Here, we present a protocol for reconstituting microtubule bundles in vitro and directly quantifying the forces exerted within them using simultaneous optical trapping and total internal reflection fluorescence microscopy. This assay allows for nanoscale-level measurement of the forces and displacements generated by protein ensembles within active microtubule networks.
Microtubule networks are employed in cells to accomplish a wide range of tasks, ranging from acting as tracks for vesicle transport to working as specialized arrays during mitosis to regulate chromosome segregation. Proteins that interact with microtubules include motors such as kinesins and dynein, which can generate active forces and directional motion, as well as non-motor proteins that crosslink filaments into higher-order networks or regulate filament dynamics. To date, biophysical studies of microtubule-associated proteins have overwhelmingly focused on the role of single motor proteins needed for vesicle transport, and significant progress has been made in elucidating the force-generating properties and mechanochemical regulation of kinesins and dyneins. However, for processes in which microtubules act both as cargo and track, such as during filament sliding within the mitotic spindle, much less is understood about the biophysical regulation of ensembles of the crosslinking proteins involved. Here, we detail our methodology for directly probing force generation and response within crosslinked microtubule minimal networks reconstituted from purified microtubules and mitotic proteins. Microtubule pairs are crosslinked by proteins of interest, one microtubule is immobilized to a microscope coverslip, and the second microtubule is manipulated by an optical trap. Simultaneous total internal reflection fluorescence microscopy allows for multichannel visualization of all the components of this microtubule network as the filaments slide apart to generate force. We also demonstrate how these techniques can be used to probe pushing forces exerted by kinesin-5 ensembles and how viscous braking forces arise between sliding microtubule pairs crosslinked by the mitotic MAP PRC1. These assays provide insights into the mechanisms of spindle assembly and function and can be more broadly adapted to study dense microtubule network mechanics in diverse contexts, such as the axon and dendrites of neurons and polar epithelial cells.
Cells employ microtubule networks to perform a wide variety of mechanical tasks, ranging from vesicle transport1,2,3 to chromosome segregation during mitosis4,5,6. Many of the proteins that interact with microtubules, such as the molecular motor proteins kinesin and dynein, generate forces and are regulated by mechanical loads. To better understand how these critical molecules function, researchers have employed single-molecule biophysical methods, such as optical trapping and TIRF microscopy, to directly monitor critical parameters such as unloaded stepping rates, processivity, and force-velocity relationships for individual proteins. The most commonly used experimental geometry has been to attach motor proteins directly to trapping beads whose spherical geometry and size mimic vesicles undergoing motor-driven transport. Numerous kinesins, including kinesin-17,8,9, kinesin-210,11,12, kinesin-313,14,15,16 kinesin-517,18, kinesin-819,20, as well as dynein and dynein complexes21,22,23,24,25, have been studied with these methods.
In many cellular processes, however, motor and non-motor proteins use microtubules both as track and cargo26,27. Moreover, in these scenarios where microtubule filaments are crosslinked into higher-order bundles, these proteins function as ensembles rather than single units. For example, within dividing somatic cells, dense filament networks self-organize to build the mitotic spindle apparatus28,29,30. The interpolar spindle microtubule network is highly dynamic and is largely arranged with minus-ends pointing toward the spindle poles and plus-ends overlapping near the spindle equator. Filaments within the spindle are crosslinked by motor proteins such as kinesin-531,32,33, kinesin-1234,35,36, and kinesin-1437,38,39, or by non-motor proteins such as PRC140,41,42,43 or NuMA44,45,46. They frequently move or experience mechanical stress during processes such as poleward flux or while coordinating chromosome centering during metaphase or chromosome segregation during anaphase47,48,49,50,51,52. The integrity of the micron-scale spindle apparatus through mitosis, therefore, relies on a carefully regulated balance of pushing and pulling forces generated and sustained by this network of interacting filaments. However, the tools needed to probe this mechanical regulation and explain how protein ensembles work in concert to coordinate microtubule motions and produce the forces needed to properly assemble the spindle have only recently been developed, and we are just beginning to understand the biophysical rules that define dynamic microtubule networks.
