1 Introduction

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused hundreds of millions of cases, and millions of deaths worldwide [1,2,3]. Recently there has been a surge in cases of COVID-19, largely due to SARS-CoV-2 Variants of Concern (VOCs), including the Alpha (B.1.1.7), Beta (B.1.351), Gamma (P.1), Delta (B.1.617.2), and Omicron (B.1.1.529) variants (Fig. 1 and Supplementary Table S1) [4,5,6,7,8,9,10,11]. The VOCs are potentially more contagious, causing increasingly severe infections, evading the host immune system, and/or inducing reinfections [12]. Notably, the Delta variant has shown a 108% increased risk for hospitalization, 235% increased risk for ICU admission, and 133% increased risk for death compared to the original virus [13]. The Omicron variant has more than double the mutations of the Delta variant and spreads twice as quickly as the Delta variant with higher infectivity and transmissibility [14,15,16,17]. Recent data also suggest that the emerging Omicron sub-lineages BA.1 and BA.2 are alarming, due to their increased prevalence worldwide and a higher risk of breakthrough infections than all other VOC lineages caused by substantially reduced vaccine protection and weakened neutralizing antibody responses [16,17,18]. Growing evidence suggests that reinfections and breakthrough infections may promote the spread of VOCs, especially Delta and Omicron variants [16, 19,20,21,22,23].

Fig. 1
figure 1

Genome of SARS-CoV-2 and signature mutations of the emerging variants [5, 6, 24]. A Genomes of wildtype SARS-CoV-2 and VOCs (Alpha variant (i.e., B.1.1.7); Beta variant (i.e., B.1.351); Gamma variant (i.e., P.1); Delta variant (i.e., B.1.617); and Omicron variant (B.1.1.529)) with indication of the most characteristic mutations, including mutations in ORF1ab, Spike, and Nucleocapsid protein. B Global distribution of COVID-19 cases with the currently predominating Delta and Omicron variants. Nextstrain clade label: 21 K indicates Omicron BA.1 lineage; 21L indicates Omicron BA.2 lineage. C Frequencies of major Delta and Omicron sub-lineages from April 2021 to March 2022. B, C Generated using Nextstrain [4, 5]

There are three different groups of proteins that are encoded in the SARS-CoV-2 genome, namely structural proteins (SPs), non-structural proteins (NSPs), and accessory proteins (APs) [24]. Spike, Nucleocapsid, Envelope, and Membrane proteins are the major SPs (Fig. 1) [24]. With respect to SPs, mutations mainly take place on the receptor-binding domain (RBD) in Spike proteins resulting in changes of its binding affinity to the membrane receptor angiotensin-converting enzyme 2 (ACE2) on host cells [25, 26]. NSPs and APs are more conserved than SPs, yet whose mutational changes may facilitate new pathways involving viral replication and release [24].

Currently available vaccines, including mRNA vaccines (e.g., Pfizer-BioNTech and Moderna) and viral vector-based vaccines (e.g., Janssen and AstraZeneca) can attenuate SARS-CoV-2 and VOC entry into host cells by inducing antibodies targeting RBD of spike protein [27]. However, the VOCs can breakthrough by escaping the target site from vaccine, and may subsequently lead to further mutations [28, 29]. Thus, repurposing and developing antiviral drugs to target the most conserved NSPs of VOCs is an immediate priority [12].

