PMC

(A) Under maintenance conditions, the lac operon is OFF (indicated by the solid red line) and the inducer TMG remains extracellular; stochastic events that lead to a transient derepression of the lac operon will initiate an autocatalytic positive-feedback response (indicated by solid blue lines). The box highlights the first three codons of the wild-type lac repressor gene and the Lys-Lys additions encoded by the A₉ and A₅GA₃ lacI alleles (in red). (B) The lacI A₉ and A₅GA₃ Lys-Lys N-terminal addition alleles encode functional lac repressors. Representative flow cytometry analyses measuring GFP fluorescence of OFF and ON populations of wild-type (red, brown), A₉ (blue, light blue) and A₅GA₃ (green, violet) lacI cells produce identical histograms; 10⁴ cells of each strain were interrogated. (C) Forward lacI ⁺→lacI ⁻ mutation frequencies. No significant difference in mutation frequency between the A₉ and A₅GA₃ strains is observed (Mann-Whitney Rank Sum Test, p = 0.23). The wild-type strain is added for comparison.

Wild-type genes (black parallel lines) make wild-type transcripts (blue wavy lines) make wild-type functional proteins (blue circles); mutant genes make mutant transcripts (red crosses) make mutant proteins (red circles); protein mis-folding can trigger phenotypic change by changing protein conformation to the prion state (red triangle) that can self-perpetuate by templating the aberrant conformation with nascent native proteins (blue triangles). From wild-type genes can also come altered mRNA (epimutation) making altered proteins that can perturb transcriptional networks in a nonlinear manner generating a heritable phenotypic change (red arrows) from a transient stochastic error in information transfer. In this case no trace of the error will remain in the lineage after the phenotypic change as indicated: while change through mutation will retain evidence of the original stochastic error in the progeny cell (mutant DNA, mutant RNA and mutant protein), change through epimutation will retain no evidence of the original stochastic error in the progeny cell (WT DNA, WT RNA and WT protein). Errors in DNA and RNA synthesis occur at rates of, very roughly, 10⁻⁹ and 10⁻⁵ errors per residue, respectively ; yeast cells in the non-prion [psi ⁻] state spontaneously switch to [PSI ⁺] at a frequency of 10⁻⁶; the great majority of cells will not have sustained any errors in information transfer.

(A) Representative flow cytometry GFP fluorescence histogram series of A₉ and A₅GA₃ lacI cells that were originally ON (green histograms) or OFF (red histograms) were sub-cultured and grown in media containing various concentrations of TMG indicated on the vertical axis (10⁴ cells interrogated). Below 5 µM TMG and above 20 µM TMG, the previous history of the cell (ON or OFF) does not affect the current state of the cell; between these TMG concentrations the system exhibits hysteresis. The shaded area highlights the maintenance concentration of 9 µM TMG for these strains. (B) Cells that were originally ON or OFF were sub-cultured and grown in media containing various concentrations of TMG, as above. Each value is the average ± SD from 5 to 15 independent cultures. The shaded area highlights the maintenance concentration of 9 µM TMG for these strains. (C) OFF A₅GA₃ lacI cells (red histograms) and A₉ lacI cells (blue histograms) were diluted and grown in media containing 9 µM TMG. After 42 h growth, flow cytometry was performed to determine the frequency of epigenetically ON cells in 20 independent cultures of each strain; the A₅GA₃ histograms are superimposed over the A₉ histograms (10⁴ cells interrogated). (D) The A₉ epigenetic-switch frequency is significantly increased over the A₅GA₃ value (Mann-Whitney Rank Sum Test, p<0.001).

(A) Single lacI A₉ cells in minimal succinate media ± maintenance TMG were grown into microcolonies in a microfluidic flow chamber (approximately 100 cell divisions per microcolony originating from a single cell; number of divisions equals final number of cells in a microcolony minus 1). Comparison of bright field images (panel series on the left) with GFP fluorescence images (panel series on the right) allows clear distinction between OFF and ON cells in the microcolony. (I) panels show microcolonies that arose from single ON cells that were subsequently grown in the presence of maintenance level TMG; (II) panels show mirocolonies that arose from single OFF cells that were subsequently grown in the presence of maintenance level TMG; (III) panels show mirocolonies that arose from single OFF and single ON cells that were subsequently grown in the presence of maintenance level TMG; (IV) panels show a mirocolony that arose from ON cells that were subsequently grown in the absence of maintenance level TMG. Exposure to fluorescence illumination was 3000 ms. (B) A single lacI A₉ cell in minimal succinate media+maintenance TMG was grown into a microcolony in a microfluidic flow chamber and monitored by time-lapse fluorescence microscopy. Presented here are four still images from a full time series of images (available as ; images shown correspond to frames 25, 28, 30, 39). Comparison of bright field images (panel series on the left) with GFP fluorescence images (panel series on the right) allows distinction between OFF and ON cells in the microcolony. (I) all cells in the microcolony are OFF (100 ms exposure to fluorescence illumination); (II) a recently divided cell is just becoming ON indicated by the arrow (100 ms exposure to fluorescence illumination); (III) the now separated cells have become ON (100 ms exposure to fluorescence illumination); (IV) further cell division has occurred in the microcolony creating one lineage of ON cells amongst many other lineages of OFF cells (3000 ms exposure to fluorescence illumination). The 100 ms exposure time images were over-exposed using the ColorSync Utility to observe the faint fluorescence signal.

(A) An operon fusion was created by first inserting a kanamycin cassette from pKD4 () into the intervening region between lacI and lacZ and then, via a flippase reaction, removing most of lac operator O₃, the complete lac promoter and lac operator O₁. Therefore, the lacIZYA fusion transcript is under the expression of the weakly constitutive lacI promoter with no interference from any lac repressor binding (lac repressor does not negatively regulate lac expression through O₂ alone) . The complete sequence of the intervening region from the TGA stop codon of the lacI gene to the ATG start codon of the lacZ gene is shown in . (B) Expression levels of the lacIZYA operon fusion strains are equivalent regardless of GreA, GreB or GreAB status indicating the absence of any or all Gre functions does not influence overall lac expression levels. Cells were grown in minimal A salts plus glucose and β-galactosidase levels were determined by the method of Miller ; the average ± SD for three independent cultures is shown. To make functional β-galactosidase in this fusion strain the transcription complex must produce at least a 4,271 nt transcript including the lacI non-translated leader, the lacI gene, the FRT scar sequence and the lacZ gene; if the transcript terminates after lacA, at the usual lac termination site, then the entire transcript will be over 6.2 kb in length. (C) At maintenance level of TMG, the absence of GreA or GreB does not increase stochastic switching over wild-type levels but when both Gre functions are absent a significant increase in stochastic switch frequency is observed . Representative flow cytometry histograms of wild-type, ΔgreA, ΔgreB and ΔgreAB cells that were originally ON (green histograms) or OFF (red histograms) were sub-cultured and grown in media containing a maintenance level of 6 µM TMG. All strains are equally responsive to TMG at this concentration (all ON populations remain ON, i.e., maintain their previous state), but only OFF ΔgreAB cells exhibit an increased stochastic switching frequency over that observed when wild-type OFF cells were grown at maintenance level of TMG; each histogram represents the interrogation of 10⁴ cells. Therefore, it is not an overall decrease of lacI expression that causes the significant increase in stochastic switch frequency in the absence of GreAB.

(A) Stochastic phenotypic switching is significantly increased when the error-prone A₉ run is in a transcription fidelity-deficient background (ΔgreA ΔgreB cells). OFF ΔgreAB A₅GA₃ lacI cells (red histograms) and ΔgreAB A₉ lacI cells (blue histograms) were diluted and grown in media containing 9 µM TMG. After 42 h growth, flow cytometry was performed to determine the frequency of epigenetically ON cells in 17–19 independent cultures of each strain; the histograms from the ΔgreAB A₅GA₃ lacI cultures are superimposed over the histograms from the ΔgreAB A₉ lacI cultures; each histogram represents the interrogation of 10⁴ cells. (B) The median for the ΔgreAB A₉ lacI strain is significantly different from the ΔgreAB A₅GA₃ lacI value (Mann-Whitney Rank Sum Test, p<0.001). (C) To model translation frameshifting in our system we have created merodiploids that provide a 10-fold excess of wild-type transcripts over ±1 frameshift transcripts (as modeled by the A₈ and A₁₀ lacI alleles). Therefore, the ratio of wild-type transcript over frameshifted transcript, at the level of transcription (10∶1), will be a very conservative approximation of the situation that would arise if during the translation of one A₉ transcript, one translational frameshift event would occur (20∶1 wild-type sub-units over frameshifted sub-units). The wild-type repressor allele is completely dominant over the frameshifted repressor alleles: left panel, the lacI allele strains without the F′; right panel, the lacI allele strains with the F′ overproducing wild-type lacI. The glucose minimal plates include Xgal (40 µg/ml) and tetracycline (Tet, 12.5 µg/ml), as indicated beneath the plate. Tet is used to maintain the F′ in the cell. (D) Quantitative measurement of the phenotype observed in (C). The level of β-galactosidase in all four strains is comparable and does not exceed 1 Miller unit, which is the basal β-galactosidase of uninduced E. coli cells ; the average ± SD for three independent cultures is shown. (E) A −1 translational frameshifting event at the A₉ sequence would cause translation to terminate at codon 4/5 (green line denotes wild-type protein; gray line denotes frameshifted protein; red X denotes translation termination; blue line denotes translation reinitiation protein; the A₉ transcript is shown as a black line with the GUG start codon in green letters and the UGA stop codon in red letters; the protein domain structure is indicated above the translation products). Therefore, no functional lac repressor sub-unit could be produced; however, it has been shown that a dominant-negative sub-unit could be produced by translational reinitiation . Reinitiation could occur at codons 23, 24, 38 or 42 , producing repressor sub-units lacking the DNA-binding domain but the core aggregation domain would be intact and able to bind and interfere with wild-type sub-unit function. Therefore, there is the possibility that one −1 translational frameshifting event would not only decrease the net total of repressor sub-units by one, but might also decrease the cell's net lac repressors by one, since it has been shown that one dominant-negative sub-unit with three wild-type sub-units may abolish the function of the tetrameric lac repressor . A +1 translational frameshifting event at the A₉ sequence (generating a A₁₀ transcript) would cause translation to terminate at codon 83/84 and would therefore only result in the net decrease of one repressor sub-unit in the cell, since no dominant-negative sub-unit can be made this far into the core domain. When the wild-type sub-unit is made at 10-fold the level of ±1 transcription frameshift events (and ±1 translation protein products), the wild-type sub-units dominate and the lac operon is repressed (as seen in the Xgal Tet F′ lacI^q plate in (C) and therefore any net decrease by one translation frameshift event is negligible when compared to the net decrease in repressor sub-units due to a transcription error. When a transcription error occurs at the A₉ sequence, all the nascent lac repressor sub-units will be non-functional (and/or dominant-negative); when a translational frameshift occurs at the A₉ sequence, less than 1/10 of all nascent lac repressor sub-units will be non-functional (and/or dominant-negative).