Adverse effects of Δ9-tetrahydrocannabinol on neuronal bioenergetics during postnatal development

Ongoing societal changes in views on the medical and recreational roles of cannabis increased the use of concentrated plant extracts with a Δ9-tetrahydrocannabinol (THC) content of more than 90%. Even though prenatal THC exposure is widely considered adverse for neuronal development, equivalent experimental data for young age cohorts are largely lacking. Here, we administered plant-derived THC (1 or 5 mg/kg) to mice daily during P5–P16 and P5–P35 and monitored its effects on hippocampal neuronal survival and specification by high-resolution imaging and iTRAQ proteomics, respectively. We found that THC indiscriminately affects pyramidal cells and both cannabinoid receptor 1+ (CB1R)+ and CB1R– interneurons by P16. THC particularly disrupted the expression of mitochondrial proteins (complexes I–IV), a change that had persisted even 4 months after the end of drug exposure. This was reflected by a THC-induced loss of membrane integrity occluding mitochondrial respiration and could be partially or completely rescued by pH stabilization, antioxidants, bypassed glycolysis, and targeting either mitochondrial soluble adenylyl cyclase or the mitochondrial voltage-dependent anion channel. Overall, THC exposure during infancy induces significant and long-lasting reorganization of neuronal circuits through mechanisms that, in large part, render cellular bioenergetics insufficient to sustain key developmental processes in otherwise healthy neurons.


Supplementary methods
Drugs Plant-derived, highly-purified THC (pTHC; 314.46 mg/mol, 95% purity) was provided in ethanol by GW Pharmaceuticals. Synthetic THC (sTHC) was obtained from THC Pharm (diluted in methanol with 98.9% purity or ethanol with 99.4% purity) and from Lipomed (in ethanol with > 97% purity). For drug preparations, ethanol/methanol was evaporated in a N2 gas flow to limit oxidization. sTHC was subsequently diluted and aliquoted in dimethyl sulfoxide (DMSO; Sigma) and stored at -80°C. For in vitro applications, the stock solutions were mixed in an ultrasonic bath before being further diluted in corresponding concentrations of cell culture media for experimental use. pTHC was either diluted in DMSO for in vitro and pharmacology applications following the same strategy as for sTHC or further diluted and aliquoted in ethanol for the preparation of in vivo applications. For in vivo application, the ethanol of pTHC was evaporated in a N2 gas flow and the substance was diluted freshly in a mixture of physiological saline (sterile, Mini-Plasco ® , Braun) and 3 v/v% Tween 80 (Sigma) as emulsifier. After ultrasonication to reach complete dissolution, a volume of 100 μl was injected subcutaneously (s.c.) immediately on each day from P5-16 or P5-35 (see below). pTHC at 1 or 5 mg/kg bodyweight was used (1), which when employing the FDA allometric scaling guidance (https://www.fda.gov/media/72309/download) corresponds to 24.3 mg/60 kg bodyweight. This dose is in the range of exposure for regular cannabis smokers when considering 5-15% THC content in regular herbal cannabis preparations (2), which results in 50-150 mg THC/g cannabis. Even higher THC intake seems relevant for cannabis resin and high potency extracts.
The control group received an equivalent volume of the vehicle solution daily. Drug doses and injection volumes were adjusted to the bodyweight of the animals, which was monitored every other day (data not shown). Notably, juvenile offspring was injected directly thus this model was neither intended to test maternal-to-offspring drug transfer nor bioavailability upon inhalation or digestion.
For ligand binding at CB1Rs, pTHC and sTHC in DMSO were ultrasonicated and diluted in 50 mM  For immunohistochemistry, sections were rinsed in 0.1M PB before being exposed to a cocktail of 5% normal donkey serum (NDS; Jackson ImmunoResearch), 1% BSA (Sigma), and 0.3% Triton X-100 (Sigma) in PB at 22-24 o C for 1h to quench non-specific immunoreactivity. Tissues were then exposed to primary antibodies (

