|Home | About | Journals | Submit | Contact Us | Français|
Kinetochores are multifunctional supercomplexes that link chromosomes to dynamic microtubule tips. Groups of proteins from the kinetochore are arranged into distinct subcomplexes that co-purify under stringent conditions and cause similar phenotypes when mutated. By co-expressing all the components of a given subcomplex from a polycistronic plasmid in bacteria, many labs have had great success in purifying active subcomplexes. This has enabled the study of how the microtubule binding subcomplexes of the kinetochore interact with both the microtubule lattice and dynamic microtubule tips. Here we outline methods for rapid cloning of polycistronic vectors for expression of kinetochore subcomplexes, their purification, and techniques for functional analysis using Total Internal Reflection Fluorescence Microscopy (TIRFM).
The faithful transmission of genetic material during cell division is essential for all life. In eukaryotes, this process utilizes a large molecular machine termed the mitotic spindle. Central to the process is the kinetochore, which mediates the link between chromosomes and microtubules. The kinetochore, however, is not simply a linker. Beyond providing a physical connection, the kinetochore controls chromosome positioning, regulates its attachment to microtubules and delays mitotic progression until properly bioriented. This multifunctional aspect is evident in its complexity with around 80 distinct proteins (Cheeseman and Desai, 2008). Because of this complexity, unraveling the roles the different proteins play and the mechanisms they use to achieve them is challenging. Here we describe methods to systematically analyze one of the most basic functions of the kinetochore, attachment to dynamic microtubule tips.
Microtubule binding assays have identified subcomplexes of the core kinetochore that bind directly to microtubules (Cheeseman et al., 2006; Miranda et al., 2005; Wei et al., 2007). These studies, however, only provide a static picture. The mechanism that enables kinetochores to remain attached to constantly remodeling microtubule tips where thousands of tubulin subunits are assembled and disassembled is perplexing. To address this question we have employed advanced biophysical techniques. Through total internal reflection fluorescence microscopy (TIRFM), we have characterized how kinetochore subcomplexes interact with both the microtubule lattice and dynamic microtubule tips at the single molecule level (Gestaut et al., 2008; Powers et al., 2009). Using a feedback-controlled optical trap, we have demonstrated that these subcomplexes can remain coupled to dynamic microtubule tips while under piconewton scale forces similar to those experienced by kinetochores in vivo (Asbury et al., 2006; Franck et al., 2007; Powers et al., 2009). The methods used for these optical trapping measurements are described in Franck et al. (2010). Collectively, these functional assays illuminate the mechanisms that allow kinetochores to maintain a floating grip on microtubule tips.
In our pursuit of these goals we have improved several technologies. Here we outline methods for rapid cloning of polycistronic vectors for expression of kinetochore subcomplexes, their purification, and techniques for functional analysis using TIRFM. Although the techniques are described in the context of studying kinetochore subcomplexes, many of them could be more broadly applied to study of other recombinant protein subcomplexes. In this chapter, a general knowledge of microtubules is assumed. For background we point the reader to the Mitchison lab website1, which has detailed protocols for the purification, labeling and polymerization of tubulin. Finally, where appropriate, the Hec1/Ndc80 subcomplex is used as an example for protocols also applicable to other kinetochore subcomplexes.
The number of identified kinetochore proteins has dramatically increased over the past decade to around80. The proteins are arranged in distinct subcomplexes that copurify under stringent conditions. These subcomplexes likely represent functional pieces of the kinetochore as mutations in different proteins of a given subcomplex often lead to similar phenotypes. Recombinant expression of individual proteins from a given subcomplex often results in poor expression levels, low solubility, and co-purification of chaperones. However in studies where the subunits of all the proteins of a subcomplex are co-expressed from a polycistron in bacteria, functional soluble complexes have been obtained (Cheeseman et al., 2006; Hori et al., 2008; Kline et al., 2006; Miranda et al., 2005; Wei et al., 2005).
A system for cloning into a polycistronic vector was developed by Tan (2001) in which genes are first cloned into a transfer vector containing upstream signals necessary for expression in the polycistron. These signals include a translational enhancer and Shine-Dalgarno sequence, which drive efficient binding and activation of ribosomes. On either end of the insertion site are a series of unique restriction sites that allow the subcloning of each gene with the flanking signals into a polycistronic vector (Figure 1A). Planning is very important as the first gene must be cloned using the most internal restriction sites in the transfer vector and go into the first position in the polycistron, the second gene in the second position with the second set of restriction sites, etc. We have found that this approach is cumbersome, and have simplified the cloning by adding upstream signals for translation to the primer used to amplify the desired gene via PCR (Figure 1B). This avoids the need for the transfer vector, and gives the researcher freedom to insert genes into the polycistron in any order (Figure 1C).
