SEARCH

SEARCH BY CITATION

Keywords:

  • Flim;
  • FRET;
  • GFP;
  • Golgi;
  • small GTPase;
  • tethering factor

Abstract

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

Peripheral tethering factors bind to small GTPases in order to obtain their correct location within the Golgi apparatus. Using fluorescence resonance energy transfer (FRET) and fluorescence lifetime imaging microscopy (FLIM) we visualized interactions between Arabidopsis homologues of tethering factors and small GTPases at the Golgi stacks in planta. Co-expression of the coiled-coil proteins AtGRIP and golgin candidate 5 (GC5) [TATA element modulatory factor (TMF)] and the putative post-Golgi tethering factor AtVPS52 fused to green fluorescent protein (GFP) with mRFP (monomeric red fluorescent protein) fusions to the small GTPases AtRab-H1b, AtRab-H1c and AtARL1 resulted in reduced GFP lifetimes compared to the control proteins. Interestingly, we observed differences in GFP quenching between the different protein combinations as well as selective quenching of GFP-AtVPS52-labelled structures. The data presented here indicate that the FRET-FLIM technique should prove invaluable in assessing protein interactions in living plant cells at the organelle level.

The Golgi apparatus plays a central role in the plant endomembrane system, as it is not only involved in the processing of protein cargo received from the endoplasmic reticulum (ER) but also in the synthesis of complex cell wall polysaccharides (1). Plant cells can contain up to hundreds of discrete Golgi stacks that are mobile in an actin-dependent manner with the surface of the ER (2–5). It is not known how Golgi stacks manage to keep their structural integrity while moving through the cytoplasm, but the existence of a plant Golgi matrix has been proposed after electron microscopy revealed the existence of intercisternal elements and a ribosome-excluding zone around the stacks (6,7). Golgi matrix proteins could also be involved in tethering Golgi stacks to the ER surface (8).

The golgins, a Golgi-localized protein family with long coiled-coil domains, are ideal candidate proteins for being part of a putative Golgi matrix, as in animal cells golgins have been implied in formation and maintenance of Golgi stack structure (9–11). One example is the Golgi reassembly stacking proteins (GRASPs) that are involved in stacking of Golgi cisternae (12,13). Many golgins, such as the cis-Golgi localized p115 and GM130, have multiple tethering functions and can act in multiprotein tethering complexes with other golgins, small GTPases and SNARE proteins (reviewed in 14). Examples are the p115-giantin-GM130-Rab1 tether and the golgin-84-CASP (CCAAT-displacement protein alternatively spliced product) tether, which are found on coat protein I (COPI) membranes and could define different subpopulations of COPI vesicles (15). Homologues of golgins have been identified in Arabidopsis and localized to the Golgi apparatus, but their function remains largely unknown (16–22), although the Arabidopsis homologue of p115 has been suggested to play a role in tethering the Golgi stack to the ER (23).

Whereas some golgins are targeted to the Golgi stack via their C-terminal transmembrane domains, peripheral golgins obtain Golgi localization by interaction with other golgins or small regulatory GTPases. One example is the human peripheral coiled-coil protein TATA element modulatory factor (TMF) that binds to Rab6 (24) and was displaced from the Golgi stack upon RNAi-mediated depletion of the Rab protein (25). Other peripheral golgins contain a conserved C-terminal sequence consisting of ∼42 amino acids. This domain was originally identified in four mammalian golgins and named GRIP (golgin-97, RanBP2a, Imh1p, and p230/golgin-245) domain according to the starting letters of those proteins (26,27). The GRIP domain is highly conserved in eukaryotes and is required for targeting of proteins to the trans-Golgi or trans Golgi network (TGN) through interaction with the ARL1 (ARF-like) GTPases (28,29).

Although the organization of the plant Golgi apparatus differs significantly from that of yeast or animal cells (30), some basic structural mechanisms appear to be conserved, as several interactions between golgins and regulatory proteins have been identified in different species. For example, the interaction between human TMF and three isoforms of Rab6 (24) was also observed in yeast between the TMF homologue Sgm1p and the Rab6 homologue YPT6 (31) and in Arabidopsis between the homologues of Rab6, AtRab-H1b /c(32) and the homologue of TMF, GC5 (golgin candidate 5, 22). GRIP domain proteins appear to interact with the small GTPase ARL1 in yeast and animals (29,33,34) as well as in plants (19,21). AtGRIP shows high homology to the mammalian golgin-97, which contains a conserved C-terminal GRIP domain that interacts with Rab6 and is necessary and sufficient to target the protein to the trans-Golgi in mammalian cells (35).

Coiled-coil proteins have not only been implied in transport pathways between the ER and the Golgi but also in trafficking between the TGN and post-Golgi compartments. In yeast, the multisubunit tethering complex GARP (Golgi-associated retrograde protein) is involved in retrograde transport from endosomes to the TGN (36,37). This complex consists of the four subunits Vps51/52/53/54p, each containing coiled-coil domains (36), and has been shown to interact with Ypt6, the yeast homologue of mammalian Rab6 (38). Similarly, it has been shown that Vps52 interacts with Rab6 in mammalian cells (39). In Arabidopsis, a gene termed pok (poky pollen tube) identified in a mutant screen shares significant homology with the yeast Vps52p gene (40). POK/AtVPS52 appears to be part of a large protein complex and located to the Golgi, pre-vacuolar compartment and an unidentified post-Golgi compartment (41), but nothing is known to date about possible interactions between POK and the Arabidopsis homologues RabH1b and RabH1c.

Characterizing the interactions between golgins and regulatory or other structural proteins provides a first step towards a better understanding of their putative role within the Golgi stack. In vitro techniques such as yeast two-hybrid screens and pull down assays (reviewed in 42,43) can provide a general indication whether two proteins have the potential to interact with each other. Such techniques can, however, be compromised by the occurrence of false positive results and the artificial environment in which the binding takes place. Proteins interacting in vitro may never do so in vivo because they might be spatially separated in different cellular compartments (44). Therefore, it is preferable that in vitro results are confirmed using in vivo approaches.

A range of techniques are available to study protein–protein interactions in vivo using fluorescence microscopy. In the bimolecular fluorescence complementation (BiFC) assay, a fluorophore is split into two non-functional halves fused to two proteins of interest and interaction between the proteins restores the fluorescent signal (45,46). Analysis of FRET (Förster or fluorescence resonance energy transfer) between a donor and an acceptor fluorophore fused to two proteins of interest is more commonly used. FRET occurs under the conditions of a sufficiently large spectral overlap between donor emission and acceptor absorption spectrum, a favourable orientation of fluorophores to each other in a proximity of 1–10 nm (47). Interacting proteins are close enough to allow non-radiative energy transfer from the donor to the acceptor fluorophore, leading to quenching of donor fluorescence and an increase in acceptor fluorescence. Besides spectral bleed-through, other problems of this technique include the dependence on fluorophore expression levels and unintended photobleaching during measurements (48).

A method that combines FRET with fluorescence lifetime imaging (FLIM) has certain advantages over conventional FRET measurements (49–52). A fluorophore has a characteristic lifetime that remains unaffected by fluorophore concentration levels or excitation intensity, and spectral bleed-through does not constitute a problem as only the donor lifetime is measured (47). The fluorophore lifetime can be influenced by changes in temperature, environment, calcium ion concentration and the occurrence of FRET (51). Quenching of the donor fluorophore lifetime indicates interaction with the acceptor fluorophore and can be measured directly. FRET-FLIM has been successfully used in planta to study protein–protein interactions in protoplasts of cowpea (53–55), petunia (56), tobacco (57), maize (58) and Arabidopsis (59,60) as well as in barley (61) and tobacco leaf epidermal cells (62,63). Whereas FRET-FLIM has been used successfully on animal cells to visualize interactions between Rab GTPases and Rab binding proteins (64,65), in plants no such in vivo analyses of Rab or golgin protein interactions have been performed so far to our knowledge.

In this study we aimed to confirm in planta interactions between several plant Golgi tethering factors and small GTPases that to date only had been performed in vitro, by analysing the fluorescence lifetime (FLIM) after FRET by two-photon excitation (66). In this system, fluorophores are excited by the absorption of multiphotons in the near-infrared wavelength, which allows deeper penetration of tissues and leads to less cell damage. Absorption is confined to a narrow focussed plane and lacks out-of-focus excitation, therefore resulting in higher resolution and further reduction of photo damage (67).

Results

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

AtRab-H1b but not AtRab-H1c interacts with AtGRIP in vitro

To investigate if the small GTPases AtRab-H1b and AtRab-H1c would interact with the plant homologue of golgin-97 in vitro, we analysed the interaction between AtGRIP and both AtRab-H1b and AtRab-H1cin vitro using the yeast two-hybrid system (68). Wild-type AtRab-H1b and AtRab-H1c were fused at their N-termini with the activation domain of the yeast transcription factor GAL4. AtGRIP was fused to the C-terminus of the GAL4-binding domain. Prey and bait constructs were sequentially transformed into yeast and the transformants were plated on selection plates. Only the GAL4BD-GRIP and GAL4AD-AtRab-H1b produced colonies and tested positive for β-galactosidase activity, indicating that these two fusion proteins interact in yeast (Figure 1A). No interaction was detected between the AtGRIP and AtRab-H1c in this assay (Figure 1B).

image

Figure 1. AtRab-H1b, but not AtRab-H1c, interacts with AtGRIP in vitro. A–E) Yeast two-hybrid analysis. Yeast transformants were plated on selection plates. The presence of two interacting partners allows yeast to grow in the absence of histidine and adenine. GAL4BD, Gal4-binding domain and GAL4AD, Gal4-activation domain. A) GAL4BD-AtGRIP + GAL4AD-AtRab-H1b, (B) GAL4BD-AtGRIP + GAL4AD-AtRab-H1c, (C) GAL4BD-AtGRIP + GDP-locked mutant GAL4AD-AtRab-H1b[T23N], (D) GAL4BD-AtGRIP + GTP-locked mutant GAL4AD-AtRab-H1b[Q68L] and (E) GAL4BD-AtGRIP + dominant-negative mutant GAL4AD-AtRab-H1b[N122I]. F–H) Confocal images of tobacco leaf epidermal cells transiently transformed with fluorescent fusion constructs of AtGRIP and AtRab-H1b wild-type and mutants. All constructs were transformed at an OD600 of 0.03 and imaged 2–3 days after inoculation. F) Co-expression of YFP-AtRab-H1b[wt] (magenta channel) and AtGRIP-GFP (green channel) shows that the construct co-located (merged image). G) Co-expression of the GTP-locked mutant YFP-AtRab-H1b[Q68L] (magenta channel) and AtGRIP-GFP (green channel) shows that the constructs co-locate (merged image). H) Co-expression of the dominant-negative mutant YFP-AtRab-H1b[N122I] (magenta channel) with AtGRIP-GFP (green channel) results in aggregation of the two constructs (merged image). All scale bars = 10 μm.

Download figure to PowerPoint

To examine whether the interaction between AtRab-H1b and AtGRIP was dependent on the state of GTP binding, the GDP-locked mutant AtRab-H1b[T23N], the GTP-locked mutant AtRab-H1b[Q68L] and the dominant-negative mutant AtRab-H1b[N122I] with reduced GDP/GTP affinity (32) were each fused to the C-terminus of the GAL4-activation domain. Again, the prey and bait constructs were sequentially transformed into yeast and plated on the selection medium (Figure 1C–E). Only GAL4AD-AtRab-H1b[Q68L] and GALBD-GRIP showed a positive interaction (Figure 1D). Neither AtRab-H1b[T23N] nor AtRab-H1b[N122I] interacted with AtGRIP (Figure 1C,E).

When transiently expressed in tobacco leaf epidermal cells, YFP-AtRab-H1b (magenta channel) co-located with AtGRIP-green fluorescent protein (GFP) (Figure 1F, green channel). Because interaction between AtRab-H1b and AtGRIP was dependent on the nucleotide-binding status of AtRab-H1b, AtGRIP was co-expressed with yellow fluorescent protein (YFP)-AtRab-H1b[T23N], YFP-AtRab-H1b[Q68L] and YFP-AtRab-H1b[N122I] in tobacco leaves (Figure 1H–I). The YFP-AtRab-H1b[Q68L] signal showed overlap with AtGRIP-GFP (Figure 1G) similar to that observed for YFP-AtRab-H1b[wt]. However, when AtGRIP was co-expressed with YFP-AtRab-H1b[N122I] (Figure 1H) and YFP-AtRab-H1b[T23N] aggregates containing both fusion proteins were formed.

Peripheral plant coiled-coil proteins co-locate with small GTPases in tobacco leaf epidermal cells

Figure 2 shows confocal images of the fluorescent fusion proteins used in this study, transiently expressed in tobacco leaf epidermal cells after transfection of leaves with agrobacterium suspension cultures. The trans-Golgi coiled-coil proteins AtGRIP-GFP (Figure 2A–D, green channel) and GFP-cGC5 (henceforth referred to as GFP-cTMF, Figure 2E,F, green channel) were observed in the cytoplasm and on punctate structures that previously have been shown to co-locate with a Golgi marker (19,22). AtGRIP-GFP and mRFP-AtARL1 signals co-located in the cytoplasm and at larger punctate structures resembling Golgi bodies as reported by Latijnhouwers et al.(19, Figure 2A). AtGRIP-GFP additionally labelled smaller punctate structures (Figure 2A, arrowheads). mRFP-AtRab-H1b was observed in the cytoplasm and punctate structures, co-locating partially with AtGRIP-GFP (Figure 2B) and fully with GFP-cTMF (Figure 2E, magenta channel). The same localization was observed for mRFP-AtRab-Hc co-expressed with AtGRIP-GFP (Figure 2C, magenta channel) or GFP-cTMF (Figure 2F, magenta channel). AtGRIP-GFP located to the same structures as AtCASP-mRFP but with only partial overlap confirming their predicted location to cis- and trans-faces of Golgi stacks (Figure 2D). GFP-AtVPS52 labelled punctate structures (Figure 2G,H, green channel), some of which partially co-located with mRFP-AtRab-H1b (Figure 2G, magenta channel, arrowheads in merged channel) or mRFP-AtRab-H1c (Figure 2H, magenta channel, arrowheads in merged channel).

image

Figure 2. The subcellular localization of coiled-coil and small regulatory proteins. Confocal images showing GFP (green) and mRFP (magenta) fusion proteins co-expressed in tobacco leaf epidermal cells 3 days after transfection with agrobacterium suspension cultures. AtGRIP-GFP, GFP-cTMF, mRFP-CASP and GFP-VPS52 were infiltrated at OD600 = 0.1, mRFP-Rab-H1b/c and mRFP-AtARL1 at OD600 = 0.05. A) AtGRIP-GFP + mRFP-AtARL1. B) AtGRIP-GFP + mRFP-Rab-H1b. C) AtGRIP-GFP + mRFP-Rab-H1c. D) AtGRIP-GFP + mRFP-AtCASP. E) GFP-cTMF + mRFP-AtRab-H1b. F) GFP-cTMF + mRFP-AtRab-H1c. Arrowheads indicate co-loalization between GFP-VPS52 and mRFP-AtRab-H1b. G) GFP-VPS52 + mRFP-AtRab-H1b. H) GFP-VPS52 + mRFP-AtRab-H1c. Arrowheads indicate co-localization between GFP-VPS52 and mRFP-AtRab-H1c. Scale bars, 10 μm.

Download figure to PowerPoint

Spatial proximity of GFP and mRFP does not affect the lifetime of GFP

To establish the lifetime of GFP in tobacco leaf epidermal cells, we first measured the lifetime of the Golgi marker STtmd-GFP (2) expressed in stable tobacco G41 plants. To stop movement of Golgi bodies during data collection, all leaf samples in this study were incubated in 25 μm latrunculin B to depolymerize the actin cytoskeleton (4). Data was collected from 84 Golgi bodies in three samples and the lifetime values ranged between 2.4 and 3.1 ns with an average of 2.72 ns (Figure 3A, Table 1). To exclude the possibility of FRET occurrence between GFP and mRFP caused by close spatial proximity in the same organelle, we transfected wild-type tobacco leaf epidermal cells with plasmids encoding ST-GFP and ST-mRFP (18), which both target to the medial and trans-Golgi. One sample with 20 Golgi bodies provided sufficient data for analysis. The lifetime distribution ranged between 2.5 and 3 ns (Figure 3A) with an average of 2.74 ns (Figure 4) and was not found to be statistically different from the control lifetime distribution (Student's t-test p > 0.05, Table 1). This result shows that the pure spatial proximity of the two fluorophores does not lead to a reduced lifetime compared to GFP expressed alone.

