|Home | About | Journals | Submit | Contact Us | Français|
CYP3A4 is an abundant and catalytically dominant human liver endoplasmic reticulum-anchored cytochrome P450 enzyme engaged in the biotransformation of endo- and xenobiotics, including >50% of clinically relevant drugs. Alterations of CYP3A4 protein turnover can influence clinically relevant drug metabolism and bioavailability and drug-drug interactions. This CYP3A4 turnover involves endoplasmic reticulum-associated degradation via the ubiquitin (Ub)-dependent 26 S proteasomal system that relies on two highly complementary E2 Ub-conjugating-E3 Ub-ligase (UBC7-gp78 and UbcH5a-C terminus of Hsc70-interacting protein (CHIP)-Hsc70-Hsp40) complexes, as well as protein kinases (PK) A and C. We have documented that CYP3A4 Ser/Thr phosphorylation (Ser(P)/Thr(P)) by PKA and/or PKC accelerates/enhances its Lys ubiquitination by either of these E2-E3 systems. Intriguingly, CYP3A4 Ser(P)/Thr(P) and ubiquitinated Lys residues reside within the cytosol-accessible surface loop and/or conformationally assembled acidic Asp/Glu clusters, leading us to propose that such post-translational Ser/Thr protein phosphorylation primes CYP3A4 for ubiquitination. Herein, this possibility was examined through various complementary approaches, including site-directed mutagenesis, chemical cross-linking, peptide mapping, and LC-MS/MS analyses. Our findings reveal that such CYP3A4 Asp/Glu/Ser(P)/Thr(P) surface clusters are indeed important for its intermolecular electrostatic interactions with each of these E2-E3 subcomponents. By imparting additional negative charge to these Asp/Glu clusters, such Ser/Thr phosphorylation would generate P450 phosphodegrons for molecular recognition by the E2-E3 complexes, thereby controlling the timing of CYP3A4 ubiquitination and endoplasmic reticulum-associated degradation. Although the importance of phosphodegrons in the CHIP targeting of its substrates is known, to our knowledge this is the first example of phosphodegron involvement in gp78-substrate recruitment, an important step in CYP3A4 proteasomal degradation.
The hepatic endoplasmic reticulum (ER)3-anchored cytochromes P450 (P450s; CYPs) are enzymes engaged in the metabolism of numerous chemically diverse endo- and xenobiotics (1, 2). Of these, CYP3A4 is not only the most abundant P450 in the human liver, it is also the most dominant catalyst, responsible for the metabolism and elimination of over 50% of clinically relevant drugs (1, 2). Alteration of its functional hepatic content either through drug-mediated induction (enhanced expression) or inhibition and/or inactivation and subsequent proteolytic degradation is associated with serious therapeutic consequences due to altered pharmacological drug responses and/or drug-drug interactions (1,–7). We have documented that the proteolytic degradation of the native structurally and functionally intact CYP3A4 and/or its rat or mouse orthologs (CYPs 3A), and their suicidally inactivated counterparts occurs via the ubiquitin (Ub)-mediated 26 S proteasomal system (UPS) (8,–12). Furthermore, two characteristic features of these hepatic drug-metabolizing P450s are worth noting. The first is that they are integral, monotopic type I ER proteins tethered to the ER membrane via a 30–33-residue-long N-terminal helix with the bulk of their catalytic domain embedded in the lipoidal ER membrane but extensively exposed to the cytosol (13,–15). The second is that they exhibit a propensity to catalytically generate reactive oxygen species (16) that often oxidatively damage their cytosolic domain. Both of these features are consistent with their UPS-mediated turnover being qualified as an ER-associated degradation (ERAD) (17,–19), specifically an “ERAD-C” process (20,–22) with CYPs 3A and 2E1 as its typical physiological substrates. This is equally true of the ethanol-inducible CYP2E1 (23,–30), a liver P450 noted for its pathogenic role in alcoholic liver disease, diabetes, obesity, and oxidative stress (31,–33).
We and others have documented that in this ERAD-C process, CYPs 3A and CYP2E1 are first phosphorylated by protein kinases (PKA and PKC) (30, 34,–38), ubiquitinated by the cytosolic UbcH5a-CHIP-Hsc70-Hsp40 and UBC7-dependent ER-integral polytopic gp78 E2-E3-Ub-ligase complexes (29, 30, 37,–40), extracted from the ER membrane by the Npl4-Ufd1-p97/VCP-AAA ATPase chaperone complex (12, 41), and then delivered to the 26 S proteasome for degradation. Herein, we document that although each of these E2-E3 systems can function quite independently in vitro, when present concurrently as in vivo, their roles in CYP3A4 ubiquitination are complementary rather than redundant. CYPs 3A and CYP2E1 thus belong to a growing list of canonical physiological substrates of UBC7-gp78 and UbcH5a-CHIP-Hsc70-Hsp40 E2-E3-Ub ligases (42,–50).
Molecular recognition of CHIP substrates is believed to occur largely via its Hsc70-Hsp40 cochaperone complex (47,–52). However, although many UBC7-gp78 heterologous substrates have now been identified (42,–46), to our knowledge little is known about the mechanisms of their molecular recognition as specific cellular targets of this E2-E3 complex. In this study, we specifically address this issue and describe our characterization of CYP3A4 intermolecular interactions with UbcH5a-CHIP-Hsc70-Hsp40 and UBC7-gp78 E2-E3-Ub-ligase complexes, using complementary approaches of chemical cross-linking coupled with peptide mapping and high performance liquid chromatographic mass spectrometric (HPLC-MS/MS) analyses along with site-directed mutagenesis of CYP3A4 and/or the E3-ligase gp78. Our findings reveal that such E2-E3 recognition of CYP3A4 as a target substrate apparently involves Lys residues residing within negatively charged clusters of phosphorylatable Ser/Thr (Ser(P)/Thr(P)) and acidic Asp/Glu (Asp/Glu) residues situated either contiguously on CYP3A4 surface loops or clustered spatially on the surface by the P450 structural fold. Thus, we found CYP3A4 Lys residues either within or flanking these Asp/Glu/Ser(P)/Thr(P) clusters not only to be chemically cross-linked to residues in the E2-E3 protein complexes but also to be ubiquitinated by these complexes. Our findings argue that these negatively charged CYP3A4 clusters provide sites for electrostatic interactions with corresponding basic residues in the E2-E3 complexes, and thus they may serve as phosphodegrons for E2-E3 recognition.
Terephthalaldehyde (TPAL) and sodium cyanoborohydride were purchased from Sigma, and bis[sulfosuccinimidyl] suberate (BS3) and 1-ethyl-3-[3-dimethylaminopropyl] carbodiimide hydrochloride (EDC) were from ThermoFisher Scientific, Inc. (Rockford, IL). The sources of most chemicals and reagents used in these studies were indicated previously (37, 38).
To determine the role of the negatively charged surface clusters in CYP3A4 ubiquitination, selective patches consisting of three, four, or five amino acids were mutated to Ala by QuikChange Lightning mutagenesis kit (Agilent Technologies, Santa Clara, CA) or by customized order of cDNA synthesis by GenScript USA, Inc. (Piscataway, NJ), and subcloned into pTOPO3A4(His)6. The mutated CYP3A4(His)6 fragments were excised from pTOPO vector using NdeI-KpnI double restriction enzyme digestion and then ligated into the pCWori+ vector for heterologous expression in Escherichia coli DH5αF strain. The primers and templates used in this mutagenesis and the Genscript orders for each construct are listed (Table 1).
To generate glutathione transferase (GST)-fused gp78 mutants, human gp78 cDNA (pGEX-gp78C, encoding C-terminal 309–643 residues) was subjected to site-directed mutagenesis by QuikChange Lightning mutagenesis kit with the indicated primers (Table 2). The resulting PCR products were subcloned into pTOPO2.1 vector, and the clones were verified by DNA sequencing and then cloned into the final GST fusion protein expression vector, pGEX-4T2, with BamHI and XhoI enzyme digestion. However, triple cloning steps were required to generate the pGEX-gp78–313PT mutant. For this, the R307A/R308A mutant was first produced by QuikChange Lightning mutagenesis kit, and after its sequence was verified, the R307A/R308A/R310A/R311A mutant was similarly constructed with this R307A/R308A mutant as the template. This second mutant was then used as a template to generate the final 313PT “patch” mutant R307A/R308A/R310A/R311A/H312A/K313A with 6 basic residues of this domain mutated to Ala.
Human cytosolic gp78C (C-terminal domain 309–643 residues; 63 kDa) (E3) and its truncated and site-directed mutants, murine UBC7 (E2) and human C terminus of Hsc70-Interacting protein (CHIP), were expressed in E. coli and purified as described previously (30, 37,–40). C-terminally His6-tagged human CYP3A4 wild type (CYP3A4WT) and all mutants were incorporated into the pCWori+ vector and expressed in DH5α cells grown in TB media at 37 °C with isopropyl 1-thio-β-d-galactopyranoside induction and continuous monitoring of P450 content by the reduced CO-binding spectral assay (53). Recombinant CYP3A4 proteins were purified to homogeneity as described previously (36,–39). The purified proteins were stored in stock aliquots at −80 °C until use.
To ensure that the mutations did not alter CYP3A4 structure and/or function, three criteria were applied. First, the relative holo-P450 content (assayed by the reduced P450-CO-binding spectrum) to total P450 protein (determined from the total protein concentration of each P450 preparation by the bicinchoninic acid (BCA) assay) was monitored to verify that each CYP3A4 mutant was structurally comparable with the corresponding wild type (CYP3A4WT) (Table 3). Second, the functional activity of each CYP3A4 mutant relative to that of CYP3A4WT was assessed using 6β-testosterone hydroxylase, a CYP3A-selective diagnostic probe (Table 3) (37). Third, the structural conformation of CYP3A4WT and its mutants was verified by blue native-PAGE as described (54).
Purified recombinant CYP3A4(His)6 protein (250 pmol) was inactivated at 37 °C for 15 min with CuOOH (1 mm), EDTA (1 mm), and GSH (1 mm) in 50 mm Hepes buffer, pH 7.4, containing 10% glycerol in a final volume of 30 μl. DTT (2 mm, final) was added to quench the reaction. A similarly reconstituted system incubated in the absence of CuOOH served as the parallel native CYP3A4 control. The mixtures were placed at room temperature for 5 min and then on ice before use, as described (37, 38).
