Arabidopsis RAB8A, RAB8B and RAB8D Proteins Interact with Several RTNLB Proteins and are Involved in the Agrobacterium tumefaciens Infection Process

ABSTRACT

Arabidopsis thaliana small GTP-binding proteins, AtRAB8s, associate with the endomembrane system and modulate tubulovesicular trafficking between compartments of the biosynthetic and endocytic pathways. There are five members in Arabidopsis, namely AtRAB8A-8E. Yeast two-hybrid assays, bimolecular fluorescence complementation assays and glutathione-S-transferase pull-down assays showed that RAB8A, 8B and 8D interacted with several membrane-associated reticulon-like (AtRTNLB) proteins in yeast, plant cells and in vitro. Furthermore, RAB8A, 8B and 8D proteins showed interactions with the Agrobacterium tumefaciens virulence protein, VirB2, a component of a type IV secretion system (T4SS). A. tumefaciens uses a T4SS to transfer T-DNA and Virulence proteins to plants, which causes crown gall disease in plants. The Arabidopsis rab8A, rab8B and rab8D single mutants showed decreased levels of Agrobacterium-mediated root and seedling transformation, while the RAB8A, 8B and 8D overexpression transgenic Arabidopsis plants were hypersusceptible to A. tumefaciens and Pseudomonas syringae infections. RAB8A-8E transcripts accumulated differently in roots, rosette leaves, cauline leaves, inflorescence and flowers of wild-type plants. In summary, RAB8A, 8B and 8D interacted with several RTNLB proteins and participated in A. tumefaciens and P. syringae infection processes.

Introduction

The soil-resident Agrobacterium tumefaciens infects various plant species and causes crown gall tumors in nature. The unique trans-kingdom DNA transfer ability of this pathogen has made it the most popular plant transformation tool in the plant biotechnology industry (Hwang et al. 2017). During A. tumefaciens infection, in addition to the transferred DNA (T-DNA), several effector virulence (Vir) proteins are transported into plant cells. Inside plant cells, numerous plant proteins are recruited to assist T-DNA transfer, integration and finally expression of oncogenes in the wild-type T-DNA, leading to the overproduction of plant hormones, cytokinin and auxin (Lacroix and Citovsky 2019).

Many environmental factors and plant compounds can affect and induce virulence (vir) gene expressions (Guo et al. 2019). When plants release various phenolic compounds, including acetosyringone (AS), from injured tissues, these phenolic compounds can be sensed by the VirA/VirG two-component system of A. tumefaciens that induces vir gene expressions and initiates the infection process (Lacroix and Citovsky 2019). With the assistance of the VirD1, VirD2, VirC1 and VirC2 proteins, the single-strand T-DNA is generated and its 5ʹ end is covalently bounded by the VirD2 protein. In addition to the VirE2, VirE3, VirD5 and VirF proteins, the T-DNA is transferred from the bacteria into plant cells through a type IV secretion system (T4SS) that is composed of a transmembrane protein complex and a T-pilus (Lacroix and Citovsky 2019, Li et al. 2019). The filamentous T-pilus structure is composed of VirB2, VirB5 and VirB7 (Li et al. 2019). The major component, VirB2, is translated as a 12.3 kD pro-pilin protein; a signal peptide of 47 amino acids of the VirB2 protein is cleaved and is processed to form a 7.2 kD pilin protein. VirB2 then undergoes a cyclization reaction that is novel to bacteria, resulting in the formation of intramolecular covalent head-to-tail peptide bond (Lai and Kado 1998, Eisenbrandt et al. 1999). The VirB5 is a minor pilin component and is important for T-pilus assembly (Schmidt-Eisenlohr et al. 1999). The VirB5 is located at the tips of T-pili and may function as the adhesin of the T4SS and coordinate host-pathogen interactions (Aly and Baron 2007).

After T-DNA enters plant cells, the VirE2 functions as a single-strand DNA-binding protein and binds to the VirD2-T-DNA to protect the single strand T-DNA from being degraded by the host defense mechanism (Duckely and Hohn 2003). The VirE2 may then migrate via the endoplasmic reticulum (ER) and F-actin network inside plant cells to facilitate T-DNA nuclear targeting (Li and Pan 2017, Yang et al. 2017). The VirD2, VirE2 and VirF interact with several plant proteins to help the T-DNA enter the nucleus and may utilize plant defense mechanisms and DNA repair machinery to help the T-DNA integrate into the plant chromosome (Gelvin 2017, Lacroix and Citovsky 2019).

Previous studies have shown that two kinds of Arabidopsis proteins interact with the A. tumefaciens VirB2, including five members of a reticulon-like (RTNL) protein family, AtRTNLB1-4 and 8, and a member of a RAB8 small GTPase subfamily, RAB8B (Hwang and Gelvin 2004, Huang et al. 2018). The AtRTNLBs are mainly localized in the ER and can form multimeric, arc-like structures that may ultimately shape and determine the diameter of ER tubules (Tolley et al. 2008, Sparkes et al. 2010), The RTNL proteins may interact with themselves or other proteins to perform specific functions, including intracellular transport, vesicle formation and membrane curvature (Oertle and Schwab 2003, Nziengui and Schoefs 2008). Another VirB2-interacting protein, the AtRAB8B, is a member of small GTPases that mediate different aspects of membrane trafficking in eukaryotic cells. The Arabidopsis genome contains 57 RAB proteins, which can be divided into eight subfamilies (RABA to RABH) based on their sequence similarities (Rutherford and Moore 2002, Vernoud et al. 2003). Each member of the RAB GTPase family is located at a different intracellular compartment where the RABs function as molecular switches to regulate vesicle trafficking between membrane-bound intracellular compartments (Martiniere and Moreau 2020, Nielsen 2020). The AtRABE subclass contains five members, namely AtRAB8A/E1c (At3g46060), AtRAB8B/E1a (At3g53610), AtRAB8C/E1d (At5g03520), AtRAB8D/E1b (At5g59840) and AtRAB8E/E1e (At3g09900) (Rutherford and Moore 2002, Vernoud et al. 2003). The AtRAB8A, 8B and 8D share high amino acid sequence similarities (Vernoud et al. 2003, Speth et al. 2009). The AtRAB8 is homologous to the mammalian protein Rab8, the Sec4 of Saccharomyces cerevisiae and the Ypt2 of Schizosaccharomyces pombe, which are proteins known or suspected to be involved in polarized vesicle transport from the trans-Golgi network to the plasma membrane (PM) (Goud et al. 1988, Craighead et al. 1993, Huber et al. 1993).

The fluorescent protein tagged-RAB8C and the endogenous RAB8C have been previously detected in the Golgi apparatus and the PM in Arabidopsis leaf cells, supporting their roles in post-Golgi secretory trafficking (Speth et al. 2009). Heterologous expressions of the dominant negative mutant of Arabidopsis RAB8C in tobacco leaf epidermis caused a secreted GFP marker to accumulate in the ER, Golgi apparatus and resulted in the mis-localization of the Green Fluorescent Protein (GFP) signal to the vacuole (Zheng et al. 2005). Moreover, the GTP-bound RAB8A can directly interact with the Arabidopsis PEX7 protein and was involved in PEX7 degradation when abnormal PEX7 was present on the peroxisomal membrane, suggesting that RAB8A may be involved in peroxisomal biogenesis in plants (Cui et al. 2013). In addition, five members of RAB8 have shown interactions with an Arabidopsis phosphatidylinositol-4-phosphate 5-kinase 2 (PIP5K2). Interactions of the fluorescently tagged PIP5K2 and RAB8C at the PM can cause the redistribution of RAB8C from the Golgi apparatus (Camacho et al. 2009). The RAB8 GTPase also showed interactions with the stomatal cytokinesis defective complex, which in turn interacted with exocyst components and mediated exocytosis during cytokinesis (Mayers et al. 2017).

