The circadian clock mutant lhy cca1 elf3 paces starch mobilization to dawn despite severely disrupted circadian clock function

Abstract

Many plants, including Arabidopsis (Arabidopsis thaliana), accumulate starch in the daytime and remobilize it to support maintenance and growth at night. Starch accumulation is increased when carbon is in short supply, for example, in short photoperiods. Mobilization is paced to exhaust starch around dawn, as anticipated by the circadian clock. This diel pattern of turnover is largely robust against loss of day, dawn, dusk, or evening clock components. Here, we investigated diel starch turnover in the triple circadian clock mutant lhy cca1 elf3, which lacks the LATE ELONGATED HYPOCOTYL and the CIRCADIAN CLOCK-ASSOCIATED1 (CCA1) dawn components and the EARLY FLOWERING3 (ELF3) evening components of the circadian clock. The diel oscillations of transcripts for the remaining clock components and related genes like REVEILLE and PHYTOCHROME-INTERACING FACTOR family members exhibited attenuated amplitudes and altered peak time, weakened dawn dominance, and decreased robustness against changes in the external light-dark cycle. The triple mutant was unable to increase starch accumulation in short photoperiods. However, it was still able to pace starch mobilization to around dawn in different photoperiods and growth irradiances and to around 24 h after the previous dawn in T17 and T28 cycles. The triple mutant was able to slow down starch mobilization after a sudden low-light day or a sudden early dusk, although in the latter case it did not fully compensate for the lengthened night. Overall, there was a slight trend to less linear mobilization of starch. Thus, starch mobilization can be paced rather robustly to dawn despite a major disruption of the transcriptional clock. It is proposed that temporal information can be delivered from clock components or a semi-autonomous oscillator.

Figure 1

Diel starch turnover in plants growing in different photoperiods and irradiance. A, Diel changes in starch content. Plants were grown in a 12-h light/12-h dark (left-hand) or 6-h light/18-h dark (right-hand) photoperiod and irradiance of 160 µmol m⁻² s⁻¹ (top), 90 µmol m⁻² s⁻¹ (middle or 60 µmol m⁻² s⁻¹ (bottom). Harvest age and sampling strategy was designed to take account of the differing growth rates and rosette size in the various conditions. Plants were harvested over a full diel cycle at 13 DAS in the 12-h light/12-h dark at 160 µmol m⁻² s⁻¹ irradiance, 14 DAS in the 12-h light/12-h dark 90 and 60 µmol m⁻² s⁻¹ treatments and 16 DAS in all 6-h light/18-h dark treatments (see Supplemental Figure S1A). At each time point, two to five samples were harvested. Background shading indicates light period (white) and night (gray). Wild-type Ws-2 and lhy cca1 elf3 are indicated by black and red symbols, respectively. The data for Ws-2 in a 12-h photoperiod at 160 µmol m⁻² s⁻¹ was collected in a separate experiment (Flis et al., 2015); otherwise all plants were grown in parallel (depicted as square and circular symbols, respectively). Symbols represent the mean value and error bars indicate the bootstrapped 95% CI. Statistical significance (Analysis of Variance [ANOVA], Sum of Squares type II) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05; subsequent Honestly Significant Difference [HSD] Tukey’s post-hoc test was significant in all cases). NA denotes that a test was not applicable due to lack of replicates. “ZT,” or “Zeitgeber” from the German language, indicates the time elapsed after the last dawn, in hours. B, Estimated absolute rates of starch accumulation. C, Estimated absolute rates of starch mobilization. D, Estimated relative rates of starch mobilization. Ws-2 and lhy cca1 elf3 are indicated by black and red, as in (A). Rates were defined as the slope of linear models, and error bars indicate the 95% CI of the standard error of the slope. Accumulation rates were calculated including all time points between dawn and dusk, and mobilization rates were calculated using time points between dusk and dawn. At each time point, two to five samples were harvested. Relative rates were calculated using starch levels as a proportion of the average starch levels at dusk in each condition. Statistical significance between wild-type Ws-2 and lhy cca1 elf3 (Analysis of Covarian [ANCOVA], Sum of Squares type III) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05). Numeric values are provided in Supplemental Data Set 1. E, Estimated time at which starch is exhausted. Symbols represent the projected time of StEx as defined by single extrapolation of the rate of starch mobilization (StEx). This was performed using time points between dusk and dawn. At each time point, two to five samples were harvested. Color indicates different genotypes, as in (A). Error bars indicate the standard error of the StEx estimation. Solid upward and downward triangles denote 90 and 160 µmol m⁻² s⁻¹ irradiance, respectively, in the 6 h/18 h and open squares, diamonds and circles denote 60, 90, and 160 µmol m⁻² s⁻¹ irradiance, respectively, in 12-h light/12-h dark photoperiod. Numeric values are provided in Supplemental Data Set 1. The time when starch would be exhausted as projected by nested linear fits is shown in Supplemental Figure S4B.

