Software for drift compensation, particle tracking and particle analysis of high-speed atomic force microscopy image series

Atomic force microscopy (AFM) image acquisition is performed by raster-scanning a faint tip with respect to the sample by the use of a piezoelectric stage that is guided by a feedback system. This process implies that the resulting images feature particularities that distinguish them from images acquired by other techniques, such as the drift of the piezoelectric elements, the unequal image contrast along the fast- and the slow-scan axes, the physical contact between the tip of nondefinable geometry and the sample, and the feedback parameters. Recently, high-speed AFM (HS-AFM) has been introduced, which allows image acquisition about three orders of magnitude faster (500-100 ms frame rate) than conventional AFM (500 s to 100 s frame rate). HS-AFM produces image sequences, large data sets, which report biological sample dynamics. To analyze these movies, we have developed a software package that (i) adjusts individual scan lines and images to a common contrast and z-scale, (ii) filters specifically those scan lines where increased or insufficient force was applied, (iii) corrects for piezo-scanner drift, (iv) defines particle localization and angular orientation, and (v) performs particle tracking to analyze the lateral and rotation displacement of single molecules.     

Direct visualization of dynamic protein-DNA interactions with a dedicated atomic force microscope.

Photolyase DNA interactions and the annealing of restriction fragment ends are directly visualized with the atomic force microscope (AFM). To be able to interact with proteins, DNA must be loosely bound to the surface. When MgCl2 is used to immobilize DNA to mica, DNA is attached to the surface at distinct sites. The pieces of DNA in between are free to move over the surface and are available for protein interaction. After implementation of a number of instrumental improvements, the molecules can be visualized routinely, under physiological conditions and with molecular resolution. Images are acquired reproducibly without visible damage for at least 30 min, at a scan rate of 2 x 2 microm2/min and a root mean square noise of less than 0.2 nm. Nonspecific photolyase DNA complexes were visualized, showing association, dissociation, and movement of photolyase over the DNA. The latter result suggests a sliding mechanism by which photolyase can scan DNA for damaged sites. The experiments illustrate the potential that AFM presents for modern molecular biology.     

Confined lateral diffusion of membrane receptors as studied by single particle tracking (nanovid microscopy). Effects of calcium-induced differentiation in cultured epithelial cells.

The movements of E-cadherin, epidermal growth factor receptor, and transferrin receptor in the plasma membrane of a cultured mouse keratinocyte cell line were studied using both single particle tracking (SPT; nanovid microscopy) and fluorescence photobleaching recovery (FPR). In the SPT technique, the receptor molecules are labeled with 40 nm-phi colloidal gold particles, and their movements are followed by video-enhanced differential interference contrast microscopy at a temporal resolution of 33 ms and at a nanometer-level spatial precision. The trajectories of the receptor molecules obtained by SPT were analyzed by developing a method that is based on the plot of the mean-square displacement against time. Four characteristic types of motion were observed: (a) stationary mode, in which the microscopic diffusion coefficient is less than 4.6 x 10(-12) cm2/s; (b) simple Brownian diffusion mode; (c) directed diffusion mode, in which unidirectional movements are superimposed on random motion; and (d) confined diffusion mode, in which particles undergoing Brownian diffusion (microscopic diffusion coefficient between 4.6 x 10(-12) and 1 x 10(-9) cm2/s) are confined within a limited area, probably by the membrane-associated cytoskeleton network. Comparison of these data obtained by SPT with those obtained by FPR suggests that the plasma membrane is compartmentalized into many small domains 300-600 nm in diameter (0.04-0.24 microns2 in area), in which receptor molecules are confined in the time scale of 3-30 s, and that the long-range diffusion observed by FPR can occur by successive movements of the receptors to adjacent compartments. Calcium-induced differentiation decreases the sum of the percentages of molecules in the directed diffusion and the stationary modes outside of the cell-cell contact regions on the cell surface (which is proposed to be the percentage of E-cadherin bound to the cytoskeleton/membrane-skeleton), from approximately 60% to 8% (low- and high-calcium mediums, respectively).     

