Introduction

Dysfunction of the prefrontal cortex (PFC) is an underlying factor in both affect and cognition-related behavioral deficits that co-occur across neuropsychiatric disorders [1,2,3,4,5,6,7,8,9,10,11,12]. Similar symptomologies are observed in individuals with chronic psychosocial or self-perceived chronic stress [13,14,15,16]. Converging evidence indicates that prolonged exposure to environmental stressors can increase the risk and severity of these neuropsychiatric disorders, highlighting a need to identify neural substrates that contribute to optimal PFC function and behavioral control.

The prelimbic cortical subdivision (PrL) of the medial PFC (mPFC) is involved in top-down regulation of behavior related to anxiety, motivation, stress, and coordination of working memory and flexible decision-making (e.g., strategy shifting) [17,18,19,20,21,22,23,24,25,26]. Optimal function of the PrL relies on coordinated activity of principle output glutamatergic pyramidal neurons. This activity is critically dependent on a dynamic balance of cell excitation:inhibition mediated largely by intrinsic (physiological) membrane properties that influence the cellular response to synaptic input [27,28,29,30,31]. Accordingly, imbalances in pyramidal neuron excitation:inhibition are thought to contribute to numerous symptoms observed in neuropsychiatric disorders [4, 28, 32,33,34,35,36,37,38,39,40].

PrL pyramidal neuron excitability and spike firing is modulated by activity of G protein-gated inwardly rectifying K+ (GIRK/Kir3) channels which produce a slow hyperpolarizing current that acts as a neuronal “off switch” in both males and females [41,2d). These data indicate that reducing PrL pyramidal neuron GIRK1-dependent signaling in males reduces normal anxiety-like responses and escape-related strategies suggestive of an anxiolytic and pro-depressive phenotype [66, 67], without altering appetitive reward motivation.

Fig. 2: Impact of PrL pyramidal neuron GIRK knockout on affect-related behavior, working memory, and cognitive flexibility in males.
figure 2

a Schematic targeting of infusion to the PrL in male mice. b Percent time in the open arm of the EPM was increased in cre+ male mice compared to cre- controls. c Total time (s) spent immobile during the FST was increased in cre+ compared to cre- mice. d Breakpoint during a PR test was similar in cre+ males versus cre- controls. e During the forced alternation T-maze paradigm, a main effect of virus was found, with cre+ male mice having reduced percent correct choice. f Comparison of trials across days showed that Cre+ male mice performed significantly worse than cre- during trials 2, 7, 9, and 11. g–n Attentional set-shifting paradigm. g There was no difference between cre- and cre+ male mice in the number of days to reach lever training criterion. h During the visual cue test, cre+ mice took significantly more trials to reach criterion compared to cre- mice. i During the ED Shift test, cre- and cre+ male mice took similar number of trials to reach criterion. j During the reversal test, cre- and cre+ mice took similar trials to reach criterion. k Cre+ and cre- male mice did not differ in trials to reach criterion for visual cue training (l) Using an alternative response-to-cue paradigm, where the visual cue test is the ED shift, Cre+ male mice took significantly more days to reach lever training criterion. m Cre+ mice required greater number of trials to reach criterion during a response test. n Cre+ mice also took more trials to reach criterion during the subsequent ED Shift visual cue test. *p < 0.05, **p < 0.01. Star in schematics denotes correct response, yellow circle denotes cue, gray rectangles denote left and right levers.

Full size image

Working memory in PrL pyramidal neuron GIRK knockout male mice

The most consistently documented cognitive deficits in neuropsychiatric disease and stress pathologies includes impaired behavioral flexibility and working memory [68, 69], thus we examined whether loss of GIRK promotes similar impairments. Using a forced alternation T-maze paradigm to assess working memory, we found no effect of day (F(5,65) = 0.24, p = 0.95), or condition (virus) by day interaction (F(5,65) = 0.66, p = 0.65) on the percent of correct trials for each day. A main effect of condition was observed (F(1,13) = 22.46, p < 0.001), with cre+ male mice showing an overall reduction in percent correct choices compared to cre- controls (Fig. 2e). When averaging performance of each trial across days, there was a significant condition by trial interaction (F(11,143) = 2.10, p = 0.24; Fig. 2f), with post-hoc comparisons indicating that the cre+ male mice performed significantly worse on trials 2 (p = 0.004), 7 (p = 0.018), 9 (p = 0.034), 10 (p = 0.008), and 11 (p = 0.008). These data indicate that PrL pyramidal neuron GIRK1-dependent signaling in male mice plays a key role in information processing related to working memory.

