Long-access heroin self-administration induces region specific reduction of grey matter volume and microglia reactivity in the rat

heroin, grey matter volume, escalation, microglia, immune response 26 27 28 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint 29

30 In opioid use disorder (OUD) patients, a decrease in brain grey matter volume (GMV) has been 31 reported. It is unclear whether this is the consequence of prolonged exposure to opioids or is a 32 predisposing causal factor in OUD development. To investigate this, we conducted a structural MRI 33 longitudinal study in NIH Heterogeneous Stock rats exposed to heroin self -administration and age -34 matched naïve controls housed in the same controlled environment. Structural MRI scans were 35 acquired before (MRI 1) and after (MRI 2) a prolonged period of long access heroin self -administration 36 resulting in escalation of drug intake. Heroin intake resulted in reduced GMV in various cortical and 37 sub-cortical brain regions. In drug-naïve controls no difference was found between MRI 1 and MRI 2. 38 Notably, the degree of GMV reduction in the medial prefrontal cortex (mPFC) and the insula positively 39 correlated with the amount of heroin consumed and the escalation of heroin use. In a preliminary gene 40 expression analysis, we identified a number of transcripts linked to immune response and 41 neuroinflammation. This prompted us to hypothesize a link between changes in microglia homeostasis 42 and loss of GMV. For this reason, we analyzed the number and morphology of microglial cells in the 43 mPFC and insula. The number of neurons and their morphology was also evaluated. The primary motor 44 cortex, where no GMV change was observed, was used as negative control. We found no differences 45 in the number of neurons and microglia cells following heroin. However, in the same regions where 46 reduced GMV was detected, we observed a shift towards a rounder shape and size reduction in 47 microglia, suggestive of their homeostatic change towards a reactive state. Altogether these findings 48 suggest that escalation of heroin intake correlates with loss of GMV in specific brain regions and that 49 this phenomenon is linked to changes in microglial morphology. 50 51 Key words: Addiction, Abuse, Neuroinflammation, Immune System, Opioids 52 53 54 55 56 57 58 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint 59

60 Opioid use disorder (OUD) is a chronic psychiatric condition characterized by compulsive drug taking, 61 severe intoxications, and abstinence episodes generally followed by relapse. Over the last twenty years, 62 the United States has witnessed a rise of overdose deaths caused by heroin and prescription opioids 63 (NIDA, 2023) that have drawn renewed attention toward OUD. This called for research efforts to 64 understand the neurobiology and neuropharmacology of OUD, necessary to widen the basis for future 65 development of OUD treatment and safer opioid-based pain therapies. 66 67 Magnetic resonance imaging (MRI) studies in patients diagnosed with OUD consistently reported brain 68 structural alterations compared to healthy controls. Grey matter volume (GMV) reduction in heroin 69 dependent patients was observed in the prefrontal (Lin et al., 2018; Qiu et al., 2013; Shi et al., 2020), 70 cingulate (Schmidt et al., 2021; Wang et al., 2012), and insular (Bach et al., 2019; Bach et al., 2021) 71 cortices or in combinations of these areas (Liu et al., 2009; Sun et al., 2017; Sun et al., 2016). Fewer 72 studies described GMV reduction in subcortical areas such as amygdala (Schmidt et al., 2021), globus 73 pallidus (Shi et al., 2020), and putamen (Bach et al., 2021). In some cases, increased GMV in the 74 somatosensory cortex (Shi et al., 2020; Sun et al., 2016) and caudate putamen (Schmidt et al., 2021) 75 of these patients were also reported. 76 77 While there is a general agreement that patients with OUD exhibit a decrease in cortical grey matter 78 volume (GMV), the specific regions involved, and the extent of the observed reduction differ across 79 studies. The heterogeneity in the results reported by human studies could stem from differences in 80 inclusion/exclusion criteria. For instance, the length of abstinence at the time MRI scans were acquired 81 vary from no abstinence (Liu et al., 2009) to more than five drug free years (Shi et al., 2020; Wang et 82 al., 2016). Some studies included patients under methadone/buprenorphine treatments (Bach et al., 83 2019; Liu et al., 2009), or included in medically assisted diacetylmorphine delivery programs (Schmidt 84 et al., 2021). In other studies, GMV alterations were not found in the general population but could be 85 observed only in groups filtered for genetic predisposition (Sun et al., 2017; Sun et al., 2016) . The 86 relative number of males and females is also a source of variability among studies, as women were 87 often not included or underrepresented. Finally, there are several factors that can significantly impact 88 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint GMV, some of which are difficult to predict, such as chemical impurities found in illicitly sold heroin, or 89 factors not directly associated with the pharmacological effects of the drug, such as poor quality of life 90 often associated with heroin use (Frischknecht et al., 2011; Puigdollers et al., 2004; Yen et al., 2011). 91 In fact, it has been demonstrated that GMV in anterior cingulate cortex, medial prefrontal cortex, and 92 insula, the three brain regions mainly affected by heroin, is positively correlated with physical and 93 general health related quality of life (Hahm et al., 2019). 94 In summary, it is difficult to disentangle the effect of heroin exposure on GMV from that of environmental 95 confounding factors inherent in human studies. Another important limitation in clinical imaging studies 96 is the difficulty in determining whether the MRI changes observed in OUD patients are the consequence 97 of prolonged drug exposure or are predisposing factors to the disease. One way to answer this question 98 and to clarify the impact of environmental variables, is to conduct longitudinal studies with imaging data 99 collected prior and after opioid exposure and dependence development. For obvious reasons, 100 conducting these studies in humans can be challenging, while it is relatively straightforward to carry 101 them out in laboratory animals. 102 In the first part of our study, we used a longitudinal approach and structural MRI to measure GMV in 103 rats before and after protracted (12 hrs. a day) exposure to operant heroin self -administration. Age 104 matched heroin naive rats were used as a control. The study was carried out under tightly controlled 105 environmental conditions. To better mimic the genetic heterogeneity of the human population, we used 106 NIH heterogeneous stock (HS) rats, an outbred line classically used for genetic and phenotypic studies 107 including GWAS analysis (Solberg Woods and Palmer, 2019). Previous research has shown that these 108 rats are suitable for studying opioid abuse-related behavior (Allen et al., 2021; de Guglielmo et al., 2019; 109 Kallupi et al., 2020; Kuhn et al., 2022). Behavioral measures aimed at evaluating the motivation for 110 heroin and the transition from its use to abuse were also taken. In the second part of the study, guided 111 by the MRI results and transcriptomic data from HS rats exposed to heroin, which suggested 112 neuroimmune system activation, we conducted an extensive immunohistochemistry investigation to 113 assess whether changes in GMV following heroin use involved alterations in microglia morphology 114 compared to neurons. 115 116

