Abstract
There is a high rate of comorbidity between binge eating (BE) and binge drinking (BD) behaviors, suggesting a common neuropathology. Recently, phosphodiesterase 4B (PDE4B) was identified as a pleiotropic gene associated with comorbid alcohol use disorder (AUD) and anorexia nervosa with BE in a genome-wide association study, implicating PDE4B as a potential contributor to shared genetic risk between these disorders. Here, we developed a novel mouse model of comorbid BE and BD in C57BL/6NJ mice in which mice underwent 10 d of BE, followed by 10 d of BD. Females exhibited cross-sensitization from BE to BD, which was apparent on the first day of ethanol access, whereas cross-sensitization emerged in males over multiple trials of BD. Accordingly immunoblotting of the nucleus accumbens tissue indicated a female-selective increase in PDE4B protein expression that was apparent on both the first and last day of BD in mice with a prior BE history. Acute pretreatment with the selective PDE4B inhibitor A33 (1.0 mg/kg) reduced the expression of cross-sensitization to BD in females on Day 1, and this effect was maintained during a 5 d A33 treatment regimen. The 5 d A33 treatment regimen also reduced expression of cross-sensitization to BD that had emerged in males over repeated sessions. These results provide preclinical, functional validation of PDE4B as a driver of food–ethanol cross-sensitization in a novel model for BE and BD comorbidity and support PDE4B in the shared genetic risk for these behavioral pathologies and as a target for pharmacotherapeutic intervention in comorbid AUD and BE behaviors.
Significance Statement
Binge eating (BE) and binge drinking (BD) are highly comorbid pathological behaviors that complicate treatment and increase risk of other psychiatric/somatic conditions and mortality. We face a knowledge gap regarding the biological bases of this comorbidity to inform prognosis and treatment-based recovery. Herein, we developed a mouse model of BE–BD cross-sensitization and showed that (1) prior BE history potentiated subsequent binge ethanol-drinking in both female and male C57BL/6NJ mice, (2) this behavioral cross-sensitization was associated with elevated expression of phosphodiesterase 4B (PDE4B) expression in the nucleus accumbens, and (3) reducing PDE4B activation via systemic pretreatment with a selective inhibitor prevented BE–BD cross-sensitization in mice of both sexes. Findings implicate enhanced PDE4B signaling in the etiology and treatment of comorbid BE and BD behaviors.
Introduction
Alcohol use disorder (AUD) is a prevalent psychiatric illness associated with substantial morbidity and mortality (Murray et al., 2012; Lozano et al., 2012). AUD frequently co-occurs with other psychiatric illnesses, including eating disorders (Grant et al., 2016). While the lifetime prevalence of AUD is higher for men than women (Grant et al., 2016), the gender gap for AUD is closing (Keyes et al., 2019), and women with AUD report a higher prevalence of comorbid psychiatric disorders than men (Zilberman et al., 2003). Eating disorders, including anorexia nervosa, bulimia nervosa, and binge eating (BE) disorder, predominantly occur in women (Hudson et al., 2007; Kessler et al., 2013), with up to 26% of women with an AUD reporting an eating disorder (Munn-Chernoff et al., 2020) and a higher prevalence of an AUD in individuals with an eating disorder than without (Hudson et al., 2007). Notably, the highest prevalence of an AUD is in women who report BE, regardless of the principal eating disorder diagnosis (Bulik et al., 2004; Dansky et al., 2000; Gadalla and Piran, 2007; Munn-Chernoff et al., 2020). AUD and eating disorder comorbidity complicates treatment and is associated with elevated co-occurring psychiatric and somatic conditions, high relapse rates, and longer recovery times and mortality than AUD or eating disorders alone (Gregorowski et al., 2013). Furthermore, treatment for this comorbidity is sorely lacking (National Center on Addiction and Substance Abuse, 2003). Thus, it is imperative that we gain a greater understanding of the mechanism(s) linking AUDs and eating disorders.
Shared genetic risk between AUD and eating disorders is suggested by twin and molecular genetic studies (Munn-Chernoff and Baker, 2016). Furthermore, twin-based genetic correlations range from 0.23 to 0.53 (Munn-Chernoff et al., 2013, 2015; Baker al., 2017), and genome-wide association studies (GWAS) have identified significant loci for problem drinking/AUD (Walters et al., 2018; Kranzler et al., 2019), anorexia nervosa (Watson et al., 2019), and most recently BE disorder (Burstein et al., 2023). Advanced genomic methods have also provided evidence of a significant positive association between AUD and anorexia nervosa, arguing that problem drinking specifically shares genetic risk with anorexia nervosa (Munn-Chernoff et al., 2021). GWAS studies in progress identified the gene encoding the cyclic AMP (cAMP)-hydrolyzing phosphodiesterase 4B (PDE4B) isozyme (Bender and Beavo, 2006; Conti and Beavo, 2007), as a pleiotropic gene associated with AUD–anorexia nervosa comorbidity, most notably with AUD and anorexia nervosa with BE, but not AUD and anorexia nervosa without BE (p = 0.02; Munn-Chernoff, 2019). Such findings argue that AUD and specifically eating disorders with BE share genetic risk that may be related to the function or expression of PDE4B.
Implicating PDE4B in problem drinking, systemic pretreatment with the selective PDE4B inhibitor A33 (Naganuma et al., 2009; Zhang et al., 2017) blunted binge ethanol-drinking under modified drinking-in-the-dark (DID) procedures in C57BL/6NJ mice (Jimenez Chavez et al., 2021). Here, we pursued the functional validation of PDE4B in a mouse model for AUD–eating disorder comorbidity. Modeling BE and binge drinking (BD) comorbidity in rodents is challenging, as few human studies focus on the etiology of comorbidity and how the interrelations between AUD and eating disorders might change during the progression of either disorder to inform model design (i.e., Does problem drinking precede BE or vice versa? Do these disordered behaviors emerge concurrently?; Escrivá-Martínez et al., 2020; Sampedro-Piquero et al., 2022). All three major eating disorders exhibit BE (Casper et al., 1980; Wardle and Beinart, 1981) and problem drinking, and BD are highly genetically correlated (Mallard et al., 2022). Thus, we coupled well-established DID BD procedures (Szumlinski et al., 2019) with a modified version of a validated BE model in C57BL/6NJ mice (Kirkpatrick et al., 2017) and developed a novel mouse model for comorbid BE and BD. In replicate studies, we show that BE cross-sensitizes with BD in a unidirectional manner. This cross-sensitization manifests more quickly in females, is associated with a female-specific elevation in PDE4B expression within the nucleus accumbens, and can be prevented by systemic pretreatment with a selective PDE4B inhibitor. Together, these results functionally validate PDE4B for behavioral cross-sensitization between BE and BD, support PDE4B in the shared genetic risk between AUD and eating disorders with BE, and pose selective PDE4B inhibitors as potential pharmacotherapeutics for comorbid BE and BD behaviors.
Materials and Methods
Subjects
Adult (8–10 weeks of age) female and male C57BL/6NJ (B6NJ) mice (catalog #005304) were obtained from The Jackson Laboratory. Mice were housed in same-sex groups of four and allowed a minimum of 7 d to acclimate to a climate- and humidity-controlled colony room, under a reverse 12 h light/dark cycle (lights off at 1100 h) prior to any experimental procedure. Mice were identified using tail markings. Food and water were available ad libitum, except during behavioral testing (see below). All home cages were lined with sawdust bedding, with nesting materials and an igloo in accordance with vivarium protocols. All experimental procedures were conducted in compliance with The Guide for the Care and Use of Laboratory Animals (2014) and approved by the Institutional Animal Care and Use Committee of the University of California, Santa Barbara (protocol number 829.3).
Modified DID BD procedures
The modified DID procedures to induce BD employed in this report were similar to those described previously by our group (Szumlinski et al., 2019; Fultz et al., 2021; Jimenez Chavez et al., 2021, 2022). Two hours after lights out (i.e., 1300 h), BD (EtOH) and water control (H2O) mice were transferred to individual drinking cages that were lined with sawdust bedding and topped with a wire lid, situated on a free-standing rack within the animal holding room. All mice were allowed to habituate to the drinking cage for 1 h, at which time, one (for Study 1; 40% ethanol, v/v) or two (for Studies 2−6; 20% and 40% ethanol, v/v) sipper tubes containing the ethanol solution(s) or water were placed on the drinking cage. When two sipper tubes were presented, their relative location was randomized daily. Animals were allowed to drink for 2 h (1400–1600 h). At 1600 h, the sipper tubes were removed from the drinking cages, and all mice were then returned to their home cages on the ventilated rack. In all experiments, the ethanol-containing sipper tubes were weighed prior to, and immediately following, each 2 h drinking session to determine the volume consumed. To increase study throughput, the volume of water consumed was not recorded because over 15 years' experience indicates that control mice consume very little water under DID procedures as they are not water-restricted. Furthermore, our BE and BD procedures were either separated by hours (Study 1) or days (Studies 2−6), during which mice had ad libitum access to water in the home cage, thus reducing the potential influence of eating-induced thirst and/or polydipsia on subsequent fluid intake during BD sessions. Indeed, as detailed below, our results for Study 1 argue against our feeding procedures inducing a high thirst state or polydipsia in our mice. However, in the absence of direct measures of water intake, we cannot make conclusive statements about thirst state or polydipsia. The ethanol/water in the bottles was refreshed, and all mice were weighed at least once weekly during the drinking procedures. The recorded body weights of the mice were used to calculate ethanol intake.
BE procedures
Similar to prior work (Babbs et al., 2018), sweetened palatable food (SPF) pellets were purchased from TestDiet (20 mg each; 5TUL diet), which contained a metabolizable energy density of 3.4 kcal/g (21% from protein, 13% from fat, and 67% from carbohydrates; see https://www.labdiet.com/ for more information). Chow control pellets were custom-made by Purina LabDiet to closely resemble the home cage diet in the vivarium (Teklad 18% Protein Diet) and represent the LabDiet 5V75 formulation, which contained a metabolizable energy density of 3.26 kcal/g (calories provided, 23% from protein, 13% from fat, and 64% from carbohydrates; see https://www.testdiet.com/ for more information). The specific parameters of the BE procedures varied across the studies as outlined below but all involved placing the mice into a testing arena (30 × 30 × 30 cm) where a small ceramic bowl was affixed to the floor by plumber's putty that contained SPF or chow or was empty (for “no-food” controls). Mice were allowed to ad libitum explore and eat for 20 min, at which time they were returned to the colony room in their home cages. For all studies, BE procedures occurred during the light phase of the circadian cycle between 0730 and 1030. Food pellets were weighed before and after the 20 min period and food intake was expressed as a function of body weight [grams (g) food consumed/body weight (kg)]. As in prior reports (Kirkpatrick et al., 2017; Babbs et al., 2018), BE was operationally defined as a significant escalation in SPF intake over bouts of SPF presentation, determined by within-subject comparisons.
