Identification of differences in human and great ape phytanic acid metabolism that could influence gene expression profiles and physiological functions

Background It has been proposed that anatomical differences in human and great ape guts arose in response to species-specific diets and energy demands. To investigate functional genomic consequences of these differences, we compared their physiological levels of phytanic acid, a branched chain fatty acid that can be derived from the microbial degradation of chlorophyll in ruminant guts. Humans who accumulate large stores of phytanic acid commonly develop cerebellar ataxia, peripheral polyneuropathy, and retinitis pigmentosa in addition to other medical conditions. Furthermore, phytanic acid is an activator of the PPAR-alpha transcription factor that influences the expression of genes relevant to lipid metabolism. Results Despite their trace dietary phytanic acid intake, all great ape species had elevated red blood cell (RBC) phytanic acid levels relative to humans on diverse diets. Unlike humans, chimpanzees showed sexual dimorphism in RBC phytanic acid levels, which were higher in males relative to females. Cultured skin fibroblasts from all species had a robust capacity to degrade phytanic acid. We provide indirect evidence that great apes, in contrast to humans, derive significant amounts of phytanic acid from the hindgut fermentation of plant materials. This would represent a novel reduction of metabolic activity in humans relative to the great apes. Conclusion We identified differences in the physiological levels of phytanic acid in humans and great apes and propose this is causally related to their gut anatomies and microbiomes. Phytanic acid levels could contribute to cross-species and sex-specific differences in human and great ape transcriptomes, especially those related to lipid metabolism. Based on the medical conditions caused by phytanic acid accumulation, we suggest that differences in phytanic acid metabolism could influence the functions of human and great ape nervous, cardiovascular, and skeletal systems.


Background
Humans and great apes (bonobos, chimpanzees, gorillas, and orangutans) share a common gut anatomy, consisting of a simple stomach, small intestine, small cecum terminating in an appendix, and a hindgut consisting of the large intestine, rectum, and anal canal [1]. Nevertheless, significant differences have been reported in their gut proportions [2][3][4]. While the large intestine represents the majority of the great ape gut volume, the majority of the modern human gut volume consists of the small intestine [2][3][4]. Initial surveys have also indicated modern humans have a smaller total gut volume to body mass ratio relative to the great apes [5][6][7]. However, this could be influenced by primate gut plasticity related to diet and genetic diversity [2,8].
It has been proposed that gut proportions changed at some point within the human lineage in response to higher quality foods which can be digested in the small intestine [2]. The diets of hominids and/or early human populations improved, in part, due to cooking [9] and the increased abundance of animal products obtained through scavenging, hunting, fishing, and dairy consumption [10][11][12][13][14][15][16][17][18][19]. In contrast, great ape species in the wild derive a significant amount of their total daily metabolic energy needs through the fermentation of lower quality plant materials in their hindguts [20][21][22][23][24][25]. Although hindgut fermentation also occurs in humans [26][27][28], there is evidence that wild great apes derive greater amount of total daily metabolic energy from this process than do humans on Western diets [20][21][22]. However, seasonal changes in great ape diets and the limited dietary diversity of the humans studied will influence the interpretation of these data sets.
To explore the systemic consequences of hindgut fermentation activities in humans and great apes, we evaluated their physiological levels and cellular metabolic activities of phytanic acid, a branched chain fatty acid that can bind to and/or activate PPAR-alpha [29][30][31][32][33][34] and RXR [35][36][37][38] transcription factors. Humans do not synthesize phytanic acid, but rather acquire it from the diet [39,40]. In ruminants, the gut fermentation of plant materials liberates phytol, a constituent of chlorophyll, which is then converted to phytanic acid and stored in fats [39,40]. While humans can covert free phytol into phytanic acid, they do not accumulate significant amounts of phytanic acid as a result of consuming plant materials [41,42]. However, they can obtain phytanic acid from ruminant fats, fish, and dairy products [41]. Humans with impaired phytanic acid catabolism can overaccumulate phytanic acid, which results in peripheral polyneuropathy, cerebellar ataxia, retinitis pigmentosa, anosmia, and hearing loss [43][44][45]. This can also result in cardiac arrhythmias, shortened metacarpals or metatarsals, and ichthyosis [43][44][45].
