Photocrosslinking and Click Chemistry Enable the Specific Detection of Proteins Interacting with Phospholipids at the Membrane Interface

Article abstract of DOI:10.1016/j.chembiol.2008.11.009

New lipid analogs mimicking the abundant membrane phospholipid phosphatidylcholine were developed to photocrosslink proteins interacting with phospholipid headgroups at the membrane interface. In addition to either a phenylazide or benzophenone photoactiv

Full text of DOI:10.1016/j.chembiol.2008.11.009

Chemistry & Biology
Article
Photocrosslinking and Click Chemistry Enable
the Specific Detection of Proteins Interacting
with Phospholipids at the Membrane Interface
Jacob Gubbens,1,4 Eelco Ruijter,2,5 Laurence E.V. de Fays,2,6 J. Mirjam A. Damen,3 Ben de Kruijff,1 Monique Slijper,3
Dirk T.S. Rijkers,2 Rob M.J. Liskamp,2 and Anton I.P.M. de Kroon1,7,
*
1Chemical Biology and Organic Chemistry, Bijvoet Center for Biomolecular Research and Institute of Biomembranes, Utrecht University,
Padualaan 8, 3584 CH Utrecht, The Netherlands
2Medicinal Chemistry and Chemical Biology, Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Sorbonnelaan 16, 3584 CA
Utrecht, The Netherlands
3Biomolecular Mass Spectrometry and Proteomics Group, Bijvoet Center for Biomolecular Research and Utrecht Institute for Pharmaceutical
Sciences, Utrecht University, Sorbonnelaan 16, 3584 CA Utrecht, The Netherlands
4Present address: Department of Biochemistry and Molecular Biology, Oregon Health & Science University, Portland, OR 97239, USA
5Present address: Department of Organic and Inorganic Chemistry, VU University, Amsterdam, The Netherlands
6Present address: Federal Agency for Drugs and Health Products, Bruxelles, Belgium
7Present address: Membrane Enzymology, Bijvoet Center for Biomolecular Research and Institute of Biomembranes, Utrecht University,
Utrecht, The Netherlands
*Correspondence: a.i.p.m.dekroon@uu.nl
DOI 10.1016/j.chembiol.2008.11.009
SUMMARY
low expression levels and unfavorable properties such as poor
solubility (Blonder et al., 2004; Santoni et al., 2000). Therefore,
Newlipidanalogsmimickingtheabundantmembrane new methodology is required for analysis of this class of proteins
by proteomics techniques.
One approach to detect peripheral and integral membrane
proteins takes advantage of their interaction with lipids, and
phospholipid phosphatidylcholine were developed to
photocrosslink proteins interacting with phospho-
lipid headgroups at the membrane interface. In
addition to either a phenylazide or benzophenone
photoactivatable moiety attached to the headgroup,
the lipid analogs contained azides attached as baits
to the acyl chains. After photocrosslinking in situ in
the biomembrane, these baits were used for the
involves the use of photoactivatable lipid analogs to perform
crosslinking in situ in the biomembrane (Brunner, 1993). Upon
ultraviolet irradiation (UV) activation, covalent bonds are formed
with adjacent molecules, thereby labeling membrane proteins
of interest. This method has been used successfully in combina-
tion with radiolabeling to detect membrane proteins interacting
with for instance 3-trifluoromethyl-3-[125I]phenyldiazirine-phos-
phatidylcholine (TID-PC) in vitro (Janssen et al., 2002), and
attachment of a fluorescent tetramethylrhodamine-
alkyne conjugate or a biotin-alkyne conjugate using
click chemistry, allowing for the selective detection [3H]cholesterol, [3H]phosphatidylcholine, and [3H]phosphatidyli-
nositol, all containing a photoreactive diazirine moiety, in vivo
and purification of crosslink products, respectively.
(Thiele et al., 2000). Although in these studies the photoactivat-
able moiety was located in the hydrophobic and relatively inert
interior of the biomembrane, other studies have reported
Proteins crosslinked to the lipid analogs in inner
mitochondrial membranes from Saccharomyces
cerevisiae were detected and subsequently identified
successful crosslinking to peripheral membrane proteins using
by mass spectrometry. Established interaction
the short chain N-([125I]iodo-4-azidosalicylamidyl)-1,2-dilauroyl-
partners of phosphatidylcholine were found, as well
as known integral and peripheral inner membrane
proteins, and proteins that were not previously
sn-glycero-3-phosphoethanolamine [125I]-ASA-DLPE (Berman
et al., 1994; Desneves et al., 1996; Gao and Bauerlein, 1987;
Montecucco, 1988; Montecucco et al., 1988) in which the photo-
considered mitochondrial inner membrane proteins.
crosslinking moiety is attached to the phosphoethanolamine
headgroup via an amide bond.
INTRODUCTION
Recently, the nonradioactive analog of this probe, N-(4-azido-
salicylamidyl)-1,2-dilauroyl-sn-glycero-3-phosphoethanolamine
Membrane proteins play a crucial role in many cellular processes, (ASA-DLPE, 1), was used in a proteomics approach aimed at
including cell signaling and membrane trafficking. Up to 30% analyzing the proteome interacting with the lipid headgroups
of all open reading frames are predicted to encode integral (Gubbens et al., 2007). The probe was incorporated in inner mito-
membrane proteins (Wallin and von Heijne, 1998). In addition, chondrial membrane vesicles from the yeast Saccharomyces
there must be numerous peripheral membrane proteins, consid- cerevisiae by addition from ethanolic solution. After photoactiva-
ering that lipid binding domains are among the most common tion, carbonate wash, and separation of the proteins on
domains found in the eukaryotic proteome (DiNitto et al., 2003). SDS-PAGE, crosslinked proteins were visualized by measuring
Nevertheless, membrane proteins are traditionally underrepre- differences in UV absorbance, originating from the attached
sented in most proteomics studies, which is caused mainly by probe molecules. Two protein bands were found to contain
Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved
3

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
incorporation in inner mitochondrial membrane vesicles from
S. cerevisiae. In addition to crosslinks to established integral
and peripheral membrane proteins, crosslinks to proteins not
previously described as membrane proteins in yeast were also
found, indicating the potential of the new lipid analogs as tools
for proteome analysis.
RESULTS
Design and Synthesis
Three possible photoreactive crosslinking moieties were consid-
ered for the new lipid analogs: a diazirine (2), a benzophenone (3)
and a phenylazide (4) as shown in Figure 2A.
Phospholipid probe 2 was synthesized via a reductive amina-
tion of the aldehyde 3-(4-formylphenyl)-3-(trifluoromethyl)diazir-
ine, synthesized by using previously reported methods (Delfino
et al., 1993), with 1,2-dilauroyl-sn-glycero-3-phospho-ethanol-
amine (DLPE). The resulting lipid analog was subsequently meth-
ylated to obtain 2. However, preliminary results showed that di-
azirine probe 2 did not establish significant crosslinking to
proteins (data not shown), most likely due to the high reactivity
of the transient carbene species generated by photoactivation,
causing reaction with water.
Figure 1. Outline of the Labeling Approach for Phospholipid-Inter-
action-Based Proteomics
A PC analog containing a photoactivatable headgroup and azides as fishing
baits at the end of the acyl chains, is incorporated in a biological membrane.
Photocrosslinking is performed, and noncrosslinked peripheral membrane
proteins are washed away by carbonate. Subsequently the system is solubi-
lized in 1% (w/v) SDS to attach a rhodamine or biotin reporter moiety (R) to
the crosslink products for detection and purification, respectively.
