Enhanced functional connectivity involving the ventromedial hypothalamus following methamphetamine exposure

Methamphetamine (MA) consumption causes disruption of many biological rhythms including the sleep-wake cycle. This circadian effect is seen shortly following MA exposure and later in life following developmental MA exposure. MA phase shifts, entrains the circadian clock and can also alter the entraining effect of light by currently unknown mechanisms. We analyzed and compared immunoreactivity of the immediate early gene c-Fos, a marker of neuronal activity, to assess neuronal activation 2 h following MA exposure in the light and dark phases. We used network analyses of correlation patterns derived from global brain immunoreactivity patterns of c-Fos, to infer functional connectivity between brain regions. There were five distinct patterns of neuronal activation. In several brain areas, neuronal activation following exposure to MA was stronger in the light than the dark phase, highlighting the importance of considering circadian periods of increased effects of MA in defining experimental conditions and understanding the mechanisms underlying detrimental effects of MA exposure to brain function. Functional connectivity between the ventromedial hypothalamus (VMH) and other brain areas, including the paraventricular nucleus of the hypothalamus and basolateral and medial amygdala, was enhanced following MA exposure, suggesting a role for the VMH in the effects of MA on the brain.


Introduction
The abuse of the psychostimulant methamphetamine (MA) causes great physiological (Wolkoff, 1997), social (Sommers et al., 2006), and financial damage (Dobkin and Nicosia, 2009). It is estimated to cost the United States approximately $23.4 billion annually (Dobkin and Nicosia, 2009). In the United States, MA was abused by approximately 439,000 people in 2011, 133,000 of which were new users (SAMHSA 2012).
MA consumption by humans is characterized by disruption of many biological rhythms including the sleep-wake cycle (Hasler et al., 2012). This circadian effect is not only seen relatively shortly following MA exposure (Honma et al., 1986(Honma et al., , 1991Uchihashi et al., 1994;Masubuchi et al., 2001;Olsen et al., 2013), but also later in life following developmental MA exposure (Eastwood et al., 2012). MA phase shifts, entrains the circadian clock, and can also alter the entraining effect of light by currently unknown mechanisms (Honma et al., 1986). The circadian rhythms of mice or rats made arrhythmic by lesions of the suprachiasmatic nucleus (SCN) can be regenerated by consumption of MA dissolved in the drinking water (Honma et al., 1986;Tataroglu et al., 2006;Mohawk et al., 2009). These reconstituted rhythms are robust and show periods significantly longer than the wild-type rhythms. These observations led to the hypothesis that a separate MethAmphetamine-Sensitive Circadian Oscillator (MASCO) exists in the brain separate from the SCN (Tataroglu et al., 2006). The MASCO is not composed of the canonical clock genes since animals made arrhythmic by knocking out these clock genes continue to demonstrate robust MA-induced rhythms (Masubuchi et al., 2001;Mohawk et al., 2009). To date, the location of this oscillator and its molecular make up remain completely unknown (Tataroglu et al., 2006). Although the SCN might play a role in MASCO, it does not seem critical as the effects of MA on circadian system are also seen in animals in which the SCN is lesioned (Tataroglu et al., 2006).
The ventromedial hypothalamus (VMH) might play a role in the circadian effects of MA. The VMH is involved in synchronization of rhythms to periodic feeding but, unlike the SCN, does not contain a self-sustained oscillator associated with food (Inouye, 1983). The VMH is also important for circadian rhythms in serum insulin, glucose, and triglycerides (Egawa et al., 1993). Disturbed circadian feeding rhythmicity and inhibition of the glucose preference reversal are observed in rats with VMH lesions, a functional state similar to that seen following chronic MA exposure (Kraeuchi et al., 1985;Mitsushima et al., 1994). The VMH might be involved in other circadian rhythms as well. Implantation of a hydro-polymer gel into the VMH to alter its neural network, likely involving neural output via a dorsal route, showed attenuation of circadian activity levels in blind female rats and of circadian changes in serum melatonin levels (Mitsushima et al., 1994).
The adrenocorticotropic hormone (ACTH) secretagogue arginine vasopressin (AVP) (Dornhorst et al., 1981;Antoni, 1986) plays a role in circadian effects. Intrahypothalamic injection of an AVP V1 receptor blocker reveals the diurnal stimulatory input of AVP to Hypothalamus-Pituitary-Adrenal (HPA) axis activation (Kalsbeek et al., 1996). Release of AVP in the paraventricular nucleus of the hypothalamus (PVN) is also seen following emotional stress (Wotjak et al., 1996), local inflammation (Turnbull and Rivier, 1996), and inflammatory cytokines (Raber et al., 1998). The PVN showed profound activation by MA and an increase in the number of c-Fos/AVP dual-labeled cells during the light phase (Zuloaga et al., 2014). It is unclear whether these effects are restricted to the light phase or seen in the dark phase as well.
In most studies, there is a bias to study a particular brain region or network that is pertinent to a specific behavioral or cognitive change seen following MA exposure or other drugs of abuse (Abekawa et al., 1996;Broening et al., 1997;Chang et al., 2002Chang et al., , 2004Chang et al., , 2009Brady et al., 2003;Koob and Volkov, 2010). In the current study, we analyzed and compared immunoreactivity of the immediate early gene c-Fos, a marker of neuronal activity (French and Pavlidis, 2011), to assess neuronal activation 2 h following MA exposure in the light and dark phases. We also compared co-localization of c-Fos with AVP in the PVN in the light and dark phases. Finally, we used network analyses of functional connectivity based on global brain immunoreactivity patterns of c-Fos, to infer functional connectivity between brain regions (Schiltz et al., 2007). We hypothesize that a brain region showing a neuronal activation pattern in the light and dark phases that resembles that of the SCN and shows a functional neuronal connectivity that is enhanced following MA exposure might suggest its involvement in MASCO.

