Stimulus-specific and differential distribution of activated extracellular signal-regulated kinase in the nucleus accumbens core and shell during Pavlovian-instrumental transfer

Abstract

The ability of reward-predictive cues to potentiate reward-seeking behavior—a phenomenon termed Pavlovian-instrumental transfer (PIT)—depends on the activation of extracellular signal-regulated kinase (ERK) in the nucleus accumbens (NAc). Here, we utilized immunohistochemistry to investigate the subregional pattern of ERK activation during PIT, and the contribution of different elements in the PIT condition to the distribution of ERK signaling in the NAc of rats. We found that the occurrence of reward-seeking behavior (lever pressing) did not affect ERK activation in either the core or the shell of the NAc. In contrast, presentation of the reward-predictive cue (auditory conditioned stimulus) caused a significant increase in ERK activation in both subregions of the NAc, with the effectbeing slightly more robustin the core than the shell. Different from the pattern evoked by the reward-predictive cue, presentation of the reward itself (food pellets) had no effect on ERK activation in the core but caused a pronounced increase in ERK activation in the shell. Taken together, our results demonstrate that ERK signaling in the NAc during PIT involves both the core and the shell and is driven by the conditioned cue irrespective of whether the situation permits engagement in reward-seeking behavior. Furthermore, our results show that the subregional distribution of ERK signaling in the NAc evoked by rewards differs from that evoked by cues that predict them. The stimulus-specific differential pattern of ERK signaling described here may present the molecular complement to stimulus-specific increases in NAc cell firing reported previously.

Introduction

Stimuli that are repeatedly paired with reward, i.e., appetitive conditioned stimuli (CSs), have the remarkable ability to potentiate operant responding to obtain the reward, a behavioral phenomenon known as Pavlovian-instrumental transfer (PIT; Rescorla 1967; Lovibond 1983). This property of CSs is generally beneficial for survival as it enhances the procurement of resources; however, it can be maladaptive when it leads to overconsumption of reward and contributes to behavioral disorders such as obesity and drug abuse (O’Brien et al. 1998; Everitt and Wolf 2002; Crombag et al. 2008; Volkow et al. 2009). Several lines of research, involving both animal and human subjects, have implicated the nucleus accumbens (NAc) as a key structure for mediating PIT (Corbit et al. 2001; Hall et al. 2001; Talmi et al. 2008). Consistent with a critical role for the NAc in PIT, we found that the activation of the signal transduction enzyme extracellular signal-regulated kinase (ERK) is increased in the NAc upon exposure to an appetitive CS, and that inhibition of CS-evoked ERK activation in the NAc completely abolishes PIT (Shiflett et al. 2008).These findings indicate that ERK activation in the NAc is required for the ability of CSs to potentiate reward seeking.

The NAc frequently is considered to serve as the limbic–motor interface (Mogenson et al. 1980; Nicola 2007). This ventral striatal structure is not a uniform nucleus but comprises two primary subregions, the core and the shell. Anatomical, pharmacological, neurochemical, and behavioral heterogeneity have been demonstrated among the NAc subregions (Voorn et al. 1989; Zahm and Brog 1992; Groenewegen et al. 1999; Zahm 2000; Di Chiara 2002; Baldo and Kelley 2007). The diverse nature of the subregions’ output targets, with the core projecting primarily to motor structures and the shell to limbic areas (Zahm and Brog 1992), suggests that the two subregions may contribute differentially to, or to different aspects of, the PIT phenomenon (Shiflett and Balleine 2011). In our previous observations of ERK activation in the NAc during PIT, we did not differentiate between the core and shell. To determine whether ERK activation is evoked in both of these subregions and to dissect out the relative contribution of the reward-seeking behavior versus the CS to the ERK activation pattern, we here examined the subregional distribution of ERK activation during PIT, as well as whether the opportunity to engage in reward-seeking behavior affects the subregional distribution of ERK activation in the NAc.

Materials and methods

Subjects

Sixty-three male Sprague–Dawley rats (Hilltop Lab Animals, Scottdale, PA) weighing between 250 and 275 g on arrival were used in this study. Rats were handled and weighed daily. Beginning 5 days prior to training and continuing until testing was completed, rats were food restricted to ~90 % of the body weight of free-feeding age-matched rats. Rats were housed individually and maintained on a 12-h light/dark cycle with ad libitum access to water. All procedures were performed during the light cycle and in accordance with the NIH Guide for the Care and Use of Laboratory Animals and with the approval of the Institutional Animal Care and Use Committee of the University of Pittsburgh. The behavioral experiments were performed in the Rodent Behavior Analysis Core of the University of Pittsburgh Schools of Health Sciences.

Behavioral apparatus

All procedures took place in standard rat operant chambers (30 cm × 23 cm × 23 cm; Med Associates, St. Albans, VT) equipped with a house light, a metal grid floor, and a loudspeaker. A food cup was mounted on the front wall and attached to a pellet dispenser that released a single food pellet when activated. A photobeam source and detector were mounted onto the sides of the food cup to monitor food cup approaches. One retractable lever was mounted ~4 cm to the left side and a second lever 4 cm to the right side of the food cup. The chambers were enclosed in light- and sound-attenuating cubicles equipped with fans that generated background noise to obscure external noise. MED-PC IV program (Med Associates) controlled the equipment and recorded food cup approaches and lever presses.

Behavioral procedures

The training protocols were adapted from Shiflett et al. (2008). Five behavioral groups were used in this study: TOY, TO, PIT, PAV, and REWARD (see Table 1). One day prior to Pavlovian training, all animals were habituated to the chamber in a single 30-min session. During this session, the only explicit cue was the illumination of the house light at the onset and its extinguishment at the termination of the session.

Table 1.

Behavioral paradigms for individual groups

Phase	Group
	TOY	TO	PIT	PAV	REWARD
Pavlovian training  (6 days)	CS only	CS only	CS + reward	CS + reward	CS + reward
Instrumental training (6 days)	Non-contingent  reward delivery	Reward contingent on  lever pressing	Reward contingent on  lever pressing	Context only	Reward contingent on  lever pressing
Test	CS; levers present	CS; levers present	CS; levers present	CS only	Reward only

Pavlovian conditioning

During the Pavlovian conditioning phase, animals from the PIT, PAV, and REWARD groups completed six daily sessions, with each daily session consisting of eight pairings between a CS (90-s 3-kHz 80-dB tone) and delivery of three to four food pellets (45-mg dustless food pellets, BioServe, Frenchtown, NJ) dispensed on a variable time (VT) schedule of 20 s. For half of the trials, the VT schedule for pellet delivery was initiated with tone onset; for the remaining half, the VT schedule was delayed by 30 s after tone onset to permit measurement of food cup approach in the absence of the US. The type of trial (food throughout vs. only during the last 60 s of the CS) was varied randomly between and within sessions, with the restriction that each type of trial was presented four times per session. The intertrial interval (CS onset to next CS onset) varied between 210 and 330 s (mean = 270 s). For each session, the number of photobeam breaks was recorded during the first 30 s of the CS periods during which food delivery was delayed, and the 30 s prior to CS onset (CS and preCS periods, respectively). The studies included two kinds of control groups, a tone-only yoked (TOY) and a tone-only (TO) group. During the Pavlovian conditioning phase, both control groups were presented with the identical number and pattern of 90-s tones as the experimental groups, but no food pellets were delivered during the sessions.

