Sex-dependent effects of preconception exposure to arsenite on gene transcription in parental germ cells and on transcriptomic profiles and diabetic phenotype of offspring

Abstract

Chronic exposure to inorganic arsenic (iAs) has been linked to diabetic phenotypes in both humans and mice. However, diabetogenic effects of iAs exposure during specific developmental windows have never been systematically studied. We have previously shown that in mice, combined preconception and in utero exposures to iAs resulted in impaired glucose homeostasis in male offspring. The goal of the present study was to determine if preconception exposure alone can contribute to this outcome. We have examined metabolic phenotypes in male and female offspring from dams and sires that were exposed to iAs in drinking water (0 or 200 ug As/L) for ten weeks prior to mating. The effects of iAs exposure on gene expression profiles in parental germ cells, and pancreatic islets and livers from offspring were assessed using RNA sequencing. We found that iAs exposure significantly altered transcript levels of genes, including diabetes-related genes, in the sperm of sires. Notably, some of the same gene transcripts and the associated pathways were also altered in the liver of the offspring. The exposure had a more subtle effect on gene expression in maternal oocytes and in pancreatic islets of the offspring. In female offspring, the preconception exposure was associated with increased adiposity, but lower blood glucose after fasting and after glucose challenge. HOMA-IR, the indicator of insulin resistance, was also lower. In contrast, the preconception exposure had no effects on blood glucose measures in male offspring. However, males from parents exposed to iAs had higher plasma insulin after glucose challenge and higher insulinogenic index than control offspring, indicating a greater requirement for insulin to maintain glucose homeostasis. Our results suggest that preconception exposure may contribute to the development of diabetic phenotype in male offspring, possibly mediated through germ cell-associated inheritance. Future research can investigate role of epigenetics in this phenomenon. The paradoxical outcomes in female offspring, suggesting a protective effect of the preconception exposure, warrant further investigation.

Introduction

Type 2 diabetes is a prevalent medical condition that poses a significant public health concern. According to the 2020 National Diabetes Statistics Report, nearly 34.2 million in the United States are affected by this chronic disease (CDC 2020). Exposure to chemicals and other environmental factors has been shown to contribute to the increasing prevalence of diabetes (Ardisson Korat et al. 2014; Thayer et al. 2012). The etiology of type 2 diabetes is complex and can elicit multiple factors, including genetics and environment interactions (Romao and Roth 2008).

Inorganic arsenic (iAs) is a ubiquitous environmental metalloid that is present in drinking water, soil, food, and other consumer products (Chung et al. 2014). In drinking water, levels of iAs are shown to exceed the WHO limit of 10 μg As/L in many regions of the United States, Bangladesh, India, China, and Taiwan, and parts of South America (Naujokas et al. 2013; WHO 2017). Chronic exposure to iAs is associated with cancer and other adverse health outcomes (Abdul et al. 2015; Kapaj et al. 2006; Maull et al. 2012). Diabetogenic effects of iAs exposures have been well studied in both humans and rodents (Maull et al. 2012). In humans, many epidemiological studies worldwide have linked iAs exposures with diabetes (Lai et al. 1994; Navas-Acien et al. 2008; Rahman et al. 1998; Tseng et al. 2000). In mice, iAs is shown to impair glucose homeostasis with or without dietary modifications (Castriota et al. 2020; Liu et al. 2014a; Paul et al. 2007b; Paul et al. 2011). Several molecular mechanisms for iAs-induced glucose dysregulation have been reported. These include oxidative stress, epigenetic alterations, and inflammatory signaling(Beck et al. 2017; Carmean and Seino 2019; Martin et al. 2017; Tseng 2004). IAs exposure is also associated with insulin resistance and β-cell dysfunction in both humans and mice (Douillet et al. 2013; Grau-Perez et al. 2017; Liu et al. 2014b; Padmaja Divya et al. 2015). A postulated underlying mechanism for arsenic-dependent insulin resistance is the down-regulation of the insulin-stimulated protein kinase B (Akt) signaling pathway (Paul et al. 2011; Walton et al. 2004; Zhang et al. 2017), while impaired mitochondrial metabolism and inhibition of calcium influx appear to underlie the inhibition of insulin secretion by β-cells (Dover et al. 2018a; Dover et al. 2018b; Huang et al. 2019). In humans, exposure to iAs often occurs over a lifetime; however, diabetogenic effects of iAs exposure at specific developmental windows, including exposure of parental germ cells prior to conception, has never been systematically examined. Some data in humans (Kile et al. 2014; Rager et al. 2014; Rojas et al. 2015) and mice (Fry et al. 2019; Huang et al. 2018) suggest that prenatal exposures play a role in diabetes development in offspring. But the specific role of preconception exposure alone on phenotypic outcomes in the offspring remains unclear.

The present study focuses on characterizing gene expression changes in germline cells from parental mice exposed to iAs as a potential mechanism by which preconception exposure to iAs may alter metabolic phenotypes in offspring. We hypothesized that preconception exposure to iAs would lead to changes in transcript levels in parental germline cells and in an adverse metabolic phenotype in offspring. We further hypothesized that offspring target tissue transcription profiles would reflect the preconception-associated germline transcription. To address this hypothesis, we conducted metabolic phenotyping and tissue-specific RNA sequencing in male and female offspring from parents that were exposed to iAs prior to conception. Our study highlights novel findings that suggest the preconception iAs exposure plays a role in glucose dysregulation in adult mice in a sex-specific manner.

