Research review paperRecent advances in human respiratory epithelium models for drug discovery
Introduction
There exists a clear and present need to improve the human relevant tools at our disposal for mechanistic investigations of respiratory pathogenesis and therapeutic drug development. Current in vivo models almost exclusively utilise non-primate animals which have been indispensable for aiding advancements in the mechanistic understanding of pulmonary pathogenesis and therapeutic drug development (Bonniaud et al., 2018). However, despite their utility, these animal models have been limited by their poor correlation to the human condition. Organisations such as the NC3Rs (UK National Centre for the Replacement, Refinement and Reduction of Animals in Research) have accelerated efforts to move beyond the use of animals for scientific purposes. Recent advances in the multidisciplinary fields of complex 3D cell culture, biofabrication and microfluidics offer unique opportunities to address the problem of generating models which faithfully replicate the biological processes of human organs in vivo. This review highlights the current state-of-the-art of respiratory epithelium modelling, and describes some of the inefficiencies of the current respiratory translational models used in drug discovery and target validation (Hendrickx et al., 2018).
Preclinical drug development processes have been optimised with the aim of determining potential toxicities and efficacies of novel compounds to reduce the inherent risk of first-in-man studies. Whilst these processes vary according to the disease and target being studied, they utilise common frameworks. For example: Initial target discovery stages serve to establish potential therapeutic roles of enzymes and membrane-bound receptors for a known pathological hallmark of disease while assessing known mechanisms of toxicity (also selectivity, tractability etc). Next, tens of thousands to millions of novel or repurposed molecules housed in ‘compound libraries’ are screened for ‘hits’ against these identified targets (‘hit identification’), usually via 2D cell-line based, high-throughput in vitro assays (Marx et al., 2016). More complex in vitro models are used during both the initial target discovery stages and subsequent target validation and lead optimisation studies, where promising early lead molecules are screened for their toxicity and efficacy and subsequently cut to candidate molecules. Finally, animal models are utilised during DMPK (drug metabolism and pharmacokinetics) and ADMET (absorption, distribution, metabolism, excretion and toxicology) studies to determine optimal dosages, potential side effects and drug-drug interactions. Only compounds delivering continual success to this stage are selected to progress into human clinical trials. However, the effectiveness of the current drug development process has been undermined by its reliance upon the implementation of inadequate models that lack translatability to the human condition.
Despite earnest efforts, drug development has increasingly failed at phase II/III of human clinical trials, attributable to the lack of predictivity of human in vivo efficacy from the current in vitro models used in early-stage studies, thus emphasising the need for better design and validation of models (Booth and Zemmel, 2004; Harrison, 2016). Novel therapeutic compounds entering clinical trials for respiratory disorders have shown a 7% and 15% chance of success from phase I and II trials respectively, and an overall success rate of less than 7% (Dowden and Munro, 2019). Though this figure is in the middle echelons of therapeutic-area dependent averages for novel compound success in clinical trials (3-16%) (Dowden and Munro, 2019), it still represents a huge drug attrition problem for the sector as a whole. Analysis into the causes of clinical failure across therapeutic areas showed between 2013 and 2015, 73% and 69% of all failures in phase II and III trials respectively, were due to insufficient efficacy and safety (48% and 25% in phase II and 55% and 14% in phase III respectively) (Harrison, 2016). Perhaps most worryingly, these figures remain largely unchanged as of 2016-2018, with 79% of overall failures due to safety and efficacy (with the remaining 21% citing operational, strategic and/or commercial reasons for failure) (Dowden and Munro, 2019).
Importantly, failures due to insufficient efficacy are almost twice as likely in phase II, and more than twice as likely in phase III, than failures due to toxicity (Harrison, 2016). This is directly attributable to the implementation of large-scale, standardised in vitro safety assays early in drug discovery. Here, the use of standardised safety assays via reductionist in vitro screening models, allow for efficient testing of molecules during hit-to-lead studies. This approach has been largely successful, and has contributed to a significant reduction in the number of failures due to toxicity in phase I and II trials (compared to efficacy failures (Dowden and Munro, 2019; Harrison, 2016)). However, this approach has yet failed to address the current drug attrition problem. Concerns pertaining to effectively determining efficacy have especially been undermined (Ledford, 2011).
