Past exam of the biological physics course of the University of Oxford / 2017 / 5

At initiation, RNA polymerase recognizes a promoter and forms a closed bound complex; bacterial promoter recognition involves its specificity factor. Local strand opening exposes the template in an open complex, after which initial RNA synthesis can precede promoter escape into elongation. Binding, opening and productive escape are distinct transitions.

For DNA unwinding measured by magnetic tweezers, magnetic tweezers report the mechanics of a torsionally constrained DNA tether through the position of an attached magnetic bead. Axial magnetic force extends the tether, while imposed rotation controls its linking constraint. Video localization measures extension changes. The important physical relation isOpening a small helical region decreases twist, so the rest of the tether acquires compensating positive writhe. On a negatively supercoiled tether this removes some writhe magnitude and raises its extension; on a positively supercoiled tether it increases writhe magnitude and lowers extension. A change with opposite signs under the two torsional conditions therefore distinguishes unwinding from simple compaction, which would shorten both tethers. Calibrating extension against imposed turns converts a mechanical step into a change in helical winding. This is a conceptual readout of local opening, rather than a direct optical image of separated bases.^[ref]

Model the target as a perfectly absorbing sphere of reaction radius $r_0$, with a fixed target and a dilute far-field polymerase density $c_{\infty }$. Steady radial diffusion satisfies ${\nabla }^{2}c=0$, $c(r_0)=0$, $c(\infty )=c_{\infty }$, henceTherefore, for a rate law multiplying number densities,For the earliest diffusive target encounter among independent searchers, with $N_p$ independent well-mixed searchers in volume $V$, the approximate first-arrival rate is $N_p{k}_a/V$, giving ${\tau }_3\simeq V/(N_p{k}_a)$. The target's effective reaction radius includes encounter geometry and is not uniquely determined by its contour length.

A numerical estimate needs temperature, cell volume and a capture-geometry convention, none of which is fixed numerically in the source. State representative estimates $T=300\mathrm{K}$ and $V=1\mu {\mathrm{m}}^3$. The Stokes–Einstein estimate for a sphere of radius $15\mathrm{nm}$ in the stipulated viscosity isUsing the target contour scale $40\times 0.34\mathrm{nm}=13.6\mathrm{nm}$ as an order-of-magnitude $r_0$ gives $k_a\simeq 2.50\times 10^{-19}{\mathrm{m}}^3{\mathrm{s}}^{-1}$ andThis is an idealized $10^{-2}$-second-scale estimate, not a unique geometric prediction. Finite cell shape, crowded motion, accessibility and orientational restrictions can reduce the ideal capture rate.

If nonspecific binding merely sequesters searchers and their free fraction is $0.2$, the three-dimensional capture hazard is reduced fivefold: $\tau \simeq {\tau }_3/0.2\simeq 0.067\mathrm{s}$ for those estimates. If bound molecules instead search productively by sliding, a free-time fraction alone cannot determine the total arrival time or its sign of change. That is the distinct facilitated-search mechanism treated in c.

Write $A=V/D_3$ and $B=L/D_1$. The given mean-search scaling is $\tau =A/\lambda +B\lambda$. Three-dimensional relocation sees an effective capture length of order $\lambda$, giving a bulk encounter time $V/(D_3\lambda )$. A sliding excursion explores length of order $\lambda$ in time of order ${\lambda }^2/D_1$; sampling order $L/\lambda$ contour segments gives a total sliding cost of order $L\lambda /D_1$. Prefactors, correlations and repeated visits are coarse-grained in this scaling model.

Small sliding lengths make relocation inefficient; very long sliding wastes time revisiting a local DNA region rather than relocating to another. The optimum balances both contributions:Since $d^2\tau /d{\lambda }^2=2V/(D_3{\lambda }^3)>0$, it is the unique minimum for unrestricted positive $\lambda$. At the optimum, bulk and sliding times are equal. With $s=\lambda /{\lambda }_{\ast }$,

The model is meaningful only over an accessible range between molecular steps and the available chromosome length. An optimum outside that range is replaced by a boundary choice; the unphysical divergences at arbitrarily small or large $\lambda$ describe limits of the scaling expression. A common multiplicative searcher-number factor would not change the optimum, but an exact many-searcher first-arrival mean is not generally obtained by dividing an arbitrary single-searcher mean by their number without a waiting-time assumption.

A conceptual single-molecule measurement follows a labelled bound molecule along an independently mapped DNA contour. Project fitted positions onto contour distance and estimate the mean-squared displacement at several lag times. For unbiased one-dimensional diffusion,Independent localization errors of variance ${\sigma }_{\mathrm{loc}}^2$ add an intercept $2{\sigma }_{\mathrm{loc}}^2$. Finite exposure adds motion blur, and drift, binding transitions and the finite DNA contour must be distinguished from free sliding. A slope fit with those limitations can estimate $D_1$; simply fitting all observed displacements to a noiseless line can bias it.

Retain the printed coefficient. The original PDF really prints $D_1=5\times 10^{10}{\mathrm{m}}^2{\mathrm{s}}^{-1}$, not a negative exponent. In sliding diffusivity with per-step dissociation, for a basepair step $a\simeq 0.34\mathrm{nm}$, an unbiased nearest-neighbour walk has step interval $a^2/(2D_1)$. If the dissociation probability per step is $P$, the mean number of steps is approximately $1/P$, givingFor the printed values, ${\tau }_{\mathrm{bound}}\simeq 1.16\times 10^{-24}\mathrm{s}$ and ${\ell }_{\mathrm{rms}}\simeq 0.34\mu \mathrm{m}$. The latter is a random-walk displacement, not the contour distance travelled in a million steps. The step interval is about $1.16\times 10^{-30}\mathrm{s}$, even shorter than the light-travel time across a basepair by many orders of magnitude. This is an unphysical extrapolation for a protein, not a meaningful experimentally measurable diffusivity.

Consequently no fluorescence camera could temporally resolve trajectories under the stated formal values, regardless of the spatial excursion's length. An actual corrected exponent cannot be inferred uniquely from the paper. If a physically plausible $D_1$ were supplied instead, resolving sliding would require exposures substantially shorter than ${\tau }_{\mathrm{bound}}$ and diffusion displacements larger than localization error, with enough detected photons. The dependence on $D_1$ is explicit above, so no guessed source correction is needed.