117 Animals 118 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint NIH heterogeneous stock (HS) rats were obtained from the Biomedical Resource Center, Medical 119 College of Wisconsin (Milwaukee, USA). Rats were housed in groups of two, in a room with artificial 120 12/12 h light/dark cycle (light off at 8 AM), constant temperature (20-22°C) and humidity (45-55%). Rats 121 had ad libitum access to tap water and food pellets (4RF18, Mucedola, Settimo Milanese, Italy) unless 122 differently indicated. All procedures were conducted in adherence to the “European Community Council 123 Directive” and “NIH Guidelines” for Care and Use of Laboratory Animals. 124 Drugs 125 Heroin (diacetylmorphine hydrochloride, NIDA drug supply program) was dissolved in sterile 126 physiological saline. 127 Experiment 1: Heroin self-administration and seeking. 128 Surgeries, self-administration apparatus and food -reinforced pre-training of operant lever responding 129 are described in Supplementary Information. Fifteen HS rats (7 males and 8 female) were implanted 130 with an indwelling catheter into the right jugular vein to receive heroin infusions. Operant heroin self -131 administration (HSA) was carried out according to a protocol adapted from (Vendruscolo et al., 2011) 132 and briefly described below. Additional ten rats (5 males and 5 female) were subjected to the occlusion 133 of the right jugular vein and received i.v. saline infusions. This yoked group served as a control for the 134 MRI experiment described below. Timelines of the experiment are depicted in Figure 1A. 135 The heroin group was initially trained to 1 -hour short access (ShA) HSA. Sessions started with the 136 insertion of both retractable levers. The right lever was designated as the active lever. Depression of 137 the active lever under fixed ratio 1 (FR1) schedule of reinforcement activated the infusion pump for 5 138 seconds. Each pump activation resulted in the intravenous delivery of 60 µg/kg of heroin in a volume 139 of 0.1 ml. A 20 second time out (TO) followed pump activation. During TO, active lever pressing had no 140 scheduled consequences. The cue light located above the active lever was activated contingently with 141 the infusion pump and remained on for the whole TO. The left lever was designated as the inactive 142 lever and functioned as control for non-specific behavior. Responses at the inactive lever were recorded 143 but had no scheduled consequences. 144 After nine FR1 ShA SA sessions, motivation for heroin was evaluated under progressive ratio (PR) 145 schedule of reinforcement similarly to that we previously described (de Guglielmo et al., 2015). In this 146 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint session the response requirements necessary to receive a single reinforcement increased after every 147 infusion according to the equation: [5e(injection numbers x 0.2)]-5. The last ratio completed was defined as the 148 break-point (BP) and was taken as a measure of the motivation for heroin expressed by the rat (de 149 Guglielmo et al., 2015; Richardson and Roberts, 1996). The session ended after four hours in total had 150 elapsed or if one hour passed since the last infusion earned, whichever occurred first. 151 After the PR test, rats were subjected to one additional ShA session, after which they entered the 12 -152 hours long access (LgA) HSA phase. LgA sessions were identical to ShA sessions except that they 153 lasted for twelve hours. After eight LgA sessions, rats were tested again in a PR schedule of 154 reinforcement as described above for the ShA phase. 155 The day after the second PR test a heroin primed reinstatement test was run according to a within 156 session extinction-reinstatement protocol (Shaham et al., 2003). This session lasted three hours in total. 157 During the first hour, the syringe pumps were off and lever pressing did not deliver heroin. This session 158 served as an extinction phase characterized by progressive decrease of lever pressing. At the 159 beginning of the second hour, syringe pumps were switched on for five minutes, during which rats could 160 self-administer a maximum of two infusions in the same condition of a SA session. If a rat failed to self-161 administer two heroin infusions within five minutes, the infusions were delivered by the operator that 162 manually activated the infusion pump. The two infusions served to prime heroin seeking behavior that 163 was monitored for the remaining 2 hrs. After the priming reinstatement test, rats returned to LgA FR1 164 baseline self -administration for an additional ten sessions. Heroin training and behavioral tests, 165 sessions were run for four days a week, from Monday to Friday with a random one-day break in within. 166 Sessions were performed during the dark phase of the light/dark cycle, and during long access training 167 tap water and chow -pellets were available in the self -administration chambers. The yoked group 168 received 35 random saline infusions paired with the same cues experienced by the heroin rats. 169 170 Experiment 2: Effect of heroin consumption on grey matter volume in HS rats 171 At fourteen-week of age, prior to enter into Experiment 1, twenty five HS rats (13 males and 12 females) 172 were subjected to the first MRI acquisition (MRI 1) at the Italian Institute of Technology (IIT, Rovereto). 173 A week later, rats were transferred to the University of Camerino. At twenty weeks of age rats were 174 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint divided into the heroin group (7 males and 8 female) and yoked saline controls (5 males and 5 female) 175 and subjected to Experiment 1. At completion of Experiment 1 rats were transferred back to the IIT for 176 a new MRI acquisition (MRI2). MRI2 acquisition started at 5 days and ended 13 days after the last heroin 177 self-administration (SA) session. Experimental timeline is summarized in Figure 1 A. For MRI 178 acquisition a 7 Tesla (T) Bruker Pharmascan (Bruker BioSpin, DE) in a double coil configuration was 179 used. A 72 ‐mm i.d. single channel transmission ‐only resonator was actively decoupled with a four ‐180 channel phased‐array receive‐only surface coil optimized for the rat brain. After a scout image, high 181 resolution T2w RARE images (TR = 5500 ms, TE = 76 ms, NEX = 8, MTX = 256 × 256 × 25, FOV = 35 182 × 35 × 25 mm, ACQtime = 7min40s) were acquired for voxel-based morphometry analysis. T2w RARE 183 was used with a multislice encoding (2D). Accurate and consistent positioning of the FOV was ensured 184 by placing the center of the first slice 0.5mm rostral to the rhinal fissure. This allowed to cover the entire 185 forebrain and midbrain regions. 