LiCl-induced sickness modulates rat gustatory cortical responses.

Gustatory cortex (GC), a structure deeply involved in the making of consumption decisions, presumably performs this function by integrating information about taste, experiences, and internal states related to the animal's health, such as illness. Here, we investigated this assertion, examining whether illness is represented in GC activity, and how this representation impacts taste responses and behavior. We recorded GC single-neuron activity and local field potentials (LFPs) from healthy rats and rats made ill (via LiCl injection). We show (consistent with the extant literature) that the onset of illness-related behaviors arises contemporaneously with alterations in 7 to 12 Hz LFP power at approximately 12 min following injection. This process was accompanied by reductions in single-neuron taste response magnitudes and discriminability, and with enhancements in palatability-relatedness-a result reflecting the collapse of responses toward a simple "good-bad" code visible in the entire sample, but focused on a specific subset of GC neurons. Overall, our data show that a state (illness) that profoundly reduces consumption changes basic properties of the sensory cortical response to tastes, in a manner that can easily explain illness' impact on consumption.

Introduction

Neural responses in primary sensory cortex are often thought to faithfully represent physical stimuli, and while recent studies (including ours) have challenged this view by documenting enhancements and decrements in stimulus-induced firing related to animals’ internal states, there has been little work setting these changes in any sort of functional, mechanistic context. Here, we show that a state (illness) that profoundly reduces consumption changes basic properties of the sensory cortical response to tastes, and then go beyond this to precisely characterize this response plasticity, connecting it to the specific perceptual changes that drive illness’ impact on consumption.

A host of external and internal contextual variables work with experience to shape an animal’s behavior in response to sensory stimulation. Prominent among these variables are the animal’s own internal states, which can be extremely positive (e.g., “euphoria”) or negative (e.g., “depression”), and which profoundly influence the animal’s interactions with its environment [1]; systemic illness, for example, such as that induced by the intake of toxins, drastically alters an animal’s behavior in relation to food stimuli [2-4]. Such states are likely instantiated in broadly distributed neural networks, and as such, they can be indexed using spectral properties of the electroencephalogram [5,6] or local field potentials (LFPs; [7,8]).

One mechanism whereby illness might influence feeding behavior is via modifications of taste perception. Sickness changes taste palatability [9,10], making even normally preferred substances aversive (potentially prolonging infirmity by demotivating the ingestion of nutrients and/or a possible cure for said illness [11]). This impact of illness on taste palatability can be long-lasting, and even permanent [12], an intimacy of interaction that makes it reasonable to propose that illness may manifest, at least in part, as changes in the function of brain regions within which one can record palatability-related taste responses.

An obvious candidate region is gustatory cortex (GC; a fairly broad swath of granular and dysgranular insular cortex anterior to visceral cortex), which is involved both in processing of palatability and illness-related learning [13,14]. As of now, however, little is known about the cortical processing of malaise. Much is known about the basic mechanisms of function for malaise-causing agents such as lithium chloride (LiCl) at the nonneural and peripheral levels [2-4,15-19], but the impact of these basic mechanisms on activity in regions critical for chemosensory learning remains uninvestigated.

The work presented here begins to fill that knowledge gap, taking a cue from studies showing that basic information about brain states can be assessed in analyses of LFPs, and more specifically from studies showing: (1) that spectral properties of LFPs shift with changes in an animal’s internal state [20,21]; and (2) that these changes are coupled with changes in single-neuron firing dynamics [22-24]. These effects apply to GC taste coding [25], which reliably fluctuate with changes in attentional state in a manner that is specifically linked to palatability coding [26]. Here, using extracellular single-neuron recordings, we test the degree to which the onset of an illness state (indexed in terms of changes in mobility) is related to changes in GC LFPs, and go on to test how these behavioral and LFP phenomena correlate with changes in identity-related and hedonic information in GC single-neuron ensemble responses.

Our results suggest that LiCl-induced changes in GC μ (7 to 12 Hz) power reflect the onset (but not the maintenance) of sickness, in that they specifically emerge around the time at which sickness-related behaviors also emerge [2-4,27,28]. This change is accompanied by reductions in the magnitude of taste-driven responses, which convey less information about taste identity in sick rats, and simultaneously by enhancement in the palatability-relatedness of the same responses, which appear to collapse toward a simple “good/bad” judgment. These results provide evidence that emesis caused by LiCl modulates network activity in GC and plays an important role in shaping the cortical taste responses that are necessary for learning and decision-making, and further expand our insight into the state dependence of sensory coding.

Materials and methods

Ethics statement

All experimental protocols were conducted according to the National Institutes of Health (NIH) guidelines for animal research and approved by the Institutional Animal Care and Use Committee (IACUC) at Brandeis University (IACUC protocol # 22007).

Subjects

Female Long–Evans rats (n = 26, 250 to 300 g at time of surgery; Charles River Laboratory, Raleigh, North Carolina) that were naïve to tastants served as subjects in this study. Animals were maintained on a 12-h light/dark schedule, with experiments performed in the light portion of the cycle from 7:00 to 11:00 AM. Rats had ad libitum access to food and water except during experimentation (see below).

Apparatus

Neural recordings were made in a custom Faraday cage (21 × 24 × 33 cm) connected to a PC and Raspberry Pi computer (Model 3B). The Pi controlled the opening time and duration of solenoid taste delivery valves. The PC controlled and saved electrophysiological recordings taken from electrode bundles via connections to an Intan data acquisition system (RHD2000 Evaluation System and Amplifier Boards; Intan Technologies, LLC, Los Angeles). Each bundle consisted of 32 microwires (0.0381 mm formvar-coated nichrome wire; AM Systems) glued to a custom-made interface board (San Francisco Circuits) and soldered to a 32-channel Omnetics connector, which was fixed to a customized drive [29].

Surgery

Rats for recording experiments were anesthetized using an intraperitoneal (IP) injection of a ketamine/xylazine mix (1 mL ketamine, 0.05 mL xylazine/kg body weight). Maintenance of anesthesia followed 1/3 induction dose every 1.25 h. The anesthetized rat was placed in a stereotaxic frame (David Kopf Instruments; Tujunga, California), its scalp excised, and holes bored in its skull for the insertion of self-tapping ground screws and an electrode bundle. Each bundle was painted with a lipophilic membrane stain (Thermo Fisher Scientific, Indianapolis, Indiana) and inserted 0.5 mm above GC (coordinates: AP +1.4 mm, ML ±5.0 mm, DV −4.4 mm from dura)—a location corresponding primarily to layer V of the granular and dysgranular segments of the insular midsection, in which we have reliably found taste responses in the past (see [30-32]). Assemblies were cemented to the skull, along with 2 intraoral cannulae (IOCs; flexible plastic tubing inserted close to the tongue in the cheek and extending upward to the top of the skull) using dental acrylic [26]. The rat’s body temperature was monitored and maintained at approximately 37°C by a heating pad throughout the duration of the surgery. Rats were given 6 d to recover from the surgery, after which electrode bundles were lowered by 30 to 50 μm/d until discriminable GC neuron waveforms were visible in multiple channels; there they stayed through the entire protocol, which lasted (at most) a single session of each type (saline and LiCl; see below).

Water restriction

During experimentation (Fig 1A), mild water restriction (ad libitum access to 20 mL every day at 4 PM, 6 h after experimental sessions), started 6 d following surgery, ensured engagement in the task. Daily records were kept of weight and water intake to ensure that rats did not fall below 85% of presurgery weight.

Fig 1

General sickness induction protocol and histological verification of recording location in GC.

(A) On Day 1, rats received a single injection of saline followed by either an injection of saline (Control Group; top) or 0.15 M LiCl (Experimental Group, bottom). GC activity was recorded for 20 min following each injection, after which tastes were delivered via IOC across an approximately 40 min session, and 48 h later, only a single saline injection was administered to both groups, 20 min before the taste testing period. (B) A coronal slice (left) from 1 rat, stained with cresyl violet. The track is an electrode bundle lesion, with the end of the lesion marking the final resting location of the wires. A coronal slice schematic (right) with circles indicating the electrode bundle resting location for each rat (experiment 1: purple, experiment 2: green). (C) The average waveform of a representative putative pyramidal cell [top] and interneuron [bottom]. Light blue shading represents the standard deviation of the noise in the waveform. GC, gustatory cortex; IOC, intraoral cannulae; LiCl, lithium chloride.

Experimental design

On the second day of water restriction, rats began 2 d of habituation to liquid delivered directly to the tongue via IOC, with 60 and 120 30-μL infusions of water delivered each day, respectively. To maximize our ability to record from neurons held stably across both (saline and LiCl) sessions, electrodes were lowered to their recording depth (approximately 4.5 mm DV from dura) on the second day of habituation and left at that depth.

For Experiment 1, on the morning following the second habituation session, in vivo recording sessions commenced: Rats received a subcutaneous (sc; 0.50% of body weight) injection of isotonic saline followed 20 min later by an injection of either a second saline injection (Control) or illness-inducing LiCl (Experimental; ); following each injection, neural and behavioral data were collected—20 min of “passive” recording followed by 120 taste delivery trials in which 1 of 4 gustatory stimuli (0.1 M sodium chloride [NaCl], 0.3 M sucrose, 0.1 M citric acid [CA], and 0.01 M quinine-HCl [QHCl]) was delivered via IOC in a pseudo-random order. Stimuli and concentrations were chosen to ensure a range of distinct taste identities and palatabilities and to maximize comparability to our [33-36] and others’ [37,38] studies. A second testing session, identical to the first session with the exception that only a single sc. injection of saline was administered 20 min prior to taste delivery, was given 48 h after the first testing session (), with the electrode bundle remaining in the same place. To test whether any observed LiCl/saline differences were not order effects, a subset of animals (N = 5) were subjected to the same experimental protocol with the exception of the order of injections being reversed (2 saline injections on the first day, LiCl injection on the second). Our analyses revealed no difference between the orders (ps > 0.05).

For Experiment 2, additional (N = 3) taste-naive animals were subjected to a single recording session in which the injection of LiCl was immediately followed by taste deliveries via the IOC.

Illness induction

Because we wished to elicit a mild illness that would not cause large scale, overt behaviors that might serve as confounding explanations for any observed changes in neural activity, we induced malaise using a mild concentration of LiCl (0.15 M, delivered sc). Multiple studies demonstrate that the malaise induced using this concentration is strong enough to cause conditioned taste aversion (CTAs) without causing gross changes in behavior [4,39].

Behavioral analyses

The assessment of lethargy provided a basic measure of sickness [2,28]. Three cameras (positioned below, above, and diagonally within the behavioral chamber (see Apparatus)) captured mobility in a separate subset of animals implanted only with IOCs (N = 4 and 5, Control and Experimental). Each video was manually scored for lateral (grid-line crossings) and vertical (rears; 2 forelimb paws off the ground for >0.5 s) movements by a trained, but blind to condition, observer. Such events have been shown to be sensitive measure of welfare in rodents—healthy, nonanxious mice/rats will perform lateral and vertical movements as a means to explore their environment [19,40]. Given the small size of the recording chamber, the rats had limited ability to move laterally, which led us to focus our analyses primarily on vertical movements. Healthy/sick differences in frequency and duration of these movements were analyzed and fit with sigmoid functions so that we could determine whether and when LiCl-treated rats became ill.

