Role of SKD1 Regulators LIP5 and IST1-LIKE1 in Endosomal Sorting and Plant Development^,

Analysis of mutant plants points to a synergistic role of two regulators of the ESCRT components SKD1 during plant development.

Abstract

SKD1 is a core component of the mechanism that degrades plasma membrane proteins via the Endosomal Sorting Complex Required for Transport (ESCRT) pathway. Its ATPase activity and endosomal recruitment are regulated by the ESCRT components LIP5 and IST1. How LIP5 and IST1 affect ESCRT-mediated endosomal trafficking and development in plants is not known. Here we use Arabidopsis mutants to demonstrate that LIP5 controls the constitutive degradation of plasma membrane proteins and the formation of endosomal intraluminal vesicles. Although lip5 mutants were able to polarize the auxin efflux facilitators PIN2 and PIN3, both proteins were mis-sorted to the tonoplast in lip5 root cells. In addition, lip5 root cells over-accumulated PIN2 at the plasma membrane. Consistently with the trafficking defects of PIN proteins, the lip5 roots showed abnormal gravitropism with an enhanced response within the first 4 h after gravistimulation. LIP5 physically interacts with IST1-LIKE1 (ISTL1), a protein predicted to be the Arabidopsis homolog of yeast IST1. However, we found that Arabidopsis contains 12 genes coding for predicted IST1-domain containing proteins (ISTL1–12). Within the ISTL1–6 group, ISTL1 showed the strongest interaction with LIP5, SKD1, and the ESCRT-III-related proteins CHMP1A in yeast two hybrid assays. Through the analysis of single and double mutants, we found that the synthetic interaction of LIP5 with ISTL1, but not with ISTL2, 3, or 6, is essential for normal plant growth, repression of spontaneous cell death, and post-embryonic lethality.

Secretory and endosomal trafficking pathways control the transport and abundance of proteins throughout the endomembrane system of cells. At the plasma membrane, proteins such as ion transporters, channels, and receptors are targeted for degradation through ubiquitination followed by endocytic internalization (Barberon et al., 2011; Kasai et al., 2011; Lu et al., 2011; Shin et al., 2013; Tanno and Komada, 2013; Martins et al., 2015). Internalized plasma membrane proteins are then delivered in endocytic vesicles to early endosomes where they can be recycled back to the plasma membrane or be sorted for degradation in late endosomes, also called multivesicular bodies (MVBs). At MVBs, the ubiquitinated plasma membrane cargo proteins are sorted into intraluminal vesicles (ILVs) that are degraded in the vacuolar lumen when mature MVBs fuse with the vacuole (Reyes et al., 2011). If the cargo proteins fail to be sequestered into ILVs, they are mis-sorted to the vacuolar membrane or tonoplast and cannot be degraded efficiently (Babst et al., 2002; Spitzer et al., 2009; Reyes et al., 2011).

The recognition, concentration, and sorting of ubiquitinated plasma membrane cargo into ILVs is mediated by Endosomal Sorting Complex Required for Transport (ESCRT) proteins. In fungi and metazoans, five multimeric ESCRT complexes called ESCRT-0 to ESCRT-III and the Vps4p/SKD1-Vta1p/LIP5 complex are involved in endosomal cargo sorting and ILV formation. ESCRT-0 binds phosphatidylinositol-3-P and clathrin on the endosomal membranes, recognizes the ubiquitinated membrane proteins, and interacts with ESCRT-I. Within eukaryotes, only fungi and metazoans contain the canonical ESCRT-0 subunits Vps27p/Hrs and Hse-1p/STAM (Leung et al., 2008). Plants and other eukaryotes lacking ESCRT-0 seem to rely on the TOM1 and TOM1-like proteins (Korbei et al., 2013), which are widely distributed in eukaryotes (Herman et al., 2011) and play the role of an ancestral ESCRT-0 module. ESCRT-I and ESCRT-II are also able to bind ubiquitin and thought to initiate or stabilize negative curvature on the endosomal membrane. ESCRT-III proteins assemble into filaments with affinity for highly curved membranes (Fyfe et al., 2011). ESCRT-III proteins do not seem to be able to bind ubiquitin and are found in a closed, inactive state in the cytoplasm, but polymerize in long filaments when recruited by ESCRT-II to the endosomal membrane where they constrict the neck of the nascent ILV and eventually mediate its release into the MVB lumen (Guizetti and Gerlich, 2012; Shen et al., 2014; McCullough et al., 2015). The ESCRT-III complex consists of four core subunits with essential functions on MVB formation: Vacuolar Protein Sorting20/Charged Multivesicular Body Protein6 (Vps20p/CHMP6), Suc Non-Fermenting7 (Snf7p)/CHMP4, Vps24p/CHMP3, and Vps2p/CHMP2, and three accessory subunits with regulatory functions, Did2p/CHMP1, Vps60p/CHMP5, and Increased Salt Tolerance1 (IST1; Babst et al., 2002; Nickerson et al., 2006; Azmi et al., 2008; Nickerson et al., 2010; Henne et al., 2011). The disassembly and recycling of the ESCRT-III components back to the cytoplasm and the continuous release of ILVs into the endosomal lumen require the ATPase mechanozenzyme Vps4p/SKD1, which co-assembles with its cofactor Vta1p/LIP5 (Shiflett et al., 2004; Lottridge et al., 2006; Azmi et al., 2008; Davies et al., 2010) onto membrane-bound ESCRT-III complexes. Vps4p/SKD1 is the only ATP-consuming enzyme within the core ESCRT machinery. Consistent with its central role in the ESCRT pathway, its activity is controlled by multiple mechanisms. Vta1p/LIP5 increases in vitro Vps4p/SKD1 ATPase activity in yeast, animals, and plants (Lottridge et al., 2006; Haas et al., 2007; Vild et al., 2015). In yeast, the C-terminal domain of Vta1p binds to Vps4p, enhancing its oligomerization (Azmi et al., 2006; Lottridge et al., 2006; Yang and Hurley, 2010; Norgan et al., 2013; Davies et al., 2014). Besides the direct activation of SKD1/Vps4p by LIP5/Vta1p, the core ESCRT-III proteins help recruit and activate the Vps4p/SKD1 ATPase complex (Saksena et al., 2009; Davies et al., 2010) by substrate engagement. Finally, the ESCRT-III accessory subunits Did2p/CHMP1, Vps60p/CHMP5, and IST1 also aid in the recruitment of Vps4p/SKD1 to the endosomal membrane and interact directly with Vta1p/LIP5 (Azmi et al., 2008; Dimaano et al., 2008; Agromayor et al., 2009; Bajorek et al., 2009; Vild et al., 2015).

