The Ca2+-binding protein PCaP2 located on the plasma membrane is involved in root hair development as a possible signal transducer


For correspondence (e-mail


Plasma membrane-associated Ca2+-binding protein–2 (PCaP2) of Arabidopsis thaliana is a novel-type protein that binds to the Ca2+/calmodulin complex and phosphatidylinositol phosphates (PtdInsPs) as well as free Ca2+. Although the PCaP2 gene is predominantly expressed in root hair cells, it remains unknown how PCaP2 functions in root hair cells via binding to ligands. From biochemical analyses using purified PCaP2 and its variants, we found that the N–terminal basic domain with 23 amino acids (N23) is necessary and sufficient for binding to PtdInsPs and the Ca2+/calmodulin complex, and that the residual domain of PCaP2 binds to free Ca2+. In mutant analysis, a pcap2 knockdown line displayed longer root hairs than the wild-type. To examine the function of each domain in root hair cells, we over-expressed PCaP2 and its variants using the root hair cell-specific EXPANSIN A7 promoter. Transgenic lines over-expressing PCaP2, PCaP2G2A (second glycine substituted by alanine) and ∆23PCaP2 (lacking the N23 domain) exhibited abnormal branched and bulbous root hair cells, while over-expression of the N23 domain suppressed root hair emergence and elongation. The N23 domain was necessary and sufficient for the plasma membrane localization of GFP-tagged PCaP2. These results suggest that the N23 domain of PCaP2 negatively regulates root hair tip growth via processing Ca2+ and PtdInsP signals on the plasma membrane, while the residual domain is involved in the polarization of cell expansion.


Root hairs, a cellular protuberance from the root epidermis, constitute a substantial proportion of the plant interface with the underground environment (Emons and Ketelaar, 2009). In addition to their important functions, e.g. nutrient uptake and root anchoring, root hairs have been intensively studied with regard to their characteristic morphogenesis. In Arabidopsis thaliana, a root hair starts as a bulge on the outer surface of the distal end of a specific root epidermal cell, and then elongates perpendicularly to the root surface with a constant diameter via tip-focused cell expansion, a process called tip growth (Gilroy and Jones, 2000; Ryan et al., 2001). In both processes (bulge formation and tip growth), the polarity of cell expansion is strictly regulated, suggesting that a robust mechanism underlies the highly polarized membrane traffic and cytoskeletal reorganization (Hepler et al., 2001; Smith and Oppenheimer, 2005; Cole and Fowler, 2006).

Genetic and physiological studies have revealed that the mechanism for establishing and sustaining cell expansion polarity involves interactions between multiple signals mediated by Ca2+, reactive oxygen species, Rho-related GTPases of plants and phospholipids, including phosphatidylinositol phosphates (PtdInsPs) (Cole and Fowler, 2006; Heilmann, 2009). Of these, Ca2+ and reactive oxygen species signals mutually promote their tip-focused localizations through a Ca2+ channel that responds to reactive oxygen species and the NADPH oxidase RHD2 that responds to Ca2+ (Foreman et al., 2003; Takeda et al., 2008). The Rho-related GTPase signal is thought to be an upstream regulator (Carol et al., 2005). Of the phospholipids, PtdIns(4,5)P2, a key signaling molecule directing polarized membrane traffic via the regulation of actin dynamics in eukaryotic cells, localizes to the plasma membrane of developing root hairs with a tip-focused gradient. Phosphatidylinositol phosphate 5–kinase 3 (PIP5K3), one of the PtdIns(4,5)P2-synthesizing enzymes in A. thaliana, also localizes specifically to the plasma membrane of the developing root hair tip, and defects in this gene cause significantly shorter root hairs than in the wild-type (Kusano et al., 2008; Stenzel et al., 2008). However, it remains to be elucidated how the PtdIns(4,5)P2 signal interacts with the other signals to establish and sustain cell polarity during root hair development.

Recently, a novel Ca2+-binding protein was found in the plasma membrane fraction of A. thaliana and named AtPCaP2 (hereafter referred to as PCaP2) (Kato et al., 2010). PCaP2 is predominantly expressed in the root, especially root hair cells. A related protein, PCaP1 of A. thaliana, is expressed in most tissues and has been characterized in detail (Ide et al., 2007; Nagasaki et al., 2008; Nagasaki-Takeuchi et al., 2008). PCaP1 and PCaP2 comprise 224 and 168 amino acid residues, respectively. Although PCaP2 shares only 28% sequence identity with PCaP1, the two proteins have similar 23 amino acid N–terminal domains, and notably high contents of proline, glutamate, valine and lysine residues in their central and C–terminal regions (Kato et al., 2010). Both N–termini have an N–myristoylation signal, and have been confirmed to be myristoylated at Gly2. PCaP1 and PCaP2 have no known functional enzymatic motifs. Therefore, the proteins are non-enzymatic proteins associated with the plasma membrane through a myristoyl moiety, although PCaP2 has been characterized as microtubule-associated protein 18 (MAP18) (Wang et al., 2007).

The ability of PCaP2 to bind PtdInsPs, Ca2+ and a Ca2+/CaM complex is characteristic. These ligands are key components of intracellular signaling (Zielinski, 1998; Meijer and Munnik, 2003; Carlton and Cullen, 2005; McCormack et al., 2005; Cheung and Wu, 2008; DeFalco et al., 2009; Kleerekoper and Putkey, 2009; Luan, 2009; Xue et al., 2009). Thus, its biochemical properties suggest that PCaP2 mediates a signal from Ca2+/CaM and PtdInsPs on the plasma membrane. The biochemical relevance of these ligands and the physiological role of PCaP2 in plant cells are unclear.

