Identification of endogenous substrates of orphan cytochrome P450 enzymes through the use of untargeted metabolomics approaches

Summary

Metabolomics provides an invaluable means to interrogate the function of “orphan” enzymes, i.e., those whose endogenous substrates are not known. Here we describe a high performance liquid chromatography-coupled mass spectrometry (HPLC-MS)-based metabolomics approach to identify an endogenous substrate of an orphan cytochrome P450.

1. Introduction

With the progress of recent genome-sequencing efforts, more than 15,000 cytochromes P450 (P450s) have been registered in the database (http://drnelson.uthsc.edu/CytochromeP450.html). Most of these P450s are considered “orphans” because no substrates or functions are associated with these enzymes. A need exists to explore these orphan P450s in order to bridge the wide knowledge gap between amino-acid sequence information and biochemical function.

Metabolomics provides an efficient approach to identify the substrates of these orphan P450s (1–3). Here we describe an untargeted metabolomics approach to identify substrates of orphan P450s from Streptomyces coelicolor. Briefly, an organic extract of the organism (or tissue) where the orphan P450 is normally found was prepared as a chemical library that may contain endogenous substrates. The extract was then incubated with the recombinant P450 in the presence of NADPH and redox partners, with subsequent metabolite profiling by LC-MS. The metabolic profile obtained after the reaction was compared with that before the reaction using specialized metabolomics software, e.g., XCMS (4, 5). Molecules that were depleted due to the enzymatic reaction were identified as potential substrates. A successful example of this approach is the recent study of P450 154A1 (6). This method can also be applied in the study of mammalian orphan P450s.

2. Materials

2.1 P450 assay

1. Recombinant orphan P450 (see ^{Note 1}).

2. Redox partners of P450: 200 μM spinach ferredoxin and 20 μM spinach NADPH-ferredoxin reductase. These enzymes can be purchased from Sigma-Aldrich. Rat NADPH-P450 reductase (NPR) can be prepared as described previously (7).

3. 10 mM NADP⁺ stock solution: 382 mg NADP⁺ in 50 ml of Milli-Q water. Store at 4 °C.

4. 100 mM glucose 6-phosphate: 3.4 g in 100 ml of Milli-Q water. Store at 4°C.

5. yeast glucose 6-phosphate dehydrogenase (10³ IU/ml): 1.0 mg in 1 ml of 10 mM Tris-acetate buffer (pH 7.4), containing 20% (v/v) glycerol and 1 mM EDTA. Store at 4 °C.

6. Extraction solvent: ethyl acetate with 0.1% (v/v) acetic acid (see ^{Note 2}).

7. Tissue extract: extract tissue with extraction solvent by using a ratio of 200 ml/g for dry tissue or 10/1 (vol/vol) for tissue culture. For preparing mammalian tissue extracts, homogenization before extraction is needed. The organic portion is filtered and dried in a rotary evaporator. Dissolve the extract in minimal acetonitrile so that no obvious pellet is observed. Store the extract at −20 °C.

8. DLPC solution: 5 mg L-α-dilauroyl-sn-glycero-3-phosphocholine in 5 ml of Milli-Q water. Sonicate (with probe, 2 × 10 s, full power). Store at 4°C. Sonicate each day before use and hold at 23 °C until use.

9. 100 mM potassium phosphate buffer, pH 7.4.

10. NADPH generating system: combine 50 parts of 10 mM NADP⁺, 100 parts of 100 mM glucose 6-phosphate, and 2 parts of yeast glucose 6-phosphate dehydrogenase (1 mg/ml). Prepare fresh daily and store on ice when not in use.

2.2 Metabolic profiling by HPLC-MS

1. HPLC: Waters Acquity UPLC system with photodiode array (PDA) detector (see ^{Note 3}).

2. Column: Acquity UPLC BEH (C₁₈) column (2 mm × 100 mm, 1.7 μm, Waters) (see ^{Note 4}).

3. Mobile phase: solvent A: 95% H₂O, 5% CH₃CN, 0.1% HCO₂H (v/v/v); solvent B: 5% H₂O, 95% CH₃CN, 0.1% HCO₂H (v/v/v).

4. Mass spectrometer: LTQ (Thermo). An ion-trap instrument is preferred in this protocol due to the broader scan range compared with quadrapole-based mass spectrometers. A q-TOF instrument (Waters) is used for high-resolution mass spectrometry (HRMS).

2.3 Processing HPLC-MS dataset using XCMS

Dell Dimension XPS-Gen 5 with Intel Pentium Quad Processor, 4 GB memory, 150 GB hard drive with R language, and XCMS installed (see ^{Note 5}).

3. Methods

3.1 Orphan P450 enzyme assay

Skip step 1 if working with bacterial P450s; mix bacterial P450 with ferredoxin and NADPH-ferredoxin reductase in potassium phosphate buffer (pH 7.4) so that the reaction mix (before addition of the NADPH generating system) contains 1 μM P450, 2 μM NADPH-ferredoxin reductase and 10 μM ferredoxin. All reactions should be performed at least in duplicate.

