INTRODUCTION
Plants exhibit remarkable regenerative capabilities
1-4, with somatic cells in specific lineages retaining pluripotency and the potential to regenerate shoots or roots under appropriate conditions
5-11. Plant regeneration is vital for generating genome-edited and transgenic crops, but current methods require improvement to enhance regeneration capacity across diverse species and genotypes
12,13.
Callus, a proliferating, partially dedifferentiated, and developmentally plastic cellular population induced by wounding or tissue culture conditions
14, is critical in regeneration protocols. In the widely used two-step tissue culture system, explants are cultured on auxin-rich callus-inducing medium (CIM) to generate pluripotent callus
14,15, then transferred to cytokinin-rich shoot-inducing medium (SIM) for shoot induction
2,6,16. In
Arabidopsis thaliana, callus originates from pericycle and pericycle-like cells, with auxin playing a pivotal role
17-19. Mutations in
Aberrant Lateral Root Formation 4 (
ALF4) block pericycle division, inhibiting callus formation on the CIM
20-22. Conversely, overexpression of the DNA binding with one finger (DOF) transcription factor
OBF Binding Protein 4 (
OBP4) promotes pericycle proliferation, leading to the formation of callus-like structures at root tips
23. Very long-chain fatty acids (VLCFAs) restrict pericycle competence for callus formation by acting as cell-layer signals
24.
Inactivation of
CALLUS FORMATION RELATED 1 (
CFR1), encoding a rate-limiting enzyme, 3-ketoacyl-CoA synthase, in the VLCFA biosynthesis pathway, allows pericycle cell division and callus formation even in the presence of auxin signaling pathway defects. Calcium signaling also mediates callus formation, where the CALMODULIN IQ-MOTIF CONTAINING PROTEIN (CaM-IQM) complex interacts with auxin signaling repressors in a Ca
2+-dependent manner
25. For instance, CaM6–IAA19 interaction releases IAA19 inhibition on AUXIN RESPONSE FACTORs 7 and 19 (ARF7/19), key transcription factors for callus initiation.
Auxin regulates callus formation via the bZIP59–LBD complex in
A. thaliana26. ARF7/19 induce expression of LBD16, LBD17, LBD18, and LBD29, while auxin stabilizes bZIP59 to enhance interaction with LBDs
27. This complex promotes callus formation by activating cell wall metabolism genes, such as
FAD-BD. Disruptions in bZIP59 or LBDs reduce auxin-induced callus formation. Other factors, including WRKY23 and bHLH041, act downstream of ARF7/19 to establish callus pluripotency
28.
In addition to stimulating cell proliferation of the callus, auxin induces pluripotency on the CIM
29. Transcriptome studies uncover that the callus shares characteristics with root developmental pathways, as root meristem genes like
WUSCHEL RELATED HOMEOBOX 5 (
WOX5),
SCARECROW (
SCR), and
SHORT ROOT (
SHR) are ectopically expressed in callus
22,30. AP2-family transcription factors (e.g., PLETHORA3 (PLT3), PLT5, PLT7) reprogram callus cells into a pluripotent state by activating
PLT1/
2 and
NAC-family genes
CUC1 and
CUC229. Mutations in
PLT3,
PLT5, and
PLT7 resulted in pronounced defects in callus formation and pluripotency. Three WOX-family TF genes also play important roles in pluripotency, with contrasting functions. While WOX11 functions by activating
LBD1631, WOX5 induces the expression of an auxin synthesis enzyme gene,
TRYPTOPHAN AMINOTRANSFERASE OF ARABIDOPSIS 1 (
TAA1), through its interaction with PLT1 and PLT2
32. Blocking these cascades leads to the loss of callus pluripotency on the CIM and impairs shoot formation on the SIM. Conversely, WOX13 negatively regulates shoot regeneration by promoting genes involved in cell expansion and cellular differentiation and repressing those associated with the shoot apical meristem (SAM)
33.
Epigenetic regulation is vital in callus formation
34,35. DNA methylation, chromatin accessibility, histone modifications, and histone variants trigger dramatic dynamic changes during somatic cell reprogramming and pluripotency acquisition
36-42. Mutants associated with DNA methylation and histone modifications display regeneration defects
37,43-45. Not all callus cells are pluripotent, and single-cell RNA sequencing (scRNA-seq) studies have revealed cellular heterogeneity in
A. thaliana32,33,46. Callus cells can be grouped into three cell layers along the direction perpendicular to the vascular bundle of the explant
32. The middle layer, characterized by a transcriptional identity similar to the root quiescent center, can regenerate shoots.
In summary, significant progress has been made in uncovering the mechanisms driving callus induction and pluripotency acquisition on the CIM. However, despite the widespread presence of the middle layer within mature callus, it remains unclear why not all regions are equally effective for shoot regeneration. Addressing this question could further enhance shoot regeneration capacity, benefiting future crop design.
To investigate this, we created a comprehensive spatiotemporal cellular map of
A. thaliana callus at single-cell resolution. We discovered that heterogeneity in cell division patterns leads to asynchronous callus development using approximately 40 cell type-specific fluorescent protein reporter lines and time-lapsed imaging. Notably, early-formed prominent protrusions were identified as ineffective regions predisposed to root development. Leveraging a gene signature associated with these ineffective regions and conducting a chemical screen, we identified U0126 as a compound significantly reducing ineffective callus regions and enhancing shoot regeneration across various dicotyledonous and monocotyledonous plants. Notably, these results establish an innovative chemical optimization framework for boosting regeneration capacity, representing a significant conceptual advancement over traditional trial-and-error approaches
13.
RESULTS
A time-resolved cell atlas of callus at single-cell resolution
Previous single-cell sequencing studies on
A. thaliana callus derived from hypocotyls were limited to a single time point
32,33, providing minimal insight into the gradual emergence of new cell types during callus initiation and development. In this study, we used
A. thaliana hypocotyls as explants and followed the regeneration protocol described by Wu
et al.
38. Explants (CIM0) and calli were sampled at three developmental stages (Fig. 1). On the CIM, pericycle cell division was evident as early as day 1 (CIM1). By day 3 (CIM3), one to five cell layers had formed, and by day 7 (CIM7), prominent callus protrusions were observed. Subsequent transfer of these calli onto the SIM induced shoot formation.
High-quality protoplasts were prepared, and single-cell transcriptome libraries were generated using the 10× Genomics Chromium scRNA-seq platform. The data were analyzed using published protocols (see Materials and Methods). Since tissue culture represents an unnatural growth condition, existing
A. thaliana cell type markers derived from root and shoot scRNA-seq datasets
47-49 may not precisely annotate callus cell types. To address this, we conducted large-scale promoter-reporter analyses. Regulatory regions of top cluster-specific genes were cloned to generate Green Fluorescent Protein (GFP) or DsRed reporter lines. The spatial expression patterns of 39 reporter lines, combined with known cell type markers, allowed us to annotate cell clusters at each time point (Supplementary Figs. S1–S7 and Text S1). Overall, annotations at CIM0 were consistent with prior anatomical characterizations of
A. thaliana hypocotyls (Fig. 1a)
50. At CIM1 and CIM3, wound-responsive cells and nascent callus clusters were readily detected (Fig. 1b, c). By CIM7, in agreement with earlier reports, three distinct callus layers — outer, middle, and inner — were clearly resolved (Fig. 1d)
32,33. Collectively, this time-resolved scRNA-seq analysis produced a comprehensive cellular atlas of
A. thaliana callus, capturing dynamic shifts in cell identity throughout callus development.
Hidden cell heterogeneity revealed by spatial transcriptomics
During our observations, to our surprise, we found that several reporters exhibited spatial heterogeneity at CIM7, which does not simply conform to the three-layer principle of callus organization proposed by Zhai
et al.
32. For instance,
INDOLE-3-ACETIC ACID7 (
IAA7,
AT3G23050) and
AT2G22510 were expressed only in specific regions of the outer layer (Supplementary Fig. S4a), suggesting previously unrecognized spatial diversity. To validate this, we employed
in situ RNA sequencing (IISS) (see Materials and Methods)
51, as commercial spatial transcriptomics platforms like 10× Genomics Visium, Curio Seeker, and Stereo-seq are unsuitable for the small tissue size of
A. thaliana calli
52-55. Based on our cell atlas annotations (Supplementary Table S1), we designed RNA probes for over 100 representative genes and generated a spatial transcriptomic map for the CIM7 callus.
To confirm the accuracy of this spatial map, we examined the localization of several known genes. For example,
ATML1 was expressed in the outer proliferating cell layer, while transcripts of
WOX5,
PLT1,
PLT2,
TAA1, and
SCR were enriched in middle layers, and
SHR was predominantly localized in inner layers (Fig. 2a; Supplementary Fig. S5a)
32,33. In contrast,
PLT3 showed uniform expression across all proliferating cells
56, further corroborating the reliability of our spatial transcriptomics and single-cell annotations.
Through spatial analysis, we identified five novel cell types (N1 to N5) that could not be spatially resolved via scRNA-seq (Fig. 2b, c; Supplementary Fig. S4b). N1 and N2 cells were localized to the outermost layers of distinct callus regions: the flat callus region (FCR) and the arched callus region (ACR), respectively (Fig. 2b, c). While the epidermal marker ATML1 failed to differentiate these populations, LONG-CHAIN ACYL-COA SYNTHETASE 2 (LACS2), GGL12, and CYP77B1 were preferentially expressed in the FCR, whereas 3-KETOACYL-COA SYNTHASE 16 (KCS16, AT4G34250) was principally restricted to the ACR (Fig. 2b; Supplementary Figs. S4b and S5b–d). These findings suggest that the outermost layers of these regions exhibit distinct surface characteristics, potentially reflecting differences in long-chain fatty acid composition. Other marker genes for N2 included IAA7 and ABCB15. While IAA7 was initially expressed in the epidermis and cortex during early callus development, its transcripts predominantly localized to the outermost cell layer of the ACR at CIM7 (Supplementary Fig. S5e, f). ABCB15 displayed a similar expression pattern (Supplementary Fig. S5g).
