Sodium Perchlorate Induces Non-Alcoholic Fatty Liver Disease in Developing Stickleback

Abstract

Perchlorate is a pervasive, water-soluble contaminant that competitively inhibits the sodium/iodide symporter, reducing the available iodide for thyroid hormone synthesis. Insufficient iodide uptake can lead to hypothyroidism and metabolic syndromes. Because metabolism, obesity and non-alcoholic fatty liver disease (NAFLD) are tightly linked, we hypothesized that perchlorate would act as an obesogen and cause NAFLD via accumulation of lipids in liver of developing threespine stickleback (Gasterosteus aculeatus). We performed an upshift/downshift exposure regime (clean water to perchlorate treated water or perchlorate treated water to clean water) on stickleback embryos at two concentrations (30 mg/L and 100 mg/L) plus the control (0 mg/L) over the course of 305 days. Adult stickleback were euthanized, H&E stained and analyzed for liver morphology. Specifically, we counted the number of lipid droplets, and measured the area of each droplet and the total lipid area of a representative section of liver. We found that perchlorate treated fish had more and larger lipid droplets, and a larger percentage of lipid in their liver than control fish. These data indicate that perchlorate causes NAFLD and hepatic steatosis in stickleback at concentrations commonly found at contaminated sites. These data also indicate the potential of perchlorate to act as an obesogen. Future studies should investigate the obesogenic capacity of perchlorate by examining organ specific lipid accumulation and whether perchlorate induces these effects at concentrations commonly found in drinking water. Work is also needed to determine the mechanisms by which perchlorate induces lipid accumulation.

Graphical Abstract

Introduction

Endocrine disrupting compounds (EDCs) are chemicals known to interfere with endocrine function of animals. Recent literature provides a growing list of EDCs that induce the accumulation of lipids and cause obesity in animals including humans, a link between chemicals and obesity known as the obesogen hypothesis (Baillie-Hamilton, 2002; Grün and Blumberg, 2009; Heindel and Schug, 2013; Janesick and Blumberg, 2016). Human rates of obesity have risen sharply leading to increases in many obesity-related illnesses and associated health care spending (Hebert et al., 2013). Although increased calorie consumption and reduced exercise may play important roles in the obesity pandemic, the obesogen hypothesis may explain the recent and dramatic increases in obesity as being due in part to exposure to certain EDCs (Heindel and Schug, 2014). Many EDCs are known to affect the regulation of sex hormones, the hypothalamic-pituitary-thyroid axis (HPT), and other endocrine axes (Grün and Blumberg, 2009; Heindel and Schug, 2014). Sex steroids in conjunction with key peptide hormones, such as growth hormone, inhibit lipid accumulation in tissues (Björntorp, 1997). Thus, disruption of this system, organs heavily involved in lipid mobilization and metabolism (liver), or endocrine axes that systemically regulate metabolism (e.g., HPT) have the potential to contribute to lipid accumulation and obesity (Lonardo et al., 2006).

Dysregulation of either androgens (Sato et al., 2003) or estrogens (Cooke and Naaz, 2004) can influence lipid accumulation and result in obesity. Estrogens can be obesogenic at certain developmental time points causing fattening and high circulating leptin levels (Ruhlen et al., 2008). Xenoestrogenic chemicals such as bisphenol-A can also cause obesity by upregulating adipocyte genes (Masuno et al., 2002). Disruption of the HPT axis can lead to depression of circulating thyroxine (T₄) and tri-iodothyronine (T₃) levels and thereby produce obese phenotypes due to downstream effects on lipid metabolism (Grün and Blumberg, 2009). For example, short-term exposure to polybrominated diphenyl ethers (PBDEs) lowers T₄ levels, and thus affects lipolysis in adipocytes (Zhou et al., 2001). Hypothyroidism with associated lower circulating T₄ levels affects overall metabolic rate and causes weight gain (Garber et al., 2012).

