Grantee Research Project Results
Final Report: High-Throughput Cellular Assays for Modeling Toxicity in the Fish Reproductive System
EPA Grant Number: R835167Title: High-Throughput Cellular Assays for Modeling Toxicity in the Fish Reproductive System
Investigators: Schultz, Irvin R.
Institution: University of Washington
EPA Project Officer: Aja, Hayley
Project Period: August 1, 2012 through September 30, 2015 (Extended to September 30, 2017)
Project Amount: $1,199,908
RFA: Developing High-Throughput Assays for Predictive Modeling of Reproductive and Developmental Toxicity Modulated Through the Endocrine System or Pertinent Pathways in Humans and Species Relevant to Ecological Risk Assessment (2011) RFA Text | Recipients Lists
Research Category: Chemical Safety for Sustainability
Objective:
The overall objective is to demonstrate that cellular in vitro assays can successfully guide parameterization of computational models of fish reproduction. Specific objectives include optimizing rainbow trout pituitary cell culture for assessing toxicant effects on follicle stimulating hormone (FSH) and luteinizing hormone (LH) synthesis. Other objectives sought to optimize in vitro ovarian follicle incubations for assessing basal and FSH induced steroidogenesis and utilize rainbow trout primary hepatocyte culture for assessing toxicant biotransformation and effects on vitellogenin (Vtg) expression.
Summary/Accomplishments (Outputs/Outcomes):
Objective 1: Pituitary cell culture. Stage Dependent Effects of E2 on FSH and LH: Primary pituitary cell cultures were prepared from previtellogenic rainbow trout (9 months pre-spawning) and from early-mid vitellogenic rainbow trout (5 months pre-spawning). Cultured pituitary cells were exposed to 0.37 to 367 nM E2 for 48 hours and then salmon gonadotropin-releasing hormone (sGnRH; 10 nM) was added to half of the cells for an additional 18 hours to investigate effects on basal versus GnRH-induced hormone secretion. Luteinizing hormone beta subunit (lhb) was upregulated 8- to 20-fold in pituitary cells from previtellogenic trout at all concentrations tested with maximal induction at 37 nM E2. E2 also increased GnRH-induced secretion of LH in cells from previtellogenic trout. In contrast, E2 had no effect while GnRH had a slight positive effect on follicle stimulating hormone beta subunit (fshb) mRNAs. However, when GnRH was combined with E2, the positive effect on fshb was not observed. In pituitary cells from early vitellogenic females, lhb was upregulated 1.5-fold at all concentrations tested, and this effect was not concentration dependent. There was no effect of E2 on basal or GnRH-induced LH secretion. Due to the greater inducibility of lhb and LH secretion by E2 in previtellogenic rainbow trout, subsequent tests were conducted with pituitary cells from previtellogenic trout.
Time course of FSH and LH response to E2 and 11-Ketotestosterone: A time course experiment was conducted to optimize the duration of pituitary cell culture for test chemical screening. E2 treatment increased lhb mRNA levels from 3 days onward, with dose- and time-dependent increases in E2-induction of lhb. E2 also increased GnRH-stimulated LH release. Pituitary fshb mRNA levels were weakly increased 1.2- to 1.4-fold when exposed to 3.7 to 37 nM E2, however, E2 decreased basal secretion of FSH and had no effect on GnRH-stimulated release of FSH. In contrast, 11-ketotestosterone (11-KT) had no significant effect on lhb mRNA or LH protein levels, but weakly decreased fshb mRNA levels in response to 3.7 -37 nM treatments. These results confirm the sensitivity of LH beta subunit gene expression to E2 stimulation. These results indicate that 11-KT alone has little effect on fshb or lhb mRNA levels. Based on these data, pituitary cell cultures should be maintained for a minimum of 3 days for chemical testing to ensure the lhb mRNA response is observed. Control levels of transcripts were observed to decline over time (data not shown) thus, longer-term pituitary cell cultures may require a media with supplementation of fetal calf serum or serum substitutes. However, this may complicate tests of the effects of steroidal chemicals because of the presence of steroids and growth factors in fetal calf serum that may alter gonadotropin gene expression.
