The role of anterior insular cortex inputs to dorsolateral striatum in binge alcohol drinking

  1. David L Haggerty
  2. Braulio Munoz
  3. Taylor Pennington
  4. Gonzalo Viana Di Prisco
  5. Gregory G Grecco
  6. Brady K Atwood  Is a corresponding author
  1. Department of Pharmacology & Toxicology, Indiana University School of Medicine, United States
  2. Medical Scientist Training Program, Indiana University School of Medicine, United States
  3. Stark Neurosciences Research Institute, Indiana University School of Medicine, United States

Abstract

How does binge drinking alcohol change synaptic function, and do these changes maintain binge consumption? The anterior insular cortex (AIC) and dorsolateral striatum (DLS) are brain regions implicated in alcohol use disorder. In male, but not female mice, we found that binge drinking alcohol produced glutamatergic synaptic adaptations selective to AIC inputs within the DLS. Photoexciting AIC→DLS circuitry in male mice during binge drinking decreased alcohol, but not water consumption and altered alcohol drinking mechanics. Further, drinking mechanics alone from drinking session data predicted alcohol-related circuit changes. AIC→DLS manipulation did not alter operant, valence, or anxiety-related behaviors. These findings suggest that alcohol-mediated changes at AIC inputs govern behavioral sequences that maintain binge drinking and may serve as a circuit-based biomarker for the development of alcohol use disorder.

Editor's evaluation

Haggerty et al. reported findings examining how changes in brain function are involved in alcohol binge drinking, with a selective focus on the synaptic and circuit alterations that occur in the anterior insular cortex (AIC) inputs within the dorsolateral striatum (DLS). They show that chronic alcohol drinking produces glutamatergic synaptic adaptations in male mice and by stimulating this circuit binge drinking could be reduced without altering either water consumption or general performance for select reinforcing, anxiogenic or locomotor behaviors. The results of this study may specifically improve our understanding of the sex-specific differences in neurocircuitry mediating excessive drinking associated with alcohol use disorder.

https://doi.org/10.7554/eLife.77411.sa0

Introduction

Binge alcohol consumption, defined as consuming at least four drinks for women and five drinks for men in a 2 hr drinking session, represents a large proportion of the deaths associated with problematic alcohol use and alcohol use disorder (AUD) (Mokdad et al., 2004; Kanny et al., 2018). Binge drinking is particularly prevalent among young adults, is behaviorally conserved in rodents, and is a theoretical entry point to the addiction cycle (Thiele and Navarro, 2014; Patrick et al., 2013; Chung et al., 2018; Koob and Volkow, 2010). Ultimately, partaking in binge drinking is one of the strongest predictors for developing AUD (Addolorato et al., 2018).

How alcohol alters neural circuitry that underlies binge drinking remains poorly understood. Previous work has shown binge drinking alters glutamate receptor function across many brain regions (Hwa et al., 2017). Few neural circuits, such as the ventral tegmental area and thalamic inputs to the extended amygdala, locus coeruleus outputs to lateral hypothalamus and rostromedial tegmental nucleus, and medial prefrontal cortex inputs to the periaqueductal gray have been directly linked to the behavioral control of binge alcohol drinking (Rinker et al., 2017; Ferguson et al., 2019; Levine et al., 2021; Dornellas et al., 2021; Burnham et al., 2021; Siciliano et al., 2019).

Numerous clinical and preclinical studies directly implicate a role for the insular cortex in encoding responses to alcohol cues and consumption (Centanni et al., 2021; Gogolla, 2017). Specifically in rodents, anterior insular cortex (AIC) inputs to nucleus accumbens core have been shown to drive aversion-resistant alcohol consumption and play a role in alcohol discrimination and self-administration (Jaramillo et al., 2018a; Jaramillo et al., 2018b; Seif et al., 2013). Others have shown that principal AIC neurons can govern fluid consumption across thirst states that influence voluntary consumption of rewarding substances, such as alcohol (Haaranen et al., 2020; Zhao et al., 2020). Yet, AIC neurons send inputs to many downstream brain regions, so resolving how alcohol alters the functional connectivity by input specificity is of great importance (Gehrlach et al., 2020). Investigating how the AIC governs behavioral control over alcohol consumption across drinking paradigms and brain states at various input locations is essential to advance our understanding of the neuroadaptations associated with AUD.

Recent studies have demonstrated a direct anatomical projection from the AIC to the dorsolateral striatum (DLS) (Hunnicutt et al., 2016; Muñoz et al., 2018; Muñoz et al., 2020; Wall et al., 2013). The DLS is a region of striatum that is implicated in AUD and behavioral responding for alcohol (Barker et al., 2015; Campbell and Lawrence, 2021). Glutamate signaling within the DLS also mediates alcohol seeking behaviors and alcohol exposure alters excitatory glutamate transmission and synaptic plasticity within the DLS (Corbit et al., 2014; Muñoz et al., 2018; Abburi et al., 2016; Johnson et al., 2020; Rangel-Barajas et al., 2021; DePoy et al., 2013). We also previously showed that glutamatergic synaptic plasticity at AIC inputs to the DLS (AIC→DLS) is uniquely vulnerable to disruption by alcohol exposure, relative to other glutamate synapses within the DLS (Muñoz et al., 2018).

Thus, we sought to explore in greater depth the impact that binge alcohol consumption has on glutamatergic transmission and synaptic plasticity specifically at these AIC inputs to the DLS. While these synapses have been investigated anatomically and functionally, they have yet to be behaviorally evaluated in any context. Therefore, we also sought to determine how these synapses may regulate binge alcohol consumption.

Results and discussion

We isolated AIC→DLS circuitry by injecting an anterograde adeno-associated virus (AAV) into the AIC, utilizing the CaMKIIa promoter to drive expression of channelrhodopsin-2 (ChR2) tagged with enhanced yellow fluorescent protein (EYFP) to visualize and modulate glutamatergic AIC inputs within the DLS (Figure 1A and Figure 1—figure supplement 1). Animals then underwent 3 weeks of the Drinking in the Dark (DID) paradigm where they had 0, 2, or 4 hr of access to alcohol or water each day, which produced ‘binge-like’ levels of alcohol intake as measured by a positive correlation between alcohol intake and blood alcohol concentration (BAC) with a subset of animals achieving BACs greater than 0.08 mg% (Figure 1B–D and Figure 1—figure supplement 2; Thiele and Navarro, 2014). Twenty-four hours after the final DID session, once 3 weeks of DID was completed, acute slices of the DLS were made to measure AIC-mediated synaptic responses.

Figure 1 with 5 supplements see all
Binge alcohol consumption alters synaptic plasticity of anterior insular cortex (AIC) inputs to the dorsolateral striatum (DLS).

(A) Representation of viral strategy and expression to record AIC input responses within the DLS. (B) Schematic of Drinking in the Dark (DID) protocol for electrophysiology experiments. (C) Group and individual animals’ alcohol consumption (n=8 animals). (D) Group and individual animals’ water consumption (n=10 animals). (E) Binge alcohol consumption increased optically evoked postsynaptic current (oEPSC) amplitude of AIC inputs within the DLS (two-way mixed analysis of variance (ANOVA), Fluid F(1,50) = 15.6084, p=0.0002; Fluid × Light Intensity F(5,250) = 6.0674, p=0.000025; alcohol: 26 recordings, n=5 animals; water: 26 recordings, n=6 animals). (F) Binge alcohol consumption reduced optically evoked AMPA (oAMPA) to NMDA (oNMDA) glutamate receptor current ratios (Mann-Whitney, U=189, p=0.0072; alcohol: 24 recordings, n=5 animals; water: 28 recordings, n=6 animals). (G) There was no main effect of binge alcohol consumption on paired-pulse ratios of oEPSCs (oPPR) compared to water consumption (two-way mixed ANOVA, Fluid F(1,55) = 2.5544, p=0.1157; alcohol: 30 recordings, n=6 animals; water: 27 recordings, n=5 animals). (H) Binge drinking alcohol decreased optically evoked population spike (oPS) amplitudes produced by photoexciting AIC inputs within the DLS (two-way mixed ANOVA, Fluid F(1,28) = 4.3484, p=0.0463; Fluid × Light Intensity F(6,168) = 2.9574, p=0.0090; alcohol: 14 recordings, n=4 animals; water: 16 recordings, n=6 animals). Error bars and shading indicate ± SEM. Box plot whiskers represent interquartile range.

Figure 1—source data 1

Binge alcohol consumption alters synaptic plasticity of anterior insular cortex (AIC) inputs to the dorsolateral striatum (DLS).

https://cdn.elifesciences.org/articles/77411/elife-77411-fig1-data1-v1.zip

An input-output assessment of optically evoked excitatory postsynaptic currents (oEPSCs) recorded from DLS medium spiny neurons (MSNs) at increasing light intensities showed that alcohol drinking animals had greater oEPSC amplitudes than water controls (Figure 1E). Binge drinking alcohol also decreased optically evoked AMPA to NMDA glutamate receptor current ratios (oAMPA /oNMDA) (Figure 1F). Given that the increased oEPSCs measured with our input-output assessment (Figure 1E) were likely AMPA receptor-mediated, we interpret the decrease in oAMPA /oNMDA ratio to mean that alcohol consumption has a larger effect on NMDA receptor currents than AMPA receptor currents. Although, assessing paired-pulse ratios of oEPSCs (oPPR) at increasing interstimulus intervals showed no significant change (Figure 1G). Binge alcohol consumption also slightly decreased amplitudes of spontaneous excitatory postsynaptic currents (sEPSCs), but there were no other differences in spontaneous synaptic activity between water and alcohol drinkers in the DLS (Figure 1—figure supplement 3).

Consistent with previous findings, we have shown that alcohol can have synapse-specific effects in dorsal striatum, which are otherwise obscured by the net effects at other synapses when synaptic transmission is measured in a non-circuit-specific manner (Muñoz et al., 2018; Muñoz et al., 2020). These data suggest that the measured alcohol-mediated plasticity changes were selective for AIC inputs. Specifically, these data indicate that binge drinking alcohol selectively potentiates postsynaptic AMPA and NMDA glutamate receptor responses at AIC→DLS synapses. We acknowledge that we cannot exclude that alcohol produces other distinct changes at non-AIC inputs to DLS.

To determine if the measured alterations in synapse-specific binge alcohol consumption were conserved in female mice, we exposed aged-matched female mice to 3 weeks of DID after virally isolating AIC→DLS circuitry to perform electrophysiology experiments that methodologically replicated our male recordings (Figure 1—figure supplement 4A-C). Interestingly, female mice that binge drank alcohol compared to water drinking controls show no changes in oEPSC amplitude, oAMPA/oNMDA ratio, or oPPR (Figure 1—figure supplement 4D-F). With both males and female datasets combined, we found a significant fluid effect, but not a significant fluid by sex effect for oEPSC amplitude (three-way mixed analysis of variance [ANOVA], Fluid F(1,75) = 8.309, p=0.0051; Fluid × Sex F(1,75) = 3.044, p=0.0851). For oAMPA/oNMDA ratios, we discovered a sex effect driven by binge alcohol exposed male mice (two-way ANOVA, Sex F(1,80) = 5.8532, p=0.0178; Fluid × Sex F(1,80) = 7.2132, p=0.0088). Since the observed synaptic plasticity changes measured were present only in male mice, we chose to focus on males for further AIC focused circuit-specific and behavioral analyses, but other female circuit-specific alterations need to be explored in future studies.

