pmc logo imageJournal ListSearchpmc logo image
Logo of prosciProtein ScienceCSHL PressJournal HomeSubscriptionseTOC AlertsThe Protein Society
Journal List > Protein Sci > v.13(10); Oct 2004
Abstract
>Full Text
PDF (185K)
Contents
Archive

PubMed articles by:
Bahns, J.
Liu, C.
Chen, L.
Protein Sci. 2004 October; 13(10): 2578–2587.
doi: 10.1110/ps.04695704.
PMCID: PMC2286549
Copyright © Copyright 2004 The Protein Society
Characterizing specific phage–protein interactions by fluorescence correlation spectroscopy
John T. Bahns, Chin-Mei Liu, and Liaohai Chen
Biosciences Division, Argonne National Laboratory, Argonne, Illinois 60439, USA
Reprint requests to: Liaohai Chen, Biosciences Division, Argonne National Laboratory, Argonne, IL 60439, USA; e-mail: lhchen/at/anl.gov; fax: (630) 252-5517.
Received February 12, 2004; Revised June 3, 2004; Accepted July 2, 2004.
>Abstract
Abstract
The interactions of several affinity reagent displayed T7 and M13 phage particles with their corresponding target molecules were examined using Fluorescence Correlation Spectroscopy (FCS). Diffusion times, relative fractions of each component in the recognition reactions at the equilibrium state, and ultimately the dissociation constants were deduced from analyzing the fluorescence autocorrelation curves. Although the sample preparation and FCS characterization of icosahedral T7-related systems were relatively straight forward, procedures with filamentous M13-related systems were complicated by the physical size of M13 and its aggregate formation. Methods that accommodate the FCS measurement of the M13 phage via changing confocal optics, fitting procedures, and aggregate discrimination are presented and discussed.
Keywords: phage–protein interaction, phage display, combinational libraries, M13 phage, T7 phage, fluorescence correlation spectroscopy, dissociation constants
 
In the postgenome era, to reveal and understand the interactions between protein molecules is increasingly becoming very important. Not only is it an indispensable part of the structural and functional genomics, but it also directly facilitates drug discovery and medical diagnostics. Phage display is one of several rather potent techniques for identifying protein–protein interactions (Kay et al. 1998; Castagnoli et al. 2001; Rodi et al. 2002). In phage display, ligands are expressed as fusions to a capsid protein present on the surface of viral particles. Libraries of millions to billions of phage particles, each displaying a different fusion protein, can be screened for members displaying the desired properties or binding affinities. Phage display offers the following advantages: (1) The peptide or proteins that are expressed on the surface of the viral particles are accessible for interactions with their targets; (2) the recombinant viral particles are highly stable; (3) the viruses can be amplified (grown); and (4) each viral particle contains the DNA encoding its recombinant genome, thereby providing a physical linkage between the genotype and phenotype. Currently, two phage systems have been widely used: icosahedral T7, and filamentous M13 phage. Both types of phage libraries can be conveniently screened by isolating viral particles that bind to targets, plaque-purifying the recovered phage, and sequencing the phage DNA inserts, the so-called “biopanning” process. In general, three rounds of panning processes will yield a set of potential affinity reagents against the target molecules. As a “quality control” step in phage display, it is important to confirm and consequently characterize the affinity reagent selected from the library quantitatively. Therefore, a rapid and reliable characterization method, which can quantify the phage target interaction, will be an asset to the phage display technology.
Among the methods used to study the biointeractions, ranging from radioisotope labeling, Enzyme-Linked Immunosorbent Assay (ELISA), Surface Plasmon Resonance (SPR) to Microcalorimetry, Fluorescence Correlation Spectroscopy (FCS) has various unique advantages (Rigler and Elson 2001; Hess et al. 2002):
  • FCS is a powerful analytic tool that enables both qualitatively and quantitatively examining biointeractions as well as their time dependency. FCS measures fluorescence fluctuations within the detection space defined by a Gaussian beam with a typical volume of a femtoliter. Such fluctuations are caused by the changes in the number or orientation of fluorescent dye-labeled biomolecules diffusing in or out of the detection volume. In a typical autocorrelation analysis, the correlation amplitude is proportional to the reciprocal of the average number of the fluorescent species, and the decay is determined by their diffusion time. Analysis of the shape and decay rate of the correlation curve of the fluctuation signal reveals information about the absolute number of dye-labeled biomolecule and its complex as well as their diffusion constants.
  • FCS can measure the equilibrium concentrations of bound versus unbound fluorescence labeled components, respectively, without the need for separating unbound from bound proteins, provided that their mass difference is large enough to produce distinctive diffusion rates (normally, mass ratios typically must exceed ~5; Muller et al. 2003). Therefore, in principle, the dissociation constant can be evaluated from a single FCS experiment.
  • Unlike ELISA or SPR, which involve multisteps with a heterogeneous environment, the FCS measurement only engages a single-step in a homogenous solution. It does not involve anchoring recognition components onto a solid surface, nor the consequent washing steps. Thus, perturbation of protein structure and function is minimized. The dissociation constant obtained from FCS is therefore more accurate and reliable.