Animals
Forty male C57BL6/J mice, 28-42 days old, were purchased from JAX Laboratories (Bar Harbor, Maine) and maintained on a 12/12 light/dark cycle with lights on at 5:00 am (Zeitgeber Time 0, ZT0). Mice were singly housed for 4 days prior to experiments and were between 50 and 70 days old at the time of testing. Rodent chow (PicoLab Rodent Diet 20, #5053; PMI Nutrition International, St. Louis, MO) and water were available ad libitum.

Animal Approval
All procedures complied with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and with IACUC approval at Oregon Health & Sciences University.

MA Treatment and c-Fos Immunohistochemistry
The mice were injected with (d)-MA hydrochloride dissolved in saline (1 mg/kg, ip.) or saline alone between 3 and 4 h after lights on (ZT3-4) or 3-4 h after lights off (ZT15-16). Two hours later, the mice were intracardially perfused with 20 ml of phosphate buffered saline (PBS), followed by 40 ml of a 4% paraformaldehyde solution. Brains were removed, stored in 4% paraformaldehyde overnight, and then transferred to a 30% sucrose solution. The 2 h time point was chosen based on our previous study that demonstrated extensive c-Fos induction in the mouse brain following administration of the same dose of MA (Zuloaga et al., 2014). The mice were injected and perfused in separate cohorts so that all mice in a cohort were perfused within 82 min. Fixed brains were coronally sectioned at 40 µm into three series using a cryostat (Microm HM505E, MICROM international GmbH, Walldorf, Germany) and processed for immunohistochemical detection of c-Fos. Sections were rinsed in PBS, incubated in 1% hydrogen peroxide and 0.3% Triton-X in PBS (PBS-TX) for 10 min, again rinsed in PBS, then incubated in 10% normal goat serum (NGS) in PBS-TX for 1 h. After rinsing in PBS, sections were incubated in primary antisera (c-Fos rabbit polyclonal: 1:5000, Santa Cruz Biotechnology, sc52) in 4% NGS and PBS-TX overnight at room temperature. Sections were then rinsed in PBS and incubated for 1 h in biotinylated goat-anti rabbit antibody in PBS-TX (1:500, Vector Laboratories, Burlingame, CA), followed by rinses in PBS and a 1 h incubation in avidin-biotin peroxidase complex (ABC Elite kit, Vector Laboratories, Burlingame, CA). Following rinses in Tris buffered saline (TBS), sections were developed for visualization of c-Fos positive cells in a hydrogen peroxide/diaminobenzidine/TBS solution for 10 min, after which sections were rinsed in PBS and immediately mounted on slides. The following day, sections were dehydrated in ethanol, defatted in xylene, and coverslipped with Permount (Sigma Chemical Co., St. Louis MO).

