Abstract
Response gain is a crucial means by which modulatory systems control the impact of sensory input. In the visual cortex, the serotonergic 5-HT2A receptor is key in such modulation. However, due to its expression across different cell types and lack of methods that allow for specific activation, the underlying network mechanisms remain unsolved. Here we optogenetically activate endogenous G protein-coupled receptor (GPCR) signaling of a single receptor subtype in distinct mouse neocortical subpopulations in vivo. We show that photoactivation of the 5-HT2A receptor pathway in pyramidal neurons enhances firing of both excitatory neurons and interneurons, whereas 5-HT2A photoactivation in parvalbumin interneurons produces bidirectional effects. Combined photoactivation in both cell types and cortical network modelling demonstrates a conductance-driven polysynaptic mechanism that controls the gain of visual input without affecting ongoing baseline levels. Our study opens avenues to explore GPCRs neuromodulation and its impact on sensory-driven activity and ongoing neuronal dynamics.
Similar content being viewed by others
Introduction
Neuromodulation of cortical processing has a substantial effect on neural response gain, which allows adaptive and flexible perception, cognition, and behavior1,2,3. However, the interactions across neuronal circuits underlying gain control remain generally puzzling and intensely debated4,5,6,7, as the results and conclusions heavily depend on experimental conditions and the configuration of the network in which neurons are embedded8. Another structural characteristic of brain modulatory systems is the co-distribution of their receptor families across various types of neurons9,10,11,12,13. This variety in expression patterns14 additionally hinders experimental and theoretical access to their coherent modulatory network function in vivo. In particular, receptors activated by dopamine, serotonin, acetylcholine, or noradrenaline can have opposing downstream effects on target neurons via their respective G protein-coupled receptors (GPCRs). Hence, the individual contribution of a single receptor type to the net functional output of a given cortical area remains largely obscure9,15,16,17,18,19.
Here, we exploited in vivo optogenetic control of the pathway of a single GPCR subtype, 5-HT2A, specifically activated in pyramidal and parvalbumin (PV) neurons in the mouse primary visual cortex (V1). Importantly, the 5-HT2A C-terminus of the optogenetic tool used enables its expression in the endogenous receptor-specific cellular domains20, providing a light-activatable functional equivalent to endogenous 5-HT2A receptor signals20,21,22,23,24,25. Moreover, targeting GPCR tools to receptor-specific domains triggers downstream kinetics with similar strength and ensures no overshoot compared to their native second messenger pathway effects25.
An increase in the cortical serotonin (5-HT) levels by microiontophoresis in V1 altered the gain of visual responses in rodents26 and in non-human primates27,28. Further pharmacological studies manipulating 5-HT receptor activation revealed a dominant contribution of the 5-HT2A receptor to gain modulation27,29,30,31 (Fig. 1a). However, these findings raise fundamental questions about the circuit mechanisms underlying such gain control. First, the 5-HT2A receptor pathway is excitatory in single cells32, leaving unanswered how the in vivo network generates an overall suppression upon its activation. Second, suppression of visual sensory activity was found to scale divisively with little to no influence on baseline20,29,30,31. Again, the circuit mechanisms reducing the magnitude gain of sensory responses without affecting spontaneous ongoing activity levels (Fig. 1b) remain to be identified. Hence, in this study, we isolate a specific coherent function of a modulatory system, such as the divisive gain modulation33,34 of sensory input by the 5-HT system, and demonstrate the network mechanisms by which this function is shaped across different types of neurons.
Results
We devised a method for optogenetic control of a single endogenous receptor pathway (5-HT2A) to explore basic polysynaptic mechanisms of GPCR signaling in two different types of cortical neurons (i.e., pyramidal and PV neurons in the visual cortex; see Fig. 1c–f). To achieve this, we expressed a chimeric construct consisting of light-activated mouse melanopsin (mOpn4L, targeted into 5-HT2A receptor domains20,35) to trigger Gq-signaling (Fig. 1g–i; Extended Data Figs. 1 and 2, Supplementary Movies 1–5) that mimics 5-HT2A receptor activation in V1 of anesthetized NEX-Cre and PV-Cre mice (Fig. 1e, f, Extended Data Figs. 3 and 4). To allow dense spatial sampling with good isolation and classification of simultaneously recorded single units as putative excitatory or inhibitory neurons (Fig. 1d), we performed recordings of extracellular activity using multi-channel silicon probes.
Cell-type-specific modulation of spontaneous activity by activation of the 5-HT2A receptor pathway
How is spontaneous activity affected when the 5-HT2A receptor pathway is activated in a cell-type-specific manner? To address this question, we isolated the modulatory effect of 5-HT2A signaling across distinct subpopulations of neurons (note that our recordings contain a mixture of directly and indirectly activated neurons). In comparison to the control spontaneous firing (S), optogenetic activation of the 5-HT2A receptor pathway in pyramidal neurons (Sph, Fig. 2a–e) led to an increase in spontaneous activity of both excitatory (Fig. 2a, dark orange trace, “direct effect”, +48% cf. Fig. 2b) and inhibitory neurons (Fig. 2a, light blue trace, indirect, “polysynaptic effect”, +70% cf. Fig. 2b). The slow rises and sustained increase in activity reflect the population dynamics as typically observed for GPCRs such as melanopsin or 5-HT receptors35. To further quantify these effects, we calculated an opto-index (OI), which scales values between −1 (infinite suppression) and +1 (infinite increase), while 0 indicates no effect on the baseline. The OI probability density functions show a similar distribution for the two neuronal populations, both peaking at a positive OI value (Fig. 2c–e) and with only a weak correlation between OI and firing rates (Extended Data Fig. 5). Quantification of activity changes shows a clear increase in activity of neurons with positive OI (Fig. 2d, e, NEX-Cre, excitatory: +76 ± 13%, inhibitory: +169 ± 59%), while a more modest decrease in OI−negative neurons was observed (NEX-Cre, excitatory: −19 ± 3%, inhibitory: −20 ± 3%).
Next, we activated the 5-HT2A receptor pathway specifically in PV interneurons. We found a slight but not significant increase in the firing of inhibitory neurons (Fig. 2f dark blue trace, +12% cf. Fig. 2g). However, it should be taken into account that the recorded sample of inhibitory neurons represent a heterogeneous group of interneurons, characterized by a distribution of the OI that is symmetrically centered around 0 (Fig. 2h, i left panel). One population of interneurons most likely represents PV neurons, which increase firing due to photoactivation of the 5-HT2A receptor (“direct effect”, see Fig. 2i solid dark blue trace, +83 ± 15% cf. Fig. 2j left panel) while subsequently suppressing other inhibitory neurons (“polysynaptic effect”, see Fig. 2i stippled dark blue trace, −39 ± 4% cf. Fig. 2j left panel; Extended Data Fig. 6 shows an example of monosynaptically connected interneurons). We also noticed a significant decrease in local field potential (LFP) power across different frequency bands caused by 5-HT2A receptor activation in PV neurons (Extended Data Fig. 7).
Activation of the 5-HT2A receptor pathway in PV interneurons led to an overall decrease in spontaneous activity of excitatory neurons (Fig. 2f, light orange traces, −35% cf. Fig. 2g). The population of excitatory neurons shows a distribution with a clear negative average OI (Fig. 2h, i right panel, light orange data; excitatory neurons with negative OI: −41 ± 2% cf. Fig. 2j right panel) with only minor increase in activity of neurons with positive OI (17 ± 6% cf. Fig. 2j right panel). The overall decrease in spontaneous activity of excitatory neurons is most likely due to the hyperpolarization caused by PV neurons via photoactivation of the 5-HT2A receptor pathway. Altogether, photostimulation of the 5-HT2A pathway separately in PV neurons produces heterogeneous bidirectional effects (suppression and facilitation) in inhibitory neurons and causes suppression in excitatory neurons. Whereas photostimulation of the 5-HT2A pathway in pyramidal neurons activates both excitatory and inhibitory populations.
