Abstract
Hedonic processing is critical for guiding appropriate behavior, and the infralimbic cortex (IL) is a key neural substrate associated with this function in rodents and humans. We used deep brain in vivo calcium imaging and taste reactivity in freely behaving male and female Sprague Dawley rats to examine whether the infralimbic cortex is involved in encoding innate versus conditioned hedonic states. In experiment 1, we examined the IL neuronal ensemble responsiveness to intraoral innately rewarding (sucrose) versus aversive (quinine) tastants. Most IL neurons responded to either sucrose only or both sucrose and quinine, with fewer neurons selectively processing quinine. Among neurons that responded to both stimuli, some appear to encode hedonic processing. In experiment 2, we examined how IL neurons process devalued sucrose using conditioned taste aversion (CTA). We found that neurons that responded exclusively to sucrose were disengaged while additional quinine-exclusive neurons were recruited. Moreover, tastant-specific neurons that did not change their neuronal activity after CTA appeared to encode objective hedonic value. However, other neuronal ensembles responded to both tastants and appear to encode distinct aspects of hedonic processing. Specifically, some neurons responded differently to quinine and sucrose and shifted from appetitive-like to aversive-like activity after CTA, thus encoding the subjective hedonic value of the stimulus. Conversely, neurons that responded similarly to both tastants were heightened after CTA. Our findings show dynamic shifts in IL ensembles encoding devalued sucrose and support a role for parallel processing of objective and subjective hedonic value.
SIGNIFICANCE STATEMENT Disrupted affective processing contributes to psychiatric disorders including depression, substance use disorder, and schizophrenia. We assessed how the infralimbic cortex, a key neural substrate involved in affect generation and affect regulation, processes innate and learned hedonic states using deep brain in vivo calcium imaging in freely behaving rats. We report that unique infralimbic cortex ensembles encode stimulus subjective and objective hedonic value. Further, our findings support similarities and differences in innate versus learned negative affective states. This study provides insight into the neural mechanisms underlying affect generation and helps to establish a foundation for the development of novel treatment strategies to reduce negative affective states that arise in many psychiatric disorders.
Introduction
Neural processing of hedonic value plays a critical role in guiding behavior (Young, 1952, 1966; Kringelbach and Berridge, 2017), and the neural mechanisms underlying basic hedonic processing are thought to contribute to complex affective states (Eisenberger et al., 2003; Kringelbach and Berridge, 2017). Aberrant hedonic processing is a key component of many psychiatric disorders including depression, substance use disorder, and schizophrenia (Koob and Le Moal, 1997; Simon et al., 2010; Der-Avakian and Markou, 2012). Notably, despair in the United States has contributed to a rise in deaths because of drug overdose and suicide (Case and Deaton, 2017; Woolf and Schoomaker, 2019), which has been exacerbated by the covid-19 pandemic (Ganesan et al., 2021). One approach to help reduce drug overdose and suicide deaths is to provide treatments that help alleviate negative emotions. To do so, it is critical to understand the neural mechanisms underlying affect generation that mediate hedonic processing, to provide a foundation for the development of new treatment strategies to reduce negative affective states evident in many psychiatric disorders.
Taste reactivity (TR) is a preclinical model used to assess hedonic processing (Grill and Norgren, 1978). In the TR model, tastants are infused directly into the oral cavity of the rats, and in response they display a set of behaviors that reflects the subjective hedonic value of the tastant. Sweet sucrose elicits appetitive TR, in which rats display licking behavior. In contrast, bitter quinine evokes aversive TR, such as gaping and mouth-to-floor-wipes; behaviors that help remove tastants from the mouth (Grill and Norgren, 1978). Further, conditioned taste aversion (CTA) can be used to shift the hedonic value of sweet solutions by pairing them with malaise-inducing lithium chloride (LiCl; Domjan, 1977). Naive rats display appetitive TR to sucrose, but after pairing the sweet with LiCl, sucrose instead evokes aversive TR (Spector et al., 1988; Flynn et al., 1991b; Spector et al., 1992). CTA provides an approach to understand the neural mechanisms involved in the development of conditioned negative affect during learned taste aversion, while sucrose and quinine provide insight into innate mechanisms of reward and aversion, respectively.
Using single-unit electrophysiology and TR, our laboratory previously found that nucleus accumbens (NAc) neurons tend to exhibit opposite activity patterns in response to rewarding and aversive tastants. Specifically, rewarding tastants predominantly inhibit NAc neurons, while aversive tastants tend to excite them (Roitman et al., 2005, 2010). The NAc is one component of a broad circuitry mediating hedonic processing (Berridge and Kringelbach, 2015; Kringelbach and Berridge, 2017). In humans, the ventromedial prefrontal cortex (vmPFC) has been linked to affective processing (Delgado et al., 2016; Hiser and Koenigs, 2018). Specifically, self-reported negative affect correlates with the degree of vmPFC activation (Zald et al., 2002) as does psychological stress (Ginty et al., 2019; Orem et al., 2019). The rat homolog of the vmPFC is the infralimbic cortex (IL), which has also been demonstrated to be a key area in the prefrontal cortex that functions to modulate reward. The IL projects to the NAc shell (Takagishi and Chiba, 1991; Brog et al., 1993; Vertes, 2004), and it has been implicated in flexible reward seeking and top-down control of appetitive and aversive behaviors (Barker et al., 2013; Richard and Berridge, 2013; Hurley and Carelli, 2020). However, how IL neuronal ensembles process hedonic value, both objective and subjective, is currently unclear.