Commonly associated with obesity is nonalcoholic fatty liver disease (NAFLD, (Braillon et al., 1985; Browning et al., 2004; Nasrallah et al., 1981)). NAFLD is associated with the accumulation of lipids in hepatic tissues at a level greater than 5 – 10% of the liver’s weight without the presence or consumption of alcohol (Bruce and Byrne, 2009). This disease represents a spectrum of disorders that range from fatty liver to nonalcoholic steatohepatitis (NASH – accumulation of lipids coupled with hepatocyte injury and inflammation) which can further progress to cirrhosis and liver failure. This disease progression is known as the two-hit model where the liver first accumulates lipids (first hit) and then experiences inflammation and scarring (second hit)(Day and James, 1998). As of 2004, approximately one third of adults in the United States were diagnosed with some form of NAFLD (Browning et al., 2004), an occurrence rate approaching the obesity rate in the United States in 2015 of nearly 40% (Spengler and Loomba, 2015).

Obesity and NAFLD are associated through genetics, lifestyle and the endocrine system (Anstee and Day, 2013; Lonardo et al., 2006). For example, thyroid function (mediation of lipid metabolism) and NAFLD are linked in humans and animal models (Lonardo et al., 2006). Overfeeding geese for foie gras is associated with both fatty liver and hypothyroidism (Janan et al., 2000). Correlative studies in humans have also found that patients with NASH were more likely to also have hypothyroidism compared to patients without any form of liver disease (Liangpunsakul and Chalasani, 2003). Sex hormones also affect lipid accumulation in the liver. Women under treatment with the estrogen receptor blocker tamoxifen have a higher risk of developing NASH (Bruno et al., 2005). Male aromatase knockout mice also develop hepatic steatosis which can be ameliorated with the addition of exogenous estrogen (Hewitt et al., 2004). These effects can also span multiple generations through epigenetic mechanisms. For example, pregnant mice exposed to tributyltin, a compound that disrupts sex hormone synthesis through inhibition of aromatase, produce offspring with fatty livers and this phenotype persists for another two generations in unexposed individuals (Chamorro-Garcia et al., 2013).

Perchlorate (ClO₄⁻) is a pervasive, water-soluble contaminant to which virtually all U.S. residents and residents of many other industrialized countries are exposed via ingestion of contaminated water and foods (EPA, 2005; FDA, 2004). As of 2009, perchlorate contamination had been detected in 45 U.S. states (GAO, 2010). It has been classified by the U.S. Environmental Protection Agency as a chemical of concern, occurring in 213 out of 285 common foods and drinks in the U.S. (Murray et al., 2008). A Centers for Disease Control national survey discovered perchlorate in urine samples of all 2,820 people tested, with the highest levels found in children. The Centers for Disease Control found that a third of American women with low iodine levels experienced reduced thyroid hormone production at perchlorate exposure levels below the Environmental Protection Agency’s 2005 “safe” threshold (Blount et al., 2007).

Perchlorate is ionically similar to iodide and competes with iodide at the sodium/iodide symporter thereby inhibiting the translocation of iodide across the basolateral membrane of thyroid follicular cells (Carr et al., 2006; Wolff, 1998). Insufficient uptake of iodide leads to hypothyroidism and to negative downstream effects on growth, development, metamorphosis and metabolism (Braverman and Cooper, 2012). Perchlorate exposure alters the development of the vertebrate thyroid leading to depleted colloid, thyrocyte hypertrophy and increased angiogenesis (Furin et al., 2015b; Patiño et al., 2003; Petersen et al., 2015), and is often associated with reduced levels of circulating T₄ and T₃ (Crane et al., 2005). Reduced levels of thyroid hormones can lead to obese phenotypes, and our research group discovered that perchlorate exposure induces lipid droplet formation around thyroid tissue in the model fish species the threespine stickleback (Gasterosteus aculeatus, hereafter “stickleback”); therefore, we determined that perchlorate is a candidate chemical obesogen (Gardell et al., 2017).

Perchlorate exposure falls within the paradigm of developmental origins of health and disease (DOHaD). Within this paradigm, early exposures to stressors may manifest later in life as specific phenotypes or diseases (Gillman, 2005). Thyroid disrupting compounds fit this paradigm (McDonald, 2002) and because perchlorate inhibits thyroid production and affects the developmental trajectories of numerous traits(Furin et al., 2015a, b; Petersen et al., 2016), it is likely that any obesogenic effects could be the result of early developmental exposures.