186 187 Experiment 3: Effect of heroin consumption on the morphology of microglia in HS rats. 188 Different groups of HS rats (3 males and 3 females) trained to LgA heroin self-administration and saline-189 yoked controls ( 2 males and 2 females) were used (Supplementary Figure S 1) for 190 immunohistochemistry experiments. Specifically, nine days after the last SA session, corresponding to 191 the average time point at which the MRI2 of the longitudinal MRI experiment was acquired, the heroin 192 and saline exposed rats were deeply anesthetized with 5% isoflurane and perfused transcardially with 193 a flush (∼30 s) of saline followed by 200 ml of 4% paraformaldehyde (PFA) in phosphate buffer (PB) 194 and post-fixed overnight in PFA at 4°C. Brains were subsequently sectioned with a vibratome (Leica) 195 into 50 μm thick coronal slices, collected in PB and 0.02% sodium azide (NaN), and stored at 4°C. 196 Four to five slices for each selected region were collected: the medial prefrontal cortex (mPFC) centered 197 at bregma (+2.20), that included the Cingulate area 1 and the Prelimbic Cortex (Cg1, PrL); the Primary 198 Motor Cortex (M1), centered at bregma (+2.20); the Insular cortex (IC), centered at bregma (+0.70), 199 that included the Granular, Dysgranular and Agranular Insular Cortex (GI, DI, AID, AIV) 200 (Supplementary Figure S 2). Coordinates refer to Paxinos & Watson (1998) (Paxinos and Watson, 201 1998). 202 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint Neuronal nuclei and parvalbumin positive neurons were double stained with NeuN and anti-parvalbumin 203 (PV) antibodies respectively; Iba1 antibody was used instead as a marker of microglia. 204 For NeuN and PV double-staining (10 animals, 2 sections/animal), sections were washed three times 205 with phosphate-buffered saline (PBS, pH 7.4), treated with a blocking solution of 3% normal goat serum 206 (NGS) and 0.1% Triton X -100 (TXT-100) for 1h at room temperature (RT), and then incubated in the 207 blocking solution containing mouse anti-NeuN antibody (1:500, Synaptic Systems 266 004) and guinea 208 pig anti-parvalbumin antibody (1:1000, ImmunoStar 24428) for 2h at RT and then overnight at 4°C. The 209 day after, sections were washed three times with PBS and treated with a blocking solution (3% NGS) 210 for 20 min at RT and incubated with 3% NGS PBS containing secondary antibodies (1:600, Alexa-Fluor 211 594 anti-mouse IgG, Invitrogen and Alexa-Fluor 488 anti-guinea pig IgG, Invitrogen) for 90 min at RT. 212 For Iba-1 staining (10 animals, 2 sections/animal), sections were washed three times with PBS (pH 213 7.4), treated with a solution of 3% bovine serum albumin (BSA) and 0.3% Triton X -100 for 1h at RT, 214 and then incubated in the blocking solution (1%BSA and 0.1% TXT -100) containing rabbit anti-Iba1 215 primary antibody (1:600, Fujifilm Wako 019 -19741) for 2h at RT a nd overnight at 4°C. Sections were 216 then washed three times with PBS and treated with a blocking solution (1% BSA) for 20 min at RT and 217 then, they were exposed for 90 min to 1% BSA in PBS containing a secondary antibody (Alexa Fluor 218 488 anti-rabbit IgG, Invitrogen) at RT. 219 Finally, all sections were washed with PBS, mounted, air -dried and cover slipped using either 220 Vectashield mounting medium (NeuN and PV) or Vectashield mounting medium with DAPI (Iba1). In 221 control slices primary antibodies were omitted. 222 Sections were examined with a confocal microscope (Nikon C2+Laser Scanning confocal) and acquired 223 as 512x512 pixel images (0.41 µm pixel size; Z -step, 1 µm; 210.12x210.12 microns) using a 60x oil 224 immersion lens for Neun/PV sections, and an air-dry 40x lens for Iba1 sections. Six fields/layer/animal 225 in layers III and V were acquired for each region (mPFC, M1, IC). 226 Images for NeuN and PV were analyzed using FIJI ImageJ software (Bellesi et al., 2017). The number 227 of NeuN and PV positive cells was counted manually. PV morphology was assessed by brain area 228 around each PV positive cell and measuring perimeter, area, and density. Image processing of Iba -1 229 was performed using a custom-made script to measure individual microglial area and perimeter length 230 in MATLAB, as described in (Bellesi et al., 2017). 231 232 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint Data analyses 233 Behavioral Experiments: Food and heroin self -administration, escalation of heroin intake, and primed 234 reinstatement were analyzed by one-way ANOVA with time as repeated measure. Break points reached 235 during the two PR sessions were compared by Wilcoxon match paired signed rank test. Statistical 236 significance was conventionally set at p<0.05. Data are presented as mean ± SEM. 237 Voxel Based Morphometry : T2w RARE images, acquired at high resolution, were analyzed to 238 investigate voxelwise differences in local GMV using a modified version of the FSL ‐VBM tool of FSL 239 (Smith et al., 2004) to account for rat brain specific issues (Tambalo et al., 2015). Briefly: an unbiased 240 GM template was created by averaging the entire group at both time points (experimental: n = 30, 241 control: n=20). Individual GMV were registered to the unbiased template using the FSL FNIRT algorithm 242 (Schnabel et al., 2003) and then modulated to correct for local deformations by the Jacobian of the 243 warp field, as reported in (Ashburner and Friston, 2000). The modulated GM images were then 244 smoothed with an isotropic Gaussian kernel (σ = 3 mm), and a nonparametric permutation test was 245 applied. The null distribution for the data in the VBM statistics was built over 5,000 permutations. To 246 define more precisely the distribution of GM density modulation in different anatomical regions of the 247 brain, a volumetric reconstruction of the digital anatomical rat brain atlas (Paxinos and Watson, 1998) 248 was coregistered with GM images in template space. 249 Histochemical Analyses: Cell density of NeuN, PV and Iba1 positive cells were compared between 250 heroin experienced and control rats using the non -parametric Mann -Whitney test. Cumulative 251 distributions of morphological parameters of Iba1 and PV positive cells were compared between heroin 252 experienced and control rats using the non-parametric Kolmogorov-Smirnov (KS) test. Male and female 253 rats were grouped together for analyses. 254 255

256 Experiment 1: Heroin self-administration and seeking in HS rats 257 After food-reinforced pretraining ( Supplementary Figure S 3), rats rapidly acquired ShA HSA 258 responding [F(9, 14)=1.0; p>0.05] (Figure 1B, left to the dashed line). Responding at the inactive lever 259 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint was always very low and did not change during training [F(9, 14)=0.9; p>0.05]. Then, when rats entered 260 the 12-hour long access (LgA) phase, ANOVA revealed an overall effect of SA sessions [F(17, 14)=5.9; 261 p<0.001], with heroin intake increasing over time. The increase in intake became statistically significant 262 starting from the 6 th LgA session, after which heroin intake remained stable ( Figure 1B, right to the 263 dashed line). Responding at the inactive lever remained stable over time [F(17, 14)=1.1; p>0.05]. This 264 indicates an escalation of heroin intake induced by long access to heroin, a marker of drug dependence 265 (Ahmed and Koob, 1998; Vendruscolo et al., 2011). Escalation is also indicated by an increase in intake 266 during the first hour of the LgA sessions (Ahmed and Koob, 1998), therefore we analyzed the heroin 267 intake at this time point comparing heroin intake under ShA and LgA condition and we found an overall 268 effect of sessions [F(27, 14)=6.0; p<0.0001]. Heroin intake at the first hour was stable during the ShA 269 and the first four days of LgA, and it started to increase from the fifth LgA session. During the last seven 270 LgA sessions, heroin intake was significantly higher than the last two days of ShA (Figure 1C). Analysis 271 of inactive lever responding found no overall effect of sessions [F(27, 14)=1.0; p>0.05], indicating that 272 the increase observed in the heroin intake did not derive from a general change in behavior and can be 273 interpreted as an escalation of heroin intake. To further evaluate the magnitude of escalation, we 274 compared the average heroin intake during the first hour of SA at four key time points: the first three 275 ShA sessions (early ShA), the last three ShA sessions (late ShA), the first three LgA sessions (early 276 LgA), the last three LgA sessions (late LgA). The ANOVA found an overall effect of time point [F(3, 277 14)=11.2; p<0.001]. Bonferroni’s post-hoc analysis revealed that the intake at late LgA was significantly 278 higher than the other three time points (Figure 1D). 