Green fluorescent protein as a marker to investigate targeting of organellar RNA polymerases of higher plantsin vivo

Note on terminology

Designation of the plant phage-type RNA polymerase (RpoT) genes follows the rules proposed by the Commission on Plant Gene Nomenclature. The ‘T’ at the end of the gene name indicates the relationship to the T7- and T3-type RNAP. The numbers refer to the different members of the gene family within a species (cf. Hess & Börner 1999).

Introduction

In addition to the nuclear eukaryotic-type RNA polymerases (RNAP), higher plants possess another set of polymerases transcribing the mitochondrial and the chloroplast genome. In fungi, the mitochondrial RNAP has been identified as an enzyme most similar to RNAP of the bacteriophages T3, T7 and SP6. It has long been suggested ( Tracy & Stern 1995) that the same type of polymerase is transcribing the plant mitochondrial genome. The other organellar genome of plants, the plastome, encodes a multisubunit eubacterial-like RNAP (plastid encoded polymerase; PEP). However, evidence has accumulated ( Maliga 1998) indicating that a second, nuclear-encoded polymerase (NEP) participates in the transcription of the plastid genome. Based on promoter studies and biochemical data ( Liere & Maliga 1999; Lerbs-Mache 1993), the NEP has been proposed to belong to the same type of polymerases as the mitochondrial RNAP. While PEP initiates transcription at typical sigma-like promoters, NEP uses a different kind of promoter, showing in several cases homology to a YRTA consensus found in numerous mitochondrial promoter sequences of higher plants (for a recent review see Hess & Börner 1999).

Recently, the first full-length cDNA encoding a phage-type RNA polymerase of a higher plant, Chenopodium album, was isolated in our laboratory ( Weihe et al. 1997), and subsequently two full-length cDNA and the corresponding genes from Arabidopsis thaliana, ARAth;RpoT;1 (Y) and ARAth;RpoT;3 (Z) could be characterized ( Hedtke et al. 1997). The amino acid sequences of the derived translation products of the plant Rpo genes show extensive similarity to the phage-like mitochondrial RNAP from fungi. The putative plant phage-type polymerases differ substantially in their extreme amino terminal sequences, which exhibit features of mitochondrial and chloroplast targeting sequences, respectively. As demonstrated by import experiments with purified pea mitochondria and spinach chloroplasts in vitro, the putative transit peptide of ARAth;RpoT;1 was capable of targeting a fusion protein to isolated mitochondria, whereas the amino terminus of ARAth;RpoT;3 conferred targeting to chloroplasts ( Hedtke et al. 1997). No experimental data have yet been provided on targeting of CHEal;RpoT, the phage-type RNAP from C. album. Computer analyses of the amino terminal sequence suggest, however, that it represents a mitochondrial presequence ( Weihe et al. 1997).

In vitro import of proteins into purified organelles can, in certain cases, give misleading results and may not reflect the actual in planta localization of the respective proteins ( Silva-Filho et al. 1997). Here we demonstrate for the first time by an in planta approach that the putative transit peptides of the phage-type organellar RNAP from A. thaliana and C. album are capable of targeting the green fluorescent protein (GFP) into mitochondria and plastids, respectively. GFP can be utilized as a powerful marker for subcellular localization and as a label for plant organelles in vivo, permitting direct observation of targeted fusion proteins in living plant cells ( Haseloff et al. 1997; Köhler et al. 1997a, 1997b; Sheen et al. 1995). We fused the putative transit peptides of the phage-type RNAP from A. thaliana and C. album (ARAth;RpoT;1, ARAth;RpoT;3 and CHEal;RpoT) to GFP and investigated the subcellular localization of the fusion proteins in transgenic Arabidopsis plants.

Results and discussion

DNA sequences encoding the putative transit peptides of the three polymerases and a modified GFP (mGFP5; Haseloff et al. 1997) were fused and set under control of a 35S promoter in the vector pGPTV-bar ( Fig. 1). As a control, mGFP5 fused to an ER signal peptide from A. thaliana basic chitinase was used ( Haseloff et al. 1997). The constructs were utilized to stably transform Arabidopsis plants. About one-third of the transgenic plants showed bright green fluorescence in root and leaf tissues. None of the constructs used produced an abnormal phenotype. All transformed plants were fertile and gave rise to normal seed (data not shown).

Using epifluorescence and confocal laser scanning microscopy we investigated the subcellular distribution of the green fluorescence in root and leaf cells of plants expressing the targeted GFP. Plants expressing GFP with the ER targeting signal showed bright green fluorescence ( Fig. 2a) localized to a network of thread-like structures excluding the nucleus and the plastids ( Fig. 3a), as described earlier by Haseloff et al. (1997).

