|Home | About | Journals | Submit | Contact Us | Français|
For currently unknown reasons, the evolution of CYP3A4 underwent acceleration in the human lineage following the split from chimpanzee. We investigated the significance of this event by comparing E. coli-expressed CYP3A4 from humans, chimpanzee, and their most recent common ancestor. The expression level of chimpanzee CYP3A4 was ~50% of the human CYP3A4, whereas ancestral CYP3A4 did not express in E. coli. Steady-state kinetic analysis with 7-benzyloxyquinoline, 7-benzyloxy-4-(trifluoromethyl)coumarin (7-BFC), and testosterone showed no significant differences between human and chimpanzee CYP3A4. Upon addition of α-naphthoflavone (25 μM) human CYP3A4 showed a slightly decreased S50 for 7-BFC, whereas chimpanzee CYP3A4 showed a > 2-fold increase. No significant differences in inhibition/activation were found for a panel of 43 drugs and endogenous compounds, suggesting that the wide substrate spectrum of human CYP3A4 precedes the human-chimpanzee split. A striking exception was the hepatotoxic secondary bile acid lithocholic acid, which at saturation caused a 5-fold increase in 7-BFC debenzylation by human CYP3A4, but not by chimpanzee CYP3A4. Mutagenesis of human CYP3A4 revealed that at least 4 out of the 6 amino acids positively selected in the human lineage contribute to the activating effect of lithocholic acid. In summary, the wide functional conservation between chimpanzee and human CYP3A4 raises the prospect that phylogenetically more distant primate species such as rhesus and squirrel monkey represent suitable models of the human counterpart. Positive selection on the human CYP3A4 may have been triggered by an increased load of dietary steroids, which led to a novel defense mechanism against cholestasis.
Cytochrome P450 3A4 (CYP3A4) is considered the most important drug-metabolizing enzyme in humans due to its abundant yet variable expression combined with an unusually wide substrate spectrum. CYP3A4 accounts on average for 30% of the total P450 expression in liver and up to 80% in the small intestine (Shimada et al., 1994; Paine et al., 2006). The individual expression of CYP3A4 varies up to 90-fold (Lamba et al., 2002), which is in part determined by still unknown gene variants in the enzyme and its transcriptional regulators (Ozdemir et al., 2000). The inherited variability in CYP3A4 expression may be further enhanced by the transcriptional induction of the CYP3A4 gene and by the inhibition of its protein product by many drugs and dietary constituents (Wilkinson, 2005).
The CYP3A4 substrate spectrum in humans is much wider and the catalytic activity frequently higher than the “minor” CYP3A isoforms CYP3A5 and CYP3A7 (Daly, 2006). CYP3A4 metabolizes an estimated 50% of the currently used drugs (Wilkinson, 2005). The enzyme is capable of accommodating large molecules such as cyclosporine and bromocriptine and exhibits non-Michaelis-Menten kinetics toward some substrates (Atkins, 2005). These characteristics are thought to be due to either multiple ligand-binding sites (Kenworthy et al., 2001; He et al., 2003), and/or to multiple, functionally distinct conformers (Davydov et al., 2003; Tsalkova et al., 2007; Fernando et al., 2008). One of the consequences of CYP3A4 ligand promiscuity is its prominent involvement in clinically relevant drug-drug interactions (Wilkinson, 2005).
Using genomic sequences from 16 species we recently investigated the evolution of Cyp3 genomic loci over a period of 450 million years (Qiu et al., 2008). We detected two noteworthy recent episodes of particularly strong positive selection acting on primate CYP3A protein-coding sequence. CYP3A7 underwent changes early in the hominoid evolution, which was accompanied by a restriction of its hepatic expression to the fetal period. CYP3A4 acquired changes to six amino acids following the split of the chimpanzee and human lineages approximately 6 million years ago (Fig. 1 and Supplemental Table 1). These findings suggested the acquisition of new, presently unknown catalytic functions of CYP3A7 and CYP3A4 especially important for the evolution of hominoids and humans, respectively.
In the present work we investigated the cumulative effect of the human-specific changes to the CYP3A4 protein sequence following the human-chimpanzee split. We hypothesized that these changes may have widened the substrate spectrum of the human CYP3A4 isozyme. Alternatively, the wide substrate spectrum may have preceded the human-chimpanzee split. In this latter case, the acceleration of the CYP3A4 evolution in the human lineage would most likely reflect the adaptation to a change in the physiology or environment of our direct ancestors. To differentiate between these hypotheses, we inferred using a phylogenetic approach the sequence of CYP3A4 in the most recent common ancestor of humans and chimpanzees. We then set out to express all three enzymes in E. coli, in order to compare their activities, including the inhibition and/or activation by a panel of exo- and endogenous CYP3A4 substrates. The uncovered differences in enzyme activities were further characterized using two chimpanzee and six human CYP3A4 mutants.