Toxicity failures may be addressed in a number of ways, such as greater use of humanized monoclonal antibodies relative to small molecules due to their reduced off-target toxicity (Paul et al., 2010). Efficacy failures have been attributed to the lack of control of bias in preclinical proof of concept studies, where removing such biases in preclinical assessments of efficacy may serve as an effective accelerator of clinical success (Lindner, 2007). Furthermore, a recent analysis of 28 projects at AstraZeneca attributed 40% of project failures to insufficient target linkage to the disease and the availability, or lack thereof, of validated models (Cook et al., 2014). Therefore, addressing the unmet preclinical needs for efficacy testing with either more appropriate animal models (Kola and Landis, 2004), or increased use of complex in vitro models, may serve to significantly improve compound efficacy studies. Certainly, there exists a requirement for more predictive models in the target validation stages of drug development, practically in the form of organotypic in vitro human assays.
A wide array of functional in vitro models have been utilised for the study of the pathologies associated with the human respiratory system (Ball and Padalia, 2019; Fraser, 2005) (see Fig. 1). Many of these systems utilise models of the respiratory epithelium, which serves to warm, moisten and remove harmful pathogens and particulates from inspired air. The most commonly utilised in vitro model of the respiratory airway is that of the tracheobronchial epithelium. Here, the extrapulmonary conducting airways are comprised of C-shaped hyaline cartilaginous rings (Kia'i and Bajaj, 2020), a collagenous submucosa (Fraser, 2005), and a pseudo-stratified ciliated epithelium supported by a fibroblast-laden lamina propria and basement membrane (Khan and Lynch, 2020). The tracheo-bronchial respiratory epithelium subsequently contains a number of specialised cellular phenotypes (see Table 1). Replicating each of these components in a model system, as well as the array of specialised cell phenotypes present, remains a challenge.
Bacterial and viral infections most frequently affect the upper respiratory tract in humans (Thomas and Bomar, 2020). The rhinovirus or common cold remains the most common of the viral infections, but others include the coronavirus, respiratory syncytial virus and the adenovirus. Inflammatory lung diseases also characteristically involve pathologies pertaining to the respiratory airway epithelia (Huang et al., 2011). These include chronic obstructive pulmonary disorder (COPD) which is currently the third leading cause of death worldwide, with estimates for a COPD-derived economic burden of £1.7 trillion by 2030 (Quaderi and Hurst, 2018).
Furthermore, asthma, the most prevalent respiratory disease in the world (GBD 2015 Chronic Respiratory Disease Collaborators, B J, et al., 2017), and the genetic, autosomal recessive disorder cystic fibrosis (CF), both develop pathologies that arise from a loss of respiratory epithelium function. The primary pathology of CF is characterised by the secretion of abnormally viscous mucus from goblet cells and serous mucus glands, which inhibits mucociliary clearance (MCC) and causes an increased risk of infection due to improper clearance of the respiratory airways (Huang et al., 2011). Similarly, a loss of MCC caused by a reduced number of ciliated cells and goblet cell hyperplasia is an underlying physiology of bronchitis and COPD (Gohy et al., 2019). Loss of functions of the respiratory epithelium, such as a defective epithelial barrier derived from inhalation of cigarette smoke and environmental insults have been linked to the onset of COPD and asthma respectively (Gon and Hashimoto, 2018a; Xiao et al., 2011). These insults can damage the protein complexes present between various cells in the respiratory epithelia and cause a breakdown of paracellular transport mechanisms and a loss of efficient control of substance diffusion in/out of the subepithelial space (Brune et al., 2015) (see Table 1). Ciliated epithelial cell dysregulation, squamous metaplasia and goblet cell hyperplasia are also associated with COPD (Gohy et al., 2019). Therefore, the use of models which effectively model these phenomena are vital for the effective development of novel therapeutics.
It is true that all models are reductionist in nature, and therefore will ultimately fail to fully recapitulate the complexity of a target organism. Therefore, it's imperative that we seek models that provide an ‘economical description of the natural phenomena’ while remaining alert to their underlying failings (Box, 1976). Current translational approaches lack the ability to provide the required understanding of disease mechanisms and signalling pathways that underpin respiratory pathogenesis. Existing in vitro models are complementary rather than alternative models to animal studies (albeit they can serve to reduce the number of animal studies required), with the sole use of multiple in vitro models remaining insufficient (Bonniaud et al., 2018). As a result, the transition from in vitro modelling and animal testing of novel compounds to first-in-man studies remains a “leap of faith” (Bonniaud et al., 2018).
The focus of this review is to highlight the current state-of-the-science of respiratory translational models used in drug discovery and target validation (Hendrickx et al., 2018). Here, the utility and limitations of in vivo and ex vivo modelling of the respiratory epithelium are described, with a focus on reviewing the current in vitro models of the tracheo-bronchial epithelium.
