Survival depends on learning to associate environmental cues with food or other natural rewards. Individual differences in learning and motivational processes support the acquisition, expression, and updating of cue-reward associations. Recent evidence suggests that distinct learning strategies are predictive of dysregulated motivation for drug-associated cues/conditioned stimuli (CS) (Chang et al., 2022; Martin et al., 2022; Pitchers et al., 2017; Saunders et al., 2013). Midbrain and striatal dopamine signals are broadly implicated in a diverse array of learning and motivational processes, including CS-reward associations underscoring the importance of dopamine (DA) in adaptive behavior that promotes survival (Langdon et al., 2018; Lee et al., 2022; Nasser et al., 2017). Yet considerably less is known about the role of DA signals in areas outside of the striatum during adaptive and maladaptive cue-reward learning. Recent advances in fluorescent sensor technology and fiber photometry permit the measurement of DA binding correlates (Labouesse et al., 2020). These new techniques can reveal understudied functions of DA signaling in non-striatal DA terminal regions like the dBNST, an extended amygdala nucleus, that is critical for dysregulated CS-triggered opioid relapse (Gyawali et al., 2020). Here, we characterize basic dBNST DA correlates by recording fluorescent GRAB_DA signals during a Pavlovian task that distinguishes two distinct relapse vulnerability phenotypes.

Recent studies identify unique learning strategies that predict heightened CS-triggered relapse vulnerability (Chang et al., 2022; Martin et al., 2022; Pitchers et al., 2017; Saunders et al., 2013). In particular, a simple Pavlovian Lever Autoshaping task distinguishes two extreme tracking phenotypes: (1) sign-tracking (ST) rats that approach and vigorously engage with the reward predictive lever cue, even though cue interaction is not necessary to obtain food reward and (2) goal-tracking rats that interact with the food cup during cue presentation where food reward is delivered after lever retraction (Boakes, 1977; Flagel et al., 2007; Hearst and Jenkins, 1974; Meyer et al., 2012). Sign-tracking rats show heightened CS-triggered drug relapse vulnerability compared to goal-trackers. A third group called intermediates approach both the food cup and lever at similar levels, and their relapse vulnerability is like that of goal-tracking rats (Saunders and Robinson, 2010). Fast scan cyclic voltammetry recording of real-time dopamine indicated that sign-, but not goal-tracking, evokes increases in phasic fluctuations in DA in the nucleus accumbens (NAc) during CS presentation (Flagel et al., 2011). NAc DA is necessary for both the expression of sign-tracking and for sign-trackers’ heightened CS-triggered drug relapse, but not for goal-trackers or their relapse behavior (Saunders et al., 2013). Given the critical role of dBNST in CS-triggered relapse, we aimed to determine whether there are similar individual differences in dBNST DA signaling in sign- and goal-tracking rats using the Pavlovian Lever Autoshaping task (Buffalari and See, 2009; Gyawali et al., 2020; Silberman and Winder, 2013b).

Midbrain dopamine neuron activity strengthens cue-outcome associations by serving as a bidirectional prediction error signal where unexpected reward delivery increases and omitted reward decreases dopamine neuron firing relative to expected reward (Montague et al., 1996; Schultz, 2015; Schultz et al., 1997). Over the course of learning, the phasic dopamine activity transfers from the unconditioned stimulus (US) to the CS (Montague et al., 1996; Schultz, 2015; Schultz et al., 1997). In the NAc, the transfer of dopamine signals from the US to the CS occurs more robustly in ST compared to GT rats (Flagel et al., 2011; Lee et al., 2018; Saddoris et al., 2016), and NAc DA antagonism reduces sign-tracking but not goal-tracking behaviors (Saunders and Robinson, 2012). Together, these studies support the Pavlovian lever autoshaping task (and sign-tracking) as a reliable framework for studying dopamine’s role in regions of the brain critically involved in cue-motivated natural and drug reward-seeking behaviors.

The dBNST receives dense dopaminergic input from several midbrain regions including the ventral tegmental area, ventral periaqueductal gray, and to a much lesser extent, the substantia nigra (Hasue and Shammah-Lagnado, 2002; Meloni et al., 2006). dBNST dopamine is associated with a variety of reward-motivated behaviors. dBNST dopamine release is increased during intra-oral sucrose infusion and in response to cues that predict intracranial self-stimulation of the medial forebrain bundle (Lin et al., 2020; Park et al., 2012; Park et al., 2013). Furthermore, dopamine antagonist injections in the dBNST reduce responding to sucrose in a binge eating paradigm (Maracle et al., 2019). All major drugs of abuse, including opioids, increase extracellular dopamine in the BNST and dBNST dopamine antagonism reduces cocaine self-administration and ethanol seeking (Carboni et al., 2000; Eiler et al., 2003; Epping-Jordan et al., 1998). Despite these studies implicating dBNST dopamine in motivated behaviors, a comprehensive characterization of endogenous dBNST dopamine dynamics in cue-induced behaviors is lacking. To address this, we used a dopamine sensor GRAB_DA in combination with fiber photometry to examine the basic properties of the dBNST dopamine signals; their role during lever autoshaping, reward violations, outcome specific-satiety and during systemic fentanyl administration (Sun et al., 2018).

Using a fluorescent dopamine sensor, GRAB_DA, we characterized phasic dBNST dopamine signals during a range of appetitive Pavlovian conditions including lever autoshaping, reward violations, specific satiety, and fentanyl injections during PLA. We found that dBNST dopamine signals are enhanced in STs compared to GT/INTs during cue presentation and shift from reward to cue across conditioning in STs but not in GT/INTs. Furthermore, dBNST dopamine signals encode bidirectional reward prediction error and are greater in GT/INTs than in STs following reward violations. Additionally, dBNST dopamine signals decrease to cues when rats are sated on food pellets associated with the cue but not when sated on homecage chow. Systemic fentanyl injections do not disrupt dBNST cue discrimination but generally potentiate dBNST dopamine signals.

Pharmacological studies establish that dopamine signaling in the dBNST maintains responding to sucrose and ethanol rewards and regulates the reinforcing properties of cocaine (Eiler et al., 2003; Epping-Jordan et al., 1998). Microdialysis and voltammetry studies show that natural and drug rewards, including opioids, increase DA in the BNST (Carboni et al., 2000; Park et al., 2012; Park et al., 2013). Although dBNST dopamine is important for a variety of appetitive motivated behaviors, little is known about cue-evoked dopamine signaling and its role in cue-triggered motivation. A recent study showed BNST GRAB_DA signals are associated with both cues and rewards (Lin et al., 2020). Our data extend these findings by showing individual differences in CS- and US-evoked BNST dopamine signaling during Pavlovian conditioning. We also demonstrate that CS-evoked BNST DA signals are state-dependent and outcome-specific.

Consistent with prior studies, we observed individual differences in sign- and goal-tracking behaviors elicited by the CS (Boakes, 1977; Hearst and Jenkins, 1974; Nasser et al., 2015; Robinson et al., 2014). Accompanied by this behavioral variation, we observed tracking differences in GRAB_DA signals to CS onset and differences in dopamine signal transfer from US to CS, both of which were stronger in sign-tracking compared to goal-tracking and intermediate rats. We observed a relationship between CS-maintained GRAB_DA signal and PCA scores, indicating sign-tracking approach and interaction with the lever cue is associated with heightened dBNST GRAB_DA signaling. These findings for the dBNST dopamine signal are consistent with prior tracking differences in NAc dopamine signals during Pavlovian lever autoshaping (Flagel et al., 2011). We also find that only in ST rats did GRAB_DA signals adhere to Sutton and Barto, 2018 reinforcement learning algorithm, which states that after learning, reward-evoked signals are temporally transferred back to antecedent cues predicting reward delivery (Nasser et al., 2017; Sutton and Barto, 2018). Consistent with this, we observed an increase in sustained GRAB_DA signal during the entire 10 s CS interaction period on Day 5 of PLA training compared to Day 1. Sustained BNST GRAB_DA signals during the cue interaction period could reflect a number of processes, including (1) ongoing lever interaction, (2) the incentive value gain of the CS, (3) the strength of CS-US association, and/or (4) the back-propagating US to CS signal. Our results suggest that dopamine signaling differences between STs and GTs are not just limited to NAc and could be present across a distributed network receiving dopaminergic projections.

To adapt to environmental changes and learn about future rewards, dopaminergic neurons calculate reward prediction errors (RPE) (Nasser et al., 2017; Schultz et al., 1997; Watabe-Uchida et al., 2017). Here, we examined if BNST GRAB_DA signals encode RPE and whether there are individual differences in dBNST GRAB_DA signals and behavioral strategies following violations of reward expectations. We found that dBNST GRAB_DA signals follow the classical bidirectional prediction error signal such that the signals increased following unexpected reward delivery and decreased following unexpected reward omission. Consistent with attention to learning theories and empirical studies, we observed that rats increase their food cup checking behavior on a trial after a positive or negative reward violation (Calu et al., 2010; Pearce and Hall, 1980; Roesch et al., 2010). Sign-tracking rats increase food cup checking on trials after both unexpected reward delivery and omission, whereas GT/INTs increase food cup checking only after reward omission. Behaviorally this suggests GT/INT rats may be more sensitive to negative reward violations than positive, which is consistent with their sensitivity to outcome devaluation and their insensitivity to conditioned reinforcement (Keefer et al., 2020; Keefer et al., 2022; Kochli et al., 2020; Morrison et al., 2015; Nasser et al., 2015; Robinson and Flagel, 2009; Smedley and Smith, 2018). Such excitatory behavioral responses (more checking for both increases and decreases in reward) before the trial are evidence for an incremental attentional processes, which reflect enhanced attention to environmental predictors for the purpose of increasing the rate of learning for either excitatory or inhibitory associations (Calu et al., 2010; Holland and Gallagher, 1993a; Pearce and Hall, 1980; Roesch et al., 2007; Roesch et al., 2010). Notably, reward prediction errors are critical for such enhanced attentional processes, and the theoretical instantiation of incremental attention for learning about positive and negative reward violations takes the absolute value of RPE signals into account (Pearce and Hall, 1980). Prior work establishes the involvement of other amygdala nuclei, namely the basolateral and central nuclei of the amygdala for encoding unidirectional prediction error signals that track enhanced attention after reward violations (Calu et al., 2010; Roesch et al., 2010). Midbrain dopamine signaling is required for such attentional encoding in the basolateral amygdala (Esber et al., 2012). Here, we identify bidirectional dopamine encoding of positive and negative reward violations in an extended amygdala nuclei, the dBNST. GT/INTs showed evidence for bidirectional RPE in the dBNST DA signal, which may be important for enhancing attention signals in downstream areas.

BNST receives heavy dopaminergic afferents from the A10 Ventral Tegmental Area (VTA) and A10dc ventral periaqueductal gray/dorsal raphe (vPAG/DR) dopaminergic cell groups, and to a lesser extent from the substantia nigra pars compacta and the retrorubral nucleus (Daniel and Rainnie, 2016; Hasue and Shammah-Lagnado, 2002; Melchior et al., 2021; Meloni et al., 2006; Vranjkovic et al., 2017). While VTA and SNc dopamine neurons classically encode bidirectional reward prediction error signals, vPAG dopamine and its projections unidirectionally encode rewarding and aversive outcomes, suggesting salient event detection (Berg et al., 2014; Lin et al., 2020; Nasser et al., 2017; Schultz et al., 1997; Walker et al., 2020; Watabe-Uchida et al., 2017). Different aspects of the dBNST DA signaling we observed lead us to postulate both dopamine projections may be contributing. For the bidirectional RPE, we observed in dBNST, we predict that VTA dopaminergic projections are the source of dopamine during reward violations. In contrast, for the greater CS signaling in ST compared to GT/INT rats may reflect salient features of the CS that support the attracting and reinforcing properties of cues in sign-tracking rats, which may also be supported by vPAG/DR→BNST dopamine. Future studies are needed to identify the extent to which VTA and PAG/DR dopaminergic inputs contribute to the BNST DA signals observed here. Dissecting the role of each dopaminergic input in driving cue and reward-related signaling and behavior will inform whether these circuits work synergistically or in competition to influence appetitive behaviors (Lin et al., 2020; Park et al., 2012; Park et al., 2013). Anatomical studies indicate that a substantial proportion of putative dopaminergic projections to BNST originate in the vPAG/DR (Hasue and Shammah-Lagnado, 2002; Meloni et al., 2006). Prior work has established glutamate and dopamine co-release from the vPAG/DR projection to BNST, positioning this input to directly influence BNST synaptic plasticity and associated behaviors (Li et al., 2016). Regardless, the dopamine dynamics reported here for BNST resemble those previously reported for nucleus accumbens in related behaviors (Clark et al., 2013; Flagel et al., 2011; Hart et al., 2014; Saddoris et al., 2015; Saddoris et al., 2016), suggesting a potential role for VTA DA in shaping BNST DA signaling. Notably, NAc DA also shows greater CS-evoked, and a greater shift from US to CS- evoked DA in ST compared to GT (Flagel et al., 2011). To our knowledge, tracking-related differences in bidirectional RPE signaling in the NAc have not been systematically tested, however, the bidirectional error encoding we observe across all rats is consistent with prior NAc voltammetry studies (Hart et al., 2014). Here, we report that GT/INT, but not ST, show evidence for bidirectional RPE DA signaling in the BNST. Whether this is also the case for NAc DA signaling remains an open question. Consistent with our findings, short inter-trial-intervals (ITI, similar to what we employ here) during autoshaping promote both classic NAc DA RPE signaling and goal-tracking, whereas longer ITIs promote NAc DA CS-salience signaling and sign-tracking (Lee et al., 2018). Pharmacology studies show D1 receptors and NAc DA signaling drive CS-salience in sign-trackers (Chow et al., 2016; Saunders and Robinson, 2012). The potentiating effects of hunger and systemic fentanyl injections on BNST DA signals observed here are in line with effects observed for NAc DA (Bassareo et al., 2013; Castro and Berridge, 2014; Cone et al., 2014; Mahler et al., 2007; Peciña and Berridge, 2005; Wilson et al., 1995). Notably, NAc primarily receives input from the VTA, whereas the BNST receives DA inputs from VTA and vPAG/DR. The tracking-specific differences in BNST dopamine signaling during simple appetitive approach and reward violations observed here suggest either (1) distinct contributions of VTA and vPAG/DR to dopamine signaling observed in BNST and/or (2) individual differences in the engagement of DA systems that bias towards CS-salience or RPE processes (Chow et al., 2016; Lee et al., 2018). Consideration of tracking-specific dopamine signaling differences in future studies that employ projection-specific manipulations will aid in interpreting each projection’s contribution to BNST dopamine signaling and behavior.