To investigate the effects of activating AIC inputs on broader DLS network activity, we recorded optically evoked population spikes (oPSs) produced by photoexciting AIC inputs within the DLS at increasing light intensities. oPS amplitudes were decreased in male mice that binge drank alcohol compared to water (Figure 1H), indicating that the net effect of binge alcohol-induced AIC→DLS synaptic adaptations is a reduced network response to AIC input activation. This is curious as we measured an increased glutamate receptor response (Figure 1E and F).

Furthermore, we tested the possibility that only a single DID session was necessary to induce changes in oEPSCs, synaptic plasticity, and altered DLS network effects (Figure 1—figure supplement 5A-D). A single alcohol DID exposure increased oEPSC amplitude compared to water controls, a finding consistent with 3 weeks of DID exposure (Figure 1—figure supplement 5E). Yet, there were no changes in oAMPA/oNDMA ratio and oPS amplitudes (Figure 1—figure supplement 5F and G). This suggests that a single DID exposure is not sufficient to recapitulate the AIC→DLS circuit changes seen after 3 weeks of DID exposure.

Together, these data indicate that enhanced glutamate transmission at AIC inputs to DLS may drive local inhibitory networks leading to overall decreased MSN activity. AIC inputs equally innervate different types of MSNs indicating that this effect is not due to a preferential engagement of one type of MSN (Wall et al., 2013). Other possibilities are that AIC inputs induce feedforward inhibition through local interneurons, such as engaging fast-spiking interneurons, which have been implicated in compulsive alcohol consumption (Patton et al., 2021; Manz et al., 2020). This is a possibility that will need to be explored in future studies.

Alterations in DLS neurotransmission are known to facilitate the development of habit learning in the context of substance-related behaviors and are also associated with the expression of numerous alcohol-related behaviors (Koob and Volkow, 2010; Lovinger and Alvarez, 2017; Corbit et al., 2012; Patton et al., 2016). Glutamate receptor signaling specifically within the DLS mediates alcohol seeking behavior (Corbit et al., 2014). We hypothesized that by modulating AIC inputs within the DLS, we could alter ongoing binge drinking behavior. Initially, we predicted that photoexciting AIC inputs during alcohol consumption would increase binge drinking.

We again isolated AIC→DLS circuitry by injecting an AAV-ChR2-EYFP vector or a control vector that solely expressed enhanced green fluorescent protein (eGFP) into the AIC and implanted a wireless, unilateral optogenetic probe into the DLS to modulate AIC inputs to the DLS during homecage DID sessions (Figure 2A and Figure 2—figure supplement 1; Shin et al., 2017). Animals then underwent 3 weeks of water or alcohol DID, via lickometers that monitored liquid intake, to acquire binge alcohol-induced plasticity changes in the absence of optical stimulation (Figure 2B–C; Godynyuk et al., 2019). Licks, bouts, and total lick duration metrics from the lickometers were significantly correlated with water and alcohol intakes and thus provided high-resolution drinking microstructure details about how and when during each DID session each animal consumed water or alcohol (Figure 2—figure supplement 2). During binge acquisition (weeks 1–3), there were baseline differences between ChR2 and eGFP controls for water, but not alcohol intake (Figure 2D–E).

Figure 2 with 6 supplements see all
Anterior insular cortex (AIC) inputs to the dorsolateral striatum (DLS) modulate binge alcohol intake, but not water intake.

(A) Representation of viral strategy and expression to modulate AIC inputs within the DLS. (B) Photographs and description of lickometer. (C) Schematic of Drinking in the Dark (DID) protocol for binge alcohol or water acquisition. (D) Group and individual animals’ water consumption differed by viral expression during binge acquisition (two-way mixed analysis of variance [ANOVA], Virus F(1,16) = 8.1951, p=0.0113; ChR2: n=8 animals, eGFP: n=11 animals). (E) Group and individual animals’ alcohol consumption did not differ by viral expression grouping during binge acquisition (two-way mixed ANOVA, Virus F(1,14) = 0.1829, p=0.6754; ChR2: n=8 animals, eGFP: n=8 animals). (F) Schematic of DID protocol for binge alcohol or water evaluation. (G) Photoexciting AIC inputs within the DLS during DID did not alter water intakes (two-way mixed ANOVA, Virus F(1,17) = 1.0087, p=0.3293; ChR2: n=8 animals, eGFP: n=11 animals), but (H) decreased alcohol intake (two-way mixed ANOVA, Virus F(1,14) = 8.5743, p=0.0110; Virus × Drinking Week F(3,42) = 4.7132, p=0.0063; Drinking Week 6, p=0.0281; ChR2: n=8 animals, eGFP: n=8 animals). Error bars indicate± SEM. All post hoc comparisons are Sidak corrected.

Figure 2—source data 1

Anterior insular cortex (AIC) inputs to the dorsolateral striatum (DLS) modulate binge alcohol intake, but not water intake.

https://cdn.elifesciences.org/articles/77411/elife-77411-fig2-data1-v1.zip

Following binge acquisition, animals underwent another 3 weeks of water or alcohol DID. In one half of the animals the detection of an alcohol or water lick triggered the activation of the unilateral optogenetic probe in a closed-loop manner, delivering blue light (470 nm) to both ChR2 and eGFP controls for the duration of their licking behavior to evoke glutamate release from AIC inputs to the DLS or as a blue light control, respectively (Figure 2F). The other half of the animals received blue light stimulation stochastically throughout each DID session (see Materials and methods), but there were no differences in water or alcohol intakes between closed-loop and open-loop light delivery, so we collapsed on these groupings (Figure 2—figure supplement 3).

To account for baseline differences between viral expression within fluid type, we summed intakes and microstructure features by week within each animal and displayed these intakes and microstructure features as a percent change from week 3 to determine how photoexciting AIC inputs within the DLS altered intakes and microstructure features. For alcohol and water intake measures for all individual sessions, see Figure 2—figure supplement 4.

For binge evaluation (weeks 4–6), there was no change in water intake between ChR2 and eGFP controls (Figure 2G). Yet, contrary to our initial hypothesis, ChR2 expressing animals drank significantly less alcohol than eGFP controls. Thus, driving glutamate release from AIC inputs within the DLS selectively decreased binge drinking in animals that consumed alcohol, but not water (Figure 2H). This was not a product of the in vivo optical stimulation inducing differential plasticity in water and alcohol drinkers. Testing the same stimulation pattern used in vivo in brain slices did not produce different effects in slices from water and alcohol drinkers or induce long-lasting glutamatergic plasticity on its own, even in the presence of the GABAA receptor antagonist picrotoxin (Figure 2—figure supplement 5).

We reasoned that since driving glutamate release during water DID sessions did not alter intakes, that alcohol-induced plasticity changes at AIC inputs to DLS were a prerequisite for decreasing alcohol intake. To confirm, we photoexcited AIC inputs to the DLS in a separate cohort of animals during binge acquisition (sessions 1–15) to see if animals would decrease their intakes or not acquire stable alcohol intakes. These animals did not show any differences between non-photoexcited alcohol drinkers suggesting alcohol-induced AIC→DLS plasticity changes are necessary to alter binge alcohol intakes (Figure 2—figure supplement 6). These data are consistent with our electrophysiology data showing that a single DID session did not produce equivalent synaptic changes as 3 weeks of DID (Figure 1 and Figure 1—figure supplement 5).

To quantify behavioral alterations associated with fluid intakes, we analyzed drinking microstructure features across drinking weeks by fluid. For water drinkers, there were no differences between ChR2 and eGFP controls in licks, lick duration, bouts, latency to drink, mean inter-drink interval, or bouts in the first 30 min of each DID session (Figure 3A–F). As expected for alcohol drinkers, there were significant decreases in licks, lick duration, and bouts in ChR2 compared to eGFP controls (Figure 3G–I). For alcohol latency to drink and mean inter-drink-interval, there were no significant effects between groups (Figure 3J and K), but the number of alcohol bouts in the first 30 min (a measurement of ‘front-loading’ behavior) decreased for ChR2 compared to eGFP controls (Figure 3L; Wilcox et al., 2014).

Alterations in drinking microstructure represent significant decreases in alcohol intake and are predictive of AIC→DLS alcohol-induced synaptic plasticity changes.

Photoexciting anterior insular cortex (AIC) inputs does not alter the (A) number of water licks (two-way mixed analysis of variance [ANOVA], Virus F(1,17) = 3.7529, p=0.0695; ChR2: 32 observations, n=8 animals, eGFP: 44 observations, n=11 animals), (B) water lick durations (two-way mixed ANOVA, Virus F(1,17) = 2.2136, p=0.1551; ChR2: 32 observations, n=8 animals, eGFP: 44 observations, n=11 animals), (C) water bouts (two-way mixed ANOVA, Virus F(1,17) = 3.0848, p=0.0971; ChR2: 32 observations, n=8 animals, eGFP: 44 observations, n=11 animals), (D) latency to drink water (two-way mixed ANOVA, Virus F(1,17) = 2.7012, p=0.1186; ChR2: 32 observations, n=8 animals, eGFP: 44 observations, n=11 animals), (E) mean inter-drink-interval for water drinking (two-way mixed ANOVA, Virus F(1,17) = 0.0272, p=0.8708; ChR2: 32 observations, n=8 animals, eGFP: 44 observations, n=11 animals), (F) or the number water bouts in the first 30 min of the Drinking in the Dark (DID) session (two-way mixed ANOVA, Virus F(1,17) = 0.5716, p=0.4599; ChR2: 32 observations, n=8 animals, eGFP: 44 observations, n=11 animals). Photoexciting AIC inputs (G) decreases the number of alcohol licks (two-way mixed ANOVA, Virus F(1,14) = 11.0142, p=0.0051; Virus × Drinking Week F(3,42) = 5.8888, p=0.0019; Week 5 p=0.0307, Week 6 p=0.0137; ChR2: 32 observations, n=8 animals, eGFP: 32 observations, n=8 animals), (H) decreases total alcohol lick duration (two-way mixed ANOVA, Virus F(1,14) = 6.9536, p=0.0195; Virus × Drinking Week F(3,42) = 3.8533, p=0.0160; Week 5 p=0.0458; ChR2: 32 observations, n=8 animals, eGFP: 32 observations, n=8 animals), and (I) decreases the number of alcohol bouts (two-way mixed ANOVA, Virus F(1,14) = 11.2086, p=0.0048; Virus × Drinking Week F(3,42) = 9.4893, p=0.0001; Week 5 p=0.0135, Week 6 p=0.0051; ChR2: 32 observations, n=8 animals, eGFP: 32 observations, n=8 animals). (J) Modulating AIC inputs does not impact the latency to drink alcohol (two-way mixed ANOVA, Virus F(1,14) = 0.0084, p=0.9281; ChR2: 32 observations, n=8 animals, eGFP: 32 observations, n=8 animals) or (K) the mean inter-drink-interval for alcohol drinking (two-way mixed ANOVA, Virus F(1,14) = 0.4686, p=0.5047; ChR2: 32 observations, n=8 animals, eGFP: 32 observations, n=8 animals), but (L) decreases front loading behaviors for alcohol drinking (two-way mixed ANOVA, Virus F(1,14) = 4.2003, p=0.0596; Virus × Drinking Week F(3,42) = 3.9078, p=0.0150; Week 6 p=0.0230; ChR2: 32 observations, n=8 animals, eGFP: 32 observations, n=8 animals). (M) Schematic for microstructure feature detection, dataset assembly, cross-validation, and network architecture. (N) Loss curve visualization for training and testing data. (O) Model accuracy for training and testing data. Error bars indicate ± SEM. All post hoc comparisons are Sidak corrected.