Here, we prepared highly purified recombinant proteins of normal PCaP2 and mutant PCaP2, and characterized their biochemical properties to identify the binding region(s) of PtdInsPs, the Ca2+/CaM complex and free Ca2+. Analysis using proteins tagged with GFP in transgenic plants revealed involvement of the N23 domain in the plasma membrane localization of PCaP2. We also investigated the physiological function of PCaP2 using mutant and transgenic plants. pcap2 knockdown mutant roots exhibited longer root hairs than the wild-type. Transgenic lines over-expressing PCaP2 under the control of the root hair-specific promoter of the EXPANSIN A7 gene (EXPA7) (Cho and Cosgrove, 2002) exhibited branched root hairs. A similar root hair phenotype was observed in transgenic lines over-expressing modified PCaP2 proteins. On the other hand, over-expression of the N–terminal domain of 23 amino acids suppressed root hair emergence and elongation. These results suggest a unique function of PCaP2 as a signal transducer between Ca2+ and PtdInsP signals on the plasma membrane. We discuss the physiological role of PCaP2 in relation to Ca2+ and PtdInsP signals in root hair development and the polarity of cell expansion.


Preparation of recombinant mutated PCaP2 and its N–terminal peptide

To determine the binding sites of PCaP2 for ligands such as PtdInsPs, we characterized the biochemical properties of PCaP2 using recombinant proteins. PCaP2 possesses a polybasic region of 23 residues at the N–terminal domain that is rich in lysine residues (N23) (Figure 1a). The region from the 24th to 168th residue (the center and C–terminal region; hereafter referred to as the CC region) is rich in proline, glutamate, valine and lysine residues (Kato et al., 2010).

Figure 1.

Preparation of recombinant PCaP2 variants.

(a) Constructs of full-length PCaP2 (PCaP2) and its variants: the N–terminal part of PCaP2 (N23) and PCaP2 without the N–terminal residues 2–23, ∆23PCaP2). These proteins and the peptide were tagged with a 6×His tag. Lysine and arginine residues are shown in bold.

(b) Recombinant ∆23PCaP2 tagged with (His)6 expressed in E. coli was purified by column chromatography and subjected to SDS–PAGE. Lane m, marker proteins; lane 1, soluble fraction prepared from E. coli expressing ∆23PCaP2; lane 2, flow-through fraction of HiTrap Phenyl HP column chromatography; lane 3, purified fraction after dialysis; lane 4, purified PCaP2; lane 5, purified ∆23PCaP2. The identity of the authentic peptide of N23 tagged with 6×His was confirmed by SDS–PAGE (lane 6).

(c) Calcium–45 overlay assay. Purified preparations of IgG (negative control), CaM (positive control), and recombinant PCaP2, ∆23PCaP2 and N23 of the indicated amounts (upper panel, 3 μg; lower panel, 2, 4 and 6 μg) were blotted on to a poly(vinylidene difluoride) membrane and incubated with 1 mm CaCl2 and 3.7 MBq 45Ca as CaCl2. An autoradiogram of the 45Ca-labeled proteins on the membrane was obtained by exposure to X–ray film.

To characterize the biochemical properties of the N–terminal domain of PCaP2, we prepared two types of recombinant PCaP2: normal PCaP2 and N–terminally truncated PCaP2 (∆23PCaP2) (Figure 1a). PCaP2 and ∆23PCaP2 were tagged with a 6×His tag at their C–termini, expressed in Escherichia coli cells, and isolated from the soluble fraction of the cell lysate. Highly purified preparations of the recombinant proteins were obtained by three-step Ni–NTA column chromatography, phenyl hydrophobic interaction and QAE [diethyl-(2-hydroxypropyl) aminoethyl] columns. The molecular size of ∆23PCaP2 was smaller than that of PCaP2 on SDS–PAGE (Figure 1b). In addition, a peptide of the N23 domain of PCaP2 tagged with 6×His was synthesized, and subjected to SDS–PAGE to confirm the molecular size and purity. These proteins were used for further biochemical characterization.

Ca2+-binding site in PCaP2

To examine the Ca2+-binding site of the PCaP2, we performed a 45Ca2+ overlay assay. Membrane sheets blotted with PCaP2, ∆23PCaP2 and N23 were incubated with 45Ca2+. PCaP2 and ∆23PCaP2, but not N23, gave a positive signal in a dose-dependent manner even in the presence of 5 mm MgCl2 and 60 mm KCl (Figure 1c). This indicates that Ca2+ binds to the CC region of PCaP2, which lacks the N23 domain, under physiological conditions.

Interaction with PtdInsPs is mediated by the N23 domain of PCaP2

The binding of negatively charged phospholipids such as PtdIns(4,5)P2 to proteins may be mediated by short sequences enriched in basic residues as discussed previously (Fivaz and Meyer, 2003). PCaP2 possesses such a polybasic region in the N–terminal domain (Figure 1a). To examine its binding capacity to PtdInsPs, we performed in vitro lipid-binding assays. PCaP2, ∆23PCaP2 and N23 were incubated with nitrocellulose membrane sheets (PIP strips), which were spotted with a series of PtdInsPs. The interaction of the proteins with lipids was detected using anti-His tag antibodies. PCaP2 preferentially bound to PtdIns(3,4)P2, PtdIns(3,5)P2, PtdIns(4,5)P2 and PtdIns(3,4,5)P3 (Figure 2a). ∆23PCaP2 did not bind to these lipids. The N23 peptide bound to PtdInsP1s, PtdInsP2s, PtdInsP3 and phosphatidic acid, but not to less polar lipids including phosphatidylcholine, phosphatidylethanolamine or phosphatidylinositol.

Figure 2.