Plant small GTPase, the RAB protein, is a vital player in various fundamental cellular activities and is involved in different regulatory processes of plant development and stress responses (Nielsen 2020). In Arabidopsis, the RAB8C showed the highest expression in rosette leaves, followed by the RAB8A (Speth et al. 2009). Co-suppression of the endogenous RAB8C in transgenic Arabidopsis plants severely affected plant size and leaf morphology (Speth et al. 2009). Additionally, virus-induced gene silencing of Nicotiana benthamiana RABE1 in infiltrated N. benthamiana plants altered leaf development, caused abnormal stomata and resulted in growth retardation and premature senescence. Ectopic expression of the dominant-negative mutant of NbRABE1 in transgenic Arabidopsis plants caused growth arrest of shoots and roots and decreased root hair formations, indicating an important role of RAB8 GTPase in plant growth and development (Ahn et al. 2013). The two-dimensional gel electrophoresis and transcriptional analysis results showed that the RAB8 gene was rapidly and transiently induced by ethylene in rosette leaves of Arabidopsis, and its gene expression level was affected in ethylene signaling mutants of Arabidopsis, suggesting that RAB8 GTPase plays a role in ethylene signal transduction pathways (Moshkov et al. 2003). In addition to its roles in plant development, RAB8 GTPase has been shown to be involved in abiotic and biotic stress responses in plants (Bogdanove and Martin 2000, Speth et al. 2009, Zhang et al. 2018). The RabE1b from Populus trichocarpa was significantly induced by salt stress, and overexpression (O/E) of the constitutively active (CA) mutant of PtRabE1b(Q74L) in transgenic poplar can enhance its salt tolerance (Zhang et al. 2018). In tomato, two RAB8 proteins, Api2 and Api3, showed interactions with an effector AvrPto of a bacteria pathogen Pseudomonas syringae pv. tomato DC3000 (Pst DC3000) (Bogdanove and Martin 2000). O/E of the CA mutant of Arabidopsis RAB8C resulted in resistance to P. syringae infection, indicating the potential role of RAB8 GTPase in plant pathogen interactions (Speth et al. 2009).

In this study, we further identified two additional RAB8 proteins, RAB8A and RAB8D, that interacted with the A. tumefaciens VirB2 protein and several Arabidopsis RTNLB proteins. Furthermore, change in RAB8A, 8B or 8D gene expression levels in O/E transgenic plants or in T-DNA insertion mutants influenced the A. tumefaciens-mediated transformation efficiencies and P. syringae infection rates. This study further reveals direct involvements of RAB8A 8B, and RAB8D in plant-micobe interactions.

Results

RAB8A, 8B or 8D interacted with A. tumefaciens VirB2 and Arabidopsis RTNLB2-4, 8 in yeasts, in vitro and in plant cells

The Arabidopsis RAB8B protein was shown to interact with the C-terminal-processed portion of VirB2 in yeast two-hybrid and in vitro assays in a previous study (Hwang and Gelvin 2004). We therefore examined if other members of the RAB8 protein family, RAB8A and 8C-8E, could interact with A. tumefaciens Vir proteins. Information regarding the RAB8A-8E genes is listed in Supplementary Table S1. In the previous study by Vernoud et al. (2003), the RAB8D gene was designated as At4g20360 and was predicted to encode a nuclear-transcribed, plastid-localized EF-Tu translation elongation factor on the basis of its sequence information. However, other publications and wider community usage assigned At5g59840 as the RAB8D gene (Supplementary Table S1; Rutherford and Moore 2002). In this study, we have specified At4g20360 as the RAB8D-4g gene (Supplementary Table S1).

We cloned the RAB8A-8E and the RAB8D-4g genes from Arabidopsis cDNA and examined if the corresponding proteins can interact with the A. tumefaciens VirB2 bait protein in yeast two-hybrid assays. Fig. 1 results show that the RARB8A and 8D prey proteins interacted with the VirB2 bait protein in yeasts. The RAB8B interacted with VirB2 in yeasts and was used as the positive control in yeast two-hybrid assays (Fig. 1; Hwang and Gelvin 2004). None of tested Vir proteins, including VirB1, ViB1*, VirB5 (the minor component of T-pili), VirD2, VirE2, VirE1 and VirF, interacted with any members of the RAB8 protein family (Supplementary Fig. S1). The RAB8A-8E and RAB8D-4g bait proteins showed no interactions with the empty vector (Fig. 1). Similarly, the unrelated Lamin C bait protein did not interact with the RAB8A-8E or RAB8D-4g prey proteins and was used as a negative control.

In a previous publication, the RAB8B bait protein was shown to interact with RTNLB1, 2 and 4 (Hwang and Gelvin 2004). Moreover, other members of the RTNLB protein family, RTNLB3 and 8, showed interactions with A. tumefaciens VirB2 and RTNLB1, 2 and 4 (Huang et al. 2018). We therefore examined if RAB8 protein family members interact with themselves and/or RTNLB1-4 and 8 in yeast. Results shown in Fig. 1 suggest that RAB8A and 8D as bait fusion proteins showed interactions with RTNLB2, 4 and 8 but not with themselves. Besides RTNLB1, 2 and 4, the RAB8B bait fusion protein interacted with the RTNLB8 prey fusion proteins (Fig. 1). Results shown in Fig. 1 revealed that RTNLB1-4 and 8 as bait fusion proteins did not interact with RAB8A, 8B or 8D prey fusion proteins in yeasts. The results presented in Fig. 1 and Supplementary Table S2 show that when tested bait proteins were swapped with prey proteins, some of positive yeast two-hybrid interactions were not observed. These inconsistent findings may be due to different conformations of the bait and prey fusion proteins in yeast, as previously reported for interactions between RTNLBs and RAB8B (Hwang and Gelvin 2004, Huang et al. 2018). The ß-Galactosidase activity assays were used to quantify the interaction strengths in these yeast strains. Yeast colonies with white colors on synthetic dropout (SD) media with X-gal substrates also displayed zero ß-galactosidase activities (Supplementary Table S2), suggesting results obtained from plate-based yeast two-hybrid assays can be validated by liquid-based ß-galactosidase activity assays. Yeast strains expressing the RTNLB4 prey fusion with RAB8A, 8B or 8D bait fusions had relatively higher ß-galactosidase activities than the yeast strains expressing the RTNLB2 or 8 prey proteins with the same tested bait proteins, indicating that the interaction strengths might be higher among RTNLB4 with other RAB8 proteins (Supplementary Table S2).

On the basis of the analyses performed with the protein 2D structure prediction program SABLE (Wagner et al. 2005), the RAB8A protein was shown to contain six α-helixes and six β-sheets. In order to determine which regions of RAB8A interact with the A. tumefaciens VirB2 protein, we generated five internal deletion mutants of RAB8A and fused them with the activation domain or the DNA-binding domain to create prey or bait fusion proteins, respectively (Supplementary Fig. S2). We designated the RAB8A protein sequence into three regions. The first region, which covered the 12th amino acid (a.a.) to the 73th a.a., included the 1st α-helix and 1st to 3rd β-sheets. The second region was from the 75th to 139th a.a and contained the 2nd to 3rd α-helixes and 4th to 5th β-sheets. The third region was from the 140th to 206th a.a. and included the 4th to 5th α-helixes and the 6th β-sheet. None of the five internal deletion mutants of RAB8A interacted with VirB2 or RTNLB1-4, 8 (Supplementary Fig. S2), indicating that each of three regions may be important for interactions between RAB8A and tested proteins. It is also possible that deletions of these regions may affect the RAB8A protein conformation and, therefore, disrupt interactions between RAB8A and examined proteins.

In order to examine direct interactions of RAB8 proteins with VirB2 or RTNLB proteins, in vitro glutathione-S-transferase (GST) pull-down assays were used. As a negative control, the GST only protein did not interact with VirB2, RAB8A, 8B, 8D and RTNLB1-4, 8 proteins (Fig. 2A). Fig. 2B results show that the GST-VirB2 fusion protein interacted with the T7-tagged-RAB8A and 8B proteins, whereas the GST-RAB8B and 8D fusion proteins, not the GST-RAB8A, showed interactions with the T7-tagged-VirB2 protein. The in vitro protein interactions between RAB8 and RTNLB1-4, 8 were also examined with GST pull-down assays using the GST fusions and T7-tagged versions of the RAB8 and RTNLB1-4, 8 proteins. GST pull-down assay results indicated that only the GST-RAB8B fusion protein interacted with the T7-tagged-RTNLB1 and vice versa (Fig. 2C). The GST pull-down assay results showed that GST-RAB8B and 8D showed interactions with the T7-tagged-RTNLB2, whereas the GST-RTNLB2 only interacted with T7-tagged-RAB8B (Fig. 2D). The results shown in Fig. 2E indicated that GST-RAB8A, 8B and 8D interacted with the T7-tagged-RTNLB3. Similarly, GST-RAB8A, 8B and 8D interacted with the T7-tagged-RTNLB4, and GST-RTNLB4 showed interactions with the T7-tagged-RAB8A and 8B (Fig. 2F). Finally, the GST-RTNLB8 only interacted with the T7-tagged-RAB8B and 8D (Fig. 2G). The yeast two-hybrid assay and GST pull-down assay results are summarized in Supplementary Table S3. The GST pull-down assay results showed that RAB8B fusion proteins interacted most with five tested RTNLB proteins, including RTNLB1-4, 8, followed by the RAB8D fusion protein interacted with RTNLB2-4, 8 and the RAB8A fusion protein interacted with RTNLB3 and 4 (Supplementary Table S3). Interestingly, we observed more positive interactions with GST pull-down assays than with yeast two-hybrid assays (Supplementary Table S3). Similar discrepancies were observed in previous studies (Hwang and Gelvin 2004, Huang et al. 2018). The RTNLB and RAB8 proteins may localize in plant endomembrane systems (Rutherford and Moore 2002, Vernoud et al. 2003, Tolley et al. 2008, Sparkes et al. 2010) Therefore, various fusions of RTNLB and RAB8 proteins in yeast two-hybrid and GST pull-down assays may not form similar conformations as native RTNLB and RAB8 proteins in plant cells.