Dual Label Immunohistochemistry
To determine the co-localization of c-Fos and AVP immunoreactivity in the mouse brain, we performed dual label immunohistochemistry. Free-floating sections were rinsed with PBS 3 times, then blocked with 4% donkey serum in PBS-TX for 90 min. Sections were then incubated in anti-c-Fos (1:250, goat, Santa Cruz Biotechnology, sc52-G, Billerica, MA) overnight. Sections were then incubated in 1:200 donkey antigoat Dylight 594 (Abcam, Cambridge, MA, USA) for 3 h at room temperature. Sections were then rinsed in PBS 4 times (20 min each rinse) after which the same protocol was repeated using AVP (1:500, rabbit, gift from Dr. Paul Plotsky) as primary antibody and 1:200 donkey anti-rabbit Alexa 488 (Life Technologies) as the secondary antibody. Sections were slide mounted and coverslipped with antifade reagent to preserve fluorescent signal (Vectashield with DAPI, Vector), light protected, and stored at 4 • C.

Microscopy
Quantification of c-Fos positive cells was performed using an Olympus IX81 microscope (Olympus, Center Valley, PA, USA) equipped with Slidebook software (Intelligent Imaging Innovations, Inc., Denver, CO, USA). Images of discrete brain regions were captured bilaterally within two sections using a 10× objective. These regions were mainly selected based on our previous study showing robust neuronal activation in these areas following MA exposure during the light phase (Zuloaga et al., 2014). Regions were identified using the mouse brain atlas (Franklin and Paxinos, 1997 For quantification of c-Fos/AVP dual labeling, confocal images of the PVN were captured bilaterally at 20× in two sections using the same Olympus IX81 confocal microscope. c-Fos/AVP dual-labeled cells were quantified within two regions of the PVN, the lateral magnocellular (circle, 135 µm diameter) and the medial parvocellular regions (circle, 135 µm diameter). Cells were considered co-localized when they expressed both red (c-Fos) and green (AVP) fluorescence. For AVP/c-fos coexpressing cells AVP was found within the extra-nuclear area, while c-fos was localized to the nucleus, resulting in a green cell body that surrounds a red nucleus. The number of c-Fos/AVP expressing cells in the magnocellular and parvocellular regions were counted within each brain and the data are expressed as the mean number of c-Fos/AVP positive cells per experimental group.

Functional Connectivity Analysis
Using the c-Fos immunohistochemical data (single labeling with biotinylated secondary antibody), we inferred a measure of functional connectivity between brain regions, following the general methodology introduced by Schiltz et al. (2007). We based our analysis on the average number of activated cells in each brain. These values were correlated across samples, and significant correlations were considered evidence of functional connectivity or coupling between the respective brain regions. These values were correlated across samples, and significant correlations were considered evidence of functional connectivity or coupling between the respective brain regions. As there was no interaction between time of day and MA treatment, we collapsed the data of the light and dark phases for the connectivity analysis. Combining the data from the light and dark phases also increased the power of the analysis. To examine the effects of MA exposure on functional connectivity, we quantified the functional connectivity between the regions for animals exposed to MA vs. saline controls. We correlated the relative activation levels of each brain region across individuals/samples and we considered these measures to reflect the levels of functional coupling between distinct brain regions. For each region, the sum of absolute levels of correlations with all other regions was denoted as the connectivity. To determine whether these connectivity values could arise by chance, we employed a permutation procedure. We randomized the order of the samples and recomputed the correlations, with the expectation that correlation values and consequently connectivity values that arise from chance randomized sample orders will be significantly lower than the values inferred from the unperturbed data. We divided the data in two sets of MA exposed and saline exposed animals, respectively, and we constructed networks from these two groups. In these separate networks, we observed differences in connectivity values of brain regions. To evaluate whether these changes reach statistical significance, we employed a bootstrapping procedure utilized in comparing MRI-derived functional connectivity networks (Hosseini and Gat, 2012), as well as in comparing gene co-expression networks (Gill et al., 2014). We combined the MA exposed and saline exposed data and we separated the combined data in two sets (n = 10,000 random resamples); for each of these resamples we constructed two different networks and we recorded the difference in connectivity. The difference between the MA exposed and saline exposed networks was compared to the distribution of differences arising from mixed-sample networks.