Perchlorate exposure also affects gonadal development (Furin et al., 2015b) and androgen levels in larval stickleback (Petersen et al., 2015). Perchlorate has been shown to alter sex ratios in zebrafish (Danio rerio) (Mukhi et al., 2007) and to masculinize stickleback (Bernhardt and von Hippel, 2008; Bernhardt et al., 2006), which may be due to perchlorate causing a reduction in primary germ cell number early in development (Petersen et al., 2016). Because perchlorate affects sex steroid levels and gonadal development, it could cause obese phenotypes through dysregulation of sex steroids or altered gonadal physiology in addition to or in concert with thyroid-mediated effects (Duarte-Guterman et al., 2014).

In this study, we investigate the effects of perchlorate exposure during development on adult liver phenotypes in stickleback. Liver is a key organ in lipid metabolism and therefore may play an important role in mediating the effects of obesogenic chemicals. Because other chemicals and disruptions to the HPT and the hypothalamic-pituitary-gonadal axis produce obese phenotypes at varying times during development (McDonald, 2002), we tested the hypothesis that perchlorate exposure over a variety of developmental time courses causes accumulation of lipids in the liver, leading to phenotypes similar to NAFLD. We assessed whether perchlorate exposed stickleback have more and larger lipid droplets and a greater proportion of lipid in the liver and other morphologies associated with NAFLD (Ludwig et al., 1980).

Methods

Fish Collection and Husbandry

Stickleback were collected from Rabbit Slough, Alaska (61.534° N, 149.266° W) on 4 June, 2008 and transported in aerated coolers to the lab at the University of Alaska Anchorage. Fish were kept in outdoor pools filled with de-chlorinated city water augmented with 3 g/L Instant Ocean©. A mass cross of eggs stripped from 40 females and sperm collected from 40 males was performed on 10 June, 2008. Embryo medium consisted of reverse osmosis purified water to which Instant Ocean© was added to 4 g/L. Contaminated (treatment) water was produced at nominal concentrations of 30 and 100 mg/L by adding the appropriate dry mass of sodium perchlorate.

Clutches from the 40 females were combined in one container in order to randomize eggs before dividing them into 38 Petri dishes (100 X 20 mm) with a previously determined water treatment (control: no perchlorate added, 30 or 100 mg/L of perchlorate). The treatments were divided among the 38 Petri dishes as follows: control = eight, 30 mg/L upshift = eight, 100 mg/L upshift = eight, 30 mg/L downshift = seven, and 100 mg/L downshift = seven. The eggs were subsequently fertilized with sperm mixed from the 40 males. Each Petri dish of embryos was then subdivided into three Petri dishes with approximately 100 embryos per dish (114 total Petri dishes; control = 24, 30 mg/L upshift = 24, 100 mg/L upshift = 24, 30 mg/L downshift = 21, and 100 mg/L downshift = 21). Embryos were incubated at 20 ± 0.5° C. For the first 10 days post-fertilization (dpf), water was changed daily and dead embryos were removed; by 10 dpf, most embryos had hatched and treatment groups from each Petri dish were transferred to 56.8 L aquaria (60 cm X 31cm X 32 cm) aerated with AZOO© multi sponge filters (65 mm diameter). All 114 aquaria started with 6 L of water because of the small size of the fish. The water level was proportionally increased as the fish grew to maintain consistent fish density across the experimental period (1 L water per 1 cm fish).

Water changes (15% of total volume) were conducted every two weeks and as needed. Two to three ml of Bacta-pur© N3000 live bacteria (IET-Aquaresearch Ltd., Quebec, Canada) were added to each tank once per week. Reverse osmosis purified water was added weekly to compensate for evaporative loss. A YSI photometer model 9100 (Yellow Springs Instrument Co., Yellow Springs, OH, USA) was used to periodically check salinity (4-5 g/L), pH (7.0-8.0), and ammonia (<2 mg/L total nitrogen); no out of range levels were detected. Perchlorate concentrations in aquaria were measured with an Acorn Ion 6 meter (Oakton Instruments, Vernon Hills, USA) with a perchlorate ion specific electrode (Cole-Parmer, Vernon Hills, IL, USA) every two weeks.