The goal of this manuscript is to demonstrate the steps required to reconstitute crosslinked microtubule pairs in vitro, immobilize these bundles in a microscopy chamber that allows for simultaneous fluorescence visualization of both the microtubules and crosslinking proteins and nanoscale force measurement, and process these data robustly. We detail the steps needed to stably polymerize fluorescence-labeled microtubules, prepare microscope coverslips for attachment, prepare polystyrene beads for optical trapping experiments, and assemble crosslinked filament networks that preserve their in vivo functionality while allowing for direct biophysical manipulation.
1. Preparation of microtubules
NOTE: When employing GFP-labeled crosslinking proteins, red (e.g., rhodamine) and far-red (e.g., biotinylated HiLyte647, referred to as biotinylated far red in the rest of the text) organic fluorophore labeling of the microtubules works well. Minimal crosstalk between all three channels can be achieved during imaging by using a high-quality quad band total internal reflection fluorescence (TIRF) filter.
2. Preparation of passivated coverslips
3. Preparation of kinesin-coated beads
4. Assembly of the microscopy chamber
5. Imaging microtubule bundles with 3-color TIRF
6. Performing optical trap experiments on microtubule bundles
7. Analysis of data and correlation of fluorescence images with optical trap records
NOTE: To optimize data collection, it is beneficial to employ two separate computer control systems: one for the optical trapping software and another for the fluorescence imaging. This setup allows for high-speed data acquisition in both experimental modalities and eliminates nano- and microsecond delays in operation execution being introduced to the data, which can arise when using a single CPU.
The preparation of microtubule bundles suitable for biophysical analysis is considered successful if several of the key criteria are met. First, imaging in three colors should reveal two aligned microtubules with a concentration of crosslinking protein preferentially decorating the overlap region (Figure 5B,C and Figure 6B). Ideally, the distance between the overlap edge and the free end of the rhodamine microtubule should be at least 5 µm to provide sufficient physical space between the trapping beam and the fluorescence-labeled proteins. Second, there should be minimal background fluorescence signal in regions that lack biotinylated far-red microtubules, particularly from the same channel that the crosslinking protein is labeled with. High background fluorescence is indicative of poor coverslip passivation and a high concentration of glass-bound proteins, which will likely interact non-specifically with the microtubules or trapping bead. Additionally, the presence of small round puncta or short fragments seen in either of the microtubule imaging channels suggests that microtubule polymerization and clarification were unsuccessful or that the microtubules were aged or damaged.
Third, the beads used for trapping should appear as single beads and not within clumps containing many beads. It is likely that a small fraction of beads will irreversibly adhere to the surface, but a significant fraction should be observed as individual particles diffused in the sample chamber. Excessive clumping can often be alleviated by sonicating the beads for at least 10 min prior to flowing into the chamber. Fourth, attaching a bead to a microtubule should result in good attachment, with the bead remaining bound when the trapping laser light is shuttered. The beads should not stick non-specifically when brought into contact with the surface, and, once a bead is attached to the microtubule, it should be possible to lightly manipulate (e.g., bend or flex) the filament prior to measurement. Finally, it should be possible to observe changes in the overlap length when force is applied, or motor activity is allowed to proceed. For example, if observing motor protein-driven sliding, the overlap length should decrease at a rate consistent with the rate of motor protein stepping. If examining passive crosslinkers, the application of force should result in a change in the overlap length. These outputs indicate that the rhodamine microtubule is attached only to the surface-bound microtubule via crosslinking proteins, rather than non-specifically sticking to the coverslip surface.
