1. Neuroscience
Download icon

mTORC1 in the orbitofrontal cortex promotes habitual alcohol seeking

  1. Nadege Morisot
  2. Khanhky Phamluong
  3. Yann Ehinger
  4. Anthony L Berger
  5. Jeffrey J Moffat
  6. Dorit Ron  Is a corresponding author
  1. University of California, San Francisco, United States
Research Article
  • Cited 1
  • Views 989
  • Annotations
Cite this article as: eLife 2019;8:e51333 doi: 10.7554/eLife.51333

Abstract

The mechanistic target of rapamycin complex 1 (mTORC1) plays an important role in dendritic translation and in learning and memory. We previously showed that heavy alcohol use activates mTORC1 in the orbitofrontal cortex (OFC) of rodents (Laguesse et al., 2017a). Here, we set out to determine the consequences of alcohol-dependent mTORC1 activation in the OFC. We found that inhibition of mTORC1 activity in the OFC attenuates alcohol seeking and restores sensitivity to outcome devaluation in rats that habitually seek alcohol. In contrast, habitual responding for sucrose was unaltered by mTORC1 inhibition, suggesting that mTORC1’s role in habitual behavior is specific to alcohol. We further show that inhibition of GluN2B in the OFC attenuates alcohol-dependent mTORC1 activation, alcohol seeking and habitual responding for alcohol. Together, these data suggest that the GluN2B/mTORC1 axis in the OFC drives alcohol seeking and habit.

Introduction

mTORC1 is a multiprotein complex that contains the serine/threonine protein kinase mTOR, the regulatory associated protein of TOR (Raptor), and other enzymes and adaptor proteins (Lipton and Sahin, 2014). Upon activation, mTORC1 phosphorylates eIF4E-binding protein (4E-BP) and the ribosomal protein S6 kinase (S6K), which in turn phosphorylates its substrate, S6 (3). These phosphorylation events lead to the translation of a subset of mRNAs to proteins (Lipton and Sahin, 2014; Saxton and Sabatini, 2017). In the CNS, mTORC1 is responsible for the local dendritic translation of synaptic proteins (Buffington et al., 2014; Santini et al., 2014). As such, mTORC1 plays an important role in synaptic plasticity, and learning and memory (Hoeffer and Klann, 2010).

Increasing lines of evidence in rodents suggest that mTORC1 is a key player in mechanisms underlying alcohol use disorder (AUD) (Neasta et al., 2014). For instance, excessive alcohol intake and reinstatement of alcohol place preference activate mTORC1 in the rodent nucleus accumbens (NAc) (Laguesse et al., 2017a; Neasta et al., 2010; Beckley et al., 2016; Ben Hamida et al., 2019), resulting in the translation of synaptic proteins, which in turn produce synaptic and structural adaptations that drive and maintain heavy alcohol use and relapse (Laguesse et al., 2017a; Neasta et al., 2010; Beckley et al., 2016; Ben Hamida et al., 2019; Liu et al., 2017). Repeated cycles of voluntary binge drinking of alcohol and withdrawal also produce a robust and sustained activation of mTORC1 in the OFC of rodents (Laguesse et al., 2017a); however, the behavioral consequences of this activation are unknown.

The OFC integrates sensory and reward information (Wallis, 2007) and is an essential player in decision making (Wallis, 2007; Meyer and Bucci, 2016; Baltz et al., 2018), in stimulus-outcome association (Gremel and Costa, 2013; Ostlund and Balleine, 2007; Gourley et al., 2013), in updating the value of predicted outcomes (Baltz et al., 2018; Fiuzat et al., 2017), in goal-directed (Gremel and Costa, 2013; Bradfield et al., 2015; Gremel et al., 2016), and compulsive behaviors (Ahmari et al., 2013; Pascoli et al., 2018). Exposure to drugs of abuse impairs the performance of OFC-dependent behavioral tasks (Schoenbaum and Shaham, 2008), and aberrant neuroadaptations induced by drugs of abuse in the OFC contribute to drug-seeking and compulsive drug taking (Schoenbaum and Shaham, 2008; Everitt et al., 2007; Lüscher, 2016). For example, withdrawal from alcohol vapor exposure produces morphological and cellular alterations in rodents (McGuier et al., 2015; Nimitvilai et al., 2016). Inactivation of the OFC enhances alcohol consumption in alcohol-dependent mice (den Hartog et al., 2016), and chronic alcohol use alters the protein landscape in the primate OFC (Nimitvilai et al., 2017). In humans, degraded OFC white matter and reduced neuronal density are associated with alcohol-dependence (Pfefferbaum and Sullivan, 2005; Miguel-Hidalgo et al., 2006), and presentation of alcohol-associated cues to alcohol-dependent subjects elicits OFC activation (Dom et al., 2005; Claus et al., 2011; Filbey et al., 2008).

Here, using a rat model system, we set out to examine whether mTORC1 in the OFC contributes to the mechanisms underlying AUD.

Results

mTORC1 in the OFC contributes to alcohol seeking

We first examined whether activation of mTORC1 in the OFC contributes to the maintenance of heavy alcohol use. Rats that underwent 7 weeks of intermittent access to 20% alcohol in a 2-bottle choice paradigm (IA-20%2BC) received an intra-OFC bilateral infusion of vehicle or the mTORC1 inhibitor, rapamycin (50 ng/μl), 3 hr prior to a drinking session and alcohol consumption was monitored (Neasta et al., 2010) (Figure 1—figure supplement 1). As shown in Figure 1—figure supplement 2 (Source data Figure 1), intra-OFC administration of rapamycin did not impact voluntary intake of alcohol. Next, rats that underwent IA-20%2BC for 7 weeks were trained to self-administer 20% alcohol in an operant self-administration paradigm (Carnicella et al., 2008). Upon establishing a baseline of alcohol lever presses, rats received a bilateral infusion of vehicle or rapamycin (50 ng/μl) into the OFC 3 hr prior to an OSA session, and the number of lever presses and alcohol consumption were measured (Neasta et al., 2010). As shown in Figure 1a–c, Source data Figure 1, inhibition of mTORC1 in the OFC failed to alter the number of lever presses (Figure 1a–b, Source data Figure 1) and alcohol consumed (Figure 1c, Source data Figure 1). Inactive lever pressing and latencies to the first and last lever presses were also unaltered (Figure 1—figure supplements 3, Source data Figure 1). We then determined whether mTORC1 plays a role in alcohol seeking by testing if rapamycin alters lever presses during an extinction session. Intra-OFC administration of rapamycin suppressed the number (Figure 1d–e, Source data Figure 1) of rats’ responding on the lever that was previously associated with alcohol. The reduction in alcohol seeking was not due to alterations in locomotion, as inter-response intervals were similar in vehicle and rapamycin-treated rats (Figure 1—figure supplements 4, Source data Figure 1). Together, these data suggest that mTORC1 in the OFC does not contribute to alcohol drinking per se, but does promote alcohol seeking.

Figure 1 with 4 supplements see all
Inhibition of mTORC1 in the OFC reduces alcohol seeking.

(a–c) Intra-OFC infusion of rapamycin does not alter self-administration of alcohol. Rats underwent 7 weeks of IA-20%2BC and were then trained to self-administer 20% alcohol using a FR3 schedule. Vehicle (white) or rapamycin (50 ng/μl, red) was infused bilaterally in the OFC 3 hr before a 30 min self-administration session. (a) RM ANOVA of cumulative lever presses did not identify a significant main treatment effect (F1,12=0.26, p>0.05). Two-tailed paired t-test revealed that the number of active lever presses (b) (t6 = 1.40, p>0.05), and amount of alcohol consumed (c) (t6 = 1.51, p>0.05) did not differ between treatment groups. (d–e) Intra-OFC infusion of rapamycin inhibits lever presses during extinction. Vehicle (white) or rapamycin (50 ng/μl, red) was infused in the OFC 3 hr before a 30 min extinction session, and responses on the previously active lever were recorded. (d) RM ANOVA revealed a significant main treatment effect (F1,16=4.40, p=0.05) of cumulative lever presses. (e) Two-tailed paired t-test revealed that the number of lever presses (per 5 min) t8 = 3.31, p<0.05, were reduced in the rapamycin-treated animals. Data are presented as individual values and mean ± SEM. *p<0.05. (a–c) n = 7, (d–e) n = 9.

