Surveying ubiquitin structure by noncovalent attachment of distance constrained bis(crown) ethers

Article abstract of DOI:10.1021/ac800177s

Selective noncovalent adduct protein probing (SNAPP) mass spectrometry was recently developed to study solution-phase conformations of proteins by exploiting the specific affinity between 18-crown-6 ether (18C6) and lysine side chains. To obtain more detailed information about protein tertiary structure, a novel noncovalent cross-linking reagent with two 18C6 molecules bridged by a covalent phenyl linker (called PBC for phenyl bis-crown) was synthesized. PBC introduces a distance constraint into SNAPP experiments where pairs of lysine side chains that are held in proximity by tertiary structure should be the most favored binding sites. Application of this method to ubiquitin reveals that PBC can bind to one lysine in a monodentate fashion or bind to two lysines via a bidentate interaction. Comparison with 18C6 can be used to reveal the mode of binding. For the native state of ubiquitin, bidentate binding of PBC is not observed. The partially denatured A-state, however, contains a single pair of lysines that are both chemically available and spaced by less than ～19 A (the maximum distance spanning the crown ether binding sites in PBC). Collision-induced dissociation and site-directed mutagenesis reveal that the bidentate PBC attaches to K29 and K33, which is in agreement with previous structural data on the A-state of ubiquitin. PBC is shown to be an effective probe of protein structure in SNAPP experiments, although assigning the specific residues to which PBC is attached can be experimentally challenging.

Full text of DOI:10.1021/ac800177s

Anal. Chem. 2008, 80, 5059–5064
Surveying Ubiquitin Structure by Noncovalent
Attachment of Distance Constrained Bis(crown)
Ethers
Tony Ly,† Zhenjiu Liu,† Brian G. Pujanauski,‡ Richmond Sarpong,‡ and Ryan R. Julian*,†
Department of Chemistry, University of California, Riverside, California 92521, and College of Chemistry, University of
California, Berkeley, California 94720
Selective noncovalent adduct protein probing (SNAPP)
mass spectrometry was recently developed to study solu-
tion-phase conformations of proteins by exploiting the
specific affinity between 18-crown-6 ether (18C6) and
lysine side chains. To obtain more detailed information
about protein tertiary structure, a novel noncovalent cross-
linking reagent with two 18C6 molecules bridged by a
covalent phenyl linker (called PBC for phenyl bis-crown)
was synthesized. PBC introduces a distance constraint
into SNAPP experiments where pairs of lysine side chains
that are held in proximity by tertiary structure should be
the most favored binding sites. Application of this method
to ubiquitin reveals that PBC can bind to one lysine in a
monodentate fashion or bind to two lysines via a bidentate
interaction. Comparison with 18C6 can be used to reveal
the mode of binding. For the native state of ubiquitin,
bidentate binding of PBC is not observed. The partially
denatured A-state, however, contains a single pair of
lysines that are both chemically available and spaced by
less than ∼19 Å (the maximum distance spanning the
crown ether binding sites in PBC). Collision-induced
dissociation and site-directed mutagenesis reveal that the
bidentate PBC attaches to K29 and K33, which is in
agreement with previous structural data on the A-state of
ubiquitin. PBC is shown to be an effective probe of protein
structure in SNAPP experiments, although assigning the
specific residues to which PBC is attached can be experi-
mentally challenging.
for examining protein structure. This statement requires clarifica-
tion because the most relevant environment for studying protein
structure is obviously in solution, yet MS is conducted exclusively
in the gas phase. Despite this fact, cleverly designed experiments
can utilize MS to track structurally dependent mass shifts that
occur in solution yet are retained and observed in the gas phase.