CYP3A4 or one of its mutants (native or CuOOH-inactivated; 250 pmol) were incubated in a reconstituted E2-E3 system consisting of Ub-activating enzyme E1 (UBA1; 0.1 μm), human UBC7 (2 μm), gp78C (1 μm), hemagglutinin (HA)-tagged Ub (120 μm) in 50 mm sodium phosphate buffer, pH 7.4, containing 20% glycerol, 0.2 mm EDTA, 0.01% sodium cholate, 10 mm magnesium chloride in a final volume of 50 μl. The reaction was initiated by the addition of an ATP-regenerating system as described previously (38).
CYP3A4 or mutant (250 pmol) was first incubated with Hsc70-Hsp40 at a final concentration of 1 μm and then mixed with UbcH5a (2 μm) and CHIP (1 μm) instead of UBC7 and gp78, sodium chloride (150 mm), and all the other reagents listed above under the reconstituted UBC7-gp78 ubiquitination system.
The reconstituted reaction mixtures were incubated at 30 °C for 90 min or for the specified time periods. Reaction mixtures that excluded CYP3A4 or ATP (−CYP3A4 or −ATP) were employed in parallel as negative controls. Aliquots (30 μl) of each reaction mixture were mixed with 4× loading buffer (containing β-mercaptoethanol (5%, v/v)) and DTT (50 mm final) and heated at 75 °C for 5 min, and then a 28-μl aliquot was subjected to SDS-PAGE on 4–15% Tris-HCl gel. In some cases, the reaction mixtures containing the ubiquitinated His6-tagged CYP3A4 proteins were “pulled down” with Talon Dynabeads, washed free of contaminating proteins first with phosphate-buffered saline (PBS) containing urea (2 m) and glycerol (10%, v/v), and then 10% glycerol/PBS, before solubilization with 40 μl of the 1× SDS-PAGE loading buffer. After SDS-PAGE, the gel was electroblotted onto a nitrocellulose membrane at ≈ 450 mA for 1.5 h. Post-transfer, the gels were stained with Coomassie Blue for verification of equivalent sample loading. The membrane was blotted with 5% nonfat milk in Tris-buffered saline (TBS) and probed with a rabbit anti-HA antibody (1:400, v/v) at room temperature overnight. The membrane was washed with 1× TBS containing 0.01% Tween 20 (TTBS) before incubation with a secondary goat anti-rabbit antibody (1:40,000, v/v) in TTBS for 1 h. The membrane was then washed six times with TTBS and rinsed with TBS before incubation with the electrochemiluminescent (ECL) substrate for pico-detection for 5 min. The film was developed and visualized using a Typhoon scanner in the chemiluminescence mode.
Please note that because these studies were time-consuming and spanned >3 years, several different batches and/or lots of purified E2 and E3 ligases as well as commercially obtained reagents (E1, UbcH5a, Hsc70, and Hsp40) were employed. Thus, because of the varied enzymatic activities of the reconstituted systems between lots, the intensity of the immunoblots tended to vary. Thus, for strict comparative purposes, the relevant −ATP and ±CYP3A4WT controls included within each immunoblot should be used as the corresponding reference points.
Incubation mixtures reconstituted as described above for either UBC7-gp78 or UbcH5a-CHIP ubiquitination analyses with either native or CuOOH-inactivated CYP3A4 or without CYP3A4 (−CYP3A4 control) were incubated at 30 °C in the presence or absence of sodium chloride at concentrations ranging from 0 to 500 mm. The reaction was initiated by ATP addition. Aliquots (30 μl) from each reaction mixture were collected at 45 or 90 min and mixed with 10 μl of 4× loading buffer, and CYP3A4 ubiquitination was examined by Western immunoblotting analyses as described above.
Five different in vitro gp78- and/or CHIP-catalyzed CYP3A4 ubiquitination systems were examined in parallel, wherein the two complete E2-E3 systems were present individually (systems 1 and 2), simultaneously (system 3), and/or sequentially (system 4 or 5) during the 90-min incubation period. In system 4, gp78 (E3) along with its cognate E2 Ub-conjugating enzyme UBC7 and CHIP were present right from the start, but CHIP-mediated ubiquitination was initiated only at 30 min by inclusion of its cognate E2 Ub-conjugating enzyme, UbcH5a. Conversely in system 5, UbcH5a-CHIP and gp78 were present from the start, but gp78-mediated ubiquitination was initiated only at 30 min by inclusion of its cognate E2 UBC7. The dual UBC7-gp78 and UbcH5a-CHIP-reconstituted ubiquitination system (system 3) consisted of UBA1 (E1, 0.1 μm), human UBC7 (2 μm), gp78C (1 μm), UbcH5a (2 μm), CHIP(His)6 (1 μm), Hsc70 (1 μm), Hsp40 (1 μm), HA-Ub (20 μm), and CuOOH-inactivated CYP3A4 (4 μm) and an ATP-regenerating system in Hepes buffer (50 mm, pH 7.4, containing 20% glycerol), EGTA (0.5 mm), EDTA (0.5 mm), with/without addition of PKC (0.016 units), and PKA (0.016 units) in a final volume of 120 μl, and it was incubated at 30 °C. Aliquots (30 μl) of the reaction mixture were collected at 30, 60, and 90 min, mixed with 10 μl of 4× loading buffer, and boiled for 5 min. Aliquots (28 μl) were subjected to SDS-PAGE and Western immunoblotting against anti-HA antibody, and immunoblots were developed and visualized as described above. Control reactions were incubated in parallel for 90 min in the absence of either ATP or CYP3A4.
All proteins used for this purpose were first dialyzed against the “cross-linking” buffer consisting of either 20 mm Hepes buffer, pH 7.4, or 25 mm sodium phosphate buffer, pH 7.4, containing 20% glycerol, EDTA (0.1 mm), and DTT (0.5 mm) before use. CYP3A4 (360 pmol), gp78C (320 pmol), and UBC7 (800 pmol) were mixed in a CYP3A4/gp78/UBC7 molar ratio of ≈1:1:2.2, in the same dialysis buffer containing DTT (10 mm) in a final volume of 75 μl, and incubated at room temperature for 30 min before the chemical cross-linking reaction. The following chemical cross-linkers were used: EDC, TPAL, and BS3. Stock solutions of EDC (20 mm) and BS3 (20 mm) were prepared in 1× phosphate-buffered saline (PBS), whereas TPAL (200 mm) was first dissolved in DMSO and then diluted 10 times with 1× PBS to a final stock concentration of 20 mm. The cross-linkers were added at the same final concentration of 2 mm, i.e. in an ≈100-fold molar excess to total proteins. Sodium cyanoborohydride (NaBH3CN) in 1× PBS at a final concentration of 20 mm was added after TPAL addition to the TPAL/cross-linking reaction mixture. All cross-linking reactions were carried out at 37 °C for 30 min in a water-bath shaker. The reactions were stopped by the addition of SDS-PAGE loading buffer and then subjected to SDS-PAGE for in-gel digestion or treated with 4 volumes of ice-cold acetone and stored at −20 °C overnight to precipitate the proteins before in-solution digestion. Digestions were carried out with both trypsin (Promega) and lysyl endoproteinase C (Lys-C; Wako) as detailed (30, 38). Briefly, for in-solution digestion, precipitated proteins were first dissolved in 50 mm ammonium bicarbonate containing 8 m urea and then treated with 10 mm Tris-(2-carboxyethyl)phosphine hydrochloride (Thermo Scientific) at 37 °C for 60 min, followed by 20 mm iodoacetamide at room temperature in darkness for 1 h to reduce and alkylate cysteine residues. After dilution with 50 mm ammonium bicarbonate to a final 2 m urea, the sample was digested overnight at 37 °C first with side chain modified trypsin (Promega), and then with Lys-C, both added at a 1:30 (w/w) ratio. The digest was acidified to 0.3% TFA, solid-phase-extracted using a C18 OMIX tip (Agilent), and vacuum-dried. In some experiments, gp78 in the cross-linking reaction mixtures was replaced with GS4B-Sepharose-bound gp78 in an attempt to enrich the gp78-interacting proteins.
Some cross-linked digests were fractionated by size exclusion chromatography on a Superdex Peptide PC 3.2 × 300-mm column (GE Healthcare) as described previously (55). Both fractionated and unfractionated cross-linked samples were analyzed on an LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific) equipped with a nanoAcquity UPLC system (Waters). Peptides were resuspended in 0.1% formic acid and separated using a chromatographic system consisting of 0.1% formic acid in water (mobile phase A), and 0.1% formic acid in acetonitrile (mobile phase B) and one of two stationary phases. For stationary phase 1, cross-linked peptide digests were loaded onto a symmetry C18 180 μm × 20-mm, 5-μm (particle size) trap column (Waters) at 5 μl/min in 3% B for 5 min at 600 nl/min before switching the output of the trap column online to a 75 μm × 150 mm, 1.7-μm (particle size) BEH130 C18 column (Waters). Peptides were eluted with a gradient of 3–27% B followed by a short wash at 50% B before returning to the initial conditions. A flow rate of 600 nl/min was maintained throughout the run. The length of the gradient varied from 30 to 150 min depending on the expected complexity of the fraction. For stationary phase 2, cross-linked peptide digests were loaded directly onto a PepMap C18 75 μm × 150 mm, 3 μm (particle size) EASY-Spray column/emitter (Thermo) for 20 min at 600 nl/min at 3% B. The flow was then reduced to 300 nl/min, and gradients were run as in stationary phase 1.
MS precursor spectra were measured in the Orbitrap from 300 to 2000 m/z at 30,000 resolving power. For TPAL cross-linked peptides, quadruply charged and higher precursors were selected and dissociated by electron transfer dissociation (ETD) or sometimes by higher energy collisional dissociation (HCD). For samples cross-linked with EDC, triply charged and higher precursors were selected and subjected to HCD activation. Both ETD and HCD product ions were measured in the Orbitrap analyzer at 7500 resolving power, although a few ETD experiments measured the product ions at low resolution in the linear ion trap.
Peak lists from all MS runs were generated as described previously (56). Peak lists were first searched against the SwissProt Database with no organism restrictions to identify proteins present in the samples. Cross-linked peptide searches were then conducted against a restricted database consisting of the 21 most abundant proteins identified in the protein search (as judged by spectral counts). These included all of the expected proteins such as CYP3A4, gp78, CHIP, Hsc70, UBC7, UbcH5a, UBA1, and other components of the in vitro ubiquitination complex, in addition to common background proteins such as trypsin, keratin, and E. coli chaperones stemming from the heterologous expression of CYP3A4 and E2-E3 proteins. Additionally, each sequence in the restricted database was randomized 10 times to generate decoy sequences. The decoy sequences were concatenated with the target database, and the combined database of 231 protein sequences (21 target + 220 decoy) was searched for cross-linked peptide spectra.