The N–terminal domain of PCaP2 interacts with PtdInsPs.

(a) The binding capacity of PCaP2 to PtdInsPs was tested using PIP strips. Recombinant PCaP2 (500 ng ml−1), ∆23PCaP2 (500 ng ml−1) and N23 peptide (100 ng ml−1) were incubated with PIP strips at 4°C for 16 h. Proteins bound to lipids were detected using the anti-His tag antibody. LPA, lysophosphatidic acid; LPC, lysophosphatidylcholine; PtdIns, phosphatidylinositol; PtdIns(3)P, phosphatidylinositol 3–monophosphate; PE, phosphatidylethanolamine; PC, phosphatidylcholine; S1P, sphingosine 1–phosphate; PA, phosphatidic acid; PS, phosphatidylserine.

(b) The affinity of PCaP2 for individual PtdInsP was determined using a PIP array™ sheet on which lipids were immobilized at the indicated amounts.

Furthermore, we assessed the PtdInsP-binding capacity quantitatively using PIP array sheets (Figure 2b). PCaP2 interacted with PtdIns(3,4)P2, PtdIns(3,5)P2, PtdIns(4,5)P2 and PtdIns(3,4,5)P3. ∆23PCaP2 showed no capacity for binding to PtdInsPs, although a background signal was observed for PtdInsP1. N23 strongly bound to all PtdInsPs examined, but not to PtdIns. These results suggest that PCaP2 binds to PtdInsPs directly through the N23 domain. N23 bound to the PtdInsPs more strongly than PCaP2 even when applied at 20% of the amount used for PCaP2 (Figure 2b). This is probably due to the difference in molecular size and/or tertiary structure between PCaP2 and N23. PtdInsPs in the spot area of the membrane may accept a larger number of the small N23 peptides but not PCaP2. There is also the possibility that the CC region interferes with interaction of the N23 domain with PtdInsPs.

Ca2+/CaM binds to the N23 domain of PCaP2

To examine the CaM binding domain of PCaP2, we performed a pull-down assay with CaM–agarose beads. Recombinant proteins were incubated with CaM–agarose in the presence of Ca2+ or EGTA, a Ca2+ chelator. The proteins associated with CaM–agarose were extracted, subjected to SDS–PAGE, and then detected by immunoblotting with antibodies against the 6×His tag to detect PCaP2. As shown in Figure 3(a), PCaP2 bound to CaM–agarose in the presence of Ca2+, but not in the presence of EGTA. Furthermore, the N23 peptide bound to CaM–agarose only in the presence of Ca2+; however, ∆23PCaP2 did not. These results indicate that the N23 domain is necessary and sufficient for binding to the Ca2+/CaM complex. The N23 domain has a possible CaM-binding motif, namely a 1–8–14 domain, in which hydrophobic residues are positioned at the 1st, 8th and 14th residues (Rhoads and Friedberg, 1997). In a helical wheel model, three corresponding residues, Val8, Leu15 and Ala21, are located at both ends of the basic region, which comprises lysine and arginine residues (Figure 3b). The basic surface of the helical wheel may interact with negatively charged phosphate moieties on the inositol ring by electrostatic bonding.

Figure 3.

A Ca2+/CaM complex interacts with the N–terminal domain of PCaP2.

(a) Recombinant PCaP2 and ∆23PCaP2 proteins and an N23 peptide (3.9 μg each) were incubated with CaM–agarose in the presence of 0.5 mm Ca2+. Proteins bound to the CaM–agarose were released by SDS and detected by immunoblotting with anti-His tag antibody. ∆23PCaP2 (left panel) and N23 (right panel) were assayed for CaM-binding ability in the presence or absence (EGTA) of Ca2+.

(b) A helical wheel model of the N–terminal domain. Lysine and arginine residues are shown in red and conserved residues in the CaM binding consensus sequence are shown in blue. The curved line shows the estimated region of the PtdInsP-binding site.

Knockdown mutation of PCaP2 resulted in a long root hair phenotype

To investigate physiological roles of the PCaP2 gene, we obtained a T–DNA tagging knockdown mutant, pcap2 (SALK_021652; Alonso et al., 2003), which contains a T-DNA insertion into the upstream of the PCaP2 gene, from the Arabidopsis Biological Resource Center, and established a homozygous line with the T–DNA insertion. Junction sequences between the T–DNA and genomic DNA were determined, revealing that the line contains the T–DNA insertion at 474 bp upstream of the PCaP2 translation start site (Figure S1). Because the insertion mutation was thought not to affect the primary structure of the gene product but to affect the level of gene expression, we examined the PCaP2 transcription level in the mutant seedlings by real-time RT–PCR, and found that the level was approximately 20% of the wild-type (Figure 4a). To assess the effect of knockdown of the gene, we compared mutant root hairs with those of the wild-type because the PCaP2 promoter is specifically active in root hair cells (Kato et al., 2010). The mutant root hairs were found to be significantly longer on average than those of the wild-type (Figure 4b,c). Plants heterozygous for the T–DNA insertion exhibited intermediate levels with regard to both PCaP2 transcript levels and root hair length between the wild-type and the homozygous mutant (Figure S2). These results indicate that the PCaP2 gene functions negatively and quantitatively in root hair elongation. Mutant root hairs developed at a similar distance from the root tip to those of the wild-type, but the elongation rate of mutant root hairs was higher than that of the wild-type, especially in late stages of root hair development (Figure S3). Except for root hair length, mutant plants did not exhibit any visible phenotypes in roots or other organs (Figure S4).

Figure 4.

The knockdown mutant pcap2 showed a long root hair phenotype.

