The maintenance of suspension culture tobacco (Nicotiana tabacum) BY-2 cells was described previously (Jiang and Rogers, 1998; Tse et al., 2004). Arabidopsis thaliana (Col-0) seeds were surface sterilized by vortexing mixture with 70% (v/v) ethanol containing 0.05% Tween 20 for 10 min. Spin down the seeds using a bench top centrifuge at max speed for 2 s, remove the supernatant and wash the seeds with 100% ethanol once for 30 s. The seeds were pipetted out onto a sterile filter paper in a sterile hood, air dried and spread on ½ MS agar plates. The plates were first incubated at 4°C for 2 days and then transferred in the plant growth chamber. The settings of growth chamber for Arabidopsis germination and growth: Light intensity: 120–150 μm m^-2 s^-1; Temperature: 22°C; Light cycle: 16 h light/8 h dark.

Briefly, the segmented promoter, target gene, fluorescent tag and terminator were amplified using primers (Supplementary Table S1) containing 5′-end overlapping sequence with the adjacent fragment by standard PCR. The DNA templates are derived from previous studies (Wang et al., 2011b, 2016). Then, the first-round PCR products above were mixed and subjected to the optimized isothermal assembly reaction at 50°C for 60 min, followed by amplifying the entire expression cassette using the reaction product as template with the outermost primers (e.g., 1-FP35S and 1-RNOS for expression cassette 1). Finally, the splicing multiple expression cassettes with ends overlapping homologous could be directly assembled into Sma I-linearized functional vectors (e.g., pUC18 or pCAMBIA1300) by the optimized isothermal recombination reaction at 50°C for 60 min. The isothermal reaction was adapted and optimized from previous studies (Gibson et al., 2009; Zhu et al., 2014). In general, 400 μl 2× master mixture was prepared by adding 160 μl homemade 5× ISO stock buffer (500 mM Tris-HCl, pH 7.5; 50 mM MgCl₂; 1 mM dNTP; 50 mM DTT; 25% PEG 8000 and 5 mM NAD) with 3 units of T5 exonuclease (Epicenter), 25 units of Phusion DNA polymerase (New England Biolabs, NEB), 2000 units of Taq DNA polymerase (NEB) and sterilized double distilled H₂O. Aliquot 15 μl per tube and store at -20°C. Add 5 μl mixture of equimolar DNA fragments to 15 μl 2× master mixture for recombination and incubate at 50°C for 60 min.

The general procedure of Agrobacterium transformation was performed as previously described (Clough and Bent, 1998; Tse et al., 2004; Zhang et al., 2006; Manavella and Chan, 2009). In brief, electroporation is used to mediate the transformation of the binary vector into Agrobacterium (EHA105). Agrobacterium competent cells were thaw on ice for 30 min, then mix the competent cells with 2 μl high-quality plasmid DNA (100–200 ng) which is extracted by plasmid extraction kit (Qiagen) and dissolved in double distilled H₂O rather than TE buffer. Sit the mixture on ice for 5 min. Transfer the mixture to a pre-chilled 0.1 cm electroporation cuvette (Bio-Rad). Insert cuvette into the Gene Pulser (Bio-Rad) and perform the electroporation with following settings: 25 μF, 1.6 kV, 600 ohms. After the electroporation, add 1 ml super optimal broth with catabolite repression (SOC) immediately into the cuvette and transfer the cells into a 2 ml Eppendorf tube and shake at 28°C, 200 rpm for 120 min. Centrifuge at 6000 rpm at room temperature (RT) for 5 min, resuspend and spread the cells on a LB plate with suitable antibiotics and incubate at 28°C for 2–3 days. To generate the transgenic plants, the resulting constructs were introduced into Agrobacterium tumefaciens and transformed into wild-type Columbia-0 by floral dip (Clough and Bent, 1998). The chimeric fluorescent fusion protein co-expression transgenic line was generated using the pCAMBIA1300 plasmid. T0 generations were screened with hygromycin. Seeds were surface sterilized and sown on plates with 1/2 Murashige and Skoog (MS) salts plus 0.8% agar. The seeded plates were kept at 4°C for 3 days before being moved to the growth chamber. The plates were incubated at 22°C under a long-day (16 h light/8 h dark) photoperiod. Seven-day old plants were checked for fluorescent signals under the fluorescent microscope prior to transferred into soil.