A methodological limitation of the current approach is that variations in the expression of fluorescent sensor and/or fiber placement along a gradient of DA input to BNST could potentially influence the magnitude of GRAB_DA measurements. Our fiber placements were largely consistent (~73% at the level of bregma) and overlapped with the densest area of viral expression of the fluorescent sensor. At this level of BNST where we measured the majority of GRAB_DA signals, there is heavy vPAG/DR DA input and to a lesser extent VTA input (Hasue and Shammah-Lagnado, 2002). Other anatomical and/or functional studies that target BNST up to 0.2 mm anterior or posterior to bregma also observe substantial putative dopaminergic input from vPAG/DR (Meloni et al., 2006; Yu et al., 2021a). Regardless, some rats presented in Figure 2—figure supplement 3 were excluded that had sufficient viral expression and fiber placement, but that did not show evidence of dopamine binding in the dBNST during these Pavlovian tasks. While we are limited from drawing conclusions from negative data, such individual differences in extended amygdala dopamine signaling may be important for interpreting differences in appetitive behaviors. In addition, we analyzed signals that were significantly different from the baseline (greater than 2z scores) in our behavioral window. We might have missed some behaviorally relevant signals due to this restriction. Future studies with control GRAB_DA virus are needed to determine how large a signal can be expected from artefactual sources (blood flow, autofluorescence, movement, etc).

The BNST is a sexually dimorphic brain region (Hisasue et al., 2010; Shah et al., 2004; Tsuneoka et al., 2017), highlighting the necessity of studying both sexes to fully understand the contribution of BNST DA to motivated behavior. Dopaminergic projections from vPAG/DR→BNST play sex-specific roles, with pathway activation associated with distinct pain and locomotor behavioral changes for males and females, respectively (Yu et al., 2021b). We used both male and female rats in the present study and analyzed our BNST DA photometry data from Pavlovian autoshaping, RPE, and satiety test sessions using Sex instead of Tracking as a factor. While we observed no sex effects here, prior studies establish BNST-mediated sex differences in pain and locomotor behaviors as well as in opioid withdrawal (Luster et al., 2020; Yu et al., 2021a; Yu et al., 2021b). While there is limited evidence for sex differences in the incubation of fentanyl seeking (a form of relapse), we find this effect to be dependent on dBNST CRFR1 receptor signaling (Gyawali et al., 2020; Reiner et al., 2019; Reiner et al., 2020). Drug-induced synaptic plasticity in the dBNST requires both dopamine and CRF and molecular and electrophysiology studies suggest that DA increases CRF release in the dBNST (Day et al., 2002; Kash et al., 2008). Given the known role of sex differences in CRF-induced relapse and opioid withdrawal, it is critical to include both sexes when studying BNST DA and CRF systems (Buffalari et al., 2012; Luster et al., 2020).

To our surprise, we found evidence for outcome-specific state-dependent BNST GRAB_DA signaling. Consistent with our prior studies, we found that rats decreased their lever responding only when they were sated on food pellets specifically associated with the lever cue, but not when sated on homecage chow (Keefer et al., 2020; Kochli et al., 2020). Similarly, we observed decreased cue-evoked BNST GRAB_DA when rats were sated on food pellets but not when they were sated on chow. All rats ate all their pellets during these reinforced sessions, and we did not see any change in GRAB_DA signals during reward consumption when sated on either food pellets or chow. Prior studies report a similar decrease in cue-evoked dopamine signals in the basolateral amygdala and dopaminergic neuron activity in the dorsal raphe during satiety (Cho et al., 2021; Lutas et al., 2019). Based on these studies that manipulated state using hunger or satiety, we expected dopamine signals to generally decrease to cues both when sated on chow or training pellets, but we found BNST dopamine signals only decreased when sated on the training pellet associated with the cue. However, other studies find evidence for sensory-specific signaling in dopamine function and signaling (Sharpe et al., 2017; Takahashi et al., 2017). This suggests BNST DA signals may carry sensory-specific information that is critical for higher-order learning processes (Burke et al., 2007; Burke et al., 2008 Keefer et al., 2021; Lichtenberg et al., 2017; Lichtenberg et al., 2021; Malvaez et al., 2015; Malvaez et al., 2019; Sias et al., 2021; Sharpe et al., 2017; Takahashi et al., 2017).

Studies show elevated BNST dopamine, dopamine-induced plasticity, and dopamine-mediated seeking behavior during and after drug administration (Carboni et al., 2000; Eiler et al., 2003; Epping-Jordan et al., 1998; Kash et al., 2008; Krawczyk et al., 2013; Krawczyk et al., 2011a; Krawczyk et al., 2011b; Melchior et al., 2021; Stamatakis et al., 2014). We extend these findings by reporting that systemic fentanyl injections do not disrupt dBNST cue discrimination but generally potentiate dBNST dopamine signals. The present study supports the need for future work aimed at fully characterizing drug-induced changes to dBNST DA cue and reward encoding during natural and opioid reward seeking.

Dopamine projections to the BNST are concentrated in the dBNST and synapse specifically onto the CRFergic neurons (Meloni et al., 2006; Phelix et al., 1994). Molecular and electrophysiology studies suggest that dopamine increases local CRF release in the dBNST and drug-induced synaptic plasticity in the dBNST requires both dopamine and CRF (Day et al., 2002; Kash et al., 2008; Silberman et al., 2013a). These anatomical and ex vivo physiology studies suggest dopamine and CRF are critically interacting to drive reward and stress-related behaviors. Indeed, our prior work indicates that CRF receptor activation in the dBNST is necessary for CS-triggered opioid relapse (Gyawali et al., 2020). Furthermore, dBNST dopamine receptor activation decreases blood corticosterone levels in mice suggesting that an increased dopamine response in the dBNST could serve as an anxiolytic signal, which could promote continued drug seeking (Daniel and Rainnie, 2016; Kash et al., 2008; Melchior et al., 2021; Meloni et al., 2006).

The present findings add substantially to the role of dBNST dopamine in motivated behaviors, providing a comprehensive characterization of endogenous dBNST dopamine dynamics in cue-induced behaviors under several natural and drug reward conditions. The fluorescent dopamine sensor GRAB_DA is a useful tool for studying real-time BNST DA dynamics in the context of motivated behaviors (Lin et al., 2020; Sun et al., 2020).

The data used in this manuscript are available on Zenodo (DOI: https://doi.org/10.5281/zenodo.7947009).

1. 1. Gyawali U
   2. Martin DA
   3. Sun F
   4. Li Y
   5. Calu DJ

   (2022) Zenodo

   Dopamine in the Dorsal Bed Nucleus of Stria Terminalis signals Pavlovian sign-tracking and reward violations.

   https://doi.org/10.5281/zenodo.7947009

1. 1. Bacharach SZ
   2. Nasser HM
   3. Zlebnik NE
   4. Dantrassy HM
   5. Kochli DE
   6. Gyawali U
   7. Cheer JF
   8. Calu DJ

   (2018) Cannabinoid Receptor-1 signaling contributions to sign-tracking and conditioned reinforcement in rats

   Psychopharmacology 235:3031–3043.

   https://doi.org/10.1007/s00213-018-4993-6

   • PubMed
   • Google Scholar
2. 1. Bassareo V
   2. Cucca F
   3. Cadoni C
   4. Musio P
   5. Di Chiara G

   (2013) Differential influence of morphine sensitization on accumbens shell and core dopamine responses to morphine- and food-conditioned stimuli

   Psychopharmacology 225:697–706.

   https://doi.org/10.1007/s00213-012-2856-0

   • PubMed
   • Google Scholar
3. 1. Berg BA
   2. Schoenbaum G
   3. McDannald MA

   (2014) The dorsal raphe nucleus is integral to negative prediction errors in pavlovian fear

   The European Journal of Neuroscience 40:3096–3101.

   https://doi.org/10.1111/ejn.12676

   • PubMed
   • Google Scholar
4. Book

   1. Boakes RA

   (1977) Performance on learning to associate a stimulus with positive reinforcement

   In: Boakes RA, editors. Operant-Pavlovian Interactions. Routledge. pp. 67–101.

   https://doi.org/10.4324/9781003150404

   • Google Scholar
5. 1. Buffalari DM
   2. See RE

   (2009) Footshock stress potentiates cue-induced cocaine-seeking in an animal model of relapse

   Physiology & Behavior 98:614–617.

   https://doi.org/10.1016/j.physbeh.2009.09.013

   • PubMed
   • Google Scholar
6. 1. Buffalari DM
   2. Baldwin CK
   3. Feltenstein MW
   4. See RE

   (2012) Corticotrophin releasing factor (CRF) induced reinstatement of cocaine seeking in male and female rats

   Physiology & Behavior 105:209–214.

   https://doi.org/10.1016/j.physbeh.2011.08.020

   • PubMed
   • Google Scholar
7. 1. Burke KA
   2. Franz TM
   3. Miller DN
   4. Schoenbaum G

   (2007) Conditioned reinforcement can be mediated by either outcome-specific or general affective representations

   Frontiers in Integrative Neuroscience 1:2.

   https://doi.org/10.3389/neuro.07.002.2007

   • PubMed
   • Google Scholar
8. 1. Burke KA
   2. Franz TM
   3. Miller DN
   4. Schoenbaum G

   (2008) The role of the Orbitofrontal cortex in the pursuit of happiness and more specific rewards

   Nature 454:340–344.

   https://doi.org/10.1038/nature06993

   • PubMed
   • Google Scholar
9. 1. Calu DJ
   2. Roesch MR
   3. Haney RZ
   4. Holland PC
   5. Schoenbaum G

   (2010) Neural correlates of variations in event processing during learning in central nucleus of amygdala

   Neuron 68:991–1001.

   https://doi.org/10.1016/j.neuron.2010.11.019

   • PubMed
   • Google Scholar
10. 1. Carboni E
    2. Silvagni A
    3. Rolando MT
    4. Di Chiara G

    (2000) Stimulation of in vivo dopamine transmission in the bed nucleus of stria terminalis by reinforcing drugs

    The Journal of Neuroscience 20:RC102.

    https://doi.org/10.1523/JNEUROSCI.20-20-j0002.2000

    • PubMed
    • Google Scholar
11. 1. Castro DC
    2. Berridge KC

    (2014) Opioid hedonic hotspot in nucleus accumbens shell: Mu, Delta, and Kappa maps for enhancement of sweetness "liking" and "wanting"

    The Journal of Neuroscience 34:4239–4250.

    https://doi.org/10.1523/JNEUROSCI.4458-13.2014

    • PubMed
    • Google Scholar
12. 1. Chang SE
    2. Krueger LD
    3. Flagel SB

    (2022) Investigating individual differences in opioid-taking and opioid-seeking behavior in male rats

    Psychopharmacology 239:1065–1080.

    https://doi.org/10.1007/s00213-021-06023-2

    • PubMed
    • Google Scholar
13. 1. Cho JR
    2. Chen X
    3. Kahan A
    4. Robinson JE
    5. Wagenaar DA
    6. Gradinaru V

    (2021) Dorsal raphe dopamine neurons signal motivational salience dependent on internal state, expectation, and behavioral context

    The Journal of Neuroscience 41:2645–2655.

    https://doi.org/10.1523/JNEUROSCI.2690-20.2021

    • PubMed
    • Google Scholar
14. 1. Chow JJ
    2. Nickell JR
    3. Darna M
    4. Beckmann JS

    (2016) Toward isolating the role of dopamine in the acquisition of incentive Salience attribution

    Neuropharmacology 109:320–331.

    https://doi.org/10.1016/j.neuropharm.2016.06.028

    • PubMed
    • Google Scholar
15. 1. Clark JJ
    2. Collins AL
    3. Sanford CA
    4. Phillips PEM

    (2013) Dopamine encoding of pavlovian incentive stimuli diminishes with extended training

    The Journal of Neuroscience 33:3526–3532.

    https://doi.org/10.1523/JNEUROSCI.5119-12.2013

    • PubMed
    • Google Scholar
16. 1. Cone JJ
    2. McCutcheon JE
    3. Roitman MF

    (2014) Ghrelin acts as an interface between physiological state and Phasic dopamine signaling

    The Journal of Neuroscience 34:4905–4913.

    https://doi.org/10.1523/JNEUROSCI.4404-13.2014

    • PubMed
    • Google Scholar
17. 1. Daniel SE
    2. Rainnie DG

    (2016) Stress modulation of opposing circuits in the bed nucleus of the Stria Terminalis

    Neuropsychopharmacology 41:103–125.

    https://doi.org/10.1038/npp.2015.178

    • PubMed
    • Google Scholar
18. 1. Day HEW
    2. Vittoz NM
    3. Oates MM
    4. Badiani A
    5. Watson SJ
    6. Robinson TE
    7. Akil H

    (2002) A 6-hydroxydopamine lesion of the Mesostriatal dopamine system decreases the expression of corticotropin releasing hormone and Neurotensin mRNAs in the amygdala and bed nucleus of the Stria Terminalis

    Brain Research 945:151–159.

    https://doi.org/10.1016/s0006-8993(02)02747-6

    • PubMed
    • Google Scholar
19. 1. Egli RE
    2. Kash TL
    3. Choo K
    4. Savchenko V
    5. Matthews RT
    6. Blakely RD
    7. Winder DG

    (2005) Norepinephrine modulates glutamatergic transmission in the bed nucleus of the Stria Terminalis

    Neuropsychopharmacology 30:657–668.

    https://doi.org/10.1038/sj.npp.1300639

    • PubMed
    • Google Scholar
20. 1. Eiler WJA
    2. Seyoum R
    3. Foster KL
    4. Mailey C
    5. June HL

    (2003) D1 dopamine receptor regulates alcohol-motivated behaviors in the bed nucleus of the Stria Terminalis in alcohol-preferring (P) rats

    Synapse 48:45–56.

    https://doi.org/10.1002/syn.10181

    • PubMed
    • Google Scholar
21. 1. Epping-Jordan MP
    2. Markou A
    3. Koob GF

