Genetic evidence suggests that indole-3-butyric acid (IBA) is converted to the active auxin indole-3-acetic acid (IAA) by removal of two side-chain methylene units in a process similar to fatty acid β-oxidation. Previous studies implicate peroxisomes as the site of IBA metabolism, although the enzymes that act in this process are still being identified. Here, we describe two IBA-response mutants, ibr1 and ibr10. Like the previously described ibr3 mutant, which disrupts a putative peroxisomal acyl-CoA oxidase/dehydrogenase, ibr1 and ibr10 display normal IAA responses and defective IBA responses. These defects include reduced root elongation inhibition, decreased lateral root initiation, and reduced IBA-responsive gene expression. However, peroxisomal energy-generating pathways necessary during early seedling development are unaffected in the mutants. Positional cloning of the genes responsible for the mutant defects reveals that IBR1 encodes a member of the short-chain dehydrogenase/reductase family and that IBR10 resembles enoyl-CoA hydratases/isomerases. Both enzymes contain C-terminal peroxisomal-targeting signals, consistent with IBA metabolism occurring in peroxisomes. We present a model in which IBR3, IBR10, and IBR1 may act sequentially in peroxisomal IBA β-oxidation to IAA.
BECAUSE the auxin indole-3-acetic acid (IAA) orchestrates many aspects of plant growth and development (Woodward and Bartel 2005b), the levels of IAA within a plant must be tightly regulated. In addition to changes in IAA biosynthesis and oxidative degradation, IAA also is transformed into alternate forms, allowing the plant to store IAA until the time and place where active auxin is needed. In one type of storage compound, IAA is conjugated to amino acids or peptides by amide bonds or to sugars by ester bonds; conjugate hydrolases break these bonds to release free IAA (Bartel et al. 2001; Woodward and Bartel 2005b). In a second potential storage form, the side chain of the indole moiety is lengthened by two methylene units to make indole-3-butyric acid (IBA), which can be shortened to IAA when necessary (Bartel et al. 2001; Woodward and Bartel 2005b).
Although both IBA and certain auxin conjugates have auxin activity in bioassays, genetic experiments suggest that in Arabidopsis this activity does not result from direct effects of the storage compounds, but rather requires release of free IAA from these precursors (Bartel and Fink 1995; Zolman et al. 2000). Both IAA conjugates and IBA appear to provide auxin during early Arabidopsis seedling development. In particular, mutants defective in IBA metabolism or IAA-conjugate hydrolysis have fewer lateral roots than wild-type plants, suggesting that the IAA released from storage forms plays a role in lateral root promotion (Zolman et al. 2001b; Rampey et al. 2004). Different auxin storage forms may have only partially overlapping activity or primarily regulate different physiological responses. For example, mutants with altered IBA responses have stronger rooting defects than IAA-conjugate mutants (Zolman et al. 2001b; Rampey et al. 2004).
Dedicated enzymes appear to be required to convert auxin storage forms to free IAA. A genetic approach identified a family of Arabidopsis hydrolases showing overlapping specificity in the conversion of various IAA–amino acid conjugates to IAA; mutants defective in each enzyme have altered responses to application of the corresponding conjugates (Bartel and Fink 1995; Davies et al. 1999; Rampey et al. 2004). We are taking a similar genetic approach to discover the enzymes required for conversion of IBA to IAA.
Even-numbered side-chain-length derivatives of IAA (Fawcett et al. 1960) and the synthetic auxin 2,4-dichlorophenoxyacetic acid (2,4-D; Wain and Wightman 1954) possess auxin activity. Wheat and pea extracts shorten these compounds in two-carbon increments (Fawcett et al. 1960). These results suggest that IBA, which is structurally identical to IAA but with two additional methylene units on the side chain, and 2,4-dichlorophenoxybutyric acid (2,4-DB), the analogous elongated derivative of 2,4-D, are converted by plants to bioactive IAA and 2,4-D, respectively. The mechanism of this conversion was suggested to be similar to the two-carbon elimination that occurs during fatty acid β-oxidation (Fawcett et al. 1960). In this model, IBA acts as a “slow-release” form of IAA (van der Krieken et al. 1997), and auxin effects promoted by IBA would be limited by the β-oxidation rate.
To elucidate the proteins necessary for IBA activity, we have isolated Arabidopsis mutants with altered IBA responses. Exogenous IBA inhibits primary root elongation; mutants that cannot sense or respond to IBA have elongated roots on IBA compared to wild type (Zolman et al. 2000). IBA-response (ibr) mutants remain sensitive to short-chain auxins (IAA and 2,4-D), whereas general auxin-response mutants defective in auxin signaling, such as axr2/iaa7 (Timpte et al. 1994; Nagpal et al. 2000) and axr3/iaa17 (Leyser et al. 1996; Rouse et al. 1998), or transport, including aux1 (Pickett et al. 1990; Bennett et al. 1996; Marchant et al. 1999), display reduced responses to both IBA and IAA (Zolman et al. 2000).
Isolation of the genes defective in ibr mutants has revealed a close connection between IBA metabolism and peroxisomal function. For example, mutants defective in the peroxins PEX4 (Zolman et al. 2005), PEX5 (Zolman et al. 2000), PEX6 (Zolman and Bartel 2004), PEX7 (Woodward and Bartel 2005a), and PEX12 (Fan et al. 2005) or in the PXA1/CTS1/PED3 transporter (Zolman et al. 2001b), which is required for import of peroxisomal substrates (Footitt et al. 2002; Hayashi et al. 2002; Theodoulou et al. 2005), are resistant to IBA. The isolation of IBA-response mutants defective in these factors, coupled with the fact that peroxisomes are the primary site of fatty acid β-oxidation in plants (Graham and Eastmond 2002; Baker et al. 2006), implicates peroxisomes as the subcellular location for IBA-to-IAA conversion. Mutants defective in peroxisomal biogenesis likely disrupt IBA responses by compromising the peroxisomal environment or by limiting peroxisomal import of enzymes necessary for IBA β-oxidation. Indeed, these ibr mutants display additional phenotypes associated with peroxisome defects, such as sucrose dependence during seedling development due to slowed β-oxidation of seed storage fatty acids (Zolman et al. 2000).
