Regulation and processing of a plant peptide hormone, AtRALF23, in Arabidopsis


*(fax 1 515 294 5256, e-mail


Arabidopsis has 34 genes encoding proteins related to rapid alkalinization factor (RALF), a peptide growth factor. One of those genes (AtRALF23) is significantly downregulated by brassinolide (BL) treatment of Arabidopsis seedlings or in mutant seedlings expressing a constitutively active form of BES1, a transcriptional effector of the brassinosteroid signaling pathway. Overexpression of AtRALF23 impairs BL-induced hypocotyl elongation in seedlings, and mature overexpressing plants are shorter and bushier. Overexpression of AtRALF23 produces slower growing seedlings, with roots that have reduced capacity to acidify the rhizosphere. AtRALF23 encodes a 138-aa protein, and when an epitope-tagged form (AtRALF23-myc) was expressed in transgenic plants, the protein was processed to release a C-terminal peptide. The presumed junction between the precursor and the processed peptide contains a recognition site for site-1 protease (AtS1P), a plant subtilisin-like serine protease (subtilase). When AtRALF23-myc was expressed in the background of a site-1 protease mutant (s1p-3), or when the AtS1P recognition site (RRIL) was mutated (RR → GG) and expressed in a wild-type background, the precursor was not cleaved, and the bushy phenotype was not produced. A fluorogenic peptide representing the presumed subtilase recognition site in AtRALF23 was cleaved in vitro by AtS1P. Thus, BL downregulates AtRALF23 expression, presumably relieving the growth-retarding effect of a peptide growth factor, which is processed from a larger precursor protein by AtS1P.


Peptides are well known growth factors in animal systems, but they have only recently been recognized for their widespread role in cell-to-cell signaling in plants (Ryan et al., 2002). In a recent review, Matsubyashi and Sakagami (2006) stated that peptide signals are involved in ‘various aspects of plant growth regulation including defense responses, callus growth, meristem organization, self incompatibility, root growth, leaf-shape regulation, nodule development and organ abscission’.

The first signaling peptide discovered in plants was systemin. Pearce et al. (1991) isolated from tomato leaves, an endogenous 18-aa peptide called TomSys, which was able to induce the synthesis of proteinase inhibitor proteins. TomSys was derived from the C-terminus of a 200-aa precursor called prosystemin (McGurl et al., 1992; McGurl and Ryan, 1992). It was found that TomSys binds with high affinity to a tomato membrane binding protein, SR160 (Meindl et al., 1998), which was identified as an LRR receptor-like kinase (Scheer and Ryan, 2002) corresponding to the tomato brassinosteroid (BR) receptor tBRI1 (Montoya et al., 2002). Two other unrelated 18-aa peptides called TobHypSys I and II (or tobacco systemins I and II), derived from the same 165-aa precursor, were identified and shown to induce proteinase inhibitor production in tobacco, and rapid alkalization in tobacco suspension culture cells (Pearce et al., 2001a).

Using the cell culture assay, Pearce et al. (2001b) identified another peptide from tobacco that was even more potent in inducing alkalization. The N terminus of the peptide was sequenced and determined to be +NH3-ATKKYISYGALQKNSVP−. A cDNA corresponding to the peptide sequence was identified, and it was found that the peptide was derived from the C terminus of a predicted 115-aa precursor. The precursor bore a signal peptide indicating that it was likely to be a secreted protein. Pearce et al. (2001b) also pointed out an RR motif two residues upstream from the N terminus of the mature peptide, and suggested that the peptide was derived through proteolysis by an enzyme with dibasic substrate specificity. Another characteristic of the peptide was that it had two disulfide bridges between cysteine 18 and 28, and between cysteine 41 and 47, that were required for activity.

Pearce et al. (2001b) called the peptide rapid alkalinization factor (RALF), and found related sequences in other plant species. They isolated RALF-like peptides from tomato and alfalfa, and reported that the purified peptides were able to induce alkalization in the corresponding plant system. A synthetic form of the tomato RALF was chemically produced, and, in oxidized form, was active in the inhibition of root growth in tomato and Arabidopsis seedlings.

