All Arabidopsis thaliana plants were of the Col-0 ecotype. The tcor1-D mutant originated from a population of ethyl methanesulfonate mutagenized M2 seeds produced for another purpose in the jar1-11 background, as described (Leyser and Furner, 1993). The mutant was backcrossed three times to wild type and restoration of the jar1-11 allele to wild type was verified as outlined (Suza and Staswick, 2008). Seed for comparison of wild type and tcor1-D phenotypes in the absence of added Trp conjugates was obtained from three plants of each genotype grown together. The sextuple mutant gh3.1,2,3,4,5,6 was previously described (Porco et al., 2016).

Plant growth conditions were as previously reported (Wei et al., 2015)_. Seedling growth on MS agar medium was in an incubator at 22°C, 12 h day/night cycles, 150 to 200 μM s^-1 m^-2 and seedling root angles, length measurements and germination assays were done as previously described (Staswick et al., 1992; Staswick, 2009). Media supplementation with compounds are indicated in the figures. Trp conjugates showing minimal or no agravitropic response were tested at 50 μM.

Synthesis of new Trp conjugates and their purification was as previously detailed, with verification of the products by thin layer chromatography and gas chromatography/mass spectrometry (Staswick, 2009). The Trp-Trp dipeptide was purchased from GenScript (Piscataway, NJ, USA). Quantification of phenylacetic acid (PAA)-Trp was done as previously described for JA-Trp, using a stable isotope internal standard synthesized from PAA and ¹³C₆-Trp (Staswick, 2009).

Genomic DNA for mapping and genotyping the mutant was prepared as previously reported (Staswick et al., 2002). F2 seeds from the cross of tcor1-D to Landsberg erecta (Ler) were grown 5 days on agar medium containing 50 μM DHJA-Trp, then plates were rotated 180°. Because the mutation was dominant we isolated homozygous individuals that were WT (not restored to the new gravity vector) after 24 to 48 h. After growth in soil leaf tissue was collected for DNA extraction. CAPS and SSLP markers used for bulk segregate analysis, and to identify flanking markers, are shown in Supplementary Table 1.

Previously described methods were used for transformation of the tcor1-D cDNA to WT and aux1-7 plants (Staswick and Tiryaki, 2004). Gene expression was driven by the AUX1 wild type promoter encompassing 2514 bp upstream of the translation start site. Primers used to obtain this sequence from genomic DNA were GGATCCTCGTTGGGTAAAATCTGCAG and GGATCCAGATCTGAGAAATAAAACAGAGCG. The sequence “AGAAGC” 13 bp upstream from translation initiation was changed to a BamH1 recognition sequence to facilitate fusion to the tcor1-D aux1 cDNA, obtained with primers GGATCCTAAAAAAATGTCGGAAGGAGTAGAAGCG and TCTAGATCAAAGACGGTGGTGTAAAGC. The CaMV 35S terminator region obtained from plasmid pRTL2 was added to the 3′ end of the full length tcor1-D coding sequence via Xba1 ends, and the entire transgene was incorporated into pZIP212 for transformation. Seeds from primary transformants were screened for the selectable marker (Kanamycin). Two transformants for each genotype that segregated 3:1 for Kan^R: Kan^S were selected. From each of these homozygous transgenic lines (WT:5-2, WT:6-5, aux:6a, aux:13b) were developed, verified for the transgene and its expression, and used for phenotypic analysis. AUX1 primers for the cDNA CAPS marker were AGAAAGTGATTGGGATGCATG and TCAAAGACGGTGGTGTAAAGC, and for the genomic CAPS marker were TTTCAGCCATTTCTTTATTTGG and CTTAGCACGCATTTAAAGGGG.

Seeds were sown on Petri plates containing a simple medium containing 1 mM KCl, 1 mM CaCl₂, 5 mM MES, and 1% agar adjusted to pH 5.7 with Bis-Tris propane. Plates with seeds were stratified for 2 days at 4°C, then maintained vertically under continuous white fluorescent light at a photon fluence rate of approximately 50 μmol m^-2 s^-1. The temperature was approximately 20°C. After 4 days of growth, the plates were placed in front of computer-controlled CCD cameras equipped with close-focusing telephoto lenses as described by Durham Brooks et al. (2010). After a 1 h period of adaptation to the recording apparatus, the plates were rotated by 90° to initiate gravitropism. Images were automatically acquired every 2 min, and automatically analyzed as described by Durham Brooks et al. (2010).