Other ibr mutants may be defective in enzymes that act in the peroxisome matrix to catalyze the β-oxidation of fatty acids, IBA, or both. In Arabidopsis, at least two genes encode related isozymes for each fatty acid β-oxidation step (Baker et al. 2006). Following substrate import into the peroxisome by PXA1, a CoA moiety is added by one of two long-chain acyl-CoA synthetases (LACS; Fulda et al. 2002, 2004). One of six acyl-CoA oxidases (ACX) catalyzes the oxidation step, adding a double bond while releasing hydrogen peroxide (Hayashi et al. 1999; Hooks et al. 1999; Eastmond et al. 2000; Froman et al. 2000; Rylott et al. 2003; Adham et al. 2005; Pinfield-Wells et al. 2005). Next, a multifunctional protein (MFP2 or AIM1), containing both enoyl-CoA hydratase and acyl-CoA dehydrogenase activity, forms a β-ketoacyl-CoA thioester (Richmond and Bleecker 1999; Eastmond and Graham 2000; Rylott et al. 2006). This substrate undergoes a thiolase-mediated retro-Claisen reaction, releasing two carbons as acetyl-CoA and producing a chain-shortened substrate that can reenter the β-oxidation spiral for continued catabolism. PED1/KAT2 encodes the most highly expressed thiolase in Arabidopsis (Hayashi et al. 1998; Germain et al. 2001).
We recently described the ibr3 mutant, which displays defective IBA and 2,4-DB responses but responds normally to other conditions tested (Zolman et al. 2007). IBR3 (At3g06810) encodes a putative peroxisomal acyl-CoA dehydrogenase or oxidase that resembles the mammalian ACAD10 and ACAD11 enzymes. Because the ibr3 mutant does not have apparent fatty acid β-oxidation defects, IBR3 may act in IBA β-oxidation to IAA (Zolman et al. 2007). In addition to ibr3, mutants defective in several fatty acid β-oxidation enzymes, including aim1, ped1, and multiple acx mutants, show altered responses to exogenous IBA (Zolman et al. 2000; Adham et al. 2005); whether these IBA-response defects reflect a direct enzymatic role on IBA β-oxidation intermediates or an indirect disruption of IBA-to-IAA conversion remains an open question.
Here, we describe the phenotypic characterization of ibr1 and ibr10, two strong IBA-response mutants that, like ibr3, have no apparent defects in peroxisomal fatty acid β-oxidation. We used map-based cloning to demonstrate that IBR1 and IBR10 encode a short-chain dehydrogenase/reductase (SDR) family enzyme and a putative enoyl-CoA hydratase/isomerase (ECH), respectively. Both IBR1 and IBR10 have peroxisomal-targeting signals. The specific IBA-response defects of ibr1 and ibr10 mutants suggest that IBR1 and IBR10 are needed for the peroxisomal conversion of IBA to IAA.
MATERIALS AND METHODS
The ibr1-1 and ibr1-2 mutants were described previously as B1 and B19, ethyl methanesulfonate (EMS)-induced IBA-response mutants in the Arabidopsis thaliana Columbia (Col-0) background (Zolman et al. 2000). Additional screens of various mutagenized populations revealed several new mutants with IBA-resistant root elongation. ibr1-3, ibr1-4, and ibr1-5 were isolated from the progeny of Col-0 seed mutagenized by fast-neutron bombardment (Zolman et al. 2007). ibr1-8 was isolated from the progeny of EMS-mutagenized Col-0 seeds (Lehle Seeds, Round Rock, TX). ibr10-1 was isolated by screening T-DNA insertion lines in the Col-0 accession (pool CS75075; Alonso et al. 2003), although our analysis indicated that the T-DNA was not linked to the IBA-resistant phenotype (data not shown). chy1-3 (Zolman et al. 2001a) and ibr3-1 (Zolman et al. 2007) were described previously and are Col-0 alleles with point mutations in CHY1/At5g65940 and IBR3/At3g06810, respectively.
The ibr1-7 (SALK_010364), hcd1/at4g14440 (SALK_012852), and echic/at1g65520 (SALK_036386) mutants in the Col-0 accession were from the Salk Institute sequence-indexed insertion collection (Alonso et al. 2003). The three insertion mutants were genotyped using PCR amplification with a combination of genomic primers and a modified LBb1 T-DNA primer (5′-CAAACCAGCGTGGACCGCTTGCTGCA-3′; http://signal.salk.edu). The T-DNA position in ibr1-7 was confirmed by sequencing the amplification product directly.
Higher-order mutants were generated by crossing and were identified using PCR-based genotyping. For ibr1-2, amplification with the oligonucleotides T1J24-8 (5′-GAAGCTTTACCTGCAGGAGAAGTATAGAGG-3′) and T1J24-12 (5′-TAAGAGATGTCTTCTGTGTTTTTGGACTCA-3′) yields a 175-bp product with one DdeI site in wild type that was absent in ibr1-2. For ibr10-1, amplification with At4g14430-1 (5′-ATTTGCACAATTCAACAACAACACGATTTC-3′) and At4g14430-2 (5′-TAGCCCTAACCAACGCCGAGAAATAATC-3′) yields a 546-bp product in wild type and a 468-bp product in ibr10-1. ibr3-1 was genotyped using a derived cleaved amplified polymorphic sequence marker (Michaels and Amasino 1998; Neff et al. 1998) in which amplification with F3E22-22 (5′-ATGGTGCAGTCTTCCAGGGCCTAACCTAGC-3′; altered residue underlined) and F3E22-23 (5′-GTTTTGATGACGCACCTCATGGACATGCTG-3′) yields a 240-bp product with one AluI site in ibr3-1 that was absent in wild type.