image

Figure 3. Lifetime distributions of GFP constructs with and without their putative binding partners. To determine the lifetime of GFP expressed with or without putative binding partners in tobacco leaf epidermal cells, fluorescence lifetime measurements were performed 2–4 days after agrobacterium-mediated infiltration. Lifetimes for individual Golgi bodies were obtained on a pixel base. This figure shows the percentage of the total analysed number of Golgi bodies with their corresponding lifetime values on a range from 1.8 to 3.1 ns and compares the lifetime distribution of control proteins against protein combinations. A) The lifetime values of the Golgi marker ST-GFP ranged between 2.4 and 3.1 ns (n = 84). Although ST-GFP and ST-mRFP both target to the medial/trans-Golgi apparatus, no quenching was observed and the lifetime values ranged between 2.5 and 3 ns (n = 20). B) The lifetime values of AtGRIP-GFP ranged between 2.2 and 2.6 ns (n = 38). Co-expression of AtGRIP-GFP with mRFP-CASP, both being located in the Golgi stack, resulted in a lifetime distribution between 2.2 and 2.6 ns (n = 35). The lifetime distribution of the positive control AtGRIP-GFP co-expressed with mRFP-AtARL1 ranged between 1.8 and 2.4 ns (n = 76). C) Compared to the lifetime distribution of the control AtGRIP-GFP (B), upon co-expression with mRFP-AtRab-H1b lifetimes between 1.8 and 2.7 ns were observed (n = 84). The lifetimes of AtGRIP-GFP expressed with mRFP-AtRab-H1c ranged between 2 and 2.5 ns (n = 27). D) GFP-cTMF had a lifetime distribution between 2.2 and 2.9 ns (n = 42). When expressed with mRFP-AtRab-H1b, the lifetime ranged between 2.1 and 2.7 ns (n = 41), whereas upon co-expression with mRFP-AtRab-H1c values between 1.8 and 2.6 ns were observed (n = 17). E) The lifetime of GFP-VPS52 ranged between 2.3 and 2.7 ns (n = 25). Co-expression of GFP-VPS52 with mRFP-AtRab-H1b resulted in a lifetime range between 1.7 and 2.6 ns (n = 123). The lifetime distribution for GFP-VPS52 expressed with mRFP-AtRab-H1c ranged between 1.9 and 2.6 ns (n = 75).

Download figure to PowerPoint

Table 1.  Average lifetime values of GFP control and GFP–mRFP interaction experiments.
DonorAcceptorAverage life time (ns) and SDnp (two-tailed Student's t-test)
  1. Average lifetimes of GFP fusion proteins expressed alone or together with mRFP-tagged putative interaction partners and corresponding standard deviations (SD) and significance values.

ST-GFP2.72 ± 0.1484
ST-GFPST-mRFP2.74 ± 0.11200.4002
AtGRIP-GFP2.49 ± 0.1238
AtGRIP-GFPmRFP-AtCASP2.42 ± 0.13350.0077
AtGRIP-GFPARL1-mRFP2.14 ± 0.17761.14 × 10−22
AtGRIP-GFPmRFP-AtRab-H1b2.17 ± 0.2841.31 × 10−16
AtGRIP-GFPmRFP-AtRab-H1c2.19 ± 0.13276.16 × 10−15
GFP-cTMF2.57 ± 0.1742
GFP-cTMFmRFP-AtRab-H1b2.44 ± 0.14410.0001
GFP-cTMFmRFP-AtRab-H1c2.3 ± 0.21176.37 × 10−08
GFP-VPS522.42 ± 0.1125
GFP-VPS52mRFP-AtRab-H1b2.3 ± 0.171230.0005
GFP-VPS52mRFP-AtRab-H1c2.37 ± 0.17750.1122
image

Figure 4. Average lifetime values of control proteins and combinations. The average lifetime values were calculated from the lifetime distributions of the GFP fusion proteins both expressed alone and in combination with their putative binding partners. Compared with AtGRIP-GFP (2.49 ns), the average lifetimes values were reduced upon co-expression with mRFP-AtARL1 (2.14 ns), mRFP-AtRab-H1b (2.17 ns) and mRFP-AtRab-H1c (2.19 ns). Looking at the average lifetime values solely, only a slight reduction of average lifetime values was observed when comparing GFP-VPS52 alone (2.42 ns) to the combinations with mRFP-AtRab-H1b (2.3 ns) and mRFP-AtRab-H1c (2.37 ns). In contrast to that, quenching of GFP could clearly be observed looking at the lifetime distribution of individual Golgi bodies (see Figure 3). GFP-cTMF had an average lifetime of 2.47 ns, which was reduced upon co-expression with mRFP-Rab-H1b (2.44 ns) or mRFP-AtRab-H1c (2.3 ns). Almost no difference could be detected for ST-GFP (2.72 ns) and ST-GFP co-expressed with ST-mRFP (2.74 ns). Error bars represent standard deviations (see Table 1).

Download figure to PowerPoint

To determine the lifetime of AtGRIP-GFP, GFP-cTMF and GFP-AtVPS52, we expressed each fluorescent fusion transiently in wild-type tobacco leaf epidermal cells. The lifetime of AtGRIP-GFP ranged between 2.2 to 2.6 ns (Figure 3B) and an average of 2.49 ns (Figure 4). Data were collected from three samples with 38 Golgi bodies. This value was slightly lower than the lifetime of the ST-GFP control. Figure 5A,B shows a representative cell expressing AtGRIP-GFP and the corresponding false-coloured lifetime map respectively with green colours representing lifetime values around 2 ns and blue shades higher lifetimes of around 3 ns. A representative decay curve for a single point within a Golgi body with a lifetime of 2.54 ns is shown in Figure 5C. The χ2 value of 1 indicates an optimal single exponential fit. To confirm that spatial proximity of fluorophores in the same compartment, but with the fluorophores exposed on the cytoplasmic face of the Golgi, does not affect the lifetime, AtGRIP-GFP was co-expressed with the cis-Golgi matrix protein mRFP-AtCASP, which is targeted to the opposite end of the Golgi stack. Data from three samples with 35 Golgi bodies were analysed with lifetime values from 2.2 to 2.6 ns (Figure 3B) and an average of 2.42 ns (Figure 4). No reduction in the lifetime of AtGRIP-GFP was detected (Student's t-test p > 0.05, Table 1). The lifetime distribution of GFP-cTMF obtained from three samples with 42 Golgi bodies ranged between 2.2 and 2.9 ns (Figure 3D), averaging 2.44 ns (Figure 4). For GFP-AtVPS52, data was analysed from three samples with 25 Golgi bodies and the lifetime distribution ranged between 2.3 and 2.7 ns (Figure 3E), averaging 2.42 ns (Figure 4).

image

Figure 5. Interactions between coiled-coil proteins and small regulatory GTPases at the subcellular level. The subcellular localizations of interactions between coiled-coil proteins fused to GFP and small regulatory GTPases fused to mRFP were visualized with pseudo-coloured lifetime maps generated in SPCImage (Becker & Hickl, Germany). A) Confocal image of a tobacco leaf epidermal cell expressing AtGRIP-GFP alone as control. B) Pseudo-coloured lifetime map of the same cell. Golgi bodies appear in blue, indicating an unquenched lifetime around 2.5 ns. C) Representative decay curve for a single point analysis of AtGRIP-GFP with a lifetime of 2.54 ns and a χ2 value of 1, indicating an optimal single exponential fit. D) Confocal image of a cell expressing AtGRIP-GFP and mRFP-AtARL1. E) Lifetime map showing quenching of AtGRIP-GFP in punctate structures co-locating with mRFP-AtARL1. The lifetime reduction is reflected in the green colour. Occasionally, large bright aggregates labelled by both proteins were observed, which did not show reduced lifetime values (arrowhead). F) Representative decay curve for a single point analysis of AtGRIP-GFP + mRFP-ARL1 with a reduced lifetime of 2.08 ns. The χ2 of 1.05 indicates a single exponential optimal fit. G) Confocal image of two cells, the lower left expressing AtGRIP-GFP and mRFP-RabH1b and the upper right only expressing AtGRIP-GFP. H) The lifetime map depicts quenching of GFP at Golgi bodies in the cell expressing both constructs, whereas Golgi bodies in the cell expressing AtGRIP-GFP show no reduced lifetime (asterisk). I) Confocal image of a cell expressing GFP-VPS52 and mRFP-AtRab-H1b. J) The lifetime map shows that quenching of GFP-VPS52 does not occur in all, but only occurs in a few punctate structures labelled by GFP-VPS52 and mRFP-AtRab-H1b (arrowhead). Scale bars, 10 μm.

Download figure to PowerPoint

AtGRIP-GFP interacts with small regulatory GTPases in planta

We have previously demonstrated the interaction between AtGRIP and AtARL1 in vitro in plants (19,21). Here we used this protein combination as positive control. AtGRIP-GFP was co-expressed with mRFP-AtARL1 (Figure 5D) and three samples with 37 Golgi bodies were analysed. Compared to the non-quenched GFP control (Figure 5A,B), clear quenching of AtGRIP-GFP was observed and a change of Golgi body colour could be visualized in the false-coloured lifetime map (Figure 5E). Figure 5F shows a representative decay curve for a single point analysis with a reduced lifetime of 2.08 ns and a χ2 of 1.05, representing a good single exponential fit. The lifetime distribution was shifted towards a lower range between 1.8 and 2.4 ns (Figure 3B), averaging 2.14 ns (Figure 4) and indicating a strong interaction (Student's t-test p < 0.05, Table 1). The low number of higher lifetime values can be explained by the presence of large bright aggregates labelled by both proteins, which do not show reduced lifetime values (Figure 5E, arrowhead).

Co-expression of AtGRIP-GFP and mRFP-AtRab-H1b also resulted in quenching of AtGRIP-GFP. Data were collected from five samples with 76 Golgi bodies and the average lifetime was 2.17 ns (Figure 4) with lifetime values ranging from 1.8 to 2.7 ns (Figure 3C), indicating protein interaction (Student's t-test p < 0.05, Table 1). Figure 5G,H shows two cells, the left one expressing both constructs (arrowhead), whereas the right cell in the upper corner only expressed AtGRIP-GFP (cell marked with asterisk in Figure 5H). Golgi bodies in the cell with both constructs showed a reduced lifetime, whereas those labelled by AtGRIP-GFP had a lifetime between 2.3 and 2.6 ns.

Similarly, expression of AtGRIP-GFP with mRFP-AtRab-H1c resulted in a reduced lifetime with values between 2.0 and 2.5 ns (Figure 3C), averaging 2.19 ns (Figure 4). This result was statistically different to the control lifetime of AtGRIP-GFP (Student's t-test p < 0.05, Table 1). Data were collected from three samples with 27 Golgi bodies. Compared to the interaction of AtGRIP with AtRab-H1b, no values below 2.0 ns were observed for AtGRIP-GFP expressed with mRFP-AtRab-H1c, which might indicate a weaker interaction between mRFP-AtRab-H1c and AtGRIP-GFP.

GFP-cTMF interacts with mRFP-AtRab-H1c and may interact with mRFP-AtRab-H1b

GFP-cTMF was expressed together with mRFP-AtRab-H1b and data were collected from three samples with 41 Golgi bodies. The majority of lifetime values were between 2.3 and 2.7 ns (Figure 3D) with an average of 2.44 ns (Figure 4), but lower lifetimes of 2.1 or 2.2 ns were observed as well. This indicates that the two proteins might interact very weakly or only rarely (Student's t-test p < 0.05, Table 1). When GFP-cTMF was co-expressed with mRFP-AtRab-H1c, the lifetime distribution was shifted towards 2.2–2.4 ns with several lower values between 1.8 and 2.0 ns (Figure 3D), averaging 2.3 ns (Figure 4). Data was collected from one sample with 17 Golgi bodies and the appearance of lower lifetimes indicates that interaction took place (Student's t-test p < 0.05, Table 1).

GFP-AtVPS52 is quenched selectively by mRFP-AtRabH1b/c

GFP-AtVPS52 was co-expressed with mRFP-AtRab-H1b (Figure 5I) and 10 samples with 123 GFP-AtVPS52-labelled structures were analysed. Lifetimes between 1.7 and 2.6 ns were observed (Figure 3E), averaging 2.3 ns (Figure 4) and indicating an interaction between the two proteins (Student's t-test p < 0.05, Table 1). Surprisingly, GFP quenching was restricted to a subset of punctate structures (Figure 5J, arrowheads). A similar result was observed for co-expression of GFP-AtVPS52 with mRFP-AtRab-H1c, where data was collected from four samples with 75 GFP-AtVPS52-labelled structures. Here the lifetime values ranged between 1.9 and 2.6 ns (Figure 3E) with an average of 2.37 ns (Figure 4). Although there appears to be no statistically significant difference between the control and the protein combination because of the wide range of lifetime values (Student's t-test p > 0.05), quenching of GFP-AtVPS52 clearly indicates protein interaction in a subset of structures.

To confirm that the relatively small but significant differences in lifetimes between control and interacting partners occurred due to FRET, we attempted a control experiment in which we tried to bleach the mRFP acceptor. In theory, photobleaching of the acceptor should prevent non-radiative energy transfer from the donor to the acceptor and restore the GFP lifetime of the donor to a value similar to the control conditions. In practice, a full reversion to control lifetimes would, however, only be achieved after complete bleaching of the acceptor. In the protein combinations tested in our study, the mRFP fusion proteins ARL1 and Rab1-Hb/c cycle between Golgi bodies and a cytoplasmic pool. To prevent fluorescent recovery from the cytoplasm, we tried to bleach whole cells expressing AtGRIP-GFP and mRFP-ARL1 using the 543 nm laser line but this proved impossible, even after continuous scanning at 100% laser strength and very low resolution (125 × 125 pixels) over several minutes.

Discussion

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

Interactions between Golgi-localized coiled-coil proteins and small GTPases can be visualized in planta

The results from this study demonstrated that interactions between proteins associated at the Golgi apparatus in plants can be confirmed using a FRET-FLIM approach. We have shown that a number of putative matrix proteins interact with small regulatory GTPases. This is in agreement with the data from yeast and mammalian cells where a number of small GTPases of the Rab, YPT (Yeast Protein transport) and ARL and ARF families have been suggested to be involved in the recruitment of tethering proteins to Golgi membranes (11). Whether this is by direct recruitment of tethers, regulating molecular interaction before membrane fusion or regulating the accessibility of SNAREs to permit fusion is not clear. Interactions with Rab proteins might not only be required for Golgi localization of coiled-coil proteins but also for correct targeting of cargo carriers or cisternae at the Golgi stack. Sinka et al. (69) mapped multiple Rab binding sites on Drosophila GRIP domain proteins and proposed a model of a ‘tentacular Golgi’, where coiled-coil proteins are anchored with their C-termini to Golgi membranes and reach out with their N-termini into the cytoplasm to capture and subsequently distribute Rab-containing membranes within the polarized Golgi stack.

We have shown that the lifetime of GFP on Golgi-targeted constructs can be obtained and quenching measured to assess protein–protein interactions by FRET. Thus, in tobacco leaf epidermal cells we were able to verify results from in vitro protein binding assays and study such protein interactions in vivo. This permits rapid analysis of transiently or stably expressed protein combinations in planta at the organelle level.

We used mGFP6 (70) and mRFP (71) as donor and acceptor pair, as the widely used cyan fluorescent protein (CFP)-YFP combination has several disadvantages (45,63). CFP and YFP are able to form heterodimers (72,73) and therefore exhibit biexponential decay kinetics (74), whereas this problem is avoided using GFP-mRFP due to the monomeric behaviour of mRFP (71). CFP has also a lower extinction coefficient than GFP, meaning that a higher excitation intensity is needed, resulting in faster photobleaching (48). The spectral overlap in emission spectra between CFP and YFP can also be problematical and requires the use of narrow band pass filters (52).

GFP lifetimes might be influenced by their subcellular environment

All lifetime values for all different protein combinations tested are summarized in Table 1. ST-GFP stably expressed in tobacco plants had an average lifetime of 2.72 ns. The average lifetime of ST-GFP transiently co-expressed with ST-mRFP was 2.74 ns. This shows that detection of protein interaction by FRET-FLIM is specific, as pure spatial proximity of GFP- and mRFP-tagged constructs targeted to the same compartment did not lead to GFP quenching and a reduction in lifetime. Similarly, when the trans-Golgi marker AtGRIP-GFP was expressed with the cis-Golgi marker mRFP-CASP, no quenching of AtGRIP-GFP was observed, indicating as predicted that no interaction took place.

The lifetimes of GFP averaged 2.49 ns when fused to AtGRIP, 2.57 ns for cTMF and 2.42 ns for AtVPS52. The reported lifetime for eGFP was 2.4 ns in Escherichia coli(75) and in tobacco leaf epidermal cells fused to the Arabidopsis EB1a (microtubule end binding protein 1a, 62). However, to our knowledge no data is available for the lifetime of mGFP6 (70), a GFP variant optimized for expression in plants and part of the pMDC vectors used in this study (76). It has been suggested that the refractive index of the environment influences the lifetime of GFP (77) and this may explain the slight differences in the lifetimes of our GFP controls.

Protein interactions visualized by FLIM

AtGRIP-GFP had an average lifetime of 2.49 ns, whereas AtGRIP-GFP expressed together with mRFP-AtARL1 had a reduced lifetime of 2.14 ns. This combination represented a positive control because the interaction between the two proteins has been suggested to be conserved in different species (29). This result has also previously been shown in vitro for the Arabidopsis proteins using a yeast two-hybrid assay and affinity chromatography (19) where it was suggested that recruitment of the GRIP domain protein to the Golgi was mediated by the ARL1 GTPase. AtGRIP-GFP was also quenched when co-expressed with mRFP-AtRab-H1b with an average lifetime of 2.17 ns or mRFP-AtRab-H1c with an average lifetime of 2.19 ns. However, results from the yeast two-hybrid analyses only indicated interaction with AtRab-H1b. A possible explanation could be that the interaction between AtGRIP and AtRab-H1c is so weak as not to be detectable in the yeast two-hybrid assay. Alternatively, the two proteins might not have interacted in vitro due to the artificial environment or the in vivo interaction might be temporally and/or spatially restricted. The data on the GRIP domain protein indicate that two GTPases may be involved in the binding of AtGRIP to the trans-Golgi. Such dual GTPase binding to GRIP domain proteins has recently been reported in mammalian cells where it was shown that Rab6 binding promotes association of ARL1 with the GRIP domain (78).

GFP-cTMF alone had an average lifetime of 2.57 ns. When GFP-cTMF was co-expressed with mRFP-AtRab-H1b and mRFP-AtRab-H1c, the average lifetimes were 2.44 and 2.3 ns, but in case of GFP-cTMF and mRFP-AtRab-H1b some Golgi bodies had lifetimes of 2.1 and 2.2 ns, and for GFP-cTMF and mRFP-AtRab-H1c significantly lower lifetimes of 1.8–2.2 ns were observed. These lifetime distributions suggest that GFP-cTMF might interact more strongly with AtRab-H1c than with AtRab-H1b. In a yeast two-hybrid system and in pulldown experiments, cTMF was interacting equally strong with AtRab-H1b and AtRab-H1c(22). This could indicate that the proteins have the principal capability to bind in vitro, but that the interaction between GFP-cTMF and the Rab GTPases might be transient or spatially restricted in planta and generally weaker between cTMF and AtRab-H1b. Another possible explanation could be that the binding site for AtRab-H1c on GFP-cTMF is located closer to its C-terminus than for AtRab-H1b, thereby resulting in a higher energy transfer from the donor to the acceptor. Ideally, equivalent concentrations of donor and acceptor would give optimal FRET analysis conditions. However, in practice saturating levels of the donor with respect to the acceptor provide a good approximation. Thus, although quenching of GFP is not generally dependent on saturating levels of the acceptor (52), it cannot be excluded that the difference in lifetime reduction could be influenced by a small difference in expression levels of mRFP-Rab-H1b and mRFP-AtRab-H1b in tobacco leaf epidermal cells.

Surprisingly, quenching of GFP-AtVPS52 co-expressed with mRFP-AtRab-H1b/c did not appear homogeneous but was restricted to a subpopulation of labelled structures. Insertional pok mutant Arabidopsis lines showed impaired pollen tube growth (40), but other than that not much is known about the function of AtVPS52. VPS52 homologues in yeast and animals have been implied in protein sorting at late Golgi compartments as part of the GARP complex (36,37,39), and as POK/AtVPS52 has been shown to locate to post-Golgi compartments in plants (41) it might be involved in similar processes (40,41). Possibly, the selective quenching of GFP-AtVPS52 could indicate transient interaction of AtVPS52-labelled late Golgi compartments with the Golgi apparatus or might even imply the existence of different Golgi body subpopulations.

Future perspectives of FRET-FLIM analysis

This study shows that FRET-FLIM analysis provides a powerful tool to confirm interactions between proteins observed in vitro as well as to study unknown protein combinations in planta at the organelle level. Summarizing our observations from in vitro and in planta approaches we conclude that in vitro methods like the yeast two-hybrid screen employed here provide a good first indication whether two proteins have the capability to interact. Using the FRET-FLIM approach in planta, however, provided us with valuable additional information. We were not only able to distinguish between stronger and weaker GFP quenching implying different strengths of interactions but in some cases we also observed selective GFP quenching in a subset of structures, which might be explained by temporally or spatially restricted interactions. Could those interactions be dependent on the activity of different Golgi stacks, or might different Golgi bodies have different functions and therefore require different sets of regulatory proteins at different times? Generally, proteins might interact only at certain developmental stages or time points within the cell or organelles and this could be studied using FRET-FLIM. If the resolution of data analysis could be increased, it might even be possible to distinguish the exact location of protein interaction within the Golgi stack.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

Plasmids

Standard molecular techniques were used as described in Sambrook and Russell (79). AtVPS52 was amplified from Arabidopsis cDNA with the following primers:

VPS52-AscI (5-AGG CGC GCC AAA AAT GTC CGA CAT TTC CAT-3) and VPS52-NotISTOP (5-GGA AAA AAG CGG CCG CTC AGA AAG TCT TGG AGT-3). The binary vector pMDC43 (76) was used for fusion to the C-terminus of mGFP5 (80). The polymerase chain reaction (PCR) fragment was cloned into pENTR1A-MCS (19) and transferred into pMDC43 using the Gateway system following instructions provided by the manufacturer (Invitrogen).

To construct pMDC83-mRFP (for fusion to the N-terminus of mRFP), mRFP was amplified with mRFP-AscI (5-AGG CGC GCC TCG AGA TGG CCT CCT CCG AGG ACG TC-3) and mRFP-SacI (5-TCC TCC GAG CTC TTA GGC GCC GGT GGA GTG GCG-3). The PCR fragment was digested with AscI and SacI and used to replace the mGFP5 gene in pMDC84. For pMDC43-mRFP (for fusion to the C-terminus of mRFP), mRFP was amplified with mRFP-KpnI (5-CGG GGT ACC ATG GCC TCC TCC GAG GAC GTC-3) and mRFP-AscI (5-AGG CGC GCC GGC GCC GGT GGA GTG GCG-3). The PCR fragment was digested with KpnI and AscI and introduced into the KpnI and AscI sites of pMDC32. AtRabH1b, AtRabH1c and AtARL1 were fused to mRFP by recombination of their respective pENTR clones (19,22) with pMDC83-mRFP or pMDC43-mRFP using the Gateway system.

