A modular cross-linking approach for exploring protein interactions

Article abstract of DOI:10.1021/ja026917a

A method is described for the elucidation of protein-protein interactions using novel cross-linking reagents and mass spectrometry. The method incorporates (1) a modular solid-phase synthetic strategy for generating the cross-linking reagents, (2) enrichm

Full text of DOI:10.1021/ja026917a

Published on Web 02/06/2003
A Modular Cross-Linking Approach for Exploring Protein
Interactions
Michelle Trester-Zedlitz,†,‡ Katsuhiko Kamada,§ Stephen K. Burley,| David Fenyo¨,
Brian T. Chait,*,‡ and Tom W. Muir*,†
Contribution from the Laboratories of Synthetic Protein Chemistry, Mass Spectrometry and
Gaseous Ion Chemistry, and Molecular Biophysics, Rockefeller UniVersity, 1230 York AVenue,
New York, New York 10021
Received May 15, 2002 ; E-mail: muirt@mail.rockefeller.edu, chait@mail.rockefeller.edu
Abstract: A method is described for the elucidation of protein-protein interactions using novel cross-
linking reagents and mass spectrometry. The method incorporates (1) a modular solid-phase synthetic
strategy for generating the cross-linking reagents, (2) enrichment and digestion of cross-linked proteins
using microconcentrators, (3) mass spectrometric analysis of cross-linked peptides, and (4) comprehensive
computational analysis of the cross-linking data. This integrated approach has been applied to the study
of cross-linking between the components of the heterodimeric protein complex negative cofactor 2.
action methodologies,3-5 chemical cross-linking irreversibly
captures binding partners, so that even transient interactions can,
Introduction
The elucidation of protein interactions is fundamental to
understanding biological processes. Within a cell, a single
protein may interact with many other molecules, and these
interactions may be under both spatial and temporal regulation.
Therefore, cataloguing the interactions of even a single protein
represents an extremely difficult task. This challenge is further
complicated by the fact that protein interactions are often quite
transient in nature or may involve large multiprotein complexes.
Despite these limitations, large-scale “proteomic” efforts are now
under way that aim to provide a complete description of all
protein interaction networks within a given cell.1,2
in principle, be detected. In addition, the covalent nature of a
chemical cross-link enables analytical techniques to be employed
that, under normal circumstances, would disrupt native protein
interactions. Two types of information can be gleaned from
cross-linking studies: the identity of the interacting proteins,
and the regions within those proteins that are involved in the
interaction. The quality of information obtained is dependent
on the accuracy, resolution, and dynamic range of the analytical
strategy used to interrogate the cross-linked system. Tradition-
ally, these readout strategies have relied heavily on chemical
or enzymatic digestion of a cross-linked complex followed by
gel-mobility or chromatographic separations in combination with
radioactive or bioaffinity “tracers”.6,7 Fragments of interest could
then be analyzed further by techniques such as Edman sequenc-
ing.
Modern biological mass spectrometry (MS) provides a
powerful addition to these classical readout approaches.8 MS
is now the method of choice for protein identification and is
seeing increasing application in the area of proteomics, for
example, by allowing the composition of multiprotein cellular
machines to be defined.9 The combination of chemical cross-
linking and mass spectrometry complements these existing
approaches by providing information regarding the architecture
One approach to discovering the identity of interacting pro-
teins is to use chemical cross-linking. Unlike other protein inter-
* To whom correspondence should be addressed. Phone: (212) 327-
7368 (T.W.M.), (212) 327-8849 (B.T.C.). Fax: (212) 327-7358.
† Laboratory of Synthetic Protein Chemistry.
‡ Laboratory of Mass Spectrometry and Gaseous Ion Chemistry.
§ Laboratory of Molecular Biophysics.
| Current address: Structural GenomiX, Inc., 10505 Roselle St., San
Diego, CA 92121.
Current address: Amersham Biosciences, Corp. 800 Centennial Ave.,
Piscataway, NJ 08855.
(1) Gavin, A. C.; Bosche, M.; Krause, R.; Grandi, P.; Marzioch, M.; Bauer,
A.; Schultz, J.; Rick, J. M.; Michon, A. M.; Cruciat, C. M.; Remor, M.;
Hofert, C.; Schelder, M.; Brajenovic, M.; Ruffner, H.; Merino, A.; Klein,
K.; Hudak, M.; Dickson, D.; Rudi, T.; Gnau, V.; Bauch, A.; Bastuck, S.;
Huhse, B.; Leutwein, C.; Heurtier, M. A.; Copley, R. R.; Edelmann, A.;
Querfurth, E.; Rybin, V.; Drewes, G.; Raida, M.; Bouwmeester, T.; Bork,
P.; Seraphin, B.; Kuster, B.; Neubauer, G.; Superti-Furga, G. Nature 2002,
415, 141-147.
(2) Ho, Y.; Gruhler, A.; Heilbut, A.; Bader, G. D.; Moore, L.; Adams, S. L.;
Millar, A.; Taylor, P.; Bennett, K.; Boutilier, K.; Yang, L.; Wolting, C.;
Donaldson, I.; Schandorff, S.; Shewnarane, J.; Vo, M.; Taggart, J.;
Goudreault, M.; Muskat, B.; Alfarano, C.; Dewar, D.; Lin, Z.; Michalickova,
K.; Willems, A. R.; Sassi, H.; Nielsen, P. A.; Rasmussen, K. J.; Andersen,
J. R.; Johansen, L. E.; Hansen, L. H.; Jespersen, H.; Podtelejnikov, A.;
Nielsen, E.; Crawford, J.; Poulsen, V.; Sorensen, B. D.; Matthiesen, J.;
Hendrickson, R. C.; Gleeson, F.; Pawson, T.; Moran, M. F.; Durocher, D.;
Mann, M.; Hogue, C. W.; Figeys, D.; Tyers, M. Nature 2002, 415, 180-
183.
(3) Miyawaki, A.; Llopis, J.; Heim, R.; McCaffery, J. M.; Adams, J. A.; Ikura,
M.; Tsien, R. Y. Nature 1997, 388, 882-887.
(4) Fields, S.; Song, O. Nature 1989, 340, 245-246.
(5) Anderson, N. G. Methods Mol. Biol. 1998, 88, 35-45.
(6) Bickle, T. A.; Hershey, J. W.; Traut, R. R. Proc. Natl. Acad. Sci. U.S.A.
1972, 69, 1327-1331.
(7) Wold, F. Methods Enzymol. 1972, 25, 623-651.
(8) Bennett, K. L.; Kussmann, M.; Bjork, P.; Godzwon, M.; Mikkelsen, M.;
Sorensen, P.; Roepstorff, P. Protein Sci. 2000, 9, 1503-1518 and references
therein.
(9) Mann, M.; Hendrickson, R. C.; Pandey, A. Annu. ReV. Biochem. 2001, 70,
437-473.