Despite the demonstrated importance of maintaining appropriate levels of phytanic acid in humans and mice [46][47][48], phytanic acid metabolism has not been studied in the great apes. Based on the biochemical evaluation of red blood cells and cultured primary skin fibroblasts, we uncovered candidate human-specific differences in the ability to obtain phytanic acid from dietary sources relative to the great apes. We also provide indirect evidence that these changes could affect physiological functions relevant to the evolution and health of these species.
To begin to address these dietary differences, we compared RBC phytanic acid levels in great apes and humans on phytanic acid-deficient diets ( Fig. 2A). The latter are based on a cohort of 16 humans on vegan diets for over one year (Additional File 1). The absolute amount of phytanic acid intake in this population is unknown; however, Figure 1 Phytanic acid catabolism in mammals. Phytanic acid in ruminant fats is derived from phytol produced during the bacterial degradation of chlorophyll in their rumen (first stomach). After conversion to its CoA thioester, phytanic acid undergoes αoxidation, yielding pristanic acid. This fatty acid then undergoes three subsequent rounds of β-oxidation in the peroxisome. The resulting medium chain fatty acid exits the peroxisome and translocates to the mitochondrion where the remaining carbon chain is degraded by β-oxidation. Abbreviations for the enzymes listed include: HACL1 (aka HPCL2) = 2-hydroxyphytanoyl-CoA lyase; PHYH = phytanoyl-CoA α-hydroxylase; PDH = pristanal dehydrogenase, whose gene is not yet known. We note that phytanic acid can also be degraded by β-oxidation; however, this activity of this pathway is relatively minor [40,44].
Given the composition of our cohort, we had adequate statistical power to screen for possible sexual dimorphism in human (Western diet) and chimpanzee RBC phytanic acid levels. Although no sex-specific differences were found in humans on Western diets, male chimpanzees showed 1.4-fold higher (P < 2 × 10 -6 ) RBC phytanic acid levels relative to females (Fig. 2B). This occurs despite the fact that our cohort of male and female chimpanzees had equivalent dietary exposure to phytanic acid and its precursor, phytol. Although our power is limited due to the numbers of vegan males in our cohort, there was also no evidence of sexual dimorphism in vegan RBC phytanic acid levels (Fig. 2B).

Comparative transcriptome analysis of peroxisomal aand b-oxidation pathways
To evaluate potential in vivo consequences of crossspecies differences in lipid metabolism, we reanalyzed  oligonucleotide microarray gene expression data from 5 male chimpanzee and 6 male human livers, brains, kidneys, heart, and testes [51]. We only considered data from probes predicted to be perfectly matched to both genomes and calculated gene expression scores only for probe sets containing at least four probes after masking [52]. We focused on genes involved in peroxisomal phytanic acid α-oxidation, β-oxidation, and PPARresponsive genes relevant to phytanic acid metabolism (Fig. 4, Additional File 3). Our criteria for differential gene expression are provided in the legend of Figure 4.
We found that cross-species differential expression of phytanic acid α-oxidation and peroxisomal β-oxidation genes in the liver, brain, and testes always reflected higher transcript levels in humans (Fig. 4). Genes in these categories also showed cross-species differential expression in heart and kidney; however, some transcripts were more abundant in humans (7 for heart and 4 for kidneys) while others were more abundant in chimpanzees (1 for heart and 8 for kidneys). Strikingly, ACOX3 transcript levels were higher in humans relative to chimpanzees for all five tissues examined. In 4 tissues (liver, heart, testes, and kidney), PHYH transcript levels were also higher in human relative to chimpanzee. To further explore this latter observation, we measured PHYH levels in human, chimpanzee, bonobo, and gorilla fibroblast cultures (6 different donors per species) via quantitative PCR (qPCR) using primers designed against a conserved region of this transcript. PHYH levels were at least 2-fold more abundant in human relative to chimpanzee, bonobo, and gorilla fibroblasts (P < 3 × 10 -4 for each individual comparison, ANOVA P < 2 × 10 -5 for human versus African great apes).
