An Extensively Humanized Mouse Model to Predict Pathways of Drug Disposition and Drug/Drug Interactions, and to Facilitate Design of Clinical Trials.

Species differences in drug metabolism and disposition can confound the extrapolation of in vivo PK data to man and also profoundly compromise drug efficacy studies owing to differences in pharmacokinetics, in metabolites produced (which are often pharmacologically active), and in differential activation of the transcription factors constitutive androstane receptor (CAR) and pregnane X receptor (PXR), which regulate the expression of such enzymes as P450s and drug transporters. These differences have gained additional importance as a consequence of the use of genetically modified mouse models for drug-efficacy testing and also patient-derived xenografts to predict individual patient responses to anticancer drugs. A number of humanized mouse models for cytochrome P450s, CAR, and PXR have been reported. However, the utility of these models has been compromised by the redundancy in P450 reactions across gene families, whereby the remaining murine P450s can metabolize the compounds being tested. To remove this confounding factor and create a mouse model that more closely reflects human pathways of drug disposition, we substituted 33 murine P450s from the major gene families involved in drug disposition, together with Car and Pxr, for human CAR, PXR, CYP1A1, CYP1A2, CYP2C9, CYP2D6, CYP3A4, and CYP3A7. We also created a mouse line in which 34 P450s were deleted from the mouse genome. Using model compounds and anticancer drugs, we demonstrated how these mouse lines can be applied to predict drug-drug interactions in patients and discuss here their potential application in the more informed design of clinical trials and the personalized treatment of cancer.

Pi-class glutathione S-transferase: regulation and function.

Our laboratory has been involved in the study of Glutathione S-transferase pi (GST pi) for many years, both in terms of regulation of gene expression and in trying to understand the endogenous function(s) of this enzyme and also what role it may play in the carcinogenic process [1]. Over-expression of GST pi has been associated with carcinogenesis and the development of many different human tumours, for example testis [2], ovarian [3] and colorectal [4] and is often inversely correlated with prognosis or patient survival [5,6]. In addition, GST Pi has been implicated in the acquisition of antineoplastic drug resistance [7-9]. In order to study the transcriptional regulation of this gene, we have utilised a multi-drug resistant derivative (VCREMS) of the human mammary carcinoma cell line, MCF7, in which GST P1 mRNA and protein are significantly elevated in the absence of gene amplification [10-13]. Interestingly, we have recently reported the discovery of polymorphisms at the GSTP1 locus, resulting in two alleles GSTP1a and GSTP1b. In the study, the GSTP1b allele was found with increased frequency in bladder and testicular cancer, while the GSTP1a allele was significantly decreased in cases of prostate cancer [14]. In an attempt to elucidate the endogenous role(s) of GST pi, we have used homologous recombination in embryonic stem (ES) cells to inactivate both murine GST Pi genes and create a mouse strain completely deficient in the expression of this enzyme. This provides us with a unique animal model with which to study the effects of the absence of GST pi expression on the metabolism and pharmacokinetics of xenobiotics.

Transcriptional and post-transcriptional mechanisms can regulate cell-specific expression of the human Pi-class glutathione S-transferase gene.

Previous studies from this laboratory have identified transcriptional mechanisms that are utilized to increase expression of the human glutathione S-transferase gene GSTP1 in a multidrug-resistant derivative (VCREMS) of the human mammary carcinoma cell line MCF7 [Moffat, McLaren and Wolf (1994) J. Biol. Chem. 269, 16397-16402]. The data presented here provide strong evidence that post-transcriptional mechanisms can also play an important role in determining cell-specific expression of the GSTP1 gene. GSTP1 mRNA levels were shown to be elevated 3.1-fold in the human bladder carcinoma cell line EJ compared with VCREMS cells. Despite this observation, transient transfection assays revealed a decreased rate of GSTP1 promoter activity in EJ cells. Indeed, GSTP1 transcriptional repressor activity, mediated by a region located between nucleotides -105 and -86 (as we have previously described in MCF7 cells), was observed in EJ cells. However, in contrast with our results in MCF7 cells, the EJ repressor activity did not displace the essential nuclear complex bound to the C1 promoter element (-73 to -54) in vitro. In addition, competition experiments indicated that an AP-1-like protein is an integral component of the C1-bound complex in EJ cells. Interestingly, experiments utilizing actinomycin D to inhibit transcription demonstrated significantly greater stability of GSTP1 mRNA in EJ cells than in VCREMS cells. These findings suggest that cell-specific differences in the rates of GSTP1 mRNA decay provide the predominant mechanism responsible for elevated expression of the GSTP1 gene in EJ cells.