Instrumental conditioning

After completion of Pavlovian conditioning, rats in the PIT, REWARD, and TO groups underwent instrumental conditioning. Each session began with illumination of the house light and extension of both levers into the chamber. One lever, designated by the experimenter as the active lever, delivered a single food pellet into the food cup when depressed. Depression of the other lever, the inactive lever, had no programmed consequences. The active lever was randomly assigned to the left or the right lever and was counterbalanced within and across cohorts. During this first session, rats received a single pellet for every depression of the active lever (termed a fixed-ratio one, or FR1, schedule). The first session terminated after rats earned a total of 50 pellets. Rats were returned to their home cages for a 45-min intersession interval after which the second training session began. For the second session, on average three presses on the active lever were required for the delivery of a pellet (random-ratio 3, or RR3). The RR3 session terminated after 50 pellets were delivered. On the subsequent 5 days, animals underwent two daily RR5 sessions with a 45-min intersession interval in their home cages. Each RR5 session terminated after 100 pellets were delivered. Levers were extended for rats of the TOY group, but depression of the levers did not produce a food pellet. Instead, these control rats received a food pellet whenever a rat from the PIT group trained concurrently earned a food pellet. Rats in the PAV group were placed into the conditioning chamber during the instrumental sessions, but no stimuli other than the house light were presented.

Pavlovian reminder and extinction

The day before the test day, rats received a Pavlovian conditioning reminder session that was identical to prior Pavlovian training sessions, as described above. The reminder session was followed by 15 min in the home cage and then by a brief extinction session. During the extinction session, both levers were extended into the chamber for all groups, except for the PAV group, but food pellets were not delivered to any of the groups. The extinction session was terminated after instrumentally conditioned rats had pressed the active lever for 10 min or made 300 unrewarded lever presses (whichever occurred first). This extinction session served to avoid a ceiling effect during the transfer test. For animals in the PAV and TOY groups, the extinction session was terminated after 10 min.

Pavlovian-instrumental transfer test

The PIT test began with the illumination of the house light and extension of both levers into the chamber for all groups, except for rats in the PAV and REWARD groups. During the first 6 min, both levers were available, but no stimuli were presented. Thereafter, four 90-s tones were presented at a fixed 270-s interstimulus interval. The number of active and inactive lever presses and food cup approaches were recorded for each 90-s CS period and the 90-s period prior to the CS (preCS period). For rats in the PAV and REWARD groups, the levers were not extended and, therefore, only food cup approaches were recorded. Rats in the PAV group received the same series of tone presentations as described above, and food cup approaches were recorded during the preCS and CS periods. Rats in the REWARD group did not receive tone presentations but instead were given food pellets on a VT schedule of 20 s during CS time periods.

Immediately after the test, rats were anesthetized with an intraperitoneal (i.p.) injection of chloral hydrate (300 mg/kg dissolved in 0.9 % saline) and either decapitated or transcardially perfused. In the case of decapitation, brains were removed quickly and briefly immersed in 2-methylbutane on dry ice and then stored at −80 °C until the excision of NAc tissue in preparation for Western blot analysis. In the case of transcardial perfusion, the perfusion medium consisted of 50 ml of 0.9 % saline containing heparin (Acros Organics, Antwerpen, Belgium; 20 i.u./ml) and sodium fluoride (2 mM), followed by 500 ml of 4 % periodate-lysine-paraformaldehyde (PLP) fixative in 0.1 M sodium phosphate buffer (PB, pH 7.4). Brains were removed and post-fixed overnight at 4 °C in PLP fixative, in preparation for immunohistochemistry.

Western blot analysis

NAc tissue was harvested as described previously (Shiflett et al. 2008, 2009). Briefly, NAc samples were excised from 1-mm thick coronal sections by placing a tissue punch (2 mm in diameter, which yields a sufficient amount of protein for analysis; Fine Science Tools, Foster City, CA) over the core and shell, so as to include about equal parts of both subregions in each sample. Samples were homogenized in a buffer containing 150 mM NaCl, 1 mM EDTA, 50 mM Tris (pH 7.4), 0.05 % SDS, 1 % Triton-X-100, 1 mM dithiothreitol (DTT), 2 mM sodium fluoride, 1 mM orthovanadate, 2 mM sodium pyrophosphate, 1 mg/ml pepstatin, and Protease inhibitor cocktail (Merck KGaA, Darmstadt, Germany). Homogenates were centrifuged for 15 min at 14,000 rpm. For each sample, protein concentration of the supernatant was determined in triplicates using a bicinchoninic acid assay (Pierce, Rockford, IL). Samples were diluted to a uniform concentration with homogenization buffer and sample buffer containing 2.5 M Tris (pH 6.8), 40 % glycerol, 8 % SDS, and 30 mg/ml DTT, and then were heated to 95 °C for 5 min. Forty-five micrograms of protein per sample were loaded for resolution by SDS-PAGE and subsequently transferred to Immobilon membranes. Membranes were blocked for 1 h at 22 °C in a Tris-buffered saline and 0.1 % Tween 20 (TBST) solution containing 5 % dried non-fat milk followed by an overnight incubation at 4 °C in TBST containing 5 % bovine serum albumin and an antibody specific for dual-phosphorylated (T202/183 and Y204/185), activated ERK1/2 (1:2,500 dilution; Cell Signaling Technology, Beverly, MA). Membranes were washed in TBST solution and incubated with an HRP-linked secondary antibody (anti-rabbit, 1:5,000; Cell Signaling Technology) for 1 h at 22 °C. Phospho-ERK2 (pERK2) immunoreactivity was visualized on blot images captured with a CCD camera (Hamamatsu Photonics, Japan) using enhanced chemiluminescence reagent (Lumiglo; Cell Signaling Technology). To probe each sample for total ERK1/2, membranes were stripped of their antigens via incubation for 45 min at 55 °C in a solution containing 62.5 mM Tris (pH 6.7), 2 % SDS, and 0.62 % β-mercaptoethanol. Membranes were blocked and reprobed with an antibody specific for phosphorylated and non-phosphorylated ERK1/2 (1:2,500; Cell Signaling Technology), as described above. Total ERK2 (tERK2) immunoreactivity was visualized as described above. pERK2 and tERK2 immunoreactivity were analyzed using densitometry software (UVP Labworks, Upland, CA). ERK2 activation was determined for each sample by dividing pERK2 immunoreactivity by tERK2 immunoreactivity for a given sample and normalizing the resulting ratios to similar ratios for samples from rats of the TOY group run on the same membrane.