Materials and methods

Preconception exposure to iAs

All procedures involving mice were approved by the UNC Institutional Animal Care and Use Committee. Twenty male and forty female C57BL/6J mice (6 weeks old) were purchased from the Jackson Laboratory (Bar Harbor, ME) and housed in the UNC animal facilities. All mice were singly caged upon arrival. The housing room was set on a 12-hour light-dark cycle. The experimental design of the study is shown in Figure 1. All mice were left to acclimatize for 2 weeks while fed AIN-93G food pellets (Research Diets Inc., New Brunswick, NJ) and drank arsenic-free deionized water (DIW). After acclimatization, all mice were arbitrarily separated into two exposure groups. Twenty female and ten male mice were exposed to sodium arsenite in deionized water (200 ug As/L, ppb) for 10 weeks. The remaining animals in the second, control group continued drinking DIW. Sodium arsenite (>99% pure) was purchased from Sigma Aldrich (St. Louis, MO). Drinking water in both groups was changed weekly. Water consumption was recorded on a weekly basis; food consumption was monitored every two weeks. The body weights were monitored biweekly. On week 9 of exposure, mice were fasted for six hrs, and tail bleeds were performed to measure fasting blood glucose (FBG) and fasting plasma insulin (FPI). After week 10, iAs exposure was terminated, and ten male and ten female mice from each treatment group were mated (1 on 1) for one week. During this time, all mice drank DIW. The ten remaining female mice in each treatment group were euthanized, and oocytes were collected from ovaries. Following the mating period, the males were also euthanized, and sperm was collected from cauda epididymis. (The methods used for oocyte and sperm isolation are described below.) The mated females were singly housed and closely monitored for pregnancy and pup births.

Figure 1.

Experimental design. After acclimatization for two weeks, 20 male and 2×20 female mice were exposed to iAs in drinking water for 10 weeks. Phenotyping was performed at week 10. At this time, oocytes were isolated from 10 females after euthanasia. The remaining 10 females and 10 males were mated (1 on 1) for 1 week. After mating, sperm was collected from males after euthanasia. Offspring born to pregnant dams were weaned at age of 3 weeks. Phenotype of the offspring was examined at age of 11 and 23 weeks. Tissues were collected from 23-week old offspring after euthanasia.

Offspring

At weaning (i.e., at 22 days of age), the G1 mice (the offspring) were weighed and singly housed. All G1 mice drank DIW ad libitum and were fed the AIN-93G diet. The body weights, food, and water consumption were monitored as described for the parental generation. Magnetic resonance imaging (MRI) using Echo MRI three-in-one Composition Analyzer and Labmaster (Echo Medical Systems, Houston, TX, USA) was conducted at 11 and 23 weeks of age to determine body composition (%fat and %lean mass). Metabolic phenotypes were also assessed at 11 and 23 weeks of age. Specifically, FBG and FPI were measured after 6-hour fasting. Fasted mice were then injected intraperitoneally with D-glucose (Sigma Aldrich, St. Louis, MO) in phosphate-buffered saline (2 g glucose/kg b.w.) and blood glucose and plasma insulin were measured 15 min post-dosing. After completion of the phenotype analysis at 23 weeks of age, mice were euthanized by cervical dislocation. Pancreatic islets were isolated as described below. The islets, along with other tissues collected after sacrifice, were flash frozen and stored at −80°C.

Analysis of blood glucose and plasma insulin

For both parental generation and the offspring, blood (~100 μL) was collected from tail cuts in ammonium heparin-coated capillary tubes from Kimble (Dover, Ohio). Glucose levels were measured in whole blood by One Touch^® Ultra^® glucometer (LifeScan, Inc., Milpitas, CA). Plasma was isolated by centrifugation at 1,700xg for 15 mins at 4°C and stored at −80°C. Plasma insulin levels were measured using ELISA kits from Crystal Chem (Elk Grove Village, IL) following the manufacturer’s protocol.

Calculation of metabolic indices

Homeostasis assessment-insulin resistance (HOMA-IR) index, an indicator of insulin resistance, was calculated using FBG (mg/dL) and FPI (μU/mL) as follows: HOMA-IR = [(FBG) x (FPI)]/405. FBG was expressed in mmol/ml for calculation of HOMA-β cell function (HOMA-B) index using the equation HOMA-B = [FPI*20]/[FBG − 3.5]. The insulinogenic index, an indicator of β cell function in response to glucose challenge, was calculated as (FPI_15min – FPI _Basal)/(FBG _15min – FBG _Basal).

Oocyte collection

The oocytes were collected by manual extraction methods from the 10 control and 10 iAs-exposed females that were not involved in mating. The media required to collect oocytes was freshly prepared before organ harvests. Minimum essential Medium (MEM) Eagle’s media, sodium bicarbonate, polyvinylpyrrolidone (PVP), sodium pyruvate, 1 M HEPES buffer, gentamycin (50 mg/mL), and Milrinone, a blood vessel vasodilator, were purchased from either Sigma Aldrich (St. Louis, MO) or Fischer Scientific (Hampton, NH). In brief, 90 mL of MEM media was taken in a beaker, and 0.22 g of sodium bicarbonate was added to it. 0.3g of PVP and 0.01g of sodium pyruvate were added to the above solution. 2.5 mL of 1 M HEPES and 0.1 mL of gentamycin (50 mg/mL) were aliquoted to the above solution. The pH was adjusted to 7.3, and the prepared media was made up to 100 mL and transferred to a bottle wrapped in foil and placed in a water bath maintained at 37°C. 2.5 mM of milrinone was added to the media to prevent oocyte maturation. Media containing milrinone was transferred to a 60 × 15 mm petri dish. Ovaries were isolated during necropsy and separated from surrounding adipose tissues. The clean ovaries were then placed in the petri dish with media. Using 25^5/8 G syringe needles, ovaries were shredded under a Nikon SMZ 800N microscope (Minato, Tokyo). A mouth pipette was assembled with a silicon tip from Drummond Scientific Company (Broomall, PA). The other end of the silicon tip was attached to a stripper or capillary plastic tube with an internal diameter of 75 μm from Cooper Surgical Inc. (Trumbull, CT). The mouth pipette setup was used to strip cumulous cells and other debris from oocytes by pipetting back and forth. The oocytes were then transferred into an Eppendorf tube containing RNase inhibitor (Qiagen, MD) and placed on dry ice.

