Degradation of Aux/IAA proteins is essential for normal auxin signalling

Authors


*For correspondence (fax +1 530 752 3085; e-mail jcallis@ucdavis.edu).

Summary

The growth substance auxin mediates many cellular processes, including division, elongation and differentiation. PSIAA6 is a member of the Aux/IAA family of short-lived putative transcriptional regulators that share four conserved domains and whose mRNAs are rapidly induced in the presence of auxin. Here PSIAA6 was shown to serve as a dominant transferable degradation signal when present as a translational fusion with firefly luciferase (LUC), with an in vivo half-life of 13.5 min in transgenic Arabidopsis seedlings. In a transient assay system in tobacco protoplasts using steady-state differences as an indirect measure of protein half-life, LUC fusions with full-length PSIAA6 and IAA1, an Aux/IAA protein from Arabidopsis, resulted in protein accumulations that were 3.5 and 1.0%, respectively, of that with LUC alone. An N-terminal region spanning conserved domain II of PSIAA6 containing amino acids 18–73 was shown to contain the necessary cis-acting element to confer low protein accumulation onto LUC, while a fusion protein with PSIAA6 amino acids 71–179 had only a slight effect. Single amino acid substitutions of PSIAA6 in conserved domain II, equivalent to those found in two alleles of axr3, a gene that encodes Aux/IAA protein IAA17, resulted in a greater than 50-fold increase in protein accumulation. Thus, the same mutations resulting in an altered auxin response phenotype increase Aux/IAA protein accumulation, providing a direct link between these two processes. In support of this model, transgenic plants engineered to over-express IAA17 have an axr3-like phenotype. Together, these data suggest that rapid degradation of Aux/IAA proteins is necessary for a normal auxin response.

Introduction

The plant growth regulator auxin has been shown to be involved in a number of processes such as tropic responses, apical dominance, cambial cell divisions and differentiation of vascular tissues ( Estelle 1992; Guilfoyle 1986; Theologis 1986; Went & Thimann 1937). One of the early responses to auxin is the transcriptional activation of multiple genes. These primary response genes are activated without a requirement for de novo protein synthesis and can be divided into five families based on amino acid identity and auxin induction kinetics: Aux/IAA, SAUR (small auxin-up-regulated), GH3-like, aminocyclopropane-1-carboxylic acid synthase (ACS) and glutathione-S-transferase (GH2/4-like) ( Abel & Theologis 1996).

The Aux/IAA family is present in diverse dicot species, and members are present in multiple organs, indicating broad expression in cells known to be affected by auxin. The proteins encoded by two members of the Aux/IAA family from Pisum sativum, PSIAA6 and PSIAA4, have short half-lives in vivo, 6 and 8 min, respectively ( Abel et al. 1994 ). Interestingly, these degradation rates were the same independent of exogenous auxin application, suggesting that the degradation machinery might be constitutively active ( Abel et al. 1994 ). Their short half-lives ensure that these proteins will not accumulate to high levels, nor will they linger after synthesis has terminated.

Aux/IAA proteins contain predicted secondary structures, a ribbon–helix–helix motif, similar to that found in a class of bacterial transcriptional regulators, implicating Aux/IAA proteins in the transcriptional regulation of downstream genes ( Abel et al. 1994 ). In a yeast two-hybrid assay, Aux/IAA proteins interacted with ARFs, proteins that bind to auxin response DNA elements ( Kim et al. 1997 ; Ulmasov et al. 1997a ). Aux/IAA over-expression in a transient assay negatively regulated expression of auxin-induced genes ( Ulmasov et al. 1997b ). These results suggest that Aux/IAA proteins function in regulating gene expression, but the exact mechanism remains unclear.

Peptide regions of some short-lived proteins have been shown to operate as transferable, dominant degradation signals. Some of the first signals characterized serve as examples to illustrate the diverse proteins that contain degradation signals. Mitotic cyclins contain a conserved nine amino acid sequence called the ‘destruction box’ which is necessary for their cell-cycle-specific degradation ( Glotzer et al. 1991 ). A 54 amino acid peptide from sea urchin cyclin B containing the destruction box is sufficient for targeting a reporter protein for cell-cycle-specific proteolysis ( Glotzer et al. 1991 ). Two different peptide regions sufficient for targeting β-galactosidase for rapid degradation were identified in the yeast transcriptional repressor, MATα2 ( Hochstrasser & Varshavsky 1990). One consists of the N-terminal 67 amino acid residues and targets proteins containing it for degradation via the ubiquitin pathway ( Chen et al. 1993 ). The N-terminal transmembrane domain of mammalian 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMG-R) and of one of the yeast homologues (HMG-R-2) is necessary for feedback-regulated degradation and sufficient for targeting a reporter protein for feedback-regulated degradation ( Chun et al. 1990 ; Hampton & Rine 1994). In response to serum deprivation, the motif KFERQ and similar peptides operate as degradation signals by targeting proteins for import into mammalian lysosomes ( Dice 1990). In contrast, in the case of the short-lived yeast transcription factor Gcn4p, no small region was found to be sufficient for targeting a fusion protein for degradation, suggesting that there may be a tertiary structural component to the degradation signal(s) contained within Gcn4p ( Kornitzer et al. 1994 ). The identified degradation signals do not share significant amino acid identity, meaning that they must be identified by functional tests.

