VGAT-ChR2 mice were purchased from the Jackson Laboratory and maintained on a C57Bl6/J background. VGAT-Cre mice were backcrossed to C57Bl6/J mice for at least six generations. For experiments in Fig. 1, a total of fifteen animals were trained, ten of which were later used to establish psychometric functions (four for divided attention; six for reversal learning). For Fig. 2, four VGAT-ChR2 mice were used for disruption of PFC and primary sensory cortices and three mice were used for inactivating LGN. In Fig. 3, four VGAT-Cre mice were used for electrophysiological recordings from optogenetically identified visTRN neurons, of which two were used for combined electrophysiological recordings with optogenetic PFC inactivation (Fig. 3c–e). An additional six mice were used for optogenetic activation or inhibition of visTRN (three per manipulation) during behaviour (Fig. 3f–g). Four wild-type mice were used for LGN recordings (Fig. 4). For fibre photometry experiments (Fig. 5), six mice were injected with AAV-hSyn-SuperClomeleon for behavioural and pharmacological experiments and three YFP control mice were used. Including all animals used for Extended Data figures, a total of 28 male mice, 1.5–6 months old, were trained on the cross-modal task. All experimental procedures involving animals were performed according to the guidelines of the Institutional Animal Care and Use Committee at the New York University Langone Medical Center and the US National Institutes of Health. Behavioural setup. Experiments were conducted in a custom-built trapezoidal testing chamber (base 1, 12 cm; base 2, 25 cm; height, 25 cm) positioned over a grid floor. The testing chamber contained three nose-pokes, each of which consisted of an infrared LED/infrared phototransistor pair (Digikey, Thief River Falls, Minnesota) for response detection. Activation of a central nose-poke located on the grid floor, 6 cm away from the reward wall, was required for trial initiation. Two headphone speakers (Skullcandy, Park City, Utah) embedded in the floor delivered biasing cues binaurally. Two white LEDs (Mouser, El Cajon, California) were mounted 6.5 cm apart on the base wall below two additional nose-pokes. Liquid reward consisting of 10 μl evaporated milk was delivered directly to these wall-mounted nose-pokes via a single-syringe pump (New Era Pump Systems, Farmingdale, New York). Access to these response nose-pokes was restricted by a rotating, servo-controlled (Tower Hobbies, Champaign, Illinois) disc (radius, 7 cm). Rewards could be accessed from these nose-pokes only when two holes in the rotating disc were aligned with the underlying nose-pokes. Trial logic was controlled by custom software running on an Arduino Leonardo microcontroller (Ivrea, Italy). Training. Mice were food restricted to 85–90% of their ad libitum body weight before training. Training consisted of multiple levels. First, mice were habituated to the test box and allowed to collect reward freely. Reward availability was signalled by the rotation of the aforementioned wall-mounted disc. The location of reward (left or right poke) was indicated by either a visual or an auditory stimulus. For ‘attend to vision’ (visual) trials, the rewarded response poke was indicated by illumination of the LED mounted underneath it. In ‘attend to audition’ (auditory) trials, an upsweep (10–14 kHz, 500 ms) indicated a reward on the left and a downsweep (16–12 kHz, 500 ms) indicated a reward on the right. To facilitate discrimination learning, sweeps were initially presented in a directional manner. Trials were given in single-modality blocks of six, with alternating block type (that is, six visual trials followed by six auditory trials; Extended Data Fig. 1, top row). The stimulus was presented until the animal collected the reward. An individual trial was terminated 20 s after reward collection and a new trial became available 5 s later. Second, mice learnt to poke to receive a reward. All other parameters remained constant. An incorrect poke had no negative consequence. By the end of this training stage, all mice collected at least 20 rewards per 30-minute session. Third, mice were trained to initiate individual trials, allowing for the establishment of a temporal window in which they could anticipate subsequent delivery of the stimuli. For successful initiation, mice had to break the infrared beam briefly (50 ms) in the initiation poke to trigger stimulus presentation and rotation of the wall-mounted disc. Mice were informed about trial availability and modality type by brown noise (10-kHz low-pass-filtered white noise, visual trial) or blue noise (11-kHz high-pass-filtered white noise, auditory trial) delivered binaurally. At this stage, modality types were arranged in a non-conflicting block design (Extended Data Fig. 1, top row). Correct poking resulted in reward delivery, whereas incorrect poking resulted in immediate termination of the trial by disc rotation, blocking access to reward. Rewards were available for 15 s following correct poking, followed by a 5-s intertrial interval (ITI). Incorrect poking was punished with a time-out, which consisted of a 30-s ITI. Mice could not initiate new trials during an ITI. To avoid development of side preferences, the target stimulus would appear at the same location as it did on the previous trial following an incorrect response. After one week of training on this stage, mice successfully associated the target stimuli with the appropriate reward location (Extended Data Fig. 1, top row). At this stage, directionality of sound stimuli did not affect performance. Fourth, mice had to resolve sensory conflict. Auditory and visual target stimuli were always presented in a conflicting manner (Extended Data Fig. 1, middle row). The brown and blue noise cues indicated the modality to be selected. During a session, four different trial types were presented in blocks in repeating order: (1) three auditory trials, (2) three visual trials, (3) six conflict trials with an auditory target and (4) six conflict trials with a visual target. To prevent modality preferences, an incorrect response resulted in the repetition of the same trial type, thereby specifically increasing the block length of the trial types with weak performance. This training stage was introduced to teach mice to attend only to the target modality during a conflict trial. Over the course of this training stage (1 week), the duration of the target stimuli was successively shortened to 3 s, 1 s and 0.5 s. In parallel, the time that mice had to break the infrared barrier in the initiation nose-poke was continually increased and randomized to a final range of 0.5–0.7 s, rendering the precise presentation time of target stimuli unpredictable. Once mice performed successfully on conflict trials (Extended Data Fig. 1, middle row) the single-modality trials were removed and block length was reduced to three trials. This change in the training paradigm was made to facilitate learning of the trial-type cueing (brown and blue noise). On the fifth and final stage of training, all block structure was removed and trial type was randomized (Extended Data Fig. 1, bottom row). We used three measures to ensure that mice followed the trial-type cueing and did not employ simple alternating strategies. In addition to computing overall accuracy (Extended Data Fig. 1, graphs on left), we quantified the number