| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Human Nutrition Unit,3 University of Sheffield, Sheffield S5 7AU, United Kingdom
Academic Unit of Child Health,4 University of Sheffield, Sheffield S10 2TH, United Kingdom
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
conceptus, pregnancy
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
There is a known, strong association between the occurrence of PROM and the presence of Ehlers-Danlos syndrome, a heritable disorder that causes defects in the synthesis and structure of collagen [5]. In pregnancies not complicated by this syndrome, some groups have reported a significant reduction in PROM membrane collagen concentrations in comparison with controls [6, 7], but others have shown no difference [8, 9]. These inconsistencies may reflect differences in study design, particularly with respect to standardization of sites of tissue collection and, in some cases, studies have very small sample sizes. Amniotic membrane strength would also be expected to be a function of collagen cross-links, which are important in stabilizing the collagen molecule [10], but no study has examined the collagen cross-link profile of amnion tissue in the context of prelabor membrane rupture.
In 1964, a relationship between maternal vitamin C status and the occurrence of PROM was first proposed [11]. Ascorbic acid plays an essential role in the synthesis of collagen as well as its stabilization by cross-linking [10]. Studies in guinea pigs have shown a log-linear relationship between vitamin C intake and bone collagen indices as well as the pyridinoline:deoxypyridinoline ratio [12, 13]. Importantly, these authors showed a progressive change in indices of bone collagen status over a physiological range of vitamin C intakes [12]. Poor vitamin C status may contribute to impaired amniotic membrane integrity through effects on collagen synthesis as well as a reduction in antioxidant activity and an associated increase in free-radical damage [14].
We hypothesized that PROM membranes are characterized by reduced collagen concentrations and altered collagen cross-link profiles as well as increased concentration of biomarkers of oxidative damage. We also hypothesized that these effects are modulated by ascorbic acid status. A case-control study was carried out to test these hypotheses.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
A case-control study was carried out in women who had PROM deliveries and those with normal term deliveries. The concentrations of collagen and collagen cross-links, ascorbic acid, and two markers of oxidative damage were measured in amniotic membranes at multiple sites. A method was developed to measure the proteolytic resistance of amniotic membrane to determine how the above parameters might affect the functional properties of this tissue.
Sample Size
The power calculation for this study was based on the hypothesis that PROM membranes have impaired membrane integrity in comparison with controls. The published data with respect to differences in collagen concentrations in normal and PROM amnion are limited and inconsistent and the functional significance of this measurement is not well understood. Impaired membrane integrity affects the mechanical strength and stress tolerance of the membrane. A 50% difference in modulus of elasticity in amniotic membrane from PROM pregnancies and from normal pregnancies has been reported and this was used as the basis of our calculation of sample size [15]. It was calculated that a sample size of 37 women per group was sufficient to detect significant differences assuming a power of 0.85 and a threshold for significance of 5%. It is uncertain whether smaller differences would be functionally significant.
Recruitment and Sampling
Following approval from the North Sheffield Research Ethics Committee and informed written consent, amniotic membranes were collected from 74 women delivering at the Jessop Wing labor ward of the Royal Hallamshire Hospital, Sheffield. This group of women consisted of 37 PROM women and 37 controls whose membranes had ruptured spontaneously during labor at term. All of the women were greater than 37 wk gestation, did not have diabetes, preeclampsia, meconium-stained liquor, or any fetal abnormalities and did not smoke or knowingly have any infection, on the basis of vaginal swabs and body temperature.
The placenta was collected within half an hour of expulsion and the amnion and chorion manually separated. On average five 2.5-cm2 samples were collected from the entire delivery-tear (rupture site, RS), which would include the initial point of rupture, while four 2.5-cm2 samples were taken from the midzone (nonrupture site, NRS), an area of the membrane found half-way between the delivery tear and where the amnion starts to cover the placenta. In addition, umbilical venous blood and maternal venous blood were collected into lithium heparin tubes immediately after delivery. Plasma was isolated by centrifugation at 200 x g for 10 min at 4°C. Plasma and tissue samples for ascorbate analysis were stored in a 5% metaphosphoric acid solution (Sigma, Poole, Dorset, UK). All samples were stored at –80°C.