Figure 3—source code 1

Alterations in drinking microstructure represent significant decreases in alcohol intake and are predictive of AIC→DLS alcohol-induced synaptic plasticity changes.

https://cdn.elifesciences.org/articles/77411/elife-77411-fig3-code1-v1.zip
Figure 3—source data 1

Alterations in drinking microstructure represent significant decreases in alcohol intake and are predictive of AIC→DLS alcohol-induced synaptic plasticity changes.

https://cdn.elifesciences.org/articles/77411/elife-77411-fig3-data1-v1.zip

Although we found significant changes in microstructure features between viral groupings for alcohol drinkers, there were group trends, variability between animals, and many more microstructure features that were not captured in our weekly analyses to characterize intake changes. To more robustly model microstructure changes, we constructed a feedforward artificial neural network using all 18 microstructure features (see Figure 3—source code 1) from each DID session for all animals to predictively classify both fluid (water vs. alcohol) and virus (ChR2 vs. eGFP) (Figure 3M; Emmert-Streib et al., 2020). After splitting our dataset into training and testing sets, performing sixfold stratified cross-validation (to ensure each training fold had a representative proportion of fluid type and viral expression samples), and training the network we achieved an average of 66.13% accuracy (2.6 times better than chance accuracy) on data previously unseen by the model to predict fluid type and viral expression from a single DID session solely from microstructure data (Figure 3N–O). Thus, AIC inputs within the DLS govern binge alcohol-related behaviors so strongly that we can reliably predict which animals received specific experimental manipulations based on how they consume fluid from a DID session.

Altogether, our drinking data indicate that alcohol-induced neuroplasticity at AIC inputs to DLS may serve as a gain-of-function that allows these synapses to maintain alcohol consumption. In the absence of alcohol, these synapses seem to play no role in modulating intake behaviors, likely due to the absence of a specific type of plasticity required to engage or bring these synapses ‘online’. Once alcohol-induced plasticity at these synapses occurs, it likely enables the governance of alcohol drinking behaviors via complex control over the patterns of licking behavior. Future work is needed to determine if consumption of other rewarding substances causes plasticity at AIC inputs to similarly produce such a gain-of-function. We note that in our previous work binge-like sucrose consumption did not produce the same effect on AIC-DLS synaptic plasticity as alcohol (Muñoz et al., 2018).

We next questioned if the observed decreases in alcohol intake following photoactivation were a product of AIC inputs simply altering behavior in general. In one cohort of animals after the final week of DID, we performed real-time place preference (RTPP) (Figure 4A). There were no differences between ChR2 and eGFP controls for time in zone or distance traveled, suggesting photoexciting AIC inputs to the DLS does not alter preference/avoidance behaviors or locomotion (Figure 4B and C). In a cohort of alcohol-naive animals, we performed a light-dark test (Figure 4D). There were no differences between ChR2 and eGFP controls in time spent on the light side of the box across LED epochs, total number of light side entries, or latency to enter the dark side (Figure 4E–G). Finally, we used optogenetic intercranial self-stimulation to determine if modulating AIC inputs itself was reinforcing (Figure 4H). There were no group differences between nosepoking behaviors for photoexcitation (Figure 4I and J). Altogether, these data indicate that modulating AIC inputs within the DLS does not decrease alcohol intake via changes in reward perception, valence, reinforcement, locomotion, or anxiety-like behavior.

Anterior insular cortex (AIC) inputs to the dorsolateral striatum (DLS) do not modulate valence, anxiety-like, or operant responding behaviors.

(A) Example animal heatmap for real-time place preference assay. Photoexciting AIC inputs within the DLS did not alter (B) real-time place preference/avoidance behaviors (two-way mixed analysis of variance [ANOVA], Virus F(1,11) = 0.2330, p=0.6388; ChR2: n=6 animals, eGFP: n=7 animals) or (C) locomotion during real-time place preference (two-way mixed ANOVA, Virus F(1,11) = 3.8457, p=0.0757; ChR2: n=6 animals, eGFP: n=7 animals). (D) Schematic for light-dark box assay. Modulating AIC inputs did not alter (E) time spent on light side across LED epochs two-way mixed ANOVA, Virus F(1,9) = 1.6826, p=0.2268; ChR2: n=6 animals, eGFP: n=6 animals, (F) the number of total light side entries (t test, t(10) = 0.2204, p=0.83; ChR2: n=6 animals, eGFP: n=6 animals), or (G) delays to enter the dark (t test, t(10) = 0.7699, p=0.4591; ChR2: n=6 animals, eGFP: n=6 animals). (H) Schematic of operant self-stimulation testing. (I) Photoexciting AIC inputs did not alter nosepoking behaviors across session (three-way ANOVA, Virus × Poke × Session F(4,96) = 0.3108, p=0.8702; ChR2: n=6 animals, eGFP: n=6 animals). (J) Active–inactive nosepokes do not differ between viral expression when modulating AIC (two-way mixed ANOVA, Virus F(1,8) = 5.0255, p=0.0553; study sufficiently powered to detect main effect of Virus as determined by power analysis, β=0.9898; ChR2: n=6 animals, eGFP: n=6 animals). Error bars indicate ± SEM. Box plot whiskers represent interquartile range.

Figure 4—source data 1

Anterior insular cortex (AIC) inputs to the dorsolateral striatum (DLS) do not modulate valence, anxiety-like, or operant responding behaviors.

https://cdn.elifesciences.org/articles/77411/elife-77411-fig4-data1-v1.zip

In conclusion, we have identified and characterized how binge consumption of alcohol induces changes in synaptic plasticity in a novel corticostriatal circuit that behaviorally governs binge alcohol drinking. Here, we used unilateral photoexcitation of AIC inputs to the DLS. Given that unilateral and bilateral striatal stimulation can have different outcomes on behavior, future work could explore whether the same holds true for bilaterally activating AIC→DLS synapses (Kravitz et al., 2010). Future work is also required to determine the mechanisms whereby alcohol-induced neuroplasticity alters circuit function to produce changes in drinking and what specific components of drinking behavior AIC→DLS synapses govern, especially how these mechanisms may differ by biological sex.

Nonetheless, the presented data suggest that there exist alcohol-induced synaptic changes within this circuitry that may serve as a critical biomarker for identifying binge alcohol consumption that leads to the future development of AUD. This identification not only will help advance basic substance use research, but may also provide translational value in the prevention and clinical treatment of AUD (Lovinger and Gremel, 2021). Specifically, the existence of a measurable circuit change that tracks with an increased number of binge drinking episodes could be used as the basis for a more reliable quantification of alcohol use and binge drinking history. The presence or absence of these circuit changes may also aid clinicians and researchers in determining a more accurate probability of when or if an individual is at an elevated risk of developing AUD. The usefulness of both measures helps create new knowledge that can foster preventative treatment approaches and helps further identify alcohol consumption patterns that may put an individual at increased risk of developing AUD (Greenfield et al., 2014). More accurately determining AUD risk can be a difficult task which has a biological basis, but is also stratified by social, societal, religious, and cultural factors that can make that calculation difficult. Providing more diverse biological measurements has proven beneficial in improving this risk calculation (Baggio et al., 2020).

New therapeutic modalities, such as non-invasive applications of brain stimulation, like transcranial magnetic stimulation (TMS) are already being investigated in humans to treat AUD. Current approaches for TMS aimed at treating AUD have provided mixed results (Perini et al., 2020; McCalley et al., 2022; Tang et al., 2021). The anatomical location of the anterior parts of the insular cortex in humans are also easily accessible by non-invasive approaches and targeted therapies to this brain region may help advance new non-abstinence-based therapeutic approaches to aid those individuals’ seeking treatment for AUD (Ekhtiari et al., 2019). The possibility that current and next-generation targeted therapeutics focused on circuit-based psychiatric approaches for AUD can modulate this specific brain circuitry to decrease binge drinking behaviors in humans has the potential to immensely impact public health and increase the overall health and quality of life for millions that suffer from AUD.

Materials and methods

Animals

Animal care and experimental protocols for this study were approved by the Institutional Animal Care and Use Committee at the Indiana University School of Medicine (IACUC #: 19017) and all guidelines for ethical protocols and care of experimental animals established by the National Institutes of Health (Bethesda, MD) were followed. Male C57BL/6J mice were ordered from the Jackson Laboratory (Bar Harbor, ME). Animals arrived at 6 weeks of age and were allowed to acclimate to the animal facility for at least 1 week before any surgery was performed. All animals were group-housed (except for mice that underwent DID experiments as outlined below) in a standard 12 hr light/dark cycle (lights on at 08:00 hr). Humidity in the vivarium was held constant at 50% and food and water were available ad libitum.

Stereotaxic surgeries

Request a detailed protocol

All surgeries were conducted under aseptic conditions using an ultra-precise small animal stereotaxic instrument (David Kopf Instruments, Tujunga, CA). Animals were anesthetized using isoflurane (3% for induction, 1.0–1.5% for maintenance at a flow rate of 25–35 ml/min of oxygen depending on body weight at the time of surgery). Viral injections were performed using a 33-gauge microinjection needle (Hamilton Company, Reno, NV). Animals were treated post-operatively for 72 hr with daily injections of carprofen (5 mg/kg) and topical lidocaine on the surgical incision. Animals were allowed to recover for at least 1 week before behavioral assays began. Animals were assigned to groups randomly and after surgery animal IDs were re-coded to blind experimenters to viral expression status.

For electrophysiology experiments, all mice were injected bilaterally with AAV9-CaMKIIa-ChR2(H134R)-EYFP (Addgene, 26969-AAV9) to drive ChR2 expression in the AIC at coordinates A/P: +2.4, M/L: ±2.30, D/V: −3.00 (50 nl/injection, 25 nl/min infusion rate).

For wireless in vivo optogenetic experiments, all mice were injected bilaterally with AAV9-CaMKIIa-ChR2(H134R)-EYFP (Addgene, 26969-AAV9) or AAV9-CaMKIIa-eGFP (Addgene, 105541-AAV9) to drive ChR2 expression or eGFP control in AIC at coordinates A/P: +2.4, M/L: ±2.30, D/V: −3.00 (50 nl/injection, 25 nl/min infusion rate). To modulate AIC inputs in the DLS, mice were unilaterally implanted with a wireless probe (4 mm depth, 470 nm blue light LED – Neurolux, Northfield, IL) at coordinates A/P: +0.7, M/L: ±1.55, D/V: −2.5. The LED orientation allowed for light to travel from anterior DLS to posterior DLS and ensured minimization of off-target light activation of AIC cell bodies while maximizing light coverage for AIC input innervation in the DLS. Probes were placed in either the right or left DLS, but there was no lateralization effect on alcohol or water intake, thus data are presented collapsed on probe location. Probes were secured to the skull using ethyl cyanoacrylate (LOCTITE 444, Henkel, Rocky Hill, CT) and the skin was closed over the top of the probe using Vetbond tissue adhesive (3M, Saint Paul, MN).

