Selective serotonergic excitation of callosal projection neurons

Serotonin (5-HT) acting as a neurotransmitter in the cerebral cortex is critical for cognitive function, yet how 5-HT regulates information processing in cortical circuits is not well understood. We tested the serotonergic responsiveness of layer 5 pyramidal neurons (L5PNs) in the mouse medial prefrontal cortex (mPFC), and found three distinct response types: long-lasting 5-HT1A (1A) receptor-dependent inhibitory responses (84% of L5PNs), 5-HT2A (2A) receptor-dependent excitatory responses (9%), and biphasic responses in which 2A-dependent excitation followed brief inhibition (5%). Relative to 5-HT-inhibited neurons, those excited by 5-HT had physiological properties characteristic of callosal/commissural (COM) neurons that project to the contralateral cortex. We tested whether serotonergic responses in cortical pyramidal neurons are correlated with their axonal projection pattern using retrograde fluorescent labeling of COM and corticopontine-projecting (CPn) neurons. 5-HT generated excitatory or biphasic responses in all 5-HT-responsive layer 5 COM neurons. Conversely, CPn neurons were universally inhibited by 5-HT. Serotonergic excitation of COM neurons was blocked by the 2A antagonist MDL 11939, while serotonergic inhibition of CPn neurons was blocked by the 1A antagonist WAY 100635, confirming a role for these two receptor subtypes in regulating pyramidal neuron activity. Selective serotonergic excitation of COM neurons was not layer-specific, as COM neurons in layer 2/3 were also selectively excited by 5-HT relative to their non-labeled pyramidal neuron neighbors. Because neocortical 2A receptors are implicated in the etiology and pathophysiology of schizophrenia, we propose that COM neurons may represent a novel cellular target for intervention in psychiatric disease.

Layer 5 pyramidal neurons (L5PNs), which provide the bulk of cortical output, have been classified into distinct subclasses based on their morphology, physiology, axonal projections, and genetic expression patterns (Arlotta et al., 2005;Molnar and Cheung, 2006;Morishima and Kawaguchi, 2006;Sugino et al., 2006;Molyneaux et al., 2007;Chen et al., 2008;Leone et al., 2008;Brown and Hestrin, 2009;Morishima et al., 2011;Oberlaender et al., 2011). Each L5PN subclass represents a unique component of the local cortical microcircuit, displaying highly selective synaptic connectivity that generates directional information flow through segregated and parallel cortical output channels (Morishima and Kawaguchi, 2006;Brown and Hestrin, 2009;Morishima et al., 2011;Otsuka and Kawaguchi, 2011). It has been hypothesized that serotonergic 2A-dependent excitation may be restricted to one or more L5PN subtypes (Spain, 1994;Weber and Andrade, 2010), thereby allowing 5-HT to selectively enhance cortical output to specific target structures. To test this hypothesis, we used retrograde labeling to identify subtypes of L5PNs in the mouse prefrontal cortex based on their axonal projection patterns (Morishima and Kawaguchi, 2006;Dembrow et al., 2010;Morishima et al., 2011). We report that 5-HT, via 2A receptor activation, selectively generates excitatory and biphasic responses in pyramidal neurons projecting to the contralateral cortex, while brainstem-projecting pyramidal neurons are inhibited by 5-HT through 1A receptors. The identification of a specific pyramidal neuron subpopulation selectively excited by 2A receptors enhances our conceptual framework regarding serotonergic regulation of cortical microcircuits, and suggests callosal-projection neurons may represent a novel cellular target for the treatment of schizophrenia and other psychiatric diseases.

ANIMALS
Wild-type C57Bl/6J (3-weeks-8-months-old) and Thy-1-YFP line H (3-7-weeks-old; Jackson Laboratory; see Feng et al., 2000) mice were used in experiments according to methods approved by the Institutional Animal Care and Use Committee of Dartmouth College.

RETROGRADE LABELING
Subpopulations of pyramidal neurons were labeled based on their axonal projections using stereotaxic injections of fluorescent beads (Retrobeads from Lumafluor, Inc.; 700 nL of undiluted solution per injection) into the contralateral medial prefrontal cortex (mPFC) or the pons (see Figure 3) using age-appropriate coordinates (Paxinos and Franklin, 2004). Animals were anesthetized for surgery with an IP injection of 2,2,2-tribromoethanol (20 mg/Kg). Surgeries lasted approximately 45 min, after which animals were allowed to recover for 72 h before use in electrophysiological experiments. Locations of dye injections were confirmed in coronal sections of the mPFC or brainstem.