Fluorometric Assay of Vitamin C
The assay is based on the oxidation of ascorbic acid to dehydroascorbic acid by incubation with ascorbic acid oxidase (Sigma, Poole, Dorset, UK). The dehydroascorbic acid produced reacts with 1,2-diphenylenediamine (Sigma, Poole, Dorset, UK, coupling reagent) to produce a fluorescent compound, the production of which can be detected at an excitation of 350 nm and emission of 450 nm. The assay was run in the presence and absence of ascorbic acid oxidase so that endogenous dehydroascorbic acid could be distinguished from total ascorbic acid and the proportion of total ascorbic acid present in the oxidized form could be determined.
The frozen amnion samples were homogenized on ice for 3
min using an Ultra-Turraz homogenizer (Janke and Kunkel KG, Ilawerke, Staufen I Breigau) and then frozen for later analysis. On the day of assay, the homogenate and plasma samples were thawed and centrifuged at 9000 x g for 5 min at 4°C. The supernatants were analyzed using a Cobas Bio centrifugal autoanalyzer with fluorescent detection (Roche Products Ltd., Welwyn Garden City, UK) [16]. The within- and between-batch coefficients of variation were both less than 5%.
Biomarkers of Oxidative Stress
Frozen amnion samples were homogenized on ice for 30 sec in a 50 mM Tris-HCl (BDH Merck Ltd, Poole, Dorset, UK), pH 7.4, 10 mM CaCl2 (BDH Merck Ltd, Poole, Dorset, UK), 0.05% Brij-35 (Sigma, Poole, Dorset, UK), 0.20% Triton X-100 (BDH Merck Ltd, Poole, Dorset, UK) extraction buffer solution using a Ultra-Turraz homogenizer. The homogenates were centrifuged at 4°C for 10 min at 9000 x g and the supernatant removed and frozen at –80°C until analysis.
Malondialdehyde
Malondialdehyde (MDA) is a product of lipid peroxidation. The method employed for its determination is based on the production of a red adduct with thiobarbituric acid (Sigma, Poole, Dorset, UK), which can then be separated by HPLC and quantified by fluorescence detection [17]. Twenty microliters of sample supernatants were injected from an autosampler held at 4°C onto a Phenomonex Luna 5-µ C18 150 x 4.6-mm column (Phenomonex, Macclesfield, Cheshire, UK) with a Waters C18 guard column (Waters Ltd., UK). The mobile phase consisted of methanol and 10 mM phosphate buffer, pH 6.8, at a ratio of 40:60. The flow rate was 1.0 ml/min, and samples were eluted between 5 and 6 min. The fluorescence was monitored with the excitation at 525 nm and emission at 547 nm using a Jasco FP 920 detector (Jasco, Great Dunmow, Essex, UK). The within-batch coefficient of variation was 8.7%, while the between-batch coefficient of variation was 11.6%.
Protein Carbonyls
Protein carbonyls are markers of protein oxidation. The method employed measures protein carbonyls by coupling them with 10 mM 2,4-dinitriphenylhydrazine (Sigma, Poole, Dorset, UK) in 2 M HCl (BDH Merck Ltd, Poole, Dorset, UK) [18]. The proteins were separated from the rest of the plasma components by precipitation with 20% (w/v) trichloroacetic acid (BDH Merck Ltd, Poole, Dorset, UK) followed by centrifugation for 5 min at 200 x g. The protein precipitate was washed by resuspending in an ethanol and ethyl acetate mixture (1:1) followed by centrifugation for 5 min at 200 x g. This process was repeated a further two times. The final protein pellet was solubilized with 6 M guanidine hydrochloride (Sigma, Poole, Dorset, UK) in 20 mM potassium phosphate (Sigma, Poole, Dorset, UK) at pH 2.3 by incubating at 37°C for 30 min. The concentration of protein carbonyls can be quantified by spectrophotometric absorption at 360 nm using a COBAS bioautoanalyzer. Absorption at 285 nm allowed determination of the nonoxidized protein concentration of the sample and the concentration of protein carbonyls could then be expressed per unit of protein. The within-batch coefficient of variation was 12.6%, while the between-batch coefficient of variation was 6%
Hydroxyproline
Amnion samples were hydrolized for 24 h at 100°C in 6 M HCl, desiccated, and dissolved in distilled water. The samples were then analyzed for hydroxyproline using a microassay technique [19]. In this assay, the Ehrlich reaction is employed so that the amount of 4-hydroxyproline present in a sample can be determined spectrophotometrically. In the reaction, chloramine-T (BDH Merck Ltd, Poole, Dorset, UK) was used as an oxidizing reagent, oxidizing 4-hydroxyproline to pyrrole. The pyrrole produced then couples to the colored reagent paramethylaminobenzaldehyde, the production of which can be read at absorption wavelength of 562 nm. The within- and between-batch coefficient of variations were both less than 10%.
