Session S20: Single-Molecule Techniques and Enzymes

Multi-scanning of single DNA Molecules in a Dual Nanopore Device

Single-molecule techniques can access to biomolecular information that is otherwise masked by ensemble averaging, and that can lead to new research discoveries or clinically applications. Solid-state nanopores offer a compact and electrical single-molecule sensing approach. However, single reads of molecules taken with solid-state nanopores generally do not possess sufficient quality for applications that otherwise require alignment of multiple reads of molecules from heterogeneous samples. Here we present a single-molecule manipulation and sensing approach based on a dual nanopore device that can linearize DNA and improve read quality by enabling repeated scanning at reduced speeds of the same molecule. A DNA molecule is first co-captured by the two pores with unbalanced voltage forces applied at the pores (tug-of-war state), and then repeatedly scanned back-and-forth by automated voltage control logic. The method enables detection of sequence-specific protein tags that are used for triggering the motion control during dual current sensing. We have achieved up to 100's of scans, and analysis of multiple scans from a single molecule demonstrates a reproducible binding pattern, or molecular barcode, with tag spacings corresponding to expected sequence positions of the tags.   

Computational design and interpretation of single-RNA translation experiments.

Recent advances in time-lapse fluorescence microscopy allow live-cell quantification of ribosome kinetics at single-molecule resolution. Here, we integrate single-molecule experimental data and stochastic models to investigate canonical and non-canonical translation processes. In a first study, we developed a new fluorescent tag system that allowed us to observe, for the first time, ribosomal frameshifting at single-molecule resolution. Our results corroborate that frameshifting is a bursty process, where the RNA stochastically switch between non-frameshifting and frameshifting state (Lyon, K., et al., 2019. Molecular Cell). In a second study, we developed novel stochastic models to estimate biophysical parameters, such as ribosomal elongation and initiation rates. Our methods were used to simulate single-molecule experiments under multiple imaging conditions and for thousands of human genes, and we evaluate which experiments provide accurate estimates of elongation kinetics. With this, we present an interpretation for the well-established experimental procedures, including Correlation Spectroscopy, ribosome Run-Off Assays, and FRAP (Aguilera, L., et al., 2019, Plos Comp Biology).   

Single-Molecule Sensitivity in Mass Spectrometry Using Nanoscale Ion Sources

Electrospray ionization has revolutionized the analysis of biomolecules by softly transforming molecules in solution into gas-phase ions, which can be analyzed by mass-spectrometry. Due to the chaotic process by which the electrosprayed droplets break down into ever-smaller droplets, eventually yielding singly charged ions via ion evaporation, only a small fraction of the sample molecules makes it to the mass analyzer. We have shown that a nanoscale ion source allows ions to evaporate directly off of the meniscus, bypassing the wasteful droplet evaporation process. The small area of the liquid-vacuum interface also prevents significant evaporative heat loss, enabling the use of volatile solvents like water. This technique has been shown to work with simple salt solutions as well as aqueous solutions of biomolecules such as amino acids, nucleic acids, and small peptides. Adoption of these nanopore ion sources could drastically improve the sensitivity of mass spectrometry experiments and open the door to a wide range of potential applications, including single-molecule protein sequencing and single cell proteomic studies.
  

Brownian motion alone may not be sufficient to supply ribosomes with tRNA during translation elongation

The construction of artificial cells holds significant promise for biotechnology and medicine. However, substantial work remains to make such cells modelable. In particular, the set of functions necessary for life remains elusive, most acutely demonstrated by the existence of genes of unknown function that are essential for life¹.

We hypothesize that some of the functions that remain unknown pertain to the physics of how molecules organize to produce cellular functions. To test our hypothesis we introduce translation elongation in E. coli as a model system. Due to its central role in protein production, elongation is likely to have undergone tremendous selective pressure to operate at the limits of physics. We construct a colloidal model of elongation and demonstrate that variable local volume densities around ribosomes at different growth rates may lead to competing trade-offs in tRNA transport and reaction. We then use lower-bounding simulations to show that tRNA transport via Brownian motion alone may be insufficient to recover experimentally determined elongation rates. Overall, we predict that currently unknown mechanisms for speeding up tRNA transport or biasing its spatial organization are implicated in translation elongation.