In other words, the mass spectrometer is used as a detector, not
a reaction vessel. One of the most common techniques employs
hydrogen/deuterium exchange to probe the chemical availability
of backbone amide hydrogens.5–7 The degree of exchange is
largely determined by intramolecular interactions (typically hy-
drogen bonds)15 within the protein and can be easily quantified
by the resulting mass shift for each exchange. In practical terms,
faster exchange is usually observed for more exposed portions
of the protein. Quenching of the exchange, followed by enzymatic
digestion, and further MS analysis of the fragments can be used
to localize fast- or slow-exchanging regions.6 Three-dimensional
structures, such as those obtained by X-ray or NMR, are not
revealed and neither are tertiary structural interactions.
This shortcoming can be partially addressed by the use of
chemical cross-linking studies.9–11 In these experiments, reactive
amino acid side chains are covalently cross-linked to one another.
The cross-linking reagents are chosen such that the maximum
distance between two amino acids can be determined.16 Following
enzymatic digestion and analysis by MS, the residues that are
cross-linked are revealed because of the mass shift caused by the
cross-linking agent. In this manner, proximity relationships
between different amino acids can be established and provide
experimental constraints on computational models of protein
structure.17 These experiments usually only provide information
Protein structure is one of the keys to understanding function
or malfunction1 in biological systems at the molecular level;
therefore, methods that enable the rapid screening of protein
structure are needed. Mass spectrometry (MS) has several
inherent advantages over X-ray2 or NMR3 in terms of speed and
sensitivity, which has led to the development of several tools4–14
(7) Wales, T. E.; Engen, J. R. Mass Spectrom. Rev. 2006, 25, 158–170
.
(8) Kaltashov, I. A.; Eyles, S. J. Mass Spectrom. Rev. 2002, 21, 37–71.
(9) Sinz, A. Mass. Spectrom. Rev. 2006, 25, 663–682.
(10) 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.
(11) Kruppa, G. H.; Schoeniger, J.; Young, M. M. Rapid Commun. Mass Spectrom.
* To whom correspondence should be addressed. E-mail: ryan.julian@ucr.edu.
† Department of Chemistry.
2003, 17, 155–162.
(12) Ly, T.; Julian, R. R. J. Am. Soc. Mass Spectrom. 2006, 17, 1209–1215.
(13) Friess, S. D.; Daniel, J. M.; Zenobi, R. Phys. Chem. Chem. Phys. 2004, 6,
2664–2675.
(14) Friess, S. D.; Daniel, J. M.; Hartmann, R.; Zenobi, R. Int. J. Mass Spectrom.
2002, 219, 269–281.
(15) Englander, S. W. Annu. Rev. Biophys. Biomol. Struct. 2000, 29, 213–238
(16) Green, N. S.; Reisler, E.; Houk, K. N. Protein Sci. 2001, 10, 1293–1304
(17) van Dijk, A. D. J.; Boelens, R.; Bonvin, A. M. J. J. FEBS J. 2005, 272,
‡ College of Chemistry.
(1) Chiti, F.; Dobson, C. M. Annu. Rev. Biochem. 2006, 75, 333–366
(2) Matthews, B. W. Annu. Rev. Phys. Chem. 1976, 27, 493–523
(3) Dyson, H. J.; Wright, P. E. Chem. Rev. 2004, 104, 3607–3622
(4) Loo, J. A.; Loo, R. R. O.; Udseth, H. R.; Edmonds, C. G.; Smith, R. D. Rapid
Commun. Mass Spectrom. 1991, 5, 101–105
(5) Katta, V.; Chait, B. T. Rapid Commun. Mass Spectrom. 1991, 5, 214–217
(6) Zhang, Z. Q.; Smith, D. L. Protein Sci. 1993, 2, 522–531
.
.
.
.
.
.
.
.
293–312.