Pyridinoline and Deoxypyridinoline
Amnion samples were weighed and hydrolized at 100°C in 6 M HCl for 24 h. The cross-links were then isolated by washing the samples through CF1 columns (Biorad Laboratory Ltd., Bio-Rad House, Hemel Hempstead, Hertfordshire, UK) containing a 5% CF1 cellulose slurry (Whatman International Ltd., James Whatman Way, Springfield Mill, Maidstone, Kent ME14 2LE). The resultant solution was dried before being reconstituted in 1% heptafluorobutyric acid (Sigma, Poole, Dorset, UK, HFBA) and analyzed by HPLC according to the method of Colwell et al. [20]. The HPLC instrument consisted of a Spectra Physics P4000 quaternary pump and AS3000 autosampler (Spectra Physics, Maidenhead, UK) and a Jasco FP-920 fluorescence detector. Samples were separated by HPLC using a C18 column, 33 x 4.6 mm with 3 µm packing (Supelco UK Ltd.). The mobile phase consisted of solvent A (10 mmol/L HFBA, pH 2.25, adjusted with 10 M NaOH) and solvent B (75% acetonitrile, 25% solvent A) at a solvent ratio of 86% A:14% B. The flow rate was 0.65 ml/min, and samples were eluted between 10 and 12 min. The fluorescence was monitored with the excitation at 295 nm and emission at 395 nm. The concentration of each cross-link was determined using an external standard prepared in our laboratory [20]. Both the within- and the between-batch coefficient of variation were less than 5%.
Functional Amnion Membrane Degradation Assay
Three 15-mm-diameter biopsy punches were taken of each tissue sample. Then 0.5 ml of fresh (31.25 U/ml) bacterial collagenase type 4 solution (Worthington Biochemical Corporation, Lakewood) were added to each biopsy. Following 1, 2, or 3 h of incubation at 37°C in a shaking water bath, the samples were centrifuged at 9000 x g for 10 min at 4°C. The supernatant was removed and the volume measured. One milliliter and 800 µl of 6 M HCl were added to the pellet and 200 µl of the supernatant, respectively, before being hydrolized at 100°C for 24 h. Following evaporation of the HCl using a hot plate at 60°C, the samples were resuspended in 1 ml of distilled water and the hydroxyproline concentration determined [19].
Statistics
An unpaired t-test was used to examine whether there were any differences in group characteristics. Paired t-tests were used to examine whether there were any differences between maternal and umbilical venous plasma within the same patient type. Biochemical data were not normally distributed and so were rank transformed, whereupon effects of tissue site and patient group were tested for using two-way ANOVA. All pairwise multiple comparison procedures were carried out using the Tukey test or the Dunn test (if group numbers were unequal) to identify groups that differed. For data that were not normalized on rank transformation, the Kruskal-Wallis one-way ANOVA on ranks was used to determine, again, effects of tissue site or patient group, and the all pairwise multiple comparison Tukey test (nonparametric analogue) was used to identify the groups that differed. Significance of associations between variables was assessed using the Spearman rank-order test for correlation. For all of the statistical tests, significance was taken to be at the 5% level and was carried out using SigmaStat.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The groups (and subgroups) were compared for the degree of matching for possible confounders. No differences were identified for patient age ([mean ± SD] control: 27.6 ± 6.1 yr; PROM: 28.8 ± 5.6 yr), gestational age ([mean ± SD] control: 40.1 ± 1.5 wk; PROM: 39.9 ± 1.1 wk), labor length (control: median 505, range 130–1815; PROM: median 550, range 550–1957), baby weight ([mean ± SD] control: 3.4 kg ± 0.5; PROM: 3.3 kg ± 0.5), or route of delivery (control: 31/37 vaginal; PROM: 30/37 vaginal).
