Endocrine disrupters : exploring novel endpoints, exposure, low-dose-and mixture-effects in humans, aquatic wildlife and laboratory animals

Cefas (partner 19) and partner 1 formed the hypothesis that oestrogens may act as anti-androgens. We then went and tested this hypothesis using single compounds and their mixtures and were successful in proving the concept. Partners 2 and 8 conducted 10 relevant exposures and partner 19 provided analysis for spiggin (the androgen and anti-androgen end-point) in stickleback kidneys. The specific tests for which we provided sample analysis were: - Four range finding exposures with oestradiol, oestrone, ethinyl-oestradiol and nonylphenol together with the androgen DHT on spiggin suppression; - Four definitive tests with oestradiol, oestrone, ethinyl-oestradiol and nonylphenol together with the androgen DHT on spiggin suppression; - Two mixtures studies on the effect of four oestrogens (E1, E2, EE2, NP) on spiggin suppression when administered together with the androgen DHT.

New endpoints (biomarkers) of DBP/phthalate action on the fetal rat testis have been identified which manifest at different fetal ages and which affect one or more of the main cell types of the fetal testis. - Germ cell differentiation. Exposure to high (500mg/kg/day) DBP doses results in delayed differentiation of fetal germ cells (GC), s manifest as delayed entry into quiescence prolongation of GC proliferation, expression of cell cycle markers and delay in switching off of the pluripotency marker OCT4. By birth there is ~40% reduction in fetal GC numbers. The delay in GC differentiation is the earliest identified effect of DBP on the fetal testis (e15.5-e17.5). These DBP effects are of particular interest because in human TDS, the CIS cells from which testicular germ cell cancers arise in adulthood are themselves thought to arise because of failure of normal fetal GC differentiation. However, in DBP-exposed rats we have not yet shown failure of fetal GC differentiation only a transient delay. Postnatally (after cessation of DBP exposure), there is also a further delay in GC differentiation with delay in their migration to the basal lamina and resumption of proliferation. Consequently there is an even bigger deficit in GC numbers (~80% reduction compared with controls) by mid-puberty, though this is corrected by adulthood in normally descended (scrotal) testes. DBP-exposed animals exhibit a high rate (~75%) of infertility in adulthood, so the functional integrity of the adult GC may be compromised as a result of their impaired differentiation in fetal life. This will be important to investigate in future studies. - Multinucleated gonocytes (MNG). Exposure to DBP in utero results in dose-dependent induction of MNG throughout the fetal testis. In contrast to the time-window within which delayed GC differentiation occurs (see 1 above), MNG only become apparent at ~e19.5 with a peak incidence at e21.5 that persists postnatally until days 4-6 but not thereafter. Exposure to DBP just from e19.5 induces MNG equally as effectively at e21.5 as does exposure from e13.5, showing that MNG formation is unrelated to delayed GC differentiation. MNG thus occur only during the time window when fetal GC are quiescent and therefore has no relationship to proliferation. The mechanisms underlying MNG formation are unknown as are its consequences. It appears that MNG induction is a valid indicator of in utero phthalate exposure and is one of the most sensitive biomarkers of phthalate action as effects are evident at 20mg/kg/day. - Fetal Leydig cell aggregation. Exposure to DBP in utero results in dose-dependent induction of Leydig cell (LC) aggregation in central regions of the testis. The number of LC is unaffected, though they are reduced notably in size (consistent with their reduced steroidogenic function). The aggregation is thought to arise because of migration of LC from more peripheral regions of the testis, but the trigger for this is unknown. LC aggregation, which can be quantified objectively, is not evident until e17.5 and becomes progressively more pronounced up to e21.5. LC aggregation is induced by doses of 100 and 500 mg/kg/day with more minor effects at 20 mg/kg/day. The most important consequence of DBP-induced LC aggregation is that it interferes with the final stages of seminiferous cord formation in central regions, and this results in formation of focal dysgenetic areas in which the cords are malformed and which contain intratubular LC (i.e. LC that are in the wrong place). These focal dysgenetic areas persist throughout postnatal life and lack germ cells in adulthood. Similar focal dysgenetic areas are evident in testes of men with testicular germ cell cancer - Reduction in Sertoli cell number. Exposure to DBP in utero results in dose-dependent reduction in Sertoli cell (SC) number at e21.5. At the highest dose tested (500 mg/kg/day DBP) SC number is reduced by ~45%. This decrease is probably secondary to the reduction in testosterone levels. The importance of the DBP-induced reduction in SC number is that, if this persists, it will result in reduced capacity to make sperm in adulthood which, in humans, would lead to a reduction in sperm counts as is found in TDS. It remains to be shown whether the DBP-induced reduction in SC number in rats persists through to adulthood.

Elements of an EDC mixtures risk assessment approach The knowledge about determinants and factors that govern the joint action of similarly acting EDC is sufficiently advanced to come to pragmatic risk assessment approaches. In the light of these considerations, a panel meeting with experts from EU environmental agencies and NGO's held in conjunction with WP16 has proposed the following modus operandi for EDC mixtures which takes account of data gaps and limited knowledge that may exist in relation to exposure scenarios: - In a particularly 'data-rich' situation, both exposure information and low dose estimates are known. In such cases, it is proposed to use the concept of dose addition (including the TEF approach) to arrive at a 'mixture no-observed-effect-level' (MNOEL) for endpoints relevant to endocrine disruption which then should be combined with a safety factor to arrive at estimates of tolerable human exposure. - In what may be described as a 'data-poor' situation, there may be fragmentary information about the number of chemicals with endocrine activities, and their exposure levels. There may also be insufficient information about their potency, and their effects at low doses, except for a few 'prototype' chemicals. In such cases, it is proposed to estimate a crude MNOEL by dividing the individual NOEL of the 'prototypical' chemical by the anticipated number of relevant similarly acting chemicals. This will be based on chemicals where information about relevant in vivo effects is documented. If this information is not available, then additional chemicals should be 'ruled in' on the basis of in vitro test results. If the number of chemicals contributing to mixture effects is unknown, it is suggested to choose a default number in the region of 10-20 until evidence to the contrary emerges. The resulting MNOEL is then combined with the usual assessment factor (e.g. 100) to estimate tolerable levels for human exposure.

Between 1996 and 1999, semen quality of young men from five European countries (889 from Denmark, 313 from Finland, 190 from Estonia, 157 from Lithuania, 221 from Norway) who were undergoing compulsory medical examination for possible military service were studied. Each man provided a semen sample, was examined by a physician, and, in collaboration with his mother, completed a questionnaire about general and reproductive health, current smoking habits, and exposure to smoking in utero. After adjustment for confounding factors, men exposed to smoking in utero had a reduction in sperm concentration of 20.1% and a reduction in total sperm count of 24.5% in comparison with unexposed men. Exposed men had a 1.15-ml (95% CI: 0.66, 1.64) smaller testis size than unexposed men. The associations were present also when data from the study centres were analysed separately except for Lithuania, where only 1% of mothers smoked during pregnancy., These findings strongly supported the hypothesis that fetal development is particularly vulnerable and important for adult reproductive health. Thus, maternal smoking may have long-term implications for the reproductive health of the offspring and this further justifies the advise for pregnant women to avoid smoking. In 2001-2005 a total of 1,925 young Danish men delivered a semen sample, had a physical examination performed and a blood sample drawn, and responded to a questionnaire. Their mothers were questioned about whether they had received fertility treatment in order to conceive their sons. Forty-seven mothers reported having received fertility treatment to conceive the index subject. After control for confounders, men whose mothers had received fertility treatment to conceive them had a 46% lower sperm concentration (95% confidence interval (CI): -63, -20) and a 45% lower total sperm count (95% CI: -64, -16). They had a smaller testis size (-0.9 ml, 95% CI: -2.2, 0.4), fewer motile sperm (-4.0%, 95% CI: -8.0, -0.1), and fewer morphologically normal spermatozoa (-2.0%, 95% CI: -4.1, 0.0). They also had a lower serum testosterone level and free androgen index (results not statistically significant). These findings should be viewed in light of the increasing use of fertility treatments. Although the cause of these findings is unknown, they raise concern about possible late effects of fertility treatment.

