key: cord-0977796-us6mkl7x
authors: Friman, Tomas; Chernobrovkin, Alexey; Molina, Daniel Martinez; Arnold, Laurence
title: CETSA® MS profiling for a comparative assessment of FDA approved antivirals repurposed for COVID-19 therapy identifies Trip13 as a Remdesivir off-target
date: 2020-07-20
journal: bioRxiv
DOI: 10.1101/2020.07.19.210492
sha: 38011ade3e2a5cd568e87d0673a3c89e07d5d32d
doc_id: 977796
cord_uid: us6mkl7x

The reuse of pre-existing small molecules for a novel emerging disease threat is a rapid measure to discover unknown applications for previously validated therapies. A pertinent and recent example where such strategy could be employed is in the fight against COVID-19. Therapies designed or discovered to target viral proteins also have off-target effects on the host proteome when employed in a complex physiological environment. This study aims to assess these host cell targets for a panel of FDA approved antiviral compounds including Remdesivir, using the cellular thermal shift assay (CETSA®) coupled to mass spectrometry (CETSA MS) in non-infected cells. CETSA MS is a powerful method to delineate direct and indirect interactions between small molecules and protein targets in intact cells. Biologically active compounds can induce changes in thermal stability, in their primary binding partners as well as in proteins that in turn interact with the direct targets. Such engagement of host targets by antiviral drugs may contribute to the clinical effect against the virus but can also constitute a liability. We present here a comparative study of CETSA molecular target engagement fingerprints of antiviral drugs to better understand the link between off-targets and efficacy.

The COVID-19 pandemic has seen a significant worldwide effort to reuse or repurpose preexisting therapies in order to combat the emerging viral threat. There have been numerous studies reported using a variety of technologies in efforts to screen panels of pre-validated molecules, many repurposed from viral therapies [1] [2] [3] [4] [5] . These studies are conducted in the hope that efficacy against the SARS-CoV-2 virus may be discovered, whilst avoiding the lengthy yet essential drug discovery pipeline that even with modern standards typically takes several years from hit and target identification, to reach clinical testing of a lead candidate drug molecule 6 .

Utilising a preexisting molecule has a significantly lower risk than rapidly developing novel chemistry, since it has already successfully navigated the prerequisite safety and toxicologic testing for use in humans. However, the original purpose of the small molecule may have undescribed off-target effects that are deemed to be tolerable when weighed against therapeutic benefit. These effects, potentially caused by drug-protein interactions, are often poorly understood 7 .

For example, many antiviral compounds are structural analogues of nucleoside triphosphates (NTPs) that have diverse biological properties and therapeutic consequences since nucleotides have an essential role in virtually all biological processes 8 . Therefore, given the abundance of nucleotide interacting proteins in the host cell, off-target interacting proteins, or an imbalance of the cellular nucleotide pool would be an expected consequence of utilising nucleotide analogues in therapy 9 .

The persistent and fundamental problem of host off-target effects arise from using a molecule to disrupt viral biology, whilst simultaneously exposing the host biology to the same chemical challenge. Methods to describe the severity of hitting off targets, rely upon in vitro and in vivo assessment or the presentation of a phenotype that can be assessed as to whether acceptable or not. But this requires knowledge and the ability to measure non-intended target biology. For example, Remdesivir is known to have an efficacy of 100nM for the viral polymerase its intended target, and 500-fold less efficacious against human polymerases 10 . It has previously not been established which other proteins may interact with and nor whether these potential interactions would elicit a response with a measurable output using conventional means.

In light of this, traditional off-target investigation relies on known functions or activities which as a prerequisite require the host proteins responsible for these activities to be studied in bias.