279 Motivation for heroin under PR measured after ShA and LgA HSA. Motivation for heroin was evaluated 280 by analyzing the break points (BP) reached in the two PR SA sessions run at the end of the ShA and 281 LgA sessions, respectively. A two-tailed Wilcoxon matched-pairs signed rank test revealed a significant 282 difference between the two BP [W(15) = 102.0; p<0.001] ( Figure 1E), indicating that rats showed a 283 higher motivation for heroin after escalation of intake. 284 Heroin primed reinstatement. Finally, a within session extinction-priming reinstatement test was run. To 285 analyze the reinstatement, data were computed in 30 min bins to compare the cumulative number of 286 lever presses after priming with lever pressing occurring during the second half of the extinction phase. 287 An ANOVA of active lever presses found an overall effect of time [F(4, 14) = 38.03; p<0.0001]. The 288 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint number of presses increased over time during the reinstatement phase and Bonferroni’s post -hoc 289 analysis revealed that it became significantly higher than extinction at 60 minutes into the reinstatement 290 phase (p<0.001). Bonferroni’s test further revealed that at each time point within the reinstatement 291 phase the number of presses was higher than the previous time point ( Figure 1F upper panel ). 292 Analysis of the inactive lever also found an overall effect of time [F(4, 14) = 8.8; p<0.01]. Bonferroni’s 293 post-hoc analysis revealed significant differences exclusively within the reinstatement phase time 294 points, specifically between 120 and 60 minutes and between 120 and 30 minutes ( Figure 1F lower 295 panel). Altogether, these analyses indicated a significant priming -induced reinstatement of heroin 296 seeking. 297 298 Experiment 2: Effect of heroin consumption on grey matter volume in HS rats 299 In heroin experienced rats we observed a negative difference between GMV in MRI2 and MRI1 indicating 300 a decrease in GMV in several cortical and subcortical regions. Specifically, GMV reduction was 301 detected in the insular cortex (IC), prelimbic cortex (PL), infralimbic (IL), and Cingulate area 1 (Cg1), 302 the caudal portion of the Retrosplenial Cortex (RsC), Orbitofrontal Cortex (OFC), Hypothalamus (Hyp), 303 Thalamus (Th), Entorhinal Cortex (EnC), Hippocampus (Hipp), bed nucleus of the stria terminalis 304 (BNST), Amygdala (amy), Nucleus Accumbens (Acb), and Dorsal Striatum (DS). (Figure 2, corrected-305 p < 0.01, blue = reduction of GM volume). No positive difference between GMV in MRI2 and MRI1 (i.e., 306 no increase in GMV) was observed at any brain level. The heroin naive group of animals, evaluated at 307 the same timepoints, showed neither positive nor negative GMV changes between the two scans, 308 indicating that GMV was not affected by manipulation of the rats and/or aging When we used drug 309 seeking and taking as covariates to assess their correlation with GMV changes (MRI 2 minus MRI 1) 310 (Figure 3) we found a negative correlation with both heroin intake and escalation (both at the 1st and 311 12th hour of HSA) in the mPFC, IL, IC, DS and Acb. For IC and DS, this correlation extended also to 312 the extinction responding and priming induced reinstatement of heroin seeking. Finally, in the caudal 313 RsC GMV negatively correlated with priming and total heroin intake at the 12th hour. 314 315 316 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint Experiment 3: Effect of heroin consumption on the morphology of microglia in HS rats. 317 To further investigate the biological phenomenon that may be at the basis of the reduction in GMV 318 observed in heroin experienced rats relative to controls, we measured the density and size of neurons 319 and microglial cells in two brain regions in which we observed GMV reduction - the mPFC and the IC - 320 and one region in which GMV reduction was not observed, the primary motor cortex (M1) (Figure 4 A-321 C). 322 Cell density. Heroin experienced rats showed higher density of PV positive neurons in the M1 [M ann-323 Whitney U = 0; p< 0.01] (Figure 4 D), while no significant differences were found between heroin and 324 saline yoked groups in the other brain regions and for the other cell types (Supplementary Figure S4). 325 Size of parvalbumin neurons 326 We estimated the area and perimeter of 701 PV positive neurons (214 mPFC, 223 IC, 264 M1) in 327 heroin-exposed animals and 232 PV positive neurons (93 mPFC, 96 IC, 43 M1) in yoked controls . In 328 the insula, we observed a rightward shift of both Area [KS D = 0.33; p<0.0001, +30.3%] and Perimeter 329 [KS D = 0.27; p<0.0001, +12.9%] distributions in heroin-exposed animals relative to controls, indicating 330 an increase in the size of these neurons ( Supplementary Figures S 5). Conversely, no significant 331 change was observed in PV size in the mPFC and M1. 332 Morphological analysis of microglia. 333 Here, we quantified cell size, perimeter length, and the area/perimeter ratio of 1236 Iba1 positive cells 334 (375 mPFC, 452 IC, 409 M1) in heroin exposed animals and 1044 cells (327 mPFC, 357 IC, 360 M1) 335 in yoked rats as a proxy of microglia state in three different brain regions. In the mPFC, we observed a 336 reduction in the area [KS D = 0.39; p<0.0001] and perimeter length [KS D = 0.39; p<0.0001] of microglial 337 cells, as shown by the leftward shift of both cumulative distributions, in the heroin experienced group 338 relative to the saline yoked rats ( Figure 5A, B). The area/perimeter ratio was instead right -shifted in 339 the heroin experienced rats [KS D = 0.29; p<0.0001] ( Figure 5C ). The higher area/perimeter ratio 340 observed in heroin experienced rats, in which both area and perimeter were smaller, is consistent with 341 a larger decrease observed in cell perimeter (-58.6%) than in cell area (-50.5%). 342 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint In the IC of the heroin experienced group, we observed a leftward shift of both cell area [KS D = 0.18; 343 p<0.0001] and perimeter [KS D = 0.17; p<0.0001] that was maintained until the 87th and 85th percentile 344 of the two cumulative distributions respectively (Figure 5D, E). Also in this case, the smaller area and 345 perimeter associated with a larger area/perimeter ratio in the heroin experienced rats [KS D = 0.11; 346 p<0.05] (Figure 5F), was due to larger decrease in cell perimeter (-10.1%) than in cell area (-6.5%) 347 In M1, similar to the other two regions analyzed, we observed left-ward shift of both microglial cells area 348 [KS D = 0.27; p<0.0001, -30% on average] and perimeter [KS D = 0.3; p0.05] (Figure 5I). 