mTORC1 in the OFC participates in habitual responding for alcohol

Reward seeking can be driven by goal-directed or habitual compulsive behaviors (Corbit and Janak, 2016; Seif et al., 2013; Pascoli et al., 2015). We next examined whether mTORC1 in the OFC plays a role in goal-directed or habitual alcohol seeking using a satiety outcome devaluation procedure. Rats that first underwent 7 weeks of IA-20%2BC were then trained to lever press for 20% alcohol using either a random ratio (RR) or random interval (RI) schedule of reinforcement, which biases lever responding toward goal-directed (RR) or habitual (RI) actions (Gremel and Costa, 2013; Gremel et al., 2016; Corbit et al., 2012) (Timeline. Figure 2a). RR and RI trained rats showed a similar number of presses throughout the training (Figure 2—figure supplements 1a,Source data Figure 2), however alcohol consumption was slightly higher during RI training (Figure 2—figure supplements 1b,Source data Figure 2), presumably due to fewer lever presses being required to obtain the same volume of alcohol, compared with RR training. Rats trained to self-administer alcohol in a goal-directed manner are sensitive to outcome devaluation, while habitually-trained rats are not, i.e. animals who habitually press a lever to obtain a reward will continue to press that lever, even when the relative value of the reward is reduced due to satiety (Gremel and Costa, 2013; Gremel et al., 2016; Corbit et al., 2012; Rossi and Yin, 2012). Thus, we conducted a devaluation procedure to differentiate between goal-directed and habitual responding (Figure 2b). On the devalued (D) day, rats had 30 min of home-cage access to 20% alcohol, the reward previously associated with lever responding, and on the non-devalued (ND) day, rats had 30 min of home-cage access to 1% sucrose, a reward that was not previously associated with lever responding (Figure 2b). Rats then underwent a 10 min extinction test, during which lever presses were recorded (Figure 2b). In order to minimize the effects of individual variability in overall lever pressing, and to graphically represent the distribution of lever presses between ND and D, we normalized lever-pressing data such that 50% indicates identical lever pressing on both days as described in Baltz et al. (2018); Gremel and Costa (2013); Gremel et al. (2016). As shown in Figure 2c (Source data Figure 2), rats trained on a RR schedule exhibit a decrease in responding following alcohol devaluation, while RI-trained rats remained insensitive to devaluation. These data show that RR training produces goal directed alcohol seeking whereas RI training produces habitual responding for alcohol.

Figure 2 with 3 supplements see all
Inhibition of mTORC1 in the OFC attenuates habitual responding for alcohol.

(a) Timeline of experiment. Rats underwent 7 weeks of IA20%−2BC. Rats were then trained to operant self-administer 20% alcohol during three overnight FR1 sessions and were pseudo-randomly assigned to two groups. One group was subjected to a progressive RI reinforcement schedule, while the other was assigned to a RR schedule. Following a stable RI-30 or RR-3 responding, rats underwent alcohol devaluation testing depicted in (b). Upon establishing the behavior, rats underwent stereotaxic surgery to bilaterally implant a guide cannula. One week later, RI-30 or RR-3 training was resumed for 2 weeks, after which rats received microinjections of vehicle or drug in a counter-balanced manner, prior to the devaluation test. (b) Alcohol devaluation test. On test day, rats were pre-fed with 20% alcohol for 30 min in their home cage during devalued day (D) or with 1% sucrose in the non-devalued day (ND) (Table 1), and normalized lever pressing (percent of total presses on D or ND, see Materials and methods) was used to evaluate goal-directed or habitual responding during a 10 min extinction session. (c) RR trained animals are goal-directed whereas RI trained animals show habitual behavior. RR trained rats (green) press less during devalued day as compared to RI trained animals (red). Two-way ANOVA with Sidak post hoc analysis indicates a significant decrease in normalized lever pressing on D vs ND in RR-trained rats (p=0.0018), but not in RI-trained rats (p=0.8014) (Main effect of devaluation: F (1, 20)=9.391, p=0.0061; reinforcement schedule X devaluation: F (1, 20)=4.736, p=0.0417). (d) Intra-OFC infusion of rapamycin does not alter goal-directed alcohol seeking. Rats that were trained using a RR schedule of reinforcement received an intra-OFC infusion of vehicle (hatched green) or rapamycin (50 ng/μl) (green) 3 hr before the alcohol devaluation procedure as in (b), and the number of lever presses were measured during extinction. Total lever press-based normalization indicates significant decreases in lever pressing between ND and D days in both vehicle-treated and rapamycin-treated RR-trained rats (Main effect of devaluation: F(1, 20)=8.826, p=0.0076). (e) Intra-OFC infusion of rapamycin reduces habitual lever presses. Rats that were trained using a RI schedule of reinforcement received an intra-OFC infusion of vehicle (hatched red) or rapamycin (50 ng/μl) (red) 3 hr before the alcohol devaluation procedure as in (b), and the number of lever presses were measured during extinction. Total lever press-based normalization indicates significant decreases upon intra-OFC infusion of rapamycin- (p=0.0202), but not vehicle (p=0.8006) in RI trained rats on D compared to ND days (Main effect of devaluation: F(1,16)=6.189, p=0.0243). Data are presented as mean ± SEM. *p<0.05, **p<0.01 . (c) n = 8–10, (d) n = 11, (e) n = 9.

Next, to test whether mTORC1 in the OFC participates in goal-directed or habitual alcohol seeking, rapamycin (50 ng/μl) or vehicle were infused bilaterally, 3 hr prior to a devaluation session, and lever presses in RR- and RI-trained groups were examined (Figure 2a, Figure 2—figure supplement 2, Table 1, Source data Figure 2). Rapamycin administration did not alter home cage alcohol intake prior to the extinction session in either group (Figure 2—figure supplements 3—source data 1, Figure 2). Intra-OFC administration of rapamycin did not impact responding in RR-trained (goal-directed) cohorts, as lever presses were reduced in rats that were pre-fed with alcohol and treated with either vehicle or rapamycin (Figure 2d, Source data Figure 2). In contrast, inhibition of mTORC1 in the OFC reduced responding in the RI-trained (habitual) rats in the devalued condition (Figure 2e, Source data Figure 2). Critically, the reduction in responding in the rapamycin-treated RI-trained rats paralleled the responding of RR-trained animals (Figure 2d vs. 2e, Source data Figure 2). These data suggest that mTORC1 inhibition in the OFC restores goal-directed alcohol seeking in animals trained to habitually self-administer alcohol.

Table 1
Mean consumption of sucrose, alcohol and water during outcome devaluation.

Rats undergoing alcohol seeking probe trials were pre-fed with sucrose (non-devalued test) or alcohol (devalued test), and water 30 min prior to a 10 min extinction session.

Alcohol Devaluation
TrainingRIRRRI
GroupVehiclerapamycinVehiclerapamycinVehicleRo25
Non-devalued day1% Sucrose (ml/kg)1.42.01.21.92.74.5
±±±±±±
0.50.80.40.61.50.9
Water (ml/kg)0.70.61.20.81.02.5
±±±±±±
0.30.10.70.10.21.2
Devalued day20% Alcohol (g/kg)0.70.71.00.70.91.0
±±±±±±
0.10.10.10.10.10.2
Water (ml/kg)0.40.60.70.61.00.9
±±±±±±
0.10.10.10.10.40.2
Sucrose Devaluation
TrainingRI
GroupVehiclerapamycin
Non-devalued day20% Alcohol (g/kg)0.90.8
±±
0.10.1
Water (ml/kg)0.91.1
±±
0.10.1
Devalued day1% Sucrose (ml/kg)4.55.1
±±
0.71.4
Water (ml/kg)1.20.9
±±
0.20.1

To determine whether mTORC1 in the OFC plays a role in habitual responding to a natural reward, animals were trained to self-administer sucrose under an RI schedule of reinforcement, and lever presses were measured in rats treated with rapamycin (50 ng/μl) or vehicle (Figure 3—figure supplement 1) during an extinction session in animals that were first offered sucrose (devalued) or alcohol (non-devalued) in their home cage (Timeline, Figure 3a–b). Rapamycin administration did not alter home cage sucrose intake prior to the extinction session (Figure 3—figure supplement 2, Table 1, Source data Figure 3). As shown in Figure 3c (Source data Figure 3), habitual responding for sucrose was similar in the vehicle- and rapamycin-treated animals. Together, these data suggest that mTORC1’s contribution to habitual behavior is not generalized to all rewarding substances.

Figure 3 with 2 supplements see all
Habitual sucrose responding is not affected by mTORC1 inhibition in the OFC.

(a) Timeline of experiment. Rats underwent 7 weeks of IA20%−2BC and were then trained to operant self-administer 8% sucrose. Sucrose concentration was progressively reduced to 1% during three overnight FR1 sessions. Rats were then subjected to a RI reinforcement schedule. Drinking data are detailed in Table 1. Following a stable RI-30 responding, rats underwent alcohol devaluation testing depicted in (b). Upon establishing the behavior, rats underwent stereotaxic surgery to bilaterally implant a guide cannula. One week later, RI-30 training was resumed for 2 weeks, after which rats received microinjections of vehicle or rapamycin (50 ng/μl) in a counter-balanced manner, prior to the devaluation test. (b) Sucrose devaluation test. Habitual behavior was probed in an identical way to the alcohol-trained cohorts with the exception that sucrose pre-feeding was considered the devaluing context and alcohol was the non-devalued substance. (c) Two-way RM ANOVA failed to show a significant effect of repeated testing (Main effect of devaluation: F(1, 12)=0.006464, p>0.9999; treatment X devaluation: F(1, 12)=0.7257, p=0.4110) on lever presses between devalued and non-devalued days when habitually responding animals were pretreated with either vehicle or rapamycin. n = 7.

