University of Pennsylvania ScholarlyCommons
Publicly Accessible Penn Dissertations
2016
An Examination Of The Cytoplasmic Dynein Stepping Mechanism At The Single Molecule Level
Lisa Gail Lippert University of Pennsylvania, [email protected]
Follow this and additional works at: https://repository.upenn.edu/edissertations
Part of the Biochemistry Commons, and the Biophysics Commons
Recommended Citation Lippert, Lisa Gail, "An Examination Of The Cytoplasmic Dynein Stepping Mechanism At The Single Molecule Level" (2016). Publicly Accessible Penn Dissertations. 2428. https://repository.upenn.edu/edissertations/2428
This paper is posted at ScholarlyCommons. https://repository.upenn.edu/edissertations/2428 For more information, please contact [email protected]. An Examination Of The Cytoplasmic Dynein Stepping Mechanism At The Single Molecule Level
Abstract Rotational motions play important roles within biological processes. These motions can drive energy production as with the F1-ATP synthase or accompany domain motions during a conformational change such as the relative rotation of the large and small ribosomal subunits during protein synthesis. Studying these motions can provide insight into the mechanics of enzyme function that cannot be obtained by measuring its localization or chemical output alone. Rotational tracking can be done in the context of single molecule studies to observe enzymatic function at the single particle level. This presents an advantage over bulk solution studies because simultaneously occurring events, such as a solution of enzymes catalyzing a reaction, are not necessarily identical. By measuring the motions of a single molecule, short-lived states and rare events that would otherwise be averaged out can be detected. Here single molecule rotational tracking is utilized to examine the stepping mechanism of the cellular transport motor, cytoplasmic dynein. Cytoplasmic dynein walks along microtubules toward the minus end and is responsible for a wide range of cellular functions including cargo transport and chromosome alignment during cell division. This work employs a position and rotational tracking method, polarized total internal reflection fluorescence (polTIRF) microscopy. This technique requires a polarized fluorescent probe that is rigidly attached to the protein domain of interest and an optical system capable of measuring the orientation of such a probe. A functionalization method was developed to water-solubilize CdSe/CdS semiconductor quantum nanorods, which have polarized fluorescence emission, and coat them with the biotin binding protein, NeutrAvidin, in order to attach them to biotinylation sites within the dynein ring. A method was also developed to quantify the number and density of functional biotin binding sites on the nanorod surface and compare it to that of commercially available streptavidin quantum dots. These nanorods were attached to cytoplasmic dynein via two inserted biotinylation sites in AAA5 and AAA6 of the ring domain and rotational motions of the dynein ring were measured in real time using a home-built optical system capable of measuring both position and orientation simultaneously. These measurements revealed small, frequent ring rotations that occurred more than twice as frequently as steps along the microtubule track. The observed ring rotations are too small to be attributed to a classic powerstroke mechanism in which large-scale tilting produces forward motion, but instead support a flexible stalk model where tension between the two dynein heads, produced by conformational changes of the linker domain, results in bending of the flexible coiled-coil stalk and hinging at the microtubule binding domain.
Degree Type Dissertation
Degree Name Doctor of Philosophy (PhD)
Graduate Group Biochemistry & Molecular Biophysics
First Advisor Yale E. Goldman
Keywords Dynein, Fluorescence, Nanoparticles, Polarization, Single molecule Subject Categories Biochemistry | Biophysics
This dissertation is available at ScholarlyCommons: https://repository.upenn.edu/edissertations/2428 AN EXAMINATION OF THE CYTOPLASMIC DYNEIN STEPPING MECHANISM
AT THE SINGLE MOLECULE LEVEL
Lisa G. Lippert
A DISSERTATION
in
Biochemistry and Molecular Biophysics
Presented to the Faculties of the University of Pennsylvania
in
Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy
2016
Supervisor of Dissertation
______Yale E. Goldman, M.D., Ph. D. Professor of Physiology
Graduate Group Chairperson
______Kim A. Sharp, Ph. D. Associate Professor of Biochemistry and Biophysics
Dissertation Committee
Michael Ostap, Ph. D. Professor of Physiology Feng Gai, Ph. D. Professor of Chemistry Erika L.F. Holzbaur, Ph. D. Professor of Physiology Michael A. Lampson, Ph. D. Associate Professor of Biology Samara L. Reck-Peterson, Ph. D. Professor of Cellular and Molecular Medicine
AN EXAMINATION OF THE CYTOPLASMIC DYNEIN STEPPING MECHANISM
AT THE SINGLE MOLECULE LEVEL
COPYRIGHT
2016
Lisa G. Lippert
This work is licensed under the Creative Commons Attribution- NonCommercial-ShareAlike 3.0 License
To view a copy of this license, visit https://creativecommons.org/licenses/by-nc-sa/3.0/us/ ACKNOWLEDGMENT
This work would not have been possible without the support of my advisor, Dr.
Yale E. Goldman. His dedication to teaching provided me with the tools necessary to complete this project, and striving to reach his high scientific standards has made me a better and more precise researcher. He has been a fantastic mentor, and I consider myself lucky to have had the opportunity to work with him.
I would also like to thank the current and former Goldman lab members as well as collaborators who contributed to this work. Tali Dadosh and Jeffrey Hallock helped to develop the methods presented here. My friends and lab-mates Betsy McIntosh and
Michael Woody provided insightful scientific and non-scientific discussions. I am honored to have collaborated with Samara Reck-Peterson, Chris Murray, Klaus Schulten and Erika
Holzbaur, each of whom made essential contributions and improved both the quality and impact of this research.
Finally, I would like to thank my family. My parents, Skip and Laura Rung, instilled in me the importance of an education. And my husband, Andrew Lippert, has patiently supported me throughout this entire process. His willingness to prioritize my career enabled me to follow the opportunities that I have been given.
iii
ABSTRACT
AN EXAMINATION OF THE CYTOPLASMIC DYNEIN STEPPING MECHANISM
AT THE SINGLE MOLECULE LEVEL
Lisa G. Lippert
Yale E. Goldman, M.D., Ph. D.
Rotational motions play important roles within biological processes. These motions can drive energy production as with the F1-ATP synthase or accompany domain motions during a conformational change such as the relative rotation of the large and small ribosomal subunits during protein synthesis. Studying these motions can provide insight into the mechanics of enzyme function that cannot be obtained by measuring its localization or chemical output alone. Rotational tracking can be done in the context of single molecule studies to observe enzymatic function at the single particle level. This presents an advantage over bulk solution studies because simultaneously occurring events, such as a solution of enzymes catalyzing a reaction, are not necessarily identical. By measuring the motions of a single molecule, short-lived states and rare events that would otherwise be averaged out can be detected. Here single molecule rotational tracking is utilized to examine the stepping mechanism of the cellular transport motor, cytoplasmic dynein. Cytoplasmic dynein walks along microtubules toward the minus end and is responsible for a wide range of cellular functions including cargo transport and chromosome alignment during cell division. This work employs a position and rotational iv tracking method, polarized total internal reflection fluorescence (polTIRF) microscopy.
This technique requires a polarized fluorescent probe that is rigidly attached to the protein domain of interest and an optical system capable of measuring the orientation of such a probe. A functionalization method was developed to water-solubilize CdSe/CdS semiconductor quantum nanorods, which have polarized fluorescence emission, and coat them with the biotin binding protein, NeutrAvidin, in order to attach them to biotinylation sites within the dynein ring. A method was also developed to quantify the number and density of functional biotin binding sites on the nanorod surface and compare it to that of commercially available streptavidin quantum dots. These nanorods were attached to cytoplasmic dynein via two inserted biotinylation sites in AAA5 and AAA6 of the ring domain and rotational motions of the dynein ring were measured in real time using a home- built optical system capable of measuring both position and orientation simultaneously.
These measurements revealed small, frequent ring rotations that occurred more than twice as frequently as steps along the microtubule track. The observed ring rotations are too small to be attributed to a classic powerstroke mechanism in which large-scale tilting produces forward motion, but instead support a flexible stalk model where tension between the two dynein heads, produced by conformational changes of the linker domain, results in bending of the flexible coiled-coil stalk and hinging at the microtubule binding domain.
v TABLE OF CONTENTS
ACKNOWLEDGMENT ...... III
ABSTRACT ...... IV
LIST OF TABLES ...... IX
LIST OF ILLUSTRATIONS ...... X
CHAPTER 1: INTRODUCTION ...... 1
1.1 Cytoplasmic dynein and intracellular transport ...... 1
1.1.1: Cellular roles of transport motors ...... 5
1.1.2 Dynein structure and ATPase cycle ...... 7
1.1.3 Mechanics of dynein forward motion ...... 13
1.1.4 Proteins that modulate dynein activity ...... 14
1.1.5 Outstanding questions about dynein function ...... 18
1.2 Single molecule and polarization microscopy ...... 19
1.2.1 Photophysics of polarized probes ...... 22
1.2.2 Selection of a polarized probe ...... 25
1.2.3 Methods for measuring probe orientation ...... 31
1.2.4 Further applications of polarized microscopy ...... 32
CHAPTER 2: FUNCTIONALIZATION OF CDSE/CDS NANORODS ...... 34
2.1 Introduction ...... 35
2.2 Determining the number of binding sites for streptavidin and NeutrAvidin...... 37
2.3 Comparing coating methods for laboratory-made quantum nanorods...... 40 vi 2.4 Quantifying streptavidin coating of commercial quantum dots ...... 46
2.5 Discussion...... 56
CHAPTER 3: MEASURING ROTATIONS OF THE DYNEIN RING 61
3.1 Introduction ...... 62
3.2 Results ...... 67
3.2.1 Labeling dynein with polarized quantum nanorods for polTIRF measurements ...... 67
3.2.2 The dynein ring tilts while stepping ...... 72
3.2.3 Tilting frequency shows little dependence on ATP concentration ...... 79
3.2.4 The dynein ring is constrained when dragging an inactive head ...... 80
3.2.5 Angle changes reflect dynein flexibility ...... 82
3.3 Discussion...... 89
3.3.1 The power-stroke model of dynein stepping ...... 90
3.3.2 The flexible stalk model of dynein stepping ...... 93
3.3.3 Mechanics of dynein tilting ...... 94
CHAPTER 4: CONCLUSIONS ...... 100
CHAPTER 5: MATERIALS AND METHODS ...... 105
5.1 Nanorod preparation and B4F assay methods ...... 105
5.1.1 Water solubilization of nanorods ...... 105
5.1.2 Nanorod NeutrAvidin functionalization ...... 107
5.1.3 Determining Nanoparticle Concentrations...... 108
5.1.4 Biotin-4-Fluorescein Quenching Assay ...... 110
5.1.5 Biotin-4-Fluorescein Data Analysis ...... 111
5.2 Yeast dynein and polarized TIRF methods ...... 112
vii 5.2.1 Yeast transformation buffers ...... 112
5.2.2 Yeast transformation ...... 112
5.2.3 Dynein purification buffers ...... 114
5.2.4 Dynein affinity purification ...... 116
5.2.5 DNA-oligo labeling of SNAP-tagged dynein heterodimers ...... 117
5.2.6 Single molecule assay buffer ...... 118
5.2.7 Flow cell construction for single molecule assays ...... 119
5.2.8 Polarized TIRF imaging of dynein flow cells ...... 121
5.2.9 Angle fitting and changepoint analysis ...... 121
5.2.10 Correction for unknown probe angle ...... 124
5.2.11 Molecular dynamics simulations ...... 124
5.2.12 Mechanical characterizations ...... 127
BIBLIOGRAPHY ...... 131
viii LIST OF TABLES
Table I ...... 4
Table II ...... 43
Table III ...... 47
Table IV ...... 52
Table V...... 53
Table VI ...... 76
Table VII ...... 98
ix LIST OF ILLUSTRATIONS
Figure 1 ...... 3
Figure 2 ...... 9
Figure 3 ...... 12
Figure 4 ...... 16
Figure 5 ...... 21
Figure 6 ...... 24
Figure 7 ...... 38
Figure 8 ...... 39
Figure 9 ...... 42
Figure 10 ...... 44
Figure 11 ...... 48
Figure 12 ...... 49
Figure 13 ...... 51
Figure 14 ...... 54
Figure 15 ...... 55
Figure 16 ...... 57
Figure 17 ...... 58
Figure 18 ...... 65
Figure 19 ...... 68
Figure 20 ...... 70
Figure 21 ...... 71
x Figure 22 ...... 73
Figure 23 ...... 74
Figure 24 ...... 78
Figure 25 ...... 81
Figure 26 ...... 83
Figure 27 ...... 85
Figure 28 ...... 86
Figure 29 ...... 91
Figure 30 ...... 97
Figure 31 ...... 129
Movie S1 ...... 87
xi CHAPTER 1: INTRODUCTION
Sections of this chapter, including some text and figures, are from our publication Shroder,
Lippert, and Goldman accepted for publication in Methods and Applications in
Fluorescence.
1.1 Cytoplasmic dynein and intracellular transport
Transport is essential for directing cargo to specific locations within a cell. Cargo can be anything from proteins destined for the cell membrane, to degradation enzymes that recycle polymers, to secreted proteins or neurotransmitters [6]. Motor-driven transport is necessary for the cargo to reach its destination at the appropriate time. Without transport motors cargo would have to depend on diffusion, but diffusion is slow when traversing long distances, and random. While it would take only 0.62 seconds for GFP, a 27 kDa protein with a measured diffusion constant of 27 μm2/s diffusion constant in eukaryotic cytoplasm [3], to diffuse across a small 10 μm eukaryotic cell, diffusion time (t) increases with the square of the displacement (d) according to t=
[8]. (Note: The Stokes-Einstein relationship is used for rough approximation, although it 1 does not necessarily apply in a complex cellular environment.) By this approximation it would take a lysosome, which has a radius between 0.05 and 0.6 μm [9], roughly 250 times as long to diffuse as a GFP protein with a radius of only 1.2 nm. This means that it would take a lysosome 2.6 minutes to diffuse across a small eukaryotic cell, and 18 days to travel the length of a short axon. Random diffusive motions would make it virtually impossible to segregate biological macromolecules in a specific pattern, such as chromosome alignment during mitosis and meiosis. Long transport times would prohibit signal transmission across a cell fast enough to, for example, produce an allergic response by releasing secretory granules from mast cells [10].
Motor-driven transport alleviates the problem of diffusion-limited transport.
Molecular motors are proteins or protein complexes that hydrolyze ATP in order to walk processively, or take multiple successive steps without detaching, along filamentous protein polymers. There are three major families of transport motors: myosins, kinesins, and dyneins [11-13]. Myosins walk along actin filaments, while kinesins and dyneins walk on microtubules, with dyneins moving toward the microtubule minus ends and most kinesins moving toward the plus ends (Figure 1A). Motor-driven transport reduces transport times relative to diffusion, especially for large particles over long distances.
Mammalian cytoplasmic dynein has a maximum velocity of 1.1 μm/s [14]. At this rate it would take only 9.1 s to travel the length of a 10 μm cell or 15 minutes to traverse the length of a 1 mm axon. The advantage of motor-driven transport increases with both distance and cargo size (Table I).
2 A
Actin
Microtubules
Kinesin 1
Cytoplasmic Dynein
Myosin V
Cargo
Microtubule Organizing Center
Nucleus
B
Selective Recruitment Tug-of-War
Selective Activation (Dynein) Selective Activation (Kinesin)
Figure 1: Motor-driven cargo transport in the cell A. Kinesin and dynein walk along microtubules. Dynein moves toward microtubule minus ends, which tend to orient near the center of the cell, while kinesins generally move toward the microtubule plus ends, which are oriented at the cell periphery. Myosin walks along actin, with different myosin classes moving in opposite directions. B. Adaptor proteins (green) can regulate cargo transport in different ways, either by selective recruitment of a single motor type, recruitment of multiple motor types resulting in a tug-of-war where the strongest motor team wins, or selective activation of certain motors.
3
Table I: Comparison of diffusion and motor-driven transport Estimated time for a small protein or vesicular cargo to travel a given distance either by diffusion or when transported with mammalian cytoplasmic dynein. Diffusion times were estimated using the measured diffusion constant of GFP in CHO cytoplasm [3] and extrapolated to different distances and for larger cargo using the Stokes-Einstein relationship. This equation provides a rough approximation since it only applies at low Reynolds number and does not account for crowding due to cytoskeletal elements such as actin and microtubules. Dynein transport times are based on velocities from in vitro measurements of mammalian cytoplasmic dynein.
4 1.1.1: Cellular roles of transport motors
Transport motors can be spatially and temporally regulated to control the destination of specific cargos, providing another advantage over diffusion alone. Different families of motors, myosins, kinesins, and dyneins, and different classes within those families have distinct biochemical and mechanical properties that enable them to perform highly specialized roles within the cell. While there are many different functions that involve the coordination of multiple motor types, one elegant example is the alignment of chromosomes during metaphase of cell division [15, 16]. In order to ensure the retention of a single chromosome copy in each of the resulting daughter cells, the cycle is heavily regulated. The cycle does not progress into anaphase until all chromosomes are aligned with all kinetochores bound to microtubules connected to centromeres on opposite sides of the cell. Transport motor motility and tension sensing play important roles in bringing the chromosomes to the cell center and signaling that all kinetochores are correctly attached to microtubules. This process involves two opposing transport motors: CENP-E (a kinesin 7 motor) and cytoplasmic dynein. Their opposing activities, CENP-E moves toward microtubule plus ends while dynein moves toward minus ends, is required for correct chromosome alignment [15]. Inhibition of CENP-E causes singly attached chromosomes to accumulate at the spindle poles, positioned at the microtubule minus ends, presumably driven by dynein motors [17]. Chromosomes in dynein-inhibited cells reach the metaphase plate at the microtubule plus ends but are not properly aligned, suggesting that dynein is important for sensing tension due to microtubule attachment [18]. The interplay between
CENP-E and dynein, as well as the many other kinetochore and spindle assembly proteins, confers the necessary accuracy to ensure that chromosome segregation occurs correctly.
5 As with kinetochores, intracellular cargos often bind multiple motors of the same or different types, providing an additional level of regulation (Figure 1B) [6, 19]. Motors of the same type may work against motors of different types moving in opposite directions or on different filaments in a tug-of-war that results in bi-directional cargo motion.
Alternately, selective activation or deactivation of motor subsets by regulatory proteins enables coordinated unidirectional motion even in the presence of multiple motor types
[20]. For example, autophagosomes in neurons move bidirectionally at the distal end of the axon driven by both kinesin and dynein motors, but as autophagosomes mature and fuse with lysosomes or late endosomes, the kinesin motors become inactive and dynein-driven motion dominates, resulting in minus-end directed transport toward the cell center [21].
Scaffolding proteins can serve to either selectively recruit or activate specific types of proteins, and this recruitment or activation can change as a function of cellular location or environment or cargo composition [6, 22]. One example of selective environment-specific selective activation is the regulation of dynein recruitment to lipid vesicles depending on the vesicle lipid composition. The scaffolding protein, RILP, can recruit the dynein binding protein p150Glued, and in turn ORP1L can recruit the RILP/p150 complex to vesicles with high cholesterol content. However, in vesicles with low cholesterol content ORP1L adopts a conformation that ultimately results in dynein dissociation and vesicle localization to the plus ends of microtubules [23]. With a wide range of such scaffolding proteins available in cells a single motor protein, such as cytoplasmic dynein, can be used to perform many different functions within the cell. Since a single dynein gene is used for all dynein-driven transport in the cytoplasm (different dynein genes are responsible for axoneme sliding and
6 transport within flagella) dynein binding proteins are important for controlling dynein- cargo interactions [24].
1.1.2 Dynein structure and ATPase cycle
One important player in cellular transport is cytoplasmic dynein. Dynein is the motor primarily responsible for minus-end directed, or retrograde, microtubule transport in eukaryotes. In fact, with exception of kinesin-14, it is the only identified minus-end directed microtubule motor (outside of some plants) [24]. As with most processive motors, dynein exists as a homodimer of two heavy chains, which contain the motor domain.
However, dynein is structurally distinct from any of the other class of transport motors.
The dynein heavy chain is composed of an N-terminal tail, a linker domain, six AAA domains, a stalk which contains the microtubule binding domain, and a C-terminal domain
(Figure 2A) [25]. The tail domain is the dimerization domain and also serves as the binding site for many dynein associated proteins including dynactin and BicD. Dynein is classified as a AAA protein due to the presence of six AAA domains arranged in a ring (Figure 2B), however dynein is unusual among this class of enzymes. Most AAA proteins are hexamers, consisting of six protein subunits, but dynein contains six unique AAA domains within a single polypeptide chain [24, 25]. It is worth noting that neither myosins nor kinesins are
AAA proteins, making dynein unique among transport motors as well. The AAA ring is interrupted by a coiled coil stalk which forms a long protrusion extending out of the ring to contact the microtubule track via the microtubule binding domain (MTBD). The buttress extends from the ring and contacts the stalk to stabilize the connection. The linker domain
7 spans the face of the AAA ring, and the C-terminal domain lies on the face opposite the linker [26].
ATP binding and hydrolysis occur in the AAA ring. Specifically, ATP binds in
AAA1 – AAA4 but is hydrolyzed only in AAA1, AAA3 and AAA4 [27]. AAA5 and
AAA6 cannot bind or hydrolyze ATP. While only ATPase activity in AAA1 is required for processive motility, AAA3 ATPase activity promotes motility and is thought to play a regulatory role [28]. AAA4 also regulates motility but to a lesser degree than AAA1 [29].
AAA1 and 3 contain both Walker A and Walker B motifs, while AAA2 contains a Walker
A but lacks a Walker B motif. The Walker A sequence is required for ATP binding, and
Walker B is required for ATP hydrolysis. This is demonstrated in AAA1 Walker A mutants that cannot bind ATP and Walker B mutants that can bind but not hydrolyze ATP [30]. In both cases, dynein motility is abolished. Comparable mutations in AAA3 dramatically slow dynein velocities by impairing dynein’s ability to release from microtubules and therefore slowing ATPase activity in AAA1, but these mutations do not eliminate motor activity [31]. Interestingly, the ATP cycles of AAA1 and 3 are not coordinated. Each AAA domain is composed of a large and small subdomain; in AAA1 these domains are open in the absence of nucleotide and close upon the binding of ATP [11, 25, 32].
8
A
N Tail Linker 1 2 3 4 Stalk 5 6 C C
B Essential ATPase
Regulatory ATPase
1 2
Tail AAA Ring 6 3 4 5
Buttress
Stalk
Microtubule Binding Domain
Figure 2: Structural domains of cytoplasmic dynein A. Primary structure of the dynein heavy chain. The tail is positioned at the N-terminus and is contiguous with the linker domain. The AAA domains (marked 1 through 6) are interrupted between AAA4 and AAA5 by the stalk which contains the microtubule binding domain. AAA6 is followed by the C-terminal domain (dark red). B. Schematic of the dimeric structure. Dynein dimerizes via the N-terminal tail. The AAA domains are arranged in a ring, which is spanned by the linker. ATP binds in AAA1 through AAA4. Only ATPase activity in AAA1 is absolutely required for motility. The coiled-coil stalk extends from between AAA4 and AAA5 into the microtubule binding domain. It is supported by the buttress, which is important for conferring conformational changes from ATP hydrolysis in AAA1.
9 The most prominent conformational change that occurs during the ATP cycle is the motion of the linker domain. As shown in figure 2 the linker is positioned toward the N- terminus of dynein and spans the face of the AAA ring [26, 33]. In the nucleotide-free (or apo) and ATP bound states the N-terminal region of the linker is docked between AAA4 and 5 near the stalk (as shown in figure 2B). Upon ATP hydrolysis into ADP and phosphate
(Pi) the linker undergoes a conformation change, referred to as “priming”, in which it moves from AAA4/5 to AAA3 [34, 35]. The reverse motion, often called the linker
“powerstroke”, is associated with the release of the hydrolysis products. This straightening of the linker is thought to pull cargo forward and enable communication between the two dynein heads. The nucleotide state of AAA1 also controls the affinity of dynein for the microtubule track.
Coordination of linker conformational changes and changes in affinity for the microtubule results in an AAA1 ATPase cycle as follows: in the absence of nucleotide dynein is strongly bound to the microtubule (Figure 3A). Binding of ATP to AAA1 causes dynein to detach from the microtubule (Figure 3B). While in the unbound state ATP hydrolysis causes reorganization of the linker domain to its unprimed or pre-powerstroke state (Figure 3C) and re-binding to the microtubule, usually forward of the previous binding site (Figure 3D). Release of phosphate results in a linker powerstroke followed by release of ADP to return to the starting position (Figure 3A) except farther along the microtubule [25, 36]. Since AAA1, the primary site of ATP hydrolysis, is ~28 nm away from the microtubule binding site, conformational changes associated with nucleotide state must be communicated allosterically over a large distance to modulate affinity of the
MTBD for microtubules. Conversely, microtubule binding must be communicated back to 10 AAA1 to stimulate the release of ADP and phosphate. This communication is achieved in part by a registry shift of the coiled coil stalk that alters the conformation of the MTBD and changes microtubule affinity [37, 38].
The function of the C-terminal domain is somewhat mysterious. It constitutes the most substantial difference between yeast and mammalian cytoplasmic dynein motor domains. In yeast the C-terminal region consists of a single α-helix that sits on the face of the ring opposite the linker, but other organisms including mammals and Dictyostelium contain an additional sequence that forms a larger globular domain that sits next to AAA5 and 6 [39]. Yeast and mammalian dynein exhibit functional differences: yeast dynein can exert higher forces than mammalian and is more processive, but it exhibits slower velocities. Truncation of the mammalian C-terminal domain increases the force production and processivity, suggesting that it is in part responsible for the functional differences between yeast and mammalian dynein [40]. It is worth noting that, in addition to C-terminal truncation, binding of the accessory proteins dynactin and BICD2 to mammalian dynein increases its stall force close to that of yeast dynein [41].