351 For both Iba-1 and PV positive cells, the morphological parameters’ distributions were similar across 352 groups and are reported in Supplementary Figures S6 and S7. 353 In summary, the population of microglial cells sampled in both groups contained a wide range of sizes 354 and morphologies ( Figure 6), but while control rats were enriched in highly ramified cell phenotypes 355 (Figure 6C), heroin experienced rats contained smaller and less ramified cells (Figure 6A). 356 357 1. DISCUSSION 358 Our results demonstrate a significant escalation of heroin intake and enhanced motivation for the drug 359 following protracted LgA self -administration. These features reflect some of the primary DSM criteria 360 for OUD diagnosis (i.e., the substance is often taken in larger amounts and over a larger period than 361 intended) (Edwards and Koob, 2013). Escalation and enhanced motivation for heroin following LgA 362 exposure were associated with a reduction of GMV in cortical and subcortical regions shown to being 363 part of the addiction neurocircuitry (Koob and Volkow, 2016). Moreover, consistent with observations in 364 heroin dependent patients (Lin et al., 2018; Qiu et al., 2013; Schmidt et al., 2021; Shi et al., 2020; Wang 365 et al., 2012), we found reduction of GMV extending from the Cg1 dorsally, to the IL ventrally (Paxinos 366 and Watson, 1998). These structures are functionally, and to a large extent anatomically, homologous 367 to the dorsolateral prefrontal cortex and anterior cingulate cortex in primates (Seamans et al., 2008; 368 Uylings et al., 2003). In addition, as previously described in human opioid abusers, we observed GMV 369 reduction in the insular cortex (Bach et al., 2019; Bach et al., 2021). Our results expand these findings 370 with human as our longitudinal approach and the use of an age matched control group support two 371 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint major conclusions: first, changes in GMV in these regions appears to be the consequence of heroin 372 exposure rather than being a predisposing factor to heroin abuse; second, due to the tightly controlled 373 environmental conditions used with the laboratory animal experiments it is conceivable to argue that 374 heroin exposure alone is adequate to induce these changes in GMV, thereby eliminating the influence 375 of common confounding factors (i.e, patient history, lifestyle etc.) intrinsically associated with clinical 376 studies. Nonetheless, the contribution of environmental factors in affecting GMV changes observed in 377 OUD patients should not be overlooked. 378 To further explore the link between heroin consumption and brain structural alterations, we also ran 379 correlation analyses between behavioral readouts and GMV changes. We found a negative correlation 380 between the total heroin intake and escalation of drug consumption with GMV reduction in cortical 381 regions and in the striatum. This finding further strengthens the hypothesis that reduced GMV is a 382 consequence of heroin consumption. However, this does not rule out the possibility that diminished 383 GMV could contribute to shaping the progression of OUD. For instance, the mPFC modulates impulse 384 controls and the immaturity or damage of this area has been associated with poor control inhibition and 385 the undertaking of risky behavior, including drug binging (Perry et al., 2011). One could argue that 386 chronic heroin consumption reduces GMV in fronto -cortical regions, which in turn results in a poor 387 behavioral control that further facilitates binging episodes, eventually leading to escalation of heroin 388 intake. In a self -powered vicious cycle, enhanced drug consumption may then contribute to further 389 reduction of GMV. In other words, it is possible that heroin consumption leads to the reduction of GMV 390 which would then promote further heroin seeking. When correlational analysis was applied to heroin 391 extinction/seeking data, a significant link with GMV reduction was found in the insula and DS. It is 392 tempting to speculate that reduced GMV in the insula may reflect changes in interoceptive perception 393 of heroin (in the case of priming) and/or in drug abstinence (Droutman et al., 2015). Also, structural 394 changes in the DS may be linked to aberrant habit learning and promotion of relapse -like behavior 395 (Everitt and Robbins, 2013). 396 397 To gain further insights on the morphological changes occurred following heroin we expanded our 398 investigation to determine possible cellular adaptations associated with GMV reduction. 399 In a preliminary analysis of an ongoing wide transcriptomic investigation in HS rats trained for LgA 400 heroin self-administration, we observed that drug exposure induced significant expression changes in 401 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint a number of genes associated with immune modulation and microglia activity in the prelimbic cortex 402 (see Supplementary Table S1). Notably, the prelimbic cortex is one of the subregions in which 403 structural changes were observed in our MRI study. In addition to this finding , previous work studying 404 the cellular correlate of GMV variations showed multiple potential mechanisms, including changes in 405 the density and size of neuronal and glial cells, but also remodeling of dendritic spines. For instance, 406 after heart failure, a decreased number of neurons and increased microglia were found in the same 407 region where GM concentration was reduced (Bach et al., 2019). However, in other cases, changes in 408 voxel-based morphometry were mostly explained by changes in dendritic spine density or cell clustering 409 rather than in cell death or proliferation (Asan et al., 2021; Keifer et al., 2015). 410 Inspired by these results we decided to carry out an immunohistochemistry analysis of neurons and 411 microglia in two regions (mPFC and IC) in which the longitudinal MRI experiment showed decreased 412 GMV, and one region (M1) in which GMV was unaltered. 413

showed that heroin self-administration did not affect neuronal and microglia cell density except 414 for an increased number of PV neurons in the motor cortex. In the insula, we observed a rightward shift 415 of both area and perimeter distributions in heroin -exposed animals relative to controls, indicating an 416 increase in the size of these neurons. Conversely, no significant changes were detected in the mPFC 417 and M1. When we looked more in depth at the microglia, we found that heroin self -administration was 418 associated with an overall shrinkage of these cells and an increased area/perimeter ratio in the IC and 419 in the mPFC. This indicated a possible retraction of microglial processes and a shift towards a more 420 amoeboid reactive state of these cells. Interestingly, in M1, where GMV was not altered by heroin, the 421 shrinkage of microglia was not accompanied by a change in the area/perimeter ratio suggesting a 422 different kind of morphological reshaping and perhaps of reactivity. De Santis and colleagues 423 demonstrated that following alcohol drinking, the switch of microglia morphology to a less ramified 424 shape decreased the extracellular space complexity in the grey matter, increasing mean diffusivity and 425 promoting neurotransmitter diffusion (De Santis et al., 2020). Likewise, the increased area/perimeter 426 ratio observed in in our heroin exposed rats may be due to a less ramified morphology of the microglia 427 resulting in a decreased extracellular space complexity and subsequent enhancement of 428 neurotransmitter diffusion from release sites. 