Ciliary neurotrophic factor signaling in the rat orbitofrontal cortex ameliorates stress-induced deficits in reversal learning

Abstract

Deficits in cognitive flexibility, i.e. the ability to modify behavior in response to changes in the environment, are present in several psychiatric disorders and are often refractory to treatment. However, improving treatment response has been hindered by a lack of understanding of the neurobiology of cognitive flexibility. Using a rat model of chronic stress (chronic intermittent cold stress, CIC) that produces selective deficits in reversal learning, a form of cognitive flexibility dependent on orbitofrontal cortex (OFC) function, we have previously shown that JAK2 signaling is required for optimal reversal learning. In this study we explore the molecular basis of those effects. We show that, within the OFC, CIC stress reduces the levels of phosphorylated JAK2 and of ciliary neurotrophic factor (CNTF), a promoter of neuronal survival and an activator of JAK2 signaling, and that neutralizing endogenous CNTF with an intra-OFC microinjection of a specific antibody is sufficient to produce reversal-learning deficits similar to stress. Intra-OFC delivery of recombinant CNTF to CIC-stressed rats, at a dose that induces JAK2 and Akt but not STAT3 or ERK, ameliorates reversal-learning deficits, and Akt blockade prevents the positive effects of CNTF. Further analysis revealed that CNTF may exert its beneficial effects by inhibiting GSK3β, a substrate of Akt and a regulator of protein degradation. We also revealed a novel mechanism of CNTF action through modulation of p38/Mnk1/eIF4E signaling. This cascade controls translation of select mRNAs, including those encoding several plasticity-related proteins. Thus, we suggest that CNTF-driven JAK2 signaling corrects stress-induced reversal learning deficits by modulating the steady-state levels of plasticity-related proteins in the OFC.

INTRODUCTION

Reversal learning is a form of cognitive flexibility dependent on the function of the orbitofrontal cortex (OFC). Tests to measure this executive process in both humans and animal models assess the ability to reverse a previously learned associative contingency. For example, after acquiring a rule associating one stimulus with a reward and another with no reward, the valence of the cues is switched, such that the cue previously associated with reward is now negative, and the previously negative cue becomes positive, signaling the reward (1). Several psychiatric conditions, including mood disorders, anxiety disorders and psychotic disorders present with deficits in reversal learning that can contribute to the persistence of maladaptive behaviors (2–6). Stress exacerbates these effects, reducing flexible learning and increasing habitual responding in both humans and in rodents (7–10). Despite the central role of reversal learning in adaptive behavior, the neurobiology of this executive function has not been studied extensively, and this may contribute to the infectiveness of available interventions to improve reversal learning deficits (4). In this study we address the neurobiological mechanisms underlying reversal learning in rats.

Janus kinase (JAK) and its best-characterized downstream effector, Signal Transducer of Activation and Transcription (STAT), comprise a family of signaling molecules crucial to cell proliferation, differentiation and survival in both the periphery and the central nervous system. JAK2 is involved in several aspects of neuronal function (11–16) and in the hippocampus it is required for the induction of N-methyl-D-aspartate receptor (NMDAR)-dependent long-term depression (LTD) (17), a form of activity-dependent synaptic plasticity associated with spatial reversal learning (18–21).

Typical activators of JAK2 signaling are cytokines and growth factors. Ciliary neurotrophic factor (CNTF) is an activator of JAK2 signaling and a neuronal survival factor highly expressed within the central nervous system (22–24). CNTF binds to a tripartite receptor comprised of CNTF receptor alpha (CNTFRα), glycoprotein-130 (gp130) and the leukemia inhibitor factor receptor (LIFR), which activates JAK2 by auto-phosphorylation. JAK2 then phosphorylates and activates the STAT family of transcription factors. Besides STATs, JAK2 can also activate other effector pathways, including Ras/ERK1/2 and PI3 kinase/Akt (25, 26). Neurons express CNTFRα, along with LIFR and gp130, and are therefore capable of responding to CNTF (27–29).

We recently found that blocking JAK2 signaling in the OFC produced reversal-learning deficits comparable to those induced by chronic intermittent cold (CIC) stress, suggesting a role of this signaling cascade in reversal learning (30). Indeed, CIC stress reduced the levels of JAK2 phosphorylation (pJAK2), but not pSTAT3 in the OFC (31). Activating the cascade by microinjecting low levels of interleukin 6 (IL-6) corrected CIC stress-induced reversal learning deficits (30). Interestingly, ketamine also restored stress-induced reversal deficits and modulated OFC responses to afferent stimulation through a mechanism involving JAK2 (31). Silencing JAK2 but not STAT3 in cortical neurons reduced the expression of activity-regulated cytoskeleton-associated protein (Arc), a modulator of synaptic plasticity and LTD, suggesting the possibility that JAK2 signaling may regulate OFC function by modulating the expression of plasticity-related proteins. In this study we examined the possibility that CIC stress negatively affects reversal learning by reducing JAK2 signaling, and that activation of this pathway by CNTF rescues reversal learning deficits in stressed rats through STAT-independent JAK2-driven effector pathways. We specifically focused on cascades that regulate protein homeostasis, such as PI3K/Akt and p38/Mnk/eIF4E. We hypothesize that CNTF/JAK2 signaling controls protein expression essential for optimal reversal learning. Portions of this work have been presented in abstract form (32).

MATERIALS AND METHODS

Animals.

Adult male and female Sprague Dawley rats (Envigo, 225–249g) were maintained on a 12/12 h light/dark cycle, with food and water ad libitum. Estrus cycle was not found to correlate with any of the measures reported in female rats in this study. All procedures were consistent with NIH guidelines and approved by the Institutional Animal Care and Use Committee of the University of Texas Health Science Center at San Antonio.