Statistical Analyses
All data are reported as means ± standard error of the mean (SEM). Data were analyzed using GraphPad Prism v.4 and SPSS v.16.0 software (Chicago, IL). Immunohistochemical data were analyzed using Two-Way ANOVA with time and treatment as factors. For the functional connectivity analyses, R software was used. Bonferroni corrected post hoc comparisons were performed when statistically appropriate. Using a bootstrapping procedure, we derived statistical significance for changes in the patterns of connectivity.

c-Fos Quantification
For each brain area, we analyzed whether there were significant effects of treatment, time of day, and a treatment × time interaction. In brain areas in which there was a time × treatment interaction, we also analyzed the effects of the two treatments (i.e., MA and saline) separately. Based on the statistical analysis, there were five distinct patterns of neuronal activation: (1) brain areas that showed an effect of treatment, but no effect of time or time × treatment interaction: (2) brain areas that showed an effect of time, but no effect of treatment or time × treatment interaction; (3) brain areas that showed an effect of treatment and a treatment × time interaction; (4) brain areas that showed Frontiers in Neuroscience | www.frontiersin.org effects of treatment and time but no time × treatment interaction; and (5) brain areas that show no effects of treatment or time and no time × treatment interaction. Each pattern is described separately below. Please see also Table 1 for an overview of the statistical effects.

Pattern 2. Brain Areas That Showed Only an Effect of Time
Three brain regions showed significantly more c-Fos positive cells in the light than dark period. A significant main effect of time, but no significant effects of MA, was found for the number of c-Fos cells in the SCN [F (1, 36) = 185.78, p < 0.001; Figure 2A], VMH [F (1, 36) = 12.92, p < 0.001; Figure 2B], and DMH [F (1, 36) = 5.48, p = 0.02; Figure 2C]. Representative images for neuronal activation in the SCN following saline or MA exposure in the day and night phases are shown in Figure 2D. There were trends toward an effect of time in the ILC (p = 0.08, Figure 1A) and CIN (p = 0.06, Figure 1E) but that did not reach statistical significance. Frontiers in Neuroscience | www.frontiersin.org

Pattern 3. Brain Areas That Showed Effects of Treatment and a Treatment × Time Interaction
Some regions showed only a profound increase in c-Fos positive cells following MA in the day. A significant interaction between treatment and time was found in the PVN [F (1, 36) = 26.51, p < 0.001; Figure 3A], PVT [F (1, 36) = 9.26, p < 0.01; Figure 3B], and ARC [F (1, 36) = 90.17, p < 0.001; Figure 3C]. In the light phase, the number of c-Fos positive cells was higher following MA than saline exposure in the PVN (p < 0.001), PVT (p < 0.001), and ARC (p < 0.001). In the dark phase, the number of c-Fos positive cells was only higher following MA than saline exposure in the PVT (p = 0.028) but not in the PVN or ARC. In the PVN, in addition to the MA × time interaction, there were also main effects of MA [F (1, 36) = 38.09, p < 0.001] and time [F (1, 36) = 12.89, p < 0.001; Figure 3A). Representative images for neuronal activation in the PVN following saline or MA exposure in the light and dark phases are shown in Figure 3D.
In addition to the MA × time interaction, there were also main effects of MA in the PVT [F (1, 36) = 50.24, p < 0.001; Figure 2B], and ARC [F (1, 36) = 15.05, p < 0.001; Figure 2C]. As there was a treatment × time interaction in these brain areas, we also assessed whether for each treatment there was a significant difference between neuronal activation in the light and dark phase. In the PVN (p < 0.0001), PVT (p = 0.01), and ARC (p = 0.03) neuronal activation following MA exposure was significantly greater in the light than dark phase. In contrast, in the ARC, neuronal activation following saline exposure was significantly greater in the dark than the light phase (p < 0.01). In these three brain regions there was a MA × time interaction. In the dark phase, the number of c-Fos positive cells was only higher following MA than saline exposure in the PVT (p = 0.028) but not in the PVN or ARC. In the PVN (p < 0.0001), PVT (p = 0.01), and ARC (p = 0.03) neuronal activation following MA exposure was significantly greater in the light than dark phase. In contrast, in the ARC, neuronal activation following MA exposure was significantly greater in the dark than the light phase (p < 0.01). In the PVN (p = 0.12) and PVT (p = 0.20), there was no difference in neuronal activation following saline exposure in the light and dark phase. Representative images for neuronal activation in the PVN following saline or MA exposure in the day and night are shown in (D). n = 10 mice/treatment/time period.
In the PVN (p = 0.12) and PVT (p =0.20), there was no difference in neuronal activation following saline exposure in the light and dark phase.  Figure 5C). There was a trend toward an interaction between MA and time in the MEA (p = 0.08) with an increase in the number of c-Fos positive cells following MA exposure in the day but decrease in the number of c-Fos positive cells following MA exposure in the night but that did not reach significance. Representative images for neuronal activation in the DG following saline or MA exposure in the light and dark phases are shown in Figure 5D.

