Quantitative Mapping of Cocaine-Induced ΔFosB Expression in the Striatum of Male and Female Rats

ΔFosB plays a critical role in drug-induced long-term changes in the brain. In the current study, we evaluated locomotor activity in male and female rats treated with saline or cocaine for 2 weeks and quantitatively mapped ΔFosB expression in the dorsal striatum and nucleus accumbens of each animal by using an anti-FosB antibody that recognizes ΔFosB isoforms preferentially. Behavioral analysis showed that while there was little difference between males and females that sensitized to cocaine, nonsensitizing rats showed a large sex difference. Nonsensitizing males showed low behavioral activation in response to cocaine on the first day of treatment, and their activity remained low. In contrast, nonsensitizing females showed high activation on the first day of treatment and their activity remained high. Western blot and immunohistochemical analyses indicated that basal levels of ΔFosB were higher in the nucleus accumbens than the dorsal striatum, but that the effect of cocaine on ΔFosB was greater in the dorsal striatum. Immunostaining showed that the effect of cocaine in both the dorsal striatum and nucleus accumbens was primarily to increase the intensity of ΔFosB immunoreactivity in individual neurons, rather than to increase the number of cells that express ΔFosB. Detailed mapping of ΔFosB-labeled nuclei showed that basal ΔFosB levels were highest in the medial portion of the dorsal striatum and dorsomedial accumbens, particularly adjacent to the lateral ventricle, whereas the cocaine-induced increase in ΔFosB was most pronounced in the lateral dorsal striatum, where basal ΔFosB expression was lowest. Sex differences in ΔFosB expression were small and independent of cocaine treatment. We discuss implications of the sex difference in locomotor activation and regionally-specific ΔFosB induction by cocaine.


Introduction
Drug abuse gives rise to long-term changes in brain function. DFosB, a truncated and long-lasting form of the immediate-early gene FosB, has been studied extensively as a mediator of long-term structural and functional changes within the reward circuitry of the brain, especially in response to psychostimulants [1]. For example, overexpression of DFosB enhances sensitivity to psychostimulants, evidenced by heightened locomotor responses to cocaine [2] and decreased threshold for conditioned place preference [2] and self-administration [3].
DFosB accumulation has been most studied in the rodent striatum [4]. The striatum is generally divided into dorsal striatum (DS, caudate-putamen) and ventral nucleus accumbens (NAc). The nucleus accumbens is further divided into core and shell, based on chemorachitecture [5,6,7] and function [8]. The principal cells in the striatum are GABAergic medium spiny neurons (MSNs) [9]. Subtypes of MSNs are classified based on their projections and chemoarchitecture. Striatonigral MSNs express D1 dopamine receptors, Substance P and dynorphin (Dyn), whereas striatopallidal cells express mostly D2 receptors and enkephalin. These classes of MSNs are largely distinct in the dorsal striatum [10,11], but less so in the NAc [12]. DFosB is preferentially, but not exclusively, expressed in striatonigral projection cells [13,14]. In addition, DFosB accumulation varies subregionally with different drugs [15] and over time during the course of drug administration and withdrawal [13].
Behavioral manifestations of drug abuse are sexually dimorphic, and women may be more vulnerable to drug abuse [reviewed in 16,17,18]. In preclinical studies with rodents, females show greater locomotor responses [19,20,21,22], motivation to take drug [23,24], and preference for drugs [25] than males do. Sex differences in physiology [26,27] and anatomy [28] of MSNs, as well as in striatal dopamine responses to drugs [reviewed in 29] likely underlie these behavioral differences, although sex differences are not fully understood.
While drug-induced DFosB expression has been described [15,30,31], detailed quantitative investigation of drug-induced DFosB accumulation among striatal suregions in males versus females has been lacking. Psychostimulants are known to alter immediate-early gene expression in regionally specific manners [see 32 for review] and amphetamine-induced c-fos expression is sexually dimorphic and regionally specific [33]. Given the important role of DFosB in the effects of psychostimulants on synaptic input to MSNs [34], more detailed description of DFosB accumulation in both sexes could aid in determining how subregional variation in the neural effects of drugs relate to drug addiction. In the current study, we quantified the effects of repeated cocaine administration on locomotor behavior and FosB immunoreactivity in the DS and NAc of male and female rats. We used an antibody selective for DFosB and unbiased stereology to obtain anatomically detailed and quantitative measures of cocaineinduced DFosB accumulation.

Ethics Statement
All animal procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Northwestern University Institutional Animal Care and Use Committee (protocol 2009-0969). All efforts were made to minimize suffering of animals used for this study, and animals were euthanized with an overdose of sodium pentobarbital.

Animals
Gonadally intact young adult male and female Sprague-Dawley rats (Harlan, Indianapolis, IN), ,50 days old at the start of experiments, were used. Animals were housed in same-sex groups of 2 or 3 per cage and kept under a 12:12 light:dark cycle. Soy-free rat chow (Teklad #2916, Harlan, Indianapolis, IN) and water were available ad libitum. Estrous cycles in females were not monitored to avoid vaginal lavage-induced stress, since stress is known to induce DFosB accumulation in many brain areas [35].