Whereas a role of Vta1p/LIP5 as a stimulator of Vps4p/SKD1 ATPase activity has been widely demonstrated, the mechanisms of action of the ESCRT-III accessory subunits in MVB sorting have been harder to decipher. For example, IST1 is a divergent ESCRT-III subunit that interacts with Did2p/CHMP1 and Vps4p/SKD1 in yeast and mammals but is not required for proper MVB sorting (Dimaano et al., 2008; Rue et al., 2008; Agromayor et al., 2009; Tan et al., 2015). However, the simultaneous deletion of IST1 and Vta1p/LIP5 enhanced trafficking defects in both yeast and mammalian cells (Dimaano et al., 2008; Rue et al., 2008; Agromayor et al., 2009), indicating a positive regulatory role of IST1 in MVB sorting. In yeast, however, the effect of Ist1p on Vps4p changes according to Ist1p concentration. At low concentration, Ist1p positively regulate recruitment of Vps4p through its interaction with Did2p but at high concentration, Ist1p forms a heterodimer with Vps4p, blocking the interaction between Vps4p and the rest of the ESCRT machinery (Dimaano et al., 2008; Jones et al., 2012). In addition, supporting a negative role in the MVB pathway, Ist1p reduces the in vitro ATPase activity of Vps4p (Dimaano et al., 2008).

Although the ESCRT system is conserved across eukaryotes, plants show an important diversification of ESCRT subunits, with the evolution of multiple isoforms for most ESCRT components as well as plant-specific ESCRT proteins, such as FYVE DOMAIN PROTEIN REQUIRED FOR ENDOSOMAL SORTING1 and POSITIVE REGULATOR OF SKD1 (PROS; Gao et al., 2014, 2015; Reyes et al., 2014). The function of SKD1 and some of its key regulators have been analyzed in Arabidopsis. The over-expression of an ATP-locked form of Arabidopsis SKD1 results in alterations of the endomembrane system (Cai et al., 2014) and leads to a lethal phenotype in plants (Haas et al., 2007) whereas its specific expression in trichomes causes vacuole fragmentation and the presence of multiple nuclei (Shahriari et al., 2010). Mutations in the two Arabidopsis CHMP1 gene copies lead to embryo and seedling lethality and drastic mislocalization of the auxin efflux facilitators PIN1 and PIN2 (Spitzer et al., 2009). PROS, a plant-specific positive regulator of SKD1 ATPase activity, is required for normal growth and cell expansion (Reyes et al., 2014). Arabidopsis LIP5, as its animal and yeast counterparts, positively regulates SKD1 enzymatic activity in vitro (Haas et al., 2007). In contrast to chmp1a chmp1b and PROS knock-down lines, lip5 mutants show relatively minor growth defects. However, LIP5 is a critical element in plant basal defense against pathogens (Wang et al., 2014).

How the coordinated action of SKD1 regulators affect membrane trafficking and development in multicellular organisms is unknown. Here we investigate the function of two SKD1 regulators, LIP5 and IST1, in Arabidopsis. We have found that LIP5 is important for controlling the constitutive degradation of plasma membrane proteins and the formation of ILVs. We also found a drastic evolutionary diversification of IST1-LIKE (ISTL) proteins in plants and investigated their interactions with LIP5 and other ESCRT components.

RESULTS

PIN Auxin Efflux Facilitators Are Mislocalized in lip5 Mutants

To investigate the function of LIP5 in endosomal trafficking, we used two Arabidopsis lines with T-DNA insertion in the LIP5 gene, lip5-1 (SAIL_854_F08) and lip5-2 (GABI_351F05; Supplemental Fig. S1A). Both lines were previously characterized as transcript-null mutants (Haas et al., 2007; Wang et al., 2014) with slightly smaller rosette leaves when grown in soil (Wang et al., 2014) but otherwise able to complete their lifecycle normally.

Previous work (Spitzer et al., 2009) showed that ESCRT proteins are important for the polar distribution and vacuolar degradation of PIN proteins. Therefore, we expressed PIN2-GFP and PIN3-GFP in lip5 mutants under their native promoters and analyzed their localization and abundance by confocal microscopy imaging. We first analyzed the plasma membrane distribution of PIN2-GFP in vertically grown lip5-1 roots and PIN3-GFP in lip5-1 columella cells before and after gravistimulation, since PIN3 polarization is enhanced toward the lower side of the columella cells as part of the root gravitropic response (Friml et al., 2002; Abas et al., 2006; Harrison and Masson, 2008; Kleine-Vehn et al., 2010). We found that lip5-1 root cells are able to polarize both PIN2-GFP and PIN3-GFP (Supplemental Fig. S2).

However, although the polar localization of PIN2-GFP did not seem to be affected, lip5-1 mutant roots over-accumulated PIN2-GFP in both the plasma membrane and tonoplast (Fig. 1, A and B). We were also able to confirm the over-accumulation of PIN2-GFP in the lip5-1 roots by western blotting of protein extracts (Fig. 1C). To test whether the accumulation of PIN2-GFP was due to higher expression of the ProPIN2::PIN2-GFP transgene in the lip5-1 background, we performed quantitative RT-PCR and found that the steady-state levels of PIN2-GFP transcripts were similar in both wild-type and lip5-1 seedlings (Fig. 1D), indicating that the accumulation of PIN2-GFP in lip5-1 was due to a post-transcriptional event.

Figure 1.

lip5-1 seedlings mis-sort and accumulate PIN proteins. A, Epidermal root cells in wild-type and lip5-1 seedlings showing the localization of PIN2-GFP. Arrows indicate PIN2-GFP mis-sorted to the tonoplast in lip5-1. WT: wild type. B, Quantification of PIN2-GFP signal intensity in epidermal root cells (n = 11 wild-type roots and 10 lip5-1 roots). Asterisks indicate significant differences between wild type and lip5-1 based on a two-tailed Student’s t-test (* P < 0.05, ** P < 0.001). Error bars indicate sd. C, Western-blot analysis of PIN2-GFP abundance in extracts from wild-type and lip5-1 roots. HISTONE3 was used as loading control. H3: HISTONE3; WT: wild type. D, Quantitative PCR of PIN2-GFP transcripts. Expression was normalized to that of UBC9. Data are from three biological replicates. Error bars indicate sd. E, Root caps from wild-type and lip5-1 seedlings showing localization of PIN3-GFP in columella cells. Arrows indicate mis-sorting of PIN3-GFP to the tonoplast in lip5-1 columella cells. WT: wild type. F, PIN3-GFP (green) in root cap cells of wild-type and lip5-1 seedlings. Propidium iodide (red) was used for counterstaining. Arrowheads indicate tiers of cells with PIN3-GFP signal found at the plasma membrane; asterisks indicate tiers of cells within the root cap with PIN3-GFP signal detectable at tonoplast but not at the plasma membrane. WT: wild type. G, Quantification of tiers of cells within the root cap with detectable PIN3-GFP signal, either at the plasma membrane, tonoplast, or both. (n = 9 wild-type and 10 lip5-1 roots). The asterisk (*) indicates significant differences between lip5-1 and wild type based on a two-tailed Student’s t-test (P < 0.005). Error bars indicate sd. Bars = 20 μm.

The lip5-1 mutation also caused the mis-sorting of PIN3-GFP to the tonoplast (Fig. 1E). An initial analysis of PIN3-GFP fluorescence indicated that PIN3-GFP in lip5-1 was present in more cell tiers of the root cap columella compared to wild type (Fig. 1, F and G). However, the extra tiers of cells presenting GFP signal in lip5-1 had most of PIN3-GFP at the tonoplast and not in the plasma membrane (Fig. 1F; asterisks), suggesting that these peripheral columella cells do not express PIN3-GFP any longer but fail to target the already existing PIN3-GFP pool to the vacuolar lumen for degradation.

