Cyclin A participates in the TSO1-MYB3R1 regulatory module to maintain shoot meristem size and fertility in Arabidopsis

Plant meristems are responsible for generating all above and below ground tissues. The identification of gene regulatory circuitries that confer and maintain the ‘stem cell’ property in plant meristems is therefore of fundamental importance. In plant shoot apical meristem (SAM), the WUSCHEL (WUS)-CLAVATA (CLV) negative-feedback loop maintains the stem cell pool and limits the meristem size. However, beyond the WUS-CLV3 pathway, there have been limited studies of other regulatory pathways for SAM regulation, and little is known about how the plant meristem activities are integrated with the cell cycle regulation.

The DREAM/MMB complex is a master cell cycle regulator in both animals and plants (Kobayashi et al., 2015; Sadasivam and DeCaprio, 2013). The animal DREAM complex consists of retinoblastoma (RB)-like proteins (P107 and P130), E2Fs and their dimerization partners DP1-3 and the MuvB core (LIN9, LIN37, LIN52, LIN54 and RBBP4). In animal quiescent cells, the DREAM complex prevents cells from entering the cell cycle while in dividing cells, the MuvB core associates with B-Myb (also called MYBL2) instead of the Rb-like proteins to form the MMB complex that promotes the G2/M phase (Fischer and Müller, 2017). Recent studies revealed that the DREAM/MMB complexes also exist in plants. The Arabidopsis genome possesses essentially all the homologous genes of the DREAM/MMB complex components, and the corresponding plant homologs can form similar complexes (Lang et al., 2021; Ning et al., 2020). Intriguingly, plants have more copies of each complex component. For example, there are eight Arabidopsis homologs of LIN54, named as TSO1, SOL1, SOL2, TCX4, TCX5, TCX6, TCX7 and TCX8, all of which encode two cysteine-rich CXC motifs separated by a linker region (Andersen et al., 2007; Hauser et al., 2000; Song et al., 2000). TCX5/6-containing DREAM complexes in Arabidopsis act to preclude DNA hypermethylation and prevent excessive cell proliferation (Ning et al., 2020), and the Arabidopsis SOL1 and SOL2 are required for efficient cell fate transitions in the stomata lineage (Simmons et al., 2019). The existence of multiple paralogs in plant genomes could contribute to different isoforms of the DREAM/MMB complex acting in different developmental contexts.

In animals, disruptions of the DREAM/MMB complex usually lead to cancer. The loss of both retinoblastoma protein (pRb) and Rb-like (p107/p130) induces retinoblastoma in mice (Wu et al., 2017), while overexpression of B-Myb is found in many cancers, including breast cancer and colorectal cancer, and is often associated with poor outcomes (Musa et al., 2017). Although plants do not suffer from cancer, plant mutants defective in the DREAM/MMB components exhibit shoot meristem fasciation: an over-proliferation of the SAM. In Arabidopsis, the tso1-1, an antimorphic mutation in the Arabidopsis homolog of LIN54, shows fasciated SAMs, and tso1-1 mutant flowers fail to differentiate into floral organs, and are therefore sterile (Liu et al., 1997; Sijacic et al., 2011; Wang et al., 2018). Meanwhile, the cells in the tso1-1 root apical meristem (RAM) exit the cell cycle early, resulting in a short root phenotype, suggesting TSO1 has different and opposite effects on shoot and root meristems. tso1-3, a loss-of-function allele, on the other hand, has weak phenotypes; it develops normal flowers and normal length root but exhibits reduced fertility (Hauser et al., 2000; Sijacic et al., 2011; Wang et al., 2018). In Arabidopsis, SOL2 is highly similar to TSO1, not only in sequence but also in expression; tso1-3; sol2-1 double loss-of-function mutants exhibit a synergistic genetic interaction and a tso1-1-like phenotype. Therefore, the tso1-1 antimorphic allele may inactivate TSO1 as well as SOL2 to cause its severe phenotype (Sijacic et al., 2011). These different alleles of TSO1 in Arabidopsis provide a useful tool for dissecting DREAM/MMB function in the context of plant meristem regulation.

To identify genes that may act in the same pathway as TSO1, we previously conducted a genetic screen for suppressors of tso1-1 (Wang et al., 2018). Seeds of tso1-1 containing an inducible TSO1 (35S::TSO1-GR) were mutagenized. When applied with dexamethasone (DEX), tso1-1; 35S::TSO1-GR M1 plants were able to overcome sterility and gave rise to M2 progeny, which were screened for suppressors. Thirty-two suppressors from the screen were found to reside in MYB3R1, which encodes one of the five Arabidopsis homologs of B-Myb. The work established that the wild-type TSO1 activity is required to repress MYB3R1 expression to prevent SAM overproliferation (Wang et al., 2018). However, questions concerning how the TSO1-MYB3R1 module regulates SAM cell proliferation and how the module interacts with cell cycle machinery remain unanswered.

In this study, we characterized a second suppressor locus of tso1-1 from the same genetic screen. We showed that mutations in an A-type cyclin named CYCA3;4 suppressed the abnormal floral organ development, shoot fasciation and sterility of tso1-1 to certain degrees. Additionally, ectopic and more abundant CYCA3;4-GUS proteins were found in tso1-1 mutant inflorescence, and CYCA3;4 overexpression enhanced the tso1-3 phenotype. The work reveals the function of a specific cyclin A in shoot meristem regulation and directly integrates cell cycle machinery with the TSO1-MYB3R1 regulatory module.

In order to identify new components of the TSO1 regulatory pathway, a second suppressor locus, defined by a single allele, A144, was analyzed. A144 suppresses tso1-1 defects in floral organ differentiation, fertility and shoot meristem size (Fig. 1A). Although tso1-1 plants develop highly abnormal carpels (Fig. 1A) that lead to no seeds per carpel (Fig. 1B), tso1-1; A144 double plants form normal carpels (Fig. 1A) that yield 1-10 seeds per silique (Fig. 1B). Furthermore, 100% tso1-1 mutant shoot meristems are fascinated (Fig. 1B), as shown by more floral buds per inflorescence (Fig. 1A). This meristem fasciation defect of tso1-1 is completely suppressed in tso1-1; A144 double mutants (Fig. 1A,B). In the weaker allele of tso1-3, even though the carpels develop properly, their fertilities are low, which is indicated by the small siliques; however, tso1-3; A144 double mutants form longer siliques than tso1-3 single mutants (Fig. 1A).

To isolate the gene defined by the A144 suppressor, a F2 mapping population was created and sequenced. Using the SIMPLE mapping pipeline (Wachsman et al., 2017), A144 was mapped to an A-type cyclin gene CYCA3;4 on chromosome 1 (Fig. S1). A G-to-A mutation occurs in the last nucleotide of the first exon of CYCA3;4 and is synonymous (E-to-E) (Fig. 1C). RT-PCR was used to examine whether this mutation affects the splicing of CYCA3;4 transcripts. A CYCA3;4 transcript of wild-type size was not detected in the A144 plants, but two transcripts of aberrant sizes were detected (Fig. 1D). Sequence analysis of the RT-PCR products showed that the longer aberrant transcript retained the first intron whereas the shorter aberrant transcript used a cryptic splice donor site in the first exon. As both the aberrant transcripts contained premature stop codons, A144 likely caused a non-functional CYCA3;4.

To confirm that the mutated CYCA3;4 is indeed the causal mutation for A144, we conducted both complementation tests and CRISPR/Cas9 knockout of CYCA3;4. The genomic sequence of CYCA3;4 (gCYCA3;4; from ∼1.4 kb upstream to the end of 3'UTR) was transformed into the A144 plants containing the tso1-1 mutation. If the transgene rescues A144, the resulting transgenic plants should regain the tso1-1 mutant phenotype, which was indeed observed, including meristem fasciation and complete sterility (Fig. S2). Second, a CRISPR/Cas9 construct with a gRNA targeting the first exon of CYCA3;4 was transformed into tso1-1; 35S::TSO1-GR plants. Several T₁ plants showed suppressed phenotypes even in the absence of DEX application (Fig. 2A). Sequencing of the CYCA3;4 locus in these T1 plants showed either homozygous or biallelic mutations in CYCA3;4 (Fig. 2E). Detailed characterization of a tso1-1; cyca3;4^CR3 homozygous line showed that these mutant plants formed normal shoots and carpels, and developed somewhat elongated siliques containing 0-5 seeds per silique (Fig. 2A,B; Fig. 1B). Therefore, CRISPR-mediated knockouts of CYAC3;4 also suppress tso1-1, although the suppression of fertility, like A144, is incomplete. Together, these data strongly support that, in addition to MYB3R1, CYCA3;4 defines a second suppressor locus of tso1-1.

CRISPR/Cas9-directed knockout of CYCA3;4 allowed us to investigate whether the cyca3;4 mutations could suppress the tso1-1 short root phenotype without worries of background mutations caused by the EMS mutagenesis. Both tso1-1 and tso1-1; cyca3;4^CR3 plants had similar root length, which is shorter than that of wild-type Ler (Fig. 2C,D), suggesting that mutations in CYCA3;4 do not suppress the short root phenotype of tso1-1. It remains to be determined whether this tissue-specific suppression of tso1-1 by the cyca3;4 mutation is due to tissue-specific function of different members of the CYCA3 family.

Among the 50 putative cyclin genes in the genome, categorized into CYCA1-3, CYCB1-3, CYCC, CYCD1-7, CYCH, CYCL, CYCP1-4 and CYCT classes (Menges et al., 2005), CYCA3 genes are believed to govern the G1-S transition and hence resemble the E-type cyclin in animals (Yu et al., 2003). However, the expression profiles of the four Arabidopsis CYCA3s vary greatly, suggesting potentially diverse functions (Takahashi et al., 2010; Willems et al., 2020). CYCA3;3 (AT1G47220) is a meiosis-specific cyclin (Bulankova et al., 2013), and the transcripts of CYCA3;1 (AT5G43080) and CYCA3;2 (AT1G47210) peak in the G1/S phase (Takahashi et al., 2010). CYCA3;4 (AT1G47230), on the other hand, has been shown to be expressed throughout the cell cycle based on RT-PCR analysis of aphidicolin synchronized cell lines (Takahashi et al., 2010). More recently, CYCA3;4-GUS protein accumulation was examined in synchronized root tip by hydroxyurea treatment and shown to accumulate prominently in the G2/M phase (Willems et al., 2020). A loss-of-function cyca3;4-1 was previously identified as a suppressor of ccs52a-1, and two T-DNA insertion lines (cyca3;4-2 and cyca3;4-3) showed strong reductions of transcripts but no obvious root or shoot phenotype (Willems et al., 2020). A144 is therefore named as cyca3;4-4.

To understand the mechanism of tso1 suppression by cyca3;4, we constructed and analyzed a translational reporter of CYCA3;4 (pCYCA3;4-gCYCA3;4-GUS), where a 1.4 kb promoter plus the coding region of CYCA3;4 was fused to GUS. The construct was transformed into plants heterozygous for tso1-1, and seven independent transgenic lines were analyzed. Reporter GUS expression was compared among the T₂ siblings from the same T₁ parent; these T₂ siblings are either tso1-1 or wild type (+/+ or tso1-1/+). In the T₂ wild-type inflorescences, light and consistent blue staining was observed in the young floral buds (Fig. 3A,B). At floral stages 9-10, the locules of anthers show intense blue staining (white arrow in Fig. 3A); at stages 10-12, intense staining occurs in ovules inside the gynoecium as well as stigma (see inset of Fig. 3A). Therefore, the CYCA3;4 protein is at the highest level in actively dividing germ cells. In tso1-1 mutant inflorescences, young floral buds showed significantly stronger and ectopic GUS staining in young floral organ primordia and floral meristems (Fig. 3C,D). In tso1-1 plants with a strong phenotype, including meristem fasciation and a lack of floral organ differentiation, strong, punctate and ectopic staining were observed in young floral meristems (Fig. 3E,F). As tso1-1 mutants do not form well differentiated stamen or carpels, we do not see strong blue staining in gynoecium or anthers.

Although tso1-1 mutant inflorescences showed an increased and ectopic CYCA3;4-GUS protein accumulation, we did not observe a corresponding increase in CYCA3;4 transcript levels in tso1-1 SAM (Fig. S3), suggesting that the tso1-1 mutation may affect CYCA3;4 protein levels indirectly. Nevertheless, the resulting increased and ectopic CYCA3;4 protein activity may excessively promote cell proliferation in tso1-1 mutant inflorescences. A second and distinct interpretation is that the increased and ectopic CYCA3;4 protein accumulation could be a consequence of tso1-1 mutant cells that are unable to complete the cell cycle and arrest at cell cycle phases that express CYCA3;4 proteins. In this second scenario, stronger CYCA3;4 protein accumulation could be simply a byproduct of the cell division defect of tso1-1.

We also compared the CYCA3;4 reporter expression in the roots. Consistent with previous published data (Willems et al., 2020), the CYCA3;4-GUS fusion proteins are located in the meristematic zone and the transition zone in all cell layers (Fig. 3G). In the elongation zone, CYCA3;4-GUS fusion is restricted in the stele. Similar to the wild type, the CYCA3;4-GUS fusion proteins are also restricted to the same tissues in tso1-1 roots (Fig. 3G). However, the meristem zone is compressed in tso1-1 root. Furthermore, whereas the CYCA3;4-GUS proteins are slightly more abundant near the transition zone in the wild-type root (bracketed area in the wild-type root, Fig. 3G), CYCA3;4-GUS seems to be more evenly distributed throughout the meristematic zone and lacks a strong staining band at the transition zone (see transition zone marked by arrows, Fig. 3G). In summary, TSO1 does not seem to repress the CYCA3;4-GUS in the root, but may affect its spatial distribution.

If the elevated CYCA3;4 activity in tso1-1 is responsible for some of the tso1-1 mutant phenotypes, overexpressing CYCA3;4 in the weak tso1-3 mutants may enhance the tso1-3 phenotype. To test this, pUBQ10::CYCA3;4 was introduced into tso1-3/+ plants and RT-qPCR showed 7- to 16-fold higher expression of CYCA3;4 in three different pUBQ10::CYCA3;4 transgenic lines (Fig. S4). tso1-3 plants containing the pUBQ10::CYCA3;4 transgene showed a higher number of defective carpels than tso1-3 plants without the transgene (Fig. 4B,C). Two of the three tso1-3 transgenic lines also made fewer seeds per silique than tso1-3 (Fig. 4C). Therefore, overexpressing CYCA3;4 enhances the floral organ defect and the fertility defect of tso1-3.

In contrast to tso1-3 plants, the inflorescences and carpels of wild-type plants with the pUBQ10::CYCA3;4 transgene exhibited no abnormal phenotype (Fig. 4A). One possible explanation is that other partners of CYCA3;4, which might not be mis-expressed in the wild-type background, are needed in order for pUBQ10::CYCA3;4 to confer a mutant phenotype. One such partner could be MYB3R1. Ectopically expressed MYB3R1 in tso1-1 has previously been shown to mediate the majority of the tso1-1 phenotypes (Wang et al., 2018). The ectopically expressed MYB3R1 protein in tso1-1 could rely on CYCA3;4-mediated phosphorylation to become functional (Fig. 4D).

We propose a model that summarizes previous as well as current findings (Fig. 4D). Specifically, TSO1 acts transcriptionally to prevent MYB3R1 expression. In tso1-1 mutants, MYB3R1 is over- and constitutively expressed, and the constitutive MYB3R1 activity underlies the tso1-1 mutant phenotype (Wang et al., 2018). However, in the tso1-1; cyca3;4 double mutants, the constitutive MYB3R1 activity is reduced or lost due to an absence of CYCA3;4 and/or CDK and hence a loss or reduction of phosphorylation of MYB3R1. This explains why merely overexpressing CYCA3;4 in wild type was insufficient to induce tso1-like phenotype (Fig. 4A). This is also supported by our previous experiment (Wang et al., 2018), when we found phosphorylated serine residues at positions 656 and 709 of MYB3R1 by mining the Arabidopsis Protein Phosphorylation Site Database (PhosPhAt 4.0). Phosphomimics on these two residues showed that MYB3R1(S656D) was able to enhance the tso1-3 fertility defect (Wang et al., 2018). Furthermore, S656, which is immediately followed by PVLDRR, matches the minimal CDK consensus phosphorylation site (S/T-P) (Örd et al., 2019), supporting the possibility that MYB3R1-S656 is likely phosphorylated by a CDK. Accordingly, CDKA;1 could be a partner of CYCA3;4 based on early research showing that the Arabidopsis CYCA3;4 can bind CDKA;1 (Van Leene et al., 2010).

In our proposed model, TSO1 is shown to repress CYCA3;4 indirectly. This is because TSO1 encodes a transcriptional regulator and its impact on CYCA3;4 protein is likely to be indirect. Future molecular, genetic and biochemical experiments will be necessary to test different aspects of our proposed model.

Plants were grown on soil (Sungrow) under a 16 h light/8 h dark cycle at 20°C. All mutants used are in Landsberg erecta (Ler) background. tso1-1, tso1-3 and plants heterozygous for tso1-1 or tso1-3 (tso1-1 +/+ sup-5 and tso1-3 +/+sup-5) have been described previously (Hauser et al., 1998; Liu et al., 1997; Sijacic et al., 2011; Wang et al., 2018). cyca3;4-4 (A144) was isolated from an EMS mutagenesis screen of tso1-1; 35S::TSO1-GR (Wang et al., 2018).

All constructs were transformed through floral dip using Agrobacterium strain GV3101. All sequences were cloned from Ler. All relevant primers are listed in Table S1.

For the complementation test construct, the genomic sequence of CYCA3;4 (from the start of the 5′UTR to the end of the 3′UTR plus ∼1.4 kb upstream of the 5′UTR) was PCR amplified with primers (Table S1), cloned into pCR8/GW/TOPO and LR-recombined into pMDC99 (Curtis and Grossniklaus, 2003).

For the CRISPR/Cas9 construct targeting CYCA3;4, the crRNA was designed using the website crispr.dbcls.jp/. Two 19 nucleotide guide sequences (Table S1) were chosen as they had no off-target sites. crRNA1 and crRNA2, respectively, target the first few nucleotides and the last few nucleotides of the first exon. To introduce the two crRNAs into the pHEE401E vector (Wang et al., 2015), cloning PCR was carried out using the pCBT-DT1T2 vector as the template and the PCR product containing the two crRNAs was introduced into pHEE401E via Gibson assembly (Gibson et al., 2009). The construct was transformed into tso1-1; 35S::TSO1-GR plants. Only crRNA2 was able to generate successful knockout of CYCA3;4. No dexamethasone (DEX) was applied when analyzing the phenotype of the transgenic plants.

A 1.4 kb promoter sequence of CYCA3;4 plus the genomic sequence of CYCA3;4 (from the start of 5′UTR to the end of the coding region) were cloned into pCR8/GW/TOPO and recombined into pMDC162 (Curtis and Grossniklaus, 2003) to make the pCYCA3;4::CYCA3;4-GUS translational fusion. The construct was transformed into tso1-1 +/+ sup-5 plants. Seven independent transgenic lines were analyzed. GUS expression pattern was compared between wild-type and tso1-1 T2 sibling plants derived from the same T₁ parent (tso1-1 +/+ sup5).

To overexpress CYCA3;4, the Arabidopsis UBQ10 promoter was cloned from the JH23 vector (Zhou et al., 2021) and used to drive full-length cDNA of CYCA3;4 (pUBQ10::CYCA3;4). pUBQ10::CYCA3;4 was first cloned into pCR8/GW/TOPO by Gibson assembly and LR recombined into pEarleyGate301 (Earley et al., 2006). pUBQ10::CYCA3;4 was introduced into agrobacterium GV3101 and used to floral dip tso1-3 +/+ sup-5 plants. T₂ plants from three independent transgenic lines, 1, 3 and 4, were further analyzed and ∼11 plants per line were scored for phenotypes.

The mapping population was created by crossing A144 (in the tso1-1; 35S::TSO1-GR background) with the parent plant (+/+, tso1-1; 35S::TSO1-GR). Leaf tissues were collected and pooled from 34 suppressed F2 plants and 50 unsuppressed F2 plants, respectively. Genomic DNAs were extracted using the NucleoSpin Plant II Midi Kit (Macherey-Nagel) and then sent for Illumina sequencing (PE-150). The sequencing depth of the suppressed plants was 202 folds. The sequencing depth of the unsuppressed group was 95 folds. The SIMPLE pipeline (Wachsman et al., 2017) was employed for mapping A144 with default settings.

The seeds were sterilized with 70% ethanol and 10% bleach, and then kept in water at 4°C in the dark for 2 days, after which the seeds were planted on ½ MS (RPI) medium and allowed to germinate under dim light environment for 2 days. Once germinated, they were transferred to the growth chamber. Five days post-germination (DPG) roots were used for quantification.

Inflorescences or 5 DPG seedlings were soaked in 90% acetone for 20 min at room temperature, followed by three washes of staining buffer [0.2% Triton X-100, 50 mM NaHPO4 Buffer (pH 7.2), 2 mM potassium ferrocyanide, 2 mM potassium ferricyanide]. They were then stained in a buffer with 2 mM X-Gluc for 3 to 3.5 h. Tissues were cleared with an ethanol series (20%, 35% and 50%) and then stored in 70% ethanol at 4°C before imaging. Subsequently, tissues were embedded and sectioned based on a published protocol (Hollender et al., 2012). Imaging was performed with Zeiss LSM980 and Zeiss Stemi SV 6.

For tso1-3 and pUBQ10::CYCA3;4 individual transgenic lines, the whole inflorescence (three inflorescences per biological replicate, two biological replicates per genotype) were collected. For wild type (L-er), tso1-1 and tso1-1; myb3r1-9 (both tso1-1 and tso1-1; myb3r1-9 possess 35S::TSO1-GR but no DEX treatment), SAMs were dissected and pooled (15-25 SAMs per biological replicate and four biological replicates per genotype).

RNA was extracted with the RNeasy Mini Kit (Qiagen) and cleaned with DNase I or ezDNase (Invitrogen). cDNAs were synthesized using RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific) or SuperScript IV VILO Master Mix (Invitrogen). RT-qPCR experiments were performed on the Bio-Rad CFX 96 machine with PowerUp SYBR Green Master Mix (Thermo Fisher Scientific). Each biological replicate was analyzed in three technical replicates. The TIP41 (AT4G34270) gene was the internal control for RT-qPCR of CYCA3;4 overexpressing lines. The PP2AA3 (AT1G13320) gene was used as the internal control for all other RT-qPCR. Primers are provided in Table S1.

We thank Dr. Umeda for their collaborative efforts, Amy Beavan at the Department of Cell Biology and Molecular Genetics Imaging Core for assistance with the Zeiss LSM980, and Dirk Joldersma for critical reading of the manuscript. Some of the text and figures in this paper formed part of F.W.'s PhD thesis in the Department of Cell Biology and Molecular Genetics at the University of Maryland in 2022.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201405.reviewer-comments.pdf