Peak lists were then searched against this database using Protein Prospector to assign cross-linked peptides (56). Carbamidomethylation of cysteine was considered as a fixed modification. N-terminal methionine loss and acetylation, N-terminal glutamine conversion to pyroglutamate, oxidation of methionine, and “dead-end” modification of lysine by the appropriate cross-linker were considered as variable modifications. A mass tolerance of 8 ppm was used for precursor ions. A tolerance of 25 ppm was used for product ions acquired in the Orbitrap analyzer and 0.7 Da for product ions measured in the ion trap. Only cross-linked products in which the more poorly matched peptide was one of the top 250 hits and had an expectation value below 50 were considered. Acceptance criteria were based on the Protein Prospector parameter “score difference.” A score difference threshold of 7.5 was selected, which resulted in a false discovery rate of 3.4% for the entire data set. The false discovery rate was calculated by dividing the number of decoy database hits by 10 to account for the larger size of the decoy database and then dividing that number by the number of matches to the target database. A cross-linked peptide was considered as a decoy match if either of the two individual peptides matched the decoy database. When a spectrum matched to both an intra-protein and inter-protein cross-link with similar scores, the intra-protein cross-link was assumed to be correct. Each cross-linked peptide was required to be at least four amino acids long, and hits in which a peptide was both cross-linked and dead-end-modified were disallowed.
Product ion spectra were inspected manually to assess whether sufficient backbone fragments were present to positively identify both of the cross-linked peptides as well as the covalently modified residues. When there was sufficient evidence to identify both of the cross-linked peptides, but the exact sites of modification were ambiguous, all possible sites were reported. Precursor ion annotations were also inspected manually.
pcDNA6 vectors encoding CYP3A4WT or its E258A/S259A/E262A/D263A/T264A or E283A/T284A/E285A mutants were transfected into HEK-293T cells. After 48 h, they were treated with or without MG132 (10 μm), and cells were harvested at 0, 4, and 8 h thereafter. Cell lysates were subjected to Talon-Dynabead “pull down” of the His6-tagged proteins and extensive washing of the beads as described above, followed by elution with SDS-PAGE loading buffer and SDS-PAGE and immunoblotting of the eluates against rabbit anti-Ub IgGs, as described previously (40).
Our previous findings indicated that knockdown of either gp78 or CHIP through lentiviral shRNA interference (shRNAi) in cultured rat hepatocytes increased the CYP3A protein and functionally active content (40). However, although shRNAi-elicited gp78 knockdown in cultured rat hepatocytes considerably reduced CYP3A export out of the ER into the cytosol, and consequently its 26 S proteasomal degradation, significant CYP3A ubiquitination (detected as higher molecular mass 35S species) still persisted both in the ER and cytosol (40). Thus, gp78 knockdown reduced but did not totally abolish CYP3A ubiquitination and ER to cytosol export (40). By remarkable contrast, shRNAi-elicited CHIP knockdown completely abrogated both CYP3A ubiquitination and its export into the cytosol (40). This is so, even though gp78 was not tampered with and was therefore present at its normal functional levels. These findings reveal that “in vivo” CHIP may serve quite early to prime CYP3A ubiquitination in the ER and that after such initial CHIP-mediated CYP3A priming, gp78 may act as an E4 to extend the polyUb chains (43, 46). However, the finding that gp78 can per se effectively ubiquitinate P450s (and its other substrates) in in vitro reconstituted systems suggests that it can function both as an E3 and an E4 in P450 ERAD/Ub-dependent proteasomal degradation (30, 37,–40).
To test this possibility more directly, we devised an in vitro approach based on functional reconstitution of purified recombinant CYP3A4 ubiquitination with either one or both E2-E3 systems (Fig. 1). Five sets of CYP3A4 ubiquitination systems were incubated in parallel at 37 °C for 0–90 min, and the relative extent of CYP3A4 ubiquitination was monitored by SDS-PAGE and immunoblotting against HA-Ub as described under “Experimental Procedures.” Concurrent incubation of CYP3A4 with either the complete functionally reconstituted UBC7-gp78 or the UbcH5a-CHIP-Hsc70-Hsp40 system led to its time-dependent ubiquitination, as revealed by the characteristic Ub ladders extending from 60 kDa and above (Fig. 1A, lanes 1–6). This indicated that each of these functionally reconstituted E2-E3 systems is capable of individually ubiquitinating CYP3A4, with the CHIP system being nominally a better catalyst of this modification than the gp78 system, judging by the ubiquitination profile extending from 60 kDa to the top of the gel at regions of >250 kDa. By contrast, inclusion of both fully reconstituted E2-E3 systems right from the start of the incubation resulted in a much more robust CYP3A4 ubiquitination pattern than that with either E2-E3 system (Fig. 1A, lanes 7–9). When UbcH5a was omitted from this dual functionally reconstituted E2-E3 system for the first 30 min of incubation, and then added at 30 min (Fig. 1A, lanes 10–12), the extent of CYP3A4 ubiquitination was greatly reduced relative to that seen with both E2-E3 systems together from the start (Fig. 1A, compare lanes 7–9 with 10–12). Conversely, if instead UBC7 was withheld for the first 30 min, and then added (Fig. 1A, lanes 13–15), the extent of CYP3A4 ubiquitination at 90 min was much more robust than that seen when UbcH5a was excluded during the first 30 min of the dual E2-E3 incubation system (Fig. 1A, lanes 10–12). However, this CYP3A4 ubiquitination was somewhat weaker than that observed with the dual system present from the very start of the incubation (Fig. 1A, lanes 7–9). Equally notable is that the CYP3A4 ubiquitination at 90 min (Fig. 1A, lane 15) is considerably greater extending to a higher molecular mass range not detected in lane 12, consistent with the generation of longer higher molecular mass Ub chains and/or additional CYP3A4 Lys residues recruited for ubiquitination. This profile approached in intensity that observed with the dual system present right from the start of the reaction (Fig. 1A, lanes 9).
A similar ubiquitination profile is observed when PKA/PKC was included in these reaction mixtures (Fig. 1B), although, as expected from our previous reports (37, 38), protein phosphorylation considerably accelerated/enhanced CYP3A4 ubiquitination by either or both E2-E3 systems. Together, these findings are entirely consistent with the possibility that when both E2-E3 systems are concurrently present, the CHIP complex may initiate the ubiquitination process, whereas the gp78 complex serves as an E4 to elongate the Ub chains initiated by CHIP.
Our previous findings that P450 protein phosphorylation indeed accelerates and enhances its ubiquitination (30, 37, 38) and that the CYP3A4 and CYP2E1 phosphorylation (Ser/Thr) and ubiquitination (Lys) sites reside within clusters of negatively charged surface Asp/Glu residues prompted us to determine whether such regions would provide interaction sites for positively charged surface residues within the components of each of the E2-E3 complexes. Indeed, similar intermolecular electrostatic interactions between P450s and their functionally relevant redox partners cytochrome P450 reductase (CPR) and cytochrome b5 (b5) have been established through site-directed mutagenesis and chemical cross-linking coupled LC-MS/MS analyses (57,–60). Thus, we posited that if such CYP3A4 negatively charged clusters were in fact to interact with positively charged residues of one or more proteins in each of these E2-E3 complexes, then such intermolecular electrostatic interactions would be disrupted in the presence of high salt concentrations in the incubation systems, thereby attenuating if not completely abrogating CYP3A4 ubiquitination. Indeed, raising the NaCl concentration from 75 to 150–300 mm in the incubation mixture greatly attenuated the time-dependent UBC7-gp78-mediated ubiquitination of the native (lanes 7–10 versus lanes 5 and 6) as well as CuOOH-inactivated CYP3A4 (lanes 15–18 versus lanes 13 and 14) in a salt concentration-dependent manner (Fig. 2A). This ubiquitination of either native (Fig. 2A, lanes 11 and 12) or CuOOH-inactivated (lanes 19 and 20) CYP3A4 was nearly abrogated at 500 mm NaCl. By contrast, the basal ubiquitination in the absence of any substrate (−CYP3A4) was attenuated only at ≥300 mm NaCl (Fig. 2A, lanes 1–4).
Corresponding analyses of time-dependent UbcH5a-CHIP-Hsc70-Hsp40-mediated CYP3A4 ubiquitination revealed a similar profile (Fig. 2B). The single notable difference was that unlike the UBC7-gp78-mediated CYP3A4 ubiquitination, the optimal NaCl concentration for this reaction was found to be 150 rather than 75 mm. NaCl concentrations above this level (i.e. 300 mm) attenuated the ubiquitination of both native and inactivated CYP3A4 species, with almost complete abrogation at 500 mm NaCl. By contrast, 300 mm NaCl enhanced the substrate-free reaction (−CYP3A4; Fig. 2B, lane 3 versus lane 1), and the basal ubiquitination even under 500 mm salt was higher than that observed at 75 mm NaCl (Fig. 2B, lane 4 versus lane 1). These findings reveal that the intermolecular CYP3A4 interactions with this CHIP-based E2-E3 complex are also electrostatic. Although tolerating higher 150 mm NaCl (salt) concentrations than those tolerated by the UBC7-gp78 complex, these intermolecular CYP3A4 interactions were also disrupted at NaCl concentrations above 150 mm. Moreover, the failure of 300 mm NaCl concentrations to affect basal substrate-free ubiquitination (Fig. 2B, lanes 1–4) suggests that this disruption is largely due to impaired intermolecular electrostatic interactions with the CYP3A4 protein rather than between any individual components of this E2-E3 Ub-ligase complex.