c-Fos-AVP Dual-labeling
Analysis of the number of c-Fos/AVP dual labeled cells in the PVN revealed comparable numbers in all four experimental groups and no effect of MA treatment (Figure 6A; see Figure 6B for a representative image of c-Fos/AVP double labeling). There was no effect of MA treatment on the number of c-Fos/AVP dual labeled cells in the SCN either [F (1, 12) = 0.16, p = 0.69, data not shown].

Functional Connectivity Analysis
We inferred a measure of functional connectivity between brain regions, following the general methodology introduced by Schiltz et al. (2007). We started by computing the correlations between all distinct brain regions; summing the absolute values of these connectivity values resulted in a measure of functional connectivity of each brain region. It is conceivable that spurious FIGURE 5 | Brain areas that showed no treatment or time effects or a treatment × time interaction. There were no significant effects of MA, time, or an interaction between MA and time in the MEA (A), BLA (B), and DG (C). There was a trend toward an interaction in the MEA (p = 0.08) but that did not reach significance. Representative images for neuronal activation in the DG following saline or MA exposure in the day and night are shown in (D). n = 10 mice/treatment/time period.
correlations between these quantities could result in false positives, or connectivity values that are due to chance. To guard against this possibility, we employed a randomization procedure. We randomized the order of the samples (N = 10, 000) and we recomputed the connectivity values for each region utilizing the randomized data. The connectivity values inferred utilizing the correct sample order were compared against the distribution of these random connectivity values and quantified utilizing Z-scores. Significant values (Z > 2) were taken to signify connectivity that is unlikely to arise by pure chance. Nearly all of the regions examined, the connectivity values were significantly higher than values derived from randomized data (see Table 2).
In particular, VMH appears highly connected following MA exposure but not after saline. Next, we directly tested the hypothesis that VMH connectivity is significantly altered following exposure to MA. As detailed in Methods, we employed a bootstrapping procedure which involved combining the MA and saline data, randomly assigning each sample to one of two mixed groups (N = 10, 000), and computing the differences between these groups. Differences between groups arranged in this manner are expected to be small and to reflect spurious changes arising from pure chance. Compared with this distribution of random differences, the difference in VMH connectivity between the MA exposed and saline exposed networks corresponds to a Z-score of 2.6 and a bootstrap p-value. We illustrate these results in Figure 7.
For a majority of regions examined, the connectivity values were significantly higher than values derived from randomized data (see Table 2).
Next, we inquired whether exposure to MA disrupts the pattern of connectivity between brain regions. The most affected region was the VMH, where the difference in connectivity between the MA exposed and saline exposed networks corresponds to a Z-score of 2.6. We illustrate these results in Figure 7.

