Leveraging Complementary in Vivo and in Vitro Gene Expression Measurements to Elucidate Uniquely Human Metabolic Processes

Leveraging Complementary In Vivo and In Vitro Gene Expression Measurements to Elucidate Uniquely Human Metabolic Processes

by Lisa Warner Pfefferle

University Program in Genetics and Genomics Duke University

Date:______Approved:

______Gregory A. Wray, Supervisor

______Robert Cook-Deegan

______Gregory E. Crawford

______Erich D. Jarvis

______Mohamed A. F. Noor

Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the University Program in Genetics and Genomics in the Graduate School of Duke University 2012

ABSTRACT

Leveraging Complementary In Vivo and In Vitro Gene Expression Measurements to Elucidate Uniquely Human Metabolic Processes

by Lisa Warner Pfefferle

University Program in Genetics and Genomics Duke University

Date:______Approved:

______Gregory A. Wray, Supervisor

______Robert Cook-Deegan

______Gregory E. Crawford

______Erich D. Jarvis

______Mohamed A. F. Noor

Abstract

The origin of man has motivated researchers to investigate differences between humans

and our non-human relatives. The striking phenotypic differences that distinguish

humans from chimpanzees are likely controlled by a relatively modest number of

genetic changes present between these species. As energy acquisition and processing

effect multiple organ systems, the dramatic changes in the human diet are thought to

underpin many of these unique phenotypes. The evolution of the human diet is

marked by omnivory with increased consumption of animal products, cereal grain and

vegetable oil associated with the Paleolithic era, domestication of plants and the

industrial revolution respectively. Multiple tailored expression techniques in tissues of

dietary relevance were used to uncover combinations of neurosensory, physiological

and morphological changes between humans and non-human primates. Taken

together, the combined power of in vitro and in vivo approaches elucidates several

genetic mechanisms important in uniquely human bioenergetic processes.

Contents

Abstract ...... iv

List of Tables ...... ix

List of Figures ...... x

1. Introduction ...... 1

1.1 Overview ...... 2

1.2 The importance of dietary changes during human evolution ...... 3

1.2.1 Shifts in the diets of modern human populations as compared with ape diets . 3

1.2.2 Dietary shifts in hominin evolution ...... 4

1.2.3 Trade-off hypotheses regarding diet ...... 10

1.3 Why study expression in primates? ...... 12

1.4 Summary of chapters ...... 15

2. Comparative expression analysis of the phosphocreatine circuit in extant primates: implications for human brain evolution ...... 22

2.1 Abstract ...... 23

2.2 Introduction ...... 24

2.3 Materials and Methods ...... 29

2.3.1 Primate sample collection ...... 29

2.3.2 Tissue preparation, total RNA isolation, and cDNA synthesis ...... 29

2.3.3 Primer design ...... 31

2.3.4 Quantitative RT-PCR ...... 36

2.3.5 Gene expression comparison ...... 38

2.3.6 Positive selection analysis ...... 39

2.3.6.1 Sequence data and annotation ...... 39

2.3.6.2 Tests for positive selection ...... 40

2.3.6.3 Determining a neutral proxy ...... 41

2.4 Results ...... 46

2.4.1 Comparative expression analysis ...... 46

2.4.2 Creatine transporter, SLC6A8 ...... 46

2.4.3 Mitochondrial creatine kinases, CKMT1 and CKMT2 ...... 47

2.4.4 Cytosolic creatine kinases, CKB and CKM ...... 48

2.4.5 Positive selection ...... 50

2.5 Discussion ...... 51

2.6 Conclusions ...... 59

2.7 Acknowledgements ...... 60

3. Functional consequences of genetic variation in primates on tyrosine hydroxylase (TH) expression in vitro ...... 61

3.1 Abstract ...... 62

3.2 Introduction ...... 63

3.3 Results ...... 66

3.3.1 Sequence analysis between non-human primates and humans ...... 66

3.3.2 Expression of primate TH in multiple neuronal cell lines ...... 69

3.3.3 Examining the functional impact of TH promoter SNPs in IMR-32 cells ...... 74

3.4 Discussion ...... 80

3.5 Experimental Procedure ...... 85

3.5.1 Cloning and sequencing ...... 85

3.5.2 Positive selection analysis ...... 85

3.5.3 TH RT-PCR ...... 86

3.5.4 Cell culture ...... 87

3.5.5 Site directed mutagenesis ...... 87

3.5.6 Transfection and expression measurements ...... 89

3.5.7 In vitro expression analysis ...... 90

3.6 Acknowledgements ...... 92

4. Coordinated changes in adipose transcriptomes accompanied dietary shifts during human origins ...... 93

4.1 Abstract ...... 94

4.2 Body of Paper ...... 96

4.3 Materials and Methods ...... 128

4.3.1 Samples ...... 128

4.3.2 Cell culture and tissue processing ...... 128

4.3.3 Next generation sequencing and processing ...... 129

4.4.4 Enrichments and analyses ...... 131

4.3.5 Staining and imaging ...... 132

vii

4.4 Acknowledgements ...... 134

5. Insights from a chimpanzee adipose stromal cell population: opportunities for adult stem cells to expand primate functional genomics ...... 135

5.1 Abstract ...... 136

5.2 Introduction ...... 137

5.3 Morphological and genetic characterization of chimpanzee ASCs ...... 140

5.4 Chimpanzee ASC pluripotency and differentiation status ...... 148

5.5 Transcriptomic differences between chimpanzee and human ASCs ...... 153

5.6 Stem cells can greatly expand the number of in vitro models for comparative primate genomics ...... 158

5.7 Materials and Methods ...... 162

5.7.1 Culturing and differentiating stromal cells ...... 162

5.7.2 Stromal cell transcriptomics ...... 162

5.7.3 Staining and imaging ...... 164

5.8 Acknowledgements ...... 166

6. Summary ...... 167

Biography ...... 191

viii

List of Tables

Table 2.1 Primate samples used in this study ...... 30

Table 2.2 Primer sequences for quantitative RT-PCR and verifications ...... 34

Table 2.3 p-values for human-specific expression ...... 45

Table 3.1 Mean and standard deviation of TH expression in cell lines ...... 72

Table 4.1 Differentiated adipocytes and white adipose tissue samples ...... 99

Table 4.2 Illumina’s HiSeq statistics for the differentiated adipocytes ...... 100

Table 4.3 ABI’s SOLiD 4 statistics for the white adipose tissue ...... 115

Table 5.1 Illumina’s HiSeq statistics for the ASC profiled in this study ...... 144

List of Figures

Figure 1.1 A timeline showing some of the temporal intersections of diet, natural selection and one of the many changes in human morphology ...... 8

Figure 2.1 Schematic representation of the phosphocreatine circuit ...... 27

Figure 2.2 Verifying cDNA purity using SDHA control primers that amplify an intronic region ...... 32

Figure 2.3 Primer validation for phosphocreatine circuit and control genes ...... 35

Figure 2.4 Phosphocreatine circuit gene expression comparisons among species ...... 43

Figure 3.1 Alignment of six natural promoter region TH haplotypes ...... 68

Figure 3.2 Normalized expression levels driven by TH cis-regulatory haplotypes in SH- SY5Y, SK-N-BE(2), and IMR-32 cell lines ...... 70

Figure 3.3 Normalized expression levels driven by TH cis-regulatory haplotypes in SH- SY5Y, SK-N-BE(2), and IMR-32 cell lines ...... 76

Figure 3.4 The human cis-regulatory variants of TH found in this study ...... 77

Figure 3.5 Functional consequences of SNP-1898 and SNP-801 ...... 79

Figure 4.1 Phenotypic and transcriptomic profiles of human and chimpanzee differentiated adipocytes ...... 98

Figure 4.2 Schematic of an adipocyte and key reactions in lipid metabolism ...... 102

Figure 4.3 Multidimensional scaling plot that distinguishes human and chimpanzee adipocytes in different challenged conditions ...... 106

Figure 4.4 Venn-diagram of human differentiated adipocytes challenged with linoleic and oleic acid...... 107

Figure 4.5 Phenotypic differences between human and chimpanzee differentiated adipocytes across several fatty acid conditions ...... 110 x

Figure 4.6 Distribution of the approximate volume of individual lipid droplets in each condition across species ...... 111

Figure 4.7 Multidimensional scaling plot that distinguishes human, chimpanzee, and macaque white adipose tissue ...... 116

Figure 4.8 Comparing differentiated adipocyte and white adipose tissue transcriptomes ...... 117

Figure 4.9 The biosynthesis of unsaturated fatty acids pathway is important during human evolution ...... 119

Figure 4.10 COX2 and it’s interactors are higher in human white adipose tissue compared to the chimpanzee ...... 123

Figure 4.11 Signature of inflammation in human white adipose tissue ...... 125

Figure 5.1 10x images of chimpanzee ASCs in culture ...... 141

Figure 5.2 20x images of a chimpanzee and a representative human ASCs in culture . 142

Figure 5.3 Transcriptomic insights into chimpanzee ASCs ...... 145

Figure 5.4 Pluripotency insights ...... 149

Figure 5.5 Visualizing the normalized ASC transcriptomes ...... 154

Figure 5.6 PANTHER gene categories enriched for differential expression by species. 155

Acknowledgements

My graduate advisor: Greg Wray, for your support and enthusiasm throughout my graduate journey.

My undergraduate advisors: April Hill, Malcolm Hill, Laura Runyen-Janecky, and William Myers, for helping me progress as a scientist and young adult.

My committee members: Greg Crawford, Bob Cook-Deegan, Erich Jarvis, and Mohamed Noor, for your scientific direction and professional advice.

Current and past Wray Lab members: for sharpening my critical thinking skills.

Shannon Looney: for your encouragement and patience. You are an exceptional friend, a sister by heart.

Austin Warner: a better brother than I deserve -- with you, I will always be home.

Mom and Dad: for your quiet sacrifices, unwavering support and unconditional love. Words cannot express my gratitude.

My husband, Adam Pfefferle, who humbly lives Ephesians 5:25-33.

xii

1. Introduction

Excerpts originally published in:

Babbitt CC, Warner LR, Fedrigo O, Wall CE, Wray GA. 2011. Genomic signatures of diet- related shifts during human origins. Proceedings of the Royal Society B 278:961-969.

Additional excerpts taken from:

Lisa R. Warner Preliminary Exam – November 12, 2009

My contribution to this work:

1) Researching “The importance of dietary changes during human evolution” section.

2) Co-wrote “The importance of dietary changes during human evolution” section with Dr. Christine Wall.

3) Editing and providing comments for entire manuscript.

1.1 Overview

Humans and chimpanzees last shared a common ancestor ~ 5.5 mya (Chen and

Li 2001) with the first widely accepted member of the genus Homo arising in Africa ~1.9

mya (Wood and Collard 1999). There are many striking adaptations associated with the

appearance of humans including an increase in brain size, upright posture, extensive

tool use, and altered social behavior (Olson and Varki 2003, Varki and Altheide 2005).

Several hypotheses have emerged stating that the appearance of human specific traits

paralleled changes in energy acquisition and processing (Aiello and Wheeler 1995,

Horrobin 1999, Broadhurst et al 1998, Leonard et al 2003, Ungar et al 2006). One

possibility is some of these changes are due to different access and utilization of specific

nutrients, as the human diet is distinctly different from chimpanzees. These dietary

specializations likely have unique interactions with genetic material, both shaping and

being shaped by the evolving genome. Interestingly, King and Wilson (1975) posited

that the genetic differences present between humans and our closest living relative the

chimpanzee are likely to be found within regulatory regions. By measuring expression,

the functional output of gene regulation, we are in a position to gain molecular insights

into the origin of our species.

1.2 The importance of dietary changes during human evolution

1.2.1 Shifts in the diets of modern human populations as compared with ape diets

There is an enormous amount of geographical and temporal variation in the modern human diet, but the underlying strategy remains omnivory. The most dramatic change in the recent diet of humans is the domestication of animals and plants (Zeder

2006). Nutrient intake varies widely among modern human populations, where some have a heavy reliance on carbohydrates in the form of cereals, roots and tubers from agriculture and gathering, while other populations have an emphasis on fat and protein extracted from animal husbandry, hunting and fishing (Murdock 1967). These diets distinguish human populations from members of the larger family of living Hominidae, which includes the gorilla, chimpanzee and bonobo. All of the African apes have a diet that includes large quantities of fruit and/or structural plant parts. This is not to say these animals are exclusively vegetarian, as we know chimpanzees, bonobos and gorillas sometimes eat invertebrates (McGrew 2001), and chimpanzees (Stanford et al 1994,

Boesch and Boesch-Achermann 2000) as well as bonobos hunt vertebrates (Surbeck and

Hohmann 2008, Hohmann and Fruth 2007).

1.2.2 Dietary shifts in hominin evolution

The connection between diet and the appearance of possibly adaptive traits in hominins is of great importance for understanding human evolutionary history as well as the health and disease consequences of these adaptations for modern humans.

Hominins, as a group, include humans and all of our ancestors arising after the human– chimpanzee divergence approximately 4.6–6.2 mya (Chen and Li 2001). Dietary traits act at several scales, from molecular to organismal and are associated with the intake and processing of food. In order to interpret the signatures of diet-related molecular changes, it is important to revisit the evolutionary history of humans at the organismal level. The phylogenetic affinities and the accompanying diets of many early hominin species remain unclear; however, ecological studies of extant primates and functional analyses of fossil remains suggest that hominins in general occupied an omnivore trophic niche (Peters 2007). The fossil remains of several australopithecine and paranthropine species show that diet varied between 4.5-1.2 mya, but overall these hominins had large molars lacking well-developed shearing crests, thick enamel and powerful jaws (Daegling and Grine 1991, Robinson 1954, Demes and Creel 1988, Grine and Martin 1988, Kay and Grine 1988). These dental traits indicate crushing of hard food items during mastication and a diet that included seeds, rich in protein and fat, but do not preclude a diet including underground storage organs, such as roots and tubers,

covered with abrasive soil and rich in carbohydrates (Peters and O’Brian 1981, Teaford and Ungar 2000, Jolly 1970). The genus Homo is first recognized in Africa approximately

2.5 mya. The evolution of the genus Homo between 2.5 and 1.9 mya is poorly understood due to difficulties in assigning fossil specimens to distinct taxa, however, there appears to be a trend of gradually increasing brain size during this period

(Schoenemann 2006, McHenry 1994, De Miguel and Henneberg 2001). Dental and skeletal traits of early Homo are difficult to interpret, though increased occlusal relief suggests an emphasis on shearing of food items during mastication (Ungar 2004). It is notable that both stable isotope and dental microwear studies suggest that it is difficult to demonstrate a highly specialized diet for early hominins (Ungar et al 2006, Ungar et al

2008, Van der Merwe et al 2008, Yeakel et al 2007, Sponnheimer and Lee-Thorp 1999).

It is with the origin of Homo erectus approximately 1.9 mya and the appearance of the Acheulean tool industry approximately 1.6 mya that we see several traits that signify a clear shift towards the modern human form and a change in diet as compared with their more robust predecessors. With H. erectus, there is an increase in body size, skeletal indicators of a striding bipedal gait, a reduction in the size of the teeth and jaws and a substantial jump in relative brain size, which together with the evidence from the archaeological record suggest a dietary strategy that included bulk processing of a significant proportion of high-quality, calorie-rich food items (Aiello and Wheeler 1995,

McHenry 1994, Shea 2007, Brown et al 1985, McHenry 1992) (Figure 1.1). There is evidence at this time for extraction of marrow and flesh from large mammals using stone tools (Bunn 2001, Bunn and Ezzo 1993, Blumenschine et al 2003), although recent evidence argues that this may have occurred much earlier in Australopithecus afarensis

(McPherron et al 2010). Early representatives of the genus Homo probably used tools for the processing of both animal and plant materials and for wooden tool production (Shea

2007, McPherron et al 2010); however, hunting weapons do not show up unequivocally in the fossil record until about 400 kya (Thieme 1997). While the archaeological record clearly shows that scavenging occurred at carnivore kill sites, there is little consensus as to what constitutes evidence for distinguishing passive scavenging, power scavenging, and hunting (Blumenschine and Pobiner 2007), where power scavenging is defined as actively driving another animal away from a carcase, as opposed to passive scavenging which involves harvesting food from an abandoned carcase (Bunn 2001). The evidence from the feeding apparatus, particularly the reduction in post-canine tooth size, the increased occlusal relief and the gracilization of the jaws, indicates continued emphasis on shearing food items during mastication with a reduction in the hardness of foods consumed and/or the use of technology to process foods prior to ingestion (Robinson

1954, Jolly 1970, Ungar 2004, Teaford et al 2002). Perhaps, more convincingly, fossil isotope studies demonstrate that hominid remains contain large quantities of the carbon

isotope C13 found in grasses (Sponheimer and Lee-Thorp 1999), but they show no capacity for digesting this material (Ungar et al 2006, Grine and Kay 1988), indicating that they were likely consuming animals that grazed on the grasses (Mann 2007). Finch and Stanford (2004) proposed that the Homo lineage would have experienced genetic adaptations, to counterbalance the pathologic effects of eating meat such as cholesterolemia, vascular disease and parasite loads. The consumption of animals was not limited to the land, in fact, evidence from the East African Rift Valley Lake indicates fish and shellfish, were consumed by our Homo ancestors (Broadhurst et al 1998). These lacustrine species are high in two essential poly-unsaturated fatty acids, arachidonic acid and docosahexaenoic acid, which are necessary for the growth and development of the human brain. It has been proposed that increased consumption of these omega-3 and omega-6 allowed for, responded to or caused the dramatic expansion of the cerebral cortex in the Homo lineage (Broadhurst et al 1998, Simopoulos 1999). The fossil and archaeological evidence suggests there was not only an increase in access to animal products during this period, but also the continued importance of plant material.

Figure 1.1 A timeline showing some of the temporal intersections of diet, natural selection and one of the many changes in human morphology

Figure 1.1) Teal bars indicate the temporal range on which different methods for scanning for selection are optimized to identify relevant changes in the genome. Blue bars indicate the times in which there is evidence for shifts in human dietary intake. The colored bubbles are a general schematic of the time and range in size of cranial capacity found in various hominin species adapted from Schoenemann 2006 with additional data from White et al 2009. 8

Taken together, the hallmark of the early Homo diet is its great versatility and consistent access to high-quality foods (Ungar et al 2006, Schoeninger et al 2001).

Regardless of the predominate meat procurement mode, the increased availability of protein and fat in the diet of H. erectus via oil-rich seeds, underground storage organs and meat (Peter 2007) would provide consistently available, high-quality, calorie-rich fuels for such energetically expensive adaptations as a large brain.

1.2.3 Trade-off hypotheses regarding diet

As discussed above, a hallmark of the evolution of human diet is the inclusion of a high percentage of nutrient-rich foods (including animal products). Several hypotheses have been put forth to connect changes in nutrition with evidence of adaptation, specifically the increase in brain size over the past 2 mya, and many of these focus on tissue mass changes. The first, and most prominent, of these hypotheses is the expensive tissue hypothesis proposed by Aiello and Wheeler (1995). They noticed that the total mass-specific basal metabolic rate of humans is well within the range of other primates, but that we have a larger brain, which results in greater energy requirements.

They predicted that some other structure or structures had to decrease in mass in order to reduce energy consumption and allow for the expansion of our metabolically demanding human brain. Aiello and Wheeler (1995) hypothesized that the increasing quality and digestibility of the hominin diet during evolution allowed for the reduction of the energetically expensive gut tissue. Then, the net gain in energy could be allocated to the human brain. The differences in gut size among mammal species are consistent with empirical observations relating digestive organs to diet quality (Chivers and

Hladik 1980); herbivores usually have large guts to better extract nutrients from plant tissue, whereas the simpler gut of humans is more typically found in carnivores.

Another trade-off prediction comes from Leonard et al (2003), who suggest that a

decrease in muscle mass and an increase in adiposity provided a potential source of energy to fuel the evolution of the human brain in two main ways. First, an energetically expensive tissue, skeletal muscle, was reduced. Second, a tissue known for its ability to store energy, fat, was increased.

These hypotheses illustrate that diet may have acted as a releaser and a challenger in human evolution. Examples of a release of energetic constraint would be the re-allocation of energy-expensive tissues, allowing for the development and maintenance of the human brain, or an increase in diet quality in terms of energy value

(Aiello and Wells 2002, Fish and Lockwood 2003, Gaulin and Kurland 1976, Leonard and Robertson 1994). Diet can also act as a challenger as foraging for foods high in quality often provides both an energetic challenge as well as a cognitive challenge

(Kaplan et al 2000, McLean 2001, Hamilton and Busse 1978, Milton and May 1976,

Ruxton and Houston 2004). When considering diet as a releaser and a challenger of metabolism, we are in the position to identify and characterize interesting genetic and molecular candidates that could be responsible for adaptive traits.

1.3 Why study expression in primates?

Original primate comparative studies focused on anatomical and behavioral differences between humans and chimpanzees, but in 1975, King and Wilson extended these comparisons using molecular biology techniques. Initially, they examined the protein sequences of forty-four genes and found that humans and chimpanzees are

99.18% similar at this level. Given the striking phenotypic differences between these primate species, King and Wilson hypothesized that uniquely human traits must have arisen due to changes in gene regulation rather than protein structure. Since 1975, our understanding and definition of transcriptional regulation has changed and expanded, but its functional output still remains gene expression. Changes in gene expression can be adaptive and result in morphological (Stern 1998, Abzhanov et al 2004, Gompel et al

2005, Shapiro et al 2004, Prud’homme et al 2006, Abzhanov et al 2006, Steiner et al 2007,

Jeong et al 2008), physiological (Tournamille et al 1995, Tishkoff et al 2007) and behavioral (Alaux et al 2009) consequences (Carroll et al 2007, Wray 2007). In humans, there is abundant material for regulatory evolution (Rockman and Wray 2002) with more than one hundred documented cases of these mutations causing phenotypic change within humans (Wray 2007, Kleinjan and van Heyningen 2004).

Further reasons to initially focus on gene expression comparisons are the biological and technical advantages it affords. Biologically, expression changes are

known to effect adaptive changes and King and Wilson (1975) specifically hypothesized that transcript abundance is relevant for phenotypic comparisons between humans and chimpanzees. Technically, expression can be easily measured at the genomic level and the primate centers are competent in collecting tissues that contain quality mRNA.

When combined, the biological and technical advantages point to expression changes as an excellent place to start identifying the basis for trait differences between humans and chimpanzees. To date, the candidate gene approach has found several cases of regulatory, or structural changes resulting in expression differences between humans and chimpanzees that have a putative adaptive consequence (Huby et al 2001, Rockman et al 2005, Pollard et al 2006, Perry et al 2007, Prabhakar et al 2008). When using the candidate gene approach, expression can be measured in numerous ways including luciferase assays to quantitate the contribution of specific regulatory regions and quantitative PCR to obtain a snapshot of mRNA abundance for a certain gene or exon.

Moving beyond single gene studies, Enard et al (2002) inaugurated the field of primate transcriptomics comparing human, chimpanzee, orangutan, and rhesus macaque expression patterns in brain, liver, and whole blood using human microarrays.

Following this initial study, many others compared different regions of the brain, kidney, heart, testis, and fibroblast cells in great apes. Prior to the initiation of my thesis work, the published studies exclusively used microarray platforms for transcriptomic

analyses. Recently, next generation sequencing has expanded on these initial expression analyses allowing for insights into alternative splicing (Blekhman et al 2010), non-coding transcripts (Babbitt et al 2010), histone modifications (Cain et al 2011), methylation (Pai et al 2011), and DNaseI hypersensitivity (Shibata et al 2012). Sequencing techniques provides several advantages including greater genomic coverage, sensitivity, precision, and less measurement bias. Several collective trends have emerged from these comparative expression studies. Hundreds of genes are differentially expressed among primates with some of these being species, age, sex, and/or tissue specific (Khaitovich et al 2005, Gilad et al 2006, Somel et al 2009, Reinius et al 2008). Additionally, genes involved in energy metabolism are consistently differentially expressed between primate species (Caceres et al 2003, Uddin et al 2004, Khaitovitch et al 2004, Blekhman et al 2008). As a complementary analysis to expression measurements, bioinformatic studies have identified putative regulatory regions of genes that are under selection in humans as compared to non-human primates (Haygood et al 2007, Haygood et al 2009).

Evidence for positive selection in genes related to dietary processes include the acquisition and metabolic processing of nutrients (Haygood et al 2007).

With many lines of data pointing to energy metabolism as a key player during the evolution of our species, comparative studies need to investigate diet responsive tissues such as the cerebral cortex, skeletal muscle, and white adipose tissue.

1.4 Summary of chapters

My dissertation work is comprised of three main data chapters, each of which draws on a specific anthropological concept to analyze expression differences in tissues of dietary relevance. The final chapter is an exposition on the future of genomic primate research and my recommendations for research programs moving forward. Below I have outlined the major highlights of this work.

Comparative expression analysis of the phosphocreatine circuit in extant primates: implications for human brain evolution

Anthropological and dietary connection: Substantial consumption of meat distinguishes the human diet from other great apes (Stanford 1999, Stanford and Bunn,

2001). Creatine, an energy source in skeletal muscle and cerebral matter is found at high levels in meat. A network of five genes dubbed the phosphocreatine circuit is responsible for regulating energy homeostasis in these tissues. The expansion of the human brain (Schoenemann 2006) during the Paleolithic era required significant metabolic support (hypothesized by Aiello and Wheeler, 1995). As the consumption of animal products likely coincided with cerebral enlargement on the Homo lineage, differential regulation of the phosphocreatine circuit genes, measured by expression might have provided the necessary fuel.

