All experiments were conducted in accordance with Marshall University's Institutional Animal Care and Use Committee (IACUC) guidelines (W.C.R. protocols 696 and 697). Adult C57/Bl6 mice heterozygous for Cacna2d1 (i.e. α2δ-1 +/-) (31) were a kind gift from Dr. Cagla Eroglu (Duke University). The mice were bred onsite at the Marshall University Animal Resource Facility. Virgin α2δ-1 +/- females and α2δ-1 +/- males were set up as mating pairs. Upon visual confirmation of the vaginal plug (established as embryonic day 0 [E0]), males and females were separated. On E6, pregnant females were given access to 1 ml of a 1:1 sweetened condensed milk/water solution served in a plastic 35 mm dish. Starting on E7, pregnant females were given free access to a once daily 1 ml solution of 1:1 condensed milk/water containing either pharmaceutical grade buprenorphine hydrochloride (CIII) (5 mg/kg; Spectrum Chemical, Gardena, CA), gabapentin (30 mg/kg; Spectrum), a combination of both drug doses, or vehicle control. Drug doses were calculated based on the weight of pregnant females on E7. Daily dosing continued through the birth of pups and ended on postnatal day 11 (P11). All animals were given ad libitum food (standard mouse chow milled on site) and water.

For wild-type (WT) mice, 7 pups (3 male and 4 female) from vehicle control treated litters, 6 (2 male, 4 female) from gabapentin treated litters, 6 (3 male, 3 female) from buprenorphine treated litters, and 7 (4 male, 3 female) from combined buprenorphine and gabapentin treated litters were included. For α2δ-1 +/- mice, 11 pups (5 male, 6 female) from vehicle control treated litters, 7 (3 male, 4 female) from gabapentin treated litters, 7 (4 male, 3 female) from buprenorphine treated litters, and 7 (4 male, 3 female) from combined buprenorphine and gabapentin treated litters were included.

The brains of newborn WT pups from dams prenatally treated with 5 mg/kg buprenorphine, 30 mg/kg gabapentin, or vehicle control (2 litters per treatment) were isolated and then shipped on dry ice to the Pharmaceutical Sciences Research Institute at the McWhorter School of Pharmacy (Samford University, Birmingham, AL) for drug concentration analysis. Mouse brain samples were homogenized 1:5 with 5 mM ammonium acetate buffer using an Omni THQ homogenizer (Atlanta, GA). Gabapentin calibration standards, blanks and QCs were prepared by spiking 20 μl of blank control brain homogenate with the appropriate amount of gabapentin to achieve concentrations in ranging from 25–10,000 ng/ml. Internal standard (1 ml of 100 ng/ml Gabapentin-d4 in methanol) was added to precipitate the proteins. After centrifugation for 5 min at 13,000 rpm, the supernatant was transferred to glass tubes and the solvent was evaporated under nitrogen at 40°C. The samples were then redissolved in 200 μl of 95/5 10 mM ammonium formate/methanol with 0.1% formic acid and transferred to limited volume autosampler vials and analyzed in positive ion mode by liquid chromatography with tandem mass spectrometry (LC/MS/MS).

Buprenorphine calibration standards, blanks, and quality controls (QCs) were prepared by spiking 100 μl of blank control brain homogenate with the appropriate amount of buprenorphine to achieve concentrations ranging from 0.2–200 ng/ml. Standards, blanks, QCs and samples were spiked with internal standard (10 μl of 50 ng/ml Terfenadine in acetonitrile). Acetonitrile (1 ml) was added to precipitate the proteins. After centrifugation for 5 min at 13,000 rpm, the supernatant was transferred to glass tubes and the solvent was evaporated under nitrogen at 50°C. The samples were then redissolved in 200 μl of 50/50 5 mM ammonium acetate/acetonitrile and transferred to limited volume autosampler vials and analyzed in positive ion mode by LC/MS/MS.

The LC/MS/MS system consisted of Shimadzu system (Columbia, MD) equipped with LC20-AD dual HLPC pumps, an SIL20-AC HT autosampler, and a DGU-20A2 in-line degasser. Detection was performed using an Applied BioSystems 4000 QTRAP (Applied Biosystems, Foster City, CA) triple quadrupole mass spectrometer operated in the positive ion mode utilizing electrospray ionization. Mass calibration, data acquisition and quantitation were performed using Applied Biosystem Analyst 1.6.2 software (Applied Biosystems).