Stickleback fry (<2 months old) were fed live brine shrimp and a mixture of Golden Pearls 100 (a commercial larval food), Artemia food (both from Aquatic Ecosystems, Apopka, FL, USA), and frozen ground brine shrimp (Artemia sp.; Brine Shrimp Direct, Ogden, UT, USA). Juvenile and adult fish were fed frozen brine shrimp daily. Perchlorate intake from food sources was assumed to be negligible. Feeding behaviors and overall consumption was not recorded during the study. The photoperiod in the aquarium room was adjusted weekly to track ambient lighting conditions for Anchorage, Alaska. Mean temperature in the aquaria throughout the experiment was maintained at 13.5 ± 0.5° C.

Experimental Design

Fish were either introduced or removed from sodium perchlorate (> 98% purity, Sigma-Aldrich, St. Louis, MO, USA) contaminated (30 mg/L or 100 mg/L) water at various time-points during development with the goal of understanding when in development perchlorate exposure induces effects. Fish were either moved from control water to perchlorate-treated water (upshift) or from perchlorate-treated water to control water (downshift) at 0, 3, 7, 14, 21, 42, 154 or 305 dpf (Fig. 1). Once fish reached approximately one year of age, they were euthanized with an overdose of pH-neutral MS-222 (Argent Chemical Laboratories, Richmond, WA, USA) and processed for later analyses.

Figure 1:

Perchlorate exposure regime used for both 30 and 100 mg/L exposures at each day post fertilization (dpf). Zero dpf downshift fish were exposed during fertilization and for 15 minutes post fertilization before their transfer to clean water (Furin et al., 2015b).

Sex Determination via Genotyping

Genomic PCR was used to determine the genotypic sex following the methods of Griffiths et al. (2000). The genotypes of 192 individual stickleback were determined with the following method. Genomic DNA was extracted from caudal fins using a DNeasy Blood and Tissue Kit (Qiagen Inc., Valencia, CA) following the manufacturer’s instructions for purification of DNA from animal tissues. The PCR reactions were carried out in 25 μ1 volumes containing 10X PCR buffer (100 mM Tris-HCL, 500 mM KCl), 0.5 μM of each primer (Ga1F: CTTCTTTCCTCTCACCATACTCA and Ga1R: AGATGACGGGTTGATAAACAG), 50 μM of each deoxyribonucleotide triphosphate, 100 to 150 ng of the target DNA, 1.25 U of TaqDNA polymerase (TaKaRa Taq, Fisher Scientific), and 1.5 mM MgCl₂. Genotypes were visualized following electrophoretic separation on a 1.5% agarose gel and genotypic sex was determined by gel banding pattern.

Histology

Five fish from each aquarium were collected for histology, while other fish were used for parallel experimental end-points (Furin et al., 2015a, b). Due to the number of treatments and fish, samples were collected over six weeks when the fish were at 344-386 dpf (median = 365 dpf). Variation in age of stickleback was evenly distributed among the treatments. The abdomen of each fish was opened with a scalpel before placing the carcass in Dietrich’s solution (Kent et al., 2012). After fixing for 2-5 days, fish were preserved in 70% isopropanol. Fish were then dehydrated in a graded series of ethanol using an auto processor and then embedded in TissuePrep-2 Embedding Media (Fisher Scientific, PA, USA), coronally sectioned at 5 pm thickness and stained with hematoxalin and eosin. Stained sections were embedded with Cytoseal XYL (Richard-Allan Scientific, MA, USA) and a coverslip was added.