LIP5 Affects Root Gravitropism But Not Root Growth

Root gravitropism depends on dynamic changes in the amounts and localizations of plasma membrane-localized auxin efflux facilitators of the PIN family (Spalding, 2013). As we detected abnormal distribution of PIN2 and PIN3 in lip5-1, we asked whether lip5 mutant roots were able to respond normally to gravity. Using a machine vision platform for measuring root growth rate and direction (Miller et al., 2007; Brooks et al., 2010), we found that the tips of both lip5-1 and lip5-2 roots reoriented faster than wild type after gravistimulation (Fig. 2A). The faster root tip reorientation was not due to a faster root elongation rate (Supplemental Fig. S3).

Figure 2.

Root gravitropism and degradation of PIN2-GFP in lip5 mutants. A, Tip-angle time course of seedlings of wild-type and lip5 alleles after 90° reorientation (gravistimulation). The graph shows the average tip angle at 2-min intervals mean ± se for each genotype. Significant differences (P < 0.05) in tip angle between wild-type and lip5 mutants are shown in dashed lines as upward deflections (n = 16 seedlings from each genotype). B, PIN2-GFP abundance at the plasma membrane and intracellular membranes including tonoplast of epidermal cells 150 min after gravistimulation. The graph shows ratios PIN2-GFP signal intensity between the lower and upper side of the gravistimulated roots (n = 18–21 seedlings for each genotype). Bars indicate sd. IC: intracellular; PM: plasma membrane; WT: wild type. C, Western-blot analysis of PIN2-GFP abundance in extracts from gravistimulated wild-type and lip5-1 roots. HISTONE3 was used as loading control. H3: HISTONE3; WT: wild type. D, Densitometric quantification of PIN2-GFP/H3 ratios based on three independent western blots. WT: wild type. E, Immunolocalization of PIN2-GFP on wild-type and lip5-1 MVBs. Arrowheads indicate the presence of 6-nm gold particles as indication of positive labeling. WT: wild type. F, Red dots were overlaid on the images displayed in E to more easily visualize the 6-nm gold particles. WT: wild type. Bars = 200 nm.

LIP5 Affects Constitutive But Not Gravistimulation-Induced Degradation of PIN2

We then analyzed whether abnormal dynamics of PIN proteins during gravistimulation were responsible for the enhanced root gravitropic response in lip5 seedlings. During gravistimulation, PIN2 is preferentially degraded in the epidermis of the upper side of the root zone, presumably helping to establish the appropriate auxin differential between the upper and lower sides of the elongation zone that produces curvature (Abas et al., 2006). Degradation of PIN2 during gravitropism depends on PIN2 ubiquitination, endocytic internalization, and vacuolar degradation (Leitner et al., 2012a, 2012b). To assess the ability of lip5 mutants to degrade PIN2, we imaged lip5-1 and wild-type seedlings expressing PIN2-GFP 150 min after gravistimulation and quantified the fluorescence intensity at the plasma membrane of the upper and lower sides of the root epidermis. We did not detect significant differences between lip5-1 and wild-type seedlings in their ability to remove PIN2-GFP from the plasma membrane at the upper side of the root (Fig. 2B). To further confirm that the lip5 mutation does not affect endocytosis rates from the plasma membrane, we monitored the internalization of plasma membrane-bound FM4–64 in lip5 and wild-type seedlings and found no differences in FM4 to FM64 endocytosis rates (Supplemental Fig. S4).

If endocytosis rates are normal in lip5 plants, we reasoned that later steps in the endosomal pathway and degradation of PIN proteins could be affected. We measured PIN2-GFP abundance during gravistimulation of wild-type and lip5-1 mutant seedlings. Consistent with previous studies (Kleine-Vehn et al., 2008), PIN2-GFP abundance in wild-type seedlings increased by more than 2-fold at 60 min (T1) after gravistimulation, followed by a sharp reduction of approximately 50% at 150 min (T2). At 210 min after gravistimulation (T3), the abundance of PIN2 was restored to T1 levels (Fig. 2, C and D). Although the lip5-1 roots contained twice as much PIN2-GFP compared to wild type before gravistimulation, the relative changes in protein abundance after gravistimulation, including a decrease of approximately 50% at 150 min, were very similar to those observed in wild-type roots (Fig. 2, C and D). Consistently, we did not detect a significant increase in the PIN2-GFP pool mis-sorted to the tonoplast in the upper-side root epidermal cells by the time PIN2-GFP was being degraded in their vacuoles (150 min after gravistimulation; Fig. 2B). These results suggest that lip5-1 roots mis-sorted PIN2-GFP to the tonoplast during its constitutive endosomal-mediated degradation. However, we could not detect alterations in gravistimulation-triggered PIN2 proteolysis in lip5-1 roots.

To determine whether cargo is sorted properly into ILVs of lip5-1 MVBs, we immune-localized PIN2-GFP on root cells. We detected positive labeling on ILVs of both wild-type and lip5-1 MVBs (Fig. 2, E and F), indicating that lip5-1 ILVs contain cargo from the plasma membrane.

IST1-LIKE Proteins in Arabidopsis

Whereas LIP5 and other ESCRT components positively modulate SKD1 ATPase activity, IST1 is the only protein proposed to act as an SKD1 negative regulator. To understand the functional integration of positive and negative regulators for the key ESCRT component SKD1, we decided to investigate the function of IST1 in plants. Through a BLAST search, we identified the Arabidopsis protein with highest homology to the yeast IST1 domain (33% identity, AT1G34220) and called it ISTL1. We expressed recombinant ISTL1–6×HIS and GST-LIP5 in bacteria and found they are able to interact in vitro (Fig. 3A). However, a more careful analysis of our IST1 domain BLAST search indicated that, contrary to most eukaryotic organisms containing only one IST1 gene copy, Arabidopsis contains 12 genes coding for IST1-domain-containing or ISTL proteins (Fig. 3B). We also identified three IST1-like genes in the moss Physcomitrella patens and five IST1-like genes in rice (Oryza sativa), suggesting an early diversification of ISTL genes in the plant lineage. In a phylogenetic analysis of ISTL protein sequences, we found that only Arabidopsis ISTL1 to ISTL6 have rice counterparts, whereas ISTL7 to ISTL12 grouped into their own clade (Fig. 3C).

Figure 3.

ISTL proteins in Arabidopsis. A, In vitro pull-down assay of recombinant Arabidopsis LIP5 and ISTL1 proteins tagged either with GST or 6×His and expressed in bacteria. GST-LIP5 or GST alone bound to beads were incubated for 2 h with His-ISTL1. Beads were washed three times. Input and output fractions were loaded on an SDS-PAGE gel and transferred to a nitrocellulose membrane. Proteins were detected with specific antibodies anti-GST or anti-6×His. I: input; O: output. B, Diagrams of IST1 proteins in humans and yeast and ISTL1 proteins in Arabidopsis. Highlighted in yellow is the position of the characteristic IST1 domain. C, Phylogenetic analysis of IST1-LIKE performed with MEGA6. Bootstrap values are shown above each branch. Scale indicates 0.5 amino-acid substitutions per site. Pp, Physcomitrella patens; At, Arabidopsis thaliana; Os, Oryza sativa; Sc, Saccharomyces cerevisiae; Hs, Homo sapiens; Tb, Trypanosoma brucei; Ls, Leishmania major.