Based on the retrosynthetic analysis of probes 3 and 4 (Fig-
ure 2B), the photoactivatable phosphocholine-like headgroup in
A was constructed from a common precursor by alkylation of
crosslink products. Analysis by liquid chromatography-tandem the tertiary amine in B with a photoreactive benzylic bromide
mass spectrometry (LC-MS/MS) yielded a list of integral and (10 or 11, Figure 3) (Wang et al., 2004). Because of the sensitivity
peripheral membrane proteins that had potentially been cross- of the photoreactive functionalities, they were introduced in the
linked to ASA-DLPE. Although this study demonstrated the final step of the synthesis. Therequired (N,N-dimethylamino)ethyl
feasibility of photocrosslinking in detecting the phospholipid- phosphate 6 for B was accessible from commercially available 5
interacting proteome, it suffered from two limitations. First, the (LeKim et al., 1979).
detection by UV absorbance was not specific because proteins
Thus, sn-glycerophosphocholine (5) was subjected to a
containing heme or flavin also produced a signal. Second, the monodemethylation reaction in the presence of 1,4-diazabicy-
presence of a large excess of uncrosslinked proteins compli- clo[2,2,2]octane (DABCO) (Figure 3) (LeKim et al., 1979; Wang
cated the identification of proteins that were crosslinked.
et al., 2004). Subsequent acylation of crude 6, with the imidazo-
To eliminate these shortcomings, we developed novel lipid lide generated in situ from 1,1’-carbonyldiimidazole (CDI) and
analogs (3 and 4) containing azides as fishing baits attached to 11-azidoundecanoic acid 8 (prepared from commercially avail-
short C11 lipid acyl chains that could be added to biological able 11-bromoundecanoic acid 7), afforded the tertiary amine 9
membranes from ethanolic solution. Azides are small and rela- in 34% overall yield over two reaction steps. Although the yield
tively apolar moieties, and therefore not expected to disturb the was moderate, this strategy afforded an advanced intermediate
packing of the lipid bilayer. After photoactivation, carbonate that would otherwise be much more difficult to synthesize in
wash, and solubilization of the membranes, they can be used a very short reaction sequence (Hermetter et al., 1989). All
for the covalent attachment of an alkyne-conjugated tag for visu- attempts to improve the yield of this reaction (e.g., by changing
alization or affinity purification (Figure 1). The chemoselective the sequence of reactions or by using alternative acylation
reaction employed, a Cu(I)-catalyzed 1,3-dipolar cycloaddition protocols) were not successful.
commonly referred to as click chemistry (Rostovtsev et al.,
Next, tertiary amine 9 was alkylated with the appropriate
2002; Tornøe et al., 2002), has proven a reliable ligation method photoreactive probes (10 and 11) as their benzylic bromides.
in proteomics studies because the utilized functionalities do not The required benzylic bromide for the synthesis of phospholipid
occur in biological systems and are inert under physiological probe 3, 4-(bromomethyl)benzophenone 10 is commercially
conditions (Ballell et al., 2005; Chan et al., 2004; Kuhlmann available. The 4-azidobenzyl bromide 11 required for the
et al., 2002; Salisbury and Cravatt, 2007; Speers and Cravatt, synthesis of phospholipid probe 4 was synthesized by diazo
2004). The novel probes were designed to mimic phosphatidyl- transfer to 4-aminobenzyl alcohol 12 (Liu and Tor, 2003), and
choline, the most abundant phospholipid in most eukaryotes, conversion of the resulting azido-alcohol into the corresponding
with the photoreactive group attached directly to the quaternary bromide. Bromide 11 was obtained in a moderate yield of 41%
ammonium moiety of the headgroup. Using rhodamine-alkyne with incomplete conversion.
and biotin-alkyne conjugates for detection and purification of
Alkylation of tertiary amine 9 with the benzylic photoreactive
crosslink products, respectively, the new lipid analogs were probes 10 and 11 was unsuccessful using a previously reported
found to crosslink specific subsets of membrane proteins after procedure (Wang et al., 2004). However, it was found that
4
Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
Detection of Crosslinked Proteins in Model Membranes
Both benzophenone-PC (3) and phenylazide-PC (4) were first
tested for their ability to crosslink to the peripheral membrane
protein apo-cytochrome c in a model membrane system. Using
the lipid analog ASA-DLPE (1), it was previously shown that
crosslinking of this protein to membrane vesicles increased its
carbonate wash-resistance (Gubbens et al., 2007). Large unila-
mellar vesicles obtained by extrusion technology (LUVET) were
prepared containing 1,2-dioleoyl-sn-glycero-3-phosphocholine
(DOPC), the anionic lipid 1,2-dioleoyl-sn-glycero-3-[phospho-
rac-(1-glycerol)] (DOPG) (7:3 mol/mol) to enhance the interaction
with the basic apo-cytochrome c, and one of the new lipid probes
(4 mol %). Apo-cytochrome c (0.12 mg/ml) was added to the
LUVET (4 mM phospholipid) and a crosslink experiment was per-
formed as described for ASA-DLPE (1) (Gubbens et al., 2007).
After UV illumination at two different wavelengths, the vesicles
were washed with carbonate, pelleted, and analyzed for associ-
ated apo-cytochrome c on SDS-PAGE gel using Coomassie
staining (Figure 4). The results show an increase in the amount
of apo-cytochrome c associated with the carbonate-washed
vesicles after UV activation of the crosslinkers for 20 min. UV
illumination at a wavelength of 254 nm was more effective in pho-
toactivation than at 366 nm (compare lanes 3 and 5). Moreover,
photoactivation of the phenylazide-PC probe (4) at 254 nm
caused a smear above the protein band (lanes 3 and 4). A similar
effect was observed previously for the ASA-DLPE probe (1), in
which case the smear was found to contain crosslink products
(Gubbens et al., 2007). For these reasons, 254 nm was selected
as the preferred wavelength to perform crosslinking with both
lipids.
Detection of Crosslinked Proteins in Biological
Membranes
Next, the probes were incorporated in inner mitochondrial
membrane vesicles (IMV) from the yeast S. cerevisiae by addition
from a solution in ethanol, and their ability to crosslink to
membrane proteins was tested. After crosslinking, the vesicles
were washed with or without carbonate and solubilized in 1%
(w/v) SDS to perform click chemistry. A tetramethylrhodamine
(TAMRA)-alkyne conjugate (Invitrogen) was used as a fluorescent
reporter. Proteins were isolated from the reaction mixture by
precipitation, separated on SDS-PAGE gels, and scanned for
fluorescence (Figure 5A). Both crosslinking with phenylazide-
PC (4, lanes 1 and 2) and with benzophenone-PC (3, lanes 7
and 8) yielded multiple fluorescent protein bands. The intensity
of most bands was unaltered by the carbonate wash, suggesting
Figure 2. Structures and Retrosynthetic Analysis of the Photoacti-
vatable Lipid Analogs
(A) Structures of ASA-DLPE (1), TPD-PC (2), benzophenone-PC (3), and phe-
nylazide-PC (4).
(B) The retrosynthetic analysis of phospholipid analogs 3 and 4.
treatment of 9 with four equiv 4-(bromomethyl)benzophenone 10 that the probes crosslink integral membrane proteins and/or
in N,N-dimethylformamide (DMF) in the presence of 2,6-lutidine peripheral membrane proteins that are rendered carbonate
(2 equiv) as base under microwave irradiation (2 3 10 min at wash-resistant by the covalent linkage to the lipid probes. Strik-
120ꢀC) led to complete conversion of the starting material to ingly, the two probes crosslink to different, albeit partially over-
the desired photoprobe 3. Under the same conditions, bromide lapping, subsets of proteins. The fluorescence signal observed
11 reacted smoothly with amine 9 to phospholipid probe 4. for benzophenone-PC showed stronger resemblance to the Coo-
Although thin-layer chromatography (TLC) analysis indicated massie-stained pattern than that for phenylazide-PC (Figure 5B).
a nearly complete conversion of the starting material, the isolated In the absence of UV activation, the benzophenone-PC probe
yields of photoprobes 3 and 4 were moderate, 45% and 60%, yielded more background fluorescence (lanes 9 and10) than phe-
respectively, due to the formation of an unidentified side product nylazide-PC (lanes 3 and 4), of which the background signal was
with similar Rf-value that complicated purification by column comparable to that obtained in the absence of crosslinker (lanes
chromatography. Purity and identity of the products were 11 and 12). The experiment was also performed using phenyla-
confirmed by TLC, 1H and 13C NMR, and ESI-MS analysis.
zide-PC at a five times reduced concentration (Figure 5A, lanes
Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved
5

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
Figure 3. Synthesis of the Photoactivatable Phospholipid Probes 3 and 4
5 and 6). Labeling intensity was less, but the labeling pattern did peroxidase (HRP) conjugate on western blot (Figure 6A, lanes 1
not differ from that detected in lanes 1 and 2. Altogether, the data and 2). The banding pattern obtained for the biotinylated proteins
suggest that phenylazide-PC is the more suitable probe to iden- resembled the pattern for the fluorescently labeled proteins
tify proteins that are crosslinked in IMV.
(compare lanes 1 and 2 of Figures 5A and 6A), although the reso-
lution and dynamic range of the fluorescence detection method
were much better. Neutravidin-HRP also bound to some proteins
Purification and Identification of Crosslinked Proteins
To identify the proteins crosslinked to phenylazide-PC (4) they when no crosslinking had been performed (Figure 6A, lanes 3 and
were reacted with biotin-alkyne conjugate (Invitrogen) to enable 4) or with no crosslinker present (lanes 5 and 6). The band at
purification on neutravidin beads. Biotinylation of crosslink prod- 250 kDa could be explained by the endogenously biotinylated
ucts was confirmed by detection with a neutravidin-horseradish mitochondrial protein Hfa1p (Hoja et al., 2004). In yeast mito-
chondria, no endogenously biotinylated proteins of smaller size
are known. The prominent band at 35 kDa probably represents
a protein to which the neutravidin-HRP conjugate binds aspecifi-
cally, because it did not bind to neutravidin beads during the
purification of crosslinked proteins (Figure 6B, lane 2).