Plant growth and phenotypic characterization:
Surface-sterilized seeds were plated on plant nutrient (PN) medium (Haughn and Somerville 1986) solidified with 0.6% (w/v) agar and supplemented with 0.5% sucrose (PNS), hormones, kanamycin, or Basta (glufosinate–ammonium; Crescent Chemical, Augsburg, Germany) as indicated. Plates were incubated under continuous light at 22°; for auxin experiments, plates were placed under yellow filters to slow breakdown of indolic compounds (Stasinopoulos and Hangarter 1990). Seedlings were transferred to soil and grown at 18°–22° under continuous illumination.
Prior to phenotypic analyses, ibr1-2 and ibr10-1 were backcrossed to the parental Col-0 accession four and two times, respectively. For root elongation assays, seeds were plated on PNS supplemented with hormones at the indicated concentrations. Roots were measured after 7 or 8 days at 22° under yellow light. For some assays, seeds first were stratified for 3 days in 0.1% agar at 4° and germinated under white light for 2 days at 22° before being moved to assay plates. For lateral root assays, seeds were stratified for 3 days at 4°, grown on PNS for 4 days under yellow light at 22°, and then transferred to PNS supplemented with the indicated hormone and grown for 4 days under yellow light at 22°. To assay sucrose-dependent hypocotyl elongation, seeds were stratified for 3 days at 4° and grown on PN or PNS for 1 day under white light followed by 5 days in the dark at 22°.
ibr10-1 was crossed to Col-0 carrying a DR5-GUS transgene (Guilfoyle 1999). Lines homozygous for DR5-GUS and ibr10-1 were selected on 12 μg/ml kanamycin and by genotyping as described above. For analysis of DR5-GUS reporter gene activation, seeds were germinated on PNS and grown for 4 days and then transferred to PNS supplemented with auxin for the indicated number of days. Histochemical localization was done by staining homozygous seedlings for 2 days at 37° with 0.5 mg/ml 5-bromo-4-chloro-3-indolyl-β-d-glucuronide (Bartel and Fink 1994).
Positional cloning and mutant complementation:
ibr1 and ibr10 alleles were outcrossed to Landsberg erecta tt4 (Ler) and Wassilewskija accessions for recombination mapping. Mutations were mapped using PCR-based molecular markers (http://www.arabidopsis.org) on DNA from IBA-resistant F2 plants. ibr1-8 was identified as an ibr1 allele on the basis of failure to complement ibr1-1. Candidate genes were PCR amplified from mutant DNA and sequenced directly using gene-specific primers.
To make the 35S-IBR1c construct, the SalI/NotI insert from the 103N7T7 EST (Newman et al. 1994) was ligated into the 35SpBARN (Leclere and Bartel 2001) plant transformation vector cut with XhoI and NotI. An IBR1 genomic rescue construct was made by digesting the T1J24 bacterial artificial chromosome containing IBR1 with BglII, yielding a 4-kb fragment containing the IBR1 coding sequence plus 1278 bp 5′ and 1118 bp 3′ of the coding region. This fragment was subcloned into the BamHI site of pBluescript KS+ to give pKS-IBR1g. An EcoRI/XbaI fragment was subcloned from pKS-IBR1g into the plant transformation vector pBIN19 (Bevan 1984) cut with the same enzymes to give pBIN-IBR1g.
For complementation of the ibr10 mutant, Col-0 DNA was amplified with PfuTurbo DNA polymerase (Stratagene, La Jolla, CA) using primers modified to add SalI and NotI restriction enzyme sites (added residues underlined): At4g14430-5 (5′-CGTCGACCATAGACCAATCAGGATAAGGTATATTTCTCTC-3′) and At4g14430-6 (5′-GAGGCGGCCGCTGTCTCTTTCTCCCAAAACATGTG-3′). The purified PCR product was cloned into the PCR4-Blunt TOPO vector following the manufacturer's instructions (Invitrogen, Carlsbad, CA) to give pTOPO-IBR10c. The insert of pTOPO-IBR10c was sequenced to ensure the absence of PCR-derived mutations. The SalI/NotI fragment from pTOPO-IBR10c was subcloned into XhoI/NotI-cut 35SpBARN to give 35S-IBR10.
Complementation constructs were introduced into Agrobacterium tumefaciens GV3101 (Koncz et al. 1992) by electroporation (Ausubel et al. 1999) and transformed into mutant alleles using the floral dip method (Clough and Bent 1998). Transformed T1 seedlings and homozygous T3 lines were identified by selecting seedlings resistant to 7.5 μg/ml glufosinate–ammonium (35S-IBR1c and 35S-IBR10) or 12 μg/ml kanamycin (pBIN-IBR1g).
ibr1 and ibr10 have IBA-response phenotypes:
Isolation and characterization of mutants with altered responses is a powerful method for identifying proteins involved in specific metabolic pathways. We screened mutagenized seed pools for mutants with long roots on normally inhibitory IBA concentrations. In addition to two previously described ibr1 alleles from EMS-mutagenized pools (Zolman et al. 2000), we isolated an additional EMS-induced ibr1 allele, three ibr1 alleles from a fast-neutron mutagenized population, and a single ibr10 allele from a T-DNA-mutagenized population (Alonso et al. 2003). Both ibr1 and ibr10 mutants were resistant to the inhibitory effects of IBA on root elongation (Figure 1A). All of the ibr1 alleles displayed a similar level of IBA resistance (Figure 1A and data not shown) and we chose ibr1-2 for more detailed analysis.
We examined IBA-response defects under several conditions, comparing ibr1 and ibr10 with the previously described chy1-3 (Zolman et al. 2001a) and ibr3-1 (Zolman et al. 2007) mutants. First, we examined the effects of increasing IBA concentrations on root elongation. Although the mutants had longer roots than wild type on intermediate IBA levels, all of the mutants responded to IBA at higher concentrations (Figure 1B). We found that ibr1 was more IBA resistant than ibr3 and that ibr10 was more IBA resistant than ibr1, ibr3, and chy1 (Figure 1B). Like ibr3 and chy1, ibr1 and ibr10 mutants responded normally to IAA (Figure 1A) and to synthetic auxins that are not β-oxidation substrates, such as 1-naphthaleneacetic acid (NAA) and 2,4-D (Figure 1C). ibr1 and ibr10 mutants also were resistant to 2,4-DB (Figure 1C), a chain-elongated version of 2,4-D that may be converted to 2,4-D by β-oxidation (Hayashi et al. 1998). The phenotypes of ibr1 and ibr10—resistance to IBA and 2,4-DB coupled with sensitivity to IAA, NAA, and 2,4-D—closely matches the ibr3 and chy1 phenotypes.