Figure 2

Diel starch turnover in non-T24 cycles. A, Diel changes in starch content. Plants were grown in 8.5-h light/8.5-h dark cycle (T17, left hand) or 14-h light/14-h dark cycle (T28, right hand) at 160 µmol m⁻².s⁻¹. Plants were harvested over a full diel cycle, 13 diel cycles after sowing. At each time point, two to five samples were harvested. Background shading indicates light period and night. Wild-type Ws-2, lhy cca1 elf3, lhy cca1, and elf3 are indicated by black, red, orange, and purple symbols, respectively (see insert). Symbols represent the mean and error bars indicate the bootstrapped 95% CI. Statistical significance (ANOVA, sum of squares type II) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05; subsequent HSD Tukey’s post-hoc test was significant in all cases) for each wild-type—mutant comparison and is denoted by the color code for that mutant. “ZT,” or “Zeitgeber” from the German language, indicates the time elapsed after the last dawn, in hours. B, Estimated absolute rates of starch mobilization. C, Estimated relative rates of starch mobilization. Plants grown in T17 and T28 conditions were compared to plants grown in 24-h diel cycle conditions (T24). Ws-2, lhy cca1 elf3, lhy cca1, and elf3 are indicated by different colored symbols, as in (A). Rates were defined as the slope of linear models, and error bars indicate the 95% CI of the standard error of the slope. These were performed excluding times beyond 8.5 h of darkness (highlighted in dotted box in Supplemental Figure S5A). This allowed comparison of T-cycles over a similar duration after dusk. Relative rates were calculated using starch levels as a proportion of the average starch levels at dusk in each condition. At each time point, two to five samples were harvested. Statistical significance between wild-type Ws-2 and lhy cca1 elf3 (ANCOVA, Sum of Squares type III) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05). Numeric values are provided in Supplemental Data Set 1. D, Estimated time at which starch is exhausted in T17, T24, and T28 cycles. Symbols represent the projected time of StEx as defined by the geometric mean of multiple projections (StEx^app). These were performed using linear models on time spans of varying lengths that started at dusk and extended for increasing lengths of time into the night (at least three time points) excluding times beyond 8.5 h of darkness (as in (B and C), highlighted in dotted box in Supplemental Figure S5A). This allowed comparison of T-cycles over a similar duration after dusk. Error bars indicate the 95% CI of the multiple estimations. At each time point, 2–5 samples were harvested. Solid and open symbols indicate results from 6-h/18-h to 12-h/12-h light–dark cycle conditions, respectively. Color indicates different genotypes, as in (A). Numeric values are provided in Supplemental Data Set 1.