We examined IL neuronal ensemble processing of rewarding and aversive tastants using in vivo deep brain calcium imaging in freely behaving rats. In experiment 1, we examined how IL neuronal ensembles respond to infusions of rewarding sucrose and innately aversive quinine. In experiment 2, we further characterized IL neuronal ensemble processing of innate and conditioned aversion by comparing IL response to sucrose conditioned to become aversive through CTA learning to the innately aversive quinine. Collectively, our results implicate a dynamic and complex role for parallel processing of hedonic value in the IL.
Materials and Methods
Subjects.
Male and female Sprague Dawley rats (Envigo) weighing 200–300 g on arrival were used. Rats were singly housed in a temperature- and humidity-controlled room on a 12 h reverse light/dark cycle (all testing occurred during the dark phase). Rats had ad libitum access to chow (Tekland 2920×, Envigo) and water except during behavioral testing where water was restricted to 20–25 ml/d. All protocols were conducted in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals and were approved by the University of North Carolina, Chapel Hill, Institutional Animal Care and Use Committee.
Materials.
The adenovirus AAV5-hSyn-GCaMPs was obtained from the University of Pennsylvania Vector Core. Supplies for calcium imaging recording were purchased from Inscopix. In experiments 1 and 2, 7.1 × 0.6 mm (length × width) and 9.0 × 1.0 mm, respectively, gradient-index (GRIN) lenses were used. Sucrose (0.5 m; Thermo Fisher Scientific) and quinine (0.05 mm; Sigma-Aldrich) were used for tastant infusions. LiCl was used to induce malaise in CTA experiments (127 mg/kg 0.3 m LiCl, i.p.).
Calcium imaging surgeries.
Calcium imaging methods were modified from a previous report (Resendez et al., 2016). Briefly, rats were anesthetized with a mixture of ketamine and xylazine (100 mg/kg ketamine, 10 mg/kg xylazine) and received a 700 nl microinjection of AAV5-hSyn-GCaMPs (1.75 × 1012 viral genomes/ml) in the IL (AP, +2.7; ML, −0.5; DV, −4.6) at a rate of 100 nl/min using a Nanofil syringe (World Precision Instruments). The injector remained for 10 min to allow virus spread. One month after virus injection, rats underwent a second surgery to implant a GRIN lens (Inscopix) and an intraoral (IO) catheter. Rats were anesthetized with ketamine/xylazine, and an IO catheter was implanted as described previously (Roitman et al., 2005; Wheeler et al., 2008). The skull was exposed, skull screws were implanted, and a trephine was used to drill a cranial window above the IL. A GRIN lens was then carefully lowered at a rate of 0.1 mm/min until 200 µm above the IL injection site. Dental acrylic was used to cement the lens in place. Rats received meloxicam (1 mg/kg/d, s.c.) for 3 d after virus and GRIN lens implantation to control postoperative pain and inflammation and enrofloxacin (5 mg/kg, s.c., twice per day) for 5 d to control infection associated with intraoral catheter implantation. Rats were allowed 2 months to recover from lens implantation for inflammation to clear beneath the lens. Rats then received baseplate installation. Rats were anesthetized with isoflurane, a baseplate (Inscopix) was attached to the microendoscope (Inscopix), and the microendoscope was positioned to view GCaMPs-expressing neurons. Once cells were in focus, the baseplate was cemented into place using dental acrylic.
Collection and analysis of calcium signal.
Calcium activity was captured using nVista HD (version 2.0.4; Inscopix). Calcium signaling was recorded at a rate of 12 frames/s with the LED power at the lowest setting that allowed for visualizing calcium activity in each rat (30–70%). Calcium imaging occurred over three imaging blocks with 5 min breaks between each block to prevent photograph bleaching. During the first block (baseline), the calcium signal was captured over a 10 min period where rats moved freely about the testing chamber. During the second and third blocks, rats received 15 IO infusions of sucrose or quinine. The order of infusions of quinine and sucrose were counterbalanced such that half of the rats received quinine in block 2 and sucrose in block 3, and the other half received sucrose before quinine. The start of each calcium imaging session was signaled to the behavioral software system (Med-PC IV, Med Associates). Processed calcium signal and time stamps of the imaging session and tastant infusions were then exported into NeuroExplorer (Plexon) to analyze calcium signaling during IO infusions (trials).