11
A Cargo Powerstroke ADP ATP Pi Detachment
Cargo D B Cargo
Cargo ATP C
Binding ADP + Pi Linker Re-Priming
Figure 3: Dynein ATPase cycle A. In the absence of nucleotide dynein is strongly bound to the microtubule B. Upon binding of ATP dynein detaches from the microtubule
C. In the unbound state ATP is hydrolyzed to ADP and phosphate (Pi), and the linker undergoes a conformational change to the primed state D. Dynein re-binds to the microtubule ahead (toward the minus end) of its previous position E. It undergoes a linker powerstroke upon the release of phosphate. After the release of ADP it adopts the apo conformation as in A except farther along the microtubule.
12 1.1.3 Mechanics of dynein forward motion
Since dynein’s structure is so different from other transport motors it is not surprising that the stepping mechanism is also different. Myosin V [42, 43] and kinesin 1
[44], both canonical examples of their respective motor classes, have been shown to walk in a hand-over-hand fashion, alternating leading and trailing heads with each step. This hand-over-hand mechanism is achieved in both cases by gating to ensure that the trailing head detaches before the leading head and only when the leading head is bound to the actin or microtubule [45, 46]. Gating is usually the result of strain between the two heads that causes different nucleotide binding, hydrolysis or release rates for the leading and trailing heads or of a mechanical preference for the detached head to bind in front of the attached head. Dynein steps using both a hand-over-hand mechanism, like myosin V and kinesin 1, as well as an “inchworm” mechanism in which forward motion occurs without the leading and trailing heads trading position [47, 48]. While the mechanism of nucleotide gating in dynein is not clear, dynein does exhibit a tension-dependent release rate from microtubules.
This is demonstrated both in the fact that the trailing head steps more frequently the larger the separation between the heads [47, 48], suggesting that tension plays a role in determining the likelihood of stepping, and by directly pulling on a dynein motor either with assisting (toward the direction of motion) or resisting load [49]. Consistent with tension-dependent gating, the dynein motor steps forward more frequently under assisting load that resisting load and, under sufficiently high resisting load, begins to walk backward
[49]. This tension-dependent microtubule release rate can explain how dynein walks processively over long distances, nearly 2 μm on average for yeast cytoplasmic dynein
[50].
13 Another feature of the strictly gated hand-over-hand mechanisms of myosin V and kinesin 1 is that they rarely take backward steps under unloaded conditions [51, 52].
Dynein, on the other hand, not only takes frequent backward steps but also steps sideways, switching between microtubule protofilaments [50]. This could be explained by weaker coupling between heads than in myosin or kinesin. While sideways and backward stepping may appear an inefficient way to travel, this irregular stepping pattern enables dynein to navigate around obstacles on the microtubule better than kinesin 1, which walks forward along a single microtubule protofilament [53].
1.1.4 Proteins that modulate dynein activity
As described earlier, transport motors are regulated in many different ways to ensure that cargo reaches the correct destination at the appropriate time. Dynein is no exception. There is a myriad of proteins that bind to dynein and regulate its activity, many of which still have unknown function and/or binding sites. Only a small number of these regulatory proteins will be discussed here. The motor activity of dynein comes from the dynein heavy chain, which contains the AAA ring, the stalk, the linker and the tail.
However, the full dynein complex contains at least eight subunits for a total molecular weight of 1.4 MDa [27]. In addition to the heavy chain dimer, this includes two copies each of the dynein intermediate chain and light intermediate chain, plus up to three light chains.
These bind to the tail region of the heavy chain and are thought to be involved in binding cargo.
14 One method of dynein regulation arises from auto-inhibition through the formation of Φ-particles. In this auto-inhibitory state, named for its resemblance to the Greek letter
Φ, the two dynein motor domains adopt a stacked conformation with the C-terminal domains of the two heads contacting each other (Figure 4A) [14]. This forces the two stalks to point in opposite directions, physically prohibiting the two microtubule binding domains from interacting with the microtubule simultaneously as required for processive motility.
In this conformation dynein diffuses back and forth along the microtubule rather than walking processively toward the minus end. The formation of the auto-inhibited state depends on the length and rigidity of the dynein dimerization domain in artificial constructs, with short or rigid connectors preventing Φ-particle formation. A proposed mechanism for alleviating this auto-inhibition in vivo is binding of accessory proteins to the dynein tail, preventing the formation of Φ-particles [22, 27].
15
A B
Dynactin BICD2
Intermediate Chain
Light Intermediate Chain
Lis1
Figure 4: Regulation of mammalian cytoplasmic dynein A. In the absence of cargo or adaptor proteins dynein can adopt an auto-inhibited conformation. In this conformation, or 휙 particle, the C-terminal domains of both heads contact each other. This positions the stalks so that the two microtubule binding domains point in opposite directions, preventing both from binding to the microtubule simultaneously. B. Components of the dynein complex, the intermediate and light intermediate chains, bind along the heavy chain tail. The dynein activator protein dynactin and the scaffolding protein BICD2 also bind along the dynein tail. The regulatory protein Lis1 is unusual in that it binds to the ring domain, specifically contacting AAA3 and AAA4.
16 The primary activator of dynein is dynactin, a 1.2 MDa complex which contains 11 unique polypeptides and at least 20 subunits [54]. Dynactin binds to the tail of dynein and enhances processive motility of full-length mammalian cytoplasmic dynein (Figure 4B)
[55, 56]. The largest subunit, p150Glued [57], consists of a long coiled-coil and a microtubule binding motif called the CAP-Gly domain. The p150Glued protein and its microtubule binding function are required for dynein function [58]. While dynactin can link dynein directly to cargo, it also binds to adaptor proteins that then bind to cargo [54]. One such adaptor protein is BICD2. BICD2, a member of the Bicaudal D family, binds both dynein and dynactin with its N-terminal region and increases the processivity of the dynein- dynactin complex, possibly by increasing the stability of the interaction or releasing dynein or dynactin auto-inhibition [22, 27, 59]. The C-terminal region binds to cargo proteins, including Rab6, a protein found on the Golgi and vesicles. In Drosophila melanogaster the
C-terminal region of BicD (the fly homologue of BICD2) is involved in auto-inhibition, and binding to cargo alleviates this auto-inhibition [22]. Other cargo adaptors which, like
BICD2, have been shown to activate dynein motility are Hook1, Hook3, Spindly, and
Rab11-FIP3 [27, 60].
Another protein that plays an interesting role in regulating dynein motility is Lis1
[27]. Rather than binding the tail of dynein like most regulatory and scaffolding proteins,
Lis1 binds to the motor domain, contacts AAA3 and 4 of the AAA ring and prevents the linker from undergoing its usual conformational change (Figure 4B) [61]. Binding of Lis1 uncouples the AAA1 ATPase cycle from changes in affinity of the microtubule binding domain for microtubules, trapping dynein in a strongly bound or force-producing state [62].
Lis1 aids in kinesin-dependent localization of dynein to microtubule plus ends and may 17 play a role in loading cargo onto microtubules and in transporting large cargo under high load [27].
Regulatory proteins are important for dynein’s function in vivo, however yeast dynein is able to walk processively along microtubules in vitro without any of these accessory proteins [27]. Understanding dynein’s mechanism without any binding proteins may improve our understanding of how regulatory proteins interact and change dynein function.
1.1.5 Outstanding questions about dynein function
Substantial progress has been made in recent years toward fully understanding the function and mechanics of dynein and its accessory proteins. Structural studies have provided critical information about conformational states and protein-protein interactions
[1, 26, 32, 34, 63-65]. Still, many questions remain about the dynein stepping mechanism.
The site of ATP hydrolysis in AAA1 and the large separation from the microtubule binding domain contributes to some of the mystery remaining behind the dynein stepping mechanism. The nucleotide state in AAA1 must be communicated through the ring and a long, flexible stalk to change microtubule binding affinity in the microtubule binding domain. This long coiled-coil stalk must also support sufficient force to pull the second motor domain and cargo forward. What forces are exerted on the stalk and tail to cause processive forward motion, and what roles do compliance in the stalk, linker and tail play?
The presence of multiple sites that hydrolyze ATP within the AAA ring opens many
18 questions about how they contribute to the conformational changes that drive dynein motility. Additionally, the C-terminal sequence and its size variability between species raises questions about its function and role in dynein regulation. And beyond the structure and function of the dynein heavy chain, there are many dynein binding proteins such as dynactin, BICD1, Lis1 and many others whose roles in dynein activation and regulation are only just beginning to be understood. The work presented in this thesis attempts to address some of these outstanding questions as well as introduce a new technique that can be used to examine some of the questions that are not directly studied here.
1.2 Single molecule and polarization microscopy
The ability to observe the motions of single proteins has played an important role in our understanding of the mechanisms of molecular motors [66-68]. Single molecule techniques enable the study of individual states within a mixed population because they eliminate the need to average over a large number of particles. Early fluorescence microscopy was restricted by the diffraction limit of light. The diffraction limit or Abbe resolution, d, is dictated by the wavelength of light, λ, and the numerical aperture of the
휆 microscope objective, NA, following the relationship 푑 = . This constrains the 2푁퐴 resolution of a typical fluorescence experiment to around 250 nm. The development of single molecule and super-resolution microscopy shattered these constraints and enabled the study of motions on the order of single nanometers, much smaller than the diffraction limit of light. The contribution of these methods was deemed significant enough to be awarded the 2014 Nobel Prize in Chemistry [69-71]. While many super-resolution
19 techniques have been developed from these initial discoveries, the simplest, and the basis for many of the more advanced techniques such as PALM and STORM, takes advantage of the predictable shape of a diffraction-limited fluorescence or scattered spot. If the diffraction-limited spots are well isolated the intensity pattern can be fit to a Gaussian distribution (Figure 5A, B). The location of the point-emitter, defined by the peak of the
Gaussian function, can be determined within a few nanometers, depending on the intensity of the emission among other factors [42]. This point tracking method can be combined with total internal reflection fluorescence microscopy (TIRF), which illuminates only a small distance (about 100 nm) into the sample (Figure 5C), to enable precise localization with low background fluorescence of single fluorescent particles.
20 A
C
SAM 100 nm
OBJ
B Intensity
Position
Figure 5: Single molecule TIRF microscopy A. Diffraction limited fluorescence emission of a single molecule measured with TIRF microscopy. Scale bar is 300 nm. B. Intensity of each pixel in a row along the x-axis through the image shown in A. Intensity can easily be fit by a Gaussian distribution both in x and y to precisely determine the center of the diffraction limited spot and therefore the fluorescent particle. C. Schematic of total internal reflection fluorescence (TIRF) illumination. The excitation beam (green) encounters the microscope objective (OBJ) offset from the center. The lenses in the objective diffract the beam so that it encounters the sample (SAM) at an angle. TIRF is achieved when the incident angle of the beam is equal to the critical angle between the glass (or quartz) slide and the aqueous sample, resulting in an evanescent wave that illuminates only about 100 nm into the sample. In contrast, an illumination beam that passes through the center of the objective (red) passes straight through the sample and illuminates a large distance into the sample.
21 While super-resolution microscopy can be used to precisely determine the position of single particles, polarization microscopy can be used in combination with single molecule assays to determine a particle’s orientation. Such measurements can provide important information about protein conformational changes that would be difficult to obtain using other single molecule methods. For example, polarized fluorescence measurements have been used to study the hand-over-hand mechanism of myosin V [43] as well as observe the diffusion of the unbound head as it searches for a binding site along actin [72]. Polarization microscopy depends on a specific property of some optical probes, namely the preferential excitation by or selective emission or scattering of polarized light.
The axis of polarized absorption or emission is referred to as a dipole moment. The detected intensity of the probe either depends on the orientation of the excitatory light wave or on the orientation of the detected light relative to the probe dipole moment.
1.2.1 Photophysics of polarized probes
Interactions between probes with dipole moments and polarized light depend on their orientation. Absorption of light is most likely when the polarization, i.e. the transverse direction of its oscillating electric field, is oriented parallel to the absorption dipole of the polarized probe (Figure 6A). The probability of photon absorption, Pa, is proportional to
2 the square of the projection of the electric field onto the dipole orientation, Pa cos (a), where a is the angle between the polarization of the exciting light and the absorption dipole
(for instance, ax in Figure 6A when the light is polarized along the x axis, x). As fluorescence is only produced when the molecule absorbs light, by varying the input
22 polarization and detecting the fluorescence for each polarization, information about the probe angle is obtained.
Similarly, the emitted photons are polarized along the emission dipole axis, so polarized detectors, often called analyzers, can also be used to determine the probe orientation (Figure 6B). The relative probability of detecting an emitted photon that is
2 captured by the collection optics and projected onto the detector is given by Pe cos (e), where e is the analyzer polarization relative to the emission dipole. The likelihood of capturing a photon is related to the emission dipole orientation relative to the propagation
2 axis of the detector by Pc sin (e), where e is the angle between the dipole and the direction of observation (ey for the observational eye on the y axis in Figure 6B). No photons are emitted along the dipole axis and the probe looks brightest when it is perpendicular to the analyzer axis (Figure 6B).
For rhodamine and other xanthene derivative fluorophores (fluorescein, eosin) and
Cy-dyes, the absorption and emission dipoles are parallel to each other and to the long axis of the chromophore [73, 74]. Green fluorescent protein (GFP) and its variants also have aligned dipoles [74], whereas elongated quantum dots, termed quantum rods, have well polarized absorption and emission dipoles, but they are not necessarily aligned with each other [75].
23
Figure 6: Orientation dependence of polarized excitation and emission 2 2 A-B. Relative probabilities for fluorophore absorption (A, cos (ax) and cos (ay)) and collection of its 2 2 emission (B, sin (ez) and sin (ey)). In the microscope coordinate frame (x, y, z), the optical axis is z.
For axial illumination (heavy arrow in A), x and y are excitation polarizations. The probe absorption and emission dipole moments (considered to be parallel) in the (x, y, z) frame are defined by p (axial angle) and p (azimuth, the angle between the projection of the dipole onto the x-y plane (grey triangle) and the positive x axis). ax and ay are angles between the probe dipole and excitation polarizations along the x- and y-axes. ey and ez are angles between the probe dipole and detector optical paths in the y- and z-axes. Adapted from Rosenberg et al., 1993 with permission.
24 In contrast to the excitation and emission of fluorescent probes, light absorption and elastic re-emission from metal nanorods is the result of coupling between the oscillating electromagnetic field and surface electrons in the particle, termed local surface plasmon resonance (LSPR). Metal nanorods are not fluorescent, but their interactions with light depend on the difference in the LSPR between the long and short axes of the rod [76].
These axes have similar relationships to input and output light as those described for fluorescent dipoles.
The properties of dipolar and elongated metal probes can be exploited to measure their spatial orientations. Changing the polarization of input light to detect the angle of the excitation dipole, detecting the polarization of the output light to determine the angle of the emission dipole, detection of the propagation distribution of emitted light, directional scattering and optical phase retardation from nanorods have all been used, separately or sometimes in tandem, to determine the angles of probes attached to biologically-relevant macromolecules.
1.2.2 Selection of a polarized probe
Selection of an appropriate probe is important for single molecule orientation experiments. Many small organic fluorescent probes, variants of GFP, semiconductor nanoparticles (such as quantum dots), and light-scattering gold nanoparticles are available for site-specifically labeling macromolecules. Desirable traits for a single molecule probe include high brightness, which includes absorption cross-section and quantum yield, low probability of blinking, resistance to photobleaching, and facility for incorporation into the
25 biological system without disturbing function. The latter characteristic depends on physical size, ligation chemistry, and the specificity of the labeling site. For angular studies, restricted rotational mobility is also important. Available fluorescent and light scattering probes vary greatly in these properties.
Fluorescent Proteins
The minimally perturbing approach for interrogating macromolecular orientation is to use the native optical properties of a fluorescent protein, such as the light-harvesting complexes from a purple bacterium probed at the single-molecule level by Bopp et al. [77].
Most biological materials, however, are not inherently fluorescent. Recent technological advances in the incorporation of unnatural amino acids have greatly increased the potential for expressing intrinsically fluorescent proteins that can monitored for site-specific changes in orientation [78]. Promisingly, a 3-hydroxyflavone dye, which exhibits dual color fluorescence depending on the polarity of its environment, and acridon-2-ylalanine, a visible FRET acceptor, have been incorporated into unnatural amino acids for ensemble microscopic analysis [79, 80].
Alternately, fluorescent proteins can be genetically fused to the target protein through a short linker. This approach has been used successfully at the single-molecule level with green fluorescent protein (GFP) [81] fused to nuclear pore complexes to determine their orientation at the nuclear membrane, and with the photoswitchable GFP variant, Dendra2, fused to actin or hemagglutinin to observe changes in rotational rate of single molecules within cells [82]. The linker length between GFP and the protein under study is a crucial parameter in this approach. Linker length must be varied to find the
26 species that gives the highest fluorescence anisotropy (lowest mobility), as linkers with several flexible bonds in series may not effectively couple protein rotational motions to those of the fluorescent reporter. In fact, FRET experiments usually rely upon the assumption that probe motion is completely isotropic. For orientation studies, though, the probe motion must follow the protein motions and so cannot be isotropic.
Other ways of covalently attaching fluorophores to proteins have been recently introduced [83], including SNAP-, Halo- and other tags [84], which are then modified with organic dyes. These protein tags are also subject to concerns regarding flexibility of the linkers.
Organic probes
Organic fluorescent probes with useful spectral properties in the visible wavelength range have extended conjugated double bonds leading to oriented absorption and emission dipole moments. They are much smaller than GFP variants and the semiconductor nanorods described below. The ideal small molecule reporter should bind specifically and rigidly to the macromolecule in a fixed or known orientation. The DNA-intercalating fluorophores YOYO-1[85] and SYTOX Orange [86] adopt fairly rigid angles perpendicular to the DNA axis and have been used for single-molecule studies of DNA orientation within cells. In addition, the membrane-binding DiI has been used at single- molecule sensitivity to detect membrane packing [87], and effects of gangliosides on the formation of lipid rafts [88].
In special cases an organic fluorophore can be introduced into the biological system by linking it to a ligand that recognizes a specific target. For example, the mushroom toxin
27 phalloidin is a small, bicyclic peptide that binds to actin filaments with nanomolar affinity and is commercially available tagged with many fluorescent probes [89, 90]. As with fluorescent proteins, the linker length between the dye molecule and the recognition motif strongly affects the probe’s rotational mobility [85]. Fluorescent antibodies have not been successful in orientation studies due to variable probe angle [85, 91].
Linker flexibility can be minimized with a short reactive moiety such as maleimide or iodoacetamide to covalently attach rhodamine [92, 93], Cy3 [94], eosin-5 [95] or other fluorophores to either native or introduced cysteines in the target protein. However, when bound to a single Cys residue, the local orientation of the probe relative to the labeling site is not known [96, 97] and may vary considerably depending on the geometry of the fluorophore and the local environment at that residue, which may allow free rotation about the attachment point or hinder motion. For this reason, Adachi et al. [94], considered several cysteine point mutations in the subunit of F1-ATP synthase labeled with Cy3- maleimide and chose the one showing the highest fluorescence anisotropy. Fortunately, single-molecule studies of rhodamine bound the most reactive cysteine in actin (Cys374)
[93, 98] show very low probe mobility because the fluorophore occupies a groove in the protein with known local orientation [92].
Probes with two separate reactive groups that bind to the macromolecule can achieve markedly reduced local motions [99], and fixed local orientation relative to the structure. Bifunctional rhodamine (BR) contains two iodoacetamide groups that flank the chromophore and can crosslink two cysteines which are 7 or 8 residues (~1.1 nm) apart on an -helix. This configuration aligns the absorption and emission dipoles parallel to the helical axis. BR was originally synthesized for labeling myosin light chains in muscle fiber 28 polarized fluorescence studies [100]. Bifunctional rhodamines have also been adopted for labeling troponin C in muscle fibers [101, 102], calmodulin for single molecule polarized total internal reflection fluorescence (polTIRF) experiments in non-muscle myosins [43,
103, 104], kinesin motor proteins [105], the subunit in F1-ATPase [106], and ribosomal elongation factor EF-G [107].
Attachment of fluorophores to one or two cysteine residues can only be used for in vitro applications, as they are not specific enough for in-cell labeling. The use of bi- arsenical probes that bind specifically to tetra-cysteine motifs, such as FlAsH [108] and
AsCy3 [109] should be considered, but they have not yet been used in published single molecule orientation experiments.
Nanoparticles
It is not always possible to express a cysteine-lite mutant capable of specifically binding an organic dye or to rigidly attach fluorescent proteins in the desired domain.
Certain applications might also require a substantially brighter probe than an organic fluorophore or fluorescent protein. In these instances, nanometer-sized inorganic particles provide an alternative. Fluorescent semiconductor nanoparticles, such as quantum dots
(QDs), are composed of CdS, CdSe, or CdTe [110], and are often coated with a shell of
CdS or ZnS to increase brightness and reduce blinking [111, 112]. The fluorescent properties of QDs are highly tunable: by changing the size and composition of the quantum dots, the excitation and emission spectra can be easily controlled [113-115] in the visible wavelength range compatible with laboratory microscopes. QDs are generally much brighter and more photo-stable than organic fluorophores and fluorescent proteins, so they
29 can be imaged for long periods under high intensity illumination with minimal changes to their spectral properties [116]. They are commercially available with a number of different surface coatings that can be used for specifically targeting protein groups such as carboxyls, amines, and antibodies, and streptavidin. Thus, they have found many applications in fluorescence imaging [113, 115, 117]. QDs are often prolate ellipsoids with aspect ratios ranging from 1.1 to 1.2, rather than perfect spheres. These small deviations from symmetry result in polarization-dependent optical emission and absorption [118].
Compared to organic fluorophores and fluorescent proteins, however, QDs are much larger, 5 – 20 nm in typical applications, which can lead to limitations or steric constraints on activity and difficulty inserting them into live cells, and they can have cytotoxic effects
[119, 120].
Under controlled conditions quantum nanoparticles can be synthesized in a wide range of shapes and sizes including elongated prolate ellipsoids, termed quantum rods
(QRs) [121, 122]. As with QDs, the emission wavelength of QRs is tunable and they confer the same brightness and photostability advantages [75, 123]. The core and shell materials, various dot-in-rod or rod-in-rod configurations, and the aspect ratio are adjusted during synthesis [124]. High aspect ratio 5:1 – 20:1 QRs exhibit well polarized absorption and fluorescence emission [125], providing high signal-to-background ratio and high angular resolution in orientation experiments. Commercially-available nanorods are not as common as QDs, and thus may require both in-house synthesis and surface modification to make them soluble in aqueous media and biologically compatible [126].
Semiconductor nanoparticles can be functionalized either by covalently modifying the shell [127, 128] or by coating the surface with an amphiphilic polymer [129, 130].
30 These surfaces can then be modified with a targeting molecule such as streptavidin [131],
NeutrAvidin [128], or halo ligand [123, 127]. In either case, the large surface area of nanoparticles accommodates the attachment of many ligands to each particle. This allows rigid multi-site attachment to the biological macromolecule if multiple target sequences are inserted with appropriate spacing [123, 128].
Unlike many of the organic dyes discussed earlier, the excitation and emission dipoles of nanoparticles are not necessarily aligned, and the offset between them may depend on the excitation wavelength and probe dimensions [75, 124]. This trait reduces the facility of measuring nanoparticle orientation using both the absorption and emission dipoles. Thus, polarized fluorescence microscopy with QDs typically uses the emission dipole only. Polarized emission TIRF microscopy with QRs has been used to observe the rotations of the myosin tail domain [123].
1.2.3 Methods for measuring probe orientation
There are many methods that can be used to detect and measure the orientation of a polarized particle, and each one confers different advantages. Factors that should be considered when selecting a method include the spherical ambiguity (i.e. how many redundant angles there are within a sphere, two-fold redundancy results in hemisphere ambiguity and four-fold redundancy gives quarter-sphere ambiguity), desired time resolution, and if position tracking is needed.
One method to measure orientation using only probe emission was theoretically proposed by Fourkas, et al. in 2001 [132]. This technique could determine orientation in three dimensions with quarter-sphere ambiguity. The intensity of emission split into, at 31 minimum, its x, y, and 45°-xy polarized components would provide sufficient information to calculate 휃p and p probe angles independent of the excitation light. If a fourth emission polarization is included, i.e. 135°, the 휃p and p angles are over-determined and can be calculated by fitting the following equations to measured intensities [132, 133]:
2 2 퐼0(휃푝, 휙푝) = 퐼푡표푡(퐴 + 퐵 sin 휃푝 + 퐶 sin 휃푝 cos 2휙푝)
2 2 퐼45(휃푝, 휙푝) = 퐼푡표푡(퐴 + 퐵 sin 휃푝 + 퐶 sin 휃푝 sin 2휙푝)
2 2 퐼90(휃푝, 휙푝) = 퐼푡표푡(퐴 + 퐵 sin 휃푝 − 퐶 sin 휃푝 cos 2휙푝)
2 2 퐼135(휃푝, 휙푝) = 퐼푡표푡(퐴 + 퐵 sin 휃푝 − 퐶 sin 휃푝 sin 2휙푝)
Emission-only detection is preferred for QRs, because, as mentioned, their absorption and emission dipoles are not aligned [75]. This theoretical system was practically applied to measure myosin V tail domain rotations [123].
Other techniques that implement excitation polarization modulation in addition to polarized emission detection can reduce the spherical ambiguity to a hemisphere, the minimum for a dipolar probe, as well as provide additional information about probe motion on microsecond timescales [43, 72, 98, 103, 104, 107, 134]. Polarized imaging can also be applied to other forms of microscopy such as differential interference contrast (DIC) microscopy [135-137] or three-dimensional super-resolution microscopy using shaped point spread functions [91, 138, 139].
1.2.4 Further applications of polarized microscopy
Single molecule polarized microscopy has been used to answer questions about the function of biological macromolecules previously inaccessible by standard fluorescence
32 microscopy. It has been used to determine the direction of actin rotation by myosin II [134], observe the hand-over-had motions of myosin V [43], and examine the rotational pausing states of the F1-ATPase [94, 106, 140, 141]. Such experiments have increased our understanding of molecular motors such as myosin II [134], myosin V [43, 72, 123, 142], myosin VI [103], F1-ATPase [94, 140, 141, 143] and the ribosome [107]. They have been used to detect conformational changes and extract molecular mechanisms from motors translocating or catalyzing reactions in real time.
