Detection of Functional Ligand-Binding Events Using ...

Detection of Functional Ligand-Binding Events
Using Synchrotron X-Ray Scattering

DIANE J. RODI,1 SUNEETA MANDAVA,1 DAVID B. GORE,2
LEE MAKOWSKI,1 and ROBERT F. FISCHETTI1,3

Small-molecule ligands that change the structure of a protein are likely to affect its function, whereas those causing no struc-
tural change are less likely to be functional. Wide-angle x-ray scattering (WAXS) can be easily carried out on proteins and
small molecules in solution in the absence of chemical tags or derivatives. The authors demonstrate that WAXS is a sensi-
tive probe of ligand binding to proteins in solution and can distinguish between nonfunctional and productive binding.
Furthermore, similar ligand-binding modes translate into similar scattering patterns. This approach has high potential as a
novel, generic, low-throughput assay for functional ligand binding. (Journal of Biomolecular Screening 2007:994-998)

Key words: ligand binding, wide-angle x-ray scattering

INTRODUCTION higher concentration.1,2 Total cycle time for data collection
once the experimental setup is complete averages from 8 to 10
AMAJOR ADVANCE IN DRUG DISCOVERY has been the devel- min per protein sample. The high flux and stable x-ray beams
opment of techniques to generate large libraries of target- available at third-generation synchrotron facilities such as the
focused probe chemicals. This development, in combination with Advanced Photon Source (APS) enable the measurement of
the ever-increasing numbers of proteins entering screening weak wide-angle scattering (out to a spacing of almost 2 Å).
programs via human genome expression profiling, has intensified The ability to obtain scattering data over a range of spacing
the need for novel rapid screening techniques that can pinpoint from 2.5 Å to about 100 Å allows the detection of structural
those molecules with biologically relevant properties (such as changes ranging from domain rotations to protein com-
knock-down or knockout activity). Most methods used to date paction/expansion to intramolecular breathing motions.1-3
rely on 1 of 2 strategies: either detection of physical binding or
impairment of target function. The former class usually requires MATERIALS AND METHODS
immobilization or tagging of 1 or more of the binding pairs and
will identify ligands that may or may not impair protein activity. WAXS data collection
The latter require specialized assays for each target function and
may be less amenable to a high-throughput approach. WAXS data were collected at the BioCAT undulator beam
line (18ID) at APS (Argonne, IL). The experimental layout was
Frequently, the functional binding of a small molecule to a arranged as previously described.1 The sample cell consisted of a
protein is accompanied by a change in the structure of the pro- thin-walled quartz capillary (1 or 1.5 mm ID) attached to a pro-
tein. Wide-angle x-ray scattering (WAXS) from proteins in grammable pump (Hamilton Microlab 500 series, Hamilton
aqueous solution is capable of detecting small changes in struc- Company, Reno, NV) that was adjusted to deliver continuous
ture1 and can be quickly collected (less than 5-s exposure time) flow through the capillary during data collection. The ambient
from any soluble protein or protein fragment at 5 mg/mL or temperature of the air surrounding the capillary and sample tub-
ing was manipulated by attachment of an ethylene glycol bath to
1Biosciences Division, Argonne National Laboratory, Argonne, Illinois. the brass capillary holder during data collection. The x-ray scat-
2BioCAT, CSRRI, and BCPS, Illinois Institute of Technology, Chicago. tering pattern was recorded with a MAR165 2k×2k CCD detec-
3GM/CA CAT, Advanced Photon Source, Argonne National Laboratory, tor. The specimen-to-detector distance was approximately 170
Argonne, Illinois. mm and was calibrated using powder diffraction rings from
either silicon or silver-behenate. The beamline is capable of
Received May 18, 2007, and in revised form Jun 18, 2007. Accepted for pub- delivering approximately 2 × 1013 photons/s/100 mA of beam
lication Jun 18, 2007. current. As previous experience on the BioCAT beamline has
demonstrated that proteins under a variety of physical conditions
Journal of Biomolecular Screening 12(7); 2007
DOI: 10.1177/1087057107306104