(a) Transcription levels of PCaP2 in the wild-type (WT) and pcap2. Homozygous seedlings were examined by real-time RT–PCR using PCaP2 gene-specific and ACT2 gene-specific primer sets. Relative mRNA contents in the wild-type and pcap2 are shown, with the wild-type contents arbitrarily set as 1. Values are means ± SE of two independent experiments.

(b) Root hair phenotypes of the wild-type and pcap2. Ten-day-old seedlings grown on vertical 1.2% agar medium are shown. Scale bars = 500 μm.

(c) Distributions of root hair lengths in the wild-type and pcap2. Ten-day-old seedlings grown on vertical 1.2% agar medium were measured. The mean values of root hair lengths (Avr.) were significantly different between the wild-type and pcap2 (Student's t test, P < 0.0001, = 174 for the wild-type and = 212 for pcap2).

The N23 domain is essential for association with the plasma membrane

Polybasic regions of the proteins have been proposed to play a role in their membrane targeting through interactions with PtdInsPs (Heo et al., 2006). To investigate how the N23 domain is involved in the plasma membrane localization of PCaP2, we expressed PCaP2 and its variants tagged with GFP in transgenic A. thaliana. In addition to PCaP2–GFP, ∆23PCaP2–GFP and N23–GFP constructs, we prepared a PCaP2G2A–GFP construct to estimate the contribution of N–myristoylation at Gly2 to the plasma membrane localization (Figure 5a).

Figure 5.

Localization of GFP-tagged PCaP2 and its variants expressed under the control of the EXPA7 promoter at the plasma membrane depends on the N23 domain.

(a) The structures of the four constructs PCaP2–GFP, PCaP2G2A–GFP, Δ23PCaP2–GFP and N23–GFP.

(b−e) Fluorescence signals of GFP-tagged proteins in root hair cells of 5-day-old seedlings were observed with a confocal laser-scanning microscope. Root hair cells (trichoblasts) expressing PCaP2–GFP (b), PCaP2G2A–GFP (c), Δ23PCaP2–GFP (d) and N23–GFP (e) at the bulge stage are shown.

(f−q) Fluorescence signals of GFP-tagged proteins and FM4–64 in root hair tips were observed 5 min after staining with FM4–64. Fluorescence signals for GFP (f, i, l, o), FM4–64 (g, j, m, p) and their overlay (h, k, n, q) in root hairs expressing PCaP2–GFP (f–h), PCaP2G2A–GFP (i–k), Δ23PCaP2–GFP (l–n) and N23–GFP (o–q) are shown. Scale bars = 20 μm.

These constructs were expressed under the control of the EXPA7 promoter, which is specific to root hair cells (Cho and Cosgrove, 2002). Localization patterns of these GFP-tagged proteins were observed in root hair cells at the bulge-forming stage, because they caused root hair phenotypes that may confuse estimation of their intracellular localization at later stages (see below). At least eight independent lines for each construct were established, and the same localization pattern of GFP-tagged protein was observed in all established lines for each construct. The results are shown in Figure 5. Fluorescence of the PCaP2–GFP, PCaP2G2A–GFP and N23–GFP constructs was detected mainly in the cell peripheral region, including the plasma membrane (Figure 5b,c,e). However, fluorescence of the ∆23PCaP2–GFP construct was detected throughout the cytoplasmic space (Figure 5d). For detailed localization of these proteins, the tip regions of root hairs without apparent abnormalities were treated with a lipophilic fluorescent dye, N–(3–triethylammoniumpropyl)-4–(6–(4–(diethylamino)phenyl)hexatrienyl)pyridinium dibromide (FM4–64), for a sufficiently short period that the dye stained mainly the plasma membrane. Then, fluorescence from GFP and FM4–64 was observed simultaneously. The overlaid images indicate that PCaP2–GFP and N23–GFP localize predominantly to the plasma membrane (Figure 5f–h,o–q), and that ∆23PCaP2–GFP localizes predominantly to the cytoplasm (Figure 5l–n). Therefore, the N23 domain is necessary and sufficient for association of PCaP2–GFP with the plasma membrane. Fluorescence of the PCaP2G2A–GFP construct was detected in both the plasma membrane and the cytoplasm (Figure 5i–k). This suggests that association of PCaP2 with the membrane is enhanced by interaction of the protein with the membrane components in addition to N–myristoylation.

Over-expression of PCaP2 and its variants affected root hair morphogenesis

To investigate the physiological roles of PCaP2 and its domains in root hair development in detail, we used a reverse genetic approach by over-expressing PCaP2 and its variants in a root hair cell-specific manner. Four types of construct (PCaP2, PCaP2G2A, ∆23PCaP2 and N23) were expressed in transgenic A. thaliana under the control of the EXPA7 promoter. At least six homozygous transgenic lines were established for each construct. The transcript levels of these transgenes were determined by real-time RT–PCR, and found to be approximately 1.5–17-fold higher than the level of the PCaP2 transcript in wild-type plants (Figure 6a).

Figure 6.

PCaP2 and its variants expressed under the control of the EXPA7 promoter caused abnormalities in root hair development.

(a) Expression levels of the transgenes in over-expressing lines. The mRNA levels of three independent transgenic lines for each construct (PCaP2, PCaP2G2A, Δ23PCaP2 or N23) were determined by real-time RT–PCR with a specific primer set for each construct. The mRNA levels are expressed as fold change compared with that of wild-type (WT). Values are means ± SE of three independent experiments.

(b) Frequency of branched root hairs in PCaP2, PCaP2G2A and ∆23PCaP2-expressing lines. Three independent transgenic lines for each construct were examined. More than 30 root hairs from each of 10 seedlings were analyzed for each line. Values are means ± SE of three independent experiments.

(c) Lengths of root hairs of N23-expressing lines. Three independent transgenic lines were examined. More than 20 root hairs from each of 10 seedlings were measured for each line. Values are means ± SE of three independent experiments.

(d–n) Root hair phenotypes of transgenic lines over-expressing PCaP2 (d, e), PCaP2G2A (f,g), ∆23PCaP2 (h–l) and of wild-type plants (m, n). Five-day-old seedlings were photographed. Scale bars = 20 μm.

(o–v) Root hair phenotypes of N23-expressing lines. Roots of N23-expressing lines #2–1 (p, s, v) and #4–12 (q, t) and of wild-type plants (o, r, u) are shown. Nine-day-old seedlings (o–t) and 5-day-old seedlings (u, v) were photographed. Scale bars = 500 μm (o−t) and 20 μm (u, v).

In all the established transgenic plant lines expressing the PCaP2, PCaP2G2A and ∆23PCaP2 constructs, root hairs with abnormal morphologies, including branches and bulbous structures, were frequently observed (Figure 6b,d–l). To statistically assess the frequencies of root hairs with abnormal morphologies, we counted total root hairs and those containing branched structures in wild-type and transgenic roots (see 'Ca2+-binding site in PCaP2'). The frequency of appearance of abnormal root hairs was significantly higher (>25%) in all examined lines for each construct than in the wild-type (<5%), but the frequency did not significantly vary between the constructs nor was it correlated with the expression level of each construct (Figure 6a,b).

However, the N23 construct conferred a different phenotype to the others. Seedlings of eight of the ten obtained transgenic lines had shorter root hairs than wild-type seedlings (Figure 6c,p,s), and those of the other two lines had no root hairs in the transit region of the root–hypocotyl boundary (Figure 6c,q,t), where wild-type seedlings form many root hairs (Figure 6o,q). A statistical analysis revealed that root hair lengths of root hair-containing lines were less than half of the wild-type on average (Figure 6c). The transcription level of the N23-coding transgene in a hairless line (#4–12) was over twice that in root hair-containing lines #2–1 and #6–9 (Figure 6a,c), suggesting that the severity of the phenotype caused by N23 is correlated with its expression level. Moreover, root hair-containing N23 lines exhibited single epidermal cells containing multiple bulges, which were never observed in the wild-type root epidermis under the conditions used in this study (Figure 6u,v).

Transgenic lines containing the GFP-tagged protein constructs PCaP2–GFP, PCaP2G2A–GFP, ∆23PCaP2–GFP and N23–GFP showed the same root hair phenotypes as caused by each corresponding construct without the GFP moiety (Figure S5). For all the constructs, the intracellular localization patterns observed in the root hair cells at the bulge-forming stage were sustained in root hairs at later stages. These facts support the idea that addition of the GFP moiety had no influence on the molecular functions of PCaP2 and its variants in root hair cells.


Functional importance of the N23 domain for interaction with PtdInsPs and Ca2+/CaM

To determine the role of PCaP2, we first focused on the N23 domain of PCaP2 and demonstrated that the region is necessary and sufficient for binding to PtdInsPs. This was clearly demonstrated using recombinant ∆23PCaP2 and synthetic N23 peptide in the PtdInsP binding assay (Figure 2). PCaP2 contains eight positively charged residues in the N23 domain (Figure 1). This polybasic cluster is thought to interact with the negatively charged PtdInsPs (McLaughlin and Murray, 2005; Heo et al., 2006). The present experiments revealed that the N–terminal basic region interacts with PtdInsPs, probably via electrostatic interaction with their phosphate moieties.

The N23 domain also has a key sequence for binding Ca2+/CaM. PCaP2 and N23 bound to CaM–agarose in the presence of Ca2+, but the mutated PCaP2 without N23 did not (Figure 3). With regard to this dual function of the N–terminal region, it should be noted that the presence of a Ca2+/CaM complex weakens the PCaP2–PtdInsP interaction (Kato et al., 2010). In a helical wheel model of the N–terminal part (Figure 3), five positive basic residues (Lys7, Arg11, Lys14, Lys18 and Lys22) are aligned at the same side. These positively charged residues may interact with PtdInsPs. Additionally, the arrangement of neutral residues at positions 1, 8 and 14 in an α–helix (1–8–14 motif) has been proposed as a candidate CaM-binding site (Rhoads and Friedberg, 1997). This motif exists in the N–terminal region and comprises Val8, Leu15 and Ala21 as the 1–8–14 motif (Figure 3). Binding of Ca2+/CaM may change the tertiary structure of PCaP2, as shown for PEP–19, a small CaM-binding protein of human Purkinje cells (Kleerekoper and Putkey, 2009). Therefore, Ca2+/CaM and PtdInsPs may competitively bind to this small N23 domain. This competition may be a key mechanism for signal transduction between the Ca2+ and PtdInsP signaling pathways.

We previously reported that PCaP2 is localized to the plasma membrane via N–myristoylation using A. thaliana cultured cells (Kato et al., 2010). The strict plasma membrane localization of PCaP2 was destroyed in lines expressing ∆23PCaP2–GFP (Figure 5). In PCaP2G2A–GFP lines, the protein was partly associated with the plasma membrane and was partly located in the cytoplasm. These data suggest that N–myristoylation and binding to PtdInsPs synergistically support plasma membrane targeting. Taken together, the N23 of PCaP2 may serve as a multi-functional domain that mediates plasma membrane localization and the interaction with the ligands as discussed below.