10.1021/ac800177s CCC: $40.75 2008 American Chemical Society
Published on Web 05/23/2008
Analytical Chemistry, Vol. 80, No. 13, July 1, 2008 5059

about the proximity of side chains, and the backbone is not directly
probed. Caution should be employed while interpreting the results,
because factors other than spatial proximity can influence cross-
linking potential including chemical reactivity18 (which may be a
function of intramolecular entanglement) and protein structural
dynamics.
easily identify the degree of bidentate binding to a protein. The
A-state of ubiquitin contains one abundant PBC adduct, which is
attached by a bidentate interaction. Determination of the precise
binding sites for PBC is more challenging. Collision-induced
dissociation (CID) experiments can be used to narrow down the
possible sites and can also be used to confirm the degree of
bidentate binding. Site-directed mutagenesis was used to pinpoint
the preferred bidentate binding location to residues Lys29 and
Lys33. These results suggest that PBC will be useful for SNAPP
experiments to probe changes in protein structure under equi-
librium conditions in a general manner; however, the further
extraction of site-specific information with the noncovalent cross-
linking approach may be challenging.
A third method for examining protein structure, known as
SNAPP (selective noncovalent adduct protein probing) has been
developed recently.12 SNAPP utilizes a selective interaction
between 18-crown-6 ether (18C6) and lysine to examine protein
structure.19 18C6 associates weakly with protonated primary
amines (such as the side chain of lysine) in solution via the
formation of three hydrogen bonds. The dissociation constant for
the 18C6-protonated lysine complex has not been determined and
may be influenced by a variety of factors including local sequence
or protein structure. Nevertheless, the binding constant should
be similar in magnitude to protonated butylamine, for which
thermodynamic data are available. Protonated butylamine binds
to 18C6 with a dissociation constant of ∼110 mM in water.20
Therefore, attachment of 18C6 to lysine depends on the degree
to which the side chain is occupied by intramolecular interactions
such as salt bridges or hydrogen bonds.21 These potentially
interfering intramolecular interactions prevent attachment of 18C6
and are inseparably connected to the structure of the protein,
making attachment of 18C6 (or lack thereof) a sensitive probe of
protein structure.12 The number of 18C6s that attach to a protein
can be easily determined with MS because the 18C6/lysine
interaction becomes strong in the gas phase (∆H ∼-150 kJ/
mol)19 and causes an easily detectable mass shift. A typical SNAPP
experiment is conducted by electrospraying a solution containing
the protein and 18C6 directly into the mass spectrometer. The
resulting mass spectrum contains an intensity distribution of
protein-18C6 complexes. If the ensemble of conformations is
heterogeneous, then the observed 18C6 distributions represent
statistical averages of the entire ensemble. Assuming that distinct
conformations have different propensities for protonation, then
the 18C6 distributions for different structures are resolvable by
differences in charge state, and therefore, different dynamic states
of the protein can be simultaneously detected. Site-specific
information can be obtained by determining which lysine residues
are capable of attaching 18C6.21 The three-dimensional relation-
ship between different lysine side chains, however, is not
examined.
MATERIALS AND METHODS
Unless otherwise noted, all commercial reagents and solvents
were used as received. Bovine ubiquitin, oligonucleotide primers,
high-performance liquid chromatography (HPLC)-grade methanol,
and HPLC-grade acetonitrile were all purchased from Sigma-
Aldrich (St, Louis, MO). Water was purified to 18.2-MΩ resistivity
using a Millipore Direct-Q (Millipore, Billerica, MA). Escherichia
coli BL21 (DE3) cells were purchased from Novagen Inc.
(Madison, WI). T4 DNA ligase, pGEM T-vector, BamH1 and NdeI
were purchased from New England Biolabs (Beverly, MA); Wizard
SV Gel and PCR Clean-up System and Plasmid Miniprep Kit were
purchased from Promega (Madison, WI). Gene Jet Plasmid
Miniprep Kit was purchased from Fermentas Life Science (Glen
Bernie, MA).
All air- or moisture-sensitive reactions were conducted in oven-
dried glassware under an atmosphere of nitrogen using dry,
deoxygenated solvents. Tetrahydrofuran (THF) and diethyl ether
were distilled under nitrogen from sodium benzophenone ketyl.