Bimolecular fluorescence complementation (BiFC) assays were used to visualize the interactions between RTNLB and RAB8 proteins in plant cells. As negative controls, only the N-terminal or C-terminal part of the enhanced YFP protein fused with the RAB8A showed no interactions with the cEYFP-or nEYFP-empty vectors (Fig. 3A-1, 2, 3B-1, 2). The CFP-Golgi and mCherry-ER markers were used as positive indicators of successful protoplast transformation (Fig. 3A-1, 2, 3B-1, 2). Figs. 3A-3 to 3A-6 show that the RAB8A protein interacted with RTNLB2, 3, 4, or RTNLB8 and YFP signals co-localized with the CFP-Golgi signals. Furthermore, Fig. 3B-3 to 3B-6 indicate that the interacting YFP protein signals colocalized with the mCherry-ER signals, suggesting that the RAB8A can interact with RTNLB2-4, 8 in the Golgi apparatus and ER networks. Similarly, the RAB8B and RAB8D showed interactions with RTNLB2-4, 8 in Golgi apparatus and ER networks (Supplementary Figs. S3, S4).

Previous studies have demonstrated that RTNLB1, 2 and 4 interacted with themselves and each other (Hwang and Gelvin 2004, Sparkes et al. 2010). In this study, we performed multicolor BiFC experiments in which RTNLB4 was tagged with cCFP and other tested RTNLB partners were tagged with either nVenus or nCerulean (Supplementary Fig. S5). Interactions of the RTNLB4 protein with tested nVenus-RTNLB proteins could generate green fluorescence, whereas interactions of the RTNLB4 protein with nCerulean-RTNLB could generate blue fluorescence (Supplementary Fig. S5). As expected, the RTNLB4 could simultaneously interact with RTNLB1/2, RTNLB1/4 or with RTNLB2/4 (Supplementary Fig. S5), whereas the RTNLB2 could simultaneously interact with RTNLB1/2, indicating the positive interactions among RTNLB1, 2 and 4 in plant cells. Because the green and blue fluorescence overlapped with the red fluorescence generated by the mCherry-ER marker, the RTNLB protein complex may localize in ER networks (Supplementary Fig. S5). Supplementary Fig. S6 show that the RTNLB4 could show simultaneous interactions with the RTNLB1, 2 or 4 and the RAB8B in ER networks. Additionally, the RTNLB2 protein could interact with RTNLB1/2 and the RAB8B protein simultaneously (Supplementary Fig. S6). When we swapped cCFP fluorescent tags on the RAB8A, 8B and 8D proteins, the RAB8 proteins showed interactions with the RTNLB1/2, RTNLB1/4 and RTNLB2/4 simultaneously (Supplementary Fig. S7). These data also suggest that tested RAB8 proteins may interact with multiple RTNLB proteins in plant cells.

The VirB2 protein is expressed in A. tumefaciens and is assembled into the A. tumefaciens T-pilus to mediate the T-DNA and Vir protein transfer into plants, so far there is no study to directly demonstrate that the VirB2 protein can be transferred into plant cells. Therefore, we did not test the interactions between the VirB2 and the RAB8 or RTNLB proteins in plant cells by BiFC assays.

The Arabidopsis rab8a, rab8b and rab8d mutants were resistant to A. tumefaciens-mediated transformation

The RAB8A, 8B and 8D showed interactions with A. tumefaciens VirB2. Therefore, we determined whether RAB8 proteins are involved in the A. tumefaciens infections. The T-DNA insertion Arabidopsis mutants (ecotype: Columbia) in RARB8A-8E or RAB8D-4g genes were obtained (Supplementary Table S1) and the susceptibilities of various rab8 mutants were tested in root- and seedling-based A. tumefaciens infection assays. At least one T-DNA insertion homozygous mutant was identified for each of the RAB8A-8E and RAB8D-4g genes (Supplementary Table S1; Fig. 4A-1 to 4A-6). The T-DNA insertion sites of the rab8a-8e and rab8d-4g single mutants are summarized in Supplementary Table S1 and Fig. 4A-1 to 4A-6. Quantitative real-time polymerase chain reaction (Q-PCR) results showed that RAB8 target gene transcript levels were reduced more than 25% or were not detectable in the rab8a-8e or rab8d-4g single mutants, suggesting that T-DNA insertions in these mutants may significantly affect the target RAB8 gene transcript stabilities and accumulations (Fig. 3B-1 to 3B-6). We next performed stable and transient A. tumefaciens-mediated root transformation to determine transformation rates of these rab8 mutants. Fig. 4C results showed that rab8a-1, 8a-2, 8a-3, rab8b-1 and rab8d-1, 8d-2 mutants had relatively lower stable and transient transformation rates than wild-type plants, whereas the other examined rab8 mutants showed comparable transformation rates as wild-type plants. These results suggested that the transformation process may be affected at steps occurring before T-DNA integration in the rab8A, rab8B and rab8D single mutants. We also generated two rab8b-1 complementation lines, rab8b-1-16C and rab8b-1-17C. Q-PCR and transformation assay results showed that RAB8B transcript levels and transformation efficiencies in the two rab8b-1 complementation lines were restored to wild-type levels, suggesting that the resistance to A. tumefaciens infection phenotypes in the rab8b-1 mutant may be due to lower RAB8B transcript levels in plants (Fig. 4B-2, 4C). We used more sensitive Arabidopsis seedlings-based transient transformation assays to examine rab8a, rab8b and rab8d mutant susceptibilities to A. tumefaciens infections (Wu et al. 2014, Huang et al. 2018). Fig. 4D results showed that β-glucuronidase (GUS) activities decreased more than 40% in all the tested rab8 mutants in comparison with that in wild-type plants, whereas two rab8b-1 complementation lines showed similar GUS activity levels as the wild-type plants. In summary, these results suggested that resistant phenotypes of three rab8a, two rab8d and the rab8b-1 mutants correlated well with reduced levels of RAB8 gene expression. In summary, these results suggest that RAB8A, 8B and 8D may participate in early step(s) of the A. tumefaciens-mediated transformation process and that RAB8C, 8E and RAB8D-4g may not be directly involved in the A. tumefaciens infections.

The RAB8A, RAB8B and RAB8D O/E transgenic plants showed higher A. tumefaciens-mediated transformation efficiencies

Since the rab8a, rab8b and rab8d single mutants showed resistance to A. tumefaciens infections, we examined if O/E of RAB8A, 8B or 8D in Arabidopsis plants (ecotype: Wassilewskija [Ws]) could increase their susceptibilities to A. tumefaciens infections. Because the Arabidopsis Ws plants show more prominent phenotypes than Columbia plants when infected with the A. tumefaciens (Zhu et al. 2003), we utilized Arabidopsis Ws plants to generate RAB8 O/E transgenic plants. Q-PCR results indicated that RAB8 transcript levels increased more than 2.5-fold in two RAB8A and RAB8B O/E transgenic plants and more than 1.6-fold in three RAB8D O/E transgenic plants (Fig. 5A-1 to 5A-3). Protein gel blot analysis using the anti-T7-tag antibody showed that T7-tagged-RAB8A, 8B or 8D recombinant proteins were highly accumulated in the T7-tagged-RAB8 O/E transgenic plants (Fig. 5B-1 to 5B-3). We then used relatively lower concentrations of A. tumefaciens, 10⁵ and 10⁶ cfu (colony forming unit)· mL⁻¹ to infect roots of RAB8A, 8B and 8D O/E transgenic plants. In Fig. 5C, RAB8A, 8B and 8D O/E transgenic plants showed 1.8- to 4.1-fold enhanced tumor formation rates and 1.4- to 4.3-fold increased transient transformation rates compared to wild-type plants. Similarly, when 10⁵ cfu · mL⁻¹ of A. tumefaciens were used to infect the RAB8 O/E transgenic plant seedlings, the RAB8A and T7-tagged-RAB8A O/E transgenic plants had 2.2- to 3.0-fold increased GUS activities; the RAB8B and T7-tagged-RAB8B O/E transgenic plants had 3.3- to 7.8-fold increased GUS activities, whereas the RAB8D and T7-tagged-RAB8D O/E transgenic plants had 3.3- to 10.2-fold increased GUS activities in comparison with wild-type plants (Fig. 5D). These data showed that when the RAB8A, 8B or 8D genes were overexpressed in transgenic plants, A. tumefaciens-mediated transformation rates increased. Furthermore, the findings indicated that the presence of the T7 tag sequence in the N-terminal region of the RAB8 protein may not affect RAB8 protein functions during A. tumefaciens infections. These data further support the importance of RAB8A, 8B and 8D during A. tumefaciens infections.