5-HT2A signaling in PV neurons controls visual gain in excitatory neurons
Next, we investigated the effect of cell-type-specific 5-HT2A activation on visually evoked responses. Visual stimuli consisting of vertical or horizontal gratings with 100% contrast were repeatedly presented at an interval of 3 s (Fig. 3). We measured neuronal activity during visual stimulation alone (Fig. 3a, b, black traces, V) and with concurrent photostimulation of the 5-HT2A receptor pathway (Fig. 3a, b, blue traces, Vph). To obtain the photostimulation-related response component over time, we subtracted the two traces. The resulting activity (Fig. 3a, b, gray traces, Vph-V) resembled the trends in spontaneous activity found during 5-HT2A photostimulation across all neuron types except when the 5-HT2A receptor pathway was stimulated in PV interneurons, and activity was measured in excitatory neurons (Fig. 3b, light orange encircled panel). There was no effect on the OI magnitude when pyramidal neurons were separately photoactivated (Fig. 3c). Only photostimulation of PV interneurons caused excitatory neurons to display transient dips in their Vph-V trace (cf. Fig. 3b, gray trace in the light orange encircled panel), indicating a substantially larger suppression of evoked activity than spontaneous activity during visual responses (Fig. 3d). In all other cases the changes in amplitude of the visual responses could be solely explained by 5-HT2A-induced effects on baseline (cf. Fig. 2). Thus, a reduction in the gain of visually evoked activity was exclusively revealed in the response magnitude (i.e., the difference between response amplitude and baseline, Fig. 3e left sketch) of excitatory neurons after activation of the 5-HT2A receptor pathway in PV neurons (Fig. 3e, light orange trace) without changing preferred orientation31 (Extended Data Fig. 8).
A linear regression model applied to the obtained magnitude values of all excitatory neurons revealed that gain suppression is divisive (16% reduction in slope, i.e., β4/β2 in Fig. 3f; see Extended Data Fig. 9 for similar results using various contrasts). We conclude that the divisive control of visual input is largely based on an “indirect” polysynaptic network effect triggered by “direct” 5-HT2A activation in PV interneurons.
Simultaneous activation of 5-HT2A in pyramidal and PV neurons allows for gain control of external input without changing levels of baseline activity
Next, we activated the 5-HT2A pathway in both pyramidal and PV neurons simultaneously (see “Methods” section). This “systemic” activation revealed two substantial deviations from a simple superposition of the above results. First, in contrast to the above 5-HT2A pathway activation in the single subpopulations, no changes in baseline amplitude was observed after activating the pathway jointly across both cell types (Fig. 4a, Extended Data Fig. 10a). Second, gain control of visual input was no longer exclusively present in excitatory neurons (cf. Fig. 3e, f) but additionally found in inhibitory neurons (Fig. 4b, d, Extended Data Fig. 10b). Thus, we obtained a striking difference to a linear effect suggesting a significant change of the cortical processing regime when the 5-HT2A pathway is concurrently activated in pyramidal and PV neurons. Moreover, the maximal suppression in response magnitude of excitatory neurons was significantly stronger in the systemic condition than for 5-HT2A pathway activation in only PV neurons (54% vs 20%, p < 0.01, two-sided two-sample t-test).
Whether the observed divisive suppression (Fig. 4c, d) displays also properties of normalization remains to be tested with different stimulus contrasts. A previous study, in which the release of 5-HT was optogenetically controlled, showed magnitude normalization of V1 responses to varying contrasts, most likely involving 5-HT2A signaling as suggested by additional pharmacological manipulations29.
Spiking network model suggests changes in cortical processing regime through 5-HT2A-induced polysynaptic mechanism
We employed a spiking cortical network model developed by Sadeh et al. 36 with slight modifications, consisting of excitatory and inhibitory units driven by realistic neuronal parameters (Table 1). 5-HT2A activation was introduced to this model by adding background excitatory input to varying fractions of units. This free parameter also accounts for the fact that PV neurons are a subpopulation of inhibitory neurons and for variations in the expression strength of our 5-HT2A viral construct.
Similar to the measured neuronal data, when activating 5-HT2A in only pyramidal cells, the model showed increased spontaneous activity of both inhibitory and excitatory units (Fig. 5a, left data panel, Extended Data Fig. 11a), demonstrating an overall excitatory effect on the network (Extended Data Fig. 12a). Conversely, separate 5-HT2A activation of inhibitory units caused an increase of their baseline activity (Fig. 5b, left data panel, dark blue), while baseline activity of excitatory units was suppressed (Fig. 5b, left data panel, light orange; Extended Data Fig. 12b). Moreover, the OI of directly stimulated inhibitory units strictly increased, while the OI of indirectly affected inhibitory units decreased (Extended Data Fig. 11b). In contrast, the solely, indirectly stimulated excitatory subpopulation maintains a fairly compact OI distribution (Extended Data Fig. 11b). This matches the experimental data (Fig. 2h) where the shape of the OI distribution of the excitatory population remains roughly similar to the control condition while the distribution of the inhibitory population becomes much broader following photostimulation. Thus, the model captured both the experimentally observed depolarizing effect within subpopulations of interneurons via the direct 5-HT2A pathway and the “indirect” (i.e., multi-synaptic) effect, that is, their inhibitory influence on excitatory units. The network model also reproduced the experimentally observed changes in visually evoked responses. Separate 5-HT2A activation in the subpopulation of excitatory units led to minor rises in evoked magnitude across the entire network (Fig. 5a, right panel). Conversely, separate 5-HT2A activation in inhibitory units led to a slight decrease in response magnitude across the population of inhibitory units (Fig. 5b, right panel, dark blue). However, magnitude gain was most prominently reduced in excitatory units (Fig. 5b, right panel, light orange) and also divisive (Extended Data Fig. 13), similar to our physiological data (Fig. 3).
Finally, we tested the model’s prediction for systemic activation of 5-HT2A receptors, mimicking their joint activation across excitatory and inhibitory neurons. When modeling the above effects in a mean-driven regime where all parameters are constant, yielded approximately an additive effect, the sum of activating 5-HT2A in each subpopulation (cf. Fig. 5a, b) separately. Even though excitatory units displayed negligible changes in baseline firing as in the experiments, significant modulation of evoked response gain was absent in these units (Fig. 5c, triangles). Contrary to the systemic activation in the experiments, the baseline activity in the inhibitory units increased without any effect on response gain (Fig. 5c, circles). Hence, we observed different results for the mean-driven regime (Fig. 4) and therefore, the simple linear model substantially fails in explaining the experimentally revealed values.
When all other parameters remain constant, increasing 5-HT2A-mediated excitatory and inhibitory conductance led to increases in baseline activity. Whereas at higher levels of conductance, baseline activity declined. Indeed, previous studies applying 5-HT2A agonist in monkey visual cortex showed bidirectional modulatory effects on visual responses dependent on initial firing rates27. However, recent pharmacological studies29,31 that employed systemic modulation of endogenous 5-HT2A receptors in mice (targeting 5-HT2A simultaneously across all cell types) suggested that spontaneous activity levels are unaffected. In these two studies, systemic activation of the 5-HT2A receptor had strong suppressive effects on evoked visual responses (−21 and −67%, respectively).