To identify the sites of intermolecular CYP3A4 interactions with each of the components of the two E2-E3 complexes, we employed a chemical cross-linking approach using reductive alkylation by TPAL in the presence of sodium cyanoborohydride to adduct proteins at Lys-ϵ amino groups as well as at free N termini. This chemistry yields highly charged reaction products to facilitate gas-phase isolation of cross-linked peptides through charge-state selection of quadruply charged and higher precursors as well as driving efficient product ion formation by ETD (61). TPAL can form cross-links between Lys residues whose α-carbons are within 21 Å of each other and thus may be used to “capture” any surface Lys residues within and/or in the close vicinity of CYP3A4 interacting regions (62,–65) and the individual partners of E2-E3 complexes (Fig. 3A). In additional/parallel studies, EDC (13.2 Å length) or BS3 (24.6 Å length), was also used as the cross-linker. With proteins that tend to oligomerize, i.e. gp78, CYP3A4, etc., such TPAL cross-links could include those of protein homodimers as well. Subsequent tryptic/Lys-C digestion of the TPAL cross-linked proteins, subfractionation of the cross-linked peptide digests by size-exclusion chromatography, followed by LC-MS/MS analyses of the TPAL cross-linked peptide digests wherein quadruply charged and higher precursors were selected and dissociated by ETD or sometimes by HCD (“Experimental Procedures”), yielded MS data from which peak lists were generated as described previously (56) and further analyzed as detailed under “Experimental Procedures.” Precursor ion annotations were also inspected manually. The MS/MS spectra of two TPAL cross-linked peptides (CYP3A4-gp78 and CYP3A-UBC7) so obtained are illustrated (Fig. 3, B and C). A list of these and additional TPAL cross-linked peptides is provided (Tables 4 and and5).5). Annotated product ion spectra derived as detailed (66) may be viewed online using the Protein Prospector 2 MS Viewer program on the UCSF website. The search key e86ve67zdj will retrieve the data corresponding to Table 4.
Inspection of these LC-MS/MS data revealed several notable cross-links between Lys residues of CYP3A4 and those of E2 (UBC7 or UbcH5a (UBC5)), E3 (gp78 or CHIP), and/or cochaperone (Hsc70) (Tables 4 and and5).5). Interestingly, six different TPAL cross-links between UBC7 and gp78 were found, 12 between UBC7 and CYP3A4, 15 between Hsc70 and UbcH5a, 2 between CYP3A4 and CHIP, 9 between CYP3A4 and UbcH5a, 1 between CYP3A4 and Hsc70, and 1 between Hsc70 and CHIP. Similarly, one verifiable TPAL cross-link was reproducibly detected between CYP3A4Lys-251-gp78Lys-313 that subscribed to the stipulated score difference selection criterion for significance. But two other interacting sites (gp78Lys-586-CYP3A4Lys-2574 and gp78Lys-595-CYP3A4Lys-424) were also detected, although the latter cross-link could not be validated. The plausibility of gp78Lys-586-CYP3A4Lys-257 and gp78Lys-595-CYP3A4Lys-424 as interaction sites is, however, strengthened by our findings on site-directed mutagenesis of the positively charged gp78-Lys-586 and Lys-595 patches (see below).
UBC7, however, was found to cross-link more extensively with CYP3A4, consistent with its active role as the ultimate Ub/Ub chain donor to several CYP3A4 surface Lys residues (38). Thus the following TPAL cross-links UBC7Lys-153-CYP3A4Lys-487, UBC7Lys-156-CYP3A4Lys-141, UBC7Lys-156-CYP3A4Lys-209, UBC7Lys-156-CYP3A4Lys-257, UBC7Lys-156-CYP3A4Lys-413, UBC7Lys-156-CYP3A4Lys-487, UBC7Lys-161-CYP3A4Lys-141, UBC7Lys-161-CYP3A4Lys-168, UBC7Lys-161-CYP3A4Lys-209, UBC7Lys-161-CYP3A4Lys-251, UBC7Lys-161-CYP3A4Lys-257, and UBC7Lys-161-CYP3A4Lys-266 were detected (Tables 4 and and5).5). Not surprisingly, these CYP3A4 Lys residues targeted by UBC7 not only lie within or in close proximity to the negatively charged Asp/Glu/Ser(P)/Thr(P) CYP3A4 surface clusters (62,–65), but many were documented to be bona fide ubiquitination sites (Fig. 4) (38). As expected, six cross-links between gp78 Lys residues and its cognate UBC7Lys-156 were also found in the C-terminal 600–639-residue domain contiguous to its G2BR domain, in a region believed not critical for UBC7-gp78 interaction (67,–70).5
Corresponding chemical cross-linking and LC-MS/MS analyses of CYP3A4 molecular interaction sites with the individual components of the UbcH5a-CHIP-Hsc70-Hsp40 E2-E3 system led to the identification of Hsc70Lys-512-CYP3A4Lys-266, UbcH5aLys-4-CYP3A4Lys-141, UbcH5aK8-CYP3A4Lys-168, UbcH5aK8-CYP3A4Lys-266, UbcH5aK8-CYP3A4Lys-421, UbcH5aLys-133-CYP3A4Lys-127, UbcH5aLys-144-CYP3A4Lys-141, UbcH5aLys-133-CYP3A4Lys-168, UbcH5aLys-133-CYP3A4Lys-257, and UbcH5aLys-133-CYP3A4Lys-266 (Tables 4 and and5).5). Similarly, CHIPLys-125-CYP3A4Lys-257 and CHIPLys-22-CYP3A4Lys-421 were also detected, thereby revealing that CHIP does indeed interact directly with CYP3A4 (Tables 4 and and5).5). Just one cross-link between Hsc70Lys-512 and CYP3A4Lys-266 was detected,6 whereas 15 cross-links between Hsc70 and UbcH5a were identified.
Preliminary structural deletion analyses to determine the gp78C domains that were relevant to CYP3A4 ubiquitination confirmed that residues 341–378 (RING, catalytic core), 456–487 (Cue, Ub-binding domain), 579–600 (G2BR, UBC7-interacting domain), and 614–620 (VIM, VCP/p97-interacting domain) (Fig. 5A) were also important for its CYP3A4 ubiquitination (Fig. 5B). The first three of these gp78 domains have been found to be relevant to its Ub ligase function, its autoubiquitination, as well as the ERAD of CD3δ, a heterologous substrate (67, 68).
gp78Lys-313, a residue identified through our chemical cross-linking/proteomic analyses to be interacting with CYP3A4Lys-251 (Tables 4 and and5),5), resides within a positively charged N-terminal patch consisting of residues Arg-307–Arg-308–Arg-310–Arg-311–His-312–Lys-313. Mutation of just gp78Lys-313 to Ala failed to affect its CYP3A4 ubiquitination (Fig. 5B). However, concurrent mutation of each of these basic gp78 residues to Ala greatly reduced time-dependent CYP3A4 ubiquitination relative to that observed with the wild type gp78C protein (Fig. 5C). To enable parallel comparisons of multiple mutants and/or conditions within the same immunoblot, a 90-min period of incubation that affords near-maximal CYP3A4WT ubiquitination was employed in all subsequent analyses (Fig. 5D). Similar mutation to Ala of another positively charged C-terminal gp78 Gln-584–Arg-585–Lys-586 domain containing the CYP3A4Lys-257-interacting Lys-586 residue also reduced UBC7-gp78-mediated CYP3A4 ubiquitination (Fig. 5D). This was also true if the positively charged gp78 C-terminal Arg-594–Lys-595–Arg-596 residues that include the putative CYP3A4Lys-424-interacting Lys-595 residue were all mutated to either Ala or Val7 (Fig. 5D). Further diminution of UBC7-gp78-mediated CYP3A4 ubiquitination was observed if all the positively charged residues in either the gp78 N-terminal Arg-307–Arg-308–Arg-310–Arg-311–His-312–Lys-313 and C-terminal Gln-584–Arg-585–Lys-586 patches or in both the gp78 C-terminal Gln-584–Arg-585–Lys-586 and Arg-594–Lys-595–Arg-596 patches were to be concurrently mutated to Ala (Fig. 5D). These findings revealed that the positively charged gp78 residues bracketing its UBC7-interacting RING domain on its N-terminal end and those within its UBC7-interacting G2BR domain at its C-terminal end are all relevant to UBC7-gp78-mediated CYP3A4 ubiquitination (Fig. 5D). Furthermore, these findings also suggest that such positively charged gp78 patches could be important for intermolecular electrostatic interactions with the negatively charged CYP3A4 Asp/Glu/Ser(P)/Thr(P) clusters consistent with our chemical cross-linking/proteomic findings as well as the high salt effects on CYP3A4 ubiquitination discussed above.
The role in UBC7-gp78-mediated CYP3A4 ubiquitination of some of its negatively charged residues in the close vicinity of its chemically cross-linked Lys residues was also examined through site-directed mutagenesis. Three negatively charged CYP3A4 surface clusters (each circumscribed by a yellow oval ring, Fig. 6A) were initially selected for our scrutiny as follows. (i) The first acidic cluster scrutinized was the Glu-283–Thr-284–Glu-285 surface loop that also contains the UBC7-gp78-ubiquitinated residue Lys-282 and the highly PKC-phosphorylated Thr-284 residue (38), and being highly disordered is not featured in the existing crystal structures (62,–65). (ii) The second acidic cluster was Glu-258–Ser-259–Glu-262–Asp-263–Thr-264, a CYP3A4 surface cluster, containing Ser-259 and Thr-264, two residues found to be highly phosphorylated by PKC in vitro (37, 38). This cluster is not only contiguous to Lys-257 but is also flanked by Lys-251 and Lys-266, all three CYP3A4 Lys residues shown through chemical cross-linking to be interacting with UBC7, gp78, Hsc70, CHIP, and/or UbcH5a (Tables 4 and and5).5). One CYP3A4 crystal structure reveals that this region is also highly disordered (63). (iii) The third acidic cluster Lys-115–Ser-116–Glu-122–Asp-123–Glu-124–Glu-125 cluster containing the UBC7-gp78-ubiquitinated Lys-115 residue was also selected for our analyses, but the corresponding Ala mutant consistently failed to express, even after coexpression of molecular chaperones to facilitate its proper folding. A similarly failed expression was previously reported with the CYP3A4 K127A mutant as well (59), revealing some critical sine qua non intrinsic requirements of these particular residues for CYP3A4 structural integrity. Thus for now, the analyses of this particular cluster had to be shelved until some viable protein becomes available. (iv) The fourth acidic cluster Thr-264–Ser-420–Ser-478 triple Ala mutant was also included as a control, as we have shown that the simultaneous site-directed Ala mutation of these residues not only impairs UBC7-gp78-mediated ubiquitination of CuOOH-inactivated CYP3A4, but also it attenuates its proteasomal turnover in yeast and HEK-293T cells relative to CYP3A4WT (37, 38). In addition, because our more recent findings of CYP3A4 ERAD in cultured hepatocytes (40) revealed that both gp78 and CHIP systems are involved in the ubiquitination of not just the CuOOH-inactivated but also of the native CYP3A4 species, we examined the relevance of each of these clusters to the ubiquitination of both the native CYP3A4 and its CuOOH-inactivated species, as well as that of their corresponding phosphorylated (+PKA/PKC) and nonphosphorylated (−PKA/PKC) forms.