BY-2 cells cultured for 3 days were used for particle bombardment. The sequential procedures for gene delivery into BY2 cells via particle bombardment were the same with the steps for pollen tube bombardment as described before (Wang et al., 2010, 2016; Wang and Jiang, 2011). The bombarded cells were kept in dark in plant growth chamber (27°C) for 6–12 h prior to observation for fluorescent signals. The expression efficiency of chimeric fusion proteins in tobacco BY-2 cells is approximate 5–7% that is consistent with previous studies (Wang and Jiang, 2011; Wang et al., 2011b, 2013).

Stock solutions of wortmannin (1 mM in DMSO; Sigma–Aldrich) and brefeldin A (BFA) (1 mM in DMSO; Sigma–Aldrich) were prepared and stored at -20°C. These drugs were diluted in growth medium to appropriate working concentrations before incubation with plant roots.

In general, the confocal fluorescence images were collected using a Zeiss 710 system with the following parameters: 63× water objective, 700 gain, 0 background, 0.168 mm pixel size, and photomultiplier tubes detector. The images from cells were collected with a laser level of 3% to ensure that the fluorescent signal was within the linear range of detection (typically 0.5% or 1% laser was used). Colocalizations between two fluorophores were calculated by using ImageJ program with the Pearson-Spearman correlation (PSC) colocalization plug-in (French et al., 2008). Results were presented either as Pearson correlation coefficients or as Spearman’s rank correlation coefficients, both of which produce r values in the range (-1 to 1), where 0 indicates no discernable correlation while +1 and -1 indicate strong positive and negative correlations, respectively.

We have developed a novel and high-efficient system which differs from the classical approaches (Figures 1A,B) for co-expressing multiple chimeric fluorescent fusion proteins in a time-saving fashion (Figures 1C,D). It is fully compatible with current approaches of transient expression and stable genetic transformation in plants. In this procedure, we employ a single expression vector which contains multiple protein expression cassettes as shown in Figures 1C,D. Each cassette has its own promoter, fluorescent tag, target protein, and terminator, respectively, and is responsible for expressing its individual chimeric fluorescent fusion protein (Figure 2). Therefore, it functions semi-independently as a basic “Lego” element and can be manipulated independently according to different expression demands including the requirement for a specific promoter and preference of N- or C-terminal fusion of fluorescent tags with target proteins. Furthermore, all the expression cassettes are linked and integrated into a final single expression vector which is designed for either transient expression or stable plant transformation depending on different experimental requirements.

Additionally, the strategy for assembly of DNA molecules to generate the cassettes for expressing chimeric fluorescent fusion proteins, and the combination of multiple protein expression cassettes with a single destination vector are all conveniently achieved by an optimized isothermal in vitro recombination method which is adapted from previous studies (Gibson et al., 2009; Zhu et al., 2014). This takes advantage of homologous recombination of overlapping short DNA sequences to create DNA fusion elements or plasmids with various types of sequence manipulation in one-single reaction. Therefore, ligation of multiple DNA molecules as well as recombination with the destination expression vector are all effectively achieved in one reaction at one time without the need of multiple steps of restriction enzyme digestion and DNA ligation.

Therefore, we have developed at once a robust system for the convenient co-expression of multiple chimeric fluorescent fusion proteins in plants. In addition, it is a highly manipulable “Lego” system that can be easily modified and re-constructed via a one-step DNA junction reaction.

