We present protocols for simple actin filament microfluidic assays, in combination with fluorescence microscopy, that allow one to accurately monitor individual actin filaments in real-time while sequentially exposing them to different protein solutions.
In order to decipher the complex molecular mechanisms that regulate the assembly and disassembly of actin filaments, it is a great asset to monitor individual reactions live in well-controlled conditions. To do so, live single-filament experiments have emerged over the past 20 years, mostly using total internal reflection fluorescence (TIRF) microscopy, and have provided a trove of key results. In 2011, in order to further expand the possibilities of these experiments and to avoid recurring problematic artifacts, we introduced simple microfluidics in these assays. This study details our basic protocol, where individual actin filaments are anchored by one end to the passivated coverslip surface, align with the flow, and can be successively exposed to different protein solutions. We also present the protocols for specific applications and explain how controlled mechanical forces can be applied, thanks to the viscous drag of the flowing solution. We highlight the technical caveats of these experiments and briefly present possible developments based on this technique. These protocols and explanations, along with today's availability of easy-to-use microfluidics equipment, should allow non-specialists to implement this assay in their labs.
The assembly and disassembly of actin filaments and actin filament networks are controlled by several biochemical reactions and depend on the mechanical context. In order to gain insight into these complex mechanisms, it is invaluable to be able to observe individual reactions on individual filaments (in sufficiently large numbers). Over the past decades, the observation of dynamic actin filaments in real-time, mostly using total internal reflection fluorescence (TIRF) microscopy, has emerged as a key technique and has provided an impressive list of results that could not have been obtained with bulk solution biochemical assays1.
To achieve this, one needs to maintain fluorescently labeled actin filaments close to the surface of the microscope coverslip while exposing them to solutions of actin-binding proteins (ABPs), which can also be fluorescently labeled. Doing so provides a means to monitor events taking place on individual filaments in well-controlled biochemical conditions, and thus quantify reaction rates. However, a number of specific limitations should be considered. Artificially maintaining filaments close to the surface, often thanks to multiple anchoring points or by using a crowding agent such as methylcellulose, can alter their behavior (e.g., causing pauses in their polymerization and depolymerization2). Tracking the contour of each filament can be challenging, particularly if new filaments or filament fragments accumulate in the field of view over time. The reactions take place in a finite volume where the concentration of actin monomers and ABPs can vary over time, potentially making it difficult to derive accurate rate constants. Finally, renewing or changing the solution of ABPs is difficult to achieve in less than 30 s and will often lead to inhomogeneous protein content in the sample.
A bit over 10 years ago, inspired by what was already done to study individual Deoxyribonucleic Acid (DNA) strands3, we introduced a new technique based on microfluidics to observe and manipulate individual actin filaments4. It allows one to circumvent the aforementioned limitations of classical single-filament techniques. In these microfluidics assays, actin filaments are grown from spectrin-actin seeds adsorbed on the coverslip. Filaments are thus anchored by one end only to the bottom of the microfluidic chamber and fluctuate above the surface without sticking. Filaments align with the flow of incoming solutions, thereby easing the monitoring of their contour length and maintaining them in a shallow region above the coverslip where TIRF can be used. Different solutions are simultaneously flowed into the chamber without mixing, and the filaments can be exposed to them sequentially and rapidly.
Here, we propose a series of basic protocols to set up single-actin-filament microfluidics assays in the lab. Coverslips and microfluidics chambers can be prepared in advance (in half a day), and the experiment itself, where several biochemical conditions can be tested, is done in less than a day.
1. Microfluidic chamber preparation
2. Glass coverslip cleaning
NOTE: Here, a standard coverslip cleaning procedure, based on a series of sonication steps, is detailed. Other glass coverslip-cleaning procedures have been described in many other publications that can achieve similar satisfying results6,7,8,9.