Figure 3

Starch degradation after a single day of low irradiance. Wild-type Ws-2 (left hand, black symbols) and lhy cca1 elf3 (right hand, red symbols) were grown in 12-h light/12-h dark at 160 µmol m⁻² s⁻¹ for 13 days and then harvested at different times during the night following a day at growth irradiance (open symbols) or a day on which the irradiance was decreased to 90 µmol m⁻² s⁻¹ from dawn on (solid symbols). At each time point, two to five samples were harvested. A, Starch content. Wild-type Ws-2 (left hand, back symbols) and lhy cca1 elf3 (right hand, red symbols) were grown in 12-h light/12-h dark at 160 µmol m⁻² s⁻¹ for 13 days and then harvested at different times during the night following a day at growth irradiance (open symbols) or a day on which the irradiance was decreased to 90 µmol m⁻² s⁻¹ from dawn on (solid symbols). At each time point, two to five samples were harvested. Background shading indicates light period (white) and night (gray). Wild-type Ws-2 and lhy cca1 elf3 are indicated by black and red symbols, respectively. Symbols represent the mean values, and the error bars indicate bootstrapped 95% CI. Statistical significance (ANOVA, Sum of Squares type II) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05; subsequent HSD Tukey’s post-hoc test was significant in all cases). “ZT,” or “Zeitgeber” from the German language, indicates the time elapsed after the last dawn, in hours. B, Estimated absolute rates of starch mobilization. C, Estimated relative rates of starch mobilization. Ws-2 and lhy cca1 elf3 are indicated by black and red symbols, as in (A). Rates were defined as the slope of linear models, and error bars indicate the 95% CI of the standard error of the slope. Mobilization rates were calculated using time points between dusk and dawn. At each time point, two to five samples were harvested. Relative rates were calculated using starch levels as a proportion of the average starch levels at dusk in each condition. Statistical significance between wild-type Ws-2 and lhy cca1 elf3 (ANCOVA, Sum of Squares type III) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05). Numeric values are provided in Supplemental Data Set 1. D, Estimated time at which starch is exhausted. Symbols represent the projected time of StEx as defined by single extrapolation of the rate of starch mobilization (StEx). This was performed using time points between dusk and dawn. At each time point, two to five samples were harvested. Error bars indicate the standard error of the StEx estimation. Solid and open symbols indicate results from “Low light” and “Control” conditions, respectively. Color indicates different genotypes, as in (A). Numeric values are provided in Supplemental Data Set 1. The time when starch would be exhausted as projected by nested linear fits is shown in Supplemental Figure S6B).

Figure 4

Starch degradation after a sudden early dusk. The lhy cca1 elf3 triple mutant was grown in 12-h light/12-h dark at 160 µmol m⁻² s⁻¹ for 13 days and then harvested at different times during the night following a day in the growth regime (open symbols) or after a sudden early dusk at ZT8 (solid symbols; “ZT,” or “Zeitgeber” from the German language, indicates the time elapsed after the last dawn, in hours). At each time point, two to seven samples were harvested. Three further experiments are shown in Supplemental Figure S7. The data for Ws-2 in a 12-h photoperiod at 160 µmol m⁻² s⁻¹ was collected in a separate experiment (Flis et al., 2019). A, Starch content. Background shading indicates light period (white), the time during which some plants were in the light and others had been already darkened (pale gray), and the time when all plants were in darkness (gray). Symbols represent the mean values and the error bars indicate bootstrapped 95% CI. Statistical significance (ANOVA, Sum of Squares type II) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05); subsequent HSD Tukey’s post-hoc test was significant in all cases). B, Estimated absolute rates of starch mobilization. C, Estimated relative rates of starch mobilization. Ws-2 and lhy cca1 elf3 are indicated by black and red symbols, as in (A). Rates were defined as the slope of linear models, and error bars indicate the 95% CI of the standard error of the slope. Mobilization rates were calculated using time points between ZT12 and ZT24 in the control condition and from ZT8 to ZT18 in the early dusk treatment (solid bars). At each time point, 2–7 samples were harvested. Solid bars represent the treatment condition whilst the hollow bars indicate control conditions. Dashed line bars indicate results in control conditions but restricted from ZT16 to ZT24. Relative rates were calculated using starch levels as a proportion of the average starch levels at dusk in each condition. Statistical significance between wild-type Ws-2 and lhy cca1 elf3 (ANCOVA, Sum of Squares type III) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05). Numeric values are provided in Supplemental Data Set 1. D, Estimated time at which starch is exhausted. Symbols represent the projected time of StEx as defined by single extrapolation of the rate of starch mobilization (StEx). This was performed excluding times beyond 8.5 h of darkness, corresponding to ZT20 and ZT16 in the “Control” and the “Early Dusk” conditions, respectively. This allowed comparison over a similar duration after dusk. Error bars indicate the standard error of the StEx estimation. At each time point, two to seven samples were harvested. Solid and open symbols indicate results from “Early Dusk” and “Control” conditions, respectively. Color indicates different genotypes, as in (A). Numeric values are provided in Supplemental Data Set 1. The time when starch would be exhausted as projected by nested linear fits is shown in Supplemental Figure S7F). Numeric values are provided in Supplemental Data Set 1.

Figure 5

Analysis of the extent to which starch mobilization is nonlinear in stable diurnal cycles, and after a sudden low light day or a sudden early dusk. If starch mobilization is linear, the content at different times in the night will lie on a straight line between dusk starch content and dawn starch content. To score the deviation of starch mobilization from linearity, at each time point the starch content expected if mobilization were linear was subtracted from the measured starch content, the difference (residue) was normalized on the content expected if mobilization was linear, and the normalized residues were averaged across all time points to give an average relative error, or score for nonlinearity. This score is zero if breakdown is linear, and close to zero if noise leads to small random deviation from linearity. A consistent deviation from linearity will lead to a negative score if mobilization is initially too fast, and a positive score if mobilization is initially too slow. A, 12-h light/12-h dark with 160 or 90 µmol m⁻² s⁻¹ irradiance (data from Figure 1). B, 6-h light/18-h dark photoperiod with 160 or 90 µmol m⁻² s⁻¹ irradiance (data from Figure 1). C, In the night after a sudden drop in irradiance from growth conditions (12-h light 12-h dark with 160 µmol m⁻² s⁻¹ (control) to 90 µmol m⁻² s⁻¹ (sudden low light) (data from Figure 3). D, In the night after a sudden 4-h advance in dusk (sudden early dusk) compared to growth conditions (12-h light 12-h dark with 160 µmol m⁻² s⁻¹, control). Ws-2 data are from Flis et al. (2019). Data for lhy cca1 elf3 are from Figure 4 and Supplemental Figure S7, Experiments #27, #4, and #3 (13 DAS) are shown from left to right; the analysis was performed for Experiment #11 despite the low density of sampling (4 h instead of 2 h intervals; at 21 DAS). The plots show the mean and 95% CI (calculated using a nonparametric bootstrap procedure) of the nonlinearity score. At each time point, two to five samples were harvested in the sudden “low light” experiment and two to seven in the “sudden early dusk” experiment. Significant statistical differences (indicated by letters; at 95% CI level, P ≤ 0.5) between groups of results in each part was analyzed using ANOVA and subsequent HSD Tukey’s post-hoc test (significant in all cases).

Figure 6

Response of C-starvation reporter transcripts in 12-h photoperiod. The C starvation reporter transcripts DIN1, DIN6, and BCAT2 were measured in the 12-h photoperiod at 160 µmol m⁻² s⁻¹ irradiance experiment of Supplemental Figure S1A (the experiment used to measure starch and sugars in this condition, see Figure 1; Supplemental Figure S8). At each time point, two to five samples were harvested. Ws-2 data from Flis et al. (2019). Transcript abundance was measured by quantitative reverse transcription polymerase chain reaction (RT-qPCR), adding artificial RNA standard before cDNA amplification to allow absolute quantification. Abundance is given as log₂(copies × 2.5 × 10⁷ g⁻¹ FW). Background shading indicates the light period (white) or night (gray). Wild-type Ws-2 and lhy cca1 elf3 are shown as black and red symbols, respectively (see insert). The symbols give the mean value, and error bars indicate the bootstrapped 95% CI. Statistical significance (ANOVA, Sum of Squares type II) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05, subsequent HSD Tukey’s post-hoc test (significant in all cases). “ZT,” or “Zeitgeber” from the German language, indicates the time elapsed after the last dawn, in hours.

Figure 7

Summary of the peak time and amplitude of the diel oscillations of clock and GBSS1 transcript abundance in different photoperiods. The display shows the mean peak time (x-axis) and the amplitude of oscillation (y-axis) for the LHY, CCA1, PRR9, PRR7, PRR5, GI, TOC1, ELF4, LUX, and GBSS1 transcripts. Peak time was estimated as described in “Materials and methods.” The origin of the time axis is aligned to dawn and runs from ZT18 to ZT16 in order to capture all dawn-phased in the same part of the display (“ZT,” or “Zeitgeber” from the German language, indicates the time elapsed after the last dawn, in hours). Horizontal error bars indicate the bootstrapped 95% CI of the estimated peak time. The amplitude of oscillation (given as log₂(copies × 2.5 × 10⁷ g⁻¹ FW)), see methods for explanation of the unit) was estimated as the difference between maximum and minimum mean transcript level in a 24-h period. Vertical error bars indicate the 95% CI of the amplitude of oscillation. Wild-type Ws-2 and lhy cca1 are indicated by black and red open symbols, respectively. Gene names are indicated by numbers. Background shading indicates the light period (white) and night (gray). The numerical values of these analyses are provided in Supplemental Data Set 3.

Figure 8

Response of selected transcripts in different photoperiods. A, PRR9. B, PRR7. C, RVE8. D, GBBS1. The three subparts show responses in a 6-h light/18-h dark, 12-h light/12-h dark and 18-h light/6-h dark cycle. Transcript measurements were carried out in the same material that was used for the starch analyses in a 12-h photoperiod in Figure 1A and in separate experiments for the 6-h light/18-h dark and 18-h light/6-h dark cycles. At each time point, two to five samples were harvested. Symbols represent the mean value and error bars indicate the bootstrapped 95% CI. Wild-type Ws-2 and lhy cca1 are indicated by black and red symbols, respectively (see insert). Statistical significance (ANOVA, Sum of Squares type II) is indicated by asterisks (0 “***” 0.001 “**” 0.01 “*” 0.05); subsequent HSD Tukey’s post-hoc test is indicated by dashes (i.e. when not significantly different). “ZT,” or “Zeitgeber” from the German language, indicates the time elapsed after the last dawn, in hours. Plots for further transcripts are provided in Supplemental Figure S9. Statistical analyses of the extent of dawn alignment and dusk alignment are provided in Figure 9 and Supplemental Figure S10. Abundance is given as log₂(copies × 2.5 × 10⁷/g FW).

Figure 9

Mutual information analysis of the extent of alignment of transcript responses to dawn or dusk in 6/18-h, 12/12-h, and 18/6-h light/dark cycles. The analyses for Ws-2 are shown in the left hand and for lhy cca1 elf3 in the right-hand block of plots. For each genotype, transcript and photoperiod the transcript time series data were organized as “dawn aligned” (i.e. profiles starting at light on) and as “dusk aligned” (i.e. profiles starting at light off; plots of the dawn- and dusk-aligned data series are provided in Supplemental Figure S9). The DIST score is defined as the Euclidean Distance between all treatments. The NORM score is defined as the Euclidean Distance divided by its propagated error (see Supplemental Data Set 4 for details). Heatmap colors indicate the average value of each score for each transcript; as explained in Supplemental Data Set 4. DIST and NORM scores were independently transformed to values between 0 and 1, with 0 indicating the best and 1 the worst alignment observed. ANOVA (Sum of Squares type II) and subsequent Tukey’s HSD was used to detect which transcripts were not significantly different between “dawn” and “dusk”; these are indicated as gray (at 95% CI level, P ≤ 0.5).