Earlier experiments showed that IAA- and JA-Trp caused agravitropic root growth in Arabidopsis seedlings, while Trp conjugates of benzoic (BA) and cinnamic acids (CA) did not (Staswick, 2009). To further clarify the structural basis for activity we synthesized a number of Trp conjugates with other carboxylic acids, shown in Figures 1A, 2A. JA and IAA themselves are structurally diverse but they have in common an acetic acid side chain forming the linkage to Trp, whereas BA and CA are linked through one and three carbon carboxylic acid side chains, respectively. Thus we tested the PAA Trp conjugate, the two carbon side chain analog of BA, and found that it caused agravitropic growth similar to that of JA-Trp (Figure 1B). The dihydro analog of JA (DHJA) was also strongly active as a Trp conjugate. In contrast, indole-3-butyric acid, linked to Trp through a four carbon side chain was much less active than PAA-Trp, while the ABA conjugate and the Trp-Trp dipeptide were inactive (Figure 2). This suggests that although the structure of the carboxylic acid itself is not critical, spacing provided by the single carbon bridging the amide linkage to the ring structures of the carboxylic acids is important for activity.

Nevertheless, not all Trp conjugates coupled through an acetic acid moiety were active. The 3,4-dihydroxy derivative of PAA-Trp was weakly active and the 12-OH-jasmonic acid conjugate was inactive (Figure 2). Three conjugates with non-cyclic carboxylic acids were also inactive. We cannot rule out the possibility that inactive conjugates were not assimilated or were more readily catabolized than the active Trp conjugates. However, these results suggest that a bulky moiety spaced by a single carbon from the amide linkage is important for activity.

To elucidate the molecular mechanism of Trp conjugate activity an EMS mutagenized Arabidopsis M₂ seedling population (representing about 13,000 M1 seeds) was screened for insensitive individuals. After 5 days growth on MS agar medium containing 50 μM DHJA-Trp, plates were rotated 180°. DHJA-Trp rather than JA-Trp was used for the screen because JA released from hydrolysis JA-Trp has root growth inhibiting activity, whereas DHJA does not. Putative insensitive seedlings were those that deviated minimally from the gravity vector in the first 5 days, and then reoriented their growth by 180° in response to the new gravity vector within 1–2 days, forming a pronounced hook.

Only one putative mutant maintained a robust phenotype in subsequent tests. The original seedling was found to be heterozygous dominant for DHJA-Trp insensitivity and the mutant is called tcor1-D. Figure 3A shows the phenotypes of WT, homozygous tcor1-D and F₁ seedlings from a cross between these two when grown under the same conditions as for the initial screen. F₂ individuals from this cross segregated 9 sensitive: 23 resistant, consistent with a dominant mutation (Figure 3B).

We next tested tcor1-D for sensitivity to the active conjugates identified in Figure 1 at 25 and 50 μM. Figure 4A shows there was no significant difference in mean root deviation from vertical compared to WT on control medium. The two jasmonate conjugates failed to affect gravitropic response of the mutant at either concentration, consistent with resistance to both conjugates. In contrast, tcor1-D had the same level of sensitivity to PAA-Trp and IAA-Trp as for WT, suggesting the insensitivity of the mutant was specific for the jasmonate conjugates.

Tryptophan conjugates were also previously shown to partially restore growth to seedling primary roots that were inhibited by exogenous IAA (Staswick, 2009). DHJA-Trp and PAA-Trp alone have only a small effect on root length after 6 days and there was no significant difference between WT and the mutant (Figure 4B). JA-Trp and IAA-Trp were not tested because conjugate hydrolysis releases small amounts of free JA and IAA, both of which are potent growth inhibitors that complicate interpretation of the results (Staswick, 2009). IAA at 2 μM strongly inhibits root growth in both genotypes, but DHJA-Trp and PAA-Trp at 50 μM restore WT growth to the level seen for these conjugates alone (compare WT in Figure 4C with 4B). In contrast the efficacy of DHJA-Trp was much lower in tcor1-D at all concentrations tested. The mutant was modestly insensitive to PAA-Trp in this assay, but not to the degree seen for DHJA-Trp. Together, these results show that the tcor1-D mutant has strong resistance to the jasmonate-Trp conjugates, but only weak resistance to PAA-Trp and IAA-Trp.

To identify the gene affected in tcor1-D (Col-0 ecotype) molecular mapping was done in a segregating F2 population resulting from a cross to the Ler ecotype. Bulk segregate analysis indicated linkage on chromosome 2, and the locus was subsequently localized between flanking CAPS markers M429 and T28M21 near the right end of chromosome 2, shown in Figure 5A. Genes associated with auxin function in this chromosomal region include the IAA cellular importer AUX1. We sequenced the AUX1 cDNA from the tcor1-D mutant to determine if any nucleotide changes were evident. The only difference from WT was a G to A substitution in the first nucleotide of the codon for Asp⁴⁵⁶, converting it to an Asn codon. Fortuitously, this eliminated a Sal1 site found in WT genomic DNA and cDNA. Genomic DNA analysis confirmed this mutation in AUX1 from tcor1-D (Figure 5B).