Testing pituitary cell response to a suite of chemicals: A series of experiments was performed using primary pituitary cell cultures prepared from previtellogenic trout and exposed to 15 different chemicals: 17α-ethynylestradiol (EE2), tamoxifen (TAM), 4-OH-tamoxifen (4OH-TAM), prochloraz (PRCHL), 11-KT, testosterone (T), trenbolone (TREN), gestodene (GEST), norgesterol (NORG), progesterone (P4), fluoxetine (FLX), sertraline (SERT), citalopram (CIT) and 3-hydroxy - 2,2,4,4, terabromodiphenyl ether (3-OH-PBDE-47). Pituitary cell cultures exposed to 0.1 – 1000 nM EE2 with or without GnRH showed altered gonadotropin gene expression. Similar to results obtained with E2, EE2 strongly stimulated lhb mRNA levels at concentrations of 1 to 1000 nM EE2, with maximal induction at 10 nM EE2. Fshb mRNA levels were weakly decreased by 100 nM EE2 alone, or 10 – 1000 nM EE2 in the presence of GnRH. In contrast to previous experiments with E2, EE2 did not significantly alter GnRH-induced release of LH into the culture media.
The next set of experiments examined the pituitary cell response to estrogen receptor antagonists. First, we specifically evaluated the anti-estrogenic activity of 4OH-TAM. Dispersed pituitary cells were exposed to a range of E2 dose levels with or without 250 nM 4OH-TAM, and the responses of Gth mRNA levels were examined. As expected, E2 concentrations of 0.37 to 183 nM increased lhb mRNA levels 8- to 27-fold relative to control levels. However, addition of 250 nM 4OH-TAM abolished the stimulation of lhb by E2 concentrations up to 3.67 nM and significantly decreased E2-stimulation of lhb at all concentrations. Next, we compared the anti-estrogenic activity of TAM, 4OH-TAM and PRCHL in trout pituitary cells with or without 4 nM E2, and examined the response of Gth mRNA levels. TAM and 4OH-TAM displayed similar anti-estrogenic activity by significantly inhibiting E2-stimulation of lhb at 1000 nM and 500 nM, respectively. PRCHL also showed anti-estrogenic activity, but a dose of 20,000 nM was required to significantly reduce E2-stimulation of lhb. Steady state mRNA levels of fshb were not significantly altered by E2 or any of the estrogen receptor antagonists in these in vitro exposures. We next exposed trout pituitary cells to the natural non-aromatizable androgen hormone, 11-KT, T (the natural aromatizable androgen) and the synthetic androgen, TREN. The only significant effect observed was an increase in lhb mRNA levels in response to T. As mentioned, the enzyme aromatase can convert T to E2 in pituitary cells. Therefore, the upregulation of lhb in response to T is likely due to an increase in E2 concentration in these cells. We next investigated the effects of selective serotonin reuptake inhibitor (SSRI) pharmaceuticals. Cytotoxicity was observed in pituitary cells exposed to 10000 nM FLX, 11,400 nM SERT, or 48,900 nM CIT or greater. These were the first chemicals observed to cause overt cytotoxicity in the 3-day pituitary cell cultures. All SSRIs tested reduced the E2 induction of lhb at relatively high doses. In addition, high concentrations of CIT caused upregulation of fshb mRNA levels. Citalopram is the only chemical tested so far that has shown a significant alteration of fshb mRNA levels. We also exposed trout pituitary cells to different doses of 3-OH-PBDE-47 with or without 4 nM E2. The expression of fshb and lhb mRNA levels was compared across treatments. In addition, thyroid stimulating hormone beta subunit (tshb) was measured due to the known effects of PBDEs on the thyroid axis. No significant effects were observed suggesting that hydroxylated PBDEs do not directly affect pituitary gonadotropin hormone expression in vitro.
In summary, the most sensitive pituitary cell endpoint to chemical exposure was altered expression of lhb, which was decreased in eight treatments and increased in two. Some chemicals exhibited complex effects, dependent on the presence or absence of E2.