    (1998) The dopamine D-1 receptor antagonist SCH 23390 injected into the Dorsolateral bed nucleus of the Stria Terminalis decreased cocaine reinforcement in the rat

    Brain Research 784:105–115.

    https://doi.org/10.1016/s0006-8993(97)01190-6

    • PubMed
    • Google Scholar
22. 1. Esber GR
    2. Roesch MR
    3. Bali S
    4. Trageser J
    5. Bissonette GB
    6. Puche AC
    7. Holland PC
    8. Schoenbaum G

    (2012) Attention-related Pearce-Kaye-hall signals in Basolateral amygdala require the Midbrain dopaminergic system

    Biological Psychiatry 72:1012–1019.

    https://doi.org/10.1016/j.biopsych.2012.05.023

    • PubMed
    • Google Scholar
23. 1. Flagel SB
    2. Watson SJ
    3. Robinson TE
    4. Akil H

    (2007) Individual differences in the propensity to approach signals vs goals promote different adaptations in the dopamine system of rats

    Psychopharmacology 191:599–607.

    https://doi.org/10.1007/s00213-006-0535-8

    • PubMed
    • Google Scholar
24. 1. Flagel SB
    2. Clark JJ
    3. Robinson TE
    4. Mayo L
    5. Czuj A
    6. Willuhn I
    7. Akers CA
    8. Clinton SM
    9. Phillips PEM
    10. Akil H

    (2011) A selective role for dopamine in stimulus-reward learning

    Nature 469:53–57.

    https://doi.org/10.1038/nature09588

    • PubMed
    • Google Scholar
25. 1. Flavin SA
    2. Winder DG

    (2013) Noradrenergic control of the bed nucleus of the Stria Terminalis in stress and reward

    Neuropharmacology 70:324–330.

    https://doi.org/10.1016/j.neuropharm.2013.02.013

    • PubMed
    • Google Scholar
26. 1. Gyawali U
    2. Martin DA
    3. Sulima A
    4. Rice KC
    5. Calu DJ

    (2020) Role of BNST Crfr1 receptors in incubation of fentanyl seeking

    Frontiers in Behavioral Neuroscience 14:153.

    https://doi.org/10.3389/fnbeh.2020.00153

    • PubMed
    • Google Scholar
27. Software

    1. Gyawali U

    (2022) Photometry, version swh:1:rev:c5989b197eb45864f354f725e4382f44fc2b8056

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:dir:9a949a5135390dc7055ed61cce1dcb4440b679d6;origin=https://github.com/ugyawali/photometry;visit=swh:1:snp:2a838f7da514ba3da41a74df52c80d35235a4be2;anchor=swh:1:rev:c5989b197eb45864f354f725e4382f44fc2b8056
28. 1. Hart AS
    2. Rutledge RB
    3. Glimcher PW
    4. Phillips PEM

    (2014) Phasic dopamine release in the rat nucleus accumbens symmetrically Encodes a reward prediction error term

    The Journal of Neuroscience 34:698–704.

    https://doi.org/10.1523/JNEUROSCI.2489-13.2014

    • PubMed
    • Google Scholar
29. 1. Hasue RH
    2. Shammah-Lagnado SJ

    (2002) Origin of the dopaminergic Innervation of the central extended amygdala and accumbens Shell: A combined retrograde tracing and immunohistochemical study in the rat

    The Journal of Comparative Neurology 454:15–33.

    https://doi.org/10.1002/cne.10420

    • PubMed
    • Google Scholar
30. Book

    1. Hearst E
    2. Jenkins HM

    (1974)

    Sign-Tracking: The Stimulus-Reinforcer Relation and Directed Action

    Austin, Tex: Psychonomic Society.

    • Google Scholar
31. 1. Hisasue S
    2. Seney ML
    3. Immerman E
    4. Forger NG

    (2010) Control of cell number in the bed nucleus of the Stria Terminalis of mice: Role of testosterone metabolites and estrogen receptor subtypes

    The Journal of Sexual Medicine 7:1401–1409.

    https://doi.org/10.1111/j.1743-6109.2009.01669.x

    • PubMed
    • Google Scholar
32. 1. Holland PC
    2. Gallagher M

    (1993a) Amygdala central nucleus lesions disrupt increments, but not Decrements, in conditioned stimulus processing

    Behavioral Neuroscience 107:246–253.

    https://doi.org/10.1037//0735-7044.107.2.246

    • PubMed
    • Google Scholar
33. 1. Holland PC
    2. Gallagher M

    (1993b) Effects of amygdala central nucleus lesions on blocking and Unblocking

    Behavioral Neuroscience 107:235–245.

    https://doi.org/10.1037/0735-7044.107.2.235

    • Google Scholar
34. 1. Hsu TM
    2. McCutcheon JE
    3. Roitman MF

    (2018) Parallels and overlap: The integration of Homeostatic signals by Mesolimbic dopamine neurons

    Frontiers in Psychiatry 9:410.

    https://doi.org/10.3389/fpsyt.2018.00410

    • PubMed
    • Google Scholar
35. 1. Kash TL
    2. Nobis WP
    3. Matthews RT
    4. Winder DG

    (2008) Dopamine enhances fast excitatory synaptic transmission in the extended amygdala by a CRF-R1-dependent process

    The Journal of Neuroscience 28:13856–13865.

    https://doi.org/10.1523/JNEUROSCI.4715-08.2008

    • PubMed
    • Google Scholar
36. 1. Keefer SE
    2. Bacharach SZ
    3. Kochli DE
    4. Chabot JM
    5. Calu DJ

    (2020) Effects of limited and extended Pavlovian training on devaluation sensitivity of Sign- and goal-tracking rats

    Frontiers in Behavioral Neuroscience 14:3.

    https://doi.org/10.3389/fnbeh.2020.00003

    • PubMed
    • Google Scholar
37. 1. Keefer SE
    2. Gyawali U
    3. Calu DJ

    (2021) Choose your path: Divergent Basolateral amygdala Efferents Differentially mediate incentive motivation, flexibility and decision-making

    Behavioural Brain Research 409:113306.

    https://doi.org/10.1016/j.bbr.2021.113306

    • PubMed
    • Google Scholar
38. Preprint

    1. Keefer SE
    2. Kochli DE
    3. Calu DJ

    (2022) Basolateral Amygdala to Insular Cortex Activity Makes Sign-Tracking Behavior Insensitive to Outcome Value

    bioRxiv.

    https://doi.org/10.1101/2022.02.24.481881

    • Google Scholar
39. 1. Kochli DE
    2. Keefer SE
    3. Gyawali U
    4. Calu DJ

    (2020) Basolateral amygdala to nucleus accumbens communication Differentially mediates devaluation sensitivity of Sign- and goal-tracking rats

    Frontiers in Behavioral Neuroscience 14:593645.

    https://doi.org/10.3389/fnbeh.2020.593645

    • PubMed
    • Google Scholar
40. 1. Krawczyk M
    2. Georges F
    3. Sharma R
    4. Mason X
    5. Berthet A
    6. Bézard E
    7. Dumont EC

    (2011a) Double-dissociation of the catecholaminergic modulation of synaptic transmission in the oval bed nucleus of the Stria Terminalis

    Journal of Neurophysiology 105:145–153.

    https://doi.org/10.1152/jn.00710.2010

    • PubMed
    • Google Scholar
41. 1. Krawczyk M
    2. Sharma R
    3. Mason X
    4. Debacker J
    5. Jones AA
    6. Dumont EC

    (2011b) A switch in the Neuromodulatory effects of dopamine in the oval bed nucleus of the Stria Terminalis associated with cocaine self-administration in rats

    The Journal of Neuroscience 31:8928–8935.

    https://doi.org/10.1523/JNEUROSCI.0377-11.2011

    • PubMed
    • Google Scholar
42. 1. Krawczyk M
    2. Mason X
    3. DeBacker J
    4. Sharma R
    5. Normandeau CP
    6. Hawken ER
    7. Di Prospero C
    8. Chiang C
    9. Martinez A
    10. Jones AA
    11. Doudnikoff É
    12. Caille S
    13. Bézard E
    14. Georges F
    15. Dumont ÉC

    (2013) D1 dopamine receptor-mediated LTP at GABA synapses Encodes motivation to self-administer cocaine in rats

    The Journal of Neuroscience 33:11960–11971.

    https://doi.org/10.1523/JNEUROSCI.1784-13.2013

    • PubMed
    • Google Scholar
43. 1. Labouesse MA
    2. Cola RB
    3. Patriarchi T

    (2020) GPCR-based dopamine sensors-A detailed guide to inform sensor choice for in vivo imaging

    International Journal of Molecular Sciences 21:8048.

    https://doi.org/10.3390/ijms21218048

    • PubMed
    • Google Scholar
44. 1. Langdon AJ
    2. Sharpe MJ
    3. Schoenbaum G
    4. Niv Y

    (2018) Model-based predictions for dopamine

    Current Opinion in Neurobiology 49:1–7.

    https://doi.org/10.1016/j.conb.2017.10.006

    • PubMed
    • Google Scholar
45. 1. Lee B
    2. Gentry RN
    3. Bissonette GB
    4. Herman RJ
    5. Mallon JJ
    6. Bryden DW
    7. Calu DJ
    8. Schoenbaum G
    9. Coutureau E
    10. Marchand AR
    11. Khamassi M
    12. Roesch MR

    (2018) Manipulating the revision of reward value during the Intertrial interval increases sign tracking and dopamine release

    PLOS Biology 16:e2004015.

    https://doi.org/10.1371/journal.pbio.2004015

    • PubMed
    • Google Scholar
46. Preprint

    1. Lee RS
    2. Engelhard B
    3. Witten IB
    4. Daw ND

    (2022) A Vector Reward Prediction Error Model Explains Dopaminergic Heterogeneity

    bioRxiv.

    https://doi.org/10.1101/2022.02.28.482379

    • Google Scholar
47. 1. Lerner TN
    2. Shilyansky C
    3. Davidson TJ
    4. Evans KE
    5. Beier KT
    6. Zalocusky KA
    7. Crow AK
    8. Malenka RC
    9. Luo L
    10. Tomer R
    11. Deisseroth K

    (2015) Intact-brain analyses reveal distinct information carried by SNC dopamine Subcircuits

    Cell 162:635–647.

    https://doi.org/10.1016/j.cell.2015.07.014

    • PubMed
    • Google Scholar
48. 1. Li C
    2. Sugam JA
    3. Lowery-Gionta EG
    4. McElligott ZA
    5. McCall NM
    6. Lopez AJ
    7. McKlveen JM
    8. Pleil KE
    9. Kash TL

    (2016) Mu opioid receptor modulation of dopamine neurons in the Periaqueductal gray/dorsal raphe: A role in regulation of pain

    Neuropsychopharmacology 41:2122–2132.

    https://doi.org/10.1038/npp.2016.12

    • PubMed
    • Google Scholar
49. 1. Lichtenberg NT
    2. Pennington ZT
    3. Holley SM
    4. Greenfield VY
    5. Cepeda C
    6. Levine MS
    7. Wassum KM

    (2017) Basolateral amygdala to Orbitofrontal cortex projections enable cue-triggered reward expectations

    The Journal of Neuroscience 37:8374–8384.

    https://doi.org/10.1523/JNEUROSCI.0486-17.2017

    • PubMed
    • Google Scholar
50. 1. Lichtenberg NT
    2. Sepe-Forrest L
    3. Pennington ZT
    4. Lamparelli AC
    5. Greenfield VY
    6. Wassum KM

    (2021) The medial Orbitofrontal cortex-Basolateral amygdala circuit regulates the influence of reward cues on adaptive behavior and choice

    The Journal of Neuroscience 41:7267–7277.

    https://doi.org/10.1523/JNEUROSCI.0901-21.2021

    • PubMed
    • Google Scholar
51. 1. Lin R
    2. Liang J
    3. Wang R
    4. Yan T
    5. Zhou Y
    6. Liu Y
    7. Feng Q
    8. Sun F
    9. Li Y
    10. Li A
    11. Gong H
    12. Luo M

    (2020) The raphe dopamine system controls the expression of incentive memory

    Neuron 106:498–514.

    https://doi.org/10.1016/j.neuron.2020.02.009

    • PubMed
    • Google Scholar
52. 1. Luster BR
    2. Cogan ES
    3. Schmidt KT
    4. Pati D
    5. Pina MM
    6. Dange K
    7. McElligott ZA

    (2020) Inhibitory transmission in the bed nucleus of the Stria Terminalis in male and female mice following morphine withdrawal

    Addiction Biology 25:e12748.

    https://doi.org/10.1111/adb.12748

    • PubMed
    • Google Scholar
53. 1. Lutas A
    2. Kucukdereli H
    3. Alturkistani O
    4. Carty C
    5. Sugden AU
    6. Fernando K
    7. Diaz V
    8. Flores-Maldonado V
    9. Andermann ML

    (2019) State-specific gating of salient cues by Midbrain dopaminergic input to basal amygdala

    Nature Neuroscience 22:1820–1833.

    https://doi.org/10.1038/s41593-019-0506-0

    • PubMed
    • Google Scholar
54. 1. Mahler SV
    2. Smith KS
    3. Berridge KC

    (2007) Endocannabinoid Hedonic Hotspot for sensory pleasure: Anandamide in nucleus accumbens Shell enhances ‘liking’ of a sweet reward

    Neuropsychopharmacology 32:2267–2278.

    https://doi.org/10.1038/sj.npp.1301376

    • PubMed
    • Google Scholar
55. 1. Malvaez M
    2. Greenfield VY
    3. Wang AS
    4. Yorita AM
    5. Feng L
    6. Linker KE
    7. Monbouquette HG
    8. Wassum KM

    (2015) Basolateral amygdala rapid glutamate release Encodes an outcome-specific representation vital for reward-predictive cues to selectively invigorate reward-seeking actions

    Scientific Reports 5:12511.

    https://doi.org/10.1038/srep12511

    • PubMed
    • Google Scholar
56. 1. Malvaez M
    2. Shieh C
    3. Murphy MD
    4. Greenfield VY
    5. Wassum KM

    (2019) Distinct cortical–amygdala projections drive reward value Encoding and retrieval

    Nature Neuroscience 22:762–769.

    https://doi.org/10.1038/s41593-019-0374-7

    • PubMed
    • Google Scholar
57. 1. Maracle AC
    2. Normandeau CP
    3. Dumont ÉC
    4. Olmstead MC