Imaging and Histomorphology

Sections were photographed and analyzed using a Lecia DM6 B and Leica Application Suite X (LASX) software for liver tissue morphology. Sections of liver were chosen based landmarks. Specifically, we used the most anterior lobe of the liver (unless not represented in the section) and selected representative samples based on homogeneity (how well the chosen section represents the majority of the liver). Sections were avoided if large vasculature was present (hepatic arteries and veins, although sinusoids were unavoidable). Because the liver stores both lipid and glycogen in similar ways, we distinguished lipid droplets from glycogen vacuoles through morphological differences. Glycogen vacuoles appear angular, flocculent, have soft margins whereas lipid droplets appear circular, possess sharp margins and when large, can displace the nucleus (Wolf and Wheeler, 2018). The hepatocytes appeared to be occupied with lipid droplets as opposed to glycogen vacuoles. In a representative section of each liver, we enumerated lipid droplets and quantified the area of each droplet and the total lipid area. Filters were set on the Leica software to exclude noise (minimum 25 pixels) and tissues that are not lipid droplets (blood vessels, glycogen vacuoles and interstitial space). Lipid droplets are generally round with a roundness factor (ratio) of 0.25 or greater (roundness ranges from 0 – 1 where 1 is a perfect sphere). In some cases, sinusoids were manually removed if these filters did not exclude them. For each liver section, the area of each droplet was ranked and the median droplet size was used in future analyses (representative lipid droplets shown in Fig. 2). The total lipid area was the summation of the area of all lipid droplets. All images were taken at 400X magnification and the total field of view and the total area was the same for all individuals. The liver sections were also assessed for the presence or absence of specific anatomical alterations due to lipid accumulation. Liver samples were assessed for displaced nuclei, cellular deformation, nuclear deformation, and disorganized hepatic cordons (Fig. 2) (Siddik et al., 2018; Wolf and Wheeler, 2018).

Figure 2:

Histological features in a typical stickleback liver (A) and liver displaying select anatomical alterations due to lipid accumulation (B). Blue outlines hepatocytes, green outlines the general hepatic chord structure, yellow outlines sinusoids filled with blood cells, and red outlines lipid droplets. Arrows (B) indicate hypertrophied nuclei.

Statistical Analyses

All statistical analyses were conducted with IBM SPSS statistics 24. Normality of data was determined by a Shapiro-Wilke test and a Levene’s test was used to test for homogeneity of variance (Table S1). The number of droplets was normally distributed and homogenous for all treatments but the total lipid area was transformed using the square root function to achieve normality and homogeneity across all treatment groups. The median droplet area was homogenous for all treatment groups but not normal and no transformation achieved normality. A multivariate analysis of variance (MANOVA) was used to assess the relationships between the three dependent variables (number of droplets, median droplet area and total lipid area) and two independent variables (treatment and sex). Differences in the three dependent variables between timing of exposure (dpf) within each exposure regime (upshift or downshift) and concentration (30 mg/L and 100 mg/L) were analyzed using a Kruskal-Wallis test. A Dunnett’s MCC post hoc with a Bonferoni correction was used to compare the dependent variables for each exposure timing to the pooled mean of all of the controls (n=55) for each of the two perchlorate concentrations. All analyses were considered significant when p<0.05 or lower for the Bonferoni corrected p-values. The frequency of displaced nuclei, cellular deformation, nuclear hypertrophy, and disorganized hepatic cordons were analyzed using chi-squared tests where all perchlorate treatments were pooled and compared to the frequency of each variable for the control treatment.

Results

Sex Effects

There were no effects of sex (n=192) on the number of lipid droplets, median droplet size or total lipid area (p=0.570, p=0.455 and p=0.174, respectively) and no interaction between sex and treatment (p=0.504, p=0.713 and p=0.517, respectively). Because there were no sex effects, this variable was removed from further analyses and a larger sample size (n = 222) was achieved by including individuals where sex was unknown (n=30).

Liver Histomorphology

Treatment had a significant effect on number of lipid droplets, size of the droplets and the total lipid area (p<0.001, p=0.008 and p<0.001, respectively). Larger effects were shown in the 30 mg/L treatment than in the 100 mg/L treatment indicating a possible non-monotonic dose-response curve. Due to low sample sizes, nonparametric statistics were used to test for differences in the timing of exposure. Fish reared in control water had relatively few and small lipid droplets and “normal” hepatocytes (polygonal shaped cells with centrally located nuclei) and liver architecture typical of other Gasterosteus fishes (tubular form - sinusoids surrounded by double layer of hepatocytes, Fig. 3A, B, E, and F)(Akiyoshi and Inoue, 2004). Stickleback that were chronically exposed to perchlorate (fish exposed for greater than 80% of their lives) had more lipid droplets with a greater size and greater total accumulated lipid area as compared to controls (Fig. 3C, D, G, and H).

Figure 3:

Histological images of stickleback liver. A - D display the general structure of the anterior lobe of the stickleback liver. E - H display a magnified subsection from the corresponding 100X image. The control panels display normal liver morphology with little lipid accumulation (A, B, E, and F). C, D, G, and H display perchlorate treated fish with significant lipid accumulation. Downshift (DN) fish began in contaminated water and were moved to untreated water on a given day post fertilization (dpf). Arrows (G and H) indicate hypertrophied nuclei.

