Ept7, a quantitative trait locus that controls estrogen-induced pituitary lactotroph hyperplasia in rat, is orthologous to a locus in humans that has been associated with numerous cancer types and common diseases

Pituitary adenoma is a common intracranial neoplasm that is observed in approximately 10% of unselected individuals at autopsy. Prolactin-producing adenomas, i.e., prolactinomas, comprise approximately 50% of all pituitary adenomas and represent the most common class of pituitary tumor. Multiple observations suggest that estrogens may contribute to development of prolactinoma; however, direct evidence for a causal role of estrogens in prolactinoma etiology is lacking. Rat models of estrogen-induced prolactinoma have been utilized extensively to identify the factors, pathways and processes that are involved in pituitary tumor development. The objective of this study was to localize to high resolution Ept7 (Estrogen-induced pituitary tumor), a quantitative trait locus (QTL) that controls lactotroph responsiveness to estrogens and was mapped to rat chromosome 7 (RNO7) in an intercross between BN and ACI rats. Data presented and discussed herein localize the Ept7 causal variant(s) to a 1.91 Mb interval of RNO7 that contains two protein coding genes, A1bg and Myc, and Pvt1, which yields multiple non-protein coding transcripts of unknown function. The Ept7 orthologous region in humans is located at 8q24.21 and has been linked in genome wide association studies to risk of 8 distinct epithelial cancers, including breast, ovarian, and endometrial cancers; 3 distinct types of B cell lymphoma; multiple inflammatory and autoimmune diseases; and orofacial cleft defects. In addition, the Ept7 locus in humans has been associated with variation in normal hematologic and development phenotypes, including height. Functional characterization of Ept7 should ultimately enhance our understanding of the genetic etiology of prolactinoma and these other diseases.

Introduction associated with risk of numerous cancer types and other common diseases. These data may ultimately enhance our understanding of the genetic etiology of prolactinoma in humans.

Care, treatment, and phenotypic characterization of animals
All procedures involving live animals were approved by the Institutional Animal Care and Use Committee of the University of Wisconsin-Madison (protocol #M005111). ACI/SegHsd (ACI) and BN/SsNHsd (BN) rats were obtained from Envigo (formerly Harlan Sprague Dawley, Inc.). Each of the unique congenic rat strains described herein was generated in our laboratory as described below. All rats were housed under controlled temperature, humidity, and 12-hr light/12-hr dark conditions in facilities that were accredited by the American Association for Accreditation of Laboratory Animal Care and operated in accordance with The Guide for the Care and Use of Laboratory Animals [37].
Female rats from each rat strain were treated with 17β-estradiol (E2) beginning at 9 weeks of age as described previously [38][39][40][41][42]. Control rats from each strain received empty implants. Implants, empty or containing 27.5 mg of E2 (Sigma-Aldrich, St. Louis, MO), were prepared from Silastic tubing (Dow Corning, Midland, MI) and sealed with Silastic Medical Adhesive (Silicone Type A; Dow Corning). Implants were placed surgically into the interscapular region of 9 week old rats. During the procedure, rats were anesthetized with isoflurane administered at a rate of 1-5% via inhalation using an anesthetic vaporizer. Because the studies described herein were performed over a five year interval, fifteen groups of ACI females, totaling 77 animals, were evaluated starting at different points across this interval to ensure that adequate numbers of ACI rats were treated contemporaneously with rats from each of the different congenic strains. The pituitary weight phenotype in the ACI rats was stable over the five year interval (S1 Fig). Therefore, data from the different groups of ACI rats were pooled for statistical comparison to rats from the different congenic strains. Research staff monitored the rats a minimum of three times per week and weighed the rats weekly. The rats were generally euthanized following 196 ± 5 days of treatment. However, an animal was euthanized prior to the intended experiment endpoint if it exhibited cachexia (loss of 15% body mass); poor body condition; loss of normal neuromuscular coordination; abdominal swelling; or a mammary tumor that grew in excess of 2 cm in longest dimension, became ulcerated or impeded normal ambulation. The pituitary gland was collected at necropsy, weighed, and photographed. Pituitary weight has been shown to correlate with pituitary DNA content and the level of circulating prolactin, and thereby serves as a surrogate indicator of absolute lactotroph number [20,29,35,36,38].