In addition to assaying root elongation inhibition, we tested the ibr1 and ibr10 mutant responses to the stimulatory effects of exogenous auxins on lateral root proliferation. At the concentrations used in our assay, wild-type plants produced similar numbers of lateral roots in response to IBA or NAA induction. Although the ibr1 and ibr10 mutants produced normal numbers of lateral roots in response to NAA (Figure 2A) or IAA (data not shown) treatment, both mutants made dramatically fewer lateral roots than wild type in response to IBA stimulation (Figure 2A).
The auxin-inducible DR5-GUS reporter (Guilfoyle 1999) is expressed in root tips and lateral root primordia, facilitating the visualization of lateral root induction. In wild-type lines carrying the DR5-GUS transgene, lateral root primordia were apparent in response to IBA induction (Figure 2B). In ibr10-1 DR5-GUS, we found fewer lateral roots and less root staining in response to IBA (Figure 2B). In contrast, we found similar GUS induction throughout ibr10-1 and wild-type DR5-GUS lines following NAA treatment (Figure 2B). Similar results were seen in ibr1-2 and ibr3-1 lines carrying the reporter (data not shown). We concluded from these experiments that, like ibr3, ibr1 and ibr10 display defective responses to IBA and 2,4-DB, but maintain normal responses to other auxins.
Higher-order mutants show similar IBA-response phenotypes:
Because ibr1, ibr3, and ibr10 mutants were all incompletely defective in IBA responses (Figure 1B), we generated double and triple mutants to determine if we could further block IBA responsiveness. In root elongation assays, all of the mutant combinations were IBA resistant compared to wild type at intermediate IBA concentrations (Figure 1B). At very high IBA concentrations, all of the double and triple mutants still responded to IBA. Interestingly, at high IBA concentrations, the ibr1 ibr3 ibr10 triple mutant was less resistant than the previously described (Zolman et al. 2007) ibr3 chy1 double mutant (Figure 1D). The limited enhancement of IBA resistance seen in higher-order ibr mutants suggests that the IBR enzymes may act in the same pathway. The stronger resistance of ibr3 chy1 suggests that the IBR pathway may promote IBA responses independently of CHY1. chy1 is believed to indirectly disrupt β-oxidation at the thiolase step (Zolman et al. 2001a; Lange et al. 2004).
The double and triple mutants still responded like wild type to NAA and 2,4-D and were 2,4-DB resistant (Figure 1C), consistent with the defect specific to chain-elongated auxins seen in the single ibr mutants. Also like the single mutants, the double and triple mutants made few lateral roots in response to IBA induction, but responded like wild type to induction by NAA (Figure 2A). Because so few lateral roots were made in the single mutants in the presence of IBA, we did not attempt to assess the significance of additive or altered effects on lateral rooting in the higher-order mutants.
Other peroxisomal pathways are not detectably disrupted in ibr1 and ibr10:
Previously isolated IBA-response mutants fall into two categories: those with general peroxisomal defects, including reduced rates of fatty acid β-oxidation and consequent sucrose dependence during germination, and those that are IBA resistant but appear to carry out other peroxisomal functions normally (Zolman et al. 2000). To test whether these new IBA-response alleles had general defects in peroxisomal metabolism manifested in reduced fatty acid β-oxidation, we tested ibr1 and ibr10 requirements for sucrose during development. Wild-type seedlings can develop without an exogenous carbon source; early seedling growth is fueled by peroxisomal catabolism of seed storage fatty acids (Hayashi et al. 1998). Mutants defective in seed oil catabolism, due to either loss of a single β-oxidation enzyme or disruption of general peroxisomal function, arrest during early seedling development. Supplementing the growth medium with sucrose, which provides the energy normally obtained from fatty acid β-oxidation, can relieve the arrest. Therefore, comparison of growth with and without exogenous sucrose can reveal defects in peroxisomal function. We found that dark-grown hypocotyls of ibr1 and ibr10 elongated normally in the absence of sucrose (Figure 3), suggesting that seed storage fatty acids are efficiently metabolized in both mutants. Indeed, previous gas chromatography–mass spectrometry analysis indicated that seed-storage fatty acids are metabolized normally during early seedling development in ibr1-1 and ibr1-2 (Zolman et al. 2000). We also assayed the ibr double and triple mutants and found that none displayed sucrose-dependent growth compared to the chy1 sucrose-dependent control (Figure 3A). We concluded that, even in combination, the ibr1 and ibr10 mutations do not appreciably disrupt fatty acid utilization during development.
In addition to IBA and fatty acids, peroxisomes are implicated in the metabolism of other substrates. CHY1 encodes a peroxisomal β-hydroxybutyryl-CoA hydrolase similar to mammalian mitochondrial enzymes acting in valine catabolism (Zolman et al. 2001a). chy1 mutants are sucrose dependent (Zolman et al. 2001a), resistant to IBA (Zolman et al. 2001a) and 2,4-DB (Lange et al. 2004), and display heightened sensitivity to root elongation inhibition caused by isobutyrate and propionate treatment (Lucas et al. 2007). These defects are thought to result from accumulation of toxic catabolic intermediates in the mutant. To determine whether the IBR1, IBR3, or IBR10 enzymes might be involved in metabolism of propionate or the branched-chain substrate isobutyrate, we assayed responses of the ibr mutants to these compounds. Unlike chy1, the ibr single-mutant roots were able to elongate like wild type on isobutyrate and propionate (Figure 3B), suggesting that the IBR proteins are not needed for metabolism of these compounds.
acx1 acx5 double mutants have decreased male fertility and reduced jasmonate production, suggesting that peroxisomal ACX enzymes are important for chain shortening of jasmonate precursors (Schilmiller et al. 2007) in addition to fatty acids (Adham et al. 2005; Pinfield-Wells et al. 2005). In contrast, the ibr1 ibr3 ibr10 triple mutant was fully fertile, suggesting that the IBR enzymes are not essential for jasmonate biosynthesis. Soil-grown ibr1, ibr3, and ibr10 single, double, and triple mutants also were not notably different than wild type in size, morphology, or pigmentation (data not shown).