Here, PSIAA6 is shown to be capable of targeting a heterologous protein for rapid degradation. The conserved domains found in Aux/IAA family members provided the framework for deciding where to truncate the PSIAA6 protein to elucidate which peptide regions are necessary and sufficient for targeting a protein for rapid degradation. In a transient assay system, a region of PSIAA6 was found which was sufficient for targeting firefly luciferase (LUC) for low protein accumulation. This property depended on absolutely conserved residues of domain II contained within this region. The importance of regulation of Aux/IAA abundance was revealed through production of transgenic plants designed to over-express the Aux/IAA protein IAA17.

Results

PSIAA6 targets a heterologous protein for rapid degradation in vivo in transgenic plants

To determine whether an amino acid sequence is not only necessary for rapid proteolysis, but also sufficient for targeting a heterologous protein for degradation, it must be present in-frame with a more stable protein and shown to decrease the half-life of the entire fusion protein. The coding region for PSIAA6, a rapidly degraded Aux/IAA protein, was placed in-frame upstream of the coding region for firefly luciferase (LUC) lacking the C-terminal peroxisomal targeting sequence, creating PSIAA6::LUC ( Fig. 1). Firefly luciferase has an accessible mobile N-terminus, suggesting that N-terminal fusions would not interfere with the folding or activity of LUC ( Conti et al. 1996 ). The modified LUC used here is not targeted to a specific organelle, and requires no post-translational modifications for activity. For expression of the PSIAA6::LUC fusion in transgenic plants, the 5′ flanking region of the Arabidopsis UBQ10 polyubiquitin gene, previously shown to direct constitutive levels of marker enzyme activity in transgenic Arabidopsis ( Sun & Callis 1997), was placed upstream of the fusion protein coding region.

Figure 1.

Diagrammatic representations of Aux/IAA::LUC fusion protein expression cassettes used for stable transformation and transient transfections.

Diagrammatic representation of LUC expression cassettes placed in both transient and plant transformation plasmids. The 5′ untranslated region from the A. thaliana UBQ10 gene serves as the promoter in front of either the coding sequences for Aux/IAA::LUC in-frame fusions or in front of LUC alone. Both coding regions are followed by the 3′ untranslated region of the A. tumefaciens nopaline synthase gene ( Bevan et al. 1993) . Both contain the same GUS expression cassette on the same DNA for normalization in the transient assay.

Pulse–chase experiments were performed on seedlings from multiple independent transgenic lines expressing either PSIAA6::LUC or LUC to determine their respective in vivo half-lives. Representative pulse–chase results are shown in Fig. 2. In transgenic LUC-expressing seedlings, the amount of labelled LUC protein after 30 min of chase was not reproducibly different from that at T0, indicating no significant loss of protein during this time period ( Fig. 2, lanes 5–8). A chase time of 2 h also did not result in a detectable loss (data not shown). Assuming that differences greater than 25% could be measured reproducibly, the half-life of unfused LUC is at least 3 h. In contrast, pulse–chase experiments with PSIAA6::LUC- expressing seedlings revealed that significant loss of the PSIAA6::LUC protein had occurred by 30 min ( Fig. 2, lanes 1–4). Quantification of the pulse–chase results from two independent lines indicated that an average of 21% of the PSIAA6::LUC protein present at T0 was present at 30 min, corresponding to a half-life of 13.5 min. This is comparable to the 6 min half-life determined for PSIAA6 in pea seedlings ( Abel et al. 1994 ). This indicated that PSIAA6 was capable of acting as a transferable degradation signal.

Figure 2.

Pulse–chase analysis of LUC and full-length PSIAA6::LUC proteins in transgenic Arabidopsis seedlings.

Pulse–chase was performed as described in Experimental procedures. Proteins were immunoprecipitated from extracts containing the indicated amount of TCA precipitable counts per minute: lanes 1 and 2, 57 × 106 cpm; lanes 3 and 4, 32 × 106 cpm; lanes 5 and 6, 29 × 106 cpm; lanes 7 and 8, 42 × 106 cpm; lane 9, 33 × 106 cpm. The molecular weight markers indicate the migration of standard 14C-labelled proteins whose masses are shown in kDa. Lanes 1, 3, 5, 7 and 9 contain the LUC immunoprecipitates from immediately after the pulse of 35S-methionine. Lanes 2, 4, 6 and 8 represent the LUC immunoprecipitates after 30 min of chase. The transgenic line is indicated above the lanes, representing two independent lines for each transgene. ‘Neg. Control’ is the non-transformed ecotype, No-0, grown and labelled under identical conditions. The migrations of PSIAA6::LUC and LUC proteins are indicated by closed and open triangles, respectively. Quantification of T30 lanes relative to their respective T0 lanes are: lane 2, 19%; lane 4, 21%; lane 6, 94%; lane 8, 99%. Lanes 1 and 2 were exposed for 24 h in order to clearly visualize the PSIAA6::LUC protein, lanes 3–6 were exposed for 6 h. The asterisk denotes a non-specific LUC immunoprecipitating band that is present in all lanes at about equivalent levels at equal exposure times, including in the non-transformed control lane (data not shown).