Experimental designs and statistical approaches
Study 1: concurrent BE–BD model
For Study 1, mice underwent a twice weekly (Tuesday, Thursday) BE procedure over the course of 3 weeks (six BE sessions total). On Monday, Wednesday, and Friday, mice were placed in a distinct testing arena and presented with an empty bowl, in a manner similar to Babbs et al. (2018). Following the end of the 20 min morning session, mice were returned to the colony room in their home cages where they remained until BD procedures were conducted later in the afternoon on each weekday (see Fig. 1A for a procedural timeline). Three groups of mice were compared in Study 1. SPF–EtOH mice were presented with SPF in the mornings and ethanol (EtOH; 20% v/v) in the afternoon. SPF–H2O mice were also presented with SPF in the mornings but water (H2O) in the afternoon to control for the effects of SPF on thirst/fluid consumption. Chow–EtOH mice were presented with chow in the mornings and ethanol in the afternoon to determine the specificity of any SPF effect observed on ethanol intake. With this design, we were also able to determine whether or not food consumption in the mornings impacted daily ethanol intake by comparing the average ethanol intake on “non-eating days” (Monday, Wednesday, Friday) versus “eating” days (Tuesday, Thursday). With this design, the average chow–SPF consumed over the six feeding sessions were analyzed using a sex × group (H2O–chow, H2O–SPF, EtOH–chow, EtOH–SPF) analysis of variance (ANOVA), while the BD data were analyzed using a sex × food type (chow vs SPF) × availability (food present vs absent in the AM) × drinking day ANOVA, with repeated measures on the drinking day factor (10 levels).
Study 1: Concurrent BE–BD procedures do not induce behavioral cross-sensitization. A, Depiction of the procedural timeline employed during concurrent BE and BD procedures in Study 1. B, The amount of food consumed by chow-fed (Chow) and SPF-fed (SPF) female and male mice across the six BE sessions conducted in the A.M. C, Comparison of the change in food intake from the first to the last BE session between female and male mice. D, The amount of ethanol (EtOH) consumed by female mice during the P.M. following A.M. presentation with an empty bowl (Empty) or presentation of a bowl containing either chow (Food/Chow) or SPF (Food/SPF). D′, Comparable data for the male mice in this study. E, Data collapsed across sex for the average ethanol intake over the 6 d that mice were presented with either an empty bowl (Empty) or chow (Chow) prior to the afternoon drinking session, compared with mice that were presented with either an empty bowl or SPF prior to the afternoon drinking session. F, Sex-collapsed time-course of ethanol intake across the 6 BE days during which mice were presented with an empty bowl or either of the foods. G, Sex-collapsed difference in ethanol intake from the first to the last BE session when the food bowl was empty or contained either of the foods. The data represent the means ± SEMs of six mice/group/sex. In panel C, *p < 0.05 versus chow–EtOH (Tukey's post hoc tests). In panels F and G, *p < 0.05 versus empty (Tukey's post hoc tests).
Studies 2 and 3: sequential BE–BD model
The results from Study 1 yielded results that were inconsistent with a model of BE–BD comorbidity (see Study 1 in Results below). Thus, we next determined if a prior BE history would augment subsequent BD using a sequential BE–BD procedure. For this, mice underwent 10 d of 20 min BE procedures, during which one-third of the mice from each sex were presented with SPF, chow, or an empty bowl (empty) on each weekday (Monday to Friday) for 2 consecutive weeks. Upon completion of the 10 d BE phase, “all” mice then underwent 10 d of 2 h BD sessions (Monday to Friday for 2 weeks) during which time they were presented with sipper tubes containing 20 and 40% ethanol to entice higher ethanol intake (see Fig. 2A for procedural timeline of Study 2). With this design, we could examine sex differences in escalation of daily chow–SPF intake, as well as the effects of prior BE history on both initial and repeated bouts of BD. With this design, the BE data were analyzed using a food type (SPF vs chow) × sex × day ANOVA, with repeated measures on the day factor (10 levels), while the BE data were analyzed using a food type (SPF, chow, empty bowl) × sex × day ANOVA, with repeated measures on the drinking day factor (10 levels).
Study 2: Sequential BE–BD procedures induce behavioral cross-sensitization. A, Depiction of the procedural timeline employed during our first study of the effects of sequential BE–BD procedures on behavioral cross-sensitization (Study 2). B, The amount of food consumed by chow-fed (Chow) and SPF-fed (SPF) female and male mice across the 10 BE sessions. For panel B, *p < 0.05 versus chow and +p < 0.05 versus males. C, The amount of ethanol (EtOH) consumed across the 10 BD sessions by female and male chow- and SPF-fed mice, as well as mice presented with an empty bowl during BE procedures (Empty). D, Comparison of the ethanol intake on Day 1 of BD procedures by female and male mice. E, Comparison of the average ethanol intake over the course of the 10 d BD phase of the study, with data collapsed across sex. For panel E, *p < 0.05 versus SPF. The data represent the means ± SEMs of eight mice/group/sex.
As the data from Study 2 indicated a cross-sensitization between BE and BD procedures (see Study 2 in Results below), a replicate study was conducted in distinct cohorts of mice (Study 3; see Fig. 3A for procedural timeline). The procedures employed in Study 3 were similar to those employed in Study 2 with the following exceptions: (1) additional laboratory personnel were recruited to conduct Study 3; (2) the group receiving the empty bowl control condition was eliminated to increase study throughput; and (3) the NAC tissue was collected for immunoblotting following either Day 1 or Day 10 of BD procedures to determine the protein correlates of the cross-sensitization. Study 3 was conducted in two cohorts of 48 mice each, with approximately half of the mice from each cohort killed for tissue collection immediately following Day 1 of BD (n = 4/sex/group/cohort) and the remaining mice killed following Day 10 of BD. Two mice/sex/group/cohort were also transcardially perfused with phosphate-buffered saline (PBS) and 4% paraformaldehyde (PFA). Brains were stored overnight in 2% PFA, followed by storage in PBS. Brains were sectioned (40 µm thick) through the prefrontal cortex (PFC), NAC, and ventral tegmental area (VTA), using a VT1000S vibrating microtome (Leica Biosystems), and sections stored at 4°C in scintillation vials containing PBS. Unfortunately, the PBS storage solution crystallized, damaging the tissue sections and precluding any immunhistochemical analyses. Thus, only immunoblotting data (n = 8 mice/sex/group) are presented. One male mouse was humanely killed during this study due to conjunctivitis. The behavioral data from Study 3 were analyzed using a food type (SPF vs chow) × sex × day ANOVA, with repeated measures on the day factor (10 levels). The details of the statistical approach to the immunoblotting data are detailed below.
Study 3: Replication of cross-sensitization under sequential BE–BD procedures. A, Depiction of the procedural timeline employed during our replicate study of the effects of sequential BE–BD procedures on behavioral cross-sensitization (Study 3). B, The amount of food consumed by chow-fed (chow) and SPF-fed (SPF) female and male mice across the 10 BE sessions. For panel B, *p < 0.05 versus chow (one sex), **p < 0.05 versus chow (both sexes) and +p < 0.05 versus males. C, The amount of ethanol (EtOH) consumed by female and male chow- and SPF-fed mice across the 10 BD sessions. D, Comparison of the ethanol intake on Day 1 of BD procedures by female and male mice. E, Comparison of the average ethanol intake by chow- and SPF-fed mice over the course of the 10 d BD phase of the study, with data collapsed across sex. F, Comparison of the time-course of ethanol intake by chow- and SPF-fed mice over the 10 BD sessions. For panel F, *p < 0.05 versus chow. The data represent the means ± SEMs of 12 mice/group/sex.
Study 4: sequential BD–BE model
Study 4 was conducted to determine whether the cross-sensitization between BE and BD was bidirectional. For this, mice first underwent 10 d of 2 h BD procedures (Monday to Friday for 2 consecutive weeks) during which half of the mice were presented with 20 and 40% ethanol (EtOH), while the other half of the mice were presented with two sipper tubes containing potable tap water (H2O). Then, “all” mice underwent 10 d of BE procedures to examine SPF intake (Monday to Friday for 2 consecutive weeks; see Fig. 4A for a procedural timeline). As described for Study 3 above, approximately half of the Study 4 mice in each 48-mouse cohort were killed immediately following Day 1 or Day 10 of BE for immunoblotting (n = 8/sex/group). The behavioral data from Study 4 were also analyzed using a drinking (H20 vs EtOH) × sex × day ANOVA, with repeated measures on the day factor (10 levels). The details of the statistical approach to the immunoblotting data are detailed below.
Immunoblotting following sequential BE–BD procedures (Study 3). Summary of the immunoblotting results from Study 3 in which female and male mice with a prior 10 d history of chow (C) or SPF (S) consumption underwent 10 d of BD procedures, with representative immunoblots for each protein. The NAC tissue was collected immediately following the 2 h drinking session on Day 1 or Day 10 of the BD procedures. The NAC tissue was assayed for (A) the monomer form of mGlu1, (B) the monomer form of mGlu5, (C) the dimer form of mGlu5, (D) the Group 1 mGlu receptor scaffolding protein Homer2a/b, and (E) PDE4. In addition, the expression and phosphorylation/activational state of CaMKII (F–H), ERK (I–K), and CREB (L–N) were examined. The data represent the means ± SEMs of 9–10 mice/group/sex. *p < 0.05 versus respective for specific comparisons Tukey's tests.