A suite of advanced tools for diagnosing and assessing endocrine disruption in fish and mammals has been successfully developed. These tools include antibodies and real-time RT-PCR protocols for quantifying EDC-inducible molecular responses (e.g., aromatase, bvitellogenin), microarray technologies to assess EDC-responsive genes in fish and mammalian tissues, and new cell-based reporter assays for detecting mixture effects of EDCs. In addition, novel targets of EDC action have been identified. Particular emphasis has been given to the signalling pathways mediating EDC effects, and to the steroidogenesis/ aromatase system. The investigations on signalling pathways have clearly shown that the genomic effects (gene regulation and mitogenicity) effects of endocrine disrupters can be regulated not only by the classical receptor-binding activity, but also by interfering with rapid cell signalling pathways that ultimately lead to the ligand-independent ER activation. This highlights the need to take such short-term effects into account when evaluating the full estrogenic potency of chemicals. With respect to aromatase, it was shown that the brain isofom of aromatase is responsive to estrogens, while the gonadal isoform is responsive to dioxin-like compounds. Environmental inhibition of the aromatase system can lead to profound changes of sexual development and reproductive functions, however, it can also affect functions outsided the reproductive system such as neurogenesis. With these results, EDEN provides researchers and testing laboratories with a suite of novel protocols for diagnosis of EDC exposure and effects, as well as information on the scopes and limits of these tools. Further, the results from this work have important implications for risk assessment of EDCs since they point to new mechanisms of action and new targets at risk by EDCs.

Four assays have been developed and validated: - a HPLC-hyperfractionation method to assess the total estrogenic xenobiotic burden (TEXB) in human tissue using the E-Screen. - the yeast estrogen screen YES, successfully used for bioassay-directed fractionation (the corresponding androgen screen YAS has been established but proofed to be less rugged). - the optimized hEST assay for measurement of fractionated fish bile. - the HRS coumestrol ERalpha screening assay for fish bile and human breast tissue samples.

Extensive low dose studies were carried out in assays sensitive to endocrine disrupters. None of the chemicals investigated in exhibited dose-response curves with non-monotonic curvatures (inverted-U) in the low dose range. In some instances, a down-turn of responses from doses associated with a plateau was observed which could be attributed to cytotoxicity (e.g. in the E-Screen assay). Concerns relating to the danger of overlooking effects at low doses could not be confirmed. Instead, detailed low dose response analyses revealed that many of the tested agents exhibited quite shallow curves in the low effect range, and this resulted in low dose estimates with often surprisingly small numerical values. Dose response curves with small gradients give rise to complications during the estimation of low effect doses. High statistical power is necessary to arrive at valid estimates. However, such resources are usually not available for in vivo testing, with the consequence that the statistical detection limit is often considerably higher. If this coincides with shallow dose response curves in the low effect range, the testing with only a few dose groups might fail to identify effects of a small size. No-observed-effect-levels were associated with effects of up to 10% of a mbioligcally maximal effect. Data showed very clearly, that NOEL cannot be equated with zero effect levels. For each assay, detailed recommendations for low dose studies were made, in terms of number of positive and negative controls, spacing and number of dose groups for regression analyses, and dose group sizes for studies based on hypothesis testing. Because a wide variety of different dose response curves occurred, no uniform dose-response regression model could be found that was suitable for EDC in all cases. This means that the best fitting model should be chosen from a pool of regression models. With the recognition that doses associated with zero effects cannot be determined empirically, the aim of low dose EDC testing can only be to derive estimates of doses that correspond to a specific effect magnitude. Thus, the starting point of low dose testing strategies should be a decision about the effect size a low dose experiment should be able to demonstrate. Only after such a choice has been made, can the strengths of hypothesis testing procedures (the ability to test certain doses with a large number of replicates) and those of regression-based approaches (the ability to assess effect trends) be exploited productively for low dose testing. Key elements of an approach combining the strengths of hypothesis testing and regression methods were outlined. The proposed integrated procedure aims to (1) identify the minimum effective dose that is statistically significant and produces an effect that it is at least of the relevant effect size and (2) if reliable, to estimate the corresponding dose for this effect size (benchmark). Firstly, a power analysis is performed with the aim of assessing whether the suggested experimental design is of sufficient sensitivity to demonstrate effect sizes of relevance. This can be achieved by comparing the statistical detection limit that is achievable with the chosen experimental design with the magnitude of the effect of relevance. The application of hypothesis testing methods is inadequate for capturing low dose effects of EDC. Whenever possible, regression-based approaches with their benchmark dose limits should replace NOEL as the basis for establishing acceptable human exposure levels. Ideally, hypothesis testing and regression methods should be combined in a framework that utilises the strengths of both methodologies by making considerations of statistical detection limits and statistical power the starting point of testing procedures. Implementation of this framework will require a significant change in toxicological testing practice.

Resulting from the gene expression experiments carried out using DNA-microarray techniques and RT-PCR we recommend to use the estrogen response element ERE to construct transgenic zebrafish. This is due to the strong regulation of vitellogenin by the ERE. We also suggest thinking of using a reporter gene under the control of 3 to 4 EREs. This may strongly enhance the signal. In addition the down regulation of aromatase renders aromatase an interesting target gene. This recommendation is corroborated by the results of the experiments on the functional significance of altered aromatase gene expression.However it should be noted that in fish there are two distinct genes for aromatase (gonadic and brain form). Unlike the gonadic form, the expression of brain aromatase gene is significantly stimulated by estrogen. The promoter region of this gene (also containing an ERE) could therefore also be a good candidate for transgenic zebrafish.

These studies have involved treatment of time-mated pregnant rats with daily doses of dibutyl phthalate (DBP) varying from 4 to 500 mg/kg/day or with vehicle (= controls) with the treatment period ranging from e13.5 to e21.5. Exposed male offspring have then been sampled at fetal ages ranging from e15.5-e21.5 and Leydig cell (LC) immunoexpression of Insl3 evaluated and compared with LC size (= cytoplasmic volume) and the expression of factors involved in different steps of steroidogenesis (SR-B1, P450ssc) or with other roles in LC (Inhibin-a, CRABP-II); at certain ages (e19.5, e21.5) intratestcular levels of testosterone have also been measured. Most attention has been focused on males exposed in utero to 500 mg/kg/day DBP because this is the only dose level that causes a high incidence (>70%) of (mainly unilateral) cryptorchidism and is the only dose to consistently reduce immunoexpression of Insl3 at certain foetal ages. In addition, we have undertaken similar evaluations in fetal males from pregnant dams that have been treated with either 100 mg/kg/day flutamide from e15.5 until e21.5. These studies have shown that Insl3 immunoexpression in fetal LC appears to change more or less hand in hand with steroidogenic capacity and with the expression of steroidogenic enzymes that are highly regulated, such as P450ssc. Thus, in controls as testosterone levels and P450ssc immunoexpression increase from e15.5 to a peak at e19.5, so does the intensity of immunoexpression of Insl3. Furthermore, exposure to 500 mg/kg/day DBP results in a similar decrease in immunoexpression of Insl3 as in P450ssc, a change that also correlates with reduction in LC size. In contrast, exposure in utero to flutamide does not affect immunoexpression of either Insl3 or P450ssc. However, it remains uncertain whether immunoexpression of Insl3 provides an accurate reflection of the level of secreted hormone, although for P450ssc there is clear evidence that reductions in immunoexpression (such as after treatment with DBP) correlate closely with reduction in testosterone production. With such fundamental reservations, and in the absence of data to correlate LC immunoexpression of Insl3 with levels of secreted Insl3, it was decided that further detailed evaluation of LC Insl3 immunoexpression in relation to other factors, would be pointless.

Expression of the hAR ligand binding domain (LBD) in E. coli has been achieved but the levels were too low to setup a HRS AR system. Many efforts have been made to obtain higher expression levels. To determine the levels of hAR LBD, specific receptor binding was estimated with [1,2,6,7-3H]-testosterone. The synthesis of a fluorescent ligand for the on-line HRS-hAR bioassay had to be postponed until high enough LBD-expression levels were reached.