How is it possible then, that following systemic and specific 5-HT2A activation, the baseline firing rate remains constant, while at the same time, response amplitudes are modulated? To reconcile our present findings, we consider that our network model operates in a fluctuation-driven regime37. In this regime, the mean membrane potential of a given unit does not change while both excitatory and inhibitory input rates increase, i.e., by balancing each other38,39. However, the fluctuations of the membrane potential change in a non-monotonic, inverted U-shaped, fashion37. For low conductance, fluctuations increase with the conductance, which leads to an increase in the firing rate of a target neuron as fluctuations exceed the spiking threshold more often. For high conductance, the fluctuations decrease, and thus, the firing rate of the target unit decreases. The model correctly predicts baseline stability and visual gain suppression following systemic 5-HT2A activation (Fig. 5d). Moreover, it reproduced that visual gain suppression is stronger in systemic activation than for isolated PV-specific 5-HT2A activation (cf. Fig. 5b, light orange trace and Fig. 5d, blue trace with triangles in the right panel). Our systemic activation data and the previously reported values of visual gain suppression29,31 are in agreement with the fluctuation-driven model when 35–85% of all excitatory and inhibitory units are activated by 5-HT2A receptors (Fig. 5d, blue area). Thus, modeling of systemic 5-HT2A receptor activation in a balanced regime (Extended Data Fig. 14) shows that the network maintains a stable baseline, buffering against spontaneous activity changes, while modulating external input gain.
Specifically, the inverted U-shape profile of conductance-dependent membrane fluctuations enables units in the model to exhibit the same baseline firing rate at two different rates of excitatory and inhibitory synaptic inputs (Fig. 5e, two left panels). The low-rate setting mimics input rates without 5-HT2A activation (Fig. 5e, 0%), while the high-rate setting mimics the input rates when the 5-HT2A receptor is activated systemically (Fig. 5e, blue bar indicates 5-HT2A receptor activation, 50% of all units in the example shown). As a result, spontaneous spiking remains constant in both cases (Fig. 5e, compare pre-stimulus times, 0–2 s in the two right panels) while the sensory gain is reduced (Fig. 5e, compare amplitudes of black (0%) and blue (50%) traces, respectively).
Discussion
Our experimental and modeling results suggest that the 5-HT2A receptor contributes to divisive control of response gain. Its activation in PV interneurons directly mediates this effect, while its simultaneous activation in pyramidal cells may further amplify suppression of response gain through increased overall conductance. The 5-HT2A receptor can separately modulate spontaneous activity upon cell-specific activation but has little net baseline effect during strong systemic drive. Hence, at the network level, the 5-HT2A receptor supports specific and independent modulation of one activity stream, i.e. visually evoked input, while leaving the other one, i.e., spontaneous ongoing activity, largely intact. Whether these findings also hold true for cortices of other modalities needs further exploration.
Further we showed that selective optogenetic activation of the 5-HT2A receptor pathway in distinct cortical populations, either pyramidal or PV neurons, uncovers a mixture of direct (i.e. receptor activated) and indirect (i.e. postsynaptic) effects of increased and decreased spiking dependent on neuron type. Combining these multiple effects in a computational model revealed that the systemic activation of cortical 5-HT2A receptors results in increased overall conductance across the network. As a consequence, the reduced driving force of sensory input produced suppression of evoked visual spiking responses. Thus, our model explains the 5-HT-induced reduction in pyramidal cell output upon visual stimulation27,28,29,30,31 by a reciprocal balanced action of the 5-HT2A receptor in antagonistic (excitatory vs. inhibitory) neuron types.
Importantly, we found that tonic balanced excitatory and inhibitory drive by 5-HT2A activation increased conductance across the network with only mild net effects on the average ongoing spiking activity. This indicates functional segregation in which 5-HT2A activation produces a coherent and selective impact on sensory input40. In addition, the tonic 5-HT2A-mediated increase in baseline conductance was much higher than the increase in conductance elicited by visual input. This suggests that sensory gain modulation comes at the cost of high metabolic turnover when 5-HT levels are elevated. Interestingly, 5-HT levels change during day-night cycle with lower values during rest41,42. We, therefore, speculate that reduced 5-HT levels partly contribute to lower energy consumption and cortical resting-state changes during sleep, when the processing of sensory stimuli is naturally reduced compared to awake state.
Our computational model potentially solves two apparently conflicting observations. First, during systemic activation, the gain of the visually evoked response is suppressed even though the baseline firing rate is unchanged. Second, during cell-type-specific modulation both spontaneous and visually evoked response change. Using a single set of modeling parameters, by modulating 5-HT2A–mediated changes in conductance, we were able to account for all experimental results. However, our model crucially depends on the network operating in a fluctuation-drive regime37. Although visual cortical neurons have been shown to operate in this regime during sensory processing43, further details of the modeling constraints remain to be explored. The reduction in gain obtained here in-silico were similar to our experimental findings and values in the literature29,31. However, the involvement of other 5-HT receptor- and cell types, most likely contributing to a further fine-tuning of network responses should be considered15,27,44,45,46,47,48,49. For example, the expression pattern of our construct does not concur with the normal complex distribution of 5-HT2A receptors across cortical layers47, which naturally serves further signal tuning within a spectrum of functions. Thus, the dependence of the mechanisms on layer-specific circuitries needs further study. Using subcutaneous injections of 2,5-dimethoxy-4-iodoamphetamine (DOI) to stimulate endogenous 5-HT2A receptors in mice, Michaiel and colleagues31 showed that suppressive effects were more pronounced in layer 2/3 and also cell type-specific. Moreover, the expression level of optogenetic tools can vary, as accounted for in our model by varying the percentage of activated units (Extended Data Fig. 14). Note also that our recordings include neurons with different preferred orientations. Thus, our visual stimulation with a single orientation yields an average effect of gain suppression over the population rather than scrutinizing effects on individual tuning curves. Hence, possible mixtures of divisive and subtractive or purely subtractive suppression at the single neuron level8 might be undetected. Interestingly, similar to our observations, Michaiel and colleagues31 found no changes in preferred orientation following DOI administration. Finally, the impact of 5-HT signaling may differ under wakeful conditions compared to the anesthetized state where inhibition and 5-HT levels are affected50,51. In fact, we showed recently that 5-HT-induced suppressive effects are less pronounced under awake conditions as compared to anesthetized preparations29. However, while the value of effect sizes can deviate between the light-activated and the native 5-HT2A system, it appears unlikely that our manipulation imposes fundamental different operations across the native functional architecture. Our findings demonstrate that joint and balanced activation of the 5-HT2A pathway (i.e. simultaneous Gq-signaling in pyramidal and PV neurons) leads to a qualitative change in the cortical state which allows for gain control of external input without changes in baseline.
We conclude that 5-HT2A-induced gain modulation of external input is achieved through balanced excitatory and inhibitory ongoing synaptic input52,53 with only minor effects on the amplitude of baseline activity. This allows independent control of the weight of external sensory input and ongoing internal communication processes29. Modulation of 5-HT2A receptor contribution54 may permit flexible segregation55 and integration56 of ongoing activity (including top-down feedback57,58) to achieve context-dependent scaling of input. This also supports the notion that these functions are sensitive and prone to malfunction when imbalances occur in the distribution or activation of 5-HT2A receptors across neuron types59,60,61,62. Altogether our results shed light on network mechanisms of gain control by modulatory systems, influencing sensory impact on cortical dynamics, and providing distinct control of various streams of information via GPCRs.
Methods
Animals
Adult male and female heterozygous NEX-Cre63 and homozygous PV-Cre64 mice were used in this study to induce expression of the injected construct in either pyramidal neurons or parvalbumin (PV) interneurons. All experimental procedures were carried out in accordance with the European Union Community Council guidelines and approved by the German Animal Care and Use Committee under the Deutsches Tierschutzgesetz (Az.: 84-02.04.2014.A439, 84-02.04.2016.A138, 84-02.04.2019.A483, 81-02.04.2019.A228, 81-02.04.2021.A412) and the NIH guidelines. Mice were group-housed in standard vivarium conditions (temperature maintained at 20–22 °C, 30–70% humidity). After stereotactic injections (detailed below), mice were housed individually, in a 12 h light/dark cycle with food and water ad libitum.