Expression measurement method: Quantitative PCR on the five genes of the phosphocreatine circuit.

Tissue source: Cerebral cortex (in vivo), cerebellum (in vivo), and skeletal muscle (in vivo) from humans, chimpanzees and rhesus macaques.

Insights gained: SLC6A8 and CKB transcript levels are higher in the human brain, which could potentially increase the frequency and amount of available energy compared to non-human primates. In concert with other adaptions, this increase in energy allocation may have supported the increased bioenergetic demands of the rapidly expanding human brain.

Functional consequences of genetic variation in primates on tyrosine hydroxylase (TH) expression in vitro

Anthropological and dietary connection: The growth of the human brain could both cause as well as respond to changes in the human diet. For example, a richer supply of nutrients associated with the hunter-gather diet might allow for the increased size and complexity of the human brain. However, the hunter-gather strategy necessitates specializations associated with a highly developed nervous system such social interactions as well as tool making and use. Here we investigated the rate-limiting enzyme in catecholamine synthesis, tyrosine hydroxylase (TH). TH uses the amino acid tyrosine, found in high-protein foods, to create L-DOPA, which can be further synthesized into dopamine. Dopamine, controls several aspects of reward-seeking behavior (Arias -Carrion and Poppel 2007), which would have been critical when our ancestors adopted new dietary strategies. This specific article however is not focused on historic anthropological connections, but rather variation within modern human populations.

Expression measurement method: Transient dual luciferase assay to assess genetic variation in the promoter region of TH.

Tissue source: DNA from three human populations: a Luhyan in Webuye, Kenya, a

Yoruba in Ibadan, Nigeria and a CEPH panel member who is a Utah resident with ancestry from northern or western Europe. DNA from non-human primates: chimpanzee, gorilla, rhesus macaque. Culturing was done in three neuroblastoma cell lines to mimic neuronal environments (in vitro).

Insights gained: Naturally occurring human sequence polymorphisms 2kb region upstream of the TH translation start site demonstrates two unique properties. One, there are trans- effects between cell lines indicating that polymorphisms may have functions in some tissues, but not others. Two, these SNPs are non-additive in cis- indicating that there are multiple functional variants in a single promoter region.

Coordinated changes in adipose transcriptomes accompanied dietary shifts during human origins

Anthropological and dietary connection: The human genome evolved under specific dietary conditions over millions of years. Modern hominid history is marked by two periods of distinct and rapid dietary change. The first occurred ~2 mya in the form of increased meat consumption (Bunn and Kroll 1986, Sponheimer and Lee-Thorp 1999,

Ungar et al 2006). Oleic acid, an important and abundant fatty acid found in animal marrow and muscle, greatly increased in the human diet as a result (Cordain et al 2002).

The second period of dietary change occurred during the agricultural and industrial revolutions (Zeder 2006, Simopoulos 1999). These changes in food processing resulted in the dramatic increase in grain intake, specifically omega-6 linoleic acid. Although the biological importance of white adipose tissue in regulating lipid metabolism is established, no evolutionary study has investigated the molecular differences between humans and nonhuman primates in this tissue.

Expression measurement method: Transcriptomics using the next generation sequencing platforms SOLiD4 and HiSeq.

Tissue source: Adipose derived stromal cells differentiated into adipocytes (in vitro) from humans and chimpanzee and subcutaneous white adipose tissue (in vivo) from humans, chimpanzee and rhesus macaques.

Insights gained: Using a combined in vitro and in vivo approach, we provide evidence that humans translocate, bind, activate and synthesize more long chain fatty acids than nonhuman primates. Furthermore, the pathway responsible for processing essential fatty acids (EFA) critical for brain development is higher in humans; with FADS2 being specifically upregulated in humans compared to chimpanzees and rhesus macaques.

We hypothesize that this adaptation would have been beneficial during human origins, but the modern shift in EFA ratio to heavily favor omega-6s like linoleic acid over omega-3 fatty acids could result in inflammation due to increased prostaglandin production. Indeed, a higher degree of inflammatory signals in the human white adipose tissue support this claim and may give insight into modern human disease.

2. Comparative expression analysis of the phosphocreatine circuit in extant primates: implications for human brain evolution

Originally published in:

Pfefferle AD*, Warner LR*, Wang CW, Nielsen WJ, Babbitt CC, Fedrigo O, Wray GA. 2011. Comparative expression analysis of the phosphocreatine circuit in extant primates: Implications for human brain evolution. Journal of Human Evolution 60:205- 212.

My contribution to this work:

1) Contributed to the conception and design of experiments.

2) Contributed to data analysis.

3) Contributed to writing the manuscript.

2.1 Abstract

While the hominid fossil record clearly shows that brain size has rapidly expanded over the last ~2.5 million years, the forces driving this change remain unclear.

One popular hypothesis proposes that metabolic adaptations in response to dietary shifts supported greater encephalization in humans. An increase in meat consumption distinguishes the human diet from that of other great apes. Creatine, an essential metabolite for energy homeostasis in muscle and brain tissue, is abundant in meat and was likely ingested in higher quantities during human origins. Five phosphocreatine circuit proteins help regulate creatine utilization within energy demanding cells. We compared the expression of all five phosphocreatine circuit genes in cerebral cortex, cerebellum, and skeletal muscle tissue for humans, chimpanzees, and rhesus macaques.

Strikingly, SLC6A8 and CKB transcript levels are higher in the human brain, which should increase energy availability and turnover compared to non-human primates.

Combined with other well-documented differences between humans and non-human primates, this allocation of energy to the cerebral cortex and cerebellum may be important in supporting the increased metabolic demands of the human brain.

2.2 Introduction

The rapid expansion of brain size that began ~2.5 mya in the lineage leading to modern humans (Schoenemann 2006, Tattersall 2008) required substantial metabolic support (Aiello and Wheeler 1995, Leigh 2004, Dunbar and Shultz 2007, Isler and van

Schaik 2009). Comparisons of both DNA sequence (Haygood et al 2007) and mRNA abundance (Uddin et al 2004, Khaitovich et al 2006a, Blekhman et al 2008, Babbitt et al

2010) indicate that extensive changes in the regulation of metabolic associated genes helps distinguish humans from chimpanzees. A dietary shift toward increased meat consumption by early hominids (Stanford 1999, Stanford and Bunn 2001, Ungar et al

2006) may have contributed to some of the bioenergetic modifications necessary to support the human brain (Milton 1987, Leonard and Robertson 1992 & 1994, Milton 1999

& 2003, Leonard et al 2007). Comparative studies of primate genetics and molecular function provide powerful tools for identifying specific molecular changes associated with human diet (Luca et al 2010), an important step in understanding how changes in physiology allowed for the dramatic expansion of our brains.

Creatine, an abundant metabolite of red meat (Williams 2007), occurs at higher concentrations in the plasma of humans who consume meat (Delanghe et al 1989,

Shomrat et al 2000) and was likely present at higher quantities in the diet of human ancestors. Interestingly, creatine metabolism has been shown to positively correlate

with brain activity (Sauter and Rudin 1993, Du et al 2008) and creatine supplements may improve mental performance in humans (Rae et al 2003, McMorris et al 2007). Thus, increased meat consumption along the hominin lineage may have influenced brain metabolism by providing additional creatine. The importance of creatine in helping maintain brain energy homeostasis (Wyss and Kaddurah-Daouk 2000, Brosnan and

Brosnan 2007, Tachikawa et al 2007) makes the phosphocreatine circuit a particularly attractive candidate for beginning to uncover the molecular changes associated with increased encephalization.

The phosphocreatine circuit begins when creatine diffuses across the plasma membrane, through the cytosol, and into the mitochondria (Figure 2.1a-b), where it is converted to phosphocreatine (Figure 2.1c), a high-energy compound (Wallimann et al

1992). Phosphocreatine serves several functions within the brain, skeletal muscle, and other energetically demanding tissues (Brosnan and Brosnan 2007) (Figure 2.1). First, phosphocreatine assists the cell in performing energy expensive intensive processes in the cytosol (Figure 2.1e-f), such as muscle contractions and maintaining membrane potentials (Wyss and Kaddurah-Daouk 2000). Second, the phosphocreatine circuit serves as an energy shuttle to cellular sites with high ATP utilization (Figure 2.1d) (Wyss and Kaddurah-Daouk 2000). Phosphocreatine is able to diffuse across the cytosol faster than ATP, allowing for efficient support of biological processes farther from the

mitochondria (Ellington 2001). Third, by donating a phosphate to ADP, the phosphocreatine circuit also acts to buffer ATP concentrations (Figure 2.1e) (Wyss and

Kaddurah-Daouk 2000). Without ATP buffering, cellular processes regulated by ATP concentrations would be adversely affected by energy fluctuations during times of rapid energy utilization, leading to cell toxicity (Matthews et al 1999). These functions are particularly important in the brain. While skeletal muscle contains large amounts of glycogen for energy storage (Fisher et al 2002), the brain has few mechanisms for storing energy (Peters et al 2004), making the phosphocreatine circuit particularly important for maintaining proper brain energy homeostasis (Wallimann and Hemmer 1994).

Given that the phosphocreatine circuit is responsible for critical cellular functions and the association between a meat-rich diet and elevated creatine levels, we hypothesize that gene regulatory changes occurred during human evolution to increase phosphocreatine circuit gene expression specifically in the brain as a way of increasing

ATP energy availability and turnover. This study compares the expression of the genes that encode the phosphocreatine circuit in humans, chimpanzees, and rhesus macaques.

The goals are to better understand the role of this circuit in these primates and to infer the functional significance of these differences in the context of human evolution.

Figure 2.1 Schematic representation of the phosphocreatine circuit

Figure 2.1) (A) Creatine enters cells through the membrane transporter SLC6A8.

(B) Creatine moves across the outer mitochondrial membrane through porin.

CKMT2. (D) Phosphocreatine moves through porin back into the cytosol where it can diffuse to site with high ATPase activity. (E) Phosphocreatine interacts with either CKB or CKM to generate ATP. (F) The resulting ATP is then available as a source of energy for cytoplasmic ATPases and creatine returns to the mitochondria. ATPases, such as the sodium-potassium pump, are proteins that typically utilize energy from ATP to perform a specific cellular function.

2.3 Materials and Methods

2.3.1 Primate sample collection

Samples were obtained from the Kathleen Price Bryan Brain Bank at Duke

University (Homo sapiens, frozen tissue), BioChain Institute Incorporated (Homo sapiens, total RNA), Southwest National Primate Research Center (Pan troglodytes and Macaca mulatta, frozen tissue), New England Regional Primate Research Center (Macaca mulatta, frozen tissue), and Yerkes National Primate Research Center (Macaca mulatta, frozen tissue) (Table 2.1).

2.3.2 Tissue preparation, total RNA isolation, and cDNA synthesis

All RNA extractions from collected tissue were performed at Duke University's

Regional Biocontainment Laboratory (RBL). Special care was taken to section each tissue consistently within and between species. Whole brain tissues maintained natural form in the freezer, allowing for anatomic landmarks, such as major sulci and gyri, to be used for proper brain region identification. Cerebral cortex tissue sections were specifically taken from the frontal lobe and skeletal muscle tissue samples were taken from the vastus lateralis. All tissue sections were superficially taken with a width of

~4mm, a depth of ~4mm. Tissues were homogenized in QIAzol lysis buffer (Qiagen) using a Tissuelyser II (Qiagen). Total RNA samples were purified using an RNeasy lipid tissue kit (Qiagen) in conjunction with RNase-free, DNase set (Qiagen).

Table 2.1 Primate samples used in this study

2.3.3 Primer design

Gene sequences for SLC6A8, CKMT1A, CKMT1B, CKMT2, CKB, and CKM were obtained from Ensembl genome browser using the Homo sapiens 36.3, Pan troglodytes 2.1, and Macaca mulatta 1.1 builds. PCR primers (Sigma-Aldrich) were designed within completely conserved exonic regions among all transcript isoforms and species (Table

2.2). Because CKMT1A and CKMT1B encode for identical proteins (Ensembl), PCR primers could not be designed to specifically amplify one gene and not the other. As such, a single PCR primer was designed to simultaneously amplify both genes and is referred to as CKMT1. Special care was taken to ensure that the amplified exonic region was found in all known transcript isoforms for each gene. This is important because transcript isoforms may be differentially expressed between species and would complicate our interpretation of the data. This potential confounding factor is controlled by ensuring all isoforms are simultaneously captured in our expression measurements.

Primers were selected using Primer3 Input v0.4.0 (Rozen and Skaletsky 2000). The primer sequences were blasted to all three species’ genomes using Ensembl BLAST and a test PCR was performed on human cDNA to ensure only one product for each primer pair (Figure 2.3).

Figure 2.2 Verifying cDNA purity using SDHA control primers that amplify an intronic region

Figure 2.2) These four gels indicate that there is no genomic DNA contamination in any of the samples that could bias the expression measurements. For the last gel, C. Cortex: cerebral cortex and S. muscle: skeletal muscle. The positive control for all three gels was human genomic DNA.

Table 2.2 Primer sequences for quantitative RT-PCR and verifications

Figure 2.3 Primer validation for phosphocreatine circuit and control genes Figure 2.3) An agarose gel ensuring the PCR amplicon size matched Ensembl BLAST predictions.

2.3.4 Quantitative RT-PCR

Quantitative RT-PCR measurements were conducted on a Mastercycler ep realplex machine (Eppendorf) in 10 µL reactions: 5.0 µl 2X QuantiFast SYBR Green PCR

Kit (Qiagen), 0.25 µl for each primer (10 µM), 0.5 µl of cDNA template, and 4.0 µl PCR quality water. The following PCR program was used for all reactions: 95oC for 5 minutes, 40 cycles of 95oC for 15 seconds and 60oC for 30 seconds, followed by a melt curve from 60 to 95oC. A single peak was detected on all melt curves, ensuring a single amplification product size for all species. Ct values were determined using the

CalQPlex setting with a baseline drift correction. For each primer pair, a standard curve was setup on human brain or skeletal muscle cDNA over a twelve point, factor of two dilution series to determine the efficiency and working Ct range of each primer set. All primer sets had an efficiency between 94 and 100 percent with r2 values greater than

0.99 (Table 2.2).

Data were collected by running each experimental and control sample in technical triplicate. For genes with medium to high expression (as defined by a mean Ct

< 33 PCR cycles), only measurements with low standard deviation across replicates (Std

Dev < 0.4 Ct) were used in the expression analysis (Karlen et al 2007). All tissue specific genes fell into this category of medium to high expression (e.g. CKB in both brain regions). For genes with low expression (as defined by a mean Ct>33 PCR cycles), a 36

higher standard deviation threshold was implemented (Std Dev <1.0 Ct), due to increased variation of measurements in this Ct range (Karlen et al 2007). All genes that fell into the category of low expression were genes that were being expressed in their non-dominant tissue (e.g. CKB in skeletal muscle). Within plates, expression was normalized with two control genes (SDHA and EEF2) that were selected based on their performance in a geNormTM analysis on brain and skeletal muscle tissue for humans and chimpanzees, as well as, having a similar expression level to the genes of interest

(Vandesompele et al 2002, Pattyn et al 2003, Fedrigo et al 2010). An inter-run calibration was conducted by running the control gene, EEF2, on IMR-32 cell cDNA on each plate

(Hellemans et al 2007). To convert the raw Ct expression into normalized relative expression, we used a modified delta-delta Ct method (our code is available at: http://www.biology.duke.edu/wraylab/wraylab/Resources.html) (Vandesompele et al

2002, Hellemans et al 2007, Fedrigo et al 2010).

2.3.5 Gene expression comparison

Although no generally accepted method exists for overlaying gene expression profiles onto a phylogenetic tree, a few studies have put forth alternatives for identifying natural selection from expression (Gilad et al 2006, Khaitovich et al 2006b). We focused on identifying gene expression differences between species (Figure 2.4) using a Mann-

Whitney test to calculate statistical significance as this approach does not assume a normal data distribution and it works well with smaller sample sets. Even though the

Mann-Whitney test works well with small sample sets, larger sample sets have more statistical power to identify significant expression differences and have the potential to give lower p-values. This should be taken into account when comparing p-values across tissues with a higher number of samples (cerebral cortex) to those with a smaller sample set (cerebellum). The motivation behind this research is to identify human-specific expression patterns, as we believe those traits are more likely to be associated with human-specific phenotypes. As such, two sets of expression comparisons were performed, human versus chimpanzee and human versus rhesus macaque (Table 2.3).

2.3.6 Positive selection analysis

2.3.6.1 Sequence data and annotation

Sequences were obtained from the University of California Santa Cruz Genome

Bioinformatics website (UCSC; http://genome.ucsc.edu/) using the human (hg19), chimpanzee (panTro2), orangutan (ponAbe2) and rhesus macaque (rheMac2) genomes.

From UCSC and refSeq human annotations, we used customized scripts to extract the 5’ flanking region, 5’ untranslated region (UTR), coding exons, 3’ UTR as well as the intronic regions of six genes: creatine transporter (SLC6A8), creatine kinase mitochondrial 1A (CKMT1A), creatine kinase mitochondrial 1B (CKMT1B), creatine kinase mitochondrial 2 (CKMT2), creatine kinase brain (CKB), and creatine kinase muscle (CKM). The compartment identity was determined by taking the intersection of all known transcript isoforms. The putative promoter sequence or 5’ flanking region was annotated by taking a 5 kb region upstream of the most 5’ transcription start site.

This 5 kb region is believed to contain most of the promoter elements (Wray et al 2003,

Blanchette et al 2006, Crawford et al 2006). Some genomic regions could not be mapped because of missing sequences or complicated UTRs that prevented reasonable homology calls (e.g. CKB 5' UTR). These regions were not tested for positive selection. Gene compartments annotated from the human genome (coding exons, transcript spans, introns, and non-coding regions) were independently mapped using pslMap and

liftover chain files from UCSC, and checked for chromosomal and location syntenies.

The annotations were checked against and found to be consistent with the Ensembl genome browser annotation of the same genomic regions. CKMT1A was not kept in subsequent analyses because of missing sequence and annotation issues in at least one species. Sequences were aligned across the four species using MUSCLE (Edgar 2004).

We masked nucleotide sites containing Ns and gaps across all species and visually checked for alignment quality.

2.3.6.2 Tests for positive selection

The modified version of the branch site models (Zhang et al 2005) was used to scan for positive selection on both the human and chimpanzee branches. The goal of this method is to detect a lineage-specific accelerated nucleotide substitution rate, relative to a neutral rate (ratio of the rates sequence of interest / neutral sequence).

Using an appropriate neutral proxy is crucial for this type of method (section 2.3.6.3).

This method contrasts two models with a likelihood ratio test (LRT): a null model with no positive selection (but allowing for relaxed constraint) and an alternative model with positive selection on the branch of interest (i.e. human or chimpanzee). The LRT significance, suggestive of signatures for positive selection, was assessed using a χ2 with one degree of freedom. The scan was performed 10 times with random starting parameters and the best likelihood scores were used to prevent local optima. All the

likelihood fittings were performed with customized and provided scripts in the HyPhy software (Pond et al 2005). Coding and noncoding sequences were analyzed with variations of this method (Wong and Nielsen, 2004; Zhang et al 2005; Haygood et al

2007).

2.3.6.3 Determining a neutral proxy

Test for signatures of positive selection on coding sequences have been performed with the commonly used synonymous substitution rate as neutral proxy. For the tests on noncoding sequences, we chose introns for neutral proxy as used in

Haygood et al (2007) because introns are the least constrained sequences in the genome

(Hellmann et al 2003, Keightley et al 2005). We selected all introns in a 100-200 kb window centered around each gene of interest. In order to account for the presence of conserved nucleotide sites in introns that may inflate the test for positive selection, we filtered the intron sequences by excluding 100 bp at each extremity of the introns (i.e. splicing signal sites; Sorek and Ast 2003). Additionally, we discarded first introns since they often contain regulatory elements and included at most 2500 bp of intronic sequences, drawn from the edges, because long introns often contain regulatory elements in their center (Blanchette et al 2006). Finally, because some conserved functional elements may be still present, 100 bootstrap replicates were performed on the

intron sequences, the tests for selection were performed for each of the bootstrap replicates, and the median p-values were considered for significance.

Figure 2.4 Phosphocreatine circuit gene expression comparisons among species

Figure 2.4) Phosphocreatine circuit gene expression comparisons among species.

Quantitative PCR measurements for the creatine transporter and kinases in humans, chimpanzees, and rhesus macaques. Individuals are each represented by a point, the horizontal bar is the mean, and the spread of the bar from the mean represents one standard deviation. SLC6A8: creatine transporter, CKMT1: creatine kinase mitochondrial 1, CKMT2: creatine kinase mitochondrial 2, CKM: creatine kinase muscle,

CKB: creatine kinase brain, Hsap: Homo sapiens, Ptro: Pan troglodytes, Mmul: Macaca mulatta.

Table 2.3 p-values for human-specific expression

2.4 Results

2.4.1 Comparative expression analysis

To our knowledge, this is the first study to comprehensively report expression measurements of the phosphocreatine circuit genes within nonhuman primates. Since we are interested in identifying which phosphocreatine circuit genes may be important for human-specific traits, we focused our attention on two comparisons: human versus chimpanzee and human versus rhesus macaque. These two comparisons allow us to identify tissue-specific expression signatures that are unique to humans among these three species.

2.4.2 Creatine transporter, SLC6A8

The phosphocreatine circuit begins with the active transport of creatine through a dedicated transmembrane protein, SLC6A8, into energetically expensive tissues, such as the brain and skeletal muscle (Snow and Murphy 2001) (Figure 2.1a). Since creatine must be transported across the plasma membrane before the cell can harvest its energy potential (Figure 2.1b-f), SLC6A8 is a critical protein for fueling the underlying the phosphocreatine circuit. We, therefore, began by measuring transcript abundance from the SLC6A8 gene that encodes this protein. Comparisons based on quantitative RT-PCR reveal higher mRNA transcript abundance of the creatine transporter gene SLC6A8 in humans than in chimpanzees and rhesus macaques in both the cerebral cortex (1.7-fold,

p-value = 0.028 and 2.0-fold, p-value = 0.007, respectively) and the cerebellum (1.8-fold, p-value = 0.05 and 2.3-fold, p-value = 0.034, respectively) (Figure 2.4). Analyzing the skeletal muscle samples reveals almost equal expression of SLC6A8 when comparing human to chimpanzee and rhesus macaque (1.3-fold, p-value = 0.917 and 1.1-fold, p- value = 0.807, respectively) (Figure 2.4).

2.4.3 Mitochondrial creatine kinases, CKMT1 and CKMT2

Once inside the cell, creatine primarily interacts with one family of proteins, the creatine kinases. Each of the four creatine kinase family members has tissue, intracellular, and substrate preferences, allowing for precise metabolic control

(Wallimann et al 1998) (Figure 2.1). Two mitochondrial kinases, creatine kinase mitochondrial 1 (CKMT1) and creatine kinase mitochondrial 2 (CKMT2), are primarily expressed in the brain and skeletal muscle, respectively. Coupled with oxidative phosphorylation, these kinases localize within the mitochondria to catalyze the production of phosphocreatine, a high-energy phosphate compound similar to ATP

(Figure 2.1c) (Vendelin et al 2004).

Consistent with previous reports in humans (Wyss and Kaddurah-Daouk 2000), we found that CKMT1 and CKMT2 are also expressed in a tissue specific fashion in chimpanzees and rhesus macaques, with CKMT1 predominant in the brain regions and

CKMT2 in skeletal muscle (Figure 2.4). Although we focus our discussion for each gene

on their primary tissue of expression, all human versus chimpanzee and human versus rhesus macaque statistical comparisons were performed and are reported in Table 2.3.

When comparing human CKMT1 expression to chimpanzee in the brain regions, we observe about equal transcript abundance in the cerebral cortex (0.9-fold, p-value =

0.242), but an increase in the cerebellum (1.7-fold, p-value = 0.028). A more consistent pattern was measured between the human and the rhesus macaque brain samples, with humans having lower CKMT1 expression in both the cerebral cortex (0.3-fold, p-value =

0.007) and the cerebellum (0.7-fold, p-value = 0.289).

The other mitochondrial kinase, CKMT2, is expressed primarily in skeletal muscle. Human to chimpanzee comparisons show decreased expression (0.7-fold, p- value = 0.754) of CKMT2 in the skeletal muscle samples, but an analysis of human to rhesus macaque shows increased expression (1.6-fold, p-value = 0.327) (Figure 2.4).

2.4.4 Cytosolic creatine kinases, CKB and CKM

After synthesis by the mitochondrial kinases, phosphocreatine diffuses out of the mitochondria (Figure 2.1d) to sites with high ATPase activity (Figure 2.1f) where it is able to interact with cytosolic creatine kinase (Figure 2.1e) (Wallimann et al 1998). Two cytosolic kinases, creatine kinase brain type (CKB) and creatine kinase muscle type

(CKM), are predominantly expressed in the brain and skeletal muscle, respectively.

Complementary to their mitochondrial counterparts, the cytosolic kinases drive the

production of ATP (Figure 2.1e). This reaction is coupled with cytosolic ATPases to provide energy for metabolic processes (2.1f), such as muscle contractions and maintaining plasma membrane potentials (Kushmerick 1998, Wyss and Kaddurah-

Daouk 2000).