9
2416
J. AM. CHEM. SOC. 2003, 125, 2416-2425
10.1021/ja026917a CCC: $25.00 © 2003 American Chemical Society

A Cross-Linking Method To Study Protein Interactions
A R T I C L E S
Scheme 1. Use of a Multidetachable Thioester Linker
Figure 1. Modular cross-linking reagent design.
of large multicomponent complexes.10 Elucidation of higher
resolution structural information, for example, defining specific
binding surfaces, may also be possible in some cases.11
Mass analyses of digested cross-linked proteins are compli-
cated by sample heterogeneity; in most cases samples contain
a complex mixture of modified and unmodified peptides.8
Potential solutions to this problem include procedures that either
enrich the cross-linked products or provide them with a
recognizable mass signature. Several research groups have found
that the inclusion of chemical moieties, such as affinity
handles,12 isotope tags,13,14 and cleavable sites,8 within their
modifying or cross-linking reagent can facilitate MS analyses.
At present, very few cross-linking reagents with all the
aforementioned properties are commercially available.15
We are interested in the development of chemical cross-
linking strategies that employ MS as the readout step. Ideally,
one would like to design a general strategy and a single reagent
that can be used for all cross-linking studies. However, this is
probably unrealistic due to the structural diversity among
proteins; no two proteins will have exactly the same three-
dimensional distribution of reactive side chains. Thus, it is
unlikely that a single reagent will work for every application.
Moreover, it may be useful to probe any given system with a
variety of reagents each with different properties. To address
these needs, we have developed a strategy, the key feature of
which is a rapid, modular approach to the synthesis of cross-
linking reagents. Modularity allows the properties of the reagent
to be systematically varied (Figure 1), making it feasible to
optimize the reagent for each different cross-linking application.
In this paper, we describe the syntheses of five trifunctional
cross-linkers, all of which were designed to be flexible and
soluble in water and to contain a biotin affinity handle. To target
different protein functional groups and enhance MS analyses,
several properties were varied within the cross-linker, specifi-
cally, spacer lengths, chemical reactivities, cleavage sites, and
isotope tags. Although we applied several of our reagents to
the heterodimeric protein complex negative cofactor 2 (NC2)
(the structure of which is known),16 we only provide details on
the most efficient of these, i.e., 4 and 6. The studies of NC2
employed a novel “one-pot” procedure that allowed straight-
forward isolation of cross-linked protein fragments suitable for
MS analysis. Interpretation of the MS data was facilitated by a
specialized computer program which exhaustively compares
experimental masses to all possible17 calculated cross-linked
combinations. Importantly, in this paper we highlight the utility
of mass spectrometry as a continuous feedback tool for
optimizing cross-linking syntheses and activation, and for
following the course of the cross-linking experiment.
Results
Cross-Linker Syntheses. Cross-linking reagents were syn-
thesized using solid-phase chemistry.18 The synthetic strategy
was tailored to be compatible with isotope labels, affinity
handles, and a variety of reactive groups at the extremities of
the reagent. Moreover, the syntheses were designed such that
the end product would be obtained directly from the resin upon
cleavage with nucleophiles (i.e., no global deprotection steps
were required).
The solid-phase approach outlined in Scheme 1 exploits the
chemical versatility of the mercaptopropionamide-PEGA resin
linker.19 An important feature of this system is the acid stability
of the alkyl thioester linkage formed between the resin linker
and the cross-linker. Reagents can thus be prepared on the solid
phase using standard tert-butyl protection schemes. Use of the
PEGA resin is also important to our strategy since it is
compatible with both organic and aqueous solvent systems.20
(10) Rappsilber, J.; Siniossoglou, S.; Hurt, E. C.; Mann, M. Anal. Chem. 2000,
72, 267-275.
(16) Kamada, K.; Shu, F.; Chen, H.; Malik, S.; Stelzer, G.; Roeder, R. G.;
Meisterernst, M.; Burley, S. K. Cell 2001, 106, 71-81.
(17) On the basis of the extent of cross-linking and enzymatic digestion observed
with MS, the user can limit the search in the following ways: (1) maximum
number of fragments cross-linked; (2) maximum number of cross-linker
attachments; (3) maximum number of missed enzymatic cleavage sites;
(4) maximum number of partial global modifications (e.g., Cys modifica-
tions).
(11) Young, M. M.; Tang, N.; Hempel, J. C.; Oshiro, C. M.; Taylor, E. W.;
Kuntz, I. D.; Gibson, B. W.; Dollinger, G. Proc. Natl. Acad. Sci. U.S.A.
2000, 97, 5802-5806.
(12) Alley, S. C.; Ishmael, F. T.; Jones, A. D.; Benkovic, S. J. J. Am. Chem.
Soc. 2000, 122, 6126-6127.
(13) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold,
R. Nat. Biotechnol. 1999, 17, 994-999.
(14) Mueller, D. R.; Schindler, P.; Towbin, H.; Wirth, U.; Voshol, H.; Hoving,
S.; Steinmetz, M. O. Anal. Chem. 2001, 73, 1927-1934.
(15) Hermanson, G. T. Bioconjugate Techniques; Academic Press: New York,
1995.
(18) Merrifield, R. B. J. Am. Chem. Soc. 1963, 85, 2149.
(19) Camarero, J. A.; Cotton, G. J.; Adeva, A.; Muir, T. W. J. Pept. Res. 1998,
51, 303-316.
(20) Meldal, M. Tetrahedron Lett. 1992, 33, 3077-3080.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2417

A R T I C L E S
Trester-Zedlitz et al.
Scheme 2. Synthesis of Amine-Reactive Cross-Linking Reagentsa
a Conditions: (A) (i) TFA; (ii) Boc-Dapa(Fmoc), HBTU, DIEA, DMF; (iii) 1% DBU; (iv) 1, HBTU, DIEA, DMF; (B) (i) TFA; (ii) succinic anhydride,
DIEA, DMF; (C) 1 M NaOH; (D) (i) tributylamine; (ii) sulfo-NHS, PyBOP, DIEA, DMF; (E) (i) TFA; (ii) 3, HBTU, DIEA, DMF; (iii) TFA; (iv) succinic
anhydride, DIEA, DMF; (F) 1 M NaOH; (G) (i) tributylamine; (ii) sulfo-NHS, PyBOP, DIEA, DMF; (H) (i) TFA; (ii) 2, PyBOP, DIEA, DMF; (I) (i) TFA;
(ii) Boc-Gly-OH, HBTU, DIEA, DMF; (iii) TFA; (iv) succinic anhydride, DIEA, DMF; (J) 30 mM Hg(OAc)2, 30% MeCN, 5% glacial acetic acid; (K) (i)
CsCl; (ii) sulfo-NHS, PyBOP, DIEA, DMF.
This greatly expands the range of cleavage conditions that can
be used to remove the final product from the solid support
(Scheme 1). For example, mild aqueous hydrolysis conditions
(Hg(OAc)2, pH 4)21 can be used, thereby allowing the inclusion
of base-labile components in the cross-linking reagents.
NaOH. The solid-phase synthesis of cross-linker 6 employed
Boc-serine derivative 2 in which the â-OH group is esterified
with 1. The presence of the ester group prevented the use of
NaOH in the cleavage step. However, hydrolysis using Hg-
(OAc)2, pH 4, was found to proceed with excellent efficiency
(89%)24 and without detectable cleavage of the ester.
