The in Vitro Estrogenic Activities of Polyfluorinated Iodine Alkanes

Background: Polyfluorinated iodine alkanes (PFIs) are important intermediates in the synthesis of organic fluoride products. Recently, PFIs have been detected in fluoropolymers as residual raw materials, as well as in the ambient environment. Objectives: High production volumes and potential environmental releases of PFIs might become a concern, but the exposure risk and toxicity of these chemicals are still unclear. In this study, we investigated the potential estrogenic effects of PFIs. Methods: We studied the estrogenic effects of fluorinated iodine alkanes (FIAs), fluorinated telomer iodides (FTIs), and fluorinated diiodine alkanes (FDIAs) using the E-screen and MVLN assays and the evaluation of estrogen-responsive genes in MCF-7 cells. Results: FIAs have an iodine atom at one end of the perfluorinated carbon chain. 1-Iodoperfluorohexane (PFHxI) and 1-iodoperfluorooctane (PFOI) promoted the proliferation of MCF-7 cells, induced luciferase activity in MVLN cells, and up-regulated the expression of TFF1 and EGR3. In these assays, other FIAs gave negative responses. FDIAs have an iodine atom at each end of the perfluorinated carbon chain, and all the FDIAs showed estrogenic effects. The estrogenic potencies of FIAs and FDIAs correlate well with the carbon chain length of the chemicals. The optimum chain length for estrogenic effects is six carbons, and then eight and four carbons. All FTIs have a single iodine atom at the end of a partially fluorinated carbon chain. None of the FTIs showed estrogenic effects in the tests. Conclusions: The estrogenic effects of PFIs are dependent on the structural features of iodine substitution and chain length. This research will be helpful in further understanding the estrogenic effects of perfluorinated compounds.


Research
Perfluorinated chemicals (PFCs) have a broad range of applications in the manufacture of various industrial and commercial products, such as fluoro polymers, surfactants, emul sifiers, and non stick coatings. PFCs have been of considerable scientific and public concern because some of them are environ mentally persistent, bio accumula tive, and widely detected in humans, wildlife, and the environ ment (Giesy and Kannan 2001;Olsen et al. 2007).
Because of environmental concerns, the 3M company voluntarily phased out electro chemical fluorinationbased fluoro chemicals in 2001 (Dupont 2005). Consequently, the current production of fluorinated poly mers and surfactant is mostly based on telomerization processes (Lehmler 2005). Polyfluorinated iodine alkanes (PFIs) are organic iodides composed of a fluorinated carbon backbone terminated by iodine substi tution (Table 1) and are important interme diates in the synthesis of various fluorinated chemicals (Brace 1999;Prevedouros et al. 2006). In the telomerization process, PFIs are used to synthesize fluoro telomer alcohols (FTOHs) and other related PFCs. In turn, FTOHs are intermediates in the production of surfactants and fluoro polymers, and these volatile compounds have been detected in the atmosphere around the world (Ellis et al. 2004). The annual production of FTOHs increased to 11-13 × 10 3 metric tons in 2002 (Ellis et al. 2003). The annual world production of PFIs has been estimated to exceed 4,000 metric tons (Organisation for Economic Cooperation and Development 2004), and the increasing demand of fluoro telomer products might increase the risk of emission of volatile PFIs to the environment (Ruan et al. 2010a(Ruan et al. , 2010b. Fluorinated iodine alkanes (FIAs) and fluorinated telomer iodides (FTIs) have been detected in air and soil samples around a fluoro chemical manu facturing plant in Shandong province in northern China (Ruan et al. 2010a). Residual FIAs and FTIs could also be incorporated into FTOH containing raw materials and fluoro telomerbased products during manu facturing. 1Iodoperfluoro octane (PFOI) and 6:2 FTI have been detected in fluoro telomer raw materials and selected fluoro telomer based products, such as urethane polymer and phosphate surfactant (Larsen et al. 2006). Furthermore, unreacted residual FTOH has also been found in commercial and industrial products and could be released to the ambient environment as well (DinglasanPanlilio and Mabury 2006;Larsen et al. 2006). Likewise, residual PFIs in fluorinated polymers and surfactants can be released into the environ ment and degrade to other persistent PFCs. Abiotic or biotic transformation of FTIs could contribute to the environmental bur den of FTOHs and perfluoro carboxylic acids (PFCAs) (Young et al. 2008). There is there fore a potential risk for release of PFIs to the environment due to direct emission during manufacturing and indirect emission from some fluorinated products.