Chronic Intermittent Cold (CIC) Stress.

CIC stress was conducted as previously described (8). Animals were transported in their home cages into a cold room (4 °C) for 6 h, then returned to housing. This was repeated for 14 consecutive days. Controls remained in the housing room.

Reversal Learning Test.

The reversal learning test was performed as previously described (30). One week before testing (i.e., day 11 of CIC), male rats were food restricted to 14g and female rats to 7g per day. One day after the end of CIC, animals were trained to dig in terracotta pots to retrieve ¼ Cheerio reward (General Mills Cereals, Minneapolis, MN). On day 2, they were trained to locate the reward within the baited pot via cues in two sensory dimensions: an odor applied to the rim of the pot, and the digging medium that filled the pot. The following day, animals were tested on a series of discrimination tasks, in which a criterion of six consecutive correct responses was required to proceed to the next task. In the first task (simple discrimination, SD), animals learned to associate one stimulus dimension (e.g., odor) with the reward-baited pot. On the second task (compound discrimination, CD), animals continued to discriminate based on that dimension, and the second irrelevant dimension (e.g., digging medium) was introduced as a distractor. During the reversal learning (R1) task, the cue/reward association was switched: the previously non-rewarded cue was now positive, and vice versa. The dependent measure was the number of trials required to meet the criterion of six consecutive correct responses (trials to criterion, TTC). All animals tested reached criterion within 50 trials. Microinjections were performed either 20 min or 24h before reversal learning. In the latter case, following training, rats remained undisturbed in their home cages for 4h before injection procedures.

Stereotaxic surgery and OFC microinjections

Rats were placed in a stereotaxic apparatus under isofluorane anesthesia (Parkland Scientific) and implanted with bilateral guide cannulae (22-gauge stainless-steel, 11 mm length, Plastic One) with their tips positioned 1 mm above the lateral OFC (from Bregma: AP + 2.9, ML ± 2.6, DV – 4.2 mm). Rats were allowed to recover for at least 1 week before further manipulations.

For OFC microinjections, obdurators were removed and microinjectors (30-gauge stainless steel) were inserted, extending 1 mm beyond the cannula tips. Microinjectors were connected by a fluid-filled line to a Hamilton syringe mounted on a syringe pump (Instech Laboratories) and injections were made in volumes of 0.5 µl at a rate of 0.25 µl/min. After delivery, injectors were left in place for 3 min to allow diffusion. At the end of the experiment, cannulae placement were verified histologically. Images depicting the sites of injections are presented in Suppl. Figure 1.

For the experiments of CNTF blockade, an anti-rat CNTF antibody (R&D systems, MAB557, ND₅₀ = 1–3 µg/ml) or isotype-matched mouse IgG_2B were injected bilaterally (50 ng / 0.5 µl saline vehicle) after completion of the CD task. Twenty minutes after injection, testing resumed with the reversal learning task. To verify that CNTF blockade in the OFC did not interfere with learning non-specifically, a separate group of rats was injected in a similar fashion prior to testing on the CD task. The effectiveness of the neutralizing antibody was confirmed by testing its capacity to prevent CNTF induced pJAK2 phosphorylation in the OFC (Suppl. Figure 2A).

For the experiments testing if CNTF rescued reversal learning deficits caused by CIC stress, rat recombinant CNTF protein (R&D Systems, 557-NT) was microinjected bilaterally at 0.5 ng/0.5 µl (50nM) 20 min or 24 hrs before reversal testing. This dose was selected from an in vivo experiment of dose-dependent CNTF effects on JAK2 signaling in the OFC (see Figure 2).

Figure 2. Exogenous CNTF restores reversal learning compromised by stress.

(A–D) Injection of recombinant CNTF activates JAK2, Akt and STAT3 but not ERK in the OFC. Recombinant CNTF was injected at different concentrations (50, 100 and 500 nM) into the OFC 30 min before dissection of OFC for western blotting. (A) pJAK2 levels were significantly elevated by 50 nM CNTF (Dunnett’s post-hoc test: *p< 0.05 vs VEH, n= 4 males and 3 females). (B) pSTAT3 levels were elevated by 500 nM CNTF (*p< 0.01 vs VEH). (C) pT-Akt levels were elevated by 50nM CNTF (*p< 0.05 vs VEH). (D) pERK levels were not affected by any concentration of CNTF. (E-J) CNTF corrects reversal-learning deficits induced by CIC stress. CNTF (50nM) injected into the OFC 20 min before the reversal learning task corrected the CIC-induced deficits (E, Mixed-sex groups, Newman-Keuls post-hoc test: *p< 0.001 vs. NS/VEH, and #p< 0.001 vs. CIC/VEH, n= 11–12; F, male rats, *p< 0.01 vs. NS/VEH, and #p< 0.01 vs. CIC/VEH, n=6–7; G, female rats, *p< 0.01 vs. NS/VEH, and #p< 0.01 vs. CIC/VEH, n= 5 per group). CNTF also improved reversal deficits in stressed rats when administered 24h before testing (H, mixed groups Unpaired t-test: *p< 0.01, n=10–11; I, male rats, *p< 0.01, n= 6–7; J, female rats, *p< 0.05, n= 4 per group).

To test the involvement of Akt in the reversal learning rescue by CNTF, the reversible PI3 kinase inhibitor LY294002 (Tocris, Cat # 1130) was diluted to a concentration of 25 µM in 0.5% v/v DMSO/saline and microinjected bilaterally at 4.3 ng/0.5 µl, 5 min before microinjecting CNTF (50nM). Control rats were sequentially injected with the respective vehicles (0.5% DMSO and saline). Rats were tested on reversal learning 24h after microinjections. The concentration of LY294002 was selected based on published evidence of in-vivo efficacy (33); inhibition of Akt phosphorylation at the threonine 308 site in the OFC was confirmed by western blotting (Suppl Figure 2B).