3. PDMS chamber assembly
4. [OPTIONAL] Direct passivation and functionalization
NOTE: Depending on the application, chambers can be passivated and functionalized either once connected to the microfluidic controlling device (see Table of Materials) or by manually injecting solutions directly into the chamber with a pipette prior to its connection to the microfluidic device. The latter offers the advantage of consuming less reagent and avoiding potential contamination by flowing the solution through the polyether ether ketone (PEEK) tubing of the microfluidic device. In all the following steps, solutions are injected by directly sticking the pipette tip into the outlet. In order to avoid creating bubbles inside the chamber, make sure to have a tiny droplet sticking out of the pipette tip when plugging the tip into the outlet of the PDMS chamber. Likewise, remove the pipette tip before the entire volume has been injected.
5. Connect microfluidic device
NOTE: Use a pressure-based microfluidic system with up to four channels to control flows in the microfluidic chamber (Figure 1A, see Table of Materials). To avoid bubbles forming in the microfluidic tubing and perturbing flow stability, degas all solutions. Place 5 mL of dH20 and 10 mL of F-buffer stock in a vacuum desiccator connected to a vacuum pump (ultimate vacuum <250 mbar) and degas for at least 1 h at RT.
Figure 1: Injecting solutions through a microfluidic chamber. (A) Standard microfluidic setup for single actin filaments experiments. Protein solutions, placed in reservoirs 1-3, are pushed into the chamber by adjusting the pressure in the gas phase. The generated flow rates are measured by flow meters. Inside the microfluidic chambers, solutions do not mix and occupy space depending on the relative pressures applied (here, equal pressure on all inlets). Typical dimensions: reservoir tubes contain up to 2 mL of solution. PEEK tubing (0.25 mm inner diameter) connects the reservoirs to the flow meters (after 10 cm of tubing) and then to the PDMS chamber (after another 70 cm). Silicon tubing and stainless steel tubing couplers are used to connect the PEEK tubing to the PDMS inlets. The main microfluidic channel is 20-60 µm high, around 1 mm wide and 1 cm long. (B,C) Flow profiles inside the microfluidic chamber. (B) The fluid generates a parabolic profile across the chamber height: v(z) = 6z(h-z)R/h3w, where h and w are the chamber height and width, and R is the total flow rate. Bottom: Single actin filament polymerized from surface-anchored spectrin-actin seeds. (C) When the chamber width is considerably larger than its height, the flow is nearly uniform across the chamber, except at the PDMS surfaces, where it goes to zero. Please click here to view a larger version of this figure.
6. Configuring the setup with standard flow rates
NOTE: The computer-controlled pressure system allows easy and precise adjustment of the pressures of all inlets/outlet connected to the PDMS chamber, therefore the control of incoming and outcoming flow rates. Preset configurations can be saved and turned on/off with a single mouse click. Below are the recommended configurations (unless otherwise stated, outlet pressure is set to 0 mbar). See Table 3 for expected flow rates for these preset configurations. The pressures indicated here must be adjusted depending on the chamber geometry and system configuration.
Figure 2: The pressure applied to each reservoir controls the partition/spatial distribution of solutions inside the microfluidic chamber. (A) With equal pressure applied to the reservoirs, each solution occupies one-third of the chamber. (B) When changing a reservoir tube (here reservoir 3), the effective pressure drops down to zero, creating a backward flow. (C,D) Increasing the relative pressure on one of the reservoirs allows exposure of the glass surface to a single solution. The field of view in the middle of the chamber can be sequentially exposed to solutions 1 and 2 by alternating between configuration Mid Flow 1 (C) and Mid Flow 2 (D). Please click here to view a larger version of this figure.
7. Changing solution 'x'
NOTE: As shown in Figure 3A-C, it is important to keep in mind that solutions take minutes to flow from a reservoir tube to the main channel of the chamber. This minimal 'dead' time is imposed by the liquid volume contained in the tubing and the flow profile within the tubing (Figure 3A-C).
