Bot. Bull. Acad. Sin. (2003) 44: 123-128

Wang et al. An NLS in LLA23 for nuclear targeting

The nuclear localization signal of a pollen-specific, desiccation-associated protein of lily is necessary and sufficient for nuclear targeting

Huei-Jing Wang1, Guang Yuh Jauh2, Yau-Heiu Hsu1, and Co-Shine Wang1,*

1Graduate Institute of Biotechnology, National Chung Hsing University, Taichung, Taiwan 40227

2Institute of Botany, Academia Sinica, Taipei, Taiwan 11529

(Received August 22, 2002; Accepted November 26, 2002)

Abstract. The lily LLA23 protein represents a novel member of water-deficit/ripening-induced protein family (Plant Cell Physiol. 41: 477-485, 2000). Examination of the C-terminal half of LLA23 reveals the presence of basic regions that are reminiscent of a nuclear localization signal (NLS). To investigate the nuclear targeting property of NLS in LLA23, a green fluorescent protein (GFP) gene fused with the C-terminal half of LLA23 (GFP-LLA23) was constructed. In addition, the GFP alone and a mutGFP-DLLA23 that had the sequence for the putative NLS deleted were also constructed. All these three constructs were separately inserted into a bamboo mosaic potexvirus (BaMV) vector. Infection of BaMV in Chenopodium quinoa caused local lesions in leaves where the green fluorescence of fusion proteins could be visualized by fluorescence microscopy. The RNA blot and immunoblot analyses of BaMV-infected leaves indicated that the recombinant subgenomic RNA and the resulting protein were strongly detected. Fluorescence microscopic studies revealed that the NLS in LLA23 exhibited a property of nuclear targeting, showing highly condensed spots of green fluorescence in leaf cells, whereas GFP alone was apparently distributed throughout the cytoplasm. In contrast, a deletion of the NLS sequence resulted in exclusively cytoplasmic localization of the fusion protein. The nuclear location of the GFP-LLA23 protein in leaf cells was further confirmed by staining with 4,6-diamidino-2-phenylindole. These results clearly demonstrate that the putative NLS in LLA23 is necessary and sufficient for import of the LLA23 protein into the nucleus.

Keywords: BaMV; Desiccation; Lilium longiflorum; NLS; Pollen-specific protein.

Abbreviations: BaMV, bamboo mosaic potexvirus; DAPI, 4,6-diamidino-2-phenylindole; GFP, green fluorescent protein; Lea, late embryogenesis abundant; NLS, nuclear localization signal; PAGE, polyacrylamide gel electrophoresis; PBS, phosphate buffer saline; PCR, polymerase chain reaction.


Pollen plays a crucial role in sexual reproduction of flowering plants. Within the anther, pollen mother cells undergo meiosis to produce a tetrad of haploid microspores that subsequently devide mitotically to form mature pollen grains. When pollen grows to full maturity, it exhibits various degrees of desiccation prior to anthesis and this desiccation process represents the last stage of pollen maturation.

The quantitative and qualitative changes in protein accumulation during pollen development have been documented in various plant species (Zarsky et al., 1985; Vergne and Dumas, 1988; Detchepare et al., 1989; Bedinger and Edgerton, 1990; Zarsky et al., 1995), but there have been few reports on pollen-specific proteins that accumulate in the maturing pollen. We have previously described the ac

cumulation of a set of newly synthesized anther/pollen-specific proteins prior to anthesis during anther development in lily (Wang et al., 1992; Wang et al., 1996). Ueda and Tanaka (1995) have found two male gamete-specific histone variants in the generative nuclei of lily pollen. Recently, pollen-specific proteins were also described in pollen of tobacco and lily (Wittink et al., 2000; Ko et al., 2002; Mogami et al., 2002).

We have characterized a number of pollen-specific proteins related to dehydration in lily pollen (Wang et al., 1996, 1999; Ko et al., 2002). Of these, the LLA23 protein represents a novel member of the water-deficit/ripening-induced protein family (Wang et al., 1998; Huang et al., 2000). The water-deficit/ripening-induced proteins reported in various plant species possess a putative nuclear localization signal (NLS) at the C-terminus (Iusem et al., 1993; Canel et al., 1995; Silhavy et al., 1995; Chang et al., 1996; Schneider et al., 1997). Subcellular fractionation experiments performed by Iusem et al. (1993) indicated that the Asr protein is located primarily in the nucleus.