429 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint Microglia are specialized resident immune cells of the central nervous system, which upon insult can 430 react to mediate pro- or anti-inflammatory processes (Paolicelli et al., 2022). A change in the microglial 431 state is accompanied by a morphological transformation, from a highly ramified phenotype with thin 432 processes extending far from the small soma , to a less ramified shape with shorter and dynamic 433 processes and a larger amoeboid-like appearance (Stence et al., 2001). Even though we did not directly 434 assess the number and length of microglial process ramifications, the increase in the cell area/perimeter 435 ratio is compatible with a reduction in their complexity. Reabsorption of microglial processes and shift 436 from a highly ramified to a less ramified shape is consistent with a possible inflammatory state in heroin 437 treated animals relative to controls in the mPFC and Insula. The presence of a pro -inflammatory state 438 associated with heroin escalation still needs to be demonstrated with specific experiments (Davis et al., 439 2017; Stence et al., 2001) , however, this would be in line with the pharmacological effects of heroin’s 440 metabolites. Heroin is de-acetylated to monoacetyl-morphine and then to morphine, which is converted 441 to morphine-6-glucuronide (M6G) and morphine-3-glucuronide (M3G) (Milella et al., 2023; Rook et al., 442 2006). While M6G contributes to the rewarding effects of heroin, M3G has no affinity for opioid 443 receptors, but binds and activates the toll -like receptor 4 (TL4) / myeloid differentiation factor 2 (MD2) 444 complex on microglia surface (Green et al., 2022), which in turn leads to the release of pro-inflammatory 445 factors like TNFα and IL -1β. This effect of M3G would be consistent with an inflammatory state 446 associated with the increased area/perimeter ratio that we observed. Noteworthy, in our preliminary 447 RNAseq analysis we found genes such as Il1r1, that encodes IL-1β cognate receptor, and Nfkbib, a 448 modulator of the TNF signaling pathway (Yazdi & Ghoreschi, 2016; Fields et al., 2019) among the genes 449 which expression was altered by heroin. It is also possible that the stimulation of microglia reactivity by 450 the heroin metabolite M3G contributed to the escalation of heroin intake. Indeed, it was reported that 451 blockade of glial reactivity reduced morphine induced release of dopamine in the ventral striatum (Bland 452 et al., 2009), possibly through a TLR4 dependent mechanism, as TLR4 knock out mice do not develop 453 opioid place preference (Hutchinson et al., 2012). Consistently, the blockade of TLR4 by (+)-naltrexone 454 decreased operant self -administration of remifentanil (Yue et al., 2020), though the selectivity of this 455 effect has been argued (Tanda et al., 2016). In addition to opioid reward, activation of TLR4 plays a 456 role in the development of opioid tolerance (Eidson and Murphy, 2013). It is thus possible that by 457 decreasing reward and increasing tolerance, the reactive state of microglia contributed to the escalation 458 of heroin intake. 459 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint The under-representation of female subjects has been a bias in both clinical and preclinical research 460 since many years . To avoid this bias , in this study we adopted a sex matched approach. However, 461 while a sex -matched experimental design allowed us to generalize between male and female , the 462 sample size of our experiments is insufficient to evaluate if sex differences in response to heroin 463 occurred (Paolicelli et al., 2022). 464 In conclusion, our data demonstrated that heroin consumption under long access contingency led to 465 region-specific reduction in GMV that correlated with the quantity and escalation of heroin intake. Such 466 reduction was confined to cortical regions known to be consequential in substance use disorder. GMV 467 reduction occurred in the same areas where alterations in microglia morphology were also observed, 468 which suggests that this cell population may contribute to gray matter alterations following heroin. 469 470

471 This work was supported by U01-DA045300 to GH, LSW, DC, PK and RC. The Authors wish to thank 472 Agostino Marchi, Rina Righi and Matteo Valzano for animal care and technical support. AB 473 acknowledges support by the Fondazione Cassa di Risparmio di Torino (CRT), R.F. 2019.0610. 474 Author Contributions: NC and RC ideated the project, NC coordinated the implementation of this 475 work, ran behavioral experiment, and analyzed data , RC supervised the project . ST acquired scans 476 and analyzed MRI data. VL ran self -administration sessions, and immunofluorescence. GS acquired 477 MRI scans, LdV supervised immunofluorescence, SA, AK, JM, and GH provided preliminary RNAseq 478 data, LCSW supervised HS rats breeding, LS and MU sampled tissues, AB supervised MRI acquisitions 479 and analyses. NC, ST, LDV and RC wrote the manuscript. All authors contributed to the article and 480 approved the submitted version. 481 Conflict of Interest: Authors declare no competing interests. 482 483 484

485 Ahmed, S.H., Koob, G.F., 1998. Transition from moderate to excessive drug intake: change in hedonic 486 set point. Science 282, 298-300. 487 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint Allen, C., Kuhn, B.N., Cannella, N., Crow, A.D., Roberts, A.T., Lunerti, V., Ubaldi, M., Hardiman, G., 488 Solberg Woods, L.C., Ciccocioppo, R., Kalivas, P.W., Chung, D., 2021. Network -Based Discovery of 489 Opioid Use Vulnerability in Rats Using the Bayesian Stochastic Block Model. Front Psychiatry 12, 490 745468. 491 Asan, L., Falfan-Melgoza, C., Beretta, C.A., Sack, M., Zheng, L., Weber -Fahr, W., Kuner, T., Knabbe, 492 J., 2021. Cellular correlates of gray matter volume changes in magnetic resonance morphometry 493 identified by two-photon microscopy. Sci Rep 11, 4234. 494 Ashburner, J., Friston, K.J., 2000. Voxel-based morphometry--the methods. Neuroimage 11, 805-821. 495 Bach, P., Frischknecht, U., Klinkowski, S., Bungert, M., Karl, D., Vollmert, C., Vollstadt -Klein, S., Lis, 496 S., Kiefer, F., Hermann, D., 2019. Higher Social Rejection Sensitivity in Opioid -Dependent Patients Is 497 Related to Smaller Insula Gray Matter Volume: A Voxel -Based Morphometric Study. Soc Cogn Affect 498 Neurosci 14, 1187-1195. 499 Bach, P., Frischknecht, U., Reinhard, I., Bekier, N., Demirakca, T., Ende, G., Vollstadt-Klein, S., Kiefer, 500 F., Hermann, D., 2021. Impaired working memory performance in opioid-dependent patients is related 501 to reduced insula gray matter volume: a voxel -based morphometric study. Eur Arch Psychiatry Clin 502 Neurosci 271, 813-822. 503 Bellesi, M., de Vivo, L., Chini, M., Gilli, F., Tononi, G., Cirelli, C., 2017. Sleep Loss Promotes Astrocytic 504 Phagocytosis and Microglial Activation in Mouse Cerebral Cortex. J Neurosci 37, 5263-5273. 505 Bland, S.T., Hutchinson, M.R., Maier, S.F., Watkins, L.R., Johnson, K.W., 2009. The glial activation 506 inhibitor AV411 reduces morphine-induced nucleus accumbens dopamine release. Brain Behav Immun 507 23, 492-497. 