Olsen et al. (2002) identified a family of 34 RALF-like genes in Arabidopsis, and found that different RALF genes were expressed in different tissues. This study focuses on one of the Arabidopsis RALFs, AtRALF23, which is downregulated in Arabidopsis in response to treatment with brassinolide (BL) (Nemhauser et al., 2004). This is of interest because AtRALFs appear to counteract BR growth-promoting effects. BRs enhance growth and cell elongation (Clouse and Sasse, 1998; Mandava, 1988), and BR-deficient or BR-insensitive mutants display dwarfism (Clouse, 1996; Li and Chory, 1999; Mussig, 2005). We show here that AtRALF23 overexpression reduces plant growth, and mature plants are dwarfed and bushy. Our findings suggest that AtRALF23 retards plant growth, and that some of BR’s growth-promoting effects may be relieved by the downregulation of AtRALF23. We also show that AtRALF23 is synthesized as a larger precursor protein, and is released in peptide form by the action of AtS1P, a subtilisin serine protease. In the absence of functional AtS1P, AtRALF23 is not processed, and does not produce its overexpression phenotype.


Brassinolide treatment and AtRALF23 overexpression

In an earlier transcript-profiling study, Nemhauser et al. (2004) demonstrated that AtRALF23 was among the genes significantly downregulated by the treatment of Arabidopsis seedlings with BL. We confirmed this observation by treating seedlings with 1 μm BL for 3 h, and examined AtRALF23 expression using qRT-PCR. In keeping with the previous study, we found that AtRALF23 RNA levels in the wild type were reduced by about half compared with mock-treated samples (Figure 1a). We also examined the levels of AtRALF23 expression in a bes1-D mutant. bes1-D is a dominant mutation that constitutively activates a major transcription factor in the BR signaling pathway (Yin et al., 2005, 2002). AtRALF23 transcript levels were reduced about 30% in 2-week-old mock-treated seedlings, compared with wild type, when treated in the same way (Figure 1a). Treatment of the bes1-D seedlings with 1 μm BL reduced the AtRALF23 transcript levels further. These findings indicated that AtRALF23 gene expression is downregulated through the operation of the BR signaling pathway. In very young (3-day-old) seedlings, AtRALF23 expression was localized to the tips of the cotyledons, as indicated by the expression of AtRALF23 promoter:GUS reporter constructs (Figure S1a). In 7-day-old seedlings, the expression moved basipetally, and was expressed strongly in the cotyledon vasculature and petiole (Figure S1b). By 14 days, expression was most intense in the vasculature and petioles of the cotyledons and leaves. However, as leaves emerged they showed patterns of expression similar to the cotyledons, with expression first at the tips of the emerging leaves (Figure S1c). When 7-day-old seedlings were grown in 50 nm BL, the seedlings clearly showed a lower level of GUS expression than those grown in the absence of BL (Figure 1b, c). These findings indicate that the BL downregulation of AtRALF23 expression is, in part, transcriptional.

Figure 1.

 Brassinosteroids (BRs) repress the expression of AtRALF23 (At3g16570).
(a) Two-week-old Col-0 (WT) or bes1-D mutant seedlings were treated without or with 1 μm brassinolide (BL) for 3 h. The expression of AtRALF23 was determined by quantitative real-time PCR.
(b) Histochemical staining and (c) quantitative 4-methylumbelliferyl β-d-glucuronide (MUG) assays of AtRALF23 promoter:GUS seedlings germinated and grown for 7 days in the presence or absence of 50 nm BL. Two biological replicates with two technical repeats each were used to determine the averages and standard deviations of the gene expression and quantitative MUG assays.

To study the function of AtRALF23, the gene was overexpressed in transgenic plants from a 35S:AtRALF23-myc construct. Several lines were obtained, and the ones producing detectable levels of transcripts by RT-PCR were selected for further study. Root growth in AtRALF-mycox seedlings was retarded in comparison with the wild type (Figure 2a). After 5 days on agar plates the length of the primary roots in wild-type seedlings averaged 13.6 ± 0.61 mm (SE), whereas the roots in AtRALF-mycox seedlings averaged 8.49 ± 0.43 mm (SE). Hypocotyl elongation in seedlings grown under low light conditions was reduced in overexpression lines (Figure 2b). The hypocotyls on 5-day-old wild-type seedlings were 5.5 ± 0.3 mm (SE) in length, whereas the hypocotyls on AtRALF-mycox seedlings were 3.4 ± 0.4 mm (SE) in length. We also tested BR responses under normal light conditions, and found that the BL-induced hypocotyl elongation is compromised in RALF23-mycox lines (Figure 2c). Shoot growth in overexpressing plants grown in soil was also reduced (Figure 2d). AtRALF23-mycox plants produced about the same number of leaves as the wild type; however, the leaves were smaller. The most prominent features of the AtRALF-mycox plants were their short stature and bushiness, with multiple inflorescences (Figure 2e). The abnormal growth in the AtRALF23-mycox lines was not caused by the epitope tag on AtRALF23, because lines overexpressing a 35S:AtRALF23 construct had the same phenotype (Figure S2).