¹Maheshwari et al. Phys. Rev. Fluids (2019)   

Enhanced diffusion and enzyme dissociation at high substrate concentration

The concept that catalytic enzymes can act as molecular machines transducing chemical activity into motion has conceptual and experimental support, but much of the claimed support comes from experimental conditions where the substrate concentration exceeds k_M (the Michaelis-Menten constant), meaning that it exceeds concentrations that are biologically relevant. Moreover, many of the enzymes studied experimentally to date are oligomeric. Urease, a hexamer of subunits, has been considered to be the gold standard demonstrating enhanced diffusion. Here we show that urease and certain other oligomeric enzymes of high catalytic activity above k_M dissociate into their smaller subunit fragments that diffuse more rapidly, thus providing a simple physical mechanism of enhanced diffusion in this regime of concentrations. Data for urease are presented in the main text and the conclusion is validated for hexokinase and acetylcholinesterase with data presented in supplementary information. For substrate concentration regimes below k_M at which these enzymes do not dissociate, we validate that enzymatic catalysis does lead to the enhanced diffusion phenomenon.   

Stochastic Simulation to Visualize Gene Expression and Error Correction in Living Cells

Stochastic simulation can make the molecular processes of cellular control more vivid than the traditional differential-equation approach by generating typical system histories, instead of just statistical measures such as the mean and variance of a population. Simple simulations are now easy for students to construct from scratch, that is, without recourse to black-box packages. In some cases, their results can also be compared directly to single-molecule experimental data. After introducing the stochastic simulation algorithm, we give two case studies, involving gene expression and error correction, respectively. For error correction, several proofreading models are compared to find the minimal components necessary for sufficient accuracy in translation. Animations of the stochastic error correction models provide insight into the proofreading mechanisms. [Ref: KYC, DMZ, PCN, "The Biophysicist" in press.]   

Single-pair measurements of nonradiative energy transfer efficiency between quantum dots and organic dyes reveals orientational variation

Förster resonance energy transfer (FRET) is a technique that can be used to measure the nanoscale distances and to report on biomolecular dynamics at the single-molecule level. Organic dyes FRET pairs are commonly used in biophysical studies, but this approach suffers from the fact that the organic dyes have short bleaching lifetimes. Quantum dots have longer lifetimes, which makes them ideal for use as FRET donors in longer experiments. However, the energy transfer between quantum dots and organic dye molecule acceptors has not formerly been characterized at the single-molecule level. Therefore, we used TIRF microscopy to observe surface immobilized FRET pairs separated by known lengths of duplex DNA. We used two methods to construct these assemblies. In both cases, we observed broad distributions of FRET efficiencies and little correlation between duplex length and the average FRET efficiency. However, when we probe multiple donor-acceptor distances for a particular quantum dot, we observe the expected decrease in FRET efficiency with increasing duplex length. We attribute these contradictory observations to variations in the orientation factor for each single quantum dot, based on its inherent dipole moment and the location at which the DNA molecule is coupled to its surface.   

Bayesian nonparametrics allow for super resolution microscopy without photo-switching

Super resolution microscopy permits direct observation of biomolecules. However, due to noise, distinguish single molecules demands specialized analysis models that facilitate the interpretation of the acquired images. In such models, the positions of the molecules are represented by random variables and conventionally the total number of random variables in a model is kept fixed and finite. However, prevailing uncertainty in the number of individual molecules contributing photons to the images, makes existing approaches inappropriate as they rely on pre-specification of the number of unknown parameters. Switching of the fluorophores between bright and dark states, induced by experimental means (STORM, PALM, PAINT), ensures that each time at most one fluorophore is visible leading to a convenient solution. Nevertheless, photo-switching requires specialized fluorophores and long experiments necessary to collect photons from multiple switching events. In the talk, I will walk through recent modeling advances and highlight how Bayesian nonparametrics can be used to achieve super resolved localization of single molecules without photo-switching and so allowing for super resolution microscopy with less specialized fluorophores and significantly shorter experiments.   

Intensity Distribution of a Dilute Solution of Point Emitters under Gaussian Detection

In single-molecule techniques, one often employs a Gaussian illumination profile to measure fluorescence intensities from a dilute solution of molecules. Typically such measurements are performed by single-photon detection and result in a sequence of single-photon arrival times. Here we develop a general framework for calculating the intensity distribution of a dilute solution of Point emitters under Gaussian Detection. We will show that the resulting Intensity distributions dramatically change character at low emitter concentrations when considering different dimensionalities of Gaussian Detection. In particular the one- and two- dimensional solutions are strongly structured, exhibiting discontinuities in their derivative at integer multiples of the brightness. This makes these distributions strong candidates for applications that distinguish and measure concentrations of mixtures of emitters of different brightnesses. Finally, we will discuss a maximum likelihood method to determine concentration and brightness from a sequence of single-photon arrival times.   

Reliable Imaging of Dynamic Structures in Live-Cell Super-Resolution

Super-resolution imaging by single-molecule localization is a powerful tool for visualizing structures inside of cells that are smaller than the diffraction limit. However, imaging in living matter poses a challenge because structures can be mobile and dynamic, yet single-molecule localization microscopy can require a long time to generate images. In this talk, I will present a metric for assessing image reliability, based on robust estimation of the density of detected molecules. This simple metric quantitatively defines the tradeoff between spatial and temporal resolutions, and their relationships to acceptable levels of noise in density estimation. I will discuss how this implies variations in resolution across a data set, and present a tool for visually assessing data quality.   

Controlled sliding of DNA knots in solid-state nanopores

Knots in DNA are useful for investigating the static and dynamic physics of biopolymers in solutions. Recently, we showed nanopores that probe the equilibrium structures of DNA knots in solutions. With unprecedented statistics, our distributions of sizes of equilibrium DNA knots and their positions provide important insights relevant from biological, physical, and technological perspectives. Although nanopores enable statistically significant single molecule investigation, these distributions get biased due to sliding of knots through nanopores, adversely affecting the conclusions. Using our experiments and parametrized simulations, we demonstrate complete control over sliding of knots in nanopores over a molarity range of 1M-3M. We show that controlling the experimental parameters enabled the detection of equilibrium knots in both linear and circular DNA molecules. These results open up an exciting opportunity for statistically significant biophysical investigation of knots in single biomolecules, generating unprecedented insights useful from both scientific and technological perspectives.   

Optimal Control and Reinforcement Learning of Regulation and Enzyme Activities

Experimental measurement or computational inference/prediction of the enzyme regulation needed in a metabolic pathway is hard problem. Consequently, regulation is known only for well-studied reactions of central metabolism in various model organisms. In this study, we use statistical thermodynamics and metabolic control theory as a theoretical framework to calculate enzyme regulation policies for controlling metabolite concentrations to be consistent with experimental values. A reinforcement learning approach is utilized to learn optimal regulation policies that match physiological levels of metabolites while maximizing the entropy production rate and minimizing the heat loss. The learning takes a minimal amount of time, and efficient regulation schemes were learned that either agree with theoretical calculations or result in a higher cell fitness using heat loss as a metric. We demonstrate the process on four pathways in the central metabolism of Neurospora crassa (gluconeogenesis, glycolysis-TCA, Pentose Phosphate-TCA, and cell wall synthesis) that each require different regulation schemes.   

Binding Energies of Fn Cas12a

In recent years, CRISPR proteins have attracted much attention due to their use as a DNA binding platform, in which the protein undergoes steps of PAM attachment, crRNA-DNA inspection, and reconfiguration. Thermodynamically, we can model determinants of binding energies for different sites of the genome with a partition function that reflects energetic costs associated with base pair mismatches. From suitably chosen weights that assign higher energetic contributions to base pair mismatches among the first $6$ positions of a DNA sequence, appropriate transition probabilities for a random walk $X$ will be defined so as to coincide with base pair mismatches. As $X$ approaches the position of binding, we stipulate that none of the transition probabilities of $X$ from all positions before $N$ vanish. Furthermore, with such a probabilistic approach, we will analyze the energy landscape of Fn Cas12a, leading to a more comprehensive understanding of how the binding energy of a sequence is dependent on individual base pairs. More broadly, this work will also study gRNA and DNA sequences that are conducive to binding in Fn Cas12a.