In cells expressing GFP fused to the putative transit peptide of ARAth;RpoT;1, fluorescence was restricted to structures of typical mitochondrial morphology, 0.5–1 μm round- or elliptical-shaped particles that were concentrated at the outer edge of the cell (in guard cells; 2, 3) or distributed throughout the cytoplasm often forming conglomerates (in root hair cells; Fig. 2c), with the number of fluorescent mitochondria-like structures in guard cells often being considerably lower than in root cells. Essentially the same results were obtained when plants were studied expressing GFP fusions that contained the transit peptide of the putative C. album RNAP, CHEal;RpoT ( 2, 3). Using the chlorophyll autofluorescence of the chloroplasts in the guard cells and confocal imaging with combined red and green channels, information was provided about the localization of the GFP fusions relative to chloroplasts. Plants transformed with the ARAth;RpoT;1 as well as the CHEal;RpoT construct, showed GFP fluorescence localized to the same subcellular structures, excluding the chloroplasts ( 2, 3). To verify that the GFP fluorescence observed was indeed localized to mitochondria, whole transgenic plants were stained with a mitochondrial-specific dye for living cells, MitoTracker CM-H₂XRos ( Haugland 1996), and the GFP and the dye fluorescence were compared within the same cells. Comparison of Fig. 2(d) and Fig. 2(e) (CHEal;RpoT) clearly reveals co-localization of the red mitochondrial-specific dye and the green GFP fluorescence. Co-localization of MitoTracker and GFP fluorescence was also observed when plants transformed with the Arabidopsis RpoT;1 construct were examined (results not shown). Our data demonstrate that the CHEal;RpoT-GFP and the ARAth;RpoT;1-GFP fusion proteins are localized specifically to mitochondria. The localization of the Chenopodium construct provides the first experimental evidence for a mitochondrial function of the C. album phage-type RNAP.

A completely different picture was obtained with plants expressing the GFP fusion with the ARAth;RpoT;3 transit peptide. Guard cells of dark-grown plants exhibited under epifluorescence illumination a yellow fluorescence of plastids caused by overlapping green GFP fluorescence and red chlorophyll autofluorescence of the developing chloroplasts ( Fig. 2f). The fluorescence pattern of these cells was strikingly different from both mitochondrial- and ER-targeted GFP. In root hair cells of plants transformed with the Arabidopsis RpoT;3 construct ( Fig. 2g), green GFP fluorescence was localized in round structures much larger than mitochondria, 2.5–4 μm in size, most probably representing non-green (lack of chlorophyll autofluorescence) plastids (leucoplasts). We have never detected fluorescent images similar to that obtained when the putative transit peptides of ARAth;RpoT;1 or CHEal;RpoT were in the fusion constructs. Further striking evidence for plastid localization of the ARAth;RpoT;3-GFP fusions provided confocal images of leaf cells ( 3, 1, e2): green GFP fluorescence co-localized clearly with red chlorophyll autofluorescence in the chloroplasts of the guard cells. Surrounding cells contained the GFP fusion protein as well; however, the abundance of chloroplasts as well as mechanical damage of the organelles caused by the mounting procedure made visualization of particular structures impossible. Our data demonstrate the ability of the transit peptide of the putative Arabidopsis RNA polymerase RpoT;3 to confer transfer of GFP to both green chloroplasts and non-green plastids in the transgenic plants.

Discrepancies between in vivo and in vitro targeting as well as co-targeting of one protein to both chloroplasts and mitochondria have been reported, and artefacts in in vitro import experiments cannot be ruled out totally ( Chow et al. 1997; Creissen et al. 1995; Silva-Filho et al. 1997). For the two A. thaliana RNAP, our data on subcellular localization, which were essentially based on in vitro import experiments ( Hedtke et al. 1997), can now be confirmed and considerably extended by demonstrating the targeting function of the transit peptides in the living plant and in a homologous system. Examination of the transgenic plants revealed no co-targeting of the individual constructs to both mitochondria and plastids. Nevertheless, we cannot exclude that a very low percentage of targeting of the mitochondrially located fusion constructs to chloroplasts (or vice versa) was occurring; the limit of such observations is set by the fluorescence detection level of GFP. The coincidence of the in vivo results of the present study and our earlier in vitro import data is a strong evidence for a final conclusion on the subcellular localization of the three putative RNA polymerases investigated. Thus, our data confirm the plastidal localization of ARAth;RpoT;3, most probably representing the postulated NEP that co-exists in the plastids together with the organelle-encoded PEP.

Recently, a third gene encoding a putative phage-type RNAP has been identified in A. thaliana (ARAth;RpoT;2, Schuster et al. unpublished data; EMBL accession number Y09432) which, based on phylogenetic analyses, seems to be closely related to the ARAth;RpoT;1 (data not shown). The targeting properties of this RNAP are under investigation, and initial in vitro experiments suggest that it might be mitochondrially localized.

Further studies will have to demonstrate the enzymatic function of the different phage-type RNA polymerases within the transcriptional machinery of chloroplasts and mitochondria.

Experimental procedures

Acknowledgements

We thank Dr J. Haseloff for the kind gift of pBIN-mGFP5-ER. The excellent technical assistance of C. Stock, E. Kwidzinski and P. Klein is gratefully acknowledged. This work is supported by a grant from the Deutsche Forschungsgemeinschaft to T.B. and A.W.