Tissue Collection.

A 2 mm coronal section was cut 2–4 mm caudal to the frontal pole of the brain, and OFC was dissected on ice, from the lateral margin of the brain to the medial boundary of the forceps minor, ventral to the forceps minor and dorsal to the rhinal sulcus.

Western Blots.

Western blots were performed as described in (30, 31). Membranes were incubated overnight at 4°C in the following pr imary antibodies: CNTF (mouse mAb, 1:500, Millipore MAB338), pY_1007–1008-JAK2 (rabbit pAb, 1:1,000, Milipore 07–606), pY₇₀₅-STAT3 (rabbit mAb 1:1,000, CST 9145), pT₃₀₈-Akt (rabbit mAb, 1:1,000 CST 4056), pS₉-GSK3β (rabbit mAb, 1:1,000, CST 9323), pT₃₈₉-S6K (rabbit pAb, 1:1,000 CST 9205), pS₂₀₉-eIF4E (rabbit pAb, 1:1,000 CST 9741), pT₂₀₂/Y₂₀₄-ERK (mouse mAb, 1:5,000 CST 9106), pT₁₈₀/Y₁₈₂-p38 MAPK (rabbit mAB, 1:1,000 CST 4511) and pT_197/202 Mnk1 (rabbit pAb, 1:1,000, CST 2111). Subsequently, membranes were stripped with Restore Plus (Fisher Scientific) and re-probed with antibodies against un-phosphorylated STAT3 (mouse mAb 1:2,000, Santa Cruz Biotechnology, sc-8019), JAK2 (rabbit mAb, 1:2,000, CST 3230), Akt (mouse mAb, 1:1,000, CST 2920), GSK3β (mouse mAb, 1:1,000, CST 9832), S6K (rabbit pAb, 1:1,000, CST 9202), eIF4E (rabbit pAb, 1:1,000, CST 9742), ERK (rabbit pAb, 1:5,000, Santa Cruz Biotechnology, sc-94), p38 MAPK (rabbit mAb, 1:2,000, CST 8690), and Mnk1 (mouse mAb, 1:1,000, Novus Biologicals NBP2–11526), or GAPDH (CST; 1:10,000). Western blot images were captured using the G:BOX-XT4 Chemi system (Syngene; Frederick, MD). Data were normalized to the respective un-phosphorylated protein or GAPDH and represented as percent baseline.

Statistical Analysis.

Data were analyzed by t-test or ANOVA, using Prism 8.0 (GraphPad Software) and are reported in Tables 1–5. Post-hoc comparisons were performed with Newman-Keuls test, Sidak’s test or Dunnett’s test. For the behavioral experiments, we first analyzed trials to criterion on the reversal task using ANOVA, including Sex as a factor. Main effects and/or interactions of the experimental variables (e.g., Stress, Drug, Inhibitor) were reported for mixed groups including both sexes. If there was a main effect of sex, we ran the same analyses for each sex separately. Additionally, to ensure that any differences observed in reversal learning were not attributable to non-specific effects on learning in general, we analyzed the tasks preceding reversal learning (i.e., simple discrimination and compound discrimination, SD and CD) using a repeated measure ANOVA on Task, including Sex as a factor. The analysis of SD and CD for experiment 1 (Suppl Figure 3A) and experiment 2 (Suppl Figure 3B) did not include Antibody or Drug as factors because the compounds were administered after the SD and CD tasks were completed.

Table 1.

Figure	Analysis	Factors	F (DFn, DFd)	p
1A	All subjects	Time (one way anova)	5.727 (2, 34)	0.0072
1B	Males	Time (one way anova)	5.714 (2, 17)	0.0126
1C	Females	Time (one way anova)	4.442 (2, 14)	0.0321
1D	All subjects	Time (one way anova)	7.174 (2, 36)	0.0024
1E	Males	Time (one way anova)	10.95 (2, 18)	0.0008
1F	Females	Time (one way anova)	6.076 (2, 15)	0.0117
1G	All subjects	Antibody	38.30 (1,18)	<0.0001
		Sex	48.82 (1,18)	<0.0001
		Antibody × Sex	2.242 (1,18)	0.1517
1H	Males		t= 5.316, df= 10	0.0003
1I	Females		t= 3.562, df= 8	0.0074
1J	All subjects (n= 3–4 males + 2 females)		t= 0.1551, df= 9	0.8801

Statistical analysis tables

Significant p values are marked in bold

Table 5.

Figure	Analysis	Factors	F (DFn, DFd)	p
Suppl 3 A	All subjects	Sex	8.442 (1, 20)	0.0087
Suppl 3 B	All subjects	Sex	16.57 (1, 42)	0.0002
		Stress	0.08631 (1, 42)	0.7704
Suppl 3 C	All subjects	Sex	13.14 (1, 17)	0.0021
		Drug	2.947 (1, 17)	0.1042
Suppl 3 D	All subjects	Sex	19.70 (1, 36)	<0.0001
		Inhibitor	0.03328 (1, 36)	0.8563

RESULTS

CNTF is necessary for optimal reversal learning.

We previously reported that CIC stress decreased JAK2 phosphorylation (31). To determine if this was potentially due to an effect of stress on CNTF, an endogenous JAK2 activator, we measured the levels of CNTF in the OFC at 24 h or 72 h following the end of CIC stress. The latter time coincides with the day we perform the reversal learning test.