994 www.sbsonline.org © 2007 Society for Biomolecular Sciences

Downloaded from jbx.sagepub.com by guest on February 28, 2016

Detection of Functional Ligand-Binding Events

are damaged after exposure times of a few tenths of a second to a adequately reflect the scattering it contributes when the capil-
few seconds at these intensity levels, in these experiments, 8 to 32 lary is filled with buffer or protein solution, due to absorption
(20-µm) aluminum foils were used (depending on the concentra- of scattering by buffer and/or protein. This was accounted for
tion of protein in the samples) as x-ray beam attenuators to con- by modeling the capillary scatter as
trol the incident beam flux. A data set consists of a series of 1-s
exposures, with 4 from buffer, 5 to 7 from protein solution, and 5 Icap = (scale factor)*Icap(observed) + (3)
from the empty capillary. This can consume anywhere from 100 constant (absorption correction).
to 150 µL of protein solution ranging from 2.5 to 20 mg/mL,
requiring an amount of protein ranging from 0.25 to 2 mg of pro- The scale factor and constant were selected by empirically
tein total. Specific concentrations used for these experiments are fitting the capillary scatter to the scatter from the buffer-filled
shown in the figure legends. Exposures from sample and buffer capillary in the scattering range, (1/d) = 0.05 to 0.2 Å-1, because
were alternated to minimize the possible effects of drift in any scatter from buffer is negligible in this range.
experimental parameter. Incident beam flux was monitored using
nitrogen gas-filled ion chambers. Integrated beam flux during Statistical significance calculation
each exposure was used to scale scattering from protein solutions
with scattering from buffer solutions. Diffraction data were col- WAXS pattern was composed of ~860 data points extending
lected using a flow cell in which no protein was exposed to a from q ~ 0055 Å-1 to 2.6 Å-1. Of this, about 725 points were
beam for more than 100 ms. At these exposure levels, the effect used to estimate the significance of differences between curves.
of radiation damage on radio-sensitive test proteins is unde- Points at q < 0.25 Å-1 were not used because small errors in
tectable.2 Outlier patterns, generated due to passage of small bub- scaling would be amplified at small angles due to the large
bles through the x-ray beam during the exposure, were removed intensities. Standard deviations were calculated on the basis of
prior to averaging. Standard deviations of the observed data were multiple (usually 7) data sets from each sample. A standard chi-
calculated, with error propagation formulas used to calculate their square test was used to derive a measure of the significance of
effect on the final estimate of scattering from protein. differences observed between 2 data sets.8 The number of
degrees of freedom was determined from the number of inde-
Scattering data analysis pendent parameters required to define the scattering pattern
from a protein of known radius.9,10 For a protein of diameter
The 2D scattering patterns were integrated radially to 1D ~30 Å, that amounts to ~22 parameters over the range of scat-
scattering intensity profiles using the program Fit2D (version tering vectors used for the chi-square calculation, correspond-
12.077).4-6 The origin of the diffraction pattern was determined ing to υ ~700 degrees of freedom. The reduced chi-square,
from the powder diffraction rings from silver-behenate and/or χυ = χ / υ, provides a measure of goodness of fit, with values
silicon powder. Scattering from samples should be separable smaller than ~1.0 indicative of a good fit (scattering curves that
into 4 individual components: that due to scattering from the are indistinguishable in this context) and values larger than 1.0
protein, from the bulk solvent, from the solvent of hydration, corresponding to scattering curves that exhibit statistically sig-
and from the capillary. Scattering from the solvent of hydration nificant differences.
(boundary layer), although potentially important at small
angles of scattering, is generally at least 2 orders of magnitude RESULTS AND DISCUSSION
weaker than any other contribution in the range of angles stud-
ied here.7 Scattering from protein was estimated according to Previous work that focused on proteins in the presence and
absence of their natural cofactors demonstrated that WAXS solu-
Iprot = Iobs – Icap – (1 – vol%)Isolvent , (1) tion patterns can detect both large and subtle shifts in protein
structure, such as the rotational change accompanying NAD+
where Iobs is the measured scattering from the protein sample, binding to alcohol dehydrogenase.1 These data, however, left in
Icap is the measured scattering from the empty capillary, and vol% doubt the potential utility of this technique to distinguish relative
is the estimated proportion of the solution taken up by the protein binding of small molecules in a population of closely related
(and thereby excluding solvent). Isolvent was estimated by potential ligands, as would be encountered in a lead drug detec-
tion study. To address these questions, we used riboflavin kinase
Isolvent = Ibkgd – Icap, (2) (RFK), an essential enzyme that has been demonstrated to bind its
2 small-molecule ligands at adjacent sites on the surface of the
where Ibkgd is the measured scattering from the capillary con- molecule.11 Each ligand modulates the protein in such a manner
taining buffer. Scattering from the empty capillary does not as to shift a surface flap to a new position. Scattering patterns

Journal of Biomolecular Screening 12(7); 2007 www.sbsonline.org 995

Downloaded from jbx.sagepub.com by guest on February 28, 2016

Rodi et al.

FIG. 1. Wide-angle x-ray scattering (WAXS) data collected from value of WAXS in the distinction of ligand-binding capacity
riboflavin kinase (RFK). Solutions of 16.4 mg/mL RFK in 50 mM between a series of structurally related small molecules.
Tris-HCL (pH 8.0), 13.3 mM MgCl2, 5 °C with relative intensity (nor-
malized for concentration) plotted versus scatter angle q in the absence Solution WAXS patterns from hen egg white lysozyme were
of added ligand (black), the same in the presence of 5 mM adenosine collected as previously described,1,2 in the absence of ligand
triphosphate (red), and 10 mM riboflavin (blue), respectively. Note the and the presence of dextrose, NAG, and (NAG)3. Figure 2a
similarity in scattering patterns between the 2 different ligand-bound demonstrates the reproducibility of solution scattering over
forms of the enzyme. time (χυ = 0.89). The minor changes measured upon addition of
dextrose and NAG (χυ = 1.14 and 1.07, respectively) are due to
from RFK were collected as previously described,1,2 in the changes in the contrast between protein and solvent on adding
absence of ligand and the presence of either adenosine triphos- ligand. By comparison, the scattering profile of lysozyme in the
phate (ATP) or riboflavin. Figure 1 demonstrates that not only presence of the tightly binding active site inhibitor (NAG)3 is
does the addition of each ligand produce a statistically significant dramatically altered. The 2 peaks at q = 0.31 Å–1 and 0.60 Å-1,
change in the scattering profile (reduced chi-square, χυ = 2.94 for as well as the trough at 0.50 Å–1, are not only shifted in relative
ATP and 2.90 for riboflavin vs. apo RFK normalized for error), intensity upon ligand binding but in scattering angle as well
but the profiles for ATP and riboflavin are virtually indistinguish- (χυ = 5.41). A calculation of the radius of gyration of apo- and
able (χυ = 0.03 between the 2 ligand-bound forms). These data (NAG)3-bound lysozyme from their atomic coordinates (1E8L
demonstrate that not only can WAXS detect a ligand-induced and 1HEW, respectively) using the software CRYSOL7 (ver-
conformation change, but it can reliably detect protein complexes sion 2.4) confirms the implications of the scattering peak shifts,
with similar conformations. as upon (NAG)3 binding, the Rg increases from 14.21 to 15.08.

A remaining question is whether the changes in scattering CONCLUSIONS
intensity can distinguish between ligands with different affin-
ity. Stable protein-carbohydrate complexes have properties in These data from well-characterized protein/ligand model
common with protein-drug complexes in the sense that they can systems demonstrate that WAXS has great potential as a drug
be formed with just a few interfacial contacts.12 Lysozyme cat- discovery tool to quickly identify those chemicals with structure-
alyzes the breakdown of a component of the cell wall of certain altering properties. WAXS has several features that make it
bacteria, a polymer of N-acetyl-glucosamine (NAG)n. The full- particularly attractive as a rapid screen for lead drugs: 1) It
length active site of lysozyme can accommodate up to 6 NAG distinguishes between structure-altering versus nonspecific
residues in 6 subsites (sites A to F). Because the catalytic (adherent) binding events; 2) both target proteins and test lig-
residues are located between subsites C and D, tetrasaccharide ands are in solution without tags or immobilized scaffolds to
or longer oligosaccharides are enzymatically cleaved, whereas potentially interfere in the protein-ligand reaction; 3) the
NAG, (NAG)2, and (NAG)3 bind to subsites A, A/B, and A/B/C, degree of WAXS pattern change correlates with the degree
respectively, acting as competitive inhibitors. Dissociation con- of structure change (note that this ability to distinguish grada-
stants at pH 5.0 for NAG, (NAG)2, and (NAG)3 decrease from tion in ligand activity allows for the identification of weak
50 mM to 0.175 mM to 6.58 µM, a range of over 104 in magni- binders as a separate category to help further drug discovery,
tude.13 This wide range offers the opportunity to assess the including the biasing of future scaffolds in combinatorial
approaches, limitations in virtual screen/structure-guided lig-
and optimization, or ligand buildup/fragment-based proce-
dures); 4) a perquisite to this method is that the pattern shift can
frequently provide information regarding the binding mode of
the ligand, as in peak/trough shifts that indicate compaction/
expansion of the protein; and 5) crystallization is not required
for the technique. With the incorporation of a robotics-based
sample delivery system, automated data analysis, and the pool-
ing of chemicals to reduce protein consumption, this novel syn-
chrotron-based screening technique could be adaptable to a
medium-throughput approach to identify functional ligands.

