3.2.2. F1 Behavior

### PFOA

Parental PFOA exposure was associated with increased swimming activity in both light and dark at the very low and low concentrations (Figures 1 and 2) (dark: *p =* 1.3 × 10−<sup>2</sup> ; *p* < 1 × 10−<sup>8</sup> ; 70 and 700 ng/L, respectively) (light: *p =* 4 × 10−<sup>2</sup> , 1.3 × 10−<sup>2</sup> , respectively).

#### PFOS

Parental PFOS exposure was associated with increased larval activity in the dark at the very low and ultra-low concentrations (*p* < 1 × 10−<sup>8</sup> for both), yet activity decreased in the light at the low concentration (*p =* 5.9 × 10−<sup>3</sup> ) (Figures 1 and 2).

#### Mixture

Parental exposure to the PFAS mixture strongly decreased swimming behavior in the dark at all concentrations (*p* < 1 × 10−<sup>8</sup> for all) and in the light as well only at the ultra-low level (*p =* 3.4 × 10−<sup>6</sup> ); *n* = 72 per concentration (Figures 1 and 2).

#### 3.2.3. F1 Transcriptomics and Pathway Analysis

The full lists of DEGs for all chemicals and concentrations in the F1 generation can be found in Table S4; the top five up- and downregulated DEGs and affected pathways are shown in Table 2. The full lists of pathways (where applicable) for all chemicals, concentrations, and generations can be found in Table S5. Venn diagrams illustrating the overlap of generation-specific DEGs (all concentrations combined) are in Figure 2. Venn diagrams illustrating the overlap of F1 DEGs for each chemical (all concentrations combined) are in Figure 3. DEGs are considered significant at *p* < 0.01 and log2FC of ≥0.75 or ≤−0.75.

#### PFOA

Parental PFOA exposure at the ultra-low level caused the significant upregulation of 12 genes and the downregulation of 5 genes (log2FC < −0.75). The most upregulated gene (log2FC = 1.03) was *dusp16* and the most downregulated *si:ch211-125e6.5* (log2FC: −0.84). At very low exposure, 106 genes were significantly differentially expressed (64 upregulated, 42 downregulated). The most upregulated gene was *dusp27* (log2FC: 1.24), and the most downregulated was the innate immunity-related *c3a.2* (log2FC: −1.44). The genes were involved in pathways of xenobiotic metabolism via the CAR pathway and estrogen receptor signaling. Other xenobiotic pathways involved were LXR, RXR, AhR, and FXR. The kinase *dusp27* was the most upregulated molecule. With low parental exposure, there were 49 DEGs, with the most highly upregulated gene of 37 genes being *tmigd1* (log2FC: 1.26) and the most downregulated *trmt1* (log2FC: −0.87), out of 12.

#### PFOS

Parental PFOS exposure at the ultra-low level caused the significant upregulation of five genes. The most upregulated gene (log2FC = 0.96) was *satb1a*. Very low parental exposure resulted in only two significant DEGs: *npas4a* (log2FC: 0.75) and *slc43a2a* (log2FC: −0.94). Low parental exposure resulted in 149 significant DEGs. The most highly upregulated of 118 genes was *zmat5* (log2FC: 1.29); the most downregulated gene of 80 genes was *mfsd14ba* (log2FC: −1.14). Pathway analysis indicated increased lipid metabolism and decreased cell death pathways.

#### Mixture

Parental exposure to the ultra-low level of PFAS mixture induced 35 significant DEGs. Of the 21 upregulated genes, *zgc:92590* was the highest (log2FC: 1.35). Of the 13 downregulated genes, *smtnl* was the most downregulated (log2FC: −1.17). Very low parental exposure was associated with the upregulation of 11 genes, with *cela1.3* being the most upregulated (log2FC: 1.19), and only 1 downregulated gene (*panx1a*, log2FC: −0.85). At

low parental exposure, five genes were upregulated, with *cpa4* the most upregulated (log2FC: 0.88), and two genes were downregulated: *rlbp1b* and *ms4a17a.8* (log2FC: −0.79, −0.89, respectively).

3.2.4. F1 Fecundity PFOA

The F1 generation of very low-level PFOA exposure lineage produced significantly fewer eggs than controls (−28.2%, *p <* 0.01). Ultra-low and low concentrations were not affected (Figure S2). Full fecundity data are shown in Table S2.