Increasing evidence has shown that some PFCs may have endocrinedisrupting potency. Some PFCs can disturb the thyroid system and neuro endocrine function, activate both peroxisome proliferatoractivated receptors and estrogen receptors (ERs), and induce develop mental toxicity in rodents (Lau et al. 2004). The estrogenic effects of some PFCs have been studied in many aspects. For example, Maras et al. (2006) demon strated the estrogenlike properties of FTOHs in MCF7 cells. Using yeast twohybrid assays, Ishibashi et al. (2007Ishibashi et al. ( , 2008 demonstrated that FTOHs can activate the male medaka (Oryzias latipes) and human ER. Liu et al. (2007) reported that vitello genin expression was induced by perfluoro octane sulfonate (PFOS), perfluoro octanoic acid (PFOA), and FTOHs in primary cultured tilapia hepato cytes, and they suggested that estrogenic effects may be mediated through the ER pathway. FTOHs also induced vitello genin in male medaka fish through the activa tion of ER, whereas PFOS and PFOA did not (Ishibashi et al. 2008).
Little information is currently available regarding the estrogenic effects of PFIs. In the present study, we used three in vitro bio assays-Escreen assay, MVLN assay, and evaluation of an estrogenresponsive gene-to comprehensively evaluate the estrogenic potencies of PFIs. The structural features responsible for estrogenic effects were identi fied by the alternations in potency derived from specific structural changes.
Background: Polyfluorinated iodine alkanes (PFIs) are important intermediates in the synthesis of organic fluoride products. Recently, PFIs have been detected in fluoropolymers as residual raw materials, as well as in the ambient environment. oBjectives: High production volumes and potential environmental releases of PFIs might become a concern, but the exposure risk and toxicity of these chemicals are still unclear. In this study, we investigated the potential estrogenic effects of PFIs. Methods: We studied the estrogenic effects of fluorinated iodine alkanes (FIAs), fluorinated telomer iodides (FTIs), and fluorinated diiodine alkanes (FDIAs) using the E-screen and MVLN assays and the evaluation of estrogen-responsive genes in MCF-7 cells.
Cell culture. Human MCF7BUS breast adeno carcinoma cells and MVLN cells were cultured in 100mm culture dishes in a humidified atmosphere of 5% CO 2 at 37°C. Cells were maintained in Dulbecco's modified Eagle's medium (DMEM)/F12 (Hyclone, Logan, UT, USA) containing 10% fetal bovine serum, 100 U/mL streptomycin penicillin, 2 mM lglutamine, and 1% insu lintransferrinselenium supplement (all from Gibco, Grand Island, NY, USA). E-screen assay. MCF7BUS cells were kindly provided by A.M. Soto and C. Sonnenschein (Tufts University School of Medicine, Boston, MA, USA). In response to ERα agonists, the mitotic effect leads to the proliferation of MCF7BUS cells. We performed the Escreen assay following a method modified from the protocol by Soto et al. (1995). Cells were trypsinized and plated into the interior 60 wells of 96well plates at the density of 3,000 cells/well. Before each experiment, cells were starved in steroidfree (SF) medium for 48 hr to minimize the basal hormonal activity during assays. SF medium consisted of phenol redfree DMEM/F12 (Hyclone) supplemented with 5% dextrancharcoaltreated fetal bovine serum (Hyclone), 100 U/mL streptomycin penicillin, and 2 mM lglutamine. Cells were treated with serial dilutions of test chemi cals (from 1 nM to 100 μM) in SF medium; a concentration range of 0.01-200 pM E 2 was used as the positive control. We used a WST1 proliferation kit (Roche Diagnostics, Mannheim, Germany) to assess proliferation after 6 days of exposure according to the kit instructions. The WST1 assay is based on the enzymatic cleavage of the tetrazolium salt WST1 to formazan by cellular mitochon drial dehydrogenases present in viable cells. The absorbance of the WST1 solution was detected by a microplate reader (Varioskan Flash, Thermo Fisher Scientific, Waltham, MA, USA) at 450 nm, with the reference wave length at 690 nm. The cell prolifera tion effect was calculated from the solvent con trol (0.1% ethanol)corrected absorbance and expressed as the percentage of maximal absor bance of the positive control. Three replicates were used in each experiment.
