Bayesian machine learning analysis of single-molecule fluorescence colocalization images.

Multi-wavelength single-molecule fluorescence colocalization (CoSMoS) methods allow elucidation of complex biochemical reaction mechanisms. However, analysis of CoSMoS data is intrinsically challenging because of low image signal-to-noise ratios, non-specific surface binding of the fluorescent molecules, and analysis methods that require subjective inputs to achieve accurate results. Here, we use Bayesian probabilistic programming to implement Tapqir, an unsupervised machine learning method that incorporates a holistic, physics-based causal model of CoSMoS data. This method accounts for uncertainties in image analysis due to photon and camera noise, optical non-uniformities, non-specific binding, and spot detection. Rather than merely producing a binary 'spot/no spot' classification of unspecified reliability, Tapqir objectively assigns spot classification probabilities that allow accurate downstream analysis of molecular dynamics, thermodynamics, and kinetics. We both quantitatively validate Tapqir performance against simulated CoSMoS image data with known properties and also demonstrate that it implements fully objective, automated analysis of experiment-derived data sets with a wide range of signal, noise, and non-specific binding characteristics.

UvrD helicase activation by MutL involves rotation of its 2B subdomain.

Escherichia coli UvrD is a superfamily 1 helicase/translocase that functions in DNA repair, replication, and recombination. Although a UvrD monomer can translocate along single-stranded DNA, self-assembly or interaction with an accessory protein is needed to activate its helicase activity in vitro. Our previous studies have shown that an Escherichia coli MutL dimer can activate the UvrD monomer helicase in vitro, but the mechanism is not known. The UvrD 2B subdomain is regulatory and can exist in extreme rotational conformational states. By using single-molecule FRET approaches, we show that the 2B subdomain of a UvrD monomer bound to DNA exists in equilibrium between open and closed states, but predominantly in an open conformation. However, upon MutL binding to a UvrD monomer-DNA complex, a rotational conformational state is favored that is intermediate between the open and closed states. Parallel kinetic studies of MutL activation of the UvrD helicase and of MutL-dependent changes in the UvrD 2B subdomain show that the transition from an open to an intermediate 2B subdomain state is on the pathway to helicase activation. We further show that MutL is unable to activate the helicase activity of a chimeric UvrD containing the 2B subdomain of the structurally similar Rep helicase. Hence, MutL activation of the monomeric UvrD helicase is regulated specifically by its 2B subdomain.

Regulation of UvrD Helicase Activity by MutL.

Escherichia coli UvrD is a superfamily 1 helicase/translocase involved in multiple DNA metabolic processes including methyl-directed mismatch DNA repair. Although a UvrD monomer can translocate along single-stranded DNA, a UvrD dimer is needed for processive helicase activity in vitro. E. coli MutL, a regulatory protein involved in methyl-directed mismatch repair, stimulates UvrD helicase activity; however, the mechanism is not well understood. Using single-molecule fluorescence and ensemble approaches, we find that a single MutL dimer can activate latent UvrD monomer helicase activity. However, we also find that MutL stimulates UvrD dimer helicase activity. We further find that MutL enhances the DNA-unwinding processivity of UvrD. Hence, MutL acts as a processivity factor by binding to and presumably moving along with UvrD to facilitate DNA unwinding.