Separation of gabapentin and the gabapentin-d4 from the brain homogenate matrix was achieved from a 10 μl injection of the samples using a Phenomenex Polar C18, 100 X 2 mm 5 μm particle column. The mobile phase was delivered at a flow rate of 400 μl/min using a gradient elution profile consisting of 5 mM ammonium acetate (A) and acetonitrile with 0.1% formic acid (B). The analyte and internal standard were detected using multiple reaction monitoring (MRM) for the following transitions: Gabapentin (m/z 172.2 → 137.0), Gabapentin-d4 (m/z 176.2 → 141.0).

Separation of buprenorphine and terfenadine from the brain homogenate matrix was achieved from a 10 μl injection of the samples using a Phenomenex Luna C18, 100 X 2 mm 5 μm particle column. The mobile phase was delivered at a flow rate of 400 μl/min using a gradient elution profile consisting of 5 mM ammonium acetate with 0.1% formic acid (A) and acetonitrile with 0.1% formic acid (B). The analyte and internal standard were detected using multiple reaction monitoring (MRM) for the following transitions: Buprenorphine (m/z 468.4 → 396.1), Terfenadine (m/z 472.4 → 436.3). Drug concentrations are summarized in Table 1, with extended summary data available online at Harvard Dataverse (https://doi.org/10.7910/DVN/QKL6OP).

At P7, the tails of the pups were clipped and collected. Toe pads were tattooed (AIMS NEO9 Animal Tattoo System, Hornell, NY) for later identification. Tissue digestion was performed on the clipped tails and DNA was isolated for PCR amplification and genotyping with a 2% agarose gel. Pups were identified as either α2δ-1 +/- heterozygous, α2δ-1 +/+ wildtype (WT), or α2δ-1 -/- knockout (KO) using the following primers: F1 (WT forward, 5′-TCTCAGTTACAAGACTATGTGG-3′), F3 (KO forward, 5′-GGCTGTGTCCTTATTTATGG-3′), and LAF-Test (reverse, 5′-AGTAGGAGAAGGTACAATCGGC-3′) (Integrated DNA Technologies, Coralville, IA).

At P21, the dam and pups were placed on a scale to be weighed (group data is summarized in Table 2) before being anesthetized with tribromoethanol/Avertin (250 mg/kg). Cardiac perfusion was performed, first with 0.24 μg/ml heparin salt in 0.1M phosphate buffered saline (PBS) for 2 min followed by 4% paraformaldehyde (PFA) for 2–5 min at a flow rate of 5 ml/min. The brains of the dam and pups were extracted and submerged in 4% PFA at 4°C for 24 h. After 24 h, the brains were rinsed in PBS and then placed in 30% sucrose:PBS solution for 48 h. Brains in sucrose solution were frozen within plastic embedding molds (Cat. #70182, Electron Microscopy Sciences, Hatfield, PA) in 2:1 sucrose solution:Tissue Freezing Medium (Cat. #72592-G, Electron Microscopy Sciences) and stored at −80°C. Brains were then cryosectioned using a Leica CM 1950 (Leica, Wetzlar, Germany) to 20 μm slices and stored in 50% glycerol:Tris-Buffered Saline (TBS) at −20°C.

Three independent coronal sections per mouse containing the PFC (bregma, 2.96 to 2.58 mm; interaural, 6.76–6.38 mm) and three sections per mouse containing the ACC and NAC core/shell (bregma, 1.42–0.86 mm; interaural, 5.22–4.66 mm) were used for analyses (32). After selection, brain slices were washed in TBS + 0.2% Triton X-10% (Cat. #11332481001, Roche Diagnostics, Mannheim, Germany) w/v (TBST) solution at room temperature. The slices were then placed in a 5% normal goat serum (NGS; Cat. #005-000-121, Jackson Immuno Research Laboratories Inc., West Grove, PA):TBST solution for 1 h to block nonspecific binding sites. After blocking was completed, the slices were then placed in 5% NGS:TBST containing primary antibodies against both pre- and post-synaptic proteins. Guinea pig anti-vesicular glutamate transporter 1 (VGlut1; Cat. #AB5905, EMD Millipore, Burlington, MA) at 1:2,000 dilution was used to identify excitatory glutamatergic presynaptic axonal regions while rabbit anti-post synaptic density protein 95 (PSD95; Cat. #51-6900, Invitrogen, Carlsbad, CA) at 1:300 dilution was used to identify postsynaptic dendritic regions. To label inhibitory GABAergic synapses, guinea pig anti-vesicular GABA transporter (VGAT; Cat. #131004, Synaptic Systems, Göttingen, Germany) at 1:1,000 dilution was used to identify presynaptic axonal regions and rabbit anti-gephyrin (Cat. #147002, Synaptic Systems) at 1:1,000 dilution was used to identify postsynaptic dendritic regions. Slices were incubated overnight on an orbital shaker at 4°C.