Immunohistochemistry

Following overnight fixation, brains were cut into 50-μm coronal sections using a vibratome (Vibratome Series 1000) and collected in PB in four adjacent series. The fixation process was quenched by placing sections in 1 % sodium borohydride in PB for 30 min and then rinsing them several times in PB. Sections were stored at −20 °C in cryopreservant solution until immunohistochemical staining. In preparation for staining, free-floating sections were rinsed first in PB overnight at 4 °C and then three times for 30 min in Tris-buffered saline (TBS; 50 mM Tris in 0.9 % NaCl, pH 7.6). After rinsing, sections were placed in blocking buffer (5 % normal goat serum and 0.3 % Triton X-100 in TBS) for 1 h at 22 °C, and then incubated with anti-phospho-ERK antibody (1:800; Cell Signaling Technology) in blocking buffer for 24–48 h at 4 °C. After primary antibody incubation, sections were washed in TBS and incubated with biotinylated goat anti-rabbit IgG secondary antibody (1:500; Vector Laboratories, Burlingame, CA, USA) in blocking buffer for 2 h at 22 °C. Sections were then washed in TBS and incubated with avidin–biotin conjugate (1:500 for each A and B reagent; Vector Laboratories, Burlingame, CA) in blocking buffer for 90 min at 22 °C. Afterward, sections were rinsed in 0.1 M sodium acetate buffer (SA, pH 6.0) three times for 10 min. Immunostaining was developed in SA buffer containing 0.022 % diaminobenzidine, 0.003 % hydrogen peroxide, and 2.5 % nickel sulfate. After three additional washes, sections were mounted on gelatin-coated slides, dried at room temperature, dehydrated in ethanol, cleared in xylene, and coverslipped with Cytoseal 60 (Richard-Allan Scientific).

Analysis of pERK1/2 immunohistochemistry

Images of sections stained for pERK1/2 were captured using a 10× objective and a digital camera (Micrometrics 3.2 MP) mounted on a light microscope (Leitz Orthoplan 2). The core and shell of the NAc, and the dorsolateral and dorsomedial striatum (DLS and DMS, respectively) were identified using a rat brain atlas (Paxinos and Watson 2007). For each subject, sections at a frequency of 200 μm between +1.1 and +2.0 mm anterior to bregma were captured for ~40 images (10 per subregion) per rat. Digitized images were used to estimate the number of pERK1/2-immunoreactive (pERK-IR) cells in the NAc core by placing a 200 μm × 400 μm counting window just dorsal to the anterior commissure. For the medial NAc shell, the number of pERK1/2-IR cells was estimated by placing a 400 μm × 200 μm counting window 400 μm medial to the anterior commissure between +1.1 and 1.2 mm anterior to bregma, 500 μm medial to the anterior commissure between +1.3 and 1.4 mm anterior to bregma, and 600 μm medial to the anterior commissure between +1.5 and 2.0 mm anterior to bregma (see Fig. 3 for examples of counting window placements). For the dorsal striatum, the number of pERK1/2-IR cells was estimated by placing a 600 μm × 600 μm counting window either next to the lateral ventricle in the most medial portion of the dorsal striatum (DMS), or at the same dorsoventral level ~2.0 mm lateral to the lateral ventricle, in the most lateral portion of the dorsal striatum (DLS). Digitized images of the counting windows were created and used to obtain cell counts with the aid of an image analysis macro written for NIH ImageJ (1.43u). To account for differences in background staining, image thresholds were adjusted prior to cell counting. Briefly, using a macro written for ImageJ, a background mask of the image was created by replacing each pixel with the median of the surrounding pixels in a 30 pixel radius. Next, the contrast of each pixel was calculated with respect to the background from the following equation: (background − image)/(background + image). Then, a binary image was created using an arbitrary threshold of 0.10 for all images. Lastly, a watershed algorithm and a particle analysis algorithm were applied to separate any overlapping cells and to discard particles that were too small to be considered as cells, respectively. Cell counts derived from ImageJ (NIH) were verified by an experimenter blind to the experimental conditions using the original image of the counting window. For each subject, the number of pERK1/2-immunopositive cells was counted in both hemispheres, yielding a total of ten estimates per region per animal.

Fig. 3.

Reward-seeking behavior alone does not affect the level of active ERK in either the core or the shell. a Representative low-magnification photomicrographs of a section taken from near the most anterior (AP +2.0; left) and the most posterior end of our counting range (AP +1.2; right). Black rectangles illustrate placements of the counting window for pERK-IR cell counts in the core and the shell, respectively. The same counting window placements were used for the data depicted in Figs. 3, 4, 5. b Mean ± SEM number of pERK1/2-IR cells per mm² in each subregion for the TOY (n = 9) and the TO (n = 8) groups. The number of pERK1/2-IR cells were comparable between the two control groups in both the core and the shell, and were similarly greater in the shell than the core (**P < 0.01). c Representative high-magnification photomicrographs of pERK-IR cells in the core from a rat of the TOY group and a rat of the TO group (scale bar see d; ac anterior commissure). d Similar representative high-magnification photomicrographs of pERK-IR cells in the shell from a rat of each of the two control groups (scale bar 40 μm; applies to c and d)

Statistical analysis

Data from Pavlovian and instrumental training were compared using analysis of variance (ANOVA) for repeated-measures with group as between-subject factor and training day as within-subject factor followed by Bonferroni post hoc comparisons of individual groups. PIT transfer and PAV test data were compared using ANOVA for repeated-measures with group as between-subject factor and interval (preCS, CS) as within-subject factor followed by post hoc comparisons. For experiment 1, pERK2/tERK2 ratios were compared with a one-way ANOVA followed by post hoc comparisons. For experiment 2, cell counts per mm² from the TOY and TO control groups were compared using ANOVA with group as between-subject factor and area (core, shell) as within-subject factor. Comparisons between pooled controls and experimental groups were conducted using one-way ANOVAs for each subregion (core, shell, DLS, DMS) separately, followed by post hoc comparisons. All statistical analyses were performed using the SPSS software package version 19.0 (Chicago, IL). For all statistical comparisons, P < 0.05 was the criterion for significance.

Results

Experiment 1: NAc ERK activation during Pavlovian-instrumental transfer: does lever-pressing matter?

We previously found that active ERK in the NAc, as indicated by the density of NAc cells immunopositive for dual-phosphorylated, active ERK (pERK), is higher in rats exhibiting PIT than in control rats that were familiarized with the CS during the Pavlovian conditioning phase but that did not receive lever-press training during the instrumental conditioning phase (Shiflett et al. 2008). Because control rats did not engage in lever pressing during the PIT test, it is possible that the difference in pERK density between groups resulted, at least in part, from the difference in instrumental responding during the test. Furthermore, we did not measure total ERK levels, thus leaving it unclear whether the increase in pERK density resulted from an increase in total ERK or from higher activation of the available ERK. To address both of these issues, we here used Western blot analysis and compared both ERK activation and total ERK levels in the NAc between experimental rats (PIT group), control rats that received lever-press training (TO group), and control rats that received food pellets yoked to the experimental group but no lever-press training (TOY group; see Table 1).