Generation of congenic rat strains
Each of the congenic strains was developed using a marker assisted selective breeding protocol as previously described [36,39]. When a male was obtained that was heterozygous for BN alleles across a desired segment of RNO7 and homozygous for ACI alleles at all background markers, that male was backcrossed to ACI females and sibling progeny carrying the same recombinant chromosome were intercrossed to produce rats homozygous for BN alleles across the locus of interest. The different congenic strains are referred to by their abbreviated names: Ept7.1 through Ept 7.18. The full strain names and Rat Genome Database (RGD) identification numbers are listed in S1 Table. Genome coordinates for the BN intervals on RNO7 carried by each congenic strain are listed in S1 Table.

Statistical analyses of data
Phenotypes exhibited by Ept7 congenic rats were compared to those of treated ACI rats. Pituitary weights and treatment duration were expressed as the mean ± standard error of the mean (SEM). Differences in pituitary weight relative to ACI were evaluated using the Wilcoxon Rank Sum test. The impact on survival of morbidities related to pituitary hyperplasia/adenoma was evaluated using the Kaplan-Meier estimator and log rank test. Animals that were removed from the study due to mammary tumor burden or any mammary cancer associated morbidity were censored in these analyses. The thresholds for statistical significance were adjusted for multiple comparisons using the Bonferroni method. MSTAT Version 6.1 was used to perform all statistical analyses [43].

Sources of microarray data
Gene expression data for Ept7 candidates were generated using Affymetrix Rat Genome 230 version 2.0 GeneChips as described previously [40]. The biological samples from sham treated and diethylstilbestrol (DES) treated male ACI and BN rats were part of a larger study focused on genetic control of estrogen action in the anterior pituitary gland and the resulting data were deposited in the GEO Database under accession number GSE4028.

Sources of genomic data
Rat and human genome coordinates in this manuscript refer to genome assemblies Rnor_6.0 and GRCh38.p10, respectively. All genetic markers, marker locations, and rat strains are registered in the Rat Genome Database [44,45]. Genomic features of the Ept7 locus are from Ensembl Rnor_6.0 release 91 [46].