Molecular characterization of the ibr1 mutants:
Using positional information, we mapped the ibr1 mutant defect to a 3-Mb region in the middle of chromosome 4 between the molecular markers F4C21 and mi167 (Figure 4A). Within this region are the chromosome 4 centromere and a large zone of reduced recombination (Copenhaven et al. 1999), as well as a chromosomal rearrangement between Col-0 and Ler (Fransz et al. 2000). We identified a candidate gene (At4g05530) within the ibr1 mapping interval that encodes a protein in the SDR family with a C-terminal peroxisomal-targeting signal; recent proteomic analysis confirmed that the At4g05530 protein is peroxisomal (Reumann et al. 2007). Because of previous connections between IBA response and peroxisomal function (see above), we hypothesized that disruption of this peroxisomal enzyme might cause the ibr1 phenotypes.
We sequenced At4g05530 using DNA from the ibr1-1 mutant and identified a C-to-T mutation at position 126 (where 1 is the A of the initiator ATG), which alters an Arg to a Cys residue (Figure 4B). Figure 5 shows an alignment of similar proteins from different kingdoms; Arg126 is conserved in 25/25 IBR1 homologs. ibr1-8 has a mutation at position 1090, which converts a Ser residue to Phe. In ibr1-2, we found a G-to-A mutation at position 828 that disrupts the 3′-splice junction following the second intron. The mutation changes the absolutely required G in the splice site, presumably destroying the splice site and altering the downstream protein. In ibr1-3, we found a 1-bp deletion at position 859, causing a frameshift in the mutant protein. The first 94 amino acids of the mutant ibr1-3 protein are identical to IBR1, followed by 50 amino acids translated in a different frame and a stop codon; the C-terminal peroxisomal-targeting signal in this protein is lost.
The ibr1-4 and ibr1-5 mutants were isolated from distinct pools of fast-neutron mutagenized seeds. Both mutants appear to harbor a large deletion removing all of IBR1, as multiple primer sets spanning the gene did not amplify products that were present in wild-type samples (Figure 4C). Furthermore, amplification reactions using primers 1500 bp upstream or 850 bp downstream did not yield products. For both ibr1-4 and ibr1-5 mutant alleles, we were able to amplify genes in other regions of the genome (Figure 4C), indicating that we had successfully prepared DNA from the mutants. However, we did not explore the precise boundaries of the deletions in either ibr1-4 or ibr1-5 and do not know whether additional genes are affected in these alleles.
We obtained a T-DNA insertion mutant (SALK_010364) in IBR1 from the Salk Institute collection (Alonso et al. 2003). Sequence analysis confirmed that this mutant (ibr1-7) had an insertion at position 469 relative to the start codon, near the end of intron 1 (Figure 4B). When we compared the IBA responses of the ibr1 alleles, we found that all were similarly defective in IBA responsiveness (Figure 1A), indicating that we are likely examining the null phenotype for this gene.
At4g05530 complements the ibr1 mutants:
To complement the mutant alleles, we made both a genomic construct driving IBR1 expression from the IBR1 5′ regulatory region and an overexpression construct driving an IBR1 cDNA from the constitutive 35S promoter from cauliflower mosaic virus. ibr1-1 was rescued similarly following transformation with either the genomic or overexpression construct (Figure 4, D and E), indicating that both constructs provided sufficient protein for wild-type function and that any overexpression did not affect the response. In addition, we found that expression of genomic IBR1 in ibr1-3, ibr1-4, and ibr1-5 restored IBA responses in root elongation and lateral root initiation (Figure 4, D and E, and data not shown). We concluded that the IBA-response defects in ibr1 result from disruption of the IBR1/At4g05530 gene.
Molecular characterization of the ibr10 mutant:
We mapped the ibr10-1 lesion to the middle of chromosome 4, 5.5 Mb below IBR1 (Figure 6A). Within the mapping region, we identified a gene encoding a putative enoyl-CoA hydratase (At4g14430). The At4g14430 protein is similar to enzymes acting in fatty acid β-oxidation (Figure 7) and recently was confirmed to localize in peroxisomes (Reumann et al. 2007). Sequencing this candidate gene revealed a 78-bp deletion in the ibr10-1 mutant (Figure 6B). This deletion is predicted to remove 26 amino acids from the 240-amino-acid protein, but leave the C terminus in frame with wild-type IBR10. The deletion removes several conserved amino acids; however, a predicted active site residue and the peroxisomal-targeting signal remain in the mutant protein (Figure 7). If the mutant protein is able to fold correctly, retention of this region may be sufficient to allow partial enzyme activity. Because publicly available T-DNA insertion collections contain no insertions in the At4g14430 coding sequence, we currently are unable to determine whether the single ibr10-1 allele displays a null phenotype.
At4g14430 complements the ibr10-1 mutant:
Because we recovered only a single ibr10 allele, we sought to confirm that the mutant phenotypes were due to At4g14430 disruption. We created an IBR10 overexpression construct that drives expression of IBR10 from the cauliflower mosaic virus 35S promoter. ibr10-1 plants transformed with the 35S-IBR10 construct displayed restored responses to IBA in both root elongation and lateral root initiation assays (Figure 6, C and D), indicating that the identified lesion in At4g14430 caused the IBA-response defects that we observed in ibr10-1.