Drinking in the Dark

Request a detailed protocol

The DID paradigm was based on the original DID procedure (Thiele and Navarro, 2014) with two modifications. First, there were four 2 hr DID sessions (Monday–Thursday) and one 4 hr DID session (Friday) and second, all DID sessions were performed out of a single bottle of water or alcohol (20% v/v in water) via lickometers (described below) inserted into the cage at the beginning of the DID session, which were removed at the end of the DID session. Mice had ab libitum access to their standard water bottles at all other times.

To summarize the performed procedures, after animals recovered from surgery, they were singly housed and allowed to acclimate in a reverse 12 hr dark/light cycle (lights off at 06:00 hr) for 1 week. Three hours into the dark cycle (09:00 hr) the standard water bottle was removed from the cage and the lickometer was inserted into the cage after the fluid bottle was weighed for 2 or 4 hr depending on the test day. Following the completion of 2- or 4-hr access, the lickometers were removed and the fluid bottles were weighed immediately after the session. Grams per kg (g/kg) of water and alcohol were computed from the difference in bottle weight and the density of water or 20% alcohol in water. Mice were weighed post-DID session on Monday, Wednesday, and Friday (Tuesday and Thursday weights were equivalent to Monday and Wednesday, respectively). For Saturday and Sunday, the mice had no access to the lickometers as no procedures were performed. These 1-week repeating DID cycles are referred to as ‘Drinking Weeks’.

Water and alcohol intakes for study inclusion were determined by summing the intake across the first 15 DID sessions. Animals that fell outside of the lower interquartile range (IQR) for their respective fluid type were excluded from the study for not consuming enough water or alcohol in the binge acquisition period. Only one alcohol animal was removed from the study for failure to meet this criterion as it only consumed a total of 1.28 g/kg of alcohol across all 15 DID sessions.

Blood alcohol concentration

Request a detailed protocol

In a subset of animals following all DID sessions as to not disturb drinking, retro-orbital blood samples were taken after an additional 2-hr DID session. BACs were determined via gas chromatography (GC-2010 plus, Shimadzu, Japan) and correlated with respective alcohol intakes from that session.

Electrophysiology

Request a detailed protocol

Brain slices were prepared 24 hr after the completion of the mice’s final sessions during week 4 (sessions 16–20) of DID. Mice were anesthetized with isoflurane and then immediately killed via decapitation. The brain was rapidly dissected out and placed in an ice-cold cutting solution containing (in mM): 194 sucrose, 30 NaCl, 26 NaHCO3, 10 glucose, 4.5 KCl, 1.2 NaH2PO4, and 1 MgCl2, saturated with a 95/5% O2/CO2 mixture. Brain slices containing the DLS were made at a thickness of 280 μm on a VT1200S vibratome (Leica, Germany). Slices were incubated in artificial cerebrospinal fluid (aCSF) containing (in mM): 124 NaCl, 26 NaHCO3, 10 glucose, 4.5 KCl, 2 CaCl2, 1.2 NaH2PO4, 1 MgCl2 (310–320 mOsm) saturated with a 95/5% O2/CO2 mixture at 30°C for 1 hr after which they were moved to incubate at room temperature until recording. DLS brain slices were transferred to a recording chamber continuously perfused with 95/5% O2/CO2-saturated aCSF solution at a rate of 1–2 ml/min at 29–32°C.

DLS brain slices were visualized on a BX51WI microscope (Olympus Corporation of America, Center Valley, PA). Whole-cell patch clamp electrophysiological recordings were made from MSNs in the voltage clamp configuration. Recordings were made using a Multiclamp 700B amplifier and a 1550B Digidata digitizer (Molecular Devices, San Jose, CA). Glass patch pipettes (borosilicate with filament, 2.0–3.5 MΩ resistance, World Precision Instruments, Sarasota, FL) were made using a P-1000 micropipette puller (Sutter Instruments, Novato, CA). Pipette internal solution contained (in mM) 120 CsMeSO3, 10 HEPES, 10 TEA-Cl, 5 lidocaine bromide, 5 NaCl, 4 Mg-ATP, 1.1 EGTA, and 0.3 Na- GTP (pH 7.2, 290–310 mOsm). MSN identity was determined visually based on soma size and confirmed via measures of membrane resistance and capacitance. MSNs were constantly voltage clamped at −60 mV (except when recording NMDA receptor-mediated currents), using sampling rate of 10 kHz and low-Bessel filter 2 kHz. No correction for liquid junction potential was used. In order to isolate excitatory transmission, picrotoxin (50 μM) was added to the aCSF solution for whole-cell recordings and a subset of field recordings. Extracellular field recordings were used to measure network responses using glass micropipettes filled with 1 M NaCl. Both whole-cell patch pipettes and field recording pipettes were placed in the regions of the DLS where AIC innervation was the most dense (Figure 1—figure supplement 1) for recordings. Data were acquired using Clampex 10.3 (Molecular Devices, San Jose, CA). For whole-cell recordings, series resistance was monitored throughout the experiment. Cells with series resistances greater than 25 MΩ or that changed more than 15% during recording were excluded from data analyses.

Optically evoked excitatory postsynaptic currents (oEPSCs) or oPSs were produced in DLS brain slices using 5 ms 470 nm blue LED light pulses delivered via field illumination through the microscope objective. oEPSCs and oPSs were evoked every 30 s. Stimulus-response measures of oEPSC were performed by increasing the intensity of blue LED light from 0 up to 1 mW. For AMPA/NMDA and paired-pulse ratio (PPR) whole-cell recording measures, light intensity was initially adjusted to produce stable oEPSCs of 100–600 pA amplitude after which experimental recordings were begun. oAMPA/oNMDA current ratio recordings were measured by first holding the cell at −80 mV and optically evoking an AMPAR-mediated EPSC. The NMDAR current was then determined by shifting the membrane potential to +40 mV and optically evoking an EPSC. Since the AMPAR component of oEPSCs at −80 mV was not apparent at 100 ms following the optical stimulus (the measured current rapidly returned to baseline), we calculated the NMDAR-mediated portion of the oEPSC at +40 mV as the average of the measured current over the following 25 ms (100–125 ms post-stimulus), similar to our previously published protocol (Fritz et al., 2018). oPPRs were measured by performing two optical stimulations separated by 50–200 ms averaging three technical replicates. oPPR was calculated by taking the average amplitude of the second pulses (using tail current as baseline) and dividing it by the average amplitude of the first pulses (using pre-stimulation current as baseline). Non-optically evoked sEPSCs were also measured using whole-cell voltage clamp recordings, with MSNs held at –60 mV using sampling rate of 50 kHz. sEPSCs were measured over the course of a 2 min gap-free recording. sEPSC amplitude, frequency, and decay and rise times were computed using Minianalysis (Synaptosoft, Decatur, GA). Stimulus-response measures of oPSs were performed by increasing the intensity of blue LED light from 0 up to 1.15 mW. To test whether the optical stimulation used in vivo produced plasticity on its own, we recorded a 10 min baseline of oPS amplitudes elicited every 30 s after which we optically stimulated the brain slices with blue LED light (again, through the objective) at 20 Hz for 10 s. oPS amplitudes were then measured again at 30 s intervals. In one set of experiments, at the conclusion of recording the effects of the 20 Hz stimulation, the AMPA receptor antagonist NBQX (5 μM) was bath applied to the brain slice. In a separate set of experiments, the same recordings of oPS amplitudes with 20 Hz stimulation were performed but in the presence of the GABAA receptor antagonist picrotoxin (50 μM).

Histology

Request a detailed protocol

Animals were anesthetized with isoflurane and trans-cardially perfused with 15 ml of ice-cold phosphate buffered saline (PBS) followed by 25 ml of ice-cold 4% paraformaldehyde (PFA) in PBS. Animals were decapitated and the brain was extracted and placed in 4% PFA for 24 hr. Brains were then transferred to 30% sucrose solution until they sank after which they were sectioned on a vibratome at a section thickness of 50 μm. Brain sections were mounted in serial order on glass microscope slides and stained with 4′,6-diamidino-2-phenylindole to visualize nuclei. Fluorescent images were captured on a BZ-X810 fluorescent microscope (Keyence, Itasca, IL) using ×4 and ×20 air objectives. Injection site, viral expression, and the location of optogenetic probe implantations were determined from matching images to the Reference Allen Mouse Brain Atlas. Animals that did not have viral expression in the brain regions of interest and/or did not have optogenetic probe placements within the brain region of interest as confirmed by histology were excluded from the study.

Lickometers

Request a detailed protocol

All alcohol and water drinking experiments were performed out of custom-built homecage fluid monitors (i.e. ‘lickometers’) that were constructed as described (Godynyuk et al., 2019) with the following modifications. Liquid monitoring was constantly sampled as a function of the state (open vs. closed) of an infrared beam directly in front of the fluid bottle valve, but data was only written to device memory at minimum every 2 s. Therefore, any tube interaction within a 2 s window (defined as a drinking bout) was recorded as the total number of beam breaks (defined as licks) for the total duration that the beam was broken (defined as lick duration) within that bout. Drinking bouts could be longer than 2 s, but never shorter.

The lickometers can hold two bottles, although we only used one for each device during these experiments. We randomized the bottle side location across animals and found no effect on drinking, thus data are presented collapsed on bottle side.

Lickometer data was cleaned by fitting a linear model comparing the number of licks by the lick duration within each bout for every DID session performed by each animal.

Bouts that had a residual value of greater than or less than 3 from model fit were removed. This cleaning procedure removed bouts that were due to slow leaks (a new tube was used if a leak was detected for the next DID session) or chews on the bottles. A total of 3,928,956 bouts were recorded across all experiments, and after cleaning 3,927,570 remained (99.9647% of bouts remained) showing that almost all bottle contacts were directed drinking behaviors, and that chews and leaks were rare.

For wireless in vivo optogenetic experiments that utilized lickometers for closed-loop control during DID sessions, the lickometer’s BNC out was connected to a TTL port on the Neurolux optogenetics system such that when a lick occurred, for the entire duration of the lick, a voltage change was sent from the lickometer to turn on the optogenetics probe. Optogenetic stimulation was delivered at 20 Hz with 5 mW of light power at 470 nm for the entire duration of the beam break.

In a subset of animals, we also performed yoked controls such that yoked animals received optogenetic stimulation at stochastic time points during DID sessions. The yoked controls’ stimulations were dependent on the drinking activity of some other animal in the room during the DID session to ensure the amount of stimulation and type (length, interval, etc.) was similar across groupings. There were no differences in intake by yoked status for water or alcohol, thus the data are presented collapsed on yoked and closed-loop stimulation. This finding suggests that changes in alcohol intake driven by optogenetic manipulations may be generalizable to when alcohol is accessible or at timepoints near alcohol consumption behaviors, and not necessarily linked to closed-looped stimulation during licking activity.

Microstructure feature analyses

Request a detailed protocol

Using licks, lick duration, bouts, and the timestamps when these events occurred during the DID session, we calculated other drinking features such as latency to drink (time to first bream break after session initiation) and mean inter-drink interval time per session (the mean time between bouts within each DID session). We also calculated features for events in the first 30 min of each DID session as a measure of front-loading behavior. Finally, we used maximum values (max bout length, max licks per bout, etc.) to characterize microstructure features within DID sessions.