Alcohol-dependent mTORC1 activation in the OFC requires GluN2B

Next, we set out to elucidate the mechanism by which long-term heavy alcohol use activates mTORC1 in the OFC. Glutamatergic inputs from cortical and subcortical structures project to the OFC (Schoenbaum et al., 2003; Lichtenberg et al., 2017). Extracellular glutamate content is elevated in cortical regions of humans and rats during alcohol withdrawal (Hermann et al., 2012). Glutamate binds the N-methyl D-Aspartate receptor (NMDAR), which is a known target of alcohol (Morisot and Ron, 2017). For example, alcohol exposure promotes the phosphorylation of the GluN2B subunit of the NMDA receptor in the hippocampus and dorsomedial striatum (DMS) resulting in the activation of GluN2B-containing NMDARs in these two brain regions (Morisot and Ron, 2017). Stimulation of GluN2B activates the guanine nucleotide releasing factor 1 (GRF1), the activator of H-Ras (Krapivinsky et al., 2003) (Figure 4—figure supplement 1). Alcohol activates H-Ras in the NAc (Ben Hamida et al., 2012), as well as AKT in striatal regions (Neasta et al., 2011; Laguesse et al., 2018), and in the OFC (Laguesse et al., 2017a). Since the H-Ras/AKT axis is upstream of mTORC1 (Manning and Toker, 2017) (Figure 4—figure supplement 1), we hypothesized that alcohol stimulates GluN2B in the OFC resulting in mTORC1 activation.

First, we examined whether alcohol increases GluN2B phosphorylation in the OFC. Rats were subjected to 7 weeks of IA-20%2BC and the phosphorylation of GluN2B was measured at the end of last 24 hr of deprivation session (Table 2). Rats that drank water only for the same duration of time were used as control. As shown in Figure 4a–b, Source data Figure 4a-b, alcohol drinking and deprivation produced a robust increase in GluN2B phosphorylation. In contrast, the total level of GluN2B was unaltered by alcohol exposure (Figure 4a–b, Source data Figure 4a-b). Together, these data suggest that repeated cycles of alcohol binge and withdrawal enhances the activity of GluN2B containing NMDARs in the OFC.

Figure 4 with 2 supplements see all
GluN2B in the OFC is required for mTORC1 activation by alcohol.

(a–b) Alcohol increases GluN2B phosphorylation in the OFC. Rats underwent 7 weeks of IA-20%2BC resulting in an average of alcohol intake of 5.01 g/kg/24 hr (Table 2). The OFC was dissected at the end of the last 24 hr alcohol withdrawal session, and GluN2B phosphorylation and protein levels were determined by western blot analysis, with actin used as a loading control. (a) Representative images depicting GluN2B phosphorylation (top) and total protein (middle) in the OFC of alcohol consuming vs. water only consuming rats. (b) ImageJ was used for optical density quantification. Data are expressed as the average ratio ± S.E.M of phospho-GluN2B to GluN2B (middle) or GluN2B to actin (right) and expressed as percentage of the water control. Significance was determined using unpaired t-test. Alcohol increased GluN2B phosphorylation (t10 = 5.829, p<0.001), but did not alter GluN2B levels (t10 = 2.025, p>0.05). (c–d) Stimulation of NMDA receptors activates mTORC1 in the OFC. Vehicle (0.1% DMSO) or NMDA (25 μM) was applied to OFC slices of naïve rats for 3 min. Phosphorylation of CaMKII, S6K and S6 were determined by western blot analysis. Total levels of the proteins were also measured, and actin was used as a loading control. (c) Representative blots of CaMKII (left), S6K (middle), and S6 (right) phosphorylation in OFC slices treated with vehicle or NMDA. (d) ImageJ was used for optical density quantification. Data are expressed as the average ratio ± S.E.M of phospho-CaMKII to CaMKII, phospho-S6K to S6K and phospho-S6 to S6, and are expressed as percentage of vehicle. Significance was determined using two-tailed unpaired t-test. (CaMKII: t12 = 5.61, p<0.001; S6K: t12 = 4.77, p<0.001; S6: t12 = 4.51, p<0.001). (e–j) Alcohol-dependent mTORC1 activation is attenuated by the GluN2B inhibitor, Ro25-6981. Rats underwent 7 weeks of IA-20%2BC resulting in an average of alcohol intake of 4.37 g/kg/24 hr (Table 2). Rats received an intra-OFC bilateral infusion of vehicle (saline) or Ro25-6981 (Ro25, 10 μg/μl), 15 min prior to the beginning of the last water only session and the OFC was dissected 24 hr later. Water only consuming rats were used as controls. (e–g) Representative images of CaMKII (e), S6 (f) and 4E-BP (g) phosphorylation in water vs. alcohol-exposed animals that were pre-treated with vehicle or Ro25-6981. (h–j) ImageJ was used for optical density quantification. Data are expressed as the average ratio ± S.E.M of phospho-CaMKII to CaMKII (h), phospho-S6 to S6 (i), and phospho-4E-BP to 4E-BP (j), and are expressed as percentage of water + vehicle group. Two-way RM ANOVA showed a significant effect of alcohol on phospho-CaMKII (F1,20=24.66, p<0.0001), phospho-S6 (F1,20=21.92, p=0.0001), and phospho-4E-BP (F1,20=48.23, p<0.0001). In addition, two-way RM ANOVAs indicated a main effect of drug treatment for phospho-CaMKII (F1,20=48.51, p<0.0001), phospho-S6 (F1,20=4.58, p<0.05), and phospho-4E-BP (F1,20=8.49, p<0.01), as well as treatment x alcohol interaction for phospho-CaMKII (F1,20=18.57, p<0.01), phospho-S6 (F1,20=7.559, p<0.05), and phospho-4E-BP (F1,20=13.22, p<0.01). Tukey’s multiple comparison post hoc analysis revealed that alcohol increased phospho-CaMKII (p<0.001), pS6 (p<0.001), and phospho-4E-BP (p<0.001) in vehicle-treated animals, and that pretreatment with Ro25-6981 reduced the phosphorylation of CaMKII (p<0.0001), S6 (p<0.05), and 4E-BP (p<0.001). *p<0.05, ***p<0.001. (a–b, e–j) n = 6 per group, (c–d) n = 7 per group.

Table 2
Individual alcohol drinking data from rodents used for biochemical experiments.

Individual alcohol drinking data of the final 4 sessions of IA-2BC20%. Alcohol intake is expressed as mean ± S.E.M.

Figure 4a, bDrinking paradigm IA20%−2BC 24h-withdrawalRat numberLast four drinking session (g/kg /24 )
15.97
Animal numbers26.40
n = 634.00
45.84
54.23
63.60
Mean ± S.E.M.5.01 ± 0.49
Figure 4e,f and gDrinking paradigm IA20%−2BC 24h-withdrawalRat numberLast four drinking session (g/kg/24 h)
Animal numbers n = 1215.35
25.52
35.54
44.84
56.41
63.16
74.37
83.45
93.48
106.24
112.71
125.72
Mean ± S.E.M.4.73 ± 0.37

Next, we tested whether NMDAR stimulation activates mTORC1 in the OFC. To do so, OFC slices were briefly treated with NMDA (25 μM), and the activation of CaMKII and mTORC1 was tested by measuring the phosphorylation of CaMKII and of mTORC1 downstream targets, S6K and S6 (Figure 4—figure supplement 1). We found that ex vivo treatment of the OFC with NMDA increased the phosphorylation, and thus activation, of CaMKII, a marker of NMDAR activation (Figure 4c–d, Source data Figure 4c-d). In parallel the phosphorylation of S6K and S6 were also increased (Figure 4c–d, Source data Figure 4c-d). These data suggest that NMDAR stimulation activates mTORC1 in the OFC.

We previously showed that mTORC1 is activated in the OFC in response to a binge drinking session which persists for at least 24 hr into alcohol abstinence (Laguesse et al., 2017a). Thus, we next tested whether GluN2B is required for alcohol-dependent mTORC1 activation during alcohol deprivation. To do so, rats that underwent 7 weeks of IA-20%2BC paradigm or that drank water only for the same duration of time, received a bilateral intra-OFC infusion of the selective GluN2B antagonist Ro-25–6981 (Fischer et al., 1997) (10 μg/µl) or vehicle, 15 min prior to the end of the last alcohol drinking session. mTORC1 activation was examined at the end of the last 24 hr alcohol deprivation session by measuring the phosphorylation of mTORC1 downstream targets, 4EBP and S6 (Figure 4—figure supplement 1, Timeleine Figure 4—figure supplement 2, Table 2). In vehicle-treated rats, alcohol produced a robust activation of CaMKII (Figure 4e,h, Source data Figure 4e-j), and mTORC1, as measured by S6 (Figure 4f,I, Source data Figure 4e-j), and 4E-BP phosphorylation (Figure 4g,j, Source data Figure 4e-j). In contrast, R025-6981-treated rats exhibited a marked reduction in alcohol-dependent CaMKII, S6 and 4E-BP phosphorylation (Figure 4e–j, Source data Figure 4e-j). Together our data suggest that alcohol stimulates GluN2B which in turn promotes mTORC1 activation.

GluN2B in the OFC is required for alcohol seeking and habit

We reasoned that if GluN2B is recruited in response to alcohol exposure to activate mTORC1, then GluN2B should also participate in habitual alcohol seeking. First, we examined whether inhibition of GluN2B alters alcohol seeking during extinction. We found that intra-OFC administration of Ro25-6981 (10 μg/μl) 15 min prior to an extinction session (Figure 5—figure supplement 1) suppressed alcohol seeking, as indicated by a reduction in cumulative (Figure 5a, Source data Figure 5), active lever presses (Figure 5b, Source data Figure 5), as well as the rate of lever pressing (Figure 5c, Source data Figure 5). The reduction of alcohol seeking by Ro25-6981 was not due to alteration of locomotion as inter-response intervals were similar in vehicle-treated and Ro25-6981-treated animals (Figure 5—figure supplement 2, Source data Figure 5).

Figure 5 with 3 supplements see all
GluN2B in the OFC promotes alcohol seeking and habitual alcohol responding.

(a–c) Intra-OFC administration of Ro25-6981 decreases alcohol seeking. Rats underwent 7 weeks of IA-20%2BC, and were then trained to self-administer 20% alcohol. Vehicle (white) or Ro25-6981 (Ro25, 10 μg/μl, blue) was infused bilaterally in the OFC 15 min prior to a 30 min extinction session. (a) Cumulative lever presses. Two-way RM ANOVA revealed a significant main treatment effect (F1,12=10.89, p<0.05). Two-tailed paired t-test revealed that the number (b) and the rate (per 5 min) (c) of lever press were significantly different in the vehicle vs. Ro25-6981 groups (both t6 = 2.62, p<0.05). (d) Timeline of experiment. Rats underwent 7 weeks of IA20%−2BC, and were then trained to operant self-administer 20% alcohol using a progressive RI reinforcement schedule. Following stable RI-30 lever presses, rats underwent stereotaxic surgery to bilaterally implant guide cannulae in the OFC. One week later, RI-30 training was resumed for 2 weeks, after which rats received microinjections of vehicle (white) or Ro25-6981 (5 μg/μl, blue) in a counter-balanced manner 15 min prior to a 30 min home cage 20%2BC alcohol drinking session, and the number of lever presses were measured during extinction. (e) Intra-OFC administration of Ro25-6981 attenuates habitual alcohol seeking. Two way RM ANOVA showed a main effect of devaluation (F1,33=13.72, p<0.001) as well as a treatment x devaluation interaction (F1,33=6.31, p<0.05). Sidak’s multiple comparison test detected a significant difference for Ro25-6981 (p<0.001) on D compared to ND days. Individual data points and mean ± SEM are shown, (a–c) n = 7, (e) n = 12. *p<0.05, **p<0.01.

Finally, we tested whether GluN2B in the OFC contributes to habitual alcohol seeking. The timeline of the experiment is depicted in Figure 5d. Intra-OFC infusion of R025-6981 (5 μg/μl) did not impact home cage alcohol consumption during the devalued days (Figure 5—figure supplement 3, Table 1, Source data Figure 5). In contrast, administration of Ro25-6981 (5 μg/μl) into the OFC reverted habitual responding for alcohol to goal-directed responding (Figure 5e, Source data Figure 5). Together, these data indicate that the recruitment of GluN2B signaling by alcohol is required for alcohol seeking and habit.

Discussion

Using operant training protocols that bias alcohol seeking towards habitual or goal-directed behavior, we present data implicating mTORC1 in the OFC in mechanisms underlying the development and/or maintenance of habitual behavior associated with alcohol use. In addition, we show that GluN2B in the OFC is required for alcohol-dependent mTORC1 activation and function. Together, our data suggest that the GluN2B/mTORC1 axis acts as a molecular switch that converts the OFC from driving goal directed to habitual alcohol use.

We found that inhibition of mTORC1 in the OFC attenuates lever pressing during extinction, suggesting that mTORC1 regulates alcohol seeking. It is important to note, however, that mTORC1 inhibition does not appear to affect lever pressing in reinforced self-administration sessions. The OFC has been implicated in updating reward anticipation based on the value of an expected outcome (Ostlund and Balleine, 2007; Schoenbaum et al., 1998), and O'Doherty et al. showed that the lateral OFC (lOFC) is activated during reward anticipation in humans (O'Doherty et al., 2002). Thus, it is plausible that one of the roles of mTORC1 in the OFC is the strengthening of context-alcohol learning, consequently promoting alcohol seeking.

We found that blockade of mTORC1 activity in the OFC converts habitual alcohol responding to a more goal directed behavior. This finding could explain, at least in part, the decrease in alcohol seeking behavior following mTORC1 inhibition. Either extended operant training (Corbit et al., 2012) or specific schedules of reinforcement (Baltz et al., 2018; Gremel and Costa, 2013) have been utilized previously to bias animals toward habitual seeking for alcohol and other substances. By first cultivating habitual alcohol seeking and then testing the effects of rapamycin, we were able to demonstrate that mTORC1 inhibtion alters rat action selection strategies; either enhancing goal-directed responding or inhibiting habitual actions. Habitual behavior is thought to rely on the strengthening of the dorsolateral striatum (DLS) over the DMS, a brain region which participates in goal-directed behaviors (Corbit et al., 2012; Belin et al., 2013; Everitt and Robbins, 2005). Renteria and colleagues showed that alcohol vapor exposure generates habitual behavior via the disruption of OFC to DMS circuit (Renteria et al., 2018). Thus, it is plausible that mTORC1 activation is weakening OFC to DMS projections. Furthermore, Gremel et al. showed that the endocannabinoid/CB1 receptor system in OFC to dorsal striatum cicruitry promotes a shift from goal directed behavior to habitual responding (Gremel et al., 2016). Interestingly, cannabinoid signaling has been shown to activate mTORC1 in hippocampal neurons (Puighermanal et al., 2009). Furthermore, THC-dependent impairment of novel object recognition was reversed by the mTORC1 inhibitor, rapamycin (Puighermanal et al., 2009). Thus, it is possible that endocannabinoids and mTORC1 are part of the same signaling cascade that gates goal-directed behaviors, thereby promoting habit. In addition, Gourley and colleagues reported that the brain-derived neurotrophic factor (BDNF) in the OFC participates in goal-directed behaviors (Gourley et al., 2013; Gourley et al., 2016; Zimmermann et al., 2017). BDNF and mTORC1 play opposing roles in AUD (Ron and Barak, 2016). For example, BDNF keeps alcohol intake in moderation, and mTORC1 is associated with excessive alcohol use (Ron and Barak, 2016). As breakdown of the BDNF signaling cascade results in compulsive alcohol use (Warnault et al., 2016). It is tempting to speculate that the balance between BDNF and mTORC1 signaling in the OFC determines whether alcohol use is kept in moderation or becomes habitual.

We previously showed that mTORC1 activation is detected in the lOFC (Laguesse et al., 2017a), which was targeted in the current study. However, the structure of the OFC is complex. The medial OFC (mOFC) and lOFC are each part of anatomically and functionally distinct corticostriatal circuits in humans (Fettes et al., 2017), and in rodents (Izquierdo, 2017). The mOFC has been reported to play a larger role in determining, updating, and storing the relative reward values of stimuli and actions (Noonan et al., 2010). The lOFC, on the other hand, is proposed to be more involved in encoding the relationship between stimuli/actions and outcomes (Noonan et al., 2010), cue-guided behavioral flexibility (Panayi and Killcross, 2018), and goal-directed behavior (Gremel and Costa, 2013; Parkes et al., 2018). Thus a careful examination of mTORC1’s role in lOFC vs. mOFC-dependent behaviors is required.

The OFC has been associated with compulsive behavior (Ahmari et al., 2013), and recently Pascoli et al. reported that OFC neurons projecting to the dorsal striatum are associated with compulsive lever pressing in response to optogenetic activation of dopaminergic ventral tegmental area neurons (Pascoli et al., 2018). Compulsive drug use is associated with maladaptive habitual responding (Everitt and Robbins, 2005; Smith and Laiks, 2018). We and others provided animal data to support the notion that alcohol consumption can become compulsive (Seif et al., 2013; Warnault et al., 2016; Augier et al., 2018). Thus, it is possible that mTORC1 signaling in the OFC is driving both habitual and compulsive alcohol seeking.

Interestingly, mTORC1 inhibition in the OFC does not alter habitual responding for sucrose, indicating that mTORC1 plays a specific role in the formation of habits in response to alcohol exposure and not natural rewards. The difference between the consequences of mTORC1 inhibition on habitual alcohol vs. sucrose seeking is striking, but it is important to note that, unlike alcohol, voluntary sucrose intake does not activate mTORC1 in the OFC (Laguesse et al., 2017a).

Corbit et al. showed that RR-trained rats exhibit habitual alcohol self-administration after 8 weeks of training on an RR-3 schedule of reinforcement (Corbit et al., 2012). In the current study, rats undergoing RR training demonstrated persistent goal-directed alcohol seeking behaviors. The difference between the two studies is most likely to be due to the length of RR training (long in Corbit et al. (Corbit et al., 2012) vs. short herein). It is possible that further training would drive RR-trained goal-directed rats to exhibit more habitual phenotypes, and further research would be required to compare the models more accurately.