In some cases, fish chronically exposed to perchlorate also showed displacement of hepatocyte nuclei from the center of the cell, cellular deformation associated with hypertrophy, hypertrophy of hepatocyte nuclei, and disorganization of typical hepatic cord structure, which are consistent with NASH, a more progressed state of NAFLD (Fig. 3D and H). Fish reared in control water rarely had these abnormalities, and the differences in frequencies of these traits between control and perchlorate-treated fish were all significant (Table 1).

Table 1:

Chi-squared results for perchlorate exposed fish based upon the frequency of occurrence for each variable in the perchlorate treated fish (all perchlorate-exposed fish pooled) vs. control fish. The table displays the percent present rather than the frequency counts.

Variable	Percent Present Perchlorate-Treated	Percent Present Control	X2	df	p
Displaced Nuclei	52.7%	23.6%	14.0	1	<0.001
Cellular Deformation	38.9%	11.0%	14.7	1	<0.001
Nuclear Hypertrophy	24.0%	5.5%	8.6	1	0.003
Disorganized Cordons	32.9%	14.5%	6.5	1	0.011

Timing of Exposure

The total lipid area of liver in perchlorate-treated fish was significantly greater than in control fish (Kruskal-Wallis test, Table 2). The Dunnett’s post hoc test indicated that total lipid area was greatest in the 30 mg/L downshift at 3 dpf and after 42 dpf. The 100 mg/L downshift showed higher lipid accumulation levels at 42 dpf (Fig. 4). Total lipid area was greater in the 30 mg/L upshift at 0 dpf and 7 dpf and in the 100 mg/L upshift at 7 dpf.

Table 2:

Kruskal-Wallis tests (p value and H statistic) of each treatment and shift. Results are compared to the control fish (no perchlorate exposure) and considered significant if p<0.05. However, after Bonferoni correction on Dunnett’s post-hoc tests, the median lipid droplet area for the 100mg/L downshift treatment was no longer significant.

	30mg/L Downshift	100mg/L Downshift	30mg/L Upshift	100mg/L Upshift
Total Lipid Area	<0.001 28.7	0.011 18.3	0.004 22.6	0.001 25.8
Number of Lipid Droplets	0.003 21.8	0.003 21.2	0.001 25.8	<0.001 30.7
Median Lipid Droplet Area	0.001 23.6	0.027 10.1	0.276 4.5	0.041 16.1

Figure 4:

Mean total lipid area of stickleback liver (+/− 1SE). Control group (n=55) is represented by the solid line (mean) surrounded by dashed lines (+/− 1SE). Downshift fish began in contaminated water and were moved to untreated water on the given day post fertilization (dpf), and Upshift fish began in untreated water and were moved to contaminated water on the given dpf. Asterisk denotes significant difference from the control group as determined by Kruskal-Wallis test with Dunnett’s post hoc with a Bonferoni corrected alpha of 0.05. Note the X-axis is not a linear scale and downshift treatments did not have individuals from day 305.

The number of lipid droplets was also significantly greater in perchlorate-treated fish than in control fish (Kruskal-Wallis, Table 2). The number of droplets was significantly greater at 3 dpf and 42 dpf in the 30 mg/L downshift and at 14 dpf in the 100 mg/L downshift (Fig. 5). Both 30 mg/L and 100 mg/L upshift treatments showed significantly more droplets in the 0 dpf and 7 dpf exposures (Fig. 5).

Figure 5:

Mean number of lipid droplets in stickleback liver (+/− 1SE). Control group (n=55) is represented by the solid line (mean) surrounded by dashed lines (+/− 1SE). Downshift fish began in contaminated water and were moved to untreated water on the given day post fertilization (dpf), and Upshift fish began in untreated water and were moved to contaminated water on the given dpf. Asterisk denotes significant difference from the control group as determined by Kruskal-Wallis test with Dunnett’s post hoc with a Bonferoni corrected alpha of 0.05. Note the X-axis is not a linear scale and downshift treatments did not have individuals from day 305.