Phenotypic characterization of 19 unique ACI.BN-Ept7 congenic rat strains
Ept7 was previously mapped to a 32.4 Mb interval on RNO7 defined by microsatellite markers D7Wox3 (71.45 Mb) and D7Rat17 (103.85 Mb) [34,36]. Emca4 (Estrogen-induced mammary cancer), a QTL that influences development of E2-induced mammary cancer, was mapped to this same region of RNO7 in a BN x ACI intercross [42]. In order to further localize the Ept7 and Emca4 causal variants we generated and characterized multiple congenic rat strains, each of which is homozygous for BN alleles across a unique segment of RNO7 introgressed onto the ACI genetic background (S1 Table). The data presented here are focused on the mapping of Ept7. Data relating to the mapping of Emca4 will be published elsewhere and are discussed only when relevant to Ept7.
Pituitary weight, a surrogate indicator of absolute lactotroph number, was measured as the primary phenotypic indicator of lactotroph responsiveness to E2 [20,29,35,36,38]. Pituitary weight for sham treated control rats across all 21 rat strains examined in this study averaged 9.8 mg (standard error of the mean = 0.2; n = 105). No significant differences were observed when basal pituitary weight for BN rats or rats from each of the 19 unique congenic strains was compared to basal pituitary weight for ACI rats, which served as the background (recipient) strain during the generation of the congenic rat strains characterized in this study (Table 1).
Continuous treatment with E2 induced pituitary lactotroph hyperplasia, as evidenced by significant increases in pituitary weight, in females of each of the 21 rat strains examined in this study. For reference, pituitary weight in the ACI strain was increased 18.1-fold, from 10.5 ± 0.7 mg in sham treated (control) female rats to 189.9 ± 8.0 mg in rats treated with E2 (Table 1). By contrast, female rats of the BN strain, which served as the donor strain during generation of the congenic rat strains described in this manuscript, exhibited a 1.9-fold increase in pituitary weight in response to prolonged treatment with E2, from 8.5 ± 0.8 mg in sham treated controls to 16.2 ± 0.7 mg. Evaluation of pituitary weight in E2 treated females from each of the 19 congenic rat strains revealed two distinct groups of strains that differed in responsiveness to E2. Rats from the congenic strains in Group 1 were highly responsive to E2 and exhibited pituitary weights that did not differ from those exhibited by ACI rats (Table 1). Group 1 included the Ept7.1, Ept7.4, Ept7.5, Ept7.7, Ept7.8, Ept7.9, Ept7.12, Ept7.13, and Ept7.15 congenic strains. Rats from the congenic stains in Group 2 exhibited pituitary weights that were significantly less than that exhibited by ACI rats, indicating that the responsiveness of the pituitary lactotroph in these rat strains is reduced relative to that of ACI rats. Group 2 consisted of the Ept7, Ept7.2, Ept7.3, Ept7.6, Ept7.10, Ept7.16, Ept7.17, and Ept7.18 congenic rat strains. The Ept7.11 and Ept7.14 congenic strains were not assigned to either Group 1 or Group 2 for reasons discussed below. Female rats were treated with E2, released from subcutaneous Silastic implants, beginning at 9 weeks of age as described in Materials and Methods. Animals were euthanized upon observation of any treatment associated morbidity or following 28 weeks of treatment. Pituitary weight measured at necropsy is indicated on the y-axis; duration of treatment is indicated on the x-axis. Each data point represents means ± SEMs; N = 9-70 for each strain. Red symbols indicate mean pituitary weights that are significantly less than exhibited by ACI rats (P < 0.0025 by Wilcoxon rank sum test). The dashed lines demarcate those congenic strains that were assigned to Groups 1 (pituitary weights not different from ACI) and Groups 2 (pituitary weights less than ACI). Average pituitary weight in Ept7.11 and Ept7.14 rats (blue symbols) did not differ from that observed in ACI rats; however, the Ept7.11 and Ept7.14 rats were treated with E2 for a longer duration (approximately 9 days) than ACI rats. https://doi.org/10.1371/journal.pone.0204727.g001 Rats were removed from the study upon evidence of morbidity resulting from pituitary hyperplasia/adenoma, mammary cancer, or unknown etiology. Therefore, we plotted pituitary weight for each strain as a function of the duration of E2 treatment (Fig 1). We also evaluated survival of E2 treated rats from each congenic strain as a function of morbidities that are known to result from lactotroph hyperplasia/adenoma and gross pituitary enlargement ( Table 1). The resulting data indicate that rats from the Group 2 congenic strains, with the exception of rats from the Ept7.16 strain, were generally treated with E2 for a longer duration than ACI rats. Thus, the inhibitory actions of BN alleles at Ept7 in the Group 2 congenic strains on E2-induced pituitary growth were likely to have been underestimated due to the longer durations of treatment (discussed further below).

Ept7 causal variant resides within a 1.91 Mb interval on RNO7
Examination of the BN genome intervals harbored by each of the congenic rat strains localized the Ept7 causal variant(s) within a 1.91 Mb segment on RNO7 (Fig 2). Most informative in this regard is the observation that each of the Group 2 congenic strains harbored BN alleles across the 1.91 Mb Ept7 locus and rats from each of the Group 2 strains exhibited a diminished pituitary growth response to E2 when compared to ACI rats. By contrast, none of the Group 1 congenic strains harbored BN alleles across the 1.91 Mb Ept7 locus and rats from each Group 1 strain exhibited a pituitary growth response that did not differ significantly from that exhibited by ACI rats. The proximal boundary of this refined Ept7 locus is defined by microsatellite marker D7Uwm37 (101.94 Mb on RNO7), at which ACI alleles were observed in the Ept7.18 congenic strain. This proximal boundary is further supported by marker D7Uwm43 (101.93 . Each black or red filled horizontal box indicates the segment of RNO7 from the donor BN rat strain that was introgressed onto the genetic background of the recipient ACI strain to generate the indicated congenic rat strains. The white segments flanking a black or red box indicate regions of recombination (i.e., unknown genotype); all other segments of RNO7 were derived from the recipient ACI strain. Black boxes indicate congenic intervals in strains that did not exhibit reduced pituitary weight when compared to E2 treated ACI rats, whereas red boxes indicate congenic intervals in strains that exhibited reduced pituitary weight (P < 0.0025 by Wilcoxon rank sum test Mb on RNO7) which marks the distal end of the BN interval in congenic strains Ept7.9 and Ept7.12 (Fig 2). The distal boundary of the Ept7 locus is defined by microsatellite marker D7Rat17, at which ACI alleles were observed in the Ept7.18 congenic strain as well in several of the other Group 2 congenic strains.
The Ept7.11 and Ept7.14 congenic strains appear to represent exceptions to the observed linkage between BN alleles across the 1.91 Mb Ept7 locus and reduced pituitary weight. Although Ept7.11 and Ept7.14 rats harbor BN alleles at Ept7 and exhibited lower average pituitary weights, relative to ACI rats, the observed differences between these congenic rats and ACI rats did not achieve the stringent threshold of statistical significance (P < 0.0025) used to compensate for multiple comparisons (Table 1). However, it is important to note that the Ept7.11 and Ept7.14 rats were treated with E2 for an average of 194.4 and 194.6 days, respectively, which is 9 days longer than the average length of treatment for ACI rats. Because pituitary weight in E2 treated ACI and ACI-derived congenic rats increases as a function of the duration of treatment, we suggest that the 9 day longer duration of E2 treatment partially eclipsed the impact of BN alleles at Ept7 on induction of pituitary lactotroph hyperplasia and associated pituitary growth during evaluation of the Ept7.11 and Ept7.14 congenic strains. Comparison of congenic rats to ACI rats based solely on genotype at Ept7 revealed that average pituitary weight in congenic rats harboring BN alleles at Ept7 was reduced by 57.6 mg (i.e., 30.3%; P = 9.5E-11), relative to ACI rats ( Fig 3A). Moreover, the congenic rats harboring BN alleles at Ept7 were treated with E2 for an average of 5.5 days longer than ACI rats (P = 3.47E-10) (Fig 3B), in large part because of a lower incidence of mortality associated with pituitary enlargement (P = 1.12E-11) (Fig 3C). By contrast, congenic rats harboring ACI alleles at Ept7 did not differ from ACI rats with respect to pituitary weight (P = 0.99), duration of E2 treatment (P = 0.92), or survival (P = 0.59).