We decided to focus on the ISTL1–6 group and tested whether they are all part of the ESCRT machinery by assessing their capability to interact with LIP5, SKD1, and CHMP1 in directed yeast-2-hybrid (Y2H) assays (Fig. 4). As the expression of full-length LIP5 resulted in autoactivation of the GAL4 expression system, we used a fragment of LIP5 (amino acids 1–159) containing two microtubule interacting and trafficking domains expected to interact with IST1 proteins (Agromayor et al., 2009; Guo and Xu, 2015). As evidence of positive interaction, we determined general growth of yeast cells on interaction-selective plates (Fig. 4A), β-galactosidase activity on a colony filter lift assay (Fig. 4B), and the ratio between colonies grown on interaction selection media and transformation selection media for interactions with positive β-galactosidase activity (Fig. 4C; see “Materials and Methods”). In the three tests, only ISTL1 and ISTL5 showed consistent strong interactions with LIP5(1–159), CHMP1, and SKD1 whereas ISTL4 showed positive interaction with CHMP1 but not with SKD1 or LIP5 (1–159; Fig. 4).

Figure 4.

Y2H assays between ISTL1–6 proteins and LIP5(1–159), CHMP1, and SKD1. A, Y2H assays between ISTL1–6 and LIP5(1–159), CHMP1, or SKD1. Plates show colonies growing on −LWH medium (selection for interaction) and on −LW medium (selection for transformation). Controls were performed by coexpressing ISTL, LIP5(1–159), CHMP1, and SKD1 with the corresponding prey or bait empty vectors. B, Detection of β-galactosidase (GUS) activity by a colony filter lift assay using X-gal as a substrate. Asterisks (*) indicate positive GUS activity detection. C, Quantification of colonies grown on selection and interaction media for those cases with positive GUS activity. Graphs show ratios between the number of colonies grown on −LWH medium and the number of colonies grown on −LW medium. The ratio values from control plates were subtracted from those testing direct interactions between ISLT proteins and LIP5(1–159), CHMP1, or SKD1. Between 250 and 400 colonies were counted in each case. The graphs show combined results from two independent experiments; error bars indicate sd.

Within the ISTL1–6 group, ISTL4 and ISTL5 are unique in being approximately twice as large as the other four proteins and preferentially expressed in pollen (Fig. 3B; Supplemental Fig. S5). Therefore, we narrowed our functional analysis to ISTL1, 2, 3, and 6, which are broadly expressed in both vegetative and reproductive tissues (Supplemental Fig. S5). We identified T-DNA lines with insertions in all four genes and isolated the istl1-1, istl2-1, istl3-1, and istl6-1 mutant alleles (Fig. 5A). All four single istl mutants grew normally in soil with no obvious developmental defects (Fig. 5B). Since the four genes have overlapping expression patterns (Supplemental Fig. S5), we suspected they could act redundantly. Therefore, we combined all four transcript-null mutations in an istl1 istl2 istl3 istl6 quadruple mutant (Fig. 5C). However, the quadruple mutant plants also showed normal growth and development (Fig. 5C).

Figure 5.

Identification and characterization of istl mutants. A, Diagrams of ISTL genes and position of T-DNA insertion in the analyzed mutants. Black rectangles indicate exons; black lines indicate introns. B, Overview of 26-d-old wild-type and single istl mutant plants grown in soil. C, Overview of 26-d-old wild-type and quadruple istl1 istl2 istl3 istl6 (quad) mutant plants; RT-PCR of ISTL1, 2, 3, and 6 transcripts from wild-type and quad mutant plants. D, Overview of 26-d-old single lip5-1 and double mutant combinations between lip5-1 and istl mutations. E, Close-up view of 26-d-old istl1 lip5-1 double mutants.

A Genetic Interaction between LIP5 and ISTL1 Controls Defense Responses and Fertility in Arabidopsis

The deletion of either IST1 or LIP5/Vta1 causes mild endosomal trafficking defects in yeast and animal cells. However, the deletion of both genes simultaneously results in severe mis-sorting of MVB cargo proteins (Dimaano et al., 2008; Rue et al., 2008; Agromayor et al., 2009). To test whether this synthetic genetic interaction is conserved in plants and to uncover functional redundancy among ISTL genes, we crossed every one of the istl mutants with lip5-1. Double mutants for lip5-1 and istl2, istl3, or istl6 were indistinguishable from single lip5-1 mutant plants (Fig. 5D). However, the combination of lip5-1 with istl1-1 resulted in plants with reduced growth, small rosette leaves, and shorter petioles. These plants showed signs of early senescence after approximately 3 weeks in soil at 22°C (Fig. 5, D and E). We also found that the istl1-1 lip5-1 cotyledons senesced earlier than those of single mutants or wild-type plants grown in the same conditions (Fig. 6A). To determine the presence of dead cell sectors and accumulation of H₂O₂ in leaves, we stained leaf samples with trypan blue and DAB (3,3′-diaminobenzidine; Shirasu et al., 1999; Noutoshi et al., 2005; Ichimura et al., 2006), respectively (Fig. 6, B and C). We found that istl1-1 lip5-1 leaves stained more strongly with both methods than single mutants and wild-type leaves (Fig. 6, B and C).

Figure 6.

Early cotyledon senescence and spontaneous cell death in istl1 lip5 mutants. A, Cotyledons from 3-week-old plants. WT: wild type. B, First rosette leaves from 3-week-old plants stained with DAB. WT: wild type. C, First rosette leaves from 3-week-old plants stained with Trypan Blue. WT: wild type. D, Quantitative PCR of PR1 transcript. Expression was normalized to that of PP2A. Bars indicate mean ± se. WT: wild type. E, Partial suppression of istl1 lip5-1 of growth defects at a high temperature. A quantity of 31-d-old plants grown at 22°C or 28°C for 17 d. WT: wild type.

Dwarfism, late lethality, early senescence of both cotyledons and leaves, and spontaneous cell death are common features of mutants exhibiting constitutive defense response (Lam, 2004; Ichimura et al., 2006), that is, the up-regulation of genes involved in pathogen responses in the absence of pathogens or elicitors. In these cases, the constitutive up-regulation of defense-related genes leads to spontaneous cell death and oxidative stress, even in the absence of pathogens. To test whether the phenotypic defects of the istl1-1 lip5-1 were connected to a constitutive pathogen response, we analyzed the accumulation of PATHOGEN-RELATED1 (PR1), a salicylic acid and defense response marker gene (Bowling et al., 1994; van Loon et al., 2006; Tsuda et al., 2008). We found a strong up-regulation in the expression of PR1 in the double ist1-1 lip5-1 compared to that in wild-type and lip5-1 seedlings (Fig. 6D).