MVLN assay. The MVLN cell line was kindly provided by J.P. Giesy (Michigan State University, East Lansing, MI, USA). This cell line was stably transfected with the luciferase reporter gene and estrogenresponsive ele ment derived from the Xenopus vitellogenin A2 gene. ER agonists can induce the pro duction of luciferase in MVLN cells (Pons et al. 1990). Cells were seeded in the inte rior 60 wells of a 96well ViewPlate (Packard Instrument Company, Boston, MA, USA) at a density of 7 × 10 4 cells/well, starved in SF medium for 48 hr, and exposed to test compounds for 2 days. A concentration range of 0.5 pM-1 nM E 2 was used as a positive control, whereas the exposure concentration range of test chemicals was 0.1-100 μM. Luciferase activity was measured with the LucLite kit (Packard Instruments) according to the manufacturer's protocol. We meas ured luminescence by microplate reader (Varioskan Flash) and integrated the luminescence signal for 10 sec. Total protein content was meas ured by the Bradford assay (Tiangen, Beijing, China) to normalize luminescent units. The results are given as relative luminescent unit per microgram protein. The maximal induc tion of positive control (corrected for sol vent control, 0.1-0.2% ethanol) was set as 100%, and the responses of other chemicals were converted to a percentage of the maxi mum level. Three replicates were used in each experi ment. The cytotoxicity of tested chemi cals was examined by WST1 kit in parallel and routinely observed under microscope to identify the exposure concentration range.

RNA isolation and reverse transcription polymerase chain reaction (RT-PCR).
MCF7BUS cells were seeded in sixwell plates at the density of 1.0 × 10 6 cells per well, starved in SF medium for 48 hr, and exposed to test compounds for 48 hr. First, cells were rinsed twice with cold phosphatebuffered saline, and total RNA was isolated using Trizol reagent (Invitrogen Inc., Carlsbad, CA, USA) following the manufacturer's protocol. The 260 nm and 280 nm absorbance reading of total RNA was performed using a Nanodrop spectrophotometer (Thermo Scientific, Waltham, MA, USA). The concentration of Table 1. The structures of tested chemicals.
Statistical analysis. All results are expressed as the mean ± SD. For statistical analysis, we used oneway analysis of variance (ANOVA) and Tukey's multiple range test to assess the significance of mean differences. Difference was considered significant at a pvalue ≤ 0.05.
The concentration-response analyses were performed with fourparameter logistic curve regression analysis according to the following formula: where y is the response value, x is the log con centration of the test compound, and EC 50 is the concentration that induces half of the maxi mum proliferation effect or luciferase activity.
EC 50 values were calculated from this non linear regression model. EC 20 were calculated as where x is 20% of the maximum effects, and the Hill slope and EC 50 were calculated from Equation 1.
All statistical analyses were performed using Sigma Plot (version 10.0; Systat Software Inc., San Jose, CA, USA).

Stimulation of MCF-7 cell proliferation.