The NMDAR is a target of alcohol (Morisot and Ron, 2017), and alcohol enhances GluN2B phosphorylation in brain regions such as the cerebellum, the DMS, and the hippocampus (Morisot and Ron, 2017). GluN2B phosphorylation enhances the activity of the channel (Trepanier et al., 2012), and we show that alcohol increases GluN2B phosphorylation in the OFC. Thus, it is plausible that the recruitment of glutamatergic signaling by alcohol in the OFC enhances GluN2B function thereby enabling mTORC1 activation. Nimitvilai and colleagues showed that chronic exposure to alcohol vapor decreases in the expression of GluN2B (Nimitvilai et al., 2016), whereas we did not observe any changes in the protein levels of GluN2B following repeated cycles of alcohol binge drinking sessions and withdrawal. One important difference between our study and Nimitvilai et al. as well as additional reports discussed herein, relates to the use of contingent vs. non-contingent alcohol administration. Specifically, prior to the OSA, rats were subjected to 7 weeks of IA-20%2BC. The IA-20%2BC paradigm models ‘problem drinkers’ i.e. human subjects that suffer from AUD phenotypes such as binge drinking, compulsive drinking and craving (Enoch and Goldman, 2002). In our case, rats achieve blood alcohol concentrations (BAC) of 80 mg/dl (Carnicella et al., 2014) a BAC equivalent to binge drinking humans (NIAAA, 2004). In contrast, the majority of other reports described herein involved passive exposure of alcohol vapor, in which BAC is 150–250 mg/dl, and in which physical dependence on alcohol is observed (Griffin, 2014). Humans who are dependent on alcohol show profoundly degraded white matter and reduced neuronal density in the OFC (Pfefferbaum and Sullivan, 2005; Miguel-Hidalgo et al., 2006), possibly an indication of alcohol-induced damage to this structure. Thus, it is likely that cycles of binge drinking do not damage the OFC which may be the case in animal models of alcohol dependence. As only a small percentage of alcohol users exhibit physical dependence (WHO, 2014), we believe that our findings have important implications for the understanding of the mechanisms underlying problem drinking and AUD.

Our results suggest that the glutamatergic system is recruited during alcohol withdrawal leading to mTORC1 activation in the OFC. One major glutamatergic input in the OFC is the thalamus (Fresno et al., 2019). It would be therefore of interest to investigate whether thalamic nuclei inputs are recruited by alcohol to activate mTORC1 in the OFC.

We show here that GluN2B activation in the OFC is sufficient to influence the formation and/or maintenance of alcohol seeking and habit. In line with the possibility that GluN2B activation contributes to the development of habit, DePoy et al. showed that habitual behavior, developed upon subchronic cocaine treatment during adolescence (DePoy et al., 2017), is inhibited by the treatment of animals with the GluN2B inhibitor, ifenprodil (DePoy et al., 2017). DePoy et al. further showed that cocaine exposure during adolescence produces structural changes in the OFC which were attenuated by ifenprodil (DePoy et al., 2017). While our data suggest that GluN2B in the OFC participates in behavioral inflexibility, Brigman et al. reported that GluN2B in the OFC plays a role in choice learning (Brigman et al., 2013). It is plausible that alcohol-dependent neuroadaptations tilt the function of GluN2B from choice learning, for example a goal-directed behavior, to habitual responding.

What could be the mechanism by which mTORC1 contributes to alcohol seeking and habit? Activation of mTORC1 triggers the translation of a subset of mRNAs to protein in the cell body and in dendrites (Saxton and Sabatini, 2017; Yoon et al., 2016). We previously showed that long-term excessive alcohol intake and reinstatement of alcohol place preference initiate the translation of several synaptic proteins in the NAc (Ben Hamida et al., 2019; Liu et al., 2017; Laguesse et al., 2017b). We also found that reconsolidation of alcohol reward memories increases the immunoreactivity of mTORC1 targets in the OFC (Barak et al., 2013). Thus, further studies are required to identified transcripts which are translated in response to alcohol exposure and determine their role in alcohol seeking and habit.

Materials and methods

Reagents

Anti-phospho-S6 (S235/236, 1:500), anti-S6 (1:1000), anti-phospho-4E-BP (T37/46, 1:500), anti-4E-BP (1:1000), anti-phospho-CaMKII (T286, 1:1000), anti-phospho-S6K (T389, 1:500), anti-S6K (1:500), anti phospho-GluN2B (Y1472, 1:500) and anti GluN2B (1:1000) antibodies were purchased from Cell Signaling Technology (Danvers, MA). Anti-CaMKII (1:500) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-Actin (1:10,000) antibodies, phosphatase Inhibitor Cocktails 2 and 3, Dimethyl sulfoxide (DMSO), D-sucrose, tetrodotoxin (TTX) and NMDA were purchased from Sigma Aldrich (St. Louis, MO). Nitrocellulose membranes were purchased from EMD Millipore (Billerica, MA, USA). Enhanced Chemiluminescence (ECL) was purchased from GE Healthcare (Pittsburg, PA, USA). Donkey anti-rabbit horseradish peroxidase (HRP), and donkey anti-mouse horseradish peroxidase (HRP) were purchased from Jackson ImmunoResearch (West Grove, PA). EDTA-free complete mini Protease Inhibitor Cocktails were purchased from Roche (Indianapolis, IN). NuPAGE Bis-Tris precast gels and Phosphate buffered saline (PBS) were purchased from Life Technologies (Grand Island, NY). Bicinchoninic Acid (BCA) protein assay kit was obtained from Thermo Scientific (Rockford, IL), and ProSignal Blotting Film was purchased from Genesee Scientific (El Cajon, CA). Ethyl alcohol (190 proof) was purchased from VWR (Radnor, PA). Rapamycin was purchased from LC laboratories (Woburn, MA) and Ro25-6981 was purchased from Tocris Bioscience (Bristol, UK).

Subjects

Male Long Evans rats (Harlan, Indianapolis, IN) were one (ex vivo experiment) or two months (in vivo experiments) old at their arrival. Animals were single-housed in a temperature- and humidity-controlled colony room (22 ± 2°C, relative humidity: 50–60%) under a normal 12 hr light/dark cycle (lights on at 7:00AM) with food and water available ad libitum. Rats were given a week of habituation to the housing conditions before the beginning of the experiments. All animal procedures were approved by the University of California San Francisco Institutional Animal Care and Use Committee (IACUC) (protocol number number AN179720-01B), and were conducted in agreement with the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC).

Preparation of solutions

Request a detailed protocol

Alcohol was diluted to 20% (v/v) and sucrose to 1–8% (w/v) in tap water. Rapamycin was dissolved in PBS. Ro25-6981 and NMDA were dissolved in 0.1% DMSO in PBS.

Intermittent access to 20% alcohol using two bottle choice and drug infusion

Request a detailed protocol

Rats underwent intermittent access to 20% alcohol in a 2-bottle choice (IA-20%2BC) paradigm for 7 weeks as described in Laguesse et al. (2017b). Specifically, rats were given 24 hr of concurrent access to one bottle of 20% alcohol (v/v) in tap water and one bottle of water. Control rats had access to water only. Drinking sessions started at 12:00pm on Monday, Wednesday and Friday, with 24- or 48 hr (weekend) of alcohol-deprivation periods in which rats consumed only water. The placement (left or right) of the water or alcohol solution was alternated between each session to control for side preference. Water and alcohol bottles were weighed at the beginning and at the end of each alcohol drinking session. Rats were weighed once a week. Rats drinking more than 3.5 g/kg/24 hr on the last week of IA-20%2BC were selected for the study.

Stereotaxic surgery

Request a detailed protocol

Following a stable baseline of alcohol drinking, rats underwent stereotaxic surgery. Specifically, rats were continuously anesthetized using isoflurane (Baxter) and bilateral guide cannulae (Plastic One) were implanted in the ventrolateral OFC (anterioposterior +3.5 mm, mediolateral ±2 mm and dorsoventral +3.9 mm; all coordinates are from bregma), secured with screws (Plastic One) and dental cement (Ortho-Jet, Lang Dental).

Alcohol seeking and drug infusion

Request a detailed protocol

First, rats underwent IA-20%2BC paradigm for 7 weeks, as described above. Rats drinking more than 3.5 g/kg/24 hr on the last week of IA-20%2BC were selected for alcohol self-administration training in an operant self-administration (OSA) paradigm (Jeanblanc et al., 2009) with modification adapted from Gremel et al. (2016); Corbit et al. (2012); Renteria et al. (2018). Training was conducted during the light cycle in operant chambers (Med-Associates; Georgia, VT) using a fixed ratio 3 (FR3) (i.e. three active lever presses result in the delivery of one alcohol reward) during daily a daily 30 min session (Mon-Fri) as previously described (Jeanblanc et al., 2009). An inactive lever was also extended during the operant sessions but had no programmed consequence. Following stable baseline of alcohol responding, rats underwent stereotaxic surgery for cannula placement in the OFC, as described above. After a one-week recovery period, operant self-administration procedure was resumed and handling for drug microinfusions began. Rats received infusion of either vehicle or rapamycin (50 ng/μl/side) 3 hr or R025-6981 (10 μg/μl/side) 15 min before a single 30 min extinction session. Drugs were injected with bilateral infusion needles (Plastics One) that projected 0.5 mm past the end of the guide cannula, at a rate of 0.5 μl/minute. Drug testing was performed using a ‘within-subject’ design, in which rats received both treatments in counterbalanced order.