An exposure study involving exposing transgenic ZER-CALUX® zebrafish to a potent estrogen (EE2) was carried out by Partner 7 to determine the usefulness of transgenic reporter fish in detecting endocrine activities of chemicals and polluted environments. As expected, the fish synthesized vitellogenin in response to estrogen exposure (EE2). Similar results would have been obtained if normal (not transgenic) fish had been used. All 3 concentrations of EE2 (up to 10ng/litre) produced significant responses, albeit only at the highest concentration was the response very pronounced. Also, as expected, luciferase was induced in the liver (the transgenic fish produce luciferase in response to exposure to estrogen). Given that vitellogenin was synthesised in the liver in response to the EE2, it would have been extremely surprising if luciferase had not been induced in that organ. Luciferase induction was also investigated in the brain and Testes. The results demonstrate that in the brain there was no luciferase induction above the control level in the fish exposed to EE2. This might imply that the brain is unresponsive to EE2, or responsive but less sensitive than the liver. Another explanation is that the EE2 could not reach the brain, possibly due to the existence of the blood:brain barrier. In complete contrast, exposure to the highest concentration of EE2 caused a major induction of luciferase in the testes. This can mean only that functional ERs exist in the testes. This tissue therefore becomes a prime candidate for effects caused by environmental estrogens. A more complete investigation of many different tissues may demonstrate that other tissues/organs are also estrogen-responsive. Partner 11 hs been working on the expression of the genes encoding aromatase (namely zfcyp19a and zfcyp19b) in zebrafish exposed to endocrine disrupting chemicals. They showed, in collaboration with Partners 14 and 15, that exposure to EDCs did not affect zfcyp19a expression, either in cells or whole fish embryos. In contrast, expression of the other aromatase gene, zfcyp19b, was strongly upregulated by estradiol, an effect that was down-regulated by co-exposure to a dioxin (TCDD). Their research, demonstrated the usefulness of in vivo zebrafish larvae, and also in vitro aromatase reporter gene assays, for evaluation of the actions of estrogenic chemicals. They also very nicely demonstrated the cross-talk between estrogenic and dioxin chemicals on the expression of the two aromatase genes in the zebrafish. Finally, the demonstration that the presence of a dioxin can reduce (possibly even eliminate) the estrogenic effects of a strong estrogen illustrates the very complex nature of how an animal responds to a mixture of chemicals, especially mixtures containing chemicals with different mechanisms of action. Zebrafish were used by partner 13 to examine estrogenic effects on gene expression, using custom-made arrays. Fish were exposed for 11 days to a range of concentrations of E2, BPA, genistein and EE2. The flow-through systems used allowed stable concentrations of the test chemicals to be maintained. As expected, the microarray data confirmed that all estrogenic chemicals up-regulated the expression of the vitellogenin gene. No other genes showed such a strong up-regulation, underlining the use of vitellogenin as a superb biomarker of exposure to estrogenic chemicals. There were many similarities in the responses to the different estrogenic chemicals. For example, E2 and genistein both up-regulated the same 45 genes. This suggests that these two natural estrogens are using the same or similar pathways to affect estrogenic responses. In other cases, however, all the tested chemicals regulated the expression of the same gene, but in different ways. And there were a few examples where only one (or two) of the chemicals affected a particular gene. The results demonstrate the complexity of the genetic response of fish to estrogenic chemicals. Interpreting this response is a difficult task. It is currently unclear whether some of the 'new' estrogen-responsive genes identified might serve as useful biomarkers, indicating exposure to perhaps an individual chemical, or perhaps a group of chemicals with similar modes of action. Important issues that could be successfully addressed using these transgenic fish: - What are the relative in vivo potencies of different estrogenic chemicals? - Can weakly estrogenic chemicals that bioconcentrate reach levels high enough to induce effects? - Which stages of the life-history of the fish are particularly sensitive to estrogenic chemicals? - Are some tissues/organs responsive to estrogens at certain stages of development, but not at others? - Can the fish be used to determine whether environments (e.g. rivers, or lakes) are estrogenic?

Mouse tumour Leydig cells (mLTC-1) were transfected with a reporter construct containing the Insl3 promoter in front of the luciferase reporter gene. This promoter possesses three SF-1 binding sites which have been shown to be important for regulation of the mouse Insl3 gene. Transfected cells were used to screen for the effects of oestradiol, testosterone and hCG on Insl3 gene expression in vitro. 1 million mLTC-1 cells were seeded in 6-well plate wells in Waymouth's medium containing 9% iHSs and 4.5% iFCS 24h prior to transfections. On the following day the transfections were performed in triplicate using 0.5µg, 1.0µg or 1.5µg of reporter construct, 0.25µg or 0.05µg mERa/empty construct and 6µl FuGENE6 per well. The amount of DNA/well was always set equal in every experiment by using empty pIRES-EGFP vector as a compensator. The day after the medium was replaced with fresh Waymouth's medium containing 10% DCC iFCS, 'stripped' serum and different concentrations of oestradiol, testosterone or hCG in ethanol or ethanol only as control. Then, 24h after hormone stimulation the cells were lysed. The lysates were centrifuged and luciferase activity was measured from the supernatants. Tranfection efficiency was measured by co-transfecting the cells with Promega's phRL-SV40 vector that encodes Renilla luciferase. The Insl3 reporter gene assay showed that there were no direct effects by estrogens, whereas androgens slightly stimulated gene activity. It is therefore concluded that this Insl3 reporter gene assay cannot be used to screen effectively for chemicals likely to have an impact on Insl3 gene expression.

A website was developed to inform the public and scientific community about the EDEN project and its progress. The site introduces the research area and the projects aims and research themes. The 'Achievements' section gives the publications arising form the EDEN project which number >100. The 'what's new' section is continually updated including a section on the Prague Declaration on Endocrine Disruption which arose form the Workshop held in Prague 2005 and there is also a separate header with the complete document. As well as the summaries for each of the four yearly reports, a poster presented to the GRC on the activities arising from EDEN is available on the 'what's new' section of the website. All eight issues of the CREDO Cluster newsletter are also available as pdfs on the main site, along with a synopsis of each issue. A separate part of the website is only accessible to EDEN members to aid communication.

Studies have involved treatment of time-mated pregnant rats with daily doses of dibutyl phthalate (DBP) varying from 4 to 500 mg/kg/day with the treatment period ranging from e13.5 to e21.5 (standard regimen used for most studies) or being restricted to a shorter window (e19.5-e21.5). Controls were treated with the vehicle alone. Male offspring have then been sampled at fetal ages ranging from e15.5-e21.5 or postnatally at ages ranging from 4-90+ days (adulthood). At each age LC expression of Insl3 has been evaluated using immunohistochemistry. Most attention has been focused on males exposed in utero to 500 mg/kg/day DBP because this is the only dose level that causes a high incidence (>70%) of (mainly unilateral) cryptorchidism. In addition, we have undertaken similar evaluations in fetal males from pregnant dams that have been treated with either 100 or 50 mg/kg/day flutamide from e15.5 until e21.5 or with 25 mg/kg/day flutamide or with 0.1 mg diethylstilboestrol (DES; days e13.5, 15.5 and 17.5) and then evaluated on e17.5 or e19.5. It has been shown that Insl3 immunoexpression in fetal LC is highly expressed in control males at e17.5 and e19.5 (during the period of trans-abdominal testis descent) and is notably lower at e21.5. In animals exposed in utero to 500 mg/kg DBP from e13.5, Insl3 immunoexpression was suppressed in most animals at e17.5-e19.5 but was not different from controls at e21.5. However, when Insl3 expression in LC was compared in normally descended and abnormally descended testes in DBP-exposed animals, there was no consistent relationship between the level of Insl3 immunoexpression and testis position. It was therefore concluded that there was no straightforward relationship between DBP-induced suppression of Insl3 in Leydig cells in the fetal testis and the normality or otherwise of testicular descent. Instead, failure of normal testis descent, as occurs with high frequency in animals exposed in utero to high doses of DBP, may be the consequence of combined suppression of Insl3 and testosterone production. Alternatively, it may be that Insl3 immunoexpression detected using the antibody generated may not accurately reflect changes in secretion (and thus the blood levels) of Insl3. The latter possibility is supported by the finding that neither in utero exposure to flutamide nor to DES resulted in consistent reduction in Insl3 immunoexpression in LC at e19.5, whereas similar treatment with flutamide did result in a significant reduction in Insl3 gene expression and a similar, and larger reduction in mRNA expression, has been reported by others after in utero exposure of mice to DES in which cryptorchidism was also induced.