Cocaine Treatment and Behavioral Monitoring
All behavioral testing was conducted during the mid-late light phase. On each treatment/testing day, rats were placed individually in 45 cm 645 cm 645 cm testing chambers, and allowed to acclimatize for 30 min. Each rat was then injected (i.p.) with saline or cocaine (15 mg/kg), immediately returned to the chamber, and allowed to move freely for 60 min after injection. Locomotor behavior was digitally recorded using LimeLight software (Actimetrics, Wilmette, IL) during the 30 min acclimation period and then for 60 min following injection. Data are expressed as mean6SEM cumulative distance traveled during the 60 min post-treatment period. Rats were treated for 2 weeks using a 5 days on 2 days off treatment regimen, and then sacrificed 24 hrs after their last injection to evaluate FosB protein levels by western blot or immunohistochemistry (IHC) for FosB immunoreactivity (FosB-IR) in the dorsal striatum (DS) and nucleus accumbens (NAc). Rats were kept in their regular housing facility until immediately before sacrifice.
Because 5G4 recognized primarily lower molecular weight bands (34-37 kD) characteristic of DFosB, 5G4 was used subsequently to evaluate DFosB protein levels in cocaine-or saline-treated male and female rats by western blot and IHC. For western blots, ImagePro Plus v.6.2 software (MediaCybernetics, Bethesda, MD) was used to measure optical density of bands in the 34-37 kD range, and this was normalized to actin. All samples were run in duplicate and the optical density values for each animal were averaged for statistical analysis.

FosB immunohistochemistry
For analysis of FosB IHC in the DS, we used 9 rats per group. For the NAc, an additional 6 rats per group were added for a total of 15 rats per group. Twenty four hrs after the last cocaine or saline injection, rats were deeply anesthetized with sodium pentobarbitol (80 mg/kg, i.p., Virbac Animal Health) and transcardially perfused with 4% paraformyladehyde in 0.1 M phosphate buffer (PB, pH 7.4). Brains were removed, post-fixed in the same fixative overnight at 4uC, cryoprotected in 30% sucrose in PB, and then 50 mm coronal sections through the NAc and DS were collected into 5 series. Brain sections were stored in 10% sucrose and 0.03% sodium azide in PB at 4uC until further processing.

Quantification of FosB-IR
Slides were coded prior to analysis. The proportions of all neuron-like cells, identified by cresyl violet staining, in the DS and the NAc containing FosB-IR nuclei were estimated using randomsystematic sampling and the optical fractionator probe [38,39].
Tissue was visualized with an Olympus BX60 microscope (Olympus America, Center Valley, PA) equipped with Stereo-Investigator 9 system (MicroBrightField, Williston, VT), including a CCD camera and a XYZ-motorized stage. The DS, NAc Core, and NAc Shell were delineated under a 4x objective and cells were counted under a 100x oil-immersion objective. Throughout the study, 30 mm630 mm68 mm counting frames were used. For the DS, labeled nuclei were counted random-systematically in counting frames distributed as 500 mm6500 mm grids in 3 sections (anterior-posterior (AP): +1.56-0.60 mm relative to Bregma), for approximately 208 counting frames (1.50610 6 mm 3 ) per animal. For the NAc, 350 mm 6350 mm grids were used in 4 sections (AP: +2.40 -1.20 mm relative to Bregma) for approximately 191 counting frames (1.38610 6 mm 3 ) per animal. The grid size was determined through a pilot study to achieve a stereologically reliable sampling density (Coefficient of Error,0.1) [40]. The intensity of staining in each labeled nucleus was categorized as either light, medium, or dark (see Fig. 3 for examples). Light staining was defined as light DAB reaction product clearly visible in the nucleus without obscuring the view of the nucleolus, medium staining was defined as DAB reaction product partially obscuring the view of the nucleolus, and dark staining was defined as DAB reaction product completely obscuring the view of the nucleolus. FosB-IR data are expressed as mean6SEM percent of all neuron-like cells. To verify the consistency of staining intensity classification, random-systematically sampled cells from a subset of animals in each group were analyzed for their mean optical density using ImageJ. There were no significant treatment-or sexrelated differences in measured optical density of nuclei within an intensity class, and each class of cells (light, medium, or dark) was separated by approximately 20 optical density units on 8-bit gray scale (0-256), on average.

Anatomical partitioning of dorsal striatum and nucleus accumbens
To evaluate subregional variation in FosB staining, the DS was divided into 4 zones, similar to the scheme used by Steiner and Gerfen ([41] Fig. 4C]. The DS ventral to the ventral limit of the lateral ventricle was classified as Ventral (V), and the remaining DS was divided into 3 zones of equal medial-lateral width for Medial (M), Intermediate (I), and Lateral (L) zones (Fig. 4C). Similarly, the NAc Core was divided into 4 quadrants and 2 anterior (A) and 2 posterior (P) sections collapsed for 8 final subdivisions (4 quadrants62 AP levels, Fig. 6A). As such, both the anterior (A) and posterior (P) NAc Core was divided into dorsomedial (DM), dorsolateral (DL), ventromedial (VM), and ventrolateral (VL) quadrants (Fig. 6A). The NAc Shell was divided into 2 halves and 2 anterior and 2 posterior sections collapsed for 4 final subdivisions (2 halves62 AP levels, Fig. 7A). Both the anterior and posterior Shell was divided into dorsomedial (DM) and ventrolateral (VL) halves (Fig. 7A). Each zone was named based on AP, dorsal-ventral (DV), and medial-lateral (ML) location within a structure, so that the dorsomedial quadrant in the anterior NAc Core was named ADM, for example (see Figs. 4C, 6A, and 7A for details).