Recordings were decompressed using Inscopix Image Decompressor Software and analyzed by Inscopix Data Processing Software (version 1.3.0). Briefly, recordings were preprocessed and downsampled, run through a spatial filter, motion corrected, and change in fluorescence was identified (ΔF/F). Putative cells were identified using a principal component/independent component analysis. Cells were identified using the auto accept/reject function in Inscopix Data Processing Software and then further validated by the experimenter to ensure that it was spatially selective of a single neuron and expressed a calcium signal indicative of neuronal activity (i.e., a sharp rise to peak signal followed by a gradual decay; Resendez et al., 2016). Rats with <10 recorded neurons were removed from calcium signal analysis, but their behavioral data were included (experiment 1: one female removed from analysis; experiment 2: one male removed from analysis).
Calcium signal fluorescence data were imported to NeuroExplorer (version 5.122), and fluorescence was averaged over IO infusions. Fluorescence data were then exported to Microsoft Excel. The difference in fluorescence from the baseline 10 s period before infusion and 10 s during infusion was calculated. Cells were determined to be phasic to IO infusion if they exhibited a change in calcium fluorescence (>2 SDs from baseline) continuously for ≥1 s during the IO infusion period. Cells were classified as inhibitory if calcium signal decreased and excitatory if signal increased. Peak response was calculated as the average response during the last 2 s of the IO infusion.
Taste reactivity.
Two experiments were completed (described below), and taste reactivity was used in each study to assess the rewarding and aversive qualities of tastants, as described previously (Grill and Norgren, 1978; Berridge, 2000). Each instance of a lateral tongue protrusion, paw lick, or bout of rhythmic tongue protrusions was scored as an appetitive response. Aversive TR consisted of mouth-to-floor-wipes, gaping, and mouth wipes. TR was analyzed as the total number of appetitive and aversive responses along with hedonic score. Hedonic score was calculated as
Experiment 1: IL ensemble processing of innate reward and aversion.
Figure 1 shows the experimental timeline. Rats (n = 4; one male, three female) were habituated to the testing chamber for ≥1 week before testing during which time they were placed into the testing chamber for 2–4 h/d. On the last day of habituation, rats were mildly water deprived for 24 h and then they received IO water infusions to acclimate to IO infusions. The next day, rats remained mildly water deprived (20–25 ml of water/24 h), then were placed in the testing chamber and allowed an hour to acclimate to the chamber, and a microendoscope was attached to their baseplate. Testing occurred over three, 10 min blocks, with a 5 min break between blocks. Calcium signal and TR were recorded during each block. In the first block, rats were allowed a 10 min acclimation period where they moved freely about the test chamber. In the second block, rats received either IO 0.5 m sucrose or 0.05 mm quinine (15 infusions, 10 s infusions, on a variable-interval 30 s schedule, 170 µl/infusion). Notably, we chose a 10 s IO infusion duration based on preliminary data indicating a 10 s duration IO infusion elicited reliable calcium signal responses. In the third block, rats received either IO quinine or sucrose, depending on the tastant they received in block 2. For example, if rats received sucrose in block 2, they received quinine in block 3. The order of sucrose and quinine infusions was counterbalanced across rats such that half of the rats received quinine-sucrose and half received sucrose-quinine.
Schematic timeline of experiments. All rats received IL GCaMP6s microinjection, and lens and baseplate implants 4 and 8 weeks later, respectively. In experiment 1, rats were then habituated to testing procedures over 1 week, received sucrose and quinine infusions (order counterbalanced across rats), and, on test day, TR and calcium signals were recorded during infusions. In experiment 2, rats were habituated to testing procedures and then received a pairing of sucrose and LiCl (CTA learning). Two days later, during the test, they received sucrose and quinine infusions (order counterbalanced across rats), and TR and calcium signals were recorded during infusions.
Experiment 2: IL ensemble processing of an innate versus conditioned aversion.
Figure 1 shows the timeline for experiment 2. Experiment 2 followed a similar protocol as in experiment 1; however, before testing rats received a sucrose devaluation day using CTA.
Here, another group of rats (n = 4, 1 male, three female) received two blocks of IO sucrose infusions (a total of 30, 10 s duration sucrose infusions on a variable interval 30 s schedule, 170 µl/infusion). Immediately after infusions, rats received an injection of LiCl (127 mg/kg 0.3 m LiCl, i.p.) to induce malaise. Two days later, rats were tested in a manner identical to that described in experiment 1.
Histology.
After behavioral testing, rats were heavily anesthetized with ketamine/xylazine (300 mg/kg ketamine, 30 mg/kg xylazine) and perfused with 4% paraformaldehyde. Brains were removed and post fixed in 4% paraformaldehyde for 24 h at 7°C. Then brains were swapped into 20% sucrose in 0.1 m phosphate buffer for at least 24 h at 7°C. Brains were sectioned with a cryostat at 50 µm, mounted onto slides, coverslipped, and visualized using a DFC 45°C wide-field microscope (Leica) or a LSM 800 Confocal Microscope (Zeiss). IL virus expression and lens placement were verified using Paxinos and Watson (2007).
