A Sec14p-nodulin domain phosphatidylinositol transfer protein polarizes membrane growth of Arabidopsis thaliana root hairs

Sec14p, the major yeast phosphatidylinositol (PtdIns) transfer protein (PITP), regulates an essential interface between lipid metabolism and protein transport from Golgi membranes to the cell surface (Bankaitis et al., 1990; Cleves et al., 1991a,b; Xie et al., 1998). This regulatory circuit, termed the Sec14p pathway, defines a signaling cascade that involves functionally uncharacterized proteins of the oxysterol binding protein family such as Kes1p (Fang et al., 1996; Li et al., 2002) and ADP-ribosylation factor GTPase activating proteins (Yanagisawa et al., 2002). These regulatory interactions couple lipid signaling to the function of core components of the vesicle trafficking machinery, but precisely how this occurs remains unknown.

Mammals express at least three soluble PITPs: PITPα, PITPβ, and rdgBβ; and all of these share primary sequence homology to each other (for review see Routt and Bankaitis, 2004). The mammalian PITP module is found throughout metazoans and is structurally unrelated to yeast PITPs (Sha et al., 1998; Yoder et al., 2001). Gene ablation experiments in mice, although suggesting an essential housekeeping function for PITPβ, demonstrate that PITPα nullizygosity results in chylomicron retention disorder, severe hypoglycemia, and a fulminating spinocerebellar neurodegenerative disease (Alb et al., 2002, 2003). As at least some forms of human chylomicron retention disease are caused by null mutations in the Sar1b GTPase that regulates coassembly of COPII coat components with ER cargo (Jones et al., 2003), PITPα is suggested to regulate a Sar1b-GTPase activating protein function on the enterocyte ER surface in the chylomicron biogenic pathway (Bankaitis et al., 2004). Indeed, the hypothesis for PITPα function in chylomicron trafficking shares basic features with that proposed for Sec14p function in yeast (Yanagisawa et al., 2002) and leaves open the possibility that structurally disparate PITPs nonetheless operate via similar mechanisms in regulating analogous membrane trafficking reactions.

Although PITPs exist in higher plants (Jouannic et al., 1998; Kearns et al., 1998a), there has been no systematic functional analysis of them. Herein, we describe a large and novel family of Sec14p-nodulin domain proteins in Arabidopsis thaliana. We show that the Sec14p domains of these proteins share functional properties consistent with those of Sec14p-like PITPs and report the first analysis of the biological function of any Sec14p-like protein in plants. We demonstrate that loss of AtSfh1p, a membrane-bound Sec14p-nodulin protein, dramatically compromises polarized root hair membrane trafficking. Derangement of polarized membrane growth occurs after the site of root hair emergence has been correctly determined and emergence initiated. The collective data suggest AtSfh1p generates phosphoinositide (PIP) landmarks that focus membrane delivery to the root hair tip plasma membrane in a manner that depends on the actin cytoskeleton. The results further suggest that the polarized secretory pathway establishes a tip-directed Ca²⁺ gradient that cues microtubule (MT) organization in a manner that further reinforces tip-directed membrane trafficking. The collective data describe the functional characterization of the role for a novel membrane-associated PITP in execution of developmentally regulated polarized membrane trafficking pathway.

A search of the National Center for Biotechnology Information and The Arabidopsis Information Resource databases identified 31 homologous A. thaliana sequences when the Sec14p primary sequence was queried. These sequences each exhibit a 239-residue domain that shares significant primary sequence homology with the Sec14p lipid binding domain (LBD), and these domains fall into two Sec14p homology groups. One Sec14p homology group encodes proteins that consist of a Sec14p domain that shares rather low (but significant) sequence identity with yeast Sec14p. The other Sec14p homology group consists of the 14 highest-scoring Sec14p LBD-like sequences. 11 of these 14 sequences represent proteins where an NH₂-terminal Sec14p domain is joined to a COOH-terminal nodulin domain of ∼100 residues (Fig. 1 A). These nodulin domains define three classes based on similarity to the Nlj16 nodulin (Fig. 1 B). Nlj16 is expressed by root cells of the legume Lotus japonicus upon infection with Rhizobium and defines a plasma membrane targeting module (Kapranov et al., 2001). Lotus expresses four Sec14p-nodulin domain genes (LjPLP-I thru LjPLP-IV; Fig. 1 B). The Sec14p-nodulin domain proteins are of special interest because of the unanticipated physical linkage of Sec14p and nodulin domains, and because these define unique examples of membrane-bound Sec14p-like PITPs.