Heat Maps
In order to display patterns of FosB-IR, heat maps were created from a subset of sections from the rats included in the analyses of Figure 1. Cocaine increased locomotor activity to a greater extent in females than in males. A) Representative locomotor activity traces during the initial 5 min following saline (Sal) or cocaine (Coc) injection on the last day (Day 10) of treatment in males (M) and females (F). Blue lines represent the movement of each animal during this period. B) The mean6SEM distance traveled during 60 min following cocaine or saline injection each day over the 2 week treatment period. Cocaine increased locomotion, especially in female rats. C) Distance traveled on the first day (D1) and the average of week 2 (Wk2) for cocaine-treated male (left) and female (right) rats. Data are displayed separately for sensitizing and nonsensitizing rats; connected points represent data for the same individual. Nonsensitizing males showed consistently low activity throughout treatment, whereas nonsensitizing females showed consistently high activity throughout treatment. *p,0.05. doi:10.1371/journal.pone.0021783.g001 both DS and NAc (9 animals per group). To create the heat maps, the xy-coordinate of each counted cell, as well as the coordinates of the lateral ventricles and the anterior commissure, were exported from StereoInvestigator and plotted in SigmaPlot 11 (Systat Software Inc. San Jose, CA). The plot of each section was sorted according to its rostrocaudal coordinates, aligned in Adobe Illustrator CS4 (Adobe Systems, Inc. San Jose, CA) using the lateral ventricle and the anterior commissure as references. A marker representing each counted cell was resized to roughly fill the counting grid that contained the respective counting frame. The markers were converted to gray scale based on their classification (dark: 60, medium: 30, light: 5, in inverted 8-bit gray scale). The generated vector graphics were converted to gray scale raster images, smoothed by applying Gaussian blur filter (sigma: DS: 30 px and NAc: 40 px), a color lookup table was applied in ImageJ (NIH, Bethesda, MD), and heat maps were superimposed on panels from the rat brain atlas [42]. Each map was generated from 12 hemispheres per rostrocaudal level, chosen on the basis of best fit with the respective panel of the brain atlas and representing 2/3 of all sections analyzed for FosB-IR.

Statistical Analysis
Locomotor Behavior: The 60 rats used for FosB-IHC were analyzed for locomotor behavior. Distance traveled on each of 10 injection days was analyzed with 3-way ANOVA (sex6drug6day). In addition, distance traveled on the first day of injection (D1), the average distance traveled in the 2 nd week (Wk2, days 6-10), and the ratio of Wk2 to D1 (Wk2/D1) were analyzed with 2-way ANOVAs (sex6drug). Cocaine-treated males and females were categorized as showing behavioral sensitization according to criteria established by Boudreau and Wolf [43]. A cocaine-treated animal was considered to show sensitization if its Wk2/D1 ratio exceeded the average C.V. for the saline animals. For additional analyses of locomotor sensitization, Wk2 and D1 data were compared between sensitizing and nonsensitizing rats with unpaired Student's t-tests. Significant interactions and main effects  in ANOVAs were further analyzed with appropriate lower order ANOVAs and Student's t-tests.
FosB-IR: For analysis of FosB protein levels with western blots, a 2-way ANOVA (sex6drug) was used, followed by unpaired Student's t-tests. For analyses of FosB-IR nuclei with IHC, we used 4-way mixed ANCOVAs (sex6drug6AP level6zone) for the DS and the NAc shell. For the NAc core, we used a 5-way mixed ANCOVA, where zones were substituted by dorsal-ventral and medial-lateral levels. Owing to the large number of animals in this study, we processed sections in two cohorts with an equal number of animals from each group. As such, cohort number was included as a covariate to account for any differences between cohorts. Significant interactions and main effects were further analyzed with appropriate lower-order ANCOVAs and/or Student's t-tests. Separately, the data from cocaine-treated animals were analyzed as above with the sensitization classification substituted for drug treatment as an independent variable.
Bonferroni correction was used for all multiple comparisons with t-tests. All statistical analyses were conducted with PASW Statistics 18 (SPSS, Inc., Chicago, IL), and p,0.05 was considered significant.

