Abstract
Liver cytochrome P450s (CYPs) of the endoplasmic reticulum (ER) are involved in the metabolism of exogenous and endogenous chemicals. The ER is not uniform, but possesses ordered lipid microdomains containing higher levels of saturated fatty acids, sphingomyelin, and cholesterol and disordered regions containing higher levels of polyunsaturated fatty acid chains. The various forms of drug-metabolizing P450s partition to either the ordered or disordered lipid microdomains with different degrees of specificity. P450s readily form complexes with ER-resident proteins, including other forms of P450. This study aims to ascertain whether lipid microdomain localization influences protein-P450 interactions in rat liver microsomes. Thus, liver microsomes were co-immunoprecipitated with CYP1A2-specific and CYP3A-specific antibodies, and the co-immunoprecipitating proteins were identified by liquid chromatography mass spectrometry proteomic analysis. These two P450s preferentially partition to ordered and disordered microdomains, respectively. More than 100 proteins were co-immunoprecipitated with each P450. Segregation of proteins into different microdomains did not preclude their interaction. However, CYP3A interacted broadly with proteins from ordered microdomains, whereas CYP1A2 reacted with a limited subset of these proteins. This is consistent with the concept of lipid raft heterogeneity and may indicate that CYP1A2 is targeted to a specific type of lipid raft. Although many of the interacting proteins for both P450s were other-drug metabolizing enzymes, other interactions were also evident. The consistent CYP3A binding partners were predominantly involved in phase I/II drug metabolism; however, CYP1A2 interacted not only with xenobiotic metabolizing enzymes, but also with enzymes involved in diverse cellular responses such as ER stress and protein folding.
SIGNIFICANCE STATEMENT This work describes the protein interactomes in rat liver microsomes of two important cytochromes P450s (CYP1A2 and CYP3A) in drug metabolism and describes the relationship of the interacting proteins to lipid microdomain distribution
Introduction
Cytochrome P450 (P450 or CYP) is a term referring to a diverse superfamily of heme-containing enzymes that catalyze the metabolism of a variety of endogenous and xenobiotic compounds (Nelson, 2003). In mammalian species, the metabolism of most drugs and other exogenous compounds is catalyzed primarily by three families of P450 that are further divided into several subfamilies. Members from each of the three families are highly expressed in the liver, and approximately fifteen major forms from eight different subfamilies of P450 can be identified in the endoplasmic reticulum (ER) of human liver (Zanger and Schwab, 2013).
Proteins are packed at a very high density in the ER membrane (Quinn et al., 1984; Watanabe et al., 1993). Many lines of evidence, indicate that the P450s tend to function as aggregates and that the rate of metabolism by a given P450 is influenced in a form-specific manner by the physical interaction with other P450s (Backes et al., 1998; Hazai and Kupfer, 2005; Subramanian et al., 2009; Reed et al., 2010; Subramanian et al., 2010; Kenaan et al., 2013; Davydov et al., 2015). Because of the multiple P450 enzymes, it is a daunting task to assess all the possible interactions of P450 for their related effects on metabolism. It would be extremely useful to identify whether a certain P450 is more likely to interact with a specific form before investing the resources to assess the potential effects of a P450–P450 interaction on metabolism.
Biologic membranes are made up of a complex array of different lipids heterogeneously arranged in an inverted bilayer (van Meer et al., 2008). In the simplest of representations, these membranes have disordered regions containing high concentrations of phosphatidylcholine and other lipids interspersed with relatively small, ordered microdomains that are transient in nature and contain high proportions of cholesterol, phosphatidylethanolamine, and sphingomyelin (London and Brown, 2000) and an abundance of fatty acyl groups with saturated or mono-unsaturated carbon chains (Lindner and Naim, 2009). In contrast, the disordered regions of the membranes are characterized by an abundance of polyunsaturated fatty acyl moieties (Lindner and Naim, 2009).
Our laboratory has established that the different P450s variably distribute between these ordered and disordered lipid microdomains in ER (Brignac-Huber et al., 2011; Park, 2014). Initially, we discovered that rabbit microsomal CYP1A2 resided primarily in ordered lipid microdomains that were resistant to solubilization by the nonionic detergent, Brij 98. These detergent resistant microdomains (DRMs) to which CYP1A2 preferentially located were dependent on cholesterol (Brignac-Huber et al., 2011). Subsequent work with purified enzymes showed that ordered lipid microdomains containing CYP1A2 also formed in reconstituted systems which were prepared so that the enzyme was integrally incorporated into lipid vesicles with the phospholipid and cholesterol composition similar to that in liver ER (Brignac-Huber et al., 2013). Both studies suggested that CYP1A2 localization also influenced the microdomain localization of the redox partner, NADPH-cytochrome P450 reductase. Western blotting showed that other rabbit liver P450s distributed differently between the lipid membrane microdomains. CYP2E1 was localized primarily to disordered regions, whereas CYP2B4 distributed evenly between DRM and disordered regions (Park et al., 2014).
In a recent study, we used Brij 98 solubilization and quantitative proteomics to categorize the microdomain localization of the proteins in liver microsomes from control and phenobarbital-treated rats. This study identified the microdomain localization of 34 rat liver P450s. In addition to CYP1A2, four members of the CYP2D subfamily localized to ordered microdomains, whereas most of the P450s, such as CYP2E1 (and 24 others), resided in disordered lipid microdomains. However, five members, including CYP2B1, also were categorized as uniformly distributed between the lipid membrane microdomains. Thus, P450s are well represented in each of the three defined modes of membrane distribution (e.g., ordered and disordered lipid microdomains as well as equally distributed between the two microdomains). This previous study also showed rats are similar with respect to the membrane localization of the similar forms of P450 in rabbit liver microsomes (Park et al., 2014).
Another major P450 enzyme of interest in drug metabolism is CYP3A. In humans, CYP3A4 is the most abundant P450 in the liver and metabolizes over 50% of all marketed drugs (Zanger and Schwab, 2013). CYP3A members in rat liver microsomes also were primarily localized to disordered lipid regions in the membrane (Reed et al., 2022).
In the current study, we use analytical proteomics to identify the P450-binding partners of representative P450s from ordered and disordered membrane regions by doing high stringency co-immunoprecipitation of CYP1A2 and CYP3A, respectively, followed by liquid chromatography tandem mass spectrometry (LC/MS/MS) to identify the tryptic peptides of the binding partners. Our objective was to ascertain whether the microdomain localization of microsomal proteins influence the heteromeric interactions of protein P450 complexes as would be implicated by a preponderance of co-immunoprecipitated (Co-IP) proteins from the domain to which the P450 partitions.