Mutants defective in the IBR10 homologs At4g14440 and At1g65520 respond normally to IBA:
IBR10 has two close Arabidopsis homologs. To determine if either of these proteins contributes to IBA responses, we examined the phenotypes of insertion mutants disrupting these two genes. The closest IBR10 relative in Arabidopsis is At4g14440 (80% identical; Figure 7), which is encoded directly adjacent to IBR10 and likely results from a recent duplication. However, At4g14440 ends with the C-terminal tripeptide “HNL,” which is not a known peroxisomal-targeting signal (Reumann et al. 2004). We isolated a T-DNA insertion line (SALK_012852) predicted to disrupt At4g14440 (Figure 6B). The mutant responded like wild type to IBA (Figure 6E). We similarly examined a T-DNA insertion allele (SALK_036386) disrupting a second IBR10 homolog At1g65520, which encodes a protein 49% identical to IBR10 that contains the canonical peroxisomal-targeting signal “SKL” (Figure 7). We found that this mutant also displayed wild-type IBA responsiveness (Figure 6E). We concluded that the two closest IBR10 homologs in the Arabidopsis genome are not required for IBA-responsive root elongation inhibition.
Expression of IBR genes:
The phenotypic similarities among the ibr mutants suggest that IBR1, IBR3, and IBR10 might be involved in similar processes. We used the Genevestigator database of microarray data (Zimmermann et al. 2004) to compare the expression patterns of IBR1, IBR3, and IBR10 with genes encoding known fatty acid β-oxidation proteins (Figure 8). This analysis revealed that the three IBR genes show similar constitutive moderate-level expression in most tissues examined with apparent increases in seeds (Figure 8B). This expression pattern is similar to that of fatty acid β-oxidation genes, except that the fatty acid β-oxidation genes generally have higher expression levels (Figure 8D). In particular, LACS6, LACS7, ACX3, MFP2, and PED1 seem to have a similar pattern as IBR gene expression, albeit at apparently higher levels. Interestingly, IBR1 and IBR3 transcripts appear to accumulate in senescing leaves, as do many of the fatty acid β-oxidation genes (Figure 8, B and D).
The isolation and characterization of two new IBA-response mutants, ibr1 and ibr10, along with the map-based identification of the defective genes, revealed two novel peroxisome-targeted proteins. On the basis of mutant phenotypes and the identity of the encoded proteins, we hypothesize that IBR10 and IBR1 may act in sequential steps in the peroxisomal conversion of IBA to IAA.
IBR10 contains an ECH domain (pfam00378) and has been annotated as ECHIb (Reumann et al. 2007). In a subset of characterized family members, including bacterial hydratases and a yeast isomerase, a conserved Glu residue is required for catalysis (Mursula et al. 2001; Wong and Gerlt 2004); this residue is present in IBR10 and retained in the ibr10 mutant protein (Figure 7). The substrates metabolized by ECH enzymes are diverse, although common structural intermediates are typical and there is strong conservation within the CoA-binding site (Agnihotri and Liu 2003). Enzymes in the ECH family often act in reversible reactions and include not only isomerases and hydratases, but also napthoate synthase, carnitine racemase, and hydroxybutyryl-CoA dehydratases. In fatty acid β-oxidation, ECH family enzymes catalyze a hydration reaction to produce hydroxyacyl-CoA thioester intermediates. For example, the Arabidopsis peroxisomal MFP2 and AIM1 multifunctional proteins acting in fatty acid β-oxidation each contain an ECH domain and have enoyl-CoA hydratase activity (Richmond and Sommerville 2000; Rylott et al. 2006). Unlike these large multifunctional proteins, which also contain l-3-hydroxyacyl-CoA dehydrogenase and d-3-hydroxyacyl-CoA epimerase domains (Richmond and Sommerville 2000; Rylott et al. 2006), IBR10 is a single domain protein, reminiscent of enzymes acting in mammalian β-oxidation.
Analysis of other sequenced genomes revealed single-domain IBR10 homologs in several plants, including both monocots and dicots (Reumann et al. 2007). The closest IBR10 relative in Arabidopsis is At4g14440, which is 80% identical to IBR10 at the amino acid level but lacks a known peroxisomal-targeting signal. This IBR10 homolog, recently named AtHCD1, can act as a hydroxyacyl-CoA dehydratase on very-long-chain substrates in wax biosynthesis (Garcia et al. 2007). An insertion mutant disrupting AtHCD1 (SALK_012852) contains reduced wax levels and altered wax crystal morphology (Garcia et al. 2007). We tested the hcd1/at4g14440 mutant and found wild-type IBA responses (Figure 6E), suggesting that these two highly related proteins are not functionally redundant in the examined responses. A second Arabidopsis homolog, At1g65520 (ECHIc), is 49% identical to IBR10 and is peroxisomal (Reumann et al. 2007); disruption of At1g65520 also did not alter IBA responses (Figure 6E). Rice contains two clear IBR10 homologs; as in Arabidopsis, one has a predicted peroxisomal localization whereas the other does not (Figure 7).
Outside the plant kingdom, IBR10 has limited similarity (∼20% identical) to a mammalian peroxisomal enoyl-CoA isomerase (PECI; Geisbrecht et al. 1999) that is localized in both peroxisomes and mitochondria (Geisbrecht et al. 1999; Zhang et al. 2002). This enzyme has Δ3Δ2enoyl-CoA isomerase activity (Geisbrecht et al. 1999), which is required for rearranging double bonds to allow complete β-oxidation of unsaturated fatty acids. The Saccharomyces cerevisiae homolog, Eci1p, also has isomerase activity but lacks hydratase activity (Geisbrecht et al. 1998). Human PECI has an N-terminal extension compared to the other homologs (Figure 7) that resembles acyl-CoA-binding proteins (Poirier et al. 2006).
IBR1 is in the SDR family, a large group of proteins with highly variable enzymatic functions. SDR proteins can have oxidoreductase, lyase, isomerase, or related activities and act on a wide variety of substrates, including steroids, sugars, and alcohols. Each enzyme requires NAD, NADP, or FAD as an electron carrier, which typically is bound by a Rossman fold within the N-terminal region of the protein; the identity of this cofactor is used as a partial determinant of enzyme classification (Kallberg et al. 2002; Kallberg and Persson 2006). The C-terminal region is typically less conserved and is important for substrate interactions.
