Spatial and temporal localization of SPIRRIG and WAVE/SCAR reveal roles for these proteins in actin-mediated root hair development

SPIRRIG and SCAR/WAVE are molecular determinants of actin-mediated root hair development in Arabidopsis.


Introduction
Root hairs are single-cell tubular projections that emerge from root epidermal cells. Root hairs increase the effective surface area of the root system by extending laterally into soil pores, thus enabling increased access to nutrients and water (Carminati et al., 2017;Ruiz et al., 2020). Root hairs have been studied extensively by plant biologists for decades because they serve as excellent models to unravel mechanisms by which cell size and shape in plants are regulated (Grierson et al., 2014). To attain their cylindrical shapes, root hairs undergo tip growth, a process in which expansion of the cell is confined to its apical domain. Tip growth involves a balance between the directed delivery of post-Golgi compartments carrying protein complexes and cell wall building blocks to the cell apex and localized cell wall loosening and recycling of excess membranes. Besides root hairs, tip growth is exhibited by other cell types such as pollen tubes, fungal hyphae, and rhizoids of mosses, liverworts, and algae (Bascom et al., 2018a).
Root epidermal cells called trichoblasts are the cell types that form root hairs. Work in Arabidopsis thaliana has shown that trichoblasts are specified to become root hairforming cells early during root development through the patterned assembly of protein complexes of transcriptional activators and repressors in different cell files Shibata et al., 2018). Upon establishing their identity, trichoblasts undergo two developmental stages that lead to root hair outgrowth. The first stage is the establishment of a root hair initiation domain (RHID) that eventually leads to a conspicuous root hair bulge at the basal (root-tip oriented) end of the trichoblast (Grierson et al., 2014;Nakamura and Grebe, 2018). Several proteins accumulate at the RHID, most prominently, the Rho of Plants (ROP) small GTPases and their guanine nucleotide exchange factor (GEF) activators (Denninger et al., 2019). The second stage is tip growth. In addition to small GTPases, the cytoskeleton, the endomembrane trafficking machinery, cytoplasmic calcium, phosphoinositide lipids, hormones (e.g. auxin and ethylene), and reactive oxygen species are other major players in signaling pathways that modulate root hair development (Bascom et al., 2018a;Nakamura and Grebe, 2018).
The filamentous-actin (F-actin) and microtubule cytoskeletons orchestrate root hair development. The actin cytoskeleton in particular has been studied widely during tip growth, as it functions as tracks for the trafficking of cellular cargo to the cell apex (Stephan, 2017;Bascom et al., 2018a;Szymanski and Staiger, 2018). Many insights into the role of the actin cytoskeleton in tip growth have been demonstrated through work with Arabidopsis mutants and pharmacological approaches involving the use of chemicals that disrupt F-actin. For example, the actin-disrupting compound latrunculin B (LatB) inhibits tip growth, while also inducing the formation of root hairs and pollen tubes with irregular shapes (Bibikova et al., 1999;Gibbon et al., 1999). In this regard, Arabidopsis mutants with defects in the root hairexpressed vegetative ACTIN2 (ACT2) gene are characterized by root hairs that mirror those treated with LatB (Ringli et al., 2002;Yoo et al., 2012). Furthermore, ACT2 and ACT7 mutants display an apical (shoot-ward) shift in the position of the RHID compared to the wild-type (Kiefer et al., 2015).
To fulfill its cellular functions, the actin cytoskeleton is organized into higher order networks that correspond to the growth strategy of the cell. This is evident in tip-growing cells, whereby the base and shank of the cell consist mostly of thick, longitudinal F-actin bundles, while the apex contains actin fringes, rings, patches, or a fine meshwork, depending on the plant species or cell type (Stephan, 2017).
To sustain tip growth, the integrity and organization of these tip-focused F-actin arrays must be maintained, a task facilitated by a diverse collection of actin-binding proteins and actin nucleators (Li et al., 2015;Paez-Garcia, 2018;Szymanski and Staiger, 2018).
In addition to trichome development, the W/SC-ARP2/3 actin filament-nucleating module has been implicated in specifying leaf pavement cell shape, light and auxin-dependent root growth, stomatal gating, gravitropism, responses to salinity stress, and plant immunity (Li et al., 2003;Basu et al., 2005;Zhang et al., 2005a;Dyachok et al., 2011;Zhao et al., 2013;Li et al., 2014;Zou et al., 2016;Isner et al., 2017;Pratap Sahi et al., 2018;Badet et al., 2019). The W/SC and ARP2/3 complexes have been shown to play essential roles in some tip growing cells, such as the protonemal cells of the moss Physcomitrium (formerly Physcomitrella) patens. This was shown when RNA interference of transcripts encoding selected subunits of the moss W/SC and ARP2/3 complexes resulted in substantial protonemal tip growth defects (Harries et al., 2005;Perroud and Quatrano, 2006;Finka et al., 2008). The importance of the W/SC andARP2/3 complexes was further demonstrated by Quatrano (2006, 2008), who showed that two of its components, BRK1 and ARPC4, accumulated in tips of moss protonemal cells. In contrast, Arabidopsis plants with mutations in components of the W/SC and ARP2/3 complexes displayed minimal or no defects in root hair or pollen tube growth (Li et al., 2003;Mathur et al., 2003), and none of the subunits have been conclusively shown to exhibit clear polar localization in these cell types. Therefore, the extent by which the W/SC and ARP2/3 complexes function in tip growing cells of plants remains to be determined.
SPIRRIG (SPI) is one of eight genes that has been considered to be a member of the DIS group, but compared to other DIS mutants, the trichome phenotypes of spi are less severe and the mutant does not display early stage cell swelling that is diagnostic of the DIS group (Schwab et al., 2003). SPI was shown to encode a 3,571 amino acid long protein with N-terminally located armadillo and concanavalin A-like lectin domains and C-terminally located pleckstrin homology, beige and Chediak Higashi (BEACH), and WD40 repeat domains (Saedler et al., 2009). BEACH domain-containing proteins are highly conserved in eukaryotes and are known to function in membrane dynamics, vesicle transport, apoptosis, and receptor signaling. This family of proteins is of clinical importance as they have been implicated in a variety of human disorders, such as cancer, autoimmunity syndrome, and autism (Cullinane et al., 2013).
In addition to mild trichome defects, Arabidopsis SPI mutants have short root hairs characterized by fragmented vacuoles, suggesting that SPI, like other eukaryotic BEACH domain-containing proteins, functions in membrane trafficking (Saedler et al., 2009). Additionally, Steffens et al. (2017) showed that SPI physically interacts with proteins involved in endosomal sorting, reinforcing its role in membrane remodeling. SPI was also demonstrated to localize to mRNA processing bodies (P-bodies) in transiently transfected Arabidopsis epidermal cells, thereby suggesting a novel role for SPI in post-transcriptional regulation (Steffens et al., 2015). Because SPI had a substantial but incomplete overlap with distorted mutant phenotypes, it was suggested that SPI might be involved in actin-mediated cell developmental processes (Saedler et al., 2009), and perhaps function in coordination with W/SC and ARP2/3 complexes. However, SPI is not a known W/SC or ARP2/3 subunit; therefore, its relationship to ARP2/3 function is unclear. Moreover, the uncertainty with regard to SPI function is confounded by the fact that its subcellular localization in root hairs, which exhibit the most profound phenotype when SPI is mutated, remains unknown.
In the current study, we addressed these questions by generating a fully functional SPI fluorescent protein fusion and documented its spatial and temporal localization during root hair development. In addition, we used live cell microscopy to simultaneously image the dynamics of SPI and W/ SC and correlate SPI localization with root hair tip-focused F-actin. Taken together, our results point to potential functional relationships among SPI, W/SC, and actin during root hair development.

