Psychoactive Pharmaceuticals Induce Fish Gene Expression Profiles Associated with Human Idiopathic Autism

Michael A. Thomas1*,  Rebecca D. Klaper2

Michael Thomas

Michael Thomas

1 Department of Biological  Sciences, Idaho State University School, Pocatello, Idaho, United States of America, 2 School of Freshwater  Sciences, University  of Wisconsin- Milwaukee, Milwaukee, Wisconsin, United States of America

Abstract

Idiopathic autism, caused by genetic susceptibility interacting with  unknown environmental triggers, has increased dramatically in  the  past 25 years. Identifying environmental triggers has been difficult  due to  poorly understood pathophysiology and subjective definitions of autism. The use of antidepressants by pregnant women has been associated with autism. These and other unmetabolized psychoactive pharmaceuticals (UPPs) have also been found in drinking water from surface sources, providing another possible exposure route and raising questions about human health consequences. Here, we  examined gene expression patterns of  fathead minnows treated with  a mixture of  three psychoactive pharmaceuticals (fluoxetine, venlafaxine & carbamazepine) in dosages intended to be similar to the highest observed conservative estimates of environmental concentrations. We conducted microarray experiments examining brain tissue of fish exposed to individual pharmaceuticals and a mixture of all three. We used gene-class analysis to test for enrichment of gene sets involved with ten human neurological disorders. Only sets associated with idiopathic autism were unambiguously enriched. We found that UPPs induce autism-like gene expression patterns in fish. Our findings suggest a new potential trigger for idiopathic autism in genetically susceptible individuals involving an overlooked source of environmental contamination.

 

Introduction

Autism spectrum disorders (ASD) are characterized by stereo- typed behaviors and impaired social skills, typically diagnosed by three years of age [1–3]. Idiopathic ASD, caused by genetic susceptibility factors [4–6] interacting with unknown environmen- tal triggers [7,8], has increased dramatically in the past 25 years [9,10]. Identifying environmental triggers has been difficult due to the poorly understood pathophysiology of ASD and broad, subjective case definitions.

In order to serve as such a trigger, a candidate teratogen must have a  biologically plausible etiological mechanism, exist in sufficient environmental  concentrations,  be  capable  of passing from mother to fetus and across the fetal blood-brain barrier (if one assumes prenatal exposure), and have experienced historical increases in environmental concentration parallel with observed increases in ASD prevalence.

Coincident with the observed increase in ASD prevalence is the introduction of modern rationally designed psychoactive pharma- ceuticals, beginning with selective serotonin re-uptake inhibitors (SSRIs) in 1987 (initially fluoxetine, FLX; now 9+ versions), and serotonin–norepinephrine  reuptake inhibitors (SNRIs) in 1994 (initially venlafaxine, VNX; now 8+ versions). These pharmaceu- ticals result in an increase in the neurotransmitter serotonin, which


is responsible for the regulation of neural activity and other physiological functions [11]. The leaky blood brain barrier of the fetus and infants is permeable to many compounds, making this population  particularly vulnerable to the effects of serotonin. Maternal exposure to SSRIs result in elevated fetal plasma serotonin levels, which has been associated with autism [12]. In other work, rat pups exposed prenatally to SSRIs exhibited behaviors associated with ASD [13].

The use of SSRIs and SNRIs by pregnant women to treat depression and other psychological disorders has been associated with low APGAR (Appearance, Pulse, Grimace, Activity, Respi- ration) scores [14], increased risk of spontaneous abortion [15], several other  consequences for children [16–20], and,  recently, ASD [21]. Several other classes of psychoactive pharmaceuticals are also known autism risk factors when taken prenatally [22,23], including valproic acid used to induce the autism rat model [24,25]. Since the 1960s, for example, pregnant women have used carbamazepine (CBZ) for control of seizures, despite its potential association with developmental issues [26]. Studies of these drugs considered only effects of maternal  usage of clinical dosages [21,27]; while widespread, clinical usage of antidepressants is insufficient to account for recent increases in autism prevalence.

There  is an alternative source of exposure to antidepressants: Unmetabolized psychoactive pharmaceuticals (UPPs) found in raw sewage, effluent from sewer treatment facilities, rivers downstream of such facilities, and, ultimately, drinking water [28,29]. Because concentrations  are  so  minute  (typically ng  to  mg?L-1),  human health consequences of UPPs remain controversial. While the highest observed concentrations of UPPs have biological effects in fish [30], these concentrations are many orders of magnitude below human clinical dosages (Table 1). While fetuses and infants

may have been exposed to UPPs through maternal water consumption, it has been assumed that UPPs have no measurable and enduring effects on human health [31,32]. However, multiple related formulations and active metabolites of UPPs present in the environment exist in complex mixtures [33,34] that together constitute much higher dosages, especially in contamination hot- spots [35]. Therefore, we describe the experimental concentrations used here as similar to the highest observed environmental concentrations, despite the fact that experimental concentrations are an order of magnitude higher than the most recent (and probably conservative) estimates of environmental concentrations (Table 1).

Here, we describe results of an experiment that explores a potential  association between  UPPs  and  idiopathic  ASD.  We tested whether chronic exposure to a mixture of UPPs induced autism-like gene expression profiles in a model organism using gene-class analysis. We used treatments  involving a  mixture of UPPs similar to that observed in aquatic systems and examined expression of genes expressed by individuals with various forms of idiopathic ASD. For comparison, we examined sets of genes expressed by individuals with other neurological disorders.

 

Gene Expression Analysis

We  exposed  fathead  minnows,  Pimephales   promelas,  to  FLX, VNX, and CBZ in a 3-component mixture. FLX, VNX, and CBZ were chosen because they represent modern pharmaceutical classes that are highly prescribed and are among the UPPs with the highest observed environmental concentrations (Table 1). We conducted gene-class analysis of expression patterns induced by the pharmaceutical treatments using Gene Set Enrichment Analysis (GSEA) [36] and an enhanced annotation of the fathead microarray platform [37]. The gene-class analysis approach tests if a set of genes, described a priori, is enriched by a given treatment. The data for the present study were derived from a previous analysis of this system [38]. In that study, we found enrichment of gene sets associated with neurological development, growth and regulation by the mixture of UPPs. These genes sets were not enriched by treatments of the pharmaceuticals administered separately, and were associated with the formation and regulation of neural circuits, which may indicate formation of altered and imprecise synaptic connections  and  presage  a  failure to  form typical mature neural circuits.