To purify crosslink products, proteins precipitated from the
click chemistry reaction mixture corresponding to samples
shown in lanes 1 and 3 of Figure 6A were dissolved by heating
in 1% (w/v) SDS, and, after appropriate dilution, bound to neutra-
vidin beads. Nonbiotinylated proteins were washed away, and
the proteins of interest were eluted from the beads by heating
in 2-fold concentrated SDS sample buffer. No biotinylated
Figure 4. Crosslinking of Apo-Cytochrome c to Model Membranes
LUVETs (DOPC/DOPG 7:3 [mol/mol] containing 4 mol % of the indicated
proteins could be detected when no crosslinking was per-
crosslinker) were incubated with apo-cytochrome c (0.12 mg/mL). Crosslink-
formed, in contrast to the situation after crosslinking with phenyl-
ing was performed at a lipid concentration of 4 mM using UV light of the indi-
azide-PC (Figure 6B). The eluted proteins were separated on
cated wavelengths for 20 min, and the vesicles were washed with or without
SDS-PAGE, digested, and identified by LC-MS/MS analysis
(Table 1; see Table S1 available online for details on the protein
carbonate. Apo-cytochrome c attached to the vesicles was precipitated,
separated on SDS-PAGE, and stained with Coomassie.
6
Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
Figure 6. Biotin Labeling and Purification of Crosslink Products
from Inner Membrane Vesicles
A crosslink experiment was performed in IMVs as described in the legend of
Figure 5 using 75 pmol phenylazide-PC (4) per microgram mitochondrial
protein, but instead of TAMRA, a biotin-alkyne conjugate was added via click
chemistry. Samples were separated on gel, and proteins were transferred to
a nitrocellulose membrane by western blotting. Biotinylated proteins were
detected using a neutravidin-HRP conjugate (A). Samples corresponding to
lane 1 (+UV) and lane 3 (ꢁUV) were incubated with neutravidin beads in the
presence of SDS. Proteins were eluted from the beads, separated on gel,
and biotinylated proteins were detected as described above (B).
Figure 5. TAMRA Labeling of Crosslink Products from Mitochondrial
Inner Membrane Vesicles
The indicated amounts of crosslinker per amount of mitochondrial protein
were added from a solution in ethanol to IMVs from yeast mitochondria. The
samples were either activated with UV light or left in the dark, and subse-
quently washed with or without carbonate, as indicated. Proteins were solubi-
lized in 1% (w/v) SDS, subjected to click chemistry using a TAMRA-alkyne
conjugate, and separated on a CRITERION XT 4%–12% (w/v) Bis-Tris gel
(Bio-Rad) in MES running buffer. The gel was scanned for fluorescence (A)
and subsequently stained with Coomassie (B). The Coomassie-stained pattern
is shown only for lanes 1 and 2, because the other lanes showed correspond-
ing banding patterns.
crosslinked to photoactivatable lipid analogs. Therefore, this
method is a clear improvement over the approach described
previously (Gubbens et al., 2007) to detect the phospholipid
headgroup-interacting proteome.
Fluorescence detection of crosslinked proteins showed that
benzophenone-PC (3) and phenylazide-PC (4) crosslink to
distinct subsets of proteins that partially overlap. Some proteins
are more efficiently or (almost) exclusively crosslinked by either
phenylazide or benzophenone, whereas others show similar
extents of crosslinking, which was confirmed by LC-MS/MS
analysis of the biotinylated and affinity-purified crosslink prod-
ucts of both probes (Tables 1 and S2). The differences in labeling
by the two probes probably reflect the different characteristics of
the photoactivatable moieties. For instance, benzophenone has
a preference to insert into C-H bonds (Dorman and Prestwich,
1994), whereas phenylazides often attack N-H bonds. Alterna-
tively, the photoactivatable moieties might adopt different local-
izations at the membrane-water interface, exposing the probe
molecules to amino acid residues that might differ in reactivity.
In addition, the hydrophobic benzophenone moiety is larger
than the phenylazide group, and therefore requires a larger
binding pocket. Finally, the benzophenone moiety might render
the membrane interface more hydrophobic, consistent with the
observed decreased binding to benzophenone-PC containing
LUVETs of cytochrome c (data not shown), which mainly interacts
electrostatically with model biological membranes (Jordi and De
Kruijff, 1996). In contrast, it was shown that apo-cytochrome c,
identifications). A total of 47 yeast proteins were found in the two
samples, of which 37 are known mitochondrial proteins. A total
of 27 proteins showed a significant increase (R2-fold) in the
number of identified spectra after they were exposed to UV light
in the presence of phenylazide-PC (4), and are therefore
presumed to have been crosslinked by this lipid analog. A
number of proteins was also detected in the absence of UV acti-
vation, which was attributed to aspecific binding to the neutravi-
din beads with the possible contribution of the unavoidable
minute exposure of the sample to light. In comparison, LC-MS/
MS analysis of the affinity-purified, biotinylated, crosslink prod-
ucts of benzophenone-PC (Table S2) demonstrated a significant
increase in the number of identified spectra of 28 proteins, of
which 13 were also found to be labeled by phenylazide-PC.
This confirmed the interpretation of the fluorescence data in
Figure 5 that the two probes crosslink to different, albeit overlap-
ping, subsets of proteins.
DISCUSSION
In this study, we demonstrated that click chemistry is a conve- which binds to the membrane due to exposed hydrophobic
nient method to attach a tag to lipid-protein crosslink products regions, binds and crosslinks equally well to vesicles containing
for either fluorescence detection or affinity purification, thus al- either probe. In IMV, benzophenone-PC produced higher back-
lowing the unambiguous identification by LC-MS/MS of proteins ground fluorescence than phenylazide-PC, which might be
Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved
7

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
Table 1. LC-MS/MS Analysis of Phenylazide-PC Crosslink Products Purified on Neutravidin Beads
Total no. of Identified Spectra
Protein
Name
Open Reading
Frame Name
Molecular
Weight (kDa) +UV
Description
ꢁUV
Locationa
Mb
Lpd1
YFL018C
Dihydrolipoamide dehydrogenase component (E3) 54
of the pyruvate dehydrogenase and 2-oxoglutarate
dehydrogenase multienzyme complexes
11
1
Gut2
Ndi1
YIL155C
Mitochondrial glycerol-3-phosphate
dehydrogenase
72
13
12
4
4
M
YML120C
NADH:ubiquinone oxidoreductase, transfers
electrons from NADH to ubiquinone in the
respiratory chain but does not pump protons
57
Matrixc
Nop1
YDL014W
Nucleolar protein, component of the small
subunit processome complex
34
3
1
Nd
—
YDR119W-A
YGR244C
YNL005C
Putative protein of unknown function
Beta subunit of succinyl-CoA ligase
7
47
43
3
3
3
1
1
1
Unknown
M
Lsc2
Mrp7
Mitochondrial ribosomal protein of the large
subunit
Matrix
Cor1
Dld1
YBL045C
YDL174C
Core subunit of the ubiquinol-cytochrome c
50
65
11
16
4
6
MIMe
MIM
reductase complex (bc1 complex)
D-lactate dehydrogenase, oxidizes D-lactate to
pyruvate
Nde1
Cyt1
YMR145C
YOR065W
Mitochondrial external NADH dehydrogenase
63
34
8
3
5
M
Cytochrome c1, component of the mitochondrial
13
MIM
respiratory chain
Atp1
YBL099W
Alpha subunit of the F1 sector of mitochondrial
F1F0 ATP synthase
59
10
4
MIM
Hhf1,2
Cox2
Cox4
Atp17
YBR009C,YNL030W Two identical histone H4 proteins
11
29
17
11
7
4
2
4
3
2
N
Q0250
Subunit II of cytochrome c oxidase
Subunit IV of cytochrome c oxidase
MIM
MIM
MIM
YGL187C
YDR377W
1
Subunit f of the F0 sector of mitochondrial
F1F0 ATP synthase
—
Ald4
YOR374W
Mitochondrial aldehyde dehydrogenase
57
14
52
4
3
3
—
—
—
M
N
Hta1,2
Lat1
YDR225W,YBL003C Nearly identical histone H2A subtypes
YNL071W
YOR136W
YPR024W
YGL206C
Dihydrolipoamide acetyltransferase component
(E2) of pyruvate dehydrogenase complex
M
Idh2
Subunit of mitochondrial NAD(+)-dependent
isocitrate dehydrogenase
40
82
3
3
2
—
—
—
Matrix
MIM
Yme1
Chc1
Subunit, with Mgr1p, of the mitochondrial
inner membrane i-AAA protease complex
Clathrin heavy chain, subunit of the major