ACKNOWLEDGMENT

The authors thank the staff and scientists of Sector 18 at the
APS for technical help and use of the facilities. This work and

996 www.sbsonline.org Journal of Biomolecular Screening 12(7); 2007

Downloaded from jbx.sagepub.com by guest on February 28, 2016

Detection of Functional Ligand-Binding Events

FIG. 2. Wide-angle x-ray scattering (WAXS) data collected from hen egg white lysozyme (HEWL). Solutions of HEWL in 50 mM Na acetate,
pH 4.5, 5 °C with relative intensity (normalized for concentration) plotted versus scatter angle q (a) in the absence of added ligand at 30 mg/mL
(solid circles) and 32 mg/mL (open circles) during 2 data collection sessions carried out 4 months apart at the Advanced Photon Source; (b) the

same in the absence (open circles) and presence (solid circles) of 10 mg/mL dextrose; (c) the same in the absence (open circles) and presence

(solid circles) of 12.5 mg/mL N-acetyl-glucosamine (NAG); and (d) the same in the absence (open circles) and presence (solid circles) of 25

mg/mL (NAG)3. Although NAG is a stable enough ligand to cocrystallize with the enzyme, only the tightly binding competitive inhibitor (NAG)3
produces a dramatic change in scattering pattern.

use of the Advanced Photon Source was supported by the US REFERENCES
Department of Energy under contract DE-AC-02-06CH11357.
BioCAT is a National Institutes of Health-supported Research 1. Fischetti RF, Rodi DJ, Gore DB, Makowski L: Wide-angle x-ray solution
Center RR-08630. The authors wish to thank Dr. Reddy's scattering as a probe of ligand-induced conformational changes in pro-
Laboratories Limited in Hyderabad, India for the kind gift of teins. Chem Biol 2004;11:1431-1443.
riboflavin kinase. The content is solely the responsibility of
the authors and does not necessarily reflect the official views of 2. Fischetti RF, Rodi DJ, Mirza A, Irving TC, Kondrashkina E, Makowski L:
the National Center for Research Resources or the National High-resolution wide-angle x-ray scattering of protein solutions: effect
Institutes of Health. of beam dose on protein integrity. J Synchrotron Radiat 2003;10(pt 5):
398-404.

Journal of Biomolecular Screening 12(7); 2007 www.sbsonline.org 997

Downloaded from jbx.sagepub.com by guest on February 28, 2016

Rodi et al.

3. Makowski L, Rodi DJ, Mandava S, Minh D, Gore D, Fischetti RF: 11. Karthikeyan S, Zhou Q, Osterman AL, Zhang H: Ligand binding-induced
Molecular crowding inhibits intramolecular breathing motions in pro- conformational changes in riboflavin kinase: structural basis for the
teins. J Mol Biol. In press. ordered mechanism. Biochemistry 2003;42:12532-12538.

4. Hammersley AP, Svensson SO, Hanfland M, Fitch AN, Hausermann D: 12. Garcia-Hernandez E, Zubillaga RA, Chavelas-Adame EA, Vazquez-
Two-dimensional detector software: from real detector to idealised image Contreras E, Rojo-Dominguez A, Costas M: Structural energetics of protein-
or two-theta scan. High Pressure Res 1996;14:235-248. carbohydrate interactions: insights derived from the study of lysozyme
binding to its natural saccharide inhibitors. Protein Sci 2003;12:135-142.
5. Hammersley AP: FIT2D: an introduction and overview. ESRF Internal
Report, ESRF97HA02T, 1997. 13. Dahlquist FW, Jao L, Raftery M: On the binding of chitin oligosaccha-
rides to lysozyme. Proc Natl Acad Sci USA 1966;56:26-30.
6. Hammersley AP: FIT2D V9.129 Reference Manual V3.1. ESRF Internal
Report, ESRF98HA01T, 1998. Address correspondence to:
Diane Rodi
7. Svergun D, Barferato C, Koch MHJ: CRYSOL—a program to evaluate
x-ray solution scattering of biological macromolecules from atomic coor- Biosciences Division
dinates. J Appl Crystallogr. 1995;28:768-773. Argonne National Laboratory

8. Bevington PR: Data Reduction and Error Analysis for the Physical Sciences. 9700 South Cass Avenue
New York: McGraw-Hill, 1969. Argonne, IL 60439

9. Makowski L: The use of continuous diffraction data as a phase constraint: E-mail: [email protected]
I. One-dimensional theory. J Appl Crystallogr 1981;14:160-168.

10. Makowski L, Rodi DJ, Mandava S, Gore DB, Fischetti RF: Estimation of
protein fold using wide angle x-ray solution scattering. 2007. Manuscript
submitted for publication.

998 www.sbsonline.org Journal of Biomolecular Screening 12(7); 2007

Downloaded from jbx.sagepub.com by guest on February 28, 2016