Transient assays demonstrate that fusion proteins containing PSIAA6 and IAA1 accumulate to much lower levels than LUC alone

Because PSIAA6 increased the rate of LUC degradation when fused to the LUC N-terminus, the use of a transient assay system was desired to assess rapidly the in vivo effects of alterations of Aux/IAA protein amino acids on fusion protein accumulation. In transient assays, the ratio between steady-state levels of two proteins is equivalent to the ratio of their half-lives if synthetic rates are identical ( Berlin & Schimke 1965). Cultured tobacco cells are a convenient transient assay system with efficient protoplast formation for transfection and high protoplast viability. Especially relevant for these experiments is that these cells require exogenous auxin for growth, indicating that an auxin signal transduction pathway is functional. In addition, protoplasts from a number of species show auxin-mediated responses ( Koshiba et al. 1995 ; Ulmasov et al. 1997b ). Previous work from our laboratory using protoplasts isolated from cultured cells in transient assays established that LUC accumulation reached steady state by 24 h post-transfection, and that this level of activity was maintained for at least another 48 h ( Worley et al. 1998 ). For all transient transfections, the same expression units used for stable transformation were introduced; the second expression unit encoding β-glucuronidase (GUS) was used as a control for differences between samples in transfection, viability and lysis efficiency ( Fig. 1).

The relative normalized activity of the full-length PSIAA6::LUC fusion protein in extracts after DNA transfection and incubation of the protoplasts to achieve steady-state enzyme activity is shown in Fig. 3. Steady-state LUC activity for PSIAA6::LUC fusion was 56-fold lower than the activity obtained for LUC alone, consistent with the short half-life of PSIAA6::LUC determined in vivo ( Fig. 2). Because differences in activity could be due to misfolding or to a reduction in catalytic efficiency itself, rather than differences in protein content, Western blot analysis was attempted to measure the protein levels directly. LUC protein was not visualized from protein extracts of transfected tobacco protoplasts using chemiluminescent detection, demonstrating that LUC protein concentration was below the detection limit for this method (data not shown).

Figure 3.

Aux/IAA protein conserved features, and LUC activity and protein accumulation of Aux/IAA::LUC fusion proteins after transient transfection of expression constructs into protoplasts.

Diagrammatic representation of Aux/IAA proteins Pisum sativum PSIAA6 and Arabidopsis IAA1 as fusions with luciferase with conserved Aux/IAA features noted. The thick lines represent the approximate location and length of the four conserved domains (I–IV) found in all Aux/IAA proteins. Patterned regions include sequences shown to function as nuclear localization signals (NLS) with identity to either the SV40-like NLS or the bipartite (BP) nucleoplasmin-like NLS ( Abel & Theologis 1995), as well as a region with biased amino acid composition (acidic). The region containing the secondary structural elements, a β-sheet followed by two α-helices, with the first two elements constituting domain III, is indicated. Activities for the fusion proteins after transfection of the corresponding DNA constructs were calculated as normalized percentage activity ± standard deviation (SD) relative to the mean activity value for normalized LUC alone (see Experimental procedures). Relative specific activities were determined by quantitative Western blot analyses after expression of the same proteins in yeast (see Experimental procedures), and activity levels were divided by the specific activities to yield the relative protein accumulation. The Aux/IAA protein is drawn to scale (as indicated); the LUC portion is not drawn to scale.

To determine whether there were changes in LUC specific activity (activity per mole) as a result of addition of PSIAA6 to the N-terminus that may contribute to the observed differences in enzymatic activity, LUC and PSIAA6::LUC proteins were expressed in yeast. LUC protein levels in these cells were sufficient to allow immunological visualization. Enzymatic activity was determined in extracts, and LUC protein content was determined by Western blot analysis in samples containing equal LUC activity (data not shown). The relative specific activity of PSIAA6::LUC to LUC was 52%. Using this value, the PSIAA6::LUC activity measured in protoplasts translated into 3.5% of the accumulation of LUC alone, or a 29-fold reduction in protein level ( Fig. 3).

