Epistatic evidence for gender-dependant slow neurotransmission signalling in substance use disorders: PPP1R12B versus PPP1R1B

Abstract

Background: Slow neurotransmission including DARPP-32 signalling is implicated in substance use disorders (SUDs) by experimental systems but not yet in the human aetiology. PPP1R12B, encoding another protein in the DARPP-32 family, hasn't been studied in the brain.

Methods: Brain-regional gene activity was assessed in three different animal models of SUDs for mRNA level alterations. Genetic associations were assessed by meta-analysis of pre-existing dbGaP GWAS datasets for main effects and epistasis with known genetic risks, followed by cell type-specific pathway delineation. Parkinson's disease (PD) was included as a dopamine-related disease control for SUDs.

Findings: In animal models of SUDs, environmentally-altered PPP1R12B expression sex-dependently involves motivation-related brain regions. In humans with polysubstance abuse, meta-analysis of pre-existing datasets revealed that PPP1R12B and PPP1R1B, although expressed in dopamine vs. dopamine-recipient neurons, exerted similar interactions with known genetic risks such as ACTR1B and DRD2 in men but with ADH1B, HGFAC and DRD3 in women. These interactions reached genome-wide significances (P_meta<10^-20) for SUDs but not for PD (disease selectivity: P = 4.8 × 10^-142, OR = 6.7 for PPP1R12B; P = 8.0 × 10^-8, OR = 2.1 for PPP1R1B). CADM2 was the common risk in the molecular signalling regardless of gender and cell type.

Interpretation: Gender-dependant slow neurotransmission may convey both genetic and environmental vulnerabilities selectively to SUDs.

Funding: Grants from National Institute on Drug Abuse (NIDA) and National Institute on Alcohol Abuse and Alcoholism (NIAAA) of U.S.A. and National Natural Science Foundation of China (NSFC).

Keywords: Adolescence; Cell type-specific; Environmental risk; Missing heritability; Polysubstance abuse; Slow neurotransmission.

Fig. 1

Study design. Left, brain regional expression (protein by IHF and mRNA by qPCR) in nine regions (1, mPFC; 2, M/PtA cortex; 3, Hippocampus; 4, CPU; 5, LHb; 6, NAc,; 7, CeA; 8, SNc and 9, VTA; red, three focused regions in the animal models) and altered gene activity (mRNA) in three regions (red) of three animal models, including P rats with SUDs, adolescent exposures to alcohol or nicotine (n = 5 males or females per group); right, genetics of signalling network with cell type-specific pathway in humans. scRNA, single cell RNA.

Fig. 2

Ppp1r12b expression in different brain regions. (a-b) Immunoreactivities in five representative rat brain regions (mPFC, CPU, LHb, M/PtA and VTA) are shown here. Left column, PPP1R12B antibody (Anti-C, in red) recognized C-terminus in a; PPP1R12B antibody (Anti-N, in green) recognized N-terminus in b; second and third columns, anti-NeuN or anti-TH (green or red), and DAPI (blue); right column shows the merged staining. Clear neuronal staining was observed in VTA, mPFC and M/PtA; diffuse staining was observed in the LHb and CPU, consistently. In the bottom rows as a closeup of VTA, arrows indicate co-localization of Ppp1r12b and TH in dopamine neurons. Scale bars, 50 µm (other regions examined but not shown here: SNc was similar to VTA with a distinct cellular pattern; CeA and NAc were similar to diffuse CPU; hippocampus had a pattern in between mPFC and LHb). (c) Densitometry analysis of Ppp1r12b immunoreactivities observed in a (in light red) and b (in light green), and the average (in dark grey) of immunoreactivities, showing regional expression which was verified by one-way ANOVA results (n = 3). Information for additional regions was collected but less consistent so not shown. (d) Mouse brain regional expression by mRNA levels. Differential expression was verified by one-way ANOVA analysis. VTA and mPFC had no significant difference by Tukey's multiple comparisons (P = 0.29) (n = 5/group).

Fig. 3

Sex- and brain region-dependant levels of Ppp1r12b mRNA in NP vs. P rats: left, mPFC; middle, CPU; right, LHb. The t-test-based exact P values of <0.05 only are showed in graphs; 3-way ANOVA implied significant model interaction with sex (P = 0.0161) (n = 5/group).

Fig. 4

Tissue-dependant regulation of Ppp1r12b mRNA by chronic ethanol (a,c; blue) or nicotine (b,d; red) exposure in male (a,b) or female (c,d) mice. Procedures for chronic treatments are indicated on top of whole figure, the symbol♂ for male and ♀ for female are showed on left of whole figure. Note that males and females had different nicotine doses; the t-tests-based exact P values of <0.05 are showed in graphs; 3-way ANOVA implied significant ethanol interaction with sex (P<0.0001) (n = 5/group).

Fig. 5

Summary of sex-dependences in model- and region-related Ppp1r12b mRNA levels. Symbol:♂ for male and ♀ for female; orientation of symbols, for up or down-regulation; size of symbol, extent of regulation (not to scale), in three regions (mPFC, CPU and LHb); different colours, different rodent models as indicated.

Fig. 6

Gender-dependant PPP1R12B (top) or PPP1R1B (bottom) interactions with other genes in SUDs: (a) males; (b) females. 46 genes, organized in a wheel here in seven categories (enzyme, receptors, signalling, structure, TF, transporter and other; also listed in Supplementary Table 1), were included in this epistasis analysis. TF, transcription factors (e.g., HIVEP2 and unpublished PLAGL1); interacting genes are labelled in red; black triangle, known risk for SUDs, per reported meta-analyses (no interaction found for PD in this 46 genes-network (see Supplementary Fig. 2; also for the full labelling of genes). All interactions shown here reached statistical significance after Bonferroni correction and most of them reached absolute genome-wide significance (P_meta<10⁻²⁰), referring to the thermometer bar; FOXA1 SNPs might represent next gene TTC6 and ADH1B SNPs might represent next gene ADH1C too; scale bar, 100 kb. (c) Selective contribution of PPP1R12B (top) or PPP1R1B (bottom) to SUDs (top P value, from gender combined data), especially in females (P value grey-underlined, OR=6.7 (PPP1R12B) and 2.1 (PPP1R1B)) comparing to PD, based on number of statistically significant interactions; gender-specific selectivity for SUDs over PD: P = 1.6 × 10⁻²⁸ in males and 4.1 × 10⁻¹⁷⁸ (OR=164) in females in PPP1R12B and P = 5.9 × 10⁻²³ (OR=50.8) in males and 4.7 × 10⁻⁴⁹ in females in PPP1R1B.

Fig. 7

Cell type-specific pathways for PPP1R12B or PPP1R1B (DARPP-32) to interact with other genetic risks in a gender-dependant manner, based on epistasis in Fig. 6. (a) Cell type specific expression of Ppp1r12b and Ppp1r1b (cell number used: n = 417 for TH+, n = 40 for DRD1+, n = 40 for DRD2+, n = 10 for TRAP DRD2+ of mouse origin, and n = 129 for human cortex neurons), based on single cell RNA (scRNA) profiling except BacTRAP strategy sequencing for TRAP DRD2+. GAPDH was used to normalize for relative density of mRNA here in each cell and had very low density comparing to beta-actin in DRD1+ and TRAP DRD2+ cells. P values were from two-tailed t-tests. (b) Pathways: upper panels, for PPP1R12B in TH+ cells and lower panels, for PPP1R1B (DARPP-32) in dorsal striatal DRD1+ cells; left for males and right for females; all activity in pathway is expressed in the indicated cell type, based on the scRNA profiling.

Fig. 8

Summary of translational findings.