After overnight incubation was completed, brain slices were removed from primary antibody solution and washed with TBST. Slices were then placed in 5% NGS:TBST containing fluorescent secondary antibodies against the primary antibodies. Alexa Fluor IgG (H+L) 594 goat anti-guinea pig (Cat. #A11076, Invitrogen, Carlsbad, CA) at 1:200 dilution was used to stain against presynaptic antibodies (either VGlut1 for excitatory or VGAT for inhibitory) and Alexa Fluor IgG (H+L) 488 goat anti-rabbit (Cat. #A11034, Invitrogen) at 1:200 dilution was used to stain against postsynaptic antibodies (either PSD95 for excitatory or gephyrin for inhibitory). Slices were incubated for 2 h at room temperature in darkness. After incubation was complete, the slices were washed with TBST.

After the final washing step, the slices were rinsed once in 2/3 TBS:1/3 dH₂0 solution. Then, the slices were transferred onto glass slides and excess liquid was suctioned off with a pipette. 1 drop of mounting media containing DAPI (Cat. #H-1200; VectaShield, Burlingame, CA) was applied to each slice prior to placement of coverslip and sealing with clear nail polish. The glass slides were allowed to dry overnight in darkness before being stored at −20°C.

Images of the slides were captured using the Leica SP5 Confocal Fluorescence Microscope housed and maintained by the Marshall University Molecular Biological and Imaging Core. Argon visible light laser (at 30% final filter power) was used with emission windows set at 493–550 nm for blue excitation and 595–647 nm for green excitation. Leica LAS AF software was used to capture 5 μm z-stacks with 15 steps (0.33 μm distance between each step) within the brain regions of interest (ROI) across all treatment groups (imaged area/scan = 19,036 μm²; 63× oil objective, 1.4 NA).

The saved images were analyzed using ImageJ (NIH) software. The custom plug-in “ProjectZ_Triple” (available by request) was used to average and convert the 15 separate images of each z-stack into 5 separate maximum projections, each representing a 1 μm “mini-stack”. Puncta Analyzer (Dr. Cagla Eroglu, Duke University, Durham, NC) was then used to quantify the number of discrete puncta for the 493–550 nm wavelength channel (corresponding to PSD95 in brain slices stained for excitatory synapses or Gephyrin for inhibitory synapses), the 595–647 nm wavelength channel (corresponding to VGlut1 in brain slices stained for excitatory synapses or VGAT for inhibitory synapses), and the co-localized puncta, where puncta from both channels overlapped within 4 pixels of each other. These regions of co-localization of pre- and post-synaptic markers were designated as synapses (33). The counted synapses were compiled into spreadsheets (Microsoft Excel, Redmond, WA) and analyzed with GraphPad Prism (San Diego, CA).

For body mass comparisons (Supplementary Table 1), dams belonging to different treatment groups were compared via One-Way ANOVA with Tukey's multiple comparisons post hoc test. For pups, a Three-Way ANOVA (treatment x genotype x sex) was performed with Tukey's multiple comparisons post hoc test. The D'Agostino and Pearson omnibus normality test (Supplementary Table 2) confirmed that most synapse counts were not normally distributed; therefore, statistical differences were analyzed using the nonparametric Kruskal-Wallis test to detect differences in mean number of synapses amongst drug treatments or between genotypes. Post-hoc Dunn's multiple comparisons tests were performed to detect significant differences (p < 0.05) amongst the mean values of each treatment group (Supplementary Tables 3, 4). Analyses were performed using GraphPad Prism.