RAB8A, 8B and 8D O/E transgenic plants were hypersusceptible to P. syringae infection

Because a previous study showed that five RAB8 proteins interacted with the Pst DC3000 effector AvrPto (Speth et al. 2009), we next examined whether O/E of RAB8A, 8B or 8D could affect the plant susceptibility of Pst DC3000. Wild-type Pst DC3000 and the hrcC mutant, a type III secretion system-defective bacteria used as the negative control, were syringe-infiltrated into Arabidopsis wild-type and O/E transgenic plants. Bacteria viable numbers in infected leaves were determined after 0-, 1-, 3-, 5-, 7-day infection. Results shown in Fig. 6A-1, 6B-1 and 6C-1 demonstrated that bacteria numbers of the wild-type Pst DC3000 in wild-type plants increased more than 10³-fold 5-days post-infection, suggesting successful infection with P. syringae in plants. On the other hand, the bacteria numbers of the hrcC mutant only slightly increased in the wild-type and O/E transgenic plants (Fig. 6A-2, 6B-2, 6C-2). Three days after infection with the Pst DC3000, bacteria numbers in the RAB8A, 8B and 8D O/E transgenic plants were at least 10-fold higher than wild-type plants (Fig. 6A-1, 6B-1, 6C-1). After being infected with the hrcC mutant, both wild-type and RAB8 O/E plants showed no difference in bacteria growth (Fig. 6A-2, 6B-2, 6C-2). We also infected three rab8a mutants, rab8b-1, two rab8b-1 complementation lines and two rab8d mutants with the wild-type Pst DC3000 to determine their susceptibilities to P. syringae. Results shown in Supplementary Fig. S8 indicated that all tested rab8 mutant plants had relatively lower viable bacterial numbers than wild-type plants from 1 d after infected with Pst DC3000, whereas two rab8b-1 complementation lines, rab8b-1-16C and rab8b-1-17C, had similar viable bacteria numbers as the wild-type plants. These data indicate that abnormally high or low expression levels of RAB8A, 8B, or 8D genes may affect plant susceptibility to Pst DC3000 infection and suggest that RAB8 proteins may play important roles in plant P. syringae interactions.

The RAB8A-E genes were expressed differently in various plant tissues

A previous study indicated that five RAB8 genes accumulated differently in Arabidopsis rosette leaves (Speth et al. 2009). We utilized the Arabidopsis eFP browser (Winter et al. 2007) to examine tissue-specific expression patterns and levels of five RAB8 genes. The eFP browser expression predictions of RAB8A-8E showed that five RAB8 genes were expressed in all Arabidopsis tissues and developmental stages at different levels (Supplementary Fig. S9). In order to further examine expression levels of RAB8A-8E in plants, the Q-PCR was conducted with gene-specific primers of RAB8A-8E to determine their transcript levels in the root, rosette leaf, cauline leaf, inflorescence, and flower of wild-type Arabidopsis thaliana (ecotype: Columbia). In five examined tissues, the RAB8A transcript levels were relatively higher than the other four RAB8 genes, whereas the RAB8C transcript levels were the lowest among five RAB8 genes (Fig. 7). The transcript levels of RAB8A, 8B, 8D and 8E in roots were relatively lower than same genes in tissues above ground (Fig. 7). In the rosette leaf, cauline leaf, inflorescence and flower, the RAB8B, 8D and 8E accumulated at similar levels (Fig. 7). We also examined the transcript levels of five RAB8 genes in A.thaliana ecotype Wassilewskija (Ws), because we have used wild-type Ws plants to generate O/E transgenic plants. Similar to results obtained in Columbia plants, the RAB8C transcript levels were the lowest among all five RAB8 genes in all tested tissues (Supplementary Fig. S10). In tissues above ground, the RAB8A, 8B and 8D gene transcript levels were higher than those of the other two RAB8 genes (Supplementary Fig. S10). The five RAB8 gene transcript levels were relatively higher in tissues above ground than in roots (Supplementary Fig. S10). Interestingly, the five RAB8 gene transcript levels were lower in Ws plants than in Columbia plants (Supplementary Fig. S10). These data indicated that all five RAB8 genes were preferentially expressed in different tissues and may play different roles during plant growth and development.

Because the Arabidopsis RAB8 is similar to the Saccharomyces cerevisiae Sec4 protein that functions in the terminal step of the secretory pathway, i.e. delivery of Golgi-derived transport vesicles to the cell surface (Novick et al. 1980, Salminen and Novick 1987), we next investigated if five RAB8 genes could functionally complement the temperature-sensitive yeast secretion (sec) sec4 mutant (Novick et al. 1980). We transformed the Sec4 mutant/NY405 with either the empty vector (V) pYX, which was used a negative control or the pNB136 plasmid (P) containing the genomic DNA fragment of the Sec4 gene from yeast, which was used as a positive control or the pYX plasmid containing the RAB8A-8E genes. In order to induce RAB8A-8E protein expressions in yeast transformants, 2% galactose and 1% raffinose were added in SD media. As expected, the Sec4 mutants showed normal growth at 25°C and 30°C in SD media with 2% glucose (Supplementary Fig. S11A-1, 11B-1), with 2% galactose (Supplementary Fig. S11A-2, 11B-2) or with 2% galactose/1% raffinose (Supplementary Fig. S11A-3, 11B-3). However, only the Sec4 mutant with the pNB136 plasmid (P) can grow at 37°C, not the Sec4 mutant with any tested RAB8 gene (Supplementary Fig. S11C-1 to S11C-3). These data indicated that none of the five RAB8 genes could functionally complement the yeast secretion (sec) sec4 mutant, suggesting that plant RAB8 genes might be more highly evolved than the yeast homologous Sec4 gene.

Discussion

The small GTP-binding protein RAB8 participated in polarized vesicle transport from the trans-Golgi network to the PM in various organisms (Goud et al. 1988, Craighead et al. 1993, Huber et al. 1993, Rutherford and Moore 2002). In this study, the AtRAB8A, 8B and 8D proteins interacted with the Arabidopsis RTNL proteins AtRTNLB1-4 and AtRTNLB8 in three protein–protein interaction assays. Furthermore, the BiFC results showed that three RAB8 co-localized with RTNLBs at ER networks and the Golgi apparatus in plant cells. The RTNLBs are involved in ER tubular structure formations and play important roles in vesicle formations for intracellular trafficking (Oertle and Schwab 2003, Tolley et al. 2008, Nziengui and Schoefs 2008, Sparkes et al. 2010). Data in this study further support the involvement of RAB8 in endomembrane trafficking through interactions with other endomembrane localized proteins in plant cells. Furthermore, the AtRAB8A, 8B and 8D interacted with the A. tumefaciens VirB2 protein in yeast two-hybrid and GST pull-down assays. When the individual RAB8A, 8B or 8D genes were over- or under-expressed, the infection rates of A. tumefaciens and Pst DC3000 were affected. These data collectively indicated that three RAB8 proteins may play a part in plant-pathogenic bacteria interactions.

In this study, we have observed that interactions between the RAB8A, 8B, 8D as bait fusion proteins and several RTNLB proteins as the prey fusion proteins. However, when tested bait fusion protein were swapped with prey proteins, we did not observe some of positive yeast two-hybrid interactions. Previous studies by a large-scale yeast two-hybrid analysis to study the yeast (Saccharomyces cerevisiae) protein interactome showed that yeast proteins involved in the autophagy, spindle pole body function and membrane fusion steps of vesicular transport process only interacted unidirectional not when the tested bait and prey fusion protein were exchanged (Ito et al. 2000, 2001), which is similar to our observation in this study. Furthermore, we have observed similar interaction results between RAB8B and RTNLB1, 2, 4 (Hwang and Gelvin 2004) and interaction results between RTNLB3, 8 with RTNLB1, 2, 4 in the yeast two-hybrid analysis (Huang et al. 2018). These results might indicate that careful design of tested bait and prey fusion protein combinations may be important in the yeast two-hybrid analysis.