To determine whether this property of conferring lowered protein accumulation is specific for PSIAA6 or characteristic of other Aux/IAA proteins, a plasmid for the expression of an additional Aux/IAA LUC fusion protein was constructed. IAA1 is an Aux/IAA protein from Arabidopsis that represents a separate subgroup distinct from the clade containing PSIAA6 and its Arabidopsis orthologues ( Abel et al. 1995 ). The coding region for full-length IAA1 was placed upstream of LUC, and protein accumulation was determined in the transient assay as described above. IAA1(1–168)::LUC accumulated to 1.0% of the level of LUC alone, not statistically different from that observed for the PSIAA6(1–179)::LUC fusion protein ( Fig. 3). This result suggests that rapid degradation may be a property of the Aux/IAA proteins.

Deletions of Aux/IAA proteins define a region necessary and sufficient for low protein accumulation

A series of C-terminal deletions of the coding region of PSIAA6 were made, each singly deleting one of the features of PSIAA6 that is shared among the Aux/IAA proteins ( Abel et al. 1994 ). DNAs containing PSIAA6 deletions were fused in-frame with LUC coding region ( Fig. 4), placed in the same plant expression cassette (see above), and introduced into plant protoplasts. Protoplast extracts were analysed for enzymatic activity as described for the full-length PSIAA6::LUC fusion, specific activity measurements were performed in an analogous manner, and the data are shown in Fig. 4.

Figure 4.

LUC activity and relative protein accumulation after transient transfection of constructs expressing C-terminal deletions of PSIAA6::LUC fusions into protoplasts.

Diagrammatic representations of portions of PSIAA6 retained in LUC fusion constructs are shown with designations as in Fig. 3. The thin lines represent the regions missing in the corresponding fusion. The construct names indicate the amino acids remaining in the fusion. The expression constructs designated NLS contain the nuclear localization signal from the squash leaf curl virus BR1 movement protein ( Sanderfoot et al. 1996 ) at their C-termini. Constructs expressing fusions with PSIAA6 amino acids 18–73 have a methionine codon at the 5′ end of the coding region to serve as a translational start site followed immediately by PSIAA6 codons 18–73. LUC activities and protein accumulation after transfection were determined as described in Fig. 3.

PSIAA6(1–150)::LUC was missing the C-terminal SV40-like NLS, and PSIAA6(1–122)::LUC was missing both the NLS and an adjacent acidic region. Together, region 122–179 constitutes domain IV which is conserved among Aux/IAA proteins ( Abel et al. 1994 ). Both proteins were present at steady state at 20–30-fold lower levels than LUC alone in the transient assay ( Fig. 4). Fusion proteins PSIAA6(1–103)::LUC, PSIAA6(1–85)::LUC and PSIAA6(1–73)::LUC represent sequential deletions of each putative secondary structural element of the βαα motif, respectively. Of these, the β-sheet and N-terminal α-helix constitute Aux/IAA conserved domain III ( Abel et al. 1994 ). None of these three proteins accumulated to levels significantly different from that of the full-length protein ( Fig. 4).

The smallest fusion protein tested, PSIAA6(1–22)::LUC retained the conserved domain I, but was missing a conserved KR sequence and the amino acids which make up the Aux/IAA conserved domain II ( Abel et al. 1994 ). Accumulation of this protein was significantly different from the full-length protein, but not significantly different than LUC alone ( Fig. 4). Because domain I was not sufficient to operate as a degradation signal, it was deleted to determine whether it was necessary for the degradation of a fusion protein. The resulting fusion protein contained amino acids 18–73 from PSIAA6, and hence retained a conserved KR sequence, domain II, and the intervening non-conserved amino acids. PSIAA6(18–73)::LUC protein accumulated to levels which were not significantly different compared to the full-length PSIAA6 fusion protein ( Fig. 4).

These data suggested that the smallest peptide region of PSIAA6 sufficient for targeting LUC for degradation was amino acids 18–73. However, one possibility for high protein accumulation with PSIAA6(1–22)::LUC was that this fusion protein was not localized in the same subcellular compartment as the others; all other constructs had retained functional nuclear localization signals and are predicted to localize to the nucleus ( Abel & Theologis 1995). To test this hypothesis, codons for a demonstrated transferable nuclear localization signal (NLS) from the squash leaf curl virus BR1 protein ( Sanderfoot et al. 1996 ) were placed at the C-terminus of several LUC coding regions. LUC with the added transferable NLS (LUC::NLS) accumulated to levels not significantly different than LUC alone, indicating that the added NLS by itself did not affect the accumulation of LUC ( Fig. 4). When the same NLS was fused to the C-terminus of the PSIAA6(1–22)::LUC, this protein accumulated to an even higher level than the same fusion without the NLS. Thus, it appears that the N-terminal 1–22 amino acids are not sufficient for directing reduced protein accumulation. In contrast, addition of the NLS to the PSIAA6(18–73)::LUC fusion protein resulted in accumulation to a low level, comparable to that of the LUC fusions containing full-length PSIAA6 or the N-terminal 1–73 amino acids. Thus, the minimal degradation signal identified in this work consists of PSIAA6 amino acids 18–73.