Impact of PrL pyramidal neuron GIRK knockout on cognitive flexibility in male mice

To determine if PrL pyramidal GIRK channels play a role in complex forms of PFC-dependent cognition, we examined the impact of PrL GIRK1 suppression on cognitive flexibility. We used a modified operant-based ASST [61, 63] that is reliant on PrL function [24, 70]. This task resembles the Wisconsin Card Sorting Task in its sensitivity to distinct components of decision making such as suppression of irrelevant strategies, acquisition and generation of novel strategies, and maintenance of effective strategies. This is the first known study to investigate cognitive flexibility in GIRK1flox/flox mice, therefore we wanted to determine if GIRK1flox/flox mice exhibit a baseline phenotype. We first compared GIRK1flox/flox mice to C57BL/6 mice receiving sham surgery and found no difference in their performance on any measure in ASST (lever train: t(11) = −0.98, p = 0.35; VC trials: t(11) = −0.35, p = 0.74; ED trials: t(11) = −0.31, p = 0.76; REV trials: t(11) = −1.27, p = 0.23; data not shown), thus data were combined for further analyses. There was no difference in days to criterion for lever training between cre- and cre + (t(19) = −1.04, p = 0.31; Fig. 2g). Following lever training, comparison of performance in the visual cue test showed that cre+ male mice required significantly greater number of trials (t(19) = −2.52, p = 0.02; Fig. 2h) and errors (t(19) = −2.56, p = 0.019; Supplementary  1a) to reach criterion compared to cre- but did not differ on omissions (U = 45.00, p = 0.58; Supplementary Table 1). Further investigation of error type revealed no difference in initial errors (U = 51.00, p = 0.94) however the cre+ male mice had significantly more regressive errors compared to cre- control mice (t(19) = −2.87, p = 0.01; data not shown). During the ED Shift, cre- and cre+ mice required a similar number of trials (U = 28.50, p = 0.09; Fig. 2i), errors (t(19) = 0.95, p = 0.35; Supplementary 1b) and omissions to reach criterion (t(19) = 1.17, p = 0.26; Supplemental Table 1). Similarly, no difference was observed in trials (t(19) = −0.35, p = 0.73; Fig. 2j), errors (t(19) = −0.87, p = 0.40; Supplementary 1c) or omissions (t(19) = −1.42, p = 0.17; Supplementary Table 1) to criterion in a subsequent reversal learning test.

Past studies have shown that pharmacological inhibition of the PrL does not impact performance in an operant or maze-based visual cue discriminative learning task in rats [63, 71]. To ensure that impaired performance during the visual cue test in male GIRK1-knockout mice did not reflect deficits in general attention to a cue or visual acuity, we next trained a new cohort of mice on the visual cue test only (i.e., no lever training) with similar criterion to test sessions (i.e., 10 consecutive correct responses). During visual cue training, there were no differences in the trials (t(16) = 0.95, p = 0.36; Fig. 2k), errors (t(16) = 0.33, p = 0.75; Supplementary 2a), or omissions (U = 32.00, p = 0.48; Supplementary Table 2) to reach criterion between cre+ and cre- mice indicating that the cre+ male mice do not have deficits in the visual cue simply due to attentional deficits to the cue or due to visual acuity.

As mice were initially lever trained in a pseudorandom fashion to press either the left or the right lever when it was presented (i.e., only attend to levers), it is possible that the addition of a visual cue rule was acting as an attentional shift (i.e., ignore previous lever presentation order and attend to visual cue) which may have resulted in the unexpected deficits during the visual cue test in male cre+ mice. To address this, additional cohorts underwent lever training during which cre+ male mice took longer to obtain criterion compared to controls (t(8)  = −3.06, p = 0.02; Fig. 2l). Notably, this significance appears largely driven by the lack of variability in days to reach criterion in cre- mice. After lever training, mice received a response test (lever opposite of bias is correct) and a subsequent visual cue test. During the response test, cre+ mice took significantly more trials (t(8) = −2.55, p = 0.03; Fig. 2m) and errors (t(8) = −3.78, p = 0.01; Supplementary 3b) and had similar omissions (t(8) = −2.05, p = 0.08; Supplementary Table 3). During the visual cue set-shift, cre+ mice also required a greater number of trials (t(8) = −2.67, p = 0.03; Fig. 2n) but did not significantly differ in errors (t(8) = −2.06, p = 0.07; Supplementary 3c) or omissions (t(8) = −1.06, p = 0.32; Supplementary Table 3) to reach criterion compared to cre- mice. These findings combined with a lack of difference to acquire a visual cue training indicate that loss of GIRK channel activity alters PrL function (i.e., cognitive flexibility) in a manner distinct from lesions.

Influence of PrL pyramidal neuron GIRK knockout on female performance in EPM, FST, and progressive ratio

Intrinsic sex differences in mPFC physiology and function may have translational significance regarding resilience or susceptibility to pathological disorders [72,73,74]. Past studies have identified sex-dependent differences in GIRK-dependent signaling in adolescence [79,80,81].