7-Benzyloxyquinoline (7-BQ), 7-hydroxyquinoline (7-HQ), 7-benzyloxy-4-(trifluoromethyl)coumarin (7-BFC), and 7-hydroxy-4-(trifluoromethyl)coumarin (7-HFC) were purchased from Molecular Probes, Inc. (Eugene, OR). NADPH, α-naphthoflavone (ANF) and all the drugs and steroids were bought from Sigma-Aldrich (St. Louis, MO). Ni-NTA affinity resin was purchased from Qiagen (Valencia, CA). All other chemicals were of the highest grade available and were obtained from standard commercial sources. Recombinant CPR and cytochrome b5 (b5) from rat liver were prepared as described previously (Harlow and Halpert, 1997).
Human CYP3A4 expression vector (pSE3A4HisTag) was double-digested with the restriction enzymes NcoI and KpnI (New England Biolabs, Frankfurt, Germany) at 37 °C for 4 hours followed by 5 minute incubation at 90 °C for enzyme inactivation. The digestion mixture was run on an agarose gel in 1 X TAE buffer and the vector band (5~6 kb) was extracted using Gel Extraction Kit (peQlab, Erlangen, Germany). cDNA derived from chimpanzee total liver RNA was from a previous study (Qiu et al., 2008). The CYP3A4 protein coding region was first amplified using a high fidelity Taq polymerase (Fermentas, St. Leon-Rot, Germany) with primers 3A4e1F (5′- AAAGAGCAACACAGAGCTG-3′) and 3A4e13R (5′- GTCCTTAGGAAAATTCAGG-3′). The products were diluted and used as templates for a second round of polymerase chain reaction (PCR) to amplify the coding region (codon 19-503). Modifications at the two ends of the open reading frame (ORF) (i.e., truncation at the amino-terminus and inclusion of four His-tags at the C-terminus) were introduced by PCR reaction via the primers (3A4-F-NcoI: 5′- GGGGGCCATGGCTCTGTTATTAGCAGTTTTTCTGGTGCTCCTCTATCTATATGGAAC-3′) and 3A4-R-KpnI: (5′- CGGCGGGGTACCTCAGTGATGGTGATGGGCTCCACTTACGGTGCC-3′). PCR products were digested with NcoI and KpnI and ligated with digested plasmid DNA at room temperature for 4 hours using T4 ligase (New England Biolabs, Frankfurt, Germany). The resulting expression construct containing the chimpanzee CYP3A4 coding region was designated as pSE-pt-3A4-HisTag.
Since the human/chimpanzee divergence, six and two nonsynonymous mutations have accumulated in the human and chimpanzee CYP3A4, respectively (Fig. 1 and Supplemental Table 1). The sequence of their most recent common ancestor was reconstructed as described (Qiu et al., 2008), i.e. using baseml included in the PAML package 3.15. As in-files we used i) an alignment comprising Cyp3A4 orthologs of M. mulatta, P. anubis, C. jacchus, P. pygmaeus, H. sapiens und P. troglodytes and ii) an unrooted tree reflecting the commonly accepted phylogeny among these species. The calculations were conducted using the HKY85 model. Ambiguous codon positions were removed from the dataset.
Chimpanzee CYP3A4 was used as a starting template to restore the CYP3A4 sequence predating the chimpanzee/human split, since only two artificial mutations (C495g and G1105a) were required. Two sequential rounds of site directed mutagenesis were conducted on the above described pSE-pt-3A4-HisTag plasmid using a QuikChange Site-Directed Mutagenesis Kit (Stratagene, Amsterdam, The Netherlands). Besides the putative ancestral CYP3A4, we also obtained plasmids separately expressing each of the two amino acid residues that differ from the chimpanzee enzyme (Asp165Glu and Val369Ile). Site-directed mutagenesis was also used to revert amino acids 54, 78, 129, 224, 478, and 489 of human CYP3A4 into those present in the ancestral CYP3A4. Sense sequences of the oligonucleotides used in mutagenesis are given in Supplemental Table 2.
