|Home | About | Journals | Submit | Contact Us | Français|
Products of lipid peroxidation are generated in a wide range of pathologies characterized by those associated with oxidative stress and inflammation. Many oxidized lipids contain reactive functional groups that can modify proteins, change their structure and function, and affect cell signaling. However, intracellular localization and protein adducts of reactive lipids have been difficult to detect and rely largely on antibodies raised against specific lipid-protein adducts. As an alternative approach for monitoring oxidized lipids in cultured cells, we have tagged the lipid peroxidation substrate arachidonic acid, and an electrophilic lipid, 15-deoxy- Δ12,14-prostaglandin-J2 (15d-PGJ2) with either biotin or the fluorophore BODIPY. Tagged arachidonic acid can be used in combination with conditions of oxidant stress or inflammation to assess the subcellular localization and protein modification by oxidized lipids generated in situ. Furthermore, we show that reactive lipid oxidation products such as 15d-PGJ2 can also be labeled and used in fluorescent and Western blotting applications. This article describes the synthesis, purification, and selected application of these tagged lipids in vitro.
Polyunsaturated fatty acids (PUFAs), such as arachidonic acid, are primary components of the diet, biological membranes, and lipoproteins and are substrates for lipid peroxidation (LPO) [1–3]. They readily undergo non-enzymatic peroxidation during oxidative stress or are oxidized through reactions involving enzymes such as cyclooxygenase (COX), cytochrome P450, or the lipoxygenases [4–8]. A role for oxidized lipids in pathophysiology is supported by the increased levels of LPO apparent in several diseases including Alzheimer’s disease [9, 10], renal failure [11, 12], and atherosclerosis [13, 14].
Many LPO products contain functional groups capable of covalently modifying proteins [15–18]. For example, both non-enzymatic and enzymatic oxidation of PUFAs result in the formation of reactive lipid species that have electrophilic carbons, usually in the β-position of an α,β-unsaturated carbonyl (Figure 1), which are reactive with protein nucleophiles. Electrophilic products of non-enzymatic lipid oxidation include aldehydes such as 4-hydroxynonenal (HNE), malondialdehyde, and acrolein, and the J-and A-series isoprostanes (Figure 1). Other products include the isoketals, which occur through non-enzymatic LPO by rearrangement of endoperoxide intermediates of the isoprostane pathway . Isoketals are highly electrophilic γ-ketoaldehydes that rapidly and covalently modify lysine residues, forming a variety of adducts including lactam, hydroxylactam, imine, and pyrrole species. The enzymatic oxidation of arachidonic acid also results in the formation of reactive LPO products. However, in this case the variety and reactivity of these species is constrained by the active site of enzymes such as lipoxygenase which limits the number of stereoisomers generated during the enzymatic reaction.
Studying the complex mixture of reactive lipids formed in vivo during oxidative stress or inflammation has been limited by existing techniques, which have focused primarily on following the reactions of a defined lipid in conjunction with a candidate target protein approach. The covalent modification of proteins by oxidized lipids is an important mechanism by which LPO may initiate cell signaling as well as contribute to tissue injury [15, 20–22]. Although many studies have shown that the formation of oxidized lipids are increased in disease, it is unclear whether they represent the footprints of oxidative stress or are mediators of pathology. A major obstacle to a complete understanding of the biological roles of oxidized lipids is the unambiguous detection and quantification of biologically relevant protein modifications. Hence, improved technologies that allow a more comprehensive definition of the reactive lipid proteome are needed to determine the mechanism, the extent, and under what circumstances oxidized lipids affect cell signaling and physiology.
Due to the immunogenic potential of some lipid peroxidation products such as isoketals, levuglandins, and aldehydes, antibodies against specific protein-lipid adducts have been raised that allow for their detection by immunological methods [23–27]. This approach has been used to identify specific lipid modifications using proteomics and mass spectrometry techniques [28–30]. However, many antibodies to protein-lipid adducts are susceptible to epitope bias thus protein modifications detected using antibodies may represent a small subset of the reactive lipid proteome. Another disadvantage of this approach is that since complex lipid mixtures are typically generated in cells and tissues during oxidative stress, an antibody to a specific lipid protein adduct will necessarily give only a partial picture of the reactive lipid proteome.
