Antiviral protease inhibitors (PIs) are peptidomimetic molecules that block the active catalytic center of viral proteases and, thereby, prevent the cleavage of viral polyprotein precursors into maturation. They continue to be a key class of antiviral drugs that can be used either as boosters for other classes of antivirals or as major components of current regimens in therapies for the treatment of infections with human immunodeficiency virus (HIV) and/or severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). However, sustained/lifelong treatment with the drugs or drugs combined with other substance(s) often leads to severe hepatic side effects such as lipid abnormalities, insulin resistance, and steatohepatitis. Molecular mechanisms underlying the PI-induced liver injury and potential therapeutic/pharmaceutical solutions are systemically reviewed her in this article.
Drug Names | Molecular Factors and Pathways | Pathological Consequences |
---|---|---|
IDV, LPV, NFV, RTV, SQV | GLUT4, IRS1&2, GLUT2, Insulin Signaling | |
AKT/PKA, PKCε, JNK1 ApoB, C/EBPα, PPARγ, SREBP |
Insulin Resistance, Dyslipidemia | |
ATV, NFV, RTV | ROS, UCP2, CYP450, Nrf2, HO1, GST, SREBP | |
IDV, NFV, RTV, SQV RTV, SQV ATV, IDV |
OCT1 MRP2/ABCC2, OATP1B3 UDPGT1A1, UDPGT1A3, UDPGT1A7 |
Dysfunction of Transporters, Hyperbilirubinemia |
RTV, LPV | FXR, PXR, SREBP, HNF4α, CYP3A4, GLUT2 G0/G1 Arrest, NFκB/Akt Signaling, BAX, BCL2 Caspase 3&8, TNFα, IL-1β, IL-6, IL-10, eNOS |
Lipid Accumulation, Cell Death, Immune Dysfunction |
RTV, Tipranavir IDV, NFV, Cannabinoids, Alcohol, ATV, DRV, LPV, Azoles SQV, RTV, Rifampicin ETV, RTV, Cobicistat, Statins, Telaprevir RTV, DRV, LPV, MPV, NMV, Amoxicillin, Interferon, Ribavirin |
CYP3A | |
CYP3A4 | ||
CYP45014DM | ||
CYP3A4 OATP1B1, P-gp, CYP3A4, CYP3A5 |
Elevation of ALT and AST, Biliary and Hepatic Injuries | |
GST, ACE2, CYP450 | ||
APV, LPV, RTV, DRV, DEX, RDV, EFV, DTG, NFV, Alcohol |
UPR, IRE1, ATF6, PERK, CHOP, ER Stress C/EBPβ, CREBP3, TFE3, Rab Proteins Golgi Stress, SREBP, ACC, FAS, SCD1 |
Cell Death, Inflammation, Steatohepatitis, Liver Fibrosis |
HIV PIs are known to induce organelle stress that, in general, links to the dysregulation of lipid metabolism in the liver [92][93][94][95][96][97][77,114–118]. RTV-boosted LPV or DRV was reported to induce the dilatation of ER and dispersed/fragmented Golgi apparatus [94][98][99][100]. RTV increased the activity of complexes I and IV, with simultaneous uncoupling and the inhibition of complex V, contributing to mitochondrial dysregulation and cell death [101][102][103]. The phosphorylation of eIF2α and activation of ATF4 was increased in the liver of mice treated with NFV [104]. In mouse and human primary hepatocytes, LPV/RTV treatment combined with alcohol-inhibited sarcoplasmic reticulum Ca2+-ATPase (SERCA) expression, which disturbed ER calcium homeostasis, exacerbating ER stress and resulted in excessive liver cell death [94][100]. The three canonical UPR branches, IRE, PERK, and ATF6, were reported as being differentially expressed in hepatocytes in response to RTV-LPV [98]. The ATF6 branch and its downstream factors were inactivated, while the other two branches of UPR, IRE1, represented by the expression of sXbp1 mRNA, and PERK and represented by the expression of CHOP, were upregulated. These observations are interesting because the activation of ATF6 is known to require trafficking from ER to Golgi for proteolytic processing and involves both ER and Golgi. In fact, the co-localization of ATF6 and the Golgi matrix protein GM130 was lower in liver cells treated with PIs than in cells treated with tunicamycin, which induced only ER stress [98]. Golgi stress markers, GCP60 and HSP47, were also increased remarkably in the hepatocytes treated with PIs. The inhibition of ATF6 nuclear translocation and subsequent induction of Golgi stress by the anti-HIV drugs were quite specific as the ER stress-inducing agents, tunicamycin and thapsigagin, did not exert similar effects, and the knockdown of TFE3 by RNA interference worsened PI-induced cell death [98][105].
