Precision‐cut lung slices: A powerful ex vivo model to investigate respiratory infectious diseases

Abstract Respiratory infections are a leading cause of mortality worldwide. Most of the research on the underlying disease mechanisms is based on cell culture, organoid, or surrogate animal models. Although these provide important insights, they have limitations. Cell culture models fail to recapitulate cellular interactions in the lung and animal models often do not permit high‐throughput analysis of drugs or pathogen isolates; hence, there is a need for improved, scalable models. Precision‐cut lung slices (PCLS), small, uniform tissue slices generated from animal or human lungs are increasingly recognized and employed as an ex vivo organotypic model. PCLS retain remarkable cellular complexity and the architecture of the lung, providing a platform to investigate respiratory pathogens in a near‐native environment. Here, we review the generation and features of PCLS, their use to investigate the pathogenesis of viral and bacterial pathogens, and highlight their potential to advance respiratory infection research in the future.


| INTRODUC TI ON
Lower respiratory tract infections such as pneumonia are among the 10 most common causes of death worldwide and are a significant burden for healthcare systems (Gibson et al., 2013;World Health Organization, 2020). The swift onset of the SARS-CoV-2 pandemic further highlights that new respiratory pathogens can emerge rapidly.
Hence, there is an urgent need to accelerate fundamental research into respiratory infection biology and to develop technologies that allow for its translation into optimized diagnosis and treatment strategies.
To date, fundamental respiratory research has mostly relied on cell culture and surrogate animal models to investigate the mechanisms underlying host-pathogen interactions and different pathologies. Twodimensional (2D) cell culture has been the most employed model as it is cheap and easy-to-implement, but it has limitations. It predominantly relies on immortalized cell lines that often display significantly altered metabolism and gene expression profiles compared to primary cells (Sanderson, 2011). The use of primary cells can circumvent some of these drawbacks, as these retain more morphological and physiological characteristics of the source tissue. However, primary cells can only be retrieved in low numbers, expanded for a finite time, display limited differentiation capacity and, as the other 2D models, still lack the cellular heterogeneity, intercellular interactions, and complex structural organization of tissues. More complex three-dimensional (3D) cell culture models address some of these limitations. For example, primary airway epithelial cells cultured at air liquid interface (ALI) develop into a differentiated, polarized airway-like epithelium (Cao et al., 2020;Zscheppang et al., 2018). Stem/progenitor cells | 579 VIANA et Al. embedded in a matrix can self-organize into complex lung tissue-like organoids (Hiemstra et al., 2019;Nikolić & Rawlins, 2017). These models recapitulate many aspects of the airway epithelium, as the heterogeneous epithelial cell composition including mucus-producing goblet and beating ciliated cells, and some degree of 3D organization. Other cells, for example macrophages, can be added to these 3D models, allowing the investigation of cellular crosstalk and immune mechanisms (Hiemstra et al., 2019). Integration with microfluidic "organ-on-a-chip" systems, which continuously supply nutrients and remove waste, further enhances their potential to mimic in vivo conditions (Gkatzis et al., 2018;Hiemstra et al., 2019). However, current models are still unable to recapitulate the true cellular richness and spatial complexity of the lung.
Surrogate animal models allow experimentation under physiological conditions in a living organism and provide invaluable knowledge about the mechanisms underlying respiratory infectious diseases. Mice are the most frequently used model, due to the large range of genetically modified strains and tools that exist to study them. Guinea pigs reflect human airway pathology better than mice (Ressmeyer et al., 2006) and were the first model used to study tuberculosis and diphtheria. Although there are less tools available to study these, guinea pigs remain important surrogate models for Legionella pneumophila, Brucella spp., and several viruses (Padilla-Carlin et al., 2008). Pigs have been increasingly employed since porcine airway anatomy, physiology and more than 80% of immune parameters closely resemble that of humans (Meurens et al., 2012). In contrast, mice only share 10% of immune parameters. Still, despite best efforts, significant genetic, anatomical, and physiological differences often hinder the direct translation of animal research to humans (Bonniaud et al., 2018;Williams & Roman, 2016).
To bridge this gap, ex vivo human lung tissue explants (HLTEs), which are usually obtained from resected tissue samples from patients undergoing surgery for example, for lung cancer, have been used for modeling infection with L. pneumophila (Jäger et al., 2014), Chlamydophila pneumoniae (Rupp et al., 2004), Streptococcus pneumoniae (Xu et al., 2008), and Haemophilus influenzae (Drömann et al., 2010). As usually only small areas of tissue are resected at one time, the number of HLTEs generated is small, limiting the number of experimental conditions that can be tested. Moreover, while tumor tissue itself is avoided, HLTEs are typically derived from a diseased lung environment.
Low availability of human tissue, the high cost of infrastructure and maintenance of laboratory animals, and substantial efforts made in recent years to implement the 3Rs principles, Refinement, Reduction, and Replacement for animal experimen- They can be generated in large numbers from diverse organs including the lung (precision-cut lung slices-PCLS), from different animals and also from live or cadaveric human tissue and organs donated for research. This enables testing of a wide range of variables in replicate, reduces the amount of source tissue or animals needed and is therefore also significantly more cost-effective when compared to in vivo work.
Here, we will describe the generation and characteristics of PCLS, review their use for studying human respiratory pathogens so far and discuss their potential for advancing respiratory infectious disease research and antimicrobial drug discovery.