Objective 2: Ovarian follicle. Initial experiments were conducted using mid-vitellogenic rainbow trout follicles ranging in size from approximately 1-3 mm diameter. Comparison of basal and time-course production of E2 production was evaluated in response to a range of concentrations of partially purified salmon gonadotropin (sGTH; a mixture of FSH and LH), using L-15 and a trout (teleost) Ringer’s saline solution as the incubation media. On the basis of these experiments, the following protocol was adopted: ovarian fragments (10 follicles/ml) are cultured in trout Ringers for 18 hours at 12-14 oC with varying concentrations of a test chemical. Test concentrations were chosen to span environmentally relevant levels to those limited by cytotoxicity or aqueous solubility. At 18 hours, the culture media is completely changed and replaced with media containing the same test concentrations with or without 500 ng/ml sGTH. At 6 hours, 12 hours, and 24 hours following the media change, 0.1 ml is collected for assessment of E2 levels by radioimmunoassay. The concentration of E2 in media is compared across treatments and time points. A reduction or increase in either the rate or total production of E2 measured in the media relative to controls would indicate adverse effects of the test chemical. We tested nine chemicals (EE2, TAM, 4OH-TAM, PRCHL, TREN, FLX, nor-fluoxetine [NFLX], Flutamide [FLUT], 2-OH-flutamide [2OHFLUT]. Of these, EE2, PRCHL and TREN caused the most significant changes on E2 production. The most potent chemical was EE2, which decreased E2 production at exposure concentrations as low as 0.1 nM after the sGTH challenge, while decreasing basal production at 100 nM. PRCHL significantly decreased both basal and the sGTH stimulated E2 production at levels down to 22.4 nM. TREN exhibited complex effects on E2 production. At treatment levels 40 nM and higher, basal E2 production was stimulated. However, trenbolone decreased E2 production after the sGTH challenge. We repeated the trenbolone experiments using a longer exposure time of 40 hr, which provided similar results. A significant decrease in relative E2 production after the sGTH challenge was seen with high concentrations of fluoxetine (10,000 nM). Incubation with the FLX metabolite NFLX indicated a similar but non-significant trend at concentrations ranging from 400 nM to 10,0000 nM. Incubation with FLUT at concentrations 4000 nM and higher appeared to decrease both basal and sGTH stimulated E2 production but was not statistically significant. However, incubation with the active flutamide metabolite 2OHFLUT had no effect even at a high exposure of 10,000 nM. TAM appeared to decrease basal E2 production at 500 nM but did not alter response to sGTH stimulation. The tamoxifen metabolite and potent anti-estrogen 4OHTAM, had no effect on E2 production.
In conclusion, our experimental results indicate that xenoestrogens (EE2, TAM), aromatase inhibitors (PRCHL) and xenoandrogens (TREN) were capable of altering E2 production in a dose-dependent manner. Other classes of contaminants such as anti-estrogens, anti-androgens and SSRI type pharmaceuticals have little or no effect on E2 production. An 18H incubation time appears adequate for the follicles to accumulate a significant amount of test chemical and impact E2 production. Overall, our results demonstrate that isolated rainbow trout ovarian follicles readily permit the rapid assessment of chemical exposure on both the total E2 production and on the rate of E2 production in response to gonadotropins.