of consecutive correct trials (Extended Data Fig. 1, middle column) and calculated the fraction of correct modality switches (Extended Data Fig. 1, graphs on right). At this final stage, rewards were available for only 5 s. For experiments determining the visual detection psychometric function, the ratio between visual and auditory trials was adjusted from the typical 1/1 to 4/1 to facilitate the acquisition of a larger number of visual trials while maintaining the divided-attention nature of the task. In addition, visual stimulus duration was shortened to 0.1 s and the light was randomly displayed at one of five different intensities (0.15, 0.3, 0.6, 0.9, 1.2 lm). To establish the comparison between psychometric functions of visual-only and divided-attention trials (Fig. 1c), we trained mice that reached criterion (>70% accuracy) on the cross-modal task to perform a visual-only task. For one week, mice were trained on a visual-only task every other day; trials containing only visual target stimuli were cued by broadband white noise. Subsequently, visual-only trials were introduced into the cross-modal task at a 1/4 ratio and in a random interleaved manner. Mice were found to differentially anticipate visual-only and visual target with auditory conflict trials (Fig. 1c), whereas they continued to perform equally well on conflict trials with an auditory target (Extended Data Fig. 3). To separate the effect of anticipating a conflicting stimulus (top-down) from the presence or absence of a distracting stimulus itself (bottom-up), we performed two experiments. In the first experiment, mice performed the cross-modal task with 70% conflict trials and 30% in which conflict was expected but auditory distraction was removed (Fig. 1d). In the second experiment, mice that had been trained on the cross-modal task had the biasing cues replaced with broadband white noise and the modality rewarded was changed on a session-by-session basis such that mice would deduce it based only on reward history (Fig. 1e). Performance in behavioural tests was assessed based on the fraction of correct responses relative to chance level or guess rate (50%, γ). The visual detection threshold (α) and maximum performance (λ) were estimated by fitting performance across stimulation intensities with a logistic function29, 30: where x corresponds to the five stimulus levels expressed as a percentage of maximum stimulus intensity. The fraction of correct trials as a percentage of all trials was summed across sessions and the overall performance as a function of stimulus intensity was fit using maximum likelihood estimation30 implemented in the Palamedes psychophysical toolbox (http://www.palamedestoolbox.org/). Estimation of the distribution of the α parameter was made via non-parametric bootstrap analysis of curve fits (Fig. 1). To adjust for variable lapse rates (Extended Data Fig. 2), the fraction of correct trials was normalized so that the minimum and maximum performance rates corresponded to 50% and 100%, respectively31. Curve fitting and estimation of the α parameter then proceeded as described above. Model selection for the number of psychophysical parameters was based on the Akaike information criterion20, 32. For experiments with optical stimulation (Fig. 2), testing conditions were equivalent to the final stage of training. Laser trains of either blue (for ChR2 activation) or yellow (for eNpHR3.0 activation) light consisting of 50-Hz 18-ms pulses (90% duty cycle) at an intensity of 5–6 mW (measured at the tip of the optic fibres) were delivered on every other trial. On laser trials, stimulation occurred either during the anticipatory period (0.5–0.7 s) or during stimulus presentation (0.5 s). Because behaviour and recording systems were automated and stimulus sequence and optogenetic manipulations varied on a trial-by-trial basis, researchers were not blinded to the conditions. In the case of multiple sequential pharmacological or optogenetic manipulations in the same animals, tests were performed in a predefined, pseudorandom order. For comparisons of multiple groups, Kruskal–Wallis one-way analysis of variance was used to assess variance across groups before pairwise comparisons. Power analysis based on effect size estimates was used to determine sample size required for statistical significance with a power of β = 0.7; more than three samples were required to detect significant differences. For combined TRN recordings with optogenetic PFC disruption (Fig. 3d, e), laser trains of blue light (as described earlier) were delivered during the anticipatory period on every other trial. For electrophysiological recordings of LGN units and fibre photometry measurements, visual stimuli were presented through illumination of diffusion-coated wide-angle 3-mm flat-top LED lights (LightHouseLEDS, Washington), fixed directly on the head of the mouse and centred 8 mm from the eyes. The LEDs mounted on the base wall of the behavioural box were turned off in this condition. These changes allowed emitted light to activate ~150° of the visual angle when the eye is centred at rest33. For retrograde optogenetic tagging and TRN manipulation, FuGB2-pseudotyped retrograde lentiviruses (RG-LV) were used as described previously20. visTRN neurons were labelled through injection (0.4–0.6 μl) of RG-LV-EF1α-DIO-ChR2-GFP (for activation) or RG-LV-EF1α-DIO-eNpHR3.0-eYFP (for inactivation) into the primary visual thalamus (anterior–posterior (A–P), −2.1 mm; medial–lateral (M–L), ±2 mm; dorsal–ventral (D–V), 2.5 mm) using a quintessential stereotactic injector (QSI, Stoelting, Wood Dale, Illinois). Coordinates are referenced to Bregma. For combined electrophysiological recordings with optogenetic PFC disruption during behaviour, 0.4 μl AAV2-hSyn-DIO-ChR2-GFP (titre, 1012 vector core per ml) was injected into the PFC (A–P, 2.6 mm; M–L, ±0.25 mm; D–V, –1.25 mm). To measure chloride flux in the LGN, the transgene SuperClomeleon (gift from G. J. Augustine25) was cloned into the AgeI and EcoRI restriction sites of an AAV-hSyn-SSFO-eYFP plasmid to obtain AAV-hSyn-SuperClomeleon. The SuperClomeleon recombinant AAV was packaged as serotype 2 (University of North Carolina, Vector core facility; titre, 1012 vector core per ml) and 0.6–0.7 μl virus was injected into the visual thalamus. Following injections, mice were allowed to recover for 2–4 weeks to allow for virus expression. Mice were anaesthetized using 1% isoflurane and mounted on a stereotactic frame. For cortical inactivation experiments, up to three pairs of 4–5-mm-long optic fibres (Doric Lenses, Quebec, Canada) were inserted bilaterally to target up to three different brain areas per mouse (prelimbic cortex, 2.6 mm A–P, ±0.25 mm M–L, –1.25 mm D–V; primary visual cortex, –3.5 mm A–P, ±2.50 mm M–L, –0.50 mm D–V; primary auditory cortex, –2.8 mm A–P, ±4.00 mm M–L, –2.00 mm D–V; AAC, 0.5 mm A–P, ±0.25 mm M–L, –1.00 mm D–V; lateral OFC, 2.6 mm A–P, ±1.50 mm M–L, –2.00mm D–V; primary visual thalamus, –2.1 mm A–P, ±2.00 mm M–L, –2.50 mm D–V; visTRN, –1.6 mm A–P, ±2.20 mm M–L, –3.00 mm D–V). Two or three stainless steel screws were implanted into the skull to anchor the implant and were fixed with dental cement. Animals were allowed to recover and training resumed one week later. For ChR2 activation a 473-nm laser was used and a 561-nm laser was used for eNpHR3.0 activation (Omicron-Laserage, Dudenhofen, Germany). Custom