Statistical analyses.
Taste reactivity data and peak calcium response were analyzed using unpaired or paired Student's t tests depending on whether experimental designs were within or between subjects. Comparisons of neuronal population data were analyzed using χ2 tests.
Results
Histology
Virus expression and lens placements are illustrated in Figure 2. Calcium imaging recordings were obtained from rats with optimal GcaMP expression in the IL. Specifically, GCaMP was expressed along the cytoplasm of the cell, indicating that cells did not overexpress or underexpress GCaMP (Fig. 2A; Resendez et al., 2016). All lenses were placed in the right IL hemisphere, above GcaMP-expressing neurons (Fig. 2B). In both experiments, recordings were obtained from four rats (experiment 1: 25, 78, 33, and 38 neurons, respectively; experiment 2: 81, 68, 94, and 63 neurons, respectively). Within each experiment, the number of neurons recorded from each rat were within 2 SDs from the mean, indicating that neuronal calcium data from a single rat were not overly influencing results.
Histologic validation of lens placement and virus expression in the IL. A, Representative histology images displaying a lens (outlined in blue) placed above GCaMP-expressing neurons in the IL (left) and a confocal image of GCaMP-expressing neurons (right). Of note, most neurons exhibited an optimal GCaMP expression pattern (GCaMPs primarily expressed along the neural cytoplasm). B, Histology placements of lenses in experiment 1 and experiment 2 based on coordinates from Paxinos and Watson (2007).
Experiment 1: IL processing of innate rewarding and aversive tastants
In experiment 1, we examined how IL neuronal ensembles process innate rewarding and aversive tastants. To do so, rats were intraorally infused with sucrose and quinine, and neuronal activation was analyzed using calcium imaging. As shown in Figure 3A, the infusion of sucrose and quinine elicited opposite hedonic TR responses, as evidenced by significantly different hedonic scores (t(4) = 112.3, p < 0.0001). Specifically, while sucrose elicited robust appetitive TR, this behavior was nearly absent when quinine was infused (Fig. 3B; t(4) = 4.271, p < 0.05). The opposite pattern was observed when analyzing aversive TR, in which only quinine elicited robust aversive reactions (Fig. 3C; t(4) = 13.63, p < 0.001).
Experiment 1: behavior. A, Mean ± SEM hedonic score, calculated as
Next, we analyzed the response of different neuronal ensembles to the infusion of rewarding sucrose and aversive quinine. We detected nine different clusters of neurons that responded differently to the tastants (Fig. 4A). Examples of excitatory and inhibitory traces (top), peristimulus histograms (middle), and peristimulus response across trials (bottom) elicited by sucrose and quinine are shown in Figure 4B. Figure 4C shows the temporal responses of each cluster to both tastants, showing that some of them responded only to quinine (2, 8), only to sucrose (4, 6), or to both sucrose and quinine (1, 3, 7, and 9). One cluster was nonresponsive to either quinine or sucrose (5). A summary of the responses of each cluster to the tastant and the number of cells is shown in Figure 4D. In total, of the 174 neurons, 74 responded to quinine and 121 to sucrose, while 100 did not respond to quinine and 53 did not respond to sucrose (Fig. 4E). Interestingly, more neurons that responded to quinine were excitatory than inhibitory, while the opposite was observed for sucrose (Fig. 4F). Thus, the inhibitory/excitatory profile was significantly different between quinine and sucrose infusions [x2 (1, N = 195) = 15.78; p < 0.0001], showing that taste stimuli of different hedonic value elicit distinct firing profiles in the IL. Since we tracked the activity of the same cells across tastant conditions, we then analyzed whether these phasic neurons were tastant-specific. To do so, we examined the proportion of neurons that selectively responded to sucrose, quinine, or both tastants (Fig. 4G). Most phasic IL neurons selectively responded to sucrose (47% of phasic neurons) or were phasic to both sucrose and quinine (39% of phasic neurons), but only a small proportion of neurons selectively encoded quinine (14% of phasic neurons). This suggests that, under normal (innate) conditions, fewer IL neurons are dedicated to process aversion while most neurons respond to either a rewarding stimulus or both. We further dissected how each neuron in the ensemble that was phasic to both sucrose and quinine responded to each tastant (Fig. 4H). In this ensemble, a proportion of neurons exhibited opposite responses to sucrose and quinine [21 neurons (38%) inhibited to sucrose and excited to quinine; only 2 neurons (4%) excited to sucrose and inhibited to quinine]. The remaining neurons exhibited identical phasic activity in response to both tastants [14 neurons (25%) excited to both; 18 neurons (33%) inhibited to both]. Collectively, these data suggest that each subgroup of IL neurons encode different aspects of hedonics. To explore this further, we completed experiment 2 to determine how IL neurons respond when there was a learned shift in the palatability of sucrose from appetitive to aversive by using CTA.