SLICE PREPARATION
Animals were anesthetized with isoflurane and decapitated. Brains were quickly removed into ice-cold artificial cerebral spinal fluid (ACSF) containing, in mM, 125 NaCl, 25 NaHCO 3 , 3 KCl, and 1.25 NaH 2 PO 4 , 0.5 CaCl 2 , 6 MgCl 2 , 25 glucose, and saturated with 95% O 2 /5% CO 2 . Coronal brain slices (200 µm thick) containing the mPFC were cut using a Leica VT 1200 slicer and transferred to a storage chamber filled with ACSF in which CaCl 2 was increased to 2 mM and MgCl 2 was decreased to 1 mM. Slices were maintained at 35 • C for approximately 1 h, then stored at room temperature for up to 8 h prior to use in experiments.

ELECTROPHYSIOLOGY
Slices were transferred to a recording chamber below an Olympus BX51WI microscope and continuously perfused with oxygenated ACSF at 35 • C. Whole-cell current-clamp recordings of L5PNs were made with patch pipettes (∼5 M ) filled with, in mM, 135 K-gluconate, 2 NaCl, 2 MgCl 2 , 10 HEPES, 3 Na 2 ATP, and 0.3 NaGTP (pH 7.2 with KOH). In some experiments, biocytin (7.5 mg/mL) was added to the pipette solution to allow for post-hoc visualization and morphological analysis of the recorded neuron (see below). Data were acquired with AxographX software (AxographX Company) using a BVC-700 amplifier (Dagan Corporation) and an ITC-18 digitizer (HEKA Instruments). Membrane potentials were filtered at 5 kHz, sampled at 25 kHz, and corrected for the junction potential of 12 mV. Input resistance (R N ) was determined from the slope of the linear portion of the steady-state voltage-current relationship established with a series of somatic current injections (generally −50 to +50 pA). The magnitude of "sag" rectification (indicative of HCN channels in pyramidal neuron dendrites; see Dembrow et al., 2010) was quantified using a hyperpolarizing current injection sufficient to generate a 20 mV peak hyperpolarization relative to the resting membrane potential. Sag was defined as the relative "rebound" (percent) from peak membrane potential hyperpolarization, as measured at steady-state. Serotonergic inhibition was quantified as the duration of cessation of action potential generation following 5-HT application (see below). 5-HTinduced excitation was quantified as the peak increase in action potential rate, relative to baseline, measured over a 5 s period occurring around the timing of peak spike frequency following 5-HT application.

DRUGS AND DRUG APPLICATION
Patch-pipettes were filled with 5-HT (100 µM) dissolved in ACSF and connected to a Toohey Spritzer (Toohey Company). Pipettes were positioned near the soma (within 50 µm) of targeted pyramidal neurons, and 5-HT was briefly applied (1 or 10 s) at low-pressure (∼4 PSI). 10 s applications were used in initial experiments in unlabeled tissue, while 1 s applications were used in tissue labeled with retrograde tracers (to reduce the prolonged duration of serotonergic inhibitory responses; compare Figure 1D with Figure 5A). Antagonists for 1A (WAY 100635; Sigma) and 2A (MDL 11939; Tocris) receptors were bath applied for 5 min prior to measurements of 5-HT

HISTOLOGY, IMAGING, AND MORPHOLOGICAL ANALYSIS
Slices containing biocytin-filled neurons were fixed in phosphatebuffered solution (PBS) containing 4% paraformaldehyde and 0.2% picric acid. After 24 h, slices were washed several times in PBS and incubated for up to 12 h in PBS containing 0.25% Triton X-100 and avidin conjugated to Alexa Fluor® 488 (20 µg/ml; Invitrogen). Slices were then dried, imbedded in FluorSave (EMD Chemicals), and imaged on a 2-photon microscope (Prairie Technologies). To compare neuron morphologies, we quantified somatic distance from the pia, maximum horizontal width of the apical tuft, the number of dendritic branch points in the tuft, and the number of primary oblique dendrites. Tracings of neurons were made using NeuronJ (freely available at http://www. imagescience.org/meijering/software/neuronj/) from z-stack projections (1 µm sections).

STATISTICAL ANALYSIS
Data are presented as mean ± SEM. Physiological measurements of membrane potential (V M ), R N , and sag rectification in 5-HT-inhibited, -excited, and -biphasic neurons were compared using One-Way ANOVAs (with Tukey-Kramer post-tests) for each parameter. One-Way ANOVAs and post-tests were also used to compare the physiology of callosal projection (5-HT-excited and -biphasic) and pontine-projecting L5PNs. Other comparisons used a 2-tailed Student's t-test, paired or un-paired, as appropriate. Significance was defined as p < 0.05.

