Interactions of Anti-COVID-19 Drug Candidates With Hepatic Transporters May Cause Liver Toxicity and Affect Drug Metabolism


 Transporters in the human liver play a major role in the metabolism of endo-and xenobiotics. Apical (canalicular) transporters extrude compounds to the bile, while basolateral hepatocyte transporters promote the uptake or expel various compounds into the venous blood stream. In the present work we have examined the in vitro interactions of some key repurposed drugs advocated to treat COVID-19 (lopinavir, ritonavir, ivermectin, remdesivir and favipiravir), with the relevant key transporters in the hepatocytes. These transporters included the ABCB11/BSEP, ABCC2/MRP2, and MATE1 in the canalicular membrane, as well as ABCC3/MRP3, ABCC4/MRP4, OCT1, OATP1B1, OATP1B3, and NTCP, residing in the basolateral membrane. Lopinavir and ritonavir in low micromolar concentrations inhibited the ABCB11/BSEP and MATE1 exporters, as well as the OATP1B1/1B3 uptake transporters. Ritonavir had a similar inhibitory pattern, also inhibiting OCT1. Remdesivir strongly inhibited ABCC4/MRP4, OATP1B1/1B3, MATE1 and OCT1. Thus, these agents may cause severe drug-drug interactions and drug-induced liver injury. Favipiravir had no significant effect on any of these transporters. Since both general drug metabolism and drug-induced liver toxicity are strongly dependent on the functioning of these transporters, the variable interactions reported here may have important clinical relevance in the drug treatment of this viral disease and the existing co-morbidities.

In the apical (canalicular) membranes of the human hepatocytes the key ATP-dependent endoand xenobiotic extrusion pumps ("ABC efflux transporters") are ABCC2/MRP2, ABCG2/BCRP, and ABCB1/MDR1/ Pgp. Among these multispecific transporters ABCB1/Pgp mostly transports hydrophobic drugs, while both ABCC2 and ABCG2 are also transporting partially detoxified amphiphilic compounds -in fact, ABCC2 is the key transporter for bilirubin conjugates. ABCB11 (bile salt export pump, BSEP), has a restricted function for extruding bile salts into the canaliculi, but numerous drugs inhibit this transporter, leading to cholestasis and drug-induced liver injury (DILI). An important SLC-type transporter in the canalicular membrane, excreting both endogenous and exogenous toxic compounds, is the MATE1 (SLC47A1 -multidrug and toxin extrusion protein 1), mainly transporting cationic drugs, but also some zwitterionic and anionic molecules 11,12 . In the basolateral membranes of the hepatocytes the major ABC-type efflux transporters are ABCC3/MRP3, and ABCC4/MRP4, with variable affinities and specificities for elimination of endo-and xenobiotics. The major function of these efflux pumps is to defend the hepatocytes from any potential overaccumulation of toxic agents and drugs by delivery to the venous blood 13 .
The key SLC-type ("uptake") drug transporters in connection with the sinusoidal blood stream include OCT1/SLC22A1, preferentially supporting the hepatocellular entry of cationic agents, thus working in a coordinated fashion with the apical MATE1 in the bile delivery of such compounds through the hepatocytes. NTCP/SLC10A1 is primarily performing a sodiumdependent uptake of bile acids in the basolateral membrane of hepatocytes, thus NTCP, together with BSEP, is a major player in the enterohepatic circulation of bile acids. However, NTCP also promotes the hepatocyte uptake of several drugs, including statins, and this protein is a recognized risk factor for hepatic drug-drug interactions 14-16 . In the basolateral membrane of the hepatocytes OATP1B1 and OATP1B3 (formerly SLCO1B1 and SLCO1B3) are the key organic anion uptake transporters. Although with somewhat different selectivities and specificities, these transporters are responsible for the hepatocellular uptake and elimination of a wide variety of drugs and toxic compounds from the blood stream. Inhibition of these transporters causes an increase in blood retention and general toxicity of many clinically applied agents [17][18][19][20] and Figure 1.
In the present study we focused on examining repurposed anti-COVID-19 drug interactions with the above transporters by using specific in vitro assays. In the case of the ABC transporters, we employed isolated membranes in which the drug-modulation of the model substrates could be quantitatively examined. In the case of SLC-type transporters we used whole-cell assays, in which the specific modulation of the selected model substrate uptake by the overexpressed transporters could be directly studied. Since in our recently published work, focusing on barrierresident transporters, we have already provided detailed studies for the ABCB1/Pgp and ABCG2/BCRP interactions of these compounds 10 , we do not include these experiments here, while refer to them in the Discussion section.

Results
Hepatic transporter interactions of potential antiviral compounds, applied in anti-COVID clinical trials, have been investigated in in vitro assays developed for specific, individual testing of these transporters. In the assays for ABC type transporters, we have used vesicular transport assays, properly reflecting the potential intracellular interactions of the tested drugs. In the case of SLC-type transporters we have applied transporter overexpressing model cells. The investigated drugs, favipiravir (FAV), ritonavir (RIT), lopinavir (LOP), remdesivir (REM) and ivermectin (IVE), were tested in a wide range, up to 50 µM concentrations, to reveal any potential interactions with the given transporters.
 Vesicular transport assays for ABC transporters MRP2/ABCC2 is the key apical membrane efflux transporter of hepatocytes, responsible for the release of bilirubin and other metabolite-or drug-conjugates into the bile canaliculi. MRP2 drug interactions were characterized here by following the inhibitory effects of drugs on ATPand benzbromarone-sensitive CDCF uptake into inverted membrane vesicles (Fig 2A). We found that IVE strongly decreased the transport rate of the probe substrate at higher than 10 µM concentration, while LOP and FAV had no effect. Interestingly, RIT and REM increased the probe substrate transport rate at high concentrations, possibly by an allosteric effect 21,22 . Since reduced glutathione (GSH) may affect MRP2-dependent substrate transport or drug interactions, we have also examined these transport processes in the presence of 2 mM GSH, but neither the vesicular uptake of CDCF, nor its inhibition by the drugs examined were affected by GSH (data not shown).

MRP3/ABCC3
is mostly a "safety lock" type conjugated metabolite transporter in the basolateral membrane of the hepatocytes transporting its drug substrates into the venous blood 23 .
In this work MRP3 drug interactions were characterized by measuring ATP-and benzbromarone sensitive CDCF uptake into inverted HEK membrane vesicles overexpressing MRP3 (Fig 2B). In the case of MRP3, only RIT inhibited the vesicular uptake function at less than 10 µM concentration, whereas LOP and IVE were effective only at high concentrations (see Table 1), and their inhibitory effects were around 50% even at the highest concentrations applied.

MRP4/ABCC4
is an ABC efflux transporter present in numerous tissues, and has a major role in the liver, blood-brain barrier and kidney to extrude various metabolites and drugs from the cells. Interestingly, while MRP4 is localized to the apical membrane of kidney tubules and brain capillaries, it is a basolateral transporter in the hepatocytes [24][25][26][27][28] . MRP4 drug interactions were characterized by measuring ATP-sensitive DHEAS uptake into inverted HEK membrane vesicles, overexpressing MRP4 ( Fig 2C). We found that most of the drugs examined had no significant inhibitory effect of MRP4-dependent substrate transport (a maximum of 10-30% inhibition was observed by up to 50 µM of RIT, LOP, IVE or FAV), while REM had a strong inhibitory action on this transporter, with an IC 50 of about 2.3 µM ( Figure 2C). BSEP/ABCB11 is the key apical membrane transporter in hepatocytes responsible for bile salt export into the canaliculi 29 . BSEP drug interactions were characterized here by measuring the effects of drugs on ATP-and glyburide sensitive taurocholate (TC) uptake into inverted HEK cell membrane vesicles overexpressing BSEP (Fig. 2D). We found that LOP and RIT strongly inhibited TC transport at low concentrations (see Table 1 for IC 50 values) whereas IVE and REM had smaller but still significant effect at 50 µM. Favipiravir did not modify TC transport.
 Cellular assays for SLC type transporters OATP1B1 and OATP1B3 are important drug and metabolite uptake transporters localized in the basolateral membranes of hepatocytes. These proteins are involved in hepatic bilirubin and bile salt uptake, and are also responsible for the uptake of various statins which may also act as inhibitors of these transporters 18 . In this work we have studied OATP1B1 ( Figure 3A) and OATP1B3 ( Figure 3B) drug inhibition by measuring the fluorescent model substrate, pyranine, into A431 cells overexpressing the related transporters.
As shown in Figure 3, the drug inhibition panel for the two OATPs examined was somewhat different: IVE was less inhibitory for OATP1B1 than for OATP1B3, and RIT, LOP and REM had strong inhibitory effect but with slightly different IC 50 values for both transporters (see Table   1). Favipiravir had no inhibitory effect on these OATPs.

MATE1 (SLC47A1) is localized to the apical membranes of the hepatocytes and has an
important function of the release of endo-and xenobiotics, mostly organic cations (e.g. metformin) into the bile canaliculi. MATE1 is working in a coordinated fashion with the OCT1 basolateral uptake transporter (see below) in the hepatocytes, having a combined role in the metabolism and excretion of organic acid type drugs. MATE1 is an SLC-type proton countertransporter, capable for working in either direction in the plasma membrane 12 , and in this study, we have measured metformin uptake in MDCKII cells overexpressing MATE1. We compared this transport activity to that seen in mock-transfected MDCKII cells. As shown in Figure 4A, LOP, RIT and REM had a strong inhibitory action on this transporter, while IVE and FAV showed practically no inhibition of the MATE1 transport activity (see also Table 1).

OCT1 (SLC22A1)
is the basolateral drug-and metabolite counterpart of the MATE1 transporter in the hepatocytes 11 . In many cases drugs are entering the liver cells from the blood stream through OCT1, and then leaving the cells into the bile by the transport activity of MATE1 (see above). In the present work, we have measured metformin uptake in HEK cells overexpressing OCT1 and compared this transport activity to that seen in mock-transfected HEK cells. As shown in Figure 4B, RIT and REM had a strong inhibitory action on this transporter, while LOP was less inhibitory and IVE and FAV had no significant inhibitory action on OCT1 transport (see also Table 1).
NTCP/SLC10A1 is the major basolateral bile acid uptake transporter in the hepatocytes, working in coordination with the apical bile efflux transporter BSEP 14 . In this work we have measured taurocholate (TC) uptake in HEK cells overexpressing NTCP, and compared this transport activity to that seen in mock-transfected HEK cells. As shown in Figure 4C, the repurposed anti-COVID agents had relatively weak inhibitory action on this uptake transporter: RIT, LOP and REM had IC 50 values at or above 50 µM, and IVE and FAV showed little inhibition and only at high concentrations.

Discussion
In the present work we have studied the transporter interactions of several repurposed, potential anti-COVID-19 agents. Here we focused on key membrane transporters involved in the liver metabolism of endo-and xenobiotics and applied well documented assay techniques to characterize these interactions. We selected lopinavir, ritonavir, ivermectin, remdesivir and favipiravir, as promising repurposed anti-COVID-19 agents for these studies. Some of the data presented here reinforce previously indicated transporter-drug interactions, while the comparative study of the hepatocyte transporters, working in a network to allow combined drug uptake and excretion should help to decipher both potential drug-induced liver injury (DILI) and alterations of general hepatic drug metabolism. In addition, in case of the clinically most promising remdesivir and favipiravir, there are only limited data available for specific transporter interactions.
IVE is a widely applied anti-parasitic drug, especially effective in various tropical diseases. IVE has also been indicated to inhibit the cellular replication of the SARS-CoV-2 virus 1 , while clinical trials showed no convincing anti-viral efficacy in this disease 30,31 . LOP and RIT are efficient HIV protease inhibitors and were shown to also have in vitro efficacy against SARS-CoV-2 replication 6,32 . However, clinical studies performed until now do not support their anti-COVID-19 efficiency, either when applied separately, or in combination with the name Kaletra 5 .
Currently none of these compounds are promoted to be used in COVID-19 outside selected clinical trials 33 .
The effects of both IVE, LOP, and RIT on membrane transporters have already been examined in numerous in vitro and in vivo studies. IVE is known to be strongly neurotoxic, and the ABCB1/Pgp and ABCG2/BCRP efflux transporters are responsible for the reduced human IVE toxicity, by restricting the absorption of this drug in the intestine and protecting the brain passage of IVE through the blood-brain barrier [34][35][36][37] . Several transporters have been reported to be strongly inhibited by these drugs, and as shown in the literature 29,[38][39][40][41][42][43][44] Table 1C). Thus IVE, LOP, and RIT, by seriously impairing ABCB1-and ABCG2-dependent drug extrusion from hepatocytes into the bile canaliculi, may cause severe drug-drug interactions (DDI) and DILI.
As documented in the results section, IVE has relatively little effect on most of the hepatic transporters examined here, while this drug strongly inhibits especially the OATP1B3 uptake transporter. A preferential inhibition of this transporter has already been indicated by 45 , by using different assay systems. We found that LOP has relatively little inhibitory effect on the ABCC2/MRP2, ABCC3/MRP3 or ABCC4/MRP4 efflux transporters, or the NTCP and OCT1 uptake transporters. In contrast, LOP in low micromolar concentrations strongly inhibits the ABCB1/Pgp, ABCG2/BCRP and ABCB11/BSEP and the MATE1 exporter proteins, as well as the OATP1B1 and OATP1B3 uptake transporters (this latter effect has already been noted by 45 . RIT has a similar inhibitory pattern regarding these transporters, although RIT also inhibits OCT1 (see Table 1). As a combined effect on the liver transporters, LOP (and/or RIT) probably has the strongest effect on the canalicular drug and bile salt extrusion, as well as the hepatocellular uptake of bilirubin, bile salts and certain pharmacological agents. OATP1B transported drugs, potentially affected by LOP and RIT, include statins, repaglinide, olmesartan and valsartan. 18,46,47 .
The BSEP/ABCB11 inhibition by LOP found in this study raises the probability that the use of LOP and drugs often applied in treating co-morbidities in COVID-19 (e.g. statins, Cyclosporin A, rifampicin, or glyburide), may result in the accumulation of bile salts in the liver, leading to cholestasis and DILI (see US FDA web site, 29 ). In case of MATE1, verified substrate drugs, the transport of which may be inhibited by LOP include metformin, some antivirals like acyclovir or ganciclovir, and anticancer drugs like oxaliplatin or topotecan 12 .
Since REM is a most promising repurposed drug in treating COVID-19, in this work we carefully assessed potential transporter inhibition by this compound in various assay system. We found that OATP1B1, OATP1B3 and OCT1 uptake transporters, present in the basolateral membrane, and the MATE1 exporter in the canalicular membrane are strongly inhibited by REM.
This combined inhibitory action may significantly affect the uptake and the biliary extrusion of both anionic and cationic drugs, thus significantly affect general drug metabolism.

MRP4 inhibitors include non-steroidal anti-inflammatory and cardiovascular drugs.
It should be noted that while in this work we focused on liver transporters, MRP4 is also highly expressed in the basolateral renal tubular membranes and here this protein has an important role in the renal excretion of uric acid and anionic drugs. In addition, MRP4 in the intestinal epithelia is involved in the oral absorption of various drugs, while in the BBB it has a protective role of the CNS from drugs 27,50 . Thus, REM inhibition of MRP4 may have multiple effects on general uric acid and drug metabolism.
As a summary (also depicted in the visual abstract and detailed in Table 1  For the OATP1B1 and OATP1B3 transporter assays, the interaction between potential anti-COVID-19 agents and OATPs was investigated in a microplate-based indirect assay employing pyranine as described 53

MATE1 (SLC47A1), OCT1 (SLC22A1) and NTCP (SLC10A1) transporter inhibition assays
were performed as follows: Overexpressing cell lines created by lentiviral method were used for these assays described in 54

Data and Statistical Analysis
All experiments were performed in duplicates and repeated in three biological replicates, thus each data point reflects six independent measurements. The group sizes were equal in all experiments. Normalization for the per cent (%) of control values was performed after subtracting the background levels as specified in the assay descriptions above for avoiding the inherent differences of the transporter activities and to properly reflect the inhibitory potential of each drug examined. IC 50 values were calculated by nonlinear regression analysis using GraphPad prism software (version 5.01, GraphPad, La Jolla, CA, US).

Data availability statement
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.          µM LOP inh >40 no effect -inh 7.8 no effect -inh 7.7 RIT inh 17 no effect -inh 3.2 increase -inh 1.4 IVE inh >50 no effect -inh ≥30 inh ≥20 no effect -REM no effect -inh 2.9 inh ≥40 increase -inh 2.3