Differential Expressions of the Alternatively Spliced Variant mRNAs of the µ Opioid Receptor Gene, OPRM1, in Brain Regions of Four Inbred Mouse Strains

The µ opioid receptor gene, OPRM1, undergoes extensive alternative pre-mRNA splicing in rodents and humans, with dozens of alternatively spliced variants of the OPRM1 gene. The present studies establish a SYBR green quantitative PCR (qPCR) assay to more accurately quantify mouse OPRM1 splice variant mRNAs. Using these qPCR assays, we examined the expression of OPRM1 splice variant mRNAs in selected brain regions of four inbred mouse strains displaying differences in µ opioid-induced tolerance and physical dependence: C56BL/6J, 129P3/J, SJL/J and SWR/J. The complete mRNA expression profiles of the OPRM1 splice variants reveal marked differences of the variant mRNA expression among the brain regions in each mouse strain, suggesting region-specific alternative splicing of the OPRM1 gene. The expression of many variants was also strain-specific, implying a genetic influence on OPRM1 alternative splicing. The expression levels of a number of the variant mRNAs in certain brain regions appear to correlate with strain sensitivities to morphine analgesia, tolerance and physical dependence in four mouse strains.


Introduction
Morphine and most clinical analgesic agents, as well as heroin, act primarily through m opioid receptors [1]. However, their actions, such as analgesia, tolerance, physical dependence and other side-effects vary among patients [2][3][4]. Similar effects have been observed in animal models [5][6][7][8][9]. These observations are consistent with the concept of subtypes of m opioid receptors that was initially proposed by pharmacological studies [10][11][12] and subsequently confirmed with the isolation of a multitude of m opioid receptor variants generated by alternative splicing [13][14][15][16][17] in rodents and humans despite the presence of a single m opioid receptor gene, as determined by chromosomal mapping [18][19][20][21] and genomic sequencing. Alternative pre-mRNA splicing is a major mechanism for generating protein diversity and is extensively used by the m opioid receptor gene. To date, studies show that the mouse, rat and human OPRM1 genes generate 29, 16 and 19 splice variants, respectively, with similar patterns (Fig. 1) (see reviews: [16,22] ). These splice variants can be categorized into three major groups based upon receptor structure: 1) Fulllength carboxyl (C-) terminal variants with 7-transmembrane (7-TM) domains; 2) Truncated variants containing 6-TM domains; and 3) Truncated variants containing a single TM. The full length 7-TM C-terminal variants differ in their region-specific expressions, receptor phosphorylation, internalization, and post-endocytic sorting [23][24][25][26][27][28][29][30][31]. Furthermore, the intrinsic activities of m drugs vary from variant to variant [32]. The significance of the exon 11-associated 6-TM variants is illustrated by their role in the analgesic actions of opioids such as morphine-6b-glucuronide (M6G), fentanyl and heroin [33], as a novel new class of opioid analgesics lacking many of the side-effects of traditional opiates [34].
The sensitivity of mice towards m opioids on a variety of phenotypic measures is genotype dependent [35][36][37][38][39][40]. For example, surveys of inbred mouse strains reveal differences in morphine dependence, assessed with naloxone-precipitated withdrawal jumping scores, and tolerance, as seen by analgesic potency shifts after chronic morphine administration [35,36,41]. 129P3/J mice are highly refractory to morphine tolerance and naloxoneprecipitated withdrawal, whereas C57BL/6J and SWR/J mice develop significant morphine tolerance and physical dependence. These inbred mouse strains also show similar differences in naloxone-precipitated withdrawal jumping scores toward other m opioids including heroin [37], M6G and methadone [42]. However, it is not clear how the genetic background of these strains influences the expression of the OPRM1 gene, a primary candidate responsible for the actions of m drugs seen in these strains.