Each day during the dosing period, pregnant mice were administered drug treatments as 1 ml of a 1:1 sweetened condensed milk/water solution served in a plastic dish. Dishes from the previous day were always observed to be empty, confirming that the dams routinely consumed the entirety of the treatment solution. To confirm drug passage from the dam to the pups, the brains of newborn pups from litters prenatally treated with buprenorphine or gabapentin were isolated, lysed, and subjected to liquid chromatography-tandem mass spectrometry to determine the brain tissue concentrations of each drug. Both buprenorphine and gabapentin were detected in the brains of the prenatally exposed pups (Table 1), prompting further investigation of brain connectivity changes as a result of the presence of these compounds.

In order to produce the required number of WT and α2δ-1 heterozygous (+/–) pups for IHC in this study, the following litter numbers were required: 4 litters dosed with vehicle control, 5 litters dosed with 30mg/kg gabapentin, 4 litters dosed with 5mg/kg buprenorphine, and 5 litters dosed with both drugs. Immediately before anesthetization and perfusion at P21, dams and all pups from each litter were weighed and their weights were recorded. The mean masses for the dams of the different litters and the pups in each sex-genotype-treatment group were calculated (Table 2). The mean body masses for dams revealed a significant main effect of treatment, F(3, 14) = 50.21, p < 0.0001. Dams treated with buprenorphine (p = 0.0122), gabapentin (p = 0.0006), and buprenorphine + gabapentin (p < 0.0001) were all significantly lower than the mean body mass of the vehicle control dams. Significant differences were also observed among the various pup groupings across treatment [F(3, 42) = 140.4, p < 0.0001], genotype [F(1, 42) = 12.66, p =0.0009], treatment × sex [F(3, 42) = 4.487, p = 0.0081], and genotype × sex [F(1, 42) = 5.627, p = 0.0223]. No significance was found for the main effect of sex [F(1, 42) = 0.5268, p = 0.4720] or the interactions of treatment × genotype [F(3, 42) = 1.301, p = 0.2866] or treatment × genotype × sex [F(3, 42) = 2.039, p = 0.1229]. Full multiple comparisons are listed in Supplementary Table 1. Of note, key differences between treatments that reached significance included: WT male vehicle vs. either WT male buprenorphine (Bup) or WT male buprenorphine plus gabapentin (BupGBP), WT female vehicle vs. WT female BupGBP; heterozygous (Het) male vehicle vs. either Het male Bup or Het male BupGBP, and Het female vehicle vs. Het female BupGBP.

We next wanted to determine whether excitatory synaptic development within the mesolimbic dopamine pathway was significantly impacted by the presence of buprenorphine and/or gabapentin in utero. Fluorescence IHC followed by confocal imaging was used to distinguish and quantify excitatory glutamatergic synapses within three different brain regions of WT mouse pups at P21 (Figure 2). Synapses were identified by the co-localization of presynaptic VGlut1 and postsynaptic PSD95 (Figure 2A). This type of imaging-based synapse analysis relies on the fact that these markers are expressed in completely different neuronal compartments (i.e., VGlut1 in axons, PSD95 in dendrites), and would only appear co-localized in confocal microscopy when directly opposed at sites of synaptic contact. Indeed, rotating the channels out of alignment with each other results in a dramatic decrease in co-localized puncta (Supplementary Figure 1). Beginning with excitatory glutamatergic synapses within the ACC (Figure 2B), we observed a significant main effect between groups, H = 99.00, p < 0.0001 (full list of multiple comparisons reported in Supplementary Table 3), as mice treated with either buprenorphine or buprenorphine in combination with gabapentin had a significantly increased mean number of synapses than mice treated with vehicle control. Interestingly, gabapentin treatment on its own did not significantly impact excitatory synapse number compared to vehicle, despite its known role as an inhibitor of prominent synaptogenic pathways (34). Similar results were also observed within the NAC (Figure 2C; H = 119.2, p < 0.0001), with buprenorphine and buprenorphine together with gabapentin having significantly more synapses than vehicle, with no significant effect of gabapentin alone. However, in the NAC, the combination drug treatment group also exhibited a significantly higher number of synapses than the gabapentin-only group. In contrast, within the PFC (Figure 2D; H = 165.0, p < 0.0001), no significant differences were observed among any of the treatment groups when compared to vehicle.