The synthesis of the isotope-tagged trifunctional cross-linking
reagent 8a,b is shown in Scheme 3. This reagent was made in
a stepwise fashion similar to that of 4 with the following excep-
tions: 4-benzoylbenzoic acid was attached to the â-amino group
of Dapa, and the NR group of Dapa was acylated with either
succinic anhydride, 8a, or 2,2′,3,3′-d4-succinic anhydride, 8b.
Nucleophilic cleavage of the reagents from the solid support
was achieved by treatment with biotin derivative 7. The ami-
nolysis step was quantitative as indicated by reversed-phase
HPLC analysis of the crude material.24
Trifunctional reagent 9 was synthesized according to Scheme
4. This reagent was made similarily to reagent 4 except for the
replacement of succinic anhydride with 3-maleimidopropionic
acid. Due to the lability of maleimides at high pH, reagent 9
was released from the support with Hg(OAc)2 at pH 4 with a
cleavage efficiency of 92%.24 Importantly, the maleimide group
was found to be stable to these cleavage conditions as indicated
by HPLC and MS.
To demonstrate the versatility of the mercaptopropionamide-
PEGA system, five different cross-linking reagents were
designed and synthesized. The amine-reactive trifunctional
cross-linking reagents’ (4-6) syntheses are shown in Scheme
2. All three syntheses began with the attachment of Boc-Gly to
the solid support through a thioester linkage. For reagents 4
and 5, Boc-Dapa(Fmoc) was added using standard in situ
activation procedures.22 Selective removal of the Fmoc group
from the â-amino group of the Dapa moiety allowed attachment
of biotin derivative 1. Deprotection of the base-labile Fmoc
protecting group, without unwanted aminolysis of the thioester
linkage, was accomplished using a modified version of the
DBU-based procedure described by Wade and co-workers.23 The
spacer length in reagent 5 was increased by the attachment of
trioxa compound 3 to the Dapa R-amino group. Following NR
deprotection and acylation with succinic anhydride, compounds
4 and 5 were obtained by treatment of the resin with aqueous
(21) Camarero, J. A.; Ayers, B.; Muir, T. W. Biochemistry 1998, 37, 7487-
7495.
(22) Schnolzer, M.; Alewood, P.; Jones, A.; Alewood, D.; Kent, S. B. Int. J.
Pept. Protein Res. 1992, 40, 180-193.
(23) Wade, J. D.; Bedford, J.; Sheppard, R. C.; Tregear, G. W. Pept. Res. 1991,
4, 194-199.
(24) Cleavage supernatants were analyzed by reversed-phase HPLC, and relevant
peak areas were compared with those obtained by subsequent NaOH
cleavage of the same beads.
9
2418 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

A Cross-Linking Method To Study Protein Interactions
A R T I C L E S
Scheme 3. Synthesis of a Cross-Linking Reagent with an Isotope
Scheme 4. Synthesis of a Sulfhydryl- and Amine-Reactive
Taga
Cross-Linking Reagenta
a Conditions: (P) (i) TFA; (ii) 3-maleimidopropionic acid, PyBOP, DIEA,
DMF; (Q) 30 mM Hg(OAc)2, 30% MeCN, 5% glacial acetic acid; (R) sulfo-
NHS, PyBOP, DIEA, DMF.
a Conditions: (L) (i) 1% DBU, DMF; (ii) 10% DIEA; (iii) 4-benzoyl-
benzoic acid, DIC, DMF; (M) (i) TFA; (ii) succinic anhydride or 2,2′,3,3′-
d4-succinic anhydride, DIEA, DMF; (N) 7, DMF, DIEA; (O) (i) tributy-
lamine, (ii) sulfo-NHS, PyBOP, DIEA, DMF.
protein (TBP) and promoter DNA has recently been determined
at 2.6 Å resolution.16 As a model system for our cross-linking
studies, we chose a related NC2 complex, incorporating residues
11-95 of the R subunit and 7-99 of the â subunit, which was
used in early efforts to crystallize the complex. Although crystals
of this latter form were not obtained, biochemical analyses
indicated that the two polypeptides form a tight complex in
solution (data not shown), which we anticipated would be
closely related to the corresponding NC2 component of the
NC2-TBP-DNA crystal structure.16
The truncated NC2 Râ heterodimer was combined with a
∼70-fold molar excess of activated 4, 6, and 8a on ice at pH 7
(Figure 3A (i)). These conditions were empirically selected on
the basis of a desire to attain single-hit conditions with the
highest yield of cross-linked heterodimer. SDS-PAGE and mass
spectrometry results of 8a with NC2 indicated that this cross-
linker was less efficient and yielded reaction products that were
significantly more difficult to analyze than that with 4 or 6.
Therefore, we will focus only on results using the latter re-
agents.
Aliquots were taken at four different time points for analyses
by SDS-PAGE (Figure 3B) and MALDI-TOF MS (Figure 3C).
Each time point was immediately quenched with 2-mercapto-
ethanol and then acidified. The SDS-PAGE analysis in Figure
3B provides a quantitative assessment of the NC2 cross-linking
time course. At just 5 min, a gel shift, composed of one Râ
complex covalently tethered with reagent 4 or 6, was observed.
Estimated cross-linking efficiencies of the time course (5-60
min) experiment ranged from 10% to 25%. MS analysis in
Figure 3C provides a qualitative assessment of the extent of
cross-linking within the Râ complex. The mass spectra show
that, by 5 min, a portion of the R and â monomers have become
Cross-Linker Activation. In preliminary studies, we explored
a variety of conditions for activation of carboxyl groups within
our cross-linking reagents. These strategies included the use of
DIC with NHS, EDC with sulfo-NHS, and PyBOP with both
NHS and sulfo-NHS. The efficiency of these activation proce-
dures was followed using an arginine trapping procedure which
allowed for straightforward MS analysis of the acylated
products. None of these activation procedures gave high yields
of the desired products when the free acid form of the cross-
linking reagents was used. A typical result is shown in Figure
2A in which PyBOP and sulfo-NHS were used in the activation
of reagent 4. MS analysis indicated that the major product
contained a single arginine addition, and tandem MS analysis
further revealed that this acylation had selectively taken place
on the glycine carboxyl group (Figure 2C,E). As shown in
Figure 2B, the yield of the diacylated adduct was greatly
improved when a tributylammonium or cesium salt of the diacid
was employed in the activation step. Presumably, salt formation
improved the nucleophilicity of the succinic acid carboxyl group
relative to the glycyl carboxyl group.25
Cross-Linking of the NC2 Heterodimer. NC2 is a protein
heterodimer (consisting of R and â subunits) that functions as
a negative regulator of basal transcription by blocking the
assembly of the preinitiation complex. The X-ray structure of
an NC2 construct, incorporating residues 1-77 of the R subunit
and 1-176 of the â subunit, in complex with TATA box binding
(25) Dijkstra, G.; Kruizinga, W. H.; Kellogg, R. M. J. Org. Chem. 1987, 52,
4230-4234.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2419

A R T I C L E S
Trester-Zedlitz et al.