We find that PrL GIRK1 ablation produced a consistent reduction in working memory in male mice as assessed using a forced alternation model. Assessment of more complex processes using the ASST showed an impaired performance in cre+ during the visual cue test in males, while not altering performance during the ED shift. These findings were unexpected as PrL lesions have previously shown to impair performance in the ED shift, but insufficient to disrupt performance in a cue-based discriminative test [14, 34]. It is important to note, however, that reduction in GIRK1-signaling and disorganized mPFC activity, both of which we show is evident in GIRK1 knockout mice, is inherently different than PrL lesioning (and therefore lack of activity). Given that loss of GIRK channel activity did not alter the ability to acquire a visual cue-based learning task, it is unlikely that this deficit reflects general attention to a cue or reduced visual acuity [82]. Rather, loss of GIRK1 alters PrL function in a manner where the addition of a visual cue contingency not present during lever training is sufficient to act as an attentional shift. In support, experiments involving response-to-cue shifts showed that cre+ male mice also performed worse when the contingency was changed from lever training to response (attend to one lever only), and that these impairments persisted during a subsequent response-to-visual cue ED Shift. Cognitive processes associated with working memory are reliant, although not exclusively, on coordinated PrL activity [83,84,85,86] therefore it is likely that increased/disorganized firing following loss of PrL pyramidal neuron GIRK underlies the observed deficits in working memory and cognitive flexibility.

PrL PYR GIRK1 knockout and CUS exposure in females

Sex differences in susceptibility to anxiety, mood, and other neuropsychiatric disorders have been established in humans [87,88,89,90]. Therefore, the influence that mPFC physiology and function has on behavior in both males and females has significant translational value for identifying both susceptibility and treatment options. We replicated findings showing that GABAB-GIRK signaling is present in female PrL pyramidal neurons [106, 107] and alter mPFC plasticity in female rats [108]. Importantly, studies showing set-shift task deficits in females following stress exposure used a task that is not operant-based, but rather employs odor and digging medium as cues [17, 109], which may have different capacities to identify stress-induced cognitive changes.

Impact of CUS on GABAB-GIRK signaling and cognitive flexibility in males

Similar to cognitive inflexibility in humans [9, 110, 111], exposure to CUS impairs performance in the ED Shift test in male mice. Deficits in flexibility aligned with a reduction in GABAB-GIRK signaling, highlighting a role in CUS-induced PrL dysfunction. Unlike GIRK1 knockouts, deficits in flexibility following CUS were specific to the ED Shift, as has previously been shown following chronic stress in male rats [101, 112]. Similar to PrL lesion studies [18, 34], CUS exposure may impair ED Shift performance rather than producing deficits in the visual cue test as it likely has more global effects than just reducing PrL pyramidal neuron GIRK1-dependent signaling [113]. As converging lines of evidence show that exposure to unpredictable stressors negatively impacts cognitive control in humans and rodents [9, 110, 111, 114, 115], the ability of systemic ML297, a GIRK1-selective agonist, to rescue this deficit has significant impact on future therapeutics aimed at treating cognition-related deficits produced by stress. Collectively, we highlight the importance of GIRK-dependent signaling in male, but not female, PrL pyramidal neurons in the regulation of both affect and cognition and demonstrate that PrL GIRK channels in males are targets of stress.

There are several limitations to the study. First, it is important to note that the collection of female data for both GIRK and stress-related studies were performed at a later date, however, all groups of animals were run with proper controls in parallel. As direct comparison of sex may confer false positive or negative data, findings were presented separately. Thus, while our data provide a proper assessment of the impact of GIRK knockout and stress in females and males alone, without a direct statistical comparison conference of sex differences in the functional role GIRK channels play should be considered with caution. Second, although the locus of ML297 effects were not identified, systemic ML297 can rescue stress-induced cognitive deficits in males and therefore demonstrates therapeutic potential. It is possible that the rescue of deficits is due to augmentation of residual GIRK1 signaling in PrL pyramidal neurons, and that this is sufficient to restore optimal neuronal activity within this region. It is also possible that the effectiveness of ML297 reflects activation of GIRK1 and thus inhibition of other cortical structures that drive behavioral rigidity. While we demonstrate that loss of GIRK1 increases output of PrL pyramidal neurons, it is unknown how viral- and stress-related elevations in activity alter an often reciprocally and/or collaterally connected mPFC microcircuitry. It is also not clear whether behavioral deficits in males are disproportionally driven by reduced GIRK signaling and increased output of pyramidal neuron subpopulations based on the downstream target. These questions have important implications as both increased (disorganized) activity as well as mPFC hypoactivity have been highlighted in neuropsychiatric disease states.

Funding and disclosure

These studies were supported by funding from the Brain and Behavior Research Foundation (#26299; MH), Marquette University Regular Research Grant (MH), the Charles E Kubly Mental Health Research Foundation at Marquette University (MH), and NIH grants DA034696 and AA027544 (KW). The authors have no biomedical financial interests or potential conflicts of interest.