Comparison of prenatal muscle tissue expression profiles of two pig breeds differing in muscle characteristics

Comparison of prenatal muscle tissue expression profiles of two pig breeds differing in muscle characteristics M. Cagnazzo, M. F. W. te Pas, J. Priem, A. A. C. de Wit, M. H. Pool, R. Davoli and V. Russo J ANIM SCI 2006, 84:1-10.

The online version of this article, along with updated information and services, is located on the World Wide Web at: http://www.journalofanimalscience.org/content/84/1/1

www.asas.org

Downloaded from www.journalofanimalscience.org at Wageningen UR Library on November 6, 2013

Supplementary Material

Supplementary material can be found at: http://www.journalofanimalscience.org/content/suppl/2005/12/16/84.1.1.D C1.html

References

This article cites 28 articles, 6 of which you can access for free at: http://www.journalofanimalscience.org/content/84/1/1#BIBL

Citations

This article has been cited by 7 HighWire-hosted articles: http://www.journalofanimalscience.org/content/84/1/1#otherarticles

Downloaded from www.journalofanimalscience.org at Wageningen UR Library on November 6, 2013

Comparison of prenatal muscle tissue expression profiles of two pig breeds differing in muscle characteristics1 M. Cagnazzo,*†2 M. F. W. te Pas,†2 J. Priem,† A. A. C. de Wit,† M. H. Pool,† R. Davoli,*3 and V. Russo* *DIPROVAL University of Bologna, Sezione Allevamenti Zootecnici, Italy; and †Wageningen University and Research Centre, Animal Sciences Group, ID-Lelystad, Division of Animal Resource Development, Animal Genomics Group, Lelystad, The Netherlands

ABSTRACT: The objective of this study was to compare purebred Duroc and Pietrain prenatal muscle tissue transcriptome expression levels at different stages of prenatal development to gain insight into the differences in muscle tissue development in these pig breeds. Commercial western pig breeds have been selected for muscle growth for the past 2 decades. Pig breeds differ for their muscle phenotypes (i.e., myofiber numbers and myofiber types). Duroc and Pietrain pig breeds are extremes; Duroc pigs have redder muscle fiber types with more intramuscular fat, and Pietrain pigs have fastergrowing and whiter muscle fiber types. Pietrain pigs are more muscular than Duroc pigs, whereas Duroc pigs are fatter than Pietrain pigs. The genomic background underlying these breed-specific differences is poorly known. Myogenesis is a complex exclusive prenatal process involving proliferation and differentiation (i.e., fusion) of precursor cells called myoblasts. We investigated the difference in the prenatal muscle-specific transcriptome profiles of Duroc and Pietrain pigs using microarray technology. The microarray contained more than 500 genes affecting myogenesis, energy metabo-

lism, muscle structural genes, and other genes from a porcine muscle cDNA library. The results indicated that the expression of the myogenesis-related genes was greater in early Duroc embryos than in early Pietrain embryos (14 to 49 d of gestation), whereas the opposite was found in late embryos (63 to 91 d of gestation). These findings suggest that the myogenesis process is more intense in early Duroc embryos than in Pietrain embryos but that myogenesis is more intense in late Pietrain fetuses than in Duroc fetuses. Transcriptomes of muscle structural genes followed that pattern. The energy metabolism genes were expressed at a higher level in prenatal Pietrain pigs than in prenatal Duroc pigs, except for d 35, when the opposite situation was found. Fatty acid metabolism genes were expressed at a higher level in early (14 to 49 d of gestation) Duroc embryos than in Pietrain embryos. Better understanding of the genomic regulation of tissue formation leads to improved knowledge of the genome under selection and may lead to directed breed-specific changes in the future.

Key words: embryo, expression profile, microarray, muscle, pig breed, transcriptome 2006 American Society of Animal Science. All rights reserved.

INTRODUCTION

J. Anim. Sci. 2006. 84:1–10

As a result, pig breeds differ in muscle traits such as muscularity, muscle fiber type, color, etc. Duroc pigs are slower growing with relatively high i.m. fat content (Sellier, 1998) and redder muscle fiber types. Pietrain pigs are faster-growing pigs with relatively low i.m. fat content (Jones, 1998; Sellier, 1998) and whiter muscle fiber types. In addition, overall fatness of Duroc pigs is greater than that of Pietrain pigs. These 2 breeds are considered to represent extremes of modern Western pig breeds.

Pig breeding during recent decades has focused on improving growth rate and muscularity (Merks, 2000). 1

The authors acknowledge the contributions and helpful discussions of the colleagues in the PorDictor EU project: K. Wimmers (University Bonn, Germany); K. C. Chang and N. de Costa (University Glasgow, UK); J. Merks and B. Harlizius (Institute for Pig Genetics (IPG); Beuningen, The Netherlands; and H. Henne (Zu¨chtungszentrale Deutsches Hybridschwein Gmbh, Lueneburg, Germany). The prenatal pig samples were delivered from IPG and Zu¨chtungszentrale Deutsches Hybridschwein Gmbh, Lueneburg, Germany. The Pietrain prenatal samples were isolated by K. Wimmers. This work was financially supported by an EU grant under Contract No. QLK5-CT-2000001363 with additional finances from the director of ID-Lelystad. 2 These authors contributed equally to this work.

3

Corresponding author: [email protected] Received January 18, 2005. Accepted August 25, 2005.

1

2

Cagnazzo et al.

Many genes regulating myogenesis are known. Often these genes affect the activity and/or expression of the muscle regulatory factors (MRF) and myocyte-specific enhancer binding factor2 (Mef2) gene families, which take central positions in the regulation of myogenesis. The MRF gene family consists of 4 genes regulating proliferation (myf-5 and MyoD), terminal differentiation (myogenin), and maintenance of muscle tissue (Olson, 1990; Weintraub et al., 1991). The Mef2 genes (Mef2-A,-B,-C,-D) are members of the MADS gene family, and Mef2-D is present in undifferentiated myoblasts and may participate in the earliest commitment events leading to myogenesis (Wagenknecht et al., 2003). The Mef2 proteins are expressed in many tissues, but Mef2 only activates transcription in developing muscle. Using microarray technology, we studied expression of the many known genes simultaneously affecting myogenesis. Microarray technology can simultaneously measure the differential expression of a large number of genes in a given tissue and may identify the genes related to the different muscle phenotypes. The aim of this study was to compare purebred Duroc and Pietrain prenatal muscle tissue transcriptome expression levels at different stages of prenatal development to get insight into the differences in muscle tissue development in these pig breeds.

known effects on myogenesis, energy metabolism, and muscle structural genes. Some of the genes were members of larger gene families (See Supplemental Table; available online at http://jas.fass.org). A few other members of such gene families also were included in the microarray. The genes were partially cloned using a pool of RNA consisting of 2 samples per gestational age. Gene-specific primers were constructed based on sequence information from GenBank and used to amplify a PCR fragment on this cDNA pool. The fragments were cloned and partially sequenced to verify identity when doubt on identity existed. Furthermore, 309 clones of an adult pig muscle cDNA library (Davoli et al., 1999, 2002) were added. GenBank accession numbers, references, primers, and pathway information of all genes on the microarray are supplied in the Supplemental Table. The clones were amplified in four 50-␮L PCR reactions to obtain 200 ␮L total volume of PCR product for each clone, collected in a single well of a 96-well microtiter plate, and purified through Sephadex G100 (Sigma Aldrich, Zwijndrecht, The Netherlands). The purified products were precipitated and resuspended in 20 ␮L spotting buffer (1M phosphate buffer, pH 5.8; 50% dimethyl sulfoxide). After that, the length and quality of each fragment were checked on a 1% agarose gel. The fragments were spotted on glass slides in duplicate.

MATERIALS AND METHODS Animals Eighty-nine purebred Duroc and 87 Pietrain sows or gilts were bred with Duroc or Pietrain purebred boars, respectively. The sows were slaughtered around d 14, 21, 35, 49, 63, 77, and 91 after insemination, and the embryos/fetuses were collected. The litters of at least 14 sows per gestational age were collected. All handling of sows, embryos, and fetuses was done according to Dutch laws. The gestational ages of the litters varied for 4 d around the sampling ages. Fourteen-day-old embryos were collected by flushing the uterus. Thus, these samples represented all the embryos in a uterine horn. Twenty-one-day-old embryos were individually collected as whole embryos. Embryonic length was determined for each embryo. Fetuses of 35 to 91 d were weighed. For embryos with gestational age of 49 to 91 d, the gender was determined. Although the LM itself was not clearly visible at 35 d of development, the area where the muscle developed appeared and was extracted. Longissimus muscle tissue was dissected from the 49- to 91-d-old embryos. All collected materials were snap frozen in liquid N2 and stored at −80°C until RNA isolation. The gestational ages covered the whole period of myogenesis.

Microarray Construction The microarray contained PCR fragments of 509 genes. A literature survey resulted in 200 genes with

Microarray Hybridization The Trizol-Phenol (Life Technologies, Breda, The Netherlands) method was used to isolate RNA from 2 embryos or fetuses per breed, except for the 14-d gestational age stage, for which Qiagen RNeasy kit (Qiagen, Cologne, Germany) was used. The RNA samples were pooled, and 2 ␮g of RNA for each stage and breed was labeled with Cy3 or Cy5 using the TSA labeling and amplification kit protocol (Perkin Elmer Life Sciences, Inc., Langen, Germany). Each microarray was hybridized with samples from a single gestational age with a sample from Duroc labeled with one dye and a sample from Pietrain labeled with the other dye. For each gestational age, hybridizations were performed in duplicate and in duplicate dye swap. A total of 7 gestational ages were investigated; thus, the number of microarray hybridizations was 28. Hybridization was done for 16 h at 65°C. After hybridization, the slides were rinsed according to the stringent washing protocol recommended by the manufacturer.

Microarray Analysis Microarrays were scanned using the GeneTac2000 scanner (Genomics Solutions, Ann Arbor, MI). Each microarray was analyzed independently using the following steps. The first step was normalization of raw scanning data, which included background correction. The background signal was determined using blank

Pig breed muscle transcriptome differences

spots. The background was normalized by including 33 blank spots per patch, which were evenly distributed. Blank spots are spots for which only spot buffer was used for printing and no DNA material. The blanks spots are therefore a better reference to account for nonspecific expression than local background surrounding expressed spots. Beyond that, we found that the number of lost spots attributable to negative values after background correction was less using blank spots for background correction. Normalization also was done 1) using all spots, and 2) per patch, all intensity-dependent using a locally weighted scatterplot smoothing (LOWESS) fit (Cleveland, 1974; Park et al., 2003), following the procedure described by Yang et al. (2002) and Pool et al. (2003). Spots are represented by their M- and A-values: M = log2 (Cy5/Cy3); thus, the M-value indicates the differential expression between the Cy5 and Cy3 labeled RNA samples; and A = (log2 [Cy5 * Cy3])/2. Therefore, the A-value indicates a weighted mean expression level of the Cy5- and Cy3-labeled RNA samples (Yang et al., 2002). Additionally, the significance of the difference is indicated by the 2-sided Pvalue tested on a log-logistic distribution. The second step in microarray analysis was that normalized spots with the difference Cy3 − Cy5 having P > 0.05 were discarded. The third step was that the remaining spots were analyzed for up- and/or downregulation of expression by comparing the differential expression in the breeds using the M-values, and for mean expression levels (i.e., A-values). The M-values were regarded significant if −1.58 < M >1.58 (i.e., if the differential expression is larger by at least 3 times, as recommended by the manufacturer of the labeling kit; Perkin Elmer). Using the Spotfire Pro 7 software (BioASP, Amsterdam, The Netherlands), clustering was performed with the K-means clustering option after a principal component analysis using the M- and A-values. Clustering is based on A-values using Euclid distances. The software was arbitrarily set to divide the spots into 8 clusters. Finally, the last step was that the results were analyzed with biological interpretation of data using available data on the physiology of the genes (i.e., involved in energy metabolism, myogenesis, or muscle structural genes). The P-values were calculated without accounting for multiple comparisons; thus, there is the need for protection against Type I errors. The Bonferroni test was not used because it is too conservative, especially for microarray data. We avoided the risk of Type I errors by considering multiple copies of the clones on the array. The false error rate of a single spot of a gene is 0.05 and is 0.05 × 0.05 = 0.0025 for 2 spots. If no effect (false positive), positive measurements are expected to occur only in 1 spot of a gene (with the chance to occur in 2 spots also being extremely small). To be included in the list of up- or downregulated genes, a gene should show 1 to 4 spots significant per microarray on at least 2 microarrays, of which at least 1 microarray should be a dye swap microarray, making a Type I error extremely

3

unexpected. Therefore, finding multiple significant upor downregulated copies of a gene on the array increases the overall chance of a gene to be differentially expressed on the array (i.e., the more copies are found significant, the smaller the risk of detecting falsely significant designated genes due to Type I errors). Furthermore, the chance of having a false positive upregulated copy would be as large as for a false positive downregulation. If the experiment had no power at all, one would expect as many false positive down-regulations as upregulations. We have accounted for this by suggesting that a group of genes only has an effect when the number of either up- or down-regulated genes is more than one-half of the genes within that group. Therefore, we believe that we avoided the risk of Type I errors in this experiment.

Quantitative Real-Time PCR To validate the results of the microarrays, 5 genes showing differential expression were selected and analyzed with real-time PCR using the Lightcycler (Roche Diagnostics, Almere, The Netherlands). Genes were chosen in different pathways: myogenesis-affecting (EPO-receptor, β-catenin, and TGFβ2), energy metabolism (GAPDH), and muscle structural (collagen 3A1). Primers were designed on the cDNA sequence to amplify a 100 to 150 bp fragment, and probes containing fluorescein were designed according to the rules set by the manufacturer (Table 1). All reactions had an annealing temperature of 60°C, except TGF-β2 (55°C), and a magnesium concentration of 3 mM, except βcatenin (5 mM). For real-time PCR, each gestational age was represented by individual RNA samples isolated from 6 embryos or fetuses different from the animals and litters used for microarray analysis. Mean values are presented. Quantitative results are usually presented normalized against an internal control, often house-keeping genes such as GAPDH or beta-actin. In our results, we show that these genes are also highly regulated during these prenatal differentiation events. It is clear that such genes cannot be used to normalize the expression of other genes (Radonic´ et al., 2004). In the microarray, normalization as a function of all genes on the microarray was used, which cannot be done with real-time PCR. Therefore, we also tried 18S rRNA; however, we showed that 18S rRNA also is strongly regulated during the 14and 21-d gestational ages, as has been shown previously (Voronina, 2002; Radonic´ et al., 2004). Using 18S rRNA normalization, the 35- to 91-d period resulted in unrealistic differences between 21 and 35 d because of extremely low 18S rRNA levels (data not shown). This period is a very important stage in primary muscle differentiation. Therefore, we decided to use nonnormalized real-time PCR values. The mean value of 6 animals is presented. The results are presented as the log(Duroc/Pietrain); the sign of the results indicates the ratio and the direction of the differential expression

4

Cagnazzo et al.

Table 1. Primers and probes used for real-time PCR verification of 5 genes showing differential expression in the microarray analysis Gene

EMBL1

Type2

Sequence

Location (nt)3

GAPDH

AF017079

Col3A1

AU059332

Epo-receptor

AF274305

TGF-β2

X70142

β-catenin

AB046171

Forward Reverse FL LC640 Forward Reverse FL LC640 Forward Reverse FL LC640 Forward Reverse FL LC640 Forward Reverse FL LC640

GGCTCTCCAGAACATCATCC CCCAGCATCAAAGGTAGAAGA CCATGCCAGTGAGCTTCCCG GAGCTCAGGGATGACCTTGCCC GCCATCCAGGACAACCAG ATCGGGACTAATGAGGCTTTC GCTTTTTCACCTCCAACACCAGCG GGCAGCAGCCCCACCACC GAACCAGCCGCAGATGATG CCAGAGCAGATGAGCAGAAGG CACAGCCTGGTGGTGATTTGGAC GGCGGCCATGGATGAAGCC CAACCGGCGGAAGAAG CGTTTTGCCGATGTAGTAGAG TAAACCCAGAAGCTTCTGCTTCCCC GCTGCGTGTCCCAGGATTTAGAACC GACGCTGCTCATCCCAC CAGCGAGCCGTTTCTACA GCTTCCAGACATGCCATCATGCG CTCCTCAGATGGTGTCTGCAATTGTACGT

938–957 1,229–1,209 1,029–1,009 1,007–986 26–43 243–223 125–102 99–82 1,198–1,216 1,393–1,373 1,319–1,341 1,343–1,362 123–138 426–406 344–368 371–395 543–559 838–821 747–769 772–800

1 EMBL = European Molecular Biology Laboratory. Forward and Reverse are the PCR primers. FL = probe with the fluorescein group attached to the 3′-end; LC640 = probe with the red fluorochrome LC640 attached to the 5′-end, and the 3′-end contains a phosphate group. 2 http://www.ebi.ac.uk/embl/ 3 Location (nucleotide numbers) of the primers and probes in the sequence of the accession number in the EMBL database.

between Duroc and Pietrain. It is important to note that no new data are produced with real-time PCR. These experiments were only used to verify the microarray data.

RESULTS Microarray Analysis Genes were grouped into 3 major groups: myogenesis, energy metabolism, and muscle structural genes. The first 2 groups have been subdivided into pathway-specific subgroups (Table 1). Results were analyzed for 1) up- or downregulation (i.e., the ratio between the expression level in Duroc and the expression level in Pietrain; M-value), and 2) for general expression level (A-value).

Duroc-Pietrain Expression Ratio Data were analyzed for spots with P < 0.05 and for spots showing more than 3 times differential expression. Although the number of spots is substantially lower in the latter analysis, the results for each group are similar. The results are shown in Table 2. Column N shows the total number of genes on a microarray for each (sub)group, whereas column N-list indicates the total number of genes for each (sub)group of genes showing statistically significant differential expression on at least 1 gestational age. To be included, a gene should show 1 to 4 spots significant per microarray on at least

2 microarrays, of which at least 1 microarray should be a dye swap microarray. A (sub)group was considered significant when at least 50% of the genes was differentially expressed on at least 1 gestational age. Note that especially in the myogenesis group, genes may be included in 2 subgroups (e.g., differentiation-inhibiting and proliferation-stimulating effects). Energy Metabolism. The results (Table 2) indicate a major difference between the energy metabolism in Duroc and Pietrain embryos and fetuses at all gestational ages. At all gestational ages except 35 d of gestation, the energy metabolism in the Pietrain was at a greater level than in the Duroc. At d 35 the situation was reversed. The genes of the fatty acid metabolism had a different general expression profile. The results indicate that fatty acid metabolism was at a higher level in early Duroc embryos (d 14 to 49 of gestational age) compared with Pietrain embryos, whereas the reverse situation was found in older fetuses from d 63 of gestation and onward. Finally, with the exception of the ATP metabolism subgroup, all subgroups had lower numbers of differentially expressed genes at 91 d of gestation. This finding may indicate that at birth the Duroc and Pietrain breeds differ less in energy metabolism. Myogenesis. Table 2 indicates that, in general, myogenesis started earlier in Duroc than in Pietrain because the expression levels of all myoblast proliferation and differentiation affecting groups of genes were greater in Duroc embryos of 14 to 35 d than the expres-

5

Pig breed muscle transcriptome differences

Table 2. Numbers of genes showing differential expression within groups of genes related to energy metabolism, myogenesis, and muscle structure at 7 gestational ages covering the whole period of myogenesis Gestational age, d Gene function grouping Energy metabolism genes ATP metabolism Oxidative phosphorylation Glycolysis Fatty acid metabolism Miscellaneous Myogenesis affecting Differentiation (Diff) stimulating Differentiation inhibiting Proliferation (Prol) stimulating Proliferation inhibiting Diff/Prol affecting Migration regulating Structural genes Muscle structural

N1

N-list2

61 16 13 14 9 10 113 31 30 24 4 29 6 62 62

54 15 12 12 5 10 81 27 30 22 2 19 6 54 54

14

21

35

49

63

77

91