Figure 2.

 AtRALF23-myc overexpression reduces Arabidopsis growth.
(a) Root growth in 5-day-old wild-type and AtRALF23-mycox seedlings grown on agar in the light (50 μmol m−2 s−1).
(b) Hypocotyl elongation in 5-day-old seedlings grown in low light (15 μmol m−2 s−1).
(c) Hypocotyl elongation assay in which wild-type and AtRALF23-mycox seeds were geminated and grown in the absence or presence of 50 nm BL for 5 days under normal lighting conditions (50 μmol m−2 s−1). Hypocotyl lengths were measured in 10–20 seedlings, and averages and standard deviations were calculated.
(d) Rosette development in 3-week-old plants grown in soil, and (e) growth of 5-week-old plants in soil.

AtRALF23-mycox seedlings also showed another phenotype, brought to light by transferring the seedlings onto plates containing the pH indicator bromocresol purple, in an unbuffered medium, with a starting pH of 6.3. The pH indicator clearly demonstrated that wild-type seedlings acidified the medium around their roots, whereas AtRALF23-mycox seedlings did not (Figure 3). Thus, the overexpression of AtRALF23 prevented the acidification of the rhizosphere under these conditions.

Figure 3.

 Acidification or alkalinization of the rhizosphere of Arabidopsis seedlings.
Seedlings were grown vertically on B5 plates for 7 days, and then transferred onto plates containing 0.006% bromocresol purple in unbuffered 1 mm CaSO4 (pH 6.3) (Wu et al., 2007). The pH indicator is generally yellow below pH 5.2, and purple above pH 6.8.

AtRALF 23 processing

AtRALF23 encodes a 138-aa preproprotein (Figure 4a), which is thought to be proteolytically processed to a bioactive peptide (Matsubayashi and Sakagami, 2006). The signal peptide on the N terminus of the preprotein is predicted to be 28 residues long (SignalP 3.0; Two bands were observed on heavily loaded western blots from root extracts of AtRALF23-mycox plants: an abundant, broad, slow-migrating band, and a less abundant, fast-migrating band, about the size expected for the myc-tagged, mature peptide (∼12 kDa) (Figure 4b). The mature RALF23 peptide has not been isolated and sequenced in Arabidopsis; however, the amino acid sequence is sufficiently similar to tobacco RALF (NtRALF) to assume that the Arabidopsis mature peptide is the same size (Figure 5a) (Pearce et al., 2001b). To demonstrate that the 12-kDa band is, indeed, a processed form of AtRALF23, we raised antiserum to a peptide in the mature AtRALF23 sequence, and demonstrated using western blot analysis that the 12-kDa band in the material immunoprecipitated by anti-myc reacted with the AtRALF23 antiserum (Figure S3).

Figure 4.

 AtRALF23-myc is processed in wild-type seedlings.
(a) Diagram of the AtRALF23-myc construct. AtRALF23 is synthesized as a prepropeptide, from which the mature (mat) peptide is proteolytically processed.
(b) Western blot probed with anti-myc showing the processing and appearance of the mature, myc-tagged peptide in overexpressing lines with a wild-type background. AtRALF23 is not processed in a AtRALF23[GG]-myc mutant and when expressed in the background of an s1p-3 mutant.

Figure 5.

 The AtRALF23-mycox phenotype depends on AtS1P.
(a) Clustal W alignment of the AtRALF23 sequence with tobacco RALF (NtRALF), and with three other closely related Arabidopsis RALFs (AtRALF1, AtRALF22 and AtRALF33). The sequences in red indicate the presumptive site-1-protease recognition site, the underlined sequences represent predicted signal peptides and the arrow indicates the actual cleavage site in AtRALF23.
(b) Phenotypes of 5-week-old plants. From left to right: wild type, AtRALF23-mycox in a wild-type background, AtRALF23[GG]-mycox in a wild-type background and AtRALF23-mycox in an s1p-3 background.

We posit that the broad slow-migrating band on western blots represents the uncleaved AtRALF23-myc precursor, although we cannot, as yet, account for the broadness of the band (Figure 4b). The abundance of the precursor relative to the cleaved product may result from the fact that it is derived from a transgene driven by a strong, constitutive promoter, and expressed at high levels. The high-level expression may exceed the capacity of the processing system.

It has been proposed that the mature AtRALF peptide is released from its precursor by proteolysis (Pearce et al., 2001b). That appears likely because there is a canonical subtilase site (Siezen and Leunissen, 1997) with a dibasic amino acid motif (RRXL) immediately upstream from the predicted N terminus of the mature AtRALF23 peptide and in nine other related AtRALFs, including the three shown here, AtRALF1, 22 and 33 (Figure 5a). The site is identical to the protease recognition site in AtbZIP17, a membrane-associated transcription factor that is cleaved by AtS1P in response to salt stress (Liu et al., 2007b). AtS1P is a Golgi-localized subtilisin-like serine protease (subtilase), similar to S1P in mammalian cells (Sakai et al., 1998).

To ask whether AtS1P is involved in processing AtRALF23 in vivo, we introduced the 35S:AtRALF23-myc construct into a s1p-3 mutant by crossing (Liu et al., 2007b). The s1p-3 mutant had previously been shown to be a full knock-out mutant (Liu et al., 2007b). As described above, we observed that 35S:AtRALF23-myc generated a characteristic dwarf, bushy phenotype in a wild-type background (Figure 5b). However, the 35S:AtRALF23-myc transgene did not produce a bushy phenotype in an s1p-3 background (Figure 5b), even though the transgene was still expressed (Figure S4). Most importantly, AtRALF23-myc processing was not detected in the s1p-3 mutant (Figure 4b). Thus, this finding indicates that AtRALF23-myc processing requires AtS1P, and without processing, the AtRALF23-myc precursor is not functional, i.e. does not produce an overexpression phenotype in 35S:AtRALF23-myc expressing plants. If processing of the endogenous AtRALF23 fails to occur in vivo, as would be expected in the s1p-3 mutant, then one might predict that growth would be more robust in the mutant. Indeed, it was found that the roots of s1p-3 seedlings grew faster on plates than the wild-type roots (Figure S5). The roots of 10-day-old wild-type seedlings were 16.5 ± 0.12 mm (SE) in length, while the s1p-3 roots were 18.77 ± 0.13 mm in length (SE). This observation is consistent with the findings above; however, it should be pointed out that there could be alternative explanations for the slight growth-promoting effect of s1p-3, as AtSIP is involved in other processes (Liu et al., 2007a,b). On pH-indicator media we also observed that s1p-3 seedlings tended to acidify the medium around their roots (Figure 3). In fact, the mutant seedlings appeared to acidify the medium more than the wild type, as judged by the size of the yellow halo around the root area. However, this observation has not been quantified. Nonetheless, it appears that failure to proteolytically process AtRALF23 in s1p-3 promotes rhizosphere acidification.

We also developed a construct in which the dibasic sequence in the canonical AtS1P recognition site (RRIL) was replaced by GG (35S:AtRALF23[GG]-myc). We analyzed eight lines bearing the construct, and all were normal: i.e. none showed a bushy phenotype (Figure 5b). One of the lines was tested for the expression of the transgene, and it was found that the slow-migrating precursor bands were expressed at normal levels, but the mature peptide could not be detected (Figure 4b).

Having shown that AtS1P was required for processing AtRALF-myc in vivo, we asked whether a semi-purified AtS1P could cleave AtRALF23 in vitro. This was done by expressing a modified form of AtS1P (35S:AtS1P-myc-HDEL) in transgenic Arabidopsis, pulling down AtS1P-myc-HDEL out of extracts with anti-myc beads, and incubating the well-washed beads with the fluorogenic peptide. The fluorogenic peptide, Abz-INRRILATRRY(NO2)D-OH, represents the putative subtilase recognition site in the RALF23 precursor. The peptide contains a fluorescent ortho-aminobenzoic acid group (Abz) and a 3-nitrotyrosine [(NO2)D-OH] fluorescence quencher. Peptide cleavage is measured by an increase in fluorescence. We observed that the fluorogenic peptide was readily cleaved by a factor in pull-downs from extracts of plants expressing AtS1P-myc-HDEL, but not in extracts from control transgenic plants harboring the empty vector (Figure 6a). The reaction products were analyzed by mass spectrometry, and a peptide of mass 903.53 amu appeared as the principal reaction product (Figure 6b). The reaction product was analyzed by tandem mass-spectrometry, and the spectrum of ions produced was consistent with the peptide representing the N-terminal half of the fluorogenic peptide, Abz-INRRIL (Figure 6c). This indicates that AtS1P-myc-HDEL cuts AtRALF23 immediately upstream of the N terminus of the mature peptide (Pearce et al., 2001b) (Figure 5a).