In the present study, we first tested the prediction that UPPs would induce  fish gene expression profiles that  mimic human expression profiles observed in individuals diagnosed with various neurological disorders (‘‘ND’’ sets, described  in  Table  2). We tested 12 sets of genes associated with idiopathic ASD (broadly defined), autism secondary to  known genetic defects (involving fragile X and Rett syndromes), Alzheimer’s disease, Parkinson’s disease, schizophrenia, multiple sclerosis, major  depression, bipolar disorder and ADHD.

Second, we tested the prediction that UPPs would induce fish gene expression profiles that  mimic human  expression profiles observed in individuals diagnosed with autism of various degrees of severity. The  idiopathic autism set used above (described in Table 2) was derived from several independent  gene expression studies with minimal overlap of gene constituents, each of which identified genes enriched in individuals diagnosed with idiopathic autism. One of these sets [39] was deconstructed into specific populations classified by severity of autism symptoms. We created a  second  collection consisting of 10  autism  sets (‘‘ASD’’  sets; Table 3): nine sets associated with gene expression in individuals diagnosed with some form of autism plus a set of susceptibility genes not associated with any known increase in gene expression but, rather, known to have either mutations (e.g., single nucleotide polymorphisms) or structural variations (e.g., copy number variation) associated with some form of autism [4].

 

Results

In the analysis of the ND sets, seven of 12 sets were up- regulated, with significant enrichment of the set associated with idiopathic autism (Table 4). We also observed enrichment of two disorders with adult onset: a set associated with Parkinson’s and one set associated with MS (but not a second, independent  MS set). There  was no overlap between the autism set and the Parkinson’s & MS sets; 4 genes occurred in both MS and Parkinson’s sets. No other set associated with human neurological disorders was enriched, including secondary autism sets.

In the analysis of ASD sets, all 10 sets were up-regulated, with significant enrichment  of five of the  nine  expression-based sets (Table 5). Very few genes are shared among enriched sets. The susceptibility set was not enriched; this is significant because one would not necessarily expect genes that underlie susceptibility to also experience differential expression. Interestingly, one of the strongly enriched  sets (Voineagu_Down) was identified by  the study authors as genes down-regulated in individuals with autism

 

Table 1. Observed values of psychoactive pharmaceuticals in various systems.

Source FLX VNX CBZ
Experimental 10 mg?L21 50 mg?L21 100 mg?L21
Raw sewage 0.073 mg?L21[50] 2.19 mg?L21[51] 6.3 mg?L21[52]
Wastewater treatment plant (WWTP) 0.509 mg?L21[53] 1.115 mg?L21[47] 17.3–22.0 mg?L21[54,55]
Effluent from WWTP 0.841 mg?L21[56] No information 1.16 mg?L21[57]
Downstream from WWTP 0.93 mg?L21[55] 0.387 mg?L21[57] 2.3 mg?L21[47]
River system 0.12 mg?L21[58] 1.31 mg?L21[51] 1.283 mg?L21[52]
Drinking water 0.014 mg?L21[59,60] No information 0.25 mg?L21[60,61]

FLX, VNX and CBZ are fluoxetine, venlafaxine and carbamazepine, respectively. Values reported indicate the highest observed concentrations from various systems. Experimental treatment dosage was selected to reflect combined dosages of multiple active metabolites for each pharmaceutical. doi:10.1371/journal.pone.0032917.t001

 

 Table 2. ‘‘ND’’: Sets of genes associated with various human neurological disorders.

Set name Description Number  of genes (set) Number  of genes (FH microarray) GEO link Reference
ASD_Idiopathic Combination of Chakrabarti, Hu & ASD_2Class; duplicates were removed(see Table S3 for details). 324
ADHD_up A comparison of molecular alterations in environmental and genetic rat models of ADHD (up-regulated genes) 50 30 GSE12457 [62]
ADHD_down A comparison of molecular alterations in environmental and genetic rat models of ADHD (down-regulated  genes) 34 20 GSE12457 [62]
Rett Genes found to be up-regulated in females with Rett Syndrome 39 25 [63]
ASD_Secondary Gene expression profiles of lymphoblastoid  cells from individuals with fragile X syndrome and dup(15q). 67 39 GSE7329 [64]
Schizophrenia Proteins consistently differentially expressed in the brains of SCZ patients. 30 23 [65]
Alzheimers Up-regulated in correlation with incipient Alzheimer’s  Disease, in the CA1 region of the hippocampus 345 237 GSE1297 [66]
Parkinsons Genes associated with Parkinson’s Disease. 162 94 KEGG HSA05012
Depression Genes upregulated in major depressive disorder (p,0.05, fold change .1.4, mean average difference .150 in at least one of the groups, called presentin greater than 20% of all samples) 45 23 GSE12654 [67]
Bipolar Genes found to be up-regulated in individuals with bipolar disorder. 71 41 [68]
MS_Bomprezzi In an attempt to identify molecular markers indicative of disease status rather than susceptibility genes for MS, the authors show that gene expression profiling of peripheral blood mononuclear  cells by cDNA microarrays candistinguish  MS patients from healthy controls. 45 28 [69]
MS_Gilli Results showed an altered expression of 347 transcripts in non-pregnantMS patients with respect to non-pregnant healthy controls. 348 216 GSE17393 [70]

doi:10.1371/journal.pone.0032917.t002

 

[40], while its complementary set (Voineagu_Up) was not enriched. The former set was endowed with an overrepresentation of gene ontology categories associated with synapse function, while the latter had an overrepresentation of categories associated with immune and inflammatory response.

Normalized enrichment scores (NES; Tables 4, 5) for the examined sets ranged from -1.172 to 1.621 for ND sets (58% up- regulated) and 1.050 to 1.537 for ASD sets (100% up-regulated). While the nominal p-value indicates significance of an NES, that value is somewhat difficult to interpret.  For comparison, we analyzed 242 curated  gene sets from the Molecular Signatures Database  [36] corresponding to each  human  cytogenetic band that has at least one gene. In that analysis, 52% of the sets were up-regulated, and NES values ranged from -1.70 to 1.73. Only 5% of scores exceeded an NES of 1.40, and no set passed the false discovery rate (FDR) threshold of 0.25.

The analyses in this report consider mainly the MIX treatment of UPPs. For comparison, we included GSEA analysis results from treatments consisting of the three pharmaceuticals considered individually. As one  might expect, those treatments  had  lower NES values and fewer enriched sets relative to the MIX treatment (Tables 6, 7), although the single-drug NES values involving Parkinson’s were all higher than the MIX treatment NES values.

 

Discussion

We found enrichment of gene sets associated with idiopathic ASD but not of sets involving autism diagnoses secondary to other disorders (Rett and fragile X syndromes) known to be caused by specific mutations.  This  is significant, because it indicates that enrichment  in  our  treatments  involve only idiopathic forms of ASD.

There was no enrichment of other neurological disorders except MS (in one of two sets) and Parkinson’s. This is significant because it indicates that  enrichment  of the idiopathic ASD set was not simply associated with general neurological processes, pathways or systems generally common to neurological disorders. The MS set has a low NES (,1.40), and a second MS set (see Table 4) is not enriched; therefore, we are not exceedingly confident in describing that set as enriched. The other enriched non-ASD set, Parkinson’s, is more interesting, with a convincing NES and an intriguing potential  connection  to  ASD involving similar phenomenology involving brain dysfunction [41].

A number of the genes contributing to enrichment of ASD gene sets have been implicated in other recent studies not included in our  analysis. For example, Suda  et al. [42] found that  relative expression levels of EFNB3, PLXNA4 and  ROBO2  were significantly different in individuals with autism than in neuroty- pical individuals; protein levels of PLXNA4 and ROBO2, but not for EFNB3, were significantly reduced in brains of individuals with autism compared to control brains. In the present study, we found

3 plexin genes (PLXNB1, PLXND1 & PLXNA3; PLXNA4 was not on the array) and ROBO2  to be strongly down-regulated in response to the MIX  treatment,  while EFNB3 and  five related genes (EFNB1, EFNA1, EFNA2, EFNA3 & EFNA5) were up- regulated. These and other genes contributing to gene set enrichment are associated with the formation of synapses, perturbation of which may indicate an altered and imprecise synaptic connections or a failure to form mature neural circuits.

The  results presented here are consistent with several recent lines of inquiry: First, the hypothesis that hyperserotonemia plays a

 

Table 3. ‘‘ASD’’: Collection  of sets of genes associated with idiopathic autism.

Set name Description Number  of genes (set) Number  of genes (FH microarray) GEO link Reference
Pinto Genes associated with genetic susceptibility to ASD but not 104 63 [4]
known to be up- or down-regulated in the disorder.
Chakrabarti Genes related to sex steroids, neural growth, and social-emotional 66 43 [71]
behavior associated with autistic traits, empathy, and Asperger’s
syndrome. Included only mild cases (no severe language impairment).
Hu Gene identified by comparisons of neurotypical  vs. ASD individuals  with 34 22 [72]
severe language impairment (with individuals having specific genetic and
chromosomal abnormalities and co-morbid disorders excluded from study).
ASD_2Class Significantly differentially  expressed genes from a 2-class SAM analysis 370 240 GSE15402 [39]
of data from combined autistic samples and neurotypical controls,
with FDR ,5%
ASD_Mild Significantly differentially  expressed genes from a 2-class SAM analysis 360 241 GSE15402 [39]
of data from the group with mild ASD (M) and neurotypical  controls (C ),
with FDR ,5%
ASD_Severe Significantly differentially  expressed genes from a 2-class SAM analysis of 191 121 GSE15402 [39]
data from the group with severe language impairment (L) and neurotypical
controls (C ), with FDR ,0.0001%
ASD_Shared Common genes to all GSE15402 sets. 70 48 GSE15402 [39]
ASD_Savant Significantly differentially expressed genes from a 2-class SAM analysis of data<

 

role in autism, affecting the developing fetus and potentially involving SSRIs [27]. In that  work, Hadjikhani explored a potential role of elevated serotonin levels perturbing  brain development  during   pregnancy   (in  which  he   assumes  that maternal serotonin ultimately passes the fetal blood brain barrier). The author speculated that elevated levels could be increased by maternal use of serotonin elevating pharmaceuticals (like SSRIs) or consumption of serotonin-rich foods. In the present study, serotonin levels were not  measured. However, all six serotonin receptor genes on the array (HTR1A, HTR1B, HTR2C,  HTR4, HTR7&  SLC6A4) were strongly down-regulated in response to the MIX  treatment.  If this implies a consequential elevation of serotonin levels, our results would seem to be consistent with the

 

Table 4. Sets associated with human neurological disorders.


Hadjikhani hypothesis [27] and  with other recent experimental work using model organisms [13].

Second, recent evidence supports an association of antidepres- sants, including SSRIs, with autism [21]. In that study, Croen and colleagues found a 2-fold increase in ASD risk associated with SSRIs, with the strongest effect occurring in the first trimester. The results of the present study are consistent with this finding. However,  maternal  SSRI  use  is not  sufficient to  explain  the increase in prevalence of ASD.

Third, there is evidence for an unambiguous environmental component involved in the etiology of autism [7]. In that study, Hallmayer  and  colleagues provide  robust  evidence that,  while having a moderate genetic component, ASD also clearly involves an environmental trigger. The results of the present study are consistent with this finding, as is the assumption that the environmental trigger acts in concert with genetic susceptibility.

Fourth,  there  is evidence of demographic changes that  may have increased the proportion of genetically susceptible individuals

 

 

Set Size NES p-value FDR q-value
AUTISM_IDIOPATHIC 324 1.621 0.000 0.064
PARKINSONS

94

1.560 0.007 0.055
MS_GILLI 216 1.375 0.011 0.137
SCHIZOPHRENIA

23

1.232 0.181 0.364
MS_BOMPREZZI

28

1.199 0.201 0.326
ADHD_UP

30

1.187 0.222 0.275
DEPRESSION

23

1.137 0.307 0.293
ADHD_DOWN

20

–0.684 0.894 0.924
RETT

25

–0.784 0.798 1.000
ALZHEIMERS 237 –0.967 0.549 0.859
ASD_SECONDARY

39

–1.083 0.332 0.764
BIPOLAR

41

–1.172 0.217 1.000

 

Table 5. Analysis of sets associated with human autism.

 

Sets are described in Table 2; size refers to the number of genes in the set; NES is the normalized enrichment  scores for the set; p-value is the nominal p-value associated with the NES; FDR q-value  is the false discovery rate ratio. doi:10.1371/journal.pone.0032917.t004

Set Size NES p-value FDR q-value
ASD_MILD 241 1.537 0.0000 0.1261
ASD_2CLASS 240 1.519 0.0000 0.0742
VOINEAGU_DOWN 121 1.514 0.0017 0.0511
ASD_SAVANT

60

1.474 0.0298 0.0537
ASD_SHARED

48

1.459 0.0391 0.0489
CHAKRABARTI

43

1.358 0.0769 0.0864
HU

22

1.352 0.1168 0.0777
ASD_SEVERE 121 1.261 0.0781 0.1233
VOINEAGU_UP 132 1.117 0.2092 0.2749
PINTO

63

1.050 0.3558 0.3558

 

Column labeled as in Table 4. Sets are described in Table 3. doi:10.1371/journal.pone.0032917.t005

 

Set NES         p-value       FDR q-value      NES         p-value       FDR q-value      NES           p-value       FDR q-value

 

Table 6. Single drug treatments & ASD sets.

 

 

FLX

FLX FLX VNX VNX VNX CBZ CBZ CBZ

 

ASD_SAVANT 1.443 0.036 0.144 1.344 0.075 0.174 1.287 0.103 0.138
ASD_MILD 1.429 0.003 0.081 1.370 0.018 0.215 1.531 0.000 0.099
ASD_SHARED 1.362 0.066 0.089 1.331 0.088 0.143 1.115 0.244 0.322
ASD_2CLASS 1.290 0.022 0.118 1.411 0.005 0.289 1.490 0.002 0.076
ASD_SEVERE 1.199 0.115 0.172 0.979 0.514 0.634 1.342 0.039 0.155
CHAKRABARTI 1.144 0.246 0.206 0.887 0.662 0.695 1.017 0.432 0.475
VOINEAGU_DOWN 20.837 0.831 0.809 1.116 0.220 0.443 1.288 0.054 0.172
VOINEAGU_UP 20.899 0.701 0.906 20.909 0.716 0.670 -1.018 0.404 0.415
PINTO 21.061 0.339 0.677 1.018 0.420 0.627 -1.065 0.337 0.659
HU 21.378 0.080 0.150 0.901 0.582 0.747 0.893 0.591 0.697

FLX, VNX and CBZ are fluoxetine, venlafaxine and carbamazepine, respectively. Column labeled  as in Table 4. Sets are described in Table 3. doi:10.1371/journal.pone.0032917.t006

 

 

in contemporary populations [43]. In that study, Baron-Cohen proposed  that  assortative mating  among  genetically susceptible individuals has increased the proportion of susceptible individuals in human  populations since the 1970s. Especially when coupled with increased levels of an environmental trigger, this would create circumstances in which one would expect an  increase in ASD prevalence. Given that SSRIs were introduced in the mid-1980s and SNRIs in the mid-1990s, coincident with increases in ASD prevalence [10], the assortative mating hypothesis provides a framework for understanding why such a trigger is able to induce such a  large effect. The  results of the  present study provide a potential source of exposure to psychoactive pharmaceuticals that does not involve maternal clinical usage of SSRIs.

Given the conserved nature (i.e., sequence and function) of the genes involved in the observed expression profiles, and given that the genes on the Fathead array are homologous to highly conserved human  genes, it is reasonable to expect induction of humans gene expression profiles similar to the Fathead profiles. This sort of approach has been effectively used for other models of


human disorders [44] and in previous investigations involving the Fathead  microarray platform [37]. Here,  many of the enriched sets involve genes associated with neuronal development and growth [38], which is consistent with systems and pathways known to be perturbed in the developing brain of individuals with autism [40,45].

The   concentrations  used  in  this  study  were  higher  than observed environmental concentrations in order to account for conservative concentration estimates and the presence of related formulations and  active metabolites [29,34,46,47]. Future  work needs to be conducted to measure the concentrations of all UPP constituents present in aquatic systems and drinking water (with appropriate  temporal and  geographic sampling) in order  to accurately assess human exposure and health consequences.

 

Conclusions

These  results provide  a  new perspective on  the  etiology of idiopathic  ASD  and  suggest new  directions  for  research  into

 

 

Set NES         p-value       FDR q-value      NES         p-value       FDR q-value      NES         p-value       FDR q-value

 

Table 7. Single drug treatments & ND sets.

 

 

FLX

FLX FLX VNX VNX VNX CBZ CBZ CBZ

 

PARKINSONS 1.650 0.000 0.058 1.780 0.000 0.011 2.110 0.000 0.000
AUTISM_IDIOPATHIC 1.505 0.000 0.219 1.412 0.004 0.390 1.5104 0.0000 0.1930
RETT 1.164 0.237 0.738 1.079 0.338 0.452 0.7529 0.8097 0.8795
SCHIZOPHRENIA 1.104 0.316 0.661 1.164 0.251 0.497 1.2040 0.2079 0.2905
ADHD_UP 1.102 0.327 0.501 1.155 0.243 0.390 1.5102 0.0397 0.0965
MS_GILLI 0.997 0.500 0.639 1.364 0.017 0.265 1.0616 0.2827 0.4790
MS_BOMPREZZI 0.904 0.589 0.755 0.857 0.687 0.732 0.9935 0.4295 0.5343
ASD_SECONDARY 0.631 0.974 0.965 –0.749 0.896 0.870 –0.622 0.970 0.968
BIPOLAR –0.774 0.832 0.856 –1.401 0.047 0.300 –0.812 0.773 1.000
ALZHEIMERS –1.118 0.204 0.447 –0.847 0.906 0.928 –0.934 0.640 0.969
DEPRESSION –1.170 0.248 0.523 0.919 0.583 0.717 1.2313 0.1871 0.3349
ADHD_DOWN –1.715 0.004 0.019 –1.375 0.110 0.177 –1.322 0.152 0.465

FLX, VNX and CBZ are fluoxetine, venlafaxine and carbamazepine, respectively. Column labeled  as in Table 4. Sets are described in Table 2. doi:10.1371/journal.pone.0032917.t007

 

Citation:Thomas MA, Klaper RD (2012) Psychoactive  Pharmaceuticals  Induce Fish Gene Expression Profiles Associated with Human Idiopathic Autism. PLoS ONE 7(6): e32917. doi:10.1371/journal.pone.0032917Editor:  Efthimios M. C. Skoulakis, Alexander  Flemming  Biomedical  Sciences Research Center, GreeceReceived September  2, 2011; Accepted  February 6, 2012; Published June 6, 2012

Copyright: © 2012 Thomas, Klaper. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted  use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: MAT was supported by a PhRMA Foundation  Sabbatical Fellowship grant, National Institutes of Health Grant Number P20 RR016454 from the INBRE Program of the National Center for Research Resources, and grant number URC-FY2010-05 from the University  Research Committee of Idaho State University. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing Interests:  The authors have declared that no competing interests exist.

* E-mail: mthomas@isu.edu

 

 

 

autism’s environmental ‘‘exposome’’ [48]. The results of the gene expression study indicate that a mixture of UPPs can induce an ASD-like gene expression profile in a model organism. Using a low-cost model system like fathead minnow, researchers can rapidly screen potential teratogens for their ability to induce ASD- like gene expression patterns  in developing brains. In  order  to clearly determine if UPPs are associated with idiopathic ASD in humans, future work needs to examine a wider palette of UPPs (and other potential teratogens) and results need to be validated by demonstrating treatment response in another model systems. This could involve using a mouse model, with which one could measure fetal brain expression patterns, UPP concentration in fetal blood, and concentrations of fetal neurohypophyseal hormones, following maternal  treatment.  Further,  epidemiological studies at the individual patient level should be conducted to confirm and specify the relationship between environmental contaminants and ASDs. The mimicry of ASD-like gene expression profiles in fish, described above, does not conclusively indicate UPP induction of ASD in humans.  It  does, however, serve as the  basis for new hypotheses regarding the etiology of idiopathic ASD.

 

Materials and Methods

 

Ethics Statement

All fish handling and treatments were performed at the Great

Lakes WATER Institute (School of Freshwater Sciences, Univer- sity of Wisconsin-Milwaukee, Milwaukee, Wisconsin) using appropriate UWM Institutional Animal Care and Use Committee (IACUC) approved protocols (approval number 0708#14).

 

Fish Treatments

Full details of the fish treatments are described in a previous

report  [38].  Briefly, three  2-gallon tanks  were  used  for  each pharmaceutical treatment along with three tanks for a mixture treatment (containing all three pharmaceuticals in the concentra- tions listed in Table 1) and three tanks for control (containing no pharmaceuticals). Each  tank  housed five juvenile fathead  min- nows. Dosages of pharmaceuticals were re-administered with each change of the tank water (every 2 days). Fish were exposed to treatments for eighteen days.

 

Gene Set Enrichment  Analysis

Fish mRNA was pooled within a tank for microarray work (for

3 replicates per treatment) but not for qPCR  validation (for 15 replicates per treatment). Details of microarray experiments, including validation by qPCR analysis of 9 genes with high rank correlation and all data files, are described in a previous report [38]. Microarray experiments conformed to MIAME guidelines and results were deposited in GEO  (GSE22261).

The  previous study [38] also described an altered phenotype associated with pharmaceutical treatment  that involved measur- ing fish behavior in response to a startle stimulus modeled after predator avoidance behavior used elsewhere [49]. We found that fish behavior was indicative of a neurologically relevant phenotype:  following a  ‘‘startle,’’ the   distance  traveled  and number  of  direction  changes  both  significantly increased  for treated fish [38].

Here,  two groups of gene sets gene-class analyses were conducted:  ND  (‘‘neurological disorder’’) and  ASD  (‘‘autism spectrum disorder’’). The gene sets in the ND collection, known


to be associated with a variety of human neurological disorders, are described in Table 2. The  gene sets in the ASD collection, associated with enriched gene expression in autism, are described in   Table   3.   Both   gene   sets  are   provided   in   Supporting Information  (Table S1, ‘‘ND gene list,’’ and  Table  S2, ‘‘ASD gene list’’). Each group consisted of a collection of gene sets, with each set tested against the ranked list of genes reflecting signal-to- noise ratio of MIX  treatment  (combining FLX, VNX  & CBZ) relative to control. Additional comparisons between the control and  treatments  consisting of the  three  pharmaceuticals consid- ered individually were included for comparison (Tables 6, 7).

Gene-class analyses used GSEA release 2.06 and MSigDB release 2.5. Weighted enrichment scores were calculated using gene expression lists ranked by signal-to-noise ratio. The  genes on the array were ranked by correlation between the MIX and CTL  treatments (those genes with the strongest up-regulation in treatment relative to control were ranked highest; those with strongest down-regulation were ranked  lowest). (See Table  S3,

‘‘Ranked gene list,’’ for these data.) The maximum gene set size was set to 500 genes; the minimum gene set size was set to 10 genes; the number of permutations was set to 1000. Permutations were  conducted  by  gene  set  (rather  than  by  phenotype). For details of GSEA parameter  usage, see Subramanian  et al. [36]. Gene  sets  were  examined  to  ensure  they  contained  only GSEA-recognized primary HUGO  symbols, rather  than  aliases or unapproved symbols. This was accomplished through the use of a custom script that compared each gene in a given set to the GENE_SYMBOLS.chip  file (from GSEA) containing  a  list of HUGO   symbols with  accepted  aliases. Gene  set  components listed as aliases in this file were replaced with the  appropriate HUGO   symbol. For  additional  details of the  annotation  and GSEA implementation using the EcoArray 15k Fathead Minnow

arrays, see Thomas et al. [37].

