KARRIKIN UP-REGULATED F-BOX 1 (KUF1) imposes negative feedback regulation of karrikin and KAI2 ligand metabolism in Arabidopsis thaliana.

SignificanceKarrikins are chemicals in smoke that stimulate regrowth of many plants after fire. However, karrikin responses are not limited to species from fire-prone environments and can affect growth after germination. Putatively, this is because karrikins mimic an unknown signal in plants, KAI2 ligand (KL). Karrikins likely require modification in plants to become bioactive. We identify a gene, KUF1, that appears to negatively regulate biosynthesis of KL and metabolism of a specific karrikin. KUF1 expression increases in response to karrikin or KL signaling, thus forming a negative feedback loop that limits further activation of the signaling pathway. This discovery will advance understanding of how karrikins are perceived and how smoke-activated germination evolved. It will also aid identification of the elusive KL.

Plants have evolved several adaptations that enable rapid regrowth in the postfire environment, including germination of buried seeds in response to chemical cues in smoke. Approximately 1,200 angiosperms are known to have positive germination responses to smoke or smoke-water treatments (1). A potent germination stimulant, 3-methyl-2H- furo[2,3-c]pyran-2-one (KAR1), was identified in smoke in 2004, followed by several similar butenolide compounds now known as karrikins (KARs) (2, 3). Unexpectedly, KAR responses are neither limited to germination nor to species from fire-prone environments. For example, KARs enhance Arabidopsis thaliana germination and seedling growth responses to light (4, 5). Smoke-water and/or KAR1 treatment have also been reported to enhance seedling vigor or accelerate development of several crop plants (6, 7). The capacity to respond to KARs may be widespread among angiosperms, with some fire-following species having evolved more sensitive germination responses to KARs (1, 8).

Two genes are required for KAR responses in A. thaliana. The first is MORE AXILLARY GROWTH2 (MAX2), which encodes an F-box protein that is also necessary for responses to the plant hormone strigolactone (SL) (9–11). As an F-box protein, MAX2 functions as a substrate-specifying component of SCF (Skp1, Cullin1, F-box) E3 ubiquitin ligase complexes, which mark proteins for degradation by attaching polyubiquitin chains. The second is KARRIKIN INSENSITIVE2 (KAI2)/HYPOSENSITIVE TO LIGHT (HTL), an ancient paralog of the SL receptor DWARF14 (D14)/DECREASED APICAL DOMINANCE2 (DAD2). KAI2 and D14 encode α/β-hydrolases with strictly conserved Ser-His-Asp catalytic triads that are important for their signaling activity (12). During SL signaling, a methylbutenolide “D-ring” moiety of SL is hydrolyzed by D14 and covalently attached to the catalytic His residue (13, 14). It is currently under debate which step in the process of SL binding, hydrolysis, and release constitutes activation of D14 (13–15). Regardless, SL induces a conformational change in D14 that promotes its association with SCFMAX2 and a subclade of proteins in the SMAX1-LIKE (SMXL) family that are orthologous to the rice protein DWARF53 (D53). D53-type SMXL proteins (i.e., SMXL6, SMXL7, and SMXL8 in Arabidopsis) are then polyubiquitinated and rapidly degraded by the 26S proteasome, activating the expression of downstream genes that carry out growth responses to SLs (13, 16–18). KAR signaling proceeds similarly. Activation of KAI2 promotes interactions with SCFMAX2 and the target proteins SMAX1 and SMXL2, leading to their polyubiquitination and proteolysis (19–23). In addition to mediating KAR signaling, Arabidopsis KAI2 responds to GR24t-5DS, a component of racemic GR24 (rac-GR24) whose D-ring is in a 2′S stereochemical configuration that is not found in natural SLs. Although rac-GR24 is commonly used as a synthetic SL analog, its constituent enantiomers GR245DS and GR24-5DS preferentially activate D14 and KAI2, respectively (24, 25).

KARs have been detected in smoke-water, postfire soil, and biochars, but there is currently no evidence that KARs are produced by living plants (3, 26, 27). Thus KARs may be strictly abiotic, or at least nonplant, signals. However, growing evidence suggests that KARs mimic an endogenous signal in plants that is recognized by KAI2 (28–30). For example, kai2 and max2 have several phenotypes that are not found in d14 or SL biosynthesis mutants (9) and are opposite to the effects of applied KAR. Although the putative KAI2 ligand (KL) remains unknown, insights into its functions can be gained by examining kai2 phenotypes and by using KARs or chemicals such as GR24-5DS or desmethyl-GR24 (dGR24) as substitutes (24, 31). Mutant kai2 seeds have increased dormancy, while kai2 seedlings have impaired photomorphogenesis that results in elongated hypocotyls and reduced cotyledon expansion (12). Recent studies have expanded the known roles of KAI2, and by proxy KAR/KL signaling, beyond germination and seedling growth. In rice, KAI2 is necessary for the formation of symbiotic interactions with arbuscular mycorrhizal fungi (32, 33). In Arabidopsis, KAI2 influences the shape of leaves, promotes several drought and abiotic stress resistance traits, increases root hair density, and promotes root hair elongation (34–37). In the liverwort Marchantia polymorpha, a KAI2-SCFMAX2-SMAX1 signaling pathway regulates thallus growth and orientation and cell proliferation in gemmae (38).

Because KAI2 is necessary for KAR responses and has been demonstrated to bind KAR1, it is often referred to in the literature as a KAR receptor. This may be misleading, however, as there is conflicting evidence that KARs can activate KAI2 directly. Several studies have demonstrated KAR1 binding by KAI2 through isothermal calorimetry, equilibrium microdialysis, intrinsic fluorescence, heteronuclear single quantum coherence, and dye–based thermal denaturation assays (39–45). These methods produced Kd estimates for KAR1 ranging from 5 µM to 147 µM for Arabidopsis KAI2, which indicates a lower affinity than expected considering KAR1 affects Arabidopsis growth at 1 µM and lower concentrations.