Sperm collection

Sperm was collected using the swim out method from 10 sires immediately after 1 week of mating. The buffer for sperm collection was freshly prepared before necropsy. Human Tubal Fluid (HTF) HEPES buffer was purchased from Fisher Scientific (Hampton, NH). Bovine serum albumin (BSA) was purchased from Sigma Aldrich (St. Louis, MO). BSA was added to the buffer so that the final concentration was 5 mg/mL. The vas deferens and cauda epididymis were transferred into 1mL of warm 1X HTF on a well plate. Using a 26^½ G syringe needle, the vas deferens and epididymis were thoroughly scored to release sperm cells into the media. The media was incubated for 10 minutes at 37°C incubator to allow sperm to swim out to the media. The media was then transferred to a 1.5mL Eppendorf tube and centrifuged at 12,000g for 5 minutes. The supernatant was removed, and the pellet containing the sperm cells was frozen in dry ice.

Isolation of pancreatic islets

Islets were isolated from pancreata of male and female offspring using the previously described in situ collagenase perfusion method followed by centrifugation in Ficoll PM 400 (GE Healthcare, Uppsala, Sweden) (Douillet et al., 2013). Collagenase P was purchased from Roche Diagnostics Corp. (Indianapolis, IN). The islets were washed with Hank’s balanced Salt Solution containing 10mM HEPES, 7.5% sodium bicarbonate and 0.25% BSA (all from Sigma-Aldrich); pH was adjusted to 7.4.

RNA sequencing and differential gene expression analysis in sperm and oocytes from parental mice

Sperm and oocyte samples, 5 per treatment group, were selected from the parental generation for RNA sequencing. Total RNA extraction, library preparation, sequencing, and reads alignment was conducted at the UNC Center for Gastrointestinal Biology and Disease (CGIBD) Advanced Analytical Core. SMART-seq v4 Ultra Low RNA kit was purchased from TaKaRa Bio USA, Inc. (Mountain View, CA) for RNA extraction. cDNA libraries were prepared using Nextera XT DNA Library Preparation Kit from Illumina^® (San Diego, CA). Sequencing was conducted using NextSeq 500/500 v2.5 (150 cycles) kits from Illumina^® (San Diego, CA). Reads were aligned to the mm9 genome using Spliced Transcripts Alignment to a Reference (STAR)(Dobin et al. 2013). Differential gene expression was assessed using the DESeq package (Love et al. 2014). Transcriptome data were filtered to genes with absolute read counts of more than 10 in all sperm and oocyte samples for differential gene expression analysis (Venkatratnam et al. 2018). Differentially expressed genes were identified based on Benjamini-Hochberg FDR<0.1. Pathway analysis was conducted using the Database for Annotation, Visualization and Integrated Discovery (DAVID) (Dennis et al. 2003; Huang et al. 2009). The differentially expressed genes with p values <0.05 were used for pathway analyses.

RNA sequencing and differential gene expression analysis in pancreatic islets and livers from offspring

RNA sequencing was carried out in islets from three male and five female offspring from the control parents and five male and five female offspring from iAs-treated parents. RNA sequencing was also carried out in livers from five male and six female offspring from control parents and five male and five female offspring from iAs-treated parents. RNA was extracted from islets using RNAqueous micro kit from Ambion^® (Austin, TX). cDNA libraries were prepared using SEQuoia Complete Stranded RNA Library Kit from Bio-Rad laboratories (Hercules, CA). RNA from livers was extracted using RNeasy mini kit from Qiagen (Germantown, MD). cDNA libraries were prepared using Next Ultra II Directional RNA Library Prep Kit from New England Biolabs Inc (Ipswich, MA). Islet and liver RNA was sequenced using NextSeq 500/500 v2.5 (77 cycles) kits from Illumina^® (San Diego, CA). Reads alignment, differential gene expression, and pathway analysis were conducted as described above.

Data analysis of phenotypic endpoints

GraphPad Prism (La Jolla, CA) and R studio (Boston, CA) were used for data analysis and visualizations. Data from metabolic phenotyping were screened for outliers using Grubb’s outlier test. These data are presented in scatterplots graphs where the bars represent the mean and standard deviation. The data represent observations in male (N=10, 25) and female (N=10, 28) control and iAs nests. Mann-Whitney non-parametric test was used to characterize differences between treatment groups and sexes; P <0.05 was used as a threshold for statistical significance.

Results

Phenotypic characteristics of the parental mice

No statistically significant difference in FBG and FPI were observed between control and iAs-exposed males or females from the parental generation (Suppl. Figure 1A,B, C,D). Similarly, no treatment related differences were found in HOMA-IR and HOMA-B (Suppl. Figure 1E,F,G,H).

Transcriptomic responses in sperm cells and oocytes of parental mice exposed to iAs

Gene expression profiles were examined in the oocytes that were collected from the control and iAs-exposed virgin females and in the sperm collected from control and iAs-exposed sires after mating. RNA sequencing identified 912 and 180 genes differentially expressed in response to iAs exposure (p <0.05) in sperm cells and oocytes, respectively (Suppl. Tables 1 and 2). After false discovery rate correction using Benjamini-Hochberg method, five genes were identified to show altered expression due to iAs exposure in sperm cells, namely protein tyrosine phosphatase non-receptor type 22 (Ptpn22), cluster of differentiation 240 (Cd240), chemokine (C-C motif) ligand 8 (Ccl8), major vault protein (Mvp), and ras-related protein (Rab3d) were statistically significant (FDR <0.1) (Table 1). Interestingly, no statistically significant effects of iAs exposure on gene expression in oocytes were found at FDR<0.1. Pathway enrichment analysis was conducted using DAVID via the KEGG and BioCarta pathways. The genes (n=912; P<0.05) altered by iAs exposure in sperm cells have known roles in several biological pathways including Focal adhesion, Insulin-like growth factor binding, Phosphoinositide 3 kinase/protein kinase B (PI3K-Akt) signaling, Extracellular matrix receptor interaction, and Wingless-related integration site (Wnt) signaling that were significantly perturbed (FDR<0.1) and are relevant to glucose dysregulation (Table 2). No statistically significant pathway perturbation was observed in oocytes.

Table 1.