PSIAA6 does not appear to contain multiple equivalent degradation signals

Rapidly degraded proteins may contain multiple redundant degradation signals ( Hochstrasser & Varshavsky 1990). To test whether the region C-terminal to amino acids 1–73 of PSIAA6 are capable of affecting LUC accumulation, codons for PSIAA6 amino acids 71–179 were placed in-frame upstream of LUC::NLS (to ensure proper intracellular localization) and analysed for protein accumulation in transient assays as described above. These PSIAA6 amino acids are C-terminal to the degradation signal determined above and contain the rest of the protein. PSIAA6(71–179) accumulated to 33% of LUC alone ( Fig. 5a). While significantly different from LUC::NLS accumulation (by Student's t test), this threefold difference is small compared to the fold differences observed between LUC::NLS and the full-length PSIAA6::LUC::NLS fusion ( Fig. 5a) and between LUC and PSIAA6(18–73)::LUC ( Fig. 4). This indicates that amino acids 71–179 are not sufficient for targeting a fusion protein for the very rapid proteolysis observed with the full-length PSIAA6::LUC fusion or fusions containing PSIAA6 amino acids 18–73.

Figure 5.

LUC activity and relative protein accumulation after transient transfection of constructs expressing altered PSIAA6::LUC fusions into protoplasts.

(a) LUC activity and protein accumulation for the PSIAA6::LUC fusion containing amino acids 1–3 and 71–179 after transient transfection of its expression construct into protoplasts was determined relative to LUC::NLS as described for Fig. 3.

(b) The amino acid sequence of domain II of wild-type Aux/IAA proteins is designated above the diagrammatic representation of full-length PSIAA6::LUC::NLS fusion protein. The absolutely conserved residues are shown in capital letters, residues found in more than half of the Aux/IAA proteins are indicated in lower case, and underlined amino acids indicate two additional amino acids present in PSIAA6 that are not present in all Aux/IAA proteins ( Abel et al. 1995 ). The amino acids mutated in axr3-1 and axr3-3 IAA17 alleles are equivalent to P61L and V62G, respectively, in PSIAA6, and the corresponding wild-type amino acids are designated in reverse type ( Rouse et al. 1998 ). LUC activities and protein accumulation after transfection of the designated constructs were determined as described in Fig. 3.

The putative degradation signal requires conserved amino acids in domain II

Genetic approaches in Arabidopsis to identify genes involved in auxin responses have identified at least two different Aux/IAA proteins. The product of the AXR3 gene is IAA17 and the product of the SHY2 locus is IAA3 ( Rouse et al. 1998 ; Tian & Reed 1999). The axr3-1, axr-3-3, shy2-2 and shy2-3 mutations are semi-dominant and elicit an enhanced auxin response, consistent with either gain of function or a dominant negative effect. The mutations in these axr3 and shy2 proteins map within domain II which is included in the region identified as important in protein accumulation (above). To test directly whether these mutations result in an alteration in protein accumulation, the same mutations found in axr3-1 and axr3-3 were created in the PSIAA6(1–179)::LUC::NLS fusion protein (producing PSIAA6P61L(1–179)::LUC::NLS and PSIAA6V62G (1–179)::LUC::NLS, respectively) ( Fig. 5b) and the accumulation of these proteins determined in the transient assay ( Fig. 5b).

The accumulation of both PSIAA6P61L(1–179)::LUC::NLS and PSIAA6V62G(1–179)::LUC::NLS was significantly higher than that observed for wild-type PSIAA6(1–179)::LUC::NLS, and not significantly different from each other. These single amino acid mutations resulted in slightly greater than 50-fold increases in protein accumulation over the non-mutated fusion protein, indicating that the wild-type amino acid residues are necessary for the most efficient recognition of the degradation signal. As expected, the mutant fusion protein accumulation relative to LUC::NLS was comparable to that of the fusion protein with the deleted N-terminal degradation signal (compare Fig. 5a,b). Thus, two of the residues of Aux/IAA protein domain II are required for function of the major signal for reduced protein accumulation in these transient assays.

Transgenic plants over-expressing IAA17 show effects similar to those observed in axr3 plants

To test further the hypothesis that increases in IAA17 protein cause the phenotypes observed in axr3 plants, wild-type Arabidopsis plants (ecotype Co-1) were transformed with a construct in which expression of the IAA17 cDNA was driven by the cauliflower mosaic virus 35S promoter. Multiple independent transgenic lines were recovered. All showed a similar range of phenotypes, reminiscent of the axr3 gain-of-function phenotypes ( Leyser et al. 1996 ). These include increased adventitious rooting ( Fig. 6, top panel), short roots ( Fig. 6, bottom panel), reduced gravitropism, and reduced root hair development (data not shown) relative to the untransformed control. In addition, transgenic tobacco plants designed to over-express IAA17 also exhibited gain-of-function axr3-like phenotypes (data not shown).

Figure 6.

Transgenic Arabidopsis plants engineered for sense expression of IAA17 mimic the axr3 phenotype.