The work presented here aims to extend the use of polarized microscopy to the study of cytoplasmic dynein. This technique allows the observation of dynein structural states and changes in real time, giving insight into dynein’s stepping mechanism. It enables the study of many of the questions posed earlier, specifically how conformational changes of dynein contribute to force production, and the role of stalk compliance in dynein motility. The application of this technique required improvement of methods to attach polarized probes to proteins, discussed in Chapter 2, and development of novel microscopy and data analysis methods utilized in Chapter 3.
33 CHAPTER 2: FUNCTIONALIZATION OF CdSe/CdS NANORODS
Sections of this chapter, including some text and figures, are from our publication [128],
Lippert et al. Bioconj Chem (2016). Required permission was obtained.
With the ultimate goal of using polarized TIRF microscopy to study the motions of cytoplasmic dynein in real time we needed a polarized fluorescent probe with which we could rigidly and specifically label dynein. Bifunctional attachment of an organic dye typically requires generating a protein construct that contains no surface cysteine residues aside from those used for the probe linkage. This is prohibitively difficult in a protein as large as dynein. For this reason, we opted to use genetically encoded target sequences which would ensure high labelling specificity without mutating many points. Early experiments with organic dyes bound to SNAP tag insertions were unsuccessful as the linkage proved too flexible for accurate angular measurements. In order to reduce the motions of the probe relative to dynein we opted to use fluorescent quantum rods which could be surface modified to attach multiple inserted target sequences. By linking the probe via multiple sites within dynein we aimed to reduce probe motions with respect to dynein.
One problem we encountered was that existing coating methods for fluorescent nanorods did not provide a high enough surface density of ligand, in this case biotin binding proteins, to ensure multiple attachment points. So we endeavored to improve the surface coating methods as well as develop a reliable method to determine the amount of ligand bound to each nanorod surface.
34 2.1 Introduction
Semiconductor quantum dots (QDs) are fluorescent nanoparticles that are widely used in biochemical assays for labeling individual proteins both for in vivo imaging [144], and in vitro for high precision tracking [50, 104, 145]. QDs present some photophysical advantages over organic dyes because they are much brighter and do not photobleach over the timescale of typical fluorescence experiments [116]. Applications for fluorescent nanoparticles are broad since their emission wavelengths can be tuned simply by changing the diameter and composition of the typically CdSe or CdTe core [115, 146]. This tunability of the emission wavelength, paired with their broad excitation spectrum, makes
QDs ideal for multi-color imaging of biological molecules using a single excitation wavelength [114]. Quantum nanorods (QRs) are elongated semiconductor nanoparticles that share many features with QDs such as material composition and bright, stable fluorescence, but unlike nearly spherical QDs, QRs exhibit polarized fluorescence emission which can be utilized to determine their three-dimensional orientation [75, 123].
A high degree of polarization, >20:1, is achieved when the aspect ratio of length to width is greater than 10:1 [124]. Disadvantages of semiconducting nanoparticles are their larger size compared to visible organic fluorescent probes and fluctuations and blinking of their fluorescence. Adding a CdS [112] or ZnS [111] shell reduces blinking and increases the brightness of nanoparticles [115, 147]. The size of the QD-coating hybrid can be minimized by choice of coating used to solubilize and conjugate the nanoparticles to the target biological system.
Quantum dots are available with a range of surface coatings, facilitating specific labeling of proteins both in vivo and in vitro. Although water soluble, functionalized QDs 35 are readily available, commercial availability of coated quantum rods is limited.
Nanoparticles labeled with a biotin binding protein, such as streptavidin or NeutrAvidin, can be used to attach them to biotinylated proteins or nucleic acids [148]. Here we present several methods to functionalize CdSe/CdS QRs with NeutrAvidin that can be readily applied for use in single molecule polarized fluorescence assays. Nanoparticles that have been synthesized in organic solvent and coated with a hydrophobic ligand [149] are transferred to aqueous solution by exchanging the hydrophobic layer with a bifunctional ligand which contains a thiolate that binds to the particle surface and a polar carboxyl group that stabilizes the particles in aqueous media [150]. The carboxyl group can be covalently cross-linked to an amine-containing compound or protein, in this case the biotin binding protein NeutrAvidin. Fluorescence polarization is retained after functionalization.
Knowing the number of binding sites available on avidin-coated QDs and QRs can be important in designing experiments requiring attachment to multiple or known numbers of proteins. Here we describe an improved method to quantify the number of avidin proteins attached to individual nanorods or quantum dots and compare the number of binding sites obtained using different methods for ligand exchange. We also compare the degree of NeutrAvidin functionalization achieved on QRs to that of commercially available functionalized QDs of different sizes and surface treatments. The same materials (CdSe,
CdS and ZnS) are used to manufacture the QDs and commercial QRs, but their shapes and sizes are different which might affect the liganding chemistry. The comparable avidin protein density achieved indicates that the shape and size are not major determinants.
36 2.2 Determining the number of binding sites for streptavidin and NeutrAvidin
Biotin-4-fluorescein (B4F) binds tightly to streptavidin and NeutrAvidin. Its fluorescence is strongly quenched (>90%, Figure 7) when bound. B4F quenching was used to determine the concentration of NeutrAvidin or streptavidin free in solution [151-153] or conjugated to nanoparticles [154] at 5 to 60 nM concentrations of protein tetramer. To verify and calibrate the technique and as a basis for reliably determining the amount of avidin in solutions of functionalized nanoparticles, we first measured B4F quenching over a range of known NeutrAvidin and streptavidin concentrations (Materials and Methods).
While each tetramer contains four biotin binding sites, all four sites are not necessarily active and/or occupied simultaneously with B4F. Known concentrations of streptavidin and NeutrAvidin (based on absorbance at 280 nm) from 0 to 60 nM tetramers were combined with B4F at concentrations spanning 0 to 200 nM and the B4F fluorescence intensity was measured (Figure 8A). In the absence of protein the fluorescence increased linearly with increasing B4F concentration, but in the presence of streptavidin or
NeutrAvidin the fluorescence was quenched until B4F binding became saturated, at which point the fluorescence increased linearly with a slope similar to that of B4F alone.
37 A 1400000 150 nM B4F 1200000 Excitation 150 nM B4F 1000000 Emission 150 nM B4F + 800000 SAv Excitation
600000 150 nM B4F + SAv Emission
Fluorescence (a.u.) 400000
200000
0 320 420 520 620 Excitation or Emission Wavelength (nm)
B 100000 90000 150 nM B4F + SAv Excitation 80000 150 nM B4F + 70000 SAv Emission 60000 50000 40000
Fluorescence (a.u.) 30000 20000 10000 0 320 420 520 620 Excitation or Emission Wavelength (nm)
Figure 7: Absorption and emission spectra of B4F with streptavidin quenching A. Excitation and emission spectra of 150 nM B4F in the presence (quenched) or absence (unquenched) of 100 nM streptavidin tetramer. B4F fluorescence is quenched more than 90% upon binding to streptavidin. B. Quenched B4F excitation and emission spectra enlarged to show detail. Excitation scan fluorescence was detected at 535 nm, and emission scan fluorescence was excited at 485 nm.
38 A 45000 5 nM NeutrAvidin 40000 20 nM NeutrAvidin 40 nM NeutrAvidin 35000 60 nM NeutrAvidin 30000 Buffer
25000
20000 Fluorescence 15000
10000
5000
0 0 50 100 150 200 B4F Concentration (nM)
B 180 Streptavidin 160 NeutrAvidin 140
120
100
80
60 y = 3.46x + 3.08 40 B4F Concentration B4F Concentration (nM) y = 2.23x + 15.49 20
0 0 20 40 60 Tetramer Concentration (nM)
Figure 8: Quantification of B4F binding sites on streptavidin and NeutrAvidin A. B4F fluorescence vs. B4F concentration with NeutrAvidin present at concentrations listed in the legend. Quenching data for each NeutrAvidin concentration are fit with a curve and a straight line (solid lines; see Methods) to determine the B4F concentration, CI, at the intersection. The B4F series without NeutrAvidin (“Buffer”) is fit to a straight line only. Error bars are standard deviations.
B. CI vs. streptavidin (blue) and NeutrAvidin (red) tetramer concentrations with fitted lines given in the boxes. The slopes give the apparent number of B4F binding sites per streptavidin (3.46) or NeutrAvidin (2.23). Error bars are 95% confidence interval as determined by bootstrapping.
39 The effective concentration of biotin binding sites was determined from the point where the curve fitted to the data at low B4F concentration and the line fitted to the data at high B4F concentration intersect (Materials and Methods). The slopes of the curves in
Figure 8B give the number of biotin binding sites per streptavidin or NeutrAvidin tetramer.
Streptavidin binds an average of 3.46 B4F molecules per tetramer, close to the maximum occupancy of four, while NeutrAvidin binds 2.23 B4Fs per tetramer, close to the lower end of the range, 2.7 to 4.2 implied by the manufacturer’s instructions
(https://tools.lifetechnologies.com/content/sfs/manuals/MAN0011245_NeutrAvidin_Biot in_BindProtein_UG.pdf). Incomplete biotin binding site occupancy could be due to steric hindrance of B4F binding or reduced activity of the lyophilized protein after resuspension in aqueous solution.
2.3 Comparing coating methods for laboratory-made quantum nanorods
The NeutrAvidin functionalization of laboratory-made QRs surface coated using different methods was measured using the B4F quenching assay on samples of QRs at known concentrations. Free NeutrAvidin was carefully removed from QR samples using sequential pelleting by centrifugation and resuspension in fresh buffer until the amount of free NeutrAvidin present in the QR solution was less than one tetramer per 30 QRs, based on the number of times the buffer was exchanged. B4F quenching was measured in either
0.25 or 0.5 nM solutions of QRs and compared to the NeutrAvidin calibration curve to determine the concentration of biotin binding sites present in each QR sample (Figure 9A).
Poly(maleic anhydride-alt-1-octadecene) (PMAOD)-coated QRs bound an average of 63.1
40 NeutrAvidin tetramers per QR, glutathione (GSH)-coated QRs had an average of 30.8 tetramers per QR, and mercaptoundecanoicacid (MUA)-coated QRs had an average of 42.2 tetramers per QR (Figure 9B, summarized in Table II). Nanorods had average dimensions of 56.3 nm long and 5.6 nm in diameter as determined by TEM (Figure 10), giving an average surface area, calculated assuming a cylindrical shape, of 1040 nm2 per QR. On the basis of this surface area, the PMAOD-, GSH-, and MUA-QRs had NeutrAvidin surface densities of 0.061, 0.030, and 0.041 NeutrAvidins per nm2, respectively (Figure 9C, Table
II). To confirm that the quenching observed with the nanorod samples was due to the bound
NeutrAvidin and not the result of an interaction with the polymer coating, we compared
B4F quenching of PMAOD, GSH, and MUA QRs before and after the NeutrAvidin conjugation reaction. The results showed that QRs do not quench B4F prior to NeutrAvidin conjugation, so the quenching observed with the NeutrAvidin functionalized QRs is due to the NeutrAvidin.
41 A 40000 0.25 nM PMAOD-NAv 35000 0.5 nM GSH-NAv 0.5 nM MUA-NAv 30000 1 nM PMAOD 1 nm GSH 25000 4 nM MUA
20000
15000 Fluorescence (a.u.) 10000
5000
0 0 50 100 150 200 B4F Concentration (nM)
) 0.1 B 100 C 2 90 0.09 80 0.08 70 0.07 60 0.06 50 0.05 40 0.04
30 0.03 NeutrAvidin NeutrAvidin QR per 20 Surface per Area (1/nm 0.02 10 0.01 0 NAv 0
Figure 9: Measurement of NeutrAvidin content of laboratory-made QRs A. B4F and QRs with different surface treatments were combined at known concentrations but with unknown surface density of NeutrAvidin. Data are fit to a curve and a line as in Fig. 1 to determine the intersection, CI. B4F fluorescence was also measured in presence of surface-treated QRs without NeutrAvidin to verify that quenching is due solely to the NeutrAvidin. Error bars are standard deviations. B, C. Average numbers of NeutrAvidins per QR (B) and per unit surface area (C) as determined from the
CI in (A) and the effective numbers of B4F binding sites per tetramer from Figure 8. Error bars are 95% confidence interval as determined by bootstrapping.
42
Table II: NeutrAvidin quantification parameters for QRs with different surface treatments. Values denoted by + and – indicated the upper and lower bounds, respectively, of the 95% confidence interval determined by bootstrapping. The concentrations of biotin binding sites per QR absorption unit or per QR are listed as CI /OD and CI/[QR], respectively. The number of NeutrAvidin tetramers per QR
(NAv/QR) were calculated as NAv/QR = (CI - 15.49)/(2.23·[QR]), coefficients determined from the linear fit of CI vs. NeutrAvidin tetramer concentration shown in Figure 8, where CI and [QR] are both given in nM. NeutrAvidins per unit surface area is listed as NAv/QR divided by the surface area of individual QRs.
43
A
B
Figure 10: TEM images of laboratory-made QRs A, B. TEM of laboratory-made CdSe QR (A) core only or CdSe/CdS/ZnS core/shell/shell QRs (B). QR core and shell dimensions were used to calculate the extinction coefficient and the surface area.
44 Throughout the optimization of the coating and functionalization methods, a number of conditions were observed that decreased stability or increased the rate of aggregation of the nanoparticles. The ligand exchange reaction was sensitive to the starting organic solvent. Beginning with the QRs in THF improved the yield compared to performing the reaction in chloroform. However, storage of QRs in THF for more than a day resulted in a transition from a brilliant pink color to brown, indicating loss of fluorescence. In contrast, QRs are stable for months to years when stored in hexane. Adding potassium tert-butoxide (KBuOt) as a base for the reaction also improved the yield relative to adding KOH (Materials and Methods). Once in aqueous solution, QRs are prone to aggregation. Using avidin instead of NeutrAvidin (deglycosylated avidin) caused the nanoparticles to precipitate even in the absence of 1-ethyl-3-[3- dimethylaminopropyl]carbodiimide (EDC) crosslinker. Zwitterions, which have a net neutral charge, have been shown to increase the stability of nanoparticles [155, 156].
NeutrAvidin, which has lower isoelectric point than glycosylated avidin (isoelectric point
10.5), may have a stabilizing effect similar to a zwitterion and so is more effective than avidin at stabilizing the nanoparticles. Nanoparticle stability was also sensitive to the ratio of EDC to NeutrAvidin; increasing EDC concentration without increasing the NeutrAvidin concentration accelerated precipitation. Exchanging aqueous QRs into phosphate buffered saline (PBS) instead of borate buffer (pH 7.4 or pH 9) also precipitated them.
45 2.4 Quantifying streptavidin coating of commercial quantum dots
To compare the functionalized QRs with commercial streptavidin-coated QDs we applied the B4F quenching method to quantify the number of streptavidins coating various samples of quantum dots obtained from Life Technologies, Inc. Mittal, et al [154] also measured the streptavidin complement of QDs, but we consider our assay, the methods for estimating QD concentrations, and our estimate of B4F-streptavidin binding stoichiometry more reliable than theirs. In contrast to the earlier study, we measured concentrations of
QD stock solutions rather than assuming the listed concentration (extinction coefficients are listed in Table III), and we experimentally determined the average number of functional biotin binding sites per streptavidin tetramer instead of assuming the maximum of four.
We used a series of QDs coated via polyethylene glycol (PEG QD evaluation kit part
#Q10151MP) and a series of QDs coated using ITK, an amphiphilic polymer. As specified in the product literature, ITK quantum dots contained more biotin binding sites than the
PEG-coated quantum dots (Figure 11), so we used a lower range of B4F concentrations for the PEG QD measurements (Figure 12).
46
Table III: QD and QR extinction coefficients Comparison of QD extinction coefficients determined using the position of the lowest energy absorption peak (“Empirical Extinction Coefficient”) and extinction coefficients provided by Life Technologies (“Extinction Coefficient from Manufacturer”). Empirical extinction coefficients are listed as N/A if the lowest energy absorption peak is not clearly defined.
47 A 45000 4 nM PEG 525 40000 4 nM PEG 565 4 nM PEG 585 35000 4 nM PEG 605 30000 4.8 nM PEG 655 4.1 nM PEG 705 25000 Buffer 20000
15000 Fluorescence (a.u.) 10000
5000
0 0 20 40 B4F Concentration (nM)
B 45000 1 nM ITK 525 40000 1 nM ITK 545 1 nM ITK 565 35000 1 nM ITK 585 1 nM ITK 605 30000 1.6 nM ITK 655 25000 1 nM ITK 705 1 nM ITK 800 20000 Buffer
15000 Fluorescence (a.u.)
10000
5000
0 0 50 100 150 200 B4F Concentration (nM)
Figure 11: Measurement of streptavidin coating of commercial QDs A, B. B4F quenching of (A) PEG and (B) ITK QDs. B4F and QDs were combined at known concentrations and the B4F fluorescence was measured. CI values were determined as before. Error bars are standard deviations.
48 A 50000 45000 5 nM Streptavidin 20 nM Streptavidin 40000 40 nM Streptavidin 35000 Buffer 30000 25000 20000 15000
Fluorescence Fluorescence (a.u.) 10000 5000 0 0 50 100 150 200 B4F Concentration (nM)
50000 B 45000 1 nM Streptavidin 4 nM Streptavidin 40000 8 nM Streptavidin 35000 Buffer 30000 25000 20000 15000
Fluorescence (a.u.) 10000 5000 0 0 10 20 30 40 B4F Concentration (nM)
Figure 12: Quantification of B4F binding sites on streptavidin A, B. B4F fluorescence vs. B4F concentration with streptavidin present at concentrations listed in the legends. Conditions and procedures as in text and Figure 8. The measurements were performed at (A) high and (B) lower concentrations of B4F and streptavidin to optimize the assay sensitivity range for measuring streptavidin coating of ITK and PEG QDs, respectively (Figure 11A and B).
49 PEG QDs had an average number of streptavidins per quantum dot ranging from
0.30 to 1.4 (Figure 13A, results summarized in Table IV), whereas the ITK quantum dots had between 7.4 and 18 streptavidins per quantum dot (Figure 13B, summarized in Table
V). Carboxylated ITK 655 quantum dots (without streptavidin) did not quench B4F, demonstrating that quenching for the main series was due to the streptavidin. Tables IV and V give raw data as biotin binding sites (intersection between curves in the B4F assay) per OD of absorption at 350 nm to enable calculation of streptavidin content with alternate assumptions about QD extinction coefficients (e.g. values given by the manufacturer, which are generally higher than those estimated from the lowest energy absorption peak, resulting in lower concentration estimates).
QD shapes and sizes were estimated using TEM (Figure 14) to determine average surface areas, and streptavidin surface densities. Surface densities on PEG-QDs ranged from 0.0011 to 0.0083 streptavidins per nm2 (Figure 15A Table IV), while the densities on
ITK quantum dots were ~10-fold higher, 0.094 and 0.17 streptavidins per nm2 (Figure 15B,
Table V). Quantum dots increase in size with increasing emission wavelength and there was a trend for the number of streptavidins per QD to increase with QD size within a given type of coating, as expected (Figure 13). The streptavidin surface density was more constant (Figure 15).
50 A 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4
Streptavidin Streptavidin per QD 0.2 0
B 25
20
15
10 Streptavidin Streptavidin per QD 5
0
Figure 13: Number of streptavidins per commercial QD
A, B. Average number of streptavidin tetramers per (A) PEG or (B) ITK QD as determined by CI values from Figure 13 and the apparent number of B4F binding sites per streptavidin tetramer (3.46) from Figure 8. Error bars are 95% confidence interval as determined by bootstrapping.
51
Table IV: Summary of PEG QD streptavidin content Summary of streptavidin quantification parameters for PEG QDs of different sizes. Biotin binding site concentrations per OD and Biotin binding sites per QD were calculated as for Table II. Streptavidins per QD were calculated from the linear fit to [B4F] vs. streptavidin tetramer concentration in Figure 8:
SAv/QD = (CI - 3.08)/(3.46*[QD]). Values denoted by + and – indicate the upper and lower bounds, respectively, of the 95% confidence interval determined by bootstrapping.
52
Table V: Summary of ITK QD streptavidin content Summary of streptavidin quantification parameters for ITK QDs of different sizes. Calculations and confidence intervals as in Table IV.
53 ITK585 ITK605
ITK655 ITK705
Figure 14: TEM images of commercial QDs Example TEMs of 585, 605, 655 and 705 ITK QDs. Dimensions were used to estimate their surface areas. Scale bars are 20 μm.
54 A 0.012
0.01
0.008
0.006
0.004
0.002
0 Streptavidins Streptavidins Unit per Surface Area (1/nm2)
B 0.3
0.25
0.2
0.15
0.1
0.05
0 Streptavidins Streptavidins Unit per Surface Area (1/nm2)
Figure 15: Density of streptavidin coating on commercial QDs A, B. Number of streptavidin tetramers per unit surface area for (A) PEG and (B) ITK QDs as determined from the CI (Figure 11), B4F sites per streptavidin molecule (Figure 8), and TEM estimation of surface area (Figure 14). Error bars indicate the 95% confidence interval as determined by bootstrapping.
55 2.5 Discussion
We compared three different ligands in order to coat and water solubilize CdSe/CdS
QRs synthesized in organic solvent. The QRs were then functionalized with NeutrAvidin using EDC and NHS. All three methods produced QRs with NeutrAvidin coating density comparable to the streptavidin coating of commercial ITK QDs. Nanorods maintained their polarization properties even after coating with NeutrAvidin (Figure 16). While QRs coated using PMAOD had the most NeutrAvidin, as measured using B4F quenching, they did not bind biotinylated yeast cytoplasmic dynein in a single molecule binding assay (Figure
17C). GSH-QRs coated with NeutrAvidin also failed to bind to biotinylated dynein in the single molecule assay (Figure 17B). MUA-coated QDs bound to biotinylated GFP-tagged dynein at approximately 0.22 QDs per GFP (Figure 17A). However, binding of MUA QRs was lower than that of commercial ITK QDs which bound at ~0.8 QDs per GFP (Figure
17D). QRs prepared using the other two coating methods were never observed bound to dynein on axonemes. We can speculate that the reason for the surprisingly low binding of
PMAOD QRs is that the amphiphilic polymer shell increased the diameter of the rods enough to prevent binding to dynein.
B4F fluorescence quenching can be used to determine the concentration of biotin binding proteins in solution and attached to nanoparticles with high sensitivity and precision. At similar concentrations of NeutrAvidin and streptavidin, we found that
NeutrAvidin binds fewer B4F molecules per tetramer than streptavidin.
56
0.2
0.18
0.16
0.14
0.12
0.1
Anisotropy 0.08
0.06
0.04
0.02
0 NAv MUA Nanorods in 655 ITK Nanorods Nanorods Chloroform Qdots
Figure 16: Anisotropy of nanorods and quantum dots Fluorescence anisotropy of coated and uncoated quantum rods and 655 ITK quantum dots. The quantum rods remain highly polarized after coating, while the quantum dots are nearly completely unpolarized. Samples were excited with 450 nm light and the emission was detected at 650 nm.
57
A B
C D
Figure 17: Binding of NeutrAvidin QRs to cytoplasmic dynein A-D. TIRF images of biotinylated GFP yeast cytoplasmic dynein bound along axonemes in the absence of ATP labeled with (A) MUA quantum rods, (B) GSH quantum rods, (C) PMAOD quantum rods, or (D) 655 ITK quantum dots. Green is GFP fluorescence and magenta is nanoparticle fluorescence. Scale bars are 2 μm. Binding efficiency was calculated as the number QRs or QDs bound along axonemes divided by the number of GFPs.
58 The PMAOD-, GSH-, and MUA-coated QRs made in-house had more avidin tetramers per QR (30 to 60) than the commercial ITK QDs (7 to 22). The QRs are larger and when normalized to surface area, QRs exhibited an avidin surface density of roughly one third that of the ITK QDs and five-fold higher than the PEG QDs.
ITK quantum dots coated with amphiphilic polymer have more streptavidins per quantum dot than the PEG alternatives. As expected from the increase in size with wavelength, the number of streptavidins per QD tended to increase with emission wavelength and size. For a similar set of QDs obtained from Invitrogen, Inc. (now Life
Technologies, Inc.) as used here, Mittal and Bruchez [154] reported 40 to 80 B4F binding sites per ITK QD and 2 to 4 B4F sites per PEG QD (except 12 sites on 800 nm PEG QDs).
They concluded that the binding capacity did not change systematically with QD size.
Several earlier studies of streptavidin content of QDs are also listed by Mittal and Bruchez
[154]. Our values of 30 to 70 sites per ITK QD and 2 to 6 per PEG QD are similar overall, but we observed a substantial increase of content with size (Tables IV and V leading to approximately constant surface density on both types (Figure 15). This is logical, as we would expect that a surface modification reaction would depend on the amount of surface present rather than the number of individual particles, assuming that surface curvature does not significantly impact reaction rates. Different bases for quantifying B4F, streptavidin and QD concentrations and the number of B4F sites per streptavidin tetramer may be the cause of this apparent discrepancy. The largest difference is their use of the manufacturer’s nominal stock QD concentrations whereas we based the QD concentrations on measurements of the extinction coefficients where possible (Table III). Except for the trend with QD size, though, the two studies are comparable. The methods for coating and 59 functionalizing QRs described here and for quantifying avidin content should be applicable to other semiconductor nanocrystal reagents and shapes.
60 CHAPTER 3: MEASURING ROTATIONS OF THE DYNEIN RING
Sections of this chapter, including some text and figures, are from our publication Lippert et al. currently under review.