drive housings were designed using 3D CAD software (SolidWorks, Concord, Massachusetts) and printed in Accura 55 plastic (American Precision Prototyping, Tulsa, Oklahoma) as described previously20, 34. Prior to implantation, each drive was loaded with 8–12 independently movable microdrives carrying up to 3 nichrome (12.5 μm) and/or tungsten (25 μm) stereotrodes (California Fine Wire Company, Grove Beach, California). Stereotrodes were pinned to custom-designed 32- or 64-channel electrode interface boards (EIB; Sunstone Circuits, Mulino, Oregon) along with a common reference wire (A-M systems, Carlsborg, Washington). For optogenetic tagging, an optical fibre was embedded adjacent to the stereotrode array. In these cases, the optic fibre extended 3.5 mm from the base of the drive so that it could be stereotactically positioned above the TRN during implantation. Targeting of the TRN or LGN was achieved by guiding stereotrodes and optic fibres through a square array of polyimide sleeves attached to the base of the drive body. Prior to surgical implantation, mice were anaesthetized with 1% isoflurane and placed in a stereotactic frame. Stainless steel screws were implanted into the skull to provide electrical and mechanical stability for the drives. For drive implantations, craniotomies (~3 mm × 2 mm) were drilled, centred at –2 mm A–P and 2.5 mm M–L for TRN recordings (15° angled implantation) and at –2.3 mm A–P and –2.5 mm M–L for LGN recordings. The dura mater was carefully removed and drives were centred at the craniotomy coordinates using a custom stereotactic arm. Drive bodies were slowly lowered into the craniotomy until stereotrode tips were ~500 μm below the cortical surface and optical fibres were positioned just above the TRN (2.5 mm D–V). For fibre-photometry-based optical recording, low-internal-fluorescence optic fibres (400 mm diameter) (Doric Lenses, Canada) were implanted just dorsal to the LGN (–2.2 mm A–P, 2.15 mm M–L and 2.6 mm D–V) following virus injection. After mice had recovered from implantation surgery, recordings were made using a Neuralynx multiplexing digital recording system (Neuralynx, Bozeman, Montana). Signals were acquired using a 32- or 64-channel digital headstage connected to the implanted EIB. Signals from each electrode were amplified, filtered between 0.1 Hz and 9 kHz and digitized at 30 kHz. Local field potential signals were obtained from a single wire per stereotrode. Following implantation, stereotrode sets were incrementally lowered from the cortex into the target thalamic structure over the course of 1–2 weeks (Extended Data Fig. 5). Spike sorting was performed offline following acquisition based on relative spike amplitude and energy within electrode pairs using the MClust toolbox (http://redishlab.neuroscience.umn.edu/mclust/MClust.html). Following manual clustering, cross-correlation and autocorrelation analyses were used to confirm adequate separation. Optogenetically tagged visTRN units were identified based on ChR2-mediated response to stimulation using a 473-nm analogue-modulated laser (Omicron-Laserage, Dudenhofen, Germany)20. Laser light was delivered by a 200-μm optic fibre targeted to the TRN (Extended Data Fig. 5) connected to a fibre optic patch cord (200-μm core, Doric Lenses, Quebec, Canada). The laser intensity was set at ~8 mW optical output power measured at the patch cord terminus. Fibres were polished before implantation so that the power at the tip was ≥50% maximum, resulting in ~4–5-mW laser light being delivered to the brain. Only neurons that showed clear transient responses to laser stimulation were included in the analysis. Changes in firing rate during task performance were assessed for 138 identified visTRN neurons recorded from four animals and 119 LGN neurons in two animals. Peri-event time histograms aligned to trial initiation and to stimulus presentation were computed using a 5-ms bin width for individual neurons in each recording session4. Separate histograms were created for correct and incorrect trials within auditory and visual target stimuli and convolved with a Gaussian kernel (8 ms half-width at half-height) to create a spike density function35, 36. The average firing rate across trials was determined during the anticipation window before stimulus presentation. The evoked response amplitude was estimated by averaging the firing rate within a 100-ms window starting 20 ms after stimulus onset. Window duration was chosen based on the latency-to-peak response for point stimuli in the mouse LGN37. For normalized rate changes in TRN neurons, firing rates during the attentional window in each trial were compared with the baseline firing rate (5-s window, 0.5 s before task initiation). Statistical comparison of firing rate changes was used to identify neurons with significant task-associated changes in firing rate via non-parametric comparison of firing rate during the attentional window and the baseline period38. The test statistic (W) was calculated based on ranking of all trials (N) and comparison was performed using the sign function (sgn): where x and x were the attentional window and baseline firing rate, respectively, and R denotes the rank. The threshold for significance was set at 0.05 and significantly modulated units were defined as neurons in which the test statistic was less than the critical value for the sample size (W ). Comparison of firing rates across trial types (for example, visual versus auditory correct) was performed using the Wilcoxon rank-sum test. Homogeneity of variance for firing rates across conditions was determined using the Fligner–Killeen test of homoscedasticity. Visual evoked potentials (VEPs) were computed from the broadband LGN local field potential (LFP; 0.1 Hz–10 kHz). The particular stereotrode used for VEP analysis in behaviour was selected based on the amplitude of responses in post-task recordings during which there were many more trials included. Task-related VEPs were averaged during correct auditory and visual trials across recording sessions. To determine peak response, the lowest negative-potential offsets associated with the visual response39 (0–250-ms window) were identified on a trial-by-trial basis. Signals from individual trials were smoothed with a 25-ms half-width filter over the response window before obtaining the peak offset40. Mice were euthanized and transcardially perfused with PBS followed by 4% paraformaldehyde. Brains were dissected, post-fixed overnight at 4 °C and sectioned using a vibratome (LEICA, Buffalo Grove, Illinois). For GFP enhancement, immunofluorescent staining was carried out on 50-µm-thick sections using chicken anti-GFP (1/1,000, GFP-1020, Aves). Sections were incubated overnight with primary antibody in PBS-T (10% normal goat serum and 0.05% Tween20) at 4 °C. Detection of primary antibodies was carried out with Alexa-Fluor-conjugated secondary antibodies (1/1,000, A-11039, Invitrogen). All sections were imaged on a Zeiss LSM510 META confocal microscope. FRET-based measurement of chloride was performed during behaviour using a custom-designed fibre photometry system24. A fibre-coupled LED (Thorlabs, Newton, New Jersey) light source, filtered using a 434-nm clean-up filter (MF434-17 Thorlabs, Newton, New Jersey), was used for CFP excitation. Excitation light was split via a long-pass dichroic mirror (DMLP425, Thorlabs, Newton, New Jersey) and