QTL mapping was performed using GeneNetwork 1.0 with a composite interval mapping function using 2000 permutations. This is a forward regression approach in which a single locus with a major effect is included in a mapping model that scans for additional additive effects and interactions with the major locus. The technique will miss higher order interactions among loci that are not detectable as main effects, but is effective when a large, consistently observed major effect locus is present. Strain mean scores were Winsorized when statistical outliers were present. We first determined which marker had the highest LRS value, using a marker regression analysis and then performed composite interval mapping, controlling for the Chromosome 10 SNP rs3721803, one of the 3 markers with the highest LRS scores.

The MGI database ( http://www.informatics.jax.org/) was used to find information about SNPs and strain polymorphisms occurring within the Chromosome 7 interval (queried –June 20, 2016). This database now includes variants detected in the sequencing of 17 mouse genomes ( Keane et al., 2011). The 3 SNPs located at the Chromosome 7 QTL peak (rs13479451, rs3724540, rs6386601) are at ~114.5 Mb in GeneNetwork and ~107.6 Mb in the MGI and NCBI databases ( http://www.ncbi.nlm.nih.gov/gene). In GeneNetwork, the relevant QTL interval on Chromosome 7 was 110–125 Mb whereas this region corresponds to 103–118 Mb in the MGI database. As finer mapping methods are developed, the exact interval location as reported here, may change. DAVID version 6.8 ( https://david.ncifcrf.gov/) was used to obtain functional annotations and pathway information for the genes within the interval and GeneWeaver 1.0 ( http://www.geneweaver.org/) to identify other drug-related phenotypes associated with the candidate genes.

As reported by Philip et al. (2010), whole genome scans produce robust QTLs mapping to Chromosome 10 at 0–30 Mb ( Figure 1A-B). The positional and functional best gene candidate is the Oprm1 gene which encodes the opioid G-protein coupled receptor mu 1 ( Philip et al., 2010) which has been previously detected ( Bergeson et al., 2001; Doyle et al., 2014).

We remapped these data for each time point using composite interval mapping to identify QTLs potentially masked by this large QTL. This revealed a second significant QTL on Chromosome 7 for locomotion (LRS= 18.0, 114.7 Mb; Figure 1C) and, in an overlapping position, for rearing (LRS=16.5, 114.3 Mb). We also observed a time dependent decrease of the peak LRS values for both traits after the 60–75 minute time interval post-morphine injection. A suggestive QTL for the rearing trait data was already apparent on Chromosome 7 from the whole genome scan, but this was not the case for the locomotion trait data ( Philip et al., 2010).

The area under the QTL peak contains 280 protein coding genes (MGI Chromosome 7: 103–118 Mb, GeneNetwork Chromosome 7: 110–125 Mb; database query-June 20, 2016, Supplementary worksheet 2). Of these genes, 162 are associated with olfactory receptors. Of the other genes with functional annotations, 11 are connected with neural processes ( Adm: neural tube development, Appb1, Tub, Calca, Cckbr, Cnag4: sensory perception and cognition, Insc, Arntl: neurogenesis, Tpp1: neuromuscular control, Rras2: regulation of neuron death and Pde3b: morphine addiction (KEGG pathway 05032).

There are 3 SNPs mapped at the location of the LRS peak and two positional candidates: Synaptogamin 9 ( Syt9: rs13479451, rs3724540) and PPFIA binding protein 2 ( Ppfibp2, rs6386601). Both of these genes differ between the parental strains (MGI database query – June 20, 2016). Syt9 functions in vesicle traffic and Ca2+ triggered exocytosis ( Ncbi, 2016b) and Ppfibp2 is involved in the regulation and development of neuronal synapses ( Ncbi, 2016a). These two candidates are also differentially expressed in the striatum of the brains of mice from strains with distinct opioid sensitivity ( Korostynski et al., 2006; GeneWeaver: Syt 9, geneset #86830; Ppfibp2, geneset #86906).