Characterization of the human cytochrome P4502D6 promoter. A potential role for antagonistic interactions between members of the nuclear receptor family.

The functional mapping of the human cytochrome P4502D6 (CYP2D6) promoter in HepG2 cells revealed the presence of both positive and negative regulatory elements. One of these regulatory elements overlapped a sequence that is highly conserved in most members of the CYP2 family. This element, which consists of a degenerate AGGTCA direct repeat spaced by 1 base pair (DR1) and is known to be a target for members of the steroid receptor superfamily, was found to bind in vitro translated hepatocyte nuclear factor 4 (HNF4) in gel retardation analysis. Using HepG2 nuclear extracts, three protein-DNA complexes were formed on the DR1 element, one of which was confirmed to be dependent on the binding of HNF4. The other DR1 complexes were shown to be due to the interaction of the orphan receptor chicken ovalbumin upstream promoter transcription factor I (COUP-TFI). Experiments in COS-7 cells showed that HNF4 could activate the CYP2D6 promoter 30-fold. Surprisingly, mutation of the DR1 element produced a relatively minor 23% decrease in activity in HepG2 cells. Additionally, COUP-TFI was shown to inhibit HNF4 stimulation of the CYP2D6 promoter in COS-7 cells, suggesting that COUP-TFI could attenuate the effect of HNF4 in HepG2 cells. However, when HNF4 levels were increased in HepG2 cells by co-transfection, it resulted in the enhancement of CYP2D6 promoter activity, indicating that HNF4 could overcome the repressive effect of COUP-TFI. Therefore, the contribution of the DR1 element in controlling the transcription of the CYP2D6 gene depends on the balance between positively and negatively acting transcription factors.

Functional characterization of the transcription silencer element located within the human Pi class glutathione S-transferase promoter.

We have previously demonstrated enhanced transcriptional activity of the human Pi class glutathione S-transferase (GSTP1) promoter in a multidrug-resistant derivative (VCREMS) of the human mammary carcinoma cell line, MCF7 (Moffat, G. J., McLaren, A. W., and Wolf, C. R. (1994) J. Biol. Chem. 269, 16397-16402). Furthermore, we have identified an essential sequence (C1; -70 to -59) within the GSTP1 promoter that bound a Jun-Fos heterodimer in VCREMS but not in MCF7 cells. These present studies have examined the negative regulatory element (-105 to -86), which acted to suppress GSTP1 transcription in MCF7 cells. Mutational analysis of this silencer element further defined the repressor binding site to be located between nucleotides -97 and -90. In vitro DNA binding assays suggested that the repressor exerted its action by causing displacement of the essential non-AP-1-like MCF7 C1 complex. However, the addition of MCF7 nuclear extract did not disrupt binding of the VCREMS Jun-Fos C1 complex to the GSTP1 promoter. Furthermore, upstream insertion of the GSTP1 silencer element failed to inhibit activity of a heterologous promoter in MCF7 cells. These results highlighted the cell and promoter specificity of the GSTP1 transcriptional repressor and implicated a functional requirement for contact between the repressor and C1 complex. In this regard, the introduction of half-helical turns between the silencer and the C1 element abrogated repressor activity, thus leading to the hypothesis that a direct interaction between the repressor and C1 complex was required to suppress GSTP1 transcription. Moreover, these findings suggest that cell-specific differences in the composition of the C1 nuclear complex may dictate repressor activity.

Sp1-mediated transcriptional activation of the human Pi class glutathione S-transferase promoter.

