Nyctinastic thallus movement in the liverwort Marchantia polymorpha is regulated by a circadian clock

The circadian clock coordinates an organism’s growth, development and physiology with environmental factors. One illuminating example is the rhythmic growth of hypocotyls and cotyledons in Arabidopsis thaliana. Such daily oscillations in leaf position are often referred to as sleep movements or nyctinasty. Here, we report that plantlets of the liverwort Marchantia polymorpha show analogous rhythmic movements of thallus lobes, and that the circadian clock controls this rhythm, with auxin a likely meditator. The mechanisms of this circadian clock are partly conserved as compared to angiosperms, with homologs to the core clock genes PRR, RVE and TOC1 forming a core transcriptional feedback loop also in M. polymorpha.


Results
the Marchantia polymorpha circadian clock controls nyctinastic thallus movements. In Arabidopsis, growth rates of both hypocotyls and leaves are rhythmic and under the control of a circadian clock 30,31 . In our attempts to detect and measure similar rhythmic growth patterns in liverworts, we noticed that young M. polymorpha gemmalings display nyctinastic movements as the lobes of young thalli waves up and down with a 24 hour rhythm in conditions of 12 hour light and 12 hour darkness (neutral days (ND); Fig. 1a,b; Supplementary Video S1). Furthermore, in gemmalings of different accessions these rhythmic movements are Wild type gemmaling in through (a) and peak (b) positions, respectively. Gemmalings were grown in a 12 hour light, 12 hour dark photoperiod (neutral days, ND) for five days, after which light was switched to continuous (LL) and imaging was started. (c) Period and amplitude for wild type (Upp10 to 14) at different temperatures. (d-g) Vertical apex region position plotted against time for wild type (Tak1, black) and Mpprr ko (gray) (d), wild type (Tak1, black) and Mptoc1 ko (gray) (e), wild type (Upp5, black) and Mplux ge -19 (gray) (f), wild type (Upp12) and Mplux ge -9 (gray) (g). Vertical position data of apex region were de-trended using a cubic smoothing spline with 12 degrees of freedom. Data are means ± SE of ten replicate gemmalings. maintained in LL (continuous light) conditions with an approximate average period of 26.1-26.5 hours for several days, supporting that they are controlled by a circadian clock (Fig. 1c-e). One key characteristic of circadian rhythms is temperature compensation, i.e., that the free-running period does not change much with ambient temperature. We thus estimated the free-running period of nyctinastic thallus movement at different temperatures. Consistent with circadian regulation, we found no significant difference in period for temperatures ranging from 18 to 24 °C (Fig. 1c).
To further investigate the role of a circadian clock for this movement, we first obtained a more detailed view on the role of MpPRR, MpRVE and MpTOC1 as core circadian clock components. Because transcriptional feedback loops are crucial for angiosperm circadian clocks, we examined temporal expression patterns of these genes using qRT-PCR over a 48 hour period. As previously reported 29 , MpPRR display rhythmic expression in the wild type in LL conditions (Fig. 2a). In Mptoc1 ko mutants, expression of MpPRR was continuously high and arrhythmic, as indicated by highly significant effects of both genotype (G) and genotype x time interaction (GxT) terms in ANOVA (P < 10 −11 ), suggesting that MpTOC1 represses MpPRR. Conversely, expression of MpPRR was low with limited amplitude in Mprve ko (P-values for both G and GxT terms were <10 −8 ), indicating that MpRVE promote the expression of MpPRR (Fig. 2a). The rhythmic expression of MpTOC1 in LL is weak, hampering interpretations of changes in its expression in mutant genotypes. Still, the expression of MpTOC1 in wild type, Mpprr ko and Mprve ko is consistent with an activating role of MpRVE and a repressing role of MpPRR analogous to the role of the Arabidopsis homologs on TOC1 expression (Fig. 2b; both G (P < 10 −4 ) and GxT (P = 0.026) were significant for Mpprr ko and G was significant for Mprve ko with P = 0.017). Furthermore, in Mpprr ko and Mptoc1 ko , expression of MpRVE remains high and arrhythmic indicating that MpRVE is repressed by both MpPRR and MpTOC1 ( Fig. 2c; P-values for G were less than 10 −9 in both cases, while GxT was marginally significant with P = 0.04 and 0.07 for Mpprr ko and Mptoc1 ko , respectively). These data collectively suggest that MpPRR, MpRVE and MpTOC1 are part of a core transcriptional feedback loop of the M. polymorpha circadian clock, and that knock-out mutants of these genes can be used to study the role of the circadian clock in the control of growth and development.