*Corresponding author. Tel: 886-4-287-4754 ext. 105; Fax: 886-4-286-1905; E-mail:

Botanical Bulletin of Academia Sinica, Vol. 44, 2003

Proteins are translocated from the cytoplasm where they are synthesized into the nucleus either to perform basic cellular processes or in response to developmental or environmental signals (Harter et al., 1994). It is generally accepted that even for small proteins, nuclear import is mediated by nuclear localization signals characterized as short clusters of basic amino acids (Garcia-Bustos et al., 1991; Raikhel, 1992). Mechanisms to control nuclear targeting involve regulation of NLS binding by protein interaction and modulation of NLS activity by biochemical modification (Fobes, 1992; Goldfarb, 1994; Hunter and Karin, 1992). The LLA23 protein contains a segment of NLS at the C terminus (Huang et al., 2000). In a continuing effort to elucidate the function of the LLA23 protein, here we examined whether this putative NLS indeed functions to direct nuclear targeting of LLA23.

Materials and Methods

Plant Materials and Inoculation

Plants of lily (Lilium longiflorum Thunb. cv Snow Queen) were grown in the field. Plants of Chenopodium quinoa were grown in the green house. Leaves of Chenopodium quinoa were used for virus inoculation as previously described (Lin and Hsu, 1994). Each inoculum contained a mixture of 0.25 g of bamboo mosaic potexvirus (BaMV) RNA per leaf.

RNA Blot

Total RNA was extracted from leaves of C. quinoa using the Ultraspec RNATM isolation system (Biotecx Laboratories Inc., Houston, USA). RNA samples were electrophoresed in 1.0% formaldehyde-MOPS gels using standard procedures (Sambrook et al., 1989) and transferred onto nylon membranes. The membranes with immobilized RNA were prehybridized for 4 h at 42C in medium containing 5 SSC (1 SSC is 0.15 M NaCl and 15 mM sodium citrate), 0.1% polyvinylpyrrolidone, 0.1% ficoll, 20 mM sodium phosphate, pH 6.5, 0.1% (w/v) SDS, 1% glycine, 50% formamide and 150 g/ml of denatured salmon sperm DNA. For hybridization, the prehybridization solution was removed and replaced with hybridization buffer that contained the same components as the prehybridization buffer except for the addition of 1% glycine, denatured salmon sperm DNA (100 g/ml) and random-primed 32P-labeled green fluorescent protein (GFP) DNA (specification 8.0 108 cpm/g). Hybridization was carried out at 42C overnight with constant agitation. The membranes were washed at 42C twice in 2 SSC, 0.1% (w/v) SDS for 20 min followed by twice in 0.1 SSC, 0.1% (w/v) SDS at 60C for 20 min. The membrane was exposed to X-ray films (Konica AX) using 1 or 2 intensifying screens (DuPont).

Generation of GFP-LLA23 Fusion Constructs and in Vitro Transcription

To generate a chimeric fusion construct of GFP and LLA23 (GFP-LLA23), a DNA fragment that encoded the C-terminal half of LLA23 (amino acid residues 55 to 142) was

