Ethylene Glycol Monomethyl Ether Altered Rat Sperm Small RNAs with Critical Developmental Roles

testis, germ cells, spermatozoa, RNA, sncRNA 13 Running Head: EGME dose-dependent sperm sncRNAs 14 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 1

15 Ethylene glycol monomethyl ether (EGME) is a testicular germ cell toxicant that selectively 16 targets spermatocytes. In rats, male-only EGME exposure reduces mating success and can 17 lead to an increase in resorbed fetuses. In a previous study, five-day exposure to 50, 60, or 75 18 mg/kg/d EGME in male rats led to a decrease in sperm motility and increase in retained 19 spermatid heads with a LOAEL of 75 mg/kg/d. At 60 mg/kg/d, EGME exposure altered the 20 proportion of sperm small RNA reads mapped to different small RNA categories and the 21 distribution of read lengths. Because there is evidence that small non-coding RNAs (sncRNAs) 22 in sperm regulate embryonic development, we analyzed sperm sncRNA data from EGME-23 treated male rats to identify differential expression at the individual RNA level. EGME treatment 24 resulted in dose-dependent increases in the expression levels of microRNAs (miRNAs), 25 piRNAs, and tRNA-derived small RNAs (tsRNAs). We identified 12 miRNAs that were 26 differentially expressed at all EGME doses, with a monotonic, dose-dependent increase. High-27 confidence targets of these 12 miRNAs are known to be expressed in pre-implantation embryos 28 and statistically enriched for Gene Ontology (GO) biological processes related to early 29 development, such as cell fate commitment and regulation of developmental growth. These 30

demonstrated that the EGME-induced changes in sperm sncRNA levels were 31 reproducible, dose-dependent, and provided a putative mechanism of paternal EGME effects on 32 embryonic development, which will be investigated in future studies. 33 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 2

34 Ethylene glycol monomethyl ether (EGME) is a well-characterized testicular toxicant in 35 the rat. Upon oral exposure, EGME is metabolized to the active metabolite, 2-methoxyacetic 36 acid (MAA) (Beattie and Brabec 1986). EGME selectively targets pachytene and meiotic 37 spermatocytes, resulting in apoptosis (Creasy and Foster 1984; Creasy et al. 1986). EGME-38 induced germ cell death is rapid, detectable at 12 hours after a dose of 250 mg EGME/kg body 39 weight (Creasy et al. 1986) and 24 hours after a dose of 150 mg EGME/kg body weight in rats 40 (Chapin et al. 1984). Perhaps more interestingly, EGME is an early example of a non-genotoxic 41 chemical (Hoflack et al. 1995) for which paternal exposure causes adverse effects on embryo 42 development. Short-duration exposure to EGME in rats leads to post-implantation embryo loss 43 that is delayed by five weeks (Chapin, Dutton, Ross, and Lamb 1985), a time frame that 44 corresponds with the time it takes for pachytene spermatocytes to mature into spermatozoa and 45 complete epididymal maturation. 46 The mechanisms of preconception exposure developmental toxicity are not well 47 understood. Given that there is no evidence that EGME or its major metabolite, MAA, are 48 genotoxic, it stands to reason that EGME has the potential to disrupt embryo development 49 through an epigenetic mechanism. One plausible epigenetic mechanism is disruption of sperm 50 RNA. Sperm contains mRNA, miRNA, piRNA, tsRNA, and other classes, which have putative 51 roles in embryo development (Sendler et al. 2013; Jodar et al. 2015; Conine et al. 2018). These 52 sperm RNAs are well-conserved across mammalian models; human, mouse, and rat sperm all 53 contain thousands of mRNAs, with one report identifying 6,684 transcripts expressed in all three 54 species (Bianchi et al. 2021). In particular, there is evidence for a critical role of miRNAs and 55 endo-siRNAs in the male germline. Germ-cell specific loss of Dicer or Dgcr8 from embryonic 56 day 15, driven by Dddx4-Cre in mice, leads to infertility and severe reduction of meiotic 57 progression and maturation of elongate spermatids (Zimmermann et al. 2014). Furthermore, 58 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 3 there is increasing evidence that altered levels of regulatory RNAs in sperm contribute to 59 disease phenotypes such as metabolic disorder in offspring (Zhang et al. 2019). 60 There is evidence that EGME alters testicular gene expression and miRNA expression, 61 which suggests that sperm RNAs may be altered following EGME exposure, as well. In 62 monkeys exposed to EGME, expression of many testicular miRNAs increases, while serum 63 miRNA levels decrease (Sakurai et al. 2015). In rats, EGME exposure results in changes in 64 expression of meiosis-related genes in the testis (Tonkin et al. 2009). Further, we previously 65 reported that low-dose EGME alters the proportions and read length distributions of several 66 classes of small RNA in rat sperm (Stermer et al. 2019). Rats were treated with 50-75 mg 67 EGME/kg body weight/d for 5 days. After 5 weeks with no treatment, no testicular 68 histopathology was observed at doses lower than 75 mg/kg/d. However, at 50 mg/kg/d, the 69 proportion of miRNA reads relative to all small RNA reads increased, and the distribution of 70 piRNA and tsRNA read length was significantly altered. Here, we conducted a new analysis of 71 that dataset to determine the impact of EGME exposure on expression of specific sperm small 72 RNAs, with the goal of identifying dose-dependent regulatory RNA changes in sperm. These 73 sensitive, dose-dependent changes have the potential to serve as biomarkers of exposure to 74 EGME and possibly other toxicants, as well as candidate RNAs in the mechanisms of 75 developmental toxicity of paternal EGME exposure. 76

77 Raw small RNA sequencing data were obtained from a previously published study 78 (Stermer et al. 2019). Briefly, single-end sequencing of small RNA was performed on the 39 79 libraries (one sample in the 60mg/kg exposure group lacked sufficient RNA to process) with an 80 IonProton sequencer (ThermoFisher Scientific). This yielded an average of 16 million reads per 81 library and an average read length of 30nt. 82 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 4 Small RNA Expression Profiling 83 Initial quality control was performed using FASTQC (Andrews 2010). Two samples with 84 extremely low read counts were identified as outliers and excluded from downstream analysis. 85 Small RNA expression levels were then profiled using sRNAbench (Aparicio-Puerta et al. 2022) 86 in genome mode. Reads were mapped against the rat genome (Rnor_6_0) and compared to 87 miRBase Release 22.1 (Kozomara et al. 2019). Default sRNAbench parameters were applied: 88 the bowtie seed alignment, seed length 20, minimum read count 2, minimum read length 15, 89 allowed 1 mismatch, and 10 multiple mappings. After expression profiling, multi-map adjusted 90 read counts for small RNAs (miRNAs, tRNAs, ncRNAs, cDNA) from sRNAbench ".grouped" files 91 were parsed using a custom Python script. 92 Differential Expression Analysis 93 Differential expression of small RNAs across dosage groups was determined using 94 DESeq2 (Love et al. 2014) with default filtering. Both the Wald test and the likelihood ratio test 95 were run for each small RNA type. For miRNAs, significantly differentially expressed miRNAs (p 96 < 0.05) were identified for each of the three dosage vs. control pairs (50mg/kg vs 0 mg/kg, 97 60mg/kg vs 0 mg/kg, and 75mg/kg vs 0mg/kg). The intersection of these significant miRNA sets 98 across all comparisons was used for further analysis. 99 As RNAseq methods such as DESeq2 have high false positive rates (Li et al. 2022), we 100 assessed robustness of our results with a Monte Carlo simulation (Dere et al. 2016; Dere et al. 101 2018). We created 1000 subsamples of our dataset where each subsample consisted of smaller 102 groups of 6 individuals in each group, where the individuals were sampled without replacement 103 from the larger set of 10 individuals per group. We then ran DESeq2 again on each of the 1000 104 subsamples and counted how many times we found the same differentially expressed miRNAs. 105 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 5 Target Gene Prediction and Functional Enrichment 106 Putative target genes for the intersection set of significant miRNAs were assembled by 107 manually querying each miRNA in miRDB (Chen and Wang 2020). Targets with scores greater 108 than 80 were retained, yielding 2073 unique genes. An over-representation analysis (ORA) was 109 performed on the full list of target genes, using the R package ClusterProfiler (Yu 2024). The 110 default background gene list (whole rat genome) was used, as gene expression was not 111 measured in this experiment. All three subontologies - biological process (BP), molecular 112 function (MF), and cellular component (CC) were investigated. All the other parameters are left 113 as default. For each subontology, we obtained a list of GO terms ranked by p-values. 114

115 EGME dose-dependently increased sperm miRNA levels 116 A total of 405 miRNAs were initially identified by sRNAbench, with 277 remaining after 117 filtering in DESeq2. Sperm miRNA levels showed a high degree of clustering by treatment level, 118 based on Euclidean distance (Fig. 1A), with two main clusters forming, consisting of (1) all 10 119 vehicle controls, four 50 mg/kg/d EGME samples, and two 60 mg/kg/d EGME samples; and (2) 120 the remaining six 50 mg/kg/d samples, seven 60 mg/kg/d samples, and all ten 75 mg/kg/d 121 samples. In principal component analysis (PCA) (Fig. 1B), there was a clear gradient from 122 vehicle to 75 mg/kg/d along the first principal component. The first two principal components 123 explained 25 and 13% of the variance in the data, respectively, and the samples became more 124 variable within treatment group as the treatment level increased, suggesting that increased 125 EGME-induced germ cell stress led to greater variability. There was distinct visual separation in 126 the PCA plot between sample groups, in particular the vehicle control and 50 mg EGME/kg/d, 127 with less separation between the 60 and 75 mg EGME/kg/d groups. 128 The total number of DE miRNAs increased from 12 in the 50 mg/kg/d group to 56 in the 129 60 mg/kg/d group and 57 in the 75 mg/kg/d group, with the majority of significant DE miRNAs 130 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 6 being upregulated in the EGME treatment groups (Fig. 1C, Table S1). There was significant 131 overlap between the treatment groups, with 12 miRNAs being shared among all three groups, 132 and a further 25 DE miRNAs shared by the 60 and 75 mg/kg/d groups (Fig. 2A). A Monte Carlo 133 analysis found that these 12 shared miRNAs were highly stable among the DE miRNAs (Fig. 134 2B). In 1000 simulations, 10 of the 12 shared DE miRNAs were in the top 12 most frequently 135 identified DE miRNAs, appearing in 748 to 992 out of 1000 iterations. The remaining two shared 136 DE miRNAs were also in the top 40 most stable DE miRNAs. The 12 shared DE miRNAs were 137 miR-200b-3p, miR-221-3p, miR-200a-3p, let-7c-5p, miR-429, miR-365-ep, miR-27a-3p, miR-138 205, miR-148-3p, miR-23a-3p, miR-24-3p, and let-7b-5p. All 12 monotonically increased across 139 the dose range (Fig. 2C). 140 EGME dose-dependent miRNA targets in embryo development 141 We used miRDB to identify the high-confidence targets of the 12 EGME dose-dependent 142 DE miRNAs, using a miRDB target score > 80 as the cutoff. At this level of confidence, we 143 identified 1,203 targets (Table S2). 787 of the 1,203 target genes were only targets of one 144 miRNA. A further 343 were targets of two miRNAs, and 50, 12, and 7 genes were identified as 145 shared targets of 3, 4, or 5 of the 12 core miRNAs, respectively (Fig. 3A). The most highly 146 shared target genes were Zeb1, Mmd, Clasp2, Hectd2, Wdr37, Sema6d, and Slc6a1, which 147 were each identified as a target of 5 of the 12 shared DE miRNAs. To assess whether these 148 target genes are likely to have roles in embryonic development, we compared the targets with 149 mRNAs enriched in the epiblast (EPI), primitive endoderm (PE), and trophectoderm (TE) of 150 human e5-7 embryos (Petropoulos et al. 2016). The targets matched 54, 31, and 36 of the EPI, 151 PE, and TE-enriched genes, respectively (Fig. 3B, Table S2), which confirms that the EGME 152 dose-dependent miRNAs target genes involved in early development and that at least a subset 153 of those genes are expressed in early embryos. That subset included well-known early embryo 154 development genes, such as Aldh1a1, Foxa2, Gata2, Gata4, Gata6, as well as fundamental 155 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 7 genes involved in cell cycle control, cytoskeleton structure and remodeling, and cell signaling. 156 Using the 1,203 target transcripts as input, we identified 321 enriched Gene Ontology (GO) 157 terms (Table S3), including 295 biological process (BP), 9 cellular compartment (CC), and 17 158 molecular function (MF) terms (Fig. 3C-D). The top 20 GO BPs included important 159 developmental functions, such as “axonogenesis,” “regulation of developmental growth,” “cell 160 fate commitment,” “cell fate commitment,” “morphogenesis of a branching structure,” and 161 “morphogenesis of a branching epithelium” (Fig. 3D). Genes under the “cell fate commitment” 162 term, as an example, have well-characterized roles in development. These include Cyp26b1, 163 Dmrt3, Foxa2, Gata2, Gata4, Gata6, Hoxd10, Kdr, Pax6, Pou3f2, Pparg, Runx1, Runx2, 164 Smad4, Tgfb2, Tgfbr1, Wnt9a, and Wnt9b, among 40 target genes (Table S3). Analysis of 165 target gene sharing among the 12 miRNAs in the GO BP gene sets revealed a high degree of 166 overlap. Of the significant GO BPs, the majority included targets of 6 or more miRNAs, with only 167 one GO BP uniquely linked to a single miRNA (Fig. 3C). 90 out of 295 pathways were shared by 168 all 12 miRNAs, although most target genes were not shared by a large number of miRNAs (Fig. 169 3A). Beyond biological processes, the top GO CCs included compartments associated with cell 170 division, including “cleavage furrow,” and “cell division site,” as well as terms related to 171 phosphatases, Golgi, and neuronal synapses (Fig. 4A). The top GO MFs included several terms 172 related to kinase and phosphatase activity, transcription, RNA binding, and SMAD binding (Fig. 173 4B). 174 EGME effects on sperm tsRNA levels 175 Small RNA reads mapped to 365 tRNA genes in sRNAbench, which corresponded with 176 41 unique tRNAs (Fig. 5, Table S4), after removing duplicate tRNA genes (Iben and Maraia 177 2014). We performed unsupervised analyses of tRNA reads, which showed separation of 178 samples by tRNA expression, similar to the pattern observed with miRNAs. In hierarchical 179 clustering based on Euclidean distance (Fig. 5A), there were four main clusters, showing a lot of 180 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 8 separation between the control and high dose group. One cluster contained eight of nine 75 181 mg/kg/d samples and three of nine 60 mg/kg/d samples, while another contained seven of nine 182 vehicle control samples, four 50 mg/kg/d samples, and one 60 mg/kg/d sample. PCA showed an 183 even clearer separation between dose levels than for miRNAs, with the vehicle and 50 mg/kg/d 184 groups clustered somewhat closely and the 60 and 75 mg/kg/d groups displaying a gradient 185 along principal component 1 (Fig. 5B). The first two principal components explained 69% of 186 variance in the dataset (46% and 23%, respectively). The number of significant DE tRNA genes 187 (Wald test, fold change ≥ 2, p-adj < 0.05) was 9, 117, and 204 in the 50, 60, and 75 mg/kg/d 188 treatment groups, respectively, relative to control (Fig. 5C). After removing duplicate tRNA 189 genes and comparing at the tRNA level, there were 2, 28, and 39 significant tRNAs at the 50, 190 60, and 75 mg/kg/d levels; at least one GluCTC and GlyCCC gene was significant at each of the 191 three dose levels; and of the 24 additional tRNAs with at least one significant gene at the 60 and 192 75 mg/kg/d dose levels, GluTTC, LysCTT, AlaAGC, and CysGCA had the lowest adjusted p-193 values at 50 mg/kg/d (Fig. 5D). We generated coverage maps to analyze the tsRNA fragments 194 present in the 2 consensus DE tRNAs and the next four most significant tRNAs (Fig. 5E), which 195 showed the greatest coverage for GlyCCC, and mostly 5’ coverage for all except for GluTTC, 196 which had more 3’ coverage. When analyzing coverage at the single-read level (Fig. 6), for 197 GluCTC, LysCTT, AlaAGC, and Cys GCA, the vast majority of reads were 5’ fragments or 198 halves, with a much smaller number of reads representing 3’ fragments. The reason for the 199 different coverage map for GluTTC was that reads consisted of an approximately 50:50 200 distribution of 5’ fragments and halves and 3’ fragments and halves. 201 EGME effects on sperm piRNAs 202 Small RNA reads mapped to 18,027 unique piRNAs in sRNAbench, using annotations 203 from piRBase (Wang et al. 2019). piRNAs separated by treatment group in unsupervised 204 analyses, although less separation was visible in hierarchical clustering (Fig. 6A) than for other 205 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 9 categories of RNAs. Similarly, PCA showed a gradient of values along PC1 with a cluster of 0, 206 50, and 60 mg/kg/d EGME samples mostly separating from a cluster of 75 mg/kg/d samples 207 (Fig. 6B). The first two PCs explained 74% of the variance in the dataset (67% and 7% in the 208 first two PCs, respectively). An increasing number of piRNAs were significant in DE analysis 209 with each treatment group (Fig. 6C), with 1, 244, and 2,181 DE piRNAs in the 50, 60, and 75 210 mg/kg/d groups, respectively. This large number of DE piRNAs is consistent with the size of this 211 RNA category in spermatogenesis, which comprises more than 1 million piRNAs that are critical 212 in the germline of rats and mice (Lau et al. 2006; Beyret et al. 2012; Watanabe et al. 2015). 213 EGME effects on other sperm small RNAs 214 EGME exposure level led to changes in sperm levels of multiple classes of small RNAs 215 in addition to miRNAs, tsRNAs, and piRNAs (Fig 7, Table S6). These included rRNAs, 216 mitochondrial rRNAs, snRNAs, lincRNAs, antisense RNAs, processed transcripts, and RNAs 217 without known identities (identified as misc_RNA in Table S6). There was strong separation by 218 unsupervised clustering analyses, similar to the pattern observed with miRNAs and tsRNAs. 219 The hierarchical clustering result included two main clusters, one containing one 50 mg/kg/d 220 sample, seven 60 mg/kg/d samples, and all nine 75 mg/kg/d samples; while the other cluster 221 contained two 60 mg/kg/d samples, nine 50 mg/kg/d samples, and all nine vehicle control 222 samples (Fig. 7A). PCA showed strong separation along the first PC. However, the first two PCs 223 explained only 48% of the variance in the dataset (37% and 11%, respectively) (Fig. 7B). The 224 number of DE RNAs in each group was 16, 79, and 133, in the 50, 60, and 75 mg/kg/d EGME 225 groups, respectively. 226

227 Ethylene glycol monomethyl ether (EGME), and its active metabolite, 2-methoxyacetic 228 acid, is a testicular germ cell toxicant that causes selective spermatocyte toxicity in the rat 229 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 10 (Chapin and Lamb 1984; Creasy and Foster 1984; Foster et al. 1984; Gray et al. 1985; Moss et 230 al. 1985; Creasy et al. 1986), resulting in apoptosis (Ku et al. 1995). Meiotic spermatocytes are 231 most sensitive to EGME, followed by pachytene spermatocytes, leptotene/zygotene 232 spermatocytes, and step 1 round spermatids (Creasy and Foster 1984; Foster et al. 1984).The 233 NOAEL dose for germ cell death has been reported as 50 mg/kg/d following 11-day exposure to 234 EGME and 250 mg/kg EGME in a single-dose experiment (Creasy et al. 1986). EGME-induced 235 germ cell death is preceded by mitochondrial damage 16 hours after a single 500 mg/kg dose 236 (Foster et al. 1983). 237 Because EGME selectively targets spermatocytes, an exposure duration of as little as 238 five days results in decreased male fertility that becomes apparent three to four weeks after 239 cessation of exposure, with the greatest magnitude of effect occurring in weeks five to six post-240 exposure, with a LOAEL of 100 mg/kg/d (Chapin, Dutton, Ross, and Lamb 1985; Chapin, 241 Dutton, Ross, Swaisgood, et al. 1985; Holloway et al. 1990). The delayed fertility effect 242 corresponds with the 4-5 weeks required for spermatocytes to mature into sperm. A dose of 200 243 mg/kg/d has also been reported to increase the number of resorptions five to six weeks after a 244 five-day exposure (Chapin, Dutton, Ross, and Lamb 1985), suggesting that paternal EGME 245 exposure can cause developmental toxicity without exposure of the dam or embryo. Given 246 evidence that EGME is non-genotoxic (Hoflack et al. 1995), a mechanism other than germ cell 247 DNA damage must be responsible for those paternal exposure effects on embryonic 248 development. 249 Stermer et al. (2019) used low-dose EGME as a model to understand how sub-cytotoxic 250 germ cell effects of EGME could alter markers of reproductive health. Exposure of male rats to 251 50-75 mg/kg/d EGME for five days, followed by a five-week withdrawal, resulted in reduced 252 sperm count and increased retained spermatid head (RSH) count, with a NOAEL of 60 mg/kg/d 253 and a LOAEL of 75 mg/kg/d (Stermer et al. 2019). In addition to these apical effects, all doses of 254 EGME that were tested (50-75 mg/kg/d) altered the proportion of miRNA, piRNA, and tsRNA 255 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 11 reads in sperm and the distribution of small RNA read lengths. An ROC curve analysis revealed 256 that average sperm small RNA read length was a better EGME dose classifier than sperm 257 motility or RSH. In the present paper, we performed a new analysis of these data to identify the 258 sperm small RNAs that were differentially abundant following EGME exposure, to obtain novel 259 mechanistic insights about EGME-driven changes in spermatogenesis and to identify RNAs that 260 may contribute to differences in embryonic development following paternal EGME exposure. We 261 identified EGME dose-dependent changes in abundance of sperm miRNAs (Figs. 1-2), tsRNAs 262 (Fig. 5-6), piRNAs (Fig. 7), and other classes of small RNA (Fig. 8). 263 miRNAs canonically act as negative translational regulators of their mRNA targets, but 264 they can act as positive regulators in some instances (O’Brien et al. 2018). There is evidence 265 that sperm miRNAs regulate pre-implantation embryonic transcription in the mouse, at the 4-cell 266 and blastocyst stages (Conine et al. 2018). In our analysis, the number of significant 267 differentially expressed (DE) miRNAs increased with EGME dose (Fig. 1), and there were 12 268 DE miRNAs present in all three doses, which all showed a monotonic increase in abundance 269 with dose, and which were highly reproducible based on a Monte Carlo analysis (Fig. 2). These 270 12 miRNAs have over 1,000 high confidence targets, many of which are involved in 271 developmentally important biological processes and expressed in human preimplantation 272 embryos (Fig. 3B-D, Tables S2, S3). Several of the EGME dose-dependent sperm miRNAs 273 have known functions in spermatogenesis, and/or embryo development, including miR-221-3p, 274 miR-27a-3p, miR23a-3p, and let-7b-5. miR-221-3p regulates germ cells to maintain 275 undifferentiated spermatogonia states (Yang et al. 2013). miR-27a-3p is overexpressed in 276 azoospermic men (Norioun et al. 2020), and miR-23a/b-3p are overexpressed in subfertile men 277 (Abu-Halima et al. 2019). Let-7 miRNAs regulate germ cell transcription, with let-7b being 278 especially highly expressed in pachytene spermatocytes (Sangiao-Alvarellos et al. 2015). We 279 hypothesize, therefore, that levels of miRNAs such as let-7b could be altered in sperm following 280 an injury to spermatocytes and could influence embryo development as a result. In addition to 281 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 12 those miRNAs that have previously characterized roles in male reproductive biology, it is 282 notable that let-7c-5p, miR-200a-3p, and miR-200b-3p increased with EGME dose in our 283 dataset (Fig. 2), but decreased during epididymal maturation in the mouse in (Sharma et al. 284 2018). 285 Although miRNAs are well-characterized, recent evidence shows that the majority of the 286 RNAs in sperm are tsRNAs and rsRNAs, which are underrepresented in sequencing results 287 because of modifications that impede reverse transcription (Shi et al. 2021; Hernandez et al. 288 2023). Although the sequencing approach from which the present EGME dataset is derived did 289 not use the PANDORA-seq methodology, a significant proportion of the reads in this dataset 290 mapped to tRNA sequences: 51.4-65.5%, depending on EGME dose (Stermer et al. 2019). 291 Consistent with that finding, we identified 41 DE tRNAs, including 2 that were DE in all EGME 292 doses, GluCTC and GlyCCC (Fig. 5). Coverage maps showed that the tsRNAs that contributed 293 to these differential read counts were predominantly 5’ tRNA fragments and halves (Figs. 5-6). 294 tsRNAs may have multiple functions, including miRNA-like RNAi functions or aptamer functions 295 (Chen and Zhou 2023) or regulation of transposable elements (Sharma et al. 2016); however, 296 there is currently no broad understanding of tsRNA functions. Although there are still many 297

to understanding the functions of tsRNAs, recent evidence shows that the sperm 298 quantity of one of our DE tsRNAs, 5’tRF-GluCTC, is negatively associated with ART outcomes 299 in humans (Hekim et al. 2025). 300 EGME caused a massive increase in the number of upregulated piRNAs from 1 301 significant DE piRNA at 50 mg/kg/d to over 1000 at 75 mg/kg/d (Fig. 7). piRNAs are important 302 for maintenance of germline genomic integrity by regulating insertion of transposable elements 303 (Wang and Lin 2021). piRNAs also control some aspects of posttranscriptional processing and 304 cleavage of RNAs (Watanabe and Lin 2014) and have a variety of other epigenetic control 305 mechanisms (Ku and Lin 2014). Hypothetically, differential expression of piRNAs could 306 indirectly destabilize the genome of germline cells, despite EGME not being a direct mutagen 307 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 13 (Hoflack et al. 1995). This hypothesis would require further testing, but it presents a possible 308 sperm sncRNA-mediated mechanism of paternal EGME effects on embryonic development. As 309 noted above, rsRNAs are also prevalent in sperm, representing over half of sperm RNA reads 310 when PANDORA-seq is employed (Shi et al. 2021; Chen and Zhou 2023). We also analyzed 311 other sperm sncRNAs, such as rRNAs, mtRNA, snRNA, and lincRNA, and we found that the 312 number of DE RNAs increased across those classes with increasing EGME dose (Fig. 8). 313 Overall, we found that short-term exposure to low doses of EGME led to changes in 314 sperm RNA that were detectable 5 weeks after cessation of exposure, consistent with the 315 known progression of EGME-induced testicular injury and reproductive performance in rats 316 (Chapin, Dutton, Ross, and Lamb 1985). This is also consistent with prior reports that EGME 317 exposure altered testicular gene and protein expression (Wang & Chapin, 2000, Fukushima et 318 al., 2005, Yamamoto et al., 2005) and miRNA expression in rats and non-human primates 319 (Fukushima et al., 2011, Sakurai et al., 2015). EGME’s property of targeting pachytene 320 spermatocytes may be the reason why it has such potent and persistent effects on testicular 321 germ cell RNA expression, as the peak of RNA synthesis during spermatogenesis occurs in the 322 pachytene spermatocytes (Monesi, 1965). However, testicular toxicants with a variety of target 323 cells and toxicity mechanisms have been reported to alter sperm mRNA expression. These 324 include pharmaceuticals (Dere et al. 2017) and industrial or environmental testicular toxicants 325 including the germ cell toxicant, carbendazim, and the Sertoli cell toxicant, 2,5-hexanedione 326 (Pacheco et al. 2012). There is also a recent report that phthalate exposure altered miRNA, 327 tsRNA, and piRNA levels in epididymal extracellular vesicles (Oluwayiose et al. 2025), some of 328 which are transferred to sperm (Sharma et al. 2018). Similarly, a 12-week exposure to a mixture 329 of per- and polyfluoro alkyl substances (PFAS) resulted in increased numbers of sperm tsRNAs, 330 miRNAs, and piRNAs at doses that caused few phenotypic changes in adults and no changes in 331 fertilization rate (Gillespie et al. 2025). Other parental treatments alter sperm miRNAs with 332 apparent intergenerational biological effects, including maternal stress (Gapp et al. 2014), 333 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 14 paternal exercise (Yin et al. 2025 Oct), and paternal metabolic disorders (Zhang et al. 2018; 334 Zhang et al. 2019). 335 In humans, there is evidence that sperm RNAs may act as markers of fertility or 336 exposures that affect fertility. For example, in one study the total amount of RNA per sperm cell 337 differed based on alcohol use (Bianchi et al. 2019). We recently reported that human sperm 338 mRNA profiles differed between study participants with different clinical fertility status (Qi et al. 339 2025 May 10). Another recent study found that miRNAs, including let-7b-5p, differed between 340 proven fertile and subfertile men (Abu-Halima et al. 2024 Sept 23). Here, our new analysis 341 identified changes in rat sperm sncRNA following EGME exposure, which is consistent with the 342 hypothesis that paternal exposure to environmental toxicants can alter sperm RNA profiles and 343 that this is a putative mechanism of reduced male fertility. Future experiments should test the 344 role of specific EGME dose-dependent candidate miRNAs on rat embryo development and 345 other sncRNA-mediated epigenetic toxicity mechanisms. In summary, EGME not only 346 sensitively alters the proportion of different classes of small RNAs present in sperm, as 347 previously reported (Stermer et al. 2019); it also leads to dose-dependent changes in specific 348 small regulatory RNAs in sperm. Investigation of the roles of those RNAs in spermatogenesis 349 and embryo development may lead to discovery of novel EGME toxicity mechanisms. 350 Data Availability Statement 351 Stored in repository: The data that support the findings of this study are openly available in the 352 NCBI Gene Ontology Omnibus (GEO) database at 353 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE313131, reference number 354 GSE313131. The analysis code is publicly available at https://github.com/compbiocore/egme. 355 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 15 Funding Information 356 This project was funded by: Brown University; and the National Institute of General Medical 357 Sciences (NIGMS) of the National Institutes of Health (NIH) [P20 GM109035; P20 GM156712]. 358 The content is solely the responsibility of the authors and does not necessarily represent the 359 official views of the National Institutes of Health. 360 Conflicts of Interest Statement 361 The authors declare that they have no conflicts of interest. 362 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 16

