| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Testis; |
Department of Human Anatomy and Genetics,5 University of Oxford, Oxford OX1 3QX, United Kingdom
and Mammary Apoptosis and Development Group,4 Department of Pathology, University of Cambridge, Cambridge CB2 1QP, United Kingdom
Human Molecular Genetics Group,3
ABSTRACT
Testicular apoptosis is involved in the regulation of germ cell numbers, allowing optimal sperm production. Apoptosis has been described to occur in response to the absence of hormonal stimulation of the testis. Here we investigate the effect of the physiological lack of gonadotropins from birth using the hypogonadal (homozygous for the mutant allele Gnrh1hpg) mouse as a model. We pursued a concerted strategy using microarray analysis and RT-PCR to assess transcript levels, TUNEL to quantify the incidence of apoptosis, and Western blotting to assess the respective contribution of the extrinsic and intrinsic apoptotic pathways. Our results indicate a large increase in apoptosis of both somatic and germ cell compartments in the hpg testis, affecting Sertoli cells as well as germ cells of all ages. We confirmed our observations of Sertoli cell apoptosis using anti-Mullerian inhibiting substance staining and staining for cleaved fodrin alpha. In the somatic compartment, apoptosis is primarily regulated via the membrane receptor (extrinsic) apoptotic pathway, while in the germ cell compartment, regulation occurs via both the mitochondrial (intrinsic) and membrane receptor (extrinsic) apoptotic pathways, the latter potentially in a stage-specific manner. This study is the first report of spermatogonial apoptosis in response to gonadotropin deficiency as well as the first report of Sertoli cell apoptosis in response to gonadotropin deficiency in the mouse.
apoptosis, follicle-stimulating hormone, gonadotropin-releasing hormone, Hpg,, leutenizing hormone, sertoli cells, testis
Initiation and maintenance of a functional reproductive axis relies on secretion of GnRH from the hypothalamus. GnRH stimulates secretion by the pituitary of the gonadotropins LH and FSH, which in turn regulate testicular function [1]. These hormones are key modulators of gene expression during spermatogenesis. FSH is required for proliferation and differentiation of Sertoli cells, the "nurse cells" of the testis. Sertoli cells span the width of the seminiferous tubules from basement membrane to lumen, and maintain close cellular contacts with all stages of developing germ cells. This allows them to mediate signaling between different cell types that are important in the maintenance of spermatogenesis [2]. LH is necessary for the normal function of Leydig cells. These cells are found in the interstitial tissue between the seminiferous tubules and produce testosterone in response to LH stimulation [1].
The hypogonadal mouse (hpg) is homozygous for the mutant allele Gnrh1hpg. This allele is an interstitial deletion in the Gnrh1 gene, and thus the homozygote completely lacks GnRH [3]. There is a consequent failure of secretion of gonadotropins (FSH and LH) and testosterone, causing infertility [4–6]. The testes are underdeveloped (2%–5% of wild-type [WT] adult size) with reduced tubule diameter and no lumen, and spermatogenesis does not progress beyond the pachytene spermatocyte stage [3]. The lack of FSH in the hpg mouse leads to impaired postnatal Sertoli cell proliferation and an alteration in Sertoli cell:germ cell ratio. This in turn results in reduced germ cell numbers and reduced support of germ cell proliferation and differentiation [7, 8]. Due to the lack of LH in the hpg mouse, Leydig cells remain sparse (
10% compared to the WT) and morphologically immature [9]. Testicular testosterone concentration is normal during fetal life, indicating that during this period, testosterone production is LH independent. However, testosterone levels are undetectable from Day 5 after birth in this model [10], thus the observed phenotype is due to the compound lack of both gonadotropins and testosterone.
The purpose of this study was to investigate the transcriptional changes associated with the abnormal developmental patterns in hpg testes, and in particular to examine the contribution of apoptotic processes to the observed phenotype. It is known that the arrested pachytene germ cells in adult hpg mice are eliminated by apoptotic mechanisms rather than by necrosis, because there is no associated inflammation of the testis. However, it is not yet known whether the changes in Sertoli cell numbers in hpg testes are due to increased apoptosis or which apoptosis pathways are involved in cell death in hpg testes.
Apoptosis is a normal part of testicular development, serving to adjust the Sertoli cell:germ cell ratio during the first wave of spermatogenesis and thus establishing normal testicular architecture [11]. Apoptosis is also a common pathway for clearing of aberrant/arrested germ cells [12]; for example in humans, maturation arrest has been associated with germ cell apoptosis [13]. Debris from degenerating germ cells is cleared through phagocytosis by Sertoli cells [4]. Testicular apoptosis is at least partly hormone dependent. Lack of FSH has been reported to increase DNA fragmentation in primary spermatocytes and spermatids without involving the caspase pathways, with no effect on Sertoli cell DNA integrity or caspase activity in the adult human male [14]. In vitro testosterone withdrawal increases DNA fragmentation and caspase activity in human Sertoli cells, but the effect on germ cells is minor [14].
Apoptosis may be initiated either by extracellular or intracellular signals, the first stimulating the extrinsic (receptor-mediated) pathway, and the second the intrinsic (mitochondrial) pathway. FAS and its ligand, FASL, form one of the best-known apoptotic receptor pathways, and have been reported to have a key role in apoptosis during the first wave of spermatogenesis [15]. However, it is not yet clear which cell types are subject to FAS-mediated apoptosis: some studies show FAS to be present on germ cells and FASL to be present on Sertoli cells [16–18], whereas others show the reverse [19–23]. The intrinsic pathway is also known to be involved in balancing the germ cell:Sertoli cell ratio. Spermatogonial apoptosis is induced by BAX and inhibited by BCL2L1 (formerly BCL-XL), BCL2L2 (formerly BCL-W), KITL (Kit ligand, also known as stem cell factor) and testosterone [24]. Spermatocyte apoptosis is induced by BAX and inhibited by BCL2L1 and testosterone; hence testosterone appears to be a crucial survival factor preventing apoptosis [24]. Further evidence of a link between testosterone and the intrinsic pathway of apoptosis is that testicular expression of Bcl2l1 and Bcl2 was shown to be altered by antiandrogen treatment of prostate cancer [25].