Section snippets
In vivo lung models
Traditionally, disease modelling of respiratory disorders in small animals has been the primary method of understanding the mechanisms and pathologies of a disorder in man. Schanker's seminal work in the development of pulmonary drug absorption and inhaled therapeutic drug deposition in in vivo respiratory models has remained a foundation for respiratory models (Burton and Schanker, 1974; Enna and Schanker, 1972a, Enna and Schanker, 1972b; Mahato and Narang, 2010; Schanker and Burton, 1976).
Ex vivo lung models
The limitations of the current animal models and 3D cell culture systems to accurately mimic the respiratory epithelium have led to the use of human ex vivo tissue models, involving the culture of explanted human lung tissue. The resulting ex vivo tissue possesses the cellular composition of the native human lung, as well as all of the correct extracellular matrix components and complexity (Liu et al., 2019a).
Isolated perfused lungs (IPL), derived from rejected lungs for transplantation, have
2D cell-line airway models
Two-dimensional (2D) monolayer culture systems are still heavily used in preclinical drug development for high-throughput toxicity screening of novel compounds due to their ease-of-use, availability and convenience e.g. rapid time to cell confluency (reviewed by Castellani et al., 2018; Faber and McCullough, 2018; Hiemstra et al., 2018; Lechanteur and das Neves, 2018; Nikolić et al., 2018). The relatively successful implementation of these reductionist models in standardised high-throughput
3D air-liquid interface respiratory epithelium models
To overcome the issues related with 2D culture of cell-line based airway models, the modern gold-standard for human respiratory airway models consist of culturing primary human bronchial epithelial cells (HBECs) (also applied to human nasal epithelial cells (HNECs)) at an air-liquid interface (ALI) (see Fig. 2). HBECs, often obtained from bronchial brushings or via cadaveric donor tissue, are cultured to confluency on transwell inserts (characteristically in a 24-well plate format). Once
Concluding remarks and future outlook for modelling the respiratory epithelium
All the methodologies presently reviewed serve various functions aimed at studying respiratory physiology. The advancements in 3D cell culture technologies have vastly improved the ability to create functional in vitro models that recapitulate in vivo organ structures (Moroni et al., 2018). Furthermore, biofabrication methods such as 3D printing (Wu et al., 2020) and bioprinting (Ma et al., 2018; Ong et al., 2017; Vanderburgh et al., 2017; Zhang et al., 2019), electrospinning (Chen et al., 2018
Author contributions
N.Y conceptualised and wrote the manuscript. G.W, MB and W.S mentored the work. All the authors read, discussed and commented on the manuscript.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the content reported in this review article.
Acknowledgements
The authors acknowledge financial support by the Biotechnology and Biological Sciences Research Council and GlaxoSmithKline (BBSRC LIDo DTP Industrial CASE studentship, BB/R505985/1). M.B. and W.S. also thank financial support by the Engineering and Physical Sciences Research Council in the United Kingdom [EP/L020904/1 and EP/R02961X/1] and Wellcome Trust (106574/Z/14/Z). Figures were created with BioRender.com.
References (317)
- et al.
A new computer-controlled air-liquid interface cultivation system for the generation of differentiated cell cultures of the airway epithelium
Exp. Toxicol. Pathol.
(2016) - et al.
Pulmonary drug delivery with aerosolizable nanoparticles in an ex vivo lung model
Int. J. Pharm.
(2009) - et al.
Genetic engineering of a 3D in vitro human airway model sensitive to carcinogens
Toxicol. Lett.
(2017) - et al.
Human primary bronchial lung cell constructs: the new respiratory models
Toxicology
(2010) - et al.
Assay establishment and validation of a high-throughput screening platform for three-dimensional patient-derived colon cancer organoid cultures
J. Biomol. Screen.
(2016) - et al.
Regulatory cytokine function in the respiratory tract
Mucosal Immunol.
(2019) - et al.
Ex vivo lung graft perfusion
Anaesth. Crit. Care Pain Med.
(2016) - et al.
Rhinovirus infection does not alter bronchodilation in human precision cut lung slices from asthma donors
J. Allergy Clin. Immunol.
(2019) - et al.
IL-27 is elevated in patients with COPD and patients with pulmonary TB and induces human bronchial epithelial cells to produce CXCL10
Chest
(2012) - et al.
Precision-cut lung slices from bleomycin treated animals as a model for testing potential therapies for idiopathic pulmonary fibrosis
Pulm. Pharmacol. Ther.
(2019)