Figure 6.

 Cleavage reaction for the fluorogenic peptide representing the putative subtilase recognition site in AtRALF23.
(a) Time-dependent increase in fluorescence during the incubation of the fluorogenic peptide, Abz-INRRILATRRY(NO2)D-OH, with bead-bound AtS1P-myc-HDEL. Various controls, as indicated, including the use of the fluorogenic peptide in which GG is substituted for RR.
(b) MALDI-TOF mass spectrometry of the cleavage pattern for the fluorogenic AtRALF23 peptide by bead-bound AtS1P-myc-HDEL. The MS spectrum was acquired after 30 min of reaction. Parent and product peptide peaks are noted.
(c) MS/MS spectrum of peptide, with a mass peak of 903.53 amu derived from the fluorogenic peptide.


BR and RALFs

The findings in this study suggest that the downregulation of AtRALF23 may be part of the mechanism by which BR stimulates plant growth and cell expansion. Others have proposed that BR promotes cell elongation, in part, by altering the mechanical properties of the cell wall (Zurek et al., 1994). It is thought that BR may be acting much like auxin, or perhaps through the auxin signaling pathway, by inducing cell-wall acidification (Hardtke, 2007; Hardtke et al., 2007). Cerana et al. (1983) showed that BR stimulated growth and proton extrusion in Azuki bean epicotyls (Vigna angularis). The effects of BR and auxin (indole-3-acetic acid, IAA) in that system appeared to be additive, suggesting that the hormones may be acting independently. BL also stimulated root growth and proton extrusion in maize root segments (Romani et al., 2006).

AtRALF23 is an inhibitor of plant growth, and is downregulated by BR. Pearce et al. (2001b) demonstrated that a chemically synthesized tomato RALF inhibited root growth in tomato and Arabidopsis seedlings. Wu et al. (2007) showed that silencing of the single RALF gene in Nicotiana attenuata (NaRALF) using an inverted repeat (ir) construct resulted in seedlings with roots that grew more rapidly, but produced abnormal root hairs. The irRALF lines tended not to acidify the medium around the tips of the root hairs, and the authors argued that the abnormal root hairs resulted from a higher apoplastic pH. They proposed that the irRALF phenotype was probably a result of pH effects, because when they grew seedlings on medium buffered to pH 5.5, they could restore the wild-type phenotype in the irRALF lines. The authors did point out that the effects of irRALF were contrary to expectations, as RALF induces alkalinization, and not acidification, of the medium in tobacco cell suspension cultures (Pearce et al., 2001b).

Our findings are quite different from Wu et al. (2007), but are consistent with the observations that RALF induces rhizosphere alkalinization. In our experiments, the AtRALF23-myc overexpressers prevented the acidification of the medium surrounding the roots. On the other hand, the roots of wild-type and s1p-3 seedlings clearly acidified the rhizosphere, based on pH indicator color changes. The s1p-3 mutant appeared even more effective than the wild type in acidifying the medium, consistent with the expectation that the mutant fails to process AtRALF23 and produce an active alkalinization factor.

RALFs tend to counteract cell-wall acidification, presumably by blocking membrane-associated proton pumps leading to rapid media alkalization (Moura et al., 2006). It is not known how RALFs elicit an alkalization response. A RALF receptor has not yet been identified, although Scheer et al. (2005) used a labeled form of a tomato RALF ([125I]azido-LeRALF), and found two membrane proteins that would bind the peptide. The authors reported that the alkalization response caused by LeRALF was effectively blocked by suramin, a heterocyclic, polysulfonated inhibitor of ligand–receptor interactions, suggesting that the alkalization response caused by LeRALF was, indeed, produced through ligand–receptor interaction (Scheer et al., 2005). However, the proteins that bind LeRALF have not yet been isolated and identified; therefore, the LeRALF receptor is not known.