Histology

At the completion of the experiment, rats were deeply anesthetized with ketamine/xylazine (200:20 mg/kg, IP) and then perfused transcardially with physiological saline followed by 10% formalin. The brains were extracted and stored in a 10% formalin/30% sucrose solution for at least 3 d before staining, after which they were frozen and sliced on a sliding microtome (Leica SM2010R, Leica Microsystems; thickness 60 μm). Slices were mounted, and sections were stained with a Nissl stain (Thermo Fisher Scientific, Indianapolis, IN) to evaluate cannulae and electrode tracks, respectively, via inspection of fluorescence on a light microscope ().

Electrophysiology methods and analyses

Differential recordings from the micro-electrodes were sampled at 30 kHz using a 32-channel analog-to-digital converter chip (RHD2132) from Intan Technologies. The signals were digitalized online at the head stage/amplifier and saved to the hard drive of the PC. The collected recordings were filtered for either single-neuron action potential isolation (300 to 3,000 Hz bandpass) or LFPs (1 to 300 Hz lowpass) and analyzed off-line. Putative single-neuron waveforms (3:1 signal-to-noise ratio) were sorted using a semi-supervised methodology: Putative action potentials were grouped into clusters by a Gaussian mixture model (GMM) with strict noise tolerance and refined manually; conservatism was enhanced via a requirement that the variability in spiking be random (and thus not reflecting biased cluster cutting; for the details of the recording system and Python scripts, see [29]). To further ensure that the isolated waveforms represented single-neuron records, we also calculated the interspike intervals and removed units for which ≥1% of the intervals disobeyed the biological constraints of refractory period. By this approach, a total of 266 and 190 GC neurons were isolated from, respectively, 7 Experimental (saline-LiCl) and 5 Control (saline-saline) rats across 24 sessions (2 sessions/animal).

All analyses, unless otherwise stated, were performed using all neurons; in a few cases (noted in the text), analysis was restricted to neurons that were taste responsive (see also Table 1).

Table 1

Animal assignment in each figure.

Figure	Animals	Neurons	Experiment: Detail
Fig 2        Panel A    Panels B/C	1 Sal-LiCl5 Sal-Sal vs. 7 Sal-LiCl rats	N/A	LFP
Fig 3        Panel B	4 Sal-Sal vs. 5 Sal-LiCl rats	N/A	Behavioral verification
Fig 4        Panel C	7 Sal-LiCl rats	Sal (n = 136)LiCl (n = 174)	Single unit spiking activity
Fig 5	7 Sal-LiCl rats	Sal (n = 136)LiCl (n = 174)	Single unit spiking activity
Fig 6        Panels A/B	7 Sal-LiCl rats	Sal (n = 136)LiCl (n = 174)	Single unit spiking activity
Fig 7        Panel B	3 LiCl rats	56 taste-responsive units	Single unit spiking activity
Fig 8        All panels	7 Sal-LiCl rats	44 held units	Single unit spiking activity

The same experimental data (all neurons recorded from the experimental group (Fig 1A: Experimental [Saline-LiCl; Sal-LiCl]; main dataset) were used for Figs 2 and 4–6 with additional control data (Fig 1A: Control [Saline-Saline; Sal-Sal]) were also used in Fig 2. Fig 7 involved collection of a new dataset from 3 rats that received LiCl injection prior recording. Fig 8 used all neurons that were held across sessions from the main dataset. LFP, local field potential; LiCl, lithium chloride.

Fig 2

LiCl administration alters GC LFP 15-min post-injection.

(A) Representative spectrograms from 1 animal (saline-LiCl condition) showing activity after saline (top) or LiCl (bottom) injection. LiCl reduces power in μ during the last half of the 20-min period. (B) The posterior distributions for a single change point model identify the probabilities of a change (shown as frequency of τ) in μ power (each color represents distribution for single animal) for saline-saline (top, N = 5) and saline-LiCl (bottom, N = 7) groups. The horizontal dashed lines indicate the 99th percentile from temporally shuffled data, and the overlaid boxplot depicts the range in peak likelihood of change in μ power for distributions with values higher than the 99th percentile: Each dot represents a single animal (the animal from panel A is noted with a red circle); the box extends from the lower to upper quartiles (red line: median, black diamond: mean) of likelihood. The mean onset of the change in μ power occurs at 15.11 min, with all change points occurring after 10-min post-injection and before 18-min post-injection. (C) The absolute difference in μ for saline-saline (purple) and saline-LiCl (teal) groups reveals a significant interaction between change in μ power and quartile (p < 0.05, 2-way ANOVA), with LiCl changing μ power significantly within the fourth quartile (* p < 0.05; 1-way ANOVA). The error bars represent SEMs. Data underlying this figure can be found in S1 Data. LFP, local field potential; LiCl, lithium chloride.

Fig 4

LiCl administration reduces magnitudes of taste responsiveness.

PSTHs of trial-averaged firing rates of 4 representative neurons recorded in (A) saline and (B) LiCl states. The examples in the top panels show excitatory responses, and the bottom examples show inhibitory responses. Vertical dashed lines indicate the time of taste delivery. Subtle variances in firing rates observed between 0–200 ms are not statistically significant (1-way ANOVA, ps > 0.05, see Methods). (C) Illness decreased the magnitude of these responses; for neurons with excitatory responses, this reduction was significant (** p < 0.01; 2-way ANOVA with Tukey post hoc). Error bars represent SEMs across neurons. Data underlying this figure can be found in S3 Data. LiCl, lithium chloride; PSTH, peristimulus time histogram.

Fig 6

LiCl administration shifts GC taste responses in the direction of a simple “good vs. bad” judgment.

(A) The time course of average correlations (Spearman rank-correlation coefficient) between firing rates and known palatability shows the expected increase as responses reached the late epoch, but this rise reached a higher peak following LiCl (n = 174) than saline injection (n = 136, Kruskal–Wallis H-test, ** p < 0.01). (B) The sickness-induced reduction of taste discriminability, juxtaposed with this sickness-induced enhancement of palatability-relatedness, suggests that illness selectively decreases the distinctiveness of tastes with similar palatability. As predicted, a PPI, which evaluates this possibility (see inlaid equation), peaks at a larger late epoch value following LiCl injection than following saline (unpaired 2-sample Wilcoxon test; * p < 0.05), reflecting sickness-induced polarization of coding. The error bars represent SEMs. Data underlying this figure can be found in S5 Data. GC, gustatory cortex; LiCl, lithium chloride; PPI, pure palatability index.

Fig 7

Reduction in taste discriminability emerges with sickness within session.

(A) A schematic of the second testing protocol, in which an injection of 0.15 M LiCl was immediately followed by delivery of tastants via IOC for approximately 40 min. (B) Bar plot shows the percent of trials in which the proffered taste was correctly identified using an LDA (trained on the first 5 trials) from GC responses. Across N = 3 animals (n = 56 taste responsive neurons), taste discrimination is significantly reduced post-sickness induction. * p < 0.05; paired t test with Bonferroni correction. The error bars represent SEMs. Data underlying this figure can be found in S6 Data. GC, gustatory cortex; IOC, intraoral cannulae; LDA, linear discriminant analysis; LiCl, lithium chloride.

Fig 8

Neurons held across post-LiCl and post-saline tasting sessions produce distinct clusters of taste response profiles.

(A) Firing rate trajectories for single neurons before (small circles) and after (large circles) taste delivery to the presentation of citric acid reveal condition-specific (sickness–y-axis, healthy–x-axis) taste response properties. Responses are colored with respect to angle of the response trajectory (0–90°, 90–180°, 180–270°, and 270–360° are yellow, teal, magenta, and blue, respectively), which highlights the distinct natures (excitatory, inhibitory) of responses in each condition. (B) A comparison of pre-stimulus activity suggests that post-injection firing rates did not differ between conditions: The peak of the distribution of saline-LiCl differences hovers around 0, and the distribution is fit well by a Gaussian (Kolmogorov–Smirnov test; p = 0.98); these facts suggest 0 difference (plus random noise). (C) A population response histogram (n = 44) held neurons depicting each neuron’s taste response (change from pre taste) as a function of vector direction in panel B confirms the presence of 3 unique peaks of vector direction (peaks indicated with red Xs). (D) A GMM was used to cluster cells based on response magnitude and trajectory. This analysis confirmed distinct cluster responses, revealed here in a polar plot (dashed lines denote 95% confidence interval of cluster classification with respect to black-filled centroids). PSTHs for representative neurons are seen adjacent to the detected clusters, indicating taste responses post-saline (black) and post-LiCl (colored, cluster specific). (E) Cluster 3 (magenta) neurons reliably and selectively experience a sickness-induced increase in palatability-relatedness in the late epoch (nonparametric 2-way repeated-measure ANOVA; *** p <0.001). Vertical lines represent SEMs. Data underlying this figure can be found in S7 Data. GMM, Gaussian mixture model; LiCl, lithium chloride; PSTH, peristimulus time histogram.

LFPs were extracted only from channels containing isolated single-neuron data, in order to avoid potential signal artifact arising from noisy/broken channels. Data were averaged across electrodes within and across session for each animal, and analyzed across the 1 to 20 Hz frequency bands, with focus on the μ bands (7 to 12 Hz, [41-43] in 1-min time bins). After normalizing (across the entire session) each animal’s μ power to between 0 and 1, the LiCl vectors were subtracted from the saline vectors to reveal changes in μ following LiCl administration, and an ANOVA was used to compare groups. Differences were considered significant only if 3 or more consecutive time bins reached a p < 0.05 criterion.

Change point analysis

To determine the position of a change point in GC μ power, we used a custom model implemented in the PyMC3 probabilistic programming package [44], with parameter estimation performed using Markov Chain Monte Carlo sampling (see details below). The LFP power time series were z-scored, and then modeled using 2 normal distributions; the model attempts to detect a single change point (tau) in the mean value of the LFP power as the time series switches between distributions. Change points were determined for each animal independently.

Since sampling returns a distribution over the parameter values explaining the data, distributions of tau with low variance (high peaks) indicate strong presence of a change point. To mark the position of these change points using the tau distribution in an unbiased manner, we compared the tau distributions (containing equal numbers of samples) of our time series (actual data) to those of 50 temporally shuffled time series (shuffled data), which, by definition, have no change points. Points in the actual data distribution higher than the 99th percentile of the shuffle distribution were marked. The average of these marked time points was taken to be the position of the change point for that time series.

The calculation of the behavioral changes and LFP change points was performed on separate cohorts of animals. We then quantified the likelihood that the close synchronization between the behavioral and neural changes seen in our data could have happened by chance, by comparing the data to simulated data produced under the assumption that the latencies of the average behavioral and LFP change points are drawn from independent, uniform distributions. We calculated the distribution of summed, absolute distances between the behavioral and LFP change points, testing the hypothesis that the change points in our data are clustered more tightly than those drawn from the simulation (a p-value less than 0.05 indicates a significantly more coupled relationship than if time points were truly independent).