 

Supporting Information

 

Table S1    ND  gene  list. A list of the neurological disorders gene sets, in the GSEA gene matrix (.GMX) file format.

(GMX)

 

Table  S2    ASD  gene   list.  A  list  of  the  autism  spectrum disorders gene sets, in the GSEA gene matrix (.GMX) file format. (GMX)

Table S3    Ranked gene  list. A Microsoft Excel spreadsheet containing all annotated array gene elements sorted by signal-to- noise ratio for the mixture v. control comparison.

(XLS)

 

Acknowledgments

 

D. Arndt and J. Crago provided support and expertise on lab techniques and fish handling. C. Ryan and E. O’Leary-Jepsen provided critical expertise in support of the qPCR  analysis. R.  Salmore, P. Hallock, L. Yang, S. St. Hillaire, and G. Kaushik provided feedback on an early draft of the manuscript.

 

Author Contributions

Conceived and designed the experiments: MAT RDK. Performed the experiments: MAT. Analyzed the data: MAT. Contributed  reagents/ materials/analysis tools: MAT RDK. Wrote the paper: MAT RDK.

 

 

 

References

 

1.  Landrigan  PJ  (2010)   What  causes  autism?  Exploring  the  environmental contribution. Current Opinion in Pediatrics 22: 219–225.

 

 

2.  Currenti  SA (2010) Understanding and Determining the Etiology of Autism.

Cellular and Molecular Neurobiology 30: 161–171.

 

 

 

 

3.  Hughes JR (2009) Update  on autism: A review of 1300 reports published in

2008. Epilepsy & Behavior 16: 569–589.

4.  Pinto D, Pagnamenta AT, Klei L, Anney R, Merico D, et al. (2010) Functional impact of global rare copy number variation in autism spectrum disorders. Nature 466: 368–372.

5.  Sebat J, Lakshmi B, Malhotra D, Troge J, Lese-Martin C, et al. (2007) Strong association of de novo copy number mutations with autism. Science 316: 445–

449.

6.  Glessner JT, Wang K, Cai GQ,  Korvatska O, Kim CE, et al. (2009) Autism genome-wide copy number variation reveals ubiquitin and neuronal genes. Nature 459: 569–573.

7.  Hallmayer J, Cleveland S, Torres A, Phillips J, Cohen B, et al. (2011) Genetic Heritability  and  Shared  Environmental  Factors  Among  Twin  Pairs  With Autism. Arch Gen Psychiatry: archgenpsychiatry.2011.2076.

8.  Johnson CP, Myers SM, Coun Children D (2007) Identification and evaluation of children with autism spectrum disorders. Pediatrics 120: 1183–1215.

9.  Rice C, Nicholas J, Baio J, Pettygrove S, Lee LC, et al. (2010) Changes in autism spectrum disorder prevalence in 4 areas of the United  States. Disability and Health Journal 3: 186–201.

10.  McDonald  ME,  Paul  JF  (2010)   Timing  of  Increased  Autistic  Disorder

Cumulative Incidence. Environmental Science & Technology 44: 2112–2118.

11.  Berger M,  Gray  JA, Roth  BL (2009) The  Expanded  Biology of Serotonin.

Annual Review of Medicine 60: 355–366.

12.  McNamara  IM,  Borella  AW,  Bialowas LA,  Whitaker-Azmitia PM  (2008) Further studies in the developmental hyperserotonemia model (DHS) of autism: Social, behavioral and peptide changes. Brain Research 1189: 203–214.

13.  Simpson KL, Weaver KJ, de Villers-Sidani E, Lu JYF, Cai ZW, et al. (2011) Perinatal antidepressant exposure alters cortical network function in rodents. Proceedings of the National Academy of Sciences of the United States of America 108: 18465–18470.

14.  Casper  RC,  Fleisher BE, Lee-Ancajas JC, Gilles A, Gaylor E, et al. (2003) Follow-up of children of depressed mothers exposed or not exposed to antidepressant drugs during pregnancy. Journal of Pediatrics 142: 402–408.

15.  Nakhai-Pour  HR,  Broy P,  Berard  A  (2010) Use  of antidepressants during pregnancy and the risk of spontaneous abortion. Canadian Medical Association Journal 182: 1031–1037.

16.  Pedersen LH, Henriksen TB, Olsen J (2010) Fetal Exposure to Antidepressants and Normal Milestone Development at 6 and 19 Months of Age. Pediatrics 125: E600–E608.

17.  Favreliere  S,  Nourrisson  A,  Jaafari  N,  Pochat  MCP  (2010) Treatment   of depressed pregnant women by selective serotonin reuptake inhibitors: Risk for the foetus and the newborn. Encephale-Revue De Psychiatrie Clinique Biologique Et Therapeutique  36: D133–D138.

18. Gentile S (2010) Neurodevelopmental effects of  prenatal  exposure to psychotropic medications. Depression and Anxiety 27: 675–686.

19.  Oberlander  TF, Gingrich JA, Ansorge MS (2009) Sustained Neurobehavioral Effects of Exposure to SSRI Antidepressants During Development: Molecular to Clinical Evidence. Clinical Pharmacology & Therapeutics 86: 672–677.

20.  Oberlander TF, Papsdorf M, Brain UM, Misri S, Ross C, et al. (2010) Prenatal Effects of Selective Serotonin  Reuptake  Inhibitor  Antidepressants, Serotonin Transporter Promoter Genotype (SLC6A4), and Maternal Mood on Child Behavior at 3 Years of Age. Archives of Pediatrics & Adolescent Medicine 164:

444–451.

21. Croen LA, Grether JK, Yoshida CK, Odouli R, Hendrick V (2011) Antidepressant Use During Pregnancy and Childhood Autism Spectrum Disorders. Arch Gen Psychiatry: archgenpsychiatry.2011.2073.

22.  Rasalam AD, Hailey H, Williams JHG, Moore SJ, Turnpenny PD, et al. (2005) Characteristics of fetal anticonvulsant syndrome associated autistic disorder. Developmental Medicine and Child Neurology 47: 551–555.

