The Input-Output Relationship of the Cholinergic Basal ... - Cell Press

Resource

The Input-Output Relationship of the Cholinergic Basal Forebrain Graphical Abstract

Authors Matthew R. Gielow, Laszlo Zaborszky

Correspondence [email protected]

In Brief Monosynaptic viral tracing in transgenic rats reveals that patterns of input cells contacting cholinergic neurons are biased according to their cortical or amygdalar output. Mapping inputs across the entire brain, Gielow and Zaborszky provide structural evidence for networks enabling differential acetylcholine release.

Highlights d

Inputs to cholinergic cells are biased, varying by cortical output target

d

Proportions of input to a given cortical cholinergic output are reproducible

d

The most prominent input to many cholinergic cells is the caudate putamen

d

Medial prefrontal cortex-projecting cholinergic cells receive little caudate input

Gielow & Zaborszky, 2017, Cell Reports 18, 1817–1830 February 14, 2017 ª 2017 The Author(s). http://dx.doi.org/10.1016/j.celrep.2017.01.060

Cell Reports

Resource The Input-Output Relationship of the Cholinergic Basal Forebrain Matthew R. Gielow1 and Laszlo Zaborszky1,2,* 1Center

for Molecular and Behavioral Neuroscience, Rutgers, the State University of New Jersey, Newark, NJ 07102, USA Contact *Correspondence: [email protected] http://dx.doi.org/10.1016/j.celrep.2017.01.060 2Lead

SUMMARY

Basal forebrain cholinergic neurons influence cortical state, plasticity, learning, and attention. They collectively innervate the entire cerebral cortex, differentially controlling acetylcholine efflux across different cortical areas and timescales. Such control might be achieved by differential inputs driving separable cholinergic outputs, although no input-output relationship on a brain-wide level has ever been demonstrated. Here, we identify input neurons to cholinergic cells projecting to specific cortical regions by infecting cholinergic axon terminals with a monosynaptically restricted viral tracer. This approach revealed several circuit motifs, such as central amygdala neurons synapsing onto basolateral amygdala-projecting cholinergic neurons or strong somatosensory cortical input to motor cortex-projecting cholinergic neurons. The presence of input cells in the parasympathetic midbrain nuclei contacting frontally projecting cholinergic neurons suggest that the network regulating the inner eye muscles are additionally regulating cortical state via acetylcholine efflux. This dataset enables future circuit-level experiments to identify drivers of known cortical cholinergic functions. INTRODUCTION The connectome of the basal forebrain (BF) is not well understood due to the anatomical complexity of the region. Patients with Alzheimer’s disease and related dementias have a significant decrease of acetylcholine (ACh) in the cortex and show pathological changes in cholinergic neurons in the BF (Zaborszky et al., 2008). Thus, a complete understanding of its functional organization is warranted. ACh is released in the cortex by neurons whose cell bodies are located in the BF. Each of these cholinergic neurons supplies a small portion of cortex (Price and Stern, 1983), and ACh release is controlled precisely across different cortical regions and timescales, thought to be responsible for the variety of its cognitive effects (for review, see Gritton et al., 2016; Mun˜oz and Rudy, 2014). However, the mechanism behind this control is unknown.

Lesions of the BF in experimental animals or humans cause enhancement of slow oscillations and severe attention and memory deficits (Botly and De Rosa, 2012; Buzsa´ki et al., 1988; Lutkenhoff et al., 2015), while BF stimulation increases the spontaneous and visually driven cortical firing rates, improving neuronal response reliability (Pinto et al., 2013). Similarly, experiments in the somatosensory (Maalouf et al., 1998) or in the auditory cortex (Leach et al., 2013; Metherate and Weinberger, 1990) suggest that ACh plays a role in improving sensory perception. Release of ACh across different cortical areas covaries between sleep and wake states (Jasper and Tessier, 1971; Phillis, 1968; Sarter and Bruno, 2000). BF cells projecting to remote regions of cortex appear to have overlapping dendritic fields (Woolf, 1991), and cholinergic cells projecting to different cortical targets often intermingle across an extended region of the BF. These data together have been taken as evidence that the cholinergic BF is a diffuse projection system. However, discrete cholinergic efflux occurs per cortical region at different time points (Fournier et al., 2004; Parikh et al., 2007; Rasmusson, 1993), and recent tracing and optogenetic activation of cholinergic neurons in the mouse BF suggest that cholinergic neurons can modulate sensory cortical areas in a modality-specific manner (Kim et al., 2016). Part of this controversy is based upon the complex organization of BF projection neurons: cholinergic and non-cholinergic projections to the neocortex are not diffuse but are organized into segregated and overlapping pools of neurons that may transmit information from specific locations in the BF to subsets of cortical areas, themselves interconnected (Zaborszky et al., 2015a). Additionally, there are subregional differences in input across the BF (Zaborszky et al., 2015b), although how this affects cortical outputs is unknown. The relationship of synaptic inputs to target-identified cell populations cannot be readily assessed on a whole-brain level with classic anatomical techniques that rely on the combination of conventional anterograde and retrograde tracing (Zaborszky et al., 2006). Here, we employed output-specific monosynaptic viral tracing in cells expressing choline acetyltransferase (ChAT) to sample the input-output relationship of the cholinergic BF, and thereby identify on a whole-brain scale the neuronal populations likely to drive differential ACh efflux in particular output structures. Genetic access to BF ChAT cells was achieved by injecting Credependent helper viruses (AAV-EF1a-FLEX-TVAmCherry and AAV-CA-FLEX-RG) (Watabe-Uchida et al., 2012) at five points across the BF in ChAT::Cre rats (Witten et al., 2011) (Figures 1A and 1B). When these same helper viruses were used in the BF of the same line of rats, expression of helper viral product only occured in cells immunopositive for ChAT (Chavez and Zaborszky,

Cell Reports 18, 1817–1830, February 14, 2017 ª 2017 The Author(s). 1817 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Figure 1. Monosynaptic Viral Tracing (A) Cre-dependent helper viruses limit starter cells to the cholinergic cell type. Virus placed at cortical terminals retrogradely labels a subset of cholinergic cells that project to the cortical injection site (figure based on Wall et al., 2010). (B and C) 12 subjects (B) received helper virus across five basal forebrain sites and (C) received retrograde virus in a single BF projection target, whether in motor cortex (red), medial prefrontal cortex (green), orbitofrontal cortex (blue), or amygdala (yellow) (figure based on Paxinos and Watson, 2007). (D) Cells are visualized by the presence of mCherry (helper virus in cholinergic cells), GFP (afferent cells, monosynaptic inputs to starter cells), or both fluorophores (cholinergic starter cells).

2016). Tracing restricted to ChAT terminals was achieved by subsequently injecting a pseudotyped, helper-dependent, replication-defective rabies vector (EnvA-DG-rabies-eGFP) (Wall et al., 2010; Wickersham et al., 2007) into a single target area (prefrontal, orbitofrontal, motor cortex, or amygdala) in each subject (n = 12; three subjects per group; Figures 1A and 1C). Complementation of the deficient rabies with the rabies glycoprotein occurred in BF ChAT cells (starter cells) projecting to the cortical area of rabies injection. This allowed for retrograde monosynaptic spread of GFP-tagged rabies from these starter cells back to the cell bodies of their synaptic inputs (Figures 1A and 1D), whereas in the control condition, ChAT::Cre with rabies injection but no helper viruses, no GFP was detected. Therefore, we could map the inputs to specific output populations, regardless of how output somata are distributed across the BF itself. RESULTS The topography of basal forebrain cholinergic (BFc) cell bodies projecting to particular cortical targets has been studied exten1818 Cell Reports 18, 1817–1830, February 14, 2017

sively, resulting in many examples of ChAT populations that project to different targets yet partially overlap in the BFc space (for ref see Zaborszky et al., 2015a). The BF topography of starter cells in the current study confirms this rule of partial overlap (Figure 2; Table S1; Movie S1). When the monosynaptic inputs to the starter cells are considered across the three cortical target groups and the basolateral amygdala (a cortex-like structure), we find specific sets of inputs differ significantly between the four groups depending on the viral-injected cholinergic projection target (Wilks’ Lambda one-way multivariate analysis of variance [MANOVA]; F = 21.21, p = 0.008) (Figure S1). To make sense of this result, we took a detailed look at differential distributions of afferents in each input region. Dorsal Striatal and Accumbens Input Disparity Dorsal striatum is on average the largest input source for cholinergic BF cells. The proportion of these inputs is far less in the case of medial prefrontal cortex (mPFC) injection (Figure 3), with an average of 5% of inputs for mPFC cases versus 47% for the motor cortex injection group (p = 0.002;

lateral shell. Taken together, this demonstrates that different basalocortical cholinergic targets receive distinct inputs from dorsal and ventral striatum.