Many other considerations made during the cloning of a polycistronic vector can lead to more successful purifications. In our hands, better expression is achieved by placing the larger genes in the more 3′ cassettes of the polycistron. This may be because ribosomes are more effectively loaded, or funneled, from the shorter transcripts onto the ribosome binding sites for the longer genes(Schoner et al., 1990)The affinity tag used to purify the complex should be inserted on a protein known to tolerate tags in vivo. Moreover, an affinity tag on the end of the protein most susceptible to degradation can allow separation of the intact protein from degraded versions. If possible, the affinity tag should be inserted on the protein expressed at the lowest levels. If all the other proteins are in excess, then any complexes containing the affinity tagged protein should be complete. If instead the affinity tagged protein is expressed at levels significantly greater than its binding partners, purification will result in a variety of incomplete complexes that may be difficult to separate from the complete complex. In addition to the affinity tag, we also add a restriction site for insertion of a fluorescent protein tag in frame. Again, the fluorescent tag should be inserted at a place known to be tolerated in vivo. Because of the size of the polycistron an 8 bp restriction site will have a greater chance of being unique to both the genes already in the polycistron and in tags to be added later.
Many of the same considerations required for expression of individual proteins also apply to expression from polycistronic vectors. Because eukaryotic genes contain codons that are rare in bacteria, the coexpression of tRNAs for these codons is essential for robust expression. Many bacterial strains are available for this purpose. Our choice is BL21 Rosetta DE3 (Novagen, EMD Biosciences, San Diego CA), which includes an inducible T7 RNA polymerase and a plasmid that expresses both the rare tRNAs and the T7 lysozyme gene for easier lysis. For expression including proteins tagged with GFP (or its equivalents) induction at a low growth temperature (23°C) enhances folding. (Remember that GFP was originally cloned from a jelly fish collected at Friday Harbor Laboratories in Washington State, where the waters average about 10°C). Low temperature inductions also turn on the expression of chaperones, which can aid in folding.
The polycistronic vector is particularly susceptible to recombination because it contains homologous sequences upstream of each gene. Ostensibly, genes that produce toxic products will have both the opportunity, and selective pressure to be excised by homologous recombination. For this reason, it is essential to use cloning strains that lack the gene for recombinase, the recA gene. Most strains used for expression, however, contain a functional recombinase because its absence causes slow growth. Although we have experienced little trouble so far from recombination during expression, a recombinase-free expression strain has recently been created (Rosettablue -Novagen) and could be considered for expression.
TIRFM has been used extensively to study the dynamics of interactions between proteins and polymers such as DNA, actin and microtubules. In TIRFM, excitation lasers are aimed at an angle such that the beam is completely reflected at the interface between two mediums (usually between a coverslip and the aqueous solution within a flow channel above the coverslip). This creates an evanescent field that decays exponentially so as to preferentially excite fluorophores within ~ 100 nm of the coverslip. This technique eliminates background fluorescence caused by excitation of fluorophores outside the plane of focus, which would normally overwhelm the signal from individual fluorophores, providing single molecule resolution. TIRFM, therefore, is a vital tool for characterizing how individual kinetochore subcomplexes interact with the microtubule lattice and dynamic microtubule tips.
To observe the interaction of kinetochore subcomplexes with microtubules we needed a two-color TIRFM setup. We chose to use a far-red laser (FTEC-635-0-25-PFQ, Blue Sky Research, Milpitas CA, USA) that excites either Cy5 or Alexa-647 for visualizing microtubules with some of their subunits covalently modified with these dyes, and a blue laser (473–30, LaserPath Technologies, Oviedo FL, USA; or, more recently, Sapphire 488–75, Coherent, Santa Clara CA, USA) for the excitation of GFP-tagged kinetochore subcomplexes. There are three major steps required for this setup, the alignment of the two lasers before entering the microscope, the building of the TIRF microscope itself, and the capture of the images from the two laser channels. Although commercial setups are available for this application, we chose to build our own which was both cost-effective and provided greater flexibility.
To align two lasers, both the angle at which the beams are aimed and the position of the beams must be the same as they enter the microscope. Control of these parameters can either be accomplished through a series of adjustable lenses and mirrors, or by the use of fibre optics, which allows both to be controlled. We chose not to use fibre optics which are more difficult to align and more susceptible to loss of alignment. The final merged beams are passed through an iris that is adjusted to allow just enough light through to excite the full field of view in the microscope. This field iris eliminates unusable light that could potentially scatter and contaminate the otherwise collimated illumination beam. Finally, the intensity of a laser beam has a Gaussian distribution, and therefore the width of the beams for both lasers should be optimized so that the change in intensity from the center of the opening in the field iris to the edge is the smallest possible (largest overall beam width), while not causing too much loss in overall beam intensity. (If the beam is widened too far it will not be effective for exciting fluorophores.) Because the lasers we used are not perfectly monochromatic, we also added filters for each laser beam upstream of the microscope to ensure a narrow range of wavelengths for excitation.
TIRFM requires that the illumination light reaching the sample is well collimated and at an angle greater than the critical angle, such that the light undergoes total internal reflection at the coverslip/sample interface. For TIRF, the objective must have a numerical aperture (NA) greater than 1.40 and a NA > 1.45 is recommended as it allows for greater flexibility in the depth of the evanescent field that illuminates the sample at the coverslip surface. (Currently, we use a Nikon objective, CFI Planapochromat. 100× 1.49NA.) The light striking the interface must be collimated otherwise the illumination beam will have a range of incidence angles leading to a less defined depth of field. To accomplish this, the beam is brought to a tight focus (a ‘waist’) at the back focal plane of the objective (for TIRF objectives this is usually just 2–3 mm from the front surface of the objective). Each objective will have its own parameters for alignment, however, one requirement that is almost universally true is that the final lens that focuses the illumination beam at the back focal plane should be as close as possible to the objective (150 mm or less). This ensures a tightly focused beam at the back focal plane (and thus a tightly collimated beam leaving the objective).
As mentioned earlier, the depth of the evanescent field created by the TIRFM scope is dependent upon the angle of incidence. This angle is dependent on the NA of the objective and how far from the centerline of the objective the illumination beam is focused. (The further from the centerline the greater the angle of incidence and shorter the depth of the evanescent field). For this reason, it is desirable to be able to control this distance. This can be accomplished by mounting a mirror into a translatable stage that can be externally controlled and a system of three lenses that maintain these translations while keeping the beam waist focused at the back focal plane of the objective (Figure 2). Two of these lenses downstream of the mirror compose a Keplerian telescope so that the translation of the mirror, which will initially be converted to angular diffractions by the first lens, is converted back to a translation by the second lens. The third lens, which is upstream of the translatable mirror, causes the collimated beam to diverge prior to reaching the first lens downstream of the mirror, this is required so that the beam leaving the final lens is focused at the focal plane, and not collimated at the focal plane.
To simultaneously collect images from both the far-red and green channels, we implemented a system that allowed both images to be projected side by side onto a CCD camera (iXon 887-BI, Andor Technology, Belfast, UK) (Figure 3). Although using a second camera would simplify the optical layout, the cameras are relatively expensive. Since the CCD is square, the first step was to pass the image leaving the microscope through an aperture that created a rectangular border at its plane of focus so that the component images would fit side by side onto the CCD. We next developed a system that utilizes two dichroic mirrors to split and rejoin the two separately colored images side-by-side onto the same camera CCD. The images are refocused using separate lenses within each path and each image also passes through an additional emissions filter. Because the images have slightly different path lengths, one image ends up being slightly magnified relative to the other. To avoid this problem an additional lens was added to the red path, which, in conjunction with the other lens, allowed both the focus and magnification to be controlled independently.
The major advantage of TIRFM stems from the elimination of background fluorescence. Because this requires imaging close to the coverslip surface, it is essential to minimize adsorption of fluorescently labeled protein to this surface. Non-specifically bound fluorescent molecules and aggregates make interactions with the microtubules difficult to distinguish from background binding. Transient nonspecific interactions with the coverslip can also affect diffusion rates, on-rates and off-rates. To avoid these issues the coverslips must be thoroughly cleaned and silanized. During silanization, silane molecules react with the glass surface and each other forming a covalently linked layer above the coverslip surface. The silane used is a functionalized form that has a linked group (in our case a short chain PEG) that ends up pointed away from the glass surface creating a hydrophilic environment at the coverslip surface in the perfusion chamber. The coverslips are dried under a steady flow of high purity nitrogen gas as silane reacts with oxygen. Silanized coverslips may also be stored under aninert gas such as nitrogen. Below we outline our protocol for coverslip silanization adapted from Cras et al.(1999) and Walba et al.(2004). Custom Teflon racks that have minimal contact points with the coverslips are used as Teflon can tolerate the various surface treatments.
As mentioned before, when studying complexes at single molecule resolution, small variations in the experimental setup can significantly alter the data. Factors that can vary between slides are:
Because of these inherent variations the experimental and control measurements should be done on the same slide. We have developed a slide design using standard disposable glass slides and coverslips that allows the generation of four perfusion chambers. The perfusion chambers are attached to a peristaltic pump to control the flow rate of buffers through the chambers and to achieve complete buffer exchange. Below we outline our protocol to produce four perfusion chambers per coverslip:
In TIRFM the area of experimentation is directly above the coverslip, so the experiment must be performed in this narrow depth of field. Directly adsorbing the microtubules to the glass slide will inhibit the action of some microtubule enzymes, notably microtubule depolymerases, and make it impossible to measure the behavior of dynamic microtubules. To restrict the microtubules to the evanescent field without directly adsorbing them to the glass, the microtubules must be tethered to the coverslip. The best method would employ a molecule that is inert to free tubulin but would rapidly tether dynamic extensions off of microtubule seeds during imaging. This will prevent the microtubule ends from “flopping” out of the evanescent field during imaging. We have used two methods, both of which take advantage of nonspecific adsorption of proteins to the coverslip. Although the silanization process we use has been carefully optimized, imperfections still exist creating spots where proteins can be nonspecifically and permanently adsorbed to the coverslip. The ratio of the tethering agent to a blocking protein (such as BSA) that can compete for binding to the imperfections is optimized to control the density of the tethering agent on the coverslip. To minimize the amount of tubulin subunits interacting with tethering agents, and maximize the accessibility of the microtubule lattice, the minimum amount of tethering agent density is used that still prevents movement of parts of the bound microtubules in and out of the narrow field of excitation above the coverslip. The first method uses a biotin-streptavidin-biotin linkage, while the second method uses a modified kinesin that binds to microtubules in a rigor state and does not release them in the presence of ATP/GTP(Rice et al., 1999).
The second method is quicker, and has the advantage that rigor kinesin only interacts with polymerized tubulin and therefore dynamic extensions are bound down (free biotinylated tubulin saturates streptavidin sites faster than polymerization of the tubulin, which prevents dynamic extensions from becoming tethered to the coverslip). A disadvantage of the second method is that the coiled-coil domain of the rigor kinesin can cause increased nonspecific interactions with the subcomplex being studied, to minimize these we use the lowest concentration of rigor kinesin that still prevents “floppiness” in the microtubules.
In single molecule studies, small changes in conditions can significantly alter the outcome of the experiment. We first optimize our buffers to prevent nonspecific binding. Nonspecific binding is detrimental in that it causes loss of your protein changing the actual working concentration (which can be significant when working at pM concentrations), and also speckles the background of your field of view making data analysis difficult. Dynamic parameters such as rate constants and diffusion coefficients can be altered by non-specific interactions with the experimental set-up. To prevent nonspecific binding we use blocking proteins such as BSA and κ-casein, which saturate imperfections in the coverslip silanization. Salt can also drive nonspecific hydrophobic interactions, and decreasing the ionic strength of the solution can help prevent these. Effective conditions for preventing nonspecific interactions seem to differ between complexes, and therefore must be optimized empirically. After finding an appropriate buffer, the concentration of the complex must also be optimized for single molecule studies. When concentrations are too low, binding events are infrequent, while at higher concentrations binding events (especially if diffusive) may overlap preventing their analysis. The on-rate and off-rate can be adjusted by complex concentration and ionic strength. Finally the optimal situation for studying subcomplexes requires a uniform species and therefore conditions in which your complex exists in a near-homogeneous state should be identified if possible (ie monomer, dimer etc…).
The simplest TIRF assay uses microtubules stabilized by paclitaxel. Prior to the assays, microtubules are assembled with a 1:100 ratio of Alexa-647 labeled to unlabeled tubulin subunit for visualization. After polymerization microtubules are stabilized by the addition of paclitaxel, pelleted, and then resuspended in BRB80 with paclitaxel. This pelleting step removes the unpolymerized tubulin, which can interact with the subcomplexes being studied during TIRFM. The microtubules are bound to the coverslip as described above, washed with BRB80 plus a blocking protein, and washed again with BRB80 plus a blocking protein and free oxygen scavenger system to slow photo bleaching.
Steps listed are for all perfusion chambers, and during any incubation step the solution for the next step should be applied to the well to prevent the well from drying.
The raw data from TIRF microscopy are a stack of TIFF files. Movies are collected at 10 frames per second normally for 200 seconds. As described above, two images are projected onto the camera, on the top half is the green channel, and on the bottom half is the far-red channel. To analyze the raw data we have developed a series of three programs in Labview, which correspond to three separate analysis steps. In the first step, the far-red channel is translated to overlap with the green channel, allowing the position of the microtubules to be seen in reference to the kinetochore sub-complex being studied. This allows the position of the microtubule to be traced by the user with a multi-segment line onto the green channel (Figure 5A), and a kymograph is generated by summing the brightness of the four pixels above and below the line. In step two, the user identifies binding events in the generated kymograph, and draws a line along the kymograph where the binding event occurs. The line is automatically adjusted to lie on the brightest pixel within five pixels of the line drawn by the user, and frames where the brightness does not surpass a defined threshold are omitted (Figure 5B). Finally, in step three the coordinates obtained in step two are used in the original TIFFstack and the brightest pixel within six pixels of this coordinate is identified. Around this pixel a 6X6 box is drawn and a brightness measurement is made by summing the pixels. To account for background fluorescence, an additional 20X20 box is drawn around the pixel and the average brightness of these pixels excluding the inner 6X6 box is used as a measurement of background per pixel (Figure 5C). From step three a file for each binding event is created containing the position, brightness and background brightness for each frame. The binding events are then loaded into Igor for quantification and further analysis.
Plots of brightness versus frame number are created for each binding event allowing the start and end times of the event to be confirmed, and bleach steps to be identified. From these values we generate a histogram of residence times. This distribution is often well-described by a single exponential decay except for the lowest bin, where events are likely missed due to the finite time resolution of the instrument. The mean from the best-fit exponential (excluding the lowest bin), or equivalently, its time constant, τ, is inverted to give the off-rate, koff = τ−1 (e.g., in s−1). For each binding event, we also calculate a mean squared displacement along the microtubule long-axis, <x2>, for every possible time lag, Δt. The <x2> values are averaged across many events to generate a plot of <x2> versus Δt for a population. With this method, individual events contribute no more than one value to the population <x2> at a particular Δt, so long events are not weighted disproportionately. The one-dimensional diffusion coefficient, D, is then computed from the slope, m, of a linear fit to the plot according to D = 1/2 m = 1/2 <x2> Δt−1 (Berg, 1993).
A second-order on-rate can also be calculated from the TIRF Data. We do this by counting the number of binding events that occur per micron of microtubule lattice over a set period of time (e.g., in μm−1 s−1) and then converting to a second-order on-rate, kon (in M−1 s−1), by dividing by the total concentration of Ndc80 complex and by the total number of tubulin dimers per micron of microtubule lattice (1585 μm−1, assuming a 13-protafilament microtubule). This calculation gives a good approximation to the second-order on-rate if four criteria are met. First, the total concentration of Ndc80 must be well below the equilibrium dissociation constant, Kd, to ensure that binding does not significantly deplete the pool of free complexes. Second, only a small fraction of lattice sites on the microtubules must be occupied, so binding does not significantly deplete the availability of sites. Third, Ndc80 loss to non-specific background binding (e.g. to tube walls, etc) must be negligible. Fourth, the linkage of the microtubules to the coverslip must not block a significant fraction of lattice sites. The first two criteria are usually true given the very low concentrations used to obtain single molecule resolution. However, the last two are more difficult to prove, so these calculations must be seen as approximations. We also note that the apparent on-rate can be reduced by the ‘missing event problem’ mentioned above (i.e., some short-lived events, with residence times comparable to the temporal resolution of the instrument, will inevitably be missed). If the residence times are distributed exponentially, the magnitude of this effect can be estimated and, in principle, corrected for by comparing the number of short-lived events scored with the number predicted based on an exponential fit to the distribution.
We have outlined methods for the cloning, purification, and functional analysis of kinetochore subcomplexes. In the first section we discuss methods for rapid cloning of a polycistron and ways to enhance protein expression. Next, we describe our design of a two-color TIRF microscope setup. Finally, we describe experimental designs for TIRFM assays and optimizations to drive reproducibility and rapid data collection. The TIRFM assays described here are quite powerful for characterizing the interactions of subcomplexes with microtubules. Although not described here, the assays may also be used to determine stoichiometry and arrangements of larger assemblies, which should prove useful as more of the kinetochore is reconstituted. Finally, we note that theoptical trap is another powerful tool for assaying the strength of these interactions, which we describe in Methods (Franck et al., 2010).