PULSIS. (A) Basic schematic of the pressure-driven flow system and imaging setup, not to scale. (B) Cartoon depiction of PULSIS. Objects are imaged with different duration pulses and by comparing the relative lengths, we can accurately measure the lengths of moving objects and distinguish point objects from elongated objects. (C) Example experimental PULSIS trajectories of fluorescently labeled linearized DNA at 50 dyn/cm2. (D) Example relationship between measured length of pulse Lm (μm) vs relative pulse duration Tp (arbitrary time units) for the 2 DNA PULSIS trajectories in panel C. For each trajectory, streak lengths are measured and a linear regression performed of the form Lm = L0 + V × Tp, with fitting errors according to York et al.37,Lm is the illuminated streak length that we measure, and Tp is the relative pulse duration defined by the pulse pattern (1, 2, or 3). The linear fit gives us the particle velocity V, and the y-intercept L0 represents the length of the molecule observed with an infinitesimally short pulse, that is, with no motion blur. The first trajectory (yellow) has a corrected length of 0.29 μm and resembles a compact object. The second trajectory has a corrected length of 1.03 μm and represents an elongated object. Error bars on pulse length are based on goodness of fit to predicted pulse shape (supplemental Methods). (E) Positive control showing histogram of PULSIS-determined lengths of double stranded M13 DNA plasmid both in supercoiled (blue) and linearized (red) state at 50 dyn/cm2, imaged in sucrose buffer. Histograms are of motion blur–corrected lengths of hundreds of single molecules. The examples (trajectories and analysis) from Figure 2C and D are 2 statistics from the linearized (red) distribution. Histograms are displayed along with kernel density estimates. Kernel density estimation is a method for smoothing histograms by applying a Gaussian kernel to each point.38 A Gaussian kernel was used with bandwidth set by the Silverman rule.39 (F) Negative control showing kernel density estimate for PULSIS motion blur–corrected beads at different shear stress (manufacture determined diameter of 0.11 μm). Raw histograms are shown in supplemental Figure 1. Number of measurements and mean length and standard deviation for each condition are shown in panels E and F.