Isolation of a new SPIRRIG mutant allele
We previously described a forward genetic screen that led to the isolation of three nonallelic recessive Arabidopsis mutants that were hypersensitive to the growth inhibitory effects of LatB. The hypersensitive to LatB1 (hlb1) and hlb3 mutants have been described previously (Sparks et al., 2016;Sun et al., 2019). hlb1 was disrupted in a gene encoding a trans-Golgi network-localized tetratricopeptide repeat protein involved in actin-mediated membrane recycling (Sparks et al., 2016), whereas the genetic lesion in hlb3 was found to encode the class II actin nucleator formin (Sun et al., 2019). Here, we report on hlb2, the third of these recessive mutants. Like hlb1 and hlb3, primary root growth was more severely inhibited by LatB in hlb2 compared to the wildtype. In the absence of LatB or at low (i.e. 25 nM) LatB concentrations, the primary root length of hlb2 was similar to the wild-type. Differences in root length between wild-type and hlb2 became apparent when seedlings were grown on a concentration of 50-nM LatB and higher (Supplemental Figure S1, A-C).
Through map-based cloning, we found that the mutation in hlb2 was confined to a region between the AT1G02740 and AT1G03410 loci. Nucleotide sequencing revealed that hlb2 had a 10-base pair deletion (Chr1 position: 720,152-720,161) in exon 14 of the AT1G03060 gene. This 10-base pair deletion led to an open-reading frame shift at the Asp 1526 codon, resulting in a truncated protein (Supplemental Figure S2A). AT1G03060 encodes the BEACH domain-containing protein SPI (Saedler et al., 2009;Supplemental Figure S2A). Because the first spi mutant alleles were reported to have short root hairs (Saedler et al., 2009), we examined root hairs of hlb2. The growth rate of hlb2 root hairs was approximately 70% slower than the wild-type, with some root hairs forming only small bulges (Supplemental Figure S2, B and C). Moreover, hlb2 had mild trichome defects that were reminiscent of the phenotypes of previously isolated spi mutant alleles (Supplemental Figure S2D; Saedler et al., 2009). To further verify if HLB2 is SPI, we obtained a mutant from the publicly available SALK collection (SALK_065311), which had a T-DNA insertion in the 10th exon of the SPI gene (Supplemental Figure S2A; Alonso et al., 2003). The SALK_065311 line we obtained is the same as the spi-3 mutant allele reported previously in Steffens et al. (2015). In addition to having similar root hair and trichome defects to hlb2, primary roots of spi-3 were hypersensitive to LatB (Supplemental Figure S2E), and a cross between hlb2 and SALK_065311 failed to complement each other in the F1 hybrid. Taken together, these results indicate that hlb2 is a new spi mutant allele. Based on earlier nomenclature (Steffens et al., 2015), we renamed hlb2 as spi-5 (Supplemental Figure S2A). The spi-5 mutant allele was used for all subsequent experiments.
SPIRRIG localizes to the tips of rapidly elongating root hairs SPI fluorescent protein fusions were previously demonstrated to associate with P-bodies and endosomes (Steffens et al., 2015(Steffens et al., , 2017. However, these constructs have not been shown to complement the defective root hair and trichome phenotypes of spi. Due to the large size of SPI, we were unable to generate native promoter-driven fluorescent protein fusions to the full-length SPI complementary DNA or genomic DNA. To overcome this problem, SPI was tagged with the 3x-Yellow Fluorescent Protein (YFP) for energy transfer (YPet; Nguyen and Daugherty, 2005) using a method called recombineering. This method involves tagging the gene of interest in the context of transformation-competent bacterial artificial chromosomes to ensure that all regulatory sequences are included in the fluorescent protein fusion (Zhou et al., 2011;Brumos et al., 2020).
Once spi was transformed with a recombineered SPI-YPet construct, primary root hypersensitivity to LatB, short root hairs, and defective trichome phenotypes were rescued, indicating that the construct was functional ( Figure 1A; Supplemental Figure S3). Transgenic complementation of spi with SPI-YPet provided additional evidence that HLB2 is SPI. Confocal microscopy of more than 150 root hairs from at least 25 seedlings revealed that SPI-YPet signal was strongest in the tips of rapidly elongating root hairs ( Figure 1, B and C; (Supplemental Movies S1 and S2). Low magnification time-lapse movies showed that 100% of rapidly elongating root hairs had robust SPI-YPet signal at the tips (Supplemental Movie S1). Weak or no SPI-YPet fluorescence was detected at the RHID and during early root hair bulge formation, but the signal intensified as the root hairs transitioned to tip growth ( Figure 1C; Supplemental Movie S3). SPI-YPet signal dissipated as the root hairs matured and the tip growth rate declined (Figure 1, B and C; Supplemental Movie S4). A linear regression analysis showed that the intensity of SPI-YPet fluorescence in root hair tips is significantly positively correlated to rapid root hair growth ( Figure 1D). These results indicate that the SPI protein has functions related to root hair elongation, which is consistent with the short root hair phenotypes of spi.

SPIRRIG is transported to the root hair tip via post-Golgi compartments
The prominent SPI-YPet signal at the tips of elongating root hairs is reminiscent of the localization patterns of post-Golgi markers such as RAB small GTPases, which are known to function in tip-directed secretion (Preuss et al., 2004). We therefore hypothesized that SPI is trafficked to the tips of root hairs via post-Golgi compartments. To test this hypothesis, we treated seedlings expressing SPI-YPet with Brefeldin A (BFA). BFA is a fungal toxin that is routinely used to investigate endomembrane dynamics because it prevents vesicle formation for exocytosis by inhibiting ADP ribosylation factor GEFs (ARF-GEFs), while still enabling endocytosis and some retrograde pathways to continue (Balu ska, 2002, Doyle et al., 2015. In Arabidopsis roots, BFA binds to the ARF-GEF GNOM, which is localized to early endosomes in the endocytic pathway, and possibly also in the trans-Golgi network (TGN; Robinson et al., 2008). Consequently, BFA treatment in Arabidopsis roots causes the formation of TGN/endosomal agglomerations called BFA-induced compartments. Treatment of seedlings expressing SPI-YPet with 50-mM BFA induced the formation of SPI-YPet agglomerates in root hairs ( Figure 2, A-C). The average number of BFA-induced compartments per root hair was 3.69 6 4.05 (SD). We also quantified the effect of BFA by obtaining the fluorescence ratio of the SPI-YPet agglomerates to the fluorescence of the root hair cytoplasm that did not contain any agglomerates ( Figure 2D, inset). The higher fluorescence ratio of BFAtreated root hairs compared to untreated controls reinforces our qualitative observations of the sensitivity of SPI-YPet to BFA ( Figure 2D). The appearance of fluorescent puncta at the shank of root hairs also suggests that SPI-YPet is trafficked to the tip via post-Golgi compartments (Supplemental Movies S2-S4). These results indicate that SPI-YPet is associated with endomembranes and is trafficked to the root hair tips via BFA-sensitive post-Golgi compartments.
The slow root hair growth of spi and the formation of SPI-YPet agglomerates after BFA treatment suggest that spi may be defective in tip-directed protein secretion, which would partly explain the short root hair phenotypes of spi.