Cocaine-induced locomotor behavior in male and female rats
Overall Locomotor Behavior. The expected cocaineinduced increase in locomotion is apparent in Fig. 1A, which shows representative tracking data from each of the 4 groups studied. The blue lines represent locomotion during the first 5 min after saline or cocaine injection on the last treatment day.
Sex-specific Patterns of Locomotor Sensitization. Among the cocaine-treated rats, similar proportions of males (10 out of 15) and females (9 out of 15) showed locomotor sensitization to cocaine as defined previously [43]. Interestingly, however, while the pattern of locomotor activity in sensitizing rats was similar between males and females (D1: t 17 = 1.0, and Wk2: t 17 = 1.8, respectively, both ns; Fig. 1C), the behavior of nonsensitizing rats showed a significant sex difference. Specifically, nonsensitizing males showed low activity on D1 and their activity remained low, whereas nonsensitizing females started high on D1 and their activity remained high. On D1, nonsensitizing males responded to cocaine similarly as sensitizing males but then failed to increase their responses over time, resulting in much lower responses during Wk2 (sensitizing vs. nonsensitizing males: F 1,13 = 8.7, p = 0.011, Fig. 1C left). Nonsensitizing females, on the other hand, exhibited greater responses to cocaine on D1 than sensitizing females did (sensitizing vs. nonsensitizing females: F 1,13 = 13.8, p = 0.003, Fig. 1C right), and their responses remained high. The difference between nonsensitizing males and females on D1 was statistically significant (t 9 = 3.7, p = 0.005). Thus, the higher locomotor response seen in females as a whole (Fig. 1B) is largely attributable to the behavior of nonsensitizing, but not sensitizing, rats.

Characterization of FosB antibodies
We investigated 3 commonly used anti-FosB antisera to determine which best recognized DFosB isoforms. As shown in Fig. 2A, 5G4 showed the most robust labeling of DFosB (MW: 34-37 kD) with only a very weak band for the full length FosB (MW: ,48 kD). The sc-48 antibody also recognized DFosB, but labeling was weaker than with 5G4. The H-75 antibody recognized multiple higher molecular weight bands, including strong bands for the full length FosB. Particularly H-75 and 5G4 also recognized a smaller isoform (MW: ,21 kD). Based on its robust labeling of DFosB, we performed subsequent western blot and immunohistochemical studies with 5G4.

FosB protein levels in the Dorsal Striatum and Nucleus Accumbens
We used western blots as an initial approach to investigate cocaine effects on DFosB protein in the DS and NAc of males and females. As expected, cocaine treatment significantly increased DFosB protein in both regions. In the DS, cocaine increased DFosB levels by ,170% (Fig. 2B, F 1,8 = 8.2, p = 0.02). The effect of cocaine was slightly larger in females (208%) than males (133%), but there was no statistically significant sex difference (sex: F 1,8 = 0.6, sex6drug: F 1,8 = 0.4, both ns). In the NAc, cocaine increased DFosB levels to a lesser extent than in the DS, by ,80% (Fig. 2C, F 1,8 = 5.4, p = 0.049). As in the DS, the cocaine-induced increase in the NAc was slightly larger in females (97%) than in males (69%), but this was not a statistically significant difference (sex: F 1,8 = 0.3, sex6drug: F 1,8 = 0.3, both ns). Interestingly, based on the DFosB to actin ratio in both brain regions, basal DFosB levels in saline-treated animals were higher in the NAc than in the DS. This regional difference in basal DFosB levels and the greater effect of cocaine in the DS, were corroborated by subsequent immunohistochemical staining (see below). Fig. 3 shows representative low magnification photomicrographs of brain sections from saline- (Fig. 3A) and cocaine-treated (Fig. 3B) rats stained for FosB using 5G4. Examples of FosBnegative cresyl violet-stained cells and FosB-positive cells with light, medium, or darkly stained nuclei are shown in Fig. 3C. From images like these, we quantified and mapped FosB-IR in the DS and NAc as the proportion of all neuron-like cells containing FosB-IR vs. non-IR nuclei and categorized each labeled nucleus as light, medium or dark. Overall, this analysis showed that the major effect of cocaine was to increase the intensity of FosB immunostaining, with relatively small effects on FosB-IR cell number. FosB-IR in the Dorsal Striatum Subregions. To investigate subregional differences in FosB labeling, we divided the DS into 4 zones: 3 dorsal and 1 ventral zone (Fig. 4C). This revealed significant differences between zones and a zone6drug interaction in all classes of FosB-IR cells (Fig. 4B). Statistically significant differences are summarized in Fig. 4D, with ''D'' indicating a drug effect and ''S'' indicating a sex difference for each class of cell in each zone. Post-hoc analyses revealed mediallateral gradients in FosB-IR and in cocaine's effects on FosB-IR. In general, the Medial, Intermediate, and Ventral zones contained more FosB-IR cells than the Lateral zone, and the effects of cocaine were greatest in the Lateral and Intermediate zones (Fig. 4B).