To test whether the mutation in AUX1 was responsible for conjugate insensitivity, the cDNA from tcor1-D was transformed into WT plants under the direction of about 2,500 bases of the WT AUX1 promoter region. Presence of the transgene was verified by amplification of genomic DNA. The size of the endogenous gene fragment with these primers was 1126 bp, while the cDNA transgene was only 446 bp, due to lack of an intron present in the endogenous gene (Figure 5C). Homozygous transformants WT:5-2 and WT:6-5 had a single insertion site based on selectable marker segregation (not shown). Both the WT and cDNA transgene DNA bands were evident in these two transgenic lines (Figure 5C). (The intermediate-sized fragment was apparently a PCR artifact resulting from the transgene). Although the transgene copy number at each insertion site is unknown, the cDNA appears to be preferentially amplified, because the amount of total DNA was equal in each amplification reaction, yet the WT band was much weaker when the transgene was present. To verify that the tcor1-D cDNA transgene was expressed we amplified total cDNA across the region containing the polymorphic Sal1 site. After restriction enzyme digestion WT cDNA yielded fragments of 351 and 91 bp, while tcor1-D DNA produced a single fragment of 446 bp corresponding with the uncleaved amplification product (Figure 5D).

Analysis of the two WT transformants phenocopied with mutant tcor1-D cDNA showed that they acquired the tcor1-D insensitive phenotype for agravitropic response when grown on DHJA-Trp medium, while growth on control medium was not significantly different from WT seedlings (Figure 5E). The apparent wild phenotype of the tcor1-D mutant in the absence of exogenous Trp conjugates suggested that this aux1 allele functions normally in auxin transport. We thus tested whether the tcor1-D transgene could restore normal gravitropic root growth in the agravitropic aux1-7 allele. Genomic and cDNA amplification for the aux:6a and aux:13b transgenic plants gave similar results as for the WT transgenic plants (Figures 5C,D). (The aux1-7 allele is indistinguishable from the WT allele in this PCR assay). Seedling root growth on control medium revealed that the two aux1-7 transformants were restored to the same, or nearly the same, gravitropic response as for WT plants (Figure 5F). Collectively, these results demonstrate that the Asp⁴⁵⁶ conversion to Asn in AUX1 is responsible for the insensitivity of tcor1-D to JA-Trp.

Obvious growth or developmental defects were not readily apparent for tcor1-D when grown on control agar medium or in soil. To assess possibly more subtle effects of the tcor1-D mutation seedling primary root length was evaluated in agar medium with and without added IAA. To minimize confounding effects from differences in seed quality, seed was obtained from WT and tcor1-D plants grown together in a growth chamber. Figure 6A shows results after 6 days of growth. Roots were 15% shorter than WT for tcor1-D when grown on control medium, and similar differences were seen on media supplemented with 50–250 nM IAA. The two genotypes were tested for germination and all were essentially the same, indicating greater root length for WT was not due to earlier germination, but rather to slower growth for the mutant (Figure 6B).

Because Trp conjugates strikingly block root gravitropism we used a machine-vision method (Miller et al., 2007) to test whether the tcor1-D mutation affects the development of the gravitropic response in subtle ways in the absence of added conjugates. Root tip angles were measured automatically by software from images acquired at 2-min intervals. The responses of seedlings produced from three seed pools were combined to generate the results shown in Figure 6C. The only apparent difference between the tcor1-D and wild-type responses was a slightly greater initial tip angle for WT, apparently due to the observed modestly slower growth (Figure 6A). The first derivatives of the tip angle curves, or swing rates, (data not shown) were indistinguishable (Durham Brooks et al., 2010). Thus, despite slightly slower growth of tcor1-D roots the time courses and end-points of the gravitropism response were unaffected by the mutation.

The presence of PAA-Trp was not previously reported in plants, but several GH3 conjugating enzymes are able to use PAA as an acyl substrate (Staswick et al., 2005). Although JA-Trp and IAA-Trp are only found at <2 pmole g^-1 FW in Arabidopsis, Figure 7 shows that PAA-Trp in seedling roots and leaves is about 17-fold higher. To determine whether GH3 enzymes synthesize PAA-Trp in vivo the sextuple mutant gh3.1,2,3,4,5,6 was also examined (Porco et al., 2016). The amount of PAA-Trp was significantly lower in the mutant than in wild type, indicating at least some of these enzymes are partly responsible for synthesis of PAA-Trp.