To address this question, we expressed the secreted (SEC)red fluorescent protein (RFP), which contains a cleavable sporamin signal peptide that accumulates in the apoplast (Faso et al., 2009), and observed changes in endocytosis Figure 1 A functional SPI-YPet fusion localizes to the tips of growing root hairs. A, SPI-YPet rescues the short root hair phenotype of spi. Bar ¼ 100 mm. B, Low-magnification image shows that SPI-YPet signal is most prominent at the tips of rapidly expanding root hairs (arrows). A black and white look up table (LUT) image is provided to increase the clarity of SPI-YPet localization. SPI-YPet signal (black color) is weak during initiation/early bulge formation and mature root hairs that have terminated tip growth (arrowheads). Bar ¼ 50 mm. C, High-magnification images of single root hairs during bulge formation until tip growth termination. A black and white look up table (LUT) image is provided to increase the clarity of SPI-YPet localization. SPI-YPet (black color) is enriched at the tip of root hairs that are rapidly growing or transitioning to tip growth (arrow). Faint SPI-YPet signal is found in bulging root hairs or those that have stopped elongating (arrowheads). Images are representative of approximately 150 root hairs from at least 25 seedlings. Bar ¼ 10 mm. D, Scatter plot showing correlation analysis of root hair tip SPI-YPet fluorescence and root hair tip growth. The mean fluorescence in the oval in region 1 divided by the oval in region 2 as shown in (C) represents the fluorescence ratio in the Y-axis. Line shows linear regression fit with R 2 value ¼ 0.308 and P ¼ 1.72 Â 10 À5 (n¼5-7 root hairs per time point) using the membrane selective dye FM 1-43 in both spi and wild-type genotypes (Bolte et al., 2004;Jelínková et al., 2010;Malínská et al., 2014). Similar to previous reports, rapidly elongating wild-type root hairs had a tip-focused gradient of SEC-RFP (Sparks et al., 2016; Figure 2E). In contrast, slowgrowing root hairs of spi lacked these tip-focused SEC-RFP gradients ( Figure 2E). We quantified tip-focused SEC-RFP in elongating wild-type and spi root hairs by obtaining the ratio of tip fluorescence to the subapical cytoplasm ( Figure 2F, inset). The lower fluorescence ratio of spi root hairs compared to the wild-type confirmed our qualitative observations ( Figure 2F). These results were corroborated by examining FM 1-43 dye uptake. spi root hairs showed a significantly reduced tip-focused FM 1-43 gradient compared to the wild-type (Supplemental Figure S4, A-C). Altogether, the loss of tip-focused secretion indicated that the spi mutants had defects in tip-directed bulk flow exocytosis.

SPIRRIG maintains root hair tip-focused F-actin
In a study of other spi mutant alleles, the similarities in phenotypes among spi, w/sc, and arp2/3 mutants suggested that SPI could function in actin-dependent cellular processes (Saedler et al., 2009). However, because the trichome defects of spi were mild compared to other w/sc and arp2/3 mutants, no obvious actin phenotypes were observed in spi trichomes (Schwab et al., 2003). To clarify the relationship between SPI and actin, we focused on investigating actin organization in root hairs, since these structures displayed the most obvious growth defects in SPI-altered plants.
To study actin organization, we expressed the live F-actin reporter UBQ10: mGFP-Lifeact in spi (Vidali et al., 2010). This particular F-actin reporter was selected because it prominently labels the tip-focused F-actin meshwork typically observed in root hairs that are rapidly growing (Sparks et al., 2016). In the wild-type, fine F-actin networks were observed in the root hair bulge, which had a weaker signal compared to the thicker actin bundles in other regions of the trichoblast ( Figure 3A). As the root hair bulge expanded and the root hair transitioned to rapid tip growth, the tip-focused Factin meshwork, which consisted of short filaments and dynamic puncta, became more conspicuous (Figure 3, B-E; Supplemental Movie S5; Sparks et al., 2016)). When wildtype root hairs stopped elongating, the tip-focused F-actin meshwork was replaced by thick F-actin cables ( Figure 3F).
F-actin organization in the tips of spi root hairs was different from that of the wild-type. As noted, some root hairs of Figure 2 SPI-YPet is localized to BFA-sensitive post-Golgi compartments. A, B, Low-magnification images showing several root hairs expressing SPI-YPet. Note that untreated root hairs (A) maintain tipfocused SPI-YPet, while those treated with BFA (B) show an abundance of fluorescence agglomerates (arrows). Bar ¼ 50 mm. C, Bright field and corresponding fluorescence image of representative untreated and BFA-treated elongating root hair showing the accumulation of SPI-YPet at the apical dome in solvent control-treated seedlings (arrow). Within 10 min after treatment with 50-mM BFA, SPI-YPet at the root hair tip dissipated and formed fluorescent agglomerates along the subapical regions (arrows). Bar ¼ 10 mm. D, Box plot of BFA-induced agglomerates of SPI-YPet in untreated control root hairs and after treatment with 50-mM BFA. Ratio values were obtained by dividing mean fluorescence in rectangular region in 1 over region 2 (inset). Box limits indicate 25th and 75th percentiles, horizontal line is the median, and whiskers display minimum and maximum values. Each dot represents individual measurements from 8 to 14 root hairs per group from 8 to 24 plants. ***P <0.001 indicates statistical significance as determined by Student's t test. BFA-treated plants had an average of 3.7 BFA-induced agglomerates per root hair, with standard deviation of 4.05, whereas control root hairs showed no BFA-induced agglomerates per root hair. E, The bulk secretory marker SEC-RFP accumulates at the tips of growing wild-type root hairs (arrowhead), but is absent in spi root hairs. Bar ¼ 10 mm. F, Box plot of SEC-RFP root hair tip accumulation expressed as fluorescence ratio. Ratio values were obtained by dividing mean fluorescence in oval region in 1 over region 2 (inset). Box limits indicate 25th and 75th percentiles, horizontal line is the median and whiskers display minimum and maximum values. ***P <0.001 indicates statistical significance as determined by Student's t test. Each dot represents individual measurement from 8 to 10 root hairs per group from 9 to 12 plants spi were only able to form small bulges due to premature termination of tip-growth (Supplemental Figure S2). In spi root hairs, the distinct F-actin meshwork observed in wildtype root hairs was unable to form. Instead, F-actin in these spi root hair bulges contained thick F-actin cables that resembled those of wild-type root hairs that had terminated growth (Figure 3, G and H). However, the thick F-actin cables in nongrowing spi root hair bulges were unstable, as Figure 3 The tip-focused F-actin meshwork is disrupted in root hairs of spi. A-D, Time course of F-actin organization in a wild-type root hair from bulge formation to rapid tip growth. Weakly fluorescing F-actin networks (double asterisks) in the root hair bulge (A) reorganize into prominent tip-focused meshworks as the root hair transitions to rapid tip growth (arrows in B-D). E, Tip-focused F-actin meshworks (arrow) remain prominent in a long, rapidly elongating wild-type root hair. F, Tip-focused F-actin meshworks in wild-type root hairs are replaced with F-actin bundles (arrowheads) that protrude to the tip when growth stops. G-J, Time course of F-actin organization in a spi root hair bulge that is unable to transition to tip growth. Distinct F-actin meshworks are unable to form in root hair bulges that terminate tip growth (double asterisks, I). Thick F-actin bundles eventually form in these short, nongrowing root hair bulges (arrowhead, J). K-N, F-actin organization in slow-growing spi root hairs. Some spi root hairs show the tip-focused F-actin meshworks typically observed in the wild-type (arrow, K, L). However, tip-focused F-actin meshworks in slow-growing spi root hairs either dissipate (double asterisks, M) or prematurely form thick F-actin bundles that protrude to the tip (arrow, N). Images from (A) to (N) are based on maximum projection images of 20-25 optical sections taken at 0.5-mm intervals. O, Representative maximum projection images and corresponding computer-generated cross-sections of wild-type and spi root hair tips. Only growing root hairs with a clear cytoplasmic cap were selected for analysis. The fluorescence ratio of the root hair tip (oval in 1) to background (oval in 2) was used to quantify tip-focused F-actin meshworks. P, Box plot showing tip-focused F-actin fluorescence ratio. Box limits indicate 25th and 75th percentiles, horizontal line is the median and whiskers display minimum and maximum values. ***P <0.001 indicates statistical significance as determined by Student's t test. Each dot represents individual measurement from 18 to 21 root hairs per group from at least five independent seedlings. Q, Comparison of root hair growth rates between wild-type and spi lines with their corresponding live F-actin reporter lines UBQ10: mGFP-Lifeact. Box limits indicate 25th and 75th percentiles, horizontal line is the median and whiskers display minimum and maximum values. *** P <0.001 indicates statistical significance as determined by one-way ANOVA. Each dot represents individual measurement from four to five root hairs per group from one to two plants. All scale bars ¼ 10 mm they dissipated ( Figure 3I) and reformed again at a later time ( Figure 3J). In the small population of spi root hairs that were able to undergo tip growth, a few exhibited a tipfocused F-actin meshwork that resembled those observed in elongating wild-type root hairs ( Figure 3K). However, most of these slow-growing spi root hairs lacked a tip-focused Factin meshwork (Figure 3, L and M; Supplemental Movie S5) or had thick F-actin bundles protruding to the tip, a feature that was reminiscent of nongrowing, mature wild-type root hairs ( Figure 3N). We quantified the disruption of the tip-focused F-actin meshwork in spi by measuring F-actin fluorescence from computer-reconstructed transverse sections of the root hair tip and by taking the ratio of the average fluorescence over background signal. A higher ratio indicates a higher signal of tip-focused F-actin ( Figure 3O). Our analysis showed that the fluorescence ratio in spi root hairs was significantly reduced compared to the wild-type root hairs, supporting visual observations that the tip-focused F-actin meshwork in spi root hairs is disrupted ( Figure 3P).
