The circadian clock modulates anti-cancer properties of curcumin
Circadian rhythms persist at low curcumin concentrations
The C6 rat glioma cell line was selected to test for an effect of the circadian clock
on curcumin’s anti-cancer properties because it displays circadian rhythms in expression
of the core circadian clock gene mPer2 in cell cultures 40] and in tumorsphere cultures 39]. Curcumin decreases NF-kB activation, inhibits C6 cell proliferation, and induces
cell death. We treated C6 cells with a low dose of curcumin (CUR) to cause limited
but significant cell death over several days while allowing enough cells to remain
for measurements 40]. Also, because curcumin acts on many intracellular signaling pathways 41] it was important to use a dosage that would not suppress the molecular mechanism
of the circadian oscillator.
We used C6 cells stably transfected with a reporter gene that generates a fusion protein
of mPER2 and firefly luciferase under control by the mPer2 gene promoter 39]. Circadian clock cells in cultures of these mPer2::mPer2:luc C6 cells were synchronized with a 2-h forskolin treatment. Starting 12 h later, to
allow acute forskolin effects to subside, medium was exchanged with medium containing
curcumin dissolved in 0.02 or 0.04 % DMSO for 5 and 10 ?M curcumin, respectively.
The DMSO concentration was well below the threshold of 10 % for toxic effects on colon
cancer cells 42], and the threshold of 0.1 %, below which solutions can be injected safely into the
vitreous of the rat eye without affecting retinal neurons 43]. This pulse of curcumin was intended to mimic a single delivery of the drug to a
cancer patient either intravenously or intracerebrally. Similar to what occurs in
the body, it was expected to be degraded in vitro over the next few hours based on
the known properties of curcumin 44], 45].
C6 cell cultures in medium with 10 % FBS are reported to have a 23.5-h circadian rhythm
when measured with a destabilized luciferase reporter gene controlled by the mPer2 promoter 40]. Similarly, in a previous study we detected a 25.2-h bioluminescence rhythm in a
C6 culture expressing the mPER2-LUC fusion protein, according to Lomb-Scargle Periodogram
(LS) analysis (p??0.001) 46]. The mPer2::mPer2:luc C6 cells also expressed circadian rhythms in cultures given 5 ?M CUR (Fig. 1a). For cultures treated with 5 ?M CUR the average period was estimated as 24.48 h
by LS (p??0.001) (Fig. 1b) and 24.47 h by FFT.
Fig. 1. Circadian rhythms in mPer2 clock gene expression persist after treatment with 5 ?M curcumin. (a) Signal from the mPer2::mPer2:luc reporter gene after a forskolin pulse used to synchronize cellular circadian oscillators
is shown as the hourly integrated light signal (relative light units averaged from
4 independent cultures). (b) A significant period of 24.48 h was detected by Lomb-Scargle Periodogram analysis
(p??0.001) is shown in the frequency spectrum
Long-term effects of curcumin on mitosis and cell death
Because circadian rhythms persisted after low-dose curcumin treatments, we synchronized
the circadian clock cells in C6 cultures and examined the pattern of individual mitotic
and apoptotic events for any effects from curcumin or the circadian clock. To determine
whether the 5 ?M curcumin treatment was sufficient to produce anticancer effects on
C6, and whether a higher dose would be more useful for this study, C6 cells were monitored
continuously by digital video imaging of cell cultures. To identify ongoing cell division
and cell death events in cultures time-lapse imaging (TLI) was performed using 5-min
intervals between frames for up to 5 days, after synchronizing cells with forskolin
and then treating with curcumin 12 h later. All events were counted from a single
field-of-view which represented the cell events occurring in the dish (Additional
file 2). The culture dish and microscope remained for days in a sealed incubator without
disturbance. During TLI, there was initially an average of 23.57?±?7.03 (SD) cells
in the field-of-view (range 18 to 30, n?=?15 cultures).
