Abbreviations

3CLpro, 3C-like proteinase of the virus
ACE, angiotensin-converting enzyme
ACE2, angiotensin-converting enzyme 2
ADRP, ADP-ribose-1”-phosphatase
ARDS, acute respiratory distress syndrome
BPS, British Pharmacological Society
CARD, caspase activation and recruitment domain
CoV, coronavirus
E, Envelope protein of the virus
FRET, Förster Resonance Energy Transfer
GtoPdb, BPS/IUPHAR Guide to PHARMACOLOGY database
IUPHAR, International Union of Basic and Clinical Pharmacology
M, Membrane glycoprotein of the virus
MERS, Middle East respiratory syndrome
N, Nucleocapsid protein of the virus
nsp, non-structural protein of the virus
PAMP, pathogen-associated molecular pattern
PLpro, papain-like proteinase of the virus
RBD, receptor binding domain
S, Spike glycoprotein of the virus
SADS, Swine Acute Diarrhoea Syndrome
SARS, severe acute respiratory syndrome
TM, transmembrane

Author contributions

The document was conceived in discussions among SPHA, JA, JD, EF, SDH, FLS, AP and MJS; it was initially drafted by SPHA and all the co-authors contributed text and checked the manuscript; all the authors read and agree to submission of the manuscript.

Conflict of Interests

None of the authors has a conflict of interest to declare.

Introduction

PubMed has already accumulated a vast repository of information on SARS-CoV-2/COVID-19, which increases on a daily basis (on 2020-03-23, there were 1369 hits for COVID-19; this number more than doubled in the space of two weeks, so that by 2020-04-06 there were 2780 hits in PubMed for COVID-19). Clearly, there is a need to summarise this information critically and prioritise the elements which are constructive and useful for each individual sector. This document suggests priorities for how drug discovery and development might be rationally focussed for the rapid identification and successful translation of therapeutic agents to treat COVID-19.
Given the urgency of the current situation, clearly initial drug discovery should focus on repurposing licensed drugs, as dosage and safety information are largely to hand. Unfortunately, there is controversy over proof of efficacy for essentially all the potential repurposed agents for which preliminary, and, in many cases, non-peer reviewed data has surfaced. Some of this controversy is addressed below but efforts are underway from both WHO and NIH to coordinate larger, higher powered and better controlled studies in an attempt to demonstrate efficacy unequivocally. As a ‘second wave’, de novodiscovery focussing on novel agents may allow future refinement and capacity to treat patients who are unable to be treated by, or are unresponsive to, the repurposed agents, but it would be very unlikely to have these new drugs available to treat the current crisis.
The IUPHAR/BPS Guide to PHARMACOLOGY (GtoPdb) is an open-access database, developed by the International Union of Basic and Clinical Pharmacology (IUPHAR) and the British Pharmacological Society (BPS). It provides expert-curated descriptions of almost 3,000 human proteins and over 10,000 ligands, including more than 1,400 approved drugs. Management of the new resource is the responsibility of the Nomenclature and Standards Committee of IUPHAR (NC-IUPHAR), which acts as the scientific advisory and editorial board. The committee has an international network of over 700 expert volunteers organized into ∼60 subcommittees dealing with individual target families. The database is notably enhanced through the continued linking of relevant pharmacology with key immunological data types as part of the IUPHAR Guide to IMMUNOPHARMACOLOGY (supported by the Wellcome Trust) and by a major new extension, the IUPHAR/MMV Guide to Malaria PHARMACOLOGY (in partnership with the Medicines for Malaria Venture). The GtoPdb team centred at the University of Edinburgh have constructed a resource (Faccenda et al.), which provides a precis of the current understanding about the virus and potential associated drug targets and drugs. As with the other databases, the emphasis of the curation process is on stringent provenancing of the information provided, although inevitably the current situation limits the capacity for triangulation of data.