23.  Bromley RL, Baker GA, Meador KJ (2009) Cognitive abilities and behaviour of children exposed to antiepileptic drugs in utero. Current Opinion in Neurology

22: 162–166.

24.  Dufour-Rainfray D, Vourc’h P, Le Guisquet AM, Garreau L, Ternant D, et al. (2010) Behavior and serotonergic disorders in rats exposed prenatally to valproate: A model for autism. Neuroscience Letters 470: 55–59.

25.  Schneider T, Roman A, Basta-Kaim A, Kubera M, Budziszewska B, et al. (2008) Gender-specific behavioral and immunological alterations in an animal model of autism induced by prenatal exposure to valproic acid. Psychoneuroendocrinol- ogy 33: 728–740.

26.  Bromley RL, Mawer G, Clayton-Smith J, Baker GA, Liverpool, et al. (2008) Autism spectrum disorders following in utero exposure to antiepileptic drugs. Neurology 71: 1923–1924.

27.  Hadjikhani N (2010) Serotonin, pregnancy and increased autism prevalence: Is there a link? Medical Hypotheses 74: 880–883.

28.  Jjemba PK (2006) Excretion and ecotoxicity of pharmaceutical and personal care products in the environment. Ecotoxicology and Environmental Safety 63:

113–130.

29. Santos L, Araujo AN, Fachini A, Pena A, Delerue-Matos C, et al. (2010) Ecotoxicological aspects related  to  the  presence  of  pharmaceuticals  in  the aquatic environment. Journal of Hazardous Materials 175: 45–95.

30.  Mennigen JA, Sassine J, Trudeau VL, Moon TW (2010) Waterborne fluoxetine disrupts  feeding and  energy  metabolism  in  the  goldfish Carassius  auratus. Aquatic Toxicology 100: 128–137.


31.  Cunningham  VL, Binks SP, Olson MJ (2009) Human  health risk assessment from the presence of human pharmaceuticals in the aquatic environment. Regulatory Toxicology and Pharmacology 53: 39–45.

32.  Schwab BW, Hayes EP, Fiori JM, Mastrocco FJ, Roden  NM,  et al. (2005) Human pharmaceuticals in US surface waters: A human health risk assessment. Regulatory Toxicology and Pharmacology 42: 296–312.

33.  Metcalfe CD, Chu SG, Judt C, Li HX, Oakes KD, et al. (2010) Antidepressants and their metabolites in municipal wastewater, and downstream exposure in an urban watershed. Environmental Toxicology and Chemistry 29: 79–89.

34. Celiz MD,  Tso J, Aga DS (2009) Pharmaceutical metabolites in the environment: Analytical challenges and ecological risks. Environmental Toxicology and Chemistry 28: 2473–2484.

35. Madureira TV, Barreiro JC, Rocha MJ, Rocha E, Cass QB, et al. (2010) Spatiotemporal distribution of pharmaceuticals in the Douro River estuary (Portugal). Science of the Total Environment 48: 5513–5520.

36.  Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, et al. (2005) Gene  set enrichment  analysis: A knowledge-based approach  for interpreting genome-wide expression profiles. Proceedings of the National Academy of Sciences of the United States of America 102: 15545–15550.

37.  Thomas  MA, Yang LB, Carter  BJ, Klaper  RD  (2011) Gene set enrichment analysis of microarray  data  from Pimephales promelas (Rafinesque), a  non- mammalian model organism. BMC Genomics 12: 66.

38.  Thomas  MA, Joshi PP, Klaper  RD  (2011) Gene-class analysis of expression patterns induced by psychoactive pharmaceutical exposure in fathead minnow (Pimephales promelas) indicates induction of  neuronal  systems. Comparative Biochemistry and Physiology Part C: Toxicology & Pharmacology 155: 109–

120.

39.  Hu VW, Sarachana T, Kim KS, Nguyen A, Kulkarni S, et al. (2009) Gene Expression Profiling Differentiates Autism Case-Controls and Phenotypic Variants of Autism Spectrum Disorders: Evidence for Circadian Rhythm Dysfunction in Severe Autism. Autism Research 2: 78–97.

40.  Voineagu I, Wang X, Johnston P, Lowe JK, Tian Y, et al. (2011) Transcriptomic analysis of autistic brain reveals convergent molecular pathology. Nature advance online publication.

41.  Hollander E, Wang AT, Braun A, Marsh L (2009) Neurological considerations: Autism and Parkinson’s disease. Psychiatry Research 170: 43–51.

42.  Nakamura K, Iwata Y, Anitha A, Miyachi T, Toyota T, et al. (2011) Replication study of Japanese cohorts supports the role of STX1A in autism susceptibility. Progress in Neuro-Psychopharmacology & Biological Psychiatry 35: 454–458.

43.  Baron-Cohen  S  (2006)  The  hyper-systemizing, assortative mating  theory  of autism.  Progress in  Neuro-Psychopharmacology & Biological Psychiatry 30:

865–872.

44.  McGary KL, Park TJ, Woods JO, Cha HJ, Wallingford JB, et al. (2010) Systematic discovery of nonobvious human disease models through orthologous phenotypes. Proceedings of the National Academy of Sciences of the United States of America 107: 6544–6549.

45.  Dufour-Rainfray D, Vourc’h P, Tourlet S, Guilloteau D, Chalon S, et al. (2011) Fetal exposure to teratogens: Evidence of genes involved in autism. Neuroscience and Biobehavioral Reviews 35: 1254–1265.

46.  Olver JS, Burrows GD, Norman TR  (2004) The treatment of depression with different formulations of venlafaxine: a comparative analysis. Human  Psycho- pharmacology-Clinical and Experimental 19: 9–16.

47.  Metcalfe CD, Miao XS, Koenig BG, Struger J (2003) Distribution of acidic and neutral drugs in surface waters near sewage treatment plants in the lower Great Lakes, Canada. Environmental Toxicology and Chemistry 22: 2881–2889.

48.  Rappaport SM, Smith MT (2010) Environment and Disease Risks. Science 330:

460–461.

49.  Painter MM, Buerkley MA, Julius ML, Vajda AM, Norris DO,  et al. (2009) Antidepressants at environmentally relevant concentrations affect predator avoidance behavior of larval fathead minnows (Pimephales promelas). Environ- mental Toxicology and Chemistry 28: 2677–2684.

50.  Batt AL, Kostich MS, Lazorchak JM (2008)  Analysis of ecologically relevant pharmaceuticals  in  wastewater and  surface water  using selective solid-phase extraction and UPLC-MS/MS. Analytical Chemistry 80: 5021–5030.

51.  Schultzt MM, Furlong ET (2008) Trace analysis of antidepressant pharmaceu- ticals and their select degradates in aquatic matrixes by LC/ESI/MS/MS. Analytical Chemistry 80: 1756–1762.

52.  Ternes TA (1998) Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32: 3245–3260.

53.  Chen  M, Ohman  K, Metcalfe C, Ikonomou MG,  Amatya PL, et al. (2006) Pharmaceuticals and endocrine disruptors in wastewater treatment effluents and in  the  water  supply  system of  Calgary,  Alberta,  Canada.   Water  Quality Research Journal of Canada 41: 351–364.

54.  Camacho-Munoz  MD,  Santos JL, Aparicio I,  Alonso E  (2010)  Presence of pharmaceutically active compounds in Donana  Park (Spain) main watersheds. Journal of Hazardous Materials 177: 1159–1162.

55.  Christensen AM, Markussen B, Baun A, Halling-Sorensen B (2009) Probabilistic environmental risk characterization of pharmaceuticals in sewage treatment plant discharges. Chemosphere 77: 351–358.

56.  Corcoran  J,  Winter  MJ,  Tyler  CR  (2010) Pharmaceuticals  in  the  aquatic environment: A critical review of the evidence for health effects in fish. Critical Reviews in Toxicology 40: 287–304.

 

 

 

 

57.  Alonso SG,  Catala  M,  Maroto  RR,  Gil JLR, de  Miguel AG, et  al. (2010) Pollution by psychoactive pharmaceuticals in the Rivers of Madrid metropolitan area (Spain). Environment International 36: 195–201.

58.  Gros M, Petrovic M, Ginebreda A, Barcelo D (2010) Removal of pharmaceu- ticals during  wastewater treatment  and  environmental  risk assessment using hazard indexes. Environment International 36: 15–26.

59.  Stackelberg PE, Gibs J, Furlong ET, Meyer MT, Zaugg SD, et al. (2007) Efficiency of conventional drinking-water-treatment  processes in  removal of pharmaceuticals and other organic compounds. Science of the Total Environment 377: 255–272.

60.  Bruce  GM,   Pleus  RC,   Snyder  SA  (2010) Toxicological  Relevance  of

Pharmaceuticals  in  Drinking  Water.  Environmental  Science & Technology

44: 5619–5626.

61.  Stackelberg PE, Furlong ET,  Meyer MT,  Zaugg SD, Henderson  AK, et al. (2004) Persistence of pharmaceutical compounds and other organic wastewater contaminants in a conventional drinking-watertreatment plant. Science of the Total Environment 329: 99–113.

62.  DasBanerjee T, Middleton FA, Berger DF, Lombardo JP, Sagvolden T, et al. (2008) A Comparison of Molecular Alterations in Environmental and Genetic

Rat Models of ADHD: A Pilot Study. American Journal of Medical Genetics

Part B-Neuropsychiatric Genetics 147B: 1554–1563.

63.  Traynor J, Agarwal P, Lazzeroni L, Francke U (2002) Gene expression patterns vary in clonal cell cultures from Rett  syndrome females with eight different MECP2 mutations. BMC Medical Genetics 3: 12.

64.  Nishimura Y, Martin  CL, Lopez AV, Spence SJ, Alvarez-Retuerto AI, et al. (2007)  Genome-wide expression profiling of  lymphoblastoid cell lines distin-

guishes  different   forms  of   autism  and  reveals  shared  pathways.  Human

Molecular Genetics 16: 1682–1698.

 

65.  Martins-De-Souza D, Dias-Neto E, Schmitt A, Falkai P, Gormanns  P, et al. (2010) Proteome  analysis of schizophrenia brain  tissue. World Journal of Biological Psychiatry 11: 110–120.

66.  Blalock EM, Geddes JW, Chen KC, Porter NM, Markesbery WR, et al. (2004) Incipient  Alzheimer’s disease: Microarray  correlation  analyses reveal major transcriptional and  tumor  suppressor responses. Proceedings of the National Academy of Sciences of the United States of America 101: 2173–2178.

67. Iwamoto K, Kakiuchi C, Bundo M, Ikeda K, Kato T (2004) Molecular characterization of bipolar disorder by comparing gene expression profiles of postmortem brains of major mental disorders. Molecular Psychiatry 9: 406–416.

68.  Matigian N, Windus L, Smith H, Filippich C, Pantelis C, et al. (2007) Expression

profiling  in  monozygotic  twins  discordant  for  bipolar  disorder  reveals dysregulation of the WNT signalling pathway. Molecular Psychiatry 12: 815–

825.

69.  Bomprezzi R,  Ringner  M, Kim  S, Bittner ML, Khan  J, et al. (2003) Gene expression profile in multiple sclerosis patients and healthy controls: identifying pathways relevant to disease. Human Molecular Genetics 12: 2191–2199.

70.  Gilli F, Lindberg RLP,  Valentino P, Marnetto  F, Malucchi S, et al. (2010)

Learning from Nature:  Pregnancy Changes the Expression of  Inflammation- Related Genes in Patients with Multiple Sclerosis. Plos One 5.

71.  Chakrabarti B, Dudbridge F, Kent L, Wheelwright S, Hill-Cawthorne G, et al. (2009) Genes Related to Sex Steroids, Neural Growth, and Social-Emotional

Behavior are Associated with Autistic Traits, Empathy, and Asperger Syndrome.

Autism Research 2: 157–177.

72.  Hu VW, Nguyen A, Kim KS, Steinberg ME, Sarachana T, et al. (2009) Gene

Expression Profiling of Lymphoblasts from Autistic and Nonaffected Sib Pairs: Altered Pathways in Neuronal Development and Steroid Biosynthesis. Plos One

4: 13.