Ectopic Expression of Retrotransposon-Derived PEG11/RTL1 Contributes to the Callipyge Muscular Hypertrophy

The callipyge phenotype is an ovine muscular hypertrophy characterized by polar overdominance: only heterozygous + Mat /CLPG Pat animals receiving the CLPG mutation from their father express the phenotype. + Mat /CLPG Pat animals are characterized by postnatal, ectopic expression of Delta-like 1 homologue (DLK1) and Paternally expressed gene 11/Retrotransposon-like 1 (PEG11/RTL1) proteins in skeletal muscle. We showed previously in transgenic mice that ectopic expression of DLK1 alone induces a muscular hypertrophy, hence demonstrating a role for DLK1 in determining the callipyge hypertrophy. We herein describe newly generated transgenic mice that ectopically express PEG11 in skeletal muscle, and show that they also exhibit a muscular hypertrophy phenotype. Our data suggest that both DLK1 and PEG11 act together in causing the muscular hypertrophy of callipyge sheep.


Introduction
The callipyge phenotype The callipyge phenotype is a muscular hypertrophy described in sheep. The augmented muscle mass (less than or equal to 50% increase) characterizing callipyge animals results from the larger proportion and diameter of glycolytic, fast twitch type II muscle fibers. The callipyge phenotype manifests itself at approximately one month of age and is characterized by a rostrocaudal gradient (i.e. more pronounced in the posterior than in the anterior part of the body).

Trans-effects underlying polar overdominance
The most remarkable feature of the callipyge phenomenon is the fact that only + Mat /CLPG Pat animals display the muscular hypertrophy. This non-Mendelian mode of inheritance is referred to as polar overdominance [7]. It differs from regular parental imprinting as homozygous CLPG/CLPG animals do not express the phenotype. This is thought to result from the post-transcriptional trans-inactivation of the paternally expressed protein-coding DLK1 and PEG11 genes by one or more of the long and/or small non-coding maternally expressed RNAs from the domain [8]. As a consequence, DLK1 and PEG11 are expressed at the protein level only in postnatal skeletal muscle of + Mat /CLPG Pat animals, the only individuals expressing the callipyge phenotype [9][10][11]. The trans-inactivation of PEG11 in CLPG/CLPG animals has been shown to result from the RISC-mediated cleavage of PEG11 mRNA by at least three perfectly complementary miRNAs (oar-miR-136, -127 and -432) processed from the maternally expressed anti-PEG11 non-coding RNA [12]. The molecular mechanisms underlying the trans-inactivation of DLK1 remain elusive [13].

Role of DLK1 and/or PEG11 protein in causing the callipyge phenotype
The distinctive features of + Mat /CLPG Pat animals (i.e. (i) the only genotype expressing the callipyge phenotype, and (ii) the only genotype expressing DLK1 and PEG11 protein in skeletal muscle after birth), strongly suggest that ectopic expression of DLK1 and/or PEG11 protein causes the callipyge muscular hypertrophy. DLK1 encodes a member of the Notch-Delta family of membrane receptors/ligands, which is known to be involved in adipogenesis, haematopoiesis, neurogenesis and adaptation to independent life (f.i. [1]). DLK1 has been implicated directly in causing the callipyge muscular hypertrophy in a murine model; transgenic mice expressing ovine DLK1 (C2 isoform) under the dependence of the muscle specific murine Myosin light chain 3F promoter and 2E enhancer (Mlc 3F/2E) exhibit a muscular hypertrophy [9]. PEG11 is an "exapted" protein derived from a eutherian-specific sushi-ichi retrotransposon, which is essential for the maintenance of the feto-maternal placental interface [14]. Notably, enhanced expression of PEG11 was shown recently to drive hepatocarcinogenesis [15]. Whether PEG11 participates in producing the callipyge phenotype has remained undetermined so far.
Here, we describe the generation of transgenic mice expressing ovine PEG11 under the dependence of the same Mlc 3F/2E promoter and enhancer combination to study the contribution of PEG11 in producing the callipyge phenotype.

Results
Generating transgenic mice expressing ovine PEG11 in skeletal muscle First, we cloned the ovine PEG11 open reading frame (oPEG11 ORF), encoding 1,333 amino acids, into a modified pMlc3F-nlacz-2E vector, placing it under the dependence of the Mlc 3F/ 2E [9]. The corresponding regulatory elements are expected to drive expression of the transgene preferentially in type IIB fast twitch muscle fibers (affected predominantly in callipyge sheep) throughout pre-and postnatal development [16]. The 7.6 Kb insert was released from the remainder of the vector by digestion with NotI and SmaI, gel-purified and microinjected into fertilized inbred FVB/N murine eggs, which were re-implanted into pseudo-pregnant foster mothers. Among 41 offspring, eight had genomic integration of the transgene. Six out of the eight founders transmitted the transgene to their offspring. We performed RT-PCR to amplify the oPEG11 ORF using total RNA extracted from quadriceps femoris muscle of two offspring per transmitting founder. Transgene expression was observed for two mouse lines (referred to as 126 and 127; data not shown) and these were analyzed further.
We combined Southern blotting, regular PCR, and "splinkerette" PCR [17] to characterize the transgene insertions in mouse lines 126 and 127. Line 126 was shown to harbor a singlecopy transgene that was integrated into an intergenic region on chromosome 16, at *407 Kb from Roundabout homolog 2 (Robo2) and *722 Kb from Lipase, member I (Lipi) (Fig 1; S1  Fig). Unexpectedly, the transgene comprised both the Mlc/oPEG11 insert, as well as the remainder of the vector including a 0.22 Kb duplicated segment flanking the transgene on either end (Fig 1). This indicated that the microinjected insert preparation was contaminated with undigested vector. The restriction patterns obtained by Southern blotting suggest that the transgene insertion was accompanied by a more than or equal to 1.6 Kb deletion at the integration site (Fig 1).
Mouse line 127 was shown to harbor two distinct integration sites (127A and 127B), both comprising a similar, complex array thought to encompass multiple tandem copies of digested insert and undigested full-length vector, confirming the contamination of the insert preparation with undigested vector (Fig 1C; S2 Fig). Expression of the transgene was only observed in mice with the 127A integration, not in mice with the 127B integration (data not shown). The 127A integration site was shown to correspond to the last intron of Coiled-coil domain containing 91 (Ccdc91) on chromosome 6 (S1 Fig). Ccdc91 was shown to be highly expressed in skeletal muscle, and its expression to be reduced strongly by the transgene insertion (p = 5.6x10 -5 ; Fig  2A).
We used RT-PCR, rapid amplification of cDNA ends (RACE), and Northern blotting to analyze further the transgene expression in mouse lines 126 and 127A. Northern blotting with total RNA extracted from quadriceps femoris of 10-week old animals revealed transgene- specific bands of expected size (*4.3 Kb) and intensity (*2:1 homozygote:heterozygote ratio) in line 126. In addition to the same band, a stronger and larger (*7 Kb) band was observed in line 127A (Fig 2B). RT-PCR analysis with overlapping amplicons encompassing the complete oPEG11 ORF revealed amplification products of expected size and sequence for line 126 ( Fig  2C). Similar results were obtained for line 127A with, however, extra amplification products for the 5' amplicon (S2/A3 in Fig 2C). The largest of these was shown by sequencing to result from retention of the construct's intron (733 bp between Mlc exon 2 and oPEG11 ORF) Relative expression levels of Ccdc91 (mean ± 1.96 x standard error of the mean (s.e.m.)) measured by QRT-PCR using RNA extracted from quadriceps femoris muscles of +/+ and +/T P mice from oPEG11 line 127A. The nearly 1.5-fold reduction in +/T P animals suggests an inactivation of the Ccdc91 gene and likely explains the absence of T P /T P mice in this cross. B. Northern blotting of mRNA extracted from quadriceps femoris of +/+, +/T P , and T P /T P mice from oPEG11 lines 126 and 127A using RNA probes corresponding to the ovine PEG11 transgene (oPEG11) and mouse β-actin for control (Actb). The bands have the approximate size expected for transcripts derived from the oPEG11 transgene (4.3 Kb, arrow) in the both mouse lines plus a larger band (~7 Kb, arrow head) in line 127. C. RT-PCR (or PCR) performed using total RNA (or DNA) extracted from quadriceps femoris (or tail) of +/+, +/T P , and T P /T P mice from lines 126 and 127A using primer combinations shown in Fig 1A. The extra-bands corresponding to unspliced transcripts observed in line 127A are differentiated from the bands expected for spliced transcripts observed in lines 126 and 127A. D. Results of 5' and 3' RACE experiments conducted with total RNA extracted from quadriceps femoris of +/+, and +/T P mice from lines 126 and 127A. Sequencing the PCR products marked by the arrows confirmed the use of the vector-specific transcription start and polyadenylation sites (data not shown). E. Relative expression level of oPEG11 (mean ± 1.96 x s.e.m.) in line 126 measured by QRT-PCR using total RNA extracted from quadriceps femoris from animals of indicated genotypes and ages, and from tissues at 10 weeks of age. Expression levels of +/T P animals were used for references. The numbers of tested samples are shown in parentheses. introducing premature termination codons. 5' and 3' RACE uncovered the expected, vectordependent transcription start and polyadenylation sites in both lines ( Fig 2D).
Because of the complex structure of the transgene insertion, the occurrence of multiple abnormal PEG11 transcripts, and the deleterious effect on Ccdc91 expression precluding the attainment of homozygous mice (data not shown) in line 127A, we decided to pursue detailed analysis of line 126 only. RT-PCR analysis indicated that the line 126 transgene was expressed preferentially in skeletal muscle, and that expression increased from birth to 10 weeks in this tissue ( Fig 2E).
Transgene genotype marginally increased live weight at 10 weeks of age (p = 0.03), as well as the weight of organs (kidney and lung, p > 0.04), although these effects were not significant when accounting for the realization of multiple tests (p ! 0.27 after Šidák correction, S2 Table). To verify further whether the effect on muscularity might just reflect a systemic growth effect rather than being genuinely muscle-specific, we tested the effect of transgene genotype on the weight of individual body parts relative to live weight either by dividing the phenotypic values by live weight (Model-1) or by adding live weight as a covariate in the mixed model (Model-2, S2 Table). The effect on hind legs and quadriceps remained highly significant (Šidák corrected p 3.7x10 -3 ). The effect on organs, however, became non-significant (Šidák corrected p ! 0.13) and their relative weight didn't increase consistently with transgene copy number (Fig 3; S2 Table).
We next examined the effect of the oPEG11 transgene on myofiber size distribution. In callipyge sheep, myofiber diameter was increased by 24%, 20% and 7% in longissimus, gluteus medius and supraspinatus muscles, respectively [18]. The ovine DLK1 transgenic mice exhibited increase of the quadriceps myofiber diameter by 7.4%, 8.7% and 10.5% in +/T and T/T (line A), and +/T (line D) genotypes, respectively [9]. We here selected to use extensor digitorum longus (EDL) and soleus (SL) muscles that are known to be rich and poor in type IIB myofibers, respectively (55.7-86.6% for EDL, 0.0-3.1% for SL [19,20]) where the Mlc 3F/2E regulatory elements are expected to be most active [16]. We dissected the two muscles from ten male mice in two litters at 7 and 10 months of age (n = 3, 5, 2 for +/+, +/T P and T P /T P , respectively) and performed hematoxylin and eosin staining of transverse sections. We measured the minimal Feret's diameter of myofibers [21] (on average 1,062 myofibers per animal-muscle combination) so as to cover the entire cross-sectional muscle area (Fig 4; S3 Fig). EDL myofiber size was increased by 15.7% and 24.6% in +/T P and T P /T P mice, respectively, compared to +/+ (p < 2.0x10 -16 ). The contrast between +/T P and T P /T P genotypes was also significant (+7.7%, p < 2.0x10 -16 ). On the other hand, none of the corresponding contrasts in SL muscle was significant (p > 5.9x10 -2 ), consistent with the lower contribution of type IIB fibers in SL muscle. Skewness of the myofiber diameter distribution in EDL muscle was changed from right-to lefttailed by increasing the transgene copy number (+/+ = 0.22, +/T P = -0.05, T P /T P = -0.17). This implies that part of the myofibers in EDL muscle (supposedly glycolytic fast twitch fibers) is enlarged by the oPEG11 transgene expression. We did not observe any apparent histological abnormality in these samples (f.i. necrosis, fibrosis, fat deposition, increase of myofibers with centrally located nuclei) (S3 Fig). Together, our data indicate that postnatal ectopic expression of ovine PEG11 in skeletal muscle of transgenic mice enhances muscle growth, and this strongly supports a contributing role for PEG11 in the muscular hypertrophy of callipyge sheep. The magnitude of the effects observed on muscle weight (less than or equal to 5.3% increase (T P /T P quadriceps)) are modest when compared to those reported in callipyge sheep (f.i. 18.8-21.0% increase in quadriceps Phenotypic effects of one (+/T P ) and two (T P /T P ) copies of the oPEG11 transgene (line 126) in the oPEG11 F1 mouse cross (+/T P x +/T P ) at 10 weeks of age. The eight graphs show the distribution of individual phenotypic values corrected for sex and litter effects (carcass weight, the weight of the hind legs, quadriceps and lung and their ratios to live weight (LW)) of mice sorted by transgene genotype. Ratios were subject to angular transformation. Genotype means are shown by black dots with error bars (± 1.96 s.e.m.). Statistically significant differences between the possible combinations of genotype computed with the mixed model and Tukey honest significant difference test are marked by braces; ***: p < 0.001, **: p < 0.01, *: p < 0.05 (S2 Table); n = 68 (+/+), 122 (+/T P ) and 55 (T P /T P ).  [22,23]). To gain some insights in the possible causes of this discrepancy, we evaluated simultaneously the expression level of ovine PEG11 in + Mat /CLPG Pat sheep as well as +/T P and T P / T P transgenic mice by Northern blotting (Fig 5). While the ovine PEG11 transcripts were clearly detectable in skeletal muscle of transgenic mice, their abundance (estimated by densitometry) was *10-fold lower than in callipyge sheep (mouse T P /T P versus sheep + mat /C pat , Fig  5), and this, in addition to the absence of ectopic expression of DLK1, may explain the differences in terms of effect size between mice and sheep.
Following Vuocolo et al. [24], we attempted to demonstrate a change in myofiber type composition in +/T P and T P /T P transgenic mice by performing QRT-PCR using quadriceps muscle RNA for Myosin heavy chain 1 (Myh1), Myh2, Myh4 and Myh7, which are specific for type IIx, IIa, IIb and I myofibers, respectively. We observed no significant effect of transgene genotype on the relative amounts of any one of these myosin heavy chain isoforms (S4 Fig).

Discussion
The question of whether ectopic expression of DLK1 and/or PEG11 is responsible for the callipyge muscular hypertrophy has been a matter of debate for some time. The notion that imprinted genes from the DLK1-GTL2 domain might affect muscle development is supported by both the observed increase in the number of myofibers with centrally located nuclei and also the enlargement of myofiber cross-sectional area in fetuses with paternal uniparental disomy of mouse chromosome 12 (f.i. [25]), and fetuses inheriting a deletion of the Dlk1-Gtl2 intergenic germline-derived differentially methylated region (IG-DMR) on their maternal chromosome [26,27]. However, these models perturb the dosage of at least three genes: Dlk1, Peg11 and Dio3. As DIO3 is not affected by the CLPG mutation in callipyge sheep, DLK1 and PEG11 remained the two most plausible candidates.
A role for DLK1 appeared well established when it was shown in transgenic mice that ectopic expression of ovine DLK1 (transmembrane C2 isoform) in pre-and postnatal skeletal muscle caused a muscular hypertrophy [9]. However, the validity of this conclusion was challenged by the observation that BAC-transgenic mice with double or triple Dlk1 dosage did not exhibit the expected skeletal muscle anomalies [28]. This discrepancy may be explained in part by the fact that the natural spatio-temporal gene expression profile is maintained in the da Rocha et al. [28] model, but not in the Davis et al. [9] model, in which DLK1 is re-expressed in a tissue (muscle) and at a developmental stage (postnatal) where it is normally silenced completely.
Direct evidence for a role of PEG11 in modulating muscle phenotype has not been reported previously. Sekita et al. [14] generated transgenic mice overexpressing Rtl1 (Peg11) alone (respecting its normal spatio-temporal expression pattern), but did not report any effect on . Northern blotting performed with 20 μg of total RNA extracted from longissimus dorsi muscles of sheep of the four possible CLPG genotypes (+/+, CLPG Mat /+ Pat , + Mat /CLPG Pat and CLPG/CLPG), or quadriceps femoris muscles of +/+, +/T P , and T P /T P mice from oPEG11 line 126 and probed with a strand-specific RNA probe corresponding first to ovine PEG11, and secondly to mouse β-actin. The bands corresponding presumably to the endogenous ovine PEG11 (sheep PEG11), transgene-derived ovine PEG11 (oPEG11), or β-actin (ACTB/Actb) are marked. Relative quantities were estimated by densitometric analysis of the autoradiograms (a bar graph on the right) suggesting that expression levels of ovine PEG11 are *10-fold lower in T P /T P mice from oPEG11 line 126 than in callipyge + Mat /CLPG Pat sheep. skeletal muscle. It is noteworthy that the majority of these animals died before or just after birth with severe placental anomalies, and that-in surviving animals-PEG11 expression would likely have rapidly been downregulated in skeletal muscle. In this study, although presenting results for one line only (oPEG11 mouse line 126), we strongly believe that the observed increase in muscle mass is due to ectopic expression of the oPEG11 transgene in skeletal muscle rather than to an unintended side-effect of the transgene insertion. First of all, the phenotype is the most prominent where (skeletal muscle) and when (after 7 weeks of age) the Mlc 3F/2E promoter and enhancer is expected to be most active and is independent of live weight [16,19]. Secondly, we observe a strong effect on the diameter of myofibers in the fast twitch muscle. Thirdly, the integration site is in the middle of a large intergenic region at more than 400 Kb from the closest annotated gene.
Our data indicate that ectopic expression of PEG11 controlled by the Mlc3f/2E promoter and enhancer elements in mice affects muscle growth and development, but the precise molecular mechanisms remain elusive. PEG11/RTL1 is a sushi-ichi-related retrotransposon homolog and predicted to have a gag (capsid-like), an aspartyl protease, and two transmembrane domains [29,30]. PEG11 appears to be preferentially localized proximal to nuclei of the endothelial cells of fetal capillaries in the labyrinth zone of placenta [14], and associated with the nuclear fraction in skeletal muscle of callipyge sheep [11]. It has been speculated that the hepatocarcinogenic effect of PEG11 overexpression might result from cleavage of an extracellular matrix component mediated by a speculative aspartyl protease activity of PEG11 [15]. Possible connections, if any, between these observations and an effect on muscle mass are unclear.
In conclusion, this work provides the first direct evidence that excess PEG11 as well as DLK1 in post-natal skeletal muscle contributes to the muscular hypertrophy of callipyge sheep. Additional transgenic mouse lines expressing PEG11 and/or DLK1 at levels equivalent to callipyge sheep would facilitate further analyses to elucidate the mechanisms underlying the muscular hypertrophy in adult skeletal muscle. That might provide new insight useful for animal breeding as well as finding new therapeutic targets to treat muscle loss related to disease and age.

Oligonucleotide information
All oligonucleotide sequences are listed in S4 Table. Transgene construction The complete ovine PEG11 ORF was amplified by PCR as seven partially overlapping fragments from BAC 229G11 [6,31,32]. These were digested at endogenous restriction enzyme sites (BspeI, BglI, NsiI, AatII, BspHI, MluI) to generate paired cohesive ends which allowed for the reconstitution by ligation of a 4,178 bp fragment spanning from codon two to 175 bp downstream of the TGA stop codon of the ovine PEG11 gene (oPEG11). The corresponding fragment was digested with BglII and ligated in frame with the vector's ATG initiation codon of the pMlc3F-nlacZ-2E vector [9]. A 7.5 Kb fragment encompassing Mlc 3F promotor, oPEG11 ORF, bovine GH polyadenylation signal and Mlc 2E enhancer was released from the remainder of the vector by digestion with NotI and SmaI, subjected to agarose gel electrophoresis, and purified with the Geneclean Spin kit (MP Biomedicals, Santa Ana, CA).

Generation of transgenic mice
Transgenic mice were generated as described previously by Davis et al. [9]. Briefly, the purified DNA fragment was microinjected into the pronuclei of fertilized inbred FVB/N (Charles River, Wilmington, MA) mouse eggs at the concentration of 1 ng/μl via standard techniques. Injected eggs were then surgically implanted into pseudopregnant B6CBAF1 (Charles River) foster mothers. The transgene-positive founders were mated with wild type FVB mice to confirm transmission of the transgene upon PCR analysis of tail DNAs from their offspring. To evaluate transgene expression, we performed RT-PCR targeting the oPEG11 ORF using total RNA extracted from quadriceps femoris skeletal muscles of two offspring per transmitting founder.

Ethical statement
All animal procedures were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of University of Liège and approved by the Animal Ethics Committees of University of Liège (Permit Number: LA1610002).

Characterizing the transgene insertion sites
Southern blotting. Genomic DNA was extracted from mouse spleen using standard procedures; proteinase K digestion followed by phenol/chloroform purification and ethanol precipitation. Ten micrograms of genomic DNA were digested with SpeI and/or MfeI overnight at 37°C, and separated on a 0.8% agarose gel. The gel was equilibrated in the transfer buffer (0.4N NaOH and 1M NaCl), and DNA was then capillary-transferred to the Hybond N+ membrane (GE Healthcare, Buckinghamshire, UK). DNA probes for Southern hybridization were obtained by PCR amplifications of the vector used for generating transgenic mouse using primers; PEG11-probe1-S1 and-A1 (mouse line 126), or PEG11-probe2-S1 and-S2 (line 127A). The PCR products were then labeled with [α-32 P] dCTP with the Prime-It II random primer labeling kit (Agilent, Santa Clara, CA). The membrane was hybridized with approximately 10 6 cpm/ml of the probe in the ULTRAhyb ultrasensitive hybridization buffer (Ambion, Austin, TX) at 55°C overnight. After washing twice with 2 × SSC and 0.1% SDS buffer at 55°C for 5 min and three times at 60°C for 15 min with 0.1 × SSC and 0.1% SDS, the membrane was exposed overnight to a phosphor screen which was subsequently scanned with the Typhoon 9400 instrument (GE Healthcare).
PCR. PCR amplifications using genomic DNAs or the vector DNA were conducted with the Phusion Hot Start II High-Fidelity DNA Polymerase (Finnzymes, Espoo, Finland) according to the manufacturer's instructions under the following conditions: (1) 98°C 2 min, (2) 5 cycles of 98°C for 10 s, 62°C for 20 s and 72°C for 50 s, (3) 5 cycles of 98°C for 10 s, 60°C for 10 s and 72°C for 50 s, (4) 25 cycles of 98°C for 10 s, 58°C for 10 s and 72°C for 40 s, followed by (5) a 5 min final extension at 72°C. The PCR products were analyzed by electrophoresis on a 1.5% agarose gel.
Splinkerette PCR. Splinkerette PCR was performed according to Potter & Luo [17]. Briefly, 1 μg of genomic DNA was digested with FatI (for line 126) or ApoI (for line 127A), and ligated to a corresponding linker consisting of double stranded splinkerette oligonucleotides with a stable hairpin loop at one side and a compatible cohesive end at another side (SPLK-A and SPLK-B oligonucleotides were used for FatI digestion, and SPLK-A and SPLK-E for ApoI digestion, respectively). Two rounds of nested PCRs were conducted using primers; SPLA1 with either S7 (line 126) or S9 (line 127A) for the first round of PCR, and SPLA2 with either S8 (line 126) or S10 (line 127A) for the second round of PCR. The PCR products were gel-purified and sequenced using a standard Sanger sequencing procedure to determine integration sites of the oPEG11 transgene in the mouse genome.

Genotyping assays for mouse lines 126 and 127A
According to the information of the 126 and 127A transgene integration sites, we designed allele-specific PCR assays for both lines (S1 Fig). PCRs targeting wild-type and transgene alleles were performed separately using the Go-Taq DNA polymerase (Promega, Madison, WI) under the following condition: (1) 95°C 3 min, (2) 35 cycles of 95°C for 30 s, 60°C for 30 s and 72°C for 30 s, followed by (3) a final extension at 72°C for 5 min. Amplification products were pooled and detected by electrophoresis on a 2.5% agarose gel.
Characterizing the expression of oPEG11 transgenes in mouse lines 126 and 127A at the RNA level RNA extraction. Total RNA was extracted using the Trizol (Invitrogen, Carlsbad, CA) and treated with the Turbo DNase (Ambion) prior to RT-PCR or RACE according to the manufacturer's instructions.
Northern blotting. Northern blotting was performed essentially by following Green and Sambrook [33]. Briefly, 20 μg of total RNA was denatured at 65°C for 15 min in three volumes of the NorthernMax formaldehyde load dye (Ambion) and subjected to electrophoresis on a 1% agarose gel containing 2.4 M formaldehyde in 1 x MOPS buffer (Lonza, Basel, Switzerland) at 5 V/cm for~5 h. The gel was soaked in 0.01 N NaOH/3 M NaCl for 20 min, and RNA was transferred to the Hybond N+ membrane using upward capillary flow of 10 × SSC buffer (150 mM sodium citrate, 1.5 M NaCl, pH 7.0) for 1 hr. RNA was cross-linked on the membrane using the Stratalinker UV crosslinker (Stratagene, La Jolla, CA). For preparing probes, part of oPEG11 ORF (199-bp) was PCR-amplified using sheep muscle cDNA and primers (antisense primers were appended with a T7 RNA polymerase promoter), ovPEG_P_sen1 and _ans1_T7, and purified with the DNA Clean & Concentrator-5 (Zymo research, Irvine, CA). RNA probe was synthesized using the MAXIscript kit (Ambion) from one microgram of the PCR product with T7 RNA polymerase with [α-32 P] UTP according to the manufacturer's instructions. After DNase I treatment and column purification using the ProbeQuant G-50 micro column (GE Healthcare), approximately 1 × 10 6 cpm/ml of the RNA probe was used for hybridization at 68°C overnight. The membrane was washed twice at 70°C for 10 min in 2 x SSC and 0.1% SDS, then 3~5 times at 70°C for 30 min in 0.1 x SSC and 0.1% SDS, exposed to a phosphor screen for a few days and scanned with the Typhoon 9400 instrument. RNA probe targeting mouse β-actin (for control) was prepared from a plasmid DNA supplied in the MAXIscript kit.
RT-PCR. cDNA synthesis was performed using 1 μg of total RNA with 40 pmol of oligo (dT) 18 , 40 pmol of random hexamer and 1 pmol each of two oPEG11-specific primers (A4 and PEG-cDNA2) using the RevertAid H Minus First Stand cDNA Synthesis kit (Fermentas, Burlington, Canada) according to the manufacturer's instructions. PCR amplification was performed using the Phusion Hot Start II High-Fidelity DNA Polymerase under the following conditions: (1) 98°C 1 min, (2) 5 cycles of 98°C for 10 s, 65°C for 20 s and 72°C for 90 s, (3) 5 cycles of 98°C for 10 s, 60°C for 10 s and 72°C for 80 s, (4) 25 cycles of 98°C for 10 s, 58°C for 10 s and 72°C for 70 s, followed by (5) 5 min final extension at 72°C. The PCR products were purified after electrophoresis on 1.5% agarose gels and validated by sequencing.
qRT-PCR. cDNA was synthesized from 1 μg of total RNA using 50 pmol each of oligo (dT) 18 and random hexamers. QRT-PCR assay was performed using the Absolute Blue QPCR SYBR Green ROX Mix (ABgene, Surrey, UK) according to the manufacturer's instructions on the 7900HT Real-Time PCR System (Applied Biosystems, Foster City, CA) under the following conditions: (1) initial denaturing at 95°C for 15 min, followed by (2) 40 cycles of 95°C 15 s, 60°C 30 s, and 72°C 30 s, and (3) a melting curve analysis with 95°C 15 s, 60°C 15 s, and 95°C 15 s. Expression levels of target genes were normalized relatively to the expression levels of two control genes (Hypoxanthine phosphoribosyltransferase 1 and Actin gamma 1) that were selected out of five tested housekeeping genes (Splicing factor 3a subunit 1, Actin beta, Topoisomerase DNA II beta) using 2-and 10-week old mouse quadriceps muscles according to Vandesompele et al. [34].
RACE. cDNAs for 5' and 3' RACE were synthesized with the GeneRacer Kit (invitrogen) from five and one micrograms of total RNA, respectively, according to the manufacturer's instructions. To identify a 5' end of transcript (5' RACE), cDNA was amplified using GeneR-ACE 5' and oPEG11-specific 5GSP-O primers for the first round of PCR, and GeneRACE 5' Nested and oPEG11-specific 5GSP-I primers for the second round of PCR. To identify a polyadenylation site (3' RACE), GeneRACE 3' and oPEG11-specific 3GSP primers were used. The

Phenotyping
Mice were weaned at 21 days of age and kept with 4~5 mice per cage. Mice were weighted at weaning and then weekly until 10 weeks of age. Animals were euthanized at 10 weeks of age (range: 70-74 days) and dissected. We measured carcass (corresponding to skinned body with head, ankles, tail, internal organs, and associated fat removed), hind legs (corresponding to skinned limbs between hip and ankle), quadriceps femoris muscles, triceps brachialis muscles, kidney, spleen, liver and heart weights. Experimentalists were unaware of the animals' genotypes upon dissection.

Statistical analysis of mouse phenotype
Data points for which the residuals (i.e. phenotype corrected for fixed and random effects) were greater or equal to 3 or less than or equal to -3 times the residual standard deviation were considered as outliers and omitted from further analyses. Ratios (phenotypes divided by live weight) were subject to angular transformation (i.e. arcsin ffiffiffiffiffiffiffiffiffi ratio p À Á Â 180 Ä π) prior to analysis. Phenotypic values were analyzed with a mixed model including sex and genotype (and live weight) as fixed effects and litter as random effect. The effect of genotype accounted for multiple comparisons was calculated with the Šidák correction. Effects in possible contrasts of genotypes were computed with the Tukey honest signifcant difference test. Statistical analyses were conducted with R [35] (packages: lme4 [36], lmerTest [37], multcomp [38]) and SAS (SAS, Cary, NC).

Morphometric analysis
Extensor digitorum longus (EDL) and soleus (SL) muscles were dissected from male mice at 7 and 10 months of age. Both muscles from an individual were pinned up on a cork plate, fixed in 4% formaldehyde at room temperature for approximately four hours, cut at the midpoint perpendicularly to the long axis of the muscles, and embedded in a paraffin block so that the cutting planes were oriented for sectioning. Transverse sections (two sections per muscle) were prepared at a thickness of 5 μm and stained with hematoxylin and eosin by a standard procedure. Digital images of the entire muscle cross-sectional area were acquired at 20x magnification using the FSX100 microscope (Olympus, Center Valley, PA). To estimate myofiber crosssectional size, approximately 500 myofibers per section so as to cover the entire area of the muscle were manually framed and their minimal Feret's diameters were measured as a geometrical parameter for myofiber cross-sectional size [21] using the ImageJ software [39]. The fiber diameter was analyzed using a linear regression model containing genotype as a fixed effect and its statistical significance of contrasts between genotypes was computed using the Tukey's honest significant difference method using R (package stat [40]).