T3 homozygous lines derived from five independent T1 plants transformed with the IAA17 cDNA under control of the 35S promoter were analysed for number of adventitious roots (top) and for mean root length (bottom) after 5 days of growth on agar. Each sample contained a minimum of 15 plants. The error bars represent the standard error of the mean.

Discussion

Two different assays were used to assess the ability of PSIAA6 to target a heterologous protein for degradation. Direct pulse–chase experiments demonstrated that PSIAA6 targets a LUC fusion protein for rapid degradation with a half-life of 10–15 min. While the half-life of LUC in plants could not be reliably determined in this study, it is at least 3 h, which is likely to be an under-estimate. LUC expressed in cultured mammalian cells was determined to have a 3 h half-life ( Thompson et al. 1991 ). Using 3 h as an approximate LUC half-life in plants, the calculated fold steady-state difference between LUC and PSIAA6::LUC would be 13.3. The second assay utilized steady-state protein accumulation differences in transient transfections as an indirect measure of protein half-life ( Berlin & Schimke 1965). This approach gave a similar result, with PSIAA6::LUC accumulating to much lower levels than LUC alone; the magnitude of the difference was 29-fold. The approximate twofold difference between the two methods is not large given the estimation of LUC half-life and the difficulty of accurate half-life determinations in plants.

This concordance between the two methods supports the validity of the transient assays as an indirect estimate despite the necessary assumption of equal protein synthetic rates between constructs. Equal synthetic rates are likely because fusion constructs were engineered to have identical 5′ and 3′ untranslated sequences and share the same LUC coding region. Other potential differences, such as differences between samples in transfection and lysis efficiency, viability of the transfected cells, and global changes in transcription and/or translation were normalized by inclusion of a second expression cassette encoding a different reporter protein. It is possible that addition of the PSIAA6 coding region could affect the stability of the mRNA, which can influence the rate of protein synthesis. However, PSIAA6 mRNA was reported to have a half-life of 75 min ( Koshiba et al. 1995 ), which is much longer than the protein half-life, and only slightly shorter than the half-life for the average mRNA in plants ( Abler & Green 1996). This indicates that alterations in mRNA stability are not likely to account for the large decrease in protein accumulation observed. Therefore, accelerated protein degradation is likely to be the largest contributor to the reduction in protein accumulation. While rapid degradation of the full-length fusion was determined directly in a pulse–chase analysis, a similar study would need to be performed to verify a reduced half-life for the LUC fusion proteins containing portions of PSIAA6.

The ability of multiple regions of PSIAA6 to direct lowered protein accumulation was investigated. The existence of multiple degradation signals is not without precedent; yeast Matα2 contains at least three distinct degradation signals ( Chen et al. 1993 ; Hochstrasser & Varshavsky 1990). However, for PSIAA6, the N-terminal region was the major region identified. When removed, the remainder of PSIAA6 had a small, but significant effect on protein accumulation. Pulse–chase experiments in transgenic plants will verify whether this difference can be detected as a protein half-life difference.

The fold difference in accumulation between PSIAA6:: LUC::NLS and LUC::NLS was approximately 8–10-fold larger than that observed for the equivalent pair without their NLSs. The reasons for this are not completely understood, but one possibility is that the additional NLS results in more efficient retention in the nucleus and this alteration in intracellular location alters the half-life of proteins containing NLSs. An alternative explanation is that the C-terminal NLS acts to specifically alter the degradation rates of Aux/IAA fusion proteins. Knowing the effect of this NLS on intracellular location of LUC proteins would help to resolve this issue. However, experiments to assess the intracellular partitioning of LUC proteins are hampered by the low abundance of these proteins.

The smallest peptide region of PSIAA6 which was sufficient to result in decreased levels equivalent to the full-length protein contained amino acids 18–73. Within this region is conserved domain II found in all Aux/IAA family members, as well as a basic region containing a conserved KR dipeptide which is thought to serve as part of a nuclear localization signal. A BLAST search ( Altschul et al. 1990 ) using amino acids 1–73 as a query identified many other family members from pea, Arabidopsis, mung bean, soybean and tomato, but no other proteins including other auxin primary response proteins were identified. This sequence then probably represents a signal specific for the Aux/IAA family. Delineation of the exact sequence necessary and sufficient is in progress.

In Arabidopsis, mutations in the absolutely conserved residues of domain II are semi-dominant and result in auxin-related phenotypes ( Rouse et al. 1998 ; Tian & Reed 1999). Two of these same mutations were introduced into PSIAA6::LUC fusion proteins and accumulation of these proteins tested in the transient assay. Both single amino acid changes tested increased protein accumulation dramatically. The increased accumulation of these mutant proteins probably represents increased stability, and leads us to propose that increases in Aux/IAA protein levels are responsible for the phenotypic effects seen in axr3 and shy2 plants. In addition, the semi-dominant behaviour of the axr3-1 and axr3-3 mutants is consistent with our model that the concentration of Aux/IAA proteins is critical for a proper auxin response. The ability to alter protein levels of the endogenous Aux/IAA protein provides a test of this model. Transgenic plants designed to over-express IAA17 mimic the mutant phenotype as the model predicts.

While naming it a degradation signal, this region could be required for post-translational modification prior to proteolysis. In that case, this region would not be required for interaction with the degradation machinery directly, but would be required for recognition and/or interaction with modification enzymes that provide either the entire recognition signal or a part. Ubiquitination of substrate proteins serves as a recognition element for degradation by the major ATP-dependent protease in the cell, the 26S proteasome ( Ciechanover & Schwartz 1998). The Aux/IAA domain II region could serve as the recognition sequence for ubiquitination. In addition, modifications prior to ubiquitination are required for degradation of multiple substrates of the ubiquitin pathway. Substrates that require the ubiquitin E3 ligase, SCF, must be phosphorylated for wild-type degradation rates ( Patton et al. 1998 ). It is the phosphopeptide region that serves as the recognition element for binding to the ubiquitination enzymes. Thus, the degradation signal of Aux/IAA proteins may include sites for multiple post-translational modifications and/or ubiquitination. The role of modification of Aux/IAA proteins in degradation is currently under investigation.

mRNAs for most of the Aux/IAA proteins increase in abundance within 30 min of exogenous auxin application. Cycloheximide treatment alone of the same tissues results in increases in Aux/IAA mRNA. This suggested that Aux/IAA proteins are early transducers of the auxin signal, and may be responsible for the auxin-induced alterations in gene expression previously observed. The short-half lives of these proteins ensure that after cessation of protein synthesis, Aux/IAA abundance is rapidly reduced. The data reported here suggest that the concentration of Aux/IAA proteins is critical for a proper auxin response, and modest changes in abundance have profound consequences on growth, development and responses of plants to their environment.

Experimental procedures

Molecular techniques

Standard molecular protocols were used for cloning ( Sambrook et al. 1989 ). Expression cassettes containing both the coding region for the experimental LUC fusion proteins and the coding region for an unmodified GUS for normalization were cloned into a single plasmid. The plant expression cassette for GUS was derived from p35SGUS ( Norris et al. 1993 ). The promoter for the experimental LUC fusion proteins was the 5′ untranslated region of the UBQ10 gene from A. thaliana which is constitutively expressed ( Norris et al. 1993 ; Sun & Callis 1997). The LUC coding region, modified with a KpnI site just upstream of the ATG, was derived from pSP-LUC+ (Promega) which encodes a luciferase enzyme with the peroxisomal targeting sequence and potential glycosylation sites removed.

PCR of the PSIAA6 coding region of PSIAA6::GUS or portions thereof ( Abel et al. 1994 ), or of AtIAA1 cDNA ( Abel et al. 1995 ), was used to amplify the desired portion, place a KpnI site at the 5′ end just upstream of the start codon, and add an NcoI site at the 3′ end. For the C-terminal PSIAA6 portion, the first three amino acids were included N-terminal to amino acids 71–179 using PCR. Overlapping PCR was used to create the single amino acid substitutions. After sequence verification, they were placed into the same expression vector as wild-type. Plasmids for yeast expression and in planta transformation were constructed by ligating the coding region of interest into pYES2 (Invitrogen) or into a pBIN19-derived plasmid ( Bevan 1984) containing the UBQ10 promoter, respectively.

To add a nuclear localization signal to the C-terminus of LUC, PCR mutagenesis was used to amplify the LUC coding region (pSP-LUC+ template), deleting the stop codon and adding an XbaI and a PstI site to the 3′ end of the PCR product. The PCR product was ligated into a plasmid containing a downstream BamHI site. Two complimentary oligos encoding the proven transferable C-terminal nuclear localization from the squash leaf curl virus movement protein BR1 ( Sanderfoot et al. 1996 ) were ligated into the XbaI/BamHI overhangs: oligo 1: 5′-CTAGATCTTACGTTA-AGACTGTTCCAAACAGAACTAGAACTTACATCAAGTTCTGAG-3′ oligo 2: 5′-GATCCTCAGAACTTGATGTAAGTTCTAGTTCTGTTTG-GAACAGTCTTAACGTAAGAT-3′. Loss of the PstI site verified the new clones, and they were sequenced from the 3′ end through an internal AvaI site in the LUC coding region. The AvaI/BamHI fragment from this modified LUC was substituted for LUC in the transient expression plasmid.

