Naltrexone blocks alcohol-induced effects on kappa-opioid receptors in the plasma membrane

Naltrexone (NTX), a homologue of the opiate antidote naloxone, is an orally active long-acting mu-opioid receptor (MOP) antagonist used in the treatment of opiate dependence. NTX is also found to relieve craving for alcohol and is one of the few FDA-approved drugs for alcohol use disorder (AUD). Reports that NTX blocks the actions of endogenous opioids released by alcohol are not convincing, suggesting that NTX interferes with alcohol actions by affecting opioid receptors. MOP and kappa-opioid receptor (KOP) are structurally related but functionally different. MOP is mainly located in interneurons activated by enkephalins while KOP is located in longer projections activated by dynorphins. While the actions of NTX on MOP are well established, the interaction with KOP and addiction is not well understood. We used sensitive fluorescence-based methods to study the influence of alcohol on KOP and the interaction between KOP and NTX. Here we report that alcohol interacts with KOP and its environment in the plasma membrane. These interactions are affected by NTX and are exerted both on KOP directly and on the plasma membrane (lipid) structures (“off-target”). The actions of NTX are stereospecific. Selective KOP antagonists, recently in early clinical trials for major depressive disorder, block the receptor but do not show the full action profile of NTX. The therapeutic effect of NTX treatment in AUD may be due to direct actions on KOP and the receptor environment.


INTRODUCTION
Moderate consumption of alcohol (i.e., social drinking) gives a sensation of elatedness and relaxation. This is very different from the drive to binge drinking, to get intoxicated. Whereas the euphoric effects of alcohol have been well studied, much less is known about factors driving to alcohol abuse. We have decided to assess the actions of alcohol at the molecular level, using sensitive methods and uorescent molecular markers. A focus is naltrexone (NTX) which is approved by the Food and Drug Administration (FDA) for the treatment of alcohol use disorder (AUD). Recently, NTX has been advocated for the treatment of alcohol misuse as "one of the most underutilized interventions in medicine" [1].
Early experimental studies reporting the potential therapeutic effects of NTX in alcohol dependence [2,3] were followed by clinical studies in human subjects. For example, NTX was found to reduce the feeling of "high" induced by alcohol in alcohol-dependent individuals [4]. However, the response was not universal and it has been suggested that differences in response may be genetically determined. The potential therapeutic effects of NTX for the treatment of AUD eventually led to an FDA approval for this indication, which is signi cant since very few medications for this condition are available. NTX was initially assumed to counteract alcohol-induced release of endogenous opioids acting on the MOP [5,6].
It is commonly assumed that AUD is a consequence of the euphoriant activity of alcohol (positive reinforcement) and of craving, the urge to resume consumption in abstinence (negative reinforcement). It has also been proposed that positive reinforcement is mainly exerted via MOP and negative reinforcement is mainly exerted via kappa-opioid receptors (KOP) [7]. Although this model may seem rational for therapeutic developments, there are principal di culties. In classic binding analysis, NTX is assumed to primarily act on MOP and has lower a nity and activity for KOP. This has been an opening for the clinical use of compounds with overlapping a nities for MOP and KOP such as pentazocine or buprenorphine. Our studies show that NTX has a signi cant in uence on alcohol interactions with KOPan activity that may be particularly relevant in binge drinking. In a series of publications using sensitive imaging technologies, we have observed that alcohol (ethanol, EtOH) in pharmacologically relevant concentrations directly affects glycosylphosphatidylinositol-enriched membrane domains, MOP, and KOP; these effects were largely blocked by NTX [8,9]. A behavioral/neurochemistry analysis in mice observed a KOP supersensitivity and a hypodynamic side of the nucleus accumbens [10].
A valuable asset for drug development is the reported structural characterization of the dynorphin/KOP system. The X-ray structure of KOP with the antagonist JDTic was one of the rst in the opioid receptor family [11]. The dynamics of the interaction between dynorphin and KOP was followed using nuclear magnetic resonance (NMR) [12]. Membrane lipids have been seen as a catalyst for dynorphin -KOP interaction: membrane attraction results in lower energy needed for the ligand to bind to the receptor [13].
These observations are probably also relevant for the NTX-KOP interaction and may be sensitive to the presence of alcohol. To probe the receptor binding sites, uorescent NTX has been used previously [14].
Here we used a newly designed uorescent NTX derivative (fNTX) and a uorescent protein conjugate of KOP to assess the ethanol effects on KOP and KOP-NTX.