Detection of the carbonyl moiety present on proteins conjugated to α,β-unsaturated aldehydes and ketones has been a useful marker of lipid peroxidation and oxidative stress [31, 32]. Aldehydes and ketones can react with protein nucleophiles such as histidine, cysteine, and lysine by a Michael addition reaction to form stable adducts . This type of adduct can be detected by derivatizing the carbonyl group using hydrazine or hydrazide chemistry to form a stable hydrazone product [31, 33, 34]. In particular, the use of avidin or streptavidin detection techniques in conjunction with biotin hydrazide increases the sensitivity for detecting low abundance proteins. One important caveat, however, is that protein carbonyls can be introduced by reaction with oxidants other than reactive lipids, decreasing the specificity of this approach for the unique detection of the oxidized lipid proteomes.
Other methods of detection, such as the radioactive labeling of precursor lipids, have been used to measure protein-lipid adducts [35–37]. Nevertheless, the intracellular targets for reactive lipid oxidation products remain largely undefined due to a lack of suitable detection reagents and protocols. Herein, we describe a non-radioactive method for the labeling, purification, and utilization of substrates for LPO (i.e., arachidonic acid) and of a specific oxidized lipid, 15-deoxy-prostaglandin J2 (15d-PGJ2). In previous studies, we and others have shown that conjugation of 15d-PGJ2 to biotin or fluorescent tags can be used to monitor both the subcellular localization of the lipid and to identify specific protein targets [16, 17, 38–42]. Using this approach, signaling proteins such as Keap-1 were shown to form covalent adducts with 15d-PGJ2, which mediated the induction of antioxidant defenses including glutathione and heme oxygenase-1 [16, 17, 38]. The insertion of the tags on the lipid via the carboxyl group has little or no impact on the potency of the lipid electrophiles to induce either antioxidant defenses at low concentrations or apoptosis at higher levels. The BODIPY analog of 15d-PGJ2 has been particularly interesting, revealing mitochondrial targeting of the lipid, and biotin-tagged 15d-PGJ2 has been used to identify the proteome modified by reactive prostaglandins in mitochondria and cells [41, 42]. This paper describes the methods for detecting protein-lipid adducts using fluorescent and Western blotting approaches.
To follow the formation of lipid adducts with proteins, we describe a protocol to conjugate biotin or BODIPY to the carboxylic acid group of the unsaturated lipid arachidonic acid or the lipid electrophile. This method was adapted from two previously published methods which describe the biotinylation of prostaglandin A2  and 15d-PGJ2 . Tagging arachidonic acid offers the important advantage of detecting proteins that are reactive with multiple LPO species. As shown in Figure 2, tagged arachidonic acid can be oxidized to lipid peroxidation products in the presence of catalytic metals, oxidants, or in response to an inflammatory stimulus. Many of these tagged products contain functional groups such as reactive aldehydes, ketones, and electrophilic carbons capable of modifying protein nucleophiles. Hence, this technique enables the detection of protein-reactive lipid products containing the BODIPY or biotin tag. Incorporation of the fluorescent BODIPY tag also allows the detection of protein modification by oxidized lipids using in-gel fluorescence, thereby eliminating the need for protein transfer and allowing a more high-throughput proteomics approach.
The structures for unlabeled and biotinylated arachidonic acid are shown in Fig. 3A. The biotin moiety is covalently attached to the fatty acid at the carboxyl group using the activating compound 1-ethyl-3-(3-dimethyl) aminopropylcarbodiimide (EDC) causing it to be reactive with the amino group of 5-(Biotinamido) pentylamine. This condensation reaction results in the formation of a stable amide bond between the fatty acid and the biotin tag. To minimize oxidation, exposure to room air and light must be avoided and all solvents should be purged with an inert gas such as argon or nitrogen prior to use. In addition, new glassware that has been pre-rinsed with ethanol and dried should be used to prevent contamination and loss of lipid. The reaction should be performed in an amber or foil-covered glass vial to further limit light-dependent oxidation of the lipid.
This synthesis of BODIPY-labeled arachidonic acid is also performed using a carbodiimide-mediated conjugation reaction. The structure of BODIPY-arachidonic acid is shown in Fig. 3A. The reaction should be performed in a new, pre-rinsed amber or foil-covered glass vial to prevent light-dependent oxidation of the unsaturated lipid.
The electrophilic prostaglandin, 15-deoxy-prostaglandin-J2, can also be conjugated to biotin; the structure of biotinylated 15-deoxy-prostaglandin-J2 is shown in Fig. 3B. The protocol for this conjugation reaction is as follows:
The protocol used for this conjugation reaction is as follows:
After synthesis, the reaction preparations have a mixture of the original unreacted lipid, the tag, the expected product, and the priming compound, EDC. High performance liquid chromatography (HPLC) with UV-Vis detectors is needed to purify the tagged lipid. Following chromatographic separation, product purity is verified using electrospray ionization mass spectrometry (ESI-MS).
