Analysis of gene expression in non-regressed and regressed bovine corpus luteum tissue using a customized ovarian cDNA array

Theriogenology 64 (2005) 1963–1976 www.journals.elsevierhealth.com/periodicals/the

Analysis of gene expression in non-regressed and regressed bovine corpus luteum tissue using a customized ovarian cDNA array Orla M. Casey a,b, Dermot G. Morris a, Richard Powell b, Joseph M. Sreenan a, Richard Fitzpatrick a,* a

Animal Reproduction Department, Teagasc, Agriculture and Food Development Authority, Athenry, Galway, Ireland b Department of Microbiology, National University of Ireland, Galway, Ireland

Received 1 February 2005; received in revised form 25 April 2005; accepted 25 April 2005

Abstract The lifespan of the bovine corpus luteum (CL) is an important factor in the control of normal ovarian cyclicity and the establishment and maintenance of pregnancy. There is increasing evidence that CL lifespan is regulated by alternative expression of genes that promote or inhibit luteolysis. To gain further insights into these events a 434 character ovarian cDNA array comprising genes attributed to key aspects of CL function including more than 100 anonymous expressed sequence tags (ESTs) was constructed and screened with a33P dATP labeled RNA isolated from non-regressed (n = 6) and regressed (n = 6) CL tissue. Significance analysis of microarrays (SAM) identified 15 genes that changed expression 1.7-fold ormorewitha false discovery rate of <5%. The differentiallyexpressed genesencodedenzymes involved in steroid biosynthesis and oxygen radical metabolism and proteins involved in extracellular matrix remodeling, apoptosis and cell structure. Results for five of the differentially expressed genes including matrix gla protein and collagen a1(I) (extracellular matrix), glutathione-S-transferase aI (oxygen metabolism), clusterin (apoptosis) and scavenger receptor BI (steroid biosynthesis) were confirmed by Northern blot analysis and found to be significantly different (P < 0.01) between non-regressed and regressed CL tissue.Collectively this study identified geneswith recognized roles in CL regression, genes with potential roles in this process and genes whose function have yet to be defined in this event. # 2005 Elsevier Inc. All rights reserved. Keywords: Bovine (Bos taurus); Corpus luteum; Luteolysis; cDNA array; Northern blots * Corresponding author. Tel.: +353 91 845849; fax: +353 91 845847. E-mail address: [email protected] (R. Fitzpatrick). 0093-691X/$ – see front matter # 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.theriogenology.2005.04.015

1964

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

1. Introduction The bovine corpus luteum (CL) is a transient endocrine gland with an average lifespan of 17–18 days [1]. Its primary function is to secrete progesterone, which is necessary for regulation of the estrous cycle and for the establishment and maintenance of pregnancy [2]. The CL undergoes luteolysis during the follicular phase of the normal estrous cycle, a process that involves both functional and structural regression. Functional regression, characterized by cessation of progesterone production, is necessary for final maturation and ovulation of the dominant follicle in the subsequent cycle [3]. Structural regression, characterized by destruction and removal of luteal cells is necessary to maintain the ovary in proportion to the rest of the reproductive tract and to facilitate subsequent preovulatory follicle growth [4]. Inadequate progesterone production by the developing CL and or premature CL regression are associated with embryo loss [5,6] while persistent CLs that fail to regress result in irregular estrous cycles in cattle [7]. An improved understanding of the molecular mechanisms regulating CL regression is clearly an important objective in reproductive biology with potential for the development of more efficient estrous cycle synchronization methods and for the enhancement of fertility in cattle and other farm species. The lifespan of the bovine CL is principally determined by the luteolytic action of uterine produced prostaglandin F2a (PGF2a) which induces CL regression [8,9]. However, the inability of PGF2a to induce luteolysis during early stages (up to day four) of the estrous cycle suggests that endocrine, paracrine and autocrine factors also act in concert with PGF2a to bring about regression of the CL [10]. Luteal regression is initiated by a series of morphological and biochemical changes in a variety of cell types including large and small luteal cells, fibroblasts, endothelial and immune cells and ultimately leads to cessation of steroidogenic capacity, cell death and extensive tissue involution [2]. These events are likely to be mediated by alternative expression of key genes that either promote or inhibit CL regression. For example, spontaneous luteolysis or luteolysis following administration of PGF2a have been associated with decreased expression of genes involved in steroidogenesis [11], angiogenesis [12], oxygen metabolism [13], and increased mRNA expression for vasoactive peptides [14], inflammatory cytokines [15] and genes involved in apoptosis [16]. Gene specific techniques including reverse transcription polymerase chain reaction (RTPCR), in situ hybridization and Northern blot analysis have in the past been used to study gene expression in the bovine CL. These approaches are limited in that only a restricted number of genes can be studied concurrently. This limitation can now be overcome using DNA arrays which permit large-scale analysis of gene expression and is thus becoming a popular technique for evaluating global gene expression and tissue function. Recent studies show the potential of this approach for elucidating the genes and mechanisms that regulate aspects of bovine reproduction [17–19] and immunology [20]. The potential of DNA array technology to identify pro- and anti-luteolytic related genes in rat CL tissue has also been reported [21]. The aim of this study was to improve our understanding of the mechanisms regulating the lifespan of the bovine CL by evaluating gene expression differences in nonregressed and spontaneously regressed CL tissue using a customized 434 character ovarian cDNA array.

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

1965

2. Materials and methods 2.1. Corpus luteum collection and characterization The corpora lutea were collected from heifers that had been scanned for reproductive normality and had exhibited at least one normal cycle prior to their allocation to the experiment. Following twice daily checks for the occurrence of estrus (day 0), CL tissue was collected from heifers immediately after slaughter from day 16 to day 19 of the estrous cycle. Average time between animal slaughter and tissue storage was 45 min. The corpora lutea were dissected from the ovaries, washed in RNase-free phosphate buffer, immediately frozen in liquid nitrogen and stored at
80 8C until RNA extraction. The average weight for the non-regressed CLs was 5.99  0.48 g and 4.74  1.10 g for the regressed CLs. The physiological status of the CLs was confirmed by measurement of systemic concentrations of progesterone using RIA (Coat-a-Count, Diagnostic Products Corporation, USA) and by the detection of oligonucleosome formation, an indicator of apoptotic cell death and structural regression. For this purpose, genomic DNA was isolated from individual CL tissues using DNeasy tissue kit according to manufacturers instructions (Qiagen Ltd., Crawley, UK) and separated (4 mg/sample) by electrophoresis through a 2% agarose gel. Gels were stained with SYBR Green I (Molecular Probes, Leiden, The Netherlands) and scanned using a Molecular Imager FX (Bio-Rad, Hercules, CA). 2.2. RNA isolation Total RNA was prepared from fragmented frozen tissue using the TRIzol reagent (Gibco BRL Life Technologies Ltd., Renfrewshire, UK). RNA quality was assessed by examining the 28S and 18S ribosomal RNA bands on ethidium bromide stained 1% agarose gels. RNA quantity was determined by absorbance at 260 nm and the degree of protein contamination assessed by absorbance at 260 and 280 nm. RNA quality and quantity were also assessed using automated capillary gel electrophoresis on a Bioanalyzer 2100 with RNA 6000 Nano Labchips according to manufacturers instructions (Agilent, Waldbronn, Germany). 2.3. Ovarian cDNA array construction A total of 434 ovarian cDNAs comprising 351 and 83 non-redundant clones derived from CL and ovarian cortex cDNA libraries, respectively, [22,23] were amplified by the polymerase chain reaction using primers T3 (50 -AATTAACCCTCACTAAAGGG-30 ) and T7 (50 -GTAATACGACTCACTATAGGGC-30 ). The PCR reactions were conducted in 96 well plates in a final volume of 25 ml and contained 0.2 ml of bacterial culture, 7.5 pmol primer, 0.2 mM dNTPs, 2 mM MgCl2 and 0.33 U Taq polymerase (Qbiogene Inc., Cambridge, UK). The process was started by a hot start of 5 min at 92 8C followed by 30 cycles consisting of 94 8C for 10 s, 58 8C for 30 s, 72 8C for 2 min with a final extension of 72 8C for 5 min. PCR efficiency in terms of quality and quantity of the amplified products was determined by 1% ethidium bromide agarose gel electrophoresis. The cDNA arrays were constructed by RZPD, Berlin, Germany. Briefly this involved printing 50–100 ng of each PCR product in a 3  3 pattern onto 8 cm  12 cm Hybond N + nylon membranes

1966

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

(Amersham Pharmacia Biotech, NJ, USA) using a Picking-Spotting Robot (Linear Drives LDT) with 400-mm pins (Genetix, Hampshire, UK). All clones were spotted in duplicate to improve reproducibility. After spotting, the arrays were carefully floated for 2 min on 0.4 M NaOH and 5 SSC (pH 7.5), air-dried and cross-linked by UV (120 mJ/cm2) using the Stratagene UV-Stratalinker 2400 (Stratagene, Amsterdam, The Netherlands). The constitutively expressed bovine GAPDH gene and the Arabidopsis thaliana CAB, RCA and rbcL genes (Stratagene, Europe) were also spotted in duplicate at four distinct positions and served as positive and spiking controls, respectively. 2.4. RNA labeling and hybridization The number of CL replicates used for the array experimentation were six per treatment group. A total of 5 mg of total RNA isolated from individual non-regressed (n = 6) and regressed (n = 6) CL tissues were reverse transcribed with 500 ng of anchored oligo (dT)18 (Ambion Europe Ltd., Cambridgeshire, UK) and 4 ml of a33P dATP (3000 Ci/mMol) (Amersham) using the Strip-EZ RT kit (Ambion Europe Ltd.) according to manufacturers instructions. RNA samples were spiked with 2 ng of A. thaliana CAB, RCA and rbcL mRNA prior to reverse transcription and served as internal controls for probe labeling. Unincorporated nucleotides were removed using NucAway spin columns (Ambion Europe Ltd.) and radiolabel incorporation measured using a Victor multilabel counter (Wallac, Turku, Finland). Membranes were prehybridized in a Hybaid, HS 9360 hybridization oven for 4 h at 60 8C in 10 ml of ULTRArray hybridization buffer (Ambion Europe Ltd.) containing 3 mg of (dA)40 oligonucleotide and 50 mg/ml of denatured herring sperm DNA. Following pre-hybridization, approximately 1  106 cpm/ml of heat-denatured (95 8C for 5 min) radiolabeled cDNA probe was added to 10 ml of fresh hybridization buffer and incubated overnight at 60 8C. Following overnight hybridization, membranes were washed twice in 2 SSC, 0.5% SDS for 30 min at 60 8C and twice in 0.5 SSC, 0.5% SDS for 30 min at 60 8C. Membranes were sealed in Saran Wrap, exposed to imaging plates (Fugi, BAS-MS) for 40–72 h and scanned at 50 mm/pixel using a Molecular Imager FX (BioRad). A total of six membranes were initially hybridized with radiolabeled RNA derived from non-regressed (n = 3) and regressed (n = 3) CL tissues. Following washing and exposure procedures the six membranes were stripped using the Strip-EZ system (Ambion Europe Ltd.). Stripping efficiency was evaluated by exposure to imaging plates and phosphor-imager analysis. Membranes were stored at 4 8C prior to rehybridization and reprobed with a further six radiolabeled RNA samples derived from regressed (n = 3) and non-regressed CL tissues (n = 3) resulting in a total of 12 individual hybridizations. 2.5. Image analysis and statistical analysis of array data The TIFF images of the scanned arrays were processed and spot and local background intensities calculated using ImageneTM 5.0 (BioDiscovery, Inc., Marina del Rey, CA). The log2 intensity values of the local background subtracted spots from each membrane were normalized using global geometric mean normalization option of MaxdView (v1.0.3) [24] to a mean intensity of zero and a trimmed standard deviation of one. Gene expression was examined using two-class analysis within the significance analysis of microarray data

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

1967

(SAM v1.21) program [25]. A false discovery rate (FDR) of <5% was used for the twoclass analysis. 2.6. Northern blot confirmation The number of CL replicates for the Northern analysis were six per treatment group. Total RNA (20 mg) isolated from individual non-regressed (n = 6) and regressed (n = 6) CL tissue was size fractionated through a denaturing 2.2 M formaldehyde 1% agarose gel in 1  MOPS buffer, transferred to nylon membranes (Schleicher & Schuell, Dassel, Germany) and baked at 80 8C for 2 h. Membranes were prehybridized for 4 h at 42 8C in 15 ml of Ultrahyb solution (Ambion Europe Ltd.) followed by overnight hybridization at 42 8C with a32P dATP-radiolabeled DNA fragments encoding glutathione-S-transferase aI (1 kb), scavenger receptor BI (1.8 kb), collagen a1(I) (700 bp), matrix gla protein (750 bp), clusterin (2 kb) and secreted protein acidic and rich in cysteine (1.5 kb). DNA probes were generated by PCR amplification from appropriate plasmid clones using T3 and T7 primers. The identity of the PCR products were confirmed by DNA sequencing and 32P-radiolabeled using StripEZ DNA kit (Ambion Europe Ltd.) according to manufacturers instructions. Membranes were successively stripped using the Strip-EZ DNA protocol (Ambion Europe Ltd.). A control hybridization using a 32P-radiolabeled DNA fragment of the 18S rRNA (493 bp) was used to normalize for RNA loading. Membranes were washed twice for 5 min in 2 SSC, 0.1% SDS at 42 8C and twice for 15 min in 0.1 SCC, 0.1% SDS at 42 8C. Membranes were exposed to imaging plates (Fugi, BAS-MS) and scanned using a Molecular Imager FX (BioRad). The mean and the standard error were generated by analysis of variance of the densitometric data associated with the six samples derived from the regressed and the six samples derived from the non-regressed groups using PROC GLM SAS [26]. 3. Results 3.1. Characterization of regressed and non-regressed CL tissues Corpora lutea deemed to be functional and non-regressed (n = 6) generated a mean systemic progesterone concentration of 11.72  1.17 ng/ml (mean  S.E.M.) within a range of 7.86–16.69 ng/ml and did not display DNA laddering (Fig. 1). In contrast CLs deemed to be regressed (n = 6) generated a mean systemic progesterone concentration of 0.84  0.27 ng/ ml (mean  S.E.M.) within a range of 0.24–1.76 ng/ml and exhibited the characteristic pattern of DNA laddering indicative of programmed cell death and structural regression (Fig. 1). Although one of the regressed CLs generated 1.76 ng/ml of systemic progesterone this was over four times less than the lowest systemic progesterone (7.86 ng/ml) produced in the non-regressed group. In addition, this CL also displayed DNA laddering indicating structural regression (Fig. 1, lane 10) and was therefore included in the regressed CL category. 3.2. Gene selection The ovarian cDNA array contained 434 genes derived from 351 non-redundant CL ESTs [22] and an additional 83 unique ESTs obtained from an ovarian cortex cDNA library

1968

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

Fig. 1. SYBR Green I stained DNA isolated from non-regressed (lane 2–7) and regressed CL tissue (lane 8–13). The occurrence of DNA laddering characteristic of apoptotic cell death and structural regression is detectable only in the regressed CLs. Lane 1 and 14 contain 100 bp molecular weight markers.

[23]. Identification was assigned to sequences with an expectation value of <1e
7 following BLASTX and BLASTN sequence similarity searches against GenBank protein and nucleotide databases as previously described [22]. The arrayed genes encoded proteins involved in a wide range of biological functions including lipid metabolism, immune response, extracellular matrix remodeling, and apoptosis. Additional information including accession number, E-value and percent identity are provided on a web accessible database (http://bioinf.may.ie/Bovine/bovine_CL_ests.pdf). 3.3. Isolation of differentially expressed genes Two-class analysis at a false discovery rate of <5% revealed 15 differentially expressed genes of which 7 genes increased and 8 decreased expression in regressed relative to nonregressed CL tissue (Table 1). The fold change in gene expression ranged from 1.74 to 2.37 for the up-regulated genes and 1.71–5.81 for the down-regulated genes (Table 1). Among

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

1969

Table 1 Genes differentially regulated at luteolysis Biological function

Gene name

Accession no.

Fold change

q-value (FDR)a

Extracellular matrix

Decorin Matrix gla protein Collagen a1(I) Collagen a2(I) Pancreatic anionic trypsinogen

CV547971 BU917233 BU917189 CV547968 BU917352

2.37 1.91 1.85 1.74
3.95

1.22 1.22 3.96 3.96 1.22

Cell structure

b-Actin

BU917286

1.83

1.22

Oxygen metabolism

Xanthine oxidoreductase Glutathione-S-transferase aI

BU917063 BU917155

1.92
3.82

1.22 1.22

Apoptosis

Clusterin

BU917282

1.76

3.96

Steroid biosynthesis

3 b-hydroxysteroid dehydrogenase Steroidogenic acute regulatory protein Scavenger receptor BI Cytochrome p450scc

BU917207 BU917068 BU917275 BU917164


5.81
4.69
1.96
1.85

1.22 1.22 2.04 2.84

Metabolism Uncharacterised EST

46 kDa mannose-6-phosphate receptor BARC 5BOV cDNA

BU917348 BU917392


3.98
1.71

1.22 2.04

a

Differentially expressed genes identified by SAM at a false discovery rate (FDR) of less than 5%.

the up-regulated genes, four encoded proteins associated with the extracellular matrix, and three had known functions relating to cell structure, oxygen radical metabolism and apoptosis. In contrast the eight down-regulated genes encoded proteins mainly involved in steroid biosynthesis and are detailed in (Table 1). To confirm the validity of the cDNA array results, 5 of the 15 differentially expressed genes including glutathione-S-transferase aI (GSTAI), scavenger receptor BI (SR-BI), collagen a1(I), matrix gla protein (MGP), and clusterin were subjected to Northern blot analysis (Fig. 2). The Northern results indicated that the mRNA for GSTAI and SR-BI decreased in regressed relative to non-regressed CL tissue (P < 0.01) a result consistent with the array data (Table 2). Genes shown to be up-regulated by array analysis (collagen a1(I), MGP and clusterin) also increased expression by Northern blot analysis (P < 0.01). The mRNA encoding secreted protein acidic and rich in cysteine (SPARC) which did not change significantly by the array analysis also did not change significantly when analyzed by Northern blot (P > 0.05) (Fig. 2). In addition, the fold change in gene expression calculated for the Northern blots correlated closely with the array results R2 = 0.992 (Table 2). Collectively these findings demonstrate that the array experimentation in terms of screening and data analysis was accurate and reproducible.

4. Discussion Using a 434 character ovarian specific cDNA array a total of 15 differentially expressed genes were identified in non-regressed compared to spontaneously regressed bovine CL tissue. Spontaneous regressed and non-regressed in vivo CL tissue were used to provide

1970

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

Fig. 2. Northern blot analysis of collagen a1(I), MGP, GSTAI, SR-BI, clusterin and SPARC. The data are expressed as mean  S.E.M. of six CL tissues per group (**P < 0.01, ***P < 0.001).

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

1971

Table 2 Comparison of fold change in gene expression for cDNA array and Northern blot analysis Gene name

Fold changea

q-value (FDR)

Fold changeb

P-value

Matrix gla protein Collagen a1(I) Glutathione-S-transferase aI Clusterin Scavenger receptor BI Secreted protein acidic and rich in cysteine

1.91 1.85
3.82 1.76
1.96 N/A

1.22 3.96 1.22 3.96 2.04 N/A

1.58 2.30
5.42 2.13
3.03 1.18

<0.01 <0.01 <0.001 <0.01 <0.01 >0.05

Fold change in gene expression as determined by aarray analysis and bNorthern analysis. N/A: not detected as significantly different at a false discovery rate (FDR) of less than 5%.

normal physiological conditions and to avoid possible artifacts derived from PGF2a induced luteolysis or from in vitro cell culture systems. The identified genes changed expression by at least 1.7-fold with a false discovery rate of <5% and encoded genes with recognized and potential roles in CL regression. 4.1. Steroidogenesis The decreased systemic progesterone concentrations in heifers with regressed CLs was accompanied by corresponding decreases in mRNA for genes involved in steroidogenesis including, 3b hydroxysteroid dehydrogenase (3bHSD), P450 cholesterol side-chain cleavage enzyme (p450scc) steroidogenic acute regulatory protein (StAR) and SR-BI. The mRNA for 3bHSD and StAR were the most highly repressed and showed an almost 5-fold expression difference in non-regressed compared to regressed luteal tissue. Similarly, p450scc mRNA decreased 1.85-fold in regressed luteal tissue. These results correlate with other bovine studies linking the repression of these genes to cessation of progesterone biosynthesis during CL regression [11,27]. The results of the cDNA array and Northern blot analysis demonstrated a greater than 1.9-fold decrease in expression for SR-BI mRNA. This receptor is involved in the uptake of high density lipoprotein (HDL) which is the major source of cholesterol for progesterone biosynthesis [28]. The decreased expression of mRNA encoding SR-BI in regressed bovine CL tissue suggests that HDL uptake by SR-BI may play an important role in the control of progesterone biosynthesis. This suggestion is supported by the alternate regulation of SR-BI by the luteotrophic action of prolactin and the luteolytic action of PGF2a in rat ovarian tissue [21,29]. 4.2. Oxygen radical metabolism Reactive oxygen species are recognized cellular signals implicated in the induction of apoptosis and are known to be generated within the bovine CL as a by-product of steroid hormone biosynthesis [30] and in the rat CL as a consequence of PGF2a induction at the end of the estrous cycle [31]. There is growing evidence that the balance between oxygen free radical generating systems and scavenging systems play a key role in regulating bovine CL lifespan [32]. In the current study mRNA encoding glutathione-S-transferase aI

1972

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

(GSTAI) was approximately 5-fold down-regulated in regressed luteal tissue. One of the principal biological functions of GSTAI is to provide protection against oxidative stress and in particular the damaging effects of lipid peroxidation [33]. Down regulation of GSTAI mRNA is, therefore, likely to increase luteal sensitivity to oxidative damage, hasten the onset of apoptosis and contribute to the demise of the CL. In contrast, mRNA encoding xanthine oxidoreductase was almost 2-fold up-regulated in regressed luteal tissue. Xanthine oxidoreductase occurs in two intercovertible forms. The oxidized form, xanthine oxidase (XO) catalysis the reduction of molecular oxygen to the reactive oxygen species, superoxide anion and hydrogen peroxide and is, therefore, likely to be pro-luteolytic [34]. The xanthine deyhydrogenase (XD) form, on the other hand, may form an important antiluteolytic function through its generation of uric acid, a potent antioxidant and defense mechanism against oxygen damage [35]. The methodology used in the current study prevented us from distinguishing which form of xanthine oxidoreductase was up-regulated in regressed CL tissue. Further studies are therefore necessary to determine the levels of XO and XD in bovine CL tissue, how this varies in non-regressed and regressed CL tissue and how this impacts on CL lifespan. 4.3. Apoptosis The mRNA encoding clusterin was elevated more than 2-fold in regressed relative to non-regressed CL tissue and was accompanied by the disappearance of systemic progesterone and the formation of DNA laddering, an indicator of apoptosis and structural regression. In rat mammary tissue, levels of clusterin mRNA have been shown to increase in response to tissue regression, involution, damage and disease [36,37] and in primate and murine cells functionally attributed to alterations in lipid transport, membrane remodeling and apoptosis [38,39]. These observations suggest that clusterin may alter CL function at several levels. For example, it is possible that up-regulation of clusterin in regressed CL tissue reduces progesterone biosynthesis through alteration of cholesterol transport and distribution, contributing to a shortage of substrate for steroidogenesis as proposed by Forni et al. [40]. Although clusterin appears to be inherently a pro-survival protein, over-expression in kidney cells is known to induce mitochondria-dependent apoptosis [38]. Increased clusterin expression has also been demonstrated in regressed rat and swine CL tissue [40,41]. These observations, combined with the results found in the current study, implicate clusterin expression with yet another incidence of programmed cell death and point towards a role for this glycoprotein protein in luteolysis. 4.4. Extracellular matrix and cell structure Structural regression of the CL is associated with extensive degradation and remodeling of the extracellular matrix (ECM) which can influence cell division, differentiation, migration and apoptosis [42]. In the current study mRNA for collagen a1(I) and collagen a2(I), components of the triple helical structure of collagen type I, were greater than 1.7fold up-regulated in regressed luteal tissue. The mRNA encoding pancreatic anionic trypsinogen also known as trypsinogen 2 was approximately 4-fold down-regulated in

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

1973

regressed relative to non-regressed CL tissue. Trypsinogen 2 is known to degrade type I collagen and can also activate latent collagenases including matrix metalloproteinases MMPs-1, -8 and -13 [43]. Taken together these results indicate that collagen biosynthesis is active during the latter stages of CL regression and supports biochemical data in the bovine showing that collagen type (I) is retained during luteolysis [44]. Decorin, an extracellular proteoglycan, was identified as the most highly induced mRNA (2.37-fold) represented on the cDNA array. Decorin has been implicated as a key regulator of matrix assembly through its ability to promote collagen fibril assembly [45]. It has also been implicated in the negative control of cell proliferation by virtue of its ability to bind and sequester the bioactivity of transforming growth factor b [46]. Up-regulation of decorin is therefore likely to promote CL regression by enhancing collagen stability and suppressing cellular growth. The mRNA for MGP, a vitamin K dependent calcium binding protein, not previously identified in the CL was elevated 1.91-fold in regressed CL tissue. One of the principal roles of MGP is to promote the calcification, differentiation and migration of bovine and murine vascular cells [47,48]. While these functions are cell type dependent [47] and are also influenced by the presence or absence of particular growth factors [47,49] it is tempting to speculate that the role of MGP in CL regression is also mediated through alteration of the luteal vascular system. Our results also show that b-actin mRNA, a component of the cytoskeleton network was 1.83-fold up-regulated in regressed compared to non-regressed luteal tissue. Fluctuations in b-actin mRNA levels have also been observed in human cell lines induced to undergo apoptosis [50]. It is probable that the enhanced expression of b-actin mRNA, as demonstrated in the current study reflects the increased apoptotic body formation and reorganization of the cytoskeleton that accompanies apoptotic cell death in regressing CL tissues [51]. These findings also call into question the use of b-actin as a control gene to normalize for RNA loading in gene expression studies, especially studies involving apoptosis. 4.5. Uncharacterized genes Two mRNAs encoding 46 kDa mannose-6-phosphate receptor which is involved in the transport of lysosomal enzymes and an uncharacterized EST accession no. BU9171392 whose ortholog was recently isolated from bovine mammary tissue were repressed in regressed luteal tissue. These genes have as yet no obvious role in CL physiology and their functional significance in CL regression therefore remains to be explored. In summary, this paper demonstrates that small-scale tissue-specific cDNA arrays provide a valid alternative to conventional molecular approaches for studying differential gene expression. Using this approach we identified a total of 15 genes with known and potential roles in luteolysis. The nature of the identified genes suggest that CL regression is a complex multi-mechanistic process that involves alterations in extracellular matrix remodeling, oxygen radical metabolism, apoptosis and steroid biosynthesis. Increased application of array technology, combined with protein and functional assays should provide further insights into the genes and mechanisms regulating the lifespan of the CL and the underlying causes of luteal insufficiency.

1974

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

Acknowledgements The authors would especially like to acknowledge Dr. Richard Powell’s essential contribution to this paper, who died suddenly during the course of this study. The technical assistance of Messrs. P. Joyce, W. Connelly, J. Nally and Mrs. A. Glynn is also gratefully acknowledged. The authors would also like to thank Dr. Alex Evans and Dr. Michael Diskin for critical reading of this manuscript. O.M.C. is currently supported by the Teagasc Walsh Fellowship scheme.

References [1] Fields MJ, Fields PA. Morphological characteristics of the bovine corpus luteum during the estrous cycle and pregnancy. Theriogenology 1996;45:1295–325. [2] Niswender GD, Juengel JL, Silva PJ, Rollyson KE, McIntosh EW, Rollyson MK. Mechanisms controlling the function and lifespan of the corpus luteum. Physiol Rev 2000;80:1–29. [3] Ireland JJ, Mihm M, Austin E, Diskin MG, Roche JF. Historical perspective of turnover of dominant follicles during the bovine estrous cycle: key concepts, studies, advancements, and terms. J Dairy Sci 2000;83:1648– 58. [4] Juengel JL, Garverick HA, Johnson AL, Youngquist RS, Smith MF. Apoptosis during luteal regression in cattle. Endocrinology 1993;132:249–54. [5] Mann GE, Lamming GE. The influence of progesterone during early pregnancy in cattle. Reprod Domest Anim 1999;34:269–74. [6] Starbuck GR, Darwash AO, Mann GE, Lamming GE. The detection and treatment of post insemination progesterone insufficiency in dairy cows. BSAS Occas Publ 26 2001;2:447–50. [7] Royal MD, Darwash AO, Flint AP, Webb R, Woolliams JA, Lamming GE. Declining fertility in dairy cattle: changes in traditional and endocrine parameters of fertility. Anim Sci 2000;70:487–501. [8] McCracken JA, Custer EE, Lamsa JC. Luteolysis: a neuroendocrine-mediated event. Physiol Rev 1999; 79:263–323. [9] Anderson LE, Wu YL, Tsai SJ, Wiltbank MC. Prostaglandin F(2alpha) receptor in the corpus luteum: recent information on the gene, messenger ribonucleic acid, and protein. Biol Reprod 2001;64:1041–7. [10] Pate JL. Lives in the balance: responsiveness of the corpus luteum to uterine and embryonic signals. Reprod Suppl 2003;61:207–17. [11] Pescador N, Soumano K, Stocco DM, Price CA, Murphy BD. Steroidogenic acute regulatory protein in bovine corpora lutea. Biol Reprod 1996;55:485–91. [12] Neuvians TP, Berisha B, Schams D. Vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF) expression during induced luteolysis in the bovine corpus luteum. Mol Reprod Dev 2004;67: 389–95. [13] Nakamura T, Ishigami T, Makino N, Sakamoto K. The down-regulation of glutathione peroxidase causes bovine luteal cell apoptosis during structural luteolysis. J Biochem (Tokyo) 2001;129:937–42. [14] Choudhary E, Costine BA, Wilson ME, Inskeep EK, Flores JA. Prostaglandin F2alpha (PGF2alpha) independent and dependent regulation of the bovine luteal endothelin system. Domest Anim Endocrinol 2004;27:63–79. [15] Neuvians TP, Schams D, Berisha B, Pfaffl MW. Involvement of pro-inflammatory cytokines, mediators of inflammation, and basic fibroblast growth factor in prostaglandin F2alpha-induced luteolysis in bovine corpus luteum. Biol Reprod 2004;70:473–80. [16] Taniguchi H, Yokomizo Y, Okuda K. Fas-Fas ligand system mediates luteal cell death in bovine corpus luteum. Biol Reprod 2002;66:754–9. [17] Yao J, Ren X, Ireland JJ, Coussens PM, Smith TP, Smith GW. Generation of a bovine oocyte cDNA library and microarray: resources for identification of genes important for follicular development and early embryogenesis. Physiol Genomics 2004;19:84–92.

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

1975

[18] Evans AC, Ireland JL, Winn ME, Lonergan P, Smith GW, Coussens PM, et al. Identification of genes involved in apoptosis and dominant follicle development during follicular waves in cattle. Biol Reprod 2004;70:1475–84. [19] Ishiwata H, Katsuma S, Kizaki K, Patel OV, Nakano H, Takahashi T, et al. Characterization of gene expression profiles in early bovine pregnancy using a custom cDNA microarray. Mol Reprod Dev 2003; 65:9–18. [20] Madsen SA, Chang LC, Hickey MC, Rosa GJ, Coussens PM, Burton JL. Microarray analysis of gene expression in blood neutrophils of parturient cows. Physiol Genomics 2004;16:212–21. [21] Stocco C, Callegari E, Gibori G. Opposite effect of prolactin and prostaglandin F(2 alpha) on the expression of luteal genes as revealed by rat cDNA expression array. Endocrinology 2001;142:4158–61. [22] Casey OM, Fitzpatrick R, McInerney JO, Morris DG, Powell R, Sreenan JM. Analysis of gene expression in the bovine corpus luteum through generation and characterisation of 960 ESTs. Biochim Biophys Acta 2004;1679:10–7. [23] Fitzpatrick R, Casey OM, Morris D, Smith T, Powell R, Sreenan JM. Postmortem stability of RNA isolated from bovine reproductive tissues. Biochim Biophys Acta 2002;1574:10–4. [24] Manchester. Microarray Bioinformatics Group. http://bioinfmanacuk/microarray/maxd/indexhtml, 2004. [25] Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001;98:5116–21. [26] SAS/STAT Users Guide, 8.2 ed. Statistical Analysis Systems Institute, Cary, NC 2001. [27] Tian XC, Berndtson AK, Fortune JE. Changes in levels of messenger ribonucleic acid for cytochrome P450 side-chain cleavage and 3 beta-hydroxysteroid dehydrogenase during prostaglandin F2 alpha-induced luteolysis in cattle. Biol Reprod 1994;50:349–56. [28] Grummer RR, Carroll DJ. A review of lipoprotein cholesterol metabolism: importance to ovarian function. J Anim Sci 1988;66:3160–73. [29] McLean MP, Sandhoff TW. Expression and hormonal regulation of the high-density lipoprotein (HDL) receptor scavenger receptor class B type I messenger ribonucleic acid in the rat ovary. Endocrine 1998;9:243–52. [30] Hanukoglu I, Rapoport R, Weiner L, Sklan D. Electron leakage from the mitochondrial NADPHadrenodoxin reductase-adrenodoxin-P450scc (cholesterol side chain cleavage) system. Arch Biochem Biophys 1993;305:489–98. [31] Tanaka M, Miyazaki T, Tanigaki S, Kasai K, Minegishi K, Miyakoshi K, et al. Participation of reactive oxygen species in PGF2alpha-induced apoptosis in rat luteal cells. J Reprod Fertil 2000;120:239–45. [32] Rapoport R, Sklan D, Wolfenson D, Shaham-Albalancy A, Hanukoglu I. Antioxidant capacity is correlated with steroidogenic status of the corpus luteum during the bovine estrous cycle. Biochim Biophys Acta 1998;1380:133–40. [33] Yang Y, Sharma R, Cheng JZ, Saini MK, Ansari NH, Andley UP, et al. Protection of HLE B-3 cells against hydrogen peroxide- and naphthalene-induced lipid peroxidation and apoptosis by transfection with hGSTA1 and hGSTA2. Invest Ophthalmol Vis Sci 2002;43:434–45. [34] Kuppusamy P, Zweier JL. Characterization of free radical generation by xanthine oxidase. Evidence for hydroxyl radical generation. J Biol Chem 1989;264:9880–4. [35] Becker BF. Towards the physiological function of uric acid. Free Radic Biol Med 1993;14:615–31. [36] French LE, Soriano JV, Montesano R, Pepper MS. Modulation of clusterin gene expression in the rat mammary gland during pregnancy, lactation, and involution. Biol Reprod 1996;55:1213–20. [37] Jones SE, Jomary C. Clusterin Int J Biochem Cell Biol 2002;34:427–31. [38] Debure L, Vayssiere JL, Rincheval V, Loison F, Le Drean Y, Michel D. Intracellular clusterin causes juxtanuclear aggregate formation and mitochondrial alteration. J Cell Sci 2003;116:3109–21. [39] Gelissen IC, Hochgrebe T, Wilson MR, Easterbrook-Smith SB, Jessup W, Dean RT, et al. Apolipoprotein J (clusterin) induces cholesterol export from macrophage-foam cells: a potential anti-atherogenic function? Biochem J 1998;331(Part 1):231–7. [40] Forni M, Zannoni A, Tamanini C, Bacci ML. Opposite regulation of clusterin and LH receptor in the swine corpus luteum during luteolysis. Reprod Nutr Dev 2003;43:517–25. [41] Kaynard AH, Periman LM, Simard J, Melner MH. Ovarian 3 beta-hydroxysteroid dehydrogenase and sulfated glycoprotein-2 gene expression are differentially regulated by the induction of ovulation, pseudopregnancy, and luteolysis in the immature rat. Endocrinology 1992;130:2192–200.

1976

O.M. Casey et al. / Theriogenology 64 (2005) 1963–1976

[42] Ny T, Wahlberg P, Brandstrom IJ. Matrix remodeling in the ovary: regulation and functional role of the plasminogen activator and matrix metalloproteinase systems. Mol Cell Endocrinol 2002;187:29–38. [43] Moilanen M, Sorsa T, Stenman M, Nyberg P, Lindy O, Vesterinen J, et al. Tumor-associated trypsinogen-2 (trypsinogen-2) activates procollagenases (MMP-1, -8 -13) and stromelysin-1 (MMP-3) and degrades type I collagen. Biochemistry 2003;42:5414–20. [44] Luck MR, Zhao Y. Identification and measurement of collagen in the bovine corpus luteum and its relationship with ascorbic acid and tissue development. J Reprod Fertil 1993;99:647–52. [45] Reed CC, Iozzo RV. The role of decorin in collagen fibrillogenesis and skin homeostasis. Glycoconj J 2002;19:249–55. [46] Yamaguchi Y, Mann DM, Ruoslahti E. Negative regulation of transforming growth factor-beta by the proteoglycan decorin. Nature 1990;346:281–4. [47] Bostrom K, Zebboudj AF, Yao Y, Lin TS, Torres A. Matrix GLA protein stimulates VEGF expression through increased transforming growth factor-beta1 activity in endothelial cells. J Biol Chem 2004;279:52904–13. [48] Luo G, Ducy P, McKee MD, Pinero GJ, Loyer E, Behringer RR, et al. Spontaneous calcification of arteries and cartilage in mice lacking matrix GLA protein. Nature 1997;386:78–81. [49] Zebboudj AF, Shin V, Bostrom K. Matrix GLA protein and BMP-2 regulate osteoinduction in calcifying vascular cells. J Cell Biochem 2003;90:756–65. [50] Naora H, Naora H. Differential expression patterns of beta-actin mRNA in cells undergoing apoptosis. Biochem Biophys Res Commun 1995;211:491–6. [51] Suarez-Huerta N, Mosselmans R, Dumont JE, Robaye B. Actin depolymerization and polymerization are required during apoptosis in endothelial cells. J Cell Physiol 2000;184:239–45.