SEROTONERGIC RESPONSES IN PREFRONTAL LAYER 5 PYRAMIDAL NEURONS
In initial experiments, we focally applied 5-HT (100 µM for 10 s) to L5PNs during periods of current-induced action potential generation in acute slices of unlabeled mPFC from 3-5weeks-old mice (Figure 1 and Table 1). 5-HT inhibited action potential generation in 84% (145 of  In L5PNs, a well characterized developmental increase in 1A receptor expression induces a phenotypic "switch" from serotonergic excitation to serotonergic inhibition during the third postnatal week (Zhang, 2003;Beique et al., 2004). Although we used neurons from animals over 3 weeks of age, we confirmed our results did not reflect delayed maturation of L5PNs by conducting additional experiments using neurons from 7-to 8-month-old mice (n = 11) and from 3-to 7-week-old Thy-1 YFP line H mice (n = 34), in which a subpopulation of L5PNs is labeled with YFP. We found 5-HT to excite similar proportions of L5PNs (18% and 20%, respectively) in both groups ( Figure 1E and Table 1).
Prefrontal L5PNs are heterogeneous in their physiological properties (Dembrow et al., 2010). To determine whether

Frontiers in Neural Circuits
www.frontiersin.org serotonergic responsiveness correlates with physiology, we compared resting V M , R N , and the amount of "sag" rectification occurring during hyperpolarizing current injections in 5-HTexcited and 5-HT-inhibited L5PNs (Figure 2 and Table 1). We found 5-HT-excited L5PNs had higher R N (100 ± 7 M ) and less sag (10 ± 1% sag) than did L5PNs inhibited by 5-HT (R N , and sag for inhibited L5PNs were 75 ± 2 M and 16 ± 0%, respectively, p < 0.05 for each when compared to values from 5-HT-excited neurons). These results suggest 5-HT may selectively excite a physiologically distinct subpopulation of L5PNs.

SELECTIVE SEROTONERGIC EXCITATION OF CALLOSAL/COMMISSURAL PROJECTION NEURONS
COM projection L5PNs connecting homologous cortical areas in the two cerebral hemispheres have relatively high R N and modest sag rectification, physiological properties similar to those found in 5-HT-excited L5PNs (Dembrow et al., 2010). To test whether COM neurons include those excited by 5-HT, we injected fluorescent Retrobeads unilaterally into the left mPFC and later recorded serotonergic responses in labeled COM neurons in slices of the contralateral cortex ( Figure 3A). Focal 5-HT application (100 µM for 1 s) generated excitatory (n = 15) or biphasic (n = 9) responses in all 5-HT-responsive COM neurons (93%; 24 of 26 tested neurons; Figures 4A,C). Purely inhibitory responses were never observed in COM neurons. Serotonergic excitation of COM neurons was blocked by MDL 11939 (n = 5; Figure 4D), confirming a functional role for 2A receptors in regulating COM neuron excitability.
We next applied 5-HT to a second, non-overlapping L5PN population projecting to the pons (CPn, see Dembrow et al., 2010) (Figure 3B). In contrast to COM neurons, all CPn neurons tested (n = 17) were inhibited by 5-HT (Figures 4B,C), and this inhibition was blocked by the 1A antagonist WAY 100635 (n = 4; Figure 4D). To confirm that selective serotonergic regulation of COM and CPn L5PNs does not result from indirect modulation of excitatory drive onto L5PNs, in additional experiments we measured serotonergic responses in the presence of blockers of fast synaptic transmission (4 mM kynurenic acid, to block ionotropic glutamate receptors, and 10 µM SR-95531, to block GABA A receptors; see Gulledge and Stuart, 2005). With fast synaptic transmission blocked, 5-HT selectively excited all COM neurons tested (n = 12), generating pure excitations (n = 7) or biphasic excitations (n = 5) indistinguishable from those generated in COM neurons in the absence of synaptic blockers (Figure 5). Similarly, in the presence of synaptic blockers, 5-HT inhibited all CPn neurons tested (n = 3) for durations comparable to those generated in the absence of blockers ( Figure 5). These data demonstrate that selective regulation of COM and CPn neuron excitability by 5-HT does not depend on changes in fast synaptic transmission.