1. Reconstitute the mammalian P450 system: sonicate the DLPC solution for 2 × 10-second bursts or until the solution is clear. Add reagents in following sequence: recombinant P450, NPR and DLPC. Gently mix the reaction components for 20–30 s and then leave the mix at room temperature for 10 min. Add an appropriate amount of 100 mM potassium phosphate buffer. A typical reaction has a volume of 1.0 ml and contains 0.1 μM P450, 0.5–1 μM NPR and 20 μg/ml DLPC. Omit the P450 from the negative controls.

2. Add an appropriate amount of tissue extract so that the total organic content is ≤1% (vol/vol). Initiate the reaction by the addition of 100 μl of the NADPH-generating system (10% of reaction volume). Incubate the reaction at 37 °C for 1 h.

3. Extract the reaction mix 3× with 5 ml of ethyl acetate, combine all organic phases, and evaporate to dryness under a nitrogen stream. Dissolve the extract in 50 μl of methanol.

3.2 Metabolic profiling by HPLC-MS

Tune and calibrate the mass spectrometer according to the manufacturer’s instructions before the experiment. Positive electrospray (+ESI) mode was used in this protocol. Other ionization methods, e.g., atmospheric chemical ionization (APCI), may be employed if the orphan P450s are suspected to metabolize compounds that do not ionize well in the ESI mode, e.g., steroids.

1. HPLC gradient: flow rate 0.35 ml/min, 0–1 min, 15% B (v/v); 1–9 min, 15%–95% (v/v) B; 9–12 min, 95% (v/v) B: 12–13 min, 95%-15% B (v/v); 13–15 min, 15% B (v/v). The column temperature is set at 40 °C.

2. Each sample (10 μl) of is injected using the partial-loop mode. After each injection, it is recommended to run a blank sample (100% methanol) to ensure there is no carry over of the previous sample.

3.3 Processing HPLC-MS dataset using XCMS

1. Convert the raw LCMS data to CDF or mzXML for processing by XCMS. The manufacturer normally provides a file-convertor function in their instrument-operation software. For example, Xcalibur (Thermo) has a “File convertor” option under the menu of “Tools,” which can convert .RAW files to .CDF files.

2. Build two file folders under the same directory and name these folders as “Assay” and “NegativeControl” respectively. Move the converted assay data to the Assay folder and the control data to the NegativeControl folder.

3. Start R, load an XCMS library, and run a standard XCMS analysis (see ^{Note 6}). Set the output number n to choose how many statistically significant differences should be reported. A good starting point is about 50.

4. Verify the reported metabolic difference by checking the original LCMS data under the extracted-ion mode. Make sure that the difference is significant and consistent among replicates (see ^{Note 7}). The metabolites that are consumed in the reaction are considered putative endogenous substrates and worthy of further follow-up. For example, a putative substrate was identified in our study of P450 107U1 (Figure 1a).

Figure 1.

3.4 Verification of the identified substrates

Partially purify the substrate candidates. This can be accomplished by collecting the HPLC elute or using solid-phase extraction. Repeat the assay using partially purified candidates to confirm the substrate will be depleted in an enzyme-dependent manner. With fewer interfering compounds present in the partially purified substrate, it is not rare to see the formation of product. In our studies of P450 107U1, a product peak derived from the desaturation of the putative substrate was very obvious (−2 amu, Figure 1b), lending further support that a bona fide substrate was identified.

3.5 Structural elucidation of identified substrate

After a substrate is identified, considerable effort may be required to elucidate the structure of the compound. Detailed instructions for structure elucidation for all compounds are beyond the scope of this protocol. Here we describe a simplified and basic systematic approach for structural characterization.

1. Obtain the HR-MS of substrate (error < 5 ppm, using a qTOF, Orbitrap, or magnetic sector instrument) and deduce the molecular formula. Sometimes multiple formulae may be possible. The use of MS fragmentation and other information, e.g., UV absorbance, can often help with selecting the correct molecular formula.

2. Search chemical databases using the derived formula (e.g., chemspider: http://www.chemspider.com/, NCBI pubchem: http://pubchem.ncbi.nlm.nih.gov/, NIST Chemistry Webbook: http://webbook.nist.gov/chemistry/, and Scifinder: https://scifinder.cas.org/scifinder); determine whether the substrate formula matches any known compound. After the integration of all known information, if any known compound appears to be the unidentified substrate then obtain the standard compound (if available commercially) and compare its HPLC elution time and MS fragmentation pattern with those of the unidentified substrate.

3. If the substrate appears to be a novel chemical entity, collect enough material and identify the structure by crystallography or (more realistically) NMR. A minimum amount for 1-dimensional ¹H-NMR work is ~ 1 μg, but an order of magnitude more is required for 2-dimensional ¹H studies and even more is needed if natural abundance ¹³C-NMR is required.