To allow observation of independent circadian rhythms in cocultured cells, we chose the mBmal1 promoter because mBmal1 gene expression shows a clear circadian rhythm in cultured fibroblasts 19.

Rat-1 cell lines stably expressing either the green- or the red-emitting luciferases under the control of the mBmal1 promoter were generated as described in the Methods. We refer to these cell lines as Bmal1–GR and Bmal1–RED, respectively. First, we measured the bioluminescence spectra of these cells, because the spectra are the most important factors for measuring each luciferase activity by splitting the emissions. The cells were trypsinized and harvested and the spectra were measured without destroying the cells. As shown in Figure 1, the bioluminescence spectra of both cell lines are almost the same as those from cellular extracts reported in a previous study 7, demonstrating that each luciferase activity can be measured even when the cells are mixed. These cell lines showed clear circadian oscillations after treatment with DEX, which is known to synchronize circadian oscillation of clock genes in Rat-1 cells 20 (Fig. 2C, D). The circadian periods (mean ± SEM) of Bmal1–GR ± Bmal1–RED cells were 26.2 ± 0.1 h (n = 8) and 23.7 ± 0.1 h (n = 8), respectively.

If individual fibroblasts oscillate following their endogenous circadian periods, as shown previously 1819, then the circadian rhythms of cocultured Bmal1–GR and Bmal1–RED cells would be expected to keep their endogenous rhythms. We first attempted to demonstrate that independent circadian oscillations of two types of fibroblasts could be simultaneously monitored using different color-emitting luciferases. To do this, we triggered antiphasic oscillation in the two cell lines and examined whether rhythmic mBmal1 expression is affected by co-culture.

Cell lines were grown separately in two wells formed by a silicon frame within a 35 mm culture dish (Fig. 2A, B). As controls, monocultures of either Bmal1–GR or Bmal1–RED cells were cultured in one of the two wells, and the other well was left empty (Fig. 2B, left and middle). The times of applying DEX to Bmal1–GR or Bmal1–RED cells were staggered by 12 h. Thus, Bmal1–RED cells were stimulated with DEX first and the Bmal1–GR cells received DEX 12 h later. The same procedure was performed on the monocultures with blank cells adjacent. During the measurement of luciferase activity, the silicon frame was removed from the 35 mm dish so that the two cell lines could establish contact through the medium. The respective luciferase activities of both cell lines were measured simultaneously. The cocultured cell lines showed antiphasic circadian rhythms of each luciferase activity and the rhythm of each cell line was identical to that of the controls (Fig. 2C–E). Reflecting the longer endogenous circadian period of the Bmal1–GR cells, the peak time of Bmal1–GR cells was delayed progressively relative to that of the Bmal1–RED cells over 3 cycles (Fig. 2F, G). There were no significant differences in peak times or rates of damping between cocultures and monocultures (P > 0.05, Mann-Whitney U test; Fig. 2F, G). Although it appears that the average rate of damping of Bmal1–GR cells was different from that of Bmal1-RED cells, the difference was not statistically significant. Thus, the varying gene expression levels in these two cocultured cell lines were monitored successfully.

We next examined whether signaling between neighboring cells plays a role in the synchronization of circadian rhythms. Taking advantage of the differences in circadian periods between the two types of Rat-1 cell lines, we cocultured these cells with direct contact and examined rhythmic mBmal1 expression.

Equal numbers of Bmal1–GR and Bmal1–RED cells were mixed and grown in a dish. As controls, monocultures of either Bmal1–GR or Bmal1–RED cells were cultured individually (Fig. 3A, B). The cells were stimulated by replacing the culture medium with medium containing DEX. Both cell lines showed clear circadian rhythms in their luciferase activities, even when cocultured with direct contact (Fig. 3C). The peak times after the start of DEX treatment and the rate of damping in amplitude were analyzed statistically (Fig. 3D, E). The peak time of Bmal1–GR cells advanced to that of Bmal1–RED cells at peak 1 but it was reversed at peak 2 and the difference was enlarged at peak 3. There were no significant differences in peak times or amplitude-damping rates between the monoculture and coculture systems (P > 0.05, Mann-Whitney U test). Thus, the independence of the circadian rhythms of Rat-1 cells was confirmed using this dual-color luciferase assay system.

To monitor the synchronization process of the fibroblasts, we changed the medium of the two cell lines oscillating in antiphase as described above, because fibroblasts are synchronized by external stimuli such as changing the medium, the application of DEX and temperature changes 18192326.

The medium was replaced 3 d after the DEX treatment of Bmal1–RED cells. After changing the medium, the independently phased circadian rhythms of Bmal1–GR and Bmal1–RED cells were shifted to a new phase and the peaks become closer to each other (Fig. 4). These results are consistent with reports that fibroblast cycles are synchronized by environmental cues rather than a coupling among fibroblasts 181923. This result shows that the phase resetting of the two cell lines could be successfully and simultaneously monitored using the dual-color luciferase assay system.

F2408 (Rat-1) cells (Health Science Research Resources Bank, Osaka, Japan) were grown in Dulbecco's modified Eagle's medium (DMEM; Sigma-Aldrich, St Louis, MO, USA) with 10% fetal bovine serum (FBS; ICN Biochemicals, Aurora, OH, USA), 0.1 mg/ml streptomycin (Nacalai Tesque, Kyoto, Japan), and 100 U/ml penicillin (Nacalai Tesque) at 37°C in a humidified 5% CO₂ incubator. Reporter plasmids, Bp/915-mGr and Bp/915-mRed, respectively expressing green- and red-emitting luciferases under the control of the mBmal1 promoter (-816/+99) 28, were constructed as reported 78. The green- and red-emitting luciferases are commercially available as SLG and SLR (MultiReporter Assay System, Tripluc^®, Toyobo, Osaka, Japan). For stable transfection, Rat-1 cells were cotransfected with the plasmid and pSV-Neo as the expression vector for the neomycin resistance gene using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. After two days of transfection, the cells were subcultured for selection with 500 μg/ml geneticin (G418; Nacalai Tesque, Kyoto, Japan). We named the subcultured G418-resistant cell line with Bp/915-mGr as Bmal1–GR and the cell line with Bp/915-mRed as Bmal1–RED.

Bmal1–GR and Bmal1–RED were grown to confluence in 10 cm dishes. The circadian rhythms were synchronized by replacing the culture medium with DEX medium consisting of DMEM without phenol red (Gibco BRL, Gaithersburg, MD, USA) plus 10% FBS and 100 nM DEX (Nacalai Tesque). The cells were trypsinized and harvested from dishes after 18 h. Cells were centrifuged and suspended in 250 μl of luciferin medium consisting of DMEM without phenol red (Gibco), 10% FBS, 100 nM DEX, and 200 μM D-luciferin (Toyobo, Osaka, Japan). Bioluminescence spectra were measured using an AB1850 spectrophotometer (ATTO, Tokyo, Japan) without destroying the cells.

Bmal1–GR and Bmal1–RED cells were grown in DMEM supplemented with 200 μg/ml G418. To culture two types of Rat-1 cells in separate wells, silicon frames with a 32 mm major diameter and a 26 mm inside diameter with a 3 mm wide divider were cut from a 3-mm-thick silicon rubber plate (Togawa Rubber, Osaka, Japan) (Fig. 3A). The silicon frames were soaked in 70% ethanol (Nacalai Tesque, Kyoto, Japan) and placed on 35 mm dishes. After the ethanol was dried up at room temperature, the frame became stuck tightly to the bottom of a 35 mm dish. Bmal1–GR (2 × 10⁵) cells and Bmal1–RED (2.4 × 10⁵) cells were seeded in separate wells of the silicon frame. After 2–3 d, the culture medium from the Bmal1–RED cells grown to 100% confluence was replaced with DEX medium (time 0 h). Bmal1–GR cells were treated with DEX medium 12 h later (time 12 h). After 2 h of DEX treatment of the Bmal1–GR cells, the medium in both wells was replaced with luciferin medium (time 14 h). The silicon frame was removed from the 35 mm dish so that the two cell lines were in contact through the luciferin medium. The luciferin medium was covered with 2 ml mineral oil (Sigma) to prevent evaporation.

To culture Bmal1–GR and Bmal1–RED cells with direct contact, the two lines were mixed and seeded in 35 mm dishes at a density of 3 × 10⁵ cells each. After 2–3 d, the cells at confluence were treated with DEX medium. After 2 h of DEX treatment, the medium was replaced with luciferin medium and covered by 2 ml of mineral oil.

The real-time monitoring of mBmal1 expression using green- and red-emitting luciferases was performed with a dish-type luminometer (AB2500 Kronos, ATTO), according to a previous report, with slight modification 78. The luminometers were placed in an incubator under 5% CO₂ in air at 20°C. The cultures were incubated at 37°C in the luminometers and bioluminescences were monitored for 1 min at 10–20 min intervals in the absence or presence of a 620 nm long-pass filter (R62 filter, Hoya, Tokyo, Japan) 78. The bioluminescence count obtained was expressed in relative light units (RLU). Each green- and red-emitting luciferase activity was calculated from the total RLU (F0) and the RLU passed through the R62 filter (F1), as reported 7. Briefly, the light intensity of green-emitting luciferase (G) and red-emitting luciferase (R) were obtained from F0 and F1 and from the optical filter's transmission coefficients for the green- (κ_G) and the red-emitting (κ_R) luciferases using the following simultaneous equation.

( F 0 F 1 ) = ( 1.0 1.0 κ G κ R ) ( G R ) MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaWaaeWaaeaafaqabeGabaaabaGaemOrayKaeGimaadabaGaemOrayKaeGymaedaaaGaayjkaiaawMcaaiabg2da9maabmaabaqbaeqabiGaaaqaaiabigdaXiabc6caUiabicdaWaqaaiabigdaXiabc6caUiabicdaWaqaaiabeQ7aRnaaBaaaleaacqWGhbWraeqaaaGcbaGaeqOUdS2aaSbaaSqaaiabdkfasbqabaaaaaGccaGLOaGaayzkaaWaaeWaaeaafaqabeGabaaabaGaem4raCeabaGaemOuaifaaaGaayjkaiaawMcaaaaa@43D2@

To determine the periods, we analyzed the data over 4 d by subtracting a 24 h moving average from the crude data and obtained a best-fit cosine curve for the data using a least squares spectrum method 29. To determine the peak times and amplitude-damping rate, serial records of bioluminescence were smoothed using a 10-point (2–3 h) moving average 15. The circadian peaks and troughs were calculated as the highest and lowest points of the smoothed data. We determined peaks 1, 2, and 3 as the highest points that appeared 10.0–25.0 h, 40.0–55.0 h, and 60.0–80.0 h after the start of DEX treatment, respectively, and troughs 1, 2, and 3 as the lowest points following peaks 1, 2, and 3, respectively. The RLU difference between peak 1 and trough 1, and peak 2 and trough 2 were determined as amplitude 1 and amplitude 2, respectively. The ratio of amplitude 2 to amplitude 1 was determined as the amplitude-damping rate.