| G ENER ATI ON OF P CL S
Tissue slices have been used in research since the 1920's.
However, it was the development of precise vibrating microtomes that enabled PCTS in 1980, increasing the homogeneity and reproducibility of the generated slices (Krumdieck et al., 1980). It took seven more years until the introduction of an agar infusion method provided enough structural support to the soft and fragile honeycomb architecture of lung tissue, making it amenable for slicing (Placke & Fisher, 1987). This method was further developed and first applied to human lung tissue in 1994 (Fisher et al., 1994). Various adaptations have been made since then, and detailed step-by-step protocols for generating PCLS from human and mouse tissue are published (Gerckens et al., 2019;Wu et al., 2019).
Central to all protocols is that the freshly retrieved whole lungs or individual lobes are infused with warm low-melting agarose dissolved in buffer or tissue culture medium, at a concentration ranging from 0.5% to 3% (Figure 1). The agarose can be administered through the trachea, for example for small animal lungs, or through smaller airways of dissected lobes. It is delivered at a constant slow rate and the total volume adjusted to the natural tissue volume to prevent tissue damage. Once the tissue is inflated, rapid solidification of the agarose is achieved by immersing the tissue in cold buffer, followed by incubation at 4℃. Upon solidification, the tissue is sliced using a vibrating microtome, yielding PCLS of uniform thickness between 150 and 500 µm. Thinner slices are typically obtained from smaller animals and thicker slices from larger animals and human tissue (Alsafadi et al., 2020). Variations of this procedure have been recently reviewed (Alsafadi et al., 2020). The PCLS are kept immersed in cell culture medium in well plates for downstream applications. Serum is either absent or added in very low quantities in order to prevent cell growth and changes to the native cellular diversity (Sanderson, 2011). The number of slices which can be generated is limited by the size of the organ, but a small human lung lobe can yield hundreds of PCLS (hPCLS). Uniform punches fitting 96 well plate wells can be generated from PCLS using biopsy punchers, facilitating high-throughput applications. Reports on PCLS viability vary as this is likely affected by diverse factors (e.g., method for animal sacrifice, time, and conditions of conservation from tissue retrieval to slicing) (Sanderson, 2011) but hPCLS have been reported to remain viable for at least 15 days .

| P CL S: A NE AR NATIVE LUNG ENVIRONMENT TO S TUDY AIRWAY B I O LO GY
The process of PCLS generation preserves most features of the native lung tissue, that is, its architecture and mechanical properties (Hiorns et al., 2016;Pybus et al., 2021) as well as a complex cellular composition, including resident lung cells, for example, fibroblasts, ATI and ATII epithelial cells, and immune cells, for example, macrophages, monocytes, and NK and T cells (Stegmayr et al., 2021). proteomics (Khan et al., 2020), and RNA isolation for transcriptome profiling (Niehof et al., 2017;Stegmayr et al., 2021).
Molecular mechanisms in PCLS can be probed with small molecule inhibitors, with the advantage that the amount of drug needed for well plate experiments is significantly reduced compared to in vivo studies, decreasing cost and facilitating assessment of drugs with limited availability. Moreover, genetic manipulation of PCLS by siRNA-mediated protein knockdown has been achieved (Ruigrok et al., 2017(Ruigrok et al., , 2018. Notably, PCLS can also be obtained from healthy and diseased tissue from both human donors or animal models, as demonstrated for idiopathic pulmonary fibrosis (Alsafadi et al., 2017), chronic obstructive pulmonary disease (COPD), and asthma (Mertens et al., 2017), enabling the assessment of the interplay between these conditions and infections. Overall, these characteristics show that PCLS combine advantages of both in vitro and in vivo models and constitute an excellent ex vivo system to study lung biology, disease, and treatments. PCLS have therefore been widely adopted for pre-clinical drug discovery and toxicological studies (Bäckström et al., 2016;Fisher et al., 1994;Hess et al., 2016;Lauenstein et al., 2014;Maihöfer et al., 2017;Neuhaus et al., 2017Neuhaus et al., , 2018Switalla, et al., 2010a) including the evaluation of immune modulators (Henjakovic et al., 2008;Leus et al., 2016;Switalla, et al., 2010b;Temann et al., 2017) and the induction of antigen-specific T cell responses by new vaccine formulations (Neuhaus et al., 2013).
Additionally, PCLS were also used to characterize tissue tropism and functional properties of gene therapy vectors and oncolytic viruses (Rosales Gerpe et al., 2018).