18-Crown-6-methanol was purchased from Alfa Aesar and R,R′-
dibromo-p-xylene was purchased from Aldrich; both were used
without further purification. 1H NMR spectra were recorded on a
Bruker AV-500 (at 500 MHz) in chloroform-d at 23 °C. Chemical
shifts were referenced to the residual chloroform-H peak, which
was set at 7.26 ppm for 1H. Data for 1H NMR are reported as
follows: chemical shifts (δ ppm), multiplicity, (s ) singlet, d )
doublet, t ) triplet, q ) quartet, dd ) doublet of doublet, dt )
doublet of triplet, m ) multiplet, br ) broad resonance), coupling
constants (Hz), and integration. IR spectra were recorded on a
Nicolet MAGNA-IR 850 spectrometer and are reported in fre-
quency of absorption (cm-1). All mass spectra were obtained using
an LTQ linear ion trap mass spectrometer (Thermo Fisher,
Waltham, MA) with a standard electrospray ionization source.
Organic Synthesis. To a stirred solution of 18-crown-6-
methanol (28.3 mg, 0.096 mmol), R,R′-dibromo-p-xylene (12.7 mg,
0.048 mmol) in THF (500 µL) was added NaH (4.3 mg, 60%
dispersion in mineral oil, 0.106 mmol) with stirring. Stirring was
continued at room temperature for 24 h. Diethyl ether (2 mL)
was added, the mixture was filtered through celite, and the solvent
was removed by evaporation under reduced pressure to yield 41
mg of crude product (product + NaCl). The crude product was
The present work extends the utility of SNAPP with the
implementation of the bis(crown) probe shown (phenyl bis-crown,
PBC). The covalent attachment of two 18C6 moieties22 introduces
an additional distance constraint factor into SNAPP experiments.
Molecular modeling suggests that the maximum distance between
the centers of the two crown ethers is ∼19 Å. Results with
ubiquitin demonstrate that PBC is a viable SNAPP reagent.
Comparison with SNAPP experiments using 18C6 can be used to
(18) Swaim, C. L.; Smith, J. B.; Smith, D. L. J. Am. Soc. Mass. Spectrom. 2004,
15, 736–749
(19) Julian, R. R.; Beauchamp, J. L. Int. J. Mass Spectrom. 2001, 210, 613–623
(20) Izatt, R. M.; Terry, R. E.; Haymore, B. L.; Hansen, L. D.; Dalley, N. K.;
.
.
Avondet, A. G.; Christensen, J. J. J. Am. Chem. Soc. 1976, 98, 7620–7626
(21) Liu, Z.; Cheng, S.; Gallie, D. R.; Julian, R. R. Anal. Chem. I
n press (DOI: 10.1021/ac800176u).
.
(22) Julian, R. R.; May, J. A.; Stoltz, B. M.; Beauchamp, J. L. Angew. Chem., Int.
Ed. 2003, 42, 1012–1015
.
5060 Analytical Chemistry, Vol. 80, No. 13, July 1, 2008

Figure 1. Comparisons of 18C6 (solid) and PBC (hatched) distributions for ubiquitin sampled from native conditions (a) and A-state conditions
(b). The average numbers of adducts observed (18C6/PBC) by charge state are (+6) 0.64/0.63, (+7) 0.69/0.68, (+9) 2.34/1.27, and (+10)
2.23/1.46. The difference between the average number of 18C6 and PBC suggests a bidentate interaction is present in the A-state.
resuspended in water and purified using reversed-phase HPLC
on a 150 mm × 30 mm, 10-µm particle size C18 Gemini column
(Phenomenex, Torrance, CA). The elution gradient was 0-90%
acetonitrile over 30 min, using water and acetonitrile as cosolvents.