Generation of plasmid constructs
Adeno-associated viral (AAV) vector construction of pAAV-CMV-floxed-mOpn4L-eGFP-5-HT2ACT was based on pAAV-eGFP-CW3SL (GenBank accession number: KJ411916.2) as a backbone plasmid, in which loxP and lox2272 were introduced to create a conditional construct. Sequences for long isoform mouse melanopsin (mOpn4L35, GenBank accession number: NM_013887.2), eGFP, and the C-terminus of human 5-hydroxytryptamine receptor 2A (5-HT2A, amino acids K385-V471, GenBank accession number: NM_000621.4) were PCR amplified with 16 base pair overhangs and inserted into the backbone plasmid using In-Fusion cloning kit (Takara).
Viral vector production
AAV stocks of serotype 2/9 were produced using the AAV helper-free system (Agilent Technologies, Santa Clara, CA). Three plasmids encoding (1) the DNA sequence of interest flanked by two inverted terminal repeats (ITRs), (2) the helper plasmid encoding AAV rep and cap genes, and (3) adenoviral helper plasmid were co-transfected in HEK293T cells using polyethyleneimine as a transfection reagent. After 72 h incubation, HEK293T cells and culture medium were harvested and centrifuged at low speed (300 × g, 10 min, 4 °C). Cells were resuspended in lysis buffer (150 mM NaCl, Tris–Cl pH 8.5) while the supernatant was incubated with 40% PEG-8000 at 4 °C on a shaker. Cells were lysed by 5 freeze-thaw cycles and incubated with DNase and MgCl2 (30 min, 37 °C). Following centrifugation (3700 × g, 20 min, 4 °C), the supernatant from the cell lysate was used to resuspend PEG-8000 precipitates. For AAV purification, the resuspension was incubated with 40% PEG-8000 for 2 h, centrifuged (3700 × g, 20 min, 4 °C), the supernatant discarded, and the pellet resuspended in 10 ml 50 mM HEPES buffer. Room-temperature chloroform (1:1 volume) was added, the mixture thoroughly vortexed for 2 min, and centrifuged (370 × g, 5 min, RT). Finally, the aqueous phase was collected, sterile filtered (0.22 µm syringe filter membrane), and concentrated by incubation with 40% PEG-8000 for 2 h and subsequent centrifugation (3700 × g, 20 min, 4 °C). The resulting AAV stock was resuspended in 100 µl of 1× PBS with 0.001% Pluronic F68, aliquoted, and stored at −80 °C.
Viral construct and stereotactic injections
The viral solution of AAV containing the 5-HT2A receptor construct, pAAV-CMV-floxed-mOpn4L-eGFP-5-HT2ACT, was injected in the mouse primary visual cortex. In order to express the construct simultaneously in both PV and pyramidal neurons, a mixed viral solution containing pAAV-CMV-floxed-mOpn4L-eGFP-5-HT2ACT and pAAV-CamKII(0.4)-mOpn4L-eGFP-5-HT2ACT was used and injections were performed in PV-Cre mice (Extended Data Fig. 15). Animals were anaesthetized with isoflurane (3% induction and 1–1.5% maintenance) via a nose mask. Animals were placed in the stereotactic frame on a heating pad (38 °C) and injected subcutaneously with 0.2–0.25 ml NaCl (0.9%) solution containing buprenorphine (10 µg/ml; 0.1 µg/g bodyweight) and atropine (5 µg/ml; 0.05 µg/g bodyweight). Lidocaine cream (2%) was applied on the scalp for local anesthesia. After the scalp was cut with a circular incision, the edges of the skin were glued to the sides of the skull with histoacryl glue. The dorsal part of the skull was thinned using a microdrill until the vascular structure was clearly visible. Two small craniotomies (~100 µm diameter) were made in each hemisphere above the visual cortex based on stereotactic coordinates (from Bregma: 4.2 mm posterior and 2.5 mm lateral; 3.5 mm posterior and 2 mm lateral). These coordinates were further adjusted beforehand for each animal, by multiplying them with a factor calculated as the ratio between the animal’s Bregma-Lambda distance and the default Bregma-Lambda distance (4.2 mm). Using a micromanipulator and a glass pipette coupled to a syringe via a silicone tube, the virus was injected into the craniotomies in three steps, at depths of 600, 400, and 200 µm, with 10 min breaks between the steps. After the injections, the skull was covered with transparent dental cement (Super-Bond C&B Set) for protection and a transparent nail polish layer to reduce light scattering. A head holder was attached using dental cement to provide stable mounting of the animals during experiments. A green silicon elastomer (Kwik-Cast™, WPI) was placed above the nail polish layer to avoid photoactivation of the optogenetic probe by ambient light.
Visual stimuli and photostimulation
A monitor (60 Hz, mean luminance 55 cd/m2, LG 24BK55WV-B) was placed 30 cm away from the stimulated eye, covering ~56 × 70 deg of the visual field. A semipermeable zero-power contact lens was used to prevent the eye from drying out. The non-stimulated eye was covered with eye cream (Visco-Vision gel, OmniVision®) and aluminum foil to prevent any input by ambient light. Each experiment consisted of 10–25 trials and included four types of conditions, presented in a block-randomized order: (1) S – spontaneous condition, during which a uniform isoluminant gray screen was presented, (2) V – visually evoked condition (control), during which moving gratings were presented, (3) Sph – spontaneous condition, during photostimulation, inducing 5-HT2A receptor signaling, and (4) Vph – visually evoked condition during photostimulation of 5-HT2A receptors. Visual stimuli consisted of vertical or horizontal square-wave gratings (0.05 cycles/deg), moving at 2 cycles/s. Stimuli were repeatedly presented (6 times at a 3 s inter-stimulus interval) for 200 ms. In photostimulation conditions, continuous light (465 nm, applied with LED driver and module system, Plexon) lasted 8 s, which in the Vph condition started 1.8 s after the 1st visual stimulus onset. The time interval between two consecutive trials was 40 s. Light was delivered via an optic fiber connected to the silicon probe (“optrode“, Cambridge NeuroTech). The fiber ended 750 µm above the electrode tip, light power at the fiber tip was 1–1.5 mW. After the photostimulated trials, a continuous yellow light pulse (590 nm, LED driver and module system, Plexon) lasting for 30 s was delivered through the same optic fiber to switch off the opsin.
In vivo extracellular recording
Extracellular single-unit activity (SUA) was recorded using acutely inserted silicon probes with a varying number of recording channels: 4-channel (Q1x1-tet-5mm-121-Q4, NeuroNexus Technologies), 16-, 32- or 64-channel silicon probes (ASSY-1 E-1, ASSY-1 P-1, ASSY-37 E-1, ASSY-37 P-1, ASSY-77 H6, Cambridge NeuroTech) coupled to the optic fiber. Signals were amplified with an amplifier board (RHD2132 or RHD2164, Intan Technologies) and recorded at 20 kHz sampling rate using a multi-channel electrophysiology acquisition board (Open Ephys65). For spike isolation, the signal was filtered between 600 to 6 kHz. Isolated spikes were detected and clustered using the Klusta suite66 (https://github.com/klusta-team/klustakwik) or the Kilosort2 program67,68 with default configuration parameters. Extracellular recordings, visual- and photo-stimulations were synchronized using a Master-8 Pulse Stimulator (A.M.P.I.) and Raspberry Pi devices, and custom-written scripts in Python.
