Fig 1.
Schematics explaining design of PBP-biosensors.
A) Crystal structures of apo (1ANF) and holo (1OMP) states of periplasmic binding proteins (PBPs), with schematic representations underneath. N and C termini are highlighted in blue and red respectively. B) Circularly permuted GFP has its N and C termini close together, so it can be inserted into PBPs to generate biosensors, but it is difficult to predict sensor activity from insertion site. C) Our method takes holo PBP structure as input, and using molecular dynamics simulations can provide insertion sites to generate functional biosensors.
Fig 2.
RMSD of molecular dynamics simulations of Maltose Binding Protein (MBP) in apo and holo forms.
A, B: Holo simulations were generated from the starting point of the MBP/maltose complex crystal structure. A) Individual traces for 10x100ns MBP holo simulations. B) The mean, minimum, and maximum RMSD values for the MBP holo simulations. C, D: Apo simulations were generated from the starting point of the MBP/maltose complex crystal structure with maltose removed. C) Individual traces for 10x200ns MBP apo simulations. D) The mean, minimum, and maximum RMSD values for the MBP apo simulations. E) Final poses of MBP in representative holo/closed (purple) and apo/open (orange) simulations, chosen for RMSD close to the mean-of-means for each condition. Maltose is shown in the holo structure in green.
Fig 3.
A comparison of metrics attempting to identify regions of MBP primary sequence known to be receptive to effector protein insertion with highlighted regions of interest.
Across all figures, the residue ranges 167–173 and 335–348 are highlighted. 167–173 contains the region of interest 169–171, with flanking residues also highlighted for ease of visualisation. A) The change in backbone dihedral angle between holo and apo crystal structures (PDB 1ANF, 1OMP respectively). B) Change in backbone dihedral angle between holo and apo molecular dynamics simulations. Neither regions 167–173 nor 335–348 contain significant values for this metric. C) Difference in RMSF of each residue between holo and apo molecular dynamics simulations. D) Pearson correlation coefficient metric for all apo MBP simulations where the closed-to-open state transition was observed, filtered to only show peaks in the top percentile for mean, median, and maximum pixel value. Negative Pearson correlation coefficient indicates residues that become closer to each other throughout the closed-to-open transition, and positive values closer.
Fig 4.
Comparison of fluorescence of designed sensors at varied maltose concentrations.
On the left of the axes are the control conditions. cpGFP: the fluorescent protein alone. MBP: the Maltose Binding Protein alone. 131 and 195: cpGFP inserted at MBP’s positions 131 and 195, which are both in areas of flat Pearson space so are predicted to generate poor quality sensors. 170 and 348: cpGFP inserted at MBP’s positions 170 and 348, both very strong sensors previously identified by Nadler et al. [12]. Presented are the means and standard deviations of 3 biological repeats per condition. Tables for all values are in supplement. A) Raw fluorescence of sensors at 0, 1, and 100 mM maltose. B) Relative change in fluorescence from 0 mM maltose to 1 or 100 mM, calculated by (F0—FX) / F0. 0.25 and -20.5 are indicated with dashed lines.