Figure 3: Delayed arrival of solutions from the reservoirs to the PDMS chamber and rapid change of biochemical conditions. (A-C) Delayed arrival of solutions from the reservoirs to the PDMS chamber. (A) Depending on the chamber geometry, the tube length, and the applied pressure at the inlet(s), the replacement of one solution by another is not instantaneous. After changing the reservoir tube to one containing a fluorescent solution (0 min), the solution progressively fills in the tubing (0.4 min) and the PDMS chamber (1-2 min). Indicative timing is given for a 150 mbar applied pressure, 80 cm PEEK tubing, and a 1600 µm wide, 20 µm high PDMS chamber. (B) The parabolic flow profile inside the PEEK tubing generates an effective gradient of fluorescence along the tubing radial profile and inside the chamber (see also Figure 1B). (C) Delayed arrival of solutions can be quantified by measuring the background epifluorescence signal in the chamber as a function of time. Experimental conditions: 0.5 µM 10% Alexa-568-labeled G-actin is injected with 150 mbar through a flow meter and 80 cm PEEK tubing. (D,E) Rapid change of biochemical conditions. (D) Pattern of incoming solutions in two Mid Flow conditions. (E) Increase in background fluorescence as a readout of actin concentration. Time t = 0 is set as the onset of fluorescence increase. Solution 1: 0.5 µM 10% Alexa-488-labeled G-actin, solution 2: F-buffer. (C,E) PDMS chamber: 20 µm high and 1600 µm wide. The epifluorescence intensity, ~2 µm above the surface, was quantified by averaging the signal over the full field of view, normalized to 0 in the absence of fluorophore and 1 at maximum intensity. Please click here to view a larger version of this figure.
8. Basic single filament experiment: Adenosine diphosphate (ADP)-actin barbed end depolymerization
NOTE: This section assumes a non-functionalized chamber (section 5 only). If the chamber has been directly functionalized (section 4), start at step 8.4.
9. Other single-filament experiments
10. Fascin-induced filament bundle formation and disassembly by ADF/cofilin
NOTE: To form actin filament bundles, make sure to have a sufficiently high filament seed density at the surface of the chamber. When exposed to fascin protein, neighboring filaments that fluctuate laterally will be dynamically cross-linked by fascin proteins. As fascin quickly unbinds from the filament side19, fascin has to be constantly present in the main flowing solution in order to maintain filament bundling.
11. Microfluidic device cleaning procedure
NOTE: To avoid any contamination from one experiment to another, it is critical to extensively clean and completely dry all the tubings and flow meters after each experiment.
12. Image analysis
NOTE: While this manuscript focuses on the method to assemble, manipulate, and visualize single actin filaments in microfluidics, a brief method to analyze acquired movies is provided here. The analysis is performed on 16-bit images, using ImageJ, following section 8.
For all the experiments described above, fluorescently labeled actin filaments should be clearly visible, with good contrast, indicative of low background fluorescence from the surface (Figure 4, see Supplementary File 1 for troubleshooting of common issues). Actin filaments should also not stick to the surface: when the dominant flow rate is low, the actin filaments' lateral fluctuations should be perceptible when observing them live and allow one to clearly determine that they are anchored by one of their ends only. Similarly, when using TIRF imaging, their vertical fluctuations should be visible by changes in intensity along their length and time. Depending on the applied flow rates, one may need to adjust the TIRF penetration depth to optimize the image quality of the actin filaments acquired by TIRF.
When exposing filaments to polymerization conditions (see section 8), filament elongation should be regular (i.e., the elongation at the end of the filament is not impeded by surface interaction or permanent sticking). In addition, the measured filament barbed end elongation rate should match the expected value according to the actin concentration in the tube1,20, indicating that the tube solution has been correctly flowed up to the microfluidic chamber (Figure 4A). Similarly, when exposed to a buffer solution, filaments should depolymerize steadily at a rate that reflects their ADP-content4 (Figure 4A). When exposing already grown actin filaments to a solution of fluorescently labeled cofilin, cofilin clusters will be nucleated and grow toward both the pointed and barbed ends (Figure 4B) at a rate that is dependent on the cofilin concentration. When assessing a potential cross-linking activity of an ABP, such as fascin (Figure 4C), close-by actin filaments forming bundles will be easily detected by their higher fluorescence intensity and a change in their lateral fluctuations.