Figure 2. Topography of Starter Cells Cholinergic cells capable of spreading virus monosynaptically (mCherry+/ GFP+) in 12 subjects following retrograde viral injection in M1/M2 (red), mPFC (green), VO/LO (blue), or amygdala (yellow). Starter cells from individual cases (labeled with differently shaped symbols) were warped into a common template brain. ac, anterior commissure; BLA, anterior basolateral amygdala; f, fornix; GP, globus pallidus; HDB, horizontal diagonal band; ic, internal capsule; lo, lateral olfactory tract; LV, lateral ventricle; mt, mammillothalamic tract; MS/VDB, medial septum/ vertical diagonal band; opt, optic tract; sm, stria medullaris; SI/EA, substantia innominata/ extended amygdala.

independent-samples t test). There is a topography of inputs from the caudate-putamen (CPu) depending on cholinergic target site; ventral orbital and lateral orbital cortex (VO/LO)-projecting cholinergic cells receive CPu inputs from a medial area rostrally and more diffusely caudally, while motor cortex-projecting cholinergic cells receive input mainly from ventral CPu, and amygdalopetal cholinergic cells receive a majority of input from caudal CPu (Figure 4). Inputs to ChAT cells also arise from the ventral striatum (Za´borszky and Cullinan, 1992) and occupy a differing topography depending on the target of cholinergic starter cells (Figure S3). Amygdalopetal ChAT cells receive more inputs from medial nucleus accumbens core/shell, while motor-projecting ChAT cells receive more from lateral core/shell, and VO/LO and mPFC-targeting cholinergic cells receive only scattered cells from the

Cortical Feedback to Cholinergic Neurons Direct cortical input to BF ChAT cells was found in a variety of cortical locations in all subjects, arising mostly from deep cortical layers, with occasional labeling in layers II–III. An average across all subjects shows 9% of inputs to cholinergic cells came from cortex. This number drops to just 1.5% from the mPFC in particular, with most cells found in layers 5 and 6. Although mPFC control of BF cholinergic cells has been suggested elsewhere (Golmayo et al., 2003; Rasmusson et al., 2007), the present results suggest a role for other cortical areas as well. For instance, the somatosensory cortex regulates cholinergic motor efflux, since in the motor cortex injected cases, somatosensory cortices S1 and S2 supply an average of 2.3% of inputs, versus 0.5% for subjects with rabies injection elsewhere. The inputs from cortex are spatially sparse, a pattern also seen when cholinergic afferents are mapped without regard for output specificity (Do et al., 2016; Hu et al., 2016). Thus, for simplicity, we categorized all cortical areas into either isocortex, mesocortex, or allocortex according to McGeorge and Faull (1989) (see classification scheme in Table S3). Although low cortical cell numbers increase within-group variance and prohibit highly specific conclusions, it seems each group of target-identified cholinergic cells receives a specific combination of cortical input: as alluded above, the S1-S2 contribution of the total cortical input in case of motor cortex group is 24%–36% (1.6%–2.7% of all inputs), practically zero in the mPFC group and similarly strong in VO/LO cases (Figures S2 and S7). Also note the allocortical contribution is about 10% of the total input in the mPFC and less than 1% in the primary/secondary motor cortex (M1/M2) cases. From the allocortical areas, the most numerous cells were found in the piriform cortex-endopiriform nucleus. In the hippocampal complex, a few input cells were found in the stratum oriens of CA1 and CA3 and the subiculum in cases with mPFC or basomedial (BMP) amygdala injections. Occasional labeling was found in the entorhinal cortex as well. Images of cortical cells appear in Figure 5. Amygdala Feedback to Cholinergic Neurons We found the amygdala holds 6% of all inputs to cholinergic cells across all subjects (imaged cells in Figure 6). Notably, this figure increases to 12% when only amygdala-injected subjects are considered (including two subjects with a majority of rabies in the anterior basolateral amygdala [BLA] and a third in the posterior basomedial amygdala [BMP]). Although the medial nucleus and the anterior amygdaloid area contained a few input cells, the largest input source was seen in all compartments of the central nucleus, a structure known to have GABAergic outputs (Figure 3). These data demonstrate that the amygdala influences its own cholinergic innervation monosynaptically via a central nucleus feedback. Afferents from the medial division of the central nucleus, excited during fear expression, are likely suppressing ACh efflux in BLA under conditions of learned fear expression, whereas interpreting the influence of lateral central amygdala (CeL) input cells will require further study, as CeL contains Cell Reports 18, 1817–1830, February 14, 2017 1819

Figure 3. Distribution of Inputs across All Brain Regions Percentage of labeled inputs to cholinergic starter cells across all brain regions is shown per subject (individual bars clustered in groups of three). Regions of sparse input (where no injection group averaged >0.5%) are not shown. See also Figures S3 and S4 and Table S2. M1/M2, motor cortex; mPFC, medial prefrontal cortex; VO/LO, ventral/lateral orbitofrontal cortices.

both fear-excited and fear-inhibited cells (Ciocchi et al., 2010). As the action of ACh in BLA is dependent on both motivational state and state of the BLA principal cells (Power and McGaugh, 1820 Cell Reports 18, 1817–1830, February 14, 2017

2002; Unal et al., 2015), cholinergic input to the BLA may facilitate memory formation by biasing the synaptic competition in favor of the strongly activated neurons (Jiang et al., 2016).

Figure 4. Distribution of Striatal Input Cells There is a topography of afferents originating across different levels (rows) in the CPu when comparing three subjects with different rabies injection sites: the amygdala (left column), motor cortex (middle), or ventral orbitofrontal cortex (right column). Various contours delineate cytoarchitectonic areas where cells appear. BFc, basal forebrain cholinergic cells; BMP, posterior basomedial amygdala; M1/M2, primary/secondary motor cortex; VO, ventral orbitofrontal cortex.

Hypothalamus, Thalamus, and Brainstem Several regions within thalamus, hypothalamus, and brainstem can be viewed as a common source of inputs to cholinergic cells, with apparent disregard to the cholinergic output target. Some notable exceptions to this, where apparent topography exists between groups, are also described here. Hypothalamic inputs make up 2%–11% of all inputs to cholinergic cells (Table S2; Table S3). A substantial proportion of hypothalamic cells were found in the lateral preoptic-hypothalamic continuum

(LPO-PLH), but numerous input cells were also found in the supramammillary nuclei, ventrolateral preoptic nucleus (VLPO), supraoptic nucleus, and occasionally in the paraventricular, ventromedial, ventrolateral, and suprachiasmatic nuclei (Figure S2). The LPO-PLH area, especially around the fornix, contains orexin and MCH cells that oppositely regulate the sleep-wake cycle (Luppi et al., 2013). Although orexin axons in the septum synapse with cholinergic neurons and depolarize them (Wu et al., 2004), GFP+ input cells in the perifornical Cell Reports 18, 1817–1830, February 14, 2017 1821

Figure 5. Input Cells from Cortex In each pair of panels, GFP appears on DAPI (blue, left) and thionin Nissl (violet, right). (A) S1 cells synapsing onto M1/M2-targeted ChAT. (B) Labeled cell in the medial entorhinal cortex (mEC) synapsing onto mPFC-targeted ChAT. (C) Endopiriform cell synapsing onto mPFC-targeted ChAT. (D) Retrosplenial cell synapsing onto mPFC-targeted ChAT. (E, F, and I) CA1 cells of differing morphologies synapsing onto mPFC-targeted ChAT. (G) VO cell synapsing onto VO-targeted ChAT. (H) CA3 cell synapsing onto mPFC-targeted ChAT. Scale bars, 100 mM. cc, corpus callosum; cg, cingulum; CPu, caudate putamen; DG, dentate gyrus; En, endopiriform nucleus; fi, fimbria; or, oriens; PaS, parasubiculum; Pir, piriform cortex; rad, radiatum; S1, primary somatosensory cortex; VO, ventral orbitofrontal cortex.

regions were not colocalized with orexin immunoreactivity following mPFC rabies injection, as assessed in the subject with the largest number of hypothalamic input cells. We also checked for the TH immunoreactivity, abundant in many hypothalamic areas (see Chan-Palay et al., 1984) but found no colocalized GFP cells. Interestingly, we found large number of 1822 Cell Reports 18, 1817–1830, February 14, 2017

labeled cells and axons in the supraoptic nucleus (Figures S2C and S2C0 ) in several cases, including prelimbic, BLA and VO/Cl rabies injections. While a supraopic-cholinergic projection has not been described before, oxytocin axons emanating from the supraoptic nucleus and terminating in CeL has been suggested to be involved in attenuating fear response

Figure 6. Input Cells from Amygdala In each pair of panels, GFP appears on DAPI (blue, left) and thionin Nissl (violet, right). (A) AA cells synapsing onto mPFC-targeted ChAT. (B) BMP cells synapsing onto mPFC-targeted ChAT. (C) CeC cells synapsing onto BMP-targeted ChAT. (D) ACo cells synapsing onto mPFC-targeted ChAT. (E) CeL cells synapsing onto M1/M2-targeted ChAT. (F) AHiPM cell synapsing onto M1/M2-targeted ChAT. Scale bars, 100 mM. AA, anterior amygdaloid area; ACo, anterior cortical amygdaloid nucleus; AHiPM, posteromedial amygdalohippocampal area; BLA, anterior basolateral amygdala; CA3, hippocampal field CA3; Ce, central nucleus, amygdala; CeC, central nucleus, amygdala, capsular part; CeL, central nucleus, amygdala, lateral part; cp, cerebral peduncle; CPu, caudate putamen; cst, commissural stria terminalis; En, endopiriform nucleus; I, intercalated nucleus, amygdala; La, lateral amygdala; LOT, nucleus of the lateral olfactory tract; LV, lateral ventricle; opt, optic tract.