2-Photon calcium imaging
NEX-Cre mice were injected as described above with viruses containing floxed constructs of mOpn4L-mCherry-5-HT2ACT and GCaMP. After 10–14 days the mice were anaesthetized and the brains were extracted. Coronal slices (250 µm thick) including V1 were prepared in a solution containing 87 mM NaCl, 2.5 mM KCl, 0.5 mM CaCl2, 7 mM MgCl2, 1.25 mM NaH2PO4, 25 mM NaHCO3, 10 mM D-glucose and 75 mM sucrose. The solution was bubbled with 95% O2 and 5% CO2 using the VT1000S Vibratome (Leica). All slices were consecutively incubated in 1 µM TTX (Tocris), 1 µM CNQX (Tocris) and 25 µM 9-cis retinal (Sigma) for 1 h at 37 °C in a solution consisting of 125 mM NaCl, 2.5 mM KCl, 2 mM CaCl2, 1 mM MgSO4, 1.25 mM NaH2PO4, 26 mM NaHCO3, and 20 mM D-glucose, bubbled with 95% O2 and 5% CO2. Finally, the PLC antagonist U73122 (10 µM) was added to the bath solution to block the Gq pathway. Calcium imaging was performed in 3 slices with a 2-photon imaging system (Bruker) using a 965 nm tunable laser and with 16× objective and 1.6× optical zoom. Prior to the recording of each condition (3 min), a 1100 nm tunable laser was used to set the focus level and to screen for mOpn4L-mCherry-5-HT2A expression. To exclude initial melanopsin activation, each slice was incubated for 10 min after setting the focus level. Calcium signals were detected using photomultipliers and fluorescent levels were quantified using the ImageJ Fiji software and the Time Series Analyzer V3 Plugin. For each cell, the fluorescent trace was normalized to its respective starting value (F0), and mean ± SEM was calculated over all conditions and across all cells. Amplitudes of the calcium signals were quantified between 7.2 s and 7.9 s after recording onset (around the time when maximal amplitudes were observed).
Cells were categorized by visual inspection as “inactive” if no response occurred during the recording (180 s), as “spontaneously active” if the activity was not directly locked to photoactivation, determined by activity occurring later than 20 s or as “light-activated” if the activity occurred within the first 20 s of recordings. For each condition, the fraction of cells that was activated by light (<20 s) and cells that were not photoactivated was quantified (Extended Data Fig. 1).
Hek tsA 201 cells were seeded in 35 mm culture dishes 48 h and transfected 24 h prior to the experiment with either Opn4L-5-HT2ACT or native 5-HT2A receptor co-transfected with GCamP6m respectively and kept in the dark at all times. The Imaging was performed in regular DMEM high glucose medium (Gibco). For PLC deactivation experiments the dishes were preincubated with 10 µM of the PLC antagonist U73122 for 30 min. Calcium imaging was performed with the same 2-photon setup (Bruker) using the 965 nm tunable laser, a 40× objective, and no further optical zoom. Focusing cell surface levels beforehand was done using red light to exclude prior melanopsin activation. To initialize 5-HT2A receptor signaling 50 µM serotonin was applied after 30 s. Data Analysis was performed using the Fiji Time Series Analyzer Plugin. Absolute fluorescent values were measured for each cell, baseline values subtracted and normalized (Extended Data Fig. 2d).
GsX assay
Hek tsa 201 cells were seeded into Poly-L lysin coated 96-well plates 48 h and co-transfected with the GloSensor (ProMega), Opn4L-5HT2ACT or 5-HT2A receptor, and the respective G protein chimera for Gs, Gi, Go or Gq 24 h prior to the GsX assay69,70 (transfection ratio 100:100:1). 1 h before the experiment cell culture medium is replaced by L-15 medium containing 2 mM beetle luciferin (and 1 µM 9-cis retinal in case of the Opn4L-5HT2ACT assay). Luminescence recordings were performed using a PerkinElmer 2030 multilabel reader VICTOR X3. After 160 s baseline recordings, a stimulation period of 5 s blue LED light stimulation (Opn4L-5HT2ACT) or application of 20 µM Ro 60-175 Fumarat (5-HT2A receptor) followed. Subsequently, luminescence was recorded every ~40 s for 1300 s. For Data Analysis baseline values were subtracted to correct for differences in expression levels. Mean fluorescence values were normalized to the maximal luminescent increase for each condition, Opn4L-5HT2ACT and 5-HT2A receptor, respectively (Extended Data Fig. 2a–c).
Immunohistochemistry
To check the specificity of the virus expression, the animals were perfused at the end of each experiment and immunohistochemical stainings were performed. The animals were anaesthetized with a mixture of Ketamine (0.1 mg/g bodyweight) and Xylazine (0.01 mg/g bodyweight) injected intraperitoneally (i.p.). After the exclusion of any signs of pain, transcardial perfusion of PBS (pH = 7.4) followed by 4% PFA was performed. The brains were removed, fixed overnight at 4 °C in 4% PFA, and cryoprotected in PBS solution with 30% sucrose. The brains were frozen in Tissue-Tek (O.C.T.™ Compound) and sectioned using a cryotome (Leica CM3050 S). Coronal slices (30 µm thick) were immersed in TBS on a 24-well plate (CytoOne®) for further immunohistochemical staining. All the incubation and rinsing steps took place on an orbital shaker and at room-temperature unless stated otherwise. Free-floating sections were rinsed 3 times (10 min) with TBS. Sections were incubated in a blocking solution containing 0.3% Triton-X-100 and 10% normal donkey serum (NDS) in TBS for 1 h, followed by primary antibody (1:300 rabbit anti-GluR2/3, 1:500 rabbit anti-GABA or 1:1000 mouse anti-PV) incubation in TBS containing 0.3% Triton-X-100 and 5% NDS for 1 day at 4 °C. After 3 steps of rinsing with TBS (10 min), the sections were incubated in the secondary antibody solution (1:500 donkey anti-rabbit DyLight650 and 1:500 donkey anti-mouse DyLight550 in 0.3% Triton-X-100 and 5% NDS in TBS) for 1.5 h while protected from light. The sections were then rinsed 3 times with TBS (10 min), mounted onto microscope glass slides (SuperFrost/Plus, Thermo Scientific), and covered with mounting solution (Roti®-Mount FluorCare, Carl Roth) and a glass coverslip. Digital images of brain sections were acquired sequentially for each fluorophore channel using a confocal laser scanning microscope (Leica TCS SP5, DMI6000 B) with a magnifying objective (10×/0.3 NA, 2× digital zoom) and coupled to a personal computer, running Leica Application Suite Advanced Fluorescence software (LAS AF 2.6). The images were further processed and/or overlaid using ImageJ (version 1.51j871).
Data analysis
Preprocessing and analysis of the data was performed with custom-written Matlab (R2020a, Mathworks) scripts. Based on the trough-to-peak time (Extended Data Fig. 16), units were classified as putative inhibitory (<0.5 ms) or putative excitatory units (>0.5 ms)47,72. Spike counts were binned to 200 ms time frames and the resulting traces were averaged across trials for each recorded condition. To avoid floor effects, units with a baseline firing rate of <0.5 Hz were excluded from the analysis.
To normalize spontaneous activity (\(S,\,{S}_{{ph}}\) in Fig. 2), the spontaneous average firing rate was calculated over 3 s after the onset of recording in individual traces during the S condition, and traces with and without photostimulation were divided by this average firing rate. To normalize the evoked responses (\({V}_{{ph}},\,{V}_{{ph}}\) in Fig. 3), the average spiking rate over 1 s before the first visual stimulus was subtracted from each trace. For further normalization of the stimulus-evoked conditions, traces were divided by the first stimulus response amplitude (i.e., 600 ms time interval after the visual stimulus onset) of the control condition (\(V\)). The resulting normalized traces were then averaged over all units.
To normalize the photostimulation effects, the opto-index (\({OI}\)) was calculated for the \({S}_{{ph}}\) trace (Fig. 2):
where \({{PRE}}_{{Firing}}\) represents the firing rate of a unit before photostimulation (average over the 3 s interval after the start of the trial) and \({{POST}}_{{Firing}}\) represents the firing rate during photostimulation (average over the last 3 s interval of the photostimulation interval). Therefore, \({OI}=1\) indicates infinite increase in activity from the baseline, \({OI}=0\) means no change, while \({OI}=-1\) indicates complete suppression of the activity.
To quantify the magnitude of visually evoked responses, the difference between response amplitude (average firing within 600 ms after stimulus onset) and the baseline (average firing within 1 s before stimulus onset) was calculated based on each unit’s normalized (Fig. 3e) or raw firing rate (Fig. 3f).
For the linear regression (Fig. 3f), the following equation was used:
Where \({r}_{{pre}}\) and \({r}_{{post}}\) are the magnitude firing rates before and during photostimulation, respectively, normalized to the unit with the highest, ph represents the presence of photostimulation and is 0 for the V condition and 1 for the Vph condition and \({\beta }_{1-4}\) are the fitting coefficients of the linear regression model. Therefore, \({\beta }_{3}\) and \({\beta }_{4}\) represent the effect of photostimulation on the intercept and the slope of the control, respectively, and can point out to a subtractive or divisive effect, respectively.
Putative monosynaptic connections between units were identified based on a fast and transient increase (excitatory connection) or decrease (inhibitory connection), occurring in the spike cross-correlogram of 2 units with ~2 ms delay73,74.
For local field potential analysis, the raw signal was downsampled at 1 kHz and low-pass filtered at 300 Hz. Signals of all channels were processed using a fast Fourier transform (FFT) of the photostimulated time interval, averaged over all trials and channels for each recording, and then normalized to their peak value. The spectrogram was calculated by computing the short-time Fourier transform of the signal. The normalized single-sided amplitude spectrum and the power/frequency were quantified by averaging over different frequency bands (slow waves: 0–1 Hz, delta: 1–4 Hz, theta: 4–8 Hz, alpha: 8–12 Hz, beta: 12–30 Hz, gamma: 30–70 Hz).
Spiking network model
The spiking neural network model consisted of conductance-based excitatory and inhibitory neurons. It was based on the network developed by Sadeh et al. 36 with two changes. Simple integrate-and-fire neurons were used instead of exponential integrate-and-fire neurons and the size of the network was reduced by a factor of 10. While our physiological experiments show indications of putative non-linear effects of lateral interactions75,76,77,78,79, these were not explicitly implemented in our model to keep the parameter ensemble as simple as possible. This may underestimate the strength of 5-HT2A-mediated synaptic effects when modeling its separate experimental activation in local subpopulations80.
In the neuron model, the membrane potential \({V}_{m}\) was described by the following first-order differential equation:
where \(C\) is the membrane capacitance, \({g}_{L}\) is the leak conductance and \({E}_{L}\) is the resting potential. When the membrane potential \({V}_{m}\) reached \({E}_{T}\), a spike was generated and the membrane potential was reset to \({V}_{r}\).
The model neurons were driven via excitatory and inhibitory synaptic inputs represented in the second and third terms, respectively. In these terms \({E}_{e}\) is the reversal potential of excitatory and \({E}_{i}\) is the reversal potential of inhibitory conductances. \({G}_{e}(t)\) and \({G}_{i}(t)\) are the total excitatory and inhibitory conductances at time \(t\), respectively, defined as:
where \({t}_{j}\) and \({t}_{k}\) are the times when excitatory and inhibitory synaptic events reach a particular post-synaptic neuron, respectively. The conductances \({g}_{e}\) and \({g}_{i}\) map the timing of synaptic events onto changes in the membrane conductance based on the following
\(H(t)\) is the Heaviside step function
\({J}_{e}\) denotes the peak excitatory conductance and \({J}_{i}\) denotes the peak inhibitory conductance. \({\tau }_{e}\) and \({\tau }_{i}\) are the synaptic time constants of excitatory and inhibitory conductances, respectively.
The network model consists of excitatory (e) and inhibitory (i) subpopulations of size \({N}_{e}=160\) and \({N}_{i}=40\), respectively, that are randomly connected using a binomial distribution with a mean connection probability of \({CP}\). The peak conductance of each connection is random and drawn from a Gaussian distribution with the mean \(J\) and the standard deviation \(J/5\). To be consistent with Dale’s law, any negative weights were set to zero.
The neurons received background excitatory and inhibitory inputs from independent homogeneous Poisson spike trains firing at rate \({\nu }_{{{bkg}}_{\left(e,i\right)}}\), included in the sum in Eq. 4 (for excitation) and Eq. 5 (for inhibition). The effect of 5HT2A activation was modeled via additional background excitatory and inhibitory input firing at rates \({\nu }_{{pert}}\) (see Table 2 for parameter values) to different fractions of neurons (note that the parameter values of the fraction of activated units and perturbation strength are in principle interchangeable (Extended Data Fig. 12)). These background inputs are independent homogeneous Poisson spike trains. To account for the different impact of the 5-HT2A receptor-dependent on its localization in pyramidal vs inhibitory neurons (dendrite vs. soma, respectively), the background input was estimated as twice as high in inhibitory units in comparison to excitatory units81,82. We modeled the stimulus-evoked responses of neurons by applying a 500-ms long homogeneous Poisson excitatory spike train firing at \({\nu }_{{stim}}\) as excitatory inputs to all neurons in both populations. Aggregate results were obtained by simulating 10 different instances of the model (different initializations of the connectivity matrix) and each model 10 times with different realizations of the inputs. The outcomes were then averaged across the 100 simulations.
Statistical analysis
Statistical analysis was performed using Matlab. Two-sided paired sample t-test (with Bonferroni correction for multiple comparisons in Fig. 4a), two-sided one-sample t-test (with Bonferroni correction for multiple comparisons in Figs. 3e and 4b), two-sample Kolmogorov–Smirnov test (Figs. 2c, h and 3c, d), two-sided Mann–Whitney U-test (Fig. 1i), one-way ANOVA (Extended Data Figs. 1 and 2) were used to assess the significance of the values (mean ± SEM).
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The data supporting the findings of this study are available within the main text, the Supplementary Information file or from the corresponding author upon request. Source data are provided with this paper.
Code availability
The code underlying the modeling is provided as open-source at https://github.com/sencheng/Serotonergic-Modulation83.
References
Dayan, P. & Abbott, L. F. Theoretical Neuroscience: Computational and Mathematical Modeling of Neural Systems (Massachusetts Institute of Technology Press, 2001).
Deco, G., Tononi, G., Boly, M. & Kringelbach, M. L. Rethinking segregation and integration: contributions of whole-brain modelling. Nat. Rev. Neurosci. 16, 430–439 (2015).
Ferguson, K. A. & Cardin, J. A. Mechanisms underlying gain modulation in the cortex. Nat. Rev. Neurosci. 21, 80–92 (2020).
Wilson, N. R., Runyan, C. A., Wang, F. L. & Sur, M. Division and subtraction by distinct cortical inhibitory networks in vivo. Nature 488, 343–348 (2012).
Lee, S.-H. et al. Activation of specific interneurons improves V1 feature selectivity and visual perception CAMKIIα-Arch Light-on Light-on. Nature 488, 379–383 (2012).
Lee, S.-H., Kwan, A. C. & Dan, Y. Interneuron subtypes and orientation tuning. Nature 508, E1–E2 (2014).
Atallah, B. V. et al. Atallah et al. reply. Nature 508, E3–E3 (2014).
Seybold, B. A., Phillips, E. A. K., Schreiner, C. E. & Hasenstaub, A. R. Inhibitory actions unified by network integration. Neuron 87, 1181–1192 (2015).
Santana, N. & Artigas, F. Laminar and cellular distribution of monoamine receptors in rat medial prefrontal cortex. Front. Neuroanat. 11, 87 (2017).
Shukla, R., Watakabe, A., Yamamori, T. & Mengod, G. mRNA expression profile of serotonin receptor subtypes and distribution of serotonergic terminations in marmoset brain. Front. Neural Circuits 8, 52 (2014).