Genes that were differentially expressed in sperm cells of sires exposed to iAs^a

Gene	Base mean	Log2 Fold change	p-value	Benjamini Hochberg adjusted p-value
Protein tyrosine phosphatase, non-receptor type 22 (Ptpn22)	133.95	−0.90	6.38E-06	0.05
Cluster of differentiation (CD248)	84.67	1.28	1.12E-05	0.05
Chemokine (C-C motif) ligand 8 (Ccl8)	297.20	1.04	1.84E-05	0.05
Major Vault Protein (Mvp)	231.27	0.48	2.13E-05	0.05
Ras-related protein Rab-3D (Rab3d)	145.55	0.58	4.35E-05	0.08

^aSperm was collected after mating from males exposed to iAs for 10 weeks prior to mating and unexposed control males. Differentially expressed genes were identified based on Benjamini-Hochberg adjusted p-value <0.1.

Table 2.

Top pathways in sperm cells that were significantly enriched with genes (n=912) altered by exposure to iAs^a

Category	Pathway	Fold enrichment	p-value	Benjamini Hochberg adjusted p-value
KEGG	Focal adhesion	3.25	9.71E-08	2.43E-05
KEGG	PI3K-Akt signaling pathway	2.53	3.57E-07	4.46E-05
KEGG	ECM-receptor interaction	3.82	6.08E-05	0.005
Gene Ontology	Collagen trimer	4.05	3.59E-05	0.001
Gene Ontology	Insulin-like growth factor binding	8.35	2.84E-05	0.006
Gene Ontology	Positive regulation of non-canonical Wnt signaling	19.54	4.57E-05	0.024

^aDifferentially expressed genes (p<0.05) were analyzed for pathways enrichment; the significantly enriched pathways were identified using a Benjamini-Hochberg adjusted p-value <0.1.

Numbers and sex of the offspring

Nine out of 10 dams exposed to iAs prior to mating gave birth to offspring. In contrast, the mating of 10 control pairs resulted only in 3 litters (Suppl. Table 3). Thus, the G1 generation consisted of 10 females and 10 males from the control parents and 29 females and 27 males from iAs-treated parents. Two males and one female from iAs-treated parents died later after weaning for unknown causes.

Phenotypic characteristics of the G1 offspring

Food and water consumption, and body weights

Body weights of the offspring were monitored weekly from weaning (i.e., week 3 of age) until week 23 of age. During this period, male and female mice in both treatment groups steadily gained weight (Suppl. Figure 2A). Male mice were heavier compared to females by the end of the study. There were no statistically significant differences between males or females due to treatment. Food and water consumption were relatively steady for most duration of the study. Food consumption varied between 2.5–3.5g/day and tended to be higher in males as compared to females (Suppl. Figure 2B). Water consumption ranged from 2.5 to 6.2g/day (Suppl. Figure 2C).

Body composition

At 11 weeks of age, no significant differences were detected in body composition (% fat and % lean mass) between the treatment groups or between males and females within each treatment group (Suppl. Figure 3). At 23 weeks of age, no treatment related differences were found in body composition of G1 males (Figure 2A,C); however, females from the iAs-exposed parents had significantly higher % fat than females from the control parents (p =0.01) (Figure 2B). There were no significant treatment related differences in % lean mass among G1 females (Figure 2D). At 23 weeks of age, G1 males had significantly higher % fat than females in the treatment group (p=0.0003), but not in the control group (p=0.052) (Figure 2A,B). At this age, % lean mass in G1 females were significantly (p=0.0092) higher than in males within the iAs-exposed parental group (Figure 2C,D).

Figure 2.

Body composition of 23-week old male and female offspring from parents exposed to iAs prior to mating and from control, unexposed parents: Percent body weigh represented by fat (Percent fat) in male (A) and female (B) offspring, and percent body weight represented by lean mass (Percent of lean mass) in male (C) and female (D) offspring. Scatter plots show mean, ±SD and individual values. P values for the statistically significant differences (p<0.05) between treatments are shown.

Blood glucose and plasma insulin

FBG and FPI, as well as blood glucose and plasma insulin levels 15 minutes after glucose challenge were measured at 11 and 23 weeks of age. At week 11, there were only few significant differences in the glucose and insulin measures between treatment groups or between sexes (Suppl. Figure 4). Specifically, FBG was lower in females from iAs-treated parents than in females from control parents (p=0.039); Δ glucose_{(15–0 min)} value (i.e., the difference between 15-min glucose and FBG) was greater in males from iAs-exposed parents as compared to males from control parents (p=0.028). The same differences were found in the offspring at 23 weeks of age, but at this time point, these differences were not statistically significant: p=0.07 for the difference in FBG in females and p=0.09 for the difference in Δ glucose_{(15–0 min)} in males (Suppl. Figure 5). However, a statistically significant difference was found in 15-min blood glucose among the female offspring; 15-min glucose was lower in females from iAs-treated parents (Suppl. Figure 5D). At 23 weeks, 15-min plasma insulin (p=0.02) was higher in males from iAs exposed parents as compared to controls (Suppl. Figure 6). In contrast, no statistically significant treatment-related differences were found among female offspring.

Because we found that % fat differed significantly between females from the control and iAs-treated parents and because adiposity is known to affect blood glucose and insulin levels, we re-assessed the differences in blood glucose and plasma insulin levels at week 23 while normalizing for % fat. We found that the normalized FBG, 15-min glucose, and Δ glucose_{(15–0 min)} were significantly lower (p<0.025) in females from the iAs-treated nests as compared to females from the control nests (Figure 3B,D,F). There were no statistically significant treatment-related differences in the glucose measures among males (Figure 3A,C,E). The normalized FPI and 15-min plasma insulin were found to be lower in females from the iAs-treated parents than in those from the control parents (Figure 4A,B) with p=0.074 and p=0.017, respectively. Differences were also found in the normalized 15-min insulin (p = 0.003) and Δ insulin_{(15–0 min)} (p=0.07) among male offspring; both values were higher in males from the iAs-treated parents (Figure 4C,D).

Figure 3.

Fasting blood glucose (A,B), 15-min blood glucose (C,D) and delta glucose (i.e., 15-min glucose – fasting glucose) (E,F) normalized to percent body fat in 23-weeks old male and female offspring from iAs-treated and control parents. Scatter plots show mean, ±SD and individual values. P values for the statistically significant differences (p<0.05) between treatments are shown.