(Knobloch et al., 2012). Knowing the massive input from the central amygdala to cholinergic projection neurons (above), the supraoptic-cholinergic link may be involved in the same behavioral response.

The thalamic input to target-identified cholinergic cells is a relatively minor component (1.3%–8.6%) of all afferents. Notable nuclei that contain input cells include the parafascicular (PF), mediodorsal, ventromedial, and paraventricular. Interestingly, Cell Reports 18, 1817–1830, February 14, 2017 1823

Figure 7. Input Cells from Brainstem (A) GFP on DAPI background shows inputs in RIP, synapsing onto mPFC-targeted ChAT cells. (A0 ) The same cells from (A) on thionin Nissl (violet). (B) GFP on DAPI background shows inputs in dorsal raphe, synapsing onto M1/M2-targeted ChAT cells. (B0 ) The same cells from (B) on thionin Nissl (violet). (C and D) GFP inputs in VTA on Nissl (violet) background synapsing onto mPFC-targeted (C) or M1/M2-targeted ChAT cells (D), respectively. (E) GFP inputs in nucleus Darkschewitsch (Dk) projecting to mPFC-targeted ChAT cells. (E0 ) The same cells from (E) on thionin Nissl (violet). (F) Macro view of GFP inputs in PAG and PPT synapsing onto M1/M2-targeted ChAT cells. Scale bars, 200 mM. 3N, oculomotor nucleus; 7, facial nucleus; g7, facial nerve; Gi, gigantocellular reticular nucleus; IP, interpeduncular nucleus; isRt, isthmic reticular formation; MB, mammillary body; MG, medial geniculate nucleus; ml, medial lemniscus; mRt, mesencephalic reticular formation; PAG, periaqueductal gray; Po, posterior thalamic nuclei; PPT, pedunculopontine tegmentum; PT, pretectal nucleus; RIP, raphe interpositus; SNC, substantia nigra pars compacta; SNR, substantia nigra pars reticulata; VTA, ventral tegmental area.

almost all cases display input cells in a narrow triangular area, medial from medial geniculate body, occupying the SG (suprageniculate), MGM (medial geniculate body, medial), the posterior (Po), posterior triangular (PoT), posterior intralaminar (PIL), and subparafascicular parvicellular (SPFPC) nuclei (Figures 7C and 7D). Cells in this region are important for auditory plasticity (Disterhoft and Stuart, 1976; Edeline, 1990; Gabriel et al., 1975; Lennartz and Weinberger, 1992; McEchron et al., 1995; O’Connor et al., 1997; Ryugo and Weinberger, 1978) and most likely, in part, corresponds to the area of the peripeduncular nucleus in primates from which a heavy projection was described to the ‘‘basal nucleus of Meynert-substantia innominata’’ (NBM) complex (Jones et al., 1976). A topography exists both in the PF and the posterior intralaminar complex such that its medial portion is labeled in amygdaloid- and mPFC-injection cases, while its lateral portion contains input cells in motor cases. In most cases, the subthalamic (STN) nucleus, part of the basal ganglia circuitry, contained labeled cells with heaviest labeling found for motor and VO/LO injection cases. Although from our material it is unclear where the cholinergic cells are 1824 Cell Reports 18, 1817–1830, February 14, 2017

located that receive STN input and project to the motor or VO/LO cortices, it is likely that these cholinergic cells are distributed in SI/EA/GP/ic regions, that contain the majority of starter cholinergic projection neurons in these cases (Table S1). Thus, our results are compatible with recent electrophysiological data (Saunders et al., 2015), indicating that cholinergic cells in and around the globus pallidus are functionally integrated into basal ganglia circuitry. It remains for future studies to inquire whether this subthalamic input is specific for motor-projecting ChAT cells around the GP or within the horizontal diagonal band (HDB) as well. The average proportion of brainstem input to cholinergic starter cells varies between 2% (amygdaloid projecting) and 11% (VO/LO innervating) of all labeled input. Substantial components of these arise from the ventral tegmental area (VTA), substantia nigra, raphe nuclei, reticular formation, periaqueductal gray (PAG), and parabrachial nucleus (Figures 7 and S5). Interestingly, following motor and orbitofrontal rabies injections, a consistent labeling of input cells were observed in the EdingerWestphal (EW) nucleus, the site of origin of parasympathetic fibers to the inner ocular muscles, as well as in the interstitial nucleus of Cajal (InC) and Darkschewitsch (Dk) nuclei, that are associational oculomotor centers. Labeled brainstem cells were often distributed diffusely, lacking a clear overall topography between groups. However, differences can be seen when isolating particular nuclei. In the reticular formation mPFC- and VO/LO-projecting cholinergic neurons receive

brainstem input from more medially located areas than do M1/ M2-projecting cholinergic neurons. Another example is the distribution of input cells in the ventral midbrain: mPFC-projecting cholinergic cells receive input more from the VTA, while motor cortex-projecting cholinergic neurons receive from the substantia nigra (Figures 7C and 7D). Additionally, amygdala-projecting cholinergic neurons receive input from more ventral brainstem areas than do prefrontal- or motor-projecting ChAT neurons. The mix of discrete and common sources of input from thalamus, hypothalamus and brainstem signals that in future studies, nuclei from these regions should be considered as independent actors on the cholinergic system rather than one homogeneous block per region. The VTA (A10 cell group), substantia nigra (A9; SN), and retrorubral field (A8) all contain dopaminergic, GABAergic and glutamatergic (Vglut2-containing) neurons displaying a distinct subregional distribution pattern. In addition, these transmitterspecific neurons contain in various degrees dopamine (DAT) and vesicular monoamine (VMAT2) transporters and D2 receptors, thus individual subpopulations of neurons are likely to be differentially involved in drug addiction and various motivational and cognitive functions (Faget et al., 2016; Hur and Zaborszky, 2005; Morales and Root, 2014; Nair-Roberts et al., 2008; Tritsch et al., 2012). Cholinergic neurons in the BF receive synapses from the VTA and substantia nigra (Gaykema and Zaborszky, 1997; Zaborszky and Cullinan, 1996). Here we found occasional TH-containing input cells in the VTA and retrorubral field, although a majority of input cells in this region were TH-negative (Figure S6). Glutamatergic (Hur and Zaborszky, SFN Abstract, 2007) and GABAergic (Gaykema and Zaborszky, 1997) projections from the VTA to BF areas rich in cholinergic cells have been described, although it is not known whether these inputs target cholinergic neurons. Due to the divergent molecular composition of SN and VTA neurons and their role in various functions and diseases, further investigation into the behavioral role of these inputs to BFc neurons is warranted. Serotonin input to the BF apparently synapses only on parvalbumin-containing neurons (Leranth and Vertes, 1999), while cholinergic neurons do not seem to receive direct input from 5HT axons (Hajszan and Zaborszky, 2000). Although a few input cells in the mesopontine tegmentum were detected in mPFC and amygdala (BMP) rabies injection cases, we found that these cells were not colabeled following ChAT immunostaining, and thus we have no evidence to demonstrate a cholinergic brainstem synapse onto cholinergic BF cells. Additionally, projections from the parabrachial nucleus may contain glutamate (Hur and Zaborszky, SFN Abstract, 2007) and could participate in the hypercapnia-induced arousal effect (Kaur et al., 2013). Local Inputs to Cholinergic Neurons Local afferents to ChAT starter cells are found across the BF cholinergic volume, often appearing in the vicinity of starter cells (Figure 1D; Table S3). Across all cytoarchitectonic BFc subregions in all subjects, the number of labeled afferents within a BFc subdivision correlates significantly with the number of starter cells within the same structure (Spearman’s Rho = 0.77; p < 0.001), suggesting that cholinergic corticopetal neurons

likely receive local inputs primarily from neighboring rather than remote regions of the BF itself. Local neurons contacting cholinergic starter neurons may contain neuropeptide Y (NPY), as such neurons have been shown to synapse with cholinergic neurons using electron microscopic (EM) (Zaborszky and Duque, 2000) and optogenetic strategies (Nelson and Mooney, 2016). Local BF afferents may also contain GABA and/or glutamate. Local inputs to cholinergic outputs both drive and inhibit ChAT cells, depending on afferent cell identity (Xu et al., 2015). Recent awake behaving physiological data indicate heterogeneous response patterns among BF units becoming activated and inhibited to different phases of a directed attention task (Tingley et al., 2014), patterns that may be carried by local subgroups of BF input impinging on different cholinergic output targets. Further studies into the role of local afferents in contributing to specific behaviors are warranted. DISCUSSION Synaptic inputs to BF cholinergic cells used to be studied one input source at a time with EM (for reference, see Zaborszky et al., 2015b), yet the output targets of the cholinergic neurons in these studies were rarely investigated, which is important since heterogeneous cholinergic neurons from any one subregion in the BF collectively project to wide cortical areas. Monosynaptic viral tracing of neuromodulatory systems has been used in the noradrenergic, dopaminergic and serotonergic systems (Beier et al., 2015; Faget et al., 2016; Lerner et al., 2015; Ogawa et al., 2014; Pollak Dorocic et al., 2014; Schwarz et al., 2015; Watabe-Uchida et al., 2012; Weissbourd et al., 2014). BFc monosynaptic tracing was also reported (Do et al., 2016; Hu et al., 2016) though without reference to target specificity. Here, we sample monosynaptic inputs from across the brain to gain insight into the networks able to control cholinergic innervation of different cortical regions and the amygdala. Several motifs are described such as common thalamic, hypothalamic, and brainstem input to the four cholinergic projection populations examined, as well as inputs distinct to particular projection populations, such as strong somatosensory cortical input to motor cortex-projecting cells, versus strong piriform input to mPFCprojecting cells. Similarly, VTA input is biased toward mPFCinnervating cholinergic cells and SN to motor cortex-innervating cholinergic neurons. Several follow-up circuit-level experiments can be designed based on this database to identify drivers of known cholinergic-dependent cortical functions. Limitations In pilot studies, we attempted to image injection sites by mixing rabies with various labeling agents, from India ink to Dye I. Unfortunately, all these attempts resulted in no labeling in the brain whatsoever. Therefore, we identified the injection site by finding the pipette tract and mapping the region of gliosis and other debris created by the rabies injection itself. Following this mapping, Nissl staining of the same section revealed the cytoarchitectonic region of injection. Therefore, although the precise location of the center of each injection site is known, the spread (of approximately 100–200 nL of virus) is unknown. Due to the large size of the rabies virus particles and small volume Cell Reports 18, 1817–1830, February 14, 2017 1825