Our previous studies suggested region-specific alternative splicing of the OPRM1 gene among brain regions using regular RT-PCR approaches [25][26][27][28]. However, regular PCR can only provide estimates of quantification due to technical limitations. To establish more accurate, reliable and sensitive assays for quantifying OPRM1 variant mRNAs we have established SYBR green quantitative PCR (qPCR) assays for the OPRM1 splice variants. Using these qPCR assays, we examined the expression levels of OPRM1 splice variant mRNAs in selected brain regions of four inbred mouse strains: C56BL/6J (B6), 129P3/J (129), SJL/J (SJL) and SWR/J (SWR).

Materials and Methods
Animal C56BL/6J (B6, stock #: 000664), 129P3/J (129, stock#: 000690), SJL/J (SJL, stock #: 000686) and SWR/J (SWR, stock #: 000689) male mice at 7-8 weeks of age were obtained from Jackson Laboratory. All mice were housed in groups of five, maintained on a 12-h light/dark cycle and given ad libitum access to food and water. All animal studies were approved by the Institutional Animal Care and Use Committee of the Memorial Sloan-Kettering Cancer Center.
Tissue dissection, RNA extraction and reversetranscription (RT) Brain regions including the prefrontal cortex (Pfc), striatum (Str), thalamus (Tha), hypothalamus (Hyp), hippocampus (Hip), brain stem (BS), periaqueductal gray (PAG), and cerebellum (Cb) were dissected on a mouse Plexiglas brain mold using the atlas of Paxinos and Franklin [43] as a reference. The spinal cord (Spc) from L1 to L5 and whole brain were also dissected. The dissected tissues were immediately homogenized in QiAzol Reagent (Qiagen). The homogenates were stored at 280uC until further RNA isolation. 2-3 mice were pooled for each region, and a total of 8-10 mice used for each strain. Total RNAs were extracted using miRNeasy kit (Qiagen) following manufacture protocol. RNA concentrations were determined using a Qubit 2.0 Fluorometer (Invitrogen). RNAs were first treated with Turbo-DNA free reagent (Ambion) following the manufacture's protocol to remove potentially contaminating genomic DNA, and reverse transcribed with Superscript II reverse transcriptase (Invitrogen) and random hexmers. The first-strand cDNA was then used as templates in SYBR qPCRs.

Primer design and SYBR green qPCR
Since many exons are shared by different variants, it is difficult to design the short amplicon required for an efficient qPCR using the TaqMan assay for most of the splice variants. SYBR Green qPCR offers more flexibility in terms of amplicon size and thermocycle conditions. Using primer design tools in GeneRunner and VectorNTI, we designed appropriate primers for each splice variants, optimized qPCR parameters, such as annealing temperature and extension time, and obtained reasonable amplification efficiency by generating standard curves using 5-7 serial 10-fold dilutions of corresponding cDNAs as templates (Table S1). In most cases, we designed primers targeting exon-exon junctions or crossing at least two exons to obtain specific amplification. Specificities of the primers were verified by qPCRs using plasmid constructs containing corresponding variant cDNAs, and by melting curve analysis and agarose gel analysis after qPCR. Table  S1 lists the names, sequences and location of primers, as well as amplicon size, efficiency and PCR parameters, for all 29 splice variants. Two workstations in two separate rooms were used exclusively for qPCR to avoid cross contamination. Three reference genes, succinate dehydrogenase subunit A (SDHA), TATA box binding protein (TBP) and glycerakdehyde 3phosphate dehydrogenase (G3PDH), were selected to obtain normalization factors using the formula[44-46]: Normalization factor (NF) = (C(t) SDHA 6 C(t) TBP 6 C(t) G3PDH ) 1 , where E is efficiency. Fold change (FC) was calculated using the formula [44]: FC = NE variant /NE reference , where a reference was mE1/2 and is was defined as 100%. mE1/2 amplified with primers from exon 1 to exon 2 represented all C-terminal variants, including the original mMOR-1. Although the original mMOR-1 transcribed through the exon 1 promoter is by far the most abundant, it is difficult to design a pair of specific primers for amplifying the exon 1 promoter-driven mMOR-1, since exons 1-4 are shared by several other variants as well.

Statistical analysis
Data are represented as the mean 6 S.E.M. from at least three independent experiments. Two-way ANOVA with Tukey's multiple comparisons test was used for analyzing qPCR data. Statistical significance was set at p,0.05. The results are listed in Table S3.