Genomic features of the Ept7 locus
The 1.91 Mb Ept7 interval harbors two protein coding genes, A1bg and Myc (Fig 4). A1bg encodes alpha-1-B glycoprotein. Searches of the PubMed and GEO databases did not reveal any associations between A1bg and estrogen action in the pituitary gland or any other cell or tissue type. Evaluation of gene expression data from our lab (GEO accession number GDS2913) indicates that A1bg (probe id = 1369509_a_at) is expressed at a low level in the anterior pituitary gland with no significant differences being apparent between male ACI and BN rats, either sham treated or treated for 12 weeks with the synthetic estrogen diethylstilbestrol (Fig 5). Myc encodes Myc proto-oncogene, bHLH transcription factor, which has been functionally linked to development of multiple cancer types. Expression of Myc in the anterior pituitary gland is known to be enhanced by estrogens, and MYC is highly expressed in prolactinoma relative to normal pituitary [47][48][49]. Twelve weeks of treatment with diethylstilbestrol increased Myc mRNA expression (probe id = 1368308_at) in the anterior pituitary glands of male ACI and BN rats 5-fold (P = 3.8E-4) and 7-fold (P = 6.8E-3), respectively, and no significant differences were observed between the two rat strains. (Fig 5).
The Ept7 locus also harbors several genes that drive expression of non-protein coding RNAs. Most noteworthy is Pvt1, which generates a diverse assortment of alternatively spliced RNAs as well as multiple microRNAs: miR1204, miR1205, miR1206, miR1207-5p, and miR1207-3p. Microarray data indicate that 12 weeks of treatment with DES increased Pvt1 expression (probe id = 1384449_at) in the anterior pituitary glands of ACI and BN rats 3.6-fold (P = 9.0E-5) and 2.8-fold (P = 0.017), respectively (Fig 5). Ept7 also harbors miR1208, which is generated from an as yet unidentified primary transcript. Although the miRNAs residing within Ept7 have not been annotated onto the rat genome, their existence in the rat was predicted and BLAST analyses indicate their presence in the rat in the same relative locations and orientations as in the human and mouse genomes [50]. Finally, it is noted that D7Rat17, the marker that defines the distal boundary of Ept7, is annotated to two distinct locations on RNO7 separated by 58,383 nucleotides. The duplicated placement of this marker on RNO7 does not impact the results or interpretation of this study.