CYP3A4 and mutants were expressed as His-tagged proteins in E. coli TOPP3 and purified using a Ni-affinity column as described previously (Domanski et al., 2001). Protein concentrations were determined using the Bradford protein assay kit (BioRad, Hercules, CA). The specific contents of the human and chimpanzee CYP3A4 were 14 and 8 nmol P450 per mg protein, respectively.
The standard NADPH-dependent assays with 7-BQ, 7-BFC, and testosterone were essentially carried out as described (Kumar et al., 2006). Enzyme inhibition or activation was measured using the 7-BFC O-deethylation assay in a final reaction volume of 100 μl at the substrate concentration of 20 μM as described previously (Domanski et al., 2001; Oezguen et al., 2008). Steady state kinetic parameters were determined by regression analysis using Sigma Plot (Jandel, San Rafel, CA). The S50 and n values were determined using the Hill equation. Each kinetic experiment included human CYP3A4 for more accurate comparison of the data. Thermal inactivation of P450 was carried out as described (Kumar et al., 2007). All data treatment and fitting of the titration curves for thermal inactivation were performed with our SpectraLab software as described (Davydov et al., 2003).
The heterologous expression of chimpanzee CYP3A4 was ~50% of the human CYP3A4 (57 vs. 120 nmol P450/liter culture). Upon purification, chimpanzee CYP3A4 showed a higher amount of P420 (19% vs. 4%) than human CYP3A4, suggesting lower protein stability. However, when tested for thermal stability (Supplemental Fig. 1), human and chimpanzee CYP3A4 showed indistinguishable Tm values of 53.8 ± 0.64 and 54.4 ± 0.67 °C, respectively. Ancestral CYP3A4 showed negligible heterologous P450 expression in E. coli. As there are only two amino acid differences, Glu165Asp and Ile369Val, between ancestral and chimpanzee CYP3A4 (Fig. 1 and Supplemental Table 1), we introduced separately these two mutants into the chimpanzee enzyme. Both Val369Ile (17 nmol P450/liter culture) and Asp165Glu (6 nmol P450/ liter culture) showed greatly decreased expression.
Steady-state kinetics of human and chimpanzee CYP3A4 oxidation of the model substrates 7-BQ, 7-BFC, and testosterone were examined in the absence and presence of ANF. The results are presented in Fig. 2 and Table 1. Chimpanzee and human CYP3A4 showed no significant difference in kcat with 7-BQ, 7-BFC, or testosterone. The only significant difference was that the chimpanzee enzyme showed > 50% lower S50 with 7-BFC than the human. In addition, there was a slight difference in the Hill coefficient with 7-BFC and testosterone (Table 1). Similar to human CYP3A4, the chimpanzee enzyme also showed heterotropic activation by ANF, which is characterized by an increase in kcat and/or decrease in n values. The most striking difference was that the human CYP3A4 showed a slightly decreased S50 in the presence of ANF for 7-BFC debenzylation and testosterone hydroxylation, whereas chimpanzee CYP3A4 showed an increase in S50 in the presence of ANF (Table 1).
To explore further whether a variety of xenobiotics and endogenous compounds that are known substrates, inhibitors, or activators of human CYP3A4 show different characteristics with chimpanzee CYP3A4, we performed activation/inhibition studies with both the enzymes using 43 compounds. The study was carried out using 7-BFC as substrate at 20 μM and appropriate inhibitor/activator concentrations. The results are presented in Table 2. As expected, a majority of ligands either inhibit or activate the metabolism of 7-BFC by human CYP3A4. In most cases, no significant differences with regard to inhibition or activation were found between human and chimpanzee CYP3A4. The most striking difference was that LCA strongly activated human CYP3A4 (3-fold), whereas no activation was observed with chimpanzee CYP3A4 (Table 2). In contrast, the primary bile acids cholic and chenodeoxycholic acid showed comparable (≤ 2-fold) activation of either P450 (Table 2). Cholic and chenodeoxycholic acid are metabolized by human hepatic cytochrome P450, especially by CYP3A4 (Deo and Bandiera, 2008). Furthermore, we determined the concentration dependence of the activation by LCA. The results are presented in Fig. 3. The LCA binding constant (KE) was 216 ± 20 for human CYP3A4 and 105 ± 25 for chimpanzee CYP3A4. The Vmax for human CYP3A4 was > 5-fold higher than the V0 (4.94 ± 0.53 vs. 27.3 ± 0.88 min−1). Chimpanzee CYP3A4 showed no significant activation (5.71 ± 0.20 vs. 6.70 ± 0.49 min−1).
