Evolution of the JULGI–SMXL4/5 module for phloem development in angiosperms

Significance

Seed plants (gymnosperms and angiosperms) share several developmental characteristics, including lateral branching and phloem formation from the cambium. Gymnosperms are perennial woody plants with a main stem showing apical dominance, whereas angiosperms exhibit a variety of growth and branching habits and can be herbaceous, woody, annual, or perennial plants. We determined that in angiosperms, the mechanisms regulating phloem formation have been uncoupled from those regulating lateral branching. This uncoupling allows angiosperms to respond with more plasticity to environmental and external stimuli through modifications in branching.

Abstract

Bifacial cambium, which produces xylem and phloem, and monopodial architecture, characterized by apical dominance and lateral branching from axillary buds, are key developmental features of seed plants, consisting of angiosperms and gymnosperms. These allow seed plants to adapt to diverse environments by optimizing resource allocation and structural integrity. In seed plants, SUPPRESSOR OF MAX2-LIKE (SMXL) family members function in phloem development and strigolactone-induced inhibition of axillary bud outgrowth. Although strigolactone signaling regulates most SMXL family members, the only known regulator of SMXL4 and SMXL5 is the RNA-binding protein JULGI. We demonstrate that in angiosperms, by directly regulating SMXL4/5 expression, JULGI uncouples SMXL4/5 activity from strigolactone signaling. JULGI and ancestral SMXLs from seedless vascular plants or SMXL4/5 from seed plants are coexpressed in the phloem tissues of vascular plants, from lycophytes to angiosperms. Core angiosperm SMXL4/5 mRNAs contain a G-rich element in the 5′ untranslated region (UTR) that serves as a target sequence for JULGI to negatively regulate SMXL4/5 expression. Heterologous expression of JULGIs from various angiosperms rescued the Arabidopsis jul1 jul2 mutant. Expressing SMXL4/5s from seed plants and ancestral SMXLs rescued Arabidopsis smxl4 smxl5. Angiosperm SMXL4/5s lack an RGKT motif for proteasomal degradation. Indeed, treatment with the synthetic strigolactone analog rac-GR24 induced proteasomal degradation of SMXL from ferns and SMXL5a from gymnosperms, but not SMXL4/5 from angiosperms. These findings suggest that in ancestral angiosperms, the 5′ UTR of SMXL4/5 gained G-rich elements, creating a regulatory module with JULGI that allows the phloem development pathway to act independently of strigolactone signaling.

Vascular plants appeared in the Early Devonian period, approximately 415 Mya (1). Like extant seedless vascular plants (lycophytes and monilophytes), these early vascular plants contained tracheids and sieve elements, but they lacked companion cells. Present-day tree ferns and fossils of lycophyte tree species show a woody habit. Notably, these seedless vascular plants did not undergo secondary phloem development and exhibited terminal branching (2, 3). The origin of bifacial cambium (producing xylem and phloem on opposite sides of the cambial cells) and monopodial architecture, characterized by a single main stem with lateral branching, traces back to the extinct progymnosperms, marking a significant event in plant evolution (4, 5). Subsequently, this cambium became a characteristic of gymnosperms and most angiosperms, supporting bifacial secondary growth (6, 7). Progymnosperms also showed an architectural shift from terminal branching to lateral branching (5, 8). Complete monopodial architecture, which involves apical dominance and lateral branching from axillary buds, occurs exclusively in seed plants and was established in gymnosperm conifers (5). Seed plants have complex vascular structures and exhibit architectural plasticity, ranging from creeping herbaceous plants to towering canopy trees that can exceed 100 m in height (9, 10).

Angiosperms exhibit greater developmental plasticity than gymnosperms which are perennial woody plants with secondary vascular development and strong apical dominance (11). Evidence suggests that ancestral arborescent angiosperms evolved into contemporary species with diverse architectures and patterns of vascular development (12–17). Among extant plants, members of orders such as the Amborellales and Austrobaileyales as well as the superorder Magnoliidae, often referred to as “basal” angiosperms, exhibit arborescent habits (12). Annual and biennial herbs, from magnoliids to numerous eudicot clades, employ a strategy to rapidly establish and dismantle shoot architecture by programmed senescence to ensure adequate reproduction within a limited time frame.

Comparative genomic studies suggest that the genetic modules involved in the phytohormonal signaling that governs vascular development and monopodial architecture either predate vascular plants or have been conserved from the earliest vascular plant taxa (4, 18–21). The signaling machinery of auxin, the primary hormonal regulator of vascular development and apical dominance, is conserved across nearly all vascular plants (22–24). The major signaling components for cytokinin, which acts antagonistically to auxin in root vascular patterning and promotes axillary bud outgrowth (25, 26), are conserved throughout bryophytes to seed plants (27–30). Strigolactone plays a key role in regulating plant architecture by inhibiting shoot branching (26, 31). The origin of strigolactone biosynthetic enzymes can be traced back to the origins of streptophytes (32, 33). The strigolactone receptor DWARF14 (D14) and the F-box protein MORE AXILLARY GROWTH2 (MAX2), which form the D14–SCF^MAX2 E3 ubiquitin ligase complex to degrade target proteins by the proteasome, were also conserved since green algae (34, 35). The target protein of the D14–SCF^MAX2 complex, which is encoded by a member of the SMXL gene family, has been conserved from bryophytes (36, 37).