To ensure that the mGFP-Lifectact probe did not interfere with normal root hair elongation or F-actin organization, we compared the growth rates of the wild-type and spi with and without the reporter. The root hair growth rates of wild-type root hairs without the mGFP-Lifeact probe were not significantly different from that of wild-type root hairs expressing the reporter. Similarly, the growth rate of spi expressing mGFP-Lifectact was not significantly different from that of spi without the reporter ( Figure 3Q).
We generated plants expressing both SPI-YPet and mRuby-Lifeact to examine the correlation between SPI and F-actin in growing root hairs. In elongating root hairs of dual labeled seedlings, SPI-YPet and the mRuby-labeled F-actin meshwork overlapped at the tips of actively elongating root hairs ( Figure 4A; Supplemental Movie S6). In one time-lapse sequence, when the root hair stopped growing at the 80-120 min time points, both SPI-YPet and mRuby-labeled F-actin meshwork dissipated from the root tip ( Figure 4A). Quantification of both SPI-YPet and mRuby-Lifeact from at least three root hair time-lapse sequences revealed that the appearances of both markers at the tip were highly correlated to each other (Figure 4, B and C). This supports the notion that root hair tip-localized SPI is strongly associated with the tip-focused F-actin meshwork, and as such is involved in sustaining normal root hair elongation in coordination with actin.

BRK1 and SCAR2 are molecular determinants of the root hair initiation domain
The identification of SPI as one of the genes in the DIS group that also included genes encoding subunits of the W/ SC and ARP2/3 complexes (Saedler et al., 2009) raises the possibility that SPI might function in root hair developmental pathways mediated by W/SC-ARP2/3. Furthermore, the observation that SPI-YPET accumulation at the root hair tip (Figure 1) mirrors the enrichment of BRK1-YFP and ARPC4-GFP at the apex of Physcomitrium protonemal cells Quatrano, 2006, 2008) raises the possibility that the W/SC and ARP2/3 complexes are localized to the root hair. In an attempt to link SPI with the W/SC and ARP2/3 pathways, we imaged roots of brk1 complemented with BRK1promoter:BRK1-YFP (hereafter referred to as BRK1-YFP; Figure 5, A-C; Dyachok et al., 2008). Unlike SPI-YPet, we did not observe a BRK1-YFP fluorescence gradient in rapidly elongating root hairs. However, closer examination of trichoblasts revealed prominent BRK1-YFP signal at the plasma membrane of the RHID that mirrored the localization of other known early root hair initiation site markers such as ROP ( Figure 5A). Unlike ROP, which has a persistent plasma membrane localization throughout root hair development (Molendijk et al., 2001;Jones et al., 2002), BRK1-YFP signal was transient and dissipated as root hairs transitioned to rapid tip growth ( Figure 5B; Supplemental Movie S7). This observation was in contrast to the intensification of SPI-YPet fluorescence as root hairs proceeded with rapid tip growth (see Figure 1). In agreement with our visual observations, a linear regression analysis showed that BRK1-YFP fluorescence is inversely proportional to the root hair growth rate ( Figure 5C).
Given that BRK1 stabilizes the entire family of SCAR proteins and is required for functional W/SC assembly (Le et al., 2006), we investigated if the SCAR protein localized to the RHID, as did BRK1. To address this question, we imaged a recombineered SCAR2-mCherry fusion expressed in the scar1 scar2 scar3 scar4 (scar1234) quadruple mutant background. Like BRK1-YFP, SCAR2-mCherry marked the RHID and dissipated when rapid root hair tip growth commenced ( Figure 5D).
The accumulation of BRK1-YFP and SCAR2-mCherry at the RHID led us to hypothesize that brk1 and scar1234 might have defects in root hair initiation. One parameter that has been studied extensively as an indicator of root hair initiation defects is planar polarity, which is a measure of root hair position along the length of the trichoblast (Nakamura and Grebe, 2018). We found that the root hair positions of brk1 and scar1234 shifted apically (i.e. toward the shoot) compared to the wild-type ( Figure 5F). The planar polarity defects of brk1 were rescued by expressing BRK1-YFP in brk1, while partial complementation of scar1234 was achieved by expressing SCAR2-mCherry in scar1234 ( Figure 5F).
We also investigated if the ARP2/3 pathway was involved in RHID, since the W/SC complex can interact with the ARP2/3 pathway. APR2/3 subunit mutants arp2, arp3, and arpc5 exhibited significantly apically shifted root hairs that were similar to those of brk1 and scar1234 ( Figure 5F). Nonetheless, we did not observe polarized accumulation of ARPC5-GFP at the RHID or at root hair tips (Supplemental Figure S5). These results suggest that the ARP2/3 pathway may interact with the W/SC complex at the RHID but is not essential for root hair positioning.
Taken together, our results reveal that BRK1 and SCAR2 are molecular determinants of the RHID that are required for specifying the position of root hair emergence and can do so in both an ARP2/3-dependent and -independent manner.
SPI is required for the depletion of BRK1 as root hairs transition to tip growth Live cell microscopy of BRK1-YFP and SPI-YPet revealed contrasting spatial and temporal dynamics, with the former intensifying and the latter dissipating as root hairs elongated (Figures 1 and 5). To observe BRK1 and SPI simultaneously within the same root hair, we generated plants expressing both SPI-YPet and BRK1-mRuby3. Similar to BRK1-YFP, BRK1-mRuby3 labeled the RHID and dissipated as the root hair bulge expanded ( Figure 6A). Within the same root hair cell, SPI-YPet fluorescence at the tip intensified as BRK1-mRuby3 dissipated ( Figure 6A; Supplemental Movie S8). To better understand the relationship between SPI and BRK1, we expressed BRK1-YFP in spi and SPI-YPet in brk1. Unlike BRK1-YFP in the complemented brk1 background ( Figure 5B), BRK1-YFP signal in spi persisted throughout the entire imaging time course ( Figure 6B). In several cases, BRK1-YFP remained visible in nongrowing root hair bulges of spi for more than 60 min. The persistence of BRK1-YFP signal was also observed in short root hair outgrowths of spi ( Figure 6C; Supplemental Movie S9). To support our visual observations with quantitative data, we selected root hairs of spi and brk1 expressing BRK1-YFP that were of approximately equal lengths. From these root hairs, the ratio of BRK1-YFP tip fluorescence to subapical fluorescence was obtained ( Figure 6D). These analyses showed that BRK1-YFP signal persisted in spi, as demonstrated by the higher fluorescence ratio ( Figure 6E). In contrast, BRK1-YFP in brk1 root hairs disappeared from RHID when root hairs experienced rapid tip growth and thus did not exhibit a fluorescence gradient at the root tips ( Figure 6, D and E). In addition, SPI-YPet in brk1 exhibited similar dynamics to SPI-YPet in the complemented spi ( Figure 6F). In brk1, SPI-YPet signal was weak during early root hair bulge formation and intensified as root hair tip growth accelerated ( Figure 6F). Taken together, these results strongly indicate that SPI contributes to the stability of BRK1 during the transition from root hair initiation to rapid tip growth. On the other hand, BRK1 does not appear to have direct effects on SPI during root hair initiation or rapid tip growth.