Materials and Methods
Materials
Protein G-Plus agarose beads, NP40, pre-cast 10% polyacrylamide gels, HiPPR Detergent Removal Resin Column Kit, and radioimmunoassay immunoprecipitation (RIPA) buffer were purchased from Thermo (Waltham, MA). Bovine serum albumin was purchased from Gemini Bioproducts (West Sacramento, CA). Sodium deoxycholate, SDS, and complete protease inhibitor tablets without EDTA were purchased from Sigma (St. Louis, MO).
Rat Liver Microsomes
Liver microsomes from control and phenobarbital-treated male rats weighing between 200 and 300 g were prepared by differential centrifugation of homogenized rat liver as described in our previous study (Reed et al., 2022). In fact, most of the analyses reported in this study involved the same preparations of rat microsomes that were used in the previous quantitative proteomic study identifying the proteins from different lipid microdomains.
Co-Immunoprecipitation of Rat Liver Microsomes with Anti-CYP3A and Anti-CYP1A2
Immunoprecipitation of proteins from rat liver microsomes using control and P450-specific antibodies was carried out as follows:
Preparation of Protein G-Plus agarose – Protein G-Plus agarose beads (50% bead slurry) were prepared for co-immunoprecipitation experiments by transferring a volume equal to the original volume of the microsomal suspension before addition of 2xRIPA buffer (Pierce-Thermo Scientific). A mark was drawn on the centrifuge tube containing the slurry before it was centrifuged at 2,500 × g for 10 seconds. The pelleted beads were then washed three times in 10 ml of phosphate buffered saline (pH 7.6) by centrifugation and resuspension to the original volume indicated by the tube marking before being centrifuged and resuspended a final time in 10 ml of phosphate buffered saline containing 3% bovine serum albumin. The bead resuspension was incubated overnight by gentle rotation at 4°C. The next day, the beads were washed two times with 10 ml of Tris-HCl buffer (pH 7.6) containing 150 mM NaCl before resuspending in RIPA buffer (Pierce-Thermo Scientific) to the original hand-drawn mark on the tube. A P450-specific antibody (ab4227 for CYP1A2 (Abcam – Cambridge, MA) and Millipore AB1253 for CYP3A (Burlington, MA)) or control goat- and rabbit-antibody Ig antibody, respectively, was then added to the bead suspension at a 1:25 dilution. For every microsomal preparation that was tested, both P450-specific and control IgG antibodies were run simultaneously. The tubes were then placed on a rotator at 4°C until use in the Co-IP experiment (usually about 15 minutes).
Immunoprecipitation of P450s – For all experiments, rat liver microsomes (50 μl) were suspended at 4 mg/ml microsomal protein in a solution with final concentrations of 50 mM Hepes-NaOH (pH 7.5), 15 mM MgCl2, and 0.1 mM EDTA. An equal volume of ice-cold, 2xRIPA buffer containing 50 mM Tris-HCl (pH 7.6), 300 mM NaCl, 2% Nonidet P-40 (Thermo Scientific), 0.2% SDS, and 500 mM glycine was then added. For reasons explained in the Results section, CYP1A2 was co-immunoprecipitated from control liver microsomes, and CYP3A proteins were co-immunoprecipitated from phenobarbital (PB)-induced microsomes. The 2 x RIPA buffer was supplemented with a complete protease inhibitor tablet (Sigma) and 200 µM PMSF. The microsomal suspensions were bath-sonicated for 5 seconds, which was found to dramatically increase the number of proteins co-immunoprecipitated with the P450s. The bead/antibody suspensions prepared as described above were then added at an equal volume to the original microsomal suspension (0.05 ml). The bead/microsomal suspensions were then set on a rotator at room temperature for 2 hours. The samples were then centrifuged at 2500 × g for 2 minutes, and the supernatant was pipetted and placed in a clean tube. The beads were then washed 3 times by centrifugation/resuspension in 1 ml of commercially purchased RIPA buffer (Pierce-Thermo, Waltham, MA) (added slowly through a large-orifice pipette tip). The beads were then resuspended/washed two times in 1 ml of 25 mM Tris-HCl buffer (pH 7.6) with 150 mM NaCl. After the final centrifugation, the protein G bead pellets were resuspended in 0.1 ml of a loading buffer consisting of 50 mM Tris-HCl (pH 7.1), 6% SDS (w/v), 24% sucrose, 0.67 M of glycine, and 0.67 M NaCl and were heated for 5 to 10 minutes at 100°C. The loading buffer was heated to 40°C before use to dissolve the components. Samples were then centrifuged, and the supernatant of each was pipetted and transferred to a clean tube. The beads were then rinsed with an additional 0.1 ml of 50 mM Tris-HCl (pH 7.5) and 150 mM NaCl, and the final supernatant was combined with the first bead extract (described above). Beta-mercaptoethanol (0.01 ml) was then added to the combined bead extracts and samples were subsequently boiled for another 10 minutes. Samples were then stored at -80°C until being processed for proteomic analysis or western blotting.
Western Blotting
Bead extracts (0.01 ml) from immunoprecipitation experiments with P450-specific and control antibodies and the supernatants resulting from the first centrifugation of beads (after the 2 hour incubation of beads and microsomes, see above) were loaded on 10% polyacrylamide pre-cast gels (Thermo-Invitrogen). The gels were run for 1.25 hours at 150 V. The gels were transferred to nitrocellulose using the Power Blotter system (Thermo), blocked for 1 hour with PBS containing 0.05% Tween 20 (Bio-Rad – Hercules, CA) and 2% bovine serum albumin. Western blots were developed using the primary antibodies – Abcam 56073 at 1:1000 and PM40 (Oxford Biochemical – Rochester Hills, MI) at 1:1000 for CYP1A2 and CYP3A, respectively. The secondary antibodies were anti-mouse conjugated horseradish peroxidase (Sigma) at 1:4000, and blots were developed with SuperSignal West Pico PLUS chemiluminescent substrate (Thermo).
Protein ID Proteomic Methods
After IP, the eluates were treated with a HiPPR Detergent Removal Resin Column Kit (Thermo), and this elution was subjected to methanol-chloroform extraction. Extracted proteins were mixed with 30 µl of 20 µg/ml of trypsin dissolved in 50 mM ammonium bicarbonate and incubated overnight at 37°C. The next day, trypsinized proteins were speed vacuumed to completion and then resuspended in 20 µl of 2% acetonitrile (ACN) and 0.1% formic acid for liquid chromatography mass spectrometry (LC-MS).
