HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Physiol. Genomics 28: 253-272, 2007. First published October 31, 2006; doi:10.1152/physiolgenomics.00121.2006
1094-8341/07 $8.00

This Article Free upon publication Abstract Full Text (PDF) All Versions of this Article: 28/3/253    most recent 00121.2006v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by Vuocolo, T. Articles by Tellam, R. L. Search for Related Content PubMed PubMed Citation Articles by Vuocolo, T. Articles by Tellam, R. L.

Identification of a gene network contributing to hypertrophy in callipyge skeletal muscle

¹ Commonwealth Scientific and Industrial Research Organisation Livestock Industries, Queensland Bioscience Precinct, St. Lucia, Queensland
² Department of Veterinary Science, The University of Melbourne, Melbourne, Victoria, Australia
³ Utah State University, Logan, Utah

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION GRANTS REFERENCES

 
The callipyge mutation in sheep results in postnatal skeletal muscle hypertrophy in the pelvic limbs and loins with little or no effect on anterior skeletal muscles. Associated with the phenotype are changes in the expression of a number of imprinted genes flanking the site of the mutation, which lies in an intergenic region at the telomeric end of ovine chromosome 18. The manner in which these local changes in gene expression are translated into muscle hypertrophy is not known. Microarray-based transcriptional profiling was used to identify differentially expressed genes in longissimus dorsi skeletal muscle samples taken at birth and 12 wk of age from callipyge and wild-type sheep. The phenotype was only expressed at the latter developmental time and associated with decreased type 1 fibers (slow oxidative) and a shift toward type IIx and IIb fibers (fast-twitch glycolytic). We have identified 131 genes in the samples taken at 12 wk of age that were differentially expressed as a function of genotype but not due to the fiber type changes. The gene expression changes occurring as a function of genotype in the samples taken at birth indicated that the transcriptional framework underpinning the phenotype was emerging prior to expression of the phenotype. Eight genes were differentially expressed as a function of genotype at both developmental times. A model is proposed describing a core network of genes and histone epigenetic modifications that is likely to underpin the fiber type changes and muscle hypertrophy characteristic of callipyge sheep.

Dlk1; microarray; histone; epigenetics; histone deacetylase-9

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION GRANTS REFERENCES

 
AN UNDERSTANDING OF THE ROLES of key proteins and genes that regulate myogenesis, particularly differentiation and hypertrophy, is crucial to the development of novel strategies for modulating muscle mass. Studies of gene expression in myogenic cell lines and skeletal muscle tissues have been extensive. These, in combination with a number of transgenic mouse models showing overexpression and knockout of specific genes as well as naturally occurring mutations, have provided considerable insight into the transcriptional mechanisms regulating muscle cell differentiation and function. Notwithstanding these studies, details of the complex multigene processes controlling muscle fiber type expression and muscle hypertrophy are not yet clear.

Callipyge sheep are characterized by a rostrocaudal gradient of skeletal muscle hypertrophy expressed in specific muscles principally located in the pelvic limbs and loins and to a much lesser extent the thoracic limbs (12, 15, 22, 28–30, 38). The affected muscles, which are not apparent until 2–3 mo after birth, are increased in size by up to 40% and characterized by an increase in type IIb fibers (fast-twitch glycolytic) (17, 35, 38). The muscles of these animals are also leaner, and there is a 10% increase in feed efficiency (28). The causative point mutation (A/G), which was originally identified by Freking et al. (23) and confirmed by Smit et al. (63), is located at the telomeric end of ovine chromosome 18 in an intergenic region between the protein encoding gene Dlk1 and the noncoding gene Gtl2. This region is contained within a cluster of highly conserved imprinted genes that includes Begain, Dlk-1, Dat, Meg3, Peg11, Peg11as, Meg8, and Dio3 (13, 14, 52, 58, 59, 67, 71, 72) (Fig. 1).

View larger version (8K): [in this window] [in a new window]   Fig. 1. Diagrammatic representation of the organization of imprinted genes located at the telomeric end of ovine chromosome 18. Paternally expressed genes are highlighted by black cross-hatch shading and maternally expressed genes by grey shading. An arrow indicates the direction of transcription of each gene. The asterisk shows the position of the callipyge mutation, and the triangle denotes the position of an intergenic differentially methylated region controlling maternal gene expression. The diagram is based on that deduced by Georges et al. (24) and supplemented by other studies (13, 14, 52, 58, 59, 67, 72), as well as annotation for a miRNA cluster (Mirg) deduced by comparative sequence analyses with the orthologous murine and human sequence regions.

The aims of the current study were to identify genes that were differentially expressed in callipyge skeletal muscle compared with wild-type muscle and subsequently link these changes with the muscle hypertrophy phenotype. Microarrays and quantitative RT-PCR were used for transcription profiling of skeletal muscle samples taken at birth, when the phenotype was not expressed, and 12 wk of age from callipyge and wild-type sheep. The gene expression changes that underlie the expression of the phenotype rather than resulting from the fiber type change in the affected muscle have been identified. A working model that links the muscle hypertrophy phenotype with a core group of transcriptional regulators is proposed.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION GRANTS REFERENCES

 
Biological samples.
All animal protocols were submitted to and approved by the Utah State University animal ethics committee. Skeletal muscle samples were obtained from a flock of Dorset-Suffolk-Rambouillet cross-bred sheep raised at Utah State University. Matings were conducted to produce the genotypes for transcript profiling comparisons, particularly the genotypes NN and NC^pat, where NN is the wild-type and NC^pat is the paternal heterozygote. The genotypes of all animals were confirmed by genotyping (23, 63). Longissimus dorsi skeletal muscle samples (LD) were obtained from 32 male lambs representing four of each genotype, within 5 days of birth (T0) and 11–12 wk of age (T12). In addition, samples were also taken of semimembranosus (SM) and semitendonosus (ST) skeletal muscles from wild-type lambs at 12 wk of age and an unaffected muscle, supraspinatus (SS), from both the NN and NC^pat genotypes at T12. The effect of genotype on the characteristics of these muscles has been extensively reported (12, 15, 17, 22, 28–30, 35, 38). All muscles of interest were dissected from the animal within 15 min of euthanasia and weighed, and samples were collected at predetermined sites before being frozen in liquid nitrogen. Muscle samples were also mounted onto cork blocks using Tragacanth Gum (5% wt/vol) in a transverse orientation and snap-frozen in liquid nitrogen-quenched isopentane. Animals were reared and euthanized in accordance with the animal ethics guidelines of Utah State University (Logan, UT).