Discussion
Our work uncovers important insights into the mechanism underlying actin-mediated root hair development. A major result from our studies is the revelation that SPI is a root hair tip-localized protein. Although SPI fused to fluorescent proteins was reported to localize to endosomes and P- Figure 4 SPI-YPet and mRuby-Lifeact co-localizes at the root tip in elongating root hairs. A, Time course of a root hair simultaneously expressing SPI-YPet and mRuby-Lifeact. Note that SPI-positive post-Golgi compartments and F-actin meshworks colocalized at the root hair apex (arrows) and dissipated at around the same time (arrowheads) at 80 min. Images are single median optical sections. Bars ¼ 10 mm. B, Method for obtaining SPI-YPet and mRuby-Lifeact ratios at the root hair tip for data shown in (C). A rectangular ROI at the tip and sub apex was used to measure fluorescence. C, Scatter plot showing correlation analysis of root hair tip mRuby-Lifeact fluorescence and SPI-YPet fluorescence within the same root hair. The mean fluorescence in the rectangle in region 1 divided by the rectangle in region 2 as shown in (C) represents the fluorescence ratio for each reporter. For each ratio value, root hair growth rate was obtained by measuring the displacement of the root hair tip after a 10 min interval. Line shows linear regression fit with R 2 value ¼ 0.725 and P ¼ 2.127 Â 10 À8 (n ¼ 26 time points from three root hair sequences) Figure 5 BRK1 and SCAR2 mark the root hair initiation domain and contribute to planar polarity. A, Maximum projection confocal micrograph of the elongation and maturation zone of an Arabidopsis primary root expressing a functional BRK1-YFP fusion. The image was generated by merging 50 Z sections taken at 0.5-mm intervals. BRK1-YFP accumulates in the basal end walls and root hair initiation domains (arrows). Bar ¼ 50 mm. B, Time course of BRK1-YFP depletion in a developing root hair. BRK1-YFP signal (arrows) is strongest prior to the formation of a root hair bulge (0 min) and gradually dissipates (arrowheads) as the root hair undergoes rapid tip growth (35-50 min). Bar ¼ 20 mm. C, A scatter plot showing an inverse relationship between BRK1-YFP signal and root hair growth rate. Ratio of fluorescence of BRK1-YFP (a) to background (b) (inset) was plotted against root hair growth rate. Line shows linear regression fit with R 2 value ¼ 0.2667 and P ¼ 1.131 Â 10 À7 (n¼ 5-7 root hairs per time point). D, Time course of SCAR2-mCherry in a developing root hair. Like BRK1-YFP, SCAR2-mCherry signal is strongest at the RHID and early stages of root hair bulge formation (0 min, arrow). SCAR2-mCherry signal dissipates when the root hair undergoes rapid tip growth (40 min, arrowhead). Bar ¼ 20 mm. E, Brightfield microscopy images of representative trichoblasts from 6-d-old seedlings showing apical shift in root hair position of brk1 compared to the wild-type, and complementation of brk1 planar polarity phenotypes by BRK1-YFP. Arrowheads mark the end walls of the trichoblast and asterisks mark the basal wall of the emerged root hair. F, Violin plots of root hair planar polarity in wild-type, brk1, scar1234, BRK1-YFP in brk1, SCAR2-mCherry in scar1234, arp2, arp3, and arpc5 genotypes. Relative root hair position was obtained by taking the ratio of the distance from the basal trichoblast wall (bottom arrowheads in E) to the basal root hair wall asterisks over the total length of the trichoblasts (i.e. length between the two arrowsheads). The plot illustrates kernel probability density in which the width represents distribution of data points. The black dot is the median and whiskers display minimum and maximum values. Statistical significance was determined using nonparametric, two sample Kolmogorov-Smirnov pairwise test. Wild-type versus brk1 (***P < 2.2 Â 10 À16 ); brk1 versus BRK1-YFP in brk1 (***P < 2.2 Â 10 À16 ); wild-type versus BRK1-YFP (P ¼ 0.0808 not significant, NS); wild-type versus scar1234 (***P < 2.2 Â 10 À16 ); brk1 versus scar1234 (P ¼ 0.699, ns); scar1234 versus SCAR2-mCherry in scar1234 (***P < 2.2 Â 10 À16 ); wild-type versus SCAR2-mCherry in scar1234 (*P ¼ 0.012); wild-type versus arp2 (***P ¼ 2.739 Â 10 À5 ); wild-type versus arp3 (***P ¼ 8.882 Â 10 À16 ); wild-type versus arpc5 (***P ¼ 9.18 Â 10 À12 ); brk1 versus arp2 (***P < 2.2 Â 10 À16 ); brk1 versus arp3 (***P ¼ 3.41 Â 10 À8 ); brk1 versus arpc5 (***P ¼ 1.304 Â 10 À11 ); arp2 versus arp3 (***P ¼ 0.00036); arp2 versus arpc5 (*P ¼ 0.001624); arp3 versus arpc5 (*P ¼ 0.0465); brk1 versus arp2 (***P < 2.2 Â 10 À16 ); brk1 versus arp3 (***P ¼ 3.41 Â 10 À8 ); brk1 versus arpc5 (***P ¼ 1.304 Â 10 À11 ). n ¼ 90-117 root hairs Figure 6 Depletion of BRK1 as root hairs transition to tip growth is delayed in spi. A, Dual imaging of BRK1-mRuby3 (arrowheads) and SPI-YPet shows that dissipation of BRK1 coincides with accumulation of SPI (arrow) as root hairs transition to rapid tip growth. Bar ¼ 20 mm. B and C, BRK1-YFP signal persists in spi root hair bulges (arrows in B) that fail to transition to tip growth and in slow-growing spi root hairs (arrows in C). Bars ¼ 20 mm. D, Method for quantification of BRK1-YFP signal persistence in the root tip of spi. Root hairs of brk1 and spi expressing BRK1-YFP that were approximately 40 mm in length were selected. Rectangular ROIs (1 and 2) were drawn to obtain fluorescence values. Ratio values were obtained by dividing mean fluorescence in rectangle region 1 over region 2 used to plot data in (E). E, Box plot of BRK1-YFP root hair tip gradient expressed as fluorescence ratio. Box limits indicate 25th and 75th percentiles, horizontal line is the median and whiskers display minimum and maximum values. **P <0.01 indicates statistical significance as determined by Student's t test. Each dot represents measurement from three to six root hairs per group from eight plants. Only root hairs of similar length were compared. The average root hair length for brk1 was 37.28 mm (standard deviation 18.05 mm) and for spi5 was 38.74 mm (standard deviation 18.34 mm). F, Accumulation of SPI-YPet at the root hair tip is not altered in brk1. Bar ¼ 20 mm bodies, such studies have been limited to transient expression assays in biolistically bombarded leaves (Steffens et al., 2015(Steffens et al., , 2017. As such, a mechanistic link between reported SPI subcellular localization patterns and cell growth phenotypes (i.e. short root hairs and distorted trichome shapes) described in the original spi mutant alleles has been unclear. The large size of SPI and the possibility that it is targeted to discrete cellular domains could have hindered transgenic complementation and subsequent in planta localization efforts. Through our work, we demonstrated that a SPI-YPet fusion generated through recombineering complemented spi (Zhou et al., 2011;Brumos et al., 2020). In doing so, we provide compelling evidence that SPI is a root hair tip-enriched BEACH domain-containing protein.
Our studies indicate that SPI mediates root hair tip growth by maintaining the tip-focused fine F-actin meshwork, although the precise mechanism by which this is accomplished is unknown. The root hair tip-focused F-actin meshwork is the functional equivalent of domain-specific actin structures found in other tip-growing cells, such as cortical actin fringes in pollen tubes and actin spots in moss protonemata (Bascom et al., 2018a). Although the tip-focused F-actin meshwork was able to form in a population of spi root hairs, it often dissipated or was replaced by thick actin cables, a feature of mature wild-type root hairs that are terminating tip growth. Although the presence of tip-focused SPI-YPet was strongly correlated with the maintenance of the tip-focused F-actin meshwork, direct causal links between SPI and F-actin have yet to be established. An alternative explanation is that tip growth-induced reduction due to the absence of SPI could lead to downstream, indirect effects on actin organization.
The root hair phenotypes and corresponding depletion of tip-focused F-actin in spi is reminiscent of studies of BEACH domain-containing proteins in mammalian cells, particularly in neurons, which have often been compared to tip-growing plant cells (Balu ska, 2010). For instance, neuronal dendritic spines are cytoplasmic protrusions that modulate excitatory synaptic transmission in the mammalian nervous system. Actin is a major component of dendritic spines that plays a central role in driving the dynamic shape changes and secretory activities of these cytoplasmic protrusions during synaptic signaling (Cingolani and Goda, 2008). In animals, neurobeachin (Nbea) is a BEACH domain-containing protein that has a similar domain architecture to SPI. Cultured neurons of Nbea mice knockouts had fewer dendritic spine protrusions and depleted F-actin at the synapse than the wildtype (Niesmann et al., 2011). Another BEACH domain-containing protein called FAN was shown to be crucial for the formation of filopodia, i.e. actin-rich plasma membrane extensions that enable motile cells to probe their environment (Mattila and Lappalainen, 2008). FAN-deficient fibroblasts had reduced filopodia formation and were unable to reorganize their actin cytoskeletons in response to upstream activation by tumor necrosis factors (Haubert et al., 2007). Taken together, these results indicate that actin-mediated regulation of cell polarity by BEACH domain-containing proteins is likely to be conserved across animals and plants.
The formation of SPI-YPet agglomerates in response to BFA indicates that SPI is associated with post-Golgi compartments. This result is consistent with observations made with BEACH domain-containing proteins in mammals. For example, Nbea was shown to localize to vesicular endomembranes adjacent to the TGN, and like SPI, the distribution of Nbea-positive post-Golgi compartments was altered by BFA (Wang et al., 2000). Furthermore, the Caenorhabditis elegans (nematode) Nbea homolog SEL-2 was demonstrated to function in endomembrane trafficking in polarized epithelial cells based on the finding that sel-2 mutants mistarget proteins normally found in the apical cell surface to the basolateral surface (de Souza et al., 2007). Here, the protein secretory marker SEC-RFP, which is typically trafficked to the tips of elongating wild-type root hairs (Sparks et al., 2016), was uniformly distributed in spi root hairs. The absence of tip-directed SEC-RFP gradients in spi root hairs shows that, as in mammals, plant BEACH domain-containing proteins are required for the proper targeting of molecular cargo to points of polarized cell growth.
Another significant result from our studies is the discovery that BRK1 and SCAR2 are molecular determinants of the RHID. The localization of BRK1 and SCAR2 to the RHID and their dissipation as root hair tip growth commenced contrasts with findings in Physcomitrium in which a functional BRK1-YFP distinctly labeled the tips of caulonemal cells (Perroud and Quatrano, 2008). Like BRK1-YFP, an ARPC4-GFP construct labeled the tips of caulonemal cells (Perroud and Quatrano, 2008), whereas no clear tip-focused labeling of ARPC5-GFP was observed in root hairs (Supplemental Figure S5). The observation that BRK1-YFP prominently localized to the RHID but dissipated during active root hair tip growth was surprising given that Arabidopsis BRK1 complemented the defective filamentous growth of Physcomitrium BRK1 knockouts (Perroud and Quatrano, 2008). Physcomitrium BRK1 and ARPC4 mutants have clear tip growth defects, while Arabidopsis mutants in the W/SC and ARP2/3 complexes have only weak to no tip growth abnormalities (Le et al., 2003;Mathur et al., 2003;Quatrano, 2006, 2008). This suggests that seed plants might have evolved a specialized function for the W/SC complex at the site of root hair emergence and early bulge formation with only a minor role in driving actin-dependent tip growth. Moreover, a recent discovery regarding the emergence of an ARP2/3 independent W/SC pathway in the regulation of F-actin dynamics for sperm nuclear migration in Arabidopsis pollen tubes further indicates that this occurrence may be more common in seed plants. On the other hand, the knockouts of an Arabidopsis SPI ortholog in the liverwort Marchantia polymorpha led to short rhizoids, indicating that SPI has conserved functions across land plants in regulating tip growth (Honkanen et al., 2016). In other single cell types such as diffusely growing trichomes, W/SC and ARP2/3 subunits likely play a more prominent role than SPI based on their more severe trichome phenotypes when the expression of their encoding genes is suppressed.
The weakening of BRK1-mRuby3 fluorescence coinciding with SPI-YPet accumulation, and the persistence of SPI-YPet signal in spi root hair tips, provides indirect evidence that SPI might play a role in mediating BRK1 stability or localized clustering at the plasma membrane of the RHID. Although it is unknown why BRK1-YFP signal persists in spi, it is tempting to speculate that SPI might modulate BRK1 via protein degradation pathways. This possibility is supported by studies in mammals pointing to a role for BEACH domain-containing proteins in protein degradation via the ubiquitination pathway. In mouse models, for example, the BEACH domain-containing protein WDR81 was shown to be essential for the removal of autophagy-dependent ubiquitinated proteins (Liu et al., 2017). In this regard, it is worth noting that the W/SC-ARP2/3 pathway was demonstrated to function in stress-induced autophagy  and that proteasome inhibitors stabilized SCAR during darkinduced primary root growth inhibition (Dyachok et al., 2011). It is possible that SPI-mediated proteolytic pathways and BRK1 coordinate their activities to specify the levels of W/SC at the RHID that enable the transition to actin-dependent rapid tip growth. However, such a scenario is complicated by the observation that the formation of tipdirected SPI-YPet does not appear to require BRK1. This suggests that SPI's appearance at the root hair tip is likely regulated by other factors besides the W/SC complex. Alternatively, the failure of BRK1 to disappear from the RHID in spi mutants may be caused by a general failure to transition to normal tip growth. Future studies will require subjecting root hairs to conditions that prematurely terminate tip growth to determine if BRK1 signals persist.
In summary, our work provides data that contribute to our understanding of actin-mediated root hair development. A crucial result from our work is the discovery that SPI and the W/SC subunits BRK1 and SCAR2 exhibit polarized localization patterns in root hairs that point to potential functional relationships among these proteins during root hair development. In the future, it will be important to evaluate genetic interactions between SPI and W/SC and ask whether SPI physically interacts with actin or subunits of W/SC to better explain the functional links between SPI and W/SC.