Ascorbic Acid Concentrations
Umbilical venous blood samples were collected from all but three patients, one control, two PROMs. Maternal blood samples could not be collected from 11 control and 3 PROM patients. Umbilical venous plasma ascorbic and dehydroascorbic acid concentrations were significantly higher than maternal plasma for both the PROM and control groups (P < 0.05). The percentage of total ascorbic acid present as dehydroascorbic acid was significantly higher in maternal plasma for both the PROM and control groups than in umbilical venous plasma (P < 0.05; Fig. 1). Ascorbic and dehydroascorbic acid concentrations in maternal and umbilical venous plasma were similar in PROM and control women (Fig. 1).
|
Total ascorbic acid concentration was significantly lower (P = 0.013) in amniotic membrane from PROM patients than controls (PROM: median = 0.0265 µmol/g tissue, range 0.008–0.129; control median = 0.034 µmol/g tissue, range 0.009–0.135). Ascorbic acid concentration was not different between rupture site and nonrupture site (PROM RS median = 0.032 µmol/g tissue, range 0.01–0.086; NRS median = 0.019 µmol/g tissue, range 0.008–0.129; control RS median = 0.038 µmol/g tissue, range 0.013–0.135; NRS median = 0.032 µmol/g tissue, range 0.009–0.099).
Markers of Oxidative Stress
While the primary objective of the study was to compare aspects of collagen synthesis and membrane integrity between patient groups, the opportunity arose to determine differences in measures of oxidative damage. However, this was only possible in patients for whom we had sufficient material. MDA measurements were made on membrane samples from 25 PROM and 26 control patients. Protein carbonyl measurements were made on 16 PROM and 15 control patients.
Malondialdehyde
MDA concentrations were significantly higher in the rupture-site region of the control patients than in nonrupture-site samples (RS median = 4.01 nmol/g tissue, range = 1.13–10.78; NRS median = 2.515 nmol/g tissue, range = 0.88–8.46) (P < 0.05; Fig. 2). There was no significant difference in MDA concentrations between the two tissue sites for the PROM patients (RS median = 4.28 nmol/g tissue, range = 0.96–14.31; NRS 4.025 nmol/g tissue, range = 1.34–16.1; Fig. 2). MDA concentrations were significantly higher in PROM nonrupture-site samples than control nonrupture-site samples (P < 0.05; Fig. 2) but no significant difference existed between the rupture-site samples (Fig. 2).
|
Protein Carbonyls
Protein carbonyl concentrations did not differ between rupture-site and nonrupture-site tissue within the same patient type (PROM RS median = 2.92 nmol/mg protein, range 0.84–4.84; NRS median = 3.135 nmol/mg protein, range 1.94–6.48; control RS median = 1.7 nmol/mg protein, range 0.49–3.59; NRS median = 2.33 nmol/mg protein, range 1.34–3.45; Fig. 3). There was, however, a significantly higher protein carbonyl concentration in PROM samples compared with controls at both tissue sites (RS: P = 0.005; NRS: P = 0.011; Fig. 3).
|
Hydroxyproline Determination
Collagen concentrations did not differ between rupture and nonrupture-site samples within the same patient type (PROM RS median = 3.92 µmol/g tissue, range 1.37– 13.18; NRS median = 4.17 µmol/g tissue, range 1.86– 11.57; control RS median = 4.49 µmol/g tissue, range 1.91–9.31; NRS median = 4.91 µmol/g tissue, range 1.95– 9.91). There was, however, a significantly lower (P = 0.027) collagen concentration in PROM amnion than control tissue (PROM median = 4.1 µmol/g tissue, range 1.37– 13.18; control median = 4.64 µmol/g tissue, range 1.91– 9.91). No significant correlation was observed between PROM latency and the collagen concentration of the amnion samples from PROM patients.