    (2019) Dopamine in the oval bed nucleus of the Stria Terminalis contributes to compulsive responding for Sucrose in rats

    Neuropsychopharmacology 44:381–389.

    https://doi.org/10.1038/s41386-018-0149-y

    • PubMed
    • Google Scholar
58. 1. Martin DA
    2. Keefer SE
    3. Calu DJ

    (2022) Investigating Discriminative stimulus modulation of opioid seeking after conflict-induced abstinence in Sign- and goal-tracking rats

    Psychopharmacology 239:3223–3236.

    https://doi.org/10.1007/s00213-022-06204-7

    • PubMed
    • Google Scholar
59. 1. Melchior JR
    2. Perez RE
    3. Salimando GJ
    4. Luchsinger JR
    5. Basu A
    6. Winder DG

    (2021) Cocaine augments dopamine mediated inhibition of neuronal activity in the dorsal bed nucleus of the Stria Terminalis

    The Journal of Neuroscience 41:5876–5893.

    https://doi.org/10.1523/JNEUROSCI.0284-21.2021

    • PubMed
    • Google Scholar
60. 1. Meloni EG
    2. Gerety LP
    3. Knoll AT
    4. Cohen BM
    5. Carlezon WA

    (2006) Behavioral and anatomical interactions between dopamine and corticotropin-releasing factor in the rat

    The Journal of Neuroscience 26:3855–3863.

    https://doi.org/10.1523/JNEUROSCI.4957-05.2006

    • PubMed
    • Google Scholar
61. 1. Meyer PJ
    2. Lovic V
    3. Saunders BT
    4. Yager LM
    5. Flagel SB
    6. Morrow JD
    7. Robinson TE

    (2012) Quantifying individual variation in the propensity to attribute incentive Salience to reward cues

    PLOS ONE 7:e38987.

    https://doi.org/10.1371/journal.pone.0038987

    • PubMed
    • Google Scholar
62. 1. Montague PR
    2. Dayan P
    3. Sejnowski TJ

    (1996) A framework for Mesencephalic dopamine systems based on predictive Hebbian learning

    The Journal of Neuroscience 16:1936–1947.

    https://doi.org/10.1523/JNEUROSCI.16-05-01936.1996

    • PubMed
    • Google Scholar
63. 1. Morrison SE
    2. Bamkole MA
    3. Nicola SM

    (2015) Sign tracking, but not goal tracking, is resistant to outcome devaluation

    Frontiers in Neuroscience 9:468.

    https://doi.org/10.3389/fnins.2015.00468

    • PubMed
    • Google Scholar
64. 1. Nasser HM
    2. Chen YW
    3. Fiscella K
    4. Calu DJ

    (2015) Individual variability in behavioral flexibility predicts sign-tracking tendency

    Frontiers in Behavioral Neuroscience 9:289.

    https://doi.org/10.3389/fnbeh.2015.00289

    • PubMed
    • Google Scholar
65. 1. Nasser HM
    2. Calu DJ
    3. Schoenbaum G
    4. Sharpe MJ

    (2017) The dopamine prediction error: Contributions to associative models of reward learning

    Frontiers in Psychology 8:244.

    https://doi.org/10.3389/fpsyg.2017.00244

    • PubMed
    • Google Scholar
66. 1. Page ME
    2. Lucki I

    (2002) Effects of acute and chronic Reboxetine treatment on stress-induced monoamine Efflux in the rat frontal cortex

    Neuropsychopharmacology 27:237–247.

    https://doi.org/10.1016/S0893-133X(02)00301-9

    • PubMed
    • Google Scholar
67. 1. Park J
    2. Kile BM
    3. Wightman RM

    (2009) In vivo Voltammetric monitoring of norepinephrine release in the rat ventral bed nucleus of the Stria Terminalis and Anteroventral thalamic nucleus

    The European Journal of Neuroscience 30:2121–2133.

    https://doi.org/10.1111/j.1460-9568.2009.07005.x

    • PubMed
    • Google Scholar
68. 1. Park J
    2. Wheeler RA
    3. Fontillas K
    4. Keithley RB
    5. Carelli RM
    6. Wightman RM

    (2012) Catecholamines in the bed nucleus of the Stria Terminalis reciprocally respond to reward and aversion

    Biological Psychiatry 71:327–334.

    https://doi.org/10.1016/j.biopsych.2011.10.017

    • PubMed
    • Google Scholar
69. 1. Park J
    2. Bucher ES
    3. Fontillas K
    4. Owesson-White C
    5. Ariansen JL
    6. Carelli RM
    7. Wightman RM

    (2013) Opposing catecholamine changes in the bed nucleus of the Stria Terminalis during intracranial self-stimulation and its extinction

    Biological Psychiatry 74:69–76.

    https://doi.org/10.1016/j.biopsych.2012.11.008

    • PubMed
    • Google Scholar
70. 1. Patriarchi T
    2. Cho JR
    3. Merten K
    4. Howe MW
    5. Marley A
    6. Xiong WH
    7. Folk RW
    8. Broussard GJ
    9. Liang R
    10. Jang MJ
    11. Zhong H
    12. Dombeck D
    13. von Zastrow M
    14. Nimmerjahn A
    15. Gradinaru V
    16. Williams JT
    17. Tian L

    (2018) Ultrafast neuronal imaging of dopamine Dynamics with designed genetically encoded sensors

    Science 360:eaat4422.

    https://doi.org/10.1126/science.aat4422

    • PubMed
    • Google Scholar
71. Book

    1. Paxinos G
    2. Watson C

    (2006)

    The Rat Brain in Stereotaxic Coordinates

    Elsevier.

    • Google Scholar
72. 1. Pearce JM
    2. Hall G

    (1980)

    A model for Pavlovian learning: Variations in the effectiveness of conditioned but not of unconditioned stimuli

    Psychological Review 87:532–552.

    • PubMed
    • Google Scholar
73. 1. Peciña S
    2. Berridge KC

    (2005) Hedonic hot spot in nucleus accumbens Shell: Where do mu-opioids cause increased Hedonic impact of sweetness?

    The Journal of Neuroscience 25:11777–11786.

    https://doi.org/10.1523/JNEUROSCI.2329-05.2005

    • PubMed
    • Google Scholar
74. 1. Phelix CF
    2. Liposits Z
    3. Paull WK

    (1994) Catecholamine-CRF synaptic interaction in a septal bed nucleus: Afferents of neurons in the bed nucleus of the Stria Terminalis

    Brain Research Bulletin 33:109–119.

    https://doi.org/10.1016/0361-9230(94)90056-6

    • PubMed
    • Google Scholar
75. 1. Pitchers KK
    2. Phillips KB
    3. Jones JL
    4. Robinson TE
    5. Sarter M

    (2017) Diverse roads to relapse: A Discriminative cue signaling cocaine availability is more effective in renewing cocaine seeking in goal Trackers than sign Trackers and depends on basal forebrain cholinergic activity

    The Journal of Neuroscience 37:7198–7208.

    https://doi.org/10.1523/JNEUROSCI.0990-17.2017

    • PubMed
    • Google Scholar
76. 1. Reiner DJ
    2. Fredriksson I
    3. Lofaro OM
    4. Bossert JM
    5. Shaham Y

    (2019) Relapse to opioid seeking in rat models: Behavior, pharmacology and circuits

    Neuropsychopharmacology 44:465–477.

    https://doi.org/10.1038/s41386-018-0234-2

    • PubMed
    • Google Scholar
77. 1. Reiner DJ
    2. Lofaro OM
    3. Applebey SV
    4. Korah H
    5. Venniro M
    6. Cifani C
    7. Bossert JM
    8. Shaham Y

    (2020) Role of projections between Piriform cortex and Orbitofrontal cortex in relapse to fentanyl seeking after palatable food choice-induced voluntary abstinence

    The Journal of Neuroscience 40:2485–2497.

    https://doi.org/10.1523/JNEUROSCI.2693-19.2020

    • PubMed
    • Google Scholar
78. 1. Robinson TE
    2. Flagel SB

    (2009) Dissociating the predictive and incentive motivational properties of reward-related cues through the study of individual differences

    Biological Psychiatry 65:869–873.

    https://doi.org/10.1016/j.biopsych.2008.09.006

    • PubMed
    • Google Scholar
79. 1. Robinson TE
    2. Yager LM
    3. Cogan ES
    4. Saunders BT

    (2014) On the motivational properties of reward cues: Individual differences

    Neuropharmacology 76 Pt B:450–459.

    https://doi.org/10.1016/j.neuropharm.2013.05.040

    • PubMed
    • Google Scholar
80. 1. Roesch MR
    2. Calu DJ
    3. Schoenbaum G

    (2007) Dopamine neurons Encode the better option in rats deciding between differently delayed or sized rewards

    Nature Neuroscience 10:1615–1624.

    https://doi.org/10.1038/nn2013

    • PubMed
    • Google Scholar
81. 1. Roesch MR
    2. Calu DJ
    3. Esber GR
    4. Schoenbaum G

    (2010) Neural correlates of variations in event processing during learning in Basolateral amygdala

    The Journal of Neuroscience 30:2464–2471.

    https://doi.org/10.1523/JNEUROSCI.5781-09.2010

    • PubMed
    • Google Scholar
82. 1. Root DH
    2. Barker DJ
    3. Estrin DJ
    4. Miranda-Barrientos JA
    5. Liu B
    6. Zhang S
    7. Wang H-L
    8. Vautier F
    9. Ramakrishnan C
    10. Kim YS
    11. Fenno L
    12. Deisseroth K
    13. Morales M

    (2020) Distinct signaling by ventral Tegmental area glutamate, GABA, and Combinatorial glutamate-GABA neurons in motivated behavior

    Cell Reports 32:108094.

    https://doi.org/10.1016/j.celrep.2020.108094

    • PubMed
    • Google Scholar
83. 1. Saddoris MP
    2. Cacciapaglia F
    3. Wightman RM
    4. Carelli RM

    (2015) Differential dopamine release Dynamics in the nucleus accumbens core and Shell reveal complementary signals for error prediction and incentive motivation

    The Journal of Neuroscience 35:11572–11582.

    https://doi.org/10.1523/JNEUROSCI.2344-15.2015

    • PubMed
    • Google Scholar
84. 1. Saddoris MP

    (2016) Terminal dopamine release Kinetics in the accumbens core and Shell are distinctly altered after withdrawal from cocaine self-administration

    ENeuro 3:ENEURO.0274-16.2016.

    https://doi.org/10.1523/ENEURO.0274-16.2016

    • PubMed
    • Google Scholar
85. 1. Saddoris MP
    2. Wang X
    3. Sugam JA
    4. Carelli RM

    (2016) Cocaine self-administration experience induces pathological Phasic accumbens dopamine signals and abnormal incentive behaviors in drug-abstinent rats

    The Journal of Neuroscience 36:235–250.

    https://doi.org/10.1523/JNEUROSCI.3468-15.2016

    • PubMed
    • Google Scholar
86. 1. Saunders BT
    2. Robinson TE

    (2010) A cocaine cue acts as an incentive stimulus in some but not others: Implications for addiction

    Biological Psychiatry 67:730–736.

    https://doi.org/10.1016/j.biopsych.2009.11.015

    • PubMed
    • Google Scholar
87. 1. Saunders BT
    2. Robinson TE

    (2012) The role of dopamine in the accumbens core in the expression of Pavlovian-conditioned responses

    The European Journal of Neuroscience 36:2521–2532.

    https://doi.org/10.1111/j.1460-9568.2012.08217.x

    • PubMed
    • Google Scholar
88. 1. Saunders BT
    2. Yager LM
    3. Robinson TE

    (2013) Cue-evoked cocaine "craving": Role of dopamine in the accumbens core

    The Journal of Neuroscience 33:13989–14000.

    https://doi.org/10.1523/JNEUROSCI.0450-13.2013

    • PubMed
    • Google Scholar
89. 1. Schultz W
    2. Dayan P
    3. Montague PR

    (1997) A neural substrate of prediction and reward

    Science 275:1593–1599.

    https://doi.org/10.1126/science.275.5306.1593

    • PubMed
    • Google Scholar
90. 1. Schultz W

    (2015) Neuronal reward and decision signals: From theories to data

    Physiological Reviews 95:853–951.

    https://doi.org/10.1152/physrev.00023.2014

    • PubMed
    • Google Scholar
91. 1. Shah NM
    2. Pisapia DJ
    3. Maniatis S
    4. Mendelsohn MM
    5. Nemes A
    6. Axel R

    (2004) Visualizing sexual Dimorphism in the brain

    Neuron 43:313–319.

    https://doi.org/10.1016/j.neuron.2004.07.008

    • PubMed
    • Google Scholar
92. 1. Sharpe MJ
    2. Chang CY
    3. Liu MA
    4. Batchelor HM
    5. Mueller LE
    6. Jones JL
    7. Niv Y
    8. Schoenbaum G

    (2017) Dopamine transients are sufficient and necessary for acquisition of model-based associations

    Nature Neuroscience 20:735–742.

    https://doi.org/10.1038/nn0817-1189e

    • PubMed
    • Google Scholar
93. 1. Sias AC
    2. Morse AK
    3. Wang S
    4. Greenfield VY
    5. Goodpaster CM
    6. Wrenn TM
    7. Wikenheiser AM
    8. Holley SM
    9. Cepeda C
    10. Levine MS
    11. Wassum KM

    (2021) A Bidirectional Corticoamygdala circuit for the Encoding and retrieval of detailed reward memories

    eLife 10:e68617.

    https://doi.org/10.7554/eLife.68617

    • PubMed
    • Google Scholar
94. 1. Silberman Y
    2. Matthews RT
    3. Winder DG

    (2013a) A corticotropin releasing factor pathway for ethanol regulation of the ventral Tegmental area in the bed nucleus of the Stria Terminalis

    The Journal of Neuroscience 33:950–960.

    https://doi.org/10.1523/JNEUROSCI.2949-12.2013

    • PubMed
    • Google Scholar
95. 1. Silberman Y
    2. Winder DG

    (2013b) Emerging role for corticotropin releasing factor signaling in the bed nucleus of the Stria Terminalis at the intersection of stress and reward

    Frontiers in Psychiatry 4:42.

    https://doi.org/10.3389/fpsyt.2013.00042

    • PubMed
    • Google Scholar
96. 1. Smedley EB
    2. Smith KS

    (2018) Evidence of structure and persistence in motivational attraction to serial Pavlovian cues

    Learning & Memory 25:78–89.

    https://doi.org/10.1101/lm.046599.117

    • PubMed
    • Google Scholar
97. 1. Stamatakis AM
    2. Sparta DR
    3. Jennings JH
    4. McElligott ZA
    5. Decot H
    6. Stuber GD

    (2014) Amygdala and bed nucleus of the Stria Terminalis circuitry: Implications for addiction-related behaviors

    Neuropharmacology 76 Pt B:320–328.

    https://doi.org/10.1016/j.neuropharm.2013.05.046

    • PubMed
    • Google Scholar
98. 1. Sun F
    2. Zeng J
    3. Jing M
    4. Zhou J
    5. Feng J
    6. Owen SF
    7. Luo Y
    8. Li F
    9. Wang H
    10. Yamaguchi T
    11. Yong Z
    12. Gao Y
    13. Peng W
    14. Wang L
    15. Zhang S
    16. Du J
    17. Lin D
    18. Xu M
    19. Kreitzer AC
    20. Cui G
    21. Li Y

    (2018) A genetically encoded fluorescent sensor enables rapid and specific detection of dopamine in flies, fish, and mice

    Cell 174:481–496.

    https://doi.org/10.1016/j.cell.2018.06.042

    • PubMed
    • Google Scholar
99. 1. Sun F
    2. Zhou J
    3. Dai B
    4. Qian T
    5. Zeng J
    6. Li X
    7. Zhuo Y
    8. Zhang Y
    9. Wang Y
    10. Qian C
    11. Tan K
    12. Feng J
    13. Dong H
    14. Lin D
    15. Cui G
    16. Li Y

    (2020) Next-generation GRAB sensors for monitoring dopaminergic activity in vivo

    Nature Methods 17:1156–1166.

    https://doi.org/10.1038/s41592-020-00981-9

    • PubMed
    • Google Scholar
100. Book

     1. Sutton RS
     2. Barto AG

     (2018)

     Reinforcement Learning: An Introduction Adaptive Computation and Machine Learning Series

     A Bradford Book.