Taste responsiveness

We defined basic responsiveness to taste delivery by first subtracting the across-trial averaged 1 s of prestimulus firing from the first 1 s of taste-evoked firing for each neuron (ignoring the first 200 ms of post-taste activity; see below), and subjecting the resultant single-neuron data to a Mann–Whitney U test to determine whether a significant change in firing rate occurred when the taste hit the tongue. These data were then pooled to ascertain the overall magnitude of response change under different conditions. But because taste delivery could reduce firing to below baseline levels in some neurons (thereby masking the true magnitude of effects in the averaging of enhanced and reduced firing rates), we divided the sample into those in which firing normalized to pre-stimulus baseline was greater than zero, and those in which it was less than zero. We refer to these as “excitatory” and “inhibitory responses,” respectively, but note that this designation only refers to the direction of firing rate change—we make no claims regarding whether or not “inhibitory responses” are caused by direct inhibitory influence on these neurons; nor do we intend to imply that these are responses found in inhibitory interneurons. We used a 2-way ANOVA to test whether LiCl administration changed taste response magnitude.

Note that in these analyses, as well as in those described below, we typically ignored the first 200 ms of post-taste activity. This was done because we have repeatedly observed that these early responses are seldom chemosensory in our paradigm, reflecting only tactile stimulation of the tongue (see [26,30-32,36,45]). This differentiates our paradigm from those in which rats perform active licking or lever-pressing for fluid delivery [46-50], wherein researchers observe a much shorter pre-chemosensory response, likely because the rodent is able to anticipate taste acquisition (see also [51]).

Taste specificity

The above analysis is performed on data averaged across tastes, and therefore serves only to estimate whether, and how, GC is responsive to oral stimuli—not whether that response depends upon stimulus identity. To determine how LiCl impacted the taste specificity of GC responses, we subjected whole ensembles to a standard linear discriminant analysis (LDA) classifier used in previous studies [30,32,52]. This classifier tests the reliability with which a trial of taste-evoked responses in an ensemble of simultaneously recorded neurons can be identified among responses to other tastes. We binned neural responses into 250 ms bins and used a linear classifier with a leave-one-out validation approach, calculating the prediction accuracy of the classifier averaged across each excluded trial for each time bin. Paired-sample t tests (with Bonferroni correction) were performed for each post-taste delivery epoch (see above) to determine the impact of LiCl.

To assess changes of taste discriminability within a single session, the ensemble-wise LDA classifier was trained on the first 5 trials (per taste) and tested on later trials. Here, and when testing the degree to which sickness impacted discriminability in real-time, we restricted the analysis to taste-responsive units (typically >70% of the total number of recorded neurons [30,31,45]; see ): Across-iteration averages were computed for each tested trial and the resulting means were binned into 5-trial blocks (approximately 6 min/block). Blocks were normalized (between 0 and 1) across animals, and results were subject to a repeated measures ANOVA and a series of paired-sample t tests (with Bonferroni correction).

Finally, it is worth noting that taste responsiveness and taste specificity provide slightly different types of information about GC activity. It is possible for a neuron to be taste responsive and not respond distinctly to different tastes, and it is possible for a neuron to respond in a manner that is taste specific without the average, across-stimuli response being significantly modulated from baseline. Both are useful.

Taste palatability

Using our now-standard palatability-correlation analysis [32,35,51,53], we evaluated the degree to which LiCl altered the amplitude of palatability-relatedness in late epoch GC taste responses. Using a moving window (window size: 250 ms, step: 25 ms), we correlated firing rates with well-established palatability ranks (sucrose > NaCl > CA > QHCl; [32,54]) and compared the magnitude of this correlation in healthy and sick rats.

In a further analysis of palatability-relatedness, we computed what we termed a “pure palatability index” (PPI), adapting methods from [55] to test the degree to which GC taste responses could be described to code a simple “good versus bad” dichotomy. Specifically, we compared the Euclidian distances between single neurons’ responses (normalized to −500 ms pre-taste delivery) to tastes with similar palatability (i.e., sucrose and NaCl, CA and QHCl) to those between tastes with different palatability (sucrose and CA, sucrose and QHCl, NaCl and CA, NaCl and QHCl). This amounts to evaluating the ratio of the distances between “different-palatability” and “same-palatability:” a lack of palatability-related information results in a PPI of 0 (because tastes of similar palatability and tastes of different palatabilities are equally distant from one another); the more polarized the response into “good versus bad”—the larger the ratio—the more positive the PPI. To avoid artificially attenuating the effect via the inclusion of information from different epochs in single averages, the PPI was calculated from firing within the middle of our standard epochs [30,31]. The results were significance tested using a Wilcoxon test.

We hasten to emphasize that these analyses of the degree to which firing is palatability-related are quite different from the above-described analysis of the degree to which firing is taste specific. In fact, taste information and palatability-relatedness are actually very different variables—one makes use of any cell-specific differences in responses to different tastes, and one is specifically measuring whether a particular pattern of responsiveness exists—necessitates these differences in analysis.

Stability of single-neuron waveforms across days

A subset of analyses required the stable tracking of neurons held across testing sessions (i.e., “held neurons,” ). To evaluate whether a neuron was “held,” we performed a spike shape analysis the likes of which has been brought to bear in several previous studies [33,35,56,57]. We applied a conservative criterion for stability based on within-session data (comparing each neuron’s waveforms from the first third of the session to those of the last third)—only neurons for which the between-session (nonparametric clustering statistic) value was less than the 95th percentile value calculated from the within-session data were deemed to have been stably held. Of the entire population of recorded GC neurons, a total of 44 (23.2%) satisfied this criterion.

Cluster detection in patterns of healthy-ill response differences

A clustering analysis was used to identify groups of held neurons for which LiCl impacted palatability-related firing similarly. A condition response was determined for each held unit (under each condition) by taking the average (minmax-normalized) pre- (−750➔250 ms) and post-delivery (250➔750 ms) firing rates (presented as a percentage of maximum responsiveness). We then calculated and plotted the distance and Cartesian direction between condition responses or each neurons’ taste response, yielding 176 (44 held units × 4 tastants) response differences (RDs)—quantification of how LiCl administration changed the excitatory or inhibitory response to the taste. As a conservative estimate of RD likeness, we determined the number of clusters that best fit a GMM probability distribution (calculating the Bayesian information criterion) as done previously [58]. Classifications were considered valid only if they fell within the 95% confidence interval from the centroid of each respective cluster (see ), after which we constrained our analyses to neurons within the same cluster.

Cluster-specific taste palatability

Using methods described above, we evaluated the effect of LiCl administration on palatability-related firing for neurons within clusters. A nonparametric 2-way ANOVA was used to reveal whether differences (saline–LiCl) in hedonic coding (Spearman rho) differed across cluster and time during taste processing.

Results

LiCl administration changes GC LFPs and behavior

We recorded activity for 20 min following subcutaneous injections of either LiCl or saline, before beginning to record taste-driven activity (); GC single-unit responses () and LFPs were acquired from drivable bundles of 32 wires. Given that network function, measured in terms of spectral properties of LFPs, appears to change with even the most innocuous of body states (e.g., sleep versus wake, [59,60]), and that single-neuron firing is altered in concert with these changes [61,62], our investigation into the characterization of illness started with assessment of changes in GC LFPs. We focused on power in the mu (μ: 7 to 12 Hz; [25]) range, because power in this frequency band has proven particularly sensitive to changes in even general states related to wakefulness and attention [20,25,26,41,63]. While the results described below were also observed in frequency ranges above (e.g., β) and below (i.e., θ) μ, they were centered on and largest in the μ range.

Save for the brief period immediately following handling and injections, the amplitude of μ in the LFPs remained relatively stable across the first half of the (20 min) post-injection recording periods for both experimental groups, but an impact of LiCl injections emerged in the second 10-min period following injections for the saline-LiCl group ( LFPs spectral power from a representative rat following either saline [top] or LiCl injection [bottom]). The precise nature of this impact varied with individual, in some cases (N = 3) involving a reduction of μ power and in some cases an increase (N = 4), but a change point analysis quantifying the time points and likelihoods of change in μ power for each animal (across both experimental groups) revealed that LiCl-induced changes in GC μ power reached significance at around the same time in the majority (5 out of 7) rats (). That is, the timing of the change in spectral properties was remarkably reliable across the 7 rats, despite the fact that in some cases the change involved an enhancement of the power in the μ range, and in some cases the opposite: had no change occurred, or had the change been gradual/subtle following injection (as in saline-saline–injected rats, , N = 5), the likelihood of change (τ; posterior distribution) would be spread uniformly across post-injection time; instead, the analysis reveals that a change in GC μ power reliably occurs between 10.89 and 17.07 min post-injection (never before and never after), with the mean likelihood occurring at 15.11 min.

An analysis of the group data confirms these rat-by-rat results, demonstrating that absolute changes in μ power following LiCl injection emerged late in the 20-min post-injection recording session (compared to saline-saline animals, ). A 2-way ANOVA on these data revealed a significant interaction between group and quartile (F(3, 33) = 3.62, p = 0.023); a subsequent 1-way ANOVA confirmed that the absolute difference reached significance only in the 16 to 20-min post-injection bin (F(1, 11) = 5.14, p = 0.04) for the saline-LiCl group. While in some cases, an apparent change in spectral power (in this case, μ power) could in fact be an artifactual effect of firing rate changes [64], our investigation failed to observe concomitant changes in firing rate at any time across the 20-min post-injection recording session (F(3, 33) = 2.39, p = 0.09; see also below); this suggests that our observed effect on μ is not simply a reflection of firing rate change, but rather a matter of network synchrony being modulated.

Overall, these results accord well with those of previous studies, in that they demonstrate: (1) that changes in LFP activity are hallmarks of the onsets of cortical state changes, regardless of the specific directionality of the changes [65-68]; and (2) that sickness-related behaviors such as immobility emerge at approximately this time point following LiCl injections [19,40,69-72].

To more completely test this last point, we compared our LFP data to (independently collected and blindly coded) video recordings of illness-related behaviors, predicting that the above-described changes in network activity would roughly coincide with times at which changes in mobility appeared following LiCl injections. Because we specifically injected a low concentration of LiCl in order to avoid gross movement changes that could confound interpretation, and because the small recording chamber limited the rats’ ability to move laterally, we focused our measurement of mobility on rearing events in which the animal lifted both forepaws off the floor simultaneously without proceeding to grooming [73,74]. We specifically hypothesized that reduction of such rearing events, which has been linked to mild LiCl-induced illness [72], would occur at around the time that we observed changes in cortical LFP μ power ().

In fact, the durations of rearing events declined following LiCl (N = 5) injection, but not following saline injection (N = 4). A 2-way repeated measures ANOVA brought to bear on the 20-min recording session (again binned into 5 min quartiles to ensure sufficient power; ) revealed a significant difference in rearing duration emerging post-injection (F(3, 182) = 3.14, p < 0.05)—a change that, like the change in cortical LFP power, became significant (according to post hoc Wilcoxon signed-rank tests) only in the 15- to 20-min post-injection bin (W = 158, Z = 0.65, p = 0.024, r = 0.50), compared to rats receiving only saline injections.