Behavior

At the beginning of Pavlovian training, discriminative food cup approach rates (number of approaches/min during the CS minus number of approaches/min during the preCS period) were equally low in all three groups; however, as training progressed, these rates increased for rats in the PIT group (n = 8), but not for rats in either the TO (n = 6) or the TOY (n = 7) control groups. ANOVA confirmed these differences, as indicated by a significant group × day interaction (F_(10,90) = 5.75; P < 0.01). Post hoc comparisons showed that the three groups did not differ from one another on the first day of training. After 6 days of training, the mean discriminative approach rate of the PIT group was significantly higher than that of either the TO or the TOY group (4.7 ± 0.9, 0.1 ± 0.4, and 0.8 ± 0.3 approaches/min, respectively; P < 0.01), whereas the rates of the two control groups did not differ significantly from one another. During instrumental training, rats in both the TO and PIT groups acquired instrumental responding at a comparable rate, and their response rates increased across training days. As expected, rats in the TOY group did not acquire lever pressing. ANOVA confirmed these trends, as indicated by a significant group × day interaction (F_(10,90) = 4.46; P < 0.01). On the last day of training, the mean lever-press rates for the PIT, TO, and TOY groups were 24.9 ± 3.8, 28.6 ± 8, and 0.1 ± 0.1 presses/min, respectively. Post hoc comparisons revealed that mean rates of the PIT and the TO groups did not differ for any of the six training days, which indicates that rats in these two groups experienced similar instrumental training prior to the PIT test.

During the PIT test, rats in the PIT group pressed the lever significantly more during the CS than the preCS period, and thus demonstrated a robust PIT effect (Fig. 1a). Rats in the TO group also displayed reasonably high lever-press rates; however, their response rates were similar during the preCS and CS periods. As expected, rats in the TOY group essentially did not press the lever during either time period. ANOVA confirmed these differential trends, as indicated by a significant group × stimulus period interaction (F_(2,18) = 21.48; P < 0.01). Post hoc comparisons revealed that the mean lever-press rate during the CS was significantly higher than that during the preCS period for the PIT group (P < 0.01), but not for the TO or the TOY groups. The mean lever-press rate during the preCS period was significantly lower for the TOY group than for either the TO or the PIT groups (P < 0.01), with no difference between the latter two group. The mean lever-press rate during the CS also was significantly lower in the TOY group than in either the TO or the PIT group (P < 0.01); however, in contrast to the preCS period, the mean lever-press rate during the CS was significantly higher in the PIT than the TO group (P < 0.05). ANOVA of the total lever presses during the test revealed a significant effect of group (F_(2,18) = 11.03; P < 0.01), and post hoc comparisons confirmed that the total number of lever presses were significantly lower for rats in the TOY group compared with rats in the PIT or TO groups (P < 0.05), with no significant difference between the latter two groups. The total number of lever presses for the PIT, TO, and TOY groups during the PIT test were 256 ± 50, 173 ± 40, and 11 ± 4, respectively.

Fig. 1.

Pavlovian-instrumental transfer, but not lever pressing, results in increased ERK2 activation in the NAc. a Mean ± SEM lever presses per minute for the preCS and CS periods of each group during the PIT test. Only for rats of the PIT group (n = 8), but not for rats of either the TOY group (n = 7) or the TO group (n = 6), were response rates during the CS significantly higher than those during the preCS period (**P < 0.01). Lever-press rates during the preCS period did not differ between the PIT and TO groups, but lever-press rates during the CS were significantly higher for rats in the PIT group than rats in the TO group (^#P < 0.05), whose lever-press rates, in turn, were significantly higher than those of rats in the TOY group (^##P < 0.01). b Mean ± SEM normalized pERK2 immunoreactivity/tERK2 immunoreactivity in NAc samples from each group. pERK2/tERK2 immunoreactivity was significantly higher in NAc samples from rats of the PIT group than NAc samples from rats of either the TO or the TOY control group (*P < 0.05). Representative pERK and tERK immunoblots for each treatment group are shown on the right. The lower (p42) band was used for analysis

Western blot analysis

Immediately after the test, rats were decapitated and the NAc was excised for the determination of activation of ERK2, the ERK isoform we showed previously to be regulated differentially during Pavlovian conditioning (Shiflett et al. 2008). As shown in Fig. 1b, we found that ERK2 activation, as indicated by pERK2 immunoreactivity/tERK2 immunoreactivity (see “Materials and methods”), in NAc samples from rats in the PIT group was about 2.5-fold greater than ERK2 activation in NAc samples from rats in either the TO or the TOY group. ANOVA indicated a significant effect of group (F_(2,20) = 6.34; P < 0.01), and post hoc comparisons confirmed that ERK2 activation was significantly higher in the PIT group than in either the TO or the TOY group (P < 0.05). Importantly, ERK2 activation did not differ between the two control groups. Thus, despite the marked difference in lever-press rates between the two control groups, NAc ERK2 activation was comparable between them. To determine whether the PIT-associated increase in ERK2 activation resulted from an increase in pERK2, rather than a decrease in tERK2, we also compared tERK2 immunoreactivity between the three groups. ANOVA revealed no significant group effect (F_(2,20) < 1). The lack of an effect on tERK2 level indicates that the observed changes in ERK2 activation stem from an increase in activated ERK2. By extension, these findings confirm that our previous observations of PIT-associated changes in pERK density (Shiflett et al. 2008) reflected changes in ERK activation. Taken together, our results show that NAc ERK2 activation is increased during PIT, and that this increase does not stem from conditioned lever-pressing per se.

Experiment 2: NAc ERK activation during Pavlovian-instrumental transfer: subregional pattern as a function of presented stimulus

In light of demonstrated dissociations between the core and shell subregions of the NAc at the anatomical, neurochemical, and functional levels (Voorn et al. 1989; Heimer et al. 1991; Berendse et al. 1992; Zahm and Brog 1992; Brog et al. 1993; Kelley 1999), it is possible that the PIT-relevant molecular signal we described previously (Shiflett et al. 2008) differs between NAc subregions. Furthermore, the apparent dependence of NAc ERK activation on exposure to the CS and not on lever pressing observed in the previous experiment may not apply to the two subregions uniformly. To address these issues, we compared the number of pERK-IR cells in the core and shell separately between the TO and TOY control groups, the PIT group, and a group of rats that received Pavlovian conditioning but no lever-press training (PAV group; see Table 1).