The inhibitory effects of PIs on ATF6 processing and activation reflect interference with the inter-organellar communication between ER and Golgi. The Golgi is part of the cellular endomembrane system, which receives secretory and membrane proteins from the ER and delivers them to various destinations via the formation of an ER–Golgi interface and bidirectional membrane trafficking between the two organelles. This trafficking requires the coat protein complex I (COPI) and COPII [106][107][108][109]. COPI participates in the retrograde route from Golgi to ER, and COPII participates in the anterograde route from ER to Golgi. The initiation of the retrograde route requires COPI, ADP-ribosylation factor 1 (ARF1), GTPases, and other co-factors [109]. The initiation of the anterograde route requires COPII, secretion-associated RAS-related 1 (SAR1), GTPases, and GTPase-activating proteins (GAPs). Both routes are required for the structural and functional integrity of the two organelles. However, the integrity of the anterograde route depends on the retrograde route, which ensures not only the retrieval of resident proteins from the ER but also facilitates the recycling of lipids and traffic machinery [109][110][111]. In RTV-LPV-treated liver cells, COPII was found to be aggregated, whereas Brefeldin A (BFA), which is known to inhibit the retrograde route [98][105], did not prevent PI-induced COPII aggregation, suggesting that RTV-LPV interfered with COPII-mediated anterograde trafficking. RTV-LPV also inhibited the distribution of Golgi-resident enzyme mannosidase alpha class 2A member 1 (MAN2A1) to Golgi. In parallel to the impaired ER-Golgi trafficking and ER/Golgi stress response, marked Golgi fragmentation, fat accumulation, and cell death were observed in the hepatocytes from mice treated with the PIs, dexamethasone (DEX), and/or remdesivir (RDV) [94][98][105][106]. Moreover, the severity of Golgi fragmentation in response to other PIs, including TFV, EFV, DTG, DRV, APV, and NFV, was well correlated with downstream hepatic injury [98][105].
One specific mechanism for lipid abnormality and lipodystrophy likely results from the off-targeting effects of HIV aspartyl PIs. HIV PIs were reported to block ZMPSTE24 (the human ortholog of STE24 protease) and interfere with the endoproteolytic processing of the mammalian farnesylated proteins, prelamin A, which alter the maturation and stability of lamin A and the nuclear localization of SREBP-1 [112]. The cholesterol content of membranes, cells, and blood can be controlled by the integral membrane zinc metalloprotease ZMPSTE24 which resides in both the ER membrane and the inner nuclear membrane [113]. HIV-PIs, including LPV, RTV, and TPV, could directly block the enzymatic activity of purified Ste24p and the yeast ortholog of ZMPSTE24 [114][115]. Newer HIV PIs, such as DRV, with less capability to induce ER stress, do not inhibit ZMPSTE24 or lead to an accumulation of farnesyl-prelamin A in cells [116]. Recent evidence implicates Ste24 as a key factor in several ER processes, including the UPR, the removal of misfolded proteins from the translocon, and lipogenic abnormalities in the liver, muscle, and adipose tissues [117][118]. In addition, mutations in ZMPSTE24 diminish its activity, giving rise to cell senescence and calcification, excess visceral abdominal fat, and genetic diseases of accelerated aging (progerias) that can be seen in HIV patients under PI treatment [119][120]. Thus, off-targeting the enzymatic activity of ZMPSTE24 and subsequent blocking prelamin A processing likely contribute to lipodystrophy in individuals undergoing HIV-PI treatment.