coat protein involved in intracellular protein
transport and endocytosis
187
Endocytic
vesicles
Phb2
YGR231C
Subunit of the prohibitin complex
34
2
—
MIM
(Phb1p-Phb2p), a 1.2 MDa ring-shaped inner
mitochondrial membrane chaperone
Atp2
YJR121W
YKL179C
Beta subunit of the F1 sector of mitochondrial
F1F0 ATP synthase
55
77
2
2
—
—
MIM
Coy1
Golgi membrane protein with similarity to
mammalian CASP
Golgi
—
YKR016W
YPL004C
Unknown protein, localized to the mitochondria
61
38
2
2
—
—
M
Lsp1
Long chain base-responsive inhibitor of protein
kinases Pkh1p and Pkh2p
Eisosome
Qcr2
Por1
YPR191W
YNL055C
Subunit 2 of the ubiquinol cytochrome c
40
30
19
18
10
11
MIM
reductase complex
Mitochondrial porin (voltage-dependent
anion channel)
MOMf
8
Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
Table 1. Continued
Total no. of Identified Spectra
Protein
Name
Open Reading
Frame Name
Molecular
Weight (kDa) +UV
Description
ꢁUV
2
Locationa
N
Hht1,2
Kgd1
YBR010W,YNL031C Two identical histone H3 proteins
15
3
YIL125W
Component of the mitochondrial alpha-
ketoglutarate dehydrogenase complex
114
11
8
Matrix
Pet9
YBL030C
Major ADP/ATP carrier of the mitochondrial
inner membrane
34
16
12
MIM
Atp18
Kgd2
YML081C-A
YDR148C
Subunit of the mitochondrial F1F0 ATP synthase
7
4
3
MIM
Dihydrolipoyl transsuccinylase, a component
of the mitochondrial alpha-ketoglutarate
dehydrogenase complex
50
12
11
Matrix
Sdh1
Sdh2
YKL148C
YLL041C
Flavoprotein subunit of succinate dehydrogenase
70
30
6
3
6
3
MIM
MIM
Iron-sulfur protein subunit of succinate
dehydrogenase
Hsp60
Qcr7
YLR259C
YDR529C
Tetradecameric mitochondrial chaperonin
61
15
3
2
3
2
M
Subunit 7 of the ubiquinol cytochrome c
MIM
reductase complex
Atp7
YKL016C
Subunit d of the stator stalk of mitochondrial
F1F0 ATP synthase
20
2
2
MIM
—
YEL025C
YJR077C
SWI/SNF and RSC interacting protein 1
136
33
11
13
5
N, CPg
M
Mir1
Mitochondrial phosphate carrier, imports
inorganic phosphate into mitochondria
4
Cox9
Aco1
YDL067C
YLR304C
Subunit VIIa of cytochrome c oxidase
7
2
3
3
MIM
Aconitase, required for the tricarboxylic acid
cycle and mitochondrial genome maintenance
85
2
Matrix
Atp4
YPL078C
Subunit b of the stator stalk of mitochondrial
F1F0 ATP synthase
27
2
4
MIM
Htb1,Htb2 YDR224C,YBL002W Nearly identical histone H2B subtypes
14
71
1
1
2
5
N
Ssc1
YJR045C
Mitochondrial matrix ATPase that is a subunit
of the presequence translocase-associated
protein import motor
MIM
Rpl13a,b YDL082W,
YMR142C
Protein components of the large (60S)
ribosomal subunit
23
—
2
CP
Samples corresponding to the samples shown in lanes 1 (+UV) and 3 (ꢁUV) of Figure 6 were incubated with neutravidin beads. Biotinylated proteins
were eluted by heating in SDS-PAGE sample buffer, separated on gel, digested, and subjected to LC-MS/MS analysis. The total number of identified
spectra as detected for each protein in two independent experiments is shown. Proteins are ranked in decreasing order of the ratio of the number of
identified spectra with / without photo-activation and sorted into groups, separated by horizontal lines, with a ratio R 2, 1 < ratio < 2, and ratio % 1. The
number of spectra for proteins detected at less than 99.0% probability in one of the two conditions is given in italics.
a Localization according to SGD (Saccharomyces genome database).
b M represents mitochondrial.
c Matrix represents mitochondrial matrix.
d N represents nucleus.
e MIM represents mitochondrial inner membrane.
f MOM represents mitochondrial outer membrane.
g CP represents cytoplasm.
caused by some crosslinking occurring in the dark or by aspecific activatable moiety in the lipid structure, the benzophenone
binding of the more hydrophobic benzophenone moiety to moiety could be an excellent alternative because of its high
membrane proteins. In addition, benzophenone-PC was found chemical stability and high selectivity even in the presence of
to preferentially crosslink to abundant proteins that were also water and strong nucleophiles (Dorman and Prestwich, 1994).
stained by Coomassie, whereas phenylazide-PC demonstrated
Several proteins crosslinked by phenylazide-PC and benzo-
crosslinking to numerous proteins, mainly of low molecular phenone-PC have been described to interact with PC or PC-
weight, that were not visible after Coomassie staining. In view analogs in mitochondrial inner membranes. For example, the
of the above, we argue that phenylazide-PC is the preferred, crystal structure of the S. cerevisiae ubiquinol-cytochrome c
more specific crosslinker to be used in the biological system reductase complex (cytochrome bc1 complex) contains two
described in this research. However, depending on the lipid of tightly bound PC molecules (Lange et al., 2001; Palsdottir et al.,
interest, type of biological membrane, and location of the photo- 2003) interacting with the subunits Cor1p, Cobp, and Cyt1p. Of
Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved
9

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
these proteins, Cor1p and Cyt1p were found to be crosslinked by
Most proteins in Table 1 are present at high copy numbers in
phenylazide-PC and benzophenone-PC. Cobp was not found, the mitochondrial inner membrane. The relative lack of low
possibly because its highly hydrophobic nature might interfere copy number proteins is probably related to the difficulty of
with detection by LC-MS/MS analysis. With the possible excep- detecting them in an LC-MS/MS approach. However, lowly
tion of Qcr2p, which showed an increase in the number of identi- expressed proteins are crosslinked by phenylazide-PC, consid-
fied spectra just below the threshold after crosslinking, no other ering the difference between the fluorescence signal and
subunits of the cytochrome bc1 complex were found to be cross- Coomassie stain obtained after crosslinking with this probe.
linked by phenylazide-PC, indicating high specificity of the phe- The highly abundant ADP/ATP carrier Pet9p does not show
nylazide-PC probe for PC binding pockets. Benzophenone-PC a significant increase in the number of identified spectra upon
did crosslink to additional subunits of the complex, including photoactivation of phenylazide-PC (Table 1), demonstrating
Qcr2p. One PC molecule has been found in the crystal structure that the probe does not specifically crosslink abundant
of bovine heart cytochrome c oxidase at a position next to proteins.
subunits II and VIc (Shinzawa-Itoh et al., 2007). Although the The carbonate wash resistance of the crosslinked peripheral
structure of yeast cytochrome c oxidase might differ, phenyla- membrane proteins described above is consistent with the
zide-PC was found to crosslink subunit II, Cox2p, in the present observation that fluorescently labeled crosslink products were
study.
carbonate wash-resistant. This indicates that the short azide-
Another mitochondrial protein, glycerol-3-phosphate dehy- containing acyl chains of the probes, which allow for the incorpo-
drogenase (Gut2p), has previously been described to be cross- ration of the lipid analogs in the biological membrane by addition
linked by TID-PC (Janssen et al., 2002), a PC analog with the from an ethanol solution, were sufficient to anchor peripheral
photoactivatable moiety attached to one of the acyl chains. proteins to the membrane. The probes are most likely able to
Recently, Gut2p was shown to be a peripheral membrane protein flip-flop across the membrane based on the finding that periph-
that was carbonate-wash sensitive and to depend on PC for eral proteins from both the matrix and the intermembrane space
efficient functioning (Rijken et al., 2007). Interestingly, phenyla- side of the membrane were crosslinked. This is consistent with
zide-PC and benzophenone-PC were able to crosslink Gut2p, the previous finding that externally added spin-labeled PC
confirming the interaction of this protein with PC, irrespective of analogs rapidly redistribute over the two leaflets of the inner
the nature of the photoactivatable moiety and its position on mitochondrial membrane (Gallet et al., 1999).
the headgroup. Therefore, the new lipid-crosslinkers appear to
be successful in crosslinking peripheral membrane proteins.