Methods that are independent of activity and in an unbiased way report on compound interaction against the entire proteome, have only in recent years been established 11, 12 . The CEllular Thermal Shift Assay (CETSA) is a powerful technique to detect protein ligand interactions in cells 13 . Coupled with mass spectrometry (MS) as a readout, CETSA MS is a technique employed in the identification of off-target effects in proteome-wide studies observing the thermal stabilization or destabilization of endogenous proteins and downstream effects after matrix and compound incubation. The method is being increasingly employed in both mechanism of action (MoA) studies and to identify primary and off-targets of candidate drug molecules. For example, quinine and drug target interactions in Plasmodium falciparum 14, 15 . In this study, we screened a panel of drugs using the CETSA MS format on HepG2 cells to identify host proteins as hopeful starting points for further research and possible inroads into the improvement or development of fortuitous therapies for SARS-CoV-2 infection.

Given the intense global interest in searching for a viable therapy combined with the wide accessibility to information sources and even raw data, efforts from a wide variety of groups have been well documented in both the scientific and non-scientific media. The inclusion of compounds for this study was directed around prominent molecules discussed in the literature and adopted for clinical trials in the earlier phases of the worldwide pandemic, namely

Remdesivir and Hydroxychloroquine. The study was bolstered by the edition of other compounds repurposed from a variety of anti-viral classes including retroviral reverse transcriptase and protease inhibitors that were available for expeditious purchase from commercial sources [16] [17] [18] [19] .

This study investigates compound effects on uninfected whole HepG2 cells. Understanding how the molecule reacts in an environment containing both viral and host cell proteins is not beyond the technique, but outside of the capacity and scope for this study that was completed utilising a preexisting in vitro platform with a per compound acquisition time of approximately ~6 hours.

The human cell line HepG2 was procured from ATCC and cultured until 70% confluency in collagen-coated flasks. The cells were cultured in DMEM/F12 (without phenol red) (ThermoFisher) supplemented with 10% FBS (ThermoFisher), 5 mM sodium pyruvate (ThermoFisher), 1X NEAA (ThermoFisher) and PEST (ThermoFisher). Cells were detached for 60 minutes at 37˚C with end-over-end rotation. Viability after 1 hour of incubation with each compound were greater than 90%.

Each of the treated cell suspensions was further divided into 12 aliquots that were all subjected to a heat challenge for 3 minutes, each at a different temperature between 44 and 66°C. After heating, all temperature points for each test condition were pooled to generate 32 individual (compressed) samples.

Precipitated proteins were pelleted by centrifugation at 30 000 x g for 20 minutes and supernatants constituting the soluble protein fraction were kept for further analysis.

The experiment was performed over three independent biological replicates.

The total protein concentration of the soluble fractions were measured by Lowry DC assay (BioRad). From each soluble fraction, a volume containing an equivalent of 20µg of total protein was taken for further sample preparation.

Samples were subjected to reduction and denaturation with tris(2-carboxyethyl)phosphine (TCEP) (Bond-breaker, Thermo Scientific) and RapiGest SF (Waters), followed by alkylation with chloroacetamide. Proteins were digested with Lys-C (Wako Chemicals) and trypsin (Trypsin Gold, Promega).

After complete digestion had been confirmed by nanoLC-MS/MS, samples were labelled with 16-plex Tandem Mass Tag reagents (TMTpro, Thermo Scientific) according to the manufacturer's protocol.

Labeling reactions were quenched by addition of a primary amine buffer and the test concentrations and room temperature control samples were combined into TMT16-plex sets such that each TMT16-multiplex set contained 12 test compounds, two positive control samples (MTX+Vincristine) and two negative controls (1% DMSO). The labelled samples were subsequently acidified and desalted using polymeric reversed phase chromatography (Oasis, Waters). LC-MS grade liquids and low-protein binding tubes were used throughout the purification. Samples were dried using a centrifugal evaporator.