Machine learning

Request a detailed protocol

A feedforward artificial neural network was used to determine if microstructure features from a single DID session could predict the experimental manipulations across fluid type (alcohol vs. water) and viral expression (ChR2 vs. eGFP). To bolster our predictive capabilities, we used all 18 microstructure features we computed per DID session.

Features were concatenated across all animals and DID sessions into a single data frame and categorical labels were one-hot encoded. Data were normalized to a range of 0–1 before cross-validation using stratified k-fold (k=6) to ensure the unequal proportion of fluid × virus labels in the dataset did not influence model training predictions. Each shuffled fold contained 832 training examples and 166 testing samples. The sequential network architecture contained 2048 nodes in the first layer, 512 nodes in the second layer, 64 nodes in third layer, and an output layer of 4 nodes. We used a rectified linear unit activation function for the first three layers and a softmax activation function for the final output layer. The model was compiled using an Adam optimizer and loss was scored using categorical cross entropy. After training all six folds, the average testing accuracy on data previously unseen by the model was 66.13% with a maximum of 70.48% accuracy for one of the training folds.

Real-time place preference

Request a detailed protocol

One week after the final DID session (session 30), animals were subjected to an RTPP assay to determine if photoexciting AIC inputs altered valence states that could explain differences in alcohol behaviors. Animals were placed in an arena (40 cm × 20 cm × 30 cm) with a divider down the middle that contained a cutout so mice could freely pass through from one side to the other. One side of the arena contained a mesh floor with vertical striped walls and the other with wire floor and horizontal striped walls. All testing was performed in the dark using red light and infrared cameras to capture behavioral videos. Two boxes were used, each with counterbalanced LED ON sides that triggered the activation of the optogenetic probe to deliver 20 Hz 470 nm light pulses at 5 mW when entry into that side of the arena occurred for the entire duration the animal was in that side of the arena. For the pretest day, animals were placed in the arena for 20 min with no optogenetic activation regardless of their location in the arena. On the RTPP testing day the same procedure was repeated except the LED ON side of the box triggered optogenetic stimulation. All location and locomotion data were computed using EthoVision XT (V15 – Noldus, Skokie, IL) from a camera placed directly above the arena which measured the center of the animal as it freely behaved across testing.

Light dark box

Request a detailed protocol

A light dark box assay was used in a separate cohort of alcohol-naïve mice to determine if photoexciting AIC inputs could alter anxiety-related behaviors. Mice were placed into an arena (40 cm × 40 cm × 30 cm) that was split in half with a covered divider that contained a cutout so mice could freely pass through from one side to the other. The open side of the box was brightly lit (300–500 lumens) with overhead light and the other covered side registered dimly lit (0–10 lumens). The 15 min session consisted of three epochs (LED ON, LED OFF, LED ON) for 5 min each. During the LED ON epochs, optogenetic stimulation was delivered at 20 Hz with 470 nm light pulses at 5 mW for the duration of the epoch. Animals were placed in the corner furthest from the dark entry to begin the assay. All animal tracking data was computed using EthoVision XT (V15 – Noldus, Skokie, IL) from a camera placed directly above the arena which measured the center of the animal as it freely behaved across testing.

Intracranial self-stimulation

Request a detailed protocol

Intracranial self-stimulation was used to determine if photoactivation of AIC inputs could maintain an operant response. Intracranial self-stimulation was performed in a homecage (30 cm × 15 cm × 15 cm) with a FED3 device (Matikainen-Ankney et al., 2021). Each FED3 device has two nose poke ports (‘active’ and ‘inactive’). Responding in the active port was reinforced with photoactivation on a fixed-ratio 1 schedule and resulted in light delivery (5 mW of 470 nm light in 25 ms pulses for a total of 20 pulses) as well as presentation of a tone (4 kHz for 300 ms) and illumination of a cue light bar located below the active nose port. During only the initial session, the back of the active nose poke was baited with crushed Froot Loops and sucrose pellets to encourage nose poking behaviors. The inactive port resulted in no photoactivation, tone, or light cue. Animals were run for five sessions, each 1 hr long and the active port was randomized and balanced across all animals.

Modeling and statistics

Request a detailed protocol

Sample sizes for all experiments were determined based on previously published experimental findings for electrophysiology and in vivo behavioral assays such as DID (Muñoz et al., 2018).

Data preprocessing and machine learning modeling utilized SciPy (Virtanen et al., 2020), Statsmodels (Seabold and Perktold, 2010), Scikit-learn (Pedregosa et al., 2011), and TensorFlow (Abadi et al., 2016). Statistical analyses were performed using GraphPad Prism (GraphPad Software – V9.2, San Diego, CA) and pingouin (V0.5.0, Vallat, 2018). Data visualization used matplotlib (Hunter, 2007) and seaborn (Waskom, 2021) libraries. For time series or repeated measures, we used two-way mixed ANOVAs, which represented time (session or week) or the repeated factor (light power, etc.) as the repeated, within-subject variable, and the factor (fluid type, viral expression, bottle side, etc.) as the between-subjects factor. All two-way mixed ANOVA data were tested to see if variances between factors were equal and normal using the Levene test. If there was a main effect for factor, we used pairwise t-tests to determine post hoc significance and p-values were Sidak corrected. For correlations between two variables, we tested multivariate normality using the Henze-Zirkler test. If samples were normal, we used a Pearson’s correlation to report r and p-values. If samples failed normality testing, we used a Shepherd pi correlation that returned the Spearman correlation after removing bi-variate outliers. For tests of two factors, we used unpaired Welch two-tailed t-tests to correct for unequal sample sizes or paired t-tests. All significance thresholds were placed at p<0.05 and all data and model fits are shown as mean ± standard error (68% confidence interval) of the mean (SEM) or by box plot with error bars indicating the IQR. Data points beyond IQRs are represented as diamonds.

Data and materials availability

Request a detailed protocol

All experimental data are available in the main text or the supplementary materials. Microstructure analysis and machine learning code is available at https://github.com/dlhagger/AIC-DLS_Microstructure_Modeling (Haggerty, 2022; copy archived at swh:1:rev:ccf174d370b21d632426e929f9bcfd5b539b211c).

Data availability

Source data files are provided for all electrophysiology and behavior studies and machine learning model code is provided to reproduce all neural network analyses. Further code to replicate figures is hosted on GitHub at https://github.com/dlhagger/AIC-DLS_Microstructure_Modeling (copy archived at swh:1:rev:ccf174d370b21d632426e929f9bcfd5b539b211c).

References

  1. Conference
    1. Abadi M
    2. Barham P
    3. Chen J
    4. Chen Z
    5. Davis A
    6. Dean J
    7. Devin M
    8. Ghemawat S
    9. Irving G
    10. Isard M
    11. Kudlur M
    12. Levenberg J
    13. Monga R
    14. Moore S
    15. Murray DG
    16. Steiner B
    17. Tucker P
    18. Vasudevan V
    19. Warden P
    20. Wicke M
    21. Yu Y
    22. Zheng X
    (2016)
    TensorFlow: a system for large-scale machine learning
    In Proceedings of the 12th USENIX conference on Operating Systems Design and Implementation. pp. 265–283.
    1. Chung T
    2. Creswell KG
    3. Bachrach R
    4. Clark DB
    5. Martin CS
    (2018)
    Adolescent binge drinking developmental context and opportunities for prevention
    Alcohol Research-Current Reviews 39:5–15.
  2. Book
    1. McCalley DM
    2. Kaur N
    3. Wolf JP
    4. Contreras IE
    5. Book SW
    6. Smith JP
    7. Hanlon CA
    (2022)
    Medial Prefrontal Cortex Theta Burst Stimulation Improves Treatment Outcomes in Alcohol Use Disorder: A Double-Blind, Sham-Controlled Neuroimaging Study
    Biological Psychiatry Global Open Science.
  3. Book
    1. Patton MH
    2. Shah AP
    3. Mathur BN
    (2016)
    Alcohol effects on the dorsal striatum
    In: Soghomonian Jean-Jacques, editors. Basal Ganglia. Springer International Publishing Cham. pp. 289–315.
    1. Pedregosa F
    2. Varoquaux G
    3. Gramfort A
    4. Michel V
    5. Thirion B
    6. Grisel O
    7. Blondel M
    8. Prettenhofer P
    9. Weiss R
    10. Dubourg V
    (2011)
    Scikit-learn: machine learning in python
    The Journal of Machine Learning Research 12:2825–2830.
  4. Conference
    1. Seabold S
    2. Perktold J
    (2010)
    Statsmodels: Econometric and Statistical Modeling with Python
    Python in Science Conference.

Decision letter

  1. Michelle Antoine
    Reviewing Editor; National Institute on Alcohol Abuse and Alcoholism (NIAAA), United States
  2. Kate M Wassum
    Senior Editor; University of California, Los Angeles, United States
  3. Andrew Lawrence
    Reviewer; The Florey Institute of Neuroscience and Mental Health, Australia

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Anterior insular cortex inputs to the dorsolateral striatum govern the maintenance of binge alcohol drinking" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Kate Wassum as the Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Andrew Lawrence (Reviewer 2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

We ask that you please address all reviewer comments and especially attend to the follow.

1. In figure 4J where n=2, please increase the n to at least 6/7 or provide power analysis justifying the n used if less than 6. See Reviewer 3 comment 3 and Reviewer 1 comment 1 for further explanation.

2. To strengthen the case for the specificity of the effect in figure 1 please repeat the experiments in Figure 1 to include female mice.

3. Please restrict the implication of the findings to addiction or the impact of these results to public health to the discussion only. In the discussion further expand on the use of AIC to DLS alterations as a biomarker for the development of alcohol use disorder.

4. For figure 1B, please clarify in the results and methods section whether brain slices were taken 24 or 48 hours after the final alcohol session. See Reviewer 2 comment 2 for further explanation.

4b. Also please report the rationale for starting the ex-vivo recording 24 h after the last binge episode. See Reviewer 1 comment 2 for further explanation.

5. Please add DAPI to the representative picture of viral injection in figure 1A. See Reviewer 1 comment 6 for further explanation.

6. Please add statistical p-values for figure 4C. See Reviewer 3 comment 3 for further explanation. Please also ensure your manuscript complies with the eLife policies for statistical reporting: https://reviewer.elifesciences.org/author-guide/full "Report exact p-values wherever possible alongside the summary statistics and 95% confidence intervals. These should be reported for all key questions and not only when the p-value is less than 0.05."

7. Please comment in the methods whether reported values include a correction for the liquid junction potential. See other comments section for reviewer 3.

8. Please clarify the description of the optogenetic stimulation paradigm in line 213.

See other comments section for reviewer 5.

Reviewer 1 (Recommendations for the authors):

Based on the comments above, these are the recommendations for the authors:

1. I strongly suggest increasing the number of mice to have at least 6/7 mice/group for each experiment.

2. Please report the rationale for starting the ex-vivo recording 24 h after the last binge episode. An alternative interpretation for the reported data could be that the glutamatergic synaptic adaptation observed are simply related to only one (i.e., the last) binge episode and not with the drinking history. Please discuss the possibility that these adaptations can occur with a single binge/alcohol exposure.

3. In line with the point above, it is not clear why for the electrophysiology recording the control virus is missing. This could be critical also for validating the behavioral approaches.