Impact of ethanol on the KOP-surrounding environment
Ethanol is known to affect membrane lipid structures [8, 19,20]. The size of nano-scale clusters and receptor density of KOP in the plasma membrane are also affected under EtOH treatment [9]. These studies indicate that EtOH may change the environment surrounding KOP and receptor self-organization in the plasma membrane. To further characterize the receptor environment, Fluorescence Lifetime Imaging Microscopy (FLIM) was carried out in PC12 cells expressing KOP-eGFP using our in-house instrument [21]. KOP-eGFP uorescence was localized in the plasma membrane with and without the treatment of 40 mM EtOH (Fig. 1A, top and middle rows), suggesting EtOH treatment does not induce receptor internalization as observed with agonist ( g. S2A). FLIM curves in the plasma membrane, Speci city of NTX binding to the KOP receptor and in uence of EtOH We have previously observed that NTX enhances formation of a larger nano-scale cluster of KOP [9]. To con rm the NTX effect on KOP in the plasma membrane, FLIM was also performed under the treatment with NTX or NTX + EtOH. NTX decreased FL in a dose-dependent manner, with a IC 50 value 7.6 nM ( Fig. 2A). To con rm that NTX binding triggered a change in the FL, we also tested two related compounds; 1) the inactive optical isomer, (+)-NTX [22] and 2) the KOP-selective agonist, nalfura ne (NFF) [23]. Each compound caused changes in FL, with the IC 50 value 19 µM with (+)-NTX and 0.15 nM with NFF ( g. S1A and S2B). These values coincide with the binding a nity of these compounds to the KOP receptor estimated in other studies [22,24,25] (Fig. 2C, black). Interestingly, the dose-response curve with NTX in presence of 40 mM EtOH was shifted to a higher concentration (Fig. 2B), whereas dose dependency disappeared in the treatment with (+)-NTX and 40 mM EtOH ( g. S1B and S1C). On the other hand, 40 mM EtOH did not change the IC 50 value of NFF ( g. S2C).
To further address whether the higher IC 50 value of FL is caused by the lower binding a nity of NTX in the presence of 40 mM EtOH, a uorescently labeled NTX (fNTX) was synthesized and tested in PC12 cells ( g. S3A). fNTX binding was reduced under the treatment with 40 mM EtOH, and was blocked by a large excess of non-labeled NTX and the KOP-selective antagonist, LY2444296 ( g. S3A, B), suggesting KOP-speci c binding of fNTX in the plasma membrane. Dose-response curves of uorescence intensity ratio (fNTX/KOP-eGFP) were determined to be 15.4 nM (fNTX) and 93 nM (fNTX + 40 mM EtOH) as EC 50 values ( Fig. 2D and 2E). NTX binding a nity to KOP was lower under the 40 mM EtOH, in agreement with our FLIM data ( Fig. 2A-C). Fluorescence Recovery After Photobleaching (FRAP) analysis demonstrated that the immobile fraction of KOP-eGFP was the same with fNTX and NTX (Fig. 2F), suggesting that fNTX retained similar functionality.

NTX reduces EtOH effect on lipid dynamics
To investigate the EtOH effect on lipid and receptor dynamics, we stained PC12 cells expressing KOP-eGFP with a lipid marker Abberior Star Red-labeled DOPE (ASR-DOPE) which clearly stained the plasma membrane and colocalized with KOP-eGFP in control cells (Fig. 3A) and EtOH-treated cells ( g. S4). Dualcolor Fluorescence Correlation Spectroscopy (dcFCS) was carried out in the plasma membrane. Autocorrelation curves recorded with ASR-DOPE (Fig. 3B 1 ) showed two distinct decay curves corresponding to diffusions of ASR-DOPE in the medium and in the plasma membrane, respectively. The counts per particle of ASR-DOPE (CPP; particle brightness in FCS) was not signi cantly changed under any condition (Fig. 3C 1 ), whereas diffusion in the plasma membrane was signi cantly faster in the presence of 40 mM EtOH (Fig. 3C 2 ). Pretreatment with 200 nM NTX signi cantly reduced the EtOH effect ( Fig. 3C 2 ). To address the NTX effect on EtOH-modulated lipid dynamics, we also tested (+)-NTX and the selective KOP antagonist, LY2444296 (LY) [26,27]. As expected from FL measurement ( g. S1A), 200 nM (+)-NTX did not suppress the EtOH effect on lipid dynamics ( g. S5C). To study whether this is an NTX-speci c or general effect of KOP antagonists, PC12 cells were treated with 100 nM LY2444296. There was no suppression of the EtOH effect ( g. S6B). This suggests that NTX modulates speci cally lipid dynamics in a KOP-unrelated pathway. ASR-DOPE diffusion ( g. S7) and membrane uidity using General Polarization (GP) analysis ( g. S8) in untransfected PC12 cells supported the EtOH and NTX effect on lipid dynamics. These results indicate that EtOH enhances lipid uidity and that NTX reduces the EtOH effect in a KOP-independent manner.