Nomenclature

Sequencing analysis of the novel virus has identified a high level of similarity with the virus identified to cause the Severe Acute Respiratory Syndrome (SARS) outbreak in China in 2002/03/04, which was known as the SARS coronavirus or SARS-CoV. Provisionally named as 2019-nCoV, the virus has been renamed SARS-CoV-2 (Viruses, 2020). For the purposes of this document, the virus is described as SARS-CoV-2, while the infectious disease is named as COVID-19 (World Health Organization, 2020). One of the positive aspects of the emergence of SARS-CoV-2 and COVID-19 is the rapidity at which aspects like genome sequencing (for example, Lu et al. , 2020) and 3D structures (for example, Yan et al. , 2020) have been described.

The viral cycle and virally-encoded potential drug targets

For general reviews of the coronaviruses, see Masters, 2006; Fehr and Perlman, 2015; de Wit et al. , 2016; Zumla et al. , 2016; Cui et al. , 2019; Desforges et al. , 2019; Song et al. , 2019. SARS-CoV-2 is a betacoronavirus; a lipid-enveloped, single-stranded, positive sense RNA virus. Other human coronaviruses include alphacoronaviruses, such as human coronavirus-229E (HCoV-229E), and betacoronaviruses, such as SARS-CoV and MERS-CoV (responsible for the Middle East respiratory syndrome) (for review, see Zumla et al. , 2016; Corman et al. , 2018; Pillaiyar et al. , 2020). More than 200 viral types have been associated with the common cold, of which 50% of infections are rhinovirus, but also include respiratory syncytial virus, influenza and coronaviruses, particularly HCoV-229E. Although HCoV-229E is regarded as ‘relatively benign’ since monocytes are much more resistant to infection, it does rapidly kill dendritic cells (Mesel-Lemoine et al. , 2012).
Classically, the viral lifecycle can be divided into six elements: cell attachment; cell entry; viral uncoating; nucleotide replication; viral assembly, and viral release (see Figure 1 ). Positive-stranded RNA viruses replicate in the cytoplasm of infected cells, in close contact with intracellular membranes. This organization allows a concentration of viral and host factors to enable virus production and to evade innate immune responses (reviewed by Yager and Konan, 2019).
The SARS-CoV-2 coronavirus 30 kb genome encodes 29 proteins (Gordonet al. , 2020). Historically, therapeutic benefit has been gained through exploitation of the differences between viral and host proteins that subserve superficially similar functions (proteases and nucleotide polymerases, for example). The rapidity with which structural elements of the SARS-CoV-2 proteome have been identified provides hope that drug discovery approaches will soon provide agents to target the virus selectively, with minimal impact on the host. Based on the evidence from orthologous proteins from other betacoronaviruses and the information currently available on SARS-CoV-2 (some of it not yet from peer-reviewed sources), we propose here the priority targets for pharmacological investigation. That should not be taken to mean that research should be limited to these targets, since there are undoubtedly a number of functions of the viral proteins still to be ascertained. It would be remiss not to conduct a thorough examination of all the viral proteome, both in isolation and in combination. The strategies we learn from investigation of the host:viral interaction from SARS-CoV-2 will stand us in good stead for future viral threats.