Study 5: acute effect of A33 on initial BE–BD cross-sensitization
The preliminary results of human GWAS studies identified PDE4B as a pleiotropic gene associated with comorbid AUD and anorexia nervosa with BE (Munn-Chernoff, 2019), and PDE4B expression was elevated within the NAC of female BE–BD mice on Days 1 and 10 of BD in Study 3 (see Results below). Thus, Study 5 used the selective PDE4B inhibitor A33 (Naganuma et al., 2009; Zhang et al., 2017; CAS number 121604-72-6; Tocris Bioscience) to determine the functional relevance of PDE4B for the initial behavioral sensitization observed under our sequential BE–BD procedures. For Study 5, female and male mice first underwent our 10 d chow–SPF BE procedure (Monday to Friday × 2 weeks). A33 was suspended in 10% dimethyl sulfoxide (DMSO) at a concentration of 100 mg/ml and bath sonicated for 45 min, followed by dilution with saline to the final concentrations of 1.0 mg/ml (0.1% DMSO), and additional sonication. The 1.0 mg/kg dose was selected for this study as it reduces both ethanol BD (Jimenez Chavez et al., 2021) and oral methamphetamine self-administration in B6NJ mice (Honeywell et al., 2022). The vehicle (VEH) solution consisted of 0.1% DMSO in saline. The Monday following the end of BE procedures, SPF mice were either pretreated intraperitoneally with 1.0 mg/kg of A33 or VEH (injection volume, 10 ml/kg), while chow-fed mice were pretreated with VEH only to provide a drinking baseline upon which to detect cross-sensitization and any effect of A33 pretreatment. Thirty minutes following pretreatment, all mice were presented with 20 and 40% ethanol and allowed to drink for 2 h. Immediately following the single drinking session, brains were extracted and the tissue from the PFC, NAC, and VTA dissected for immunoblotting. The behavioral data from Study 5 were analyzed using sex × group (chow-VEH, SPF-VEH, and SPF-A33) ANOVAs and the statistical approach to the immunoblotting data from Study 5 are detailed below, Immunoblotting and statistical analyses of immunoblotting data.
Study 6: repeated A33 effects on BE–BD cross-sensitization
BE females exhibited behavioral cross-sensitization on Day 1 of BD procedures in Studies 3–5, and the results of Study 5 indicate that this cross-sensitization is completely blocked by A33 pretreatment. In contrast to females, behavioral cross-sensitization manifested in BE males at some point during BD procedures in both Studies 3 and 4. As the precise timing of the manifestation of BE–BD cross-sensitization in males was unpredictable, Study 6 examined the effects of repeated pretreatment with 1.0 mg/kg A33, over the course of the first 5 d of BD procedures, on the expression of cross-sensitization in female and male mice. As in Study 5, 1.0 mg/kg A33 or VEH was administered 30 min prior to each of the first five BD sessions. To examine potential carry-over effects of repeated A33 pretreatment on ethanol intake, the mice in Study 6 then underwent an additional 5 d of BD procedures without any pretreatment. The behavioral data were analyzed using sex × food type (chow vs SPF) × pretreatment (VEH vs A33) ANOVAs.
Immunoblotting and statistical analyses of immunoblotting data
To determine the protein correlates of the cross-sensitization between BE and BD, immunoblotting was conducted on the NAC of mice from Study 3 (sequential BE–BD) and Study 4 (sequential BD–BE) on Days 1 or 10 of the second procedure (i.e., during BD procedures in Study 3 or during BE procedures in Study 4). For these studies, the NAC tissue was immunoblotted for PDE4B, as well as glutamate receptor-related proteins that are reported to drive BD behavior (Cozzoli et al., 2009). PDE4B regulates cAMP levels in the brain, potentially affecting cAMP-dependent protein kinase A (PKA)-dependent phosphorylation of CREB at serine-133 (Ser 133; Naqvi et al., 2014). Thus, we also immunoblotted for p(Ser133)-CREB expression. As CREB can also be phosphorylated at Ser 133 by calcium-/calmodulin-dependent kinase II (CaMKII) and extracellular signal-regulated kinase (ERK; Naqvi et al., 2014), the levels of phosphorylated CaMKII and ERK were also examined to index their relative activity in relation to cross-sensitization in Study 3 and 4. Immunoblotting was also conducted in Study 5 to examine for the effects of A33 pretreatment on both cross-sensitization–related changes in protein expression derived from Study 3 and the activational state of PDE4B. As no selective antibody exists to detect the phosphorylated state of PDE4B, we employed a phospho-PDE4B/C/D antibody in Study 5 to index PDE4B phosphorylation, as conducted in other published reports of PDE4B activity (Komatsu et al., 2013). The large number of mice employed in our experiments necessitated immunoblotting of the tissue separately in females and males, and thus, the results pertaining to cross-sensitization–related changes in protein expression and the effects of A33 upon protein expression are presented separately for female and male subjects.
The procedures for preparing tissue homogenates and detecting and quantifying the expression of our proteins of interest were very similar to those employed in recent work (Huerta Sanchez et al., 2023; Denning et al., 2024). The following rabbit primary antibodies were used: mGlu5 (metabotropic glutamate receptor 5; 1:1,000 dilution; Merck Millipore; AB5675), Homer2a/b (1:1,000 dilution; Synaptic Systems; 160 203), p(Thr286)-CaMKII (1:500 dilution; Cell Signaling Technology; 12716S), PDE4B (1:750 dilution; GeneTex; GTX14612), PDE4B/C/D (MyBioSource.com; MBS9458796), p(Ser133/119/190)-PDE4B/C/D (Bioassay Technology Laboratory; BT-AP12873), p(Ser133)-CREB (1:500 dilution; Invitrogen, PA5-85645), and ERK1/2 (1:1,000 dilution; Invitrogen, MA5-15605). The following mouse primary antibodies were also employed: mGlu1 (1:500 dilution; BD Biosciences; 610965), CaMKIIα (1:1,000 dilution; Merck Millipore; 05–532), CREB (1:500 dilution; Invitrogen, 35–0900), and p(Tyr204)ERK1/2 (1:1,000 dilution; R&D Systems; AF1018). Calnexin expression was employed to control for protein loading and transfer using a rabbit or mouse primary anti-calnexin antibody (for rabbit primary, 1:1,000 dilution; Enzo Life Sciences; ADI-SPA-860; for mouse primary, 1:500 dilution; Enzo Life Sciences; ADI-SPA-860-D).
Following primary antibody incubation, the membranes were washed with Tris-buffered saline with Tween, incubated in either a goat anti-rabbit IRDye 800CW secondary antibody (1:10,000 dilution; LI-COR Biosciences; 925–3221) or a goat anti-mouse IRDye 680RD secondary antibody (1:10,000 dilution; LI-COR Biosciences; 925–68070), and imaged on an Odyssey Infrared Imaging System (LI-COR Biosciences). All immunoblots were visually inspected prior to image analyses, and blots exhibiting any sign of signal anomalies (i.e., misshaped band, high background signal, or air bubbles) were not included in the final statistical analyses of the data. Raw values for each band were measured and first normalized to their corresponding calnexin signal and then to the average value of the control group (e.g., chow-water or chow-VEH) on each gel. The normalized data from Study 3 were then analyzed using a food type (chow vs SPF) × drinking day (1 vs 10) ANOVA, while the normalized data from Study 4 were analyzed using a drink (H2O × EtOH) × eating day (1 vs 10) ANOVA. For Study 5, the effects of A33 pretreatment on protein expression following Day 1 of BD were analyzed using one-way ANOVAs across group (CNT-VEH, SPF-VEH, and SPF-A33).
General statistical approach
All data were analyzed using multivariate ANOVAs, and α was set at 0.05 for all analyses. Significant interactions were deconstructed along the relevant variable and then analyzed using tests for simple main effects or Tukey's HSD tests, as appropriate. Statistical outliers were identified using the ±2.5 IQR rule for extreme outliers and excluded from analyses. All data depicted in figures represent mean ± SEM of the number of animals indicated in parentheses. Analyses were performed using the SPSS v.23 statistical software (IBM).
Results
Study 1: concurrent BE (in AM) and BD (in PM) model
BE in AM
Overall, the females in Study 1 consumed more food than males across the six BE sessions (Fig. 1A,B; sex effect, F(2,30) = 6.817; p = 0.014; n = 6/sex/group). Furthermore, both female and male mice consumed more SPF than chow, and only SPF mice exhibited an escalation of food intake over the 3 week course of testing (Fig. 1A; group effect, F(2,30) = 20.919; p < 0.0001; group × session, F(10,150) = 4.249; p < 0.0001; no interactions with the sex factor, p's > 0.28). The escalation in SPF intake was confirmed by the results of within-subject linear contrasts conducted on the data collapsed across sex (for SPF–EtOH, F(1,11) = 15.89; p = 0.002; for SPF–H2O, F(1,11) = 33.49; p < 0.0001; for chow–EtOH, F(1,11) = 0.055; p = 0.819). However, post hoc analyses failed to indicate any difference in the average SPF intake between SPF–EtOH and SPF–H2O mice (Tukey HSD, SPF–H2O vs SPF–EtOH, p = 0.90; chow–EtOH vs SPF–H20 or SPF–EtOH, p < 0.0001). Furthermore, a comparison of the difference in SPF intake from Day 1 to 6 of BE procedures indicated a comparable escalation of SPF intake between SPF–EtOH and SPF–H2O mice (Fig. 1C; group effect, F(2,35) = 10.152; p < 0.0001; no sex effect or interaction, p's > 0.169; Tukey's HSD, chow–EtOH vs SPF–H2O, p = 0.001; chow–EtOH vs SPF–EtOH, p = 0.002; SPF–H2O vs SPF–EtOH, p = 0.900). These data confirm that SPF intake escalates under an intermittent BE protocol in both female and male B6NJ mice. Moreover, they demonstrate that neither SPF intake nor its escalation is impacted by daily BD procedures.
BD in PM
The drinking patterns exhibited by female and male SPF–EtOH and chow–EtOH mice over the 6 d during which mice underwent BE procedures in the morning versus the 6 d prior to these BE sessions when mice were presented with an empty bowl are presented in Figure 1D for females and Figure 1D′ for males. A sex × food type × availability × drinking day ANOVA failed to indicate any main effect of food type (i.e., SPF vs chow) or sex (p's > 0.441) on daily ethanol intake. However, two 2-way interactions indicated that the morning BE sessions influenced ethanol intake later in the day (food × availability, F(1,20) = 8.005; p = 0.010; availability × drinking day, F(5,100) = 4.678; p = 0.001). To determine how the availability of SPF or chow differentially influenced ethanol intake, the data were collapsed across sex, averaged across the 6 drinking days within each availability condition [i.e., food (SPF or chow) in AM vs empty bowl in AM] and then analyzed along the food-type factor. Paired samples t tests revealed only statistical trends for more and less ethanol intake, respectively, by chow–EtOH (t(11) = 1.869; p = 0.088) and SPF–EtOH mice (t(11) = 1.98; p = 0.073) on days when they engaged in morning eating behavior (Fig. 1E). To deconstruct the availability × drinking day interaction, we collapsed the data across both sex and food type and then analyzed along the day factor, separately for each availability condition. As depicted in Figure 1F, ethanol intake declined across the days when mice underwent morning BE procedures (day effect, F(5,115) = 2.80; p = 0.02), with significantly lower ethanol intake on Days 4 and 6 of drinking (for Day 4, t(23) = 2.846; p = 0.009; for Day 6, t(23) = 2.322; p = 0.029; other days, p's > 0.074). In contrast, ethanol intake was relatively stable across the days when mice were presented with an empty bowl in the morning (day effect, F(5,115) = 2.159; p = 0.063). The influence of food availability on the change in ethanol intake over the course of study was also apparent when the data were expressed as the difference in ethanol intake from the first to the last drinking day (Fig. 1G; t(23) = 3.372; p = 0.003). Taken together, these ethanol data from Study 1 indicate that morning BE sessions lower ethanol intake similarly in both male and female B6NJ mice, regardless of the type of food presented. Such observations are inconsistent with an animal model of BE–BD comorbidity, and thus, we discontinued employing this “concurrent” procedure.
