We have greatly expanded previous surveys of bile salt structural diversity in reptiles and mammals. The variation of bile salts is consistent with current molecular-based phylogenies of reptiles and mammals, including the recent reclassification of squamate groups based on DNA sequencing data and studies showing that pinnipeds form a monophyletic group with Carnivora . Perhaps the most unusual finding in our study was that the Paenungulates (elephants, manatees, and the rock hyrax) have a very different bile salt profile from the Rufous sengi and South American aardvark, two other mammals classified with Paenungulates in the cohort Afrotheria by molecular phylogenetic analysis. It will be of interest to determine the molecular basis for these differences. As we discuss below, this finding could be accounted for by differences in a single enzyme, CYP27A1.
Our analysis did not reveal any obvious relationship between diet and bile salt profiles, although a major caveat to this analysis is the non-random nature of the animals found in our survey of reptiles and mammals. In reptiles, the analysis is further limited by the preponderance of carnivores in the surveyed sample (e.g. snakes, varanid lizards, crocodilians), with far fewer herbivores or omnivores. We also did not find any evidence that bile salt complexity is associated with a more complex omnivorous diet in either mammals or reptiles. The lack of association of bile salt profile and diet is also suggested by groups of animals that have very similar bile salt profiles but different diets. Examples include primates, Ursidae (pandas as herbivores, other bears as carnivores or omnivores), other animals in Carnivora excluding pinnipeds, and Testudines. Overall, common bile acids such as CA and CDCA, or other C24 bile acids such as ursodeoxycholic acid, are found in species with varied diets. With uncommon bile acids, there is insufficient data to make generalizations. For example, pythocholic acid (Boidae and Pythonidae) or varanic acid (Varanidae) could theoretically be advantageous to a carnivorous diet, pending more extensive surveys of reptiles and mammals.
Our analysis of ancient human and extinct giant ground sloth coprolites demonstrates the stability of bile acids in biological specimens preserved in arid climates. Previous studies have analyzed bile acids and sterols from human coprolites or internal organs preserved within mummies, with ages ranging from approximately 500 to 3,200 years old [40–42]. Our results suggest that bile salts may be useful markers in select paleontological studies.
Our use of multiple analytical techniques (e.g., HPLC, ESI/MS/MS, HPLC, NMR) allowed us to precisely resolve complicated bile salt profiles. We speculate on what these findings mean for the understanding of the evolution of bile salt synthesis, a complex and key biochemical pathway that permits cholesterol excretion to be regulated and at the same time generates amphipathic molecules with multiple physiological functions. In analyzing the patterns of bile salt variation across vertebrates, there appears to be at least two major pathways in the evolutionary transition from C27 bile alcohols (ancestral) to C24 bile acids (derived): (a) a 'direct' pathway and (b) an 'indirect' pathway that uses C27 bile acids as an 'intermediate' step. Evidence to support pathway (a) comes mainly from teleost fish families (e.g., Perciformes) where either type III (C27 bile alcohols and C24 bile acids but no C27 bile acids) or type VI (C24 bile acids only) are found, but not fish with any appreciable amount of C27 bile acids . This suggests that the ancestors of these fish followed an evolutionary transition from C27 bile alcohols directly to C24 bile acids. This differs from reptiles (e.g., Crocodylia, Testudines, varanid lizards), some mammals, and amphibians, where C27 bile acids are common [2, 3, 5].
One of the major goals of our study is to provide testable hypotheses for determining what bile salt enzyme differences underlie cross-species differences in bile salt profiles. Cross-species comparisons have already yielded some insights into different in bile salt profiles. An example is the discovery in pigs of CYP4A21, an enzyme that catalyzes 6α-hydroxylation of bile acids , a bile acid modification not found in humans. One unanswered question is how a bile salt synthetic pathway 'stops' at C27 bile alcohols, as must occur in animals such as Paenungulates, rhinoceroses, and zebrafish whose bile salt profiles do not show C27 or C24 bile acids. One possible explanation would involve the multi-functional capabilities of CYP27A1. Animals synthesizing only C27 bile alcohols would have CYP27A1 enzymes that can 27-hydroxylate but not oxidize the side-chain of the resulting C27 bile alcohols. It would be of interest to test the substrate specificities of CYP27A1 from species that synthesize mainly bile alcohols. No high-resolution crystal structure of CYP27A1 is currently available, but a homology model of human CYP27A1 has been developed  which could be used for comparative purposes to rationalize cross-species differences in substrate and catalytic specificity.