Similar to previous reports in humans (Wyss and Kaddurah-Daouk, 2000) and to the mitochondrial kinases measured in this study, expression analysis of both cytosolic creatine kinases reveals a corresponding tissue-specific expression pattern in chimpanzee and rhesus macaque (Figure 2.4). CKB is dominant in both brain regions of interest while CKM is dominant in skeletal muscle. As with the mitochondrial kinases, for each of these genes we concentrate on the tissue in which they are most highly expressed.

We found that CKB expression has undergone a human-specific increase in the cerebral cortex and cerebellum compared to chimpanzee (2.0-fold, p-value = 0.008 and

2.5-fold, p-value = 0.014 respectively) and rhesus macaque (1.8-fold, p-value = 0.011 and

1.9-fold, p-value = 0.077 respectively) (Figure 2.4).

Consistent with the other skeletal muscle genes (SLC6A8 and CKMT2), CKM lacks a species-specific expression pattern in our skeletal muscle samples. Average CKM expression in skeletal muscle shows an increase in abundance in humans compared to

chimpanzee (1.9-fold, p-value = 0.465), but about equal expression compared to rhesus macaque (1.1-fold, p-value = 0.462).

2.4.5 Positive selection

We aligned the homologous regions for all five phosphocreatine circuit genes: creatine transporter (SLC6A8), creatine kinase mitochondrial 1 (CKMT1B), creatine kinase mitochondrial 2 (CKMT2), creatine kinase brain (CKB), and creatine kinase muscle (CKM). We compared four regions of each gene (5’ flanking, 5’ UTR, coding, and

3’ UTR) by scanning for positive selection on both the human and chimpanzee phylogenetic branches. Some homologous regions of these genes could not be aligned because of missing sequences or complicated UTRs that prevented reasonable homology calls (e.g. CKB 5' UTR). Our scans for positive selection on these five genes did not identify any regions that showed evidence of positive selection (p<0.05).

2.5 Discussion

In an influential paper published in 1975, King and Wilson posited that changes in gene regulation generated many of the phenotypic differences that distinguish humans from chimpanzees (King and Wilson 1975). Controlling the abundance of mRNA is one of the most important aspects of gene regulation, as fluctuations in the expression of specific genes are known to produce a wide variety of phenotypic consequences (Wray 2007). Comparing gene expression between primate species provides an initial approach for understanding observed physiological differences.

In this study, we measured expression of the phosphocreatine circuit genes to determine if they are differentially expressed between primate species. Since the brain is such an energetically expensive organ, the approximately two-fold increase in cranial capacity that occurred during the past ~2 million years (Schoenemann 2006) imposed a substantially larger metabolic demand (Aiello and Wheeler 1995, Leonard et al 2007). A shift toward increased meat consumption may have contributed towards meeting that increased demand (Milton 1999, Stanford 1999, Stanford and Bunn 2001, Ungar et al

2006). Knowing that creatine is an abundant nutrient in red meat (Williams 2007) and that phosphocreatine is critical to metabolically active cells (Wyss and Kaddurah-Daouk

2000, Brosnan and Brosnan 2007, Tachikawa et al 2007), we hypothesized that a brain- specific increase in of phosphocreatine circuit gene expression arose in the lineage

leading to humans. Higher expression of this circuit in humans would have provided additional ATP energy to brain cells by increasing ATP turnover and transport efficiency

(Wyss and Kaddurah-Daouk 2000, Snow and Murphy 2001).

Our results show that there is higher expression of genes encoding two key components of the phosphocreatine circuit in the cerebral cortex and the cerebellum of humans (Figure 2.4). The first of these genes, SLC6A8, encodes a protein that mediates creatine transport across the plasma membrane (Snow and Murphy, 2001). The importance of intracellular creatine for normal brain anatomy, physiology, and cognition is revealed by creatine transporter deficiency syndromes (MIM ID #300352), which involve impaired transport of creatine across the blood brain barrier and lead to serious health consequences in humans, including mental retardation, language impairment, seizures, and microcephaly (deGrauw et al 2003, Schiaffino et al 2005,

Anselm et al 2006). These phenotypes indicate that the transport of creatine into the brain is important within humans and suggest that differences in intracellular creatine concentrations between primate species may also be significant.

Expression comparisons reveal that SLC6A8 is expressed at about twice the level in the cerebral cortex and cerebellum of humans as it is in chimpanzees (Figure 2.4).

Given that SLC6A8 expression is positively correlated with intracellular creatine concentrations (Wyss and Kaddurah-Daouk 2000), the observed increase in SLC6A8 expression, combined with increased meat intake, would likely increase the transport of 52

creatine into the brain. In contrast, SLC6A8 expression levels in skeletal muscle are not significantly different between human and chimpanzee (Figure 2.4). Thus, not only did more creatine become available to the body from the dietary shift toward meat consumption, but a greater proportion of this creatine is likely transported into the brain as opposed to another metabolically demanding tissue, skeletal muscle. Higher intracellular creatine concentrations in the brain would fuel the phosphocreatine circuit

(Brosnan and Brosnan 2007).

The second phosphocreatine circuit gene whose expression differs between humans and chimpanzees is CKB. This gene encodes a kinase that generates ATP from

ADP using phosphocreatine as a source of high-energy phosphate in the cytosol. CKB protein plays a critical role in maintaining proper brain energy homeostasis (Wyss and

Kaddurah-Daouk 2000) and brain activity positively correlates with CKB function

(Sauter and Rudin 1993, Du et al 2008). In rats, creatine kinase regenerates ATP twelve times faster than through oxidative phosphorylation (Wallimann et al 1992). This rapidly available energy is important in regulating neurotransmitter release, maintaining membrane potentials, assisting growth cone migration, and restoring energy homeostasis

(Wallimann et al 1992, Wallimann and Hemmer 1994). Further evidence for CKB's importance comes from CKB -/- knockout mice. These mice show decreased spatial learning and decreased habituation behavior, in addition to other neurological conditions

(Jost et al 2002).

Our expression comparisons show that humans have approximately twice as much CKB mRNA in both the cerebral cortex and cerebellum compared to chimpanzees and rhesus macaques. The corresponding gene that is expressed in muscle, CKM, shows no difference in transcript abundance between these three species. Because CKB expression positively correlates with CKB protein activity (Ishikawa et al 2005), higher

CKB transcript abundance in humans may allow for more efficient ATP regeneration during energy utilization, helping support the increased metabolic demands of the human brain (Wallimann et al 1992, Wallimann and Hemmer 1994). The reaction catalyzed by CKB provides energy for a diverse set of enzymatic activities in the cytoplasm (Figure 2.1e-f), potentially providing humans with the ability to support a greater number of simultaneous enzymatic reactions in the brain.

Thus, gene expression underlying two key components of the phosphocreatine circuit are elevated in the brains of humans relative to chimpanzees and rhesus macaques. Interestingly, these observations are tissue-specific, suggesting differential allocation of a key metabolite between two metabolically demanding tissues in the body.

Furthermore, the expression changes in these two genes are likely to be synergistic by transporting more creatine into cells and increasing the capacity to utilize phosphocreatine as a source of energy for ATP-dependent enzymatic reactions. These reactions are consistent with the hypothesis that humans utilize the phosphocreatine

circuit more heavily than chimpanzees and rhesus macaques to support our energy demanding brain (Peters et al 2004).

An important concern for this study and follow-up experiments is whether gene expression differences are the result of genetic changes between species or the result of environmental effects, such as different diets. It is clear that environmental factors can influence gene expression (Idaghdour et al 2008, Somel et al 2008, Gibson 2008, Hodgins-

Davis and Townsend 2009). Of direct relevance to this study, creatine levels and creatine metabolism are influenced by dietary intake of creatine (Wyss and Kaddurah-

Daouk 2000, Snow and Murphy 2001, Brosnan and Brosnan 2007). Although it is not possible to carry out studies in chimpanzees and humans that fully control for dietary differences, these kinds of studies can readily be conducted with mice. Somel and colleagues (2008) investigated the effects of diet on gene expression in mouse. After feeding adult mice four different diets for eight weeks (chimpanzee diet, cafeteria food diet, McDonald's diet, and pellet diet), the authors measured transcript abundance using microarrays (Somel et al 2008). Importantly for the present study, they observed no significant difference in the expression levels of either SLC6A8 or CKB among the four diets (M. Somel, personal communication). While these data do not rule out the possibility that diet can influence the expression of these two genes, they do support the interpretation that the differences in transcription levels we observe are unlikely to be

exclusively due to diet. An interesting question for future studies will be parsing the relative influences of genetic factors, environmental influences, or a combination of both, on creatine distribution and utilization in primates.

At this point, it is not possible to conclude whether the gene expression differences we observe in the phosphocreatine circuit were adaptive. Comparisons of gene expression have been used to infer adaptation by concluding that interspecies differences that are significantly larger than intraspecies variation are more likely to result from positive selection than drift (Blekhman et al 2008). To date, these techniques have been optimized and applied on a genome wide scale, making it difficult to apply these same methods to our study. However, the pattern of expression changes in both

SLC6A8 and CKB loosely meet these criteria (Figure 2.4). The false positive rate using this test for selection is not well understood, and we do not draw firm conclusions from it. A second line of evidence regarding adaptation comes from analysis of DNA sequences. We sought evidence of positive selection on all five phosphocreatine circuit genes by examining three regions that can house gene regulatory elements (5’ flanking region, 5’ UTR and 3’ UTR) and one that encodes the protein function (coding region).

We found no evidence of positive selection for mutations in or around any of the five genes. It is important to bear in mind, however, that these methods are generally underpowered, and are unable to identify selection on single point mutations, any other

kind of mutation, or epigenetic modifications, any of which could influence gene expression. In fact, expression of CKMT1, CKM, and CKB can be influenced by epigenetic regulation (Uzawa et al 2006, Caretti et al 2004, Ishikawa et al 2005).

Although our data do not speak directly to the question of adaptation, they do focus attention on changes in specific molecular processes that may have contributed to a shift in energy allocation towards the brain during human evolution. Energy trade-off hypotheses predict that metabolic reallocation from other energetically demanding tissues to the brain allowed for greater encephalization in humans (Aiello and Wheeler

1995, Leonard et al 2003, Isler and van Schaik 2009). The tissue- and species-specific differences in SLC6A8 and CKB expression we reported here are consistent with these predictions. Perhaps the most convincing evidence that these expression differences are functionally important comes from genetic data showing that reducing the amount of normal SLC6A8 and CKB protein produces pathologic phenotypes related to the brain

(Jost et al 2002, deGrauw et al 2003, Schiaffino et al 2005, Anselm et al 2006). The implication to our study being that elevated expression of these genes would have increased the metabolic scope of the brain.

Comparative gene expression studies in primates provide an exciting opportunity to complement the extensive body of work investigating energetic trade- offs at the level of tissue mass (Aiello and Wheeler 1995, Leonard et al 2003, Isler and

van Schaik 2009) by giving molecular insight into the physiological underpinnings of those tissues. It seems highly unlikely that only a small set of molecular changes accounted for differential energy allocation among tissues during human evolution.

Indeed, our earlier genome-wide analysis of noncoding sequences in the same three species examined here suggested that diverse genes involved in carbohydrate metabolism experienced positive selection on regulatory sequences during human origins (Haygood et al 2007). Large-scale surveys of gene expression have begun to identify numerous genes whose expression differs among primate species (Uddin et al

2004, Khaitovich et al 2006a, Blekhman et al 2008, Babbitt et al 2010), greatly expanding our view of the specific molecular changes that accompanied the origin of humans as a distinct species. An important challenge for the coming years will be identifying which of these changes were associated with the evolution of uniquely human traits.

2.6 Conclusions

While it is well known that the anatomy and physiology of the human brain differs from other great apes in numerous regards (Deacon 1997, Schoenemann 2006), the underlying molecular mechanisms responsible for those differences have remained elusive. Gene expression analysis provides a rapid and powerful tool for identifying functional differences among primate species (Khaitovich et al 2006a). Our analysis of the phosphocreatine circuit has revealed two genes, SLC6A8 and CKB, in the phosphocreatine circuit that are consistently and differentially expressed between humans, chimpanzees, and rhesus macaques specifically within the cerebral cortex and cerebellum. Given the bioenergetic importance of this circuit and its association with dietary intake, increased expression of SLC6A8 and CKB in the human brain may have a profound influence on brain energy homeostasis today and during human origins by increasing ATP energy availability and turnover.

2.7 Acknowledgements

We thank the Kathleen Price Bryan Brain Bank, J. Pecotte and M.J. Aivaliotis at the Southwest National Primate Research Center (NIH-NCRR grant P51 RR013986), E.

Curran at the New England Regional Primate Research Center (NIH base grant

RR00168), K.L. Summerville at the Yerkes National Primate Research Center (NIH base grant RR000165) and J. Horvath at Duke University for biological materials used in this study. We would also like to thank R. Frothingham for his direction and guidance with tissue extractions in the NIAID Regional Biocontainment Laboratory at Duke University

(UC6-AI058607) and J. Tung and C. Wall for helpful discussions. NSF Grant NSF-BCS-

08-27552 (HOMINID) to GAW supported this study.

3. Functional consequences of genetic variation in primates on tyrosine hydroxylase (TH) expression in vitro

Originally published in:

Warner LR, Babbitt CC, Primus AE, Severson TF, Haygood R, Wray GA. 2009. Functional consequences of genetic variation in primates on tyrosine hydroxylase (TH) expression in vitro. Brain Research 1288:1-8.

My contribution to this work:

1) Contributed to the conception and design of experiments.

2) Performed all mutagenesis experiments including cell culture.

3) Contributed to data analysis.

4) Wrote manuscript.

3.1 Abstract

Tyrosine hydroxylase, the rate-limiting enzyme in catecholamine synthesis, is known to contain naturally occurring genetic variation in the promoter region that associates with a number of neuropsychological disorders. As such, examining non- coding regions is important for understanding tyrosine hydroxylase function in human health and disease. We examined ~2 kb upstream of the translation start site within humans and non-human primates to obtain a fine resolution map of evolutionarily and functionally relevant cis-regulatory differences. Our study investigated Macaca mulatta,

Pan troglodytes, Gorilla gorilla, and Homo sapiens haplotypes using transient dual- luciferase transfection in three neuroblastoma cell lines. In addition to trans- effects between cell lines, there are several significant expression differences between primate species, but the most striking difference was seen between human haplotypes in one cell line. Underlying this variation are numerous sequence polymorphisms, two of which influence expression within humans in a non-additive and cell line-specific manner.

This study highlights functional consequences of tyrosine hydroxylase genetic variation in primates. Additionally, the results emphasize the importance of examining more than one cell line, the existence of multiple functional variants in a given promoter region and the presence of non-additive cis- interactions.

3.2 Introduction

The presence of functional, non-coding genetic variation throughout the human genome has been well established (Pastinen et al 2006, Rockman and Wray 2002). Some of these cis- regulatory variants have functional consequences with respect to disease and fitness (De Gobbi et al 2006, Enattah et al 2008, Hamblin and Di Rienzo 2000,

Rockman et al 2005, Tishkoff et al 2007). One approach to assess the impact of non- coding sequence variation is to use unbiased scans to look for signatures of positive selection either between humans and other primates or between different human populations (Haygood et al 2007, Voight et al 2006). Another method is to identify regulatory regions for a candidate gene of interest experimentally. Here we assess the functional importance of genetic variation in the promoter region of tyrosine hydroxylase

(TH) using transfection assays.

Tyrosine hydroxylase is widely studied, as it is the first and rate-limiting enzyme in the synthesis of catecholamine neurotransmitters (Nagatsu et al 1964, Nagatsu 1989).

Due to the role TH plays in the synthesis of dopamine, norepinephrine and epinephrine,

TH coding and non-coding variants have been implicated in psychiatric disorders

(Leboyer et al 1990, Lobos and Todd 1997, Meloni et al 1995, Serretti et al 1998), and neurological disease (Hoffmann et al 2003, Ludecke et al 1995, Ludecke et al 1996). For instance, there is a detectable increase in tyrosine hydroxylase levels between

individuals that suffer from major depressive order and age matched controls (Zhu et al

1999) and the administration of certain antidepressants decreases the transcript abundance of tyrosine hydroxylase (Nestler et al 1990). As promoter regions have been associated with some of these disorders (Rao et al 2007, Ribases et al 2007, Verbeek et al

2007), understanding natural genetic variation and it’s role in regulation could have important clinical consequences.

Tyrosine hydroxylase is intricately regulated in vertebrates (Kumer and Vrana

1996). In humans, individual TH promoter elements confer different levels of cellular specificity highlighting the importance of both cis- and trans- effects in TH expression

(Kim et al 2006, Kim et al 2003, Romano et al 2005). Additionally, transcription factor binding sites have been experimentally validated in the TH promoter (Kessler et al 2003,

Kim et al 2006). When the ~11 kb region upstream of the mouse, rat and human TH transcription start site are aligned there are only five evolutionarily conserved regions

(Kessler et al 2003). This lack of conservation indicates that the promoter region may be changing relatively rapidly and phylogenetic comparisons closer than human-mouse are necessary to understand important changes on the human branch. Finally, within human populations, a recent study resequenced ~1.2 kb 5’ of the TH translation start site in 80 ethnically diverse subjects (Rao et al 2007). They found four common (minor allele frequency > 10%) single nucleotide polymorphisms (SNP) within this region, SNP-824

(rs10770141), SNP-801 (rs10840490), SNP-581 (rs10770140), and SNP-494 (rs11042962)

(Rao et al 2007). The effect of these common polymorphisms in a neural environment has not been characterized.

Given the importance of the promoter region to tyrosine hydroxylase expression and regulation, as well as its association with fitness and disease, we used transfection assays to assess the ~2 kb region upstream of the translation start site, a region known to contain functional polymorphisms. Transfection assays in human cell lines have been successfully used to identify and verify cis- regulatory variants effecting expression between primate species (Chabot et al 2007, Rockman et al 2005). Our study is the first to investigate genetic variation in tyrosine hydroxylase using non-human primates to polarize human specific changes in expression. We tested the impact of interspecific and intraspecific variation as well as the functional consequence of cis-regulatory element interaction.

3.3 Results

3.3.1 Sequence analysis between non-human primates and humans

In order to assess naturally occurring primate sequence variation 5’ of the TH translation start site, we isolated ~2 kb of this region for six haplotypes. The non-human primates investigated were macaque (AG07109), gorilla (AG05251), and chimpanzee

(AG06939). The human variation examined comes from three HapMap individuals, representing three geographically distinct human populations: a Luhyan in Webuye,

Kenya (GM19360), a Yoruba in Ibadan, Nigeria (GM18522), and a CEPH panel member who is a Utah resident with ancestry from northern or western Europe (GM12154).

When all six haplotypes, which represent a single haplotype from the population, are aligned, there are over two hundred sequence differences (Figure 3.1). We also obtained

~2 kb of TH intronic regions from macaque, orangutan, chimpanzee, and human and compared divergence in the 5’-flanking region to that of the intronic sequences

(excluding first introns, which often contain regulatory elements and intron ends, which generally contain splice sites), in order to test for signatures of positive selection between species (Haygood et al 2007). We found no significant acceleration of substitutions in the TH promoter region relative to the introns. Our finding suggests that the promoter region has experienced little or no positive selection along the human lineage (human p-value = 0.801, chimpanzee p-value = 0.081, orangutan p-value = 1.00,

gorilla p-value = 1.00). Moreover, measures of sequence polymorphism and linkage disequilibrium show no evidence of positive selection within human in this region based on the HapMap analysis panels (Voight et al 2006). Nevertheless, as tests for positive selection are generally underpowered and given the importance of tyrosine hydroxylase as the rate-limiting step in catecholamine synthesis, we wanted to functionally characterize the abundant variation in this regulatory region.

Figure 3.1 Alignment of six natural promoter region TH haplotypes

3.3.2 Expression of primate TH in multiple neuronal cell lines

To determine which aspects of these sequence differences have a functional impact, we performed in vitro dual luciferase assays for each of the six haplotypes in three neuroblastoma cell lines SH-SY5Y, SK-N-BE(2), and IMR-32, each of which is commonly used as a proxy for neurons in cell culture. These transfection assays were repeated on three different days with eight wells of replication per construct. We fitted a mixed-model ANOVA to the background normalized transfection expression measurements, for each haplotype in each cell line. By these means, we sought to eliminate not only random but also systematic errors, such as those arising from changes in the condition of the cell cultures from one day of assays to another. The fitted fiducial expression levels served as inputs to subsequent analyses.

To investigate trans- effects, the constructs were transiently transfected into three neuroblastoma cell lines. The rank order of expression when comparing the six constructs is different between the cell lines, indicating that different cellular environments affect TH gene expression (Figure 3.1). For example, the gorilla haplotype has the highest TH expression in the IMR-32 cells, but the lowest expression in the SH-

SY5Y and SK-N-BE(2) cells (Figure 3.2).

Figure 3.2 Normalized expression levels driven by TH cis-regulatory haplotypes in SH-SY5Y, SK-N-BE(2), and IMR-32 cell lines

Figure 3.2) There are repeatable differences between and within cell lines. The value plotted for each construct is the fitted overall mean (µ) plus the fitted construct effect

(Hi) in our mixed-model ANOVA. Error bars represent standard errors of fitted construct effects. Constructs not labeled at top with the same letter are significantly different from one another according to the Tukey-Kramer HSD test.

Table 3.1 Mean and standard deviation of TH expression in cell lines

When expression is compared between non-human primates and humans, there are differences between cell lines. The standard deviation of expression is greater among the non-human primates for SH-SY5Y and SK-N-BE(2) as compared to among humans (Figure 3.2 and Table 3.1). This is what one would expect, given there are many more sequence differences among the non-human primates than among the humans.

However, the reverse is true in the IMR-32 background where there is a greater standard deviation within humans as compared to non-human primates (Figure 3.2 and Table

3.1).

Additionally, there are significant and repeatable differences between haplotypes within each cell line. The letters above the bar graph indicate significant differences for

SH-SY5Y, SK-N-BE(2), and IMR-32 cell lines as tested by the Tukey-Kramer HSD test, which incorporates adjustment for multiple testing. Bars that do not have the same letter are significantly different from one another (SH-SY5Y p-value < 0.04; SK-N-BE(2) p-value < 0.001; and IMR-32 p-value < 0.05). When the constructs are analyzed in the

SH-SY5Y cell line, there are differences between the non-human primates with the constructs for the macaque and chimpanzee being significantly different from the gorilla, but there are no intraspecific differences among humans (Figure 3.2a). This is not the case when the TH 2 kb upstream region is assessed in the IMR-32 cell line.

Strikingly, in this background there are significant intraspecific differences within

humans. The GM19360 and GM18522 haplotypes are higher expressed as compared to the GM12154 individual (Figure 3.2c).

Together, these results indicate that multiple sequence differences, both within humans and between humans and non-human primates affect TH transcription, and that these consequences are to some extent dependent on cell type. In all three cell lines, the range of functional variation within humans is similar to, but lower than, interspecies differences. Below, we focus on identifying specific functional variants within humans, and their possible interactions.

3.3.3 Examining the functional impact of TH promoter SNPs in IMR-32 cells

To assess the functional consequences of the previously identified four common polymorphisms (Rao et al 2007) on TH expression in neuroblastoma cells, we used site- directed mutagenesis to alter these SNPs in the GM12154 background and tested these artificial haplotypes in IMR-32 cells. These transfection assays were repeated on three different days with eight wells of replication per construct. We again fitted a mixed- model ANOVA to the relevant transfection expression measurements, to obtain a fiducial expression level for each haplotype. The letters above the bar graph indicate significant differences (p < 0.05) by the Tukey-Kramer HSD test (Figure 3.3). The haplotype with all four major alleles had the highest expression (Figure 3.3).

Furthermore, the major allele haplotype is significantly different from the one containing the minor allele of SNP-801 or SNP-494 (Figure 3.3).

In order to identify sequence features underlying the expression differences seen between the natural human haplotypes in the IMR-32 cell line, we examined the sequence features in the 2 kb region directly upstream of the translation start site. A total of eleven variants exist among the three human constructs, with seven specific to

GM19360, two specific to GM18522, and two specific to GM12154 (Figure 3.4). The only two sequence features that are common to the high expressing GM19360 and GM18522 constructs and different from the lower expressing GM12154 construct are SNP-1898

(rs7115640) and SNP-801 (rs10840490) (grey columns, Figure 3.4). As such, these two features are the best candidates for explaining TH expression differences between humans in the IMR-32 cellular environment.

Figure 3.3 Normalized expression levels driven by TH cis-regulatory haplotypes in SH-SY5Y, SK-N-BE(2), and IMR-32 cell lines

Figure 3.3) The frequencies are taken from Rao et al 2007. The haplotype containing all of the major alleles has the highest expression level and is statistically different from the haplotypes containing the minor allele of SNP-801 or SNP-494 and is similar to the haplotypes containing the minor SNP-824 or SNP-581. All constructs are made in the

GM12154 background. Values reported, error bars, and letter labels are described in

Figure 3.2.

Figure 3.4 The human cis-regulatory variants of TH found in this study

Figure 3.4) Eleven polymorphic features occur in the 2 kb region upstream of the TH translation start site among the GM19360, GM18522, and GM12154 haplotypes. The arrow indicates where transcription starts. The derived alleles (vertical lines) and insertion/deletions (white boxes) are determined by the chimpanzee sequence. The sequence state is given for those features that are further investigated in this study. The

SNP number is the location upstream of the translation start site (+1) as compared to the reference sequence ENST00000352909. The grey columns indicate the sequence features investigated in Figure 3.5.