Study 2: sequential BE–BD model
BE phase
As the results of Study 1 failed to support the face validity of a concurrent model of BE and BD, Study 2 employed a sequential model of BE “then” BD to determine whether a prior history of BE would promote subsequent BD in a manner more consistent with a model of comorbidity. For this, mice first underwent 10 d of our BE procedure during which a third of the mice were presented with SPF, a third of the mice were presented with chow and the final third presented with an empty bowl (n = 8/sex/group). As in Study 1, eating procedures or presentation of an empty bowl occurred in individual activity arenas distinct from the home cage. The BE and BD sessions occurred on a daily basis. Specifically, we employed a 10 d BE phase (5 d/week × 2 weeks) followed by 10 d of our BD procedure (also conducted 5 d/week × 2 weeks; Fig. 2A).
In contrast to the data from the twice weekly BE procedure employed in Study 1 (Fig. 1), an analysis of chow versus SPF intake over the 10 consecutive days of BE in Study 2 indicated a significant food type × sex × day interaction (F(9,252) = 3.197; p = 0.0001). As reported previously (Babbs et al., 2018), deconstruction of the significant three-way interaction along the food-type factor indicated higher chow intake overall in males versus females (sex effect, F(1,14) = 12.086; p = 0.004), with no sex difference detected in the pattern of chow intake over the course of study (Fig. 2B; day effect, F(9,126) = 2.997; p = 0.003; sex × day interaction, p = 0.597). However, opposite chow intake, SPF intake was higher overall in females versus males (sex effect, F(1,14) = 5.942; p = 0.029), and a sex difference was detected for the change in SPF intake over the course of the 10 d procedure (Fig. 2B; sex × day interaction, F(9,126) = 5.493; p < 0.00001). Deconstruction of the sex × day interaction along the sex factor indicated a linear increase in SPF intake by females (day effect, F(9,63) = 10.583; p < 0.0001; test for within-subject contrasts for linearity, F(1,7) = 18.158; p = 0.004), while the SPF intake by males rose progressively during the first 5 d of feeding and then stabilized ∼7.5 g/kg/day (day effect, F(9,63) = 2.847; p = 0.007; tests for within-subject contrasts for Order 5 pattern, F(1,7) = 5.852; p = 0.046). Thus, prior to ethanol presentation, both females and males binge-ate SPF, with females exhibiting more robust BE behavior than males.
BD phase
The time-course of ethanol intake by female and male mice-fed SPF, mice-fed chow or presented with an empty bowl are presented in Figure 2C. To determine whether or not a prior BE history altered the initial propensity to binge-drink ethanol, we first compared ethanol intake on Day 1 of the 10 d drinking period. As highlighted in Figure 2D, this analysis revealed a significant sex × food-type interaction (F(2,47) = 4.433; p = 0.018), which reflected higher BD in SPF females, relative to both their chow-fed and empty bowl counterparts (one-way ANOVA, F(2,23) = 8.186; p = 0.0001; Tukey's HSD, SPF vs chow, p = 0.0003; SPF vs empty, p = 0.016; chow vs empty, p = 0.724), with no group differences detected for Day 1 ethanol intake by males (one-way ANOVA, p = 0.578).
An examination of ethanol intake over the entire 10 d drinking period indicated that, on average, females exhibited higher ethanol intake than males (sex effect, F(2,42) = 4.562; p = 0.039) and that, overall, prior BE history increased ethanol consumption (food effect, F(2,42) = 5.744; p = 0.006). Although inspection of Figure 2C suggested higher ethanol intake by SPF females versus chow and empty controls over the 10 d course of drinking, the statistical results did not support a significant three-way interaction (sex × food × day interaction, F(18,378) = 1.561; p = 0.067; other p's > 0.20). Thus, the data for ethanol intake were collapsed across both the day and sex and Tukey's HSD post hoc comparisons revealed significantly higher mean ethanol intake in the SPF mice versus those presented with an empty bowl (p = 0.004), with the ethanol intake of chow mice not different from either empty bowl controls (p = 0.243) or SPF-fed animals (p = 0.197; Fig. 2E). Together, these data from Study 2 indicate that a prior SPF BE history increases the propensity to binge-drink and that females may be more sensitive to this cross-sensitization.
Study 3: replicate of the sequential BE–BD model
To demonstrate the reproducibility of the BE–BD cross-sensitization observed during Study 2, we conducted a replicate experiment (Study 3) ∼6 months later in distinct cohorts of mice. The procedures were very similar between Study 2 and Study 3 (Fig. 3A), with three exceptions: (1) the empty bowl control was not included in Study 3 to increase study throughput; (2) new laboratory personnel were recruited to conduct both the BE and BD phases of Study 3; and (3) the tissue was collected from half of the mice following Day 1 of BD and the other half of the mice following Day 10 of BD to determine the protein correlates of the cross-sensitization. As such, Study 3 commenced with a sample size of n = 24/sex/group, with a final sample size of n = 12/sex/group. One male slated for the Day 1 chow-fed condition exhibited malocclusion and was killed prior to the start of Study 4, resulting in an initial sample size of n = 23 for this group.
BE phase
The time-course of chow and SPF intake over the 10 d BE period is presented in Figure 3B. As observed in Study 2, we detected a significant food type × sex × day interaction (F(9,810) = 1.90; p = 0.049) and deconstruction of this interaction along the food-type factor replicated higher chow intake overall in males versus females (sex effect, F(1,44) = 7.591; p = 0.009), with no sex difference detected in the pattern of chow intake over the course of study (Fig. 3B; day effect, F(9,396) = 2.411; p = 0.011; sex × day interaction, p = 0.282). Also replicating Study 2, a sex difference was detected in the escalation of SPF intake over the 10 d of BE, with no overall sex difference in SPF intake, but a significant sex × day interaction (Fig. 3B; sex effect, p = 0.772; sex × day, F(9,414) = 6.613; p < 0.0001). Deconstruction of the sex × day interaction along the sex factor indicated a linear increase in SPF intake by females over the 10 d course of BE (day effect, F(9,207) = 33.428; p < 0.0001; test for within-subject contrasts for linearity, F(1,23) = 65.907; p < 0.0001). However, in this study, the SPF intake by males rose progressively until the seventh day of BE and then stabilized ∼10.0 g/kg/day (day effect, F(9,207) = 29.520; p < 0.0001; test for within-subject contrasts for linearity, F(1,7) = 74.040; p < 0.0001). Thus, prior to ethanol presentation, both the males and females in Study 3 binge-ate SPF, although the sex difference in the escalation of SPF intake was less robust than that observed in Study 2.
BD phase
The time-course of ethanol intake by the female and male SPF- and chow-fed mice that underwent the entire 10 d drinking procedure are presented in Figure 3C. An analysis of Day 1 ethanol intake by all 95 mice revealed a replication of the female-selective potentiation of initial ethanol intake by a prior history of BE Figure 3D (sex × food type, F(1,94) = 6.935; p = 0.010; for females, t(46) = 2.882; p = 0.006; for males, t(45) = 0.691; p = 0.493).
An examination of the data from the mice that underwent the entire 10 d drinking period (n = 12/sex/group) replicated higher ethanol intake in females versus males (Fig. 3C, left vs right; sex effect, F(1,44) = 15.229; p < 0.0001; sex × drinking day, F(9,396) = 3.759; p < 0.0001). For this dataset, the effect of prior BE on the average ethanol intake was not statistically significant (Fig. 3E; food effect, p = 0.061). However, prior BE history did alter the time-course of ethanol consumption in both male and female mice (food type × drinking day, F(9,396) = 2.732; p = 0.004; three-way interaction, p = 0.774), and this interaction reflected higher ethanol intake during the second week of BD by SPF- versus chow-fed mice (Fig. 2F; tests for simple main effects: Days 1–6, 9, p's > 0.05; Days 7,8,10, p's < 0.05). Furthermore, tests for within-subject contrasts conducted on the ethanol intake over the 10 d course of drinking indicated a linear increase in ethanol intake for male and female SPF-fed mice (F(1,23) = 23.058; p < 0.001), while that for chow-fed animals exhibited a quadratic pattern of intake (F(1,23) = 13.147; p = 0.001). These data from Study 3 show that the cross-sensitization between BE and BD observed in B6NJ mice is replicable across time and across experimenters, strengthening the validity of this model.
Immunoblotting in NAC of the BE–BD model
To determine the protein correlates of the cross-sensitization under our sequential BE–BD model, the NAC tissue was collected immediately following the 1st and 10th ethanol-drinking session (n = 8/sex/group). Due to some technical issues during immunoblotting (e.g.,the presence of air bubble or high background), the final samples sizes ranged from n = 6–8/sex/group as indicated by the individual data points in Figure 4. Based on the behavioral results indicating a female-selective cross–sensitization on Day 1 of drinking, we anticipated that SPF females would exhibit higher NAC expression of PDE4B and perhaps known drivers of BD (e.g., elevated mGlu1, mGlu5, and Homer2; Cozzoli et al., 2009, 2012, 2014; Goulding et al., 2011) than their chow controls, along with corresponding changes in the activational state of their downstream effectors in this region. As cross-sensitization emerged in males during repeated bouts of drinking, we predicted that the protein profile of SPF males by the end of the 10 d BD period would be similar to that expressed by Day 1 SPF females (i.e., higher PDE4B, Group1 mGlu receptors, Homer2, and intracellular signaling) and that the protein profile of SPF females could persist.