A sigmoidal curve fit to the normalized LiCl-saline difference in rearing duration () confirmed that the appearance of this illness-related change in behavior (the asymptote minima of fit; r2 = 0.72) aligned well with the above-noted change in GC μ power (compared to ), again suggesting that the onset of LiCl-induced illness is reflected in the function of the GC network. The fact that the Figs data were collected from separate groups of rats only increases the conservatism of this interpretation—the likelihood that this close alignment of behavioral and neural changes would occur if LFP change points of the first cohort and behavioral change points of the second cohort were random (uniformly distributed) and uncoupled is extremely low (difference between real and modeled data, p = 0.03). While this result does not conclusively rule out the possibility that LiCl could independently cause illness and cortical changes (see Discussion), the pattern of results implicates the emergence of illness (as reflected in behavior) in the change in GC spectral properties.

LiCl administration causes GC taste responses to collapse into a “good-bad” distinction between tastes

shows 2 example of GC single-neuron taste responses—one for which taste administration caused inhibitory firing rate changes, one excitatory—from each type of session (saline/healthy and LiCl/illness). As an initial look at how illness impacts these responses, we evaluated the magnitudes of responses elicited by tastes delivered directly into the mouth via IOC. While initial paired-sampled t tests (with Bonferroni correction for multiple comparisons) performed across the entire pooled sample failed to reveal an impact of LiCl, this apparent non-effect proved to be an artifact of averaging across neurons with “excitatory” and “inhibitory” taste responses: When we performed separate analyses of those 2 types of taste responses, it became clear that LiCl-induced illness reduced the magnitude of both excitatory and inhibitory GC taste responses (). A 2-way ANOVA performed on these data revealed a significant interaction between the condition and taste response direction (F(1,3191) = 6.05, p = 0.014); a Tukey post hoc comparison showing the main effect of illness was larger for excitatory responses (p = 0.003) but not for inhibitory responses (p > .05).

The above analysis, however, examines only responsiveness, providing no information regarding whether the impact of LiCl-induced illness differs for different specific tastes. We therefore performed an LDA, quantifying the reliability with which GC responses to one taste could be differentiated from responses to other tastes. As shown in (blue trace and bars below), LDA enables us to correctly identify administered tastes from middle epoch responses on over 40% of the trials in healthy sessions (well above chance, which = 25%); this percentage quickly rises to over 50%, despite the use of a small time bin sliding window analysis that left the data at the mercy of unsmoothed trial-to-trial variability. More to the point, this distinctiveness (i.e., classifiability) of responses was significantly diminished following systemic LiCl administration (coral trace and bars in ), with the decrement becoming significant in the “middle epoch” and continuing into the “late epoch” (Mann–Whitney U (1-tailed) = 1,952, M = 34.06%/43.77%, SD = 11.09%/14.58%, p = 0.0003; U = 2,045, M = 39.78%/47.83%, SD = 15.72%/14.64%, p = 0.001, respectively). Taste responses carry less identity-related information when the rat is ill.

LiCl administration reduces taste discriminability in neural ensembles.

The time course of trials in which the taste could be correctly identified by ensemble firing, according to LDA, shows reduced discriminability of responses when an animal is sick (LiCl; coral, n = 174) in comparison to when they are healthy (saline; purple, n = 136). The vertical dashed lines delineate approximate epochal boundaries, as defined in previous research; collapsed into these epochs (bottom bars, the effect of illness only becomes significant within the middle (400–700 ms) and late (800–1,100 ms) epochs post taste delivery (** p < 0.01; pairwise t test with Bonferroni correction). The error bars represent SEMs. Data underlying this figure can be found in S4 Data. LDA, linear discriminant analysis; LiCl, lithium chloride.

The fact that this decrement in discriminability extends into the “late” epoch led us to ask whether LiCl-induced illness might also change the palatability-relatedness of coding as well (a feature that is known to emerge at this point in GC taste responses; see [30,31]). We began with the simple hypothesis that reduced discriminability should imply reduced palatability-relatedness, testing this hypothesis in the standard manner [32,34,51,53]—i.e., calculating moving window correlations between firing rates and the known canonical palatability ranks of the administered taste stimuli.

As has been observed many times previously, palatability-related information in our GC taste responses climbed (regardless of condition) across the period leading into the late epoch (; note that we currently have no explanation for the small, low magnitude, but significant saline-LiCl session difference in palatability-related firing in the earliest responses, but see Discussion); despite the well-known, oft-commented upon variability of sensory neural activity [75,76], this property of GC taste responses shines through reliably in assessments of large numbers of neurons and across multiple animals. The climb calculated following saline and LiCl injections diverged significantly; however, particularly as the late epoch was reached—peak correlations following LiCl injections were higher than those observed in saline condition within this epoch (H(1) = 8.68, p = 0.003). This means that our initial hypothesis was disconfirmed: Whereas illness reduces taste response discriminability, it increases late epoch taste response palatability-related content.

The above results beg the question “how can illness reduce information pertaining to identity while enhancing the palatability-relatedness of the same taste responses?” We hypothesized that these seemingly contradictory results could be reconciled if (and only if) illness causes GC coding to collapse toward a simple “good versus bad” judgment, preferentially decreasing the differences between the coding of tastes with similar palatabilities—making sucrose and NaCl (the palatable tastes) responses more similar and making quinine and citric acid (the aversive tastes) more similar—and leaving GC responses closer to a simplistic, “pure” code of whether or not a taste is palatable. We tested this hypothesis by quantifying the distance in Euclidean space between neural responses for all pairs of similar (e.g., sucrose and NaCl) and dissimilar (e.g., sucrose and quinine) taste stimuli.

Specifically, we calculated what we term a “pure palatability index”—the difference of the distances (Dist−Dist) between responses to “different-palatability” and “same-palatability” tastes. By this analysis, the more polarized the response into “good versus bad”—the larger the difference—the more positive the PPI; a PPI of one (or greater) would indicate that GC neural responses are determined purely on the basis of whether the taste in question is pleasant or aversive. shows the results of this analysis. One-sample t tests comparing each PPI to a null result revealed that the PPI reaches significance only after the early epoch in both conditions (ps < 0.01); meanwhile, a Wilcoxon signed-rank test (performed because the data differed significantly from normal, p >0.05) revealed that LiCl-induced illness significantly enhances the PPI in the late epoch (W = 68379, Z = 0.54, p = 0.037, r = 0.09), reflecting enhanced polarization of the coding of palatable and aversive tastes. Thus, our second hypothesis was confirmed: Illness shapes the coding of taste hedonics by enhancing the polarization between similar and dissimilar tastes in the late epoch GC taste responses, and thereby simultaneously increases the overall palatability correlation and reduces the average discriminability of the responses.

Illness-related changes in taste coding occur in single neurons

The above results suggest that ensembles of GC neurons assayed approximately 20 min after LiCl administration code tastes differently than ensembles of GC neurons assayed in healthy rats. The implication of these results is that coding in individual neurons changes as sickness emerges, but the analyses above provide only indirect evidence for this implication; they stop short of directly testing whether single-neuron responses truly change with the onset of illness.

We therefore moved on to performing this direct test—2 such tests, in fact. First, we collected a new dataset using a modified experimental protocol in which we administered (and acquired spiking responses for n = 74 GC neurons to) tastes starting immediately after the injection of LiCl (). As already established, illness-related behaviors and changes in GC μ power emerge by 20 min after LiCl injection (Figs ); we therefore hypothesized that the above-described coding changes would emerge within single ensembles at approximately this same time point—that coding in trials delivered before sickness onset (“Pre”) would differ from that in trials delivered after sickness onset (“Post”; ).

We again used LDA to test whether and when taste response coding changed by testing the similarity of the first 5 “healthy” trials of each tastant to each subsequent taste delivery (binned; 5 trials/taste/bin). As expected, a repeated measures ANOVA performed between bins (bin = 6 min, approximately 3 trials/min) on taste responsive neurons (n = 56) revealed that the classifiability of tastes decreased significantly as rats became ill (F(4, 8) = 5.42, p = 0.02, np2 = 0.64), with this disruption becoming significant approximately 30 min into the session (, ps < 0.05). This result is a fairly good match for the above-described changes in GC μ power and illness-related behaviors (Figs ), particularly given the necessarily small sample of trials. While one could speculate that this effect is driven simply by time (as opposed to illness), this explanation is rendered unlikely by previous results demonstrating that, in the absence of illness, within-session taste discriminability does not vary significantly across much longer time spans [26]. The far more likely explanation is that taste responses change following the onset of illness.

We went on to more closely examine the precise nature of these changes in the subsample of our recorded neurons that were held across sessions—neurons in which we were able to assay responses after both LiCl and saline injections (see Methods for criteria). For each of 44 held neurons, we first examined session-specific firing rates for the 500 ms before and after taste delivery (starting after the initial 200 ms, which, as expected, was shown to be devoid of taste information in paired t tests). Visual examination of the data suggests that pre-stimulus firing rates in these neurons were not affected by LiCl administration (note that the small dots, which plot pre-stimulus firing, form a cloud going up the diagonal of , which plots the data for 1 representative taste [CA]). A paired t test confirmed that the mean pre-stimulus activity did not differ between saline (M = 6.68, SD = 4.71) and LiCl (M = 5.46, SD = 5.02) sessions (t(43) = 1.46, p = 0.15), and a Kolmogorov–Smirnov test confirmed the lack of a condition-specific impact on baseline firing rates (D(44) = 0.045, p = 0.98; ), revealing that saline-LiCl difference in pre-stimulus firing rates were normally distributed and centered around 0.

Further scrutiny of , meanwhile, suggests 3 distinct “types” of impact. Cluster analyses confirmed these appearances, revealing 3 distinct and discrete clusters of taste-response changes wrought by sickness (identified with distinct colors attached to neurons in each cluster): (1) in some neurons, taste responses were excitatory in saline sessions but inhibitory during LiCl sessions; (2) in others, the reverse was true; and (3) in a third group of neurons, responses were excitatory in both states, but these excitatory responses were generally less strong following LiCl injections. Ancillary analysis confirmed the presence of these clusters in the full dataset, showing 3 peaks in the angle of the vectors ( the latter with representative examples of each type, as well as of the few neurons that fail to cluster).

These 3 clusters of neurons, which were sorted without regard to the types of information within the taste response, nonetheless differed with regard to how illness-inducing LiCl impacted palatability-relatedness in the dynamics of the taste responses. A 3-way nonparametric mixed ANOVA run on the palatability content of all neurons represented in revealed a significant interaction between cluster, epoch, and condition, F(4, 478) = 4.80, p = 0.001, indicating that the relationship between cluster (whether the neurons were in cluster 1, 2, or 3) and the coding of palatability within epoch was significantly different across the saline and LiCl conditions. Specifically, the late epoch enhancement of palatability-related firing described earlier proved to be primarily a function of one group of neurons—those for which normally excitatory responses were turned into inhibitory responses by LiCl-induced illness (F(1, 130) = 33.47, p = 5.12 × 10−8; ). This result both confirms the validity of the separation of neuron types and reveals that this separation is related to the impact of illness state on palatability processing.

Discussion

Nutritional requirements, environmental conditions, and experience work together to guide consummatory behavior [2,11,13], in cooperation with an animal’s internal state (e.g., hunger, illness) [1]. The study of how physiological states impact perception is an important one, because this relationship intrinsically controls the animal’s likelihood of survival. The fact that neurons are embedded in networks, and that the states of such networks can be indexed in terms of EEG power (particularly in θ and μ frequency ranges [25,26,62]), motivated our decision to characterize changes in LFP activity within GC as an animal shifts from a healthy state into one of emesis, and to then relate this shift to changes in the dynamics of taste coding. This analysis reveals a modulation of μ power occurring approximately 15 min after LiCl injection—a result consistent with human [77] and rat data [78].