The sample was run in duplicate on a Thermo-Fisher U3000 nano flow system coupled to a Thermo Fusion mass spectrometer. Each sample was subjected to chromatographic separation, employing a gradient from 2% to 25% acetonitrile in 0.1% formic acid (ACN/FA) over the course of 16 minutes, from 25% to 35% ACN/FA for an additional 15 minutes, from 35% to 50% ACN/FA for an additional 4 minutes, a step to 90% ACN/FA for 4 minutes, and a re-equilibration into 2% ACN/FA. Chromatography was carried out in a “trap-and-load” format using an EASY-Spray source (Thermo); trap column C18 PepMap 100, 5 µm, 100 A and the separation column was an EASY-Spray PepMap RSLC C18 2 µm, 100 A, 75 µm × 25 cm (Thermo Fisher Dionex, Sunnyvale, CA). The entire run was at a flow rate of 0.3 µl/min. Electrospray was achieved at 1.8 kV.
MS1 scans were performed in the Orbitrap utilizing a resolution of 240,000, and data-dependent MS2 scans were performed in the Orbitrap using high energy collision dissociation of 30% using a resolution of 30,000. Data analysis was performed using the Proteome Discoverer 2.4 using SEQUEST HT scoring. The Protein FASTA database was Rattus norvegicus, SwissProt tax ID=10116, version 2017-10-25 and contained 9616 sequences. Dynamic modification of oxidation of methionine (=15.9949) was considered. Parent ion tolerance was 10 ppm, fragment mass tolerance was 0.02 Da, and the maximum number of missed cleavages was set to 2. Only high scoring peptides were considered, utilizing a false discovery rate of 1%. Proteins that were specifically associated with either CYP1A2 or CYP3A were identified by the parallel immunoprecipitation reactions using control IgG antibodies. Only the proteins that were specifically associated with the P450s (i.e., were not identified in the LC/MS/MS analysis of the matched immunoprecipitates using control IgG antibodies) are reported in this study, with one exception. A small proportion of CYP3A was immunoprecipitated (IP) using control IgG antibody from one microsomal sample that had exceptionally high levels of CYP3A after PB treatment (Fig. 2B, lane 5).
The final list of CYP1A2 Co-IP proteins was comprised of proteins identified in six IP/LC-MS experiments using the CYP1A2-specific antibody as described above. Three preparations of rat liver microsomes were analyzed in duplicate. The IP/LC-MS analyses with the CYP3A-specific antibody used six different preparations of PB microsomes. Three of the PB microsomal samples were prepared from the same cohort of animals used in the preparation of control microsomes for the analyses with the CYP1A2-specific antibody. The other three IP/LC-MS analyses came from an older set of PB-treated rat liver microsomes prepared by the same method previously described (Reed et al., 2022). The proteins identified in the immunoprecipitates using this older set of liver microsomes were not included in the CYP3A Co-IP unless the proteins were also identified in the immunoprecipitates from the experiments with the common cohort of animals.
Results
The goal of this study was to determine if microdomain localization of P450 proteins affected their ability to form physical complexes with other ER-resident proteins. In other words, do proteins that reside in ordered microdomains more effectively interact with other ordered-domain proteins, or do they also interact readily with proteins from the disordered regions. This was examined by co-immunoprecipitating and comparing those proteins that formed complexes with CYP1A2 (a P450 that associated with the ordered microdomains) to those co-immunoprecipitated with CYP3A (a P450 associated with disordered lipid regions of the membrane). Co-IP, under high stringency to increase the specificity of the interactions, was performed as described in Materials and Methods. Figs. 1 and 2 show the western blots confirming that there was specific immunoprecipitation of CYP1A2 and CYP3A, respectively. Although microsomes from both control and phenobarbital-treated rats were available, immunoprecipitation of CYP1A2 was only performed using the control microsomes because CYP1A1 is not expressed constitutively (Guengerich et al., 1982). However, we have detected induction of CYP1A1 in rats after PB-mediated induction using the treatment protocol of this study (unpublished). The antibody used for CYP1A2 immunoprecipitation reacts with both CYP1A1 and CYP1A2. As a result, we did not perform Co-IP experiments for CYP1A in phenobarbital microsomes because our objective was to compare the binding partners of the DRM-resident, CYP1A2, and those of a P450 from disordered lipid domains, CYP3A. Fig. 1A shows that CYP1A2 was immunoprecipitated from each microsomal suspension from three rats. Fig. 1B shows the results using a control goat IgG antibody with the same three suspensions of microsomes and indicates that the CYP1A2 was not immunoprecipitated non-specifically.
Western blots showing immunoprecipitation of CYP1A2 from control rat liver microsomes. Immunoprecipitation of CYP1A2 from rat liver microsomes was performed as described in Methods. Panel A shows the results using CYP1A2-specific antibody with microsomes from three control rats. Lanes 2–5 were loaded with 30 µl of the final protein G- plus agarose bead extract from an immunoprecipitation reaction. Lanes are described as follows: lane 1, 2 pmol of rat CYP1A2 expressed in Supersomes (Gentest-Corning, Woburn, MA); lane 2, CYP1A2-specific antibody (no microsomes); lanes 3, 4, and 5, anti-CYP1A2 immune precipitates from control liver microsomal preparations (three different animals); lane 6, blank; lanes 7–9, supernatants from the respective immunoprecipitation reactions showing protein that did not bind to anti-CYP1A2. Panel B shows the gel with the same arrangement of lanes as in panel A, but the immunoprecipitations of the same microsomal samples were conducted with control goat IgG antibody (Panel B).
Western blots showing immunoprecipitation of CYP3A from control and PB-treated rat liver microsomes. Immunoprecipitation of CYP3A from rat liver microsomes was performed as described in Methods. Panel A shows the results using CYP3A-specific antibody with microsomes from two control rats and two PB-treated rats. Lanes are described as follows: lane 1, blank; lane 2 CYP3A-specific antibody (no microsomes); lanes 3 and 4, anti-CYP3A immunoprecipitates from saline-treated rat liver microsomes; lanes 5 and 6, anti-CYP3A immunoprecipitates from PB-treated rat liver microsomes; lanes 7–10, supernatants from the respective immunoprecipitation reactions that did not bind to anti-CYP3A. Panel B shows the same arrangement of lanes as in panel C except the immunoprecipitations were performed with control rabbit IgG antibody.
CYP3A immunoprecipitation was performed with microsomes from both control and PB rats as shown in Fig. 2. Although the immunoprecipitation was successful (Fig. 2A) and specific (Fig. 2B), a lower proportion of the CYP3A was pulled down from control rats than from PB rat microsomes. Furthermore, only fifteen proteins were co-immunoprecipitated from control rat microsomes with the CYP3A-specific antibody, whereas, over 120 proteins co-immunoprecipitated with CYP3A from PB rat microsomes. The different result with control and PB rats is most likely due to the levels of CYP3A expression. The highest proportion of CYP3A was immunoprecipitated from the microsomal sample with the highest CYP3A expression (lane 6 in Fig. 2A). Thus, the immunoprecipitation process was more efficient at higher levels of target enzyme expression. This is consistent with antibody-ligand binding kinetics. Efforts to circumvent this problem with CYP3A immunoprecipitation from control microsomes by increasing the antibody and/or microsome concentrations were not successful in increasing the yield of immunoprecipitated protein and, in fact, lowered the proportion of immunoprecipitated enzyme. Because CYP3A mainly partitioned to disordered lipid regions in both control and phenobarbital-treated rat microsomes, the influence of lipid microdomain localization on protein interactions with CYP3A can be assessed in both types of microsomes. Because only a limited number of proteins were co-immunoprecipitated from the control microsomes, the LC/MS/MS results only report on the CYP3A binding partners that were co-immunoprecipitated from phenobarbital microsomes.