| P CL S A S MODEL FOR H UMAN RE S PIR ATORY INFEC TI OUS D IS E A S E S
The first report of the use of PCLS to study the molecular mechanisms of infection focused on two respiratory pathogens, the Gramnegative bacterium C. pneumoniae and paramyxovirus Respiratory Syncytial Virus (RSV) (Ebsen et al., 2002). C. pneumoniae is associated with wide range of diseases, for example, pulmonary emphysema, and RSV is the most common cause of viral infections in the lower respiratory tract of infants and children. Immunofluorescence

| Paramyxoviruses
Paramyxoviruses are RNA viruses transmitted via inhalation of airborne droplets and able to cause diverse diseases in humans and animals. Apart from RSV, prominent members of this group are measles virus, the causative agent of measles, and metapneumovirus, an important cause of respiratory disease in particular in children (Park & Tishkowski, 2021;Troy & Bosco, 2016). PCLS from dogs, cynomolgus macaques, ferrets, and cotton rats, were assessed for their ability to support infection and discern differences in cell tropism and replication kinetics between the three human paramyxoviruses and canine distemper virus, the cause of distemper in carnivores (de Vries et al., 2018;Nguyen et al., 2013). Measles virus replicated optimally in macaque slices, metapneumovirus and RSV in cotton rat slices, and the canine distemper virus in slices from dogs and ferrets.
This work not only established effective PCLS models for these viruses but, as PCLS were infected with recombinant virus expressing fluorescent proteins, also highlighted the potential of PCLS for realtime monitoring of infection.

| Enteroviruses
Rhinoviruses (RVs) are the cause of the common cold, the most frequent human viral infection, and severe exacerbations of asthma and COPD (Troy & Bosco, 2016). Quantification of the RV load by quantitative PCR in hPCLS from asthmatic and healthy donors showed no differences between the groups at infection peak. However, RV triggered higher expression of cytokines, for example, IL25, TSLP, and IL13, in the tissue from asthmatics, indicating that RV-induced exacerbations F I G U R E 2 Comparison between human (a) and mouse (b) precision-cut lung slices (PCLS) structure (Blue: DAPI staining of nuclei). (c) Section of an hPCLS infected with Legionella pneumophila (Red: expressing mScarlet-I) 24 hr post infection (White: CellMask staining of cytoplasmic membranes). * mark alveolar spaces in (a) and (c). (a, c) were acquired by FV and (b) was acquired by FV and John Stegmayr may be linked to an altered immune response (Kennedy et al., 2018).
As hPCLS retain ciliary beating and are responsive to treatments with contractile stimuli such as the drug carbachol, the effect of RV infection on airway constriction was also analyzed, showing an enhanced contraction in infected asthmatic donor tissue (Parikh et al., 2020).
Increased carbachol-induced airway narrowing was also observed in mPCLS infected with rhinovirus C15 (RV-C15). Both studies support a model in which altered immune responses in asthmatic airways following RV infection drive airway hyperresponsiveness.
PCLS derived from a mouse model of allergic asthma showed a compromised immune response to RV, that is, increased release of IL-4, IL-6, and IL-10, which could be partially inhibited by the antiviral drug Rupintrivir . Bronchobini, a homeopathic multicomponent drug preparation marketed for inflammatory respiratory diseases,