The observed retention time for PBC was ∼13.9 min (∼42%
the driving force to maximize separation between the two crown
ether centers without significant distortion of PBC. The Arabi-
dopsis mutants were prepared as described in detail elsewhere.21
RESULTS AND DISCUSSION
1
acetonitrile). H NMR (500 MHz, CDCl3) δ 7.31 (s, 4H), 4.55 (s,
Ubiquitin is an important signaling protein that has been well
characterized by a variety of mass spectrometric and other
techniques.24 The results for SNAPP experiments with ubiquitin
using PBC and 18C6 as the probe reagents are shown in Figure
1. The structure of ubiquitin is quite stable25 but can be denatured
in the presence of acid and organic cosolvent to yield the
“A-state”.25,26 In Figure 1a, the natively folded structure is
represented in the +6 and +7 charge states. The A-state structure
is partially denatured and is observed in higher charges states,
such as +9 and +10 shown in Figure 1b.12,27 Comparison of the
SNAPP distributions for the two structures reveals that both PBC
and 18C6 attachment are structure dependent. This conclusion
is drawn from the fact that both the total number of crowns
attaching and the most abundant number of adducts shift to a
higher number of adducts for the A-state with PBC and 18C6.
Therefore, both reagents can be used for SNAPP experiments to
evaluate changes in protein structure, but is there additional
information provided by the distance constraint component of
PBC?
4H), 3.6-3.9 (m, 50H). FT-IR (thin film) 2923.1, 2853.4, 1923.0,
1437.9, 1228.4. MS (m/z): [M + Na]+ calcd for [C34H59O14Na]+,
713.37; found, 713.48.
Selective Noncovalent Adduct Protein Probing Mass
Spectrometry. Solutions were prepared either with 50/50 water/
methanol + 1% acetic acid to generate the A-state or with water
alone to sample the native structure. To ensure stoichiometric
excess of adduct, solutions were prepared such that the number
of crown moieties was twice the number of possible binding sites.
For 18C6, the concentrations used were 10 µM protein/160 µM
18C6. For PBC, which contains two crowns, the molar ratio was
10 µM protein/80 µM 18C6. Ions were generated by direct
infusion into the electrospray inlet of the LTQ linear ion trap mass
spectrometer. Source and ion lens parameters (electrospray
voltage, ion-transfer capillary temperature, tube lens voltage) were
optimized for gentle ionization and maximal adduct intensities and
are similar to parameters used previously.12 Once determined,
these instrument parameters remained constant for a given
analyte. Ions of interest were then isolated in the trap and
subjected to a resonance excitation rf voltage to produce frag-
mentation after many collisions with He buffer gas. Multiple stages
of CID (MSn) can be achieved in the trap by reisolating and
exciting subsequent dissociation products. MSn experiments were
used to locate the noncovalent attachment of PBC.
Careful comparison of the 18C6 and PBC distributions reveals
that some of the PBC adducts are attached in a bidentate fashion.
If two lysines are within ∼19 Å (the maximum separation as
determined by modeling the distance between two ammonium
ions attached to PBC at the PM3 semiempirical level), then one
PBC can attach to both residues. If two lysine residues are not
within this distance, then two PBC adducts can attach, each in a
Semiempirical calculations (PM3)23 were performed as imple-
mented in Gaussian 03 version 6.1 revision D.01. To determine
the maximum binding distance for PBC, a ternary complex of PBC
and two ammonium adducts were minimized in the gas phase.
Coulombic repulsion between the two ammoniums should provide
(24) Hershko, A.; Ciechanover, A. Annu. Rev. Biochem. 1998, 67, 425–479
(25) Brutscher, B.; Bruschweiler, R.; Ernst, R. R. Biochemistry 1997, 36, 13043–
13053
(26) Hoerner, J. K.; Xiao, H.; Kaltashov, I. A. Biochemistry 2005, 44, 11286–
11294
(27) Pan, J. X.; Wilson, D. J.; Konermann, L. Biochemistry 2005, 44, 8627–8633
.
.
.
(23) Stewart, J. J. P. J. Comput. Chem. 1989, 10, 209–220
.
.