In order to identify the molecular determinants of the activation of human CYP3A4 by LCA, we reverted individually the 6 amino acid mutations acquired by human CYP3A4 into the respective ancestral ones. The P450 expression of these CYP3A4 mutants was similar to the wild-type enzyme, except for V489I (Supplemental Table 3). Because of low expression and poor binding to the affinity column, V489I could not be purified in a sufficient amount to assess enzyme activity. The activity of the remaining 5 mutants towards 7-BFC was similar to the activity of wild-type human and chimpanzee CYP3A4 (Table 3). Four mutants (H54R, Q78R, L129I, and T224I) showed a significant decrease in the activation of CYP3A4 activity by LCA compared to wild-type human CYP3A4. Importantly, none of the mutations completely abolished the activity by LCA. S478R showed no significant decrease in the activation by LCA compared with wild-type human CYP3A4 (Table 3). Overall, the results suggest that most of the amino acid acquired by human CYP3A4 contribute to the activation by LCA.
The purpose of this work was to clarify the reasons for and effects of the unusually rapid changes to the protein sequence of human CYP3A4 following the human-chimpanzee split. Contrary to our original hypothesis, these changes are unlikely to be responsible for the wide substrate spectrum of human CYP3A4, which thus appears to have preceded the human-chimpanzee split. Instead, the amino acid substitutions may have been triggered by an increased load of dietary steroids, leading to a novel defense mechanism against cholestasis.
The above conclusions can be reached despite our inability to express the ancestral CYP3A4 made by introducing Asp165→Glu and Val369→Ile mutations into chimpanzee CYP3A4. This failure may reflect the limitations of the heterologous P450 expression in E. coli, due to diminished protein stability, decreased level of heme incorporation, and/or increased level of P420 formation (Kumar et al., 2007). Alternatively, the prediction of the ancestral CYP3A4 sequence may have been erroneous, resulting in the inclusion of an amino acid incompatible with expression. Since this latter possibility is extremely unlikely considering the small number of differing amino acids, we investigated whether Asp165→Glu or Val369→Ile substitutions disrupted salt bridges and/or hydrogen bonds. To this end, we created models of chimpanzee and ancestral CYP3A4 based on the crystal structure of human CYP3A4 as shown in Fig. 1. Detailed in silico analysis suggested no apparent change in electrostatic interactions and/or hydrogen bonds except that Val-489 of human CYP3A4 has an additional H-bond with Gly-480 (data not shown). This is in contrast to recent results with CYP2B4dH where an Asp192→Ala substitution abolished hydrogen bonds and decreased Tm by > 10° C (Oezguen et al., 2008).
Although the ancestral CYP3A4 was not available for activity analysis, the wide substrate spectrum of human CYP3A4 likely precedes the human-chimpanzee split. This conclusion is supported by the paucity of major activity differences towards a large effector panel between human and chimpanzee CYP3A4 which differs from the ancestral CYP3A4 in only two amino acids. The absence of activity differences is consistent with the sympatric evolution of human and chimpanzee lineages in East Africa until approximately 100,000 years ago, since any major changes in the exposure to environmental chemicals would be expected to affect either lineage.
The most striking differences between human and chimpanzee CYP3A4 comprise the responses to ANF and LCA. ANF had opposite effects on the S50 values for 7-BFC debenzylation or testosterone hydroxylation by the human compared with the chimpanzee enzyme. Of the eight residues that differ between chimpanzee and human CYP3A4, Ile369 in the human enzyme is in the active site, as evidenced by X-ray crystallography (Yano et al., 2004). In addition, a Val substitution at that position, as in the chimpanzee enzyme, alters progesterone hydroxylation and sensitivity to activation by ANF (He et al., 1997). Otherwise, a comparison of models of the human and chimpanzee enzymes suggests no significant difference in their structures with an RMS value of 0.3 Å (Fig. 1). Accordingly, the structural basis for the opposite effects of ANF on the S50 values for 7-BFC and testosterone oxidation by the human as opposed to chimpanzee enzyme remains unclear.