High temperatures have been shown to partially or completely suppress dwarfism in mutants exhibiting constitutive defense responses (Bieri et al., 2004; Ichimura et al., 2006; Zhu et al., 2010; Wang et al., 2011). We found that the dwarfism of istl1-1 lip5-1 plants was partially suppressed at 28°C, and that mutant plants outlived by 4 weeks their counterparts grown at 22°C (Fig. 6E). Consistently, istl1-1 lip5-1 plants grown at 28°C showed highly reduced levels of PR1 transcripts compared to plants grown at 22°C (Fig. 6D). Although istl1-1 lip5-1 mutant plants grown at 28°C were able to form inflorescences, they were unable to produce seeds.

ISTL1 and LIP5 Control ILV Size and MVB Formation

To determine whether the lack of both LIP5 and ISTL1 function affects MVB formation, we analyzed MVB morphology in high-pressure frozen/freeze-substituted wild-type, lip5-1, istl1-1, and lip5-1 istl1-1 MVBs (Fig. 7A). When the identification of MVBs was not possible solely on morphological recognition, we used immunogold labeling with antibodies against MVB-localized proteins such as RABF2A or VSR1 (Haas et al., 2007; Gao et al., 2014). We found that the diameter of MVBs in the four genotypes was similar, with mean diameters ranging from 260 nm to 280 nm (Fig. 7B), but lip5-1 and istl1-1 lip5-1 MVBs contain fewer intraluminal vesicles compared to those of wild type and istl1-1 (wild type, 5 ILV/MVB section, sd ± 2.4, n = 49 MVBs; lip5-1, 2 ILV/MVB section, sd ± 1.2, n = 87 MVBs; istl1-1, 4.3 ILV/MVB section, sd ± 2.8, n = 44; istl1-1 lip5-1, 1.4 ILV/MVB section, sd ± 1.3, n = 41 MVBs). This represents approximately 60% and 70% decrease in the number of ILVs in lip5-1 and istl1-1 lip5-1 MVBs, respectively. Interestingly, whereas the ILVs from wild-type, istl1-1, and lip5-1 MVBs are similar in size with mean diameters ranging between 35 nm and 38 nm (Fig. 7B), the double istl1-1 lip5-1 mutant ILVs were approximately 27% larger in diameter (mean diameter = 45 nm, SD ± 10.7 nm, n = 40) and consequently, 52% larger in surface area compared to the other three genotypes (Fig. 7B). These results indicate that LIP5 is necessary for ILV formation, and that together LIP5 and ISTL1 control ILV size.

Figure 7.

Analysis of MVBs in roots. A, Overview of multivesicular bodies in high-pressure frozen/freeze-substituted roots. WT: wild type. B, Quantification of morphological features of MVBs (n = 41–87 MVBs and 40–200 ILVs from at least two roots for each genotype). Letters above the bars represent statistical significance (one-way ANOVA followed by Tukey, P < 0.05); bars sharing a letter are not significantly different from one another. Bars indicate sd. Bars = 200 nm. WT: wild type.

DISCUSSION

We have found that the ESCRT component LIP5 is critical for the constitutive degradative sorting of plasma membrane-localized auxin efflux facilitators of the PIN family, but not for the gravity-induced degradation of PIN2. We also found that the enhanced gravitropic response observed in lip5 roots is likely due to changes in the abundance of PIN2 at the plasma membrane. We investigated six of the 12 Arabidopsis ISTL proteins and showed that, among the ISTL1–6 group, ISTL1 showed robust interactions with LIP5, CHMP1, and SKD1 in a Y2H system. Consistently, ISTL1 shows a strong synthetic interaction with LIP5 that negatively regulates spontaneous cell death, early lethality, and constitutive defense responses.

ESCRT, Gravitropism, and PINs

Root gravitropic response provides exceptional tools to analyze endosomal protein trafficking since it involves drastic changes in the distribution and degradation rates of plasma membrane-localized PIN proteins. Auxin redistribution to the lower side of the root upon gravistimulation depends on changes in the distribution and abundance of PIN efflux facilitators (Friml et al., 2002; Harrison and Masson, 2008; Kleine-Vehn et al., 2010). PIN proteins undergo constitutive endosomal recycling (Geldner et al., 2003; Abas et al., 2006; Dhonukshe et al., 2007; Kleine-Vehn et al., 2008, 2011) and are targeted to ILVs and vacuolar degradation by the ESCRT machinery (Spitzer et al., 2009).

Either enhanced PIN3-GFP polarization toward the lower side of the columella cells or an expansion of its expression domain within the root cap could relate to enhanced root gravitropism. We did not observe an increase in the number of root cap cells with PIN3-GFP at the plasma membrane (Fig. 1F), but we noticed a slight increase in the number of vertically grown roots showing lateral polarization of PIN3-GFP in lip5 seedlings (Supplemental Fig. S2). However, how this increase in the random lateral polarization of PIN3-GFP in non-gravistimulated roots can affect gravitropism, is unclear.

Consistent with PIN2 being an ESCRT cargo (Spitzer et al., 2009), PIN2 ubiquitination is important for both its constitutive and gravistimulation-related vacuolar degradation (Leitner et al., 2012a, 2012b). The localization and endosomal recycling rates of PIN2 depend on auxin and other hormones (Paciorek et al., 2005; Willige et al., 2011; Löfke et al., 2013; Marhavý et al., 2014), light (Laxmi et al., 2008; Wan et al., 2012), temperature (Shibasaki et al., 2009; Hanzawa et al., 2013), cell types (Löfke et al., 2015), and gravity (Abas et al., 2006; Rahman et al., 2010).

We have found that lip5 mutant seedlings respond faster to reorientation of root growth induced by gravistimulation. In lip5-1 root cells, PIN2-GFP was not only mis-sorted to the vacuolar membrane, but also over-accumulated at the plasma membrane. Although not analyzed in this study, loss-of-function of LIP5 could affect the abundance and distribution of many other plasma membrane proteins that control auxin distribution in roots [e.g. the AUXIN RESISTANT1 influx carrier (Marchant et al., 2002), other PIN proteins (Blilou et al., 2005; Wisniewska et al., 2006), and members of the ATP BINDING CASSETTE SUBFAMILY B family (Noh et al., 2001; Lewis et al., 2007; Titapiwatanakun et al., 2009; Cho et al., 2014)]. However, the sole increase of PIN2 abundance at the plasma membrane has been shown to enhance root gravitropic response. For example, enhanced gravity root responses have been associated with increased accumulation of PIN2 at the plasma membrane due to (1) enhanced PIN2 endosomal recycling associated with gibberellic acid accumulation (Löfke et al., 2013) or high temperature (Hanzawa et al., 2013); (2) reduced PIN2 endocytosis in ROP6 gain-of-function lines (Chen et al., 2012); and (3) stabilization of PIN2 at the plasma membrane by over-expression of the tonoplast transporter ZINC INDUCED FACILITATOR-LIKE1.1 (Remy et al., 2013). Thus, although it is reasonable to think that enhanced gravitropism in lip5 roots could be due to the combined changes in abundance and localization of many plasma membrane proteins, the over-accumulation of PIN2 at the plasma membrane itself could explain the observed root-tropic responses to gravity.