Behavior
We tested spontaneous animal behaviors a day prior to sample collection for iTRAQ proteomics to confirm no acute THC effects. We chose the elevated plus-maze test that is frequently used to assess anxiety-like phenotypes brought about by psychoactive drugs (9). Each session lasted for 4 min under dimmed lighting conditions. Video records were analyzed off-line (Panlab, Smart Video Tracking Software 3.0.03).
iTRAQ proteomics 5 Male C57Bl6/J mice were treated with pTHC (1 mg/kg or 5 mg/kg) or vehicle from P5 to P35. At P25, all animals were weaned, but remained group-housed for the duration of the study. Food and water were available ad libitum. The first batch of animals was sacrificed by decapitation on P48 ('midadolescence'; n = 5/group), whereas the second batch was processed on P120 ('adulthood'; n = 5 per group). Both hippocampi were dissected, flash-frozen in liquid N2 separately and stored at -80 °C.
iTRAQ was performed on the right hippocampus of each subject according to published protocols (1, first filtered for a confidence interval above 99% and thereupon selected using the ProteinPilot software.
This procedure was performed separately for each of the iTRAQ layouts. Subsequently, protein levels above the threshold were filtered for common proteins which occurred in each of the chips. At P48, 2,076 proteins were consistently identified. For P120, 3,542 commonly-occurring proteins were found.
All logscale data-points were normalized to 'vehicle 1'. The description and classification of proteins is based on the gene ontology (GO) database www.uniprot.org.

Western blotting
The left hippocampus of each mouse was lysed and homogenized in HEM + -buffer (25mM HEPES, 1mM EDTA, 6mM MgCl2, 1mM DTT, 1x protease inhibitor) before preparing membrane and cytosolic fractions. Initial centrifugation at 500 g at 4 °C for 5 min was performed to remove nuclei and cellular debris. Cytosolic fractions were obtained after additional centrifugation at 14,400 g at 4°C for 30 min.
Membrane fractions were isolated after an additional centrifugation step (14, Neurons were isolated from E14.5 C57Bl6/J mouse cortices (11) and grown in Neurobasal A medium (GIBCO, Life Technologies) supplemented with L-glutamine (2 mM), B27 supplement (2%) and penicillin-streptomycin (1%). Cells were plated in 96-well plates (25,000 cells/well; 0030730119, Eppendorf) pre-treated with poly-D-lysine in 0.1M borate buffer overnight. Drug challenges were initiated on day 2 in vitro (DIV). Antagonists and rescue protocols were applied 30 min before THC application. An IncuCyte live-cell imaging device (Essen Bioscience), itself placed in an incubator with stable 5% CO2 intake and temperature control at 37 °C, as used for live cell (including neurite) imaging.
Phase contrast images were taken at hourly intervals (or every other hour). 'The area of cell viability', that is the surface area occupied by cell bodies (mm 2 /mm 2 ) was taken as a measure of cell viability. The growth rate of neurites was obtained by using the 'neurite length' mask (mm/mm 2 ). Optionally, neurite branching was also defined as branch points per surface area (n/mm 2 ; data not shown). All measures were optimized in preliminary experiments, including drug doses and treatment paradigms. In vitro imaging of antibody-labeled primary neurons was conducted on an LSM 880 confocal laser scanning microscope (ZEISS), while super-resolution microscopy of TOM20-stained mitochondria was performed on an ELYRA PS.1 system (Zeiss) as previously described (12).
Abbreviations: E, embryonic day; LA, lateral amygdaloid nucleus; P, postnatal day. were expressed as means ± s.e.m. Colored asterisks label statistically significant differences of the respective treatment groups vs. control; *p < 0.05; two-way ANOVA followed by Bonferroni post-hoc correction. (D) All THC preparations (10 M) eliminated the mitochondrial membrane potential. Data are after 1h exposure and were expressed as means ± s.d. Colored circles denote independent biological replicates. **p < 0.05 vs. control; two-way ANOVA for (B,C) and one-way ANOVA for (D), both followed by Bonferroni's post-hoc comparison.