Forward-genetic screening and map-based cloning
The A. thaliana hlb mutants from which the spi-5 mutant allele was identified were isolated from a population of T-DNA seeds (Arabidopsis Biological Research Center stock CS31100). Plants generated from these seeds were transformed with the activation tagging pSKI015 plasmid (Weigel et al., 2000;Sparks et al., 2016;Sun et al., 2019). Briefly, mutagenized seeds were surface sterilized in 95% (v/v) ethanol and 20% bleach (v/v), followed by three washes in sterile deionized water. A solution of 0.5Â Murashige and Skoog (MS) basal salt medium with vitamins (PhytoTech Labs, USA) and 1% (w/v) sucrose was prepared, and the pH of the solution was adjusted to 5.7. After adding 0.5% agar (w/ v) (Sigma-Aldrich), the solution was autoclaved and allowed to cool at room temperature. Upon reaching 55 C, a stock solution of 10-mM LatB (CalBiochem-EMD Chemicals) in 100% dimethyl sulfoxide (DMSO) was added to make a final LatB concentration of 100 nM. Sterilized seeds were suspended in the MS-agar-LatB medium and gently swirled to evenly distribute the seeds. The seed-MS-agar-LatB mixture was poured to a thickness of 2 mm on the bases of 10 cm Â 10 cm Petri dishes. After incubating the plates at 4 C for 48 h, they were positioned vertically in a Conviron plant growth chamber set to 24 C with a 14-h light (120 mmol m À2 s À1 )/10-h dark cycle. Six days after transfer to the Conviron growth chamber, seedlings that exhibited severe growth inhibition were transplanted to LatB-free medium and grown to maturity.
For LatB hypersensitivity assays, seeds from selected plants and the hlb2 mutant were planted on the surface of a 3 mm layer of polymerized 0.5Â MS-1% agar growth medium in gridded square 10 cm Â 10 cm Petri plates and grown in the same Conviron used for screening. Four-dayold seedlings with primary roots that were $1 cm long were selected. Selected seedlings were transplanted to a new set of square Petri dishes containing MS medium supplemented with 50-nM LatB or MS supplemented with the appropriate volume of LatB solvent (i.e. DMSO). During transplantation, the tip of the root was positioned at the grid line of the Petri dish and maintained in a vertical orientation. Four days after transplanting, images of the roots were captured with a Nikon Insight digital camera mounted on a copy stand. Primary root length was expressed as the distance between the position of the root tip 4 d after transplantation and the grid line where the root tip was positioned during transplantation (Supplemental Figure S1).
To identify the HLB2 gene, homozygous hlb2 (Col-0 ecotype) was out-crossed to the Landsberg erecta ecotype to generate seeds for map-based cloning because attempts to identify the responsible mutation for the hlb2 phenotype using TAIL-PCR had been unsuccessful. Segregating F2 seedlings were surface sterilized as described above and grown for 3 d. These seedlings were then transferred to MS medium containing 50 nM LatB and root lengths were marked on the plates to track root growth. Transferred seedlings were grown vertically for an additional 4 d in the growth chamber and seedlings that showed root hypersensitivity to 50-nM LatB were selected for mapping. Briefly, DNA was extracted (Edwards et al., 1991) from approximately 2,600 LatB sensitive seedlings and cloned with simple sequence length polymorphism and cleavage of amplified polymorphic site markers to chromosome one between the AT1G02740 and AT1G03410 loci (Lukowitz et al., 2000), which spanned 248 kb and contained 77 annotated genes. Primers were then designed to several candidate genes based on the sequenced 248-kb region. Nucleotide sequencing revealed that the hlb2 mutation had a 10-base pair deletion (Chr1 position 720, 152-720, 161) in exon 14 of the AT1G03060 gene. In addition, we obtained a T-DNA insertional mutant (SALK_065311) from the ABRC with a predicted insertion in SPI. After genotyping, SALK_065311 was subjected to similar LatB hypersensitivity and growth assays as hlb2 as mentioned above. Allelism was determined by examining the F1 progeny from a cross between hlb2 Â SALK_065311. Following the nomenclature of Steffens et al. (2015), SALK_065311 and hlb2 were named spi-3 and spi-5, respectively.

Generation of fluorescent protein-tagged constructs and plant lines
The SPI protein was tagged at the C-terminus with the fluorescent protein 3x-YPet via the recombineering method (Zhou et al., 2011). Agrobacterium tumefaciens UIA143 pMP90 harboring a SPI-YPet recombineering construct was used to transform the spi-5 mutant by the floral dip method (Clough and Bent, 1998). Transgenic lines were isolated based on resistance to kanamycin and restoration of the wild-type root length. The SPI-YPet transgene was amplified with primers (5 0 -ATTCCACAAGCAACCAGTCAC-3 0 and 5 0 -AACAGAGTTGAGAGTGGCTCG-3 0 ) and sequenced to confirm the correct configuration. Final validation of SPI-YPet expression was accomplished by screening selected lines under the confocal microscope for YPet fluorescence and complementation of the short root hair phenotype.
SCAR2-mCherry lines were also generated by recombineering in which an mCherry tag was fused to an internal region of SCAR2 (Sharan et al., 2009). The mCherry tag was inserted after amino acid 585 in SCAR2 to generate SCAR2-i2mCherry. Recombineering primers were designed as follows: the forward primer contained 50 SCAR2 nucleotides upstream from the insertion site of SCAR2 and 18 nucleotides of the 5 0 -end of the mCherry cassette, which had a 5Â Glycine, 1 Alanine linker. The reverse primer contained 50 nucleotides after the insert site of SCAR2 and 24 nucleotides of the 3 0 -end of the mCherry cassette. The JAtY75L14 clone was transformed into the recombineering competent Escherichia coli strain SW105 and recombined with the SCAR2-i2mCherry recombineering cassettes. A flippase recombination reaction removed the ampicillin resistance marker. The sequence between the two test primers was verified by DNA sequencing. The clones were transformed into Agrobacterium and then into scar1234 plants (Dyachok et al., 2008). Because both scar1234 plants and the recombineering clones were Basta resistant, plants on MS plates were screened by eye for rescued trichome phenotypes. Rescued plants were genotyped with the recombineering test primers and screened for fluorescence under a confocal microscope.
Plants expressing BRK1-YFP were crossed with spi-5 to generate BRK1-YFP in the spi-5 background. In parallel, plants expressing SPI-YPet were crossed with brk1 to obtain SPI-YPet in the brk1 background. Spi-5 was directly transformed with the UBQ10:mGFP-Lifeact construct by the floral dip method (Clough and Bent, 1998). Spi-5 expressing SEC-RFP was generated by crossing spi-5 with SEC7-RFP-expressing wild-type plants, and progeny in subsequent generations that exhibited fluorescence and the spi root hair phenotypes were selected for analysis.

Generation of dual fluorescent protein-labeled plant lines
To generate plant lines co-expressing SPI-YPet and BRK1-mRuby3, SPI-YPet in spi-5 was crossed with BRK1-mRuby3 in the brk1-2 background. Progeny from subsequent generations that exhibited yellow and red fluorescence and rescued root hair and trichome phenotypes were selected for the analysis. For lines expressing SPI-YPet and mRuby3-Lifeact (Bascom et al., 2018b), SPI-YPet in the spi-5 background was directly transformed with a UBQ10:Lifeact-mRuby3 construct. Seedlings that showed both YPet and mRuby3 fluorescence were selected under the confocal microscope.

Evaluation of root hair growth rate
Seeds were surface-sterilized in ethanol and bleach as described above. To evaluate root hair growth rate, seeds were planted on 48 Â 64 mm coverslips coated with 0.5Â MS, 1% sucrose, and 0.4% (w/v) Gelzan CM (Sigma Aldrich, USA) according to Dyachok et al. (2016). The coverslips were placed in 9-cm round Petri dishes and the seeds stratified at 4 C for 48 h. After stratification, the coverslip system with planted seed were kept in a 24 C growth chamber under a 14-h light (120 mmol m À2 s À1 )/10-h dark cycle for 5-6 d.
To quantify root hair growth rate, time-lapse sequences of elongating root hairs at intervals of 10 min over a period of 60 min from a region of the primary root located between 2 and 3 mm from the root tip were captured with a Nikon Eclipse TE300 inverted microscope using a 10Â objective. Root hair lengths at various time points were extracted from digital images using ImageJ (v1.51) software (https:// imagej.nih.gov/ij/). The displacement of the root hair tip after each 10-min interval was obtained and divided by 10 to obtain growth rate as mm/min.

Chemical treatments of root hairs
For BFA treatment, a stock solution of 10 mM was made by dissolving BFA powder (Sigma-Aldrich) in DMSO. Subsequently, the working solution of 50 mM was diluted in 0.5Â MS, 1% sucrose solution, and loaded into a 1-mL syringe. The BFA solution was injected directly next to the roots of 4-to 5-d-old Arabidopsis seedlings grown on coverslips and incubated at room temperature for 10 min prior to microscopy.
FM 1-43 dye (ThermoFisher Scientific) was dissolved in DMSO to make a 10-mM stock solution. The dye was diluted to 4 mM in 0.5Â MS, 1% sucrose solution. The dye was loaded into a 1-mL syringe, and the solution was injected next to the roots as described above.

Microscopy and image analysis
Live cell imaging of root hairs using confocal microscopy was performed on 4-or 5-d-old seedlings grown on the 48 mm Â 64 mm coverslip system described above. Coverslips with seedlings were placed horizontally on the stage of an inverted Leica SP8-X point scanning confocal microscope (Leica Microsystems, Buffalo Grove, IL, USA) or an UltraView ERS spinning-disc confocal microscope (Perkin Elmer Life and Analytical Sciences, Waltham, MA, USA) equipped with 40 Â water (numerical aperture ¼ 1.10) or 100 Â oil (numerical aperture ¼ 1.40) immersion objectives. SPI-YPet and YFP-BRK1 were imaged by illuminating roots growing along the coverslip surface with the 514 nm line of the SP8-XArgon laser and emission detected at 527 nm. Images of root hairs expressing SCAR2-mCherry, BRK1-mRuby3, Lifeact-mRuby, and SEC7-RFP were acquired by illumination with the tunable SP8-X white light laser (560-580 nm) and detecting emission at 610 nm. Excitation and emission parameters for GFP (GFP-ARPC5 and mGFP-Lifeact) were 488 nm and 510 nm, respectively. Time-lapse movies or single time point images were collected using Volocity acquisition version 6.3.5 (Improvision) or SPX-8 LAS software, for the UltraView and Leica SP8-X, respectively.
Quantification of fluorescence from root hair images was conducted on 8-bit confocal images acquired at a single fixed focal plane that spanned the median of the cell (SPI-YPet, BRK1-YFP, SEC-RFP, and BFA and FM 1-43 treatments) or from maximum projected images (mGFP-Lifeact). For SPI-YPet and SEC-RFP, an oval region of interest (ROI) at the root hair tip was drawn and mean fluorescence within this area was acquired using ImageJ. Fluorescence was expressed as the ratio of mean fluorescence within the root tip ROI to background fluorescence. For SPI-YPet, the background used was the region adjacent to but outside of the root hair tip ( Figure 1C), while for SEC-RFP, the background used was an area on the sub-apical region of the root hair tip ( Figure 2F). For BRK1-YFP, a ROI was drawn along the apical-most root hair tip that was approximately 20 pixels wide using the selection brush tool of ImageJ. The ratio of fluorescence within this area to background fluorescence was obtained ( Figure 2C). For BFA, a variable ROI was selected based on the size of BFA-induced agglomerates using the selection brush tool of ImageJ. The ratio of fluorescence was calculated based on similarly sized ROI in the cytoplasmic region. For FM 1-43 experiment, the same quantification technique as for SEC-RFP was used.
Root hair growth rate data for scatter plots in Figures 1  and 5 were derived from the same root hair images in which fluorescence images were acquired under a Leica SP8-X. For the former, an image of a root hair was taken at time 0 and every 5 min thereafter. Scatter plots to determine the relationship between growth rate and tip-focused fluorescence were generated in R (R Core Team, 2019) using the ggplot function in the ggplot2 package (Wickham, 2016). Linear regression analysis between two variables was performed using the lme4 package in R software (Bates et al., 2015;R Core Team, 2019).
Root hair growth rate for Figure 3Q was obtained directly on a Nikon TE300 from root hairs growing in 0.5Â MS, 1% sucrose, and 0.4% Gelzan in 5-cm diameter round Petri dishes. Time-lapse images were taken every 10 min over a period of 60 min. The growth rate refers to the displacement of the root hair tip in micrometer divided by time elapsed (min).
To quantify tip-focused F-actin, 25 optical sections were taken at 0.5-mm intervals using an UltraView spinning-disc confocal microscope. Raw Ultraview Z-stacks were exported to Imaris image analysis software version 9.2.0 (Bitplane). Transverse sections of the root hair tip were obtained from Z-stacks using the surpass view interface of Imaris software and exported as 8-bit TIFF files. From these images, an ROI spanning the circular area of the root hair tip was drawn using ImageJ and mean fluorescence was extracted ( Figure 3O). The ratio of the tip fluorescence to background was obtained from 18 to 21 root hairs.
Pairwise t tests and ANOVA tests were performed using the lsmeans package (Lenth, 2016) in R (R Core Team, 2019). Violin plots for planar polarity in Figure 5F were designed with the ggplot function in the ggplot2 package (Wickham, 2016). Pairwise Kolmogorov-Smirnov test was selected to compare the shapes of two empirical cumulative distributions for planar polarity of two genotypes using the basic package in R (R Core Team, 2019). Statistical results of t tests and ANOVA are shown in Supplemental Tables S1 and S2, respectively.

Supplemental data
The following materials are available in the online version of this article.
Supplemental Figure S1. Primary root growth of hlb2 seedlings is hypersensitive to LatB.
Supplemental Figure S2. HLB2 encodes the beach domain containing protein SPIRRIG.
Supplemental Figure S4. FM1-43 uptake assays in growing wild-type and spi root hairs.
Supplemental Figure S5. The ARP2/3 pathway does not mark the root hair initiation domain.
Supplemental Table S1. Statistical test results for t test.
Supplemental Table S2. Statistical test results for ANOVA.
Supplemental Movie S1. Time-lapse confocal microscopy of SPI-YPet in multiple elongating root hairs.
Supplemental Movie S2. Time-lapse confocal microscopy of SPI-YPet in a single rapidly elongating root hair.
Supplemental Movie S3. Time-lapse confocal microscopy of SPI-YPet in a root hair bulge transitioning to tip growth.
Supplemental Movie S4. Time-lapse confocal microscopy of SPI-YPet in a root hair during termination of tip growth.
Supplemental Movie S5. Spinning-disc confocal microscopy of Lifeact-mGFP in elongating root hairs of wild-type and spi.
Supplemental Movie S7. Time-lapse confocal microscopy of BRK1-YFP root hair bulge transitioning to tip growth.
Supplemental Movie S8. Time-lapse confocal microscopy of a root hair bulge transitioning to tip growth and coexpressing SPI-YPet and mRuby3-Lifeact.
Supplemental Movie S9. Time-lapse confocal microscopy of a spi root hair expressing BRK1-YFP.