#### PFOS

Parental PFOS exposure had no effect on F1 fecundity (Figure S2). Full fecundity data are shown in Table S2.

#### Mixture

Parental mixture exposure had no effect on F1 fecundity (Figure S2). Full fecundity data are shown in Table S2.

#### 3.2.5. F1 Sex Ratio

#### PFOA

At every concentration (ultra-low, very low, low), there was a significant increase in the male:female ratio of adult fish (*p =* 7.4 × 10−<sup>4</sup> , 2.6 × 10−<sup>9</sup> , 3.2 × 10−<sup>9</sup> , respectively). The authors note the abnormal lack of males in the control group, which may have led to a false-positive result (PFOA control male ratio: 0.17, PFOS control male ratio: 1.53, mixture control male ratio: 1.12). PFOA F1 contained significantly fewer males than PFOS and mixture F1 (*p* = 4.1 × 10−<sup>4</sup> , 4.24 × 10−<sup>5</sup> , respectively (chi-square test)). There was no difference in control PFOS and mixture male ratio (*p* = 0.59). Additionally, due to a lack of access to research animals during the SARS-CoV-2-related institutional shutdown, only one cohort of fish (*n* = 56–66) could be observed.

#### PFOS

No sex ratio shift was observed (*n* = 36–86).

#### Mixture

At the very low exposure level, there was a significant increase in the male:female ratio (4.45 compared to 1.12 in controls, *p =* 1.8 × 10−<sup>3</sup> , *n* = 55–63). The authors note that due to a lack of access to research animals during the SARS-CoV-2-related institutional shutdown, only two cohorts of fish could be observed.

#### *3.3. F2 Generation*

Table 1 shows significant endpoints for all chemicals and concentrations.

#### 3.3.1. F2 Abnormalities and Survival

No statistically significant abnormalities or mortality were observed in any concentration of any chemical or mixture.

#### 3.3.2. F2 Behavior

PFOA

Transgenerational behavioral effects of legacy PFOA exposure manifested as hypoactivity at each concentration (Figures 1 and 2) (ultra-low: *p* < 1 × 10−<sup>8</sup> (dark); very low: *p* < 1 × 10−<sup>8</sup> (light); low: *p =* 1 × 10−<sup>2</sup> (dark); *p* < 1 × 10−<sup>8</sup> (light)).

#### PFOS

Transgenerational behavioral effects of legacy PFOS exposure manifested at only the very low exposure concentration, where hyperactivity was observed (Figures 1 and 2) (dark: *p =* 7.6 × 10−<sup>3</sup> ; light: *p* < 1 × 10−<sup>8</sup> ).

#### Mixture

Transgenerational behavioral effects of legacy exposure to the PFAS mixture presented as hypoactivity following ultra-low exposure (Figures 1 and 2) (dark: *p =* 4.2 × 10−<sup>2</sup> ; light *p =* 3.2 × 10−<sup>4</sup> ) and hyperactivity only at the very low concentration and only in the dark (*p* < 1 × 10−<sup>8</sup> ).

#### 3.3.3. F2 Transcriptomics and Pathway Analysis

The full lists of DEGs for all chemicals and concentrations in the F2 generation can be found in Table S5; the top five up- and downregulated DEGs and affected pathways are shown in Table 2. The full lists of pathways (where applicable) for all chemicals, concentrations, and generations can be found in Table S6. Venn diagrams illustrating the overlap of generation-specific DEGs (all concentrations combined) are in Figure 2. Venn diagrams illustrating the overlap of F2 DEGs for each chemical (all concentrations combined) are in Figure 3. DEGs are considered significant at *p* < 0.01 and log2FC of ≥0.75 or ≤−0.75.

#### PFOA

Ultra-low-level ancestral exposure to PFOA resulted in 112 significant DEGs in the F2 generation (30 upregulated, 82 downregulated). The most upregulated gene was *amy2al2* (log2FC: 4.41) and the most downregulated was *ms4a17a.8* (log2FC: −2.21). Pathway analysis revealed dysregulation of mitochondrial membrane potential and increased organismal injury, including cancer. In total, 106 DEGs resulted from ancestral PFOA exposure at the very low level (38 upregulated, 68 downregulated). As with the ultra-low concentration, the most upregulated gene here was the carbohydrate-metabolism-related *amy2al2* (log2FC: 3.58); the most downregulated was the sodium channel gene *scn2b* (log2FC: −2.62). Pathway analysis implicated cholesterol synthesis via CYP51A1 and other canonical pathways of sterol synthesis. Ancestral low PFOA exposure resulted in 302 significant DEGs in the F2 generation (124 upregulated, 178 downregulated). The most upregulated gene was again *amy2al2* (log2FC: 4.22), and the most downregulated was *lhx2b* (log2FC: −2.10). Pathways of immune function were upregulated; the top five most upregulated pathways all regard the trafficking of various immune cell types; 9 of the top 20 most upregulated pathways also feature cellular movement, 7 of these in immune cells specifically. The 20 most downregulated pathways feature 9 cell-death-related functions and 5 involved in the dysregulation of glucose homeostasis. Comparisons with GSEA datasets concerning epigenetic and/or chromatin regulation returned multiple DEGs (Table 3).

#### PFOS

Ultra-low-level ancestral exposure to PFOS resulted in 484 significant DEGs in the F2 generation (209 upregulated, 275 downregulated). The most upregulated gene was *cela1.5* (log2FC: 2.07), and the most downregulated was *pgk1* (log2FC: −2.34). Pathway analysis shows increased lipid metabolism, with 8 of the top 20 most upregulated pathways having to do with the synthesis or metabolism of steroids and terpenoids. The most downregulated pathway was bone mineral density (bias-corrected z-score: −2.75); other connective tissue pathways were overrepresented in the 20 most downregulated pathways. In total, 23 DEGs resulted from ancestral PFOS exposure at the very low level (16 upregulated, 7 downregulated). As with the ultra-low concentration, the most upregulated gene here was *cela1.5* (log2FC: 1.52); however, the most downregulated was *b3gntl1* (log2FC: −0.87). Ancestral low PFOS exposure resulted in seven significant DEGs in the F2 generation (four upregulated, three downregulated). The most upregulated gene was an unnamed/unannotated

gene on chromosome 3 (*ENSDARG00000115830*) (log2FC: 1.38), and the most downregulated was *cfp* (log2FC: −1.00). Comparisons with GSEA datasets concerning epigenetic and/or chromatin regulation returned multiple DEGs (Table 3).

**Table 3.** Significant DEGs involved in epigenetic processes in the F2 generation of zebrafish (*Danio rerio*).


Mixture

Ultra-low-level ancestral mixture exposure was associated with 69 significant DEGs in the F2 generation. Of the 27 upregulated genes, *fzd6* had the highest log2FC (1.31). Of the 42 downregulated genes, *tcap* had the lowest log2FC (−1.19).

At the very low concentration, F2 larvae exhibited only one significantly upregulated DEG (*purab*, log2FC: 0.76) and four downregulated. The most downregulated gene was *hbae5* (log2FC: −1.36).

Similarly, the low level exposure had few DEGs. Two were upregulated (*fgfbp2b* and *si:dkey-102c8.3*, log2FC: 0.77, 0.76, respectively), and seven were downregulated (most downregulated: *si:ch211-281l24.3*, log2FC: −0.93). Comparisons with GSEA datasets concerning epigenetic and/or chromatin regulation returned one DEG, *h2bc1* (Table 3). No generation alone produced a sufficient number DEGs for pathway analysis; however, when all generations and concentrations were collated, pathways of cell death and immune dysfunction emerged (Table 4).


**Table 4.** Pathway analysis (IPA) of all DEGs from each concentration and generation of mixtureexposed zebrafish (*Danio rerio*) larvae combined.

#### **4. Discussion**

In this study, numerous endpoints were examined across three generations of zebrafish exposed to environmentally relevant PFAS concentrations. Locomotion, gene expression, and fecundity were significantly altered across all generations by at least one concentration of PFOA, PFOS, and/or their mixture.

Environmental levels of PFAS exposure, as expected, did not cause significant mortality in any generation. Mortality with PFAS exposure is typically not observed in zebrafish under 10 mg/L (10<sup>7</sup> ng/L) [23,24]. Jantzen et al. [25] also found no significant death or abnormalities using similar exposure methods to PFOS and PFOA. Gross morphological abnormalities were not increased by exposure, agreeing with literature noting abnormalities following ≥1 mg/L (10<sup>6</sup> ng/L) PFAS exposure [26]. In the F1 generation of PFOA larvae, decreased abnormalities and increased survival were observed in the low and ultra-low groups, respectively. It is possible these unexpected outcomes may be due to exposure solutions, which were carefully derived from commercially available certified stock solutions but were not analytically verified, which may lead to variability in dosing; additionally, neither stock solution was available at 100% purity, ranging from 95% (PFOA) to 99.4% (PFOS) purity. Impurities of unknown origin, constituting up to 5% of PFOA exposure (0.35, 3.5, and 35 ng/L of the ultra-low, very low, and low concentrations, respectively) and 0.6% of PFOS exposure (0.14, 1.4, and 14 ng/L), could potentially have influenced the results. Overall, these results do not point to a severe risk of bodily harm from environmental-level exposure, though analytical verification of our exposure doses would support a higher-confidence assessment.

A persistent endpoint across all chemicals and generations was alterations in the behavioral response to light and dark stimuli. Larval swimming behavior is used as an indicator of neurotoxicity [27]. By 5 dpf, all major organ systems, including the brain, are functional [27]. Larvae are naturally more inclined to reserve swim bouts for dark periods, where they are less susceptible to predators than in the light [28]. Exposure-induced excitability or lethargy may be modulated by CNS function, which could translate to negative health implications in humans, and erratic behavior could have ecological consequences in aquatic wildlife consistently exposed to PFASs. We report multigenerational behavioral effects in the F1 generation and report for the first time transgenerational PFAS-associated behavioral changes in the completely unexposed F2 generation. The presence of a behavioral phenotype in the F2 generation suggests epigenetic changes induced by F0 exposure. More research is needed to plot the mechanism of this phenotypic inheritance, as well as how animal and human health and ecology are affected by continued PFAS exposure over multiple generations.

PFOA-exposed F0 larvae were hypoactive in both light and dark; this pattern was reversed in the indirectly exposed F1 generation, then returned to the F0 pattern in the unexposed F2 generation. Hyperactivity is typically seen in moderately to highly exposed F0 larvae [12,29,30]; however, exposure sometimes has no behavioral effect [31,32]. It is possible that PFOA exerts a non-monotonic response, wherein ng/L concentrations produce the observed hypoactivity, while higher doses produce hyperactivity. More research at low doses in the F0 generation will be required to draw conclusions. To our knowledge, this is the first study to examine behavior in the F1 and F2 generations of PFOA-exposed F0. The reversal in each generation of the direction of behavior (hypoactivity in F0; hyperactivity in F1; hypoactivity in F2) also suggests the possibility of a neuromodulatory compensation mechanism overcorrecting for the previous generation's propensity for erratic behavior.

PFOS-exposed F0 larvae were hypoactive in the light only; this persisted in the F1 generation. Additionally, F1 larvae were hyperactive in the dark, and F2 larvae were hyperactive under both conditions. In F0 larvae, hyperactivity is generally observed at moderate to high doses [25,29,31–35]. One study at 2 mg/L (20<sup>6</sup> ng/L) found hypoactivity [12], but to our knowledge, this is the first study within a ng/L range, which may account for the diverging effect. Few studies have examined F1, and none at low doses. At moderate doses, Chen et al. observed the exact pattern that we observed of hyperactivity in the dark and hypoactivity in the light [33]; hyperactivity was also observed in other studies [36]. In contrast to PFOA-lineage F2, transgenerational PFOS effects present as totally different from the F0 pattern. The differences in the structure of sulfonic and carboxylic acids are known to exert different effects [31]; this phenomenon appears to continue into the F2 generation.

Mixture-exposed F0 larvae were hyperactive, F1 were hypoactive, and F2 possessed a variable response to dark and light stimuli. Though PFOS was present in a higher concentration than PFOA, PFOS did not appear to overpower PFOA's presence or drive the mixture results as the mixture endpoints were quite different from the PFOS endpoints. Mixtures are generally understudied. Despite PFOS and PFOA being two of the most thoroughly investigated individual PFAS chemicals, their mixture at human levels has not been well-studied for locomotor behavior. However, a complex mixture including both chemicals induced hyperactivity at putative human serum levels [32], though the presence of other chemicals likely influenced the outcome. Very high exposure to a >1 g/L (10<sup>9</sup> ng/L) mixture of nine PFASs in equal amounts was associated with hypoactivity [12]. However, this concentration may have caused lethargy-inducing toxicity as the LC50 of a 1:1 ratio of PFOA:PFOS has been demonstrated at ~37 mg/L (37<sup>6</sup> ng/L) in zebrafish [11]. Given that we are never exposed to a single PFAS alone, and PFAS mixtures have been measured in amniotic fluid [37,38], the lack of knowledge on mixtures across generations necessitates more research, especially at environmentally relevant levels.

Gene expression dysregulation was the most sensitive and persistent endpoint observed across all chemicals and generations. While PFASs are not directly genotoxic, they are known to cause transcriptomic changes [25,29,39]. As exposure to low concentrations of PFASs is understudied, we chose to explore the full transcriptome using RNA-seq rather than targeted expression analysis. This study may provide genes of interest for future biomarkers of effect and for targeted analysis in low-exposure schemes. We report multiand transgenerational effects in gene expression. In fact, for every chemical, more genes were differentially expressed as the generations progressed. The F1 generation showed 2–4× more DEGs than F0. The F2 generation showed 2–3× more DEGs than F1 (and 8–9× more than F0) even in the absence of exposure in the F1 and F2 generations, suggesting epigenetic regulation of the transcriptome. The affected pathways ranged from the immune system, xenobiotic metabolism, and steroid metabolism and synthesis (PFOA) to movement disorders and bone mineral density (PFOS). Surprisingly, the mixture caused relatively few DEGs compared to the single chemicals alone. In general, each chemical was associated with a unique set of DEGs in each generation. However, common to all chemicals in the F1 generation was the dysregulation of *si:dkey-14d8,* and to the F2 generation, *wbp2nl*. Little

is known about the *Danio rerio* gene *si:dkey-14d8*; it is predicted to be involved in collagen fibril organization [40]. The F2 gene *wbp2nl* encodes a sperm protein that promotes oocyte fertilization [41]. This gene was upregulated by PFOS and mixture exposure (log2 fold change 1.20, 0.85, respectively) but downregulated by PFOA exposure (log2 fold change −1.23). In line with our null findings of changes in fecundity, *WBP2NL* expression was not associated with reproductive outcomes in a human study seeking prognostic fertilization factors [42]. *Wbp2nl* is silenced during early development [43]; thus, its activation in the F2 larvae by PFOS and the mixture may indicate aberrant epigenetic programming. While *wbp2nl* is mainly expressed by sperm, it is also found in the breast and kidney [44]—areas known to be PFAS targets [45,46]. It will be interesting for future studies to further phenotypically anchor the diverse transcriptomic pathways of each chemical and the mixture and establish biomarkers of effect for PFAS exposure.

In PFOA-exposed F0 larvae, *rpe65a* and *atp6v0e1* were downregulated. Downregulation of *rpe65a* is associated with retinal degeneration in zebrafish [47], and loss of RPE65 function leads to blindness in humans [48]. *Atp6v0e1* is involved in visual–motor behavior [49]. The downregulation of optical-related *rpe65a* and *atp6v0e1* may have contributed to the hypoactivity we observed in PFOA-exposed F0 larvae. In the F1 generation, xenobiotic pathways predominated, driven by the upregulation of *cyp3a7*. PFASs are known xenobiotic inducers of PXR and CAR pathways in humans and rodents; in zebrafish, these receptors have been shown to be unresponsive to PFOA [50]. However, as these data seek to inform human health, the change in *cyp3a7* expression implicating PXR and CAR activation is still meaningful. Additionally, *CYP3A7* in humans is enriched in fetal liver [51], underscoring the relevance of the embryonic zebrafish exposure model to human developmental health. Another upregulated molecule in xenobiotic pathways was *dusp16*, which has a role in immune function [52]. Though immune dysregulation does not feature prominently in the PFOA F1 pathway profile, PFASs are a demonstrated immunotoxicant [53], and the *dusp* genes appeared in DEGs of the F0 generation and additionally in the pathway analysis of the F2 generation. In addition to immune system pathway disruption in the F2 generation, steroid synthesis was affected, and the glucose-homeostasis-related pancreatic gene *amy2al2* was the most upregulated DEG at every concentration. PFOA has been shown to increase steroid hormone levels in zebrafish larvae [54], and links between PFOA serum levels and diabetes risk have been established in humans [55–57]. Immune dysfunction appears to be a significant outcome of low-level PFOA exposure, though effects may not be seen until later generations.