Median lipid droplet size was significantly larger in the 30 mg/L downshift and 100 mg/L upshift exposures than in control fish, but there were no effects detected in the 30 mg/L upshift and 100 mg/L downshift exposures (Kruskal-Wallis, Table 2). The Dunnett’s post hoc test indicated significantly larger droplets at 42 dpf and 154 dpf in the 30 mg/L downshift and at 7 dpf in the 100 mg/L upshift (Fig. 6).

Figure 6:

Median lipid droplet area of stickleback liver (+/− 1SE). Control group (n=55) is represented by the solid line (mean) surrounded by dashed lines (+/− 1SE). Downshift fish began in contaminated water and were moved to untreated water on the given day post fertilization (dpf), and Upshift fish began in untreated water and were moved to contaminated water on the given dpf. Asterisk denotes significant difference from the control group as determined by Kruskal-Wallis test with Dunnett’s post hoc with a Bonferoni corrected alpha of 0.05. Note the X-axis is not a linear scale and downshift treatments did not have individuals from day 305.

Discussion

Our results indicate that perchlorate exposure at concentrations experienced in contaminated sites (Trumpolt et al., 2005) induces a spectrum of NAFLD phenotypes in the liver of developing stickleback. Here we show varying degrees of the “first-hit” of NAFLD (hepatic steatosis - fat accumulation) but we are unable to assess whether this fat accumulation affects liver function in these animals. Specifically, stickleback chronically exposed to perchlorate during development had greater total lipid content, larger lipid droplets, and a greater number of lipid droplets in the liver as compared to control fish (Figs. 4–6). In some cases (~10%), stickleback liver displayed a more severe case of NAFLD where the accumulation of lipid was coupled with marked changes to the hepatocytes - specifically, hypertrophy of hepatocytes and displacement of nuclei from the center of the cell as a result of the large lipid vacuoles (Fig. 3G and H). These liver samples more closely resemble NASH, which is a further progression of the disease that can lead to fibrosis, cirrhosis and liver failure (Browning et al., 2004). The onset of NASH is coupled with Kupffer cells secreting inflammatory cytokines (Day and James, 1998). We cannot speculate beyond morphology if perchlorate-exposed livers progressed into NASH because we did not measure these cytokines. None of the liver samples investigated displayed any further progression towards fibrosis or cirrhosis, which are pathogenic states that compromise the architecture of the liver. Stickleback liver is generally tubular with a double lining of hepatocytes surrounded by sinusoids (Fig. 3E and F) which is distinct from human liver histology (Akiyoshi and Inoue, 2004).

The mechanisms driving lipid accumulation in the liver of perchlorate-exposed fish are unknown. Perchlorate caused the accumulation of lipids in the liver of both sexes indicating that the pathway causing this accumulation may not be sex hormone mediated, but more likely is mediated via disruption of the HPT axis. However, substantial crosstalk occurs between sex hormone synthesis and the HPT axis (Duarte-Guterman et al., 2014) and we did not collect the data necessary to disentangle these potential mechanisms. Fish exposed to perchlorate in our study also showed marked alterations of thyroid morphology, but did not show concomitant changes in total thyroid hormone content (Furin et al., 2015b).

That the 30mg/L treatment fish had more lipid in their livers than did the 100mg/L treatment fish suggests a non-monotonic dose response of perchlorate on liver lipid accumulation. This finding is consistent with other endocrine disrupting studies and non-monotonicity is particularly common when the chemical has multiple mechanisms of action (Vandenberg et al., 2012). Because perchlorate can affect the HTP axis (Carr et al., 2006; Wolff, 1998) as well as gonadal development (Bernhardt et al., 2006; Furin et al., 2015b; Petersen et al., 2015; Petersen et al., 2016), we speculate that the observed non-monotonicity may result from cross-talk of these systems. Future work should expand the range of concentrations, especially to lower dose exposures (Vandenberg, 2014).