| Adenoviruses
Adenoviruses cause diverse pathologies. Type 7 Adenovirus (Ad7) is associated with severe lower tract infection and pneumonia in both healthy and immunocompromised individuals, especially among children or adults living in crowded conditions (Clementi et al., 2021). Infection leads to infiltration of neutrophils in the lower respiratory tract and alveoli, followed by monocytes and later lymphocytes. Bovine and human PCLS support Ad7 replication and viral protein was detected in alveolar epithelial cells by immunohistochemistry (Booth et al., 2004). As in 2D models, PCLS infection triggered production and release of interleukin-8 (IL-8), the main chemoattractant for neutrophils, via the RAS/RAF-I/MEKI/ERK pathway. A subsequent study in hPCLS, revealed that IL-8 is mainly produced by type I alveolar epithelial cells, whereas IP-10, a chemokine attracting monocytes and lymphocytes, is produced by epithelial cells and macrophages and seems to require communication of both cell types (Wu et al., 2010).
These studies evidence the potential of PCLS to dissect cell tropism of pathogens and intercellular communication in the host response.

| Influenza viruses
Influenza A virus is a major cause of global morbidity and mortality. The mPCLS model also enables the assessment of complex interactions of infection, drug treatments, and environmental factors, such as smoking, a main risk factor for COPD. Influenza A virus is a known driver of exacerbations of COPD and asthma. PCLS from smoke-exposed mice revealed that smoke had no effect on clearance of H1N1 virus, but was associated with higher expression of inflammatory chemokines such as MCP-1/-3, KC, MIP-2, and GCP-2  (Meng et al., 2013). In pPCLS, swine influenza A virus (H3N2) replicated better than two avian influenza viruses (H9N2 and H7N7), but H9N2 replicated robustly in lung epithelial cells, indicating potential for interspecies transmission (Punyadarsaniya et al., 2011). Adaptive evolution experiments in pPCLS showed that after three passages avian influenza virus H9N2 evolved its HA protein to bind α2,6-linked sialic acids, typical surface receptors for swine and human viruses, in addition to the α2,3-linked sialic acids commonly used by avian subtypes (Yang et al., 2017). These mutants displayed enhanced replication in mice but not in pPCLS, suggesting that other adaptive mutations might still be required for virulence in pigs. These studies demonstrate the usefulness of PCLS to model zoonotic infections and assess the risk for emergence of new human pathogens.

| Coronaviruses
The SARS-CoV-2 pandemic generated the urgent need for relevant preclinical drug discovery platforms. hPCLS were employed for the characterization of the antiviral effect of Camostat mesylate, an inhibitor of TMPRSS2 and related proteases, which are required for proteolytic activation of the viral S protein and cell invasion (Hoffmann et al., 2021), and in a drug repurposing screen for inhibitors of SARS-CoV-2 replication (Zimniak et al., 2021). SARS-CoV-2 replicated in hPCLS as measured by quantitative RT-PCR and viral titer. Camostat mesylate and Fluoxetine, a selective serotonin reuptake inhibitor used for treatment of depression, reduced viral burden. Notably, Lopinavir, which repressed viral replication in in vitro 2D infection models, failed to reduce the viral titers in hPCLS. This shows the value of this more complex, ex vivo tissue-based infection model for the effective discovery of antimicrobials and for studying the pathogenesis of SARS-CoV-2 in the future.

| Bacterial pathogenesis in PCLS
Since the first ever use of PCLS as a model for respiratory infection with C. pneumoniae (Ebsen et al., 2002), an increasing number of studies have exploited PCLS to investigate how bacterial pathogens cause disease in humans. Additional studies on zoonotic pathogens, for example, Bordetella bronchiseptica, that focus on aspects of pathogenesis in the natural animal hosts have been discussed elsewhere (Vötsch et al., 2021).

| Staphylococcus aureus
Staphylococcus aureus is a multifaceted Gram-positive pathogen of high concern due to increasing multidrug resistance. Methicillinresistant S. aureus (MRSA) and Methicillin-sensitive strains (MSSA) cause community-pneumonia in healthy individuals, but MSSA are associated with a lower mortality rate. As murine macrophages were found to be growth permissive, it was speculated that macrophages modulate pathogenesis in humans. However, both S. aureus types survive, but do not replicate in isolated human alveolar macrophages, and in hPCLS the bacteria are mostly found in S. aureus-containing phagosomes in epithelial cells and interstitial regions (Brann et al., 2019). Only limited numbers are detected in alveolar macrophages. Thus, human alveolar macrophages are most likely not a key permissive replicative niche, pointing to other cell types as drivers of S. aureus lung infection. This study highlights the potential of PCLS to investigate cell tropism and key cellular interactions of bacterial pathogens.