There is the additional question of when did the fatty acid and bile salt β-oxidation enzymes develop in peroxisomes. Three of these enzymes have strong candidates for invertebrate orthologs, suggesting a long evolutionary history, but the parallel ability for bile salt β-oxidation must have been a strictly vertebrate innovation given the occurrence of bile salts only in vertebrates. The zebrafish is an example of an animal that produces only C27 bile salts, yet possesses putative orthologs to genes for all four peroxisomal enzymes involved in shortening of the bile acid side-chain in mammals . Evolutionary changes of peroxisomal transporters such as ATP-binding cassette transporter D3 (ABCD3) [45, 46] may have been important in this regard by facilitating substrate entry into the peroxisomes (beyond passive diffusion), possibly explaining how an animal can maintain a bile salt profile of mainly C27 bile acids while having all the enzymes capable of C27 bile salt side-chain cleavage to form a C24 bile acid. There have been limited studies of peroxisomes in non-mammalian species, and we are not aware of any published studies of this organelle in reptiles. Detailed studies of peroxisomes in reptile groups that produce only C27 bile acids (e.g., turtles, varanid lizards, crocodilians) could be especially helpful in understanding cross-species differences in bile salt biosynthesis, particularly modification of the side-chain.
Genomic comparisons may also provide some potential insight into the synthesis of 5α-bile salts. The sporadic appearance of 5α-bile salts in otherwise evolutionarily recent animals suggests that it is the result of mutation in a biosynthetic enzyme (a likely candidate is AKR1D1). Trace levels of 5α-bile acids are ubiquitous in bile, but the origins of these small concentrations are likely produced by microbial flora and not from biosynthesis . The presence of 5α-bile acids in the bile of germ-free rabbits indicates that at least some mammals are capable of synthesizing 5α-bile acids . The hepatic enzymes(s) (if present) that perform 5α-reduction of bile salt intermediates like 4-cholesten-7α,12α-diol-3-one are as yet uncharacterized .
Model animals such as the anole lizard and zebrafish may be helpful in determining the enzymatic pathways for generating 5α bile salts. For example, we did not find a putative ortholog to the gene for AKR1D1 in the zebrafish genome. We did, however, find likely orthologs to the mammalian genes for 5α-reductase 1 (SRD5A1) and 2 (SRD5A2) in the zebrafish genome. It is possible that either or both of these enzymes, or another enzyme altogether, catalyzes 5α-reduction of the steroid nucleus in zebrafish bile salt biosynthesis. In contrast, a putative ortholog for AKR1D1 was identified in the genome of the anole lizard. Studies of anole lizard AKR1D1 are required to see if this enzyme catalyzes 5α-reduction of bile acid intermediates (unlike human and rodent AKR1D1) or 5β-reduction for the trace amounts 5β-bile acids present in anole bile. There are also ongoing sequencing projects for sea lamprey (Petromyzon marinus) and coelacanth (Latimeria calumnae), two animals that exclusively produce 5α bile alcohols [9, 12, 49]. Comparison of the genomes of anole lizard, zebrafish, lamprey, and coelacanth may also provide insight into how they achieve a bile salt pool of 5α-bile salts. Additional File 5 includes a summary of model animals whose study may provide insight into the mechanisms underlying cross-species differences in bile salt profiles.
Our study does not shed light on how animals regulate their bile salt profiles. In Additional File 3 we include some spectra focusing on the minor bile salts of various species revealing there can be a complex mixture of bile alcohols at low concentration even in species whose biliary bile salts are comprised of greater than 95% bile acids. Whether these minor bile salts perform physiological functions or are simply vestigial is currently unknown. It is logical to assume to that both bile acid biosynthesis and intestinal conservation are mainly controlled by transcriptional regulators, although this is not well-understood even in mammals.
There is still little understanding of what drives variation in bile salt chemical diversity. As our analysis of reptiles and mammals shows, there is essentially no apparent link between diet and bile salt structure. Typical 5β-C24 bile acids are found in reptiles and mammals with a wide variety of diets. It should be noted that all described hydroxylation sites occur on the hydrophilic face of the bile salt molecule, keeping the amphipathicity intact . It is also unclear what benefit may be conferred to those reptiles (e.g., in the group Iguania) that use 5α bile salts. Studies in mammals have shown that 5α bile acids can be toxic. One striking example of this occurs in the rabbit. Rabbits fed 5α-cholestanol (the saturated homolog of cholesterol) form 5α-cholic acid, which is 7-dehydroxylated by gut bacteria to form 5α-deoxycholic acid and subsequently conjugated with glycine in the liver. However, the glycine conjugate of 5α-deoxycholic acid acid precipitates from solution in the gallbladder, forming gallstones [50, 51]. Animals with a bile salt pool of mostly 5α-bile acids evidently have mechanisms to avoid toxicity.