FosB-IR in the Dorsal Striatum
Specifically, in the Medial zone, the relative proportions of heavily stained (medium+dark) FosB-IR cells increased (+209%) with cocaine-treatment, and the proportion of lightly-stained FosB-IR cells decreased (220%). As a result, the overall proportion of cells that were FosB-IR was unchanged in this zone. In contrast, in the Lateral zone, the proportions of all categories, light (+45%), medium (+293%), and dark (+2,262%) cells, increased with cocaine treatment. Consequently, the overall Figure 5. Heat maps of FosB-IR in the dorsal striatum. Numbers in the center indicate AP levels relative to Bregma. The cocaine-induced increase in FosB-IR is apparent at all levels in both male (left) and female (right) rats. Greater FosB-IR was seen in the medial, relative to the lateral, part of the dorsal striatum. Females exhibited higher FosB-IR than males, which is more apparent in the saline-treated rats. In addition, the cocaine-induced FosB-IR increase was larger in the more caudal (1.08 and 0.60) sections than in the rostral (1.56) section. Coc: cocaine, and Sal: saline. doi:10.1371/journal.pone.0021783.g005 proportion of FosB-IR cells in the Lateral zone was significantly higher in cocaine-treated rats (saline: 12.360.9% vs. cocaine: 19.461.3%, F 1,31 = 23.3, p,0.001). In the Intermediate zone, the pattern of FosB-IR combined features of the Medial and the Lateral zones. As was the case in the Medial and the Lateral zones, the proportions of heavily stained cells were significantly higher in cocaine-treated rats (medium: +517%, dark: +7,582%, relative to saline-controls). However, the proportion of lightly stained cells was unchanged. As a result, the total proportion of all cells that were FosB-IR was only modestly higher in the cocaine-treated rats. The pattern of FosB-IR and cocaine-induced changes in FosB-IR in the Ventral zone was similar to the Medial zone; the proportion of darkly stained cells increased (+410%) significantly with cocaine treatment, but the total proportion of all cells containing FosB-IR was unchanged.
In addition to these subregional differences in FosB-IR, we observed an anterior-posterior level x drug interaction in the proportion of heavily stained (medium+dark) cells (F 2,62 = 5.5, Sex Differences in the Dorsal Striatum Subregions. The proportion of neuron-like cells containing FosB-IR in the DS was slightly higher overall in females than in males (Fig. 4A, males: 24.761.1% vs. females: 27.461.0%, F 1,31 = 4.7, p = 0.04). Broken down by staining intensity, this trend was observed in all 3 classes (light, medium, and dark) of FosB-IR cells, but statistically significant only in the medium stained cells (males: 2.360.3% vs. females: 3.260.4%, F 1,31 = 10.1, p = 0.003). These sex differences were much smaller than cocaine-induced changes, however. For example, cocaine increased the proportion of darkly-stained cells by ,1,000% overall compared to only a ,20% difference between sexes. No significant sex differences were observed in subregional patterns of FosB-IR.
Dorsal Striatum Heat Maps. Most of the quantitative differences described above are visible in heat maps of FosB-IR in the DS (Fig. 5), which summarize patterns of labeling in each group. The significantly higher FosB-IR in cocaine-treated rats and the medial-lateral gradient in FosB-IR are visible in the heat maps. While not as prominent as cocaine-saline or medial-lateral differences, higher FosB-IR in females, especially in saline-treated rats, is also evident. Finally, the greater effect of cocaine on FosB-IR in the posterior sections is evident in the heat maps.
FosB and Locomotor Sensitization: Dorsal Striatum. Comparison of FosB-IR in sensitizing vs. nonsensitizing cocaine-treated rats showed only minor differences. Specifically, the proportion of lightly stained cells differed depending on sex and sensitization classification (sex6sensitization: F 1,13 = 6.5, p = 0.02). Among nonsensitizing rats, the proportion of lightly stained cells was higher in females than males (males: 17

FosB-IR in the Nucleus Accumbens Core
Overall, proportion of cells that were FosB-IR was higher in the NAc Core (30-50%) than the DS (15-35%). In contrast to the clear effects of cocaine in the DS, however, the effects of cocaine on FosB-IR were smaller in the NAc, including the Core, and were limited to darkly stained cells (Fig. 6A). The proportion of darkly stained cells was 93% higher in cocaine-treated rats than in the saline-treated rats (saline: 2.060.2% vs. cocaine: 3.860.4%, F 1,55 = 25.6, p,0.001). There were no drug-induced changes in light or medium stained cells, nor in the proportion of all neuronlike cells that contained FosB-IR. All statistically significant differences in the NAc Core are indicated in the tables in Fig. 6.
FosB-IR in the Nucleus Accumbens Core Subregions. The zones used for subregional analyses within the NAc Core are shown in Fig 6A (left: anterior NAc, and right: posterior NAc). The proportions of all neuron-like cells containing FosB-IR in each zone are shown in Fig. 6B (anterior NAc Core) and Fig. 6C (posterior NAc Core). Statistically significant differences in FosB-IR within individual zones of the NAc Core are summarized in Fig. 6B (anterior) and Fig. 6C (posterior). Subregional analyses revealed that, in both the anterior and the posterior NAc Core, cocaine increased the proportion of darkly stained cells in all zones, except for anterior ventrolateral and posterior ventromedial (p = 0.052 and p = 0.15, respectively).
As observed in the DS, we found a significant medial-lateral and anterior-posterior gradients in FosB-IR the NAc Core. The proportion of darkly stained cells was higher in the medial than lateral NAc Core (medial: 3.360.3% vs. lateral: 1.960.2%, F 1,55 = 7.0, p = 0.011), and the proportion of darkly stained cells was higher in the posterior than anterior NAc Core (anterior: 2.160.2% vs. posterior: 3.260.2%, F 3,165 = 6.3, p,0.001). In addition, there was a dorsal-ventral gradient; the proportion of medium stained cells was higher in the dorsal NAc Core than ventral NAc Core (dorsal: 3.260.3% vs. ventral: 2.160.3%, F 1,55 = 4.43, p = 0.04).
Sex Differences in Nucleus Accumbens Core FosB-IR. Whereas cocaine-induced differences in FosB-IR were found in the proportions of darkly stained cells, sex differences were found in the proportions of light and medium stained cells, and in the dorsal but not ventral NAc Core (see tables in Fig.6B and 6C). Unlike the DS where females exhibited higher FosB-IR, males exhibited slightly higher FosB-IR in the NAc Core overall, but this was not statistically significant (males: 41.761.5% vs. females: 39.061.3%, F 1,55 = 2.0, ns). Considering zones separately, we saw significantly higher proportions of medium (in the anterior dorsolateral, F 1,55 = 6.6, p = 0.01) and light (in the posterior dorsomedial, F 1,55 = 4.1, p = 0.048) cells in males than in females. Furthermore, the total proportion of FosB-IR cells was higher in males than in females in anterior dorsal medial zone (F 1,55 = 4.7, p = 0.03). These sex differences were quite small in magnitude, however, and there were no sex6drug interactions among any of the zones in the Core.