     • Google Scholar
101. 1. Takahashi YK
     2. Batchelor HM
     3. Liu B
     4. Khanna A
     5. Morales M
     6. Schoenbaum G

     (2017) Dopamine neurons respond to errors in the prediction of sensory features of expected rewards

     Neuron 95:1395–1405.

     https://doi.org/10.1016/j.neuron.2017.08.025

     • PubMed
     • Google Scholar
102. 1. Tsuneoka Y
     2. Tsukahara S
     3. Yoshida S
     4. Takase K
     5. Oda S
     6. Kuroda M
     7. Funato H

     (2017) Moxd1 is a marker for sexual Dimorphism in the medial Preoptic area, bed nucleus of the Stria Terminalis and medial amygdala

     Frontiers in Neuroanatomy 11:26.

     https://doi.org/10.3389/fnana.2017.00026

     • PubMed
     • Google Scholar
103. 1. Vranjkovic O
     2. Pina M
     3. Kash TL
     4. Winder DG

     (2017) The bed nucleus of the Stria Terminalis in drug-associated behavior and affect: A circuit-based perspective

     Neuropharmacology 122:100–106.

     https://doi.org/10.1016/j.neuropharm.2017.03.028

     • PubMed
     • Google Scholar
104. 1. Walker RA
     2. Wright KM
     3. Jhou TC
     4. McDannald MA

     (2020) The ventrolateral Periaqueductal grey updates fear via positive prediction error

     The European Journal of Neuroscience 51:866–880.

     https://doi.org/10.1111/ejn.14536

     • PubMed
     • Google Scholar
105. 1. Watabe-Uchida M
     2. Eshel N
     3. Uchida N

     (2017) Neural circuitry of reward prediction error

     Annual Review of Neuroscience 40:373–394.

     https://doi.org/10.1146/annurev-neuro-072116-031109

     • PubMed
     • Google Scholar
106. 1. Wilson C
     2. Nomikos GG
     3. Collu M
     4. Fibiger HC

     (1995) Dopaminergic correlates of motivated behavior: importance of drive

     The Journal of Neuroscience 15:5169–5178.

     https://doi.org/10.1523/JNEUROSCI.15-07-05169.1995

     • PubMed
     • Google Scholar
107. 1. Yu W
     2. Caira CM
     3. Del R Rivera Sanchez N
     4. Moseley GA
     5. Kash TL

     (2021a) Corticotropin-releasing factor neurons in the bed nucleus of the Stria Terminalis exhibit sex-specific pain Encoding in mice

     Scientific Reports 11:12500.

     https://doi.org/10.1038/s41598-021-91672-8

     • PubMed
     • Google Scholar
108. 1. Yu W
     2. Pati D
     3. Pina MM
     4. Schmidt KT
     5. Boyt KM
     6. Hunker AC
     7. Zweifel LS
     8. McElligott ZA
     9. Kash TL

     (2021b) Periaqueductal gray/dorsal raphe dopamine neurons contribute to sex differences in pain-related behaviors

     Neuron 109:1365–1380.

     https://doi.org/10.1016/j.neuron.2021.03.001

     • PubMed
     • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "Dopamine in the Dorsal Bed Nucleus of Stria Terminalis signals Pavlovian sign-tracking and reward violations" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Michael Taffe as the Senior Editor. The reviewers have opted to remain anonymous.

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:

1) Additional data and analyses for the fentanyl self administration experiments are needed given the low sample size. Relatedly, some explanation for exclusion of 26 rats is warranted.

2) There should be a more thorough discussion of inputs from different DA sources.

3) Analysis of shuffled vs. time-locked signal to control for spontaneous fluorescence is a minimum requirement, though ideally some pilot data with a control sensor (see individual reviewer comments).

4) Better comparison of sign-tracking vs. goal-tracking in the fentanyl experiment and tighter integration with reward prediction error (RPE) analyses.

Reviewer #1 (Recommendations for the authors):

This study is well conducted and provides new information as to how DA is signaling within the BNST during reward-relevant behaviors. At present though I have some questions that I think would guide/improve the study:

The big question is where is the DA coming from during these relevant behaviors. The BNST receives DA from both the VTA and the PAG but the anatomical distribution of this innervation is distinct https://pubmed.ncbi.nlm.nih.gov/26792442/. The discussion gets into this a bit, however, the lack of pathway-specific studies is an absence to fully understanding how what is being observed in the BNST is functionally distinct from what is happening with DA encoding in the NAc.

26 rats were excluded for various reasons, 2 additional for catheter patency.

The methods state that transients were excluded when they were not equal to or greater than 2 z-scores above baseline. This seems excessive and may miss smaller events? Perhaps this wording is confusing in the methods bc while the transients are over 2, the "sustained DA" as measured in AUC seems to be under 2z. Further, the transients in Figure 2B don't appear to be over 2z in the representative traces. This being said, I've recently been made aware that there are issues with autofluorescence, and blood flow that can result in optical signals that do not correspond to the detection of an analyte. The authors should demonstrate that they do not observe behaviorally relevant signals with the control GRAB DA virus. (https://www.addgene.org/140555/)

In discussing dBNST NE input, it would be good to cite the following where NE release was directly measured in the dBNST: https://pubmed.ncbi.nlm.nih.gov/20128849/ (see Figure 3).

Were there any differences in dBNST DA signaling with the fentanyl SA between the sign trackers and the goal trackers?

The authors state: "Microdialysis studies establish that several classes of drugs of abuse, including opioids, increase tonic DA in the BNST (Carboni et al., 2000)." As microdialysis cannot distinguish tonic vs. phasic dopamine signals, I would not describe this as "tonic" DA.

For discussion of phasic dopamine in the BNST with voltammetry and opioids (Line 561), this should be cited: https://www.nature.com/articles/npp2016135

The discussion on sex differences within dopaminergic signaling in the BNST would benefit from the following papers examining PAG dopamine neurons, pain, and sex differences: https://www.nature.com/articles/s41598-021-91672-8, https://pubmed.ncbi.nlm.nih.gov/33740416/

Reviewer #2 (Recommendations for the authors):

1) The authors interestingly find that, after intermittent fentanyl self-administration, cues associated with fentanyl reward appear to trigger enhanced BNST dopamine release. However, the small group of rats (n=3) makes it difficult to appropriately make this assertion in the same thorough manner as in earlier experiments. This interesting trend would be further strengthened with a larger group of rats that allow enough power to detect interaction in Figure 5C, for example. It would also be interesting to know whether the 3 rats tested in the fentanyl probe test were sign-tracking or goal-tracking rats.

2) How do the GRABDA expression/optic fiber placements overlap with dopaminergic fibers in dBNST? – For example, how does the density of dopaminergic terminals span dBNST, from rostral to caudal? This could potentially account for differences in the magnitude of the signal seen between rats, or between goal and sign trackers (although most of their placements look to be among mostly overlapping Bregma points).

3) In the methods section (lines 114-115), the text states that rats were excluded if there was a 'lack of significant GRABDA signal during behavioral event compared to baseline (N=26)". Please clarify what behavioral event was used. Did the experimenters use a lack of response to reward or reward CS+ in the PLA as being indicative of a low/no signal? This could be potentially important if only those rats that showed dopamine response to reward and reward cues during PLA were included and others were not. In that case, it would mean a substantial proportion of animals where dBNST dopamine release does not occur in dBNST during said tasks. In such a case, that proportion should be reported and traces depicted in Figure 1 (or supplement to Figure 1).

4) Regarding the analysis in Figure 3, I appreciate that the authors chose to show z-scores for each of the 2s time bins following reward delivery/non-delivery. This is important in light of the fact that judging from the traces, there seems to be a biphasic signal during negative reward error trials (a brief positive response followed by a negative signal). Yet, it is unclear why this analysis strategy is abandoned for the insets of Figures3G-H that show analysis that averages the signal across the entire 6s bin (as opposed to the above Figure 3D which parses each time bin). As a result, it appears that there is no effect of positive or negative RPE signals (especially in STs). Instead, in panels G-H, it might make more sense to compare z scores during each bin among STs and then among GTs. Otherwise, it's difficult to discern whether STs show any negative reward prediction signaling in dBNST. Relatedly, for the same set of analyses in Fig3G-H (also in text lines 406-412), comparing positive to negative trial z scores among each type of rat (goal- or sign- tracking) seems an odd choice – because these trials are independent and it's unclear what is gained from comparing the two. Rather, a more insightful analysis could be to see whether z-scores on positive or negative trials among each group differ significantly from zero (i.e., >2) or from z-scores on 'expected' trials.

Reviewer #3 (Recommendations for the authors):

I have comments on some of the analysis and interpretations of the data overall.

The RPE manipulation is a nice addition and really broadens the scope of the dopamine investigation. I'm a little confused about the approach to determining if a positive or negative RPE is signaled by the recorded GRAB fluorescence, however. Generally, it seems like the changes in dopamine based on expectation violations are temporally specific, which motivates the data binning in Figure 3 – but it's a bit unclear what statistical comparisons are significant. Maybe an area under/below the curve analysis would help this a little.

Related, the positive vs negative RPE comparisons for STs vs GT/INT is also confusing – it seems that only positive vs negative trials are contrasted statistically (Figure 3G+H). This is where the difference between tracking groups comes – with STs not differing between positive and negative, but GT/INTs showing elevated signals in positive vs negative trials. First, which part/bin/time of the signal being compared here is not clear. Second, to me in order to really say that a positive or negative RPE has been signaled the dopamine response would need to be different in positive vs expected and expected vs negative conditions. The fact that GT/INT dopamine more clearly discriminates against positive and negative expectation violations is still meaningful but it doesn't seem quite the same as "encode bidirectional RPE" without further analysis. Overall I feel like the analysis of this section could be beefed up and expanded. I also think the extension of RPE encoding questions to the BNST is the most impactful part of the data.

More or less, the dopamine signals recorded in BNST follow classic striatal/midbrain dopamine encoding. That is interesting and to me a little unexpected, given the role BNST has in stress, anxiety, and other negative states. Perhaps a little more discussion of how these signals do and do not compare to classic striatal dopamine is warranted. Also given the quite distinct dopamine signals seen in the tail of the striatum, which also come from nigra dopamine neurons (rather than raphe and another place), it is also surprising to see such "normal" RPE-related dynamics in a non-striatal region.

The satiety experiment and fentanyl results are interesting, but in the scope of the paper in the current form, they felt disconnected, especially given that the ST/GT tracking component of the investigation is not carried through. It just feels a bit like two different papers, perhaps these elements of the data could be better linked.

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

Thank you for resubmitting your work entitled "Dopamine in the Dorsal Bed Nucleus of Stria Terminalis signals Pavlovian sign-tracking and reward violations" for further consideration by eLife. Your revised article has been evaluated by me as 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:

The authors have introduced new questions about rat exclusion criteria that need to be clarified. This could be addressed by showing more of the excluded rat data since it is unclear if authors may have excluded rats that had a good signal but missed placements. This would also bolster the central RPE conclusions and specificity of the signals.

Reviewer #1 (Recommendations for the authors):

I am still enthusiastic about this study and the contributions it makes to the field. I appreciate the authors' careful and thoughtful reply to the reviewers; however, I am still a bit concerned about some facets of the paper and look forward to discussing this with the other reviewers in the consult session.

In particular, I am concerned that the authors did not conduct the shuffling analysis, which the reviewers requested was important. The bootstrapping method is interesting, but I am not sure if animals should be excluded based on it, perhaps the other reviewers can elaborate. In the same vein, the distinct BNST inputs (VTA and PAG) release DA very differently, therefore smaller signals may be relevant, and they are still excluding small signals. The fiber placement is too large to determine if there is input coming into the oval BNST (more PAG input) vs. the juxtacapsular (more VTA input); and, it is concerning that there were some rats with no signals at all (contributing to a large number of animals that were excluded from the study).

Finally, the removal of the fentanyl SA study, while focusing on the paper, does eliminate some of the excitement. I do hope the authors add additional N and either publish with this manuscript or in a subsequent manuscript

Reviewer #2 (Recommendations for the authors):

The authors here addressed all my concerns, regarding the inclusion of fentanyl self-administration, analysis of ST/GT RPE photometry, and expansion of important discussion points regarding dopamine inputs to dBNST. I also appreciate their added explanation of exclusion criteria in photometry experiments. The overall revisions help strengthen the authors' conclusion that dBNST dopamine contributes to cue-induced motivated behavior and is influenced by factors such as satiety and opioids. These findings add new and important insight to the role of dBNST in reward-related behavior beyond the more classical role of negative/stress-related motivation.