Calcium signal in response to quinine and sucrose. A, Heatmap of neuronal response to quinine and sucrose. Nine distinct clusters that responded differently to the tastants were observed. B, Example of excitatory and inhibitory calcium signal (ΔF/F in top panels), perievent histograms (middle panels), and response (z score) across trials for quinine and sucrose. C, Mean ± SEM response (z score) during quinine (red) and sucrose (blue) infusion for each of the 9 clusters. D, Summary table showing the response profile to each tastant and the number of neurons that comprise each cluster. E, Distribution of responsive and nonresponsive (N/R) neurons to quinine and sucrose. F, Inhibitory/excitatory profile. Pie charts show the number (percentage) of phasic IL neurons that exhibited excitatory (red) or inhibitory (blue) calcium activity patterns in response to quinine (left) and sucrose (right). G, Percentage of phasic neurons that responded to sucrose only (blue), quinine only (red), or both tastants (purple). H, Categorization of the IL neuronal ensemble phasic to both sucrose and quinine. Scatter plot of neurons divided according to the average response (z score) to sucrose (y-axis) and quinine (x-axis).
Experiment 2: IL processing of innate versus conditioned aversive tastants
In experiment 2, we compared how IL neuronal ensembles process innate (quinine) versus conditioned aversive taste stimuli; the latter was established by pairing sucrose with an injection of LiCl to produce CTA. The results show that both devalued sucrose and quinine predominantly elicited aversive TR as reflected by nearly equivalent negative hedonic scores (t(4) = 1.514, p = 0.20; Fig. 5A). Both tastants produced negligible levels of appetitive TR (t(4) = 1.581, p = 0.19; Fig. 5B) and robust aversive TR (Fig. 5C). Quinine tended to elicit more aversive TR, although this was not significantly different compared with devalued sucrose (t(4) = 2.427, p = 0.07).
Experiment 2: behavior. A, Mean ± SEM hedonic score, calculated as
We then examined the population response across quinine and devalued sucrose. Again, we detected nine different clusters of neurons that responded differently to the tastants (Fig. 6A). Examples of excitatory and inhibitory traces (top), peristimulus histograms (middle), and peristimulus response across trials (bottom) elicited by devalued sucrose and quinine are shown in Figure 6B. Figure 6C shows the temporal responses of each cluster to each tastant, and again we observed that some of them responded only to quinine (2, 8), only to devalued sucrose (4, 6), or to both devalued sucrose and quinine (1, 3, 7, and 9). One ensemble was nonresponsive to either sucrose or devalued sucrose (5). A summary of the responses of each cluster to the tastant and the number of cells is shown in Figure 6D. Overall, of the 306 recorded neurons, 156 were responsive to quinine and 185 to devalued sucrose, while 150 did not respond quinine, and 121 to devalued sucrose (Fig. 6E). The inhibitory/excitatory neuronal profile depicted in the pie charts in Figure 6F shows that, contrary to rewarding sucrose, devalued sucrose elicited an excitatory response of IL neurons and quinine predominantly elicited excitation of IL neurons. However, these profiles significantly differed [x2 (1, N = 341) = 6.327, p < 0.05], suggesting that, within the IL, innate aversion produces a more robust aversive profile than conditioned aversion. Next, we analyzed whether activation of these same neurons was tastant specific (Fig. 6G). We observed that the greatest proportion of IL neurons encoded both devalued sucrose and quinine (41%), the next largest proportion encoded devalued sucrose alone (35%), and the smallest proportion encoded quinine only (24%). Finally, in Figure 6H we show that within the ensemble that encoded both devalued sucrose and quinine, the largest proportion of neurons exhibited identical responses to both tastants [46 neurons (46%) excited to both; 20 neurons (20%) inhibited to both]. The remaining neurons exhibited opposite firing patterns: 22 neurons (22%) were inhibited to sucrose and excited to quinine, while 11 neurons (12%) were excited to sucrose and inhibited to quinine. Collectively, these results suggest that, although there is an overlap in the encoding of both innate and conditioned aversive tastants in the IL, there is still some divergence between the two conditions. To dissect these findings further, we then compared the results obtained in experiment 1 and experiment 2.
Calcium signal in response to quinine and devalued sucrose. A, Heatmap of neuronal response to quinine and devalued sucrose. Nine distinct clusters that responded differently to the tastants were observed. B, Example of excitatory and inhibitory calcium signal (ΔF/F in top panels), perievent histograms (middle panels), and response (z score) across trials for quinine and devalued sucrose. C, Mean ± SEM response (z score) during quinine (red) and devalued sucrose (blue) infusion for each of the 9 clusters. D, Summary table showing the response profile to each tastant and the number of neurons that comprise each cluster. E, Distribution of responsive and nonresponsive (N/R) neurons to quinine and devalued sucrose. F, Inhibitory/excitatory profile. Pie charts show the number (percentage) of phasic IL neurons that exhibited excitatory (red) or inhibitory (blue) calcium activity patterns in response to quinine (left) and devalued sucrose (right). G, Percentage of phasic neurons that responded to devalued sucrose only (blue), quinine only (red), or both tastants (purple). H, Categorization of the IL neuronal ensemble phasic to both devalued sucrose and quinine. Scatter plot of neurons divided according to the average response (z score) to devalued sucrose (y-axis) and quinine (x-axis).
Overlap and divergence in IL processing in conditioned and unconditioned rats
To examine how the IL processes hedonics in innate (unconditioned) versus after induction of CTA (conditioned), we first compared the inhibitory/excitatory profiles obtained in experiments 1 and 2 (Fig. 7). When sucrose was conditioned to be aversive using LiCl (devalued sucrose), it increased the number of excitatory neurons relative to normal rewarding sucrose (x2 (1, 306) = 13.66, p < 0.001; Fig. 7A, left). No significant difference was observed between experiments 1 and 2 with respect to the percentage of cells responsive to quinine (x2 (1, 230) = 0.75, p = 0.39; Fig. 7A, right). Next, we determined which neurons drove the change in the firing profile of IL neurons after sucrose was devalued via CTA. Of note, the inhibitory/excitatory profile is composed of both tastant-specific neurons and neurons that responded to both sucrose and quinine. Hence, in Figure 7B, we analyzed the inhibitory/excitatory profiles of sucrose-only neurons (left) and neurons that responded to both tastants (right). Here, no differences were observed in the proportion of inhibitory/excitatory sucrose-only neurons between experiments 1 and 2 (Fig. 7B, left). However, a significant difference between experiment 1 and experiment 2 (x2 (1, 164) = 17.26, p < 0.0001) was observed in those neurons that responded to both sucrose and quinine (Fig. 7B, right). These results suggest that neurons that showed a similar response (i.e., sucrose-only neurons) after conditioning may be involved in encoding the objective hedonic value of the tastant. In contrast, those that shifted from a rewarding-like to an aversive-like profile (i.e., neurons that responded to both tastants) may be involved in tracking the subjective hedonic value that shifts after CTA.
Comparison of the inhibitory/excitatory profiles between experiment 1 and experiment 2. A, IL inhibitory (light gray)/excitatory (dark gray) profiles in response to sucrose/devalued sucrose (left) and quinine (right). While sucrose and devalued sucrose showed different neuronal profiles (left), no difference was observed in response to quinine between experiments 1 and 2. B, Inhibitory (light gray)/excitatory (dark gray) profile for sucrose-only (left) and sucrose/quinine neurons (right) in response to sucrose or devalued sucrose. While no change was observed in sucrose-only neurons, a shift in the profile was observed in neurons that responded to both tastants across experiments. *** p < 0.001; n.s.: non significant.
We then compared the proportion of neurons that were tastant specific, and a significant difference was observed between experiments 1 and 2 in the proportion of neurons that encoded sucrose, quinine, and both tastants (x2 (2, 382) = 7.5, p < 0.05; Fig. 8A). After conditioning, there was an increase in the percentage of quinine-only neurons from 14% to 24% that was proportional to a reduction in sucrose-only neurons from 47% to 35%. Next, we studied differences in the response of these neurons during tastant infusion by analyzing the peak response (average calcium activity during the last 2 s of infusion), and no significant differences were observed between experiments 1 and 2 (Fig. 8B) in tastant-specific neurons. These results suggest that during the development of learned (conditioned) negative affect, neurons that track rewarding stimuli are disengaged while more neurons that track aversion are recruited, but the overall response of these ensembles remains unchanged. These tastant-specific neurons then either increase or decrease in proportion, but as observed, they do not significantly change the inhibitory/excitatory proportions after conditioning (Fig. 7) or their response (Fig. 8). Thus, these ensembles appear to track the objective hedonic value of the stimulus.
IL processing in conditioned versus unconditioned rats. A, The proportion of taste-specific neurons in unconditioned (experiment 1, left) and conditioned (experiment 2, right) states. The pie charts show the number and proportion of neurons that responded to sucrose only, quinine only, and both sucrose and quinine for each experiment. B, Mean ± SEM peak response (z score) for quinine only (left) and sucrose only (right) neurons between experiment 1 (E1) and experiment 2 (E2). C, The proportion of excitatory and inhibitory neurons that responded to both tastants in unconditioned (experiment 1, left) and conditioned (experiment 2, right) states. The pie charts show the number and proportion of neurons that were excitatory to both tastants (light gray), excitatory to quinine and inhibitory to sucrose (black), inhibitory to quinine and excitatory to sucrose (white), and inhibitory to both tastants (dark gray). D, Mean ± SEM peak response (z score) each for each subset of neuronal ensembles that responded to both tastants between experiment 1 and experiment 2. QE/SE, Excitatory to both; QE/SI, excitatory to quinine and inhibitory to sucrose; QI/SE, inhibitory to quinine and excitatory to sucrose; QI/SI, inhibitory to both tastants. **p < 0.01.