Figure 2. Trapping the activated cross-linking reagent with arginine: (A) MALDI-TOF MS of Arg-quenched sulfo-NHS-activated 4; (B) MALDI-TOF MS
of the Arg-quenched sulfo-NHS-activated tributylamine salt of 4; (C) MALDI-QqTOF MS/MS of 4′ ) [4 + Arg]; (D) MALDI-QqTOF MS/MS of 4′′ ) [4
+ 2 Arg]; (E) proposed cleavages inferred from MS/MS data of 4′ ) [4 + Arg]; (F) proposed cleavages inferred from MS/MS data of 4′′ ) [4 + 2 Arg].
tethered together with one cross-link. And by 60 min, up to
three cross-links were evident.26
Analysis of Cross-Linked Proteins. To isolate cross-linked
protein fragments suitable for MS analysis, we devised the
procedure that is summarized in Figure 3A. Importantly,
microconcentrators were used as a porous surface to facilitate
excess cross-linker removal, retention of cross-linked protein
complexes, easy exchange of experimental conditions, and the
potential for size exclusion.27 MS provided a quick and
informative tool for assessing each step of the process. Upon
searching for combinations of cross-linked protein digest
fragments, it became apparent that we were looking for needles
in a haystack. To locate and identify these “needles”, we utilized
(26) Observation of the cross-linked product requires that only one of the three
be intermolecular. The other two links can be either inter- or intramolecular.
(27) Note: Some sample loss on filters should be anticipated.
9
2420 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

A Cross-Linking Method To Study Protein Interactions
A R T I C L E S
Figure 3. Procedure and analyses of cross-linking the NC2 complex. (A) Cross-linking procedure: (i) cross-link the protein complex; (ii) add the protein
complex to the microconcentrator; (iii) wash away excess cross-linking reagent; (iv) reduce, alkylate, and then apply the endoproteinase(s); (v) add neutravidin
resin (grey) and transfer to the spin column; (vi) wash away un-cross-linked digested protein fragments; (vii) cleave with NH2OH and elute cross-linked
protein fragments. (B) SDS-PAGE (Tris-glycine, 4-20%) of the cross-linking time course with 4 and the NC2 heterodimer. (C) MALDI-TOF spectra of
the cross-linking time course with reagent 4 and the NC2 heterodimer (Râ). Note: x, cross-linking reagent that attaches at both ends; y, cross-linking reagent
at only one end. (D) MALDI-QqTOF of the Asp-N and trypsin digest of the NC2 heterodimer with reagent 6. (E) MALDI-QqTOF of the Asp-N and trypsin
digest post affinity purification with neutravidin beads and base cleavage of reagent 6.
a biotin-avidin affinity purification strategy in combination with
high-accuracy MS (MALDI-QqTOF) and a program (Pepti-
deMap) for exhaustive identification of cross-linked peptides.
Biotin affinity handles incorporated into our cross-linking
reagents enabled us to selectively extract cross-linked proteolytic
fragments and thereby reduce the complexity of the mass
spectra. To bring the proteolytic fragments into the mass range
of the MALDI-QqTOF (m/z 4000), sequential enzymatic diges-
tions were used. MALDI was selected for the ease of sample
preparation, while the QqTOF mass analyzer was selected for
its high mass accuracy (5-50 ppm). High mass accuracy is
critical in reducing the number of cross-link assignments at a
given nominal m/z value.
information about interactions between different polypeptide
chains. Intramolecular cross-links refer to connections made
within a single polypeptide chain. Dead-end cross-links do not
provide connectivity information but do provide information
about the solvent accessibility of specific amino acid residues.
Figure 4 provides a summary of these three cross-link types
observed after enzymatic treatment (Glu-C, Asp-N, Asp-N/
trypsin) of the NC2 complex with reagent 6. Confident
assignments of the cross-linked peptides were made when the
measured mass matched a search hit unambiguously. For the
most abundantly cross-linked proteolysis fragments a correlative
mass shift (indicated with the arrows connecting Figure 3D to
Figure 3E) was used as an additional constraint in identifying
cross-linked protein fragments. When reagent 6 is used, this
correlative mass shift arises from the loss of the cleaved biotin
Three types of cross-links were identified: intermolecular,
intramolecular, and dead-end. Intermolecular cross-links provide
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2421

A R T I C L E S
Trester-Zedlitz et al.
Figure 4. Summary of cross-links: (A) data points identified in the cross-linking experiment of the NC2 heterodimer with reagent 6; (B) intermolecular
cross-link transposed upon the crystal structure. In the space-filling model, NC2 R is shown in blue, NC2 â is shown in purple, Lys residues are shown in
yellow, N-termini are shown in orange, and the observed cross-link of NC2 R with Lys20/23 of NC2 â is indicated in red.
Discussion
arm (i.e., 456 Da). Additional information was obtained by
comparing the appearance, reduction, or disappearance of mass
signals prior to and after cross-linking.
We have demonstrated a facile, modular strategy for synthe-
sizing cross-linking reagents on the solid phase. With the various
building blocks in hand (1-3, 7), it proved feasible to synthesize
several cross-linking reagents in parallel in a single day.
Modularity was demonstrated through the synthesis of five
cross-linking reagents that included two or more of the following
groups: an amine-reactive sulfo-NHS ester, a sulfhydryl-reactive
maleimide, a photochemically reactive benzophenone, an isotope
tag, a biotin affinity handle, and/or a base-labile ester cleavage
site. All of the aforementioned reagents are water soluble and
have attributes that improve the detectability of cross-linker
containing peptides by MS. The versatility of the present
synthetic procedure allows ready incorporation of other reactive
groups, cleavable sites, and/or affinity handles. Thus, the
researcher can optimize cross-linking reagents to fit the protein
complex of interest, the question(s) of interest, and the ac-
companying analytical techniques.
To maximize MS coverage of NC2 cross-linked with the
amine-reactive reagent 6, several endoproteases were applied
(i.e., Asp-N, Glu-C, and the combination of Asp-N followed
by trypsin). A cross-link was identified between N-terminal
residues, R (6-16) and â (6-24), using Glu-C, each containing
three cross-linker reactive sites (i.e., NR, Lys18, and Lys19 or
Nâ, Lys20, and Lys 23). Upon digesting cross-linked NC2 with
both Asp-N and trypsin, we were able to reduce the site of the
cross-link to the N-terminus of R (fragment residues 6-16) with
either Lys20 or Lys23 of â (fragment residues 16-24). The
observed mass of this product was 2324.15 Da (calcd 2324.24
Da). Additional information observed before and after cross-
linking included the reduction and disappearance of signals,
respectively, of the unmodified R (6-16) and â (16-24)
components. Formation of dead-end cross-links within these
individual R and â components was also observed.