Expression in yeast of AtSfh1p-, AtSfh2p-, AtSfh4p-, or AtSfh6p-LBDs rescued growth defects associated with sec14-1^ts (Fig. 2 A) and haploid-lethal sec14Δ alleles (not depicted). These results were scored in phospholipase D (PLD)–proficient (SPO14) or –deficient (spo14Δ) genetic backgrounds. PLD deficiencies exacerbate Sec14p defects, and assessment of rescue in both SPO14 and spo14Δ genetic backgrounds reports quality of rescue. As an example, expression of AtSFH19 (AT5G47730.1) or AtSFH20 (AT1G01630.1) (i.e., representatives of the second Sec14p homology group) rescued sec14-1^ts alleles in SPO14 but not spo14Δ yeast strains (Fig. 2 A). Rescue of sec14 growth defects by AtSfh1p-, AtSfh2p-, AtSfh4p-, or AtSfh6p-LBDs extended to restoration of invertase secretion from Sec14p-deficient Golgi membranes (Fig. 2 B) and normal morphology to Sec14p-deficient cells (Fig. 2 C). The toroid structures observed in sec14-1^ts cells incubated at restrictive temperature represent defective Golgi compartments engorged with secretory cargo. AtSfh1p-LBD expression restores wild-type morphology to >90% of sec14-1^ts cells (Fig. 2 C).

AtSfhp-LBDs exhibit intrinsic PITP biochemical activities. AtSfh1p- and AtSfh2p-LBDs catalyzed phosphatidylcholine (PtdCho)- and PtdIns-transfer in vitro (Fig. 2 D). Significance of the PtdIns-transfer activities of AtSfh1p- and AtSfh2p-LBD was confirmed in measurements of PIP synthesis in a Sec14p-deficient yeast strain with basal PIP levels that expressed each LBD. In these experiments, the appropriate yeast strains were radiolabeled to steady-state with myo-[2-³H] inositol. Bulk PIPs were extracted, deacylated to yield the corresponding glycerophosphoinositol derivatives of extracted PIP species, and the water-soluble glycerophosphoinositols were chromatographically resolved and quantified (see Materials and methods). AtSfh1p-LBD expression elicited significant elevations in steady-state PtdIns-3-phosphate, PtdIns-4-phosphate, and PtdIns-4,5-P₂ levels in the yeast tester strain compared with mock control (Fig. 2 E). These elevations in bulk PtdIns-4-phosphate and PtdIns-4,5-P₂ content were of the magnitude recorded for the YEp(SEC14) positive control. AtSfh2p-LBD expression also elevated bulk PIP levels (unpublished data).

Sec14p-nodulin domain proteins are uncharacterized and expansion of this family suggests tissue-specific functions for its members. AtSfh1p was chosen for detailed analysis because the AtSfh1p-LBD is most homologous to Sec14p. RT-PCR analyses indicated essentially root-specific expression of AtSFH1 (unpublished data), a result in accord with microarray data (http://www.cbs.umn.edu/arabidopsis/). β-Glucuronidase (GUS) histochemical staining confirmed and extended these results. AtSFH1 was expressed solely in root trichoblast cell files engaged in root hair growth (Fig. 3, C–E), hydathodes, shoot apical meristem, and apical cells of the root cap (Fig. 3, F–H).

Atsfh1::T-DNA plants (Fig. 4 A) were substantially normal and fertile. However, mutant plants elaborated short root hairs. Mutant single root hairs from 3-d-old seedlings were one-third the length of age-matched wild-type structures (69 ± 13 vs. 224 ± 51 μm, n = 55) and exhibited half the surface area (3494 ± 902 vs. 6924 ± 1045 μm², n = 55). Mutant and wild-type single root hairs exhibited similar volumes (14866 ± 5550 vs. 17113 ± 3510 μm³, n = 55), as did double root hairs (unpublished data). These disparities persisted throughout the lifetime of the plant (Fig. 4 B). In addition, Atsfh1::T-DNA root hairs appeared flaccid. This observation was in contrast to the rigid profiles exhibited by age-matched wild-type root hairs (Fig. 4 C). Although there is robust AtSFH1 expression in apical cells of the root cap, Atsfh1::T-DNA primary roots exhibit no defects in gravitropism (unpublished data).

Reduced lengths of Atsfh1::T-DNA root hairs reflect failures in polarized membrane growth. High frequencies of Atsfh1::T-DNA root hairs with two growing tips were observed and a significant number exhibited three. Wild-type plants rarely elaborated two growing tips, and we never observed any with three (Fig. 4, C and D). Double root hairs exhibited volumes similar to those of wild-type root hairs, but only 50% of the surface area. The data indicate AtSfh1p-deficient root hairs exhibit impaired secretion efficiency and failure in restricting active growing points to a single site that leads to isotropic root hair cell expansion. Yet, Atsfh1::T-DNA plants correctly specified site of root hair emergence from the distal plasma membrane of parent epidermal cells (Fig. 4 E). Moreover, consistent with the tissue restriction of AtSFH1 expression, nullizygous plants did not exhibit obvious defects in other organs (such as leaf trichomes and pollen tubes) whose development requires polarized membrane trafficking (unpublished data).