Reviewer #3 (Recommendations for the authors):

Thanks to the authors for this revised manuscript. My original comments have been addressed and I think the paper is stronger and more focused. I have a few remaining comments mostly about the excluded rats.

A little clarification. In the rebuttal, the authors say "We had 20 rats that were not excluded for technical reasons. Of these, 4 rats were excluded for not meeting our (revised) standards of having sufficient photometry signals (see below)." But then in the revised manuscript text, it says – "We excluded rats because of lack of viral expression (N = 4), incorrect fiber optic placements (N = 6), and headcap loss (N = 4). Additionally, we excluded rats (N = 8) that showed food cup entry artifacts before we optimized our photometry setup. The artifacts resulted in loss of signal due to the patch cord hitting the wall of the foodcup. Finally, rats (N = 4) presented in Figure 2 —figure supplement 3 were excluded that had sufficient viral expression and fiber placement but did not show robust photometry signals".

- Unless I'm misunderstanding it sounds like only 4 rats were excluded for non-technical reasons and the other 22 had various technical issues.

For the 6 rats with misplaced optic fibers – did these rats have measurable dopamine signals, or is this a matter of missing the virus expression completely? If there are rats with viable signals that happen to be outside of the BNST that would be an interesting control comparison. Alternatively, if these 6 rats have measured photometry signals but the fibers aren't above virus expression, you could include them as controls with the new confidence interval analyses in a supplement – presumably there will be no meaningful autofluorescence signals in these rats. Either way, this could potentially further strengthen the main dataset.

Did the 4 excluded rats with no cue signals develop conditioned behavior at similar levels to the main data set rats? I would suggest including the behavioral data for these rats alongside the photometry data in the corresponding supplemental figure.

8 rats were excluded because of signal artifacts associated with the port entry. It's unclear from the behavior data presented when rats are making port entries (presumably this varies substantially by tracking phenotype) but if there is substantial CS onset data that is not contaminated by port entry, that could be a meaningful addition given the large size of this exclusive group.

Essential revisions:

1) Additional data and analyses for the fentanyl experiments are needed given the low sample size. Relatedly, some explanation for exclusion of 26 rats is warranted.

As detailed below in point by point response to reviewers, we have refocused the manuscript on sign- and goal-tracking differences during Pavlovian approach and reward violations.

We have removed the fentanyl self-administration data, which not only were conceptually extraneous, but also had low sample size. We have detailed our justification of rat exclusions and include a supplemental figure that parallels figure 2 for the rats with signals that failed to reach our statistical threshold for inclusion. We added a statement in the discussion that describes the possibility that some rats may have been excluded that had sufficient viral expression and fiber placement, but that did not show evidence of dopamine binding in the dBNST.

2) There should be a more thorough discussion of inputs from different DA sources.

We have added discussion on dopaminergic inputs to BNST.

3) Analysis of shuffled vs. time-locked signal to control for spontaneous fluorescence is a minimum requirement, though ideally some pilot data with a control sensor (see individual reviewer comments).

To address concerns regarding the strength and timing of our signals relative to cue presentation, we reanalyzed all animals with any discernible signals on Day 5 of PLA during CS+ presentations using a bootstrap analysis to generate 95% confidence intervals across time. We implemented a criterion that the signals must significantly increase above the baseline signal within 70 ms of CS+ onset and stay above baseline for a minimum of 40 ms, consecutively. This second requirement greatly reduces the possibility that a chance signal crossing above 0 is mistaken for a real signal and reduces the corresponding false positive rate. All animals included in our original analysis meet these criteria, while 4 animals were excluded (Note these animals had all previously been excluded for not reaching the 2 z-score peak height minimum) – as now stated in the Methods section and shown in Figure 2 —figure supplement 3. While we cannot address the issue of blood flow changes or other biological phenomena leading to fluorescence changes without conducting lengthy additional experiments with a control sensor, we have no reason to doubt that our signals reflect dopamine given the extensive characterization of these GRAB_DA sensors in other brain areas in which verified dopamine signals share very similar time courses of fluorescence changes as the signals we observe here.

4) Better comparison of sign-tracking vs. goal-tracking in the fentanyl experiment and tighter integration with reward prediction error (RPE) analyses.

We have substantially revised our RPE analysis based on reviewer 2’s thoughtful feedback. These analyses now include time bin as a factor and explicitly compare error trial types to expected reward which strengthen the conclusion that goal- but not sign-, trackers show evidence for dBNST dopamine bidirectional reward prediction error signaling.

Reviewer #1 (Recommendations for the authors):

This study is well conducted and provides new information as to how DA is signaling within the BNST during reward-relevant behaviors. At present though I have some questions that I think would guide/improve the study:

The big question is where is the DA coming from during these relevant behaviors. The BNST receives DA from both the VTA and the PAG but the anatomical distribution of this innervation is distinct https://pubmed.ncbi.nlm.nih.gov/26792442/. The discussion gets into this a bit, however, the lack of pathway-specific studies is an absence to fully understanding how what is being observed in the BNST is functionally distinct from what is happening with DA encoding in the NAc.

We agree this is an interesting question and something we plan to explore in the future. Ultimately, identifying the source of dopamine within the BNST during reward relevant behaviors is out of the scope of the present manuscript. However, we hope that the work presented here will spur future research on this topic. We expand on this topic in the Discussion section:

“Future studies are needed to identify the extent to which VTA and PAG/DR dopaminergic inputs contribute to the BNST DA signals observed here. Dissecting the role of each dopaminergic input in driving cue and reward related signaling and behavior will inform whether these circuits work synergistically or in competition to influence appetitive behaviors (Lin et al., 2020; Park et al., 2012, 2013). Anatomical studies indicate that a substantial proportion of putative dopaminergic projections to BNST originate in the vPAG/DR (Hasue and Shammah-Lagnado, 2002, Meloni et al., 2006). Prior work has established glutamate and dopamine co-release from the vPAG/DR projection to BNST, positioning this input to directly influence BNST synaptic plasticity and associated behaviors (Li et al., 2016). Regardless, the dopamine dynamics reported here for BNST resemble those previously reported for nucleus accumbens in related behaviors (Clark et al., 2013; Flagel et al., 2011; Hart et al., 2014; Saddoris et al., 2015, 2016), suggesting a potential role for VTA DA in shaping BNST DA signaling…” “…The tracking specific differences in BNST dopamine signaling during simple appetitive approach and reward violations observed here suggest either (1) distinct contributions of VTA and vPAG/DR to dopamine signaling observed in BNST and/or (2) individual differences in the engagement of DA systems that bias towards CSsalience or RPE processes (Chow et al., 2016, Lee et al., 2018). Consideration of tracking-specific dopamine signaling differences in future studies that employ projection-specific manipulations will aid in interpreting each projection’s contribution to BNST dopamine signaling and behavior.”

26 rats were excluded for various reasons, 2 additional for catheter patency.

The methods state that transients were excluded when they were not equal to or greater than 2 z-scores above baseline. This seems excessive and may miss smaller events? Perhaps this wording is confusing in the methods bc while the transients are over 2, the "sustained DA" as measured in AUC seems to be under 2z. Further, the transients in Figure 2B don't appear to be over 2z in the representative traces. This being said, I've recently been made aware that there are issues with autofluorescence, and blood flow that can result in optical signals that do not correspond to the detection of an analyte. The authors should demonstrate that they do not observe behaviorally relevant signals with the control GRAB DA virus. (https://www.addgene.org/140555/)

We agree more detail is needed to justify rat exclusions. We have now added more robust exclusion criteria for rats that displayed viral expression and proper fiber placement, but not high levels of signal during training (4 rats) in the form of a consecutive, confidence interval test. We have now expanded the reasons in the Subjects subsection of Methods and the Photometry Analysis subsection of the Methods section:

Methods

“Subjects

…We excluded rats because of lack of viral expression (N = 4), incorrect fiber optic placements (N = 6), and headcap loss (N = 4). Additionally, we excluded rats (N = 8) that showed food cup entry artifacts before we optimized our photometry setup. The artifacts resulted in loss of signal due to the patch cord hitting the wall of the foodcup. Finally, rats (N = 4) presented in Figure 2 —figure supplement 3 were excluded that had sufficient viral expression and fiber placement but did not show robust photometry signals in the dBNST during CS+ presentation by day five of PLA training (see Photometry Analysis subsection for more details).”

“Photometry Analysis

…We defined significant transients in our behavioral window if the peak amplitude during the trials (0 to +20 s relative to cue onset) was 2 z-score (p=0.05) above baseline (5s prior to cue onset) during the entire behavioral window on Day 1 or Day 5 of PLA. Furthermore, to ensure these signals were time-locked to cues and not spurious, we calculated 95% confidence intervals using bootstrapped resampling (1000 resamples) of all trials' photometry data for each rat across CS+ trials of Day 5 of PLA. Most rats displayed a consistent, robust increase in signal reaching significantly above baseline within 70 milliseconds of CS+ onset, that stayed above baseline for a minimum of 40 milliseconds, consecutively. 4 rats did not meet either of these criteria (greater than 2 z-score peak signal or 40 ms of consecutive time with significantly elevated signal at CS+) and were excluded. Their data is in Figure 2 —figure supplement 3. All included rats met both criteria. We also removed trials where the patch cord disconnected from further signal processing.”

The addition of the consecutive-confidence-interval criteria should rule out spurious signals, and potentially autofluorescence artifacts. Some rats (N = 4) had viral expression and correct fiber placement but did not show peak amplitude 2 z-score above baseline (during the trial) on either Day 1 or Day 5 of PLA. In Figure 2—figure supplement 3, we now show parallel analysis to Figure 2 from excluded rats that do not meet the 2 z-score peak and threshold and confidence interval analyses. Based on these analyses we are confident we are not missing smaller events, as excluded rats with expression do not show robust event related activity changes. We also added a brief discussion about excluded rats that had sufficient viral expression and fiber placement, but that did not show evidence of dopamine binding in the dBNST. As noted above in this response letter, our signals display time courses consistent with well characterized dopamine signaling using this sensor in other brain areas, leading us to conclude that blood flow changes are unlikely to be the source of the signal.

In discussing dBNST NE input, it would be good to cite the following where NE release was directly measured in the dBNST: https://pubmed.ncbi.nlm.nih.gov/20128849/ (see Figure 3).

We now include this reference and add the finding about relatively slow NE clearance in the dBNST:

“Even though the GRAB_DA construct we used is 15-fold more sensitive to dopamine than norepinephrine (NE), BNST norepinephrine plays an important role in motivated behaviors and dBNST receives dense noradrenergic input with relatively slow NE clearance measured in vivo (Egli et al., 2005; Flavin and Winder, 2013; Park et al., 2009; Sun et al., 2020).”

Were there any differences in dBNST DA signaling with the fentanyl SA between the sign trackers and the goal trackers?

This is a very interesting question, but since our sample size for this phase of the study was low, we have now removed the fentanyl self-administration data. Instead, we now focus the manuscript on sign- and goal-tracking differences during Pavlovian approach and reward violations.

The authors state: "Microdialysis studies establish that several classes of drugs of abuse, including opioids, increase tonic DA in the BNST (Carboni et al., 2000)." As microdialysis cannot distinguish tonic vs. phasic dopamine signals, I would not describe this as "tonic" DA.

We agree with the reviewer and have changed the sentence accordingly:

“Microdialysis studies establish that several classes of drugs of abuse, including opioids, increase DA in the BNST (Carboni et al., 2000).”

For discussion of phasic dopamine in the BNST with voltammetry and opioids (Line 561), this should be cited: https://www.nature.com/articles/npp2016135

We now add the citation the reviewer suggests.

The discussion on sex differences within dopaminergic signaling in the BNST would benefit from the following papers examining PAG dopamine neurons, pain, and sex differences: https://www.nature.com/articles/s41598-021-91672-8, https://pubmed.ncbi.nlm.nih.gov/33740416/

We agree with the reviewer and as such have edited existing text and added a brief discussion based on the findings from these studies.

“The BNST is a sexually dimorphic brain region (Hisasue et al., 2010; Shah et al., 2004; Tsuneoka et al., 2017), highlighting the necessity of studying both sexes to fully understand the contribution of BNST DA to motivated behavior. Dopaminergic projections from vPAG/DR-BNST play sexspecific roles, with pathway activation associated with distinct pain and locomotor behavioral changes for males and females, respectively (Yu et al., 2021). We used both male and female rats in the present study and analyzed our BNST DA photometry data from Pavlovian autoshaping, RPE, satiety test sessions using Sex instead of Tracking as a factor. While we observed no sex effects here, prior studies establish BNST-mediated sex differences in pain and locomotor behaviors as well as in opioid withdrawal (Luster et al., 2020; Yu et al. 2021a; Yu et al., 2021b).”

Reviewer #2 (Recommendations for the authors):

1) The authors interestingly find that, after intermittent fentanyl self-administration, cues associated with fentanyl reward appear to trigger enhanced BNST dopamine release. However, the small group of rats (n=3) makes it difficult to appropriately make this assertion in the same thorough manner as in earlier experiments. This interesting trend would be further strengthened with a larger group of rats that allow enough power to detect interaction in Figure 5C, for example. It would also be interesting to know whether the 3 rats tested in the fentanyl probe test were sign-tracking or goal-tracking rats.

The three rats tested were N = 2 intermediate (mix of sign- and goal-tracking approach) and N = 1 sign-tracker. The self-administration phase of the experiment was not a primary aim of the study, though we agree the data are interesting. Only one cohort of rats was trained in fentanyl selfadministration, and unfortunately, we were not sufficiently powered to look within tracking groups. Conceptually, this departs from our aim to characterize individual differences in extended amygdala dopamine dynamics. Furthermore, we do not currently have the expertise (first author graduated) or resources to replicate this phase of the study, so we have removed data from the self-administration phase of the study from the manuscript. We look forward to pursuing these questions in future studies.

2) How do the GRABDA expression/optic fiber placements overlap with dopaminergic fibers in dBNST? – For example, how does the density of dopaminergic terminals span dBNST, from rostral to caudal? This could potentially account for differences in the magnitude of the signal seen between rats, or between goal and sign trackers (although most of their placements look to be among mostly overlapping Bregma points).