Lastly, we examined a group of five genes (FABP1, DBI, SLC27A1, GOT2, and CAT) previously used to evaluate PPAR-activity in mice with phytanic acid disorders [53] Figure 4 Differential expression of genes related to peroxisomal lipid metabolism. We reanalyzed of Affymetrix GeneChip U133v2.0 expression profiles of human and chimpanzee tissues [51] using the masking strategy stated in the text. We used standard F-tests (FDR-adjusted using the Benjamini and Hochberg approach) to test for differences in the distributions by species for the 5 tissues. The fold change (FC) of human (Hsa) versus chimpanzee (Ptr) geometric mean gene expression scores are provided. Differentially expressed genes (≥1.2 FC in either direction with a Student's t-test, two-tailed P-value ≤0.05 after Bonferroni correction) for a given tissue are highlighted in red (higher in human) or green (higher in chimpanzee). Probe sets with (i) F-tests yielding a ≤5% FDR and (ii) differential expression in at least one tissue are shown. All data are provided in Additional File 3. and PEX11A, a direct transcriptional target of PPARalpha [54]. All differentially expressed transcripts in heart (DBI and SLC27A1) and testes (DBI and GOT2) were more abundant in humans. FABP1, DBI, and PEX11A transcript levels were higher in human liver; however, CAT transcripts were more abundant in chimpanzee liver. None of these six genes showed cross-species differential expression in brain or kidney.

Discussion
Microbial fermentation in human [26][27][28] and great ape [20][21][22][23][24][25] hindguts are responsible for the break down of complex carbohydrates not processed and absorbed in the small intestine. The major end products of hindgut fermentation are short chain fatty acids, which provide energy-yielding substrates for colonic mucosa that regulate its growth and blood flow and promotes sodium and water absorption [26,27]. It has been estimated that wild chimpanzees derive 21-33% and wild orangutans derive 7-57% of their total daily metabolic energy from fiber fermentation [22]. This could be as high as 30-60% for wild Western gorillas [21]. In contrast, humans on Western diets are estimated to obtain no more than 10% of their daily energy needs through hindgut fermentation; however, this is likely higher in populations with lower quality diets and instances of small intestinal malabsorption [28].
Although we contrast human and the great ape dietary strategies, it is important to consider diversity within the great apes. For example, gorillas and orangutans can and often do subsist on lower quality foods (e.g. mature foliage and unripe fruits) relative to chimpanzees and bonobos, which can consume ample amounts of succulent ripe fruits [2]. Furthermore, chimpanzees hunt colobus monkeys and other small mammals; however, this comprises a minor component of their diets and varies among individuals and communities [55][56][57][58][59]. Thus, we propose that it is likely that the human, chimpanzee, bonobo, gorilla and orangutan lineages have all acquired molecular adaptations relevant to their diets.
In the studies presented here, we began to evaluate the broader systemic consequences of hindgut fermentation and related lipid metabolic activity in humans and great apes. In principle, RBC phytanic acid can be an effective biomarker of hindgut fermentation of plant materials as long as phytanic acid or its precursor phytol is not appreciably present in the human or great ape diets. The phytanic acid content of Western diets (50-100 mg daily intake) [43] is estimated to be at least 10 × greater than its free phytol content [41,60]. Since phytanic acid is primarily found in ruminant fats (organic beef fat has >325 mg/100 g), dairy products (45% fat cream cheese has >125 mg/100 g), and fish (canned salmon has >250 mg/100 g) [41], human vegans are appropriate for our studies. In this regard, a survey of fresh vegetables and fruits all showed minimal phytol levels (≤2 mg phytol/100 g) [41]. We also note a prior study demonstrating that an individual who ingested 3.5-kg of boiled spinach leaves within a 60 hour period as a supplement to an uncontrolled mixed diet later showed serum phytanic acid levels within normal limits [42]. Our captive great ape cohort should be exposed to similar low levels of dietary phytanic acid (< 2 mg daily intake) and phytol [41].
We discovered elevated RBC phytanic acid levels in all great apes relative to humans on both high-(Western) and low-phytanic acid (vegan) diets (Fig. 2). In principle, these could be influenced by differences in phytanic acid retention and/or biosynthesis. Relevant to the retention hypothesis, both humans and great apes showed robust cellular rates of phytanic and pristanic acid oxidation (Fig. 3). In fact, there were no differences in phytanic acid oxidation rates in human relative to chimpanzee or bonobo fibroblasts (Fig. 3A). Although it is possible that the lower phytanic acid oxidation rates in gorilla and orangutan relative to human fibroblasts (Fig. 3A) relate to the dietary strategies of these species [2], these observations need to be validated in larger-scale studies prior to commentary. We also cannot exclude the possibility that the cultured fibroblasts do not fully reflect in vivo activities; however, they are invaluable for the clinical diagnosis of phytanic acid oxidation disorders [61]. Likewise, we cannot exclude the possibility that great apes, unlike humans, show robust biological retention of trace dietary phytanic acid. However, laboratory mice do not accumulate significant physiological levels of phytanic acid when fed standard diets not supplemented with additional phytol or phytanic acid [53]. In addition, rats on standard diets can rapidly catabolize artificially elevated stores of phytanic acid in their tissues [62].
Our preferred hypothesis is that all the great apes, unlike humans [45,63], are capable of accumulating significant amounts of phytanic acid from the hindgut fermentation of plant materials. We propose that the great apes have maintained the ancestral condition and that humans diverged since their most recent common ancestor with chimpanzees and bonobos and now have a derived metabolism. We believe that changes in gut anatomy provide a likely basis for the proposed reduction of hindgut fermentation activity in the human lineage. In this regard, it has been proposed that improved dietary quality contributed to hindgut reduction in modern humans relative to great apes (i.e. the hindgut comprises 20% of the total human gut volume, but >50% of the total great ape gut volume) [2,3] and the smaller total gut volume to body mass ratio in modern humans relative to great apes [6,7,64,65]. However, we also recognize that the human and great ape gut microbiomes, whose composition are influenced by the diets of these species and other environmental interactions [66,67], could also contribute to the proposed differences in hindgut fermentation activity. Thus, in principle, the improved dietary quality in the human lineage could have influenced the major factors (gut anatomy and microbial activities) that we propose are most likely to be causally responsible for the observed differences in human and great ape RBC phytanic acid levels.
Given that sexual dimorphism in human lipid metabolism is well-documented [68,69], we screened for sexspecific differences in RBC phytanic acid levels. Here, we found no evidence of sexual dimorphism in RBC levels in humans on Western or vegan diets. However, we observed 1.4-fold elevated RBC phytanic acid levels in male relative to female chimpanzees (Fig. 2B). The evolutionary implications and physiological origin of this sexual dimorphism are unclear. Phytanic acid profiling in additional males and females from other non-human primates could shed light on the extent of this phenomenon. As discussed above, phytanic acid levels are affected by multiple factors, including gut anatomies, microbiomes, and gene expression, which provide candidate mechanisms for the observed sexual dimorphism in chimpanzees.
We note that sexual dimorphism in phytanic acid metabolism has been documented in rodents. C57BL/6J laboratory mice show sexually dimorphic phytanic acid accumulation in response to phytol-enriched diets; however, unlike chimpanzees, females showed higher phytanic acid levels than males [53,70]. There is no evidence as to whether this is a metabolic adaptation to diet or a nonspecific result of inbreeding. It is thought this sexual dimorphism is influenced by the higher liver expression of sterol carrier protein-x (SCP-x) in males relative to females [53]. SCP-x is required for branched chain fatty acid catabolism through its keto-acyl CoA thiolase activity [53,70]. Sexual dimorphism in liver SCP-x expression has also been observed in FVB mice [53], BALB/c mice [71], and laboratory rats [72]. This suggests that sexual dimorphism in phytanic acid metabolism could be present in additional rodent species. To date, SCP-x expression has not been measured in human and chimpanzee livers, even in a recent study involving both sexes from each species [73].