Objective 3: Hepatocyte culture. We used hepatocytes freshly isolated from juvenile male rainbow trout. Initial studies cultured hepatocytes without added E2, which permitted estimation of the estrogenic potential of a contaminant. Chemical exposure of hepatocytes began with EE2, which was incubated with cells for 24 and 48 hrs. Endpoints measured included Vtg mRNA expression (measured by Q RT-PCR), Vtg protein (measured by Vtg ELISA), and ERα1 mRNA (measured by Q RT-PCR). A robust, dose response for both of the mRNAs was observed at each time point. The Vtg protein endpoint was only apparent in the medium at the two higher exposure levels after 48 hours. Based on these results, it was determined that Vtg and ERα1 transcriptional response after 24 hrs of exposure was adequate to detect induction. Next, we evaluated strain differences in the consistency and reproducibility of the response to EE2 exposure. These studies used an “outbred” strain of rainbow trout maintained at local hatcheries and an isogenic (genetically identical) strain of male rainbow trout. The latter was tested as it was hypothesized less inter-fish variability in the response would be observed compared to outbred trout. Hepatocytes were prepared from three individual trout of each strain and subsequently exposed to a series of EE2 concentrations. We observed that ERα1 induction by EE2 is similar among individuals with a consistent dose-response. Greater inter-individual variability was observed for Vtg where peak induction varied four-fold among individuals. A similar pattern was observed with outbred trout, suggesting variability in the Vtg response is unrelated to genetic differences among the test fish. In general, hepatocytes prepared from either the isogenic and outbred trout appeared to respond similarly to EE2 exposure. Based on these results, we only used outbred trout for future hepatocyte preparations.
We then tested the response after exposure to five concentrations of TAM, FLX, and FLUT. None of the chemicals consistently altered expression of ERα1 and Vtg. FLX appeared to decrease the expression ERα1 by 50% at the highest treatment level. In two fish, TAM slightly increased Vtg expression at higher treatment levels. Subsequent analysis of Vtg protein synthesis indicated low-level induction by TAM in fish 1 and 2, but not fish 3. This would suggest that TAM elicits a weak estrogenic effect in the hepatocytes. Subsequent experiments explored co-treatment of test chemicals with E2, which is added as a chemical competitor to stimulate a basal Vtg synthesis comparable to mid-vitellogenic female trout. We evaluated different doses of E2, timing of toxicant introduction, and cell harvesting. We found that an E2 dose of 3670 nM induced Vtg expression to a level similar to that observed in mid-vitellogenic female trout and was selected to add to the cells either 24 hours before or simultaneously with toxicant introduction. When hepatocytes were exposed simultaneously with E2 and either TREN or 4OH-TAM, the expression of ERα1 and Vtg was decreased. 4-OH-TAM was able to decrease Vtg expression at test levels down to 49 nM. We found that pretreatment of hepatocytes with E2 for 24 hrs followed by the 24 hr chemical treatments was less satisfactory, often providing conflicting results. This was attributable in part, to the near maximal induction of Vtg mRNA that occurs by 24 hrs of E2 exposure.
In summary, our testing suggests 24 hrs is an adequate time period to assess induction of Vtg and ERα1 mRNA in trout hepatocytes. The expression of Vtg exhibits considerably more inter-fish variation than ERα1. Strain differences among rainbow trout does not appear to be a factor in observed variability. With regard to interference of E2 stimulation of Vtg expression, our testing suggests a protocol that uses freshly prepared hepatocytes simultaneously co-treated with E2 and the test chemical is more sensitive than E2 pretreatment. Integrating results from different tests is valuable as it helps to understand complex responses of some chemicals like tamoxifen, which can elicit a weak estrogenic mode of action and an anti-estrogenic mode of action. The latter appears attributable to the rapid formation of the anti-estrogen metabolite, 4-OH-tamoxifen during hepatocyte incubations.
Objective 4A: Computational modeling. This objective focused on using a computational model of the female rainbow trout hypothalamus-pituitary-ovarian-liver (HPOL) axis. We used the results from in vitro testing of specific contaminants to guide adjustment of model parameters and then through model simulations, predict effects on oocyte growth and ovulation (considered by the model to be equivalent to spawning). We then used the HPOL model to simulate oocyte growth under series of hypothetical changes on key model parameters such as E2 synthesis in ovarian follicles, VTG synthesis in hepatocytes and GTH synthesis in the pituitary. These simulations indicated when E2 synthesis is reduced by 25 % during the entire cycle, oocyte growth is incomplete and failure to ovulate is predicted. These simulations permitted estimation of the threshold for adverse effect (delayed or incomplete oocyte growth) when specific model parameter(s) are altered. We then used the in vitro results to determine the extent of change in a parameter value that is proportional to the in vitro difference from control values.