injected at each position, we feel the size of the spread is not likely to stray outside of the cytoarchitectonic areas reported. To this point, we have acted conservatively, since in cases where the injection site encroached upon a bordering area (e.g., claustrum or central amygdala), we included this bordering area in the description of the injection site to err on the side of caution. Injection sites in cortex and amygdala are centered on slightly different cytoarchitectonic zones when comparing subjects within group, such that different injections targeting the same gross region may receive cholinergic innervation from differing BF neurons. Although each subject therefore can be viewed as a complete and independent dataset, the grouping of subjects allows the results to be interpreted in a coherent manner, comparing four groups instead of twelve individuals. Limitations of this technique are known (Callaway and Luo, 2015). Additionally, since only one in every four sections throughout the brain was mapped and analyzed, each distribution of input cells should be viewed as a sample of a sample for this viral tracing, as previously suggested (Ginger et al., 2013). In spite of these various methodological limitations, we still see some within-group similarities and notable significant between-group difference in inputs collecting three rats per group for the four regions studied. Since most of the forebrain afferents, including cortical, amygdaloid and striatal axons appear to have a preferential distribution across cytoarchitectonically defined BF regions (Zaborszky et al., 2015b), additional specificity might be garnered if helper viruses were introduced into specific cytoarchitectonic BF subregions, followed by the same cortical rabies injections. Another potential caveat arises when we consider the possibility of cholinergic-cholinergic synapses spreading virus locally. This does not appear to be the case, given that the BF location of the majority of starter cells in each case is restricted only to BF subregions known to house cholinergic projection neurons innervating the cortical region in question. In other words, the pattern of starter cells across the BF matches the pattern occurring following classical retrograde tracer injection in the same cortical area. If, on the other hand, we were to find starter cells in large numbers all across the BF, occurring in areas not known to project to the cortical region in question, this would have caused us concern in our interpretation of the results. However, this was not the case. If we suppose that it is nevertheless possible that spread occurred between connected cholinergic cells, and that inputs of these secondarily infected cells were labeled, the resulting mono- and di-synaptic labeled population still comprises a circuit eventually impinging on the rabies-injected cortical area. Specific Cholinergic Input-Output Streams Our study demonstrates that although each cortical and amygdaloid area receives the majority of cholinergic input from cell bodies located in partially overlapping areas of the BF (Figure 2), each of these target-identified groups of cholinergic output neurons is innervated by a partially unique set of afferents (Figure 3). In addition to numerical differences of input cells in these networks (Table S3), convolved maps of inputs to output-identified cholinergic neurons reveal a grossly segregated topography supporting the partial separability of these networks (Movie 1826 Cell Reports 18, 1817–1830, February 14, 2017

S2). For example, in the rostral forebrain at the level of the septum, the amygdala-targeted input cells occupy a middle space between the mPFC-targeted input cells within the septum and more laterally located motor-targeted input cells occupying much of the striatum (see Figure S3). While the distribution of these three networks is visible rostral to the crossing of the anterior commissure, with the appearance of the amygdaloid body caudally, there is a lateralward shift of the input cells of the amygdala as more of such cells become visible in the central amygdala and the dorsal striatum. Eventually, the motor-targeted input cells in the striatum disappear and between 3.0 and 3.4 mm caudal to bregma, the dorsal striatum is almost exclusively occupied by amygdala-targeted input cells (Figure S4). While Figure S3 clearly demonstrates segregated amygdalaand motor-targeted input cells in the nucleus accumbens, there is also a substantial overlap of amygdala-, mPFC- and motor-targeted input cells in the dorsal striatum, interstitial nucleus of the anterior commissure (IPAC), LPO-PLH, and the PF of the thalamus, suggesting these latter areas may play more of a role in modulation of global cholinergic tone. Septal areas (excluding medial septum) prominently contact cholinergic cells that project to mPFC, representing on average 18% of all inputs, yet are comparatively minor inputs (1%) for other output groups (p < 0.0001; independent-samples t test) (Figure S1). The main component of these septal inputs arises from the dorsal lateral septum, a small area that on its own comprises an average of 7% of all input for mPFC cases. The primate homolog of this area, the anterodorsal septum is known to code for reward uncertainty (Monosov and Hikosaka, 2013), a behavioral measure disrupted in rats following immunolesion of BFc neurons (Cordova and Chiba, 2004, SFN Abstract). Furthermore, reward uncertainty information is carried by basal forebrain units (Ledbetter et al., 2016). This raises the possibility that reward uncertainty is updated in the mPFC with ACh, as driven monosynaptically by the likely GABAergic (Hur and Zaborszky, SFN Abstract, 2007) dorsal lateral (anterodorsal) septum. Cellular and Circuit Mechanisms of Cortical State Control Classically, three cortical states are differentiated, including wake, SWS and REM sleep (for reference, see Brown et al., 2012; McCormick et al., 2015). However, recent studies suggest a more complex scenario that is characterized by a continuum of states rather than sharp transitions and, both in SWS and waking, finer distinctions can be made in terms of internal cortical dynamics and responsiveness to external stimuli (Harris and Thiele, 2011; McGinley et al., 2015; Vyazovskiy et al., 2011). The mechanism of how cortical arousal, movement-related activity (Harrison et al., 2016), and pupil micro-dilations (McGinley et al., 2015) are linked remains unexplained. Input cell labeling in the EW, Dk, and InC nuclei in the upper midbrain tegmentum of some of the motor, VO/LO, and PrL/IL rabies injection cases suggests that BFc neurons receive information about pupil diameter and reflex gaze coordination and can broadcast these signals to frontal cortex, potentially to modulate attention. On another level, input cells to BFc neurons, in the pedunculopontine tegmental (PPT, see Figure 7), cuneiform, and parabrachial nuclei (see Table S3), largely

corresponding to the mesencephalic locomotor region, are good candidates to convey fast movement-related information that accompany cortical desynchronization and arousal (Bennett et al., 2014; Kaur et al., 2013; Lee et al., 2014; Nelson and Mooney, 2016). The brainstem motor-related inputs together with inputs from the locus coeruleus and dorsal raphe neurons, known to show state-related activity (Jones, 2005; Luppi et al., 2013), as well as top-down input from M2 cortex to cholinergic neurons may be a common mechanism by which sensory cortical neurons, including those in the auditory cortex, receive bottom-up and top-down signals (Chavez and Zaborszky, 2016; Nelson and Mooney, 2016). Putative GABAergic Circuits Further extrapolation of our results can be gained when considering afferent groups with known transmitter content. The largest input source we found is the dorsal striatum, known to elicit sustained GABAergic output activated by a variety of stimuli and actions, including by outcome-predictive cues (Adler et al., 2013; Kita and Kitai, 1988). While the accumbens, also GABAergic in its output, is contacting cholinergic outputs across all four groups in the present study, the relative absence of CPu input for mPFC-projecting ChAT cells, compared with the abundance of this projection in the other three groups (Figure 3), underscores the special role ACh plays in the mPFC. We can now hypothesize that while suppression of cholinergic outputs to multiple brain areas is likely highly dependent on the CPu, at the same time the dorsal striatum is likely playing little to no role in the inhibition of cholinergic tone in the mPFC. Given that discrete cholinergic depletion of the PFC disrupts working memory and attentional processes (Croxson et al., 2011; Dalley et al., 2004), this putative striatal dichotomy of control over cholinergic output modules is a prime candidate of topic for future experiments. Another input source known to be GABAergic (Hur and Zaborszky, SFN Abstract, 2007) arises from the lateral septum, constituting the largest source of afferents outside the BF in the mPFC injection cases and containing few to no cells in other groups (Figure 3). Further inputs, most likely GABAergic (NairRoberts et al., 2008), arise from the VTA and SN pars compacta contacting cholinergic outputs in all experimental groups. Other afferents arising from the ventral anterior and ventrolateral thalamic nuclei, the glutamatergic thalamocortical nuclei receiving basal ganglia output, selectively contact cholinergic outputs to motor cortex. Therefore, different cholinergic output streams may be individually controlled subcortically by differing loops of the basal ganglia, including large numbers of different GABAergic cells. This hypothesis is in agreement with data demonstrating BF cells reliably responding to particular events in a behavioral task, while some cells seem to be active during the exact same periods when others are silent (Tingley et al., 2014). In this way, differing input-output streams of the cholinergic BF may comprise modules important for differing behavioral demands. Conclusions Future optogenetic studies utilizing viral tracing and genetic methods are necessary to uncover the fine functional details that this BF input-output organization subtends. Toward this