The mobile phases used for HPLC separations are prepared as follows:
A summary of all HPLC parameters used for each lipid is given in Table 2. For all purifications, the column is equilibrated in the initial mobile phase which varies depending on which tagged lipid is being separated (Table 1). The mass of each lipid, typical retention times, and optimum ionization mode for mass spectrometry are given in Table 1.
For purification of tagged arachidonic acid, reversed-phase HPLC should be performed on a C18 semi-preparative column as indicated in the Instruments section. Typically, an aliquot of the reaction mix containing approximately 1 mg of starting material is eluted at a flow rate of 4.5 ml/min using a linear gradient of 50–95% B after equilibration in 50% solvent A/50% solvent B (Table 2, Gradient I). Fractions containing the peak of interest are collected and verified using electrospray ionization mass spectrometry (ESI-MS).
Figure 4A represents an example separation of the Bt-AA reaction mixture by HPLC, with the product typically eluting at a retention time of 13 min (peak b), arachidonic acid eluting late (27 min, peak c), and oxidized products of arachidonic acid eluting early (peak a). To verify separation of Bt-AA, a wavelength scan from 200–300 nm (Figure 4B) should be performed using a spectrophotometer. Figure 4C is a characteristic spectrum obtained from ESI-MS in the positive ionization mode with the major peak at 616 (M+H)+, the m/z value for Bt-AA.
BD-AA can also be separated to high purity by HPLC (Figure 5A) using the same HPLC gradient profile as Bt-AA. As with Bt-AA, oxidized products containing the tag elute before the product (peak a). BD-AA elutes at approximately 29 min, and purity can be verified by UV/Vis spectrophotometry (Figure 5B) at 200–600 nm. The maximum absorbance is observed at 504 nm (for BODIPY) with the second largest peak at 205 nm (for arachidonic acid). Other components in the spectra are contributions of the BODIPY tag. Product purity must be verified using ESI-MS in the negative ionization mode as shown in Figure 5C. The m/z value for BD-AA is 619 (M+H)+.
For purification of biotinylated and BODIPY-tagged 15d-PGJ2, HPLC separation should be carried out on a C18 preparative column as indicated in the Instruments section. Approximately 2 mg of starting material can be separated using Gradient II, at a flow rate of 20 ml/min. The absorbance of the eluant should be measured at 306 nm (λmax for 15d-PGJ2). Fractions containing the peak of interest are collected and verified using ESI-MS.
BD-15d-PGJ2 is purified using HPLC Gradient profile II, which begins at a concentration of 100% A/0% B and after 13 min increases over a linear gradient to 5% A/95% B. BD-15d-PGJ2 elutes during the linear gradient as Peak c with a retention time of 26 min (Figure 6A). A characteristic UV/Vis wavelength scan from 200–600 nm is shown in Figure 6B, with major peaks observed at 504 nm (BODIPY) and 306 nm (15d-PGJ2). Purity of BD-15d-PGJ2 should be verified using ESI-MS in the negative ionization mode (Figure 6C). BD-15d-PGJ2 is apparent at 632 (M-H)−. A formic acid adduct formed during the mass spectrometry can also be seen in the spectra (m/z = 678).
Following HPLC purification, the tagged lipids are in a solution containing acetic acid, acetonitrile, and water. To eliminate water and exchange solvents, it is necessary to extract the product using the following protocol:
Once synthesized, the tagged lipids are purified and fully characterized by mass spectrometry as shown for Bt-AA in Figure 4C. Typically, derivatives of arachidonic acid are diluted to 1 μM in ethanol and injected in an infusion mobile phase of 50% acetonitrile containing 10 mM ammonium acetate for ESI-MS. Tagged 15d-PGJ2 can be diluted to 1 μM in ethanol and injected in an infusion mobile phase of 50% acetonitrile containing 0.1% formic acid. The mass for each lipid and optimal ionization mode for each is reported in Table 1.
Since tagged lipids are typically used for in vivo or in vitro detection of lipid products, lipid derivatives should be resuspended and stored at −80°C in a vehicle suitable for cell culture, such as ethanol or DMSO. As with any new experimental treatment, it must be verified that the vehicle used does not change the biological outcome observed in your particular model.
Due to evaporation of the storage solvent, the concentration of tagged lipids may change over time and should be measured prior to each experiment using a spectrophotometer. For this reason, it is also not recommended to make small aliquots of the stock, since they evaporate more quickly. The absorbance and extinction coefficient for calculating the concentration of each lipid in ethanol is given in Table 1. In addition, a wavelength scan of arachidonic acid derivatives can be used to detect oxidation of the lipid which will be apparent through the presence of a peak at 233 nm due to conjugated dienes [45, 46].