Other studies using hormonal interventions or other pharmacological treatments (e.g., estradiol, GnRH antagonists, the antiandrogen Flutamide, or the Leydig cell toxicant ethane disulfonate [EDS]) have also led to observations of increased apoptosis [26–31], thus it appears clear that appropriate hormonal support is necessary to prevent mass apoptosis in the testis. The use of the naturally occurring hpg mutant mouse, effectively a GnRH knockout, provides a physiological model for the effect of constitutive hormone absence, in contrast to the use of pharmacological inhibitors.
The development of testis-specific microarrays has allowed us to investigate all possible apoptotic activity within the testis. Using microarray analysis, we highlight gene expression alterations related to apoptosis due to the congenital deficiency of gonadotropins in the hpg mouse. As spermatogenesis in the hpg mouse arrests at pachytene spermatocytes, the microarray study was concentrated on the early stages of spermatogenesis, where the key gene expression changes involved in the spermatogenic failure are likely to occur. These findings show an increase in expression of genes and proteins involved in both the intrinsic and extrinsic apoptotic pathways and are consistent with previous studies using antagonists of hormone action.
The hpg mice used in this study were from the original colony discovered at the MRC Laboratories, Harwell, Oxford [3] and were bred in the Department of Physiology, Anatomy and Genetics, Oxford under Home Office license. The Gnrh1hpg mutation was identified by PCR analysis of tail DNA as previously reported [32].
All procedures were carried out in accordance with the Animals (Scientific Procedures) Act 1986 and with the approval of a local ethical review committee. Testes were collected on Postnatal Days 3, 8, 13, 19, and 27 and from 8-wk-old adult mice. The day of birth was taken as Day 1. Mice were killed by cervical dislocation, the testes dissected out, weighed, and snap frozen in liquid nitrogen for RNA and protein analysis. Tissues were stored at –70°C until use. Testes taken from a further set of mice were collected for histological examination and immunohistochemistry.
XXSxrb [33] testes were a kind gift from Dr. Burgoyne at the National Institute for Medical Research, Mill Hill, U.K. XXSxrb mice carry the testis-determining gene Sry and thus are physiologically male; however, the second X chromosome is toxic to pro-spermatogonia, leading to complete absence of germ cells from the seminiferous tubules.
Total mRNA was extracted from WT and hpg mice at each time point, and the tissues were divided into two pools, forming biological replicates A and B for Postnatal Day 3 (WT: n = 36, hpg: n = 36), Day 8 (WT: n = 18, hpg: n = 18), Day 13 (WT: n = 13, hpg: n = 13), and adult (8 wk; WT: n = 5, hpg: n = 32). Total mRNA was extracted by homogenizing (UltraTurrax, IKA) in TRI reagent (Sigma) and purified using MinElute columns (Qiagen, Valencia, CA) following the manufacturer's instructions. Following DNase treatment (Qiagen), RNA quality was checked using an Agilent Bioanalyser 2100 (Agilent Technologies, Palo Alto, CA), and RNA concentration was measured with an ND-1000 Spectrophotometer (NanoDrop). Total protein was extracted from frozen testes at Postnatal Day 3 (WT: n = 3, hpg: n = 3), Day 8 (WT: n = 2, hpg: n = 2), Day 13 (WT: n = 2, hpg: n = 2), and adult (8 wk; WT: n = 2, hpg: n = 2 ) for WT and hpg mice as previously described [34]. Testes from the same time point were pooled in order to obtain sufficient amounts of protein.
Complementary DNA array hybridizations were performed on the MmcDNAv1 chip. The gene set printed on this array is drawn from three sources: 1) Two subtracted cDNA libraries generated in-house as part of a previous project [35] and enriched for testis-specific genes (MTn library) and germ cell-specific genes (Sxrb library); 2) Plates 8825–39/8846–50/9339–42/13869–74 of the IMAGE cDNA collection, which are drawn from a set of six testis-purified cell type libraries [36] and contain genes expressed in testicular cell types; and 3) The NIA7.4 set [37], which is a curated set of clones representing 7400 different genes expressed in various stages of mouse embryogenesis.
The MmcDNAv1 array is a largely nonredundant cDNA clone set containing all genes present in any of the above libraries: 11 000 different mouse genes in total. Clones were printed in duplicate on the array to improve the accuracy of quantitation.
For one crucial time point (Postpartum Day 13), RNA samples were also hybridized on the mouse oligo chip MEEBO (Mouse Exonic Evidence Based Oligonucleotide) representing 25 000 mouse genes. The Centre for Microarray Resources, Pathology Department, Cambridge University, U.K., provided the arrays [38]. The cDNA target was amplified, labeled, and hybridized as described by Petalidis et al. [39] except that 1) a constant number of 19 cycles was used and 2) for the labeling step, 1 µl of Cy3 or Cy5-dCTP was used with 22 µl of second-strand cDNA. The labeled products were purified using G50 columns, according to the manufacturer's instructions (Amersham Biosciences U.K., Ltd.). Labeled samples were combined (as per experimental design, see below), and competitive hybridization to the arrays was performed for 16 h at 48°C for the mmcDNAv1 and 50°C for the MEEBO in the presence of 4 µl of Cot-1 mouse DNA, 1 µl PolyA (8 mg/ml), and 1 µl yeast tRNA (4 mg/ml).
Arrays were scanned using an ArrayWorx CCD-based scanner (Applied Precision, LLC), and spot quantification was performed using BlueFuse software (BlueGnome, Ltd., Cambridge, U.K.). For the cDNA chip, where each spot is duplicated in the array, the fused option of the BlueFuse output data was chosen for further analysis. This combines the value of replicates of the same clone for each experiment.
Labeled cDNAs prepared from staged hpg testes were compared to age-matched WT. Microarray hybridizations were performed on two biological replicates for each comparison. For each biological replicate, four technical replicates were obtained (including dye reversal), thus giving a total of eight hybridizations for each comparison.
Knowledge Discovery Environment (InforSense, Ltd., London, UK) was used to analyze the microarray data, which were normalized to the median of overall intensities for the slide and filtered by sorting into 10 equal-width bins across the absolute fluorescence intensity spectrum. The most variable clones (the 10% with the highest coefficients of variation) in each bin were excluded from further analysis [40]. Upon further analysis, some of these clones were reincluded following identification and elimination of a single replicate outlier. In order to identify differentially expressed genes in the age-matched comparison, we used an intensity-dependent Z score. This measures the number of standard deviations a particular data point is from the mean relative to other clones expressed at a similar absolute level [41]. Clones with ratios of two standard deviations from the mean of the window were considered differentially expressed (95% confidence level). All genes were annotated using SOURCE (Genetics Department, Stanford University), a gene ontology package, providing information on potential biological function [42].