Antiviral drugs can target different stages of viral replication cycles such as viral entry and fusion, uncoating, transcription, translation, and virion release [30]. In COVID-19, antiviral therapies such as Remdesivir, neutralizing antiviral antibodies (i.e., Casirivimab/Imdevimab), and plasma therapy were granted emergency use by FDA [30,31,32]. Remdesivir is a nucleoside analogue that was clinically investigated for Ebola virus prior to COVID-19 [30]. As a phosphoramidate, Remdesivir undergoes metabolic transitions to become the active nucleoside triphosphate (NTP; GS-443902) (Fig. 2) [33, 34]. In its metabolically active stage, NTP directly interrupts viral replication by inhibiting viral RNA-dependent RNA Polymerase (RdRp) [33, 35]. Although Remdesivir has exerted some efficacy in treating COVID-19 patients [36], it has shown no decrease in all-cause mortality in patients with severe COVID-19, with most trials demonstrating marginal antiviral benefit [36,37,38,39]. On the other hand, GS-441524 is the parent C’-adenosine analogue of Remdesivir, and theoretically has several pharmacological advantages over Remdesivir in the treatment of SARS-CoV-2 infection [28, 40]. In the present article, we focus our discussion on the potential efficacy of GS-441524 as a therapeutic agent in breakthrough SARS-CoV-2 VOC infections [19,20,21, 23, 41, 42].

Fig. 2
figure 2

Mechanism of action of GS-441524 and Remdesivir following SARS-CoV-2 infection. The diagram depicts the structures and potential disposition of GS-441524 in SARS-CoV-2 infected cytoplasm of a Type II pneumocyte. Remdesivir is converted into GS-441524 extracellularly and into NTP intracellularly, while the only major metabolic pathway of GS-441524 is its conversion into NTP in cells. The bioactivation of Remdesivir involves liver-specific enzymes Carboxylesterase 1 (CES1), Cathepsin A (CTSA), and Histidine Triad Nucleotide Binding Protein 1 (HINT1). NTP disrupts RdRp activities, inhibits viral replication, and prevents the translation and assembly of viral proteins (e.g., Spike protein, Envelope protein, Membrane protein) to ultimately minimize further damages of the human body from the amplifying SARS-CoV-2 virus. Recreated from “Remdesivir: Potential Repurposed Drug Candidate for COVID-19”, by BioRender.com (2021). Retrieved from https://app.biorender.com/biorender-templates

2 The Non-phosphoramidate GS-441524 Could be Superior Over Remdesivir (Contains the Functional Group Phosphoramidate) Against SARS-CoV-2 and VOCs

Although the catalytic NTP that interacts and interferes with RdRp can be derived from both GS-441524 and Remdesivir, the complicated bioactivation pathway, preferential expression of Remdesivir prodrug bioactivating enzymes in the liver, and short half-life of Remdesivir (~ 1 h) render GS-441524 (~ 3–5 h) a better therapeutic candidate [33, 34, 43]. GS-441524 is bioactivated by nucleoside kinases, which are expressed more evenly across all tissues in the body [33]. Considering that it is quite common to observe comorbidities in patients with severe SARS-CoV-2 infections, GS-441524 could be potentially more potent than Remdesivir and could be an antiviral therapeutic option against SARS-CoV-2 variants with greater patient tolerability.

2.1 Nucleoside Kinase is the Only Enzyme Required for GS-441524 Conversion to Active NTP

The molecular basis of bioactivation demonstrates differences in enzymatic requirement between GS-441524 and Remdesivir. GS-441524 requires only nucleoside kinase for bioactivation [33]. In contrast, Remdesivir requires carboxylesterase 1 (CES1), cathepsin A (CTSA), and histidine triad nucleotide-binding protein 1 (HINT1) that are expressed in kidney and liver tissues to be metabolised involving esterase and phosphoramidase pathways for bioactivation [33, 44, 45]. In severe COVID-19 patients with underlying comorbidities, the liver and kidney are likely to malfunction for the conversion of Remdesivir to bioactive NTP against SARS-CoV-2 (Fig. 2).

2.2 Orally Bioavailable Prodrug of GS-441524

In in vitro studies, GS-621763, an orally bioavailable prodrug of GS-441524, has been shown to have low cytotoxicity and a similar EC50 to GS-441524 [46, 47]. Recent studies suggested that administration of GS-621763 is efficacious against SARS-CoV-2 in ferrets and mice [45, 48]. A pharmacokinetic study revealed higher and more consistent plasma concentrations of GS-441524 in ferrets receiving oral GS-621763 compared to those receiving intravenous administration of Remdesivir or GS-441524 [48]. GS-621763 supports the exploration of GS-441524 oral prodrug in the management of breakthrough COVID-19.