end, we suggest two, not mutually exclusive principles: (1) segregated inputs to the BFc convey specific cognitive operations, and overlapping, common inputs mediate state-related changes; alternatively, (2) the complex set of inputs described in our study reflect multiple inbuilt templates for flexible control of cortical states in a modality-specific fashion via the tonic and phasic firing of cholinergic neurons (Hangya et al., 2015; Harrison et al., 2016). EXPERIMENTAL PROCEDURES Subjects Animals were treated in accordance with the National Research Council’s ‘‘Guide for the Care and Use of Laboratory Animals.’’ Experiments were performed with the approval of the Institutional Animal Care and Use Committee of Rutgers University. ChAT::Cre rats expressing cre under the ChAT promoter (Witten et al., 2011), a gift from Dr. Karl Deisseroth at Stanford University, were backcrossed with wild-type Long-Evans (Harlan). 13 naive ChAT::Cre adults (3+ months of age) were used in virus tracing experiments; 12 were used for cortically targeted tracing, and one was used as a no-helper-virus control. All were housed in cages of one to three animals on a 12-hr-light/12-hr-dark cycle. Viruses and Surgeries Subjects were anesthetized with 1%–4% isoflurane inhalation in O2. 12 rats received intracranial injection of 2.64 mL of each helper virus (both from University of North Carolina [UNC] vector core: rAAV5/EF1a-Flex-TVA-mCherry, titer 4.3e12 VP/mL; rAAV5/CA-Flex-RG, titer 2e12 VP/mL) (Watabe-Uchida et al., 2012), injected via micropipette across five locations filling the right BF (coordinates in mm relative to bregma, from pia, +0.5 AP, 1.05 mediolateral [ML], 6.7 dorsoventral [DV] for males, 6.0 DV females; 0.0 AP, 1.65 ML, 7.1 DV males, 6.5 DV females; 0.87 AP, 2.3 ML, 5.3 and 7.1 DV males, 5.2 and 6.5 DV females; 1.72 AP, 3.6 ML, 5.8 DV males, 1.7 AP, 3.4 ML, 5.6 DV females). 21 days later, EnvA-DG-rabies-eGFP (Salk vector core, titer 4.3e8 transducing units [TU]/mL) (Wickersham et al., 2007) was injected ipsilaterally into motor cortex (111nL at +1.1 AP, 3.2 ML, 1.0 and 1.3 DV from pia), mPFC (222 nL at +3.15 AP, 0.5 ML, 3.9 and 4.7 DV from skull), VO/LO (111 nL at +3.6 AP, 2.23 ML, 4.7 DV from skull), or amygdala (222 nL at 2.16 AP, 4.756 ML, 7.55 DV from pia). Each litter of animals was spread as evenly as possible across all injection groups, and both females and males appear in each group. One additional control animal (ChAT::Cre+) received rabies injections in all four of the above cortical/amygdalar loci, without receiving any helper viruses. Tissue Preparation and Imaging 7 days following rabies injections, animals were transcardially perfused with 0.1 M PB followed by 4% paraformaldehyde in 0.1 M PB. Brains were kept in this fixative overnight at 4 C and then sunk in 30% sucrose in 0.1 M PB. 50-mm sections were collected via sliding microtome, and every fourth section was mounted onto gelatin-subbed slides. All experimental animals showed good fluorescence, whereas the single control animal (rabies only; no helper virus injected) showed no fluorescence anywhere in the brain. Fluorescent images were acquired via light or confocal microscope, and the location of cell bodies was mapped with Neurolucida software (MBF Bioscience). Mapping was performed with the experimenter blind to the rabies injection location. Coverslips were then removed, and the same slides were prepared with a thionin Nissl stain. Nissl-stained sections were illuminated with bright-field and aligned to existing images or to existing Neurolucida maps previously created under fluorescence, and cytoarchitectonic boundaries were created in Neurolucida, allowing for precise identification of the regions where fluorescent cells were mapped. The raw cell counts per cytoarchitectonic area appear in Table S3. For the purpose of comparing cell distributions across subjects, maps of individual subjects were warped into a common volume using Neurolucida software. Data Availability Raw cell count data appear in Table S3. Any other relevant data are available from the authors.

Cell Reports 18, 1817–1830, February 14, 2017 1827

Code Availability The requisite combination and separation of Neurolucida files was performed via custom MATLAB scripts, available at http://zlab.rutgers.edu/modules/ members/warp.zip. Immunostaining Sections were exposed to primary antibodies overnight, utilizing either mouse-anti-tyrosine hydroxylase (EMD Millipore # MAB318; 1:1,000 dilution), goat-anti-orexin A (Santa Cruz Biotechnology # sc-8070; 1:500 dilution), or goat-anti-ChAT (EMD Millipore # AB144P; 1:500 dilution). Sections were then washed three times for 5 min each in 0.1 M PB (pH 7.4). Subsequently, sections were exposed to CY3-conjugated secondary antibodies raised against the host species of the corresponding primary antibody (Jackson ImmunoResearch) for 3 hr. Following three more washes in PB, sections are mounted to gelatin-subbed slides. Statistical Analyses Two-tailed independent-samples t tests were performed comparing afferents from specific regions between groups, with a significance level of a = 0.01. Additionally, a Wilks’ Lambda one-way MANOVA was performed on data from all regions of monosynaptic input, comparing between all four groups of cortical rabies injection, with a significance level of a = 0.01. Finally, a Spearman’s Rho test of correlation was performed on all cholinergic BF subregions in all animals (72 individual regions), comparing the subregional numbers of starter cells and their local afferents, with a significance level of a = 0.01. SUPPLEMENTAL INFORMATION Supplemental Information includes four figures, three tables, and two movies and can be found with this article online at http://dx.doi.org/10.1016/j.celrep. 2017.01.060. AUTHOR CONTRIBUTIONS Conceptualization, Data Analysis, and Writing, M.R.G. and L.Z.; Methodological Design and Experimental Procedures, M.R.G. ACKNOWLEDGMENTS We thank K. Deisseroth for providing ChAT::Cre rats for breeding, C. Chavez for technical assistance, A. Chiba, D. Sullivan, T. Koos, and D. Pare for manuscript comments, and R. Samulski, L. Lisowski, the UNC Vector Core, and the Salk Institute Vector Core for research materials. This work was supported by The Graduate School-Newark (M.R.G.) and an NINDS grant to L.Z. (NS023945). Received: August 6, 2016 Revised: December 5, 2016 Accepted: January 24, 2017 Published: February 14, 2017 REFERENCES Adler, A., Katabi, S., Finkes, I., Prut, Y., and Bergman, H. (2013). Different correlation patterns of cholinergic and GABAergic interneurons with striatal projection neurons. Front. Syst. Neurosci. 7, 47. Beier, K.T., Steinberg, E.E., DeLoach, K.E., Xie, S., Miyamichi, K., Schwarz, L., Gao, X.J., Kremer, E.J., Malenka, R.C., and Luo, L. (2015). Circuit architecture of VTA dopamine neurons revealed by systematic input-output mapping. Cell 162, 622–634. Bennett, C., Arroyo, S., and Hestrin, S. (2014). Controlling brain states. Neuron 83, 260–261.