coupled to a 400-μm, 0.48-NA (pharmacology) optic patch cord (Doric lenses, Canada) linked to a 400-μm chronically implanted optical fibre. Excitation and emission light were conveyed by a single patch cord linking the fibre photometry system to the implanted fibre. SuperClomeleon CFP and YFP emissions25 were separated using a single-edge beam splitter (FF511-Di01, Semrock, Rochester, New York). Each emission wavelength was independently focused onto a separate femtowatt silicon photoreceiver (Newport, Irvine, California) using custom optics (12.7-mm focal length plano-convex lens mounted in Thorlabs SM1NR05 lens tube). The light signal was digitized and recorded using a TDT signal acquisition system (Tucker-Davis Technologies, Alachua, Florida). Signal bandwidth was limited to <750 Hz based on the photoreceiver response characteristics. The fluorescence ratio was calculated across the recording period. To minimize the effect of slow fluctuations, normalized delta fluorescence (df/F) was calculated for evoked responses relative to the baseline fluorescence level before each event (1-s window). Traces were smoothed with a convolution filter (50 ms half-width). Peak response (Extended Data Fig. 10) was estimated as the minimum over a 500-ms window following stimulus onset. For pharmacological activation of GABA receptors with 4,5,6,7-tetrahydroisoxazolo(5,4-c)pyridin-3-ol (THIP), baseline fluorescence was estimated over 5 minutes before injection. For visual stimulation, light pulses of 100-ms duration were displayed to the ipsi- or contralateral side of the recorded LGN. Effects of the GABA receptor antagonist flumazenil on visual evoked responses were quantified by comparing the average peak response from 5 minutes before injection (baseline) to one within a 5-minute time window around the maximal response suppression (maximal drug effect) and at the end of the recording session (recovery, at least 100 min after injection). For optogenetic manipulations of frontal cortical structures, smaller-diameter patch cords (200 µm, 0.37 NA) were used to allow movement and prevent tangling. For these recordings, power analysis was performed to determine sample size required to detect significant differences with a power of β = 0.7 based on the observed differential signal in correct auditory and visual trials under baseline conditions. Analysis indicated that more than four independent samples would be required to detect a change in these differential responses.
News Article | November 23, 2015
Despite the lack of evidence, the Obama Administration has revived the encryption debate, pointing to encryption as an aid to the terrorists behind the Nov. 13 Paris attacks. Investigators from France and the U.S. have conceded that there has been no evidence backing up their conclusion that the terrorist behind the attacks relied on the latest, high-level encryption techniques being offered to consumers by Google and Apple. Yet, the debate over government-grieving encryption is back in high gear. The Great Encryption debate kicked into full swing about a year ago, when current and former chiefs of the U.S. Department of Justice began calling on Apple and Google to create backdoors in iOS 8 and Android Lollipop. The encryption built for the two mobile operating systems is so tough, that the world's best forensic scientists in all of computing wouldn't be able to crack devices running the software in time for a seven-year statute of limitations. While it's possible to crack the encryption in less time, each misstep would push back the subsequent cool-down period before the software would allow for another go. A few weeks before the Nov. 13 attacks on Paris, the DOJ employed a new strategy to coerce Apple into handing over the keys to iOS – and it's a good one. The tech world is still awaiting Apple's counterpunch. Roughly a year ago, then U.S. Attorney General Eric Holder frame the debate on encryption and stated the DOJ's stance while speaking at the Global Alliance Against Child Sexual Abuse Online. "Recent technological advances have the potential to greatly embolden online criminals, providing new methods for abusers to avoid detection," Holder said, adding that there are those who take advantage of encryption in order to hide their identities and "conceal contraband materials and disguise their locations." The Information Technology Industry Council, which speaks on behalf of the high-tech industry, sees all of the above issues as reasons everyone needs encryption. "Encryption is a security tool we rely on everyday to stop criminals from draining our bank accounts, to shield our cars and airplanes from being taken over by malicious hacks, and to otherwise preserve our security and safety," said Dean Garfield, president and CEO of ITI. While stating the ITI's deep "appreciation" for the work done by law enforcement and the national security community, Garfield said there is no sense in weakening the security just to improve it. "[W]eakening encryption or creating backdoors to encrypted devices and data for use by the good guys would actually create vulnerabilities to be exploited by the bad guys, which would almost certainly cause serious physical and financial harm across our society and our economy," he explained. In the wake of the recent Paris Attack, U.S. officials have again reissued their call for software developers – Apple, Google and others – to provide law enforcement agencies with keys to the backdoor of operating systems with government-grade encryption. While there is still no evidence that law enforcement agencies, with encryption keys in hand, could have given police on the ground in Paris a game-changing heads up of the attacks. Nevertheless, Paris has been turned into a talking point said Michael Morell, a former deputy director of the CIA, who stated that the tragic events will reshape the encryption debate. "We have, in a sense, had a public debate [on encryption]," said Morell. "That debate was defined by Edward Snowden." Although, instead of what the former NSA contractor and leaker had done, the issue of encryption will now be "defined by what happened in Paris."
Based on power analyses that assumed a normal distribution, a 20% change in mean and 15% variation, we determined that at least 9 mice per group would be needed for behavioral experiments. This was adhered to as far as possible, except in cases where mice had to be removed owing to misplaced injections or lost headcaps. Mice were randomly assigned to groups and attempts were made to balance groups according to variables such as age and housing condition. The investigators were not blinded to allocation during experiments, but were blinded to outcome assessment for all behavioral experiments. Mice were used in all experiments. For experiments involving Cre lines, mice were crossed for several generations to C57 mice before using. All wild-type mice were C57BL/6 mice obtained from The Jackson Laboratory (Bar Harbour, ME). For all behavioural experiments except those involving Htr2cCre mice, male mice ranging in age from 8–16 weeks were used. Female Htr2cCre mice were used in chemogenetic manipulations. Both male and female mice aged 6–20 weeks were used for slice electrophysiology and anatomical tracing experiments. All behavioural studies or tissue collection for ex vivo slice electrophysiology were performed during the light cycle. All