Maternal Pregnancy-Associated Circulating MicroRNA predictors of infant birth outcomes are Determinants of Placental Maturation

We previously identified 11 miRNAs which were significantly elevated in blood plasma of pregnant mothers who subsequently gave birth to infants affected by prenatal alcohol exposure (PAE, Heavily Exposed Affected: HEa) compared to those with infants who were exposed but apparently unaffected (Heavily Exposed Unaffected: HEua) or unexposed (UE). These maternal HEamiRNAs were predicted to originate in part, as a paracrine placental signal to influence and coordinate placental epithelial-mesenchymal transition (EMT), a pathway essential for endometrial invasion and development. We now report that PAE inhibits expression of placental EMT pathway members in rodent and primate voluntary alcohol consumption models, with HEamiRNAs collectively mediating placental EMT inhibition. To directly investigate the interaction between HEamiRNAs and ethanol on placenta, we assessed their effects on human placental trophoblast cell lines. When administered together, HEamiRNAs retarded trophoblast cell cycle, significantly impaired expression of core EMT pathway members, and reduced invasiveness, pointing to their collective role in modulating placental growth and invasion deficits seen with PAE. HEamiRNAs additionally interfered with maturation-dependent intracellular calcium dynamics, while promoting syncytialization-dependent increases in placental hormone expression. Finally, a single systemic administration of the pooled murine-expressed HEamiRNA subpopulation, to pregnant mice, decreased fetal and placental growth and inhibited expression of EMT pathway mRNA transcripts in placenta. Taken together, our data suggests that, following PAE, HEamiRNAs interfere with placental development and may contribute to the pathology of Fetal Alcohol Spectrum Disorders.

exposure (PAE, Heavily Exposed Affected: HEa) compared to those with infants who were 48 exposed but apparently unaffected (Heavily Exposed Unaffected: HEua) or unexposed (UE). 49 These maternal HEamiRNAs were predicted to originate in part, as a paracrine placental signal to 50 influence and coordinate placental epithelial-mesenchymal transition (EMT), a pathway 51 essential for endometrial invasion and development. We now report that PAE inhibits Introduction 66 Alcohol use during pregnancy remains prevalent despite published prevention 67 guidelines (1). A recent meta-analysis determined the global prevalence of alcohol use during 68 pregnancy at 9.8% (2). Within the United States, a 2013 study found 18% of women reported 69 alcohol consumption during pregnancy, and 6.6% engaged in binge-drinking episodes, which 70 are particularly damaging to fetal development (3). In the state of Texas, we recently reported 71 an average state-wide third-trimester rate of alcohol exposure of 8.4%, with rates as high as 72 17.7% in some localities (4). These data collectively suggest that prenatal alcohol exposure 73 (PAE) is common. 74 PAE harms the developing embryo and fetus and can result in a constellation of adverse 75 infant outcomes, including craniofacial dysmorphologies, growth retardation, and 76 neurobehavioral abnormalities, collectively termed Fetal Alcohol Spectrum Disorders (FASDs) 77 (5). Estimates for the prevalence of FASDs range from 2-5% of the school age population in the 78 US and Western Europe (6) to 13.6-20.9% in South Africa (7). Estimates for the global 79 prevalence of fetal alcohol syndrome (FAS), the most severe end of the FASD continuum, 80 characterized by cranio-facial dysmorphologies, growth restriction, and CNS abnormalities, 81 range from 0.15 to 0. 3% (8, 9), but are much higher in specific populations, for example, as high 82 as 5.9-9.1% in the Cape Coloured (mixed ancestry) community in South Africa (7). 83 Given the lifelong debility and economic burden associated with FASDs (10), there is an 84 urgent need to develop sensitive and specific methodologies to identify affected children early 85 in development, as well as develop interventions which can mitigate the effects of PAE on the 86 in 2 nd and 3 rd trimester maternal circulation distinguished infants who were affected by in-utero 88 alcohol exposure (Heavily Exposed Affected: HEa) from those who were apparently unaffected 89 at birth by PAE (Heavily Exposed Unaffected: HEua), or those who were unexposed (UE) (11) . 90 We predicted that these HEamiRNAs   [hsa-miR-204-5p], MIMAT0002869 [has-miR-519a-3p]) in addition to being biomarkers of infant 95 outcome following PAE, influence signaling pathways crucial for early development, particularly 96 the epithelial-mesenchymal transition (EMT) pathway (11). 97 Placental development involves maturation of the cytotrophoblasts at the tip of 98 anchoring villi into invasive extravillous trophoblasts, as well as fusion of cytotrophoblasts into 99 multinucleate, hormone-producing syncytiotrophoblasts (12). Maturation into extravillous 100 trophoblasts, which invade the maternal decidua and remodel the uterine spiral arteries into 101 low-resistance high-flow vessels that enable optimal perfusion for nutrient and waste 102 exchange, requires cytotrophoblasts to undergo EMT (13). Impaired placental EMT, as well as 103 orchestration of the opposing mesenchymal-epithelial transition pathway, has been found in 104 conditions resulting from placental malfunction, primarily preeclampsia (14)(15)(16)(17)(18)(19). While there 105 have been no previous studies directly investigating the effects of PAE on placental EMT, a 106 rodent study demonstrated that PAE, during a broad developmental window, reduced the 107 number of invasive trophoblasts within the mesometrial triangle, a region of the uterine horn 108 directly underlying the decidua (20). Furthermore, both human and rodent studies have found

118
HEamiRNAs are implicated in placental-associated pathologies 119 Outside of our report that elevated HEamiRNA levels predict infant outcomes following 120 PAE, few other studies have investigated expression of these miRNAs in the context of PAE or 121 other potentially toxic prenatal exposures ( Figure 1A) (29)(30)(31)(32)(33)(34)(35)(36). Given our prediction that 122 HEamiRNAs interfere with signaling pathways governing fetal and placental development (11), 123 we conducted a literature review of reports on HEamiRNA levels in gestational pathologies 124 which, as with PAE, include aberrant placentation as an etiological factor (37)(38)(39). Surprisingly, 125 placental and plasma levels of 8 out of 11 of these HEamiRNAs were significantly dysregulated in 126 one or more of these gestational pathologies with expression of the majority of these 8 miRNAs 127 altered in both fetal growth restriction and preeclampsia ( Figure 1B) (33,34,36,(40)(41)(42)(43)(44)(45)(46)(47)(48)(49)(50)(51)(52)(53)(54)(55)(56)(57), both of 128 which are characterized by poor placental invasion as an etiological factor (58-64). 129 HEamiRNAs explain variance in infant growth outcomes due to PAE 130 Given the association of individual HEamiRNAs with gestational pathologies, we sought to 131 determine if circulating HEamiRNAs levels could explain the variance in sex and gestational age-132 adjusted neonatal height, weight and head circumference in our Ukrainian birth cohort, which 133 are growth measures sensitive to in utero environment (65). We found that 8 of the HEamiRNAs, 134 each significantly explained between 7 to 19% of infant variation in these growth measures 135 (Table 1). Furthermore, 7 of these miRNAs were also associated with fetal growth restriction 136 and preeclampsia as identified by our literature review. Additionally, only 1 of the 3 miRNAs not 137 previously-reported to be associated with gestational pathologies was correlated with 138 measures of fetal growth and development in our dataset ( Figure 1B). Interestingly, a 139 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. our studies utilized these markers to assess the effects of alcohol and HEamiRNAs on trophoblast 163 EMT. 164 In the first analysis, using a murine model of PAE that mimicked moderate to binge-type 165 alcohol consumption throughout early and mid-pregnancy, we fractionated GD14 placenta into 166 three zones: the cytotrophoblast and syncytiotrophoblast rich labyrinth zone, the glycogen and 167 spongiotrophoblast rich junctional zone, and the decidual zone comprising the endometrial 168 contribution to the placenta ( Figure 2A). Multivariate analysis of variance (MANOVA) for 169 expression of these five core genes in the EMT pathway within placental trophoblasts, revealed 170 a significant effect of ethanol exposure on EMT pathway member expression selectively within 171 the labyrinth zone (Pillai's trace statistic, F(5,21)=6.85, p<0.001, Figure 2B)  To determine if PAE's effects on EMT pathway members in placenta are broadly 188 conserved throughout mammalian evolution, we adopted a non-human primate (macaque) 189 model of moderate to binge-type alcohol consumption. Placental tissues were isolated from 190 GD85, GD110, and GD 135 placenta ( Figure 2D), which spans the human equivalent of mid-191 second to mid-third trimester (Supplementary Figure 2C). There was a significant effect of 192 ethanol exposure on expression of core EMT mRNA transcripts by MANOVA (Pillai's trace 193 statistic, F(4,9)=4.229 p=0.045, Figure 3B Figure 2E). Interestingly, accounting for expression of individual HEamiRNAs did not explain the 201 effects of PAE on placental EMT, suggesting that HEamiRNAs act in concert to mediate the effect 202 of PAE on EMT in primate placenta ( Figure 2F). Cytotrophoblasts normally undergo EMT to become invasive trophoblasts, which we also 215 modelled using the HTR8 human extravillous trophoblast cell line. We initially overexpressed 216 each of the 11 HEamiRNAs individually, to determine whether any of them could alter mRNA 217 transcripts for EMT pathway members. However, we did not observe any significant effects due  Figure 4E). We were unable to detect 227 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint SNAI2 transcript expression or vimentin protein expression in these cells, consistent with 228 previous reports (71). 229 Whereas transfection of HEamiRNA mimics increased CDH1 expression, transfection of 230 pooled antagomirs to HEamiRNAs, significantly reduced CDH1 expression, only in the context of 231 320 mg/dL ethanol co-exposure (post-hoc Tukey's HSD, p=0.005), consistent with an overall 232 statistically significant interaction effect between ethanol exposure and HEamiRNA inhibition 233 (F(1,36)=13.51, p=0.0008, Figure 4F). However, expression of TWIST was also decreased in this 234 context and there was no significant difference in E-Cadherin protein expression relative to the 235 control ( Figure 4G-J). Thus, our data suggest that, increasing HEamiRNA levels impairs EMT 236 pathway members in cytotrophoblasts whereas inhibiting their action has a more restricted 237 effect on EMT pathway members.  Figure 5E). Interestingly, there was also a main effect of alcohol 249 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.  The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. interfere with the EMT pathway ( Figure 7A). Contrastingly, transfecting HEamiRNA antagomirs 292 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint increased invasion in the context of 320 mg/dL ethanol co-exposure, though this effect was 293 only marginally significant (t(14)=1.805, p=0.093, Figure 7B).  The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. Adaptations to cellular stress can be seen in alterations to cellular energetics in 376 response to ethanol, as ethanol-exposed BeWO cells showed decreased baseline and stressed 377 oxygen consumption rates (OCR) (F(1,28)=15.55 and 16.91, p=0.0005 and 0.0003 respectively) 378 and increased extracellular acidification rates (ECAR) (F(1,28)=4.868, p=0.036). However, 379 HEamiRNAs had minimal effects on metabolic activity ( Figures 10D-10G). 380 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint Extracellular ATP has been shown to inhibit trophoblast migration (78)  The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint effect of HEamiRNA antagomirs on CGA and CGB expression, although we did observe a decrease 403 in IGF2 transcript expression, following syncytialization, relative to controls (post-hoc Tukey's 404 HSD, n=10 samples per group, p=0.001) ( Figure 11F-I). 405 Given that HEamiRNAs promotes syncytialization-dependent hormone production, we  The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint

444
In a prior study, we reported that gestational elevation of 11 maternal plasma miRNAs, 445 in a Ukrainian clinical cohort, predicted which PAE infants would exhibit adverse outcomes at 446 birth (11). These HEamiRNAs were elevated throughout mid and late-pregnancy, encompassing 447 critical periods for fetal development, and were predicted to target key developmental 448 pathways (11). Given that HEamiRNAs are placentally enriched and their dysregulation is 449 associated with placental-associated gestational pathologies, we sought to determine if they 450 also contributed to the pathophysiology of FASDs. 451 To mimic the alcohol consumption patterns in our Ukrainian cohort, we adopted 452 moderate alcohol self-administration paradigms during mouse and macaque gestation. Though 453 these species share a hemochorial placenta, fundamental differences in placental anatomy 454 remain. Specifically, macaque trophoblasts exhibit deeper invasion into the uterine horn than 455 their mouse counterparts (83-86). Despite these differences, we found PAE inhibited 456 mesenchymal transition-associated mRNA transcripts in both our models, indicating a 457 conserved effect of PAE on placental development. Additionally, we found that HEamiRNAs 458 mediated the effects of PAE on core EMT pathway members in the placenta, and collectively 459 though not individually, inhibited EMT in human cytotrophoblast and extravillous trophoblast 460 culture models. The components of the EMT pathway were selected for assessment based on a 461 substantial literature that implicates these as core EMT components (13,17,18,(67)(68)(69)(70). 462 However, analysis of their 3'UTRs indicates that these are unlikely to be the direct targets of 463 HEamiRNA action. Additional studies will be needed to dissect out the signaling networks that 464 connect HEamiRNAs to the assessed EMT components. 465 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint Interestingly, HEamiRNAs also promoted syncytialization (forskolin)-dependent hCG 466 expression, mirroring the elevation of third trimester maternal hCG levels in the PAE group 467 within our Ukrainian birth cohort. This late-gestation elevation of hCG levels may serve as a 468 compensatory mechanism to prevent the preterm birth associated with PAE, as hCG during late 469 gestation is hypothesized to promote uterine myometrial quiescence (87,88). In support of this 470 hypothesis, we found significant negative associations between both hCG levels and alcohol 471 consumption with gestational age at delivery. Furthermore, there was a significant interaction 472 between periconceptional alcohol exposure and hCG levels, with higher hCG levels 473 corresponding to a smaller effect of alcohol exposure at conception on gestational age at 474 delivery, indicating that hCG moderates the effect of alcohol on age at delivery (Supplementary   475   Table 3). 476 Since HEamiRNAs collectively prevented trophoblast EMT, we hypothesized that, as a 477 functional consequence, these maternal miRNAs would also inhibit fetal growth. When we 478 delivered 8 out of the 11 HEamiRNAs known to be present in mouse, to pregnant dams during 479 the period of placental branching morphogenesis and endometrial invasion , when EMT is 480 particularly active, we found that HEamiRNAs reduced fetal growth. Importantly, ethanol 481 exposure during this period has also been shown to result in fetal growth deficits and 482 dysmorphia in rodent PAE models (89, 90) suggesting that maternal miRNA-mediated deficits in 483 trophoblast invasion may mediate some of the effects of PAE on fetal growth. In support of this, 484 we found placentas from the HEamiRNA treated group had impaired expression of core EMT 485 pathway members. It is also feasible that HEamiRNAs disrupt fetal growth through placental 486 vascular dynamics. The non-human primate tissue analyzed here was derived from animals that 487 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint were characterized in vivo using MRI and ultrasound imaging, which demonstrated that 488 maternal blood supply to the placenta was lower in ethanol-exposed animals compared to 489 controls, and that oxygen availability to the fetal vasculature was reduced (91). Thus, one 490 possibility is that compromised trophoblast invasion contributes to impaired maternal blood 491 flow in ethanol-exposed individuals, as we have also previously observed in mouse models (92). further investigation is warranted to determine if HEamiRNAs contribute to this alcohol induced 502 impairment in trans-placental nutrient transport, and ultimately whether this impairment of 503 nutrient transport represents another avenue through which PAE disrupts fetal development. 504 It is likely that HEamiRNAs may mediate other pregnancy associated pathologies, aside 505 from PAE. We identified numerous studies that reported increased circulating and placental 506 levels of at least 8 out of 11 HEamiRNAs in gestational pathologies arising from placental 507 dysfunction. For example, elevated levels of one HEamiRNA, miR-519a-3p, a member of the 508 placentally-expressed C19MC family cluster, was reported in placentae of patients with pre-509 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint eclampsia, recurrent spontaneous abortion, and intrauterine growth restriction (40,41,54,55). While we did not investigate the effects of PAE on EMT in non-placental organs, it is 530 likely that PAE broadly disrupts EMT in multiple fetal compartments. Developmental ethanol 531 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. in these miRNAs yielded net resistance to apoptosis following ethanol exposure (116). In that 553 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint study, we also showed that coordinate knockdown of these three miRNAs was required to 554 induce mRNA for Jagged-1, a ligand for the Notch cell signaling pathway, an outcome that was 555 not recapitulated by knocking down each miRNA individually (116). More recently, combined 556 administration of miR-21 and miR-146a has been shown to be more effective in preserving 557 cardiac function following myocardial infarction than administration of either of these miRNAs 558 alone (117). While miRNA synergy has not been explored in detail, these data show that new 559 biology may emerge with admixtures of miRNAs, and that therapeutic interventions may 560 require the use of such miRNA admixtures rather than single miRNA molecules as have been 561 used in clinical studies to date. Our study was restricted to exploring the effects of the 11 562 HEamiRNAs on development, which represent only the significantly increased fraction of miRNAs 563 elevated in the HEa group maternal plasma within our Ukrainian cohort. Indeed, the next 5 564 most elevated miRNAs are also abundantly expressed in the placenta (data not shown), 565 indicating they may also mediate the effects of PAE on placental biology. 566 In conclusion, we have observed how a set of 11 miRNAs, predictive of adverse infant 567 outcomes following PAE, collectively mediate the effects of alcohol on the placenta. Specifically, 568 elevated levels of these miRNAs promote an aberrant maturational phenotype in trophoblasts 569 by inhibiting core members of the EMT pathway and promoting syncytialization-dependent 570 hormone production. Functionally, these miRNAs are clinically correlated with measures of 571 fetal development and directly cause intrauterine growth restriction when administered in vivo. 572 Our work suggests that a greater understanding for the role of HEamiRNAs during development, 573 and their role in coordinating the EMT pathway in the placenta and other developing tissues, 574 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint will benefit the understanding of FASDs and other gestational pathologies and potentially lead 575 to effective avenues for intervention. 576 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint injected via tail vein. These human miRNAs were selected because no mouse homologs are known to 599 exist and consequently, estimates for organ distribution of exogenous miRNAs in the mouse are unlikely 600 to be contaminated by the expression of endogenous murine miRNAs. GD10 is a time point near the 601 beginning of the developmental period of branching morphogenesis, immediately following 602 chorioallantoic attachment, during which the placenta invades the maternal endometrium 603 (120). At GD18, pregnancies were terminated with subsequent quantification of fetal weight, Non-human primate model of PAE: 620 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint As previously described in detail (91), adult female rhesus macaques were trained to orally self-621 administer either 1.5 g/kg/d of 4% ethanol solution (equivalent to 6 drinks/day), or an isocaloric 622 control fluid prior to time-mated breeding. Each pregnant animal continued ethanol exposure 623 until gestational day 60 (GD60, term gestation is 168 days in the rhesus macaque) (122). BeWO cells were treated with 20 µm forskolin to induce syncytialization, as previously 641 described (124,125). BeWO and HTR8 cells were also subjected to four separate ethanol 642 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint treatment conditions: 0 mg/dL, 60 mg/dl (13 mM),120 mg/dl (26 mM) or 320 mg/dl (70 mM). BeWO cells were also plated onto glass coverslips in 24 well plates at a density of 30,000 685 cells/well. After exposure to ethanol and/or forskolin in culture, cells were prepared for calcium 686 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. Quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) analysis: 704 Total RNA was extracted from tissue, well as BeWO and HTR8 cells, using the miRNeasy Mini kit 705 (Qiagen, Cat No. 217004). For miRNA qPCR assays, cDNA was synthesized from 200 ng of total 706 RNA using the miRCURY LNA Universal RT cDNA synthesis kit (Exiqon, Cat No. 203301/Qiagen, 707 Cat No. 339340, Germantown, MD) and expression was assessed using miRCURY LNA SYBR 708 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint Green (Exiqon, Cat No. 203401/Qiagen, Cat No. 339345). For mRNA qPCR assays, cDNA was 709 synthesized from 500 ng of total RNA using the qScript™ cDNA Synthesis Kit (Quanta/Qiagen, The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.  The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint characterizing the individual effects of HEamiRNAs against the control miRNA or antagomirs, 797 individual 2-tailed Student's t-test with 5% FDR correction was applied to account for multiple 798 comparisons. All statistical tests, sample-sizes, and post-hoc analysis are appropriately reported 799 in the results section. A value of p < 0.05 was considered statistically significant, and a value of 800 0.1 < p < 0.05 was considered marginally significant. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint   The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. For post-hoc analysis, ***p<0.001 by Tukey's HSD.

878
All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. For post-hoc analysis, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by Tukey's HSD. 920 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. Results are expressed as expressed as the mean ± SEM; n=10 samples per group; *p<0.05 by 925 Unpaired T-test 926 927 All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder.

1002
All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/409854 doi: bioRxiv preprint