FCS has been extensively used to study the interactions of protein–protein and protein–small molecule (Schwille et al. 2000; Rigler and Elson 2001; Hess et al. 2002) both qualitatively and quantitatively. Rigler’s group first extended the FCS studies to the phage system, and was able to characterize the protein–M13 phage interactions a few years ago (Lagerkvist et al. 2001). To the best of our knowledge, there have been few reports of FCS studies on phage–protein interactions since then. Herein, we describe detailed, systematical FCS studies of the specific interactions of both M13 and T7 phage particles with the corresponding proteins with sizes ranging from 12 kDa (S-protein) to 150 kDa (IgG antibodies). Our results indicated that FCS is a very reliable tool to characterize protein–T7 interactions. For the M13 phage system, considering its filamentous structure (~800 nm in length and 8 nm in diameter), new measures have been developed to characterize protein–M13 interactions quantitatively by FCS. Although we could adapt the method established by Lumma et al. (2003), which was successful to probe the dynamics of a large semiflexible chain such as a λ-phage DNA by FCS, we used a simple approach of enlarging the confocal detection volume to accommodate the M13 phage and still used the standard FCS correlation function. In addition, we took the advantage of the large mass ratio between the phage and the protein; thus, were able to obtain the diffusion time of the protein–phage complex from the phage particle itself, and used it as a fixed parameter when fitting the autocorrelation curves of the protein–phage system. Under these conditions, we were able to reliably deduce the dissociation constant of the protein–phage interaction from the corresponding FCS spectra.
>Results and Discussion
Results and Discussion
FCS studies of T7 phage systems
Our T7 system contains a special T7 phage, on which copies of a 15 amino acid (aa) S-peptide were displayed on its capsid. The wild type of T7 phage particles, which do not contain S-peptides, were used as the negative control. As revealed from a TEM image (Fig. 1A [triangle]), the T7 phage has an icosahedral shape with a diameter of ~50 nm. At the concentrations used in this study (~1 nM), purified T7 phages (via CsCl gradient) are well dispersed in the solution, and there was no evidence of aggregate formation. Because the overall dimension of T7 phage is much smaller than the confocal detection volume (the beam waist radii was estimated to be 500 nm), T7 phage systems were characterized using FCS setup with a conventional 50-μm pinhole and a 60×/1.2 NA, water immersion objective. The interaction of the S-peptide peptide with a 104 aa S-protein derived from pancreatic ribonuclease A has been well characterized in the literature with a nanomolar dissociation constant (Raines et al. 2000). By labeling the S-protein with Alexa-647 dye, the interaction of the S-protein and S-peptide-displayed T7 was examined by FCS.
Figure 1.Figure 1.
TEM images of T7 and M13 phage particles.
Diffusion time determination for pure dye, dye-labeled proteins, and T7 phage particles
To measure the diffusion time of Alexa-647 dye, 10 μL of 1 nM of dye in PBS buffer solution was placed on the top of a cover glass (150 μm thickness). At least five FCS spectra were recorded with average counting rate of 400 KHz (laser power 250 μW). The diffusion time of Alexa-647 in PBS buffer solution was deduced using a one-component model, including the triplet term from each of five FCS spectra, and the average values reported in Table 1 (0.19 msec). The diffusion time of Alexa-647 was then used to obtain the diffusion time of dye-labeled proteins via a two-component model.
Table 1.Table 1.
Results of FCS-labeled T7 phage measurements using 60×/1.2 NA, with 50 μm pinhole at 633 nm
To measure the diffusion time of dye-labeled S-protein, 10 μL of 10 nM of Alexa-647-conjugated S-protein in PBS buffer solution was placed on the top of the cover glass. Again, at least five FCS spectra were recorded with an average counting rate of 70 KHz (laser power 250 μW). It was found that it is impossible to fit the autocorrelation curve with a one-component model even though the sample was twice purified with a size exclusion column, indicating the presence of unreacted free dye. Thus, we used a two-component model, while keeping the diffusion time of free dye as a fixed parameter to fit the autocorrelation curves. The diffusion time of Alexa-647-conjugated S-protein in PBS (0.72 msec), as well as the mole fraction of free dye were deduced from each of six FCS spectra, and the average values were reported in Table 1. The average values were used as fixed parameters for the FCS studies of the interaction between S-peptide-displayed T7 and S-protein.
Dye conjugation of the T7 phage was achieved by the reaction of succinimidyl ester of the red fluorescent dye Alexa-647 to the epsilon-amino groups of the lysine residues in T7 capsids. The labeled T7 phage were readily purified (removal of unreacted free dye) to ~98% using a gel filtration spin-column. To determine the diffusion time, 10 μL of 0.3 nM of Alexa-647-conjugated T7 phage in PBS buffer solution was placed on the top of the cover glass. At least five FCS spectra were recorded with an average counting rate of 20 KHz (laser power 250 μW). Again, we used the two-component model, while keeping the diffusion time of free dye a fixed parameter to fit the autocorrelation curves. The diffusion time of Alexa-647-conjugated T7 phage in PBS (3.8 msec), as deduced from each of five FCS spectra, is reported in Table 1. The diffusion time of T7 phage was used as a fixed parameter for the FCS study of the interaction between S-peptide-displayed T7 and S-protein.