In PFOS-exposed F0 larvae, *irs2a* and inflammatory response gene *irg1l* were upregulated, and the kinesin gene *kif3c* was downregulated. Besides its known glucose metabolism function [56–58], *irs2a* has an emerging function in hypoxia protection [59,60]. Hypoxia and inflammation, in combination with decreased expression of the photoreceptor *kif3c* gene [61], may have contributed to the hypoactivity we observed in PFOS-exposed larvae. Others have found downregulation of the histamine H1 receptor [32] and steroidogenic enzymes [29,62] at moderate to high exposure. Relatively few genes were differentially expressed in the F0 generation as compared to the F1 and F2 generations. In the F1 generation, DNA-binding genes *satb1a* and *npas4a* were two of the most upregulated genes. *Satb1a* and *npsa4a* expression is localized to the CNS in larvae [63,64]. In line with the present behavioral results of hyperactivity from very low exposure, *npas4* expression is increased in response to neuronal activity [64]. Less is known about *satb1a* in zebrafish. In humans, *SATB1* remodels chromatin in thymocyte differentiation into T-cells [65,66]. Pathway analysis results included lowered chemotaxis of immune cells, increased steroid synthesis, survival of neuronal cell types, and movement disorders. F1 larvae were the only group across all chemicals and generations where the direction of behavior (hyper- or hypoactivity) had no agreement between light and dark conditions. The upregulation of neuronal activity gene *npas4* may have contributed to hyperactivity in the dark; the movement disorder pathway is one molecular indication of the contrasting responses to light and dark. Other studies in zebrafish have not examined and compared for gene expression with

F1-lineage behavior. As in F1, F2 larvae showed increased pathways involving steroids and, additionally, other lipids. The pancreatic gene *cela1.5* was the most upregulated DEG at ultra-low and very low exposures and was moderately upregulated at low exposure. Lipid metabolism disruption has been previously linked to PFOS exposure at moderate levels in F0 fish [67], but we did not observe this effect until the F2 generation. Bone mineral density and other connective tissue pathway disruptions were also a PFOS F2-specific occurrence. Increased lipid pathways and decreased connective tissue pathways do not seem to explain the observed hyperactivity in the F2 generation; however, a non-specific movement disorder pathway was also increased. More research is required in F2 larvae to fully understand the scope of the ancestral effects of PFOS exposure. In most generations, PFOS exposure was associated with pathways of increased lipid synthesis, which complements the thoroughly-studied PFOS-associated high cholesterol in humans [68–70].

Mixtures are a rapidly expanding field of research, and low levels are highly relevant to human health. In mixture-exposed larvae, few DEGs were expressed in all generations compared to individual PFASs (Figure 2). The F0 generation exhibited dysregulation of genes involved in basic cellular processes, with no obvious influence on the observed hyperactivity. Similarly, in their assessment of behavior and gene expression in a complex mixture including PFOS and PFOA, Khezri et al. [30] could not rationalize a clear link between exposure-associated hyperactivity and DEGs. More research is certainly needed to elucidate the complex transcriptomic dynamics underpinning behavioral outcomes in mixtures. The F1 generation showed dysfunction in pancreatic genes *zgc:92590*, *cela1.5,* and *cpa4*. F1 downregulation in optic-related gene *rlbp1b* [71] could have contributed to the observed hypoactivity of larvae in light and dark. F2 larvae downregulated the muscular gene *tcap*, yet upregulated growth genes *fzd6* and *fgfbp2*. When significant DEGs from all generations were collated, pathway analysis revealed immune dysfunction and developmental deficits predicting organismal and cell death; however, mortality was unaffected in any generation. The present results suggest mixture exposure does not cause overt harm in any generation; however, the transcriptome of developmentally exposed fish may be an early indicator of latent embodied effects. Perhaps a longer experiment with aged fish would reveal mixture-associated latent mortality.