Behavior

During Pavlovian training, rats in the PIT (n = 8) and PAV (n = 9) groups developed discriminative food cup approach behavior, whereas control rats in either the TO (n = 8) or the TOY group (n = 9) did not. ANOVA revealed a significant group × day interaction (F_(5,150) = 3.82; P < 0.01), and post hoc tests confirmed that, although there was no significant difference in approach behavior between groups at the beginning of training, rats in the PIT and PAV groups exhibited significant and similar discriminative approach behavior on the last day of Pavlovian training, whereas rats in the TO or TOY group did not (P < 0.01). The mean discriminative approach rates for the PIT, PAV, TO, and TOY groups on the final day of Pavlovian conditioning were 6.3 ± 1.2, 4.9 ± 1.2, −0.2 ± 0.2, and 0.0 ± 0.4 approaches/min, respectively. During instrumental training, rats in both the PIT and the TO group acquired lever pressing at comparable rates across training sessions, whereas rats in the TOY group did not lever press. ANOVA indicated a significant group × day interaction (F_(5,110) = 8.18; P < 0.01), and post hoc comparisons confirmed no differences between the PIT and TO groups for any of the instrumental training days, but both groups engaged in lever pressing at significantly higher rates than the TOY group on each of the training days (ps < 0.01). On the final day of instrumental conditioning, the lever-press rates were 33.6 ± 6.1, 31.5 ± 5.9, and 0.2 ± 0.1 presses/min for rats in the PIT, TO, and TOY groups, respectively.

Lever-press rates during the PIT test are shown in Fig. 2a. Similar to the results from the previous experiment, rats in the PIT group pressed the lever about twice as frequently during the CS compared to the preCS period. In comparison, lever-press rates for rats in the TO group did not differ between the preCS and CS periods, and rats in the TOY group did not press the lever during either the preCS or CS periods. ANOVA of the lever-press rates revealed a significant group × period interaction (F_(2,22) = 11.82; P < 0.01). Post hoc tests confirmed that only the PIT group exhibited significantly enhanced lever pressing during the CS compared to the preCS period (P < 0.01). Lever-press rates during the preCS period did not differ between the TO and PIT groups, but were significantly higher in both of these groups compared with the TOY group (P < 0.05). In contrast, lever-press rates during the CS were significantly higher in the PIT group compared to the TO group (P < 0.01), which, in turn, was significantly higher than in the TOY group (P < 0.05). The total number of lever presses for the PIT, TO, and TOY groups during the PIT test were 218 ± 32, 152 ± 33, and 6 ± 2, respectively, and did not differ between the PIT and TO groups, but both of these groups pressed significantly more than rats in the TOY group (F_(2,22) = 18.55; P < 0.01; post hoc test, P < 0.01). Discriminative food cup approach rates during the PIT test for rats in the PAV and TOY groups are shown in Fig. 2b. Whereas rats in the PAV group exhibited pronounced discriminative approach behavior, rats in the TOY group showed no differential response. A t test for independent groups confirmed that discriminative responding was significantly higher in the PAV group compared to the TOY control group (t₍₁₆₎ = 4.06; P < 0.01). However, the total number of food cup approaches during the test did not differ between the two groups (51 ± 8 and 45 ± 4 approaches/min for the PAV and TOY groups, respectively; t₍₁₆₎ < 1).

Fig. 2.

Appetitive conditioning leads to PIT and discriminative food cup approach during testing. a Mean ± SEM lever presses per minute for the preCS and CS periods of the TOY group (n = 9), the TO group (n = 8), and the PIT group (n = 8) during the PIT test. Only for rats of the PIT group, but not for rats of either the TOY or the TO group, were response rates during the CS significantly higher than those during the preCS period (**P < 0.01). Lever-press rates during the preCS period did not differ between the PIT and TO groups, but lever-press rates during the CS were significantly higher for rats in the PIT group than rats in the TO group (^##P < 0.01), whose lever-press rates, in turn, were significantly higher than those of rats in the TOY group (^#P < 0.05). b Mean ± SEM discriminative food cup approach for rats of the TOY group and rats of the PAV group (n = 9). Discriminative food cup approach was significantly higher for rats in the PAV group than rats in the TOY control group (*P < 0.01)

pERK1/2 immunohistochemistry

Immediately after the test, animals were perfused to assess pERK1/2 immunostaining in the core and shell. As illustrated in Figs. 3b, 4b and d, pERK immunoreactivity was observed throughout the cell body and in the larger dendritic branches of accumbal neurons in both subregions. The great majority of immunopositive cells appeared to have a medium-sized soma (~14 μm). To determine whether lever pressing in the absence of a CS influences the pattern of pERK immunolabeling across the two NAc subregions, we first compared the number of pERK1/2-IR cells in the core and the shell between the TO and TOY control groups. pERK-IR cells were observed throughout the NAc core of rats in both groups, and the distribution appeared equally sparse in the two groups. In comparison, pERK immunolabeling appeared to be populous in the medial shell (~500 μm medial to the ac). Again, this effect was observed similarly in the two control groups (Fig. 3a, b; see Table 2). ANOVA of the number of pERK-IR cells confirmed these observations, as there was a significant effect of area (F_(1,15) = 32.20; P < 0.01), but neither the main effect of group nor the group 9 area interaction was significant. Interestingly, we observed comparable numbers of pERK-IR cells in these two subregions of the NAc of rats that merely received daily handling but no conditioning (n = 3; 9.6 ± 1.7 and 34.2 ± 3.6 pERK-IR cells per mm² for the core and the shell, respectively; see Table 2). In light of these findings, it is reasonable to conclude that the level of pERK immunoreactivity observed in the TO and TOY control groups reflects basal pERK immunoreactivity. Because the results did not differ between our two control groups, we pooled their data for comparison with the experimental groups below.

Fig. 4.

Presentation of a CS during PIT increases the level of active ERK in both the core and the shell. a Mean ± SEM number of pERK1/2-IR cells in the NAc core for the PIT (n = 8), the PAV (n = 9), and the pooled CTRL group (n = 17; pooled data depicted in Fig. 3). The number of pERK1/2-IR cells was significantly greater in both experimental groups than in the CTRL group (*P < 0.05, **P < 0.01). b Representative photomicrographs of pERK-labeled cells in the core from a rat of the PIT group (left) and a rat of the PAV group (right) (scale bar see d; ac anterior commissure). c Similar data as shown in a are depicted for the NAc shell. The number of pERK1/2-IR cells was significantly greater for the PIT group than the CTRL group (**P < 0.01). A similar, albeit less pronounced trend was observed for the PAV group (see text for more details). d Representative photomicrographs of pERK-labeled cells in the shell from a rat of the PIT group (left) and a rat of the PAV group (right) (scale bar 40 μm)

Table 2.