behavioural experiments in Htr2ccre mice were conducted at the University of Aberdeen and in accordance with the United Kingdom Animals (Scientific Procedures) Act of 1986. All in vivo electrophysiology experiments were conducted in accordance with all rules and regulations at the National Institute for Alcohol Abuse and Alcoholism at the National Institutes of Health. All other procedures were conducted in accordance with the National Institutes of Health guidelines for animal research and with the approval of the Institutional Animal Care and Use Committee at the University of North Carolina at Chapel Hill. All animals were group housed on a 12 h light cycle (lights on at 7 a.m.) with ad libitum access to rodent chow and water, unless described otherwise. CRF-ires-Cre (Crfcre) were provided by Bradford Lowell (Harvard University) and were previously described21. C57BL/6J mice were obtained from the Jackson Laboratory (Bar Harbour, ME). To visualize CRF-expressing neurons, CrfCre mice were crossed with either an Ai9 or a Cre-inducible L10-GFP reporter line (Jackson Laboratory)22 to produce CRF-Ai9 or CRF-L10GFP progeny, referred to throughout the manuscript as CRF-reporters. SertCre mice (from GENSAT) were a generous gift from Bryan Roth. Htr2cCre mice were supplied by Lora Heisler and are described in detail elsewhere7. Male mice were used for in vivo optogenetic behavioural experiments and for assessing the involvement of BNST CRF neurons on fluoxetine-induced enhancement of fear. Female 5-HT -Cre mice were used in chemogenetic manipulations. Both male and female mice were used for slice electrophysiology and anatomical tracing experiments. All behavioural studies or tissue collection for ex vivo slice electrophysiology were performed during the light cycle. All AAV viruses except INTRSECT constructs were produced by the Gene Therapy Center Vector Core at the University of North Carolina at Chapel Hill and had titres of >1012 genome copies per ml. For ex vivo and in vivo optical experiments, mice were injected with rAAV5-ef1α-DIO-hChR2(H134R)-eYFP or rAAV5-ef1α-DIO-eYFP as a control. Red IX retrobeads (Lumafluor) were used to fluorescently label LH- and VTA-projecting BNST neurons during ex vivo slice electrophysiology recordings. The retrograde tracer Fluoro-Gold (Fluorochrome) was used for anatomical mapping. Choleratoxin B (CTB) 555 and CTB 657 retrograde tracers (Invitrogen; C34776, and C34778, respectively) diluted to 0.5% (w/v) in sterile PBS were used per injection site for anatomical mapping of collateral projections from BNST to LH and VTA. For chemogenetic manipulations, mice were injected with 400 nl of rAAV8-hsyn-DIO-hM3D(Gq)-mCherry, rAAV8-hsyn-DIO-hM4D(Gi)-mCherry, or rAAV8-hsyn-DIO-mCherry bilaterally. HSV-hEF1α-mCherry, HSV-ef1α-LSL1-mCherry-IRES-flpo, and HSV-ef1α-IRES-Cre (supplied by Rachel Neve at the McGovern Institute for Brain Research at MIT) were injected bilaterally into the VTA and LH at a volume of 500 nl per site. The INTRSECT construct AAVdj-hSyn-Con/Foff-hChR2(H134R)-EYFP was infused at 500 nl per side into the BNST. All AAV constructs had viral titres >1012 genome particles per ml. All surgeries were conducted using aseptic technique. Adult mice (2–5 months) were deeply anaesthetized with 5% isoflurane (v/v) in oxygen and placed into a stereotactic frame (Kopf Instruments) while on a heated pad. Sedation was maintained at 1.5–2.5% isoflurane during surgery. An incision was made down the midline of the scalp and a craniotomy was performed above the target regions and viruses and fluorescent tracers were microinjected using a Neuros Hamilton syringe at a rate of 100 nl min−1. After infusion, the needle was left in place for 10 min to allow for diffusion of the virus before the needle was slowly withdrawn. Injection coordinates (in mm, midline, Bregma, dorsal surface): BNST (±1.00, 0.30, −4.35), LH (±0.9 to 1.10, −1.7, −5.00 to −5.2), VTA (−0.3, −2.9, −4.6), DR (0.0, −4.65, −3.2 with a 23° angle of approach). When using retrobeads, injection volumes into the LH and VTA were 300 nl and 400 nl, respectively. Fluoro-Gold injection volumes were 200 nl per target site. CTB volumes were 200 nl per target site. An optical fibre was implanted in the BNST (±1.00, 0.20, −4.15) at a 10° angle for in vivo photostimulation studies. After fibre implantation, dental cement was used to adhere the ferrule to the skull. Following surgery, all mice returned to group housing. Mice were allowed to recover for at least 3 weeks before being used for chemogenetic behavioural studies, or 6 weeks for in vivo optogenetic studies. RS-102221, 5-HT and mCPP were from Tocris (Bristol, UK). For electrophysiology experiments, RS-102221 was made up to 100 mM in DMSO and then diluted to a final concentration of 5 μM in aCSF. 5-HT and mCPP were stocked at 10 and 20 mM, respectively, in ddH O and diluted to their final concentations in aCSF. For electrophysiology experiments, clozapine-N-oxide (CNO; from Bryan Roth) was stocked at 100 mM in DMSO and diluted to 10 μM in aCSF. For behaviour experiments, CNO was dissolved in 0.5% DMSO (in 0.9% saline) to a concentration of 0.1 mg ml−1 or 0.3 mg ml−1 and injected at 10 ml per kg for a final concentration of 1 or 3 mg per kg, i.p. Fluoxetine (Sigma) was made up in 0.9% NaCl to a concentration of 1 mg ml−1 and then injected at 10 ml per kg for a final concentration of 10 mg per kg, i.p. Surgical procedures. Mice were anaesthetized with 2% isoflurane (Baxter Healthcare, Deerfield, IL) and implanted with 2 × 8 electrode (35 μm tungsten) micro-arrays (Innovative Neurophysiology, Durham, NC) targeted at the BNST (ML: 0.8 mm, AP: ± 0.5 mm, and DV: −4.15 mm relative to Bregma). Following surgery, mice were singly housed and allowed at least one week to recover before behavioural testing. Fear conditioning. Fear conditioning took place in 27 × 27 × 11 cm conditioning chambers (Med Associates, St. Albans, VT), with a metal-rod floor (context A) and scented with 1% vanilla. Mice received 5 parings of a pure tone CS with a 0.6 mA foot shock. 24 h following conditioning, mice underwent a CS recall test (10 presentations of the CS alone, 5 s ITI), which was conducted in a Plexiglas cylinder (20 cm diameter) and scented with 1% acetic acid (context B). Stimulus presentations for both tests were controlled by MedPC (Med Associates, St. Albans, VT). Cameras were mounted overhead for recording freezing behaviour, which was scored automatically using CinePlex Behavioural Research System software (Plexon, Dallas, TX). Electrophysiological recording and single unit analysis. Electrophysiological recording took place during both fear conditioning and CS recall tests. Individual units were identified and recorded using Omniplex Neural Data Acquisition System (Plexon, Dallas, TX). Neural data was sorted using Offline Sorter (Plexon, Dallas, TX). Waveforms were isolated manually, using principal component analysis. To be included in the analyses, spikes had to exhibit a refractory period of at least 1 ms. Autocorrelograms from simultaneously recorded units were examined to ensure that no cell was counted twice. Single units were analysed by generating perievent histograms (3 s bins) of firing rates from 30 s before CS onset until 30 s after CS offset (NeuroExplorer 5.0, Nex Technologies, Madison, AL). Firing rates were normalized to baseline (30 s before CS onset) using z-score