The finding that rearing durations, a measure of healthy exploratory behavior [2,70,71,78-80], drop at approximately the same time is consistent with decades of work showing that poisoning-related behaviors, and specifically those brought on by systemic nausea, emerge (depending on dose) at roughly 10 to 15 min post administration [2-4,78]. In our hands, the emergence of this LiCl-induced behavioral change closely mirrors the change in GC μ power induced by the same LiCl dose, implicating the process whereby an animal experiences illness in generation of these neural changes.

Of course, the comparable timing of GC LFP activity and the emergence of sickness (as indicated by behavioral changes) cannot exclude the possibility that changes in GC activity could conceivably reflect, not illness per se, but rather a direct effect of LiCl on GC neurons. While unlikely, this alternative explanation would persist as a possibility even if we had shown illness-related behaviors and GC changes in the same rats (we used separate groups because the intentional subtlety of the illness-related rearing changes, which allowed us to avoid the possibility of GC changes reflecting massive changes in oral fluid handling, cannot be observed in rats attached to the electrophysiology/IOC harness). To completely disapprove this hypothesis, an additional set of experiments would be required, in which we somehow block every pathway whereby visceral inputs carry information regarding LiCl-induced gastrointestinal distress and test whether this manipulation also blocks cortical changes. In the absence of the huge expenditure of time and resources required for these experiments, the best solution involved testing a particularly rigorous and risky prediction: Rather than just predicting that LiCl would change cortical activity, we specifically predicted that LiCl would change cortical activity at about the same latency that it caused illness (measured in rearing behavior). Given the success of this prediction (see Figs 2, 3, and 7), the most likely explanation (by far) is that the 2 are linked—that LiCl-induced illness is reflected in cortical activity.

Fig 3

LiCl administration reduces rearing approximately 12 min after LiCl injection.

(A) An example rat’s movement through time around the cage space (color coding shows the rat’s position in the box, Quadrant 1–4) overlain with rearing (black bars) durations after saline (top) and LiCl injections (bottom). (B) The average normalized LiCl–saline difference in rearing duration across animals (N = 4 and 5, saline and LiCl) reveals (top) a significant reduction of rearing occurring in the fourth quarter of the session (* p < 0.05, Wilcoxon-rank sum). A more fine-grained analysis (bottom) suggests that the change in behavior (black line) begins at 12-min post-injection; the dashed red line illustrates the sigmoidal fit (r2 = 0.72). The error bars represent SEMs. Data underlying this figure can be found in S2 Data. LiCl, lithium chloride.

The change in GC μ power is phasic, meaning that it is not simply the case that health connotes one μ amplitude and illness a different μ amplitude. This concept is not novel—while body states can be indexed in terms of LFP power fluctuations, behavioral states (e.g., sleep/wake) and muscle movements do not always correspond with a particular amplitude of field potential discharge [81] nor do significant changes in LFP power necessarily imply an observable change in body states [82]. Like many dynamical systems, the cortex experiences transient periods of instability at the time of state changes; we postulate that the observed transient impact of LiCl administration on GC μ power signals the onset of an illness state, and that the relaxation to a lower power regime shortly (approximately 7 min) thereafter nonetheless leaves the network changed in a way that impacts the processing of internal and external stimuli. In this, GC is akin to an automobile engine: Shifting from one gear to the next requires a brief transition from a steady state to a transient and then back to a steady state for optimal efficiency [83,84].

In addition to altering μ power, LiCl also significantly altered GC taste responses. This fact dovetails nicely with data presented by Arieli and colleagues [85], who examined how GC activity changes following CTA learning, identifying 2 types of impact resulting from the pairing of taste with LiCl-induced illness—an “immediate” impact of LiCl administration on taste responses in GC single neurons and a “delayed” impact on GC ensemble population dynamics in taste response. Arieli and colleagues hypothesized that the late impact is driven by CTA formation, and that the immediate impact is likely caused by the illness state induced by LiCl injection. Our current findings are consistent with theirs.

We went on to show that the impact of LiCl administration on taste coding is epoch dependent. The drug disrupted identity coding while (somewhat incongruously) enhancing the palatability-relatedness of the responses, with the latter effect localized to the oft-described late epoch of the responses [30-32]. We also observed a small but significant illness-related increase in palatability coding in the early epoch. While an explanation for this unexpected result awaits further experimentation, we would speculate, based on work showing that expectation of stimulus availability can reduce that latency of gustatory coding (effectively eliminating the 200 ms non-chemosensory period observed in our work, see [38,46-50]), that illness might similarly enhance the “readiness” of cortex and reduce the latency of taste-specific coding; since illness biases GC toward palatability-related coding (see below), it could very well be palatability coding that appears early when the animal is ill.

These findings may at first blush appear to be in conflict with previous studies examining behavioral changes following LiCl-induced illness [86-88], which suggested that sickness has no effect on palatability (e.g., as measured via evaluation of taste reactivity to a single taste [86]). In fact, however, our findings regarding sickness polarizing palatability processing are largely consistent with these results, in that they do not suggest wholesale changes in palatability of individual tastes, or even changes in the order of preference. Rather, we argue that illness simplifies taste coding, making aversive tastes more similarly aversive and palatable tastes more similarly palatable.

The results described in Figs 4–6 suggest, but do not prove, that single neurons change their coding as a function of illness. It is always possible that neurons coding taste identity and palatability during illness represent a distinct set of neurons from those coding these properties in healthy rats, and that the differences in overall sample responses reflected the addition of this new set of “illness-only” responses. In the first test of the hypothesis that illness changes the processing of taste stimuli in individual neural ensembles, we showed that taste discriminability is significantly changed as an animal enters the emetic state. And since the small number of trials available in this analysis rendered it impossible to reliably compare palatability correlations, we performed a second analysis in which we leveraged our ability to hold a subset of our single neurons across both testing sessions. This analysis revealed that LiCl-induced illness enhances palatability-relatedness in an epoch-specific manner but went beyond this to reveal 3 distinct subpopulations of illness-related response profiles.

We cannot, as of yet, provide a definitive explanation of this functional dissociation of GC neurons, but we can speculate as to several mechanisms that could explain our results. The most parsimonious of these explanations might be that the clusters come from either functionally or physiologically distinct populations of cells which in turn code specific features about the state of the animal (and thus responses to external stimuli) in unique, yet beneficial ways. However, cross-correlational analysis (methods adapted from Li and colleagues [54]) failed to reveal differences in the strength of the functional connectivity between (nor within) the individual clusters, which would have been expected with such an explanation [54]. Furthermore, analysis of wave shapes and basal firing rates failed to reveal groups to be made up of different percentages of putative pyramidal and interneurons. While it is always risky to reach conclusions on the basis of a null result, particularly in small datasets, we find it unlikely that these clusters represent distinct types of GC neurons.

An alternative explanation for the observed differences in palatability coding across these putative clusters has to do with the possible sources of input to these neurons. It has recently been shown that specific area postrema neuron types are responsible for providing illness-related information to brain regions [18], and that these influences may reach GC by way of either the nucleus of the solitary tract [89] or other regions (e.g., amygdala) known to be impacted by the physiological state induced by LiCl [90]. It is possible that the palatability-coding differences between GC clusters are a downstream result of cell type–specific projections. Future work will examine this possibility, incorporating the use of video recordings of oral-facial reactivity to establish the link more concretely between these putative clusters and their behavioral relevancy in consummatory behaviors.

Regardless of the answers to these questions, the work presented here demonstrates that illness, or at least that caused by LiCl, impacts cortical function in a far more refined manner than via a simple wholesale disruption of activity. The effects of illness on GC activity are complex, but this complexity seems linked to an animal’s need, when ill, to simply tell good from bad. In fact, this is not the first case we have found a manipulation that biased GC response toward stimulus palatability. Fontanini and Katz [26] also found that when the behavioral state of an animal changed from a task-oriented to a disengaged (or inattentive) state, GC response became less discriminative between palatability-similar tastes but more distinct for tastes with opposite valences. The similarity in impact of illness and engagement states provides further support for the general finding that GC is involved in the making and putting into action of decisions related to palatability [32,91]: While the most basic circuit involved in these decisions is found in the brainstem [73,74,92], the taste system is very much like other vertebrate and invertebrate sensorimotor systems [92]—in situ, top-down modulation plays a major role in determining behavior. Here, GC plays that role, and this is possibly why body states are reflected in GC as well.

S1_Data.xlsx includes the data underlying Fig 2.

(XLSX)

Click here for additional data file.

S2_Data.xlsx includes the data underlying Fig 3.

(XLSX)

Click here for additional data file.

S3_Data.xlsx includes the data underlying Fig 4.

(XLSX)

Click here for additional data file.

S4_Data.xlsx includes the data underlying Fig 5.

(XLSX)

Click here for additional data file.

S5_Data.xlsx includes the data underlying Fig 6.

(XLSX)

Click here for additional data file.

S6_Data.xlsx includes the data underlying Fig 7.

(XLSX)

Click here for additional data file.

S7_Data.xlsx includes the data underlying Fig 8.

(XLSX)

Click here for additional data file.

10 Jan 2022

Dear Dr Katz,

Thank you for submitting your manuscript entitled "LiCl-induced sickness modulates spontaneous activity and response dynamics in rat gustatory cortex." for consideration as a Research Article by PLOS Biology.

Your manuscript has now been evaluated by the PLOS Biology editorial staff, as well as by an academic editor with relevant expertise, and I am writing to let you know that we would like to send your submission out for external peer review. Please accept my apologies for the delay in communicating this decision to you. We are working through our post-holiday backlog.

Before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed the checks it will be sent out for review. To provide the metadata for your submission, please Login to Editorial Manager (https://www.editorialmanager.com/pbiology) within two working days, i.e. by Jan 12 2022 11:59PM.

If your manuscript has been previously reviewed at another journal, PLOS Biology is willing to work with those reviews in order to avoid re-starting the process. Submission of the previous reviews is entirely optional and our ability to use them effectively will depend on the willingness of the previous journal to confirm the content of the reports and share the reviewer identities. Please note that we reserve the right to invite additional reviewers if we consider that additional/independent reviewers are needed, although we aim to avoid this as far as possible. In our experience, working with previous reviews does save time.

If you would like to send previous reviewer reports to us, please email me at ggasque@plos.org to let me know, including the name of the previous journal and the manuscript ID the study was given, as well as attaching a point-by-point response to reviewers that details how you have or plan to address the reviewers' concerns.

During the process of completing your manuscript submission, you will be invited to opt-in to posting your pre-review manuscript as a bioRxiv preprint. Visit http://journals.plos.org/plosbiology/s/preprints for full details. If you consent to posting your current manuscript as a preprint, please upload a single Preprint PDF.

Given the disruptions resulting from the ongoing COVID-19 pandemic, please expect some delays in the editorial process. We apologise in advance for any inconvenience caused and will do our best to minimize impact as far as possible.

Feel free to email us at plosbiology@plos.org if you have any queries relating to your submission.