Incubation of native CYP3A4WT at 37 °C for 90 min in a complete UBC7-gp78 ubiquitination system resulted in significant ubiquitination of the native protein, and this was further enhanced upon inclusion of PKA/PKC in the incubation mixture, consistent with a phosphorylation-enhanced effect (Fig. 6B, lanes 2 and 3). As described previously (37, 38), mutation of CYP3A4S478 to Ala reduced this ubiquitination, but inclusion of PKA/PKC was able to restore some of this loss, possibly due to CYP3A4 phosphorylation at sites other than Ser-478 (Fig. 6B, lanes 6 and 7). Ser-478 mutation to the phosphomimetic Asp residue considerably enhanced the ubiquitination of the native CYP3A4 (Fig. 6B, lane 8), as observed previously with the CuOOH-inactivated CYP3A4S478D mutant (38). Indeed, this S478D mutation alone maximally increased native CYP3A4S478D ubiquitination to levels much greater than those seen with the PKA/PKC-phosphorylated native CYP3A4WT (Fig. 6B, lane 8 versus lane 3). PKA/PKC inclusion in this incubation produced no further enhancement of CYP3A4S478D ubiquitination (Fig. 6B, lane 8 versus lane 9). The UBC7-gp78-mediated ubiquitination of the native CYP3A4 T264A/S420A/S478A mutant as reported previously (37, 38) was also greatly impaired (Fig. 6B, lane 10 versus lane 2). This impairment was slightly remedied upon inclusion of PKA/PKC in the incubation, as seen in the case of the S478A mutant (Fig. 6B, lanes 11 and 7). UBC7-gp78-mediated ubiquitination of the CuOOH-inactivated CYP3A4WT and its S478A, S478D, and T264A/S420A/S478A mutants is consistent with our previously reported findings (37, 38). The ubiquination of the CuOOH-inactivated CYP3A4 proteins was more robust relative to that of the corresponding native proteins, consistent with the likelihood that CuOOH-mediated structural damage exposes additional degrons and/or ubiquitination sites for E2-E3 recognition. Together, these findings yet again underscore the critical importance of Ser-478 phosphorylation to the UBC7-gp78-mediated ubiquitination of the native as well as inactivated CYP3A4 proteins.
Moreover, the UBC7-gp78-mediated ubiquitination of the native CYP3A4 E283A/T284A/E285A mutant despite retaining the UBC7-gp78-ubiquitinatable Lys-282 and the PKA-phosphorylatable Ser-478 residue was greatly impaired (Fig. 6B, lanes 4 and 5 versus lanes 2 and 3). PKA/PKC inclusion slightly improved this impairment (Fig. 6B, lane 5 versus lane 3), and further slight remediation was detected upon CuOOH inactivation of this CYP3A4 E283A/T284A/E285A mutant (Fig. 6B, lane 16 versus lane 5), which may expose additional Ser/Thr-based phosphodegrons and/or ubiquitinatable Lys residues for E2-E3 recognition (Fig. 6B, lanes 15 and 16 versus lanes 4 and 5). Together, these findings reveal the critical importance of this disordered CYP3A4 surface loop to the UBC7-gp78-mediated ubiquitination of both the native and the CuOOH-inactivated proteins.
In preliminary studies, we examined the relative roles of Ser-259, Thr-264, and the Glu-258–Glu-262–Asp-263 cluster in the UBC7-gp78-mediated CYP3A4 ubiquitination through their Ala mutation (data not shown). Ser-259 and Thr-264 are two highly PKC-phosphorylatable residues belonging to the Glu-258–Ser-259–Glu-262–Asp-263–Thr-264 structural cluster, whose phosphorylation would further augment the negative charge of this cluster. Relative to CYP3A4WT, the native S259A mutant exhibited reduced ubiquitination that was remedied to some extent by PKA/PKC inclusion in the reaction mixture. The ubiquitination of the native T264A and the S259A/T264A mutants was not appreciably impaired in the presence/absence of PKA/PKC. Conversely, the UBC7-gp78-mediated ubiquitination of the native CYP3A4 E258A/E262A/D263A mutant was only slightly decreased relative to that of CYP3A4WT (Fig. 6C, lanes 4 and 5 versus lanes 2 and 3). However, the additional mutation of Ser-259 and Thr-264 within this cluster further reduced the extent of this ubiquitination (Fig. 6C, lanes 6 and 7 versus lanes 2 and 3). When this E258A/S259A/E262A/D263A/T264A mutant was directly compared against the E283A/E285A and E283A/T284A/E285A mutant (Fig. 6C, lanes 6 and 7 versus lanes 8 and 9), the Glu-258–Ser-259–Glu-262–Asp-263–Thr-264 residues were not as critical as either Glu-283 and Glu-285 or Glu-283–Thr-284– Glu-285 (Fig. 6C, lanes 8–11) to UBC7-gp78-mediated CYP3A4 ubiquitination. Similar conclusions could also be drawn from the ubiquitination profiles of CuOOH-inactivated CYP3A4WT and its S259A, T264A, and S259A/T264A (data not shown) or E258A/E262A/D263A and E258A/S259A/E262A/D263A/T264A mutants (Fig. 6C, lanes 13–22).
In preliminary studies, we also assessed the relative roles of Glu-258–Glu-262–Asp-263 cluster versus that of CYP3A4 Ser-259 and Thr-264 residues (singly and in combination) to CHIP-mediated ubiquitination (Fig. 7A). CHIP-mediated ubiquitination of native S259A or T264A mutant in the presence or absence of PKA/PKC was not appreciably altered (Fig. 7A, lanes 4–7 versus lanes 2 and 3) but that of the double S259A/T264A mutant was clearly reduced relative to that of the native CYP3A4WT (Fig. 7A, lane 8 versus lane 2). This was particularly noticeable in the presence of PKA/PKC (Fig. 7A, lane 9 versus lane 3), thereby underscoring the relative importance of phosphorylation at either Ser-259 or Thr-264 site to the CHIP-mediated ubiquitination of native CYP3A4. Furthermore, the ubiquitination of the native CYP3A4 E258A/E262A/D263A mutant was greatly attenuated relative to that of the native CYP3A4WT, irrespective of the presence or absence of PKA/PKC in the incubation (Fig. 7A, lanes 10 and 11 versus lanes 2 and 3). Essentially similar findings were observed upon CuOOH inactivation, with the possible exception of the S259A mutant whose ubiquitination in the absence of PKA/PKC seemed to be slightly more affected than that of the corresponding CYP3A4WT protein (Fig. 7A, lane 13 versus lane 15). Together these findings underscore the relevance of the negatively charged Glu-258–Ser-259–Glu-262–Asp-263–Thr-264 cluster to CHIP-mediated ubiquitination of the native as well CuOOH-inactivated CYP3A4.
To further assess the relative relevance of the negatively charged CYP3A4 Glu-258–Glu-262–Asp-263, Glu-258–Ser-259–Glu-262–Asp-263–Thr-264 and Glu-283–Thr-284–Glu-285 clusters to CHIP-mediated CYP3A4 ubiquitination in vitro, the ubiquitination of their corresponding Ala mutants by this system was monitored in parallel (Fig. 7B). Relative to that of the native CYP3A4WT, the ubiquitination of the native CYP3A4 E258A/E262A/D263A mutant was decreased to a greater extent than that of the native CYP3A4 E283A/T284A/E285A mutant (Fig. 7B, lanes 4 and 8 versus lane 2). This difference was maintained (Fig. 7B, lanes 5 and 9 versus lane 3) even after P450 phosphorylation that greatly increased the overall level of CHIP-mediated ubiquitination of native CYP3A4WT and its two mutants. Furthermore, additional Ala mutation of Ser-259 and Thr-264 within this Glu-258–Glu-262–Asp-263 cluster nearly abrogated CHIP-mediated native CYP3A4 ubiquitination relative to that of CYP3A4WT (Fig. 7B, lane 6 versus lane 2). By contrast, the Ala mutation of Thr-284 within the Glu-283–Thr-284–Glu-285 cluster had considerably lesser effects (Fig. 7B, lane 8 versus lane 2 or 10). Upon CuOOH-mediated CYP3A4 inactivation, although the overall level of ubiquitination was increased (Fig. 7B, lanes 13–22), similar differences between the CYP3A4WT and these mutants were observed, although the inclusion of PKA/PKC in the ubiquitination system to phosphorylate the P450 proteins further improved the overall extent of their ubiquitination. These findings thus reveal that of these two negatively charged clusters, the Glu-258–Ser-259–Glu-262–Asp-263–Thr-264 cluster may play a greater role than the Glu-283–Thr-284–Glu-285 cluster in native CYP3A4 ubiquitination by the CHIP complex.
Given that these E2-E3 systems act in a complementary rather than a redundant fashion when they are present together (Fig. 1), we explored their dual effects on CYP3A4 ubiquitination upon mutation of some of the individual Asp/Glu/Ser/Thr residues in each of these clusters to Ala (Fig. 8). Relative to that of the native CYP3A4WT, an appreciable impairment in such ubiquitination was evident with S259A and T264A mutants, which was further magnified when these mutations were coupled in the S259A/T264A mutant (Fig. 8A, lanes 4, 6, and 8 versus lane 2). These differences were maintained even after inclusion of PKA/PKC in the incubation to phosphorylate the available Ser/Thr residues (Fig. 8A, lanes 5, 7, and 9 versus lane 3). Upon CuOOH inactivation, some of these effects were to some extent mitigated as the observed differences appeared less pronounced than those observed with the native proteins (Fig. 8A, lanes 15–19 versus lane 2 and 3), irrespective of the presence or absence of PKA/PKC.
However, a much more pronounced impairment of this dual ubiquitination was observed with the native CYP3A4 E258A/E262A/D263A mutant both in the phosphorylated or nonphosphorylated state (Fig. 8A, lanes 10 and 11 versus lanes 2 and 3), and this seemed to be the case with the corresponding CuOOH-inactivated protein (Fig. 8A, lane 21 versus lane 13). Inclusion of PKA/PKC in the latter incubation to some extent restored this loss (Fig. 8A, lane 22 versus lane 14). Together, these findings indicate that this Glu-258–Ser-259–Glu-262–Asp-263–Thr-264 surface cluster that is relevant to CHIP-mediated CYP3A4 ubiquitination is also relevant to the concerted CYP3A4 ubiquitination by both E2-E3 systems.