Buzsa´ki, G., Bickford, R.G., Ponomareff, G., Thal, L.J., Mandel, R., and Gage, F.H. (1988). Nucleus basalis and thalamic control of neocortical activity in the freely moving rat. J. Neurosci. 8, 4007–4026. Callaway, E.M., and Luo, L. (2015). Monosynaptic circuit tracing with glycoprotein-deleted rabies viruses. J. Neurosci. 35, 8979–8985. Chan-Palay, V., Za´borszky, L., Ko¨hler, C., Goldstein, M., and Palay, S.L. (1984). Distribution of tyrosine-hydroxylase-immunoreactive neurons in the hypothalamus of rats. J. Comp. Neurol. 227, 467–496. Chavez, C., and Zaborszky, L. (2016). Basal forebrain cholinergic-auditory cortical network: Primary versus nonprimary auditory cortical areas. Cereb. Cortex, Published online April 12, 2016. Ciocchi, S., Herry, C., Grenier, F., Wolff, S.B.E., Letzkus, J.J., Vlachos, I., Ehrlich, I., Sprengel, R., Deisseroth, K., Stadler, M.B., et al. (2010). Encoding of conditioned fear in central amygdala inhibitory circuits. Nature 468, 277–282. Croxson, P.L., Kyriazis, D.A., and Baxter, M.G. (2011). Cholinergic modulation of a specific memory function of prefrontal cortex. Nat. Neurosci. 14, 1510– 1512. Dalley, J.W., Theobald, D.E., Bouger, P., Chudasama, Y., Cardinal, R.N., and Robbins, T.W. (2004). Cortical cholinergic function and deficits in visual attentional performance in rats following 192 IgG-saporin-induced lesions of the medial prefrontal cortex. Cereb. Cortex 14, 922–932. Disterhoft, J.F., and Stuart, D.K. (1976). Trial sequence of changed unit activity in auditory system of alert rat during conditioned response acquisition and extinction. J. Neurophysiol. 39, 266–281. Do, J.P., Xu, M., Lee, S.-H., Chang, W.-C., Zhang, S., Chung, S., Yung, T.J., Fan, J.L., Miyamichi, K., Luo, L., et al. (2016). Cell type-specific long-range connections of basal forebrain circuit. eLife 5, 1–17. Edeline, J.M. (1990). Frequency-specific plasticity of single unit discharges in the rat medial geniculate body. Brain Res. 529, 109–119. Faget, L., Osakada, F., Duan, J., Ressler, R., Johnson, A.B., Proudfoot, J.A., Yoo, J.H., Callaway, E.M., and Hnasko, T.S. (2016). Afferent inputs to neurotransmitter-defined cell types in the ventral tegmental area. Cell Rep. 15, 2796–2808. Fournier, G.N., Semba, K., and Rasmusson, D.D. (2004). Modality- and regionspecific acetylcholine release in the rat neocortex. Neuroscience 126, 257–262. Gabriel, M., Saltwick, S.E., and Miller, J.D. (1975). Conditioning and reversal of short-latency multiple-unit responses in the rabbit medial geniculate nucleus. Science 189, 1108–1109. Gaykema, R.P., and Zaborszky, L. (1997). Parvalbumin-containing neurons in the basal forebrain receive direct input from the substantia nigra-ventral tegmental area. Brain Res. 747, 173–179. Ginger, M., Haberl, M., Conzelmann, K.-K., Schwarz, M.K., and Frick, A. (2013). Revealing the secrets of neuronal circuits with recombinant rabies virus technology. Front. Neural Circuits 7, 2. Golmayo, L., Nun˜ez, A., and Zaborszky, L. (2003). Electrophysiological evidence for the existence of a posterior cortical-prefrontal-basal forebrain circuitry in modulating sensory responses in visual and somatosensory rat cortical areas. Neuroscience 119, 597–609. Gritton, H.J., Howe, W.M., Mallory, C.S., Hetrick, V.L., Berke, J.D., and Sarter, M. (2016). Cortical cholinergic signaling controls the detection of cues. Proc. Natl. Acad. Sci. USA 113, E1089–E1097. Hajszan, T., and Zaborszky, L. (2000). Serotoninergic innervation of basal forebrain neurons in the rat. In Serotonin: From the Molecule to the Clinic. A Serotonin Club/Brain Research Bulletin Conference (Elsevier), p. 97.

Botly, L.C.P., and De Rosa, E. (2012). Impaired visual search in rats reveals cholinergic contributions to feature binding in visuospatial attention. Cereb. Cortex 22, 2441–2453.

Hangya, B., Ranade, S.P., Lorenc, M., and Kepecs, A. (2015). Central cholinergic neurons are rapidly recruited by reinforcement feedback. Cell 162, 1155– 1168.

Brown, R.E., Basheer, R., McKenna, J.T., Strecker, R.E., and McCarley, R.W. (2012). Control of sleep and wakefulness. Physiol. Rev. 92, 1087–1187.

Harris, K.D., and Thiele, A. (2011). Cortical state and attention. Nat. Rev. Neurosci. 12, 509–523.

1828 Cell Reports 18, 1817–1830, February 14, 2017

Harrison, T.C., Pinto, L., Brock, J.R., and Dan, Y. (2016). Calcium imaging of basal forebrain activity during innate and learned behaviors. Front. Neural Circuits 10, 36.

Maalouf, M., Miasnikov, A.A., and Dykes, R.W. (1998). Blockade of cholinergic receptors in rat barrel cortex prevents long-term changes in the evoked potential during sensory preconditioning. J. Neurophysiol. 80, 529–545.

Hu, R., Jin, S., He, X., Xu, F., and Hu, J. (2016). Whole-brain monosynaptic afferent inputs to basal forebrain cholinergic system. Front. Neuroanat. 10, 98.

McCormick, D.A., McGinley, M.J., and Salkoff, D.B. (2015). Brain state dependent activity in the cortex and thalamus. Curr. Opin. Neurobiol. 31, 133–140.

Hur, E.E., and Zaborszky, L. (2005). Vglut2 afferents to the medial prefrontal and primary somatosensory cortices: A combined retrograde tracing in situ hybridization study [corrected]. J. Comp. Neurol. 483, 351–373.

McEchron, M.D., McCabe, P.M., Green, E.J., Llabre, M.M., and Schneiderman, N. (1995). Simultaneous single unit recording in the medial nucleus of the medial geniculate nucleus and amygdaloid central nucleus throughout habituation, acquisition, and extinction of the rabbit’s classically conditioned heart rate. Brain Res. 682, 157–166.

Jasper, H.H., and Tessier, J. (1971). Acetylcholine liberation from cerebral cortex during paradoxical (REM) sleep. Science 172, 601–602. Jiang, L., Kundu, S., Lederman, J.D., Lo´pez-Herna´ndez, G.Y., Ballinger, E.C., Wang, S., Talmage, D.A., and Role, L.W. (2016). Cholinergic signaling controls conditioned fear behaviors and enhances plasticity of cortical-amygdala circuits. Neuron 90, 1057–1070. Jones, B.E. (2005). From waking to sleeping: Neuronal and chemical substrates. Trends Pharmacol. Sci. 26, 578–586.

McGeorge, A.J., and Faull, R.L. (1989). The organization of the projection from the cerebral cortex to the striatum in the rat. Neuroscience 29, 503–537. McGinley, M.J., Vinck, M., Reimer, J., Batista-Brito, R., Zagha, E., Cadwell, C.R., Tolias, A.S., Cardin, J.A., and McCormick, D.A. (2015). Waking state: Rapid variations modulate neural and behavioral responses. Neuron 87, 1143–1161.

Jones, E.G., Burton, H., Saper, C.B., and Swanson, L.W. (1976). Midbrain, diencephalic and cortical relationships of the basal nucleus of Meynert and associated structures in primates. J. Comp. Neurol. 167, 385–419.

Metherate, R., and Weinberger, N.M. (1990). Cholinergic modulation of responses to single tones produces tone-specific receptive field alterations in cat auditory cortex. Synapse 6, 133–145.

Kaur, S., Pedersen, N.P., Yokota, S., Hur, E.E., Fuller, P.M., Lazarus, M., Chamberlin, N.L., and Saper, C.B. (2013). Glutamatergic signaling from the parabrachial nucleus plays a critical role in hypercapnic arousal. J. Neurosci. 33, 7627–7640.

Monosov, I.E., and Hikosaka, O. (2013). Selective and graded coding of reward uncertainty by neurons in the primate anterodorsal septal region. Nat. Neurosci. 16, 756–762.

Kim, J.-H., Jung, A.-H., Jeong, D., Choi, I., Kim, K., Shin, S., Kim, S.J., and Lee, S.-H. (2016). Selectivity of neuromodulatory projections from the basal forebrain and locus ceruleus to primary sensory cortices. J. Neurosci. 36, 5314– 5327.

Morales, M., and Root, D.H. (2014). Glutamate neurons within the midbrain dopamine regions. Neuroscience 282, 60–68. Mun˜oz, W., and Rudy, B. (2014). Spatiotemporal specificity in cholinergic control of neocortical function. Curr. Opin. Neurobiol. 26, 149–160.

Kita, H., and Kitai, S.T. (1988). Glutamate decarboxylase immunoreactive neurons in rat neostriatum: Their morphological types and populations. Brain Res. 447, 346–352.

Nair-Roberts, R.G., Chatelain-Badie, S.D., Benson, E., White-Cooper, H., Bolam, J.P., and Ungless, M.A. (2008). Stereological estimates of dopaminergic, GABAergic and glutamatergic neurons in the ventral tegmental area, substantia nigra and retrorubral field in the rat. Neuroscience 152, 1024–1031.

Knobloch, H.S., Charlet, A., Hoffmann, L.C., Eliava, M., Khrulev, S., Cetin, A.H., Osten, P., Schwarz, M.K., Seeburg, P.H., Stoop, R., and Grinevich, V. (2012). Evoked axonal oxytocin release in the central amygdala attenuates fear response. Neuron 73, 553–566.

Nelson, A., and Mooney, R. (2016). The basal forebrain and motor cortex provide convergent yet distinct movement-related inputs to the auditory cortex. Neuron 90, 635–648.