Transient expression, protein extracts and enzyme activity assays

Plasmids used for transfection into protoplasts were harvested by alkaline lysis of large-scale preparations and purified by polyethylene glycol precipitation ( Sambrook et al. 1989 ). A 100 μg aliquot of a single plasmid encoding both the experimental protein as well as a second reporter for normalization was introduced into protoplasts derived from NT1 tobacco cells using PEG-mediated DNA transfer ( Altman et al. 1992 ). Protoplasts were incubated for 24–48 h; during this time period, LUC levels were at steady-state ( Worley et al. 1998 ). Cells were lysed by sonication in 300 μl extraction buffer (100 m m potassium phosphate pH 7.8, 1 m m EDTA, 7 m mβME, 1 m m PMSF and 20 μg/ml each of antipain, aprotinin, chymostatin, leupeptin and pepstatin A), spun at 10 000 g at 4°C for 10 min, and supernatants used for assays.

Duplicate LUC activity assays were performed on each sample as described ( Norris et al. 1993 ). Duplicate luminescent GUS activity assays were performed with the GUSLIGHT kit (Tropix) according to the manufacturer's instructions. A significant difference between samples described in the text refers to > 95% probability based on Student's t test. Determination of the activity for each protein after transient assay represents the mean and standard deviation of at least three independent experiments with typically three replicas per experiment. The total number of replicas was typically nine, ranging from 7 to 19.

Yeast expression and Western blot analysis

Plasmids encoding the desired fusion proteins were transformed into S. cerevisiae strain WCG4α ( Richter-Ruoff et al. 1992 ) using the lithium acetate method ( Ito et al. 1983 ) with URA3 selection. After galactose induction, cells were lysed in 300 μl LUC extraction buffer (see above) by glass bead agitation. Extracts were spun at 10 000 g at 4°C for 10 min, and A280 of the supernatant was measured to determine relative total protein content ( Bollage & Edelstein 1991). Protein concentration was verified by SDS–PAGE and Coomassie blue staining. Supernatants were diluted with extraction buffer to equal A280, and dilutions were assayed for LUC activity. Linearity of the assay was verified with serial dilutions of the supernatant.

Yeast extracts containing equal reporter activity and equal total protein were prepared for SDS–PAGE and Western blotting ( Beers et al. 1992 ). Equal total protein between extracts was achieved by addition of negative control yeast extract to samples with lower protein content. Anti-LUC (Cortex Biochem) polyclonal antibodies were used to visualize their respective immunogens by chemiluminescence ( Gallaher et al. 1994 ). The signal was quantified by densitometry, and the dilution series included on the same gel verified linearity of the signal. Each construct was expressed in two independently transformed yeast strains for specific activity determination. Two separate blots were performed for each construct and their densitometry values averaged.

In planta transformation

For expression of PSIAA6::LUC fusion proteins, A. thaliana ecotype No-0 plants were transformed by the in planta method ( Bechtold et al. 1993 ). Three homozygous transgenic lines expressing PSIAA6::LUC and two homozygous lines expressing LUC segregating 3:1 for kanamycin resistance were analysed. For over-expression of IAA17, the coding region of IAA17 was placed downstream of the CaMV 35S promoter in a pBIN19-derived plasmid ( Bevan 1984) and transformed into Arabidopsis ecotype Co-1 as above. Five independent lines, for which segregation analysis in T2 indicated a single site of transgene insertion, were selected for further analysis. Phenotypic measurements were performed as described previously ( Leyser et al. 1996 ).

Pulse–chase analysis

Continuous light-grown transgenic Arabidopsis seedlings expressing LUC and PSIAA6::LUC were used for pulse–chase experiments 5–7 days after imbibition. Seedlings were incubated with 700 μCi of 35S-labelled methionine and cysteine (NEN EASYTAG) for 2 h, and chased with 1 m m cysteine, 1 m m methionine and 200 μg/ml cycloheximide in H2O. At each time point, seedlings were ground in ice-cold NP-40 extraction buffer: 150 m m NaCl, 1.0% nonidet P-40, 50 m m Tris (pH 8) and 1 m m EDTA ( Harlow & Lane 1988) containing a protease inhibitor cocktail (final concentration: 1 m m phenylmethylsulphonyl fluoride (PMSF) and 20 μg ml−1 each of antipain, aprotinin, chymostatin, leupeptin and pepstatin A).

LUC and PSIAA6::LUC were immunoprecipitated from aliquots of the cleared supernatants containing equal TCA precipitable counts for T0 and T30 samples using 10 μl anti-LUC antibodies (Cortex Biochem) and 50 μl PANSORBIN cells (Calbiochem) following standard protocols ( Harlow & Lane 1988). The precipitates were separated by SDS–PAGE in 7% polyacrylamide gels, exposed to phosphorimager plates, and bands quantified (MacBas program). Pulse–chase analysis was performed twice with two different transgenic lines for each transgene.

Acknowledgements

We appreciated the assistance of Brian McDougle in constructing deletions and the expert technical assistance of Barbara McArdle. This work was supported by National Science Foundation grant (IBN 98-08791) to J.C., by UCD Jastro-Shields fellowships to C.K.W., N.Z. and J.R., and by the Biotechnology and Biological Sciences Research Council of the UK to O.L. We thank other members of the Callis lab for careful reading of the manuscript and helpful discussions.

Ancillary