Arabidopsis IBR1 is in the dehydrogenase subcategory with different or unknown specificities (FabG/COG1028 and COG4221; Marchler-Bauer and Bryant 2004). On the basis of protein size and amino acid residues at conserved domains, IBR1 has been classified as a “classical” SDR (Kallberg et al. 2002, 2007) and also has been annotated as SDRa (Reumann et al. 2007). IBR1 has an apparent nucleotide-binding site with a predicted preference for an NADP(H) cofactor (http://www.ifm.liu.se/bioinfo/; Kallberg and Persson 2006). In addition, the three SDR catalytic amino acids (Kallberg et al. 2002) are conserved in IBR1 (Ser146, Tyr159, and Lys163), along with other less tightly conserved amino acid sequences indicative of SDR classification (Figure 5; Kallberg et al. 2002; Reumann et al. 2007).
IBR1 homologs are present in animals, bacteria, and plants. According to the classification system of Kallberg et al. (2007), IBR1 is the only Arabidopsis family member in cluster 5c. Human HEP27 (also known as DHRS2) is the most characterized member of this cluster and is 45% identical to IBR1. HEP27 is a nuclear protein (Pellegrini et al. 2002) and can reduce dicarbonyl compounds, but not sugars, steroids, or retinoids (Shafqat et al. 2006). The mammalian protein NRDR (also known as DHRS4, HEP27-like, RRD, PHCR, SDR-SRL, and SCAD-SRL) is a closely related homolog that acts as a retinol dehydrogenase and a retinal reductase regulating retinoid metabolism (Lei et al. 2003). This enzyme has an “SRL” peroxisomal-targeting signal, like Arabidopsis IBR1, and has been shown to localize to peroxisomes using a fluorescent reporter protein (Fransen et al. 1999). Furthermore, NRDR activity has been isolated from mouse peroxisomes and increases in the presence of peroxisome proliferators (Lei et al. 2003). Although characterization of the human NRDR protein lends credence to the importance of IBR1 in peroxisomal metabolic pathways, retinol (vitamin A) or retinal are unlikely substrates for IBR1, as these molecules are not observed in plants (Wolf 1984).
Recent proteomic analysis of soybean peroxisomal proteins revealed GmSDR, which is 78% identical to IBR1 (Arai et al. 2008), and analysis of sequenced databases identified IBR1 homologs in a variety of monocot and dicot species (Reumann et al. 2007). None of the closely related IBR1 plant homologs have known substrates, although a Solanum tuberosum tropinone reductase (34% identical to IBR1) has been characterized (Keiner et al. 2002). Tropinone reductases are NADPH-dependent enzymes acting in alkaloid metabolism; these enzymes do not appear to be localized to peroxisomes. More distant Arabidopsis relatives include ABA2, which acts in ABA biosynthesis (Cheng et al. 2002), and the ATA1 enzyme homologous to TASSELSEED 2 (Lebel-Hardenack et al. 1997), involved in sex determination in maize (Delong et al. 1993).
We can envision both direct and indirect mechanisms by which defects in peroxisomal enzymes might reduce IBA responsiveness. In the direct model, IBR enzymes would catalyze IBA β-oxidation reactions. The specificity of the ibr defects and the nature of the mutated genes are consistent with the possibility that IBR1 and IBR10 act directly in the conversion of IBA to IAA (Figure 8A). Because elimination of two carbons could convert IBA to IAA, we expect that the responsible enzymes will resemble those acting in fatty acid β-oxidation. In plants, fatty acid β-oxidation occurs by addition of a CoA moiety by an acyl-CoA synthetase followed by the repeating activities of ACX, MFP, and thiolase enzymes (Figure 8C). Previously, we suggested that IBR3 might act as an IBA-CoA dehydrogenase or oxidase, carrying out the second step in IBA β-oxidation (Figure 8A, step 2; Zolman et al. 2007). The multifunctional proteins AIM1 (Richmond and Bleecker 1999) and MFP2 (Rylott et al. 2006) are isozymes with different chain-length specificities that carry out steps 3 and 4 of fatty acid β-oxidation, which involve a hydration and subsequent NAD+-dependent dehydrogenation. AIM1 also might act in the conversion of IBA, as the aim1 mutant is resistant to IBA (Zolman et al. 2000) and 2,4-DB (Richmond and Bleecker 1999). However, identification of IBR10, which encodes an enoyl-CoA hydratase/isomerase family member, and IBR1, which encodes a dehydrogenase/reductase-like protein, suggests that IBR10 and IBR1 may together provide AIM1 activity in IBA β-oxidation, either in addition to or instead of AIM1.
The activities of biochemically characterized enzymes similar to IBR1 and IBR10 fit with what would be required for IBA β-oxidation. ibr1 and ibr10, along with ibr3 (Zolman et al. 2007), have specific defects in IBA responses, displaying reduced responses to IBA in root elongation, lateral root initiation assays, and IBA-induced reporter gene expression, but normal responses to other forms of auxin (Figures 1 and 2). Because neither ibr1 nor ibr10 have phenotypes typical of fatty acid β-oxidation defects (sucrose dependence during early development; Figure 3A), the disruptions of these genes are unlikely to affect general peroxisomal function or fatty acid β-oxidation. Genetic evidence suggests that IBA-to-IAA conversion is peroxisomal; therefore, we expect enzymes catalyzing this conversion to be peroxisomally targeted. Both IBR1 (“SRL”) and IBR10 (“PKL”) have predicted C-terminal type 1 peroxisomal-targeting signals (Kamada et al. 2003; Reumann et al. 2004) and recent proteomic analysis of purified peroxisomes and YFP-localization studies confirm that both IBR1 and IBR10 are peroxisomal (Reumann et al. 2007), consistent with a role in IBA metabolism.