Analytical Chemistry, Vol. 80, No. 13, July 1, 2008 5061

monodentate fashion. A single bidentate attachment is expected
to be more favorable than two monodentate attachments, due to
the lower entropic costs of forming the bidentate crown-lysine
complex. Additionally, monodentate attachment of PBC may be
somewhat less favorable than attachment of 18C6 due to scaveng-
ing interactions between the unoccupied crown of PBC with free
ions in solution. The unoccupied crown of PBC acts as a receptor
for free ions, effectively decreasing binding at lysine for the other
crown due to Coulombic repulsion. These scavenging interactions
may become more common in the late stages of ionization, i.e.,
evaporating droplets, where concentrations of free ions are higher
than in bulk solution. In contrast, 18C6 can only interact with a
single lysine; therefore, comparison of the two adduct distributions
should reveal information about the degree of bidentate binding
because each bidentate PBC will be replaced by two 18C6s while
each monodentate PBC will be matched by a single 18C6. It is
not anticipated that there will be an exact quantitative correlation;
however, the net result should be a reduction in the number of
PBC adducts by approximately one relative to 18C6 for each
bidentate PBC.
Returning to the data in Figure 1, the average number of
adducts can be calculated for each charge state to determine the
extent of bidentate attachment for PBC. The average numbers
by charge state for 18C6/PBC are 6+, 0.64/0.63; 7+, 0.69/0.68;
9+, 2.34/1.27; and 10+, 2.23/1.46. For the folded structure, there
is almost no change in the number of adducts that attach between
PBC and 18C6 (although the shape of the distributions do vary).
This suggests that there are no lysine-lysine pairs that are
available to bind 18C6 and held within ∼19 Å for the native-state
structure. Chemical cross-linking studies have found lysine-lysine
pairs within this distance.11,28 There are several experimental
differences that may account for this discrepancy. First, nonco-
valent attachment of 18C6 may be more sensitive to interfering
intramolecular interactions than covalent reactions.29 Second,
SNAPP experiments sample equilibrium conditions, whereas
covalent cross-linking is not reversible and may trap transient
states or exaggerate the importance of minor conformations. In
either case, a more dramatic shift occurs in the A-state, suggesting
that this structure exhibits a significantly higher degree of
bidentate binding. The difference between the average number
of 18C6 and PBC adducts in the A-state suggests that one PBC is
probably interacting with ubiquitin via bidentate binding. More
detailed information, such as the specific binding site for the
bidentate PBC, cannot be extracted from the SNAPP distribution
alone and requires further experiments.
Figure 2. CID spectra of (a) [ubiquitin + PBC + 11H]11+, (b) PBC-
retaining fragment product [y58 + 9H + PBC]9+, and (c) [y58-y27
+
PBC + 5H]5+. Observation of [y58-y27 + PBC + 5H]5+ localizes the
bidentate attachment to residues 20-52. However, more specific
information could not be obtained by CID alone, due to competitive
noncovalent loss of PBC from [y58-y27 + PBC + 5H]5+. Bold down
arrows indicate peaks being subjected to collisional excitation.
plexes of PBC and ubiquitin at lower charge states results in loss
of PBC as the exclusive dissociation pathway. However, for the
+11 charge state, several fragments are produced in substantial
abundance with PBC still attached. In the comparable experiment
with 18C6, no fragments with 18C6 still attached are detected,
suggesting that bidentate attachment is required for PBC to be
retained following fragmentation of the protein. Following reiso-
lation and fragmentation of the [y58 + PBC + 9H]9+ fragment,
loss of PBC again yields the base peak as shown in Figure 2b.
However, there is a small amount of a secondary fragment
(y58-y24) produced with PBC still attached. In Figure 2c, an MS4
experiment is conducted to verify the assignment of this peak.
Indeed, loss of PBC from this fragment yields the base peak,
confirming that PBC was attached to the precursor. These results
suggest that the possible locations for bidentate binding are most
likely between residues 20-52. Unfortunately, there are four lysine
residues in this region of the protein, and further experiments
CID experiments were performed to evaluate potential binding
sites. Fortuitously, protonated ubiquitin ions that carry a high net
charge (g+10 charges) have a facile cleavage point at Pro19
(which has been identified in previous work)30 that facilitates
fragmentation of the protein. The results for fragmenting [Ubi +
PBC + 11H]11+ are shown in Figure 2a. The base peak results
from simple loss of the PBC adduct, as would be expected given
the noncovalent attachment. Indeed, CID of noncovalent com-
(28) Novak, P.; Young, M. M.; Schoeniger, J. S.; Kruppa, G. H. Eur. J. Mass
Spectrom. 2003, 9, 623–631
(29) For example, compare ref 21 and Novak, P.; Kruppa, G. H.; Young, M. M.;
Schoeniger, J. J. Mass Spectrom 2004, 39, 322–328
(30) Reid, G. E.; Wu, J.; Chrisman, P. A.; Wells, J. M.; McLuckey, S. A. Anal.