To determine potential germ or somatic cell expression of genes on the array in cases where this information was not available from the scientific literature, we also hybridized the total testis mRNA from a germ cell-deficient mouse model XXSxrb [43]. This, in combination with previously published work from our laboratory, allows broad assignment to testicular cell types [35, 44]. The ratio observed in this germ cell-deficient mouse model gives an indication of whether the gene is mainly expressed in the somatic compartment, germ cell compartment, or both.
The expression levels of seven genes identified as differentially expressed from our microarray data were verified by qRT-PCR, together with the expression levels of another nine genes not present on the array. This allowed us to extend our knowledge of the transcriptional changes related to apoptosis occurring in the hpg testis. Beta actin was used for normalization of the RNA concentration and each qRT-PCR was performed in triplicate. Real-time RT-PCR was performed in 96-well white plates (Abgene) using a QuantiTect SYBR green RT-PCR kit (Qiagen) according to the manufacturer's protocols; the resulting fluorescence was quantified using an iCycler system (Bio-Rad). For each reaction, 100 ng total RNA was used. The RT step was performed at 50°C for 30 min, followed by activation of the PCR enzyme at 95°C for 15 min. The PCR cycling conditions were 95°C for 15 sec, 53°C for 30 sec, 72°C for 30 sec, and 77°C for 10 sec. Real-time fluorescence data were captured during the 77°C step of the cycle. Melt curve data were obtained to confirm amplification of the correct product in each well. The threshold cycle value (CT) was obtained for each well as the cycle number at which the measured fluorescence crossed the arbitrary threshold of 100 units (at which all the values are in log phase). The average CT was calculated for each gene in each sample. Data were normalized by reference to beta actin, with 
CT calculated as follows: CT = CT(test) – CT(actin). CT values were then calculated as the change in CT for the hpg genotype relative to the CT value for WT. The CT value is inversely proportional to the abundance of the transcript. In order to allow comparison to the microarray data, the inverse of CT was plotted on the graphs (primer sequences are shown in Table 1).
|
Samples were resolved in 8%–15% SDS-polyacrylamide gels (Bio-Rad), transferred to PVDF membranes (Immobilon-P, Millipore), and blocked in PBS with 0.05% Tween-20 (PBST) and 5% nonfat powdered milk for 30 min before incubation with primary antibody overnight at 4°C. Membranes were quickly washed twice with dH2O, then horseradish peroxidase (HRP)-conjugated secondary antibody was added (goat anti-rabbit HRP, DAKO) and incubated for 1 h at room temperature. Following two further quick H2O washes, membranes were washed twice for 5 min in PBST, and proteins were detected by enhanced chemiluminescence. Beta actin was used as a loading control. Primary antibodies used were caspase 9 (1:1000; Abcam 32539), caspase 8 (1:1000; Cell Signaling 4927), cleaved caspase 3 (1:1000; Cell Signalling 9664), caspase-3 (1:1000; Upstate) and beta actin (1:1000; Abcam).
Standard wax sections (1 µm) were stained with hematoxylin and eosin. In all cases, two biological replicates were used for histological data. Images were taken using a Leica DMLB microscope (Leica Microsystems). Measurements were taken using image J [45].
Fluorescent Immunohistochemistry
Sections of paraffin-embedded-buffered formalin fixed testis were dewaxed in xylene and rehydrated in alcohol. Antigen retrieval was by pressure cooker for 10 min at high pressure in 10 mM citric acid buffer (pH 6.0). Sections were blocked in 10% goat serum for 2 h at room temperature. Primary antibody, diluted (1:100) in 5% goat serum, was added to the sections and incubated at 4°C. Cleaved caspase 3 antibody was incubated for 48 h, Caspase 8 antibody was incubated for 12 h, and cleaved fodrin antibody (asp 1185; Calbiochem) was incubated for 8 h. Following PBS washes, secondary Alexa Fluor goat anti-rabbit 488 antibody (1:200 in 5% goat serum; Molecular Probes Europe BV, Leiden, The Netherlands) was incubated on the sections for 2 h. Sections were washed in PBS, and the nuclei were counterstained with Hoechst stain (bisbenzimide-Hoechst 33342, Sigma), mounted in 50% glycerol, and imaged. Staining was visualized with an Axiovert-S100TV inverted microscope (Zeiss).
Sections of paraffin-embedded Bouins fixed testis were dewaxed in xylene and rehydrated in alcohol. TUNEL staining was performed using the ApopTag plus peroxidase in situ apoptosis detection kit (Chemicon, Millipore), according to the manufacturer's protocol. Nuclei were counterstained for 5 min with Carazzi hematoxylin and washed for 5 min in Scott water. Sections were dehydrated in methanol and cleared in xylene before mounting in DPX and visualized with a Leica microscope.
Identification of Sertoli Cells Using Immunostaining for MIS
Sections were dewaxed in xylene and dehydrated in ethanol, and antigen retrieval was in 10 mM citric acid buffer (pH 6.0) for 15 min. Following PBS washes, slides were immersed in 2 M HCl for 20 min and again rinsed in PBS. To inactivate internal peroxidases, sections were covered in methanol + 3% hydrogen peroxide for 10 min. Blocking of the section was performed with 10% normal goat serum (in PBS) for 1 h at room temperature. Sections were incubated for 12 h with anti-MIS (Mullerian inhibiting substance) antibodies C-20 (goat polyclonal; 1:200 dilution; Santa Cruz Biotechnology). Histofine Simple Stain MAX PO (G) kit (Nichirei, Tokyo, Japan) was used for secondary antibody staining following the manufacturer's instructions. Nuclei were counterstained with Carazzi stain and washed in Scott solution. Sections were visualized with a Leica microscope.
Statistical Evaluation of the TUNEL Results
The proportion of positive seminiferous tubules was counted on multiple sections from two different mice for each experimental condition (n = 6 for WT samples and hpg Day 8 and Day 13, n = 7 for hpg Day 3 testes) and the average number of positive cells per positive tubule was calculated by looking at all positive tubules in each of the sections [46]. A t-test was performed to identify statistically significant differences; *, P < 0.01; , P < 0.05.
