Fertilization controls tiller numbers via transcriptional regulation of a MAX1-like gene in rice cultivation

The study addresses how fertilization regulates rice tiller number, a key determinant of yield, through hormonal and genetic mechanisms. While chemical fertilizers have historically increased yields, overuse leads to environmental pollution and economic burdens, motivating sustainable practices that reduce fertilizer inputs. Fertilization affects branching (tillering), panicle number, and organ sizes, likely via changes in phytohormone balance, including strigolactones (SLs), which regulate branching and are induced under nutrient-poor conditions. Prior work shows SL biosynthesis/signaling mutants in rice (e.g., d17, d27, d3) are insensitive to nutrient depletion and have increased tillering. In rice, multiple MAX1-like genes exist (Os900, Os1400, Os1500, Os1900, Os5100), with Os1500 likely nonfunctional. A QTL involving Os900 and Os1400 influences tiller number and SL biosynthesis, and Os900/Os1400 catalyze steps from carlactone (CL) to carlactonoic acid (CLA), 4-deoxyorobanchol (4DO), and orobanchol (ORO). However, canonical SLs may not be the major determinant of tillering, and the roles of each MAX1-like gene in tillering remain unresolved. The authors hypothesize that fertilization controls tiller number via transcriptional regulation of the MAX1-like gene Os1900 and investigate its function (with Os5100) in SL biosynthesis and tillering, including the impact of promoter cis-regulatory variation on fertilizer responsiveness.

• Nutrient stress and branching: SLs are carotenoid-derived hormones induced by phosphate (P) and nitrate deficiency, regulating shoot branching/tillering. SL-deficient and signaling mutants in rice (d17/D17, d27/D27, d3/D3) are insensitive to nutrient depletion and show elevated branching.
• P limitation and mycorrhiza: Inorganic phosphate is often insoluble in soils; plants rely on arbuscular mycorrhizal (AM) fungi for P acquisition. SLs act as rhizospheric signals to promote AM symbiosis.
• SL biosynthesis: CL forms from all-trans-β-carotene and is converted to CLA, then to 4DO and ORO through cytochrome P450 enzymes. In dicots (Arabidopsis, tomato), a single MAX1 catalyzes CL→CLA. In monocots (rice, maize), multiple MAX1-like genes exist.
• Rice MAX1-like genes: Five homologs (Os900, Os1400, Os1500, Os1900, Os5100) with Os1500 likely loss-of-function. Os900/Os1400 region is a QTL for tiller number and SL derivatives. Os900 and Os1400 show strong in vitro activities for CL→CLA and downstream conversions (CLA→4DO by Os900; 4DO→ORO by Os1400). Os1900/Os5100 showed low conversion to 4DO in vitro. P starvation induces Os900 and Os1900 (not Os1400/Os5100), with Os900 mainly in roots and Os1900 in shoots.
• Canonical SLs vs tillering: Recent work suggests canonical SLs (e.g., 4DO) are important rhizospheric signals but not the major determinants of tillering in rice, leaving the genetic basis for fertilizer-responsive tillering unclear. These studies set the context for testing whether Os1900 transcriptional regulation mediates fertilizer effects on tillering and for dissecting functional redundancy among MAX1-like genes.

• Field transcriptomics under real cultivation: Rice (cv. Koshihikari) grown in adjacent paddy fields with three fertilization regimes (None, Once, Twice) under normal and delayed transplanting. Leaf samples collected at eight time points (10:00 or 16:00) post-transplanting. Agilent 180k custom arrays were used; data log2-transformed, q-spline normalized, and analyzed via paired t-tests with FDR control. To isolate fertilizer effects under fluctuating environments, fold-change (FC) and FC/SD metrics were computed across samples, and repeated paired t-tests across fertilization contrasts identified fertilizer-responsive genes (FDR<0.01).
• Genome editing and genotyping: CRISPR/Cas9 generated mutants: os1900, os5100, os1900&os5100 double, and os900&os1400 double (Koshihikari background). For promoter dissection, 19 gRNAs targeting 5 kb upstream of Os1900 ATG were assembled (three gRNAs per construct) to create promoter deletion alleles (Nipponbare background). MAX1-like coding mutants were Sanger-genotyped (T0); promoter mutants screened by PCR for >100 bp deletions (T0) and sequenced in T1. Off-target assessments used k-mer coverage comparisons (k=20) against IRGSP1.0.
• Growth conditions and phenotyping: Plants grown under varied photoperiod/temperature: long/short day × high/low temperature. Tiller numbers were counted every 3 days across leaf ages. Fertilization levels: normal, minor, trace, and none (defined by proportions of fertilized vs non-fertilized soil). Liquid fertilization (Kimura’s B) used for expression assays; total dissolved solids (TDS) monitored. Statistical analyses used GLM/GLMM with ANOVA; Student’s t-tests for pairwise comparisons.
• Gene expression analyses: Real-time qPCR with ubiquitin normalization for Os900, Os1400, Os1900, Os5100 in shoots/roots and promoter mutants under defined TDS regimes (sampling at 0, 1, 3, 6, 24 h post-fertilization). RNA-seq (NovaSeq PE100) on leaves (7th blade) and first tiller bases (8-leaf stage) from WT and os1900&os5100 under minor fertilization; reads processed with Trimmomatic, STAR, RSEM, and edgeR (filterByExpr; glmQLFTest). Phytohormone-related gene sets interrogated; heatmaps via ggplot2.
• In situ hybridization: Localization of Os1900 and Os5100 transcripts in shoot apices (17 days after germination) before/after fertilization using DIG-labeled antisense probes; histone H4 as positive control.
• LC-MS/MS of SLs: Quantification of CL, CLA, MeCLA, and 4DO in shoots and roots of 32-day-old seedlings (activated by Pi starvation) using SPE fractionation and X500R QTOF LC-MS/MS with stable isotope-labeled internal standards.
• Bioassay: Application of synthetic SL analog (+)-GR24 (0–1 μM) in hydroponic culture to assess axillary bud inhibition (first and second tiller lengths measured).

• Field transcriptomics identified 107 fertilizer-responsive genes in rice leaves under natural paddy conditions; Os1900 (MAX1-like) stood out with high FC and FC/SD and was repressed by fertilization in leaves.
• Genetic analyses revealed functional redundancy and specificity: os1900 or os5100 single mutants did not significantly increase tiller numbers under normal fertilization, whereas the os1900&os5100 double mutant showed a robust and significant increase in tillers across diverse photoperiod and temperature regimes (e.g., GLMM/ANOVA P=2.038e-10 for tiller number vs WT across leaf ages). Increased tillers were mainly secondary and tertiary.
• In contrast, os900&os1400 double mutants exhibited a much smaller effect on tiller number under normal fertilizer, and under minor/trace fertilization their tiller numbers were similar to WT, indicating Os1900&Os5100 dominate tiller control.
• Biochemical profiling (LC-MS/MS) showed massive accumulation of carlactone (CL) in os1900&os5100: shoots ~92,000 pg/g fresh weight (P≈2.1e-05 vs WT) and elevated in roots (P=0.0003). CLA levels were very low in mutants (shoots and roots; both significantly reduced vs WT), indicating a block in CL→CLA conversion. Small amounts of 4DO were detected in mutant shoots (none in WT shoots) and higher in mutant roots, consistent with Os900 activity using accumulated CL; ORO was undetectable.
• MAX1-like gene expression: Os900 and Os1400 transcript levels were similar between WT and os1900&os5100 in shoots and roots, suggesting the CL accumulation difference (shoots>roots) reflects tissue-specific expression/consumption; Os1900 and Os5100 transcripts were reduced in mutants (likely due to nonsense-mediated decay).
• GR24 complementation: (+)-GR24 inhibited axillary bud elongation in os1900&os5100, confirming SL deficiency phenotype; inhibition was weaker at low GR24 concentration.
• Spatial expression: In situ hybridization localized Os1900 and Os5100 transcripts to axillary bud apex regions at the tiller base; Os5100 showed a slightly distinct pattern near vascular tissues.
• Transcriptome consequences: In os1900&os5100 leaves and tiller bases, SL signaling and branching regulators were altered: D53 (SL signaling repressor) decreased; D14 (SL receptor) slightly increased; OsTB1 (branching suppressor) downregulated. Broader hormonal networks were modulated: cytokinin biosynthesis/signaling genes (LOG/LOGL, RR2) upregulated; CKX genes downregulated; auxin transport genes (PIN5a, PIN9, AUX4) downregulated; ABA/JA signaling components (SAPKs, PP2C, JAZ) upregulated; SA-related genes induced.
• Os1900 fertilizer responsiveness: Under controlled TDS regimes, Os1900 expression remained low with moderate fertilization and increased as TDS declined (e.g., pronounced induction at 30 ppm TDS), then rapidly decreased to basal within 1 h after fertilization pulses.
• Promoter cis-regulatory dissection: Nine Os1900 promoter deletion alleles (M1–M9) spanning 5 kb upstream displayed distinct baseline expression and fertilizer responses. Some deletions overlapped predicted nutrient-responsive cis-elements (WRKY/MYB family sites; P1BS/PHR1-binding motifs). M4 showed consistently reduced Os1900 expression across conditions; M3, M6, and M9 showed attenuated nutrient-deficiency induction at 30 ppm.
• Agronomic outcomes under reduced fertilization: Under minor fertilization, M3, M4, and M6 had increased tiller numbers vs WT; under trace fertilization, M1 and M6 had slight decreases. Grain yield assessments indicated only M4 significantly increased yield under laboratory conditions, whereas M3 and M6 suffered reduced seed fertility despite higher tillering. Collectively, Os1900 and Os5100 function in vivo to mediate CL→CLA conversion central to SL biosynthesis controlling tillering; fertilization regulates tiller number via Os1900 transcription, and specific promoter deletions (e.g., M4) can enhance tillering and yield under low fertilizer.

The findings directly address the hypothesis that fertilization modulates tiller number through transcriptional regulation of a MAX1-like gene. Os1900 is a fertilizer-responsive gene whose expression is repressed by fertilization; when Os1900 and its close homolog Os5100 are disabled, CL accumulates and active SL production is curtailed, leading to enhanced tillering. This positions Os1900/Os5100 as key in vivo catalysts of the CL→CLA step in rice SL biosynthesis, a role previously less appreciated relative to Os900/Os1400 based on in vitro data and QTL analyses. The contrasting minor tillering effects of os900&os1400 double mutants imply Os900/Os1400 contribute more to rhizospheric SL signaling and natural variation in SL profiles than to primary control of tillering under standard cultivation. Promoter engineering demonstrates that cis-regulatory variation in Os1900 can tune fertilizer responsiveness and baseline expression, thereby modulating tiller number and, in the case of M4, improving yield under reduced fertilizer. The RNA-seq results show that perturbing SL biosynthesis via Os1900/Os5100 impacts a broad phytohormone network (cytokinin, auxin transport, ABA, JA, SA), consistent with known crosstalk in branching control and reinforcing the centrality of SLs in integrating nutrient status with shoot architecture. Spatial expression at tiller bases supports a model where local CL-to-CLA conversion and potential shoot-derived SL flux contribute to axillary bud dormancy and elongation decisions. Thus, fertilizer signals likely repress Os1900 transcription via specific cis-elements, reducing SL deficiency cues, and maintaining axillary bud dormancy; under nutrient scarcity, Os1900 induction promotes SL biosynthesis to suppress branching. Targeted promoter editing can decouple this response to sustain productive tillering under low-input systems.

This work identifies Os1900 as a central fertilizer-responsive regulator of rice tiller number and establishes, genetically and biochemically, that Os1900 together with Os5100 catalyzes the CL→CLA step in SL biosynthesis in vivo. Field-derived transcriptomics pinpointed Os1900 among 107 fertilizer-responsive genes, and CRISPR mutants revealed that only the os1900&os5100 double mutant exhibits strong tiller increases, while os900&os1400 showed limited effects on tillering. Massive CL accumulation and low CLA in the double mutant confirm the blocked step, and GR24 rescue supports SL deficiency as the cause of increased tillering. Promoter dissection uncovered cis-regulatory regions that govern Os1900’s fertilizer responsiveness; notably, the M4 deletion reduces Os1900 expression, increases tiller number, and improves grain yield under minor fertilization. These results demonstrate a route to breeding rice with lower fertilizer demand via promoter engineering of Os1900. Future directions include: precise mapping and functional validation of cis-elements and trans-factors (e.g., PHR1/P1BS, WRKY/MYB) mediating fertilizer responses; spatiotemporal localization of Os1900/Os5100 proteins and SL intermediates in axillary tissues; evaluation of promoter alleles across diverse genetic backgrounds and environments in field trials; and assessment of impacts on AM symbiosis and nutrient acquisition.

• Yield and agronomic performance were primarily assessed under controlled/laboratory conditions; comprehensive multi-location paddy field trials are needed to validate stability and benefits of promoter alleles (e.g., M4).
• The exact cis-regulatory elements and binding transcription factors controlling Os1900’s fertilizer response remain to be functionally validated; predictive motif overlaps (e.g., P1BS) did not uniformly explain expression phenotypes.
• The study focused on japonica cultivars (Koshihikari, Nipponbare); natural variation and transferability to indica and other backgrounds require testing.
• Potential pleiotropic effects of promoter deletions (e.g., seed fertility reductions in some alleles) indicate trade-offs that must be optimized.
• While off-target editing was examined for MAX1-like loci via k-mer analysis, broader genome-wide off-target effects were not exhaustively ruled out.
• Rhizospheric signaling roles (e.g., AM symbiosis efficacy) and environmental interactions (P availability dynamics) were not directly measured.