Kind regards,

Gabriel

Gabriel Gasque

Senior Editor

PLOS Biology

ggasque@plos.org

7 Feb 2022

Dear Dr Katz,

Thank you for submitting your manuscript "LiCl-induced sickness modulates spontaneous activity and response dynamics in rat gustatory cortex." for consideration as a Research Article at PLOS Biology. Your manuscript has been evaluated by the PLOS Biology editors, by an Academic Editor with relevant expertise, and by four independent reviewers.

In light of the reviews (below), we will not be able to accept the current version of the manuscript, but we would welcome re-submission of a much-revised version that takes into account the reviewers' comments. We cannot make any decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is also likely to be sent for further evaluation by the reviewers.

We expect to receive your revised manuscript within 3 months.

Please email us (plosbiology@plos.org) if you have any questions or concerns, or would like to request an extension. At this stage, your manuscript remains formally under active consideration at our journal; please notify us by email if you do not intend to submit a revision so that we may end consideration of the manuscript at PLOS Biology.

**IMPORTANT - SUBMITTING YOUR REVISION**

Your revisions should address the specific points made by each reviewer. As you will see, the reviewers agree that your study is potentially important but point out a few issues that should be addressed. Most importantly, there is a general lack of clarity in the description of the results and methods. The referees also make a number of suggestions to further improve the manuscript and its conclusions. There is considerable overlap between reviews, and you should pay special attention to all major comments raised by reviewer 1, as well as points 3 and 4 by reviewer 2 (point 4 is related to reviewer 1's point 3).

Please submit the following files along with your revised manuscript:

1. A 'Response to Reviewers' file - this should detail your responses to the editorial requests, present a point-by-point response to all of the reviewers' comments, and indicate the changes made to the manuscript.

*NOTE: In your point by point response to the reviewers, please provide the full context of each review. Do not selectively quote paragraphs or sentences to reply to. The entire set of reviewer comments should be present in full and each specific point should be responded to individually, point by point.

You should also cite any additional relevant literature that has been published since the original submission and mention any additional citations in your response.

2. In addition to a clean copy of the manuscript, please also upload a 'track-changes' version of your manuscript that specifies the edits made. This should be uploaded as a "Related" file type.

*Re-submission Checklist*

When you are ready to resubmit your revised manuscript, please refer to this re-submission checklist: https://plos.io/Biology_Checklist

To submit a revised version of your manuscript, please go to https://www.editorialmanager.com/pbiology/ and log in as an Author. Click the link labelled 'Submissions Needing Revision' where you will find your submission record.

Please make sure to read the following important policies and guidelines while preparing your revision:

*Published Peer Review*

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out. Please see here for more details:

https://blogs.plos.org/plos/2019/05/plos-journals-now-open-for-published-peer-review/

*PLOS Data Policy*

Please note that as a condition of publication PLOS' data policy (http://journals.plos.org/plosbiology/s/data-availability) requires that you make available all data used to draw the conclusions arrived at in your manuscript. If you have not already done so, you must include any data used in your manuscript either in appropriate repositories, within the body of the manuscript, or as supporting information (N.B. this includes any numerical values that were used to generate graphs, histograms etc.). For an example see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5

*Blot and Gel Data Policy*

We require the original, uncropped and minimally adjusted images supporting all blot and gel results reported in an article's figures or Supporting Information files. We will require these files before a manuscript can be accepted so please prepare them now, if you have not already uploaded them. Please carefully read our guidelines for how to prepare and upload this data: https://journals.plos.org/plosbiology/s/figures#loc-blot-and-gel-reporting-requirements

*Protocols deposition*

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for your submission to our journal. We hope that our editorial process has been constructive thus far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Gabriel

Gabriel Gasque

Senior Editor

PLOS Biology

ggasque@plos.org

*****************************************************

REVIEWS:

Reviewer #1: Comments:

LiCl-induced conditioned taste aversion is a fundamental mechanism to study taste-visceral interactions. However, how malaise impacts cortical processing of this sensory information is unknown. This report provides novel insights into the effects of LiCl-induced malaise on neural activity in the gustatory cortex. The authors show that spontaneous changes in LFPs at the µ frequency (7-12 Hz) arise on a similar timescale as the onset of illness-related behaviors. Furthermore, LiCl-malaise altered single-unit responses in the gustatory cortex; with reductions in response magnitude and discriminability, but an enhancement in the representation of taste palatability. In summary, the authors detail how a change in state (malaise) that reduces consumption alters the representation of taste signals in sensory cortex.

While the question is important to the field and the authors interpretations are measured and interesting, I had a hard time grasping exactly what datasets were being analyzed throughout the Results. This made judging the impact of the findings extremely difficult. However, the authors can alleviate my concerns by addressing the following points.

Major Concerns:

1. Both in the text of the Results and the figures, there is a dearth of experimental detail related to the number of animals and/or neurons used in the analyses. As far as I could tell, only the section titled "Illness-related changes in taste coding occur in single neurons", and its corresponding figure (Fig. 7), lists the number of animals (n=3), number of total recorded neurons (n=74), and the subset of taste responsive neurons (n=56) used for the analyses. It is entirely unclear to me what populations of single-units are being compared in the section titled "Illness causes taste responses to resemble a good-bad distinction" (Fig. 4, Fig. 5, and Fig. 6). Are these populations from the same rats, but different days? If so, are they recorded at the same depth? Are the analyses performed across all neurons, only taste responsive neurons, or only the taste responsive neurons with stable waveforms across days (putative single-neurons)? What constitutes a "change in firing rate"? Is this a neuron by neuron comparison in taste-evoked activity between LiCl- and Saline-sessions? Or is this a comparison between all recorded neurons from the different sessions? Is baseline firing rate taken into account, because a 5Hz change in activity may impact the system more for a neuron with a 5hz baseline than one with a 50Hz baseline. The lack of specific information impedes a clear understanding of the results and mars the author's interpretations

2. Relatedly, the presentation of the data would be significantly enhanced by the addition of representative examples to the figures describing the single-unit activity (i.e. Fig. 4, Fig. 6) and especially the figures showing changes in single-unit activity within a session (Fig. 7 and Fig. 8).

3. Would the authors please explain why the first 300ms of the taste evoked activity is removed for some analysis (i.e. directionality), but then used as a key component of the analyses of taste epochs. The timeframe of 300ms seems questionable, given the many studies from this group showing that taste identity begins to emerge ~200ms after stimulus delivery (Katz et al., 2001; Fontanini and Katz, 2006; Sadacca et al., 2012; Sadacca et al., 2016). What is the justification for ignoring data for one analyses, but including it for others? Does including the data eliminate the significant difference in the directionality analyses?

Minor concerns:

1. I agree that it's interesting the illness-related behavior from a separate group of rats correlates with the onset of changes in LFP. However, can the authors provide any evidence that the rats in the electrophysiological group experienced malaise? Especially since only 5 of the 7 exhibited a significant change in LFP? For example, LiCl-induced malaise can evoke a number of aversive oromotor and somatic responses (Spector et al., 1988), including gape, chin rub, head shake, and forelimb flail. Did either group (ephys. or behavior) perform more of these behaviors after LiCl-induced malise compared to Saline controls?

2. For the analyses represented by Figure 8, the authors say "pre-stimulus firing rates fall generally along the diagonal, demonstrating that spontaneous firing rates do not differ strongly between condition". Individual variability aside, do the spontaneous firing rates of putative single neurons (as a population) differ between the LiCl- and Saline-sessions? Statistical confirmation would further support the stability of the population across sessions.

3. The authors should take care when describing activity suppressed by taste delivery as "inhibitory". I have received this same criticism in the past because some readers wrongly assume that the authors are suggesting that these neurons are themselves inhibitory.

4. The manuscript would benefit from a spelling and grammar check, as there were a number of typos throughout the article.

Reviewer #2: In this paper, Stone et al., made electrophysiological recordings from the gustatory cortex (GC) of healthy

and ill (due to LiCl injections) female rats. They evaluated how a behavioral state (illness) alters the way GC

neurons represent the chemosensory and hedonic properties of tastants delivered via intra-oral cannulae. Overall,

the results show that being in a sickness state changes basic properties of the sensory responses to basic taste

qualities. Many others reports (also from the author's laboratory) have already showed that GC responses are

plastic and are strongly affected by behavioral states (even behavioral states less "salient" than illness behavior).

Nevertheless, this is a fine paper that describes how GC neurons alter their taste-evoked spiking activity after

LiCl injection - changes that somehow correlate with the onset of sickness behavior. The authors might consider

the following points:

1 - It would be helpful to know some more information about the recording strategy to better interpret the data

presented in figures 4-6. That is, were the electrodes lowered after each NaCl and LiCl recording sessions to

record new ensembles of GC neurons? If that is the case, I have two main concerns: 1) I wonder if the results

observed in figures 4-6 are affected by potential differences in the sensory properties of different neurons across

the GC (the heterogeneity of GC responses within a recording session is highlighted by the authors in Fig 8;

thus, we can expect even more heterogeneity across sessions and across animals). 2) While I do understand the

challenges of an experimental design in which the same neurons are recorded (and held) during and after NaCl

and LiCl injections, I am also convinced that the current approach may bias the results shown in figures 4-6.

For instance, I found that the data presented in figures 7 and 8 (even without the NaCl session) are much more

relevant to the overall (i.e., sickness changes the response profile of GC neurons) interpretation of the data.

2 - I am somewhat confused by the criteria used to evaluate taste responsiveness. In the method session, the

author indicated that "Data were collapsed all four tastes, such the ""responsivity"" indicates purely that a GC

neurons responds to taste delivery and reveals nothing about taste specificity". If all the evoked responses to the

different tastes are collapsed, what will happen if 1) a neuron shows an increase (or decrease) in response to only

one taste and/or 2) a neuron is excited by two tastes and inhibited by the other two? What are the chances that

these two hypothetical neurons will be deemed non-responsive based on your paired sample t-test (pre vs. post

taste) after collapsing all the tastes?

3 - Figures 5 and 6 showed how sickness reduces taste discriminability (across all tastes; Fig. 5) and increases

the palatability-relatedness (Fig. 6) of taste responses. I do not understand why the data in Figure 6 are not analyzed

with the same analysis showed in Figure 5 (LDA). The author could have combined sweet-salt (good) and

acid-bitter (bad) and run a time series of LDA to show that sickness allows the GC neurons to better discriminate

good vs. bad taste, as per the main conclusion of the manuscript. Using different analyses to run "chemosensory"

(Figure 5) and "palatability" (Figure 6) coding is confusing, to say the least.

4 - When evaluating how LiCl changed the "sign" of taste responses, the authors decided to ignore the first 300 ms

of post taste activity. They justified this choice by indicating that evoked responses within 300 ms after stimulus

delivery do not contain chemosensory information. This is absolutely true in the case of the many reports from

the authors (and others) when tastes are delivered intra-orally. However it is not the case when stimuli are licked

from a spout (Stapleton 2006, 2007; Gutierrez 2010; Graham 2014; Bouaichi 2020; Dikecligil 2020). I suggest

that the author highlight that 300ms of post-taste activity is seldom chemosensory in the context of intra-oral

cannulae delivery. Also, this seems like a lost opportunity. If there is no chemosensory information in that time

window, the author should use it as a control and show that there is no change in the sign of the responses in that

early epoch in LiCl.