Leach, N.D., Nodal, F.R., Cordery, P.M., King, A.J., and Bajo, V.M. (2013). Cortical cholinergic input is required for normal auditory perception and experience-dependent plasticity in adult ferrets. J. Neurosci. 33, 6659–6671. Ledbetter, N.M., Chen, C.D., and Monosov, I.E. (2016). Multiple mechanisms for processing reward uncertainty in the primate basal forebrain. J. Neurosci. 36, 7852–7864. Lee, A.M., Hoy, J.L., Bonci, A., Wilbrecht, L., Stryker, M.P., and Niell, C.M. (2014). Identification of a brainstem circuit regulating visual cortical state in parallel with locomotion. Neuron 83, 455–466. Lennartz, R.C., and Weinberger, N.M. (1992). Frequency-specific receptive field plasticity in the medial geniculate body induced by pavlovian fear conditioning is expressed in the anesthetized brain. Behav. Neurosci. 106, 484–497. Leranth, C., and Vertes, R.P. (1999). Median raphe serotonergic innervation of medial septum/diagonal band of broca (MSDB) parvalbumin-containing neurons: Possible involvement of the MSDB in the desynchronization of the hippocampal EEG. J. Comp. Neurol. 410, 586–598. Lerner, T.N., Shilyansky, C., Davidson, T.J., Evans, K.E., Beier, K.T., Zalocusky, K.A., Crow, A.K., Malenka, R.C., Luo, L., Tomer, R., and Deisseroth, K. (2015). Intact-Brain Analyses Reveal Distinct Information Carried by SNc Dopamine Subcircuits. Cell 162, 635–647. Luppi, P.H., Cle´ment, O., and Fort, P. (2013). Paradoxical (REM) sleep genesis by the brainstem is under hypothalamic control. Curr. Opin. Neurobiol. 23, 786–792. Lutkenhoff, E.S., Chiang, J., Tshibanda, L., Kamau, E., Kirsch, M., Pickard, J.D., Laureys, S., Owen, A.M., and Monti, M.M. (2015). Thalamic and extrathalamic mechanisms of consciousness after severe brain injury. Ann. Neurol. 78, 68–76.

O’Connor, K.N., Allison, T.L., Rosenfield, M.E., and Moore, J.W. (1997). Neural activity in the medial geniculate nucleus during auditory trace conditioning. Exp. Brain Res. 113, 534–556. Ogawa, S.K., Cohen, J.Y., Hwang, D., Uchida, N., and Watabe-Uchida, M. (2014). Organization of monosynaptic inputs to the serotonin and dopamine neuromodulatory systems. Cell Rep. 8, 1105–1118. Parikh, V., Kozak, R., Martinez, V., and Sarter, M. (2007). Prefrontal acetylcholine release controls cue detection on multiple timescales. Neuron 56, 141–154. Paxinos, G., and Watson, C. (2007). The Rat Brain in Stereotaxic Coordinates (Academic Press). Phillis, J.W. (1968). Acetylcholine release from the cerebral cortex: Its role in cortical arousal. Brain Res. 7, 378–389. Pinto, L., Goard, M.J., Estandian, D., Xu, M., Kwan, A.C., Lee, S.-H., Harrison, T.C., Feng, G., and Dan, Y. (2013). Fast modulation of visual perception by basal forebrain cholinergic neurons. Nat. Neurosci. 16, 1857–1863. €rth, D., Xuan, Y., Johansson, Y., Pozzi, L., Silberberg, G., Pollak Dorocic, I., Fu Carle´n, M., and Meletis, K. (2014). A whole-brain atlas of inputs to serotonergic neurons of the dorsal and median raphe nuclei. Neuron 83, 663–678. Power, A.E., and McGaugh, J.L. (2002). Cholinergic activation of the basolateral amygdala regulates unlearned freezing behavior in rats. Behav. Brain Res. 134, 307–315. Price, J.L., and Stern, R. (1983). Individual cells in the nucleus basalis—diagonal band complex have restricted axonal projections to the cerebral cortex in the rat. Brain Res. 269, 352–356. Rasmusson, D.D. (1993). Cholinergic modulation of sensory information. Prog. Brain Res. 98, 357–364.

Cell Reports 18, 1817–1830, February 14, 2017 1829

Rasmusson, D.D., Smith, S.A., and Semba, K. (2007). Inactivation of prefrontal cortex abolishes cortical acetylcholine release evoked by sensory or sensory pathway stimulation in the rat. Neuroscience 149, 232–241. Ryugo, D.K., and Weinberger, N.M. (1978). Differential plasticity of morphologically distinct neuron populations in the medical geniculate body of the cat during classical conditioning. Behav. Biol. 22, 275–301. Sarter, M., and Bruno, J.P. (2000). Cortical cholinergic inputs mediating arousal, attentional processing and dreaming: Differential afferent regulation of the basal forebrain by telencephalic and brainstem afferents. Neuroscience 95, 933–952.

Wickersham, I.R., Lyon, D.C., Barnard, R.J.O., Mori, T., Finke, S., Conzelmann, K.K., Young, J.A.T., and Callaway, E.M. (2007). Monosynaptic restriction of transsynaptic tracing from single, genetically targeted neurons. Neuron 53, 639–647. Witten, I.B., Steinberg, E.E., Lee, S.Y., Davidson, T.J., Zalocusky, K.A., Brodsky, M., Yizhar, O., Cho, S.L., Gong, S., Ramakrishnan, C., et al. (2011). Recombinase-driver rat lines: Tools, techniques, and optogenetic application to dopamine-mediated reinforcement. Neuron 72, 721–733. Woolf, N.J. (1991). Cholinergic systems in mammalian brain and spinal cord. Prog. Neurobiol. 37, 475–524.

Saunders, A., Oldenburg, I.A., Berezovskii, V.K., Johnson, C.A., Kingery, N.D., Elliott, H.L., Xie, T., Gerfen, C.R., and Sabatini, B.L. (2015). A direct GABAergic output from the basal ganglia to frontal cortex. Nature 521, 85–89.

Wu, M., Zaborszky, L., Hajszan, T., van den Pol, A.N., and Alreja, M. (2004). Hypocretin/orexin innervation and excitation of identified septohippocampal cholinergic neurons. J. Neurosci. 24, 3527–3536.

Schwarz, L.A., Miyamichi, K., Gao, X.J., Beier, K.T., Weissbourd, B., DeLoach, K.E., Ren, J., Ibanes, S., Malenka, R.C., Kremer, E.J., and Luo, L. (2015). Viralgenetic tracing of the input-output organization of a central noradrenaline circuit. Nature 524, 88–92.

Xu, M., Chung, S., Zhang, S., Zhong, P., Ma, C., Chang, W.-C., Weissbourd, B., Sakai, N., Luo, L., Nishino, S., and Dan, Y. (2015). Basal forebrain circuit for sleep-wake control. Nat. Neurosci. 18, 1641–1647.

Tingley, D., Alexander, A.S., Kolbu, S., de Sa, V.R., Chiba, A.A., and Nitz, D.A. (2014). Task-phase-specific dynamics of basal forebrain neuronal ensembles. Front. Syst. Neurosci. 8, 174. Tritsch, N.X., Ding, J.B., and Sabatini, B.L. (2012). Dopaminergic neurons inhibit striatal output through non-canonical release of GABA. Nature 490, 262–266. Unal, C.T., Pare´, D., and Zaborszky, L. (2015). Impact of basal forebrain cholinergic inputs on basolateral amygdala neurons. J. Neurosci. 35, 853–863. Vyazovskiy, V.V., Olcese, U., Hanlon, E.C., Nir, Y., Cirelli, C., and Tononi, G. (2011). Local sleep in awake rats. Nature 472, 443–447. Wall, N.R., Wickersham, I.R., Cetin, A., De La Parra, M., and Callaway, E.M. (2010). Monosynaptic circuit tracing in vivo through Cre-dependent targeting and complementation of modified rabies virus. Proc. Natl. Acad. Sci. USA 107, 21848–21853.

Za´borszky, L., and Cullinan, W.E. (1992). Projections from the nucleus accumbens to cholinergic neurons of the ventral pallidum: A correlated light and electron microscopic double-immunolabeling study in rat. Brain Res. 570, 92–101. Zaborszky, L., and Cullinan, W.E. (1996). Direct catecholaminergic-cholinergic interactions in the basal forebrain. I. Dopamine-beta-hydroxylase- and tyrosine hydroxylase input to cholinergic neurons. J. Comp. Neurol. 374, 535–554. Zaborszky, L., and Duque, A. (2000). Local synaptic connections of basal forebrain neurons. Behav. Brain Res. 115, 143–158. Zaborszky, L., Wouterlood, F.G., and Lanciego, J.L. (2006). Neuroanatomical Tract-Tracing 3 (Springer). Zaborszky, L., Hoemke, L., Mohlberg, H., Schleicher, A., Amunts, K., and Zilles, K. (2008). Stereotaxic probabilistic maps of the magnocellular cell groups in human basal forebrain. Neuroimage 42, 1127–1141.

Watabe-Uchida, M., Zhu, L., Ogawa, S.K., Vamanrao, A., and Uchida, N. (2012). Whole-brain mapping of direct inputs to midbrain dopamine neurons. Neuron 74, 858–873.

Zaborszky, L., Csordas, A., Mosca, K., Kim, J., Gielow, M.R., Vadasz, C., and Nadasdy, Z. (2015a). Neurons in the basal forebrain project to the cortex in a complex topographic organization that reflects corticocortical connectivity patterns: An experimental study based on retrograde tracing and 3D reconstruction. Cereb. Cortex 25, 118–137.