Upon establishing a baseline of IA-20%2BC, a separate cohort of animals received an infusion of R025-6981 (10 μg/μl/side) or vehicle 15 min prior to the end of the last withdrawal session. The OFC was harvested at the end of the 24 hr withdrawal session and were processed for western blot analysis.

Habitual and goal-directed alcohol seeking and drug infusion

Request a detailed protocol

After 7 weeks of IA-20%2BC, high-drinking rats (alcohol intake >3.5 g/kg/24 hr) were trained to self-administer 20% alcohol in an OSA paradigm as described above. Self-administration training was initiated under fixed ration 1 (FR1) for three consecutive overnight sessions. Afterwards, the sessions lasted for an hour and took place during the light cycle. Only one lever was available during the operant sessions. Rats were pseudo-randomly assigned to two groups and were subjected to a random interval (RI), or a random ratio (RR) schedule of reinforcement which biases responding towards habitual or goal directed actions, respectively (Gremel and Costa, 2013). Only rats that self-administered alcohol more than 0.3 g/kg/hour were included in the study.

RI training: During RI training, each reward was delivered following one lever press occurring after variable reward delivery time intervals (Gremel and Costa, 2013). Three sessions were under RI-5 (i.e. inter-reward time was 5 s on average, ranging from 1 to 10 s), followed by three RI-15 sessions (i.e. inter-reward time was 15 s on average ranging from 10 to 30 s). Reinforcement was then shifted to RI-30 (i.e. inter-reward time was 30 s on average ranging from 10 to 50 s) for at least 20 sessions (Timeline, Figure 2a).

RR training: During RR training, reward was delivered following a variable number of lever responses (Gremel and Costa, 2013). Six sessions under RR-2 (one reward delivery following on average two lever presses with number of presses ranging from 1 to 3) followed by RR-3 (one reward delivery following on average three lever presses with number of presses ranging from 1 to 5) for at least 20 sessions (Timeline, Figure 2a).

Following stable RI-30 or RR-3 responding, rats underwent alcohol devaluation testing as described in Corbit et al. (2012) with modifications. Rats were given home-cage 2BC access to 1% sucrose on non-devalued (ND) days or 20% alcohol on devalued (D) days for 30 min (Figure 2b). On each day, immediately after home-cage pre-feeding, rats underwent a 10 min extinction test during which unrewarded lever presses were recorded. A ‘within-subject’ design was used in which ND or D sessions were in a counterbalanced order, with two standard retraining sessions between the tests. To minimize the effects of individual variance in total lever pressing, we normalized ND and D lever presses (Baltz et al., 2018; Gremel and Costa, 2013; Gremel et al., 2016). Normalized lever pressing was calculated as ND or D lever presses/total number of lever presses, respectively. Normalizing devaluation lever presses produces a distribution of data, where 50% is equivalent to equal lever pressing between ND and D (Gremel and Costa, 2013). At the end of the training period and upon acquiring the behavior (RR, goal directed behavior and RI, habitual behavior), rats underwent stereotaxic surgery to bilaterally implant a guide cannula in the ventrolateral OFC, as described above. One week later, RI-30 or RR-3 training was resumed for 2 weeks after which rats received microinjections of either vehicle or rapamycin (50 ng/μl/side) 3 hr or R025-6981 (5 μg/μl/side) 15 min prior an alcohol devaluation testing.

Habitual sucrose seeking

Request a detailed protocol

Habitual sucrose training followed an approach similar to that described above for habitual alcohol seeking. After 7 weeks of IA-20%2BC, high-drinking rats (alcohol intake >3.5 g/kg/24 hr) were trained to self-administer 8% sucrose in an OSA paradigm. RI-30 training was similar to the procedure described for alcohol except that 8% sucrose was initially used and progressively reduced to 1% across training sessions. Only ten RI30 sessions (2 weeks) were conducted prior to the initial outcome devaluation test. Rats were then implanted with a guide cannula in the OFC as described above. After a week recovery period, sucrose self-administration was resumed for two weeks, after which rats received microinjections of vehicle or drug in a counter-balanced manner prior to the devaluation test. Sensitivity to changes in sucrose value was tested using a sucrose devaluation procedure which was similar to alcohol devaluation, except that rats had 2BC access to 20% alcohol on ND day and 1% sucrose on D day (Timeline, Figure 3a–b).

Histology

Request a detailed protocol

For verification of cannula placement, brains were post-fixed in 4% PFA for one week, then rapidly frozen and sectioned into 50 μm coronal slices using a Leica CM3050 cryostat (Leica Biosystems). Slices were stained using cresyl violet and examined for cannula placement using a bright-field microscope. Rats with correct guide cannula placement were included in the study.

Ex vivo NMDA treatment

Request a detailed protocol

Naïve rats were euthanized by isoflurane and decapitated. Brains were removed and sectioned into 300 μm-thick slices at 4°C in a solution of aCSF containing (in mM) 200 sucrose, 1.9 KCl, 1.4 NaH2PO4, 0.5 CaCl2, 6 MgCl2, 10 glucose, 25 NaHCO3, four ascorbic acid; 310–330 mOsm. Immediately after sectioning, OFC-containing slices were recovered in aCSF (in mM: 125 NaCl, 2.5 KCl, 1.4 NaH2PO4, 25 NaHCO3, 2 CaCl2, 1.3 MgCl2, 25 glucose, 0.4 ascorbic acid) heated at 32°C for 30 min and then moved to an aCSF solution containing 1 μM TTX at room temperature for 30 min (Sutton and Chandler, 2002). The OFC was dissected and incubated for 30 min in aCSF containing 1 μM TTX at room temperature. OFC sections were then treated for 3 min with vehicle (0.1% DMSO) or NMDA (25 μM) in TTX-containing aCSF , and western blot analysis was conducted as described below.

Collection of OFC for western blot

Request a detailed protocol

Animals were euthanized and brains rapidly dissected on ice using a 1 mm brain block. One mm thick coronal sections located between +4.20 mm and +3.20 mm anterior to bregma were removed and the OFC was dissected on an anodized aluminum block on ice, using the Paxinos and Watson stereotaxic atlas 4th edition as a reference. Tissue was collected into 1.5 ml Eppendorf tubes and immediately homogenized in 300 μl RadioImmuno Precipitation Assay (RIPA) buffer (in mM: 50 Tris-HCl, pH 7.6, 150 NaCl, 2 EDTA, and 1% NP-40, 0.1% SDS and 0.5% sodium deoxycholate, protease and phosphatase inhibitor cocktails).

Western blot analysis

Request a detailed protocol

Tissue were homogenized in ice-cold RIPA buffer , using a sonic dismembrator. Protein content was determined using BCA kit. Tissue homogenates (30 μg per sample) were resolved on NuPAGE 10% Bis-Tris gels at 100 V for 2 hr and transferred onto nitrocellulose membranes at 30V for 2 hr. Blots were blocked with 5% milk-phosphate-buffered saline with 0.1% tween-20 at room temperature and then probed with primary antibodies overnight at 4°C. Membranes were washed and probed with HRP-conjugated secondary antibodies for 2 hr at room temperature, and bands were visualized using ECL. Band intensities were quantified by ImageJ (National Institutes of Health, MD, USA).

Data analysis

Request a detailed protocol

All data are expressed as the mean ± SEM. GraphPad Prism 7.0 (GraphPad Software, Inc, La Jolla, CA, USA) was used to plot and analyze the data. D’Agostino–Pearson normality test and F-test/Levene test were used to verify the normal distribution of variables and the homogeneity of variance, respectively. The sample sizes were based on previous publications (Neasta et al., 2010; Barak et al., 2013). Data between two groups were compared using two-tailed paired or unpaired t-test. Data from multiple groups against one group were compared using a two-way repeated-measures analysis of variance (RM-ANOVA) followed when appropriate by a Tukey multiple comparisons test or a Sidak’s multiple comparisons test. Statistical significance was set at p<0.05.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
    Problem drinking and alcoholism: diagnosis and treatment
    1. MA Enoch
    2. D Goldman
    (2002)
    American Family Physician 65:441–448.
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
    Ro 25-6981, a highly potent and selective blocker of N-methyl-D-aspartate receptors containing the NR2B subunit. Characterization in vitro
    1. G Fischer
    2. V Mutel
    3. G Trube
    4. P Malherbe
    5. JN Kew
    6. E Mohacsi
    7. MP Heitz
    8. JA Kemp
    (1997)
    The Journal of Pharmacology and Experimental Therapeutics 283:1285–1292.
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
    NIAAA council approves definition of binge drinking
    1. NIAAA
    (2004)
    NIAAANewsletter, Winter 2004.
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
    The Mouse Brain in Stereotaxic Coordinates (2)
    1. G Paxinos
    2. KBJ Franklin
    (2001)
    San Diego: Academic Press.
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
    Mechanisms of translation control underlying long-lasting synaptic plasticity and the consolidation of long-term memory
    1. E Santini
    2. TN Huynh
    3. E Klann
    (2014)
    Progress in Molecular Biology and Translational Science 122:131–167.
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
    Global status report on alcohol and health
    1. WHO
    (2014)
    World Health Organization.
  81. 81
  82. 82

Decision letter

  1. Mary Kay Lobo
    Reviewing Editor; University of Maryland, United States
  2. Kate M Wassum
    Senior Editor; University of California, Los Angeles, United States
  3. Michelle Mazei-Robison
    Reviewer; Michigan State University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This study defines a novel role for mechanistic target rapamycin complex 1 (mTORC1) action, potentially through GluN2B activation, in the orbitofrontal cortex (OFC) in alcohol seeking and habitual intake. The study demonstrates that inhibition of mTORC1 activity attenuates alcohol seeking and restores sensitivity to outcome devaluation in rats that habitually seek alcohol. This effect was specific to alcohol as mTorc1 inhibition did not alter habitual responding to a natural reward. Overall, the study provides new insight into a mechanism, via GluN2B and mTORC1, in mediating a shift from goal-directed to habitual alcohol seeking.