transformation. Analysis included a total of 139 cells over three days of recording. Data reported for raw firing rates include only putative principal neurons (<10 Hz). The formula for computing the suppression ratio was (average freezing rate) / (average freezing rate + average movement rate). Each cell was calculated individually. A value of 0.5 = no change in rate). Ex vivo slice electrophysiology. Brains were sectioned at 0.07 (mm per s) on a Leica 1200S vibratome to obtain 300 μm coronal slices of the BNST, which were incubated in a heated holding chamber containing normal, oxygenated aCSF (in mM:124 NaCl, 4.4 KCl, 2 CaCl , 1.2 MgSO , 1 NaH PO , 10.0 glucose, and 26.0 NaHCO ) maintained at 30 ± 1 °C for at least 1 h before recording. Slices were transferred to a recording chamber (Warner Instruments) submerged in normal, oxygenated aCSF maintained at 28–30 °C at a flow rate of 2 ml min−1. Neurons of the BNST were visualized using infrared differential interference contrast (DIC) video-enhanced microscopy (Olympus). Borosilicate electrodes were pulled with a Flaming-Brown micropipette puller (Sutter Instruments) and had a pipette resistance between 3–6 MΩ. Signals were acquired via a Multiclamp 700B amplifier, digitized at 10 kHz and analysed with Clampfit 10.3 software (Molecular Devices, Sunnyvale, CA, USA). In SertCre or CrfCre mice, fluorescently labelled neurons expressing ChR2 were visualized and stimulated with a blue (470 nm) LED using a 1 Hz, 2 Hz, 5 Hz, 10 Hz, and 20 Hz stimulation protocol with a pulse width of 0.5 ms. Evoked action potentials were recorded in current clamp mode using a potassium gluconate based internal solution (in mM: 135 K+ gluconate, 5 NaCl, 2 MgCl , 10 HEPES, 0.6 EGTA, 4 Na ATP, 0.4 Na GTP, pH 7.3, 285–290 mOsmol). In CrfCre mice with ChR2 in the BNST and retrograde tracer beads in the VTA or LH, we visualized non-ChR2-expressing, beaded neurons using green (532 nm) LED. Recordings were conducted in voltage clamp mode using a caesium-methansulfonate (Cs-Meth) based internal solution (in mM: 135 caesium methanesulfonate, 10 KCl, 1 MgCl , 0.2 EGTA, 2 QX-314, 4 MgATP, 0.3 GTP, 20 phosphocreatine, pH 7.3, 285–290 mOsmol) so that we could detect EPSCs (−55 mV) and IPSCs (+10 mV) in the same neuron. After confirming the absence of a light-evoked EPSC signal, we measured light-evoked IPSCs during a single 5-ms light pulse of 470 nm. In a subset of these experiments, SR95531 (GABAzine, 10 μM) was bath applied for 10 min to block IPSCs. Crf-reporter mice were injected with retrograde tracer beads into the VTA (ML −0.5, AP −2.9, DV −4.6). We then recorded from beaded (VTA-projecting) and non-beaded (non-projecting) CRF neurons in the BNST. Acute drug effects were determined in current clamp mode in the presence of TTX using a potassium gluconate-based internal solution. After a 5-min stable baseline was established, 5-HT (10 μM) or mCPP (20 μM) was bath applied for 10 min while recording changes in membrane potential. The difference in membrane potential between baseline and drug application at peak effect (delta or Δ MP) was later determined. In a subset of mCPP experiments, slices were incubated with RS-102221 (5 μM) for at least 20 min before experiments began. Spontaneous inhibitory postsynaptic currents (sIPSCs) were assessed in voltage clamp using a potassium-chloride gluconate-based intracellular solution (in mM: 70 KCl, 65 K+-gluconate, 5 NaCl, 10 HEPES, 0.5 EGTA, 4 ATP, 0.4 GTP, pH 7.2, 285–290 mOsmol). IPSCs were pharmacologically isolated by adding kynurenic acid (3 mM) to the aCSF to block AMPA and NMDA receptor-dependent postsynaptic currents. The amplitude and frequency of sIPSCs were determined from 2 min recording episodes at −70 mV. The baseline was averaged from the 4 min preceding the application of 5-HT (10 μM) or mCPP (10 μM) for 10 min. In a subset of these experiments, RS-102221 (5 μM) was added to the aCSF and slices were incubated in this drug solution for at least 20 min before experiments began. For miniature IPSCs (mIPSCs), TTX was included in the aCSF to block network activity. In SertCre::ChR2BNST mice with retrograde tracer beads in the VTA, sIPSCs were recorded as described above. After achieving a stable baseline, a 10 s, 20 Hz photostimulation was applied. For assessment of spontaneous excitatory postsynaptic currents (sEPSCs), a caesium gluconate-based intracellular solution was used (in mM: 135 Cs+-gluconate, 5 NaCl, 10 HEPES, 0.6 EGTA, 4 ATP, 0.4 GTP, pH 7.2, 290–295 mOsmol). AMPA -mediated EPSCs were pharmacologically isolated by adding 25 μM picrotoxin to the aCSF. sEPSC recordings were acquired in 2 min recording blocks at −70 mV. Electrodes were fabricated as previously described and cut to 50–100 μm in length23. Animal and slice preparation were as described above for electrophysiology and slices were perfused on the rig in ACSF. Using a custom-built potentiostat (University of Washington Seattle), 5-HT recordings were made in the BNST using TarHeel CV written in laboratory view (National Instruments). Briefly a triangular waveform (−0.1 V to 1.3 V with a 10% phase shift at 1,000 V per s, versus Ag/AgCl) was applied to the carbon fibre electrode at a rate of 10 Hz. Slices were optically stimulated with 20 5-ms blue (490 nm) light pulses at a rate of 20 Hz down the submerged 40× objective. 10 cyclic voltammograms were averaged before optical stimulation for background subtraction. Voltammograms were digitally smoothed one time with a fast Fourier transform following data collection and analysed with HDCV (UNC Chapel Hill). Fluoxetine (10 μM) was bath applied following a stable baseline (20 min). For chemogenetic manipulations, mice were transported to a holding cabinet adjacent to the behavioural testing room to habituate for at least 30 min before being pretreated with CNO (3 mg per kg, i.p. for CrfCre mice and 1 mg per kg, i.p. for Htr Cre mice). All behavioural testing began 45 min following CNO treatment, with the exception of fear conditioning training, which occurred 30 min after a CNO injection. When assessing the effect of fluoxetine on fear conditioning, fluoxetine (10 mg per kg, i.p.), or vehicle, was administered 1 h before training (30 min before CNO treatment). For optogenetic manipulations, mice received bilateral stimulation (473 nm, ~10 mW, 5 ms pulses, 20 Hz) when specified. Unless specified, all equipment was cleaned with a damp cloth between mouse trials. All sessions were video recorded and analysed using EthoVision software (Noldus Information Technologies) except where noted. Mice were placed in the centre of an elevated plus maze and allowed to explore during a 5 min session. Light levels in the open arms were ~14 lux. During optogenetic manipulations mice received bilateral stimulation during the entire 5 min session. Mice that left the maze were excluded from analysis (n = 2 control, 1 ChR2 from optogenetic experiments). Mice were placed into the corner of a white Plexiglas open field arena (25 × 25 × 25 cm) and allowed to freely explore for 30 min. The centre of the open field was defined as the central 25% of the arena. For optogenetic studies the 30 min session was divided into three 10-min epochs consisting of stimulation off, stimulation on, and stimulation off periods. 