The flow of liquid applies a viscous friction force on actin filaments that are anchored to the surface of the microfluidic chamber. The friction force coefficient on F-actin is η = 6.10-4 pN·s/µm2, expressed per filament micron length14. At intermediate flow rates, as filament height fluctuates around a constant average of 250 nm above the surface, there exists a force gradient from the free-floating end up to the filament anchoring point. One can therefore compute the applied tension at any point along the filament, using F = 6ηπLv, where v is the local flow velocity 250 nm above the surface (Figure 1B) and L is the downstream filament segment length (i.e., from the considered point up to the free end). For higher flow rates, the filament average height is not constant but increases linearly from the anchoring point to the free end, remains below 250 nm on average, and will vary depending on the flow rates, thus leading to a more complex tension force profile along the filament21.
Figure 4: Representative results. Typical experiments in which actin filaments are polymerized from spectrin-actin seeds and exposed to different ABPs. For the sake of clarity, only a fraction of the field of view is shown. (A) Outcome from the basic polymerization-depolymerization experiment (section 8). Filaments are polymerized with a solution of 0.8 µM 10% Alexa-488 labeled G-actin, aged for 15 min to convert all subunits into ADP-actin (not shown), and depolymerized when exposed to F-buffer only. Bottom: kymographs used to quantify the polymerization and depolymerization rates. Acquired at 1 frame/5 s, 200 ms exposure time, 150 mW 488 nm laser at 9% power, TIRF (laser penetration depth 250 nm). (B) Fragmentation of single actin filaments by 500 nM mCherry-cofilin-1. Actin is labeled with ATP-ATTO48822 (yellow) and cofilin-1 is fused to mCherry (blue). Top: fraction of a field of view. Note: protein aggregates on the surface. Bottom: kymograph showing the binding of cofilin-1 to a filament (arrowheads show cofilin-1 domain nucleation events), leading to a fragmentation event (lightning symbol). Acquired at 1 frame/4 s, 200 ms exposure, 150 mW 488 nm laser at 16% and 100 mW 561 nm laser at 12% power, epifluorescence. (C) Bundling of actin filaments by fascin (section 10.5). Filaments were first polymerized with 0.8 µM 5% Alexa-488 labeled G-actin and bundled with 200 nM fascin. Compared with single filaments, filament bundles appear two- to threefold brighter and not perfectly aligned with the flow. Acquired at 1 frame/10 s, 200 ms exposure, 20% 200 W Mercury lamp intensity, epifluorescence. (A–C) The background was subtracted with ImageJ's ad hoc function. Please click here to view a larger version of this figure.
Protein name | species | Uniprot ref (sequence) | original purification protocol ref. | comments |
actin | rabbit | P68135 (full length) | 23 | For fluorescent labeling, see ref 24 |
profilin1 | human | P07737 (full length) | 25 | see also ref 11 |
Spectrin-actin seed | human | N/A | 26, 27 | see also ref 11 |
cofilin1 | mouse | P18760 (full length) | 28 | |
gelsolin | human | P06396 (full length) | 29 | |
mDia1 formin | mouse | O08808 (aa 552–1255) | 13 | more detailed protocol in ref 24 |
fascin1 | human | Q16658 (full length) | 30 |
Table 1: Actin and actin-binding proteins23,24,25,26,27,28,29,30
Reagent | concentration |
Tris-HCl pH 7.4 | 5 mM |
KCl | 50 mM |
MgCl2 | 1 mM |
EGTA | 0.2 mM |
ATP | 0.2 mM |
DTT | 10 mM |
DABCO | 1 mM |
Table 2: F-buffer composition. DABCO and a relatively high concentration of DTT are used to limit photo-induced damage to filaments due to light exposure during fluorescence microscopy experiments.
Setting names | Pressure (mBar) | Flow rate (nL/min) |
Max pressure | 300 | ~ 30 000 (in dominant channel) |
High pressure | 150 | ~ 15 000 (in dominant channel) |
Mid pressure | 12 | ~ 1500 (in dominant channel) |
‘Change’ pressure | 12 for all inlets, 5 for outlet |
~ 500 (in each inlet) |
Table 3: Correspondence between applied pressures and measured flow rates. The resulting flow rates highly depend on the experimental setup. Values are given for a microfluidic chamber with a 1 cm long main channel of cross-section 20 µm x 800 µm (height x width), connected to each reservoir with 80 cm long PEEK tubing.
Supplementary File 1: Classical issues, causes, and solutions. They commonly encountered issues when working with microfluidics and/or single actin filaments. Please click here to view a larger version of this figure.
Compared to standard single-filament methods where actin filaments are anchored to the surface by multiple points along their length or maintained close to it by a crowding agent such as methylcellulose, microfluidics offers a number of advantages. As interactions with the surface are minimal, the artificial pauses these interactions can induce during both elongation and depolymerization are avoided. The filaments are aligned by the flow, parallel to each other, easing their monitoring and the measurement of their lengths. The solution around the filaments is constantly renewed, exposing them to constant protein concentrations. Being able to rapidly (<1 s, Figure 3D,E) switch between different protein solutions to which the filaments are exposed allows one to perform time-controlled sequential experiments, which are often instrumental for kinetic studies. Finally, the viscous drag exerted by the flowing solution on the filaments can be exploited to apply controlled mechanical stress on the filaments (Representative Results section). One should note that moderate fluid flows (Mid Flow pressure settings) bring filaments close enough to the surface (~250 nm) to efficiently image them with TIRF while generating minimal tension (<1 pN)14.
Compared to classical single-filament assays, however, microfluidics will require larger volumes of protein solutions: typically a few 100 µL, when a standard experiment could be done with less than 10 µL. This can be a limitation when using precious proteins. Classical experiments can be used to help establish the relevant experimental conditions (e.g., absolute or relative concentrations of different proteins) before starting a series of microfluidic experiments. Another limitation, as for any other single-filament techniques in vitro, comes from the imperfect passivation of the coverslip surface. Reproducibility in coverslip cleaning and binding of the passivation layer (BSA, PEGylation, etc.) is always hard to control. A one-step passivation technique based on PEG-silane surface treatment has become the technique of choice in many laboratories7,15. As such, the effective density of filament seeds may vary between experiments by roughly twofold, even when repeated as accurately as possible. One should aim for a satisfactory range of filament surface density and be prepared to repeat the experiment if needed. Commonly encountered issues when working with microfluidics and/or single actin filaments are discussed in Supplementary File 1.
For the basic protocol presented here, one should note that spectrin-actin seeds, which can be viewed as short stabilized filaments, are randomly oriented as they stick to the surface. As a consequence, as the filaments grown from these seeds align with the flow, their portion closest to the seed will be sharply bent, each with its own angle. The length over which filaments are bent is usually very small when filaments are exposed to mid or high flows. In fact, this length will generally be smaller than the diffraction limit (~200 nm) and will thus not be easily detected. Importantly, the ABPs that are sensitive to filament curvature will bind and function differently in this highly bent region. To avoid biasing results, the simplest is to exclude this region from the analysis21.
Before we started using microfluidics to manipulate and visualize single actin filaments, it had already been used to study single DNA filaments3, which are far more flexible. This can give rise to notable differences, as the flow can dramatically unwind DNA and change its apparent length dramatically. Microfluidics can also be used, very similarly to the method presented here, to study microtubules; these are much stiffer but can nonetheless be made to align with the flow in order to measure their elongation and depolymerization, taking advantage of the rapid switch of conditions31,32, or be bent by a perpendicular flow to measure microtubule plasticity7.
We have presented here the protocol for the basic experiment, where filaments are anchored by one end only and where the flow direction in the field of view is the same throughout the experiment. These two characteristics can be varied. For example, filaments can be anchored by multiple points in order to generate a different force profile along the filament. Likewise, the flow direction can be varied (in the vicinity of the junction between the entry channels, in the flow chamber) to locally bend filaments, as the unanchored part will be pointing in a different direction than the anchored filament segment21. Filaments elongating from randomly anchored spectrin-actin seeds can also be exposed to cross-linking proteins to form bundles33 (see section 10). By combining microfluidics with other techniques (micropatterning, optical tweezers, etc.) or designing microfluidic chambers with compartments to modify flow lines, multiple configurations can be created to study specific ABP activity onto single filaments or to form small actin networks34. The number of combinations, together with the advantage and versatility of microfluidics, offers many tools to researchers in order to decipher the spatio-temporal regulation of actin networks at the molecular scale.
The authors have nothing to disclose.
We are grateful to the B. Ladoux and R.-M. Mège lab for the use of their UV-cleaner equipment, and J. Heuvingh and 0. du Roure for the initial training we received on preparing molds on silicon wafers and providing tips on microfluidics. We acknowledge funding from European Research Council Grant StG-679116 (to A.J.) and Agence Nationale de la Recherche Grants Muscactin and Conformin (to G.R.-L.).
β-Casein | Merck | C6905 | Used at 8 mg/mL |
Biopsy punch (with plunger) | Ted Pella | 15115-2 | ID 0.75 mm, OD 1.07 mm |
Biotin-BSA | Merck | A8549 | Used at 1 mg/mL |
BSA | Merck | A8022 | Used at 50 mg/mL |
Coverslip Mini-Rack Teflon holder |
Invitrogen | C14784 | for 8 coverslips |
Coverslips 22x40mm Thickness #1.5 |
Menzel Gläser | 631-1370 | |
DABCO | Merck | D27802 | component in f-buffer |
DTT | Euromedex | EU0006-D | component in f-buffer |
Ester NHS Alexa Fluor 488 | Invitrogen | A20000 | Fluorophore for actin labeling on Lys328. |
EZ-Link Sulfo-NHS-Biotin | Thermo Scientific | 21338 | To biotinylate actin on Lys328 |
Hellmanex III | Hellma | 9-307-011-4-507 | Glass cleaning detergent |
ImageJ | NIH | N/A | open source software |
Laboport | KNF | 811kn.18 | vacuum pump (ultimate vacuum: 240 mbar) |
Magic invisible tape | Scotch | 7100024666 | standard transparent office tape |
Micrewtube | Simport | T341-6T | 2 mL microfluidic reservoir tubes |
Microfluidic device Part 1: Flow Unit S | Fluigent | FLU-S-D-PCKB | Flowmeter |
Microfluidic device Part 2: Fluiwell-4C-2 mL | Fluigent | 14002001PCK | Reservoir holder |
Microfluidic device Part 3: MFCS-EZ | Fluigent | EZ-11000001 EZ-00345001 |
Pressure controller |
Model 42 – UVO-Cleaner | Jelight Inc. | 42-220 | Ultraviolet cleaner |
N6-(6-Aminohexyl)-ATP-ATTO-488 | Jena Bioscience | NU-805-488 | ATP-ATTO used to label actin |
neutravidin | Thermo Scientific | 31000 | |
PLL-PEG | SuSoS | PLL(20)-g[3.5]- PEG(2) | Use at 1 mg/mL in PBS. |
Polydimethylsiloxane (PDMS) Sylgard 184 Silicon Elastomer | Dow Corning | 1673921 | Contains PDMS base and curing agent |
Polyetheretherketone (PEEK) tubing | Merck | Z226661 | “Blue” : I.D. = 0.25 mm |
Safety blow gun | Coilhose Pneumatics | 700-S | filtered air |
Silicon tubing | VWR | 228-0701P | connect PEEK to coupler |
Stainless steel catheter coupler | Prime Bioscience | SC22/15 | Inserted into PDMS inlets and outlet to connect to PEEK tubing |
Thermoplastic film | Sigma Aldrich | PM996 | Standard "parafilm" |
Ultrapure ethanol | VWR | 64-17-5 | |
Ultrasonic cleaning bath | VWR | USC200TH | To accomodate 1 L beakers |
Vacuum dessicator | SP Bel-Art | F42022-0000 | to degas the PDMS or solutions |