Bioluminescence imaging of individual fibroblasts reveals persistent, independently phased circadian rhythms of clock gene expression

Abstract

Circadian (ca. 24 hr) oscillations in expression of mammalian "clock genes" are found not only in the suprachiasmatic nucleus (SCN), the central circadian pacemaker, but also in peripheral tissues. Under constant conditions in vitro, however, rhythms of peripheral tissue explants or immortalized cells damp partially or completely. It is unknown whether this reflects an inability of peripheral cells to sustain rhythms, as SCN neurons can, or a loss of synchrony among cells. Using bioluminescence imaging of Rat-1 fibroblasts transfected with a Bmal1::luc plasmid and primary fibroblasts dissociated from mPer2(Luciferase-SV40) knockin mice, we monitored single-cell circadian rhythms of clock gene expression for 1-2 weeks. We found that single fibroblasts can oscillate robustly and independently with undiminished amplitude and diverse circadian periods. Cells were partially synchronized by medium changes at the start of an experiment, but due to different intrinsic periods, their phases became randomly distributed after several days. Closely spaced cells in the same culture did not have similar phases, implying a lack of functional coupling among cells. Thus, like SCN neurons, single fibroblasts can function as independent circadian oscillators; however, lack of oscillator coupling in dissociated cell cultures leads to a loss of synchrony among individual cells and damping of the ensemble rhythm at the population level.

Figure 1

Rat-1 Cell Rhythms (A) Long-term luminometer recording of luminescence in a 35 mm culture dish of Rat-1 cells acutely transfected with the mBmal1::luc circadian reporter plasmid. Cells were transfected one day before the recording began, and treated with a serum shock (50% serum × 2 hrs) just prior to recording. Note the initial circadian oscillations which damp out quickly in the first few days, followed by a slow increase in luminescence to a broad peak at 2-3 weeks after transfection. Even when the luminescence begins to decline, changing medium restores circadian oscillations. (B) Phases of individual Rat-1 cell luminescence rhythms were significantly clustered at the start of each experiment, but randomly distributed by the end. Each blue triangle indicates the phase of one cell, following the convention that 0˚ is the phase of the fitted peak of the luminescence rhythm. The radial line indicates the average start phase (127˚, or 15.7 circadian hrs before peak), and the arc indicates the 95% confidence interval. (C) Plots of bioluminescence over the course of a 2 week experiment beginning 5 days after transfection. Medium was changed just prior to starting the experiment. Plots are of two representative Rat-1 cells (#22, 48), the sum of all 100 cells monitored in the experiment, and the entire microscope field. Note that the two individual cells begin with similar phases, peaking ~15 hrs after the start of the experiment, but that due to differences in period and phase instability the cells have drifted completely out of phase by the end of the experiment. The amplitudes of the individual cell rhythms are variable but do not damp significantly. Cell 22 actually increases its amplitude toward the end of the experiment. A movie of Cell 48 can be seen in Figure S1. As the cells become desynchronized, they begin to cancel out one another’s rhythms, and this is reflected in the rapidly damped oscillations seen in the arithmetic sum of luminescence from all 100 cells monitored, as well as in the luminescence from the entire microscope field. Finally, in accordance with the expected increase in plasmid expression at 5-10 days post-transfection (see A), there is an upward sloping baseline evident in the ensemble rhythms.

Figure 2

Primary Fibroblast Rhythms (A) Bioluminescence images of primary fibroblasts dissociated from tails of mPER2::LUC-SV40 knockin mice, showing circadian rhythms of luminescence. Because the cells emitted only a few photons per minute, detection of single cell luminescence required 30 min exposures and binning of pixels 8 × 8 to reduce read noise per pixel. Cells 1-4 peak near 0 and 24 h elapsed time, whereas cells 5-8 peak about 15 h later. See the movie in Figure S2 for a dynamic view of single cell luminescence rhythms in a larger field of view. (B) Representative circadian bioluminescence rhythms from individual primary fibroblasts, over the course of an experiment lasting more than 8 days. Compared to the Rat-1 cells, the primary fibroblast rhythms are much more robust, with greater phase and period stability, and more regular amplitude. No damping of single cell rhythms was evident.

Figure 3

Desynchrony of Primary Fibroblasts (A) Individual primary fibroblasts expressed diverse circadian periods. Above is a histogram of circadian period values for luminescence rhythms of 178 primary fibroblasts cultured from mPER2::LUC-SV40 knockin mice. Periods averaged 25.65 hrs, but ranged widely from 22.4 to 29.7 hrs. Below is a raster plot showing two cells with clearly different periods. In the raster plot, time of day is plotted left to right, and successive days down the page, such that vertically adjacent points are 24 hrs apart. Each row is extended to 48 hrs, duplicating data in the next row, so that patterns crossing midnight can be appreciated. One cell with a period > 24 hrs is plotted in red, another cell with a period < 24 hrs is plotted in blue, and thick bars designate times when the luminescence for a cell was above the mean for each row. Due to different circadian periods, the cells’ phase relationship changes over time. (B) Phases of fibroblast luminescence rhythms were significantly clustered at the start of each experiment, but randomly distributed by the end. Each blue triangle represents the phase of one cell, following the convention that 0˚ is the phase of the fitted peak of the luminescence rhythm. The radial line indicates the average start phase (295˚, or 4.3 circadian hrs before peak), and the arc indicates the 95% confidence interval. (C) Damping population rhythms emerge from undamped single cell rhythms, as illustrated by plots of bioluminescence over the course of an 11 day experiment. Medium was changed just prior to starting the experiment. Plots are of two representative fibroblasts (#6,10), the sum of all 25 cells monitored, and the entire microscope field. Damped luminescence rhythmicity previously recorded from the same culture dish in the luminometer is also plotted for comparison. Note first that the two individual cells begin with similar phases, peaking ~4-6 hrs after the start of the experiment, but that due to differences in period, the cells have drifted completely out of phase by the end. As cells become desynchronized, they begin to cancel out one another’s rhythms, and this is reflected in the rapidly damped oscillations seen in the arithmetic sum of luminescence from all 25 cells monitored, as well as in the luminescence from the entire microscope field, and finally also in the luminometer recording from the entire dish.

Figure 4

Primary Fibroblast Rhythms Sorted by Start Phase Luminescence rhythms of all 75 primary fibroblasts from one experiment are represented in this plot. Each horizontal raster line represents a single cell, with elapsed time plotted left to right. Luminescence intensity data from all cells were normalized for amplitude, and then color-coded: higher than average values are red, and lower than average values are green. The cells are sorted in order of start phase, so that the emergence of desynchrony can be more easily appreciated.

Figure 5

Primary Fibroblast Rhythms Normalized by Period and Phase Luminescence rhythms of all 25 primary fibroblasts from one experiment were normalized by period and phase, by plotting luminescence as a function of circadian cycles for each cell, in order to reveal any overall trends in amplitude or waveform. Individual cell rhythms are plotted in the upper panel, a different color for each cell. The mean is plotted below, revealing a reasonably stable mean amplitude and waveform over the course of the 11 day experiment, and no damping.