Previous studies in this laboratory have identified an essential AP-1 recognition sequence (C1 region; -69 to -63) in th human Pi class glutathione s-transferase (GSTP1) promoter and a negatively acting regulatory element (-105 to -86) that acts to suppress GSTP1 transcription in the human mammary carcinoma cell line, MCF7 (1). The data presented here further delineate the functional characteristics of the GSTP1 promoter by examining the significance of two potential binding sites for the transcription factor, Sp1 (-57 to -49 and -47 to -39). The introduction of mutations within these Sp1-like elements and the use of Sp1 antisera in electrophoretic mobility shift assays demonstrated that Sp1 was bound to this region of the GSTP1 promoter in three different cell lines, MCF7, VCREMS, and EJ. Moreover, these in vitro studies indicated that only one of the two putative Sp1 response elements was utilized. Transient transfection assays using GSTP1 promoter constructs that incorporated mutations of the Sp1 elements clearly demonstrated that binding of Sp1 to the GSTP1 promoter was absolutely required for optimal levels of GSTP1 transcription. In particular, disruption of the distal Sp1 recognition motif (-57 to -49) markedly reduced GSTP1 promoter activity in each cell line, thus indicating preferential binding of Sp1 to the distal site. However, insertion of the repressor binding site (-105 to -86) into these constructs suggested that Sp1 was not involved in mediating the suppressive effects of the GSTP1 transcriptional repressor in MCF7 cells, because inhibition of Sp1 binding did not alleviate repressor activity. Therefore, these studies provide strong evidence that Sp1 plays a central role in regulating basal levels of GSTP1 transcription.

Involvement of Jun and Fos proteins in regulating transcriptional activation of the human pi class glutathione S-transferase gene in multidrug-resistant MCF7 breast cancer cells.

Elevated levels of the human pi class glutathione S-transferase (GSTP1-1) have been implicated in the development of antineoplastic drug resistance. Using GSTP1 promoter deletion constructs we have shown that enhanced GSTP1 transcription (up to 18-fold) is the predominant mechanism responsible for increased GSTP1-1 levels in a multidrug resistant derivative (VCREMS) of the human mammary carcinoma cell line MCF7. Furthermore, disruption of a putative AP-1 response element within the GSTP1 promoter (nucleotides -69 to -63) abrogated GSTP1 transcription in both cell lines. In addition, band shift assays demonstrated binding of a VCREMS nuclear complex to the promoter region C1 (-73 to -54) which could be competed for by a DNA fragment containing a known AP-1 binding site from the human collagenase promoter. However, no such competition was observed for the major MCF7 C1 complex. The role of a Fos-Jun-like complex in regulating GSTP1 transcription in VCREMS cells was further emphasized by the introduction of point mutations within the C1 region which were known to inhibit AP-1 binding and the interaction of antisera raised against human c-Jun and c-Fos with the major C1 complex in VCREMS cells. These studies therefore highlight cell-specific differences in the binding pattern of Jun and Fos proteins to the GSTP1 promoter which are likely to play an important role in regulating transcriptional activation of the GSTP1 gene in drug-resistant breast cancer cells.

Relative expression of cytochrome P450 isoenzymes in human liver and association with the metabolism of drugs and xenobiotics.

Cytochrome P450s play a central role in the metabolism and disposition of an extremely wide range of drugs and chemical carcinogens. Individual differences in the expression of these enzymes may be an important determinant in susceptibility to adverse drug reactions, chemical toxins and mutagens. In this paper, we have measured the relative levels of expression of cytochrome P450 isoenzymes from eight gene families or subfamilies in a panel of twelve human liver samples in order to determine the individuality in their expression and whether any forms are co-regulated. Isoenzymes were identified in most cases on Western blots based on the mobility of authentic recombinant human cytochrome P450 standards. The levels of the following P450 proteins correlated with each other: CYP2A6, CYP2B6 and a protein from the CYP2C gene subfamily, CYP2E1 and a member of the CYP2A gene subfamily, CYP2C8, CYP3A3/A4 and total cytochrome P450 content. Also, the levels of two proteins in the CYP4A gene subfamily were highly correlated. These correlations are consistent with the relative regulation of members of these gene families in rats or mice. In addition, the level of expression of specific isoenzymes has also been compared with the rate of metabolism of a panel of drugs, carcinogens and model P450 substrates. These latter studies demonstrate and confirm that the correlations obtained in this manner represent a powerful approach towards the assignment of the metabolism of substrates by specific human P450 isoenzymes.

Identification of the human liver cytochrome P-450 responsible for coumarin 7-hydroxylase activity.