LC/MS/MS of Co-IP Proteins – Protein Microdomain Distribution
The proteins that co-immunoprecipitated with CYP1A2 and CYP3A were cross-referenced against the quantitative proteomic data of Brij 98-solubilized microsomes (Reed et al., 2022) to determine the relative partitioning of the P450 binding partners to different lipid membrane microdomains (also shown in Supplemental Tables 1 and 2, respectively). Figs. 3 and 4 also provide an overview of the microdomain distribution of microsomal proteins from control and PB-induced rat liver (Figs. 3A and 4A, respectively) and those proteins that Co-IP with CYP1A2 and CYP3A (Panels C & D in Figs. 3 and 4, respectively). Other studies have shown that liver microsomes actually contain proteins from a variety of organelles in addition to the ER (Galeva and Altermann, 2002; Peng et al., 2012). Thus, the microdomain distributions of proteins with “endoplasmic reticulum” as a molecular component in Uniprot are also shown in Figs. 3 and 4 (Panel B). The collection of ER proteins from control and PB rat liver microsomes were identified from our previous study using quantitative LC-MS to analyze the Brij 98-mediated solubilization of rat liver microsomes (Reed et al., 2022). The list of selected endoplasmic reticulum proteins in panel B were further narrowed down by removing the proteins with “Golgi” as a molecular component. Although P450s have been identified in the Golgi (Neve et al., 1996), the levels of expression are believed to be very low (Yamamoto et al., 1985).
Lipid microdomain distribution of CYP1A2 Co-IP proteins versus that of total liver microsomes. The microdomain localization of proteins from control rat liver microsomes were compared with the localization of proteins that co-immunoprecipitated with CYP1A2. The assignments of proteins into ordered lipid microdomains (DRM), disordered lipid membrane, and uniformly distributed are determined by reference to our previous study (Reed et al., 2022). The numbers in each segment indicate the number of proteins represented by that segment, with parentheses showing their percentages versus the total number of proteins. Panel A: Microdomain distribution of total microsomes (Reed et al., 2022). Panel B: Distribution of ER proteins in control microsomes. Panel C: Microdomain distribution of CYP1A2 Co-IP proteins, including those of unknown microdomain localization (i.e., were not detected in our previous study). Panel D: Microdomain distribution of CYP1A2 Co-IP proteins when the proteins with unknown microdomain localization were excluded.
Lipid microdomain distribution of CYP3A Co-IP proteins versus that of total liver microsomes. The microdomain localization of proteins from PB-treated rat liver microsomes were compared with the localization of proteins that co-immunoprecipitated with CYP3A. Panel A: Microdomain distribution of PB-treated microsomes (Reed et al., 2022). Panel B: Distribution of ER proteins in PB-treated microsomes. Panel C: Microdomain distribution of CYP3A Co-IP proteins, including those of unknown microdomain localization (i.e., these were not detected in our previous study (Reed et al., 2022)). Panel D: Microdomain distribution of CYP3A Co-IP proteins when the proteins with unknown microdomain localization were excluded.
As stated in our previous study (Reed et al., 2022), PB treatment had very little effect on the protein distribution into lipid microdomains. The lipid microdomain distributions of ER proteins were also very similar in control and PB rats but very different from the corresponding distributions of proteins in total microsomes (compare panels A and B in Figs. 3 and 4). The total ER proteins have higher proportions of proteins in ordered lipid microdomains and smaller proportions of proteins that uniformly distribute between the ordered and disordered microdomains compared with the total proteins in the liver microsomes.
Both CYP1A2 and CYP3A immunoprecipitated with proteins from both ordered and disordered lipid microdomains and with proteins that uniformly distribute between the two domains. Interestingly, a higher percentage of proteins from ordered lipid domains were pulled down with the disordered lipid domain-associated CYP3A (34%) than with the DRM-associated, CYP1A2 (22%). Unfortunately, 12%–16% of the Co-IP proteins identified using both of the P450-specific antibodies were not detected in the proteomic data from Brij 98-solubilized microsomes from the previous study (Reed et al., 2022), so the domain localization of these proteins could not be determined.
Fig. 5 depicts the number of common proteins in the two groups of Co-IPs in addition to the levels of overlap for each group of proteins according to lipid microdomain localization. Supplemental Table 3 shows the proteins that were identified in both sets of Co-IPs along with their membrane localization. The proportions of proteins that are common to the total, disordered, and mixed groups of Co-IP are very similar for both sets of Co-IP. There were also similar proportions of common proteins that had unknown localization within the membrane for the two Co-IPs (data not shown). The proportions of proteins that were common to those from ordered lipid microdomains were very different for the two Co-IPs. There were only 11 proteins that were predominantly localized to ordered lipid regions of the membrane in the CYP1A2 Co-IP that were not observed in the CYP3A Co-IP. In fact, 56% of the CYP1A2 Co-IP were also observed in the CYP3A Co-IP. Conversely, the CYP3A Co-IP proteins from ordered lipid microdomains included 27 proteins that were not identified in the CYP1A2 Co-IP, and only 34% of the CYP3A Co-IP proteins from the ordered lipid environment were also observed in the CYP1A2 Co-IP. The implications of these findings are addressed in the Discussion.
Proportions of proteins identified in both groups of P450 Co-IPs. The circles represent the populations of proteins contained in the indicated groups (total proteins in the CYP1A2- and CYP3A-Co-IPs, complete; proteins that partition to disordered lipid domains, disordered; proteins that evenly distribute between ordered and disordered lipid domains, mixed; and proteins that partition to ordered microdomains, ordered). The areas of the circles are proportional to the number of proteins in each group and are calculated as percentages of the total proteins in the CYP3A Co-IP. The overlapped regions in the Venn diagrams for each of the categories reflect the numbers of proteins that are common to both Co-IPs. The numbers drawn to the side of the overlapped regions represent the numbers and percentages (in parentheses) of proteins that are not common to both groups of proteins. The proteins common to both Co-IP are listed in Supplemental Table 3.