In order to assess the function of SNP-1898 and SNP-801 we mutated these two

SNPs separately and together from their GM19360/GM18522 state to the GM12154 state in the GM12154 background and tested these in the IMR-32 cell line. These transfection assays were repeated on three different days with eight wells of replication per construct. We again fitted a mixed-model ANOVA to the relevant transfection expression measurements, to obtain a fiducial expression level for each haplotype in

IMR-32 cells. As before, letters above the bar graph indicate significant differences (p <

0.001) by the Tukey-Kramer HSD test (Figure 3.5). We would anticipate that if a particular SNP is functional, the expression of the mutated GM12154 construct would be increased to levels similar to the GM19360 and GM18522 constructs. We indeed saw this with both of the single mutations: -1898T>C and -801C>G each raise the expression to levels significantly different from the natural GM12154 haplotype and statistically indistinguishable from the GM19360 and GM18522 haplotypes (Figure 3.5).

Interestingly, a construct containing both of these SNPs (-1898T>C; -801C>G) drives expression at a level similar to the natural, lower-expressing, GM12154 haplotype

(Figure 3.5). This result indicates that the interaction between these two SNPs is non- additive.

Figure 3.5 Functional consequences of SNP-1898 and SNP-801

Figure 3.5) Individually mutating SNP-1898 and SNP-801 results in increased expression; however, the double mutant is statistically the same as the natural haplotype. Values reported, error bars, and letter labels are as in Figure 3.2.

3.4 Discussion

This investigation set out to determine whether genetic differences in cis- affect

TH expression between primate species as well as within humans, using a method that has been successful in identifying and verifying cis- regulatory elements between primate species (Chabot et al 2007, Rockman et al 2005). In addition to confirming previous functional polymorphisms and identifying new variants in the 2 kb 5’ flanking region upstream of the TH translation start site (Figure 3.1), this study revealed several interesting results.

Looking beyond sequence signals to expression, prior studies have revealed that the same TH haplotype transfected into different cell lines can result in different levels of expression (Kim et al 2003). Our study, is the first to examine this in multiple primate species using three neuroblastoma cell lines, and demonstrates that variant-by-cell type interactions are widespread at this locus. We recognize that the significant expression differences observed in vitro might not necessarily translate to biologically meaningful differences in vivo. However, differences in transcript abundance in tyrosine hydroxylase have been associated with psychiatric (Zhu et al 1999) and neurological phenotypes (Joyce et al 1997).

By comparing humans to other non-human primates, we found that although there are clearly statistically significant differences between primate species, humans are

not categorically unusual (Figure 3.2). Some of the differences detected are attributed to trans- effects, as the same construct can show different levels of expression depending on the cellular context. This is seen as the rank order of expression for the six haplotypes changes depending on the cell line (Figure 3.2) and the standard deviations between non-human primates and humans are different depending on cellular context (Table

3.2). These findings highlight the importance of investigating expression in multiple cellular environments in order to reveal trans- effects. The cell lines we used originate from different biopsies, and neuroblastomas are known to be somewhat representative of different environments (Ciccarone et al 1989). If TH promoter activity were just studied in SH-SY5Y or SK-N-BE(2) cells, we would have missed an interesting case of sequence polymorphisms driving expression variation within humans (Figure 3.2c).

To extend the intraspecific analyses, we investigated the functional impact of standing common genetic variation in the TH promoter region. Previously, four common polymorphisms were discovered in the 1.2 kb region upstream of the translation start site (Rao et al 2007). We isolated each of these polymorphisms in the

GM12154 background and then examined their expression in IMR-32 cells. This approach is different from the previous study, which examined haplotypes in chromaffin cells and was concerned with associating SNPs with in vivo biochemical and physiological measurements (Rao et al 2007). There are significant differences between

these four common SNPs. Specifically; the minor allele of SNP-801 and SNP-494 has significantly different expression from the major alleles (Figure 3.3). This result in conjunction with those presented in Figure 3.5 demonstrates that multiple functional polymorphisms within a regulatory region can be segregating in human populations.

To date, only a few other studies have searched for multiple promoter variants that affect expression (Enattah et al 2002, Enattah et al 2008, Horan et al 2003, Rao et al 2007,

Tao et al 2006, Tishkoff et al 2007), so it is not clear how common this phenomenon is.

To further examine the natural intraspecific expression differences in IMR-32 cells, we investigated the sequence features associated with three human haplotypes

(Figure 3.4). In order to pinpoint functional variation, we focused on nucleotides where the two higher expressing haplotypes (GM19360 and GM18522) have the same state and the lower expressing haplotype (GM12154) has a different allele. The haplotypes we tested contain only two instances of this, at SNP-1898 and SNP-801, making these the most likely candidates for causing the expression differences (grey columns, Figure 3.4).

When either SNP was isolated individually in the GM12154 background, we saw a significant increase in expression from the natural GM12154 state (Figure 3.5). Although these single nucleotide mutations were artificially created, we know that these combinations of SNPs segregate naturally in human populations based on HapMap genotypes (International HapMap Consortium, Rel_23A).

Surprisingly, when these two SNPs were mutated together in the GM12154 background, the expression remained statistically unchanged (Figure 3.5). This is highly suggestive of non-additive interactions between SNP-1898 and SNP-801 and further analysis in other genetic background could potentially provide additional insight. The current results indicate that testing a single sequence feature will not always reveal its functional consequences and that one needs to examine other variants that are segregating in the surrounding regions in order to assess the possibility of combinatorial effects. This example adds to a growing number of genes where non-additive interactions between cis- acting SNPs have been found, including GH1 (Horan et al

2003), KRT1 (Tao et al 2006), and PDYN (Rockman et al 2005). The number of examples is most likely due to the small number of studies that have carried out fine-scale functional studies rather than a reflection of biological reality. Indeed, the extent of non- additive interaction between regulatory variants on gene regulation is poorly understood in general.

Knowing that these two SNPs impact expression in a transfection assay, transcription factor binding could be modifying TH transcription as transfection experiments can separate the effects of transcription factors from other types of expression regulation. Using biochemical approaches, Kim and colleagues (2006) identified three NRSF/REST binding sites in the 5’ regulatory region of TH, one of

which, NRSE-R, overlaps with SNP-1898 and is known to interact with the TH promoter in vitro. The GM19360 and GM18522 constructs have the same allele as the NRSE-R consensus sequence given for NRSF/REST binding at SNP-1898 (Kim et al 2006), suggesting a possible molecular mechanism for the expression differences between haplotypes. Further work is needed to validate the impact of transcription factor binding events at SNP-1898 and SNP-801 on TH expression.

We identified many inter- and intraspecific variants in the 5’ flanking region of

TH and examined their effects on expression in neuroblastoma cell lines. The expression differences we observed between human haplotypes in IMR-32 cells can be attributed partly to the functional variants SNP-1898 and SNP-801. Furthermore, we demonstrated the importance of using multiple cell lines due to trans- effects, the presence of non- additive interactions between regulatory variants, and the existence of multiple functional polymorphisms within a promoter region. Finally, this study provides some novel functional candidate variants for examining associations between tyrosine hydroxylase and neurological and psychiatric disease.

3.5 Experimental Procedure

3.5.1 Cloning and sequencing

Constructs were made from non-human primate fibroblast purified DNA from macaque, chimpanzee and gorilla (respectively, Macaca mulatta - AG07109; Pan troglodytes - AG06939; Gorilla gorilla - AG05251) and human lymphoblast purified DNA

(Luhya from Webuye, Kenya - GM19360; Yoruba from Ibadan, Nigeria - GM18522;

CEPH a Utah resident with ancestry from Northern or Western Europe - GM12154) received from Coriell Institute. The 2-kb TH cis-regulatory haplotypes were obtained through PCR-amplification (primers used: 5’ AGGCAAATCCCTCCAACGC 3’ and 5’

GGCTCAGTGTGGAGGTC 3’) using the high-fidelity polymerase Phusion (Finnzymes).

Individual PCR-amplification products were cloned into pGL4.1 vector using a KpnI restriction site on the 5’ end and a XhoI restriction site on the 3’ end. Constructs were then prepared using the Wizard midi-prep kit (Promega) and sequenced.

3.5.2 Positive selection analysis

The 5’ flanking regions for chimpanzee and macaque came from AG06939 and

AG07109. The 5’ flanking regions for human (ENST00000352909) and orangutan (Pongo pygmaeus, ENSPPYT00000003515), as well as all of the intronic sequences (introns: 3, 4, 5,

7, 8, 9, 10, and 11) came from Ensembl (human - ENST00000352909; chimpanzee -

ENSPTRT00000006077; orangutan - ENSPPYT00000003515; and macaque -

ENSMMUT00000005543). We fitted single-nucleotide substitution models by likelihood maximization using HyPhy (http://www.hyphy.org), taking the best of 10 fits starting from random points to guard against local maxima of the likelihood function (Haygood et al 2007). The models are equivalent to the HKY85 model (Hasegawa et al 1985) modulated within the 5’-flanking region relative to the intronic sequences in the same way that preferred models are modulated at non-synonymous sites relative to synonymous sites (Zhang et al 2005). The null model accommodates relaxation of negative selection on the so-called foreground lineage where the alternate model allows positive selection. We considered each terminal lineage (macaque, orangutan, chimpanzee, human) in turn as the foreground lineage. In each case, we tested the fit of the alternate model relative to that of the null model using a likelihood ratio test conservatively approximated as a chi-squared test with one degree of freedom (Zhang et al 2005). In each case, we analyzed 100 bootstrap replicates over the intronic alignment to guard against small-sample stochasticity; each p-value in Table S1 is the median over the replicates.

3.5.3 TH RT-PCR

To verify that SH-SY5Y, SK-N-BE(2), and IMR-32 cells express TH, fragments from these cell lines were separately reverse transcribed and then PCR amplified. RNA was extracted from SH-SY5Y, SK-N-BE(2), and IMR-32 cells with an Aurum total RNA

extraction kit (Bio-Rad). The cDNA single strand synthesis used a High Capacity cDNA Archive kit (Applied Biosystems). Primers were designed to amplify the four main TH alternative transcripts (Alternative Splicing Database at EMBL-EBI) (primers used: 5’ACTGCTGCCACGAGCTG 3’ and 5’ TGGACAGCTTCTCAATTTCCT 3’). PCR was then completed with the Absolute qPCR SYBR kit (Abgene). In all three cell types, the expected amplicon was observed at 123 bp (Figure 3.6). These results indicate that these cell lines are appropriate for assessing the expression of TH and that the readings obtained are not artifacts from this experiment.

3.5.4 Cell culture

SH-SY5Y and SK-N-BE(2) cells were cultured in a 1:1 mixture of Ham’s F12K and

MEME (ATCC) supplemented with 10% FBS (HyClone). IMR-32 cells were cultured in

MEME (Sigma-Aldrich) supplemented with 1 mM sodium pyruvate, 0.1 mM non- essential amino acids (Gibco), and 10% FBS (ATCC). All cell lines were acquired from

ATCC or the Cell Culture Facility at Duke University and were maintained at 37 oC with

5% CO2.

3.5.5 Site directed mutagenesis

The 2-kb TH cis-regulatory GM12154 haplotype in the pGL4.1 vector was methylated per GeneTailor Site-Directed Mutagenesis System (Invitrogen) and was then used as the DNA template. The SNP locations are given in reference to the human TH

sequence from Ensembl ENST00000352909. Six different single nucleotide changes were obtained through PCR-amplification using Platinum Taq DNA Polymerase high fidelity

(Invitrogen) using the following primer sets, -494G>A, (5’

AGACACACGGCCTGGAATCTTCTGGAG 3’ and 5’

CCAGGCCGTGTGTCTTGTAGACGTGGTT 3’), -581A>G (5’

GGGCGAGCTTGGGAAGCCGCTGCAAG 3’, and 5’

TTCCCAAGCTCGCCCCGTGGGGTCCAGAAT 3’), -801C>G (5’

CTAGCTCCTGGCTTCCCTCGGGGTCCTGT 3’, and 5’

AGGGAAGCCAGGAGCTAGCAGTGGGCCATA 3’), -824C>T (5’

AAAGAAGGGGCCACAGGACCCCCAGGG 3’, and 5’

CTGTGGCCCCTTCTTTAAAGAGCACGAACG 3’), -1898T>C (5’

GTGTTCCCCAAAGGCCTCCACCTCCTGT 3’, and 5’

CTTTGGGGAACACCGTGGAGGGGCATA 3’). The PCR products were transformed into One-Shot MAX Efficiency DH5αT1R E. coli. Constructs were then prepared using the Wizard midi-prep kit (Promega) and sequenced. In order to ensure the mutagenesis reaction did not alter the backbone vector, the plasmid was cut using KpnI and XhoI restriction enzymes and this TH fragment was ligated into a new pGL4.1 vector and transformed. These constructs were then prepared using the Wizard midi-prep kit

(Promega) and sequenced.

3.5.6 Transfection and expression measurements

To control for variation each construct was transfected in eight wells each day and each experiment was repeated three times on separate days. The first set of experiments used the six natural haplotypes and all three cell types, SH-SY5Y, SK-N-

BE(2), and IMR-32 (Figure 3.2). The second set of experiments used the three human natural haplotypes and the six mutagenesis created GM12154 constructs, in IMR-32 cells

(Figure 3.3 and Figure 3.5).

Transfections were performed in 24-well plates and the cells were seeded at 8x105 cells/ml in a volume of 500 µl of supplemented medium. Cells were transfected 24 hours after seeding using Lipofectamine 2000 (Invitrogen). The transfection mixture used was

2 µl Lipofectamine 2000, 100 µl OPTI-MEM, 800 ng reporter construct and 200 ng of

Renilla-TK as a co-reporter. Another control was 427 ng of empty pGL4.1 was added to the transfection mixture, which is the molar equivalent of the other constructs. Forty- eight hours after transfection the cells were lysed with 100 µl of Passive Lysis Buffer

(Promega) per well for 30 minutes. Lysates were read in the automated 96-well Veritas

Luminometer (Turner Biosystems) with auto-injection following procedures obtained from Promega, using a 2 second delay and 10 second read time.

3.5.7 In vitro expression analysis

In order to distinguish true biological signal from the background noise inherent in gene expression experiments, we fitted our measurements to a mixed-model

ANOVA. We first calculated the mean of the promoterless pGL4 vector over eight replicates for each cell type on each day, subtracted this from each expression measurement for a TH cis-regulatory haplotype in that cell type on that day, and computed the base-2 logarithm of the difference (Babbitt et al 2009). For each cell type, we then used restricted maximum likelihood (REML) to fit the “normalization model” yijk = µ + Hi + Dj + (HD)ij + εijk, where yijk is the logarithmically transformed, background- subtracted value of measured expression for haplotype i (1–12), day j (1–6), and well k

(1–8), µ is the overall mean, Hi is the fixed main effect of haplotype i, Dj is the random main effect of day j, (HD)ij is the random interaction effect of haplotype i and day j, and

εijk is the residual. This model, which resembles models commonly used in the analysis of microarray data (Wolfinger et al 2001), accounts for not only purely technical, well-to- well variation, but also systematic effects of day arising from, for example, day-to-day variation in the cell cultures. The fits indicate that these effects are substantial, accounting for between 35% and 70% of expression variation. For each cell type, we used the Tukey–Kramer procedure to assess the significance of each pairwise difference between fitted Hi's. The fitted Hi's constitute our best measures of the typical expression

from the haplotypes, so we call them “fiducial”. We performed these computations using the R system for statistical analyses. Our R code is available on request.

All relevant measured data from the six days are included in the different analyses except for one data point from the macaque haplotype (AG07109) in the IMR-32 cell line. This point was not included as its expression level was similar to the empty vector control.

3.6 Acknowledgements

We thank Adam Pfefferle and Gregory Nachtrab for review of this manuscript and helpful suggestions. This work was supported by the Institute for Genome Sciences

& Policy at Duke University and the National Institutes of Health [T32-HD040372 to

L.R.W.].

4. Coordinated changes in adipose transcriptomes accompanied dietary shifts during human origins

My contribution to this work:

1) Conceived and designed experiments.

2) Obtained funding for cell culture work.

3) Performed all cell culture work, imaging experiments, most of mRNA

extractions and library preparations.

4) Co-conducted data analyses.

5) Co-wrote manuscript.

4.1 Abstract

The appearance of metabolically expensive human adaptations, such as increased brain size, is proposed to have coincided with changes in energy acquisition and processing during human origins. A dietary shift toward increased lipid consumption during human origins, continues to differentiate us from chimpanzees today. Two periods during human history were likely marked by rapid increases in specific long-chain fatty acids: 1) meat consumption around 2 mya resulted in greater oleic acid and 2) increased cereal grain production during the agricultural and industrial revolutions introduced high levels of linoleic acid, an essential omega-6 fatty acid, into our diet. Defining the biological implications of these rapid dietary shifts is important for understanding human origins and modern day disease susceptibilities. White adipose tissue and its specialized cell type, the adipocyte, are essential for regulating lipid metabolism; yet, no comparative primate study has investigated the genetic basis of either in great apes. We found that human adipocytes are genetically primed to transport, bind and activate more long-chain fatty acids compared to the chimpanzee when cultured under the same conditions. When challenged with high levels of oleic or linoleic acid, humans express higher levels of genes important for synthesizing long chain fatty acids. Combining the in vitro approach with the in vivo tissue measurements,

FADS2, a rate-limiting step in the biosynthesis of unsaturated fatty acids was identified

as specifically upregulated in humans as compared to nonhuman primates. Increased

FADS2 could have been important for the appearance of several human adaptations during human origins. During the Paleolithic era, when essential fatty acid intake was balanced between omega-3 and omega-6, this expression change could have provided important building blocks for the rapidly growing human brain. In a striking case of maladaption in modern humans this pathway might be leading to the overproduction of proinflammatory prostaglandins, due to the increase of omega-6 in our diet. In support of this, inflammation gene ontology categories are enriched in human white adipose tissue. This study highlights the power of using a combined in vitro and in vivo approach to further our understanding of human origins.

4.2 Body of Paper

Food sources shifted substantially during the evolutionary origin of humans,

from a primarily vegetarian diet centered on fruit to a decidedly omnivorous diet

(Babbitt et al 2011). About 2 mya, the appearance of stone tools, large mammal bones

with cut-marks, shifts in stable isotope ratios, and dentition changes all point toward

increased consumption of meat and marrow (Bunn and Kroll 1986, Sponheimer and

Lee-Thorp 1999, Ungar et al 2006). Much later, about 10 kya, the dawn of the

agricultural revolution marked a substantial increase in the intake of grains by

anatomically modern humans (Simopoulos 1999, Zeder 2006). These shifts marked

dramatic changes in overall diet, in particular consumption of lipids like long-chain

fatty acids (LCFA) (Eaton 1992, Wells 2009). In addition to altering disease risk in

modern humans (Lawrence 2010), this evolutionary legacy likely had a profound

influence over the evolution of human metabolism (Aiello and Wells 2002), life

history (Finch and Stanford 2004), and brain size (Aiello and Wheeler 1995,

Broadhurst et al 1999).

Despite the central role of lipid metabolism in health and disease (Pond 1998),

next to nothing is known about changes in the molecular function of white adipose

tissue (WAT) or its specialized cell type, the adipocyte during human origins. WAT

integrates organismal energy balance through the regulation of food intake, storage,

and expenditure (Rosen and Spiegelman 2006). Adipocytes are the single cell type that distinguishes WAT from other tissues, and comprise approximately half of its constituent cells (Fruhbeck 2007). In this study, we adopted complementary in vitro and in vivo approaches to identify human-specific features of WAT and adipocyte gene regulation.

Adipocytes are readily cultured in vitro (Figure 4.1a), providing two key advantages: 1) the ability to control environmental conditions exposes genetic differences and 2) the ability to manipulate culture conditions allows experimental investigation of specific dietary differences. WAT, in contrast, provides less information about genetically based differences; however, it represents more biologically realistic in vivo function and in particular, reflects the influence of natural environmental differences. We cultured adipose stromal cells from humans and the only available chimpanzee cell line and induced them to differentiate into adipocytes in vitro (Figure 4.1a) (Table 4.1). A comparison of adipocyte mRNA abundance for 11,188 orthologous gene regions between human and chimpanzee using the Illumina platform revealed 909 transcripts that are expressed at significantly higher levels in humans (red in Figure 4.1b) (Table 4.2).

Figure 4.1 Phenotypic and transcriptomic profiles of human and chimpanzee differentiated adipocytes

Figure 4.1) (A) Adipose derived stromal cells before the differentiation and mature differentiated adipocytes (red, lipid droplets and blue, nucleus). (B) MA-plot of differentiated adipocytes highlights the 909 genes that are up in human and the 1035 genes that are down in human at a FDR-adjusted p-value of < 0.05.

Table 4.1 Differentiated adipocyte and white adipose tissue samples

Adipose Stromal Cell Samples

Species Identifier Sex Age Source

Human AG19304 Male 35 Coriell Institute for Biomedical Research Human AG20471 Male 34 Coriell Institute for Biomedical Research Human L040903 Male 39 Zen-Bio, Inc.

Chimpanzee S008396 Male ~22 Coriell Institute for Biomedical Research

White Adipose Tissue Samples

Species Identifier Sex Age Source

Human 602 Male 27 NICHD Brain and Tissue Bank Human 1104 Male 35 NICHD Brain and Tissue Bank Human 1134 Male 41 NICHD Brain and Tissue Bank Human 1442 Male 22 NICHD Brain and Tissue Bank Human B401298 Male 41 BioChain

Chimpanzee 4X0327 Male 37 Southwest National Primate Research Center Chimpanzee 4X0499 Male 23 Southwest National Primate Research Center Chimpanzee 4X0505 Male 34 Southwest National Primate Research Center Chimpanzee 4X0519 Male 20 Southwest National Primate Research Center Chimpanzee 95A014 Male 13 New Iberia Research Center

Macaque 22515 Male 9 Oregon National Primate Research Center Macaque 22646 Male 9 Oregon National Primate Research Center Macaque A08298 Male 5 New England Primate Research Center

Table 4.2 Illumina’s HiSeq statistics for the differentiated adipocytes

Species Human Human Human Chimpanzee Identifier AG19304 AG20471 L040903 S008396 d14 d14 d14 d14 Type Adipocyte Adipocyte Adipocyte Adipocyte Total Reads 45265315 47015397 53483585 53852567 TopHat Mapped 43259564 44931187 51229736 50545452 Counts (H-P) 12085917 11815431 13757235 14176504

Species Human Human Human Chimpanzee Identifier AG19304 AG20471 L040903 S008396 d14 d14 d14 d14 Adipocyte Adipocyte Adipocyte Adipocyte Type Oleic Acid Oleic Acid Oleic Acid Oleic Acid Total Reads 45115929 60816534 55300272 51668491 TopHat Mapped 42471020 58617397 53409701 49306972 Counts (H-P) 11592171 16186008 15107531 13951246

Species Human Human Human Chimpanzee Identifier AG19304 AG20471 L040903 S008396 d14 d14 d14 d14 Adipocyte Adipocyte Adipocyte Adipocyte Type Linoleic Acid Linoleic Acid Linoleic Acid Linoleic Acid Total Reads 53758300 41495344 52054928 50860938 TopHat Mapped 51999640 40139983 50338524 48563153 Counts (H-P) 14646098 11051444 14422643 13771696

100

Within this set, we found consistent indicators of elevated expression in the initial steps of fatty acid metabolism in humans (Figure 4.2). Fatty acids must first cross the adipocyte membrane through dedicated fatty acid transporters. Two genes encoding transporters specific for long-chain fatty acids (LCFA), SLC27A1 and SLC27A6, are expressed significantly higher in humans relative to chimpanzee (Figure 4.2) (Stahl

2004, Watkins et al 2006). After entering the cell, a fatty acid is bound by a member of the fatty acid binding protein (FABP) family (Storch and Corsico 2008).

Two family members, FABP5 and FABP7, are expressed at significantly higher levels in humans with FABP5 shows preference for 18C LCFAs (Storch and Corsico

2008) (Figure 4.2). In order to trap fatty acids inside the adipocyte, uptake is coupled to coenzyme A activation by acyl-CoA synthetase (encoded by ACS) (Watkins et al 2006).

Three family members encoding proteins specific for activating LCFAs (ACSL1, ACSL4, and ACSL5), are significantly higher in human adipocytes, whereas ACSF2, which is significantly lower and prefers medium-chain fatty acid substrates (Watkins et al 2006)

(Figure 4.2). The LCFA transporters mentioned above SLC27A1 and SLC27A6, also serve to activate very long chain fatty acids (Watkins et al 2006).

101

TAG DGAT

CHPT1 DAG

PPAP2B PA

AQPx MOGAT1 glycerol DGAT2 MAG glycerol DAG TAG HSL MGLL

ABHD5 PNPLA3 CPTIA ACBP ACSx O

CoA FABPx long-chain fatty acid O acetyl-CoA

HO SLC27Ax malonyl-CoA O HO CM

O ACACA HO FASN VLDL LPL SLC2Ax

CD36 O

glucose

ALB O HO long-chain fatty acid

Figure 4.2 Schematic of an adipocyte and key reactions in lipid metabolism

102

Figure 4.2) Colored enzymes denote significant difference between humans and chimpanzee adipocytes (correct p-value < 0.05) with red indicating higher expression in humans and blue indicating lower expression in humans. Genes that are hashed signify this difference is only present in the fatty acid challenge condition.