As the reviewer keenly points out, our placements were largely consistent (~73% (N=11) at level of bregma and only ~13% (N=2) slightly anterior(+0.12) and ~20 % (N=3) slightly posterior (-0.12) to bregma), limiting our ability to analyze the data with sufficient power to determine whether variability in expression and/or fiber placement contributed to differences in the signal along the anterior to posterior axis of BNST. Regardless, we think this is an important consideration and have consulted anatomical and functional studies (1 relevant case, no BNST coordinates given, estimated target– based on lateral ventricle, fornix and anterior commissure landmarks in Hasue and Shammah-Lagnado, 2002: Figure 7B– around 0.0, bregma, in rat), Meloni et al., 2006 (stereotaxic target -0.2 posterior to bregma in rat), Yu et al., 2021 (stereotaxic target +0.23 anterior to bregma in mouse). From these studies that demonstrate DA input anatomically and/or functionally that spans anterior (Yu et al., 2021) to posterior (Meloni et al., 2006) of our reported cases mostly around bregma (ie: Hasue and Shammah-Lagnado, 2002), we conclude there is substantial vPAG/DR DA input and to a lesser extent VTA input at the level of BNST that we measure GRABDA signals. We appreciate any further insight the reviewer may have for anatomical or functional studies that increase resolution on rostral-caudal distribution of dopaminergic inputs to BNST. We have added the following sentences to the discussion to address this important methodological limitation of our approach:

“A methodological limitation of the current approach is that variations in expression of fluorescent sensor and/or fiber placement along a gradient of DA input to BNST could potentially influence the magnitude of GRAB_DA measurements. Our fiber placements were largely consistent (~73% at level of bregma) and overlapped with the densest area of viral expression of the fluorescent sensor. At this level of BNST that we measure the majority of GRAB_DA signals, there is heavy vPAG/DR DA input and to a lesser extent VTA input (Hasue and Shammah-Lagnado, 2002). Other anatomical and/or functional studies that target BNST up to 0.2 mm anterior or posterior to bregma also observe substantial putative dopaminergic input from vPAG/DR (Meloni et al., 2006, Yu et al., 2021).”

3) In the methods section (lines 114-115), the text states that rats were excluded if there was a 'lack of significant GRABDA signal during behavioral event compared to baseline (N=26)". Please clarify what behavioral event was used. Did the experimenters use a lack of response to reward or reward CS+ in the PLA as being indicative of a low/no signal? This could be potentially important if only those rats that showed dopamine response to reward and reward cues during PLA were included and others were not. In that case, it would mean a substantial proportion of animals where dBNST dopamine release does not occur in dBNST during said tasks. In such a case, that proportion should be reported and traces depicted in Figure 1 (or supplement to Figure 1).

We regret that we did not list all the reasons for animal exclusions. We have now expanded the reasons for exclusions in the Subjects and Photometry Analysis subsection of the Methods section. These changes are also detailed above in response to Question 2 by Reviewer #1.

The reviewer’s concerns are valid regarding exclusions of animals without a dopamine response. 4 rats had viral expression and correct fiber placement but did not meet our (revised) standards of having sufficient photometry signals. In Figure 2—figure supplement 3, we now show parallel analysis to Figure 2 including signals from excluded rats that do not meet the 2 z-score threshold and confidence interval analyses. Based on these analyses we are confident we are not missing smaller events, as excluded rats with expression do not show event related activity changes. We had 20 rats that were not excluded for technical reasons. Of these, 4 rats were excluded for not meeting our (revised) standards of having sufficient photometry signals (see below). As we do not know why these animals did not show sufficient signal, we hesitate to speculate that this lack of signal really does reflect low dopamine in these animals, as it may be due to unidentified technical issues instead. Nevertheless, the reviewer makes a valid point, as the lack of signal may be indeed biological. We have added this point and the ratio of the animals kept vs discarded (4/20 = 0.2) in the discussion.

Methods

“Subjects

...We excluded rats because of lack of viral expression (N = 4), incorrect fiber optic placements (N = 6), and headcap loss (N = 4). Additionally, we excluded rats (N = 8) that showed food cup entry artifacts before we optimized our photometry setup. The artifacts resulted in loss of signal due to the patch cord hitting the wall of the foodcup. Finally, rats (N = 4) presented in Figure 2 —figure supplement 3 were excluded that had sufficient viral expression and fiber placement but did not show robust photometry signals in the dBNST during CS+ presentation by day five of PLA training (see Photometry Analysis subsection for more details).”

“Photometry Analysis

…We defined significant transients in our behavioral window if the peak amplitude during the trials (0 to +20 s relative to cue onset) was 2 z-score (p=0.05) above baseline (5s prior to cue onset) during the entire behavioral window on Day 1 or Day 5 of PLA. Furthermore, to ensure these signals were time-locked to cues and not spurious, we calculated 95% confidence intervals using bootstrapped resampling (1000 resamples) of all trials' photometry data for each rat across CS+ trials of Day 5 of PLA. Most rats displayed a consistent, robust increase in signal reaching significantly above baseline within 70 milliseconds of CS+ onset, that stayed above baseline for a minimum of 40 milliseconds, consecutively. 4 rats did not meet either of these criteria (greater than 2 z-score peak signal or 40 ms of consecutive time with significantly elevated signal at CS+) and were excluded. Their data is in Figure 2 —figure supplement 3. All included rats met both criteria. We also removed trials where the patch cord disconnected from further signal processing.”

Discussion

“…Regardless, some rats presented in Figure 2 —figure supplement 3 were excluded that had sufficient viral expression and fiber placement, but that did not show evidence of dopamine binding in the dBNST during these Pavlovian tasks. While we are limited from drawing conclusions from negative data, such individual differences in extended amygdala dopamine signaling may be important for interpreting differences in appetitive behaviors.”

4) Regarding the analysis in Figure 3, I appreciate that the authors chose to show z-scores for each of the 2s time bins following reward delivery/non-delivery. This is important in light of the fact that judging from the traces, there seems to be a biphasic signal during negative reward error trials (a brief positive response followed by a negative signal). Yet, it is unclear why this analysis strategy is abandoned for the insets of Figures3G-H that show analysis that averages the signal across the entire 6s bin (as opposed to the above Figure 3D which parses each time bin). As a result, it appears that there is no effect of positive or negative RPE signals (especially in STs). Instead, in panels G-H, it might make more sense to compare z scores during each bin among STs and then among GTs. Otherwise, it's difficult to discern whether STs show any negative reward prediction signaling in dBNST. Relatedly, for the same set of analyses in Fig3G-H (also in text lines 406-412), comparing positive to negative trial z scores among each type of rat (goal- or sign- tracking) seems an odd choice – because these trials are independent and it's unclear what is gained from comparing the two. Rather, a more insightful analysis could be to see whether z-scores on positive or negative trials among each group differ significantly from zero (i.e., >2) or from z-scores on 'expected' trials.

We thank the reviewer for pointing out this analysis shortfall. We have conducted suggested analyses which appear on the manuscript and below:

“First, to determine whether BNST GRAB_DA signals encode bidirectional reward prediction error (RPE), we compare signals on expected, positive and negative trials. Notably, because lever retraction occurs simultaneously with reward delivery, and sign- and goal-trackers may be in different locations at this time, we examine the signals during the six seconds (three 2-s bins) after reward delivery or omission, which captures the period corresponding to violations in reward expectations (Figure 3D). We performed a repeated measures ANOVA on z scores during the RPE session including Trial Type (Expected, Positive, Negative) and Bin (three 2 s bins (0-2 s, 2-4 s, 4-6 s)) as factors. We observed a difference in dBNST GRAB_DA signaling between the three trial types in the bins following reward delivery/omission (Figure 3D, Bin: F_(2,72) = 13.65, p<0.001, Bin x Trial Type: F_(4,72) = 13.99, p<0.001, Trial Type: F_(2,36) = 3.49, p=0.041). Post-hocs confirm that in the second 2-second bin (ie. 2-4 s) after reward delivery/omission, BNST GRAB_DA signals differed from one another for all three trial types, Expected vs. Positive (population traces in Figure 3E; p=0.013), Expected vs. Negative (population traces in Figure 3F; p=0.043) and Positive vs. Negative (p=0.0004). Across all rats, we observe that dBNST GRAB_DA signals reflect bidirectional reward prediction errors.

Then to determine whether there are tracking differences in dBNST RPE signals, we separately analyzed the z scores during RPE sessions in the two tracking groups. Again we examine how GRAB_DA signaling differs for the three trial types (expected, positive, negative) during the three 2second bins after reward delivery (population traces for STs and GT/INTs Figure 3G-H). In GT/INT rats we observed main effects of Trial (F(2,12) = 8.2, p = 0.006) and Bin (F(2,12) = 4.9, p = 0.027) and a Trial x Bin interaction (F(4,24) = 25.7, p <0.001). GT/INT rats showed evidence for both positive RPE (Trial (Expected, Positive) x Bin interaction F(2,12) = 14.5, p = 0.001) and negative RPE Trial (Expected, Negative) x Bin interaction (F(2,12) = 9.9, p = 0.003; Figure 3G inset). In GT/INT rats, we next examined the time course and found dBNST GRAB_DA signaling on both positive and negative trials differs from expected trials during the second 2-second bin (ie 2-4 s) after lever retraction/pellet delivery/omission (positive vs. expected p = 0.04, negative vs. expected p = 0.021). This suggests GT/INT rats show evidence for dBNST GRAB_DA bidirectional RPE signaling.

In a parallel analysis in ST rats considering all trial types (Expected, Positive, Negative) we also observed main effect of Bin (F(2,10) = 10.4, p = 0.004) and Trial x Bin interaction (F(4,20) = 5.4, p = 0.004). ST rats showed evidence for positive RPE (Trial (expected, positive) x Bin interaction F(2,10) 6.8, p = 0.014) but not negative RPE (Trial (expected, negative) x Bin interaction F(2,10) = 2.8, p = 0.153, Figure 3H inset). Post-hoc analyses in ST rats on the time-course failed to identify which bin GRABDA signals distinguished by trial type, however a planned analysis on the relevant second 2-s bin (ie. 2-4 s after lever retraction/pellet delivery/omission) indicates a main effect (F(2,10) = 5.3, p = 0.027) is marginally driven by Expected vs. Positive trial types (F(1,5)=5.0, p = 0.075) and not Expected vs. Negative trial types (F(1,5) = 2.0, p=0.216). This analysis suggests ST rats fail to show evidence for dBNST GRAB_DA bidirectional RPE signaling.”

Reviewer #3 (Recommendations for the authors):

I have comments on some of the analysis and interpretations of the data overall.

The RPE manipulation is a nice addition and really broadens the scope of the dopamine investigation. I'm a little confused about the approach to determining if a positive or negative RPE is signaled by the recorded GRAB fluorescence, however. Generally, it seems like the changes in dopamine based on expectation violations are temporally specific, which motivates the data binning in Figure 3 – but it's a bit unclear what statistical comparisons are significant. Maybe an area under/below the curve analysis would help this a little.

We thank the reviewer for pointing out the ambiguity that previously existed for this analysis and associated Figure 3. We have revised both the figure and the text to clarify the updated statistical analysis for the RPE sessions. The entire section of revised text appears in manuscript and also in response to the final point of reviewer 2, just above. We paste subsections of this revised text in the next point to address the specific issues outlined in the point below.

Related, the positive vs negative RPE comparisons for STs vs GT/INT is also confusing – it seems that only positive vs negative trials are contrasted statistically (Figure 3G+H). This is where the difference between tracking groups comes – with STs not differing between positive and negative, but GT/INTs showing elevated signals in positive vs negative trials. First, which part/bin/time of the signal being compared here is not clear. Second, to me in order to really say that a positive or negative RPE has been signaled the dopamine response would need to be different in positive vs expected and expected vs negative conditions. The fact that GT/INT dopamine more clearly discriminates against positive and negative expectation violations is still meaningful but it doesn't seem quite the same as "encode bidirectional RPE" without further analysis. Overall I feel like the analysis of this section could be beefed up and expanded. I also think the extension of RPE encoding questions to the BNST is the most impactful part of the data.

We thank the reviewer for the constructive feedback and positive assessment of the work. We have addressed these comments and performed repeated measures ANOVA on the timecourse of the RPE signals in ST and GT/INT rats. Below is a sub-section of the revised text accompanying Figure 3.

“In GT/INT rats we observed main effects of Trial (F_(2,12) = 8.2, p = 0.006) and Bin (F(2,12) = 4.9, p = 0.027) and a Trial x Bin interaction (F_(4,24) = 25.7, p <0.001). GT/INT rats showed evidence for both positive RPE (Trial (Expected, Positive) x Bin interaction F_(2,12) = 14.5, p = 0.001) and negative RPE Trial (Expected, Negative) x Bin interaction (F(2,12) = 9.9, p = 0.003; Figure 3G inset). In GT/INT rats, we next examined the time course and found dBNST GRAB_DA signaling on both positive and negative trials differs from expected trials during the second 2-second bin (ie 2-4 s) after lever retraction/pellet delivery/omission (positive vs. expected p = 0.04, negative vs. expected p = 0.021). This suggests GT/INT rats show evidence for dBNST GRAB_DA bidirectional RPE signaling.

In a parallel analysis in ST rats considering all trial types (Expected, Positive, Negative) we also observed main effect of Bin (F_(2,10) = 10.4, p = 0.004) and Trial x Bin interaction (F_(4,20) = 5.4, p = 0.004). ST rats showed evidence for positive RPE (Trial (expected, positive) x Bin interaction F_(2,10) = 6.8, p = 0.014) but not negative RPE (Trial (expected, negative) x Bin interaction F_(2,10) = 2.8, p = 0.153, Figure 3H inset). Post-hoc analyses in ST rats on the time-course failed to identify which bin GRABDA signals distinguished by trial type, however a planned analysis on the relevant second 2-s bin (ie. 2-4 s after lever retraction/pellet delivery/omission) indicates a main effect (F_(2,10) = 5.3, p = 0.027) is marginally driven by Expected vs. Positive trial types (F_(1,5)=5.0, p = 0.075) and not Expected vs. Negative trial types (F_(1,5)= 2.0, p=0.216). This analysis suggests ST rats fail to show evidence for dBNST GRAB_DA bidirectional RPE signaling.”