Why would a mutant with normal endocytosis but abnormal MVB function have more PIN2 at the plasma membrane? It is logical to hypothesize that the strong reduction in ILV formation in lip5 mutants leads to reduced sequestration of PIN2 into the lumen of MVBs and its accumulation at the MVB limiting membrane. As plasma membrane proteins are in continuous endocytic and recycling fluxes through the endomembrane system, the failure to efficiently sequester PIN2 into ILVs would likely affect PIN2 recycling rates as well. The endosomal recycling of PIN2 depends on the retromer proteins SORTIN NEXIN1 and VACUOLAR PROTEIN SORTING29 (Jaillais et al., 2006, 2007; Kleine-Vehn et al., 2008; Hanzawa et al., 2013). The rate of PIN2 recycling can be affected by environmental factors. High temperature has a positive effect on the SORTIN NEXIN1-dependent PIN2 recycling from late endosomes and, similar to the lip5-1 mutation, leads to enhanced root gravitropic responses (Hanzawa et al., 2013). Thus, it is possible that the deficient sorting of PIN2 into the lip5 MVB lumen leads to more PIN2 available for recycling and an accumulation of PIN2 at the plasma membrane.

Whereas we found that lip5-1 roots grown vertically over-accumulate PIN2-GFP in both the plasma membrane and tonoplast, we did not detect defects in the gravity-triggered degradation of PIN2-GFP. Both by confocal imaging and western blotting, we measured similar patterns of PIN2-GFP dynamics in wild-type and lip5-1 roots after gravistimulation. The accumulation of PIN2-GFP in vertically grown roots is likely the result of the cumulative endosomal mis-sorting of PIN2-GFP during its constitutive degradation. It is possible that gravity-induced PIN2 proteolysis also carries some degree of cargo mis-sorting at endosomes as the constitutive degradation pathway, but it is not strong enough to be detected with our tools. Alternatively, other components regulating MVB sorting could act in situations when rapid proteolysis of plasma membrane proteins is required, overriding the need for LIP5. In either case, LIP5 seems to be more critical for the constitutive turnover of PIN2 than for its gravity-induced proteolysis, suggesting that the two proteolytic processes may be regulated differently.

Diversification of ISTL Genes in Arabidopsis

Yeast IST1 consists of an N-terminal IST1 domain of approximately 150 amino acids that binds CHMP1 and a C-terminal region that binds the N-terminal domain of Vps4p/SKD1. In yeast, the single vta1Δ, vps60Δ, and ist1Δ mutant strains exhibit either mild or undetectable MVB sorting defects. However, simultaneous deletion of IST1 and either Vta1p or Vps60p leads to strong mis-sorting of endosomal cargo to the vacuolar membrane. It has been postulated that IST1 and Did2p/CHMP1 form a complex that positively regulates Vps4p/SKD1 ATPase activity, in concert or redundantly with the Vta1p/LIP5–Vps60p complex (Dimaano et al., 2008; Rue et al., 2008).

We have identified 12 genes in Arabidopsis coding for predicted proteins with conserved IST1 domains. We found that ISTL1 interacted strongly with both CHMP1 and SKD1 in Y2H assays. Consistently, ISTL1, but not ISTL2, 3, or 6, showed a strong genetic interaction with LIP5. These results suggest that, out of the analyzed ISTL genes, only ISTL1 plays a central role in the MVB pathway. The synthetic effects between mutations in LIP5 and ISTL1 resulted in reduced growth, late lethality, and spontaneous cell death due to a constitutive pathogen response as evidenced by the strong up-regulation of the PR1 gene. Interestingly, although previous studies have shown that LIP5 is important for signaling in response to pathogens (Wang et al., 2014), lip5 mutants do not show up-regulation of PR1, early senescence, or spontaneous cell death, suggesting that either LIP5 or ISTL1 is sufficient to negatively regulate constitutive pathogen responses. Several plasma membrane-localized receptor-like kinases with important functions in plant development and growth have been implicated in constitutive pathogen responses. Thus, BRASSINOSTEROID INSENSITIVE1-ASSOCIATED RECEPTOR KINASE1, SOMATIC EMBRYOGENESIS RECEPTOR KINASE4, and BRASSINOSTEROID INSENSITIVE1-ASSOCIATED RECEPTOR KINASE1-INTERACTING RECEPTOR-LIKE KINASE1 negatively regulate spontaneous cell death and seedling lethality, whereas CYS-RICH RECEPTOR-LIKE KINASE acts as positive regulators (de Oliveira et al., 2016). It will be interesting to investigate in future studies whether abnormal trafficking of some of these proteins in the double mutant can explain the observed phenotypes.

Plants have evolved new proteins with ESCRT function such as FYVE DOMAIN PROTEIN REQUIRED FOR ENDOSOMAL SORTING1 (Gao et al., 2014, 2015) and PROS (Reyes et al., 2014) and increased the number of ESCRT-like subunits, some of which do not seem to act in the MVB pathway. As an example, three genes code for VPS2-like proteins in Arabidopsis, VPS2.1, VPS2.2, and VPS2.3 (Winter and Hauser, 2006). Different from VPS2.1, which seems to play a bona fide VPS2 function, the other two Arabidopsis VPS2 proteins are neither able to interact with the ESCRT-related AMSH3 deubiquitinase enzyme nor to be recruited to endosomes when an ATP-locked form of SKD1 is expressed in protoplasts (Katsiarimpa et al., 2011). A similar scenario may be found in the Arabidopsis ISTL gene family. Whereas ISTL1 seems to act as a canonical IST1 ESCRT-III subunit, the other ISTL proteins may have tissue-specific roles, for example ISTL4 and ISTL5 that are primarily expressed in pollen, or that may have evolved ESCRT-independent functions.

Formation of MVB in Plants

Different from animal and plant cells, yeast cells lacking core ESCRT components have flattened stacked endosomes called “class E” compartments (Rieder et al., 1996), making direct structural comparisons between mutant MVBs in plants and yeast difficult. In contrast, deletions of the yeast ESCRT-III-related subunits cause relatively milder defects in MVB structure, mostly affecting ILV size and luminal membrane surface area (Nickerson et al., 2010). ILVs in yeast MVBs are smaller (mean diameter 24–27 nm; Nickerson et al., 2010; Adell et al., 2014) than those in plants (mean diameter 35–38 nm). However, the variation in ILV sizes due to mutations in LIP5/Vta1p, IST1, and other ESCRT-III accessory subunits shared some similarities. For example, both in plants and yeast, the single lip5/vta1Δ and istl1/ist1Δ mutations caused no changes in ILV diameter. The double istl1 lip5 mutant in plants and the ist1Δ did2Δ double mutant in yeast show an increase in approximately 30% of the ILV diameter, which in terms of ILV membrane surface area translates into an increase of 52% in plants and 68% in yeast.