Locomotor Sensitization and FosB: Nucleus Accumbens
Core. Comparison of FosB-IR in the Core of sensitizing vs. nonensitizing cocaine-treated rats showed a 3-way interaction in the proportion of darkly stained cells (DV x ML x sensitization: F 1,25 5.2, p0.03), but no post-hoc test was statistically significant.

FosB-IR in the Nucleus Accumbens Shell
The pattern of FosB-IR in the NAc Shell was similar to that in the NAc Core. Namely, there was a cocaine-induced increase in darkly-stained cells (saline: 1.660.3% vs. cocaine: 2.660.3%, Fig. 7A) and the magnitude of this increase was smaller than in the DS. The proportions of medium or lightly stained cells were not significantly different between groups. All statistically significant differences in the NAc Shell are indicated in the tables in Fig. 7.
FosB-IR in the Nucleus Accumbens Shell Subregions. The zones used for subregional analyses are shown in Fig. 7A (left: anterior NAc; right: posterior NAc). The proportions of FosB-IR in each zone are shown in Fig. 7B (anterior NAc Shell) and Fig. 7C (posterior NAc Shell). As was the case in NAc Core, cocaineinduced increases in the proportions of darkly stained cells were found in most zones, but in the shell, these were statistically significant only in the anterior sections (Fig. 7B).
Sex Differences in Nucleus Accumbens Shell FosB-IR. As in the NAc Core, males exhibited slightly higher FosB-IR in the NAc shell than females did. This sex difference was significant only in medium stained cells and was apparent only in the posterior sections (Fig. 7C). Males showed significantly higher proportions of medium stained cells in the posterior ventrolateral zone (F 1,55 8.6, p0.005) and a strong trend in the posterior dorsomedial zone (p0.059). In the NAc Shell as a whole, males exhibited higher proportions of FosB-IR cells overall ( Fig. 7A; medium, dark, total FosB), but the difference was not statistically significant (medium: F 1,55 = 0.9, dark: F 1,55 = 0.9, total FosB: F 1,55 = 0.8, all ns).

Locomotor Sensitization and FosB: Nucleus Accumbens
Shell. Similar to analysis in the Core, comparing FosB-IR the Shell of in sensitizing vs. nonsensitizing rats showed a 4-way interaction (AP x zone x sex x sensitization: dark: F 3,75 = 15.2, p = 0.04 and total FosB: F 3,75 = 209.9, p = 0.02), but no post-hoc test was statistically significant.
Nucleus Accumbens Heat Maps. Heat maps displaying FosB-IR in the NAc of saline-vs. cocaine-treated rats (Fig. 8) show few differences overall, consistent with the relatively small effects of cocaine and sex differences in FosB-IR in the NAc. Nonetheless, medial-lateral and dorsal-ventral gradients in FosB-IR are visible. The heat maps also show subregional variation in FosB-IR, with Numbers in the center indicate AP levels relative to Bregma. Unlike the dorsal striatum, the cocaine-induced increase in FosB-IR is relatively small for both males (left) and females (right). Particularly high FosB-IR near the lateral ventricle is evident in most maps. Males exhibited higher FosB-IR, which is more apparent in the saline-treated rats. In addition, medial-lateral, as well as dorsal-ventral gradients in FosB-IR are visible at most levels. Coc: cocaine, and Sal: saline. doi:10.1371/journal.pone.0021783.g008 the greatest proportion of FosB-IR cells in the dorsomedial core, adjacent to the lateral ventricle. Additionally, the generally higher level of FosB-IR in the NAc compared to the DS is apparent, particularly in the controls (compare Fig. 5 and Fig. 8).