5 - One of the main results of this paper is that, sickness causes the firing properties of GC neurons to collapse

to a pure good vs. bad code of whether or not a taste is palatable. Would this be behaviorally relevant? Is the

cortex (or the entire forebrain for that matter, [see for example Grill and Norgren 1978]) needed to compute such

basic and salient sensory information? Could the authors comment on these points in the Discussion section?

6 - There is almost no information on how the single neurons are sorted. Is the SNR the only quality metric

used? I think it would be better to add some more details even if they are described elsewhere. Likewise, I would

replace all the "data not shown" with the actual data.

7 - It would be helpful to see some examples of single neuron activity (rasters and PSTHs) associated with

either Figure 3 or 4.

8 - The author provide a relatively low-quality image of one "representative" electrode track (please, do add a

scale bar). It might be beneficial to show all the electrodes tracks to give the reader a better sense of the recording

locations.

9 - It is not clear how many single units were recorded per animal and per session (NaCl and LiCl). The

authors should provide this information.

Minor:

1 - The description of Fig. 2B in the text is wrong.

2 - Overall, the manuscript is a pleasure to read because it is very well written. There are a few places where I

can find typos but these are super-minor (e.g., decide to use "μ" or "mu" and be consistent).

Reviewer #3: Stone et al., recorded single unit activity and LFP from IC of rats in correlation with CTA formation (i.e. before and during drinking novel taste and before and after LiCl injection i.p. as UCS). They report alteration in LFP power 11 min following the injection together with reduction in single neuron taste response. In addition, they report for reduced discriminability and enhancement in palatability. This is interesting novel and an important work. The research is based heavily on previous publication from the same lab. The authors recorded from small numbers of rats but it seems that the rigor analysis they did justify such a small number. A major point the authors do not take into account in all figures and analysis that it is not only the presentation of the taste and its valance! One would assume that a learning process of an association between malaise and taste takes place also within the very same neurons they record from. Its not only about taste but about prediction and association. This make the interpretations for the data more complex (I propose they will refer to this in the discussion). Nevertheless, the manuscript is important and innovative.

Major

1. The authors refer to taste cortex; this is not a clear term. I suggest they will talk about the relevant area in the insular cortex: Left? Right? anterior? Agranualr? disgranular? granular?, inner or outer layers. We know much more about subdivisions within the IC that should be take into account.

2. What does it mean that half cases amplitude of activity was reduced and half was increased?

3. Can you call the measured activity spontaneous?

4. Does the behavioral facial expression the authors recorded can be analyzed as in Gogola to better define sickness state?

5. The fact that data for fig 2 and 3 was collected from different groups is less convincing. The difference in time for changes in power or behavior within groups could use to validate the results. One would assume that first will appear the change in power and only later in behavior. Data from each rat could validate this point (i.e. the connection between change in behavior and change in power).

6. Fig 4, though the authors claim only the middle epoch was sig different between saline and LiCl, all epochs show same direction and about the same size? Can the authors claim for specificity of the middle epoch?

7. Following the above in Fig 4, why the authors did not analyze the time scale between 1 and 1.5 second? I would assume they find also a sig difference in fraction of trials correct at the "post" late epoch. What does it mean?

Minor

1. Fig 1 and 2 can unify.

2. The author should present all brain imaging following all treatments.

Reviewer #4: In this manuscript, Stone et al. assess the impacts of LiCl induced illness on neural activity in the gustatory cortex (GC). They found that LiCl injections resulted in a robust change in the mu frequency range, ~15 min post injection, within the local field potentials (LFP), which corresponded to a change in the activity of the rats as assessed by the presence of rearing behavior. The authors then analyzed taste-evoked activity within GC following illness and found subtle, yet meaningful, effects on GC activity. Notable changes included a reduction in the number of significantly inhibited neurons, a reduction in taste discriminability within GC neurons, and an increase in taste palatability. The overall decrease in discriminability and increase in palatability appear to be very robust. These findings, along with the distinction of a specific cluster of neurons within GC that seem to be most strongly affected by the illness, are very intriguing and cast substantial light onto the general process of illness-induced activity changes in the brain.

This work provides an excellent backdrop for understanding the baseline effects of how a gastro-induced illness may influence activity within GC. What I find to be the most intriguing and insightful aspect of the design of this experiment, which the authors conspicuously did not raise in the introduction or discussion, is how it provides a fundamental starting point for studying learning of taste-illness pairings, or conditioned taste aversion. This experiment reveals how illness itself affects coding in GC without the taste-illness learning process interacting with the effects. More specifically, the changes in palatability coding in the absence of taste-illness learning is extremely intriguing and helps constrain the possible mechanisms that may lead to this type of learning. I'll leave it to the authors whether they feel it is worth discussing this in the paper.

In general the manuscript was well-written and analysis were on-point and well-performed, although there were a few parts in the manuscript where it was a bit unclear on the specifics of the design and the analyses performed. These are detailed below. Otherwise, my concerns revolve more around the discussion and interpretation of the results.

1. As mentioned above, I was surprised how the authors did not relate their findings to learning about CTAs. A recent paper by the Moran group looked at changes in GC processing following CTA. It seems appropriate to discuss their findings given what the authors found here. Specifically, these results indicate that several of the findings by the Moran group are not necessarily due to changes in learning, but are changes in activity induced by the illness that set the stage for learning. It seems that a re-assessment of previous results on changes in activity within GC during CTA, both from the Katz group as well the recent study from the Moran group, would be very worthwhile in light of the findings in this paper. Of course, a full consideration might require a separate review, but at minimum a mention seems pertinent to the discussion.

2. The analyses performed in reference to figure 4 are a bit confusing and may lead the reader down the wrong interpretation of the study. My general understanding was that the authors found no general changes in activity between conditions (the specifics of this analysis were a bit unclear, see below), although they did find changes in the percentages of neurons that were significantly inhibited/excited/unchanged. The authors then reanalyze changes in activity across these groups and find modest activity changes. As a reader, it seemed that the authors were stretching to show a significant effect on activity across the two conditions, yet this seemed to be a moot point after reading the rest of the paper as there were clear and robust changes in the coding of taste stimuli across conditions. It's not clear that figure 4 as a whole adds substantially to the paper and even perhaps acts as something of a red herring. The small changes in firing rates between the 'excited' groups is underwhelming, particularly when there are more neurons in the group with the smaller average. It seems some other analysis would fit better here. Maybe a comparison of the empirical cumulative distributions of the firing rates between conditions? As for confusion regarding the analysis performed:

What exactly were the authors comparing here? Is it baseline vs evoked activity for each neuron for each taste?:

'As an initial look at how illness impacts taste-evoked spiking activity of GC neurons, we evaluated the quantity and magnitudes of responses elicited by tastes delivered directly into the mouth via IOC. Paired-sampled t-tests (with Bonferroni correction for multiple comparisons) revealed that LiCl did not significantly impact the magnitude of responses in a taste-specific manner (ps <0.05).'

Wording here is confusing. By putting the (within condition) it sounds like the averages are for each condition. Is it that you took the average differences of activity, compared to baseline, for all tastes for each neuron and performed a Chi square test?:

"For this reason, we averaged taste responses (within condition) and a subsequent chi-square test of independence revealed a significant relationship between condition and direction of response on neuron count"

3. I'm not sure I understood the following argument. Could the authors provide a bit more clarification?:

'While one could speculate that illness' effect on GC μ power merely represents changes in firing rates (FRs) across time, a two-way ANOVA revealed a there was no significant interaction between the experimental group (saline-LiCl versus saline- saline) and quartile on change in FR (F(3, 33) = 2.39, p = 0.09); this suggests that our observed effect on μ is not simply a gating of firing rate direction, but rather a matter of synchrony being modulated.'

4. This may relate to the above. The authors make a clear point that the changes in mu power are transient, yet the observed effects in the paper raise the question about whether a good portion of the observations may be due to changes of evoked taste activity that occurs within mu. The reason I'm suggesting this is that prior work from the Katz group has shown that tastes delivered during mu results in a similar effect which is observed in this paper - that discrimination across similarly palatable tastes decreases while palatability coding increases. This may not necessarily matter to the main point of the paper, but if more tastes are delivered during the mu state, this result would certainly fit with prior work from the lab.

5. In Figure 1A, did the authors forget to include the control group, shouldn't there also be a group of rats with saline injections at the same time as the LiCl injections? If I'm not misunderstanding the design, I would certainly add this since readers might assume that all comparisons were to the NaCl injections on the second day.

6. Figure 2B references in the results section are flipped (top - control; bottom - experimental)

7. In the results, when discussing figure 2C, the authors mention 'subsequent testing'. What exactly was performed here?

8. The authors ran an LDA classifier and compared the results across conditions using a Mann-Whitney U test (page 9). My guess is that a classifier was run for each neuron rather than on a pseudo-ensemble, but it's not clear which is the case either looking in the results or the methods sections. If the latter, please add more clarification about how separate samples, and how many, were included in the Mann-Whitney U-Test.

9. Editing error: 'a two-way ANOVA revealed a there was no'

2 May 2022

Submitted filename: PLOS Authors response_Final.docx

Click here for additional data file.

24 May 2022

Dear Dr Katz,

Thank you for your patience while we considered your revised manuscript "LiCl-induced sickness modulates field potential spectral power and response dynamics in rat gustatory cortex." for consideration as a Research Article at PLOS Biology. Your revised study has now been evaluated by the PLOS Biology editors, the Academic Editor and the original reviewers.

In light of the reviews, which you will find at the end of this email, we are happy to continue consideration of your study at PLOS Biology but do not feel that additional revisions are still needed based on feedback we've received from Reviewers 1 and 2. In particular, Reviewer 1 continues to have concerns with the experimental set-up looking at distinct cohorts of rats, and has asked if you can provide more direct evidence of a change in behavior related to illness in order to rule out that LiCl itself is responsible for the changes in neural activity. We ask that you address this concern as best you are able, and address the additional issues raised by Reviewer 1 and Reviewer 2. We will then assess your revised manuscript and your response to the reviewers' comments with our Academic Editor. We will aim to avoid further rounds of peer-review, although we might need to consult with Reviewer 1 depending on the nature of the revisions.

We expect to receive your revised manuscript within 1 month. Please email us (plosbiology@plos.org) if you have any questions or concerns, or if you feel you would need an extension in order to fully address the remaining concerns.

At this stage, your manuscript remains formally under active consideration at our journal; please notify us by email if you do not intend to submit a revision so that we withdraw the manuscript.

**IMPORTANT - SUBMITTING YOUR REVISION**

Your revisions should address the specific points made by each reviewer. Please submit the following files along with your revised manuscript:

1. A 'Response to Reviewers' file - this should detail your responses to the editorial requests, present a point-by-point response to all of the reviewers' comments, and indicate the changes made to the manuscript.

*NOTE: In your point-by-point response to the reviewers, please provide the full context of each review. Do not selectively quote paragraphs or sentences to reply to. The entire set of reviewer comments should be present in full and each specific point should be responded to individually.

You should also cite any additional relevant literature that has been published since the original submission and mention any additional citations in your response.

2. In addition to a clean copy of the manuscript, please also upload a 'track-changes' version of your manuscript that specifies the edits made. This should be uploaded as a "Revised Article with Changes Highlighted " file type.