Degree of Replication Within the Experimental Analyses
Unfortunately, hpg mice are sterile in the homozygous state, and only one hpg male in eight results from the heterozygous breeding; therefore, only a limited degree of biological replication was possible for some of these experiments. The histological and Western blot data are thus based on small numbers of animals (two to three, see above for details), and further experimentation will be necessary to confirm the statistical significance of these results. The expression data are based on a larger cohort of animals (see above). For technical reasons, these were pooled to give two biological replicates; however, the pooling process should render our expression results robust to individual variation.
Histological Analysis of hpg Testes
As in previous studies [3, 9], we found a marked reduction in average testis weight in hpg mice relative to WT mice from the same colony. This reduction was seen as early as Postnatal Day 3, and was statistically significant by Day 8 and at subsequent ages (P < 0.01). At Day 3, testis weight was reduced by 13%, by 46% at Day 8, by 64% at Day 13, and by 87% at Day 19 (see Supplementary Fig. 1; all supplemental material is available online at www.biolreprod.org). We also observed a statistically significant decrease (P < 0.01) in tubule diameter from Postnatal Day 8 onwards. The average diameter was reduced by 9% at Day 8, 27% at Day 13, 51% at Day 19, and 66% at Day 27. There was no significant difference between the two biological replicates for any of the experimental conditions. Representative images of hpg and matched WT testes are available online (Supplementary Fig. 2).
|
|
Detection of Apoptosis in hpg Testes
TUNEL staining was used to detect apoptosis in sections from hpg testes in comparison to age-matched WT control (Fig. 1). There was no significant difference between biological replicates for any experimental condition. There was a significant rise in the proportion of TUNEL-positive tubules at all ages in the hpg samples, and from Postnatal Day 8 onwards there was also an increase in the number of positive cells per positive tubule. This is especially pronounced at Day 8 and Day 13, with 30% and 32% of hpg tubules, respectively, showing apoptosis, compared to 10% and 11% of WT tubules. After Day 13, there is a reduction in apoptosis in the hpg testes; however, it still remains significantly elevated relative to WT.
This technique identifies germ cells undergoing apoptosis (Fig. 2) and also occasional Leydig cells (Supplementary Fig. 3). However, Sertoli cells and myoid cells are less easily detected. In particular, techniques that label the nucleus of apoptotic cells are poor at detecting Sertoli cell apoptosis, because it is often unclear whether a nucleus belongs to a Sertoli cell or to an associated germ cell. We used two techniques to detect whether Sertoli cells undergo apoptosis in hpg testes. Firstly, we carried out a co-staining experiment using an anti-MIS antibody to specifically label Sertoli cells, with Carazzi counterstain to detect condensed nuclei indicative of apoptosis. This allows a clearer identification of any given apoptotic nucleus as a Sertoli cell nucleus. Secondly, we carried out immunostaining for fodrin cleaved at Asp1185. This is a caspase cleavage product of the membrane protein fodrin (non-erythroid alpha spectrin) and is thus a specific marker of apoptosis. This antibody labels the membrane and cytoplasm of apoptotic cells and thus shows a clear distinction in the pattern of staining for Sertoli cells and germ cells.
|
At Postpartum Days 8 and 13, we observed condensed nuclei in cells with MIS staining, demonstrating Sertoli cell apoptosis in the hpg mouse testis (Fig. 3). Using the anti-cleaved-fodrin antibody, staining of large cells with branching cytoplasm (Sertoli cells) is occasionally observed at Postpartum Days 8 and 13 in the hpg mouse testis, but is never seen in WT testis (Fig. 4). This method also detects apoptosis in germ cells, which are the smaller, rounder cells labeled in Figure 4.
|
Transcriptional Analysis of hpg Testes
We carried out two-color microarray expression profiling comparing expression in developing hpg testes to age-matched WT controls, using a cDNA clone set (mmcDNAv1) containing 11 000 genes. Because the spermatogenic arrest in hpg mice first becomes histologically observable in the Day 13 sample, we carried out a second examination of this sample using the more extensive MEEBO oligonucleotide array. The data set generated in this experiment was submitted to the ArrayExpress database, accession number E-MEXP-1153.
Verification of Transcriptional Results and Pathway Annotation
We used gene ontology (GO) data to filter the lists of deregulated genes at each time point and determine which of the deregulated genes were likely to be involved in apoptotic processes. Full lists of genes that were deregulated at each time point are available online as Supplementary Tables 1–4. In all, 168 apoptosis-related genes were found to be deregulated at Day 3, 132 at Day 8 and 137 at Day 13, using the cDNA clone set. 83% of the apoptosis-related genes found to be deregulated on the cDNA clone set were confirmed in the MEEBO analysis. We used quantitative RT-PCR to verify our array data for Casp3, Casp8, Casp9, Diablo, Daxx, Fadd, and Fasl, in all cases obtaining results consistent with the array analysis.
Having observed deregulation of apoptosis-related genes in hpg testes, we wished to see which apoptotic pathways were being activated, whether the genes involved had pro- or antiapoptotic activity, and in which cell types within the testis they were expressed. This type of pathway information is very poorly represented in gene ontology databases; therefore, we supplemented the GO information with literature searches using the NCBI's Gene and PubMed databases in order to determine pathway information for each of the deregulated genes. A number of key apoptosis genes from various pathways failed to give signal from the arrays, presumably due to low transcript abundance, so we extended the more sensitive quantitative RT-PCR screen to gather data for these genes as well. The genes included in this extended PCR screen were Apaf1, Bax, Bbc3 (formerly Puma), Bcl2l10 (formerly Diva), Bid, Birc3, Casp6, Fas, and Tnf. Of these, all except Birc3, Casp6, and Fas showed altered expression in at least one of the time points investigated.