363 Abu-Halima M et al. 2019. Differential expression of miR-23a/b-3p and its target genes in male 364 patients with subfertility. Fertil Steril. 112(2):323-335.e2. 365 https://doi.org/10.1016/j.fertnstert.2019.03.025 366 Abu-Halima M et al. 2024. Single Sperm RNA signatures reveal MicroRNA biomarkers for male 367 subfertility. J Assist Reprod Genet. [published online ahead of print] [accessed 2024 Oct 368 1]. https://link.springer.com/10.1007/s10815-024-03264-w. 369 https://doi.org/10.1007/s10815-024-03264-w 370 Andrews S. 2010. FastQC: A Quality Control Tool for High Throughput Sequence Data [Online]. 371 https://www.google.com/url?q=http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ 372 Aparicio-Puerta E et al. 2022. sRNAbench and sRNAtoolbox 2022 update: accurate miRNA and 373 sncRNA profiling for model and non-model organisms. Nucleic Acids Res. 374 50(W1):W710–W717. https://doi.org/10.1093/nar/gkac363 375 Beattie PJ, Brabec MJ. 1986. Methoxyacetic acid and ethoxyacetic acid inhibit mitochondrial 376 function in vitro. J Biochem Toxicol. 1(3):61–70. https://doi.org/10.1002/jbt.2570010307 377 Beyret E, Liu N, Lin H. 2012. piRNA biogenesis during adult spermatogenesis in mice is 378 independent of the ping-pong mechanism. Cell Res. 22(10):1429–1439. 379 https://doi.org/10.1038/cr.2012.120 380 Bianchi E et al. 2019. Spermatozoal large RNA content is associated with semen 381 characteristics, sociodemographic and lifestyle factors. PLoS One. 14(5):e0216584. 382 https://doi.org/10.1371/journal.pone.0216584 383 Bianchi E et al. 2021. Highly conserved sperm function-related transcripts across three species: 384 human, rat and mouse. Reprod Toxicol. 104:44–51. 385 https://doi.org/10.1016/j.reprotox.2021.06.012 386 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 17 Chapin RE et al. 1984. The effects of ethylene glycol monomethyl ether on testicular histology in 387 F344 rats. J Androl. 5(5):369–80 388 Chapin RE, Dutton SL, Ross MD, Swaisgood RR, et al. 1985. The recovery of the testis over 8 389 weeks after short-term dosing with ethylene glycol monomethyl ether: histology, cell-390 specific enzymes, and rete testis fluid protein. Fundam Appl Toxicol. 5(3):515–25 391 Chapin RE, Dutton SL, Ross MD, Lamb JC. 1985. Effects of ethylene glycol monomethyl ether 392 (EGME) on mating performance and epididymal sperm parameters in F344 rats. 393 Fundam Appl Toxicol. 5(1):182–9 394 Chapin RE, Lamb JC. 1984. Effects of ethylene glycol monomethyl ether on various parameters 395 of testicular function in the F344 rat. Env Health Perspect. 57:219–24. 396 https://doi.org/10.1289/ehp.8457219 397 Chen Q, Zhou T. 2023. Emerging functional principles of tRNA-derived small RNAs and other 398 regulatory small RNAs. J Biol Chem. 299(10):105225. 399 https://doi.org/10.1016/j.jbc.2023.105225 400 Chen Y, Wang X. 2020. miRDB: an online database for prediction of functional microRNA 401 targets. Nucleic Acids Res. 48(D1):D127–D131. https://doi.org/10.1093/nar/gkz757 402 Conine CC et al. 2018. Small RNAs Gained during Epididymal Transit of Sperm Are Essential 403 for Embryonic Development in Mice. Dev Cell. 46(4):470-480 e3. 404 https://doi.org/10.1016/j.devcel.2018.06.024 405 Creasy DM, Beech LM, Gray TJ, Butler WH. 1986. An ultrastructural study of ethylene glycol 406 monomethyl ether-induced spermatocyte injury in the rat. Exp Mol Pathol. 45(3):311–22 407 Creasy DM, Foster PM. 1984. The morphological development of glycol ether-induced testicular 408 atrophy in the rat. Exp Mol Pathol. 40(2):169–76 409 Dere E et al. 2017. Identification of sperm mRNA biomarkers associated with testis injury during 410 preclinical testing of pharmaceutical compounds. Toxicol Appl Pharmacol. 320:1–7. 411 https://doi.org/10.1016/j.taap.2017.02.001 412 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 18 Dere E et al. 2018. Effects of continuous bisphenol A exposure from early gestation on 90 day 413 old rat testes function and sperm molecular profiles: A CLARITY-BPA consortium study. 414 Toxicol Appl Pharmacol. 347:1–9. https://doi.org/10.1016/j.taap.2018.03.021 415 Dere E, Wilson SK, Anderson LM, Boekelheide K. 2016. From the Cover: Sperm Molecular 416 Biomarkers Are Sensitive Indicators of Testicular Injury following Subchronic Model 417 Toxicant Exposure. Toxicol Sci. 153(2):327–340. https://doi.org/10.1093/toxsci/kfw137 418 Foster PM et al. 1983. Testicular toxicity of ethylene glycol monomethyl and monoethyl ethers in 419 the rat. Toxicol Appl Pharmacol. 69(3):385–99 420 Foster PM, Creasy DM, Foster JR, Gray TJ. 1984. Testicular toxicity produced by ethylene 421 glycol monomethyl and monoethyl ethers in the rat. Env Health Perspect. 57:207–17. 422 https://doi.org/10.1289/ehp.8457207 423 Gapp K et al. 2014. Implication of sperm RNAs in transgenerational inheritance of the effects of 424 early trauma in mice. Nat Neurosci. 17(5):667–669. https://doi.org/10.1038/nn.3695 425 Gillespie L et al. 2025. Exposure of mice to environmentally relevant per- and polyfluoroalkyl 426 substances (PFAS) alters the sperm epigenome. Commun Biol. 8(1):1487. 427 https://doi.org/10.1038/s42003-025-08865-4 428 Gray TJ, Moss EJ, Creasy DM, Gangolli SD. 1985. Studies on the toxicity of some glycol ethers 429 and alkoxyacetic acids in primary testicular cell cultures. Toxicol Appl Pharmacol. 430 79(3):490–501 431 Hekim N, Gunes S, Ayas B, Alkan EN. 2025. Seminal 5′ tRF-Glu-CTC as a Biomarker for 432 Oligozoospermia and ART Outcome Prediction. Basic Clin Androl. 35(1):33. 433 https://doi.org/10.1186/s12610-025-00283-0 434 Hernandez R et al. 2023. PANDORA-Seq unveils the hidden small noncoding RNA landscape 435 in atherosclerosis of LDL receptor-deficient mice. J Lipid Res. 64(4):100352. 436 https://doi.org/10.1016/j.jlr.2023.100352 437 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 19 Hoflack JC, Lambolez L, Elias Z, Vasseur P. 1995. Mutagenicity of ethylene glycol ethers and of 438 their metabolites in Salmonella typhimurium his-. Mutat Res. 341(4):281–287 439 Holloway AJ, Moore HD, Foster PM. 1990. The use of rat in vitro fertilization to detect 440 reductions in the fertility of spermatozoa from males exposed to ethylene glycol 441 monomethyl ether. Reprod Toxicol. 4(1):21–7 442 Iben JR, Maraia RJ. 2014. tRNA gene copy number variation in humans. Gene. 536(2):376–443 384. https://doi.org/10.1016/j.gene.2013.11.049 444 Jodar M et al. 2015. Absence of sperm RNA elements correlates with idiopathic male infertility. 445 Sci Transl Med. 7(295):re6. https://doi.org/10.1126/scitranslmed.aab1287 446 Kozomara A, Birgaoanu M, Griffiths-Jones S. 2019. miRBase: from microRNA sequences to 447 function. Nucleic Acids Res. 47(D1):D155–D162. https://doi.org/10.1093/nar/gky1141 448 Ku H-Y, Lin H. 2014. PIWI proteins and their interactors in piRNA biogenesis, germline 449 development and gene expression. Natl Sci Rev. 1(2):205–218. 450 https://doi.org/10.1093/nsr/nwu014 451 Ku WW et al. 1995. Spermatocyte toxicity of 2-methoxyethanol (ME) in rats and guinea pigs: 452 evidence for the induction of apoptosis. Toxicol Appl Pharmacol. 134(1):100–10. 453 https://doi.org/10.1006/taap.1995.1173 454 Lau NC et al. 2006. Characterization of the piRNA Complex from Rat Testes. Science. 455 313(5785):363–367. https://doi.org/10.1126/science.1130164 456 Li Y et al. 2022. Exaggerated false positives by popular differential expression methods when 457 analyzing human population samples. Genome Biol. 23(1):79. 458 https://doi.org/10.1186/s13059-022-02648-4 459 Love MI, Huber W, Anders S. 2014. Moderated estimation of fold change and dispersion for 460 RNA-seq data with DESeq2. Genome Biol. 15(12):550. https://doi.org/10.1186/s13059-461 014-0550-8 462 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 20 Moss EJ et al. 1985. The role of metabolism in 2-methoxyethanol-induced testicular toxicity. 463 Toxicol Appl Pharmacol. 79(3):480–9 464 Norioun H, Motovali-bashi M, Morteza Javadirad S. 2020. Hsa-miR-27a-3p overexpression in 465 men with nonobstructive azoospermia: A case-control study. Int J Reprod Biomed IJRM. 466 18(11):961–968. https://doi.org/10.18502/ijrm.v13i11.7963 467 O’Brien J, Hayder H, Zayed Y, Peng C. 2018. Overview of MicroRNA Biogenesis, Mechanisms 468 of Actions, and Circulation. Front Endocrinol. 9 [accessed 2025 July 17]. 469 https://www.frontiersin.org/article/10.3389/fendo.2018.00402/full. 470 https://doi.org/10.3389/fendo.2018.00402 471 Oluwayiose OA et al. 2025. Phthalate exposure to adult mice alters non-coding RNAs in cauda 472 epididymosomes. Environ Pollut. 386:127227. 473 https://doi.org/10.1016/j.envpol.2025.127227 474 Pacheco SE et al. 2012. Sperm mRNA transcripts are indicators of sub-chronic low dose 475 testicular injury in the Fischer 344 rat. PLoS One. 7(8):e44280. 476 https://doi.org/10.1371/journal.pone.0044280 477 Petropoulos S et al. 2016. Single-Cell RNA-Seq Reveals Lineage and X Chromosome 478 Dynamics in Human Preimplantation Embryos. Cell. 165(4):1012–1026. 479 https://doi.org/10.1016/j.cell.2016.03.023 480 Qi X et al. 2025. Downregulation of spermatogenesis‐ associated transcripts in the spermatozoa 481 of idiopathic infertile men. Andrology. andr.70060. https://doi.org/10.1111/andr.70060 482 Sakurai K et al. 2015. MicroRNA profiling in ethylene glycol monomethyl ether-induced monkey 483 testicular toxicity model. J Toxicol Sci. 40(3):375–82. https://doi.org/10.2131/jts.40.375 484 Sangiao-Alvarellos S et al. 2015. Testicular expression of the Lin28/let-7 system: Hormonal 485 regulation and changes during postnatal maturation and after manipulations of puberty. 486 Sci Rep. 5:15683. https://doi.org/10.1038/srep15683 487 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 21 Sendler E et al. 2013. Stability, delivery and functions of human sperm RNAs at fertilization. 488 Nucleic Acids Res. 41(7):4104–4117. https://doi.org/10.1093/nar/gkt132 489 Sharma U et al. 2016. Biogenesis and function of tRNA fragments during sperm maturation and 490 fertilization in mammals. Science. 351(6271):391–396. 491 https://doi.org/10.1126/science.aad6780 492 Sharma U et al. 2018. Small RNAs Are Trafficked from the Epididymis to Developing 493 Mammalian Sperm. Dev Cell. 46(4):481-494 e6. 494 https://doi.org/10.1016/j.devcel.2018.06.023 495 Shi J et al. 2021. PANDORA-seq expands the repertoire of regulatory small RNAs by 496 overcoming RNA modifications. Nat Cell Biol. 23(4):424–436. 497 https://doi.org/10.1038/s41556-021-00652-7 498 Stermer AR, Reyes G, Hall SJ, Boekelheide K. 2019. Small RNAs in Rat Sperm Are a 499 Predictive and Sensitive Biomarker of Exposure to the Testicular Toxicant Ethylene 500 Glycol Monomethyl Ether. Toxicol Sci. 169(2):399–408. 501 https://doi.org/10.1093/toxsci/kfz041 502 Tonkin EG et al. 2009. Testicular gene expression profiling following 2-methoxyethanol and 2-503 ethoxyethanol exposure in male rats reveals abnormal expression of the actin binding 504 protein cortactin in degenerating spermatocytes. Toxicol Lett. 190(2):193–201. 505 https://doi.org/10.1016/j.toxlet.2009.07.017 506 Wang C, Lin H. 2021. Roles of piRNAs in transposon and pseudogene regulation of germline 507 mRNAs and lncRNAs. Genome Biol. 22(1):27. https://doi.org/10.1186/s13059-020-508 02221-x 509 Wang J et al. 2019. piRBase: a comprehensive database of piRNA sequences. Nucleic Acids 510 Res. 47(D1):D175–D180. https://doi.org/10.1093/nar/gky1043 511 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 22 Watanabe T, Cheng E, Zhong M, Lin H. 2015. Retrotransposons and pseudogenes regulate 512 mRNAs and lncRNAs via the piRNA pathway in the germline. Genome Res. 25(3):368–513 380. https://doi.org/10.1101/gr.180802.114 514 Watanabe T, Lin H. 2014. Posttranscriptional Regulation of Gene Expression by Piwi Proteins 515 and piRNAs. Mol Cell. 56(1):18–27. https://doi.org/10.1016/j.molcel.2014.09.012 516 Yang QE et al. 2013. MicroRNAs 221 and 222 regulate the undifferentiated state in mammalian 517 male germ cells. Development. 140(2):280–90. https://doi.org/10.1242/dev.087403 518 Yin X et al. 2025. Paternal exercise confers endurance capacity to offspring through sperm 519 microRNAs. Cell Metab. S1550413125003882. 520 https://doi.org/10.1016/j.cmet.2025.09.003 521 Yu G. 2024. Thirteen years of clusterProfiler. The Innovation. 5(6):100722. 522 https://doi.org/10.1016/j.xinn.2024.100722 523 Zhang Y et al. 2019. Sperm RNA code programmes the metabolic health of offspring. Nat Rev 524 Endocrinol. 15(8):489–498. https://doi.org/10.1038/s41574-019-0226-2 525 Zhang Yunfang et al. 2018. Dnmt2 mediates intergenerational transmission of paternally 526 acquired metabolic disorders through sperm small non-coding RNAs. Nat Cell Biol. 527 20(5):535–540. https://doi.org/10.1038/s41556-018-0087-2 528 Zimmermann C et al. 2014. Germ cell-specific targeting of DICER or DGCR8 reveals a novel 529 role for endo-siRNAs in the progression of mammalian spermatogenesis and male 530 fertility. PLoS One. 9(9):e107023. https://doi.org/10.1371/journal.pone.0107023 531 532 Figure Legends 533 Figure 1. Changes in sperm miRNA levels following EGME exposure. 534 A. Hierarchical clustering of EGME- and vehicle control-treated samples showing Euclidean 535 distance based on sperm miRNA read counts. B. Principal component analysis of EGME- and 536 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 23 vehicle-control-treated samples based on sperm miRNA read counts. C. Differential expression 537 of sperm miRNA between vehicle control and 50, 60, or 75 mg/kg/d EGME-treated samples 538 (DESeq2 with the Wald test, p-adj 2). 539 540 Figure 2. EGME dose-dependency of sperm miRNAs. 541 A. Venn diagram showing overlap of differentially expressed miRNAs in comparisons between 542 control and 50, 60, and 75 mg/kg/d EGME. 12 miRNAs were common to all three treatment 543 levels. B. Top 40 most commonly differentially expressed sperm miRNAs in a Monte Carlo 544 analysis. Ten of the 12 common differentially expressed miRNAs (shown in red) were in the top 545 12, and the remaining two were within the top 40 most reproducibly differentially expressed 546 miRNAs. C. Normalized read counts for the 12 common differentially expressed miRNAs. In all 547 cases, the response trend was monotonic increase with increasing dose. 548 549 Figure 3. Targets of EGME dose-dependent miRNAs. 550 Using mirDB, we identified 1,203 high-confidence targets of the 12 common differentially 551 expressed miRNAs. A. Histogram showing the distribution of target genes by the number of DE 552 miRNAs that share the target. B. Venn diagram showing the overlap of the 1,203 common DE 553 miRNA target genes with genes known to be expressed in the epiblast (EPI), primitive 554 endoderm (PE), or trophectoderm (TE) of the early human embryo, from a single-cell RNA-seq 555 analysis (Petropoulos et al. 2016). The analysis identified a total of 119 early human embryo 556 genes among the 1,203 targets of common EGME differentially expressed miRNAs. Note that 557 one gene shared by the EGME-miR-target, PE, and EPI gene list and one shared by the EGME-558 miR-target, PE, and TE gene lists is not accounted for in the diagram because of graphing 559 constraints. C. Histogram showing the number of core miRNAs with target genes common to 560 each of the enriched GO BPs. D. Enrichment analysis of Gene Ontology Biological Process 561 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 24 terms (GO BPs) for the 1,203 targets of the 12 common differentially expressed miRNAs. Right 562 panel shows hierarchical arrangement of enriched GO BPs. 563 564 Figure 4. Functional analysis of EGME dose-dependent miRNA targets. 565 Using the 1,203 high-confidence targets of the 12 common differentially expressed miRNAs. A. 566 Enrichment analysis of Gene Ontology Cellular Component terms (GO CCs) for the 1,203 567 targets of the 12 common differentially expressed miRNAs. B. Enrichment analysis of Gene 568 Ontology Biological Molecular Function terms (GO MFs) for the 1,203 targets of the 12 common 569 differentially expressed miRNAs. Right panels show hierarchical arrangement of enriched GO 570 MFs. 571 572 Figure 5. Changes in sperm tsRNA levels following EGME exposure. 573 A. Hierarchical clustering of EGME- and vehicle control-treated samples showing Euclidean 574 distance based on sperm tRNA read counts. B. Principal component analysis of EGME- and 575 vehicle-control-treated samples based on sperm tRNA read counts. C. Differential expression of 576 sperm tRNA between vehicle control and 50, 60, or 75 mg/kg/d EGME-treated samples 577 (DESeq2 with the Wald test, p-adj 2). D. Venn diagram showing overlap of 578 differentially expressed tRNAs in comparisons between control and 50, 60, and 75 mg/kg/d 579 EGME, after removal of duplicate tRNA genes. Two unique tRNAs were common to all three 580 treatment levels. E. Sequence coverage map showing read coverage of tRNA reads for the two 581 tRNAs that were differentially expressed at all doses, GluCTC and GlyCCC, and the next four 582 tRNAs with at least one significant gene at the 60 and 75 mg/kg/d dose levels, GluTTC, 583 LysCTT, AlaAGC, and CysGCA, based on the lowest adjusted p-value at 50 mg/kg/d, scaled by 584 read count on the y-axis. 585 586 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint 25 Figure 6. Detailed coverage maps for the most significant tsRNAs. 587 Coverage maps showing all unique tRNA reads for the tRNAs that were differentially expressed 588 at all doses, GluCTC and GlyCCC, and the next four tRNAs with at least one significant gene at 589 the 60 and 75 mg/kg/d dose levels, GluTTC, LysCTT, AlaAGC, and CysGCA, based on the 590 lowest adjusted p-value at 50 mg/kg/d. Plots show unique reads with color key based on the 591 number of copies of each unique read. The majority of reads were 5’ fragments and halves. 592 593 Figure 7. Changes in sperm piRNAs following EGME exposure. 594 A. Hierarchical clustering of EGME- and vehicle control-treated samples showing Euclidean 595 distance based on sperm piRNA read counts. B. Principal component analysis of EGME- and 596 vehicle-control-treated samples based on sperm piRNA read counts. C. Differential expression 597 of sperm piRNA between vehicle control and 50, 60, or 75 mg/kg/d EGME-treated samples 598 (DESeq2 with the Wald test, p-adj 2). 599 600 Figure 8. Changes in other sperm small RNAs following EGME exposure. 601 A. Hierarchical clustering of EGME- and vehicle control-treated samples showing Euclidean 602 distance based on remaining sperm read counts. After separating miRNA, tRNA, and piRNA 603 reads, the remaining DE RNAs were from RNA classes including rRNA, mitochondrial rRNA, 604 lincRNA, snRNA, B. Principal component analysis of EGME- and vehicle-control-treated 605 samples based on sperm sncRNA read counts. C. Differential expression of sperm sncRNAs 606 between vehicle control and 50, 60, or 75 mg/kg/d EGME-treated samples (DESeq2 with the 607 Wald test, p-adj 2). 608 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted December 29, 2025. ; https://doi.org/10.64898/2025.12.29.693789doi: bioRxiv preprint