There was little overlap between the crosslink candidates
found for ASA-DLPE (1) (Gubbens et al., 2007) and the proteins
Proteins that were not previously described to interact with PC crosslinked to phenylazide-PC found in the present study, with
include the two subunits of the peripheral part of the F1F0 ATP Gut2p and Cyt1p as exceptions. Because ASA-DLPE and phe-
synthase, Atp1p and Atp2p, which were crosslinked by phenyla- nylazide-PC have an almost identical photoactivatable moiety,
zide-PC. Also new PC-interacting proteins were identified that the most likely explanation for these observations is the different
have not previously been described as membrane proteins in headgroup structure of the lipid analogs: ASA-DLPE carries a net
S. cerevisiae or whose membrane association is still disputed. negative charge, whereas phenylazide-PC has a zwitterionic
Two components of the pyruvate dehydrogenase complex, headgroup similar to PC. Binding sites of PC have been found
Lpd1p and Lat1p, were found to be crosslinked by both probes. to differ from binding sites of negatively charged phospholipids
Lpd1p also serves as subunit of the homologous 2-oxoglutarate in that the latter contain more polar and positively charged resi-
dehydrogenase complex. Both complexes function in the tricar- dues, absent in PC binding pockets so as to accommodate the
boxylic acid cycle in the mitochondrial matrix. Although in bulky and positively charged choline moiety (Palsdottir and
bacteria they are recovered in the soluble cytoplasmic fraction, Hunte, 2004). In support of this view, in LUVETs, the PC-analogs
they were found to be associated with the mitochondrial inner used in this study did not crosslink to holo-cytochrome c, which
membrane in plants (Millar et al., 1999) and mammals (Maas interacts electrostatically with anionic lipids (data not shown).
and Bisswanger, 1990), requiring detergents for solubilization. It The partial overlap observed between the subsets of proteins
was speculated that this association might occur through crosslinked by phenylazide-PC and benzophenone-PC, indi-
complex I (Millar et al., 1999), but this complex is not present in cating that these probes target similar sites, in combination
S. cerevisiae. Therefore, association of these complexes to the with the ASA-DLPE data, suggests that the phospholipid head-
mitochondrial inner membrane through lipid-protein interactions group structure and charge are more important than the nature
might provide an alternative explanation. Other likely peripheral of the photoactivatable moiety. More information about specific
membrane proteins that were crosslinked by phenylazide-PC lipid-protein interactions will have to be obtained to characterize
include the NADH:ubiquinone oxidoreductases Ndi1p and the lipid binding motifs involved.
Nde1p, which transfer electrons from NADH to quinones and
In summary, the new lipid analogs developed in this study
have been proposed to bind to the membrane through an amphi- allow for the straightforward detection and identification of the
pathic helix (Melo et al., 2004). Mrp7p, which is part of the large lipid headgroup-interacting proteome. Phenylazide-PC was
subunit of mitochondrial ribosomes, was also found to be cross- found to be the most suitable photoactivatable lipid for this
linked, consistent with the previously observed membrane asso- purpose. It crosslinks to both established interaction partners
ciation of mitochondrial ribosomes (Jia et al., 2003; Szyrach et al., of PC and possible new interaction partners in IMV, regardless
2003). A last example of a potential peripheral membrane protein of whether they are peripheral or integral membrane proteins.
is Ald4p, an aldehyde dehydrogenase that converts acetalde- In addition, new peripheral membrane proteins, not previously
hyde to acetate and is required for growth on ethanol.
described as membrane associated, were identified.
10 Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
(ethylenediaminetetraacetic acid), 5 mM HEPES (4-[2-hydroxyethyl]-N0-piper-
azineethanesulfonic acid) (pH 7.4). TLC plates (silica gel 60F254 on glass) were
purchased from Merck (Whitehouse Station, NJ).
SIGNIFICANCE
Here, a new proteomics approach is described to identify
proteins interacting with phospholipid headgroups at the
membrane interface. Different analogs of phosphatidylcho-
line, an abundant phospholipid in eukaryotes, were synthe-
sized from the commercially available precursor sn-glycero-
phosphocholine, thus allowing the addition of custom acyl
chains, containing azides for click chemistry, and the addi-
tion of different photoactivatable moieties to the quaternary
ammonium group of phosphatidylcholine. The probes were
validated in a model membrane system in which they were
found to render the peripheral membrane protein apo-cyto-
chrome c carbonate wash-resistant, as described previously
for the photoactivatable probe ASA-DLPE (Gubbens et al.,
2007). Photocrosslinking was performed in situ in the inner
mitochondrial membrane from S. cerevisiae after adding
the probes from an ethanol solution. The system was solubi-
lized to attach a reporter molecule containing a terminal
alkyne to crosslink products via click chemistry. Lipid
analogs containing benzophenone and phenylazide as
photoactivatable moieties crosslinked to specific, partially
overlapping subsets of proteins as detected by a fluorescent
reporter. Compared with phenylazide, benzophenone gave
rise to a higher background signal in the absence of photo-
crosslinking. The crosslink products were biotinylated via
click chemistry and purified on neutravidin beads. Mass
spectrometry analysis of crosslinked proteins identified
both established and potential new interaction partners of
phosphatidylcholine, and confirmed the partial overlap in
the subsets of proteins crosslinked by both probes. Estab-
lished interaction partners included protein subunits adja-
cent to this lipid in the crystal structure of membrane protein
complexes, in agreement with the existence of phosphati-
dylcholine binding pockets. New interaction partners
included proteins not previously described as membrane
proteins in yeast, and which might be attached peripherally
to the mitochondrial inner membrane. These results demon-
strate the high potential of our approach to identify a phos-
pholipid-interacting proteome. In principle, the method can
be extended to any membrane or any lipid.
General Methods
Microwave experiments were performed in a microwave reactor (Biotage
Initiator, Uppsala, Sweden). Visualization of TLC spots was achieved with
UV light, by staining with an aqueous KMnO4 (1%) containing 2% K2CO3, or
with molybdenum blue spray reagent (Sigma, 1.3% molybdenum oxide in
4.2 M aqueous H2SO4). NMR spectra were recorded on a Varian Mercury
300 spectrometer. Chemical shifts were calibrated to tetramethylsilane
(1H, d = 0.00 ppm), CDCl3
( (
13C, d = 77.0 ppm), CFCl3 19F, d = 0.00 ppm),
and H3PO4
(
31P, d = 0.00 ppm). Electrospray mass spectra of synthesized
compounds were measured on a Shimadzu LCMS-QP8000 single quadrupole
bench-top mass spectrometer, which was operated in the positive ionization
mode. Phospholipid and phospholipid-crosslinker concentrations were
measured by phosphorous determination (Rouser et al., 1970) after destruc-
tion in perchloric acid at 180ꢀC using KH2PO4 (0–50 nmol) as a standard.
Protein concentrations were determined using the BCA method (Pierce) with
0.1% (w/v) SDS added and bovine serum albumin as a standard. Incorporation
of the photoactivatable probes in LUVETs and IMVs was performed under red
safety light conditions, and their photoactivation was performed using a Min-
eralight UVGL-58 UV lamp (UVP Inc., San Gabriel, CA). In gel detection of
TAMRA was performed on a Typhoon 9400 imaging system (GE Healthcare
Bio-Sciences AB, Uppsala, Sweden) using a 532 nm laser for excitation and
a 580 15 nm band-pass filter for emission.
Synthesis
Compounds 2, 8, and 11
See Supplemental Experimental Procedures.