Summary of behavioral and immunohistochemical results (group mean ± SEM) for the individual groups in experiment 2

Measure	Group
	Handled	TOY	TO	PIT	PAV	REWARD
Behavior during test	n/a	−0.1 ± 0.5a	0.2 ± 1.7b	7.6 ± 1.5b	3.0 ± 0.6a	2.8 ± 0.5c
pERK+ cell count CORE	9.6 ± 1.7	15.0 ± 2.9	10.7 ± 2.4	30.3 ± 4.0	39.5 ± 6.9	11.1 ± 2.0
pERK+ cell count SHELL	34.2 ± 3.6	36.2 ± 5.7	27.8 ± 5.7	78.8 ± 15.6	52.0 ± 7.9	138.3 ± 16.3
pERK+ cell count DMS	n.d.	6.9 ± 2.8	6.5 ± 1.5	9.8 ± 3.5	11.6 ± 5.1	n.d.
pERK+ cell count DLS	n.d.	21.0 ± 9.1	16.5 ± 4.5	19.9 ± 4.1	31.2 ± 5.9	n.d.

n/a not applicable, n.d. not determined

^aDiscrimination rate of Pavlovian conditioned response; rate of food cup approaches during the CS minus rate of food cup approaches during the preCS

^bPIT score; rate of active lever presses during the CS minus rate of active lever presses during the preCS

^cFood cup approaches during the period of intermittent food delivery (coincident with the CS period for the other groups) minus rate of food cup approaches during a time period equivalent to the preCS period. See text for further details

Similar to our observations with the controls, pERK-IR cells were distributed relatively sparsely in the core of rats from the PIT and PAV groups. Further, compared with immunolabeling in the dorsal, ventral, and medial regions of the core, labeling was scant in the lateral core in both trained groups, and this pattern was observed throughout the rostral-caudal extent of the NAc (see Fig. 4b). Overall, the density of pERK immunolabeling was greater in the two trained groups compared to the controls (Fig. 4a; see Table 2). ANOVA of the pERK-IR cell counts in the NAc core of rats from the PIT, PAV, and pooled control (CTRL) groups confirmed a significant group effect (F_(2,31) = 13.26; P < 0.01), and post hoc comparisons revealed that the number of pERK-IR cells was significantly higher for both experimental groups compared to the CTRL group (P ≤ 0.01). Importantly, the number of pERK-IR cells did not differ between the PIT and the PAV groups. The difference in the number of pERK-IR cells between the experimental and control groups tended to bemore pronounced in the anterior compared to posterior portion of the core because of relatively lower immunolabeling in the anterior (>+1.7 AP) compared to the posterior core of control rats. To determine whether this trend was statistically significant, we compared pERK-IR cell counts from the two most anterior sections versus those from the two most posterior sections collected from each rat between the three groups. However, ANOVA using groups as between-subject factor and anterior versus posterior subdivision as within-subject factor did not reveal a significant group × subdivision interaction (F_(2,31) < 1), which suggests that the differential patterns of pERK labeling were overall comparable across the anterior–posterior extent examined here.

In the NAc shell, the great majority of pERK-IR cells were located in the medial shell, and very few pERK-IR cells were found in the lateral portion of the shell in rats from both the PIT and PAV groups. The distribution of pERK-IR cells within the medial shell was characterized by patches of sparse and dense staining (see Fig. 4d). Although clusters of pERK-IR cells were found in rats from all four groups, they appeared to be more frequent and often contained more immunopositive cells in rats from the PIT and PAV groups than in control rats. Consistent with this trend, pERK-IR cells overall appeared to be denser in the shell of experimental than control rats (Fig. 4c; see Table 2). ANOVA of the pERK-IR cell counts in the NAc shell confirmed a significant group effect (F_(2,31) = 8.22; P < 0.01); however, post hoc comparisons revealed that only the PIT group expressed significantly more pERK-IR cells than the CTRL group (P < 0.01). Nevertheless, the number of pERK-IR cells did not differ between the two experimental groups, similar to what we observed in the core. Moreover, as suggested by the trend for a greater number of pERK-IR cells in the PAV compared to the CTRL group (see Fig. 4c), a separate comparison between these two groups indicated a significantly greater number of pERK-IR cells in the PAV than the CTRL group (Student’s t test for independent groups, t₍₂₄₎ = 2.49, P < 0.02, two-tailed), and this difference remained statistically significant after adjustment of the P value for a second comparison (e.g., with the PIT group). In contrast, a similar separate comparison between the PIT and PAV groups confirmed the lack of a significant difference between the two experimental groups (t₍₁₅₎ = 1.59, P >0.1). Similar to the rostro–caudal pattern noted in the core, the difference in the number of pERK-IR cells in the shell between the experimental and control groups was more prominent in the anterior compared to the posterior portion of the shell; however, different from the core, the gradient in effect size resulted from overall stronger immunolabeling in the anterior compared to the posterior portion of the shell in experimental rats. To determine whether this qualitative trend in the shell was borne out statistically, we compared pERK-IR cell counts from the two most anterior sections versus those from the two most posterior sections collected from each rat, as described for the core above. ANOVA confirmed a group-dependent difference in pERK-IR cell count between the anterior and the posterior portion of the shell, as indicated by a significant group × subdivision interaction (F_(2,31) = 3.99, P < 0.05), and post hoc comparisons indicated significantly higher immunolabeling in the anterior compared to the posterior subdivision of the shell in rats from the PIT and PAV groups (P < 0.01), but not the shell of rats from the CTRL group.

Taken together, the results indicate a robust increase in the number of pERK-IR cells in both the core and the shell of rats from the PIT group and the core of rats from the PAV group, and a mild increase in the shell of rats from the PAV group. To place these increases in pERK immunolabeling upon exposure to a reward-predictive CS into perspective, not just relative to controls but also relative to ERK signaling evoked by the reward itself, we also examined the number of pERK-IR cells in the core and shell of rats that were trained identically to the PIT group, but that received food pellets in place of tone presentations during the test (REWARD group; n = 8; see Table 1). Similar to controls but different from both experimental groups, the number of pERK-IR cells was very sparse in the core of rats from the REWARD group (see Fig. 5b; Table 2). In contrast, the number of pERK-IR cells in the shell of rats from the REWARD group was very pronounced and, similar to the pattern observed in the two experimental groups, located predominately in the medial shell (see Fig. 5d; Table 2). This distribution pattern within the shell was observed throughout the rostral–caudal extent we examined, although the density of pERK-IR cells was greater in the anterior than the posterior portion of the shell, similar to the trend we observed in the PIT and PAV groups. Comparing numbers of pERK-IR cells in the core of rats from the three experimental groups in terms of fold change from control level, ANOVA indicated a significant group effect (F_(2,24) = 8.46, P < 0.01), and post hoc comparisons revealed that the fold change from control level for rats from the REWARD group was significantly lower than that for rats from either the PIT or the PAV group (P < 0.05; Fig. 5a), with no significant difference between the latter two groups. Indeed, the number of pERK-IR cells in the core of rats from the REWARD group (11.1 ± 2.0 cells) did not present a significant fold change from control level (Student’s t test assuming a population mean of 1.0 or no difference from control, t₍₇₎ < 1.0), whereas the fold change for both the PIT and the PAV group was significant (t₍₇₎ = 4.35 and t₍₈₎ = 3.84, respectively, P < 0.01, two-tailed). Similar comparisons of fold changes in the number of pERK-IR cells in the shell also indicated a significant group effect (F_(2,24) = 10.86; P < 0.01); however, in the case of the shell, the fold change from the control level was significantly greater for rats from the REWARD group than rats from either the PIT or the PAV group (post hoc comparison, P < 0.05; Fig. 5c), again with no significant difference between the latter two groups. Furthermore, the number of pERK-IR cells in the shell of rats from the REWARD group (138.3 ± 16.3 cells) presented a significant fold change from control level (REWARD: t₍₇₎ = 6.52, P < 0.01, two-tailed), as did the changes observed in the PIT and PAV groups (t₍₇₎ = 2.98 and t₍₈₎ = 2.51, respectively, P < 0.05, two-tailed). In light of our findings from experiment 1 that tERK levels are unaffected by prior conditioning history and remain stable during PIT, our observations described here likely reflect differences in ERK activation. Thus, the cross-regional pattern of ERK activation evoked by a reward-predicting CS differs markedly from that evoked by the reward itself, with the signal being stronger for a CS than the reward in the core and stronger for a reward than the CS that predicts it in the shell.

