science.sciencemag.org/cgi/content/full/science.

abe2305/DC1

Supplementary Materials for

Molecular mechanism of cytokinin-activated cell division in Arabidopsis

Weibing Yang, Sandra Cortijo, Niklas Korsbo, Pawel Roszak, Katharina Schiessl,
Aram Gurzadyan, Raymond Wightman, Henrik Jönsson*, Elliot Meyerowitz*
*Corresponding author. Email: meyerow@caltech.edu (E.M.); henrik.jonsson@slcu.cam.ac.uk (H.J.)

Published 25 February 2021 on Science First Release

DOI: 10.1126/science.abe2305

This PDF file includes:

Materials and Methods

Mathematical Modeling
Figs. S1 to S15
Captions for Tables S1 to S4
Captions for Movies S1 to S9
References

Other Supplementary Material for this manuscript includes the following:

(available at science.sciencemag.org/cgi/content/full/science.abe2305/DC1)

MDAR Reproducibility Checklist (.pdf)

Tables S1 to S4 (.xlsx)
Movies S1 to S9 (.mp4)
Materials and Methods
Plant materials and growth conditions
Arabidopsis thaliana plants were grown in a growth chamber under the following conditions: long-
day photoperiod (16-h light/8-h dark), light intensity 170 μmoles m-2 s-1, day/night temperature
21°C/17 °C. The Arabidopsis thaliana ecotype Columbia (Col-0) was used as the wild-type control.
The mutant lines myb3r1, myb3r4, myb3r1 myb3r4 (diploid plants as measured by flow cytometry),
and ckx3-1 ckx5 were described previously (16, 20). The impa3 (impa3-1, SALK_025919) and
impa6 (impa6-2, GK_435H12) were ordered from the Nottingham Arabidopsis Stock Centre. All
seeds were surface-sterilized with 70 % ethanol/0.5% Triton X-100 for 5 min and briefly rinsed in
95 % ethanol for 1 min. The seeds were dried under a sterile hood and germinated on horizontal
or vertical plates containing half-strength solid Murashige-Skoog medium. 11-day-old seedlings
were transferred to soil (Levington F2).

Plasmid construction and plant transformation

The MultiSite Gateway Three-Fragment Vector Construction system (Invitrogen) was used to
generate plasmid constructs. For pMYB3R1::GFP-MYB3R1, a 2,466 bp promoter upstream of
MYB3R1 start codon was amplified with primers MYB3R1P-GW-F and MYB3R1P-GW-R. The
PCR product was inserted into pDONR P4-P1R by BP reaction, resulting in 1R4-MYB3R1. A
5,600 bp genomic fragment containing the whole coding sequence of MYB3R1 as well as 995 bp
3’ region was amplified with primers MYB3R1_2R3_F and MYB3R1_2R3_R, and the PCR
product was inserted into pDONR P2R-P3, resulting in 2R3-gMYB3R1. The enhanced GFP
(EGFP) coding sequence was amplified using primers GFP_GW_F and GFP_GW_R, and the
product was inserted into pDONR 221 by BP reaction, resulting in 221-GFP. These three entry
constructs were incorporated into the binary vector pH7m34-GW by LR reaction. A similar
strategy was applied to generate pMYB3R4::GFP-MYB3R4. A 2,279 bp promoter region of
MYB3R4 were amplified with primers MYB3R4P-GW-F and MYB3R4P-GW-R. A 5,232 bp
genomic fragment containing the whole coding sequence of MYB3R4 as well as a 999 bp 3’ region
was amplified with primers MYB3R4_2R3_F and MYB3R4_2R3_R. The constructs were named
as 1R4-MYB3R4 and 2R3-gMYB3R4, respectively. 1R4-MYB3R4, 221-GFP, and 2R3-
gMYB3R4 were introduced into pH7m34-GW by LR reaction. To generate the constitutively
expressed GFP-MYB3R4, we used the Arabidopsis UBQ10 promoter. A 1,500 bp UBQ10
promoter was amplified with UBQ10-1500p-1R4-F and UBQ10-1500p-1R4-R. The PCR product
was inserted into pDONR P4-P1R by BP reaction, resulting in 1R4-UBQ10. The entry constructs
1R4-UBQ10, 221-GFP, and 2R3-gMYB3R4 were then incorporated into pB7m34-GW by LR
reaction, resulting in pUBQ10::GFP-MYB3R4.
For the MYB3R4 deletion assay, the truncated MYB3R4 genomic fragments fused with GFP
coding sequence were amplified using primers GFP_GW_F and MYB3R4-del-1-R to MYB3R4-
del-10-R. The PCR products were inserted into pDONR 221 by BP reaction, resulting in the entry
constructs 221-GFP-MYB3R4-del-1 to 221-GFP-MYB3R4-del-10. A 255 bp Nos terminator
sequence was amplified with primers Nos_BamHI_SalI_2R3_F and Nos_2R3_R. The PCR
product was introduced into pDONR P2R-P3 by BP reaction, resulting in 2R3-Nos. The entry
constructs 1R4-MYB3R4, 221-GFP-MYB3R4-del-1 to 221-GFP-MYB3R4-del-10, and 2R3-Nos
were incorporated into pB7m34-GW by LR reaction. These deletion constructs were named as
pMYB3R4::GFP-MYB3R4-del-1 to pMYB3R4::GFP-MYB3R4-del-10. To generate MYB3R4 C-

2
terminal regions fused with GFP, the coding sequence of MYB3R4 C-terminal part (amino acids
562-961) was amplified using primers MYB3R4-C-Terminal-2R3-F and MYB3R4_2R3_R. The
PCR product was inserted into pDONR P2R-P3, resulting in 2R3-MYB3R4C-ter. The entry
constructs 1R4-MYB3R4, 221-GFP, and 2R3-MYB3R4C-ter were incorporated into pB7m34-GW
by LR reaction. The resulting construct was named pMYB3R4::GFP-MYB3R4C-ter. MYB3R4mNES
was amplified by fusion PCR using primers MYB3R4_SalI & R4-NES mutation-R and R4-NES
mutation-F & MYB3R4_PstI. The PCR product was fused with GFP coding sequence and ligated
into pDONR 221, resulting in 221-GFP-MYB3R4mNES. The entry constructs 1R4-MYB3R4, 221-
GFP-MYB3R4mNES, and 2R3-Nos were incorporated into pB7m34-GW by LR reaction. The
resulting construct was named pMYB3R4::GFP-MYB3R4 mNES. The entry constructs 1R4-UBQ10,
221-GFP-MYB3R4mNES, and 2R3-Nos were incorporated into pB7m34-GW by LR reaction. The
resulting construct was named pUBQ10::GFP-MYB3R4 mNES.
To generate IMPA3 and IMPA6 promoter-driven fluorescence reporter gene expression lines, a
2,000 bp IMPA3 promoter region was amplified with primers IMPA3-Promoter-1R4-F and
IMPA3-Promoter-1R4-R. A 2,300 bp IMPA6 promoter region was amplified with primers
IMPA6-Promoter-1R4-F and IMPA6-Promoter-1R4-R. The PCR products were introduced into
pDONR P4-P1R by BP reaction, resulting in 1R4-IMPA3 and 1R4-IMPA6. To clone CYCB1;1
D box fused with GFP-N7, the coding sequence of D box was amplified with primers D-Box-F
and D-Box-R, the GFP-N7 coding sequence was amplified with primers GFP-N7-F and GFP-N7-
R. The PCR products were first introduced into pBluescript II SK (-) by KpnI/SalI and SalI/BamHI
digestion and ligation, resulting in SK-D box-GFP-N7. The primers D-Box-221-F and N7-GW-R
were then used to amplify the D box-GFP-N7 fragment, which was inserted into pDONR 221,
resulting in 221-D box-GFP-N7. The entry clones 1R4-IMPA3 or 1R4-IMPA6, 221-D box-GFP-
N7, and 2R3-Nos were incorporated into pB7m34-GW by LR reaction. The resulting construct
was named pIMPA3::D box-GFP-N7 and pIMPA6::D box-GFP-N7. To generate the IMPA3
translational fusion reporter, a 3,719 bp genomic region of IMPA3 was amplified with primers
IMPA3-DNA-2R3-F and IMPA3-DNA-2R3-R. The PCR product was inserted into pDONR P2R-
P3, resulting in 2R3-IMPA3. The entry clones 1R4-IMPA3, 221-GFP, and 2R3-IMPA3 were
incorporated into pB7m34-GW by LR reaction. The resulting construct was named
pIMPA3::GFP-IMPA3. All constructs were transformed into Arabidopsis plants via
Agrobacterium-mediated transformation.

