(a) Power spectral density (PSD) for temporal signal fluctuations during sub-saturating ligand stimulation of 18 cells with wild-type receptor complement (light blue, CheRB-, TSS58) and 58 cells expressing only chemoreceptor Tsr (orange, CheRB- Tsr+, TSS1964/pPA114). Also shown, for comparison, are PSDs from experiments without ligand stimulation for WT cells and CheRB- cells (red and dark blue, respectively; same data as in Figure 4). Error bars represent standard error of the mean. We note that the Tsr+ experiment had a larger FRET amplitude scaling factor (see Materials and methods) compared to the standard conditions under which the other strains were measured, and to account for this difference, the Tsr+ power spectrum has been scaled by a factor , where to account for this difference. (b) (Top) Stimulus protocol for modulation of the L-serine ligand concentration ([L](t)). Cells were incubated either in buffer ([L]=0, white) or a subsaturating stimulus ([L] = 50 μM, gray) for 1> hr. A saturating stimulus ([L] = 1 mM, black) is applied at the end of the experiment. (Bottom) Population- averaged time series for adaptation-deficient cells with wildtype receptor complement (CheRB-, TSS58) for experiments with (18 cells, light blue) and without (17 cells, dark blue) a sustained 50 μM L-serine stimulus during the time interval used to compute the PSDs in panel a (indicated by the red bar). (c) (Top) Stimulus protocol for L-serine concentration ([L](t)). At the start of the experiment, a saturating concentration ([L] = 1 mM, black) is applied for a short time. After flushing buffer ([L]=0, white), an intermediate concentration ([L] = 25 μM, gray) is sustained for 10 min. (Bottom) Population-averaged time series of 58 adaptation-deficient cells expressing Tsr as the sole chemoreceptor (RB-Tsr+; TSS1964/pPA114) under the stimulus protocol indicated above. (d) Selected single-cell time series of the population shown in panel (c), each normalized to its activity level before adding the first stimulus. To the unfiltered data (gray) a 7 s moving average filter is applied and superimposed (colored according to categories in panel (e)). All time series and corresponding activity histograms of the same experiment are shown in Figure 5 - Supplement 1 and 5. (e) Classification of RB-Tsr+ single cell fluctuation phenotypes by the number of stable activity levels observed during the sustained subsaturating stimulus. Many cells show only one stable activity level (yellow), corresponding to either full-amplitude response () or no response (). Some cells show two (red) or more (purple) apparently stable states. In other cells, fluctuations appeared chaotic with no discernibly stable state (black). (f) Definitions for analysis of two-state switching dynamics. The transition timescales and were determined by fits of a symmetric exponential function (see main text) to the upward (cyan) and downward (purple) switching transients, respectively. Residence times were defined as the interval between two successive transitions, at 50 activity. (g) Histogram of transition timescales, (4.2 ± 2.2 s, 26 events, cyan) and (3.5 ± 3.2, 29 events, purple) from 10 two-state switching cells of a single experiment with 1 Hz acquisition frequency. (h) Mean residence times and for two-state switching cells as a function of the average activity bias . The slopes are and , and the crossover point at s defines a characteristic switching timescale. Data of 17 cells from three independent experiments (one at 1 Hz acqusition, two at 0.2 Hz acquisition).