The procedure was applied to a study of the transcription
repressor NC2. The NC2 R and â components rapidly cross-
linked, as predicted from the related crystal structure.16 This
result was confirmed both qualitatively and quantitatively, by
mass spectrometry and SDS-PAGE. After enzymatic digestion,
MS analysis revealed an intermolecular interaction between the
N-terminal regions of NC2 R and â, as well as three intramo-
lecular and several dead-end cross-links. The present results
using the truncated NC2 construct are consistent with the X-ray
After affinity purification, the mass spectra included large
signals identified as proteolytic peptides that contained dead-
end cross-linker modifications, and small signals for the intra-
and intermolecular cross-links (Figure 3E). All proteolytic
fragments within NC2 that contained modifiable lysine residues
were observed by MS. However, signal intensities varied
strongly from one peptide fragment to the next. It is noteworthy
that Lys 84 of NC2 R was not observed with a dead-end cross-
linker modification (Figure 4).
9
2422 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

A Cross-Linking Method To Study Protein Interactions
A R T I C L E S
MALDI-MS was performed on either a MALDI time-of-flight mass
spectrometer Voyager-DE STR (PE Biosystem, Foster City, CA) or a
modified Sciex prototype QqTOF (Centaur).29 These instruments are
equipped with nitrogen lasers delivering pulses of ultraviolet light
(wavelength 337 nm) at, respectively, 3 and 20 Hz to the matrix spot.
Acquisitions by the MALDI time-of-flight mass spectrometer Voyager-
DE STR were made in both linear and reflectron, delayed extraction
modes. MALDI-TOF data were smoothed and calibrated using Data
Explorer (PE Biosystem). All other spectra were smoothed, calibrated,
and analyzed using the program M-over-Z (Genomic Solutions, Ann
Arbor, MI). The MALDI matrixes used were 4-hydroxy-R-cyanocin-
namic acid (4HCCA) and 2,5-dihydroxybenzoic acid (DHB). ESI-MS
was performed on a commercial ion-trap mass spectrometer (Thermo
Finnigan LCQ-DECA, San Jose, CA).
crystallographic findings of the ternary NC2-TBP-DNA
complex (Figure 4).
Other interactions that might be predicted from the crystal
structure but were not observed include Lys29 of R with Lys63/
64 of â and Lys18/19 of R with the N-terminus of â. Possible
reasons why these interactions were not observed include local
environmental factors that may reduce reaction rates, suboptimal
reaction conditions, or low MS response to certain cross-linked
peptides. Failure to observe a cross-link should thus be
interpreted with caution. In such cases, it may prove worthwhile
to consider the application of different cross-linking reagents
and/or conditions. However, in this study two different reagents
were applied, giving the same result.
Synthesis. (a) N-Boc-3,6-dioxaoctane-1,8-diamine. A solution of
di-tert-butyl dicarbonate (di-Boc) (12 g, 55 mmol, 0.5 equiv) in DCM
(274 mL) was added dropwise to a mixture of tris(ethylene glycol)-
1,8-diamine (15.2 g, 103 mmol, 1.0 equiv) and DIEA (10 mL, 57 mmol,
1.0 equiv) at room temperature over ∼45 min. The reaction was stirred
for 2 h more and concentrated in vacuo. Purification by flash
chromatography using 0.8:0.5:8.7 methanol/propylamine/DCM as the
eluent gave 6.9 g (50%) of product as an oil: 1H NMR (400 MHz,
CD3OD) δ 3.6 (s, 4H), 3.54 (t, 2H), 3.53 (t, 2H), 3.24 (t, 2H), 2.8 (t,
2H), 1.4 (s, 9H); ESI-MS (MH+) m/z calcd for C11H25N2O4 249.3, found
249.0
Advantages of the present procedure include the control
afforded by the facile, modular synthetic strategy, rapid feedback
using MS data at every stage, and straightforward interpretation
of the resulting MS data using PeptideMap. The utilization of
microconcentrators to digest cross-linked protein complexes
circumvents the use of SDS-PAGE, which can exhibit smearing
of cross-linked products over extended areas of the gel.
Application of this integrated chemical cross-linking procedure
has the potential to reveal direct protein interactions within large
protein complexes and to differentiate between their confor-
mationally distinct functional states. Finally, the use of solid-
phase synthesis procedures will allow for the preparation of
focused cross-linking reagent libraries. In principle, this should
provide a means to optimize cross-linking studies for any given
system of interest.
(b) N-Boc-N′-biotinyl-3,6-dioxaoctane-1,8-diamine. A solution of
biotin (4.4 g, 18 mmol, 1.5 equiv), HBTU (6 g, 16 mmol, 1.3 equiv),
and DIEA (3.5 mL, 20 mmol, 1.6 equiv) in DMF (332 mL) was stirred
for 10 min at room temperature before being adding dropwise to a
solution of N-Boc-3,6-dioxaoctane-1,8-diamine (3 g, 12 mmol, 1.0
equiv). The reaction was stirred for 1 h at room temperature, after which
the DMF was removed in vacuo to give an oil. The crude product was
purified by flash chromatography using 0.8:0.5:8.7 methanol/propy-
lamine/DCM as the eluent. The reaction was quantitative with a yield
of 5.8 g: 1H NMR (400 MHz, CD3OD) δ 4.5 (m, 1H), 4.3 (m, 1H),
3.6 (s, 4H), 3.54 (tt, 4H), 3.39 (t, 2H), 3.26 (t, 2H), 2.9 (dd, 1H), 2.7
(d, 1H), 2.2 (t, 2H), 1.7-1.5 (m, 8H), 1.4 (s, 9H); MALDI-QqTOF
(M + Na+) m/z calcd for C21H38N4O6SNa 497.241, found 497.251.
(c) N-Biotinyl-3,6-dioxaoctane-1,8-diamine (7). N-Boc-N′-biotinyl-
3,6-dioxaoctane-1,8-diamine (5.8 g, 12 mmol) was dissolved in 50 mL
of neat TFA containing 0.5 mL of anisole and stirred for 1 h at room
temperature. The TFA was evaporated under reduced pressure to give
an oil which was then mixed with toluene and reevaporated to remove
any residual TFA. Analytical HPLC and ESI-MS indicated the
deprotection reaction was quantitative, so the crude material was used
without further purification: ESI-MS (MH+) m/z calcd for C16H31N4O4S
375.5, found 375.0.
Experimental Section28
Materials. Boc-amino acids, PyBOP, and HBTU were obtained from
Novabiochem (San Diego, CA). PEGA resin was purchased from
Polymer Laboratories Inc. (Amherst, MA). DMF (synthesis grade) and
acetonitrile (HPLC grade) were purchased from Fisher (Pittsburgh, PA).
Trifluoroacetic acid was purchased from Halocarbon (River Edge, NJ).
All other reagents were obtained from Aldrich Chemical Co. (Mil-
waukee, WI). Sequencing grade endoproteases Glu-C (Staphylococcus
aureus V8), Asp-N (Pseudomonas fragi), and modified bovine trypsin
were purchased from Roche (Basel, Switzerland). Flash chromatography
was performed on silica gel, 200-400 mesh (Merck).