Next, we compared the proportions of neurons within the ensemble that were phasic to both sucrose and quinine, and their response to each specific tastant. We found a significant difference in the distribution of neurons between experiments 1 and 2 [x2 (3, 154) = 11.82, p < 0.01]. Specifically, when comparing unconditioned and conditioned situations (Fig. 8C), there was a reduction in the percentage of neurons that were inhibited to sucrose and excited to quinine (Fig. 8C, black) that was proportional to an increase in neurons that were excited to both tastants (Fig. 8C, light gray). Also, after conditioning there was an increase in neurons that were excited to sucrose and inhibited to quinine (Fig. 8C, white) that was proportional to a reduction in neurons that were inhibited by both tastants (Fig. 8C, dark gray). Additionally, when we compared the peak calcium response of each group (Fig. 8D), we observed a significant increase only in the response to devalued sucrose in those neurons that were excitatory to both tastants (t(58) = 3.12, p < 0.01). Overall, the predominant difference between conditioned and unconditioned rats was a shift from an inhibitory toward an excitatory response to sucrose, which mostly occurred in neurons that were excitatory to both tastants and showed an increase in response to devalued sucrose, and this may be associated with the change in the perceived palatability.
Discussion
The hedonic properties of stimuli guide behavior. While the IL is a key structure in hedonic processing in rodents, its precise role remains unclear. We used deep brain, in vivo calcium imaging in freely moving rats to assess how IL neuronal ensembles process innately rewarding sucrose versus aversive quinine and how IL neurons process sucrose when it is conditioned to become aversive through CTA learning. Our findings reveal complex and dynamic parallel processing of tastant information including objective hedonic value and subjective hedonic value after learned aversion by IL neuronal ensembles.
Neuronal ensembles in the IL differentially processes innately rewarding and aversive tastants
In experiment 1, we examined how IL ensembles process information about innately rewarding (sucrose) versus aversive (quinine) tastants. IL neurons differentially processed sucrose and quinine such that sucrose predominantly inhibited IL neurons, whereas quinine predominantly excited IL neurons. Our finding parallels electrophysiology data in the NAc showing that IO sucrose and quinine produced mostly neuronal inhibition and excitation, respectively (Roitman et al., 2005). We also found that a large portion of IL neurons selectively processed sucrose, whereas a small percentage exclusively encoded quinine, showing that, in a normal state, there are fewer neurons in the IL dedicated to process aversion. However, ∼40% of the cells belonged to a neuronal ensemble that encoded both sucrose and quinine. Within this ensemble, a subset of neurons showed a dual profile, being inhibited by sucrose but excited by quinine. In contrast, a separate ensemble exhibited similar responses to sucrose and quinine, being either excited or inhibited by both stimuli. These subsets of neuronal ensembles may be differentially involved in responding to various aspects of innate hedonic processing. For example, the ensemble that is inhibited by sucrose but excited by quinine may be involved in tracking the hedonic value of the taste stimulus. In contrast, the ensemble that is excited in response to both tastants could be involved in tracking stimulus salience, as it has been observed that mPFC excitation facilitates recognition of behaviorally relevant events (Popescu et al., 2016). However, when we examined the neuronal response across each of the 15 trials for these cells, we did not observe any pattern associated with salience coding (e.g., increased activity during early vs later trials).
The IL differentially processes innate versus conditioned aversion
We assessed how neuronal ensembles in the IL encoded conditioned aversion by devaluing a rewarding sucrose solution using CTA. Overall, devalued sucrose tended to activate IL neurons. Although these results mirror previous findings showing neuronal excitation in the NAc shell in response to a conditioned aversive taste solution (Roitman et al., 2010), the activation pattern in the IL evoked by devalued sucrose was less robust than quinine as only 55% of the neurons were excited to devalued sucrose. These results parallel the behavioral data, in which quinine tended to promote greater aversive TR than devalued sucrose. When characterizing the response of phasic neurons, only a small number responded exclusively to the innate aversive stimulus, and most IL neurons were phasic to both devalued sucrose and quinine (41%) or exclusively to the devalued sucrose (35%). Investigating the neuronal ensemble that encoded both sucrose and quinine, we found that the largest proportion of neurons exhibited excitations to both tastants (46%). These findings indicate that there is both overlap and divergence in the processing of innate and conditioned aversive tastants. While some neurons in the IL may process aversion in general, other neurons process more complex characteristics such as the nature of the stimulus or whether it is innate or conditioned.