103

These results indicate that in a controlled environment, human adipocytes are genetically programmed to transport, bind, and activate more LCFA than chimpanzee adipocytes through increased expression of genes encoding proteins critical for the initial steps of their processing. These changes in LCFA lipid metabolism may represent an evolutionary response to the increase in animal consumption that began ~2 mya

(Bunn and Kroll 1986, Sponheimer and Lee-Thorp 1999, Ungar et al 2006). Indeed, lipid intake estimates for wild chimpanzees (2.5%), Paleolithic humans (8.5%), and contemporary humans (15%), indicate that lipid consumption not only increased during human evolution but also continues to distinguish humans from chimpanzees today

(Conklin-Brittain et al 2002). We reasoned that quantitative differences in bulk lipid processing is unlikely to represent a complete picture, however, given differences in the lipid composition of human and chimpanzee diets.

Therefore, we investigated how adipocytes respond to dietary lipids representing major dietary shifts toward increased meat and grain consumption. Analysis of a contemporary antelope, a prominent animal source during the Paleolithic era (Bunn and

Kroll 1986) reveals that the LCFA oleic acid (C18:1) represents the highest proportion of any fatty acid in marrow (49.49%) and muscle (20.66%) (Cordain et al 2002). Another fatty acid of particular interest during human history is the long-chain essential omega-6 fatty acid, linoleic acid (C18:2 n-6). The consumption of linoleic acid increased dramatically

104

following the cultivation of cereal grains and more recently with the invention of the screw-press and industrial production of vegetable oils (Simopoulos 1999). We separately challenged human and chimpanzee adipocytes with oleic acid and linoleic acid and profiled their transcriptional responses (Figure 4.3). When comparing these twelve transcriptomes by multidimensional scaling, the first axis separates the two species and the second distinguishes lipid concentration (Figure 4.3). Both kinds of differences are likely to be genetically based as they are manifest under controlled environmental conditions. Linoleic acid elicited a more substantial transcriptomic response (1101 specific genes) than oleic acid (143 specific genes) in human adipocytes, which is perhaps reflective of their distinct evolutionary histories in humans (Figure 4.4).

105

Figure 4.3 Multidimensional scaling plot that distinguishes human and chimpanzee adipocytes in different challenged conditions

Figure 4.3) Multidimensional scaling plot of Euclidean distances among the twelve differentiated adipocyte transcriptomes investigated in this study. The first and second dimensions explain 68.08% and 12.70% of the distance between samples.

106

Figure 4.4 Venn-diagram of human differentiated adipocytes challenged with linoleic and oleic acid.

Figure 4.4) Only genes that were significantly different between the challenge the control state in humans were used (FDR-adjusted p-value < 0.05).

107

Comparing these responses among species, several components of de novo lipid synthesis show elevated expression in humans specifically under fatty acid challenge conditions (Figure 4.2). The sugar transporters SLC2A1 and SLC2A5 are expressed at significantly higher levels in humans (Figure 4.2). Acetyl-CoA produced in the mitochondria from the transported sugar must be relocated to the cytosol through the citrate shuttle, before fatty acid synthesis can begin. Genes encoding three key enzymes in this process (PC, ME3, and ACLY) show significantly elevated expression in human adipocytes under both fatty acid challenges (Figure 4.2). Additional evidence of elevated rates of LCFA synthesis in human adipocytes under fatty acid challenge conditions comes from upregulation in humans of ACACA, the gene encoding the rate-limiting enzyme in fatty acid synthesis that converts acetyl-CoA to malonyl-CoA (Figure 4.2). The presumptive presence of additional malonyl-CoA in human adipocytes is consistent with the observed decreased expression of CPT1 in humans, as this gene is specifically inhibited by malonyl-CoA (Figure 4.2). Finally, FASN, which encodes the enzyme that converts chain acetyl-CoA and malonyl-CoA into LCFAs is expressed at higher levels in human relative to chimpanzee adipocytes under all conditions we assayed (Figure 4.2).

These results indicate that in the presence of linoleic and oleic acid, human adipocytes synthesize their own LCFA at a higher level than chimpanzee adipocytes, in spite of the presence of additional LCFA in the cell culture media.

108

Based on the elevated expression of genes involved in the intake of LCFA at rest and the production of LCFA during fatty acid challenges, we anticipated increased fatty acid storage in human adipocytes. Upon examining lipid droplet size, however, we found that chimpanzee adipocytes contain significantly larger droplets and greater overall lipid droplet volume across all conditions relative to humans (Figure 4.5 and Figure 4.6). This apparent contradiction becomes less puzzling upon examination of other steps in lipid metabolism. Genes encoding the rate-limiting lipolytic enzymes responsible for the breakdown of triacylglycerols (TAG) stored in the lipid droplets into diaclygylcerols

(DAG) and acyl-CoA are expressed at significantly higher levels in human adipocytes:

PNPLA2 under all conditions and PNPLA3 specifically under both lipid challenges (Figure

4.2) (Miyoshi et al 2008). Interestingly, PNPLA2 directly increases basal lipolysis rates in adipocytes, which decreases lipid droplet size and increases fatty acid release (Miyoshi et al 2008). These results suggest that human adipocytes maintain a large pool of DAG and fatty-acyl CoA that extends beyond catabolic processes. Support for this inference come from MOGAT1, which encodes the anabolic enzyme responsible for synthesizing DAG, and is expressed at significantly higher levels in human adipocytes across all conditions

(Figure 4.2). In addition, DGAT2, which encodes the anabolic enzyme that catalyzes the terminal and committed step in TAG synthesis using available DAG and acyl-CoA substrates, shows significantly lower expression in human adipocytes (Figure 4.2).

109

Figure 4.5 Phenotypic differences between human and chimpanzee differentiated adipocytes across several fatty acid conditions

Figure 4.5) Brightfield images of differentiated adipocytes (d14) grown in standard conditions as well as a linoleic and oleic acid challenge.

110

Figure 4.6 Distribution of the approximate volume of individual lipid droplets in each condition across species Figure 4.6) Each within condition comparison between species is significant (p < 0.0001)

by a t-test. Red is the average of three humans and blue is the chimpanzee.

111

These results suggest that an elevated DAG and acyl-CoA pools are maintained through several independent mechanisms in human adipocytes: including the breakdown of TAG in lipid droplets, increased DAG production in the endoplasmic reticulum, and decreased TAG production in the endoplasmic reticulum. These events parallel the lipid droplet differences between species across all adipocyte conditions, and may explain consistently smaller lipid droplets in human adipocytes, despite increased fatty acid intake, translocation, and de novo synthesis. Taken together, the differences in lipid droplets and underlying gene expression indicate that, relative to chimpanzee adipocytes, human adipocytes are more preoccupied with maintaining higher concentrations of DAG and LCFA than storing TAG in lipid droplets. Given differences in diet between the two species and the gene expression differences we observed, it seems unlikely that these are simply quantitative differences in generic pools of DAG and fatty acids. Instead, they are likely to be enriched in human adipocytes for specific LCFAs, namely palmitic (C16:0), oleic (C18:1), and linoleic (C18:2 n-6) acids. DAGs in the form of phospholipids comprise a major component of cell membranes and act as precursors for signaling molecules such as prostaglandins (Carrasco and Merida 2007). CHPT1 and CEPT1 encode enzymes responsible for synthesizing the phospholipids phosphatidylethanolamine (PE) and phosphatidylcholine (PC) from DAGs (Carrasco and Merida 2007). Both of these enzymes are expressed at significantly higher levels in human adipocytes across all conditions

112

(Figure 4.2). The composition of PE and PC are of consequence as these can include C16:0,

C18:1, and C18:2 n-6.

To further understand the genetics of adipocytes it is important place them in context. Adipocytes do not function by themselves, but rather within white adipose tissue

(WAT), where they operate in concert with endothelial cells, macrophages, and fibroblasts to mediate bioenergetic processes (Fruhbeck 2007). We, therefore, profiled the transcriptome of subcutaneous WAT from human, chimpanzee, and rhesus macaque using the SOLiD platform (Tables 4.1 and 4.3). Data from the latter species provides the ability to infer increases or decreases in expression specific to the human phylogenetic branch. The major axes in a multidimensional scaling plot separate the three species by phylogenetic distance, the first splitting the two great apes from the old world monkey, and the second distinguishes humans from the other primate species (Figure 4.7).

Correlations in gene expression between adipocytes and WAT are 0.6933 and 0.7150 for humans and chimpanzees respectively (Figure 4.8). These values are quite high, given that the comparison involves a single cell type versus a complex tissue, and that expression measurements were made on two different platforms. These correlation results convincingly demarcate adipocytes as the cell type contributing the major signal in

WAT transcriptomes (Figure 4.8).

113

When using the macaque to polarize, the gene fatty acid desaturase 2 (FADS2) is of special interest as it shows a signal of human branch-specific increased expression

(Hsap-Ptro p-value = 1.74E-02 and Hsap-Mmul p-value = 3.21E-14). This is especially interesting when connecting back to the differentiated human adipocyte data as FADS2 functions in the most enriched KEGG pathway ‘biosynthesis of unsaturated fatty acids’

(p-value = 0.001) (Figure 4.1b, genes in red). This pathway is responsible for converting the essential omega fatty acids linoleic and linolenic into important building blocks.

FADS2 and FADS1 encode the rate limiting enzymes responsible for the desaturation steps in this pathway (Nakamura 2004). The magnitude of the differences is substantial in

WAT between humans and chimpanzee (FADS1 FC = 2.38, FADS2 FC = 2.85).

114

Table 4.3 ABI’s SOLiD 4 statistics for the white adipose tissue

Species Human Human Human Human Human Identifier 602 B401298 1442 1104 1134 Total Reads 60667107 56247821 61640355 50203856 57795847 TopHat Mapped 21868831 20145397 13775357 9118214 27724055 Counts (H-P) 5706987 6481540 4125130 2706325 10007423 Counts (H-P-M) 5227770 5914254 3767303 2492473 9201969

Species Chimpanzee Chimpanzee Chimpanzee Chimpanzee Chimpanzee Identifier 4X0519 4X0327 4X0505 95A014 4X0499 Total Reads 79514618 74704247 30079174 87178307 53271125 TopHat Mapped 33625972 22667607 14310622 49140636 26758163 Counts (H-P) 7934687 3692934 1428564 2727544 5718824 Counts (H-P-M) 7357295 3427427 1322308 2524611 5179106

Species Macaque Macaque Macaque Identifier 22515 A08298 22646 Total Reads 52051970 64643778 80366305 TopHat Mapped 23179375 19504589 24977825 Counts (H-P) 2657503 2935875 3192472

115

Figure 4.7 Multidimensional scaling plot that distinguishes human, chimpanzee, and macaque white adipose tissue

Figure 4.7) Multidimensional scaling plot of Euclidean distances among the thirteen white adipose tissue transcriptomes investigated in this study. The first and second dimensions explain 41.59% and 20.87% of the distance between samples.

116

Figure 4.8 Comparing differentiated adipocyte and white adipose tissue transcriptomes

Figure 4.8) Pearson correlation of FPKM comparing differentiated adipocytes to white adipose tissue for humans and chimpanzees.

117

The unsaturated fatty acid synthesis pathway transforms the essential fatty acids linolenic (C18:3 n-3) and linoleic (C18:2 n-6) acid into key metabolites, including docosahexaenoic acid (DHA) and arachindonic acid (AA) (Figure 4.9). Both are indispensable for human health. Omega-3 DHA is present at high levels in the grey matter of the cerebral cortex, where it helps maintain structure and provides neuroprotective properties (Bradbury 2011). On the other hand, omega-6 AA is present in white matter in the brain and elsewhere in the body, where it functions in growth and reproduction (Yehunda 2003).

Although these are essential for health and development, both the level and proportion of omega-3 and omega-6 fatty acids can influence disease risk. The two classes of fatty acids compete for FADS2 and FADS1 enzyme activity (Lands 2012) and are thought to function best when present in approximately equal amounts (Kang et al

2001). We hypothesize that increased expression of key enzymes involved in the metabolism of unsaturated fatty acids during human origins, had profound consequences for human evolution and health (Lawrence 2010) (Figure 4.9).

118

Figure 4.9 The biosynthesis of unsaturated fatty acids pathway is important during human evolution

119

Figure 4.9) Omega-3 and omega-6 fatty acids compete for the same pathway of genes.

The expression of these genes are represented as the human/chimpanzee fold change from in vitro and in vivo experiments as indicated by the color bar. Expression of FADS2 was significantly higher in humans compared to both nonhuman primates (p-value <

0.05), highlighted here by an additional box. The dietary omega-6:omega-3 ratio is represented by the thickness of each arrow, the scale graphic and is displayed numerically. Key: A: adipocyte, A-LA: adipocyte challenged with linoleic acid, WAT: white adipose tissue.

120

Fatty acids in hominid diets likely contained a fairly balanced omega-3:omega-6 fatty acid ratio of ~1:0.7 (Simopoulos 2011) (Figure 4.9). The elevated expression of

FADS1 and FADS2 that we observe in modern humans is consistent with adaptation to redirect essential fatty acids into the biosynthesis of unsaturated fatty acids pathway.

The increased expression of these two rate-limiting steps in the pathway may have a common genetic basis. They are located only 11 kb apart, are divergently transcribed from a shared cis-regulatory region, and two modern human haplotypes are associated with coordinately distinct expression (Ameur et al 2012). Segregating variation the

FADS1/FADS2 regulatory region affects the ability to convert linolenic and linoleic acids into downstream products such as DHA and AA, with the human specific haplotype allowing for additional synthesis of essential LCFA (Ameur et al 2012). Taken together the increased production of essential structural components (DHA and AA) in paleolithic humans, due to the upregulation of rate-limiting enzymes, may have allowed for brain expansion in early Homo (Yehunda 2003).

Following the agricultural revolution, the dietary ratio of omega-3:omega-6 fatty acids rapidly shifted, reaching 1:20 in some modern human societies (Simopoulous

2011). This dramatic change in composition occurred because cereal grains and vegetable oils are rich in omega-6 fatty acids. With this qualitative change in fatty acid intake, elevated FADS1 and FADS2 expression may have become problematic, since the

121

two enzymes catalyze parallel steps and metabolites compete for their activity. Mass action alone would reduce the production of omega-3 compounds and increase omega-6 products like AA (Figure 4.9).

After AA is produced it can be processed by the PTGS (COX) family, which encodes the rate-limiting enzymes that convert omega fatty acids derivatives into eicosanoids. When converted to eicosanoids, the products from omega-3s serve as anti- inflammatory signals, while those from omega-6s are typically pro-inflammatory signals

(Stables and Gilroy 2011). There are two main COX enzymes: COX1, which is constitutive and is responsible for basal levels of conversion and COX2, which is inducible and acts as an acute responder when the immune system is facing a specific challenge. Interestingly, only the inducible COX2 is significantly different and higher in humans (FDR-adjusted p-value = 0.013 and FC = 4.6x higher in humans) (Figure 4.10).

We found that several known regulators of COX2 are expressed at significantly elevated levels in human WAT, as are several of its known downstream targets (Figure 4.10). The large set of inhibitors that have been developed as pharmacological agents against COX2 activity alone speaks to its importance in modulating inflammation (Fitzgerald 2003).

With COX2 clearly being activated, an overabundance of AA would lead to a substantial inflammatory response, due to high levels of prostaglandins according to this model

(Figure 4.10).

122

Figure 4.10 COX2 and it’s interactors are higher in human white adipose tissue compared to the chimpanzee

Figure 4.10. Differentially expressed (WAT human-chimp FDR-adjusted p-value <0.05) upstream regulators (direct) and downstream targets (indirect) of PTGS2. Expression of these genes is represented as the human/chimpanzee fold change with darker red indicting higher expression in humans and darker blue indicating lower in humans. The thickness of the arrows, the graphical scale, and numerical ratios, represents an estimated omega-6:omega-3 ratio of modern humans consuming a Western diet. 123

We reasoned that increased omega-6 arachidonic acid production should produce a pro-inflammatory signal in the transcriptome. Consistent with this hypothesis, we found that GO categories related to ‘positive regulation of acute phase inflammatory response’ are enriched in genes that are expressed significantly higher in human compared to chimpanzee WAT (FDR-adjusted p-value <5%) (Figure 4.11).

Indeed, of all the over represented categories those related or plausibly related to inflammatory response dominate the genes highly expressed in humans (Figure 4.11).

These results are particularly relevant in light of a growing appreciation for the role of inflammation in the onset and progression of multiple human pathologies (Hotamisligil and Erbay 2008). Interestingly, some of these diseases show an elevated incidence or unique manifestations in humans, and are likely consequences of either genetic differences or dietary changes unique to our species (Varki and Altheide 2005). Type-2 diabetes is an especially interesting example, as inflammation and high levels of DAG are thought to contribute to insulin resistance (Erion and Shulman 2010). Our discovery of evolutionarily recent increases in long chain fatty acid processing (Figure 4.2), more specifically that of essential fatty acids (Figure 4.9) as well as an inflammatory response signature in human WAT samples (Figure 4.11), may provide insights into the molecular basis for a variety of human diseases.

124

Figure 4.11 Signature of inflammation in human white adipose tissue

Figure 4.11) Gene ontology enrichments in the 286 significantly higher expressed genes in human white adipose tissue (WAT) as compared to chimpanzee (FDR-adjusted p- value < 0.05) were analyzed using GOrilla and visualized using ReVIGo. Boxes shown are significantly enriched (p-value <0.05), with the size of each box corresponding to the

–log10 p-value.

125

Although only a few medical conditions, such a sickle cell anemia, are widely understood to have a clear and understandable basis in our evolutionary history, the actual number is almost certainly much larger (Domazet-Loso and Tautz 2008). The vast majority of human diseases have a subtler genetic and ecological basis than sickle cell anemia, making it challenging to identify the underlying evolutionary changes in molecular function. The ability to apply functional genomic methods to both in vitro cell lines and in vivo tissue samples from humans and great apes -- and especially the ability to carry out controlled experimental manipulations with the former -- opens the door to investigating the molecular and genetic basis for the evolutionary origins of many additional human diseases.

In this study, we identify several gene expression differences that likely confer lipid metabolism changes unique to humans. Prominent among these are elevated fatty acid uptake, binding and activation of LCFA as well as increased de novo fatty acid synthesis in the presence of high levels of external fatty acids (Figure 4.2). There is a preoccupation of human adipocytes to maintain DAG pools, which could be used in phospholipid or prostaglandin production. Furthermore, both in vitro and in vivo studies converged on an elevated expression of genes involved in the processing of the omega-6 and omega-3 EFA (Figure 4.9). A subsequent shift in dietary intake of omega-6 fatty acids likely resulted in increased levels of their derivatives such as prostaglandins

126

through increased COX2 processing, which could account for the signatures of inflammation we see in the human white adipose tissue samples. These changes in lipid metabolism are a significant genomic legacy of our evolutionary history that continues to shape disease risk and severity today.

127

4.3 Materials and Methods

4.3.1 Samples

Male subcutaneous white adipose tissue samples were obtained from five humans (602, 1104, 1134, 1442, B401298), five chimpanzees (4X0327, 4X0499, 4X0505,

4X0519, 95A014) and three macaques (22515, 22646, A08298). Adult male adipose stromal cells (AG19304, AG20471, L040903) were obtained from three humans and one chimpanzee (S008396). No primate was sacrificed for the purpose of this research.

These samples were obtained from the National Institute for Child Health and Human

Development Brain and Tissue Bank (602, 1104, 1134, 1442), BioChain (B401298),

Southwest National Primate Research Center (4X0327, 4X0499, 4X0505, 4X0519), New

Iberia Research Center (95A014), Oregon National Primate Research Center (22515,

22646), New England Primate Research Center (A08298), Coriell Institute for Biomedical

Research (AG19304, AG20471, S008396) and ZenBio (L040903). Details regarding these samples can be found in Table 4.1.

4.3.2 Cell culture and tissue processing

The adipose stromal cells were recovered from cryofreeze and immediately placed in MesenPro RSTM Medium (Invitrogen) supplemented with 200 mM L-

Glutamine (Invitrogen) and 1% Penicillin-Streptomycin Solution (Invitrogen) and incubated at 37oC, 5% CO2. When the cells had expanded to approximately 70%

128

confluence, they were detached using TryPLE Express (Invitrogen) and plated at 40,625 cells/cm2 in 6-well plates (Corningstar) for differentiation. ASC were cultured until confluent (24-48 hours) and then differentiated into adipocytes using ZenBio’s

Differentiation (DM-2-PRF) and Maintenance (AM-1-PRF) media following standard methods (ZBM0001.03). For the challenge experiments, oleic acid (Sigma O3008) or linoleic acid (Sigma L9530) was added to the AM-1-PRF media to a final concentration of

250 µM. Adipocytes were considered mature after 14 days in culture as verified by Oil

Red O staining. The white adipose tissue samples were flash frozen, shipped to Duke

University and stored at -80oC until RNA extraction.

4.3.3 Next generation sequencing and processing

Total RNA was isolated from adipocytes and white adipose tissue using the miRNeasy Kit with the optional DNaseI step. RNA quality was assessed using the

Experion (BioRad). All sequencing was conducted at the Duke Institute for Genome

Sciences & Policy’s Genome Sequencing & Analysis Core Resource. 1 ug of differentiated adipocyte total RNA was used in Illumnia TruSeq SBS library construction, with each sample having a unique barcode. These went through cluster generation, were sequenced using Illumina’s HiSeq TM 50 bp paired end technology with four libraries multiplexed per lane. On average, 51 million reads were obtained from each library with 49 million mapping to the respective genome (Table 4.2). 10 ug of

129

white adipose tissue total RNA was poly(A) selected and libraries were constructed using the SOLiD Whole Transcriptome Analysis Kit; each sample was individually barcoded. These went through cluster generation and were sequenced using ABI’s

SOLiD4 50bp x 35bp paired end technology with eight libraries multiplexed per flow cell. On average, 62 million reads were obtained from each library with 26 million mapping to the respective genome (Table 4.3). All reads were mapped to the appropriate genome (hg19, panTro3 or RheMac2) with TopHat v1.4.1 using default settings (Trapnell et al 2009). Mapped reads were counted with htseq-count 0.5.1p1 using the settings union and strandedness. Orthologous gene models for were constructed using the publically available Primate Exon Orthology Database version 2

(http://giladlab.uchicago.edu/orthoExon/). This collection was further filtered to eliminate genes with unclear homologies. We removed genes that had a Human-

Chimpanzee or Human-Macaque ensembl homology of one2many and many2many as well as the ribosomal families, RPL, RPS, MRPL and MRPS. Additional genes were removed where the database chromosome assignment did not match the Ensembl assignment as well as genes where the same HGNC name has multiple Ensembl gene ID pairings. To reduce spurious findings we removed low expressed genes where the library had less than one count per ten million reads for the HiSeq data and less than one count for the SOLiD4 data before normalization. As such, the number of genes

130

analyzed changes depending on the samples used in the comparison: differentiated adipocytes – 11,188; differentiated adipocytes with fatty acid challenges – 10,833; human and chimpanzee white adipose tissue – 9,589; human and chimpanzee white adipose tissue and adipocyte comparisons – 8,832; and human, chimpanzee and macaque white adipose tissue – 6,759. The final counts were normalized using a negative binomial distribution by estimating the common and then tagwise dispersion using edgeR 1.6.0

(Robinson et al 2010). Significance was established using an extact test and FDR corrections for multiple comparisons were calculated using the Benjamini-Hochberg method, built into edgeR (Benjamini and Hochberg 1995). Counts were converted to fragments per kilobase per million (FPKM).

4.4.4 Enrichments and analyses

The DAVID (v6.7) KEGG Pathway tool (Huang et al 2009) was used to investigate pathway enrichment in the 909 highly expressed genes in human adipocytes as compared to chimpanzee adipocytes (FDR-adjusted p-value <0.05) using the default settings against all genes in the dataset as background. GOrilla was used to investigate gene ontology (GO) enrichment in these datasets. In all cases, the ‘two unranked list of genes’ tool was used with default settings against all genes in the dataset as background, and ReVIGo (April 2012) was used to plot their similarities with the size of the box corresponding to the –log10 p-value of the biological enrichment. GOrilla was

131

specifically used to identify enrichments in the 1506 genes that changed expression in both the OA and LA challenge conditions (FDR-adjusted p-value < 0.05) and the 286 highly expressed genes in human WAT as compared to chimpanzee (FDR-adjusted p- value < 0.05). Ingenuity (IPA) Systems (Summer Release 2012) was used to identify expression changes of direct upstream regulators and direct/indirect downstream targets of COX2 associated genes, with a specific focus on genes that have FDR-adjusted p- value < 0.05.

4.3.5 Staining and imaging

Black and white images were taken with the Zeiss Axio Observer A1 inverted stand microscope with a Zeiss HBO arc lamp and power supply using a Hamamatsu

Orca ER digital camera. Color images were taken with a Leica DM IRB microscope using a Zeiss AxioCam ICc1 digital camera. Mature differentiated adipocytes (day14) were stained for lipid content using the Oil Red O Stain Kit and associated protocol

(ScyTek ORK-1). Post-processing was done using ImageJ for batches of similar images.

Adipocyte lipid droplets were analyzed using 20x black and white images. A method similar to Or-Tzadikario et al 2010 was conducted using ImageJ instead of MATLAB.