Effects of EtOH and NTX on KOP dynamics and dimerization
Receptor homodimerization has key roles for receptor functionality [28][29][30]. We computed counts per particle (CPP) of KOP-eGFP for analysis of its homodimerization. Recorded in untreated cells, CPP was signi cantly higher than eGFP (Fig. 3D 1 ), suggesting that KOP is partially homodimerized. EtOH signi cantly reduced the CPP of KOP-eGFP to the level of eGFP, suggesting that KOP homodimers dissociate to the monomeric state. On the other hand, NTX enhanced KOP homodimerization and suppressed the EtOH effect during combined treatment. This effect was also observed under the treatment with 100 µM (+)-NTX, but not with 200 nM (+)-NTX ( g. S5D). The selective KOP-antagonist, LY2444296 did not interfere with EtOH-induced dissociation of KOP dimer ( g. S6C). We suggest that EtOH and NTX modulate nano-scale cluster formation and number of KOP in the nano-scale cluster, affecting KOP homodimerization.
Our previous data indicate that KOP is non-randomly organized into nano-domains [18]. We also performed total internal re ection microscopy-based FCS (TIR-FCS) to determine the KOP-eGFP con nement in membrane domains by diffusion law analysis ( g. S9D). The intercept of linear regression under 40 mM EtOH was signi cantly reduced, suggesting that EtOH reduces KOP-eGFP con ned to the membrane domains. Diffusion coe cient (DC) was insensitive under each treatment in PC12 cells, whereas signi cant difference in U2OS cells (Fig. 3D 2 and g. S9C). FRAP analysis addressed a reduction of the immobile fraction of KOP-eGFP with EtOH and an increase of that fraction under the treatment with NTX and NTX + EtOH ( g. S10). This may suggest that membrane domain localization of KOP-eGFP contributes to generate immobile receptor in the plasma membrane and EtOH redistributes KOP-eGFP to mobile fraction.

EtOH and NTX modulate calcium signaling via KOPdependent and independent pathways
To address the impact of lipid dynamics modulated by EtOH and NTX for the function of KOP, we performed Ca 2+ imaging in KOP-transfected cells with Fura Red (Fig. 4A). After K + depolarization, uorescence intensity of Ca 2+ -bound Fura Red dramatically increased while uorescence intensity of Ca 2+ -unbound Fura Red was decreased (Fig. 4B 1 ). The Fura Red ratio was calculated as the ratio of Ca 2+bound intensity and Ca 2+ -unbound intensity (Fig. 4B 2 ). EtOH-treated KOP cells showed dose-dependent increment of the Fura Red ratio up to 80 mM EtOH, followed by a sudden drop to untreated levels with over 100 mM EtOH (Fig. 4C 1 ). The IC 50 value was determined to be 13.1 mM EtOH (Fig. 4C 2 ), in good agreement with the IC 50 value in FL (10.3 mM) (Fig. 1C). Cholesterol depletion caused similar changes of the Fura Red ratio as EtOH treatment ( g. S11), suggesting that the EtOH modulated Ca 2+ in ux via the modulation of the environment of KOP-eGFP including deformation of cholesterol-enriched membrane domains. To clarify whether this was a KOP-mediated pathway or not, Ca 2+ imaging was performed in untransfected PC12 cells, showing a higher Fura Red ratio in untreated (black dashed line) compared with EtOH-untreated KOP-expressing cells (Fig. 4C 1 ) and a gradual decrease of Fura Red ratio ( g. S12). This is in good agreement with previous studies that a L-type channel is inhibited and a non-L-type channel is partially inhibited by EtOH [31]. This suggests that KOP inhibits Ca 2+ signaling at a basal level and that the enhancement of Ca 2+ in ux with EtOH treatment is KOP-mediated. To clarify the impact of NTX on Ca 2+ in ux, we pre-treated KOP-expressing cells and untransfected cells with 200 nM NTX. EtOHuntreated KOP-expressing cells showed higher Fura Red ratio and a constant Fura Red ratio even under the over 100 mM EtOH conditions (Fig. 4C 1 ). The constant Fura Red ratio was also con rmed in untransfected cells ( g. S12). These data suggest that NTX affects live cells in two aspects; 1) direct binding to KOP in the plasma membrane and 2) KOP-unmediated modulation of cholesterol-enriched membrane domains which is in good agreement with our FCS data (Fig. 3). To further characterize the two effects of NTX on Ca 2+ in ux, we tested (+)-NTX ( g. S13) and LY2444296 ( g. S14). 200 nM (+)-NTX clearly showed no effect on Ca 2+ in ux ( g. S13, red), while 100 µM (+)-NTX slightly increased Ca 2+ uptake without EtOH treatment, following similar changes under the co-treatment with EtOH ( g. S13, blue). LY2444296 (100 nM) was only found to enhance Ca 2+ in ux as a direct KOP antagonistic effect (Fig S14), via direct binding to KOP, but no KOP-unmediated effect (Fig. S6). Altogether FL data (Figs. 1 and 2) and FCS data (Fig. 3), suggest that EtOH modulates Ca 2+ in ux via changes of membrane environment, in particular deformation of cholesterol-enriched membrane domains. NTX does not only show the KOP-mediated antagonistic effects on Ca 2+ in ux, but also KOP-unmediated effects linking lipid dynamics and the receptor surrounding environment.