508 Davis, B.M., Salinas -Navarro, M., Cordeiro, M.F., Moons, L., De Groef, L., 2017. Characterizing 509 microglia activation: a spatial statistics approach to maximize information extraction. Sci Rep 7, 1576. 510 de Guglielmo, G., Kallupi, M., Sedighim, S., Newman, A.H., George, O., 2019. Dopamine D(3) Receptor 511 Antagonism Reverses the Escalation of Oxycodone Self -administration and Decreases Withdrawal -512 Induced Hyperalgesia and Irritability -Like Behavior in Oxycodone -Dependent Heterogeneous Stock 513 Rats. Front Behav Neurosci 13, 292. 514 de Guglielmo, G., Melis, M., De Luca, M.A., Kallupi, M., Li, H.W., Niswender, K., Giordano, A., 515 Senzacqua, M., Somaini, L., Cippitelli, A., Gaitanaris, G., Demopulos, G., Damadzic, R., Tapocik, J., 516 Heilig, M., Ciccocioppo, R., 2015. PPARgamma activation attenuates opioid consumption and 517 modulates mesolimbic dopamine transmission. Neuropsychopharmacology 40, 927-937. 518 De Santis, S., Cosa-Linan, A., Garcia-Hernandez, R., Dmytrenko, L., Vargova, L., Vorisek, I., Stopponi, 519 S., Bach, P., Kirsch, P., Kiefer, F., Ciccocioppo, R., Sykova, E., Moratal, D., Sommer, W.H., Canals, 520 S., 2020. Chronic alcohol consumption alters extracellular space geometry and transmitter diffusion in 521 the brain. Sci Adv 6, eaba0154. 522 Droutman, V., Read, S.J., Bechara, A., 2015. Revisiting the role of the insula in addiction. Trends Cogn 523 Sci 19, 414-420. 524 Edwards, S., Koob, G.F., 2013. Escalation of drug self -administration as a hallmark of persistent 525 addiction liability. Behav Pharmacol 24, 356-362. 526 Eidson, L.N., Murphy, A.Z., 2013. Blockade of Toll -like receptor 4 attenuates morphine tolerance and 527 facilitates the pain relieving properties of morphine. J Neurosci 33, 15952-15963. 528 Everitt, B.J., Robbins, T.W., 2013. From the ventral to the dorsal striatum: devolving views of their roles 529 in drug addiction. Neurosci Biobehav Rev 37, 1946-1954. 530 Fields, J.K., Günther, S., Sundberg, E.J.. 2019. Structural Basis of IL-1 Family Cytokine Signaling. Front 531 Immunol, 20:10:1412. 532 Frischknecht, U., Beckmann, B., Heinrich, M., Kniest, A., Nakovics, H., Kiefer, F., Mann, K., Hermann, 533 D., 2011. The vicious circle of perceived stigmatization, depressiveness, anxiety, and low quality of life 534 in substituted heroin addicts. Eur Addict Res 17, 241-249. 535 Green, J.M., Sundman, M.H., Chou, Y.H., 2022. Opioid -induced microglia reactivity modulates opioid 536 reward, analgesia, and behavior. Neurosci Biobehav Rev 135, 104544. 537 Hahm, S., Lotze, M., Domin, M., Schmidt, S., 2019. The association of health-related quality of life and 538 cerebral gray matter volume in the context of aging: A voxel -based morphometry study with a general 539 population sample. Neuroimage 191, 470-480. 540 Hutchinson, M.R., Northcutt, A.L., Hiranita, T., Wang, X., Lewis, S.S., Thomas, J., van Steeg, K., 541 Kopajtic, T.A., Loram, L.C., Sfregola, C., Galer, E., Miles, N.E., Bland, S.T., Amat, J., Rozeske, R.R., 542 Maslanik, T., Chapman, T.R., Strand, K.A., Fleshner, M., Bachtell, R.K., Somogyi, A.A., Yin, H., Katz, 543 J.L., Rice, K.C., Maier, S.F., Watkins, L.R., 2012. Opioid activation of toll-like receptor 4 contributes to 544 drug reinforcement. J Neurosci 32, 11187-11200. 545 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint Kallupi, M., Carrette, L.L.G., Kononoff, J., Solberg Woods, L.C., Palmer, A.A., Schweitzer, P., George, 546 O., de Guglielmo, G., 2020. Nociceptin attenuates the escalation of oxycodone self -administration by 547 normalizing CeA-GABA transmission in highly addicted rats. Proc Natl Acad Sci U S A 117, 2140-2148. 548 Keifer, O.P., Jr., Hurt, R.C., Gutman, D.A., Keilholz, S.D., Gourley, S.L., Ressler, K.J., 2015. Voxel -549 based morphometry predicts shifts in dendritic spine density and morphology with auditory fear 550 conditioning. Nat Commun 6, 7582. 551 Koob, G.F., Volkow, N.D., 2016. Neurobiology of addiction: a neurocircuitry analysis. Lancet Psychiatry 552 3, 760-773. 553 Kuhn, B.N., Cannella, N., Crow, A.D., Roberts, A.T., Lunerti , V., Allen, C., Nall, R.W., Hardiman, G., 554 Woods, L.C.S., Chung, D., Ciccocioppo, R., Kalivas, P.W., 2022. Novelty -induced locomotor behavior 555 predicts heroin addiction vulnerability in male, but not female, rats. Psychopharmacology (Berl) 239, 556 3605-3620. 557 Lin, H.C., Wang, P.W., Wu, H.C., Ko, C.H., Yang, Y.H., Yen, C.F., 2018. Altered gray matter volume 558 and disrupted functional connectivity of dorsolateral prefrontal cortex in men with heroin dependence. 559 Psychiatry Clin Neurosci 72, 435-444. 560 Liu, H., Hao, Y., Kaneko, Y., Ouyang, X., Zhang, Y., Xu, L., Xue, Z., Liu, Z., 2009. Frontal and cingulate 561 gray matter volume reduction in heroin dependence: optimized voxel -based morphometry. Psychiatry 562 Clin Neurosci 63, 563-568. 563 Milella, M.S., D'Ottavio, G., De Pirro, S., Barra, M., Caprioli, D., Badiani, A., 2023. Heroin and its 564 metabolites: relevance to heroin use disorder. Transl Psychiatry 13, 120. 565 NIDA, 2023. Drug Overdose Death Rates. 566 Paolicelli, R.C. et al., 2022. Microglia states and nomenclature: A field at its crossroads. Neuron. 2022 567 Nov 2;110(21):3458-3483. 568 Paxinos, G., Watson, C., 1998. The rat brain. 569 Perry, J.L., Joseph, J.E., Jiang, Y., Zimmerman, R.S., Kelly, T.H., Darna, M., Huettl, P., Dwoskin, L.P., 570 Bardo, M.T., 2011. Prefrontal cortex and drug abuse vulnerability: translation to prevention and 571 treatment interventions. Brain Res Rev 65, 124-149. 572 Puigdollers, E., Domingo -Salvany, A., Brugal, M.T., Torrens, M., Alvaros, J., Castillo, C., Magri, N., 573 Martin, S., Vazquez, J.M., 2004. Characteristics of heroin addicts entering methadone maintenance 574 treatment: quality of life and gender. Subst Use Misuse 39, 1353-1368. 575 Qiu, Y.W., Jiang, G.H., Su, H.H., Lv, X.F., Tian, J.Z., Li, L.M., Zhuo, F.Z., 2013. The impulsivity behavior 576 is correlated with prefrontal cortex gray matter volume reduction in heroin -dependent individuals. 577 Neurosci Lett 538, 43-48. 578 Richardson, N.R., Roberts, D.C., 1996. Progressive ratio schedules in drug self-administration studies 579 in rats: a method to evaluate reinforcing efficacy. J Neurosci Methods 66, 1-11. 580 Rook, E.J., Huitema, A.D., van den Brink, W., van Ree, J.M., Beijnen, J.H., 2006. Pharmacokinetics 581 and pharmacokinetic variability of heroin and its metabolites: review of the literature. Curr Clin 582 Pharmacol 1, 109-118. 583 Schmidt, A., Vogel, M., Baumgartner, S., Wiesbeck, G.A., Lang, U., Borgwardt, S., Walter, M., 2021. 584 Brain volume changes after long -term injectable opioid treatment: A longitudinal voxel -based 585 morphometry study. Addict Biol 26, e12970. 586 Schnabel, J.A., Tanner, C., Castellano-Smith, A.D., Degenhard, A., Leach, M.O., Hose, D.R., Hill, D.L., 587 Hawkes, D.J., 2003. Validation of nonrigid image registration using finite-element methods: application 588 to breast MR images. IEEE Trans Med Imaging 22, 238-247. 589 Seamans, J.K., Lapish, C.C., Durstewitz, D., 2008. Comparing the prefrontal cortex of rats and 590 primates: insights from electrophysiology. Neurotox Res 14, 249-262. 591 Shaham, Y., Shalev, U., Lu, L., de Wit, H., Stewart, J., 2003. The reinstatement model of drug relapse: 592 history, methodology and major findings. Psychopharmacology (Berl) 168, 3-20. 