AtCPSF30 does not appear to possess an inherent nuclear localization signal
To study the subcellular distribution of the Arabidopsis
CPSF30, the AtCPSF30 coding region was fused to the C-termini of the DsRed2 or GFP genes (a list of the constructs used is provided in Table ). Control constructs encoded unmodified DSRed2 [19
], as well as fusion proteins containing the product of the Arabidopsis
NDA2 gene (a mitochondrial marker; [20
]), a synthetic endoplasmic reticulum localization sequence [21
], and a zinc finger protein (AtZFP11) known to localize to the nucleus [22
]. In all of these constructs, expression was under control of the cauliflower mosaic virus 35S promoter; the choice of overexpression using the 35S promoter was based on prior observations indicating that the levels of AtCPSF30 in wild-type plants are exceedingly low [16
], and thus detection of the products of transgenes driven by the promoter from the AtCPSF30 gene (At1g30460) was not feasible. Overexpression in transient assays has been used by others to study protein localization and protein-protein interactions, and often reveals unexpected aspects of the functioning of protein complexes [23
]. It was thus expected that this approach would provide information about the inherent subcellular location signals carried by AtCPSF30. Accordingly, various combinations of these plasmids were introduced into tobacco leaf cells using a biolistics apparatus and the protein expression and localization assessed using confocal microscopy.
List of localization constructs used in this study
In cells co-transfected with DSRed2-AtCPSF30 and GFP-AtZFP11, no accumulation of DSRed2-AtCPSF30 could be seen in the nucleus, (Figure , panels A-C). Instead, a distribution of DSRed2-AtCPSF30 outside of the nucleus, in distinct foci, was apparent. A similar pattern was not seen in cells that express unmodified DSRed2 (Figure , panel M). These foci did not correspond to chloroplasts (visualized using the fluorescent properties of the plastid; not shown). Moreover, they did not co-localize with the GFP-NDA2 marker (Figure , panels D-F), indicating that they were not mitochondrially-localized. These extranuclear foci were noticeably mobile (see Additional File 1
), and were coincident with the endoplasmic reticulum (Figure , panels G-I). However, some locales of the ER marker (the nuclear envelope and nucleolus) were devoid of DSRed2-AtCPSF30. Invariably, while DSRed2-AtCPSF30 could not be seen in the nucleus to any appreciable extent, one or more of the cytoplasmic DSRed2-AtCPSF30-containing foci abutted the nucleus (as indicated with the arrows in panels A and C of Figure ).
Figure 1 Subcellular distribution of AtCPSF30 in tobacco cells. In this figure, the overlays of the two images corresponding to pairs of fusion proteins are shown in the column on the right (panels C, F, I, and L). The AtCPSF30 is visualized in panels A, D, G, (more ...)
As an important cytoplasmic location for RNA metabolism is the so-called processing body, or P-body [25
], the co-localization of AtCPSF30 with an Arabidopsis
P-body marker was studied. For this, DSR was fused to the Arabidopsis
Dcp2 protein-coding region and GFP was fused to AtCPSF30. Dcp2 is a component of the Arabidopsis
] and serves in this study as a marker for this structure. In cells co-expressing the DSRed2-AtDcp2 and GFP-AtCPSF30 fusion proteins, the DSRed2-AtDcp2 was found in distinctive foci (Figure , panel K), much as has been observed by others [28
]. These foci were distinct from mitochondria and chloroplasts (not shown) in these cells. The DSRed2-AtDcp2 foci were coincident with the GFP-AtCPSF30 foci (Figure , panels J-L). This indicates that, in cells overexpressing the two proteins, GFP-AtCPSF30 and DSRed2-AtDcp2 are present in the same structures.
The interaction between AtCPSF30 and AtCPSF160 promotes nuclear localization of AtCPSF30
Because messenger RNA 3' end formation is a nucleus-localized RNA processing event, the absence of DSRed2-AtCPSF30 from the nucleus was surprising. One possible explanation for this observation is that AtCPSF30 by itself possesses no nuclear localization information, but rather is recruited to the nucleus via interactions with other CPSF subunits. In this case, in cells transiently expressing DSRed2-AtCPSF30 from the 35S promoter, the protein might be present in a vast excess over other interacting partners and would thus localize to a "default" location in the cell. To explore this possibility, the DSRed2-AtCPSF30 fusion protein was co-expressed with GFP fusion proteins containing AtCPSF100, AtCPSF160, AtCPSF73(I), and AtCPSF73(II), respectively. AtCPSF160 and AtCPSF100 were chosen because they have been reported to interact with AtCPSF30 [18
]. AtCPSF73(I) and AtCPSF73(II) are two relatives of the 73 kD subunit of CPSF in mammals and the yeast homolog Ysh1. While AtCPSF73(I) and AtCPSF73(II) apparently do not interact with AtCPSF30 in two-hybrid assays [18
], they are part of the Arabidopsis
CPSF complex and localize to the nucleus [30
In plant cells that co-express the GFP-AtCPSF160 and DSRed2-AtZFP11 fusion proteins, both proteins accumulated in the nucleus (Figure , panels A-C). This confirms the expected location of AtCPSF160 in plant cells, and indicates that AtCPSF160 possesses nuclear localization signals. Invariably, when GFP-AtCPSF160 was co-expressed with DSRed2-AtCPSF30, the latter accumulated exclusively in the nucleus (as did GFP-AtCPSF160; Figure , panels D-O). However, the distribution of these proteins within the nucleus was often different from that seen for GFP-AtCPSF160 by itself. In some cases (Figure , panels D-F), the distribution of the two proteins in the nucleus was similar to the nuclear marker (Figure , panel A). However, the more frequent result was that GFP-AtCPSF160 and DSRed2-AtCPSF30 co-localized to distinctive structures within the nucleus (two representative cells are shown in Figure , panels G-L); these structures also contained DNA, as they could be stained with Hoechst stain (Figure , panels M-O).
Figure 2 Localization of AtCPSF30 in cells that also express AtCPSF160. GFP-AtCPSF160 was co-expressed with DSRed2-AtZFP11 (panels A-C) or DSRed2-AtCPSF30 (panels D-O). The column labeled "CPSF160" denotes the images of the GFP-AtCPSF160 fusion protein, and the (more ...)
AtCPSF30 consists of three distinct domains [14
] – a central core that includes the evolutionarily-conserved triad of CCCH zinc finger motifs that is flanked by novel N- and C-terminal domains (this general structure is illustrated in Figure ). To determine the part(s) of AtCPSF30 that are important for the interaction inferred by the effects of AtCPSF160 on AtCPSF30 localization, two deletion derivatives of AtCPSF30 were studied. These derivatives, termed m4 and m9 after Delaney et al
], lack either the C-terminal or N-terminal domains, respectively; together with the full-sized protein they permit an assignment of interactions to one of the three domains of the protein.
Figure 3 Localization of mutant AtCPSF30 isoforms in the presence of AtCPSF160. A. Illustration of the two deletion mutants (m4 and m9) used in this study; these mutants have been described in detail elsewhere [13,14]. B. Localization of fusion proteins in cells (more ...)
In cells expressing both GFP-AtCPSF160 and the DSRed2-m4 mutant, a range of results was obtained. In some cells (Figure , panels A-C), GFP-AtCPSF160 was retained in the nucleus, while the DSRed2-m4 mutant protein displayed the subcellular distribution seen with DSRed2-AtCPSF30 in the absence of GFP-AtCPSF160 (as seen in Figure ). In such cells, the distribution of GFP-AtCPSF160 in these nuclei was similar to that shown in Figure , panel B, as opposed to the concentration in subnuclear domains seen in Figure , panels H, K, and N. In some cells (Figure , panels D-F show a representative one), in addition to cytoplasmic DSRed2-m4, nuclear accumulation reminiscent of that seen in experiments performed with full-sized DSRed2-AtCPSF30 (Figure , panels G-L) was also seen. These results reveal a total (panels A-C) or partial (panels D-F) loss of the interaction between AtCPSF30 and AtCPSF160, such that some cytoplasmic DSRed2-m4 protein could be seen in cells co-expressing GFP-AtCPSF160. Thus, the C-terminal domain of AtCPSF30 seems to be important for efficient interactions with AtCPSF160.
In contrast, GFP-AtCPSF160 co-localized with the DSRed2-m9 mutant within the nucleus, and no extranuclear DSRed2-m9 mutant could be seen in cells co-expressing these two proteins (Figure , panels G-I). However, the novel subnuclear domains in which the two wild-type proteins accumulated were never seen in the experiments with the DSRed2-m9 mutant. Thus, the DSRed2-m9 mutant protein retains the ability to interact with GFP-AtCPSF160, but has lost the "ability" to change the subnuclear distribution of this protein. In the absence of AtCPSF160, the DSRed2-m4 and DSRed2-m9 variants were distributed in cells much as were the AtCPSF30 fusion proteins (not shown).
Altered subcellular distributions in cells co-expressing AtCPSF30 with AtCPSF100 or AtCPSF73(I)
Consistent with what has been reported elsewhere [30
], GFP-AtCPSF100 co-localized with the nuclear marker (DSRed2-AtZFP11) when both were expressed in tobacco cells (Figure , panels A-C). Remarkably, co-expression of GFP-AtCPSF100 with DSRed2-AtCPSF30 changed the location of GFP-AtCPSF100, such that was it was largely in cytoplasmic foci (Figure , panel E). These foci also contained DSRed2-AtCPSF30 (Figure , panels D-F). This dramatic change in subcellular distribution of GFP-AtCPSF100 indicates that the interaction between this protein and AtCPSF30 [18
] has the potential to interfere with the nuclear localization of AtCPSF100.
Figure 4 Localization of AtCPSF30 and the m4 and m9 mutants in cells that also express AtCPSF100. GFP-AtCPSF100 was co-expressed with DSRed2-AtZFP11 (panels A-C) or DSRed2 fused to AtCPSF30 (panels D-F), the m4 mutant (panels G-I), or the m9 mutant (panels J-L). (more ...)
Similar results were obtained in cells co-expressing GFP-AtCPSF100 and the DSRed2-m4 variant (Figure , panels G-I). However, the nuclear localization of GFP-AtCPSF100 was largely restored in cells co-expressing the DSRed2-m9 variant along with GFP-AtCPSF100 (Figure , panels J-L). In these cases, the DSRed2-m4 and DSRed2-m9 variants remained in cytoplasmic foci, much as was seen with the wild-type DSRed2-AtCPSF30 (e.g., Figure ) and with these two proteins when expressed without any CPSF partner (not shown). Thus, the redistribution of GFP-AtCPSF100 due to co-expression with DSRed2-AtCPSF30 requires the N-terminal part of AtCPSF30.
As was seen with GFP-AtCPSF160 and GFP-AtCPSF100, GFP-AtCPSF73(I) was located in the nucleus when co-expressed with the nuclear marker (Figure , panels A-C). In cells co-expressing GFP-AtCPSF73(I) and DSRed2-AtCPSF30, GFP-AtCPSF73(I) remained largely in the nucleus, although cytoplasmic foci containing GFP-AtCPSF73(I) were also discernible (Fig , panel E). In these cells, DSRed2-AtCPSF30 co-localized with GFP-AtCPSF73(I) in the nucleus and cytoplasm (Figure , panels D-F). Interestingly, nucleus-localized GFP-AtCPSF73(I) and DSRed2-AtCPSF30 accumulated in numerous large foci (inset of Figure , panel F).
Figure 5 Localization of AtCPSF30 and the m4 and m9 mutants in cells that also express AtCPSF73(I). GFP-AtCPSF73(I) was co-expressed with DSRed2-AtZFP11 (panels A-C) DSRed2 fused to AtCPSF30 (panels D-F), the m4 mutant (panels G-I), or the m9 mutant (panels J-L). (more ...)
In cells co-expressing the DSRed2-m4 variant along with GFP-AtCPSF73(I), the DSRed2-m4 protein was found in the cytoplasm but not the nucleus (Figure , panels G-I). In these cells, GFP-AtCPSF73(I) was found in the nucleus, but the distribution within the nucleus was more similar to the protein in cells that do not express DSRed2-AtCPSF30 (Figure , panel B). Additionally, GFP-AtCPSF73(I) could also be found in cytoplasmic foci that also contained the DSRed2-m4 variant (Figure , panels H and I). In cells co-expressing the DSRed2-m9 mutant and GFP-AtCPSF73(I) proteins, GFP-AtCPSF73(I) was found predominantly in the nucleus, and the DSRed2-m9 protein in the cytoplasm (Figure , panels J-L), suggestive of a lack of any interaction between these proteins. These results indicate that AtCPSF30 interacts with AtCPSF73(I) in a manner that promotes a substantial nuclear localization of AtCPSF30, and that the nuclear localization of AtCPSF30 in AtCPSF73(I)-expressing cells requires the N- and C-terminal domains of AtCPSF30.
Consistent with a previous study [30
], GFP-AtCPSF73(II) localized to the nucleus when co-expressed with the nuclear marker (Figure , panels A-C). Co-expression of GFP-AtCPSF73(II) with DSRed2-AtCPSF30 did not alter this distribution (Figure , panels D-F). Moreover, the cytoplasmic location of DSRed2-AtCPSF30 was not affected by co-expression with GFP-AtCPSF73(II); in particular, no DSRed2-AtCPSF30 could be seen in nuclei. This result is consistent with other results [18
] that indicated that these two proteins do not interact.
Figure 6 Localization of AtCPSF30 in cells that also express AtCPSF73(II). GFP-AtCPSF73(II) was co-expressed with DSRed2-AtZFP11 (panels A-C) or DSRed2-AtCPSF30 (panels D-F). The column labeled "CPSF73-II" denotes the images of the GFP-AtCPSF73(II) fusion protein, (more ...)