Discussion
Repeated exposure to psychostimulants causes long-term changes in the brain. Previous studies have demonstrated the importance of DFosB in these changes [reviewed in 1]. In the current study, we provide a detailed quantitative analysis of cocaine-induced DFosB accumulation in subregions of the dorsal striatum and nucleus accumbens in behaviorally characterized male and female rats. As others have found [44,45], females showed significantly greater locomotor activity in response to cocaine than males. Interestingly, however, breaking down the data by sensitizing and nonsensitizing rats showed that this was due largely to sex differences among nonsensitizing rats. Nonsensitizing males showed a low initial response to cocaine that stayed low throughout treatment, whereas nonsensitizing females showed a high initial response that stayed high. The behavior of sensitizing males and females was quite similar except that maximal locomotor activity was higher in females.
We identified 5G4 anti-FosB as being slightly better than sc-48 and much better than H-75 for recognizing DFosB isoforms. Using 5G4, we found that cocaine clearly increased total DFosB levels in the DS and NAc, as assessed by western blot. The cocaine-induced increase was greater in the DS than in the NAc, as previously shown [46]. Mapping DFosB positive cells showed that basal DFosB levels were highest in the medial DS and the dorsomedial NAc, areas adjacent to the lateral ventricles, but that the largest cocaine-induced changes were in the lateral DS, where basal DFosB levels were low. Overall, DFosB was expressed in up to half of the neurons in the striatum (10-20% in low areas vs. ,50% in high areas), consistent with the proportion of striatonigral cells in which DFosB is preferentially expressed [13,14]. The effect of cocaine was primarily to increase the intensity of DFosB immunostaining rather than to increase the proportion of cells with DFosB positive nuclei. Namely, the overall proportions of cells containing any FosB-IR did not change in most areas, while the proportion of heavily stained cells increased with cocainetreatment. Sex differences in DFosB accumulation were small and regionally specific; DS levels were higher in females whereas NAc levels were higher in males. Cocaine-induced increases in DFosB and these small sex differences were additive, as no interactions were observed.

Sex differences in psychostimulant-induced locomotion
One interesting finding from our behavioral analysis was that the wide range of initial responses to cocaine in females was closely linked to subsequent development of sensitization. Understanding the factors that lead to locomotor sensitization is important because there is evidence that the changes underlying locomotor sensitization also mediate potentiation of drug taking following prior drug exposure [see 47 for review]. The multiple drug treatments typically used to assess sensitization are known to disrupt the estrous cycle [reviewed in 48] and likely cause brain changes on their own, masking pre-existing differences. Thus being able to differentiate between females that would go on to show sensitization (or not) using a single drug treatment could be useful to identify pre-existing differences that predict sensitization or the lack of sensitization.
One factor uncontrolled in the current study that may have significant influence is the hormonal status of females, as estrogens have been shown to enhance the locomotor-activating effects of psychostimulants [reviewed in 16,17,29] and the rate of sensitization in females [49]. However, it is unlikely that the wider range of initial responses in females that we saw was due solely to their estrous phase, as Walker et al., [45] have shown that even ovariectomized females show a much wider range of initial responses than males. In addition, that sensitization was similar between males and females in our study argues against a major role for estrogens in modulating sensitization of locomotor behavior. Others have also suggested that individual differences in vulnerability to sensitization and hormonal influences are independent and can be dissociated [19,50]. It remains possible, however, that hormonal modulation of behavioral responses to psychostimulants depends on what type of behavior is assayed (e.g., locomotor behavior vs. rotational behavior).
In contrast to the wide range of behavioral responses observed upon the initial injection in females, males showed a wider spread of behavior during the second week. This pattern is reminiscent of the bimodal distribution of behavior and associated differences in dopaminergic activity observed only in males following repeated psychostimulant injections [19]. Thus, perhaps females have a wider range of pre-existing differences that affect initial responses to psychostimulants, while males have more diverse psychostimulant-induced responses to repeated treatment.
While there were significant differences in DFosB accumulation between saline-and cocaine-treated animals, we found no clear relationship between locomotor sensitization and DFosB accumulation in the DS or NAc within the animals treated with cocaine. This observation is consistent with Kelz, et al., [2] in which overexpression of DFosB potentiated cocaine-induced locomotion but did not affect locomotor sensitization. It is possible that the design of the current study, with a single cocaine dose, did not produce a sufficient range of behavior or DFosB accumulation to detect such a relationship. Alternatively, it could be that accumulation of DFosB beyond a certain threshold level is necessary to trigger changes associated with locomotor sensitization. Because there likely are multiple steps between DFosB accumulation and behavioral output, other intervening factors may be more closely associated with locomotor sensitization. For example, we have recently reported a correlation between locomotor behavior and MSN spine density in the NAc [51].