In addition to functionalization a second difference between nanoparticles and organic dyes previously used in polarization measurements is the misalignment of the absorption and emission dipoles [75]. Previous polarized TIRF measurements of myosin
V [43, 72, 142] and the ribosome [107] from our lab depended on rhodamine’s aligned absorption and emission dipoles to calculate the probe orientation using both excitation and emission polarization. The offset of the dipoles seriously complicates the probe angle calculations, so we opted to develop a novel polarized TIRF microscope that relies only on the emission dipole to calculate the probe orientation. The theory for this technique was proposed by Fourkas in 2001 [132] and was demonstrated in practical applications to myosin V by Ohmachi, et al. in 2012 [123]. To facilitate the data analysis, we developed custom software for tracking the polarized particles, detecting transition events and calculating probe angles (see Materials and Methods).
Equipped with NeutrAvidin functionalized polarized nanorods and an optical system capable of tracking them we were able to begin measurements of dynein angular motions. We opted to label the dynein ring because of the relative ease of mutating it compared to other regions within the motor protein. We selected AAA5 and AAA6 since they are the only ring domains without ATP binding activity and chose sites on the side of
61 the ring away from the linker domain so that rod binding would not interfere with motions of the linker.
3.1 Introduction
Dynein is a molecular motor that walks processively toward the minus end of microtubules (MTs) in an ATP-dependent manner [25, 27, 157]. Axonemal dyneins drive the motility of eukaryotic cilia and flagella, while cytoplasmic dynein is responsible for a wide range of functions within eukaryotic cells including the retrograde transport of cargo such as autophagosomes in neurons [21], alignment of the mitotic spindle [158, 159] and chromosome segregation during mitosis [160]. Disruption of dynein-mediated neuronal transport has been implicated in neurodegeneration [161], and mutations in dynein and dynein associated proteins can cause a range of diseases including lissencephaly [162] and
Charcot-Marie-Tooth disease type 2 (reviewed in [163]). Despite the importance of dynein function, the mechanism by which dynein walks along the MT is not yet well understood.
The motor domains of dynein are formed from six concatenated AAA domains, interrupted by a long anti-parallel coiled-coil stalk that emerges from AAA4 and terminates in the microtubule binding domain (MTBD) and the buttress that extends from AAA5 [26,
164]. Two motor domains form a dimer via their N-terminal tails [25, 27]. ATP binding and hydrolysis drives dynein mechano-chemistry; the binding of ATP to AAA1 induces a conformational change leading to dissociation of the MTBD from the MT. Hydrolysis on the dissociated head induces a primed conformation that then rebinds to the MT. The forward step, or power-stroke, is coupled to release of phosphate and ADP [30, 165, 166]. 62 The “linker domain”, which spans the face of the AAA ring and connects the tail to AAA1, undocks and moves across the face of the ring upon ATP binding to AAA1 [63] and has been shown to play an important role in the translocation mechanism [34, 167]. How this motion of the linker leads to cargo translation toward the minus end of the MT is unclear.
While high-resolution electron microscopy (EM) and x-ray experiments have provided detailed information about structural changes among the different nucleotide states of dynein, single molecule studies have provided dynamic information on motions along the MT. Dynein walks along MTs with a variable step size [50] using a combination of hand-over-hand and inchworm stepping [47, 48]. Step size and the likelihood of the leading or trailing head to step are regulated by head-head separation [47, 48].
Multiple models have been proposed to explain the mechanism of dynein stepping.
These models are based on structural studies as well as comparison to mechanisms of other transport motors such as myosin and kinesin. A classic power-stroke model suggests that, analogous to myosin V [43], the stalk rotates about a fixed point, the MTBD, and acts as a lever arm to produce translation of the ring and cargo (Figure 18A) [168]. EM studies provide some support for stalk rotation [1, 64, 169]. In contrast, other structural studies report that the stalk and ring domain remain in a fixed orientation relative to the MT throughout different nucleotide states [170]. In an alternate model to a power-stroke mechanism, the stalk and ring adopt a more fixed orientation relative to the MT and the linker acts as a lever to produce force by straightening and thereby pulling cargo forward
(Figure 18B) [64, 164, 171]. This model is sometimes referred to as a winch mechanism
[171, 172].
63 In order to elucidate the relative movements of the ring and stalk domains as dynein generates force along the MT, and thus to discriminate among different models describing the underlying molecular mechanisms, we used a single molecule method combining polarized TIRF (polTIRF) microscopy with nanometer localization to measure the three- dimensional orientation and position of the ring domain of Saccharomyces cerevisiae cytoplasmic dynein while walking along MTs in real time. Fluorescent semiconductor QRs exhibit polarized emission and were used to label the AAA ring at a fixed angle and track its position and orientation over time. We observed that the ring undergoes small, frequent angle changes with a mean of 8°. Surprisingly, angle changes were only weakly coupled to stepping. Angle changes occurred more than twice as often as steps along the MT track, suggesting unexpected flexibility of the dynein stalk. We used molecular dynamics (MD) simulations to further investigate the degree of flexibility within the stalk and hinge region connecting the MTBD. These data again emphasize the role of flexibility, and argue against a mechanism in which the working stroke results from the tilting of the MT-binding stalk.
Instead, these data support a flexible linker-lever model in which inter-head strain produces opposing torques in the two heads resulting in bending of the stalk and hinging at the
MTBD. Together with recent EM data [1, 65], our observations support a unique stepping mechanism for an essential cellular motor based on docking of the linker and flexing of the stalk.
64 A B Cargo Cargo
Cargo
Cargo
2 1 1, 2
C
N C
D E
zMT 1 2 α
6 3 4 5
β yMT
xMT
MTBD
65
Figure 18: Stepping models for cytoplasmic dynein A. Power-stroke stepping model. The ring and the stalk tilt with respect to the microtubule as a rigid lever arm resulting in translocation of both the cargo and the motor domain from position 1 to 2. In this model, the fluorescent probe used in this study (pink arrow) would translocate from position 1 (pre-powerstroke) to position 2 (post-powerstroke) and undergo a rotation of the same magnitude as the ring and stalk. B. Linker swing or winch stepping model. The ring and stalk maintain a relatively fixed orientation with respect to the microtubule. Translocation of the cargo is a result of linker (violet segment) straightening in a working stroke and docking to the ring. The probe undergoes little or no translocation or rotation during this conformational change. C. Domain organization of the 331 kDa tail-truncated and doubly biotinylated dynein construct. Biotinylation target sequences (light purple, “BioTag”) are inserted in AAA5 and AAA6 of the ring domain. D. Schematic of the heterodimeric dynein construct. Dynein is dimerized via short, complimentary DNA oligonucleotides covalently attached to SNAP tags at the dynein N-terminus. Heterodimeric constructs contain one doubly biotinylated head, as shown in C, and one “wild type” head which lacks the BioTags. E. Definition of probe orientation angles with respect to the microtubule frame of reference. The x axis points toward the direction of dynein motion (the minus end of the microtubule). The y axis is in the plane of the microscope slide. Angles are expressed in terms of α (green), the azimuthal angle of the probe around the microtubule, and β (red), the probe angle relative to the microtubule axis.
66 3.2 Results
3.2.1 Labeling dynein with polarized quantum nanorods for polTIRF measurements
To visualize ring rotations and translations with high resolution we tightly coupled polarized QRs to the dynein ring at specific locations. We used a well-characterized 331 kDa tail-truncated cytoplasmic dynein construct in which the dynein heads were either homodimerized with GST or heterodimerized by DNA oligonucleotides. These constructs were previously shown to exhibit velocities and run lengths similar to those of the full length molecule [47, 50]. The dynein ring was labeled by inserting biotinylation sites [148] in AAA5 and AAA6 (Figure 18C, D). Doubly biotinylated constructs showed ATP- dependent velocities similar to the constructs lacking the insertions (Figure 19A), indicating that insertion of the biotinylation sites does not disrupt motor activity.
NeutrAvidin coated QRs, which have well-polarized fluorescence emission, were attached bifunctionally to the two biotinylation sites using a biotin-avidin linkage [75, 128].
Bifunctional attachment ensures a fixed orientation of the QR relative to the dynein ring.
The inserted biotin target sequences did not disrupt processive motor activity, but decreased observed translocation velocities (Figure 19B).
67 A
B
Figure 19: Mutant dynein velocities with and without quantum nanorods A. Velocities of GST homodimeric dynein with (“mutant homodimer”) or without (“wild type”) biotinylation sites inserted in AAA5 and AAA6. Homodimeric constructs were used to measure the effect of the insertions on velocity because heterodimeric constructs containing one “wild type” head may be able to compensate for slower activity in the biotinylated head by dragging the mutated head [2]. BioTag insertions have only slight impact on motor velocity. B. Velocity of heterodimeric dynein with one doubly biotinylated head and one “wild type” head bound to QR. Dynein velocity is significantly reduced both in the heterodimer and the homodimer (not shown) by binding to a QR.
68
In order to measure tilting of the dynein ring domain concurrent with stepping we developed a method for polTIRF imaging using an EMCCD camera as a detector. This technique was modified from methods previously used to measure the tail rotations of myosin V [123]. Here, circularly polarized light excites the fluorescent sample, and the emitted fluorescence is split into four components, polarized at 0°, 45°, 90°, and 135° around the optical axis, and is projected, spatially separated, onto the camera detector
(Figure 20). The orientation of the probe emission dipole, parallel to the long axis of the quantum rod [75], is determined from the relative intensities of the four component polarized fluorescence intensities in each image [133]. Orientation is expressed in terms of angles and relative to the MT (Figure 18E). Position is measured at sub-pixel resolution by fitting a Gaussian distribution [42] to the total image of each QR, accurately combined from all four channels. We confirmed that this approach can appropriately detect orientation changes of the probes by imaging labeled dynein bound to axonemes on a rotating stage in the absence of ATP (Figure 21).
69
EMCCD
135°
SAM 45° OBJ WGP 90° Slit 0° DC
PBS 50:50 BS Laser QWP
Figure 20: Wide-field polarized TIRF and sub-pixel localization microscope Polarized fluorescent samples (SAM) are illuminated with 532 nm laser light (green rays) beyond the critical angle for total internal reflection through a microscope objective (OBJ). Excitation light is circularly polarized using a quarter wave plate (QWP). The emission fluorescence (red lines) is separated using a dichroic mirror (DC) and resolved into four different polarized components (0°, 45°, 90° and 135°) relative to the microscope x-axis using a 50:50 pellicle beamsplitter (BS) followed by either a polarizing beamsplitter (PBS) or a wire grid polarizer (WGP). The four component polarized emissions are imaged on an EMCCD camera. A slit at the primary image plane narrows the field of view to accommodate the four images onto the camera detector. Position and intensity are measured simultaneously for each point. Sub-pixel localization is achieved by fitting a Gaussian to the intensity from the four channels accurately combined using a custom ImageJ script.
70
A B 250 Theta θ 200 Phi 150
100 x
Probe Angle Probe Angle (deg) 50 φ 0 -40 -20 0 20 40 60 80 y z Stage Angle (deg)
C D 100 120 80 100 60 80 60 40 40 20 20 0 0
-20 Theta -20 Theta
Probe Angle Probe Angle (deg) Probe Angle (deg) -40 Phi -40 Phi -60 -60 -60 -40 -20 0 20 40 -60 -40 -20 0 20 40 60 80 Stage Angle (deg) Stage Angle (deg)
Figure 21: Probe angles are accurately measured using the polTIRF system QR-labeled dynein molecules bound to axonemes in the absence of nucleotide were physically rotated around the microscope optical axis using a rotational stage positioned on the polTIRF microscope. A. Schematic of probe angles expressed as θ (blue) and φ (red) with respect to the microscope stage in the x-y plane. B. - D. Examples of QRs bound to single dynein motors on an incrementally rotated stage. The azimuthal stage rotation was determined from the change in orientation of the dynein-labeled axoneme. As expected, the measured angle, φ, of the probe is directly proportional to changes in stage angle around the optical axis, with a slope of close to 1.0, while measured θ values relative to the optical axis are unchanged. Error bars are the standard deviations of the N = 1188 angles measured at each stage position.
71 3.2.2 The dynein ring tilts while stepping
QR-labeled dynein motors were tracked using polTIRF while moving processively along MTs. Position and orientation of single dynein motors were measured simultaneously (Figure 22). Consistent with previous work, single molecule tracking of dynein’s position along the MTs revealed forward and backward steps of variable size
(Figure 23A) [49, 50]. During motility, the dynein ring tilts both azimuthally around the
MT, in , and axially in the plane of the MT, in (Figure 22B). Change point analysis
[173] was used to objectively detect abrupt changes in or (or both) with 95% confidence. Contrary to expectations based on EM imaging in the plane of the dynein ring
[1, 64, 65, 169], rotational changes in and β were similar both in frequency and magnitude (Figure 23B, C), with <|Δα|> = 6.7° ± 0.17º [s.e.m.] and <|Δβ|> = 5.47° ± 0.1º
[s.e.m.]. The measured angle changes assume that the polarized probe is oriented in the plane of the dynein ring. To determine the possible effect of the probe binding to dynein out of the plane of the ring, we calculated the angular distributions if the probe had been randomly oriented. The adjusted values are <|Δα|> = 11.4º and <|Δβ|> = 9.95º. The actual magnitudes of the ring rotations are likely in between these adjusted values and the measured values listed above since both represent extreme cases. This adjustment does not affect our conclusions, so, for consistency, all further analysis assumes the probe is oriented in the plane of the ring.
72
Figure 22: Position and orientation of a single dynein motor walking along a microtubule A. Polarized fluorescence intensities measured at 0° (red), 45° (magenta), 90° (blue) and 135° (green) with respect to the microscope x axis and normalized for differential channel sensitivity. Dots represent the values measured in each camera frame, and bold lines are the same data averaged between identified state transitions or change points (see Methods). B. Probe angles α (green) and β (red) calculated from the intensities in A. Dots denote angles calculated from frame by frame intensities, while solid lines show the angles calculated from the intensities averaged between change points. C. Distance traveled along the path of the microtubule (blue) or the sideways deviations from that path (side steps, magenta) by the same dynein motor shown in A and B. Dots show the positions of the dynein measured in each frame, and solid lines are the positions averaged between steps identified as change points. Triangles mark the times of detected angle changes (in B) or steps along the microtubule path (in C). Correlated steps and angle changes are marked with solid red triangles, while uncorrelated angle changes (B) and steps (C) are marked with open triangles. Data were collected at 100 μM MgATP.
73 A B
C D
E F
Figure 23: Distributions of measured step sizes and angle changes Histograms of all step sizes and angle changes for all events measured at 100 μM ATP. A. Probability distribution of step sizes either correlated with an angle change (green) or not correlated with an angle change (“uncorrelated”, dark green). A step and angle change were considered correlated if they occurred within one frame (50 ms) of each other. Ncorrelated = 3,972, Nuncorrelated = 4,361. B. Probability difference between the step size distributions plotted in A, correlated step probability minus uncorrelated. C-E. Probability distributions of α (C, green) β (D, red) or total included (E, blue) angle changes either correlated with a step (“correlated”, darker lines) or not correlated with a step (“uncorrelated”, lighter lines). Ncorrelated = 3,972, Nuncorrelated = 15,010. F. Probability difference between total included angle change distributions plotted in E, correlated angle change probability minus uncorrelated.
74 Next, we calculated changes in the total included angle, defined as the spherical arc between the orientations before and after each step (Figure 23D). This analysis indicated that tilting events were more frequent than steps: on average 2.2 angle changes occur per step at 100 μM ATP (Table VI). Typical angle changes were small, 8.3º ± 0.1º [s.e.m.], with 95% of total included angle changes less than 21º. We measured the same values for dynein walking along axonemes instead of MT and found the total included angle to be only slightly larger: 9.13° ± 0.19°. These angle changes are markedly smaller than predicted by a power-stroke model (see Discussion). As a control we increased the threshold for angle change detection in our analysis. While this increased the mean angle change size slightly it did not alter the shape of the distribution nor the ratio of steps and angle changes. This suggests that our results are not contingent on hyper-sensitive angle change detection.
We compared angle changes that were correlated with steps along the MTs to those that were not correlated with steps. A step and an angle change were considered correlated if they occurred within one frame (50 ms) of each other as determined by change point analysis. Only 20.9% of angle changes were correlated with steps at 100 μM ATP; conversely 47.7% of steps were correlated with angle changes at the same ATP concentration (Table VI). This indicates that ~80% of the measured angle changes occur while the labeled head is bound to the MT. While some of these uncorrelated angle changes can be attributed to stepping of the unlabeled head, which would not be seen in the position trace, the majority appear to reflect the pronounced flexibility of the dynein motor domain.
75
Table VI: Step and angle change rates Step and angle change rates as determined by fitting single or double exponentials, respectively, to dwell time distributions. Numbers of angle changes per step are listed regardless of correlation. Percent of correlated steps or angle changes were calculated as numbers of correlated steps or angle changes divided by the total numbers of steps or angle changes. Since there were more angle changes than steps, the correlated events were a smaller fraction of all angle changes than they were of steps. Upper (+) and lower (-) 95% confidence intervals were determined by bootstrapping.
76 Angle changes that are correlated with steps tend to be slightly larger than those that are uncorrelated; this can be seen in a difference histogram as a depletion of the population of smaller angle changes and an increase in the population of larger angle changes (Figure 23E). Similarly, both forward and backward steps that are correlated with angle changes tend to be larger than uncorrelated steps (Figure 23F). Within the subset of events exhibiting both a step and an angle change, the magnitudes of steps and angle changes are loosely correlated with large variability in the size of angle changes (Figure
24A). In correlated events, the dynein ring tilts 0.22° (95% CI = 0.1874° - 0.2693°) per nm of step size along the axis of the MT, demonstrating a very slight positive correlation between step size and the magnitude of angle change (Figure 24A). The step-dependence of angle change can be used to estimate mechanical compliance of the MT binding stalk and the connector between the two rings (see Discussion). Together these data indicate that changing the relative position of the two heads, i.e. taking a step, tends to produce larger angle changes than those that occur when both heads are stably bound and that the magnitude of the angle change depends on the size of the step. However, the small differences between the two populations and the occurrence of many uncorrelated events suggests that steps and angle changes are only weakly coupled.
77 A 30
25
) ° 20
15
10 Beta Beta Angle Change ( 5
0 0 10 20 30 40 Step Size (nm)
B
Figure 24: Step size relationship and angular dwell time distribution at 100 μM ATP A. Magnitudes of beta angle changes correlated with forward steps as a function of step size. Data were binned in 2 nm increments and error bars are the standard deviation of the angle changes within each bin. A line was fitted to the unbinned data, and the 95% confidence interval of the fit (dotted lines) was determined by bootstrapping. The same data is re-plotted in Figure 6A. B. Angular dwell time distribution at 100 μM ATP. The unbinned angular dwell times were fit by a double exponential decay (A·k1exp(-k1t)
+ (1 - A)·k2exp(-k2t)), using maximum likelihood estimation [4] with a minimum acceptable dwell time of 150 ms (three camera frames). Data contributing to the two lowest histogram bars (light purple) were ignored. The dwell times were fit significantly better by the double exponential decay than by a single exponential, with rate constants of 6.38, +0.501, -0.453 and 1.76, +0.204, - 0.189 [95% CI] at 100 μM ATP (see also Table VI).
78 3.2.3 Tilting frequency shows little dependence on ATP concentration
If dynein undergoes a working stroke in which the stalk and ring tilt to produce the translocation (the classic power-stroke model), the dwell time distribution of angle changes should be tightly coupled to each step along the MT. We measured dynein stepping and tilting rates at a range of ATP concentrations from the absence of nucleotide up to 200 μM
ATP as well as at 100 μM ADP in the absence of ATP. Stepping dwell time distributions are fit well by single exponentials. Stepping rates are ATP dependent (Table VI), although even in the absence of ATP some steps are observed. The fraction of backward steps increases at low ATP concentrations accounting for the surprisingly high stepping rate in the absence of ATP despite velocities close to zero. This bidirectional stepping pattern in the absence of ATP may be indicative of flexible motions that are no longer biased in the forward direction by ATP.
Distributions of dwell times between angle changes were fit significantly better by double exponential decays than single exponentials (p<0.01), possibly indicating two distinct populations of stepping events (Table VI, Figure 24B) [4]. Surprisingly, both the faster and slower components of the dwell time distributions are only weakly dependent on nucleotide. While stepping and tilting rates are dependent on ATP, the ratio of steps to angle changes was constant across the nucleotide concentrations tested. Similarly, the fraction of total steps and angle changes that were correlated remained constant. These data suggest that ring rotations are only weakly coupled with the ATPase cycle and are therefore unlikely to be the result of a mechanism such as a power-stroke model in which rotational changes would be predicted to be tightly correlated with ATP hydrolysis and with stepping.
79
3.2.4 The dynein ring is constrained when dragging an inactive head
Ring rotations were also measured in the context of a heterodimer with one inactive or “dead” head (Figure 25A, B), since it has been shown that a dynein construct having only a single active head is still able to move processively as long as the second head retains the ability bind to the MT [2]. We used a heterodimeric construct in which the active head was doubly biotinylated and labeled with a QR, while the other, unlabeled head contained a P-loop (Walker A) K to A mutation in AAA1 which renders it unable to bind or hydrolyze
ATP [31]. This dead head construct has smaller step sizes and angle changes than QR labeled “wild-type” dynein (Figure 25C-F). Ring rotations are reduced from 9.13° ± 0.19° to 5.22° ± 0.16° [s.e.m.s] total included angle change when walking along axonemes. The dead head is presumably dragged behind the active one [174], apparently limiting the flexibility or range of motions of the active head. The processive motility of this QR- labeled dead head heterodimer, albeit at a slower velocity, indicates that the QR does not eliminate activity of the labeled head (Figure 25A, B).
80 A B
C D
E F
Figure 25: Steps and angle changes of dynein walking with one disabled head Steps and angle changes of a “dead head” heterodimeric construct measured at 100 μM MgATP on axonemes. A. Schematic of the dynein heterodimer construct used in this experiment. One head contains a lysine to alanine mutation at residue 1802, which renders it unable to bind ATP in AAA1, creating a dead head. The active head contains the two biotinylation sites as shown in Figure 1C. The two heads are dimerized using complimentary DNA oligonucleotides bound to their respective N-terminal SNAP tags. B. Distance traveled along the path of an axoneme (blue) or sideways deviations from the path (side steps, magenta) by a single dynein motor tracked using quantum rod fluorescence. Dots are the position measured in each camera frame, and solid lines are positions averaged between detected steps. C. Probability distribution of step sizes of a QR-labeled heterodimer paired with either a dead head mutant
(gray) or an active “wild type” head (blue) on axonemes. Ndead = 310, Nactive = 1,616. D. Probability difference of the data plotted in C, dead head construct steps minus active construct. E. Probability distribution of total included angle changes of a QR-labeled heterodimer paired with either a dead head mutant (gray) or an active “wild type” head (purple). Ndead = 646, Nactive = 3,107. F. Probability difference of the data plotted in E, dead head construct minus active head construct.
81 3.2.5 Angle changes reflect dynein flexibility
Forces from the dynein ring are applied to the MT through the -helical coiled-coil stalk and globular MTBD domain during translocation. Rotational motions of the ring thus depend on these forces and the mechanical compliance of the stalk and MTBD hinge. To examine the dynamic behavior of these structural features and their implications for dynein’s mechanical properties, an MD simulation was performed for a system comprising the dynein stalk and MTBD. While an earlier simulation study [175] calculated motions of the stalk and MTBD over 50 ns, we extended the timescale to 3.3 μs (among other differences, see Materials and Methods) to more thoroughly sample motions at thermal equilibrium. The mechanical characterizations based on simulation data considered 8.8 nm of the distal stalk coiled-coil region, approximately the flexible portion of the stalk that extends outward from the buttress, as well as the MTBD (residues 2918-3165). Movie S1 shows an animated rendering of the MD simulation.
The persistence length, Lp, of the dynein coiled-coil, calculated by tangent correlation analysis of conformations sampled in the MD simulation was estimated to be
250 nm (Figure 26). Considering the dynein stalk as a rigidly anchored cantilever, the mechanical stiffness at the free end for the first bending mode is given by s = 3 Lp kBT /
3 -21 Lc , where kBT = 4.28 x 10 J at the 310 K temperature of the MD simulation and Lc = 8.8 nm contour length. The corresponding stiffness of the stalk at the free end is 4.7 pN/nm.
82
Figure 26: Tangent correlation analysis based on MD simulations Persistence length was estimated by tangent correlation analysis, as described in Methods. The angle (δ) between tangent vectors taken along all possible arc lengths (s) of the dynein coiled-coil are compared with their analog in the average dynein structure observed in MD simulation. The negative inverse slope of the linear regression through the origin yields persistence length. Notably, the data in the plot do not form a perfectly straight line, suggesting that flexibility is not homogenous along the coiled-coil region of the dynein stalk.
83 Notably, the data in the tangent correlation plot (Figure 26) do not form a perfectly straight line, suggesting that the flexibility of the coiled-coil is not homogenous along its length. Frames from the MD simulation were aligned with respect to the heptad at the base of the stalk to evaluate lateral fluctuations of the distal tip of the stalk, measured as the joint between the stalk and MTBD (the hinge vertex). As suggested by tangent correlation analysis, the fluctuations are not symmetric (Figure 27A), but are somewhat higher in amplitude in the x-direction (parallel to the plane of the dynein ring, Figure 28A) than in the y-direction (perpendicular to the ring, Figures 27A and 28B). The standard deviations
2 2 and values of variance of these fluctuations are x = 0.97 nm,
0.73 nm,
84
A
4
2
0
Deflection (nm) Deflection -2
- y
-4 -4 -2 0 2 4 x-Deflection (nm)
B C 8 8 8 8
110 110-8 110-8 110 /Hz)
2 10 10
9 -99 -99 9 110 10110 10110 110
10 10 10 10 110 110-10 110-10 110 P P P P n2 1 n2 1 10 n2 2 10n2 2 PSD P A f PSD P A f nf2 0 1 c1 1 PSDPnf2 0A1fc11 nf2 0 2 c2 2 PSDPnf2 0A2fc22 11 -11 11 -11 11 11 110 10110 10110 110
12 -12 12 -12 12 12 110 10110 10110 110
13 13 13 13 110 110-13 110-13 110 6 7 6 8 7 9 8 10 9 11 10 11 6 7 6 8 7 9 8 10 9 11 10 11 110 11010 1 101610 1101710 1101810 1101910 110 10 11010 1 10 6 110 7 110110 8 110110 9 110110 10 110110 110 110 10 P P10 10 10 10 10 10 10P P P10 10 n2 0 nf2 0 Pn2 0Pnf2 0 n2 0 nf2 0 nf 0 Pn2 0Pnf2 0Pnf 0
FluctuationPower Density (nm Frequency (Hz) Frequency (Hz)
Figure 27: Asymmetry of stalk flexibility observed in MD simulations A. Fluctuations of the distal tip of the stalk (hinge vertex), measured based on alignment of the stalk to the heptad at its base (as described in Methods), show more flexibility in the x-direction (parallel to the ring domain) than in the y-direction (perpendicular to the ring domain). Positions from the simulation are plotted every 400 ps. B, C. Power density spectra of fluctuations at the hinge vertex in the x- (B) and y- (C) directions. Spectra calculated based on simulation data collected at 10 ps intervals, were smoothed by a 15-point moving average filter and plotted.