We further discovered the discrepancies in RAB8 and RTNLB interaction results in yeast two-hybrid assays and GST pull-down assays. Similarly, previous studies have demonstrated that the Arabidopsis RABA4B showed interactions with the full-length PLANT U-BOX13 (PUB13) protein and the phosphatidylinositol 4-OH kinase, PI-4Kβ1, in the biochemical assays but no interactions were observed in the yeast two-hybrid assays (Preuss et al. 2006, Antignani et al. 2015). Another study also revealed that the deletion of the C-terminal prenylation sites of the RAB8C/RABE1d to prevent membrane association increased the interaction strength of RAB8C/RABE1d and the PIP5K2 protein, perhaps by facilitating nuclear rather than membrane localization of the bait protein during the yeast two-hybrid assays (Camacho et al. 2009). Both RAB8 and RTNLB proteins are membrane proteins, associated with the endomembrane system and are involved in vesicle trafficking process in cells (Oertle and Schwab 2003, Zheng et al. 2005). Therefore, it is possible that membrane-associated properties of the RTNLB and RAB8 protein might affect the nuclear localization of the interacting partner to activate the reporter genes in traditional yeast two-hybrid assays (Stagljar et al. 1998; Stagljar and Fields 2002). Additionally, based on the amino acid sequences of the RTNLB proteins, they contained two putative transmembrane regions in the C terminal RHD domain that help the RTNLB protein to form a wedge-like conformation and insert into the enodmembranes (Tolley et al. 2008, Nziengui and Schoefs 2008, Sparkes et al. 2010). In the bacterial expression system, the conformation of the RTNLB protein might not be same as in the eukaryotic cells that may therefore affect the interaction studies of GST pull-down assays. Due to limitations and drawbacks of the yeast two-hybrid and GST-pull down assays, we further examined the interactions between the RAB8 and RTNLB proteins in plant cells by BiFC assays.

In this study, we have demonstrated that the RTNLB1, 2 and 4 could simultaneously interact with themselves and the other two RTNLB proteins in ER networks of plant cells by multicolor BiFC results. This finding is consistent with previous published results showing that RTNLB1-4 and 13 can interact with each other, colocalize and help constrict tubular ER structure formations (Tolley et al. 2008, 2010, Sparkes et al. 2010). Additionally, BiFC results showed that the RAB8A, 8B and 8D could concurrently interact with three RTNLB proteins in the endomembrane structures, including the ER and Golgi apparatus, of plant cells although some of the tested RTNLB fusion proteins showed no interactions with three RAB8 proteins in yeast and in vitro. These discrepancies might be due to differences in the conformations or expression levels of tested fusion proteins in three analysis systems.

Previous studies have shown that O/E of AtRTNLB1, 3, 4 and 8 in Arabidopsis affect A. tumefaciens-mediated transformation efficiencies and plant susceptibilities to Pst DC3000 infection (Hwang and Gelvin 2004, Lee et al. 2011, Huang et al. 2018, Huang and Hwang 2020). Furthermore, the plant defense gene expressions, H₂O₂ production and seedling growth inhibition induced by A. tumefaciens elf18 and VirB2 peptides were influenced by the abnormal expression levels of AtRTNLB4 in transgenic and mutant plants, indicating that AtRTNLB4 may play important roles in elf18 and VirB2 peptide-induced defense responses (Huang and Hwang 2020). Additionally, AtRTNLB1 and 2 showed interactions with the Arabidopsis Flagelin-sensitive2 (FLS2) protein, one of the pattern recognition receptors (PRRs) for the bacterial flagellin of Pst DC3000 (Lee et al. 2011). The translocation of the newly synthesized FLS2 from the ER to the PM and the FLS2-induced plant immune responses were mediated by the AtRTNLB1 and 2 (Lee et al. 2011). These data indicated the involvement of RTNLBs in plant defense responses during A. tumefaciens and Pst DC3000 infection. Because the RAB8A, 8B, 8D proteins showed interactions with the Arabidopsis RTNLB proteins and the A. tumefaciens VirB2 protein, possibilities of involvements of the RAB8A, 8B and/or 8D in the A. tumefaciens VirB2 peptide-induced defense responses await further investigation.

Consistent with these published observations, reductions and O/E of the RTNLB-interacting proteins, the RAB8A, 8B and 8D in plants influenced the infection rates of A. tumefaciens and Pst DC3000, suggesting their vital roles in interactions between plants and pathogenic bacteria. Other studies have also shown that other AtRAB subgroups were involved in plant-defense responses against other phytopathogen infections, such as the participation of AtRABA/AtRAB11 in the transport of de novo-synthesized FLS2 to PM (Choi et al. 2013) and suggested that it may be a host target of the RxLR24 effector of Phytophthora brassicae (Tomczynska et al. 2018). Additionally, the RabG3b may be a positive regulator of autophagy and promoted hypersensitive cell death in response to the avirulent bacteria Pst DC3000 (AvrRpm1) in Arabidopsis (Kwon et al. 2013). A plant-specific RAB5 GTPase, ARA6/RABF1, accumulated at specialized membranes, called the extrahaustorial membranes, that surrounded the infection hyphae in Arabidopsis cells infected by biotrophic fungi or oomycete (Inada et al. 2016). These data along with our study results indicated that RAB GTPase in plant endomembrane trafficking systems play critical roles when plants encounter bacteria, fungi and oomycete infections.

The RAB small GTPase also affected the virus movement. The transport of the Semliki forest virus (SFV) from the early endosomes to late endosomes required RAB7 assistance. When the dominant-negative RAB7 mutant was expressed in vero cells, it caused the accumulation of SFV in early endosomes (Vonderheit and Helenius 2005). Another member of the RAB7 subgroup, NaRABG3f from Nicotiana benthamiana, was involved in the movement of the Bamboo mosaic virus from the Golgi membrane to its replication site (Huang et al. 2016). The triple-gene block proteins of the Potato mop-top virus that are important for virus movement can localize in vesicles with ARA7/RAB5, suggesting that virus intracellular movement may utilize the host cell endocytic trafficking system (Haupt et al. 2005). Additionally, the secretory pathway is crucial for the intercellular movement of the Turnip mosaic virus. Expression of dominant-negative mutants of the RABE1d in plants reduced virus intercellular movement from post-Golgi to the PM (Agbeci et al. 2013). Previous studies have demonstrated that pathogens can exploit the host endocytosis or autophagy systems by affecting RAB protein functions to achieve successful infection (Rivero et al. 2019, Elliott et al. 2020). Furthermore, the plant endocytosis system and ER-actin networks assisted nuclear targeting of the A. tumefaciens VirE2 protein during infection (Li and Pan 2017, Yang et al. 2017, Tu et al. 2018). Therefore, it is possible that the RAB8A, 8B and 8D may influence the infection process of A. tumefaciens and P. syringae by affecting the intracellular movement in plant cells.

A previous study showed that RAB8C/RABE1d had the highest expression level in Arabidopsis ecotype Columbia (Col-0) rosette leaves (Speth et al. 2009) by RT-PCR analysis, which is consistent with the expression pattern of RAB8C in the Arabidopsis eFP browser (Supplementary Fig. S9). However, in our Q-PCR results, the RAB8C transcript level was relatively lower than those of the other four RAB8 genes in roots, leaves and flowers. These inconsistent results may be due to differences in the analysis approaches and/or differences in the growth stages of plant materials used in these studies. Notably, we observed that transcript levels and expression patterns of five RAB8 genes were different in Arabidopsis ecotype Ws plants compared to Columbia plants. These differences in the two ecotypes may be caused by ecotype-specific expressions and/or alternative splicing of examined genes (Lemoine 2000, Sauer et al. 2004, Feuerstein et al. 2010).

The small GTPase RAB8 protein family consists of 57 members in Arabidopsis, which are important regulators of intracellular membrane trafficking (Rutherford and Moore 2002, Vernoud et al. 2003). However, the precise functions and the functional redundancy of each member in the eight RAB subfamilies remain largely unknown. In our study, the rab8a, rab8b and rab8d single mutants were less susceptible to A. tumefaciens and Pst DC3000 infections, implying that RAB8A, 8B and 8D may not show functional redundancy during pathogen infection. Prior studies have also revealed unique functions among different family members of small GTPases. In zebrafish, the small GTP-binding protein, Arf-like (Arl), was expressed during early cell division and gastrulation and regulated the cilia biogenesis process (Song and Perkins 2018). The arl13b mutation resulted in retinal degeneration, and the Arl13a, a close paralog of Arl13b, could not fully compensate for Arl13b function in zebrafish retina (Song and Perkins 2018). Another study indicated that two RAB6 isoforms, RAB6a and RAB6a’, regulated retrograde transport to the trans-Golgi network and performed distinct functions in mammal cells (Mallard et al. 2002). Recent research results showed that alternative splicing of stress-related genes increased protein and transcript diversities to help plants to cope with pathogen infection (Liu et al. 2016, Zhang et al. 2019). In conclusion, although the RAB8A, 8B and 8D shared high sequence similarities, they may play distinct roles when plants confront phytopathogens.