Egg production in females was measured to estimate fecundity in the F0 and F1 generations. Changes in fecundity may have implications for reproductive health, the offspring, as well as the ecosystem. In humans, there is no consensus on fecundity and PFAS exposure, possibly owing to the multiple ways of defining fecundity in humans. Multiple epidemiological studies have found a decrease in fecundity with PFOA or PFOS [72–75], while some have found no effect in either [76,77]. There was no effect on the fecundity of F0 exposure to PFAS. Of note, controls in the PFOA F0 group produced significantly fewer eggs than controls in the PFOS (*p =* 4 × 10−<sup>3</sup> ) but not mixture (*p =* 0.15) groups (one-tailed *t*-test) (PFOS and mixture controls were not significantly different (*p =* 0.22, two-tailed *t*-test)). When egg production in PFOA-exposed larvae was compared to control data from the PFOS and mixture larvae, there was actually a significant decrease in egg production at low exposure (*p =* 2.1 × 10−<sup>2</sup> ) (data not shown). Decreased zebrafish egg production was observed in another study on low-level exposure to PFOA [37] and at moderate exposure in the crustacean *Daphnia magna* [78]; however, in wild-caught fish, hepatic levels of PFOA had no association with fecundity [79]. In the F1 generation, very low exposure lineage fish produced significantly fewer eggs. In the only other transgenerational study of fecundity, Marziali et al. [80] found no effect in F0–F2 in harlequin flies. In PFOS studies, fecundity is found to be either decreased at moderate doses [78,81] or to have no effect [34,77], including no effect in F1 and/or F2 [34,78]. In an F0–F2 study of a moderate dose PFAS mixture containing low doses of PFOA and PFOS on Japanese medaka, Lee et al. [82] reported no significant effect. Overall, the present results suggest little to no effect on fecundity in F0–F1 zebrafish exposed to low levels of PFAS.

Sex determination in laboratory zebrafish is polygenic and is thought to be influenced by their environment, which can include exposure to contaminants [83]. Alterations in the male:female ratio are thus a common endpoint in endocrine disruption studies, with a shift in either direction indicating disruption. In the F0 generation, there was no significant change in sex ratio following exposure. In the F1 generation, the very low level of the mixture and every concentration of PFOA caused an increase in the ratio (significantly more males). The authors note the abnormal lack of males in the PFOA control group, which may have led to a false-positive result. Additionally, due to a lack of access to research animals during the SARS-CoV-2-related institutional shutdown, only one cohort of PFOA F1 and two cohorts of mixture F1 fish could be observed. No changes were observed in the PFOS-exposed sex ratio in either generation. Other studies have observed a decreased ratio following F0 PFOS exposure [34]. Exposing the F0 and F1 generations to a mixture of four PFAS, including PFOA and PFOS, did not result in any shift [82]. The scarcity of replicates for all groups and abnormal PFOA control fish do not allow meaningful conclusions to be drawn from this endpoint in the present study.

As each generation was differentially affected by each chemical, it is pertinent to summarize the similarities and differences of each chemical's discussed effects in the F0, F1, and F2 generations separately. Additionally, human health research and policies are mainly concerned with the directly exposed subject. In the F0 generation, PFOA exposure was associated with hypoactivity, with a potential visual–motor impact occurring via the downregulation of vision-related genes *rpe65a* and *atp6v0e1* [47,49]. Similarly, PFOS-exposed larvae were also hypoactive in the light and had a downregulated photoreceptor gene (*kif3c*) [61]. Ophthalmic health should be observed more closely in future studies examining behavior response to visual cues, especially as zebrafish eyes are in constant contact with the exposure solution. The mixture-exposed larvae were, in contrast, hyperactive and showed no clear disruption of a particular pathway, dysregulating instead the genes involved in basic cellular processes. No chemical was associated with a significant change in adult fecundity, body weight, length, or sex ratio. In the F1 generation, PFOA exposure was associated with hyperactivity and xenobiotic response. PFOS-related behavior varied by light or dark status in the only locomotor disagreement in the study; upregulated CNSrelated genes could account for the hyperactivity in the dark. The mixture larvae displayed hypoactivity and dysfunctional pancreatic genes. Additionally, the downregulation of optic-related gene *rlbp1b* [71] in the mixture larvae could complicate behavior results, as in the F0 generation. Overall, each chemical was associated with disparate pathways in the F1 generation, in line with different behavioral patterns across the chemicals. No chemical was associated with a reliable change in adult fecundity or sex ratio. In the F2 generation of PFOA- and PFOS-exposed larvae, pancreatic genes were most affected, likely leading to the observed alterations in hormone-related pathways. Additionally, immune pathways were affected in the PFOA and mixture groups. Each chemical in the F2 generation was associated with a different behavioral pattern (PFOA: hypoactivity; PFOS: hyperactivity), with the mixture showing both hyper- and hypoactivity. In sum, the evidence points to varying effects of PFASs depending on both the specific chemical and degree of exposure.