We then analyzed rhythm movements during constant conditions in knockout mutants of MpPRR and MpTOC1 that we identified as part of core feedback loops in the M. polymorpha circadian clock 29 (Fig. 2). In Mpprr ko and Mptoc1 ko mutants the rhythm is completely lost in LL (Fig. 1d,e). Because mutations of EC members result in an arrhythmic Arabidopsis clock we also created two independent mutant alleles for MpLUX, the single M. polymorpha homolog to the Arabidopsis EC component LUX ( Supplementary Fig. S1). The two mutants, Mplux ge -9 and Mplux ge -19, also showed the arrhythmic phenotype in LL observed in the other clock mutants (Fig. 1f,g). These results strongly support that the circadian clock controls "gemmaling waving", and that estimates of this movement can be used to monitor the M. polymorpha circadian clock.
The circadian clock regulates IAA levels and expression of the auxin biosynthesis gene MpTAA. Most likely, the rhythmic movement is growth related and similar to cotyledon movement in Arabidopsis, suggesting a role for rhythmic auxin production, transport and/or signaling in its control. Available data suggest that most auxin in liverworts such as Lunularia cruciata and M. polymorpha is produced in the apical region and transported basipetally through the midrib region producing an auxin gradient [32][33][34][35] . To assay temporal auxin biosynthesis patterns, we analysed gene expression of two genes coding for key enzymes in auxin biosynthesis, MpTAA and MpYUC2 35,36 . In wild-type plants a clear circadian expression pattern was observed for MpTAA (Fig. 3a). This pattern was also strongly affected in Mptoc1 ko , Mpprr ko and Mprve ko mutants (Fig. 3a). In Mpprr ko and Mptoc1 ko , expression was reduced and the rhythm dampened, while a higher expression with dampened amplitude was observed for Mprve ko . Highly significant effects of genotype (G) was obtained in ANOVA for all cases (P < 0.003), while P-values for GxT terms were 0.024, 0.009 and 0.053 for Mpprr ko , Mprve ko and Mptoc1 ko , respectively. For MpYUC2 expression in the wild type we could not detect a rhythmic pattern in LL, and only Mprve ko showed a slightly higher overall expression level with P = 0.006 for G term (Fig. 3b).
Because MpTAA share expression domain in the apical part of the thallus with MpPRR, MpTOC1 and MpRVE, as well as with other putative clock genes such as MpLUX and MpELF3 29,35 , it seems plausible that the clock could regulate IAA levels at least in part via MpTAA. To test if IAA levels display circadian oscillations we measured free IAA in whole young gemmalings of wild type (Upp14) and the Mplux ge -9 knock-out mutant of MpLUX ( Fig. 3c) 29 . To measure IAA levels, gemmalings were entrained in ND for three days after which material was sampled every four hours during 32 hours (nine time points), starting four hours after switch to LL. Statistical tests with JTK_CYCLE clearly indicate that IAA levels in the wild type are circadian (P = 1.8 × 10 −7 ), while IAA-levels in the Mplux ge -9 mutant are not (P = 0.12) (Fig. 3c). In Arabidopsis, LUX is part of the EC which acts as a repressor of e.g., PRR7/9 as well as TOC1 [21][22][23] . To verify that MpLUX regulates MpPRR we analysed MpPRR transcript levels in the Mplux ge -9 and Mplux ge -19 mutants using qRT-PCR. We found a significantly higher expression of MpPRR and also a strongly dampened rhythm in the Mplux ge mutants compared to wild type (Fig. 3d; both G and GxT terms in ANOVA have P < 10 −12 ), supporting a role for the M. polymorpha LUX gene in the control of the transcription of circadian clock genes.