The same IHC procedure used for glutamatergic excitatory synapses was next used to stain for GABAergic inhibitory synapses (identified by the co-localization of presynaptic VGAT and postsynaptic gephyrin; Figure 3A) within the same brain regions of the previously analyzed WT pups. As with excitatory synapses, we observed a main effect of treatment, H = 196.5, p < 0.0001 (full list of multiple comparisons reported in Supplementary Table 4), for inhibitory synapses within the ACC (Figure 3B). A significant decrease in mean number of inhibitory synapses was observed in the combination drug treated mice compared to all other treatment groups. As with excitatory synapses, similar observations were made in the NAC (Figure 3C; H = 219.8, p < 0.0001), with the buprenorphine plus gabapentin group having significantly fewer inhibitory synapses than the other groups. The PFC, however, was the only region that exhibited an increase in inhibitory synapse number with any of the drug treatments (Figure 3D; H = 121.8, p < 0.0001), with the buprenorphine treatment group having significantly more VGAT/gephyrin co-localized puncta than the vehicle control group. Yet, as in the ACC and NAC, the combination treated mice had significantly fewer synapses within the PFC than mice from either of the single drug treatment groups, but not compared to vehicle control.

Gabapentin has previously been shown to inhibit excitatory synapse formation by binding to the neuronal calcium channel subunit α2δ-1 (34), and this mechanism is proposed to underlie the efficacy of gabapentin in the alleviation of symptoms in both epilepsy and neuropathic pain (35, 36). We previously showed deficits in synaptic connectivity in the brains of young mice lacking α2δ-1 (31). To determine whether α2δ-1-mediated synaptic development is impacted by early life drug exposure, we expanded our synapse analysis to compare our WT synapse numbers with those from mice that were haploinsufficient for α2δ-1. Comparing excitatory synaptic development, within the ACC (Figure 4A; H = 99.00, p < 0.0001), only the gabapentin treated α2δ-1 +/- mice showed a significantly greater number of synapses compared with WT mice that received the same treatment. A similar gabapentin-induced increase was observed within the NAC (Figure 4B; H = 119.2, p < 0.0001). The same held true for the α2δ-1 +/- PFC (Figure 4C; H = 165.0, p < 0.0001), which also exhibited more synapses than WT mice with buprenorphine treatment.

Compared to excitatory synaptic development, much less is known about the roles of α2δ-1 and gabapentin in the formation of inhibitory circuits. We concluded our analyses by comparing mean inhibitory synapse numbers in α2δ-1 +/- mice to WT mice across the 4 different treatment groups (Figure 5). In the vehicle control, gabapentin, and buprenorphine groups, α2δ-1 +/- mice had significantly fewer synapses within the ACC compared to their WT counterparts (Figure 5A; H = 196.5, p < 0.0001). Only in the combination treatment group did we observe a significantly higher number of synapses in the ACC of α2δ-1 +/- mice compared to WT. Within the NAC (Figure 5B; H = 219.8, p < 0.0001), a similar trend was observed among the vehicle control, gabapentin, and buprenorphine groups as was seen in the ACC, with significantly fewer VGAT/gephyrin co-localized synapses in α2δ-1 +/- mice than WT. However, in the NAC, the α2δ-1 +/- mice in the combination treatment group did not differ significantly from the WT mice. Finally, within the PFC (Figure 5C; H = 121.8, p < 0.0001), the most dramatic significant difference was observed in the buprenorphine treatment group, with α2δ-1 +/- mice showing significantly fewer inhibitory synapses than WT mice. Taken together, these results indicate that α2δ-1 plays an important regulatory role in both excitatory and inhibitory synaptic development in the context of early life drug exposure.