DISCUSSION
Alcohol abuse and dependence remain the most signi cant substance abuse problems worldwide and in the US alone, more than 140,000 people are dying from alcohol-related causes annually [32]. There is a choice whether psychotherapy or medication have superiority for treatment, the majority of cases with AUD receive no treatment at all.
Earlier studies of medication in AUD have been subjected to a meta-analysis, showing the NTX and acamprosate are superior to placebo; NTX is particularly effective in heavy drinking and prevention of recurrence [33]. A more recent survey supports the e cacy of NTX. Moreover, effectiveness of extendedrelease NTX medication [34] was recently demonstrated. However, NTX-based prescriptions were primarily given to higher income males with private insurance, leaving women and minorities without such intervention [35]. A certain rise in AUD have been recorded during the covid-19 pandemic [36], prompting an editorial advocating higher rate of NTX prescriptions [1].
Behavioral effects mediated by KOP differ markedly from those of the other opioid receptors, MOP and the delta-opioid receptor (DOP). Kappa-agonists are not self-injected and clinical use is compromised by psychotomimetic side effects. It has even been proposed that the overt euphorigenic effects of MOP and DOP pathways are related to positive reinforcement whereas effects on KOP are balancing and related to the negative reinforcement (craving) [37]; both effects have been related to AUD.
NTX has found a therapeutic niche for the treatment of AUD and is one of the very few medications that can be prescribed for this indication. One characteristic of AUD in humans is that dependent subjects will consume alcohol to relieve or avoid withdrawal symptoms. Similarly, in preclinical studies, alcohol postdependent rats exhibit an alcohol dependence syndrome that is characterized by both somatic and motivational withdrawal symptoms that usually begin after 6 to 8 hours of abstinence and engage in excessive drinking when alcohol is made available again. Using a rat model of alcohol dependence (i.e, chronic intermittent alcohol vapor exposure) we showed that NTX decreased alcohol intake in nondependent rats, regardless of sex and abstinence time point [38]. In postdependent rats, NTX signi cantly decreased the exaggerated alcohol intake only at a delayed abstinence time point (i.e., 6 weeks) in males, whereas it similarly reduced alcohol drinking in females at 8 h, 2 weeks, and 6 weeks abstinence time points. These ndings further support targeting the endogenous opioid system to prevent excessive drinking that is characteristic of AUD, even after long periods of abstinence and further suggest that alcohol dependence causes neuroadaptations [38].
The access of a uorescent derivative of NTX was a priority in the study. While uorescent NTX derivatives have been described before [14], the strategy here was to extend the separation of the uorescent marker to NTX by a longer linker, considering the X-ray analysis data showing that the JDTic binds in a deep pocket [11] and NMR analysis of the KOP/dynorphin interaction [12].
NTX has been reported to have lower a nity for KOP as compared to MOP and may therefore not be considered a KOP antagonist. However, under our conditions, a nity is strong and more in line with previous studies using competition assays in transfected cell cultures that identi ed approximately equal a nity of NTX for KOP and MOP [39]. Constitutive activity and inverse agonism was also observed, which increased after agonist pretreatment [39]. To approach the effects of alcohol on both MOP and KOP at a molecular level, we introduced high-resolution molecular imaging with FCS to follow the dynamics in cell culture of MOP and KOP labeled with uorescent tags. The addition of pharmacologically relevant concentrations of EtOH in uenced their lateral movements in the plasma membrane [40]. Signi cantly, EtOH-induced effects showed differences between MOP and KOP, with higher presence of MOP in the membrane, whereas KOP presence declined. Differences related to EtOHinduced effects were also observed with super-resolution microscopy [8,9].