For each TMT16-multiplex set, the dried labelled sample was dissolved in 20 mM ammonium hydroxide (pH 10.8) and subjected to reversed-phase high pH fractionation using an Agilent 1260 Bioinert HPLC system (Agilent Technologies) over a 1.5 x 150 mm C18 column (XBridge Peptide BEH C18, 300 Å, 3.5 µm particle size, Waters Corporation, Milford, USA). Carbamidomethylation of Cys, TMTpro-modification of Lysine and peptide N-termini were set as static modifications. For protein identification, validation was done at the peptidespectrum-match (PSM) level using the following acceptance criteria; 1 % FDR determined by Percolator scoring based on Q-value, rank 1 peptides only. For quantification, a maximum coisolation of 50 % was allowed. Reporter ion integration was done at 20 ppm tolerance and the integration result was verified by manual inspection to ensure the tolerance setting was applicable. For individual spectra, an average reporter ion signal-to-noise of >20 was required.

Only unique or razor peptides were used for protein quantification.

Quantitative results were exported from Proteome Discoverer as tab-separated files and analyzed using R version 4.0.2 software. Protein intensities in each TMT channel were log2transformed and normalized by subtracting median value per each TMT sample and each TMT channel (column-wise normalization). For each protein and each compound, thermal stability changes were assessed by comparing normalized log2-transformed intensities to DMSO treated control using moderated t-test implemented in "limma" R-package version 3.44.1 20 . Proteins were quantified via isobaric labelling liquid chromatography coupled to massspectrometry (LC-MS). The resulting dataset covers more than 8,000 protein groups, of them 5,873 protein groups were reliably quantified in more than 17 out of 22 treatments with at least two unique peptides.

In order to assess compound induced protein thermal stability changes, for each treatment we Given during the activation of the prodrug includes an intracellular esterase hydrolysis step, an interaction is not surprising.

In contrast, and most notable from this study is the destabilization of Pachytene checkpoint protein 2 homolog (TRIP13). Trip13 is a hexameric AAA+ ATPase and a key regulator in chromosome recombination and structural regulation, such as crossing over and DNA double strand breaks 23 . Trip13 is essential in the spindle assembly checkpoint and is expressed in a number of human cancers where its reduction has been linked with effects on proliferation and hence therapeutic benefit 24 . It is plausible that Remdesivir, in its fully synthesized triphosphate form is competitive with endogenous ATP binding with Trip13, disrupting or affecting multimerization with itself or downstream on the spindle assembly complex.

Interestingly, GS-441524, a metabolite of Remdesivir had no significant hits in this study.

There could be multiple explanations for this, but in this case, it is established that unfavorable compound properties of GS-441524 result in limited cellular uptake. Especially when a 60minute incubation protocol is considered. In our experience, addition of nucleosides often has impact on several proteins involved in cellular nucleoside homeostasis.

As apparent from Figure The remaining compounds either induce no shifts or do so for very few proteins. The latter make up cluster 2 in Figure 2 where the lack of pronounced molecular fingerprint does not allow for further division into separate or unique groupings. Lamivudine treatment resulted in in a stabilizing shift for DCK, which is known to be responsible for the intracellular phosphorylation of the drug 28 which provide confidence of cellular uptake. These data may well constitute useful information when taken in the context of further study.

This study intended to help better understand any off-target effects of Remdesivir and Chloroquine as two prominently repurposed drugs for targeting SARS-CoV-2, with the view to identify potential biological inroads for further investigation.

This is an intact cell study and therefore conducted in a highly biological context. In that, proteins exist at their endogenous expression and environment. There are no previous studies using CETSA MS to comparatively analyze a panel of anti-viral compounds. Given the primary purpose of the majority of these drugs is to interact or inhibit viral proteins, there was no expectation that common host targets would be identified.

In contrast, the Chloroquine molecules are known to have substantial effects to the endosomal compartment and the expectation was significant and broad shifts in these samples that was not observed. Aside from the described shifts, the bulk of cluster 2 represented less defined changes to a broad range of biological activities, not allowing for a definitive molecular fingerprint to be elucidated.

This study was designed to identify previously unidentified proteins that could have critical importance for the reported activity of Remdesivir and other compounds in the context of COVID-19. The inclusion of a panel of molecules allows for the cross comparison against hits specific to one molecule, which has facilitated the novel finding that Remdesivir uniquely destabilizes Trip13.