In contrast, N3 cells were located in the L2–L5 layers of the ACR and marked by genes such as
YUCCA3 (
YUC3),
YUC8, and
CYP79B2 (Fig. 2b, c; Supplementary Figs. S4b, S6a, b). Interestingly, both N2 and N3 were parts of cluster 3 in the CIM7 cell atlas (Fig. 1d), indicating similar cellular states. Remarkably, several root cap genes, including
BEARSKIN 1 (
BRN1),
BRN2, and
SMB57 were highly enriched in cluster 3 at CIM7 and specifically expressed in N2 as shown by spatial transcriptomics (Fig. 2b; Supplementary Figs. S4b and S6c). Furthermore, the top marker genes of cluster 3 at CIM7 were significantly enriched in the root cap cluster from published root cell atlases (Supplementary Fig. S6d)
58. These findings indicate that N2 and N3 in the ACR represent a unique cell type with root cap-like properties. Consequently, we designated N2 and N3 as the "root cap-like cells I and II", and N1 as the "outermost layer cells at the FCR".
N4 cells were identified using
AT2G22510 and
PRO-GLU-LEU|ILE|VAL-PRO-LYS 1 (
PELPK1) as markers (Fig. 2b, c; Supplementary Figs. S4b and S6e–g). At CIM3, these cells (cluster 13 cells at the CIM3, Fig. 1c) were categorized as "overlying ground tissue cells" that are later penetrated by inner proliferating callus cells (Fig. 2b, c; Supplementary Fig. S6e–g). At CIM7, they (cluster 13 cells at the CIM7, Fig. 1d) maintained specific localization within the callus, exhibiting larger volumes compared to adjacent proliferating cells (Supplementary Fig. S6e–g). Notably, these cells were enriched in genes associated with cell death and senescence, including
BIFUNCTIONAL NUCLEASE 1 (
BFN1),
SENESCENCE ASSOCIATED GENE 29 (
SAG29),
RIBONUCLEASE 3 (
RNS3),
METACASPASE 9 (
MC9),
EXITUS 1 (
EXl1), and
DUF679 DOMAIN MEMBRANE PROTEIN 4 (
DMP4) (Supplementary Fig. S6h)
59,60.
A recent study by the Sugimoto and Ikeuchi labs identified "highly expanded globular cells" in the SIM6 callus, which lack regenerative capabilities
33. Due to morphological similarities, we hypothesized that these cells may transcriptionally correspond to N4 at CIM7. Supporting this,
AT2G22510 and
PELPK1 were highly expressed in cluster 17 of the SIM6 cell atlas by Ogura
et al., which matched one of the "highly expanded globular cell" clusters (Supplementary Fig. S6i). Additionally, genes enriched in CIM7 cluster 13 were most highly expressed in cluster 17 of the SIM6 atlas (Supplementary Fig. S6i). Thus, N4 cells, marked by
AT2G22510 and
PELPK1, represent "fully differentiated callus cells" that have lost their pluripotency (Fig. 2c). We therefore named this cell type as the "vacuolized cells".
N5 cells were predominantly located in the inner ACR and vascular tissues of explants and were characterized by genes linked to vascular development, such as SMAX1-LIKE 3 (SMXL3), XYLOGEN PROTEIN 2 (XYP2), HOMEOBOX GENE 8 (HB8), and ACL5 (Fig. 2b, c; Supplementary Figs. S4b and S7a). ACL5 and ABNORMAL SHOOT 5 (ABS5) reporters were initially expressed in the stele's procambium from CIM0 to CIM3 (Supplementary Fig. S7b, c). By CIM3, ACL5 was ectopically expressed within the ACR, and ABS5 expression was also detected in the ACR at CIM7. Although no anatomical features typical of vascular tissue, such as SE or TE formation, were observed in the ACR, ACL5 and ABS5-positive cells exhibited a striped pattern resembling vascular organization (Fig. 2b; Supplementary Fig. S7b, c). These results indicate that N5 represents cells with vascular-like attributes; therefore, we assigned them as "vascular-like cells".
In summary, integrating time-resolved scRNA-seq and spatial transcriptomics revealed previously unrecognized spatial heterogeneity within callus tissues. Notably, callus organization can be categorized not only into outer, middle, and inner layers perpendicular to vascular bundles but also into FCR and ACR regions parallel to vascular bundles (Fig. 2c). In particular, the ACRs exhibit an organization resembling root primordium-like structures based on marker gene expression.
This heterogeneity likely results from asynchronous cell division patterns within explants. To test this hypothesis, we employed a constitutive nuclear localization reporter (
pAT2G18020:H2B-mCherry)
61 to monitor the dynamics of cell layer formation in ACRs and FCRs. The results revealed that cell layer proliferation occurred more rapidly in ACRs than in FCRs (Fig. 2d; Supplementary Fig. S8a). Notably, the formation of two distinct cell layers in ACRs predominantly appeared at CIM2, whereas in FCRs, this process typically began at CIM4 (Fig. 2e; Supplementary Fig. S8a). These observations were further corroborated through live-cell imaging using PlaCCI (Supplementary Fig. S8c, d; Supplementary Video S1)
62, which showed a higher frequency of G2–M reporter signals — indicative of active mitosis — in ACRs during CIM3 to CIM5.
To further assess the initiation of cell division in pericycle cells corresponding to ACRs and FCRs, we utilized Cytrap
63, a synthetic dual-color reporter that simultaneously visualizes the S-to-G2 and G2-to-M transitions of the cell cycle. At CIM1, approximately 87% of ACR regions had already initiated cell division, in contrast to only 28% of FCR regions (Fig. 2f; Supplementary Fig. S8b).
Together, these findings demonstrate that the differential formation of ACRs and FCRs arises from both the distinct onset timing of pericycle cell division and the varying proliferation rates of the resulting callus cells. These two factors collectively drive the more rapid development of ACRs, leading to a greater number of cell layers compared with FCRs.
Identification of ACRs as ineffective callus regions within the callus
The newly identified N2, N3, and N5 cells correspond to dome-shaped ACRs, traditionally considered effective callus areas capable of shoot formation during the SIM stage
32. As such, we tracked ACR fate at CIM7 using wild type explants (Supplementary Fig. S9a–c) and
ACL5,
YUC3, and
ABCB15 reporters (Fig. 3a–c). Surprisingly, we found that these ACRs were committed into roots after transfer to the SIM. Specifically,
ACL5-marked callus regions transitioned into vascular tissues of the root during the SIM stage (Fig. 3c), while
YUC3- and
ABCB15-marked callus regions differentiated into root cap cells (Fig. 3a, b).
A previous study demonstrated that lateral root meristems can be reprogrammed into shoot meristems in a specific timeframe through SIM
64, prompting us to examine whether ACRs retain residual plasticity to transition into shoots during
de novo shoot organogenesis (DNSO). To address this, we performed a more refined, long-term cell fate tracking experiment from SIM0 to SIM17 using the
ACL5 reporter line and spinning disk confocal microscopy (Fig. 3d; Supplementary Fig. S9d, e), which reduced imaging time and phototoxicity. The ACL5
+ callus regions were those exhibiting a striped expression pattern of
ACL5 (Supplementary Fig. S7b). As shown in Fig. 3e, only 2.1% of
ACL5-labeled ACRs developed into shoots (
n = 10/471), while 88.3% developed into roots, consistent with earlier tracking results (Fig. 3a–c; Supplementary Fig. S9a–c). By contrast, simultaneously tracking of the FCRs (i.e., non-ACL5-labeled regions) revealed that FCRs either developed into shoots or retain their original cell fate (Supplementary Fig. S10a–e).
Collectively, these findings indicate that not all callus regions possess pluripotency for shoot regeneration. Contrary to prior assumptions, the majority of multilayered ACRs fail to form shoots and do not transition into shoot meristems under SIM conditions (Fig. 3f). Nevertheless, ACRs retain limited cell fate plasticity, similar to lateral root meristems
64, exhibiting a rare capacity for shoot conversion within a transient developmental window (Fig. 3e).
As outlined above, callus development and ACR formation occur asynchronously. Would transferring calli earlier than CIM7 — before extensive ACR formation — enhance shoot regeneration? To address this, we performed a time-course experiment, transferring calli cultured on the CIM for varying durations to the SIM to evaluate their shoot regeneration potential (Supplementary Fig. S11a). Our results showed that calli earlier than CIM4 lacked shoot regeneration capacity, while those at CIM5 and CIM6 exhibited shoot regeneration rates comparable to CIM7.
These findings indicate that a minimum of four days is required for some callus cells to complete reprogramming and acquire pluripotency. Such prolonged culture time necessary for pluripotency acquisition inevitably lead to ACR formation, reflecting an unresolved conflict between asynchronous callus development and the extended time required to acquire pluripotency. Interestingly, epigenetic profiling results indicate that most ACRs may have adopted a root-like identity while simultaneously losing the open chromatin configuration required for shoot regeneration (Supplementary Text S1 and Fig. S11b–n). Notably, a long-term cell fate tracking experiment revealed that FCRs are likely not an early developmental state of ACRs (Supplementary Text S1 and Fig. S10d, e). Thus, the above results indicate that shoot regeneration competence is only restricted to discrete cellular states within Arabidopsis callus.
Chemical screening identifies compounds to reduce ineffective callus regions
Our findings indicate that ACRs represent ineffective callus regions that preferentially develop into roots on the SIM. Interestingly, we observed a strong negative correlation between the number of roots and shoots formed (Fig. 4a, b). Plants with higher shoot numbers tended to form fewer roots, whereas those with fewer shoots exhibited greater root development. Moreover, root numbers significantly exceeded shoot numbers (Supplementary Fig. S12a), suggesting that reducing the proportion of ACRs within callus tissues could enhance shoot regeneration efficiency. Inspired by advances in chemically induced pluripotent cells in animal research
65, we sought to identify compounds that could effectively diminish ACR proportions through chemical screening.
To establish a high-throughput screening system, we leveraged marker genes associated with ineffective callus regions. As noted, ACL5 expression predominates in ACRs (Fig. 3c). We hypothesized that ACL5 reporter fluorescence intensities could serve as a rapid proxy for assessing compound efficacy, circumventing the need to measure shoot regeneration efficiency directly and improving time use. To facilitate compound application and limit the costs of compounds, we developed a 24-well liquid culture system for calli (Fig. 4c). In this setup, groups of eight explants were cultured per well with liquid CIM supplemented with specific compounds. Preliminary tests confirmed that calli regenerated normally from hypocotyl explants after seven days under liquid CIM conditions in the absence of chemical treatment (Supplementary Fig. S12b). Expression patterns of reporters like WOX5, PLT1, and ATML1 remained unchanged (Supplementary Fig. S12c).
We screened 92 compounds from a chemical library targeting factors involved in signal transduction (Supplementary Fig. S12d and Table S1). In the first round, while most compounds did not alter ACL5 reporter fluorescence, 27 compounds significantly reduced reporter activity while incompletely blocking callus proliferation (Fig. 4d; Supplementary Fig. S12e, f). Validation in a second round identified eight candidate compounds (Fig. 4c, f), from which six were selected for further analysis, including assessments of their effects on shoot regeneration.
Consistent with the results of the liquid culture, candidate compounds significantly reduced ACL5+ cell populations in calli cultured on the CIM compared to DMSO controls (Fig. 4d). However, while all compounds reduced root formation, not all enhanced shoot regeneration (Fig. 4d–f). For instance, Pemetrexed and Vismodegib did not improve shoot regeneration. Conversely, Ki8751, Patupilone, and CYC116 significantly reduced shoot regeneration capacity, with patupilone severely impairing callus proliferation. Among the tested compounds, U0126-EtOH was the sole compound that significantly enhanced shoot regeneration efficiency.
U0126 enhances shoot regeneration efficiency by reducing ineffective callus regions
The screened compound U0126-EtOH contains ethanol molecules. To rule out ethanol’s influence on shoot regeneration, we compared the effects of U0126-EtOH and U0126 (ethanol-free). Both forms actually exhibited equivalent efficacy in inhibiting root development and promoting shoot regeneration (Supplementary Fig. S12g, h). To determine the optimal concentration of U0126, we tested concentrations ranging from 0.5 to 50 μM on the CIM and evaluated shoot regeneration on the SIM (Fig. 5a, b). Results showed peak shoot regeneration at 5 μM, with higher concentrations offering no significant improvement. Thus, 5 μM was chosen as the optimal concentration for subsequent phenotypic and mechanistic studies.
We then sought to investigate the reason why U0126 significantly enhances shoot regeneration efficiency. As shown in Fig. 5c, d, U0126-treated calli exhibited reduced fluorescence signals of the ACL5 reporter and fewer cell layers. These findings suggest that U0126 effectively minimizes ineffective callus regions. Additionally, U0126 treatment substantially reduced YUC3+ and ABCB15+ cell populations in calli (Fig. 5c, d), indicating a broad reduction of ineffective regions rather than specific inhibition of ACL5 expression.
Using the PlaCCI system combined with cell layer tracking, we compared cell division dynamics between mock-treated and U0126-treated explants. U0126 treatment markedly reduced the division rate of callus cells (Fig. 5f–h; Supplementary Fig. S13a and Videos S2 and S3). Because ACR cells divide earlier and more rapidly than FCR cells (Fig. 2d–f; Supplementary Fig. S8), the inhibitory effect of U0126 was more pronounced in ACRs. Detailed analyses of cell cycle phase reporters and 5-ethynyl-2'-deoxyuridine (EdU) incorporation revealed that U0126 primarily induced cell cycle arrest at the G1–S checkpoint or within the S phase by reducing DNA synthesis (Fig. 5h; Supplementary Fig. S13a–c).
To determine whether U0126’s function was solely due to suppressed cell division, we treated calli on CIM with flavopiridol (FVP) and hydroxyurea (HU) — which inhibit cell cycle progression at the G1–S and G2–M phases
56,61, and nocodazole (NOCO), a cyclin inhibitor that blocks the G2-M transition in
A. thaliana callus
56,66. Unlike U0126, FVP, and HU, NOCO was cytotoxic, causing abnormal cell morphology (Supplementary Fig. S13d). Although FVP and HU exhibit a stronger capacity to inhibit cell division than U0126 (Supplementary Figs. S13e, f), they only modestly enhanced shoot regeneration efficiency (Supplementary Fig. S13g).
Previous studies suggest that middle-layer formation in callus tissue is crucial for pluripotency and shoot formation
32. Mutants of middle-layer genes, such as
WOX5,
WOX7,
PLT1, and
PLT2, exhibit reduced shoot regeneration capacity
29,32,56. However, U0126 treatment did not significantly alter the expression patterns of
WOX5 or
PLT1 reporters in calli at CIM7 (Fig. 5e; Supplementary Fig. S14a). Similarly, the expression of outer-layer gene reporters, such as
ATML1 and
KCS10, remained unchanged (Fig. 5e; Supplementary Fig. S13h). Thus, U0126 does not appear to enhance shoot regeneration by directly influencing pluripotency genes on the CIM.
To further investigate U0126’s effects, we cultured wild-type explants on the CIM with U0126 or DMSO (control) and conducted RNA-seq on calli at CIM7 and SIM6 (the stage of SAM commitment). At CIM7, U0126-treated calli exhibited upregulated expression of genes involved in immune response, stress response, toxin metabolism, and endoplasmic reticulum-associated protein degradation pathways (Supplementary Fig. S14a–c). Downregulated genes were linked to mitosis, cell wall synthesis, root and vascular development, and auxin synthesis and transport (Supplementary Fig. S14a–f). At SIM6, U0126-treated calli showed increased expression of SAM-related genes, including WUS, CLAVATA3 (CLV3), and SHOOT MERISTEMLESS (STM), while genes associated with vascular development and transport, such as SWEET11 and UMAMIT20, were downregulated (Supplementary Fig. S14d).
Finally, to test whether U0126 enhances shoot regeneration by accelerating the emergence of WUS
+ cells on the SIM, we used the
pWUS::DsRed-N7 reporter line
67. Calli cultured on the CIM with U0126 and transferred to the SIM did not exhibit earlier WUS
+ cell appearance compared to controls (Supplementary Fig. S14g). Taken together, these findings indicate that U0126 reduces the level of ineffective callus regions by synchronizing callus growth and limiting root-like characteristics within ACRs (Fig. 6h; Supplementary Fig. S14a, c, d).
U0126 reduces ineffective callus regions via the MKK4/5–MPK3/6 pathway
Based on compound annotation and prior plant studies
68,69, U0126 is an inhibitor of MAPK kinases (MPKKs, also known as MEKs) that phosphorylate mitogen-activated protein kinases (MPKs)
70. Its effects are unlikely to reflect off-target activity, as additional MEK inhibitors elicited comparable, though weaker, enhancements in shoot regeneration (Supplementary Fig. S15). In
A. thaliana, 10 MPKKs are classified into four groups based on sequence similarity
71. Two key MPKK–MPK signaling cascades with distinct roles in plant development have been identified: (1) the MKK4/5–MPK3/6 pathway, which regulates processes such as stomatal development, floral organ abscission, lateral root emergence, hormone response, hypoxia signaling, and immunity
72-74; and (2) the MKK1/2–MPK4 cascade, which is associated with plant immunity, cold responses, and aluminum resistance
75.
Comparative studies revealed that the
mkk1/
2 double mutants exhibited lower shoot regeneration efficiency compared to the wild-type (Supplementary Fig. S16c). Since the
mkk4/
5 double mutants are lethal, conditional
mkk4/
5 mutants (
mkk4,5RNAi-est) were used
73. Silencing
MKK4/5 via estradiol (ES) treatment resulted in slender calli resembling those treated with U0126 (Fig. 6a, b; Supplementary Fig. S16a, b). These calli displayed increased shoot regeneration efficiency, reduced root formation (Fig. 6c), fewer ineffective callus protrusions, and diminished fluorescence intensities of the ACL5 reporter (Fig. 6d, e). GO and PO enrichment analyses demonstrated substantial overlap in gene expression changes between U0126-treated and
mkk4,5RNAi-est calli (Supplementary Figs. S14b, c, and S16d–i), indicating that U0126 primarily functions through the MKK4/5 pathway.
To validate these findings, we performed scRNA-seq on mkk4,5RNAi-est calli at CIM7. ES-treated calli exhibited reduced cell heterogeneity, indicative of synchronized growth (Fig. 6f; Supplementary Fig. S16j). Integration analyses revealed a lower proportion of callus cells (clusters 2 and 5) (Supplementary Fig. S16k), enriched in vascular tissue-related genes such as ACL5 and SMXL3 (Supplementary Fig. S16l).
MPK3/6 serves as the principal downstream target of MKK4/5
72-74,76. We therefore examined whether
mpk3/6 mutants display regeneration phenotypes similar to
mkk4,5RNAi-est plants. As homozygous
mpk3/6 double mutants are lethal, we utilized the conditional mutant line
MPK3SR72. Consistent with expectations, silencing
MPK3/6 resulted in the formation of slender calli exhibiting enhanced shoot regeneration efficiency and reduced root formation (Supplementary Fig. S16m), closely resembling the phenotype induced by U0126 treatment. Notably, U0126 treatment did not further enhance the shoot regeneration efficiency of
MPK3SR and
mkk4,5RNAi-est explants (Fig. 6g), indicating that the MKK4/5-MPK3/6 cascade is likely the primary target through which U0126 exerts its effects.
It is worth noting that MKK4/5 and MPK3/6 show no tissue or cellular specificity within the callus (Supplementary Fig. S14h). Differential phosphoproteomic analysis further revealed substantial decreases in the phosphorylation of proteins implicated in epigenetic regulation, the cell cycle, and auxin transport following U0126 treatment (Supplementary Text S1 and Fig. S17).
Collectively, these results demonstrate that U0126 suppresses the formation of ineffective callus regions through an MKK4/5–MPK3/6-associated pathway (Fig. 6h). In agreement with the transcriptomic and scRNA-seq data, U0126 likely acts through multiple MPK3/6 downstream targets — coordinating callus development not only by restricting cell division but also by regulating genes associated with auxin biosynthesis and transport, cell wall remodeling, and vascular tissue differentiation. Nonetheless, the specific MPK3/6 targets mediating these effects require further investigation.
U0126 promotes shoot regeneration in other species
To assess U0126’s efficacy across species, we first tested it within the Brassicaceae family. In
Crucihimalaya wallichii, a close relative of
A. thaliana that has recently been used in studies on perennial life history
77, U0126 treatment during CIM induced slender calli similar to those in
A. thaliana (Supplementary Fig. S18a). These calli produced fewer roots and more shoots upon transfer to the SIM (Fig. 7a; Supplementary Fig. S18b), indicating U0126’s broad applicability in the
Arabidopsis clade. Next, we evaluated U0126 in the oil crop
Brassica napus, testing two varieties with distinct life histories: ZS11 (semi-winter-type), and Tapidor (winter-type). As shown in Fig. 7b and Supplementary Fig. S18f, U0126 significantly enhanced shoot regeneration in both varieties.
To extend the applicability of our method beyond the Brassicaceae family, we conducted regeneration experiments on the plants from other families, including Solanaceae and Poaceae. Notably, we found that U0126 treatment during CIM could also significantly boost shoot regeneration efficiency in Capsicum annuum (Solanaceae) and Setaria viridis and Oryza sativa (Poaceae) (Fig. 7c–e).
It is important to recognize that plant species differ markedly in genotypes and developmental programs. Moreover, tissue culture-based regeneration is a multistep process, and defects during the SIM stage can further constrain regenerative capacity
11. Accordingly, compounds such as U0126, which primarily reduce the proportion of ineffective callus regions during CIM, are unlikely to constitute a universal solution (Supplementary text S1 and Figs. S18–S20). Nevertheless, our results establish a four-step optimization framework that represents a substantial practical advance over traditional trial-and-error approaches, which typically focus on modifying medium composition or adjusting culture duration and temperature (Supplementary Fig. S21)
13. This framework can be readily extended to large-scale compound screening across species or regeneration stages. In the near future, it may facilitate the identification of species- or family-specific compounds or cocktails (see Discussion below), ultimately alleviating persistent bottlenecks in plant regeneration.
DISCUSSION
Our findings indicated that high auxin concentrations in the CIM, along with rapid cell division in ACR, causes a swift cell reprogramming toward root identity and a loss of shoot regeneration potential. In contrast, FCR cells retain pluripotency, potentially due to fewer rounds of cell division, preventing them from reaching the epigenetic threshold required for root identity. U0126 treatment reduced cell division, synchronized callus development, and reduced ACR proportions. However, cell division alone is not the sole determinant of pluripotency. Simply inhibiting cell division does not guarantee that all callus cells acquire or maintain pluripotency (Supplementary Fig. S13e–g). These findings suggest a critical time window for pluripotency acquisition and are consistent with recent observations that stem cell activity is transiently acquired and subsequently lost in the callus
78. Given that numerous studies have documented pronounced heterogeneity in the regenerative capacity of rice callus, with regeneration progressively declining during prolonged culture
14,15, it is highly plausible that similar principles apply to calli derived from other tissue sources. However, critical questions remain unresolved. For instance, do similar ineffective regions exist in calli derived from immature or mature embryos of monocots like rice, wheat, and maize? If so, what genes mark these regions?
The asynchrony in
A. thaliana callus development also warrants further exploration. Could inherent differences in pericycle cell responsiveness to auxin signals explain this phenomenon? Are spatial variations in auxin concentration a contributing factor, or is this asynchrony purely stochastic? To date, long-term cell fate tracking has not revealed definitive patterns. Our findings suggest that U0126 functions through the MKK4/5–MPK3/6 signaling cascade, whose phosphorylation of multiple downstream targets (Fig. 6h), collectively slows cell division, synchronizes developmental progression, and preserves the pluripotent cell state required for efficient shoot regeneration. Interestingly, the MAPK pathway also regulates cell fate manipulation in animals
65,79, hinting at shared principles of cellular pluripotency or totipotency between plants and animals.
This study opens new research avenues akin to the field of stem cells in animals. Chemical reprogramming
65,80,81 may enable cell fate transitions, producing stem cells and diverse plant organs. Notably, other MEK inhibitors were less effective than U0126 (Supplementary Fig. S15). Future efforts could focus on developing more potent compound cocktails to further enhance regeneration efficiency. Indeed, epigenetic inhibitors have shown promise in improving plant regeneration capabilities
82-86. Additionally, recent studies report that small molecules like adenosine monophosphate enhance
A. thaliana regeneration competence, while plant peptides promote regeneration across species
87,88. High-throughput chemical screenings in crops remain challenging but achievable. Identifying marker genes for effective and ineffective callus regions using single-cell multi-omics, spatial transcriptomics, and cell fate tracking will be crucial. Such markers will facilitate compound candidate screening via platforms similar to those described here, ultimately advancing plant tissue culture and regeneration technologies.
MATERIALS AND METHODS
Plant materials and growth conditions
The
Arabidopsis thaliana (L.)
Heynh. plants were used as the experimental model in the study. Columbia-0 (Col-0) ecotype was used as wild type with transgenic lines in this background. The Cytrap, PlaCCI,
pAT2G18020:H2B-mCherry,
MPK3SR,
mkk1/
2 mutants,
pWUS::DsRed-N7, and
mkk4,5RNAi-est lines were described
61-63,67,72,73,75. We crossed
mkk4,5RNAi-est plant with
pACL5::sGFP-N7 plant to obtain
mkk4,5RNAi-est
pACL5::sGFP-N7 line. The seeds for DNSO assays including
C. wallichii,
S. lycopersicum varieties,
B. napus varieties,
S. viridis ME034V,
C. annuum CM334,
Oryza sativa ZH11, and
M. sativa Zhongmu No.1 were either maintained in our lab
77 or obtained from other labs.
Wild type plants were grown in growth room at 21 °C under long-day (16-h light/8-h dark) conditions. The calli were grown in Percival or RUIHUA incubators at specified temperatures under long-day or dark (wrapped in aluminum foil) conditions.
Generation of transgenic plants
To construct the fluorescent reporter lines that mimic the endogenous expression pattern, we defined the necessary upstream and downstream sequences of each cell-cluster enriched gene based on their local chromatin accessibilities. Details regarding the lengths of the upstream and downstream sequences for each gene, along with the primers used for vector construction, are provided in Supplementary Table S1.
The overall procedure involves amplifying the fragments using a high-fidelity enzyme and inserting the upstream and downstream DNA sequences of the gene into the binary vector via homologous recombination. Fragment amplification was carried out using 2× Phanta® Max Master Mix (Vazyme Biotech, Cat No./ID: P515-02). Fragment recovery was conducted using the FastPure Gel DNA Extraction Mini Kit (Vazyme Biotech, Cat No./ID: DC301-01). The recombination reactions were performed using the ClonExpress® MultiS One Step Cloning Kit (Vazyme, Cat No./ID: C113-02) or the ClonExpress® II One Step Cloning Kit (Vazyme, Cat No./ID: C112-02). For plasmid extraction, we used the High Purity Plasmid Mini Kit (Hlingene, Cat No./ID: NG201S).
The resultant binary constructs were then introduced into
Agrobacterium tumefaciens GV3101 (pMP90) by the freeze-thaw method. Transgenic Arabidopsis plants were generated by the floral dipping method
89 and screened with 0.05% glufosinate (Basta) on soil, or 50 mg/L kanamycin on the half-strength Murashige and Skoog (1/2 MS) agar plates.
Shoot regeneration assays for Arabidopsis thaliana and Crucihimalaya wallichii
Shoot regeneration assays of
A. thaliana and
C. wallichii were performed as described previously
38,90. Briefly, seeds were sterilized using 20% bleach and germinated on the 1/2 MS agar plate in the dark. After 7 days, 1 cm long hypocotyls of seedlings were excised, placed on the CIM, and cultured in a Percival incubator at 22 °C under long-day conditions with a light intensity of 45 μmol m
–2·s
–1. For
A. thaliana, the hypocotyls were cultured on the CIM for 7 days, while for
C. wallichii, the duration was extended to 14 days. The resultant calli were transferred to the SIM and cultured in a Percival incubator at 22 °C under long-day conditions with a light intensity of 80 μmol m
–2·s
–1. U0126 was only applied during the CIM stage.
The formulation of the 1/2 MS agar medium is as follows: 2.21 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), and 8 g/L agar (Sangon Biotech, Cat No./ID: A505255-0250), pH 5.8. The formulation of CIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 20 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 2.2 µM 2,4-dicholorophenoxyacetic acid (PhytoTech, Cat No./ID: D295-100mL), 0.2 µM kinetin (Sigma, Cat No./ID: 48130-1G-F), and 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.7. The formulation of SIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 20 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 0.9 µM indole-3-acetic acid (Sigma, Cat No./ID: 45533), 5 µM 2-isopentenyladenine (Sigma, Cat No./ID: D7674), and 1.8 g/L phytagel (Sigma, Cat No./ID: P8169-250G), pH 5.8.
For chemical treatment in A. thaliana, candidate compounds and cell cycle inhibitors (such as FVP, HU, and NOCO) and inducers (such as ES and NA-PP1) are directly added to CIM before it solidifies. The hypocotyl explants were cultured on CIM containing the corresponding compounds or inhibitors for 7 days. The resulting calli were either phenotyped or further transferred to SIM without compounds or inhibitors to assess the efficiency of shoot regeneration. The concentrations used are specified in the figure legends.
Shoot regeneration assays for Brassica napus
The shoot regeneration assays for
B. napus were performed as described previously, with some modifications
91. The seeds were first washed with 75% ethanol for 1 min, then rinsed with sterile water for 5 times. The seeds were then washed with a 10% sodium hypochlorite solution (2 drops of Tween-20 per 50 mL) for 10 min, followed by 5 times of rinsing with sterile water, and placed in the dark at 4 °C. After 1 day, seeds were sown on the 1/2 MS agar plate and cultured in an RUIHUA incubator at 26 °C in the dark for 6–8 days. The hypocotyls of the seedlings were excised to a fixed length of approximately 1.5 cm and cultured on the CIM in a RUIHUA incubator at 26 °C in the dark for 21 days. If
n-hexane (Macklin, Cat No./ID: H811147) treatment is required, the excised hypocotyls were placed on a dish containing
n-hexane for 1 min and then transferred to the CIM. If squeezing treatment is required, the excised hypocotyls were placed on a petri dish, squeezed by ophthalmic tweezers, and then cultured on the CIM. To induce shoot regeneration, the resultant calli were transferred onto the SIM and cultured in a Percival incubator at 26 °C under long-day conditions with a light intensity of 80 μmol m
–2·s
–1. If the varieties ZS11 and Tapidor were used, the SIM was refreshed approximately every 14 days. If the variety Westar was used, no refreshment is necessary. U0126 was only applied during the CIM stage.
The formulation of the 1/2 MS agar medium is the same as that for A. thaliana. The formulation of CIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 18 g/L mannitol (Sigma, Cat No./ID: M4125-500G), 1 mg/L 2,4-dicholorophenoxyacetic acid (PhytoTech, Cat No./ID: D295-100mL), 0.3 mg/L kinetin (PhytoTech, Cat No./ID: K483-100mL), 1:10000 ratio STS (0.1 M Na2S2O3:0.1 M AgNO3 = 4:1), and 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.8. The formulation of SIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 10 g/L glucose (Sigma, Cat No./ID: G7528-1KG), 0.25 g/L xylose (Sinopharm Chemical Reagent, Cat No./ID: 63012037), 0.6 g/L MES (Abcone, Cat No./ID: M38689-500G), 0.1 mg/L indole-3-acetic acid (PhytoTech, Cat No./ID: I364-100mL), 2 mg/L zeatin (PhytoTech, Cat No./ID: Z860-10ML), 3 mg/L AgNO3 (Sinopharm Chemical Reagent, Cat No./ID: 81012365), and 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.8.
Shoot regeneration assays for Solanum lycopersicum
The shoot regeneration assays for
S. lycopersicum were conducted as previously described, with some modifications
92. First, seeds were washed in 75% ethanol for 1 min and then rinsed with sterile water five times. Next, seeds were treated with 50% bleach (2 drops of Tween-20 per 50 mL) for 10 min, followed by five more rinses in sterile water. The seeds were subsequently sown on the 1/2 MS agar plate and incubated in a RUIHUA incubator at 26 °C in the dark for 7–10 days. The hypocotyls of the seedlings were excised to a fixed length of approximately 1 cm and cultured on the CIM in the RUIHUA incubator at 26 °C in the dark for 14 days. If squeezing treatment is necessary, the excised hypocotyls were placed on a petri dish and gently squeezed with ophthalmic tweezers before being transferred to the CIM. To induce shoot regeneration, the resulting calli were transferred onto the SIM and cultured in the RUIHUA incubator at 26 °C under long-day conditions with a light intensity of 140 μmol m
–2·s
–1. The SIM was refreshed approximately every 17 days. U0126 was only applied during the CIM stage.
The formulation of the 1/2 MS agar medium is the same as that for A. thaliana. The formulation of CIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 18 g/L mannitol (Sigma, Cat No./ID: M4125-500G), 2 mg/L 6-benzylaminopurine (PhytoTech, Cat No./ID: B130-100mL), 0.2 mg/L 1-naphthaleneacetic acid (PhytoTech, Cat No./ID: N605-100ML), 1:10000 ratio STS (0.1 M Na2S2O3:0.1 M AgNO3 = 4:1), and 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.8. The formulation of SIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 1 mg/L zeatin (PhytoTech, Cat No./ID: Z860-10ML), and 2.5 g/L phytagel (Sigma, Cat No./ID: P8169-250G), pH 5.8.
Shoot regeneration assays for Solanum pimpinellifolium
The shoot regeneration assay for
S. pimpinellifolium were conducted as described previously, with some modifications
92. First, seeds were washed with 75% ethanol for 1 min and rinsed 5 times with sterile water. Next, seeds were treated with a 10% sodium hypochlorite solution (2 drops of Tween-20 per 50 mL) for 10 min, followed by five more rinses in sterile water. The seeds were subsequently sown on the 1/2 MS agar plate and cultured in an RUIHUA incubator at 26 °C in the dark for 5–7 days. The hypocotyls of seedlings were excised to a fixed length of approximately 1 cm and cultured on the CIM in a RUIHUA incubator at 26 °C in dark for 14 days. If squeezing treatment is required, the excised hypocotyls were placed on a petri dish and gently squeezed with ophthalmic tweezers, and then placed on the CIM. The calli were cultured on the SIM in a RUIHUA incubator at 26 °C under long-day conditions with a light intensity of 140 μmol m
–2·s
–1M stage.
The formulation of the 1/2 MS agar medium is the same as A. thaliana. The formulation of CIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 18 g/L mannitol (Sigma, Cat No./ID: M4125-500G), 2 mg/L 6-benzylaminopurine (PhytoTech, Cat No./ID: B130-100mL), 0.2 mg/L 1-naphthaleneacetic acid (PhytoTech, Cat No./ID: N605-100ML), 1:10000 ratio STS (0.1 M Na2S2O3:0.1 M AgNO3 = 4:1), and 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.8. The formulation of SIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 0.1 mg/L indole-3-acetic acid (PhytoTech, Cat No./ID: I364-100mL), 2 mg/L zeatin (PhytoTech, Cat No./ID: Z860-10ML), 2.5 g/L phytagel (Sigma, Cat No./ID: P8169-250G), pH 5.8. The formulation of the SIM for refreshment is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (ABCONE, Cat No./ID: S48029-1KG), 1 mg/L zeatin (PhytoTech, Cat No./ID: Z860-10ML), and 2.5 g/L phytagel (Sigma, Cat No./ID: P8169-250G), pH 5.8.
Shoot regeneration assays for Capsicum annuum
The shoot regeneration assays for
C. annuum CM334 were conducted as previously described, with some modifications
93,94. Briefly, seeds were washed in 75% ethanol for 1 min, followed by five rinses with sterile water. The seeds were then treated with a solution of 50% bleach (containing 2 drops of Tween-20 per 50 mL) for 10 min, after which they were rinsed five more times in sterile water. The seeds were subsequently sown on the 1/2 MS agar plate and incubated in a RUIHUA incubator at 26 °C under long-day conditions with a light intensity of 140 μmol m
–2·s
–1 for 14 days. After germination, the cotyledons were excised and cultured on CIM in the RUIHUA incubator at 26 °C under long-day conditions with a light intensity of 70 μmol m
–2·s
–1 for 14 days. To induce shoot regeneration, the resulting calli were transferred onto SIM and cultured in the RUIHUA incubator at 26 °C under long-day conditions with a light intensity of 140 μmol m
–2·s
–1, with the SIM being refreshed approximately every 2 weeks. U0126 was only applied during the CIM stage.
The formulation of the 1/2 MS agar medium is the same as that for A. thaliana. The formulation of CIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 4 mg/L 6-benzylaminopurine (PhytoTech, Cat No./ID: B130-100mL), 3 mg/L indole-3-acetic acid (PhytoTech, Cat No./ID: I364-100mL), 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.8. The formulation of SIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 4 mg/L 6-benzylaminopurine (PhytoTech, Cat No./ID: B130-100mL), 1 mg/L indole-3-acetic acid (PhytoTech, Cat No./ID: I364-100mL), 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.8.
Shoot regeneration assays for Medicago sativa
The shoot regeneration assays for
M. sativa Zhongmu No.1 were conducted as previously described, with some modifications
95. Briefly, seeds were washed in 75% ethanol for 1 min and then rinsed with sterile water five times. The seeds were then treated with 15% sodium hypochlorite solution (2 drops of Tween-20 per 50 mL) for 10 min, followed by five more rinses in sterile water. The seeds were subsequently sown on the 1/2 MS agar plate and incubated in a RUIHUA incubator at 26 °C in the dark for 5–7 days. After germination, the hypocotyls were excised to a fixed length of approximately 1 cm and cultured on CIM in a RUIHUA incubator at 26 °C in dark for 6 weeks, with the CIM being refreshed approximately every 3 weeks. To induce shoot regeneration, the resulting calli were transferred onto SIM and cultured in the RUIHUA incubator at 26 °C under long-day conditions with a light intensity of 140 μmol m
–2·s
–1. The SIM was refreshed approximately every 2 weeks until as many green proembryos as possible appeared. The calli were then transferred to hormone-free MS medium containing 20 g/L sucrose. U0126 was only applied during the CIM stage.
The formulation of the 1/2 MS agar medium is the same as A. thaliana. The formulation of CIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 4 mg/L 2,4-dicholorophenoxyacetic acid (PhytoTech, Cat No./ID: D295-100mL), 0.5 mg/L kinetin (PhytoTech, Cat No./ID: K483-100mL), 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.8. The formulation of SIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 1 mg/L kinetin (PhytoTech, Cat No./ID: K483-100mL), 0.5 mg/L 6-benzylaminopurine (PhytoTech, Cat No./ID: B130-100mL), 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.8.
Shoot regeneration assays for Setaria viridis
The shoot regeneration assays for
S. viridis ME034V were conducted as previously described, with some modifications
96. Briefly, the dehusked seeds were sterilized in a solution of 10% bleach plus 0.1% Tween-20 for 10 min on a rotating mixer, followed by five more rinses in sterile water. Excess water was then removed, and the sterilized seeds were dried on sterile filter paper. The seeds were subsequently cultured on CIM in a RUIHUA incubator at 26 °C with a photoperiod of 16 h for 1 week with a light intensity of 70 μmol m
–2·s
–1 followed by 4–5 weeks in the dark, with the CIM being refreshed approximately every 2 weeks. To induce shoot regeneration, the resulting calli were transferred onto SIM and cultured in the RUIHUA incubator at 26 °C under long-day conditions with a light intensity of 140 μmol m
–2·s
–1, with the SIM being refreshed approximately every 2 weeks. U0126 was only applied during the dark treatment phase at the CIM stage.
The formulation of CIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 40 g/L maltose (BBI, Cat No./ID: A604950-0500), 2 mg/L 2,4-dicholorophenoxyacetic acid (PhytoTech, Cat No./ID: D295-100mL), 0.5 mg/L kinetin (PhytoTech, Cat No./ID: K483-100mL), 0.6 mg/L copper sulfate pentahydrate (Sigma, Cat No./ID: C3036-250G), 4 g/L Gelzan (Sigma, Cat No./ID: G1910-250G), pH 5.8. The formulation of SIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 20 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 1 mg/L 2,4-dicholorophenoxyacetic acid (PhytoTech, Cat No./ID: D295-100mL), 1 mg/L kinetin (PhytoTech, Cat No./ID: K483-100mL), 0.6 mg/L Copper sulfate pentahydrate (Sigma, Cat No./ID: C3036-250G), 7 g/L agar (Duchefa, Cat No./ID: P1001.0500), pH 5.8.
Shoot regeneration assays for Oryza sativa
The shoot regeneration assays for
O. sativa ZH11 were conducted as previously described, with some modifications
97. Briefly, the dehusked seeds were washed in 75% ethanol for 1 min and then rinsed with sterile water five times. The seeds were then treated with 20% sodium hypochlorite solution (2 drops of Tween-20 per 50 mL) for 15 min, followed by five more rinses in sterile water. Excess water was then removed, and the sterilized seeds were dried on sterile filter paper. The seeds were subsequently cultured on CIM in a RUIHUA incubator at 26 °C in dark for 6 weeks, with the CIM being refreshed approximately every 3 weeks. To induce shoot regeneration, the resulting calli were transferred onto SIM and cultured in the RUIHUA incubator at 26 °C under long-day conditions with a light intensity of 140 μmol m
–2·s
–1, with the SIM being refreshed approximately every 2 weeks. U0126 was only applied during the dark treatment phase at the CIM stage.
The formulation of CIM is as follows: 4.43 g NB basal medium w/vitamins powder (PhytoTech, Cat No./ID: N492-50L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 0.5 g/L L-Proline (Sigma, Cat No./ID: P5607-100G), 0.5g/L L-Glutamine (Sigma, Cat No./ID: G8540-100G), 0.3g/L Casamino acids (Caisson, Cat No./ID: C004-100GM), 2.5 mg/L 2,4-dicholorophenoxyacetic acid (PhytoTech, Cat No./ID: D295-100mL), 3.5 g/L phytagel (Sigma, Cat No./ID: P8169-250G), pH 5.8. The formulation of SIM is as follows: 4.43 g MS basal medium w/vitamins powder (PhytoTech, Cat No./ID: M519-100L), 30 g/L sucrose (Abcone, Cat No./ID: S48029-1KG), 30 g/L sorbitol (Abcone, cat No./ID: S18766-500G), 0.5 g/L L-Proline (Sigma, Cat No./ID: P5607-100G), 0.5g/L L-Glutamine (Sigma, Cat No./ID: G8540-100G), 0.3g/L Casamino acids (Caisson, Cat No./ID: C004-100GM), 0.1 g/L myo-inositol (BBI, Cat No./ID: A600536-0100), 2 mg/L kinetin (PhytoTech, Cat No./ID: K483-100mL), 2 mg/L 6-benzylaminopurine (PhytoTech, Cat No./ID: B130-100mL), 0.2 mg/L indole-3-acetic acid (PhytoTech, Cat No./ID: I364-100mL), 0.2 mg/L 1-naphthaleneacetic acid (PhytoTech, Cat No./ID: N605-100ML), 3.5 g/L phytagel (Sigma, Cat No./ID: P8169-250G), pH 5.8.
scRNA-seq library construction
The protoplasts of explants and calli were prepared as previously described with modifications
47. Briefly, the hypocotyls (CIM0) and calli (CIM1–CIM7) were harvested and cut into small pieces using scissors. Each sample was then separately digested in an RNase-free enzyme solution consisting of 1.5% cellulase R10, 0.4% macerozyme R10, 0.5% snailase, 1% Driselase, 0.4 M mannitol, 0.1 M 4-morpholineethanesulfonic acid, 10 mM KCl, 10 mM CaCl
2, and 0.1% BSA, at room temperature for 2 h. During the enzymatic digestion, a syringe was used to mechanically disrupt the calli under a dissecting microscope to accelerate the digestion process. To completeness of digestion was confirmed under a stereomicroscope. The protoplasts were then filtered 3–4 times using cell strainers (40 μm in diameter, Falcon, Cat No./ID: 352340), concentrated, and washed 3–4 times with 8% mannitol at room temperature. Protoplast viability was assessed by trypan blue staining, with the ratio of viable cells to total cells in each sample exceeding 85%. The protoplast suspension was counted by a hemocytometer, and the final concentration of protoplasts was adjusted to 1,500–2,000 cells/μL.
scRNA-seq libraries were prepared as previously described
47. Briefly, the cell suspensions were loaded onto a Chromium Single Cell Instrument (10× Genomics, USA) to generate single-cell GEMs. The scRNA-seq libraries were then constructed using the Chromium Single Cell 3ʹ Gel Bead and Library Kit v3 (10× Genomics, Cat No./ID: P/N 1000075 and 1000073). The detailed experimental procedure followed the user guide for the Chromium Single Cell 3ʹ Reagent Kits v3 (CG000183 Rev A). Qualitative analyses of DNA libraries were performed using an Agilent 2100 Bioanalyzer. The libraries were sequenced on an Illumina NovaSeq X Plus (Shanghai Genomics Technology, China) using two 150 bp paired-end kits. The resulting raw scRNA-seq data comprised 28 bp Read1, 150 bp Read2, and 8 bp i7 index reads.
Identification of protoplasting-responsive genes
To minimize the impact of protoplasting on scRNA-seq, we performed RNA-seq on calli at four time points. The bulk RNA-seq data on the samples at CIM0, 3, and 7 without protoplasting were described
38. By comparing differentially expressed genes between normal condition and protoplast, we identified protoplasting-responsive genes at each time point (log
2FoldChange > 2 and
padj < 0.05) (Supplementary Fig. S1b). The resulting shared set of 284 genes was used as the final gene list (Supplementary Table S1). We then mitigated the effect of protoplasting on scRNA-seq by filtering these genes (see below).
Pre-processing of raw scRNA-seq data
The raw scRNA-seq data were initially processed using Cell Ranger (v7.0.0, 10× Genomics, USA). This software pipeline includes aligning reads and generating gene-cell matrices from the raw scRNA-seq data. The genome and GTF annotation files of A. thaliana were obtained from TAIR (arabidopsis.org/), specifically using the TAIR10 version of the genome and the Araport11 version of the annotation files. To build the reference, we executed the "cellranger mkref" command with the "--genome, --fasta and –genes" arguments. Subsequently, we ran the "cellranger count" command with the "--id, --transcriptome, --fastqs, --sample" arguments to generate single-cell gene counts. The gene-cell matrices directory generated by Cell Ranger (v7.0.0), named "filtered_feature_bc_matrix" by 10× Genomics, served as the raw data for further analysis.
scRNA-seq data analysis for a single sample
Single-cell data analyses were conducted following the guidelines outlined in the Bioconductor community's manual, "Orchestrating Single-Cell Analysis with Bioconductor (OSCA)"
98. First, we used the "read10xCounts" function from the DropletUtils package (v1.16.0) to import the single-cell expression matrix obtained from Cell Ranger, converting it into a SingleCellExperiment object. Next, we applied the "scDblFinder" function from the scDblFinder package (v1.10.0) to identify doublets within the dataset
99. Following this, we conducted a data quality assessment using the "perCellQCMetrics" and "addPerCellQCMetrics" functions from the scuttle package (v1.6.2). Subsequently, we performed cell outlier detection based on the quality metrics obtained in the previous step using the "isOutlier" function from the robustbase package (v0.95). We filtered cells based on their classification as outliers, the proportion of chloroplast and mitochondrial content (with a threshold set at 5%), and the presence of doublets.
We used the "quickCluster" function from the scran package (v1.24.0) for rapid cell clustering. Following the clustering, we applied the "computeSumFactors" function to adjust for sequencing depth in each cell, resulting in a factor-scaled and log-transformed single-cell expression matrix. Finally, we utilized the "modelGeneVar" function from the scran package (v1.24.0) to model the variation of each gene across different cells, identifying the top 1,000 highly variable genes for subsequent analysis. Among these highly variable genes, we removed protoplast-responsive genes to mitigate the effects of protoplasting on our data.
We employed the "runPCA" function from the scater package (v1.27.9) to perform dimensionality reduction on the single-cell data. Following this, we used the "runTSNE" and "runUMAP" functions, also from the scater package (v1.27.9), to create t-SNE and UMAP plots for visualization. Subsequently, we applied the "clusterCells" function from the scran package (v1.24.0) to cluster all cells in the atlas according to the PCA results. We selected the "Louvain" method for clustering and used the "jaccard" method for boundary weight calculations. We experimented with different numbers of neighboring cells — specifically 5, 10, 20, and 50 — to obtain varying cluster counts, selecting specific cluster numbers based on our annotation requirements. Finally, we utilized the "scoreMarkers" function from the scran package (v1.24.0) to calculate marker genes for each cell cluster. To assist in identifying marker genes, we also calculated a score using the results generated by the "scoreMarkers" function. The scoring formula is as follows: "score = (self.average – other.average) / (other.average + 1e-5) * median.logFC.detected". We defined candidate marker genes as those with self.detected > 0.05 and self.average – other.average > 0, along with the highest scores.
Two biological replicates were performed for each time point except CIM7 (Supplementary Table S1). The integration of the callus atlas at CIM7 (from this study) with the published atlas at CIM6 revealed a high degree of similarity (Supplementary Fig. S3j)
32.
Integrated analysis of scRNA-seq data for multi-samples
We utilized the "multiBatchNorm" function from the batchelor package (v1.12.3) to correct for sequencing depth across the single-cell data as mentioned above, allowing for integration across multiple samples. Following this, we employed the "combineVar" function from the same batchelor package to integrate the highly variable genes identified in the above single simple analysis, resulting in a consolidated set of highly variable genes. From this set, we selected the top 6,000 genes for subsequent data analysis. The protoplast-responsive genes were removed during this process.
Next, we performed dimensionality reduction on the corrected single-cell data using the "runPCA" function, followed by visualization with "runTSNE" and "runUMAP". We assessed the results from t-SNE and UMAP to determine if further integration was necessary, avoiding the risk of over-integration. In case where multiple samples could not be effectively integrated in the t-SNE or UMAP plots, we applied the "fastMNN" function from the batchelor package (v1.12.3) for integration. The order of merging was based on our sample choice: for samples derived from calli at multiple time points, we merged them in the order of CIM 0, 1, 3, and 7; for a comparison of a control group with a treatment group, the merging occurred in the order of control followed by treatment.
For further data visualization, we selected the "corrected "dimensions in the "dimred" option of the "runTSNE" and "runUMAP" functions. We subsequently employed the "clusterCells" function to conduct clustering on the integrated single-cell data. If the integration was achieved using fastMNN, we specified the "corrected" dimensions in the "use.dimred" option; otherwise, we used the "PCA" dimensions. The remaining clustering parameters remained consistent with those used in the above "scRNA-seq data analysis for a single sample" section.
Finally, if multiple samples represented multiple biological replicates, we activated the "block" option in the "scoreMarkers" function when calculating marker genes. This enabled us to fully leverage the replicate information, thereby enhancing the identification of robust marker genes across biological replicates.
Chemical screening
We used liquid CIM media for chemical screening in 24-well plates based on three main reasons: 1) Using normal Petri dishes (9 cm × 9 cm) would require large quantities of each compound, increasing costs and making large-scale screening impractical; 2) It is difficult to evenly distribute compounds in solid CIM medium in 24-well plates; 3) Using solid media in a 24-well plate risks cross-contamination from tweezers when placing explants in different compounds, affecting results. In contrast, using liquid media allows us to add compounds after placing the explants, avoiding cross-contamination.
The chemical compound library (Selleck New Compound Library) was obtained from Core Technology Facility of Center (CAS Center for Excellence in Molecular Cell Science, sjzx.sibcb.ac.cn/Cn/Index/listView/catid/58.html) (Supplementary Table S1). The steps for compound screening are as follows (Supplementary Fig. S12f): 1) Preparation of Culture Plates: add 1.8 mL of liquid CIM to each well of a 24-well cell culture plate, and then place eight hypocotyl explants into each well; introduce 0.9 μL of a 10 mM stock solution of the compound to each well, while designating one well on each plate as a control by adding 0.9 μL of DMSO. 2) Incubation: seal the cell culture plates with Parafilm to prevent contamination and incubate them in a Percival incubator at 22 °C under long-day conditions, with a light intensity of 45 μmol m–2·s–1 for 7 days. 3) Fluorescence Observation: use a confocal microscope to examine the number of pACL5::sGFP-N7 fluorescence foci in the calli treated with DMSO; establish the threshold by identifying the lowest number of fluorescent foci observed in 8 calli; subsequently, evaluate the fluorescence in the callus treated with each compound. If the count of fluorescent foci exceeds the threshold, record it as 1; if it falls below the threshold, record it as 0. 4) Selection of Promising Compounds: choose the compounds with the highest number of 0s for a second round of screening. The protocol for the second round follows the same method as the first round. 5) Selection of Candidate Compounds: after completing both rounds of screening, select candidate compounds for statistical analysis of fluorescence foci on the CIM and assess their shoot regeneration efficiency on the SIM. 6) Validation of Candidate Compounds I: add 5 μM of the candidate compound to the CIM and place 24 explants. After 7 days of culture, count the number of fluorescent foci in each callus and capture representative images. 7) Validation of Candidate Compounds II: add 5 μM of the candidate compound to the CIM. After 7 days of culture, transfer the explants to the SIM. After 16 days of culture, count the number of roots and shoots produced.
The chemical compound library of MEK inhibitors was obtained from TargetMol (Supplementary Table S1). The steps for compound screening followed the previously described protocol, with one exception: during the fluorescent observation step, we utilized a larger quantity of calli (n = 15 instead of 8) for each MEK compound. This adjustment allowed us to more effectively identify MEK inhibitors with stronger effects.
Frozen sectioning
To prepare frozen sections of fresh callus tissues for in situ RNA sequencing (IISS), follow these steps: 1) Immerse fresh callus tissues in Optimal cutting temperature compound (OCT) (Skura, Cat No./ID:4583) and apply a vacuum for 1–2 minutes while on ice, ensuring that no bubbles form. 2) Add an appropriate amount of OCT into an embedding mold (Citotest, Cat No./ID: 80203-0007). Then, transfer the tissue onto the surface of the OCT. 3) Quickly place the mold containing the tissue onto dry ice until the material solidifies. 4) Use a cryostat (Leica CM3050S) for sectioning, setting the blade temperature to –15 °C and the chamber temperature to –20 °C. Slice the tissue into sections with a thickness of 10 µm. 5) Lay the tissue slices flat on low-adhesion glass slides (Citotest, Cat No./ID: 188105) for further analysis.
In situ RNA sequencing
The IISS was performed as previously described
51. In brief, fresh frozen sections underwent an initial treatment at 37 °C for 5 min, followed by fixation in 4% (w/v) paraformaldehyde for 5 min. The sections were then washed twice with diethyl pyrocarbonate (DEPC)-treated 1× phosphate buffered saline (PBS). Subsequently, the sections were dehydrated through a series of ethanol concentrations, followed by three washes with Wash Buffer (Dynamic Biosystems, Suzhou, China), and then treated with 0.1 M HCl at 37 °C for 5 min.
The sections were incubated sequentially with the following solutions: custom target gene probes (Dynamic Biosystems, Suzhou, China) labeled with a specific barcode for each RNA molecule in hybridization mix at 37 °C for 4 h, a ligation mix at 37 °C for 30 min, splint primers in a circularization mix at 37 °C for 30 min, and an amplification mix for rolling circle amplification at 30 °C overnight. After each incubation step, the slides were washed three times with Wash Buffer at room temperature.
For sequencing, the sections were incubated in ligation buffer containing anchor primers, labeled sequencing interrogation probes, and ligase at 30 °C for 45 min. They were then mounted with SlowFade Gold Antifade Mountant (ThermoFisher, Cat No./ID: S36938) mixed with 0.5 μg/mL 4',6-diamidino-2-phenylindole (DAPI). Imaging was conducted using a 3D-Hitech Pannoramic® MIDI II equipped with a pco.edge 4.2 bi-camera, utilizing a 20× objective lens. A reference image was established through a round of DAPI imaging, and the SIFI algorithm was employed to extract features from different rounds of DAPI images. Corresponding feature points across rounds were matched, and these matched points were used for affine transformation to register images from various rounds. Background filtering was performed using the "tophat" function from OpenCV. By distinguishing genuine signals and decoding them, the spatial localization and quantification of RNAs were ultimately attained. The analytical pipeline utilized in this study can be accessed at github.com/DynamicBiosystems/DynamicISS.
Tissue embedding and sectioning
Samples were collected and immediately placed in the vials with ice-cold PBS containing 2.5% paraformaldehyde (PFA, pH 7.0). The samples were infiltrated for 30 min by vacuum at ice. Tissues were then washed with PBS solution for three times, embedded with 6% low melting agarose, and sliced with a Lecia Sliding Microtome 1200S at the thickness of 70 to 130 µm.
Cell outline staining
For the hypocotyl (CIM0) and calli at different stages (CIM1 to CIM5), transfer the samples into water containing propidium iodide (PI, Meilunbio, Cat No./ID: MA0137_1ML) at a concentration of 5–10 µg/mL. Incubate at room temperature for 5–10 minutes, then rinse three times with water. For the calli at CIM7, first fix the samples using 2.5% PFA as described above. Then transfer the samples to ClearSee solution
100, and incubate overnight at room temperature. Add SCRI Renaissance 2200 (SR2200, Renaissance Chemicals, Cat No./ID: SR2200) reagent to the ClearSee solution to a final concentration of 0.1%, and incubate for 1 h at room temperature. Afterwards, wash the samples once with ClearSee solution.
Microscopy
For confocal imaging, we used FV3000 confocal microscopic system (Olympus, Japan). Proper filter sets and lasers were selected for fluorescence signal scanning. Specifically, sGFP and VENUS were stimulated with an argon laser at 488 nm, with emission filtered at 500 to 540 nm; DsRed was excited at 561 nm and detected at a wavelength between 570 to 620 nm; PI was excited at 561 nm and detected at a wavelength between 590 to 680 nm; SR2200 was excited at 405 nm and detected at a wavelength between 430 to 470 nm. The images were processed with ImageJ.
Time-lapse recording of callus development under white light conditions
We used FV3000 confocal microscopic system (Olympus, Japan) to conduct time-lapsed tracking of specific callus region of the fluorescence reporter lines. For each experiment, the callus was placed in a glass-bottom cell culture dish containing 300 μL of the SIM phytagel medium (Biosharp, Cat No./ID: BS-20-GJM). We labeled specific callus regions by marking the bottom of the dish. Once the tracking at each time point was complete, the callus was returned to the normal culture conditions.
We used BX63 microscope (Olympus, Japan) equipped with a DP74 digital camera for time-lapsed tracking of the entire callus. For each experiment, the callus was placed in a glass-bottom culture dish containing 300 μL of SIM phytagel medium. We used the "automatic XY-Positions/MIA acquisition process" function in OLYMPUS CellSens Dimension software (v4.1.1) to scan the entire callus and merged all the images into one stitched image. Once the tracking at each time point is complete, the callus is returned to the normal culture condition.
Time-lapse recording of ACL5+ callus regions
We combined SpinSR spinning-disk microscope (Olympus, Japan) and BX63 microscope (Olympus, Japan) for time-lapse tracking of ACL5+ callus regions. The ACL5+ callus regions were defined as those exhibiting a striped expression pattern of ACL5 (Supplementary Fig. S7b). Specifically, we used SpinSR spinning-disk microscope, which shortens imaging time and minimizes phototoxicity, to define the ACL5+ and ACL5– callus regions at CIM7 and employed the "automatic XY-Positions/MIA acquisition process" function in OLYMPUS CellSens Dimension software (v4.3) of SpinSR to scan the entire callus and merged all the images into one stitched image. During the SIM stages, we used BX63 microscope for time-lapsed tracking of the corresponding callus regions and used the "automatic XY-Positions/MIA acquisition process" function in OLYMPUS CellSens Dimension software (v4.1.1) of BX63 to scan the entire callus and merged all the images into one stitched image. By comparing the ACL5+ and ACL5– callus regions at CIM7 with the corresponding callus regions at the SIM stages, we determined their cell fates. For each experiment, the callus was placed in a glass-bottom cell culture dish containing 300 μL of the SIM phytagel medium (Biosharp, Cat No./ID: BS-20-GJM). Once the tracking at each time point was complete, the callus was returned to the normal culture conditions. Fluorescence acquisition settings for pACL5::sGFP-N7 reporter: 488-nm excitation; emission collected at 500–550 nm.
Live and time-lapse imaging using fluorescent reporters
A 35-mm glass-bottom dish was pre-rinsed with 200 μL of 0.1% BSA to reduce surface tension. To minimize any impact of medium on fluorescence imaging, 50 μL of solid CIM medium was used. After the medium solidified, callus tissue was placed on the surface. The tissue was overlaid with a thin layer of 1% agar, followed by 200 μL of liquid CIM medium.
Live imaging with PlaCCI was performed at room temperature on SpinSR spinning-disk confocal microscope (Olympus, Japan) at 40-minute intervals for a total of 66 cycles. Z-stacks were acquired with a step size of 3 μm. Fluorescence acquisition settings: pCDT1a::CDT1a-eCFP: 445-nm excitation; emission 465–495 nm. pHTR13::HTR13-mCherry: 561-nm excitation; emission 565–595 nm. pCYCB1;1::NCYCB1;1-YFP: 488-nm excitation; emission 500–550 nm.
Time-lapse imaging with pAT2G18020:H2B-mCherry and Cytrap was performed at room temperature on FV3000 microscope (Olympus, Japan), with Z-stacks acquired eight times daily. After tracking at each time point was completed, the callus was returned to normal culture conditions. The fluorescence acquisition settings for pAT2G18020:H2B-mCherry and pCYCB1::CYCB1-GFP were the same as those described above.
EdU labeling experiment
For the 5-ethynyl-2'-deoxyuridine (EdU) labeling experiments, calli at different growth stages were transferred to a small dish containing liquid CIM, 10 μM of EdU, and 5 μM of U0126 (with DMSO as a control). After incubating for 4 h at 22 °C (or 8 h for the explants at CIM0), the culture media were replaced with 1 mL of fix solution (4% PFA). The samples were then subjected to vacuum infiltration for 15 min and washed twice with 1 mL of wash buffer (PBS containing 3% BSA) for 3–5 min each. Subsequently, 1 mL of permeabilization buffer (PBS containing 0.3% Triton X-100) was added into each well, and the samples were incubated at room temperature for 10–15 min. The calli were washed 1–2 times with 1 mL of wash solution for 3–5 min each. The remaining procedures were performed using EdU Cell Proliferation Kit (Beyotime, Cat No./ID: C0071S) according to the manufacturer’s instructions.
RNA-seq experiments
Total RNAs were extracted from calli using RNeasy Plant Mini Kit (Qiagen, Cat No./ID: 74904). Library construction and 150-bp paired-end sequencing were performed using the Illumina NovaSeq X plus Platform according to manufacturer’s protocols (Personal Biotechnology, Shanghai, China). Two or three biological replicates were performed.
Data preprocessing and reads alignment for RNA-seq
For each library, raw.fastq was first trimmed by fastp (v0.20.0)
101 with default parameters. FastQC (v0.11.7) and MulitQC (v1.6) were then performed to obtain clean fq files
102. Reads were finally aligned to the
A. thaliana (TAIR10),
B. napus (ZS11), and
S. pimpinellifolium (LA1589) genomes using hisat2 (v2.1.0)
103. The TAIR10 genome sequence was downloaded from TAIR. The ZS11 and LA1589 genome sequences were downloaded from BnPIR
104 and Figshare
105, respectively. The resulting SAM files were converted into the BAM format, sorted, and indexed by Samtools (v1.9)
106.
RNA-seq data analyses
The number of paired mapping reads overlapping each annotated gene (
A. thaliana:
Araport11;
B. napus:
ZS11;
S. pimpinellifolium:
LA1589) was quantified using featureCounts (v1.6.2) with the parameter "-p"
107,108. The resulting counts files served as inputs for differential gene expression analysis performed with DESeq2 (v1.36.0)
109. Log fold change shrinkage was estimated using the ashr package
110.
For data visualization, we extracted the normalized count values generated by DESeq2 and applied max-min scaling. Gene Ontology (GO) and Gene Set Enrichment Analysis of differentially expressed genes were conducted using clusterProfiler (v4.4.1)
111 and org.At.tair.db (v3.15.1). The Plant Ontology (PO) set was downloaded from the Planteome github repository
112, and the gene-PO annotations were sourced from TAIR. We used ontologyIndex (v2.12)
113 to select sets of ontological terms, and the "enricher" function provided by ClusterProfiler (v4.4.1) to conduct PO enrichment analysis.
Mass-spectrometry-based phosphoproteomic analysis
Callus tissues were treated with either DMSO (mock) or U0126 for 7 days on CIM, with each biological replicate weighing between 0.1 and 0.2 grams. Total protein was extracted with 6 M guanidine hydrochloride buffer (100 mM Tris-HCl, 10 mM EDTA, 10 mM TCEP, 1 mM PMSF, pH 8.0) and quantified by bicinchoninic acid (BCA) assay. Two milligrams of protein were digested overnight at 37 °C with trypsin (Promega, Cat. No./ID: V5280). Peptides were quantified, allocating 10 μg for global proteomics and 1.5 mg for phosphoproteomics. Phosphopeptides were enriched using the High-Select TiO2 Phosphopeptide Enrichment Kit (Thermo Fisher Scientific, Cat No./ID: A32993) according to the manufacturer’s instructions. LC-MS/MS was performed on a timsTOF Pro 2 mass spectrometer (Bruker Daltonics) in data-independent acquisition (DIA) mode with parallel accumulation–serial fragmentation (PASEF). Raw data were processed in Spectronaut 19 (Biognosys) with default settings for label-free quantification against the A. thaliana reference proteome (UniProt UP000006548). Five biological replicates were performed.
Phosphoproteomics mass spectrometry data analysis
We used the NAguideR package (v0.2.1) to perform missing value imputations and ensure data quality for both proteomics and phosphoproteomics datasets. The data were then transformed to log2 scale for subsequent analysis and statistical evaluation
114. To correct for variations, we adjusted the phosphopeptide data by dividing the abundance of phosphopeptides by the corresponding abundance of their respective proteins. Differential analysis of the corrected phosphopeptides was carried out using the limma package (v3.58.1)
115. For GO term analysis of the differentially expressed phosphopeptides, we employed clusterProfiler (v4.4.1)
111 and org.At.tair.db (v3.15.1). Additionally, we conducted motif enrichment analysis of the differential phosphopeptides using pLogo
116 and motif-x 2.0
117.
Quantification and statistical analysis
For Fig. 2d, a generalized linear mixed model (GLMM) with a generalized Poisson distribution was fitted using the glmmTMB package to account for underdispersion and repeated measurements, with material type, time point, and their interaction treated as fixed effects, and explant (Callus) and individual tracking ID (Location) treated as random effects. Model residuals were validated using the DHARMa package, showing acceptable dispersion and overall distribution fit. Post-hoc pairwise comparisons between ACR and FCR at each specific day were subsequently conducted based on estimated marginal means using the emmeans package, with P values adjusted for multiple testing using the False Discovery Rate (FDR) method.
For Fig. 2e, to evaluate the initiation timing of cell division, a mixed-effects Cox proportional hazards model was applied with the explant batch (Callus) included as a random effect. The results revealed that ACR initiated division (reaching ≥ 2 cell layers) significantly earlier than FCR (P < 0.001, z = 11.84). The hazard ratio (HR) for ACR relative to FCR was 11.98 (95% CI: 7.94–18.07), indicating that ACR cells progressed to two or more layers at an approximately 12-fold higher rate compared to FCR.
For Fig. 2f, to compare the proportion of cells undergoing the first division between ACR and FCR across time points (CIM1, CIM2, and CIM3), a linear mixed-effects model (LMM) was fitted using the lme4 and lmerTest packages in R. The model included material type (ACR vs FCR), time point (Day), and their interaction (type × Day) as fixed effects, while individual explant batches (CallusID) were treated as a random intercept to account for the paired experimental structure. Pairwise comparisons between ACR and FCR at each specific day were performed using estimated marginal means via the emmeans package, with P values adjusted for multiple testing using the False Discovery Rate (FDR / Benjamini-Hochberg) method.
We used Student’s t-test or one-way ANOVA and Tukey's multiple comparisons to determine the statistical significance between different samples in quantification and phenotypes. All statistical results and graphs were generated by R. The numbers of samples and types of statistical analyses are given in figure legends. The normality of the data distribution and the homogeneity of variances (homoscedasticity) were evaluated using the Shapiro–Wilk test or Levene’s test, respectively. Based on the evaluation results, appropriate parametric tests (e.g., Student’s t-test and ANOVA) or non-parametric tests (e.g., Mann–Whitney U test and Kruskal–Wallis test) were applied for subsequent comparisons.
DATA AVAILABILITY
The scRNA-seq and RNA-seq data (BioProject PRJCA034290) were deposited in Beijing Institute of Genomics Data Center (bigd.big.ac.cn). The raw data related to phosphoproteomics and spatial transcriptomics were deposited on Figshare through doi.org/10.6084/m9.figshare.33159995.
The Author(s) 2026. Published by Higher Education Press. This is an Open Access article distributed under the terms of the CC BY license (https://creativecommons.org/licenses/by/4.0/).