PDE4B. Overall, SPF females exhibited elevated PDE4B expression, relative to chow female controls (Fig. 4A; food effect, F(1,30) = 13.610; p = 0.001; other p's > 0.238). In contrast, males exhibited a nonsignificant reduction in PDE4B expression over the course of BD (Fig. 4A; drinking day effect, p = 0.065; other p's > 0.445). This indicated that both the initial and persistent BE–BD cross-sensitization observed in SPF females was associated with elevated PDE4B expression within the NAC, whereas the emergent BE–BD cross-sensitization observed in SPF males did not relate to increased PDE4B levels in this region.
Group1 mGluRs and Homer Proteins. As illustrated in Figure 4C, we detected no group differences in the expression of the mGlu1 monomer in the NAC of either female (food × drinking day ANOVA, all p's > 0.185) or male mice on Day 1 or 10 of BD (food × drinking day ANOVA, all p's > 0.135). However, a significant interaction was detected with respect to NAC expression of the mGlu5 monomer in female mice (F(1,29) = 7.698; p = 0.010), which reflected lowered mGlu5 monomer expression in SPF versus chow females following Day 1 of BD (t(13) = 4.154; p = 0.001), with no difference observed between feeding groups following the 10th day of BD (Fig. 4D; t test; p = 0.863). Overall, female mice with a history of SPF intake exhibited elevated levels of the mGlu5 dimer (Fig. 4E; food effect, F(1,30) = 5.001; p = 0.034; other p's > 0.540), as well as one of its major scaffolding proteins Homer2a/b (Fig. 3B; food effect, F(1,30) = 11.613; p = 0.002; other p's > 0.175). In contrast, males exhibited no differences in the expression of the mGlu5 monomer (Fig. 4C; food × drinking day ANOVA, all p's > 0.519), dimer (Fig. 4E; all p's > 0.438), or Homer2 (Fig. 4B; all p's > 0.346). These results indicate that the initial and persistent BE–BD cross-sensitization observed in SPF females is associated with increased expression of the active, dimer, form of mGlu5, and its major scaffolding protein Homer2 in the NAC, while in contrast to our hypothesis, we did not observe this protein profile in SPF males following the emergence of cross-sensitization to BD.
Kinase Activation. Female mice failed to exhibit group differences in the NAC expression or activational state of any of the kinases examined (Fig. 4F–N, left panels; food × drinking day ANOVAs, for CaMKII, all p's > 0.266; for p-CaMKII, all p's > 0.231; for p-CaMKII–CaMKII ratio: all p's > 0.501; for ERK, all p’s > 0.599; for p-ERK, all p's > 0.263; for p-ERK–ERK ratio, all p's > 0.648; for CREB, all p's > 0.507; for p-CREB, all p's > 0.459; for p-CREB–CREB ratio, all p's > 0.680). Similarly, we did not detect group differences in kinase expression or activity in males (Fig. 4F–N, right panels; food × drinking day ANOVAs, for CaMKII, all p's > 0.211; for p-CaMKII, all p's > 0.427; for p-CaMKII–CaMKII ratio, all p's > 0.786; for ERK, all p's > 0.123; for p-ERK, all p's > 0.482; for p-ERK–ERK ratio, all p's > 0.218; for CREB, all p's > 0.346; for p-CREB, all p's > 0.259; for p-CREB–CREB ratio, all p's > 0.622). These results do not support a link between the activational state of ERK, CaMKII, or CREB within the NAC and the manifestation of BE–BD cross-sensitization in either male or female mice.
Study 4: sequential BD–BE model
As the results of Studies 2 and 3 indicated BE–BD cross-sensitization, Study 4 examined whether this cross-sensitization was bidirectional. For this, we employed an opposite sequential model in which half the mice first underwent BD procedures for 10 d, while the other half of the mice were presented with water. Then all mice underwent SPF BE procedures (Fig. 5A). In Study 4, half of the mice were killed following Day 1 of BE procedures to examine for biochemical correlates of any initial cross-sensitization. The remainder of the mice (n = 12/sex/group) continued BE procedures for a total of 10 d to examine for the effects of a prior BD history on the escalation of SPF intake, and the tissue was collected immediately following the 10th day of BE procedures.
Study 4: The test for cross-sensitization under sequential BD–BE procedures. A, Depiction of the procedural timeline employed during our study of the effects of sequential BD–BE procedures on behavioral cross-sensitization (Study 4). B, The amount of ethanol (EtOH) consumed by female and male mice across the 10 BD sessions. C, The amount of SPF consumed by female and male water-drinking (H2O) and ethanol-drinking (EtOH) mice on Day 1 of BE procedures. D, Comparison of the amount of SPF consumed by water- versus ethanol-drinking mice of each sex over the course of the 10 d BE phase of the study. The data represent the means ± SEMs of 12 mice/group/sex. +p < 0.05 versus female (sex difference).
BD phase
The time-course of ethanol intake over the 10 d of BD is presented in Figure 5B. Overall, females tended to consume more ethanol than males, but this sex difference was not statistically significant (sex effect, F(1,46) = 3.55; p = 0.066), and there was no sex difference in the pattern of ethanol intake over this phase of the study (day effect, F(9,414) = 15.674; p < 0.0001; sex × day, p = 0.876).
BE phase
As a female-selective cross–sensitization between BE and BD was detected during the initial drinking session in Studies 2 and 3, we first compared the effects of a history of BD on initial SPF intake. When all 96 mice were included in the analysis, we detected no effects of either sex or drinking condition on Day 1 SPF intake (Fig. 5C; sex × drinking ANOVA, all p's > 0.68; n = 24/sex/group), with all mice consuming ∼5 g/kg SPF in a manner consistent with the data from the ethanol-naive mice in Study 2 (Fig. 1B). We next examined the effects of prior BD procedures on the escalation of SPF intake over the course of the 10 d BE phase of the study. As illustrated in Figure 5D, we replicated the sex differences in SPF intake (sex effect, F(1,43) = 11.091; p = 0.002) and its escalation with repeated opportunities to eat (sex × day, F(9387) = 6.638; p < 0.0001). However, we detected no effect of a prior BD upon SPF intake (drinking effect and interactions, p's > 0.17). Thus, the data were collapsed across the drinking condition for post hoc analyses. Deconstruction of the sex × day interaction along the sex factor indicated a linear increase in SPF intake by females (day effect, F(9,198) = 17.789; p < 0.0001; test for within-subject contrasts for linearity, F(1,22) = 39.383; p < 0.001), as well as males (day effect, F(9207) = 13.110; p < 0.0001; tests for within-subject contrasts for linearity, F(1,23) = 44.338; p < 0.0001). However, the magnitude of the escalation was greater in females than males as indicated by a comparison of difference scores for SPF intake (Day 10 to Day 1, for males, 3.13 ± 0.817 g/kg; for females, 8.11 ± 1.60; t(45) = 2.816; p = 0.007). These data from Study 4 clearly indicate that the BE–BD cross-sensitization observed in Studies 2 and 3 is not bidirectional.
Immunoblotting in the NAC of the BD–BE model
As the combined experimental histories of the mice in Study 3 (BD–BE) and Study 4 (BE–BD) were comparable (i.e., 10 total days of eating + 10 total days of drinking), but the mice in Study 4 failed to express behavioral cross-sensitization, we conducted comparable immunoblotting procedures on the NAC tissue from the mice in Study 4 to examine for the specificity of protein changes for the cross-sensitized state. For Study 4, the NAC tissue was collected immediately following the 1st and 10th BE session (n = 8/sex/group) in a manner consistent with tissue collection following BD in Study 3. For Study 4, we predicted that the protein profiles of BD–BE mice would be distinct from those observed in BE–BD cross-sensitized mice. Furthermore, we predicted that we would detect water–ethanol differences in NAC protein expression that are like those observed in our prior immunoblotting studies conducted in BD mice (Cozzoli et al., 2009, 2012, 2014; Lee et al., 2016, 2017) but that these water–ethanol differences would likely not be affected by subsequent BE experience as no behavioral cross-sensitization was observed.
PDE4B. In females, a prior BD history increased (drink effect, F(1,29) = 9.473; p = 0.005), whereas 10 d of BE procedures reduced (eating day, F(1,29) = 10.042; p = 0.004), PDE4B expression within the NAC, with no interaction detected between these factors (Fig. 6A, left; interaction, p = 0.481). In contrast, we detected no significant group differences in NAC PDE4B expression in male subjects (Fig. 6A, right; drink × eating day ANOVA, all p's > 0.087). These data indicate that BD history elevates PDE4B expression within the NAC selectively in females and that this effect can persist for at ∼2 weeks following drinking cessation. These data also provide the first evidence that BE in females induces a decrease in NAC PDE4B expression.
Immunoblotting following sequential BD–BE procedures (Study 4). Summary of the immunoblotting results from Study 3 in which female and male mice with a prior 10 d history of water (W) or ethanol (E) consumption underwent 10 d of BE procedures, with representative immunoblots for each protein. The NAC tissue was collected immediately following the 20 min eating session on Day 1 or Day 10 of the BE procedures. The NAC tissue was assayed for (A) the monomer form of mGlu1, (B) the monomer form of mGlu5, (C) the dimer form of mGlu5, (D) the Group 1 mGlu receptor scaffolding protein Homer2a/b, and (E) PDE4. In addition, the expression and phosphorylation/activational state of CaMKII (F–H), ERK (I–K), and CREB (L–N) were examined. The data represent the means ± SEMs of 9–10 mice/group/sex. *p < 0.05 versus respective for specific comparisons Tukey’s tests).