Distinct mitotic and apoptotic events were visible and counted following exposure
to 0, 5, and 10 ?M CUR (Fig. 2). When mitotic events during day 1 of imaging were compared, 10 ?M was significantly
more effective at suppressing cell division than 5 ?M (ANOVA, F?=?4.537, Fisher post hoc test, p?=?0.0148) and the control (p?=?0.0216) (Fig. 2a). The 5 ?M group was, however, not significantly different from the control (p?=?0.806). When the total mitotic events for the first four days were examined, 10 ?M
again resulted in a significant suppression of mitosis (p?=?0.0485) relative to control (0 ?M: 62.67?±?36.439, n?=?6; 5 ?M: 60.00?±?30.470, n?=?6; 10 ?M: 15.00?±?0.358, n?=?3). By day 4 (5
th
day in culture), cell confluence in the control dishes limited our ability to detect
cell division events, so they were not counted.
Fig. 2. Effects of curcumin on mitosis and apoptosis of C6 glioma cells. Suppression of mitotic
rate (a) and induction of cell death rate (b) at two curcumin concentrations. Average hourly rates were imaged in single fields-of-view
for 5 days after a forskolin pulse and a single curcumin treatment
When the apoptotic events occurring in the 0, 5, and 10 ?M groups during day 1 were
compared, 10 ?M produced significant cell death (F?=?18.751, p??0.001), but not 5 ?M (Fig. 2b). However, when total apoptotic events over days 1–4 were compared (F?=?6.398, p?=?0.0128) the 5 ?M treatment caused significant cell death (p?=?0.00384). Cell death rates were overall lower in 10 ?M-treated cells, and more
cell death occurred in the first day.
Circadian modulation of curcumin effects on mitosis and cell survival
To determine whether ongoing events of cell division and cell death in cultures exposed
to curcumin are modulated by circadian timing, we measured the period and phase of
any significant circadian rhythms (Fig. 3). Because individual events within the field-of-view were few the data were pooled
from all cultures in each treatment group. According to LS analysis, a significant
circadian rhythm in mitosis was detected in the control culture (Table 1). Most rhythms were similarly identified by FFT.
Fig. 3. Circadian clock regulation of curcumin efficacy. a–c: Mitotic events in C6 cultures showed circadian rhythms in 0 and 10 ?M but not 5 ?M
CUR. d–f: Cell death rates (apoptotic events/hr) displayed circadian rhythms in 0 and 5 but
not 10 ?M CUR (blue circles: events, red line: after adjacent averaging). g–i: Apoptotic and mitotic rates were inversely correlated as shown by linear regressions.
j, k: The phase of cell death events displayed significant clustering (p??0.001) in 5 ?M CUR on days 2 and 3 (left and right, Z?=?14.62 and 7.399 by Rayleigh
Test, respectively). Long arrow indicates phase of mean vector. Short arrow (green)
indicates peak of mPER2::LUC rhythm from Fig. 1. The curcumin treatment began at 12:00 AM (0:00) on Day 1. Each dot indicates 2 events
Table 1. Period analysis of C6 cell cultures
To provide a more precise estimate of the circadian period than what LS or FFT can
provide we used the Maximum Entropy Method to find periods with greater resolution.
According to MEM the untreated cultures displayed an average period of 21.3 h for
mitotic events. This rhythm in cell division was similar to the doubling time of 22 h
reported for C6 cells 47], suggesting that the forskolin treatment may have synchronized individual cell cycles.
The 10 ?M CUR-treated cultures displayed a 20.5-h period. Thus, circadian rhythms
were observed in the mitotic events of the control and 10 ?M cultures (Fig. 3a, c), but not in the 5 ?M group (Fig. 3b). In the presence of 5 ?M CUR the cell division cycles and circadian rhythms appeared
to be uncoupled. Mitosis displayed a rhythm of about 15 h, and these shorter ultradian
rhythms (defined here as having periods less than 18 h) may have resulted from curcumin
acting on cell cycle oscillations. To further test the periods of these cultures,
we analyzed the mitotic events with FFT, which yielded periods of 21.3 h for the control
group, 15.0 h for 5 ?M, and 18.3 h for the 10 ?M group.
Apoptotic events occurring in the cell cultures were analyzed to detect any circadian
rhythms. A rhythm with a period of 18.9 h was detected in the untreated group (Fig. 3d), which is at the edge of a typical circadian range of 19–30 h (Table 1). The apoptotic events occurring in 5 ?M CUR-treated cells followed a circadian rhythm
with a period of 22.3 h (Fig. 3e). All the period estimates for this treatment group fell in the circadian range (Table 1). Circadian rhythms were absent in 10 ?M CUR-treated cells (Fig. 3f), which instead had an ultradian rhythm of 11–14 h (Table 1). Circadian rhythms in apoptosis persisted in the 5 but not 10 ?M cultures, indicating
that the clock can modulate cell death at the lower curcumin dosage.