We used the Escreen assay to investi gate the estrogenic activities of 12 PFIs in MCF7 cells. FIAs are mono iodized fluori nated alkanes with evennumbered chains that have 4-12 carbons. PFBI [4 carbons in its alkyl chain (C4)], PFDI (C10), and PFDoI (C12) did not show proliferation effects within the concentration ranges, whereas PFHxI (C6) and PFOI (C8) pro duced full concentration-response curves compared with E 2 ( Figure 1A); EC 50 values were 0.63 μM and 1.15 μM for PFHxI and PFOI, respectively ( Table 2). The prolifera tion effects appeared to be dependent on the chain lengths of FIAs. Fluorinated diiodine alkanes (FDIAs) have evennumbered chains with 4 to 8 carbons. All FDIAs produced full concentration-response curves in the Escreen assay ( Figure 1B). The relative proliferation effects were in the following order: PFHxDI (C6) > PFODI (C8) > PFBDI (C4). Likewise, prolifera tion potency also seems to be related to the specific carbon chain length of FDIAs. The EC 50 values of PFBDI (1.45 μM), PFHxDI (7.5 nM), and PFODI (43.3 nM) were much lower than those of the corresponding FIAs with the same chain length. The order of their relative proliferation potencies was PFHxDI > PFODI > PFHxI

Concentration (M) Concentration (M)
ND, not detected. a Percentage of the maximum proliferation effects of tested compounds to that of E 2 . b Ratio of EC 50 of E 2 to that of test compounds. c Percentage of the maximum induction effects of tested compounds to that of E 2 . d The maximum relative proliferation effect was < 7%; the absorbance ratio of E 2 (100 pM) to control was 2.1 ± 0.2. e The maximum relative luciferase activity was < 5%; the relative luminescent unit ratio of E 2 (1 nM) to control was 13.5 ± 2.4.
volume 120 | number 1 | January 2012 • Environmental Health Perspectives > PFOI > PFBDI. These compounds are thus considered to behave like xeno estrogens in the Escreen assay. Transactivation in MVLN cells. The MVLN assay has been widely used to study ER activity of test compounds (Freyberger and Schmuck 2005). In the present study, we used the MVLN assay to further inves tigate ER activity and estrogenic potency of PFIs. Before the MVLN assay, we tested the cytotoxic effects of each compound using the WST1 assay. Exposure of MVLN cells to FIAs, FTIs, or FDIAs did not produce sig nificant cyto toxicity within the concentra tion ranges, and no cytotoxic effects were observed by microscopic examination (data not shown).
Because these compounds did not show maximum induction compared with E 2 , rela tive potency based on the EC 20 would be more reliable than that derived from the EC 50 (Villeneuve et al. 2000). The estrogenic effects of PFIs revealed by the MVLN assay were in accordance with the results of the Escreen assay. As shown in Figure 2A, the induction of luciferase activity by PFBI, PFDI, and PFDoI were at the basal level (< 5%), whereas PFHxI and PFOI induced luciferase activity in a doserelated manner. PFHxI (EC 20 = 14.1 μM) showed higher estrogenic activity than did PFOI (EC 20 = 20.4 μM) in MVLN cells, with maximum induction val ues of 47% and 25%, respectively (Table 2). Luciferase activity induced by FDIAs seems to be related to the specific carbon chain length ( Figure 2B). PFHxDI (EC 20 = 0.38 μM) showed stronger estrogenic potency than did PFODI (EC 20 = 1.07 μM) and PFBDI (EC 20 = 13.8 μM), with the maximum induc tion values of 73%, 38%, and 21%, respec tively. Because the difference of EC 50 or EC 20 values among PFHxI, PFOI, and PFBDI were small, we compared the estrogenic potency with the maximum induction value in an MVLN assay. The order of estrogenic potency was PFHxDI > PFHxI > PFODI > PFOI > PFBDI, which is comparable to results from the Escreen assay. Similarly, FDIAs possessed stronger estrogenic potency than did FIAs in the MVLN assay (PFBDI > PFBI; PFHxDI > PFHxI; PFODI > PFOI), which indicated that iodine substitution at the end of a fluorinated chain may enhance the estrogenic potency of FIAs. The optimum chain length for estrogenic activity was six carbons for FIAs and for FDIAs in both of these estrogenscreening assays.