Distinctions between innate versus conditioned aversion processing have been reported in other brain regions. In both innate and conditioned aversive situations, the quality of the taste is initially processed by gustatory neurons in the pontine parabrachial nucleus. Lesions of the pontine parabrachial nucleus suppress the development of aversive TR during CTA, but not in response to innately aversive quinine (Flynn et al., 1991b). Furthermore, innate versus conditioned aversive behaviors are modulated differently, as the gut hormone glucagon-like peptide-1 attenuates aversive TR and the intake of quinine but has no effect on either TR or the consumption of an LiCl-paired saccharin solution in rats (Douton et al., 2021). Finally, the IL has robust connections with the gustatory cortex (Gabbott et al., 2003), and both structures process taste quality and taste palatability (Jezzini et al., 2013). Lesions in the rat gustatory cortex impaired quinine sensitivity but did not prevent the expression of CTA once acquired (Bales et al., 2015), whereas lesions in the ventromedial prefrontal cortex, including the IL, increased aversive responses to quinine but reduced the development and expression of CTA in rats (Flynn et al., 1991a,b; Berta et al., 2018). Hence, while some aspects of aversion are processed similarly by IL neurons, the activation of specific neuronal ensembles in the IL will ultimately depend on the nature of the stimulus (i.e., innate vs learned).
Overlap and divergence in IL processing in conditioned and unconditioned rats
A comparison of the findings across both experiments reveals similarities and differences in sucrose and quinine processing in conditioned and unconditioned rats. First, innately rewarding sucrose predominantly inhibited IL neurons, but this shifted to an aversive-like profile (mostly excitatory) when sucrose was devalued via CTA, consistent with prior findings in the NAc shell (Roitman et al., 2010). However, a comparison of experiments 1 and 2 revealed differences in the proportion of neurons responsive only to sucrose, only to quinine, and to both sucrose and quinine. Here, a reduction in the percentage of neurons that responded exclusively to sucrose was proportional to the increase in the percentage of neurons that responded to quinine following CTA. This finding suggests that, during the development of an aversive state, some neurons that respond to rewarding stimuli are disengaged, while more neurons that track aversive stimuli are recruited, making the system more susceptible to track aversion. Importantly, neither of these tastant-specific ensembles changed their response in conditioned rats. Second, when we analyzed the inhibitory/excitatory profile, we found that the shift from appetitive-like to aversive-like activity only occurred in neurons that responded to both sucrose and quinine and not in sucrose-only or quinine-only neurons. This finding suggests that neuronal ensembles that do not change during CTA (i.e., sucrose only and quinine only) may function to track the objective hedonic value of the tastant (i.e., its natural properties). Conversely, those neurons that responded to both tastants and did change during the development of CTA, may serve to track the subjective hedonic value (i.e., the properties perceived by the rat). This is supported by a change in the proportions of this neuronal ensemble across conditions. That is, in unconditioned rats, most of these neurons were inhibited in response to sucrose and excited in response to quinine, while in conditioned rats most of these neurons shifted toward an excitatory response to both tastants. Importantly, neurons that were excited by both tastants showed an increased response to devalued sucrose after the development of CTA. Hence, even within the neurons that respond to both tastants, a small subgroup that did not shift its response after CTA continues to track objective hedonic value. However, a larger subgroup comes to encode subjective hedonic value, which shifts toward an aversive-like profile and a stronger excitatory response to sucrose after conditioned taste aversion learning.
Finally, it is possible that the increased excitation profile following CTA may be associated with an increased susceptibility to aversion. However, although there was an increased proportion of neurons responding only to quinine in conditioned rats, this change was not translated into a differential tracking of innate aversion, as no difference was observed in the inhibitory/excitatory profile or in the peak response to quinine between conditioned and unconditioned rats. Collectively, these findings indicate that there is a dynamic hedonic processing within the IL that takes into account both the objective value of the stimulus and the subjective experience of the rat, and that can be modified with experience.
In summary, hedonic processing is complex, requiring the coordinated function of several brain regions. In this study, we found that neuronal ensembles in the IL track both the objective and the subjective hedonic values of taste stimuli. Furthermore, we found that, during the development of negative affect, there are shifts in the neuronal profile that may be involved in adapting the behaviors necessary to reject the aversive stimulus. Hence, when sucrose is devalued using LiCl (1) a subset of neurons continue to track objective hedonic value; (2) some sucrose-only neurons stop responding while more quinine-only neurons are engaged; and (3) a shift in the profile of sucrose/quinine neurons that tracks the subjective hedonic value is observed, in which most neurons are now excited in response to both quinine and devalued sucrose. Collectively, this research supports a role of the IL in affect regulation and contributes to our understanding of its dynamic role in the development of learned negative affect.
Footnotes
This work was supported by Department of Health and Human Services | National Institutes of Health | National Institute on Drug Abuse Grants DA-014339 and DA-052108. We thank the Genetically-Encoded Neuronal Indicator and Effector (GENIE) Project and the Janelia Research Campus of the Howard Hughes Medical Institute, including Drs. Vivek Jayaraman, Douglas S. Kim, Loren L. Looger, and Karel Svoboda, for providing viral vectors. We also thank Dr. Travis Moschak for aiding with data analysis. In addition, we thank Inscopix and specifically Dr. Shanna Resendez for assistance with establishing calcium imaging in the Carelli laboratory.
The authors declare no competing financial interests.
- Correspondence should be addressed to Regina M. Carelli at rcarelli{at}unc.edu