This study provides the first report on the multigenerational effects of environmentallevel PFOA exposure on zebrafish behavior and of a mixture of the two chemicals on behavior and gene expression. Further, it is the first report of the transgenerational effects of PFOA, PFOS, and a 1:1 mixture in terms of behavior and transcriptomics. The next steps in this line of research will be to examine the epigenetic influences set in motion by these PFASs. Effects onto the F2 generation have been reported in PFBS [84] and PFOSalternative F-53B exposure [85] at moderate levels. Intriguingly, low-level PFAS exposure in the present study continued to exert effects generations after exposure cessation. Gene expression dysregulation increased as the generations progressed, with F2 exhibiting far more DEGs than F0, suggesting epigenetic regulation of expression in the absence of a chemical stressor. In general, the DEGs in each generation and in each chemical had little overlap. Interestingly, the mixture had a relatively small influence on the number of DEGs compared to the individual PFASs. The unique suites of DEGs underscore the differential effects of different functional groups of PFASs and individual PFASs versus a mixture and

suggest different mechanisms of action in the production of the observed transcriptomic signatures and behavioral phenotypes.

It is the authors' aim that these results inform decision-making regarding safe contaminant limits in drinking water, food sources, and aquatic habitats. Future studies into the mechanisms of epigenetic dysregulation under exposure will be of great interest. PFAS replacements, including "short-chain" alternatives to PFOA and PFOS, will be critical to study as well, both individually and in environmentally relevant mixtures.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/toxics10060334/s1. Figure S1: Schematic of experimental design for obtaining embryos and for fecundity assay. Figure S2: Box plots of F0 and F1 fecundity data. Table S1: Replicates and *n*-values for all experiments. Table S2: Endpoint data and significance reporting for survival, abnormalities, behavior, fecundity, and sex ratio. Table S3: F0 DEGs. Table S4: F1 DEGs. Table S5: F2 DEGs. Table S6: IPA summaries. File S1: Custom R code for analysis of behavioral data.

**Author Contributions:** Conceptualization, T.R.B. and A.H.; methodology, A.H., C.A., D.N.M., M.A., M.C., A.K. and D.J.; formal analysis, A.H. and C.-C.W.; resources/funding acquisition, T.R.B.; writing: A.H.; editing: T.R.B. and A.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** Funding was provided by Great Lakes Water Authority; the National Center for Advancing Translational Sciences (K01 OD01462 to T.R.B.), the National Institute of Environmental Health Sciences (R01 ES030722 to T.R.B., A.H., and D.N.M.; P30 ES020957 to D.N.M. and T.R.B.; F31 ES030278 to D.N.M.), the National Institute of General Medicine Sciences (R25 GM 058905 to D.J.), and the WSU reBUILD program (to M.C. and M.A.).

**Institutional Review Board Statement:** All zebrafish use protocols were approved by the Institutional Animal Care and Use Committee at Wayne State University, according to the National Institutes Health Guide to the Care and Use of Laboratory Animals (Protocol 16-03-054; approved 4 August 2016).

**Acknowledgments:** We acknowledge Emily Crofts, Kim Bauman, and all members of the Warrior Aquatic, Translational, and Environmental Research (WATER) lab at Wayne State University for help with zebrafish care and husbandry. We would like to acknowledge the Wayne State University Applied Genomics Technology Center, especially Katherine Gurdziel, for providing sequencing services and the use of Ingenuity Pathway Analysis software.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