Four lines of evidence demonstrate the full spectrum of Atsfh1::T-DNA root hair phenotypes results from a single fully penetrant recessive mutation. First, cross of Atsfh1::T-DNA homozygotes to wild-type plants yielded only wild-type progeny. Second, the root hair phenotype cosegregated with Atsfh1::T-DNA through multiple (>3) backcrosses. Third, transgenic Atsfh1::T-DNA plants bearing ectopic AtSFH1 exhibited normal root hairs (Fig. 4 F). Four, examination of 2947 F2 progeny from three independent Atsfh1::T-DNA/Atsfh1::T-DNA X AtSFH1/AtSFH1 crosses yielded 701 mutant (23.8%) and 2,246 (76.2%) wild-type phenotypes, respectively. To assess the functional importance of the Sec14p domain, we generated an NH₂-terminal GFP-fusion to the AtSfh1p-LBD that inactivates Sec14p domain activities. GFP-AtSfh1p, when placed in the context of full-length AtSfh1p and expressed in plants, fails to complement Atsfh1::T-DNA (Fig. 4 F). Similarly, a COOH-terminal GFP-fusion that preserves Sec14p domain function, but abuts the nodulin domain, also fails to complement Atsfh1::T-DNA (unpublished data). Thus, both functional Sec14p and nodulin domains are critical for AtSfh1p function in plants.

The AtSfh1p nodulin domain exhibits high primary sequence identity to the Nlj16 nodulin. As Nlj16 functions as a plasma membrane targeting domain (Kapranov et al., 2001), we expected AtSfh1p would also localize to membranes. Consistent with expectation, the GFP-AtSfh1p chimera (with the caveat that it harbors a nonfunctional Sec14p domain) distributed in an apex-directed spiraling arrangement along the root hair cortical plasma membrane in otherwise wild-type plants (Fig. 5 A, top left; and Video 1). Optical cross sections taken through the root hair at positions removed from the apex also indicated a plasma membrane localization for GFP-AtSfh1p (Fig. 5 A, top right). Optical sectioning of the apex plasma membrane at the root hair tip reported a clear enrichment of GFP-AtSfh1p staining on the plasma membrane at that site as well (Fig. 5 A, bottom panel). That this profile reflects plasma membrane staining was confirmed in FM1-43 double label experiments. Under conditions where FM1-43 selectively labels plasma membrane, FM1-43 and GFP-AtSfh1p staining were coincident (unpublished data). Strong enhancement of GFP-AtSfh1p reporter fluorescence was also recorded in the tip cytoplasm (Fig. 5 A, top left and bottom panel; and Video 1). For the reasons detailed in the section Ultrastructure of the Atsfh1 tip cytoplasm, we interpret this staining to reflect an AtSfh1p pool that is localized on post-Golgi vesicles. Expression of GFP or YFP alone gave diffuse staining (Figs. S1 and S2).

The ability of AtSfh1p-LBD to stimulate PIP synthesis suggests AtSfh1p mediates PIP-dependent regulation of polarized membrane transport in A. thaliana. Because polarized membrane transport in yeast and plants requires involvement of the actin cytoskeleton and actin dynamics are responsive to PtdIns(4,5)P₂, we focused on the role of AtSfh1p in modulating PtdIns(4,5)P₂ homeostasis. Using a phospholipase Cδ1 (PLCδ1) pleckstrin homology (PH) domain–YFP reporter to infer PtdIns(4,5)P₂ status, we found wild-type root hairs exhibited a tip-directed (4,5)P₂ gradient on the root hair plasma membrane. Indeed, PtdIns(4,5)P₂ was distributed in a pattern similar to that recorded for GFP-AtSfh1p in wild-type root hairs. Discrete PtdIns(4,5)P₂–enriched domains were also recorded along the cortical root hair plasma membrane (Fig. 5 B, top left). Pseudocolor rendering of PH_PLCδ1–YFP fluorescence revealed a tip-directed spiraling arrangement along the cortical root hair plasma membrane (Fig. 5 B, top right and bottom left; and Video 2), and optical sectioning of the tip plasma membrane revealed a high concentration of PtdIns(4,5)P₂ at the very apex of the root hair (Fig. 5 B, bottom right). Again, FM1-43 double-labeling experiments confirmed the plasma membrane localization of the PH_PLCδ1–YFP reporter fluorescence (see the following paragraph). Strikingly, as was observed for AtSfh1p-GFP, optical cross sections of the root hair apex demonstrated strong enhancement of PH_PLCδ1–YFP reporter fluorescence in the tip cytoplasm.

Given the ability of AtSfh1p-LBD to stimulate PIP synthesis, we anticipated Atsfh1::T-DNA root hairs would exhibit PIP deficiencies. We focused on PtdIns(4,5)P₂ as this phospholipid is an established regulator of polarized membrane trafficking. Indeed, the tip-directed PtdIns(4,5)P₂ gradient was compromised, and the prominent cytoplasmic PH_PLCδ1–YFP reporter fluorescence was absent from mutant tip cytoplasm (Fig. 5, C and D; and Videos 3 and 4). That PtdIns(4,5)P₂ is enriched at the tip plasma membrane of wild-type root hairs, and that this PtdIns(4,5)P₂ enrichment is lost in mutant tip plasma membrane, was indicated by ratiometric imaging of PH_PLCδ1–YFP/FM1-43 fluorescence in double-label experiments. From those experiments, we record a threefold tip enrichment of PtdIns(4,5) P₂ in wild-type root hair tip plasma membrane relative to cortical plasma membrane. We detected loss of PtdIns(4,5)P₂ tip enrichment in the Atsfh1 tip plasma membrane relative to wild-type and estimate a 5–10-fold reduction in relative PtdIns(4,5) P₂ in mutant tip plasma membrane (Fig. 5 E). No obvious PtdIns(4,5) P₂ deficiencies were recorded in the cortical plasma membrane of Atsfh1::T-DNA root hairs.