Compound 3
To a solution of 9 (66 mg, 0.10 mmol) in DMF (3 ml) was added 4-(bromome-
thyl)benzophenone (10, 110 mg, 0.40 mmol) and 2,6-lutidine (23 ml, 21 mg,
0.20 mmol). The mixture was heated to 120ꢀC by microwave irradiation for
20 min. TLC analysis indicated full conversion. The solvent was removed in va-
cuo and the residue was purified by flash chromatography (CHCl3/MeOH/25%
aqueous NH3 90:10:1 v/v/v / CHCl3/MeOH/25% aqueous NH3 80:20:1 v/v/v)
to give pure 3 (39 mg, 0.046 mmol, 46%) as a slightly yellow wax. Rf(CHCl3/
MeOH/25% aqueous NH3 80:20:1 v/v/v): 0.43; 1H NMR (300 MHz, CDCl3)
d 1.23–1.32 (m, CH2, 24H), 1.46-1.52 (m, CH2, 8H), 2.21-2.28 (m, CH2-CO,
4H), 3.14 (s, N-CH3, 6H), 3.17–3.24 (m, CH2-N3, 4H), 3.64 (t (J 4.4Hz),
N-CH2-CH2, 2H), 3.96 (dd (J 6.6 Hz, J 5.8 Hz), O-CH2, 2H), 4.11 (dd (J 12.1 Hz,
J 6.9 Hz), O-CH2, 1H), 4.32 (m, CH2-CH2-O, 2H), 4.37 (dd (J 12.1 Hz, J 3.0
Hz), O-CH2, 1H), 4.69 (s, benzyl CH2, 2H), 5.19 (m, CH-O, 1H), 7.49 (t (J 7.4
Hz), arom CH, 2H), 7.62 (t (J 7.4 Hz), arom CH, 1H), 7.73 (d (J 7.4Hz), arom
CH, 2H), 7.75 (d (J 8.5 Hz), arom CH, 2H), 7.84 (d (J 8.5 Hz), arom CH, 2H);
13C NMR (75.5 MHz, CD3OD) d 26.2, 28.0, 30.0–31.4, 35.1, 35.3, 51.8, 52.6,
60.6, 63.9, 65.1, 65.9, 69.7, 72.0, 129.9, 131.3, 131.7, 133.1, 134.5, 134.7,
138.4, 141.1, 174.8, 175.1, 197.6; ES-MS (50 eV) calcd for C43H66N7O9P:
855.47, found: m/z [M + H]+ 856.95 (100%), [M + Na]+ 878.55 (21%).
Compound 4
EXPERIMENTAL PROCEDURES
Materials
All solvents were distilled before use or were high-pressure liquid chromatog-
To a solution of 9 (66 mg, 0.10 mmol) in DMF (3 ml) was added p-azidobenzyl
bromide (11, 85 mg, 0.40 mmol) and 2,6-lutidine (23 ml, 0.20 mmol). The mixture
was heated to 120ꢀC by microwave irradiation for 35 min. The solvent was
removed in vacuo and the residue was purified by flash chromatography
raphy grade. Anhydrous solvents were obtained by storing the solvents over
˚
activated 4 A molecular sieves. BOP (benzotriazol-1-yl-oxy-tris[dimethylami-
no]phosphonium hexafluorophosphate) was purchased from GL Biochem
(Shanghai, China). sn-Glycero-3-phosphocholine was purchased from
Brunschwig Chemie (Amsterdam, the Netherlands). Horse heart cytochrome
c was purchased from Sigma (St. Louis, MO). Apocytochrome c was derived
from cytochrome c by removing the heme group, as described elsewhere
(Fisher et al., 1973), and stored in small aliquots (7.2 mg/ml) in Na2HPO4 buffer
(10 mM, pH 7.3) at ꢁ20ꢀC. DOPC, DOPG, and DLPE were from Avanti Polar
Lipids (Alabaster, AL). Photoactivatable probes were stored at a concentration
of 5 mM in a 2:1 chloroform/methanol (v/v) solution shielded from light at
ꢁ20ꢀC. IMVs from S. cerevisiae were isolated as described previously
(Gubbens et al., 2007) from the wild-type strain BY4742 (MATa his3D1
leu2D0 lys2D0 ura3D0), and stored in aliquots at ꢁ80ꢀC at a protein concen-
tration of approximately 4 mg/ml in H/K buffer (10 mM KCl, 2.5 mM EDTA
(CHCl3/MeOH/25% aqueous NH3 90:10:1 v/v/v
/ CHCl3/MeOH/25%
aqueous NH3 85:15:1 v/v/v) to give pure 4 (47 mg, 0.060 mmol, 60%) as a color-
less wax. Rf (CHCl3/MeOH/25% aqueous NH3 80:20:1 v/v/v): 0.42; 1H NMR
(300 MHz, CDCl3) d 1.27-1.40 (m, CH2, 24H), 1.54-1.64 (m, CH2, 8H), 2.23-
2.29 (m, CH2-CO, 4H), 3.23-3.28 (m, CH2-N3, 4H), 3.25 (s, N-CH3, 6H), 3.89
(m, N-CH2-CH2, 2H), 3.95 (m, O-CH2, 2H), 4.11 (dd (J 12.1 Hz, J 7.4 Hz),
O-CH2, 1H), 4,37 (dd (J 12.1 Hz, J 2.7 Hz), O-CH2, 1H), 4.42 (m, CH2-CH2-O,
2H), 4.86 (s, benzyl CH2, 2H), 5.19 (m, CH-O, 1H), 7.07 (d (J 8.6 Hz), arom
CH, 2H), 7.65 (d (J 8.6 Hz), arom CH, 2H); 13C NMR (75.5 MHz, CD3OD)
d 26.2, 28.0, 30.0-31.4, 35.1, 51.4, 52.7, 60.5, 63.9, 65.1, 65.5, 69.7, 72.0,
121.0, 125.4, 136.2, 144.5, 174.8, 175.1; ES-MS (50 eV) calcd for
C36H61N10O8P: 792.44, found: m/z [M + H]+ 793.30 (100%).
Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved 11

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
Compound 9
tions. For TAMRA detection, 50 mg protein was used per lane, and for biotin
detection, 20 mg protein was used per lane. In case of TAMRA detection, the
gels were washed in H2O and scanned for fluorescence before gel staining
with Coomassie brilliant blue. In case of biotin detection, proteins were trans-
ferred to a nitrocellulose membrane by western blotting, blocked for 1 h with
gelatin (3%, w/v, Bio-Rad) in TBST (50 mM Tris/HCl, 100 mM NaCl, 0.05%
[v/v] Tween 20, pH 8.0), incubated for 1 h with neutravidin-HRP (1 mg/ml;
Pierce, Rockford, IL) in TBST containing gelatin (0.5%, w/v), and visualized,
after washing, using ECL solutions.
To a solution of sn-glycero-3-phosphocholine (5, 1.3 g, 5.0 mmol) in DMF
(75 ml), DABCO (3.37 g, 30.0 mmol) was added. The mixture was heated to
reflux for 5 hr and the solvent was removed in vacuo to give crude sn-glyc-
ero-3-phospho-N,N-dimethylethanolamine (6) as a brown oil that was used in
the next step without further purification. To a solution of 11-azidoundecanoic
acid (8, 2.27 g, 10.0 mmol) in dry tetrahydrofuran (THF, 25 ml) CDI (1.78 g,
11.0 mmol) was added. After CO2 evolution had ceased and the solution
became clear, the resulting imidazolide solution was added to a solution of
crude 6 (ꢂ5.0 mmol) and 4-(N,N-dimethylamino)pyridine (DMAP, 122 mg,
0.20 mmol) in DMF (100 mL). The mixture was stirred overnight at room temper-
ature and the solvent was removed in vacuo. The crude product was purified by
flash chromatography using a gradient: CHCl3/MeOH/25% aqueous NH3
90:10:1 v/v/v / CHCl3/MeOH/25% aqueous NH3 50:50:1 v/v/v as eluents,
to give 9 (1.11 g, 1.68 mmol, 34%) as a colorless wax. Rf(CHCl3/MeOH/25%
aqueous NH3 80:20:1 v/v/v): 0.35; 1H NMR (300 MHz, CDCl3) d 1.15-1.45 (m,
CH2, 24H), 1.55-1.62 (m, CH2, 8H), 2.29 (t (J 7.4 Hz), CH2, 2H), 2.31 (t (J 7.4
Hz), CH2-CO, 2H), 2.87 (s, N-CH3, 6H), 3.21 (m, N-CH2-CH2, 2H), 3.26 (2 3 t
(J 6.9 Hz), CH2-N3, 2 3 2H), 4.03 (dd (J 7.4 Hz, J 5.5 Hz), O-CH2, 2H), 4.16
(dd (J 11.8 Hz, J 6.6 Hz), O-CH2, 1H), 4.25 (m, CH2-CH2-O, 2H), 4.39 (dd (J
11.8 Hz, J 3.3 Hz), O-CH2, 1H), 5.25 (m, CH-O, 1H); 13C NMR (75.5 MHz,
CD3OD) d 26.2, 28.0, 30.0-31.4, 35.1, 35.3, 43.8, 44.0, 52.6, 59.4, 60.7, 63.8,
65.2, 72.0, 174.8, 175.1; ES-MS (50 eV) calcd for C29H56N7O8P: 661.39, found:
m/z [M + H]+ 662.30 (100%), [M + Na]+ 684.80 (8%), [2M + H]+ 1323.80 (11%).