Co-IP of Microsomal Proteins
The Co-IP/LC-MS analyses identified numerous co-immunoprecipitated proteins with each P450. In addition, each set of Co-IP was approximately equal in the number of proteins (116 for CYP1A2 Co-IP (Supplemental Table 1) and 134 for the CYP3A Co-IP (Supplemental Table 2)). Thus, our study has identified many more binding partners than in previous studies utilizing co-immunoprecipitation of P450s (Li et al., 2011; Fujiwara and Itoh, 2014; Davydov et al., 2022). Like the other studies that have used cross-linking and/or immunoprecipitation to identify protein interactions with P450s, both P450s were immunoprecipitated with a wide variety of microsomal proteins that are related to many cellular functions. The CYP1A2 Co-IP contained proteins that participate in a variety of processes that include but are not limited to metabolism, biosynthesis, cellular responses, and protein folding. Proteins related to xenobiotic metabolism include a variety of P450s, flavin monooxygenases, and epoxide hydrolase. There are also a variety of phase II enzymes, such as glutathione S-transferase and various forms of UDP-glucuronosyl transferase (UGT), in addition to enzymes related to endogenous metabolism of fatty acids and steroids. Biosynthetic enzymes include proteins related to N-glycosylation and CoA ligation. Enzymes involved in various cellular responses were also co-immunoprecipitated with CYP1A2, including some of those related to the responses to endoplasmic reticulum stress, exposure to alcohol and metal ions, and corticosteroid stimulation. Finally, CYP1A2 was associated with 11 proteins involved in protein folding, including a number of heat shock proteins and protein isomerases. There was a high degree of overlap in the proteins contained in the two groups of Co-IP. Like CYP1A2 Co-IP, the CYP3A Co-IP also included some of the same enzymes related to metabolism, biosynthesis, cellular responses, and protein folding. The CYP3A Co-IP was distinguished by the inclusion of seven enzymes directly related to steroid metabolism.
Of the co-immunoprecipitated proteins, the most obvious trend is that the two enzymes interact with other forms of P450. CYP3A was associated with a larger number of P450s than CYP1A2 (16 versus 14 in addition to the four forms of CYP3A that were immunoprecipitated – Supplemental Tables 1 and 2). The interaction of CYP1A2 with CYP3A was detected in the experiments with both P450-specific antibodies (Supplemental Tables 1 and 2). These results are consistent with previously reported CYP1A2–CYP3A interaction (Alston et al., 1991; Yamazaki et al., 1997).
Enzymes that were Repeatedly Co-Immunoprecipitated with the P450s
The proteomic analysis used in this study was qualitative and did not provide information on the tendency of proteins to interact with the two P450s. To gain further insight regarding the specificity of protein interactions with the P450s, the co-immunoprecipitations with each antibody were repeated six times in total with at least three different preparations of rat liver microsomes. Tables 1 and 2 show the Co-IP proteins that were detected in at least three of the six analyses. Approximately, 30 proteins were identified as “consistent” binding partners with each P450 using this criterion. It is interesting to note that most of the consistent binding-partners were drug-metabolizing enzymes, especially other P450s.
Proteins that co-immunoprecipitated with CYP1A2 at least three times out of six immunoprecipitation reactions with three sets of control rat liver microsomes
Microsomes prepared from three different rat livers were subjected to immunoprecipitation (two times for each microsomal preparation) with CYP1A2-specific and control antibodies. Co-immunoprecipitated proteins were identified by LC/MS/MS. The table shows proteins that were detected as co-immunoprecipitates in three or more of the experiments. The table also shows the predominant lipid microdomain of localization as identified in the preceding study (Reed et al., 2022). These proteins and proteins that co-immunoprecipitated less than three times experiments are color-coded according to their lipid microdomain localization in Supplemental Table 1. Proteins that were detected in any given experiment using control antibody were not considered specific co-immunoprecipitates and are not shown in the tables.
Proteins that co-immunoprecipitated with CYP3A at least three times out of six immunoprecipitation reactions with PB-treated rat liver microsomes
Microsomes prepared from six different PB-treated rat livers were subjected to immunoprecipitation with CYP3A-specific and control antibodies. At least one form of CYP3A was pulled down in each IP experiment. Co-immunoprecipitated proteins were identified by LC/MS/MS. The table legend is as described in that for Tables 1. These proteins and proteins that co-immunoprecipitated less than three times out of six experiments are color-coded according to their lipid microdomain localization in Supplemental Table 2.
Microsomes prepared from three different rat livers were subjected to immunoprecipitation (two times for each microsomal preparation) with CYP1A2-specific and control antibodies. Co-immunoprecipitated proteins were identified by LC/MS/MS. The table shows proteins that were detected as co-immunoprecipitates in three or more of the experiments. The table also shows the predominant lipid microdomain of localization as identified in the preceding study (Reed et al., 2022). These proteins and proteins that co-immunoprecipitated less than three times experiments are color-coded according to their lipid microdomain localization in Supplemental Table 1. Proteins that were detected in any given experiment using control antibody were not considered specific co-immunoprecipitates and are not shown in the tables.
Microsomes prepared from six different PB-treated rat livers were subjected to immunoprecipitation with CYP3A-specific and control antibodies. At least one form of CYP3A was pulled down in each IP experiment. Co-immunoprecipitated proteins were identified by LC/MS/MS. The table legend is as described in that for Table 1. These proteins and proteins that co-immunoprecipitated less than three times out of six experiments are color-coded according to their lipid microdomain localization in Supplemental Table 2.
The potential for P450 proteins to form complexes with other drug-metabolizing enzymes has been well established; however, the breadth of these interactions is not known. For example, CYP1A2 has been shown to interact with CYP2E1, CYP2B4, and CYP3A4 (Alston et al., 1991; Yamazaki et al., 1997; Cawley et al., 2001; Kelley et al., 2006; Reed et al., 2010). CYP3A has also been reported to form complexes with other P450 enzymes, including CYP2C9 and CYP2E1 (Subramanian et al., 2010; Davydov et al., 2015; Dangi et al., 2020; Davydov et al., 2022). Furthermore, both CYP1A and CYP3A have been shown to interact with phase II enzymes, such as UGTs and epoxide hydrolase. Many of these interactions were shown to influence their metabolic function (Taura et al., 2000; Taura Ki et al., 2002; Takeda et al., 2005; Miyauchi et al., 2020; Miyauchi et al., 2021).
Of the P450s repeatedly detected in immunoprecipitations with the CYP1A2-specific antibody, CYP2D26, CYP2E1, and CYP3A2 were identified in every IP experiment. The other P450s that were consistent binding partners of CYP1A2 were CYP2D1, CYP2A2, CYP2C6, and CYP2C23. When co-immunoprecipitation was repeated with CYP3A, nine different P450s co-immunoprecipitated three or more times with CYP3A forms. All of the consistent CYP3A binding partners were from the CYP2 family. In particular, CYP2C (six forms total) showed a tendency to co-immunoprecipitate with CYP3A. Thus, multiple interactions between CYP3A and other P450s were observed, with some having a greater tendency to interact than others. CYP3A seems to bind to other P450s more readily than CYP1A2 as evidenced by the number of repeatedly detected P450 interactions (Tables 1 and 2). Although CYP1A2 also co-immunoprecipitated with CYP3A2 in all of the IP experiments with the CYP1A2-specific antibody (as discussed above, see Supplemental Table 2), it was only detected in two of the six IP experiments with the CYP3A-specific antibody, so it is not shown in Table 2.