Fig. 5.

Presentation of food reward evokes a different subregional pattern of active ERK than does a CS predicting the reward. a Mean ± SEM fold change from CTRL level in the number of pERK1/2-IR cells in the NAc core for the REWARD group (n = 8), the PIT group (n = 8), and the PAV group (n = 9). Data for the PIT and PAV groups are based on those depicted in Fig. 4 and expressed relative to the mean count of pERK1/2-IR cells in the pooled CTRL group. The fold changes were significantly higher for the PIT and PAV groups than the REWARD group (*P < 0.05 and **P < 0.01, respectively), for which pERK-IR cells counts did not differ from CTRL level. b Representative photomicrograph of pERK-IR cells in the core of a rat from the REWARD group (scale bar see d; ac anterior commissure). c Similar data as shown in a are depicted for the NAc shell. The fold change was significantly higher for the REWARD group than either the PIT or the PAV group (**P < 0.01). d Representative photomicrograph of pERK-IR cells in the shell of a rat from the REWARD group (scale bar 40 μm)

To examine whether the increases in pERK immunolabeling upon exposure to a reward-predictive CS are specific to the ventral striatum or also occur in the dorsal striatum at the same anterior–posterior level as the NAc, we also counted the number of pERK-IR cells in the DMS and the DLS in a subset of the rats. We chose to distinguish between the DLS and DMS, as these subregions of the dorsal striatum were shown to play distinct roles in PIT (Corbit and Janak 2007, 2010). Compared to the NAc, the staining in the dorsal striatum was sparser. Within the dorsal striatum, the density of pERK-IR cells was higher in the DLS compared to the DMS (see Table 2). We first compared the number of pERK-IR cells in the DMS and the DLS between the TO (n = 4) and the TOY (n = 4) control groups. ANOVA revealed a significant effect of area (F_(1,6) = 7.79, P < 0.05), indicative that pERK labeling was higher in the DLS than the DMS in rats from both groups. However, neither the effect of group nor the group × area interaction was significant. Therefore, we pooled the results from the two control groups for comparison with the PIT (n = 6) and the PAV groups (n = 6), and conducted separate analyses for the DMS and the DLS. Different from the pattern we observed in the NAc, the number of pERK-IR cells in the dorsal striatum was comparable across groups (see Table 2), and ANOVA confirmed the lack of a significant group effect for both the DMS (F_(2,17) < 1) and the DLS (F_(2,17) = 1.45; P < 0.2). Taken together, these results suggest that CS-evoked increases in ERK signaling are restricted to the NAc and do not extend to the dorsal striatum, at least not within the anterior–posterior range examined here.

Discussion

Our examination of the subregional pattern of ERK activation in the NAc during PIT revealed that in both the core and the shell, ERK activation is increased about two- to threefold above control levels. The increase in ERK activation in both subregions was similar regardless of whether animals were presented with the CS only or also given the opportunity to display PIT. In fact, our data indicate that reward-seeking behavior itself does not contribute to ERK activation during PIT in either the core or the shell. Interestingly, exposure to the reward alone caused a pronounced increase in ERK activation in the shell while leaving ERK activation in the core equivalent to control levels.

We previously demonstrated that blockade of ERK activation in the NAc abolishes PIT (Shiflett et al. 2008). Our present observations of increased ERK activation in both the core and the shell suggest that both subregions are recruited during PIT. In support of this suggestion, lesions of the core were found to interfere with general PIT (Hall et al. 2001) and lesions of the shell with reward-specific PIT (Corbit et al. 2001; Shiflett and Balleine 2010; Corbit and Balleine 2011). General PIT refers to the potentiating effect of a CS on reward seeking regardless of whether the CS previously was paired with the same reward as the one for which the animal learned to emit a particular seeking behavior (e.g., lever press). On the other hand, reward-specific PIT refers to the potentiating effect of a CS when the situation affords multiple reward-seeking actions (e.g., pressing the left lever vs. the right lever), each of which leads to a different reward, and a CS enhances selectively the seeking behavior for the reward with which the CS previously was paired. Although the PIT paradigm we used here aligns more closely with paradigms for general PIT, the inclusion of an active and inactive lever in our paradigm may also have led to the recruitment of elements operative during reward-specific PIT. Indeed, dopamine receptor blockade in either the core or the shell was found to disrupt PIT using a paradigm similar to the one we used here (Lex and Hauber 2008).

Our findings of increased ERK activation in both the core and the shell during PIT suggest that inputs to and outputs from both subregions are part of the neural circuit that underlies PIT. The ventral tegmental area (VTA) provides dopaminergic input to both the core and the shell (Brog et al. 1993), and manipulations that disrupt VTA innervation to the NAc were found to abolish both forms of PIT (Murschall and Hauber 2006; Corbit et al. 2007). In light of evidence of positive coupling between dopamine D1 receptor activation and the ERK signaling cascade in NAc neurons (Valjent et al. 2004, 2005; Borgkvist et al. 2008; Fricks-Gleason and Marshall 2011), the pattern of PIT-associated ERK activation we described here may be mediated, at least in part, by input to both NAc subregions from the VTA. The basolateral amygdala (BLA) also has been implicated in PIT (Blundell et al. 2001; Corbit and Balleine 2005) and sends glutamatergic afferents to both subregions (Kelley et al. 1982; Brog et al. 1993; Groenewegen et al. 1999). Activation of N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors was shown to drive ERK signaling in striatal neurons (Perkinton et al. 1999; Mao et al. 2004; Haberny and Carr 2005; Valjent et al. 2005); the ERK activation pattern observed here may thus also be attributable to input from the BLA, potentially to the same cells as those innervated by the VTA (Johnson et al. 1994; Floresco et al. 2001; Valjent et al. 2005). Much less is known about the role of structures downstream of the NAc in PIT. Whereas the core projects primarily to the dorsolateral division of the ventral pallidum, the subthalamic nucleus, and the medial substantia nigra pars reticulata, the shell projects primarily to the ventromedial division of the ventral pallidum, the VTA, and the lateral hypothalamus (Heimer et al. 1991; Zahm and Brog 1992; Groenewegen et al. 1999; Tripathi et al. 2010). Accordingly, CS-evoked ERK activation in the core is likely to serve a different behavioral aspect of PIT than does CS-evoked ERK activation in the shell. For instance, in light of our findings of a marked increase in ERK activation in the shell upon exposure to the food reward, increased ERK activation in the shell during PIT may reflect a CS-evoked molecular representation of the reward. Consistent with this idea, reward-specific PIT, in which the CS is thought to trigger retrieval of specific sensory properties of the reward, relies on intact NAc shell functioning (Corbit et al. 2001; Corbit and Balleine 2011). Future studies will serve to pinpoint which of the various output targets critically participate in PIT, and the particular behavioral aspects mediated by the pERK immunopositive cells in the NAc projecting to these targets.