In a mixed sex analysis, CNTF levels were significantly lower 24 h (Figure 1A and Table 1) and pJAK2 levels were lower 24 h and 72 h (Figure 1D and Table 1) after the end of stress compared to non-stressed animals. Although analysis of the data using sex as a factor did not return a significant effect of sex, when analyzed separately a potential difference in the duration of effects emerged between males and females. Thus, CNTF levels were reduced in the OFC of both males and females 24 h after the end of stress compared to non-stressed rats; by 72 h after stress CNTF levels had returned to normal in female but not in male rats (Figure 1B and 1C). Similarly, JAK2 phosphorylation was decreased in both sexes at 24 h, and but only in males at 72 h (Figure 1E and 1F and Table 1).

Fig. 1.

CNTF is necessary for optimal reversal learning. (A–F) CIC stress reduces levels of CNTF and pJAK2 in the OFC. Rats were exposed to CIC stress and OFC was collected 24h or 72h after the end of stress. CNTF levels were decreased 24h after the end of CIC stress, in mixed sex groups (A, n=11–14) and in both sexes analyzed separately (B, n= 4–8 C, n=4–7). CNTF levels were still significantly decreased 72 h after stress in male rats (B), but not in female rats (C). pJAK2 levels were lower at 24 h after stress (D, E, F). At 72 h levels of pJAK2 were lower in mixed sex groups (D, n= 11–16) and in male rats (E, n= 4–9) but not in female rats (F, n= 4–7). (Dunnett’s post-hoc test: *p< 0.05 vs. No Stress; #p<0.05 vs. 72h). (G-J) Blockade of endogenous CNTF in the OFC causes reversal learning deficits. A neutralizing antibody against CNTF injected into the OFC 20 min before reversal learning caused deficits in reversal learning (G, mixed-sex groups, unpaired t-test, *p< 0.01, n=10–12; H, male rats *p< 0.001, n=6 per group; I, female rats *p< 0.01, n=4–6). (J) Antibody injected before the compound discrimination (CD) task did not impair performance on that task (Mixed groups: 3–4 males + 2 females per group). TTC= trials to criterion.

We then asked if endogenous CNTF signaling was necessary for reversal learning by injecting a neutralizing antibody against CNTF into the OFC immediately after completion of the CD task, 20 min before reversal learning. In groups containing rats of both sexes, blocking CNTF in OFC significantly impaired reversal learning compared to control rats injected with an isotype-matched IgG (Figure 1G, and Table 1). As there was a significant effect of sex, we then conducted separate analyses for the two sexes. Females required fewer trials to reach criterion overall than males. Nonetheless, anti-CNTF impaired performance on the reversal task comparably in both male and female rats (Figure 1H and 1I, and Table 1). Analysis of the tasks preceding reversal learning also indicated a sex difference, with females again requiring fewer TTC than males (Supplemental Figure 3A and Table 5). Finally, injection of the anti-CNTF antibody before CD did not impair performance in this task, indicating that CNTF blockade in the OFC specifically impacts reversal learning (Figure 1J and Table 1). These data indicate that CNTF is necessary for optimal reversal learning, and that this effect is comparable in males and females. Therefore, it is possible that CNTF may rescue the behavioral deficits produced by CIC stress. This was tested in the next experiment.

Exogenous CNTF restores reversal learning compromised by stress.

To verify that exogenous CNTF administration activates JAK2 signaling in the OFC, and to select an appropriate concentration to test its ability to rescue reversal learning deficits, we injected recombinant CNTF at different concentrations (50, 100 and 500 nM) into the OFC of both male (n=4) and female rats (n=3). There was no effect of sex but there was a significant effect of CNTF on pJAK2, pSTAT23 and pT-Akt (Table 2). pJAK2 levels were significantly increased by 50 nM CNTF compared to vehicle controls (Figure 2A and Table 2), whereas pSTAT3 was significantly increased only by 500 nM but not by 50 nM CNTF (Figure 2B). Similar to pJAK2, pT-Akt was increased by 50 nM CNTF (Figure 2C), whereas pERK did not respond to any concentration of CNTF (Figure 2D). In previous work we observed that silencing JAK2 but not STAT3 in cortical neurons decreased the levels of the activity-dependent cytoskeleton associated protein, Arc (31) suggesting that STAT3 may not be the primary mediator of JAK2 effects on plasticity; therefore, in the subsequent behavioral experiments, we administered the concentration that induced JAK2 but not STAT3 (50 nM CNTF).

Table 2.

Figure	Analysis	Factors	F (DFn, DFd)	p
2A	All subjects	Dose (one way anova)	3.701 (3, 24)	0.0255
2B	All subjects	Dose (one way anova)	4.477 (3, 24)	0.0124
2C	All subjects	Dose (one way anova)	3.471 (3, 24)	0.0318
2D	All subjects	Dose (one way anova)	1.151 (3, 24)	0.3487
2E	All subjects	Stress	14.35 (1, 38)	0.0005
		Drug	8.515 (1, 38)	0.0059
		Sex	29.21 (1, 38)	<0.0001
		Sex × Stress × Drug	0.8448 (1, 38)	0.3638
2F	Males	Stress	10.71 (1, 22)	0.0035
		Drug	3.638 (1, 22)	0.0696
		Stress × Drug	7.479 (1, 22)	0.0121
2G	Females	Stress	6.560 (1, 16)	0.0209
		Drug	9.163 (1, 16)	0.008
		Stress × Drug	7.380 (1, 16)	0.0152
2H	All subjects	Drug	20.29 (1, 17)	0.0003
		Sex	22.46 (1, 17)	0.0002
		Sex × Drug	1.611(1, 17)	0.2214
2I	Males		t=4.277, df=11	0.0013
2J	Females		t=2.556, df=6	0.0432