2.3 Safety of GS-441524 Over Remdesivir

Data in both cell culture and in animal models indicate that GS-441524 is much less cytotoxic in cells and better tolerated in animals compared to Remdesivir [33, 43, 49, 50]. The latter has shown to induce adverse effects in rhesus macaques (e.g., renal tubular atrophy) and patients (e.g., liver and kidney inflammation) [33, 44]. The non-phosphoramidate GS-441524 may minimise liver and kidney adverse events, enables the drug to be administered in higher doses. The first human study of orally administered GS-441524 for COVID-19 (Trial ID: NCT04859244) in a healthy woman has also shown sustained plasma concentrations and excellent safety profile [51, 52].

2.4 Socio-economic Benefits of GS-4414524 Over Remdesivir

The structural complexity of Remdesivir makes drug production costly and difficult. The minimum production cost of Remdesivir is USD$9.30 for a 10-day treatment course (100 mg two times on Day 1 and 100 mg one time on Days 2–9), which is much more expensive than many other repurposed antiviral drugs for COVID-19 such as fluvoxamine [53,54,55]. In contrast, GS-441524, with its simpler structure (3 functional groups less than Remdesivir) and as the prodrug of Remdesivir during the production procedure and in tissue metabolism, would be significantly less expensive than Remdesivir to produce [33].

3 Combination Therapy of GS-441524

As SARS-CoV-2 variants continue to emerge, there have been increasing interests in developing combination therapies (both virus- and host-targeted) through repurposed drugs, with the goal of better inhibiting viral infections by targeting different mechanistic pathways [56, 57]. As each drug has different yet specific mechanism of action, one advantage of utilizing combination therapies is the potential of achieving drug synergy, offering a treatment that performs better than when administering the individual drugs alone. This also raises the possibility of using lower effective concentrations of each drug in a combination therapy to minimize drug toxicity, side effects, and costs.

3.1 Combination Therapy with Functional Inhibitors of Acid Sphingomyelinase (FIASMA)

FIASMA (e.g., fluoxetine, amiodarone, and imipramine) is a group of psychotropic medications that inhibits the lysosomal enzyme acid sphingomyelinase and regulates the homeostasis of the endolysosomal host–pathogen interface [58, 59]. In in vitro models, FIASMA has been found to efficiently inhibit SARS-CoV-2 entry and propagation via mechanisms such as impairing endolysosomal acidification and inducing cholesterol accumulation within the endosomes [59, 60]. The antiviral potency of FIASMA is further supported by recent clinical studies [55, 58, 61, 62]. For example, in patients with psychiatric disorders hospitalized for severe COVID-19, those receiving FIASMA medications at baseline had significantly reduced risk of intubation or death as compared to those receiving non-FIASMA antidepressants (p < 0.01) [58]. Together, these data suggest the potential antiviral potency of FIASMA in treating SARS-CoV-2 infections.

Recently, many papers revealed the synergistic antiviral potential of therapies that combine remdesivir and/or GS-441524 with FIASMA for treating SARS-CoV-2 variants [56, 57, 59]. For example, in in vitro models of SARS-CoV-2, combined therapy of GS-441524 with fluoxetine showed a more superior viral titer reduction (over 99% reduction) than using fluoxetine (60–70% reduction) or GS-441524 treatment alone (90% reduction) [57]. More importantly, the synergistic antiviral effects are not only observed in the SARS-CoV-2 parental strain, but also in the Alpha and Beta variants, providing support for the applicability of such combination treatments for the currently prevailing Omicron variant infections [57]. Omicron variants have also been found to rely heavily on the endocytic pathways for viral entry, which further supports the use of host endolysosome-directed FIASMA with viral replication-directed remdesivir/GS-441524 for treating SARS-CoV-2 infections [63].