Here, we found that multiple genes reported to be influenced by PPAR-alpha activity, including those involved in peroxisomal αand β-oxidation of fatty acids, show crossspecies expression differences in the chimpanzee and human tissues surveyed (Fig. 4). Such genes were primarily more highly expressed in humans relative to chimpanzee for all tissues except kidney, where higher expression in chimpanzees was frequently found. Exploratory work, not involving non-human primates, highlighted the lower fatty acid α-oxidation capacity in the kidneys, but not livers, of carnivorous relative to herbivorous animals [74]. Also of interest, others have provided gene expression-based evidence that PPAR-signaling is higher in human relative to chimpanzee liver [75]. The elevated FABP1 transcript levels in human liver suggests an increased capacity for humans to transport phytanic acid to peroxisomes for degradation [53]. Nevertheless, we recognize phytanic acid is just one of several activators of PPAR transcription factors and that the dietary information and RBC lipid profiles of the humans and chimpanzees in the gene expression profiling studies are unknown.
We have previously noted that the toxicity associated with abnormal phytanic acid accumulation in humans could provide selective pressure to maintain or enhance its degradation [76]. For example, Refsum disease is an autosomal recessive disorder in humans that can result from mutations in the PHYH gene and subsequent accumulation of phytanic acid in tissues [46,47]. This typically affects the peripheral nerves (i.e. polyperipheral neuropathy), movement (i.e. cerebellar ataxia), and senses (e.g. retinitis pigmentosa, anosmia, and hearing loss) [43][44][45]. These individuals can also show cardiac arrhythmias, skeletal malformations (e.g. shortened metacarpals or metatarsals), and skin changes (i.e. ichthyosis) [43][44][45]. Given that cardiac arrhythmias are suspected to be responsible for the sudden deaths of Refsum disease patients [43], the commonly observed sudden cardiac death, presumed secondary to fatal arrhythmias, in captive male chimpanzees [77] could be influenced by their physiological levels of phytanic acid. However, the RBC phytanic acid levels in our cohort of male chimpanzees are far below those of Refsum disease patients, which was approximately 100 times higher than normal in one case we evaluated. Nevertheless, we raise the possibility that phytanic acid metabolism could influence physiological differences and medical conditions amongst human and non-human primates.

Conclusions
The candidate differences in human and great ape phytanic acid metabolism are consistent with prior reports that the regulation of genes involved in lipid metabolism likely evolved under directional selection in humans [75] and that the promoter regions of nutrition-related genes are under positive selection in humans [78]. It is also relevant that the effects human diets have on mouse liver transcriptomes can be used as a model for expression differences between humans and chimpanzees [79]. We note that even if transcriptome data of humans and chimpanzees on matching diets were obtained, differences in how each species digests their food and extract regulators of transcription would have a major impact on the observed cross-species and sex-specific differences.
To further explore such hypotheses, it will be necessary to determine if the RBC phytanic acid profiles in our cohort extend to other cell types. The comparative analyses of the lipid composition of human and great ape nervous and cardiovascular systems will be of special interest given the clinical phenotypes observed in humans with phytanic acid metabolic [43][44][45] and other peroxisomal disorders [86]. In addition, prostate tissue could be of interest given the differences of phytanic acid metabolic activities in normal and cancerous tissues [87][88][89] and male reproductive differences across human and non-human primates [90]. Lastly, we propose that larger-scale lipid profiling studies will likely identify additional candidate metabolic adaptations to human and non-human diets and that investigations into the microbial contributions toward specific metabolic differences is warranted.