Expression levels of OPRM1 splice variant mRNAs in four inbred mouse strains
Using the established RT-SYBR green qPCR assays, we examined the expression levels of OPRM1 splice variant mRNAs in nine brain regions of four inbred mouse strains. The overall expression levels among the variants are shown in a heatmap clustered based upon Z-scores from the values (Log 2 (E 2DC(t) ) of individual variants across different strains/regions, plotting from the highest level (left, red) to the lowest level (right, green) ( Fig. 2) (All E 2DC(t) values are listed in Table S2). The heatmap reveals a wide range of expression levels among the variants, as indicated by Z-score values from,24 to .4. mE1-2, which represents the entire repertoire of full-length 7-TM variants, was expressed at the highest level in all brain regions and strains. The expression of the other splice variants differed dramatically among the regions and strains.
To visualize expression patterns more closely, we compared five C-terminal splice variants and two 6-TM and two 1-TM variants, as well as mE1-2, by plotting their (21)/log 2 (E 2DC(t) ) values in the brain regions (Fig. 3). The various splice variant expression levels among the brain regions ranged widely. For example, in the hypothalamus of 129 mice, the relative expression levels ranged from 0.1% for mMOR-1M to 22% for mMOR-1A, when their E 2DC(t) values were used to compare with those of mE1-2 designated as 100% ( Fig. 3 & Table S2). The full length Cterminal splice variants mMOR-1A and mMOR-1B1 in 129 mice were highly expressed among most brain regions, corresponding to approximately 10-20% of mE1-2 levels in their E 2DC(t) values ( Fig. 3 & Table S2). On the other hand, in 129 mice the levels of the truncated 6-TM variant mMOR-1G and mMOR-1M and single TM variants, mMOR-1S and mMOR-1T, and the full length mMOR-1D, ranged from about 0.1% to 5% of mE1-2 levels ( Fig. 3 & Table S2). Similar expression patterns were observed in other mouse strains ( Fig. 3 & Table S2).

Region-specific expressions of OPRM1 splice variant mRNAs in four inbred mouse strains
Overall, expression levels in whole brain were not predictive of strain differences in specific regions. Within each strain, there are major differences in the regional expression of the variants. The heatmap segregating the different strains shows the overall differential expression pattern of all variants across the brain regions (Fig. 4). Plotting the expression level (E 2DC(t) ) for each variant among the regions illustrates the differences quantitatively and provides a visual pattern of regional expression (Figs. 5, S1 & S2). For example, the expression pattern of the full length variant mMOR-1A differed among the four strains. Its levels were high in the thalamus and PAG of B6 mice, with lower levels in the hippocampus and hypothalamus. In contrast, mMOR-1A expression in 129 mice was highest in the hypothalamus with similar high levels in the striatum, thalamus, hippocampus and PAG. In SWR mice mMOR-1A levels were highest in the hypothalamus with lowest levels in the hippocampus.
In general, variant expression was low in the cerebellum and prefrontal cortex, consistent prior in situ hybridization studies with the full-length MOR-1 probe [47]. However, this was not uniformly the case. In SWR mice, mMOR-1U expression was greatest in the cerebellum (Fig. S2) while mMOR-1O was highest in the prefrontal cortex (Fig. 5).
Strain-specific expression of OPRM1 splices variant mRNAs in selected brain regions Segregating variant expression by individual regions provides another approach to visualize strain differences in different splice variants, and reveals interesting comparisons (Figs. 6, 7, S3 & S4). The heatmap shows broad differences in expression of individual variants among the strains in most brain regions (Fig. 6), as does quantitative graphing where the expression level (E 2DC(t) ) of each variant was plotted by strains within each region (Figs. 7, S3 & S4). Within a region on the heatmap, each column provides the relative expression of the different variants within the strain whereas the horizontal comparisons reveal interstrain differences for a specific variant.