Group1 mGluRs and Homer Proteins. In female mice, 10 d of BE procedures lowered the expression of Homer2 (Fig. 6B), as well as the monomer forms of mGlu1 (Fig. 6C) and mGlu5 (Fig. 6D) and within the NAC (eating day effects, for Homer2, F(1,31) = 21.003; p = 0.002; for mGlu1, F(1,31) = 7.255; p = 0.0512; for mGlu5, F(1,31) = 18.977; p < 0.001). To our surprise, we failed to detect any effects of prior BD history on the expression of these proteins in female mice as indicated by no water–ethanol differences (drink effect and interaction, for Homer2, p's > 0.367; for mGlu1, p's > 0.278; for mGlu5 monomer, p's > 0.436). In males, 10 d of BE procedures also lowered mGlu1 and mGlu5 monomer expression, irrespective of BD history (Fig. 6C,D); however, the mGlu1 monomer effect was smaller in males and did not reach statistical significance (eating day effect, for mGlu1 p = 0.07; other p's > 0.311; for mGlu5 monomer, F(1,29) = 8.901; p = 0.006; other p's > 0.190). Ten days of BE also significantly lowered Homer2 expression in males (Fig. 6B; eating day effect, F(1,29) = 18.565; p < 0.001; other p's > 0.370). As indicated by the results of the statistical analyses, we also failed to detect any effect of prior BD history on the expression of these proteins in males. In contrast, no group differences were detected with respect to the NAC expression of the mGlu5 dimer for either sex (Fig. 6E; drink × eating day ANOVA: for females, all p's > 0.721; for males, all p's > 0.072). Thus, a 10 d BE history reduces the inactive, monomer, forms of both mGlu1 and mGlu5 within the NAC, and their Homer2 scaffolding protein, without influencing the expression of the active, dimer form of mGlu5. Thus, contrary to our hypothesis, our prior BD procedures did not elicit an enduring increase glutamate-related protein expression within the NAC of either male or female mice, and a 10 d history of SPF intake lowered, rather than increased, glutamate-related protein expression within the NAC in a sex-independent manner.
Kinase Activation. Although no group differences were detected for total CaMKII expression within the NAC of females mice (drink × eating day ANOVA, all p's > 0.228), total CaMKII expression increased in the NAC of males over the 10 d BE period (Fig. 6F; eating day, F(1,30) = 11.130; p = 0.002; other p's > 0.321). In contrast, BE reduced p-CaMKII expression in both females and males (Fig. 6G; eating day, for females, F(1,29) = 11.510; p = 0.002; for males, F(1,30) = 26.128; p < 0.0001). Also, p-CaMKII expression was overall lower in females with a prior history of BD (drink effect, F(1,29) = 7.550; p = 0.011; interaction, p = 0.326), an effect not apparent in males (drink effect and interaction, p's > 0.487). An analysis of the relative p-CaMKII expression indicated a pattern of effects that were similar to those observed for total p-CaMKII levels (Fig. 6H; eating day effect, for females, F(1,29) = 7.830; p = 0.010; for males, F(1,31) = 31.936; p < 0.0001; drink effect, for females, F(1,29) = 4.234; p = 0.050; for males, p = 0.999; interaction, for both males and females, p's > 0.865).
Females exhibited no significant group differences in total ERK expression (Fig. 6I; drink × eating day ANOVA, all p's > 0.632), while p-ERK expression was significantly lower on Day 10 versus Day 1 of BE procedures (Fig. 6J; eating day effect, F(1,31) = 19.796; p < 0.001; other p's > 0.213). Examination of their relative expression failed to indicate any change in the activational state of ERK in the NAC of female subjects (Fig. 6K; drinking × eating day ANOVA, all p's > 0.355). In contrast to females, prior BD history and the 10 d BE procedure both elevated total ERK expression in males (Fig. 6I; drink effect, F(1,30) = 4.753; p = 0.038; eating day effect, F(1,30) = 22.968; p < 0.001; interaction, p = 0.139), while neither variable influenced NAC levels of p-ERK in males (Fig. 6J; drink × eating day ANOVA, all p's > 0.325). Consequently, an examination of the relative p-ERK expression yielded an overall effect of BE in male mice (eating day effect, F(1,31) = 4.978; p = 0.034), with a less robust effect of prior BD history observed (Fig. 6K; drink effect, p = 0.057; interaction, p = 0.147).
BE procedures also reduced total CREB expression in both females and males (Fig. 6L; eating day effects, for females, F(1,31) = 32.216; p < 0.0001; for males, F(1,30) = 36.740; p < 0.0001), with no effect of prior BD history detected (drink effect and interaction, for females, p's > 0.073; for males, p's > 0.128). BE procedures reduced p-CREB expression to a lesser extent in females than in males (Fig. 6M; eating day effect, for females, p = 0.052; for males, F(1,29) = 17.182; p < 0.0001), with no evidence of any effect of prior BD history in either sex (drink effect and interaction, for females, p's > 0.196; for males, p's > 0.510). When the p-CREB:t-CREB ratio was considered, our 10 d BE procedures increased this index of CREB activation in both females and males, with a larger effect size detected for females (Fig. 6N; eating day effect, for females, F(1,29) = 11.163; p = 0.003; for males, F(1,30) = 5.773; p = 0.023). Although the results of the ANOVA failed to indicate any effect of prior BD history on this measure (drinking effect and interaction, for females, p's < 0.0142; for males, p's > 0.382), inspection of Figure 6N suggested that BD history increased CREB activity in females selectively on the first day of BE procedures, and an exploration of this apparent group difference confirmed a higher p-CREB:t-CREB ratio in female mice on BE Day 1 (t(14) = 5.024; p < 0.0001), but no water–ethanol differences in the females on Day 10 (t test, p = 0.989) or the males on either BE day (t tests, for Day 1, p = 0.368; for Day 10, p = 0.631).
Together, these data point to a reduction in the activational state of CaMKII or an increase in the activational state of CREB in the escalation of SPF intake over the course of the BE procedures.
Study 5: effects of acute PDE4B inhibition on initial BE–BD cross-sensitization
We next examined the effects of acute PDE4B inhibition on the expression of BE–BD cross-sensitization. For this, female and male mice underwent our 10 d BE procedures and then were pretreated acutely with 1.0 mg/kg A33, 30 min prior to a single BD session. The tissue was collected immediately following the session to examine A33's effects on cross-sensitization–related protein expression and PDE4B activation within the NAC, PFC, and VTA. Control mice consumed chow and were pretreated with VEH to index baseline drinking and basal protein expression.
BE phase
Although it appeared that male SPF-fed mice slated to receive A33 pretreatment exhibited higher SPF intake than their counterparts slated to receive VEH pretreatment (Fig. 7A, right), the analysis of the time-course of eating indicated a significant group × day interaction (F(18,594) = 5.296; p < 0.001), with no significant sex effect or interactions with the sex factor observed (p's > 0.053). Thus, the data were collapsed across the sex factor for post hoc analyses. As illustrated in Figure 7A’, only the SPF-VEH mice exhibited higher food intake than chow-VEH controls on Day 2 of feeding (Tukey's test, p = 0.004 and p = 0.015, respectively), while both SPF-fed groups consumed more food than chow controls for the remainder of the BE phase of the study (Tukey’s tests, p's < 0.001) and no VEH-A33 differences were detected for SPF intake at any point during the BE phase (Tukey's tests, p's > 0.113). To confirm that the magnitude of SPF intake was comparable between SPF-fed mice slated to receive VEH versus A33 prior to the BD phase of the study, we also examined the change in food intake across the 10 d eating period (Fig. 7B). This analyses indicated a group effect only (F(2,71) = 43.479; p < 0.001; sex effect and interaction, p's > 0.183). Thus, the data were collapsed across the sex factor (Fig. 7B’) and Tukey’s post hoc tests confirmed the escalation in intake by both SPF-fed groups, relative to chow-fed controls (p's < 0.001) but also a larger magnitude increase in SPF intake in mice slated to receive A33 versus VEH (p = 0.030). Thus, by random assignment, A33-slated mice exhibited more BE behavior than their VEH controls prior to the BD phase of the study.
Study 5: Effects of acute PDE4B inhibition on behavioral cross-sensitization and protein expression in the NAC. A, Comparison of food intake by female and male chow-fed (Chow) mice slated to receive vehicle (VEH) pretreatment, as well as SPF-fed (SPF) mice slated to receive either VEH or 1.0 mg/kg A33 (A33) over the course of the 10 d BE procedure. A′, The data from panel A, collapsed across sex. B, The data from panel A, expressed as the difference in food intake from Day 1 to 10 of BE procedures. B′, The data from panel B, collapsed across sex. C, Comparison of the effects of acute A33 or VEH pretreatment on ethanol (EtOH) intake by female and male mice during a single bout of BD. Results from subsequent immunoblotting studies of the NAC for (D) PDE4B, (E) p-PDE4B/C/D, (F) their relative expression, (G) Homer2a/b, as well as the (H) monomer and (I) dimer forms of mGlu5. Representative immunoblots for the different proteins examined are also provided. The data represent the means ± SEMs of 12 mice/group/sex for behavior and of 8 mice/group/sex for protein expression. *p < 0.05 chow versus SPF (food-type effect), +p < 0.05 VEH versus A33 (pretreatment effect).
BD phase
Interestingly, while no sex difference was detected with respect to SPF intake (Fig. 7A,B) and despite VEH-A33 group differences in SPF BE prior to pretreatment (Fig. 7B′), we detected a significant sex × group interaction for total ethanol intake on the drinking day (Day 1; F(2,71) = 4.653; p = 0.013). As illustrated in Figure 7C, this interaction reflected a female-selective reduction in ethanol intake by A33, which was confirmed by the results of univariate ANOVAs (for females, F(3,250) = 52.532; p = 0.002; for males, p = 0.723). Furthermore, Tukey's post hoc tests conducted on the data for females confirmed BE–BD cross-sensitization in VEH-pretreated female controls (chow-VEH vs SPF-VEH, p = 0.007), while the ethanol intake of A33-pretreated SPF–fed females was not only significantly lower than that of their VEH-pretreated counterparts (SPF-VEH vs SPF-A33: p = 0.005) but was also statistically indistinguishable from that of chow-fed females (chow-VEH vs SPF-A33, p = 0.992). These data provided our first evidence that acute pretreatment with A33 reduces BE–BD cross-sensitization selectively in female B6JN mice.