The dosing for buprenorphine in this study was in line with a study by Martin et al. that showed 5 mg/kg to be capable of producing significant increases in pain threshold in rats (44). The gabapentin dosage was chosen based on the finding that 30 mg/kg gabapentin was sufficient to produce analgesia in mice (45). The dosing schedule, starting on E7 and ending on P11, was chosen to correspond to a period of significant synaptogenesis within mice (46–48), a period which also roughly correlates to the second trimester of human fetal development (49, 50). At birth, a mouse pup is considered to be at a developmental stage similar to that of a human fetus in the late second trimester. Despite the fact that newborn mouse pups are no longer receiving direct exposure to drugs of abuse via the placental blood supply, there were still precedents in the literature for postnatal drug transference to the brains of pups being nursed by drug-exposed dams. Kongstorp et al. demonstrated that buprenorphine accumulates and remains in the brain tissue of newborn rodents for several days following birth (51). In addition, both buprenorphine and gabapentin have been shown to pass from mother to infant via breast milk, albeit in low concentrations compared to the mothers' blood plasma levels (52–54). We confirmed the presence of both buprenorphine and gabapentin in the brains of prenatal drug exposed pups, though the levels in our study were admittedly at the lower ends of the ranges reported in the referenced works. Regardless, the results of this study show that the chosen treatment paradigm was capable of disrupting normal CNS synaptic development in mouse pups. The fact that we detected relatively lower concentrations in our animals may even suggest that the observed deficits may have been further exacerbated if the accumulated drug levels had been higher.

Unexpectedly, single drug treatment with gabapentin did not result in any significant differences in excitatory or inhibitory synapses within any of the examined brain regions in the WT. This finding would appear to contradict the findings of previous studies showing that gabapentin inhibits normal excitatory synapse development by interfering with pathways mediated by α2δ-1. Though originally designed as a structural analog for the neurotransmitter GABA, gabapentin does not bind to GABA _A or GABA _B receptors. It is instead proposed that gabapentin decreases neurotransmitter release from the presynaptic terminal by inhibiting Ca⁺² influx through L-type channels (55, 56). Gabapentin performs this action by binding to α2δ-1 on the surface of neurons at the site of synaptic terminals (35, 57). α2δ-1 is also the binding site for a class of astrocyte-secreted extracellular matrix glycoproteins known as thrombospondins (TSPs), which promote synapse formation and maturation during early development (31, 34). It may be possible that, despite interference by gabapentin in normal TSP/α2δ-1 mediated synaptic development, the net level of synaptogenesis is maintained via other signaling mechanisms as a means to compensate for the gabapentin-induced decrease in glutamate release.

In an attempt to elucidate the impact of α2δ-1 on synaptic development in the drug exposed brain, the original goal of this study was to mate α2δ-1 +/- males and females to produce pups of all three possible α2δ-1 genotypes, i.e. α2δ-1 +/- (Het), α2δ-1 +/+ (WT), and α2δ-1 -/- (knockout; KO). However, after several attempts to breed α2δ-1 KO pups, we found that such pups were either not being born or did not survive to age P21, particularly in the buprenorphine or gabapentin treatment groups. Given the widespread presence of α2δ-1 throughout the mouse CNS (58, 59), skeletal muscle (60, 61), and cardiac muscle (62), the global knockout of this Ca²⁺ channel subunit, in combination with exposure to drugs of abuse during fetal development, may have been potentially lethal.

Despite the difficulty in generating the necessary number of drug exposed α2δ-1 KO pups for this study, we still observed significant drug treatment effects on synaptic development with the loss of just a single copy of α2δ-1, which may indicate that astrocyte-mediated synaptogenic pathways were disrupted with the drug treatment paradigms. Indeed, it has been previously shown that opioids can directly dysregulate the expression and secretion of synaptogenic TSP by astrocytes (63, 64). Far from being a passive bystander in the CNS, astrocytes have been increasingly shown to influence and maintain the excitatory/inhibitory balance necessary for homeostasis (65). Not only are astrocytes able to mediate both excitatory (47) and inhibitory (66) synaptogenesis, they are also able to mediate synapse elimination via phagocytosis mediated by astrocytic cell surface receptors such as MEGF10 and MERTK (67). Outside of synapse regulation, astrocytes are also capable of both uptake (68–70) and release (71, 72) of glutamate and GABA. μ-opioid receptors have been confirmed on the surface of astrocytes within the hippocampus, NAC, and VTA (73), while morphine and other μ-opioid receptor agonists have been shown to alter DNA synthesis (74) and increase expression of glial fibrillary acidic protein (GFAP), an indicator of activation and reactivity, in astrocytes (75). Greater understanding of the role of astrocytes in maintaining excitatory/inhibitory synaptic balance, coupled with increased awareness of the effects of drugs of abuse on astrocytes during development, may facilitate novel insight into how neurological function may be impacted in cases of early life drug exposure.