Fluorescence Lifetime Cross Correlation Spectroscopy Resolves

Jun 29, 2010 - Jose M. Paredes , Luis Crovetto , Angel Orte , Sergio G. Lopez , Eva M. ... Li Wang , Hong Li , Xueqin Lv , Tong Chen , Ruili Li , Yiqu...
22 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 2010, 82, 6415–6421

Fluorescence Lifetime Cross Correlation Spectroscopy Resolves EGFR and Antagonist Interaction in Live Cells Jiji Chen and Joseph Irudayaraj* Birck Nanotechnology & Bindley Bioscience Center, Department of Agricultural & Biological Engineering, Purdue University, West Lafayette, Indiana 47906 Fluorescence correlation or cross-correlation spectroscopy (FCS or FCCS), a single molecule technique, has the ability to provide highly sensitive information on interaction and dynamics of biomolecules both in vitro and in vivo. However, the inherent drawback of FCS is that species with similar molecular weight could not be differentiated. Although FCCS could resolve this through cross-correlation, it suffers from nonideal confocal volume overlap and spectral cross-talk which limits its application. In this work, we demonstrate for the first time the applicability of fluorescence lifetime correlation spectroscopy (FLCS) to monitor the interaction of an antagonist antibody with the epidermal growth factor receptor (EGFR) in live cells. As a proof of concept, we demonstrate the interaction of Cy5 labeled IgG and Alexa633 labeled anti-IgG using a single laser source (636 nm excitation) in vitro. The autocorrelation functions were separated based on their respective lifetime with a single detector and their Kd value was determined to be 11 ( 3 nM. An in vivo application constituting the interaction of EGFR neutralizing antibody labeled with Alexa488 and EGFR-GFP in live HEK293 cells was successfully demonstrated. The binding specificity of EGFR neutralizing antibody was confirmed by fluorescence lifetime crosscorrelation measurements and fluorescence lifetime imaging (FLIM). The dissociation constant of this complex was found to be 9.2 ( 2.7 nM. A quantitative assessment of receptor density calculations show that the density of EGFR significantly decreased, from 540 ( 64 receptors/ µm2 to 38 ( 7 receptors/µm2 upon addition of the neutralizing EGFR antibody, indicating that the antagonist could induce receptor internalization. The demonstrated work not only opens up new opportunities in studying protein-protein interactions in solutions and in live cells but also provide new insights in biology to understand how the antagonists influence EGFR through live cell quantification and imaging. Fluorescence correlation spectroscopy (FCS) provides single molecule information,1-3 and in recent years, has become a powerful technique for studying protein,4,5 nucleic acids,6,7 and * To whom correspondence should be addressed. E-mail: josephi@ purdue.edu. (1) Rigler, R.; Elson, E. Fluorescence Correlation Spectroscopy: Theory and Applications; Springer: New York, 2001. 10.1021/ac101236t  2010 American Chemical Society Published on Web 06/29/2010

receptor-ligand interactions8-10 in live cells. However, FCS can only separate two diffusing species differing in their diffusion times by a factor of at least 1.6,11 which translates to an approximate difference in molecular weight by a factor of at least 4 in the multicomponent system. This restriction could be overcome by dual-color fluorescence cross correlation spectroscopy (FCCS) where the two interacting molecules could be labeled with two distinct fluorophores12 excitable by different sources and the signals collected in the different detectors could be cross correlated and evaluated independent of their molecular mass. However, for a successful FCCS experiment, a perfect overlap between the two different excitation laser beams is necessary and the spectral overlap and signal bleed through must be accounted for appropriately.13 FCCS could also be performed with a single wavelength excitation (SW-FCCS)14,15 or with a two photon excitation.16 SW-FCCS is possible only with special fluorophores since it requires similar excitation spectra but spectrally different emission along with mathematical considerations to assess crossexcitation and cross-talk. Two photon excitation FCCS is possible but the high cost of the laser source limits its applicability. Fluorescence lifetime correlation spectroscopy (FLCS), integrates FCS with the time-correlated single photon counting (TCSPC), to separate FCS autocorrelation based on different lifetimes.17-19 (2) Maiti, S.; Haupts, U.; Webb, W. W. Proc. Natl. Acad. Sci. U. S. A. 1997, 94, 11753–11757. (3) Qian, H.; Elson, E. L. Appl. Opt. 1991, 30, 1185–1195. (4) Chen, Y.; Muller, J. D.; Ruan, Q.; Gratton, E. Biophys. J. 2002, 82, 133– 144. (5) Dittrich, P.; Malvezzi-Campeggi, F.; Jahnz, M.; Schwille, P. Biol. Chem. 2001, 382, 491–494. (6) Winter, H.; Korn, K.; Rigler, R. Curr. Pharm. Biotechnol. 2004, 5, 191– 197. (7) Ohrt, T.; Schwille, P. Curr. Pharm. Des. 2008, 14, 3674–3685. (8) Rauer, B.; Neumann, E.; Widengren, J.; Rigler, R. Biophys. Chem. 1996, 58, 3–12. (9) Whelan, R. J.; Wohland, T.; Neumann, L.; Huang, B.; Kobilka, B. K.; Zare, R. N. Anal. Chem. 2002, 74, 4570–4576. (10) Hegener, O.; Jordan, R.; Haberlein, H. Biol. Chem. 2002, 383, 1801–1807. (11) Meseth, U.; Wohland, T.; Rigler, R.; Vogel, H. Biophys. J. 1999, 76, 1619– 1631. (12) Schwille, P.; Meyer-Almes, F. J.; Rigler, R. Biophys. J. 1997, 72, 1878– 1886. (13) Bacia, K.; Schwille, P. Nat. Protoc. 2007, 2, 2842–2856. (14) Hwang, L. C.; Gosch, M.; Lasser, T.; Wohland, T. Biophys. J. 2006, 91, 715–727. (15) Hwang, L. C.; Wohland, T. J. Chem. Phys. 2005, 122, 114708. (16) Heinze, K. G.; Koltermann, A.; Schwille, P. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 10377–10382. (17) Kapusta, P.; Wahl, M.; Benda, A.; Hof, M.; Enderlein, J. J Fluoresc 2007, 17, 43–48.

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6415

The unique advantage of FLCS is that it could distinguish species with similar diffusion times not resolvable by traditional FCS11 with potential to separate up to four signal components with a single excitation and detector setup.20 Recently, FLCS was used to evaluate metal-enhanced fluorescence interactions and to separate multiple species with different lifetimes in a solution mixture.21 This technique has also been used to remove the effect of after pulsing of the autocorrelation function which could interfere with FCS analysis in a short time scale.22 A significantly useful but under explored concept is the ability to obtain crosscorrelation information from signal components that can be differentiated based on their lifetime values. Benda A et al. used FLCS to study dye exchange between unilamellar vesicles by cross-correlation analysis.23 By making use of the lifetimes of fluorophores in the folded and unfolded domains of DNA, the dynamics of DNA condensation was revealed through auto- and cross-correlation generated by FLCS.24,25 The epidermal growth factor receptor (EGFR) which is usually overexpressed in the epithelia cancers26 is thought to be associated with tumor development and poor prognosis. Therapeutic anticancer agents that target and inhibit EGFR family are under intense study in biomedicine. These agents include the widely used humanized monoclonal antibody Herceptin which specifically targets the HER-2 overexpressing breast cancer cells27 and the neutralizing monoclonal antibodies that target EGFR.28,29 A number of antibodies directed against EGFR are in different phase of clinical evaluation30 including cetuximab,31 antibody drug IMC11F832 and gefitinib,33 all of which have demonstrated antitumor effects in patients. These antibodies could bind specifically to the extracellular domain of EGFR and block the epidermal growth factor (EGF) binding and downstream signaling pathways resulting in the inhibition of cell growth and induction of apoptosis. Despite the choice of drugs that act as inhibitors of EGFR signaling, the interaction between EGFR and its antagonist antibody is still unclear. (18) Gregor, I.; Enderlein, J. Photochem. Photobiol. Sci. 2007, 6, 13–18. (19) Benda, A.; Fagul’ova, V.; Deyneka, A.; Enderlein, J.; Hof, M. Langmuir 2006, 22, 9580–9585. (20) Ruttinger, S.; Kapusta, P.; Patting, M.; Wahl, M.; Macdonald, R. J. Fluoresc. 2009. (21) Ray, K.; Zhang, J.; Lakowicz, J. R. Anal. Chem. 2008, 80, 7313–7318. (22) Enderlein, J.; Gregor, I. Rev. Sci. Instrum. 2005, 76, -. (23) Benda, A.; Hof, M.; Wahl, M.; Patting, M.; Erdmann, R.; Kapusta, P. Rev. Sci. Instrum. 2005, 76, -. (24) Humpolickova, J.; Beranova, L.; Stepanek, M.; Benda, A.; Prochazka, K.; Hof, M. J. Phys. Chem. B 2008, 112, 16823–16829. (25) Humpolickova, J.; Benda, A.; Sykora, J.; Machan, R.; Kral, T.; Gasinska, B.; Enderlein, J.; Hof, M. Biophys. J. 2008, 94, L17–19. (26) Woodburn, J. R. Pharmacol. Ther. 1999, 82, 241–250. (27) Slamon, D. J.; Leyland-Jones, B.; Shak, S.; Fuchs, H.; Paton, V.; Bajamonde, A.; Fleming, T.; Eiermann, W.; Wolter, J.; Pegram, M.; Baselga, J.; Norton, L. N. Engl. J. Med. 2001, 344, 783–792. (28) Salomon, D. S.; Brandt, R.; Ciardiello, F.; Normanno, N. Crit. Rev. Oncol. Hematol. 1995, 19, 183–232. (29) Li, S.; Kussie, P.; Ferguson, K. M. Structure 2008, 16, 216–227. (30) Fletcher, L. Nat. Biotechnol. 2001, 19, 599–600. (31) Herbst, R. S.; Kim, E. S.; Harari, P. M. Expert Opin. Biol. Ther. 2001, 1, 719–732. (32) Lu, D.; Zhang, H.; Koo, H.; Tonra, J.; Balderes, P.; Prewett, M.; Corcoran, E.; Mangalampalli, V.; Bassi, R.; Anselma, D.; Patel, D.; Kang, X.; Ludwig, D. L.; Hicklin, D. J.; Bohlen, P.; Witte, L.; Zhu, Z. J. Biol. Chem. 2005, 280, 19665–19672. (33) Bruzzese, F.; Di Gennaro, E.; Avallone, A.; Pepe, S.; Arra, C.; Caraglia, M.; Tagliaferri, P.; Budillon, A. Clin. Cancer Res. 2006, 12, 617–625.

6416

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

Although the capability of performing cross-correlation by FLCS has been demonstrated, no reports have yet demonstrated the utility of FLCS in studying protein interaction in vitro or for that matter in live cells. In this study for the first time, we demonstrate the FLCS concept for cross-correlation experiments using lifetime data to study protein-receptor interaction in live cells using a single wavelength excitation with a single detector. As a proof of concept, first the binding of IgG labeled with Cy5 and anti-IgG labeled with Alexa633 was demonstrated in solution. We note that the amplitude of cross-correlation is directly proportional to the concentration of the binding complexes. Living cell experiments were then conducted to evaluate the interaction of the neutralizing anti-EGFR Alexa488 labeled antibody with EGFRGFP transfected HEK293 cells. The dissociation constant calculated using FLCS for this reaction was 9.2 nM. Furthermore, by estimating the receptor density on the cell surface we found that the neutralizing antibody could accelerate the endocytosis of EGFR on the cell membrane. Compared to FCCS, the FLCS strategy is simpler because there was no need to correct for spectral cross-talk, signal bleedthrough, and nonideal volume overlap of the two excitation sources. Thus, FLCS could potentially become a powerful technique for investigating biomolecule interaction in live cells using a single excitation. EXPERIMENTAL SECTION Cell Culture and Transient Gene Expression. pcDNA3EGFR-GFP plasmid obtained from Dr. John G. Koland’s Laboratory (University of Iowa, College of Medicine) was used to transfect HEK293 cells. Anti-EGFR, neutralizing, clone LA1, Alexa Fluor 488 conjugate (anti-EGFR-Alexa488) was purchased from Millipore (Billerica, MA). Cy5 conjugated human IgG (IgG-Cy5) was purchased from Jackson Immunoresearch Laboratories Inc. (West Grove, PA) and Alexa633 conjugated anti-IgG (anti-IgGAlexa633) was purchased from Invitrogen Inc. (Carlsbad, CA). IgG and anti-IgG was mixed in a 1:1 ratio and incubated at room temperature for different time periods to form a complex for initial experiments. HEK293 cells were grown in RPMI-1640 (ATCC) media supplemented with 10% fetal bovine serum. Cells were maintained at 37 °C with 5% CO2. For living-cell FLCS experiments, cells were seeded onto sterilized No. l coverslip (VWR International, Batavia, IL) and placed inside a 6-well plate. After the cells reached 80% confluence, DNA was transfected using 1 µL Lipofectamine 2000 (Invitrogen, Carlsbad, CA), according to manufacturer’s specifications. The amount of DNA used per well was 250 ng of EGFR-GFP. After transfection, cells were cultured for another 36 h to let the EGFR-GFP transient express in HEK293 cells. Then the cells were incubated with 5 nM antiEGFR-Alexa488 antibody for 50 min for the following studies. FLCS and Fluorescence Lifetime Imaging (FLIM) Instrumentation. FLCS and FLIM were performed using time-resolved scanning confocal microscopy (Microtime 200 from Picoquant GmbH, Berlin, Germany). Details of the instrumentation are presented elsewhere.34 Two picosecond pulsing lasers at 467 and 636 nm were used as excitation sources (20 MHz, Picoquant, Berlin, Germany). The laser beam was delivered to the sample through an apochromatic 60×, 1.2 NA water immersion objective (34) Chen, J.; Irudayaraj, J. ACS Nano 2009, 3, 4071–4079.

and the emitted fluorescence was collected using the same objective and separated from the excitation beam by a dual band dichroic (z467/638rpc, Chroma). A 50 µm pinhole was used to reject the off-focus photons from the excitation volume, and the overall fluorescence was collected and filtered by suitable emission filters before being detected by single photon avalanche photodiodes (SPAD) (SPCM-AQR, PerkinElmer Inc.). Fluorescence was measured using the time-correlated single photon counting (TCSPC) module in the time-tagged time-resolved (TTTR) mode (Time Harp200, PicoQuant GmbH, Germany). Data Analysis for FCS. Fluorescence fluctuations of δF(t) around the average fluorescence 〈F〉 were recorded in real time and the normalized autocorrelation was calculated as in eq 1.

G(τ) )

〈δF(t)

× δF(t + τ)〉

G(τ) in eq 2, can be rewritten to obtain a continuous distribution of diffusion times using MEMFCS as follows: G(τ) )

∫ R (1 + ττ )

(

-1

i

D

1 + κ2

τ τD

)

-1/2

dτD

(4)

S is defined as, S)

∑ p In p , where p i

i

i

) Ri /

i

∑R

i

(5)

Data Analysis for FLCS. The separation of mixed components in FLCS is based on their different TCSPC histograms and the fluorescence lifetime was obtained by fitting the TCSPC histograms with the multiexponential model.

(1)

〈F(t)〉2

∑ R exp(- τt ) n

where δF(t) ) F(t) - 〈F(t)〉. The autocorrelation curve of fluorescence molecules diffusing in solution was fitted to a 3D diffusion model using one or two components with the SymphoTime software (PicoQuant GmbH Berlin, Germany) and Origin Lab using eq 2

G(τ) )

∑ N1 (1 + ττ )

(

-1

i

Di

1+

τ τDi.κ2

)

-1/2

(2)

Here, Ni and τDi are the number of fluorescent molecules in the detection volume and diffusion time of component i, respectively. κ is defined as the ratio of the axial beam size z and radius ω of the laser beam. For calibration of the effective confocal volume, aqueous solution of Rhodamine 123 and Atto 655 (Invitrogen, Carlsbad, CA) was used. Assuming a 3D Gaussian observation volume approximated by Veff ) π3/2ω2z, the confocal volume of the 467 and 636 nm laser excitation was estimated to be 0.38 and 0.96 fL respectively. Diffusion of free anti-EGFR above the cell surface was described by the 3D diffusion function, whereas diffusion of bound anti-EGFR on the cell membrane was fitted by the 2D diffusion function. Hence, the autocorrelation of anti-EGFR on the cell membrane was given by,

G(τ) )

(

(

τ 1 (1 - y) 1 + free 〈N〉 τd

)( -1

1+

τ τdfree.κ2

(

)

-1/2

y 1+

τ

i

(6)

i

i)1

Where the values of the lifetime, τi and amplitude Ri are obtained through nonlinear least-squares fitting using the Symphotime software. The goodness-of-fit was determined by the χ2 value. For the in vitro and in vivo experiments, two different lifetime components exist and the measured fluorescence intensity Ij was expressed as follows:17-19 Ij(t) ) w1(t)p1j + w2(t)p2j

(7)

Where w1 and w2 denote the respective contributions from two different fluorescent species obtained from the total fluorescence signal, pj1,2 is the normalized fluorescence decay histogram. The pj1,2 patterns were obtained by fitting the measured histogram of the mixture by an iterative convolution technique using the instrument response function (IRF).18 By introducing each signal component, an associated filter L function Fj(i) was defined with the property: 〈∑j)1 Fj(i)Ij〉 ) w(i), where L is the total number of TCSPC channels, and the brackets denote averaging over an infinite number of measurements. In addition, F(i) j has been chosen to minimize the relative errors L

+

τdbound

I(t) )

〈(

))

∑f

(i) j Ij

- w(i))2〉

(8)

j)1

-1

(3)

Where τfree is the diffusion time for the unbound anti-EGFR, d is the diffusion time of the bound anti-EGFR antibody and y is the fraction of bound anti-EGFR antibody diffusing at time τdbound. A multicomponent model using the maximum entropy method, MEMFCS35 was employed to further validate the autocorrelation functions generated from FLCS. MEMFCS is based on minimizing χ2 as well as maximizing the entropy S to obtain an optimal fit when species with different diffusion times are involved. Thus,

The FLCS filters then can be numerically calculated from Ij and from the TCSPC fluorescence decay histogram pj1,2 using matrices.36 The auto and cross-correlation functions (G12) can then be calculated and the concentration of bound molecules (C12) can be estimated (Supporting Information (SI) eqs S1-S3). The analysis was performed using Symphotime software (Picoquant, version 5.13) and Origin Lab. Temporal Analysis of Membrane EGFR Expression. EGFRGFP transfected cells were treated with 5 nM antagonist antibody at different time points (0, 5, 30, 60, 120 min). The membrane EGFR receptor expression level was quantitatively determined

(35) Sengupta, P.; Garai, K.; Balaji, J.; Periasamy, N.; Maiti, S. Biophys. J. 2003, 84, 1977–1984.

(36) Bohmer, M.; Wahl, M.; Rahn, H. J.; Erdmann, R.; Enderlein, J. Chem. Phys. Lett. 2002, 353, 439–445.

τdbound

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6417

Figure 1. (A) Autocorrelation functions of IgG-Cy5 (b) and anti-IgGAlexa633 (9) were separated from a mixture through FLCS based on their different lifetime values. Their respective fits are shown in solid line (Red). (B) Cross-correlation plot of IgG-Cy5 binding with anti-IgG-Alexa633 through FLCS at different time periods.

through cross-sectional fluorescence intensity analysis. All membrane fluorescence intensity was normalized using untreated cells as 100% expression level. Data obtained were expressed as mean ± standard deviation and analyzed by one-way ANOVA. A difference between means was considered significant if the p value is less than 0.05. RESULTS AND DISCUSSION Binding Analysis of IgG and Anti-IgG. As a proof of concept, we first demonstrate the utility of FLCS to study protein-protein interaction in vitro using a 633 nm excitation source and a single detector channel. IgG labeled with Cy5 and anti-IgG labeled with Alexa633 were first used in the demonstration to evaluate the binding constant. The lifetime of IgG-Cy5 and anti-IgG-Alexa633 was determined as 1.57 and 3.35 ns, respectively, by fitting the time correlated single photon counting (TCSPC) data with single exponential equation. IgG-Cy5 and anti-IgG-Alexa633 were mixed in a 1:1 ratio and incubated for 50 min to form an IgG-Anti-IgG complex and an autocorrelation analysis was conducted (SI Figure S1). It was not possible to resolve the IgG and anti-IgG diffusion times from this mixture through autocorrelation (SI Figure S1) because their molecular weights are similar (∼150 KDa). However, the autocorrelation curves can be obtained from the TCSPC histogram based on their lifetime using statistical filters derived from TCSPC histogram for each antibody in the solution mixture. The separated autocorrelation for IgG and anti-IgG is shown in Figure 1A. The concentration and diffusion time of IgG and anti-IgG could be obtained by fitting the autocorrelation function using eq 2 with a single component. The respective diffusion time calculated for IgGCy5 and anti-IgG-Alexa633 was 0.57 and 0.54 ms, respectively. The IgG and anti-IgG mixture could not be distinguished by FCS autocorrelation function however, FLCS analysis could clearly differentiate these entities based on their lifetime values. The concentration of IgG and anti-IgG recovered by FLCS was estimated to be 1.5 nM and 1.9 nM, close to 1:1 ratio, demonstrating that lifetime correlation could be a reliable technique for separating autocorrelation functions from a solution mixture. Subsequently, IgG-Cy5 and anti-IgG-Alexa633 were incubated for different time periods. A drop of the solution mixture was placed on the coverslip and measurement was taken 30 µm above the coverslip to exclude nonspecific binding between the antibodies and the coverslip. The cross-correlation curve was calculated using eq S1 (Supporting Information) using SymphoTime as shown in Figure 1B. The gradual increase in cross-correlation amplitude with time shows an increase in the concentration of 6418

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

bound (IgG-anti-IgG) complexes in solution since the amplitude of cross-correlation is proportional to the concentration of the binding complexes.13 The dissociation constant of IgG binding to anti-IgG can be estimated from the equation [IgG][Anti-IgG]/ [IgG-Anti-IgG]. The dissociation constant calculated as 11 ± 3 nM using FLCS was in excellent agreement with the results from dual-color fluorescence cross-correlation spectroscopy.37 The value calculated using FCCS was 15 ± 2 nM. EGFR Interaction with Neutralizing Antibody by FLCS. Based on our in vitro demonstration of FLCS analysis to assess protein-protein interaction, we develop a procedure to study the kinetics of binding between EGFR and its antagonist antibody in live cells. Human embryonic kidney 293 cells (HEK293) was transiently transfected with pcDNA3-EGFR-GFP plasmid to express EGFR-GFP. It is worth mentioning that the HEK293 cells were transfected with a small amount of plasmid DNA (250 ng) so that the EGFR-GFP is expressed in a relatively lower concentration to render FCS measurement possible. A neutralizing EGFR antibody38,39 was chosen as an inhibitor of the stimulation of EGFR by blocking the ligand/receptor interaction due to binding of EGFR and its antagonist. The mechanism of antagonist antibodies inhibiting EGFR is thought to be related in its ability to inhibit EGF binding, down regulate receptor expression, and interfere with the formation of receptor dimerization. The binding between EGFR and neutralizing antibody was confirmed by fluorescence lifetime imaging (FLIM) (Figure 2A, merge image). The TCSPC decay curve corresponding to the fluorescence lifetime image was fitted using a biexponential decay function using the instrument response function (IRF) to extract fluorescence lifetimes of different components as shown in Figure 2B. Two distinct lifetime values were possible corresponding to anti-EGFR-Alexa488 and EGFR-GFP (Figure 2C). The fluorescence lifetime image was then generated based on the fitting results with a maximum likelihood estimator. The lifetime image was shown in a red-green-blue (RGB) mode where antiEGFR-Alexa488 is shown in green and EGFR-GFP in red (Figure 2A). It is found that the majority of EGFR localized on the cell surface and the yellow color in the merged image pointing to potential binding sites between anti-EGFR-Alexa488 and EGFRGFP. A small portion of antibody-EGFR complex was found in the cytoplasm suggesting that the antibody-EGFR complex could be internalized after binding. The expression levels of protein coded by plasmid DNA that was introduced through transient transfection is typically much higher compared to its naive condition since they can replicate independent of the chromosomal DNA. As shown by the “+” symbol in SI Figure S3A, FLCS measurements were taken from a relatively dim location (lower fluorescence intensity) on the cell membrane in its physiological state. The autocorrelation function of neutralizing antibody in solution is compared with that bound to EGFR (SI Figure S2). The free anti-EGFR-Alexa488 in solution could be fitted with one component using the freely diffusing 3D model and the diffusion time was determined as 0.27 ms. On the contrary, an obvious longer diffusion time shift of the autocorre(37) Varghese, L. T.; Sinha, R. K.; Irudayaraj, J. Anal. Chim. Acta 2008, 625, 103–109. (38) Buchanan, F. G.; Wang, D.; Bargiacchi, F.; DuBois, R. N. J. Biol. Chem. 2003, 278, 35451–35457. (39) Ganti, A. K.; Potti, A. Expert Opin. Biol. Ther. 2005, 5, 1165–1174.

Figure 2. (A) From left to right, fluorescence intensity image, FLIM of anti-EGFR-Alexa488 shown in red-green-blue (RGB) mode, FLIM of EGFR-GFP shown in RGB mode, and the merged FLIM images to show the relative localization of EGFR and its antagonist antibody. (B). TCSPC data was fitted by double exponential components with the instrument response function (IRF). (C) A double exponential fit with two fixed lifetimes, 2.10 and 4.12 ns from the lifetime histogram of HEK293 cells was noted.

lation curve of bound antibodies was noted indicating the binding of antibodies with EGFR. The autocorrelation function of bound antibody could be fitted well with eq 3 instead of the single component 3D freely diffusing model. The diffusion times of these two components were 0.23 and 39 ms corresponding to free neutralizing antibody and bound neutralizing antibody, respectively. The relative amount of unbound vs bound ligand molecules is given by the ratio (1 - y)/y, where y was obtained through eq 3. The percentage of bound fraction of anti-EGFR to EGFR could be estimated to be in the range between 45 and 60%. The autocorrelation function of EGFR-GFP when not interacting with its neutralizing antibody as fitted by a two component curve is shown in SI Figure S3B. Two different diffusion times of EGFR-GFP were obtained, τ1 ) 0.21 ± 0.08 ms and τ2 ) 46 ± 12 ms. Past studies using total internal reflection fluorescence (TIRF)40,41 show that EGFP fusion proteins are not distributed homogeneously on cell surfaces. Rather, EGFR fuses to yield preformed dimers before binding to a ligand allowing the possibility of different subpopulations to be present in the plasma membrane. The faster component (smaller/free complex) is indicative of the diffusion of free EGFR while the slower component shows a restricted mobility because of its presence in a complex environment or denote its association with other proteins. An iterative algorithm18 was used to reconstruct the measured TCSPC histogram into three different patterns: EGFR-GFP, Anti-EGFP-Alexa488, and background contributions from dark counts and detector afterpulsing as shown in Figure 3A. Statistical filter functions (Figure 3B) were then numerically calculated from the respective reconstructed fluorescence decay patterns used to separate the autocorrelation function of the neutralizing antibody and EGFR-GFP. The autocorrelation of the neutralizing antibody and EGFR in HEK293 cells and their cross-correlation functions are (40) Teramura, Y.; Ichinose, J.; Takagi, H.; Nishida, K.; Yanagida, T.; Sako, Y. EMBO J. 2006, 25, 4215–4222. (41) Yu, C.; Hale, J.; Ritchie, K.; Prasad, N. K.; Irudayaraj, J. Biochem. Biophys. Res. Commun. 2009, 378, 376–382.

Figure 3. (A) TCSPC histogram from FLCS measurement (Black) decomposed into three patterns: EGFR-GFP (blue), AntiEGFR-Alexa488 (red curve), and background (green). (B) Corresponding anti-EGFR-Alexa488 (red), EGFR-GFP (blue), and background (green) filters calculated from different fluorescence decay patterns to extract the emissions from different species. (C) Crosscorrelation (2), anti-EGFR-Alexa488 (b) autocorrelation, EGFR-GFP (9) autocorrelation from FLCS and the corresponding curve fitted (solid line) are shown. (D) Continuous distribution of the diffusion time of EGFR-GFP (blue curve), neutralzing antibody (red curve), and cross-correlation (black curve) by MEMFCS analysis.

shown in Figure 3C. The cross-correlation function of the binding of neutralizing antibody with EGFR could be fitted well with two components and the complex diffusion time was determined as τc1 ) 0.23 ± 0.10 ms and τc2 ) 35 ± 6 ms. The faster component (τc1) was closer to the free EGFR diffusion time (τ1 ) 0.21 ± 0.08 ms) discussed above. The diffusion time for the slower component of the antibody-EGFR complex (τc2 ) 35 ± 6 ms) was slightly lower than the slower component of EGFR (τ2 ) 46 ± 12 ms) observed earlier. To further assess the auto- and cross-correlation functions generated from FLCS, a multicomponent model using MEMFCS Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6419

was applied for data analysis. From MEMFCS, two peak positions could be observed for both the autocorrelation functions of EGFRGFP and neutralizing antibody and cross-correlation (Figure 3D). Specifically, for the cross-correlation, one peak centered at 0.19 ms depicting the faster diffusing species and the other peak centered at 32 ms depicting the slowly diffusing species constituting the antibody-EGFR complex, which is consistent with our two component fitting results. Collectively, these data suggests that the two pools of antibody-EGFR complexes may exist on the cell membrane with different diffusion behavior. A one to one stoichiometry of anti-EGFR binding to EGFR was assumed in the calculations of the binding kinetics for this reaction. The advantage of cross-correlation is that a direct estimate of the concentration of products and reactants is possible. The dissociation constant of anti-EGFR binding to EGFR could be easily calculated from [Anti-EGFR][EGFR]/[EGFR-AntiEGFR] as 9.2 ± 2.7 nM, similar to the published work on EGFEGFR studies in cells.42 Mapping EGFR Density on Live Cells. To better answer the question of whether the antibody brings about receptor internalization, FLCS was used to quantify receptor density changes on the cell membrane before and after the addition of the neutralizing antibody. Single point FLCS measurements were obtained from the confocal volume with bound antibody-Alexa488 and EGFR-GFP on the cell membrane. The autocorrelation function of EGFR-GFP was separated through FLCS and the total number of EGFR (N) was estimated by fitting the autocorrelation function (eq 2). Assuming the cell membrane was in the half position of the ellipsoidal confocal volume, the radius of the surface area on the cell membrane illuminated by the laser was assumed to be equal to the radial beam size (ω) of the laser. The effective confocal volume was determined as 0.38 fL using Rhodamine 123 dye with a known diffusion coefficient 300 µm2/s. The radius of the laser beam ω was then calculated as 0.21 µm. The circular area on the cell membrane illuminated by the laser beam was estimated as πω2 ) 0.14 µm2. Therefore, the density of receptors expressed could be given by density )

N πω2

Figure 4. Comparison of EGFR density changes before and after the addition of EGFR neutralizing antibody in HEK293 cells.

Figure 5. (A) Cross-sectional fluorescence intensity measurements of EGFR before (Left panel) and after (Right panel) treatment with antagonist antibody. Scale bar )10 um. (B) Temporal analysis of surface EGFR expression using untreated cells normalized as 100% expression level. Cells treated with antagonist antibody were monitored at 5, 30, 60, 120 min. Error bars, ( SD (n ) 20).

The above results demonstrate that the neutralizing antibody could induce receptor internalization. Our findings are in agreement with a previous study which used Cetiximab to target EGFR and have been found to internalize through an uncertain route.43 The induction of EGFR internalization by the neutralizing antibody may bring about endocytosis and degradation of active EGFRs responsible for the attenuation of growth-promoting signals.44

The EGFR density was evaluated using above equation in 30 different HEK293 cells. Assuming a homogeneous distribution, the EGFR expressed in HEK293 cells was determined as 540 ± 64 receptors/ µm2. Upon addition of the neutralizing antibody, the EGFR density showed a significant decrease and reached an estimated value of 38 ± 7 receptors/µm2 (p < 0.05, Figure 4). We further employed fluorescence intensity cross-sectional analysis to demonstrate that EGFR redistributed from cell membrane (Figure 5A, left) to the cytoplasmn (Figure 5A, right) after binding with the antagonist antibody due to internalization. A temporal fluorescence fluctuation analysis to determine the level of EGFR at the cell surface revealed that cells treated with the antagonist antibody for 2 h showed a 52% reduction compared to untreated cells (Figure 5B.).

CONCLUSION By employing FLCS, for the first time we quantitatively evaluate IgG and anti-IgG interaction in solution and EGFR binding with its antagonist antibody in live cells. We demonstrate that FLCS could be used to distinguish species with different lifetimes from their autocorrelation functions when present in solution as well as on cell membranes. The cross-correlation functions could be readily calculated and the diffusion coefficient and dissociation constant could subsequently be computed. Experiments indicate that EGFR density significantly decreased upon addition of the neutralizing antibody and the antagonist antibody could bring about receptor internalization, inhibiting EGFR activation. Compared to dual-color FCCS, FLCS can be technically much simpler with significant potential to study up to three protein interactions using a single laser source in vitro and in live cells.

(42) Pramanik, A.; Rigler, R. Biol. Chem. 2001, 382, 371–378. (43) Liao, H. J.; Carpenter, G. Cancer Res. 2009, 69, 6179–6183. (44) Prewett, M.; Rockwell, P.; Rockwell, R. F.; Giorgio, N. A.; Mendelsohn, J.; Scher, H. I.; Goldstein, N. I. J. Immunother. Emphasis Tumor Immunol. 1996, 19, 419–427.

ACKNOWLEDGMENT Funding from Purdue Center for Cancer Research and Purdue Research Foundation scholarship to J.C. is acknowledged. pEGFR-

6420

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

GFP plasmid was obtained from Dr. John Koland, from the University of Iowa, College of Medicine. Professors Maiti and Periaswamy (TIFR, Mumbai) are acknowledged for providing the MEMFCS software. Finally, the authors thank Professor Nagendra K. Prasad for the helpful insights.

izing antibody in solution and those bound to EGFR, the autocorrelation function of EGFR-GFP when not interacting with antibody and location of FLCS measurement. This material is available free of charge via the Internet at http://pubs.acs.org.

SUPPORTING INFORMATION AVAILABLE Supporting figures of autocorrelation function of mixture IgG and anti-IgG solution, comparison of autocorrelation functions of neutral-

Received for review February 24, 2010. Accepted June 15, 2010. AC101236T

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6421