Other evidence, nonetheless, supports the proposition that RALFs are ligands for receptor-mediated responses. Alkalization in response to a tobacco RALF, NtRALF, is very fast, peaking at less than 5 min after adding the peptide to tobacco cell-suspension cultures (Pearce et al., 2001b). In addition, NtRALF elicits a rapid MAP kinase response that is coincident with the alkalization response (Pearce et al., 2001b). However, it is not known whether RALFs affect proton pumping as a final effector in a signaling pathway, or whether RALFs interact directly with a membrane-associated proton pump. Such a pump mediating the BR-induced acidification response has not been firmly identified. On the other hand, RALF signaling may be very indirect, affecting other signaling processes that in turn impact on the operation of membrane-associated proton pumps. In this regard, it is interesting that Haruta et al. (2008) found that AtRALF1 elevates cytoplasmic Ca2+ levels in Arabidopsis seedlings, as indicated by aequorin luminescence. It is possible that changes in cytoplasmic Ca2+ levels might trigger other ion pumps or gates, leading to the alkalization response associated with RALFs.

In any case, BRs regulate RALF23 by downregulating its expression. BRs are well known for their ability to regulate gene expression, and the BR signaling pathway controls transcription factors that up- or downregulate BR-responsive genes (see recent reviews: Belkhadir and Chory, 2006; Gendron and Wang, 2007; Li and Jin, 2007). The BR signaling pathway affects the phosphorylation and activity of BES1, one of two transcription factors that activate or repress BR-responsive genes (Yin et al., 2005, 2002). We observed that AtRALF23 was downregulated in a constitutively activated bes1-D mutant, indicating that AtRALF23 is regulated by the BR signaling pathway. We do not know yet whether the AtRALF23 is a direct target of BES1; however, the presence of BES1 binding sites in the AtRALF23 gene promoter imply that BES1 may directly regulate AtRALF23 expression (RS, J-XL, HG, YY and SHH, unpublished data).

The repression of AtRALF23 by BR signaling may have several implications for BR-induced cell elongation. Most obviously, downregulation of a negative regulator (AtRALF23) for cell elongation may be required for optimal BR-induced growth. The AtRALF23 overexpression phenotypes strongly support this possibility. AtRALF23 overexpression is characterized by a reduction in vegetative growth, particularly in root elongation and leaf expansion in seedlings, as well as BR-induced hypocotyls elongation. AtRALF23 overexpressing plants also show a reduction in inflorescence stem elongation, and a loss in apical dominance. An AtRALF23 loss-of-function phenotype has not yet been characterized because a knock-out in the gene has not been identified, and RNAi constructs specific for individual RALFs are challenging to produce, because some of the sequences of the 34 members of the RALF gene family are similar. Some of the other members of the RALF gene family are also negative growth regulators. Like AtRALF23, AtRALF1 overexpression results in semidwarf plants with reduced leaf and root growth (Matos et al., 2008). Unlike AtRALF23, AtRALF1 expression is not downregulated by BL treatment; however, AtRALF1 is normally expressed at much lower levels than AtRALF23 (Arabidopsis eFP browser,

AtRALF23 processing

In their study of AtRALF1, Matos et al. (2008) observed that a canonical subtilase site (dibasic AA) upstream from the presumed start of the mature peptide was important for the processing of the precursor. They mutated the site and demonstrated that the precursor was not processed, and the construct did not produce the overexpression phenotype. They showed that the AtRALF1 precursor can be processed in extracts by microsomal fractions, but were unable to identify the enzyme responsible for the proteolysis. Nonetheless, their findings are consistent with the possibility that a protease associated with the secretory pathway might be responsible for the processing of AtRALF1.

An important finding in our study was the discovery that AtS1P is involved in AtRALF23 processing. The first function reported for AtS1P in plants was the processing of membrane-associated transcription factors (Liu et al., 2007b). AtS1P participates in processing AtbZIP17, a factor involved in endoplasmic reticulum (ER) stress responses. AtS1P is largely located in the Golgi apparatus (Liu et al., 2007b), and it is thought that in response to ER stress, these transcription factors are released from the ER and translocated to the Golgi, where they undergo proteolytic processing.

RALFs are thought to be secreted peptides, and since we showed that AtRALF23 requires AtS1P for processing, it is likely that the AtRALF23 propeptide is processed in the Golgi on its way through the secretory pathway. Escobar et al. (2003) tracked RALF after transfecting Nicotiana benthamiana leaves with NtRALF-GFP via a viral vector. They first found NtRALF-GFP in the ER, and later in the apoplast. It would seem possible that some AtS1P is secreted, and that AtRALF23 propeptide is proteolytically processed in the apoplast. AtS1P has a C-terminal transmembrane domain that is thought to anchor the enzyme to the membranes of the secretory pathway. In mammalian cells it was found that a small quantity of S1P undergoes autocatalytic cleavage to release the truncated enzyme, which is then secreted into the media (Cheng et al., 1999; Espenshade et al., 1999). It is unlikely that proAtRALF23 is cleaved by AtS1P in the apoplast, because the Arabidopsis enzyme has a pH optimum of 7.5 (RS, unpublished data), and little or no activity at the acidic pH of the apoplast. However, if a small quantity of proAtRALF23 is processed, and begins to promote the alkalization of the apoplast, then that might generate an apoplastic environment conducive to the processing of AtRALF23 by a released form of AtS1P.

In the case of Arabidopsis PHYTOSULFOKINE 4 (AtPSK4), the expression of both the gene and the processing activity are upregulated by wounding. When the AtPSK4 transgene is constitutively expressed, processing regulates the appearance of the mature AtPSK4 peptide (Srivastava et al., 2008). Although processing is required for AtRALF23 activity, it is unlikely that processing regulates AtRALF23 in response to BL. The microarray data of Nemhauser et al. (2004) indicate that AtRALF23 expression is regulated by BL, but that the expression of AtS1P is not.

AtS1P processing cuts the AtRALF23 propeptide cleanly at the N terminus of the predicted mature peptide. Hence, AtS1P cleavage is probably the only proteolytic step needed to activate AtRALF23. This is in contrast to the processing of AtPSK4 by AtSBT1.1, which releases the peptide from the larger precursor by cutting three amino acid residues from the N terminus of the predicted mature peptide (Srivastava et al., 2008). The activation of AtPSK4 is likely to require the trimming of three residues from the N terminus and five residues from the C terminus of the released peptide, by more than one protease. However, results from this study demonstrate that after the removal of the signal peptide, AtRALF23 maturation appears to be dependent only on one protease, AtS1P.

Experimental procedures

Plant material and growth conditions

T-DNA insertion mutant lines for s1p-3 were obtained from the Arabidopsis Biological Resource Center (ABRC, Seeds were surface sterilized, rinsed with sterile water and stratified at 4°C for at least 2 days in 0.1% agar. Seeds were germinated and grown vertically on agar plates containing Gamborg’s B5 medium (Gamborg et al., 1968). To test for media acidification, seedlings were transferred 7 days after growth to plates containing bromocresol purple pH indicator, according to the method described by Wu et al. (2007).

Plasmid construction

PreproAtRALF23 was amplified from the RNA of 1-week-old Arabidopsis seedlings by RT-PCR using primer PSKMRalf23 (Table S1), and was then cloned into the AscI and SpeI sites of pSKM36 in frame with a 4X epitope myc tag (EQKLISEEDLRN). cDNA clones with error-free copies were named ppAtRALF23. A mutated form of AtRALF23 (RR → GG) was generated using a Quick-Change Site-directed Mutagenesis Kit (Stratagene,, with primers SDMRalf23 and PSKMRalf23 as a template. Promoter:GUS constructs for AtRalf23 were generated by amplifying 2040 nucleotides using the primers PBIRalf23. The amplified promoter was ligated into the BamHI and SalI sites of pBI101.2.

Detection of AtRALF23 processing in vivo

Total protein was extracted from transgenic and wild-type plants with extraction buffer: 0.1 m Hepes-KOH, pH 7.0, 20 mm 2-mercaptoethanol, 0.1 mg ml−1 phenylmethylsulphonyl fluoride (PMSF), 0.1% (v/w) Triton X-100, 1 mm ethylenediamine tetraacetic acid (EDTA), 20% (v/w) glycerol and protease inhibitor cocktail (Sigma-Aldrich, An aliquot of total protein was precipitated with trichloroacetic acid and quantified by the Bradford method (Bradford, 1976). Reaction products were resolved on 12% SDS-PAGE gels and visualized by western blotting using c-myc antibody (9E10; Santa Cruz Biotechnology,, or with antiserum raised against a peptide (ATRRYISYGALRRNTIPCSR) in the mature form of AtRALF23 and an ECL kit (GE Healthcare,