To assess the relative relevance of each of the three CYP3A4 clusters examined in the individual E2-E3 systems to this dual E2-E3 Ub-ligase ubiquitination system and thus determine their potential in vivo relevance, we directly compared the native and CuOOH-inactivated CYP3A4 E258A/E262A/D263A, E258A/S259A/E262A/D263A/T264A, E283A/T284A/E285A, E283A/E285A, and T264A/S420A/S478A mutants with the corresponding native or inactivated CYP3A4 WT in parallel incubations with both E2-E3 systems (Fig. 8B). Relative to the native CYP3A4WT, the ubiquitination of the native E258A/E262A/D263A mutant was reduced, and this was further reduced when Ser-259 and Thr-264 mutations were included in the E258A/S259A/E262A/D263A/T264A mutant (Fig. 8B, lane 2 versus lanes 4 and 6, respectively). Inclusion of PKA/PKC in these incubations mitigated these reductions to some albeit incomplete extent. However, relative to the native CYP3AWT, the ubiquitination of the native E283A/T284A/E285A and E283A/E285A mutants was decreased considerably (Fig. 8B, lane 2 versus lanes 8 and 10, respectively), and the inclusion of PKA/PKC in the incubations failed to substantially restore this loss. Similar conclusions could be drawn with the corresponding CuOOH-inactivated CYP3A4 proteins, although as expected the relative extent of their ubiquitination was enhanced by the inactivation (Fig. 8B, lanes 13–22). The minor relevance of the T284A mutation to this dual CYP3A4 ubiquitination only became apparent upon CuOOH inactivation and PKA/PKC phosphorylation of the protein (Fig. 8B, lane 22 versus lanes 14 and 20). By contrast to these clusters, but consistent with our previous reports (37, 38), the dual E2-E3 ubiquitination of the CYP3A4 T264A/S420A/S478A mutant was greatly impaired whether in the native (Fig. 8B, lanes 23 and 24 versus lanes 2 and 3) or CuOOH-inactivated form (Fig. 8B, lanes 25 and 26 versus lanes 13 and 14), yet again underscoring the critical relevance of the Ser-478 residue to CYP3A4 ubiquitination.
Independently of their enhanced ubiquitination exhibited over the corresponding −ATP and −CYP3A4WT controls included in every immunoblot, the CYP3A4His6-tagged proteins were verified to be genuinely ubiquitinated in these in vitro systems through Talon-Dynabead pulldown of the His6-tagged proteins (Fig. 9, A and B). The relative physiological ubiquitination of these proteins was also similarly verified upon HEK-293T cell transfection of pcDNA6 vectors encoding CYP3A4WT or each of these mutant proteins at 0, 4, and/or 8 h in the presence or absence of the proteasomal inhibitor, MG132 (Fig. 9C). These observations of the differential modification of these CYP3A4 proteins by the concurrently present intracellular ubiquitination systems are fully consistent with our findings of their differential susceptibility to in vitro ubiquitination (Fig. 8).
Collectively, our findings on CYP3A4 ubiquitination in vitro with two functional E2-E3 systems reveal that although these two systems can function entirely independently of each other, when they are simultaneously present they are synergistic rather than redundant. In the intact hepatocyte where they coexist physiologically, the CHIP-based E2-E3 system may be important for the initial ubiquitination of a P450 Lys residue, with subsequent elongation of the nascent Ub chain by UBC7-gp78 E2-E3 system acting as an E4 enzyme. Such functional synergism is not only consistent with our previous findings of shRNAi-mediated knockdown of either gp78 or CHIP on overall intracellular CYP3A4 ubiquitination discussed earlier (40), but it is also supported by our in vitro findings of concurrent CYP3A4 ubiquitination by this dual E2-E3 system (Figs. 1 and and8).8). Accordingly, the presence of both E2-E3 systems resulted in a robust CYP3A4 ubiquitination profile with a greater number of higher molecular mass poly-Ub chains extending to >250 kDa, which is also considerably more intense than that seen individually with either of these two E2-E3 systems. This profile may also reflect additional CYP3A4 Lys residues possibly recruited for ubiquitination by each of the E2-E3 systems. Physiologically, this concerted synergistic function of both E2-E3 systems would both accelerate and enhance CYP3A4 ubiquitination and its subsequent 26 S proteasomal degradation. A similar functional synergism between different cellular Ub ligases acting cooperatively as E3s and E4s in the ubiquitination of heterologous substrates has been documented (71,–75).
Based on our previous identification of CYP3A4 and CYP2E1 PKA/PKC-Ser/Thr phosphorylation sites and UBC7-gp78- and CHIP complex Lys ubiquitination sites, and recognition of their specific location within or vicinal to Asp/Glu surface clusters, we have proposed that such negatively charged surface clusters are critical to P450 molecular recognition by these E2-E3 systems (30, 38, 76). Several converging lines of evidence based on well noted biological features of these interacting proteins and/or similar ubiquitination processes supported this possibility. These include the following: (i) intermolecular electrostatic interactions are the sine qua non of CYP3A4 functional interactions with its redox partners CPR and cytochrome b5 (57,–60); (ii) PKA/PKC-mediated CYP3A4 phosphorylation not only greatly accelerates/enhances its ubiquitination by both E2-E3 systems but also Ser-259, Thr-264, and Thr-284 residues residing within two of these negatively charged CYP3A4 Asp/Glu surface clusters are indeed highly phosphorylated by PKC in vitro (37, 38). This phosphorylation would further boost the negative charge of each of their respective CYP3A4 Asp/Glu clusters; (iii) the ubiquitination of ER-integral inositol 1,4,5-triphosphate receptors also occurs at Lys residues residing within their cytosol-exposed surface loops (77); (iv) a positively charged residue in the immediate vicinity of the catalytic E2-Cys that discharges the activated Ub-thioester species, serves as a “hook” for an acidic residue in the proteins ubiquitinated solely by Ubc4 or Ubc5 (78). This infers similar intermolecular electrostatic interactions between these E2s and their corresponding substrates. (v) gp78-mediated ubiquitination of 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase preferentially occurs at Lys residues within its KEEE or KNEEEE motifs (79); (vi) in addition to phosphorylatable residues, yeast integral proteins exhibit two consecutive acidic residues at −2 and −1 of an ubiquitinatable Lys residue (0-position) (78). Indeed, these combined considerations led us to propose that the molecular recognition of CYP3A4 by UBC7-gp78 and/or CHIP systems as a ubiquitination substrate was dictated by similar principles. Therefore, the Ser/Thr phosphorylation within the negatively charged CYP3A4 surface Asp/Glu clusters by raising the negative charge of the cluster engendered phosphodegrons for enhanced molecular recognition by positively charged E2-E3 residues (30, 37, 38, 76).
Consistent with our proposal, the global proteomic analyses of protein ubiquitination utilizing the elegant Ub remnant profiling approach has indeed revealed the relatively high propensity for the detection of Asp/Glu/Ser/Thr residues flanking ubiquitinated Lys residues (80,–83). In this approach, a monoclonal antibody rather selectively recognizes the Ub-derived diGly remnant (GGK) on the ubiquitinated Lys residues derived from the tryptic digestion of the Ub-modified proteins, enabling the 1000-fold enrichment of GGK-modified peptides, with a characteristic added mass of 114.0429 Da (84). This GGK remnant mass signature of ubiquitinated proteins has enabled the unprecedented global identification of ubiquitinated cellular proteins by LC-MS/MS analyses (80,–84), leading to the identification of linear peptide sequences in this vast “ubiquitinome” in which the ubiquitinated Lys residues are often flanked by Asp/Glu/Ser/Thr residues (81, 82).
However, using the structurally defined CYP3A4 protein (62,–65) as a model ubiquitination substrate, we now document that such P450 Lys ubiquitination not only occurs on contiguous linear motifs (i.e. Lys-282–Glu-283–Thr(P)-284–Glu-285) along cytosol-exposed surface loops that conform to canonical phosphodegrons but also involves interactions with Asp/Glu surface clusters spatially associated with phosphorylatable Ser/Thr residues (i.e. Glu-258–Ser(P)-259–Glu-262–Asp-263–Thr(P)-264; Lys-115–Ser(P)-116–Glu-122–Asp-123–Glu-124–Glu-125) that are assembled together by the P450 tertiary structure and thus represent “conformational” phosphodegrons (Fig. 4). As proposed earlier (30, 37, 38, 76), such P450 Ser/Thr phosphorylation by imparting additional negative charge would further magnify the overall negative charge of a specific spatially associated Asp/Glu cluster and thus serve as a molecular switch to enhance P450 recognition as an E2-E3 ubiquitination substrate, thereby accelerating its proteasomal degradation and controlling the timing of its ERAD.
The validity of our proposal was tested through four complementary approaches as follows. (i) In vitro high salt treatment aimed at disrupting such intermolecular electrostatic interactions of CYP3A4 with the components of each of the two E2-E3 systems. (ii) Characterization of these intermolecular interactions through chemical cross-linking of the surface Lys residues coupled with LC-MS/MS analyses of the proteolytic digests of these protein complexes to identify their specific interaction sites. (iii) Assessment of the role of CYP3A4 Asp/Glu/Ser/Thr clusters in its ubiquitination upon site-directed mutagenesis of these residues individually and/or in combination. (iv) Examination through site-directed mutagenesis of the role of gp78 E3-ligase positively charged “basic patches” containing its CYP3A4-cross-linked Lys residues in CYP3A4 ubiquitination.
Our chemical cross-linking LC-MS/MS analyses revealed two CYP3A4-CHIP and nine CYP3A4-UbcH5a cross-linked Lys residues that could be consistently identified (Tables 4 and and5).5). It is possible, as suggested previously (47,–51), that CHIP interacts largely indirectly with CYP3A4 via Hsc70 and/or UbcH5a.8 However, the detection of just one cross-link between Hsc70 and CYP3A4, albeit surprising, could be entirely attributed to the well recognized nature of this chaperone's essentially hydrophobic rather than electrostatic substrate interactions and thus the absence of any strategically located Lys residues vicinal to their intermolecular interacting regions. Another plausible reason is that unlike the quite successful detection of constitutive subunit interactions of relatively stable protein complexes via cross-linking/LC-MS/MS analyses, the capture and subsequent identification of transient and substoichiometric interactions are much more difficult via this approach. The rather rigorous selection criteria employed in our cross-linking analyses is yet another factor. Nevertheless, all the cross-linked Lys residues thus identified are also found on the cytosol-exposed CYP3A4 surface, in close proximity to its negatively charged Asp/Glu/Ser(P)/Thr(P) clusters, and some (Lys-127 and Lys-168) indeed have been previously identified as CHIP ubiquitination sites (Fig. 4B) (38).