It must be considered that evidence of a molecule binding to a receptor is not necessarily evidence that the receptor is activated. Differential scanning fluorimetry (DSF) has emerged as a useful tool to identify ligands that activate D14. Upon exposure to a SL, D14 undergoes a shift in melting temperature that presumably reflects a conformational change or propensity for structural shifts (46). The abilities of different SL analogs to trigger a melting temperature shift in D14 correlates well with their bioactivities, although the extent of the temperature shift can vary in different D14 proteins (13, 15, 47, 48). Similarly, KAI2 undergoes destabilization in DSF assays in the presence of rac-GR24 and GR24-5DS, consistent with KAI2-mediated growth responses to these chemicals, and an intact catalytic triad is required to do so. However, KAI2 shows no response to KAR1 or KAR2 in DSF assays (8, 29). Similarly, yeast two-hybrid (Y2H) interactions between KAI2 and SMAX1 are promoted by rac-GR24 and GR24-5DS, but not by KAR1 or KAR2 (22). Therefore, in yeast and in DSF assays, KARs are not equivalent agonists to GR24.

Recent evidence suggests that this is also true in plants. Coimmunoprecipitation (Co-IP) of SMAX1 and SMXL2 by KAI2 is triggered by GR24-5DS, but KAR1 and 2′R-configured GR24 stereoisomers (e.g., GR245DS) are ineffective at stimulating KAI2-SMXL2 interactions. KAR1 is also much slower than GR24-5DS to induce polyubiquitination and degradation of SMXL2 (23). This raises the possibility that KAR1 requires metabolism in plants to become an active ligand for KAI2, but GR24-5DS does not.

There are two additional reasons to question KAR1 as a KAI2 ligand. First, KAR1 has been crystallized in complex with KAR-responsive KAI2 proteins from A. thaliana (AtKAI2) and the parasitic plant Striga hermonthica (ShKAI2iB) (39, 42). Although these proteins have highly similar structures, the KAR1 molecules have very different orientations in their ligand-binding pockets. Thus, it is unclear which structure, if either, correctly shows KAR1 binding to the receptor. Notably, neither KAI2 structure captures a substantial conformational change compared to the unbound apo-protein. Second, KAI2 is an extant representative of the evolutionary ancestor of D14. SL perception evolved again more recently in a clade of KAI2 paralogs that emerged in the parasitic Orobanchaceae (12, 49, 50). In both D14 and SL-responsive KAI2d proteins in parasites, the mechanism of SL perception involves hydrolysis of SL and covalent attachment of the cleaved butenolide ring to the His residue in the catalytic triad (13, 14, 47). The triad is essential for KAI2 activity, and KAI2 hydrolyzes GR24-5DS in vitro (25, 29). Although it has not yet been reported whether Arabidopsis KAI2 signals after forming a covalently linked intermediate molecule (CLIM) from GR24-5DS, as D14 and KAI2d proteins do with GR245DS, the most parsimonious explanation is that the hydrolysis-based signaling mechanism is conserved. If so, it is unclear how KARs may activate KAI2, as the butenolide ring in KARs cannot be cleaved through the same mechanism.

These observations led us and others to hypothesize that KAR1 must be modified in vivo before it can be recognized by KAI2 (29, 43). Here we show support for this hypothesis that emerged unexpectedly during a reverse genetic study of KARRIKIN-UP-REGULATED F-BOX1 (KUF1)/SKP1/ASK-interacting protein 25 (SKIP25). KUF1/SKIP25 (At1g31350) is a transcriptional marker of KAR/KL signaling that was first identified among 121 transcripts up-regulated by KAR1 treatment in a microarray analysis of primary dormant A. thaliana seed (5). KUF1 was later shown to be induced by KAR treatment in seedlings as well (5, 9, 12, 19, 20). Up-regulation of KUF1 by KAR1 in Arabidopsis requires KAI2 and MAX2. Furthermore, KUF1 transcripts are reduced in kai2 and max2 mutants, and increased in smax1 smxl2 (5, 9, 12, 20, 35, 51, 52), which have inactive and constitutive KAR/KL signaling pathway activities, respectively. KUF1 transcripts are also induced by rac-GR24. This response is partially mediated by D14 (12, 53). However, KUF1 expression is normal in the d14 mutant, suggesting that KUF1 is not regulated by endogenous SL signaling. Instead, the response to exogenous GR24 is likely due to D14 cross-talk with SMXL2 and/or SMAX1 (23). Although KUF1 expression is up-regulated in smax1 smxl2 seedlings, it is the same as wild type in smxl6 smxl7 smxl8. Furthermore, KUF1 expression in smax1 smxl2 is not further affected by treatments that activate D14 (23) (). Therefore, KUF1 up-regulation is a specific consequence of SMAX1 and SMXL2 degradation, which is typically caused by KAR/KL signaling through KAI2-SCFMAX2. Here we present evidence that this transcriptional response creates a negative feedback loop that putatively regulates the conversion of KAR1 into a bioactive signal and KL biosynthesis.

Results

Loss of KUF1 Causes Constitutive KAR/KL Responses in Seedlings.

KUF1 is predicted to have an N-terminal F-box domain, which mediates association with Skp1 in an SCF complex. Prior yeast two-hybrid screens found putative interactions between KUF1 and the Arabidopsis Skp1 proteins ASK1, ASK2, and ASK18 (54, 55). Consistent with these results, we observed F-box domain–dependent interactions between KUF1 and ASK1 in directed yeast two-hybrid assays (). This suggests that KUF1 is a functional F-box protein. UniProt also predicts that KUF1 has five C-terminal Kelch repeats, which are likely to mediate protein–protein interactions with downstream targets. However, a de novo structural prediction by AlphaFold indicates that KUF1 may have a six-bladed rather than five-bladed β-propeller structure () (56).

We hypothesized that KUF1 may have a role in activating downstream growth responses to KAR/KL. To investigate this idea, we used an egg cell–specific promoter-controlled CRISPR-Cas9 system (57) to introduce loss-of-function mutations in KUF1. Two sites in KUF1 were targeted simultaneously. We identified a 200-bp deletion between these sites. This allele, kuf1-1, is likely to be null, as it causes a translational frameshift that retains only the first 36 amino acids (aa) of the 395-aa protein (Fig. 1). Contrary to our expectations, kuf1-1 seedlings grown under continuous red light had reduced hypocotyl elongation and enlarged cotyledons, similar to smax1 and KAR-treated wild-type (Col-0) seedlings (Fig. 1). The kuf1-1 (hereafter, kuf1) hypocotyl phenotype is recessive and was rescued by introduction of a wild-type KUF1p:KUF1 transgene (Fig. 1 and ).