Fifty adipocytes from each treatment and cell line combination were analyzed. Briefly, adipocyte cell boundaries were outlined and binary images were used to threshold the lipid droplets within cells. Like Or-Tzadikario et al 2010, only cells in the field of vision

132

that had clearly defined cell boundaries and had no evidence of lipid droplet multilayers were counted. Additionally, lipid droplets were counted if they were over 3 µm2 in area and had a minimum circularity of 0.3 to ensure that debris was not counted. Using the size of individual lipid droplets (µm2), the approximate radius and volume were determined for each droplet. A t-test was used to determine significance for the size of lipid droplets present in each cell between species for each condition and across species.

All species comparison in each challenge had a p-value <0.0001.

133

4.4 Acknowledgements

We thank the Duke Institute for Genome Sciences & Policy’s Genome Sequencing

& Analysis Core Resource, especially Olivier Fedrigo, Nick Hoang, and Fangfei Ye for library preparation and sequencing assistance. Additionally, we thank Courtney Babbitt library construction assistance. We thank the Light Microscopy Core Facility at Duke

University, especially Sam Johnson and Yasheng Gao for microscopy training and imaging support. We thank the team at ZenBio, especially Ben Buehrer, Jim Nicoll and

Renee Lea-Currie for advice on adipocyte differentiation and a human ASC line. We thank the Coriell Institute for Medical Research, especially Bernie Goldstein, for assistance with ordering chimpanzee and human ASC lines. For the in vivo samples, we thank the Southwest National Primate Research Center, Yerkes Regional Primate

Research Center, New England Primate Research Center, BioChain and the NICHD

Brain and Tissue Bank.

This work was supported by Duke Primate Genomics Initiative Fellowship to

L.W.P.; a Wenner Gren Dissertation Fieldwork Grant to L.W.P.; and HOMINID NSF

Grant NSF-BCS-08-27552 to G.A.W.

134

5. Insights from a chimpanzee adipose stromal cell population: opportunities for adult stem cells to expand primate functional genomics

My contribution to this work:

1) Conceived and designed experiments.

2) Obtained funding.

3) Performed all cell culture work, imaging experiments and mRNA

extractions.

4) Conducted data analyses.

5) Co-wrote manuscript.

135

5.1 Abstract

Comparisons between closely related primates, such as the human and chimpanzee, are essential for understanding traits unique to each species. However, linking important phenotypic differences to underlying molecular changes is often challenging. The ability to generate, differentiate, and profile adult stem cells provides a powerful, but underutilized opportunity to investigate the molecular basis for trait differences between species, within specific cell types, in a controlled environment.

Here we profile adipose stromal cells (ASCs) from Clint, the chimpanzee whose genome was first sequenced. Using imaging and RNA-Seq, we compare the chimpanzee ASCs to three comparable human cell lines. Consistent with previous studies on ASCs in humans, the chimpanzee cells have fibroblast-like morphology and express components of the extracellular matrix at high levels. Differentially expressed genes are enriched for distinct functional classes between species: immunity and protein processing are higher in chimpanzees, while cell cycle and DNA processing are higher in humans. Adult stem cells provide a powerful and practical way to investigate the profound disease susceptibility differences that exist between humans and our closest living relatives, as well as informing chimpanzee conservation efforts. By allowing for experimental perturbations in relevant cell types, adult stem cells promise to complement classic comparative primate genomics in vivo research.

136

5.2 Introduction

The advent of next generation sequencing has resulted in an explosion of exploratory genomic studies that have identified many candidate genes waiting further examination. For a number of species, such as the mouse, fruit fly and zebrafish, these candidates can be directly investigated in the living animal using transgenic technologies. However, for some organisms -- including endangered species, animals with husbandry challenges, and those with ethical concerns -- investigating the evolution of gene function requires the development of additional tools. Adult stem cells allow for experiments in multiple cell types, physiological status and disease states in vitro, providing the ability to link genomic data and organismal traits. Two species in particular that will benefit from this approach are chimpanzees and humans.

Sequencing the chimpanzee and human genomes allowed for new layers of analysis, including expression (Enard et al 2002, Khaitovich et al 2005), alternative splicing

(Blekhman et al 2010), non-coding transcripts (Babbitt et al 2010), histone modifications

(Cain et al 2011), methylation (Pai et al 2011), and DNaseI hypersensitivity (Shibata et al

2012). These comparative genomic studies have provided important initial insights, greatly furthering our understanding of functional genomic differences relevant to the human and chimpanzee condition. All, however, were based on in vivo tissue samples

137

that are difficult to obtain, composed of multiple cell types, and do not control for disparate environmental influences.

Moving forward, detailed follow-up studies are necessary to unveil specific molecular mechanisms influencing trait differences between species. These experiments will need to address the lack of control over environmental variables and the presence of a heterogeneous cellular milieu. Working with cells in culture can overcome both challenges. A pioneering study by Barreiro and colleagues investigated immune response to a lipopolysaccharide challenge in human, chimpanzee, and macaque monocytes in cell culture (Barreiro et al 2010). Currently, this type of investigation is limited to commercially available fibroblast and lymphoblast cell lines. Because very few other cell lines exist from nonhuman primates, it is difficult to address phenotypes outside of the epithelium and bone marrow. In particular, differentiated adipocytes are an appropriate model to investigate phenotypes such as: physiological changes in white adipose tissue, differences in fatty acid processing, and Type 2 Diabetes.

Adult stem cells offer an efficient approach to overcoming the limited availability of cell lines from nonhuman primates. A single adult stem cell line can differentiate into many different cell types, opening the door to experimental analysis of fundamental questions regarding human origins. In a seminal publication, Cheng and colleagues isolated and profiled adult progenitor cells from the dental pulp of an adult female

138

chimpanzee, demonstrating that these cells can differentiate into osteoblasts, adipocytes, and chondrocytes (Cheng et al 2008). Another type of adult stem cell, the adipose derived stromal cell (ASC), is obtained by mechanically and enzymatically processing white adipose tissue (Zuk et al 2001). In cell culture, ASCs are capable of differentiating into adipocytes, osteoblasts, chondrocytes, hepatocytes, myocytes (smooth, skeletal and cardiac), endothelial cells, neural cells, epithelial cells and pancreatic B-cells (Cawthorn et al 2012). Here, we profile the only publically available population of chimpanzee

ASCs and compare it to three human ASC lines. Interestingly, the chimpanzee ASCs were derived from Clint, the first chimpanzee whose genome was sequenced (The

Chimpanzee Sequencing and Analysis Consortium 2005). To our knowledge, this is the first investigation into the biology of chimpanzee adipose stromal cells. In addition to providing insights into fundamental differences between chimpanzee and human adult stem cells, our results demonstrate the need for additional public resources for this kind of research.

139

5.3 Morphological and genetic characterization of chimpanzee ASCs

We visualized cultured chimpanzee ASCs using several different staining protocols (Figure 5.1 and Figure 5.2). Most chimpanzee ASCs adopt a fibroblast-like phenotype in vitro (Figure 5.1 and Figure 5.2), consistent with previous reports of human

140

Figure 5.1 10x images of chimpanzee ASCs in culture

Figure 5.1) (A) Fluorescence image depicting the nucleus (blue: DAPI) and the actin filaments (red: phallodian). (B) Brightfield image of ASCs at confluence before collection. (C) Brightfield image depicting the nucleus (blue: Mayer’s Hematoxylin) and lipid droplets (red: Oil Red O).

141

Figure 5.2 20x images of a chimpanzee and a representative human ASCs in culture

Figure 5.2) The top represents brightfield images of ASC at confluence before collection.

The bottom panel contains brightfield images depicting the nucleus (blue: Mayer’s

Hematoxylin) and lipid droplets (red: Oil Red O). Arrows indicate examples of chimpanzee ASCs that contain small lipid droplets. No droplets were seen in any of the human ASC lines.

142

ASCs (Zuk et al 2001) and the human ASCs profiled in this study (Figure 5.2, human data not shown). However, the chimpanzee’s ASCs are not uniform, displaying a range of sizes and shapes (Figure 5.1 and Figure 5.2). When grown to confluence, the chimpanzee ASCs migrate on top of one another and appear to exhibit lower levels of contact inhibition than the human ASCs (Figure 5.2). Decreased contact inhibition has been seen in other stem cell populations, such as embryonic stem cells (Burdon et al

2002). Interestingly, small lipid droplets are present in some of the chimpanzee ASCs, but were not seen in any of the human stromal cells profiled in this study (example

Figure 5.2b, see arrows).

To uncover fundamental properties of the chimpanzee ASC transcriptome, RNA extracted from the confluent chimpanzee stromal cells (Figure 5.1b and Figure 5.2a) was made into Illumina HiSeq RNA-Seq libraries. Approximately 48 million reads were mapped to the panTro3 chimpanzee genome (Table 5.1). The highest expressed genes in the chimpanzee ASC are overwhelmingly extracellular matrix (ECM) components

(Figure 5.3a). Cells of the connective tissue produce, organize, and degrade the ECM. In turn, the ECM provides organization, strength, and signaling mechanisms for cells of the connective tissue. The abundance of extracellular matrix genes we observe in chimpanzee ASCs is consistent with a previous study of human ASCs (Katz et al 2005).

Additionally, the top ten highest expressed chimpanzee genes were also

143

Table 5.1 Illumina’s HiSeq statistics for the ASC profiled in this study

Species Chimpanzee Human Human Human Identifier S008396 AG19304 AG20471 L040903 Total Reads 51354259 48213838 47716699 43307149 TopHat Mapped 48229125 45779949 45560720 41304174 Counts 15310086 14415828 15086585 12969130

144

Figure 5.3 Transcriptomic insights into chimpanzee ASCs

145

Figure 5.3) (A) Highest expressed genes using Fragments Per Kilobase of transcript per

Million mapped reads (FPKM). (B) KEGG pathway: ECM-receptor interaction.

Highlighted genes have an FPKM ≥ 100 and the connections that contain a complete

ECM-ligand and receptor pair are bolded. Immunoglobin superfamily not shown. found within the top fourteen highest expressed human genes in the current study

(Figure 5.3a). It is not surprising that collagen, the most abundant protein in mammals and primary source of strength and structure in the ECM, represents six of the top ten highest expressed genes (Figure 5.3a) (Alberts et al 2008).

146

We next sought to determine what is unique about the collection of highest expressed chimpanzee ASC genes (FPKM >= 100, n = 614). Using DAVID, we identified the ECM-receptor interaction KEGG pathway as the most enriched pathway when compared to all ASC expressed genes (p-value = 2.9E-10 and FDR-adjusted p-value =

4.1E-8) (Huang et al 2009). The hallmark of the ECM-receptor interaction pathway is the relationship between structural molecules, like collagen, and the α/β integrins (Figure

5.3b). Integrins are responsible for mediating a physical and chemical connection between the internal ASC cytoskeleton (actin shown in Figure 5.1b) and the external matrix environment. This is necessary for critical cell behaviors including proliferation, migration, adhesion, differentiation, and apoptosis (Alberts et al 2008). The specific α/β heterodimer dictates which ECM ligand(s) the integrin interacts with (Figure 5.3b). In our analysis, there were four complete integrin pairings among the highest expressed chimpanzee genes: fibronectin-α5β1, fibronectin-αVβ1, osteopontin-αVβ1, and osteopontin-α5β5 (Figure 5.3b). As cells of the connective tissue, much of the structure and function of ASCs are mediated by ECM-receptor interactions.

147

5.4 Chimpanzee ASC pluripotency and differentiation status

One of the hallmarks of stem cells is their pluripotency, or ability to differentiate into cell types present in all three germ layers. Although the ability of ASCs to self- renew and differentiate into ectodermal lineages in vivo has not been definitively established (reviewed in Cawthorn et al 2012), these cells can be used to investigate many different cell types in culture. We have successfully differentiated Clint’s ASCs into mature adipocytes in vitro using a cocktail of adipocyte differentiation media

(Figure 5.4a). A marker of mature adipocytes is the presence of lipid droplets, here defined by Oil Red O staining. Although this result does not confirm pluripotency for chimpanzee ASCs, it does demonstrate their ability to differentiate into mesodermal lineages.

148

Figure 5.4 Pluripotency insights into ASC

Figure 5.4) (A) Brightfield image of adipocytes after 14 days of differentiation depicting the nucleus (blue: Mayer’s Hematoxylin) and lipid droplets (red: Oil Red O).

(B) Schematic of ASC differentiation into adipocytes, modified from Cawthorn et al

2012. (C) Relative age of cell lines in this study measured by passage number and estimated population doubling level (ePDL).

149

ASCs and preadipocytes (a slightly more differentiated state) are occasionally conflated in the literature, as both are members of the white adipose tissue expansion continuum and share many of the same cell surface markers (Figure 5.4b) (Katz et al

2005, Cawthorn et al 2012). As mentioned earlier, Clint’s ASCs show evidence of lipid accumulations that were not detected in the human cells we profiled (Figure 5.2). These lipid droplets are substantially smaller than those found in Clint’s differentiated adipocytes (Figure 5.4a). Three possibilities could explain the presence of lipid droplets in the chimpanzee, but not the human ASCs: 1) the chimpanzee cells are further differentiated than the human cells, 2) an artifact was introduced during collection prior to receipt of the cell lines, or 3) these lipids represent a species-specific difference in

ASCs.

To investigate the differentiation state of these cell lines, we examined the expression of two transcriptional regulators that mark committed preadipocytes in white adipose tissue: PPARγ (upregulated in the chimpanzee ASCs, FDR-adjusted p- value = 0.0316, FC = 2.694) and Zfp467 (not expressed in the chimpanzee or human

ASCs) (Cawthorn et al 2012). The significant difference in expression for PPARγ, a transcription factor that controls lipid uptake, indicates that perhaps the population of chimpanzee cells is more differentiated than human ASCs. However, the absence of

Zfp467 expression in both species suggests that the story is more complicated. We also

150

examined a third gene, MMP3, a metalloprotease produced by committed preadipocytes

(Cawthorn et al 2012). This gene is expressed at much higher levels in human than chimpanzee ASCs, and shows the second greatest fold-difference between the two species transcriptome-wide (FDR-adjusted p-value = 5E-09, FC = 182..278). Based on these three incongruent expression signatures for a limited selection of markers, it remains unclear where the chimpanzee and human cells are on the white adipose tissue expansion continuum (Figure 5.4b). Another insight comes from the estimated population doubling level and passage number, which act as a proxy for cell age. This is critical as the length of time ASCs are in culture changes their immunophenotypes

(Mitchell et al 2006), increases senescence (Gruber et al 2012), and is inversely related to their pluripotency (Katz et al 2005, Wall et al 2007). For the cells used in this experiment, there are no substantial species differences between passage number and estimated population doubling level (Figure 5.4c).

A species-specific collection bias is improbable, as the adipose stromal cells were collected from at least three different institutions and harvested by different investigators. Despite widely known differences in ASC processing strategies, several groups have commented on the consistency in immunophenotype and molecular profiles of ASCs across studies (Katz et al 2005, Gimble et al 2007). Therefore, it is unlikely that the approach employed by the scientists and veterinarians that harvested

151

Clint’s cells was so radically different that they induced a phenotypic change in the ASC line.

There are several documented differences in adipose tissue derived mesenchymal stem cells between humans and the non-human primate Macaca mulatta.

Izadpanah and colleagues (2006) profiled three critical transcription factors involved in pluripotency and found that the macaque ASCs showed a marked decrease in Rex-1 mRNA as well as Oct-4 and Sox-2 protein levels at passage 20 compared to the humans.

Consistent with this finding, the human ASCs retained their adipogenic potential longer than the macaque (Izadpanah et al 2006). Although these results do not directly speak to differences in lipid droplet formation (no droplets were detected in any of the undifferentiated ASCs in that study) they do demonstrate differences in ASCs among primate species. The lipid droplet difference in ASCs reported in the current study could be indicative of biological differences between humans and chimpanzee ASCs

(Figure 5.2), but without more chimpanzee adult stem cell resources the nature of these differences remains unclear.

152

5.5 Transcriptomic differences between chimpanzee and human ASCs

In order to find genes that are differentially regulated between the human and chimpanzee, we next compared Clint’s ASC transcriptome to three human ASC transcriptomes. The human ASC samples had on average 44 million reads mapped hg19

(see Table 5.1 for details). 10,021 orthologs were analyzed between the humans and chimpanzee. As expected, the three human ASCs are more similar in expression to each other than any of them is to the chimpanzee ASC, with the major axis in a MDS plot clearly separating the species and explaining 69.12% of the distance between samples

(Figure 5.5a). Next, we sought to determine which genes distinguish Clint’s ASCs from the human ASC samples. At a 5% FDR (Benjamini and Hochberg 1995), 679 genes are expressed at significantly higher levels in chimpanzee ASCs (Figure 5.5b, red) and 486 genes are expressed at significantly higher levels in human ASCs (Figure 5.5b, blue).

To uncover functional differences between chimpanzee and human ASCs, we interrogated the red (chimpanzee higher) and blue (human higher) genes in Figure 5.5b using the PANTHER tools database (Mi et al 2005). These categorical enrichments include biological processes, molecular functions, and protein classes. Chimpanzee

ASCs have higher expression for genes involved in immunity (dark red) and protein processing (light red), whereas human ASCs have higher expression for genes involved in the cell cycle (dark blue) and DNA processing (light blue) (Figure 5.6). 153

Figure 5.5 Visualizing the normalized ASC transcriptomes

Figure 5.5) (A) Multidimensional scaling plot of Euclidean distances among the four transcriptomes investigated in this study. The first and second dimensions explain

69.12% and 18.73% of the distance between samples. (B) MA-plot where each dot represents a gene and those that are significantly differentially expressed at a FDR- adjusted p-value <0.05 are red (up in chimpanzee) or blue (up in human).

154

Figure 5.6 PANTHER gene categories enriched for differential expression by species.

155

Figure 5.6) The queried genes include those significantly higher in the chimpanzee

ASCs (red in Figure 5.2b) and significantly higher in the human ASCs (blue in Figure

5.2b). These were assessed against the background set of all 10,221 ASC genes in this study. Shown are the top five most significantly enriched categories for both human and chimpanzee, italicized categories are not statistically significant for the species they are enriched in. The sign next to the nominal p-value indicates whether the category is enriched (+) or depauperate (-) for the given species. (A) GO Molecular Function enrichments, (B) GO Biological Process enrichments, and (C) PANTHER Protein Class enrichments.

156

Strikingly, every one of the broader highlighted categories is distributed perfectly onto either the human or chimpanzee branch; for instance, all six cell cycle sub- categories are enriched on the human branch. The most significant enrichments for the chimpanzee are processes involved in the development and functioning of the immune system, which responds to potential invasive or internal threats (Figure 5.6) (Gene

Ontology Consortium 2000). Another complementary category that is enriched in the chimpanzee and significantly depauperate in humans is cytokine activity (Figure 5.6).

The chemokines, a class of cytokines, elicit homing behavior in bone marrow stem cells by sensing tissue injury and migrating to the site of damage (Shyu et al 2006). Higher expression of genes involved in immunity and cytokine activity is consistent with anecdotal evidence that both captive and wild chimpanzees have faster epidermal wound healing abilities compared to humans (Hedlund et al 2007). Analyses like these can give us a glimpse into the molecular differences underlying the human and chimpanzee condition.

157

5.6 Stem cells can greatly expand the number of in vitro models for comparative primate genomics

Recently, the National Research Council (National Academy of Sciences, USA) reaffirmed the important role of chimpanzees in comparative genomic research, highlighting numerous insights that have and will likely continue to come from this data

(Altevogt et al 2011). This report set forth two criteria for research involving chimpanzees: 1) that the studies provide otherwise unattainable insight and 2) that all experiments are performed on acquiescent animals in a manner that minimizes distress

(Altevogt et al 2011). The vast majority of comparative genomic research using chimpanzee adult stem cells not only meets both of these criteria, but also offers the opportunity to significantly expand the number of available approaches for fruitful inquiry. Moving forward, the use of adult stem cells from chimpanzees can complement existing in vivo research by providing access to a single cell type from that tissue, where experiments can be carried out in a controlled in vitro setting. Combined in vivo and in vitro comparative functional genomic analyses can provide a unique perspective with the potential to uncover novel results that would not otherwise be accessible.

Moving forward, adult stem cells promise to transform comparative primate genomics. Here, we profiled just one type of adult stem cell, the adipose stromal cell.

The primary nature of ASCs makes them an especially attractive candidate for medical applications (Gimble et al 2012). Outside of in vivo strategies, primary cells are the 158

closest representative a cell type as they have been taken directly from the living organism and have not been genetically transformed or reprogrammed. Another type of adult stem cell, the induced pluripotent stem cell (iPSC), offers further advantages for comparative functional studies. iPSCs are artificially derived through the genomic reprogramming of fibroblasts (Takahashi et al 2007). Romero et al 2012, commented on the potential role of iPSC in their recent review of evolutionary genomics approaches.

These may be the most promising source of adult stem cells from chimpanzees. There are established methods for de-differentiating fibroblasts into iPSCs, and several companies have developed kits specifically for this purpose (Takahashi et al 2007). This is important because a large catalog of chimpanzee fibroblasts are currently available

(for instance, the Coriell Institute has ~50 lines), compared with just one chimpanzee

ASC line. The available chimpanzee fibroblasts are derived from both sexes and span a wide range of ages, allowing for characterization of biological variation, which is unfortunately absent from the current study. In addition, iPSCs can be passaged many times, allowing for a steady supply of cells. In contrast, ASCs can only be cultured for a few passages before their ability to differentiate is diminished (Wall et al 2007).

Obtaining the amount of cells one needs for a complete analysis with ASCs is difficult when working with the chimpanzee, an endangered animal with minimal body fat. This limitation makes follow-up experiments a challenge when relatively large numbers of

159

cells are needed, as in DNase-Seq experiments (Shibata et al 2012) or when carrying out physiological challenge experiments in vitro. For example, interesting potential follow- ups to this study include eliciting an immune response by challenging cells with immunomodulating chemokines and carrying out classic in vitro scratch migration assays (Figure 5.6). These experiments could provide molecular insights into the presumed wound healing differences between human and chimpanzees (Hedlund et al

2007). The toolkit of experimental manipulations available for in vitro studies is vast, and includes a great variety of physiological and hormonal challenges, gene knock- downs, co-culturing multiple cell types, environmental manipulations, etc. It is also possible to measure a number of cellular phenotypes in vitro, including proliferation and apoptosis rates, migration ability, cell size and shape, organelle content, etc. There are innumerable combinations of functional assays, molecular manipulations, cell types, and phenotypes to measure that can be exploited using culture systems.

The utility of adult stem cells extends to conservation efforts as well. The chimpanzee is listed by the U.S. Fish and Wildlife Service as threatened in captivity and endangered in the wild (U.S. Fish & Wildlife Service 2012), while the International

Union for Conservation of Nature considers them as endangered with a declining population (Oates et al 2008). ASC can assist in chimpanzee conservation efforts by protecting their genetic diversity. The idea that cells can be used in this manner is

160

becoming more widely recognized as the utility of “frozen zoos” is gaining credibility

(Ben-Nun et al 2011). Recently, Ben-Nun et al, 2011 created iPSC from two highly endangered species, the northern white rhinoceros and the drill monkey, where they hypothesized that these resources could “facilitate the reintroduction of genetic material into the population” in the future. In fact, a transcriptomic approach, a recent study found that an endangered primate population contained considerable genetic variation, which could be capitalized on for conservation efforts (Perry et al 2012).

An important next step in primate comparative genomics will be using adult stem cells to carry out controlled experiments aimed at investigating molecular differences between humans and our closet living relatives. In vivo approaches based on tissue samples have proved valuable and will undoubtedly continue to provide useful information. However, the ability to work with cell culture systems provides opportunities for functional studies that would otherwise be impossible for practical or ethical reasons. These in vitro approaches provide a powerful complementary set of experimental tools that will likely become an increasingly important component of primate evolutionary genomics.

161

5.7 Materials and Methods

5.7.1 Culturing and differentiating stromal cells

Adult male ASCs from two different species are investigated in this study.

Specifically, one chimpanzee - S008396 (from the Coriell Institute for Biomedical

Research) and three humans - AG19304, AG20471 (from the Coriell Institute for

Biomedical Research), L040903 (from Zen-Bio). The stromal cells were recovered from cryofreeze in MesenPro RSTM Medium (Invitrogen) supplemented with 200 mM L-

Glutamine (Invitrogen) and 1% Penicillin-Streptomycin Solution (Invitrogen). These cells were allowed to expand to 70% confluency and then were removed using TryPLE

(Invitrogen) and plated at 40,625 cells / cm2 in 6-well plates (Corningstar). ASC were cultured for 24-48 hours until confluent and the chimpanzee were then differentiated into adipocytes using ZenBio’s Adipocyte Differentiation Medium (DM-2-PRF) and

Adipocyte Maintenance Medium (AM-1-PRF) supplemented with 250 µm of oleic acid

(Sigma) following Zen Bio’s adipocyte differentiation and maintenance protocol.