1. We have constructed a full-length human liver cytochrome P450IIA cDNA from a partial-length clone by oligonucleotide-directed mutagenesis, and subcloned it into the monkey kidney (COS-7) cell expression vector, pSVL. 2. The cDNA encodes a 49 kDa protein with coumarin 7-hydroxylase (COH) activity which cross-reacts with antisera to the mouse cytochrome P-450 isoenzyme responsible for COH activity and comigrates with a human liver microsomal protein. 3. Western blot analysis of a panel of human livers indicates that the level of the 49 kDa protein, detected using antisera to either the mouse COH P-450 or rat P450IIA1 protein, correlates very highly with COH activity. 4. Antisera to the rat P450IIA1 protein can inhibit COH activity in human liver microsomes. Taken together, these data indicate that a member of the P450IIA subfamily is responsible for most, if not all, of the COH activity in human liver.

Alternative splicing in the human cytochrome P450IIB6 gene generates a high level of aberrant messages.

Polymorphisms within the human cytochrome P450 system can have severe clinical consequences and have been associated with adverse drug side effects and susceptibility to environmentally linked diseases such as cancer. Aberrant splicing of cytochrome P450 mRNA has been proposed as a potential mechanism for these polymorphisms. We have isolated aberrantly, as well as normally, spliced mRNAs (cDNAs) from the human P450IIB6 gene which either contain part of intron 5 and lack exon 8 or which contain a 58-bp fragment (exon 8A) instead of exon 8. Sequence analysis of the P450IIB6 gene demonstrates the presence of cryptic splice sites in intron 8 which will account for the generation of exon 8A. The mRNAs were therefore generated by alternative splicing. These data gain significance as the mRNAs will not encode a functional P450 enzyme and appear to represent a high proportion of the P450IIB6 mRNA population. Analysis of mRNA from fifteen individual human livers and cDNA libraries constructed from a variety of human tissues using the polymerase chain reaction shows that the aberrant splicing occurs in all cells and all individuals tested. This suggests a high level of infidelity in the processing of P450IIB6 mRNAs and demonstrates that the presence of abnormal transcripts does not imply the presence of a functionally inactive gene.

Close linkage of the human cytochrome P450IIA and P450IIB gene subfamilies: implications for the assignment of substrate specificity.

We have isolated from human liver libraries two cytochrome P450 cDNA clones (lambda MP14 and lambda MP3) which are highly similar (83% over the coding region) to mouse testosterone 15 alpha hydroxylase and are therefore part of the cytochrome P450IIA gene subfamily. The P450IIA (CYP2A) gene subfamily was found to be closely linked to the P450IIB (CYP2B) subfamily and their chromosomal location could not be distinguished using somatic cell hybrids containing fragments of chromosome 19 between 19q12 and 19q13.2. Pulsed field gel electrophoresis indicates that both gene subfamilies are contained within 350-kb genomic DNA fragments, but were separated using various restriction enzymes. Northern blot analysis identified three P450IIA mRNAs each showing a wide inter-individual variation in their levels in the liver. High levels of P450IIA transcript were associated with high levels of P450IIB transcript suggesting that common factors may influence the expression of genes within these subfamilies. Genetic analysis has suggested previously that a member of the P450IIB subfamily is responsible for coumarin hydroxylase activity in the mouse. We discuss the possibility, based on our findings of tight linkage of the human P450IIA and IIB subfamilies, that a member of the IIA subfamily is a better candidate for this enzyme activity.

A novel human cytochrome P450 gene (P450IIB): chromosomal localization and evidence for alternative splicing.

We have isolated from a single human liver cDNA library two clones which are highly homologous (78% over the coding region) to the major phenobarbital-inducible P450 from rat (P450IIB1). This is the first direct demonstration of the presence of the P450IIB gene subfamily in humans. This subfamily is much less extensive than the rodent homologues, but does appear to contain at least two genes. Of the cDNA clones isolated one is apparently normally spliced, whereas the other lacks exon 8 and retains all or part of intron 5. Both clones contain transcribed Alu sequences. The human P450IIB gene has been located to chromosome 19q12----19q13.2 using a probe derived from intron 5, and is close to the CYP 2A locus encoding cytochrome P450IIA2. Restriction fragment length polymorphisms have been found with the enzymes BamHI and MspI which will enable linkage to be determined between these two loci.