Cross-correlated TIRF/AFM reveals asymmetric distribution of force-generating heads along self-assembled, "synthetic" myosin filaments

Abstract

Myosin-II's rod-like tail drives filament assembly with a head arrangement that is often considered to be a symmetric bipole that generates equal and opposite contractile forces on actin. Self-assembled myosin filaments are shown here to be asymmetric in physiological buffer based on cross-correlated images from both atomic force microscopy and total internal reflection fluorescence. Quantitative cross-correlation of these orthogonal methods produces structural information unavailable to either method alone in showing that fluorescence intensity along the filament length is proportional to height. This implies that myosin heads form a shell around the filament axis, consistent with F-actin binding. A motor density of approximately 50-100 heads/micrometer is further estimated but with an average of 32% more motors on one half of any given filament compared to the other, regardless of length. A purely entropic pyramidal lattice model is developed and mapped onto the Dyck paths problem that qualitatively captures this lack of length dependence and the distribution of filament asymmetries. Such strongly asymmetric bipoles are likely to produce an unbalanced contractile force in cells and in actin-myosin gels and thereby contribute to motility as well as cytoskeletal tension.

Figure 1

TIRF and AFM images of self-assembled myosin filaments on PMMA-coated glass in buffer with physiological salt. Myosin molecules were labeled with an average of 1.2 dyes per heavy chain and predominantly at Cys-707 in the motor domain. (a) The TIRF image has limited resolution both because of diffraction effects and because of the finite pixel size of the camera. The inset shows the crystal structure of the myosin motor domain with part of the converter domain. The red star indicates Cys-707. (b) Tapping mode AFM height image of the same region shows the elongated and tapered structure of individual myosin filaments that are now clearly resolved. The white overlay represents the edge of the thresholded TIRF image to demonstrate registration. (c) Cropped and scaled TIRF image from the region indicated with the black box in b. (d) Rescan of the same region by AFM reveals still higher resolution and finer structural details of the filaments.

Figure 2

Fluorescence intensity (and therefore number of myosin heads) correlates with height. (a) TIRF image. Inset shows the optical microscope's point spread function (PSF) determined by fitting a Gaussian (standard deviation 260 nm) to a single spot in a TIRF image. (b) Tapping-mode AFM height image of the same region. The “shadows” vertically displaced above the filaments are most likely a tip artifact, but because they are several filament widths away from the principal image, this artifact is unlikely to affect the tip tracking and height profiles along the filament lengths. (c) Maximum height (black) and intensity (red) profiles measured along the three filaments in the images. The blue curve is the convolution of the AFM trace with the PSF to simulate broadening of fluorescence by diffraction and to facilitate a more appropriate comparison between height and intensity. The number in the top right of each plot is the fractional asymmetry of each filament calculated assuming the filaments have a circular cross section. It is an estimate of how much more volume the filament has on one half than on the other divided by the total volume.

Figure 3

Height-intensity scaling and its structural implications. (a) Plot of intensity versus height calculated using the broadened AFM height data. The black curve is a power law fit with a resulting exponent of 1.0 ± 0.1, suggesting that the best scaling is linear; indeed, the Akaike information criterion shows that a linear fit is three times more likely to account for the data than the power law fit and 3 × 10⁶ times more likely than a quadratic fit. Linear scaling implies that the myosin heads are arranged in a shell around the filament, if this were not the case, the number of dyes present in a given diffraction-limited cross section would scale as a higher power of the diameter (and therefore height) as illustrated schematically in (b).

Figure 4

Fractional asymmetry from raw AFM height profiles as a function of length. Black points are averages over 500-nm bins of the individual data points shown in gray, except for the longest three points, which correspond to the three filaments imaged in Fig. 2. There is large scatter about the mean of 0.32 shown by the horizontal line, but there is no clear dependence on filament length. The right panel is a histogram showing the probability distribution of the filament asymmetries, δ. Small asymmetries are the most likely (the maximum of the distribution occurs at the smallest bin), but the decay with δ is slow (approximately linear).

Figure 5

Pyramidal lattice model of myosin filaments. Myosin filaments are modeled using a lattice with staggered sites without allowing overhanging edges. The problem of counting the number of ways of stacking bricks on this lattice is equivalent to counting the number of “Dyck paths” (a). Two model filaments are shown in (b). To see the equivalence of these two problems, note that the path drawn behind the first filament is the same as that shown in a but rotated by 3π/4. The base lengths L and fractional asymmetries δ are written in the insets. The height profiles of these filaments are equivalent to so-called Dyck paths as shown by the rotated plot under the first schematic filament. (c) Plot of δ as a function base length. Points show all possible model filaments of a given length. The line shows the average δ. (d) Histogram of the points plotted in (b) showing dependence of P(δ) on δ.

Figure 6

Asymmetric myosin filaments could be motile as shown in this schematic of an asymmetric myosin filament at an F-actin junction. If the myosin filament (top) is bipolar about its center, and the actin filaments (bottom, minus ends toward center) are relatively immobile, then the contractile forces exerted by the myosin filament on the actin filaments will not balance. If there is sufficient asymmetry, the myosin filament will move toward the side with more heads (in this case to the right).

Figure 7

(a) Geometry for scaling correction calculation. The region with fluorophores for the shell model is the arc of the circle between the points labeled 1 and 2. The region with fluorophores for the solid model is the segment indicated by the thin lines. (b) The results of the numerical integration for the shell model (black points) and solid model (gray points) taking into account the decay of the TIRF field. The solid lines are fits to the data—linear in the case of the shell model and quadratic (no constant or linear term) in the case of the solid model.

Figure 8

(a) Average intensity over time for a 5 by 5 pixel square showing a one-step bleaching event. (b) A Gaussian fit to a spot that subsequently bleached in a single step to determine that spot's total intensity.