The BODIPY-FL EDA moiety allows for visualization of lipid products in situ. This can be combined with tracker dyes to determine the subcellular localization of the lipids during experimentation. To detect BODIPY-tagged lipids using microscopy, the following protocol is used:
The detection of covalent protein adducts by oxidized lipids using tagged lipids is facilitated by the biotin or BODIPY moiety present on the lipid. Both types of tagged lipids offer unique advantages and should be considered depending on the research goal, available laboratory equipment, and the experimental conditions. Biotinylated lipids allow for sensitive detection as well an accurate quantification of protein modification using Western blot analysis . BODIPY-labeled lipids can also be detected using Western blot analysis; however, an advantage of using fluorescent tags is that the gel can be imaged immediately after electrophoresis using “in-gel” imaging, eliminating the need for transfer to a membrane. For both biotin as well as BODIPY-tagged lipids, high resolution digital imaging techniques are advantageous In our experience, the use of film is rarely optimal, primarily due to the fact that, even at low exposures the film easily saturates. The result is that quantification obtained from a film image is often nonlinear and may underestimate experimental changes. The dynamic range of digital cameras and fluorescent imagers (e.g., Typhoon imagers) are far superior and are therefore recommended for Western blotting applications using biotin tags and/or in-gel fluorescent imaging. Protocols specific for each tagged technology (biotin or BODIPY) are detailed below:
One advantage of using biotinylated lipids is that the amount of lipid-protein adducts may also be calculated using a standard, such as biotinylated cytochrome c (bt-cyt c). It is recommended to include such a standard on the same gel to quantitate the biotin content in experimental samples . Since each mole of lipid is tagged with one mole of biotin, the quantification of biotin on a Streptavidin blot serves as a quantitative marker for the lipid adduct, therefore allowing the calculation of the mol lipid/mol protein in proteomics format. The following protocol is used to detect proteins modified by biotinylated lipids:
The following protocol is used to detect proteins modified by BODIPY-tagged lipids by Western blotting. Alternatively, if access to a fluorescent imager is available, in-gel fluorescence imaging may be used. Refer to [(III) Visualizing BODIPY-tagged adducts using in-gel fluorescence].
Since Bromophenol blue is a fluorescence quenching agent [48, 49], a dye-free SDS sample loading buffer is recommended if using fluorescence imaging. Our laboratory uses a 5× sample buffer containing 250 mM Tris HCl (pH 6.8), 25% glycerol, and 10% sodium dodecyl sulfate. If another loading buffer is used, it should be adapted so that components that interfere with BODIPY fluorescence are removed or substituted with a similar but non-interfering reagent. The Typhoon imaging system or another analogous system may be used for imaging BODIPY-labeled proteins. In addition, we recommend staining the gels with Coomassie blue or another stain such as Deep Purple (GE Healthcare) after acquiring the BODIPY image to verify even protein loading. This allows for simultaneous or post-visualization of the total protein amount as well as the modifications.
As shown in Figure 7, BD-15d-PGJ2 and BD-AA are capable of entering bovine aortic endothelial cells (BAEC) after 2 h. In this experiment, BD-15d-PGJ2 fluorescence showed a reticular pattern, consistent with previous studies demonstrating its ability to concentrate in mitochondria [41, 42]. After entering cells, the fluorescent tag of BD-AA retains its excitation/emission properties and does not appear to be oxidized due to hemin treatment.
Both the BODIPY and biotin moieties allow the detection of protein modification by endogenously produced lipid peroxidation products in lysates prepared from in vitro and in vivo samples. Hemin has been shown to increase lipid peroxidation resulting in the formation of protein reactive lipid species; however, the formation of protein-lipid adducts has not been previously reported. As shown in Figure 8A, Bt-AA can be used to detect protein modifications by lipid peroxidation products in cells. In this experiment, RAW 264.7 macrophages were treated with the ethanol vehicle (control), Bt-AA, or hemin and Bt-AA. In control lysates, a small number of proteins can be detected on the Strep-HRP blot corresponding to endogenous biotin-containing carboxylases. Bt-AA only slightly increases the reactivity with Strep-HRP, suggesting that arachidonic acid itself is not reactive with proteins, as expected. However, addition of hemin, which catalyzes lipid peroxidation, results in an increase in Strep-HRP signal. BD-AA can be used in a similar manner and also facilitates the detection of protein reactive lipid species. In addition, other controls, such as the addition of the chain-breaking antioxidants butylated hydroxytoluene (BHT) and α-tocopherol, can be used to decrease LPO and protein adduct formation (data not shown).
The electrophilic lipid, 15d-PGJ2, has been shown to modify proteins in several experimental models [17, 20, 50–54] and is capable of modifying the mitochondrial proteome . The mitochondrial proteome modified by BD-15d-PGJ2 can be investigated using proteomic techniques. Figure 8C shows the results from such an experiment. Briefly, isolated cardiac mitochondria were treated with BD-15d-PGJ2, and the proteins were separated by 2D Blue Native PAGE. This method uses the protein-binding and charge-conferring characteristics of Coomassie dye to facilitate protein separation in the 1st dimension [55–57]. After separation in the 2nd dimension, the proteins remain bound to much of the Coomassie dye. Hence, both the BODIPY and Coomassie fluorescence can be detected using the guidelines in (III) Visualizing BODIPY-tagged adducts using in-gel fluorescence. As shown in Figure 8C, equal protein loading was verified on 2D Blue Native PAGE gels of isolated heart mitochondria using the intrinsic Coomassie fluorescence. BODIPY fluorescence could be detected with a Typhoon imager in samples treated with BD-15d-PGJ2. No interfering fluorescence of proteins was detected at this wavelength in control mitochondria. The BN-PAGE gels shown in Figure 8C demonstrate the utility of using tagged electrophilic lipids to detect reactive subproteomes. Similar experiments with these tagged lipids can be performed using other 1D and 2D gel-based separation techniques.
Protein modification by biotin tagged lipids can be quantified using standards such as bt-cyt c. The development of this method is described in . Briefly, addition of bt-cyt c allows for the absolute determination of the moles of biotin per milligram of protein in a biological sample. Due to the addition of the biotin tag, this increase in biotin is assumed to be due to protein modifications by the biotin-containing oxidized lipid. Therefore, each biotin molecule is equivalent to one lipid covalently adducted to a protein. Since covalent modification of proteins is an important regulatory mechanism in cell signaling, reliable measurement of these modifications is advantageous for these studies [20, 40, 52, 53]. As shown in Figure 8A, proteins modified by oxidized products of Bt-AA were detected on Strep-HRP western blot. Quantification of adducts (nmol biotin/mg protein) was facilitated using the internal standard, bt-cyt c and is shown in Figure 8B.
Tagged derivatives should be compared with the untagged version of the same lipid for all new experimental models to ensure that the biological effects are not changed by the tag. For example, induction of antioxidant defenses, and the cytoprotection elicited by the electrophilic lipid, 15d-PGJ2, are not significantly different for the tagged derivatives of this electrophile . Importantly, the covalent addition of BODIPY and biotin occurs at the carboxylic acid distant from the electrophilic carbons on the molecule, and will not affect the electrophilicity of the lipid. However, modification of the carboxyl group may change other properties of the lipid . For example, the biotinylated derivative of 15d-PGJ2 is not an effective ligand for PPARγ.
While the use of tagged unsaturated fatty acids allows the identification of lipid protein adducts derived from lipid peroxidation there are a number of important limitations. For example, modifications of the fatty acid other than oxidation which lead to covalent modification of proteins could also result in increased protein adduct formation. As a control for this the attenuation of modification by chain breaking antioxidants, which should inhibit lipid peroxidation but not other modifications of the fatty acid such as acetylation can be used. In interpreting the accumulation of lipid-protein adducts a number of mechanisms should be considered. For example, it is known that electrophilic lipids can inhibit the proteasome and thus the turnover of adducted proteins may be inhibited [30, 51].
Use of the tagged lipids described herein enable the discovery and identification of protein targets of reactive LPO products. A limitation of this approach is the products of LPO in which the tag is lost or oxidized will not be detected. For example, 4-hydroxy-2-nonenal generated from the oxidation of the PUFA, will not contain a tag, which is attached to the carboxyl group. In addition, it should be noted that using the tagged substrate arachidonic acid and the oxidation systems described (i.e. hemin) will not enable the investigator to know the specific oxidized lipid species that modified a particular protein without coupling this technology to mass spectrometry analysis. However, use of the tagged substrate does allow a more comprehensive and thorough approach to identifying the proteins modified by LPO in response to oxidative stress or inflammation.
The authors would like to thank Ray Moore for mass spectrometric data acquired at the Comprehensive Cancer Center Mass Spectrometry Shared Facility, University of Alabama at Birmingham.
This research was supported by NIH grants ES10167, DK 75865, DK 079337 (VDU), T32 HL007457 (BGH) T32 HL007918 (BPD), and American Heart Association funding through the Scientist Development Grant 0635361N (AL), and Predoctoral Fellowship 0815177E (ANH).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.