Pyridinoline and Deoxypyridinoline
Measurements were made on all of the tissue samples for 15 randomly selected PROM and 10 randomly selected control patients. Pyridinoline and deoxypyridinoline cross-link concentrations were similar in PROM and control samples regardless of the sampling area (Fig. 4). Pyridinoline concentrations were not significantly different between the two sampling regions within each patient type (PROM RS median = 8.02 nmol/µmol Hpr, range 5.09–17.32; NRS median = 7.9 nmol/µmol Hpr, range 4.96–10.96; control RS median = 8.3 nmol/µmol Hpr, range 5.8–11.4; NRS median = 8.15 nmol/µmol Hpr, range 5.16–10.83; Fig. 4). However, rupture-site deoxypyridinoline levels were about twofold higher in rupture-site samples than in nonrupture-site samples, and this difference was highly significant (PROM RS median = 0.9 nmol/µmol Hpr, range 0.28– 2.01; NRS median = 0.26 nmol/µmol Hpr, range 0.14– 0.55; control RS median = 1.02 nmol/µmol Hpr, range 0.39–1.57; NRS median = 0.31 nmol/µmol Hpr, range 0.14–0.53) (P < 0.001; Fig. 4). This significantly altered the rupture-site cross-link ratio in favor of deoxypyridinoline (P < 0.001). No significant correlation was observed between cross-link ratio and amnion total ascorbic acid concentration.
|
Proteolytic Resistance
Data for the proteolytic resistance of amnion expressed as the percentage hydroxyproline released into the supernatant after 1, 2, and 3 h of incubation are shown in Figure 5. There was a consistent difference in the amnion proteolytic susceptibility when comparing PROM and control samples, which reached statistical significance at 2 h and was maintained at 3 h. Following 2 h (PROM RS vs. control RS P = 0.045; PROM NRS vs. control NRS P = 0.004) and 3 h of incubation (PROM RS vs. control RS P = 0.016; PROM NRS vs. control NRS P = 0.002) percentage hydroxyproline release was greater in PROM tissue compared with control tissue for both rupture and nonrupture-site samples compared separately. Within each patient type at each time point, there was a higher percentage of total hydroxyproline released into the supernatant from rupture-site samples compared with nonrupture-site samples, but this was only statistically significant for the control group (P < 0.05).
|
The proteolytic susceptibility of the amnion showed a strong negative association with the hydroxyproline content of the amnion at 1 h (r = –0.319; P < 0.001) and 2 h (r = –0.227; P = 0.006) of incubation but not after 3 h. Tissue deoxypyridinoline concentrations were positively correlated with percentage total hydroxyproline released after 1 h (P = 0.012) and 2 h of incubation (P = 0.023) but not after 3 h of incubation. Pyridinoline concentration showed no significant association with hydroxyproline release. Incubation for 24 h with 0 U/ml and 31.25 U/ml bacterial collagenase resulted in 9.5% and 92% of total hydroxyproline release into the supernatant, respectively.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Membranes that rupture prelabor or otherwise have within them a zone of extreme altered morphology [23], a region that is characterized by dissociation of collagens [4]. This region can be found within the postdelivery tear and is believed to act as the rupture-initiation site [4]. We have reported no regional variation in collagen or ascorbate concentration within the amnion, suggesting that this weak point is not associated with reduced collagen or ascorbic acid levels. Instead, it would appear that a regional variation in collagen cross-link ratio in favor of deoxypyridinoline is responsible for the development of a rupture-initiation site. Observed changes in collagen cross-links are consistent with reduced collagen stability. A shift in cross-link ratio in favor of deoxypyridinoline is seen in patients with Ehlers-Danlos syndrome type VI and is thought to be responsible for the connective tissue hyperelasticity seen in this condition [24]. A similar shift in the ratio of these cross-links has been reported in Ullrich-Turner syndrome, another connective-tissue disease [25]. This study is the first to examine amnion collagen cross-link concentration. We report that, despite a possible involvement in the formation of the rupture-initiation site, collagen cross-linking is not involved in the etiology of PROM.