Materials and Methods

Bacterial strains and culture

The Agrobacterium tumefaciens and Escherichia coli strains used in this study are summarized in Supplementary Table S4. A. tumefaciens strains were cultured in 523 medium or on 523 agar with appropriate antibiotics (rifampicin 50 μg · mL⁻¹, gentamycin 50 μg · mL⁻¹, kanamycin 20 μg · mL⁻¹, carbenicillin 50 μg · mL⁻¹ or tetracycline 20 μg · mL⁻¹) at 28°C, while all E. coli strains were grown in 2X YT medium with antibiotics (ampicillin 100 μg · mL⁻¹, or kanamycin 50 μg · mL⁻¹) at 37°C.

Yeast two-hybrid assays

The bait and prey plasmids used for the yeast two-hybrid assays are listed in Supplementary Table S4. The bait plasmid pSST91 (Huang et al. 2018) was used to express various LexA-RAB8 or LexA-RTNLB recombinant proteins that were under the control of the yeast ADH1 promoter. The prey plasmid pGAD424 (Huang et al. 2018) was used to generate different GAL4-RAB8 or GAL4-RTNLB recombinant proteins. In order to generate the bait and prey plasmids, the DNA fragments of the RAB8A-8E and RAB8D-4g coding sequences from Arabidopsis were obtained with PCR using Arabidopsis cDNA as templates, the high-fidelity Phusion DNA polymerase (New England Bio Labs Inc., Ipswich, MA, USA) and appropriate primers (Supplementary Table S5). The PCR products were digested with EcoRI/PstI (for DNA fragments containing the RAB8A, RAB8C or RAB8D), with EcoRI/BamHI (for the DNA fragment containing the RAB8E) or with BamHI/PstI (for the DNA fragment containing the RAB8D-4g), and cloned into same restriction enzyme sites of the pSST91 or the pGAD424 plasmids with in-frame fusion to the LexA or GAL4 coding sequences (Supplementary Table S4).

In order to generate the internal deletion mutants of RAB8A, the EcoRI-PstI fragment of the RAB8A coding sequences were acquired from PCR and subcloned into same restriction enzyme sites of the pBluescript to create the plasmid pBluescript-RAB8A (Supplementary Table S4). This plasmid was then used as the template to perform inverse PCR with the high-fidelity Phusion DNA polymerase and appropriate primers (Supplementary Table S5), which generated five different internal deletion mutants of the RAB8A (Supplementary Table S4). The purified PCR products were digested with SalI, after which the digested DNA fragments were self-ligated and transformed into competent E. coli DH10B cells (Supplementary Table S4). The successfully identified clones were confirmed by sequencing; the DNA fragments of RAB8A deletion mutants were then digested with EcoRI/PstI and the inserts were cloned into the same restriction enzyme sites of the pSST91 or the pGAD424. The resulting five different bait and prey plasmids that encode the RAB8A deletion mutants are listed in Supplemental Table S4.

The yeast strain CTY10-5d (Supplementary Table S4) was transformed with the bait and prey plasmids using a lithium acetate method. All yeast strains were grown in SD medium with yeast nitrogen-base, glucose, and all but the selective amino acids at 30°C. The yeast transformants were screened for protein interactions on the basis of the colony color phenotype on SD medium containing the chromogenic substrate X-Gal (5-bromo-4-chloro-3-indolyl-ß-D-galactopyranoside) but lacking leucine and tryptophan. The ß- galactosidase enzyme activity assays were also used to quantify yeast two-hybrid interactions. The liquid cultures of yeast transformants with ONPG (ortho-nitrophenyl-β-D-galactopyranosidase) as substrates were utilized to conduct enzyme activity assays as described (Huang et al. 2018). The one unit of ß-galactosidase enzyme activity was defined as the amount that can hydrolyzes 1 μmol of ONPG to o-nitrophenol and D-galactose per min per cell (Huang et al. 2018).

GST protein affinity purification assays

Plasmids and bacteria used for the GST pull-down assays are listed in Supplementary Table S4. The plasmids pGEX4T-1 or pET42a were used to express recombinant proteins fused in frame with the GST tag in the E. coli strain BL21(DE3). The plasmids pET23a and pET28a were used to generate and express the T7-tagged fusion proteins in bacteria. The PCR products containing the coding sequences of the RAB8A and RAB8D were obtained using the Arabidopsis cDNA as templates and appropriate primers (Supplementary Table S5); the PCR products were subsequently cloned into the pBluescript plasmid to create pBluescript-RAB8A and pBluescript-RAB8D (Supplementary Table S4). The EcoRI-NotI DNA fragments containing RAB8A coding sequences from the plasmids pBluescript-RAB8A were cloned into the pET23a or the pET42a as in-frame fusion to the T7 tag or GST coding sequence. Similarly, the EcoRI-NotI fragments from the pBluescript-RAB8D were cloned into the pET23a or the pGEX4T-1 to express the T7-tagged or GST fusion proteins in bacteria. Expression and purification of GST fusion proteins and affinity purification of proteins binding to GST fusion proteins were performed as described previously with minor modifications (Hwang and Gelvin 2004, Huang et al. 2018). The isolated protein complexes were analyzed in 12.5% SDS-polyacrylamide gels, and immunoblot analyses were performed using a 1:1000 dilution of anti-T7 tag primary antibody (Merck, Danvers, MA, USA) or using a 1:15000 dilution of anti-GST primary antibody (GE Healthcare, Piscataway, NJ, USA), followed with a 1:20000 dilution of goat anti-rabbit IgG conjugated to horseradish peroxidase (HRP) secondary antibody (PerkinElmer Life and Analytical Science, Boston, MS, USA), or with a 1:15000 dilution of donkey anti-goat IgG-HRP conjugated secondary antibody (Santa Cruz Biotechnology, Dallas, TX, USA), to confirm the identities of these fusion proteins. Finally, membranes were developed with chemiluminescent detection methods and subjected to autoradiography.

BiFC assays

Plasmids for the BiFC assays are listed in Supplementary Table S4. The BiFC plasmid pSAT4-nEYFP or pSAT4-cEYFP (Lee et al. 2008) was used to express various RAB8A, 8B, 8D or RTNLB2, 3, 4 or 8 recombinant proteins fused in frame with the N or C terminal part of yellow fluorescent protein in plant cells under the control of the Cauliflower mosaic virus (CaMV) double 35S promoter. The PCR products containing the coding sequences of the RAB8A, 8B, or 8D, RTNLB2, 3, 4 or 8 were obtained using the Arabidopsis cDNA as templates, the high-fidelity Phusion DNA polymerase and appropriate primers (Supplementary Table S5). The PCR products were then digested with EcoRI/BamHI (for DNA fragments containing the RAB8A, RTNLB3 or RTNLB8) or with XhoI/KpnI (for the DNA fragment containing the RAB8B, RAB8D or RTNLB2) or with XhoI/EcoRI (for the DNA fragment containing the RTNLB4) and cloned into same restriction enzyme sites of the pSAT4-n/cEYFP plasmids (Supplementary Table S4).

The multi-color BiFC plasmid pSAT1-nVenus-C/pE3228 (Lee et al. 2008) was used to express RAB8A, 8B or 8D or RTNLB1, 2 or 4 recombinant protein fused in frame with the N terminal part of the Venus fluorescent protein in plant cells. The multi-color BiFC plasmid pSAT4-cCFP-C/pE3243 (Lee et al. 2008) was used to generate and express different cCFP-RAB8 or cCFP-RTNLB recombinant proteins in plant cells. Similarly, the BiFC plasmid pSAT1-nCerulean-C/pE3415 was utilized to express various nCerulean-RAB8 or nCerulean-RTNLB recombinant proteins in plants. The PCR products containing the coding sequences of the RAB8A, 8B or 8D, RTNLB1, 2 or 4 were obtained using the Arabidopsis cDNA as templates, the DNA polymerase, and appropriate primers (Supplementary Table S5). The PCR products were then digested with EcoRI/KpnI (for DNA fragments containing the RAB8A), with XhoI/KpnI (for the DNA fragments containing the RAB8B and 8D) or with XhoI/EcoRI (for the DNA fragment containing the RTNLB1, 2, or 4), and cloned into same restriction enzyme sites of the pE3228, pE3243 or the pE3415 plasmids (Supplementary Table S4).