The Co-IP results from CYP1A2 and CYP3A immunoprecipitation are further distinguished with respect to the proteins repeatedly detected in the IP analyses. The consistent CYP3A binding partners are predominantly involved in phase I/II drug metabolism, whereas those repeatedly associated with CYP1A2-CoIP represent a range of cellular roles (although many of the proteins are also related to phase I drug metabolism). More specifically, CYP3A was repeatedly associated with two forms of flavin monooxygenase (FMO forms 3 & 5), with FMO3 also repeatedly associating with CYP1A2. The repeatedly detected proteins of the CYP3A Co-IP also is comprised of phase II enzymes, including four forms of UDP-glucuronosyltransferases (three other forms of UGTs were identified in less than three IP experiments) and microsomal glutathione S-transferase 1. The list of CYP3A Co-IP that were repeatedly co-immunoprecipitated also included another unusual phase II drug metabolism enzyme, methyltransferase-like protein 7B, that has been shown to transfer methyl groups to hydrogen sulfide and captopril using SAM as a cofactor (Maldonato et al., 2021). Retinol dehydrogenase 7 and bile acyl-CoA synthetase were identified in four or more of the IP repeats with the rat liver microsomes. Both of the total lists of Co-IPs also detected epoxide hydrolase although this enzyme was detected in less than three of the six IP experiments with each P450-specific antibody (Supplemental Table 1). No phase II enzymes were co-immunoprecipitated more than three times with CYP1A2. Finally, cytochrome b5 has been shown to be an important redox partner that stimulates the rates of metabolism of many substrates by P450s (Jansson et al., 1985; Schenkman and Jansson, 2003). This redox partner was detected in all six IP experiments with the CYP1A2-specific antibody but in only one of the IP experiments with the CYP3A antibody.
The consistent binding partners of both P450s also included a number of enzymes related to fatty acid oxidation/synthesis (many involving substrates or products that are acyl-CoA derivatives). For CYP1A2, these included very long-chain acyl-CoA synthetase, bile acyl-CoA synthetase, long-chain-fatty-acid–CoA ligase 1, and fatty aldehyde dehydrogenase. The CYP3A consistent binding-partners that participate in fatty acid metabolism also include bile acyl-CoA synthetase in addition to peroxisomal acyl-CoA oxidase 2 and very long-chain enoyl-CoA reductase.
The P450s were also repeatedly associated with enzymes regulated by calcium binding and/or involved in calcium homeostasis in the cytosol (Taniguchi et al., 2000). The CYP1A2 consistent binding partners include nucleobindin-2, which is believed to be involved in calcium homeostasis, whereas CYP3A was repeatedly immunoprecipitated with transmembrane protein 33, which also plays a role in calcium homeostasis in the ER tubular network (Arhatte et al., 2019). Furthermore, endoplasmin, which is included in the CYP1A2 consistent binding partners, functions as a chaperone and is regulated by calcium binding (Marzec et al., 2012).
Finally, complement C3 was associated with both P450s and was identified in all six IP experiments with CYP1A2. These interactions may be incidental as this protein of the immune system readily binds to proteins through a reactive thioester bond (Sahu and Lambris, 2001). Thus, it would be predicted to have a propensity to be co-immunoprecipitated nonspecifically. The two keratin proteins repeatedly co-immunoprecipitated with CYP1A2 are also notorious contaminants in Co-IP experiments (Li et al., 2016).
The consistent binding partners of CYP1A2 are distinguished from those of CYP3A by a protein involved in N-linked protein glycosylation (dolichyl-diphosphooligosaccharide–protein glycosyltransferase subunit) in addition to protein disulfide isomerase A6 involved in protein folding and two protein chaperones (heat shock cognate 71 kDa protein and endoplasmin (discussed above)). Finally, CYP1A2, unlike CYP3A, repeatedly co-immunoprecipitated with a number of signaling and regulatory proteins. These include nucleobindin-2 and A-kinase anchoring protein. These are involved in calcium- and protein kinase A-signaling pathways, respectively. In addition, piwi-like protein 4 influences DNA-methylation and, in turn, gene transcription processes that have been implicated in cancer progression (Litwin et al., 2017). Finally, two types of alpha globulins were repeatedly co-immunoprecipitated with CYP1A2. Alpha-2-macroglobulin is synthesized in liver but is excreted into blood where it may act as a carrier protein for growth factors and cytokines (Vandooren and Itoh, 2021). Along this line, hemoglobin subunit beta-1 also serves to carry oxygen in the blood, and ceruloplasmin, which is expressed in the liver, is known to be a carrier of copper in the blood. Major urinary protein functions as a pheromone that also enhances insulin-evoked Akt signaling (Hui et al., 2009). The association with these proteins suggests CYP1A2 may have diverse roles in cellular signaling/regulation that are mediated through its protein–protein interactions. These types of proteins are largely absent in the list of CYP3A Co-IP that were repeatedly detected in the six IP experiments with CYP3A-specific antibody – the exception being transmembrane protein 33 (discussed above).
The consistent binding partners of the P450s also do not indicate relationships between protein–protein interactions and membrane lipid microdomain localization. Most of the repeatedly identified binding partners mainly resided in the disordered domain for both P450s, with many being P450s and other drug-metabolizing enzymes. Although two forms of CYP2D (which like CYP1A2 are localized predominantly to ordered lipid microdomains) were identified in more than three of the anti-CYP1A2-IP experiments, CYP2E1 and CYP3A (which are highly partitioned to disordered lipid domains) were also identified in all six IP experiments with the CYP1A2-specific antibody. Similarly, the other proteins in all six of the CYP1A2 IP experiments were from disordered lipid regions. Although most of the consistent binding-partners of CYP3A were from disordered lipid domains, CYP2D26, microsomal glutathione S-transferase, and retinol dehydrogenase 7, which were identified in four or more of the six IP experiments with the CYP3A-specific antibody, were from ordered lipid microdomains. Thus, the microdomain localization of a P450 does not appear to preclude its interaction with enzymes from other lipid domains.
Discussion
This study compares the interactomes of two P450s (CYP1A2 and CYP3A). The P450s chosen for this comparison were previously shown to preferentially localize to different lipid microdomains in the ER (Reed et al., 2022). Because of our findings that cytochromes P450 localize to different types of lipid microdomains in a form-specific manner (Park et al., 2014; Reed et al., 2022), we wanted to ascertain whether this localization influenced their protein-protein interactions. To our knowledge, this is the first study to address the possibility that microdomain localization might influence protein–protein interactions of drug-metabolizing enzymes. It is also the first study comparing the interactomes of different P450s.