Figure 4.

Fasting plasma insulin (A,B), 15-min plasma insulin (C,D), and delta insulin (i.e., 15-min insulin – fasting insulin) (E,F) normalized to percent body fat in 23-weeks old male and female offspring from iAs-treated and control parents. Scatter plots show mean, ±SD, and individual values. P values for the statistically significant differences (p<0.05) are shown along with values p values that are higher than 0.05 but lower than 0.1.

While focusing on the effects of iAs exposure, we also found statistically significant differences between males and female offspring within each treatment group. Specifically, the normalized FBG was significantly higher in females compared to males regardless of treatment (p=0.0082). The normalized 15-min glucose and normalized Δ glucose_{(15–0 min)} were significantly higher in females than males from the control parents (p=0.007). The normalized 15-min plasma insulin was significantly higher in females than males from the control parents (p=0.0288), but was significantly lower in females than males from the iAs-treated parents (p=0.0007). The normalized Δ insulin_{(15–0 min)} was higher in males compared to females from the iAs-treated parent, but this difference did not reach statistical significance (p=0.0522).

Insulin resistance and β-cell function

We used FBG and FPI values to calculate HOMA-IR, an indicator of insulin resistance, and HOMA-B, an indicator of β-cell function. Insulinogenic index was calculated to evaluate β-cell function after glucose challenge. We found no statistically significant differences in HOMA-IR or HOMA-B values among the treatment groups (Suppl. Figure 7). However, the insulinogenic index was higher in males from iAs-treated parents as compared to controls (p = 0.05); no difference was found among females (Suppl. Figure 7E,F). When normalized for % fat HOMA-IR was found to be significantly lower in females from the iAs-treated parents as compared to the control parents (p =0.04) (Figure 5B); no significant differences were found in the normalized HOMA-B or insulinogenic index (Figure 5D,F). In contrast, the preconception exposure had no significant effects on the normalized HOMA-IR and HOMA-B in the male offspring (Figure 5A,C). However, the normalized insulinogenic index for males from the iAs-treated parents was significantly higher than for males from the control parent (p =0.0269) (Figure 5E). Sex-related differences were found only among offspring from iAs-treated parents. The normalized HOMA-IR was significantly higher in males compared to females (p=0.0287) and normalized HOMA-B was higher in females than males (p=0.0234). In addition, the normalized insulinogenic index was significantly higher in males compared to females (p=0.0473).

Figure 5.

HOMA-IR (A,B), HOMA-B (C,D), and insulinogenic index (E,F) normalized to percent body fat in 23-weeks old male and female offspring from iAs-treated and control parents. Scatter plots show mean, ±SD, and individual values. P values for the statistically significant differences (p<0.05) between treatments are shown.

Transcriptomic responses in islets and livers of male and female offspring from iAs-treated parents

We compared transcript levels in the islets and livers of male and female offspring from iAs-treated parents to those in the islets and livers of the control offspring using RNA sequencing.

In the islets of female and male offspring, 99 and 100 genes were differentially expressed in response to the preconception iAs exposure (p<0.05), respectively (Suppl. Tables 4 and 5). However, no statistically significant differences between the transcript levels in the islets from the control and treatment groups were found after adjusting for multiple testing using Benjamini-Hochberg method. Pathway enrichment analysis of the n=100 differentially expressed genes (p<0.05) identified multiple pathways altered by the preconception exposure to iAs in male islets; pathways involved in translational processes were among the most significantly affected (FDR<0.1) (Suppl. Table 6). No pathways were significantly altered in female islets.

In the liver, the transcriptomic responses were considerably stronger than in the islets. The preconception exposure altered 1407 and 843 genes significantly (p<0.05) in the livers of the female and male offspring, respectively (Suppl. Tables 7 and 8). After adjusting for false discovery rates, eight genes in the female livers, and one gene in the male livers were found to be significantly perturbed (FDR<0.1) (Table 3). In female livers, these genes included AarF domain containing kinase 5 (Adck5), Hydroxypyruvate isomerase (Hyi), BTG3 associated nuclear protein (Banp), ATP-binding cassette sub-family A member 9 (Abca9), Cytochrome B-245 beta chain (Cybb), Ubiquitin-like modifier activating enzyme 6 (Uba6), ADP ribosylation factor Like GTPase 4D (Arl4d), and Asparaginase (Aspg). Expression of G0/G1 switch gene 2 (G0s2) was found to be altered in male livers. Pathway analysis using differentially expressed genes with p values <0.05 (n=843) identified several pathways that were altered in male livers by the preconception iAs exposure, including extracellular matrix (ECM) receptor interaction, focal adhesion, and PI3K-Akt signaling pathways (Table 4). The differentially expressed genes in female livers (n=1407) were enriched in pathways associated, for example, with ribosomal structure and function, with mitochondrial respiration and oxidative phosphorylation, non-alcoholic fatty liver disease, and NADH dehydrogenase activity (Table 5).

Table 3.

Genes that were differentially expressed (n=9) in livers of female and male offspring from parents exposed to iAs^a

Gene (male livers)	Base mean	Log2 Fold change	p-value	Benjamini Hochberg adjusted p-value
G0/G1 switch gene 2 (G0s2)	2364.91	−1.820	7.91E-06	0.087
Gene (female livers)	Base mean	Log2 Fold change	p-value	Benjamini Hochberg adjusted p-value
AarF domain containing kinase 5 (Adck5)	550.79	−0.74029	5.69E-07	0.006271
Hydroxypyruvate isomerase (Hyi)	1062.55	−0.77357	2.42E-06	0.008925
BTG3 associated nuclear protein (Banp)	209.52	−1.01104	2.43E-06	0.008925
ATP-binding cassette subfamily A member 9 (Abca9)	80.86	0.732965	1.49E-05	0.040947
Cytochrome B-245 beta chain (Cybb)	465.07	0.619609	1.86E-05	0.041049
Ubiquitin-like modifier activating enzyme 6 (Uba6)	112.97	0.788383	5.34E-05	0.097647
ADP ribosylation factor Like GTPase 4D (Arl4d)	349.28	−0.71922	7.06E-05	0.097647
Asparaginase (Aspg)	3185.9	−0.42721	7.09E-05	0.097647

^aDifferentially expressed genes were identified based on Benjamini-Hochberg adjusted p-value <0.1.