*Resubmission Checklist*

When you are ready to resubmit your revised manuscript, please refer to this resubmission checklist: https://plos.io/Biology_Checklist

To submit a revised version of your manuscript, please go to https://www.editorialmanager.com/pbiology/ and log in as an Author. Click the link labelled 'Submissions Needing Revision' where you will find your submission record.

Please make sure to read the following important policies and guidelines while preparing your revision:

*Published Peer Review*

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out. Please see here for more details:

https://blogs.plos.org/plos/2019/05/plos-journals-now-open-for-published-peer-review/

*PLOS Data Policy*

Please note that as a condition of publication PLOS' data policy (http://journals.plos.org/plosbiology/s/data-availability) requires that you make available all data used to draw the conclusions arrived at in your manuscript. If you have not already done so, you must include any data used in your manuscript either in appropriate repositories, within the body of the manuscript, or as supporting information (N.B. this includes any numerical values that were used to generate graphs, histograms etc.). For an example see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5

*Blot and Gel Data Policy*

We require the original, uncropped and minimally adjusted images supporting all blot and gel results reported in an article's figures or Supporting Information files. We will require these files before a manuscript can be accepted so please prepare them now, if you have not already uploaded them. Please carefully read our guidelines for how to prepare and upload this data: https://journals.plos.org/plosbiology/s/figures#loc-blot-and-gel-reporting-requirements

*Protocols deposition*

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for your submission to our journal. We hope that our editorial process has been constructive thus far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Kris

Kris Dickson, Ph.D. (she/her)

Neurosciences Senior Editor/Section Manager

PLOS Biology

kdickson@plos.org

----------------------------------------------------------------

REVIEWS:

Reviewer's Responses to Questions

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

Reviewer #4: Yes: Matthew Gardner

Reviewer #1: I thank the Authors for addressing the majority of my concerns. The improved manuscript was significantly easier to understand. However, I do not believe a key concern was adequately addressed. That is, the supposition that the injection of LiCl induced illness/sickness/gastrointestinal malaise in the rat's subjected to neural recordings.

I do not believe this assumption to be insignificant, since a majority of the results and discussion focus on the results from this group. I do not agree with the Authors position, as indicated by their response to a similar critique raised by Reviewer 3, that changes in neural activity and behavior in separate groups of rats is "more conservative and compelling when reached on the basis of datasets that were collected separately." What would be compelling is showing the relationship between changes in neural activity and behavior within the same animals. In the methods, the Authors state that "Three cameras (positioned below, above, and diagonally within the behavioral chamber (see Apparatus)" were used to record mobility in a separate behavior group. Were these cameras used to record behavior during neural recording sessions? Can the Authors do a similar analysis of rearing behavior (Figure 3) for the neural recording group?

Without evidence of a change in behavior related to illness, one could simply argue that LiCl itself is responsible for the changes in neural activity. If the Authors cannot provide evidence supporting their assumption that the rats in the neural recording group were in fact made sick by the LiCl-injection, they should temper their use of terms like "illness" or "sickness" when describing the effects induced by LiCl-injection. However, I will acquiesce if the Editor and other Reviewers feel that the behavioral data provided showing that a LiCl-injection modified rearing behavior in a separate group of rats is sufficient to support the declaration that the rats in the neural recording group were made sick by a similar injection of LiCl.

Additional concerns:

The newly provided representative PSTHs in Figure 4 raise some questions. The authors posit that taste information is not represented within the first 200ms. However, three of the four representative examples clearly show that tastes evoke different responses during the first 200ms. Furthermore, the authors state that there was no overall difference in baseline activity between Saline and LiCl sessions. However, the baseline activity for the inhibitory response shows a baseline response of ~6Hz on Saline day and ~18Hz on LiCl day, a three-fold increase. Is the y-axis on the LiCl day correct? It would be helpful to the reader if the PSTHs used the same scale.

It's possible that these examples are outliers and do not represent the population responses described by the Authors. If so, I suggest they choose different representative examples that align with the broader data set. If these examples are in fact representative, then the authors must reanalyze their data to account for the taste-evoked activity occurring within the first 200ms.

Reviewer #2: Overall, the authors have addressed my previous concern in an adequate manner.

It would be helpful to know some more information about the new figure 4. I must admit that this figure confused me for two reasons. In the legend, the authors stated that in panel A "...(PSTH) for pairs of representative neurons....". 1) If they are pairs, why are the y axes (for the two PSTHs in the top row and for the two PSTHs in the bottom row) not the same? 2) If these are representative neurons, why is the neural activity evoked by the different stimuli in the first 200 ms clearly different for 3/4 of the PSTHs? Based on the authors' claims (in this and other manuscripts), there should be no chemosensory activity in this early time window. If the stimulus-specific differences observed in the PSTHs of the new Fig. 4 are not chemosensory in nature, what can explain them? In the method section, the authors say that the activity in the first 200ms reflect tactile stimulation of the tongue. Thus, I was expecting to see the responses evoked by the different stimuli within the first 200ms to be mostly overlapping. I admit that this is not a major issue (unlike the concerns raised by two other reviewers about the lack of behavioral signatures of mild sickness, or malaise, in the electrophysiological group). However, in light of the new Fig. 4, the rationale of ignoring the first 200ms of post-taste activity for some analysis is weaker and can create confusion for the reader. One option could be to change the representative neurons showed in Fig. 4 or, even better, to include the first 200ms in all analysis.

Reviewer #3: The authors answered my questions.

Reviewer #4: The authors fully addressed all of my concerns.

8 Jun 2022

Submitted filename: Stone et al - Response letter 2.docx

Click here for additional data file.

17 Jun 2022

Dear Dr Katz,

Thank you for your patience while we considered your revised manuscript "Systemic administration of illness-inducing LiCl modulates field potential spectral power and response dynamics in rat gustatory cortex." for publication as a Research Article at PLOS Biology. This revised version of your manuscript has been evaluated by the PLOS Biology editors and the Academic Editor.

I am happy to inform you that, based on our editorial evaluation of your revisions, we are likely to accept this manuscript for publication. At this stage, we simply need you to satisfactorily address the necessary data and other policy-related requests.

In addition, we'd ask that you consider a possible title change to ensure the study is more broadly accessible to our wide audience. Maybe something along the lines of the following (appreciating that this might be too simplistic and not convey all of the nuances you are aiming for...):

Sickness modulates rat cortical taste responses

LiCl-induced sickness modulates rat gustatory cortical responses

As you address the data and policy items, please take this last chance to review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the cover letter that accompanies your revised manuscript.

We expect to receive your revised manuscript within two weeks.

To submit your revision, please go to https://www.editorialmanager.com/pbiology/ and log in as an Author. Click the link labelled 'Submissions Needing Revision' to find your submission record. Your revised submission must include the following:

- a cover letter that should detail your responses to any editorial requests, if applicable, and whether changes have been made to the reference list

- a Response to Reviewers file that provides a detailed response to the reviewers' comments (if applicable)

- a track-changes file indicating any changes that you have made to the manuscript.

NOTE: If Supporting Information files are included with your article, note that these are not copyedited and will be published as they are submitted. Please ensure that these files are legible and of high quality (at least 300 dpi) in an easily accessible file format. For this reason, please be aware that any references listed in an SI file will not be indexed. For more information, see our Supporting Information guidelines:

https://journals.plos.org/plosbiology/s/supporting-information

*Published Peer Review History*

Please note that you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out. Please see here for more details:

https://blogs.plos.org/plos/2019/05/plos-journals-now-open-for-published-peer-review/

*Press*

Should you, your institution's press office or the journal office choose to press release your paper, please ensure you have opted out of Early Article Posting on the submission form. We ask that you notify us as soon as possible if you or your institution is planning to press release the article.

*Protocols deposition*

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please do not hesitate to contact me should you have any questions.

Sincerely,

Kris

Kris Dickson, Ph.D. (she/her)

Neurosciences Senior Editor/Section Manager,

kdickson@plos.org,

PLOS Biology

------------------------------------------------------------------------

DATA POLICY:

You may be aware of the PLOS Data Policy, which requires that all data be made available without restriction: http://journals.plos.org/plosbiology/s/data-availability. For more information, please also see this editorial: http://dx.doi.org/10.1371/journal.pbio.1001797

Note that we do not require all raw data. Rather, we ask that all individual quantitative observations that underlie the data summarized in the figures and results of your paper be made available. We appreciate that you have indicated that data are fully available from the first or last authors. However, we require that the data also be provided in one of the following forms:

1) Supplementary files (e.g., excel). Please ensure that all data files are uploaded as 'Supporting Information' and are invariably referred to (in the manuscript, figure legends, and the Description field when uploading your files) using the following format verbatim: S1 Data, S2 Data, etc. Multiple panels of a single or even several figures can be included as multiple sheets in one excel file that is saved using exactly the following convention: S1_Data.xlsx (using an underscore).

2) Deposition in a publicly available repository (e.g. Zenodo etc). Please also provide the accession code or a reviewer link so that we may view your data before publication.

Regardless of the method selected, please ensure that you provide the individual numerical values that underlie the summary data displayed in the following figure panels as they are essential for readers to assess your analysis and to reproduce it:

Fig 2B,C; Fig 3B; Fig 4A-C; Fig 5; Fig 6A,B; Fig 7B; Fig 8a-E

NOTE: the numerical data provided should include all replicates AND the way in which the plotted mean and errors were derived (it should not present only the mean/average values).

***Please also ensure that FIGURE LEGENDS in your manuscript include information on where the underlying data can be found, and ensure your supplemental data file/s has a legend. (NB: This step is often forgotten!!!)

***Please ensure that your Data Statement in the submission system accurately describes where your data can be found.

------------------------------------------------------------------------

SPECIES INDICATED IN THE ABSTRACT?

- Please note that per journal policy, the model system/species studied should be clearly stated in the abstract of your manuscript.

------------------------------------------------------------------------

DATA NOT SHOWN?

- Please note that per journal policy, we do not allow the mention of "data not shown", "personal communication", "manuscript in preparation" or other references to data that is not publicly available or contained within this manuscript. Please either remove mention of these data or provide figures presenting the results and the data underlying the figure(s).

------------------------------------------------------------------------

Reviewer remarks: N/A

29 Jun 2022

Dear Dr Katz,

Thank you for the submission of your revised Research Article "LiCl-induced sickness modulates rat gustatory cortical responses" for publication in PLOS Biology. On behalf of my colleagues and the Academic Editor, Thorsten Kahnt, I am pleased to say that we can in principle accept your manuscript for publication, provided you address any remaining formatting and reporting issues. These will be detailed in an email you should receive within 2-3 business days from our colleagues in the journal operations team; no action is required from you until then. Please note that we will not be able to formally accept your manuscript and schedule it for publication until you have completed any requested changes.

Please take a minute to log into Editorial Manager at http://www.editorialmanager.com/pbiology/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with biologypress@plos.org. If you have previously opted in to the early version process, we ask that you notify us immediately of any press plans so that we may opt out on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

Thank you again for choosing PLOS Biology for publication and supporting Open Access publishing. We look forward to publishing your study.

Sincerely,

Kris

Kris Dickson, Ph.D. (she/her)

Neurosciences Senior Editor/Section Manager

PLOS Biology

kdickson@plos.org