Our findings of comparable ERK activation in the NAc of rats from the PAV and PIT groups, with the increase above control levels being more robust in the core than the shell, suggest that exposure to the CS, and not the occurrence of reward-seeking behavior, is responsible for the increase in ERK activation during PIT. This conclusion is consistent with our findings of no difference in NAc ERK activation and no increase relative to handled controls between control rats that did (TO controls) and control rats that did not engage in reward seeking (TOY controls). Taken together with our previous findings that inhibition of NAc ERK activation blocks PIT but not reward-seeking per se (Shiflett et al. 2008), our present findings suggest that CS-evoked ERK activation in the NAc core and shell is necessary but not sufficient for CS potentiation of reward seeking. One possible mechanism through which CS-evoked ERK activation may enable enhanced reward seeking is through ERK action on voltage-gated channels that regulate NAc cell excitability (Tkatch et al. 2000; Yuan et al. 2002; Day 2008; Shiflett and Balleine 2011).

We found the pattern of increased ERK activation to vary differentially across NAc subregions depending on the type of stimulus to which animals were exposed. Presentation of reward caused no change in ERK activation in the core but a very pronounced increase in the shell; in contrast, presentation of the CS predictive of the reward caused a more reliable increase in the core than in the shell. Similar to our observations with a food reward, some investigators found that the psychostimulant cocaine causes a substantial increase in ERK activation predominately in the shell (Bertran-Gonzalez et al. 2008). However, others found that a range of psychostimulants and other addictive drugs greatly elevate ERK activation in both the core and the shell (Valjent et al. 2000, 2004; Corbille et al. 2007; Ibba et al. 2009). The difference between the subregional activation patterns we observed and the ones observed in these latter studies may reflect a difference in the magnitude of reward, or they may indicate that natural rewards and addictive drugs recruit different groups of NAc cells. In support of the latter explanation, different groups of NAc neurons were found to exhibit a change in firing rate during cocaine-reinforced responding versus food- or water-reinforced responding (Pennartz et al. 1994; Carelli 2002; Carelli et al. 2003; Wheeler and Carelli 2009). In agreement with our observations of a robust CS-evoked increase in ERK activation in the core, numerous studies support the involvement of the core in mediating the incentive salience of a CS (Di Ciano et al. 2001; Cardinal et al. 2002; Fuchs et al. 2004; Hollander and Carelli 2007). Furthermore, complementary to our observations of a preferential but not an exclusive CS-evoked increase in ERK activation in the core, neuronal firing has been reported as increased more frequently, but by no means exclusively, in the core than in the shell upon exposure to a CS (Day et al. 2006; Cacciapaglia et al. 2011; Saddoris et al. 2011). It is tempting to speculate that the NAc neurons whose firing rate is increased during presentation of a CS are the same neurons that exhibit a CS-evoked increase in ERK activation. Enhanced pERK immunoreactivity might, thus, serve as a useful marker to identify the phenotype of cells that respond to presentation of a CS in an excitatory fashion.

Various groups recently have begun to examine the involvement of the dorsal striatum in PIT (Corbit and Janak2007, 2010; Homayoun and Moghaddam 2009; Pielock et al. 2011). We previously found no evidence for CS-evoked ERK activation in the dorsal striatum (Shiflett et al. 2008), which suggested to us that the recruitment of ERK signaling during PIT is specific to the NAc and does not extend to the dorsal striatum, similar to the requirement of dopamine signaling in the NAc but not the dorsal striatum during PIT (Wyvell and Berridge 2000; Lex and Hauber 2008; Pielock et al. 2011). Our present findings of no significant effect of CS exposure on the ERK signal detected in either the DMS or the DLS confirm and extend our previous conclusion by showing that ERK signaling in the anterior medial and lateral striatum dorsal to the NAc is not differentially engaged upon exposure to a reward-predictive CS or during PIT. Similar to our findings in the NAc, we did not detect a change in pERK immunoreactivity in either the DMS or the DLS as a function of instrumental responding, as indicated by a lack of a significant difference between the TO and TOY groups, or between the PIT and PAV groups (see Table 2). In each case, the former of the two groups engaged in lever pressing during the test, whereas the latter did not. The lack of an effect of instrumental responding on ERK signaling in the dorsal striatum was surprising in light of recent observations that ERK activation is increased in both the DMS and the DLS after instrumental training, and that ERK inhibition in either of these striatal subregions interferes with the reduction of instrumental responding upon reward devaluation (Shiflett et al. 2010). The apparent discrepancy between observations may be attributable to differences in the anterior–posterior level at which ERK activation was examined: whereas we focused on the striatum dorsal to the NAc and thus the anterior dorsal striatum, Shiflett et al. examined and blocked ERK signaling in more posterior regions of the dorsal striatum. Additional procedural differences, such as the amount of instrumental training and the absence/presence of food reward during the test, may also have contributed to the difference in findings between studies. It should be noted that our observations of no change in ERK signaling in the anterior dorsal striatum during PIT do not preclude a specific role for either the DLS or the DMS in PIT (Corbit and Janak 2007, 2010); they merely suggest that the PIT-critical intracellular signaling steps in these subregions are unlikely to involve ERK.

In summary, we demonstrate that ERK signaling during PIT encompasses ensembles of neurons in both the core and the shell but not the anterior dorsal striatum, and that the CS is the principal stimulus responsible for the ERK activation pattern during PIT. Whereas the ERK signal in the core appears to reflect CS-specific properties, the ERK signal in the shell may reflect CS-evoked retrieval of reward-specific properties. CS-evoked ERK activation can serve as a useful molecular marker to study further the neuronal ensembles implicated in PIT, including their terminal regions and distinct roles in PIT.