Toward our primary objective, we immunoprecipitated CYP1A2 (a P450 that localizes predominantly to ordered lipid microdomains) and CYP3A (a P450 that prefers disordered domains) and identified the associated Co-IP proteins. Our results suggest that microdomain localization does not preclude the protein–protein interactions from different microdomains. For example, CYP1A2 does not predominantly interact with proteins from ordered microdomains. In fact, more than 50% of CYP1A2 binding partners were from disordered microdomains.
In consideration of the various forms of CYP3A that were identified in this study (e.g., CYP3A1/3A23, CYP3A2, CYP3A9, CYP3A18), each was highly partitioned to disordered lipid domains. Thus, the generality of our conclusions regarding whether proteins from different lipid microdomains interact should not be affected by multiple CYP3A forms. In the same trend observed with the CYP1A2 Co-IP, CYP3A does not interact with proteins exclusively from disordered lipid regions. A higher proportion of proteins from ordered microdomains associated with CYP3A than with CYP1A2. Considering that each P450 shows a preference for different microdomains, it does not appear that protein–protein interactions are dramatically skewed toward proteins that also partition into the same lipid environment.
There could be many reasons for proteins not being effectively segregated by their distribution into different lipid domains. First, ordered, lipid microdomains are small (10–200 nm diameter in size) regions that are transient and dynamic in nature (Loura et al., 2009; Lingwood and Simons, 2010; Grassi et al., 2020). Thus, proteins from different domains could interact at the boundaries of ordered and disordered regions or could interact due to the transient nature of ordered microdomains. The results could also be a function of the experimental conditions, which showed only a fraction of the total microsomal CYP1A2 was immunoprecipitated. Thus, the interaction of proteins from different domains may reflect biased sampling of CYP1A2 from disordered and not ordered lipid regions of the membranes.
Interestingly, there were similar proportions of proteins that were immunoprecipitated by both anti-CYP1A2 and anti-CYP3A when compared on a total basis (52% and 56% of the CYP1A2 Co-IP and CYP3A Co-IP, respectively, were identified in the other Co-IP). The same trends were observed for proteins that preferentially partitioned to disordered lipid regions and those that uniformly distributed between ordered and disordered regions. The only exception was observed for the proteins from ordered lipid microdomains where 56% of these proteins in the CYP1A2 Co-IP were also observed in the CYP3A Co-IP proteins. Conversely, only 34% of the CYP3A Co-IP proteins from ordered domains were also observed in the CYP1A2 Co-IP.
The explanation for this discrepancy in common proteins from ordered lipid domains may relate to the concept of heterogeneity within the lipid raft population of a membrane (Schuck et al., 2003; Pike, 2004; Ismair et al., 2009). Many studies have demonstrated the unique protein compositions of various types of lipid rafts (Röper et al., 2000; Ballek et al., 2012). Presumably, the different types of microdomains or rafts have different functional roles.
Thus, the differences between the Co-IPs with respect to the proteins from ordered lipid environment may be the result of CYP1A2 being targeted to a specific type of ordered lipid microdomain that, in turn, favors its interaction with a specific subset of proteins within the total population of proteins isolated by our methodology (i.e., detergent solubilization). Conversely, CYP3A may interact randomly with the various types of ordered lipid microdomains as it distributes between the ordered and disordered lipid regions. Thus, our findings suggest that the interactions among the proteins that partition to the ordered lipid microdomains may be influenced by their relative affinities to specific domains within a diverse population of ordered environments, displaying subtle differences in protein/lipid composition. However, proteins that preferentially partition to disordered microdomains may interact randomly with the proteins from both the ordered and the disordered lipid environments.
Although there is not a clear relationship between microdomain localization and protein interactions involving CYP1A2 and CYP3A, the proportions of proteins associated with each domain in the two groups of immunoprecipitated proteins clearly diverge from those of total rat liver microsomes and total ER proteins. CYP1A2 was associated with a much higher proportion of proteins from disordered lipid regions than that associated with either total microsomes or total ER proteins. Conversely, the proportions of both the proteins from ordered microdomains and those uniformly distributed between ordered and disordered microdomains associated with CYP3A are much higher than the respective proportions observed in both PB liver microsomes and PB ER. Thus, the P450 binding partners are not reflective of a general sampling of the milieu of proteins in the membrane (Tables 1 and 2). Thus, the microdomain distributions of Co-IP proteins do not merely represent a random sampling of drug-metabolizing enzymes in the ER but appear to be reflective of overlapping but distinct binding specificities of the two P450s. Interestingly, substrate binding of different P450s is also characterized as overlapping but distinct with regard to the compounds that are recognized in the active sites of the P450s.
The ER displays protein/lipid heterogeneity with respect to its nucleus- and Golgi-facing ends. Thus, the localization of a protein in the ER may greatly influence its protein–protein interactions. It is possible that the lipid microdomain distributions of the Co-IPs differ from overall ER because of restricted localization within the organelle. This also could explain the similar levels of overlap (approximately 50%) in the common proteins in the two groups of Co-IPs by having the two P450s in different relative regions of the ER that have 50% overlap with one another. Following this paradigm, it could be speculated that CYP1A2 (which was associated with more Golgi-related proteins) would localize closer to the Golgi body, and CYP3A (which was associated with more ribosomal proteins) would be closer to the nucleus. Thus, our data cannot parse out the differences in Co-IP groups that are due to relative binding specificities of the P450s and variable localization of the P450s within the ER.
Our study used co-immunoprecipitation followed by LC/MS/MS to identify the binding partners of two P450s (CYP1A2 and CYP3A). Some aspects of this approach are similar to previous studies. Binding partners of human CYP3A4 were previously identified by an immunoprecipitation/shotgun LC-MS-MS approach that is similar to the one used in this study (Fujiwara and Itoh, 2014). In another study, mouse CYP2C2 was expressed as a FLAG-tagged enzyme in mouse liver by recombinant DNA techniques and adenovirus infection. After immunoprecipitation from microsomes using FLAG tag antibody, co-precipitated proteins were identified by LC/MS/MS (Li et al., 2011). Finally, a recent study explored the interactome of a modified (N-truncated and C-terminal His-tagged) human CYP2E1 (Davydov et al., 2022) that was capable of light-sensitive cross linking. After the modified CYP2E1 was incorporated into human liver microsomes, the cross-linker was activated and CYP2E1 complexes were identified by mass spectrometry. All of these studies identified many of the same P450 binding partners that were observed in this study.