DNA breakage detection-fish (DBD-FISH): effect of unwinding time

DBD-FISH is a new procedure that allows detection and quantification of DNA breakage in situ within specific DNA target sites. Cells embedded in an agarose matrix on a slide are treated in an alkaline unwinding solution to transform DNA breaks into single-stranded DNA (ssDNA). After removal of proteins, DNA probes are hybridized and detected. DNA breaks increase the ssDNA and relax supercoiling of DNA loops, so more probe hybridizes, thereby increasing the surface area and fluorescence intensity of the FISH signal. The probe selects the chromatin area to be analysed.In order to restrict the extension of unwound ssDNA to a region closer to the origin of the DNA break, human leukocytes were processed for DBD-FISH with a whole genome probe, after a 10 Gy dose of X-rays, for various unwinding times: 5, 2 min and 30s. Two cell populations were detected after 30s, but not with the 5 or 2 min unwinding times. One cell group had small to medium haloes corresponding to the relaxation of DNA supercoiling after DAPI staining, and strong DBD-FISH labelling of induced DNA breaks, whereas the other cell group showed big haloes of DNA loop unfolding and an absence of DBD-FISH labelling. The latter group was similar to cells processed by DBD-FISH without the unwinding step. Thus, they should correspond to cells unaffected by the alkaline unwinding solution, possibly because very brief unwinding times do not allow the diffusion of the alkali into the cells deep within the gel, thus biasing the results. Taking this into account, 2 min seems to be the minimum unwinding time required for an accurate detection of a signal by DBD-FISH.     

Compartmentalized structure of the plasma membrane for receptor movements as revealed by a nanometer-level motion analysis

Movements of transferrin and alpha 2-macroglobulin receptor molecules in the plasma membrane of cultured normal rat kidney (NRK) fibroblastic cells were investigated by video-enhanced contrast optical microscopy with 1.8 nm spatial precision and 33 ms temporal resolution by labeling the receptors with the ligand-coated nanometer-sized colloidal gold particles. For both receptor species, most of the movement trajectories are of the confined diffusion type, within domains of approximately 0.25 microns2 (500-700 nm in diagonal length). Movement within the domains is random with a diffusion coefficient approximately 10(-9) cm2/s, which is consistent with that expected for free Brownian diffusion of proteins in the plasma membrane. The receptor molecules move from one domain to one of the adjacent domains at an average frequency of 0.034 s-1 (the residence time within a domain approximately 29 s), indicating that the plasma membrane is compartmentalized for diffusion of membrane receptors and that long- range diffusion is the result of successive intercompartmental jumps. The macroscopic diffusion coefficients for these two receptor molecules calculated on the basis of the compartment size and the intercompartmental jump rate are approximately 2.4 x 10(-11) cm2/s, which is consistent with those determined by averaging the long-term movements of many particles. Partial destruction of the cytoskeleton decreased the confined diffusion mode, increased the simple diffusion mode, and induced the directed diffusion (transport) mode. These results suggest that the boundaries between compartments are made of dynamically fluctuating membrane skeletons (membrane-skeleton fence model).     

Lowering the barriers to random walks on the cell surface

We used fluorescence recovery after photobleaching (FRAP) and single particle tracking (SPT) techniques to compare diffusion of class I major histocompatibility complex molecules (MHC) on normal and alpha-spectrin-deficient murine erythroleukemia (MEL) cells. Because the cytoskeleton mesh acts as a barrier to lateral mobility of membrane proteins, we expected that diffusion of membrane proteins in alpha-spectrin-deficient MEL cells would differ greatly from that in normal MEL cells. In the event, diffusion coefficients derived from either FRAP or SPT analysis were similar for alpha-spectrin-deficient and normal MEL cells, differing by a factor of approximately 2, on three different timescales: tens of seconds, 1-10 s, and 100 ms. SPT analysis showed that the diffusion of most class I MHC molecules was confined on both cell types. On the normal MEL cells, the mean diagonal length of the confined area was 330 nm with a mean residency time of 40s. On the alpha-spectrin-deficient MEL cells, the mean diagonal length was 650 nm with a mean residency time of 45s. Thus there are fewer barriers to lateral diffusion on cytoskeleton mutant MEL cells than on normal MEL cells, but this difference does not strongly affect lateral diffusion on the scales measured here.     

Mobility of integrin alpha5beta1 measured on the isolated ventral membranes of human skin fibroblasts

We have measured the lateral mobility of individual alpha5 integrin molecules in ventral plasma membranes of fibroblasts, which were prepared by removal of apical surfaces and nuclei followed by elimination of actin filaments with gelsolin, an actin-severing protein. The cytoplasmic domain of individual integrin molecules was tagged with 100 nm fluorescent polystyrene bead, and motion of the bead was observed and video-recorded. Position of the bead in each frame was determined from the centroid of the fluorescence image, from which plots of the mean-square displacement against time intervals were derived. Within short intervals of time (<100 ms) the mean-square displacement was proportional to the time interval, and the averaged translational diffusion coefficient of (5.3+/-4.4) x 10(-10) cm2/s was obtained with a broad distribution of (1.3-20) x 10(-10) cm2/s. The broad distribution might reflect the oligomerized state of integrin. The largest diffusion coefficient was comparable to that of lipid molecules previously measured in cells and probably represented the diffusion of a single integrin molecule in the presence of little interference of actin cytoskeleton or extracellular matrix. In longer time intervals (>100 ms) the motion of the bead was confined in an area, the average diameter of which was 410+/-160 nm. This was similar to the values described in previous reports, in which the motion of other membrane receptors labeled on their extracellular domain was measured in living cells.     

Size-dependent DNA mobility in cytoplasm and nucleus

The diffusion of DNA in cytoplasm is thought to be an important determinant of the efficacy of gene delivery and antisense therapy. We have measured the translational diffusion of fluorescein-labeled double-stranded DNA fragments (in base pairs (bp): 21, 100, 250, 500, 1000, 2000, 3000, 6000) after microinjection into cytoplasm and nucleus of HeLa cells. Diffusion was measured by spot photobleaching using a focused argon laser spot (488 nm). In aqueous solutions, diffusion coefficients of the DNA fragments in water (D(w)) decreased from 53 x 10(-8) to 0.81 x 10(-8) cm(2)/s for sizes of 21-6000 bp; D(w) was related empirically to DNA size: D(w) = 4.9 x 10(-6) cm(2)/s.[bp size](-0.72). DNA diffusion coefficients in cytoplasm (D(cyto)) were lower than D(w) and depended strongly on DNA size. D(cyto)/D(w) decreased from 0.19 for a 100-bp DNA fragment to 0.06 for a 250-bp DNA fragment and was <0.01 for >2000 bp. Diffusion of microinjected fluorescein isothiocyanate (FITC) dextrans was faster than that of comparably sized DNA fragments of 250 bp and greater. In nucleus, all DNA fragments were nearly immobile, whereas FITC dextrans of molecular size up to 580 kDa were fully mobile. These results suggest that the highly restricted diffusion of DNA fragments in nucleoplasm results from extensive binding to immobile obstacles and that the decreased lateral mobility of DNAs >250 bp in cytoplasm is because of molecular crowding. The diffusion of DNA in cytoplasm may thus be an important rate-limiting barrier in gene delivery utilizing non-viral vectors.     

Numerical simulation of gel electrophoresis of DNA knots in weak and strong electric fields

Gel electrophoresis allows one to separate knotted DNA (nicked circular) of equal length according to the knot type. At low electric fields, complex knots, being more compact, drift faster than simpler knots. Recent experiments have shown that the drift velocity dependence on the knot type is inverted when changing from low to high electric fields. We present a computer simulation on a lattice of a closed, knotted, charged DNA chain drifting in an external electric field in a topologically restricted medium. Using a Monte Carlo algorithm, the dependence of the electrophoretic migration of the DNA molecules on the knot type and on the electric field intensity is investigated. The results are in qualitative and quantitative agreement with electrophoretic experiments done under conditions of low and high electric fields.     

The rotational diffusion of chloroplast phosphate translocator and of lipid molecules in bilayer membranes

The rotational mobility of the phosphate translocator from the chloroplast envelope and of lipid molecules in the membrane of unilamellar azolectin liposomes has been investigated. The rotational dynamics of the liposome membrane were investigated by measuring the rotational diffusion of eosin-5-isothiocyanate(EITC)-labeled L-alpha-dipalmitoylglycerophosphoethanolamine (Pam2 GroPEtn) in the lipid phase of the vesicles, either in the presence or absence of the reconstituted phosphate translocator. The temperature dependence of the anisotropy decay showed that above 25 degrees C the main contribution to the anisotropy decay was caused by uniaxial anisotropic rotation of the labelled lipid molecules around the axis normal to the membrane plane. The rate of rotation of the labelled lipid molecules was strongly dependent on the viscosity of the medium (eta 1). Extrapolation to eta 1 = 0 Pa.s yielded a correlation time of phi = 20 +/- 5 ns, t = 30 degrees C, for lipid rotation with respect to the membrane normal. The rotational diffusion coefficient of the lipid molecules was calculated to be Dr = 2.0 x 10(9) rad2.s-1 and the apparent microviscosity in the vesicle membrane, as derived from the rotational correlation time, was eta 2 approximately 12 mPa.s. The rotational correlation time of the phosphate translocator in the membrane was only slightly dependent on the viscosity of the medium. The temperature dependence of the protein rotation also indicated that the rotation of the protein in the membrane was largely restricted and occurred mainly about the axis normal to the membrane plane. Measurements at a medium viscosity of eta 1 = 1 mPa.s yielded a value of phi r approximately 450 ns corresponding to Dr = 8.8 x 10(7) rad2.s-1 for protein rotation with respect to the membrane normal. From this value and the data of the lipid rotation, the cross-sectional area of the protein part embedded in the membrane was calculated to be approximately 9 nm2. This cross-sectional area is large enough to include at most 14 membrane-spanning helices. Our results also indicated that at lipid/protein molar ratios greater than or equal to 1.5 x 10(4): 1 aggregation occurred in the model membranes below 30 degrees C. However, above 30 degrees C and at a high dilution of the protein in the membrane it appeared that the membrane viscosity monitored by lipid and protein rotational diffusion were identical.     

Extracting Transition Rates in Particle Tracking Using Analytical Diffusion Distribution Analysis

Single-particle tracking is an important technique in the life sciences to understand the kinetics of biomolecules. The analysis of apparent diffusion coefficients in vivo, for example, enables researchers to determine whether biomolecules are moving alone, as part of a larger complex, or are bound to large cellular components such as the membrane or chromosomal DNA. A remaining challenge has been to retrieve quantitative kinetic models, especially for molecules that rapidly switch between different diffusional states. Here, we present analytical diffusion distribution analysis (anaDDA), a framework that allows for extracting transition rates from distributions of apparent diffusion coefficients calculated from short trajectories that feature less than 10 localizations per track. Under the assumption that the system is Markovian and diffusion is purely Brownian, we show that theoretically predicted distributions accurately match simulated distributions and that anaDDA outperforms existing methods to retrieve kinetics, especially in the fast regime of 0.1–10 transitions per imaging frame. AnaDDA does account for the effects of confinement and tracking window boundaries. Furthermore, we added the option to perform global fitting of data acquired at different frame times to allow complex models with multiple states to be fitted confidently. Previously, we have started to develop anaDDA to investigate the target search of CRISPR-Cas complexes. In this work, we have optimized the algorithms and reanalyzed experimental data of DNA polymerase I diffusing in live Escherichia coli. We found that long-lived DNA interaction by DNA polymerase are more abundant upon DNA damage, suggesting roles in DNA repair. We further revealed and quantified fast DNA probing interactions that last shorter than 10 ms. AnaDDA pushes the boundaries of the timescale of interactions that can be probed with single-particle tracking and is a mathematically rigorous framework that can be further expanded to extract detailed information about the behavior of biomolecules in living cells.     

Spatial distribution and diffusive motion of RNA polymerase in live Escherichia coli

By labeling the β' subunit of RNA polymerase (RNAP), we used fluorescence microscopy to study the spatial distribution and diffusive motion of RNAP in live Escherichia coli cells for the first time. With a 40-ms time resolution, the spatial distribution exhibits two or three narrow peaks of 300- to 600-nm full width at half-maximum that maintain their positions within 60 nm over 1 s. The intensity in these features is 20 to 30% of the total. Fluorescence recovery after photobleaching (FRAP) measures the diffusive motion of RNAP on the 1-μm length scale. Averaged over many cells, 53%±19% of the RNAP molecules were mobile on the 3-s timescale, with a mean apparent diffusion constant of 0.22±0.16 μm2-s(-1). The remaining 47% were immobile even on the 30-s timescale. We interpret the immobile fraction as arising from RNAP specifically bound to DNA, either actively transcribing or not. The diffusive motion of the mobile fraction (fmobile) probably involves both one-dimensional sliding during nonspecific binding to DNA and three-dimensional hopping between DNA strands. There is significant cell-to-cell heterogeneity in both DRNAP and fmobile.     

Microscopic mechanism of DNA damage searching by hOGG1

The DNA backbone is often considered a track that allows long-range sliding of DNA repair enzymes in their search for rare damage sites in DNA. A proposed exemplar of DNA sliding is human 8-oxoguanine (^oG) DNA glycosylase 1 (hOGG1), which repairs mutagenic ^oG lesions in DNA. Here we use our high-resolution molecular clock method to show that macroscopic 1D DNA sliding of hOGG1 occurs by microscopic 2D and 3D steps that masquerade as sliding in resolution-limited single-molecule images. Strand sliding was limited to distances shorter than seven phosphate linkages because attaching a covalent chemical road block to a single DNA phosphate located between two closely spaced damage sites had little effect on transfers. The microscopic parameters describing the DNA search of hOGG1 were derived from numerical simulations constrained by the experimental data. These findings support a general mechanism where DNA glycosylases use highly dynamic multidimensional diffusion paths to scan DNA.     

Removal of pyrimidine dimers from Saccharomyces cerevisiae nuclear DNA under nongrowth conditions as detected by a sensitive, enzymatic assay.

A sensitive and quantitative procedure for the detection of pyrimidine dimers in yesast nuclear DNA is described. The assay employs dimer-specific, endonuclease activities from Micrococcus luteus together with DNA sedimentation through calibrated, alkaline sucrose gradients to detect endonuclease-induced, single-strand breaks. Breaks were induced in a dose-dependent manner from 0 to 80 J m-2 at 254 nm and in numbers equivalent to the numbers of dimers induced by similar doses (Unrau et al., Biochim. Biophys. Acta, 312 (1973) 626--632). This procedure also allows the use of [6-3H] uridine to label cellular nucleic acids, but dose not require extensive DNA purification to eliminate concomitantly labeled RNA. Endonuclease-sensitive sites in the wild-type, haploid strain S288C, after irradiation with 5 J m-2 (254 nm), were removed in less than 5 min when cells were incubated in buffer (pH 7.0) at 28 degrees C. After irradiation with doses from 30 to 100 Jm-2 site removal in S288C required longer postirradiation incubations and was about 90% complete. In a radiation-sensitive strain carrying the mutant allele rad4-3 the number of endonuclease-sensitive sites remained constant for 6 h after irradiation with 5 Jm-2. The retention of sites in this strain indicates that it is defective in the excision of pyrimidine dimers.     

Production and subsequent second-order decomposition of protein disulfide anions: Lengthy collisions between proteins

By a technique of radiation chemistry, one can convert small groups on a protein (surface disulfide bonds) into reactive centers (disulfide anion radicals) in less than 100 μs. The reactive centers on different molecules then proceed to react with each other and are destroyed. The anion radical is formed by electron transfer from transient CO₂^? generated by pulse radiolysis (Adams et al., 1972). The formation and decay of the disulfide anion radical were studied by monitoring its absorption at 410 nm. The transient disulfide anion absorption is a reliable qualitative but not quantitative test for the presence of surface disulfide groups.The area of this reactive center is very small compared to the total surface area of the protein. Nevertheless the rate of this reaction for 13 proteins ranges from 2% to 50% of the diffusion controlled rate of 3 × 10⁹m^?1 s^?1. Billiard ball type collisions during which the protein molecules briefly touch each other do not account for these fast rates. We conclude that the collision has substantial duration. A radially symmetric potential holds the molecules together but allows them to roll on each others' surfaces. The model of Sole & Stockmayer (1973) can be applied to this reaction, showing that the collision duration is typically 10 to 100 μs. This conclusion, valid for 12 of 13 proteins studied, seems to be very general. Whenever two proteins approach each other, attractive forces hold them together long enough for mutual rotation to bring the reactive groups into contact, a necessary but sometimes not sufficient condition for their reaction.     

Physico-chemical properties of deoxyribonucleoprotamine from sturgeon sperm heads]

Deoxyribonucleoprotamine (DNPn) from sonicated nuclei of sturgeon sperm heads was studied by means of ring dichroism. A derivative analysis of DNA and DNPn melting curves in 1 mM Tris. HCl pH 8.0 revealed the fraction of protein-free DNA being about 30% and suggested the preferable binding of protamine molecules with AT-rich DNA regions. The latter is also confirmed by the data on ring dichroism of protein-poor soluble DNAPn fraction in 0,14 M NaCl. Ring dichroism of DNA and DNPn in 1 mM Tris coinsides at the wavelength of 310-240 nm at concentrations of 500-50 mkg/ml. Dilution of DNPn to 5 mkg-ml resulted in the decrease of the ellipticity at 275 nm and produced no effect at 260-210 nm. The effect observed is suggested to be due to a partial transition of DNA in DNPn into C-form under the dilution as a result of a higher molecule hydration and a destruction of some hydrogen bonds between guanidine residues of arginine and oxygen of phosphate groups, stabilyzing DNA in the B-form. Ring dichroism spectrum of protamine, calculated by the subtraction of DNA spectrum from DNPn spectrum at the region of 240-210 nm coinsides with that of free protamine and indicates the absence of an ordered structure in protamine molecules in DNPn.     

Molecular crowding reduces to a similar extent the diffusion of small solutes and macromolecules: measurement by fluorescence correlation spectroscopy

Aqueous environments in living cells are crowded, with up to >50 wt% small and macromolecule-size solutes. We investigated quantitatively one important consequence of molecular crowding--reduced diffusion of biologically important solutes. Fluorescence correlation spectroscopy (FCS) was used to measure the diffusion of a series of fluorescent small solutes and macromolecules. In water, diffusion coefficients (D(o)w) were (in cm2/s x 10(-8)): rhodamine green (270), albumin (52), dextrans (75, 10 kDa; 10, 500 kDa), double-stranded DNAs (96, 20 bp; 10, 1 kb; 3.4, 4.5 kb) and polystyrene nanospheres (5.4, 20 nm diameter; 2.3, 100 nm). Aqueous-phase diffusion (Dw) in solutions crowded with Ficoll-70 (0-60 wt%) was reduced by up to 650-fold in an exponential manner: Dw = D(o)w exp (-[C]/[C]exp), where [C]exp is the concentration (in wt%) of crowding agent reducing D(o)w by 63%. FCS data for all solutes and Ficoll-70 concentrations fitted well to a model of single-component, simple (non-anomalous) diffusion. Interestingly [C]exp were nearly identical (11+/-2 wt%, SD) for diffusion of the very different types of macromolecules in Ficoll-70 solutions. However, [C]exp was dependent on the nature of the crowding agent: for example, [C]exp for diffusion of rhodamine green was 30 wt% for glycerol and 16 wt% for 500 kDa dextran. Our results indicate that molecular crowding can greatly reduce aqueous-phase diffusion of biologically important macromolecules, and demonstrate a previously unrecognized insensitivity of crowding effects on the size and characteristics of the diffusing species.     

Measuring the translational diffusion coefficients of small DNA molecules by capillary electrophoresis

The apparent translational diffusion coefficients of four 20 base pair (bp) DNA oligonucleotides with different sequences have been measured by capillary electrophoresis, using the stopped migration method. The diffusion coefficients of the four oligomers were equal within experimental error, and averaged (120 +/- 10) x 10(-8) cm(2) s(-1) in 40 mM Tris-acetate-EDTA buffer at 25 degrees C. Since this value is nearly identical to the translational diffusion coefficient determined for a different 20-bp oligomer using other methods, the stopped migration method can accurately measure the diffusion coefficients of small DNA oligomers. The apparent diffusion coefficient of a 118-bp DNA restriction fragment was also measured by the stopped migration method. However, the observed value was approximately 25% larger than expected from other measurements, possibly because the diffusion coefficients of larger DNA molecules are somewhat dependent on the ionic strength of the solution.