Araneda, R. & Andrade, R. 5-Hydroxytryptamine2 and 5-hydroxytryptamine1A receptors mediate opposing responses on membrane excitability in rat association cortex. Neuroscience 40, 399–412 (1991).
Puig, M. V. & Gulledge, A. T. Serotonin and prefrontal cortex function: neurons, networks, and circuits. Mol. Neurobiol. 44, 449–464 (2011).
Grandjean, J. et al. A brain-wide functional map of the serotonergic responses to acute stress and fluoxetine. Nat. Commun. 10, 350 (2019).
Hansen, J. Y. et al. Mapping neurotransmitter systems to the structural and functional organization of the human neocortex. Nat. Neurosci. 25, 1569–1581 (2022).
Shimegi, S. et al. Cholinergic and serotonergic modulation of visual information processing in monkey V1. J. Physiol. Paris 110, 44–51 (2016).
Colangelo, C., Shichkova, P., Keller, D., Markram, H. & Ramaswamy, S. Cellular, synaptic and network effects of acetylcholine in the neocortex. Front. Neural Circuits 13, 24 (2019).
Gulledge, A. T., Park, S. B., Kawaguchi, Y. & Stuart, G. J. Heterogeneity of phasic cholinergic signaling in neocortical neurons. J. Neurophysiol. 97, 2215–2229 (2007).
Jimenez-Martin, J., Potapov, D., Potapov, K., Knöpfel, T. & Empson, R. M. Cholinergic modulation of sensory processing in awake mouse cortex. Sci. Rep. 11, 17525 (2021).
Pinto, L. et al. Fast modulation of visual perception by basal forebrain cholinergic neurons. Nat. Neurosci. 16, 1857–1863 (2013).
Eickelbeck, D. et al. CaMello-XR enables visualization and optogenetic control of Gq/11 signals and receptor trafficking in GPCR-specific domains. Commun. Biol. 2, 60 (2019).
Spoida, K., Masseck, O. A., Deneris, E. S. & Herlitze, S. Gq/5-HT2c receptor signals activate a local GABAergic inhibitory feedback circuit to modulate serotonergic firing and anxiety in mice. Proc. Natl Acad. Sci. USA 111, 6479–6484 (2014).
Hasegawa, E. et al. Serotonin neurons in the dorsal raphe mediate the anticataplectic action of orexin neurons by reducing amygdala activity. Proc. Natl Acad. Sci. USA 114, E3526–E3535 (2017).
Masseck, O. A. et al. Vertebrate cone opsins enable sustained and highly sensitive rapid control of Gi/o signaling in anxiety circuitry. Neuron 81, 1263–1273 (2014).
Marcus, D. J. & Bruchas, M. R. Optical approaches for investigating neuromodulation and G protein–coupled receptor signaling. Pharmacol. Rev. 75, 1119–1139 (2023).
Oh, E., Maejima, T., Liu, C., Deneris, E. & Herlitze, S. Substitution of 5-HT1A receptor signaling by a light-activated G protein-coupled receptor. J. Biol. Chem. 285, 30825–30836 (2010).
Waterhouse, B. D., Ausim Azizi, S., Burne, R. A. & Woodward, D. J. Modulation of rat cortical area 17 neuronal responses to moving visual stimuli during norepinephrine and serotonin microiontophoresis. Brain Res. 514, 276–292 (1990).
Watakabe, A. et al. Enriched expression of serotonin 1B and 2A receptor genes in macaque visual cortex and their bidirectional modulatory effects on neuronal responses. Cereb. Cortex 19, 1915–1928 (2009).
Seillier, L. et al. Serotonin decreases the gain of visual responses in awake macaque V1. J. Neurosci. 37, 11390–11405 (2017).
Azimi, Z. et al. Separable gain control of ongoing and evoked activity in the visual cortex by serotonergic input. eLife 9, e53552 (2020).
Azimi, Z. et al. Subtraction and division of visual cortical population responses by the serotonergic system. Preprint at bioRxiv https://doi.org/10.1101/444943 (2018).
Michaiel, A. M., Parker, P. R. L. & Niell, C. M. A hallucinogenic serotonin-2A receptor agonist reduces visual response gain and alters temporal dynamics in mouse V1. Cell Rep. 26, 3475–3483.e4 (2019).
Hannon, J. & Hoyer, D. Molecular biology of 5-HT receptors. Behav. Brain Res. 195, 198–213 (2008).
Carandini, M. & Heeger, D. J. Summation and division by neurons in primate visual cortex. Science 264, 1333–1336 (1994).
Salinas, E. & Thier, P. Gain Modulation: a major computational principle of the central nervous system. Neuron 27, 15–21 (2000).
Spoida, K. et al. Melanopsin variants as intrinsic optogenetic on and off switches for transient versus sustained activation of G protein pathways. Curr. Biol. 26, 1206–1212 (2016).
Sadeh, S., Silver, R. A., Mrsic-Flogel, T. D. & Muir, D. R. Assessing the role of inhibition in stabilizing neocortical networks requires large-scale perturbation of the inhibitory population. J. Neurosci. 37, 12050–12067 (2017).
Kuhn, A., Aertsen, A. & Rotter, S. Neuronal integration of synaptic input in the fluctuation-driven regime. J. Neurosci. 24, 2345–2356 (2004).
Rosenbaum, R. & Doiron, B. Balanced networks of spiking neurons with spatially dependent recurrent connections. Phys. Rev. X 4, 021039 (2014).
van Vreeswijk, C. & Sompolinsky, H. Chaos in neuronal networks with balanced excitatory and inhibitory activity. Science 274, 1724–1726 (1996).
Moreau, A. W., Amar, M., le Roux, N., Morel, N. & Fossier, P. Serotoninergic fine-tuning of the excitation–inhibition balance in rat visual cortical networks. Cereb. Cortex 20, 456–467 (2010).
Wilkinson, L., Auerbach, S. & Jacobs, B. Extracellular serotonin levels change with behavioral state but not with pyrogen-induced hyperthermia. J. Neurosci. 11, 2732–2741 (1991).
Tyree, S. Optogenetic investigation of arousal circuits. Int. J. Mol. Sci. 18, 1773 (2017).
Borg-Graham, L. J., Monier, C. & Fregnac, Y. Visual input evokes transient and strong shunting inhibition in visual cortical neurons. Nature 393, 369–373 (1998).
Lőrincz, M. L. & Adamantidis, A. R. Monoaminergic control of brain states and sensory processing: Existing knowledge and recent insights obtained with optogenetics. Prog. Neurobiol. 151, 237–253 (2017).
Celada, P., Puig, M. V. & Artigas, F. Serotonin modulation of cortical neurons and networks. Front. Integr. Neurosci. 7, 25 (2013).
Weber, E. T. & Andrade, R. Htr2a gene and 5-HT2A receptor expression in the cerebral cortex studied using genetically modified mice. Front. Neurosci. 4, 36 (2010).
de Filippo, R. et al. Somatostatin interneurons activated by 5-HT2A receptor suppress slow oscillations in medial entorhinal cortex. eLife 10, e66960 (2021).
Avesar, D. & Gulledge, A. T. Selective serotonergic excitation of callosal projection neurons. Front. Neural Circuits 6, 12 (2012).
Stephens, E. K., Avesar, D. & Gulledge, A. T. Activity-dependent serotonergic excitation of callosal projection neurons in the mouse prefrontal cortex. Front. Neural Circuits 8, 97 (2014).
Mukaida, K., Shichino, T., Koyanagi, S., Himukashi, S. & Fukuda, K. Activity of the serotonergic system during isoflurane anesthesia. Anesth. Analg. 104, 836–839 (2007).
Ma, H. et al. The states of different 5-HT receptors located in the dorsal raphe nucleus are crucial for regulating the awakening during general anesthesia. Mol. Neurobiol. https://doi.org/10.1007/s12035-023-03519-0 (2023).