48 h before testing, mice were provided with access to a single piece of Froot Loops cereal (Kellogg’s) in their home cage. 24 h before testing, home cage chow was removed and mouse body weights were recorded. Water remained available ad libitum. Beginning at least one hour before testing, mice transferred to new clean cages so they were singly housed for the test session and body weights were recorded. During the test session mice were placed into an arena (25 × 25 × 25 cm) that contained a single Froot Loop on top of a piece of circular filter paper. Mice were monitored by a live observer and the latency for the mouse to begin eating the pellet was measured, allowing up to 10 min. All mice began eating within this time. Following the initiation of feeding, mice were removed from the arena and placed back into their home cages. Mice were then provided with 10 min of access to a pre-weighed amount of Froot Loops for a post-test feeding session. After this 10 min post-test, the remaining Froot Loops were weighed and mice were returned to ad libitum home cage chow. Mice were returned to group housing at the end of this session. For optogenetic experiments, mice received constant 20 Hz optical stimulation during both the latency to feed assay and the 10 min post-test. During optogenetic experiments, one control mouse did not feed during the 10 min NSF session and was excluded from the results. SertCre mice were food deprived for 24 h. On the day of the experiment, mice were acclimated to the behaviour room for 1 h. A single pre-weighed food pellet was placed in the home cage and the mice were allowed to eat for 10 min during optogenetic stimulation. At the end of the experimental session, the pellet was removed and weighed and mice were given ad libitum access to food. Htr Cre mice were acclimated in metabolic chambers (TSE Systems, Germany) for 2 days before the start of the recordings. After acclimation, mice were food deprived for 24 h. Following fasting, mice received an i.p. injection of CNO 30 min before food presented again. Mice were recorded for 12 h with the following measurements being taken every 30 min: water intake, food intake, ambulatory activity (in x and z axes), and gas exchange (O and CO ) (using the TSE LabMaster system, Germany). Energy expenditure was calculated according to the manufacturer’s guidelines (PhenoMaster Software, TSE Systems). We used a three-day protocol to assess both cued and contextual fear recall. On the first day, mice were placed into a fear conditioning chamber (Med Associates) that contained a grid floor and was cleaned with a scented paper towel (19.5% ethanol, 79.5% H O, 1% vanilla). After a 3 min baseline period, mice were exposed to a 30 s tone (3 kHz, 80 dB) that co-terminated with a 2 s scrambled foot shock (0.6 mA). A total of 5 tone-shock pairings were delivered with a random inter-tone interval (ITI) of 60–120 s. For optogenetic studies, light stimulation occurred only during the 30-s tones of this session. Following delivery of the last foot shock, mice remained in the conditioning chamber for a 2-min consolidation period. 24 h later, mice were placed into a separate conditioning box (Med Associates) that contained a white Plexiglas floor, a striped pattern on the walls, and was cleaned and scented with a 70% ethanol solution. After a 3 min baseline period, mice were presented with 10 tones (30 s, 3 kHz, 80 dB) with a 5 s ITI. Mice remained in the chamber after the last tone for a two-minute consolidation period. 24 h later (48 h after training), mice were returned to the original training chamber for 5 min. For each session, freezing behaviour was hand-scored every 5 s by a trained observer blinded to experimental treatment as described previously24. Freezing was defined as a lack of movement except as required for respiration. All mice used for behavioural and anatomical tracing experiments were anesthetized with Avertin and transcardially perfused with 30 ml of ice-cold 0.01 M PBS followed by 30 ml of ice-cold 4% paraformaldehyde (PFA) in PBS. Brains were extracted and stored in 4% PFA for 24 h at 4 °C before being rinsed twice with PBS and stored in 30% sucrose and PBS until the brains sank. 45 μm slices were obtained on a Leica VT100S and stored in 50/50 PBS/Glycerol at −20 °C. DREADD or ChR2-containing sections were mounted on slides, allowed to dry, coverslipped with VectaShield (Vector Labs, Burlingame, CA), and stored in the dark at 4 °C. We stained free-floating dorsal raphe sections using indirect immunofluorescence sequentially for first tryptophan hydroxylase (TPH) and Fluoro-Gold (FG) and then c-fos. For TPH/FG, we washed sections 3× for 5 min with 0.01 M PBS, permeabilized them for 30 min in 0.5% Triton/0.01 M PBS, and washed the sections again 2× with 0.01 M PBS. We blocked the sections for 1 h in 0.1% Triton/0.01 M PBS containing 10% (v/v) normal donkey serum and 1% (w/v) bovine serum albumin (BSA). We then added primary antibodies (1:500 mouse anti-TPH (Sigma Aldrich T0678) and 1:3,000 guinea-pig anti-Fluoro-Gold (Protos Biotech NM101)) to blocking buffer and incubated the sections overnight at 4 °C. The next day, we washed the sections 3× for 5 min with 0.01 M PBS, then incubated them with 1:500 with Alexa Fluor 647-conjugated donkey anti-mouse and Alexa Fluor 488-conjugated donkey anti-guinea pig secondary antibodies for 2 h at room temperature, and washed the sections 4× for 5 min with 0.01 M PBS. We then proceeded directly to the c-fos tyramide signal amplification based immunofluorescent staining. We permeabilized the sections in 50% methanol for 30 min, then quenched endogenous peroxidase activity in 3% hydrogen peroxide for 5 min. Followed by two 10 min washes in 0.01 M PBS, we blocked the sections in PBS containing 0.3% Triton X-100 and 1.0% BSA for 1 h. c-fos primary antibody (Santa Cruz Biotechnology, -sc-52) was added to sections at 1:3,000 and sections were incubated for 48 h at 4 °C. On day 3, we washed the sections in TNT buffer (0.1 M Tris-HCl pH 7.5, 0.15 M NaCl, 0.05% Tween-20) for 10 min, blocked in TNB buffer (0.1 M Tris-HCl pH 7.5, 0.15 M NaCl, 0.5% blocking reagent – PerkinElmer FP1020) buffer for 30 min. We then incubated the sections in secondary antibody (goat anti-rabbit HRP-conjugated PerkinElmer) 1:200 in TNB buffer for 30 min, washed the sections in TNT buffer 4× for 5 min, and then incubated the sections in Cy3 dye diluted in TSA amplification diluents for 10 min. We washed the sections 2× in TNT buffer, mounted them on microscope slides. We coverslipped the slides using Vectashield mounting medium. We acquired 4–5 of 2 × 4 tiled z-stack (5 optical slices comprising 7 μm total) images of the dorsal raphe from each naive and shock mouse on a Zeiss 800 upright confocal microscope. Scanning parameters and laser power were matched between groups. Images were preprocessed using stitching and maximum intensity projection and then analysed using an advanced processing module in Zeiss Zen Blue that allows nested analysis of multiple segmented fluorescent channels within parent classes. Double-labelled and triple-labelled cells were validated in a semi-automated fashion. At least 4 sections per mouse were counted in this way. One mouse was identified as a significant outlier in the shock group and was excluded from further analysis. To verify expression of ChR2-expressing fibres in the BNST originating from DRN serotonergic neurons, 300 μm slices used for ex vivo electrophysiological recordings containing the DRN and BNST were stored in 4% paraformaldehyde