Chem. 2001, 73, 3274–3281
.
.
.
5062 Analytical Chemistry, Vol. 80, No. 13, July 1, 2008

Figure 4. (a) Data from seven single residue mutants of Arabidopsis
ubiquitin. Differences in PBC distributions for the 10+ charge state
between wild-type and each mutant were summed. Mutations that
affected the PBC distribution most, K29N and K33N, are the putative
sites for PBC bidentate attachment. (b) Ribbon cartoon of the A-state
of ubiquitin depicting a loose tertiary structure composed of two R
helices and a ꢀ sheet. Random coil and highly flexible regions are
shaded light gray, while distinct secondary structures are shaded dark
gray. A single PBC was to found bridge K29 and K33 (circled), which
are on the same R helix.
as shown in Figure 3b. No additional fragmentation with retention
of both adducts was observed. It is known that the A-state of
ubiquitin is dynamic in nature, and it is possible that the second
bidentate PBC attaches to one of the less abundant conformations
or only attaches to the protein transiently due to a less favorable
interaction. In either case, the amount of the second bidentate
adduct appears to be minimal from these results and analysis of
Figure 1. The PBC triple adduct was subjected to CID as shown
in Figure 3c. In this case, only loss of PBC is observed, confirming
that some of the PBC adducts are most likely attached to the
protein by a monodentate interaction. Furthermore, this result
demonstrates that it is not possible for any arbitrary PBC adduct
to adopt a bidentate state in the gas phase during the CID process.
In order to further pinpoint the primary bidentate binding site,
experiments were performed with a series of ubiquitin mutants
from Arabidopsis thaliana. Arabidopsis ubiquitin yields results
nearly identical to bovine ubiquitin for every structural probing
technique that has been applied to both proteins.21 Furthermore,
the overall backbone structure is highly conserved within the
ubiquitin family.31 In fact, the ubiquitin-related protein, RUB1, has
29 amino acid substitutions (as compared to bovine), yet yields a
structure with an essentially identical backbone structure.32 The
primary sequences of ubiquitin derived from bovine and Arabi-
dopsis differ by only three residues (S19P, D24E, A57S), indicating
that results from both proteins should be highly comparable. In
the Arabidopsis mutants, each lysine is mutated to asparagine one
at a time. This enables the contribution of each lysine as a binding
site for 18C6 or PBC to be evaluated independently. The results
are summarized in Figure 4a, where the magnitude of the peak
Figure 3. (a) CID of [ubiquitin + 11H + 2PBC]11+ producing [y58
+
8H + 2PBC]8+ in low abundance, suggesting that the second
bidentate attachment is more weakly bound than the first. (b)
Confirmation of the assignment is provided by CID of the PBC-
retaining product [y58 + 8H + 2PBC]8+, which results in loss of
noncovalently bound PBC without backbone fragmentation. (c) CID
of [ubiquitin + 11H + 3PBC]11+ and other higher order ubiquitin-PBC
complexes (data not shown) exclusively produces noncovalent losses
of PBC and [PBC + H]+. Bold down arrows indicate peaks being
subjected to collisional excitation.
(which are described below) are necessary to pinpoint the location.
Fragmentation of higher charge states produced similar product
ions.