The effects of developmental timing of perchlorate exposure on lipid accumulation in stickleback liver indicate that chronic exposures cause the greatest lipid accumulation (earlier time points in the upshifts and later time points in the downshifts; Figs. 4–6). However, it is likely that any exposure at the tested concentrations of perchlorate can yield NAFLD phenotypes in stickleback liver because long lasting effects were shown in the 3 dpf and 14 dpf downshift exposures (fish were only exposed for the first 3 and 14 days, respectively) and these effects were not ameliorated by exposure to control water from these time-points until one year of age. Therefore, the liver appears to be even more sensitive to early developmental exposures to perchlorate than the thyroid (Furin et al., 2015b). It is possible that perchlorate may operate directly on the liver and thus is not thyroidally mediated, but this has not been explored. Our observed effects of exposure to perchlorate during early development on lipid accumulation in the liver fit the paradigm of “Developmental Origins of Health and Disease” (DOHaD) whereby early developmental exposures to contaminants and other stressors lead to later-onset diseases (Gillman, 2005). These results are similar to findings of other studies where early-life exposures to thyroid disrupting compounds (e.g., PBDEs (McDonald, 2002)) or other EDCs cause obese phenotypes in adults (Newbold et al., 2007; Zhou et al., 2001). Although our results fit the paradigm of DOHaD, we have no data on the possible mechanism by which exposure as an embryo could lead to lipid accumulation in adult stickleback or why some exposures (e.g., 21 dpf for any shift) did not lead to lipid accumulation.

Our data suggest that perchlorate exposure can cause NAFLD in stickleback, and we have previously shown that perchlorate exposure induces lipid accumulation in stickleback thyroid tissues (Gardell et al., 2017). Are there other tissues or organs that accumulate lipid in response to perchlorate exposure? Substantial overlap exists between the endocrine system, metabolism and the development of obesity and NAFLD (Lonardo et al., 2006) and it is likely that perchlorate would induce obesogenic effects systemically, but effects on other tissues and in other model animals have not been investigated. Further study is needed to determine whether perchlorate causes whole-body accumulation of lipids rather than a redistribution of lipids between organs and to determine organ-specific effects beyond the liver and thyroid potentially indicating disease states beyond NAFLD and obesity. More sophisticated methods such as gas chromatography or lipid extractions could also be used to assess whole body lipid content or total liver lipid content.

Investigation of perchlorate exposure and lipid accumulation using additional animal models is necessary to determine if lipid accumulation in the liver and thyroid from perchlorate exposure is specific to stickleback, or general to vertebrates, and therefore likely to also occur in humans. Although the concentrations studied here are relevant to contaminated sites (Trumpolt et al., 2005), it is unknown how lower concentrations such as those commonly encountered in drinking water (low parts per billion range) would affect liver morphology. Our finding that the 30 mg/L exposed fish had greater lipid accumulation than the 100 mg/L exposed fish suggests a non-monotonic dose response; thus, it is possible that at lower concentration exposures effects might be even greater. Therefore, future work should examine whether perchlorate has obesogenic effects at concentrations commonly found in drinking water, milk and leafy vegetables in order to determine if perchlorate may play a role in the on-going obesity epidemic. Lastly, because NAFLD is a spectrum disorder and is caused by a multitude of factors, the mechanisms that underlie effects caused by perchlorate exposure need to be elucidated.

Conclusions

This study examines the sensitivity of the developing stickleback liver to varying concentrations and exposure windows of sodium perchlorate. Perchlorate exposure increased the number and size of lipid droplets leading to an overall increase of lipid in the adult stickleback liver. Exposed stickleback also displayed a spectrum of NAFLD phenotypes that ranged from mild (low level hepatic steatosis) to extreme modifications to liver tissue (displaced nuclei, cellular deformation, nuclear hypertrophy and disorganized hepatic cordons) resembling NASH. These data also indicate that the developing stickleback liver is highly sensitive to perchlorate exposure as effects were often not ameliorated after returning fish to control water even at a young age. To our knowledge, these data are the first to show the role of perchlorate exposure on liver development and lipid metabolism. Further studies are needed to elucidate the role of perchlorate and mechanism of action as a possible obesogenic chemical.

Key Finding.

Sodium perchlorate induces non-alcoholic fatty liver disease in developing stickleback through the accumulation of lipids in hepatocytes.

Research Highlights.

1. Perchlorate induces non-alcoholic fatty liver disease in developing stickleback.

2. Perchlorate exposure increases lipid accumulation in stickleback liver.

3. Stickleback liver appears to be more sensitive to perchlorate than thyroid tissue.