SELECTIVE SEROTONERGIC EXCITATION OF COM NEURONS IN LAYER 2/3
In the mPFC, COM neuron somata are found in both layer 5 and layer 2/3 (Morishima and Kawaguchi, 2006). We focally applied 5-HT to COM-labeled and neighboring non-labeled layer 2/3 pyramidal neurons (L2/3PNs) to test whether selective serotonergic excitation of COM neurons is conserved across cortical lamina (Figure 7). Indeed, out of 21 COM-labeled L2/3PNs, 18 showed excitatory or biphasic responses. No COM L2/3PNs had purely inhibitory responses to 5-HT (3 COM L2/3PNs were not responsive to 5-HT). On the other hand, two-thirds of non-labeled L2/3PNs (n = 12 of 18) were inhibited by 5-HT, while another third (n = 6) exhibited excitatory responses. Physiological properties of COM and unlabeled L2/3PNs were, for the most part, similar and independent of their serotonergic responsiveness. The exception was the finding of a higher input resistance in COM L2/3PNs, relative to their unlabeled neighbors ( Table 2).
Together, these data demonstrate that selective excitation of COM Frontiers in Neural Circuits www.frontiersin.org neurons is not restricted to layer 5, but appears instead to be a common feature of prefrontal COM neurons.

DISCUSSION
We found that 5-HT selectively increases the excitability of COM neurons by activation of 2A receptors, while inhibiting output from the vast majority of cortical pyramidal neurons, including all CPn neurons, via 1A receptor activation. Because 2A receptors in the prefrontal cortex are implicated in the pathophysiology of schizophrenia (Willins and Meltzer, 1997;Gonzalez-Maeso and Sealfon, 2009;Benekareddy et al., 2010), our data suggest excessive activation of COM neurons, which provide the bulk of long-range intra-and inter-hemispheric cortico-cortical projections (Otsuka and Kawaguchi, 2011), may play a central role in psychosis. We, therefore, propose that COM neurons represent a novel cellular, rather than molecular, target for the treatment of psychiatric diseases involving dysregulation of 2A receptor signaling.

SELECTIVE EXCITATION OF COM NEURONS
We found three distinct responses to 5-HT in L5PNs from unlabeled tissue: 1A-mediated inhibition, 2A-mediated excitation, and biphasic responses resulting from coactivation of 1A and 2A receptors. 5-HT-excited L5PNs had physiological properties characteristic of COM neurons (Dembrow et al., 2010), and experiments targeting labeled COM neurons confirmed that this population is selectively excited by 5-HT via 2A receptors. On the other hand, 5-HT inhibited all labeled CPn neurons via 1A receptor activation. These data are significant in demonstrating, for the first time, selective serotonergic activation of a specific cortical output channel. 2A receptor-dependent serotonergic excitation of COM neurons may also explain the parallel rostral-to-caudal gradients found for cortical 2A receptor expression (Pazos et al., 1985;Weber and Andrade, 2010) and COM neuron density (Chao et al., 2009). Two observations suggest that 5-HT may excite additional, non-COM L5PN populations. First, differences in membrane physiology were more exaggerated when comparing 5-HTinhibited CPn neurons with excited and biphasic COM neurons than when comparing unlabeled L5PNs grouped according to their serotonergic response. For instance, CPn neurons were significantly more hyperpolarized than were COM neurons, but V M was not well correlated with serotonin responsiveness in unlabeled L5PNs ( Table 1, compare also Figure 2B with Figure 6B). Second, biphasic and excited COM neurons exhibited physiological properties distinct from the populations of excited and biphasic neurons found in unlabeled tissue. For instance, COM neurons had higher R N values than did unlabeled L5PNs excited by 5-HT (Table 1).
It is possible that some of the physiological differences observed in labeled and unlabeled L5PNs reflect unintentional selection bias when visually choosing unlabeled neurons for whole-cell recording. For instance, the tendency to target larger, more prominent somata could contribute to the lower input resistances found in unlabeled 5-HT-excited neurons. Similarly, targeting of larger neurons in unlabeled tissue might bias selection toward L5PNs projecting toward the brainstem (Morishima and Kawaguchi, 2006;Dembrow et al., 2010), which may explain why only 15% of unlabeled L5PNs showed 2A-dependent excitation even while COM neurons, which were overwhelmingly excited by 5-HT, make up closer to 25% of all L5PNs (Hattox and Nelson, 2007). Alternatively, differences in the physiology of COM neurons and unlabeled 5-HT-excited L5PNs are consistent with the hypothesis that additional L5PN subpopulations are excited via 2A receptor activation. Additional studies will be needed to test serotonergic responsiveness of other L5PN projection subtypes, such as those projecting to the thalamus, the ventral tegmentum, and dorsal raphe. It will also be necessary to test whether selective serotonergic activation of COM neurons is conserved across species. For instance, Beique et al. (2007) found that "approximately one-third" of deep layer pyramidal neurons in the rat mPFC are excited by 2A receptor activation. Our results suggest that the 5-HT-excited subpopulation of pyramidal neurons in the rat cortex may include COM neurons. Such a finding would suggest that 2A-dependent excitation of COM neurons may be a conserved feature of the mammalian prefrontal cortex.