Comparison of PFCs with similar structures. FTIs are partially fluorinated alkyl iodides, which are produced by the ethylation of FIAs in telomerization processes. Compared with FIAs and FDIAs, FTIs with various chain lengths did not show estrogenic effects in the Escreen or MVLN assays within the tested concentration ranges (0.01-200 μM). We used a nonfluorinated hydrocarbon, 1iodo hexane (C6), as the control to study the effects of fluorination on estrogenic effects. Three eight carbon PFCs-PFOA, PFOC, and PFOBthat contain no iodine substitution on the car bon chain were used as compari sons to study the effects of iodine substitution on estrogenic effects. As we suspected, 1iodohexane, PFOA, PFOC, and PFOB showed negative results in the estrogenscreening assays (Table 2). These results further emphasize that a perfluo rinated alkyl chain and iodine substitution are important structural features for the estrogenic effects of PFIs.
Coexposure assay with OHT in MVLN assay. We used OHT, a strong estrogen antago nist in the mammary gland, to block the ER in the MVLN assay. OHT was co exposed with PFHxI, PFOI, PFBDI, PFHxDI, or PFODI. We used the highest induction con centrations obtained from MVLN assay in the coexposure experi ments and the gene expres sion assay. As shown in Figure 3, co exposure of OHT with the tested chemicals resulted in marked reduction of luciferase activity, which further confirmed that these xenoestrogens can activate the ER.
Expression of estrogen-responsive genes. After MCF7 cells were exposed to a series of PFCs for 48 hr, the expression levels of two estrogenresponsive genes (EGR3 and TFF1) were analyzed by realtime PCR. The TFF1 gene is involved in cell proliferation and also serves as a biomarker gene responding to estrogens (Brown et al. 1984;Jorgensen et al. 2000). As one of the ERmediated estrogen inducing genes, EGR3 belongs to the early growth response family and plays an impor tant role in the estrogendependent induc tion of the immune evasion system ). The expression levels of EGR3 and TFF1 are upregulated by natural and synthetic estrogens in MCF7 cells (Terasaka et al. 2004). PFHxI, PFOI, PFBDI, PFHxDI, and PFODI, which showed estrogenic effects in the Escreen and MVLN assays, signifi cantly upregulated the estrogenresponsive genes by 4.4, 2.7, 2.7, 5.7, 8.5, and 7.7fold for TFF1 and by 2.4, 3.6, 2.4,

Concentration (M) Concentration (M)
E 2 PFBI PFHxI PFOI PFDI PFDoI E 2 PFBDI PFHxDI PFODI -13 -12 -11 -10 -7 -9 -6 -5 -4 -3 Figure 3. Coexposure effects of the tested chemicals with OHT in the MVLN assay. The ER antagonist OHT (10 nM) was coexposed with 1 nM E 2 , 100 μM PFHxI, 100 μM PFOI, 100 μM PFBDI, 20 μM PFHxDI, or 40 μM PFODI. The relative luciferase activities are expressed as the mean ± SD of triplicate measurements in one representative experiment. 9.1, and 11.2fold for EGR3 ( Figure 4A). PFOC, PFOB, 1iodohexane, PFBI, PFDI, PFDoI, and FTIs did not affect the expres sion of EGR3 or TFF1, but PFOA slightly upregulated TFF1 by 1.57fold. The levels of EGR3 and TFF1 mRNA were greatly ele vated by 8.7 and 9.2fold upon exposure to E 2 ( Figure 4B). Therefore, PFHxI, PFOI, PFBDI, PFHxDI, and PFODI showed estro genic activity in these assays. These xeno estrogens activated ER, which was followed by increased expression of the estrogen responsive genes. PFBDI showed weaker estrogenic potency than did PFHxDI or PFODI in the Escreen and MVLN assays, whereas the expression of EGR3 and TFF1 induced by PFHxDI and PFODI was much higher than that induced by PFBDI. The upregulation of EGR3 and TFF1 was also greater for FDIAs compared with the mono iodized FIAs. This adds further evidence that an increase in iodine substitution at the end of the fluorinated chain can enhance the estrogenic potencies of FIAs.