Discussion
Ept7 is one of 19 QTL that have been linked to estrogen-induced increases in pituitary weight, a well-defined surrogate phenotype for lactotroph hyperplasia/adenoma [28, 30, 33, 34, 51]. Data presented in this manuscript refine the location of Ept7 to a 1.91 Mb interval on RNO7. Ept7 is the first QTL linked to estrogen action in any cell type, tissue or organ system to be localized to a resolution of less than 2 Mb. The data presented strongly suggest that a genetic variant(s) that resides within Ept7 impacts expression or function of one or more linked genes and thereby influences estrogen action on the pituitary lactotroph. The actions of the Ept7 variant(s) could occur directly within the lactotroph itself (i.e., cell autonomous actions). This assertion is supported by in vivo and in vitro studies that demonstrate that estrogens act directly on the lactotroph to regulate proliferation and prolactin gene expression [25, [52][53][54]. Expression of at least two of the genes that reside within the Ept7 locus, Myc and Pvt1, is increased by administered estrogen in vivo. Therefore, we consider Myc and Pvt1 to be strong Ept7 candidate genes. Alternatively, Ept7 could act within another estrogen responsive pituitary or extrapituitary cell type that in turn influences lactotroph proliferation; e.g., the tuberoinfundibular neurons of the hypothalamus, which produce and release dopamine, a major negative regulator of lactotroph proliferation and prolactin gene transcription [55][56][57].
Lactotroph homeostasis is controlled by multiple endocrine factors in addition to estrogens. Prolactin, the primary protein product of the lactotroph, exhibits a wide variety of biological duration of E2 treatment (days ± SEM) (B); and survival over the course of E2 treatment (C) are illustrated for ACI rats (N = 70 (A), 70 (B) and 77 (C)), congenic rats that are homozygous for ACI alleles at Ept7 (N = 212 (A), 212 (B) and 234 (C)), and congenic rats that are homozygous for BN alleles at Ept7 (N = 239 (A), 239 (B) and 257 (C)). Animals removed from the study due to mammary cancer related phenotypes were censored in order to clearly illustrate the impact of pituitary tumor related morbidity on survival (ACI, 4 animals censored due to mammary tumor burden; AA congenics, 22 animals censored due to tumor burden; BB congenics, 30 animals censored due to tumor burden).
https://doi.org/10.1371/journal.pone.0204727.g003 functions, including regulation of mammary gland development and lactation, regulation of maternal behavior, and regulation of the immune system [58]. We postulate that the genetic variants that reside within Ept7 and the other QTL that control responsiveness of the lactotroph to estrogen exist within the Rattus norvegicus population to ensure that a subset of the rat population is able to maintain lactotroph homeostasis under geographical and/or temporal variations in environmental conditions. The pathogenic potential of these genetic variants, as revealed in inbred laboratory rat strains under conditions of prolonged stimulation of estrogen receptor dependent pathways, is informative regarding the molecular mechanisms of estrogen action on the pituitary lactotroph. Because Homo sapiens and Rattus norvegicus have co-existed and shared environmental challenges over many thousands of years, it is possible that population level variation within orthologous genomic elements in humans and rats may enhance genetic fitness of both species.
Rat models have been employed to map genetic variants that control estrogen action in multiple tissues, including the anterior pituitary [28, 30, 32-36, 59], mammary gland [39,40,42,60], uterus [61,62], thymus [63], and testis [64]. For the most part, these QTLs map to distinct regions of the rat genome, suggesting the underlying quantitative trait variants influence estrogen action in a cell type specific manner. Ept7 appears to be the exception in that it localizes to the same region of RNO7 as Emca4, a QTL that influences development of estrogeninduced mammary cancer [34,36,42]. Data presented localize the Ept7 quantitative trait variant(s) to the 1.91 Mb interval defined by markers D7Uwm37 and D7Rat17. By contrast, data to be published elsewhere indicate that Emca4 is a composite QTL that harbors multiple genetic variants that interact with one another to influence development of E2-induced mammary Male ACI and BN rats were treated with the synthetic estrogen diethylstilbestrol (DES), released from subcutaneous Silastic implants, beginning at 9 weeks of age as described previously [33,36]. Gene expression data were generated from individual rat anterior pituitary glands using Affymetrix GeneChip Rat Genome 230 2.0 Arrays as described previously [40] and the resulting data were deposited in the Gene Expression Omnibus (GEO) database (GEO accession number GDS2913). Data for A1bg (probe id = 1369509_a_at), Myc (probe id = 1368308_at), and Pvt1 expression (probe