Table 4.

Top pathways in livers of male offspring that were significantly enriched among genes (n=843) altered by preconception exposure to iAs^a

Category	Pathway	Fold enrichment	p-value	Benjamini Hochberg adjusted p-value
KEGG	Focal adhesion	2.69	9.71E-06	0.001
KEGG	ECM-receptor interaction	3.41	1.90E-04	0.006
KEGG	PI3K-Akt signaling pathway	1.89	8.38E-04	0.023
Gene ontology	Cell cycle	1.98	8.46E-06	0.006
Gene ontology	Platelet-derived growth factor binding	12.32	6.73E-05	0.011
Gene ontology	Extracellular matrix structural constituent	5.40	2.04E-04	0.017
KEGG	Protein digestion and absorption	2.92	0.002381	0.049
Gene ontology	Collagen trimer	3.052	0.005332	0.079
Gene ontology	Leukocyte cell-cell adhesion	8.29	3.11E-05	0.014
Gene ontology	Membrane to membrane docking	19.89	6.07E-04	0.086
Gene ontology	ATP binding	1.45	2.92E-04	0.023
Gene ontology	Nucleotide binding	1.38	3.37E-04	0.024
Gene ontology	Cadherin binding involved in cell-cell adhesion	2.20	4.45E-04	0.027
Gene ontology	Cell-cell adherens junction	2.00	0.00175	0.031
Gene ontology	MCM complex	14.07	2.58E-04	0.007
Gene ontology	DNA unwinding involved in DNA replication	12.43	4.46E-04	0.070
KEGG	DNA replication	4.29	0.004963	0.086
Gene ontology	Chromosome condensation	9.32	3.11E-04	0.061
Gene ontology	Membrane to membrane docking	19.89	6.07E-04	0.086
KEGG	Bladder cancer	5.23	8.46E-05	0.005
Gene ontology	Phosphatidylinositol-3,4-bisphosphate binding	6.72	0.001597	0.071
Gene ontology	GO:0030261~chromosome condensation	9.32	3.11E-04	0.061

^aDifferentially expressed genes (p<0.05) were analyzed for pathways enrichment; the significantly enriched pathways were identified using a Benjamini-Hochberg adjusted p-value <0.1.

Table 5.

Top pathways in livers of female offspring that were significantly enriched among genes (n=1407) altered by preconception exposure to iAs^a

Category	Pathway	Fold enrichment	P-value	Benjamini Hochberg adjusted p-value
Gene Ontology	Ribosome	4.55	4.07E-21	1.34E-18
KEGG	Ribosome	4.66	5.97E-20	1.57E-17
Gene Ontology	Intracellular ribonucleoprotein complex	3.27	1.84E-17	3.02E-15
Gene Ontology	Structural constituent of ribosome	3.42	4.95E-16	5.21E-13
Gene Ontology	Translation	2.36	6.75E-10	2.36E-06
Gene Ontology	Cytosolic large ribosomal subunit	3.83	1.64E-07	8.29E-06
Gene Ontology	Mitochondrial inner membrane	3.19	6.91E-20	1.51E-17
Gene Ontology	Respiratory chain	6.28	2.03E-12	1.67E-10
KEGG	Oxidative phosphorylation	3.74	3.56E-12	4.68E-10
Gene Ontology	Mitochondrial respiratory chain complex I	6.74	1.73E-11	1.26E-09
KEGG	Parkinson’s disease	3.40	1.40E-10	1.23E-08
KEGG	Huntington’s disease	2.84	3.10E-09	1.63E-07
KEGG	Non-alcoholic fatty liver disease (NAFLD)	2.87	1.27E-07	5.55E-06
KEGG	Alzheimer’s disease	2.70	2.06E-07	7.73E-06
Gene Ontology	NADH dehydrogenase (ubiquinone) activity	4.98	4.07E-05	0.007
Gene Ontology	NADH dehydrogenase activity	8.54	7.20E-05	0.011
Gene Ontology	Cytosolic small ribosomal subunit	4.78	5.81E-07	2.72E-05
Gene Ontology	Ribosomal small subunit assembly	7.41	2.59E-06	0.003
Gene Ontology	Small ribosomal subunit	4.92	3.17E-04	0.009
Gene Ontology	Peroxisome	3.02	2.03E-06	8.90E-05
KEGG	Peroxisome	3.39	3.72E-06	1.22E-04
Gene Ontology	Peroxisomal membrane	3.29	0.001	0.038
KEGG	Valine, leucine and isoleucine degradation	3.07	0.001	0.040
Gene Ontology	Proteasome complex	3.12	7.69E-04	0.020
Gene Ontology	Sex chromatin	10.57	0.004	0.087

^aDifferentially expressed genes (p<0.05) were analyzed for pathways enrichment; the significantly enriched pathways were identified using a Benjamini-Hochberg adjusted p-value <0.1.

Genes differentially expressed in both the parental germ cells and the tissues of the offspring

We compared genes that were differentially expressed in the parental germline cells with genes differentially expressed in the islets and livers of the offspring using the p<0.05 threshold. We identified 95 genes in livers of male and 81 genes in livers of female offspring that were also differentially expressed in paternal sperm (Suppl. Tables 9 and 10). Similarly, we found six genes in male livers and 20 in female livers that were also differentially expressed in maternal oocyte (Suppl. Table 11). There were very few overlaps between genes differentially expressed in the germ cells and in the islets of the offspring (data not shown). We performed pathway analysis using genes that were differentially expressed in both paternal sperm and in livers of the male offspring (n=95) and female offspring (n=81). We identified 16 biological pathways in livers of the male offspring that were enriched for these genes. (Table 6). As discussed below, some of these pathways are involved in the regulation of glucose homeostasis and/or are linked to pathologies associated with diabetes. No pathways were significantly enriched for the genes that were differentially expressed in the sperm and the livers of female offspring.