at 4 °C for 24 h before being rinsed with PBS, mounted, and coverslipped with Vectashield mounting medium. Images showing eYFP fluorescence from the DRN and BNST were obtained on a Zeiss 800 upright confocal microscope using a 10× objective and tiled z stacks. To validate the INTRSECT construct, mice received injections of HSV-hEF1α-mCherry or HSV-ef1α-LSL1-mCherry-IRES-flpo to both the LH and VTA bilaterally (n = 4 and 5, respectively). Both groups received AAVDJ-hSyn-Cre-on/Flp-off-hChR2(H134R)-EYFP to the BNST bilaterally. Six weeks following injection, mice were perfused and tissue was collected as described above. To visualize YFP expression in the BNST of CrfCre::IntrsectBNST mice, free-floating slices containing the BNST were rinsed three times with PBS for 5 min each. Slices were then incubated in 50% methanol for 30 min then incubated in 3% hydrogen peroxide for 5 min. Following three 10-min washes in PBS, slices were incubated in 0.5% Triton X-100 for 30 min followed by a 10 min PBS wash. Slices were blocked in 10% normal donkey serum/0.1% Triton X-100 for 1 h, and then they were incubated overnight at 4°C with a primary chicken anti-GFP antibody (GFP-1020, Aves) at 1:500 in blocking solution. Following primary incubation, slices were rinsed three times with 0.01M PBS for 10 min each and incubated with a fluorescent secondary antibody (AlexaFluor 488 donkey anti-chicken) at 1:200 in PBS for 2 h at room temperature. Slices were then rinsed with four 10-min PBS washes before being mounted onto glass slides and coverslipped with Vectashield with DAPI. A 3 × 4 tiled z stack (7 optical sections comprising 35 μm total) image from both the left and right hemispheres of the BNST was obtained at 20× magnification using a Zeiss 800 upright confocal microscope. Scanning parameters and laser power were matched between groups. Images were preprocessed using stitching and maximum-intensity projection. The number of fluorescent cells in the dorsal and ventral aspects of the BNST were counted by a blinded scorer using the cell counter plug-in in FIJI (ImageJ). Each hemisphere was considered independently per mouse. One mouse in the flp-expressing group was a significant outlier for number of cells expressed in a ventral BNST hemisphere (ROUT, Q = 0.1%) and all data from that mouse were excluded. 3 male CRF-L10a reporter mice were injected with 200 nl of CTB 555 and CTB 647 bilaterally to the LH and VTA, respectively, as described above. 5 days following injection, mice were perfused as described above, the brains were extracted, and were stored in 4% paraformaldehyde for 24 h at 4 °C before being rinsed with PBS and transferred to 30% sucrose until the brains sank. 45 μm sections containing the BNST were collected as described above. Sections containing the BNST were mounted on glass slides and coverslipped using Vectashield. An image from the left and right hemispheres of a medial section of the BNST was obtained on a Zeiss 800 upright microscope using a 20×objective and 3 × 5 tiled z stacks (5 optical slices comprising 7 μm total). Images were preprocessed using stitching and maximum intensity projection, and were then analysed using the cell counter function in FIJI (ImageJ). Only cells positive for GFP (putative CRF neurons) were considered. Cells were scored exclusively as either 555+ only (LH-projecting), 647+ only (VTA-projecting), 555+ and 647+ (projecting to both LH and VTA), or 555− and 647− (unlabelled; neither LH- nor VTA- projecting). The total number of CRF neurons scored was calculated as the sum of all four groups, and percentages of each type were calculated from this value. Each hemisphere was scored and plotted independently (n = 6 images from 3 mice), and the dorsal and ventral BNST were considered separately. The average values were plotted as pie charts (Extended Data Fig. 5). For validation of 2C-cre line and comparison of CRF/2C mRNA cellular co-localization, mice were anesthetized using isoflurane, rapidly decapitated, and brains rapidly extracted. Immediately after removal, the brains were placed on a square of aluminium foil on dry ice to freeze. Brains were then placed in a −80°C freezer for no more than 1 week before slicing. 12 μm slices were made of the BNST on a Leica CM3050S cryostat (Germany) and placed directly on coverslips. FISH was performed using the Affymetrix ViewRNA 2-Plex Tissue Assay Kit with custom probes for CRF, 5-HT , and Cre designed by Affymetrix (Santa Clara, CA). Slides were coverslipped with SouthernBiotech DAPI Fluoromount-G. (Birmingham, AL). 3 × 5 tiled z stack (15 optical sections comprising 14 μm total) images of the entire 12 μm slice were obtained on a Zeiss 780 confocal microscope for assessment of CRF/2C colocalization. A single-plane 40× tiled image of a CRF/2C slice was obtained on a Zeiss 800 upright confocal microscope for the magnified image shown in Extended Data 6b, right. 3 × 5 tiled z stack (7 optical sections comprising 18 μm) images of 2C/Cre slices were obtained on a Zeiss 800 upright confocal microscope for the 2C/Cre validation. All images were preprocessed with stitching and maximum intensity projection. An image of the BNST from 3 mice in each condition was hand counted for each study using the cell counter plugin in FIJI (ImageJ). Cells were classified into three groups: probe 1+, probe 2+, or probe 1 and 2+. Only cells positive for a probe were considered. Results are plotted as average classified percentages across the three images. No specific method of randomization was used to assign groups. Animals were assigned to experimental groups so as to minimize the influence of other variables such as age or sex on the outcome. Pre-established criteria for excluding mice from behavioural analysis included (1) missed injections, (2) anomalies during behavioural testing, such as mice falling off the elevated plus maze, (3) damage to or loss of optical fibres, (4) statistical outliers, as determined by the Grubb’s test. A power analysis was used to determine the ideal sample size for behaviour experiments. Assuming a normal distribution, a 20% change in mean and 15% variation, we determined that we would need 8 mice per group. In some cases, mice were excluded due to missed injections or lost optical fibres resulting in fewer than 8 mice per group. For electrophysiology experiments, we aimed for 5–7 cells from 3–4 mice. Data are presented as means ± s.e.m. For comparisons with only two groups, P values were calculated using paired or unpaired t-tests as described in the figure legends. Comparisons across more than two groups were made using a one-way ANOVA, and a two-way ANOVA was used when there was more than one independent variable. A Bonferroni post-test was used following significance with an ANOVA. In cases in which ANOVA was used, the data met the assumptions of equality of variance and independence of cases. If the condition of equal variances was not met, Welch’s correction was used. Some of the sample groups were too small to detect normality (<8 samples) but parametric tests were used because nonparametric tests lack sufficient power to detect differences in small samples (Graphpad Statistics Guide – http://www.graphpad.com). The standard error of the mean is indicated by error bars for each group of data. Differences were considered significant at P values below 0.05. All data were analysed with GraphPad Prism software.