id = 1384449_at) were extracted from this dataset and were illustrated using CIMminer (Genomics and Pharmacology Facility, Developmental Therapeutics Branch, National Cancer Institute, National Institutes of Health). DES treatment significantly increased expression of Myc and Pvt1 in male ACI and BN rats. By contrast, expression of A1bg did not differ as a consequence of treatment. Also illustrated are expression data for Esr1, which encodes estrogen receptor alpha. cancers. Although one or more of the Emca4 quantitative trait variants resides within the same genomic interval as Ept7, the others reside proximal to Ept7 on RNO7. Based upon these observations, it is possible that Ept7 and Emca4 may exhibit some overlap in their biological functions.
The Ept7 locus harbors two annotated protein coding genes, A1bg and Myc, in addition to genes that generate non-protein coding transcripts, most notably Pvt1. Zero published studies were retrieved when the PubMed database was queried using the search terms 'A1bg and pituitary' or 'A1bg and estrogen'. Moreover, although A1bg resides proximal to Myc in the rat and mouse genomes, in the human genome A1BG is not linked to MYC on chromosome 8q24 but instead resides on chromosome 19. Consequently, we do not consider A1bg to be a high priority Ept7 candidate.
By contrast, MYC is a well characterized protooncogene that is aberrantly overexpressed in many cancer types [65]. Estrogens enhance Myc expression in the rat anterior pituitary gland as well as in GH3 rat pituitary tumor cells [49,66]. Moreover, MYC is overexpressed in prolactinomas, relative to the normal anterior pituitary gland [49]. Together, these data suggest that MYC may contribute to prolactinoma development in humans and estrogen treated rats. Pvt1 is representative of a type 3 supergene because it generates a variety of transcripts through use of multiple promoters and alternative transcript splicing/processing to yield multiple functional RNA products [67]. Emerging data suggest the RNA products generated from Pvt1 coordinately regulate large networks of genes [68]. Multiple studies reveal functional interactions between MYC and PVT1. For example, MYC and PVT1 are almost always co-amplified in cancers that exhibit somatic copy number gains at 8q24, and mouse models have been used to demonstrate that a single copy gain of a chromosome segment harboring Myc and Pvt1 enhances mammary tumorigenesis driven by an MMTV-NEU transgene, whereas single copy gains of Myc or Pvt1 alone do not [69]. In addition, MYC has been demonstrated to activate transcription from the PVT1 promoter in cultured cells [70]. More recently, the MYC and PVT1 promoters have been shown to compete for engagement with a set of enhancers located within PVT1 [71]. We are working to define the functions of Pvt1 and the different miRNAs residing within the Ept7 locus in the rat mammary and pituitary glands.
Data presented herein indicate the Ept7 locus harbors a variant(s) that influences the actions of estrogens in the regulation of pituitary lactotroph proliferation as reflected in development of E2-induced lactotroph hyperplasia/adenoma. Data to be published elsewhere indicate that a variant(s) within the same genome interval are part of a multipartite genetic element that controls estrogen action in the rat mammary gland and influences susceptibility to E2-induced mammary cancer. The region of the human genome that is orthologous to Ept7 has been linked in genome wide association studies to multiple cancer types, developmental defects, and physiological phenotypes (S2 Table). For example, SNPs residing within the Ept7 orthologous region have been associated with breast cancer [72][73][74], endometrial cancer [75], and epithelial and high grade serous ovarian cancers [76]. Estrogens have been implicated in the development of each of these cancer types. Therefore, molecular characterization of the genes and variants that reside in Ept7 may yield novel insights into the etiology of multiple diseases.
Supporting information S1 Fig. Pituitary weights in E2 treated ACI females by date of necropsy. Fifteen groups of ACI females were evaluated starting at different points over a five-year period contemporaneous to their congenic counterparts. Female rats were treated with E2, released from subcutaneous Silastic implants, beginning at 9 weeks of age as described in Materials and Methods.
Animals were euthanized upon observation of any treatment associated morbidity or following 28 weeks of treatment. Pituitary weight measured at necropsy is indicated on the y-axis; date of necropsy is indicated on the x-axis (month-year). Each data point represents one individual. The dashed line indicates the mean for the population (189.9 mg); the shaded region demarcates one standard deviation from the mean (± 67.0 mg). The dotted line indicates best fit by linear regression in Microsoft Excel 2016 (y = -0.0157x + 832.19); N = 70. (TIF) S1