3.2 Combination Therapy with Other Drugs

Additionally, combining remdesivir with many non-FIASMA drugs, including itraconazole, baricitinib, and MEK1/2 Inhibitor ATR-002 (Zapnometinib) have also shown synergistic antiviral effects against SARS-CoV-2 [56, 64, 65]. In polarized Calu‐3 cell culture model, itraconazole-remdesivir combination inhibits the production of infectious SARS‐CoV‐2 particles by over 90% and shows synergistic effects [56]. In a double-blind, randomized, placebo-controlled trial, baricitinib-remdesivir treatment reduces recovery time of patients with COVID-19 and accelerates their improvement in clinical status as compared to using remdesivir alone [64]. Treatment combinations of ATR-002 with remdesivir have also been found to display synergistic antiviral effects [65]. The above drugs are promising targets to be used in conjunction with the direct-acting antiviral remdesivir against SARS-CoV-2 and in vivo and clinical studies are critical in further validating their potency.

3.3 Limitations of Combination Therapy

Patients with severe SARS-CoV-2 infections often hold other comorbidities that may be exacerbated when giving additional repurposed drugs or combination therapies [66]. While both GS-441524 and FIASMA have been reported with little adverse effects on organs, careful evaluations need to be taken about the suitability and safety of GS-441524-FIASMA treatments before making a treatment decision [57, 66, 67].

4 Interactions Between GS-441524 and VOCs

4.1 Inhibition of RdRp (NSP12, 7, 8) Polymerization

RdRp is a multi-unit transcription complex consisting of NSP12, NSP7, and NSP8, which is essential for the replication of the SARS-CoV-2 genome (Table 1) [35, 68]. NTP is a high-affinity substrate for RdRp through interaction with NSP12 [35, 69] and inhibits viral replication through incorporation by RdRp into nascent viral RNA, predominately resulting in chain termination at the i + 3 position with a steric clash of the NTP 1′-CN group with residue R858 of RdRp [70]. The efficient incorporation of NTP into the newly synthesized viral RNA chain is due to the superior selectivity of NTP compared to ATP and other nucleoside analogues [71, 72]. The effective inhibition of RdRp is dependent on the complementarity between NTP and RdRp. When mutations in amino acid residues that interact with NTP are present, the binding properties may be altered to reconstruct the interactions between NTP and RdRp [73].

Table 1 Biological functions and GS-441524 interacting residues of NSP12-7–8, NSP5, NSP3, and NSP14

To evaluate the mutations in amino acid residues of major GS-441524-interacting SARS-CoV-2 NSPs (RdRp, NSP3, NSP5, and NSP14), the online database Nextstrain (https://nextstrain.org/ncov/global) was used (up to March 20th, 2022) [4, 5]. As of March 20th, 2022, no mutation in amino acids of RdRp that interact with GS-441524 had been found in VOCs (Table 2) (https://nextstrain.org/ncov/global) [4, 5]. The more critical RdRp amino acids V557 and D618 that directly affect the affinity of RdRp-NTP binding and stringency of base pairing are also well-preserved [35, 40, 71]. The lack of mutations in the NTP-interacting sites in RdRp suggests the plausible inhibition of RdRp by GS-441524 in VOCs (Alpha, Beta, Gamma, Delta, and Omicron variants) [74].