Fluorescent reporter imaging by confocal microscopy in living plants

Shortly after bolting (with stem length ~1 cm), the shoot apices were cut and the SAMs were
dissected. For each fluorescence reporter, at least four independent transgenic lines that exhibit
similar expression patterns were analysed. The meristems were transferred to a square box
containing fresh MS medium (Duchefa Biochemie-MS basal salt mixture) supplemented with
vitamins (myoinositol 100 mg/ml, nicotinic acid 1 mg/ml, pyridoxine hydrochloride 1 mg/ml,
thiamine hydrochloride 1 mg/ml, glycine 2 mg/ml) and 1% sucrose. The cell wall was stained with
0.1% propidium iodide (PI) to show the cell boundaries. Confocal z-stacks of SAMs were acquired
with a Leica SP8 using a 25 × NA 0.95 water dipping objective. Laser excitations were 488 nm
(PI, GFP) and 555nm or 561nm (RFP). 3-D rendering was carried out using LAS X (Leica)
confocal microscope software. GFP fluorescence intensity was measured in Fiji Image J. The cell
size and cell number were analyzed using MorphoGraphX (42).

3
Quantification of RNA FISH fluorescence signals in fixed tissue section images
For CDC20 RNA FISH expression analysis, the nuclear (DAPI staining) channel was first masked
with respect to the expanded region of CDC20 FISH signals. A few nuclei ascribed to neighbouring
cells that express no CDC20 were filtered manually. The dividing nuclei were filtered out by
adaptive mean intensity thresholding [35 pixel (px) radius with -20 subtracted intensity]. Mean
Gaussian filter with 1 to 2 px radius was applied once or twice. Nuclear segmentation was
performed using the Costanza ImageJ plugin (https://www.quantitative-
plant.org/software/costanza) with minor manual post-corrections for endoreplicated cells. The
ratio of the binary domain of CDC20 expression above a fixed noise-related threshold and the
nuclear domain was calculated per cell.

Chemical treatments
Cytokinin treatment on growing shoot apices was performed as described previously (43) with
slight modifications. One week after bolting, 10 µl BA solution with 0.05% Tween 20 was
carefully applied onto the top of the shoot apices. The same amounts of Tween 20 were applied in
the mock treatments. Cytokinin was applied every two days, and the meristem phenotypes were
analyzed after three treatments.
For assessing the cytokinin effect on MYB3R4 subcellular localization, the main inflorescence
meristems were cut in 1 cm from the tip and dissected under stereoscopic microscopy to remove
the flowers. The meristems were kept in a 6-well petri dish containing various concentrations of
cytokinin solutions diluted from 100 mM stock. To prepare cytokinin stock solutions, BA and iP
were dissolved in 1 mM NaOH, and tZ in 100% DMSO. During treatment, similar volumes of
NaOH or DMSO were used as controls. A similar method was used for NAA and sucrose
treatments. All treatments were repeated at least three times.
For Leptomycin B treatment, the dissected meristems or 9-day-old seedlings were inoculated
with 100 nM Leptomycin B (Sigma). For CHX treatment, the dissected meristems were first kept
in 100 µM CHX for 2 hours, and then in a mixture of 100 µM CHX and 100 µM tZ for another 6
hours before confocal scanning.

mRNA in situ hybridization

In situ hybridisation was carried out as described previously (44). The RNA probes were prepared
by in vitro transcription using the DIG RNA Labeling Kit (Roche). The shoot apices were fixed in
FAA (3.7% formaldehyde, 5% acetic acid, 50% ethanol) and embedded in wax. The samples were
sectioned into 8-µm slices. After dewaxing, rehydration and dehydration, the sections were
hybridized with gene-specific RNA probes at 55 °C overnight. After washing in SSC, the slices
were incubated with an anti-digoxigenin-AP antibody (Roche) for 2 hours at room temperature.
The hybridisation signals were detected by NBT/BCIP (Roche) colour reaction. Images were taken
in Zeiss AxioImager M2 microscope with a PlanApochromat 20x/0.8 NA objective and a Zeiss
Axiocam MRc colour camera.
For fluorescence in situ hybridisation, anti-digoxigenin-POD (Roche) antibody was used. The
hybridization signals were detected using the TSA Plus Cy5 Fluorescence System (Perkin Elmer).

4
Slides were mounted with 1mg/ml DAPI shortly before observing the hybridization signals.
Images were taken in a Zeiss LSM700 confocal microscope with a 20 × 0.8NA dry objective.
Laser excitations were 405 nm (DAPI) and 633 nm (Cy5).

ChIP-seq and data analysis

ChIP-seq was performed on dissected shoot apices of pMYB3R1::GFP-MYB3R1,
pMYB3R4::GFP-MYB3R4, and pMYB3R4::GFP-MYB3R4mNES plants. For each replicate, around
1000 shoot apices were pooled for chromatin extraction. Plant materials were immediately frozen
in liquid nitrogen upon collection. The ChIP experiment was performed as described previously
(45) with the following modifications. 1) Cross-linking was performed on ground material in the
first buffer of chromatin extraction. The powder was resuspended into 25ml of Extraction buffer
1 (0.4 M sucrose; 10 mM Hepes pH 7.5; 10 mM MgCl2; 5 mM β-mercaptoethanol; 0.1 mM PMSF;
1 tablet of Complete protease inhibitor for 50ml of buffer). 700 μl of 37% formaldehyde was then
added to reach a final concentration of 1% formaldehyde. Samples were cross-linked for 7 minutes
at room temperature and the cross-linking reaction was stopped by adding 1.7 ml of 2 M glycine.
2) After chromatin extraction, fragmentation was performed by sonication using 18 cycles of 30
sec ON/ 30 sec OFF on a Bioruptor (Diagenode). 3) ChIPs were performed in a buffer containing
20 mM Tris-HCl (pH, 8.0), 150 mM NaCl, 2 mM EDTA, 1% Triton X-100 and 1× protease
inhibitor cocktail using an anti-GFP antibody (Ab290, Abcam) coupled to a 1:1 mix of protein-A
and protein-G Dynabeads (life technologies, 510001D and 10003D). 4) All samples were
subsequently washed twice with low salt buffer [20mM Tris buffer (pH, 8.0), 150 mM NaCl, 0.1%
SDS, 1% Triton X-100], twice with high salt buffer [20 mM Tris buffer (pH, 8.0), 500 mM NaCl,
0.1% SDS, 1% Triton X-100], once with LiCl buffer [10 mM Tris buffer (pH, 8.0), 250 mM LiCl,
1% Sodium deoxycholate, 1% NP40], and twice with TE buffer [10 mM Tris buffer (pH, 8.0), 1
mM EDTA]. The elution was done by incubating the samples at 65°C for 15min with 100 μl of
elution buffer (1% SDS, 0.1M NaHCO3). 5) DNA was extracted using AMPure beads with a ratio
of 2.3 × as described previously (46). The efficiency of ChIP was assessed by qPCR prior to in-
house library preparation, using the TruSeq ChIP sample preparation kit (Illumina, IP-202-1012)
following the manufacturer’s instructions. The libraries were sequenced on HiSeq2100 by
Novogene using paired-end 150bp length reads.
Sequenced data were analyzed using a combination of publicly available software and in-house
scripts. We first assessed the quality of reads using FastQC
(www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were then aligned to the TAIR10
genome using Bowtie2 (47). Potential optical duplicates were removed using Picard tools
(https://github.com/broadinstitute/picard). Peak calling was performed for each ChIP using
MASC2 (48), with the corresponding INPUT used as a reference. A 60 bp regions around the
summit of the top 300 peaks were used to perform motif discovery, using the MEME tool
(http://meme-suite.org/tools/meme) (49). Targets for each ChIP were identified for the significant
peaks using PAVIS2 (https://manticore.niehs.nih.gov/pavis2/) (50). Snapshots of ChIP-seq signal
around targets were shown using the Integrative Genomics Viewer (IGV) (51). Averaged profiles
and heatmap of the ChIP-seq signal around the Transcription Start Site (TSS) of target genes were
generated using deepTools (52, 53). The genes that appear for at least 2 times in the 3 replicates
were considered as true targets.

5
Transcriptome analysis
Total RNA was isolated from dissected inflorescence meristems using RNeasy Mini Kit (QIAGEN
Cat No./ID: 74104), following the manufacturer’s instructions. 36 inflorescence meristems were
pooled as one replicate. RNA quality and integrity were assessed on the Agilent 2200 TapeStation.
The libraries were prepared and sequenced using Illumina methods by Novogene, producing
paired-end 150 bp length reads. The quality of the sequenced RNA-seq data was assessed using
FastQC (www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were then aligned to the
TAIR10 transcriptome using salmon (54). Expression levels were extracted for each gene using
the tximport library in the statistical programme R. Heatmaps representing hierarchical clustering
of genes based on their expression were produced showing the log2 (mutant/wild-type) ratio using
the function hclust on 1-Pearson correlation in the statistical programme R. Gene ontology (GO)
enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were
conducted in ShinyGO (55). Analysis of gene expression for the ChIP-seq targets throughout the
cell cycle was performed using publicly available data (27).