For yeast two-hybrid analysis, the cDNA encoding the full length of the AtGRIP (At5g66030) was amplified by PCR using the primers Y2H-F-GRIP (5-TCC AGC ACC ATT CCC GGG GAT GTC CGA AGA CAA G-3) and Y2H-R-GRIP (5-CCA GCA CAG TTT CCC GGG GTT ATG AAA ACG AGA ATC T-3) containing the SmaI restriction site (underlined) for cloning purposes. The PCR product was cloned into pGBKT7 bait vector (BD Biosciences Clontech). The AtRab-H1b and the mutants AtRab-H1b[T23N], AtRab-H1b[Q68L] and AtRab-H1b[N122I] were amplified from the plasmids pVKH18-En6-Rab-H1b, pVKH18-En6-Rab-H1b[T23N], pVKH18-En6-Rab-H1b[Q68L] or pVKH18-En6-Rab-H1b[N122I] (32) by PCR using the primers Y2H-RabH1.b-F (5-G TCC AGC ACC ATT CCC GGG ATG GCT CCG GTC TCG GCA-3 and Y2H-RabH1.b-R (5-CCA GCA CAG TTT CCC GGG GCT AAC AAG AGC ATC CTC C-3). The primers Y2HRabH1.c-F (5-A TCC AGC ACC ATT CCC GGG ATG GCT TCG GTT TCA CCT TTG GCA-3) and Y2HRabH1.c-R (5-CCA GCA CAG TTT CCC GGG TCA ACA AGA ACA GCC TCC ACC ACC-3) were designed to amplify AtRab-H1c from the plasmid pVKH18-En6-Rab-H1c. The PCR fragments were cloned in the pGADT7-Rec prey vector, again using SmaI as a restriction site.

Yeast two-hybrid analysis

Yeast two-hybrid analysis was performed as described recently (22). Yeast strain AH109 was sequentially transformed with pGBKT7-GRIP and pGADT7-Rec-Rab-H1b, pGADT7-Rec-Rab-H1c, pGADT7-Rec-Rab-H1b[T23N], pGADT7-Rec-Rab-H1b[Q68L] or pGADT7-Rec-Rab-H1b[N122I] using a lithium acetate method (81). Colonies were selected in plates lacking histidine, tryptophan, leucine and adenine up to a period of 10 days. Positive yeast transformants were replated on β-galactosidase (BD Biosciences Clontech) plates to test the expression of the reporter gene MEL1.

Expression of fluorescent protein fusions in tobacco plants

Transient expression of fluorescent protein fusions in tobacco leaves was performed using agrobacterium-mediated infiltration of lower leaf epidermal cells (82). Nicotiana tabacum sp. plants were grown in the greenhouse at 21^C, 14 h light, 10 h dark and were used for agrobacterium infiltration after 5–6 weeks. Leaf samples were analysed 2–4 days after infiltration. Stable ST-GFP tobacco plants were created in our lab as described recently (82).

Latrunculin B treatment

To inhibit Golgi movement, tobacco leaf samples were treated with the actin-depolymerising agent latrunculin B (4). Latrunculin B (Calbiochem) was dissolved in dimethyl sulphoxide (DMSO) at 1 mm and stored at–20^C. Leaf samples were incubated in a 25 μm working solution for at least 15 min, until Golgi movement had stopped.

Confocal microscopy

High-resolution confocal images were obtained using an inverted Zeiss LSM 510 confocal laser scanning microscope (CLSM) microscope and a 100x oil immersion objective. For imaging GFP in combination with mRFP, excitation lines of an argon ion laser of 488 nm and a helium ion laser of 543 nm were used alternately with line switching, using the multitrack facility of the CLSM. Fluorescence was detected using a 488/543 dichroic beam splitter, a 505–530 band pass filter for GFP and a 560–615 band pass filter for mRFP.

FRET-FLIM data acquisition

FRET-FLIM analysis was carried out using a two-photon microscope setup constructed in the Central Laser Facility of the Rutherford Appleton laboratory (66). Custom built XY galvanometers (GSI Lumonics) were used for the scanning system. Laser light at a wavelength of 920 ± 5 nm was obtained from titanium sapphire laser (Mira, Coherent) pumped by a frequency doubled vanadate laser (Coherent Lasers) producing 180 fs pulses at 75 MHz. The laser beam was focussed to a diffraction limited spot through a water immersion ultraviolet corrected objective [Nikon VC × 60, numerical aperture (NA) 1.2, water immersion]. Specimens were illuminated at the microscope stage of a modified Nikon TE2000-U. Fluorescence emission was collected without descanning, by-passing the scanning system, and passed through a bandpass filter (BG39, Comar). The scan was operated in the normal mode and line, frame and pixel clock signals were generated and synchronized with an external fast microchannel plate photomultiplier tube (Hamamatsu R3809U) used as the detector. These were linked via a time-correlated single photon counting (TCSPC) PC module SPC830 (Becker and Hickl).

Expression of the fluorescent markers was checked using a two-channel Nikon eC1 confocal scanning system coupled to an argon ion 488 nm laser for GFP excitation and a green HeNe at 543 nm for the mRFP excitation. For FLIM measurements of control constructs, cells with high expression levels were preferred. When analysing GFP and mRFP combinations, cells with similar expression levels of both fluorophores were preferred to obtain approximately equal levels of interaction partners within the cell. Fluorescence lifetime imaging micrographs were analysed using the SPCImage analysis software (Becker and Hickl, http://www.becker-hickl.com/). Different parameters were varied to obtain the optimal curve fitting for the decay graph for the majority of Golgi bodies, and χ2 values over 1.5 indicating poor fitting of the decay data were rejected for further data collection. (An exponential decay fit was considered good when the χ2 value was 1.) As the current software did not permit automatic analysis of whole Golgi bodies, lifetime values were collected on a single pixel basis from the centre of individual Golgi bodies and lifetimes were recorded on a Microsoft Excel worksheet. The collected data values were used to generate histograms depicting the lifetime distribution of Golgi bodies within the samples. Lifetime values of controls and protein combinations were statistically analysed by performing a two-tailed Student's t-test in Microsoft Excel.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References

We would like to thank Mark Curtis for the pMDC vector constructs and Roger Tsien for the mRFP construct. The Biotechnology and Biological Sciences Research Council are acknowledged for grant funding some of this work. Access to the Central Laser Facility and the multiphoton laboratory was funded by the Science and Technology Facilities Council.

References

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References