Reduction and Replacement of animal models for antiviral testing using 3D human respiratory epithelia

The use of animals in preclinical drug testing is a matter of public interest, as it often involves suffering and discomfort. Animal models not always predict human responses, as argued by those who oppose in vivo research. Laboratory animals exist as inbred lines to increase the statistical power of preclinical tests. For each human disease there are several strains, that often produce inconsistent results, as observed for SARS-CoV-2 mouse models. Besides, this range of models does not reflect the clinical variety of human patients, not allowing to pinpoint differential disease development to find new cures. It emerges from this scenario that more accurate and standardized in vitro models should systematically complement animal models. The ongoing pandemic offers the unique opportunity to implement preclinical studies. The lack of medications after SARS-CoV-2 outbreak forced the healthcare systems to repurpose drugs based only on preclinical tests run in cell lines. Some drugs then failed clinical trials, while others displayed in patients a profile similar to that observed in cell lines, implying that tests in animals may be partially replaced by in vitro approaches. Three-dimensional (3D) tissue culture models offer a compelling complement to animal use, because they can be developed from human cells, thus improving the translation of the results to humans. In agreement with the 3R principles, we propose 3D human airway epithelia (HAE) as a complement to in vivo systems, for the testing of antivirals. Cultured at the air-liquid interface, HAE reproduce the architecture, the composition and the barrier defense mechanisms of the in vivo epithelium. They are reconstituted from human primary cells in serum-free medium, thus reducing the use of animal-derived components. The main asset of HAE is that they permit to study the original pathogenicity of clinical viral specimens, which would be lost using immortalized cell lines. Also, they have a significantly longer lifespan compared to other 3D systems (such as ex vivo lung slices or organoids) and can be used to characterize the long-term effects of the manipulations. HAE represent an ideal system to test antivirals targeting respiratory viruses. They are polarized and antiviral administration from the air-exposed apical or liquid-exposed basal side nicely mimic the in vivo air-borne or blood-borne incoming pathway. To implement the Replacement and Reduction principles, we present a multidisciplinary project involving virology, pharmacology, single cell genomics and machine learning. Our goal is to assess the power of HAE, in comparison to mouse models, to predict in humans the toxicity and efficacy of antivirals against respiratory viruses. We will pursue this goal in the context of SARS-CoV-2 infection. Specifically we will 1) determine, in HAE and in mice, the non-toxic dose-range of a collection of antivirals used in clinical trials as SARS-CoV-2 therapeutics; 2) determine the efficacy and the molecular effects of the antivirals against the most relevant SARS-CoV-2 variant, in HAE and in mice, by single cell RNA-sequencing (scRNA-seq); 3) develop a machine learning framework to minimize the number of preclinical tests in mouse models based on the toxicity and efficacy readout in HAE.