Fig. 1.

Isolation of a kuf1 mutant in A. thaliana. (A) Diagram of KUF1 protein domains and KUF1 gene structure. Filled arrow, exon; red triangles, gRNA target sites; dashed lines, boundaries of genomic deletion in kuf1-1. (B) Seedlings grown 4 d in red light on 0.5× MS medium supplemented with 0.1% (vol/vol) acetone, 1 μM KAR1, or 1 μM KAR2. KUF1 kuf1-1 is a KUF1p:KUF1 kuf1-1 transgenic rescue line. (C) Expression of DLK2, BBX20/STH7/BZS1, IAA6, and SMXL2 relative to CACS reference transcripts in seedlings grown 4 d in red light on 0.5× MS medium supplemented with 0.1% (vol/vol) acetone (gray) or 1 μM KAR1 (orange), measured by qRT-PCR. Relative expression values are scaled to Col-0. (n = 5 pools of seedlings; mean ± SD). Letters indicate statistical groups, P < 0.05, Dunnett’s T3 multiple comparisons test.

To determine whether kuf1 and smax1 shared more than morphological similarities, we investigated the expression of transcriptional markers of KAR/KL response in kuf1 seedlings grown under red light. D14-LIKE2 (DLK2), B-BOX DOMAIN PROTEIN 20/SALT TOLERANCE HOMOLOG 7/bzr1–1D SUPPRESSOR (BBX20/STH7/BZS1), and SMAX1-LIKE 2 (SMXL2) are positively regulated by KAR/KL signaling, whereas INDOLE-3-ACETIC ACID 6 (IAA6) is negatively regulated by KAR/KL signaling (5, 12, 19, 23, 29). All four genes were differentially expressed in kuf1 relative to wild type to a similar or stronger extent than we observed in smax1 and KAR1-treated wild-type seedlings (Fig. 1). KAR1 treatment of kuf1 seedlings further enhanced the differential expression of these genes (Fig. 1). Therefore, kuf1 has constitutive KAR/KL responses, similar to smax1, but remains responsive to KAR1 treatment. Collectively, the seedling growth and gene expression phenotypes of kuf1 indicated that KUF1 inhibits the activity or output of the KAR/KL signaling pathway. This suggested that KUF1 up-regulation after KAI2-SCFMAX2–mediated degradation of SMAX1 and SMXL2 may be part of a negative feedback mechanism.

KUF1 Acts Upstream of KAI2 and MAX2.

We initially considered three hypotheses to explain kuf1 phenotypes: 1) KUF1 may control seedling growth independently of the KAR/KL signaling pathway, 2) kuf1 may reduce the abundance or activity of SMAX1/SMXL2 proteins independently of KAI2-SCFMAX2, or 3) kuf1 may enhance SMAX1/SMXL2 degradation by increasing the abundance or activity of KAI2 or MAX2 proteins. The first two hypotheses predict that kuf1 will at least partially suppress the seedling phenotypes of kai2 and max2, which over accumulate SMAX1 and SMXL2. To test this, we generated double mutants between kuf1 and max2, kai2, and d14. We found that kai2 and max2, which have elongated hypocotyls and small cotyledons, are epistatic to kuf1 (Fig. 2). In contrast, kuf1 d14 mutant seedlings were similar to kuf1. This indicated that a functional KAI2-SCFMAX2 signaling mechanism, but not a SL signaling mechanism, is necessary for expression of the kuf1 phenotype. These results led us to reject the first two hypotheses.

Fig. 2.

Epistasis analysis of kuf1 and KAR/SL signaling mutants. (A) Hypocotyl lengths (n = 80) and (B) individual cotyledon areas (n = 73 to 78) of seedlings grown 4 d in red light on 0.5× MS medium. Box plots have Tukey whiskers. *P < 0.01, Dunnett’s T3 multiple comparisons test; ns, not significant.

kuf1 Seedlings Are Hypersensitive to KAR1 but Not Other KAI2-Mediated Signals.

To investigate the third hypothesis, we first compared the abundance of KAI2 protein in wild-type and kuf1 seedlings grown under red light. We did not observe a significant difference in KAI2 protein abundance in kuf1 (Fig. 3). We also used a ratiometric, dual-fluorescent reporter system to examine how KUF1 affects KAI2 abundance (22, 58). We transiently coexpressed 35S:KUF1 or a 35S:kuf1Δ mutant with an AtKAI2 ratiometric reporter in Nicotiana benthamiana leaves (Fig. 3). After 3 d, we measured fluorescence from the KAI2-mScarlet-I fusion protein relative to fluorescence from a simultaneously expressed reference protein, Venus. We found that coexpression of wild-type KUF1 had no effect on the relative abundance of KAI2-mScarlet-I, consistent with our observations in Arabidopsis seedlings (Fig. 3). Therefore, KAI2 is unlikely to be a target of SCFKUF1, which agrees with the prior finding that KAI2 is not polyubiquitinated or degraded after KAR2 treatment by the 26S proteasome (59).

Fig. 3.

KAI2 protein abundance under KUF1 perturbation. (A) Western blot analysis of KAI2 protein abundance in 60 μg total protein extracted from wild-type and kuf1 Arabidopsis seedlings grown 4 d in red light. Actin and Rubisco large subunit (RbcL) were used as loading controls and imaged by Western blot or 2,2,2-trichloroethanol staining, respectively. Relative KAI2 abundance was scaled to Col-0. (n = 3 pools of seedlings; mean ± SD) (B) Ratio of KAI2-mScarlet-I to Venus fluorescence in N. benthamiana leaves coinfiltrated 3 d with the pRATIO3212-KAI2 ratiometric reporter plasmid and the 35S-driven overexpression plasmids pGWB402-kuf1Δ or pGWB402-KUF1. *F2A causes separation of KAI2-mScarlet-I and Venus proteins during translation. The kuf1Δ mutation has a deletion that causes premature truncation after the N-terminal 46 aa of KUF1. (n = 4 leaves, each leaf value is the mean relative fluorescence of six leaf discs; mean ± SD). ns (not significant), P > 0.05, two-tailed unpaired t test.

Nonetheless, the possibilities remained that MAX2 is more abundant in kuf1, or that kuf1 somehow enhances the activity of KAI2 and/or SCFMAX2, putatively resulting in more effective degradation of SMAX1/SMXL2. A reasonable prediction of either case is that kuf1 seedlings would be hypersensitive to all chemical signals that activate KAI2. Therefore, we examined the seedling growth responses of kuf1 and KUF1p:KUF1 kuf1 rescue lines to KAR1, KAR2, and rac-GR24. As demonstrated by dose–response experiments, the degree to which these treatments inhibit hypocotyl elongation offers a simple and sensitive readout of MAX2-dependent signaling activity (5, 60). We found that KAR responses were altered in kuf1 seedlings in unexpected ways. Prior studies have consistently shown that Arabidopsis is more sensitive to KAR2 than KAR1 (4, 5, 8, 59). In kuf1 seedlings, however, 1 µM KAR1 caused a stronger reduction in hypocotyl elongation than 1 µM KAR2 (Fig. 4). This was mostly due to an enhanced response to KAR1, as the percentage of growth inhibition by 1 µM KAR1 on kuf1 hypocotyls was nearly double that of wild type. However, growth inhibition by 1 µM KAR2 was also weaker in kuf1 than wild type. By contrast, the growth response to 1 µM rac-GR24 was similar in kuf1 and wild-type hypocotyls (Fig. 4 and ). The unusual KAR responses of kuf1 hypocotyls were rescued by a wild-type KUF1 transgene (Fig. 4 and ).

Fig. 4.

Hypocotyl elongation response of kuf1 to KARs and rac-GR24. (A) Representative images and (B) hypocotyl lengths of seedlings grown 4 d in red light on 0.5× MS medium supplemented with 0.1% (vol/vol) acetone, 1 μM KAR1, 1 μM KAR2, or 1 μM rac-GR24. Box plots have Tukey whiskers. Percent growth inhibition relative to the mock-treated control for each genotype is indicated below box plots of experimental treatments. (n = 100). *P < 0.01, Dunnett’s multiple comparisons test, compared to Col-0 with the same treatment; ns, not significant. #P < 0.01, Dunnett’s multiple comparisons test, KAR1 vs. KAR2 within the genotype.

We also compared the ability of KAR1, KAR2, and rac-GR24 to stimulate degradation of a bioluminescent reporter of SMAX1 abundance, SMAX1D2-LUC, in wild-type and kuf1 backgrounds. This reporter is a translational fusion of the C-terminal D2 domain of SMAX1, which is sufficient for SCFMAX2-induced degradation, to firefly luciferase (22). Supporting the hypocotyl experiment results, KAR1 caused a stronger decline in SMAX1D2-LUC abundance in kuf1 seedlings than in wild type during the first 12 h after treatment. By contrast, there was no statistically significant difference (P > 0.01) in the response of kuf1 and wild-type seedlings to either KAR2 or rac-GR24 ().

We found that the hypersensitive response to KAR1 in kuf1 seedlings was influenced by the intensity of red light. At 3.8 to 15 µmol m−2 s−1 red light, the percentage of growth inhibition of kuf1 hypocotyls by 1 µM KAR1 was approximately double that observed in wild-type seedlings (). At higher intensities of red light, the response of wild-type seedlings to KAR1 increased; under 60 µmol m−2 s−1 red light the relative responses of wild type and kuf1 to KAR1 were the same. In all red light intensities tested, kuf1 hypocotyls were shorter than wild type (). In the dark, however, kuf1 mutants grew to the same length as wild type and did not show hypocotyl growth responses to KAR1 or KAR2 (). We investigated whether changes in KUF1 expression might explain the varied effects of KAR1 on hypocotyl elongation observed under different light fluences. While we found some evidence for higher KUF1 expression in wild-type seedlings grown in the dark than in red light (P = 0.029 to 0.057), there was no clear difference in KUF1 expression across low to high red light fluences (). Therefore, other genetic factors likely contribute to KAR1 responsiveness.

Because KAR2 and rac-GR24 are normally more potent stimulants of KAI2-dependent signaling in Arabidopsis seedlings than KAR1, we reasoned that responses to these signals may appear compressed in kuf1 if a maximum growth inhibition were approached. That is, kuf1 might also be hypersensitive to KAR2 and rac-GR24, but an enhanced response cannot be detected if KAI2 activation is already saturated. Therefore, we tested kuf1 responses to lower concentrations of KAR2 and rac-GR24 (Fig. 5). At all concentrations, KAR1 caused stronger growth inhibition of kuf1 hypocotyls than wild type. In contrast, kuf1 responses to KAR2 were similar to those of wild type at 100-nM and 300-nM concentrations and less than wild type at 1 µM (Fig. 5). At all concentrations tested, kuf1 showed growth responses to rac-GR24 that were similar to wild type (Fig. 5). Notably, we observed in wild-type seedlings that 100 nM rac-GR24 caused an equivalent degree of hypocotyl growth inhibition as 1 µM KAR1 (23% vs. 24%, respectively; Fig. 5 ). The kuf1 seedlings showed a wild-type response (25% growth inhibition) to this low concentration of rac-GR24, despite having an enhanced response to 1 µM KAR1 (47% growth inhibition). Therefore, kuf1 is hypersensitive to KAR1 but not KAR2 or rac-GR24.

Fig. 5.

Response of kuf1 to low concentrations of KARs and rac-GR24. Hypocotyl lengths of wild-type and kuf1 seedlings grown 4 d in red light on 0.5× MS medium supplemented with 0.1% (vol/vol) acetone, 100 nM, 300 nM, or 1 μM (A) KAR1 or KAR2 (n = 60), or (B) rac-GR24 (n = 80). For (A) *P < 0.01, Tukey’s multiple comparisons test; ns, not significant. (C) Hypocotyl lengths of seedlings grown 4 d in red light on 0.5× MS medium supplemented with 0.1% (vol/vol) acetone or 1 μM rac-GR24. (n = 80). Box plots have Tukey whiskers. Percent growth inhibition relative to the mock-treated control for each genotype is indicated below box plots of experimental treatments.

We considered the possibility that the similar responses of kuf1 and wild-type hypocotyls to rac-GR24 could reflect the ability of rac-GR24 to activate both KAI2 and D14, whereas KAR1 only signals through KAI2. This led us to test the rac-GR24 responses of kuf1 seedlings in the absence of signaling contributions from D14. As previously demonstrated, d14 was less responsive to rac-GR24 than wild type (12) (Fig. 5). However, we found that hypocotyl elongation of d14 and kuf1 d14 seedlings was inhibited to a similar degree by rac-GR24 (39% vs. 36%, Fig. 5). Therefore, the KAI2-SCFMAX2 signaling pathway in kuf1 responds normally to rac-GR24.

A Fourth Hypothesis for KUF1 Function.

A previous study showed that ∼25-fold overexpression of KAI2 causes Arabidopsis seedlings to become hypersensitive to both KAR2 and rac-GR24, but does not impact hypocotyl length in the absence of such treatments (60). In contrast, we observed that 1) kuf1 is specifically hypersensitive to KAR1 but not KAR2 or rac-GR24, and 2) kuf1 affects seedling growth in the absence of treatments in a KAI2-SCFMAX2–dependent manner. This implied that kuf1 does not increase the overall sensitivity or responsiveness of the KAI2 signaling pathway as proposed by the third hypothesis above, but instead modifies the ability of specific signals to be perceived by KAI2. Therefore, we hypothesized that KUF1 negatively regulates metabolism of KAR1 into a bioactive signal. In addition, the phenotypes of kuf1 seedlings are similar to those of smax1 or KAR-treated wild-type seedlings, suggesting that the KAR/KL signaling pathway is more active than normal (Fig. 1). This could be explained if KUF1 also negatively regulates KL biosynthesis or availability. Such a model would be consistent with the observation that kuf1 phenotypes are dependent on MAX2 and KAI2 (Fig. 2).

KL Abundance Is Likely Increased in kuf1.

It was not possible for us to directly test whether KL biosynthesis and KAR1 metabolism are enhanced in kuf1 because KL and KAR1 metabolites have not yet been identified. Instead, we used three strategies to indirectly investigate KUF1 effects on KL levels and KAR1 metabolism. First, we took advantage of the diversified ligand preferences found among KAI2 paralogs in root parasitic plants in the Orobanchaceae. A gene from witchweed (S. hermonthica), ShKAI2i, encodes a subfunctionalized receptor that responds well to KAR1 and appears to be less sensitive to KL (28, 49). Introducing ShKAI2i into an Arabidopsis kai2 mutant produces seedlings that retain similar morphology to kai2. However, the ShKAI2i kai2 transgenic line is fully responsive to KAR1 treatment, and the degree of hypocotyl elongation inhibition by KAR1 is in fact stronger than wild type, indicating that ShKAI2i is functional in Arabidopsis (28). At a later stage of development, ShKAI2i partially rescues the altered rosette morphology of kai2 mutants grown in short day conditions. Putatively, ShKAI2i does not rescue kai2 seedlings and only partially rescues kai2 leaf development because it has reduced sensitivity to KL (28). We reasoned that the ShKAI2i kai2 line may provide a way to test whether KL levels are increased in kuf1 seedlings. According to our model, loss of KUF1 should not reduce the hypocotyl elongation of ShKAI2i kai2 as strongly as it does wild type because ShKAI2i would be less sensitive to increased KL abundance than the native AtKAI2 protein. Responses to KAR1, however, should be increased by kuf1 at least as strongly when ShKAI2i is the available KAI2 receptor as when AtKAI2 is.

To test this, we used CRISPR-Cas9 to recreate the identical 200-bp kuf1-1 deletion in ShKAI2i kai2 and its corresponding wild-type ecotype, Ler. Similar to what we previously observed in the Col-0 ecotype, the kuf1 mutation enhanced the response to KAR1 treatment in both Ler and ShKAI2i kai2 backgrounds (Fig. 6). However, under mock-treated conditions—when only endogenous KL is available to seedlings—hypocotyl elongation was reduced by 41% in kuf1 compared to Ler and by only 25% in kuf1 ShKAI2i kai2 compared to ShKAI2i kai2 (Fig. 6). These results are consistent with the idea that KL is increased in kuf1 mutants, but has a smaller effect on the growth of ShKAI2i kai2 seedlings than wild type because ShKAI2i protein is less able to perceive it than AtKAI2.

Fig. 6.

Indirect tests of KUF1 effects on KAR and KL metabolism. (A) Hypocotyl lengths of seedlings grown 4 d in red light on 0.5× MS medium supplemented with 0.1% (vol/vol) acetone or 1 μM KAR1. The kuf1 allele is identical to kuf1-1 but it was made in the Ler ecotype and the CRISPR-Cas9 T-DNA has not been removed. ShKAI2i kai2 expresses KAI2i from S. hermonthica under control of an Arabidopsis KAI2 promoter in the kai2-2 mutant background. Percent growth inhibition relative to the mock-treated control for each genotype is indicated below box plots of KAR1 treatments. Percent growth inhibition due to kuf1 is indicated above box plots. (n = 80 to 100) (B) Arbitrary relative luminescence units of 9-d-old SMAX1D2-LUC Col-0 and kuf1 seedlings grown in 16 h white light:8 h dark, normalized to each seedling’s leaf and cotyledon surface area. Wild-type and kuf1 leaf/cotyledon surface areas were not different (P > 0.05). (n = 36 seedlings) *P < 0.01, Student’s two-tailed t test. (C) Ratio of SMAX1-mScarlet to Venus fluorescence in N. benthamiana leaves coinfiltrated 3 d with the pRATIO1212-SMAX1 ratiometric reporter plasmid and the overexpression plasmids pGWB402 (empty), pGWB402-kuf1Δ, or pGWB402-KUF1, after a 16-h treatment with 0.02% (vol/vol) acetone or 10 μM KAR1. The kuf1Δ mutation has a deletion that causes premature truncation after the N-terminal 46 aa of KUF1. (n = 14 to 18). Letters indicate statistical groups, P < 0.01, Tukey’s multiple comparisons test. Box plots have Tukey whiskers.

Second, we compared the abundance of the SMAX1D2-LUC reporter in untreated wild-type and kuf1 seedlings. We observed lower luminescence in kuf1 seedlings than in wild type, suggesting reduced abundance of SMAX1D2-LUC and presumably SMAX1, which would be consistent with higher KL levels (Fig. 6).

Third, we used the ratiometric reporter system to examine how KUF1 affects SMAX1 abundance and its degradation after KAR1 treatment (22, 58). We transiently coexpressed 35S:KUF1 with a SMAX1 ratiometric reporter in N. benthamiana leaves for 3 d, treated excised leaf discs with KAR1, and measured the ratio of SMAX1-mScarlet-I to Venus fluorescence after 16 h. We noted across many experiments that the abundance of the SMAX1 reporter was consistently higher in mock-treated leaves coexpressing wild-type KUF1 compared to those coexpressing an empty vector or a kuf1Δ frameshift allele (Fig. 6 and ). This observation is consistent with the hypothesis that KUF1 overexpression blocks production of KL, thus promoting stabilization of SMAX1.

We also found that coexpression of wild-type KUF1 prevented KAR1-induced degradation of the SMAX1 reporter (Fig. 6). This led us to test whether KUF1 overexpression impacts the ability of other treatments to trigger SMAX1 degradation. In wild-type N. benthamiana plants, KUF1 overexpression blocked degradation of the SMAX1 reporter by KAR2 as well as KAR1, but rac-GR24 was still active (). To determine whether the remaining rac-GR24 response was mediated by D14 or KAI2, we also performed the experiment in an Nbd14a,b double mutant that has loss-of-function mutations in the two D14 homoeologs in N. benthamiana (61). We found that the SMAX1 reporter was not degraded in response to KAR1, KAR2, or rac-GR24 in Nbd14a,b leaves when KUF1 was overexpressed (). This suggests that KUF1 overexpression inhibits all KAI2-mediated degradation of SMAX1 in N. benthamiana, which was not expected from the characterization of Arabidopsis kuf1. This effect is not due to the loss of KAI2 protein (Fig. 3) and is at the moment difficult to explain given that GR24 should not require metabolism to activate KAI2.

Are KUF1 Expression Patterns a Cause or an Effect of KL Distribution?

Because KUF1 expression is an output of KAR/KL signaling, its expression pattern in plants may reveal where KL is most abundant, KAI2 is most active, and/or SMAX1 and SMXL2 abundance is lowest. An alternative, nonmutually exclusive possibility is that KUF1 expression patterns are imposed in certain tissues or by other cellular pathways as a way to restrict KL abundance and KAI2 activity. We generated transcriptional reporters for KUF1 by using the −2,967-bp region upstream of KUF1 to drive expression of the β-glucuronidase (GUS) reporter. This upstream region provides appropriate cis-regulation for KUF1 to rescue the kuf1 mutant (Fig. 4 and ).

We found that in seedlings, the KUF1 promoter region drives GUS expression in the root cap, newly emerged leaves, and mature vasculature in the root, but not in the differentiation or elongation zones of the root. GUS activity was also generally lacking in the hypocotyl (Fig. 7). In some strongly expressing lines, we observed GUS activity in the stomata of cotyledons. In 5-wk-old adult plants, the KUF1 promoter was active in leaves, rosette axillary buds, sepals, and anthers (). We investigated whether KUF1 expression is limited to specific tissues in seedlings because of ligand (i.e., KL) availability or because only some tissues are competent to respond to KAI2 agonists. These possibilities may be distinguished, respectively, by whether KAR and rac-GR24 treatments increase the spatial distribution or only the intensity of KUF1 expression. We observed that KAR2 and rac-GR24 tended to intensify, rather than expand, the regions with GUS activity in 5-d-old seedlings grown in red light. Therefore, the ability to express KUF1, or KAI2-SCFMAX2 signaling itself, may be limited to specific tissues. Interestingly, the expression pattern of KUF1 corresponds well with that of SMAX1 (34).

Fig. 7.

KUF1 expression patterns in Arabidopsis seedlings. Representative micrographs of β-glucuronidase staining (blue) in (A) cotyledons, (B) shoot apices, (C) mature root cells, (D) lateral root primordia, and (E) primary root tips of KUF1p:GUS seedlings grown 4 d in red light on 0.5× MS supplemented with 0.1% (vol/vol) acetone, 1 μM KAR2, or 1 μM rac-GR24.

KUF1 Negatively Regulates KAI2-Mediated Root Hair Development.

The expression of KUF1 in roots led us to consider whether it has a role in root development. Because Arabidopsis root hair density and elongation are positively regulated by the KAR/KL pathway, we investigated the effects of kuf1 on root hair growth (37). As previously demonstrated, we observed that KAR1 and KAR2 increased root hair density and elongation in a KAI2- and MAX2-dependent manner (Fig. 8). Moreover, kai2 and max2 mutants showed reduced root hair density and elongation under mock treatment. We found that kuf1 seedlings have increased root hair density and root hair elongation compared to wild type under mock treatment, which is consistent with a constitutive KAR/KL response (Fig. 8). This phenotype was rescued by KUF1p:KUF1. In addition, KAR1 enhanced root hair density more than KAR2 in kuf1 but not wild-type roots, similarly to our observations of KAR1 hypersensitivity in kuf1 hypocotyls (Fig. 8). Therefore, KUF1 imposes negative feedback regulation on multiple developmental processes regulated by KAI2-SCFMAX2 in Arabidopsis.

Fig. 8.

Root hair density and elongation in kuf1. (A) Root hair density (n = 10) and (B) root hair length (n = 10 roots, represented by average of 10 root hairs from each) of 5-d-old seedlings grown vertically in white light (120 μmol m−2 s−1) on 0.5× MS medium supplemented with 1% (wt/vol) sucrose and 0.07% (vol/vol) methanol, 1 µM KAR1, or 1 µM KAR2. The kai2-2 allele was introgressed into the Col-0 background. Box plots have Tukey whiskers. *P < 0.01, Dunnett’s multiple comparisons test, compared to Col-0 with the same treatment. #P < 0.01, Dunnett’s multiple comparisons test, compared to mock treatment within the genotype. ns, not significant.

Discussion

KUF1 is one of ∼103 F-box proteins with C-terminal Kelch repeats in A. thaliana. It is the sole member of one of eight “superstable” clades of F-box Kelch genes that have been defined based on having orthologs in eudicots A. thaliana, Populus trichocarpa, and Vitis vinifera; the monocots Oryza sativa and Sorghum bicolor; the lycophyte Selaginella moellendorffii; and the bryophyte Physcomitrium patens (62). These superstable F-box genes are under purifying selection and therefore have been proposed to carry out developmental or physiological functions that are conserved in land plants (62). Here we identified roles for KUF1 in Arabidopsis seedling and root hair development that likely arise from its molecular function as a negative regulator of responses to KAR1 and KL. This raises the possibility that KUF1 may affect other aspects of plant growth and development regulated by KAR/KL signaling such as seed germination, drought tolerance, and symbiotic interactions with arbuscular mycorrhizal fungi.

It is possible that up-regulation of KUF1 after activation of SCFMAX2-regulated signaling pathways is a widely conserved response in land plants. In P. patens, at least one of the four KUF1 orthologs (Pp3c2_34130v3.1) shows induced expression following rac-GR24 treatment and is down-regulated in a Ppmax2 mutant. Induction of KUF1 by rac-GR24 is dependent on PpMAX2, at least when the moss is grown in the light (63). In O. sativa (rice), OsKUF1 (LOC_Os06g49750) is up-regulated severalfold in smax1 and is also induced by KAR1 and GR24-5DS in a KAI2/D14L-dependent manner (64). In Pisum sativum (pea) buds, PsKUF1 (Psat2g143160) is reported in a preprint to be up-regulated by rac-GR24 in the SL-deficient rms1/ccd8 and rms5/ccd7 backgrounds, by GR245DS in rms1/ccd8, and, unexpectedly, by KAR1 in rms4/max2 only (65). In Lactuca sativa (lettuce), we have found that at least one of its three KUF1 homologs shows increased expression in achenes imbibed for 6 h with 1 µM KAR1 or KAR2 relative to mock treatment ().

The kuf1 loss-of-function mutant shows several developmental and transcriptional phenotypes that are consistent with hyperactive KAR/KL signaling. We considered several hypotheses for how KUF1 may reduce the effects of the KAR/KL pathway on plant growth and development. Although KUF1 expression is negatively regulated by SMAX1 and SMXL2, epistasis analysis demonstrated that KUF1 acts upstream of KAI2 and MAX2. This suggests a negative feedback loop. We also observed that kuf1 seedlings are hypersensitive to KAR1 but not rac-GR24. Therefore, KUF1 affects the ability of different molecules to be perceived by KAI2. This leads us to hypothesize that KUF1 negatively regulates metabolism of KAR1 into a bioactive ligand of KAI2 (Fig. 9). Because kuf1 phenotypes under control conditions are similar to smax1 and KAR-treated wild-type seedlings, we further propose that KUF1 negatively regulates biosynthesis of the unknown endogenous KAI2 ligand, KL. One way this could occur is through SCFKUF1-mediated polyubiquitylation and proteolysis of an enzyme(s) that metabolizes KAR1 and/or KL. Indeed, if there were an enzyme that carried out both KAR1 metabolism and KL biosynthesis, it could help explain the potential for KAR response in nonfire following angiosperms such as Arabidopsis, for which the selective pressure to maintain a KAR response mechanism is not obvious. Alternatively, negative regulation of a KAR1/KL-metabolizing enzyme(s) by KUF1 could be less direct, such as by modulating the abundance of a transcription factor(s) that controls expression of the enzyme(s) or production of an enzyme cofactor. In this model, up-regulation of KUF1 expression in response to SCFMAX2-mediated degradation of SMAX1/SMXL2 represents a negative feedback mechanism for KL homeostasis that occurs through proteolysis rather than transcriptional repression (Fig. 9).

Fig. 9.

Model of a KUF1-mediated negative feedback loop. Activation of KAI2 by an unknown endogenous ligand (KL), a putative KAR1 metabolite (*KAR1), or GR24-5DS triggers its association with SCFMAX2 and SMAX1/SMXL2. SMAX1/SMXL2 proteins are polyubiquitinated and degraded by the 26S proteasome, causing an increase in KUF1 transcripts. KUF1 likely acts within an SCF-type E3 ubiquitin ligase complex to target (an) unknown protein(s) for polyubiquitination and proteasomal degradation. The KUF1 target(s) positively regulate(s) KL biosynthesis and conversion of KAR1 into a bioactive signal. Thus, up-regulation of KUF1 following SMAX1/SMXL2 degradation dampens the plant’s capacity to produce additional ligands for KAI2 but does not affect its ability to respond to GR24-5DS. For simplicity, components of the E3 ubiquitin ligase complex other than Arabidopsis Skp1 (ASK1) are not shown.

Our results raise the question of why rac-GR24 responses remain normal in kuf1. We propose that rac-GR24, or more specifically the GR24-5DS stereoisomer, is a “ready-to-go” ligand that is able to be recognized directly by KAI2 without metabolism. This is supported by the observation of thermal stability shifts in KAI2 after GR24-5DS treatment in DSF assays and by yeast two-hybrid interactions between KAI2 and SMAX1 in the presence of rac-GR24 and GR24-5DS (8, 22, 29). In addition, GR24 has a hydrolyzable butenolide D-ring that KARs lack, which has been shown to be important for signaling activity in many D14 agonists. KAR2 responses in kuf1, which are not enhanced and even seem to be decreased under 1-µM treatments, are difficult to explain at this time. Presumably KAR2 also requires metabolism by plants, as it is inactive on KAI2 in DSF and yeast two-hybrid assays, but this might be carried out by a protein or mechanism not regulated by KUF1. Different responses to KAR1 and KAR2 also occur in Lotus japonicus, where both compounds affect hypocotyls but only KAR1 affects root growth (66). Because the receptor that mediates KAR2 responses, LjKAI2a, is active in both hypocotyls and roots, tissue-specific metabolism of KAR2 that is independent of KAR1 metabolism might explain this phenomenon.

A notable weakness of the proposed model (Fig. 9) is the inability to directly evaluate KL abundance. Until KL is identified, alternative explanations of these data that we have not considered remain possible. One observation that is currently difficult to reconcile with the model is the effect of KUF1 overexpression on SMAX1 degradation in N. benthamiana, which blocks responses to KAR1, KAR2, and rac-GR24 rather than just KAR1. One possibility is that a metabolic precursor of KL overaccumulates when KUF1 is overexpressed that competitively inhibits activation of KAI2 by other molecules.

It is notable that although kuf1 has a shortened hypocotyl phenotype, KUF1 expression was either absent or very low in Arabidopsis hypocotyls compared to other seedling tissues. This suggests that KUF1 effects on hypocotyl elongation are not locally mediated. Similarly, another transcriptional marker of KAR/KL signaling, B-box domain protein BBX20/STH7, regulates hypocotyl elongation but has very low expression in seedling hypocotyls compared to other tissues (52). There are a number of possibilities for how KUF1 could act at a distance to affect hypocotyl elongation, including direct or indirect control of a mobile signal or protein. For example, KL may be transported from its site(s) of synthesis to act in other tissues. Auxin is another attractive candidate for a mobile signal. KAI2 remodels auxin distribution in seedlings during photomorphogenesis, putatively by controlling the abundance of the auxin efflux carriers PIN1, PIN3, and PIN7 at the plasma membrane. This in turn influences hypocotyl elongation and root development (67).

An outstanding goal for the SL and KAR field is the identification of KL, which will give insights into the function of KAI2 as well as the evolutionary history of the SL and KAR signaling pathways. This characterization of KUF1 reveals strategies to achieve this goal. First, kuf1 tissues may provide a more abundant source of KL than wild type for bioassay-guided fractionation approaches to isolate purified KL. Second, identification of the protein(s) targeted for degradation by SCFKUF1 should provide insights into how KL and KAR1 metabolism occur and may even uncover enzymes that are directly involved in KL biosynthesis. Biochemical approaches to identify proteins that interact with KUF1 and/or genetic approaches to identify kuf1 suppressors may prove fruitful for discovering these targets.

Materials and Methods

Materials.

A. thaliana alleles in the Col-0 ecotype used here were d14-1, htl-3 (“kai2” here for clarity), d14-1 htl-3, smax1-2, max2-1, max2-2, and smax1-2 smxl2-1 (12, 19, 20, 41, 68). The kai2-2 allele (Ler ecotype) and the ShKAI2i kai2-2 transgenic line are also described previously (12, 49). SMAX1 was derived through crossing kuf1-1 into the previously described UBQ10:SMAX1 transgenic Arabidopsis line (22). Fig. 8 uses a kai2-2 allele backcrossed six times into the Col-0 background. KAR1, KAR2, and rac-GR24 were synthesized and provided by Gavin Flematti and Adrian Scaffidi, University of Western Australia, Perth, Australia. KAR and rac-GR24 stocks were prepared in acetone, stored at −20 °C, and freshly diluted before use. Oligonucleotide primer sequences for cloning, genotyping, and expression analysis are described in . Additional detailed methods are described in .

Isolation and Rescue of kuf1.

Based on cDNA and expressed sequence tag support, the gene model At1g31350.1 (395 aa) was used for KUF1 instead of At1g31350.2 (431 aa). Two KUF1 sites, 5′-GAAAACTCCGCTTTAATCGAAGG-3′ and 5′-CGGTTTATGTGTTTCAACCCGG-3′ (protospacer adjacent motif underlined), with a minimum mismatch of five nucleotides (nt) to the next closest genomic sequence were selected from the CRISPR-PLANT gRNA database for gene editing by CRISPR-Cas9 (69). Both gRNA sequences were introduced into pHEE401E through GoldenGate cloning as previously described, but with Q5 DNA polymerase (New England Biolabs) during cassette amplification steps (57, 70). Arabidopsis was transformed by floral dip with Sanger sequence-verified pHEE401E-KUF1. Seed was screened on 0.5× Murashige-Skoog (MS) media with hygromycin B (20 µg/mL) to identify T1 plants. The kuf1-1 allele (200-bp deletion between +107 and +307 of the coding sequence) was identified by PCR () and verified by Sanger sequencing. The kuf1-1 allele generated in Col-0 and Ler backgrounds was identical at the nucleotide level. The pHEE401E-KUF1 T-DNA was removed from the kuf1-1 allele in Col-0 ecotype through segregation, but was not removed in the Ler ecotype kuf1 lines. To rescue kuf1, a genomic KUF1 sequence (from −2,967 upstream of the translational start codon to the stop codon) was cloned into pDONR221A, a Gateway entry vector modified to have ampicillin resistance), sequence-verified, and transferred into pGWB501 (71) with LR clonase II (Invitrogen). The resulting pGWB501-KUF1p:KUF1 was introduced into kuf1-1 by floral dip. Homozygous, single-insertion transgenic lines were characterized.

Seedling Growth Assays.

Surface-sterilized seeds were plated on solid 0.5× MS media supplemented with KARs, rac-GR24, or an equivalent volume of solvent as indicated. Seeds were stratified in the dark for 3 d at 4 °C, then moved to a HiPoint DCI-700 LED Z4 growth chamber to grow at 21 °C under white light (150 µmol m−2 s−1) for 3 h, dark for 21 h, and red light (30 µmol m−2 s−1 unless noted otherwise) for 4 d. Multiple fluences were simultaneously achieved with foil-lined boxes covered with neutral density gel sheets (Lee filters 209, 210, 211, 299) placed under 60 µmol m−2 s−1 red light. Hypocotyl lengths of photographed seedlings were measured with ImageJ (NIH). Cotyledon surface areas were measured in ImageJ from photographs of seedlings mounted in water on glass slides.

Statistical Analysis.

Statistical analysis was performed in Prism 9 (GraphPad). Post hoc statistical comparisons were performed after ANOVA or two-way ANOVA. Box plots show the median, 25th percentile, and 75th percentile. Tukey whiskers on box plots extend 1.5 times the interquartile range beyond the 25th/75th percentile up to the minimum/maximum value in the dataset. Outlier data beyond Tukey whiskers are shown as individual points.