Table 2 summarizes the genes used in our investigation of apoptotic pathways and indicates the data sources available for each. In all, there were 29 genes that showed differential expression in either the array or qRT-PCR studies and for which full pathway data were available (and with agreement between techniques of detection). We will refer to this restricted set of 29 genes as "pathway-mapped, differentially expressed" genes.
|
Determination of Cellular Expression Distribution
The first key question is whether these pathway-mapped, differentially expressed genes are expressed in germ cells or somatic cells, or are ubiquitous. Where this information was not already available in the literature, we determined this by examining the expression levels in testes from XXSxrb mice, which completely lack germ cells (summarized in Table 2, see also Supplementary Table 5 for full data from the XXSxrb comparison). Although the expression of apoptosis genes in XXSxrb testes is likely to be disturbed due to the Sertoli-cell-only phenotype, it is safe to conclude that apoptosis genes that are highly expressed in XXSxrb must be expressed in somatic cell types in this model, and thus that somatic cells are capable of activating the corresponding pathway. Similarly, low expression in XXSxrb indicates preferential expression in germ cells and thus that the corresponding pathways may be involved in germ cell apoptosis.
Consistent with existing data on germ cell apoptosis, we found that genes from the intrinsic pathway tended to be expressed in germ cells. Of the 15 genes annotated to this pathway, five (Bag1, Bcl2l11, Bcl2l14, Casp9, and Tnfrsfip1) appeared from the literature search and/or XXSxrb data to be expressed in germ cells, two (Bcl2l2 and Bax) had a mixed pattern, and only one (Pak7) was likely to be overexpressed in somatic cells. Data were unavailable for the other seven genes (including Bcl2, which is not expressed in the WT testis). Importantly, the genes from this pathway preferentially expressed in germ cells included the critical effector caspase for this pathway, Casp9.
Intriguingly, we found that several genes from the extrinsic pathway were expressed in somatic cells. Of the 10 genes annotated to this pathway, four (Casp8, Dapk1, Tnfrsf1a, and Traf6) appeared to be strongly expressed in germ cells, four (Daxx, Eif5a, Fasl, and Ripk1) showed a mixed pattern of expression, and only one (Tnf) appeared to be germ cell specific. No localization data were available for Fadd. Importantly, one of the critical effector caspases for this pathway, Casp8, was overexpressed in XXSxrb, indicating that the extrinsic pathway is likely to be active in somatic cells.
Time Course Examination of Pathway-Mapped Gene-Expression Data
Figure 5 summarizes the transcriptional data for all pathway-mapped differentially expressed genes that gave a significant result by cDNA microarray or qRT-PCR. Table 3 shows all pathway-mapped differentially expressed genes that gave a significant result on the MEEBO oligonucleotide array. We compared the pathway annotation derived from the literature with the cellular distribution data obtained from the XXSxrb comparison and with our data from hpg testis development.
|
|
The common downstream effector caspase, Casp3, was detected by qRT-PCR as significantly upregulated in hpg testis at all ages. The BIRC family is a family of antiapoptotic proteins, formerly called IAP (Inhibitor of Apoptosis Protein). This family is represented by six members on the arrays, but only one shows differential expression in the hpg mouse testis. Birc4 has a protective role against apoptosis mediated by FAS or TNF alpha and is downregulated at all time points in the hpg mouse testis. Diablo, an antiapoptotic mediator, was found to be upregulated at Day 3 and downregulated by Day 13.
The key effector of this pathway, Casp8, showed no clear pattern of changes in transcript abundance in hpg compared to WT, although it was slightly downregulated at Day 8 and slightly upregulated at Day 13. However, a number of regulators of this pathway were altered in expression. Tnfrsf1a and Eifa5a (proapoptotic) are significantly upregulated at all time points in the hpg mouse relative to WT littermates. Fadd (regulatory) and Tnf (proapoptotic) are significantly upregulated at Day 3, whereas Ripk1 (proapoptotic) is downregulated at Day 8. Both the microarray and qRT-PCR data indicated a slight downregulation of Fasl at all time points, but this was only statistically significant on the microarrays. As noted above, no significant change was observed in Fas at any time point using qRT-PCR. The slight downregulation of Fasl may be due to increased activity of the TNF alpha pathway [43], since we observed the upregulation of Tnfrsf1a and Tnf itself.
The key effector of this pathway, Casp9, is upregulated in hpg testis at all time points, as is the proapoptotic regulator Sphk2. At Day 3, other upregulated genes include Apaf1 and Bcl2l11 (proapoptotic), while downregulated include Bcl2l14 (proapoptotic). At Day 8, upregulated genes include Bcl2l10, Bid, Apaf1, Bcl2l11 (all proapoptotic), and Bbc3 (regulator), while downregulated genes include Bax (proapoptotic). At Day 13, upregulated genes include Bcl2l14 (proapoptotic), Bcl2, and Bag1 (antiapoptotic), while downregulated genes include Bax (proapoptotic) and Diablo (antiapoptotic).
Endoplasmic Reticulum-Mediated Apoptosis
Casp12, the major effector of the endoplasmic reticulum (ER) apoptotic pathway, is upregulated in hpg testis at Day 13. From previously published data, the endoplasmic reticulum pathway seems to be important in apoptosis of both the germ cell and the somatic cell compartment [44]. Casp12 has been implicated in spermatocyte and spermatid ER-mediated apoptosis [47] and its action is mediated through activation of Casp9.
Western Blot and Immunohistochemical Investigation
We examined the developmental changes in protein expression for three key caspases in order to verify our above histological and transcriptional observations. Casp3 is the common downstream effector caspase, Casp8 is a major caspase in the extrinsic pathway, and Casp9 is a major caspase in the intrinsic pathway.
Cleaved Caspase 3 Protein Expression
We used two distinct antibodies for detection of caspase 3, one specific for the cleaved form and one that recognizes all three forms (proform and the two cleaved forms). The latter antibody was used in Western blot analysis, and the pro- and the cleaved forms were identified by their different molecular masses. The cleavage-specific antibody used on testis sections permits assessment of the distribution of activated caspase 3 within and between seminiferous tubules.