The bile salt biosynthetic pathway has apparently evolved in vertebrates in a considerably different fashion from the adrenocortical and sex steroid hormones, another set of compounds derived from cholesterol. Estrogens are the terminal products of the steroid hormone pathway whereas other steroids such as glucocorticoids or mineralocorticoids are intermediates. The actions of the steroid hormones are mediated by nuclear hormone receptors. Thornton and colleagues have used a variety of techniques, including phylogenetic reconstruction and expression of ancestral proteins, to provide evidence that estrogen receptors have a longer evolutionary history than receptors for intermediate products (e.g., glucocorticoids) in the steroid hormone pathways [52–54] (although their ancestral reconstruction in invertebrates have been challenged by other research ). In the 'molecular exploitation' model of a biochemical pathway, the terminal products of a synthetic pathway mediate the more ancestral activities. During evolution, functions (and the receptors that mediate these functions) develop for the intermediate products of the pathway, and structural diversity involving intermediates in the pathway can increase over time. In the case of the steroid hormones, the major evolutionary changes probably occurred in early vertebrate evolution or even earlier.
The diversity of bile salts, starting in fish and amphibians  and combined with the data for reptiles and mammals in the present study, suggests a different model of evolution of bile salts from steroid hormones. In the case of bile salts, the terminal products are increasing in diversity throughout all vertebrate classes. A parallel to the bile salt synthetic pathway may be found in the enzymatic pathways involved in the formation of triterpenoids, isoprene-derived 30-carbon molecules that serve as a building block for many other compounds. Studies of bacteria, plants, and animals demonstrate at least two independent pathways for the formation of isoprene units, each involving multiple organelles . Similar to bile salts, the synthetic intermediates in the triterpinoid pathways are low in diversity but the end-products have impressive biochemical diversity and mediate an array of different physiological functions. Like triterpinoids, the end-products of bile salt synthesis have substantial structural diversity by variations in hydroxylation patterns, unsaturation, and side-chain length. Snakes and marsupials are two groups that demonstrate ongoing 'innovations' in bile acids not seen in fish and amphibians (e.g., 1α- and 1β-hydroxyation in the spotted cuscus and feathered glider, respectively; and Δ22 and C23 bile acids in true viper snakes).
Bile acid synthesis is now known to be catalyzed in humans and rodents by at least two independent pathways (acidic and neutral), with tight regulation keeping intermediate products at very low concentrations in the hepatocytes and biliary bile [6, 7]. In some animals such as the crotaline snakes, the result is a bile salt pool consistently of almost 100% of a single bile salt molecule such as CA. The bile salt synthetic pathway has evolved from a relatively simple pathway that produces C27 bile alcohols (by simple saturation and polyhydroxylation of cholesterol as in hagfish)  to more complex pathways that utilize multiple organelles (cytoplasm, endoplasmic reticulum, mitochondria, and peroxisomes) to produce C24 bile acids. In the process, many ray-finned fish and most land animals have bile containing almost exclusively C24 bile acids and very low amounts of the synthetic precursors including C27 bile alcohols, a process achieved by the evolution of regulatory control and enzyme modifications.
Although there is extensive data on the physicochemical and physiological properties of common bile acids such as CA and CDCA (and their conjugated derivatives), there has been much less study of other classes of bile salts. In part, this relates to the difficulty of synthesizing or isolating sufficient material to perform detailed studies. The complicated variation of bile salts may also relate to other functions of bile salts that are not well understood, including communication, influencing of the microbial environment in the gut, and regulation of hepatobiliary development and regeneration . It has been proposed that bile acid structural variation may be driven by attempts to prevent gut microbial 'damage' of bile acids that can lead to toxic secondary bile acids such as DCA and LCA [1, 8, 57]. Novel modifications to the stem bile salt structures throughout vertebrate evolution make it difficult for bacteria to alter the bile salt structure into molecules that are toxic or have poor aqueous solubility or both.
Finally, the extensive variation of bile salt structures across vertebrate species predicts that protein receptors that bind bile salts will also show variation in structure and ligand-binding specificity. Indeed, research has already shown distinct patterns of ligand-receptor co-evolution for three nuclear hormone receptors (farnesoid X receptor, FXR; pregnane X receptor, PXR; and vitamin D receptor, VDR) that regulate various aspects of bile salt synthesis, transport, or metabolism [10, 14, 58–60]. Model species that have bile salt profiles different from humans, such as anole lizard, rock hyrax, and African elephant, may be particularly interesting to study, as their bile salt-regulating nuclear hormone receptors would have to bind different ligands (e.g., bile alcohols or 5α-bile salts) than their human or rodent orthologs which need to recognize C24 5β-bile acids. Ultimately, we aim to build up a complete multi-dimensional picture of ligand, protein, commensal microbial, and species evolution involving bile salts.