More or less, the dopamine signals recorded in BNST follow classic striatal/midbrain dopamine encoding. That is interesting and to me a little unexpected, given the role BNST has in stress, anxiety, and other negative states. Perhaps a little more discussion of how these signals do and do not compare to classic striatal dopamine is warranted. Also given the quite distinct dopamine signals seen in the tail of the striatum, which also come from nigra dopamine neurons (rather than raphe and another place), it is also surprising to see such "normal" RPE-related dynamics in a non-striatal region.

We agree and speculate a bit in the discussion about how individual differences in BNST DA signaling may arise from differences in the relative contribution of distinct inputs (VTA-BNST in bidirectional RPE seen in GT/INT group, and potentially vPAG-BNST involvement in strong CS incentive salience encoding of ST group.) Unlike the tail of the striatum, BNST receives relatively little DA input from substantia nigra (Hasue and Shammah-Lagnado, 2002; Meloni et al., 2006). In addition to the prior text briefly touching on this matter, we have added the following to the discussion.

“Regardless, the dopamine dynamics reported here for BNST resemble those previously reported for nucleus accumbens in related behaviors (Clark et al., 2013; Flagel et al., 2011; Hart et al., 2014; Saddoris et al., 2015, 2016), suggesting a potential role for VTA DA in shaping BNST DA signaling. Notably, NAc DA also shows greater CS-evoked, and a greater shift from US to CS- evoked DA in ST compared to GT (Flagel et al., 2011). To our knowledge, tracking-related differences in bidirectional RPE signaling in the NAc have not been systematically tested, however the bidirectional error encoding we observe across all rats is consistent with prior NAc voltammetry study (Hart et al., 2014). Here we report that GT/INT, but not ST, show evidence for bidirectional RPE DA signaling in the BNST. Whether this is also the case for NAc DA signaling remains an open question. Consistent with our findings, short inter-trial-intervals (ITI, similar to what we employ here) during autoshaping promote both classic NAc DA RPE signaling and goaltracking, whereas longer ITIs promote NAc DA CS-salience signaling and sign-tracking (Lee et al., 2018). Pharmacology studies show D1 receptors and NAc DA signaling drive CS-salience in sign-trackers (Chow et al., 2016; Saunders and Robinson 2012). The potentiating effects of hunger and systemic fentanyl injections on BNST DA signals observed here are in line with effects observed for NAc DA (Bassareo et al., 2013; Castro and Berridge 2014; Cone et al., 2014; Mahler et al., 2007; Pecina and Berridge 2005; Wilson et al., 1995). Notably, the NAc primarily receives input from the VTA, whereas the BNST receives DA inputs from VTA and vPAG/DR. The tracking specific differences in BNST dopamine signaling during simple appetitive approach and reward violations observed here suggest either (1) distinct contributions of VTA and vPAG/DR to dopamine signaling observed in BNST and/or (2) individual differences in the engagement of DA systems that bias towards CS-salience or RPE processes (Chow et al., 2016; Lee et al., 2018). Consideration of tracking-specific dopamine signaling differences in future studies that employ projection-specific manipulations will aid in interpreting each projection’s contribution to BNST dopamine signaling and behavior.”

The satiety experiment and fentanyl results are interesting, but in the scope of the paper in the current form, they felt disconnected, especially given that the ST/GT tracking component of the investigation is not carried through. It just feels a bit like two different papers, perhaps these elements of the data could be better linked.

We agree with the reviewer, particularly about the self-administration results, which were conceptually inconsistent with our goal to characterize individual differences in BNST DA signaling during Pavlovian behaviors. The self-administration phase of the experiment was not a primary aim of the study, though we agree the data are interesting. Only one cohort of rats was trained in fentanyl self-administration, and unfortunately, we were not sufficiently powered to look within tracking groups. Furthermore, we do not currently have the expertise (first author graduated) or resources to replicate this phase of the study, so we have removed data from the self-administration phase of the study from the manuscript. We look forward to pursuing these questions in future studies.

Given we have refocused the manuscript on BNST DA signaling during natural reward seeking, we revise the manuscript to better incorporate the satiety and i.p. fentanyl data. We feel these experiments are in line with our goal to characterize BNST DA signaling in natural reward seeking. While we do not observe tracking differences in the satiety data, we are cautious to interpret the lack of main effects or interactions with Tracking factor, due to the smaller sample size (pellet satiety N = 11, ST N = 4 , GT/INT N = 7 and chow satiety N = 7, ST N = 3 , GT/INT N = 4) in this phase of the study. We now report these N’s in the revised text. We have also added more rationale and associated citations to better incorporate these findings to what is known for the striatal dopamine system. A few excerpts are included below.

“In the current and following sections, we report the number of ST and GT/INT rats for each experimental phase but do not report tracking differences due to decreased statistical power to detect group differences. Prior studies indicate that the midbrain and striatal dopamine system tracks motivational state through satiety-dependent changes in the magnitude of dopamine responses (Cone et al., 2014; Hsu et al., 2018; Wilson et al., 1995). Here we determined whether motivational state also decreases task-related BNST GRABDA signals during lever autoshaping.”

“These results further bolster our finding that BNST GRAB_DA signals encode cue-outcome associations, which similar to striatal dopamine signaling, is blunted when the animal has reduced motivational drive (Cone et al., 2014; Wilson et al., 1995).“

“Opioids potentiate NAc activity and NAc DA responses to natural rewards and natural reward associated cues (Bassareo et al., 2013; Castro and Berridge 2014; Mahler et al., 2007; Pecina and Berridge 2005). Here we sought to determine whether opioids also potentiate task-related BNST GRAB_DA signals during natural reward seeking in lever autoshaping.”

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

Reviewer #1 (Recommendations for the authors):

I am still enthusiastic about this study and the contributions it makes to the field. I appreciate the authors' careful and thoughtful reply to the reviewers; however, I am still a bit concerned about some facets of the paper and look forward to discussing this with the other reviewers in the consult session.

In particular, I am concerned that the authors did not conduct the shuffling analysis, which the reviewers requested was important. The bootstrapping method is interesting, but I am not sure if animals should be excluded based on it, perhaps the other reviewers can elaborate.

We have conducted the shuffling analysis, as we interpreted the editor’s request, the result of which is shown below. In this time shuffled analysis, for each trial we randomized actual z-scored dBNSTs GRABDA signal values, such that the values occur at random times. We then average each rats’ CS+ Trials across the session. Then we averaged across rats for the group level (GT and ST) analysis. Displayed below are the average curves (real (blue) vs. shuffled (red) CS+ data) across the groups (Shaded is the SEM) for the first (left column) fifth (right column) session of Pavlovian Lever autoshaping. The result of the time shuffled data analysis is the expected increase in baseline signal, with no event-related fluctuations. We do not feel this analysis adds substantive information, and include it as Author response image 1.

Author response image 1

Download asset Open asset

Only 4 rats were excluded for non-technical reasons (sufficient viral expression and fiber placement but did not show robust photometry signals). These rats’ data is presented in Figure 2 —figure supplement 3. We are not suggesting that smaller GRABDA signals are irrelevant. However, because smaller events are more likely to be contaminated by noise (ie. autofluorescence), we apply a standard approach in signal analysis of examining signals in our behavioral window that are significantly different from baseline (p<0.05 or z>2). Our additional confidence interval analysis further supports our approach. We have added a couple of sentences in the Discussion section to highlight this caveat:“A methodological limitation of the current approach … In addition, we analyzed signals that were significantly different from baseline (greater than 2z scores) in our behavioral window. We might have missed some behaviorally relevant signals due to this restriction. Future studies with control GRABDA virus are needed to determine how large a signal can be expected from artefactual sources (blood flow, autofluorescence, movement, etc).”

In the same vein, the distinct BNST inputs (VTA and PAG) release DA very differently, therefore smaller signals may be relevant, and they are still excluding small signals. The fiber placement is too large to determine if there is input coming into the oval BNST (more PAG input) vs. the juxtacapsular (more VTA input); and, it is concerning that there were some rats with no signals at all (contributing to a large number of animals that were excluded from the study).

We agree with the reviewer that distinct BNST inputs (VTA vs PAG) release dopamine differently in the BNST. Indeed, the fiber placement area is too large to determine the source of dopamine. Our goal of the manuscript was not to determine the source of dopamine, nor make any claim on what the source of dopamine signals are. Instead, our study points to the general role of BNST dopamine in response to reward associated cues and reward prediction error in sign and goal tracking rats. Here are snippets from Discussion that highlight the need for future investigation related to this point:

“Future studies are needed to identify the extent to which VTA and PAG/DR dopaminergic inputs contribute to the BNST DA signals observed here. Dissecting the role of each dopaminergic input in driving cue and reward related signaling and behavior will inform whether these circuits work synergistically or in competition to influence appetitive behaviors (Lin et al., 2020; Park et al., 2012, 2013)…”

“… The tracking specific differences in BNST dopamine signaling during simple appetitive approach and reward violations observed here suggest either (1) distinct contributions of VTA and vPAG/DR to dopamine signaling observed in BNST and/or (2) individual differences in the engagement of DA systems that bias towards CS-salience or RPE processes (Chow et al., 2016; Lee et al., 2018). Consideration of tracking-specific dopamine signaling differences in future studies that employ projection-specific manipulations will aid in interpreting each projection’s contribution to BNST dopamine signaling and behavior.”

Finally, the removal of the fentanyl SA study, while focusing on the paper, does eliminate some of the excitement. I do hope the authors add additional N and either publish with this manuscript or in a subsequent manuscript

We thank the reviewer for their interest in this future work extending our findings to drug reward signaling.

Reviewer #3 (Recommendations for the authors):

Thanks to the authors for this revised manuscript. My original comments have been addressed and I think the paper is stronger and more focused. I have a few remaining comments mostly about the excluded rats.

A little clarification. In the rebuttal, the authors say "We had 20 rats that were not excluded for technical reasons. Of these, 4 rats were excluded for not meeting our (revised) standards of having sufficient photometry signals (see below)." But then in the revised manuscript text, it says – "We excluded rats because of lack of viral expression (N = 4), incorrect fiber optic placements (N = 6), and headcap loss (N = 4). Additionally, we excluded rats (N = 8) that showed food cup entry artifacts before we optimized our photometry setup. The artifacts resulted in loss of signal due to the patch cord hitting the wall of the foodcup. Finally, rats (N = 4) presented in Figure 2 —figure supplement 3 were excluded that had sufficient viral expression and fiber placement but did not show robust photometry signals".

– Unless I'm misunderstanding it sounds like only 4 rats were excluded for non-technical reasons and the other 22 had various technical issues.

The reviewer is correct. 4 rats were excluded for non-technical reasons and the other 22 had various technical issues. This was a typo in the rebuttal. We now also include behavior data from those 4 rats in Figure 2 —figure supplement 3.

For the 6 rats with misplaced optic fibers – did these rats have measurable dopamine signals, or is this a matter of missing the virus expression completely? If there are rats with viable signals that happen to be outside of the BNST that would be an interesting control comparison. Alternatively, if these 6 rats have measured photometry signals but the fibers aren't above virus expression, you could include them as controls with the new confidence interval analyses in a supplement – presumably there will be no meaningful autofluorescence signals in these rats. Either way, this could potentially further strengthen the main dataset.

All 6 rats had dBNST viral expression but misplaced fiber placements outside of the dBNST. Most of these rats were run for 3-5 sessions before they were no longer imaged due to lack of real time signal and limited photometry setups. Regardless, we analyzed the fluorescence signals from the last recording session from these rats (session 3-5 depending on when they were no longer run) and qualitatively found no consistent fluctuations in signal in the behavioral window. We also performed a confidence interval analysis on these 6 rats. Only 1/6 of these rats with misplaced cannula exhibited a signal significantly and consistently above baseline following CS+ onset (according to the parameters of our bootstrapping confidence interval test described in the methods) on the last day of training, however, the average maximum of this signal was very small, and did not exceed 2z above baseline. Therefore, all six rats with misplaced cannula had little to no appreciable signal.

Did the 4 excluded rats with no cue signals develop conditioned behavior at similar levels to the main data set rats? I would suggest including the behavioral data for these rats alongside the photometry data in the corresponding supplemental figure.

Yes, these rats developed conditioned behavior similar to the main data set rats. The data are now added in Figure 2 —figure supplement 3.

8 rats were excluded because of signal artifacts associated with the port entry. It's unclear from the behavior data presented when rats are making port entries (presumably this varies substantially by tracking phenotype) but if there is substantial CS onset data that is not contaminated by port entry, that could be a meaningful addition given the large size of this exclusive group.

Out of the 8 rats, there were 2 STs and 6 GT/INTs. The population photometry signals and associated behavior from these rats are now added as a supplement in Figure 2 —figure supplement 4. Since we also do unconditioned stimulus (US) analyses, we exclude these rats completely from the study since we do not have reliable signal from the time of US delivery for these rats. However, qualitatively we observe that these rats show similar patterns of CS-evoked dBNST dopamine signals and behavior relative to rats included in the reported data set (Figure 2) that have the continuous uncontaminated signals across the trial.

Author details

1. Utsav Gyawali

   1. Program in Neuroscience, University of Maryland School of Medicine, Baltimore, United States
   2. Department of Anatomy and Neurobiology, University of Maryland School of Medicine, Baltimore, United States

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing – review and editing

   For correspondence

   ugyawali@umaryland.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-5072-6780

2. David A Martin

   Department of Anatomy and Neurobiology, University of Maryland School of Medicine, Baltimore, United States

   Contribution

   Software, Formal analysis, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

3. Fangmiao Sun

   State Key Laboratory of Membrane Biology, Peking University School of Life Sciences; PKU-IDG/McGovern Institute for Brain Research; Peking-Tsinghua Center for Life Sciences, Beijing, China

   Contribution

   Resources

   Competing interests

   No competing interests declared

4. Yulong Li

   State Key Laboratory of Membrane Biology, Peking University School of Life Sciences; PKU-IDG/McGovern Institute for Brain Research; Peking-Tsinghua Center for Life Sciences, Beijing, China

   Contribution

   Resources

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-9166-9919

5. Donna Calu

   1. Program in Neuroscience, University of Maryland School of Medicine, Baltimore, United States
   2. Department of Anatomy and Neurobiology, University of Maryland School of Medicine, Baltimore, United States

   Contribution

   Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing – review and editing

   For correspondence

   dcalu@som.umaryland.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2377-9494

• 953

  Page views

• 81

  Downloads

• 2

  Citations

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