Table 2 VOC lineages with mutated GS-441524 interacting residues

4.2 Complementary Binding to NSP3 for Inactivation

NSP3 is one of the two major proteases in SARS-CoV-2 that facilitates the cleavage of the polyprotein into NSP1, 2, and 3 (Table 1) [4, 5]. The NSP3 macrodomain is conserved across coronaviruses and generally binds adenosine-5-diphosphoribose (ADP-ribose) [75, 76]. The similar sizes of ADP-ribose (substrate of NSP3) and NTP suggest that NTP may have the potential to interact with NSP3 [75, 76]. Indeed, there is in silico evidence demonstrating the interaction of GS-441524 with the macrodomain of NSP3 through hydrogen bonding and hydrophobic interaction [76, 77]. The binding mode of GS-441524 to the macrodomain is also highly similar to that of the ADP-ribose adenosine moiety [75, 76]. Moreover, GS-441524 is more structurally complementary to the ADP-ribose binding pocket in NSP3 than the adenosine substrate [51]. This suggests a secondary mechanism of inhibition by GS-441524 in addition to bioconversion to the NTP analogue.

Mutations in NSP3 may influence the interactions between GS-441524 and NSP3. As of March 20th, 2022, two different mutations (lineage significance cut-off at 10% global prevalence) in two samples of Omicron variant lineages (A154T: BA.1 and A154N: BA.1.1) were identified in an single amino acid residue of NSP3 that interacts with NTP (Table 2) (https://nextstrain.org/ncov/global) [4, 5]. The A154 residue is not one of the major amino acids highlighted in the in silico studies, and neither A154N nor A154T is a characteristic mutation of the BA.1 or BA.1.1 lineage [75, 76]. We also found several other mutations in GS-441524-NSP3 interacting sites of Delta and Omicron variant cases on Nextstrain; however, all these mutations occurred in patchy cases and with low prevalence (< 0.5–2%) (Table 2 and Supplementary Table S2). Although all the mutated residues exert relatively weak Van der Waals forces to enhance GS-441524-NSP3 interaction, two of them are in sub-lineages with over 10% global prevalence [76, 78]. Moreover, the overall number of samples with mutations in GS-441524-NSP3 interacting residues is significantly higher than in NSP5, RdRp, and NSP14. Therefore, the interactions between GS-441524 and NSP5 of VOCs would likely be affected.

4.3 Inhibition of NSP5

NSP5 mediates the processing of NSPs at 11 cleavage sites (NSP4-11, NSP12-15) (Table 1) [24, 79]. In conjunction with NSP3, the two proteases cleave SARS-CoV-2 encode precursor polyproteins pp1a and pp1b into 16 NSPs to assemble the viral replicase complex [24]. Among crucial NSP5 residues, C145 and H164 exhibit strong hydrogen bonding with Remdesivir [80, 81]. Current data suggest H164 as an essential active site for NSP5 function, which when disrupted, may potentially halt its proteolytic activity [80]. As the structure of the interaction site in Remdesivir is conserved in GS-441524, such interactions between GS-441524 and the NSP5 active site are highly probable [81]. Thus, it is important to investigate whether mutations in the active site are observed that could potentially disrupt GS-441524-NSP5 interactions.

Similar to RdRp, since no mutation of NTP-interacting residues in NSP5 had been found as of March 20th, 2022, the interactions between GS-441524 and NSP5 of VOCs (Alpha, Beta, Gamma, Delta, and Omicron variants) would appear similar to that between GS-441524 and Wuhan wild-type NSP5 (Table 2) (https://nextstrain.org/ncov/global) [4, 5].

4.4 Blockage of the Active Site in NSP14

As a 3′-to-5′ exoribonuclease and a guanine-N7-methyltransferase, NSP14 is a crucial component securing the replication of SARS-CoV-2 (Table 1) [82, 83]. The exoribonuclease domain of NSP14 is critical for viral replication given that mutant exoribonuclease knockout SARS-CoV-2 results in interruption of viral replication [84]. NSP10, the replicative cofactor of NSP14, stabilizes and stimulates enzymatic activities through interaction with exoribonuclease [85]. Furthermore, NSP10 has been proposed to interact with NSP12 to undergo RNA repair processes that may arise during RNA synthesis, indicating the possibility of interactions between NSP10, NSP12, and NSP14 [83]. Studies provided evidence suggesting that NSP14 interacts with NTP, where the cyano group at the 1′-ribose position of NTP fit complementarily with the active site of NSP14 exoribonuclease [40, 83]. The distorted base of NTP is predicted to prevent the proper distances for efficient two-metal ion catalysis, thus disrupting the function of exoribonuclease [40]. Due to the importance of complementarity in ensuring the effect of NTP on NSP14 activities, it is worth investigating the NSP14 mutations in VOCs to assess their influences on the potency of NTP.