A comparison of the adduct distributions and the CID experi-
ments discussed above reveals that there is a single lysine-lysine
pair bridged by a bidentate PBC. To confirm that there is only
one bidentate PBC attachment, we performed CID experiments
on complexes where the number of PBC molecules attached is
greater than one, as shown in Figure 3. Figure 3a shows the MS2
of the ternary complex of ubiquitin and two PBC molecules, which
yielded primarily loss of one or both PBC adducts. Nevertheless,
a very small amount of the y58 fragment with two PBC adducts
attached was produced. This assignment was confirmed with MS,3
which generated several peaks corresponding to the loss of PBC
(31) Walters, K. J.; Goh, A. M.; Wang, Q. H.; Wagner, G.; Howley, P. M. Biochim.
Biophys. Acta 2004, 1695, 73–87
(32) Rao-Naik, C.; de la Cruz, W.; Laplaza, J. M.; Tan, S.; Callis, J.; Fisher, A. J.
J. Biol. Chem. 1998, 273, 34976–34982
.
.
Analytical Chemistry, Vol. 80, No. 13, July 1, 2008 5063

is proportional to the degree of binding at a particular lysine. The
results suggest that K29 and K33 are the most likely sites for
PBC binding. Since bidentate binding is predicted to be strongly
preferred over monodentate binding, these sites are also the most
likely pair for the bidentate PBC. This conclusion is in good
agreement with the results obtained by CID in Figure 3.
Furthermore, predictions based on NMR data for the A-state
structure identified organized secondary structural regions, but
an absence of stable tertiary structure.25 The cartoon structure
shown in Figure 4b represents the secondary structural elements
that were predicted by NMR with an absence of any tertiary
structure.25 Lys29 and Lys33 are the same side of the central
R-helix. The predicted distance between these residues is in the
10-15-Å range, depending on the relative orientation of the side
constraints between residues that are spatially close, information
which is useful in constructing theoretical models. However, by
using noncovalent interactions, this approach avoids kinetic traps
that can potentially occur during covalent modification and
provides complementary information. Our SNAPP data for the
A-state of ubiquitin suggest a single lysine-lysine pair within ∼19
Å. Data from CID experiments and Arabidopsis ubiquitin mutants
confirm the bidentate attachment and localize the pair to K29 and
K33, in agreement with the NMR studies of the A-state.25 These
results also demonstrate that some noncovalent interactions in
solution are strong enough in the gas phase to be competitive
with protein backbone fragmentation using conventional CID.
However, CID experiments may only provide a rough estimate
of where the noncovalent adduct is attached and are limited by
the number of facile cleavage sites on a protein. Nevertheless,
PBC is shown to be a useful SNAPP reagent that can rapidly and
easily probe protein structure. The noncovalent approach for
probing distance constraints may find utility in systems where
dynamics are important, or where the required pH may prove
problematic for traditional covalent cross-linking studies. Both
issues are likely important for studying the A-state of ubiquitin.
chains. Therefore,
a sustainable bidentate interaction is
possible between these residues in the absence of tertiary
structure. The absence of a strong bidentate PBC bridging any
two of the secondary structure elements in Figure 4b agrees with
the NMR prediction of a highly dynamic structure lacking stable
tertiary features. It is interesting to note that Lys29 and Lys33
are on the same helix in the folded structure as well; however,
results with 18C6 indicate that these residues are intramolecularly
entangled. Analysis of the crystal structure suggests that the
sequence remote residues 52 and 14 are the interaction sites. This
information also suggests that the A-state lacks any stable tertiary
structure.
ACKNOWLEDGMENT
T.L., Z.L., and R.R.J. acknowledge funding from the University
of California, Riverside, and the American Society for Mass
Spectrometry Research Award. R.S. and B.G.P. acknowledge
support from the University of California, Berkeley, and the Roche
Predoctoral Fellowship (B.G.P.).
CONCLUSIONS
We have studied whether noncovalent attachment of a novel
bis(crown) ether molecule is specific toward lysine pairs within a
distance imposed by a semirigid linker. This approach is similar
to chemical cross-linking in that it potentially provides distance
Received for review January 24, 2008. Accepted April 22,
2008.
AC800177S
5064 Analytical Chemistry, Vol. 80, No. 13, July 1, 2008