Discussion
The results of this study demonstrated that, within the limited number of the brain areas analyzed, there are at least five distinct patterns of neuronal activation ( Table 1): (1) brain areas such as the CEA, BNST, CA1, CIN, and NACc that show only an effect of treatment. In the CA1, more neuronal activation following exposure to saline is seen in the dark than light phase (p < 0.05) and there was a trend toward such an effect in the ILC (p = 0.054). This circadian-dependent effect was not seen in the CEA, BNST, and NACc; (2) brain areas such as the SCN and VMH that showed only an effect of time; (3) brain areas such as the PVN, PVT, and ARC, which show an effect of treatment and a treatment × time interaction with a more profound activation during the light compared to the dark phase; (4) brain areas such as CA3 and NACs which show effects of treatment and time; and (5) brain areas such as MEA, BLA, and DG that show no neuronal activation following MA exposure and comparable neuronal activation in the light and dark phases. We recognize that in brain areas with elevated basal levels of c-Fos no effects might have been seen due to a potential ceiling effect. The pattern of neuronal activation following MA exposure that shows more profound activation during the light compared to the dark phase highlights the importance of time of drug taking (for review, see Hasler et al., 2012).
The difference in connectivity between the MA and saline exposed networks was the greatest in the VMH. The enhanced functional connectivity between the VMH and the PVN, BLA, and MEA following MA exposure is consistent with a stress response (Dunn, 1987;Sánchez et al., 1995;Walker and Davis, 1997;Fendt et al., 2005). Consistent with this notion, MA activates the HPA axis (Williams et al., 2000;Acevedo et al., 2008;Zuloaga et al., 2014) and chronic stress augments the acute and long-term effects of MA (Matuszewich and Yamamoto, 2004). In brain areas like the PVN, the circadian-dependent pattern of activation following MA exposure resembles that seen for stress-induced activation of the HPA axis. The pattern of HPA axis activation shows low baseline stress hormone levels and a stronger stress response in the early light period, compared to a high baseline and reduced stress response in the late light period, close to the onset of the dark period (Raber et al., 2000).
Within the limitations of the current study, AVP does not seem to play a significant role in the MA-induced neuronal activation in the PVN in the light phase. However, we recognize that AVP might play role in the effect of MA at other time points or in other brain areas, such as the VMH in which AVP was shown to provide a diurnal rhythm of stimulatory input in rats (Kalsbeek et al., 1996). Future efforts are warranted to determine the role of AVP in MA-induced circadian-dependent neuronal activation.
When neuronal activation was analyzed in individual brain regions, MA exposure did not affect neuronal activation in the VMH in either the light or dark phase compared to saline exposure, a pattern similar to that seen in the SCN. However, the functional connectivity data with a profound difference  in connectivity between the MA exposed and saline exposed networks might suggest a role for VMH in MASCO. Consistent with this hypothesis, the VMH is involved in synchronization of rhythms to periodic feeding (Inouye, 1983) and in circadian rhythms of circulating levels of insulin, glucose, and triglycerides (Kraeuchi et al., 1985;Egawa et al., 1993;Mitsushima et al., 1994). However, we recognize that a brain region involved in MASCO might show a response in neuronal activation following MA exposure in either the light and/or dark phases and that at present too little is known about MASCO to draw any definite conclusion.
The alterations in coordinated activity in brain following MA exposure might play an important role in the detrimental effects of MA. In addition to environmental factors, genetic factors also play an important role in the strengths of network connectivity (Glahn et al., 2010;Fornito et al., 2011) and might play a role in modulating individual susceptibility to effects of MA. Recently, synaptic genes and polymorphisms in these genes were shown to be particularly important in determining functional connectivity in brain and to be associated with resting state measures of functional magnetic resonance imaging (fMRI) and neurological diseases (Richiardi et al., 2015). Future efforts are warranted to determine how the connectivity data revealed using c-Fos immunoreactivty in this study compare to MRI measures following MA exposure.
In summary, the data of this study show distinct patterns of neuronal activation following MA exposure. In several brain areas, neuronal activation following exposure to MA is stronger in the light than the dark phase highlighting the importance of considering circadian periods in defining experimental conditions and understanding the mechanisms underlying detrimental effects of MA exposure to brain function. The enhanced functional connectivity between the VMH and areas like the PVN, BLA and MEA following MA exposure is consistent with a stress response. Future efforts are warranted to determine the role of the VMH in effects of MA on the brain.

Author Contributions
DZ: experimental design; acquisition and analysis of data, interpretation of the data and input to the manuscript. OI: analysis of data, interpretation of the data, and input to the manuscript. SW: acquisition and analysis of data, and input to the manuscript. DE: acquisition and analysis of data, and input to the manuscript. TM: acquisition of data and input to the manuscript. BS: acquisition of data and input to the manuscript. CA: interpretation of the data and input to the manuscript. JR: experimental design, interpretation of the data and input to the manuscript.