ROLE OF SEROTONIN IN CORTICAL FUNCTION AND PATHOLOGY
The quest to identify a role of 5-HT in regulating cortical activity has been marked by the apparent paradox that pyramidal neurons express both inhibitory (1A) and excitatory (2A) receptors (Amargos-Bosch et al., 2004;Santana et al., 2004;Puig et al., 2010), yet the majority of adult pyramidal neurons are functionally inhibited by 5-HT (Araneda and Andrade, 1991;Beique et al., 2004;Zhang and Arsenault, 2005;Beique et al., 2007;Goodfellow et al., 2009;Zhong and Yan, 2011 (Yuen et al., 2008;Zhong et al., 2008a,b). Our data demonstrate 2A receptors selectively enhance action potential generation in Frontiers in Neural Circuits www.frontiersin.org COM neurons independent of synaptic input, but do not preclude the possibility that 2A receptors may also have effects on synaptic transmission in these neurons. Our finding opposing actions of 5-HT in COM and CPn neurons add to a growing body of evidence pointing to highly selective serotonergic modulation of cellular components within cortical microcircuits (Zhou and Hablitz, 1999;Foehring et al., 2002;Xiang and Prince, 2003;Amargos-Bosch et al., 2004;Puig et al., 2004;Kruglikov and Rudy, 2008;Lee et al., 2010). How does selective activation of COM neurons facilitate information processing in the prefrontal cortex? COM neurons are synaptically coupled to each other, and provide unidirectional excitatory input to CPn neurons (Morishima and Kawaguchi, 2006), suggesting that serotonergic activation of COM neurons may increase glutamate release onto both COM and CPn neurons, while direct 1A-dependent inhibition of CPn output would have little impact on ongoing COM neuron activity. If 5-HT-excited neurons are reciprocally connected (Morishima et al., 2011), 5-HT could promote reverberatory excitation in small networks of interconnected COM neurons (see also Williams et al., 2002), whose collateral connections to other COM and CPn neurons might underly 2A-dependent increases in spontaneous excitatory synaptic transmission (Aghajanian and Marek, 1997;Zhou and Hablitz, 1999;Lambe et al., 2000;Beique et al., 2007). Concurrent serotonergic inhibition of CPn and other L5PN subtypes would further enhance the overall signal-to-noise of cortical output to the brainstem. Serotonin also regulates the activity of cortical GABAergic interneurons (Zhou and Hablitz, 1999;Murakoshi et al., 2001;Foehring et al., 2002;Xiang and Prince, 2003;Weber and Andrade, 2010) and suppresses transmitter release from axon terminals (Tanaka and North, 1993;Zhou and Hablitz, 1999;Torres-Escalante et al., 2004;Kruglikov and Rudy, 2008;Troca-Marin and Geijo-Barrientos, 2010), both of which might selectively gate information flow in cortical microcircuits. For instance, ionotropic 5-HT 3A receptors are expressed by most, if not all, layer 1 interneurons (Foehring et al., 2002;Weber and Andrade, 2010), where 5-HT 3A -mediated excitation would be expected to enhance feed-forward inhibition of pyramidal neuron apical dendrites. Similarly, much as selective serotonergic regulation of pyramidal neuron excitability accounts for the heterogeneity of 5-HT responses reported in previous studies, selective control of transmitter release at synapses according to their pre-and postsynaptic identity may explain the diversity of effects of 5-HT on synaptic transmission (e.g., as in Troca-Marin and Geijo-Barrientos, 2010). Dedicated studies will be necessary to characterize selective serotonergic control of these other components of cortical microcircuits.
Finally, the precision with which 5-HT selectively regulates cortical output channels may explain why disruption of Frontiers in Neural Circuits www.frontiersin.org serotonergic signaling contributes to a wide variety of mental health disorders. In particular, excessive 2A receptor activation is implicated in the etiology of many psychiatric conditions, including depression, anxiety, and schizophrenia, while 2A antagonists are efficacious in treating these disorders (Naughton et al., 2000). The selective 2A-dependent excitation of COM neurons described here suggests these neurons may represent a novel cellular target for intervention in psychiatric disease.