Rather than directly catalyzing IBA β-oxidation, IBR1 and IBR10 might act in other peroxisomal processes that, when disrupted, indirectly impede IBA metabolism. An indirect model for IBA β-oxidation disruption is not without precedent. Such a mechanism might result from accumulation of toxic intermediates, as in the chy1 mutant (Zolman et al. 2001a). CHY1 encodes a β-hydroxyisobutyryl-CoA hydrolase acting in peroxisomal valine (Zolman et al. 2001a; Lange et al. 2004) or isobutyrate (Lucas et al. 2007) catabolism. chy1 mutants are sucrose dependent (Zolman et al. 2001a), IBA resistant (Zolman et al. 2001a), 2,4-DB resistant (Lange et al. 2004), and hypersensitive to isobutyrate and propionate (Lucas et al. 2007). These phenotypes are thought to result from accumulation of a toxic valine catabolic intermediate that impedes thiolase activity (Zolman et al. 2001a; Lange et al. 2004). The ibr1, ibr3, and ibr10 mutants respond like wild type to exogenous isobutyrate and propionate (Figure 3B), suggesting that the IBR proteins do not act in the CHY1 pathway.
A second potential indirect mechanism for disruption of IBA β-oxidation is decreased cofactor availability. For example, disrupting the second step of fatty acid β-oxidation (Figure 8) in certain acx mutants leads to accumulation of fatty acyl-CoA esters (Rylott et al. 2003; Pinfield-Wells et al. 2005). These activated, but unprocessed, substrates might deplete the CoA pool available to IBA and thereby limit IBA metabolism to IAA (Adham et al. 2005).
If IBR1 acts directly in IBA β-oxidation, it is curious that the protein is in the short-chain alcohol dehydrogenase/reductase family rather than resembling acyl-CoA dehydrogenases, as might be expected for an enzyme acting at this step of the pathway. However, the SDR family has a broad list of accepted substrates, perhaps indicating a catalytic pocket that could accommodate the IBA β-oxidation intermediate. Previous bioinformatic analysis led to the suggestion that IBR1 might act on β-hydroxyisobutyrate (Reumann et al. 2004); this model is attractive because the potential substrate lacks a CoA side chain. However, it is unclear how a block this late in valine catabolism would impede IBA responsiveness. Unlike the chy1 mutant that is disrupted in valine catabolism, ibr1 mutants are not sucrose dependent (Figure 3A). Moreover, in contrast to chy1, ibr1 does not display increased sensitivity to isobutyrate or propionate (Figure 3B).
Ultimately, testing these competing hypotheses will require biochemical assays to determine the preferred substrates and specific activity of each enzyme. If potential substrates become available, we will be able to determine the nature of the activity of the IBR enzymes on intermediates in IBA metabolism, revealing each as a direct enzymatic player or as an indirect participant acting in a pathway that overlaps or interferes with IBA metabolism in the peroxisome. In addition to determining the enzymatic activity of IBR1, IBR3, and IBR10, enzymes catalyzing other steps of the pathway remain to be identified. The activating enzyme that adds CoA onto IBA (Figure 8, step 1) has not been identified; LACS6 and LACS7 act on fatty acid substrates but appear not to act on IBA (Fulda et al. 2004). In addition, the enzyme acting in the thiolase step (Figure 8, step 5) has not been established. Although PED1 is the best-characterized and most active thiolase, two other thiolases (KAT1 and KAT5) are encoded in the Arabidopsis genome (Germain et al. 2001; Carrie et al. 2007) and could act on IBA intermediates. Furthermore, we expect that an enzyme to remove the CoA moiety (Figure 8, step 6) will be required; several thioesterases with unknown substrates are predicted to reside in the peroxisome (Reumann et al. 2004) and may catalyze this reaction. Finally, IAA acts outside of the peroxisome and therefore a transporter is expected to export the active auxin; whether this transporter removes IAA or IAA-CoA also is an open question.
Although our genetic evidence in Arabidopsis suggests that IBA must be converted to IAA to exert auxin activity, IBA is proposed to promote lateral root formation in rice independently of IAA signaling (Chhun et al. 2003). Interestingly, IBR3 (Zolman et al. 2007), IBR1 (Figure 5), and IBR10 (Figure 7) all have close homologs in rice that are predicted to be peroxisomal; the conservation of these proteins hints that IBA β-oxidation may play a role in auxin metabolism in rice as well. Further examination of these proteins and IBA metabolism in a variety of plants will reveal whether different species have evolved different roles for IBA in auxin metabolism and action.
Appropriate production, transport, and storage of IAA are essential for normal plant development. Our genetic studies implicate the peroxisomal enzymes IBR1 and IBR10 as likely players in the β-oxidation of IBA to free IAA, in conjunction with the previously characterized IBR3. Continued elucidation of the regulation and mechanism of this conversion will provide a more complete understanding of inputs into the auxin pool and the control of auxin homeostasis.
We thank the Arabidopsis Biological Resource Center at Ohio State University for seeds and DNA stocks; the Salk Institute Genomic Analysis Laboratory for generating the sequence-indexed Arabidopsis T-DNA insertion mutants; and Matthew Lingard, Sarah Ratzel, Lucia Strader, and Andrew Woodward for critical comments on the manuscript. This research was supported by the University of Missouri at St. Louis start-up funds, the University of Missouri Research Board, the National Science Foundation (NSF) (IBN-0315596 and MCB-0745122), and the Robert A. Welch Foundation (C-1309). N.M. was supported in part by the Rice-Houston Alliance for Graduate Education and the Professoriate program (NSF HRD-0450363) and the National Institutes of Health (NIH) (F31-GM081911), A.M. was supported in part by an American Society of Plant Biologists' Summer Undergraduate Research Fellowship, and A.R.A. was supported by the NIH (F31-GM066373) and a Houston Livestock Show and Rodeo Fellowship.
↵2 Present address: University of California, San Francisco, CA 94158.
Communicating editor: D. Voytas
- Received April 17, 2008.
- Accepted July 8, 2008.
- Copyright © 2008 by the Genetics Society of America