To test this system, we chose two marker proteins: the vacuolar sorting receptor (VSR) and the secretory carrier membrane protein (SCAMP) which participate in the secretory and endocytic pathways, respectively (Tse et al., 2004; Lam et al., 2007a; Wang et al., 2010, 2011a). VSRs are type-I integral membrane family proteins with a single transmembrane domain (TMD) and a short cytosolic tail (CT). VSR mediates the transport of cargo proteins which contain vacuolar sorting determinants (VSDs) to the vacuole via the secretory pathway. The VSR N-terminus is responsible for cargo protein recognition and binding whereas the TMD and CT are sufficient for its correct targeting and localization (Tse et al., 2004; Shen et al., 2014). Subcellular localization studies by confocal immunofluorescent and immunogold EM have shown that VSR proteins are mainly localized to prevacuolar compartments (PVCs) which have been identified as multivesicular bodies (MVBs) in tobacco BY-2 cells (Tse et al., 2004). In contrast, SCAMPs are type-IV membrane proteins localizing to the plasma membrane (PM) and early endosome (EE) or trans-Golgi network (TGN) in the endocytic pathway. SCAMP has been used as a reporter protein to follow the endocytic process in plant cells (Lam et al., 2007a). Although the protein secretory pathway and endocytosis are two distinct pathways for protein trafficking and targeting, it has been revealed that TGN serves as a junction for the secretory and endocytic pathways in plant cells (Lam et al., 2007b). Nevertheless, the underlying molecular mechanism regulating the coordination of the two pathways in terms of cell development and morphogenesis remains unsolved. Therefore, co-expression of VSR and SCAMP proteins should provide essential hints toward a better understanding of the spatiotemporal interactions in the secretory and endocytic pathways in plant cells.

By applying this newly developed co-expression approach of multiple chimeric fluorescent fusion proteins, RFP-AtVSR2 and AtSCAMP4-GFP were first inserted into two independent expression cassettes as shown in Figure 2. All of the DNA fragments in these two expression cassettes were joined via the improved isothermal recombination reaction by short overlapping DNA sequences (Figure 2 and Supplementary Table S1). Thereafter, the expression cassettes which express RFP-AtVSR2 and AtSCAMP4-GFP, respectively, were linked and integrated into the destination vector pUC18 which is used for transient expression or pCAMBIA for plant stable transformation by applying the same DNA assembly strategy (Figure 2). Transient co-expression of RFP-AtVSR2 and AtSCAMP4-GFP in N. tabacum BY2 suspension cells was tested by biolistic bombardment. RFP-AtVSR2 gave rise to a punctate pattern as shown in Figure 1C. In contrast, AtSCAMP4-GFP localized on the PM with only some cytosolic punctate dots. As shown in Figure 3A, the subcellular localization of RFP-AtVSR2 is therefore clearly distinct from that of AtSCAMP4-GFP. In addition, RFP-AtVSR2 and AtSCAMP4-GFP were stably co-expressed in Arabidopsis plants via Agrobacterium-mediated genetic transformation. In transgenic root cells and root hairs as shown in Figures 3B,E, RFP-AtVSR2 and AtSCAMP4-GFP were successfully co-expressed and their perspective subcellular localizations were consistent with the results obtained from BY2 cells. Moreover, pharmaceutical treatments of the transgenic Arabidopsis roots and root hairs with wortmannin and BFA confirmed that the RFP-AtVSR2-labeled PVC responded to wortmannin by forming ring-like structures (through homotypical fusion of MVBs and AtSCAMP4-GFP-labeled TGN) and responded to BFA (by forming Golgi aggregates) (Figures 3C,D,F,G). Therefore, co-expression of fluorescent tagged VSR and SCAMP using this novel protein expression system via either transient expression or stable plant transformation reliably delivers correct subcellular localizations in plant cells (Tse et al., 2004; Lam et al., 2007a; Wang et al., 2010, 2011b).