4. Regarding the issue with the unilateral stimulation, ideally the authors should repeat the experiments (at least some of the behavioral controls) to demonstrate similar results. Alternatively, the authors could add a "methodological consideration" section to explain these potential limitations.

5. I strongly suggest adding female mice to this manuscript to increase the impact of the results.

6. Please add DAPI to the representative picture of viral injection. It is difficult to appreciate anatomical landmarks for these pictures.

7. Please keep both introduction and discussion related to the circuit investigation and remove implications for human transition to addiction or the impact of these results on public health.

Reviewer 2 (Recommendations for the authors):

I enjoyed reading this manuscript and have a number of comments for the authors to consider.

1. In the introduction, please qualify the criteria of a binge as per NIAAA guidelines, namely 4 drinks in 2 hours for women and 5 drinks in 2 hours for men. Also, there is a recent review on the role of the insula in AUD that updates the cited Barker et al. 2015 review (see Campbell and Lawrence 2021).

2. In the text you state that brain slices are taken 24 hours after the final alcohol session; however, figure 1B implies that slices are taken 48 hours after alcohol. Which is it?

3. In rodents silencing insula inputs to the accumbens core reduces alcohol self-administration (Jaramillo et al., 2018), and in human heavy drinkers there is increased coupling between the right anterior insula and right nucleus accumbens (Grodin et al., 2018) during high threat alcohol cues. With regards the latter point, in rats the AI mediates relapse-like behavior in a punishment associated context (Campbell et al., 2019). Have the authors performed analogous recordings at AI→accumbens synapses to compare vs the AI→DLS synapses? Also, how much spread of virus was there along the A-P axis of the AI? In this regard rodent studies suggest that functionally inhibiting the caudal portion of the insular cortex reduced alcohol consumption whereas functionally inhibiting the anterior insular cortex. was without effect (Haaranen et al., 2020; Pushparaj and Le Foll, 2015).

4. The authors conclude that "binge drinking alcohol specifically potentiates postsynaptic AMPA and NMDA glutamate receptor responses at AIC→DLS synapses". There is evidence that AI→ventral striatum projections are involved in compulsive food eating (Spierling et al., 2020). As the prior comment, have the authors assessed AI→ventral striatum synapses? Also, while I acknowledge the authors previously showed the disruption of opioid receptor mediated LTD at AIC→DLS synapses was specific to alcohol, and did not occur after sucrose binge (Munoz et al., 2018), have they assessed the specificity of the current data set in relation to alcohol vs other (non-drug or drug) rewards?

5. Note for discussion, within the DLS in both humans and rodent models, muscarinic M4 receptors are subjected to alcohol-induced adaptations and implicated in alcohol seeking (Walker et al., 2020). Have the authors considered this possibility in terms of binge behavior?

6. Figure 2 G,H – why are the data presented as %? Moreover, in figure 2H does alcohol intake increase across weeks in the eGFP group?

7. Supplemental figure 7B – in the ChR2 group the baseline alcohol intake drops both times prior to the final two binge tests (see sessions 21-30). Does this impact the % change data? Over the same period the baseline intake in the eGFP group is more stable.

8. Supplemental figure 9 – the drinking profile of the LED off group is unstable, with intake seeming to escalate with time. Is this the case and if so why?

9. In terms of a molecular mechanism, have the authors considered investigating the potential role of BDNF / p75 (see Darcq et al., 2016)?

10. Binge drinking in adolescent females is a growing problem, have the authors studied female mice?

Reviewer 3 (Recommendations for the authors):

1) PPR at 25 ms should not be included. The very low values reflect poor fidelity of ChR2 at stimulation frequencies >30 Hz, rather than biological properties of AIC->DLS synapses. Also, the interpretation of these data in line 120 should use specific language to indicate directionality in place of "alter".

2) I suggest switching orders of graphs throughout so that the control groups (Water, eGFP, etc) precede the experimental groups (Alcohol, ChR2, etc)

3) I do not understand how the DID intake values for the Water group are tenfold higher than the Alcohol group. Also, "g/kg" is an unusual way to display Water Intake, I think "mL" is more informative and conventional.

4) The representative AMPA/NMDA ratio trace in Figure 1F displays significant AMPA receptor rectification. MSNs do not typically display significant AMPAR rectification under control conditions. Furthermore, the internal solution does not appear to contain spermine, so voltage dependent block of CP-AMPARs should not be a major factor. Did the authors correct for the liquid junction potential?

5) The description of the optogenetic stimulation paradigm (line 213) is misleading. As written, it reads as if most of the animals underwent closed-loop stimulation but only a minor portion were open-loop and eventually folded into the main data set. By contrast, the cohort was split n=4/4 open/closed as shown in Figure S6. This section should be re-written accordingly.

6) The authors should be commended for assessing the effects of their optogenetic stimulation on long-term plasticity in brain slices. Experiments like this are rarely performed but can provide useful, if not necessary, information for interpreting the effects of in vivo optogenetic stimulation. While no long-term effects were detected, it appears to me as if there could be an Alcohol x Picrotoxin interaction immediately after the 20 Hz stimulation. It seems as if picrotoxin might have decreased the post-tetanus potentiation in slices from Water but not Alcohol mice. If so, this could provide some insight into changes in AIC->DLS feedforward inhibition following DID. Also, these experiments should be described in greater detail in the Methods.

7) While the summarized oPS timecourses are completely blocked by NBQX, the representative oPS traces do not appear to be.

8) There seems to be an effect of AIC->DLS stimulation during the first DID cycle in Figure S9.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Anterior insular cortex inputs to the dorsolateral striatum govern the maintenance of binge alcohol drinking" for further consideration by eLife. Your revised article has been evaluated by Kate Wassum (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

1. Amend the title of the manuscript to "The role of anterior insular cortex in binge alcohol drinking" (or similar).

2. Add to the main text that unilateral opto manipulation was used.

3. Include missing citation on Line 1002.

4. Add the "n" values to the figure 3 legend. Please double-check that all figure legends contain such "n" value information.

5. Perform power analysis to determine whether the ICSS dataset is adequately powered. State the result (whether or not ) concisely in the text. For example ( p-value, study sufficiently powered as determined by power analysis).

6. Add missing citations to the conclusion.

Reviewer 3 (Recommendations for the authors):

The addition of female mice is very welcome, and I agree that the lack of effects on physiology provides justification for limiting the behavioral studies to male mice. The introduction is much more thorough and provides greater context for the studies in this manuscript. The expanded conclusion is an improvement as well. I appreciate the additional replicates included in Figure 4 as well; however, I am not convinced that the ICSS experiment is fully powered.

The ICSS dataset, to me, appears to indicate a strong trend (.055) for AIC->DLS stimulation in decreasing discrimination of active vs inactive lever. This could be a very interesting finding in that it would suggest that the AIC->DLS pathway is aversive or regulates reward learning. Did the authors perform a power analysis for this experiment to be confident that N=6-7 is sufficient?

https://doi.org/10.7554/eLife.77411.sa1

Author response

Essential revisions:

We ask that you please address all reviewer comments and especially attend to the follow.

1. In figure 4J where n=2, please increase the n to at least 6/7 or provide power analysis justifying the n used if less than 6. See Reviewer 3 comment 3 and Reviewer 1 comment 1 for further explanation.

Thank you for pointing out this weakness in our initial submission. For this revised manuscript we reran this experiment with 6 animals per group and updated Figure 4 I and J and the accompanying methods section titled “Intracranial self-stimulation” to reflect the change. We also note that the new, correctly powered experiment confirmed the previous claim that AIC inputs to the DLS do not modulate operant responding behaviors.

2. To strengthen the case for the specificity of the effect in figure 1 please repeat the experiments in Figure 1 to include female mice.

Thank you for this request. We took age-matched female mice that were injected with AAV-ChR2 into AIC and had them undergo the same 3 weeks of Drinking in the Dark to replicate the male data displayed in Figure 1 with an experimental focus on AIC inputs. We then performed whole cell patch clamp electrophysiology in DLS brain slices from these female mice. We measured optically evoked input-output responses (oEPSCs), AMPA/NMDA current ratios (oNMDA/oAMPA), and paired pulse ratios (oPPR). These data are presented in Figure 1—figure supplement 4. In contrast to males, we did not observe any effect of alcohol consumption on AIC inputs into the DLS of female mice compared to males. We also combined both male and female datasets to statistically determine if we had sex differences for these specific measures by the existence of a main effect and/or a sex x fluid interaction. We report these statistics in text from lines 189 to 204, where we note that we did not have a sex x fluid effect for oEPSCs but did note that we had a sex x fluid effect for our oNMDA/oAMPA synaptic plasticity measure. This finding further justifies the behavioral data and circuit manipulations being conducted in solely male mice.

While this is a fascinating sex difference and important data for the field, this manuscript is not specifically about exploring sex differences per se. We believe we have done our due diligence and correctly reported the existence of sex differences, or the possibility of sex differences, but the electrophysiological findings that we later modulate in vivo are only present in males. We point out that future work is needed to determine the contribution of circuit-specific changes in females at these synapses. Ultimately it will take much more work to fully elucidate sex difference circuit-specific mechanisms that we feel are far beyond the scope of this manuscript.

3. Please restrict the implication of the findings to addiction or the impact of these results to public health to the discussion only. In the discussion further expand on the use of AIC to DLS alterations as a biomarker for the development of alcohol use disorder.

We have removed language in the body of the manuscript and expanded on the implications of our findings at the end of our results and discussion from lines 553 to 587.

4. For figure 1B, please clarify in the results and methods section whether brain slices were taken 24 or 48 hours after the final alcohol session. See Reviewer 2 comment 2 for further explanation.

4b. Also please report the rationale for starting the ex-vivo recording 24 h after the last binge episode. See Reviewer 1 comment 2 for further explanation.

We have revised the text to make it clear that this was 24 hours after the last exposure in lines 102104.

We decided the best way to address this comment was to perform an additional set of experiments where we took mice that had only one single alcohol drinking session and then recordings were again made 24 hours after that single exposure. We report these new findings in Figure 1—figure supplement 5.

We found that while a single session of alcohol DID exposure does produce a directionally similar change in oEPSCs amplitudes, changes in synaptic plasticity are not equivalent to those that occur following three weeks of DID. One single DID session produced an enhanced oEPSC response, the magnitude of which was larger than that produced by 3 weeks of DID, suggesting that while there are plastic changes of the same measure, their magnitudes (and therefore mechanisms) are longitudinally different. Furthermore, measurements of oNMDA/oAMPA ratios were not different across fluid exposure, nor were oPS amplitudes, suggesting the key measurements that we saw change after three weeks of DID were not sufficiently changed by a single DID exposure. Altogether these data argue that our prior findings are not due to simply a 24-hour withdrawal from alcohol, regardless of alcohol drinking history nor are they recapitulated by a single DID exposure. These data also support the differences in behavioral responses to AIC→DLS in vivo stimulation in mice with 3 weeks of alcohol drinking history compared to those that began receiving stimulation from the very first drinking session (Figure 2—figure supplement 6) which further speaks to the nature of repeated DID sessions having a longitudinal effect on plasticity at AIC inputs in the DLS.

5. Please add DAPI to the representative picture of viral injection in figure 1A. See Reviewer 1 comment 6 for further explanation.