Transactivation assay
The transcriptional activities of MYB3R1 and MYB3R4 on their target genes were assessed by a
dual luciferase assay. 1,606 bp of the CYCB1;2 promoter was amplified using primers CYCB1;2-
P-F and CYCB1;2-P-R. 1,675 bp of the CDC20 promoter was amplified using primers CDC20-P-
F and CDC20-P-R. All PCR products were cloned into the pGreenII 0800-LUC vector (56) with
KpnI and BamHI. In these constructs, LUC was expressed under the control of target gene
promoters. REN expression was controlled by 35S promoter, serving as an internal control. These
dual luciferase reporters were introduced into tobacco leaves together with UBQ10::GFP-
MYB3R1, UBQ10::GFP-MYB3R4, UBQ10::GFP-MYB3R4mNES, and the RNA silencing
suppressor P19 through Agrobacterium infiltration. The activities of LUC and REN were
quantified using the Dual-Luciferase® Reporter Assay System (Promega) in the SpectraMax iD3
Multi-Mode Microplate Reader (Molecular Devices). The LUC activity was normalized to the
REN activity (LUC/REN). The means and standard errors of LUC/REN were calculated from six
independent biological replicates.

BiFC
To generate MYB3R1 and MYB3R4 proteins fused with each half of YFP, the coding sequences
of YFP N-terminus (amino acids 1-155) was amplified with primers nYFP-221-F and nYFP-221-
R. The C-terminus (amino acids 156-240) of YFP was amplified with primers cYFP-221-F and
cYFP-221-R. The PCR products were inserted into pDONR221 by BP reaction, resulting in 221-
YFPN and 221-YFPC. The entry constructs 1R4-UBQ10, 221-YFPN or 221-YFPC, and 2R3-
gMYB3R1 were incorporated into pB7m34-GW by LR reaction, resulting in UBQ10::YFPN-
MYB3R1 and UBQ10::YFPC-MYB3R1. A similar strategy was used to generate UBQ10::YFPN-
MYB3R4 and UBQ10::YFPC-MYB3R4. UBQ10::YFPN-MYB3R1, UBQ10::YFPC-MYB3R4, and
the P19 silencing suppressor were co-introduced into N. benthamiana leaves through
Agrobacterium-mediated transformation. The infiltrated tobacco leaves were grown for three days
before confocal microscopic scanning. A similar experiment was performed for the construct pair
UBQ10::YFPN-MYB3R4 and UBQ10::YFPC-MYB3R1, as well as UBQ10::YFPN-MYB3R1 and

6
UBQ10::YFPC-MYB73. YFP fluorescence signals were acquired in Leica SP8 with 25 × NA 0.95
water dipping objective. Laser excitation for YFP was 488 nm.

Co-Immunoprecipitation (Co-IP)
GFP-MYB3R1 and HA-MYB3R4 fusion proteins were used for the Co-IP experiments. The
UBQ10::GFP-MYB3R1 construct was the same as that used in the transactivation assay. To
generate HA-MYB3R4, the coding sequencing of a 3 × HA tag was amplified with primers
HA_221_F and HA_221_R. The PCR product was introduced into pDONR221 by BP reaction.
The resulting 221-HA, were incorporated together with 1R4-UBQ10 and 2R3-gMYB3R4 into
pB7m34-GW by LR reaction, resulting in UBQ10::HA-MYB3R4.
UBQ10::GFP-MYB3R1 and UBQ10::HA-MYB3R4 constructs were co-infiltrated into N.
benthamiana leaves with P19 through Agrobacterium-mediated transformation. After growing for
another 3 days, the infiltrated leaves were ground into powder in liquid nitrogen and homogenized
in Lysis Buffer [10% Glycerol, 25 mM Tris-Cl pH 7.5, 1 mM EDTA pH 8.0, 150 mM NaCl, 2%
(w/v) polyvinylpyrrolidone (PVP), 1mM DTT, 0.1% Tween-20, proteinase inhibitor cocktail].
Following a brief centrifugation at 3,000 rpm for 3 minutes, the supernatant was incubated with
anti-GFP beads (GFP-Trap, ChromoTek) for 2 hours at 5°C with gentle rotation. The beads were
collected by centrifugation at 800 g for 2 minutes. After washing 5 times with IP Buffer (10%
Glycerol, 25 mM Tris-Cl pH 7.5, 1 mM EDTA pH 8.0, 150 mM NaCl, 0.1% Tween-20), the
proteins were eluted from the beads by boiling in 2 × Laemmli Sample Buffer for 10 min. The
proteins were then separated by SDS–PAGE and transferred to a PVDF membrane. Proteins were
detected using rabbit anti-GFP antibody (Abcam, catalogue ab290), horseradish peroxidase
(HRP)-conjugated anti-rabbit antibody (Abcam, catalogue ab205718), mouse anti-HA antibody
(Sigma, catalogue H9658), and rabbit horseradish peroxidase (HRP)-conjugated anti-mouse
antibody (Sigma, catalogue A9044). The co-immunoprecipitation experiments were repeated three
times with similar results.

MYB3R4 and IMPA3 co-expression in tobacco

To generate RFP-IMPA3, the coding sequence of RFP was amplified using primers RFP-221-F
and RFP-221-F. The PCR product was inserted into pDONR 221 by BP reaction, resulting in 221-
RFP. The entry constructs 1R4-UBQ10, 221-RFP, and 2R3-IMPA3 was ligated into pB7m34-GW
by LR reaction, resulting in UBQ10::RFP-IMPA3. UBQ10::GFP, UBQ10::GFP-MYB3R4,
UBQ10::GFP together with UBQ10::RFP-IMPA3, or UBQ10::GFP-MYB3R4 together with
UBQ10::RFP-IMPA3 were co-infiltrated into N. benthamiana leaves with P19 through
Agrobacterium-mediated transformation. Fluorescence signals were examined 3 days after
transformation.

7
Fig. S1. Dose-dependent cytokinin response of the SAM.
(A) Diagram showing the course of cytokinin treatment at the shoot apex, and a photograph of
cytokinin treatment of the shoot apex.
(B and C) Dose-dependent cytokinin response of the SAM. Shown are the shoot apex (top panels)
and SAM (bottom panels) morphologies of wild-type plants treated with different concentrations
of cytokinin (B), and the non-treated ckx3 ckx5 double mutant (C). The SAM domains are outlined
with dashed lines (in magenta). Scale bar in top panels, 1 mm; bottom panels, 20 µm.
(D and E) Quantification of meristem size. SAM size is estimated by measuring the radius (r) of
3-D projections of confocal stacks as indicated in (D), which shows a wild-type SAM stained with
propidium iodide (PI, displayed in magenta). ****P < 0.0001 (two-tailed t-test).

8
Fig. S2. MYB3R1 and MYB3R4 are required for meristem maintenance under cytokinin
treatment.
(A) The mRNA expression patterns of MYB3R genes, as revealed by in situ hybridisation. Only
MYB3R1 and MYB3R4 are detectably expressed in the SAM, displaying a patchy pattern typical
to cell cycle-controlled genes. Scale bar, 50 µm.
(B-E) Shoot apices of wild-type (B), myb3r1 (C), myb3r4 (D), and myb3r1 myb3r4 (E) plants after
mock or 100 µM BA treatment. The cytokinin-treated myb3r1 myb3r4 plants are grouped into two
types: Type I, with meristem; Type II, meristem early termination. Scale bars, 1 mm.
(F) 3-D projection of a terminated myb3r1 myb3r4 meristem. The SAM is indicated with an arrow.
Scale bar, 50 µm.

9
Fig. S3. Conserved functions of MYB3R1 and MYB3R4 in root cell division.
(A) Growth phenotypes of 5-day-old wild-type and myb3r1 myb3r4 seedlings.
(B) Cell division defect in the root apical meristem of myb3r1 myb3r4 double mutant. The highly
elongated vascular cells (outlined) are regularly found only in the double mutant. Cell walls are
stained with PI and shown in red. Scale bar, 50 μm.
(C) Quantification of protophloem and metaxylem cell length. The first 8 cells from the quiescent
centre (QC) were measured. ****P < 0.0001 (two-tailed t-test).
(D) Cross sections of wild-type and myb3r1 myb3r4 roots at the elongation zone. myb3r1 myb3r4
shows defects in formative cell divisions. Asterisks indicate epidermal cells. Scale bar, 25 μm.
(E) Quantification of the number of cell files in each tissue type.
(F) Root patterning defect in the vascular tissue (outlined with dotted line) of myb3r1 myb3r4.
Arrows indicate protophloem sieve elements. Scale bar, 25 μm.

10
Fig. S4. MYB3R1 and MYB3R4 promote gene expression.
(A) Overlap of MYB3R1 and MYB3R4 binding regions in proximity to the transcription start site
(TSS) of their target genes. Top panels: averaging profile of MYB3R1 and MYB3R4 occupancy
at their target genes. Bottom panels: heatmap of MYB3R1 and MYB3R4 enrichment. Red means
a high level and blue means a low level of transcription factor binding.
(B) UpSet plot showing the shared up-regulation and down-regulation of MYB3R1 and MYB3R4
common target genes in myb3r1, myb3r4, and myb3r1 myb3r4.
(C) Boxplot showing the relative expression of MYB3R1 and MYB3R4 target genes in the shoot
apices of myb3r1 and myb3r4 single mutants, and the myb3r1 myb3r4 double mutant (respectively
shown as red, green and blue) as compared to wild type.
(D) Transcriptional activity of MYB3R1 and MYB3R4. Shown are the ratios of firefly luciferease
(LUC) to Renilla luciferase (REN) activity in tobacco leaf cells co-expressing different reporter
constructs together with MYB3R1 or MYB3R4. The expression of LUC is controlled by the
promoters of target genes, while the expression of REN is driven by the 35S promoter. Data are
from six biological replicates. Error bars, mean ±s.e.m.

11
Fig. S5. MYB3R4 interacts with MYB3R1 in the nucleus.
(A) BiFC assay showing interaction of MYB3R1 with MYB3R4. The N- and C-terminal portion
of YFP were fused with MYB3R1 or MYB3R4, and co-expressed in tobacco leaf cells. The nuclear
localized MYB transcription factor MYB73 was used as a negative control. Scale bars, 50 µm.
(B) Co-immunoprecipitation (Co-IP) of GFP-MYB3R1 and HA-MYB3R4. IP was performed
using an anti-GFP antibody. Proteins were extracted from tobacco leaves expressing the constructs
listed above the gel images.

12
Fig. S6. MYB3R1 and MYB3R4 activate the expression of mitotic cell cycle genes.
(A) The expression dynamics of MYB3R1 and MYB3R4 target genes during the cell cycle in
synchronized Arabidopsis cell cultures. Data were retrieved from Menges et al. (27) and re-
analysed. While the expression of a majority of MYB3R1 target genes is not cell cycle regulated,
MYB3R4 (or MYB3R1/4 common) targets show increased expression at G2/M transition.
(B and C) Gene Ontology (GO) analysis of MYB3R1 and MYB3R4 target genes. Shown are the
percentage of target genes in the enriched GO terms (P value cutoff, 0.001). See Table S2 for a
complete list of GO terms.
(D) RNA FISH to compare the mRNA expression of MYB3R1 and MYB3R4 target genes in wild-
type and myb3r1 myb3r4 SAMs. Gene specific RNA probes were labelled with DIG and detected
by TSA-CY5 (magenta signals). The nuclei are stained with DAPI and shown in blue. FISH signals
are shown in the Fire format. Most of the genes are specifically expressed in mitotic cells and the
expression levels are greatly reduced in myb3r1 myb3r4. Scale bars, 50 µm.
(E) Quantification of the RNA FISH fluorescence intensities in wild-type and myb3r1 myb3r4
SAMs. Note that the expression levels of TANGLED, AURORA2, CCS52B, ZERZAUST mRNAs
are greatly reduced in myb3r1 myb3r4 SAMs, which were under the detection limit in our assays.
N.D., not detected.

13
Fig. S7. Subcellular localization patterns of MYB3R1 and MYB3R4.
(A) Complementation of the myb3r1 myb3r4 mutant phenotypes with GFP-MYB3R1 and GFP-
MYB3R4 fusion proteins. GFP-MYB3R1 and GFP-MYB3R4 are expressed under the control of
their own promoters. Scale bar, 2 cm.
(B and C) Subcellular localization patterns of GFP-MYB3R1 and GFP-MYB3R4 in the cells of
the SAM. MYB3R1 is localized only in the nucleus (B), whereas MYB3R4 is distributed
throughout the nucleus and the cytoplasm (C). Right panels show enlarged views of individual
cells. + PI indicates the addition of PI to stain the cell walls. Scale bars in left panels, 20 μm; in
right panels, 5 μm.

14
Fig. S8. MYB3R4 undergoes active nuclear export.
(A) The subcellular localization patterns of GFP-MYB3R4 in the SAM and its response to
leptomycin B (LMB) treatment. LMB is a specific nuclear export inhibitor. Scale bars, 20 µm.
(B) The expression patterns of GFP-MYB3R4 in the root cells. Scale bars, 50 µm.
(C) Ectopically expressed GFP-MYB3R4 under the control of UBQ10 promoter exhibits nucleo-
cytoplasmic localization in root dividing cells. The enlarged panels show two prophase cells. Scale
bars in left panels, 50 µm; in right panels, 20 µm.
(D) Nucleo-cytoplasmic localization of GFP-MYB3R4 in root elongating cells. Scale bars, 50 µm.

15
Fig. S9. Nucleo-cytoplasmic shuttling of MYB3R4 during cell division.
(A) The dynamics of GFP-MYB3R4 in a cell undergoing mitosis. The zero time points are when
most MYB3R4 proteins transiently re-localized into the nucleus. NEBD, nuclear envelope
breakdown. GFP signals are displayed in the Fire format. Scale bar, 5 μm.
(B) The nuclear (nuc)-to-cytoplasmic (cyt) ratio of GFP-MYB3R4. Fluorescence intensities were
quantified in Fiji. Dashed line indicates the stages where the nuclear envelope breaks down (and
thus when the nuc/cyt ratio cannot be measured).

16
Fig. S10. GFP-MYB3R4 protein localization in response to various treatments.
(A-C) Localization of GFP-MYB3R4 following light/dark (A), sucrose (B), and nutrient (C)
treatments. Dissected SAMs were treated for 12 hours in (A and B). For nutrient treatment in (C),
plants were grown on soils of different nutritive quality, and SAMs were observed immediately
after dissection. All scale bars, 20 μm.
(D) Quantification of MYB3R4 nuclear (nuc)-to-cytoplasmic (cyt) ratio. *P < 0.05, **
P < 0.01,
****
P < 0.0001 (two-tailed t-test).

17
Fig. S11. Cytokinin promotes MYB3R4 nuclear localization.
(A) The effect of cytokinin on GFP-MYB3R4 subcellular localization. SAMs were incubated with
tZ solutions for 6 hours. High concentration of tZ promotes MYB3R4 nuclear localization. Scale
bars, 20 μm.
(B) The localization patterns of GFP-MYB3R4 at different times after 100 μM tZ treatment. Scale
bars, 20 μm.
(C) Auxin does not affect GFP-MYB3R4 localization. Dissected SAMs were treated with 100 μM
NAA for 6 hours. Scale bars, 20 μm.
(D) CHX does not affect cytokinin-triggered MYB3R4 nuclear localization. Dissected meristems
were first treated with 100 µM CHX for 2 hours, and then in a mixture of 100 µM CHX and 100
µM tZ for another 6 hours before confocal scanning. Scale bars, 20 μm.
Nuclear (nuc)-to-cytoplasmic (cyt) ratio of MYB3R4 was quantified and shown on the right side
of each panel. *P < 0.05, ****P < 0.0001 (two-tailed t-test).

18
Fig. S12. Systematic analysis of importin and exportin gene expression in the SAM.
(A) The expression levels of importin and exportin genes in different tissues. Each sample has two
replicates. TPM values were retrived from our RNA-seq data published previously (28).
(B) mRNA in situ hybridisation to show the expression patterns of importin and exportin genes.
The patchy pattern of IMPA3 and IMPA6 indicates cell cycle-controlled expression. See Table S3
for importin and exportin gene information. Scale bar for whole shoot apex, 50 μm; for enlarged
regions of IMPA3 and IMPA6, 20 μm.

19
Fig. S13. MYB3R1 and MYB3R4 activate IMPA3 and IMPA6 expression in dividing cells.
(A) Comparison of IMPA3 and IMPA6 expression in wild-type and myb3r1 myb3r4 double mutant
SAMs by RNA FISH. Scale bars, 50 µm.
(B) Quantification of RNA FISH fluorescence signals. ****P < 0.0001 (two-tailed t-test).
(C) The expression patterns of pIMPA3::D box-GFP-N7 and pIMPA6::D box-GFP-N7. N7 is a
nuclear targeting peptide. Fusion of the CYCB1;1 D-box domain ensures timely elimination of
fluorescent proteins after cell division. Scale bars, 50 μm.
(D) Genome browser tracks of MYB3R1 and MYB3R4 ChIP-seq coverage at IMPA3 and IMPA6
genomic regions. The binding affinity of MYB3R4 to IMPA3 and IMPA6 promoters is weaker
than that of MYB3R1.
(E) The localization patterns of GFP-MYB3R4 in wild type, impa3-1, and impa6-2 SAMs. Scale
bars, 20 µm.
(F) Quantification of MYB3R4 nuclear (nuc)-to-cytoplasmic (cyt) ratio. ****P < 0.0001 (two-tailed
t-test).

20
Fig. S14. A MYB3R4-IMPA3/6 positive feedback loop enables robust MYB3R4 nuclear re-
localization.
(A) GFP-MYB3R4 protein distribution patterns in early prophase cells. The cells are classified
into two groups. In Group I cells, GFP-MYB3R4 is localized mainly in the cytoplasm, and the two
representative cells show cytoplasmic enrichment (left) and homogeneous distribution (right). In
Group II cells, GFP-MYB3R4 is localized mainly in the nucleus, and the two representative cells
show nuclear enrichment (left) and exclusive nuclear localization (right). Scale bar, 5 µm.
(B) Measurements of nuclear and cytoplasmic GFP-MYB3R4 fluorescence. K-means clustering
reveals two distinct groups corresponding to the Group I and Group II cells shown in (A).
(C) Time-course dynamics of models wherein a pulse of cytokinin induces nuclear import (‘import
model’) or inhibits nuclear export (‘export model’). Simulation of the cytokinin (CYT) peak
during cell division is shown in the top panel. The levels of nuclear localised MYB3R4 (N) and
the ratio between nuclear (N) and cytoplasmic (C) MYB3R4 is shown in the middle and bottom
panels, respectively. The target peak value of N/C (shown in grey) is estimated from the mean and
standard deviation of Group I measurements in (B). Model parameters and conditions were
inferred from biological measurements. The uncertainties from those measurements were
propagated to the simulation and are shown as transparent ribbons.
(D) The nucleo-cytoplasmic shuttling equilibration time and the equilibrium values of nuclear
MYB3R4 (N) for the import and export models when exposed to different levels of cytokinin
(CYT).
(E) Cell cycle-regulated expression of the importin IMPA3. Shown is GFP-IMPA3 expressed
under the control of its own promoter in the SAM. GFP signals are displayed in the format of Fire.

21
GFP-IMPA3 shows high expression in dividing cells and low expression in non-dividing cells.
Scale bar, 10 µm.
(F) Schematic illustration to show the positive feedback loop between MYB3R4 nuclear import
and IMPA3/6 gene expression.
(G) The change-rate of nuclear (N) and cytoplasmic (C) MYB3R4 ratio under various levels of
nuclear MYB3R4. Equilibrium is found where the model lines cross 0 (grey line). In the absence
of CYT, the system equilibrates to N ≈ 500. A sudden increase to a CYT input of 400 (c.f. peak)
would jump the system over to the dashed lines.
(H) The MYB3R4-IMPA3/6 feedback leads to stronger nuclear re-localization of MYB3R4 in
response to CYT input. A similar effect can be seen in both export and import models.

22
Fig. S15. Constitutive MYB3R4 nuclear localization promotes target gene expression.
(A) Schematic diagram displaying the constructs of GFP-tagged MYB3R4 deletion proteins that
were tested.
(B) The subcellular localization of different proteins. GFP-fused MYB3R4 truncated proteins were
expressed in wild-type plants under the control of the MYB3R4 promoter. Nucleo-cytoplasmic
localization was detected for del-1, del-2, del-4, del-5, and del-6, whereas del-7, del-8, del-9, and
del-10, containing the N-terminal regions, were localized only in the nucleus. The GFP tagged C-
terminal part of MYB3R4 showed exclusive cytoplasmic localization. Scale bar, 20 μm for SAM
overview; 4 μm for enlarged cells.
(C) Dual luciferase assay to test the transcriptional activity of MYB3R4 and MYB3R4mNES. Shown
are the ratios of firefly luciferease (LUC) to Renilla luciferase (REN) activity in tobacco leaf cells
co-expressing different reporter constructs with MYB3R4 and MYB3R4mNES. Data are from six
biological replicates. Error bars, mean ±s.e.m
(D) Averaging profile (top panel) and heatmap (bottom panel) of MYB3R4 and MYB3R4mNES
occupancy at MYB3R4 target genes. Mutation of the NES does not affect MYB3R4 chromosome
binding in ChIP-seq experiments.
(E) Heatmap showing the hierarchical clustering of MYB3R4 target genes based on their
expression levels in myb3r4 and pMYB3R4::GFP-MYB3R4mNES SAMs as compared to wild type.
(F and G) Growth phenotypes of 35S::CKX1 (F) and ahk2 ahk3 (G) plants expressing GFP-
MYB3R4 or GFP-MYB3R4mNES under the control of MYB3R4 promoter. 35S::CKX1 plants are
cytokinin-deficient plants caused by the overexpression of a cytokinin oxidase CKX1. The ahk2
ahk3 plants are cytokinin receptor mutants with reduced cytokinin response. The shoot growth

23
retardation of both 35S::CKX1 and ahk2 ahk3 could be partially rescued by the constitutively
nuclear localized form of MYB3R4 (MYB3R4mNES) (8 out of 18 lines for 35S::CKX1 and 7 out of
16 lines for ahk2 ahk3). Scale bars, 1cm.

Table S1.
A list of MYB3R1 and MYB3R4 target genes identified by ChIP-seq.

Table S2.
GO analysis of MYB3R1 and MYB3R4 target genes. Shown are the top 100 GO terms.

Table S3.
Complete list of importin and exportin genes expressed in the SAM.

Table S4.
Primers used in this study.

Movie S1.
pMYB3R1::GFP-MYB3R1 expression pattern in the SAM. This movie contains a full z-stack of
confocal images showing fluorescence signals from GFP-MYB3R1 (in green) expressed under the
control of its native promoter in the same inflorescence meristem as in Fig. S7B. The cell walls
are stained with propidium iodide (PI) and displayed in magenta.

Movie S2.
pMYB3R4::GFP-MYB3R4 expression in the SAM. This movie contains a full z-stack of confocal
images showing fluorescence signals from GFP-MYB3R4 (in green) driven by its native promoter
in the same inflorescence meristem as in Fig. S7C. The cell walls are stained with propidium iodide
(PI) and displayed in magenta.

Movie S3.
pMYB3R4::GFP-MYB3R4 expression in a genetic background with a transgenic nuclear-
localized reporter and a membrane reporter. This movie contains a full z-stack of confocal images
showing GFP-MYB3R4 green fluorescence signals, together with the red fluorescence signals
from H2B-RFP and the blue signals from myr-CFP, in the same inflorescence meristems as in Fig.
2E.

24
Movie S4.
GFP-MYB3R4 subcellular localization pattern after LMB treatment. This movie contains full z-
stack of confocal images showing green fluorescence signals from GFP-MYB3R4, together with
H2B-RFP. LMB is a nuclear export inhibitor, which leads to complete nuclear localization of GFP-
MYB3R4.

Movie S5.
The expression pattern of GFP-MYB3R4 after tZ treatment. This movie contains a full z-stack of
confocal images showing fluorescence signals from GFP fused with MYB3R4 in the same
inflorescence meristem as in Fig. 3A (left panel). The cell walls are stained with propidium iodide
(PI) and shown in magenta. Cytokinin treatment leads to complete nuclear re-localization of GFP-
MYB3R4.

Movie S6.
The expression pattern of GFP-MYB3R4 after iP treatment. This movie contains a full z-stack of
confocal images showing fluorescence signals from GFP fused with MYB3R4 in the same
inflorescence meristem as in Fig. 3A (middle panel). The cell walls are stained with propidium
iodide (PI) and shown in magenta.

Movie S7.
The expression pattern of GFP-MYB3R4 after BA treatment. This movie contains a full z-stack
of confocal images showing fluorescence signals from GFP fused with MYB3R4 in the same
inflorescence meristem as in Fig. 3A (right panel). The cell walls are stained with propidium iodide
(PI) and shown in magenta.

Movie S8.
The expression pattern of GFP-MYB3R4C-ter in the SAM under the control of MYB3R4 promoter.
This movie contains a full z-stack of confocal images showing fluorescence signals from GFP
fused with the C-terminal region of MYB3R4 (aa 562-961), corresponding to Fig. S15B. The cell
walls are stained with propidium iodide (PI) and shown in magenta. All GFP signals can be seen
only in the cytoplasm.

Movie S9.
The expression pattern of GFP-MYB3R4mNES together with the nuclear reporter H2B-RFP in the
SAM. This movie contains a full z-stack of confocal images showing fluorescence signals from
GFP fused with a mutated version of MYB3R4, in which two conserved residues of the NES were
substituted with Ala. All GFP signals could be detected only in the nucleus.

25
Supplementary Modelling Information
Based on the experimental evidence, we develop a hypothesis about how cytokinin,
MYB3R4, and IMPA3/6 interact to control the localisation of MYB3R4 and thus,
the onset of cell division. Here, we formalise the hypothesis in mathematical terms
and test whether it is capable of capturing experimentally observed dynamics.

CYT-affected nuclear flux of MYB3R4

In order to model cytokinin effect on MYB3R4 nucleo-cytoplasmic shuttling,
we consider two scenarios: i) cytokinin promotes MYB3R4 nuclear import, and
ii) cytokinin inhibits MYB3R4 nuclear export. The consequence of both would
be similar, i.e. an increase in the nuclear proportion of MYB3R4 proteins. We
first studied the dynamical difference between cytokinin increasing import or
decreasing export. To do this, we created a simple deterministic model, based on
the law of mass action, wherein MYB3R4 is shuttled in and out of the nucleus
(with respective import and export rate constants ri and re ). We separate
nuclear-localised MYB3R4, N , and cytosolic MYB3R4, C, and we assume that
the total amount of MYB3R4, N + C = R4tot , is constant. We further make the
simple assumption that the rate of import and export is directly proportional
to the availability of cytosolic and nuclear MYB3R4, respectively. From these
assumptions, we created a model where CYT increases the rate of import (the
“import model”):
CYT contrib. Cytosolic MYB3R4
dN (t) z }| { z }| {
= ri · (k + CY T ) · (R4tot − N ) − re · N . (1)
dt | {z } | {z }
Nuclear import Nuclear export

Here, CYT is a model input and k is a parameter that helps set the constitutive
import rate. This constitutive import mainly affects the model dynamics at low
CYT inputs.
We also created a model where CYT decreases the export rate (the “export
model”):
dN (t) re
= ri · (R4tot − N ) − · N. (2)
dt k + CY T

Here, k also determines the model behaviour by limiting the effect of low CYT,
but this time through ensuring a maximal export-rate in the absence of CYT.
With identical parameter values, the steady states of these two models are also
identical.
ri · (k + CY T ) · R4tot
Neq = (3)
ri · (k + CY T ) + re

However, by comparing the derivatives for the two models, labelled Ni and Ne
for the import or the export model, respectively, we get

1
 dNi dNe
= (k + CY T ) ·
dt dt
The two models share limit-behaviour but the dynamics of the export model
slows down in comparison to the import model when CYT is applied.
Equilibration is only reached asymptotically, meaning that the system never
actually gets there. Therefore, we measure the equilibration time, teq as the
time it takes to get half-way to the equilibrium from one state.
For the import model, we have that

log (2)
teq = . (4)
re + ri · (k + cyt)

while the export model equilibrates according to

log (2)
teq = re (5)
(k+cyt)+ ri

Thus, if CYT induced nuclear import then increasing the CYT level would speed
up equilibration while equilibration would be slowed down if CYT repressed
nuclear export.
For any given value of CYT, the export model can be as fast as the import model
if both the import and export rate parameters, ri and re , are a factor (k + cyt)
larger than that of the import model. However, the two models will still behave
differently if cyt varies between or within simulations. Tuning parameters such
that the two models match for one cyt input will ensure that they do not match
for any other.
Three pieces of experimental data were important for the model specification
and parameter inference. The first one regards the equilibration time, teq , of the
MYB3R4 localisation. To estimate this, we used a previously measured equili-
bration time for importin-mediated nuclear import in mamalian cells (Riddick
and Macara, 2005). From that data, we assume that

teq = 6 ± 3min. (6)

Here, and in the rest of the document x ± y should be read to mean an estimated
value of x with an estimated uncertainty of y. The second measurement was the
ratio of nuclear to cytosolic localisation of MYB3R4 before the CYT peak. This
was measured through quantification of fluorescence intensity using ImageJ in
the shoot apical meristems expressing GFP-MYB3R4 together with a nuclear
marker H2B-RFP. Since the measured data included cells in all stages of cell
division we used k-means clustering to separate cells with a high or low nuclear
to cytoplasmic ratio. We assume that, unlike the group with the low ratio, the

2
group with the high ratio has already been exposed to the CYT peak. From the
mean and standard deviation of the low ratio cells (Fig. S15A) we estimate that
the pre-CYT nuclear to cytosolic localisation of MYB3R4, ∆, was
Neq
∆≡ = 0.81 ± 0.23. (7)
Ceq

The third measurement was the ratio between the peak concentration of CYT
to the basal CYT concentration (Redig et al., 1996),

CY Tpeak CY Tpeak
σ≡ = = 5 ± 2. (8)
CY Tbasal k

To mimic the CYT peak prior to cell division, we used an impulse input, at
t = 1, with a gamma-distributed delay to get (Korsbo and Jönsson, 2020)
(
0 t<1
CY T (t) = γ·rn ·(t−1)n−1 ·e−r·(t−1) (9)
(n−1)! t ≥ 1.

The shape parameters, n and r, were set to 6 and 10, respectively, to loosely
mimic the observed CYT peak shape. γ is a scaling parameter for which the
main dynamical effect comes from its ratio to the k parameter. This ratio is
constrained by equation 8 since γ controls the CYT peak height. With this
constraint, and with the way we calculate ri and re , the actual values of k
and γ does not matter much. For deterministic simulations, they only serve
to uniformly scale the solution. For stochastic simulations their scaling effect
determines the degree to with noise affects the CYT peak. The values were set
to k = 100 and γ = 2500 ± 1000, which fulfills equation equation 8 and ensures
a somewhat realistic signal to noise ratio of stochastic simulations of the CYT
peak.
The import and export rate parameters were calculated based on ∆ and teq .
We assumed that the measured ∆ describes the nuclear to cytosolic ratio at
equilibrium, when there is no CYT peak. This enables us to combine equation 3
with equation 4 for the import model or equation 5 for the export model. For
the import model, we then get

log (2)
ri = 1

= 0.031 ± 0.016 (10)
teq · (k + CY Teq ) · 1 + ∆

and
log (2)
re = = 3.8 ± 2 (11)
teq · (1 + ratio)
For the export model, we instead get

log (2)
ri = 1

= 3.1 ± 1.6 (12)
teq · 1 + ∆

3
and
log (2) · (k + CY Teq )
re = = 380 ± 200. (13)
teq · (1 + ∆)
With this, all the parameters that are relevant for the dynamics of deterministic
model simulations have been anchored to experimental observations. Upon
exposure to a peak of CYT, both models lead to a rapid and distinct nuclear
localisation of MYB3R4, but the import model is stronger, whereas the dynamics
of the export model is weaker.

Table 1: Parameter values for the import and export models (no feedback).
Parameter Import model Export model
ri 0.031 ± 0.016 3.1 ± 1.6
k 100.0 100.0
re 3.8 ± 2.0 380.0 ± 200.0
R4tot 1000.0 1000.0

The effect of MYB3R4-IMPA3/6 feedback on MYB3R4 nu-

clear re-localisation
Here, we probe the dynamical consequences of the observed MYB3R4-
IMPA3/IMPA6 feedback. This is done in a simplified way by extending the
models above with a nuclear import feedback term that increases in strength.
For the import model, we then get
 
dN (t) v·N
= ri + · (R4tot − N ) · (k + CY T ) − re · N. (14)
dt α+N
| {z }
Constitutive + feedback

For the export model, we instead get

 
dN (t) v·N re · N
= ri + · (R4tot − N ) − . (15)
dt α+N (k + CY T )
| {z }
Constitutive + feedback

Here, the new parameter α determines the value of N at which the feedback
has half its maximal effect. This is the only parameter value that we lack some
evidence for but its dynamical effect is rather limited. For α << R4tot , we
get that the feedback is always on at its maximal value, effectively turning the
feedback term into a constant and removing the actual feedback from the model.
Rejecting this scenario, we set α = R4tot , which ensures that the feedback term
never comes close to saturation in the simulation. Further increases of α can
mostly be compensated for by a matching increase in v, rendering the dynamics
mostly unaffected. The parameter v scales the contribution of the feedback
term. We think of the feedback term to represent the action of IMAP3/IMPA6
while the constitutive import term, ri , represents the action of other importins.

4
From the confocal images of GFP-IMPA3 (Fig. S15E), we estimate that the
abundance of importins in dividing cells is about five times higher than that of
cells outside of mitosis. To emulate this, we set the v parameter to ensure that
the feedback term had a maximal contribution that was about five times higher
than the constitutive import rate, ri (Table 2).
The import and export rate parameters, ri and re , were calculated from the
same measurements used in the non-feedback models. However, since the model
equations are different, so are the expressions which map these observations
to the parameter values. To derive the new expressions, we first assumed that
the value of teq was measured in the absence of a feedback v = 0. This means
that the equations 4 and 5 still hold for the feedback import and export models,
respectively. This, combined with the model equations 14 and 15 can be solved
for ri and re . For the import model with feedback, we get
log(2) β
teq ·(k+cyteq ) − ∆
ri = 1 (16)
1+ ∆

where  
∆·R4tot
v· 1+∆
β=   (17)
∆·R4tot
α+ 1+∆

and
log(2)
teq + β · (k + cyteq )
re = . (18)
1+∆
For the export model with feedback, we instead get
log(2) β
teq − ∆
ri = 1 (19)
1+ ∆

and  
log(2)
teq + β · (α + cyteq )
re = . (20)
1+∆

Table 2: Parameter values for feedback models

Parameter Import model Export model
vi 0.1 10.0
ri 0.014 ± 0.016 1.4 ± 1.6
α 1000.0 1000.0
k 100.0 100.0
re 5.5 ± 2.0 550.0 ± 200.0
R4tot 1000.0 1000.0

5
The values of k and R4tot were the same as for the non-feedback models and
the final values of all parameters are listed in Table 2.
Except for when the concentration of nuclear MYB3R4 is very low, for both
import and export models the addition of the feedback results in faster and
stronger MYB3R4 nuclear re-localisation in response to CYT pulse. Without
feedback, the import model seemed to have a slightly better fit with data.
However, with the feedback, its response to CYT is both a bit faster and stronger
than experimental observations. The export model, on the other hand, benefits
from the addition of feedback and reproduces the target data very well. We take
this to support our hypothesis that the CYT peak shortly before cell division
acts -through a positive feedback conferred by IMPA3/6- to shift the localisation
of MYB3R4 from the cytosol to the nucleus where the MYB3R4 can trigger
downstream signalling.

6
References and Notes
1. E. M. Meyerowitz, Genetic control of cell division patterns in developing plants. Cell 88, 299–
308 (1997). doi:10.1016/S0092-8674(00)81868-1 Medline
2. T. Greb, J. U. Lohmann, Plant stem cells. Curr. Biol. 26, R816–R821 (2016).
doi:10.1016/j.cub.2016.07.070 Medline
3. C. O. Miller, F. Skoog, M. H. Von Saltza, F. M. Strong, Kinetin, a cell division factor from
deoxyribonucleic acid. J. Am. Chem. Soc. 77, 1392 (1955). doi:10.1021/ja01610a105
4. G. E. Schaller, I. H. Street, J. J. Kieber, Cytokinin and the cell cycle. Curr. Opin. Plant Biol.
21, 7–15 (2014). doi:10.1016/j.pbi.2014.05.015 Medline
5. G. E. Schaller, A. Bishopp, J. J. Kieber, The yin-yang of hormones: Cytokinin and auxin
interactions in plant development. Plant Cell 27, 44–63 (2015).
doi:10.1105/tpc.114.133595 Medline
6. L. Moubayidin, R. Di Mambro, S. Sabatini, Cytokinin-auxin crosstalk. Trends Plant Sci. 14,
557–562 (2009). doi:10.1016/j.tplants.2009.06.010 Medline
7. L. De Veylder, T. Beeckman, D. Inzé, The ins and outs of the plant cell cycle. Nat. Rev. Mol.
Cell Biol. 8, 655–665 (2007). doi:10.1038/nrm2227 Medline
8. A. P. Mähönen, M. Bonke, L. Kauppinen, M. Riikonen, P. N. Benfey, Y. Helariutta, A novel
two-component hybrid molecule regulates vascular morphogenesis of the Arabidopsis
root. Genes Dev. 14, 2938–2943 (2000). doi:10.1101/gad.189200 Medline
9. T. Inoue, M. Higuchi, Y. Hashimoto, M. Seki, M. Kobayashi, T. Kato, S. Tabata, K.
Shinozaki, T. Kakimoto, Identification of CRE1 as a cytokinin receptor from
Arabidopsis. Nature 409, 1060–1063 (2001). doi:10.1038/35059117 Medline
10. I. Hwang, J. Sheen, Two-component circuitry in Arabidopsis cytokinin signal transduction.
Nature 413, 383–389 (2001). doi:10.1038/35096500 Medline
11. J. J. Kieber, G. E. Schaller, Cytokinin signaling in plant development. Development 145,
dev149344 (2018). doi:10.1242/dev.149344 Medline
12. G. V. Reddy, M. G. Heisler, D. W. Ehrhardt, E. M. Meyerowitz, Real-time lineage analysis
reveals oriented cell divisions associated with morphogenesis at the shoot apex of
Arabidopsis thaliana. Development 131, 4225–4237 (2004). doi:10.1242/dev.01261
Medline
13. T. Werner, V. Motyka, V. Laucou, R. Smets, H. Van Onckelen, T. Schmülling, Cytokinin-
deficient transgenic Arabidopsis plants show multiple developmental alterations
indicating opposite functions of cytokinins in the regulation of shoot and root meristem
activity. Plant Cell 15, 2532–2550 (2003). doi:10.1105/tpc.014928 Medline
14. K. Miyawaki, P. Tarkowski, M. Matsumoto-Kitano, T. Kato, S. Sato, D. Tarkowska, S.
Tabata, G. Sandberg, T. Kakimoto, Roles of Arabidopsis ATP/ADP
isopentenyltransferases and tRNA isopentenyltransferases in cytokinin biosynthesis.
Proc. Natl. Acad. Sci. U.S.A. 103, 16598–16603 (2006). doi:10.1073/pnas.0603522103
Medline
15. M. Higuchi, M. S. Pischke, A. P. Mähönen, K. Miyawaki, Y. Hashimoto, M. Seki, M.
Kobayashi, K. Shinozaki, T. Kato, S. Tabata, Y. Helariutta, M. R. Sussman, T.
Kakimoto, In planta functions of the Arabidopsis cytokinin receptor family. Proc. Natl.
Acad. Sci. U.S.A. 101, 8821–8826 (2004). doi:10.1073/pnas.0402887101 Medline
16. I. Bartrina, E. Otto, M. Strnad, T. Werner, T. Schmülling, Cytokinin regulates the activity of
reproductive meristems, flower organ size, ovule formation, and thus seed yield in
Arabidopsis thaliana. Plant Cell 23, 69–80 (2011). doi:10.1105/tpc.110.079079 Medline
17. M. Ito, Conservation and diversification of three-repeat Myb transcription factors in plants. J.
Plant Res. 118, 61–69 (2005). doi:10.1007/s10265-005-0192-8 Medline
18. M. Umeda, S. S. Aki, N. Takahashi, Gap 2 phase: Making the fundamental decision to divide
or not. Curr. Opin. Plant Biol. 51, 1–6 (2019). doi:10.1016/j.pbi.2019.03.001 Medline
19. N. Haga, K. Kato, M. Murase, S. Araki, M. Kubo, T. Demura, K. Suzuki, I. Müller, U. Voss,
G. Jürgens, M. Ito, R1R2R3-Myb proteins positively regulate cytokinesis through
activation of KNOLLE transcription in Arabidopsis thaliana. Development 134, 1101–
1110 (2007). doi:10.1242/dev.02801 Medline
20. N. Haga, K. Kobayashi, T. Suzuki, K. Maeo, M. Kubo, M. Ohtani, N. Mitsuda, T. Demura,
K. Nakamura, G. Jürgens, M. Ito, Mutations in MYB3R1 and MYB3R4 cause pleiotropic
developmental defects and preferential down-regulation of multiple G2/M-specific genes
in Arabidopsis. Plant Physiol. 157, 706–717 (2011). doi:10.1104/pp.111.180836 Medline
21. K. Kobayashi, T. Suzuki, E. Iwata, N. Nakamichi, T. Suzuki, P. Chen, M. Ohtani, T. Ishida,
H. Hosoya, S. Müller, T. Leviczky, A. Pettkó-Szandtner, Z. Darula, A. Iwamoto, M.
Nomoto, Y. Tada, T. Higashiyama, T. Demura, J. H. Doonan, M.-T. Hauser, K.
Sugimoto, M. Umeda, Z. Magyar, L. Bögre, M. Ito, Transcriptional repression by
MYB3R proteins regulates plant organ growth. EMBO J. 34, 1992–2007 (2015).
doi:10.15252/embj.201490899 Medline
22. O. S. Lau, K. A. Davies, J. Chang, J. Adrian, M. H. Rowe, C. E. Ballenger, D. C. Bergmann,
Direct roles of SPEECHLESS in the specification of stomatal self-renewing cells.
Science 345, 1605–1609 (2014). doi:10.1126/science.1256888 Medline
23. P. Chen, H. Takatsuka, N. Takahashi, R. Kurata, Y. Fukao, K. Kobayashi, M. Ito, M. Umeda,
Arabidopsis R1R2R3-Myb proteins are essential for inhibiting cell division in response to
DNA damage. Nat. Commun. 8, 635 (2017). doi:10.1038/s41467-017-00676-4 Medline
24. N. Takahashi, N. Ogita, T. Takahashi, S. Taniguchi, M. Tanaka, M. Seki, M. Umeda, A
regulatory module controlling stress-induced cell cycle arrest in Arabidopsis. eLife 8,
e43944 (2019). doi:10.7554/eLife.43944 Medline
25. C. Bourbousse, N. Vegesna, J. A. Law, SOG1 activator and MYB3R repressors regulate a
complex DNA damage network in Arabidopsis. Proc. Natl. Acad. Sci. U.S.A. 115,
E12453–E12462 (2018). doi:10.1073/pnas.1810582115 Medline
26. W. Wang, P. Sijacic, P. Xu, H. Lian, Z. Liu, Arabidopsis TSO1 and MYB3R1 form a
regulatory module to coordinate cell proliferation with differentiation in shoot and root.
Proc. Natl. Acad. Sci. U.S.A. 115, E3045–E3054 (2018). doi:10.1073/pnas.1715903115
Medline
27. M. Menges, S. M. de Jager, W. Gruissem, J. A. Murray, Global analysis of the core cell cycle
regulators of Arabidopsis identifies novel genes, reveals multiple and highly specific
profiles of expression and provides a coherent model for plant cell cycle control. Plant J.
41, 546–566 (2005). doi:10.1111/j.1365-313X.2004.02319.x Medline
28. W. Yang, R. Wightman, E. M. Meyerowitz, Cell cycle control by nuclear sequestration of
CDC20 and CDH1 mRNA in plant stem cells. Mol. Cell 68, 1108–1119.e3 (2017).
doi:10.1016/j.molcel.2017.11.008 Medline
29. J. Heyman, L. De Veylder, The anaphase-promoting complex/cyclosome in control of plant
development. Mol. Plant 5, 1182–1194 (2012). doi:10.1093/mp/sss094 Medline
30. M. Fornerod, M. Ohno, M. Yoshida, I. W. Mattaj, CRM1 is an export receptor for leucine-
rich nuclear export signals. Cell 90, 1051–1060 (1997). doi:10.1016/S0092-
8674(00)80371-2 Medline
31. A. R. Jones, M. Forero-Vargas, S. P. Withers, R. S. Smith, J. Traas, W. Dewitte, J. A. H.
Murray, Cell-size dependent progression of the cell cycle creates homeostasis and
flexibility of plant cell size. Nat. Commun. 8, 15060 (2017). doi:10.1038/ncomms15060
Medline
32. I. Meier, E. J. Richards, D. E. Evans, Cell biology of the plant nucleus. Annu. Rev. Plant
Biol. 68, 139–172 (2017). doi:10.1146/annurev-arplant-042916-041115 Medline
33. K. Palma, Y. Zhang, X. Li, An importin α homolog, MOS6, plays an important role in plant
innate immunity. Curr. Biol. 15, 1129–1135 (2005). doi:10.1016/j.cub.2005.05.022
Medline
34. N. Nishinari, K. Syōno, Induction of cell division synchrony and variation of cytokinin
contents through the cell cycle in tobacco cultured cells. Plant Cell Physiol. 27, 147–153
(1986).
35. P. Redig, O. Shaul, D. Inzé, M. Van Montagu, H. Van Onckelen, Levels of endogenous
cytokinins, indole-3-acetic acid and abscisic acid during the cell cycle of synchronized
tobacco BY-2 cells. FEBS Lett. 391, 175–180 (1996). doi:10.1016/0014-5793(96)00728-
4 Medline
36. P. Dobrev, V. Motyka, A. Gaudinová, J. Malbeck, A. Trávníčková, M. Kamínek, R.
Vaňková, Transient accumulation of cis-and trans-zeatin type cytokinins and its relation
to cytokinin oxidase activity during cell cycle of synchronized tobacco BY-2 cells. Plant
Physiol. Biochem. 40, 333–337 (2002). doi:10.1016/S0981-9428(02)01383-9
37. C. Riou-Khamlichi, R. Huntley, A. Jacqmard, J. A. Murray, Cytokinin activation of
Arabidopsis cell division through a D-type cyclin. Science 283, 1541–1544 (1999).
doi:10.1126/science.283.5407.1541 Medline
38. F. Laureys, W. Dewitte, E. Witters, M. Van Montagu, D. Inzé, H. Van Onckelen, Zeatin is
indispensable for the G2-M transition in tobacco BY-2 cells. FEBS Lett. 426, 29–32
(1998). doi:10.1016/S0014-5793(98)00297-X Medline
39. B. Wybouw, B. De Rybel, Cytokinin–a developing story. Trends Plant Sci. 24, 177–185
(2019). doi:10.1016/j.tplants.2018.10.012 Medline
40. K. Wulfetange, S. N. Lomin, G. A. Romanov, A. Stolz, A. Heyl, T. Schmülling, The
cytokinin receptors of Arabidopsis are located mainly to the endoplasmic reticulum.
Plant Physiol. 156, 1808–1818 (2011). doi:10.1104/pp.111.180539 Medline
41. K. Caesar, A. M. Thamm, J. Witthöft, K. Elgass, P. Huppenberger, C. Grefen, J. Horak, K.
Harter, Evidence for the localization of the Arabidopsis cytokinin receptors AHK3 and
AHK4 in the endoplasmic reticulum. J. Exp. Bot. 62, 5571–5580 (2011).
doi:10.1093/jxb/err238 Medline
42. P. Barbier de Reuille, A.-L. Routier-Kierzkowska, D. Kierzkowski, G. W. Bassel, T.
Schüpbach, G. Tauriello, N. Bajpai, S. Strauss, A. Weber, A. Kiss, A. Burian, H.
Hofhuis, A. Sapala, M. Lipowczan, M. B. Heimlicher, S. Robinson, E. M. Bayer, K.
Basler, P. Koumoutsakos, A. H. K. Roeder, T. Aegerter-Wilmsen, N. Nakayama, M.
Tsiantis, A. Hay, D. Kwiatkowska, I. Xenarios, C. Kuhlemeier, R. S. Smith,
MorphoGraphX: A platform for quantifying morphogenesis in 4D. eLife 4, 05864 (2015).
doi:10.7554/eLife.05864 Medline
43. D. L. Lindsay, V. K. Sawhney, P. C. Bonham-Smith, Cytokinin-induced changes in
CLAVATA1 and WUSCHEL expression temporally coincide with altered floral
development in Arabidopsis. Plant Sci. 170, 1111–1117 (2006).
doi:10.1016/j.plantsci.2006.01.015
44. W. Yang, C. Schuster, N. Prunet, Q. Dong, B. Landrein, R. Wightman, E. M. Meyerowitz,
Visualization of protein coding, long noncoding, and nuclear ds by fluorescence in situ
hybridization in sections of shoot apical meristems and developing flowers. Plant
Physiol. 182, 147–158 (2020). doi:10.1104/pp.19.00980 Medline
45. A. V. Gendrel, Z. Lippman, C. Yordan, V. Colot, R. A. Martienssen, Dependence of
heterochromatic histone H3 methylation patterns on the Arabidopsis gene DDM1.
Science 297, 1871–1873 (2002). doi:10.1126/science.1074950 Medline
46. R. Blecher-Gonen, Z. Barnett-Itzhaki, D. Jaitin, D. Amann-Zalcenstein, D. Lara-Astiaso, I.
Amit, High-throughput chromatin immunoprecipitation for genome-wide mapping of in
vivo protein-DNA interactions and epigenomic states. Nat. Protoc. 8, 539–554 (2013).
doi:10.1038/nprot.2013.023 Medline
47. B. Langmead, C. Trapnell, M. Pop, S. L. Salzberg, Ultrafast and memory-efficient alignment
of short DNA sequences to the human genome. Genome Biol. 10, R25 (2009).
doi:10.1186/gb-2009-10-3-r25 Medline
48. Y. Zhang, T. Liu, C. A. Meyer, J. Eeckhoute, D. S. Johnson, B. E. Bernstein, C. Nusbaum, R.
M. Myers, M. Brown, W. Li, X. S. Liu, Model-based analysis of ChIP-Seq (MACS).
Genome Biol. 9, R137 (2008). doi:10.1186/gb-2008-9-9-r137 Medline
49. T. L. Bailey, M. Boden, F. A. Buske, M. Frith, C. E. Grant, L. Clementi, J. Ren, W. W. Li,
W. S. Noble, MEME SUITE: Tools for motif discovery and searching. Nucleic Acids Res.
37, W202–W208 (2009). doi:10.1093/nar/gkp335 Medline
50. W. Huang, R. Loganantharaj, B. Schroeder, D. Fargo, L. Li, PAVIS: a tool for Peak
Annotation and Visualization. Bioinformatics 29, 3097–3099 (2013).
doi:10.1093/bioinformatics/btt520 Medline
51. J. T. Robinson, H. Thorvaldsdóttir, W. Winckler, M. Guttman, E. S. Lander, G. Getz, J. P.
Mesirov, Integrative genomics viewer. Nat. Biotechnol. 29, 24–26 (2011).
doi:10.1038/nbt.1754 Medline
52. F. Ramírez, F. Dündar, S. Diehl, B. A. Grüning, T. Manke, deepTools: A flexible platform
for exploring deep-sequencing data. Nucleic Acids Res. 42, W187–W191 (2014).
doi:10.1093/nar/gku365 Medline
53. F. Ramírez, D. P. Ryan, B. Grüning, V. Bhardwaj, F. Kilpert, A. S. Richter, S. Heyne, F.
Dündar, T. Manke, deepTools2: A next generation web server for deep-sequencing data
analysis. Nucleic Acids Res. 44, W160–W165 (2016). doi:10.1093/nar/gkw257 Medline
54. R. Patro, G. Duggal, M. I. Love, R. A. Irizarry, C. Kingsford, Salmon provides fast and bias-
aware quantification of transcript expression. Nat. Methods 14, 417–419 (2017).
doi:10.1038/nmeth.4197 Medline
55. S. X. Ge, D. Jung, R. Yao, ShinyGO: A graphical gene-set enrichment tool for animals and
plants. Bioinformatics 36, 2628–2629 (2020). doi:10.1093/bioinformatics/btz931 Medline
56. R. P. Hellens, A. C. Allan, E. N. Friel, K. Bolitho, K. Grafton, M. D. Templeton, S.
Karunairetnam, A. P. Gleave, W. A. Laing, Transient expression vectors for functional
genomics, quantification of promoter activity and RNA silencing in plants. Plant
Methods 1, 13 (2005). doi:10.1186/1746-4811-1-13 Medline