5.7.2 Stromal cell transcriptomics

The stromal cells were imaged (see section 5.7.3) and than collected at confluency. We isolated RNA from the stromal cells using QIAzol (Qiagen) followed by miRNeasy Mini extraction kit (Qiagen) with a DNase I treatment, and verified RNA quality using the Agilent Bioanalyzer 2100 (RIN = 10). Illumina TruSeq SBS libraries

162

were constructed with 1 ug of RNA. We used 50 bp, paired end Illumina HiSeqTM

Sequencing, with all four libraries multiplexed in one lane. Sequencing took place at

Duke Institute for Genome Sciences & Policy’s Genome Sequencing & Analysis Core

Resource. Reads were mapped to hg19 and panTro3 with TopHat v1.4.1 using default settings (Trapnell et al 2009). The mapped reads were counted with htseq-count 0.5.1p1 using the settings union and strandedness

(http://www-huber.embl.de/users/anders/HTSeq/doc/count.html). Gene models for each species were constructed using Primate Exon Orthology Database version 2

(http://giladlab.uchicago.edu/orthoExon/). These models were further filtered using the

Ensembl database (http://useast.ensembl.org/index.html). To remove genes with unclear homologies we eliminated Human-Chimpanzee homology types one2many and many2many as well as the ribosomal families RPL, RPS MRPL and MRPS. We also removed genes where the original chromosome assignment did not match the Ensembl chromosome assignment and where multiple Ensembl gene ID’s were assigned to the same HGNC gene name. Genes with less then five counts per ten million fragments were removed from every library. The counts were normalized by estimating the tagwise dispersion and significance was calculated used the program edgeR 1.6.0

(Robinson et al 2010) in R. FDR corrections for multiple comparisons were calculated using the Benjamini-Hochberg method, built into edgeR (Benjamini and Hochberg 1995).

163

Expression level was calculated as Fragments Per Kilobase of transcript per Million mapped reads (FPKM).

To interrogate the highly expressed chimpanzee ASC genes (FPKM >= 100, n =

614) we used the Database for Annotation, Visualization and Integrated Discovery

(DAVID) v6.7 KEGG Pathway tool, with the background list being 10,221 expressed in this study ( Huang et al 2009). The p-values were FDR-adjusted using the Benjamini-

Hochberg method (Benjamini and Hochberg 1995). When comparing the chimpanzee and human transcriptomes, categorical gene enrichments were calculated with the

PANTHER tools gene expression data analysis feature using the compare gene lists function (http://www.pantherdb.org/tools/). The queried genes are those significantly upregulated in the chimpanzee ASC and significantly upregulated in the human ASC at an FDR-adjusted p-value <0.05 (red and blue respectively in Figure 5.5). These were assessed against the background set of all 10,221 ASC genes in this study.

5.7.3 Staining and imaging

For the florescence imagining, 22x22 mm glass coverslips were coated with FNC mix (AthenaES) and cells were plated at 5,263 cells / cm2 in 6-well plates. Cells were cultured for 24 hours, fixed for 10 minutes with 4% paraformaldehyde (Electron

Microscopy Sciences), washed with 1x PBS (GIBCO), and stained with TRITC-phalloidin

(Sigma) & DAPI (Sigma), washed with 1x PBS and mounted on slides. The florescence

164

and black and white images were taken with the Zeiss Axio Observer A1 inverted stand microscope with a Zeiss HBO arc lamp and power supply using a Hamamatsu Orca ER digital camera in the Light Microscopy Core Facility at Duke University. These images were obtained using the MetaMorph software (v 7.6.5). ASC and adipocytes were stained for lipid content using the Oil Red O Stain Kit and protocol (ScyTek ORK-1) and imaged prior to confluence and on day 14 of differentiation. The color images were taken with a Leica DM IRB microscope using a Zeiss AxioCam ICc1 digital camera.

These images were obtained using the AxioVision software.

165

5.8 Acknowledgements

We thank Courtney Babbitt, Adam Pfefferle, Dan Runcie, and Jennifer Wygoda for comments about this study and manuscript. We thank the Duke Institute for

Genome Sciences & Policy’s Genome Sequencing & Analysis Core Resource, especially

Olivier Fedrigo and Fangfei Ye for library preparation and sequencing assistance. We thank the Light Microscopy Core Facility at Duke University, especially Sam Johnson and Yasheng Gao for microscopy training and imaging support. We thank the team at

ZenBio, especially Ben Buehrer, Jim Nicoll and Renee Lea-Currie for advice on adipocyte differentiation and a human ASC line. We thank the Coriell Institute for Medical

Research, especially Bernie Goldstein, for assistance with ordering chimpanzee and human ASC lines.

This work was supported by Duke Primate Genomics Initiative Fellowship to

L.W.P.; a Wenner Gren Dissertation Fieldwork Grant to L.W.P.; and HOMINID NSF

Grant NSF-BCS-08-27552 to G.A.W.

166

6. Summary

Nutrition is essential for life and is unique as it both shapes, and is shaped by the genome. Given this complex interaction, teasing out actors and responders in the genome-diet relationship is a challenge. These questions are made even more intriguing when adding the layer of evolutionary change. For decades, the anthropological community sought explanations for adaptation through gross anatomy, behavioral observations, and morphology. This literature provided a wealth of explanations outlining the driving forces behind human evolution. Many of these hypotheses converge on the impact of dietary change and it’s ability to create and sustain metabolically expensive adaptations.

The focus of my work was addressing these hypotheses on a molecular level across tissues of dietary and evolutionary importance. I took several expression approaches by interrogating regulatory regions, candidate networks and genomes to uncover unique human metabolic processes. These types of insights were gained by leveraging the power of controlled in vitro environments and native in vivo experiments.

Moving forward, cell culture methods will provide a much needed and exciting next step, as targeted questions can be asked and answered. When investigating complex interactions (e.g. genome-diet), this framework will be especially beneficial.

167

The molecular experiments presented here support several of the anthropological hypotheses suggesting that, both dietary and genomic changes place and respond to selective demands on human metabolic processes.

168

References

Abzhanov A, Kuo WP, Hartmann C, Grant BR, Grant PR, Tabin CJ. 2006. The calmodulin pathway and evolution of elongated beak morphology in Darwin’s finches. Nature 442:563-567.

Abzhanov A, Protas M, Grant BR, Grant PR, Tabin CJ. 2004. Bmp4 and morphological variation of beaks in Darwin's finches. Science 305:1462-1466.

Aiello LC, Wells JCK. 2002. Energetics and the evolution of the genus Homo. Annual Review of Anthropology 31:323–338.

Aiello LC, Wheeler P. 1995. The expensive-tissue hypothesis: The brain and the digestive system in human and primate evolution. Current Anthropology 38:199- 221.

Alaux C, Sinha S, Hasadsri L, Hunt GL, Guzman-Novoa E, DeGrandi-Hoffman G, Uribe-Rubio JL, Southey BR, Rodriquez-Zas, Robinson GE. 2009. Honey bee aggression supports a link between gene regulation and behavioral evolution. Proceedings of the National Academy of Sciences 106:15400-15405.

Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter P. 2008. Molecular biology of the cell. Garland Science, Taylor & Francis- Group, LLC. New York.

Altevogt BM, Pankevich DE, Shelton-Davenport MK, Kahn JP. 2011. Chimpanzees in biomedical and behavioral research: Assessing the necessity. The National Academies Press. Washington, D.C..

Ameur A, Enroth S, Johansson A, Zaboli G, Igl W, Johansson AC, Rivas MA, Daly MJ, Schmitz G, Hicks AA, Meitinger T, Feuk L, vanDuijn C, Oostra B, Pramstaller PP, Rudan L, Wright AF, Wilson JF, Campbell H, Gyllensten U. 2012. Genetic adaptation of fatty-acid metabolism: a human-specific haplotype increasing the bioynthesis of long-chain omega-3 and omega-6 fatty acids. American Journal Human Genetics 4:809-820.

Arias-Carrión O, Pöppel E. 2007. Dopamine, learning and reward-seeking behavior. Acta Neurobiologiae Experimentalis 67:481–488.

169

Babbitt CC, Warner LR, Fedrigo O, Wall CE, Wray GA. 2011. Genomic signatures of diet- related shifts during human origins. Proceedings of the Royal Society B 278:961-969.

Babbitt CC, Fedrigo O, Pfefferle AD, Boyle AP, Horvath JE, Furey TS, Wray GA. 2010. Both noncoding and protein-coding RNAs contribute to gene expression evolution in the primate brain. Genome Biology and Evolution 2:67-69

Barreiro LB, Marioni JC, Blekhman R, Stephens M, Gilad Y. 2010. Functional comparison of innate immune signaling pathways in primates. PLoS Genetics 6:e1001249.

Ben-Nun IF, Montague SC, Houcj ML, Tran HT, Garitaonandia I, Leonardo TR, Wang, Y-C, Charter SJ, Laurent LC, Ryder OA, Loring JF. 2011. Induced pluripotent stem cells from highly endangered species. Nature Methods 8:829-U889.

Benjamini Y, Hochberg Y. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 57:289-300.

Blanchette M, Bataille AR, Chen X, Poitras C, Laganiére J, Lefèbvre C, Deblois G, Giguère V, Ferretti V, Bergeron D, Coulombe B, Robert F. 2006. Genome-wide computational prediction of transcriptional regulatory modules reveals new insights into human gene expression. Genome Research 16:656-668.

Blekhman R, Marioni JC, Zumbo P, Stephens M, Gilad Y. 2010. Sex-specific and lineage-specific alternative splicing in primates. Genome Research 20:1-11.

Blekhman R, Oshlack A, Chabot AE, Smyth GK, Gilad, Y. 2008. Gene regulation in primates evolves under tissue-specific selection pressures. PLoS Genetics 4:1-13.

Blumenschine RJ, Peters CR, Masao FT, Clarke RJ, Deino AL, Hay RL, Swisher CC, Stanistreet IG, Ashley GM, McHenry LJ, Sikes NE, Van Der Merwe NJ, Tactikos JC, Cushing AE, Deocampo DM, Njau JK, Ebert JI. 2003. Late Pliocene Homo and hominid land use from western Olduvai Gorge, Tanzania. Science 299:1217-1221.

170

Blumenschine RJ, Pobiner BL. 2007. Zooarchaeology and the ecology of Oldowan hominin carnivory. In Evolution of the human diet: the known, the unknown and the unknowable (ed PS Ungar), p167-190. Oxford, UK: Oxford University Press.

Boesch C, Boesch-Achermann H. 2000. The chimpanzees of the Taı¨ forest. New York, NY: Oxford University Press.

Bradbury J. 2011. Docosahexaenoic acid (DHA): an ancient nutrient for the modern human brain. Nutrients 3:529-554.

Broadhurst CL, Cunnane SC, Crawford MA. 1998. Rift Valley lake fish and shellfish provided brain-specific nutrition for early Homo. British Journal of Nutrition 79:3- 21.

Brown F, Harris J, Leakey R, Walker A. 1985. Early Homo erectus skeleton from West Lake Turkana, Kenya. Nature 316:788–792.

Bunn, HT. 2001. Hunting, power scavenging, and butchering by Hadza foragers and by Plio-Pleistocene Homo. In Meat eating and human evolution (eds CB Stanford, HT Bunn), p199-281. Oxford, UK: Oxford University Press.

Bunn HT, Ezzo JA. 1993. Hunting and scavenging by Plio-Pleistocene hominids: nutritional constraints, archaeological patterns, and behavioral implications. Journal of Archaeological Sciences 20:365-398.

Bunn HT, Kroll EM. 1986. Systematic butchery by Plio-Pleistocene hominids at Olduvai Gorge, Tanzania. Current Anthropology 27:431-452.

Burdon T, Smith A, Savatier P. 2002. Signalling, cell cycle and pluripotency in embryonic stem cells. Trends in Cell Biology 12:432-438.

Cáceres M, Lachuer J, Zapala MA, Redmond JC, Kudo L, Geschwind DH, Lockhart DJ, Preuss TM, Barlow C. 2003. Elevated gene expression levels distinguish human from non-human primate brains Proceedings of the National Academy of Sciences 100:13030-13035.

171

Cain CE, Blekhman R, Marioni JC, Gilad Y. 2011. Gene expression differences among primates are associated with changes in a histone epigenetic modification. Genetics 187:1225-1234.

Carrasco S, Mérida I. 2007. Diacylglycerol, when simplicity becomes complex. Trends in Biochemical Science 32:27-36.

Carroll SB, Grenier J, Weatherbee S. 2007. From DNA to Diversity: Molecular Genetics and the Evolution of Animal Design. Wiley-Blackwell, Malden, Massachusetts.

Cawthorn WP, Scheller EL, MacDougald OA. 2012. Adipose tissue stem cells meet preadipocyte commitment: going back to the future. Journal of Lipid Research 53:227-246.

Chabot A, Shrit RA, Blekhman R, Gilad Y. 2007. Using reporter gene assays to identify cis- regulatory differences between humans and chimpanzees. Genetics 176:2069- 2076.

Chen F-C, Li, W-H. 2001. Genomic divergences between humans and other hominoids and the effective population size of the common ancestor of humans and chimpanzees. American Journal of Human Genetics 68:444-456.

Cheng PH, Snyder B, Fillos D, Ibegbu CC, Huang AH, Chan AW. 2008. Postnatal stem/progenitor cells derived from the dental pulp of adult chimpanzee. BMC Cell Biology 9:1-11.

Chivers D J, Hladik CM. 1980. Morphology of the gastrointestinal tract in primates: comparisons with other mammals in relation to diet. Journal of Morphology 166:337-386.

Ciccarone V, Spengler BA, Meyers MB, Biedler JL, Ross RA. 1989. Phenotypic diversification in human neuroblastoma cells: Expression of distinct neural crest lineages. Cancer Research 49:219-225.

Conklin-Brittain NL, Wrangham R, Smith CC. 2002. in Human diet: Its origin and evolution (eds. Ungar PS, Teaford MF) p61-76, Bergin & Garvey, Westport Connecticut.

172

Cordain L, Watkins BA, Florant GL, Kelher M, Rogers L, Li Y. 2002. Fatty acid analysis of wild ruminant tissues: evolutionary implications for reducing diet-related chronic disease. European Journal of Clinical Nutrition 56:181-191.

Crawford GE, Holt IE, Whittle J, Webb BD, Tai D, Davis S, Margulies EH, Chen Y, Bernat JA, Ginsburg D, Zhou D, Luo S, Vasicek TJ, Daly MJ, Wolfsberg TG, Collins FS. 2006. Genome-wide mapping of DNase hypersensitive sites using massively parallel signature sequencing (MPSS). Genome Research 16:123-131.

Daegling DJ, Grine FE. 1991 Compact-bone distribution and biomechanics of early hominid mandibles. American Journal of Physical Anthropology 86:321-339.

De Gobbi M, Viprakasit V, Hughes JR, Fisher C, Buckle VJ, Ayyub H, Gibbons RJ, Vernimmen D, Yoshinaga Y, de Jong P, Cheng JF, Rubin EM, Wood WG, Bowden D, Higgs DR. 2006. A regulatory SNP causes a human genetic disease by creating a new transcriptional promoter. Science 312:1215-1217.

Demes B, Creel N. 1988. Bite force, diet, and cranial morphology of fossil Hominids. Journal of Human Evolution 17:657-670.

De Miguel C, Henneberg M. 2001. Variation in hominid brain size: how much is due to method? Homo 52:3-58.

Domazet-Loso T, Tautz D. 2008. An ancient evolutionary origin of genes associated with human genetic diseases. Molecular Biology and Evolution 25:2699-2707.

Eaton SB. 1992. Humans, lipids and evolution. Lipids 27:814-820.

Edgar RC. 2004. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research 32:1792-1797.

Enard W, Khaitovich P, Klose J, Zöllner S, Heissig F, Giavalisco P, Nieselt-Struwe K, Muchmore E, Varki A, Ravid R, Doxiadis GM, Bontrop RE, Pääbo S. 2002. Intra- and interspecific variation in primate gene expression patterns. Science 296:340- 343.

173

Enattah NS, Jensen TG, Nielsen M, Lewinski R, Kuokkanen M, Rasinpera H, El-Shanti H, Seo JK, Alifrangis M, Khalil IF, Natah A, Ali A, Natah S, Comas D, Mehdi SQ, Groop L, Vestergaard EM, Imtiaz F, Rashed MS, Meyer B, Troelsen J, Peltonen L. 2008. Independent introduction of two lactase-persistence alleles into human populations reflects different history of adaptation to milk culture. American Journal of Human Genetics 82:57-72.

Enattah NS, Sahi T, Savilahti E, Terwilliger JD, Peltonen L, Järvelä I. 2002. Identification of a variant associated with adult-type hypolactasia. Nature Genetics 30:233-237, 2002.

Erion DM, Shulman GI. 2010. Diacylglycerol-mediated insulin resistance. Nature Medicine 16:400-402.

Fedrigo O, Pfefferle AD, Babbitt CC, Haygood R, Wall CE, Wray GA. 2011. A potential role for glucose transporters in the evolution of human brain size. Brain Behavior and Evolution 78:315-326.

Fernandes G. 1994. Dietary lipids and risk of autoimmune disease. Clinical Immunology and Immunopathology 72:193-197.

Finch CE, Stanford CB. 2004. Meat-adaptive genes and the evolution of slower aging in humans. The Quarterly Review 79:3-50.

Fish JL, Lockwood CA. 2003. Dietary constraints on encephalization in primates. American Journal of Physical Anthropology 120:171-181.

Fitzgerald GA. 2003. COX-2 and beyond: Approaches to prostaglandin inhibition in human disease. Nature Reviews Drug Discovery 2:879-890.

Fruhbeck, G. 2007. Adipose Tissue Protocols Second Edition (ed. Yang, K.) p1-22 Humana Press, London.

Gaulin SJC, Kurland JA. 1976. Primate predation and bioenergetics. Science 191:314- 317.

Gene Ontology Consortium. 2000. Gene ontology: tool for the unification of biology. Nature Genetics 25:25-29.

174

Gilad Y, Oshlack A, Smyth GK, Speed TP, White KP. 2006. Expression profiling in primates reveals a rapid evolution of human transcription factors. Nature 440:242-245.

Gimble JM, Bunnell BA, Guilak F. 2012. Human adipose-derived cells: an update on the transition to clinical translation. Regenerative Medicine 7:225-235.

Gimble JM, Katz AJ, Bunnell BA. 2007. Adipose-derived stem cells for regenerative medicine. Circulation Research 100:1249-1260.

Gompel N, Prud'homme B, Wittkopp PJ, Kassner VA, Carroll SB. 2005. Chance caught on the wing: cis-regulatory evolution and the origin of pigment patterns in Drosophila. Nature 433:481-487.

Grine FE, Kay RF. 1988. Early hominid diets from quantitative image analysis of dental microwear. Nature 333:765-768.

Grine FE, Martin LB. 1988. Enamel thickness and development in Australopithecus and Paranthropus. In Evolutionary history of the ‘robust’ australopithecines (ed. FE Grine), p3-42. New Brunswick, NJ: Transaction Publishers.

Gruber HE, Somayaji S, Riley F, Hoelscher GL, Norton HJ, Ingram J, Hanley EN Jr. 2012. Human adipose-derived mesenchymal stem cells: serial passaging, doubling time and cell senescence. Biotechnic & Histochemistry 87:303-311.

Hamblin MT, Di Rienzo A. 2000. Detection of the signature of natural selection in humans: Evidence from the duffy blood group locus. American Journal of Human Genetics 66:1669-1679.

Hamilton WJ III, Busse CD. 1978. Primate carnivory and its significance to human diets. BioScience 28:761-766.

Hasegawa M, Kishino H, Yano TA, 1985. Dating of the human-ape splitting by a molecular clock of mitochondrial DNA. Journal of Molecular Evolution 22:160-174.

Haygood R, Babbitt CC, Fedrigo O, Wray GA. 2010. Contrasts between adaptive coding and non-coding changes during human evolution. Proceedings of the National Academies of Sciences 107:7853–7857.

175

Haygood R, Fedrigo O, Hanson B, Yokoyama KD, Wray GA. 2007. Promoter regions of many neural- and nutrition-related genes have experienced positive selection during human evolution. Nature Genetics 39:1140-1144.

Hedlund M, Tangvoranuntakul P, Takematsu H, Long JM, Housley GD, Kozutsumi Y, Suzuki A, Wynshaw-Boris A, Ryan AF, Gallo RL, Varki N, Varki A. 2007. N- glycolylneuraminic acid deficiency in mice: Implications for human biology and evolution. Molecular Cellular Biology 27:4340-4346.

Hellmann I, Zollner S, Enard W, Ebersberger I, Nickel B, Paabo S. 2003. Selection on human genes as revealed by comparisons to chimpanzee cDNA. Genome Research 19:831-837.

Hoffmann GF, Assmann B, Bräutigam C, Dionisi-Vici C, Häussler M, de Klerk JB, Naumann M, Steenbergen-Spanjers GC, Strassburg HM, Wevers RA. 2003. Tyrosine hydroxylase deficiency causes progressive encephalopathy and dopa- nonresponsive dystonia. Annals of Neurology 54:S56-S65.

Hohmann, G, Fruth B. 2007. New records on prey capture and meat eating by Bonobos at Lui Kotale, Salonga National Park, Democratic Republic of Congo. Folia Primatologica 79:103-110.

Horan M, Millar DS, Hedderich J, Lewis G, Newsway V, Mo N, Fryklund L, Procter AM, Krawczak M, Cooper DN. 2003. Human growth hormone 1 (GH1) gene expression: complex haplotype-dependent influence of polymorphic variation in the proximal promoter and locus control region. Human Mutation 21:408-423.

Horrobin DF. 1999. Lipid metabolism, human evolution and schizophrenia Prostaglandins, Leukotrienes and Essential Fatty Acids 60:431-437.

Hotamisligil GS, Erbay E. 2008. Nutrient sensing and inflammation in metabolic diseases. Nature Reviews Immunology 8:923-934.

Huang DW, Sherman BT, Lempicki RA. 2009b. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nature Protocols 4:44-57.

176

Huby T, Dachet C, Lawn RM, Wickings J, Chapman MJ, Thillet J. 2001. Functional analysis of the chimpanzee and human apo(a) promoter sequences: identification of sequence variations responsible for elevated transcriptional activity in chimpanzees. Journal of Biological Chemistry 276:22209-22214.

Izadpanah R, Trygg C, Patel B, Kriedt C, Dufour J, Gimble JM, Bunnell BA. 2006. Biologic properties of mesenchymal stem cells derived from bone marrow and adipose tissue. Journal of Cellular Biochemistry 99:1285-1297.

Jansen BJ, Gilissen C, Roelofs H, Schaap-Oziemlak A, Veltman JA, Raymakers RA, Jansen JH, Kögler G, Figdor CG, Torensma R, Adema GJ. 2010. Functional differences between mesenchymal stem cell populations are reflected by their transcriptome Stem Cells and Development 19:481-489.

Jeong S, Rebeiz M, Andolfatto P, Werner T, True J, Carroll SB. 2008. The evolution of gene regulation underlies a morphological difference between two Drosophila sister species Cell 132:783–793.

Jolly CJ. 1970. The seed-eaters: a new model of hominid differentiation based on a baboon analogy Man 5:1-6.

Joyce JN, Smutzer G, Whitty CJ, Myers A, Bannon MJ. 1997. Differential Modification of Dopamine Transporter and Tyrosine Hydroxylase mRNAs in Midbrain of Subjects With Parkinson’s, Alzheimer’s With Parkinsonism, and Alzheimer’s Disease. Movement Disorders 12:885-897.

Kang JX, Leaf A. 1994. Effects of long-chain polyunsaturated fatty acids on the contraction of neonatal rat cardiac myocytes. Proceedings of the National Academies of Sciences 91:9886–9890.

Kaplan H, Hill K, Lancaster J, Hurtado AM. 2000 A theory of human life history evolution: diet, intelligence, and longevity. Evolutionary Anthropology 9:156-185.

Karaman MW, Houck ML, Chemnick LG, Nagpal S, Chawannakul D, Sudano D, Pike BL, Ho VV, Ryder OA, Hacia JG. 2003. Comparative analysis of gene-expression patterns in human and african great ape cultured fibroblasts. Genome Research 13:1619-1630.

177

Katz AJ, Tholpady A, Tholpady SS, Shang HL, Ogle RC. 2005. Cell surface and transcriptional characterization of human adipose-derived adherent stromal (hADAS) cells. Stem Cells 23:412-423.

Kay RF, Grine FE. 1988. Tooth morphology, wear and diet in Australopithecus and Paranthropus from Southern Africa. In Evolutionary history of the robustaustralopithecines (ed. F. E. Grine), p427-447. Hawthorne, NY: Aldine De Gruyter.

Keeley LH, Toth N. 1981 Microwear polishes on early stone tools from Koobi Fora, Kenya. Nature 293:464-465.

Keightley PD, Lercher MJ, Eyre-Walker A., 2005. Evidence for widespread degradation of gene control regions in hominid genomes. PLoS Biology 3:e42.

Kessler MA, Yang M, Gollomp KL, Jin H, Iacovitti L, 2003. The human tyrosine hydroxylase gene promoter. Molecular Brain Research 112:8-23.

Khaitovich P, Muetzel B, She X, Lachmann M, Hellmann I, Dietzsch J, Steigele S, Do HH, Weiss G, Enard W, Heissig F, Arendt T, Nieselt-Struwe K, Eichler EE, Pääbo S. 2004. Regional patterns of gene expression in human and chimpanzee brains. Genome Research 14:1462-1473.

Khaitovich P, Hellmann I, Enard W, Nowick K, Leinweber M, Franz H, Weiss G, Lachmann M, Pääbo S. 2005. Parallel patterns of evolution in the genomes and transcriptomes of humans and chimpanzees. Science 309:1850-1854.

Kim SM, Yang JW, Park MJ, Lee JK, Kim SU, Lee YS, Lee MA. 2006. Regulation of human tyrosine hydroxylase gene by neuron-restrictive silencer factor. Biochemical and Biophysical Research Communications 346:426-435.