Reversed-Phase HPLC. Analytical gradient HPLC was performed
on a Hewlett-Packard 1100 series instrument with UV detection using
a Vydac C18 column (5 µm, 4.6 × 150 mm) at a flow rate of 1 mL/
min. Semipreparative and preparative gradient HPLC were performed
on a Waters DeltaPrep 4000 system fitted with a Waters 486 tunable
absorbance detector. Semipreparative HPLC was run on a Vydac C18
column (10 µm, 10 × 250 mm) at a flow rate of 5 mL/min. Preparative
HPLC was performed on a Vydac C18 column (15-20 µm, 50 × 250
mm) at a flow rate of 30 mL/min. All runs used linear gradients of
0.1% aqueous TFA (solvent A) vs 90% acetonitrile, 10% water plus
0.1% TFA (solvent B). Low-pressure C18 columns (Michrom BioRe-
sources, Inc., Auburn, CA) were utilized for manual cleanup of the
cross-linked samples.
(d) N-Succinimidyl-N′-biotinyl-3,6-dioxaoctane-1,8-diamine (1).
A solution containing succinic anhydride (2 g, 20 mmol, 1.7 equiv)
and DIEA (10 mL, 57 mmol, 4.7 equiv) in MeCN (10 mL) was added
to amine 7 (5.8 g, 12 mmol, 1.0 equiv). The reaction was allowed to
proceed for around 1 h, at which point the white precipitate was
dissolved and acidified with 0.1% aqueous TFA. Purification by
preparative HPLC followed by lyophilization gave the desired product
1 (2.7 g, yield 48%): 1H NMR (400 MHz, CD3OD) δ 4.5 (m, 1H), 4.3
(m, 1H), 3.6 (s, 4H), 3.55 (tt, 4H), 3.37 (tt, 4H), 3.21 (m, 1H), 2.93
(dd, 1H), 2.7 (d, 1H), 2.53 (m, 4H), 2.2 (t, 2H), 1.66 (m, 6H), 1.5 (m,
2H); 13C NMR (400 MHz, CD3OD) δ 177.3, 176.8, 175.8, 175.6, 71.4,
70.7, 63.4, 61.7, 41.1, 41.0, 40.4, 40.3, 36.8, 31.0, 29.0, 27.0, 26.7;
ESI-MS (MH+) m/z calcd for C20H35N4O7S 475.2, found 475.3.
Mass Spectrometry. MALDI-MS and tandem MS analyses were
applied routinely to characterize all synthetic cross-linking reagents.
(28) Abbreviations: Boc, tert-butyloxycarbonyl; di-Boc, di-tert-butyl dicarbon-
ate; DIEA, N,N-diisopropylethylamine; EDC, N-ethyl-N′-(3-dimethylami-
nopropyl)carbodiimide; HBTU, 2-[1H-benzotriazolyl]-1,1,3,3-tetramethyl-
uroniumhexafluorophosphate;HPLC,high-performanceliquidchromatography;
MALDI, matrix-assisted laser desorption ionization; NHS, N-hydroxysuc-
cinimide; PEGA, poly(ethylene glycol)-poly(N,N-dimethylacrylamide);
PyBOP, (benzotriazol-1-yloxy)tripyrrolidinophosphonium hexafluorophos-
phate; SPS, solid-phase synthesis; sulfo-NHS, sulfo-N-hydroxysuccinimide;
TDA, 4,7,10-trioxatridecane-11,13-diamine. Standard IUPAC single and
triple letter codes for amino acids are used throughout.
(e) N′-Boc-O-(N-succinimidyl-N′-biotinyl-3,6-dioxaoctane-1,8-di-
amine)serine (2). A mixture containing 1 (238 mg, 0.50 mmol, 1.0
equiv), HATU (171 mg, 0.4 mmol, 0.9 equiv), and DIEA (0.1 mL, 0.8
(29) Krutchinsky, A. N.; Zhang, W.; Chait, B. T. J. Am. Soc. Mass Spectrom.
2000, 11, 493-504.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2423

A R T I C L E S
Trester-Zedlitz et al.
mmol, 1.5 equiv) in DMF (23.4 mL) was stirred for 10 min at room
temperature before being added to a solution of Boc-Ser-OH (1 g, 5
mmol, 10.0 equiv) in DMF (4.2 mL) containing a catalytic amount of
DMAP. The esterification reaction was allowed to proceed at 4 °C for
5 h, at which point the DMF was removed in vacuo. The resulting oil
was dissolved in 1:9 acetonitrile/water containing 0.1% TFA and
purified by preparative scale HPLC to give 2 as a white solid (114.9
mg, yield 40%): 1H NMR (400 MHz, CD3OD) δ 4.5 (m, 1H), 4.4 (s,
2H), 4.3 (m, 1H), 3.6 (s, 4H), 3.57 (t, 4H), 3.37 (t, 4H), 3.21 (m, 1H),
2.97 (dd, 1H), 2.7 (d, 1H), 2.6 (m, 4H), 2.2 (t, 2H), 1.66 (m, 6H), 1.45
(m, 2H), 1.45(s, 9H); 13C NMR (400 MHz, CD3OD) δ 175.7, 175.1,
173.4, 173.2, 79.9, 71.1, 70.4, 63.4, 61.6, 54.1, 40.7, 40.3, 40.2, 36.6,
31.2, 30.5, 28.7, 28.6, 28.6, 26.4; MALDI-QqTOF (MH+) m/z calcd
for C28H48N5O11S 662.307, found 662.301.
3.6, (tt, 4H), 3.55 (t,4H), 3.49 (dd, 2H) 3.35 (tt, 4H), 3.35 (m, 1H), 3.2
(tt, 4H), 2.9 (dd, 1H), 2.7 (d, 1H), 2.59 (t, 2H), 2.55 (t, 2H), 2.45 (s,
4H), 2.2 (t, 2H), 1.75 (tt, 4H), 1.66 (m, 6H), 1.59 (m, 2H), 1.55 (m,
2H), 1.33 (m, 2H); MALDI-QqTOF (MH+) m/z calcd for C43H74N9O17S
1020.493, found 1020.493.
Data for compound 6: 1H NMR (400 MHz, CD3OD) δ 4.75 (t,
1H), 4.5 (dd, 1H), 4.4 (dd, 1H), 4.37 (d, 1H), 4.32 (m, 1H), 3.9 (s,
4H), 3.63 (s, 4H), 3.55 (t, 4H), 3.37 (t, 4H), 3.21 (m, 1H), 2.93 (dd,
1H), 2.71 (d, 1H), 2.64 (dd, 4H), 2.54 (m, 4H), 2.2 (t, 2H), 1.66 (m,
6H), 1.45 (m, 2H); 13C NMR (400 MHz, CD3OD) δ 175.3, 175.1, 175.0,
173, 171.8, 171.6, 171.2, 170.8, 73.0, 71.4, 70.4, 69.7, 64.6, 64.4, 63.1,
62.2, 61.7, 60.7, 56.9, 56.4, 56.2, 55.8, 53.6, 43.8, 41.8, 41.0, 40.9,
40.3, 39.5, 36.7, 35.8, 31.2, 30.5, 26.7, 29.7, 29.8, 29.5, 28.9, 28.6,
25.8; MALDI-QqTOF (MH+) m/z calcd for C31H50N7O14S 777.322,
found 777.322.