We tried to post-fix an AIC injection from a brain that was used for an electrophysical experiments in DAPI, but due to limitations in slice thickness (~250 um) and the microscopes we have access to, the images were, in our opinion, worse than the example images provided. Since we have performed the same injections for electrophysiological experiments and in vivo experiments (in which we used a different method of histology to slice thinner sections, check hits, and stain with DAPI for which we provide an example image) and provided hit maps for all injections that show where in the brain the AIC is across bregma levels, there should not be an issue or confusion of what circuitry we are isolating.

6. Please add statistical p-values for figure 4C. See Reviewer 3 comment 3 for further explanation. Please also ensure your manuscript complies with the eLife policies for statistical reporting: https://reviewer.elifesciences.org/author-guide/full "Report exact p-values wherever possible alongside the summary statistics and 95% confidence intervals. These should be reported for all key questions and not only when the p-value is less than 0.05."

We have added to all figures and/or their corresponding figure legends, including Figure 4C, statistical descriptions for all graphs shown, even if p values are greater than 0.05.

Please note that as we changed the analyses for oPPR as requested by Reviewer 3, we discovered there was a typo in our methods regarding our ephys parameters which should have been currents were elicited between 100 pA and 600 pA (not 200 pA to 600 pA). Secondly, removing the 25 ms interval from oPPR caused the significant interaction between fluid x interstimulus interval to change to non-significant. We have amended the text accordingly to this change in conclusion. As this was not a major part of our overall conclusions, we do not feel this detracts from the study’s primary conclusions.

In addition, in full disclosure, as we analyzed all of our new recording data in females and single drinking session males, we went back over all of our 3-week drinking male data to be absolutely sure every recording was within our allowable parameters (between 100 pA and 600 pA and less that 15% change in resistance). We identified a few cells/technical replicates that were previously missed that were outside our analysis parameters. These have been removed from the prior datasets and new n’s are reported as well as new statistical analyses results where those changes occurred. That being said, there were only a few of these recordings that were identified, and the removal of these recordings did not in any way change any conclusions (i.e. no change between significant/nonsignificant and negligible changes in effect sizes).

7. Please comment in the methods whether reported values include a correction for the liquid junction potential. See other comments section for reviewer 3.

We do not correct for liquid junction potential. We have revised the manuscript to specifically state this (line 701).

8. Please clarify the description of the optogenetic stimulation paradigm in line 213.

See other comments section for reviewer 5.

We have provided this clarification (lines 341-350).

Reviewer 1 (Recommendations for the authors):

Based on the comments above, these are the recommendations for the authors:

1. I strongly suggest increasing the number of mice to have at least 6/7 mice/group for each experiment.

Please see our response in Essential revisions 1.

2. Please report the rationale for starting the ex-vivo recording 24 h after the last binge episode. An alternative interpretation for the reported data could be that the glutamatergic synaptic adaptation observed are simply related to only one (i.e., the last) binge episode and not with the drinking history. Please discuss the possibility that these adaptations can occur with a single binge/alcohol exposure.

Please see our response in Essential revisions 4.

3. In line with the point above, it is not clear why for the electrophysiology recording the control virus is missing. This could be critical also for validating the behavioral approaches.

We appreciate the reviewer’s desire to see the control GFP virus that we used in the behavioral experiments in the ex vivo measures. We must point out though, that as much as we would love to do this experiment it is technically impossible to evoke glutamate release via blue light from solely GFP labeled AIC inputs into the DLS. The data that we present are the best that we can do within the realm of feasibility as far as comparing our in vivo and ex vivo data with each other.

4. Regarding the issue with the unilateral stimulation, ideally the authors should repeat the experiments (at least some of the behavioral controls) to demonstrate similar results. Alternatively, the authors could add a "methodological consideration" section to explain these potential limitations.

This is an interesting point and one that we considered over the course of these experiments. We will explain here our rationale for why we feel this is not a good use of animal life to prove negative data in addition to the high financial cost of doing so. (1) We show that unilateral stimulation of these AIC inputs can control drinking behaviors in specific conditions after plasticity has occurred, but not prior to that plasticity being induced. (2) There is no evidence of any lateralization effect in our data. (3) There are many experiments across the breadth of neuroscience research in which modulating striatal dynamics unilaterally alters behavior, mainly locomotion, which we also show data for being unaltered in control behavioral experiments. The idea that unilateral stimulation is not sufficient to produce a behavioral effect is not supported by our data. (4) If we agree that these synapses are being activated by blue light in vivo in similar ways that they are ex vivo in slice (where we measured the plasticity changes), the takeaway is that the alcohol-induced changes in synaptic responses to optical stimulation do not have an influence on these specific control behaviors – which is the claim we make. (5) Adding a methodological consideration for the in vivo optical stimulation discounts the in vivo effects that we identified that were modulated by unilateral stimulation in our alcohol drinking and lickometry measurements. In the future we can consider performing bilateral stimulation with newer devices that are now available, which we did not have access to when we ran these experiments. In such a study we would specifically test the hypothesis that bilateral activation of AIC input stimulation produces different behavioral responses to unilateral stimulation. Testing that hypothesis, however, is beyond the scope and intent of the present study.

5. I strongly suggest adding female mice to this manuscript to increase the impact of the results.

Please see our response in Essential revisions 2.

6. Please add DAPI to the representative picture of viral injection. It is difficult to appreciate anatomical landmarks for these pictures.

Please see our response in Essential revisions 5.

7. Please keep both introduction and discussion related to the circuit investigation and remove implications for human transition to addiction or the impact of these results on public health.

Please see our response in Essential revisions 3.

Reviewer 2 (Recommendations for the authors):

I enjoyed reading this manuscript and have a number of comments for the authors to consider.

1. In the introduction, please qualify the criteria of a binge as per NIAAA guidelines, namely 4 drinks in 2 hours for women and 5 drinks in 2 hours for men. Also, there is a recent review on the role of the insula in AUD that updates the cited Barker et al. 2015 review (see Campbell and Lawrence 2021).

Thank you. We have updated our language (lines 34-37) and added this citation (lines 70-75).

2. In the text you state that brain slices are taken 24 hours after the final alcohol session; however, figure 1B implies that slices are taken 48 hours after alcohol. Which is it?

Please see our response in Essential revisions 4.

3. In rodents silencing insula inputs to the accumbens core reduces alcohol self-administration (Jaramillo et al., 2018), and in human heavy drinkers there is increased coupling between the right anterior insula and right nucleus accumbens (Grodin et al., 2018) during high threat alcohol cues. With regards the latter point, in rats the AI mediates relapse-like behavior in a punishment associated context (Campbell et al., 2019). Have the authors performed analogous recordings at AI→accumbens synapses to compare vs the AI→DLS synapses? Also, how much spread of virus was there along the A-P axis of the AI? In this regard rodent studies suggest that functionally inhibiting the caudal portion of the insular cortex reduced alcohol consumption whereas functionally inhibiting the anterior insular cortex. was without effect (Haaranen et al., 2020; Pushparaj and Le Foll, 2015).

These experiments were designed based on our prior work that hinted that AIC inputs to DLS are especially sensitive to the effects of alcohol (Muñoz et al., Nat. Comm. 2018) as noted in the next comment by the reviewer. We have not performed recordings of AIC→accumbens synapses as a comparison, although this would be an interesting avenue for future experimentation. Regarding spread of the viral injections, we show bregma levels for spread in the AIC and discuss selection criteria for exclusion based on hits, we note that more caudal portions of the AIC decrease their density in projection to the DLS. Even so, we are doing all of our work at the terminals where the papers cited are CNO injections where collaterals begin to send projects to the BLA and other brain regions that can conflate projection-specific findings.

4. The authors conclude that "binge drinking alcohol specifically potentiates postsynaptic AMPA and NMDA glutamate receptor responses at AIC→DLS synapses". There is evidence that AI→ventral striatum projections are involved in compulsive food eating (Spierling et al., 2020). As the prior comment, have the authors assessed AI→ventral striatum synapses? Also, while I acknowledge the authors previously showed the disruption of opioid receptor mediated LTD at AIC→DLS synapses was specific to alcohol, and did not occur after sucrose binge (Munoz et al., 2018), have they assessed the specificity of the current data set in relation to alcohol vs other (non-drug or drug) rewards?

Again, we have not assessed AIC→ventral striatum synapses, although we acknowledge this will be an exciting comparison. Considering our prior lack of effect on sucrose consumption (Muñoz et al., 2018), we decided not to explore that here, although we cannot rule out that there may be a difference in these experiments. This is an area we feel is suited for future exploration.

5. Note for discussion, within the DLS in both humans and rodent models, muscarinic M4 receptors are subjected to alcohol-induced adaptations and implicated in alcohol seeking (Walker et al., 2020). Have the authors considered this possibility in terms of binge behavior?

We have specifically focused on glutamate transmission based upon our prior work. We have discussed looking at other neurotransmitter systems in the future as we will eventually work to dissect out the mechanisms underlying the alcohol-induced synaptic adaptations, which could involve M4 receptors or BNDF/p75 as the reviewer suggested here and in their question 9.

6. Figure 2 G,H – why are the data presented as %? Moreover, in figure 2H does alcohol intake increase across weeks in the eGFP group?

Data are percent change within animal to help remove baseline differences in drinking intakes, and to measure solely the effect of AIC input modulation of the DLS on future intake behaviors. We also show the raw data in supplemental figures. Intakes are relatively stable, but obviously they are mice and there will always be an inherent noisiness to the data even at larger N’s. Yes, eGFP increases across weeks in regards to alcohol intake. We did not compute any within group statistics for subsets of data as that was not the point of our study, which was to assess between group differences.

7. Supplemental figure 7B – in the ChR2 group the baseline alcohol intake drops both times prior to the final two binge tests (see sessions 21-30). Does this impact the % change data? Over the same period the baseline intake in the eGFP group is more stable.

We calculate from week 3 and are using sums, so yes they are taken into account. There is likely some cumulative effect of ChR2 in alcohol decreasing by session, but we were not powered to detect it, which is another reason why we chose to measure weekly percent change.

8. Supplemental figure 9 – the drinking profile of the LED off group is unstable, with intake seeming to escalate with time. Is this the case and if so why?

The LED OFF animals were not pretrained to drink from the bottles, which we explain in the methods. We pretrained the LED ON animals on water drinking to ensure any potential decreases with consumption were not related to the inability of animals to drink out of, or neophobia to, the bottles.

9. In terms of a molecular mechanism, have the authors considered investigating the potential role of BDNF / p75 (see Darcq et al., 2016)?

Please see our response to this reviewer’s question 5.

10. Binge drinking in adolescent females is a growing problem, have the authors studied female mice?

Please see our response in Essential revisions 2.

Reviewer 3 (Recommendations for the authors):

1) PPR at 25 ms should not be included. The very low values reflect poor fidelity of ChR2 at stimulation frequencies >30 Hz, rather than biological properties of AIC->DLS synapses. Also, the interpretation of these data in line 120 should use specific language to indicate directionality in place of "alter".

We removed 25 ms, which negated the statistical effect. We changed our conclusions accordingly.

2) I suggest switching orders of graphs throughout so that the control groups (Water, eGFP, etc) precede the experimental groups (Alcohol, ChR2, etc)

We thank the reviewer for their style recommendations. Some of the authors agreed with this reviewer’s opinion, but others disagreed. We put it up for a vote in a democratic process and the result of the vote to decide was to keep the style the same. We hope this does not offend the reviewer.