| Coxiella burnetii
Coxiella burnetii is an obligate intracellular pathogen that causes Q-fever in humans, which can progress from flu-like symptoms into a life-threatening endocarditis (Dragan & Voth, 2020). In cell culture models, the bacteria replicate inside acidic parasitophorous vacuoles (PVs) in human alveolar macrophages and also non-phagocytic cells. hPCLS infection established that while individual bacteria could be detected in different cell types, replication occurred in PVs in alveolar macrophages (Dragan et al., 2019;Graham et al., 2016). Replication in PCLS depended on the Dot/ Icm type 4 secretion system, which is essential to deliver effector proteins for manipulation of host processes and biogenesis of the PV. No substantial structural damage to the PCLS tissue was observed even after 72 hr of infection. Infection induced the production and caspase-1-dependent release of mature IL-1β from PCLS.
Notably, these experiments were performed with the attenuated, laboratory strain C. burnetii NMII (RSA439, avirulent clone 4) (Millar et al., 2017), which induces a markedly different inflammatory response than wild type bacteria. Nevertheless, this study highlighted how hPCLS represent a promising model to investigate the C. burnetii pathogenesis in a physiological context.

| Bacillus anthracis
Bacillus anthracis is a spore-forming Gram-positive pathogen.
Pulmonary infection with B. anthracis spores causes inhalation anthrax, the most severe and deadly form of the disease. Early innate immune responses to B. anthracis spores were investigated in hPCLS, showing Mitogen-Activated Protein Kinase (MAPK)driven proinflammatory cytokine and chemokine production by macrophages and lung epithelia, including IL-8, MCP-1 and MIP--1α/β, potent chemoattractants for neutrophils, and monocytes (Chakrabarty et al., 2007). This was the first evidence of an active involvement of the lung epithelium in the innate immune response to B. anthracis spores and contrasts with the lack of cytokine production observed in a complex 3D cell model of primary small airway epithelial cells cultured in collagen matrices in presence of peripheral blood monocytes (Radyuk et al., 2003). While hPCLS likely represent a more accurate lung model and thus show more realistic early innate immune responses, this awaits validation by autopsy data.

| Mycobacteria
Mycobacterial infections are a global health problem (Gopalaswamy et al., 2020). While Mycobacterium tuberculosis (MTb) remains the Overall, these two studies show that PCLS can be a useful tool for fundamental mycobacterial research and drug discovery.

| Pseudomonas aeruginosa
Pseudomonas aeruginosa is a versatile opportunistic pathogen causing acute and chronic respiratory infections, particularly in immunocompromised, COPD, and CF patients. The bacteria use different virulence factors including a flagellum and type 2 and type 3 secretion systems to secrete diverse toxins and enzymes (Jurado-Martín et al., 2021). However, how both these virulence factors and pathogen-associated molecular patterns such as LPS, drive the immune response in the human lung is not fully understood. Using

| CON CLUS I ON & OUTLOOK
Bridging the gap between in vitro and in vivo models, PCLS are a reliable and powerful model to investigate respiratory infectious diseases. hPCLS in particular deliver highly relevant data reflecting human disease, eliminating the often speculative extrapolation of findings from surrogate models to humans. As many human pathogens are often unable to infect and replicate in certain animal models, hPCLS can also provide an important platform for modeling viral and bacterial co-infections, abrogating the host-specificity issues that hinder such studies. Some constraints remain, for example, limited genetic tractability, lack of cell infiltration, and of adaptive immunity, as well as the short life span of the slices which hinders their use for studying chronic infections, but these limitations are being actively tackled, for example, by embedding PCLS in hydrogels that expand their life span . Given the exceptional accessibility of PCLS for live imaging and the option to compare many pathogen strains or treatments, PCLS promise to become an invaluable tool for dissecting the molecular mechanisms of host-pathogen interactions within the physiological context of lung tissue and cellular microbiology research in the future.

ACK N OWLED G M ENTS
This work is part of the project "OPTIMISE," which has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement 844197. We thank Dr Darcy Wagner for hosting FV for a training period in her lab at Lund University, Sweden, during which FV and Dr John Stegmayr acquired image 2b.

CO N FLI C T O F I NTE R E S T
FV and GNS declare that they have no conflicts of interest with the contents of this article. COK has received consultancy fees from Insmed for the treatment of pulmonary non tuberculous mycobacterial infection.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this literature review as no new data were generated. Micrographs in Figure 2 are merely included for illustrative purposes.