Weissbourd, B., Ren, J., DeLoach, K.E., Guenthner, C.J., Miyamichi, K., and Luo, L. (2014). Presynaptic partners of dorsal raphe serotonergic and GABAergic neurons. Neuron 83, 645–662.

} , P., Na´dasdy, Z., and SomoZaborszky, L., Duque, A., Gielow, M., Gombko¨to gyi, J. (2015b). Organization of the basal forebrain cholinergic projection system: Specific or diffuse? In The Rat Nervous System (Elsevier), pp. 489–505.

1830 Cell Reports 18, 1817–1830, February 14, 2017

Cell Reports, Volume 18

Supplemental Information

The Input-Output Relationship of the Cholinergic Basal Forebrain Matthew R. Gielow and Laszlo Zaborszky

Figure S1, Related to Figure 3. Global input to output-specific ChAT cells Afferents, excluding local BF inputs, are represented proportionally by brain region for each subject. The rabies injection site appears at the top of each pie along with the subject number, and each row of three subjects is considered an experimental group. See also Figures S3-S4 and Table S2. BLA, anterior basolateral amygdala; BMP, posterior basomedial amygdala; CeC, central nucleus, amygdala, capsular part; Cl, claustrum; DP, dorsal peduncular cortex; DTT, dorsal tenia tecta; IL, infralimbic cortex; LO, lateral orbitofrontal cortex; M1, primary motor cortex; M2, secondary motor cortex; MO, medial orbitofrontal cortex; PrL, prelimbic cortex; VO, ventral orbitofrontal cortex.

Figure S2, Related to Figure 3. Input cells from hypothalamus In each pair of panels, GFP appears alone and on thionin Nissl (violet). (A) VLPO cells synapsing onto mPFC-targeted ChAT. (B) RChL cells synapsing onto mPFC-targeted ChAT. (C) SO cells synapsing onto mPFC-targeted ChAT. (D) Anterior VMH cells synapsing onto mPFC-targeted ChAT. (E) Gross distribution of input cells in the lateral hypothalamic area, periventricular gray and PF nucleus of the thalamus synapsing onto mPFC-targeted ChAT. Scale 100 µM. 3V, third ventricle; AH, AHA, anterior hypothalamus; Arc, arcuate hypothalamic nucleus; AHi, amygdalohippocampal area; cp, cerebral peduncle; f, fornix; fr, fasciculus retroflexus; LH, lateral hypothalamus; Me, medial amygdala; ml, medial lemniscus; MPA, medial preoptic area; mt, mammillothalamic tract; och, optic chiasm; ot, optic tract; Pa, paraventricular hypothalamus; PF, parafascicular thalamus; RCh, retrochiasmatic area; RChL, lateral retrochiasmatic area; SCh,

suprachiasmatic nucleus; SO, supraoptic nucleus; STh, subthalamic nucleus; VM, ventromedial thalamus; VMH, ventromedial hypothalamic nucleus; VLPO, ventrolateral preoptic nucleus; ZI, zona incerta.

Figure S3, Related to Figures 3 and S1. Distribution of cell bodies of inputs to target-identified cholinergic cells. Rostral forebrain levels All subjects are warped into a common volume to demonstrate their relative position. Markers are color-coded by group: motor (n=3, red), mPFC (n=3, green), VO/LO (n=3, cyan), and amygdala (n=3, yellow). ac, anterior commissure; cc, corpus callosum; lo, lateral olfactory tract; LO, lateral orbitofrontal cortex; LV, lateral ventricle; mPFC, medial prefrontal cortex; och, optic chiasm, VO, ventral orbitofrontal cortex.

Figure S4, Related to Figures 3 and S1. Distribution of cell bodies of inputs to target-identified cholinergic cells. Middle forebrain levels All subjects are warped into a common volume to demonstrate their relative position. Markers are color-coded by group: motor (n=3, red), mPFC (n=3, green), VO/LO (n=3, cyan), and amygdala (n=3, yellow). cc, corpus callosum; cp, cerebral peduncle; D3V, third ventricle; f, fornix; fi, fimbria; fr, fasciculus retroflexus; ic, internal capsule; LV, lateral ventricle; mt, mammillothalamic tract; opt, optic tract; pm, mammillary tract; rf, rhinal fissure; sm, stria medullaris.

Figure S5, Related to Figures 3 and S1. Distribution of cell bodies of inputs to target-identified cholinergic cells. Brainstem levels All subjects are warped into a common volume to demonstrate their relative position. Markers are color-coded by group: motor (n=3, red), mPFC (n=3, green), VO/LO (n=3, cyan), and amygdala (n=3, yellow). 4V, fourth ventricle; Aq, cerebral aqueduct; Cb, cerebellum; cp, cerebral peduncle; DG, dentate gyrus; DTg, dorsomedial tegmental area; fmj, forceps major of the corpus callosum; IC, inferior colliculus; MG, medial geniculate nucleus; ml, medial lemniscus; PAG, periaqueductal gray; Pn, pontine nuclei; rf, rhinal fissure; s5, trigeminal nerve; SC, superior colliculus.

Figure S6, Related to Figure 7. Colocalization of afferent cells with tyrosine hydroxylase immunostaining An image of monosynaptically-labeled GFP (A) is overlaid with TH immunostaining (B) in the dopaminergic midbrain to reveal a double-labeled cell (C), indicating the location of a TH+ cell which synapses onto a BF cholinergic cell. The region shown in A-C lies within the box drawn on an image of same section following Nissl staining (D). IP, interpeduncular nucleus; ml, medial lemniscus; PAG, periaqueductal gray; PBP, parabrachial pigmented nucleus of the VTA; R, red nucleus; RRF, retrorubral field; SNc, substantia nigra pars compacta.

Figure S7, Related to Figure 3. Distribution of inputs across cortex Percent of labeled inputs to cholinergic starter cells across all cortical regions is shown per subject (individual bars clustered in groups of three). Regions of sparse input (where no subject had > 0.5%) are not shown. M1/M2, motor cortex; mPFC, medial prefrontal cortex; VO/LO, ventral/lateral orbitofrontal cortices.

MSVDB HDB VP SI/EA GP ic

M1/M2 1 15 1 18 16 48

VO/LO 3 17 2 29 10 39

mPFC 37 45 6 9 2 0

Amygdala 3 6 16 41 24 10

Table S1, Related to Figure 2. Starter cell distribution The proportion of cholinergic starter neurons in each BF subregion (rows), as a percent of total BF starter cells in each subject, is averaged per rabies injection group (columns, n=3 per group). GP, globus pallidus; HDB, horizontal diagonal band of Broca; ic, internal capsule; M1/M2, primary/secondary motor cortex; mPFC, medial prefrontal cortex; MSVDB; medial septum and vertical diagonal band of Broca; VO/LO, ventral and lateral orbitofrontal cortices.

BF Accumbens Septal Nuclei Striatum IPAC Amygdala BNST Isocortex Mesocortex Allocortex Claustrum STN Thalamus Hypothalamus Brainstem

M1/M2 13.38 4.83 0.06 47.31 2.52 4.40 0.77 4.86 2.29 0.79 3.07 3.50 4.49 2.10 5.62

±SEM 5.82 2.42 0.05 6.09 0.65 2.28 0.08 1.12 0.30 0.41 0.47 2.74 1.69 0.27 0.79

mPFC 26.43 6.38 18.41 3.93 1.55 5.86 1.69 0.38 4.63 10.04 0.01 0.10 1.33 10.82 8.46

±SEM 2.60 1.99 4.29 0.56 0.37 1.67 0.40 0.19 2.25 1.09 0.01 0.06 0.23 0.98 2.06

VO/LO 22.05 3.18 1.69 25.11 0.56 3.06 2.52 1.59 4.10 5.38 1.76 1.04 8.61 8.58 10.77

±SEM 5.26 1.63 0.06 9.91 0.12 0.66 1.85 1.08 1.88 3.94 0.66 0.53 4.87 3.26 3.70

Amygdala 15.15 16.53 0.82 37.01 1.30 12.26 3.31 0.69 0.87 0.87 0.35 0.26 2.04 6.34 2.22

±SEM 4.57 8.11 0.26 8.33 0.40 1.49 1.08 0.32 0.10 0.14 0.02 0.16 0.71 1.18 0.64

Table S2, Related to Figures 3 and S1. Percentage distribution of labeled inputs For each brain region, the proportion of input for each subject was averaged with its rabies injection site group (n=3 per group) and presented +/- SEM. BF, basal forebrain; BNST, bed nucleus of stria terminalis; IPAC, interstitial nucleus of the posterior anterior commisure; M1/M2, primary/secondary motor cortices; mPFC, medial prefrontal cortex; STN, subthalamic nucleus; VO/LO, ventral/lateral orbitofrontal cortex.

Supplemental Table, Related to Figure 3 and Tables S1-S2. Raw Data The raw cell counts for all subjects, including with a guide to cortical nomenclature. Supplemental Video 1, Related to Figure 2. 3D view of starter cells The relative position of cholinergic starter cells for all subjects, warped into a common volume. Supplemental Video 2, Related to Figures S3-S5. 3D view of monosynaptic afferents The relative position of afferent cell bodies which synapse onto differing populations of ascending cholinergic cells for all subjects, warped into a common volume.