Western blotting showed that the quantity of procaspase 3 is constant between Days 3 and 8, and the concentration is similar in the hpg mouse testis and in the WT mouse testis. Cleaved caspase 3 is first detected in hpg mouse testes at Day 8, whereas it does not appear in the WT control until Day 13 (Fig. 6). This increase in caspase 3 activation in hpg mouse is also observed by immunostaining (Fig. 7). The cleaved caspase 3 signal is visible in apoptotic germ cells (all stages) at Days 8 and 13 in hpg and WT testes. A lower-intensity cytoplasmic staining is visible in germ cell cytoplasm at Day 3 in hpg but not in WT. The same low-intensity cytoplasmic staining is seen in Sertoli cells at Day 3 in both hpg and WT. This diffuse cytoplasmic distribution of cleaved caspase 3 has been described as an early event in the apoptotic process [48]. At later ages, Sertoli cell staining cannot be clearly observed. There is also Leydig cell labeling visible in some sections; however, this is not specific, as expression was also seen in the negative control (data not shown).
|
|
Cleaved Caspase 8 Protein Expression
Western blotting showed that the concentration of caspase 8 proform increased over the first 2 wk after birth in both control and hpg mouse testes, but more significantly in the latter (see Fig. 6). The larger of the two cleaved forms (45 kDa) is elevated in hpg mouse testes relative to the WT and increases further from Day 3 to 13. The smaller cleaved form (18 kDa) is not detected in the WT, and is present in hpg mouse testes from Day 8 onwards. Immunostaining showed few caspase 8-positive germ cells at Day 3 in either hpg mouse or WT testes, whereas labeling of Sertoli cells is apparent in almost all tubules in both hpg mouse and WT testes (Fig. 8). At this age, Sertoli cells are apparent as large cytoplasm-rich cells toward the center of the tubule, which has not yet canalized to form a lumen, and are distinguishable from the smaller, rounder germ cells by their shape. At later time points, caspase 8 staining is visible in germ cells of both hpg mouse and WT testes. The pattern of staining shows variation between tubules (data not shown), which may reflect different tubule stages of spermatogenesis. This can be seen at Days 8 and 13, where we see some tubules with extensive labeling of the germ cells (at different stages) and some with far fewer positive cells. Groups of adjacent stained cells may be part of a syncytium of sister cells that are known to persist up until the elongating spermatid stage in the WT mouse. It is, however, important to note that this caspase 8 antibody recognizes both the pro- and cleaved forms, and thus the immunohistochemical results do not differentiate the active from the inactive form.
|
No cleaved form of caspase 9 was visible by Western blotting, only the proform (46 kDa) (see Fig. 6), which is active in the apoptosome. The concentration of the proform of caspase 9 is constant in the first 2 wk after birth in the WT mouse testis. From Day 8, there is an increase in the protein concentration of capase 9 in the hpg mouse samples compared to WT age matched controls.
Lack of hormonal stimulation from birth has an effect very early on in testis development. Leydig cell function is independent of gonadotropin action in utero, and thus the hpg mouse undergoes normal male sexual differentiation during gestation, but becomes androgen deficient at birth [10], in addition to being constitutively gonadotropin deficient. The differences we observed in tubule size and testis weight in hpg mice relative to WT are consistent with the findings of Baker et al. [9]. However, the change in testis size and cell proportion may be due either to reduced proliferation, increased apoptosis, or both. We were interested to know whether apoptotic mechanisms are important in the response to hormonal insufficiency in hpg mouse testes and, if so, which apoptotic pathways were implicated in the germ cell and somatic cell compartments. We carried out TUNEL staining to show apoptosis in the germ cell population and also used two different methods (MIS/Carazzi staining and staining for cleaved fodrin) to show apoptosis in the Sertoli cell population. Germ cell apoptosis was apparent as early as Day 3, with apoptotic Sertoli cells being detected from Day 8 onwards. Our observations suggest that testicular apoptosis in hpg mouse peaks at Days 8 and 13 and remains elevated into adulthood.
In order to achieve the correct germ cell:Sertoli cell ratio for optimal sperm production, extensive but precise germ cell apoptosis is observed during the first wave of spermatogenesis in WT testis. In the hpg mouse, we observe a disruption of this balance in favor of increased apoptosis, leading to a failure of spermatogenesis progressing beyond the pachytene spermatocyte stage. The level of germ cell apoptosis in the testis is likely to be underestimated as a consequence of efficient phagocytic activity of Sertoli cells [49, 50]. The increased germ cell apoptosis we observe may, therefore, be due to a genuine increase in apoptosis, reduced clearance of apoptotic cells, or a combination of these factors. FSH has been shown to regulate the formation of adherens junctions and gap junctions in the testis, whereas testosterone has no effect [51, 52]. Thus, the lack of FSH stimulation of the Sertoli cell is likely to result in a nonoptimal physiological cellular environment.
On average, each Sertoli cell is associated with 30 germ cells in WT mice. In the hpg mouse this ratio falls to 1:1 [7]. This strongly suggests that the increase in observed apoptosis is not simply a consequence of reduced clearing. It is possible to identify different cell types involved in this process. In fertile men, all three classes of germ cells (spermatogonia, spermatocytes, and spermatids) undergo spontaneous apoptosis [30]. In the hpg mouse, at Days 3 and 8, spermatogonia are the only germ cell type in the tubules. By Day 13 spermatocytes have differentiated and, together with spermatogonia, they are subject to apoptosis (Fig. 2). To our knowledge, this is the first time that apoptosis in spermatogonia is reported to be linked with a lack of hormone stimulation in vivo.
We also observe Sertoli cells undergoing apoptosis at Days 8 and 13 in the hpg mouse model. Sertoli cell apoptosis has not previously been documented as an in vivo response to gonadotropin insufficiency. Sertoli cell apoptosis has been observed in the hypogonadal (Hgn) rat [53], whose adult testes reach 1% of the WT weight [54] and whose germ cells degenerate before entering meiosis [55]. However, adult Hgn rats display low levels of testosterone and high levels of gonadotropins, so this model is not a good analogue of the situation in the hpg mouse. Rather, the Hgn phenotype appears to be the result of a Leydig cell dysfunction [55]. Furthermore, the majority of Sertoli cell loss in this model occurs prenatally [56].
FSH has been described to have an antiapoptotic role in Sertoli cells by enhancing the activity of PKA and the aromatase-dependent activation of FAAH [57]. However, it is unlikely that the Sertoli cell apoptosis we observe is a consequence solely of FSH deficiency, because no report of Sertoli cell apoptosis was made in the FSHβKO and FSHRKO mouse models, which, respectively, lack FSH and its receptor [58, 59]. Passive immunization against FSH in prepubertal rats [43] also did not result in Sertoli cell apoptosis. The hpg mouse also lacks LH and thus testosterone. In vitro, testosterone withdrawal increases DNA fragmentation and caspase activity in human Sertoli cells [14], and testosterone has previously been shown to protect human Sertoli cells against experimentally induced apoptosis in vitro [60]. It is thus possible that Sertoli cell apoptosis in hpg is a consequence of testosterone withdrawal. However, studies in homozygous ARTfm mice (which lack the androgen receptor [AR]) [61], SCARKO mice (which specifically lack the AR in Sertoli cells) [62], and LuRKO mice (which lack the LH receptor) [63] have not reported Sertoli cell apoptosis. It thus seems unlikely that testosterone deficiency is the cause of Sertoli cell death in hpg testes. An alternative explanation is that the Sertoli cell apoptosis is a downstream consequence of germ cell apoptosis: the amount of debris to be phagocytosed might "overwhelm" the Sertoli cells in contact with multiple apoptotic germ cells. As shown in this study, the number of tubules with germ cells undergoing apoptosis at Days 8 and 13 represents one third of the tubules on each section. This is unlikely to be the sole explanation as there are many mutant models with impaired spermatogenesis and large-scale germ cell apoptosis, and Sertoli cell death is not usually reported as a consequence.
We suggest that Sertoli cell apoptosis in hpg is a result of the compound deficiency of gonadotropins and testosterone, which is why it has not been seen in other models investigating these pathways separately. To support this, the loss of germ cells following hypophysectomy leads to the disorganization of the cytoskeleton in the Sertoli cells [64]. However, another study reported that Sertoli cell apoptosis did not occur in adult rats treated with GnRH antagonist [27]. It may be that immature, mitotically dividing Sertoli cells are more sensitive to hormone deficiency than mature Sertoli cells and that the apoptosis is thus more apparent in developing testes.
There are two major pathways controlling apoptosis: the intrinsic (mitochondrial) pathway and the extrinsic (receptor-mediated) pathway. Our transcriptional data show an increase in expression of both pro- and antiapoptotic mediators of both pathways, suggesting that the control of testicular apoptosis is complex. In particular, we observed upregulation of the transcripts for the key caspases 8 (extrinsic pathway), 9 (intrinsic pathway), and 3 (shared between both pathways). Interestingly, our results indicate that the pathways may be differentially activated in germ cell and somatic compartments. Our transcriptional data indicate a preferential overexpression of extrinsic pathway genes in XXSxrb testes, indicating that these genes are expressed in somatic cells. Conversely, our transcriptional data indicate that intrinsic pathway genes are preferentially expressed in germ cells. Consistent with this, we observe caspase 8 staining from Day 3 in Sertoli cells, though at the same age it is present in very few germ cells. The differential germ cell staining between tubules at later ages suggests that there may be a tubule stage-specific mechanism involved (data not shown). In contrast, cleaved caspase 3 protein was only detected on Western blots from Day 8 in hpg mouse testes, so the significance of the Sertoli cell staining at Day 3 is not clear.
The intrinsic (mitochondrial) pathway involves the release of caspase activators from the inner membrane of the mitochondria, in particular cytochrome c. The major events in this pathway are the loss of mitochondrial transmembrane potential and changes in cellular oxidation-reduction. The intrinsic pathway is regulated by cytochrome c, APAF1 and pro-caspase 9 [65]. BCL2L10 and numerous other molecules regulate the formation of the apoptosome. The BCL2 family is responsible for the tight control of this pathway. It is well-documented that this pathway is involved in germ cell apoptosis, and this is borne out by our results. In hpg mouse testes, the activation of the intrinsic pathway appears to be due to a shift in the balance of pro- and antiapoptotic regulators. This is most pronounced at Day 8, where there is upregulation of several proapoptotic genes, which coincides with the onset of apoptosis in this model. Notably, there is a staged reduction in Diablo transcript levels across the first wave in hpg testes, with upregulation at Day 3 relative to WT and downregulation at Day 13. DIABLO is an antiapoptotic regulator that has been reported to be involved in germ cell apoptosis due to testosterone withdrawal [66]. Its regulation from Days 3–13 agrees well with the marked rise in germ cell apoptosis over the same time course.
Interestingly, despite the known involvement of BAX in germ cell apoptosis during the first wave of spermatogenesis [11, 12, 23, 67–71], we observed a downregulation of Bax transcription at all time points. Thus, in this model the testicular apoptosis does not appear to be BAX-dependent. BAX activity has, however, been described to be androgen dependent in prostate cancer [72]; therefore, this downregulation of Bax transcripts in hpg mice may be related to androgen deficiency. Intrinsic pathway apoptosis in hpg appears to be driven by other proapoptotic family members, with changes in transcript levels of Bcl2l11, Apaf1, Bcl2l10, Bid, and Bcl2l14. Of these genes, Bcl2l11 has been shown to be upregulated under flutamide exposure [66], implying this is a direct effect of the absence of testosterone in the hpg mouse. BCL2L11 has also been implicated in growth factor withdrawal apoptosis [73]. During the first wave of the WT spermatogenesis, BCL2L11 is involved in the elimination of supernumerary germ cells that do not receive trophic support from Sertoli cells [74]. The gene BAG1, which enhances the antiapoptotic effects of BCL2, represents a link between the growth factor receptors and antiapoptotic mechanisms [75]. BCL2L10 interacts with BCL2, BCL2L1, and BAX, both preventing the release of cytochrome c and promoting Bax multimerization.
Caspase 9 is the downstream caspase of the mitochondrial pathway. Its activation can be achieved either by cleavage or by association with a cofactor (APAF1) to form the active apoptosome [76]. We did not observe cleaved caspase 9 at any time point investigated by Western blotting, though we did observe upregulation of transcripts for both Casp9 and Apaf1 in our array data, indicating that the latter mechanism is likely to be operational. Our Western blot results again support this hypothesis, with an increase in Casp9 levels seen in hpg mouse testes compared to WT.
The extrinsic (receptor-mediated) pathway involves members of the death domain receptor family. When these receptors bind their cognate ligand, several adaptor molecules are recruited to form the intracellular death-inducing signaling complex (DISC). This in turn leads to the cleavage of caspase 3 and apoptosis. We observed transcriptional changes consistent with an upregulation of receptor-mediated apoptosis. Our Western blotting and immunochemical results confirm this activation of the extrinsic pathway, with an increase in total protein levels for caspases 8 and 3 and an increase in the activated (cleaved) forms of both caspases in hpg relative to WT testes. Our transcriptional data indicate that the extrinsic pathway genes deregulated in hpg testis are more likely to be expressed in the somatic compartment of the testis and thus are likely to relate to the observed Sertoli cell apoptosis.