Immunoblotting following acute A33
NAC. We next determined whether the behavioral effect of A33 pretreatment was associated with changes in PDE4B expression and/or activation within the NAC immediately following the initial drinking session. We detected group differences in total PDE4B expression in both female and male mice (Fig. 7D; for females, F(2,35) = 15.249; p < 0.001; for males, F(2,34) = 3.314; p = 0.049). In females, the group difference reflected higher PDE4B expression in SPF-VEH versus both chow-VEH controls and SPF-A33 mice (Tukey's tests, respectively, p < 0.001 and p = 0.001), with no difference detected between chow-VEH controls and SPF-A33 females (p = 0.344). In contrast, Tukey’s post hoc tests failed to indicate significant group differences PDE4B levels in males, although SPF-A33 males tended to exhibit higher PDE4B expression than chow-VEH males (p = 0.053; other p's > 0.146). A similar pattern of group differences for females was detected for NAC expression of p-PDE4B/C/D (Fig. 7E; F(2,35) = 7.459; p = 0.002; for males, F(2,34) = 8.061; p = 0.001). Tukey's post hoc analyses confirmed higher p-PDE4B/C/D expression in SPF-VEH females versus chow-VEH and SPF-A33 females (respectively, p = 0.008 and p = 0.004), with no difference between chow-VEH controls and SPF-A33 females (p = 0.974). Given the similar pattern of group differences, there were no group differences in the relative expression of these proteins in the NAC of female mice (Fig. 7F; F(2,35) = 0.113; p = 0.893). Like females, the group difference in p-PDE4B/C/D expression in male mice reflected higher p-PDE4B/C/D expression in SPF-VEH males than both chow-VEH and SPF-A33 males (Fig. 7E; respectively, p = 0.005 and p = 0.004), with no group difference between SPF-A33 and chow-VEH controls (p = 0.973). Consequently, an analysis of relative protein expression yielded significant group differences in the male mice (Fig. 7F; F(2,34) = 4.870; p = 0.014), which reflected lower relative p-PDE4B/C/D expression between SPF-A33 and both chow-VEH and SPF-VEH controls (respectively, p = 0.047 and p = 0.018). However, no difference in relative p-PDE4B expression was detected between chow-VEH and SPF-VEH males (p = 0.905). Taken together, these data indicate that acute systemic pretreatment with A33 effectively reduces the activational state of PDE4B within the NAC of both male and female mice, at least as indexed by p-PDE4B/C/D expression. Furthermore, these results replicate the female-specific association between BE–BD cross-sensitization on Day 1 of drinking and elevated expression of total PDE4B expression within the NAC and argue that the capacity of A33 to reduce total protein expression and activational state of PDE4B in the NAC could underlie the reduction in BE–BD cross-sensitization.
One-way ANOVAs indicated group differences in the expression of Homer2a/b in the NAC of females (Fig. 7G, left; F(2,35) = 4.014; p = 0.028). Tukey's post hoc tests indicated higher Homer2a/b levels in SPF-VEH versus chow-VEH females (p = 0.024), but not between chow-VEH and SPF-A33 (p = 0.149) or between SPF-VEH and SPF-A33 females (p = 0.678). In contrast, no group difference in Homer2a/b expression was observed in the NAC of male mice (Fig. 7G, right; F(2,35) = 0.091; p = 0.913). Thus, we replicated the female-specific association between BE–BD cross-sensitization on Day 1 of drinking and Homer2 expression within the NAC and showed that acute systemic pretreatment with A33 is capable of lowering the expression of this known driver of ethanol intake.
Although we observed increased mGlu5 monomer expression within the NAC of SPF females in Study 5, this effect was less robust than that observed in Study 3, and the results of the statistical analyses indicated no significant chow–SPF differences in mGlu5 monomer expression nor any effect of A33 pretreatment on monomer expression (Fig. 7H, left; F(2,35) = 3.196; p = 0.054). However, significant group differences were detected for the mGlu5 dimer (Fig. 7I, left; F(2,35) = 7.349; p = 0.002), which reflected higher dimer expression in SPF-A33 females than chow-VEH females (Tukey's test, p = 0.002; other p's > 0.074). In contrast to females, no group differences were observed with respect to the NAC expression of either the mGlu5 monomer (Fig. 7H, right; F(2,35) = 0.287; p = 0.753) or dimer in male mice (Fig. 7I, right; F(2,35) = 0.091; p = 0.913). Taken together, these mGlu5 data argue that while elevated mGlu5 expression might contribute to the expression of BE–BD cross-sensitization in female mice, the ability of A33 to block this cross-sensitization does not relate to reduced mGlu5 expression.
VTA and PFC. In contrast to the NAC, we detected no group difference in the expression of any of the protein examined within the VTA or PFC of either female or male mice (Fig. 8). Taken together, these immunoblotting data point further to a key role for the NAC as an important neural locus for BE–BD cross-sensitization.
Study 5: Null effects of acute PDE4B inhibition on protein expression within the VTA and PFC. The table summarizing the means ± SEMs and results of the statistical analyses of the data pertaining to protein expression within the VTA (top) and PFC (bottom) of the female and male mice in Study 5. Representative immunoblots of the protein expression within each region are also provided. Sample sizes were either seven (PFC) or eight (VTA) females/group and eight males/group. No significant group differences were detected.
Study 6: effects of repeated PDE4B inhibition on the cross-sensitization between BE and BD
Given that male mice tend to manifest SPF–EtOH cross-sensitization upon repeated bouts of BD, we conducted a follow-up study examining the effects of repeated A33 pretreatment on SPF–EtOH cross-sensitization. For this, following the 10 d BE phase, both female and male chow- and SPF-fed mice underwent our 10 d BD procedure, during which they were pretreated 30 min prior to bottle presentation with either SAL vehicle or 1.0 mg/kg A33 over the course of the first 5 d of drinking. This experimental design also enabled determination of any A33 effects on chow intake of relevance to the selectivity of the effect, as well as the potential development of tolerance to any A33 effects on ethanol intake. To examine for potential carry-over effects, mice continued to undergo BD procedures for an additional 5 d in the absence of any further A33 pretreatment.
BE phase
Replicating our results, an analysis of chow versus SPF intake during the BE phase of this study indicated a significant sex × food type × day interaction (F(9,648) = 2.091; p = 0.028). Thus, the data were constructed along the sex factor to confirm differences in SPF versus chow intake. As illustrated in Figure 9A, a food × day interaction was detected in females (F(9,324) = 10.257; p < 0.001), which reflected a large increase in SPF intake over the first week of eating that plateaued for the remainder of the 10 d eating period (one-way ANOVA, F(9,171) = 8.691; p < 0.001), while chow-fed females exhibited a progressive decline in chow intake across days (one-way ANOVA, day effect, F(9,171) = 2.232; p = 0.023). A food × day interaction was also detected in males (Fig. 9B; F(9,324) = 13.205; p < 0.001), which reflected a progressive increase in SPF (one-way ANOVA, day effect, F(9,19) = 15.378; p < 0.001), with no change in chow intake (one-way ANOVA, p = 0.071). Although inspection of Figure 9C suggested that females exhibited a larger magnitude in escalation of SPF intake than males, the results of the statistical analysis of the change in food intake indicated only a trend for a sex difference in this regard (food-type effect, F(1,79) = 10.045; p < 0.001; sex effect, 0.191; sex × food type, p = 0.062). Importantly, and in contrast to our acute A33 study (Fig. 7B,B’), at no point during the statistical analyses of the eating data did we detect any differences in either chow or SPF intake between the mice slated to receive VEH or A33 during the BD phase of the study (Fig. 9A–C; all p's > 0.161). Thus, both the VEH and A33-slated mice exhibited similar eating behavior prior to A33 pretreatment and BD procedures.
Study 6: Effects of repeated PDE4B inhibition on behavioral cross-sensitization. Comparison of food intake by (A) female and (B) male chow-fed (Chow) and SPF-fed (SPF) mice slated to receive repeated pretreatment with either vehicle (VEH) or 1.0 mg/kg A33 (A33) over the course of the 10 d BE procedure. C, The data from panel A, expressed as the difference in food intake from Day 1 to 10 of BE procedures. D, Comparison of the effects of acute A33 or VEH pretreatment on ethanol (EtOH) intake by female and male mice on Day 1 of BD procedures. (D′) The data from panel D, collapsed across sex. E, Comparison of the effects of repeated A33 or VEH pretreatment on the average EtOH intake over the first 5 d of BD procedures. E′, The data from panel E, collapsed across sex. F, Comparison of the after-effects of prior repeated A33 or VEH pretreatment on the average EtOH intake over the last 5 d of BD procedures. The data represent the means ± SEMs of 10 mice/group/sex. *p < 0.05 chow versus SPF (food-type effect), +p < 0.05 VEH versus A33 (pretreatment effect).
BD phase
As females exhibit BE–BD cross-sensitization on the first day of drinking (Figs. 2D, 3D, 7C), we first examined the effects of PDE4B inhibition on initial ethanol intake. Replicating this sex difference in initial cross-sensitization, we detected a significant sex × food-type interaction (F(1,79) = 6.506; p = 0.013), which reflected higher ethanol intake by SPF- versus chow-fed females (Fig. 9D; t(38) = 2.567; p = 0.014), with no chow–SPF difference detected in males (t test, p = 0.674). Analyses of the Day 1 drinking data also indicated a significant pretreatment × food-type interaction (F(1,79) = 8.837; p = 0.004), but no significant three-way interaction (p = 0.581), indicating that the effect of A33 on initial ethanol intake was similar in both sexes. Thus, the data were collapsed across the sex factor for post hoc analyses of this interaction. As illustrated in Figure 9D′, acute pretreatment with 1.0 mg/kg A33 lowered ethanol intake in the SPF-fed mice (t(38) = 2.959; p = 0.005), but not in chow-fed controls (t test, p = 0.937). Together, these data for Day 1 of drinking replicate the results of the acute A33 study (Study 5) with respect to a female-selective blockade of SPF–EtOH cross-sensitization and indicate that A33 may also lower ethanol intake selectively in SPF-fed males that do not yet express cross-sensitization.
Female B6NJ mice exhibited higher ethanol intake on average than males during repeated A33 pretreatment (sex effect, F(1,79) = 90.125; p < 0.001); however, cross-sensitization was apparent in both sexes as indicated by a significant food effect (F(1,79) = 9.675; p = 0.003), but no sex × food-type interaction (Fig. 9E; p = 0.194). As also illustrated in Figure 9E, repeated A33 pretreatment reduced the average ethanol intake by SPF-fed mice during the first 5 d of drinking, as indicated by a significant pretreatment × food-type interaction (F(1,79) = 17.074; p < 0.001; three-way interaction, p = 0.583). Given the lack of any sex differences, the data were collapsed across the sex factor upon deconstruction of the significant pretreatment × food-type interaction and follow-up analyses indicated a selective effect of repeated A33 pretreatment on the ethanol intake by SPF-fed, but not chow-fed, mice (Fig. 9E′; for SPF, t(38) = 2.729; p = 0.005; for chow, p = 0.294).