Unlike the previous studies looking at P450-protein interactions by immunoprecipitation and LC/MS that used low stringency (CYP3A4) (Fujiwara and Itoh, 2014) and medium stringency (CYP2C2) (Li et al., 2011) conditions, our method used high stringency immunoprecipitation conditions that included high salt concentration and both non-ionic and ionic detergents. Thus, the protein–protein interactions identified in our study must be high-affinity to withstand the solubilizing effects of the detergents used in the wash steps. As another means to assess the relative importance of specific protein interactions with the P450s, the immunoprecipitation and analyses were repeated two times each with at least three different sets of control and PB-treated rat microsomes. Some protein interactions were detected in all six immunoprecipitations, with each of the P450-specific antibodies suggesting that some of the interactions are quantitatively more likely than others, and as a result, may have more of an impact on the cellular activities carried out by interacting proteins.
Our study has been able to compare numerous binding partners (>100 proteins for each P450) for two P450s. To distinguish consequential and incidental protein–protein interactions in the microsomes, we repeated the co-immunoprecipitation experiments six times with each antibody. Presumably, the more predominant interactions would be detected in multiple IP experiments. We arbitrarily chose interactions that were detected in three or more repetitions of the IP experiments as being consequential protein–protein interactions for each P450 (Tables 1 and 2). Both P450s showed a proclivity to bind other P450s (18 P450s were identified) with some P450–P450 interactions being prevalent for each P450. The P450–P450 interactions detected in multiple IP experiments might provide points of reference to prioritize in the examination of the potential effects of these interactions on drug metabolism by CYP1A2 and CYP3A. For CYP1A2, this would mean focusing on interactions with CYP2D, CYP2E1, and CYP3A. The CYP3A form identified as a frequent binding partner for CYP1A2 was CYP3A2 which is a male-specific, PB-inducible form (Gonzalez et al., 1986) and is also likely to be the major, constitutive CYP3A form in untreated male rats (Agrawal and Shapiro, 2003).
The interpretation regarding the specificity of protein interactions with CYP3A is complicated by the fact that there are six CYP3A forms in rats. Four forms were immunoprecipitated in the replicate experiments with CYP3A-specific antibody (CYP3A1/3A23, CYP3A2, CYP3A9, CYP3A18). Thus, the study design prevents drawing distinctions for the binding specificities of the various forms of CYP3A in rat PB microsomes. However, studies examining the potential effects of protein interactions of other P450s on metabolism by CYP3A should focus on those with CYP2B and CYP2C forms. Interestingly, both of these forms and CYP3A1, CYP3A2, and CYP3A9 are induced by PB treatment, whereas CYP3A18 was not (Mahnke et al., 1997).
As discussed above, the Co-IP were similar on a broad scale with a lot of overlap in the proteins identified. There were minor differences in the specific proteins identified in the two groups of Co-IP (such as the specific P450s that were pulled down with CYP1A2 and CYP3A) although both P450s broadly interacted with proteins related to metabolism, biosynthesis, protein folding, and the cellular responses to various forms of stress and hormonal stimulation. These processes are typical of the endoplasmic reticulum.
Because the two groups of proteins could not be readily distinguished when comparing the lists of proteins in total, the two Co-IPs were compared by considering only the proteins that co-immunoprecipitated three or more times out of six IP experiments with each P450-specific antibody. Examination of these groups showed a preponderance of proteins involved in drug and fatty acid metabolism. The CYP1A2 Co-IP also contains proteins involved in both N-glycosylation, mRNA processing, and protein folding, processes that are important in the early stages of protein translation. These associations raise the possibility that CYP1A2 can influence these processes via the specificity of its protein interactions. Most likely, this would involve inhibition of a process (such as protein folding) by competing with other protein interactions that participate in carrying out the process. The CYP1A2 Co-IP contained 11 proteins related to protein folding. Thus, CYP1A2 may affect this process at many levels through the protein interactions with the various proteins related to protein folding, and with conditions that elevate CYP1A2 expression. From these data, it seems that prolonged elevation of CYP1A2 expression could favor an unfolded protein response if these interactions are indeed inhibitory to protein folding.
Of course, it is not necessary to interact with multiple proteins to influence any given process. A protein could also affect a process by interacting with proteins that have key regulatory or functional roles. This could be the case for protein interactions with CYP1A2 as the P450 was associated three or more times in the IP experiments with proteins such as nucleobindin-2, A-kinase anchoring protein, piwi-like protein 4, alpha-2-macroglobulin, hemoglobin subunit beta-1, ceruloplasmin, and major urinary protein.
In contrast to the CYP1A2 Co-IP, CYP3A–protein interactions were almost exclusively related to fatty acid and phase I/II drug metabolism. Although the entire population of the CYP3A Co-IP included seven proteins related to steroid metabolism (which may indicate an important regulatory role in this process), none of these enzymes co-immunoprecipitated three times or more in the six different experiments. Thus, CYP3A may interact at a secondary level to regulate steroid metabolism. Although this most likely would be an inhibitory role, some P450–P450 interactions have stimulated metabolism. Similarly, CYP3A could stimulate some of these steroid reactions via its protein-protein interactions. As a result, our findings show a clear distinction in the interactomes of the two P450s which has not been reported before. Future work using gene ontology assignments will attempt to elucidate the functional roles resulting from the protein interactions involving the two P450s.
Data Availability
All supporting proteomic data are available on request made to the corresponding author.
Authorship Contributions
Participated in research design: Reed, Guidry, Backes.
Conducted experiments: Reed, Guidry.
Performed data analysis: Reed, Guidry.
Wrote or contributed to the writing of the manuscript: Reed, Backes.
Footnotes
- Received February 2, 2023.
- Accepted June 7, 2023.
This work was supported by United States Public Health Services grants from National Institutes of Health National Institute of General Medical Sciences [Grant R01 GM123253] and National Institute of Environmental Health Sciences [Grant P42 ES013648]. The Proteomics Project described was supported by grants from the National Center for Research Resources [Grant P20 RR018766] and National Institutes of Health National Institute of General Medical Sciences [Grant P20 GM103514] and currently by the National Institute of General Medical Sciences [Grant P30 GM103514].
No author has an actual or perceived conflict of interest with the contents of this article.
↵
This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- ACN
- acetonitrile
- Co-IP
- co-immunprecipitated
- CYP or P450
- cytochrome P450
- DRM
- detergent-resistant membrane
- ER
- endoplasmic reticulum
- FA
- formic acid
- IP
- immunoprecipitated
- LC-MS
- liquid chromatography-mass spectrometry
- LC/MS/MSliquid
- chromatography tandem mass spectrometry
- PB
- phenobarbital-induced
- RIPA
- radioimmunoprecipitation assay
- UGT
- UDP glucuronosyl transferase
- Copyright © 2023 by The American Society for Pharmacology and Experimental Therapeutics