As of March 20th, 2022, one mutated NSP14 residue that interacts with GS-441524 is observed on Nextstrain in a sample of the Omicron variant sub-lineage BA.1 (https://nextstrain.org/ncov/global) [4, 5]. The H95Y mutation is not characteristic for the BA.1 lineage, implying that it would not be present in all samples of the sub-lineage (Table 2). However, it is attention-worthy due to the high global prevalence (12%) of the BA.1 sub-lineage (Supplementary Table S2). Therefore, a more conservative conclusion is that the interactions of GS-441524 with NSP14 would be changed in the Omicron variant but would remain relatively conserved across the other VOCs.

5 Preclinical and Clinical Studies Using GS-441524 in SARS-CoV-2 Infection

GS-441524 has shown potency in lowering SARS-CoV-2 replication in in vitro human lung Calu-3 cell infection [34], in mice with an increased viral clearance 2 days post-infection and reduced weight loss [43], and has demonstrated exceptional safety, tolerability, and pharmacokinetics in one human (case report; Cmax: 12·01 μM, surpassing the concentration required to eradicate SARS-CoV-2 in vitro [51, 52]) and several preclinical species [50]. A human study of orally administered GS-441524 for COVID-19 is underway [51, 52]. Taken together, clinical studies of GS-441524 on VOCs are of great interest, given its antiviral potentials.

6 Limitations

The data for the NSP amino acid residues that interact with GS-441524 are based largely on in silico studies, and continual vigorous analysis are needed for further verification [40, 71, 77, 86]. The effect of mutations on the binding affinity of NSPs to GS-441524 remains to be monitored in emerging VOCs. The potential discrepancy may exist between the microscopic effect of the mutated amino acids and their macroscopic influence on NSP structure and protein–protein interactions. Nevertheless, our analysis provides a foundation for future clinical trial testing of GS-441524 in breakthrough VOCs. The promising results of combination therapies in recent literature also suggest that combining virus-directed and host-directed drugs may partially help to counteract the possible reduction in potency of antiviral drugs against the emerging SARS-CoV-2 variants.

7 Conclusion

Given the recent rise of breakthrough SARS-CoV-2 cases and the emerging Alpha, Beta, Gamma, Delta, and Omicron variants that have shown spike protein mutations, there is an urgent need to examine antiviral candidates that could contain these VOCs from escaping vaccines. The major amino acid sites of NSPs (NSP3, 5, 12, and 14) that interact with the parental nucleotide GS-441524 are not altered in the emerging VOCs. As such, we believe that the ready-to-use GS-441524 is a potential antiviral approach against the breakthrough VOCs (Fig. 3).

Fig. 3
figure 3

Graphical Summary of the Advantages of GS-441524 over Remdesivir against VOCs. SPs (E, M, N, S) and APs (ORF3a-10), especially S protein (Spike), are more active in forming mutations than NSPs. All COVID-19 vaccines target the most mutated parts of S protein and eventually lead to breakthrough infections in vaccinated individuals. GS-441524 and Remdesivir are both validified to target RdRp, yet GS-441524 can potentially interact with NSP3, 5, and 14, which are less mutated among VOCs. Compared to Remdesivir, GS-441524 takes fewer steps to be converted into the active metabolite NTP. GS-441524, as a precursor in the chemical synthesis of Remdesivir and with a longer half-life, is considerably more economical with less production procedures involved. NSPs, non-structural proteins; SPs, structural proteins; Aps, accessory proteins; VOCs, variants of concern