Loss of PH_PLCδ1–YFP fluorescence staining in the tip cytoplasm of AtSfh1p-deficient root hairs was a striking phenotype. Because incubation of metabolically active wild-type root hairs for 20 min with FM1-43 yielded robust staining of the tip cytoplasm, in a fashion that recapitulated the pattern of GFP-AtSfh1p and PH_PLCδ1–YFP tip fluorescence (unpublished data), we interpreted PH_PLCδ1–YFP fluorescence in tip cytoplasm to reflect the status of small membrane-enclosed structures in this region. In this regard, tip cytoplasm is enriched for post-Golgi secretory vesicles and is referred to as the vesicle-rich zone (VRZ; Braun et al., 1999). Therefore, we inspected wild-type and mutant root hair apices by electron microscopy. Wild-type and Atsfh1::T-DNA Golgi stack ultrastructures were indistinguishable in appearance (Fig. 5 F) and in physical parameters (Table I). The structural integrity of Atsfh1::T-DNA Golgi membranes suggested these systems are not grossly defective in secretory function. Golgi stack distribution throughout the cytoplasm was also unaffected in the mutant root hairs (Fig. S3). However, two unusual properties of mutant tip cytoplasm were apparent. First, the concentration of vesicles per unit tip cytoplasm was reduced sixfold in mutant relative to wild-type VRZ (Fig. 5, G and H). We interpret this result, and the loss of tip cytoplasm fluorescence as recorded by the GFP-AtSfh1p and PH_PLCδ1–YFP reporters (see the section Localization of AtSfh1p in developing root hairs), to indicate a dispersal of vesicles from the VRZ throughout the mutant root hair cytoplasm. Second, dramatic vacuolation of the mutant VRZ and tip cytoplasm was observed (Fig. 5 G).

Vacuolation of Atsfh1::T-DNA root hair tip cytoplasm recapitulated the effects recorded in root hairs where the actin cytoskeleton was disrupted (Čiamporová et al., 2003). To assess f-actin status in Atsfh1::T-DNA root hairs, we used GFP-talin as a reporter for f-actin. These imaging experiments indicated that wild-type root hairs exhibit a discrete cortical actin meshwork and a tip-concentrated f-actin microfilament network in the VRZ (Fig. 6 A and Video 5), as described previously (Baluška et al., 2000; Smith, 2003). However, selective defects in the f-actin cytoskeleton were apparent in mutant root hairs. Although the cortical actin cytoskeleton of mutant root hairs remained intact, the tip-directed f-actin microfilament component was lost (Fig. 6, B and C; and Videos 6 and 7). As tip f-actin microfilament networks focus transport vesicle delivery to the hair apex (Mathur and Hülskamp, 2002; Ketelaar et al., 2003), defects in this actin network are expected to result in dispersal of transport vesicles throughout the root hair cytoplasm. Indeed, loss of the f-actin microfilament network in Atsfh1::T-DNA root hairs coincides with loss of PH_PLCδ1–YFP fluorescence in the tip cytoplasm.

One critical contributing cue that controls polarized root hair growth is a tip-directed Ca²⁺ gradient. Defects in polarized membrane trafficking are expected to randomize ion (e.g., Ca²⁺) channel delivery to the plasma membrane, with the consequence that spatial regulation of Ca²⁺ signaling will be compromised. To investigate whether or not dysregulation of Ca²⁺ signaling was occurring in Atsfh1::T-DNA root hairs, we used Indo-1 loading strategies to image Ca²⁺ signaling in wild-type and isogenic Atsfh1::T-DNA root hairs. Striking derangements in Ca²⁺ signaling to the growing apex of the root hair plasma membrane were observed (Fig. 7 A). Ca²⁺ imaging recorded a single tip-focused Ca²⁺ gradient in wild-type root hairs, as reported previously (Wymer et al., 1997), with tip cytoplasmic Ca²⁺ reaching concentrations in excess of 600 nM. Cytoplasmic Ca²⁺ fell to concentrations below 100 nM very rapidly away from the root hair tip. In marked contrast, precocious Ca²⁺ signaling was evident along the Atsfh1 root hair cortical plasma membrane with local Ca²⁺ concentrations reaching 600 nM or greater (Fig. 7 A).

One potential mechanism for spatial dysregulation of Ca²⁺ signaling is the isotropic delivery or distribution of active Ca²⁺ channels in the mutant root hair surface. To test this prediction, we used the scanning ion-selective electrode technique (SIET) to monitor Ca²⁺ fluxes along the root hair surface in wild-type and Atsfh1 nullizygous seedlings (see Materials and methods). Inward Ca²⁺ fluxes exceeding 4 pmol cm⁻² s⁻¹ were recorded around the apex region of growing wild-type root hairs, and Ca²⁺ influx decreased dramatically as the self-referencing probe was positioned away from the root hair apex (Fig. 7, B and C). The flux profiles for Atsfh1 nullizygous single root hairs were dramatically altered, however. Large inward Ca²⁺ fluxes were not restricted to the tip region. Rather, robust fluxes exceeding 4 pmol cm⁻² s⁻¹ were detected all along the root hair surface. These flux profiles are completely congruent with the Indo-1 Ca²⁺ imaging data demonstrating well-defined tip-directed cytoplasmic Ca²⁺ gradient in wild-type, but not in Atsfh1 nullizygous, root hairs.