However, the changes in size and number of ILVs do not seem to correlate directly with the severity of the mis-sorting defects seen in ESCRT mutants. For example, the MVBs of Arabidopsis lip5 and chmp1a chmp1b mutants (Spitzer et al., 2009) are structurally similar, both with reduced number but normal-sized ILVs. If the underlying cause of MVB sorting defects in both mutants is the reduction in ILV surface area and consequently, the reduced capacity to accommodate MVB cargo in ILVs for degradation, both mutants should show similar mis-sorting defects. However, whereas the chmp1a chmp1b mutant shows strong mis-sorting of plasma membrane to the tonoplast, and complete loss of PIN2 polarization, embryo, or early seedling lethality as well as general abnormal development (Spitzer et al., 2009), lip5 plants showed partial mis-sorting of PIN2 and PIN3 to the tonoplast, reduced growth of aerial parts, and normal lifecycle. We have detected abundant PIN2-GFP cargo on ILVs of both wild-type and lip5 MVBs, whereas in PIN1-GFP it was detected in ILVs of wild type but not of chmp1a chmp1b MVBs (Spitzer et al., 2009). This means that although both CHMP1 and LIP5 are required for the formation of ILVs, CHMP1 but not LIP5 has a critical role in cargo sequestration into ILVs.

MATERIALS AND METHODS

Plant Material

The following lines were used in this study: lip5-1 (SAIL_854_F08) and lip5-2 (GABI_315F05, Haas et al., 2007; Wang et al., 2014); istl1-1 (SALK_021562), istl2-1 (SALK_060592), istl3-1 (SAIL_84_C12), istl6-1 (SALK_101433), and ProPIN2:PIN2-GFP (Laxmi et al., 2008); and ProPIN3:PIN3-GFP (Zádníková et al., 2010). Fluorescent protein expression cassettes were introgressed into loss of function lines by crossing.

Unless otherwise stated, plants were grown in growth chambers at 22°C under 16 h of light and 8 h of dark cycle. Seeds were pretreated in 70% ethanol for 10 min, surface-sterilized in 50% bleach for 1 min, and washed in distilled water for at least four times. Seeds were sowed on plates containing 0.5× Murashige and Skoog salts supplemented with 1% Suc, stratified at 4°C for 2–4 d, and set to germinate.

Quantitative PCR and RT-PCR

RNA was isolated from roots of 5-d-old wild-type and lip5-1 seedlings expressing PIN2-GFP or from rosette leaves of soil-grown wild-type, lip5-1, and istl1 lip5-1 plants using a Trizol reagent following the manufacturer’s instructions. Reverse transcription was carried out using the RETROscript kit (Ambion, Foster City, CA). Amplification and detection were performed on a model no. MX3000P qPCR system (Stratagene, San Diego, CA) to monitor double-strand DNA synthesis. Each reaction contained 1 μl of cDNA, 0.5 μL of each of the two gene-specific primers, and 10 µl of MAXIMA SYBR Green/ROX qPCR Master Mix (Thermo Fisher Scientific, Foster City, CA) in a final volume of 20 μl. Results were analyzed with LinRegPCR (V. 2013.0; http://www.hartfaalcentrum.nl/index.php?main=files&sub=LinRegPCR). The relative value for expression levels of each gene was calculated by the equation Y = 2^−ΔΔCt (Livak and Schmittgen, 2001). The calculated relative expression values are normalized to the wild-type expression levels (wild type = 1). Two different technical replicates and three independent sets of plants were used for analysis. Primers used for quantitative PCR analysis are listed in Supplemental Table S1.

Root Gravitropism

We used a morphometric method developed by Miller et al. (2007). Plates containing 5-d-old seedlings were mounted vertically and transverse to the optical axes of CCD cameras. Rotating the plate until the root tip was parallel to the camera’s horizon (90°) initiated the gravitropic response. Each time series of images representing the gravitropic response of a single root was processed by an algorithm that determined the midline of the root in each frame and computed the root tip angle.

Confocal Laser Scanning Imaging

For FM4-64 internalization experiments, 5-d-old seedlings grown on plates were incubated in 4 μm FM4-64 (Molecular Probes, Invitrogen, Carlsbad, CA) in 0.5× Murashige and Skoog salts supplemented with 1% Suc for 5 min, washed in the same medium without the dye, then mounted and imaged at 10 min after initial incubation with the dye. Seedlings were observed on a model no. LSM 510 with a Plan-Apochromat 63× objective [NA = 1.4 oil-immersion, differential interference contrast (DIC); both by Carl Zeiss, Jena, Germany]. A 488-nm argon laser line was used for excitation, and emission was collected using a LP 560 filter (Carl Zeiss). Images with no saturated pixels and captured under similar microscope settings were analyzed with the aid of FIJI (Schindelin et al., 2012). Internalization of FM4-64 is presented as the ratio of fluorescence intensity at the cytoplasm normalized by the intensity at the plasma membrane.

For PIN2-GFP and PIN3-GFP imaging, 4-5-d-old mutant and control seedlings grown on plates or directly on coverslips were observed by using a model no. LSM 510 with a C-Apochromat 40× objective (NA = 1.2 water-immersion; Carl Zeiss) or Plan-Apochromat 63× objective (NA = 1.4 oil-immersion, DIC; Carl Zeiss), and a model no. LSM 780 with a Plan-Apochromat 63× objective (NA = 1.4 oil-immersion, DIC; Carl Zeiss). A 488-nm argon laser line was used for excitation, and emission was collected between 500 and 530 nm or 490 and 562 nm.

Protein Purification and Interaction Assays

Recombinant GST-LIP5 and 6×HIS IST1 LIKE (ISTL1) proteins were expressed in Escherichia coli BL21 and purified as described previously in Spitzer et al. (2009) and Reyes et al. (2014). For the in vitro interaction assay, equivalent amounts of the purified proteins were incubated 2 h at 4°C in a 50-mm Tris-HCl, 0.1% Triton-X-100, protease inhibitor cocktail “Complete” (Roche Diagnostics, Indianapolis, IN), 1 mm sodium orthovanadate, pH 7.6 (i.e. Input). The glutathione-agarose beads were then rinsed three times with the same buffer described above (i.e. Output). Samples were denatured using Laemmli buffer, separated by SDS-PAGE and transferred onto nitrocellulose membrane. The proteins were detected using commercial anti-6×His and anti-GST antibodies (Sigma-Aldrich, St. Louis, MO).

Immunoblot Analysis

Seedling roots were frozen in liquid nitrogen and homogenized in a buffer containing: 250 mm sorbitol; 50 mm HEPES-BPT at pH 7.4 (Sigma-Aldrich); 25 mm ascorbic acid; 1 mm DTT (Fluka/Sigma-Aldrich); 6 mm EGTA 1.2% (w/v) Polyvinyl porrolidone-40 (Sigma-Aldrich); and protease inhibitor cocktail “Complete” (Roche Diagnostics). A 1:2 (w/v) ratio of tissue and homogenization media was used. The homogenization media was filtered through miracloth and centrifuged at 10,000g for 15 min. The supernatant was recovered and kept aside. The pellet was extracted with 1/10 of the original volume of homogenization buffer used in the first extraction and then centrifuged at 600g for 3 min. The supernatant from this second extraction was added to the first supernatant. The combined supernatants were mixed, centrifuged at 600g for 3 min, and the cleared homogenate was retained as supernatant. These cleared supernatants were mixed with acetone (1 ml of acetone per 200 μl of cleared homogenate), incubated for 10 min at −20°C, and centrifuged at 20,000g for 10 min. The resulting supernatant was discarded and the pellet was suspended in 50 μl of resuspension buffer containing 250 mm sorbitol, 25 mm HEPES-BPT pH 7.4, 1 mm DTT, and protease inhibitor cocktail. Protein concentration was measured with a DC Protein Assay (Bio-Rad Laboratories, Hercules, CA) according to manufacturer’s instructions.