These studies have involved in vivo exposure of pregnant rats to various doses of bibutyl phthalate (DBP) in specific time windows or have used fetal testis explants from e14.5 rats cultured for 1-3 days with various doses of diethyl hexyl phthalate (DEHP) or its main metabolite MEHP. From our findings and those being published in the literature, the action of DBP, and certain other phthalates, on the fetal testis involves major suppression of Leydig cell (LC) hormone secretion. In males exposed to 500mg/kg/day DBP administered to their mothers, testosterone levels within the testis are suppressed by 70-85% at e19.5, when the peak of testosterone normally occurs. A similar pattern of suppression of Insulin-like factor 3 (Insl3) protein expression in LC is evident. These changes are explained by reduced expression of genes involved in cholesterol uptake (scavenger receptor B1) and steroidogenesis (P450 side chain cleavage enzyme, C17-20 lyase) as well as Insl3. The adverse effects of DBP on LC gene expression are first evident in vivo at e17.5. In contrast, in e14.5 testis explants, significant suppression of testosterone production is evident within 24h after culture with MEHP (but not with DEHP). The suppression of LC hormone secretion in studies in vivo is clearly important in explaining the high incidence of cryptorchidism and the occasional cases of hypospadias as well as the consistent reduction in anogenital distance (AGD) in DBP-exposed male offspring. Suppression of testosterone levels within the testis may also account for the major reduction in Sertoli cell number in DBP-exposed males at e21.5 and possibly for other changes in the fetal testis. The endpoint disorders related to suppression of fetal LC hormone production are directly relevant to human TDS. DBP exposure also causes abnormal migration/aggregation of the fetal LC in central regions of the testis. The cause is unknown but is presumed to result from altered production/action of cell attraction/motility factors. The most important effect of the DBP-induced LC aggregation is that it resuls in the focal formation of malformed seminiferous cords which may aberrantly trap fetal LC within them. These 'focal dysgenetic areas' containing intratubular LC, persist throughout postnatal life and always lack GC. Similar focal dysgenetic areas have been found in the testes of humans with testicular germ cell cancer. DBP also exerts major effects on fetal germ cell (GC) development and also on Sertoli cell (SC) development, effects that occur within different time windows. DBP exposure transiently delays fetal GC differentiation and entry into quiescence, and this occurs in a time-frame (e15.5-e17.5) that precedes the effects of DBP on LC function above; this delay results in a 40% reduction in GC numbers by e21.5. In e14.5 testis explants cultured with MEHP, there is also a reduction in gonocyte numbers. There are also postnatal consequences of the in vivo effects of DBP, with delayed reinitiation of GC proliferation. The DBP-induced delay in fetal GC differentiation may have relevance to the origins of carcinoma-in-situ germ cells (CIS cells) in humans which are the precursors of testicular germ cell cancer, as the CIS cells almost certainly result from complete or partial failure of the GC to differentiate normally. At a later stage in gestation (e19.5-e21.5) another effect of DBP treatment on GC becomes evident with the widespread occurrence of multinucleated GC (MNG) coupled with the abnormal aggregation of the fetal GC in the centre of the seminiferous cords. These effects only first become evident at e19.5 and can be induced by e21.5 even if DBP treatment is not started until e19.5. These GC effects are completely separate from, and unrelated to, the earlier effect of DBP on fetal GC differentiation above. It is likely that formation of MNG and GC aggregation reflect abnormal SC function and interaction with the GC. At the same time as MNG form there is also a big deficit in SC numbers in DBP-exposed males such that by e21.5, SC numbers are reduced by ~45%. The most likely explanation for this decrease is that it is secondary to the reduction in testosterone levels in the testis as a similar reduction in SC numbers is observed in androgen receptor knockout mice.

Expression analyses were conducted in selected rat hypothalamic and pituitary samples generated within the EDEN consortium in order to define a set of molecular biomarkers of exposure/action of EDCs during critical periods of sexual differentiation, in terms of disruption of hypothalamic-pituitary (HP) unit, and to identify the potential effects of such exposures, in terms of changes in gene expression of these markers at the HP unit. Our analyses in models of exposure to estrogenic (as well as androgenic) compounds during critical periods of sexual differentiation in the rat have allowed us to identify a selected group of molecular markers of endocrine disruption at the HP unit. Among those, identification of persistent changes in expression of KiSS-1 system at the hypothalamus (and their functional relevance) has proven the principle that this system might be considered as a potential target (and suitable gene marker) of exposure/action of EDCs with estrogenic and/or androgenic activity. Analyses so far conducted in samples from large-scale animal (rat) experiments (conducted by EDEN partner 21), testing the effects of early exposures to selected anti-androgenic EDCs alone or in mixtures, have failed to identify consistent changes in hypothalamic KiSS-1 gene expression at early periods of postnatal development (PND 16). However, further analyses at later age-points are eagerly needed in order to define the potential impact of such exposures on the activation of gonadotropic axis at puberty, and its function later in life.

It was hoped that during EDEN one or more lines of useful transgenic fish would have been established, and that some preliminary ecotoxicological assessment of the line (or lines) conducted, to gauge how useful it/they might be. However, this has not been achieved. However, significant progress towards developing one has been made, and it is possible that success will be achieved in the not-too-distant future. The production of lines of stable transgenic fish has proved to be more difficult than was envisaged at the outset of the project. It is possible that Partner 7 is close to establishing one or more lines of transgenic zebrafish possessing an estrogen-responsive promoter (an ERE) connected to the green fluorescent protein (GFP) gene. However, one further round of screening of the F1 fish from Batch 2 is required, followed by breeding a further generation (F2 fish), which will need to be screened for the presence of the transgene. Unequivocal evidence of the presence of the transgene in some (probably 50%) of the F2 fish would indicate germline transmission of the transgene, and allow a transgenic line to be established. Three factors seem to play key roles in the production of transgenic fish: the number of eggs injected, the time involved in raising and screening offspring, and luck. The more eggs that are injected with the transgene, the greater the chance of producing a useful transgenic line of fish. A great deal of time, and costs, are associated with rearing fish to adulthood, when they can reproduce, and produce the next generation. After a batch of eggs has been injected, it takes 4 to 6 months for the fish to reach sexual maturity. Only by screening the next generation can it be ascertained whether or not the transgene has been transmitted to the next generation. Even if it has, it will be to only a small proportion of the offspring, and these need to be identified, then reared to adulthood so that they in turn can reproduce. A minimum of 3 generations, and hence about 18 months, are required to unequivocally establish whether or not a transgene has been stably integrated into the germline. We will continue to study these fish, in the hope that a useful line of transgenic fish can be obtained.

There is a need to deal with risk characterisations for EDC mixtures based on recent evidence showing that endocrine disrupting chemicals (EDCs) can act jointly to produce combination effects, and at doses that individually do not produce observable effects. Data requirements for EDC mixture risk assessment (RA) are - Define which chemicals should be considered to be EDCs; - Identify which combinations of EDC should be regarded as being of concern; - Data on EDC exposures are essential to judge what relevant doses or concentrations present in combinations are. Also important are reflections on settings for which mixture RA might be needed. The following relevant settings for EDC mixtures being: - Medium- or population-specific RA e.g. human exposure via food, air, water; - Process-specific RA e.g. sewage effluent into the aquatic environment; - Product-specific RA e.g. cosmetics; - Generic RA for certain groups of chemicals e.g. PCB's and dioxins, steroidal hormones. For humans, knowledge about the total exposure via food, air, water, and consumer products is necessary. Attention should be paid to the identification of specific highly exposed sub-populations with consideration of vulnerable windows of exposure. In terms of data requirements, this task will be greatly aided by data about internal exposure to EDC e.g. tissue levels. Some of these data are currently gathered as part of the EDEN project but literature searches are essential to supplement this. Data about combined exposures of one and the same human population would be particularly valuable here but are very hard to obtain. A comprehensive exposure assessment for humans is complicated by the fact that there is still considerable uncertainty about the nature of EDC humans may have contact with. Comparatively, wildlife exposure scenarios are better defined. Hazard assessments for EDC mixtures can be sub-divided into how mixture effects should be determined and their assessment and prediction. The approaches for mixture effect determination are: - Complex mixture testing or 'whole mixture testing - e.g. whole effluent testing; - Reconstituted laboratory mixtures termed 'similar or simple mixture testing'; - Mechanistic approaches. The data requirements of each of these approaches differ considerably. Whole mixture approaches rely on fairly well-defined sampling protocols that account for the possibility that the precise composition of the mixture to be studied (which remains unknown) may vary depending on sampling conditions. Whole mixture approaches may help characterising whether there are joint effects at all but without knowledge of the chemical composition and the effects of individual mixture components, this approach is only useful on a case-by-case basis. For viable reconstituted laboratory mixtures, data about the effects of each individual mixture component are required. In some cases, knowledge about specific effect doses e.g. ED50 may be sufficient but often entire dose-response relationship information is essential if judgements in terms of additivity, synergisms or antagonisms are to be made. Mechanistic approaches rely on data about the underlying mechanisms of action of chemicals in the mixture, with information often fragmentary. Two main concepts exist for the quantitative assessment of mixture effects in terms of additivity, synergism or antagonism. These are concentration or dose addition (DA) and independent action or sometimes referred to as response addition (IA). Often, but not always, these two approaches yield the same quantitative mixture effect prediction. Thus, the question arises as to whether it is possible to select one of the two concepts as a default approach for mixture effect assessment. In ecotoxicology, systematic comparative studies of the mixture effect predictions produced by DA and IA have shown that DA yielded the more conservative predictions. However, overall the quantitative differences between both concepts were relatively small. Here, the case can be made for using DA as the default approach for mixture assessments thus avoiding lengthy and largely fruitless discussions about establishing modes of action. Such a modus operandi would have two advantages: - Data requirements for proper use of dose addition are less stringent than those for IA. While the former works well on the basis of effect doses, the use of IA usually requires knowledge of entire dose-response curves, particularly in the low effect range; - Prospective mixture effect assessments should be compliant with the precautionary principle. This favours the concept that typically yields the more conservative predictions i.e. DA. While the case for dose addition is validated in ecotoxicology, the situation is not so clear-cut in human toxicology. Here, the relevant information is largely missing and research efforts are currently directed into conducting studies to fill these gaps.

Genomic response have a high potential to improve and enhance existing toxicity test guidelines since they can unravel whether an adverse effect of a chemical on growth, development or reproduction is based on an endocrine mode of action. In addition, genomic responses can be used at an early stage of testing to direct further hazard assessment and to make it more targeted. Finally, genomic tools offer the possibility to detect multiple actions of one substance as well as interactions of multiple substances. To conclude, genomic responses clearly increase the ability to detect biological and ecological functions at risk, support informed testing strategies and render both diagnostic and predictive hazard assessment more reliable. Currently, the major drawback in the utilization of genomic responses is that these technologies are still at the beginning of their application, and require further knowledge and experience on wise use and interpretation before they can be incorporated into standard testing protocols.

Mixtures composed of similarly acting EDC were assessed by using cell-based assays (E-Screen), an extended developmental toxicity model in the rat, vitellogenin induction in the stickleback and in zebrafish and suppression of testosterone-induced spiggin synthesis in the stickleback. Concentration (dose) response curves for each individual chosen mixture component were recorded with the aim of using this information to calculate additivity expectations for mixtures of defined composition. It was found that the joint effects of EDC belonging to the same class (estrogenic, antiandrogenic) were dose or concentration additive. These findings are relevant to mixtures risk assessment: They show how information about the potency of chemicals can be used to arrive at valid predictions of their joint effects. They lay the foundations for using a default concept for mixtures risk assessment.

The objectives of WR9 were to deliver standard operating procedures for performing studies with the stickleback assay and to carry out range-finding studies as a basis for low dose and mixture experiments in WP13 and 15Progress. The partners involved in this project met together for the first time to plan the testing protocol at on 27/28 January 2003. The partners met for a second time (also attended by the zebrafish groups and the project leader, Dr Kortenkamp) for further protocol planning on 5/7 May 2003. As a result of these meetings and further correspondence, two protocols were agreed upon for carrying out the exposures and the chemical analyses (details of protocols are provided at the ANNEX of the first year report). As planned, experiments were carried out by partner 19 to establish whether homogenates of either the heart or the liver (or of whole fish) were a good substitute for measurement of VTG in plasma. The results were very positive in that heart homogenates are providing a suitable tissue for VTG measurement. This work is in the process of being written up for publication. Also as planned, partner 19 drafted a SOP for aquatic fish maintenance, condition and exposure. Once this was agreed between all partners involved, the partners started to jointly conduct range-finding studies for a number of oestrogens. The experimental plan for most exposures involved the following: Partners 2 and 8 follow the SOP produced by partner 19 of fish maintenance and exposure to the agreed chemicals. Partners 2 and 8 dissect out the heart and kidneys of sticklebacks and sent the tissues to partner 19 for assay of VTG and spiggin, respectively. Partner 19 performed VTG and spiggin measurements on sticklebacks provided by partners 2 and 8 and communicated the results to partners 2, 8 and 1. Both partners 2 and 8 processed water samples from their experiments and sent the extracts to partner 19 for measurement of 'actual' concentrations of the compounds, which the fish had been exposed. Partner 19 and the subcontractor performing the chemical analysis provided detailed protocols and procedures on water collection and processing. The relevant experiments involved range-finding tests with estradiol (E2) and ethinyl oestradiol (EE2), oestrone (E1) and nonylphenol (NP). An additional live-fish experiment involving exposure of relatively immature fish to E2 only in order to establish an appropriate VTG time-course (28 days maximum) was carried out by partner 2. The relatively low level of mortalities observed in those first tests suggested that the agreed protocol was robust. Partner 19 has also established radioimmunoassay standard curves and extraction protocols for E2, EE2 and E1 and also tested all reagents required to measure exposure rates in water. The analysis of water samples collected from the first set of experiments showed that in general the recovery of steroid concentration is water samples was about 50% of the nominal. During the year, Dr Mayer (partner 8) transferred his base of operations from Stockholm University to Bergen University. Also during the course of the year, Dr Pottinger (partner 2) transferred his office from Lake Windermere to Lancaster University. Although these organisational changes caused delays to the start of the live fish experiments, they did not affect the first objective or the main deliverable which was to deliver a standard operating procedure for performing studies with the stickleback.

Human studies To characterize the relationship between serum insl3 levels and male reproductive function, serum insl3 levels were measured in blood samples collected from cryptorchid boys and from boys with normally descended testes. Insl3 levels were also measured in a large number of blood samples from male and female children of various ages from the neonatal period through puberty to adulthood. Additionally, blood samples from adult men during and after treatment-induced gonadotrophin suppression (which will suppress Leydig cell function) were evaluated for Insl3 levels. The studies have shown that serum INSL3 in normal men is not acutely sensitive to exogenous FSH or LH/hCG, but nevertheless is dependent on the stimulatory activity of gonadotrophins on Leydig cells. Suppression of endogenous gonadotrophins results in a marked decline in serum INSL3, following which INSL3 is acutely sensitive to LH action. Subsequent to long-term gonadotrophin suppression, INSL3 does not recover to the same degree as does testosterone, suggesting that INSL3 may be more sensitive than testosterone to impaired Leydig cell function in adulthood. Reduced blood levels of INSL3 in cryptorchid boys suggests impairment of Leydig cell function in cryptorchid testes and supports the hypothesis that cryptorchidism is associated with a primary testicular disorder; however, the magnitude of difference between blood Insl3 levels in normal and cryptorchid boys when measured postnatally is modest (levels overlap for the two groups). This may imply that Insl3 alone is not the most critical determinant of testis descent. Alternatively, it is possible that measurement of Insl3 levels postnatally does not accurately reflect levels in fetal life when Insl3 actions on the gubernaculum plays its role in testis descent. Measurement of Insl3 levels in amniotic fluid and relating levels found to subsequent normality or not of testis descent may resolve this issue. Rat studies To characterize the relationship between Insl3 mRNA or protein expression levels in fetal life and testicular descent and non-descent (cryptorchidism), Insl3 expression in fetal testes and in the Leydig cells of such testes has been evaluated in rats exposed in utero to dibutyl phthalate, flutamide or to other EDCs (TCDD, p-p-DDE). Immunoexpression of Insl3 in the adult generation of Leydig cells has also been evaluated in normal rats during progression though puberty into adulthood or after various treatments likely to affect Leydig cell function; blood samples from some of the latter treatment groups have been evaluated for Insl3 levels during development and validation of the rat Insl3 assay. Finally, a transactivation assay for Insl3, with which to screen chemicals for Insl3-gene modifying activity was developed and tested. Results for blood levels of Insl3 in the rat have been very limited and the rat Insl3 assay has still not completed validation and optimisation. However, the results obtained are largely in tune with the findings from the human in showing that factors that suppress gonadotrophin secretion and/or Leydg cell development postnatally suppress Insl3 levels. In the fetus, it has been shown that a maternal treatment (DBP) that induces a high incidence of cryptorchidism in the male offspring (due primarily to impaired development of the gubernaculum), results in major suppression of Insl3 gene and protein expression. However, no consistent relationship was found between cryptorchidism and the level of Insl3 expression, perhaps suggesting that the parallel suppression of testosterone production by DBP may also be an important factor in determining failure of testis descent. Various pieces of evidence suggest that testosterone, or other androgens, might play a role in stimulating Insl3 expression, though this needs further investigation. The Insl3 reporter gene (transactivation) assay shows that there are no direct effects of oestrogens in this system whereas androgens may be weakly stimulatory. The human and rat studies support a role for Insl3 in testis descent, but have also shown that there does not appear to be a simple, straightforward relationship between the normality of testis descent and Insl3 expression in the testis or when measured (postnatally) in blood. However, this conclusion is based on limited material. Postnatally, and particularly in adulthood, Insl3 levels may provide important information on the normality or not of Leydig cell function and more detailed comparison of testosterone and Insl3 levels should establish the relative utility and usefulness of these two markers. Transactivation assays for Insl3 probably have limited utility.

The assay was built up using: - An anti-ratInsl3 antiserum (CR15) generated using genetic immunisation of rabbits. - A synthetic rat Insl3 peptide (see Kawamura et al. 2004, PNAS 101, 7323-28) supplied from Richard Ivell (Adelaide) and Ross Bathgate (Melbourne) as a calibration standard. - The same peptide labelled by us with a europium chelate to give fluorescence signals when used as a tracer to compete in the assay with calibration samples or with Insl3 molecules in actual rat blood samples, for binding sites on the anti-ratInsl3 antibodies attached to the measuring well. In the first experiments the use of antiserum CR15 resulted in disappointingly low uncompeted tracer binding (B0) values. So our other antibodies were tried (2-8F, a monoclonal against synthetic bovine Insl3, cross-reacting both with human and rat Insl3, a gift from Erika Bullesbach and Chris Schwabe; CR 14, made by us through genetic immunisation of rabbits against human Insl3; and 3/3A, our antiserum that we used for the human Insl3 assay), but these turned out to be even worse, both in B0 values as well as in sensitivity. After optimizing buffer and incubation conditions, CR15 finally led to sufficiently high B0 values and gave reliable standard curves. The study materials so far, aside from the antisera and synthetic Insl3 peptides mentioned above, comprised sera from normal adult female and male rats (as controls) gathered by us in Hamburg, sera from adult castrated rats sent to us by partners, and blood samples from EDS-, DBP- and otherwise treated as well as control male rats. The sera from castrated rats should most presumably be free of Insl3 and thus serve as blank sera, which can be spiked with the synthetic calibration standard to generate calibration curves in the same matrix as the actual sample. This allows more accurate measurements than would be possible with calibration curves run in buffer alone and circumvents the extraction of Insl3 from the samples. The reliable measuring range for the assay encompasses 5 to 300 pg rat Insl3 per well (which takes 100µl of sample). This is similar to the measuring range of our immunoassay for human Insl3. Samples can be measured either directly, as undiluted 100µl units of serum or plasma in the case of low Insl3 content, or can be diluted 1:5 or 1:10 in assay buffer when the amount of Insl3 exceeds the measuring range in 100µl serum/plasma. Specificity of the assay was controlled by spiking assay buffer with defined amounts of rat Insl3 and related peptides and comparing the respective competition curves. Only human Insl3, the closest relative to rat Insl3 in the peptides used, shows a slight reduction of tracer binding. Females, castrated and EDS-treated males (EDS destroys the Insl3-producing Leydig cells in the testes) are at the baseline, as expected. In male rats, there is an increase in Insl3 levels at around puberty (d 20/21), reaching a peak at day 43 and then declining somewhat in adulthood, a pattern that parallels that found for testosterone in numerous earlier studies. It can be assumed that Insl3 must have a function in the adult rat in the postpubertal age. Also, male rats appear to have, according to our preliminary studies, roughly ten times more Insl3 in their blood than male humans. Drastically reduced are Insl3 levels in blood of adult male rats having been exposed to DBP in utero, regardless of their actual testicular status. This has to be investigated in further detail and points towards an obstruction of Leydig cell or general testis function by DBP already in the fetal stage and lasting past puberty, also in rats in which the testes descend normally into the scrotum.

Various tissues were collected from rats at ages spanning from fetal life through to adulthood, especially from the reproductive tract (testis, efferent ducts, epididymis, gubernaculum), to enable evaluation of the distribution and level of expression of receptors for Insl3 (the LGR8 receptor) using RT-PCR and immunohistochemistry; some of these same tissues were used to assess, using RT-PCR, whether Insl3 mRNA expression was not detectable in the gubernaculum as we had shown by immunohistochemistry that Insl3 protein expression was detectable in this tissue. More limited evaluation of LGR8 immunoexpression in the testis and reproductive tract (including the gubernaculum) was also undertaken. It was shown that mRNA for LGR8 was expressed heavily in the gubernaculum throughout fetal and postnatal life, a finding confirmed at the protein level in more than one laboratory. These findings confirm that the gubernaculum is a key target for Insl3 in fetal life when testis decent is occurring. It was also shown that no mRNA for Insl3 can be detected in the gubernaculum of the rat, despite localisation of the protein by immunohistochemistry. It was concluded that the Insl3 protein detected in the gubernaculum was protein bound to its LGR8 receptor. Another key finding was that LGR8 mRNA and protein expression is not confined to the gubernaculum but is evident in other reproductive tract tissues, notably in the efferent ducts and epididymis during fetal and postnatal development. This implies that Insl3 from the fetal Leydig cells may play a wider role in development of the male reproductive tract, a process previously considered to be largely androgen-regulated (the androgens also emanating from the fetal Leydig cells). We also showed, unexpectedly, that LGR8 protein is expressed in fetal germ cells, implying a potential (but unknown) role for Insl3 in development of these cells. Some of the abnormalities in developing fetal germ cells found in rats exposed in utero to dibutyl phthalate, which coincide with suppression of Insl3, could be attributable to altered action of Insl3. It is concluded that, in the rat, the gubernaculum is an important target for the actions of Insl3 during the period when the testis descends through the abdominal cavity consistent with a role for Insl3 in testis descent. However, judged on the wider expression of Insl3 receptors throughout the fetal reproductive tract and in fetal germ cells in the testis, it seems likely that there are other, unknown functions of Insl3 in development of the male reproductive system.

An Expert Panel Workshop was convened on 15-19 May 2006 in Granada to discuss the options for incorporating knowledge on low-dose and mixture effects in testing strategies and regulatory efforts. Invited attendees represented national regulatory agencies, NGOs and research organisations. A representative from Schering AG, Berlin, Germany was contacted and invited, but his company was uninterested and did not grant permission to attend. The workshop itself was very productive. The panelists worked on a stepped procedure for mixtures regulation and risk assessment. First considerations were to which chemicals should be included for endocrine disrupter chemical (EDC) mixture regulation, and a decision on this issue depends on an answer to the question as to how to define what an EDC should be. It was felt that it is important to distinguish 'mode of action' from 'effects'. 'Endocrine disruption' is a mode of action relevant to developmental and reproductive toxicity, but not an effect. Instead, this mode of action can give rise to a plethora of different effects. Mechanistic considerations open the way for grouping EDCs according to test results in 'mode of action' screens, as with the customary classification into 'estrogens', '(anti)androgens', 'thyroid-actives' (EAT). However, the panelists recognised that this classification, with its emphasis on steroid and thyroid receptor interactions, does not capture all known endocrine effects and therefore suggested to include as additional category 'others', thus: EATO. The 'mode of action' screen classification leads into a dilemma: The EATO grouping is of limited relevance for risk assessment, because the predictive value of such screening outcomes for the occurrence of (adverse) effects is questionable. For example, not all androgen receptor antagonists produce responses typical of disruption of male sexual development (such as retained nipples, altered anogenital distance etc). For estrogens, in vivo effect models capturing the conditions of concern in the human (e.g. breast cancer) are not available. Furthermore, screens are imperfect, as exemplified by the case of certain phthalate esters which disrupt male sexual development by interfering with steroid synthesis, and not by receptor interactions, yet many anti-androgen screening tools do not detect these phthalates. On the other hand, a grouping of EDCs in a phenomenological fashion according to effects is not currently possible, because too few chemicals have been tested in relevant toxicity studies, or because the appropriate tests are not yet available. This situation is not likely to change in the foreseeable future. An all-encompassing solution is not on the horizon, however, for specific chemicals, effects and exposure scenarios it is now well established that EDCs can act in an additive fashion. Therefore, there are possibilities for utilising existing knowledge and data for making progress with risk assessment and regulation that takes account of mixture effects. Several examples and exposure settings were discussed including: - vitellogenin induction in fish from cumulative exposure to estrogenic chemicals (so-called mixture maps); - retained nipples and changes in anogenital distances in the rat; - Spiggin induction in the stickleback and - internal exposure to estrogens in women. Conclusions were that concentration addition is a powerful risk assessment tool for cumulative exposure in these examples. The panelists highlighted that knowledge about realistic exposure scenarios for EDC mixtures is fragmentary. In most cases, we simply do not know enough about the identity of EDC that co-occur in food, environmental matrices or human tissues, let alone their levels. This presents a formidable bottleneck to rational risk assessment for combined exposures - some participants even thought that this lack of data is the obstacle to making progress. Thus, there is a need to deal with knowledge gaps by making informed assumptions, and a step-wise approach, depending on the quality of data available, was discussed. The following situations could be distinguished: - In some cases, sufficient knowledge about the identity and levels of relevant EDCs is available - in these situations dose addition can be applied; - More often, however, there is uncertainty as to whether specific EDCs are able to induce the effects under consideration, but there may be some reason for concern. In these cases, a 'mixture assessment factor' could be applied for such EDC, so as to enable risk assessors to proceed and - If no data or information is available, it was considered to apply a default mixture assessment factor, making certain assumptions about the likely number of chemicals in the mixture.

Serum samples from 5350 Danish and 3270 Finnish men, who participated in large population studies conducted in Denmark between 1983-1999 and in Finland between 1972-2002 were analysed for testosterone, lutenising hormone (LH) and sex hormone binding globulin (SHBG) in order to test the hypothesis that a secular trend in male reproductive hormones exists. A general linear model was used to investigate the effect of age, period of sampling, and period of birth (cohort effect) on the hormone levels. Evaluation of this cross-sectional data indicate that, in addition to the well-known age effect of falling testosterone levels with increasing age, an age-independent population-level decline in testosterone levels seems evident with higher testosterone levels in the oldest cohort. A significant cohort effect was also observed for LH and SHBG with more recently born men having lower levels of all three hormones. Thus, Danish men born in the 1930ies had on average 19% higher T, 28% higher LH, and 41% higher SHBG levels than Danish men born between 1949 and 1954. Similar results were found for the Finnish men. In contrast, no cohort effect was evident for free T and T/LH levels suggesting that the cohort related changes in hormone levels were interrelated. The most pronounced cohort effect observed was for SHBG. The Danish and Finnish data could not be directly compared, as there was no overlap in the timing of collection of samples in the Danish and Finnish studies. The aetiology behind the observed cohort effect remains to be resolved. The effect was most pronounced for the oldest cohorts. Thus, SHBG levels in the men born between 1921-26>1931-36>1939-46>1949-56=1959-61=1969-70, indicating that whatever caused this effect seems to have levelled out in the more recent period.

Animal (rat) studies were performed to define the sensitivity of the hypothalamic-pituitary (HP) unit to the disrupting effects of estrogenic (and eventually androgenic) compounds after exposure during critical periods of sexual development, and to identify novel molecular end-points (biomarkers and mechanistic factors) involved in these disrupting events. Results from these experimental studies evidenced that the developing HP unit is highly sensitive to the (organizing) effects of synthetic estrogens, as monitored by molecular biomarkers (expression of PR and KiSS-1 genes at the hypothalamus and that of globin genes at the pituitary). Dose-responses curves allowed us to define that neonatal exposure to doses as low as 1-10microg estradiol benzoate per rat was able to induce significant changes in the relative mRNA levels of KiSS-1 (decrease) and PR (increase) at the hypothalamus, as well as in the expression levels (increase) of alpha- and beta-globin genes at the pituitary. The fact that these changes were tightly related to concomitant changes (decrease) in basal serum levels of LH, i.e. a conventional marker for disruption of function of the gonadotropic axis, strongly suggests that these might be regarded as genuine markers for potential endocrine disrupting events at the HP. Moreover, some of the above markers were also altered in expression after neonatal exposure to the aromatizable androgen, testosterone propionate, the anti-androgen, flutamide, and/or the putative xeno-estrogen, BPA, which reinforces the contention that expression of those genes (PR, KiSS-1, globins) at the HP unit might serve as reliable, highly sensitive, continuous biomarkers of exposure (and eventual disruption at this level) to sex steroid-like compounds during critical periods of sexual differentiation. From a mechanistic perspective, some of the identified changes in gene expression in the experimental models of exposure to synthetic estrogenic and related compounds might not only be relevant in terms of setting thresholds and sensitivity for endocrine disruption at the HP unit, but they may provide also novel information regarding end-points and mechanisms for endocrine disruption at this level of the reproductive axis. The most salient example of this contribution is the definition of the hypothalamic KiSS-1 system not only as a pivotal element in the central regulation of the gonadotropic axis in normal conditions, but also as a putative target for endocrine disruption by exposure to estrogenic (and possibly androgenic) compounds during critical periods of sexual differentiation of the HP unit. Likewise, altered expression of PR at the hypothalamus, and eventually of globin genes at the pituitary, may prove mechanistically relevant to explain some of the alterations in development and/or function of the reproductive axis following early exposure to sex steroid-like acting compounds.

Semen quality of the Finnish men born at 1983 and examined during 2002-2003 showed an adverse trend when compared to previous cohorts studied in 1998-1999. The median sperm concentration was only 47 million/ml that is similar to Danish figures while in the previous years the Finnish sperm concentrations were clearly higher (60) than Danish (47). The Finnish trend mirrors the increasing incidence of testicular cancer in the same age group of men, whereas in Denmark the increase in testicular cancer incidence appears to have levelled off and semen quality has remained unchanged over the last ten years. Thus, the difference in semen quality between Danes and Finns appears to vanish, and both share a serious problem in male reproductive health. Considering all cohorts of young men (n=4236) studied during the last ten years, semen quality in Finland (Turku) is still better than in Denmark (Copenhagen): median sperm concentration 46 million per ml in Copenhagen vs. 53 in Turku, total sperm counts 156 vs. 181 millions, respectively, and normal morphology (%) 6.5 vs. 8.8, respectively. However, the recent decline in Finnish semen quality gives a new warning signal that reproductive health can be endangered in all parts of Europe. The data of the three longitudinal examinations of the men born in 1979-1981 showed unchanged semen quality with increasing age indicating that one can already at the age of 18 or 19 years have a good estimate of life-time semen quality. These figures are lower than those of fertile Finnish men and those published from other Finnish study groups previously, suggesting a clear decline.

A hyperfractionation method was standardized to assess the total effective xenoestrogen burden (TEXB) in human tissues, by extraction and effective HPLC separation of two fractions containing lipophilic xenoestrogens (alpha fraction) and endogenous hormones (beta fraction), assessing their estrogenicity in MCF-7 breast cancer cell-based E-Screen and Yeast Estrogen Screen (YES) bioassays. Both bioassays can be used as end-points to assess TEXB. However, the E-Screen can be recommended because of greater experience of its use and its superior validity to identify and quantify estrogenicity in human placenta. Regression analysis of hyperfractionation data is a straightforward approach that leads to a result that is easy to understand and can provide a mechanistic basis for the hormonal activity of the combined fractions. Using this approach the estrogenicity of the whole alpha and beta fractions can be explained in part by the combined estrogenicity/anti-estrogenicity of individual HPLC fractions. In 1 out of 3 sites, atrophy of spawning tubercles shows a positive correlation with vitellogenin levels in male fish. One possible conclusion is that both effects are caused by the same stressor, a xenoestrogen, -antiandrogen, and/or -androgen. The comparison between bream that were affected by ovotestis and bream without ovotestis showed, as only detectable difference, a significantly higher concentration of estrone and ethinylestradiol in the bile of ovotestis fish, which however was not reflected in the total bile estrogenicity as determined using the YES, a possible indication for the presence of other EDCs. Hence, ovotestis gonads did not influence vitellogenin levels or the total estrogenicity measured with the YES. Estrogenicity and steroid concentrations in the bile did not significantly differ between the pools of fish with high and low plasma vitellogenin levels. This observation indicates that we either did not find all xenoestrogens causing induction of vitellogenin, or that the time point of bile sampling did not correspond to the situation that led to high levels of vitellogenin. The study finally shows that the natural steroids E1 and E2 make up the main part of the estrogenicity found in bile or adipose tissue of male bream using the YES. Ethinylestradiol was found as well but in much smaller concentrations. Fish bile and adipose tissue samples have been successfully measured with the hEST inhibition assay. No clear differences in hEST inhibition between cases and controls were observed. No specific fraction that potently inhibits hEST could be identified. However, these findings match the results obtained with the YES. Fish bile samples have also been successfully measured using the coumestrol-based HRS ERalpha setup. The presence of compound(s) having low affinity for the ERalpha in the full extracts was observed and we suggest these compounds to be either E2 or EE2. These compound(s) are also present in fraction 6. No other compounds present in the samples having ERalpha affinity could be detected, probably due to concentrations below the detection limits of the HRS ERalpha. Again these finding match the results obtained with the YES.

Based on the observations that - some EDCs have estrogen-like effects in vivo and, therefore, might interfere with developmental processes such as puberty in the female; and - in foreign girls migrating for adoption, sexual precocity is common, and the possible association with previous exposure to EDCs has been hypothesized, we conducted animal (rat) studies to characterize the effects and threshold doses (sensitivity) of EDCs, such as the insecticide dichlorodiphenyltrichloroethane (DDT), in terms of induction of precocious activation of the HP unit, using the rat hypothalamic explant incubation paradigm to investigate for novel molecular mechanisms for such a precocious activation. In addition, specific studies to define animal models for the in vivo characterization of the above phenomenon were conducted. By combination of in vitro and in vivo analyses, we have shown that some EDCs, including DDT isomers, can stimulate the GnRH secretion in vitro in the immature female hypothalamus through both rapid and/or slow effects and they involve both estrogen and dioxin receptor pathways. These effects can account for the precocious sexual development observed after early exposure to estradiol or DDT in vivo, thus providing the molecular basis for the clinical observation of precocious puberty in foreign girls migrating for adoption in several European countries.

A network of research called the Cluster of Research on Endocrine Disruption in Europe (CREDO) was established under the jurisdiction of the EDEN project (http://www.credocluster.info/intro.html). CREDO would facilitate effective research collaborations across a core of 4 projects (EDEN, COMPRENDO, EURISKED and FIRE) and a further 8 projects funded in the last round of the Fifth Framework programme. A brochure profiling CREDO was first available detailing the aims of the cluster and research programme (http://www.credocluster.info/docs/credobrochure.pdf). Eight newsletters were published from September 2003 to November 2006 communicating research features from the projects and information on workshops. The final double issue (7&8)provided summaries of the findings from these projects, along with the perspectives for environment and health research activity in the Seventh Framework Programme (FP7).

Men at the age of 18-20 from the former West (Hamburg) and East (Leipzig) Germany were recruited for a study on male reproductive health when they attended the compulsory medical examination for military eligibility. In total 1,004 men were initially included into the study. However, 213 were excluded because either they or their mothers did not meet the eligibility criteria for place of birth. Thus, 334 men from Hamburg, and 457 men from Leipzig were included in the final analyses. Clinical examination and semen analysis were standardised according to a rigid protocol established in previous European studies on male reproductive health. Serum levels of reproductive hormones were measured and the men also completed a questionnaire. Numerous confounding factors (e.g. age, year of birth, duration of abstinence, time of sampling) affecting semen and hormone parameters were all tested in a multiple linear regressions analysis, taking possible interactions into account. No significant differences were observed for sperm concentration and total sperm count, whereas semen volume, sperm motility and morphology were significantly different between men from the two areas; higher frequency of morphologically normal forms and higher semen volume in the Hamburg group and higher frequency of motile spermatozoa in the Leipzig group. For the hormones the inhibin-B level was significantly higher for the men from Leipzig and Estradiol and LH higher for men from Hamburg, whereas there was no difference in the Testosterone level. Our study shows that the semen quality of young men from the German general populations from Leipzig and Hamburg is at a low level, that there is no biologically relevant difference in semen parameters between the two groups and that their sperm counts are at the same level as young men from Denmark and Norway.

International experts and scientists representing many different disciplines came together in Prague on 10 - 12 May 2005 for a workshop on chemicals that interfere with hormone systems, so-called endocrine disrupters. The workshop 'Endocrine Disrupters: Exposure Assessment, Epidemiology, Low-dose and Mixture Effects' was organised by the EDEN and FIRE consortia and was convened to discuss recent European research on the health risks associated with these chemicals. At this international workshop, scientists from across Europe presented the latest research findings in endocrine disrupter research. Much of this work emanated from large research projects funded by the European Union, and joined together in the cluster for research on endocrine disrupters, CREDO. There were discussions to promote the transfer of know-how between scientists on the issues of exposure assessment, epidemiology, low-dose and mixture effects. The programme included sessions on male reproductive health in Europe, human and wildlife exposure to endocrine disrupters, novel endpoints and biomarkers, as well as on low-dose and mixture effects of endocrine disrupters and their assessment. The results presented at the Prague workshop have reinforced concerns over the long-term consequences of exposure to endocrine disrupters to humans and wildlife. Outcomes from the workshop resulted in a Special Issue in Environmental Health Perspectives which was available online from June 2007. The workshop also gave rise to the 'Prague Declaration on Endocrine Disruption' which has been signed by over 200 scientists (http://www.edenresearch.info/public/PragueDeclaration%2026%20May%202006.pdf)

Analyses of human and fish tissue specimens for chemicals with endocrine-disrupting potential has been done. Frequencies and levels of EDC in humans (Children and women) and fish tissues were done. The data set from Southern Spain women is unique because it also involved the follow-up of women which have undergone a major surgery for breast cancer. Three successive tissue samples were obtained in a 18-months follow-up period. All this information is organized in a database on EDC in human and fish tissues that is one of the largest database on EDC exposure ever made. Our results confirm the high frequency of detection of EDC residues in tissues from breast cancer women and controls and in adipose tissue samples from fishes. In addition, exposure data to EDC in infants or in very young children makes this study very valuable since there is only few EDC data on children available worldwide. Our results demonstrate that in infants, because of small body size, the body burden of EDCs can easily reach similar levels what has been measured in adults due to exposure in the prenatal period and following lactation. Development of body burden in early childhood, influence on lactation to body burden and possible association of POPs and cryptorchidism can be studied when all relevant information about subjects have been added to this data set. The following groups of endocrine dissrupting chemicals (EDC) have been investigated in human and fish tissues:Organochlorine pesticides, Poplychlorinated byphenils (PCBs), Poplybrominated diphenyl ethers (PBDE), Polybrominated byphenils (PBBs), Dioxins and furans, Bisphenol-A and chlorinated derivatives, Alkylphenols, Natural endogenous sex-steroids, Benzophenone, phytoestrogens and parabens and Phthalates.

Livrables

Partager cette page

Télécharger