Unlike the paucity of Hsc70-CYP3A4 cross-links, 15 cross-links between Hsc70 and UbcH5a were identified. We find it notable that five of these Lys sites are also UbcH5a-CHIP-Hsc70 ubiquitination sites,9 consistent with Hsc70 itself serving as a target in the absence of a canonical substrate (49). Moreover, these observations also argue against the plausible spurious/artifactual nature of these cross-linking interactions. These findings are also entirely consistent with the well recognized function of Hsc70 as a chaperone in the triage of cellular proteins (47,–52), as well as of UbcH5a, one of the few CHIP-committed E2s, in this process (47,–51). Furthermore, our findings that CYP3A4 phosphorylation enhances its CHIP-mediated ubiquitination are also entirely consistent with protein phosphorylation engendering phosphodegrons, a noted feature of CHIP-mediated recognition of several substrates (i.e. pTau, α-synuclein, and androgen receptor, etc.) as ubiquitination targets (85,–90).
Our analyses of CYP3A4 interactions with the UBC7-gp78 system were far more revealing of its intrinsic molecular interactions with either of these proteins. These findings indicated that CYP3A4 interacts extensively with UBC7, the ultimate Ub donor in this system, through several of its surface Lys residues (Tables 4 and and5).5). Of these, Lys-168 and Lys-487 are previously identified CYP3A4 sites ubiquitinated by gp78 and/or CHIP systems (38), whereas others (Lys-141, Lys-251, Lys-257, Lys-266, and Lys-421) lie within and/or are vicinal to its negatively charged Asp/Glu surface clusters (Fig. 4). More importantly, these analyses also revealed that CYP3A4 interacts through its Lys-251 and Lys-421 residues with gp78-Lys-313, its Lys-257 with gp78-Lys-586, and its Lys-424 with gp78-Lys-595. We wish to note that some of these interactions were reproducibly detected through several independent iterations of similar approaches employing different chemical cross-linkers (Tables 4 and and5),5), underscoring the likelihood that such interactions are also real and entirely predictable, rather than spurious due to nonspecific chemical cross-linking.
Inspection of the CYP3A4 crystal structure (62,–65) intriguingly reveals that its negatively charged Glu-258–Ser(P)-259–Glu-262–Asp-263–Thr(P)-264 surface cluster is spatially flanked by its Lys-251 residue on one end and by its Lys-266 residue on the other end (Fig. 4A). Both of these residues are known to interact with the components of gp78 and CHIP complexes (Tables 4 and and5).5). Thus, if the P450 molecule were to be “butterflied” longitudinally across the surface diametrically opposite to this Lys-251–Glu-258–Ser(P)-259–Glu-262–Asp-263–Thr(P)-264–Lys-266 cluster, it would reveal that the ubiquitinated CYP3A4 Lys residues are somewhat symmetrically positioned on each of its two cytosol-exposed surfaces splayed on either side of this cluster (Fig. 4B). Notably, each of these surfaces also includes other Lys residues (Lys-209 and Lys-413 on one CYP3A4 surface and Lys-141, Lys-257, and Lys-421 on the other diametrically opposite surface) identified through our cross-linking analyses as UBC7-gp78 interaction sites (Tables 4 and and5;5; Fig. 4). Accordingly, electrostatic interactions with the CYP3A4 acidic surface clusters and these additional Lys interaction sites would enable the UBC7-gp78 oligomeric complex to target specific Lys ubiquitination sites in each of these diametrically opposite CYP3A4 surfaces (Fig. 4). Thus, electrostatic interactions with CYP3A4 Glu-283–Thr-284–Glu-285 and Ser-116–Glu-122–Asp-123–Glu-124–Glu-125–Glu-294 acidic clusters, as well as its Lys-141, Lys-257, and Lys-421 residues would enable the UBC7-gp78 complex to ubiquitinate Lys-282 and Lys-115 residues on one CYP3A4 surface (Fig. 4). Similar interactions with the Glu-163–Glu-165, Ser-478–Glu-486, and Glu-333–Glu-494–Asp-497 acidic clusters as well as Lys-209, Lys-288, and Lys-413 on the diametrically opposite CYP3A4 surface would enable the ubiquitination of its Lys-168 and Lys-492, as reported previously (38). Intriguingly, the latter CYP3A4 surface also contains Ser-478, a residue not only critical for UBC7-gp78-dependent CYP3A4 ubiquitination, but the only residue other than Ser-259 on this CYP3A4 surface to be significantly phosphorylated (Fig. 4) (38). Notably, given that UBC7-gp78 operate as oligomeric E2-E3 complexes (44, 46, 91, 92), their functional interactions with each CYP3A4 surface may entail the concurrent action of two or more distinct E2-E3 heteromeric units.
Several topological features of these intermolecular interactions are worth noting. First, the CYP3A4 Lys residues identified as UBC7-gp78 ubiquitination and interaction sites are also all located in its cytosol-exposed domain distal to its ER-embedded domain and thus fully accessible to the largely cytosolic ubiquitination machinery (Fig. 4). Second, it is intriguing that of these ubiquitinated Lys residues, Lys-282 resides in a highly disordered and negatively charged CYP3A4 Glu-283–Thr-284–Glu-285 surface loop, whereas Lys-115 lies within the Ser(P)-116–Glu-122–Asp-123–Glu-124–Glu-125 spatially associated surface cluster. Furthermore, both CYP3A4 Glu-258–Ser(P)-259–Glu-262–Asp-263–Thr(P)-264 surface cluster and the Glu-283–Thr-284–Glu-285 surface loop map to its structurally highly mobile “Plastic Region” PR4 (62,–64), as defined for the structurally similar CYP2B4 protein (93). On the CYP3A4 surface opposite that of Lys-282 and Lys-115, the known UBC7-gp78 ubiquitination sites Lys-168 and Lys-492 along with the UBC7-interacting Lys-487 also lie in close vicinity of its C-terminal surface loop (composed of residues 475–485), another region similarly noted for its highly pronounced structural mobility/plasticity (62,–64). Thus, the UBC7-gp78 complex apparently favors CYP3A4 Lys residues on its structurally plastic domains as ubiquitination targets. Third, many of these CYP3A4 Lys interaction/ubiquitination sites and Ser(P)/Thr(P) residues lie in the “positively charged Arg/Lys bowl,” a P450 interface involved in intermolecular interactions with corresponding CPR or b5 negatively charged residues (33, 57,–60, 62,–64, 94). Phosphorylation of these residues would not only firmly disrupt such P450 functional interactions but also foster lethal interactions with the E2-E3 ubiquitination complexes thereby committing the P450 protein firmly to ERAD.
This was assessed through site-directed mutagenesis of residues within each CYP3A4 surface cluster either individually or in combination. Although the Ala mutation of Asp/Glu residues within each of the two negatively charged CYP3A4 clusters scrutinized impaired its ubiquitination by either E2-E3 system, both were not found equally relevant to each system. Thus, judging by the impairment of CYP3A4 ubiquitination upon specific Ala mutation, its Glu-283–Thr-284–Glu-285 cluster appeared more important for the UBC7-gp78-mediated modification of the native protein irrespective of its phosphorylation state (Fig. 6B). This underscored the importance of both Glu-283 and Glu-285, as well as Thr-284 to this UBC7-gp78-mediated ubiquitination of Lys residues on this CYP3A4 surface that include Lys-115 and Lys-282, the latter within this very cluster. Thus, impairment of the native CYP3A4 E283A/T284A/E285A mutant ubiquitination was slightly more pronounced than that of the native T264A/S420A/S478A mutant (Fig. 6B). The difference between these two Ala mutants was however largely abrogated upon CuOOH inactivation and PKA/PKC phosphorylation (Fig. 6B). By contrast, this Glu-283–Thr-284–Glu-285 cluster was not quite as relevant to CYP3A4 ubiquitination by the CHIP system, as CHIP-mediated ubiquitination of CYP3A4 E283A/T284A/E285A was only slightly impaired relative to that of CYP3A4WT, and this was to a large extent restored upon PKA/PKC-mediated protein phosphorylation (Fig. 7B). However, when both E2-E3 systems were concurrently present (Fig. 8B), the relevance of this particular CYP3A4 cluster to its UBC7-gp78-mediated ubiquitination was underscored by the marked impairment of CYP3A4 ubiquitination, irrespective of CuOOH inactivation and/or PKA/PKC-mediated protein phosphorylation.
However, the Glu-258–Ser-259–Glu-262–Asp-263–Thr-264 surface cluster, although not quite as critical to the UBC7-gp78-mediated ubiquitination of either native or CuOOH-inactivated CYP3A4 (Fig. 6C), was quite relevant to the ubiquitination of these CYP3A4 species by the CHIP system. This is inferred from the markedly attenuated CHIP ubiquitination of the E258A/S259A/E262A/D263A/T264A mutant relative to that of the native or CuOOH-inactivated CYP3A4WT (Fig. 7B, lanes 6 and 7 versus lanes 2 and 3 or lanes 13 and 14). Furthermore, the relevance of this cluster to CHIP-mediated CYP3A4 ubiquitination is also evident in the ubiquitination profile of the native CYP3A4 species observed in the presence of both E2-E3 systems (Fig. 8B, lanes 6 and 7 versus lanes 2 and 3). Such impairment of CYP3A4 ubiquitination is also detectable upon CuOOH-mediated inactivation (Fig. 8B, lane 17 versus lane 13) but is attenuated to some extent when the CuOOH-inactivated protein is phosphorylated (Fig. 8B, lane 18 versus lane 17). Notably, the relevance of this particular cluster to CHIP-mediated CYP3A4 ubiquitination is also underscored by the specific cross-linking interactions observed between CYP3A4Lys-257 with UbcH5a and CHIP, and CYP3A4Lys-266 with UbcH5a and Hsc70 (Tables 4 and and5),5), two residues flanking this CYP3A4 cluster.
Collectively, these findings reveal that although both of these negatively charged surface clusters are relevant to each E2-E3 system, they are not equivalently important to each individual system assessed in vitro. However, our findings of their relevance to intracellular ubiquitination in HEK-293T cells (Fig. 9C) are not only consistent with our in vitro findings with the dual E2-E3 system but also argue that this relevance would be just as pronounced in the in vivo context, wherein both E2-E3 systems operate concurrently and cooperatively.