Auxin could be a mediator of circadian control of thallus movements. To further evaluate a role for auxin in nyctinasty, we conducted nyctinastic movement experiments manipulating auxin levels, distribution and response. For chemical treatments, wild-type gemmae were first grown on standard growth medium for five days to allow for initiation of growth and dorsiventral polarity establishment. Actively growing gemmalings www.nature.com/scientificreports www.nature.com/scientificreports/ were subsequently transferred to media supplemented with the auxin transport inhibitor 2-[4-(diethylamino)]-2-hydroxybenzoyl benzoic acid (BUM) 37 , or low concentrations of indole 3-acetic acid (IAA).
Low doses of IAA (10 and 100 nM) resulted in a reduced angle of growth and thus a more flattened thallus (Fig. 4a,c). Rhythmic waving was detectable throughout the experiment on mock and 10 nM IAA, but dampened earlier on 100 nM IAA (Supplementary Videos S2-S4). We cannot exclude that this dampening is due to contact with the solid medium (Supplementary Video S4). This dampening hampered statistical analyses of rhythmicity resulting in a very low sample size of rhythmic plants at 100 nM IAA (n = 3). Still, analyses with Spectrum  39 . Experiments (e,f) were performed using younger gemmalings in (f), which is why the controls are not identical. Graphs (e,f) shows means ± SE of three replicates. **P < 0.01, ***P < 0.001 for two-tailed t-test against control (wild type or mock). (2020) 10:8658 | https://doi.org/10.1038/s41598-020-65372-8 www.nature.com/scientificreports www.nature.com/scientificreports/ Resampling 38 indicated an increased period and RAE with increasing IAA doses (P < 0.001 and P < 0.01, respectively).
Conversely, an increased growth angle was observed for growth on BUM in a dose-dependent manner (Fig. 4b,d). The apparent early dampening of rhythmic waving on BUM could be the result of contact between the two lobes due to the high growth angle. Statistical analysis indicated an increased period and RAE but reduced amplitude with increased BUM doses (P < 0.05, P < 0.05 and P < 0.01, respectively).
Even though these chemical treatment data does not show very convincing changes in rhythmicity, an important conclusion is that IAA and BUM have clear opposing effects on growth angle.
To study the effect of reduced auxin levels, we analysed waving in MpSHI pro :iaaL plants. These lines express the bacterial auxin-conjugating enzyme IaaL from a promoter mainly active in the apical notch, that result in plants displaying e.g., slow growth and narrow thalli 39 . No difference in period of the rhythmic waving, as compared to wild type, was observed for MpSHI pro :iaaL plants (P > 0.75), but the analysed lines displayed a significantly lower amplitude (P < 9.2 × 10 −10 ; Fig. 4e). Growth on high concentrations of the TAA inhibitor L-Kynurenine (L-Kyn; 250 µM) gave a similar result: a significantly reduced amplitude (P < 1.4 × 10 −4 ; Fig. 4f) but no significant change in period (P > 0.09). Similarly, a reduced amplitude but similar period was observed for EF1 pro :amiR-MpARF1 Mpmir160 plants 39 (P > 0.96 and P < 2 × 10 −16 , for period and amplitude, respectively; Fig. 4e), suggesting that reduced auxin sensitivity also attenuate rhythmic waving. Collectively these results support a role for auxin as an output pathway for the M. polymorpha circadian clock.

Discussion
In angiosperms the circadian clock regulates a wide range of processes, including those affecting metabolism, growth, abiotic and biotic stress, and various photoperiodic responses ( 40 and references therein). Our previous work suggests an early acquisition of a complex circadian network in plant evolution, but also important differences in the wiring and function of the M. polymorpha circadian clock 29 .
One unexpected handle of the M. polymorpha circadian clock is the rhythmic waving of gemmaling thallus lobes. This movement is most likely related to the non-reversible alternating growth of adaxial and abaxial sides of e.g., cotyledons of Arabidopsis 4 . It has been suggested that this type of nyctinastic movement was ancestral, and that motor cells and pulvini developed later as a means of enhancing leaf movement 41 . If so, our results suggest an early acquisition of nyctinasty already in the ancestor of bryophytes and seed plants. The cause of this type of alternating growth is poorly known, but may be related to the circadian elongation response of hypocotyls 42 . This rhythmic elongation involves circadian regulated hormone production, transport and/or signaling 43 . Auxin is a good candidate for such a hormone in Arabidopsis and also in M. polymorpha. Early studies showed that plants respond differently to applied auxin depending on time of day 44 , and subsequent experiments revealed that auxin levels oscillate in several species 43,45,46 . Furthermore, the circadian clock controls many genes that affect auxin synthesis and signaling in Arabidopsis [47][48][49] .
Our gene expression data suggests that the circadian clock in M. polymorpha regulates the first step of auxin biosynthesis, which might contribute to rhythmic auxin levels primarily in the apical region where MpTAA is expressed and most auxin in M. polymorpha is found 35 . If auxin levels were to control alternating growth of dorsal and ventral sides we would need to hypothesize a rhythm in dorsal/ventral distribution of the apically produced hormone during the 24-hour cycle. Such a role for auxin in these rhythmic movements is consistent with the results of our manipulations of auxin levels, distribution and response. Application of IAA and BUM had opposite effects on the angle of growth that in turn is directly connected to nyctinastic movements. Assuming that BUM affects ABCB-mediated auxin transport as in Arabidopsis, the increased angle on gemmalings growing on BUM can be interpreted as a result of decreased auxin transport from the ventral to the dorsal side, leading to increased ventral auxin concentration and cell elongation. This hypothesis requires an initial uneven dorsal/ ventral auxin distribution, perhaps through a basipetal transport of auxin mainly on the dorsal side. Under this scenario, addition of a significant portion of exogenous auxin (IAA) is expected to result in a more even dorsal/ ventral distribution and hence more flat growth (reduced growth angle in our experiments). This is supported by the epinastic growth of gemmalings on high concentrations of auxin or overexpression of the auxin synthesis enzyme MpYUC2 35,39 . Conversely, L-Kyn-treatment, overexpression of IaaL and the use of amiRNA's to knock down the expression of auxin biosynthesis genes, resulted in hyponastic growth of gemmalings 35,39 . In our experiments, directly decreasing the amount of active auxin by expressing IaaL or by reducing the function of MpTAA by adding L-Kyn resulted in lowered amplitude of nyctinastic movement, further supporting the role for auxin in these movements.
Our present results support conservation of the function for the M. polymorpha homologs of PRR, TOC1 and RVE, each with only one copy in M. polymorpha (MpPRR, MpRVE and MpTOC1). Each of them seems to be crucial for maintaining a transcriptional feedback loop in constant light conditions. For MpPRR and MpTOC1 we also observed abolished circadian nyctinastic thallus movement, verifying the importance of these genes in generating circadian rhythms in M. polymorpha. The stronger effects of mutating these genes in M. polymorpha, as compared to Arabidopsis, is likely due to the lack of functionally related paralogs in M. polymorpha.
Work on green algae suggest that one homolog of the PRR/TOC1 clade and one of the CCA/LHY/RVE clade constituted the core transcriptional feedback loop in the earliest plants 50,51 . Our work thus supports a continuous use of pairs of CCA/LHY/RVE clade genes and PRR/TOC1 genes at the core of plant circadian clock networks. However, the exact function of these genes within the network seems to have varied over time, partly due to addition of copies of existing genes or new genes to the network, or even deletion of core clock genes.
The only homolog in the whole CCA1/LHY/RVE clade present in M. polymorpha, MpRVE, belongs to the LCL sub-clade 29 , as do RVE4, RVE6 and RVE8. Accordingly, the MpRVE gene does not show an early morning expression, nor acute light induction, which is typical of genes in the CCA1/LHY sub-clade 29 . In addition, our data support a role for MpRVE as transcriptional activator as opposed to the role of CCA1/LHY genes as repressors 11,13 . (2020) 10:8658 | https://doi.org/10.1038/s41598-020-65372-8 www.nature.com/scientificreports www.nature.com/scientificreports/ Thus, MpRVE seems to have retained a function typical for the LCL subfamily 29 , despite the loss of the CCA1/LHY gene that is absent in all liverworts.
Our identification of a circadian regulated thallus movement provides a practical and easy to use tool for further studies of the evolution of plant circadian clocks, including the effects of frequent gene duplication and circadian gene loss observed during land plant evolution 29,52 . Gene expression analysis. RNA was extracted using an Rneasy Plant Mini Kit (Qiagen). cDNA was synthesized using SuperScript III Reverse Transcriptase (Thermo Fisher) and analysed by qRT-PCR as previously described 29 . Primers are listed in Supplementary Table S1. MpEF1α, MpACT and MpAPT3 were used for normalization 54 .

Methods
For sampling of RNA we used biological replicates (plants of individual transgenic lines or individual wild type lines), and technical replicates (pools of individually grown plants of the same transgenic line or wild type line). We did one cDNA from each RNA sample. In all sampling we used gemmalings harboring adult tissues, but with no visible gemma cups.
For time series experiments with wild type (Tak1), Mpprr ko , Mprve ko and Mptoc1 ko , three replicate samples of each entity were harvested at six-hour intervals from the second day of LL for two days (in total eight time points). Test of statistically significant expression differences between lines were performed with a linear model in R 55 (aov). The model included time, genotype and their interaction.
Analyses of nyctinastic thallus movements. 25-well square petri dishes (Fisher Scientific) were filled with Gamborg's B5 medium, after which half of the medium in each well was removed to allow placement and growth of gemmalings. One gemmaling was placed in each well, and plates were placed vertically in a Sanyo growth cabinet (MLR-350) to allow imaging from the side (see Supplementary Fig. S2). Light was supplied from either cool white fluorescent light (experiments shown in Fig. 1a,b,f,g), or blue and red LEDs (experiments shown in Figs. 1d,e and 4) at 20 °C constant temperature, or at the temperatures indicated in the text. Plants where entrained for three to five days in ND (12: 12 hour, light: dark cycles) before exposure to constant conditions and imaging. Because younger (smaller) gemmalings waves with a higher amplitude than larger plants, the controls in the two experiments presented in Fig. 4e,f cannot be directly compared as we used smaller gemmalings for the L-Kyn experiment (Fig. 4f) than for the IaaL/ARF1 experiment (Fig. 4e). Within experiments all gemmalings were of the same age. For auxin related experiments, plants were exposed to the chemical in ND for one (IAA) or three (BUM) additional days before transfer to LL. Apex region position data was extracted from images using ImageJ 56 . Images were converted to binary ones and a rectangular selection automatically following the apex region horizontally was used with the command "Analyse Particles" to extract the center of mass in each image. The obtained data on vertical position (relative center of mass of apex region) were de-trended using a cubic smoothing spline with 12 degrees of freedom with the R package smooth.spline. Estimates or circadian period, amplitude and RAE were conducted in MATLAB with a method based on Spectrum Resampling 38 .
Generation of Mplux ge -9 and -19 mutants. Two gRNAs targeting the coding region of MpLUX in or just upstream of the DNA-binding domain were constructed by annealing oligos CPEP60 + CPEP61 and CPEP62 + CPEP63, respectively 57 (listed in Supplementary Table S1). Annealed oligos were inserted into the BsaI site of plasmid pMpGE En03 58 . The constructs were subsequently cloned into pMpGE010 and pMpGW401, respectively, using GateWay LR cloning. All plasmids were sequenced.
DNA was extracted from transformed sporelings (generation T1) and their offspring (generation G1 gemmalings) using a DNeasy Plant Mini Kit (Qiagen). PCRs to amplify the targeted regions were performed using primers ME486 + ME487. PCR products were purified using a QIAquick PCR Purification Kit (Qiagen) and sequenced.
Quantification of IAA. Wild type (Upp14) and Mplux ge -9 gemmalings were first grown on solid Gamborg's B5 medium for 10 days in LD 20 °C. Gemmalings were then entrained in ND 20 °C for 3 days before being harvested every four hours over 32 hours (nine time points) sampling five replicates per genotype per time point, starting four hours after switch to LL conditions. Plant material (around 25-30 mg fresh weight) was purified as previously described 60 , and 500 pg 13 C 6 -IAA internal standard was added to each sample before homogenization and extraction. Free IAA was quantified in the purified samples using combined gas chromatography-tandem mass spectrometry.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.