The SMXL family diversified from a single ancestral clade in seedless vascular plants into multiple specialized clades during the evolution of seed plants. These clades are thought to have distinct roles, particularly in architecture formation and phloem development (38). In modern angiosperms, the SMXL4/5 clade is crucial for phloem development, and the SMXL6/7/8 clade is responsible for strigolactone signaling (37, 39). To date, the only known regulator of SMXL4/5 gene expression is JULGI, a RanBP2-type zinc finger protein that binds to a G-rich sequence in the 5′ untranslated region (UTR) of SMXL4/5 mRNA (40, 41). In Arabidopsis and tomato (Solanum lycopersicum), JULGI inhibits the translation of SMXL4/5 mRNA via a G-rich element in its 5′ UTR, and downregulating JULGI significantly increased phloem development (40, 41). Therefore, identifying the evolutionary point at which JULGI and SMXL formed a regulatory module will shed light on the coordination of vascular development, monopodial architecture, and strigolactone signaling during the evolution of seed plants.

In this study, we identified distinctive G-rich elements in the 5′ UTRs of core angiosperm SMXL4/5 homologs. These elements are targeted by angiosperm JULGI proteins. Heterologous expression of JULGIs from various angiosperms suppressed downstream gene expression and mitigated the excessive primary phloem development characteristic of the Arabidopsis jul1 jul2 knockout mutant. Gymnosperm SMXL4/5s possess an RGKT degradation motif, which mediates the proteasomal degradation of SMXL family proteins (36). These proteins were degraded in the presence of strigolactone, but angiosperm SMXL4/5 proteins without this motif were not affected. Overall, these findings suggest that the JULGI-SMXL4/5 module regulates the accumulation of SMXL4/5 in angiosperms, thus ensuring appropriate phloem development amid strigolactone signaling.

Results

Phylogenetic Analysis Reveals the Emergence of JULGI-SMXL4/5 Module Components During the Evolution of Angiosperms.

To better understand the conservation of JULGI-SMXL4/5 module components throughout the evolution of vascular plants, we conducted phylogenetic analysis using sequences from 52 species of Viridiplantae (green plants). We aligned the protein sequences of JULGI homologs and generated a phylogenetic tree (SI Appendix, Fig. S1). All identified JULGI homologs from vascular plants formed a monophyletic group, indicating that these homologs were inherited from the common ancestor of vascular plants. Some nonvascular plants also contain JULGI homologs, representing potential ancestral forms of JULGI. All JULGI homologous proteins contain three zinc finger domains connected in tandem (Fig. 1). The RNA-binding and base-stacking residues in the RanBP2-type zinc finger domain are well documented (42). Except for OlJUL, these residues were conserved among all JULGI homologs examined (Fig. 1). We also observed a conserved proline residue in the first zinc finger of JULGI homologs in seedless vascular plants, bryophytes, and green algae at a position away from the RNA-binding and base-stacking residues (Fig. 1). Changes in several residues, including the position of proline residues, rather than large-scale domain composition, may have allowed ancestral JULGI homologs to gain binding specificity for G-rich single-stranded RNA (43).

Fig. 1.

Phylogenetic tree of representative JULGI homolog proteins depicted with first zinc fingers. The RNA-binding and base-stacking residues within the peptide sequences are designated with red wedges. Proline residues exclusively appearing among the seedless plants’ JULGI proteins are designated with blue wedges.

An examination of the phylogenetic diversity of SMXL homologs revealed that these homologs are exclusively present in land plant taxa (SI Appendix, Fig. S2). Based on prior phylogenetic studies of SMXL homologs (37), we annotated the clades as the bryophytes, lycophytes, monilophytes, SMAX1/SMXL2, SMXL4/5, SMXL3/9, and SMXL6/7/8 clades (SI Appendix, Fig. S2A). Ancestral SMXLs (bryophyte, lycophyte, and monilophyte homologs) formed a sister relationship with either the SMAX1/SMXL2 or SMXL4/5 clades. The SMXL4/5 clade was found only in seed plant taxa, suggesting its neofunctionalization in seed plants (39). The SMXL6/7/8 clade was exclusively composed of angiosperm taxa, suggesting that this clade evolved from ancestral angiosperms and now includes a new function in response to strigolactone signaling (34, 44).

For each SMXL homolog, we analyzed the genomic region 50-bp upstream of the translation start site to identify putative binding sites for JULGI. Consensus sequence analysis revealed that G-rich sequences are predominantly located 7- to 35-bp upstream from the ATG translation start codon within the 5′ UTRs of core angiosperm SMXL4/5 genes (Fig. 2 and SI Appendix, Table S1 and Fig. S2B; hereafter referred to as G-rich 5′ UTRs). We hypothesized that the JULGI-SMXL4/5 module has evolved to regulate the levels of SMXLs (41). To further understand how the regulation of SMXLs has evolved, we looked for RGKT motifs in the sequences of SMXL homologs. These motifs act as degradation motifs for SMAX1/SMXL2 and SMXL6/7/8 via the proteasome in response to strigolactone or karrikin signaling (45). The RGKT motifs were conserved in ancestral SMXLs, SMAX1/SMXL2s, and gymnosperm SMXL4/5s (Fig. 2). However, they were not conserved in angiosperm SMXL4/5 or angiosperm-specific SMXL3/9 clade members (Fig. 2), which is consistent with prior descriptions (37, 46). In summary, only angiosperms exhibited all three components of the JULGI-SMXL4/5 module: JULGI, SMXL4/5, and the G-rich 5′ UTR in SMXL4/5 mRNA. Whereas JULGI genes are conserved in all vascular plants, the G-rich 5′ UTR of SMXL4/5 represents a unique trait found only in angiosperms.

Fig. 2.

Phylogenetic tree of representative SMXL homolog proteins depicted with expression regulatory elements. RGKT motifs within the peptide sequences are shown with consensus sequence frequencies determined by at least 25% identity from the sequence alignment. Nucleotide sequences located 50-bp upstream from the translation start codon are illustrated and denoted as putative 5′ UTR. G-rich elements in angiosperm SMXL4/5 5′ UTR are shown in bold.

Angiosperm JULGI Homologs Share Conserved Molecular Functions with AtJUL1.

To investigate whether JULGIs from various plant taxa share molecular functions with Arabidopsis JULGI1 (AtJUL1), we performed a green fluorescent protein (GFP) reporter assay using a protoplast transient expression system (40). We coexpressed the GFP open reading frame (ORF) under the control of the AtSMXL5 5′ UTR with each JULGI homolog from a variety of taxa, including nonvascular plants. JULGIs from Zea mays, Cinnamomum kanehirae, Amborella trichopoda, and Ginkgo biloba suppressed the expression of GFP via the AtSMXL5 5′ UTR, similar to AtJUL1 (Fig. 3A and SI Appendix, Fig. S3A). Other JULGI homologs from Oryza sativa, Thuja plicata, Ceratopteris richardii, Selaginella moellendorffii, Sphagnum fallax, Marchantia polymorpha, Klebsormidium nitens, and Ostreococcus lucimarinus showed less translational suppression activity compared to AtJUL1 (Fig. 3A and SI Appendix, Fig. S3A). Although JULGI homologs were expressed in protoplasts using the cauliflower mosaic virus 35S promoter, variations in the expression levels of each homolog were observed (Fig. 3A). To assess the suppressive activity of each JULGI homolog in protoplasts, GFP fold changes were normalized to their respective expression levels. The suppressive activities of OsJUL2, ZmJUL1, CkJUL, AmtJUL, GbJUL, CrJUL1, and SmJUL1 were found to be comparable to that of AtJUL1 on the AtSMXL5 5′ UTR (SI Appendix, Fig. S3B). In summary, the protoplast GFP suppression assay revealed that among all JULGI homologs we investigated, SfJUL1, MpJUL, and OlJUL had no suppressive activities.

Fig. 3.

Conservation of molecular activity in JULGI homologs from angiosperms. (A) Translation suppression activity of JULGI homologs in GFP protoplast reporter assays. GFP reporter under AtSMXL5 5′ UTR coexpressed with JULGI homologs (Left). (Scale bar, 200 μm.) Immunoblotting shows GFP and JULGI proteins (Right). The Rubisco large subunit (RbcL) is a loading control. (B) Complementation of the Arabidopsis jul1 jul2 mutant by JULGI homologs expressed under the AtJUL1 promoter. Cross-sections of stem bases show recovery on phloem development (Top). (Scale bars, 200 μm.) Representative rosettes (Middle) and aerial tissues (Bottom) show developmental recovery. Scale bars, 1 and 5 cm, respectively. (C) Quantification of phloem development in JULGI-expressing jul1 jul2 plants (mean ± SD, n = 4 to 5). Different letters indicate significant differences (P < 0.05, one-way ANOVA, Tukey’s HSD test). (D) Protoplast GFP reporter assays with domain-swap chimeric JULGI proteins. AtJUL1, SmJUL1, and domain-swap chimeric proteins (AtZF-SmL, SmZF-AtL) tested as effectors (Left). (Scale bar, 200 μm.) Immunoblotting shows GFP and JULGI effectors (Right). RbcL is a loading control. (E) Protoplast GFP reporter assays with point-mutated JULGI proteins. AtJUL1, CrJUL1, and point-mutated proteins (AtJUL1^R10P, CrJUL1^P11R) tested as effectors (Left). (Scale bar, 200 μm.) Immunoblotting shows GFP and JULGI effectors (Right). RbcL is a loading control.

To investigate the genetic functions of JULGI homologs in planta, we attempted to complement Arabidopsis jul mutants by heterologous expression of these homologs. We mutated AtJUL1 in the Arabidopsis jul2 knockout mutant using CRISPR-Cas9-mediated genome editing, creating the loss-of-function jul1 jul2 double mutant (SI Appendix, Fig. S4 A–D). The jul1 jul2 mutant exhibited several abnormalities, including curly and bumpy leaves, short inflorescences, increased branching, and decreased rosette diameter (SI Appendix, Fig. S4 E and F). Furthermore, this mutant showed excessive phloem development, with interconnected fascicular phloem tissues forming a cylindrical ring akin to secondary phloem development. By contrast, wild-type (Col-0) plants at the same stage of development displayed discrete vascular bundles and showed no evidence of secondary vascular development (SI Appendix, Fig. S4G).

We then generated JULGI-expressing lines in the jul1 jul2 background driven by the AtJUL1 promoter. Despite variations in the expression levels of JULGI homologs, heterologously expressing the angiosperm JULGIs, GbJUL, and KnJUL (hereafter referred to as active JULGIs) rescued the jul1 jul2 phenotype (Fig. 3B and SI Appendix, Fig. S5). Moreover, the short inflorescence internode phenotype was reversed, allowing the inflorescence stems to grow longer than those of jul1 jul2 at the same developmental stage (Fig. 3B). In jul1 jul2 plants expressing active JULGIs, the vascular bundles in the inflorescence stem showed a reduction in phloem tissue area relative to jul1 jul2 plants (Fig. 3C). These transgenic plants exhibited an interfascicular region devoid of phloem tissue, with phloem tissues restricted to individual vascular bundles, instead of forming a cylindrical arrangement across the stele (Fig. 3B). Conversely, plants expressing either TpJUL, CrJUL1, SmJUL1, SfJUL, MpJUL, or OlJUL (collectively referred to as inactive JULGIs) displayed excessive phloem development extending into the interfascicular region, similar to what was observed in the jul1 jul2 mutants (Fig. 3B). These results imply that the genetic function of JULGI in vascular tissue is to suppress phloem development, thereby prohibiting the excessive accumulation of phloem tissue in inflorescence stems.

To determine which residues caused the differences in translational suppression activity among JULGI homologs, we examined sequence alignments of regions corresponding to the three zinc fingers and the two linker domains in JULGI. Notably, the first linker domains connecting the first and second zinc fingers of JULGI homologs in nonangiosperms were relatively long (SI Appendix, Fig. S6). To examine the effect of these linkers, we created chimeric proteins with swapped domains that had either AtJUL1 zinc fingers with SmJUL1 linkers (AtZF-SmL) or vice versa (SmZF-AtL). These chimeric proteins were expressed in protoplasts together with GFP under the control of the AtSMXL5 5′ UTR. The translational suppression and protein expression levels of AtZF-SmL and SmZF-AtL were similar to those of AtJUL1 and SmJUL1, respectively, indicating that linker domains have limited effects on translational suppression (Fig. 3D and SI Appendix, Fig. S7 A and B). Next, we investigated the effect of proline residues in the first zinc fingers of JULGIs from seedless vascular plants. We expressed point-mutated JULGI proteins in which the 10th arginine of AtJUL1 was substituted with proline (AtJUL1^R10P) and the 11th proline of CrJUL1 was changed to arginine (CrJUL1^P11R) in protoplasts, together with GFP under the control of the AtSMXL5 5′ UTR. AtJUL1^R10P exhibited a slightly reduced ability to suppress gene expression compared to AtJUL1, while CrJUL1^P11R showed slightly increased activity compared to CrJUL1 (Fig. 3E and SI Appendix, Fig. S7C). As mentioned, we observed disparities in protein levels among JULGI homologs in the protoplast GFP reporter assays, despite using the same cauliflower mosaic virus 35S promoter (Fig. 3A). Protein levels appeared to be dependent on the presence of the zinc finger domains rather than the linker domains (Fig. 3D and SI Appendix, Fig. S7B). Thus, a posttranscriptional regulatory mechanism unique to each JULGI homolog must exist, at least in Arabidopsis. In summary, the zinc finger domains of JULGI homologs is critical to determine the genetic functions of JULGI proteins, such as molecular activities and expression levels in planta. Active JULGIs expressed under the control of the AtJUL1 promoter successfully complemented the developmental defects observed in jul1 jul2 plants. This strongly suggests that active JULGIs exhibit conserved molecular activity comparable to AtJUL1, with similar genetic effects in Arabidopsis.

JULGI-SMXL4/5 Is a Functional Genetic Module that Directs Phloem Development in Angiosperms.

To elucidate the action of JULGI and SMXL4/5 as a genetic module, we first observed the localization of Arabidopsis JUL1 and SMXL5 expression using AtJUL1_pro:3xGFP and AtSMXL5_pro:erYFP transcriptional fusion reporters (47, 48). Expression of each AtJUL1 and AtSMXL5 was observed not only in the fascicular phloem bundles but also in the interfascicular phloem of each inflorescence stem (Fig. 4A), pointing to a regulatory role for JULGI-SMXL4/5 in phloem development during both primary and secondary growth.

Fig. 4.

Expression of JULGI, ancestral SMXL, and seed plant SMXL4/5 in phloem tissues. (A) Confocal images of Arabidopsis inflorescence stem bases expressing AtJUL1_pro:3xGFP and AtSMXL5_pro:erYFP. Cross-sections (0 to 0.5 mm above the ground) were cleared using Clear See. Red, green, and orange boxes in whole inflorescences indicate area of magnification below. X: xylem; P: phloem; F: fiber tissue. [Scale bars, 500 μm (whole inflorescences), 20 μm (vascular bundles and interfascicular tissues).] (B) In situ hybridization of JULGI, ancestral SMXL, and SMXL4/5 in vascular plants. Cross-sections from A. thaliana, O. sativa, G. biloba, C. richardii, and S. moellendorffii were hybridized with digoxigenin-labeled antisense or sense RNA probes showing dark blue signals. X: xylem; P: phloem. (Scale bars, 100 μm.)

To determine whether JULGI and SMXL4/5 are also coexpressed in the homologous tissues of each taxon, we performed in situ hybridization using developing vascular tissues from model plants representing each major group of vascular plant probed with JULGI and SMXL antisense RNA probes. We collected inflorescence stems from Arabidopsis and O. sativa, shoot subapical regions from G. biloba and S. moellendorffii, and growing petioles from C. richardii and hybridized them with the RNA probes. Signals from all antisense probes of both JULGI and SMXL were detected in the phloem tissues of all species examined, except for GbSMXL5a (Fig. 4B). Signals from JUL1 and SMXL5 in Arabidopsis, and JULGI in G. biloba, were detected in the phloem near the cambium. In O. sativa and C. richardii, signals were detected in the phloem within each bundle and the parenchymal tissue surrounding the bundle. In S. moellendorffii, signals were present throughout the phloem tissue surrounding the xylem inside the protostele. CrSMXL and SmSMXL, which lack the G-rich elements in their 5′ UTRs, were localized to developing phloem tissue. This raises the possibility that, even before JULGI-SMXL4/5 module formation, the ancestral SMXLs may have functioned in phloem development. Unfortunately, we detected no signal in vascular tissue using GbSMXL5a probes, although the presence of mRNA could be confirmed by RT-PCR (SI Appendix, Fig. S8 A and B). Interestingly, for the ancestral SMXLs and SMXL4/5, sense probes also showed signals in the phloem tissue area (Fig. 4B and SI Appendix, Fig. S8B). In Arabidopsis, SMXL4/5 are posttranscriptionally silenced by antisense transcription via RNA-DEPENDENT RNA POLYMERASE 6 (RDR6), followed by the production of small RNA via DICER-LIKE 2 (DCL2) (49). This suggests that the silencing mechanism may have been functional even before the diversification of ancestral SMXL throughout vascular plants.

The G-rich 5′ UTR of SMXL4/5 Homologs Is A Functional Component of the JULGI-SMXL4/5 Module.

To confirm that the G-rich element within the 5′ UTR of AtSMXL5 directly regulates phloem development in Arabidopsis, we deleted the G-rich element in the SMXL5 5′ UTR in the Arabidopsis genome using CRISPR/Cas9. We obtained a mutant (referred to as Δ39) characterized by the truncation of 39 nucleotides within the SMXL5 5′ UTR, including the G-rich element (SI Appendix, Fig. S9 A and B). The Δ39 mutant had longer primary roots compared to Col-0 and jul1 jul2 plants (Fig. 5 A and D). But, Δ39 plants exhibited a marginal increase in phloem area compared to Col-0 (Fig. 5 B and C). Both Δ39 and jul1 jul2 plants showed reduced foliar dimension (leaf length versus width) compared to Col-0, whereas the leaf area of Δ39 was greater than that of Col-0 and jul1 jul2 (Fig. 5 B, E, and F). Intriguingly, deletion of the G-rich element did not affect the transcription levels of either SMXL5 or its orthologs, SMXL3 and SMXL4 (SI Appendix, Fig. S10) (50, 51). These findings imply that the disruption of the G-rich element within the 5′ UTR of SMXL5 affected plant development. However, given that Δ39 plants did not phenocopy jul1 jul2 plants, especially regarding phloem development, and considering that SMXL5 is currently the only identified downstream target of JULGI, additional targets of JULGI likely exist in the Arabidopsis genome. To determine that the Δ39 mutation affects the translation, we coexpressed wild-type and Δ39 AtSMXL5 5′ UTR:GFP with AtJUL1 in protoplasts. AtJUL1 was unable to suppress GFP expression through the Δ39-mutated AtSMXL5 5′ UTR (SI Appendix, Fig. S9 D and E). In summary, the G-rich element within the AtSMXL5 5′ UTR is a genetic element that regulates the expression of the downstream ORF.

Fig. 5.

The G-rich 5′ UTR is essential for JULGI-SMXL4/5 modulation in core angiosperms. (A) Seven-day-old seedlings of Col-0, jul1 jul2, smxl4 smxl5, and Δ39. (Scale bar, 1 cm.) (B) Δ39 mutants show pleiotropic phenotypes, including increased phloem development (stem cross-sections, first column; Scale bar, 100 μm.) and altered aerial tissues. (rosettes, aerial tissues, and sixth rosette leaves; second to fourth column; Scale bars, 1 cm, 5 cm, and 1 cm, respectively.) Plants were grown under short days (SD) for 4 wk and long days (LD) for 2 wk for rosette leaves, and 4 wk for aerial tissue. (C) Quantification of phloem development as phloem tissue area/total stem area (mean ± SD, n = 14 to 15). (D) Root length of 7-d-old seedlings (mean ± SD, n = 8 to 12). (E and F) Sixth leaf dimensions (length-to-width ratio) and areas from plants grown under SD (4 wk) and LD (2 wk) (n = 8 to 10). For C–F, different letters indicate significant differences (P < 0.05, one-way ANOVA, Tukey’s HSD). (G) Inflorescence stem cross-sections from smxl4 sxml5 plants expressing AtSMXL5 under the AtSMXL5 promoter with or without the G-rich 5′ UTR. Light microscopy (Top Left; Scale bar, 200 μm.) and transmission electron microscopy (Top Right; Scale bars, 10 μm.) show sieve elements (red), companion/parenchymal cells (yellow), bundle sheath (purple), and cambium (green). SMXL5 protein levels were determined by immunoblotting (Bottom). RbcL is a loading control. mRNA levels were determined by reverse transcription-PCR. WT: AtSMXL5_pro:AtSMXL5-HA smxl4 smxl5; ΔG: AtSMXL5ΔG_pro:AtSMXL5-HA smxl4 smxl5. (H) Protoplasts GFP reporter assays with the 5′ UTR from SMXL4/5s and ancestral SMXLs with their respective JULGI (Left). (Scale bar, 200 μm.) JULGI and GFP were determined by immunoblotting (Right). RbcL is a loading control.

To assess the effect of SMXL5 accumulation in planta, we ectopically expressed AtSMXL5 in the Arabidopsis smxl4 smxl5 mutant under the control of its own promoter (lacking the G-rich element, AtSMXL5ΔG_pro) (SI Appendix, Fig. S9C). The Col-0 wild-type and smxl4 smxl5 plants expressing AtSMXL5 under the control of AtSMXL5_pro contained discrete fascicular phloem bundles (Fig. 5G), whereas smxl4 smxl5 plants expressing AtSMXL5 under the control of AtSMXL5ΔG_pro exhibited the vascular phenotype of jul1 jul2, characterized by notably enlarged phloem tissues forming a cylindrical structure across the stele (Fig. 5G). These results point to a link between phloem cell differentiation rates and the overaccumulation of SMXL5. Notably, the AtSMXL5ΔG_pro:AtSMXL5 smxl4 smxl5 plants showed higher translation efficiency compared to the AtSMXL5_pro:AtSMXL5 smxl4 smxl5 plants, as demonstrated by the ratio of transgenic protein and RNA levels (SI Appendix, Fig. S9F). To investigate whether the G-rich 5′ UTRs in SMXL4/5 homologs from angiosperm taxa could suppress the expression of the downstream ORF via AtJUL1 (Fig. 1 and SI Appendix, Table S1), we performed GFP reporter assays in protoplasts by coexpressing GFP conjugated to the 5′ UTR of each SMXL4/5 homolog or ancestral SMXL homolog along with AtJUL1. The G-rich 5′ UTRs of SMXL4/5 homologs from Arabidopsis, O. sativa, Z. mays, and Aristolochia contorta suppressed downstream GFP expression in the presence of AtJUL1. In contrast, the 5′ UTRs of SMXL4/5 from C. kanehirae, Liriodendron tulipifera, A. trichopoda, Nymphaea colorata, Picea abies, Pinus taeda, Thuja plicata, and G. biloba, and the 5′ UTRs of ancestral SMXL from C. richardii, Alsophila spinulosa, Azolla filiculoides, S. moellendorffii, and Diphasiastrum complanatum did not exhibit this suppression (SI Appendix, Fig. S11 A and B). These results indicate that the degree of suppression of SMXL4/5 by the 5′ UTR is directly related to its phylogenetic proximity to AtSMXL5.

We then used a protoplast transient expression system to examine whether angiosperm JULGI could suppress the expression of its corresponding SMXL4/5 via the G-rich 5′ UTRs. The GFP ORF under the control of the 5′ UTR from each seed plant SMXL4/5 and seedless vascular plant ancestral SMXL homolog was coexpressed with its respective JULGI. The 5′ UTRs of AtSMXL5, OsSMXL5a, ZmSMXL5a, and CkSMXL5 mediated the suppression of GFP reporter expression by their corresponding JULGI homologs (Fig. 5H and SI Appendix, Fig. S12). In contrast, the 5′ UTR from AmtSMXL5, TpSMXL5a, GbSMXL5a, CrSMXL, and SmSMXL did not inhibit reporter expression with its respective JULGI homolog. In summary, JULGI and SMXL4/5, which are coexpressed in developing phloem tissues from early vascular plants, formed a genetic module via the integrated G-rich element in the 5′ UTR of SMXL4/5 during the evolution of core angiosperms.

SMXL4/5 and Ancestral SMXL Can Replace the Function of AtSMXL5 in Arabidopsis.

Our phylogenetic analysis revealed the presence of RGKT degradation motifs in SMAX1/SMXL2 of seed plants, SMXL6/7/8 of angiosperms, SMXL4/5 of gymnosperms, and ancestral SMXL of nonseed plants but, intriguingly, not in SMXL4/5 of angiosperms (Fig. 2), which is consistent with previous findings (34, 37–39). RGKT degradation motifs are required to degrade SMXL protein via the SCF^MAX2 complex when D14 and KARRIKIN INSENSITIVE 2 (KAI2) receive strigolactone and karrikin ligands, respectively (44, 45, 50, 52, 53). The establishment of canonical strigolactone signaling, marked by the appearance of SMXL6/7/8 and the disappearance of the RGKT motif from SMXL4/5, coincides with the evolution to angiosperms (34). Therefore, we examined the genetic functionality of gymnosperm SMXL4/5 and ancestral SMXL before the gene duplication event for phloem development and reactivity to strigolactone and karrikin. To investigate whether gymnosperm SMXL4/5 homologs and ancestral SMXL homologs also possess conserved genetic functions in vivo, we expressed SMXL4/5 homologs from seed plants and ancestral SMXL homologs under the control of the AtSMXL5 intrinsic promoter in the smxl4 smxl5 double knockout mutant background. The smxl4 smxl5 mutant is characterized by the prominent presence of incompletely differentiated sieve elements (SI Appendix, Fig. S13A), resulting in short primary roots (Fig. 6A). Transgenic smxl4 smxl5 plants expressing AtSMXL5, OsSMXL5a, GbSMXL5a, CrSMXL, or SmSMXL exhibited significantly longer primary roots than smxl4 smxl5 plants (Fig. 6A). Consistent with these results, in the phloem tissue of SMXL-expressing smxl4 smxl5, differentiation of sieve elements was rescued from the defects seen in smxl4 sxml5 (SI Appendix, Fig. S13A). The expression levels of ALTERED PHLOEM DEVELOPMENT (APL), a phloem marker gene, were also higher in all SMXL-expressing smxl4 smxl5 plants compared to smxl4 smxl5 (SI Appendix, Fig. S13B), indicating a phenotypic recovery upon complementation with a single gene into the double smxl4 sxml5 mutant. AtSMAX1, another member of the SMAX1-like protein family that responds to both strigolactone and karrikin, was previously shown to substitute for AtSMXL5 when expressed under the control of the AtSMXL5 intrinsic promoter (51). Similarly, CrSMXL and SmSMXL, which are thought to be ancestral to the SMAX1/SMXL2 and SMXL4/5 clades, respectively (SI Appendix, Fig. S2A), compensated for the absence of SMXL5 in smxl4 smxl5 (Fig. 6A and SI Appendix, Fig. S13A). These results demonstrate that ancestral SMXL and SMXL4/5 homologs share similar genetic functions with AtSMXL5 in Arabidopsis, highlighting their conserved roles in phloem development across the vascular plant lineage (Fig. 4B).

Fig. 6.

Functional divergence in strigolactone sensitivity of ancestral SMXL, gymnosperm SMXL4/5, and angiosperm SMXL4/5 proteins. (A) Eight-day-old seedlings of Col-0, max2-3, smxl4 smxl5, and SMXL-expressing smxl4 smxl5 grown on mock or 2 μM rac-GR24 media. HA-tagged SMXL homologs were expressed under the AtSMXL5 promoter. (Scale bars, 1 cm.) (B) Root length quantification (mean ± SD, n = 9 to 39). Different letters indicate significant differences (P < 0.05, two-way ANOVA, Tukey’s HSD). (C) The stability of SMXL homolog proteins in SMXL-expressing smxl4 smxl5 seedlings treated with mock, 10 µM rac-GR24, or rac-GR24 with 50 µM MG-132 for 1 h. Immunoblotting shows SMXL levels with β-tubulin as a loading control (Top). Relative immunoblot signals were quantified (Bottom; mean ± SD, n = 3). Asterisks indicate significant differences (*P < 0.05, **P < 0.01, two-way ANOVA, Tukey’s HSD). (D) Model of the JULGI-SMXL4/5 module evolution. The translational suppressive activity of JULGI, the genetic function of SMXL, the loss of the RGKT motif in SMXL4/5, and the G-rich 5′ UTR have collectively been introduced to core angiosperms (Left). During the evolution of angiosperms, ancestral seed plant SMXL4/5 had lost the RGKT motif and gained the G-rich element within the 5′ UTR of its mRNA as a target site for JULGI, thereby completing the uncoupling of SMXL4/5 from strigolactone signaling (Right).

SMXL4/5 from Angiosperms Are Insensitive to Protein Degradation by Strigolactone and Karrikin.

We then assessed the proteolysis of ancestral SMXL and gymnosperm SMXL4/5 proteins in response to strigolactone or karrikin in Arabidopsis protoplasts expressing GFP-fused SMXL homologs. Within 1 h of treatment with rac-GR24, a synthetic strigolactone and karrikin analog (52), GFP signals from protoplasts expressing GbSMXL5a-GFP were drastically reduced, whereas GFP signals from angiosperm SMXL4/5s and ancestral SMXLs were not altered (SI Appendix, Fig. S14 A and B). We then examined Arabidopsis smxl4 smxl5 plants expressing SMXL homologs treated with rac-GR24. If rac-GR24 induces the degradation of the introduced SMXLs, the complemented seedlings should revert to the stunted growth pattern characteristic of smxl4 smxl5. Root growth was slightly reduced in seedlings of Col-0, smxl4 smxl5 expressing angiosperm SMXL4/5, and smxl4 smxl5 expressing ancestral SMXL grown on 2 µM rac-GR24, but the roots of these seedlings were still longer than those of the smxl4 smxl5 mutant (Fig. 6 A and B). However, the root growth of the max2-3 mutant, which is insensitive to strigolactone, was not affected by rac-GR24 treatment. These results indicate that the mild growth reduction observed in seedlings of Col-0 and angiosperm SMXL5-expressing smxl4 smxl5, and ancestral SMXL-expressing smxl4 smxl5 was due to the canonical strigolactone signaling and was not related to SMXL4/5. Notably, rac-GR24 treatment reverted the root growth of GbSMXL5a-expressing smxl4 smxl5 seedlings to levels similar to those of smxl4 smxl5. To confirm that the reduced root growth of SMXL-expressing smxl4 smxl5 seedlings was due to proteasomal degradation induced by strigolactone signaling, we treated the seedlings with rac-GR24 together with MG-132, a proteasome inhibitor. The levels of both GbSMXL5a and CrSMXL were reduced by rac-GR24 treatment, and these reductions were blocked by MG-132 treatment (Fig. 6C). In contrast, the levels of angiosperm SMXL4/5 did not change due to rac-GR24 treatment (Fig. 6C). To further investigate the genetic functions of SMXLs harboring the RGKT motif, multiple lines of GbSMXL5a, CrSMXL, and SmSMXL-expressing smxl4 smxl5 seedlings were grown under 2 µM rac-GR24. All GbSMXL5a-expressing smxl4 smxl5 roots were reverted to the smxl4 smxl5 root phenotype in response to rac-GR24 treatment, whereas all SmSMXL-expressing smxl4 smxl5 roots did not (SI Appendix, Fig. S15 A and B). Interestingly, two out of five CrSMXL-expressing smxl4 smxl5 lines were not reverted to the smxl4 smxl5 root phenotype in response to 2 μM rac-GR24. However, these two insensitive lines were also reverted to the smxl4 smxl5 root phenotype when grown under 10 μM rac-GR24, while wild-type and AtSMXL5-expressing smxl4 smxl5 roots remained longer than those of smxl4 smxl5 under 10 μM rac-GR24 (SI Appendix, Fig. S15 C and D). Consistently, under 10 μM rac-GR24, CrSMXL protein levels were dramatically reduced, whereas AtSMXL5 protein levels remained unchanged (SI Appendix, Fig. S15 E and F).

To investigate whether the JULGI-SMXL4/5 module suppresses SMXL expression in response to rac-GR24 treatment, we measured the transcript levels of JULGI-SMXL4/5 module components in seedlings treated with rac-GR24. The expression of strigolactone-responsive BRANCHED1 (BRC1) was induced by rac-GR24 to a similar degree in Col-0, jul1 jul2, and smxl4 smxl5 plants, indicating that the JULGI-SMXL4/5 module is not involved in canonical strigolactone signaling (SI Appendix, Fig. S12). Moreover, rac-GR24 treatment did not alter the transcript levels of JUL1, SMXL4, or SMXL5 in Col-0, jul1 jul2, or smxl4 smxl5 plants (SI Appendix, Fig. S16). These results indicate that the JULGI-SMXL4/5 complex functions as a genetic module that maintains phloem development independently from strigolactone and karrikin signaling. In summary, it appears that ancestral angiosperm SMXL4/5 have lost their RGKT motifs and have adopted JULGI as a negative regulator, making them independent of strigolactone- and karrikin-mediated proteasomal degradation (Fig. 6D).

Discussion

Angiosperm exhibit remarkable plasticity in vascular and architectural development, distinguishing them from gymnosperms (3, 4). In this study, we demonstrated that the JULGI-SMXL4/5 module is an angiosperm-specific regulator of phloem development. During the evolution of angiosperms, neofunctionalization following gene duplication resulted in uncoupling of SMXL4/5 from strigolactone signaling: the loss of the RGKT motif followed by the introduction of a G-rich element into the 5′ UTR. JULGI, which specifically binds to G-rich single-stranded RNA, has been adopted as a negative regulator of SMXL4/5. These evolutionary processes highlight how minor modifications in regulatory motifs could be a potent driving force for evolution, as opposed to large-scale modifications in protein-coding sequences (4, 54–57). The possible evolutionary advantages for angiosperms conferred by securing phloem development in organs where strigolactone signaling prevails may lie in the regulation of axillary bud outgrowth. The JULGI-SMXL4/5 module contributes to the architectural plasticity of angiosperms by forming phloem within the dormant axillary bud, thereby promoting bud outgrowth when strigolactone signaling disappears.

Strigolactone signaling inhibits axillary bud outgrowth by degrading SMXL6/7/8 (31, 45, 58–60). Because SMXL4/5 in angiosperms can be expressed independently of strigolactone in dormant axillary buds, phloem can be formed in dormant axillary buds in the presence of strigolactone signaling. The presence of phloem-equipped dormant buds likely improves survival because phloem formation is essential for outgrowing organs. If strigolactone degrades both SMXL6/7/8 and SMXL4/5, axillary buds will become dormant and lack phloem, which is conceptually consistent with dormancy. Does this occur in gymnosperms with SMXL4/5 containing RGKT? Notably, the dormant axillary buds of some gymnosperms lack vascular connections with the main vascular cylinder (9, 61). Recent study of the gymnosperm G. biloba revealed that internally synthesized strigolactone regulates axillary bud outgrowth via SMAX1/SMXL2 genes, instead of SMXL6/7/8 genes absent in gymnosperm genome (62). As AtSMAX1 in Arabidopsis can be degraded in response to both karrikin and strigolactone, perhaps SMAX1/SMXL2 from gymnosperms can also be degraded in response to strigolactone (53, 63). This suggests that the strigolactone signaling pathway for apical dominance might have existed before the divergence of SMXL6/7/8, dating back to gymnosperms. Given that GbSMXL5a can be degraded following strigolactone/karrikin signaling in Arabidopsis, gymnosperms appear to lack a mechanism connecting the phloem to the axillary bud via SMXL4/5 when buds are dormant under the strigolactone signaling. However, angiosperms can maintain SMXL4/5 expression in dormant axillary buds to sustain phloem formation. This offers significant advantages, such as increased survival rates of the dormant buds and rapid growth upon the release of dormancy. Indeed, dormant axillary buds in angiosperms contain phloem tissues (64, 65).

Phloem tissues in angiosperms might contribute to architecture formation beyond transporting photoassimilates to axillary buds. When strigolactone inhibits axillary bud outgrowth in pea (Pisum sativum) and rice, the receptor D14 protein must be transported through the phloem (66). In order for this to occur, D14 must not degrade SMXL4/5, which is crucial for phloem development. Additionally, high sucrose levels reduce apical dominance and induce axillary bud outgrowth in angiosperms (67–69). Given that JULGI integrates carbon availability and sucrose-induced phloem development (40, 70, 71), it is reasonable to assume that the JULGI-SMXL4/5 module promotes axillary bud outgrowth by preforming phloem in the dormant bud. For example, branching is increased in jul1 jul2 (SI Appendix, Fig. S3E), but decreased in smxl4 smxl5 compared to Col-0 (72).

Therefore, the diversification of regulatory mechanisms for SMXL proteins has likely influenced the diversification of growth habits in angiosperms. Phloem-enabled axillary buds have contributed to the creation of a diverse morphological architecture in angiosperms triggered by developmental stimuli and environmental changes. This led to the speciation of over 200,000 angiosperm species with various growth habits, in contrast to gymnosperms, which are solely perennial arborescent plants (11). For instance, deciduous perennials must regenerate lateral organs annually, requiring the rapid and massive transport of carbon reserves to axillary buds. Annual herbaceous plants such as Arabidopsis must complete reproduction within a limited time frame, requiring rapid axillary bud initiation in case the shoot apex is lost. In contrast, many gymnosperms are perennial evergreens with infrequent bud outgrowth and strong apical dominance. However, the existence of deciduous gymnosperms, including G. biloba, suggests the presence of a more dynamic strigolactone-SMXL signaling pathway to regulate axillary bud initiation.

Materials and Methods

A complete overview of the materials and methods used in this research can be found in SI Appendix. This includes detailed descriptions of phylogenetic analyses and growth conditions for various model plants. Molecular genetic assays including protoplast transient expression assays, histological analyses, confocal microscopy, immunoblotting, RT-PCR experiments, and in situ hybridization are described. The generation of transgenic plants and mutant plants using CRISPR/Cas9 are also included.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.