Overall, the expression level of the variants varied among the mouse strains. For example, in the spinal cord, the expression of mMOR-1A in SJL mice was higher than the other strains (Fig. 7). However, in the striatum, 129 mice had the highest expression level. In the thalamus the 6-TM variant mMOR-1N expression was much higher in B6 mice than in other tree mouse strains, contrasting to its expression in the PAG where SJL levels were greater than other three strains (Fig. 7). The expression of the single TM variant mMOR-1S was higher in SWR mice in most brain regions including the spinal cord, hypothalamus, prefrontal cortex, striatum, PAG and brainstem. Although this single TM variant does not directly bind opioid receptors, it is pharmacologically relevant in that it stabilizes the full length mMOR-1 variant and significantly contributes to morphine analgesia [48].

Discussion
The identification of a vast array of splice variants of the m opioid receptor gene has shown a complexity in opioid pharmacology far beyond what had previously been imagined. Autoradiography provided the earliest means of examining the anatomical distribution of m binding sites [49][50][51][52]. The distribution of m opioid receptors was then examined immunohistochemically and with in situ hybridation following the cloning of the receptor [53][54][55][56][57][58][59][60][61][62][63]. While these studies have provided valuable insights into opioid action, they all have limitations. Autoradiographic approaches are functional in the sense that they require active binding of ligand. Since the early studies utilized traditional m radioligands, they detected all the full length 7-TM variants, but none of the truncated ones, and were not able to distinguish among the 7-TM variants. The early immunohistochemical studies targeted the C-terminus of MOR-1, with most using an epitope encoded by exon 4. These probes will detect the full length variants MOR-1, MOR-1H, MOR-1i and MOR-1J, as well as the 6-TM variant MOR-1G which all share the exon 4 epitope, but none of the other variants. Some studies have examined epitopes from alternatively spliced C-terminals, but they, too, may have labeled more than one species. For example, early studies examined the labeling of a 7/8/9 epitope, assuming it corresponded to mMOR-1C. Yet, subsequent cloning studies identified mMOR-1M, a truncated 6-TM variant that also contains the 7/ 8/9 epitope. The in situ hybridization studies using probes targeting exons 2 and 3 detected all the 7-TM and 6-TM splice variants, but not the single TM ones. Thus, each approach gives a unique set of labeling. Theoretically, qPCR offers the ability to identify an individual mRNA, but even PCR faces selectivity challenges due to both 39 and 59 splicing and extensive overlap in sequence among the variants.
We previously used traditional RT-PCR approaches to examine the expression of OPRM1 splice variant mRNAs in brain regions [25][26][27][28]. However, these early approaches provided only an approximation of levels since they were not quantitative. Many of these studies also were carried out prior to the identification of additional variants with overlapping sequences. The present study using SYBR green qPCR overcomes these limitations and more precisely quantifies the expression of OPRM1 splice variant mRNAs. Using these assays, we investigated the expression profiles of all OPRM1 splice variant mRNAs in various brain regions of four inbred mouse strains, for the first time, a complete expression mRNA profile. However, even this approach does not address their expression at the protein level.
Although our current approach provides selectivity, dissecting tissues from a limited of brain regions does not provide anatomical resolution at the cellular level and provides only an incomplete assessment of variant expression. Future studies using single cell qPCR or more sophisticated in situ hybridization approaches may overcome these limitations. However, the current studies provide the first step toward our understanding of OPRM1 splice variant expression in brain.