In our studies, we have also used high-resolution technologies to investigate the effects of EtOH on both MOP and KOP. As expected, NTX blocks the activation of MOP. In cell culture at the ultrastructure level, EtOH affects the distribution of both MOP and KOP (induces the formation of smaller and less occupied MOP and KOP nanodomains) [9]. These studies also revealed that NTX induces formation of larger and more occupied KOP nanodomains and that NTX pretreatment has protective effects against EtOHinduced changes in nano-organization of both receptors [9].
Another approach to the speci city of the studies effects is the use of stereoisomers. The (+) isomer of NTX ((+)-NTX) was available to us. This isomer has its own pharmacologic pro le and shows equipotent binding for MOP and the toll-like receptor TR4, and interacts with opioid (morphine) analgesia [41,42] as well as drug reward [43] The results are clear, the (+) isomer is much less active demonstrating that the NTX effects we observe are mediated by interaction with KOP and not due to off-target effects (Fig. 2).
Based on this work, we have developed a model of direct/indirect actions of NTX to KOP (Fig. 5). In Under the NTX condition, KOP forms larger nano-clusters and KOP is preferably present as homodimers.
NTX also shows antagonistic effect on KOP, following the enhancement of Ca 2+ in ux. These NTX effects are sustained even with EtOH. Both NTX and EtOH affect the cholesterol-enriched membrane domains (KOP nano-clusters) as observed in previous study [9]. With EtOH cholesterol-enriched membrane domains of intermediate size are formed, and KOP is preferentially in homodimers even though to a lesser extent than with NTX alone. The Ca 2+ in ux remains the same in the whole EtOH concentration range, and lipid dynamics does not change. The EtOH-induced enhancement of Ca 2+ in ux was suppressed by NTX. Also, EtOH-induced distortion of Ca 2+ in ux under higher EtOH concentration was inhibited by NTX.
For comparison, we also included a known KOP antagonist a homologue of JNJ-67953964, LY2444296, and a NTX-related agent, nalfura ne, a KOP agonist recently introduced in the treatment of itch (as developed in patients receiving opiates chronically) [44]. We con rm that nalfura ne induces KOP internalization as a KOP agonist and like the natural ligand dynorphin. The observed activity of the LY2444296 con rmed KOP antagonism and as is shown here, NTX and LY2444296 share binding sites.
It can be noted that KOP has been identi ed as one of the strongest genetic linkages in major depressive disorder along with the D2R (dopamine receptor 2) [45]. It is noticeable that there is a striatonigral dynorphin pathway reciprocal to the classic nigrostriatal dopamine pathway [46,47]. The close connection between two potentially relevant neurotransmitter systems may be an indication of a functional relationship. A recent "fast-fail" study of JNJ-67953964 (a.k.a. CERC-501 and LY2456302) in major depressive disorder showed activity in anhedonia [48]. It is now in Phase III trial as Aticaprant.
Another chemically distinct KOP antagonist BTRX-335140, Navacaprant is in Phase II clinical trial in depression [49]. To our knowledge, there are so far no clinical studies of KOP antagonists in AUD.

Concluding remarks
NTX is an analog of naloxone -a well-known opiate antidote in emergencies. NTX was developed as a long-acting MOP antagonist to protect against further intoxication. Its activity in AUD has been hard to conceptualize. As shown here, NTX has potent activity on KOP as an inverse agonist and takes actions on MOP. These receptors are structurally related. However, the neuronal pathways with both receptors are functionally highly different and there is no reason to see them as connected. A formal model with alcohol activating MOP (and euphoria) has been linked to release of endogenous opioids (enkephalins).
The KOP system is seen as the origin of craving (and release of dynorphin peptides). Our model suggests that alcohol acts directly on KOP and that NTX acts as an inverse agonist blocking constitutive and alcohol-induced activity. Binge drinking (to intoxication) which is of medical concern is very different from social alcohol use [50].
The interactions of alcohol and NTX and the opioid receptors, particularly MOP, have been studied with a variety of technologies [6,51]. The current data illustrates that the KOP receptor is also a relevant target.  respectively. Reported binding a nity was referred from [22,24,25]. Binding a nity of (+)-NTX to KOP was estimated from binding a nity of (+)-NTX to MOP and ratio of binding a nity of NTX. Black: