• Gubernaculum;
  • undescended testis;
  • gene expression profiling;
  • fetus


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References

ABSTRACT: Development of the fetal gubernaculum is a prerequisite for testicular descent and dependent on insulin-like 3 and androgen, but knowledge of downstream effectors is limited. We analyzed transcript profiles in gubernaculum and testis to address changes occurring during normal and abnormal testicular descent in Long Evans wild-type (wt) and cryptorchid (orl) fetuses. Total RNA from male wt and orl gubernacula (gestational days [GD]18–20), wt female gubernacula (GD18), and testis (GD17 and 19) was hybridized to Affymetrix GeneChips. Statistical analysis of temporal, gender, and strain-specific differences in gene expression was performed with the use of linear models analysis with empirical Bayes statistics and analysis of variance (gubernaculum) and linear analysis (testis). Overrepresented common gene ontology functional categories and pathways were identified in groups of differentially expressed genes with the Database for Annotation, Visualization, and Integrated Discovery. Transcript profiles were dynamic in wt males between GD18–19 and GD20, comparatively static in orl GD18–20 gubernaculum, and similar in wt and orl testis. Functional analysis of differentially expressed genes in wt and orl gubernaculum identified categories related to metabolism, cellular biogenesis, small GTPase-mediated signal transduction, cytoskeleton, muscle development, and insulin signaling. Genes involved in androgen receptor signaling, regulated by androgens, or both were overrepresented in differentially expressed gubernaculum and testis gene groups. Quantitative reverse transcription polymerase chain reaction (RT-PCR) confirmed differential expression of genes related to muscle development, including Myog, Tnnt2, Fst, Igf1, Igfbp5, Id2, and Msx1. These data suggest that the orl mutation results in a primary gubernacular defect that affects muscle development and cytoskeletal function and might alter androgen-regulated pathways.

Cryptorchidism, or undescended testis, is one of the most common congenital anomalies in humans, occurring in 2%–3% of all boys (Barthold and Gonzalez, 2003). The cause of nonsyndromic cryptorchidism, in most cases, is unknown; however, the prevalence of sporadic and familial nonsyndromic cryptorchidism supports multifactorial susceptibility on the basis of contributions from specific genetic loci interacting with environmental factors, which could include endocrine-disrupting chemicals having antiandrogenic, estrogenic, or both effects (Mahood et al, 2006). Genetic contributions to cryptorchidism are not well understood, but the anomaly might be inherited in as many as 25% of cases, with autosomal dominant inheritance being the most common pattern and the mean heritability in first-degree male relatives calculated to be .67 (Czeizel et al, 1981; Elert et al, 2003).

Completion of testicular descent in mammals is dependent on the gubernaculum, which is an appendage of the anterior abdominal wall comprising a core of mesenchymal cells with associated extracellular matrix and localized striated muscle (Radhakrishnan et al, 1979; Costa et al, 2002). In the rat fetus, the gubernaculum is visible at gestational day 14 (GD14) in both sexes (Radhakrishnan et al, 1979). The female gubernaculum contains both mesenchymal and poorly organized muscle cells, and further growth fails to occur after GD16. In males, the gubernaculum enlarges after GD16, increases dramatically in size between GD18 and 20, then becomes exteriorized by everting into an extra-abdominal location around the time of birth (GD22). The mesenchymal portion of the gubernaculum disappears, leaving an outer layer of muscle, which persists as a sac of cremaster muscle and surrounds the scrotal testis. Eversion of the gubernaculum-cremaster complex occurs rapidly, but the mechanisms controlling its development and motility are poorly understood.

In vitro studies of gubernacular development and phenotypic analysis of cryptorchid genetic mouse models suggest that the testis is required for proper development of the ipsilateral gubernaculum and implicate secretion of the Leydig cell hormones insulin-like factor 3 (Insl3) and, to a lesser degree, testosterone (Emmen et al, 2000) in gubernacular development. Targeted deletion of either Insl3 or Rxfp2 is associated with high intra-abdominal testes in homozygous male mice and delayed testicular descent in heterozygotes (Zimmermann et al, 1999; Overbeek et al, 2001). Development of the fetal gubernaculum is feminized in homozygous Insl3/Rxfp2 mutants (Tomiyama et al, 2003). By contrast, mice and rats with spontaneous androgen receptor defects or that have been exposed to the antiandrogen flutamide (Spencer et al, 1991) show a milder phenotype. Although a model of testicular descent separates INSL3- and androgen-dependent phases into distinct events (Hutson and Hasthorpe, 2005), both hormones stimulate proliferation of fetal gubernacular cells. Moreover, generalized expression of both the INSL3 receptor RXFP2 (relaxin/insulin-like family receptor peptide 2, also known as LGR8 or GREAT) and the androgen receptor is present in the fetal gubernaculum (Emmen et al, 2000; Scott et al, 2005). Canonical Insl3/Rxfp2 signaling involves the cAMP/protein kinase A (PKA) pathway via activation of the cAMP response element (CRE; Halls et al, 2005), but information regarding downstream effectors is limited.

The Long Evans orl rat strain is an inbred colony at high risk for spontaneous cryptorchidism (Mouhadjer et al, 1989). Approximately two-thirds of offspring are affected, and up to 75% of cases occur unilaterally, with the left side more frequently affected (unpublished observations); overall, approximately 35%–40% of testes fail to descend (Barthold et al, 2006). The orl gubernaculum is reduced in size between GD18 and 20, but the testis descends normally during this time (Barthold et al, 2006). By the first day of life, however, normal eversion fails to occur in about half of orl gubernacula, and subsequent aberrant lateral migration occurs with final localization of the ipsilateral testis in the superficial inguinal pouch, anterior to the rectus muscle. This is a unique animal model of cryptorchidism in that the phenotype is similar to that seen most commonly in the human population.

Because of the complexity of genetic pathways and their interactions that are known in the various models of cryptorchidism, a comprehensive screen of transcript profiles is useful to address the changes associated with the gubernaculum and testis in orl rats. In this study, we use microarray analysis to study gene expression in the developing fetal gubernaculum and testis. Our data indicate that expression of genes involved in energy pathways and in the functionally related categories of muscle development, cytoskeleton organization and biogenesis, and small GTPase-mediated signal transduction is altered in normal and orl fetuses during prenatal growth of the gubernaculum. By contrast, we found fewer strain-specific differences in fetal testicular gene expression, suggesting that genetic variants with gubernaculum-specific effects predispose orl rats to cryptorchidism.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References


Breeding colonies of orl and wt rats were maintained in a reverse light cycle room following protocols approved by the institutional Animal Care and Use Committee. Female estrus cycles were assessed with vaginal smears, and animals were mated in the afternoon to generate timed pregnancies, which were identified by visualization of sperm in vaginal smears the following day, defined as GD1. Pregnant females were euthanatized via CO2 inhalation during the late morning of GD17–20. The caudal half of each fetus was immediately collected in RNAlater (Applied Biosystems, Foster City, California) and stored at 4°C for at least 24 hours to facilitate microdissection. Fetal testes and gubernacula were separated by microdissection, and samples were collected as noted in Table 1. For GD18–19, left-sided samples were used because of a higher incidence of left cryptorchidism observed in the orl strain (unpublished observations) and the possibility of intrinsic left-right asymmetry in males. At GD20, both left and right gubernacula were removed from 3 fetuses. In females, left and right gubernacula from the same fetus were pooled.

Table 1. . Samples used for microarray analysis
 Gestational Age, dSideLitter No.
  1. Abbreviations: wt, Long Evans wild-type strain; orl, Long Evans cryptorchid strain.

  2. a Female (all others male).

        W18a—da18Right + left1
        W17a—e17 8
        W19a—e19 3
        O17a—e17 9
        O19a—e19 6

RNA Extraction

Total RNA was purified from single gubernacula or testes with the RNeasy Mini Kit (Qiagen, Valencia, California) and the RNase-free DNase Set (Qiagen). RNA was quantified on the basis of A260 with the use of an ND-1000 Ultraviolet-visible spectrophotometer (NanoDrop Technologies, Wilmington, Delaware). Overall integrity of the total RNA was verified with a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, California) before processing for microarrays to assure consistency across samples.

Microarray Sample Processing

RNA samples (Table 1) were assessed with Affymetrix Rat Expression Array 230A (Affymetrix, Santa Clara, California). This microarray contains 15 866 probe sets representing approximately 10 500 genes and 2700 ESTs. Some genes are represented by more than 1 probe set. Whereas NCBI Entrez Gene lists approximately 38 000 genes for Rattus norvegicus, the 230A GeneChips interogates nearly one-third of known rat genes. For testes, 1 μg of total RNA from single organs was labeled with the One-Cycle cDNA Synthesis Kit (Affymetrix). This involved cDNA synthesis followed by in vitro transcription with T7-RNA polymerase and biotinylated nucleotide. Because of the smaller yield of RNA from gubernacula, 30 ng of total RNA from single organs was amplified and labeled with the GeneChip Two-Cycle cDNA Synthesis Kit (Affymetrix). After cDNA synthesis and in vitro transcription with T7-RNA polymerase, the resulting cRNA was used as a template for a second round of cDNA synthesis, which was followed by in vitro transcription in the presence of biotinylated nucleotide. Biotinylated cRNA was hybridized to 230A GeneChips. Arrays were washed, stained with strepavidin phycoerythrin conjugate, and scanned at DuPont Haskell Laboratory for Health and Environmental Sciences in a Hybridization Oven 640 (Affymetrix), GeneChip Fluidics Station (Affymetrix), and GeneArray Scanner (Affymetrix) with Affymetrix protocols and reagents. Standard Affymetrix quality control measures were consistent across all hybridizations of the same tissue type.

Analysis of Microarray Data

As a measure of the quality of hybridization, the raw and normalized probe intensity distributions for each GeneChip were determined with histogram plots within the AffylmGUI interface for the limma package of Bioconductor (Wettenhall et al, 2006). Representations of 5′ and 3′ regions of transcripts in the labeled cRNA were examined with the Affy package of Bioconductor to verify consistency within tissue types. Expression values were calculated with the MAS 5 algorithm from raw probe intensities using GCOS (Affymetrix). Expression values were also calculated with the GC robust multiarray average (GC-RMA) algorithm (Wu et al, 2004) from raw probe intensities within AffylmGUI. All further analyses were performed with the GC-RMA expression values unless otherwise noted. Global gene expression patterns and overall variability between samples were examined by principal component analysis (PCA), which was performed in MeV version 3.1 (Saeed et al, 2003). Two methods were used to identify differentially expressed genes: 1) the LIMMA linear models approach with the empirical Bayes statistic (B ≥ 3) and the multiple testing adjustment method of Holm, used within AffylmGUI (referred to as linear analysis), and 2) calculation of MAS5 expression ratios using GD18 wt as a reference denominator followed by the scaling of log2-transformed values to a median of 0.0 and standard deviation of 0.50 with statistical analysis in GeneSpring version 7.2 using analysis of variance with a Benjamini and Hochberg false discovery rate of .001 (referred to as reference denominator method). Differentially expressed genes were filtered for a GC-RMA average expression value of greater than or equal to 50 in at least 1 sample group in the analysis and then separated into groups according to the log2 ratio of average expression values for groups of samples. Some groups of genes were further separated with K-means clustering in MeV. Plots of expression profiles were created with the statistical package R ( Gene groups with mean expression levels as well as the entire data set are available via Accession number GSE7755 at the Gene Expression Omnibus (http:www.ncbi.nlm.nih.govgeo). Groups of differentially expressed probe sets were examined for statistical overrepresentation of Gene Ontology (GO) Biological Function categories and biological pathways as defined in the Kyoto Encyclopedia of Genes and Genomes (KEGG; http:www.genome.jpkegg) with the Database for Annotation, Visualization, and Integrated Discovery (DAVID) 2007 (Dennis et al, 2003;

Real-Time Reverse Transcription Polymerase Chain Reaction

Real-time reverse transcription polymerase chain reaction (RT-PCR) was used to validate trends in selected array-derived data from gubernaculum. cDNA was synthesized from 150 ng of total RNA (n > 6 samples per group) with the High-Capacity cDNA Archive Kit (Applied Biosystems). Amplifications were performed in triplicate using TaqMan Gene Expression Assays (see Table 2 for details) and TaqMan Universal PCR Master Mix in an ABI Prism 7900HT. Levels of target mRNA expression were determined by the 2−ΔΔCT method (Livak and Schmittgen, 2001) with tripeptidyl peptidase 2 (Tpp2) as control and total rat embryonic RNA (Agilent Technologies) as calibrator. Nonparametric statistical analyses of differences between strains were performed in SPSS (version 14.0; SPSS Inc) as indicated.

Table 2. . Real-time reverse transcription polymerase chain reaction assays
ABI AssayGene SymbolGene Name
  1. Abbreviation: MMTV, mouse mammary tumor virus.

Rn00432087_m1Bmp4Bone morphogenetic protein 4
Rn00518185_m1Dusp6Dual-specificity phosphatase 6
Rn01495280_m1Id2Inhibitor of DNA binding 2
Rn00710306_m1Igf1Insulin-like growth factor 1
Rn00563116_m1Igfbp5Insulin-like growth factor—binding protein 5
Rn00667535_m1Msx1Homeobox, msh-like 1
Rn01399583_m1Nfkb1Nuclear factor of kappa light chain gene enhancer in B-cells 1, p105
Rn01438455_m1Olfm1Olfactomedin 1
Rn01483694_m1Tnnt2Troponin T2, cardiac
Rn01437410_m1Tpp2Tripeptidyl peptidase II
Rn00584577_m1Wnt4Wingless-related MMTV integration site 4


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References

Global Gene Expression

PCA analysis using all probe sets on the RAE 230A array was performed to examine global trends in gene expression for gubernaculum and testis samples (Figure 1). Gubernaculum samples (Table 1) clustered into 4 main groups: 1) GD18 females, 2) GD18–19 wt males, 3) GD20 wt males, and 4) 13 of the 15 orl samples. The 2 remaining orl samples, O18a and O19d, clustered with GD20 and GD18–19 wt males, respectively. These samples were considered outliers and were excluded from further analysis. Linear analysis of left vs right GD20 samples failed to reveal significant differences; however, further analyses were limited to left-sided samples. Testis samples clustered into GD17 and 19 groups with intermixing of the 2 strains at each time point. After initial global data analysis, our overall strategy was to 1) identify expression profiles of developmentally regulated genes in a defined window of gestation in wt males, 2) identify genes differentially expressed between strains, and 3) analyze these groups of genes functionally.


Figure 1. . Principal component analysis (PCA) of all gubernacular (A) and testicular (B) samples for all probe sets on the Affymetrix Rat 230A GeneChip. Black and gray dots represent wild-type (wt) and cryptorchid (orl) samples, respectively. Labels refer to samples within adjacent dotted circles except as noted. (A) For gubernaculum, all gestational day (GD)18 and 19 wt (W) samples and all but 2 orl (O) samples were noted to cluster together. The 2 outlier orl samples cluster between GD20 wt and GD18 female (F) samples (gray dots), which are closely aligned. (B) For testis, GD17 and 19 samples cluster together with no clear separation between strains.

Download figure to PowerPoint

Expression Profiles in Normal and Cryptorchid Strains


We studied changes in gene expression across normal development of the gubernaculum by comparison of the 3 male wt groups (12 samples). The gene expression profile in GD18–19 samples was remarkably similar, with only 11 differentially expressed genes identified. By contrast, comparison of GD18 and 20 wt samples returned 1023 probe sets that were differentially regulated, suggesting a major switch in gene expression between GD18–19 and GD20. With the use of K-means clustering, we further classified these genes into 2 lists with declining expression (n = 371) or increasing expression (n = 652) after GD18 (Figure 2). The expression profiles show marked sexual dimorphism of these genes at GD18, suggesting that they participate in male-specific gubernacular development. Surprisingly, by GD20, the expression profile in normal males approximates that of GD18 females, as we observed in PCA analysis (Figure 1).


Figure 2. . Expression profiles of genes differentially expressed between GD18 and 20 in wt gubernaculum (n = 1023). Genes are classified on the basis of K-means clustering: (A) Increasing expression between GD18 and 20 (n = 652) and (B) decreasing expression between GD18 and 20 (n = 371). Error bars show mean Z-scores ± SD for probe sets in each subgroup. Vertical lines separate the samples by gender (F indicates wt female; W, wt male) and gestational age in days (18-20).

Download figure to PowerPoint

Because of minimal differences between GD18 and 20, we analyzed strain-specific differences at GD18 and 20 only. Linear analysis identified 2401 probe sets that were differentially expressed between wt and orl males at these time points. The reference denominator method, which takes into account all time-, gender-, and strain-specific differences in expression relative to GD18, returned 3707 probe sets. After filtering for low expression, these lists were combined to generate a final list of 3589 probe sets associated with wt versus orl differences.

Log2 ratio of expression values was used to divide the final list into groups of genes with either higher (orl-high, n = 1681) or lower (orl-low, n = 1908) expression in orl relative to wt samples. With the use of K-means clustering, we identified 3 major expression profiles in each of these groups (Figure 3). The observed pattern suggests little change in expression of differentially regulated genes in orl fetuses between GD18 and 20. Interestingly, a tendency toward reciprocal patterns of female, wt, and orl expression is seen in the corresponding clusters from each group. Two clusters (cluster 1 of each group) show the greatest differences between normal males and females at GD18. It is noteworthy that the corresponding orl expression profiles in these clusters are feminized.


Figure 3. . Expression profiles of differentially expressed gene groups in orl and wt gubernaculum are shown: (A) Genes with higher expression in orl (orl-high) and (B) genes with lower expression in orl (orl-low). Profiles were generated by K-means clustering of all samples in the 2 gene groups, and the 3 major resultant clusters for each gene group are shown. Error bars show mean Z-scores ± SD for probe sets in orl-high clusters 1 (n = 577), 2 (n = 502), and 3 (n = 448) and for orl-low clusters 1 (n = 604), 2 (n = 851), and 3 (n = 451). Vertical lines separate the samples by gender and strain (F indicates wt female; O, orl male; W, wt male) and gestational age in days (18-20).

Download figure to PowerPoint


Comparable linear analysis of GD17 and 19 wt samples identified fewer (n = 818) differentially regulated genes at these 2 time points in testis compared with gubernaculum. More significantly, few genes were differentially expressed in the testicular samples of wt compared with orl fetuses. Linear analysis of strain differences at GD17 or 19 yielded only 349 differentially expressed probe sets when combining the 2 lists. Of these, expression was higher in wt males in 248 sets and lower in 101.

Functional Annotation of Differentially Expressed Genes

We performed functional analysis of groups of genes using DAVID and analyzed common GO Biological Process annotations. Two groups of probe sets were analyzed separately: genes differentially expressed between wt and orl in gubernaculum (n = 3589; converted to 3428 unique DAVID IDs) and the group differentially expressed between wt and orl in testis (n = 349, converted to 347 DAVID IDs). The analyses from each group were compared for recurring functional themes.

Selected nonredundant categories are shown in Table 3, and expression data for selected genes in these groups are shown in Table 4. Multiple categories related to general physiologic processes such as metabolism and biosynthesis were identified. When analyzing all genes differentially expressed between wt and orl, small GTPase signal transduction was the most significantly represented signaling pathway in GO. We also identified categories related to small GTPase signaling, including muscle development and cytoskeletal organization and biogenesis. These data are consistent with the known morphological changes occurring in the gubernaculum during this time frame, including significant growth and maturation of muscle (Radhakrishnan et al, 1979; Cain et al, 1995). Comparatively few GO annotations were identified in the differentially expressed testis gene group. However, these include muscle development, which might indicate selective effects of the mutation in myoid cells.

Table 3. . Selected functional gene ontology (GO) biological process annotations represented by differentially expressed genesa
GO Biological ProcessNo. of GenesP ValueNo. of GenesP Value
  1. a Modified Fisher exact P values are shown for selected overrepresented GO biological process terms identified by analysis of differentially expressed genes in fetal gubernaculum (n = 3428 DAVID IDs) and testis (n = 347 DAVID IDs) with the use of Database for Annotation, Visualization, and Integrated Discovery (DAVID; http:niaid.abcc.ncifcrf.govhome.jsp).

Metabolism12053.8 × 10−54
Cell organization and biogenesis3921.2 × 10−27
Biosynthesis2984.7 × 10−22
Cytoskeleton organization and biogenesis1043.4 × 10−10
Cell cycle1412.3 × 10−9
Localization5024.4 × 10−8
Transport4331.8 × 10−7
Muscle development442.0 × 10−6104.9 × 10−4
Generation of precursor metabolites and energy1265.7 × 10−6
Actin filament—based process481.6 × 10−5
Small GTPase-mediated signal transduction632.1 × 10−5
Cell division342.6 × 10−5
Muscle contraction416.7 × 10−5
Apoptosis1211.3 × 10−4
Phosphorylation1221.6 × 10−4
Transcription from RNA polymerase II promoter1272.8 × 10−4
Growth534.0 × 10−4112.7 × 10−3
Cellular morphogenesis915.9 × 10−4142.1 × 10−2
Development3796.3 × 10−3603.3 × 10−5
Table 4. . Selected genes differentially expressed in wt and orl fetal gubernaculuma
Probe Set IDGene TitlebGene SymbolLog2(orl/wt)
  1. Abbreviations: GAP, GTPase-activating protein; GDP, guanosine diphosphate; MMTV, mouse mammary tumor virus; NMDA, N-methyl-d-asparate.

  2. a Greatest difference in Affymetrix 230A probe set mean expression levels between orl and wt samples at gestational day 18 or 20 expressed as log2(orl/wt).

  3. b Bold gene titles indicate that real-time reverse transcription polymerase chain reaction also was performed.

  4. c Probe set with E annotation.

  5. d Gene associated with human cryptorchidism according to Online Mendelian Inheritance in Man (OMIM; http:www.ncbi.nlm.nih.govomim).

  6. e Gene associated with cryptorchidism in mice.

Muscle development   
    1368725_atJagged 1Jag13.96
    1368302_atHomeobox, msh-like 1Msx12.44
    1387232_atBone morphogenetic protein 4Bmp41.9
    1374904_atSine oculis homeobox homolog 1 (drosophila)Six11.88
    1367652_atInsulin-like growth factor—binding protein 3Igfpb31.94
    1388185_atRetinoblastoma 1Rb1−0.81
    1386993_atMyosin, heavy polypeptide 7, cardiac muscle, betaMyh7−1.01
    1387181_atMyogenic factor 6Myf6−1.08
    1388335_atTransgelin 2Tagln2−1.14
    1398248_s_atMyosin, heavy polypeptide 6, cardiac muscle, alphaMyh6−1.2
    1369928_atActin, alpha 1, skeletal muscledActa1−1.7
    1372569_atFour and a half LIM domains 3 (predicted)Fhl3_predicted−1.99
    1369375_a_atCalpain 3Capn3−1.29
    1388298_atMyosin, light polypeptide 9, regulatory (predicted)Myl9_predicted−2.52
    1387348_atInsulin-like growth factor—binding protein 5Igfbp5−3.21
    1367628_atLectin, galactose-binding soluble 1Lgals1−4.62
Muscle contraction   
    1367617_atAldolase AAldoa−1.15
    1367592_atTroponin T2, cardiacTnnt2−1.79
    1370857_atSmooth muscle alpha-actinActa2−2.32
    1367572_atMyosin, light polypeptide 3Myl3−2.33
    1368838_atTropomyosin 4Tpm4−2.54
    1368724_a_atTropomyosin 1, alphaTpm1−2.57
    1371239_s_atTropomyosin 3, gammaTpm3−2.85
    1387787_atMyosin, light polypeptide 2Myl2−3.25
Cytoskeleton organization and biogenesis   
    1367654_atFat tumor suppressor homolog (Drosophila)Fath4.31
    1387080_atChondroitin sulfate proteoglycan 6Cspg62.2
    1372692_atTyrosine kinase, non-receptor, 2Tnk22.09
    1370875_atVillin 2Vil21.98
    1387227_atWiskott-Aldrich syndrome protein—interacting proteinWaspip1.51
    1383822_atBicaudal D homolog 2 (Drosophila)Bicd21.32
    1368893_atCAP, adenylate cyclase—associated protein, 2 (yeast)Cap2−1.62
    1371885_atCytoskeleton-associated protein 1 (predicted)Ckap1_predicted−1.98
    1367605_atProfilin 1Pfn1−2.1
    1399105_atBridging integrator 3Bin3−2.22
    1374523_at6-phosphogluconolactonase (predicted)Pgls_predicted−2.26
    1388460_atCapping protein (actin filament), gelsolin-likeCapg−2.26
    1370184_atCofilin 1, non-muscleCfl1−2.31
    1398900_atDynactin 3 (predicted)Dctn3_predicted−2.75
Small GTPase-mediated signal transduction   
    1373215_atActive BCR-related gene (predicted)Abr_predicted2.78
    1389292_atRAB18, member RAS oncogene familyRab182.38
    1367475_atCell division cycle 42Cdc422.12
    1374239_atFERM, RhoGEF and pleckstrin domain protein 2 (predicted)Farp2_predicted1.71
    1388892_atRAB2B, member RAS oncogene familyRab2b1.5
    1388730_atCDC42 effector protein (Rho GTPase-binding) 4 (predicted)Cdc42ep4_predicted1.28
    1368217_atRalA-binding protein 1Ralbp11.27
    1389710_atSon of sevenless homolog 1 (Drosophila)dSos10.96
    1371255_atHarvey rat sarcoma viral (v-Ha-ras) oncogene homologHras−1.04
    1368096_atRAB7, member RAS oncogene family-like 1Rab7l1−1.58
    1372521_atRho family GTPase 2Rnd2−1.84
    1372513_atRas-related C3 botulinum toxin substrate 1Rac1−1.88
    1398838_atRAB7, member RAS oncogene family RhoGAP involved in beta-catenin-N-cadherin and NMDA receptor signalingRab7−2
    1373881_atRho, GDP dissociation inhibitor (GDI) beta Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activationArhgdib−2.15
    1370168_atprotein, theta polypeptideYwhaq−2.16
    1368041_atSynaptojanin 2 binding proteinSynj2bp−2.23
    1370130_atRas homolog gene family, member ARhoA−3.05
    1388729_atHarvey rat sarcoma oncogene, subgroup R (predicted)Rras_predicted−3.08
    1375532_atInhibitor of DNA binding 2cId25.15
    1368641_atWingless-related MMTV integration site 4Wnt43.2
    1377064_atDual-specificity phosphatase 6Dusp62.75
    1369008_a_atOlfactomedin 1Olfm12.57
    1388856_atKit ligandKitl2.31
    1370221_atWNT1 inducible signaling pathway protein 1Wisp12.04
    1390119_atSecreted frizzled-related protein 2Sfrp21.96
    1376755_atRetinoic acid receptor, betaRarb1.93
    1370747_atFibroblast growth factor 9Fgf91.57
    1372447_atFibroblast growth factor receptor 1dFgfr11.55
    1370968_atNuclear factor of kappa light chain gene enhancer in b-cells 1, p105Nfkb11.53
    1373829_atFibroblast growth factor receptor 2dFgfr21.26
    1368395_atGlypican 3dGpc31.11
    1370224_atSignal transducer and activator of transcription 3Stat31.06
    1375043_atFBJ murine osteosarcoma viral oncogene homologFos0.9
    1367712_atTissue inhibitor of metalloproteinase 1Timp1−1.07
    1388154_atE2F transcription factor 5E2f5−1.29
    1389403_atBone morphogenetic protein 7Bmp7−1.81
    1386940_atTissue inhibitor of metalloproteinase 2Timp2−2.18
Insulin signaling pathway   
    1376779_atForkhead box O1AFoxo1a1
    1386950_atProtein phosphatase 1, catalytic subunit, beta isoformPpp1cb−0.72
    1398799_atEukaryotic translation initiation factor 4EEif4e−0.77
    1367573_atRibosomal protein S6Rps6−1.52
    1368116_a_atRibosomal protein S6 kinase, polypeptide 1Rps6kb1−2.33
    1386888_atEukaryotic translation initiation factor 4E binding protein 1Eif4ebp1−2.61
Focal adhesion   
    1369955_atProcollagen, type V, alpha 1Col5a12.34
    1370267_atGlycogen synthase kinase 3 betaGsk3b1.88
    1372905_atVinculin (predicted)Vcl_predicted1.68
    1370333_a_atInsulin-like growth factor 1Igf11.38
    1367760_atMitogen-activated protein kinase 1Map2k11.15
    1370427_atPlatelet-derived growth factor, alphaPdgfa1.12
    1370155_atProcollagen, type I, alpha 2Col1a21.03
    1389723_atPhosphoinositide-3-kinase, regulatory subunit 4, p150 (predicted)Pik3r4_predicted1.01
    1383075_atCyclin D1Ccnd10.88
    1388138_atThrombospondin 4Thbs40.73
    1387777_atIntegrin-linked kinaseIlk−0.93
    1386863_atProtein phosphatase 1, catalytic subunit, alpha isoformPpp1ca−1.5
    1368385_a_atGrowth factor receptor—bound protein 2Grb2−1.68
    1398836_s_atActin, betaActb−2.76
    1387346_atIntegrin beta 1 (fibronectin receptor beta)Itgb1−2.97
    1390638_atSimilar to Eph receptor A4 (predicted)eRgd1560587_predicted2.46
    1372964_atAT-rich interactive domain 5B (MRF1-like) (predicted)ceArid5b_pred2.26
    1390355_atRyanodine receptordRyr11.85
    1368509_atBardet-Biedl syndrome 2 homolog (human)dBbs2−1.18
    1389670_atSimilar to homeobox protein A10 (predicted)eRgd1566402_predicted−2.1

With the use of DAVID, we identified overrepresented KEGG pathways for the 2 groups of genes differentially expressed in wt and orl gubernaculum and testis. Selected pathways are shown in Table 5, with the most genes associated with focal adhesion in both gubernaculum and testis. We separately analyzed overrepresentation of 755 known androgen-regulated and androgen-signaling pathway genes (http:www.netpath.orgpathwayspath_idNetPath_2; Bolton et al, 2007) not included in DAVID with a Fisher's exact test patterned after EASE (Expression Analysis Systematic Explorer) methodology (Hosack et al, 2003). Of 622 present on the microarray, 199 (P = .039) and 30 (P = .004) androgen-associated genes are differentially expressed in gubernaculum and testis, respectively. Together, the functional category and pathway analyses suggest altered regulation of related processes and pathways linked to energy and metabolism, muscle and cytoskeleton organization, and altered expression of androgen-regulated genes.

Table 5. . Selected Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways represented by differentially expressed genesa
KEGG PathwayNo. of GenesP ValueNo. of GenesP Value
  1. a Modified Fisher exact P values are shown for selected overrepresented KEGG pathway terms identified by analysis of differentially expressed genes in fetal gubernaculum (n = 3428 DAVID IDs) and testis (n = 347 DAVID IDs) with the use of DAVID (http:niaid.abcc.ncifcrf.govhome.jsp).

Ribosome411.8 × 10−17
Oxidative phosphorylation484.1 × 10−9
Proteasome191.9 × 10−9
ATP synthesis197.5 × 10−6
Valine, leucine, and isoleucine degradation174.2 × 10−459.6 × 10−3
Glycolysis/gluconeogenesis194.0 × 10−3
Fatty acid metabolism164.2 × 10−3
Insulin signaling pathway351.7 × 10−2
Focal adhesion495.0 × 10−2126.2 × 10−3

Expression of Genes Linked to Cryptorchidism or Testicular Descent

We analyzed the expression patterns of candidate genes annotated on the 230A chip. Hoxa10, Epha4, and Arid5B, genes associated with cryptorchidism in mice with spontaneous or targeted mutations ( are present in the list of differentially expressed genes (Table 4). Of the previously reported tyrosine kinases expressed in the fetal GD16.5 mouse gubernaculum (Verma-Kurvari and Parada, 2004), several failed to show significant gubernacular expression during the interval studied (Rous sarcoma virus [c-Src], spleen tyrosine kinase [Syk], v-Erb erythroblastic leukemia viral oncogene homolog 4 [Erbb4], and Eph receptor B4 [Ephb4]) using the microarray methodology. Others were expressed during this time frame at comparable levels in females and both male strains (platelet-derived growth factor receptor alpha [Pdgfra], insulin-like growth factor 1 receptor [Igf1r], c-src tyrosine kinase [Csk], kinase insert domain protein receptor [Kdr, also known as Vegfr2 or Flk1], and v-abl Abelson murine leukemia viral oncogene homolog 1 [Abl1; data not shown]). Expression of protein tyrosine kinase 2 (Ptk2, encoding FAK or focal adhesion kinase) is sexually dimorphic at GD18 and increases significantly in wt males between GD18 and 20 but is not differentially expressed between strains. None of the functional annotation analyses of testicular gene expression identified patterns suggestive of altered hormone synthesis in orl fetal testis and representative Leydig cell—specific genes, including probe sets for isoforms of Cyp11a1, Cyp17a1, Hsd3b, and Hsd17b, which were highly, and not differentially, expressed in both strains (data not shown). Insl3 expression was higher in orl fetal GD17 and 19 testis, but the differences were not significant on the basis of our linear analysis.

Comparison of Differentially Expressed Gubernaculum and Testis Genes

Of the 349 testis genes differentially expressed between the 2 strains, 117 were also differentially expressed in wt compared with orl gubernaculum. Few of these genes were down- or up-regulated in both testis and gubernaculum of orl fetuses. These include insulin-like growth factor—binding protein 5, interferon-induced transmembrane proteins 1 and 3, osteoglycin, and calcium/calmodulin-dependent protein kinase II; delta (lower expression in orl); and Epha4 (higher expression in orl). Many genes encoding ECM proteins, including several procollagens, laminins, basigin, matrix Gla protein, chondroitin sulfate proteoglycan 2, and spondin 1, show reduced expression in orl testis.

We directly compared mean normalized expression of 62 ribosomal and mitochondrial ribosomal genes in gubernaculum and testis (Figure 4). In contrast to the highly differential expression in gubernaculum, expression levels of these genes are comparable in testis samples from the 2 strains. Absolute expression levels between testis and gubernaculum were not directly comparable because of the differences in the protocols used (ie, single compared with double amplification of RNA samples). However, these data together with global expression data suggest that altered expression profiles are much more prominent in orl gubernaculum than in testis. These data suggest that altered signaling in the orl strain has a more profound effect on gene expression in fetal gubernaculum than testis.


Figure 4. . Expression profiles of ribosomal genes in gubernaculum and testis. Mean Z-scores ± SD for 62 probe sets in (A) male gubernaculum and (B) testis are shown. Vertical lines separate the samples by strain (O indicates orl; W, wt) and gestational age in days (18-20).

Download figure to PowerPoint

Validation of Array-Derived Expression Profiles

To determine whether the expression profiles obtained from the microarrays were consistent with the relative amounts of mRNA present in parallel samples, real-time RT-PCR validation was carried out. Expression levels of selected genes in specific pathways and functional groups (Table 4, bold) were analyzed. We identified 2 candidate control genes, tripeptidyl peptidase 2 (Tpp2) and lumican, with mean GC-RMA expression levels showing minimal variation across all groups; real-time RT-PCR results showed differences in raw Ct values of less than 0.5 for both genes with more consistency seen in Tpp2 expression across samples (data not shown). We analyzed target genes related to Tgfβ/Wnt/Hedgehog (Bmp4, Id2, Msx1, Wnt4, Fst), MAPK (Dusp6, Nfkb1), and insulin-like growth factor signaling (Igf1, Igfbp5); neurogenesis (Olfm1) and myogenesis (Myog, Tnnt2, Des) in gubernacula, testes (6–12 samples/group from 2–3 litters), or both relative to Tpp2. Compared with the microarray data, we identified similar expression patterns for most wt and orl samples at the 2 time points, and differences in mRNA levels by RT-PCR were statistically different for many of these genes at GD20 but not GD18 (Figure 5). The most significant differences between strains were noted for gubernacular genes associated with neuromuscular development (Olfm1, Msx1, Myog, Tnnt2, and Id2).


Figure 5. . Comparison of expression of selected genes in gubernaculum from wt (W) and orl (O) rat strains at GD18, GD20, or both according to microarray analysis (Array) and real-time reverse transcription polymerase chain reaction (RT-PCR) (expression level relative to Tpp2, tripeptidyl peptidase 2), x̄ ± 95% confidence interval, * indicates P = .06; ** P < .05; *** P < .0.01 for differences between strains at respective time points. Wnt4 indicates wingless-related mouse mammary tumor virus (MMTV) integration site 4; Fst, follistatin; Bmp4, bone morphogenetic protein 4; Msx1, msh-like homeobox 1; Olfm1, olfactomedin 1; Dusp6, dual-specificity phosphatase 6; Nfkb1, nuclear factor of kappa light-chain gene enhancer in B-cells 1, p105; Igf1, insulin-like growth factor 1; Igfbp5, insulin-like growth factor—binding protein 5; Myog, myogenin; Tnnt2, troponin T2; Des, desmin; Id2, inhibitor of DNA binding 2.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References

We characterized transcript profiles of fetal gubernaculum and testis in an animal model of inherited cryptorchidism and a wild-type strain to identify genetic pathways that are activated during rapid growth of fetal gubernaculum. Global analysis of samples suggests delayed, feminized, or both patterns in the mutant orl gubernaculum compared with the wt strain. We observe 2 major trajectories in the normal rat gubernaculum between GD18 and 20 and sexually dimorphic expression of these genes at GD18, with little change in gene expression between GD18 and 19 in males. Functional analysis of the normal pattern of gene expression is consistent with growth, cellular proliferation, and muscle development that are known to occur in the rat gubernaculum during this time frame. The dynamic changes resolve at GD20 to a level of expression similar to that of the GD18 female, suggesting completion of male-specific development. Similarly, by GD20, expression of Rxfp2 (Insl3 receptor) mRNA is markedly diminished (Barthold et al, 2006) and antiandrogen exposure does not prevent testicular descent (Spencer et al, 1991), suggesting that the critical phase of Insl3 and androgen stimulation of the gubernaculum occurs prior to this time.

By contrast, GD18–20 gene expression in orl fetuses is substantially less varied, with significant overlap in the function of genes that are more highly expressed in orl males and in females, suggesting that loss of male-specific signaling is already present at GD18. The observation that strain-specific expression profiling of fetal testis shows relatively few differences also supports a model of cryptorchidism in the orl strain in which delayed or incomplete development of the gubernaculum is a major contributing factor. Our functional analysis suggests that many general pathways related to metabolism, energy and growth are altered in the orl gubernaculum, consistent with the decreased size of the fetal orl gubernaculum (Barthold et al, 2006). We also identified several specific, related pathways and biological processes represented by differentially expressed gubernacular genes, including small GTPase signal transduction, focal adhesion, actin- filament—based process, cytoskeleton organization and biogenesis, and muscle development. Although no database of androgen-regulated genes in the fetal gubernaculum exists, we observed that genes regulated by androgens, involved in androgen receptor signaling in other cell types, or both were overrepresented in our lists of differentially expressed genes from both testis and gubernaculum. These data suggest that androgen receptor signaling may be altered in the orl fetus, although additional studies are needed.

The respective roles of Insl3 and androgen in cell-specific development of the gubernaculum remains undefined. In vitro, Rxfp2 activation increases cAMP production via the stimulatory G-protein Gαs, activates the CRE reporter, and, in cooperation with testosterone, stimulates proliferation of fetal gubernacular cells (Emmen et al, 2000; Halls et al, 2007). Rxfp2 is one of many G- protein—coupled receptors that activates cAMP/PKA, a response that regulates multiple developmental processes, including neurogenesis and myogenesis (Lonze and Ginty, 2002; Chen et al, 2005). However, little is known of the downstream effects of cAMP/PKA signaling in fetal gubernaculum beyond cellular proliferation. In vivo, hyperplasia and extracellular matrix production in the fetal gubernaculum is followed by maturation of muscle precursors that become peripherally oriented to form the striated cremaster muscle (Radhakrishnan et al, 1979; Wensing, 1986). Cultured GD17 rat gubernacula enlarge in response to synthetic androgen R1881 without Insl3 but contain poorly organized myosin-positive cells within the mesenchymal core. However, when cultured with testis, they contain a defined outer layer of muscle (Emmen et al, 2000). Marked atrophy of the fetal gubernaculum with loss of the inner mesenchymal core is characteristic of both Insl3 and Rxfp2 null mice (Kubota et al, 2001). These data suggest that Insl3 may play a role in the regulation of both matrix remodeling and muscle development within the gubernaculum.

Other rodent data support a role for Insl3/Rxfp2 and additional candidate genes in regulation of myogenesis during development of the gubernaculum. Rxfp2 mRNA is present throughout the gubernaculum at GD16 in rat (Scott et al, 2005), but by GD19, binding sites for Insl3 are localized to the outer muscle layer (McKinnell et al, 2005), whereas the androgen receptor continues to be expressed in both mesenchyme and muscle (Staub et al, 2005). A cell-specific developmental role for androgens in the gubernaculum is not clear; however, after prenatal exposure to the antiandrogen flutamide prior to GD17, both mesenchymal and muscular compartments of the GD20 rat gubernaculum are reduced in size (Cain et al, 1995) and embryonic muscle isoforms persist in adult cremaster muscle (Tobe et al, 2002). Hoxa10 transcripts are also expressed throughout the GD15.5 gubernaculum, and histological studies of the postnatal cremaster muscle in Hoxa10 (−/–) males show disordered myogenesis (Satokata et al, 1995).

Expression patterns of specific genes that are involved in muscle development, contraction, or both (Table 4; Figure 5) support our global functional analysis results and suggest that terminal differentiation of muscle is delayed or disrupted in orl gubernaculum. Expression levels of Myog and Myf6, myogenic regulatory factors that control later stages of muscle differentiation (Sartorelli and Caretti, 2005), are reduced, whereas several genes that are down-regulated during or inhibit terminal differentiation of muscle (Melnikova et al, 1999; Ohkawa et al, 2006), including Igf1, Id2, Msx1, and representatives of the fibroblast growth factor family, show increased expression in the orl fetal gubernaculum. We also identified altered expression of several genes that promote skeletal muscle development, including Igfbp5, Ilk, Bmp7, and Fst (Huang et al, 2000; Amthor et al, 2002). Expression of Rps6kb1, Eif4e, and Eif4ebp1 are reduced in orl gubernaculum. These genes are effectors of insulin and a mammalian target of rapamycin (mTOR) signaling that regulate protein synthesis and cell size (Ruvinsky and Meyuhas, 2006); mTOR signaling is also critical for myoblast fusion (Park and Chen, 2005). Reduced expression of these genes is consistent with the global reduction in protein synthesis, as well as the reduced expression of muscle-specific genes that we observed in orl gubernaculum, with previous microarray data showing increased expression of energy and metabolism genes and decreased expression of genes involved in DNA replication and transcription during skeletal myotube maturation (Park and Chen, 2005).

In addition to muscle-specific genes, we identified altered expression of genes related to small GTPase signal transduction, cytoskeleton organization and biogenesis, and focal adhesion. The Rho GTPases encode proteins that are responsive to G- protein—coupled receptor and receptor tyrosine kinase signaling and are critical for cytoskeletal reorganization, cell motility, axon guidance, and myogenesis (Kjoller and Hall, 1999; Bishop and Hall, 2000; Bryan et al, 2005). Several, including RhoA, Rac1, Cdc42, and RhoC, are differentially expressed between strains (Table 4). Focal adhesions are sites of cell attachment to the extracellular matrix comprising integrins, cytoskeletal proteins, and signaling molecules (Sastry and Burridge, 2000). Possible roles for focal adhesion signaling in the developing gubernaculum include regulation of myoblast maturation (Clemente et al, 2005), migration (Mitra et al, 2005), or both; formation of costameres (Z-bands anchoring myofibrils to the sarcolemma) (Quach and Rando, 2006); and axon pathfinding (Robles and Gomez, 2006). Expression of the mRNA for several genes that participate in focal adhesion signaling, including Ptk2, Kdr (Flk1), Src (v-src), and Csk (Sastry and Burridge, 2000; Mitra et al, 2005) is present in the GD16.5 mouse gubernaculum (Verma-Kurvari and Parada, 2004). Csk encodes a tyrosine kinase linked to focal adhesion turnover and regulation of the actin cytoskeleton (McGarrigle et al, 2006), and the corresponding protein is localized to both mesenchymal and muscle layers in GD16.5 mouse gubernaculum. Ilk, a key component of integrin-mediated signaling that plays a role in the switch from myogenic proliferation to differentiation (Huang et al, 2000), is expressed at lower levels in orl gubernaculum.

Although the genetic basis for human cryptorchidism remains largely unknown, review of gene defects associated with cryptorchidism as compiled in Online Mendelian Inheritance in Man (OMIM) supports our present data. Several syndromes that include cryptorchidism are linked to genes that participate in small GTPase signaling (SOS1, KRAS, FGD1), actin cytoskeleton regulation (FLNA, FLNB), muscle development (ACTA1), or muscle contraction (RYR1). Expression of some of these genes is altered in orl gubernaculum (Table 4). In humans, the gubernaculum is comprised primarily of mesenchyme, but striated muscle is present within its distal portion in addition to the surrounding cremaster muscle (Tayakkanonta, 1963; Barteczko and Jacob, 2000; Costa et al, 2002), whose role, if any, in testicular descent is unclear. It is notable, however, that cryptorchidism is present in multiple forms of congenital myopathy (OMIM) and is also present in males with Prune-Belly syndrome (Jennings, 2000) and at a higher frequency in males with abdominal wall defects (Kaplan et al, 1986). Moreover, altered structure and function of the cremaster muscle has been reported in cryptorchid boys (Tanyel et al, 2000). These observations taken in combination with our present study suggest that muscle patterning might play a more important role in development and function of the gubernaculum than previously recognized.

Limitations of this study include our analysis of tissue-specific compared with cell-specific gene expression and the requirement for amplification of gubernaculum but not testis. Although the amplification, analysis, or both could be biased toward a particular cell type, we have been unable to identify any clear differences in cellular composition of wt compared with orl gubernaculum using cell-specific immunostaining (data not shown). Because of differences in RNA processing, we avoided direct comparisons of gene expression in testis and gubernaculum. Also, the possibility exists that early transcriptosomal changes associated with male specific gubernacular development were missed because gene expression is already sexually dimorphic in wt at GD18. Therefore, although we can identify global expression profiles that reflect development of wt and orl gubernacula, we cannot determine whether differences in gene expression are the cause or result of abnormal development. Moreover, because testicular descent does not occur until the postnatal period in the rat, we are unable to determine at the fetal stage which gubernacula (less than half) will be associated with cryptorchid testes. Despite this, the expression profiles of individual orl samples are highly similar and markedly different from wt in gubernaculum but not testis and therefore likely phenotype-specific as opposed to strain-specific. The basis for reduced penetrance of the phenotype remains unknown at this time but might be related to a dosage effect determined by environmental factors, modifying loci, or both. To date, we have no evidence that maternal- or paternal-specific factors determine phenotype because the frequency of cryptorchidism in offspring does not appear to be related to paternal phenotype or identity of the dam (unpublished observations).

Analysis of gene expression in fetal tissues of wild-type and cryptorchid orl mutant rats suggests that a primary gubernacular defect that directly or indirectly affects muscle function might predispose to cryptorchidism in the affected strain. Further studies will be necessary to elucidate the mechanism of gubernacular dysfunction in the orl rat.


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References
  1. Supported by NIH grant P20 RR-020173–01. The authors have nothing to disclose.