This activation of the extrinsic pathway appears to be due to TNF alpha activity with upregulation of Tnfrsf1 and Tnf. TNF alpha has previously been shown to induce Fas expression, which in turn activates mitogen-activated protein kinases and NF-
β (Nuclear Factor kappa beta) pathways to trigger apoptosis in Sertoli cells [77–79]. Consistent with this, we also observed downregulation of antiapoptotic Birc4, a cofactor in the NF-β pathway [44]. Alteration in the NF-β signaling pathway has been reported in concordance with the loss of germ cell/Sertoli cell junctions [15], and testosterone withdrawal has been shown to induce detachment of germ cells [80]. We therefore suggest that the observed downregulation of Birc4 may be a response to the reduction of androgens in the hpg mouse testis, rather than the lack of gonadotropins per se.
There is no consensus in the literature on which pathways are involved in Sertoli cell apoptosis. The intrinsic pathway has been shown to be triggered in Sertoli cells by a range of insults, including x-irradiation [81, 82] and endocrine disruption using OP (4-tert-octylphenol) [83]. Sertoli cell apoptosis is also described in male mice deficient in Bcl-w, an antiapoptotic member of the BCL2 family, once again implicating the intrinsic pathway [84]. However, further groups have shown FAS-dependent apoptosis of Sertoli cells [22], indicating that the extrinsic pathway is also active in this cell type. It is probable that the pathway leading to Sertoli cell apoptosis is dependent on the nature of the damage that triggers the apoptosis. Our results indicate that in the case of gonadotropin withdrawal, it appears that the extrinsic pathway is active in Sertoli cells.
Previous studies have highlighted changes in the transcriptional pattern of apoptosis-related genes following hormonal change in the testis. However, none of the genes that have been highlighted by our work were described in three major microarray studies looking at the effect of testosterone in the testis. The study of the ABP (Androgen Binding Protein) mouse model that has a reduced level of intratesticular testosterone highlighted a few transcriptional changes in growth factor receptor expression, but not of molecules directly involved in apoptosis [85]. A microarray study of adult hpg mice injected with testosterone did not reveal any changes relevant to the apoptotic pathways between 4 and 24 h after injection [86]. Similarly, injection of testosterone into neonatal mice did not cause any changes in the apoptosis-related gene transcription. We suggest that this is due to the expression profiling analysis being performed too soon after testosterone injection (4–16 h). The AR, upon binding of testosterone, translocates to the nucleus and binds to AR target genes activating their transcription [87]. This suggests that the effect of androgen receptor on apoptotic gene transcription takes longer than 24 h, or that apoptosis-related genes are not the direct targets of this transcriptional activation.
Endocrine hormones such as testosterone and the gonadotropins (FSH and LH) have long been known to influence germ cell fate [88]. Their removal induces germ cell apoptosis in an indirect fashion, as their receptors are only located on somatic cells [89]. Although germ cell apoptosis is the major consequence of the lack of hormonal support, this is likely to arise from the failure of hormonal support at the level of the Sertoli cell and other somatic cells within the testis, because germ cells are not directly responsive to these hormones. The reduced number of Sertoli cells in the hpg male inevitably results in a decrease in overall germ cell numbers. Over and above this the lack of FSH stimulation is likely to reduce germ cell carrying capacity even more as demonstrated in FSHβKO (knockout of the beta subunit of FSH) males [90]. In the hpg mouse, germ cell numbers are decreased even further as a result of the additional lack of Leydig cell activation and testosterone production in the absence of LH stimulation. This absence of hormonal input to the somatic cells of the testis appears to allow upregulation of the proapoptotic pathways that act to regulate germ cell numbers, resulting in the very reduced Sertoli cell:germ cell ratios, and in addition appears to initiate apoptosis in the supporting somatic cells. This study has used a mouse model to provide molecular insight into the pathology arising from gonadotropin deficiency and could provide a better understanding of testicular function in hypogonadal patients. In summary, Sertoli cells and multiple stages of germ cells undergo apoptosis in hpg mice: germ cells use both intrinsic and extrinsic pathways, whereas Sertoli cells appear to use the extrinsic pathway.
ACKNOWLEDGMENTS
We would like to thank Dr. Carole Sargent, Dr. Robert A. Furlong, and Ms. Lydia Ferguson for discussions of the manuscript and data. We thank Kevin Neoh for the immunohistochemistry protocols; Dr. Mark Aarends for help with the identification of cell types; and Dr. Karyn Megy, Dr. Emily J. Clemente, and Julien Bauer for their help with the bioinformatics. We also thank Dr. Robert A. Furlong for curation of the annotation of the MmcDNAv1 library.
FOOTNOTES
1Supported by BBSRC 8/EGH16106 (to O.E.C., P.J.I.E., and N.A.A.). 
Correspondence: 2Nabeel Affara, Human Molecular Genetics Group, Department of Pathology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QP, U.K. FAX: +44 1223 333700; e-mail: na{at}mole.bio.cam.ac.uk
Received: 20 February 2007.
First decision: 24 March 2007.
Accepted: 17 July 2007.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  R. Sa, R. Neves, S. Fernandes, C. Alves, F. Carvalho, J. Silva, N. Cremades, I. Malheiro, A. Barros, and M. Sousa Cytological and Expression Studies and Quantitative Analysis of the Temporal and Stage-Specific Effects of Follicle-Stimulating Hormone and Testosterone During Cocultures of the Normal Human Seminiferous Epithelium Biol Reprod, November 1, 2008; 79(5): 962 - 975. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. M Ruwanpura, R. I McLachlan, P. G Stanton, K. L Loveland, and S. J Meachem Pathways involved in testicular germ cell apoptosis in immature rats after FSH suppression J. Endocrinol., April 1, 2008; 197(1): 35 - 43. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. M. Ruwanpura, R. I. McLachlan, P. G. Stanton, and S. J. Meachem Follicle-Stimulating Hormone Affects Spermatogonial Survival by Regulating the Intrinsic Apoptotic Pathway in Adult Rats Biol Reprod, April 1, 2008; 78(4): 705 - 713. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