Data for compound 8a: 1H NMR (400 MHz, DMSO) δ 8.6 (t,
1H), 8.3 (t, 1H), 8.2 (d, 1H), 7.98 (d, 2H), 7.81 (d, 2H), 7.7 (m, 3H),
7.58 (d, 2H), 6.4 (m, 2H), 4.45 (m, 1H), 4.29 (t, 1H), 4.12 (m, 1H),
3.7 (s, 2H), 3.6 (dd, 2H), 3.5 (s, 4H), 3.38 (tt, 4H), 3.18 (tt, 4H), 3.07
(m, 1H), 2.8 (dd, 1H), 2.79 (d, 1H), 2.6 (d, 1H), 2.4 (m, 4H), 2.1 (t,
2H), 1.5 (m, 4H), 1.3 (m, 2H); 13C NMR (400 MHz, CD3OD) δ 196.3,
172.9, 170.1, 169.7, 166.9, 139.9, 138.4, 137.4, 133.8, 132.5, 129.3,
129.2, 128.3, 127.2, 70.4, 60.9, 59.1, 55.2, 42.1, 40.8, 39.5, 38.3, 34.9,
29.3, 27.9, 25.9; MALDI-QqTOF (MH+) m/z calcd for C39H52N7O11S
826.345, found 826.345.
Data for compound 8b: 1H NMR (400 MHz, DMSO) δ 8.6 (t,
1H), 8.3 (t, 1H), 8.2 (d, 1H), 7.98 (d, 2H), 7.81 (d, 2H), 7.7 (m, 3H),
7.58 (d, 2H), 6.4 (m, 2H), 4.45 (m, 1H), 4.29 (t, 1H), 4.12(m, 1H), 3.7
(s, 2H), 3.6 (dd, 2H), 3.5 (s, 4H), 3.38 (tt, 4H), 3.18 (tt, 4H), 3.07 (m,
1H), 2.8 (dd, 1H), 2.6 (d, 1H), 2.1 (t, 2H), 1.5 (m, 4H), 1.3 (m, 2H);
13C NMR (400 MHz, CD3OD) δ 196.3, 172.9, 170.1, 169.7, 166.9,
139.9, 138.4, 137.4, 133.8, 132.5, 129.3, 129.2, 128.3, 127.2, 69.0,
60.9, 59.1, 55.2, 42.1, 40.8, 39.5,38.3, 34.9, 27.9, 25.0; MALDI-QqTOF
(MH+) m/z calcd for C39H48D4N7O11S 830.370, found 830.371.
Data for compound 9: 1H NMR (400 MHz, DMSO) δ 8.20 (m,-
1H), 8.1 (m,1H), 7.9 (m, 1H), 7.8 (m, 1H), 7.6 (s,1H), 7.0 (s, 2H), 6.4
(s, 1H), 6.3 (s, 1H), 4.3 (m, 2H), 4.13 (m, 1H), 3.62 (dd, 1H), 3.6 (m,
2H), 3.49 (s, 4H), 3.39 (tt, 4H), 3.19 (tt, 4H), 3.09 (m, 1H), 2.79 (d,
1H), 2.87 (dd, 1H), 2.5 (m, 6H), 2.07 (t, 2H), 1.57 (m, 2H), 1.5 (m,
2H), 1.28 (m, 2H); 13C NMR (400 MHz, CD3OD) δ 175.1, 173.4, 173,
171.4, 171.2, 170.8, 170.6, 134, 69.2, 68.9, 60.9, 59.0, 55.4, 52.3, 41.0,
40.9, 40.6, 38.3, 34.9, 33.7, 32.9, 31.9, 28.0, 25.0; MALDI-QqTOF
(MH+) m/z calcd for C32H49N8O12S 769.319, found 769.307.
Preparation of the NC2 Heterodimer. A truncated form of the
human NC2 heterodimeric complex was obtained by coexpression of
the R (residues 11-95)31 and â (residues 7-99)32 subunits in Escheri-
chia coli (BL21[DE3]) using the expression vectors pGEX-6p-1
(Pharmacia) and pET29a (Novagen), respectively. Following cell lysis,
the complex was purified from the soluble fraction by glutathione-
Sepharose chromatography. After cleavage of the GST fusion using
PreScission Protease (Pharmacia), the complex was further purified by
SP-Sepharose ion exchange chromatography essentially as described.16
The measured molecular weights of NC2 R, 9954 (calcd 9956.7), and
NC2 â, 10460 (calcd 10462.0), confirmed that the complex was neither
truncated nor posttranslationally modified during expression or purifica-
tion. Gel filtration results using a Sephadex 75 column (Pharmacia)
confirmed that the two subunits formed the expected heterodimer in
solution (data not shown). The protein complex was stored at -65 °C
as a 3 mg/mL stock solution in 10% glycerol, 200 mM NaCl, 1 mM
DTT, 20 mM Hepes, pH 7, and 0.1 mM PMSF.
(f) N-Boc-N′-succinimidyl-4,7,10-trioxatridecane-1,13-diamine (3).
Di-Boc (1.75 g, 8 mmol, 0.5 equiv) in DCM (30 mL) was added
dropwise to a solution of 4,7,10-trioxatridecane-1,13-diamine (3.52 g,
16 mmol, 1.0 equiv) and triethylamine (1.2 mL, 8.8 mmol, 0.6 equiv)
in DCM (40 mL). The reaction was stirred for 4 h at room temperature
and then concentrated in vacuo. Purification by flash chromatography
using 0.5:1:8.5 propylamine/methanol/DCM as the eluent gave 1.8 g
(75%) of the product N-Boc-4,7,10-trioxatridecane-1,13-diamine (Boc-
TDA): 1H NMR (400 MHz, CD3OD) δ 3.62 (m, 4H), 3.59 (m, 4H),
3.5 (m, 4H), 3.1 (t, 2H), 2.7 (t, 2H), 1.8-1.7 (m, 4H), 1.5 (s, 9H).
Boc-TDA (1.1 g, 3.6 mmol, 1.0 equiv) was combined with succinic
anhydride (428 mg, 4.3 mmol, 1.2 equiv) in MeCN (4 mL), pyridine
(4 mL), and DCM (4 mL). The reaction was stirred for 3.5 h at room
temperature before being concentrated in vacuo. Purification by flash
chromatography using 0.01:1:9 acetic acid/methanol/DCM as the eluent
afforded 1.3 g (91%) of the desired product. Note, trace amounts of
AcOH were removed by repeated evaporation from cyclohexane (six
times) followed by lyophilization: 1H NMR (400 MHz, CD3OD) δ
3.65 (m, 4H), 3.60 (m, 4H), 3.54 (tt, 4H), 3.28 (t, 2H), 3.14 (t, 2H),
2.59 (t, 2H), 2.48 (t, 2H), 1.8-1.7 (m, 4H), 1.45 (s, 9H); 13C NMR
(400 MHz, CD3OD) δ 175.3, 174.2, 173.9, 158.1, 70.5, 70.4, 68.9,
37.6, 37.0, 30.7, 29.8, 29.4, 27.9; MALDI-QqTOF (M + Na+) m/z
calcd for C19H36N2O8Na 443.237, found 443.236.