If the reduced levels of CNTF and pJAK2 (Figure 1) are responsible for the impairment in reversal learning in stressed rats, administering CNTF should correct the deficit. Following CIC stress, rats showed significant deficits in reversal learning and injection of CNTF (50 nM) into the OFC 20 min before reversal testing rescued the CIC stress-induced impairment (Figure 2E and Table 2). As in the previous experiment, there was a significant effect of Sex. Again, female rats had overall lower TTC values than males, but CIC stress significantly increased the trials to criterion comparably in both sexes, and CNTF corrected this effect in both sexes (Figure 2F and 2G and Table 2). Stress did not affect the learning tasks preceding reversal, indicating that the effects on reversal were not due to general learning deficits. There was a significant effect of sex, with females again requiring fewer TTC overall, primarily in the CD Task (Suppl Fig 3B and Table 5).

CNTF was also effective in correcting CIC-induced deficits when administered 24h before reversal learning in both sexes (Figure 2H–J, Table 2). There were no effects of CNTF in the tasks preceding reversal, but, as seen in the preceding experiments, there was a significant effect of sex (Suppl Figure 3C, Table 5). Together these data suggest that CNTF exerts both short-term and long-lasting effects on OFC function in both sexes.

Microinjection of LY294002 into the OFC prevents the beneficial effects of CNTF.

The concentration of CNTF that rescues the reversal deficits in stressed rats (50nM) did not significantly induce either STAT3 or ERK in the OFC, but did increase levels of activated Akt (pT-Akt; Figure 2). Therefore, we investigated the possibility that PI3K/Akt signaling mediated the downstream effects of CNTF on reversal. We first determined if levels of pT-Akt were affected by stress. Stress decreased pT-Akt at both 24 h and 72 h in mixed sex groups (Figure 3A). Both male and female rats showed reduced pT-Akt 24 h after stress. By 72 h, pT-Akt levels had returned to baseline in females, but remained significantly reduced in males (Figure 3B and 3C and Table 3). We then examined whether the PI3K inhibitor LY294002 blocks the positive effects of CNTF on reversal learning in stressed rats. Using a dose of LY294002 that prevents the induction of pT-Akt by CNTF in the OFC (Suppl Figure 2B), we microinjected the inhibitor five minutes before CNTF administration and we observed that it blocked the beneficial effects of CNTF on reversal learning deficits tested 24h later (Figure 3D and Table 3). When the data were analyzed separately for the two sexes, stressed male rats receiving CNTF showed a significant improvement in performance that was prevented by the prior administration of LY294002 (Figure 3E and Table 3). In female rats, the inhibitor alone worsened performance at baseline, in the absence of CNTF, and only partially prevented the beneficial effects of CNTF, suggesting a differential effect of PI3K inhibition at baseline in male and female rats after CIC stress. (Figure 3F and Table 3). The inhibitor did not affect performance on the tasks preceding reversal learning, but as in the previous experiments, there was a significant effect of sex, with females performing better overall (Suppl. Figure 3D and Table 5).

Figure 3. Akt signaling mediates the effects of CNTF on reversal learning.

(A–C) CIC stress reduces the levels of pT-Akt in the OFC. pT-Akt is lower 24h and 72h after the end of CIC stress in a mixed-sex analysis (A, n=11–15) and in male rats (B, n=4–8) (Dunnett’s post-hoc test: *p< 0.05, and **p<0.01 vs. No Stress). In female rats pAkt is significantly decreased 24h after stress (C, *p< 0.05 vs. No Stress, n= 4–7). (D-F) The PI3K inhibitor LY294002 blocks the effects of CNTF in the OFC. LY294002 blocks the beneficial effects of CNTF on reversal learning in stressed rats (D, mixed groups, Newman-Keuls post-hoc test: *p< 0.05 vs. VEH/VEH, #p<0.05 vs. VEH/CNTF n= 9–10; E, male rats, *p< 0.05 vs. VEH/VEH, #p<0.05 vs. VEH/CNTF, n=5; F, female rats, *p< 0.05 vs. VEH/VEH, **p< 0.01 vs. VEH/VEH, #p<0.05 vs. VEH/CNTF, $p<0.01 vs. LY/VEH, n=4–5).

Table 3.

Figure	Analysis	Factors	F (DFn, DFd)	p
3A	All subjects	Time (one way anova)	8.824 (2, 35)	0.0008
3B	Males	Time (one way anova)	7.931 (2, 17)	0.0037
3C	Females	Time (one way anova)	6.819 (2, 15)	0.0078
3D	All subjects	Treatment	10.37 (1, 31)	0.003
		Inhibitor	15.65 (1, 31)	0.0004
		Sex	39.13 (1, 31)	<0.0001
		Sex × Treat × Inhib	8.666 (1, 31)	0.0061
3E	Males	Treatment	1.513 (1, 16)	0.2365
		Inhibitor	2.813 (1, 16)	0.113
		Treatment × Inhibitor	11.25 (1, 16)	0.004
3F	Females	Treatment	16.39 (1, 15)	0.001
		Inhibitor	22.31 (1, 15)	0.0003
		Treatment × Inhibitor	0.1139 (1, 15)	0.7405

CNTF elicits a p38-dependent protein translation pathway

Akt regulates protein homeostasis by increasing mTOR/S6K dependent protein synthesis and reducing activation of GSK3β, a kinase that targets proteins for proteasomal degradation. Thus, we sought to examine the effects of CNTF on S6K and GSK3β in non-stressed and stressed rats. Additionally, to investigate the selectivity of CNTF for Akt-dependent signaling, we examined the phosphorylation of eIF4E, a regulator of translation initiation activated by the MAPK/Mnk1 pathway. Microinjection of CNTF in non-stressed rats increased pJAK2, pT-Akt, pS6K, pS9-GSK3β and peIF4E in both male (Figure 4A–F) and female rats (Figure 4G–L). In stressed rats of each sex, CNTF increased levels of pJAK2, pT-Akt, pS9-GSK3β and peIF4E but not pS6K (Figure 4A–L and Table 4). Although the effects of stress in vehicle-treated rats were not significant in this experiment, there were trends toward a decrease in some of the measures, such as pJAK2, pT-Akt, peIF4E and pS-GSK3β.

Figure 4. CNTF elicits a p38-dependent protein translation pathway.

Rats were either left undisturbed in their home cages (No Stress) or were subjected to CIC stress (CIC), and injected with saline (VEH) or 50nM CNTF (CNTF) 30 min before collection of OFC tissue. Male rats: CNTF increased pJAK2, pT-Akt, pS9-GSK3β and peIF4E in both non-stressed and stressed animals; CNTF increased pS6K only in non-stressed rats (A, B, C, D, E, *p< 0.05 and **p<0.01 vs. respective vehicles, n=4–6). Female rats: CNTF increased pJAK2, pT-Akt, pS9-GSK3β and peIF4E in both non-stressed and stressed animals; CNTF increased pS6K only in non-stressed rats (G, H, I, J, K, *p< 0.05, **p<0.01 and ***p<0.001 vs. respective vehicles, n=4–6). Panels F and L show images of representative western blots. CNTF induces p38 (M) and Mnk1 (P) and this effect is not sensitive to the PI3K inhibitor LY294002 (*p< 0.05, and ***p<0.001 vs. VEH, n= 4–5). CNTF increases p-p38 and pMnk1 in both non-stressed and stressed male (N, Q) and female rats (O, R, *p< 0.05, **p<0.01 vs. respective vehicles, n=4–6). S. Proposed model of CNTF actions. CNTF (50nM) binding to gp130/LIFR/CNTFRα activates JAK2, which in turns activates PI3K/Akt. Akt phosphorylates and inhibits GSK3β. Akt also inhibits TSC, an inhibitor of mTOR, thereby activating this pathway. mTOR phosphorylates S6K, a regulator of global protein synthesis. This branch is induced by CNTF in non-stressed rats, but it does not seem to be activated in stressed rats. In parallel to regulation of Akt, CNTF also induces the p38/ Mnk1/ eIF4E cascade that controls specialized mRNA translation. Finally, at 50nM, CNTF does not seem to activate ERK or STAT3 in the OFC (dotted arrows).

Table 4.

Figure	Analysis	Factors	F (DFn, DFd)	p
4A	Males	Stress	0.44 (1,14)	0.52
		Drug	33.28 (1,14)	<0.0001
		Stress × Drug	0.51 (1,14)	0.49
4B	Males	Stress	2.77 (1,14)	0.11
		Drug	28.76 (1,14)	0.0001
		Stress × Drug	0.09 (1,14)	0.76
4C	Males	Stress	0.41 (1,14)	0.53
		Drug	5.88 (1,14)	0.03
		Stress × Drug	4.89 (1,14)	0.04
4D	Males	Stress	0.09 (1,14)	0.76
		Drug	20.38 (1,14)	0.0005
		Stress × Drug	0.008 (1,14)	0.93
4E	Males	Stress	0.66 (1,14)	0.43
		Drug	22.98 (1,14)	0.0003
		Stress × Drug	0.41 (1,14)	0.53
4G	Females	Stress	15.59 (1,14)	0.0015
		Drug	48.97 (1,14)	<0.0001
		Stress × Drug	3.55 (1,14)	0.08
4H	Females	Stress	0.60 (1,14)	0.45
		Drug	20.36 (1,14)	0.0005
		Stress × Drug	0.60 (1,14)	0.45
4I	Females	Stress	6.47 (1,14)	0.02
		Drug	3.06 (1,14)	0.1
		Stress × Drug	5.33 (1,14)	0.04
4J	Females	Stress	6.93 (1,14)	0.02
		Drug	23.79 (1,14)	0.0002
		Stress × Drug	1.52 (1,14)	0.23
4K	Females	Stress	4.95 (1,14)	0.04
		Drug	67.18 (1,14)	<0.0001
		Stress × Drug	0.11 (1,14)	0.74
4M	Males	Treatment (one way anova)	13.16 (2,10)	0.0016
4N	Males	Stress	0.21 (1,14)	0.65
		Drug	16.61 (1,14)	0.001
		Stress × Drug	1.22 (1,14)	0.29
4O	Females	Stress	2.311 (1,14)	0.15
		Drug	21.66 (1,14)	0.0004
		Stress × Drug	0.79 (1,14)	0.39
4P	Males	Treatment (one way anova)	6.96 (2,9)	0.015
4Q	Males	Stress	1.16 (1,14)	0.3
		Drug	16.17 (1,14)	0.0013
		Stress × Drug	0.05 (1,14)	0.82
4R	Females	Stress	15.89 (1,14)	0.0014
		Drug	71.02 (1,14)	<0.0001
		Stress × Drug	2.80 (1,14)	0.12

Following the unexpected observation that CNTF increased peIF4E, we considered the possibility that an alternate non-canonical signaling pathway was elicited by CNTF. eIF4E is phosphorylated by MAP kinase-interacting serine/threonine protein kinase 1 (Mnk1), which in turn is activated by ERK and p38 MAPK. Therefore we explored the activation of these kinases. First, we observed that CNTF indeed activated p38 (Figure 4M) and Mnk1 (Figure 4P), and these effects were not blocked by LY294002. CNTF activated Mnk1 and p38 in non-stressed and stressed rats of both sexes, similarly to what was observed for peIF4E (Figure 4, N, O, Q, R; for statistical analysis see Table 4).

DISCUSSION

In this study we showed that CIC stress reduced levels of CNTF, pJAK2 and pAkt in the OFC. We also found that endogenous CNTF signaling is necessary for optimal reversal performance and that microinjection of a low dose of recombinant CNTF is sufficient to restore cognitive performance compromised by CIC stress. The beneficial effects of CNTF are mediated, in part, by Akt-dependent mechanisms that include regulation of GSK3β activity. In addition, we discovered that CNTF elicits p38/Mnk/eIF4E signaling and this may constitute an additional mechanism for its regulation of cognitive function in the OFC. We postulate that through these mechanisms, CNTF/ JAK2/ Akt signaling controls the levels of proteins underlying OFC plasticity that is necessary for optimal reversal learning. Together with previous work identifying JAK2 as a mediator of the beneficial effects of ketamine on reversal learning, we suggest that the CNTF /JAK2/ Akt system is a novel pathway that can be targeted therapeutically to improve reversal learning deficits in disorders associated with OFC dysfunction.

JAK2 signaling as a mediator of OFC plasticity leading to reversal learning

Our previous data indicated that JAK2 signaling was required for optimal reversal learning (30). Specifically, we observed that either pharmacological inhibition of JAK2 or neutralizing antibodies against interleukin 6 (IL-6, a ligand of gp130/JAK2 receptor complexes) produced deficits in reversal learning, and that delivery of low doses of IL-6 could restore reversal learning compromised by CIC stress. CIC stress, however, did not seem to produce long-lasting changes in levels of IL-6 mRNA or protein (34, 35). For this reason, and because of the potential for IL-6 to elicit inflammatory responses at higher concentrations, we turned our attention to CNTF, another endogenous ligand of gp130/JAK2 signaling that has an important role as a regeneration and survival factor in neurons (36–38). Here we show that CNTF and pJAK2 levels within the OFC are markedly decreased in both male and female rats 24h after the end of CIC stress, and they remain low 72h after stress in male rats, suggesting that CNTF-driven JAK2 signaling is directly impacted by stress and this may in part be responsible for the negative effects of stress on reversal learning. Blockade of endogenous CNTF generated reversal deficits similar to those of stress, and a low dose of CNTF improved reversal learning in stressed rats within 30 min of injection, as well as 24 h after injection, suggesting that CNTF elicits both short-term effects and long-lasting plastic changes in the OFC. These data align with the results of a previous study where ketamine given 24 h before testing rescued stress-induced reversal deficits in a JAK2-dependent manner (31). Thus, the beneficial effects of ketamine on cognitive flexibility may in part be due to plasticity engendered by activation of JAK2 signaling. Along these lines, we demonstrated that ketamine induced a depotentiation of evoked potentials in the OFC that was blocked by inhibiting JAK2. This is particularly interesting in that temporary depotentiation of synapses in the hippocampus accompanies spatial reversal learning (18–21), and JAK2 plays a role in certain forms of hippocampal LTD (17, 39). Thus, we postulate that activation of JAK2 produces synaptic depotentiation, rendering previous memories labile, thereby reducing perseverative responding and enabling flexible choice during reversal learning. Other cytokine receptors that function through the JAK2 signaling pathway, such as the granulocyte-colony stimulating factor receptor and the leptin receptor, have been recently linked to increased synaptic plasticity and improvements in cognitive function (40–42).

Signaling events elicited by CNTF/JAK2 activation in the OFC.

JAK2 kinase is classically associated with 3 main effector pathways: STAT transcription factors, Ras/ERK, and PI3/Akt signaling. To establish the signaling response to exogenous CNTF within the OFC, we performed a dose response experiment. CNTF did not activate Ras/ERK signaling at the concentrations tested. With low concentrations of CNTF (50nM), we observed a dissociation of the activation of JAK2 and STAT3. The Akt activation profile was similar to that of JAK2, with significant increases at 50nM CNTF. Because of these observations and earlier evidence suggesting STAT3 is unlikely to mediate JAK2 effects on plasticity (31), we focused on Akt. When Akt signaling was inhibited using LY294002, CNTF failed to exert beneficial effects on reversal learning in stressed rats. These observations align with ample literature implicating Akt-dependent mechanisms in the therapeutic effects of other neurotrophins such as BDNF (43, 44). Thus, Akt emerges as a potential site of convergence for the neuronal effects of diverse neurotrophins, like BDNF and CNTF that have distinct cellular origins and signaling pathways, but may share common effector signals. CNTF may therefore add a level of modulation that complements the actions of other neurotrophins on plasticity.