In addition to the f-actin cytoskeleton, MT networks also function in maintaining polarized tip growth (Bibikova et al., 1999; Baluška et al., 2000; Stevenson et al., 2000; Smith, 2003). MTs appear to consolidate the results of polarized membrane deposition, and MT action in supporting tip growth is spatially regulated by Ca²⁺ gradients (Bibikova et al., 1999; Baluška et al., 2000; Smith, 2003). To assess whether or not the derangements in Ca²⁺ signaling observed in Atsfh1::T-DNA root hairs coincided with functional derangement of MT systems, we probed root hair MT organization in AtSfh1p-deficient root hairs by GFP-MAP4 imaging (Marc et al., 1998). As reported by Smith (2003), cortical MTs were organized into discrete filaments. These MT filaments were arranged in spiraling profiles parallel to the longitudinal axis of the cortical plasma membrane in wild-type root hairs (Fig. 8 A and Video 8). In contrast, Atsfh1 root hairs exhibited only diffuse GFP-MAP4 staining profiles (Fig. 8, B and C; and Videos 9 and 10). Neither discrete filaments nor obvious spiraling profiles were seen. However, the body of the trichoblast from which the mutant root hair emanates did exhibit organized MTs (Fig. 8 C and Video 10). Thus, Atsfh nullizygous root hairs elaborate defects in MT assembly and/or organization that are limited to the growing root hair itself.

Root hair development requires polarized membrane growth from a precise position on the root epidermal cell plasma membrane. Herein, we demonstrate that loss of AtSfh1p, a PtdCho- and PtdIns-binding/transfer protein with the ability to regulate PIP metabolism, deranges root hair growth. AtSfh1p dysfunction compromises tip-directed plasma membrane PtdIns(4,5)P₂ and Ca²⁺ gradients, elicits tip actin defects, and disorganizes root hair MT networks. The result is a derangement of polarized membrane growth after the site of root hair emergence has been correctly specified. We propose the primary function of AtSfh1p is to generate PIP landmarks that couple to components of the f-actin cytoskeleton, thereby focusing membrane delivery to the root hair tip plasma membrane. The polarized secretory pathway restricts insertion of cargo (e.g., ion channels) to the root hair apex, thereby establishing a tip-directed Ca²⁺ gradient. This gradient cues an organized MT assembly that further reinforces and maintains tip-directed membrane trafficking (Fig. 9).

The data presented emphasize AtSfh1p-mediated regulation of PIP metabolism. What is the mechanism that underlies such regulation? One simple mechanism is that AtSfh1p directly couples its intrinsic AtSfh1p-LBD PtdIns binding/transfer activity to PtdIns kinase action. By this model, AtSfh1p generates PIP signaling pools that regulate the action of multiple effector proteins that couple to actin dynamics and the activities of signaling enzymes (such as PLD). With regard to PLD, an alternative possibility is that AtSfh1p couples its intrinsic PtdCho-binding/transfer activity to PIP synthesis in the plant. In this regard, single or multiple PLD isoforms represent attractive candidates for AtSfh1p effectors. PLDs are PtdIns(4,5)P₂-activated PtdCho hydrolases that generate phosphatidic acid (PtdOH), itself a lipid stimulator of PtdIns-4-phosphate 5-kinase. By presenting PtdCho to PLD, AtSfh1p may initiate a robust positive feedback loop that links PLD activity to PIP synthesis via PtdOH signaling. Obviously, PIP signaling mediated by PtdIns-bound AtSfh1p could cooperate with such a regulatory loop. In support of an AtSfh1p-PLD coupling model, Ohashi et al. (2003) found that PLDζ1 activity regulates root hair growth, and that, like AtSfh1p and PtdIns(4,5)P₂, PLDζ1 is enriched in the tip cytoplasm of growing root hairs. Those results suggest a role for PLDζ1, and perhaps other PLD isoforms, in root hair morphogenesis. A general precedent for a PITP-PLD coupling also exists in yeast, where nonclassical Sec14p-like PITPs function to optimally activate PLD. In that case, PITPs do so by stimulating PIP synthesis in the absence of direct PITP-regulated PtdCho signaling (Xie et al., 1998; Li et al., 2000).

How might AtSfh1p-stimulated PtdIns(4,5)P₂ synthesis interface with the actin cytoskeleton and membrane trafficking? PtdIns(4,5)P₂ synthesis may recruit actin to the Golgi surface and modulate its assembly in a polymerization reaction that potentiates vesicle budding. Evidence for an actin involvement in vesicle formation from mammalian Golgi membranes has been reported (Fucini et al., 2000). However, as neither Golgi morphology nor Golgi distribution is perturbed in Atsfh1::T-DNA root hairs, we conclude that the membrane trafficking defects occur at a post-Golgi stage. We speculate AtSfh1p stimulates PLD activity, and ultimately PtdIns(4,5)P₂ synthesis, on formed (or forming) secretory vesicles. Such a regulatory loop promotes an on-demand PtdIns(4,5)P₂-driven actin polymerization on the transport vesicles and engages nascent vesicles with an f-actin pool that imposes polarized trafficking of those post-Golgi vesicles to the root hair tip plasma membrane. It follows that defects in such a lipid signaling program would compromise a specific f-actin component dedicated to vesicle trafficking. The consequence is imposition of kinetic and polarity defects on membrane trafficking to the Atsfh1::T-DNA root hair plasma membrane. GTPases of the Rac/Rho/Cdc42 family, and actin binding proteins (e.g., profilin), are attractive candidates as downstream effectors of AtSfh1p-mediated lipid signaling (Braun et al., 1999; Molendijk et al., 2001). It remains possible, perhaps likely, that AtSfh1p sits at the nexus of more complicated lipid signaling cascades. For example, PtdOH modulates PtdIns 3-OH kinase signaling in polar root tip growth (Anthony et al., 2004).

We further suggest that highly polarized membrane deposition at the root hair plasma membrane of wild-type plants sets the tight Ca²⁺ tip gradient by restricting the distribution of functional Ca²⁺ channels to a focused site on the tip plasma membrane. The net effect is a tip-restricted mode of Ca²⁺ entry into the developing root hair from the extracellular milieu. Our demonstration that nullizygous root hairs engage in precocious and isotropic Ca²⁺ entry across the plasma membrane is consistent with delocalized Ca²⁺ channel distribution. We suggest this is a direct consequence of isotropic fusion of post-Golgi vesicles to the root hair plasma membrane. We also note the observed derangements in Ca²⁺ signaling are not consistent with a role for PtdIns(4,5)P₂-specific phospholipase C–mediated generation of IP₃ in the gating of Atsfh1 nullizygous root hair plasma membrane Ca²⁺ channels. Were IP₃-gated Ca²⁺ channels involved in generating the cytoplasmic Ca²⁺ gradients, reductions in Ca²⁺ fluxes would have been expected, not the robust and isotropic Ca²⁺ influxes recorded. As the Ca²⁺ tip gradient then cues appropriate organization of the root hair MT cytoskeleton so that polarized membrane delivery to the root hair apex is further reinforced and consolidated (Bibikova et al., 1999), spatial derangement of Ca²⁺ signaling accounts for the lack of organized cortical MT assembly in mutant root hairs.

Herein, we identify AtSfh1p as a key regulator of polarized membrane trafficking in root hairs. A corollary to these findings is that AtSfh1p, or other members of the Sec14p-nodulin domain family, serve as attractive targets for intervention when root hair membrane growth is naturally reoriented, i.e., as occurs in legumes in response to Rhizobium NOD factors. The regulation of a Sec14p-nodulin domain protein (LjPLP-IVp) during nodulogenesis in the legume Lotus japonicus provides an interesting case in point. LjPLP-IVp is the Lotus orthologue of AtSfh1p (Fig. 1, A and B). During nodulation, the promoter driving full-length LjPLP-IV transcription is silenced and an internal bidirectional promoter is activated. The result is high level expression, in nodules, of the Nlj16 nodulin and of antisense transcripts directed against the LjPLP-IVp-LBD coding region (Kapranov et al., 2001). We suggest LjPLP-IV transcriptional reprogramming during nodulogenesis is designed to efficiently subvert the normal highly polarized root hair growth program by three converging mechanisms that functionally inactivate a master polarity regulator (LjPLP-IVp). First, formation of new transcripts encoding a functional LjPLP-IVp is terminated. Second, production of antisense RNAs directed against the LjPLP-IVp-LBD coding region silence existing full-length LjPLP-IVp transcripts encoding. Third, residual LjPLP-IVp activity is suppressed via high level expression of a dominant-interfering plasma membrane–targeting module that represents the nodulin itself.

Finally, our functional characterization of AtSfh1p raises the intriguing possibility that Sec14p-nodulin domain proteins define a family of polarized membrane growth regulators in plants. We suggest individual members of this family are dedicated to specific polarity establishment events such as those involving organogenesis and control of intracellular organelle morphogenesis. Because the mammalian genome encodes multiple uncharacterized Sec14p domain proteins, Sec14p domain proteins may represent conserved features of lipid-signaling mechanisms that control polarized membrane biogenic programs in eukaryotic cells.

Sequences were analyzed by BLAST (http://www.ncbi.nlm.nih.gov/BLAST/ and http://arabidopsis.org/Blast/). Sequence alignments were generated in ClustalX and multiple alignments were shaded with BOXSHADE 3.21 (http://www.ch.embnet.org/software/BOX_form.html).

Strains included are as follows: CTY182 (MATa ura3-52 lys2-801 Δhis3-200), CTY1-1A (CTY182 sec14-1^ts), CTY1079 (CTY1-1A spo14Δ::HIS3), and CTY303 (MATa ura3-52 lys2-801 Δhis3-200 Δsec14, cki1::HIS3) (Cleves et al., 1991b; Phillips et al., 1999; Li et al., 2000; Yanagisawa et al., 2002).

Yeast media, genetic techniques, invertase assays, EM, phospholipid transfer assays, and PIP determinations have been described previously (Kearns et al., 1998b; Guo et al., 1999; Phillips et al., 1999; Li et al., 2000; Yanagisawa et al., 2002). Site-directed mutagenesis used QuickChange (Stratagene). Primers were obtained from the University of North Carolina Lineberger Comprehensive Cancer Center Oligonucleotide Synthesis Core.

100 μg of total mRNA was prepared from 100 mg of 3-d-old seedlings using the RNeasy Plant Mini Kit (QIAGEN). The 717-bp mRNAs for each AtSFH-LBD were amplified for RT-PCR (SuperScript First-Strand Synthesis System; Invitrogen). BamHI and HindIII-restricted cDNAs were cloned into the corresponding sites of a yeast episomal plasmid derived from YEplac195 (http://genome-www2.stanford.edu/vectordb/vector_descrip/COMPLETE/YEPLAC195.SEQ.html). AtSFH-LBD expression was driven by a SEC14 promoter and subject to SEC14 termination signals.

5-d-old seedlings were stained for GUS activity using standard protocols (Jefferson et al., 1987). The GUS gene was placed under the control of the AtSFH1 promoter (P_AtSFH1) and transgenic lines were generated by Agrobacterium-mediated transformation of wild-type plants using the floral dip method. P_AtSFH1::GUS expression was recorded after staining under vacuum for 5 min at 25°C followed by 1 h at 37°C. P_AtSFH1 represented a 1958-bp DNA fragment directly 5′ to the AtSFH1 initiator codon.

Light microscopy was done with a microscope (model MZFLIII; Leica) using a cooled CCD camera (model EOSD30; Canon) interfaced with capture image software (RemoteCapture 1.1; Canon), a dissecting microscope (model SMZ-U; Nikon), or a Microphot microscope (Nikon) interfaced with a color CCD camera (model DXM1200; Nikon). Pictures were processed in Photoshop 7.0. Environmental scanning EM (ESEM) used living seedlings mounted in 0.8% (wt/vol) top agar visualized with an ESEM TMP instrument (model XL 30; Philips). Images were processed in Photoshop 7.0. Cytoplasmic Ca²⁺ measurements were performed by microinjecting root cells with Indo-1 conjugated to a 10-kD dextran (Molecular Probes) coupled with confocal ratio imaging (Wymer et al., 1997). GFP-talin and GFP-MAP4 reporters have been described previously (Kost et al., 1998; Bibikova et al., 1999). Confocal and Nomarski microscopy was performed with seedlings mounted in water and covered with number 1.5 coverslips. Fluorescence was scanned with an inverted confocal microscope (model 510 meta; Carl Zeiss MicroImaging, Inc.; 63× C-Apochromat 1.2 NA water immersion lens). GFP experiments used standard FITC settings. For YFP, laser excitation and dichroic filters were set at either 458 or 488 nm, and a 505–530-nm bandpass emission filter was used. The confocal pinhole setting was 1 Airy disk unit and z-stack step size was 0.44 μm. z-Stacks were observed unprocessed. All static images were flattened using an average projection. Volume rendering used Volocity 2 software (Improvision). Plants exhibiting comparable GFP or YFP fluorescence were identified using an inverted fluorescent microscope (model DMIRB; Leica) with standard FITC bandpass filter sets (10 0.3 NA objective). For each experiment, seven independent seedlings were analyzed and <2 root hairs were imaged from each seedling.

PH_PLCδ1–YFP fluorescence was normalized to bulk plasma membrane by dividing intensities of YFP fluorescence (emission bandpass 505–530 nm) by fluorescence of a bulk plasma membrane marker FM 1–43 (Molecular Probes; emission bandpass 530–600 nm) in superimposed images of double-labeled root hairs. Excitation was at 458 nm for both dyes. Endocytosis of FM 1–43 was blocked by pretreating plants with 10 mM NaN₃ for 20 min. Plasma membrane staining was imaged immediately after FM 1–43 (1 μM) was added to bathing medium. The PH_PLCδ1–YFP plasmid used in generating the transgenic plants was donated by T. Munnik and W. van Leeuwen (University of Amsterdam, Amsterdam, Netherlands).

3-d-old seedlings were processed for EM essentially as described previously (Čiamporová et al., 2003). Ultrathin sections were observed with an electron microscope (model Tecnai 12; FEI) interfaced with a multiscan camera (model 794; Gatan). All images were processed in Photoshop 7.0.