The function of Trip13 does not lend itself to that of an obvious benefit or hinderance to viral infection, as would be considered by a protein with known host innate viral immunity activity.

But the fact it interacts with nucleotides and forms a homohexamer which if diminished removes activity, lends it to the possibility the interaction with Remdesivir may in fact be tangible 29 .

Further in vitro biophysical investigation probing the interaction could elucidate evidence into the role of Trip13 in Remdesivir therapy. The functional relevance of such an interaction in the context of viral infected tissue could yield crucial information as to whether its potential off target behavior is tolerable, beneficial or indeed a hindrance to the molecule's efficacy against Sar-CoV-2.

This study has highlighted the power of utilizing unbiased whole proteome approaches and the information that can be rapidly gained from describing proteome wide target engagement of drug molecules.

DMM is a co-founder and shareholder of Pelago and co-inventor of patents originating from PCT/GB2012/050853. All authors are employees of Pelago Bioscience AB, Sweden. The work was carried with internal funding.

Current Status of Epidemiology, Diagnosis, Therapeutics, and Vaccines for Novel Coronavirus Disease 2019 (COVID-19)

Virtual Screening and Repurposing of FDA Approved Drugs against COVID-19 Main Protease

An Update on Current Therapeutic Drugs Treating COVID-19

COVID-19 Treatment by Repurposing Drugs until the Vaccine Is in Sight

The Host Cell ViroCheckpoint: Identification and Pharmacologic Targeting of Novel Mechanistic Determinants of Coronavirus-Mediated Hijacked Cell States. bioRxiv : the preprint server for biology

Pharmacogenomics of Off-Target Adverse Drug Reactions

Chemical Synthesis of Nucleoside 5′-Triphosphate

Addressing the Selectivity and Toxicity of Antiviral Nucleosides

A Review of Its Discovery and Development Leading to Emergency Use Authorization for Treatment of COVID-19

Tracking Cancer Drugs in Living Cells by Thermal Profiling of the

Thermal Profiling Reveals Phenylalanine Hydroxylase as an Off-Target of Panobinostat

Monitoring Drug Target Engagement in Cells and Tissues Using the Cellular Thermal Shift Assay

Identifying Purine Nucleoside Phosphorylase as the Target of Quinine Using Cellular Thermal Shift Assay

Cellular Thermal Shift Assay for the Identification of Drug-Target Interactions in the Plasmodium Falciparum Proteome

Biggest COVID-19 Trial Tests Repurposed Drugs First. Nature biotechnology

Coronavirus Puts Drug Repurposing on the Fast Track

Rapid Repurposing of Drugs for COVID-19. Science (80-. )

Drug Repositioning Is an Alternative for the Treatment of Coronavirus COVID-19

Limma Powers Differential Expression Analyses for RNA-Sequencing and Microarray Studies

Depth Characterization of Staurosporine Induced Proteome Thermal Stability Changes. bioRxiv 2020, 2020.03.13.990606

Metabolism and Pharmacokinetics of the Anti-Hepatitis C Virus Nucleotide Prodrug GS-6620

TRIP13 Regulates DNA Repair Pathway Choice through REV7 Conformational Change

Overexpression Uncovers a Critical Role for TRIP13 in Mitotic Exit

Calcium-Dependent Regulation of Tumour Necrosis Factor-Alpha Receptor Signalling by Copine

Transcriptional Profiling Suggests That Nevirapine and Ritonavir Cause Drug Induced Liver Injury through Distinct Mechanisms in Primary Human Hepatocytes

HIV Protease Inhibitors Stimulate Hepatic Triglyceride Synthesis

The Intracellular Activation of Lamivudine (3TC) and Determination of 2

Triphosphate (DCTP) Pools in the Presence and Absence of Various Drugs in HepG2

TRIP13 Impairs Mitotic Checkpoint Surveillance and Is Associated with Poor Prognosis in Multiple Myeloma