Sex differences in DFosB accumulation
We found that sex differences in DFosB accumulation are small and largely independent of cocaine-treatment, likely reflecting preexisting differences rather than differential responses to cocaine. In the DS, sex differences were consistent with the behavioral differences: females showed greater locomotor activation and higher DFosB levels than males. In contrast, in the NAc, males showed higher DFosB levels. Within the subregions of the DS and NAc, sex differences were generally small and non-significant in most cases. The sex differences in DFosB levels independent of cocaine-treatment are consistent with sex differences in locomotion upon the initial exposure to psychostimulants in the current and previous studies [22,44,45].

Subregional variation in DFosB expression
The NAc is strongly implicated in locomotion, while the DS is more closely associated with stereotypy [52]. As previously mentioned, we saw no clear relationship between locomotor sensitization and DFosB in either region, at least within cocaine treated animals. Further studies are required to determine if individual variations in cocaine-induced DFosB accumulation in specific striatal subregions are associated with other cocaine-induced behaviors, including stereotypy. Additionally, cocaine was experimenter administered in this study, thus it is unknown how regional specificity of DFosB may be influenced by self-administered cocaine. Nonetheless, the regionally specific patterns of DFosB accumulation that we observed parallel known differences in afferents to striatal subregions [53]. Because DFosB expression in MSNs requires both dopaminergic and glutamatergic stimulation [2,31,54,55], differences in one or both types of input could correspond to variation in basal DFosB levels and accumulation in response to cocaine. In turn, such differences in connectivity are related to the involvement of striatal subregions in different aspects of behavior.
Striatal regions with high basal DFosB-IR, the NAc Core and the medial DS, receive dopaminergic inputs mainly from the lateral ventral tegmental area and the medial substantia nigra pars compacta (SNc) [56, reviewed in 57]. Glutamatergic inputs come from medial prefrontal cortex (prelimbic and infralimbic areas) [58], subiculum [59], basolateral amygdala [60,61], as well as midline and medial intralaminar thalamic nuclei [62]. In contrast, the lateral DS, which had low basal DFosB-IR but showed the greatest effects of cocaine, receives dopaminergic inputs from the lateral SNc [56, reviewed in 57] and glutamatergic inputs from somatosensory cortex [58] and lateral intralaminar thalamic nuclei [62]. The medial DS has been implicated in goal-directed behaviors, while the lateral DS is involved in habit formation [see 63 for review,64].
In the context of drug abuse, functional differences between striatal subregions have been linked to acquisition of operant selfadministration (medial DS) and its eventual transition to habitual responding that is resistant to reward devaluation (lateral DS) [65]. Effects in the goal-directed circuit observed in the current study likely contribute to the potentiation of drug taking by prior drug exposure [reviewed in 47]. The even larger changes we found in the habit circuit suggest that the potentiation of habit formation by prior exposure to drug [see 66 for example] may be more pronounced than the potentiation of drug taking, which could facilitate transition to habitual drug taking.

DFosB, MAPK, and structural plasticity of MSNs
DFosB is closely linked to psychostimulant-induced changes in MSN dendritic spine density [14], which is thought to reflect rewiring of reward circuitry related to addiction. Psychostimulant-induced structural plasticity of MSN dendrites also requires activation of the MAPK signaling pathway [67,68,69] and MAPK is important in the development of behavioral sensitization to psychostimulants [70,71]. Indeed, the pattern of DFosB accumulation we observed corresponds closely with other studies of cocaine-induced MAPK activation [30,70], as demonstrated by ERK phosphorylation.
Given the relationship between DFosB and downstream consequences of psychostimulant exposure, striatal subregions with more pronounced cocaine-induced DFosB accumulation might be expected to show more robust drug-induced structural changes. Interestingly, Jedynak, et al. [72] found that repeated methamphetamine injections increase MSN spine density in the lateral DS, where we saw the greatest cocaine effect on DFosB, while decreasing spine density in the medial DS, where we saw the smallest effect on DFosB. Thus our findings may reflect subregional variation in the underlying processes leading to psychostimulant-induced rewiring of neural circuits.

Conclusion
Our results suggest that sex differences in psychostimulantinduced locomotion and DFosB accumulation are more likely due to pre-existing differences between males and females rather than to sex-specific responses to repeated psychostimulant exposure. Additionally, our behavioral analyses indicate that, especially in females, the locomotor response upon initial exposure to psychostimulants may be useful as a proxy for subsequent development of sensitization. It has been argued that locomotor sensitization shares underlying mechanisms with other behavioral changes related to drug addiction [47,73]. As such, responses to an initial cocaine exposure may help to identify preexisting individual differences associated with susceptibility to addiction.
The regionally specific patterns of DFosB accumulation observed in this study add to the existing literature on the heterogeneous nature of striatal responses to psychostimulants. The medial and the lateral DS exhibited a striking difference in DFosB accumulation, which is consistent with functional and anatomical differences between these subregions. Thus, distinguishing between medial and lateral DS in future studies, rather than treating them as a single structure, should benefit further elucidation of the how the DS is involved in drug abuse, especially in the transition to habitual drug taking.

Author Contributions
Conceived and designed the experiments: CSW. Performed the experiments: SMS AFM AMW. Analyzed the data: SMS AFM AMW CSW. Wrote the paper: SMS CSW.