Isolation of RNA and preparation of cDNA.
Total RNA was extracted from 4 g of each muscle sample tissue with TRIzol reagent (Invitrogen), followed by DNase1 (Ambion) treatment. The RNA was then further purified with an RNeasy Mini Kit (Qiagen, Basel, Switzerland) including an on-column DNase1 treatment (Qiagen DNase I) to remove traces of genomic DNA. The RNA was quantified by spectrophotometric measurements at 260 and 280 nm, and its integrity was verified by the optical density at 260 and 280 nm (OD₂₆₀/OD₂₈₀) absorption ratio (>1.8) and by visualization on an agarose gel. cDNA synthesis was performed with either total RNA isolated from each muscle sample from an individual or pooled RNA derived by combining equal quantities of total RNA from the same muscle type and developmental time for the four individuals corresponding to each genotype. cDNA synthesis was undertaken with 5 µg of isolated RNA per sample using MMLV Superscript III reverse transcriptase (Invitrogen) and an anchored oligo-T₁₈ primer combined with random hexamers (70).

Quantitative real-time reverse transcription PCR.
Quantitative real-time reverse transcription PCR (qRT-PCR) with the Sybr Green-based fluorescent detection system and the ABI Prism 7900 Sequence Detection System (PE Applied Biosystems, Foster City, CA) was used to measure mRNA abundance (65, 69). A constant amount of cDNA, corresponding to 10 ng of reverse transcribed RNA derived from each sample was used. Four technical replicates were performed for each gene investigated. This process allowed quantification of the target gene relative to a constant reference gene in each sample using threshold cycle (Ct) data. 18S ribosomal RNA (18S rRNA, GenBank accession AY779625) was used as the reference gene following demonstration that its expression was constant in all samples. Primer pairs are listed in Table 1. Database searches, alignments, and sequence analyses were performed with the aid of National Center for Biotechnology Information (NCBI, http://www.ncbi.nlm.nih.gov/) and interactive bovine in silico single nucleotide polymorphism (IBISS, http://www.livestockgenomics.csiro.au/ibiss/) systems.

Microarray experimental design.
Gene expression analyses were primarily performed on LD from wild-type (NN) and callipyge (NC^pat) sheep using Bovine Affymetrix GeneChip microarrays and qRT-PCR. Two developmental time points were investigated in this study: newborn (within 5 days of birth; T0) and 11–12 wk postbirth (T12). The muscle hypertrophy phenotype developed over the first 2–3 mo and was associated with a significant change in muscle fiber type (12, 15, 16, 35). One of the objectives of the gene expression analysis was to discriminate between those genes that underlie the muscle hypertrophy in callipyge sheep and those that directly result from the fiber type change in the affected muscle. To address this issue a comparison was undertaken at T12 of gene expression in wild-type skeletal muscles with differing fiber type compositions to identify fiber type-specific genes. The three skeletal muscles used in this analysis were SM, ST, and LD (8). Samples from the C^matN (maternal heterozygote) and callipyge homozygote (CC) genotypes at T0 and T12 as well as the unaffected muscle SS for the NN and NC^pat genotypes were also used in the initial analysis to provide increased confidence in all reporting probe sets (flag calls; see below) but data from these additional analyses are not discussed herein. The overall experimental design is shown in Fig. 2. The circles reflect all samples that contributed to the initial data analysis (i.e., contributed to flag calls), whereas the shaded circles represent samples that contributed the gene expression data and analysis discussed herein.

View larger version (17K): [in this window] [in a new window]   Fig. 2. Microarray experimental design. Bovine GeneChip microarrays were used to analyze gene expression in ovine skeletal muscle samples. The primary samples were derived from longissimus dorsi skeletal (LD) muscles at T0 (within 5 days of birth) and T12 (11–12 wk) for the NN (wild-type), NCpat(paternal heterozygote), CmatN, and CC genotypes. Each sample shown by a circle represents an RNA sample from 4 individuals of the same genotype, muscle type, and age analyzed in duplicate. Initial analyses were also performed on additional samples from SM and ST skeletal muscles from NN animals at T12, and samples from SS skeletal muscle (an unaffected muscle) at T12 for the NCpat and NN genotypes. In all, 24 microarrays contributed to the primary data set. The confidence of the GeneChip flag calls (present, marginal, and absent) for each probe set was increased by inclusion of primary data from all samples. Subsequently, data derived from only the shaded subset of samples were further analyzed, as outlined in EXPERIMENTAL PROCEDURES.

Microarray transcriptional profiling.
Ovine RNA samples were analyzed with the Bovine GeneChip microarray (Affymetrix), which contains 24,027 bovine probe sets representing 19,000 UniGene clusters (Bovine UniGene Build 57, March 24, 2004; http://www.affymetrix.com) and 101 probe sets representing control elements. A preliminary investigation of the use of the microarray with ovine samples was performed using five different ovine tissues (each in triplicate), and in addition, a direct comparison was made between ovine and bovine LD samples (manuscript in preparation). MAS5.0 software was used for data analysis, which delivers both a quantitative expression value and a flag call defining confidence in probe set detection i.e., "present" (P), "marginal" (M), or "absent" (A). Comparison between ovine and bovine LD muscle RNA samples revealed that 75% of the probe sets reporting P in the bovine skeletal muscle sample were also efficiently reporting gene expression with the comparable ovine sample. The absolute signals for the common reporting probe sets were similar (r² = 0.87). There is a mean transcript sequence identity of 97% between ovine and bovine orthologs, which probably underlies the usefulness of the microarray with ovine samples (36). Nonreporting probe sets therefore reflect lack of gene expression, poor performance of the probe set with an ovine sample, or high variation in the signal derived from the probe set. Other studies have also successfully used a variety of species-specific Affymetrix GeneChips with samples from closely related species (31). RNA processing, reverse transcription, labeled cRNA preparation and hybridization to the bovine GeneChip were all performed according to the manufacturer's recommendations. All quality control measures built into the bovine GeneChip and its associated Test Array were met. Microarray data from this experiment were submitted to the NCBI's Gene Expression Omnibus according to MIAME standards. The data can be accessed at NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/).

Microarray data processing.
The user-defined expression parameters were set to Affymetrix default values for image signal processing. The gene expression signal for a specific probe set was calculated from the weighted average of all probe signals in a probe set using one-step Tukey's biweight estimates. The weight of each probe was determined by how close its signal was to the mean value of the probe set. Hence, the software deemphasizes outliers. P, M, and A flag calls were determined by a Wilcoxon signed-rank test comparing the perfect match and mismatch probe pair intensities, where "presence" was assigned a P value < 0.05. All chips were scaled (global scaling) to a target intensity of 200 to minimize differences between chips caused by physical differences in chips, hybridization efficiencies, and manual laboratory procedures (10). No probe masking was applied.

Statistical analyses of microarray data.
Data analysis only considered probe sets that were called P by MAS5.0 in at least 75% of the GeneChips from the initial and larger analysis that included additional data from the remaining two genotypes (C^matN and CC) at T0 and T12, as well as data from the following samples: ST and SM (NN genotype at T12); and SS (NN and NC^pat genotypes at T12). A total of 24 GeneChips therefore contributed to the overall analysis although the results from only a subset of this data set are reported herein. Signals from all GeneChips were log-transformed and further normalized by fitting an analysis of variance (ANOVA) mixed model. Mixed ANOVA models provide a more general framework when there are more than two conditions to compare and embody a general and powerful approach to allow full utilization of the information available in microarray experiments with multiple factors and/or a hierarchy of sources of variation (18, 53). The model included the fixed effect of GeneChip and flag (P, M, or A) within GeneChip with 72 levels (from 24 chips by 3 flags), and the random effects of probe (with 24,128 levels), probe x genotype x age/muscle (with 289,536 levels), and residual (i.e., within probe variance). Restricted maximum likelihood estimates of variance components were obtained using analytical gradients with VCE software (http://www.tzv.fal.de/e.g./vce4/vce4.html). In terms of model parameters, measures of differential gene expression were obtained from the t-statistic computed by the difference in normalized expression [or best linear unbiased prediction (BLUP)] of each gene at the two conditions of genotype by age being contrasted. To identify differentially expressed genes, the linear combinations of BLUPs defining the contrasts of interest were processed through model-based clustering via a mixture of distributions with EMMIX software (44, 53). In addition, differentially expressed genes were clustered using PermutMatrix software (11). The human gene symbols associated with the lists of differentially expressed genes and those annotated on the bovine GeneChip were analyzed by GOstat (http://gostat.wehi.edu.au/) using default parameters (5). Fisher's exact test was used to determine significance. P values <0.05 were considered significant.

Annotation of the bovine GeneChip.
The supplied annotations for the Affymetrix Bovine GeneChip are limited, as only 14,589 of the 24,027 probe sets were definitive. To enhance the level of annotation, the Affymetrix consensus sequences that formed the basis of probe set designs were used to search the GenBank human nucleic acid and protein reference sequence datasets at NCBI using BLASTN and BLASTX with default settings [Expect (E) < 1.00E-30], where E was >1.00E-30. Further sequence annotations were undertaken by BLASTN analysis of IBISSv4 clustered bovine EST sequences using default settings (25). In this case E scores <1.00E-20 were accepted. The annotations associated with the IBISSv4 contigs were then used. All annotations were validated by a second annotation strategy whereby the Affymetrix probe set sequences were mapped to the bovine genome assembly contigs (version 1), and then these were used to identify corresponding human genes (34). Using this process, we found the definitive annotation rate to increase from 14,589 to 16,777 of the 24,016 probe sets. The unidentified probe set sequences probably represent divergent 3'-untranslated regions (UTRs), species-specific splice variants, and novel transcripts. The expression patterns of the myosin heavy chain isoforms were of particular interest as markers for muscle fiber types. However, some myosin heavy chain probe sets did not physically differentiate between isoforms (e.g., Bt.12300.1.S1.a_at and Bt.4867.1.A1_at). Hence, qRT-PCR analyses were used to supplement the Affymetrix GeneChip array results for these genes. In addition, qRT-PCR was used to supplement the microarray results for specific genes of interest that were not reporting on the GeneChip and to validate the differentially expressed genes identified by microarray analysis. Gene symbols are shown in italics but when used in the context of the encoded protein they are denoted by normal text.

Tissue histology and immunolocalizations.
Serial 8-µm sections of the frozen blocks were collected onto room temperature slides. The orientation of each block was confirmed as transverse by hematoxylin and eosin staining. Slides were stored at –20°C until required for staining. These sections were then stained with nicotinamide adenine dinucleotide-nitro-blue tetrazolium reductase (NADH-TR) to differentiate between oxidative and glycolytic fibers (51). In the former case the staining highlights slow oxidative (type I) and fast oxidative glycolytic (type IIa) fibers. The sections were also used to calculate average myofibril cross-sectional area. Muscle sections were incubated in NADH-TR solution for 30 min at 37°C, followed by three water washes and three exchanges of increasing (30, 60, and 90% vol/vol) and decreasing acetone solutions (30 s each). The slides were rinsed with water and mounted with polyvinyl acetate aqueous mounting media. A total of 10 nonoverlapping images were randomly collected from each NADH-TR-stained section using an automated XY-microscope stage controlled by Stage Pro software. All fibers in each image were counted in three categories: light (fast glycolytic), intermediate, and dark (slow oxidative). The number of fibers in each category was then expressed as a percentage of the total number of fibers in each section. The data from the 10 separate images were pooled in each case.

Immunohistochemical staining to identify type I (slow oxidative) muscle fibers was performed by firstly bringing sections to room temperature and rehydrating them in PBS for 20 min. Sections were then incubated with 5% (vol/vol) goat or donkey serum in wash buffer (PBS containing 0.5% BSA and 0.1% saponin) for 60 min. They were then incubated with the two primary antibodies, mouse anti-slow human myosin heavy chain (MAB1628, Chemicon) and rabbit anti-mouse laminin (Pan-laminin staining, Sigma), each diluted 1:200 in wash buffer and incubated overnight at 4°C. The sections were then washed (3x) in wash buffer. The secondary antibodies, anti-mouse Ig-ALEXA488 or anti-rabbit Ig-ALEXA594 (Molecular Probes), were each diluted 1:250 in wash buffer and incubated with the sections for 60 min at room temperature. Sections were again washed (3x) in wash buffer. Nuclei were then stained using Hoechst 33342 for 20 s, and sections were mounted with a permanent aqueous mountant (polyvinyl alcohol). Slides were stored in the dark at 4°C until analysis. Control staining (in the absence of primary antibodies) of a second section on each slide was performed to confirm the specificity of staining. The sections were examined under an Axiophot fluorescent microscope (Zeiss, Oberkochen, Germany). The cross-sectional area for each fiber type was calculated by tracing its perimeter using Image Pro Plus (v5.1) image analysis software. A minimum of 250 fibers were measured for each sample. Images were false pseudocolored with V++ software and were optimized for brightness, contrast, and gamma correction without altering the original interpretation. All data are presented as means ± SE. Between-group differences were analyzed using one-way ANOVA; P < 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION GRANTS REFERENCES

 
Figure 1 shows a diagrammatic representation of the organization of the imprinted genes surrounding the site of the callipyge mutation. Several of these genes are poorly defined as they do not encode proteins and potentially undergo extensive mRNA splicing and/or processing. The overall experimental design used for microarray analysis of the callipyge muscle samples is shown in Fig. 2, although as discussed in the EXPERIMENTAL PROCEDURES, only a subset of the data (shaded circles) is reported herein. To further validate the use of the bovine GeneChip with ovine samples we have employed qRT-PCR using ovine-specific primers for 16 genes to measure their relative gene expression in ovine NN and NC^pat LD samples at T0 and T12 and correlated this information with data obtained from the bovine Affymetrix GeneChip for the same genes and samples (Fig. 3). The data are plotted on equivalent scales i.e., (Ct_NCpat – Ct_NN) vs. log₂(NC^pat/NN)_GeneChip, respectively (31). The correlation coefficient for the line drawn through the data is 0.91, indicating that the gene expression data derived from the two different platforms are strongly correlated. In view of the use of multiple short (25 bp) oligonucleotides in the design of the probe sets used on the Affymetrix GeneChip and their potential for cross-species sequence variation, it is likely that the demonstrated usefulness of the bovine GeneChip with ovine samples is directly attributable to the selection of probe sets with minimal sequence variation across species and the influence of the MAS5.0 analysis software, which de-emphasizes individual probes with outlying intensities.

View larger version (9K): [in this window] [in a new window]   Fig. 3. Relationship between gene expression data measured by the bovine GeneChip and quantitative real-time reverse transcription (qRT)-PCR. Data for 16 genes were obtained from ovine LD samples analyzed using the bovine GeneChip and by qRT-PCR at T0 and T12. The data are presented on equivalent scales i.e., log2 (fold change), where the fold change was measured between the NCpat and NN genotypes. Ct, threshold cycle measured by qRT-PCR.

View larger version (25K): [in this window] [in a new window]   Fig. 4. Expression of Dlk-1, Gtl2, and Peg11 plus Peg11as in longissimus dorsi skeletal muscle samples from NCpat and NN lambs. Histograms show the mean expression of Dlk1, Gtl2, Peg11+Peg11as, MEG8 and the reference genes Rplp0 (acidic ribosomal protein) and Gpd1 (glyceraldehyde 3 P dehydrogenase 1) relative to 18S ribosomal RNA in LD muscle for samples taken from individual animals as measured by qRT-PCR. Expression levels are reported for two developmental time points, T0 and T12. Error bars for each animal denote 1 SD. Below each genotype group are the mean ± SD. NN (dark shading); NCpat (light shading). The qRT-PCR assay measured expression of both Peg11 and Peg11as, which showed expression in all 4 NN animals, but their levels were low relative to the corresponding NCpat animals and therefore cannot be discerned in the figure.

View larger version (40K): [in this window] [in a new window]   Fig. 5. Changes in gene expression for LD skeletal muscle samples from NN and NCpat lambs at T0 and T12. Bovine GeneChip microarrays were used to compare total RNA from ovine LD samples taken from NN and NCpat lambs at T0 and T12. Genes are listed in rows, and genotypes and developmental time points are shown in columns. Column 1, T0 NCpat; column 2, T0 NN; column 3, T12 NCpat; column 4, T12 NN. Red and green indicate increases and decreases in expression, respectively. The genes listed were selected on the basis of significant differential expression as a function of genotype at either T0 or T12. The expression of each of these genes at the alternative time point has also been included even though many show no significant changes. Coexpressed genes were clustered using PermutMatrix (11). Some genes were represented by >1 probe set. Anonymous genes are identified by their probe-set annotation.

Figure 5 also highlights the changes in developmental expression for many of these genes. Of the 159 unique probe sets that were differentially expressed as a function of genotype at either T0 or T12, 59 (37.1%) were also differentially expressed between T0 and T12 in the NN or NC^pat genotypes. A variety of different patterns of expression changes occurred during development for the genes that were differentially expressed as a function of genotype at either T0 or T12. Of the eight genes showing differential expression as a function of genotype at both T0 and T12, all except Hdac9 were also differentially expressed as a function of development within each genotype. Thus, the effect of genotype was strongly associated with those genes that also showed changes in expression during development.

Figure 6A shows the developmental changes for the genes that were differentially expressed at both time points. There is no single pattern to the developmental changes seen with these eight genes. While Gtl2, Dlk1, and Lrrc2 were upregulated at both times in the NC^pat genotype, Hdac9 was downregulated. Furthermore, Fos and Atf3 were both downregulated at T0 and upregulated at T12, whereas Rasd1 showed the reciprocal pattern. Clearly, the callipyge mutation is causing complex changes in the regulation of the expression of these eight genes as well as the full repertoires of differentially expressed genes at T0 and T12. Figure 6B shows similar patterns of differential expression when these eight genes were also examined by qRT-PCR. Thus, the differential expression of these genes as a function of genotype was independently confirmed.

View larger version (38K): [in this window] [in a new window]   Fig. 6. Changes in the expression of specific genes measured by GeneChip microarray analysis and qRT-PCR. A: changes in the expression of 8 genes that were differentially expressed at both T0 and T12 in NCpat (unshaded) compared with NN (shaded) LD samples. The ordinate shows the relative signal intensity (RSI) derived from GeneChip microarray data for each sample. The gene name is followed by the probe set designation in brackets. B: qRT-PCR data corresponding to the genes described in A. MNE, mean normalized expression. Opposite: C: changes in the expression of muscle structural genes at T0 and T12 in NCpat compared with NN LD muscle samples. Data were derived from GeneChip microarray analyses. D: changes in the expression of muscle structural genes at T0 and T12 in NCpat compared with NN LD samples. Data were derived from qRT-PCR analyses because the relevant probe sets for these genes were not present on the GeneChip, not reporting from the GeneChip presumably due to ovine-bovine sequence differences or the relevant probe sets were not specific for individual members of a gene family.

The LD sample at T0 also showed significant expression of myosin heavy chain embryonic (Myh3) and myosin heavy chain fetal (Myh8). There was significantly less Myh8 expression in the callipyge sample at T0 (0.21-fold change) but no genotypic difference in the expression of Myh3. For both of these myosin isoforms there was significant and strong downregulation of their expression between T0 and T12 (ranging between 0.048- and 0.009-fold change). These data are consistent with their predominant fetal-like expression patterns (data not shown).

In addition to the differential expression of the myosin heavy chain isoforms as a function of genotype, there is also differential expression of a number of myosin binding proteins. In particular, myosin light chain 2 (Mylk2) is upregulated in the NC^pat genotype by 4.3-fold at T12. This gene, like Myh4, is specifically associated with type IIb muscle fibers. A similar pattern is also seen for myosin-binding protein H (Mybph), but its association with the type IIb and IIx fiber types has not been previously documented. Myosin light chain 3 (Myl3), which is characteristic of type IIa fibers was downregulated at both time points.

Changes in fiber type defined by histochemistry and immunolocalization.
Previous studies using histochemical staining have demonstrated a shift toward type IIb muscle fibers in callipyge skeletal muscle, which only appears 2–3 mo after birth (28–30). Figure 7 indirectly confirms this conclusion at T12 with NADH-TR histochemistry, which is a measure of mitochondrial oxidative activity, a characteristic of type 1 slow oxidative fibers and type IIa fast oxidative/glycolytic fibers. There was significantly less staining in the NC^pat compared with the NN LD samples at T12 (P < 0.05) (Fig. 7, C and D). The mean percentages of positively staining fibers in LD samples from the NN and NC^pat genotypes were significantly different (32.6 and 16.6%, respectively; P < 0.01). The histochemical results are consistent with a shift from a more oxidative fiber (type 1 and type IIa fibers) to a glycolytic fiber (type IIb and type IIx fibers). In the LD muscle at birth there was no significant genotypic difference in the histochemical staining patterns.

View larger version (167K): [in this window] [in a new window]   Fig. 7. Muscle fiber type histochemistry. Nicotinamide adenine dinucleotide-nitro-blue tetrazolium reductase (NADH-TR) histochemistry was performed on LD muscles from animals sampled at T0 (A, B) and T12 (C, D). A and C: NN genotype; B and D: NCpat genotype.

View larger version (124K): [in this window] [in a new window]   Fig. 8. Immunohistochemical staining of myosin heavy chain slow (Myh7) in LD skeletal muscle. Myh7 was immunolocalized (green fluorescence) in LD muscle at T0 (A, B) and T12 (C, D). A and C: NN genotype; B and D: NCpat genotype. Laminin localization (red) in the endomysium was used to visualize myofibrils. Nuclei are stained blue.

View larger version (17K): [in this window] [in a new window]   Fig. 9. Identification of genes that were differentially expressed as a function of genotype at T12 but not due to changes in fiber type. The scales on both axes represent log10 (fold change). The horizontal axis shows the genotypic contrast for LD muscle at T12, while the vertical axis represents the muscle type comparison for the NN genotype at T12. Horizontal and vertical lines correspond to the 99% confidence intervals for the contrasts in the x- and y-axes, respectively. The genes identified by diamond symbols represent those that were differentially expressed at T12 in LD muscle as a function of the NCpat genotype but independent of skeletal muscle type. A: LD and ST contrast (both NN); B: LD and SM contrast (both NN); C: ST and SM contrast (both NN).

Gene ontology analysis.
Gene ontologies (GO) associated with the 131 genes that were differentially expressed as a function of genotype but independent of muscle type were used to identify over- and underrepresented GO terms using Gostat (5). The GO terms "muscle development" (GO:0007517, 5 genes) and "oxidoreductase activity" (GO:6546, 11 genes) were significantly overrepresented in the gene list (P = 0.001 and 0.004, respectively).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION GRANTS REFERENCES

 
The following discussion is predicated on the assumption that gene expression changes result in corresponding protein expression changes. Although this assumption may not always be true, in this study it is correct for Myh7 (Fig. 8) and Dlk1 (White J and Tellam RL, unpublished observations). It has also been assumed that the functions of ovine genes and their encoded proteins are the same as their orthologs in humans, mice, and other mammals.

The callipyge skeletal muscle hypertrophy phenotype is expressed 2–3 mo postbirth and has significant effects on the expression of six imprinted genes flanking the site of the mutation (6, 7, 13, 24, 48, 69). The current study has confirmed earlier findings of a fiber type change in rear muscles of callipyge sheep at this time (17, 35, 38) and importantly also implicated Myh1, the most predominant myosin heavy chain isoform. Increased expression of Myh4 was apparent, but it was expressed at markedly lower levels than Myh1. Thus, there is a shift toward fast glycolytic fibers, but this is primarily mediated by increased expression of Myh1 and largely at the expense of slow oxidative fibers. Since the muscle hypertrophy phenotype was only expressed postnatally, then it is possible that the callipyge mutation is affecting innervation sensitive gene pathways, which are normally activated by skeletal muscle use, particularly animal movement. Consistent with this transcriptional changes in myosin heavy chain genes occurred in the samples taken just after birth, before the muscle hypertrophy phenotype was apparent but at a time when extensive innervation was likely to be occurring.

It has been suggested that the mutation results in changes in both cis- and trans-acting regulatory systems at this locus, which cause coordinate increases in the expression of several imprinted genes while maintaining their imprinted expression status (13, 24). The paternally expressed gene Dlk1 is a strong candidate for the primary effector of the muscle hypertrophy phenotype as its differential expression as a function of genotype, development, muscle type, and muscle fiber type strongly corresponds with the emergence of the skeletal muscle hypertrophy phenotype (48, 69). Furthermore, overexpression of Dlk1 in mouse skeletal muscle results in muscle hypertrophy (19). Dlk1 is a member of a family of intercellular signaling proteins that have key roles in determining cell fate for multiple cell lineages (2). Some of these family members have been shown to be regulators of myogenesis, albeit often negative regulators (21). Members of this family have also been implicated in anterior-posterior segmentation during development, a process that has parallels with the rostrocaudal gradient of expression of the callipyge phenotype (55). However, the genetic mechanism causing the increased Dlk1 expression in callipyge sheep and the gene expression differences that underlie the muscle hypertrophy phenotype are unknown. The current study has identified genes that contribute to the expression of the callipyge phenotype and are postulated to be the result of the change in expression of Dlk1. (Although there is strong evidence that Dlk1 is the primary effector of the phenotype, the involvement of Peg11, a paternally expressed retrotransposon-like gene, cannot be ruled out.)

The current study has demonstrated significant gene expression changes in NC^pat lambs in both the T12 sample, where the muscle hypertrophy phenotype was expressed, and also in the T0 sample, where the gross muscle phenotype was not apparent. In the latter instance there were significant differences in myosin heavy chain isoform expression both at the transcriptional and protein levels, indicating that the basis for the muscle hypertrophy phenotype was emerging at this time. Presumably, the gene expression changes occurring in the sample taken at T0 provided the developmental framework that underpinned subsequent gene expression changes resulting in the hypertrophy phenotype at T12. The restricted number of genes differentially expressed at both T0 and T12 suggests that the callipyge mutation may change a developmental program between T0 and T12. Indeed, a disproportionate number of genes that were differentially expressed as a function of genotype at either T0 or T12 were also differentially expressed during normal development over this period.

A number of major themes are apparent in the differentially expressed genes at T0 and T12. There is substantially enhanced expression of the imprinted genes Dlk1 and Gtl2, which flank the site of the callipyge mutation. qRT-PCR has also demonstrated that Peg11+Peg11as and Meg8 are also upregulated in the NC^pat genotype. These results confirm and extend previous information (7, 13, 69). Importantly, in the NC^pat genotype, the increased expression of Gtl2 and Meg8, two normally maternally expressed genes, indicates that either the normally silent paternal alleles are activated or trans-acting factors derived from the paternal locus are upregulating the maternal alleles. The differential expression of Dlk1 at both T0 and T12, in combination with evidence that the Dlk1 protein may be the effector responsible for initiating the muscle hypertrophy phenotype in NC^pat animals, suggests that other differentially expressed genes at both of these times may be functionally linked to Dlk1 and consequently core contributors to the muscle hypertrophy phenotype. Further emphasizing this point, of the eight genes identified by microarray analysis that were differentially expressed at both time points, six (Dlk1, Gtl2, Hdac9, Fos, Atf3, and Rasd1) were also included in the list of 131 differentially expressed genes independent of muscle type and, by extension, fiber type. Thus, these eight genes probably play a pivotal role in establishing and expressing the muscle hypertrophy phenotype in callipyge sheep.

The aberrant expression of multiple imprinted genes surrounding the callipyge mutation suggests a local change in the epigenetic status of this 1 Mbp region. Indeed, hypomethylation of the region surrounding the mutation has been demonstrated (49, 68). More extensive but indirect epigenetic changes controlling the expression of muscle-specific genes and mediated indirectly by increased Dlk1 expression may be a primary factor causing the muscle hypertrophy phenotype. Two dynamic and interdependent epigenetic mechanisms regulating gene expression are DNA methylation and posttranslational modifications, particularly acetylation, of core histones of the nucleosome (42). There is strong evidence that epigenetic changes regulating gene expression play a central role in myogenesis and muscle function (26, 42, 45). Treatment of a murine fibroblast cell line with an inhibitor of DNA methylation, 5-azacytidine, promoted a skeletal muscle phenotype, indicating that the skeletal muscle-specific program of gene expression is regulated by DNA methylation (32). Fos, one of the eight differentially expressed genes at both developmental times, has been implicated in the transformation of fibroblasts through alterations in DNA methylation and histone deacetylation, and its expression in neuronal cells is regulated by histone deacetylation (3, 64). Thus, Fos is intimately involved in regulation of the transcription of other genes via epigenetic mechanisms and is itself controlled in this manner.

The histone deacetylase Hdac9 was significantly downregulated at both developmental time points in callipyge samples. An in vitro cell culture model and a muscle injury/regeneration model demonstrated that histone deacetylase inhibitors increase muscle cell size by promotion of myoblast recruitment and fusion through the induction of the secreted protein follistatin (Fst) (27). However, these inhibitors had no effect on follistatin expression in uninjured muscle. While the mRNA level for Fst was unaffected in the NC^pat genotype, it is interesting and unexpected to note that myostatin (Gdf8) mRNA expression was significantly increased in the T12 NC^pat sample, but this was revealed only after consideration of fiber type changes associated with callipyge skeletal muscle. Myostatin, a myoblast secreted protein, is a strong negative regulator of myogenesis that binds follistatin, a positive regulator of myoblast recruitment and fusion. The interaction results in reciprocal inhibition of the activity of each protein (27, 39). The enhanced level of myostatin mRNA in callipyge samples concomitant with the expression of the muscle hypertrophy phenotype is unexpected and suggests that myostatin may have promyoblast fusion activity in uninjured and differentiated skeletal muscle.

Hdac9, a class II histone deacetylase, is characterized by a COOH-terminal histone deacetylase domain and an NH₂-terminal domain that binds Mef2, a pivotal transcription factor that potentiates the activities of the MyoD family of basic-helix-loop-helix transcription factors (45, 74). The latter proteins, in conjunction with Mef2, orchestrate the transcription of numerous muscle-specific genes and play a pivotal role in myogenesis (40, 42, 46, 74). Recently, it was demonstrated that there are specific Mef2 response elements in the promoter of murine Myh4, suggesting that this transcription factor promotes formation of type IIb fibers (1). In addition, Mef2 has also been implicated in myocyte hypertrophy (43). The interaction of Hdac9 with Mef2 represses the latter's ability to promote transcription of muscle-specific genes, thereby resulting in the suppression of myogenesis (40, 74). By analogy with other type II HDAC family members, the nuclear export sequence located in the COOH-terminal domain of Hdac9 coupled with site-specific phosphorylation mediated by calcium-dependent kinases may regulate translocation of the protein into the cytoplasm with consequent derepression of Mef2 and promotion of myogenesis (42). Closer inspection of the Hdac9 probe set on the GeneChip reveals that it is specific for a unique skeletal muscle-specific splice variant (Hdrp; Hdac-related protein; or MITR, Mef2-interacting transcription repressor). Hdrp encodes only the Mef2 binding domain and is the predominant Hdac9 transcript in skeletal muscle (74, 75; data not shown). This splice variant has also been shown to be a potent Mef2 repressor that presumably does not undergo nuclear-cytoplasmic trafficking due to the absence of a nuclear export signal sequence, although it is inactivated by phosphorylation (73). Furthermore, Hdrp has also been shown to recruit the class I histone deacetylases Hdac1 and Hdac3 (45, 74, 75). The latter enzyme is expressed in LD muscle (result not shown). The decreased expression of the Hdrp form of Hdac9 in LD muscle from NC^pat animals may result in less recruitment of Hdac3, and together this causes derepression of Mef2 activity and consequently enhanced transcription of muscle-specific genes. Interestingly, Hdac9 has also been shown to be downregulated upon skeletal muscle denervation, which is associated with a change in muscle fiber type (45). Thus, there is substantial indirect evidence implicating Hdac9 in the skeletal muscle hypertrophy phenotype expressed in callipyge skeletal muscle.

The mitogen-activated protein kinase (MAPK) family of calcium-dependent kinases is known to positively couple Mef2 activity to multiple signaling pathways and is intimately involved in myogenesis (43). Mapk6 was upregulated 3.5-fold in NC^pat skeletal muscle, suggesting that this enzyme, which is localized in the nucleus, may be phosphorylating Mef2 and enhancing its promyogenic transcriptional activity. Mapk6 could also phosphorylate Hdrp (Hdac9), thereby inactivating its Mef2 inhibitory activity.

Both Atf3 and Fos contribute to the formation of the activator protein 1 (AP1) transcription factor complex, which regulates the expression of a wide range of genes involved in control of cell proliferation, differentiation, and programmed cell death (60). AP1 is a heterodimer consisting of a complex array of family members from the Jun, Fos, Maf, and Atf families. The patterns of transcriptional changes for Fos and Atf3 in the callipyge samples at T0 and T12 were identical and therefore consistent with the view that the AP1 complex is involved in regulating gene expression changes in callipyge LD muscle. Atf3 is a member of the mammalian activation transcription factor/cAMP responsive-element binding family of transcription factors and is involved in the transcriptional repression of genes that contain an Atf binding element in their promoters (60). Hence, the decreased expression of Atf3 and Fos in the NC^pat genotype at birth may allow activation of a specific myogenic developmental program that underpins the callipyge phenotype at T12. The enhanced levels of Atf3 and Fos at T12 may then reflect the shutdown of this developmental program and emphasis on an alternative gene expression program that is involved in the maintenance of the callipyge phenotype. In support of this hypothesis it is noted that the genes that are differentially expressed as a function of genotype at T0 are, in general, considerably different from those at T12.

Rasd1 encodes a poorly defined Ras-related protein that is involved in dexamethasone-induced alterations in cell morphology, growth, and cell-extracellular matrix interactions and is presumably involved in intracellular signaling (33). A Ras-MAPK signaling pathway has been implicated in the induction of slow type I muscle fibers in regenerating muscle (47). Consistent with this Rasd1 is downregulated at T12 concomitant with a decrease in type I fibers.

Two strongly differentially expressed transcripts in NC^pat skeletal muscle at T12 are located in adjoining regions of the bovine genome. The transcript defined by the EST CK968244 (Bt.14371.1.A1_at) shows strong upregulation at both T0 and T12 (2.86- and 12.86-fold, respectively) and maps to the 3'-UTR of the maternally expressed imprinted gene Slc22a3 in the Igf2r imprinted locus. Pacrg is an anti-sense transcript to the Parkin gene (Park2), which is located close to Slc22a3, but its imprinted status, if any, has not been investigated. Pacrg is not differentially expressed at T0, but its expression at T12 shows a 0.08-fold change (i.e., 12.5-fold decrease) in callipyge LD samples. The functions of these transcripts are not clear, but these results suggest aberrations in the regulation of this imprinted locus.

The genes that are differentially expressed at both developmental times and the 131 genes differentially expressed at T12, but independent of muscle fiber type changes, are likely to underpin the muscle hypertrophy phenotype in NC^pat LD muscle. However, the mechanism whereby Dlk1, the proposed primary effector, causes this cascade of gene expression changes is unclear. The cell surface receptor for Dlk1 is thought to be Notch1, although it is noted that the former protein does not contain the DSL motif characteristic of Notch ligands and alternative receptors may be a possibility (55). The interaction between Dlk1 and Notch1 has been shown to result in repression of the latter's function (4). Notch signaling is intimately involved in somite formation, muscle development, and the proliferation and cell fate determination occurring during muscle regeneration (41, 55). Notch1 has also been shown to be a negative regulator of myogenesis (50, 61). Activation of Notch1 by intercellular ligand binding results in enzymatic cleavage and release of an intracellular fragment, Notch intracellular domain, which translocates to the nucleus where it binds and activates the CSL family of transcription factors (55). One of the targets for these transcription factors is Hes1, a negative regulator of MyoD-induced myogenesis (41, 57). Thus, increased quantities of Dlk1 expressed on the surface of one muscle cell could inhibit the activity of Notch1 on an adjacent cell, thereby derepressing a default myogenic pathway in that cell. One prediction of this model is that there would be suppression of direct Notch1 target genes. Consistent with this, the expression of Hes1 in NC^pat LD muscle at T12 was changed by 0.51-fold, although this was not significant. Notch1 is also known to generate intracellular signals independent of the CSL pathway (61). It is also suggested that Dlk1-mediated repression of Notch1 signaling in skeletal muscle alters the expression of key regulatory genes such as Atf3, Fos, Rasd1, Lrrc2, and Hdac9, thereby promoting the skeletal muscle gene expression program that ultimately leads to muscle hypertrophy in specific skeletal muscles of NC^pat sheep. Both Notch1 and Hes1 downregulate Dlk1 expression in murine preadipocytes, thereby highlighting the inverse relationship between Notch signaling and Dlk1 expression (54). This observation suggests that upon intercellular Dlk1-Notch1 interaction there may be bidirectional intracellular signaling.

Figure 10 proposes a simple working model describing a core group of regulatory proteins that are probably involved in generating the skeletal muscle hypertrophy in callipyge sheep. Many other proteins are certainly involved, but a fully integrated model involving all components is not possible at this time. The model is based on previously reported protein regulatory interactions combined with data from the present study. In summary, it is proposed that the increased expression of Dlk1 protein in NC^pat sheep represses Notch1 signaling in an adjacent cell, thereby altering the expression of key transcription factors that promote myogenesis, particularly at the level of muscle hypertrophy. It is hypothesized that the increased expression of the transcription factors Atf3 and Fos in NC^pat promote hypertrophy by undefined mechanisms, while decreased levels of Hdac9 in conjunction with inhibition of Hdac3 nuclear activity derepresses the activity of Mef2, a key transcription factor promoting the expression of type IIb muscle fibers and potentially also directly contributing to the increased content of type IIx fibers and hypertrophy. The model implies that histone modifications targeting Mef2-regulated genes may be the primary epigenetic processes underpinning the gene expression alterations contributing to these changes in muscle structure. One characteristic of this model and aspects of the previous discussion is the abundance of negative regulators. These may be highlighting the need in normal muscle for control circuits that allow simultaneous fine regulation of muscle function while maintaining the potential for considerable metabolic and structural plasticity as the muscle responds to external stimuli such as training, injury and aging (9).

View larger version (18K): [in this window] [in a new window]   Fig. 10. Proposed model showing the core proteins contributing to muscle hypertrophy in callipyge sheep. A simple model is proposed that relates some of the differentially expressed genes identified at both T0 and T12 with the muscle hypertrophy phenotype in callipyge sheep. The callipyge mutation results in enhanced expression of Dlk1 either by alteration in the activity of a long-range regulatory element or by a change in the epigenetic status of the imprinted gene locus at the telomeric end of chromosome 18. Dlk1 is proposed as the primary effector of the system and Notch1 is its receptor on an adjacent cell. Overexpression of Dlk1 results in inhibition of the activity of Notch1, a known negative regulator of myogenesis. This inhibition of Notch1 activity also results in enhanced activity of the AP1 transcription factor [activating transcription factor-3 (Atf3) and Fos] and decreased expression of histone deacetylase (Hdac)9. The decreased Hdac9 expression, possibly linked with concomitant translocation of Hdac3 from the nucleus to the cytoplasm, derepresses myocyte enhancer factor-2 (Mef2) activity, which in conjunction with myogenic regulatory factors (MRFs) such as myoblast determination protein (MyoD) promotes the formation of type IIb fast oxidative fibers and muscle hypertrophy (see DISCUSSION for references). A: NN; B: NCpat. The dashed lines show weaker linkages while the change in the size of a gene symbol reflects its relative expression level. The large horizontal arrow symbolizes the transcriptional changes that directly underpin muscle hypertrophy.

	GRANTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION GRANTS REFERENCES

<