Akt has a prominent role in regulating protein synthesis by inhibiting TSC-1, itself an inhibitor of mTOR/S6K signaling (45). Deficits in mTOR-dependent translation have been associated with depressive behavior (46) and cognitive dysfunction in fragile X syndrome, autism spectrum disorder and schizophrenia (47–49). Indeed, the rapid antidepressant effects of ketamine have been ascribed to BDNF activating Akt/mTOR/S6K signaling (46, 50–52). Akt also phosphorylates serine residue 9 on GSK3β, which inactivates this kinase (53). GSK3 β functions, including endocytosis, synaptic pruning, modulates a variety of neuronal spine remodeling and neuronal excitability (54–57). A substantive literature has shown the association of GSK3 β kinase with mood disorders, memory impairments and psychosis in both humans and animal models (58–62), and it has been suggested that the mechanism of action of several psychoactive drugs includes GSK3 β inhibition (60). Mechanistically, GSK3β activity has been associated with protein degradation. In non-neuronal cells, proteins phosphorylated by GSK3 β are targeted to the proteasome (63). In neurons, Arc has been identified as a substrate for GSK3 β-dependent proteolytic degradation (64). Therefore Akt can control stability of plasticity-related proteins through regulation of GSK3 β. Finally, in support of our observations, the ability of JAK2 to induce pS9-GSK3 β has been reported also in non-neuronal cells (65). In the present study, CNTF increased GSK3β phosphorylation in both non-stressed and stressed rats, whereas it increased activation of S6K only in non-stressed rats. This illustrates the fact that signaling is influenced by physiological state, and the importance of testing in both normal and compromised conditions. The data also suggest that, following stress, CNTF may not engage the full range of Akt signals in the OFC (see sex difference discussion below). At any rate, the results indicate that one mechanism behind the beneficial effects of CNTF is its ability to downregulate GSK3β activity via Akt.

Somewhat unexpectedly, CNTF also elicited a non-canonical pathway involving p38 and Mnk1 activation, leading to phosphorylation of eIF4E. Both p38 and the downstream effectors Mnk1/eIF4E have been described as important players in synaptic plasticity and in different forms of LTD (66–68). Alterations in their expression or function has been seen in mood disorders and in relation to cognitive deficits associated with autism spectrum disorders (69–71). Although p38 is not classically considered an effector of JAK2 signaling, there is evidence that CNTF can activate p38 in a JAK-dependent manner in neurons (72). eIF4E is the rate limiting step of CAP-dependent translation initiation. Phosphorylation by Mnk1 affects eIF4E binding to initiation partner eIF4G, increasing translation (73). In recent studies, Mnk1-dependent phosphorylation of eIF4E relieved binding of the translational repressor CYF1P1 from eIF4E, increasing translation of a subset of mRNA involved in cytoskeleton remodeling and neurotransmission (74). We have previously reported that knock-down of JAK2 but not STAT3 in cortical neurons reduced steady state levels of Arc (31), a regulator of LTD and target of Mnk1/eIF4E-selective translation (75). Thus, one hypothesis to be tested in future work is that CNTF/JAK2 signaling regulates plasticity in the OFC by modulating degradation and/or selective translation of synaptic proteins, such as Arc.

Sex differences in stress and CNTF effects.

These experiments revealed interesting response differences between male and female rats. Firstly, the stress-induced decreases in CNTF, pJAK2 and pT-Akt persisted up to 72h in male but not female rats, who showed reduction only at 24h after stress. This suggests that the cascade is impacted by stress in both sexes, but recovers faster in females. However, this does not imply complete recovery of function in females, as they still displayed reversal learning deficits that were corrected by administration of exogenous CNTF, suggesting that although endogenous CNTF levels had returned to control levels, they may not have yet been sufficient to overcome the impairment in reversal learning after stress. Consistent with an apparent recovery of JAK2/Akt signaling by 72h, we observed baseline effects of LY294002 at that time in females but not males (Figure 3E). Further, Akt inhibition did not fully prevent the beneficial effects of CNTF on reversal learning in females, suggesting that alternate Akt-independent pathways (possibly p38/Mnk/eIF4E) are more readily engaged in females than in males. Additionally, throughout our behavioral tests females completed the tasks in fewer trials than males, suggesting a baseline difference in performance between the two sexes. Thus, although the behavioral effects of stress and JAK2 activation are ultimately analogous in the two sexes, the specific underlying mechanisms and/or timing may partially diverge.

In conclusion, from these data, we propose a model to explain the beneficial effects of CNTF in stressed rats (Figure 4S). We postulate that in the OFC, CNTF activates JAK2, which activates Akt, resulting in inhibition of GSK3β. In parallel, CNTF/JAK2 activate a non-canonical p38-Mnk1 pathway that leads to phosphorylation of the translation initiation factor eIF4E. By these mechanisms, CNTF/JAK2 may control the degradation and translation of specialized mRNAs involved in plasticity that are necessary for functional changes in the OFC during reversal learning.

Together with our earlier studies, the present data suggest that JAK2 signaling may be targeted therapeutically. In clinical trials, encapsulated CNTF prevents loss of retinal cells in macular degeneration and other retinal degenerative diseases (76). Leptin, another JAK2/Akt activator, is used in replacement regimens for individuals with ob gene mutations, and has been shown to increase brain volume and plasticity, and improve cognition (77, 78). Challenges remain for the safe use of these neuronal growth factors in humans (79, 80), but targeted modulation of JAK2 signaling may help the development of more effective treatments for cognitive deficits in psychiatric disorders.

Highlights.

• Reversal learning is impaired by stress in many psychiatric conditions

• Stress reduces CNTF levels and JAK2 signaling in the orbitofrontal cortex

• Restoring CNTF/JAK2 signaling after stress overcomes reversal learning deficits

• These effects require Akt, GSK3β and p38/Mnk1 signaling

• CNTF may regulate plasticity-related proteins required for reversal learning

Abbreviations:

CNTF

ciliary neurotrophic factor

JAK2

Janus kinase 2

Akt

AKT serine/threonine kinase 1

STAT3

signal transducer and activator of transcription 3

ERK

extracellular signal-regulated kinase

S6K

ribosomal protein S6 kinase

GSK3β

glycogen synthase kinase 3β

Mnk1

mitogen-activated protein kinase interacting protein kinase 1

eIF4E

eukaryotic initiation factor 4E

OFC

orbitofrontal cortex