Decision letter after peer review:

Thank you for submitting your article "mTORC1 in the orbitofrontal cortex promotes habitual alcohol seeking" for consideration by eLife. Your article has been reviewed by Kate Wassum as the Senior Editor, a Reviewing Editor, and three reviewers. The following individuals involved in review of your submission have agreed to reveal their identity: Michelle Mazei-Robison (Reviewer #3).

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

Essential revisions:

While the reviewers find the study to be of interest, they raise concerns that must be adequately addressed before the paper can be accepted. The reviewers agree that these revisions should not require additional experiments.

1) Please provide a compelling rationale for the exclusion criteria and provide details about numbers excluded further clarity on the exclusion criterion (did not acquire the behavior is vague) and why the exclusions do not create bias. This is in reference to the following statement: "did not acquire the behavior and were pressing less than 5 lever presses per session on ND day" (subsection “Habitual and goal-directed alcohol seeking and drug infusion”).

2) Please address fundamental issues with the below pattern of results, which appear to be empirically and conceptually discordant. This will aid the reader in understanding these discrepancies and clarify the overall significance of the current findings.

a) Please address how a manipulation disrupts alcohol seeking under normal conditions (without any satiety treatment, Figure 1D,E,F and Figure 4A,B,C) but not during ND tests (when rats are sated on irrelevant outcome, Figure 2E,G and Figure 5E). Could important procedural differences help to explain these differences?

b) Please address how disrupting habitual control can lead to a disruption of alcohol seeking, if the current and published (e.g., Yin et al., 2004; Corbit et al., 2014) data indicates that this does not lead to a response decrement, apparently due to compensation by the goal-directed system. While not all findings must fit tightly these issues are fundamental to judging the significance of the findings and their value to the field.

3) Please ensure that similar and appropriate details about instrumental training procedures are provided for all experiments, including differences; e.g., subsection “Alcohol seeking and drug infusion”. For instance, one difference is that rats are sated on sucrose prior to ND tests, while another difference is test duration.

4) Please provide a rationale for why phosphorylated GluN2B was not measured? The data in Figure 4C-J nicely links GluN2B function to alcohol-induced mTORC1 and the authors address alcohol-induced changes in GluN2B phosphorylation in other brain regions in the discussion. However, the activation of GluN2B is not specifically addressed in the study.

5) In Figure 2E, post-hoc comparisons are not necessary because a main effect between two conditions was found (not an interaction). Thus, the main effect can be indicated in the figure to emphasize that the groups don't differ in their preference for the valued outcome, but the direct comparisons should be eliminated.

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

Thank you for resubmitting your work entitled "mTORC1 in the orbitofrontal cortex promotes habitual alcohol seeking" 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 before acceptance, as outlined below:

1) The reviewers felt that the discussion was unfocused and speculative. While they do understand that this may reflect the goal to address the previous concerns, a more focused discussion would improve the manuscript. Below are the concerns that the reviewers felt should be addressed:

a) Discussion section, "reward anticipation" is invoked. This is problematic for the habit argument because habits are by nature uncoupled from reward value and an anticipation of the outcomes of one's behaviors. Thus, reward anticipation would not be a factor driving habitual behavior. If the authors want to make this argument, this should be clearly dissociated from the habit argument. Alternatively, rather than referring to action-outcome based reward anticipation, the authors may be referring to context-alcohol learning, which is likely a major source of motivation for habitual alcohol seeking. It is known that natural reward-paired cues preferentially motivate habitual reward-seeking actions (Holland, 2004; Wilgen et al., 2012) in a manner that does not depend on reward value (Rescorla, 1994; Holland, 2004).

b) Discussion section, the authors argue that mTORC1 could be strengthening OFC projections to the DLS. These projections do not exist – the OFC projects to large swaths of the striatum, but not the habit-associated DLS, referring to the dorsal-most region of the lateral striatum.

c) Discussion section remarks, "it would be of interest to determine the normal role of mTORC1, and whether mTORC1 suppresses other decision-making tasks." This line was confusing given the author's investigations with sucrose reinforcers assessed this prospect.

d) Discussion section concludes with the phrase "with less involvement coming from the lOFC." The cited references (Bradfield, Hart and Balleine, 2018; Gourley et al., 2016) do not investigate the lOFC, and thus, do not provide evidence for this notion. This section as a whole appears unfocused. It is suggested that this section be significantly edited down to simply make the point that the lOFC and mOFC have different functions in the context of goal-directed behavior. Currently, the arguments regarding the balance between the two are difficult to follow and very speculative, given the sole focus on lOFC here.

e) The authors' responses to comments 2 and 3 touch on important issues that help address confusion in understanding some apparent discrepancies in the data. However, the pieces are not entirely put together. The authors suggest that rapamycin and Ro25 treatments disrupt alcohol seeking late in nonreinforced tests, which explains why no deficits are found in short ND tests (also nonreinforced). However, the significance of this is not addressed. The reviewers suggest explicitly raising the possibility that habit processes may preferentially contribute to persistence of alcohol seeking in extinction. While persistence in extinction is not a critical test of habit behavior, it is certainly consistent with this account. Providing such a discussion would help link these findings to the devaluation results, which would benefit many readers.

2) There were concerns with the exclusion of subjects based on arbitrarily low performance of the nondevalued action (only) at test because this, by design, creates a bias, increasing the likelihood of detecting a difference between devalued vs. nondevalued actions. The rationale was that low nondevalued action performance at test meant that animals had "generally not escalated their lever-pressing during training". Based on this it is suggested that the authors exclude based on low response rate during training, which doesn't create a systematic bias and is in line with accepted practice. Alternatively, it suggested that the authors include the two excluded subjects.

https://doi.org/10.7554/eLife.51333.sa1

Author response

Essential revisions:

While the reviewers find the study to be of interest, they raise concerns that must be adequately addressed before the paper can be accepted. The reviewers agree that these revisions should not require additional experiments.

1) Please provide a compelling rationale for the exclusion criteria and provide details about numbers excluded further clarity on the exclusion criterion (did not acquire the behavior is vague) and why the exclusions do not create bias. This is in reference to the following statement: "did not acquire the behavior and were pressing less than 5 lever presses per session on ND day" (subsection “Habitual and goal-directed alcohol seeking and drug infusion”).

This is an important point, which we have addressed by indicating the number of excluded animals for each experiment in the figure legends and clarifying the reasoning for the exclusion criteria in the Materials and methods section. In essence, this study is focused on alcohol seeking behaviors, and we have attempted to exclude animals from the study who do not reliably seek alcohol. We set the cutoff at 5 lever presses, because animals who press at this rate have generally not escalated their lever-pressing during training, which could mean that they have not properly learned the task or may not find alcohol sufficiently rewarding to elicit seeking behavior. Including these animals can also skew the normalized lever pressing data in either direction due to very slight changes in pressing, much smaller than the standard deviation. Only two rats were excluded from the study.

2) Please address fundamental issues with the below pattern of results, which appear to be empirically and conceptually discordant. This will aid the reader in understanding these discrepancies and clarify the overall significance of the current findings.

a) Please address how a manipulation disrupts alcohol seeking under normal conditions (without any satiety treatment, Figure 1D,E,F and Figure 4A,B,C) but not during ND tests (when rats are sated on irrelevant outcome, Figure 2E,G and Figure 5E). Could important procedural differences help to explain these differences?

We thank the reviewers for raising this question. It is important to note that extinction does disrupt alcohol seeking in general, with or without satiety treatment. One difference between the extinction tests is that tests depicted in Figure 1 and Figure 5A last 30 minutes, while the extinction tests in Figure 2 and Figure 5E (devaluation tests) last only 10 minutes. In order to minimize any general effects of satiety treatment, none of the rats in any of the experiments were food or water restricted at any time. Although it is possible that the home cage satiety treatment on an irrelevant outcome (ND tests) could impact seeking behavior, we did not compare alcohol seeking in these animals with an additional group that underwent no satiety treatment, as this experiment is beyond the scope of the questions addressed in this manuscript. We believe that even if satiety treatment on an irrelevant outcome does impact seeking in either direction, this would not change the main conclusions drawn from the data presented here.

b) Please address how disrupting habitual control can lead to a disruption of alcohol seeking, if the current and published (e.g., Yin et al., 2004; Corbit et al., 2014) data indicates that this does not lead to a response decrement, apparently due to compensation by the goal-directed system. While not all findings must fit tightly these issues are fundamental to judging the significance of the findings and their value to the field.