593 Shi, H., Liang, Z., Chen, J., Li, W., Zhu, J., Li, Y., Ye, J., Zhang, J., Xue, J., Liu, W., Wang, F., Wang, 594 W., Li, Q., He, X., 2020. Gray matter alteration in heroin -dependent men: An atlas -based magnetic 595 resonance imaging study. Psychiatry Res Neuroimaging 304, 111150. 596 Smith, S.M., Jenkinson, M., Woolrich, M.W., Beckmann, C.F., Behrens, T.E., Johansen -Berg, H., 597 Bannister, P.R., De Luca, M., Drobnjak, I., Flitney, D.E., Niazy, R.K., Saunders, J., Vickers, J., Zhang, 598 Y., De Stefano, N., Brady, J.M., Matthews, P.M., 2004. Advances in functional and structural MR image 599 analysis and implementation as FSL. Neuroimage 23 Suppl 1, S208-219. 600 Solberg Woods, L.C., Palmer, A.A., 2019. Using Heterogeneous Stocks for Fine -Mapping Genetically 601 Complex Traits. Methods Mol Biol 2018, 233-247. 602 Stence, N., Waite, M., Dailey, M.E., 2001. Dynamics of microglial activation: a confocal time -lapse 603 analysis in hippocampal slices. Glia 33, 256-266. 604 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint Sun, Y., Liu, L., Feng, J., Yue, W., Lu, L., Fan, Y., Shi, J., 2017. MAOA rs1137070 and heroin addiction 605 interactively alter gray matter volume of the salience network. Sci Rep 7, 45321. 606 Sun, Y., Zhao, L.Y., Wang, G.B., Yue, W.H., He, Y., Shu, N., Lin, Q.X., Wang, F., Li, J.L., Chen, N., 607 Wang, H.M., Kosten, T.R., Feng, J.J., Wang, J., Tang, Y.D., Liu, S.X., Deng, G.F., Diao, G.H., Tan, 608 Y.L., Han, H.B., Lin, L., Shi, J., 2016. ZNF804A variants confer risk for heroin addiction and affect 609 decision making and gray matter volume in heroin abusers. Addict Biol 21, 657-666. 610 Tambalo, S., Peruzzotti -Jametti, L., Rigolio, R., Fiorini, S., Bontempi, P., Mallucci, G., Balzarotti, B., 611 Marmiroli, P., Sbarbati, A., Cavaletti , G., Pluchino, S., Marzola, P., 2015. Functional Magnetic 612 Resonance Imaging of Rats with Experimental Autoimmune Encephalomyelitis Reveals Brain Cortex 613 Remodeling. J Neurosci 35, 10088-10100. 614 Tanda, G., Mereu, M., Hiranita, T., Quarterman, J.C., Coggiano, M., Katz, J.L., 2016. Lack of Specific 615 Involvement of (+) -Naloxone and (+) -Naltrexone on the Reinforcing and Neurochemical Effects of 616 Cocaine and Opioids. Neuropsychopharmacology 41, 2772-2781. 617 Uylings, H.B., Groenewegen, H.J., Kolb, B., 2003. Do rats have a prefrontal cortex? Behav Brain Res 618 146, 3-17. 619 Vendruscolo, L.F., Schlosburg, J.E., Misra, K.K., Chen, S.A., Greenwell, T.N., Koob, G.F., 2011. 620 Escalation patterns of varying periods of heroin access. Pharmacol Biochem Behav 98, 570-574. 621 Wang, L., Zou, F., Zhai, T., Lei, Y., Tan, S., Jin, X., Ye, E., Shao, Y., Yang, Y., Yang, Z., 2016. Abnormal 622 gray matter volume and resting -state functional connectivity in former heroin -dependent individuals 623 abstinent for multiple years. Addict Biol 21, 646-656. 624 Wang, X., Li, B., Zhou, X., Liao, Y., Tang, J., Liu, T., Hu, D., Hao, W., 2012. Changes in brain gray 625 matter in abstinent heroin addicts. Drug Alcohol Depend 126, 304-308. 626 Yazdi, A.S., Ghoreschi, K., 2016. The Interleukin-1 Family, Adv Exp Med Biol 941, 21-29. 627 Yen, C.N., Wang, C.S., Wang, T.Y., Chen, H.F., Chang, H.C., 2011. Quality of life and its correlates 628 among heroin users in Taiwan. Kaohsiung J Med Sci 27, 177-183. 629 Yue, K., Tanda, G., Katz, J.L., Zanettini, C., 2020. A further assessment of a role for Toll -like receptor 630 4 in the reinforcing and reinstating effects of opioids. Behav Pharmacol 31, 186-195. 631 632 633 634 635 636 637 638 639 640 641 642 643 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint Figure legends 644 645 Figure 1. A) Flowchart of the longitudinal MRI experiment timeline. B) Heroin intake during ShA and 646 LgA self-administration training. Rats showed stable heroin intake during the 1h ShA phase (left side of 647 dashed line). During the 12h LgA phase heroin intake escalated over time, becoming statistically higher 648 than the first two LgA session starting from the 6th session, when it reached a plateau. C) Heroin intake 649 during the 1h ShA sessions and the 1st hour of the 12h LgA sessions. During the LgA phase the 1st hour 650 intake increased over time becoming statistically different from the last two days of ShA from the 6 th 651 session. D) Escalation of heroin intake expressed as the difference between the average intake at four 652 key time points: first three and last three ShA sessions (early and late ShA respectively), the first three 653 and last three LgA sessions (early LgA and late LgA respectively). The intake in late LgA was 654 significantly higher than the other three time points. E) Motivation for heroin expressed by the break 655 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint point reached in PR sessions run during the ShA and LgA phase. Rats showed increased break point 656 after LgA training. F) Heroin primed reinstatement of heroin seeking in a within session extinction -657 reinstatement protocol. Heroin priming reinstated active lever pressing, which increased over time 658 compared to the last 30 minutes of extinction (upper panel). Inactive lever remained constantly low and 659 was not statistically different from extinction (lower panel). Data are expressed as mean ± SEM. 660 Significant differences: A-C) **p<0.01; D) ***p<0.001 vs ShA; E) *** p<0.001 vs extinction (ext), °p<0.05, 661 °°p<0.01 and °°°p<0.001 vs previous reinstatement time point. 662 663 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint 664 665 Figure 2 . Spatial distribution of negative modulation of grey matter volume after drug exposure. 666 Significant reduction of grey matter volume is thresholded at 1 - pvalue > 0.99, TFCE corrected. 667 668 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint 669 Figure 3: Voxelwise correlation maps of drug seeking and taking behavioral readouts against grey 670 matter volume MRI2 - MRI1 difference. Significant negative correlations are thresholded at 1 - pvalue 671 > 0.95, TFCE corrected. 672 673 674 675 676 677 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint 678 Figure 4. Representative image of A) NeuN staining and B) PV staining taken in the Insular Cortex and 679 medial Prefrontal Cortex of a heroin treated rat, respectively; C) Iba1 staining taken in the Insular Cortex 680 of a control rat. D) Heroin experienced rats showed higher PV-positive cell density in the Primary Motor 681 cortex. A-C: scale bars = 20 µm; D: whiskers represent mean ± SEM, statistical significance **p < 0.01. 682 683 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint 684 Figure 5: Cumulative distribution of Iba -1 positive Cell Area, Cell Perimeter and Area/Perimeter ratio. 685 A-C) In the mPFC of heroin experienced rats there was a leftward shift in Cell Area ( A) and Perimeter 686 (B), and a rightward shift of Area/Perimeter ratio (C). D-F) In the Insula of heroin experienced rats there 687 was a leftward shift in Cell Area (D) and Perimeter (E), and a rightward shift of Area/Perimeter ratio (F). 688 G-I) In the Primary Motor cortex of heroin experienced rats there was a leftward shift in Cell Area ( G) 689 and Perimeter (H), but no difference in the Area/Perimeter ratio (F). Statistical significance: *p<0.05 and 690 ****p<0.0001 between groups. 691 692 693 FIGURE 6: Examples of automatically segmented microglial cells showing low ( A), medium ( B) and 694 highly (C) ramified phenotypes, scale bar 20 μm. 695 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted February 28, 2024. ; https://doi.org/10.1101/2024.02.26.582024doi: bioRxiv preprint