Because mitotic and apoptotic rates appeared to reach their maxima at different times,
we examined this relationship in the untreated, 5, and 10 ?M groups. The apparent
inverse relationship between apoptotic and mitotic events was confirmed by a linear
correlation test comparing the two time-series data sets: Pearson’s correlation r values were ?0.350, ?0.599, and ?0.437 for the 0, 5, and 10 ?M treatments, respectively
(Fig. 3g-i). The mitotic rate was highest in the untreated group as apoptotic events were fewer
in that group, while the 10 ?M group had lower mitotic rates and higher rates of apoptosis
during the initial days of treatment.
Along with the period analyses, we also examined the phase relationships between the
mPER2 rhythm shown in Fig. 1 and the mitotic and apoptotic rhythms. We applied circular statistics to identify
significant clustering of apoptotic events in curcumin-treated groups over the first
three 24-h cycles of imaging. When examining the timing of these events relative to
the forskolin treatment, significant clustering was observed at 18.3 and 18.6 h during
the 2
nd
and 3
rd
days of imaging, respectively, in the 5 ?M group (Fig. 3j, k). When comparing these phases with the mPER2 rhythm, they occurred on the rising
phase, about 6 and 11 h before the corresponding circadian peaks in mPER2 protein
expression.
There was no significant clustering of cell death events in the 10 ?M group, in agreement
with the loss of circadian periodicity of apoptosis. The phase of cell deaths in the
control group (0 ?M) was significantly clustered (p??0.05) on the second day, but the mean vector was not significantly different from
that of the 5 ?M group, indicating that curcumin did not produce a measurable phase
shift of the rhythm (control: 16:24 with a 99 % confidence interval of 12:31 and 20:17 h
on days 2 and 3, respectively; 5 ?M: 18:18 with 99 % confidence intervals of 16:34
and 20:03 h on days 2 and 3). There was no significant clustering of apoptotic or
mitotic events in the remaining groups.
As an additional test of whether cell death events vary according to the circadian
cycle, we quantified the percentage of cells expressing activated caspase-3, a late
apoptotic marker 48], 49] in C6 cells given 5 ?M CUR. Three times were selected to coincide with the second
peak, the following trough, and the third peak observed in the rhythm in apoptosis
(Fig. 3e). The three phases examined showed relative differences in cell death matching the
oscillations in apoptotic events in the time series (Fig. 4a). The percentage of apoptotic cells was 61.15?±?0.03 % at the 45
th
hr and 47.62?±?0.04 % at the 69
th
hr, which are both peak phases in the circadian rhythm of death rate. During the trough
phase (57
th
hr) the percentage of apoptotic cells declined to 23.50?±?0.02 %. Visibly, there were
more cells stained with anti-caspase3 antibody at the peak phases (Fig. 4b, I and II) than at a trough phase (Fig. 4b, III and IV).
Fig. 4. Expression of activated caspase-3 according to phase of the circadian cycle. a: Percentage of apoptotic cell counts at 45, 57, and 69 h after adding 5 ?M CUR. The
relative changes in the percent cell death agreed with the peaks and trough in the
circadian rhythm of death rate for cells treated with 5 ?M CUR (red line, from Fig. 3e, shown here for comparison). b: Immunostaining cleaved caspase-3 in apoptotic cells (green) at two circadian phases,
the peak at 45 h (i) and the trough at 57 h (iii), after 5 ?M CUR treatment; ii and
iv: The same cells merged with Hoechst nuclear staining (red). Scale bar?=?10 ?m
Stability and localization of curcuminoids and their metabolites
Although curcumin is being tested as an anticancer drug in a number of clinical trials
5], 50], its use is limited because of fast degradation at neutral and alkaline pH and poor
tissue absorption 51], 52]. Studies have shown that curcumin is relatively more stable in culture media containing
10 % fetal bovine serum (FBS), compared to phosphate buffer or culture media without
FBS 53]. Despite the expected loss of curcumin, our TLI data showed that apoptosis continued
for several days after initial treatment with 5 or 10 ?M CUR. Using a spectrophotometer
we found evidence of curcumin in culture media during the days the cells were imaged
(Fig. 5). A standard curve was created at curcumin’s maximal absorbance near 430 nm 54]. Nevertheless, curcumin levels declined within the first day, becoming nearly undetectable
(Fig. 5a). During the first 24 h, the curcumin declined with a half-life of about 1.7 h by
degrading or entering cells (Fig. 5a inset). The curcumin levels in media also showed an unexpected small increase after
the second day.
Fig. 5. Stability of the 3 curcuminoids in cell culture. Curcumin degraded with a half-life
of about 1.7 h in cell culture medium with cells present. a: Curcumin absorbance in culture medium with C6 cells for 4 days measured with a spectrophotometer.
Inset: Absorbance during the first 24 h. b: 5 ?M CUR produced fluorescence in live C6 cells within 1 h after its introduction.
Scale bar?=?5 ?m. c: Curcuminoid fluorescence persisted in cells for at least 24 h (same scale as B).
d: A representative HPLC chromatogram showing complete baseline separation of the 3
curcuminoids from cell culture medium containing 0.3 % DMSO. The first group of merged
peaks near 1.6 min represents the chromophoric compounds from the media. The second
peak at 6.45 min is BDMC, the peak at 7.04 min is that of DMC, and the last peak at
7.68 min is that of curcumin. (mAU?=?milliabsorbance units). This is the same CUR
preparation used for treating the cells. e: HPLC measurements show curcuminoids DMC and BDMC persist longer than curcumin in
cell culture medium and degrade even slower in DMSO. f: HPLC measurements of curcuminoids in culture medium with C6 cells present for 5 days.
Inset: The same data normalized to initial levels. Line colors are as in e.
Autofluorescence imaging of curcumin’s cellular distribution in live cells showed
that curcumin is present in the nucleus one hour after application (Fig. 5b). Although curcumin levels decreased considerably in the culture media, curcuminoids
were visible in C6 cells for at least 24 h after application (Fig. 5c). The autofluorescence study was performed in live C6 cells 1 h and 24 h after curcumin
treatment. To confirm the nuclear localization of curcumin 24 h after curcumin treatment,
cells were fixed and stained with Hoechst nuclear stain (Additional file 3). It was observed that curcumin was present in the nucleus and was concentrated in
distinct intra-nuclear sites, as described previously 55].
The curcumin treatment used in the study contained the additional congeners DMC and
BDMC (Fig. 5d), which may have contributed to the apoptotic or mitotic effects. The relative amount
of curcumin present in the CUR treatment was similar to that reported by the manufacturer
(?65 %). To better understand the potential contributions of the congeners to the
cell effects we used HPLC to measure curcumin, DMC, and BDMC concentrations in cell
culture medium at 0, 2, 4, 6 and 12 h after treatment. Samples containing curcuminoids
in complete medium (10 % FBS serum and 0.3 % DMSO) were preserved at ?20 °C in the
dark. To prevent further degradation each sample was allowed to thaw before 10 ?l
was injected without further purification or delay.
As predicted from previous studies 56], the HPLC results indicated that curcumin degraded by over 75 % within 12 h in culture
medium at room temperature, but the two congeners degraded more slowly (Fig. 5e), around 20 % for DMC and only 8 % for BDMC, suggesting that they could have been
responsible for cell death along with curcumin after the first day of treatment. The
degradation rates of the congeners in DMSO were relatively slower than in culture
media (Fig. 5e). Degradation patterns of the three curcuminoids were also measured in samples of
medium from cultures containing C6 cells once per day for 4 days after the curcumin
treatment (Fig. 5f). All conditions of these samples were kept similar to those of the previous TLI
cultures. There was a rapid decline of curcumin in culture medium during the first
24 h at 37 °C (to about 15 %). DMC decreased to about 16 % in the first 24 h of treatment,
whereas BDMC declined to about 28 % (Fig. 5f inset).
Although the culture medium had very little curcumin or congener present, the cells
retained curcuminoids, as shown in Figs. 5b and c, which could have been responsible for apoptosis and other cellular effects. To determine
whether the medium retains an anti-cancer property after curcumin levels decline,
we examined C6 cells treated with a conditioned medium (CM) that was withdrawn from
a C6 culture one day after treatment with 5 ?M curcumin (Additional file 4). Significant cell death or mitotic arrest was not observed in response to CM treatment.