For SDS/PAGE, we used a 5% (w/v) acrylamide stacking gel and 12% resolving gel in a MiniProtean 3-Electrophoresis System (Bio-Rad Laboratories). Resuspended pellets aliquots containing 10 μg of protein were incubated for 15 min at room temperature before loading on the gel. After electrophoresis, gels were either stained with Coomassie Brilliant Blue R-250 (Sigma-Aldrich) or transferred to nitrocellulose membranes (GE Healthcare Lifesciences, Pittsburgh, PA) for immunoblotting. Membranes were stained with 0.15% (w/v) Ponceau Red (Sigma-Aldrich) to check for protein transfer and then blocked with 5% (w/v) dried nonfat milk in TBS (Tris-buffered saline) buffer with 0.1% TWEEN 20. The antibodies used for western-blot analysis revealed, in each case, single bands at the expected molecular masses. The primary antibodies used were mouse anti-GFP (1:1000; Roche USA) and rabbit anti-H3 (1:5000; Abcam, Cambridge, UK). Blots were incubated with the appropriate IgG-HRP-conjugated secondary antibody. Antigen–antibody complexes were developed using either electrochemiluminescence western-blotting substrate or Super Signal West Femto Maximum Sensitivity Substrate (Pierce, Rockland, IL). Blots were exposed for different times; exposures in the linear range of signal were selected for densitometric evaluation. Optical densities of the immunoreactive bands were measured using FIJI (Schindelin et al., 2012).

Electron Microscopy and Immunolabeling

Root tips were high-pressure-frozen/freeze-substituted for transmission electron microscopy analysis as described previously in Spitzer et al. (2009). Briefly, roots tips from 1-week-old seedlings were dissected and frozen in a model no. HPM 010 high-pressure freezer (Baltec, Canonsburg, PA). Samples were freeze-substituted in 0.2% glutaraldehyde plus 0.2% uranyl acetate in acetone at −90° for 4 d in an AFS Automated Freeze Substitution device (Leica, Wetzlar, Germany) and embedded in Lowicryl HM20 (Electron Microscopy Sciences, Hatfield, PA). Sections were mounted on formvar-coated nickel grids and blocked for 20 min with a 5% (w/v) solution of nonfat milk in TBS containing 0.1% TWEEN 20. The sections were incubated in the primary polyclonal antibodies against RABF2A/RHA1 (Haas et al., 2007), anti-VSR (Rose Biotechnology, Winchendon, MA), or GFP (Torrey Pines Antibodies, San Diego, CA) for 1 h, rinsed in TBS containing 0.5% TWEEN20, and transferred to the secondary antibody (anti-rabbit IgG 1:10) conjugated to either 15- or 6-nm gold particles for 1 h. Controls lacked the primary antibodies.

Morphometric analysis of multivesicular bodies and intraluminal vesicles was performed with FIJI (Schindelin et al., 2012).

Histochemistry

Two-week Arabidopsis plants were vacuum-infiltrated with 2.5 mm DAB (3,3′-diaminobenzidine) solution, cleared in a solution of 80% (v/v) ethanol, 20% (v/v) chloroform, and 0.15% (w/v) trichloroacetic acid for 24 h, and mounted in 50% (v/v) glycerol solution (Pogány et al., 2004). For detection of cell death, we prepared a Trypan Blue stock solution consisting of 10 g of phenol, 10 ml of glycerol, 10 ml of lactic acid, 10 ml of distilled water, and 0.02 g of Trypan Blue (Sigma-Aldrich). The stock solution was diluted 1:2 with ethanol 96% (Pogány et al., 2009). Arabidopsis leaves were incubated in the lactophenol-Trypan Blue working solution, boiled in a water bath for 1 min, incubated in the working solution for 24 h, and cleared with 1:2 lactophenol/ethanol.

Y2H Assays

CHMP1, LIP5(1–159), and SKD1 cDNAs were cloned in-frame with the GAL4 DNA binding domain in the vector pBD-GAL4 and ISTL1–6 into the GAL4 DNA activation domain of the vector pAD-GAL4 (Stratagene), respectively. Yeast cells were transformed using a lithium acetate-based protocol and grown on synthetic dextrose media. For two-hybrid assays, cells were cotransformed with two plasmids. The ability to drive the HIS3 reporter gene was assessed by growing transformants on selective medium lacking Trp, Leu, and His (−LWH). To quantify the interaction between partners, the number of colonies grown on −LWH medium (selection for interaction) was divided by the number of colonies grown on −LW medium (selection for transformation) to calculate the percentage of colonies showing interaction. The ratio values from control plates were subtracted from those testing direct interactions among ISLT proteins and LIP5(1–159), CHMP1, or SKD1. Between 250 and 400 colonies were counted in each case. Two independent replicates were used for quantification. For each tested interaction, we also detected the activity of the lacZ reporter gene by a colony filter lift assay using X-gal (5-bromo-4-chloro-3-n-dolyl-β-d-galactoside) as a substrate of β-galactosidase.

Phylogenetic Analysis of ISTL1 Protein Sequences

The IST1 and ISTL1 protein sequences were obtained from the National Center for Biotechnology Information protein database and aligned using T-Coffee (http://www.tcoffee.org/Projects/tcoffee/; Notredame et al., 2000). Maximum likelihood-based inference and bootstrap analyses with 10,000 replicates to estimate clade support were performed with MEGA6 (http://www.megasoftware.net/; Tamura et al., 2013). Sequences from Trypanosoma and Leishmania were chosen as outgroup.

Accession Numbers

ISTL1 At1g34220; ISTL2 At1g25420; ISTL3 At4g35730; ISTL4 At4g29440; ISTL5 At2g19710; ISTL6 At1g13340; ISTL7 At4g32350; ISTL8 At1g79910; ISTL9 At1g52315; ISTL10 At2g14830; ISTL11 At3g15490; ISTL12 At1g51900; LIP5 AT4G26750.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Twenty-six-day-old lip5-1 and lip5-2 plants with reduced rosette size compared to wild-type Col-0.

• Supplemental Figure S2. PIN2-GFP and PIN3-GFP polarization in wild-type and lip5-1 roots.

• Supplemental Figure S3. Root growth rate in lip5-1 mutant and wild-type seedlings.

• Supplemental Figure S4. Normal internalization of FM4-64 in lip5 mutants.

• Supplemental Figure S5. Expression patterns of ISTL1–6 genes in Arabidopsis.

• Supplemental Table S1. Primer sequences.

Glossary

DIC

differential interference contrast

ESCRT

Endosomal Sorting Complex Required For Transport

ILV, ILVs

intraluminal vesicles

MVB, MVBs

multivesicular bodies

Y2H

yeast-2-hybrid