Characterization of the interaction between S-peptide-displayed T7 and S-protein
To quantify the interaction of S-protein with S-peptide-display T7 phage, a stock T7 solution of 2.4 × 1011 phage per mL (0.4 nM) was used to prepare constant total volume mixtures of stock dye-S-protein with T7 and PBS buffer such that the concentration of dye-S-protein was held constant, at a variable T7 phage concentration. Unlabeled T7 was found to provide a weak fluorescent background of ~2 KHz; however, the amplitude contributed to G was negligible during FCS integration times used. After ~1 h of incubation at room temperature, ~30-sec duration FCS measurements were made. Figure 2 [triangle] shows typical autocorrelation curves obtained for the free dye, dye-labeled S-protein and S-protein–T7 complex. The FCS autocorrelation curves were fitted with a three-component model. As before, the diffusion times from the previous measurements (free dye, dye-labeled protein, and phage particles), as well as the mole fraction of the free dye were held fixed in this model (Fig. 2 [triangle], dotted line).
Figure 2.Figure 2.
Normalized FCS spectra of free Alexa-647 and Alexa-647-S-protein, S-protein-T7 complex and the autocorrelation fitting curve (three components) for the S-protein–T7 complex.
Data analysis of FCS of the S-protein–T7 complex is based on the following three assumptions:
  • The binding of small protein molecules to T7 phage particles has a negligible effect to the diffusion time of phage particles due to the huge mass ratio of the phage particle to protein molecule; thus, we can use the diffusion time of the Alexa-647-labeled T7 phage as the diffusion time of a single S-protein bound T7.
  • Nonspecific interactions of free dye with the phage particle are negligible; thus, the mole fraction of free dye could be fixed during the fitting procedure. This was confirmed by the FCS study of T7 particles with the Alex-647 dye. The fluorescence correlation curve of Alex-647 dye (10 nM in PBS buffer) remains unchanged upon the addition of T7 phage particles up to 10 nM. Therefore, no significant nonspecific interactions were observed for the dye molecules and T7 phage particles at the above concentrations.
  • Equation 4 (see Materials and Methods) was used for the determination of the dissociation constants under the condition of 1:1 stoichiometric reaction. Although the S-peptide-displayed T7 phage has multiple binding sites for both the S-protein and antibody, and because only a nanomolar of the reagents was used in the binding study, it is reasonable to assume that the binding of the S-protein or T7 antibody to the S-peptide displayed T7 phage is primarily 1:1 stoichiometry.
This assumption was reinforced by the observation that the average counting rate as well as fluctuation amplitudes of the dye-labeled protein (both the S-protein and anti-T7 antibody) remain constant with increasing phage concentration (after deducting the weak autofluorescence of the phage particle). This suggested that a 1:1 stoichiometry of the binding reaction was observed. A higher binding stoichiometry between the fluorescent protein and the phage would lead to either a reduction of total fluorescence due to the self-quenching of the fluoreorphores on the phage when the interfluorophore distant is closed enough on the phage surface, or a larger counting rate fluctuation due to the increase of fluorescence per phage when the interfluorophore distant is far away on the phage surface. In addition, the fact that the dissociation constant deduced from FCS spectra is insensitive to the initial protein concentration also suggested a 1:1 stoichiometry of binding reaction when the initial protein concentration is less than 10 nanomolar.
Accordingly, data were analyzed by nonlinear least-squares fits to the three-component version of equation 1 using measured diffusion times of 0.19, 0.72, and 3.8 msec for the free dye and dye-S-protein and S-protein–T7 complex, respectively (Table 1). The derived mole fractions for the complex as a function of T7 concentration are shown in Figure 3A [triangle]. The increased mole fraction of bound S-protein as the function of phage concentration clearly indicate the strong binding between S-protein and S-peptide displayed T7 phage at the given concentration. The dissociation constant for the complex was determined by first deducing the value of [P]0 – [T]0y and y/(1 – y) from Figure 3A [triangle] and then fitting the data to equation 4 (Fig. 3B [triangle]). The derived dissociation constant (kd) is 2.2 nM, which matches the literature value of the dissociation constant of the S-protein and S-peptide (Raines et al. 2000).
Figure 3.Figure 3.
Binding of the S-protein to the T7 phage. (A) Mole fraction of bound dye labeled S-protein to the T7 phage vs. the concentration of T7; (B) plot of [T7]0 - [S-protein]0y vs. y/(1 - y) (y = mole fraction). The experimental data were linearly fitted according (more ...)
Control experiments were performed using a wild-type T7 phage that lacks the S-peptide displayed on its surface. Again, the fluorescence correlation curve of dye-S-protein (2 nM in PBS buffer) remains unchanged within the experimental error upon the addition of control T7 phage particles up to 10 nM. Thus, the unchanged mole fraction of the bound S-protein as the function of the phage concentration clearly indicated no binding was observed between the S-protein and control T7 phage.
Characterization of T7–T7 antibody interaction
The S-protein is a small protein with a molecular weight of 11.5 kDa. To examine the interaction of the T7 phage with relatively larger proteins, we chose to study the interaction of the T7 antibody and T7 phage particle. The commercial T7 antibody is a monoclonal antibody directed against a short peptide (11 amino acid, gene 10 leader peptide) from the amino-terminal of the T7 phage capsid protein. By conjugating the Alexa-647 fluorescence dye to the antibody, the interaction of the T7 phage and T7 monoclonal antibody were characterized by FCS. In brief, dilute samples of free dye (1 nM of Alexa-647), purified and dye-labeled antibody (2 nM), as well as purified and dye-labeled T7 phage (0.3 nM) were first analyzed by FCS to obtain their diffusion times (τj) and the free dye mole fraction of labeled antibody (shown in Fig. 4 [triangle]). From fitting the FCS curves (Fig. 4 [triangle], dotted line), the diffusion times for the free dye, dye-labeled antibody, as well as dye-labeled T7 phage are 0.24 msec, 1.4 msec, and 3.8 msec, respectively. Subsequently, a series of fluorescence correlation curves for the T7 antibody–T7 phage complex were recorded with a constant concentration of dye-labeled antibody and variable T7 concentration ranging from 0.9 nM to 2.8 nM. The resulted fluorescence autocorrelation curves were fitted with the three-component model of equation 1 with the diffusion times of free dye, dye-labeled protein, and phage particles and free dye mole fraction held constant. The derived mole fractions dye-labeled antibody bound to the T7 phage as a function of the T7 concentration are shown in Figure 5A [triangle]. The increased mole fraction of the bound anti-T7 antibody as the function of the phage concentration proves the strong binding between the antibody and antigen (T7) at the given concentration. The dissociation constant for the complex was determined by deducing the value of [P]0 – [T]0y and y/(1 – y) from Figure 5A [triangle] and fitting the data to equation 4 (Fig. 5B [triangle]). The derived dissociation constant (kd) is 1.4 nM.
Figure 4.Figure 4.
Normalized FCS spectra of free Alexa-647, the Alexa-647–antibody, antibody–T7 complex, and the autocorrelation fitting curve (three components) for the anti-T7–T7 complex.
Figure 5.Figure 5.
Binding of the T7 antibody to the T7 phage. (A) Mole fraction of bound antibody to the T7 phage vs. the concentration of T7; (B) plot of [T7]0 - [Antibody]0y vs. y/(1 – y) (y = mole fraction). The experimental data were linearly fitted according (more ...)
Unlike the results obtains from the S-protein–T7 system, the fitting curve in Figure 5B [triangle] is deviated from the origin. A dissociation constant of 5.2 nM is obtained when the fitting curve was forced to pass through the origin. However, such a fitting line exhibited unacceptable deviation especially with the data point of a high phage concentration. This may imply that the dynamic range of equation 4 for the anti-T7–T7 system is quite small. Nevertheless, the dissociation constant for anti-T7 is in the range of the average dissociation constants of monoclonal antibodies, which are in the order of a micromole to nanomole (Bender and Gizeli 2003).
FCS studies for M13 phage system
Wild-type M13 and the interaction of the monoclone M13 antibody with M13 were used for the FCS study. As illustrated in Figure 1B [triangle], the filament-like M13 phage is ~800 nm in length and ~8 nm in diameter. Due to its filamentous shape, which can be easily tangled, M13 tends to aggregate at a higher concentration (Lagerkvist et al. 2001). In a typical FCS system, the beam waist radii for the detection volume are less than ~0.5 microns. Thus, the length of a M13 phage exceeds the FCS beam waist (wo) (see Discussion section), and cannot fulfill the condition of d < wo for FCS measurements. In addition to the above factors, the FCS study of M13 phage systems was somewhat more challenging because (1) M13, with a length/diameter aspect ratio of ~100, is not expected to be rigid in solution; thus, rotational motion may contribute to the autocorrelation function; (2) aggregates of labeled M13 could easily overwhelm FCS spectra, because even an individual aggregate of M13 can contribute large nonstatistical amplitude to G with contributions from ~100 nsec to fractions of a second. Consequently, such spectra cannot be reliably modeled by equation 1 (or other standard FCS relations). Because our FCS setup (same as most of the FCS setups nowadays, except the setup used in Dr. Enrico Gratton’s group) engage the use of a hardware correlator, the system does not have the option of editing large amplitude “spikes” in FCS fluctuations prior to autocorrelation. Although the “spikes” can be eliminated from G by some other means of discrimination (e.g., non-multiple τ methods; Muller et al. 2003), the simplest way to work around is to reject any FCS spectra, which contain a “large spike.” Accordingly, several measures have been taken in our studies to address these issues; including (1) we used objective/pinhole combination (b) (i.e., a 40×/0.55 objective plus an 100-μm pinhole) with 633 nm excitation. The beam waist radius (wo) is ~0.8 micron, which is large enough to host an entire M13 particle; (2) we only used low concentration of M13 for FCS studies. Our TEM studies indicated that most of purified M13 particles (via CsCl gradient) remain monomeric when the phage concentration is less than ~1 nM or 1012 pfu/mL; (3) we eliminated any FCS spectra that has larger spikes in its count rate trace (our criterion is to reject spectra that contain spikes with amplitudes or temporal durations exceeding twice the values measured from the purified labeled receptor FCS spectra (during τ determinations) because these fluctuations are associated with phage aggregates).
Diffusion time determination for pure dye, anti-M13–dye conjugation, and M13 phage particles
The diffusion time determination procedure for pure dye and the dye-labeled antibody is the same as previously described in the T7 system except a 40×/0.55 objective and a 100-μm pinhole were used for M13 system. The diffusion times were 0.42, 2.5 msec for free dye and the dye-labeled antibody, respectively (Table 1). The preparation of Alexa-647-labeled M13 was described in the experimental section. The labeled phage was repeatedly purified by spin columns, and analyzed by FCS at 633 nm using the 40×/0.55 objective and the 100-μm pinhole. The obtained FCS spectra were analyzed with the two-component version of equation 1 with a fixed free dye diffusion time and yielded a diffusion time of ~8.2 msec for the M13 monomer (Table 2). The diffusion rate for M13 phage derived from the diffusion time measured and our instrument geometry factor matches the value measure by Rigler’s group at 1.34 msec (40×/1.2, 30-micron pinhole) (Lagerkvist et al. 2001).
Table 2.Table 2.
Results of FCS-labeled M13 phage measurements using 40×/0.55 NA, with 100 μm pinhole at 633 nm
Characterization of the M13–M13 antibody interaction
To quantify the interaction of the anti-M13 with the M13 phage, a stock M13 solution of 4 × 1011 phage per mL (0.7 nM) was used to prepare constant total volume mixtures of stock dye conjugated anti-M13 with M13 phage solution such that the concentration of dye conjugated anti-M13 was held constant at 10 nM, at a variable M13 phage concentration ranging from 0.1 nM to 0.8 nM. After ~1 h of incubation at room temperature, ~30-sec duration FCS measurements were made. Figure 6 [triangle] shows typical autocorrelation functions obtained for the free dye, labeled anti-M13, and anti-M13–M13 complex. After the acquisition of a series of fluorescence correlation curves with a constant concentration of dye-labeled antibody and variable M13 concentrations, data were analyzed by nonlinear least-squares fits to the three-component version of equation 1 using measured diffusion times of 0.42, 2.5, and 8.2 msec for the free dye and dye–anti-M13, and M13, respectively (Fig. 6 [triangle], dotted line).
Figure 6.Figure 6.
Normalized FCS spectra of Alexa-647, the Alexa-647 labeled antibody, the anti-M13–M13 complex, and the autocorrelation fitting curve (three components) for the anti-M13–M13 complex.
It was reasonable to assume that the binding of an antibody molecule to an M13 phage particle has negligible effect on the diffusion time of phage particles due to the huge mass ratio of the phage particle to the antibody; thus, we can use the diffusion time of Alxea-647-labeled M13 phage as the diffusion time of the anti-M13–M13 complex. The derived mole fractions for the complex as a function of M13 concentration are shown in Figure 7 [triangle]. Specific binding (~20% mole antibody–M13 complex) was observed at subnanomolar concentrations of M13; however, the observed mole fractions of bound anti-M13 again did not vary significantly with the increase of M13 concentrations.
Figure 7.Figure 7.
Mole fraction of the bound anti-M13 antibody to M13 phage vs. the concentration of M13.
It is very interesting to notice that the fraction of bound antibody to the phage particles remains unchanged with the increase of the concentration of M13 phage particles, although, clearly, the antibody and M13 phage particles are specifically interacted at the nanomolar range evident by Figure 7 [triangle]. The interaction of the antibody and M13 phage particles at the nanomolar range was also confirmed by an enzyme-linked immunosorbent assay using the concentrations similar to the FCS measurement (data not shown). One possible explanation for the observation of flat binding as the concentration of the phage particle is increased is the antibody-induced phage aggregation. As the phage particle concentration increases, the aggregate concentration also increases. Indeed, we observed more spikes in the count rate traces as the phage concentration was increased. On the other hand, the “larger spike” phenomenon is barely observed in the pure dye-labeled M13 system. Because we had to avoid spikes in our FCS experiments, which were corresponding to the large, aggregated particles, the effective concentration of phage particles (monomer) was not necessary increased proportionately.
On a separated study, the interactions of ATP binding M13 phage with Alexa-647-labeled ATP were examined by FCS with similar experimental conditions and identical setup. The mole fraction of bound ATP to the M13 phage was increased as a function of M13 concentration; thus, the dissociation constant of the ATP binding M13 phage was deduced to be 15 nM. The measured dissociation constant was confirmed by a fluorescence anisotropy experiment (L. Makowski, S. Mandava, J. Bahns, L. Chen, and D.J. Rodi, in prep.). Thus, the method described in this artricle is suitable for characterizing the M13 phage system.
General remarks
In agreement with Rigler’s report, we also found that to use FCS to characterize phage particles, it is extremely important to prepare samples of pure monomeric phage particles. The sample preparation steps must include prelayered cesium chloride density gradient centrifugation after PEG precipitation. Our TEM images indicated that most phage particles in nanomolar concentrations remain as monomeric after CsCl purification.