Discussion
The endocrinedisrupting effects elicited by industrial chemicals have been of extensive concern (Colborn et al. 1993). Exposure to xeno estrogens may lower sperm count and male fertility and increase the incidence of breast and testicular cancer in humans (Toppari et al. 1996). Most of the adverse effects of these compounds are thought to be mediated through ER activation. Although the environmental behaviors of PFIs are not known, these volatile and highproduction volume chemicals could be released into the ambient environment during production, storage, and transport. The atmospheric oxi dation of PFIs may contribute to the increased levels of other PFCs in the environment. Studies of the potential toxicities of PFIs are therefore needed for health risk evaluation.
In this study, we investigated the estrogenic effects of PFIs by the Escreen and MVLN assays and the expression of estrogenrespon sive genes. Our results showed that PFHxI, PFOI, PFBDI, PFHxDI, and PFODI exert estrogenic effects through activation of the ER. The relative estrogenic potencies obtained from the Escreen and MVLN assays are both related to the specific carbon chain length of FIAs and FDIAs. The optimum chain length for estrogenic effects is six carbons, and iodine substitution on the perfluorinated chain was crucial for the estrogenic effects. Those potent compounds were able to fully stimulate cell proliferation of MCF7 cells, but this was not the case for the induction of reporter gene expression in MVLN cells. This discrepancy might be due to the difference of initial seed ing density, exposure time, and sensitivity between the two assays. The expression of the estrogenresponsive gene by these PFIs further confirmed the results. The estrogenic poten cies of FDIAs were higher than that of the FIAs, indicating that the increasing number of iodine substitutions on FIAs renders the chemical more potent in inducing estrogenic activity. PFHxDI (C6), with two iodine substitutions (one at each end) of the per fluorinated chain, showed the highest potency among the PFIs. Considerable evidence has indicated that chain length determines the biological effect of PFCs (Hu et al. 2002;Liao et al. 2009;Upham et al. 1998). Bioconcentration and bioaccumulation of PFCs are related to the length of the fluorinated chain in different species . Cytotoxic end points of PFCs such as in vitro cytotoxic effects, the alteration of cell membrane poten tial, and cytosolic pH are directly related to perfluorinated chain length (Kleszczynski and Skladanowski 2009). The inhibition of per fluorinated fatty acids on gap junction inter cellular communication also depends on chain length; shorter PFCs, including perfluoro butane sulfonate and perfluoro hexane sulfonate, did not show effects, whereas PFOS signifi cantly inhibits gap junction inter cellular com munication (Hu et al. 2002;Upham et al. 1998). The interference of PFCs on cultured rat hippocampal neurons was also related to the carbon chain length and functional groups (Liao et al. 2009). Our findings sug gest that FDIAs and some of the FIAs exert estrogen effects through the activation of ER. Because the solubility of non polar FIAs in culture media decreased with increasing chain length, the lack of estrogenic effects for PFDI and PFDoI might be attributed, in part, to decreased solubility and bioavailability of longchain FIAs.
We used the non fluorinated organic iodide 1iodohexane to study the effect of fluori na tion on estrogenic effects compared with PFHxI. In the screening assays, 1iodo hexane did not exert estrogenic effects, indi cating that fluorination is an important structural feature for estrogenic activity. The hydrophobic property of the fluorinated chain imparts the proteino philic and lipo philic property of PFCs and results in the inter action of PFCs with multiple biological molecular targets in various species. PFOA did not show proliferation effects in MCF7 cells, as previously reported by Maras et al. (2006). In the present study, we found that PFOA, PFOB, and PFOC also lack estrogenic effects. By comparing the structure-activity relationship between these PFCs, we propose that the iodine substitution is a key attribute for the estrogenic effect. The estrogenic effect was also lower for mono iodized fluorinated alkanes than for diiodized fluorinated alkanes, which further supports our assumption.