Abbott, L. F. & Chance, F. S. Drivers and modulators from push-pull and balanced synaptic input. Prog. Brain Res. 149, 147–155 (2005).
Chance, F. S., Abbott, L. F. & Reyes, A. D. Gain modulation from background synaptic input. Neuron 35, 773–782 (2002).
Kringelbach, M. L. et al. Dynamic coupling of whole-brain neuronal and neurotransmitter systems. Proc. Natl Acad. Sci. USA 117, 9566–9576 (2020).
Lee, S., Park, J. & Smirnakis, S. M. Internal gain modulations, but not changes in stimulus contrast, preserve the neural code. J. Neurosci. 39, 2012–2018 (2019).
Sadaghiani, S., Hesselmann, G., Friston, K. J. & Kleinschmidt, A. The relation of ongoing brain activity, evoked neural responses, and cognition. Front. Syst. Neurosci. 4, 20 (2010).
Naumann, L. B., Keijser, J. & Sprekeler, H. Invariant neural subspaces maintained by feedback modulation. eLife 11, e76096 (2022).
Pollak Dorocic, I. et al. A whole-brain atlas of inputs to serotonergic neurons of the dorsal and median raphe nuclei. Neuron 83, 663–678 (2014).
Kringelbach, M. L. & Deco, G. Brain states and transitions: insights from computational neuroscience. Cell Rep. 32, 108128 (2020).
Preller, K. H. et al. Effective connectivity changes in LSD-induced altered states of consciousness in humans. Proc. Natl Acad. Sci. USA 116, 2743–2748 (2019).
Grandjean, J. et al. Psilocybin exerts distinct effects on resting state networks associated with serotonin and dopamine in mice. Neuroimage 225, 117456 (2021).
Urban, D. J. et al. Elucidation of the behavioral program and neuronal network encoded by dorsal raphe serotonergic neurons. Neuropsychopharmacology 41, 1404–1415 (2015).
Goebbels, S. et al. Genetic targeting of principal neurons in neocortex and hippocampus of NEX-Cre mice. Genesis 44, 611–621 (2006).
Hippenmeyer, S. et al. A developmental switch in the response of DRG neurons to ETS transcription factor signaling. PLoS Biol. 3, e159 (2005).
Siegle, J. H. et al. Open Ephys: an open-source, plugin-based platform for multichannel electrophysiology. J. Neural Eng. 14, 045003 (2017).
Rossant, C. et al. Spike sorting for large, dense electrode arrays. Nat. Neurosci. 19, 634–641 (2016).
Pachitariu, M., Steinmetz, N. A., Kadir, S., Carandini, M. & Harris, K. D. Fast and accurate spike sorting of high-channel count probes with Kilosort. Adv. Neural. Inf. Process Syst. 4448–4456 (2016).
Stringer, C. et al. Spontaneous behaviors drive multidimensional, brainwide activity. Science 364, eaav7893 (2019).
Ballister, E. R., Rodgers, J., Martial, F. & Lucas, R. J. A live cell assay of GPCR coupling allows identification of optogenetic tools for controlling Go and Gi signaling. BMC Biol. 16, 10 (2018).
Zhou, F. et al. Optimized design and in vivo application of optogenetically functionalized Drosophila dopamine receptors. Nat. Commun. 14, 8434 (2023).
Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods https://doi.org/10.1038/nmeth.2089 (2012).
Senzai, Y., Fernandez-Ruiz, A. & Buzsáki, G. Layer-specific physiological features and interlaminar interactions in the primary visual cortex of the mouse. Neuron 101, 500–513.e5 (2019).
English, D. F. et al. Pyramidal cell-interneuron circuit architecture and dynamics in hippocampal networks. Neuron 96, 505–520.e7 (2017).
Petersen, P. C. & Buzsáki, G. Cooling of medial septum reveals theta phase lag coordination of hippocampal cell assemblies. Neuron 107, 731–744.e3 (2020).
Bush, P. C. & Sejnowski, T. J. Effects of inhibition and dendritic saturation in simulated neocortical pyramidal cells. J. Neurophysiol. 71, 2183–2193 (1994).
Häusser, M., Spruston, N. & Stuart, G. J. Diversity and dynamics of dendritic signaling. Science 290, 739–744 (2000).
Holmes, W. R. & Woody, C. D. Effects of uniform and non-uniform synaptic ‘activation-distributions’ on the cable properties of modeled cortical pyramidal neurons. Brain Res. 505, 12–22 (1989).
Koch, C., Douglas, R. & Wehmeier, U. Visibility of synaptically induced conductance changes: theory and simulations of anatomically characterized cortical pyramidal cells. J. Neurosci. 10, 1728–1744 (1990).
Jancke, D. et al. Parametric population representation of retinal location: neuronal interaction dynamics in cat primary visual cortex. J. Neurosci. 19, 9016–9028 (1999).
Ebsch, C. & Rosenbaum, R. Imbalanced amplification: a mechanism of amplification and suppression from local imbalance of excitation and inhibition in cortical circuits. PLoS Comput. Biol. 14, e1006048 (2018).
Breton, J. D. & Stuart, G. J. Somatic and dendritic GABAB receptors regulate neuronal excitability via different mechanisms. J. Neurophysiol. 108, 2810–2818 (2012).
Jadi, M., Polsky, A., Schiller, J. & Mel, B. W. Location-dependent effects of inhibition on local spiking in pyramidal neuron dendrites. PLoS Comput. Biol. 8, e1002550 (2012).
Nejad, M. M., Barzan, R., Jancke, D. & Cheng, S. Gain control of sensory input across polysynaptic circuitries in mouse visual cortex by a single G protein-coupled receptor type (5-HT2A). Zenodo https://doi.org/10.5281/zenodo.12691563 (2024).
Acknowledgements
We thank Stefan Dobers and Henning Knoop and the RUB mechanical shop for technical support, Caroline Naber for assistance with performing GsX-assays, and Callum White for proofreading the manuscript. This work was supported by Deutsche Forschungsgemeinschaft (DFG) grants: Project ID 122679504 - SFB 874, S.C., S.H., D.J.; Project ID 316803389 - SFB 1280, K.S. and S.H. (Subproject A07), M.D.M. (Subproject A21); JA 945/5-1, D.J.; HE 2471/21-1, S.H.; MA 5806/2-1, M.D.M.; MA 5806/1-2, M.D.M., Project number 492434978 - GRK 2862/1, Subprojects (01, S.H.; 05, M.D.M.; 07, K.S.; 09, I.S.; 10, D.J.); BMBF, ERA-Net Neuron “Horizon 2020”, 01EW2104B, D.J.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Contributions
Conceptualization: R.B., Z.A., S.H., and D.J. Data curation: R.B. and D.J. Formal analysis: R.B. and D.J. Electrophysiological data acquisition: R.B. and B.B. Electrophysiological data visualization: R.B. and D.J. Neural model: M.M.N. and S.C. 2-Photon data acquisition, GsX assay, analysis, and visualization: L.R. and I.S. Construction and validation of AAV-virus: T.S., H.B., D.E., M.G., and K.S. Histology: R.B., B.B., K.S., and S.T.S. Software: R.B. and Z.A. Supervision: M.D.M, S.H., and D.J. Funding acquisition and resources: M.D.M., S.H., and D.J. Writing—original draft: R.B. and D.J. Writing—review & editing: R.B., S.C., and D.J.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks Philip Parker, and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Barzan, R., Bozkurt, B., Nejad, M.M. et al. Gain control of sensory input across polysynaptic circuitries in mouse visual cortex by a single G protein-coupled receptor type (5-HT2A). Nat Commun 15, 8078 (2024). https://doi.org/10.1038/s41467-024-51861-1
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-024-51861-1
- Springer Nature Limited