Table 6.

The pathways that were significantly enriched among genes (n=95) altered by preconception exposure to iAs in both the parental sperm and the liver of male offspring ^a

Category	Pathway	Fold enrichment	p-value	Benjamini Hochberg adjusted p-value
Gene Ontology	extracellular matrix	16.16	1.47E-19	2.26E-17
Gene Ontology	proteinaceous extracellular matrix	15.04	6.50E-19	5.00E-17
Gene Ontology	extracellular region	3.82	8.42E-11	4.32E-09
Gene ontology	extracellular space	3.44	1.90E-07	4.87E-06
Gene ontology	collagen trimer	20.82	1.04E-07	3.20E-06
Gene ontology	platelet-derived growth factor binding	90.86	1.88E-07	3.77E-05
Gene Ontology	extracellular matrix structural constituent	31.91	1.10E-06	1.11E-04
KEGG	ECM-receptor interaction	16.53	2.79E-06	2.71E-04
KEGG	Protein digestion and absorption	16.53	2.79E-06	2.71E-04
KEGG	PI3K-Akt signaling pathway	6.51	3.03E-06	1.47E-04
KEGG	Focal adhesion	9.03	3.81E-06	1.23E-04
Gene Ontology	collagen fibril organization	27.27	3.15E-05	0.010914
Gene Ontology	cellular response to amino acid stimulus	10.86	4.24E-05	0.00981
Gene Ontology	protein heterotrimerization	44.78	8.88E-05	0.012305
Gene Ontology	wound healing	15.84	4.93E-06	0.003428
Gene Ontology	cellular response to amino acid stimulus	10.86	4.24E-05	0.00981

^aGenes (p<0.05) were analyzed for pathways enrichment; the significantly enriched pathways were identified using a Benjamini-Hochberg adjusted p-value <0.1.

Discussion

The association between chronic iAs exposure from drinking water and type 2 diabetes has been supported by results of multiple cross-sectional and prospective population studies (Thayer et al. 2012). Almost all these studies have been carried out in adult cohorts, using the levels of iAs in drinking water at the time of the study as an indicator of iAs exposure. However, participants in many of these cohorts were likely exposed to iAs for a lifetime, and possibly across generations. Yet, no population study to date has systematically evaluated association between type 2 diabetes and prenatal exposure to iAs. Data from three recent studies suggest the iAs exposure during this developmental window may significantly contribute to the diabetogenic phenotypes observed in adults. In the first study, gestational iAs exposure was shown to alter CpG methylation of genes in cord blood of newborns in Bangladesh; some of these genes were associated with the Maturity onset diabetes of the young (Mody) pathway (Kile et al. 2014). Similarly, Rojas and associates showed that prenatal exposure to iAs modified DNA methylation of 2919 genes in the cord blood of newborns in a birth cohort in Mexico (Rojas et al. 2015). Notably, for 16 of these genes the DNA methylation correlated with the transcript levels. Among these genes was imprinted gene, KCNQ1, which encodes potassium voltage-gated channel subfamily Q member 1, a key regulator of insulin secretion in β-cells. Finally, in the same cohort, maternal exposure to iAs was shown to alter diabetes-associated microRNAs in cord blood of newborns (Rager et al. 2014).

When modeling iAs-associated type 2 diabetes in laboratory settings, most studies have focused on iAs exposure in adult laboratory animals, mainly mice. Very few studies have assessed effects of prenatal exposures. In general, results of these studies suggest that in utero exposures alone or combined with early life exposure to low (ppb) levels of iAs in drinking water result in an increased body weight and percent fat, and produce adverse metabolic phenotypes characterized by impaired glucose and lipid metabolism in offspring, and that these effects are sex-dependent (Ditzel et al. 2016; Pablo Sanchez-Soria 2014; Rodriguez et al. 2016). Similarly, we have recently shown that combined preconception and in utero exposures to 100 ppb iAs lead to impaired FBG in adult male, but not female C57BL/6 offspring (Huang 2018). In a recent study, Fry and associates found that the combined preconception and in utero exposures produced insulin resistance in adult male CC004 mice from the Collaborative Cross (Fry et al. 2019). In contrast, combined in utero and postnatal exposures had no effects on insulin sensitive/resistance in this mouse strain. These results suggest that the preconception exposure alone may result in a diabetic phenotype in offspring. The goal of the present study was to test this hypothesis and to explore a possibility that the effects of the preconception exposure may be associated with iAs-induced alterations in transcriptomes of the parental germ cell.

We exposed the parent mice to the environmental relevant 200 ppb iAs in drinking water. For comparison, chronic exposure to 100 ppb did not produce diabetic phenotype in adult C57BL/6 mice in our published studies (Douillet et al. 2017). We found no significant differences in diabetes indicators between the control and iAs-treated mice in the parental group. In clinical studies, diabetes during pregnancy, namely gestational diabetes, has been a risk factor for development of diabetes in offspring later in life (Damm 2009; Damm et al. 2016; Metzger 2007). A hyper-insulinogenic and hyper-glycemic fetal environment in pregnant women with diabetes, which results in fetal epigenetic re-programming of genes involved in energy metabolism is thought to drive the diabetic phenotype in the offspring (Dabelea 2007; Damm et al. 2016; Metzger 2007; Nicholas et al. 2016). The fact that in this study, dams exposed to iAs did not develop diabetes suggests that the altered phenotypes of the offspring from these dams are caused by other mechanisms.

We found that the preconception exposure to 200 ppb iAs had significant effects on diabetic indicators in C57BL/5 offspring, but these effects were strictly sex dependent. Interestingly, the 23-week old female offspring from parents exposed to iAs prior to mating had higher % fat but lower FBG and 15-min blood glucose levels than the female offspring from the control parents. This finding seems to contradict a well-established observation that adiposity is a risk factor for diabetes in both humans and mice (Farkhondeh et al. 2019; Jeon et al. 2015; Navas-Acien et al. 2006). Notably, the preconception exposure had no significant effects on %fat, FBG or 15-min glucose levels in the male offspring. However, the higher Δ glucose_{(15–0 min)} and Δ insulin_{(15–0 min)} values in the male offspring from iAs-exposed parents are suggestive of an impaired glucose tolerance and insulin resistance.