85
Figure 28: All-atom molecular dynamics simulations of the microtubule-binding stalk All-atom molecular dynamics simulations of the dynein stalk demonstrate its flexibility along the coiled- coil region and at the MTBD hinge. A. and B. Overlay of seven representative frames each from the ensemble of 330,000 collected over 3.3 ms, selected to illustrate 99% of the structural distribution of stalk positions in the plane of the ring (A) and perpendicular to the plane of the ring (B). Distributions are determined with respect to alignment of the stalk to the heptad at its base (as described in Methods) and capture the distal tip of the stalk (hinge vertex) at its point closest to zero displacement and three equally spaced displacements out to +/- 2.5 [s.d.] in each plane. C. and D. Individual frames from A and B, with displacement of the distal tip of the stalk (hinge vertex) indicated in nm. Images were rendered using VMD 1.9.2 [5].
86
Movie S1: Animated rendering of final 1 μs from MD simulation Model alignments and orientations correspond to those shown in Figure 5, illustrating motions parallel (left) and perpendicular (right) to the plane of the ring. Portions of the stalk and MTBD considered in the mechanical characterizations are depicted as a cartoon backbone trace in a glass-bubble surface, while solid surfaces represent the reference structure and are included for context. Also shown are the 60-bead coiled-coil CAT (yellow spheres) heptad at the base of the coil (used for alignment and marked at its terminals with black spheres), hinge vertex (large black sphere), coil-arm of the hinge (marked at its terminal with a black sphere), and an additional vector (black) drawn across the MTBD to indicate its orientation. Rendering shows the final 1 μs of stalk/MTBD dynamics at a 50-frame stride rate and +/-3 frame smoothing window. Movie was rendered using VMD 1.9.2 [5].
87 At the base of the coiled-coil, just below the stalk-buttress joint, the two component
α-helices line up with each other nearly parallel to the y-z plane, leading to high flexibility in the x-z plane, parallel to the ring (Figure 28A, C). Approximately half to two-thirds of the way farther along the stalk, the flat face of the coiled-coil ribbon is rotated ~90°, such that the two helices are parallel to the x-z plane, conferring flexibility in the y-z plane, perpendicular to the ring (Figure 28B, D). The composite superimposed views of frames from the simulation (Figure 28A, B) show these flexions; the majority of lateral fluctuations in the x-z plane stems from bending at the base of the coiled-coil, from which the extended stalk represents a longer lever; the majority of lateral fluctuations in the y-z plane stems from the bending farther along the stalk, following the ~90° rotation of the ribbon.
Apparent stiffness of the tip of the stalk was calculated based on lateral fluctuations of the distal end of the coiled-coil (stalk/MTBD hinge vertex) applying the fluctuation-
1 1 〈 2〉 1 〈 2〉 dissipation relation, 2푘퐵푇 = 2푠푥 푥 or 2푠푦 푦 , producing values of sx and sy of 4.56 and 8.05 pN/nm in the x- and y-directions respectively. These stiffness values are pertinent to considering tilting motions in relation to intermolecular forces between the rings. As the stiffness value computed from Lp more closely matches the stiffness estimated in the x- direction based on the rigid cantilever approach, this analysis indicates that overall stalk flexibility is dominated by fluctuations in the plane of the ring, which correspond roughly to the plane of the MT.
Similar analysis of the fluctuations in longitudinal twisting of the stalk leads to t
= 75.9 pN·nm per radian torsional stiffness. The corresponding torsional persistence length, Lpt = t·Lc / kBT = 156 nm, is comparable to the value, 100 nm, for a coiled coil 88 estimated by Wolgemuth and Sun [176] using a coarse-grained model of two inter-wound a-helices. Frames from the MD simulation were aligned with respect to a theoretically bound MT to characterize the compliance of the stalk/MTBD hinge. Rotational stiffness of the hinge in the plane of the MT (related to β angle) and azimuthally around the MT (related to angle) are hx = 41.5 and hy = 136.1 pN·nm per radian, respectively. The axis of rotation of the hinge is very close (within ~3º on average) to parallel to the MT axis, i.e. rotation of the stalk about that hinge is mainly in the plane of the MT. These values are used in the Discussion to estimate the forces in the linkers and segment connecting the rings related to tilting of the rings.
3.3 Discussion
Using our method of combined polarized TIRF and sub-pixel localization we determined the first three-dimensional orientation measurements of the dynein motor during active translocation in real time. We simultaneously measured the position and orientation of the AAA ring domain of S. cerevisiae cytoplasmic dynein and found that the ring tilts during processive stepping, enabling us to expand upon previously proposed stepping models. Rotations of the ring domain are quite small, frequent and only loosely coupled with stepping. These dynamic observations allow us to rule out a classic power- stroke mechanism of stepping because the measured angle changes are too small and uncorrelated. Instead our results suggest a flexible stalk model defined by frequent small angle changes that are only weakly correlated with stepping and are related to flexibility of the stalk and MTBD hinge as well as tension between the two heads. This is consistent
89 with the elegant EM studies looking at dynein bound to MTs in both primed and unprimed conformations. These structural approaches suggest only small differences in orientation of the ring relative to the MT between the two states [1, 65].
3.3.1 The power-stroke model of dynein stepping
A power-stroke model of stepping, like that of the myosin V motor lever arm, involves the tilting of the lever arm in the leading filament-bound head generating tension on the trailing head which swings forward during a step. Applying this type of model to dynein, when the trailing head detaches from the MT and moves forward to produce a step, the stalk and ring of the attached head tilt relative to the MT, possibly hinging at the MTBD
(Figure 29A). The biotinylation sites in our dynein construct are positioned approximately
28 nm from the MTBD. Based on this geometry, the simplest tilting stalk model would predict greater than 30º rotation of the ring in the plane of the MT (β) in order to produce a typical 16 nm step, or 25º of rotation to produce our observed mean forward step size of
12 nm. These predictions are in striking contrast to our observation that the 95% of β angle changes are less than 15º, with a mean of 5.4º.
90
40 A Δx B
35 Δz
) 30 ° Δβ 25 휃MT
20
15
Beta Beta Angle Change ( 10
5
0 0 10 20 30 40 Step Size (nm)
Correlated Data ThetaMT 20deg ThetaMT 45deg ThetaMT 60deg ThetaMT 75deg ThetaMT 90deg
3 2 1
Figure 29: Comparison of proposed dynein stepping mechanisms A. Magnitudes of beta angle changes correlated with forward steps as a function of step size (teal points). Magnitudes of correlated forward steps and b angle changes are gathered and averaged in 2 nm bins. Error bars are standard deviations of the angle changes in each bin. The data are fit by a line with a slope of 0.22° (95% CI = 0.1874° - 0.2693°) per nm of fluorophore (ring) translocation along the MT. Confidence intervals for the fitted line, determined by bootstrapping (Woody et al., 2016) are marked by dotted lines. Purple lines show the expected beta angle change (Δβ) as a function of step size predicted if dynein were to step using a rigid-stalk power stroke (see inset). The predicted shape of the step size and angle change relationship depends on the angle of the stalk relative to the microtubule (qMT, see inset), shown by different shades of purple lines with qMT increasing as the lines become darker. The data does not fit the power-stroke model, which would predict larger angle changes with magnitudes closely related to the step size. B. Flexible stalk model supported by observed small and frequent angle changes. Stalk flexing due to inter-head torsion causes small ring rotations both when the unlabeled head steps (probe position 1 to position 2), resulting in an angle change not correlated with a step, and when the labeled head steps (probe position 2 to position 3), resulting in an angle change correlated with a step.
91 An additional prediction of the tilting power-stroke model is that rotations of the ring are tightly coupled to stepping and the ATPase cycle. A translocation should be observed each time the ring tilts due to the long distance between the probe and the postulated hinge point of rotation at the MTBD junction near the MT. There should also be a direct correlation of step size and in-plane rotation (β) following the equation Δx = 2l
∙ sin(Δβ/2), where Δx is the step size, l is the distance from the point of rotation to the fluorescent probe, approximately 28 nm, and Δβ is the change in the β angle. This relationship is nearly linear for angle changes less than about 50°. At 100 μM ATP less than 21% of the observed angle changes occur simultaneously with a step, but even if we compare only the subset of angle change events that occur with steps, the size of steps and angle changes are only weakly correlated, and the relationship predicted by the stalk-stroke model (violet lines in Figure 29A markedly overestimates the measured rotations (green points).
Together these results provide strong evidence against a stepping mechanism for dynein in which force is produced exclusively by a tilting stroke of the MT-binding stalk.
Instead our data suggests a less strict mechanism where rotations of the ring are not tightly coupled to steps but are instead a product of dynein’s flexibility and inter-head tension.
Our results are consistent with EM data obtained from axonemal dynein, which suggest that the angle of the dynein stalk relative to the MT does not change markedly with nucleotide state in the head [65, 170].
92 3.3.2 The flexible stalk model of dynein stepping
In our proposed stepping model (Figure 29B) the stalk and ring remain at an almost fixed orientation relative to the MT while changes in nucleotide state regulate dynein’s affinity for the MT and cause conformational changes of the linker. Upon attachment to the MT and subsequent phosphate release from AAA1, the major working structural change of dynein, a straightening of the linker domain (Figure 29B middle), pulls the cargo at an acute angle toward the MT axis and toward the minus end. Movement of the linker thus changes the tension between the two heads and biases motion in the forward direction.
This mechanism is consistent with previous work showing that the size and magnitude of steps are related to the inter-head distance [47, 48] and that the stepping rate depends on the magnitude and direction of an applied force [49].
In a mechanism where linker straightening results in translocation, the ring and stalk remain at a relatively fixed orientation with respect to the MT [170, 171]. However, the model does not require that they remain completely fixed. Instead intramolecular forces between the two heads of the double-headed attached state are predicted to cause the small tilting motions observed in this study. There are two possible ways that the orientation of the ring domain could change in the context of a flexible stalk mechanism. The first is hinging at the MTBD. This possibility is supported by EM analysis of Dictyostelum cytoplasmic dynein which detected variable angles between the MT and dynein [1].
Hinging at the MTBD could result in orientation changes both in the plane of the MT (β) and around it (), although according to our results, the hinge and stalk flexibilities are greater in than in . This hinging is distinct from a stalk power-stroke mechanism in
93 which the tilting is the main cause of translocation and a requirement for stepping. In the flexible stalk mechanism, dynein can step without hinging.
3.3.3 Mechanics of dynein tilting
The mechanical characteristics of the stalk and stalk/MTBD hinge can be related to the tilting of the dynein ring measured from the polarized fluorescence intensities of the
QRs bound to the ring assuming the stalk emerges from the stalk-buttress joint at a fixed angle. Mechanical compliance (Cs) in the direction of the MT axis is given by 퐶푠 =
1 퐿2 sin2(휃) ( + 푐 ) = 0.93 nm·pN-1. Where is the angle of the stalk relative to the MT, 푠푥 ℎ푥
sx (pN/nm) is the bending stiffness of the stalk, and hx (pN·nm per radian) is the torsional stiffness of the stalk-MTBD hinge. Approximately 90% of this compliance is derived from the hinge and 10% of it is due to bending of the stalk.
When the two stalk heads are bound a total distance, dt, from each other along the
MT, force, Fc, in the elastic connection between the two rings (e.g. the linkers and dimerization domains), will pull the front head backward and the rear head forward (Figure
30B, F) by a distance dr = Fc·Cs = dc·kc·Cs, where dc and kc are the extension and stiffness of the ring-ring connection. Although there is high variance, the rotational angle change per unit step size given by Figure 29A is S = 0.22° (95% CI = 0.1874° - 0.2693°) per nm.
Ring rotation per nm of motion along the MT, due to stalk bending and tilting, is given by
1 푆∙퐿푐 sin(휃) dr = 1 / (Lc sin ()). Combining this with Cs gives 푘푐 = ∙ = 0.025 퐶푠 (1−푆∙퐿푐 sin(휃)) pN·nm-1. This value and the estimates of other mechanical parameters are very similar to
94 those estimated from ring tilting in electron micrographs by Imai et al. (2015), Table VII.
While Imai et al. do not discuss stalk bending, they do note that c is too small to correspond to the dimerized GST peptide or the hybridized DNA connecting the rings in our constructs, implying extra flexibility in the connection between the two rings, such as in the linker domains themselves.
At 8.3, 16.6 and 24.9 nm of separation, dt, between the MT binding sites, the deflection of the rings, dr, along the MT are calculated to be 0.18, 0.37 and 0.55 nm, respectively, dt = dc + 2·dr = Fc·(1/kc + 2·Cs), and corresponding step sizes measured at the ring, sr = dc, are 7.9, 15.9 and 23.8 nm, thereby broadening the distribution of measured step sizes. Intramolecular forces at the three values of dt are 0.2, 0.4 and 0.6 pN, well within the 3 – 5 pN force generating capability of the motor [2, 29, 40, 49].
The consequences of the linker-lever model of dynein stepping are pictured in
Figure 30. After a head binds the MT (blue leading head in Figure 30A), its linker domain straightens to pull the cargo and partner head forward, toward the minus end of the MT
(Figure 30B). Due to hinging at the stalk/MTBD hinge and due to cantilever bending of the stalk, the two rings tilt toward each other (panel B). This strain is relieved when the trailing head (pink in panel C) detaches. Re-priming of the linker position in the detached head swings it forward (D), although the detached head is likely to undergo considerable fluctuations, not depicted. Reattachment (E) is followed by the linker straightening in the new leading head (red in panel F) again tilting the two rings. Forward progress in this scheme is due to the linker-lever pulling toward the MT minus end and due to the re- priming motion of the detached head. The attachment can occur at any of several of the tubulin subunits. The rings tilt with each step regardless of whether the head labeled with
95 the QR or its partner is stepping. Only when the labeled head steps, however, is there a translocation along the MT, thereby providing an explanation for approximately twice as many tilting motions as steps detected in the experiments.
Overall, our results support a flexible stalk linker-lever model in which the amplitude of dynein ring and stalk rotational motions are fairly small once a step is complete and both dynein heads are bound to MT subunits. The duration of stepping is very short relative to the time spent in the two-head bound configuration, causing our orientation measurements to be dominated by the angles of the dynein rings while both of them are bound. The orientation changes we do observe are consistent with hinge tilting and stalk bending caused by the intermolecular force in the connecting domains and linkers which pull the trailing head forward and the leading head backward by 0.2 – 0.6 pN.
Previous experimental evidence for rotation of the rings was derived solely from static images obtained by EM [1, 170], whereas we provide dynamic measurements collected in real time during stepping. Nevertheless, the results from the earlier EM studies were consistent with the rather small angle changes we observed.
96
A Attached D Re-Priming
B E Linker Stroke Attached
C F Detachment Linker Stroke
Figure 30: Flexible stalk mechanism of dynein stepping The cartoons illustrate hypothetical steps in dynein walking. A. Two heads in the initial attached configuration. B. Conformational changes of the linker as it straightens from the primed to unprimed states causes an increase in inter-head tension. This tension pulls the two rings toward each other causing flexing of the stalk, bending at the microtubule binding domain, and tilting of the rings. C. Upon detachment of the trailing head (C, red) interhead tension is relieved biasing the detached head forward. D, Linker re-priming in the unbound (red) head provides more forward bias to the step, increasing the likelihood that it will bind ahead (E) of its previous position. F. The new leading (red) head undergoes its power stroke, tilting the heads toward each other again.
97
Table VII: Comparison of measured flexibility parameters to values from cryo-EM Comparison of the flexibility parameters determined here from molecular dynamics simulations and polarized TIRF microscopy (Present Work) to those calculated from cryo-EM images [1]. Δq/nm is the in-plane rotation of the ring per nanometer of forward motion, khx is the rotational stiffness of the stalk/stalk-head hinge in the plane of the microtubule, sr are the step translations of the ring along the MT for 1, 2, and 3 MT doublets of separation between the MTBDs, i.e. 8.3, 16.6 and 24.9 nm of separation, kc is the stiffness of the ring-ring connection, and Cs is the mechanical compliance of the stalk in the direction of the microtubule axis. The parameters are in good agreement.
98 As suggested by our analyses, the motion that produces force or steps forward is not a stalk-power-stroke (Figure 18A), but a consequence of straightening of the N- terminal linker region between the ring and the connecting dimerization domain (Figures
18 and 30). In that case, the flexibility we detected in the MTBD hinge and the shaft of the stalk become important in enabling attachment to the next MT site. Limited flexibility may produce a bias of attachment in the minus-end direction because the MTBD is oriented relative to the stalk at the acute angle required for forward binding, but nor rearward binding. The amount of thermal wobbling that occurs during the “search” for this site is unknown. In myosin V, the predominant evidence [72, 177] is that the detached head swivels freely and sweeps out a large orientational space until it encounters the next actin subunit. High speed angular measurements of the dynein ring and stalk will be required to elucidate the dynamics of thermal motions while dynein heads are detached during a step.
99 CHAPTER 4: CONCLUSIONS
In this work we developed a method for water-solubilizing CdSe/CdS quantum nanorods (QRs) and coating them with a biotin binding protein. Using fluorescence quenching of biotin-4-fluorescein (B4F) we were able to quantify the number of available biotin binding sites present on each QR. These methods can be applied to nanoparticles of varying shapes and sizes, making non-spherical nanoparticles a viable option for labeling biological macromolecules. QRs are very bright, leading to the possibility that they could be used for orientation measurements within cells if properly functionalized. Our described methods could, in theory, be applied to link other small proteins to the QR surface since the EDC reaction requires only the carboxyl groups on the water-soluble QR surface and amine groups on the desired protein ligand. It may be possible to functionalize QRs with nanobodies for a highly specific and brightly fluorescent polarized probe.
The B4F quenching assay is widely applicable to measure functionally available biotin binding sites both in solution or bound to nanoparticles. This can be particularly important in experiments such as in vitro motility assays in which streptavidin-coated quantum dots (QDs) are frequently used to label motor proteins. In assays where a single motor protein per QD is desired existing methods typically depend on calculations from protein dilutions and the fraction of motile QDs to verify that single motor levels have been achieved. If instead researchers used QDs with a low number of biotin binding sites per
QD (for example the PEG QDs tested in Chapter 2), single motor levels could be achieved with more confidence. Conversely, in experiments where the objective is multiple motors per QD, using QDs with many surface streptavidins is preferable. In either case a simple 100 B4F assay could be used to verify that the nanoparticle surface coating density is within the desired range.
We applied these QR coating methods to examine the ring rotations of cytoplasmic dynein while translocating along microtubules in real time. The NeutrAvidin-coated QRs were attached bifunctionally to two biotinyation sites inserted in the dynein ring domain at
AAA5 and AAA6. This marks the first dynamic study of dynein’s three-dimensional orientation. We observed small, frequent rotations of the dynein that were weakly correlated with steps. From this we concluded that dynein does not walk using a powerstroke mechanism because such a model dictates that steps and rotations would occur simultaneously with 1:1 correlation. Additionally, the magnitudes of the measured angle changes are too small to fit the predicted relationship with correlated step size. Instead we propose a flexible stalk model in which tension between the two dynein heads generated by linker rearrangement upon phosphate release results in flexing of the stalk and hinging at the microtubule binding domain. This mechanism is supported by molecular dynamics simulations which show that the stalk and hinge are highly flexible. Mechanical calculations of the stiffness are consistent with our observed angle changes. Together these data provide a new mechanism for dynein stepping and rule out some previously proposed mechanisms.
While the methods developed and used in this study, the work presented here does have limitations. This work utilized non-native dynein constructs in vitro and therefore only demonstrates a stepping mechanism for dynein in the absence of regulatory binding proteins and without the surroundings of the complex cellular environment. The presented
101 labeling technique would not be feasible for use in live cells without drastic modification due both to the size and cytotoxicity of the QRs as well as the presence of many biotinylated proteins within the cell. Additionally, the size of the QR does perturb dynein motions possibly due to restricting ring contractions between AAA5 and AAA6, and a single labeling site also only provides one point of reference within the motor. Bolstering this work with different fluorescent probes such as small organic dyes on other sites of the dynein motor will be beneficial and could provide information about how domains move with respect to one another. A dynein construct lacking exposed cysteine residues could be used to introduce a bifunctional organic probe at different sites within dynein, minimizing the impact of the probe on dynein function and possibly enabling labeling of other sites such as the linker or stalk.
Despite these limitations these methods allowed us to measure structural properties of the dynein motor that could not be obtained from static observations. They shed light on previously unknown aspects of the enigmatic dynein motor, providing quantitative analysis of its mechanical properties. Such measurements of dynein’s flexibility may be important in understanding the effects of cargo binding, auto-inhibition, or binding of associated proteins. These techniques enable us to ask a number of new questions that had been previously inaccessible by standard single molecule or structural methods. What impact does load have on the rotational motions, and does the increased drag due to cargo binding restrict these motions? Does binding of accessory proteins, such as Lis1, modulate the flexibility, and what effect does this have on dynein’s ability to navigate obstacles? Are there disease mutations that affect dynein’s flexibility, and what is their physiological impact? Further experiments could be used to address these questions. 102 Studies of the effects of adaptor and regulatory proteins on dynein would, for the most part, require a full-length mammalian dynein construct. Based on our conclusions that tension between the rings, communicated via the dimerization domain, applies torque and restricts dynein flexing, the full-length construct itself could be expected to have more flexibility than our tail-truncated constructs, seen as an increase in the magnitude or frequency of ring rotations. Addition of dynactin and an adaptor such as BICD2 may stabilize the tail and reduce flexibility as compared to the full length construct. These experiments could be used to identify the roles of adaptor proteins in increasing dynein processivity and relieving auto-inhibition. Lis1 could have a similarly stabilizing effect when binding to the ring domain, locking dynein in a strongly bound conformation.
Further experiments to examine the rotational motions of other dynein domains could be done using a bifunctional organic dye, such as tetramethyl rhodamine. With a small molecule dye it may be possible to insert it into previously inaccessible domains, such as the stalk, linker, or MTBD without substantial reduction in motor activity.
Comparing rotational motions of the stalk to those of the ring domain would provide information about the rigidity of the connection between the stalk and ring stabilized by the buttress. While we assume ring rotations to be representative of stalk motions, this may not necessarily be the case. Polarization measurements of the linker could provide direct evidence for the linker-stroke model. We expect that a probe positioned on the distal end of the linker, closer to the tail, would exhibit large rotations tightly coupled to the ATPase cycle, while a probe placed closer to the attachment site at AAA1 would experience less rotation due to its position proximal of the linker hinge. And finally, orientation measurements of the MTBD would provide dynamic information about the conformational 103 changes that lead to the change in affinity for the MT. Most likely the observed changes would depend strongly on where the probe was placed within the domain.
Polarized TIRF microscopy is a powerful tool with a wide range of applications, not only to provide insight into the roles of flexibility and rotational motions in dynein’s stepping mechanism and regulation, but in also other protein systems. It can provide critical information about how conformational changes occur in real time. The methods developed and used in this study demonstrate a way in which to bridge the gap between structure and function.
104 CHAPTER 5: MATERIALS AND METHODS
Sections of this chapter, including some text and figures, are from our publication [128],
Lippert et al. Bioconj Chem (2016), or from Lippert et al. currently under review at Cell.
Required permission was obtained.
5.1 Nanorod preparation and B4F assay methods
5.1.1 Water solubilization of nanorods
CdSe nanorod cores (14.6 × 5.3 nm) were synthesized according to published methods
[178], coated with and elongated shell of CdS [124] and a thin layer of ZnS in trioctylphosphine oxide (TOPO) [149] for a final size of 56.3 × 5.6 nm. The hydrophobic
TOPO ligand was either exchanged with hydrophilic ligands, MUA or GSH [127] or coated with PMAOD, and amphiphilic polymer that intercolates between the alkyl changes [129,
130]. Aqueous nanorods are stable for months at room temperature.
MUA Nanorods
. Add 10 mg MUA to 500 μL of ~4 μM nanorods in THF
. Heat to 60° C in a water bath
. Add 10 mg KBuOt to the nanorod solution and vortex
. Centrifuge at 14,000 × g for 10 minutes to precipitate nanorods
. Remove and discard the THF supernatant
. Resuspend nanorods in 1 mL dH2O and vortex
105 . Sonicate for 15 minutes and centrifuge at 3000 × g to remove aggregates
GSH Nanorods
. Dissolve 2 mg of GSH in 200 μL of dH2O
. Add 500 μL of ~4 μM nanorods in THF
. Heat to 60° C in a water bath
. Centrifuge at 14,000 × g to pellet nanorods
. Remove and discard the supernatant
. Resuspend nanorods in 1 mL of dH2O
. Add 5 mg of KBuOt to the nanorod solution
. Sonicate for 15 minutes and centrifuge at 3000 × g to remove aggregates
PMAOD Nanorods
. Dissolve 10 mg PMAOD in 1 mL of chloroform
. Add 1 mL of ~2 μM nanorods in chloroform
. Stir for 2 hours at room temperature
. Evaporate the chloroform under vacuum
. Resuspend rods in 2 mL of 50 mM sodium borate, pH 8.3
. Sonicate for 10 minutes at heat to 60° C for 10 minutes
. Centrifuge at 3000 × g for 10 minutes to remove aggregates
106 5.1.2 Nanorod NeutrAvidin functionalization
Aqueous nanorods coated with either MUA, GSH or PMAOD contain exposed carboxyl groups that can be covalently linked to amines using EDC, a zero-length crosslinker. NHS increases the reaction efficiency.
. Centrifuge nanorods at 62,000 × g at 4° C for 30 minutes to pellet nanorods and
remove excess surface ligand
. Resuspend rods in one tenth to one third of the original volume of 50 mM sodium
borate, pH 8.3
. Dissolve EDC and NHS together in water at concentrations of 1 mM and 5 mM,
respectively
. Combine 30 μL of buffer exchanged nanorods and 30 μL of EDC/NHS solution
. Incubate at room temperature for 5 minutes
. Dissolve NeutrAvidin at 10 mg/mL in 10 mM sodium borate, pH 7.4
. Add 30 μL of NeutrAvidin to the nanorod/EDC/NHS solution
. Incubate at room temperature for 5 minutes
. Continue incubation at 4° for 2 hours to overnight
Free NeutrAvidin was removed from the nanorod solution by sequentially centrifuging and resuspending the nanorods.
. Centrifuge 90 μL of NeutrAvidin nanorod solution at 35,000 × g and 4°C for 20
minutes to pellet nanorods
. Remove 80 μL of supernatant without disrupting the pellet
107 . Replace the solution with 80 μL of 50 mM sodium borate, pH 8.3 and resuspend
the pellet
. Repeat this for four centrifuge cycles
After the final cycle the nanorods can be resuspended in a smaller volume of buffer if a higher nanorod concentration is desired.
5.1.3 Determining Nanoparticle Concentrations
QDs conjugated to streptavidin using PEG and emitting fluorescence at 525, 565,
585, 605, 655 and 705 nm (termed PEG QDs) were purchased as an evaluation kit from
Life Technologies, part # Q10151MP. QDs conjugated to streptavidin using ITK and emitting fluorescence at 525, 545, 565, 585, 605, 655, 705 and 800 nm (termed ITK QDs), part #s Q10041MP, Q10091MP, Q10031MP, Q10011MP, Q10001MP, Q10021MP,
Q10061MP, and Q10071MP, respectively, were kindly donated to us by Life
Technologies, Inc. Measurements of the number of Streptavidin or NeutrAvidin molecules conjugated to the QDs or QRs depended on their estimated concentrations. Molar extinction coefficients (ε) for nanoparticles depend on their size, shape, and composition
[179]. Molar extinction coefficients of QDs as a function of their longest wavelength absorption peak have been well characterized [179], and this method was used to determine the concentrations of the commercial 525, 565, 585, and 605 QDs, each of which had a distinct absorption peak 10 - 25 nm below their quoted emission peak. The 655, 705 and
800 QDs, however, did not exhibit a distinguishable lowest energy absorption peak. For these QDs, an extinction coefficient of 1,700,000 M-1cm-1 at 550 nm, as provided by the
108 manufacturer, was used to determine concentration. Additional extinction coefficients provided by Life Technologies at other wavelengths are listed in table III for comparison.
In most cases, the spectral method for determining molar extinction resulted in somewhat higher estimated concentrations than those provided with the commercial samples.
Although the molar extinction coefficients of QDs have been calculated experimentally, no such calibrations are available for the more complex CdSe/CdS/ZnS- type core/shell/shell QRs as used in this study. Therefore, the extinction coefficient for
CdSe/CdS/ZnS core/shell/shell QRs was calculated by combining information on the sizes of the CdSe core and the CdS shell as determined by TEM imaging and the extinction coefficient of the individual components, and adding their contributions together (Figure
12). To determine the contribution of the CdSe nanorod core to the extinction coefficient, a literature calibration was used based upon TEM measurements of the nanorod size: at
350 nm, the absorption of the CdSe core scales with the volume,24 which was measured to be 3.22×10-21 cm3 on average, giving an extinction coefficient at 350 nm of 1.09×107 M-
1cm-1 for the CdSe core alone. No direct measurement for the extinction coefficient of CdS rods is currently available, but the wavelength-dependent linear extinction coefficient of
CdS (α(λ), in units of cm-1) can be estimated from the reported imaginary index of refraction k of 5.3 nm CdS QDs according to α(λ) = 4πk/λ [180]. Using this literature report of the value of k (0.389) at 350 nm, we obtained α(λ) = 1.40×105 cm-1. The linear extinction coefficient may be converted into a molar extinction coefficient (M-1 cm-1) if the volume
3 V of the material (e.g. CdS, in cm ) is known according to ε(λ) = NAVα(λ)/1000ln(10), in
-1 which NA is Avogadro’s number (mol ), the factor ln(10) converts the extinction coefficient from a base e exponential (standard for linear absorption) to a base 10 109 exponential common for molar extinction coefficients; and the factor of 1/1000 converts volume in cm3 to L [181]. Using TEM to calculate the volume of the total structure and subtracting the volume of the core, we obtained a volume of 1.145×10-20 cm3 and a molar extinction coefficient at 350 nm attributable to the CdS shell of 4.17×107 M-1cm-1. The extinction coefficients of the CdSe core and CdS shell can be added together resulting in
7 -1 -1 the extinction coefficient of the whole nanorod ε350(rod) = 5.26×10 M cm . The concentration of QRs in solution was determined using this extinction coefficient based upon the absorption measured at 350 nm. The amount of ZnS in the QRs and its absorption at 350 nm are both negligible and therefore contribution was not included.
5.1.4 Biotin-4-Fluorescein Quenching Assay
Powdered B4F (Invitrogen) was resuspended to an approximate concentration of
2.5 mg/mL, or ~3.9 mM in 30 mM sodium borate, pH 8.3 and filtered through a 0.2 μm syringe filter. Absorbance at 495 nm was used to determine the actual concentration of the stock solution using an extinction coefficient of 68,000 M-1cm-1 [154].
Streptavidin (Thermo Scientific) and NeutrAvidin (Thermo Scientific) were dissolved in 10 mM sodium borate, pH 7.4, and the concentrations were measured using the absorbance at 280 nm and extinction coefficients of 41,940 M-1cm-1 per monomer and
23,615 M-1cm-1 per monomer calculated from their amino acid sequences [182].
Fluorescence of the B4F was determined using a Tecan GENios plate fluorescence reader with 485 nm excitation and 535 nm emission. Solutions and cartridges for the plate reader were prepared in a 4° C cold room. 180 μL of each solution with known or unknown 110 avidin protein concentration was added to wells in a 96-well plate. 20 μL of B4F at a range of concentrations was added to each well. Final dye concentrations after mixing ranged from either 0 nM to 40 nM or 0 nM to 200 nM depending on the approximate concentration of avidin protein in the sample. The plates were incubated overnight at 4° C and measured the following morning in the plate reader.
5.1.5 Biotin-4-Fluorescein Data Analysis
Because biotin-avidin affinity is very high, the concentration, CI, of added biotin at which quenching saturates and fluorescence begins increasing linearly gives a good estimate of the concentration of binding sites on the avidin protein in the sample. Below
CI, fluorescence increased gradually as B4F increased according to F = [B4F] FSat / ([B4F]
+ Khalf), the non-linearity presumably due to mutual quenching of B4Fs in addition to quenching by the avidin [153], where FSat is the maximum fluorescence at high [B4F] and
KHalf is the half-saturating B4F concentration (Figure 10). Above CI, Fluorescence increased linearly according to F = S [B4F] + Int, where S is the slope, similar to that in the absence of any avidin protein, and Int is an intercept. The intersection between the quenched curve at low [B4F] and the unquenched line at high [B4F] was found by minimizing least squares fits of the curve and the linear functions fit to the quenched and unquenched regions, respectively. A MatLab routine successively tested partitioning the data between quenched and unquenched regions, fitting the two relations for each partition to the data and tabulating the resulting correlation coefficient, R2. For the partitioning with the highest R2 value, the [B4F] value at the intersection between the two curves was taken
111 to be CI. The chosen partitioning was also required to contain CI between quenched and unquenched B4F concentration regions. Data sets with fewer than three points in the linear regime were excluded due to unreliability of the fit. Confidence intervals for CI were determined by bootstrapping using the same fitting algorithm.
5.2 Yeast dynein and polarized TIRF methods
5.2.1 Yeast transformation buffers
5x TE Buffer 100 mM Tris 10 mM EDTA pH to 7.5 with HCl Sterile filter Store at room temp.
LiAcetate Solution 100 mM LiAcetate in 1x TE Sterile filter Store at room temp.
5.2.2 Yeast transformation
Prepare 2 mL of yeast peptone plus 2% glucose (YPD) media in a sterile culture tube. Inoculate with the yeast parent strain using a sterile stick. Shake overnight at 225 rpm at 30 C. In the morning prepare 10 mL of YPD in an autoclaved flask. Inoculate with 975
112 L of overnight culture from the previous day. Shake at 225 rpm at 30 C for about 5 hours.
Transfer culture into a 15 mL conical tube and centrifuge at 4000 × g for 5 min. at 25 C to pellet the cells. Aspirate off the supernatant and resuspend cells in 1 mL of Li-Acetate solution, gently pipeting to mix. Transfer the cell mixture to an Eppendorf tube. In a small, tabletop centrifuge, spin at maximum speed, room temperature, for 30 seconds to repellet the cells.
Prepare the following DNA mixture at room temperature
. 10 L phenol-chloroform extracted linear DNA sequence for insertion
. 10 L salmon sperm DNA (Invitrogen) heated at 95° C for 3 minutes
Aspirate off the supernatant from the repelleted cells and resuspend in ~80 L of Li-
Acetate solution, pipeting gently to mix. Add 20 L of DNA mixture to the resuspended cells. Add 700 L of 40% PEG in Li-Acetate to the cell/DNA mixture and mix by inverting the tube. Incubate the cell mixture at 30 C for 30 min. After incubation, add 100 L DMSO to the tube and mix by inverting. Heat shock cells in a 42 C water bath or heat block for
22 min. Pellet cells by centrifuging at maximum speed, room temp, for 30 sec. Aspirate off the supernatant. Resuspend cells in 400 L of autoclaved water, carefully stirring with the pipet to mix. For Ura3 insertion, plate cells on Ura- selection plates. For tag insertion
(replacing Ura3), plate on YPD. Add 200 L of cell mixture to each plate and spread cells using a sterile pipet tip. Allow plates to dry before inverting and incubating at 30 C overnight (for YPD plates) or for ~2 days (Ura- plates). For tag insertion, cells are replica plated onto 5-FOA (MP Biosciences) selection plates the next day. Dynein was expressed
113 using galactose over-expression. Cells were harvested, frozen in liquid nitrogen and stored at -80° C.
5.2.3 Dynein purification buffers
5x TEV Buffer 50 mM Tris-HCl 750 mM KCl 50% glycerol Adjust to pH 8.0 with KOH Store at room temp.
5x Lysis Buffer 150 mM Hepes 250 mM K-Acetate 10 mM Mg-Acetate 5 mM EGTA 50% glycerol Adjust to pH 7.4 with KOH Store at room temp.
4x Lysis Buffer 120 mM Tris-HCl, pH 7.4 200 mM K-Acetate 8 mM Mg-Acetate 4 mM EGTA 40% glycerol 4 mM DTT 114 0.4 mM ATP 2 mM PMSF in EtOH 0.2% triton X-100
1x Lysis Buffer 30 mM Tris-HCl, pH 7.4 50 mM K-Acetate 2 mM Mg-Acetate 1 mM EGTA 10% glycerol 1 mM DTT 0.1 mM ATP 0.5 mM PMSF in EtOH 0.05% triton X-100
Wash Buffer 30 mM Tris-HCl, pH 7.4 50 mM K-Acetate 2 mM Mg-Acetate 1 mM EGTA 10% glycerol 1 mM DTT 0.1 mM ATP 0.5 mM PMSF in EtOH 250 mM KCl 0.1% triton X-100
115 1x TEV Buffer 10 mM Tris-HCl, pH 8.0 150 mM KCl 10% glycerol 0.1% triton X-100 1 mM DTT 0.5 mM PMSF in EtOH
5.2.4 Dynein affinity purification
Recombinant dynein was purified using ZZ-tag affinity purification with IgG beads. Grind ~15 g of frozen cell paste using a KitchenAid coffee grinder cooled with liquid nitrogen. While heating in a 37° C water bath dissolve the ground yeast powder in
4× lysis buffer so that the final buffer concentration is 1×. Transfer the lysate to a cold ultracentrifuge tube and centrifuge at 209,406 × g for 30 min at 4 C. Transfer the supernatant to the cold 50 mL tube, avoiding the cloudy layer at the top of the sample.
Combine the recovered supernatant with 5-10 L of IgG (GE Biosciences) beads per mL of lysate equilibrated in 1x lysis buffer. Rock the sample at 4 C for 1 hour to allow the dynein to bind to the beads. Add the beads and cell lysate to a gravity filtration column either on ice or in a cold room. Once the lysate has flowed through do the following washes
. 2 x 3 mL 1x wash buffer
. 2 x 3 mL 1x TEV buffer
. Labeling step done here when appropriate
. 5 mL 1x wash buffer
116 . 3 x 3 mL 1x TEV buffer
Add 1 L of 1x TEV buffer per 1 L of beads to the column capped gravity column, transfer to a small tube. Rinse the column with additional 1x TEV recover beads as needed.
Let the beads settle, then aspirate off the bubbles and extra buffer, reducing the total volume to twice the bead volume. On ice add 1 L of TEV protease per 50 L of beads and mix.
Incubate at 16 C for 1 hour. Mix every ~10 min by gently inverting the tube. Transfer the contents of the tube to the cooled centrifugal filter tube (Millipore). Centrifuge in a room temperature tabletop centrifuge for 30 seconds at maximum speed. The flow through is the purified dynein product. Aliquot into 3 L aliquots in pre-cooled tubes and flash-freeze in liquid nitrogen. Store purified dynein in liquid nitrogen.
5.2.5 DNA-oligo labeling of SNAP-tagged dynein heterodimers
This reaction attaches SNAP substrate to the end of a 3’ or 5’ amino-labeled oligonucleotide (IDT or Invitrogen). This method has been used previously by the Reck-
Peterson lab [47].
5’ amino-modified primer sequence: GGT AGA GTG GTA AGT AGT GAA
3’ amino-modified primer sequence: TTC ACT ACT TAC CAC TCT ACC
Dissolve SNAP-NHS reagent in anhydrous DMSO to 20 mM. Store dissolved
SNAP ligand at -20 C, sealed with desiccant. Combine 4 L 2mM amine oligo (3’ or 5’ amine), 8 L 200 mM HEPES, pH 8.5 and 12 L 20 mM SNAP-NHS in DMSO. Incubate
117 at room temperature for 30 minutes. Use MicroSpin6 columns (BioRad) to remove the unreacted SNAP reagent. Store the final product at -20°C.
During the labeling step of the dynein purification combine 30 L SNAP oligonucleotide and 70 L of TEV buffer. Add 100 L of this mixture to the capped gravity filtration column containing the dynein and IgG beads. Mix to suspend the beads, and incubate at room temperature for 15 min. Mix the beads again and incubate for another 15 min. Continue with the purification washes.
5.2.6 Single molecule assay buffer
BRB80 80 mM PIPES, pH 6.8 1 mM MgCl2 1 mM EGTA
Dynein Lysis Buffer (DLB) 30 mM hepes, pH 7.4 2 mM magnesium acetate 1 mM EGTA 10% glycerol
DLB-Casein 30 mM hepes, pH 7.4 2 mM magnesium acetate 1 mM EGTA 10% glycerol 118 1 mM DTT 1.25 mg/mL casein
Dynein Motility Mix 30 mM hepes, pH 7.4 2 mM magnesium acetate 1 mM EGTA 10% glycerol 1.25 mg/mL casein 100 μM ATP (unless specified) 300 μg/mL glucose oxidase (Sigma) 120 μg/mL catalase (Sigma) 0.4% glucose
5.2.7 Flow cell construction for single molecule assays
Flow cells were constructed following previously established methods [183]. A flow channel was assembled by adhering a glass slide to a glass coverslip using double sided tape (Scotch 3M). Single molecule TIRF assays were performed using either microtubules or axonemes. For microtubule assays flow cells were constructed with silanized coverslips and all solutions denoted by “+ taxol” contain 20 μM taxol
(Cytoskeleton, Inc.). Steps surrounded by parenthesis are only included when labeling with nanorods.
For axoneme experiments solutions were added the flow cell as follows:
. 13 μL 1:20 axonemes in BRB80
. Wait 2 minutes
119 . 2 × 20 μL BRB80
. 2 × 20 μL DLB-casein
. Wait 5 minutes
. 20 μL ~100 pM dynein in DLB-casein
. Wait 2 minutes
. (20 μL ~20 nM nanorods in DLB-casein)
. (Wait 5 minutes)
. (4 × 20 μL DLB-casein)
. 20 μL dynein motility mix
For microtubule experiments solutions were added to the flow cell as follows:
. 13 μL 1:20 anti-tubulin antibody (Sigma) in BRB80
. Wait 5 minutes
. 20 μL DLB-casein + taxol
. Wait 5 minutes
. 20 μL 1:250 microtubules in BRB80 + taxol
. Wait 5 minutes
. 2 × 20 μL DLB-casein + taxol
. Wait 2 minutes
. (20 μL ~20 nM nanorods in DLB-casein + taxol)
. (Wait 5 minutes)
. (4 × 20 μL DLB-casein + taxol)
. 20 μL dynein motility mix
120 5.2.8 Polarized TIRF imaging of dynein flow cells
Polarized TIRF imaging was done on a Nikon TI-E TIRF microscope modified to allow direct laser illumination. Samples were illuminated with a 532 nm laser
(CrystaLaser) through custom optics. Emission illumination was divided into polarized components using a home-built optical system (Figure 7A). The fluorescence emission beam was split independent of polarization by a pellicle beamsplitter (ThorLabs). The two resulting beams were split into 0° and 90° or 45° and 135° analyzer components using a polarizing beamsplitter cube (ThorLabs) and a wire grid polarizer (ThorLabs), respectively. Polarized beam paths were directed onto a Photometrics Evolve EMCCD camera spatially separated so that the beams did not overlap.
Precise alignment of the emission analyzers was determined by placing a polarizer on the microscope stage at known orientations in 5° increments () and illuminating with unpolarized light. Intensity curves were fit with sin2() functions to determine the orientation, φ, of the analyzers from the fitted curve for each channel. Relative channel sensitivities were calibrated daily by imaging an isotropic solution of nanorods to determine the relative intensity of an unpolarized sample in each channel (Iφ). A sample of water was imaged to determine the background (Bφ). Channel sensitivity (Cφ,) was calculated relative to the 0° channel: Cφ = (Iφ – Bφ)/(I0° - B0°). C0° = 1 by definition.
5.2.9 Angle fitting and changepoint analysis
Data analysis was performed using custom MATLAB scripts. Angles θ and φ were fit using polarized intensities from the four imaging channels using equations modified 121 from Fourkas, et al. [132] and given in Supplemental Methods. The equations were adjusted to account for the imperfect alignment of the emission analyzers and cross-talk.
Angles were transformed into the MT coordinate frame, α and β (Figure 20E), using the orientation of the microtubule determined from the start and end positions of dynein movement tracks.
Three dimensional orientation was over-determined by fitting θ and φ (Figure 23) angles to the four following intensity equations modified from Fourkas, et al [132]
2 2 퐼1 = 퐼푡표푡[퐴 + 퐵(sin 휃) + 퐶(sin 휃 cos(2휑 − 휓1))]
2 2 퐼2 = 퐼푡표푡[퐴 + 퐵(sin 휃) + 퐶(sin 휃 cos(2휑 − 휓2))]
2 2 퐼3 = 퐼푡표푡[퐴 + 퐵(sin 휃) + 퐶(sin 휃 cos(2휑 − 휓3))]
2 2 퐼4 = 퐼푡표푡[퐴 + 퐵(sin 휃) + 퐶(sin 휃 cos(2휑 − 휓4))]
Where Ij are the corrected and normalized intensities measured in channel j, ψj is the analyzer alignment of channel j, and Itot is a scaling factor related to the total intensity of the probe emission. 휓1−4 = -2.3º, 36.4º, 90.3º, and 142.4º, respectively. A, B and C are constants incorporating the numerical aperture of the objective (NA) and the refractive index of the glass (n), as listed in Fourkas, et al [132].
푁퐴 훼 = sin−1 ( ) 푛
1 퐴 = (2 − 3 cos 훼 + 6 cos3 훼) 12 1 퐵 = (cos 훼 − cos3 훼) 8 1 퐶 = (7 − 3 cos 훼 − 3 cos2 훼 − cos3 훼) 48
122 Angular state and positional transitions (steps) were identified using modified four channel [173] and single channel [184] change point analysis, respectively. Intensities are then averaged between the identified change points. The nanorod orientations are displayed using angles calculated from averaged (horizontal lines) and unaveraged intensities (noisier angle points in Figure 24). In the case of the angular change point detection, the likelihood of an intensity change occurring at each frame is calculated. The likelihood function takes account of the intensity, signal-to-noise ratio, and time between flanking change-points.
Change points are recorded if the sum of the log likelihoods for each polarized intensity channel exceed a threshold corresponding to the 95% confidence interval with equal numbers of false positives and false negatives as determined from simulated data [173].
To verify that the change point detection method was not overly sensitive and thereby filling the angle change distributions with small, false positive events, we increased the threshold for change point detection so that 35% of the previously accepted angle change events were discarded. Using the same thresholding criteria, 31% of the previously detected stepping events were discarded. While the magnitude of the angle changes increased slightly in this more restrictive analysis, the shape of the angular distributions remained the same as well as the minor difference in magnitude between events correlated or not correlated with steps. The average angle change in experiments on individual MTs increased only slightly from 8.82º ± 0.13º to 10.45º ± 0.17º [s.e.m.s]. All of these results were similar when dynein was translocating along individual MTs vs. sea urchin axoneme
MT bundles.
123 5.2.10 Correction for unknown probe angle
QRs giving stable angle values mostly bound to both biotin sites on the dynein ring as evidenced by much more variable and fluctuating angles observed in earlier experiments with QRs having less NeutrAvidin on their surface [128]. Nevertheless, the exact angle of each QR relative to the ring is unknown. We calculated the effect that variation of the binding orientation of the QRs relative to the dynein ring would have on measured β angles using a custom Mathcad program. To obtain a maximum limit of this possible effect, a worst-case scenario was assumed of a completely random binding orientation where the probability of the local angle is Pprobe = sin(λ) where λ is the angle between the long axis of the QR and the plane of the dynein ring. Angle changes were converted angular distributions based on the probability of the QR being at a given orientation using the
-1 2 equation βactual = cos ((cos(βmeasured) - 1)/(cos(λ) ) + 1), where βmeasured is the experimentally observed β angle change of the QR during a step and βactual is the real β angle change of the dynein ring that would have produced βmeasured for a given . In the extreme example, if
= 90º, in-plane ring rotation would not be registered as any βmeasured. However the likelihood of high angles diminishes as sin(). This correction for unknown local QR orientation increases the mean β angle change in events correlated with stepping from 5.47° to 9.95° and does not significantly impact our conclusions.
5.2.11 Molecular dynamics simulations
An all-atom model comprising the dynein stalk and MTBD was prepared based on the crystal structure of human cytoplasmic dynein-2 in the primed conformation [63]. The
124 model encompassed residues 2847-3234 and included two short -helices connected via loops on either side of the stalk/ring interface in AAA4 to serve as positional anchors for the dynein fragment. Missing side chains in the stalk were added with SCWRL4 [185].
Assessment using SOCKET 3.03 [186] confirmed that stalk residues CC1:2918-2977 and
CC2:3106-3165 were packed according to the knobs-in-holes arrangement characteristic of a coiled-coil, with coil registration in the β+ state as appropriate for a primed dynein stalk [187]. Protonation states and hydrogen coordinates were assigned to the model using the propKa 3.0 [188] option of PDB2PQR 2.0.0 [189] for an environmental pH of 7. The
CIonize plugin of VMD 1.9.2 [5] was used to place Na+ and Cl− ions around the model according to the local electrostatic potential. The model was then suspended in an 18.0 x
21.6 x 11.0 nm box of explicit water molecules with sufficient additional Na+ and Cl− ions to produce charge neutrality at 150 mM NaCl. The solvated dynein stalk/MTBD system,
253,000 atoms total, was parameterized with the CHARMM36 [190, 191] force field and the CHARMM TIP3P water model [192]. MD simulations were performed with NAMD
2.10 [193] as described in SI, producing a final production trajectory totaling 3.3 μs.
To prepare the all-atom model for simulation, a steepest decent energy minimization protocol (5,000 cycles) was applied first to the solvent, then to the solvent and protein side chains. Upon initiating dynamics, the simulated temperature was gradually increased from 60 K to 310 K over an interval of 5 ns, applying Cartesian restraints of 5 kcal/mol to the protein backbone. Continuing under isothermal, isobaric conditions, backbone restraints (except those on the anchor helices, residues 2847-2865 and 3219-
3234) were gradually removed over an additional interval of 5 ns. A subsequent production simulation was performed for a timescale of 3.3 μs, saving frames every 10 ps.
125 Temperature regulation during the simulation was performed with the Langevin thermostat algorithm in NAMD, employing a damping coefficient of 1 ps−1. The Nosé-
Hoover Langevin piston control was applied to maintain constant pressure of 1 bar, allowing isotropic cell scaling, with piston oscillation period of 200 fs and damping timescale of 100 fs. All covalent bonds containing hydrogen were constrained, allowing a simulation time step of 2 fs. Long-range electrostatics were split from short-range at a cutoff of 1.2 nm according to a quintic polynomial splitting function and computed with the particle-mesh Ewald method, as implemented in NAMD. Full electrostatic evaluations were performed every two time steps.
A scatter plot of tip deflections in the x- and y- directions (parallel and perpendicular to the AAA ring, respectively) shows higher amplitude along x, consistent with lower stiffness, 4.56 pN/nm, in the x-direction than in y, 8.05 pN/nm, as discussed in the text. Power density spectra of the fluctuations at the stalk/MTBD hinge vertex from the
̅̅̅̅̅̅̅ MD simulation (푃푥(푓) = ℱ(푥) ∙ ℱ(푥), and corresponding 푃푦(푓), where f = frequency,
ℱ(푥) is the Fourier transform of the fluctuations, and ℱ̅̅̅(̅̅푥̅̅) is its complex conjugate, Figure
29B, C) show critical roll-off frequencies of 25.3 and 32.2 MHz corresponding to time constants of 6.29 and 4.24 ns in the x- and y- directions. The damping time constant
1 expected for the stalk/MTBD model is approximately 휏푠 = (훽푠⁄20 + 훽푠ℎ), where s is 푠 the viscous drag of the 2 nm diameter, 8.8 nm long stalk moving through water
-11 perpendicular to its axis, ~3.7 x 10 N·s/m [194, 195], sh is the viscous drag of the MTBD
-11 (considered as a sphere, 3.5 nm in diameter), 2.9 x 10 N·s/m and s is the x- or y- stalk bending stiffness value listed above. Expected sx and sy are then 6.8 ns and 3.9 ns (Figure
126 29C), reasonably close to the time constants (6.29 and 4.24 ns) given by the critical frequencies of the fluctuation power spectra in the MD simulations. This calculation strongly supports the validity of the calculated bending stiffness.
Fluctuations of a body tethered on a linear spring in viscous Newtonian fluid should display a Lorentzian power spectrum which declines at high frequencies 102-fold per decade increase of frequency. The power spectra decline much slower at high frequencies
(~101.4-fold per decade frequency increase, Figure 29B, C) This sub-diffusive characteristic is presumably due to local fluctuations along the shaft and other subtle motions arising from the additional complexity of an all-atom structure beyond a simplified, lumped mechanical model.
5.2.12 Mechanical characterizations
Persistence length of the dynein stalk coiled-coil (CC1:2918-2977 and CC2:3106-
3165) was calculated from simulation data using the so-called “dynamic” tangent correlation method, as similarly applied to estimate bending flexibility in the tropomyosin coiled-coil [196, 197]. The coil structure was reduced to a per-residue beaded trace along its central axis by applying the TWISTER algorithm [198]. Tangent vectors were approximated over spans of 15 beads along the central axis trace (CAT, see Figures 30 and
33), leading to 46 tangent correlation measurements over the internal portion of the coil.
The cosine of the angle between tangent vectors and their analogs in the average structure, cos(), was averaged for all possible arc lengths, s, over the CAT for each frame of the simulation trajectory, then over the entire trajectory ensemble of 330,000 frames. For
127 thermal bending fluctuations of a homogeneous rod, the ensemble average of cos([s]) decays exponentially with increasing arc length, s, along the CAT according to
(−푠⁄퐿푝) 〈cos(휃[푠])〉푠 = 푒 , where Lp is the persistence length.
The lateral fluctuations of the distal end of the stalk were assessed by tracking the relative translation of the stalk/MTBD joint (hinge vertex, see below) in the xy-plane, where x and y represent the directions parallel and perpendicular to the dynein ring, respectively. This was accomplished by orienting a reference structure [63] with the proximal bead of the CAT at the origin, the beads corresponding to the heptad at the base of the coiled-coil region extending along the –z-axis, and the x-direction passing through the plane of the ring (Figure 31A); each frame from the simulation trajectory was aligned to the reference structure based on the heptad at the base of the coiled-coil. Fluctuations of the coil in the x- and y-directions were used to estimate stiffness parallel and perpendicular to the plane of the ring.
128
Figure 31: Alignments and references used in analysis of MD simulation data A. Lateral fluctuations of the distal tip of the stalk were measured based on the translation of the stalk/MTBD hinge vertex in the xy-plane, given alignment of the heptad at the base of the coil to a reference structure. The reference structure was oriented with the proximal bead of the CAT at the origin, the CAT corresponding to the heptad at the base of the coiled-coil region extending along the –z-axis, and the x-direction passing through the plane of the AAA ring. B. Longitudinal twisting of the coiled-coil was measured based on a sum over angles between CC1-CC2 vectors along the coil, defined by connecting corresponding beads from the CAT of each α-helix. C. The angle of the stalk/MTBD hinge was measured based on theoretical attachment to a MT, estimated by alignment to an experimental reference structure containing a bound tubulin fragment. Images were rendered using VMD 1.9.2 [5].
129 The longitudinal twisting of the coiled-coil was measured by reducing the individual -helices of the coil to their respective CATs with the TWISTER algorithm
[198], then defining vectors through paired residue beads from CC1 to CC2 along the coil
(Figure 31B). The angular deviations between successive vectors over the length of the coiled-coil were summed to produce an overall angle of longitudinal twist. Longitudinal twisting of the coiled-coil was used to estimate torsional stiffness.
The hinge vertex of the stalk/MTBD junction was defined near the midpoint of the
CAT beads corresponding to the conserved proline residues, adjusted slightly to align with the distal heptad of the CAT, which was designated as the coil-arm of the hinge. The angle of the stalk/MTBD hinge was measured as the angle between the coil-arm of the hinge and a vector of equivalent length representing the theoretical direction of the MT (Figure 31C).
Relative orientation to a theoretically bound MT was determined by aligning the MTBD from each frame from the simulation trajectory to the MTBD of an experimental reference structure containing a bound tubulin fragment [199].
130 BIBLIOGRAPHY
[1] H. Imai, T. Shima, K. Sutoh, M.L. Walker, P.J. Knight, T. Kon, S.A. Burgess, Direct observation shows superposition and large scale flexibility within cytoplasmic dynein motors moving along microtubules, Nat Commun, 6 (2015) 8179.
[2] V. Belyy, N.L. Hendel, A. Chien, A. Yildiz, Cytoplasmic dynein transports cargos via load-sharing between the heads, Nat Commun, 5 (2014) 5544.
[3] M.B. Elowitz, M.G. Surette, P.E. Wolf, J. Stock, S. Leibler, Photoactivation turns green fluorescent protein red, Curr Biol, 7 (1997) 809-812.
[4] Michael S. Woody, John H. Lewis, Michael J. Greenberg, Yale E. Goldman, E.M. Ostap, MEMLET: An Easy-to-Use Tool for Data Fitting and Model Comparison Using Maximum-Likelihood Estimation, Biophysical Journal, 111 273-282.
[5] W. Humphrey, A. Dalke, K. Schulten, VMD: visual molecular dynamics, J Mol Graph, 14 (1996) 33-38, 27-38.
[6] M.M. Fu, E.L. Holzbaur, Integrated regulation of motor-driven organelle transport by scaffolding proteins, Trends Cell Biol, 24 (2014) 564-574.
[7] D. Debanne, E. Campanac, A. Bialowas, E. Carlier, G. Alcaraz, Axon Physiology, Physiological Reviews, 91 (2011) 555-602.
[8] J.T. Edward, Molecular volumes and the Stokes-Einstein equation, Journal of chemical education, 47 (1970) 261.
[9] K. Glunde, S.E. Guggino, M. Solaiyappan, A.P. Pathak, Y. Ichikawa, Z.M. Bhujwalla, Extracellular acidification alters lysosomal trafficking in human breast cancer cells, Neoplasia, 5 (2003) 533-545.
[10] A. Efergan, N.P. Azouz, O. Klein, K. Noguchi, M.E. Rothenberg, M. Fukuda, R. Sagi-Eisenberg, Rab12 Regulates Retrograde Transport of Mast Cell Secretory Granules by Interacting with the RILP-Dynein Complex, J Immunol, 196 (2016) 1091-1101.
[11] A.P. Carter, Crystal clear insights into how the dynein motor moves, J Cell Sci, 126 (2013) 705-713.
[12] R.D. Vale, The molecular motor toolbox for intracellular transport, Cell, 112 (2003) 467-480.
[13] H. Park, E. Toprak, P.R. Selvin, Single-molecule fluorescence to study molecular motors, Q Rev Biophys, 40 (2007) 87-111.
131 [14] T. Torisawa, M. Ichikawa, A. Furuta, K. Saito, K. Oiwa, H. Kojima, Y.Y. Toyoshima, K. Furuta, Autoinhibition and cooperative activation mechanisms of cytoplasmic dynein, Nat Cell Biol, 16 (2014) 1118-1124.
[15] Y. Mao, D. Varma, R. Vallee, Emerging functions of force-producing kinetochore motors, Cell Cycle, 9 (2010) 715-719.
[16] A. Musacchio, E.D. Salmon, The spindle-assembly checkpoint in space and time, Nat Rev Mol Cell Biol, 8 (2007) 379-393.
[17] B.F. McEwen, G.K. Chan, B. Zubrowski, M.S. Savoian, M.T. Sauer, T.J. Yen, CENP-E is essential for reliable bioriented spindle attachment, but chromosome alignment can be achieved via redundant mechanisms in mammalian cells, Mol Biol Cell, 12 (2001) 2776-2789.
[18] D. Varma, P. Monzo, S.A. Stehman, R.B. Vallee, Direct role of dynein motor in stable kinetochore-microtubule attachment, orientation, and alignment, J Cell Biol, 182 (2008) 1045-1054.
[19] W.O. Hancock, Bidirectional cargo transport: moving beyond tug of war, Nat Rev Mol Cell Biol, 15 (2014) 615-628.
[20] M.A. Welte, S.P. Gross, M. Postner, S.M. Block, E.F. Wieschaus, Developmental regulation of vesicle transport in Drosophila embryos: forces and kinetics, Cell, 92 (1998) 547-557.
[21] S. Maday, K.E. Wallace, E.L. Holzbaur, Autophagosomes initiate distally and mature during transport toward the cell soma in primary neurons, J Cell Biol, 196 (2012) 407-417.
[22] C.C. Hoogenraad, A. Akhmanova, Bicaudal D Family of Motor Adaptors: Linking Dynein Motility to Cargo Binding, Trends Cell Biol, 26 (2016) 327-340.
[23] N. Rocha, C. Kuijl, R. van der Kant, L. Janssen, D. Houben, H. Janssen, W. Zwart, J. Neefjes, Cholesterol sensor ORP1L contacts the ER protein VAP to control Rab7- RILP-p150 Glued and late endosome positioning, J Cell Biol, 185 (2009) 1209-1225.
[24] J.R. Kardon, R.D. Vale, Regulators of the cytoplasmic dynein motor, Nat Rev Mol Cell Biol, 10 (2009) 854-865.
[25] A.J. Roberts, T. Kon, P.J. Knight, K. Sutoh, S.A. Burgess, Functions and mechanics of dynein motor proteins, Nat Rev Mol Cell Biol, 14 (2013) 713-726.
[26] A.P. Carter, C. Cho, L. Jin, R.D. Vale, Crystal structure of the dynein motor domain, Science, 331 (2011) 1159-1165.
132 [27] M.A. Cianfrocco, M.E. DeSantis, A.E. Leschziner, S.L. Reck-Peterson, Mechanism and regulation of cytoplasmic dynein, Annu Rev Cell Dev Biol, 31 (2015) 83-108.
[28] T. Kon, M. Nishiura, R. Ohkura, Y.Y. Toyoshima, K. Sutoh, Distinct functions of nucleotide-binding/hydrolysis sites in the four AAA modules of cytoplasmic dynein, Biochemistry, 43 (2004) 11266-11274.
[29] C. Cho, S.L. Reck-Peterson, R.D. Vale, Regulatory ATPase sites of cytoplasmic dynein affect processivity and force generation, J Biol Chem, 283 (2008) 25839-25845.
[30] K. Imamula, T. Kon, R. Ohkura, K. Sutoh, The coordination of cyclic microtubule association/dissociation and tail swing of cytoplasmic dynein, Proc Natl Acad Sci U S A, 104 (2007) 16134-16139.
[31] M.A. DeWitt, C.A. Cypranowska, F.B. Cleary, V. Belyy, A. Yildiz, The AAA3 domain of cytoplasmic dynein acts as a switch to facilitate microtubule release, Nat Struct Mol Biol, 22 (2015) 73-80.
[32] H. Schmidt, E.S. Gleave, A.P. Carter, Insights into dynein motor domain function from a 3.3-Å crystal structure, Nat Struct Mol Biol, 19 (2012) 492-497, S491.
[33] T. Kon, K. Sutoh, G. Kurisu, X-ray structure of a functional full-length dynein motor domain, Nat Struct Mol Biol, 18 (2011) 638-642.
[34] A.J. Roberts, B. Malkova, M.L. Walker, H. Sakakibara, N. Numata, T. Kon, R. Ohkura, T.A. Edwards, P.J. Knight, K. Sutoh, K. Oiwa, S.A. Burgess, ATP-driven remodeling of the linker domain in the dynein motor, Structure, 20 (2012) 1670-1680.
[35] T. Kon, T. Mogami, R. Ohkura, M. Nishiura, K. Sutoh, ATP hydrolysis cycle- dependent tail motions in cytoplasmic dynein, Nat Struct Mol Biol, 12 (2005) 513-519.
[36] D. Tsygankov, A.W. Serohijos, N.V. Dokholyan, T.C. Elston, Kinetic models for the coordinated stepping of cytoplasmic dynein, J Chem Phys, 130 (2009) 025101.
[37] I.R. Gibbons, J.E. Garbarino, C.E. Tan, S.L. Reck-Peterson, R.D. Vale, A.P. Carter, The affinity of the dynein microtubule-binding domain is modulated by the conformation of its coiled-coil stalk, J Biol Chem, 280 (2005) 23960-23965.
[38] Y. Nishikawa, M. Inatomi, H. Iwasaki, G. Kurisu, Structural Change in the Dynein Stalk Region Associated with Two Different Affinities for the Microtubule, J Mol Biol, 428 (2016) 1886-1896.
[39] N. Numata, T. Shima, R. Ohkura, T. Kon, K. Sutoh, C-sequence of the Dictyostelium cytoplasmic dynein participates in processivity modulation, FEBS Lett, 585 (2011) 1185-1190.
133 [40] M.P. Nicholas, P. Höök, S. Brenner, C.L. Wynne, R.B. Vallee, A. Gennerich, Control of cytoplasmic dynein force production and processivity by its C-terminal domain, Nat Commun, 6 (2015) 6206.
[41] V. Belyy, M.A. Schlager, H. Foster, A.E. Reimer, A.P. Carter, A. Yildiz, The mammalian dynein-dynactin complex is a strong opponent to kinesin in a tug-of-war competition, Nat Cell Biol, 18 (2016) 1018-1024.
[42] A. Yildiz, J.N. Forkey, S.A. McKinney, T. Ha, Y.E. Goldman, P.R. Selvin, Myosin V walks hand-over-hand: single fluorophore imaging with 1.5-nm localization, Science, 300 (2003) 2061-2065.
[43] J.N. Forkey, M.E. Quinlan, M.A. Shaw, J.E. Corrie, Y.E. Goldman, Three- dimensional structural dynamics of myosin V by single-molecule fluorescence polarization, Nature, 422 (2003) 399-404.
[44] A. Yildiz, M. Tomishige, R.D. Vale, P.R. Selvin, Kinesin walks hand-over-hand, Science, 303 (2004) 676-678.
[45] T. Sakamoto, M.R. Webb, E. Forgacs, H.D. White, J.R. Sellers, Direct observation of the mechanochemical coupling in myosin Va during processive movement, Nature, 455 (2008) 128-132.
[46] N.R. Guydosh, S.M. Block, Backsteps induced by nucleotide analogs suggest the front head of kinesin is gated by strain, Proc Natl Acad Sci U S A, 103 (2006) 8054- 8059.
[47] W. Qiu, N.D. Derr, B.S. Goodman, E. Villa, D. Wu, W. Shih, S.L. Reck-Peterson, Dynein achieves processive motion using both stochastic and coordinated stepping, Nat Struct Mol Biol, 19 (2012) 193-200.
[48] M.A. DeWitt, A.Y. Chang, P.A. Combs, A. Yildiz, Cytoplasmic dynein moves through uncoordinated stepping of the AAA+ ring domains, Science, 335 (2012) 221- 225.
[49] A. Gennerich, A.P. Carter, S.L. Reck-Peterson, R.D. Vale, Force-induced bidirectional stepping of cytoplasmic dynein, Cell, 131 (2007) 952-965.
[50] S.L. Reck-Peterson, A. Yildiz, A.P. Carter, A. Gennerich, N. Zhang, R.D. Vale, Single-molecule analysis of dynein processivity and stepping behavior, Cell, 126 (2006) 335-348.
[51] A. Vilfan, Five models for myosin V, Front Biosci (Landmark Ed), 14 (2009) 2269- 2284.
[52] M.J. Schnitzer, S.M. Block, Kinesin hydrolyses one ATP per 8-nm step, Nature, 388 (1997) 386-390. 134 [53] R. Dixit, J.L. Ross, Y.E. Goldman, E.L. Holzbaur, Differential regulation of dynein and kinesin motor proteins by tau, Science, 319 (2008) 1086-1089.
[54] T.A. Schroer, Dynactin, Annu Rev Cell Dev Biol, 20 (2004) 759-779.
[55] S.R. Gill, T.A. Schroer, I. Szilak, E.R. Steuer, M.P. Sheetz, D.W. Cleveland, Dynactin, a conserved, ubiquitously expressed component of an activator of vesicle motility mediated by cytoplasmic dynein, J Cell Biol, 115 (1991) 1639-1650.
[56] J.L. Ross, K. Wallace, H. Shuman, Y.E. Goldman, E.L. Holzbaur, Processive bidirectional motion of dynein-dynactin complexes in vitro, Nat Cell Biol, 8 (2006) 562- 570.
[57] E.L. Holzbaur, J.A. Hammarback, B.M. Paschal, N.G. Kravit, K.K. Pfister, R.B. Vallee, Homology of a 150K cytoplasmic dynein-associated polypeptide with the Drosophila gene Glued, Nature, 351 (1991) 579-583.
[58] S.J. King, T.A. Schroer, Dynactin increases the processivity of the cytoplasmic dynein motor, Nat Cell Biol, 2 (2000) 20-24.
[59] R.J. McKenney, W. Huynh, M.E. Tanenbaum, G. Bhabha, R.D. Vale, Activation of cytoplasmic dynein motility by dynactin-cargo adapter complexes, Science, 345 (2014) 337-341.
[60] M.A. Olenick, M. Tokito, M. Boczkowska, R. Dominguez, E.L. Holzbaur, Hook Adaptors Induce Unidirectional Processive Motility by Enhancing the Dynein-Dynactin Interaction, J Biol Chem, 291 (2016) 18239-18251.
[61] J. Huang, A.J. Roberts, A.E. Leschziner, S.L. Reck-Peterson, Lis1 acts as a "clutch" between the ATPase and microtubule-binding domains of the dynein motor, Cell, 150 (2012) 975-986.
[62] K. Toropova, S. Zou, A.J. Roberts, W.B. Redwine, B.S. Goodman, S.L. Reck- Peterson, A.E. Leschziner, Lis1 regulates dynein by sterically blocking its mechanochemical cycle, Elife, 3 (2014).
[63] H. Schmidt, R. Zalyte, L. Urnavicius, A.P. Carter, Structure of human cytoplasmic dynein-2 primed for its power stroke, Nature, 518 (2015) 435-438.
[64] A.J. Roberts, N. Numata, M.L. Walker, Y.S. Kato, B. Malkova, T. Kon, R. Ohkura, F. Arisaka, P.J. Knight, K. Sutoh, S.A. Burgess, AAA+ Ring and linker swing mechanism in the dynein motor, Cell, 136 (2009) 485-495.
[65] J. Lin, K. Okada, M. Raytchev, M.C. Smith, D. Nicastro, Structural mechanism of the dynein power stroke, Nat Cell Biol, 16 (2014) 479-485.
135 [66] P.R. Selvin, T. Ha, Single-molecule Techniques: A Laboratory Manual, Cold Spring Harbor Laboratory Press2008.
[67] W.E. Moerner, Y. Shechtman, Q. Wang, Single-molecule spectroscopy and imaging over the decades, Faraday Discuss, 184 (2015) 9-36.
[68] A. Kusumi, T.A. Tsunoyama, K.M. Hirosawa, R.S. Kasai, T.K. Fujiwara, Tracking single molecules at work in living cells, Nat Chem Biol, 10 (2014) 524-532.
[69] E. Betzig, Single Molecules, Cells, and Super-Resolution Optics (Nobel Lecture), Angewandte Chemie-International Edition, 54 (2015) 8034-8053.
[70] S.W. Hell, Nanoscopy with Focused Light (Nobel Lecture), Angew Chem Int Ed Engl, 54 (2015) 8054-8066.
[71] W. Moerner, Single-Molecule Spectroscopy, Imaging, and Photocontrol: Foundations for Super-Resolution Microscopy (Nobel Lecture), Angewandte Chemie- International Edition, 54 (2015) 8067-8093.
[72] J.F. Beausang, D.Y. Shroder, P.C. Nelson, Y.E. Goldman, Tilting and wobble of myosin V by high-speed single-molecule polarized fluorescence microscopy, Biophys J, 104 (2013) 1263-1273.
[73] J.R. Lakowicz, Principles of Fluorescence Spectroscopy, Springer US2007.
[74] D.M. Jameson, J.A. Ross, Fluorescence polarization/anisotropy in diagnostics and imaging, Chem Rev, 110 (2010) 2685-2708.
[75] B. Diroll, T. Dadosh, A. Koschitzky, Y. Goldman, C. Murray, Interpreting the Energy-Dependent Anisotropy of Colloidal Nanorods Using Ensemble and Single- Particle Spectroscopy, Journal of Physical Chemistry C, 117 (2013) 23928-23937.
[76] L. Xiao, E. Yeung, R. Cooks, J. Pemberton, Optical Imaging of Individual Plasmonic Nanoparticles in Biological Samples, Annual Review of Analytical Chemistry, Vol 7, 7 (2014) 89-111.
[77] M.A. Bopp, A. Sytnik, T.D. Howard, R.J. Cogdell, R.M. Hochstrasser, The dynamics of structural deformations of immobilized single light-harvesting complexes, Proc Natl Acad Sci U S A, 96 (1999) 11271-11276.
[78] C.C. Liu, P.G. Schultz, Adding New Chemistries to the Genetic Code, Annual Review of Biochemistry, 79 (2010) 413-444.
[79] V.Y. Postupalenko, O.M. Zamotaiev, V.V. Shvadchak, A.V. Strizhak, V.G. Pivovarenko, A.S. Klymchenko, Y. Mely, Dual-fluorescence L-amino acid reports insertion and orientation of melittin peptide in cell membranes, Bioconjug Chem, 24 (2013) 1998-2007. 136 [80] L.C. Speight, A.K. Muthusamy, J.M. Goldberg, J.B. Warner, R.F. Wissner, T.S. Willi, B.F. Woodman, R.A. Mehl, E.J. Petersson, Efficient synthesis and in vivo incorporation of acridon-2-ylalanine, a fluorescent amino acid for lifetime and Förster resonance energy transfer/luminescence resonance energy transfer studies, J Am Chem Soc, 135 (2013) 18806-18814.
[81] M. Kampmann, C.E. Atkinson, A.L. Mattheyses, S.M. Simon, Mapping the orientation of nuclear pore proteins in living cells with polarized fluorescence microscopy, Nat Struct Mol Biol, 18 (2011) 643-649.
[82] T.J. Gould, M.S. Gunewardene, M.V. Gudheti, V.V. Verkhusha, S.R. Yin, J.A. Gosse, S.T. Hess, Nanoscale imaging of molecular positions and anisotropies, Nat Methods, 5 (2008) 1027-1030.
[83] H. Wang, X. Chen, Site-specifically modified fusion proteins for molecular imaging, Front Biosci, 13 (2008) 1716-1732.
[84] C. Freidel, S. Kaloyanova, K. Peneva, Chemical tags for site-specific fluorescent labeling of biomolecules, Amino Acids, 48 (2016) 1357-1372.
[85] C.A. Valades Cruz, H.A. Shaban, A. Kress, N. Bertaux, S. Monneret, M. Mavrakis, J. Savatier, S. Brasselet, Quantitative nanoscale imaging of orientational order in biological filaments by polarized superresolution microscopy, Proc Natl Acad Sci U S A, 113 (2016) E820-828.
[86] A.S. Backer, M. Lee, W.E. Moerner, Enhanced DNA Imaging Using Super- Resolution Microscopy and Simultaneous Single-Molecule Orientation Measurements, Conference on Lasers and Electro-Optics, Optical Society of America, San Jose, California, 2016, pp. JTh4B.4.
[87] P. Livanec, R. Dunn, Single-Molecule Probes of Lipid Membrane Structure, Langmuir, 24 (2008) 14066-14073.
[88] K.P. Armendariz, R.C. Dunn, Ganglioside influence on phospholipid films investigated with single molecule fluorescence measurements, J Phys Chem B, 117 (2013) 7959-7966.
[89] T. Wieland, [50 years of phalloidine: its discovery, characterization and current and future applications in cell research], Naturwissenschaften, 74 (1987) 367-373.
[90] M. Balaz, A. Månsson, Detection of small differences in actomyosin function using actin labeled with different phalloidin conjugates, Anal Biochem, 338 (2005) 224-236.
[91] A. Backer, M. Backlund, A. von Diezmann, S. Sahl, W. Moerner, A bisected pupil for studying single-molecule orientational dynamics and its application to three- dimensional super-resolution microscopy, Applied Physics Letters, 104 (2014).
137 [92] L.R. Otterbein, P. Graceffa, R. Dominguez, The crystal structure of uncomplexed actin in the ADP state, Science, 293 (2001) 708-711.
[93] I. Sase, H. Miyata, S. Ishiwata, K. Kinosita, Axial rotation of sliding actin filaments revealed by single-fluorophore imaging, Proc Natl Acad Sci U S A, 94 (1997) 5646- 5650.
[94] K. Adachi, R. Yasuda, H. Noji, H. Itoh, Y. Harada, M. Yoshida, K. Kinosita, Stepping rotation of F1-ATPase visualized through angle-resolved single-fluorophore imaging, Proc Natl Acad Sci U S A, 97 (2000) 7243-7247.
[95] S.M. Blackman, C.E. Cobb, A.H. Beth, D.W. Piston, The orientation of eosin-5- maleimide on human erythrocyte band 3 measured by fluorescence polarization microscopy, Biophys J, 71 (1996) 194-208.
[96] T.P. Burghardt, K. Ajtai, Effect of negative mechanical stress on the orientation of myosin cross-bridges in muscle fibers, Proc Natl Acad Sci U S A, 86 (1989) 5366-5370.
[97] C. Sabido-David, B. Brandmeier, J.S. Craik, J.E. Corrie, D.R. Trentham, M. Irving, Steady-state fluorescence polarization studies of the orientation of myosin regulatory light chains in single skeletal muscle fibers using pure isomers of iodoacetamidotetramethylrhodamine, Biophys J, 74 (1998) 3083-3092.
[98] J.N. Forkey, M.E. Quinlan, Y.E. Goldman, Measurement of single macromolecule orientation by total internal reflection fluorescence polarization microscopy, Biophys J, 89 (2005) 1261-1271.
[99] O. Julien, Y.-B. Sun, A.C. Knowles, B.D. Brandmeier, R.E. Dale, D.R. Trentham, J.E.T. Corrie, B.D. Sykes, M. Irving, Toward Protein Structure In Situ: Comparison of Two Bifunctional Rhodamine Adducts of Troponin C, Biophysical Journal, 93 (2007) 1008-1020.
[100] J.E. Corrie, B.D. Brandmeier, R.E. Ferguson, D.R. Trentham, J. Kendrick-Jones, S.C. Hopkins, U.A. van der Heide, Y.E. Goldman, C. Sabido-David, R.E. Dale, S. Criddle, M. Irving, Dynamic measurement of myosin light-chain-domain tilt and twist in muscle contraction, Nature, 400 (1999) 425-430.
[101] R.E. Ferguson, Y.B. Sun, P. Mercier, A.S. Brack, B.D. Sykes, J.E. Corrie, D.R. Trentham, M. Irving, In situ orientations of protein domains: troponin C in skeletal muscle fibers, Mol Cell, 11 (2003) 865-874.
[102] L. Fusi, E. Brunello, I.R. Sevrieva, Y.B. Sun, M. Irving, Structural dynamics of troponin during activation of skeletal muscle, Proc Natl Acad Sci U S A, 111 (2014) 4626-4631.
138 [103] Y. Sun, H.W. Schroeder, J.F. Beausang, K. Homma, M. Ikebe, Y.E. Goldman, Myosin VI walks "wiggly" on actin with large and variable tilting, Mol Cell, 28 (2007) 954-964.
[104] Y. Sun, O. Sato, F. Ruhnow, M.E. Arsenault, M. Ikebe, Y.E. Goldman, Single- molecule stepping and structural dynamics of myosin X, Nat Struct Mol Biol, 17 (2010) 485-491.
[105] H. Sosa, E.J. Peterman, W.E. Moerner, L.S. Goldstein, ADP-induced rocking of the kinesin motor domain revealed by single-molecule fluorescence polarization microscopy, Nat Struct Biol, 8 (2001) 540-544.
[106] T. Masaike, F. Koyama-Horibe, K. Oiwa, M. Yoshida, T. Nishizaka, Cooperative three-step motions in catalytic subunits of F1-ATPase correlate with 80[deg] and 40[deg] substep rotations, Nat Struct Mol Biol, 15 (2008) 1326-1333.
[107] C. Chen, X. Cui, J.F. Beausang, H. Zhang, I. Farrell, B.S. Cooperman, Y.E. Goldman, Elongation factor G initiates translocation through a power stroke, Proc Natl Acad Sci U S A, (2016).
[108] B.A. Griffin, S.R. Adams, R.Y. Tsien, Specific covalent labeling of recombinant protein molecules inside live cells, Science, 281 (1998) 269-272.
[109] H. Cao, Y. Xiong, T. Wang, B. Chen, T.C. Squier, M.U. Mayer, A red cy3-based biarsenical fluorescent probe targeted to a complementary binding peptide, J Am Chem Soc, 129 (2007) 8672-8673.
[110] C.B. Murray, M. Nirmal, D.J. Norris, M.G. Bawendi, Synthesis and structural characterization of II–VI semiconductor nanocrystallites (quantum dots), Zeitschrift für Physik D Atoms, Molecules and Clusters, 26 (1993) 231-233.
[111] M. Hines, P. Guyot-Sionnest, Synthesis and characterization of strongly luminescing ZnS-Capped CdSe nanocrystals, Journal of Physical Chemistry, 100 (1996) 468-471.
[112] X. Peng, M. Schlamp, A. Kadavanich, A. Alivisatos, Epitaxial growth of highly luminescent CdSe/CdS core/shell nanocrystals with photostability and electronic accessibility, Journal of the American Chemical Society, 119 (1997) 7019-7029.
[113] A. Alivisatos, Semiconductor clusters, nanocrystals, and quantum dots, Science, 271 (1996) 933-937.
[114] M. Bruchez, M. Moronne, P. Gin, S. Weiss, A.P. Alivisatos, Semiconductor nanocrystals as fluorescent biological labels, Science, 281 (1998) 2013-2016.
[115] B.O. Dabbousi, J. RodriguezViejo, F.V. Mikulec, J.R. Heine, H. Mattoussi, R. Ober, K.F. Jensen, M.G. Bawendi, (CdSe)ZnS core-shell quantum dots: Synthesis and 139 characterization of a size series of highly luminescent nanocrystallites, Journal of Physical Chemistry B, 101 (1997) 9463-9475.
[116] B. Dubertret, P. Skourides, D.J. Norris, V. Noireaux, A.H. Brivanlou, A. Libchaber, In vivo imaging of quantum dots encapsulated in phospholipid micelles, Science, 298 (2002) 1759-1762.
[117] V. Biju, T. Itoh, A. Anas, A. Sujith, M. Ishikawa, Semiconductor quantum dots and metal nanoparticles: syntheses, optical properties, and biological applications, Analytical and Bioanalytical Chemistry, 391 (2008) 2469-2495.
[118] S.A. Empedocles, R. Neuhauser, M.G. Bawendi, Three-dimensional orientation measurements of symmetric single chromophores using polarization microscopy, Nature, 399 (1999) 126-130.
[119] S. Gomes, C. Vieira, D. Almeida, J. Santos-Mallet, R. Menna-Barreto, C. Cesar, D. Feder, CdTe and CdSe Quantum Dots Cytotoxicity: A Comparative Study on Microorganisms, Sensors, 11 (2011) 11664-11678.
[120] N. Chen, Y. He, Y. Su, X. Li, Q. Huang, H. Wang, X. Zhang, R. Tai, C. Fan, The cytotoxicity of cadmium-based quantum dots, Biomaterials, 33 (2012) 1238-1244.
[121] L. Carbone, C. Nobile, M. De Giorgi, F. Sala, G. Morello, P. Pompa, M. Hytch, E. Snoeck, A. Fiore, I. Franchini, M. Nadasan, A. Silvestre, L. Chiodo, S. Kudera, R. Cingolani, R. Krahne, L. Manna, Synthesis and micrometer-scale assembly of colloidal CdSe/CdS nanorods prepared by a seeded growth approach, Nano Letters, 7 (2007) 2942- 2950.
[122] Y.-w. Jun, S.-M. Lee, N.-J. Kang, J. Cheon, Controlled Synthesis of Multi-armed CdS Nanorod Architectures Using Monosurfactant System, Journal of the American Chemical Society, 123 (2001) 5150-5151.
[123] M. Ohmachi, Y. Komori, A.H. Iwane, F. Fujii, T. Jin, T. Yanagida, Fluorescence microscopy for simultaneous observation of 3D orientation and movement and its application to quantum rod-tagged myosin V, Proc Natl Acad Sci U S A, 109 (2012) 5294-5298.
[124] A. Sitt, A. Salant, G. Menagen, U. Banin, Highly emissive nano rod-in-rod heterostructures with strong linear polarization, Nano Lett, 11 (2011) 2054-2060.
[125] I. Hadar, G.B. Hitin, A. Sitt, A. Faust, U. Banin, Polarization Properties of Semiconductor Nanorod Heterostructures: From Single Particles to the Ensemble, J Phys Chem Lett, 4 (2013) 502-507.
[126] E. Katz, I. Willner, Integrated nanoparticle-biomolecule hybrid systems: Synthesis, properties, and applications, Angewandte Chemie-International Edition, 43 (2004) 6042- 6108. 140 [127] T. Jin, F. Fujii, Y. Komai, J. Seki, A. Seiyama, Y. Yoshioka, Preparation and characterization of highly fluorescent, glutathione-coated near infrared quantum dots for in vivo fluorescence imaging, Int J Mol Sci, 9 (2008) 2044-2061.
[128] L.G. Lippert, J.T. Hallock, T. Dadosh, B.T. Diroll, C.B. Murray, Y.E. Goldman, NeutrAvidin Functionalization of CdSe/CdS Quantum Nanorods and Quantification of Biotin Binding Sites using Biotin-4-Fluorescein Fluorescence Quenching, Bioconjug Chem, 27 (2016) 562-568.
[129] T. Pellegrino, L. Manna, S. Kudera, T. Liedl, D. Koktysh, A. Rogach, S. Keller, J. Radler, G. Natile, W. Parak, Hydrophobic nanocrystals coated with an amphiphilic polymer shell: A general route to water soluble nanocrystals, Nano Letters, 4 (2004) 703- 707.
[130] E.V. Shtykova, X. Huang, X. Gao, J.C. Dyke, A.L. Schmucker, B. Dragnea, N. Remmes, D.V. Baxter, B. Stein, P.V. Konarev, D.I. Svergun, L.M. Bronstein, Hydrophilic Monodisperse Magnetic Nanoparticles Protected by an Amphiphilic Alternating Copolymer, J Phys Chem C Nanomater Interfaces, 112 (2008) 16809-16817.
[131] A. Watson, X.Y. Wu, M. Bruchez, Lighting up cells with quantum dots, Biotechniques, 34 (2003) 296-+.
[132] J.T. Fourkas, Rapid determination of the three-dimensional orientation of single molecules, Opt Lett, 26 (2001) 211-213.
[133] C. Lu, D. Vanden Bout, Analysis of orientational dynamics of single fluorophore trajectories from three-angle polarization experiments, Journal of Chemical Physics, 128 (2008).
[134] J.F. Beausang, H.W. Schroeder, P.C. Nelson, Y.E. Goldman, Twirling of actin by myosins II and V observed via polarized TIRF in a modified gliding assay, Biophys J, 95 (2008) 5820-5831.
[135] G. Wang, W. Sun, Y. Luo, N. Fang, Resolving Rotational Motions of Nano-objects in Engineered Environments and Live Cells with Gold Nanorods and Differential Interference Contrast Microscopy, Journal of the American Chemical Society, 132 (2010) 16417-16422.
[136] L. Xiao, J.W. Ha, L. Wei, G. Wang, N. Fang, Determining the full three- dimensional orientation of single anisotropic nanoparticles by differential interference contrast microscopy, Angew Chem Int Ed Engl, 51 (2012) 7734-7738.
[137] Y. Gu, W. Sun, G. Wang, K. Jeftinija, S. Jeftinija, N. Fang, Rotational dynamics of cargos at pauses during axonal transport, Nat Commun, 3 (2012) 1030.
[138] A. Backer, M. Backlund, M. Lew, W. Moerner, Single-molecule orientation measurements with a quadrated pupil, Optics Letters, 38 (2013) 1521-1523. 141 [139] M.P. Backlund, M.D. Lew, A.S. Backer, S.J. Sahl, G. Grover, A. Agrawal, R. Piestun, W.E. Moerner, Simultaneous, accurate measurement of the 3D position and orientation of single molecules, Proc Natl Acad Sci U S A, 109 (2012) 19087-19092.
[140] R. Yasuda, H. Noji, M. Yoshida, K. Kinosita, H. Itoh, Resolution of distinct rotational substeps by submillisecond kinetic analysis of F1-ATPase, Nature, 410 (2001) 898-904.
[141] D. Spetzler, J. York, D. Daniel, R. Fromme, D. Lowry, W. Frasch, Microsecond time scale rotation measurements of single F1-ATPase molecules, Biochemistry, 45 (2006) 3117-3124.
[142] J.H. Lewis, J.F. Beausang, H.L. Sweeney, Y.E. Goldman, The azimuthal path of myosin V and its dependence on lever-arm length, J Gen Physiol, 139 (2012) 101-120.
[143] S. Enoki, R. Iino, Y. Niitani, Y. Minagawa, M. Tomishige, H. Noji, High-speed angle-resolved imaging of a single gold nanorod with microsecond temporal resolution and one-degree angle precision, Anal Chem, 87 (2015) 2079-2086.
[144] A.R. Hodges, C.S. Bookwalter, E.B. Krementsova, K.M. Trybus, A nonprocessive class V myosin drives cargo processively when a kinesin- related protein is a passenger, Curr Biol, 19 (2009) 2121-2125.
[145] D.M. Warshaw, G.G. Kennedy, S.S. Work, E.B. Krementsova, S. Beck, K.M. Trybus, Differential labeling of myosin V heads with quantum dots allows direct visualization of hand-over-hand processivity, Biophys J, 88 (2005) L30-32.
[146] A.P. Alivisatos, Semiconductor clusters, nanocrystals, and quantum dots, Science, 271 (1996) 933-937.
[147] Y. Chen, J. Vela, H. Htoon, J.L. Casson, D.J. Werder, D.A. Bussian, V.I. Klimov, J.A. Hollingsworth, "Giant" multishell CdSe nanocrystal quantum dots with suppressed blinking, Journal of the American Chemical Society, 130 (2008) 5026-5027.
[148] J.D. Kulman, M. Satake, J.E. Harris, A versatile system for site-specific enzymatic biotinylation and regulated expression of proteins in cultured mammalian cells, Protein Expression and Purification, 52 (2007) 320-328.
[149] D. Talapin, E. Shevchenko, C. Murray, A. Kornowski, S. Forster, H. Weller, CdSe and CdSe/CdS nanorod solids, Journal of the American Chemical Society, 126 (2004) 12984-12988.
[150] J. Aldana, Y. Wang, X. Peng, Photochemical instability of CdSe nanocrystals coated by hydrophilic thiols, Journal of the American Chemical Society, 123 (2001) 8844-8850.
142 [151] G. Kada, H. Falk, H.J. Gruber, Accurate measurement of avidin and streptavidin in crude biofluids with a new, optimized biotin-fluorescein conjugate, Biochim Biophys Acta, 1427 (1999) 33-43.
[152] G. Kada, K. Kaiser, H. Falk, H.J. Gruber, Rapid estimation of avidin and streptavidin by fluorescence quenching or fluorescence polarization, Biochim Biophys Acta, 1427 (1999) 44-48.
[153] H.J. Gruber, G. Kada, M. Marek, K. Kaiser, Accurate titration of avidin and streptavidin with biotin-fluorophore conjugates in complex, colored biofluids, Biochim Biophys Acta, 1381 (1998) 203-212.
[154] R. Mittal, M.P. Bruchez, Biotin-4-fluorescein based fluorescence quenching assay for determination of biotin binding capacity of streptavidin conjugated quantum dots, Bioconjug Chem, 22 (2011) 362-368.
[155] T. Mizuhara, K. Saha, D.F. Moyano, C.S. Kim, B. Yan, Y.K. Kim, V.M. Rotello, Acylsulfonamide-Functionalized Zwitterionic Gold Nanoparticles for Enhanced Cellular Uptake at Tumor pH, Angew Chem Int Ed Engl, 54 (2015) 6567-6570.
[156] N. Zhan, G. Palui, M. Safi, X. Ji, H. Mattoussi, Multidentate zwitterionic ligands provide compact and highly biocompatible quantum dots, J Am Chem Soc, 135 (2013) 13786-13795.
[157] B.M. Paschal, R.B. Vallee, Retrograde transport by the microtubule-associated protein MAP 1C, Nature, 330 (1987) 181-183.
[158] F.R. Cottingham, L. Gheber, D.L. Miller, M.A. Hoyt, Novel roles for saccharomyces cerevisiae mitotic spindle motors, J Cell Biol, 147 (1999) 335-350.
[159] Y.Y. Li, E. Yeh, T. Hays, K. Bloom, Disruption of mitotic spindle orientation in a yeast dynein mutant, Proc Natl Acad Sci U S A, 90 (1993) 10096-10100.
[160] B.J. Howell, B.F. McEwen, J.C. Canman, D.B. Hoffman, E.M. Farrar, C.L. Rieder, E.D. Salmon, Cytoplasmic dynein/dynactin drives kinetochore protein transport to the spindle poles and has a role in mitotic spindle checkpoint inactivation, J Cell Biol, 155 (2001) 1159-1172.
[161] B.H. LaMonte, K.E. Wallace, B.A. Holloway, S.S. Shelly, J. Ascaño, M. Tokito, T. Van Winkle, D.S. Howland, E.L. Holzbaur, Disruption of dynein/dynactin inhibits axonal transport in motor neurons causing late-onset progressive degeneration, Neuron, 34 (2002) 715-727.
[162] J.R. Levy, E.L. Holzbaur, Cytoplasmic dynein/dynactin function and dysfunction in motor neurons, Int J Dev Neurosci, 24 (2006) 103-111.
143 [163] S. Maday, A.E. Twelvetrees, A.J. Moughamian, E.L. Holzbaur, Axonal transport: cargo-specific mechanisms of motility and regulation, Neuron, 84 (2014) 292-309.
[164] T. Kon, T. Oyama, R. Shimo-Kon, K. Imamula, T. Shima, K. Sutoh, G. Kurisu, The 2.8 Å crystal structure of the dynein motor domain, Nature, 484 (2012) 345-350.
[165] E.L. Holzbaur, K.A. Johnson, ADP release is rate limiting in steady-state turnover by the dynein adenosinetriphosphatase, Biochemistry, 28 (1989) 5577-5585.
[166] E.L. Holzbaur, K.A. Johnson, Microtubules accelerate ADP release by dynein, Biochemistry, 28 (1989) 7010-7016.
[167] C. Cho, R.D. Vale, The mechanism of dynein motility: insight from crystal structures of the motor domain, Biochim Biophys Acta, 1823 (2012) 182-191.
[168] R. Mallik, B.C. Carter, S.A. Lex, S.J. King, S.P. Gross, Cytoplasmic dynein functions as a gear in response to load, Nature, 427 (2004) 649-652.
[169] S.A. Burgess, M.L. Walker, H. Sakakibara, P.J. Knight, K. Oiwa, Dynein structure and power stroke, Nature, 421 (2003) 715-718.
[170] H. Ueno, T. Yasunaga, C. Shingyoji, K. Hirose, Dynein pulls microtubules without rotating its stalk, Proc Natl Acad Sci U S A, 105 (2008) 19702-19707.
[171] A. Sarlah, A. Vilfan, The winch model can explain both coordinated and uncoordinated stepping of cytoplasmic dynein, Biophys J, 107 (2014) 662-671.
[172] S.A. Burgess, P.J. Knight, Is the dynein motor a winch?, Current Opinion in Structural Biology, 14 (2004) 138-146.
[173] J.F. Beausang, Y.E. Goldman, P.C. Nelson, Changepoint analysis for single- molecule polarized total internal reflection fluorescence microscopy experiments, Methods Enzymol, 487 (2011) 431-463.
[174] F.B. Cleary, M.A. Dewitt, T. Bilyard, Z.M. Htet, V. Belyy, D.D. Chan, A.Y. Chang, A. Yildiz, Tension on the linker gates the ATP-dependent release of dynein from microtubules, Nat Commun, 5 (2014) 4587.
[175] Y. Nishikawa, T. Oyama, N. Kamiya, T. Kon, Y.Y. Toyoshima, H. Nakamura, G. Kurisu, Structure of the entire stalk region of the Dynein motor domain, J Mol Biol, 426 (2014) 3232-3245.
[176] C.W. Wolgemuth, S.X. Sun, Elasticity of alpha-helical coiled coils, Phys Rev Lett, 97 (2006) 248101.
[177] A.R. Dunn, J.A. Spudich, Dynamics of the unbound head during myosin V processive translocation, Nat Struct Mol Biol, 14 (2007) 246-248.
144 [178] E. Shaviv, A. Salant, U. Banin, Size dependence of molar absorption coefficients of CdSe semiconductor quantum rods, Chemphyschem, 10 (2009) 1028-1031.
[179] W. Yu, L. Qu, W. Guo, X. Peng, Experimental determination of the extinction coefficient of CdTe, CdSe, and CdS nanocrystals, Chemistry of Materials, 15 (2003) 2854-2860.
[180] M. Alves-Santos, R. Di Felice, G. Goldoni, Dielectric Functions of Semiconductor Nanoparticles from the Optical Absorption Spectrum: The Case of CdSe and CdS, Journal of Physical Chemistry C, 114 (2010) 3776-3780.
[181] J. Jasieniak, L. Smith, J. van Embden, P. Mulvaney, M. Califano, Re-examination of the Size-Dependent Absorption Properties of CdSe Quantum Dots, Journal of Physical Chemistry C, 113 (2009) 19468-19474.
[182] S.C. Gill, P.H. von Hippel, Calculation of protein extinction coefficients from amino acid sequence data, Anal Biochem, 182 (1989) 319-326.
[183] J.F. Beausang, Y. Sun, M.E. Quinlan, J.N. Forkey, Y.E. Goldman, Construction of flow chambers for polarized total internal reflection fluorescence microscopy (polTIRFM) motility assays, Cold Spring Harb Protoc, 2012 (2012) 712-715.
[184] J. Chen, A.K. Gupta, On change point detection and estimation, Communications in Statistics-Simulation and Computation, 30 (2001) 665-697.
[185] G.G. Krivov, M.V. Shapovalov, R.L. Dunbrack, Improved prediction of protein side-chain conformations with SCWRL4, Proteins, 77 (2009) 778-795.
[186] J. Walshaw, D.N. Woolfson, SOCKET: a program for identifying and analysing coiled-coil motifs within protein structures1, Journal of Molecular Biology, 307 (2001) 1427-1450.
[187] T. Kon, K. Imamula, A.J. Roberts, R. Ohkura, P.J. Knight, I.R. Gibbons, S.A. Burgess, K. Sutoh, Helix sliding in the stalk coiled coil of dynein couples ATPase and microtubule binding, Nat Struct Mol Biol, 16 (2009) 325-333.
[188] C.R. Søndergaard, M.H. Olsson, M. Rostkowski, J.H. Jensen, Improved Treatment of Ligands and Coupling Effects in Empirical Calculation and Rationalization of pKa Values, J Chem Theory Comput, 7 (2011) 2284-2295.
[189] T.J. Dolinsky, P. Czodrowski, H. Li, J.E. Nielsen, J.H. Jensen, G. Klebe, N.A. Baker, PDB2PQR: expanding and upgrading automated preparation of biomolecular structures for molecular simulations, Nucleic Acids Res, 35 (2007) W522-525.
[190] A.D. MacKerell, D. Bashford, M. Bellott, R.L. Dunbrack, J.D. Evanseck, M.J. Field, S. Fischer, J. Gao, H. Guo, S. Ha, D. Joseph-McCarthy, L. Kuchnir, K. Kuczera, F.T.K. Lau, C. Mattos, S. Michnick, T. Ngo, D.T. Nguyen, B. Prodhom, W.E. Reiher, B. 145 Roux, M. Schlenkrich, J.C. Smith, R. Stote, J. Straub, M. Watanabe, J. Wiórkiewicz- Kuczera, D. Yin, M. Karplus, All-Atom Empirical Potential for Molecular Modeling and Dynamics Studies of Proteins, The Journal of Physical Chemistry B, 102 (1998) 3586- 3616.
[191] A.D. MacKerell, M. Feig, C.L. Brooks, Improved Treatment of the Protein Backbone in Empirical Force Fields, Journal of the American Chemical Society, 126 (2004) 698-699.
[192] W. JORGENSEN, J. CHANDRASEKHAR, J. MADURA, R. IMPEY, M. KLEIN, COMPARISON OF SIMPLE POTENTIAL FUNCTIONS FOR SIMULATING LIQUID WATER, Journal of Chemical Physics, 79 (1983) 926-935.
[193] J.C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa, C. Chipot, R.D. Skeel, L. Kalé, K. Schulten, Scalable molecular dynamics with NAMD, J Comput Chem, 26 (2005) 1781-1802.
[194] J.G. Garcia de la Torre, V.A. Bloomfield, Hydrodynamic properties of complex, rigid, biological macromolecules: theory and applications, Q Rev Biophys, 14 (1981) 81- 139.
[195] D. Leith, Drag on Nonspherical Objects, Aerosol Science and Technology, 6 (1987) 153-161.
[196] X.E. Li, W. Lehman, S. Fischer, The relationship between curvature, flexibility and persistence length in the tropomyosin coiled-coil, J Struct Biol, 170 (2010) 313-318.
[197] X.E. Li, K.C. Holmes, W. Lehman, H. Jung, S. Fischer, The shape and flexibility of tropomyosin coiled coils: implications for actin filament assembly and regulation, J Mol Biol, 395 (2010) 327-339.
[198] S.V. Strelkov, P. Burkhard, Analysis of α-Helical Coiled Coils with the Program TWISTER Reveals a Structural Mechanism for Stutter Compensation, Journal of Structural Biology, 137 (2002) 54-64.
[199] S. Uchimura, T. Fujii, H. Takazaki, R. Ayukawa, Y. Nishikawa, I. Minoura, Y. Hachikubo, G. Kurisu, K. Sutoh, T. Kon, K. Namba, E. Muto, A flipped ion pair at the dynein-microtubule interface is critical for dynein motility and ATPase activation, J Cell Biol, 208 (2015) 211-222.
146