Sturek, J. M., Donlan, A. N., Petri, W. A., and virus shedding in hospitalized patients with coronavirus disease-2019 (COVID-19). Mann, B. J. (2020). Evaluation of K18-hACE2 mice as a model of SARS-CoV-2 Nat. Commun. 12:267. doi: 10.1038/s41467-020-20568-4 infection. Am. J. Trop. Med. Hyg. 103, 1215–1219. doi: 10.4269/ajtmh.20-0762 Vidal, E., Rodon, J., Brustolin, M., Cantero, G., Guallar, V., Muñoz-Fontela, C., Dowling, W. E., Funnell, S. G. P., Gsell, P. S., Izquierdo-Useros, N., et al. (2022). Chronological brain lesions after SARS- Riveros-Balta, A. X., Albrecht, R. A., et al. (2020). Animal models for COVID-19. CoV-2 infection in hACE2-transgenic mice. Vet. Pathol. 59, 613–626. doi: Nature 586, 509–515. doi: 10.1038/s41586-020-2787-6 10.1177/03009858211066841 Muñoz-Fontela, C., Widerspick, L., Albrecht, R. A., Beer, M., Carroll, M. W., de Watson, P. F., and Petrie, A. (2010). Method agreement analysis: a review of Wit, E., et al. (2022). Advances and gaps in SARS-CoV-2 infection models. PLOS correct methodology. Theriogenology 73, 1167–1179. doi: 10.1016/j. Pathog. 18:e1010161. doi: 10.1371/journal.ppat.1010161 theriogenology.2010.01.003 On the Calculation of TCID50 for Quantitation of Virus Infectivity | Wölfel, R., Corman, V. M., Guggemos, W., Seilmaier, M., Zange, S., Müller, M. A., SpringerLink (2022). Available at: https://link.springer.com/article/10.1007%2 et al. (2020). Virological assessment of hospitalized patients with COVID-2019. Fs12250-020-00230-5 (Accessed January 28, 2022). Nature 581, 465–469. doi: 10.1038/s41586-020-2196-x Oumarou Hama, H., Aboudharam, G., Barbieri, R., Lepidi, H., and Drancourt, M. Zhen, W., Manji, R., Smith, E., and Berry, G. J. (2020). Comparison of four (2022). Immunohistochemical diagnosis of human infectious diseases: a review. molecular in vitro diagnostic assays for the detection of SARS-CoV-2 in Diagn Pathol. 17:17. doi: 10.1186/s13000-022-01197-5 nasopharyngeal specimens. J. Clin. Microbiol. 58:8. doi: 10.1128/JCM.00743-20 R Core Team (2022). R: A language and environment for statistical computing. Zhu, N., Wang, W., Liu, Z., Liang, C., Wang, W., Ye, F., et al. (2020). Morphogenesis R Foundation for Statistical Computing, Vienna, Austria. Available at: https://www.R- and cytopathic effect of SARS-CoV-2 infection in human airway epithelial cells. Nat. project.org/ Commun. 11:3910. doi: 10.1038/s41467-020-17796-z Frontiers in Microbiology 10 frontiersin.org