Affinity Purification
Protein pellets of biotinylated crosslink products, corresponding to approxi-
mately 125 mg protein, were dissolved in Tris/HCl (25 ml, 50 mM, pH 8.0) con-
taining SDS (1%, w/v) by incubating at 70ꢀC for 10 min while shaking. An aliquot
of 5 ml was stored as a control sample in SDS-PAGE sample buffer, and the
remainder was diluted to obtain a 100 ml solution in TBSS (50 mM Tris/HCl,
100 mM NaCl, 0.2% (w/v) SDS, pH 8.0). Neutravidin agarose beads (30 mL,
Pierce) were washed two times in TBSS, the protein solution was added, and
the beads were incubated for 2h at 4ꢀC while rotating. The beads were washed
three times with TBSS and proteins were eluted in concentrated SDS-PAGE
sample buffer (62.5 mM Tris/HCl, 2.5% [w/v] SDS, 10.9% [v/v] glycerol,
50 mM dithiothreitol, Bromophenol Blue, pH 6.8, 28 ml) by heating to 95ꢀC for
10 min. To verify the presence of crosslinked proteins, 7 ml was separated on
SDS-PAGE gel and biotinylated proteins were detected on western blot as
described above. The remainder of the samples was run on an SDS-PAGE
gel using 11% (w/v) acrylamide over a distance of approximately 1.5 cm. After
staining with Coomassie brilliant blue, the lanes were excised, and cut in three
pieces for analysis by LC-MS/MS.
Crosslinking in LUVETs
7:3 (mol/mol) DOPC/DOPG LUVETs, containing 4 mol% of either the benzo-
phenone or phenylazide probe or no photoactivatable probe, as indicated,
were prepared as described (Gubbens et al., 2007). Crosslinking with apo-
cytochrome c (0.12 mg/ml) was carried out as described (Gubbens et al.,
2007) at a lipid concentration of 4 mM. For crosslinking at a wavelength of
254 nm the samples were exposed directly to the UV light for 20 min by
removing the caps of the tubes and placing the lamp above the samples. After
carbonate washing, the LUVETs were resuspended in H-buffer (50 mM NaCl,
1 mM EDTA, 10 mM HEPES, pH 7.5), and the proteins were precipitated in
chloroform / methanol (Wessel and Flugge, 1984). Protein pellets were resolu-
bilized by boiling in SDS-PAGE sample buffer containing dithiothreitol (25 mM)
and analyzed by SDS-PAGE using 15% (w/v) acrylamide. Gels were stained
with Coomassie brilliant blue.
LC-MS/MS Analysis
The gel pieces were subjected to digestion with trypsin and peptides analyzed
by nanoscale LC-MS/MS by coupling an Agilent 1100 Series LC system to
a LTQ XL quadrupole ion trap mass spectrometer (Finnigan, San Jose, CA),
as described previously (Gubbens et al., 2007). The only modification was
that, after digestion, the peptides were extracted using 5% (v/v) formic acid
instead of acetic acid. Tandem mass spectra were extracted and charge state
deconvoluted by BioWorks (Thermo Scientific, Waltham, MA; version 3.3). All
MS/MS samples were analyzed using Mascot (Matrix Science, London, UK;
version 2.2.03) and X! Tandem (www.thegpm.org; version 2007.01.01.1),
both set up to search the Yeast SGD database (5779 entries) with a parent
ion tolerance of 0.5 Da and a fragment ion mass tolerance of 0.9 Da. Fixed
and variable modifications were the iodoacetamide derivative of cysteine
and oxidation of methionine, respectively. Scaffold (version 01_07_00, Pro-
teome Software, Portland, OR) was used to validate MS/MS based peptide
and protein identifications. Peptide identifications were accepted if they could
be established at greater than 80.0% probability as specified by the Peptide
Prophet algorithm (Keller et al., 2002). Protein identifications were accepted
if they could be established at greater than 99.0% probability as specified
by the Protein Prophet algorithm (Nesvizhskii et al., 2003) and contained at
least two identified peptides in one of the samples. Proteins that contained
similar peptides and could not be differentiated based on MS/MS analysis
alone were grouped to satisfy the principles of parsimony.
Crosslinking in Mitochondrial Inner Membrane Vesicles
IMVs corresponding to 200–250 mg protein were thawed and diluted to a total
volume of 0.5 ml with H/K-buffer containing PMSF (phenylmethanesulfonyl
fluoride, 1 mM). The photoactivatable probe indicated was added from a solu-
tion in ethanol (10 ml) to either 75 pmol/mg mitochondrial protein or 15 pmol/mg
protein. For mock treatment only ethanol (10 ml) was added. The mixtures were
incubated at 37ꢀC for 30 min, immediately chilled on ice and, to remove non-
incorporated probe, layered on top of sucrose (0.5 M) in H/K-buffer (500 ml).
After centrifugation at 165,000 3 g for 1 h at 4ꢀC, the pellets were washed in
H/K-buffer (200 ml) and centrifuged for 20 min at 200,000 3 g and 4ꢀC. The
IMV pellets were resuspended in H/K-buffer at a protein concentration of
1 mg/ml, split, and either left in the dark or treated with UV light (254 nm) on
ice for 20 min. Subsequently, either Na2CO3 (100 mM) was added to a final
concentration of 40 mM Na2CO3 or HEPES/KOH (10 mM, pH 7.4) was added
to the same final volume. Both solutions contained KCl (10 mM). After 10 min
on ice, samples were centrifuged for 20 min at 355,000 3 g and 4ꢀC. Pellets
were resuspended in Tris (Trishydroxymethylaminomethane)/HCl (50 mM,
50 ml, pH 8.0) containing SDS (1%, w/v) and used as a substrate for the TAMRA
or biotin Click-iT Detection Kit (Invitrogen, Carlsbad, CA) according to the
manufacturer’s instructions. The resulting labeled protein pellets were stored
at ꢁ20ꢀC until further analysis.
SUPPLEMENTAL DATA
The Supplemental Data include Supplemental Experimental Procedures and
two tables and can be found with this article online at http://www.cell.com/
chemistry-biology/supplemental/S1074-5521(09)00003-9.
ACKNOWLEDGMENTS
This work was supported by the Netherlands Proteomics Centre.
Detection of Crosslinked Proteins
Protein pellets corresponding to 50 mg of protein were heated to 70ꢀC in XT
sample buffer (Bio-Rad, Hercules, CA) supplemented with XT reducing agent
(Bio-Rad) for 10 min under vigorous shaking to completely solubilize all
proteins. Samples were separated on Criterion XT 4%–12% (w/v) Bis-Tris
gels (Bio-Rad) in MES running buffer according to the manufacturer’s instruc-
Received: June 27, 2008
Revised: October 24, 2008
Accepted: November 11, 2008
Published: January 29, 2009
12 Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
REFERENCES
Kuhlmann, J., Tebbe, A., Volkert, M., Wagner, M., Uwai, K., and Waldmann, H.
(2002). Photoactivatable synthetic Ras proteins: ‘‘baits’’ for the identification of
plasma-membrane-bound binding partners of Ras. Angew. Chem. Int. Ed.
Engl. 41, 2546–2550.
Ballell, L., Alink, K.J., Slijper, M., Versluis, C., Liskamp, R.M., and Pieters, R.J.
(2005).
A new chemical probe for proteomics of carbohydrate-binding
proteins. ChemBioChem 6, 291–295.
Lange, C., Nett, J.H., Trumpower, B.L., and Hunte, C. (2001). Specific roles of
protein-phospholipid interactions in the yeast cytochrome bc1 complex struc-
ture. EMBO J. 20, 6591–6600.
Berman, A., Shearing, L.N., Ng, K.F., Jinsart, W., Foley, M., and Tilley, L.
(1994). Photoaffinity labelling of Plasmodium falciparum proteins involved in
phospholipid transport. Mol. Biochem. Parasitol. 67, 235–243.
LeKim, D., Heidemann, G., and Betzing, H. (1979). Verfahren zur Demethylier-
ung von cholinhaltigen Phospholipiden. German patent DE 2728893.
Blonder, J., Conrads, T.P., and Veenstra, T.D. (2004). Characterization and
quantitation of membrane proteomes using multidimensional MS-based
proteomic technologies. Expert Rev. Proteomics 1, 153–163.
Liu, J., and Tor, Y. (2003). Simple conversion of aromatic amines into azides.
Org. Lett. 5, 2571–2572.
Brunner, J. (1993). New photolabeling and crosslinking methods. Annu. Rev.
Maas, E., and Bisswanger, H. (1990). Localization of the alpha-oxoacid dehy-
drogenase multienzyme complexes within the mitochondrion. FEBS Lett. 277,
189–190.
Biochem. 62, 483–514.
Chan, E.W., Chattopadhaya, S., Panicker, R.C., Huang, X., and Yao, S.Q.