Another potential host off-target of the HIV PIs is the Ras converting CaaX endopeptidase 1 (RCE1), which is a glutamyl protease that can be classified as a member of the zinc metalloproteinase family [121][122]. RCE1 was initially identified through the total RNA-sequencing of RNA from HepG2 cells treated with RTV and LPV [105]. The levels of the RCE1 protein were inhibited not only in cultured HepG2 but also in primary hepatocytes and in the liver of mice treated with the anti-HIV drugs. Neither Rce1 transcription nor the RCE1 protein level was inhibited by Brefeldin A, which also induced organelle stress in the liver cells, suggesting that RCE1 could be specifically inhibited by the anti-HIV drugs. In addition, knocking down Rce1 with RNA interference increased RTV and LPV-induced cell death and the subsequent expression of Golgi stress response markers, TFE3, HSP47, and GCP60, in both primary hepatocytes and mouse liver, and deteriorated alcohol-induced ALT and fatty liver injury in animal models [105][106]. Further, the HIV PI-induced effects on RCE1 were accompanied by the inhibition of two potential substrates of RCE1, the small GTP binding proteins Rab13 and Rab18. Rab13 and Rab18 are special GTPases that contain the CaaX motif (where C is cysteine, A is an aliphatic amino acid, and X is any amino acid) [122]. Rab proteins are known to play an important role in regulating/coordinating the ER-Golgi traffic, which involves many effectors such as vesicle tethers, SNAREs, membranes, and motor proteins [123][124]. The disruption of the coordination has been shown to cause ER/Golgi stress, excessive lipogenesis, and abnormal lipid turnover [125][126][127]. The HIV PI and alcohol-induced Golgi fragmentation, Golgi stress response, and cell death could be reduced in the primary human hepatocytes overexpressing Rab13 [105][106]. While the direct inhibition of RCE1 enzyme activity by the HIV PIs needs to be tested further, these pieces of evidence support the host protease RCE1 as an off target.
Targeting Insulin Resistance, Cellular Stress and Dyslipidemia with Natural Compounds. Natural compounds have been used as potential therapeutic agents or as dietary supplementation for antiviral PI-induced deleterious effects on the liver. First, thymoquinone extracted from black seed oil has been shown to protect HIV-infected patients against ATV, NFV, or SQV-induced oxidative stress, glucose intolerance, and the impairment of insulin signaling and lipodystrophy [128][129][130][131][132]. Thymoquinone is beneficial for COVID-19 prevention and cure in patients under anti-SARS-CoV-2 treatment [132]. Second, flavonoids and isoflavones are also natural products possessing anti-inflammatory, antioxidant, and anti-apoptosis activities and can be used to alleviate the side effects of HIV PI drugs. Naringin, a grapefruit-derived flavonoid, reversed weight loss, polydipsia, elevated fasting blood glucose, and reduced levels of fasting plasma insulin, the expression of phosphorylated IRS-1 and Akt proteins, and hepatic glucokinase in rats treated with ATV or SQV [133]. Naringin protected against lipid abnormalities in SQV or DRV-treated rat liver slices and human liver hepatocytes [134]. In addition to HIV PI-induced hepatic injury, naringin protected against NRTI-induced mitochondrial toxicities in zidovudine-treated rats by improving antioxidant enzyme activities, reducing ROS-induced mtDNA damage, and increasing the expression of the complex IV protein [135]. Alcoholic extracts of lotus leaves, rich in flavonoids, have the potential to treat dyslipidemia in rats treated with LPV and RTV [136]. Third, the red clover isoflavones, formononetin, and biochanin A were shown to modulate NFκB/pAkt signaling molecules and protect against RTV-induced hepatotoxicity in in vivo animals [137][138][139]. Some flavonoids, such as quercetin, not only protect against drug-induced hepatotoxicity but are also effective as antiparasitic and anti-HIV/SARS-CoV-2 agents [140].
HIV PI-induced organelle stress can also be a therapeutic target for drug development. However, it can be challenging because, in addition to PI drugs, certain viruses, such as HIV, HBV, HCV, and coronavirus modulate ER stress response or preferentially activate the different pathways of UPR [141][142][143][144][145]. For instance, HCV infection activates the ATF6 pathway while blocking the IRE1 pathway and HBV infection, which stimulates both ATF6 and IRE1 signaling but has no effects on PERK signaling [141][143]. Targeting any specific UPR branches could lead to potential side effects on the normal cell functions of other UPR branches and limit the usage of drugs when targeting organelle stress proteins. Despite this challenge, there are a couple of chemical chaperones such as 4-phenylbutyric acid (4-PBA) and tauroursodeoxycholic acid (TUDCA), which can alleviate ER stress by increasing their protein-folding capacity. PBA has a beneficial role in coping with hepatic fat accumulation and lipotoxicity resulting from ER/Golgi stress [146][147]. TUDCA can also alleviate organelle stress stabilizing UPR, reduce oxidative stress and cell death, and decrease inflammation in many in vitro and in vivo models of metabolic syndromes [148][149]. In addition to chemical chaperones, salubrinal, an eIF2α dephosphorylation inhibitor, reduces xenotoxicant-Induced cellular stress and damage [150], which could be useful for the treatment of ER stress–associated liver diseases.
Improving Drug Delivery and Bioavailability with Nanotechnologies. Inefficient drug delivery and poor bioavailability necessitate high drug doses or a combination of drugs with pharmacokinetic enhancers and/or special formulations. For instance, RTV has widely been used as a booster for other antiviral PIs, which often cause severe hepatotoxicity. Improving PI drug pharmaceutical properties for efficient drug delivery or high bioavailability can provide rational solutions to mitigate drug-induced hepatic side effects. There are two potential approaches, the development of prodrugs and the application of nanotechnology for the improvement of pharmacokinetic, delivery, and side effect profiles. Peptide prodrug was reported to improve oral absorption, transport across the intestinal epithelium, mitigate CYP3A4-mediated metabolism and improve solubility profiles of LPV [151][152]. Phosphate prodrugs have been explored to address absorption limited by solubility, and amino acid prodrugs have been shown to improve drug permeability by engaging with active transport mechanisms. For instance, phosphate and amino acid ester prodrugs have improved the oral bioavailability and plasma concentration of ATV by more than fivefold [153][154]. However, the reported prodrugs lack detailed in vivo characterization, and hence, the preclinical or clinical benefits of the prodrugs have yet to be fully determined.
Nanotechnology has provided opportunities to achieve better-targeted PI delivery and enhanced bioavailability. To overcome its intense lipophilicity and extensive metabolism by liver microsomal enzymes CYP3A4, LPV has been prepared with surface-stabilizing nanoparticles, which enhanced bioavailability by more than 3-fold without the coadministration of RTV in a rat model [155]. A children-friendly and flexible solid dosage form of LPV-RTV was also created utilizing novel in situ self-assembly nanoparticles that form granules when in contact with water [156]. PI drug granules were stable under physiological conditions for over 8 h and displayed a nearly 3-fold increase in bioavailability in rats. To effectively reduce the HIV viral load in the brain that often forms an independent viral reservoir resulting in latent infection or debilitating neurological complications, PI drugs were formulated with specific brain-targeting nanocarriers, including polymeric nanoparticles, liposomes, solid lipid nanoparticles, micelles, and macrophage-based nanoparticles. This facilitated drug transport into the brain via endocytic pathways, inhibited the ATP-binding cassette (ABC) transporters expressed at the brain barrier sites, and dramatically increased local bioavailability to the brain [157][158][159][160]. HIV PIs, including ATV, IDV, RTV, and SQV, were successfully delivered across the blood–brain barrier at concentrations that did not cause hepatotoxicity in animal models [158][159][160]. In addition, sustained or prolonged release of PI drugs, including ATV, DRV, and RTV, was achieved in vivo using pH-responsive nanoparticles, nanoparticles of mPEG-PCL (methoxy poly (ethylene glycol)-poly (e-caprolactone)), or nanoparticles encapsulating both hydrophilic and hydrophobic PI drugs [160][161][162][163][164].
Designing Safer Drugs Based on Molecular Mechanisms of PI-induced Hepatotoxicity. The host glutamyl proteases, RCE1 and STE24, belong to the zinc metalloproteinase family [165][166], which are normally not targeted by the HIV-1 aspartyl protease inhibitors. However, recent studies with crystallization and analysis of STE24 revealed that the molecular structure of the STE24 enzyme protein is different from most other zinc metalloprotease members. The internal cavities of the STE24 protease hold several water molecules that enable structural stability and flexibility [167][168]. Such a hydrophilic environment could enhance the tetrahedral coordination of the zinc atoms at the enzyme active site, allowing off-target access by anti-HIV/SARS-CoV-2 protease inhibitors. RCE1 protein may have a molecular structure similar to STE24 as both are CaaX endopeptidases. In addition, the chemical properties of the individual antiviral PIs are different, which makes some PIs more prone to off-target access than others resulting in varying degrees of hepatic side effects. For instance, RTV-induced Golgi fragmentation and injury in the liver cells were severer than LPV, and LPV–induced liver damages were severer than DRV [98][105][106]. The overall chemical structures of RTV and LPV are nearly the same, except that the core region of LPV contains a hydroxy ethylene dipeptide isostere group, whereas RTV holds a longer side isopropyl thiazolyl group for inhibiting CYP3A4 and boosting the circulating concentration of other antiviral PIs [169][170]. On the other hand, DRV has a benzyl group that may hinder off-target access to the host protease. Thus, the next generation of PIs should incorporate fused ring polycyclic ethers and aromatic heterocycles to promote hydrogen bonding interactions with the backbone atoms of aspartyl HIV-1 protease as well as van der Waals interactions with residues in the enzyme internal cavities. DRV was developed with such structural consideration [171] and has been proven to have much less off-target effect than that of LPV or RTV. Another approach for safer PIs is screening a large database of active molecules or designing multiple analogs of current PI drugs through structure-based molecular docking and simulation analysis [172]. An analog of ATV has been screened out with a greater inhibitory capacity on HIV-1 aspartyl protease and less off-target effect/hepatotoxicity than the original ATV drug [173].
The molecular mechanisms underlying PI-induced liver injury involve glucose transporters (e.g., GLUT2 and GLUT4), organic ion transporters (e.g., OATP1B1 and OCT1), drug-metabolizing P450 isoenzymes, efflux transporter P-glycoprotein, lipid transporting ApoB, transcription regulators (e.g., FXR, HNF4α, and PXR), lipogenic regulators (e.g., C/EBPs, PPARγ, and SREBPs), insulin signaling adapter protein IRS1/2, AKT/PKB signaling, bilirubin-conjugating UDPGT, redox regulator Nrf2, ROS, inflammatory cytokines, off-target proteases (e.g., RCE1 and STE24) and their substrates small GTPase Rab proteins that regulate ER-Golgi trafficking, prolong unfolded protein response, activation of CHOP, and increase of hepatocellular apoptosis. In addition to the PIs, other general factors could add layers of complexity to the hepatotoxicity. First, HIV/SARS-CoV2 PIs are metabolized extensively by the liver and have potentially important interactions with other types of antiretroviral agents, alcohol consumption, or substance uses such as cannabinoid and methadone that compete for liver metabolizing enzymes. Second, there are potentially complex interactions of the antiviral protease inhibitors with multimorbidity in an aging population of PLWH who are under life-long ART. Third, PIs could also have interactions with underlying hepatic impairment from ongoing virus infections as both viral infections and anti-viral therapies cause organelle stress, and there is a paradox as to whether ER stress/UPR activation should be manipulated for cell survival, which could either reduce drug hepatotoxicity or favor virus replication. Despite these complications, new generations of antiviral protease inhibitor drugs with efficient delivery capacity, high bioavailability, enhanced affinity to viral proteases, and fewer interactions with host off-targets provide promising pharmaceutical solutions to the PI-associated liver injury.