In order to perform BiFC assays in plant cells, the protoplast isolation and transfection were conducted according to the methods of Wu et al. (2009). In brief, leaves of 5–6 week-old Arabidopsis (ecotype: Columbia) plants were used to isolate mesophyll protoplast with the Tape-Arabidopsis Sandwich method. The lower epidermis of leaves was peeled away by the 3M Magic tape, while the upper epidermis was still adhered to the time tape. The strips of time tapes with upper epidermal cell layers of peeled leaves were then transferred into enzyme solutions with gentle shaking for 1 to 2 hours until protoplasts were released into solutions. The protoplasts were centrifuged and washed twice in W5 solutions (54 mM NaCl, 125 mM CaCl₂, 5 mM KCl, 5 mM glucose and 2 mM MES, pH 5.7) and were finally counted with a hemocytometer under a light microscope. The protoplasts were washed and resuspended in MMg solutions (0.4 M mannitol, 15 mM MgCl₂ and 4 mM MES, pH 5.7) to a final concentration of 4–5 × 10⁵ cells · mL⁻¹.

The isolated protoplasts were then transfected with desired plasmids (Supplementary Table S4) using 40% PEG (MW 4000; Fluka, Loughborough, UK) solutions (including 0.1 M CaCl₂ and 0.2 M mannitol) at room temperature for 5–6 min. After incubation, the protoplasts were washed two times in the W5 solution; finally, they were resuspended in W5 solutions and kept in 6-well plates at room temperature for 16 hours in light.

After protoplast transfections, protoplasts were observed using a Zeiss LSM510 META laser scanning confocal microscope with a multi-track channel mode. A series of filters was used to collect fluorescent images and a Nomarski differential interference contrast (DIC) lens was used to capture transmitted light images. Excitation wavelengths and emission filters were 488 nm/band-pass 505–530 nm for YFP or Venus, 458 nm/band-pass 465–530 nm for CFP or Cerulean and 543 nm/band-pass 565–615 nm for mCherry. Image analysis was conducted with a LSM 510 version 4.2 (Zeiss, Oberkochen, Germany).

DNA isolation from Arabidopsis plants and genomic DNA PCR analysis

The Arabidopsis T-DNA insertion mutants of rab8A-8E and rab8D-4g (ecotype: Columbia CS60,000) were identified with the SIGnAL ‘T-DNA Express’ Arabidopsis Gene Mapping Tool (http://signal.salk.edu/, Alonso et al. 2003). The rab8a-8e and rab8d-4g mutant seeds were acquired from the Arabidopsis Biological Resource Center (ABRC). Leaves of 3-week-old mutant plants grown in Gamborg’s B5 medium were used to isolate genomic DNA according to Dellaporta et al. (1983). A PCR-based approach (Alonso et al. 2003) and the SIGnAL ‘T-DNA Express’ Gene Mapping Tool (http://signal.salk.edu/) were used to determine the homozygosity of Arabidopsis rab8 mutants. The PCR was performed in a 50 μL reaction volume with two units of GenTaq polymerase (GMbiolab Co., Taiwan), a 2.5 mM dNTP mixture, the 1× Taq polymerase reaction buffer and 0.25 μm of the PCR primers, using the following amplification cycle: 95°C for 1 min (1 cycle); 94°C for 30 sec, 56°C for 40 sec, 72°C for 1 min (30 cycles) and 72°C for 5 min (1 cycle). Primers used for genomic DNA PCR analysis are listed in Supplementary Table S5.

RNA isolation from Arabidopsis plants and Q-PCR analysis

Root and aerial plant tissues from 4 to 5 week-old wild-type plants (ecotypes: Columbia and Wassilewskija [Ws]), rab8 mutant, rab8b-1 complementation plants (ecotype: Columbia) and RAB8 over-expression transgenic plants (ecotype: Ws) were used to isolate RNA. Plant tissues were then ground with liquid nitrogen and mixed with TRIZOL LS reagents (Total RNA Isolation Reagent for Liquid Samples from Invitrogen, Carlsbad, CA, USA) to isolate RNA according to the manufacturer’s instructions. Then, 1–3 μg of RNA was treated with DNase I (Thermo Fisher Scientific Inc., Waltham, MA, USA) according to manufacturer’s instructions, and reactions were stopped with additions of EDTA and heat inactivation.

The 1 μg RNA samples from various plants were reverse transcribed using the oligo-dT primer to generate cDNA. Then, 100 ng of cDNA was used to perform Q-PCR using IQ² SYBR Green Fast qPCR System Master Mix (Bio-genesis Technologies Inc., Taipei, Taiwan) in the MS3000P QPCR system (Agilent Technologies, Santa Clara, CA, USA). Primers used for Q-PCR in this study are listed in Supplementary Table S5. The UBQ10 (polyubiquitin 10) transcript level was used as an internal control in each Q-PCR. The PCR amplification cycle was 99°C for 1 min (1 cycle); 94°C for 30 sec, 56°C for 40 sec, 72°C for 1 min (50 cycles); 99°C for 1 min (1 cycle); 55°C for 3 min (1 cycle) and 95°C for 30 sec (1 cycle). More than three independent real-time PCRs were performed with RNA samples isolated from at least 6–12 different Arabidopsis plants.

Generation of RAB8A, RAB8B and RAB8D O/E A.thaliana transgenic plants and rab8b-1 complementation plants

In order to overexpress the RAB8A, RAB8B or RAB8D gene in Arabidopsis transgenic plants, the binary vector, pE1798, with a double CaMV 35S promoter, a Nos (nopaline synthase) terminator and a hptII (hygromycin resistance gene) gene as a selectable marker in the T-DNA region was used (Hwang and Gelvin 2004, Huang et al. 2018). The PCR products containing the coding sequences of the RAB8A, 8B and 8D (Supplementary Fig. S12) were obtained using the Arabidopsis cDNA as templates, the high-fidelity Phusion DNA polymerase and appropriate primers (Supplementary Table S5); the PCR products were subsequently cloned into the pBluescript plasmid to create pBluescript-RAB8A, pBluescript-RAB8B and pBluescript-RAB8D, respectively (Supplementary Table S4). The KpnI-SacI fragments from the pBluescript-RAB8A, pBluescript-RAB8B or pBluescript-RAB8D were independently cloned into same sites of the pE1798 plasmid (Supplementary Table S4).

The plasmids pET23a-RAB8A and pET23a-RAB8D were used as templates to perform PCR with primers listed in Supplementary Table S5 and the high-fidelity Phusion DNA polymerase; the PCR products were then digested with KpnI and SacI and, subsequently, cloned into the pE1798 plasmid to create plasmids pE1798-T7-tag-RAB8A and pE1798-T7-tag-RAB8D (Supplementary Table S4). In order to overexpress the T7-tagged-RAB8B in Arabidopsis transgenic plants, the XabI-KpnI fragment of pET23a-RAB8B (Supplementary Table S4; Hwang and Gelvin 2004) was cloned into the same restriction enzyme sites of the pE1798, yielding the plasmid pE1798-T7-tag-RAB8B (Supplementary Table S4). These pE1978 series plasmids were separately transformed into the non-tumorigenic A. tumefaciens strain GV3101(pMP90) (Koncz and Schell 1986) and the A. tumefaciens strains were used to transform wild-type plants (ecotype: Ws) to generate O/E transgenic Arabidopsis plants via a floral dip method (Clough and Bent 1998). The A. tumefaciens strain containing the pE1798-RAB8B plasmid was used to transform the rab8b mutant to obtain the rab8b-1 complementation plants.

Protein extraction from Arabidopsis plants and protein gel blot analysis

Seedlings of RAB8A, 8B and 8D O/E transgenic Arabidopsis plants and wild-type plants were used to isolate proteins. Plant tissues were ground with liquid nitrogen and mixed with CelLytic P (Sigma Chemical Co., St. Louis, MO, USA) with a protease inhibitor cocktail (1:100 dilution) from Sigma according to the manufacturer’s instructions. Concentrations of final protein extracts were measured with a BCA protein assay kit (Pierce, Rockford, IL, USA) and a spectroscopy (PARADIGM Detection Platform, Beckman Coulter Inc., Indianapolis, IN, USA). Equal amounts of plant proteins were analyzed in 12.5% SDS-polyacrylamide gels. Protein gel blot analyses were subsequently conducted with a 1:1000 dilution of T7-tag antibody (Abcam, Cambridge, UK), then with a 1: 20000 dilution of HRP-conjugated goat anti-rabbit IgG antibody (PerkinElmer Life and Analytical Sciences, Boston, MA, USA). The membranes were developed by a chemiluminescent detection (PerkinElmer Life and Analytical Sciences, Boston, MA, USA) and signals were captured with X-ray-films.

Agrobacterium tumefaciens-mediated stable, transient root and seedling transformation assays of rab8 mutant plants and RAB8 O/E Arabidopsis transgenic plants

Seeds from wild-type, rab8 mutants, rab8b-1 complementation plants, and RAB8 O/E transgenic plants were surface-sterilized and grown on Gamborg’s B5 medium (PhytoTechnology laboratories, Carlsbad, CA, USA) solidified with 0.75% Bactoagar (BD Biosciences, Lenexa, KS, USA) with the appropriate antibiotics (kanamycin 50 μg · mL⁻¹ for rab8 mutant plants and hygromycin 20 μg · mL⁻¹ for O/E transgenic plants). Seedlings were grown on the solidified B5 medium in baby food jars without antibiotics for 3–4 weeks to perform stable and transient root transformation assays according to previously published methods (Hwang and Gelvin 2004, Huang et al. 2018).

All A. tumefaciens strains (Supplementary Table S4) were cultured in 523 medium (Hwang et al. 2010) with the appropriate antibiotics (rifampicin 50 μg · mL⁻¹, kanamycin 50 μg · mL⁻¹) at 28^oC. Overnight-grown bacterial cultures were subcultured into 25 mL of 523 medium with antibiotics and grown to a density of 10⁹ cfu · mL⁻¹. Bacterial cells were spun down and washed with 0.9% sodium chloride to remove antibiotics and medium; bacterial cultures were finally resuspended in 0.9% sodium chloride at 10⁵, 10⁶ or 10⁸ cfu · mL⁻¹ for root transformation assays.

For stable root transformation assays, root segments of 3–4 week-old plants were cut and transferred onto solidified MS medium and infected with the tumorigenic strain A. tumefaciens A208 (Supplementary Table S4) for 2 d at 22 to 24^oC. After co-incubation periods, root segments were separated and transferred to MS medium (lacking hormones) with the antibiotic timentin (100 μg · mL⁻¹) for 1 month to score tumor formation efficiencies. For transient root transformation assays, root segments were co-incubated with A. tumefaciens At849 containing the pBISN1 binary vector (Supplementary Table S4) for 2 d. After infection, root segments were transferred on the callus induction medium with timentin for 4 additional days at 22 to 24^oC. Roots were then stained with X-gluc (5-bromo-4-chloro-3-indolyl beta-D-glucuronic acid) staining solutions for 1 d at 37^oC and were observed under a stereoscopic microscope to obtain transient transformation efficiencies. For root transformation assays, 15–20 different Arabidopsis plants were infected with each A. tumefaciens strain and 60–80 root segments were examined for each plant for each independent transformation assay. The transformation efficiencies were presented as average values from at least three independent experiments. Error bars were calculated using Excel STDEVP function. The significance test between treatments was based on the pairwise Student t-test, P < 0.05.

The transient seedling transformation assays (Agrobacterium-mediated enhanced seedling transformation, AGROBAST) were performed using previously described methods with minor modifications (Wu et al. 2014). The Arabidopsis seedlings were first germinated in a 6-well plate with the 1/2 MS medium (pH5.7) and 0.5% sucrose at 22 to 24^oC for 7 d. The A. tumefaciens C58C1(pTiB6S3ΔT) strain with a pBISN1 binary vector (Supplementary Table S4) was first grown in 523 medium with the appropriate antibiotics (rifampicin 50 μg · mL⁻¹, kanamycin 50 μg · mL⁻¹) at 28^oC overnight; bacterial cultures were further grown at 28^oC in acidic AB-MES medium with 200 μM AS (AS) for 24 hours to induce vir gene expressions (Hwang et al. 2010, Huang et al. 2018). After AS inductions, bacterial cells were washed and resuspended in infection solutions (half-strength of the MS medium [pH 5.7], one-quarter of the AB-MES medium (pH 5.5, 0.5% sucrose, and 50 μM AS) at 10⁷ cfu · mL⁻¹ for seedling transformation assays.

The Arabidopsis seedlings were infected with AS-induced bacteria cells at 22 to 24^oC for 3 d. After infection, seedlings were ground with liquid nitrogen and mixed with extraction buffers to perform the fluorescent MUG (4-methylumbelliferyl-ß-D-glucuronide) assays as described previously (Wu et al. 2014, Huang et al. 2018). The fluorescence was determined with a 96 microplate reader (PARADIGM Detection Platform) at 365 nm excitation and 455 nm emission. The BCA protein assay kit and a spectroscopy were used to determine the protein concentration for each protein sample. The relative GUS activity was the fluorescence signal normalized by an equal amount of proteins. For seedling transformation assays, 15–20 different Arabidopsis seedlings were infected with the A. tumefaciens strain for each independent transformation assay and more than three independent transformation assays were performed. The transformation efficiencies were average values from at least three independent experiments. Error bars were calculated using the Excel STDEVP function. The significance test between treatments was based on pairwise Student t-test, P < 0.05.

P. syringae infection assays of Arabidopsis rab8 mutants and RAB8 O/E transgenic plants

Leaves of the 4- to 5 week-old pot-grown Arabidopsis plants were infected with P. syringae strains (Supplementary Table S4) by syringe infiltrations as described previously (Huang et al. 2018). The P. syringae strains were grown at 28^oC to the mid to late log phase and were used to infiltrate plants at 10⁴ cfu · mL⁻¹. To determine bacterial populations in plant leaves, leaf disks were excised from infiltrated leaves by a 0.6 cm² cork borer at 0, 1, 3, 5 and 7 d after infiltration. The leaf disks were ground with a plastic pestle in 5 mM magnesium chloride solutions. The bacterial suspensions were serially diluted with magnesium chloride and cultured on King’s medium B (KB) agar plates with rifampicin (20 μg · mL⁻¹) and cycloheximide (10 μg · mL⁻¹) to determine viable cell numbers. At least 15 different Arabidopsis plants were infected with bacteria for each independent infection assay and more than three independent infection assays were performed.

Yeast complementation assays

The yeast expression vector pYX (Supplementary Table S4) was used to express RAB8A-8E proteins that were under the control of the yeast inducible GAL1 promoter. The DNA fragments of the RAB8A-8E coding sequences from Arabidopsis were obtained with PCR using Arabidopsis cDNA as templates, the high-fidelity Phusion DNA polymerase, and appropriate primers (Supplementary Table S5). The PCR products containing the RAB8A, RAB8C or RAB8D were digested with EcoRI/PstI and were cloned into same restriction enzyme sites of the pBluescript to create the plasmids pBluescript-RAB8A, pBluescript-RAB8C and pBluescript-RAB8D (Supplementary Table S4). The EcoRI-BamHI fragments from plasmids pBluescript-RAB8A, pBluescript-RAB8C and pBluescript-RAB8D were digested and subcloned into the pYX (Supplementary Table S4) to express the RAB8 protein in yeast. Similarly, the PCR products containing the RAB8B or RAB8E were digested with EcoRI/XhoI or EcoRI/BamHI, respectively, and were cloned into same restriction enzyme sites of the pYX to create the plasmids pYX-RAB8B and pYX-RAB8E (Supplementary Table S4).

The yeast strain Sec4 mutant/NY405 (Novick et al. 1980) was transformed with the pYX (used as a negative control), pNB136 (used as a positive control) or pYX-RAB8A-8E plasmids (Supplementary Table S4) with a lithium acetate method. In order to induce the RAB8A-8E protein expressions in yeast, 2% galactose and 1% raffinose were added in SD media. The yeast transformants were then examined for viability when grown at 25°C, 30°C and 37°C for 2–4 d.

Supplementary Data

Supplementary data are available at PCP online.

Data Availability

The data underlying this article are available in the article and in its online supplementary material.

Funding

This research was funded by the Ministry of Science and Technology, Taiwan [MOST 105-2313-B-005-008; MOST 107-2321-B-005-009]. This research was supported in part by the Ministry of Education, Taiwan, under the Higher Education Sprout Project.

Acknowledgements

The authors thank Dr Erh-Min Lai for providing A. tumefaciens strains; Dr Peter Novick for providing the yeast sec4 mutant; Dr Stanton Gelvin and Dr Andreas Nebenfuhr for providing plasmids; Fu-Hui Wu, Shu-Chen Shen and lab members of the Plant Tech Core Facility at the Agricultural Biotechnology Research Center (ABRC) of Academia Sinica for their technical help with BiFC experiments and the Hwang lab members for discussion and technical assistance. This research was funded by the Ministry of Science and Technology, Taiwan [MOST 105–2313-B-005-008; MOST 107–2321-B-005-009]. This research was supported in part by the Ministry of Education, Taiwan, under the Higher Education Sprout Project.

Author Contributions

F.C.H. and H.H.H. conceived, contributed to experiment design and wrote the manuscript. F.C.H., S.F.C., Y.T.L., P.R.C., Y.T.L., H.N.C., and C.S.L. conducted experiments and analyzed data. All authors read and approved the manuscript.

Disclosures

The authors have no conflicts of interest to declare.

References