The abundance of the OPRM1 splice variant mRNAs covers a wide range. These mRNA levels presumably reflect steady-state levels that depend upon many factors that may vary over time, including transcription, post-transcriptional modification and mRNA degradation. The OPRM1 gene has two distinct promoters, exon 1 and exon 11 promoters. Most of the exon 1 promoter-controlled variants involve 39 splicing, yielding unique C-terminal sequences, and are quite abundant. They include mMOR-1A, mMOR-1B1, mMOR-1C, mMOR-1O and mMOR-1V. In contrast, the exon 11 promoter-controlled variants generate 6-TM variants and are expressed at lower levels. While the promoters may influence mRNA expression, they cannot explain all the differences. Within each promoter group, mRNA levels vary, as shown by the far lower levels of mMOR-1B2 and mMOR-1P levels than those of mMOR-1A, mMOR-1B1 and mMOR-1C in the PAG, hypothalamus and hippocampus of B6 mice and the differences in the exon 11 promoter-controlled 6-TM variants mMOR-1G, mMOR-1K and mMOR-1N. mRNA stability also may impact expression levels. Of particular interest is nonsense-mediated decay (NMD) that targets an mRNA containing a premature stop codon located more than 50 nt upstream of the last exon-exon junction [64,65]. Eight variants, including mMOR-1E, mMOR-1F, mMOR-1Q, mMOR-1R, mMOR-1T, mMOR-1V, mMOR-1W, and mMOR-1Z, are Figure 5. Region-specific expressions of ten-selected OPRM1 splice variants. Each panel represents the regional expression of one variant in four inbred mouse strains.   potential targets of NMD. However, the variability of expression among these variants implies that NMD is not solely one factor among many influencing expression levels.
mRNA levels may not correlate with either their corresponding protein expression or with their pharmacological importance. For example, immunostaining by an exon 7/8/9-specific antisera that label mMOR-1C and mMOR-1M was as robust as that seen with the exon 4-specific antisera labeling mMOR-1, mMOR-1H, mMOR-1i and mMOR-1J despite great differences in their respective mRNA levels [23,24]. The lack of correlation between mRNA and protein levels also may involve many factors. The association of the different mRNAs with polyribosomes can vary widely (Z. Lu, J. Xu, G.W. Pasternak, and Y.X. Pan, unpublished data), suggesting the differential regulation of the variants at the translational level. Other evidence indicates that the stability of the variants at the protein level is dependent upon a range of factors. For example, internalization and recycling of the receptors varies among the variants [31] and co-expression of single TM splice variants stabilizes MOR-1 protein levels, decreasing its degradation, extending its half-lives and thereby increasing its overall levels [48]. Therefore, it will be important to develop a reliable and an accurate method for determining protein expression of these OPRM1 splice variants in the future. mRNA levels do not necessarily correlate with pharmacological relevance. For example, mMOR-1D has been implicated in morphine-induced itch through its dimerization with gastrinreleasing peptide receptor (GRPR) [66] despite its low mRNA levels. Exon 11-associated 6-TM variants are also among the low abundant transcripts. Yet, an exon 11 knockout mouse model indicates a key role of these 6-TM variants in the analgesic actions of a range of opioids, including heroin, morphine-6b-glucuronide, fentanyl, levorphanol and buprenorphine [33,34]. These 6-TM variants also are critical in the actions of a novel class of opioid analgesic exemplified by 39-iodobenzoyl-6b-naltexamide (IBNtxA), which is interesting since IBNtxA produces potent analgesia without many traditional opioid side-effects such as respiratory depression, physical dependence and reward behavior [34]. The importance of 6-TM variants is further supported by recent studies showing the rescue of IBNtxA analgesia in knockout mice lacking 6-TM variants that are insensitive to IBNtxA by the lentiviral-mediated delivery of a 6-TM variant, mMOR-1G (Z. Lu, J. Xu, G. Rossi, G.W. Pasternak, and Y.X. Pan, unpublished observations).
Our data also demonstrate variability in the expression of OPRM1 splice variants in different strains. The importance of genetic backgrounds in opioid pharmacology is becoming increasingly appreciated. The four strains chosen for evaluation differ in their sensitivity to morphine analgesia, with AD 50 values ranging from 6.8 mg/kg in C57BL/6 and 7.2 mg/kg in SJL to 13.7 mg/kg in 129P3 and 18.7 mg/kg in SWR mice [36]. They also differ in the development of tolerance to chronic morphine dosing of over three days, with analgesic shifts of 0.8 in the 129P3 mice to 2.2 in the SJL mice, 5 in SWR mice and 7.2 in the C56BL/6 mice [36]. Dependence, measured by jumping following naloxone precipitated withdrawal, also varies with the 129P3 and SJL mice showing virtually no jumping with moderate levels of jumping in the C57BL/6 mice and the most in the SWR mice [35]. Although four strains is a very limited sample, it is interesting that linear regression analysis correlating mRNA expression levels with 1) morphine analgesia using half-maximal analgesic doses (AD 50 ), 2) morphine tolerance using the shift in AD 50 fold with chronic dosing over three days or [36] or 3) physical dependence determined by jump scores from naloxone-precipitated withdrawal [35] revealed strong correlations for some of the variants (r 2 .0.90; Fig. S5), raising questions regarding the functional relevance of the variants in these brain regions. While intriguing, these correlations must be interpreted cautiously and require additional studies. Recently, we observed that long-term morphine treatment in CD-1 mice markedly upregulates a number of OPRM1 splice variants in various brain regions, as determined by the qPCR assays described here (J. Xu, A.J. Faskowitz, G.C. Rossi, M. Xu, Z. Lu, Y.-X Pan and G.W. Pasternak, submitted). Considering the variability in mRNA expression among the current strains of mice, it will be interesting to compare morphine effects on the expression of OPRM1 splice variants in these inbred mice.
Why the pattern of the variant expression in these four inbred mouse strains containing the same single-copy OPRM1 gene varies is not clear. Presumably, the strains differ in their modulation of the promoters and alternative splicing machinery. These may reflect genetic variations, such as single nucleotide polymorphisms (SNPs), deletions, insertions and transposon elements. Recently, we identified a heroin addiction severityassociated intronic SNP that modulates alternative splicing of the human OPRM1 variants via hnRNPH interaction [67]. Further identifying the genetic variations that modulate the strain-specific OPRM1 alternative splicing will greatly advance our knowledge not only on the OPRM1 alternative splicing, but also on alternative splicing in general. Figure S1 Expression levels of additional ten OPRM1 splice variants in brain regions of B6, SJL and SWR mice. Each panel represents the regional expressions of one variant in four inbred mouse strains.   50 )). The AD 50 (mg/kg) values in B6, SJL, 129 and SWR mice were 6.8, 7.2, 13.7 and 18.7, respectively [36]. C & D: Correlation of the expression levels of OPRM1 splice variant mRNAs with morphine tolerance in four inbred mouse strains. The square of the correlation coefficient (r 2 ) listed was obtained from linear regression analysis (Prizm 5.0) between expression level (E 2DC(t) *10 4 ) values (mean 6 S.E.M.) for each variant in the indicated brain region and morphine AD 50 shift before and after chronic morphine treatment (log (Morphine AD 50 shift)). The AD 50 shift values in 129, SJL, SWR and B6 mice were 0.8, 2.2, 5 and 7.2, respectively [36]. E & F: Correlation of the expression levels of OPRM1 splice variant mRNAs with morphine-induced physical dependence in four inbred mouse strains. The square of the correlation coefficient (r 2 ) listed was obtained from linear regression analysis (Prizm 5.0) between expression level (E 2DC(t) *10 4 ) values (mean 6 S.E.M.) for each variant in the indicated brain region and jumping scores after chronic morphine treatment followed by naloxone precipitated withdrawal (log (Jumps/15 min)). The jumping scores in 129, SJL, B6 and SWR mice were 7, 16, 60 and 200, respectively [35]. All the variants listed have an r 2 value over 0.90. A positive slope indicates that the more expression of the variants, the more AD 50 values (the less potent) or the more AD 50 value shifts (the more tolerant) or the more jumps (the more dependent) are. A negative slope shows opposite relationships. Pfc: prefrontal cortex; Str: striatum; Hyp: hypothalamus; Hip: hippocampus; PAG: periaqueductal gray; BS: brainstem; Spc: spinal cord; WB: whole brain. (TIF)

Supporting Information
Table S1 SYBR green qPCR assay. SYBR green qPCR assay for each variant was established as described in Materials and Methods. Optimized conditions including primers and their locations, thermal cycle condition, amplicon size and efficiency are listed. SP: sense primer; AP: antisense primer. (XLSX)