Seeds were plated in 0.8% (wt/vol) top agar (low-melt agarose in 1× Murashige and Skoog Salt and Vitamin Mixture media [MS; GIBCO BRL]). Seedlings were stratified at 4°C for 4 d, and then grown vertically under constant light (90 μM m⁻² s⁻¹) at 22°C for 3 d. Adult roots were extracted after 45 d of growth in soil, cleaned in MS media, and stained either with Ponceau red or Coomassie blue for 30 min. The Atsfh1::T-DNA line of A. thaliana (Brassica family, Columbia ecotype) was obtained from the Arabidopsis Biological Resource Center via TAIR (http://arabidopsis.org; Alonso et al., 2003). The T-DNA insertion was mapped at the Salk Institute Genomic Analysis Laboratory “T-DNA Express” Arabidopsis Gene Mapping Tool (http://signal.salk.edu/cgi-bin/tdnaexpress). Atsfh1::T-DNA mutants were selected for kanamycin resistance (50 μg/ml). Genotypes were confirmed by PCR, Southern blotting, and DNA sequencing of junctional borders. Agrobacterium tumefaciens–mediated transformation floral dip protocols were routinely used to generate transgenic plant lines (Clough and Bent, 1998).

Seeds were incubated for 72 h at 4°C and sterilized in 70% (wt/vol) ethanol for 2 min and 30% (wt/vol) bleach containing 0.01% (wt/vol) Triton X-100 for 25 min. Seeds were placed on sterilized filter paper strips (Fisher Scientific) in Petri dishes (Fisher Scientific) with ∼120 seeds per dish (15 seeds per strip on 8 strips). Approximately 3 ml of sterilized liquid growth media was added to each Petri dish, and dishes were sealed with parafilm and positioned vertically on a rack. Seedlings were germinated in a Pervical growth chamber at 22°C with a 16:8 h light/dark cycle and 68% relative humidity conditions. Growth medium was sterilized before use and was comprised of 0.1 mM KCl, 0.1 mM CaCl₂, 0.1 mM MgCl₂, 0.5 mM NaCl, 0.3 mM MES, 0.2 mM Na₂SO₄, and 6% sucrose, pH 6.0. Filter paper strips containing three to five seedlings were cut off from the Petri dish and glued to the bottom of a measurement Petri dish. Approximately 5 ml of fresh growth medium was added to the chamber and equilibrated for at least 1 h. To avoid acidification of the medium, media were again replaced and the system allowed to stabilize for 15–20 min before Ca²⁺ flux measurements.

The SIET (Applicable Electronics, Inc.) determines both static ionic/molecular concentrations and concentration gradients by using ion-selective microelectrodes (Kühtreiber and Jaffe, 1990; Schiefelbein et al., 1992). The concentration gradient is measured by moving the electrode repeatedly between two positions in a predefined excursion (5–30 μm) at a fixed frequency in the range of 0.3 to 0.5 Hz. The ion-selective electrode was constructed as follows: glass micropipettes (2 μm aperture) were pulled from 1.5-mm-diam glass capillaries (TW150-4; World Precision Instruments, Inc.) with an electrode puller (P2000; Sutter Instrument Co.) to provide microelectrodes with a 2-μM aperture using a four-step protocol. Micropipettes were silanized with N–N dimethyltrimethylsilamine (Fluka) at 120°C for 50 min, back-filled with 100 mM CaCl₂, and then front-filled with Liquid Ion Exchanger (Fluka for Ca²⁺ electrode) to generate the Ca²⁺-selective probe. The micropipette was placed into an Ag/AgCl wire holder (WPI; reconditioned every time before measurement with self-constructed 9 V DC circuit). The reference was a solid, low leakage electrode (WPI). Ca²⁺ electrodes were calibrated using a series of 1-, 0.1-, and 0.01-mM CaCl₂ solutions. Only electrodes with Nernstian slopes >25 mV were used. Ca²⁺ ion flux was calculated from Fick's law of diffusion:

Fig. S1 and Video 1 show localization of GFP-AtSfh1p to the plasma membrane and VRZ of wild-type root hairs of 3-d-old transgenic plants. Fig. S2 and Videos 2–4 show YFP-PH_PLCδ1 distribution and report the derangement of PtdIns(4,5)P₂ distribution in nullizygous versus wild-type root hairs. Fig. S3 shows Golgi distribution is similar in wild-type and mutant root hairs. Videos 5–7 show GFP-talin imaging in wild-type and nullizygous root hairs and demonstrate loss of the fine tip actin microfilament network in nullizygous root hairs. Videos 8–10 show GFP-MAP4 imaging in root hairs of wild-type and nullizygous root hairs and demonstrate comprehensive loss of organized MTs in nullizygous root hairs.

We are indebted to Joe Kieber for his invaluable advice, insights, sharing of expertise, and access to technology in his laboratory. We thank Claire Hutchison, Jean Deruere, Alan Jones, Andrew Morris, and Brenda Temple (University of North Carolina, Chapel Hill, NC) for valuable discussions and technical advice; Teun Munnik and Wessel van Leeuwen for the PH_PLCδ1–YFP plasmid; John York and June de la Cruz (Duke University, Durham, NC) for access to automated HPLCs for PIP analyses; and Al Shipley (Applicable Electronics, Inc., Forestdale, MA) for critical reading of the manuscript. ESEM was performed at the Duke Biological Sciences EM Facility.

This work was supported by National Institutes of Health grant GM44530 (V.A. Bankaitis) and fellowship “Bourse Lavoisier” from the French Foreign Ministry (P. Vincent). The authors have no conflicting financial interests.