Cellular attachment and entry; replication, assembly and release

Coronavirus binds to cell surface proteins on target cells and, following proteinase priming of spike proteins on the virus surface, the virus is internalized into endosomal fractions that are subsequently acidified, or accumulates through a non-endosomal route (Fehr and Perlman, 2015). The endosomal route appears to involve clathrin (Inoueet al. , 2007), but there are contradictory reports of the importance of the intracellular C-terminus of ACE2 in this mechanism (see below) (Inoue et al. , 2007; Haga et al. , 2008). A fusion domain permits insertion of a key protein (Spike, see below), which then allows mixing of the viral and cellular membranes and subsequent release of the coronaviral genome into the cytoplasm.
Following entry into the host cell cytoplasm, the replicase gene of the viral RNA is translated. The genome of coronaviruses consists of a single, continuous, linear, ssRNA, capped at the 5’ end and with a 3’-polyA tail (Fehr and Perlman, 2015). Translation occurs from open reading frame (ORF) 1a and 1b at the 5’ terminus, with a ribosomal frameshifting mechanism allowing the overlap between ORF1a and ORF1b to generate the two polyproteins pp1a and pp1ab (Fehr and Perlman, 2015; Perlman and Netland, 2009; Snijder et al., 2003; Thiel et al., 2003). In SARS-CoV-2, the polyproteins are long, 4405 and 7096 aa, respectively. Encoded within the polyproteins of betacoronaviruses are two proteinases: papain-like proteinase, PLpro, and chymotrypsin-like proteinase, 3CLpro. In SARS-CoV, PLpro has three endoproteinase target sites, which release non-structural proteins 1-3 (Thiel et al. , 2003). 3CLpro has 11 cleavage sites to release the remaining non-structural proteins. In the family, these proteinases process the polyproteins to generate 16 functional non-structural proteins identified as nsp1-16 (Anand et al. , 2003; Thiel et al. , 2003; Ziebuhr et al. , 2007; Kindler et al. , 2016; Cuiet al. , 2019).
Downstream of the ORF1a and 1b are genes encoding four structural proteins (Spike, Envelope, Membrane and Nucleocapsid) (seeFigure 2 ) and a short series (described as at least 13 in total, Srinivasan et al. , 2020) of other proteins (see below). Once sufficient protein and RNA accumulate, coronavirus assembly takes place, centred on the structural proteins. The release of coronavirus particles involves the secretory pathway of the endoplasmic reticulum and Golgi apparatus and vesicular exocytosis (for review, see de Haan and Rottier, 2005; Fehr and Perlman, 2015), and it is likely, but as yet unconfirmed, that SARS-CoV-2 adopts this mechanism also.
To date, there is more evidence about the molecular detail involved in (and the possibilities to influence) viral recognition, entry and replication compared to uncoating, assembly and release, hence the attention paid here to the former three mechanisms.

Targetting virus recognition and cellular entry

The cell-surface anchor - ACE2

Among the coronaviruses, the spike protein interacts with proteinases to anchor on host cell surfaces. The cell-surface anchoring point for the alphacoronavirus HCoV-229E is aminopeptidase N (also known as CD13, Yeager et al. , 1992). For the betacoronavirus MERS-CoV, dipeptidylpeptidase 4 (also known as CD26, Raj et al. , 2013) is an anchor. Analysis of the co-crystal structure suggested that the SARS spike protein binds to the active site of angiotensin converting enzyme 2 (ACE2, Li et al. , 2005). Binding of SARS-CoV spike to ACE2 seems to require cholesterol-rich rafts in the host cells (Glende et al. , 2008). Recent evidence points to the spike protein of SARS-CoV-2 also binding to ACE2. Both SARS-CoV (Li et al. , 2003) and SARS-CoV-2 (Hoffmann et al. , 2020; Letko et al. , 2020) have been described to require ACE2 to enter cells. A particular domain of the spike protein of SARS-CoV-2, a so-called Receptor-Binding Domain (RBD), has been shown to facilitate binding to ACE2 (Hoffmann et al. , 2020). The ACE2 peptidase active site is located remotely from the cell membrane (Li et al. , 2005; Wrapp et al. , 2020; Yanet al. , 2020), into which the Spike protein binds. The RBD of the Spike protein is located in the S1 ectodomain, approximately a third of the way along the protein. ACE2 is a carboxypeptidase, which means it removes the terminal amino acid from the C-terminus of oligopeptides, and so it seems unlikely that the Spike protein is a substrate for ACE2.
In SARS-CoV-infected mouse lung, ACE2 protein expression was downregulated compared to uninfected mice (Kuba et al. , 2005). Following SARS-CoV Spike protein administration to mice, angiotensin II was increased in the lungs (Kuba et al. , 2005). These observations led to the suggestion that this was the molecular mechanism for the frequent development of acute respiratory distress syndrome (ARDS) during SARS-CoV infections (Imai et al. , 2005; Kubaet al. , 2005).
ACE2 activity has been reported to be released from plasma membranes by proteolysis, thought to be through the action of TNFα convertase (ADAM17, A Disintegrin And Metalloproteinase domain containing protein 17, Lambert et al. , 2005). ACE2, and ACE, activity can be measured in human plasma (Ocaranza et al. , 2006; Herath et al. , 2007; Lew et al. , 2008). Human plasma ACE2 activity is reported to be ‘masked’ by the presence of endogenous inhibitors (Lew et al. , 2008), which don’t yet appear to have been precisely defined. Blood ACE2 activity can be altered in pathology; for example, serum ACE2 was found to be decreased in patients following acute ischemic stroke (Bennionet al. , 2016).
The expression of ACE2 mRNA and enzyme activity in cardiac tissues were increased following repeated oral administration of the AT1angiotensin II receptor antagonist losartan, while oral administration of an ACE inhibitor lisinopril only increased cardiac mRNA expression, but not enzyme activity (Ferrario et al. , 2005).
Studies using disruption of the ace2 gene in mice indicated an increase in circulating angiotensin II levels and a severe cardiac contractility defect, which could be ‘rescued’ with simultaneous genetic disruption of ACE (Crackower et al. , 2002). An early investigation of ACE2 polymorphisms in man failed to show an association with hypertension (Benjafield et al. , 2004) and a study of SARS victims and ACE2 polymorphisms failed to find a correlation with patient outcomes (Chiu et al. , 2004).