Solid-Phase Synthesis. All cross-linking reagents were synthesized
manually according to the in situ neutralization/HBTU activation
protocol for Boc solid-phase peptide synthesis.22 Coupling yields were
monitored by the quantitative ninhydrin determination of residual free
amine.30 All compounds were prepared on a Boc-Gly-[COS]-PEGA
support.20 The â-amino group of Dapa was orthogonally protected with
the Fmoc group, which was selectively removed by treatment with 1%
DBU in DMF for 1 min. Following chain assembly, cross-linking
reagents were cleaved from the support by base hydrolysis (1 M NaOH,
5-20 min, room temperature), acid hydrolysis (30 mM Hg(OAc)2 in
3:7 MeCN/water containing 5% acetic acid, 18 h, room temperature),
or aminolysis (60 mM 7, 195 mM thiophenol, 90 mM DIEA in DMF,
overnight, room temperature). In each case, dried resin was treated with
the cleavage cocktail, and the beads were filtered and washed with
0.1% aqueous TFA (buffer A), or 1:1 MeCN/water and 0.1% TFA.
All compounds were purified by preparative HPLC.
Data for compound 4: 1H NMR (400 MHz, D2O) δ 4.57 (m, 1H),
4.5 (m, 1H), 4.35 (dd, 1H), 3.9 (s, 2H), 3.63 (s, 4H), 3.6 (tt, 4H), 3.49
(dd, 2H), 3.35 (m, 1H), 3.35 (tt, 4H), 2.9 (dd, 1H), 2.7 (d, 1H), 2.62
(tt, 2H), 2.6 (tt, 2H), 2.5 (s, 4H), 2.2 (t, 2H), 1.59 (m, 2H), 1.55 (m,
2H), 1.35 (m, 2H); 13C NMR (400 MHz, CD3OD) δ 177.6, 177.2, 175.6,
175.1, 174.8, 172.1, 70.0, 69.8, 62.7, 60.9, 55.8, 53.8, 42.6, 40.3, 39.3,
39.0, 36.1, 31.6, 30.0, 28.6, 28.3, 25.5; MALDI-QqTOF (MH+) m/z
calcd for C29H48N7O12S 718.308, found 718.307.
Chemical Cross-Linking. (a) Activation of Cross-Linking Re-
agents. Tributylammonium or cesium salts of the cross-linking reagents
Data for compound 5: 1H NMR (400 MHz, D2O) δ 4.57 (m, 1H),
4.51 (m, 1H), 4.35 (dd, 1H), 3.81 (s, 4H), 3.65 (s, 8H), 3.63 (s, 4H),
(31) The truncated NC2 R protein contains residues 11-95 of the native protein
in addition to an N-terminal leader sequence (GPLGS).
(30) Sarin, V. K.; Kent, S. B.; Tam, J. P.; Merrifield, R. B. Anal. Biochem.
1981, 117, 147-157.
(32) The truncated NC2 â protein contains residues 7-99 of the native protein
in addition to a Met at its N-terminus.
9
2424 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

A Cross-Linking Method To Study Protein Interactions
A R T I C L E S
were prepared using a cation exchange step (CM-50, Sigma). In a
typical activation procedure, the corresponding cross-linker salt (25-
50 µg, 1 equiv) was dissolved in DMF (10 µL) along with sulfo-NHS
(3 equiv), DIEA (2 equiv), and PyBOP (2 equiv). The activation was
allowed to proceed on ice for 10 min, at which point the solvent was
removed with a stream of argon gas using hexane as the azeotrope.
(b) Cross-Linking Proteins with Activated Reagents. Buffered
protein (49 µM NC2 R, â) was immediately added to dried activated
cross-linking reagent (60-300 equiv, 4 °C). At various time points,
aliquots of the cross-linking reaction were taken for MALDI-TOF (50-
100 ng of NC2 Râ), SDS-PAGE (1-2 µg of NC2 Râ), and enzymatic
digest analyses (1-5 µg of NC2 Râ) analyses. For the photoactivatable
reagent 8a, light exposure time points were taken using a hand-held
254 nm light source. Quenching of the sulfo-NHS cross-linking reaction
was accomplished with 2-mercaptoethanol followed by lowering of the
pH.
(c) Digesting and Extracting Cross-Linked Protein Complex. To
isolate cross-linked protein fragments suitable for MS analysis, we
devised the procedure that is summarized in Figure 3A. NC2 previously
cross-linked with reagent 4 or 6 was added to a microconcentrator (step
ii), excess deactivated cross-linking reagent was washed through the
filter (step iii), the protein complex was reduced with TCEP (1 µmol,
pH 7, 35 °C, 30 min), alkylated with iodoacetamide (1 µmol, pH 7,
room temperature, dark, 60 min), and digested with enzymes (Glu-C,
Asp-N, Asp-N/trypsin) (step iv), biotinylated fragments were bound
to neutravidin beads (250 µL slurry, 4 °C, 30 min) and transferred to
a spin column (Bio-Rad) (step v), and unmodified protein digest
fragments were washed away with solutions of high salt and detergent
(16 volumes, 400 mM NaCl, 46 mM N-octylglucoside, 50 mM Hepes)
(steps vi and vii). Finally, the cross-linker-containing protein fragments
were released from the neutravidin support by cleavage of reagent 6
with hydroxylamine (200 mM, pH 8.5 at 37 °C). MS provided a quick
and informative tool for assessing each procedural step.
Cross-Linking Search Program. The PeptideMap program (Ge-
nomic Solutions) uses the following steps for an exhaustive calculation
of all possible cross-linked and non-cross-linked peptide masses. The
program (a) digests the protein sequences with the cleavage rules of
the indicated specific enzyme(s), (b) calculates the peptide masses taking
into account partial digestions, (c) adds the specified modifications,
also taking into account partial modifications, and finally (d) calculates
the masses of all possible cross-linked peptides, accounting for both
complete and incomplete cross-links.33
Acknowledgment. We thank Sarah Keegan for assistance
with the cross-linking program PeptideMap, Mike Goger for
assistance with the NMR studies, Julio Camarero, Derek
McLachlin, Martine Cadene, and Julio Cesar Padovan for useful
discussions, and Andrew Krutchinsky for assistance with the
QqTOF-MS analyses. The present work was supported in part
by grants from the Burroughs Wellcome Fund, the National
Institutes of Health (Grant RR00862 from the National Center
for Research Resources and Grant R33CA89810 from the
National Cancer Institute), the Irma T. Hirshl Trust, and the
Rockefeller University.
JA026917A
(33) Complete cross-links include those that occur between different proteolytic
peptide fragments as well as those that occur within a given peptide.
Incomplete or dead-end cross-links occur only within single peptide
fragments.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2425