Kim TE, Park MJ, Choi EJ, Lee HS, Lee SH, Yoon SH, Oh CK, Lee BJ, Kim SU, Lee YS, Lee MA. 2003. Cloning and cell type-specific regulation of the human tyrosine hydroxylase gene promoter. Biochemical and Biophysical Research Communications 312:1123-1131.

King MC, Wilson AC. 1975. Evolution at two levels in humans and chimpanzees. Science 188:107-116.

178

Kleinjan D, van Heyningen, V. 2004. Long-range control of gene expression: emerging mechanisms and disruption in disease. American Journal of Human Genetics 76:8- 32.

Kumer SC, Vrana KE. 1996. Intricate regulation of tyrosine hydroxylase activity and gene expression. Journal of Neurochemistry 67:443-462.

Lands, WE. 2012. Consequences of Essential Fatty Acids. Nutrients 4 :1-11.

Lawrence, GD. 2010. The Fats of Life: Essential Fatty Acids in Health and Disease. Rutgers University Press, New Brunswick.

Leboyer M, Malafosse A, Boularand S, Campion D, Gheysen F, Samolyk D, Henriksson B, Denise E, des Lauriers A, Lepine JP. 1990. Lancet 335:1219.

Leonard WR, Robertson ML. 1994. Evolutionary perspectives on human nutrition—the influence of brain and body-size on diet and metabolism. American Journal of Human Biology 6:77-88.

Leonard WR, Robertson ML, Snodgrass JJ, Kuzawa CW. 2003. Metabolic correlates of hominid brain evolution. Comparative Biochemistry and Physiology Part A 136:5-15.

Lobos EA, Todd RD. 1997. Cladistic analysis of disease association with tyrosine hydroxylase: Application to manic-depressive disease and alcoholism. American Journal of Medical Genetics 74:289-295.

Luca F, Perry GH, Di Rienzo A. 2010. Evolutionary adaptations to dietary changes. Annual Review of Nutrition 30:291-314.

Ludecke B, Dworniczak B, Bartholome K, 1995. A point mutation in the tyrosine hydroxylase gene associated with Segawa's syndrome. Human Genetics 95:123- 125.

Lüdecke B, Knappskog PM, Clayton PT, Surtees RA, Clelland JD, Heales SJ, Brand MP, Bartholomé K, Flatmark T. 1996. Recessively inherited L-DOPA-responsive parkinson in infancy caused by a gene point mutation (L205P) in the tyrosine hydroxylase gene. Human Molecular Genetics 5:1023-1028.

179

Mann N. 2007. Meat in the human diet: An anthropological perspective. Nutrition & Dietetics 64:S102–S107.

Martin RD. 1996. Scaling of the mammalian brain: the maternal energy hypothesis. News Physiological Science 11:149-156.

McGrew WC. 2001 The other faunivory: primate insectivory and early human diet. In Meat-eating and human evolution (eds CB Stanford, HT Bunn),p160-178. New York, NY: Oxford University Press.

McHenry HM. 1992. Body size and proportions in early hominids. American Journal of Physical Anthropology 87:407-431.

McHenry HM. 1994. Tempo and mode in human evolution. Proceedings of the National Academies of Sciences 91:6780-6786.

McLean AN. 2001. Cognitive abilities—the result of selective pressures on food acquisition? Applied Animal Behavioral Science 71:241-258.

McPherron SP, Alemseged Z, Marean CW, Wynn JG, Reed D, Geraads D, Bobe R, Béarat HA. 2010 Evidence for stone-tool-assisted consumption of animal tissues before 3.39 million years ago at Dikika, Ethiopia. Nature 466:857-860.

Meloni R, Laurent C, Campion D, Ben Hadjali B, Thibaut F, Dollfus S, Petit M, Samolyk D, Martinez M, Poirier MF. 1995. A rare allele of a microsatellite located in the tyrosine hydroxylase gene found in schizophrenic patients. Comptes Rendus de l Academie des Sciences. Serie III 318:803-809.

Mi H, Lazareva-Ulitsky B, Loo R, Kejariwal A, Vandergriff J, Rabkin S, Guo N, Muruganujan A, Doremieux O, Campbell MJ, Kitano H, Thomas PD. 2005. The PANTHER database of protein families, subfamilies, functions and pathways. Nucleic Acids Research 33:5.

Milton K, May, ML. 1976. Body weight, diet and home range area in primates. Nature 259:459-462.

180

Mitchell JB, McIntosh K, Zvonic S, Garrett S, Floyd ZE, Kloster A, Di Halvorsen Y, Storms RW, Goh B, Kilroy G, Wu X, Gimble JM. 2006. Immunophenotype of human adipose-derived cells: Temporal changes in stromal-associated and stem cell-associated markers. Stem Cells 24:376-385.

Miyoshi H, Perfield JW, Obin MS, Greenberg AS. 2008. Adipose triglyceride lipase regulates basal lipolysis and lipid droplet size in adipocytes. Journal of Cellular Biochemistry 105:1430-1436.

Murdock GP. 1967. Ethnographic atlas: a summary. Ethnology 6:109-236.

Nagatsu T. 1989. The human tyrosine hydroxylase gene. Cellular and Molecular Neurobiology 9:313-321.

Nagatsu T, Levitt M, Udenfriend S. 1964. Tyrosine hydroxylase: The initial step in norepinephrine biosynthesis. The Journal of Biological Chemistry 239:2910-2917.

Nakamura MT, Nara TY. 2004. Structure, function, and dietary regulation of delta6, delta5, and delta9 desaturases. Annual Review of Nutrition 24:345-376.

Nestler EJ, McMahon A, Sabban EL, Tallman JF, Duman RS. 1990. Chronic antidepressant administration decreases the expression of tyrosine hydroxylase in the rat locus coeruleus. Proceedings of the National Academies of Sciences 87:7522- 7526.

Oates JF, Tutin CEG, Humle T, Wilson ML, Baillie JEM, Balmforth Z, Blom A, Boesch C, Cox D, Davenport T, Dunn A, Dupain J, Duvall C, Ellis CM, Farmer KH, Gatti S, Greengrass E, Hart J, Herbinger I, Hicks C, Hunt KD, Kamenya S, Maisels F, Mitani JC, Moore J, Morgan BJ, Morgan DB, Nakamura M, Nixon S, Plumptre AJ, Reynolds V, Stokes EJ, Walsh PD. 2008. IUCN red list of threatened species: Version 2011.2. International Union for Conservation of Nature and Natural Resources. www.iucnredlist.org. 2012 June 11.

Olson MV, Varki A. 2003. Sequencing the chimpanzee genome: insights into human evolution and disease. Nature Reviews Genetics 4:20-28.

181

Or-Tzadikario S, Sopher R, Gefen A. 2010. Quantitative monitoring of lipid accumulation over time in cultured adipocytes as function of culture conditions: toward controlled adipose tissue engineering. Tissue Engineering Part C Methods 16:1167-1181.

Pai AA, Bell JT, Marioni JC, Pritchard JK, Gilad Y. 2011. A genome-wide study of DNA methylation patterns and gene expression levels in multiple human and chimpanzee tissues. PLoS Genetics 7:e1001316.

Pastinen T, Ge B, Hudson TJ. 2006. Influence of human genome polymorphism on gene expression. Human Molecular Genetics 15:R9-R16.

Perry GH, Melsted P, Marioni JC, Wang Y, Bainer R, Pickrell JK, Michelini K, Zehr S, Yoder AD, Stephens M, Pritchard JK, Gilad Y. 2012. Comparative RNA sequencing reveals substantial genetic variation in endangered primates. Genome Research 22:602-610.

Perry GH, Dominy NJ, Claw KG, Lee AS, Fiegler H, Redon R, Werner J, Villanea FA, Mountain JL, Misra R, Carter NP, Lee C, Stone AC. 2007. Diet and the evolution of human amylase gene copy number variation. Nature Genetics 39:1257-1261.

Peters CR. 2007. Theoretical and actualistic ecobotanical perspectives on early hominin diets. In Evolution of the human diet: the known, the unknown, and the unknowable (ed. PS Ungar), p233-261. Oxford, UK: Oxford University Press.

Peters CR, O’Brian EM. 1981. The early hominid plant-food niche: insights from an analysis of plant exploitation by Homo, Pan, and Papio in eastern and southern Africa. Current Anthropology 22:127-140.

Pfefferle AD, Warner LR, Wang CW, Nielsen WJ, Babbitt CC, Fedrigo O, Wray GA. 2011. Comparative expression analysis of the phosphocreatine circuit in extant primates: Implications for human brain evolution. Journal of Human Evolution 60:205-212.

Pollard KS, Salama SR, Lambert N, Lambot MA, Coppens S, Pedersen JS, Katzman S, King B, Onodera C, Siepel A, Kern AD, Dehay C, Igel H, Ares M Jr, Vanderhaeghen P, Haussler D. 2006. An RNA gene expressed during cortical development evolved rapidly in humans Nature 443:167-172.

182

Pond SL, Frost SD, Muse SV. 2005. HyPhy: hypothesis testing using phylogenies. Bioinformatics 21:676-679.

Prabhakar S, Visel A, Akiyama JA, Shoukry M, Lewis KD, Holt A, Plajzer-Frick I, Morrison H, Fitzpatrick DR, Afzal V, Pennacchio LA, Rubin EM, Noonan JP. 2008. Human-specific gain of function in a developmental enhancer. Science 321:346-1350.

Prud'homme B, Gompel N, Rokas A, Kassner VA, Williams TM, Yeh SD, True JR, Carroll SB. 2006. Repeated morphological evolution through cis-regulatory changes in a pleiotropic gene. Nature 440:1050-1053.

Rao F, Zhang L, Wessel J, Zhang K, Wen G, Kennedy BP, Rana BK, Das M, Rodriguez- Flores JL, Smith DW, Cadman PE, Salem RM, Mahata SK, Schork NJ, Taupenot L, Ziegler MG, O'Connor DT. 2007. Tyrosine hydroxylase, the rate-limiting enzyme in catecholamine biosynthesis: Discovery of common human genetic variants governing transcription, autonomic activity, and blood pressure in vivo. Circulation 116:993-1006.

Reinius B, Saetre P, Leonard JA, Blekhman R, Merino-Martinez R, Gilad Y, Jazin E. 2008. An evolutionarily conserved sexual signature in the primate brain. PLoS Genetics 4:e100100:1-e100100:13.

Ribases M, Serrano M, Fernandez-Alvarez E, Pahisa S, Ormazabal A, Garcia-Cazorla A, Perez-Duenas B, Campistol J, Artuch R, Cormand B. 2007. A homozygous tyrosine hydroxylase gene promoter mutation in a patient with dopa-responsive encephalopathy: Clinical, biochemical and genetic analysis. Molecular Genetics and Metabolism 92:274-277.

Robinson JT. 1954. Prehominid dentition and hominid evolution. Evolution 8:324-334.

Robinson MD, McCarthy DJ, Smyth GK. 2010. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26:139-140.

Rockman MV, Hahn MW, Soranzo N, Zimprich F, Goldstein DB, Wray GA. 2005. Ancient and recent positive selection transformed opioid cis-regulation in humans. PLoS Biology 3:2208-2219.

183

Rockman MV, Wray GA. 2002. Abundant raw material for cis-regulatory evolution in humans. Molecular Biology and Evolution 19:1991-2004.

Romero IG, Ruvinsky I, Gilad Y. 2012. Comparative studies of gene expression and the evolution of gene regulation. Nature Reviews Genetics 13:505-516.

Romano G, Suon S, Jin H, Donaldson AE, Iacovitti L. 2005. Characterization of five evolutionary conserved regions of the human tyrosine hydroxylase (TH) promoter: Implications for the engineering of a human TH minimal promoter assembled in a self-inactivating lentiviral vector system. Journal of Cell Physiology 204:666-677.

Rosen ED, Spiegelman BM. 2006. Adipocytes as regulators of energy balance and glucose homeostasis. Nature 444:874-880.

Ruxton GD, Houston DC. 2004. Obligate vertebrate scavengers must be large soaring fliers. Journal of Theoretical Biology 228:431-436.

Schoenemann PT. 2006. Evolution of the size and functional areas of the human brain. Annual Review of Anthropology 35:379-406.

Schoeninger MJ, Bunn HT, Murray S, Pickering T, Moore J. 2001. Meat-eating by the fourth African ape. In Meat-eating and human evolution (eds CB Stanford, HT Bunn), p179-198. New York, NY: Oxford University Press.

Serretti A, Fabio M, Verga M, Cusin C, Pedrini S, Smeraldi E. 1998. Tyrosine hydroxylase gene associated with depressive symptomatology in mood disorder. American Journal of Medican Genetics 81:127-130.

Shapiro MD, Marks ME, Peichel CL, Blackman BK, Nereng KS, Jónsson B, Schluter D, Kingsley DM. 2004. Genetic and developmental basis of evolutionary pelvic reduction in three spine sticklebacks. Nature 428:717-724.

Shea JJ. 2007. Lithic archaeology, or, what stone tools can (and can’t) tell us about early hominin diets. In Evolution of the human diet. The known, the unknown, and the unknowable (ed. PS Ungar), p212-232. Oxford, UK: Oxford University Press.

184

Shibata Y, Sheffield NC, Fedrigo O, Babbitt CC, Wortham M, Tewari AK, London D, Song L, Lee BK, Iyer VR, Parker SC, Margulies EH, Wray GA, Furey TS, Crawford GE. 2012. Extensive evolutionary changes in regulatory element activity during human origins are associated with altered gene expression and positive selection. PLoS Genetics 8:e1002789.

Shyu WC, Lee YJ, Liu DD, Lin SZ, Li H. 2006. Homing genes, cell therapy and stroke. Frontiers in Bioscience 11:899-907.

Simopoulos AP. 1999. Essential fatty acids in health and chronic disease. American Journal of Clinical Nutrition 70:560S-569S.

Simopoulos AP. 2011. Importance of the omega-6/omega-3 balance in health and disease: evolutionary aspects of diet. World Review of Nutrition and Dietetics 102:10-21.

Somel M, Franz H, Yan Z, Lorenc A, Guo S, Giger T, Kelso J, Nickel B, Dannemann M, Bahn S, Webster MJ, Weickert CS, Lachmann M, Pääbo S, Khaitovich P. 2009. Transcriptional neoteny in the human brain. Proceedings of the National Academy of Sciences 106:5743-5748.

Sorek R, Ast G. 2003. Intronic sequences flanking alternatively spliced exons are conserved between human and mouse. Genome Research 13:1631-1637.

Sponheimer M, Lee-Thorp JA. 1999. Isotopic evidence for the diet of an early homoinid, Australopithecus africanus. Science 283:368-370.

Sponnheimer M, Lee-Thorp JA. 1999. Oxygen isotope ratios in enamel carbonate and their ecological significance. Journal of Archaeological Science 26:723-728.

Stables MJ, Gilroy DW. 2011. Old and new generation lipid mediators in acute inflammation and resolution. Progress in Lipid Research 50:35-51.

Stahl A, Evans JG, Pattel S, Hirsch D, Lodish HF. 2002. Insulin causes fatty acid transport protein translocation and enhanced fatty acid uptake in adipocytes. Developmental Cell 2:477-388.

185

Stanford CB, Wallis J, Matama H, Goodall J. 1994. Patterns of predation by chimpanzees on red colobus monkeys in Gombe National Park, 1982-1991. American Journal of Physical Anthropology 94:213–228.

Stanford CB. 1999. The Hunting Apes: Meat Eating and the Origins of Human Behavior. Princeton University Press, New Jersey.

Stanford CB, Bunn HT. 2001. Meat-Eating & Human Evolution. Oxford University Press, New York, New York.

Steiner CC, Weber JN, Hoekstra HE. 2007. Adaptive variation in beach mice produced by two interacting pigmentation genes. PLoS Biology 5:1880-1889.

Stern DL. 1998. A role of ultrabithorax in morphological differences between Drosophila species. Nature 396:463-466.

Storch J, Corsico B. 2008. The Emerging Functions and Mechanisms of Mammalian Fatty Acid-Binding Proteins. Annual Review of Nutrition 28:73-95.

Surbeck M, Hohmann G. 2008. Primate hunting by bonobos at LuiKotale, Salonga National Park. Current Biology 18:R906-R907.

Supek F, Bošnjak M, Škunca N, Šmuc T. 2011. REVIGO summarizes and visualizes long lists of Gene Ontology terms. PLoS ONE 6:e21800.

Takahashi K, Tanabe K, Ohnuki M, Narita M, Ichisaka T, Tomoda K, Yamanaka S. 2007. Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell 131:861-872.

Tao H, Cox DR, Frazer KA. 2006. Allele-specific KRT1 expression is a complex trait. PLoS Genetics 2:0848-0858.

Teaford MF, Ungar PS. 2000. Diet and the evolution of the earliest human ancestors. Proceedings of the National Academy of Sciences 97:13506-13511.

186

Teaford MF, Ungar PS, Grine FE. 2002. Paleontological evidence for the diets of African Plio-Pleistocene hominins with special reference to early Homo. In Human diet: its origin and evolution (eds PS Ungar, MF Teaford), p143-166. Westport, CT: Bergin & Garvey.

The Chimpanzee Sequencing and Analysis Consortium. 2005. Initial sequence of the chimpanzee genome and comparison with the human genome. Nature 437:69-87.

Thieme H. 1997. Lower Paleolithic hunting spears from Germany. Nature 385:807-810.

Tishkoff SA, Reed FA, Ranciaro A, Voight BF, Babbitt CC, Silverman JS, Powell K, Mortensen HM, Hirbo JB, Osman M, Ibrahim M, Omar SA, Lema G, Nyambo TB, Ghori J, Bumpstead S, Pritchard JK, Wray GA, Deloukas P. 2007. Convergent adaptation of human lactase persistence in Africa and Europe. Nature Genetics 39:31-40.

Tournamille C, Colin Y, Cartron J, Levankim C. 1995. Disruption of a GATA motif in the Duffy gene promoter abolishes erythroid gene expression in Duffy negative individuals. Nature Genetics 11:224-228.

Trapnell C, Pachter L, Salzberg SL. 2009. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 25:1105-1111.

Uddin M, Wildman DE, Liu G, Xu W, Johnson RM, Hof PR, Kapatos G, Grossman LI, Goodman M. 2004. Sister grouping of chimpanzees and humans as revealed by genome-wide phylogenetic analysis of brain gene expression profiles. Proceedings of the National Academy of Sciences 101:2957-2962.

Ungar PS, Grine FE, Teaford MF. 2008. Dental microwear and diet of the Plio- Pleistocene hominin Paranthropus boisei. PLoS ONE 3:e2044.

Ungar PS, Grine FE, Teaford MF, Zaatari SE. 2006. Dental microwear and diets of African early Homo. Journal of Human Evolution 50:78-95.

Ungar PS, Grine FE, Teaford MF. 2006. Diet in early Homo: A review of the evidence and a new model of adaptive versatility. Annual Reviews of Anthropology 35:209- 228.

187

Ungar P. 2004. Dental topography and diets of Australopithecus afarensis and early Homo. Journal of Human Evolution 46:605-622.

U.S. Fish & Wildlife Service. 2012. Chimpanzee (Pan troglodytes). http://ecos.fws.gov/speciesProfile/profile/speciesProfile.action?spcode=A06C.

Van der Merwe NJ, Masao FT, Bamford MK. 2008. Isotopic evidence for contrasting diets of early hominins Homo habilis and Australopithecus boisei of Tanzania. South African Journal of Science 104:153-155.

Varki A, Altheide TK. 2005. Comparing the human and chimpanzee genomes: Searching for needles in a haystack. Genome Research 15:1746-1758.

Verbeek MM, Steenbergen-Spanjers GCH, Willemsen MAAP, Hol FA, Smeitink J, Seeger J, Grattan-Smith P, Ryan MM, Hoffmann GF, Donati MA, Blau N, Wevers RA. 2007. Mutations in the cyclic andenosine monophosphate response element of the tyrosine hydroxylase gene. Annals of Neurology 62:422-426.

Voight BF, Kudaravalli S, Wen X, Pritchard JK. 2006. A map of recent positive selection in the human genome. PLoS Biolology 4:446-458.

Wall ME, Bernacki SH, Loboa EG. 2007. Effects of serial passaging on the adipogenic and osteogenic differentiation potential of adipose-derived human mesenchymal stem cells. Tissue Engineering 13:1291-1298.

Watkins PA, Maiguel D, Jia Z, Pevsner J. 2007. Evidence for 26 distinct acyl-coenzyme A synthetase genes in the human genome. Journal of Lipid Research 48:2736-2750.

Wells JCK. 2010. The Evolutionary Biology of Human Body Fatness: Thrift and Control. Cambridge University Press, Cambridge, United Kingdom.

White TD, Asfaw B, Beyene Y, Haile-Selassie Y, Lovejoy CO, Suwa G, WoldeGabriel G. 2009. Ardipithecus ramidus and the paleobiology of early hominids. Science 326:75-86.

188

Wolfinger RD, Gibson G, Wolfinger ED, Bennett L, Hamadeh H, Bushel P, Afshari C, Paules RS. 2001. Assessing gene significance from cDNA microarray expression data via mixed models. Journal of Computational Biology 8:625-637.

Wong WSW, Nielsen R. 2004. Detecting selection in noncoding regions of nucleotide sequences. Genetics 167:949-958.

Wood B, Collard M. 1999. The human genus. Science 287: 65-71.

Wray GA. 2007. The evolutionary significance of cis-regulatory mutations. Nature Reviews Genetics 8:206-216.

Wray GA, Hahn MW, Abouheif E, Balhoff JP, Pizer M, Rockman MV, Romano LA. 2003. The evolution of transcriptional regulation in eukaryotes. Molecular Biological Evolution 20:1377-1419.

Yeakel JD, Bennett NC, Koch PL, Dominy NJ. 2007. The isotopic ecology of African mole rats informs hypotheses on the evolution of human diet. Proceedings of the Royal Society B 274:1723-1730.

Yehuda S. 2003. Omega-6/omega-3 ratio and brain-related functions. World Review of Nutrition and Dietetics 92:37-56.

Zhao M, Chen X, Gao G, Tao L, Wei L. 2009. RLEdb: a database of rate-limiting enzymes and their regulation in human, rat, mouse, yeast and E. coli. Cell Research 19:793-795.

Zeder M A. 2006. Documenting domestication: new genetic and archaeological paradigms. Berkeley, MA: University of California Press.

Zhang J, Nielsen R, Yang Z. 2005. Evaluation of an improved branch-site likelihood method for detecting positive selection at the molecular level. Molecular Biology and Evolution 22:2472-2479.

Zhu MY, Klimek V, Dilley GE, Haycock JW, Stockmeier C, Overholser JC, Meltzer HY, Ordway GA. 1999. Elevated levels of tyrosine hydroxylase in the locus coreruleus in major depression. Biological Psychiatry 46:1275-1286.

189

Zuk PA, Zhu M, Mizuno H, Huang J, Futrell JW, Katz AJ, Benhaim P, Lorenz HP, Hedrick MH. 2001. Multilineage cells from human adipose tissue: implications for cell-based therapies. Tissue Engineering 7:211-228.

190

Biography

Born October 8, 1985 in Durham North Carolina to Christopher and Michelle Warner.

Education 2012 Ph.D.: University Program in Genetics and Genomics Certificate in Developmental and Stem Cell Biology Duke University, Durham, NC 2007 B.S.: Honors Biology with a concentration in Neuroscience Westhampton College at the University of Richmond, Richmond VA

Publications Babbitt CC, Warner LR, Fedrigo O, Wall CE and Wray GA. (2011) Genomic signatures of diet-related shifts during human origins. Proc. R. Soc. B. 278:961–969. Pfefferle AD*, Warner LR*, Wang C, Nielsen WJ, Babbitt CC, Fedrigo O and Wray, GA. (2011) Comparative expression analysis of the phosphocreatine circuit in extant primates: implications for human brain evolution. J. Hum. Evol. 60:205-212. Fedrigo O, Warner LR, Pfefferle AD, Babbitt CC, Cruz-Gordillo P and Wray GA. (2010) Selecting accurate RT-QPCR control genes for comparing gene expression in primates. PLoS ONE 5(9):e12545. Hill A, Boll W, Reis C, Warner L, Osswalt M, Hill M and Noll M. (2010) Origin of Pax and Six gene families in sponges: Single PaxB and Six1/2 orthologs in Chalinula loosanofﬁ. Dev. Biol. 343:106-123. Warner LR, Babbitt CC, Primus AE, Severson TF, Haygood R and Wray GA. (2009) The functional consequences of genetic variation on tyrosine hydroxylase (TH) expression in primates. Brain Res 1288:1-8. Haga SB, Warner LR and O’Daniel J. (2009) The Potential of a Placebo/Nocebo Effect in Pharmacogenetics. Public Health Gen. 12:158–162. Runyen-Janecky L, Dazenski E, Hawkins S and Warner L. (2006) Role and Regulation of the Shigella flexneri Sit and MntH Systems. Infect. Immun. 74:4666-4672.

Honors & Grants • Phi Beta Kappa, Omicron Delta Kappa, Mortar Board, James B. Duke Fellowship, Ethyl and Albemarle Science Scholar • HHMI Duke Vertical Integration Partners Program, Graduate Researcher (2011) Wenner-Gren Dissertation Fieldwork Grant and Osmundsen Initiative (2010) Primate Genomics Initiative Graduate Research Fellowship (2009) 191