Interestingly, a pretreatment × food-type interaction was detected when the average ethanol intake over the next 5 d of drinking was examined (i.e., Days 6–10), in the absence of any further A33 pretreatment (Fig. 9F; food-type effect, F(1,79) = 17.233; p < 0.001; sex effect, F(1,79) = 15.801; p < 0.001; pretreatment × food type, F(1,79) = 13.706; p < 0.001; sex × food type, p = 0.459; sex × pretreatment, p = 0.459; three-way interaction, p = 0.507). However, deconstruction of this significant pretreatment × food-type interaction indicated higher ethanol intake post-A33 in the chow-fed mice (Fig. 9F′; t(38) = 3.041; p = 0.004), with a less statistically robust attenuation of ethanol-drinking in the SPF-fed animals (t(38) = 1.986; p = 0.054). Thus, PDE4B inhibition by 1.0 mg/kg A33 reduced the expression of BE–BD cross-sensitization in both male and female SPF-fed mice with evidence for a modest carry-over effect a week following repeated pretreatment. Although repeated A33 administration did not impact ethanol-drinking in chow-fed mice, withdrawal from repeated A33 pretreatment elevated ethanol intake.
The results of Study 6 indicate that acute blockade of PDE4B activity lowers BE–BD cross-sensitization in both male and female mice; however, in mice of both sexes, a mild tolerance develops to this effect upon repeated A33 pretreatment with no evidence for any carry-over effect observed in cross-sensitized mice.
Discussion
Consistent with evidence for shared functional neuroanatomy, biochemistry, and behaviors between BE and BD (Boswell and Kober 2016; Vadnie et al., 2014), we found a unilateral cross-sensitization between BE and BD behaviors that is associated with, and dependent upon, activation of PDE4B, a protein coded by PDE4B—a GWAS hit for AUD-AN comorbidity (Munn-Chernoff, 2019).
Comparison of the face and predictive validity of “concurrent” versus “sequential” BE–BD procedures
Study 1 combined a BE procedure that induces escalated SPF intake in mice (Kirkpatrick et al., 2017; Babbs et al., 2018) concurrent with a DID model of BD where mice are presented with 20% (v/v) ethanol (Rhodes et al., 2005). We replicated escalated SPF intake in female and male mice; however, neither SPF nor chow intake differed between water controls and mice with daily BD experience. Notably, the total daily ethanol intake observed in Study 1 was relatively low (< 2 g/kg/day), approximately half the amount observed in subsequent experiments when mice chose between 20 and 40% ethanol (v/v), and this likely reflected the employment of a single-bottle DID procedure as reported for C57BL6/J mice (Cozzoli et al., 2014; Szumlinski et al., 2019). While the low ethanol intake of the mice in Study 1 might account for a lack of any observable effect of drinking on food intake, we did not predict the small, but statistically reliable, “reduction” in ethanol intake on afternoons when mice engaged in a morning feeding session. The fact that the morning feeding session lowered afternoon ethanol intake, irrespective of the food type, differences in amount of SPF versus chow consumed, and sex of mouse, indicated to us that the Study 1 procedure lacked both face and predictive validity for an animal model of BE–BD comorbidity.
Termed “cross-sensitization,” exposure to one drug can increase the behavioral response to a second drug (Vanderschuren and Kalivas, 2000; Berridge, 2004), implicating a common neurobiological mechanism (Nona et al., 2018). Under a cross-sensitization framework involving sequential BE and BD procedures, females consumed more SPF and exhibited a larger escalation of SPF intake than males, the latter of which indicates a greater propensity of females to BE than males. This sex difference was observed under both BE–BD and BD–BE paradigms, indicating that it is directly related to the once-daily BE procedure. Supporting shared mechanisms between BE and BD, a prior 10 d BE history increased subsequent BD (i.e., cross-sensitization) in four distinct experiments. Female mice were also reliably more sensitive than males to BE–BD cross-sensitization as evidenced by increased ethanol intake in SPF females versus chow females on Day 1 of ethanol-drinking. In contrast, for SPF males, BE–BD cross-sensitization developed gradually over BD bouts. We propose that the mechanism underlying BE–BD cross-sensitization is inherently primed or “presensitized” in females but can be incubated over BD bouts in males. Importantly, the increased BE propensity of females, coupled with their increased sensitivity to BE–BD cross-sensitization align with both the strong gender bias in the prevalence of EDs in girls and women (Hudson et al., 2007; Kessler et al., 2013) and the increased prevalence of comorbid AUD-ED in women (Munn-Chernoff et al., 2020). These findings provide both face and predictive validity for our cross-sensitization procedure for studying the neurobiological bases of BE and as a model for BE–BD behavioral comorbidity.
BE–BD cross-sensitization is unidirectional
BE predicts higher ethanol consumption as indicated by longitudinal studies in humans and meta-analyses, whereas prior BD or problem drinking does not necessarily predict the diagnosis or severity of BE (Micali et al., 2015; Escrivá-Martínez et al., 2020; Sampedro-Piquero et al., 2022). Consistent with these observations, BE–BD cross-sensitization in our mouse model was unidirectional: despite mice consuming amounts of ethanol predicted to elevate blood ethanol concentrations above 0.8 mg/ml (Rhodes et al., 2005), our 2 week DID procedure did not induce sensitization of subsequent SPF intake. The unilateral nature of BE–BD cross-sensitization argues that BE in our model induces cellular, biochemical, and/or molecular adaptations within neural circuits that drive ethanol consumption, whereas any BD-related neuroadaptations either occur in distinct neural circuits/systems that do not gate SPF intake or are of insufficient magnitude to produce overt effects on BE behavior. Comparative protein profiling of the NAC following BE–BD versus BD–BE procedures indicated distinct patterns of protein expression between these paradigms, which might underlie differences in cross-sensitization. Interestingly, a female-specific increase in PDE4B expression was detected in BD-experienced mice on both Day 1 and 10 of BE procedures, which could not be explained by sex differences in prior ethanol consumption as no sex differences in ethanol consumption were observed. Currently, the behavioral relevance of this female-specific change in PDE4B expression is not clear and may reflect a correlative biomarker of behavioral sensitization.
Several BE-induced changes in protein expression within the NAC did not vary with prior BD history (including reduced mGlu1 and mGlu5 monomers, Homer2, PDE4B, p-CaMKII and p-ERK, and elevated p-CREB). Some BE-related protein changes were sex-selective (e.g., reduced mGlu1 and PDE4B occurred only in females, while reduced p-ERK was detected only in males), which may reflect the higher SPF consumption exhibited by the female mice in this study. CREB can be phosphorylated by PKA, CaMKII, and ERK (Naqvi et al., 2014), but p-CREB expression in the NAC of BE mice was either unrelated or inversely related to our indices of ERK and CaMKII activation. We hypothesize that the increased p-CREB levels might reflect reduced PDE4B function and, thus, increased cAMP-dependent PKA activity. Individuals with bulimia nervosa exhibit higher mGlu5 expression across the brain, compared with healthy controls (Mihov et al., 2020), and administration of an mGlu5 negative allosteric modular to baboons reduces excessive candy intake (Bisaga et al., 2008). Thus, the seemingly discrepant reduction of mGlu1 and mGlu5 monomers and Homer2 in BE-experienced mice may reflect our study of whole-cell homogenates. Important next steps in characterizing the neurobiological underpinnings of BE–BD cross-sensitization is to examine for changes in cell surface protein expression to provide a more comparable measure of receptor expression to that obtained using positron emission tomography in human subjects with EDs and determine the cell-type specificity of protein changes.
Biomolecular correlates of BE–BD cross-sensitization
BE–BD cross-sensitization on Day 1 of BD argues that prior BE induces neuroadaptations that increase BD propensity to which females are more predisposed. Immunoblotting detected elevated PDE4B expression within the NAC of SPF-fed females on both Day 1 and 10 of BD. Likewise, only SPF females exhibited increased expression of the active dimer form of mGlu5 and Homer2a/b, and we replicated the increase in NAC levels of both PDE4B and Homer2 in SPF-fed female mice on Day 1 of BD in Study 5, with a less robust SPF-induced increase in mGlu5 expression. Although males developed BE–BD cross-sensitization over the course of BD procedures, no changes in protein expression were detected within the NAC between SPF- and chow-fed males in Study 3. Furthermore, no group differences were detected in the expression or activation state of ERK, CaMKII, or CREB within the NAC of either sex, raising question(s) as to the downstream molecular effectors impacted by BE–BD cross-sensitization–related changes in PDE4B activation.
In contrast to the female-selective changes in mGlu5, Homer2, and PDE4B expression, both SPF-fed females and males exhibited a similar increase in p(Ser133/119/190)-PDE4B/C/D within the NAC on Day 1 of BD procedures, relative to chow-fed controls. Furthermore, acute administration of the selective PDE4B inhibitor A33 reduced NAC levels of these phospho-proteins to a similar extent in both sexes, indicating that the cross-sensitization of p(Ser133/119/190)-PDE4B/C/D reflected, in large part, phosphorylated PDE4B. As no change in phospho-protein expression was detected in the VTA or PFC of BE–BD mice, we conclude that our 10 d BE procedures sensitize PDE4B activation to an acute bout of BD—an effect that may be selective for the NAC of both female and male mice. Interestingly, while A33 reduced p(Ser133/119/190)-PDE4B/C/D expression in both sexes, A33 pretreatment lowered ethanol intake in females only in Study 5. While such findings might argue that females are more sensitive than males to the effects of PDE4B inhibition on BE–BD cross-sensitization, acute A33 lowered ethanol intake by SPF-fed female and male mice on Day 1 of BD procedures in Study 6 and no tolerance developed to this effect upon repeated A33 treatment. Conversely, no A33 effect on ethanol intake was observed in chow-fed mice. This latter finding was unexpected as 1.0 mg/kg A33 reduced ethanol intake in SPF-naive B6J and B6NJ mice (Jimenez Chavez et al., 2021). As the total daily ethanol intake by B6NJ mice in the present study was higher than that observed previously (Jimenez Chavez et al., 2021), the null effect herein might reflect insufficient A33 dosing in the absence of any priming by prior BE history. Nevertheless, our results point to the potential clinical utility of repurposing current FDA-approved PDE4 inhibitors for treating not only AUDs (Kohne et al., 2024) but AUD-ED comorbidity and the importance of advancing selective PDE4B inhibitors from rodents to nonhuman primates and humans to confirm their relative safety and efficacy.
Footnotes
This work was funded by National Institutes of Health Grant AA02044 to K.K.S. L.E.M. and I.K. were supported by Undergraduate Research and Creative Activities Awards from the University of California Santa Barbara.
The authors declare no competing financial interests.
- Correspondence should be addressed to Karen K. Szumlinski at szumlinski{at}ucsb.edu.