Candidates for PtdInsPs that interact with PCaP2

PCaP2 interacted with PtdIns(3,4)P2, PtdIns(3,5)P2, PtdIns(4,5)P2 and PtdIns(3,4,5)P3 (Figure 2). We propose that PtdIns(4,5)P2 is the species that interacts with PCaP2 in plants, because PtdIns(4,5)P2 is the most abundant PtdInsP2 and it predominantly accumulates in the plasma membrane (Meijer and Munnik, 2003). In contrast, PtdIns(3,4,5)P3 has not been found in plants (Mueller-Roeber and Pical, 2002; Michell, 2008). PtdIns(3,4)P2 has been found in the plasma membrane, but the enzymes responsible for its formation and it physiological roles are unknown (Mueller-Roeber and Pical, 2002). PtdIns(3,5)P2 accumulates in the lysosomes of animal cells and the vacuoles of plant cells (Gary et al., 1998).

In growing root hairs, PtdIns(4,5)P2 has been reported to be localized at the plasma membrane of the root hair tip (Vincent et al., 2005; van Leeuwen et al., 2007), and has been proposed to mediate the exocytosis of membrane vesicles that is essential for polar tip growth (Heilmann, 2009). Loss-of-function mutations of PIP5K3, a PtdIns(4,5)P2-generating enzyme localized in the plasma membrane of root hair tips, results in significantly shorter root hairs (Kusano et al., 2008; Stenzel et al., 2008). Furthermore, T–DNA insertion mutations of the PtdIns transfer protein AtSfh1p led to root hairs that are defective as a result of disruption of tip-focused PtdIns(4,5)P2 distribution, the Ca2+ gradient and cytoskeleton organization (Vincent et al., 2005). These observations indicate the importance of PtdIns(4,5)P2 in root hair development, and suggest that phenotypes of lines over-expressing PCaP2 and polypeptide containing the N23 domain are due to disorders in signaling through PtdIns(4,5)P2.

Involvement of PCaP2 in root hair development

We analyzed a T-DNA insertion mutant line, in which the PCaP2 transcription level was approximately 20% of the wild-type, and found that root hairs of the knockdown mutant line were significantly longer than those of the wild-type but their morphology was apparently normal (Figure 4, and Figures S3 and S4). Moreover, heterozygous mutant plants, in which the PCaP2 transcription level was approximately 60% of the wild-type, had root hairs with a mean length intermediate between the wild-type and the homozygous mutant. These facts suggest that PCaP2 acts as a negative and quantitative regulator of root hair tip growth. However, the mutant phenotype was not closely related to the phenotype of abnormal root hair morphologies observed in the PCaP2 over-expressing line. Because the morphological phenotype was not correlated with the level of over-expression and was observed even in a line expressing the PCaP2 transcript at a level approximately 1.5 times that of the wild-type, the morphological phenotype may be caused not by over-expression but by ectopic expression of PCaP2 using the EXPA7 promoter. As another possibility, the residual activity of PCaP2 in the T–DNA mutant may be enough to prevent effects on root hair morphogenesis. Although it would be interesting to assess effects of the loss of PCaP2 gene function on root hair development, it may be difficult to obtain knockout lines of the gene because the PCaP2 gene probably also functions in pollen tubes (Kato et al., 2010).

Root hairs of transgenic lines over-expressing PCaP2, PCaP2G2A, ∆23PCaP2 or their GFP-tagged proteins all showed aberrant polarity of tip growth (Figures 5 and 6). The fact that over-expression of ∆23PCaP2 or its GFP-tagged protein, which were mis-localized in the cytoplasm and did not bind PtdInsPs, also caused aberrant polarity of tip growth indicates that this phenotype is probably independent of PtdInsP signals on the plasma membrane. It has been reported that A. thaliana MAP18, which is identical to PCaP2, destabilizes cortical microtubules (Wang et al., 2007). It is possible that ectopic expression of PCaP2 and its variants caused the phenotype through destabilization of microtubules, because destabilization of microtubules in root hairs impairs directionality of growth and leads to formation of multiple growth points in a single root hair (Bibikova et al., 1999). The abnormal morphology may be caused through interaction of a proline, glutamate, valine and lysine-rich (PEVK) domain in the CC region with actin, because the PEVK domain, the structure of which is intrinsically disordered (Labeit and Kolmerer, 1995), interacts with actin in some cases (Linke et al., 2002; Nagy et al., 2004).

Lines over-expressing N23 or its GFP-tagged protein exhibited a totally different phenotype compared with that of ∆23PCaP2 (Figures 5 and 6). The major phenotype was inhibition of root hair tip growth, and its severity appeared to be correlated with the N23 expression level. This phenotype is probably reciprocal to the long root hair phenotype of the knockdown mutant. Given the PtdInsP binding ability and plasma membrane localization of N23, it is likely that N23 quantitatively suppresses root hair tip growth by binding to PtdInsPs on the plasma membrane, especially PtdIns(4,5)P2 as mentioned above. In support of this idea, the suppression phenotype caused by N23 was rescued by expression of a YFP fusion protein of PIP5K3 (Kusano et al., 2008) in F1 seedlings from a cross between an N23 line and a transgenic line expressing PIP5K3–YFP under the control of the PIP5K3 promoter (Figure S6). These facts strongly suggest that PCaP2 negatively regulates root hair tip growth through binding of the N23 domain to PtdIns(4,5)P2.

Working hypothesis and conclusion

Based on the findings in this study, we propose a model for the function of PCaP2 as a signal transducer between Ca2+ and PtdIns(4,5)P2 signals. In the model, PCaP2 binds to PtdInsPs in the plasma membrane at baseline [Ca2+]cyt. When [Ca2+]cyt is elevated in the root hair apical region, Ca2+ and CaM form a complex that stimulates dissociation of PtdIns(4,5)P2 from PCaP2, and then the free PtdIns(4,5)P2 promotes actin reorganization and membrane trafficking in the apex (Heilmann, 2009; Pei et al., 2012), and regulates particular ion channels and a proton pump in the tip plasma membrane (Liu et al., 2005). As the PtdIns(4,5)P2 signal is thought to be localized to the tip plasma membrane to stabilize the tip growth polarity, PCaP2, which is uniformly distributed on the plasma membrane, is thought to mask PtdIns(4,5)P2 in basal regions where [Ca2+]cyt is low.

In this study, we obtained genetic evidence for the involvement of PCaP2 in regulating root hair tip growth, and revealed biochemical and physiological properties of the domains in PCaP2. These results suggest that the N23 domain is involved in quantitative regulation of root hair tip growth via signal transduction between Ca2+ and PtdInsP signals on the plasma membrane, and that the residual CC domain is involved in regulating the polarity of cell expansion. Further study of PCaP2 is required to elucidate the molecular mechanism by which Ca2+ and PtdInsP signals are integrated into spatial signaling for root hair morphogenesis.

Experimental Procedures

Plant material and growth conditions

Arabidopsis thaliana strain Col–0 was used as the wild-type. Seeds of the T–DNA tagging mutant SALK_021652 were obtained from Arabidopsis Biological Resource Center (Columbus, OH), and a homozygous line containing a T–DNA insertion upstream of the PCaP2 gene was established. PCR primers used for the genotyping experiments are listed in Table S1. The seeds were surface-sterilized and germinated on sterile gel plates containing Murashige & Skoog (MS) salts, 2.3 mm MES/KOH (pH 5.8), 2% w/v sucrose and 1% Ina agar (Funakoshi, Seedlings were grown at 21°C under long-day conditions (light/dark regime of 16 h per 8 h, cool-white lamps, 90 μmol m−2 s−1).

Construction of transgenic plants

To construct PCaP2 variants under the control of the EXPA7 promoter, the 2261 bp upstream DNA fragment from the first Met of EXPA7 was amplified by PCR using the primers 5′-CACCATATTACTGAATTATCGATAC-3′ (forward) and 5′-TCTAGCCTCTTTTTCTTTATTCTTAGGG-3′ (reverse) (Ohashi et al., 2003). The amplified fragment was inserted into pENTR/D–TOPO (Invitrogen, to obtain EXPA7p/pENTR/D–TOPO. Four variants of PCaP2 were amplified by PCR, namely full-length PCaP2, PCaP2 with a substitution of Gly2 by alanine (PCaP2G2A), PCaP2 that lacked the N23 except for the first methionine (∆23PCaP2), and N23. The fragments were inserted into linearized EXPA7p/pENTR/D–TOPO vectors using a CloneEZ® PCR cloning kit (GenScript, EXPA7p::PCaP2 variant fragments were ligated into the binary vector pGWB401 (Nakagawa et al., 2007). PCR primers used for these procedures are listed in Table S1.

To construct a PCaP2–GFP fusion protein under the control of the EXPA7 promoter, EXPA7p::PCaP2 variant fragments (without the stop codon) were amplified by PCR. The resulting fragments were inserted into pGWB404 (Nakagawa et al., 2007). PCR primers used for amplification of EXPA7p::PCaP2 variants fragments are listed in Table S1.

All chimeric constructs were transformed into A. thaliana (Col–0) by the Agrobacterium tumefaciens-mediated floral dip method (Clough and Bent, 1998). After selection with kanamycin, T2 and T3 plants were used for physiological experiments.

F1 plants that express both N23 and PIP5K3–YFP coding genes were prepared by crossing N23 line 6–9 with the transgenic lines 4 and 31 expressing PIP5K3–YFP under the control of the PIP5K3 promoter (Kusano et al., 2008).

RNA preparation and mRNA quantification

Total RNA was prepared from 9-day-old whole plants using NucleoSpin® RNA Plant (Macherey-Nagel, RNA (1 μg) was converted into cDNA using an iScript cDNA synthesis kit (Bio–Rad, Real-time RT–PCR analysis was performed using an iCycler iQ real-time PCR system (Bio-Rad) using SYBR Green real-time PCR master mix (Toyobo, The primer sets used for real-time RT–PCR are listed in Table S1. The specificity of these primers was confirmed by PCR. Copy numbers of the products were calculated from the threshold cycles of triplicate real-time RT–PCR assays using standard curves. Relative mRNA contents were normalized to the ACTIN2 (ACT2) transcript.

Preparation of recombinant PCaP2

The full-length PCaP2 with a 6×His tag at the C–terminus (PCaP2) was prepared as described previously (Kato et al., 2010). N23 with a 6×His tag was synthesized by Operon Biotechnology (

∆23PCaP2 was constructed by directly amplifying pET23b/∆23PCaP2 by PCR using a pair of PCaP2-specific primers using pET/PCaP2 as the template. The primer set is listed in Table S1. The amplified DNA fragment was self-ligated and subjected to nucleotide sequencing. The expression vector was then introduced into the E. coli BL21 (DE3) strain (Novagen, Transformants were grown in LB broth for 3 h at 30°C after induction with 0.4 mm isopropyl-thio-β–d–galactoside. E. coli cells expressing ∆23PCaP2 were centrifuged at 6000 g, 4°C for 5 min and suspended in 20 mm Tris-acetate (pH 7.5) containing 20% v/v glycerol, 0.1 mg ml−1 DNase I, 0.2 mg ml−1 lysozyme, 10 mM β–mercaptoethanol and protein inhibitor cocktail (0.5 × Complete™, EDTA-free; Roche, The cells were disrupted by sonication for 12.5 min on ice. After removal of cell debris by centrifugation at 100 000 g for 30 min, the supernatant was applied to an Ni–NTA Superflow column (Qiagen, equilibrated with 20 mm imidazole, 20 mm Tris-acetate (pH 7.5), 20% v/v glycerol and 2 m NaCl. ∆23PCaP2 protein was eluted with 300 mm imidazole, 20 mm Tris-acetate (pH 7.5) and 2 m NaCl. PCaP2-enriched fractions were applied to a HiTrap Phenyl HP column (GE Healthcare) equilibrated with 20 mm Tris-acetate (pH 7.5) and 2 m NaCl. ∆23PCaP2 was recovered in the flow-through fraction. After desalting using a Sephadex G–25 gel filtration column (GE Healthcare), the protein was applied to a QAE–550C Toyopearl column (Tosoh, equilibrated with 10 mm Tris-acetate (pH 7.5), and eluted from the column using 10 mm Tris-acetate (pH 7.5) containing 200 mm NaCl. The fraction was dialyzed against 10 mm Tris-acetate (pH 7.5) for 6 h. The resulting purified ∆23PCaP2 protein was confirmed by SDS–PAGE.

Protein quantification, SDS–PAGE and immunoblotting

The protein concentration was determined using a BCA protein assay reagent kit (Pierce Biotechnology, PCaP2 and ∆23PCaP2 were subjected to SDS–PAGE in a 12.5% w/v polyacrylamide gel and N23 peptide was subjected to SDS–PAGE in a 15% w/v polyacrylamide gel (Atto, Immunoblotting was performed using anti-His tag antibody and horseradish peroxidase-coupled rabbit IgG and Western blotting detection reagents (GE Healthcare).


The GFP fluorescence was visualized using a confocal laser-scanning microscope (Fluoview FV1000-D, Olympus, using a set of BA473 (excitation) and BA505−550 (emission) filters. For imaging the plasma membrane, root hairs were treated with 1.6 μm FM4–64 and observed within 5 min. GFP and FM4–64 signals were observed simultaneously using an Axioplan2 microscope (Zeiss, equipped with a confocal laser-scanning unit (CSU–X1, Yokogawa, A 488 nm laser and a 499-529 nm emission filter for GFP, and a 561 nm laser and a 612–660 nm emission filter for FM4–64 were used. Images of root hairs were captured using a CCD camera (VB–7010, Keyence, and a microscope with a digital camera (BX51/DP72, Olympus).

Calcium–45 overlay assay

The 45Ca2+ overlay assay was performed as reported previously (Maruyama et al., 1984). Purified preparations of recombinant PCaP2, ∆23PCaP2 and N23 were blotted onto a polyvinylidene difluoride membrane using a slot blot apparatus (Bio–Rad). The membrane sheet was washed twice with 10 mm MES/KOH (pH 6.5), 5 mm MgCl2 and 60 mm KCl at 50 rpm at 25°C, and incubated in the same buffer (1 ml) supplemented with 1 mm CaCl2 and 3.7 MBq of 45Ca2+ as CaCl2 (Perkin-Elmer, at 25°C for 30 min. The membrane was subsequently washed twice in 10 ml of 50% v/v ethanol, and dried at room temperature. An autoradiogram of the 45Ca2+-labeled proteins on the membrane was obtained by exposure to X–ray film for 6 days at –80°C.

Lipid-binding assay

The lipid-binding assay was performed using PIP strips™ and a PIP array™ (Echelon Biosciences, as described previously (Kato et al., 2010). The sheets were treated with 10 mm Tris/HCl (pH 8.0), 150 mm NaCl and 0.1% w/v Tween–20 (TBST) supplemented with 3% fatty-acid free BSA for 1 h at 25°C for blocking, and then incubated for 16 h with 100 or 500 ng ml−1 recombinant PCaP2 variants in TBST at 4°C. The sheets were then washed with TBST and immunoblotted using anti-His tag antibody. The blots were subsequently visualized using horseradish peroxidase-coupled rabbit IgG (Nacalai Tesque, and Western blotting detection reagents (Nacalai Tesque).

Calmodulin binding assay

CaM binding was examined using CaM–agarose (Sigma-Aldrich, Recombinant PCaP2, ∆23PCaP2 and N23 (3.9 μg) were incubated with 30 μl CaM–agarose suspended in 20 mM Tris/HCl (pH 7.5), 150 mm NaCl and 0.5 mm CaCl2 for 1 h at 4°C (Matsubara et al., 2004). In some experiments, buffer containing 2 mm EGTA instead of 0.5 mm CaCl2 was used. After centrifugation at 1470 g for 2 min, the supernatant was removed. The CaM–agarose was washed three times with the same buffer, and then proteins bound to CaM–agarose were eluted using SDS sample buffer containing 0.5% SDS. The bound fractions were subjected to SDS–PAGE and subsequent immunoblotting.

Statistical analyses of root hair length and morphology

Images of the root surface were captured with a CCD camera. For root hair length, root hairs horizontally elongating in the focus plane were measured on the image. For root hair morphology, total root hairs and branched root hairs visible on the image were counted. Immature root hairs or bulges with a length less than 100 μm were omitted. Root hairs with a branch longer than 20 μm were counted as branched root hairs.


We are grateful to Tsuyoshi Nakagawa (Center for Integrated Research in Science, Shimane University, Japan) for providing the Gateway vectors, Katsuhiro Shiratake (Graduate School of Bioagricultural Sciences, Nagoya University, Japan) for his help with observations using the Keyence CCD camera. This work was supported partly by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan to M.M., a Grant-in-Aid for Scientific Research on Innovative Areas number 23119511 to T.A., and a grant from the Salt Science Foundation to M.M.