FTOHs exert estrogenic activity in MCF7 cells and aquatic organisms (Ishibashi et al. 2008;Maras et al. 2006). FTOHs behave as estrogens because of the similarity of their chemical structure and properties to other xeno estrogens, such as 4nonyl phenol. In the telomerization processes, fluoro telomer iodides Figure 4. Effects of tested chemicals on mRNA expression of estrogen-responsive genes TFF1 (pS2) and EGR3 in MCF-7 cells. (A) Cells exposed to 0.1% ethanol (control), 50 μM PFBI, 50 μM PFHxI, 50 μM PFOI, 40 μM PFDI, 40 μM PFDoI, 50 μM PFBDI, 20 μM PFHxDI, 40 μM PFODI, or 100 pM E 2 for 48 hr. (B) Cells exposed to 0.1% ethanol, 50 μM PFOA, 50 μM PFOC, 50 μM PFOB, 50 μM 1-iodohexane, 50 μM 4:2 FTI, 50 μM 6:2 FTI, 40 μM 8:2 FTI, 40 μM 10:2 FTI, or 100 pM E 2 for 48 hr. Results are expressed as the mean ± SD of triplicate measurements in one representative experiment. *p < 0.05, and **p < 0.01, compared with the control. ANOVA and Tukey's multiple range test were used to assess the significance of mean differences.  PFHxDI PFODI E 2 E 2 PFHxI volume 120 | number 1 | January 2012 • Environmental Health Perspectives are oxidized to produce FTOHs. Compared with FTOHs, none of the FTIs induced cell proliferation, which indicated that the hydroxyl group is more important for the estrogenic effects than is iodine substitution in partially fluorinated chemicals. Some of the PFIs activated the ER and induced luciferase activity in MVLN cells. However, it is ques tionable whether these PFIs are able to directly bind to and activate the ER. Structural fea tures such as a phenol ring and a hydrophobic group attached para to the hydroxyl group are essential for the estrogenic effects (Blair et al. 2000;Laws et al. 2006;Suzuki and Shapiro 2007). Furthermore, hydroxylated analogs of poly brominated diphenyl ethers and poly chlorinated biphenyls have been shown to exert estrogenic effects (Bergeron et al. 1994;Fielden et al. 1997;Meerts et al. 2001). Therefore, it may be reasonable to expect that hydroxylated forms of FIAs and FDIAs could also be estrogenic.
Compared with PFOI, both PFOB and PFOC showed no estrogenic activity. It is likely that bond strength also determines their reactivity. The strength of the bonds is C-F (467 kJ/mol) > C-H (453 kJ/mol) > C-Br (290 kJ/mol) > C-I (228 kJ/mol). Among the four halogens, fluorine is the most electro negative and iodine the least. The polariza tion of the C-I bond is lower than that of the C-H bond and the other carbon-halogen bonds. Because iodine is a good leaving atom and because of the chemical reactivity of the C-I bond, it would be easier for PFIs to be converted to their hydroxylated analogs during the exposure studies. Oxidation of FIAs can result in the formation of PFCAs (Lehmler 2005). In this reaction, C n F 2n+1 OH is thought to have been formed by the cleav age of C-I bonds in FIAs and addition of OH (Yamamoto et al. 2007). We hypothesize that C n F 2n+1 I is hydrolyzed to C n F 2n+1 OH in the culture media or inside the cells, and the deg radation products or the metabolites of PFIs are the possible targets for ER, thereby exert ing estrogenic activity. However, the under lying mechanisms for the estrogenic effects of PFIs have not been completely clarified, and further studies are also warranted to charac terize possible catabolites of PFIs, which might also exhibit estrogenic activity.
The main functions of hormones are to maintain homeo stasis and regulate reproduc tion and development. Exposure to endocrine disrupting chemicals may cause adverse effects to the organs and glands that secrete hormones, further resulting in endocrine toxicity such as impaired reproduction and develop ment. PFIs are volatile chemicals and have been detected around fluoro chemical manufacturing areas (Ruan et al. 2010a). As important precursors for the synthesis of organic fluoride products, PFIs could be incorporated into fluoro telomer raw materials and fluorotelomerbased products as residues (Larsen et al. 2006). Occupational and indoor environments might be exposure risk zones, and inhalation could be a possible exposure route.

Conclusion
Some PFIs could act on ERs and potentially cause detrimental effects on reproductive and develop mental systems. To our knowl edge, this is the first study to find estrogenic activity of PFIs using three in vitro methods. Considering the current large and increasing production volume of telomerizationbased PFCs, more extensive studies should be con ducted on the environmental distribution and toxicological effects of PFIs.