Objective 4B: Toxicokinetic modeling. This objective focused on evaluating applications of clearance-volume toxicokinetic (CVTK) and physiologically based toxicokinetic (PBTK) models for reverse dosimetry calculations. The CVTK modeling activities focused on tamoxifen, prochloraz and SSRI pharmaceuticals. In vivo toxicokinetic data was collected using a static water exposure system (90 L in volume) that lasted for 72 or 84 hrs. Individual mid-vitellogenic female trout were placed in the exposure system and water samples removed periodically for chemical analysis. At the end of the exposure the fish were euthanized, tissues removed and analyzed for the test chemical. The CVTK model with in vitro and QSAR guided model parameters was then used to simulate water levels and accumulation into the fish at the end of the exposure. The CVTK model adequately described the uptake of tamoxifen into the fish but tended to overestimate accumulation. The latter may be due to greater metabolism of tamoxifen than was estimated from in vitro studies. The CVTK model was able to describe uptake of prochloraz and its accumulation assuming metabolic clearance was 10 ml hr-1 g-1, which was consistent from in vitro studies. Of the three SSRI type pharmaceuticals tested, sertraline was much more rapidly absorbed and was accurately described by the model. We subsequently focused PBTK modeling efforts on chemicals that exhibit complex / challenging toxicokinetic behavior such as bioactivation with enterohepatic circulation or carrier mediated tissue accumulation. These efforts are ongoing and have initially focused on tamoxifen and it anti-estrogen metabolite 4-OH-tamoxifen. Model simulations using largely in vitro derived parameters was able to accurately predict tissue accumulation of tamoxifen and bile levels of 4-OH-tamoxifen.
In vivo evaluation of model predictions. We performed a yearlong exposure of female rainbow trout to four test chemicals: trenbolone, tamoxifen, prochloraz and fluoxetine. This experiment was designed to provide in vivo experimental data for comparison with the in vitro test data and model predicted effects on reproduction. We fed the trout a feed ration of 1% estimated body weight per day, which encouraged the large weight gain in all fish. Most trout ovulated after 52-60 weeks. A notable exception was trenbolone exposed trout, which did not spawn. The goandosomatic index was significantly less in the tamoxifen and trenbolone treatments. In trout that ovulated, there did not appear to be any significant differences in fecundity or fertility. The primary effect of exposure was a decrease in egg mass or diameter, which for some chemicals such as tamoxifen, exposure caused a 30% reduction in oocyte diameter and mass at spawning. This is consistent with the HPOL and PBTK model predictions that indicated sufficient 4-OH-tamoxifen is formed during tamoxifen exposures to interfere with hepatic synthesis of Vtg and reduce oocyte growth rates.
Conclusions:
-
The development of adverse outcome pathway (AOP) models requires knowledge of dose-dependent perturbations across varying biological scales (cells→organs→systems→ organism→populations). In vitro, cell-based assays are expected to provide the bulk of experimental data for future AOP models while computational or quantitative physiological models are used to extrapolate across biological scales. Our hypothesis is that in vitro assays can provide valid estimates of specific model parameters used in computational models of fish reproduction. To test this, we developed a multi-scale mathematical model of the female rainbow trout hypothalamus-pituitary-ovary-liver (HPOL) axis to extrapolate results from select in vitro studies of tissues comprising the reproductive axis (Gillies et al. 2016). The model describes the essential endocrine components of the reproductive axis, including the stage specific growth of maturing oocytes and permits the presence of sub-populations of oocytes at different stages of development. This feature allows the HPOL model to accurately predict oocyte growth and the impact of toxicants that perturb physiological processes associated with oocyte growth and maturation.