The coronaviral Spike protein

The spike protein is the largest viral structural protein (~1200-1400 aa) and is heavily glycosylated, forming extended trimeric structures providing the characteristic ‘crown’ feature of coronaviruses (Belouzard et al. , 2012) (seeFigure 2 ). The ectodomain is divided into the S1 domain responsible for binding to ACE2, whereas the S2 domain is responsible for the fusion machinery. Following binding of the S1 domain to ACE2, a deformation of the pre-fusion trimer results (Wrapp et al. , 2020). Surface plasmon resonance of the binding of human ACE2 to the immobilized SARS-CoV-2 indicated an affinity (Kd value) of 15 nM, an order of magnitude larger than SARS-CoV binding to ACE2 (Wrapp et al. , 2020). Using a related label-free technique, biolayer interferometry, affinities of 5 and 1.2 nM for binding of SARS-CoV and SARS-CoV-2 spike protein, respectively, to human ACE2 has been reported (Walls et al. , 2020).
Although a proteolytic cleavage site at the S1/S2 boundary of the SARS-CoV Spike protein is the best characterised, a second site upstream of the fusion peptide in the S2 domain, called S2’ has also been described (Belouzard et al. , 2009). This raises the possibility that multiple other proteases might be targetted to influence coronavirus activation (Millet and Whittaker, 2015).
The SARS-CoV S2 domain has a pair of α-helices, which may participate in coiled:coil structures during membrane fusion (Petit et al. , 2005). The host complex of ZDHHC9 (Link to UniProt) with GOLGA7 (Link to UniProt), a palmitoyltransferase, which modifies the low molecular weight G proteins NRAS and HRAS (Swarthout et al. , 2005), also palmitoylates the cysteine-rich S2 endodomain of the SARS-CoV to facilitate membrane fusion (Petit et al. , 2007).
Very recently, in a comparison of the S2 domains of SARS-CoV and SARS-Cov-2, an enhanced capacity of the novel virus’ S2 domain for membrane fusion was observed and suggested to result from eight differing amino acids (Xia et al. , 2020). Using a series of oligopeptides conjugated to lipid entities, high affinity (IC50 values in the nanomolar range) inhibitors of cell fusion were identified.

Interfering with the ACE2:Spike interaction

Given that the spike protein binds to the active site of ACE2 (Liet al. , 2005), in theory, any alteration in the availability of the active site should influence the binding of the spike protein and, hence, interfere with SARS-CoV-2 infection. One option would be to provide an excess of an endogenous peptide substrate, or more conventionally to apply a selective enzyme inhibitor.
Endogenous substrates of ACE2
ACE2, discovered in 2000 (Donoghue et al. , 2000), shares 40% sequence similarity to ACE within the N -terminal domain and is a type I transmembrane metallopeptidase. Unlike ACE, it functions as a zinc carboxypeptidase to cleave single C-terminal amino acids from peptides, particularly hydrolysing Pro-Phe residues in angiotensin-(1-8) to angiotensin-(1-7), [Pyr1]-apelin 13 to [Pyr1]-apelin-(1-12) and [des-Arg9]-bradykinin to bradykinin-(1-8) with high efficiency. It may also cleave other peptides less effectively (Vickers et al. , 2002), shown below: