5-Aryl-substituted IAAs

In our previous report, we showed that the affinity of 5-aryl-substituted IAA for TIR1 ^F79G varies with the substitution on the 5-phenyl group ( Fig.�1A ) ( Uchida et�al. 2018 ). Through a brief screening of 5-aryl-substituted IAAs, we discovered that 5-(3-methoxyphenyl)-IAA ( 2 ) has excellent specificity for TIR1 ^F79G under physiologically relevant conditions. Based on this result, we explored a series of 5-aryl-substituted IAAs to find a new synthetic IAA with higher binding affinity for TIR1 ^F79G . The Suzuki–Miyaura coupling of 5-bromo-IAA methyl ester with various arylboronic acids, followed by hydrolysis of the ester, provided a series of 5-aryl-IAAs ( Fig.�1B ). The activity of these compounds for inducing TIR1–AUX/IAA interaction was estimated by yeast two-hybrid assay ( Fig.�1C ). 5-(3-Ethoxyphenyl)-IAA ( 3 ) possessing a larger substituent than 2 results in a decrease in the affinity for TIR1 ^F79G . In addition, a smaller substituent such as the hydroxyl group in 5-(3-hydroxyphenyl)-IAA ( 4 ) also showed rather weaker activity. Therefore, we abandoned our initial attempts to screen for compounds with stronger activity among the oxygen-linked derivatives.

Instead, we next pursued alkyl-substitution of 5-phenyl-IAA. In this case, the larger substituents also resulted in a decrease in the activity; 5-(3-ethylphenyl)-IAA ( 6 ) has weaker activity than 5-(3-methylphenyl)-IAA ( 5 ). The nitrogen-substituted 5-phenyl-IAAs including 5-(3-nitrophenyl)-IAA ( 7 ), 5-(3-aminophenyl)-IAA ( 8 ) and 5-[3-(dimethylamino)phenyl]-IAA ( 9 ) were not as active as 2 . Finally, we evaluated halogen-substituted 5-phenyl-IAAs such as 5-(3-fluorophenyl)-IAA ( 10 ), 5-(3-chlorophenyl)-IAA ( 11 ) and 5-(3-bromophenyl)-IAA ( 12 ). Among the three halogenated compounds, 11 showed the highest activity, i.e. 100-fold stronger than 2 . However, the activity of 11 is only slightly stronger than that of simple phenyl-substituted IAA ( 13 ) ( Fig.�2A, B ). 5-(2-Naphthyl)-IAA induced the interaction between TIR1 ^F79G and AUX/IAA at moderately low concentration (0.1 �M) in our previous report ( Uchida et�al. 2018 ). Therefore, we tested 5-(2-benzothiophenyl)-IAA ( 14 ), which occupies a slightly different area in the binding pocket of TIR1. However, the activity of 14 was not superior to that of 11 ( Figs.�1C, 2 B).

IAAs with non-aromatic substitutions

Next, we turned our attention to IAAs with non-aromatic substitutions. We synthesized 5-cyclohexyl-IAA ( 15 ) as the saturated six-membered ring occupies larger 3-dimensional space than the aromatic ring ( Scheme�1 ). p -Cyclohexylaniline ( 16 ) was treated with methyl chloroformate to produce 17 , which was converted to 18 by palladium-catalyzed iodination ( Moghaddam et�al. 2016 ). Sonogashira–Hagihara coupling of 18 with trimethylsilyl acetylene afforded 19 . Base treatment of 19 gave 5-cyclohexyl indole ( 20 ). Compound 20 was treated with oxalyl chloride and then water to produce the ketoacid, which was subsequently reduced to obtain 15 . This compound 15 exhibited much higher activity than simple phenyl-substituted auxin ( 13 ) ( Fig.�2B ).

We further expanded the thickness of the substituent to adamantane. 5-Adamantyl-IAA ( 21 ) was synthesized in a similar manner to 15 ( Scheme�2 ). Acetanilide ( 22 ) was treated with 1-bromoadamantane in the presence of zinc chloride to give 23 ( Zurabishvili et�al. 2008 ). The iodination of 23 , followed by Sonogashira–Hagihara coupling and deprotection, afforded 5-adamantyl indole ( 26 ). 26 was converted to the methyl ester ( 27 ), which was hydrolyzed to 5-adamantyl-IAA ( 21 ). As shown in Fig.�2B , 21 showed higher activity than 15 as shown by a stronger β-Gal response at 0.001 �M, which is 1,000 times lower than that of the natural IAA–TIR1 pair (0.1 �M). In fact, the effect of 21 on root growth inhibition in Arabidopsis expressing TIR1 ^F79G is stronger than that of IAA at 0.1 �M, although 21 only slightly affected root growth in Arabidopsis expressing TIR1 ^WT and wild-type plants ( Fig.�3 ; Supplementary Fig. S1 ).

Adjustment of the binding pocket

To look for the strongest pair of synthetic auxin and receptor, we changed the substitution of F79 of TIR1 to residues other than glycine. We previously reported that 2 -induced complexation of TIR1 ^F79A and TIR1 ^F79S is stronger than that of TIR1 ^F79G ( Uchida et�al. 2018 ). Therefore, we tested several compounds synthesized in this study with these engineered receptors. In most cases, synthetic IAA-induced interaction of TIR1 ^F79S with AUX/IAA is 10-fold stronger than that of TIR1 ^F79G , and that of TIR1 ^F79A is a further 10-fold stronger in yeast two-hybrid assay ( Figs.�1, 2, 4 ). Among them, the 5-adamantyl-IAA–TIR1 ^F79A pair showed the strongest interaction; it works at 10 pM, which is 10,000 times lower than the natural auxin–TIR1 pair (100 nM), in the yeast system. The orthogonality of the 5-adamantyl-IAA–TIR1 ^F79A pair to the natural auxin–TIR1 pair is also much higher than that of the previously reported 5-(3-methoxyphenyl)-IAA–TIR1 ^F79G pair. 21 induces the interaction between wild-type TIR1 and IAA3 at 0.1 �M ( Fig.�2B ), which means that the synthetic auxin–TIR1 pair developed in this study works orthogonally to the natural system in the concentration range from 10 pM to 10 nM.

To examine the direct effect of 21 on TIR1–AUX/IAA interaction, we performed a pull-down assay for in vitro interactions of TIR1 ^WT and TIR1 ^F79A with the IAA7 DII peptide in the presence of IAA and 21 , respectively. In our experimental condition, the 21 -induced TIR1 ^F79A –DII interaction was about 100-fold stronger than the IAA-induced TIR1 ^WT –DII interaction ( Fig.�5 ).

Plant culture and treatment

The Arabidopsis thaliana Columbia (Col) accession was used as the wild type. Transgenic plants with the engineered TIR1 constructs were described in our previous study ( Uchida et�al. 2018 ). For root growth assays, Arabidopsis seeds were sterilized, stratified at 4�C for 2–3 d, transferred to 0.5� Murashige and Skoog (MS) liquid medium and grown on a shaker set at 140 r.p.m. under continuous white light at 22�C. After incubation for 1 d, various concentrations of auxin or its analogs were added to the growth media. Root length was measured after an additional week incubation.

Yeast assays

Yeast two-hybrid assays were performed as reported previously ( Uchida et�al. 2018 ). The plasmids were also described before ( Uchida et�al. 2018 ). Briefly, the EGY48 strain was transformed with the LacZ reporter (pSH18-34), DNA-binding domain-fused TIR1 series (pGLex313-based plasmid) and the transcriptional activator-fused IAA3 DI+II domain (pJG4-5-based plasmid). Transformed strains grown on agar plates (–His/–Trp/–Ura) were transferred to liquid –His/–Trp/–Ura medium. After overnight incubation, medium was replaced with Gal/Raf/–His/–Trp/–Ura medium containing 50 mM Na-phosphate buffer (pH 7.0), 80 �g ml ^–1 X-gal (Wako) and the compounds indicated in the figure legends. After a 3 d incubation, culture media were transferred to white 96-well plates and observed.

Pull-down assays

Pull-down assays were performed using biotinyl-DII [biotinyl-(NH)-AKAQVVGWPPVRNYRKN] peptide, Dynabeads M-280 streptavidin beads (Invitrogen) and C-terminally FLAG-tagged TIR1 proteins which were synthesized with a wheat germ extract cell-free system (NUProtein) ( Calder�n Villalobos et al. 2012 , Uchida et�al. 2018 ). mRNAs for FLAG-tagged TIR1 ^WT and TIR1 ^F79A proteins were synthesized by reverse transcription with PCR products amplified by first- and second-strand PCR according to the manufacturer’s instruction (NUProtein). First PCR products were obtained using specific primers as well as pGLex313/TIR1 ^WT and pGLex313/TIR1 ^F79A as templates ( Uchida et�al. 2018 ).

Dynabeads attached to the biotinyl peptides were incubated with the protein synthesis solution containing the synthesized TIR1 proteins and an equal volume of the binding buffer (50 mM Tris–HCl, pH 8.0, 200 mM NaCl, 10% glycerol, 0.1% Tween-20), as well as various concentrations of compounds indicated in the corresponding figures. The beads were washed three times with the binding buffer supplemented with compounds. The proteins extracted from the beads by SDS–PAGE sample buffer were separated by SDS–PAGE, and immunoblot analysis was performed using anti-FLAG antibody (Sigma, F3165), SuperSignal WestPico Chemiluminescence reagent (Thermo Scientific) and a Light-Capture cooled CCD camera system (ATTO). The signal intensity of each immunoreactive band was quantified using ImageJ software.