In parallel to model development, we testeda series of model endocrine toxicants on isolated pituitary cells, hepatocytes and ovarian follicles. Key findings from these studies include estrogen (estradiol-17β; E2) induces the LHβ mRNA levels in pituitary cells from previtellogenic trout at concentrations as low as 0.01 ng/mL (0.037 nM). In contrast, 11-KT had no effect on LHβ mRNA levels and had weak, inconsistent effects on FSHβ mRNA levels. However, effects of 11-KT (the most important male sex hormone in trout) on FSH and LH secretion cannot be ruled out yet. These data suggest that the induction of LHβ in previtellogenic trout is specific to estrogens. Ovarian follicle studies indicated several chemicals affected estrogen synthesis. For example, a significant decrease in E2 production was seen with the pharmaceutical fluoxetine (10 µM), and incubation with the fluoxetine metabolite norfluoxetine showed similar although non-significant trends at concentrations ranging from 0.4 µM to 2.0 µM. The rate of E2 production was drastically reduced after exposure of follicles to environmentally-relevant concentrations of ethinyl estradiol at concentrations as low as 0.1 nM. The synthetic androgen trenbolone was also observed to alter vitellogenin synthesis in hepatocytes, which is in addition to other documented effects on E2 synthesis and FSH release, previously reported by the project (Schultz et al. 2013). As a demonstration of the complete AOP process, we used the HPOL model with data from trenbolone experiments to simulate effects on oocyte growth and spawning success (Gillies et al. 2016). The model simulations indicated trenbolone effects on E2 synthesis by itself are not sufficient to cause reproductive failure in trout and additional effects occurring simultaneously elsewhere in the system are needed. In other words, relying on results from a single in vitro assay (e.g. hepatocytes) would not adequately predict the observed effects of trenbolone exposure in vivo.
- We achieved the development and application of a computational model of the trout hypothalamus – pituitary – ovary – liver axis (HPOL).The development of adverse outcome pathway (AOP) models requires knowledge of dose-dependent perturbations across varying biological scales (cells→organs→systems→ organism→populations). In vitro, cell-based assays are expected to provide the bulk of experimental data for future AOP models while computational or quantitative physiological models are used to extrapolate across biological scales. Our hypothesis is that in vitro assays can provide valid estimates of specific model parameters used in computational models of fish reproduction. To test this, we developed a multi-scale mathematical model of the female rainbow trout hypothalamus-pituitary-ovary-liver (HPOL) axis to extrapolate results from select in vitro studies of tissues comprising the reproductive axis (Gillies et al. 2016). The model describes the essential endocrine components of the reproductive axis, including the stage specific growth of maturing oocytes and permits the presence of sub-populations of oocytes at different stages of development. This feature allows the HPOL model to accurately predict oocyte growth and the impact of toxicants that perturb physiological processes associated with oocyte growth and maturation.
Journal Articles on this Report : 4 Displayed | Download in RIS Format
Other project views: | All 24 publications | 6 publications in selected types | All 4 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Gillies K, Krone SM, Nagler JJ, Schultz IR. A computational model of the rainbow trout hypothalamus-pituitary-ovary-liver axis. PLoS Computational Biology 2016;20;12(4):e1004874 (27 pp.). |
R835167 (2015) R835167 (2016) R835167 (Final) |
Exit Exit Exit |
|
Jia Y, Cavileer TD, Nagler JJ. Acute hyperthermic responses of heat shock protein and estrogen receptor mRNAs in rainbow trout hepatocytes. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology 2016;201:156-161. |
R835167 (2015) R835167 (2016) R835167 (Final) |
Exit Exit Exit |
|
Robert A, Schultz IR, Hucher N, Monsinjon T, Knigge T. Toxicokinetics, disposition and metabolism of fluoxetine in crabs. Chemosphere 2017;186:958-967. |
R835167 (Final) |
Exit Exit Exit |
|
Schultz IR, Nagler JJ, Swanson P, Wunschel D, Skillman AD, Burnett V, Smith D, Barry R. Toxicokinetic, toxicodynamic, and toxicoproteomic aspects of short-term exposure to trenbolone in female fish. Toxicological Sciences 2013;136(2):413-429. |
R835167 (2013) R835167 (2014) R835167 (2015) R835167 (2016) R835167 (Final) |
Exit Exit |
Supplemental Keywords:
Recrudescence, maturation, gonadotropin, disruption, water, modeling toxicityProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.
Project Research Results
- 2016 Progress Report
- 2015 Progress Report
- 2014 Progress Report
- 2013 Progress Report
- Original Abstract
4 journal articles for this project