3) I do not understand how the DID intake values for the Water group are tenfold higher than the Alcohol group. Also, "g/kg" is an unusual way to display Water Intake, I think "mL" is more informative and conventional.

The what looks like higher water consumption can likely be accounted for by the fact that the alcohol group is drinking a solution that is only 20% alcohol while the water group is drinking a solution that is 100% water. Once that is accounted for, any discrepancies from what one would expect are likely due to the complex physiological responses to consuming each substance.

While we acknowledge that g/kg of water might be unconventional, to train our machine learning model the two fluids had to have the same units. In addition, the lickometer measures are validated by g/kg as well. In choosing whether to use ml/kg or g/kg for our inputs we felt that g/kg of alcohol was more important to report than ml/kg of water. Our code and thus the math we use to produce these values is publicly available for anyone that wants to convert the data for their own analyses.

4) The representative AMPA/NMDA ratio trace in Figure 1F displays significant AMPA receptor rectification. MSNs do not typically display significant AMPAR rectification under control conditions. Furthermore, the internal solution does not appear to contain spermine, so voltage dependent block of CP-AMPARs should not be a major factor. Did the authors correct for the liquid junction potential?

Thank you for pointing this, we have replaced the traces, because the appearance of what looks like rectification could be due to methodological reasons. We did not correct for liquid junction potential. We have added that point to our methods.

5) The description of the optogenetic stimulation paradigm (line 213) is misleading. As written, it reads as if most of the animals underwent closed-loop stimulation but only a minor portion were open-loop and eventually folded into the main data set. By contrast, the cohort was split n=4/4 open/closed as shown in Figure S6. This section should be re-written accordingly.

We have re-written this for better accuracy to indicate half of the animals experienced the paradigm (see lines 341 – 350).

6) The authors should be commended for assessing the effects of their optogenetic stimulation on long-term plasticity in brain slices. Experiments like this are rarely performed but can provide useful, if not necessary, information for interpreting the effects of in vivo optogenetic stimulation. While no long-term effects were detected, it appears to me as if there could be an Alcohol x Picrotoxin interaction immediately after the 20 Hz stimulation. It seems as if picrotoxin might have decreased the post-tetanus potentiation in slices from Water but not Alcohol mice. If so, this could provide some insight into changes in AIC->DLS feedforward inhibition following DID. Also, these experiments should be described in greater detail in the Methods.

Thank you. We have described this in more detail in the methods and added a comment about these data in relation to feedforward inhibition. We also tested those timepoints (both just the peak right after the stimulation and the total time before the response returns to baseline) and there are no significant effects. Picrotoxin has an effect on oPS amplitude, but there is no interaction with fluid type.

7) While the summarized oPS timecourses are completely blocked by NBQX, the representative oPS traces do not appear to be.

In a further subset of experiments, we applied TTX after NBQX to be sure that NBQX was indeed blocking the entire response. We did not see any further change as a result of TTX application. What may appear to be an incomplete block, is likely part of the stimulus artifact. Our software accounts for this in determining the oPS amplitude. Please see Author response image 1.

Author response image 1

8) There seems to be an effect of AIC->DLS stimulation during the first DID cycle in Figure S9.

This is due to methodology and not an effect of stimulation. Please see our response to reviewer 2 point 8 for an explanation.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

1. Amend the title of the manuscript to "The role of anterior insular cortex in binge alcohol drinking" (or similar).

We have updated the title to remove the word govern.

2. Add to the main text that unilateral opto manipulation was used.

We’ve added unilateral to the main text in lines 290 and 348.

3. Include missing citation on Line 1002.

We have rechecked all citations and the reference section should be inclusive of all in text and methods citations.

4. Add the "n" values to the figure 3 legend. Please double-check that all figure legends contain such "n" value information.

We have added this information to the figure 3 legend.

5. Perform power analysis to determine whether the ICSS dataset is adequately powered. State the result (whether or not) concisely in the text. For example (p-value, study sufficiently powered as determined by power analysis).

We performed a power analysis using the data in Figure 4 J to explicitly determine if we were adequately powered to discover a between factor effects for Virus with our current data. To do so, we used the partial eta squared value from the mixed ANOVA to directly compute the Virus effect size in G*Power 3.1. Using that effect size, sample size, number of groups, number of repeated measures, and computing the correlation amongst the repeated measures data, we achieved a power of = 0.9898339, indicating our sample size was correctly powered to discover a main effect of Virus. Also, we were also curious to see if we had the power to discover a within-between interaction and repeated the same approach using the partial eta squared value of the interaction of Virus x Session from our mixed ANOVA analysis to compute effect size, the nonsphericity epsilon value, and the additional values noted above. We achieved a power of = 0.5582789, suggesting although we are powered to discover a main effect for Virus, our ability to discover a Virus x Session interaction is below the threshold of = 0.8.

Thus, we are confident that our claim AIC→DLS photoexcitation does not alter operant responding remains valid given the experimental procedure employed in this manuscript. We note that future analyses may need to address whether stimulation of this pathway over time produces differential effects in operant responding in different contexts. We have added the main effects β value to the Figure 4 legend after the p value for Figure 4J.

6. Add missing citations to the conclusion.

We have added citations and additional text to the conclusion.

https://doi.org/10.7554/eLife.77411.sa2

Article and author information

Author details

  1. David L Haggerty

    Department of Pharmacology & Toxicology, Indiana University School of Medicine, Indianapolis, United States
    Contribution
    Conceptualization, Resources, Data curation, Software, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1455-2557
  2. Braulio Munoz

    Department of Pharmacology & Toxicology, Indiana University School of Medicine, Indianapolis, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  3. Taylor Pennington

    Department of Pharmacology & Toxicology, Indiana University School of Medicine, Indianapolis, United States
    Contribution
    Data curation, Formal analysis, Methodology
    Competing interests
    No competing interests declared
  4. Gonzalo Viana Di Prisco

    Department of Pharmacology & Toxicology, Indiana University School of Medicine, Indianapolis, United States
    Contribution
    Data curation
    Competing interests
    No competing interests declared
  5. Gregory G Grecco

    1. Department of Pharmacology & Toxicology, Indiana University School of Medicine, Indianapolis, United States
    2. Medical Scientist Training Program, Indiana University School of Medicine, Indianapolis, United States
    Contribution
    Conceptualization, Software, Formal analysis, Funding acquisition, Validation, Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0700-8633
  6. Brady K Atwood

    1. Department of Pharmacology & Toxicology, Indiana University School of Medicine, Indianapolis, United States
    2. Stark Neurosciences Research Institute, Indiana University School of Medicine, Indianapolis, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Methodology, Project administration, Writing – review and editing
    For correspondence
    bkatwood@iu.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7441-2724

Funding

National Institutes of Health (R01AA027214)

  • Brady K Atwood

National Institutes of Health (F31AA029297)

  • David L Haggerty

National Institutes of Health (F30AA028687)

  • Gregory G Grecco

National Institutes of Health (T32AA07462)

  • David L Haggerty

Indiana University Health

  • Brady K Atwood

Indiana University-Purdue University Indianapolis (Stark Neurosciences Research Institute)

  • David L Haggerty
  • Gregory G Grecco
  • Brady K Atwood

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Kaitlin C Reeves, Brandon M Fritz, and Fuqin Yin for experimental assistance and Erin A Newell for technical assistance. We also thank Lex Kravitz and Ethan Tyler for modified drawings that were accessed via scidraw.io. National Institutes of Health grant R01AA027214 (BKA). National Institutes of Health grant F31AA029297 (DLH). National Institutes of Health grant F30AA028687 (GGG). National Institutes of Health grant T32AA07462 (DLH). Stark Neurosciences Research Institute (BKA, DLH, GGG). Indiana University Health (BKA).

Ethics

Animal care and experimental protocols for this study were approved by the Institutional Animal Care and Use Committee at the Indiana University School of Medicine (IACUC #: 19017) and all guidelines for ethical protocols and care of experimental animals established by the National Institutes of Health (Maryland, USA) were followed.

Senior Editor

  1. Kate M Wassum, University of California, Los Angeles, United States

Reviewing Editor

  1. Michelle Antoine, National Institute on Alcohol Abuse and Alcoholism (NIAAA), United States

Reviewer

  1. Andrew Lawrence, The Florey Institute of Neuroscience and Mental Health, Australia

Publication history

  1. Preprint posted: January 28, 2022 (view preprint)
  2. Received: January 28, 2022
  3. Accepted: July 27, 2022
  4. Version of Record published: September 13, 2022 (version 1)

Copyright

© 2022, Haggerty et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,245
    Page views
  • 257
    Downloads
  • 4
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. David L Haggerty
  2. Braulio Munoz
  3. Taylor Pennington
  4. Gonzalo Viana Di Prisco
  5. Gregory G Grecco
  6. Brady K Atwood
(2022)
The role of anterior insular cortex inputs to dorsolateral striatum in binge alcohol drinking
eLife 11:e77411.
https://doi.org/10.7554/eLife.77411

Further reading

    1. Neuroscience
    Sarah Mondoloni, Claire Nguyen ... Alexandre Mourot
    Research Article Updated

    Nicotine intake is likely to result from a balance between the rewarding and aversive properties of the drug, yet the individual differences in neural activity that control aversion to nicotine and their adaptation during the addiction process remain largely unknown. Using a two-bottle choice experiment, we observed considerable heterogeneity in nicotine-drinking profiles in isogenic adult male mice, with about half of the mice persisting in nicotine consumption even at high concentrations, whereas the other half stopped consuming. We found that nicotine intake was negatively correlated with nicotine-evoked currents in the interpeduncular nucleus (IPN), and that prolonged exposure to nicotine, by weakening this response, decreased aversion to the drug, and hence boosted consumption. Lastly, using knock-out mice and local gene re-expression, we identified β4-containing nicotinic acetylcholine receptors of IPN neurons as molecular and cellular correlates of nicotine aversion. Collectively, our results identify the IPN as a substrate for individual variabilities and adaptations in nicotine consumption.

    1. Neuroscience
    Movitz Lenninger, Mikael Skoglund ... Arvind Kumar
    Research Article Updated

    According to the efficient coding hypothesis, sensory neurons are adapted to provide maximal information about the environment, given some biophysical constraints. In early visual areas, stimulus-induced modulations of neural activity (or tunings) are predominantly single-peaked. However, periodic tuning, as exhibited by grid cells, has been linked to a significant increase in decoding performance. Does this imply that the tuning curves in early visual areas are sub-optimal? We argue that the time scale at which neurons encode information is imperative to understand the advantages of single-peaked and periodic tuning curves, respectively. Here, we show that the possibility of catastrophic (large) errors creates a trade-off between decoding time and decoding ability. We investigate how decoding time and stimulus dimensionality affect the optimal shape of tuning curves for removing catastrophic errors. In particular, we focus on the spatial periods of the tuning curves for a class of circular tuning curves. We show an overall trend for minimal decoding time to increase with increasing Fisher information, implying a trade-off between accuracy and speed. This trade-off is reinforced whenever the stimulus dimensionality is high, or there is ongoing activity. Thus, given constraints on processing speed, we present normative arguments for the existence of the single-peaked tuning organization observed in early visual areas.