The gp78 Lys residues cross-linked to CYP3A4 Lys residues reside in positively charged patches composed of Arg-307–Arg-308–Arg-310–Arg-311–His-312–Lys-313 N-terminal to its RING domain and Gln-584–Arg-585–Lys-586 and Arg-594–Lys-595–Arg-596 in its C-terminal domain. gp78Lys-313 was found cross-linked to CYP3A4Lys-251, a residue vicinal to the Glu-258–Ser-259–Glu-262–Asp-263–Thr-264 acidic cluster, whereas gp78Lys-586 was cross-linked to CYP3A4Lys-257, another residue flanking this very cluster. gp78Lys-595 cross-linking to CYP3A4Lys-424 is considered tentative as it could not be validated. Nevertheless, mutation of these positively charged gp78 residues to Ala (or Val) in either or both gp78 patches impaired CYP3A4 ubiquitination without impairing its basal functional interactions with UBC7 (Fig. 5E). These findings reveal that these positively charged gp78 patches are most likely relevant to electrostatic interactions with CYP3A4 acidic surface clusters.
Of the three previously identified CYP3A4 phosphorylatable residues (Thr-264, Ser-420, and Ser-478), Ser-478 was shown to be critical for the UBC7-gp78-mediated but not CHIP-mediated ubiquitination of the CuOOH-inactivated protein, as well as the Ub-dependent proteasomal degradation of the native, heterologously expressed CYP3A4WT in yeast and HEK-293T cells (37, 38). Indeed, CYP3A4S478 phosphorylation was previously shown to both accelerate/enhance the UBC7-gp78-mediated but not CHIP-mediated ubiquitination of CuOOH-inactivated CYP3A4 (37, 38). Such ubiquitination of CuOOH-inactivated CYP3A4 was further increased by the phosphorylation of its Thr-264 and Ser-420, two residues substantially phosphorylated by PKC in vitro (36,–38). Consistent with our previous findings, CYP3A4Ser-478 mutation to the phosphomimetic residue Asp greatly enhanced its in vitro UBC7-gp78-mediated ubiquitination of both the native and CuOOH-inactivated enzymes, whereas CYP3A4Ser-478 to Ala mutation indeed greatly impaired this process. This is so, even though the S478A mutant retains all its negatively charged Asp/Glu surface clusters and Lys residues for UBC7-gp78-mediated interactions. Furthermore, this marked impairment of CYP3A4 T264A/S420A/S478A mutant ubiquitination persisted upon incubation with the dual E2-E3 system (Fig. 8B), even though we have previously documented that its CHIP-mediated ubiquitination is not impaired per se (38). These findings thus reveal that in addition to these negatively charged surface clusters, the phosphorylatable Ser-478 residue is a key feature of the UBC7-gp78 ubiquitination process (Figs. 6B and and88B). This may be due to its ability to promote CYP3A4 Lys-487 and Lys-413 interactions with UBC7, and thus UBC7-gp78 ubiquitination of its vicinal Lys-168 and Lys-492 residues. Interestingly, the Ser-478 position on the diametrically opposite CYP3A4 surface is occupied by Ser-100 and/or Ser-116, and it remains to be determined whether these residues would play similar roles. Alternatively, given its interactions with Gln-484, a residue flanking 485PEKP488, an Hsp90-binding motif, it is conceivable that Ser-478 phosphorylation (or Asp mutation) disrupts CYP3A4 interactions with this pro-folding chaperone (52). This would render its structurally mobile C-terminal surface loop even more accessible for attack by UBC7-gp78, thus inextricably committing the CYP3A4 protein to ERAD. This is consistent not only with the documented interaction of Lys-487 with UBC7 (Tables 4 and and5)5) but also the UBC7-gp78-mediated Lys-492 ubiquitination (Fig. 4) (38). Whatever the precise mechanism involved, the major role of Ser-478 in UBC7-gp78-mediated CYP3A4 ubiquitination is irrefutable.
Collectively, our present findings lend credence to our previously proposed hypothesis that the negatively charged CYP3A4 Asp/Glu/Ser(P)/Thr(P) surface clusters are important for its intermolecular electrostatic interactions with corresponding positively charged patches in the E2-E3 complexes. This we propose is essential for the molecular recognition of CYP3A4 as a substrate by these E2-E3 complexes. Our findings discussed above are indeed consistent with this possibility. First, CYP3A4 Ser/Thr phosphorylation as previously documented (37, 38) and now reconfirmed (Figs. 1, and and666–8) both accelerated and enhanced the ubiquitination of the native as well as the CuOOH-inactivated protein. This is possibly due to the increased affinity of the phosphorylated CYP3A4 for these E2-E3 complexes. Unfortunately, however, our attempts to directly monitor the intermolecular interactions of gp78 with CYP3A4, its Asp/Glu Ala mutants, or CYP3A4S478D in real time through Biacore surface plasmon resonance analyses were unsuccessful due to the great propensity of gp78 to oligomerize and aggregate upon purification to homogeneity. Second, not only does every ubiquitinated CYP3A4 Lys residue (Lys-115, Lys-127, Lys-168, Lys-173, Lys-282, Lys-466, and Lys-487) reside within each of these linear or spatially associated and negatively charged surface clusters, but also every chemically “trapped” CYP3A4 Lys residue (Lys-141, Lys-209, Lys-251, Lys-257, Lys-266, and Lys-421) cross-linked by TPAL (BS3 and/or EDC) to an E2 (UBC7 and UbcH5a) or E3 (gp78 and CHIP) Lys residue, lies within or in close vicinity of one of these clusters. Third, mutation of the Asp/Glu residues in these clusters to Ala significantly impaired the ubiquitination of the native and/or CuOOH-inactivated CYP3A4 by either E2-E3 system, even though all the ubiquitinatable Lys residues were present and, in principle, available for ubiquitination. Fourth, the Lys residues of gp78 cross-linked to CYP3A4 Lys residues reside in positively charged patches, whose mutation disrupted CYP3A4 ubiquitination without impairing its functional interactions with UBC7 (Fig. 5E).
Although we have not quite as extensively characterized the CYP2E1 molecular interactions with these E2-E3 systems, we have previously reported that CYP2E1 Ser/Thr phosphorylation does indeed enhance/accelerate its ubiquitination by the gp78 and CHIP systems (30). Furthermore, inspection of its crystal structure (33, 94) also reveals that every one of its Lys residues ubiquitinated by these E2-E3 systems similarly resides within spatially associated and negatively charged Asp/Glu/Ser(P)/Thr(P) surface clusters (30, 76). It remains to be determined at present whether these are common structural features of the ubiquitination of not only all P450 proteins but also that of an expanding list of other canonical substrates of these particular E2-E3 ubiquitination complexes. The identification through Ub remnant profiling approach of Asp/Glu/Ser/Thr as predominant residues flanking ubiquitinated Lys sites in the global protein ubiquitinome considerably strengthens this possibility (81, 82). Nevertheless, to our knowledge, our findings provide the very first mechanistic example of UBC7-gp78 substrate recognition.
We sincerely thank Dr. A. M. Weissman (NCI, National Institutes of Health) for the UBC7 and gp78C expression plasmids; Dr. R. DeBose-Boyd (University of Texas SouthWestern) for the mammalian gp78 plasmid; Dr. C. Patterson (University of North Carolina) for the CHIP expression plasmid; Dr. L. Waskell (University of Michigan) for the cytochrome b5 expression plasmid; and Dr. C. B. Kasper (University of Wisconsin) for the CPR expression plasmid. We thank Prof. P. Ortiz de Montellano (University of California at San Francisco) for helpful comments in the course of this project. We acknowledge the mass spectrometric analyses provided by the Bio-Organic Biomedical Mass Spectrometry Resource at University of California at San Francisco (A. L. Burlingame, Director) supported by the Biomedical Technology Research Centers Program of the NIGMS, National Institutes of Health Grant 8P41GM103481.
*This work was supported, in whole or in part, by National Institutes of Health Grants GM44037 and DK26506 (to M. A. C.) and Grant 8P41GM103481 from NIGMS (to A. L. B.).
4Although the gp78Lys-586-CYP3A4Lys-257 is below the score threshold and contains numerous unmatched ion signals, the doubly charged c-ion series clearly identify the CYP3A4 peptide with a modification at the initial Lys. The presence of ion signals resulting from hemolytic cleavage of the benzylammonium bonds between TPAL and lysine (denoted as “P” and “P+L” ions) further identify a cross-linked product, and the gp78 peptide is identified by z2• and z3• ions as well as c2.
5Note no crystal structure for the intact gp78-C domain is currently available. The crystal structure of Ube2g2 (UBC7) complexed with the isolated G2BR domain of gp78 reveals extensive highly distributed contacts between the UBC7 “backside” and the helical G2BR region composed of residues Ser-574 to Lys-600 that include hydrogen bonds, salt bridges, and hydrophobic interactions that facilitate “frontside” interactions with gp78-RING finger domain and/or UBA1 (69, 70). The UBC7-interacting residues in this G2BR region include both Lys-586, a residue we found to be cross-linked to CYP3A4Lys-257, and Lys-600 that was cross-linked to UBC7Lys-156. These interactions are subject to considerable conformational effects that enable loading of the activated Ub-thioester (70) and could conceivably undergo further conformational alteration to accommodate CYP3A4 as a heterologous substrate, thereby enabling the cross-linking of Lys-586 to CYP3A4Lys-257.
6Such direct CHIP and UbcH5a interactions with CYP3A4Lys-257 and those of Hsc70 and UbcH5a with Lys-266 are entirely consistent with the Lys-257–Glu-258–Ser-259–Glu-262–Asp-263–Thr-264–Lys-266 surface cluster being important for its CYP3A4 ubiquitination. Similarly CHIP and UbcH5a interactions with CYP3A4Lys-421 may promote their ubiquitination of CYP3A4 Lys-168, Lys-466, and Lys-492 just around on the opposite surface.
7Val mutation was intended to preserve the hydrophobicity of this UBC7-interacting G2BR patch.
8Although we also examined the specific roles of various CHIP domains in CYP3A4 ubiquitination through deletion constructs, our studies confirmed that its catalytic U-box and the tetratricopeptide domain required for Hsc70 interactions were required for CYP3A4 ubiquitination.
9We have found through our LC-MS/MS analyses that in the absence of a heterologous substrate, the CHIP system ubiquitinates human Hsc70 at Lys-108, Lys-500, Lys-507, Lys-512, Lys-524, Lys-557, Lys-561, and Lys-597. CHIP is also known to ubiquitinate human Hsp70 on its Lys-325, Lys-451, Lys-524, Lys-526, Lys-559, and Lys-561 residues (95).
3The abbreviations used are: