SINGLE MOLECULAR AND ATOMIC MANIPULATION OF CONFORMATION AND DYNAMICS

Jin Cao

A Dissertation

Submitted to the Graduate College of Bowling Green State University in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

December 2014

Committee:

H.Peter, Lu, Advisor

Massino Olivucci

John R. Cable Graduate Faculty Representative

Ray A. Larson ii ABSTRACT

H.Peter, Lu, Advisor

The present thesis describes the study of protein conformation dynamics probed by single techniques. Protein structural dynamic have been studied for many years. Interesting topics such as protein folding/unfolding, protein denature, and protein-protein interactions had been widely studied. Scientists are still interested in questions like: what exact contribute to the stability of protein secondary and tertiary structure; why different protein has different function based on their structure. Single molecular spectroscopy has the advantage to observe inhomogeneous behavior of protein molecule compared to ensemble measurement. The typical approach of single molecular spectroscopy is to record one single molecule at a time and repeat this process at one or multiple that are chemically identical. Then use statistics way to analyze these data.

The binding dynamics of CaM-C28W complex was studied using single molecule FRET experiment. The statistics results indicate the inhomogeneous nature of the process and the evidence of a slower process gauging the binding dynamics has been claimed.

Furthermore, the responding mechanical dynamics of CaM molecule under AFM tip manipulation has been investigated by AFM-FRET correlated measurement. The repeated

“collapsing” pattern in the force curve leads us to deeper study on the process. The nature of the process and the dynamics of the underlying conformation change have also been detailed described.

Lastly, a similar AFM manipulation experiment has been performed on EGFR dimer complex that has more biological potential in terms of dimer formation and aggregation dynamics. In order to fit the nature of the molecule, we used two identical dyes to label either monomer.

Instead of single molecule FRET, photo-stamping measurement has been used to build an AFM- iii optical correlated measurement. The statistics results shows the presents of multiple intermediate states. With the help of known crystal structure and the knowledge of EGFR dimer complex, we managed to find clue from our correlated results and assign possible conformation of each intermediate states. A possible conjugated state upon collapsing has been claimed.

. iv ACKNOWLEDGMENTS

There have been many brilliant people involved in the research who helped me to conduct the projects during my Ph. D adventure, and to whom I own my gratitude.

I would like to express my deep appreciation and gratitude to H. Peter Lu, advisor. for all the support, patient guidance, encouragement, and understanding he provided to me, a young

Chinese student living abroad, throughout all these years. I would like to thank to my committee members: Dr. Massimo Olivucci, Dr. John R. Cable, and Dr. Ray Larson for their valuable time.

Special thanks to Dr. Yufan He for his effort to help me improve my experiment skill during day-to-day work throughout the year. Special thanks to Dr. Desheng Zheng for being such a intelligent person to discuss experiment design and detail with me. I want to acknowledge other past and present group members for their help and friendly working environment: especially

Qing Guo, Zijian Wang, Xiaonan Han and Dr. Yuanmin Wang.

I am also thankful to the faculty and staff of the Center for Photochemical Sciences and the

Department of Chemistry, particularly Alita Frater, Mary Toth, Charles Codding, and Doug

Martin.

I would like to thank to my family for their love and support. I am thankful to my mother, who traveled across the earth and took care of me for one month during the busiest days. I am thankful to my girlfriend Lingxiao Ge for being such a wonderful person and supporting me all these years with love and understanding. v

TABLE OF CONTENTS

Page

CHAPTER I. INTRODUCTION...... 1

1.1 Brief History of Single Molecule Technique...... 1

1.2 Ensemble Measurement versus Single Molecule Measurement...... 3

1.3 Manipulate Single Molecule...... 5

1.4 Classic Biochemistry versus Real Biological Environment ...... 6

1.5 References ...... 9

CHAPTER II. EXPERIMENTAL SECTION ...... 13

2.1 Theoretical Background...... 13

2.1.1 Con-focal spectroscopy...... 13

2.1.2 Single-molecular FRET spectroscopy ...... 15

2.1.3 AFM basic: imaging and force spectroscopy...... 25

2.2 Experiment Setups ...... 32

2.2.1 Single-molecular FRET measurement...... 32

2.2.3 AFM setup and correlated measurement ...... 33

2.2.3 GROMACS based MD simulation ...... 39

2.3 Materials and sample preparation ...... 41

2.4 References ...... 44

CHAPTER III. SINGLE MOLECULE FRET STUDY OF CAM-PEPTIDE BINDING

DYNAMICS ...... 50 vi

3.1 Introduction...... 50

3.2 Experiment Section...... 51

3.2.1 Materials and sample preparation ...... 51

3.2.2 Single-molecular FRET spectroscopy ...... 52

3.3 Results and Discussion ...... 53

3.4 Conclusion ...... 64

3.5 References ...... 66

CHAPTER IV. PROTEIN CONFORMATION SUDDEN COLLAPSES UNDER STRESS

FORCE MANIPULATION BY SINGLE-MOLECULE AFM-FRET NANOSCOPY 68

4.1 Introduction...... 68

4.2 Experiment Section...... 69

4.2.1 Materials and sample preparation ...... 69

4.2.2 AFM-FRET correlated measurement ...... 70

4.2.3 MD simulation ...... 71

4.3 Results and Discussion ...... 73

4.4 Conclusion ...... 87

4.5 References...... 89

CHAPTER V. PROBING THE RESPONSE OF EGFR EXTRACELLULAR DOMAIN

DIMERS TO EXTERNAL v

STRESS FORCE USING SINGLE-MOLECULE CORRELATED MEASUREMENT

AND EVALUATING THE POSSIBLE REARRANGEMENT OF A COLLAPSED

STRUCTURE ...... 92

5.1 Introduction...... 92

5.2 Experiment Section...... 94

5.2.1 Materials and sample preparation ...... 94

5.2.2 AFM-Photo-stamping correlated measurement...... 96

5.3 Results and Discussion ...... 98

5.4 Conclusion ...... 113

5.5. References ...... 114 vi

LIST OF FIGURES

Figure Page

2.1 Basic principal of a confocal microscopy...... 15

2.2 The basic mechanism of fluorescence resonance transfer ...... 19

2.3 The relation between FRET efficiency and donor-acceptor distance...... 22

2.4 Spectrum of Cy3 and Cy5...... 25

2.5 The schematic representation of the principal of AFM microscope...... 28

2.6 The schematic representation of the process of the force curve measurement in force

spectroscopy experiment...... 31

2.7 The schematic representation of the experiment set-up used in chapter iii, and the optical

element of the set-up used in chapter iv...... 33

2.8 The schematic representation of the experiment set-up used in chapter iv ...... 35

2.9 The procedures of AFM-focus-point alignment in the AFM-FRET correlated

measurement...... 38

2.10 The immobilization method used in chapter iii, and iv.: ...... 43

2.11 The relation between FRET efficiency and donor-acceptor distance...... 43

3.1 The conformation of calmodulin molecule...... 51

3.2 Experimental setup and fret trajectory...... 53

3.3 Two different statistical analysis of the same FRET trajectory...... 55

3.4 Representation of how the trajectory is split into two states by judging the jumping of

each data point...... 59

3.5 The relation between adjacent close state duration and fret efficiency...... 59 vii

3.6 The schematic representation of the experiment set-up used in chapter iii, and the optical

element of the set-up used in chapter iv...... 62

3.7 The APNS electrostatic distribution of different forms of cam C or N domain...... 63

4.1 Structure of calmodulin and the structure of simulation box...... 73

4.2 Correlated measurement setup and results...... 76

4.3 Correlated real time trajectories of the AFM-FRET correlated measurement...... 77

4.4 The relation between approaching speed and collapse induced force...... 79

4.5 Potential energy changes during the squeezing perform by MD simulation...... 83

4.6 Evaluate the recovery of collapsed Cam by RMSD...... 86

4.7 A typical feature segment of force curve contains a collapse event...... 87

5.1 Schematic representation of the structure of EGFR ...... 94

5.2 Schematic representation of the AFM-photo-stamping correlated measurement setup and

photo-stamping result...... 98

5.3 Synchronized experiment result...... 101

5.4 Fitted life time mean and width of lifetime distribution histogram...... 102

5.5 Statistical facts of the magnitude and duration of the lifetime shift...... 104

5.6 Fitted contour length distribution of different scenarios...... 108

5.7 Hypothesis of possible outcomes of rearrangement of a collapsed EGFR dimer complex

on surface ...... 112 1

CHAPTER I. INTRODUCTION

This chapter is dedicated to the introduction of the basic of single molecule experiment and protein conformation dynamics.

1.1 Brief history of Single Molecule Technique

Scientists have been dreaming to observe molecular level behavior in chemical reaction for years. In recent decades, researchers successfully perform optical spectroscopy and microscopy of single molecules in condensed phase. Later on, the challenge of utilizing this technique in aqua condition was surpassed by the use of total internal reflection fluorescence microcopy. These efforts exploded the entire field of single molecular microscopy. Single molecular microscopy is essentially a cross field technique and concept. It draws the attention from varieties of specific field such as , cellular biology, and material and science.

The first indirect measurement of enzymatic activity at the single-molecule level was done by Rotman et al. in 19611. The experiment was about fluorescent reaction products generated by a single -galactosidase enzyme acting on a substrate analog. Later on in 1976,

Hirschfeld et al2. managed to detect a single by labeling 80 fluorophores. And in the 1980s Moerner and Kador(1989)3 and Orrit and Bernard(1990)4 successfully detected the absorbance and fluorescence, respectively, of single pentacene molecules doped into crystal of p-terphenyl at ultralow temperatures. Detecting fluorescence emission has become the primary choice when observing single molecule as the great sensitivity it provides5-9. In the

1980s, Keller et al. (1987)10 and Stryer et al. (1989)11 detected single molecules of the fluorescent protein phycoerythrin at room temperature in aqueous solutions. The most 2

importantly, thanks to the development of near-field scanning optical microscopy (Betzig and

Chichester 1993)12 and its implementation within a simplified confocal optical microscope geometry (Rigler and Mets 199213; Macklin et al. 199614) that made the application of single-molecule measurement more accessible.

In order to detect a light-emitting single molecule surrounded in solution by huge numbers of other molecules, three main problems have to be solved: 1) lower the observed volume, 2) confined the molecule in such a volume for a reasonable amount of time, depends on the process desired to observed, 3) maximize the luminous efficacy of each molecule. Since the number of photons emitted by a single fluorophore is limited, high-efficiency and low-background photon detection techniques have to be implemented. The efficiency can be maximized by increasing the photon collection efficiency. High efficiency optics such as high NA (numerical aperture) objectives, high efficiency filters and lens are often used.

Depending on the goal of study, researchers choose single photon detectors such as avalanche photodiodes to achieve high time and spatial resolution, or CCD cameras to achieve high frame rate. Dealing with background photons is also crucial for single molecular techniques.

Background photons come from sources such as scattering and fluorescence from buffers and impurities. Straightforwardly, minimizing the detection volume is the most efficient approach. Two most popular geometries for single molecule fluorescence detections are confocal spectroscope and total internal reflection (TIR) spectroscope. Confocal spectroscope utilized light to achieve a very small extinction volume, however it can only light up and detect one point at a time. In TIR spectroscope, a thin layer above glass surface is emitted by evanescent wave and fluorescence is detected by CCD. It allows 3

simultaneous excitation of a large surface while maintaining low background, which is ideal for study on biological samples.15-19

1.2 Ensemble Measurement versus Single Molecule Measurement

Traditional chemistry and biochemistry experiments in solution phrases study many molecules; even in 1L of water solution of 1M concentration, there are 1012 solute molecules and 1013 water molecules. These solute molecules are considered identical from a classic point of view. However, these solute molecules are dynamically interacting with each other and water molecules. During any given short period of time, each one of them is unique. When it comes down to particular measurement, using fluorescence intensity as an example, the ensemble level measurement will get a constant value if no reaction occurs.

Alternatively, the rate of fluorescent product formation in a chemical reaction can be calculated using the changing fluorescence intensity. Clearly, we learned a great deal about the system we study from these measurements, but there is much more we are curious about and essential to better understanding of the system.

Ensemble averages measurement signal is made of the unsynchronized average of the contribution of every molecule in the sample. This tends to smoothen the processes like and translation and make these processes appear to be continuously varying events. This simplified picture is misleading as molecular level dynamics could be masked.

At the single molecule level, stochastic dynamics intrinsically exist, because any chemical reaction involves the thermally induced random crossing of a free-energy barrier. This is even more important for biological system. Within a , which is the basic unit of most living system, transcription and translation are executed by only a few thousands of 4

molecules or complexes of molecules. The intrinsic stochastic fluctuation plays an even bigger role. We want to have a better understand of how an overall directional process of a given biological process could arise from naturally random events, and to what degree these random events determine the phenotypic fate of a cell. We are also interested in investigating the trajectories of these processes and identifying singular events, intermediates state, speed limiting step, and other heterogeneous factors of the system along the reaction pathways that may control the reaction’s outcome.

Here is some example to compare single molecule experiment and ensemble experiment:

First, we compare the case in a time-independent system, or a system that reaches equilibrium. College chemistry teaches us that chemical equilibrium does not mean that all reaction have stopped, only that the forward and backward reaction rates are equal, or the reactants and products are formed and consumed at a equal rate. Thus, the concentration of each component in the system remains constant. As a result, the measurable properties of the system could be constant. To think this backward, if a system under investigation turns out to have a series of constant parameters (whatever those might be), the system itself could contain only one type of molecule, or it could be a mixture of non-reacting molecules, or it could be a mixture of reacting molecules at equilibrium. The ensemble measurement cannot distinguish among these three possibilities. In contrast, single-molecule measurements can quickly distinguish among them.

Secondly, we also compare the case in a time-dependent system, where the value of single-molecule measurement becomes more outstanding. The ensemble way to understand the kinetic of a chemical reaction is describing varieties of kinetic constants. If only one 5

reactant is being observed, it is simply impossible to know when the reaction will occur.

However, the concentration change of the reactant and product in the system can be plotted using rate constants and initial concentrations of reactant and product. From the view of each individual reactant, the reaction could only occur until it accumulates enough energy to overcome the activation barrier. The molecules do not communicate with each other, but their behavior as a group follows certain rules that are often gauged by the environment and the nature of the reaction itself. The stochastic nature of chemical reactions determines that reactants, products and intermediates will all be present throughout the reaction. It will be difficult to identify and characterize all of the species in this complex mixture. However, by recording the trajectory of a single molecule, each on-pathway intermediate on the way from reactant to product become visible. Also off-pathway mechanisms can be distinguished, as long as the time resolution of the measurement technique is good enough to observe the short-lived species. Nevertheless, it is much easy to identify a single species in a trajectory than to try to resolve it from a mixture of many species in the case of ensemble averaging.

It is also easier to understand the relationship between each species from single-molecule time resolved data.

1.3 Manipulate Single Molecule

Methods to manipulate single molecules in solution, to apply force to them, and to watch them change shape; have been available since the 1990s. Arthur Ashkin invented in 1970s at IBM20; he found that a tightly focused laser beam could be used to trap and move micron-sized particles. The application of it on biological problems starts a few years afterward, and the full potential of this technique has been exploded in the last decade 6

after the booming of single-molecule technique took off. Steven Chu won himself the

Nobel Prize in 1997 for trapping individual gas-phase molecules with the tweezers. The scanning tunneling microscope (STM) was developed by Gerd Binning and Heinrich Rohrer in the early 1980s at IBM research, Zurich21. They also won the Nobel Prize for Physics in

1986. Based on STM, as an attempt to expand the technique to non-conducting material, the idea of AFM was described by Binning et al. in 198621. Later on, this idea was improved by using light-lever technique and STM is no longer required. This improvement allowed the application of AFM to be much widened and the methods had been used to solve problems in biochemistry and molecular biology in the new century22-26.

Interestingly enough, the techniques that are used to manipulate single molecule are also detection methods in single molecule experiment27-34. For example, when a reacting enzyme is specifically linked to beads trapped inside an optical tweezers, any subset in the reaction that causes the beads to move will be affected by force; the force can aid the motion or counteract the motion34-36. Plotting the force curve in real time can spot sub-steps of a reaction and point out possible pathway and intermediate states.

1.4 Classic Biochemistry Versus Real Biological Environment

In order to fully understand protein dynamics or the dynamics of any bio-molecules in general, we have to have a better understanding about the real cellular environment. There is a big gap between the simple solution environments defined by classic biochemistry and the real living environment. Whenever scientists want to study the property of a particular protein or a particular biochemistry reaction outside of living system, they often prepare the sample solution to duplicate the properties of living system, and they focus on the following 7

factors: concentration of the core species, temperature, pH, ionic strength, and maybe implementing some kind of biological segment such as membrane. These attempts are helpful but far from the real living environment. To take protein as an example, protein perform their functions in cells environment where macromolecular solutes has concentration greater than 300g/L and occupy greater than 30% of the volume. These macromolecules could participate in the process in study or they could just act as a bystander. The existence of these excess amounts of macromolecules generates an overall crowded environment and this is where the most biochemistry reaction took place. Imagine a person in a subway car.

If it is the night time and there is only a few other passengers around, one can do whatever you want to put his feet or raise his arm to grab the overhead strap. However, if it is the busiest hours of a day like 8am or 5pm in New York City, such simple motions aren’t so easy anymore in a packed and crowed car. This is just like compare the behavior of a protein molecule in dilute solution and real crowded living environment, as almost any function of protein involves complex motions of the protein conformation that could affected by other surrounding species. I’d like to use a quote from one of the article written by Professor

Pielas from UNC, “If you think about it, studying macromolecular crowding is anti-biochemistry.”37-40

The crowded intracellular environment affects the almost all the basic phenomena of protein and how protein behave. First, it has been proved that generally protein severely reduce the rotational diffusion of other . This is done through the presence of weak nonspecific, non-covalent chemical interaction between proteins. Secondly, macromolecular crowding affects the stability of protein. The stability of protein can be affected through 8

different ways. The fact that the native state occupies less space than the denatured state makes most people believe that crowding environment stabilizes protein. This is so called excluded-volume effects. However, experiment result showed the opposite. Protein crowders can be mildly destabilizing due to nonspecific interactions, including electrostatic interactions. The completing mechanism between these two factors ends up with a tunable stability. Last, the protein dynamics can be affected by macromolecular crowding. Li et al.41 reports that the ordered protein is more sensitive to the crowded environment than is the disordered protein. The inherent properties of disordered proteins are often linked to some neurodegenerative disease.

Other than describing the effect of macromolecular crowding by volume occupancy factor or the size of crowders, Ping et al. link the effects of macromolecular crowding to depletion force as mechanical force is the most fundamental way these macromolecules actually interact with each others. From their conclusion, the magnitude of depletion force fall into the range of AFM force manipulation. This opens up the possibility to simulate this effect to a certain degree using AFM manipulation. The advantage of this idea is that both conformational response and mechanical response can be recorded in real time. The concept of macromolecular crowding highly correlated with the experiments in Chapter IV and V. 9

1.5 References

1. Rotman, B. MEASUREMENT OF ACTIVITY OF SINGLE MOLECULES OF ß-D-GALACTOSIDASE. Proceedings of the National Academy of Sciences 1961, 47, 1981-1991.

2. Hirschfeld, T. Optical microscopic observation of single small molecules. Appl. Opt. 1976, 15, 2965-2966.

3. Moerner, W.; Kador, L. Optical detection and spectroscopy of single molecules in a solid. Phys. Rev. Lett. 1989, 62, 2535.

4. Orrit, M.; Bernard, J. Single pentacene molecules detected by fluorescence excitation in a p-terphenyl crystal. Phys. Rev. Lett. 1990, 65, 2716.

5. Minsky, M. Memoir on inventing the confocal scanning microscope. Scanning 1988, 10, 128-138.

6. Minsky, M. Microscopy Apparatus: US 3,013,467 .

7. Shotton, D. M. Confocal scanning optical microscopy and its applications for biological specimens. J. Cell. Sci. 1989, 94, 175-206.

8. Denk, W.; Strickler, J. H.; Webb, W. W. Two-photon laser scanning fluorescence microscopy. Science 1990, 248, 73-76.

9. Paddock, S. W. Further developments of the laser scanning confocal microscope in biomedical research. Proc. Soc. Exp. Biol. Med. 1996, 213, 24-31.

10. Benítez, J. J.; Keller, A. M.; Ochieng, P.; Yatsunyk, L. A.; Huffman, D. L.; Rosenzweig, A. C.; Chen, P. Probing transient copper chaperone-Wilson disease protein interactions at the single-molecule level with nanovesicle trapping. J. Am. Chem. Soc. 2008, 130, 2446-2447.

11. Peck, K.; Stryer, L.; Glazer, A. N.; Mathies, R. A. Single-molecule fluorescence detection: autocorrelation criterion and experimental realization with phycoerythrin. Proc. Natl. Acad. Sci. U. S. A. 1989, 86, 4087-4091.

12. Betzig, E.; Chichester, R. J. Single molecules observed by near-field scanning optical microscopy. Science 1993, 262, 1422-1425.

13. Rigler, R. Fluorescence correlations, single molecule detection and large number screening applications in biotechnology. J. Biotechnol. 1995, 41, 177-186.

14. Macklin, J.; Trautman, J.; Harris, T.; Brus, L. Imaging and time-resolved spectroscopy of single molecules at an interface. Science 1996, 272, 255-258. 10

15. Poole, C. A.; Ayad, S.; Gilbert, R. T. Chondrons from articular cartilage. V. Immunohistochemical evaluation of type VI collagen organisation in isolated chondrons by light, confocal and electron microscopy. J. Cell. Sci. 1992, 103 ( Pt 4), 1101-1110.

16. Paddock, S. W. Tandem scanning reflected-light microscopy of cell-substratum adhesions and stress fibres in Swiss 3T3 cells. J. Cell. Sci. 1989, 93 ( Pt 1), 143-146.

17. Bustamante, C.; Bryant, Z.; Smith, S. B. Ten years of tension: single-molecule DNA mechanics. Nature 2003, 421, 423-427.

18. Moerner, W.; Fromm, D. P. Methods of single-molecule fluorescence spectroscopy and microscopy. Rev. Sci. Instrum. 2003, 74, 3597-3619.

19. Förster, T. Zwischenmolekulare Energiewanderung und Fluoreszenz. Annalen der Physik 1948, 437, 55-75.

20. Ashkin, A. Acceleration and trapping of particles by radiation pressure. Phys. Rev. Lett. 1970, 24, 156.

21. Binnig, G.; Quate, C. F.; Gerber, C. Atomic force microscope. Phys. Rev. Lett. 1986, 56, 930.

22. Lee, N. K.; Kapanidis, A. N.; Koh, H. R.; Korlann, Y.; Ho, S. O.; Kim, Y.; Gassman, N.; Kim, S. K.; Weiss, S. Three-color alternating-laser excitation of single molecules: monitoring multiple interactions and distances. Biophys. J. 2007, 92, 303-312.

23. Weiss, S. Fluorescence spectroscopy of single biomolecules. Science 1999, 283, 1676-1683.

24. Ha, T.; Enderle, T.; Ogletree, D. F.; Chemla, D. S.; Selvin, P. R.; Weiss, S. Probing the interaction between two single molecules: fluorescence resonance energy transfer between a single donor and a single acceptor. Proc. Natl. Acad. Sci. U. S. A. 1996, 93, 6264-6268.

25. Kapanidis, A. N.; Laurence, T. A.; Lee, N. K.; Margeat, E.; Kong, X.; Weiss, S. Alternating-laser excitation of single molecules. Acc. Chem. Res. 2005, 38, 523-533.

26. Ha, T.; Ting, A. Y.; Liang, J.; Caldwell, W. B.; Deniz, A. A.; Chemla, D. S.; Schultz, P. G.; Weiss, S. Single-molecule fluorescence spectroscopy of enzyme conformational dynamics and cleavage mechanism. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 893-898.

27. Michalet, X.; Weiss, S.; Jäger, M. Single-molecule fluorescence studies of protein folding and conformational dynamics. Chem. Rev. 2006, 106, 1785-1813. 11

28. Seidel, R.; Dekker, C. Single-molecule studies of nucleic acid motors. Curr. Opin. Struct. Biol. 2007, 17, 80-86.

29. Smiley, R. D.; Hammes, G. G. Single molecule studies of enzyme mechanisms. Chem. Rev. 2006, 106, 3080-3094.

30. Zhuang, X. Single-molecule RNA science. Annu. Rev. Biophys. Biomol. Struct. 2005, 34, 399-414.

31. Best, R. B.; Merchant, K. A.; Gopich, I. V.; Schuler, B.; Bax, A.; Eaton, W. A. Effect of flexibility and cis residues in single-molecule FRET studies of polyproline. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 18964-18969.

32. Liu, R.; Hu, D.; Tan, X.; Lu, H. P. Revealing two-state protein-protein interactions of calmodulin by single-molecule spectroscopy. J. Am. Chem. Soc. 2006, 128, 10034-10042.

33. He, Y.; Li, Y.; Mukherjee, S.; Wu, Y.; Yan, H.; Lu, H. P. Probing single-molecule enzyme active-site conformational state intermittent coherence. J. Am. Chem. Soc. 2011, 133, 14389-14395.

34. Lu, H. P. Single-molecule photon stamping FRET spectroscopy study of enzymatic conformational dynamics. Physical Chemistry Chemical Physics 2013, 15, 770-775.

35. He, Y.; Lu, M.; Cao, J.; Lu, H. P. Manipulating protein conformations by single-molecule AFM-FRET nanoscopy. ACS nano 2012, 6, 1221-1229.

36. Wang, J.; Oliveira, R. J.; Chu, X.; Whitford, P. C.; Chahine, J.; Han, W.; Wang, E.; Onuchic, J. N.; Leite, V. B. Topography of funneled landscapes determines the thermodynamics and kinetics of protein folding. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 15763-15768.

37. Miklos, A. C.; Sarkar, M.; Wang, Y.; Pielak, G. J. Protein crowding tunes protein stability. J. Am. Chem. Soc. 2011, 133, 7116-7120.

38. Wang, Y.; Li, C.; Pielak, G. J. Effects of proteins on protein diffusion. J. Am. Chem. Soc. 2010, 132, 9392-9397.

39. Pincus, D. L.; Thirumalai, D. Crowding effects on the mechanical stability and unfolding pathways of ubiquitin. J Phys Chem B 2009, 113, 359-368.

40. Benton, L. A.; Smith, A. E.; Young, G. B.; Pielak, G. J. Unexpected effects of macromolecular crowding on protein stability. Biochemistry 2012, 51, 9773-9775.

41. Li, C.; Charlton, L. M.; Lakkavaram, A.; Seagle, C.; Wang, G.; Young, G. B.; Macdonald, J. M.; Pielak, G. J. Differential dynamical effects of macromolecular crowding on an 12 intrinsically disordered protein and a globular protein: implications for in-cell NMR spectroscopy. J. Am. Chem. Soc. 2008, 130, 6310-6311. 13

CHAPTER II. EXPERIMENTAL SECTION

This chapter is dedicated to the brief explanation of the experimental techniques and the general procedures used for the sample preparation during the experiments.

2.1 Theoretical Background

2.1.1 Confocal spectroscopy

Confocal microscopy, first patented in 1957 by Marvin Minsky1,2, is an optical imaging technique designed to overcome some limitation of traditional wide-field fluorescence microscopes. Conventional wide-field fluorescence microscopes uses light source to illuminate the entire specimen. The light source will excite all parts of the specimen in an optical path at the same time. Therefore, the photodetector or camera of the microscopes’ system will record fluorescence not only from the parts in interest but also a large unfocused background part, causing much lower resolution3-5 . This limitation is critical for biological sample as they always have complex 3-D structure and varieties of fluorophores in a relative small space. Some of these limitations could be partially overcome by video image processing3,5 and deconvolution. In contrast, point illumination is used in a confocal microscope by placing a pinhole in front of the light source. And out-of-focus signal is blocked by placing a pinhole in an optically conjugate plane in front of the detector. The name “confocal” stems from this configuration. Because only one “point” (small region) of the specimen is excited and only light from the area very close to the focal plane can be detected, the image’s optical resolution is much better than that of wide-field microscopes, particularly in the z-direction (the direction that is vertical to the focal plane) 3,4,6,7 In practice, laser source is used to avoid the use of a pinhole in illumination. Fig. 2.1 is a demonstration

14

of the basic principal of a laser based confocal microscopy1,2. There is only one objective lens in the system, because illumination and detection share the same objective. A long pass dichromatic mirror is used to prevent the laser source from going into the detector, in order to improve the signal-to-noise ratio. The idea of blocking the out-of-focus signal also comes with the cost of blocking the light from the sample in interest, which leads to decreased signal intensity. So a photomultiplier tube or avalanche photodiode will be needed as a detector.

In our experiment, we mainly use APD (Avalanche Photodiode Detector). APD is a highly sensitive electronic device that exploits the photoelectric effect and avalanche breakdown effect of semiconductor. It has higher quantum efficiency than photomultiplier tubes in the visible range.

The achievable thickness of the focal plane is defined mostly by the wavelength of the used light source and numerical aperture of the objective lens1,3.6This configuration is described as diffraction limit in Equation 2.1.  d  2nsin (2.1)

15

Fig. 2.1 Basic principal of a confocal microscopy. The green lines are laser source light. The red lines are fluorescence from the specimens after excitation.

16

The numerator is the wavelength of the light source. The denominator nsin is called the numerical aperture (NA). Term “n” is the refractive index of the medium that the light travels through. In practical confocal microscopy that targeting the molecules on top of cover glass, this “n” is the refractive index of cover glass holding the sample and the immersion oils used in between the cover glass and objective lens. The highest theoretical numerical aperture obtainable with immersion oil is 1.51 (when sin  1 ). The most common numerical apertures range from 1.0 to 1.35. 8-10 Considering green laser light with

532nm wavelength, which is the light source we used in our experiment, the diffraction limit is roughly 250nm.

As only one point in the specimen is illuminated at a time, scanning over a regular raster is required to obtain 2D or 3D imaging. There are three types of confocal microscopes: confocal laser scanning microscopes, spinning-disk confocal microscopes, and programmable array microscopes (PAM). The differences among them are the ways to achieve scanning.

As a result, each of these classes is particularly optimized for recoding speed or high spatial resolution. To obtain the highest optical resolution, confocal laser scanning microscopes will be used in order to have a programmable sampling density9-11. In contrast, spinning disk and PAM use a fixed sampling density determined by the camera’s resolution, but has much higher imaging frame rates because multiple spots are being illuminated at a time. In our experiment, we choose confocal laser scanning microscopy that utilized a 2-D scanning stage to achieve sample scanning instead of laser scanning. Our goal is to obtain single molecule fluorescence trajectory from a diluted sample instead of speed imaging of a complex sample. Highest spatial resolution is required to maximize the signal-to-noise ratio.

17

To be specific, we collect single molecular FRET trajectory of a single dye-labeled protein molecule to study the protein conformation dynamics in the processes of protein-ligand binding, protein-protein interaction, or enzymatic reaction.

2.1.2 Single-molecular FRET spectroscopy

Single-molecular FRET spectroscopy (smFRET) is a powerful tool for measuring distances between two chromophores and recording dynamics at the molecular level. Researchers always label two or more specific sites on a non-fluorescent molecule with chosen fluorescent dye, and then record the distance change between these individual fluorescent labels by measuring how efficiently electronic energy is transferred between them. The efficiency, which is a number between 0 to 1, will then be transformed into the representation of distance as precise as ~0.3nm. This technique has been applied in many different fields in physics, chemistry, and biology to address varieties of questions and topics.

The most attractive advantage of this technique is its unique ability to reveal the detailed kinetics of complex biochemical systems. In these systems, conformational fluctuations, folding pathways, macromolecular interactions and catalytic kinetics can be detailed at the single-molecule level reliably and new information could be discovered instructively. The biological systems that have been studied by smFRET technique include but not limited to

DNA rulers, staphylococcal nuclease, S15 binding to RNA junction, calmodulin, T4 lysozyme, and Rep helicase. This list is still expanding rapidly, thanks to the easy to use modern commercial high resolution microscopes.

The theoretical foundations of smFRET were laid in the 1940s by German scientist

Theodor11, whose name is used in the term of “Förster resonance energy transfer”. When

18

both chromophores are fluorescent, the term “fluorescence resonance energy transfer” is often used instead. The name “fluorescence resonance energy transfer” is used more commonly in scientific literature, although its meaning carries an erroneous interpretation.

The energy is not actually transferred by fluorescence but non-radiactive dipole-dipole coupling. The mechanism of this phenomenon is described in Fig. 2.2.

A donor chromophore, after being excited into its electronic excited state (S1), may transfer energy to an acceptor chromphore through non-radiactive dipole-dipole coupling.

FRET is similar to near-field communication. The radius of interaction is much smaller than the wavelength of the light emitted. The excited donor chromophore emits a virtual photon that is instantly absorbed by the acceptor chromophore. Because this virtual photon is not detectable, the quantum yield of the energy transfer transition, more often being referred as FRET efficiency (E), is calculated from recording the photons coming from both the donor and the acceptor chromophores.

Equation 2.2 is the expression of FRET efficiency, where kET is the rate of energy transfer, kf the radiactive decay rate, and the ki’s are the rate constant of any other de-excitation pathways. k E  ET   (2.2) k f kET B ki

19

Fig. 2.2 The basic mechanism of fluorescence resonance energy transfer: The donor chromophore is excited by incoming excitation light in blue. The FRET transition is in purple. Both donor and acceptor chromophores relax back to S0 through non-radiactive pathways in brown, and fluorescence in green and red respectively.

20

The FRET efficiency can also be expressed in terms of the lifetime of donor molecule as

 '  ' Equation 2.3, where D and D are the donor fluorescence lifetime in the presence and absence of an acceptor.

   E 1 D '/ D (2.3)

The most common expressions are Equation 2.4 and 2.5, where F’D and FD are the donor fluorescence intensities with and without an acceptor, respectively. FA is the acceptor fluorescence intensity.   E 1 F'D / FD (2.4) F E  A  (2.5) FD FA

The FRET efficiency depends on mainly the distance between two chromophores, the spectral overlap of between the donor and the acceptor absorption spectrum, the relative orientation of the donor emission dipole moment and the acceptor absorption dipole moment, and local environment. When the pair of donor and acceptor is chosen and the local environment is determined, the FRET efficiency will then be expressed as a function of the distance between the donor and the acceptor. This is described in

Equation 2.6. 1 E   6 (2.6) 1 (r / R0 )

Where R0 is the Förster distance of this pair, in other words the distance at which the energy transfer efficiency is 50%. The FRET efficiency is inversely proportional to the sixth power of the distance between donor and acceptor. The distance is normally in the

21

range of 1-10nm.

The other three factors are described in R0 in Equation 2.6.

9Q (ln10) 2 J R 6  0 0 5 4 (2.7) 128 n N A

J is the spectral overlap integral, n is the refractive index of medium.Q0 is the fluorescence quantum yield of the donor in the absence of the acceptor. NA is Avogadro’s number. 2 is the dipole orientation factor. When both chromophores are freely rotating, a number of 2/3 is often assumed. In most cases, there will not be a large error to use this number. This is because that even modest reorientation of the dyes results in enough orientation averaging.

This is mostly true in practical smFRET, when researchers label two organic dyes onto specific sites of bigger biological molecules such as protein, peptides, DNA, and

RNA12-5428-74.

Equation 2.6 is plotted in Fig. 2.3. It is clear that the FRET efficiency is extremely sensitive to small changes in distance, when the distance is near R0. This feature makes smFRET an ideal tool to study molecular level conformation dynamics in the range of 2-8nm.

25,26,28,34,35,41

22

Fig. 2.3 The relation between FRET efficiency and donor-acceptor distance. The efficiency ranges from 0 to 1 and it increases as the distance decreases. The FRET efficiency is most sensitive when it is close to 50%, or the distance between donor and acceptor is close to R0.

23

The fluorescence signal from single molecules is weak. To observe this type of signal, high-powered illumination and means of reducing background fluorescence are required. In the arsenal of modern high resolution microscopes, there are numbers of techniques that can fulfill these requirements, such as confocal microscopes, total internal reflection fluorescence(TIRF), highly inclined, and the laminated optical sheet (HILO), near field scanning optical (NSOM), or zero-mode wave guides21,25,27,30,32,. In the work of my thesis, we used laser scanning confocal microscopy to performthe experiment.

There are numbers of characteristics of ideal dyes used for smFREM studies. They need to be photo-stable, bright, and excitable and emitting in the visible wavelength. They’d better to be relatively small to the host biological molecule so that any possible perturbation is minimized. 28-32, 35, 37, 42 They also need to be commercially available in the form that can be conjugated to biomolecules. Easy conjugation chemistry is preferred. Most importantly, solubility is often the limiting factor here. Other than small size organic dyes, quantum dots and fluorescent protein can also be used in smFRET, but they have more limitation due to their size and worse photo-stability in the case of protein. The most popular way to link these organic dyes are conjugating them onto the cysteine group of the biomolecules. For proteins, in many cases, there is not enough cysteine groups on the protein, or the positions of the cysteine groups are not desirable. This limitation can be overcome by using gene mutation and gene expression. Researchers will insert or replace other residue by cysteine group at carefully chosen position in order to monitor the relative motion between two or more specific domains.

One of the most commonly used pair of donor and acceptor in smFRET experiment is Cy3

24

and Cy5. The spectrum is shown in Fig. 2.4. From the graph, it is clear that the emission spectrum of Cy3 and extinction spectrum of Cy5 overlap in the region from 550-625nm pretty well. These pair has a R0 about 5nm. And the separation between their emission spectrum is very significant. The fluorescence signal from these two dyes can be easily separated by placing a 615nm-625nm long pass dichromatic mirror at the entrance of the detectors. There will be still some degree of cross-talking that can be later adjusted by modifying the equation of FRET efficiency calculation. The leakage in between two channels can be compromised to collection efficiency of each channel. Combining with the quantum yield of the each channel, correlation factor " is produced. It is calculated as the ratio of change in the acceptor intensity A, to change in the donor intensity, D upon acceptor photo-bleaching. The final expression of FRET efficiency is calculated as:

1 C I S E  D1  D T D T (2.8) E I A U

25

Fig 2.4 Spectrum of Cy3 and Cy5. Blue curves are absorption spectrums, and red curves are emission spectrums. The overlap of FRET transition is showed in orange. 2.1.3 AFM basic: imaging and force spectroscopy

Atomic force microscopy (AFM) was first developed by Gerd Binning and Heinrich

Rohrer in the early 1980s. They won the Nobel Prized for Physics in 1986 by making this huge contribution to the scientific society. AFM is one of the most important tools for imaging, measuring, and manipulating matter at nanometer scale. Its resolution is on the order of sub-nanometer level, which is 1000 times better than the optical diffraction limit.

Because it utilized mechanical probe to detect the height of the surface, it can be widely used to investigate the electrically non-conductive materials, which STM cannot measure. The

AFM consists of a cantilever with a sharp tip at its end. The tip apex curvature ranges from

10-50nm depending on the material and coating. According to Hooke’s law, forces between the tip and sample lead to a deflection of the cantilever. The deflection is detected and

26

recorded while scanning and used to produce a surface image. There are numbers of way to detect the deflection of cantilever. The very first designed AFM used a STM tip on top of the cantilever to detect the up and down motion of the cantilever. In 1987, Wickramsinghe et al. first use light-lever mechanism to detect the deflection. The modern commercial AFM are mostly using a modified version of light lever. The backside of the cantilever serves as a mirror to reflect the light shining from the top to a photodiode. In order to project the one dimensional movement of cantilever defection, the photodiode is divided into at least two parts. For example, the photodiode can be split into top and bottom two parts. The collected photon intensity on this two parts change while the cantilever bends upward and downward.

Most commercial AFM used a photodiode split into four parts. These allow the AFM to detect the left and right twisting motion of AFM cantilever, which is caused by scanning through a complex surface.

While obtaining image, the scanning can be achieved by mounting a 3-D piezoelectric scanner onto either the sample or the cantilever. If the tip was scanned at a constant height, there is a risk that the tip collides with the surface, causing damage to a soft and fragile surface or damage to the tip itself from a hard and rough surface. Hence, a feedback mechanism is employed to control the tip to sample distance while maintain a constant force between the tip and the sample. This mechanism is demonstrated in Fig. 2.5. The Z-axis control of the peizo driver is responsible to provide feedback, while the X-Y axis control is responsible to perform raster scan. Measuring the Z-axis reading (tip-to-sample distance) at each (x,y) data point allows the scanning software to construct a topographic image of the sample surface.

27

The mechanism discussed above is also called contact mode AFM. There are other two types of AFM mode: non-contact mode and tapping mode. Instead of sensing the height of the surface and how heavily the cantilever is bent, these two modes are sensing the frequency and amplitude while the tip is oscillating on top of the surface. The different sample surface condition will change the frequency and amplitude of these oscillations since the tip will encounter different attractive or repulsive regime.

In non-contact mode, the tip is oscillated at the resonance frequency and the amplitude of the oscillation is kept constant. Tapping mode is somewhere in between the contact and non-contact mode. The tip will periodically touch the surface; the amplitude is often set at about 50-60% of the amplitude of non-contact mode.

Force constant of the AFM tip ranges from 0.01-100N/m, and resonance frequency ranges from 5-500kHz. The high constant and low frequency tips are used in contact mode; the low constant and high frequency tips are used in non-contact and tapping mode.

In summary, contact mode imaging is heavily influenced by frictional and adhesive forces, and can damage samples and distort image data. Non-contact imaging provides low resolution and can also be hindered by surface contaminant like water layer, which can interfere with oscillation. Tapping mode imaging takes the advantages of the two above.

It eliminates frictional forces by intermittently contacting the surface and oscillating with sufficient amplitude to avoid the influence of contaminant layer.

Other than nanometer resolution imaging, another very important application of AFM is force spectroscopy. Force spectroscopy in general is a set of technique for the interaction force between individual molecules. The techniques include the most commonly used three:

28

AFM, optical tweezers, and ; and other invention of methods such as micro-needle manipulation, biomembrane force probe and flow-induced stretching. These methods measure the mechanical properties of single molecule, especially biomolecules.

Force apply force and

Fig 2.5 The schematic representation of the principal of AFM microscope. The detection laser is normally sent from the top side of the cantilever. After reflection from the cantilever, the direction of the laser is detected by a photodiode. The feedback system is used to maintain the position of the laser spot at a fix position.

29

detect displacement at single molecular level. The capacity ranges 10-10-10-4m in terms of length and 10-14-10-8N in terms of force. The object of manipulation ranges from single cell to single base pair advanced by RNA polymerase.

For the most popular three, each of them has their own advantages: AFM can be coupled with high-resolution imaging, and it benefits from the easiest and rapidest sample preparation among the three; optical tweezers has very low-noise and low drift dumbbell geometry; magnetic tweezers apply the cleanest manipulation as there is less chemical or optical interaction other than the mechanical perturbation in study. They also have their disadvantages: AFM is a high stiffness probe and there will be troubles to deal with non-specific interaction in some cases; optical tweezers could potentially have photo-damage to the sample, and the prolonged measurement could become unstable due to sample heating; magnetic tweezers has a drawback of force hysteresis.

In my thesis, I chose AFM as the methodology to perform the experiment. Force spectroscopy mode of AFM is very similar to contact mode. The same type of tip is used.

(High force constant and low resonance frequency) However, instead of lock-in the tip-to surface distance when acquiring image, the cantilever actual goes through a circle of bent-release-stretch. In these experiments, single molecules in study are immobilized on surface. The AFM tip is approaching from the top. The general process is: (1) the AFM tip is brought to the top of the single molecule in study; (2) the AFM tip moves downward and approaches the molecule, gradually, repulsive force is built up and the cantilever is bent backward; (3) after reaching a pre-set limit, the AFM tip starts to retrace and gradually releases the repulsive force; (4) the AFM tip finally detaches from the surface, but

30

tip-to-molecule interaction keeps the molecule attached to the tip and the molecule will be stretched upward. Throughout the process, the attractive and repulsive force built upon the

AFM tip will be recorded and used to plot a “force curve”. The conceptual figure of this whole process is demonstrated in Fig 2.6.

In force spectroscopy experiments, in order to achieve specific interaction between the sample molecule and the handles (cantilever tips and surfaces) for stretching measurement, the ends of the molecule have to be attached specifically41-47. The attachment between the molecule and surface is very similar to other single molecular optical measurement. The attachment between molecule and cantilever tip could rely on physical attraction between the molecules and a coated tip, or special mechanical characteristics such as: , avidin-biotin bonds, , oligomer of Titin I27 domain, and so on37, 41-43. The detailed experiment design will be discussed in the later chapter respectively.

In the force curve, the stretching curves naturally draw the most attention from researchers.

These single molecule force-extension measurements provide valuable information on the structure, folding unfolding processes and even activity of the biomolecules. However, in the last two chapter of my thesis, the main focus is the approaching part, where the molecule has been squeezed. There has been some work reporting measures of response of a single virus structure to AFM tip squeezing. We tried to apply a smaller force on single protein molecule to see the response. Very interesting behavior has been found and explained.

Conclusion and hypothesis have been made. The details will be discussed in these chapters later. In terms of technique, the way we obtained the force spectroscopy is no different from the experiment focusing on the single-molecule force-extension curves.

31

Fig. 2.6 The schematic representation of the process of the force curve measurement in force spectroscopy experiment. The semicircle represents the tip curvature of the AFM tip. Green, blue and red coils represent a protein molecule at relax (nature), pressured, and stretched state respectively. The force curve can either be plotted as force vs. time or force vs. tip displacement (pieozo driver to surface distance).

32

2.2 Experiment setups

2.2.1 Single-molecular FRET measurement

In chapter III, the measurement has been performed by a classic two-channel laser scanning confocal microscopy. And this microscope is also used to serve as the optical part of the correlated measurement setup used in chapter IV. The only difference between them is the dichromatic mirror used before sending the signal light to the detectors, as two experiment used different pairs of FRET donor and acceptor. The experiment in chapter III uses

Rhodamin 6G/Texas red, and the experiment in chapter IV uses Cy3 and Cy5. The detail will be discussed detailed in the later chapters.

The setup is demonstrated in Fig. 2.7. A continuous-wave (CW) green color solid state laser

(532nm) is used as an excitation source. The light is sent into the optical box by two reflection mirrors, and then a 545 long pass dichromatic mirror(Chroma) is used to separate the excitation light from the fluorescence signal from the sample. The objective lens used here is a Zeiss 100 x 1.5 NA oil immersion oil objective lens. Because the signal comes from single molecular fluorophore, another high performance 545 long pass filter is used to further eliminate the leakage of the relative strong excitation light before it is sent in to the detectors.

The FRET signal is split via a 595nm long pass dichromatic mirror and collected by two separate avalanche photodiode detectors (APD- Perkin Elmer, SPCM-AQR-13). The sample positioning and raster scanning is achieved by using a 2-D 16bit 100X100 )m piezoelectric scanner from MadCity lab.

33

Fig. 2.7 The schematic representation of the experiment set-up used in chapter III, and the optical element of the set-up used in chapter IV. In Figure A, L is the 532nm CW laser, DF is the density filter, M1 is a reflection mirror bringing the light into optical box, M2 is a reflection mirror bringing the light into microscope, FL is a focus lens, F is a 545nm long pass filter, DM is a dichromatic mirror, the green and red rectangular in the box are APDs of donor and acceptor channel respectively, and the blue box is a representation of the sample stage. Figure B is a side view of the inside structure of microscope. Obj is the objective lens, DM is the dichromatic mirror, M is a reflection mirror, and sample chamber is located on top of the 2D-scanning stage. 2.2.2 AFM setup and correlated measurement

In order to study the conformational response of protein molecule under the mechanical perturbation of AFM tip, we developed a correlated measurement technique to combine the smFRET microscopy and the AFM microscopy. The motivation of this development is to setup a unique method that has both spatial and optical resolution at the single molecular level. On top of that, the advantage of high spatial resolution of AFM could be use to apply molecular level perturbation on single protein molecule on desire. The experiment setup of the correlated measurement is shown in Fig. 2.8.

Based on the smFRET microscopy described in the previous section, the AFM part of the set-up is equipped via a home built aluminum alloy integrated stage. It is made of two parts: the bottom part is fixed on the microscope body, and the top part could be moved by a 2-D positioner. This is used to adjust the relative position between the AFM tip and sample.

34

The AFM scanner is mounted on this stage via a home built holder that is designed to fit the shape of the commercial available AFM scanner, in order to minimize vibration and drifting of the scanner. There is a window on top of the AFM scanner and a top-side camera on top of it to monitor the position of the tip cantilever. In order to minimize the interference between AFM detection light from the top and laser excitation light from the bottom, we switch the wavelength of the AFM detection light from visible region to infrared region.

35

Fig 2.8 The schematic representation of the experiment set-up used in chapter IV, where the integrated stage is put on top of the microscope, the green object is the AFM scanner that is equipped onto the stage via a home built aluminum holder. There is a 2-D positioner located at the side of the stage to adjust the position of the holder and scanner. The purple cylinder is a top-side camera to monitor the relative position of the AFM tip and laser focus.

36

The main challenge of the experiment is the alignment of AFM tip and laser focus with a precision comparable to the radius of the AFM tip curvature. Because only by making sure both the optical and AFM instrument are collecting signal from the exact same single molecule, could the response of the single molecule under perturbation be monitored. The procedures of AFM-focus-point alignment are demonstrated in Fig. 2.9. Practically, the relative position of tip and laser focus in step A, B, and C can be viewed through top-side camera. In the figures, the size of laser focus is overstated in order to have a better view.

First, positioner of the integrated stage is used to move the tip on top of the laser focus (A to

B). Secondly, an image around the laser focus is obtained and single molecules immobilized on the cover glass will be spotted. Then, we use the 2-D piezoelectric scanner to move the sample stage around and let a targeted single molecule be focused by the laser light (B to C). After this step, the relative position between tip and laser focus is in the range of couple hundred of nanometers. Finally, an AFM tip scanning is implemented around the tip position in an area of about 300 *300nm. Instead of using regular AFM cantilever deflection feedback, optical signal collected by APD is used to plot the resulting image. With the help of this image, the piezoelectric scanner of the AFM scanner will be used to move the AFM tip towards the center of the image spot as precisely as tens of nanometers. The size and shape of the image spot is not determined by the shape of laser focus point and diffraction limit but the topographic of AFM tip curvature. The AFM tip serves as a micro mirror to reflect photon from the single molecule underneath and collected by APD. The reflection is strongest when the sharpest point of the tip is right on top the molecule. So, by moving the AFM tip to the center of the image, AFM tip, the target single

37

molecule, and the laser focus point are aligned as precise as tens of nanometers. After these procedures, the X-Y position of the AFM tip will be fixed, and moving the sample stage alone can allow us to switch different single molecules and implement further measurements.

Due to instrumental drifting, further routine adjustments are required, but they are relatively minor compared to the initial alignment.

38

Fig. 2.9 The procedures of AFM-focus-point alignment in the AFM-FRET correlated measurement. The black rectangular is the AFM cantilever. The green circle is the laser focus point. The sizes of laser focus in the figures are drawn larger than their actual size. The red spots are single molecules immobilized on the cover glass.

39

2.2.3 MD simulation using GROMACS

Molecular dynamic simulation is a computer simulation of physical movements of atoms and molecules in the context of N-body simulation. The forces between particles and potential energy are defined by molecular mechanics force fields. The method is widely used in field like chemical physics, material science and modeling of biomolecules.

There are numbers of packages or applications for MD simulation purposes. Depends on the type of molecule in study, different package is chosen. GROMACS is primarily designed for simulation of proteins, lipids, and nucleic acids. It perfectly fit our usage. It was originally developed in the Biophysical Chemistry department of University of

Groningen, and it is an open source software maintained by contributors in academic institution around the world. It is one of the fastest and most popular software packages available in this field. The version we used was 4.5.1. We performed the simulation on the platform provided by Ohio Supercomputer Center.

The simulation is normally run under the following steps:

(1) Use PDB file (protein data bank) containing the molecular coordinates to generate a GRO

file that uses the formats it uses internally. This file not only contains the information of

molecule coordinates, but also the interacting force field or constraints between atoms

within the molecules.

Force field refers to the form and parameters of mathematical function used to describe

the potential energy of a system of particles. Different force field is based on different

assumption or the molecular level interaction. Different force field is suitable for

different molecule and system in study. GROMACS supports many different force fields.

40

We used GROMOS 96 force field, which is mainly used to simulate solution phase

bimolecular interactions and dynamics.

(2) Add in solvent, ion, and or other molecules into the GRO field to generate a simulation

box, then apply periodical boundary condition.

(3) Apply further constraints upon the system. For example, artificially lock some atoms or

molecules for the purpose of simulate real environment without bring in too much

complexity.

(4) Describe temperature and pressure condition of the system.

(5) Set addition property of the system, like deforming. Set simulation time.

(6) Perform the calculation, and record the trajectory.

(7) Analyze the output trajectory. A variety of information can be analyzed: the position of

each atom or molecule at any given time of the simulation, RSMD, potential energy, and

so on.

It is important to note that in step (2) a boundary condition has to be applied to the simulation box, as a system cannot be described in a way that the space outside of the simulation box is vacuum. The most commonly used boundary condition to simulate biological system is xyz periodical boundary condition, where there are 6 identical simulation boxes at 6 directions around one simulation box. And this is repeated infinitely. So the interaction across the simulation has to be concerned. For example, if the box is designed to be too small. The interaction between molecule in one simulation box and the one next to it will become significant and the result will be distorted.

We used GROMACS to perform a simulation that applies constant speed deformation

41

along one direction of the simulation box, and observe the respond of the protein molecule inside. This is for the purpose of understanding the potential energy profile of the single protein molecule collapse under the stress pressure of the AFM tip. More detail of the simulation we performed is discussed in Chapter IV.

2.3 Materials and Sample Preparation

2.3.1 Cover glass coating for single molecular optical measurement.

In order to get single molecular level resolution, the molecule in study has to be diluted in sample to make sure only one molecule could be detected at a time. In my experiments, these molecules are immobilized on glass surface with a density of one per tens of )m2.

This is achieved by using a mixture of active and non-active site reagent to coat the glass surface. A ratio of 1:10000 is typically used, in order to keep the surface density of the active cross-linking sites on glass surface very low. Applying cross-linking reagent and protein molecule on the coated glass surface will end up with a very low surface density of immobilized protein molecules. Under confocal microscope, the possibility of finding more than one molecule in study within an area defined by diffraction limit will be close to zero.

The cover slips was first treated by (3-Aminopropyl)trimethoxysilane +

Isobutyltrimethoxysilane (1:10000) 10% in DMSO for 4h. Different method will be used after this step depends on the molecule to tether.

In chapter III, and IV, we used covalent cross-linking, the coated glass will then be treated by 10nM [Dimethyl Suberimidate•2HCl] in 50mM PBS buffer (PH=8) for 1h, and incubated in PBS buffer (PH=7.8) containing dye labeled protein molecule for 2h. Methanol and water was used to wash the cover slips in the previous two steps, and PBS buffer (PH=7.5) was

42

used to wash after the final step. This method is shown in Fig. 2.10.

In chapter V, we used chelate effect of his-tag, Ni2+ion, and NTA. His-tag is polyhistidine-tag, which is an amino acid motif in protein that consists of at least six histidine residues. The initial usage of this structure is protein purification. Scientist often engineered this structure onto the domain of a protein that naturally doesn’t have a his-tag, in order to purify the protein that is difficult to purify otherwise. It later is used as a binding assay. This binding is highly specific and mechanically strong. This method is shown in

Fig. 2.11. Instead of using a symmetry bi-functional linker, an asymmetry bi-functional linker (AT-NTA) is used here to contribute 4 out of 6 dipolar bonds in the chelate ring of nickel ion.

More details about the sample like buffer conditions will be discussed in later chapters individually.

43

Fig. 2.10 The immobilization method used in chapter III, IV.

Fig. 2.11 The immobilization method used in chapter V.

44

2.4 References

1. Minsky, M. Memoir on inventing the confocal scanning microscope. Scanning 1988, 10, 128-138.

2. Minsky, M. Microscopy Apparatus: US 3,013,467 .

3. Shotton, D. M. Confocal scanning optical microscopy and its applications for biological specimens. J. Cell. Sci. 1989, 94, 175-206.

4. Paddock, S. W. Further developments of the laser scanning confocal microscope in biomedical research. Proc. Soc. Exp. Biol. Med. 1996, 213, 24-31.

5. Denk, W.; Strickler, J. H.; Webb, W. W. Two-photon laser scanning fluorescence microscopy. Science 1990, 248, 73-76.

6. Poole, C. A.; Ayad, S.; Gilbert, R. T. Chondrons from articular cartilage. V. Immunohistochemical evaluation of type VI collagen organisation in isolated chondrons by light, confocal and electron microscopy. J. Cell. Sci. 1992, 103 ( Pt 4), 1101-1110.

7. Schatten, G.; Pawley, J. B. Advances in optical, confocal, and electron microscopic imaging for biomedical researchers. Science 1988, 239, G164,G48.

8. Wilke, V. Optical scanning microscopy—the laser scan microscope. Scanning 1985, 7, 88-96.

9. Petran, M.; Hardavsky, M.; Egger, M. D.; Galambos, R. Tandem-scanning reflected-light microscope. JOSA 1968, 58, 661-664.

10. Paddock, S. W. Tandem scanning reflected-light microscopy of cell-substratum adhesions and stress fibres in Swiss 3T3 cells. J. Cell. Sci. 1989, 93 ( Pt 1), 143-146.

11. Förster, T. Zwischenmolekulare Energiewanderung und Fluoreszenz. Annalen der Physik 1948, 437, 55-75.

12. Ha, T.; Ting, A. Y.; Liang, J.; Caldwell, W. B.; Deniz, A. A.; Chemla, D. S.; Schultz, P. G.; Weiss, S. Single-molecule fluorescence spectroscopy of enzyme conformational dynamics and cleavage mechanism. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 893-898.

13. Kapanidis, A. N.; Laurence, T. A.; Lee, N. K.; Margeat, E.; Kong, X.; Weiss, S. Alternating-laser excitation of single molecules. Acc. Chem. Res. 2005, 38, 523-533.

14. Michalet, X.; Weiss, S.; Jäger, M. Single-molecule fluorescence studies of protein folding and conformational dynamics. Chem. Rev. 2006, 106, 1785-1813.

45

15. Seidel, R.; Dekker, C. Single-molecule studies of nucleic acid motors. Curr. Opin. Struct. Biol. 2007, 17, 80-86.

16. Zhuang, X. Single-molecule RNA science. Annu. Rev. Biophys. Biomol. Struct. 2005, 34, 399-414.

17. Deniz, A. A.; Laurence, T. A.; Beligere, G. S.; Dahan, M.; Martin, A. B.; Chemla, D. S.; Dawson, P. E.; Schultz, P. G.; Weiss, S. Single-molecule protein folding: diffusion fluorescence resonance energy transfer studies of the denaturation of chymotrypsin inhibitor 2. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 5179-5184.

18. Best, R. B.; Merchant, K. A.; Gopich, I. V.; Schuler, B.; Bax, A.; Eaton, W. A. Effect of flexibility and cis residues in single-molecule FRET studies of polyproline. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 18964-18969.

19. Merchant, K. A.; Best, R. B.; Louis, J. M.; Gopich, I. V.; Eaton, W. A. Characterizing the unfolded states of proteins using single-molecule FRET spectroscopy and molecular simulations. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 1528-1533.

20. Schuler, B.; Eaton, W. A. Protein folding studied by single-molecule FRET. Curr. Opin. Struct. Biol. 2008, 18, 16-26.

21. Ha, T.; Chemla, D.; Enderle, T.; Weiss, S. Single molecule spectroscopy with automated positioning. Appl. Phys. Lett. 1997, 70, 782-784.

22. Sabanayagam, C. R.; Eid, J. S.; Meller, A. High-throughput scanning confocal microscope for single molecule analysis. Appl. Phys. Lett. 2004, 84, 1216-1218.

23. Hohng, S.; Ha, T. Single䇲㻹 㼛㼘㼑㼏㼡㼘 㼑㻌㻽㼡㼍㼚㼠㼡㼙䇲 㻰㼛㼠㻌㻲㼘㼡㼛㼞㼑㼟㼏㼑 nce Resonance Energy Transfer. ChemPhysChem 2005, 6, 956-960.

24. Hohng, S.; Ha, T. Near-complete suppression of quantum dot blinking in ambient conditions. J. Am. Chem. Soc. 2004, 126, 1324-1325.

25. Aitken, C. E.; Marshall, R. A.; Puglisi, J. D. An oxygen scavenging system for improvement of dye stability in single-molecule fluorescence experiments. Biophys. J. 2008, 94, 1826-1835.

26. Ryu, Y.; Schultz, P. G. Efficient incorporation of unnatural amino acids into proteins in Escherichia coli. Nature methods 2006, 3, 263-265.

27. Higuchi, R.; Krummel, B.; Saiki, R. K. A general method of in vitro preparation and specific mutagenesis of DNA fragments: study of protein and DNA interactions. Nucleic Acids Res. 1988, 16, 7351-7367.

46

28. Pennington, M. W. In Site-specific chemical modification procedures; Peptide Synthesis Protocols; Springer: 1995; pp 171-185.

29. Cha, T.; Guo, A.; Zhu, X. Enzymatic activity on a chip: the critical role of protein orientation. Proteomics 2005, 5, 416-419.

30. Rhoades, E.; Gussakovsky, E.; Haran, G. Watching proteins fold one molecule at a time. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 3197-3202.

31. - Lele, T. P.; - Pendse, J.; - Kumar, S.; - Salanga, M.; - Karavitis, J.; - Ingber, D. E. - Mechanical forces alter zyxin unbinding kinetics within focal adhesions of living cells. - Journal of Cellular Physiology , - 187.

32. Joo, C.; McKinney, S. A.; Nakamura, M.; Rasnik, I.; Myong, S.; Ha, T. Real-time observation of RecA filament dynamics with single monomer resolution. Cell 2006, 126, 515-527.

33. Luo, G.; Wang, M.; Konigsberg, W. H.; Xie, X. S. Single-molecule and ensemble fluorescence assays for a functionally important conformational change in T7 DNA polymerase. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 12610-12615.

34. McKinney, S. A.; Joo, C.; Ha, T. Analysis of single-molecule FRET trajectories using hidden Markov modeling. Biophys. J. 2006, 91, 1941-1951.

35. Munro, J. B.; Altman, R. B.; O'Connor, N.; Blanchard, S. C. Identification of two distinct hybrid state intermediates on the ribosome. Mol. Cell 2007, 25, 505-517.

36. Myong, S.; Bruno, M. M.; Pyle, A. M.; Ha, T. Spring-loaded mechanism of DNA unwinding by hepatitis C virus NS3 helicase. Science 2007, 317, 513-516.

37. Yang, S.; Cao, J. Direct measurements of memory effects in single-molecule kinetics. J. Chem. Phys. 2002, 117, 10996-11009.

38. Andrec, M.; Levy, R. M.; Talaga, D. S. Direct determination of kinetic rates from single-molecule photon arrival trajectories using hidden Markov models. The Journal of Physical Chemistry A 2003, 107, 7454-7464.

39. Andrecka, J.; Lewis, R.; Bruckner, F.; Lehmann, E.; Cramer, P.; Michaelis, J. Single-molecule tracking of mRNA exiting from RNA polymerase II. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 135-140.

40. Heinze, K. G.; Jahnz, M.; Schwille, P. Triple-color coincidence analysis: one step further in following higher order molecular complex formation. Biophys. J. 2004, 86, 506-516.

47

41. Lee, N. K.; Kapanidis, A. N.; Koh, H. R.; Korlann, Y.; Ho, S. O.; Kim, Y.; Gassman, N.; Kim, S. K.; Weiss, S. Three-color alternating-laser excitation of single molecules: monitoring multiple interactions and distances. Biophys. J. 2007, 92, 303-312.

42. Lang, R.; Pauleau, A.; Parganas, E.; Takahashi, Y.; Mages, J.; Ihle, J. N.; Rutschman, R.; Murray, P. J. SOCS3 regulates the plasticity of gp130 signaling. Nat. Immunol. 2003, 4, 546-550.

43. Liu, R.; Hu, D.; Tan, X.; Lu, H. P. Revealing two-state protein-protein interactions of calmodulin by single-molecule spectroscopy. J. Am. Chem. Soc. 2006, 128, 10034-10042.

44. He, Y.; Li, Y.; Mukherjee, S.; Wu, Y.; Yan, H.; Lu, H. P. Probing single-molecule enzyme active-site conformational state intermittent coherence. J. Am. Chem. Soc. 2011, 133, 14389-14395.

45. Lu, H. P. Single-molecule photon stamping FRET spectroscopy study of enzymatic conformational dynamics. Physical Chemistry Chemical Physics 2013, 15, 770-775.

46. He, Y.; Lu, M.; Cao, J.; Lu, H. P. Manipulating protein conformations by single-molecule AFM-FRET nanoscopy. ACS nano 2012, 6, 1221-1229.

47. Heymann, B.; Grubmüller, H. Dynamic force spectroscopy of molecular adhesion bonds. Phys. Rev. Lett. 2000, 84, 6126.

48. Florin, E. L.; Moy, V. T.; Gaub, H. E. Adhesion forces between individual ligand-receptor pairs. Science 1994, 264, 415-417.

49. Rief, M.; Oesterhelt, F.; Heymann, B.; Gaub, H. E. Single Molecule Force Spectroscopy on Polysaccharides by . Science 1997, 275, 1295-1297.

50. Smith, B. L.; Schäffer, T. E.; Viani, M.; Thompson, J. B.; Frederick, N. A.; Kindt, J.; Belcher, A.; Stucky, G. D.; Morse, D. E.; Hansma, P. K. Molecular mechanistic origin of the toughness of natural adhesives, fibres and composites. Nature 1999, 399, 761-763.

51. Izrailev, S.; Stepaniants, S.; Balsera, M.; Oono, Y.; Schulten, K. Molecular dynamics study of unbinding of the avidin-biotin complex. Biophys. J. 1997, 72, 1568-1581.

52. Carrion-Vazquez, M.; Oberhauser, A. F.; Fowler, S. B.; Marszalek, P. E.; Broedel, S. E.; Clarke, J.; Fernandez, J. M. Mechanical and chemical unfolding of a single protein: a comparison. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 3694-3699.

53. Lu, H.; Isralewitz, B.; Krammer, A.; Vogel, V.; Schulten, K. Unfolding of titin immunoglobulin domains by steered molecular dynamics simulation. Biophys. J. 1998, 75, 662-671.

48

54. Marszalek, P. E.; Lu, H.; Li, H.; Carrion-Vazquez, M.; Oberhauser, A. F.; Schulten, K.; Fernandez, J. M. Mechanical unfolding intermediates in titin modules. Nature 1999, 402, 100-103.

50

CHAPTER III. SINGLE MOLECULE FRET STUDY OF CAM-PEPTIDE BINDING

DYNAMICS

3.1 Introduction

Calmodulin(CaM) is a small protein approximately 148 amino acids long. It is expressed in many cell types and involved in processes such as intracellular signaling, smooth muscle contraction, and memory. In general, CaM serves as a calcium sensor and signal transducer.

Many of the proteins that CaM binds are unable to bind calcium themselves and rely on the binding with calcium activated CaM to catalyze biological reactions.

CaM has two approximately symmetrical domains, connected by a flexible -helix chain.

Each domain is an EF-motif hand, and can bind up to two Ca2+. The binding with Ca2+ will induce conformation change of the EF-hand and make it capable to bind with target proteins of CaM. Normally, the Ca2+ loaded EF-hand at C-domain of the CaM has a much higher affinity with binding sequence, especially for the case of ATPase1-18.

The effective binding sequence of Ca-ATPase, C28W, is a 28 amino acid oligmer. Studies uses C28W to stimulate the interaction between CaM and its target protein had been reported7,16. A two-step binding process has been hypothesized. After C-domain binds to

C28W, the N-domain is capable of binding around C28W peptide. Moreover, this binding tends to switch between close and open states. This flexible association has been probed by

NMR measurement and single-molecular polarization spectroscopy. The conformations of three forms of calmudulin are shown is Fig. 3.1.

51

Fig. 3.1 The conformation of calmodulin molecule: the left one is apo-CaM that is the calcium free form of calmodulin (1cfd); the middle one is the form of calmodulin binding with 4 calcium ions (1cll); the right one is CaM-C28W binding complex (2kne). Green and pink beads in the complex are shown to indicate the labeling position of the pair of FRET donor and acceptor respectively. 3.2 Experiment Section

3.2.1 Materials and sample preparation

The NHS ester rhodamine labeled CaM is used in this experiment. The labeling position is near the N-domain of the CaM. The sample is purchased from A.G. Scientific. The

C28W has Texas Red labeled onto its N-domain. The labeling positions in the complex are shown in Fig. 3.1. Texas Red’s absorption spectrum is well overlapped with the emission spectra of rhodamine. NHS ester rhodamine has a much narrower emission spectroscopy around 580nm than rhodamine 6G. The narrowed emission bandwidth is capable to be well isolated from the emission of Texas Red above 600nm. By using beam splitter at 595nm, the signal from two channels can be separated without too much cross talking.

A home built assay cell was used to hold the sample solution. Reaction mixture was dissolved in HEPES buffer (20mM HEPES, 100mMKCl, 5mM MgCl2, 100mM CaCl2,

PH=7.5). In control experiments, labeled Cam, Texas Red labeled C28W, and HEPES

52

buffer was used respectively.

Dye labeled CaM was covalently tethered onto glass surface by the methods described in the previous publication. The cover slips was first treated by

(3-Aminopropyl)-trimethoxysilane + Isobutyltrimethoxysilane (1:10000) 10% in DMSO for

4h, followed by 10nM [Dimethyl Suberimidate•2HCl] in 50mM PBS buffer(PH=8) for 4h, and incubated in PBS buffer(PH=7.8) containing dye labeled CaM for 2h. Methanol and water was used to wash the cover slips in the previous two steps, and PBS buffer (PH=7.5) was used to wash after the final step39.

3.2.2 Single-molecular FRET spectroscopy

Single molecule FRET experiment was performed on confocal microscope (Zeiss A1).

The experiment schematics are shown in Fig. 3.2. Laser source at 532nm was used as excitation light. The beam passed a beam expander and then a pin hole in order to decrease the power. In the microscope the excitation beam is reflected up by a diachronic beam splitter (Chroma 545DCLP), and focused by a 100X oil immersion objective lens (NA=1.5,

Zeiss) onto the upper surface of a microscope cover slip. The fluorescence was collected by the same objective lens, passed through a long-pass filter (Chroma HQ545), split by a diachronic beam splitter (Chroma 595DCLP) into two components, and detected by two separated APDs. The arrival time of each photon in two channels were sampled and treated into trajectory with desired binning time. A close-loop XY two dimensional scanning stage was used to hold the sample chamber. It is capable to scan over a square of 100X100μm with the resolution of a few nm. After imaging an area and locating several isolated CaM tethered on glass surface, raster scan software can drive the scanning stage and move a target

53

CaM right on top of the focus point.

Fig. 3.2 Experimental setup and FRET trajectory (A) The schematics of experiment setup using 532nm excitation laser, confocal scanning microscope, and two-channel photon detection; (B)Single molecule dye labeled CaM tethered on glass surface by bi-functional linker; (C)Single molecule FRET signal from Rhodamin(Green)/Texas Red(Red) respectively labeled on CaM and C28W; (D),(E),(F) correlation function of FRET trajectory, (D): auto-correlation of accepter channel, (E) auto-correlation of donor channel, (F) cross-correlation of two channels. 3.3 Results and Discussion

The buffer solution contained C28W with a respectively high concentration (10nM) in order to enhance the initial binding reaction between the C28W and the C-domain of the

CaM. Control measurement had been done by using the buffer containing labeled C28W alone, which was used as background subtraction for the CaM/C28W sample. After considering the cross-talking between two channels, the photons coming from two dyes could be calculated separately, and then used them to calculate the FRET efficiency and distance between two dyes respectively. We also tested samples with labeled CaM alone. After the same treatment with the data of CaM/C28W sample, the C28W sample gave FRET efficiency

54

about 0, and CaM sample gave FRET efficiency about 1. These were used to prove that contribution of the unbound CaM and C28W to the FRET is negligible. Even though a quantitative assignment of an exact distance range has significant error bars because of the relatively weak signal, spectral fluctuation, and photo-bleaching, the qualitative assignment of isolated states is still observable and essential to understanding the dynamics19-21. Other than mapping the distribution of the efficiency, we used some unique statistical way to present the data differently to show different features of the dynamics. Without quantitatively knowing the exact distance between two dyes at any time, lots of interesting dynamics information was revealed in this work.

Because FRET normally occurs in a range of 1-10nm, while there is no direct interaction between CaM and C28W, the possible FRET between two dyes is negligible. We also did control experiment with different concentration of Texas-Red labeled C28W. If we decreased the concentration by half, the photon counting level in the background decreased respectively, but the photon counting level from one specific target protein didn’t change much. This indicated that our data was probing the FRET between dyes on CaM/C28W that has direct interaction, the possible FRET from the unbound TexasRed-C28W passing by the target CaM did not contribute to the signal. This fact also proved that during our measurement time C28W stayed on the target CaM. It is highly possible that the stronger interaction between C-domain of CaM and C28W maintain the complex and FRET.

55

Fig. 3.3 Two different statistical analysis of the same FRET trajectory (A) Calculated FRET trajectory with 10ms binning time; (B) Distribution of FRET efficiency: each slide along x-axis is a distribution of a 5s time window, Y axis is FRET efficiency which share marks with (C). The color represents the occurrence of different efficiency; (C) Distribution of FRET efficiency of the whole trajectory.

56

The FRET trajectory is shown in Fig. 3.2, the green line represents the signal of donor channel and the red line represents the signal of acceptor channel. The binning used in the data is 10ms. The average photo-counting level is about couple hundreds of photon per binning. Fig. 3.2 D, E, F shows the auto-correlation function of individual channel and cross-correlation function of the two channels. The signal of donor and acceptor tend to anti-correlate with each other because increasing of one channel comes along with the decreasing of another channel, while the distance between two dyes is changing over time.

The expression of calculating FRET efficiency is showed in Equation 3.1.

F E  A  (3.1) FD FA

We can use this equation to calculate the FRET efficiency from the signal of donor and acceptor channels. By generating the FRET efficiency trajectory, we can study the dynamics of this complex in real time. Due to thermal fluctuation, the FRET efficiency in a long trajectory may drift up and down. This could cause some dynamics unclearness or undistinguishable states could appear. For example, if we used the whole trajectory in our experiment (50s or longer) to build a histogram and evaluate the distribution property of the

FRET efficiency, only a board Gaussian like distribution could be seen (Fig. 3.3 C). But by carefully looking at the detail of the trajectory, the FRET efficiency actually move up and down quite regularly and the motion is beyond the error bar. Basically, the “moving average” of the trajectory drifts up and down, and a high and a low could always be seen throughout the trajectory. This type of fluctuation cannot be visualized and analyzed by a simple distribution evaluation.

57

In order to resolve the fluctuation of E and the domain motion of this complex in real time, we treated the trajectory differently. Instead of consider the whole trajectory at one time, we spitted it into many overlapping segments, calculated their histograms, and finally combined them together onto a 2-D color table. In Fig. 3.3B, the occurrence of different E level is represented by color. Each slide along X axis is a calculated histogram of E within a time window of 5s, and the step between two slides is 500ms. We designed the step smaller than the time window in order to represent how the distribution is changing gradually over time.

The Y axis is FRET efficiency, and the occurrence in each histogram is represented by color.

We can see that FRET efficiency E majorly distributes around two levels or two bars.

Although the E level of each bar is shifting up and down over time, they can still be considered as two isolated states with no doubt, because they are separated by the less dense region in the middle along the Y axis and they are separately connecting themselves along X axis.

The upper bar in the color table is corresponding to the major peak in the global histogram or the close state of the complex. In contrast, the lower bar is corresponding to the minor peak and the open state. The fact that the higher bar is a much broader than the lower one indicates that the higher E state is a favorable state in the equilibrium compared to the lower

E state. This proves that the complex favor the close state in the experiment condition.

The higher E state can also be called as “close state” of the complex, and the lower E state can be called as “open state” of the complex. In this method, the global histogram is extended in another dimension that is time. This gives a better view of the fluctuation of E and better understanding about the dynamics of the complex.

58

Because the 2-D color table clearly shows the distinguished two states, we moved one step forward to spit the FRET trajectory into two states to further analyze their property separately.

We did not make a threshold and assign two states by evaluating whether the E is higher or lower than the threshold. However, we evaluated the jumping of each data point from the previous data point. Whenever there is a continuous upward jumping, we considered it to be a switch from low to high state. Whenever there is a continuous downward jumping, we considered it to be a switch from high to low state. By doing so, the FRET trajectory is resolved to be continuous switching between two states. The time duration of each state and the E level of each state were recorded. The Fig. 3.4A shows a segment of the trajectory.

Although the corresponding FRET efficiency of both states are changing over time, statically they are still distinguishable, which is showed in the previous 2-D color table. Because each data point is a binning of 10ms, the result shows that each state tends to stay for a range from tens of milliseconds to hundreds of milliseconds. This is consistence with our data reported previously. By looking at the distribution of the two states, we found that the close state

(state with higher E) tends to have a larger range of duration comparing to the open state

(state with lower E), which mainly exists for less than hundred milliseconds.

59

Fig. 3.4 Representation of how the trajectory is split into two states by judging the jumping of each data point. (A) A segment of the FRET trajectory; (B) The distribution of the duration of each state: close state in red, and open state in Green.

Fig. 3.5 The relation between adjacent close state duration and FRET efficiency (A) The 2-D color table shows the adjacent feature of the close state, the X axis is the duration of close state N, and the Y axis is the duration of close state N+1. The color represents the concurrency; (B) The intensity graph shows the relation between the FRET efficiency of each close state and its time duration. As the close state is the major form of the complex, we analyzed its behavior by evaluating the changing of time duration that each time the complex stays in close state. In Fig 3.5A, the X axis is the duration of close state N, and the Y axis is the duration of close state N+1, which is the close binding state right after N. We found a clear diagonal feature in this graph. The protein-protein complex tends to stay in close state for a similar amount of time with the previous one in despite of the large ranger of this duration. This observation indicates possible regulation of the binding affinity by another process, which is slower than

60

the binding process. This type of behavior is very common in biology system. Another graph is about the relationship between the time duration and average E of close states.

Interestingly, from the graph we found that N-domain of CaM tends to tightly bind with

C28W for a longer time while it has a higher E or closer distance between dyes. This correlation between binding duration and binding conformation can be explained by the following facts. C-domain of CaM has a much higher affinity with Ca2+ than N-domain;

C-domain will bind Ca2+ first and switch to its activated form. Moreover, the Ca2+ loaded

C-domain has about 100 fold higher affinities with target peptide C28W than the Ca2+ loaded

N-domain16,22-25. It tends to bind with C28W first following by N-domain. Experiments have been reported to prove that C-domain tends to bind with the N-domain of C28W first rather than C-domain. After the binding, the effective concentration of N-domain around

C28W becomes much higher and the 100 fold weaker binding process becomes significant and measurable. In our experiment, the accepter dye Texas red is labeled on N-domain of

C28W, and donor dye rhodamine is labeled near the N-domain of CaM. Because the calculated FRET efficiency E reflects the distance between donor and acceptor dye, the relation between close time and average E may be due to the fact that the N-domain undergo an equilibrium that contains more than one sub-states before binding onto C28W.

Fig. 3.6 shows the model we hypothesized based on the statistics result of FRET trajectory analysis. In the figures, (A) and (B) are the open states that are showed as the states with lower FRET efficiency in Fig. 3.3, where the peptide C28W has already bind to the C-domain of CaM while the N-domain is still not yet form any significant interaction with the peptide.

It is easy to predict that the N-domain of the CaM could potentially oscillate among a series

61

of different conformations. We only show the two extreme cases in the figures and these sub-states are represented by the darkness of the blue color. (C) and (D) are the close states that are corresponding to the states that are a higher FRET efficiency value in Fig. 3.3, and these states occupy the majority of the duration in any given trajectory we obtained. We predict that different sub-states of the N-domain could leads to different level of “affinity potentials”. This means that at one end of the sub-states series, which is showed in darker blue in the figures, will serve as better candidates to bind onto C28W. After binding, the resulting complex will have a “tighter” structure, smaller distance between two domains, and longer duration or lifetime.

We tried to understand the conformational or chemical nature of these projected sub-states of N-domain. We examined the electrostatic distribution of known crystal structure of different form of CaM. The most interesting ones that potentially support our claim are showed in Fig. 3.7.

62

Fig 3.6 Schematic representation of the hypothesized equilibrium of CaM-C28W binding. Two domains of CaM are both EF-hand motif, which is also a popular secondary structure in many other proteins. It is know that upon binding with calcium ions, EF-hand motif will open up. Fig. 3.7(A) is the N-domain structure of apo-CaM (calcium free CaM), and (B) is the calcium loaded form of it. It is easy to see from the electrostatic surface that once opened up; a region of negative charge region (in red color) will be exploded to be ready to interact with potential binding sequences. In the close form (apo-CaM), this region is hiding inside.

63

Fig. 3.7 The APNS electrostatic distribution of different forms of CaM C or N domain. (A) N-domain of apo-CaM in ribbon and surf style; (B) N-domain of Ca2+ activated CaM in ribbon and surf style. (C) C-domain of CaM-C28W complex, with and without showing peptide. (D) N-domain of CaM-C28W complex, with and without showing peptide. By examining the electrostatic of CaM-C28W binding complex (2kne), we found convincing relationship between the area of this exploded negative charge region and the binding affinity with C28W. (C) is the EF-hand motif on C-domain, which reported to have a much higher binding affinity with most of CaM’s binding peptides, has a much larger “red color area” compared to the EF-hand motif on N-domain. It is straightforward to draw the conclusion that an order of binding affinity from high to low is (C)ˈ(D), (B), and (A).

Based on the result in Fig. 3.5, instead of seeing two isolated form of close state complexes, we observed a sequence of continuous distributed close state complexes. It is highly likely that the differential binding affinity is due to more factors excluding the number of binding calcium ion.

In order to fully explain these observations, further experiment need to be done. However, our data already shows some interesting behavior of the CaM-C28W complex. This type of interaction is always inhomogeneous. Especially while multiple binding domains and

64

multiple sub-states of the same binding domain are involved, the affinity of any of them may be dramatically influenced by the equilibrium that intrinsically exists.

3.4 Conclusion

Using designed labeling of dyes, single-molecular FRET spectroscopy, and some unique mathematical treatment, we studied the interaction between Ca2+ activated CaM and C28W peptide in Ca2+ saturated condition. The tethering of single-molecule on glass surface allows the long time confocal measurement. Well isolated emission spectra of donor and acceptor can enhance the FRET signal and lower the noise level. The study revealed the two-step binding behavior of CaM and C28W. C-domain tends to bind with C28W first with a much higher affinity, followed by the second step, which is the binding of N-domain.

Due to the comparative low affinity, the N-domain switches between tightly and open states while C-domain remains close with C28W. Such fluctuations between these two states are found to be in a time scale ranging from tens of milliseconds to hundreds of milliseconds.

This result is consistent with our previous study using single-molecular polarization spectroscopy. We used a 2-D color table to demonstrate how the drifting FRET efficiency in a long trajectory can still present the two-state feature of the complex, and then analyze the properties of close states as the majority form. By plotting the adjacent graph and distribution of average E versus time duration, we found that N-domain seems to have more than one sub-state, and they show different affinity with C28W. We hypothesized that the intrinsic equilibrium of N-domain conformational fluctuation plays an important role in the process of CaM-peptide binding. Further study needs to be done in order to fully explain this regulation. These dynamics can only be observed by single-molecule measurement. It

65

is likely that the inhomogeneous interaction involving multiple domain and binding sites are general and essential for protein-protein interactions.

66

3.5 References

1. Chin, D.; Means, A. R. Calmodulin: a prototypical calcium sensor. Trends Cell Biol. 2000, 10, 322-328.

2. James, P.; Vorherr, T.; Carafoli, E. Calmodulin-binding domains: just two faced or multi-faceted? Trends Biochem. Sci. 1995, 20, 38-42.

3. Babu, Y. S.; Sack, J. S.; Greenhough, T. J.; Bugg, C. E.; Means, A. R.; Cook, W. J. Three-dimensional structure of calmodulin. Nature 1985, 315, 37-40.

4. Chattopadhyaya, R.; Meador, W. E.; Means, A. R.; Quiocho, F. A. Calmodulin structure refined at 1.7 A resolution. J. Mol. Biol. 1992, 228, 1177-1192.

5. Kuboniwa, H.; Tjandra, N.; Grzesiek, S.; Ren, H.; Klee, C. B.; Bax, A. Solution structure of calcium-free calmodulin. Nat. Struct. Biol. 1995, 2, 768-776.

6. Vetter, S. W.; Leclerc, E. Novel aspects of calmodulin target recognition and activation. Eur. J. Biochem. 2003, 270, 404-414.

7. Chen, B.; Mayer, M. U.; Markillie, L. M.; Stenoien, D. L.; Squier, T. C. Dynamic motion of helix A in the amino-terminal domain of calmodulin is stabilized upon calcium activation. Biochemistry 2005, 44, 905-914.

8. Chang, S. L.; Szabo, A.; Tjandra, N. Temperature dependence of domain motions of calmodulin probed by NMR relaxation at multiple fields. J. Am. Chem. Soc. 2003, 125, 11379-11384.

9. Baber, J. L.; Szabo, A.; Tjandra, N. Analysis of slow interdomain motion of macromolecules using NMR relaxation data. J. Am. Chem. Soc. 2001, 123, 3953-3959.

10. Evenas, J.; Forsen, S.; Malmendal, A.; Akke, M. Backbone dynamics and energetics of a calmodulin domain mutant exchanging between closed and open conformations. J. Mol. Biol. 1999, 289, 603-617.

11. Johnson, J. D.; Snyder, C.; Walsh, M.; Flynn, M. Effects of myosin light chain kinase and peptides on Ca2+ exchange with the N- and C-terminal Ca2+ binding sites of calmodulin. J. Biol. Chem. 1996, 271, 761-767.

12. Kasturi, R.; Vasulka, C.; Johnson, J. D. Ca2+, caldesmon, and myosin light chain kinase exchange with calmodulin. J. Biol. Chem. 1993, 268, 7958-7964.

13. Malmendal, A.; Evenas, J.; Forsen, S.; Akke, M. Structural dynamics in the C-terminal domain of calmodulin at low calcium levels. J. Mol. Biol. 1999, 293, 883-899.

67

14. Peersen, O. B.; Madsen, T. S.; Falke, J. J. Intermolecular tuning of calmodulin by target peptides and proteins: differential effects on Ca2+ binding and implications for kinase activation. Protein Sci. 1997, 6, 794-807.

15. Kranz, J. K.; Flynn, P. F.; Fuentes, E. J.; Wand, A. J. Dissection of the pathway of molecular recognition by calmodulin. Biochemistry 2002, 41, 2599-2608.

16. Sun, H.; Yin, D.; Squier, T. C. Calcium-dependent structural coupling between opposing globular domains of calmodulin involves the central helix. Biochemistry 1999, 38, 12266-12279.

17. Seaton, B. A.; Head, J. F.; Engelman, D. M.; Richards, F. M. Calcium-induced increase in the radius of gyration and maximum dimension of calmodulin measured by small-angle X-ray scattering. Biochemistry 1985, 24, 6740-6743.

18. Heidorn, D. B.; Seeger, P. A.; Rokop, S. E.; Blumenthal, D. K.; Means, A. R.; Crespi, H.; Trewhella, J. Changes in the structure of calmodulin induced by a peptide based on the calmodulin-binding domain of myosin light chain kinase. Biochemistry 1989, 28, 6757-6764.

19. Ikura, M.; Clore, G. M.; Gronenborn, A. M.; Zhu, G.; Klee, C. B.; Bax, A. Solution structure of a calmodulin-target peptide complex by multidimensional NMR. Science 1992, 256, 632-638.

20. Osawa, M.; Tokumitsu, H.; Swindells, M. B.; Kurihara, H.; Orita, M.; Shibanuma, T.; Furuya, T.; Ikura, M. A novel target recognition revealed by calmodulin in complex with Ca2+-calmodulin-dependent kinase kinase. Nat. Struct. Biol. 1999, 6, 819-824.

21. Nie, S.; Zare, R. N. Optical detection of single molecules. Annu. Rev. Biophys. Biomol. Struct. 1997, 26, 567-596.

22. Vorherr, T.; James, P.; Krebs, J.; Enyedi, A.; McCormick, D. J.; Penniston, J. T.; Carafoli, E. Interaction of calmodulin with the calmodulin binding domain of the plasma membrane Ca2+ pump. Biochemistry 1990, 29, 355-365.

23. Yao, Y.; Gao, J.; Squier, T. C. Dynamic structure of the calmodulin-binding domain of the plasma membrane Ca-ATPase in native erythrocyte ghost membranes. Biochemistry 1996, 35, 12015-12028.

24. Chen, B.; Mayer, M. U.; Markillie, L. M.; Stenoien, D. L.; Squier, T. C. Dynamic motion of helix A in the amino-terminal domain of calmodulin is stabilized upon calcium activation. Biochemistry 2005, 44, 905-914.

68

CHAPTER IV. PROTEIN CONFORMATION SUDDEN COLLAPSES UNDER

STRESS FORCE MANIPULATION BY SINGLE-MOLECULE AFM-FRET

NANOSCOPY

4.1 Introduction

Proteins undergo dynamic transitions including folding, unfolding, and conformational rearrangements to perform biological functions in living cells1,2. The transitions are typically from an unfolded protein state, an expanded random coils and disordered structure, to a folded protein state, a compact globules and ordered structure. It has been suggested that a molten globular state of a protein is distinct from the folded state, and is more similar to the collapsed state of a polypeptide, a predominantly disordered conformational state3-5.

The intermediate molten globular state of the protein molecules can go downhill folding or intrinsic disorder, which is closely related to their biological function, such as catalysis activity6. In the processes of conformational transition from unfolding to folding state, there is a probability for mis-folding and aggregation, for example, forming the Alzheimer’s disease-specific amyloid fibrils7-9. Therefore, understanding the stability of protein molecules and the mechanisms of protein molecule collapse in a free-energy funnel-shape landscape is of great interest and importance in biophysics.10-12

Protein molecular structure such as amino-acid sequence and its local environment such as solvent molecules, neighborhood biological molecules, metal ions, and local ionic strength and pH are the main factors to determine the pathway a protein molecule folds and forms its functional structure. In addition, the impact of external mechanical force on protein folding and unfolding is often significant. Mechanical force on proteins can be significant in living

69

cell, being originated from the biological component movements and interactions, such as, molecular aggregation, molecular partition crowding, hydrodynamic stress, and cell surface tension13-15. There have been a number of research works focusing on understanding the relation between mechanical force and biological behaviors13-18. Even though most of them are discussed on the cellular level, they suggest that the importance of understanding the molecular principles within these interplaying parameters and factors. Protein is the carrier of majority of biological behaviors. If mechanical stress force can change the property of a biological system, there must be an approach directly changing protein structures. In this work, we study the response of single protein molecular conformation under the stress of external mechanical force applied by atomic force microscope tip. We demonstrate that general molecular level force (~15-20pN) can directly induce protein conformational collapse.

4.2 Experiment Section

4.2.1 Materials and sample preparation

In this experiment, Calmodulin molecules were tethered onto a cover glass and then incubated in a homemade chamber. The cover glass was first cleaned and silanized with a mixture of (3-aminopropyl) trimethoxysilane, isobutyltrimethoxysilane with a ratio of

1:10000 dissolved in DMSO (10% v/V) for 4 hours, and then incubated in 10nM dimethyl suberimidate in 50 mM PBS (pH 8.0) for 1h. After washing by water and methanol, the glass slide was incubated in 10nM protein (either dyed labeled or label free) solution in a

PBS buffer (pH 7.4) for 2h. To make sure of studying a specific single-molecule CaM molecule at a time, the CaM molecules were tethered to the glass surface with diluted

70

distribution by controlling the mixture of (3-aminopropyl) trimethoxysilane, isobutyltrimethoxysilane with ratio of 1:10000 in the first step of the sample preparation.

The possibility of two protein molecules locating underneath AFM tip or in the same laser focus spot of around 300 nm diameters is almost zero. During the experiment, we used the

HEPES buffer (pH 7.4) containing 0.1mM CaCl2, the same buffer as most of the experiments in studying single-molecule CaM dynamics. The 0.1mM CaCl2 was added to generate saturated Ca2+ environment allowing the single-molecule CaM to be fully activated by calcium ions in binding with four Calcium ions for each CaM molecule.

In this experiment, even though we also used Calmodulin as a model of protein we didn’t use C28W in the buffer solution, because we focus on the conformation change of CaM molecule itself. And the labeling is different. We used Cy3 and Cy5 as our FRET pair and label them onto residue 34 and 114 respectively. The labeling positions are on the separated two domains, which can reflect the conformational changes of the CaM molecule more precisely under external force manipulation.

4.2.2 AFM-FRET correlated measurement

In order to collect the force curve and optical signal of one dye-labeled protein molecule spontaneously, we used our home built AFM-FRET microscope to do the correlated measurement. The setup is demonstrated in chapter III. The essential part to utilize our home built set-up is to co-axis the AFM tip and laser focus, by doing so one single protein molecule can be targeted by both instruments16.

We put the AFM scanner on top of our microscope and the position of the AFM can be adjusted by a separated stage. After put the sample on the microscope and assemble the

71

AFM, the first step is to move the AFM tip on top of the laser focus. This can be done by observing the microscope using naked eye, and it will make sure the AFM tip is less than 10 microns away from the center of the laser focus. And then, we send the photo counting signal to the AFM controller through a gated photon counter, SR400 (Stanford instrument).

The image of the optical intensity was taken during an AFM tip scanning

(10*10micrometers). A bright spot would appear, because the AFM tip can be considered as a micro mirror and reflect more photon back through the objective. This bright spot essentially indicates the relative position of AFM tip and laser focus. Through this method we can align the AFM tip and the center of the laser focus as close as less than 100nm.

4.2.3 MD simulation

We used MD simulation as an attempt to understand the possible protein conformational change and address the potential energy landscape during the squeezing process. To reproduce our experiment process in MD simulation, we have to implement a pushing speed much faster than the experiment condition.

We used Gromos 96 force field under Gromacs( 4.6 version) environment to accomplish our simulation. Firstly, we used a united CH4 atom slab to represent the surface interacting with protein molecule. The slab is made of 48X48X10 united CH4 atom. Due to the repeatability of simulation box, the width and depth of this surface is designed to minimize the undesired interaction across the boxes. This is a considerable efficient way to simulation hard physical surface. The reason behind is that we could not only avoid artificial boundary generated by using well option instead, but also stand on the fact that in our experiment we had a glass surface covered mainly by isobutyltrimethoxysilane. Our

72

“hand-made” CH4 slab is chemically similar to the surface we used to support our protein molecule in real experiment. We applied position constraint upon the slab to keep it at a constant shape as the real surface does. Secondly, we put our CaM molecule on top of the slab, and filled the simulation box with water molecules and necessary ions. We allowed the simulation box to deform along z-axis (vertical to the surface) at the speed of 0.02pm/ps.

The defined deforming basically generated an even stress force along z-axis to simulate the pushing down of AFM tip. And then, we divided the whole process into multiple steps letting the double surface approaching each other through a total of 160 steps and 0.01nm each. Because the approaching speed is 10^4 folds faster than the experiment reality, we allowed the system to go through a 1ns relaxation and energy minimization after each step of approaching. By doing so, the stress between protein and water molecules, unfavorable orientation of protein molecule due to fast deforming, and any other undesirable interaction could be released. We also used anisotropy pressure coupling to keep the simulation box at constant pressure and almost constant total volume in order to let all the water molecules to be settled reasonably and without stress inside the deforming simulation box. After reassembling all the deforming trajectories, we ended up with a whole trajectory of the protein collapsing process having a length of 90ns total.

We also try to start the simulation from different orientation. Later on realized that even the first one third of the trajectory looks different in this case, when the protein really start to be squeezed heavily, the initial orientation of the protein doesn’t matter too much in terms of potential energy change.

73

Fig. 4.1 Structure of Calmodullin and the structure of simulation box. (A) The structure of Calmodulin molecule and the dye labeling position: donor (green); acceptor (red). (B) The simulation box used for the MD simulation part. Water molecules are not drawn. 4.3 Results and Discussion

Combined atomic force microscopy and single-molecule fluorescence resonant energy transfer spectroscopic imaging (AFM-FRET) has been-demonstrated to be a powerful approach on manipulating and probing protein structure19-25. Using the sub-nanoscale spatial resolution and force manipulation capability of AFM, we are able to apply specific picoNewton mechanical force onto targeted individual protein molecules; simultaneously, we probe the conformational responses of the single-molecule protein by both AFM force spectroscopy and single-molecule FRET spectroscopy20,26-28 (Fig. 4.2 A and B).

Unlike most study of single molecular force spectroscopy, in which the force curve is demonstrated by a plot of force versus z-piezo displacement, we demonstrated the same force curve in a different way as force versus experiment time (Fig. 4.2C). This is because that we focused on the correlated response from FRET spectroscopy and force spectroscopy. In

Fig. 4.2D, the correlation of a time-stamped force curve with FRET trajectory is shown. As it is seen in the figure, while the AFM tip is moving down, the photon counts in both donor and acceptor channels rise due to stronger refection from the metal coated tip when the tip is

74

closer to the molecule, and vice versa. Both force trajectory and FRET trajectory share the same experimental time. The trajectories often contain a numbers of continuous approaching attempts.

We focused on the pushing curve instead of the pulling curve. During the pulling process, protein molecule’s conformation is expected to be gradually stretched out and eventually ruptured (if enough boundary between protein and tip is applied such as covalent linking).

However, during the pushing region, randomness such as the curvature of the AFM apex, orientation of the molecule on surface, interaction between molecule and surface could lead to more complex and interesting process. We report an observation of a repeating process during the pushing region using our correlated measurement technique (Fig. 4.3). Within those force curves, stress force up to a level of about 15pN is gradually built up on AFM tip and eventually released. We suspect that this observation indicates a protein conformation sudden collapse induced by AFM tip. The picture of undergoing conformation change is supported by the FRET trajectory, and a possible thermal dynamic scenario is demonstrated by MD simulation.

AFM force spectroscopy is capable of revealing how much stress force is built up on the cantilever before the force suddenly drops. However, AFM force spectroscopy alone can’t provide direct information about the conformational change of the protein during the stress-force loading process. Nevertheless, our previously reported AFM-FRET nano-scopic approach has demonstrated the capability of performing simultaneous force spectroscopy and correlated FRET measurements4,6,29. With the correlated single-molecule

AFM-FRET measurements, single-molecule FRET spectroscopy is capable of providing

75

conformational information for identifying the protein structure collapse events, which is complementary with the AFM force spectroscopy as the AFM tip is approaching. Fig. 4.3 shows a typical result of the correlated single-molecule AFM-FRET measurements obtained as the tip approaching at the velocity of 0.1nm/ms, and the total recording time of about 1.6 seconds. The force curve (Fig. 4.3C) is plotted with the same chronic time as the recorded

FRET trajectory (Fig. 4.3A and B). The area marked by black box shows the region where the AFM tip apex interacts with protein molecule before the force curve is dominated by the repulsion between AFM tip and glass surface. The area marked by red box shows the FRET efficiency of the system decreased at about the same time when the force dropped indicating the moment the CaM molecule collapsed. The FRET efficiency remained at a low level after collapse occurred for hundreds of milliseconds and then recover. The AFM tip touched a protein molecule at 1.6+/-0.2nm away from the surface depends on the orientation of CaM on surface, and the force kept increasing as the tip kept approaching the surface until reaching a threshold, where the force applied on the protein molecule was about 15 pN at the approaching velocity of 0.1 nm/ms. Then the force was released abruptly (Fig. 4.3D). The feature shown in the force curve was repeatable for a significant amount of trials. We did observe the same behavior from more than 25% of the molecules tested.

The sudden decrease of FRET efficiency reflects the distance between two labeling position on CaM suddenly increased or the two domain of the CaM molecule suddenly fallen apart. The fact that the sudden conformational change observed by FRET coincided with the sudden release of the stress force built up on the AFM cantilever leads us to believe that the stress force triggered a spontaneous conformational change of the CaM molecule.

76

Fig. 4.2 Correlated measurement setup and results (A) AFM-confocal microscopy correlated experimental set up. Two real time trajectories are collected separated from AFM and optical channel; (B) AFM tip apex interacting with CaM molecule whiling approaching; (C) Comparison of the same force curve represented in two different ways: experiment time (left), and Z-piezo displacement (right) (D) Correlated measurement trajectories sharing the same time axis. The red boxes indicated the focus of our study: the interaction between AFM tip apex and protein molecule during the AFM pushing process.

77

Fig. 4.3 Correlated real time trajectories of the AFM-FRET correlated measurement. (A) Optical signal containing both donor(green) and acceptor(red) channels; (B) Calculated FRET efficiency trajectory from (A); (C) real time force curve of a pushing and pulling cycle; (D) Zoom in portion of the force curve indicating the sudden collapse; (A) (B), and(C) share the same time axis.

78

The stress force induced sudden collapse of CaM itself is not surprising. Given the protein molecule had decreasing available space to occupy, a forced conformational change is warranted. However, the fact that the observed process appears to be a spontaneous process is interesting. And the thermally dynamics picture of the underlying process becomes worth studying.

The triggering force magnitude of the force-induced protein structure sudden collapse is

related to the tip approaching velocity, because the process is a non-equilibrium process.

To evaluate the relation between external force and tip approaching velocity, we repeated the

experiments under different approaching velocity. Experiments at 5 different velocities were

performed, about 25% of the force curves gave the reproducible feature during the

approaching phase based on about 1000 sets of data. Fig. 3B-3D show the statistic

distribution of the force at 3 different apparent loading rates (approaching velocity by force

constant of the cantilever). Fig. 4.4A, a plot of the force versus loading rate, shows the

linear relation between force and apparent loading rate (ALR).

The force required for single protein CaM collapse increase from 15pN to 35pN as the approaching velocity increase from 0.05nm/ms to 1nm/ms, and the force increases linearly along with log (ALR). This dependence of the triggering force upon the loading rate indicates that the protein structure collapsing events and the stress force loading process have the similar characteristics with the two-state conformational transition under an external force18. Our results suggest that under the stress force applied by AFM tip, a single-molecule CaM undergoes a transition from its natural folded state to a predictable

“collapse state”, which in turn, corresponds to the release of the stress force built up and the

79

observation of sudden drop of the measured stress force loaded on the cantilever.

Fig. 4.4 The relation between approaching speed and collapse induced force: (A) the linear relation between the forces that cause the protein collapse and the tip approaching speed.; (B,C,D) statistic results of the induced forces applied on protein at different tip approaching speed and apparent loading rate.

80

MD simulation of the collapse process and examine of the potential energy curve during the deformation. Based on the correlated experiment results, we understand that the primary cause of the collapse event is the stress force applied to an individual protein molecule by an

AFM tip; when the stress force reaches to a threshold; the structure of the protein molecule spontaneously collapses. From the previous session, we also have evidence to show that this process has similar characteristics of the two-state conformational transition under external force, in other word similar energy landscape. It is intriguing that the protein collapse proceeds as an energetically downhill process, and it is the AFM applied stress force to pushes the protein to the turning point of energetic and structural transition state. To further understand the dynamics and energetic of the sudden structural collapse of proteins, we have applied the MD simulation to identify the potential energy landscape involved in the observed protein structure sudden collapse. We use Gromos 96 force field under Gromacs environment for our MD simulation. We designed a simulation box containing a united

CH4 atom slab with a single CaM molecule on top of it, and filled the system with water molecules.

Firstly, we used a united CH4 atom slab to represent the surface interacting with protein molecule. This is a considerable efficient way to simulation hard physical surface. The reason behind is that we could not only avoid artificial generated by using well option instead, but also stand on the fact that in our experiment we had a glass surface covered mainly by isobutyltrimethoxysilane. Our “hand-made” CH4 slab is chemically similar to the surface we used to support our protein molecule in real experiment. We applied position constraint upon the slab to keep it at a constant shape as the real surface does. Secondly, we put our

81

CaM molecule on top of the slab and filled the simulation box with water molecules. We allowed the simulation box to deform along z-axis (vertical to the surface) at the speed of

0.02pm/ps. We also used anisotropy pressure coupling to keep the simulation box at constant pressure and almost constant total volume in order to let all the water molecules to be settled reasonably and stress-less inside the deforming simulation box. By implementing a deforming under a constant speed, we generate a stress force in between two surfaces to simulate the AFM tip pressuring a targeted protein molecule. Because the approaching speed is 104 times faster than that in the real experiment, we divide the whole process into multiple steps and let the simulated double surface approaching each other through a total of

160 steps, 0.01nm each30. We allow the system to have a 1ns relaxation and energy minimization after each step of approaching, giving a relaxation time for the stress force between the protein and water molecules, the unfavorable orientation of protein molecule due to the fast deforming, and the other undesirable interactions. Because the experiment time scale is much longer than the time scale of the MD simulation, the go-and-relax procedure allows the simulation result involves artificial impacts as less as possible. After reassembling all the deforming trajectories, we obtain a whole MD simulation trajectory of

90ns total time.

We evaluated the MD simulation result mainly through two aspects. First, we analyzed the potential energy change during the deformation of the simulation box that reflects the squeezing process in real experiment. Because of the limitation when building a simulation box, not all the potential terms is relevant to the real experiment. We focus on the following three terms that could represent three essential interactions during this process: protein to

82

protein, that represents the chain conformation energy; protein to water, that represents the solvation energy; protein to surface, that represents the stability contribution from hydrophobic surface interacting with a collapsing protein molecule. We extract these three terms and plotted them separately in Fig. 4.5 A, and we sum over them in Fig 4.5B. The result is very interesting.

83

Fig. 4.5 Potential energy changes during the squeezing perform by MD simulation. (A) Potential energy changes contributed by different interaction during the processes: protein chain conformation energy (purple), solvation energy (green), hydrophobic energy (blue); (B) The combined effect of the above three. A significant downhill energetic curve is found in the middle of the production simulation (b-c), which is the reflection of the combined effect in the black dashed box in (A).

84

We found that during this intense process of squeezing, the protein’s chain conformation energy change was quite insignificant (Fig. 4.5A Purple) compared to its interaction with surrounding environment: water and hydrophobic surface underneath. While the protein gradually collapsed due to the pressure from decreasing space, more and more hydrophobic part of it started to unfold and become exploded to the outside. This leaded to a diminishing stability effect from solvation (Fig. 4.5A Green). At the same time, the increasing exploded surface, which is hydrophobic, started to have increasingly interaction with the hydrophobic surface holding the protein molecule (Fig. 4.5A Blue). We found that the hydrophobic interaction strengthen uniformly during the process. In contrast, the solvation energy lost seemed to stall in the middle of the process. Since the chain conformation energy stayed neutral, the solvation and hydrophobic interaction were the two main competing factors. The combined effect of this competition can be seen in Fig. 4.5B, where there is a very significant downhill in the middle of the range. The sum-over curve of the three energy term can be viewed as three phases: in the first phase(a-b), the shortage of space and increasing stress force from the approaching surface heavily destabilized the protein molecule system and the potential energy rises; in the second phase(b-c), the marginal gain from hydrophobic stability effect exceeded the marginal loss of solvation energy, and the system was net stabilized; in the final phase(c-d), the competition reversed and the system started to be net destabilized from point c. Towards the very end of the production simulation, the system ran out of space. The limitation of simulation box became overwhelming in a way that the protein chains even partially stick into the hydrophobic surface at the very end of the simulation. At this stage, the system becomes totally distorted, and the MD simulation we performed could

85

not evaluate the process anymore. The competing mechanism between decreasing stability due to solvation energy loss and increasing stability due to hydrophobic effect gain, hints the possibility that the collapsing event due to a combination of stress force and diminishing space could be a spontaneous process in the real experiment.

After finding the important role that the hydrophobic surface played in the collapsing process, we became interested in how it affected the recovery mechanism, which could potentially support our finding from another aspect. We did another MD simulation test to try to let the collapsed molecule recover itself by suddenly decompress the simulation box.

Since the production simulation was divided into multiple steps, at the end of each step the state of the system is basically a snapshot of different stages of the underlying continuous process. We used these snapshots as the starting point of the test. To do the test, we resized the simulation box, filled necessary amount of water molecules, and let the system relax. After 10ns of relaxation from each snapshot, changes of RMSD were calculated to evaluate the protein molecule’s tendency and ability to recover towards its folded native state.

We performed the above steps under two conditions: with and without the presents of hydrophobic surface (Fig. 4.6). The results were consistence with the conclusion we make above.

In Fig. 4.6A, without removing the surface, the protein molecule becomes resilient to the recovery in the middle of the process. In contrast, if we remove the surface and let the molecule recover surrounding by water molecules only. The willingness of recovery uniformly increases, which means that the harder we compress the molecule, the more uncomfortable the molecule becomes, and the more heavily it recovers itself after all the

86

restriction suddenly being withdrawn.

Fig. 4.6 Evaluate the recovery of collapsed CaM by RMSD: (A) the RMSD change indicating recovery potential of each step of the simulation if the simulation box suddenly regained its size and the hydrophobic surface stayed; (B) the RMSD change indicating recovery potential of each step of the simulation if the simulation box suddenly regained its size and the hydrophobic surface stayed. Figure 4.7 is another summary of our MD simulation in the view of protein conformation.

Along the experiment time, we see the secondary structure of protein molecule gradually lost while the stress force is built up, and eventually the protein turns into a collapse state that the entire a-helix chain connecting two domains and most of the four EF-hands are broken. The structures in the figure are colored based on the secondary structure; we see less and less purple color (-helix) down the process.

87

Fig. 4.7 A typical feature segment of force curve contains a collapse event. The protein’s deformation process is showed along the curve. All the conformations come from the MD simulation result discussed later. Secondary structure based coloring method are used to demonstrate the disappearing of helix chains (purple). Due to the limitation of time scale, our MD simulation itself cannot reproduce the dynamic information underlying; Nevertheless, our simulation still enables us to draw a conclusion that the protein structural collapse event is strongly related to the short lived energy favorable state, which is the comprehensive result combining three different interactions. Our simulation result provides a semi-quantitatively description for the spontaneity of the protein collapse events.

4.4 Conclusion

In this work, we demonstrate a new way to perturb single protein molecule, besides the direct pulling and stretching, applying mechanical stress force on top of single protein molecule can also have enough perturbation to artificially denature the protein. The force that collapsed the protein increase while increasing the approaching velocity, it is in the range of 15-35 pN and the required force for protein collapse and logarithmic of the force loading rate follows single linear relation. The collapse of protein molecule under external stress

88

force is an abrupt processes, the time for the protein collapse is under our experimental time resolution of 10ms. We claimed that under the stress of AFM tip, Calmodulin molecule undergoes a transition from its natural state to a predictable “collapse state” at the moment the force suddenly dropped. The collapse state may be an intermediate state between folded and unfolded protein, and about 25% of the protein collapse in the tip approach. So we have to mention the fact that in single molecule force spectroscopy measurement, the collapsed state of protein molecule may appear before the force pulling, therefore, it is possible that a collapsed intermediate state is the initial state for the AFM force pulling, which may be very important for the study of the mechanism and dynamics of protein folding-unfolding by AFM force spectroscopy.

89

4.5.References

1. Whitford, P. C. Disorder guides protein function. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 7114-7115.

2. Whitford, P. C.; Sanbonmatsu, K. Y.; Onuchic, J. N. Biomolecular dynamics: order–disorder transitions and energy landscapes. Reports on Progress in Physics 2012, 75, 076601.

3. Wang, J.; Oliveira, R. J.; Chu, X.; Whitford, P. C.; Chahine, J.; Han, W.; Wang, E.; Onuchic, J. N.; Leite, V. B. Topography of funneled landscapes determines the thermodynamics and kinetics of protein folding. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 15763-15768.

4. Gumpp, H.; Puchner, E. M.; Zimmermann, J. L.; Gerland, U.; Gaub, H. E.; Blank, K. Triggering enzymatic activity with force. Nano letters 2009, 9, 3290-3295.

5. Kuo, T. L.; Garcia-Manyes, S.; Li, J.; Barel, I.; Lu, H.; Berne, B. J.; Urbakh, M.; Klafter, J.; Fernandez, J. M. Probing static disorder in Arrhenius kinetics by single-molecule force spectroscopy. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 11336-11340.

6. Mo, Y.; Bao, P.; Gao, J. Energy decomposition analysis based on a block-localized wavefunction and multistate density functional theory. Physical Chemistry Chemical Physics 2011, 13, 6760-6775.

7. Stirnemann, G.; Kang, S. G.; Zhou, R.; Berne, B. J. How force unfolding differs from chemical denaturation. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 3413-3418.

8. Zhou, R.; Berne, B. J.; Germain, R. The free energy landscape for beta hairpin folding in explicit water. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 14931-14936.

9. Kishino, A.; Yanagida, T. Force measurements by micromanipulation of a single actin filament by glass needles. 1988.

10. Yang, S.; Cao, J. Direct measurements of memory effects in single-molecule kinetics. J. Chem. Phys. 2002, 117, 10996-11009.

11. Hinterdorfer, P.; Baumgartner, W.; Gruber, H. J.; Schilcher, K.; Schindler, H. Detection and localization of individual antibody- recognition events by atomic force microscopy. Proc. Natl. Acad. Sci. U. S. A. 1996, 93, 3477-3481.

12. Schwesinger, F.; Ros, R.; Strunz, T.; Anselmetti, D.; Guntherodt, H. J.; Honegger, A.; Jermutus, L.; Tiefenauer, L.; Pluckthun, A. Unbinding forces of single antibody-antigen complexes correlate with their thermal dissociation rates. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 9972-9977.

90

13. Fazal, F. M.; Block, S. M. Optical tweezers study life under tension. Nature photonics 2011, 5, 318-321.

14. Svoboda, K.; Mitra, P. P.; Block, S. M. Fluctuation analysis of motor protein movement and single enzyme kinetics. Proc. Natl. Acad. Sci. U. S. A. 1994, 91, 11782-11786.

15. Sen Mojumdar, S.; Chowdhury, R.; Chattoraj, S.; Bhattacharyya, K. Role of Ionic Liquid on the Conformational Dynamics in the Native, Molten Globule, and Unfolded States of Cytochrome C: A Fluorescence Correlation Spectroscopy Study. The Journal of Physical Chemistry B 2012, 116, 12189-12198.

16. Borghi, N.; Sorokina, M.; Shcherbakova, O. G.; Weis, W. I.; Pruitt, B. L.; Nelson, W. J.; Dunn, A. R. E-cadherin is under constitutive actomyosin-generated tension that is increased at cell-cell contacts upon externally applied stretch. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 12568-12573.

17. - Lele, T. P.; - Pendse, J.; - Kumar, S.; - Salanga, M.; - Karavitis, J.; - Ingber, D. E. - Mechanical forces alter zyxin unbinding kinetics within focal adhesions of living cells. - Journal of Cellular Physiology , - 187.

18. Lele, T. P.; Thodeti, C. K.; Ingber, D. E. Force meets chemistry: Analysis of mechanochemical conversion in focal adhesions using fluorescence recovery after photobleaching. J. Cell. Biochem. 2006, 97, 1175-1183.

19. Leckband, D.; Israelachvili, J. Intermolecular forces in biology. Q. Rev. Biophys. 2001, 34, 105-267.

20. Kohn, J. E.; Millett, I. S.; Jacob, J.; Zagrovic, B.; Dillon, T. M.; Cingel, N.; Dothager, R. S.; Seifert, S.; Thiyagarajan, P.; Sosnick, T. R.; Hasan, M. Z.; Pande, V. S.; Ruczinski, I.; Doniach, S.; Plaxco, K. W. Random-coil behavior and the dimensions of chemically unfolded proteins. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 12491-12496.

21. Mayor, U.; Guydosh, N. R.; Johnson, C. M.; Grossmann, J. G.; Sato, S.; Jas, G. S.; Freund, S. M.; Alonso, D. O.; Daggett, V.; Fersht, A. R. The complete folding pathway of a protein from nanoseconds to microseconds. Nature 2003, 421, 863-867.

22. Berkovich, R.; Garcia-Manyes, S.; Urbakh, M.; Klafter, J.; Fernandez, J. M. Collapse dynamics of single proteins extended by force. Biophys. J. 2010, 98, 2692-2701.

23. Antikainen, N. M.; Smiley, R. D.; Benkovic, S. J.; Hammes, G. G. Conformation coupled enzyme catalysis: single-molecule and transient kinetics investigation of dihydrofolate reductase. Biochemistry 2005, 44, 16835-16843.

91

24. Pisliakov, A. V.; Cao, J.; Kamerlin, S. C.; Warshel, A. Enzyme millisecond conformational dynamics do not catalyze the chemical step. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 17359-17364.

25. Xie, X. S. Biochemistry. Enzyme kinetics, past and present. Science 2013, 342, 1457-1459.

26. Haran, G. How, when and why proteins collapse: the relation to folding. Curr. Opin. Struct. Biol. 2012, 22, 14-20.

27. He, Y.; Lu, M.; Cao, J.; Lu, H. P. Manipulating protein conformations by single-molecule AFM-FRET nanoscopy. ACS Nano 2012, 6, 1221-1229.

28. Zhao, J.; Davis, J. J.; Sansom, M. S.; Hung, A. Exploring the electronic and mechanical properties of protein using conducting atomic force microscopy. J. Am. Chem. Soc. 2004, 126, 5601-5609.

29. Kuo, T. L.; Garcia-Manyes, S.; Li, J.; Barel, I.; Lu, H.; Berne, B. J.; Urbakh, M.; Klafter, J.; Fernandez, J. M. Probing static disorder in Arrhenius kinetics by single-molecule force spectroscopy. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 11336-11340.

30. Roos, W. H.; Gibbons, M. M.; Arkhipov, A.; Uetrecht, C.; Watts, N. R.; Wingfield, P. T.; Steven, A. C.; Heck, A. J.; Schulten, K.; Klug, W. S.; Wuite, G. J. Squeezing protein shells: how continuum elastic models, molecular dynamics simulations, and experiments coalesce at the nanoscale. Biophys. J. 2010, 99, 1175-1181.

92

CHAPTER V. PROBING THE RESPONSE OF EGFR EXTRACELLULAR

DOMAIN DIMERS TO EXTERNAL STRESS FORCE USING

SINGLE-MOLECULE CORRELATED MEASUREMENT AND EVALUATING

THE POSSIBE REARRANGEMENT OF A COLLAPSED STRUCTURE

5.1 Introduction

Epidermal growth factor and its receptor were discovered by Stanley Cohen, who went on and won the 1986 Nobel Prize in Medicine with Rita Levi-Montalcini for their discovery of growth factors.

Epidermal growth factor is a growth factor that stimulates the processes of cell growth, proliferation, and differentiation. It serves as a signaling protein by binding its extracellular-receptor that exists on the cell surface. These receptors are a family of receptors that is made of EGFR (ErbB1), HER2/neu(ErbB2), HER-3(ErbB3), and

HER4(ErbB-4). The detailed mechanism of the signaling process is not fully understood, but researchers over years have been able to draw a picture of the essential pathway. It is believed that they become activated by forming homodimers or heterdimers with other family members. Even though there is some evidence that inactive dimers may also exist without ligand binding, the activation is initiated by binding their growth factor ligands (mainly EGF). There is also evidence to suggest clustering is important for activation, as these on membrane protein can freely diffuse on the membrane and distribute non-uniformly on the membrane1-7.

EGFR and ErbB family receptor are biologically important. The lost of their proper signaling is associated with the development of neurodegenerative disease,

93

such as Alzheimer’s disease, and over-expression of them is associated with the development of many types of cancers in solid tumors such as lung cancers and anal cancers.

The biochemistry property of EGFR is also very interesting to scientists. It is unique in many aspects. Its structure can be break down into three components: an extracellular domain that is responsible for binding with message protein like EGF, an intracellular domain that is responsible for the catalysis for the downstream signaling process, and a cross-membrane domain that penetrates through the cell membrane and connecting the two domains above. The simplified structure is demonstrated in Fig.

5.1A. The dimerization mechanism is also very surprising. For other similar receptors, a small molecule often binds in the middle of the complex and serves as glue helping two or more bigger molecules to form complex. Instead, EGF binds on either side of the receptor and plays a role to tune the receptor into a proper conformation for dimerization8-11. The tuning happens in a asymmetry head-to-tail fashion, which could be seen in the side view of the complex in Fig. 5.1C.

We found EGFR as a good candidate to study on and try to answer the question we left to be answered in the previous experiment. Even though we had a good idea of the driving force of the protein molecule collapse and the possible contributor of the satiability of the collapsed state, we do not have much information about how the collapsed state is recovered. Whether or not the nature state is the ultimate and only direction of the recovery is yet to be known. We need a candidate molecule that has more structure complexity than calmodulin molecule and some biological

94

relevance related to the complexity of the structure would be bonus. EGFR becomes a good molecule to study on in these manners. 12-14

Fig. 5.1 Schematic representation of the structure of EGFR: (A) the conceptual demonstration of EGF initiated EGFR dimerization process, and the on membrane structure of EGFR; (B) the complex of extracellular domain of EGFR and EGF (Green) in Monomer form; (C) the complex in dimer form 5.2 Experiment Section

5.2.1 Materials and sample preparation

Since whole size EGFR protein in scale is hard to obtain and purified.

Researchers tend to use two separate approaches to study its property: either using living cell sample to study it in vivo or using isolated domain to do in situ measurement. For the latter one, extracellular domain is used to study the topic related to dimerization and intracellular domain is used if one wants to study the catalysis behavior of EGFR.

Our target is to study the behavior of EGFR dimer complex under stress force applied by AFM tip. We used purified extracellular domain of EGFR as our sample and in

95

order to obtain optical signal, tetramethyl-rhodamin labeled EGF is used to indirectly detect the conformational information of the complex. In order to avoid the complexity of sample preparation of asymmetry labeled EGFR dimer complex, we measured the self-quenching effect to detect the distance information between two identical dyes that are on either side of the complex, instead of using pair of FRET dye. Another reason not to use FRET is because that the extracellular domain has

624 amino acid residues, and the sizes of complex makes the distance between two

EGF molecules exceed 7nm, which is close the upper level of typical FRET distance.

Work has been done by Yasushi et al. in attempt to measure the FRET efficiency between Cy3/Cy5 labeled on each EGF in vivo. Their result showed that the efficiency is close to zero, which proves that the size of the molecule indeed makes typical single molecular FRET experiment design very difficult.

In order to implement single molecular optical detection, we immobilized our sample onto glass surface with the help of an asymmetry bi-functional linker and

Ni-NTA-His-tag chelate mechanism that is described in Chapter II. Dimers of soluble receptor extracellular domain were prepared using a method described in previous work studying the ensemble kinetics of the dimerizatin. Briefly, 10)Mof pure EGFR extracellular domain (Sino tech) were incubated with 20)M mouse

EGF(Invitrogen) in 20mM HEPES(pH 7.4( and 150mM NaCl for 1h at room temperature. 0.25mM of DSS(disuccinimidyl suberate) was added to covalently lock-up the dime r complex. The cross-linking reaction was then terminated by adding

10mM of Tris0H Cl buffer. A simplified size based chromatography (Superose 12

96

column) was used to separate the dimer complex and existing non-rea cted monomer.

The purchased pure extracellular domain protein was attached with his-tag residues on C-terminal. This is the exact position connecting the extracellular domain and cross-membrane domain of EGFR in its native form in vivo (see Fig. 5.1C). Even though each the soluble dimers has two his-tag on it in principle. The density of active site on modified glass surface gauges the immobilization reaction. The possibility of two his-tag on one dimer complex both being immobilized is close to zero. Therefore the conceptual conformation of dimer complex immobilized on glass surface in measurement condition is demonstrated in Fig. 5.2A, B. The optical measurement is performed under solution phase with 20mM HEPES buffer (100mM

NaCl, pH=7.4).

5.2.2 AFM-Photo-stamping correlated measurement

We again used correlated measurement to study the behavior of single protein molecule under force manipulation. In this experiment, we replaced the confocal smFRET microscopy by single molecule photo-stamping microscopy aiming the fluctuation of chromophore lifetime. In this particular case, the lifetime fluctuation is due to self-quenching effect between two dyes labeled on the EGF molecules that bind on either side of the EGFR dimer complex. The detected lifetime of the fluorescence signal will drift higher when the self-quenching effect becomes stronger and vice versa. All other parameters being equal, the strength of self-quenching is related to the distance between two dyes, which can be indirectly link to the conformation of the dimer complex.

97

The experiment is similar to the correlated measurement setup described in Chapter

II, and IV. The optical part is replaced by a single channel photo-stamping microscopy. The output of a mod-locked femtosecond Ti:sapphire laser

(CoherentMira 900D,1.6W,2 200fs fwhm, 76MHz) was used to pump an optical parametric oscillator(APE-OPO,Coherent. Inc.), and the output laser at 1064nm was then frequency-doubled to 532nm by an LBO nonlinear optical crystal as the excitations source. The beam then passed through a pair of prisms to eliminate the fundamental IR light. The confocal beam with a power of about 0.2-2.0 )W was finally delivered into a confocal microscope with an avalanche photodiode detector

(APD). The fluorescence signal was recorded through a PicoHarp 300(PicoQuant) time correlated single photo counting (TCSPC) module. The arrival time and the delay time between the laser excitation pulse and the fluorescence photon emission of each photon are recorded. Both intensity time trajectory and photon-stamping trajectory can be generated. The photo-stamping trajectory was later converted to a

10ms bin lifetime trajectory to reflect the lifetime fluctuation of the fluorescence signal (Fig. 5.2C). And the AFM part of the correlated setup, which is the same as the setup in ChapterIII, is response for the collection of AFM force curve (Fig.5.2D).

98

Fig. 5.2 Schematic representation of the AFM-photo-stamping correlated measurement setup. (A) Conceptual demonstration of the single EGFR extracellular domain molecule (red) sitting in between AFM tip and objective; (B) the photon by photon demonstration of the lifetime trajectory; (C) the calculated lifetime trajectory based on 10ms binning. 5.3 Results and Discussion

The data ends up with a series of combination of AFM force curve and optical trajectory (in the form of either intensity or lifetime). Each though two sources of signals are collected separated; they can be synchronized with the help of intensity fitting, due to mico-mirror effect (described in Chapter II, IV). Because of the intrinsic randomness of AFM force manipulation in the form of stress force, the response of dimer complex is observed in two distinguishable catalogs. Some are without any response in the pushing part of the force curves; the rest (~25%) have feature collapsing patterns that are very similar to the ones observed in the previous experiment on calmodulin. Not surprisingly, the patterns in this experiment are longer than the ones in calmodulin, likely due to a larger size of dimer complex (642 residues) than calmodulin (148 residues). However, the force height is not significantly higher, likely due to the nature of the interaction within the dimer

99

complex. (Fig. 5.2D)

Statistical analysis was implemented. Even though longer trajectories were collected from both sources, only 7 seconds long segment of each trajectory is necessary to be evaluated. These are made of 1 second before the AFM tip starting to approach, 3 seconds that was used to implement two continuous AFM manipulation attempts, and 3 seconds after the AFM tip completely detached from the molecule. By doing so, a complete picture of the dimer complex’s response to the AFM engagement can be drawn. Fig. 5.3B serves as a guideline to help to demonstrate the motion of the

AFM tip during this 7 seconds long segment. The time period when AFM engagement occurred is shaded by blue color. In order to increase the statistical significance of the lifetime trajectory, we used a 2-D color table to visualize the fluctuation and distribution of the observed lifetime around the time of AFM manipulation. The color table is made in a way that the first vertical slide of the color table in Fig. 5.3A is a histogram reflecting the distribution of lifetime of all the trajectories that have force response in it from synchronized time 0 to 100ms. And the second vertical slide is corresponding to 50 to 150ms, and so on. This continuously generated color table is plotted throughout 7 seconds, and the overview of the lifetime fluctuation is well summarized by one graph.

From these color tables, we found that the lifetime of fluorescence signal in B is materially shifted away from the mean, while there is no such behavior being observed in A. This shifting occurs coincide with the implementation of AFM tip manipulation. And this result built a strong correlation between the lifetime shifting

100

and potential collapsing of dimer complex that is indicated by the collapsing pattern, which we used to characterize the trajectory in the first place.

Besides the color tables, we demonstrated the shifting in another way in Fig. 5.4.

For the trajectories with force response, we extracted the histogram of lifetime distribution with a time increment of 0.5 seconds, fitted the distribution according to a single Gaussian curve, and plotted the resulting mean (black) and width (red) of the fitting curves. Dramatic downward shifting of lifetime and width widening of the distribution are reflected in this figure. These figures let us draw the conclusion that significant downward shifting of lifetime occurs during the time of AFM engagement.

A state of dimer complex that has a shorter distance between two EGF molecules is populated during the manipulation and the formation of a potential collapsed state of dimer complex.

101

Fig. 5.3 Synchronized experiment result. (A) The 2-D color tables represent the lifetime fluctuation in the correlated experiment with collapsing pattern to be seen. (B) The guideline indicating the position of the AFM tip versus the experiment time. (C) One set of force curves observed during the correlated measurement and its zoom-in of the part when the protein collapse accuses

102

Fig. 5.4 The graphs show the fitted lifetime mean(black) and width(red) of lifetime histogram throughout different time period of experiment. The fitting is based on single Gaussian distribution. Based on the conclusion above, we were also interested in the details such as: the magnitude of the lifetime shifting and duration of these states. We plotted another distribution in Fig. 5.5 to investigate this information. In Fig. 5.5A, we first analyzed each individual lifetime trajectory that links to collapsing pattern in the force curve, and tried to extract a portion where the lifetime is significantly deviated from the overall mean of each trajectory. We then registered a calculated mean, start time, and end time for each of these portions. And then we plotted two distribution based on these registered properties of individual trajectory. One way is to count the calculated shifted-mean from any given trajectory regardless of its duration as one data point; we get the histogram in filled black bars. Alternatively, we treated each

10ms lifetime reading throughout all trajectories equally, and plotted a distribution of aggregated 10ms lifetime reading from the extracted portions that have significant lifetime shift; we get the histogram in unfilled red bars. Interesting behavior of the

103

lifetime shifting was observed, as we can see that the shifting is not limited to the downside as seen in the previous graphs. The black histogram is telling a story that the shifting could be either upward or downward. In Fig. 5.5B, the direction of the shifting is plotted comparing the mean of extracted portion and overall mean of each individual trajectory. The shifting ranges from -3.5 to +2.5ns. As there is a broad distribution from -3.6 to 0.6, we used the estimated cut-off of -0.6 and +0.6 to evaluate the duration of the shifting in each case: shifting lower(C), unchanged (D), and higher (E). Summarizing these graphs, we found that in the case of shifting lower, the duration of shifted lifetime form a Poisson like distribution with the maximum at about 400-500ms. For the other two scenarios, the number of trajectories is limited. However, in the cases of shifting higher, the average duration is longer than the case of shifting lower. And in Fig. 5.5D, the duration is broadly distributed. This is not surprising. Because, when there is no significant shift from the overall mean, it is impossible to define the duration of that “shifted state”. It is just a random noise related to the method used to define whether there is a significant shift.

104

Fig. 5.5 Statistical facts of the magnitude and duration of the lifetime shift. (A) Trajectory number counted distribution of shifted lifetime (Black) and 10ms bin counted distribution of shifted lifetime (Red); (B) Magnitude of shifting related to the overall mean of each evaluated trajectory; (C) (D), and (E) the distribution of shift duration in the case of shifting lower, unchanged, and higher respectively.

105

Besides evidence from optical results, we also found evidence from analyzing the pulling curve of the protein. Standard AFM pulling experiment is used to analyze the force-extension curves of either EGFR monomer or dimer. Fittings of the WLC function to the force-extension curves that lead to each force peak resolve the elasticity of the underlying protein molecules15,16 (Equation 5.1). In the function,

F(x) and x are the observed force and extension define the curves; L is the fitted contour length; lp is persistence length that is 0.38nm for protein (the average length of amino acid in chain), kB is Boltzmann constant; and T is the absolute temperature.

C S D T k T D 1 1 x T F(x)  B   (5.1) l D C x S 4 L T p D 4D1  T2 T E E L U U

Since we didn’t use specific linkage between AFM tip and EGFR molecule, the ability of picking up a protein molecule is restricted. Compared to the experiment using specific linkage at both end of the stretching points, we observed boarder distribution of the contour length, and the highest observed contour length is limited by the strength of the interaction between AFM tip apex and single protein molecule.

However, the resulting distribution could still review the picture of the pulling process, especially through comparing different scenarios (Fig. 5.6).

Force-extension curves of monomer and dimer EGFR were collected. By checking the pushing part, we could further determined whether or not there was a collapse event occurred. Presumably, collapse event triggered a conformation change, and the following pulling would be applied on a molecule that is different from the native state before collapse. We only fitted major peaks in the collected

106

force-extension curves and calculate contour length of even peak respectively. By comparing the force-extension curves under different scenarios, we found that the monomer has a single major peak force-extension curves and the resulting contour length L0 is centered at around 28.6nm (Fig. 5.6A). When the force-extension was following a collapse event at the pushing part, the resulting contour length remains unchanged at around 28.1nm (Fig. 5.6A). In the case of dimer complex tethered onto glass surface, a secondary peak was seen to follow the major peak. We called it secondary because the rupture force of this peak is smaller than the major peak came first. And these secondary peaks didn’t show at all the force-extension curves. The contour length of the major peak of dimer is centered at 30.3nm (Fig. 5.6A), and the following secondary peak has an average L=L1-L0 of 13.4nm. If the force-extension curves came after a collapse event, the resulting contour length increased sizably to around 36.0nm, and more interestingly the chance of seeing a secondary peak decreases (Fig. 5.6C, D vs. Fig. 5.6E, F). The contour length is a reflection of the number of amino acids that contribute to generate the restoring force, and the separated peaks reflect the mechanical structure of the molecule17,18. The major peak of dimer’s force-extension curves has a slight longer contour length compared to the single peak of monomer’s force-extension curves. These peaks are likely due to the AFM attempting to stretch the EGFR extracellular domain. The secondary peaks that were seen in some of the dimer’s force-extension curves are likely due to an incomplete attempt to stretching another EGFR extracellular domain.

Just like the experiment of pulling a sequence of identical molecules, they tend to be

107

ruptured one by one and give multiple peaks. We didn’t use specific binding in between the AFM apex and molecule, so the AFM could possibly reached the limit of picking up the protein molecule at the end of the secondary peaks. The most interesting fact is that the collapse event seemed to decrease the chance of even seeing these secondary peaks. Furthermore, the distribution of the contour length in Fig.

5.6E is much broader compared to the rest. From our experience, typical contour length distribution should looks more like a Poisson distribution than a Gaussian distribution. It is hard say what distribution Fig. 5.6E is, however the broadness and somewhat double peak looking distribution leaded us to believe that the process got complicated and it is safe to say the major peak of the force-extension peak is likely belong to a complex states but not a simply rupture of the first force bearing EGFR molecule inside the dimer. The underlying states and its complexity are due to the collapse event happened beforehand.

108

Fig. 5.6 Fitted contour length distribution of (A) Monomer EGFR without collapse event; (B) Monomer EGFR with collapse event; (C) Major peak of dimer EGFR without collapse event; (D) Secondary peak of dimer EGFR without collapse event; (E) Major peak of dimer EGFR with collapse event; (F) Secondary peak of dimer EGFR with collapse event.

109

Because the relationship between the direction of the lifetime shifting and distance of two EGF molecules can be drawn, we believed that the experiment result hints a strong possibility that multiple outcome of the conformation change occurs during the

AFM tip induced dimer collapsing. Due to the size and complexity of the dimer complex in study, we expected that the multiple outcomes are likely related to conformation rearrangement after the collapsing of the dimer complex. In order to find support to our speculation and draw the outline of the possible conformation of each outcome, we need to further discuss the conformational information from previous publications.

Firstly of all, it is widely believed that the back-to-back form of dimer(Figure 5.1C) is the fundamental of the EGFR dimerization, even though whether it is ligand binding induced dimerization or predominant non-binding dimers exist in the first place is yet to fully understand. There are two possible orientations of EGFR extracellular domain dimer complex on the membrane. The traditional picture suggests that dimers stand proud of the membrane (Fig. 5.1A). However, more recent work reported the membrane could favor a more stable orientation that the dimers alien almost flat on the membrane. Considering we only have one anchor point at the bottom of the dimer and the hydrophobic nature of the modified glass surface, we suspect that before any manipulation was applied the most possible orientation of the dimer complex in our experiment condition is the latter one, in which the complex laying on top the surface and engage with the surface through a large area that including mainly domain I, III and also part of V. (Fig. 5.7) We also

110

need to know all the known interaction in terms of intra-monomer interaction. In the back-to-back dimer, the interaction between domains II of two monomers is the major force keeps them connected. Studies about the model of EGFR oligomerization also suggest two possible interactions. Clayton et al. proposed an active tetramer model with side by side arrangement of two back-to-back dimers, in which the interaction between domain I and I’, and between III and III’ is the main linkage. Garrett et al. proposed another possible tetramer model that is made of a head-to-head arrangement of two dimers, in which the interaction between domain I and III’, and between I’ and

III is the main linkage. The work was about EGFR dimer active by transforming growth factor , which has a very similar structure and function with EGF. This head-to-head tetramer is later further supported by the work of Kästner et al. using

MD simulation. These interactions were supported by varieties of different aspects including crystal structure, NMR measurement, in vivo FRET measurement, and MD simulation. These interactions are essential for the formation of high order oligamers of EGFR, which is proposed by many experiments and believed to play an important role in the activity of the signaling process carried by EGFR. We believe that in order to form stable oligamer higher than four, intra-monomer interactions most likely are not limited by the two discussed above. This is even truer if the flat orientation of the back-to-back dimer is favorable, as the stable assembly of more than four asymmetry structures in one planer is impossible without interaction along different directions. The ability to form intra-monomer tertiary structural such as

H-bond should involve many different combination of intra-monomer domain

111

interactions. Based on this information and our experiment results, a hypothesis is proposed in Fig. 5.7. State A is driven by I-I’ and III-III’ interaction along the over head direction, this interaction could be a requirement to assemble more than four monomer to form a higher order oligomer. State B is driven by I-I’ and III-III’ interaction side by side, this is similar to the structure discussed in the work of

Clayton et al, just with a different orientation. A and B is corresponding to the lower shifting of lifetime in the experiment results. State C is a fast recovery of the collapsed state, which is corresponding to the unchanged lifetime scenario in the experiment. State D is due to failures of possible rearrangement attempts, this is very similar to the proposed collapsed state in the study of calmodulin, which has a much simpler structure and less potential of rearrangement. Unfortunately, due to the constraints from covalent cross-linking in the process of sample preparation, linkage of domain II-II’ and domain V-V’ is unlikely to be break down. So the interaction in the head-to-head form of tetramer is unlikely to be triggered. Also, the

EGF is most likely linked to the residue Lys336 on domain III (Fig. 5.7A).

112

Fig. 5.7 Hypothesis of possible outcomes of rearrangement of a collapsed EGFR dimer complex on surface: The most possible orientation of dimer complex on surface from two different viewpoints is on the left. After collapsing, an initial collapsed state is formed with loosen tertiary structure of each domain. This intermediate state then rearranges itself towards four most possible outcomes. (A) Interaction between the same domain of each monomer leads to a open-up form laying on the surface; (B) Interaction between the same domain of each monomer and possible dragging force from AFM tip leads to a lift-up form of conformation; (C) Reversal of the collapsing process leads to a fast recovery of the complex; (D) Failures of all the possible rearrangements leads to a prolonged collapsed state. The domain labeling is also showed, while domain IV is just represented by a black stick.

113

5.4 Conclusion

In this experiment, we used AFM-optical correlated measurement to study the response of EGFR extracellular domain under AFM force manipulation. We open the door of AFM correlated measurement excluding FRET. The optical part of the correlated setup can be switch to many different techniques based on the purpose of experiment. Signatures force pattern indicating molecular collapsing in the pushing region of the force curve was observed in this experiment. Because of the complexity of the EGFR dimer complex, multiple recovery pathways after collapsing were proposed base on the lifetime shifting distribution. From the experiment result and the popular models of EGFR dimerization and oligomerization, we made a reasonable speculation that more combination of intra-monomer domain interaction is highly possible. These interaction could be materialized by rearrangement of structure after a collapsed state induced by AFM manipulation. We could also make a more aggressive speculation that in biological condition, where vertical direction squeezing force is unlikely on cell membrane, overexpression and aggregation of

EGFR on cell membrane could lead to regional crowding of the molecules and generate squeezing force in the direction parallel to the membrane planer that can cause behavior similar to AFM induced collapsing. This process can potentially initiate more intra-monomer interactions as seen in the proposed model and accelerate oligomerization.

114

5.5 References

1. Yarden, Y.; Sliwkowski, M. X. Untangling the ErbB signalling network. Nat. Rev. Mol. Cell Biol. 2001, 2, 127-137.

2. Yarden, Y. The EGFR family and its ligands in human cancer. signalling mechanisms and therapeutic opportunities. Eur. J. Cancer 2001, 37 Suppl 4, S3-8.

3. Riese, D. J.,2nd; Stern, D. F. Specificity within the EGF family/ErbB receptor family signaling network. Bioessays 1998, 20, 41-48.

4. Harris, R. C.; Chung, E.; Coffey, R. J. EGF receptor ligands. Exp. Cell Res. 2003, 284, 2-13.

5. Bublil, E. M.; Yarden, Y. The EGF receptor family: spearheading a merger of signaling and therapeutics. Curr. Opin. Cell Biol. 2007, 19, 124-134.

6. Tzahar, E.; Waterman, H.; Chen, X.; Levkowitz, G.; Karunagaran, D.; Lavi, S.; Ratzkin, B. J.; Yarden, Y. A hierarchical network of interreceptor interactions determines signal transduction by Neu differentiation factor/neuregulin and epidermal growth factor. Mol. Cell. Biol. 1996, 16, 5276-5287.

7. Schlessinger, J. Signal transduction by allosteric receptor oligomerization. Trends Biochem. Sci. 1988, 13, 443-447.

8. Singh, A. B.; Harris, R. C. Autocrine, paracrine and juxtacrine signaling by EGFR ligands. Cell. Signal. 2005, 17, 1183-1193.

9. Wiley, H. S. Trafficking of the ErbB receptors and its influence on signaling. Exp. Cell Res. 2003, 284, 78-88.

10. Ogiso, H.; Ishitani, R.; Nureki, O.; Fukai, S.; Yamanaka, M.; Kim, J. H.; Saito, K.; Sakamoto, A.; Inoue, M.; Shirouzu, M.; Yokoyama, S. Crystal structure of the complex of human epidermal growth factor and receptor extracellular domains. Cell 2002, 110, 775-787.

11. Cho, H. S.; Leahy, D. J. Structure of the extracellular region of HER3 reveals an interdomain tether. Science 2002, 297, 1330-1333.

12. Martin-Fernandez, M.; Clarke, D. T.; Tobin, M. J.; Jones, S. V.; Jones, G. R. Preformed oligomeric epidermal growth factor receptors undergo an ectodomain structure change during signaling. Biophys. J. 2002, 82, 2415-2427.

115

13. Webb, S. E.; Needham, S. R.; Roberts, S. K.; Martin-Fernandez, M. L. Multidimensional single-molecule imaging in live cells using total-internal-reflection fluorescence microscopy. Opt. Lett. 2006, 31, 2157-2159.

14. Martin-Fernandez, M. L.; Clarke, D. T.; Tobin, M. J.; Jones, G. R. Real-time studies of the interactions between epidermal growth factor and its receptor during endocytic trafficking. Cell. Mol. Biol. (Noisy-le-grand) 2000, 46, 1103-1112.

15. Li, H.; Oberhauser, A. F.; Redick, S. D.; Carrion-Vazquez, M.; Erickson, H. P.; Fernandez, J. M. Multiple conformations of PEVK proteins detected by single-molecule techniques. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 10682-10686.

16. Marszalek, P. E.; Lu, H.; Li, H.; Carrion-Vazquez, M.; Oberhauser, A. F.; Schulten, K.; Fernandez, J. M. Mechanical unfolding intermediates in titin modules. Nature 1999, 402, 100-103.

17. Marko, J.,F.; Siggia, E.,D. Stretching DNA. Macaromolecules 1995, 28, 8759-8700.

18. Rivetti, C.; Walker, C.; Bustamante, C. Polymer chain statistics and conformational analysis of DNA molecules with bends or sections of different flexibility. J. Mol. Biol. 1998, 280, 41-59.