When these criteria are all met, it is possible to perform a wide range of experiments to extract the essential biophysical parameters that define ensemble mechanics. This assay or similar assays have been used extensively to examine mitotic crosslinking proteins in previous studies. For example, we have shown that ensembles of the essential mitotic motor protein kinesin-5 can regulate microtubule sliding by generating both pushing and braking forces that scale linearly with overlap length64 (Figure 5). Microtubules in either an antiparallel or parallel geometry were crosslinked by a small ensemble of kinesin-5 molecules. As these motor proteins stepped toward the microtubule plus-ends, the filaments were either slid apart (antiparallel) or rapidly fluctuated back-and-forth (parallel). By monitoring the mechanics within this mini-spindle geometry, we found the magnitude of force scaled both with the length of filament overlap, as well as the number of crosslinking motor proteins. When microtubules were moved at a velocity faster than kinesin-5's unloaded stepping rate, the motor proteins provided a resistive braking force. It was also found that the C-terminal tail domain is required for efficient crosslinking and force generation61 (Figure 5B,C). The full-length protein is able to generate sustained forces that plateau when all motor proteins reach their individual stall force (Figure 5D). However, the kinesin-5 motor lacking its C-terminal tail generates maximum forces that are nearly five-fold smaller in magnitude (Figure 5E,F). Together, these results reveal how kinesin-5 proteins work in ensembles to push microtubules poleward during spindle assembly and elucidate the biophysical function of essential regulatory protein domains within the molecule.
We have also demonstrated that ensembles of the non-motor mitotic protein PRC1 operate as a mechanical dashpot to resist microtubule sliding in anaphase spindle midzones54 (Figure 6A–C). PRC1 generates frictional resistance that scales linearly with pulling speed, just as a mechanical dashpot produces velocity-dependent resistive forces. These resistive forces do not depend on the length of overlap regions between microtubule pairs or the local crosslinker density, but they do strongly depend on the total concentration of engaged PRC1 molecules (Figure 6D,E). From these results, it is proposed that PRC1 ensembles act as a leaky piston, wherein compression of diffusive crosslinkers produces a velocity-dependent resistance, but the loss of crosslinkers at microtubule plus-ends alleviates large compressive forces.
Similar assay geometries were also used to examine the mitotic kinesin-12 protein Kif1562. By measuring the force produced as microtubules were pushed apart, Reinemann et al.62 found that Kif15 ensembles can push apart filaments up to a critical plateau force and require the C-terminal tail tether region of the protein in order to efficiently crosslink microtubules and build sustained loads. Reinemann et al. also elegantly demonstrated that the kinesin-14 HSET, a minus-end directed kinesin, can similarly slide apart antiparallel microtubules63. When mixed with equal amounts of the plus-end directed kinesin-5 Eg5 protein, HSET serves to inhibit Eg5 sliding activity, engaging in a tug-of-war for microtubule sliding motion within the overlap region. Together, these results all demonstrate the power of direct force measurement across active microtubule bundles and make clear the need for characterizing protein function in the biologically appropriate network geometry.
Figure 1: Passivation of glass coverslips. (A) Schematic depicting the sequential wash steps, drying, and conjugation of No. 1.5 glass coverslips for amino-silane conjugation. (B) Schematic depicting the protocol for covalently linking PEG and biotin-PEG groups to the glass coverslip, washing and drying, and sealing the coverslips under vacuum for short-term storage. Some portions of the schematic are modified images from 65. Please click here to view a larger version of this figure.
Figure 2: Assembly of a sample chamber. (A) Schematic depicting the assembly of a sample flow chamber using passivated coverslips and a standard 3 in microscope slide. (B) Assembly and typical dimensions of single- and dual-channel flow chambers that are optimized for optical trapping and fluorescence imaging of microtubule bundles. Some portions of the schematic are modified images from 65. Please click here to view a larger version of this figure.
Figure 3: Generation of surface-immobilized microtubule bundles for optical trapping and TIRF-based imaging. Schematics depicting the stepwise assembly of optically trappable microtubule bundles. (A) Reagents are flowed in from the entry port of the chamber and fluid is wicked out from the exit channel. (B) Streptavidin (blue) first binds to the biotin-PEG sites on the coverslip, and (C) casein (red) is introduced as an additional blocking agent. (D) Biotinylated far red-labeled microtubules (magenta) are introduced and allowed to incubate for ~5 min, followed by the addition of (E) crosslinking protein (green), (F) non-biotinylated rhodamine-labeled microtubules (red), and, finally, (G) rigor kinesin-coated microspheres suspended in the appropriate reaction buffer for attachment to the bundle and optical trap-based manipulation. (H) The chamber is sealed with clear nail polish to prevent evaporation of the sample buffer during experiments. Some portions of the schematic are modified images from 65. Please click here to view a larger version of this figure.
Figure 4: Schematic of the optical tweezers/TIRF microscope system. A single-beam optical trap is introduced into the optical path of an objective-based TIRF imaging system within an inverted microscope. A short pass filter inserted just below the objective allows the trap beam to be directed into the back aperture of the objective, where it will be focused just above the surface of the sample coverslip, forming an optical trap. A position-sensitive photodetector will collect the transmitted light, allowing for position and force detection of the bead. Multiple channels of fluorescence data can be acquired via TIRF imaging and recorded on a high-sensitivity EMCCD or sCMOS camera. A broad-spectrum light source is used to image trapping beads during the setup of the experiments. Please click here to view a larger version of this figure.
Figure 5: Representative example of measuring force production by kinesin-5. (A) Schematic depicting the experimental geometry, showing microtubules crosslinked by kinesin-5 motor proteins that generate force as they slide the filaments apart, which is measured with an optical trap. Representative fluorescence images of bundles composed of surface-immobilized microtubules, GFP-kinesin-5, and rhodamine-labeled transport microtubules. (B) Full-length and (C) C-terminal tail truncated kinesin-5 constructs were employed. Scale bars = 5 µm. Sample records of force production for (D) full-length and (E) tailless are shown, with average force plotted as a function of time. (F) Plots of the maximum plateau force and corresponding overlap length for both constructs are shown, revealing that the full-length kinesin construct generates overlap-length dependent forces that are substantially larger in magnitude than the tailless kinesin construct. This figure has been modified from61. Please click here to view a larger version of this figure.
Figure 6: Representative example of measuring frictional forces by PRC1. (A) Schematic depicting the experimental geometry, showing microtubules crosslinked by GFP-PRC1 (protein regulator of cytokinesis) crosslinking proteins that generate resistance as the filaments are slid apart by moving the coverslip at constant velocity. (B) Representative time series fluorescence images showing the moving surface-immobilized far-red microtubule, GFP-PRC1 molecules condensing within the shrinking overlap, and the free rhodamine microtubule held with the bead. Time interval between frames = 6 s; Scale bar = 5 µm. (C) Example force trace showing frictional force during the sliding event, with disruption event and force drop to zero (~55 sec) once the overlap reached 0 µm. (D) Correlation of mean force and integrated GFP intensity within the overlap at four different sliding velocities. (E) Mean slopes and calculated fit errors of traces in (D) reveal that the force per PRC1 molecule increases linearly as a function of sliding velocity. This figure has been modified from66. Please click here to view a larger version of this figure.
Microtubule networks are employed by myriad cell types to accomplish a wide range of tasks that are fundamentally mechanical in nature. In order to describe how cells function in both healthy and disease states, it is critical to understand how these micron-scale networks are organized and regulated by the nanometer-sized proteins that collectively build them. Biophysical tools such as optical tweezers are well suited to probing the mechanochemistry of key proteins at this scale. Reflecting the diversity of microtubule network function, complex experimental geometries that employ optical tweezers have been explored, including force-dependent analyses of microtubule tip dynamics67,68, the generation of forces by kinetochore components69,70,71, and two-beam dumbbell trap geometries to probe microtubule filament mechanics72,73. In this protocol, we have described novel methods for reconstituting fundamental microtubule network motifs such as bundles and applying the biophysical tools of single-molecule fluorescence microscopy and optical trapping to assess critical information about cellular function.
While most individual steps in this protocol are technically straightforward, taken together there are numerous potential points of failure that can arise, and substantial care must be taken at each point to ensure successful measurements. First, as with many microscope-based single-molecule experimental assays, proper surface passivation is key, as non-specific binding of proteins or beads to the coverslip surface nearly always prevents the successful completion of these experiments. Second, the expression and purification of the crosslinking protein(s) of interest must be optimized to ensure high quality single-species products, as contaminants can interfere with the assay in numerous ways, such as inducing excess bundling of filaments or interfering with bead and microtubule attachment, not to mention introducing complexities in the interpretation of force data. Third, maintaining robust bead-microtubule attachment requires high quality rigor kinesin motors bound at high density to the trapping bead, such that multiple kinesin-filament attachment points can be made. These bonds can withstand 10 s of pN of force for several minutes, which is suitable for a wide range of potential systems of study employing this technique.
We also note that there are many ways in which this protocol can be modified to best suit the end user's instrumentation or biological system needs. While we have described experiments that employ GFP-labeled crosslinking proteins and red/far red-labeled microtubules, it is also possible to swap fluorescent labeling as needed. For example, if the user does not have enough laser lines to perform three-color TIRF microscopy, it is possible to get high quality data by differentially labeling microtubules with the same fluorophore but using different concentrations (e.g., dim vs. bright) for each species of microtubule. Similarly, it may not always be possible or advantageous to add a fluorescent label to the crosslinking protein. In this situation, certain experimental information such as crosslinker concentration or localization would not be accessible, but the determination of parameters such as microtubule overlap length could still be made. We anticipate that this protocol is highly adaptable for many different types and combinations of microtubule crosslinking proteins. The introduction of multiple crosslinking proteins would likely necessitate either the removal of fluorescence tags from one of the proteins or the exclusion of fluorescent labels from one of the two microtubule types to free up imaging capabilities in an appropriate bandwidth (e.g., using a red or far-red protein-encoded label for the crosslinking protein). It is unlikely that more than three fluorescent channels could be employed with this TIRF-based assay, though there is certainly flexibility in how those are distributed, which should be carefully considered when planning an experiment. Finally, it is likely that this assay could be modified to study different network geometries and types of cytoskeletal filament. The analysis of the mechanics of microtubule branching, mediated by augmin and the gamma-tubulin ring complex, could readily be probed and imaged. In addition, other crosslinked cytoskeletal networks, such as those involving actin, septins, or intermediate filaments, would be quite amenable to study with these tools, assuming proper modifications and orthogonal attachment strategies to the coverslip and trapping bead are made.
In conclusion, we have described a protocol for the reconstitution of active force-generating microtubule networks, which can be imaged using single-molecule fluorescence microscopy tools and manipulated via a single-beam optical trap. We have demonstrated the usefulness of this method with datasets that reveal the ensemble mechanics of both a mitotic motor and non-motor protein. We anticipate that many types of highly organized microtubule networks, such as those found in neurons, epithelial tissue, or cardiomyocytes, can be analyzed with these tools, revealing new principles of biophysical function for the dynamic cytoskeleton.
The authors have nothing to disclose.
The authors wish to acknowledge support from R21 AG067436 (to JP and SF), T32 AG057464 (to ET), and Rensselaer Polytechnic Institute School of Science Startup Funds (to SF).
10W Ytterbium Fiber Laser, 1064nm | IPG Photonics | YLR-10-1064-LP | |
405/488/561/640nm Laser Quad Band Set for TIRF applications | Chroma | TRF89901v2 | |
6x His Tag Antibody, Biotin Conjugate | Invitrogen | #MA1-21315-BTIN | |
Acetone, HPLC grade | Fisher Scientific | 18-608-395 | |
Alpha casein from bovine milk | Sigma | 1002484390 | |
ATP | Fisher Scientific | BP413-25 | |
Benzonase | Novagen | 70746-3 | |
Biotin-PEG-SVA-5000 | Laysan Bio, Inc. | NC0479433 | |
BL21 (DE3) Rosetta Cells | Millipore Sigma | 71-400-3 | |
Catalase | MP Biomedicals LLC | 190311 | |
CFI Apo 100X/1.49NA oil immersion TIRF objective | Nikon | N/A | |
Chloramphenicol | ACROS Organics | 227920250 | |
Coverslip Mini-Rack, for 8 coverslips | Fisher Scientific | C14784 | |
Delicate Task Wipers | Kimberly-Clark | 34120 | |
Dextrose Anhydrous | Fisher Scientific | BP3501 | |
D-Sucrose | Fisher Scientific | BP220-1 | |
DTT | Fisher Scientific | BP172-25 | |
Ecoline Immersion Thermostat E100 with 003 Bath | LAUDA-Brinkmann | 27709 | |
EDTA | Fisher Scientific | BP118-500 | |
EGTA | Millipore Corporation | 32462-25GM | |
FIJI / Image J | https://fiji.sc/ | N/A | |
Frosted Microscope Slides | Corning | 12-553-10 | 75mmx25mm, with thickness of 0.9-1.1mm |
Glucose Oxidase | MP Biomedicals LLC | 195196 | Type VII, without added oxygen |
GMPCPP | Jena Biosciences | JBS-NU-405S | Can be stored for several months at -20 °C and up to a year at -80 °C |
Gold Seal-Cover Glass | Thermo Scientific | 3405 | |
HEPES | Fisher Scientific | BP310-500 | |
Imidazole | Fisher Scientific | 03196-500 | |
IPTG | Fisher Scientific | BP1755-10 | |
Laboratory dessicator | Bel-Art | 999320237 | 190mm plate size |
Kanamycin Sulfate | Fischer Scientific | BP906-5 | |
KIF5A K439 (aa:1-439)-6His | Gilbert Lab, RPI | N/A | doi.org/10.1074/jbc.RA118.002182 |
Kimwipe | Kimberley Clark | Z188956 | lint-free tissue |
Immersion Oil, Type B | Cargille | 16484 | |
Lens Tissue | ThorLabs | MC-5 | |
LuNA Laser launch (4 channel: 405, 488, 561, 640nm) | Nikon | N/A | |
Lysozyme | MP Biomedicals LLC | 100834 | |
Magnesium Acetate Tetrahydrate | Fisher Scientific | BP215-500 | |
Microfuge 18 | Beckman Coulter | 367160 | |
MPEG-SVA MW-5000 | Laysan Bio, Inc. | NC0107576 | |
Neutravadin | Invitrogen | PI31000 | |
Nikon Ti-E inverted microscope | Nikon | N/A | Nikon LuN4 Laser |
Ni-NTA Resin | Thermo Scientific | 88221 | |
Oligonucleotide – CACCTATTCTGAGTTTGCGCGA GAACTTTCAAAGGC |
IDT | N/A | |
Oligonucleotide – GCCTTTGAAAGTTCTCGCGCAA ACTCAGAATAGGTG |
IDT | N/A | |
Open-top thickwall polycarbonate tube, 0.2 mL, 7 mm x 22 mm | Beckman Coulter | 343755 | |
Optima-TLX Ultracentrifuge | Beckman Coulter | 361544 | |
Paclitaxel (Taxol equivalent) | Thermo Fisher Scientific | P3456 | |
PIPES | ACROS Organics | 172615000 | |
PMSF | Millipore | 7110-5GM | |
Porcine Tubulin, biotin label | Cytoskeleton, Inc. | T333P | |
Porcine Tubulin, HiLyte 647 Fluor | Cytoskeleton, Inc. | TL670M | far red labelled |
Porcine Tubulin, Rhodamine | Cytoskeleton, Inc. | TL590M | |
Porcine Tubulin, Tubulin Protein | Cytoskeleton, Inc. | T240 | |
Potassium Acetate | Fisher Scientific | BP364-500 | |
Prime 95B sCMOS camera | Photometric | N/A | |
Quadrant Detector Sensor Head | ThorLabs | PDQ80A | |
Quikchange Lightning Kit | Agilent Technologies | 210518 | |
Sodium Bicarbonate | Fisher Scientific | S233-500 | |
Sodium Phosphate Dibasic Anhydrous | Fisher Scientific | BP332-500 | |
Square Cover Glasses | Corning | 12-553-450 | 18 mm x 18 mm, with thickness of 0.13-0.17 mm |
Streptavidin Microspheres | Polysciences Inc. | 24162-1 | |
Superose-6 Column | GE Healthcare | 29-0915–96 | |
TCEP | Thermo Scientific | 77720 | |
TLA-100 Fixed-Angle Rotor | Beckman Coulter | 343840 | |
Ultrasonic Cleaner (Sonicator) | Vevor | JPS-08A(DD) | 304 stainless steel, 40 kHz frequency, 60 W power |
Vectabond APTES solution | Vector Laboratories | SP-1800-7 | |
Windex Powerized Glass Cleaner with Ammonia-D | S.C. Johnson | SJN695237 |