Assay for AtS1P activity in vitro

AtS1P-myc was affinity purified from transgenic Arabidopsis plants as follows: 500 g of seedlings were ground in liquid nitrogen and suspended in 25 mm Tris-HCl, pH 7.2, 150 mm NaCl, 0.1% Nonidet P-40 and 10% glycerol. A 200-μl volume of anti-c-myc agarose conjugate (A7470 from Sigma-Aldrich) was added to the filtered lysate and incubated for 2 h at 4°C with continuous rotation. The agarose beads with bound AtS1P-myc were recovered by centrifugation at 1000 g for 3 min at 4°C. The beads were washed three times with washing buffer (25 mm Tris-HCl, pH 7.2, and 150 mm NaCl), and were finally suspended in 25 mm 2-(N-morpholine)-ethanesulphonic acid (MES)-sodium acetate buffer (pH 6.0). The reactions were carried out at 32°C in the same buffer supplemented with 2.5 mm CaCl2. Parallel purification was performed using extracts from non-transgenic plants to obtain material for control reactions.

For fluorogenic peptide assays, 40 μl of bead-bound AtS1P-myc were added to a solution containing a final concentration of 50 μm fluorogenic peptide, in a buffer consisting of 25 mm MES-sodium acetate, pH 6.0, supplemented with 2.5 mm CaCl2. Kinetic assays were performed at 32°C, and were monitored as fluorescence emission at 420 nm (10-nm slit) following excitation by 320 nm (10-nm slit) in a BioTek Spectrophotometer. The reaction was carried out in 96-well plates (Nunc, Control reactions were performed using a mutant fluorogenic peptide in which RR had been replaced by GG in the protease recognition site.

Gene expression analysis

Total RNA was isolated from ground plant tissues using an RNeasy kit, treated with RNase-free Dnase I, according to the manufacturer’s instructions (Qiagen,, and was quantified by 260/280-nm UV light absorption. A 1-μg portion of total RNA was reverse transcribed using the Supertranscript III RT kit (Invitrogen). A 2-μl volume of cDNA was used for RT-PCR, and cDNA diluted ten times was used for real-time quantitative RT-PCR. All primers are listed in Table S1. For quantitative real-time RT-PCR, the efficiency of amplification of various RNAs was assessed relative to the amplification of transcripts for two actin genes (actin2, At3g18780, and actin8, At1g49240) or an unknown gene (At4g18810) that did not show BR regulation. RNA samples were assayed in triplicate. Expression levels were calculated relative to reference levels using a comparative threshold cycle method (CT method) with ΔΔCt =ΔCt-reference − ΔCt-sample, where ΔCt-sample was the Ct value for the assay sample normalized to actin, and ΔCt-reference is the Ct value for calibration, also normalized to actin (Liu et al., 2007a).

GUS assays

Seedlings were harvested from B5 medium and incubated for 6 h in GUS staining solution [100 mm Tris/NaCl buffer, pH 7, 2 mm ferricyanide, 2 mm 5-bromo-4-chloro-3-indolyl-d-glucuronide (X-gluc), 2 mm ferrocyanide, 10 mm EDTA and 0.1% Triton) at 37°C in the dark. The staining solution was removed, and the tissues were dehydrated in an ethanol series from 70% (v/v) to absolute ethanol. Samples were visualized under the light microscope. Quantitative GUS assays were performed essentially as described by Jefferson et al. (1987). Briefly, total proteins from about 100 mg of seedlings were extracted with extraction buffer (100 mm potassium phosphate, pH 7.8, 1 mm EDTA, 1% Triton X-100, 10% glycerol and 1 mm DTT). The assay buffer is the same as the extraction buffer plus 1 mg ml−1 MUG (methylumbelliferyl β-d-glucuronide). The products (MU) were quantified with a BIO-TEK Synergy HT multidetection microplate reader, with excitation at 360 nm and emission at 460 nm. The relative GUS activities were normalized against protein concentrations for each sample, which were determined by Bradford assay.

Mass spectrometry

MALDI-TOF MS and MS/MS analyses were performed using a QSTAR XL quadrupole TOF mass spectrometer (AB/MDS Sciex, equipped with an oMALDI ion source. The mass spectrometer was operated in the positive ion mode. Mass spectra for MS analysis were acquired over a mass to charge ratio (m/z) = 500–2500. After every regular MS acquisition, MS/MS acquisition was performed against the most intensive ions. The molecular ions were selected by information-dependent acquisition (IDA) in the quadrupole analyzer, and were fragmented in the collision cell.


This work was supported by a grant to SHH from the NSF 2010 program (IBN0420015) and a NSF grant to YY (IOS0546503).