This comment raises a salient point on the impact of shifting action control from a more habitual to goal-directed strategy in general. As far as translational goals are concerned, we are not preoccupied with necessarily reducing alcohol consumption in all cases, but rather with decreasing habitual, excessive, and compulsive drinking. While the overall treatment goal with other drugs of abuse may be complete abstinence, alcohol is unique in that moderate alcohol consumption under acceptable circumstances meets societal standards in many regions of the world. For this reason, more research is needed into the respective roles of habitual and goal-directed alcohol seeking in human cohorts. It is likely that there is at least some compensation from the goal-directed system when you disrupt habitual control. Indeed, we show in Figure 1 that inhibition of mTOR in the OFC does not alter alcohol consumption or self-administration except in extinction sessions, which would indicate that these animals still drink at the same level, but their action selection strategy has shifted.

3) Please ensure that similar and appropriate details about instrumental training procedures are provided for all experiments, including differences; e.g., subsection “Alcohol seeking and drug infusion”. For instance, one difference is that rats are sated on sucrose prior to ND tests, while another difference is test duration.

We have updated the Materials and methods section of the paper to minimize confusion between the different devaluation protocols for sucrose and alcohol.

4) Please provide a rationale for why phosphorylated GluN2B was not measured? The data in Figure 4C-J nicely links GluN2B function to alcohol-induced mTORC1 and the authors address alcohol-induced changes in GluN2B phosphorylation in other brain regions in the discussion. However, the activation of GluN2B is not specifically addressed in the study.

We thank the reviewer for the suggestion. Not measuring GluN2B phosphorylation was an oversight on our behalf especially since it is a true indication of GluN2B activation. We added new data to the revision showing that alcohol significantly increases GluN2B phosphorylation with no accompanying change in total GluN2B protein level. These results are now presented in Figure 4.

5) In Figure 2E, post-hoc comparisons are not necessary because a main effect between two conditions was found (not an interaction). Thus, the main effect can be indicated in the figure to emphasize that the groups don't differ in their preference for the valued outcome, but the direct comparisons should be eliminated.

Per the reviewer’s request, we revised the statistics and figure legend for Figure 2E (now Figure 2D).

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

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

1) The reviewers felt that the discussion was unfocused and speculative. While they do understand that this may reflect the goal to address the previous concerns, a more focused discussion would improve the manuscript. Below are the concerns that the reviewers felt should be addressed:

a) Discussion section, "reward anticipation" is invoked. This is problematic for the habit argument because habits are by nature uncoupled from reward value and an anticipation of the outcomes of one's behaviors. Thus, reward anticipation would not be a factor driving habitual behavior. If the authors want to make this argument, this should be clearly dissociated from the habit argument. Alternatively, rather than referring to action-outcome based reward anticipation, the authors may be referring to context-alcohol learning, which is likely a major source of motivation for habitual alcohol seeking. It is known that natural reward-paired cues preferentially motivate habitual reward-seeking actions (Holland, 2004; Wilgen et al., 2012) in a manner that does not depend on reward value (Rescorla, 1994; Holland, 2004).

We appreciate the reviewers’ point and the validity of their argument. We modified the term “action-outcome based reward anticipation” to “context-alcohol learning”.

b) Discussion section, the authors argue that mTORC1 could be strengthening OFC projections to the DLS. These projections do not exist – the OFC projects to large swaths of the striatum, but not the habit-associated DLS, referring to the dorsal-most region of the lateral striatum.

We have removed this part of the Discussion section.

c) Discussion section remarks, "it would be of interest to determine the normal role of mTORC1, and whether mTORC1 suppresses other decision-making tasks." This line was confusing given the author's investigations with sucrose reinforcers assessed this prospect.

We apologize for the confusion. By normal we are referring to known OFC-dependent behaviors such as decision making (Wallis, 2007; Meyer and Bucci, 2016; Baltz et al., 2018), stimulus-outcome association (Ostlund and Balleine, 2007; Gourley et al., 2013; Gremel and Costa, 2013) and updating the value of predicted outcomes (Fiuzat et al., 2017; Baltz et al., 2018). The word “normal” has been deleted from the text.

d) Discussion section concludes with the phrase "with less involvement coming from the lOFC." The cited references (Bradfield, Hart and Balleine, 2018; Gourley et al., 2016) do not investigate the lOFC, and thus, do not provide evidence for this notion. This section as a whole appears unfocused. It is suggested that this section be significantly edited down to simply make the point that the lOFC and mOFC have different functions in the context of goal-directed behavior. Currently, the arguments regarding the balance between the two are difficult to follow and very speculative, given the sole focus on lOFC here.

This part of the Discussion section was edited according to the reviewers’ suggestion.

e) The authors' responses to comments 2 and 3 touch on important issues that help address confusion in understanding some apparent discrepancies in the data. However, the pieces are not entirely put together. The authors suggest that rapamycin and Ro25 treatments disrupt alcohol seeking late in nonreinforced tests, which explains why no deficits are found in short ND tests (also nonreinforced). However, the significance of this is not addressed. The reviewers suggest explicitly raising the possibility that habit processes may preferentially contribute to persistence of alcohol seeking in extinction. While persistence in extinction is not a critical test of habit behavior, it is certainly consistent with this account. Providing such a discussion would help link these findings to the devaluation results, which would benefit many readers.

This comment raises an important point. While the reviewers are correct in noting that persistent alcohol seeking during extinction may be consistent with habitual behavior, we do not have sufficient data to support this claim, as the animals in Figure 1 and Figure 5A-C did not undergo RI training or even extended FR training, which would condition untreated rats toward more habitual seeking. The differences in seeking behavior between rapamycin- or Ro25-treated animals in the 30-minute nonreinforced tests are likely to be due to, at least in part, to changes in action-selection strategy, but speculation on enhancing already goal-directed behavior might generate uncertainty in readers. We addressed this concern in the Discussion section by highlighting the possibility that a shift in action-selection strategy may already be present in the 30-minute nonreinforced sessions, but we need to use RI training and devaluation to confirm this.

2) There were concerns with the exclusion of subjects based on arbitrarily low performance of the nondevalued action (only) at test because this, by design, creates a bias, increasing the likelihood of detecting a difference between devalued vs. nondevalued actions. The rationale was that low nondevalued action performance at test meant that animals had "generally not escalated their lever-pressing during training". Based on this it is suggested that the authors exclude based on low response rate during training, which doesn't create a systematic bias and is in line with accepted practice. Alternatively, it suggested that the authors include the two excluded subjects.

Per the reviewers’ request, we now include in the analysis the two rats which are pressing below threshold. Inclusion of the subjects did not change the interpretations and/or conclusions, the text and figures have been updated accordingly.

https://doi.org/10.7554/eLife.51333.sa2

Article and author information

Author details

  1. Nadege Morisot

    Department of Neurology, University of California, San Francisco, San Francisco, United States
    Present address
    Charles River Laboratories, South San Francisco, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
  2. Khanhky Phamluong

    Department of Neurology, University of California, San Francisco, San Francisco, United States
    Contribution
    Data curation, Formal analysis, Investigation
    Competing interests
    No competing interests declared
  3. Yann Ehinger

    Department of Neurology, University of California, San Francisco, San Francisco, United States
    Contribution
    Data curation, Formal analysis, Validation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9314-6575
  4. Anthony L Berger

    Department of Neurology, University of California, San Francisco, San Francisco, United States
    Contribution
    Data curation, Investigation
    Competing interests
    No competing interests declared
  5. Jeffrey J Moffat

    Department of Neurology, University of California, San Francisco, San Francisco, United States
    Contribution
    Formal analysis
    Competing interests
    No competing interests declared
  6. Dorit Ron

    Department of Neurology, University of California, San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Project administration
    For correspondence
    dorit.ron@ucsf.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5161-967X

Funding

National Institute on Alcohol Abuse and Alcoholism (P50 AA017072)

  • Dorit Ron

National Institute on Alcohol Abuse and Alcoholism (R01 AA027474)

  • Dorit Ron

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

Acknowledgements

This research was supported by the National Institute of Alcohol Abuse and Alcoholism, P50 AA017072 (DR) and R01 AA027474 (DR).

Ethics

Animal experimentation: All animal procedures were approved by the University of California San Francisco Institutional Animal Care and Use Committee (IACUC) and were conducted in agreement with the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC).(protocol number number AN179720-01B).

Senior Editor

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

Reviewing Editor

  1. Mary Kay Lobo, University of Maryland, United States

Reviewer

  1. Michelle Mazei-Robison, Michigan State University, United States

Publication history

  1. Received: August 25, 2019
  2. Accepted: December 9, 2019
  3. Accepted Manuscript published: December 10, 2019 (version 1)
  4. Accepted Manuscript updated: December 11, 2019 (version 2)
  5. Version of Record published: January 14, 2020 (version 3)

Copyright

© 2019, Morisot 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

  • 989
    Page views
  • 158
    Downloads
  • 1
    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)

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

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