The other difference was the strong activation of human (but not chimpanzee) CYP3A4 by LCA, a very hepatotoxic secondary bile acid generated by intestinal bacteria. LCA is a rare example of a toxic endobiotic, and several mechanisms have evolved to solve the problem of its efficient detoxification. The efficient sulfation of LCA, exclusive for humans and chimpanzees, has reduced its toxicity in these species (Hofmann, 2004). The conversion of LCA by CYP3A4 to several hydroxylated metabolites (Bodin et al., 2005) further facilitates their conjugation and excretion. Both LCA and its major metabolite, 3-dehydro-LCA (Bodin et al., 2005), are activators of the pregnane X receptor (Staudinger et al., 2001) and Vitamin D receptor (Makishima et al., 2002). In fact, the increasing complexity of the bile acid composition during vertebrate evolution may have affected the ligand specificity of PXR (Krasowski et al., 2005), although the resulting induction of CYP3A4 is surprisingly modest (Stedman et al., 2004).
The activation of human CYP3A4 by LCA reported in our present work would be expected to increase the detoxification of this and other bile acids metabolized by the enzyme. In contrast to LCA, no activation differences were detected in response to the less toxic LCA precursor chenodeoxycholic acid and the other major primary bile acid, cholic acid. This suggested a previously unknown defense mechanism against LCA-mediated cholestasis, which evolved following the split of the common human-chimpanzee lineage. The physiological necessity of such a mechanism may be related to our ancestors, beginning with Homo erectus some 1.8 million years ago, having adapted to an energy-dense, meat-based diet. Contemporary human foraging populations derive more than half of their dietary energy from animal foods, in comparison to 5-10% observed in chimpanzees. This adaptation may have been a pre-requisite for the subsequent dramatic increase in the brain size in the human lineage (Leonard et al., 2007). Importantly, meat-based diet increased the load of animal steroids and thus the risk of cholestasis.
Alternatively to the increased load of animal steroids, the activation of human CYP3A4 by LCA may represent a side-effect of the adaptive response to another selecting stimulus. In the former case, most of the amino acid differences between human and ancestral CYP3A4 would be expected to affect the magnitude of CYP3A4 activation by LCA. In the latter case, most of the LCA responsiveness would be expected to be a chance effect of only one amino acid substitution (Zhang et al., 2002). In order to distinguish between these possibilities, we reverted individually the 6 amino acid mutations acquired by the human CYP3A4 following the split from the chimpanzee. Mutating amino acid 478 from Ser to the ancestral Glu had no effect on activation by LCA, suggesting that Ser478 was selected by a factor different from LCA. This interpretation is consistent with this residue having undergone positive selection not only in the human lineage, but also across the whole primate CYP3A phylogeny (Qiu et al., 2008). The selecting factor may have been related to the overall activity and stereoselectivity, since mutating residue 478 reduced the activity of CYP3A4 by 80–90% and changed its regioselectivity toward aflatoxin B1 metabolites (Wang et al., 1998).
All four remaining expressing mutants displayed reduced activation in response to LCA, clearly supporting this substance as a selecting agent. The individual effects of the other four mutations on the activation by LCA were smaller than the difference between the human and chimpanzee. This suggests the collective effect of all sequence differences, although this would need to be verified by expressing combinations of these mutants as well as of the two amino differences (165 and 369) exclusive for human and chimpanzee. The importance to LCA activation of the latter two residues, which differ between human and chimpanzee but not between human and ancestral sequences, was not investigated, due to their apparently complex and species-specific effect on the CYP3A4 protein expression level. Indeed, 369Ile and 165Glu greatly decreased the expression of the chimpanzee CYP3A4, whereas they were compatible with high expression levels of human CYP3A4. Taken together, our results strongly suggest that LCA may have driven the evolution of CYP3A4 in the human lineage. This may constitute a rare example of incremental adaptive changes in response to a selective agent (Zhang et al., 2002).
Due to the importance of CYP3A4 in human drug metabolism, it is of substantial interest to find its appropriate animal model. Transgenic CYP3A4 mice (Gonzalez and Cheung, 2008) have their limitations, due to differences between rodents and primates in the regulation of CYP3A gene expression, as well as in the expression and activity of many other genes important to drug response. Our data indicate that chimpanzee CYP3A4 is an excellent model of its human counterpart. For ethical, logistical, and economical reasons, the use of chimpanzee in drug discovery should generally be avoided. However, the strong similarity between the chimpanzee and human CYP3A4 raises the possibility of function conservation between CYP3A4 of humans and of a primate more widely used in medical research, such as rhesus or squirrel monkey.
This research was supported in part by NIH Grant to J.R.H. [GM54995] and DFG grant WO505/2-2 (to L.W).