To differentiate between the diabetogenic effects of fat and those of the preconception iAs exposure, we normalized the blood glucose and insulin indicators in the 23-weeks old offspring and the functional indices derived from these indicators for % fat. As expected, the normalization did not eliminate the treatment-related differences in FBG and 15-min blood glucose levels of the female offspring. In fact, these differences became even more statistically significant. In addition, the difference in Δ glucose_{(15–0 min)} also became statistically significant. Interestingly, the normalization revealed significant differences in FPI and 15-min plasma insulin levels of the female offspring; both values were lower in females from parents exposed to iAs than in those from control parents. Taken together, these results suggest that the female offspring from iAs-exposed parents, while having higher % fat, were able to manage the blood glucose levels during fasting and after glucose challenge more effectively than the female offspring from control parents, perhaps due to greater insulin sensitivity. This conclusion is further supported by our finding that the %fat-normalized HOMA-IR, the indicator of insulin resistance, was lower in females from iAs-exposed parents than in those from control parents. In comparison, in our previous study, a combined preconception and in utero exposure to 100 ppb iAs had only a marginal effect on body composition of male or female offspring and normalization for %fat eliminated significant differences in FPI and HOMA-IR between iAs-exposed and control male offspring (Huang et al. 2018b). In that study, no effects of the combined exposure on diabetes indicators in female offspring were found.

Because body composition of the male offspring was not significantly affected by preconception exposure, normalizing the diabetes indicators for % fat did not change the key observations. The 15-min insulin level remained significantly higher in males from iAs-exposed parents (p = 0.003). The Δ insulin_{(15–0 min)} value was also higher, but this difference did not reach statistical significance (p=0.07). No difference was found in the normalized HOMA-IR. The difference in insulinogenic index, a marker of insulin secretion by β-cells in response to glucose challenge (Singh and Saxena 2010), which was not significant became statistically significant after normalization for % fat. The normalized insulinogenic index was significantly higher in the male offspring from iAs-exposed parents. These results indicate that the male offspring from parents exposed to iAs were able to maintain glucose homeostasis during fasting but required greater amounts of insulin to support glucose utilization after glucose challenge.

Notably, no significant effects of the preconception exposure were observed in relation to HOMA-B in male and female offspring either before or after normalizing for % fat. HOMA-B is a marker of β-cell function but is based on FBG and FPI values. Thus, it does not reflect mechanisms that are activated during glucose challenge.

Exposures to environmental contaminants have been shown to alter gene expression profiles in germ cells leading to altered phenotypes in offspring (Allegrucci et al. 2005; Bollati and Baccarelli 2010; Nilsson et al. 2018). In laboratory studies, a daily i.p. injection of iAs (6 mg/kg b.w.) for 26 days decreased testicular mass and impaired spermatogenesis in Wistar rats by increasing sperm DNA damage and decreasing testicular testosterone, epididymal sperm count, and number of germ cells at stage VII (Choudhury et al. 2011; Sarkar et al. 2003). In mice, a daily oral dose of iAs (7.5mg/kg b.w.) for 35 days has been shown to decrease sperm motility, increased sperm morphological abnormality, and decreased sperm viability (Ferreira et al. 2012). Adverse effects of iAs have also been reported in female germ cells. For example, i.p. injection of iAs (16 mg/kg b.w.) for 14 days resulted in an increased abnormal spindle and chromosomal fragmentation in mature CD-1 oocytes at metaphase II, leading to meiotic disruption (Navarro et al. 2004). These data suggest that iAs exposure impairs structure and function of germ cells. However, associations between the iAs-induced damage to germ cells and the phenotypic outcomes in offspring remain unclear.

The present study is the first to provide evidence of a link between iAs-induced differential gene expression in germ cells and altered gene transcription profiles in tissues, particularly liver, of the offspring. Specifically, we found that many of the genes and pathways that were altered by iAs exposure in paternal sperm and maternal oocytes were also altered in the livers of male and female offspring born to the iAs-exposed parents. Notably, among these genes and pathways were those that are known to be associated with regulation of glucose homeostasis and diabetes. For example, PI3K/Akt signaling pathway and pathways involved in maintenance and remodeling of the extracellular matrix. The PI3K/Akt signaling is activated by insulin during the postprandial stage (White and Kahn 1994). The insulin-dependent activation (phosphorylation) of Akt leads to the translocation of glucose transporter type 4 receptor (GLUT4) from the perinuclear compartment to the plasma membrane and to stimulation of glucose uptake and utilization (White and Kahn 1994). Inhibition of Akt phosphorylation or suppression of Akt expression is associated with impaired insulin signaling and insulin resistance (White and Kahn 1994). The extracellular matrix (ECM), which is composed by diverse networks of protein and proteoglycans, is involved in numerous cellular processes, including signal transduction and activation of intracellular signaling (e.g., PI3K/Akt signaling) (Williams et al. 2015). The pathways involved in maintenance and remodeling of the extracellular matrix are known to be impaired by diabetes (Law 2012) and this impairment is thought to contribute to insulin resistance (Williams et al. 2015). Notably, the diabetes-associated pathways were dysregulated in the paternal sperm as well as in the liver of male offspring, which developed the diabetic phenotype. Thus, the iAs-induced differential expression of genes in germ cells may be directly responsible for diabetic phenotype in male offspring from the iAs-exposed sires.

In summary, our results show that the preconception exposure to 200 ppb iAs in drinking water altered diabetes indicators in both male and female offspring that were not directly exposed to iAs. Notably, the effects of preconception exposure in the offspring were strictly sex-specific. While the findings in the male offspring suggest an increase requirement for insulin in response to glucose challenge, the glucose and insulin indicators in the female offspring are consistent with a protective phenotype, despite increased adiposity. In addition, this study is the first to link the diabetic phenotype of the male offspring to altered gene expression in the parental germ cells. Molecular mechanisms that underlie this link will be investigated in future studies.