(2004). Developing photoactive affinity probes for proteomic profiling: hydrox-
amate-based probes for metalloproteases. J. Am. Chem. Soc. 126, 14435–
14446.
Melo, A.M., Bandeiras, T.M., and Teixeira, M. (2004). New insights into
type II NAD(P)H:quinone oxidoreductases. Microbiol. Mol. Biol. Rev. 68,
603–616.
Delfino, J.M., Schreiber, S.L., and Richards, F.M. (1993). Design, synthesis,
and properties of a photoactivatable membrane-spanning phospholipidic
probe. J. Am. Chem. Soc. 115, 3458–3474.
Millar, A.H., Hill, S.A., and Leaver, C.J. (1999). Plant mitochondrial 2-oxogluta-
rate dehydrogenase complex: purification and characterization in potato.
Biochem. J. 343, 327–334.
Desneves, J., Berman, A., Dynon, K., La Greca, N., Foley, M., and Tilley, L.
(1996). Human erythrocyte band 7.2b is preferentially labeled by a photoreac-
tive phospholipid. Biochem. Biophys. Res. Commun. 224, 108–114.
Montecucco, C. (1988). Photoreactive lipids for the study of membrane-
penetrating toxins. Methods Enzymol. 165, 347–357.
Montecucco, C., Schiavo, G., Gao, Z., Bauerlein, E., Boquet, P., and Das-
Gupta, B.R. (1988). Interaction of botulinum and tetanus toxins with the lipid
bilayer surface. Biochem. J. 251, 379–383.
DiNitto, J.P., Cronin, T.C., and Lambright, D.G. (2003). Membrane recognition
and targeting by lipid-binding domains. Sci. STKE 2003, re16.
Dorman, G., and Prestwich, G.D. (1994). Benzophenone photophores in
Nesvizhskii, A.I., Keller, A., Kolker, E., and Aebersold, R. (2003). A statistical
model for identifying proteins by tandem mass spectrometry. Anal. Chem.
75, 4646–4658.
biochemistry. Biochemistry 33, 5661–5673.
Fisher, W.R., Taniuchi, H., and Anfinsen, C.B. (1973). On the role of heme in
the formation of the structure of cytochrome c. J. Biol. Chem. 248, 3188–
3195.
Palsdottir, H., and Hunte, C. (2004). Lipids in membrane protein structures.
Biochim. Biophys. Acta 1666, 2–18.
Gallet, P.F., Zachowski, A., Julien, R., Fellmann, P., Devaux, P.F., and Maftah,
A. (1999). Transbilayer movement and distribution of spin-labelled phospho-
lipids in the inner mitochondrial membrane. Biochim. Biophys. Acta 1418,
61–70.
Palsdottir, H., Lojero, C.G., Trumpower, B.L., and Hunte, C. (2003). Structure
of the yeast cytochrome bc1 complex with a hydroxyquinone anion Qo site
inhibitor bound. J. Biol. Chem. 278, 31303–31311.
Rijken, P.J., De Kruijff, B., and De Kroon, A.I. (2007). Phosphatidylcholine is
essential for efficient functioning of the mitochondrial glycerol-3-phosphate
dehydrogenase Gut2 in Saccharomyces cerevisiae. Mol. Membr. Biol. 24,
269–281.
Gao, Z., and Bauerlein, E. (1987). Identifying subunits of ATP synthase TF0.F1
in contact with phospholipid head groups, a-subunits are labelled selectively
by a new photoreactive phospholipid designed for hydrophilic photolabelling.
FEBS Lett. 223, 366–370.
Gubbens, J., Vader, P., Damen, J.M., O’Flaherty, M.C., Slijper, M., de Kruijff,
B., and de Kroon, A.I. (2007). Probing the membrane interface-interacting
proteome using photoactivatable lipid cross-linkers. J. Proteome Res. 6,
1951–1962.
Rostovtsev, V.V., Green, L.G., Fokin, V.V., and Sharpless, K.B. (2002). A
stepwise huisgen cycloaddition process: copper(I)-catalyzed regioselective
‘‘ligation’’ of azides and terminal alkynes. Angew. Chem. Int. Ed. Engl. 41,
2596–2599.
Hertmetter, A., Stutz, H., Franzmair, R., and Paltauf, F. (1989). 1-O-Trityl-
sno-glycero-3-phosphocholine: a new intermediate for the facile preparation
of mixed-acid 1,2-diacylglycerophosphocholines. Chem. Phys. Lipids 50,
57–62.
Rouser, G., Fleischer, S., and Yamamoto, A. (1970). Two dimensional thin layer
chromatographic separation of polar lipids and determination of phospholipids
by phosphorus analysis of spots. Lipids 5, 494–496.
Salisbury, C.M., and Cravatt, B.F. (2007). Click chemistry-led advances in high
Hoja, U., Marthol, S., Hofmann, J., Stegner, S., Schulz, R., Meier, S., Greiner,
E., and Schweizer, E. (2004). HFA1 encoding an organelle-specific acetyl-CoA
carboxylase controls mitochondrial fatty acid synthesis in Saccharomyces
cerevisiae. J. Biol. Chem. 279, 21779–21786.
content functional proteomics. QSAR Comb. Sci. 26, 1229–1238.
Santoni, V., Molloy, M., and Rabilloud, T. (2000). Membrane proteins and
proteomics: un amour impossible? Electrophoresis 21, 1054–1070.
Shinzawa-Itoh, K., Aoyama, H., Muramoto, K., Terada, H., Kurauchi, T., Tade-
hara, Y., Yamasaki, A., Sugimura, T., Kurono, S., Tsujimoto, K., et al. (2007).
Structures and physiological roles of 13 integral lipids of bovine heart
cytochrome c oxidase. EMBO J. 26, 1713–1725.
Janssen, M.J., van Voorst, F., Ploeger, G.E., Larsen, P.M., Larsen, M.R., de
Kroon, A.I., and de Kruijff, B. (2002). Photolabeling identifies an interaction
between phosphatidylcholine and glycerol-3-phosphate dehydrogenase
(Gut2p) in yeast mitochondria. Biochemistry 41, 5702–5711.
Speers, A.E., and Cravatt, B.F. (2004). Profiling enzyme activities in vivo using
Jia, L., Dienhart, M., Schramp, M., McCauley, M., Hell, K., and Stuart, R.A.
(2003). Yeast Oxa1 interacts with mitochondrial ribosomes: the importance
of the C-terminal region of Oxa1. EMBO J. 22, 6438–6447.
click chemistry methods. Chem. Biol. 11, 535–546.
Szyrach, G., Ott, M., Bonnefoy, N., Neupert, W., and Herrmann, J.M. (2003).
Ribosome binding to the Oxa1 complex facilitates co-translational protein
insertion in mitochondria. EMBO J. 22, 6448–6457.
Jordi, W., and De Kruijff, B. (1996). Apo- and holocytochrome c-membrane
interactions. In Cytochrome C: A Multidisciplinary Approach, R.A. Scott
and A.G. Mauk, eds. (Sausalito, CA: University Science Books), pp. 449–
472.
Thiele, C., Hannah, M.J., Fahrenholz, F., and Huttner, W.B. (2000). Cholesterol
binds to synaptophysin and is required for biogenesis of synaptic vesicles.
Nat. Cell Biol. 2, 42–49.
Keller, A., Nesvizhskii, A.I., Kolker, E., and Aebersold, R. (2002). Empirical
statistical model to estimate the accuracy of peptide identifications made by
MS/MS and database search. Anal. Chem. 74, 5383–5392.
Tornøe, C.W., Christensen, C., and Meldal, M. (2002). Peptidotriazoles on
solid phase: [1,2,3]-triazoles by regiospecific copper(i)-catalyzed
Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved 13

Chemistry & Biology
Detecting the Lipid-Interacting Proteome
1,3-dipolar cycloadditions of terminal alkynes to azides. J. Org. Chem. 67,
Wang, P., Blank, D.H., and Spencer, T.A. (2004). Synthesis of benzophenone-
3057–3064.
containing analogues of phosphatidylcholine. J. Org. Chem. 69, 2693–2702.
Wallin, E., and von Heijne, G. (1998). Genome-wide analysis of integral
membrane proteins from eubacterial, archaean, and eukaryotic organisms.
Protein Sci. 7, 1029–1038.
Wessel, D., and Flugge, U.I. (1984). A method for the quantitative recovery of
protein in dilute solution in the presence of detergents and lipids. Anal. Bio-
chem. 138, 141–143.
14 Chemistry & Biology 16, 3–14, January 30, 2009 ª2009 Elsevier Ltd All rights reserved