It has been proposed that variations in collagen structure or content will affect the properties of the membrane [9]. To explore this, we developed a functional assay to evaluate the proteolytic resistance of the amnion to bacterial collagenase. This does not specifically reflect susceptibility to degradation by mammalian MMPs. However, it does provide a measure of the susceptibility of the tissue to collagenolysis and therefore the stability of the tissue. Results showed very clearly that prelabor ruptured membranes are more susceptible to proteolytic degradation than those that undergo timely membrane rupture. In addition, within the same amniotic sac, there was variation in proteolytic resistance, with the rupture-site region of the amnion being less resistant.
The amnion contains a wide variety of different proteases. These are important in the normal process of extracellular matrix remodeling that occurs within the amniotic sac as pregnancy progresses. However, the difference in proteolytic susceptibility means that the potential for degradation may be greater in PROM membranes. The proteolytic susceptibility of the amnion was strongly negatively associated with collagen content. Maintaining an adequate collagen concentration in the membrane would be one means by which to delay membrane rupture, by increasing both the physical strength of the membrane as well as its proteolytic resistance.
Ascorbic acid is essential for the production and maintenance of collagen. It modulates collagen mRNA transcription [26], plays an essential role both in the formation of the collagen triple helix and in its stabilization [27], and is involved in the formation of strengthening collagen cross-links [9]. Inadequate vitamin C status during pregnancy might interfere with the development of the extracellular matrix of the amniotic membrane. It has been reported that ascorbic acid can downregulate MMP2 mRNA expression in WISH cells in culture [28]. However this observation was not substantiated by proteomic studies and needs to be confirmed in amnion tissue. We report a significantly lower concentration of ascorbic acid in amniotic tissue from PROM women compared with controls. However, despite the well-established role that vitamin C plays in collagen formation, the concentration of collagen in amniotic membrane showed no significant association with amniotic-membrane ascorbate concentration. It is possible that the process of labor and its associated variable mechanical and oxidative stresses may have confounded any prelabor relationship. Indeed, an 80% decline in plasma ascorbic acid with labor has been reported [29]. The failure to detect a difference in ascorbic acid concentrations in the maternal circulation between PROM and control women suggests that there may be differences in the utilization of ascorbic acid between these two groups, possibly mediated by differences in glucose transporter-receptor expression. Alternatively, our inability to detect a difference in plasma ascorbic acid concentration may be reflective of the limitations of using plasma ascorbic acid as a measure of status in nonfasted individuals.
Ascorbic acid also has a well-established antioxidant function through its free-radical scavenging activity. We have shown, for the first time, an increase in the concentration of markers of oxidative damage in tissue from PROM deliveries compared with controls. Specifically, protein carbonyls were increased in PROM tissue in comparison with controls. MDA concentrations were significantly elevated in PROM non–rupture-site tissue in comparison with the equivalent tissue from control patients. Differences were not observed at the rupture site, but this may reflect a higher baseline of oxidative stress at the rupture site and a proportionally smaller increment in lipid peroxidation at this site in association with PROM delivery.
To clarify an area of significant uncertainty in amnion biology, we have conducted a detailed systematic examination of fetal membranes collected from PROM and normal deliveries. We conclude that prelabor rupture of fetal membranes is a function of enhanced degradation of membrane collagen and net collagen loss.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Correspondence: Hilary Powers, Human Nutrition Unit, University of Sheffield, Coleridge House, Northern General Hospital, Herries Road, Sheffield S5 7AU, UK. FAX: 44 011 4261 0112; h.j.powers{at}sheffield.ac.uk
Received: 25 June 2004.
First decision: 20 July 2004.
Accepted: 27 August 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