Cohort for red blood cell lipid profiling
Bloods from adult humans with Western diets were collected from healthy individuals attending an international conference. Bloods from adult humans on vegan diets for over one year were collected in conjunction with a blood donor center. Appropriate Institutional Review Board (IRB) approval from the University of Southern California and Johns Hopkins Medicine was obtained for all human subjects research. Chimpanzee blood was collected at the Alamogordo Primate Facility. All chimpanzees took part in daily enrichment activities to maintain psychological well-being. Animals were maintained in accordance with the Guide for the Care and Use of Animals (U.S. Dept. of Health and Human Services, Public Health Service, Bethesda, MD., 1996). The APF and its program were fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, International (AAA-LAC). Other great ape bloods were collected at the Zoological Society of San Diego. The gender and ages of blood donors are provided in Additional File 2. One gorilla blood donor, described in reference [91], carries a deletion of the distal q arm of chromosome 3 (the homolog of human chromosome 4) as indicated in Additional File 4. Great ape diets contain fresh fruits, vegetables, and nutritional biscuits. Capillary gas chromatography (GC) -electron-capture negative-ion mass spectrometry [92] analysis of nutritional biscuits indicated the great apes were exposed to <2 mg of daily phytanic acid intake.

Red blood cell lipid profiling
Whole blood samples were collected from fasting individuals and stored in EDTA blood collection tubes. RBCs were collected by centrifugation, washed twice with physiological saline, transferred to freezer vials, flushed with nitrogen, and stored at -80°C until analysis. RBCs were thawed briefly before 100 μl aliquots were taken for analysis of the total lipid fatty acid content by capillary GC -electron-capture negative-ion mass spectrometry [92]. Processed data are provided in Additional File 4.

Primary fibroblast cultures
Great ape dermal fibroblasts were obtained from the Zoological Society of San Diego while human dermal fibroblasts were obtained from the Coriell Institute for Medical Research or the Kennedy Krieger Institute. All individuals are thought to be unrelated. The gender, age, and biopsy site of all fibroblast donors and corresponding biochemical analyses are provided in Additional File 2. Fibroblasts were cultured as previously described [76].

Phytanic and pristanic acid oxidation in cultured fibroblasts
Fatty acids were dried under nitrogen and solubilized with β-cyclodextrin (Sigma; 10 mg/ml in 10 mM Tris, pH 7.5) by warming to 37°C and sonication. Fibroblasts were grown to approximately 80% confluence in 12-well culture plates under standard conditions. Afterwards, the media was removed and replaced with 0.6 mL of serum-free culture media supplemented with 10 μM [1-14 C]-phytanic acid (~40,000 dpm/nmol) or 10 μM [1-14 C]-pristanic acid (~20,000 dpm/nmol). The cells were incubated for two hours (pristanic) or four hours (phytanic) at 37°C in 5% CO 2 . Reactions were terminated by addition of 0.12 ml of 2.6 N HClO 4 . Water-soluble reaction products were separated from the labeled substrates and quantified by liquid scintillation counting. The specific activity of phytanic acid or pristanic acid oxidation (pmoles/hour/mg protein) were calculated based on the amount of water-soluble radioactivity in the test sample minus the blank per mg of protein, determined using the Lowry method [93]. All primary data are provided in Additional File 5.

Statistical considerations
We analyzed all data on the log2 scale. We used analysis of variance (ANOVA) to compare average blood lipid data across humans and great apes. Heterogeneity P-values are reported for the test that the mean level is different in at least one of the great apes groups, and Wald P-values for tests comparing the average level in individual non-human primate groups to humans. Under ANOVA, statistical tests use an estimate of within-group variation from all samples. Due to the unbalanced group sizes, this estimate is driven by the variation in humans and in chimpanzees. For gene expression studies, we tested for differential expression (Fig. 4) using ANOVA and standard F-tests. To account for testing differential expression across multiple genes, we adjusted p-values using the Benjamini and Hochberg approach for controlling the false-discovery rate (FDR). All analyses were done using the R programming language.