Although FCS measurements are relatively straight forward in most experimental circumstances, situations do arise that tend to complicate or obscure such measurements. For brevity, we list a few problems encountered, and discuss in greater detail those relevant to these experiments. Generally, problem areas tend to fall into three areas: (1) the sample (e.g., contaminants, solubility, turbidity, aggregation, etc.); (2) the FCS system (e.g., system geometry, alignment, detection volume, location of beam waist in the sample, detector afterpulsing, laser scatter, etc.); (3) the interpretation of the data (e.g., unidentified factors that adversely influence the shape of the autocorrelation function, G(τ), baseline drift, optical pumping effects such as photo-bleaching, triplet decay, blinking, or rotational structure, noise, laser scatter, large particles, background fluorescence, etc.). Such factors can introduce artifacts and adversely influence G(τ). In the worst cases, these can introduce ambiguities in the fitting procedure, particularly when more than a single independent parameter is involved. Control of such effects is critical when multicomponent systems (i.e., containing more than one fluorescent component) are involved, because these require the accurate independent determinations of diffusion times for each component. The major difference of this work from other work in terms of data analysis is that we measured the diffusion times of all components in the system as well as the mole fraction of free dye, and held them as fixed parameters to obtain the relative mole fractions of bound and free target molecules. We deduced the dissociation constants by taking advantage of the phage particle’s much larger mass than the target proteins; thus, its diffusion time can be determined from dye-labeled phage particles. Although this cannot be achieved when FCS is used to study protein–protein or protein–small molecule interactions, we found it is extremely important for the phage system. Otherwise, it is very difficult to deduce a reliable and repeatable mole fraction of bound target proteins using contemporary FCS.
One unique problem associated with FCS measurements of large (e.g., M13 phage) fluorescent particles results when one or more of the physical dimensions (d, long radius) of the particle exceeds one or more of the physical dimensions (wo) of FCS sample volume itself. Typically, FCS measurements satisfy the condition d [double less-than sign] wo. However, when d is approaching wo, it becomes possible for a single large particle to completely “fill” the sample volume in one or more dimensions (at the complete exclusion of other particles), while only a fraction of the particle is ever actually being observed by the FCS system, although it is only the fluorescently labeled portion of the particle that is actually imaged. Although this is less problematic when the labeled particle is rigid, it can be a problem when the labeled particle is flexible (i.e., dynamic conformation changes) because it is then possible for a small fluorescently labeled portion of the particle to enter and leave the FCS sample volume without translational diffusion of the entire particle (i.e., without translation of the particle’s center of mass). This can call into question the validity the derived diffusion time because it no longer relates to bulk translational diffusion of the particle. To address this problem, the observation volume was increased sufficiently to observe the motion of the entire particle in all three dimensions. However, to keep N (the number of observed particles) constant, increases in the system volume would need to be compensated for by proportionate reductions in sample concentration, thus leading to the analysis of increasingly more dilute (and less biologically relevant) solutions as the particle size is increased. In addition, for G to remain statistically significant, sample integration time would need to be increased proportionately. Furthermore, as diffusion times increase, laser intensity would need to be reduced to minimize optical pumping effects (e.g., photobleaching, triplet formation, etc.) during particle residence time. It is worth noting that when d is approaching wo, the possibility of optically hindered diffusion, trapped Brownian motion becomes more probable (Osborne et al. 1998; Chirico et al. 2002).
Conclusions
We have performed FCS measurements of labeled phage diffusion times, and determined equilibrium dissociation constants for labeled complexes of T7 and M13 phage. Although experiments with T7 were relatively straightforward, two major complications were encountered with M13: (1) difficulties obtaining a sample suitable for determining τ for the M13 monomer; (2) volume sampling appropriate for M13 translational diffusion. Modifications to experiments and methodology are discussed to handle key problems associated with the FCS of large M13 phage particles.
>Materials and methods
Materials and methods
Phage particle preparation and purification
A derivative of wild-type T7 phage, T7Select415-1 with 415 copies of S-peptide displayed on the surface of phage particles as a fusion to the capsid proteins (hereafter S-peptide T7 phage) was obtained from Novagen. Wild-type M13 phage particles were purchased from Progen. In brief, both phage particles were amplified by infecting a mid-log-phase Escherichia coli culture (250 mL, OD600 0.6). After cell lysis, the phage lysate was made in 0.5 M of NaCl and clarified by centrifugation. The phage particles were first purified by PEG precipitation twice, then isolated by banding in a cesium chloride density gradient. T7 phage particles band at the layer of 42% CsCl, while M13 phage particles band at the layer of 21% CsCl. The resulting phage particles were then dialyzed against PBS buffer and concentrated to 1 × 1012 pfu/mL (~1 nM) for FCS studies. Phage titers were determined by plaque assays.
Conjugation of phage particles, as well as anti-M13 and anti-T7 tag monoclonal antibodies with Alexa-647 dye
Purified phage particles as well as anti-M13 and anti-T7 tag monoclonal antibodies (Amersham Biosciences) were conjugated to the Alexa-647 dye by using an Alexa-647 labeling kit from Molecular Probes according to the protocols provided. In brief, a 90-μL PBS buffer solution containing M13 or T7 phage particles (1012 pfu/ mL) was mixed with add 10 μL of 1 M bicarbonate (pH ~8.3). The resulting phage solution was transferred into a vial containing reactive dye (Alexa Fluor 647 carboxylic acid, succinimidyl ester). The mixture was gently agitated for one half hour in the dark at room temperature. Free dye was removed by size-exclusion spin column provided with the kit. The labeled phage particles were further purified using a gel filtration spin-column (Centrisep, Princeton Separations). For the conjugation of antibodies to the dye, a similar procedure was used with the following differences: 0.1-mg antibodies were resuspended in 450 μL PBS buffer and mixed with 50 μL of 1 M bicarbonate (pH ~8.3). The labeled proteins were finally eluated from free dye by size-exclusion spin column. The degree of labeling was calculated to be 4 moles of Alexa-647 per mole of antibody.
Observation of T7 and M13 phage particles under transmission electron microscope (TEM)
Electronic microscopic images were obtained from a Philips CM-120 transmission electron microscope, operated at 120 kV. The samples were supported on 400 mesh carbon coated grids, freshly glow discharged (Evaporator: Edwards Auto 306) for 45 sec. Specimens were negatively stained by applying a drop (5 μL) of phage particles to the grid. The grid was then washed with water, stained for 30 sec with aqueous solution of uranyl acetate (1%), and then wicked off with filter paper and allowed to dry. Micrographs were recorded on a Gatan CCD digital camera.
FCS setup
We set up a fluorescence auto correlation spectrometry base on a modified confocal/near field scanning optical (NSOM)/atomic force microscope (Nanonics-100, Nanonics), using an inverted microscope (IX 70, Olympus). A 5-mW Helium Neon laser (633 nm, Spectrooptics) was used for the excitation. The laser beam is directed to the microscope through a laser port equipped with a dichroic mirror. The beam output width is expanded to fill the back aperture of the microscope objective (a 60×/1.2 NA, water immersion, and a 40×/0.55 NA, Olympus) to provide a diffraction-limited spot. The fluorescence signal of the sample is collected with the same objective and is transmitted by a band pass filter (Chroma, Ltd.) to reduce the background signal. A pinhole (50 or 100 μm) was installed in an image plane of the microscope to discriminate against out of focus signals. The collected fluorescence light is then focused onto an avalanche photo diode (SPCM-AQR-15 (EG&G), Perkin-Elmer) detector. Autocorrelation functions are generated on line with a high-speed multiple tau digital correlator (ALV-6010, ALV-GmbH) and fitted off-line to the autocorrelation function equation. The diffusion time relaxation constant is related to the diffusion constant D by τ = wo2/4D, where wo is the lateral radius of the detection volume. The power of the laser beam could be adjusted with neutral density filters to provide high fluorescence signals per molecule, yet low enough to avoid significant photobleaching.
The geometric parameters for our home-built FCS apparatus were determined by using dilute (0.1–10 nM) standard dye (Alexa-647) with 633-nm excitation. A holographic filter (Kaiser Optical System) combined with a long pass (λ > 645 nm) filter were used to block the reflect laser beam and transmit fluorescence (645–700 nm). FCS experiments were performed with objective/pinhole combinations: (a) 60×/1.2 (water immersion) objective with a 50-micron pinhole or (b) 40×/0.55 objective with an 100-micron pinhole. Ratios of measured diffusion times for free dye were found to be ~2.4 for (b) relative to (a). Combination (b) was used to prepare a sampling volume appropriate for measuring M13 phage diffusion. The beam waist radii were estimated to be ~0.5 and ~0.8 microns for (a) and (b), respectively.
FCS model
FCS measures fluctuations of the fluorescence intensity within minute (femtoliter) detection volumes and a temporal range from microsecond to seconds. The random fluctuations represented by the autocorrelation function of the fluctuation signal can be analyzed using equation 1 (Hess et al. 2002),
equation M1
(1)
where
equation M2
represent the relative fluorescence efficiencies Fj and molar fractions Yj of j; τT is the triplet state lifetime; T is the triplet fraction; τj is the characteristic diffusion time of component j; total detected signal; IB is the intensity of the uncorrelated background signal; ω0 and z0 are the radial and axial waist radii defining the detection volume; and α is the compensate parameter for artificial effects in the autocorrelation (such as photobleaching) with time constants exceeding 1 sec.
FCS measurements
For all binding experiments, we measured the free dye first to determine its diffusion time using a one-component model, including the triplet term (equation 1). Because it was impossible to eliminate all free dye when purifying the dye labeled protein, we obtained the diffusion time of dye-labeled protein with a two-component model (equation 1) that contained the independently determined free-dye diffusion time as fixed parameter, assuming the fluorescence efficiencies (Fj) and triplet terms of Alexa-647 and Alexa-647-labeled protein are the same. For protein–phage complex systems, the FCS autocorrelation curves were analyzed with a three-component model. In this case, the diffusion times for the phage and phage–protein complex were also assumed to be the same due to the negligible mass difference between phage particles and protein–phage complexes. With predetermined diffusion times and free dye mole fraction held fixed in this model, the relative mole fractions of bound and unbound phage with target molecules in the solution were derived from nonlinear least squares fits to the experimental G(τ).
Data analysis for dissociation constant measurements
To measure the dissociation constant of the dye labeled target protein (T) with phage particle (P), T and P were first incubated sufficiently long to reach equilibrium (0.5–1 h). Because it is cumbersome to produce a large quantity of purified phage and because phage particles [P] tends to aggregate (Xu et al. 2001) at higher concentrations (which is not suitable for FCS studies), all experiments associated with dissociation constant measurements are under the conditions of maintaining a constant concentration of T (in the order of nM) while varying the concentration of P in the approximate concentration range of 0.3 nM to 3 nM. Thus, the conventional pseudofirst-order condition ([P] [dbl greater-than sign] [T]) used for binding studies could not be satisfied here. Under these conditions, the interaction between the target protein (T) and ligand displayed phage particles (P) can be described by equation 2 provided a 1:1 stoichiometry:
equation M3
(2)
The autocorrelation function can be expressed in a two- or three-component model (equation 1) [J = 1(free dye), 2(dye-labeled protein), 3(dye-protein: phage complex)]. Consequently, the mole fraction (y) of bound T can be derived from the fitting of the autocorrelation curve. The dissociation constant can be written as the following equation (equation 3):
equation M4
(3)
where [T]0 and [P]0 are the initial (total) concentrations of dye labeled target molecules and phage particles, respectively, and y is the fraction of bound target molecule against all target molecules in the solution:
equation M5
Rearrange equation 3 as equation 4:
equation M6
(4)
A plot of [P]0 – [T]0y against y/(1 – y) can be linearly fitted according to equation 4. Consequently, Kd can be determined from the slope of the plot.
 
Acknowledgments
Funding for this work was provided by the Argonne National Laboratory LDRD (Laboratory Directed Research and Development) grant. We thank Ms. Yimei Chen at the University of Chicago for helping with the TEM imaging. We are grateful to Dr. Peter Goodwin from Los Alamos National Laboratory for helpful suggestions while setting up our FSC apparatus. We also acknowledge Dr. Enrico Gratton (University of Illinois at Urbana-Champaign), Dr. Lee Makowski (Argonne), and Dr. Brian Kay (Argonne) for inspiring discussions.
The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC section 1734 solely to indicate this fact.
 
Notes
Article and publication date are at http://www.proteinscience.org/cgi/doi/10.1110/ps.04695704.
>References
References
  • Bender, S.K. and Gizeli, E. 2003. Comparative study of IgG binding to proteins G and A: Nonequilibrium kinetic and binding constant determination with the acoustic waveguide device. Anal. Chem. 75: 835–842. [PubMed].
  • Castagnoli, L., Zucconi, A., Quondam, M., Rossi, M., Vaccaro, P., Panni, S., Paoluzi, S., Santonico, E., Dente, L., and Cesareni, G. 2001. Alternative bacteriophage display systems. Comb. Chem. High Throughput Screen. 4: 121–133. [PubMed].
  • Chirico, G., Fumagalli, C., and Baldini, G. 2002. Trapped Brownian motion in single- and two-photon fluorescence correlation experiments. J. Phys. B 106: 2508–2519.
  • Hess, S.T., Huang, S., Heikal, A.A., and Webb, W.W. 2002. Biological and chemical application of fluorescence correlation spectroscopy: A review. Biochemistry 41: 697–705. [PubMed].
  • Kay, B.K., Kurakin, A., and Hyde-DeRuyscher, R. 1998. From peptides to drugs via phage-display. Drug Discov. Today 3: 370–378.
  • Lagerkvist, A.C., Foldes-Papp, A., Persson, M.A. and Rigler, R. 2001. Fluorescence correlation spectroscopy as a method for assessment of interactions between phage displaying antibodies and soluble antigen. Protein Sci. 10: 1522–1528. [PubMed].
  • Lumma, D., Keller, S., Vilgis, T., and Radler, J.O. 2003. Dynamics of large semiflexible chains probed by fluorescence correlation spectroscopy. Phys. Rev. Lett. 90: 218301–1–218301–4. [PubMed].
  • Muller, J.D., Chen, Y., and Gratton, E. 2003. Fluorescence correlation spectroscopy. In Methods in enzymology (eds. G. Marriott and I Parker), Vol. 361, pp. 69–93. Academic Press, New York.
  • Osborne, M.A., Balasubramanian, S., Furey, W.S., and Klenerman, D. 1998. Optically biased diffusion of single molecules studied by confocal fluorescence microscopy. J. Phys. Chem. B 102: 3160–3167.
  • Raines, R.T., McCormick, M., VanOosbree, T.R., and Mierendorf, R.C. 2000. The S.Tag fusion system for protein purification. Methods Enzymol. 326: 362–376. [PubMed].
  • Rigler, R. and Elson, E.S., eds. 2001. Fluorescence correlation spectroscopy: Theory and applications. Springer-Verlag, Berlin.
  • Rodi, D.J., Makowski, L., and Kay, B.K. 2002. One from column A and two from column B: The benefits of phage display in molecular-recognition studies. Curr. Opin. Chem. Biol. 6: 92–96. [PubMed].
  • Schwille, P., Kummer, S., Heikal, A.A., Moerner, W.E., and Webb, W.W. 2000. Fluorescence correlation spectroscopy reveals fast optical excitation-driven in tramolecular dynamics of yellow fluorescent proteins. Proc. Natl. Acad. Sci. 97: 151–156. [PubMed].
  • Xu, H., Frank, J., Trier, U., Hammer, S., Schroder, W., Behlke, J., Schafer-Korting, M., Holzwarth, J.F., and Saenger, W. 2001. Interaction of fluorescence labeled single-stranded DNA with hexameric DNA-helicase repA: A photon and fluorescence correlation spectroscopy study. Biochemistry 40: 7211–7218. [PubMed].
Articles from Protein Science : A Publication of the Protein Society are provided here courtesy of
The Protein Society