first amplified by polymerase chain reaction (PCR) using LLA23 cDNA as a template using a 5-primer 1 (5-CGGCATGGACGAGCTGTACAAGGACTACGAGAAAG AGAAGAAGCAC-3) and 3-primer 3 (5-TGCCTTAT CGCCGGCGTCGCTT AACCGAAGAAGTGG-3) pair as shown in Figure 1A. In addition to containing a segment of LLA23 sequence at the N-terminus (underlined), the 5-primer 1 also contains a C-terminal coding sequence of the GFP gene (without the terminator codon). The resulting PCR fragment of LLA23 (designated megaprimer 1) was fractionated on a 1% agarose gel and stained with ethidium bromide. The fragment was recovered by the NucleoTrap DNA Purification Kit (Clontech Laboratories Inc., Palo Alto, CA). Next, the GFP gene was used as a template and amplified by PCR with a 5-primer of GFP (5-AGATATCATGGTGAGCAAGGGCGAGGAGC-3) and 3-megaprimer 1 pair as described above. The resulting GFP-LLA23 fusion fragment was purified on a 1% agarose gel, stained with ethidium bromide. The fragment was again recovered by the NucleoTrap DNA Purification Kit and cloned into pGEM-T easy vector (Promega, Madison, WI). In the construction of mutGFP-DLLA23, a segment of 20 amino acids from residues 123 to 142 was deleted. This deletion resulted in a GFP-DLLA23 mutant lacking the NLS. The construction of mutGFP-DLLA23 was similar to that of GFP-LLA23 except that the 3-primers 2 (5-TGCCTTATCGCCGGCGTCGCTTACTCGTGGTGCTCA TGGAAGGTGT-3) was used instead. The resulting 3-megaprimer 2 was then paired with 5-primer of GFP for PCR amplification using GFP as a template. The GFP alone was also amplified by PCR using a 5-primer and 3-primer (5-CCCGGGCGGCGCGTTTACTTGTACAGCTCGTCC-3) pair of GFP. After the nucleotide sequence of each construct was verified by DNA sequencing, the GFP, GFP-LLA23 and mutGFP-DLLA23 fragments were digested with Eco RV and Not I and cloned into the corresponding sites of pUC119 vector, which includes the T7 promoter and the whole genomic sequence of BaMV. Each plasmid construct was purified and its concentration was determined. Conditions for in vitro transcription of linearized plasmids were as described for brome mosaic virus (French and Ahlquist, 1987). The quantity and quality of the synthesized transcripts were verified by agarose gel electrophoresis before inoculation.

Protein Preparation, Electrophoresis and Immunoblotting

Phenol extraction of total protein was performed according to Wang et al. (1992). The infected leaves of C. quinoa were ground into a fine powder in liquid nitrogen with a motar and pestle. Protein concentration was determined by the dye binding Bio-Rad protein assay according to the suppliers instructions. Total protein was fractionated by SDS-polyacrylamide gel electrophoresis (PAGE) and either stained with Coomassie blue or electroblotted onto nitrocellulose (0.45 m, Gelman Sciences, Ann Arbor, MI) (Wang et al., 1996). Blots were immunostained using a 1:800 dilution of anti-GFP antiserum.

Wang et al. An NLS in LLA23 for nuclear targeting

Fluorescence Microscopy

The infected leaves of C. quinoa were directly inspected using a fluorescence microscope. The infected leaves were treated with 0.5% Triton X-100 for 30 min before they were stained with 2.5 g/ml 4,6-diamidino-2-phenylindole (DAPI) in phosphate buffer saline (PBS) for 30 min at room temperature, and then washed with PBS. Fluorescence microscopy and photography were performed using BX51TRF and PM30 (Olympus, Tokyo, Japan), respectively.


The protein encoded by LLA23 (accession number AF077629) belongs to a family of water-deficit/ripening-

induced proteins (Huang et al., 2000). It contains a segment of the sequence KKTLKKENEEVEGKK near the C-terminus of the molecule (Figure 1A). It shows a putative bipartite motif of basic residues, K or R, characteristic of NLS. In both plants and animals, proteins are targeted to the nucleus by specific NLS characterized as short amino acid regions that are rich in basic residues (Garcia-Bustos et al., 1991).

To investigate the function of the putative NLS in LLA23, a GFP gene was used as a reporter. Using the LLA23 cDNA as a template, megaprimers were amplified by PCR using a 5-primer 1 and either 3-primer 2 or 3-primer 3 (Figure 1, A and B). Subsequently, the GFP was used as a template and amplified by PCR using a 5-primer of GFP and either purified 3-megaprimer 1 or 3-megaprimer 2. The resulting constructs contained the GFP sequence fused with the C-terminal sequence of LLA23 (GFP-LLA23) and the GFP sequence with the C-terminal sequence of LLA23 without the NLS sequence (mutGFP-DLLA23), respectively (Figure 1C). The fusion constructs were inserted into a bamboo mosaic potexvirus (BaMV) vector and BaMV RNAs from each construct was prepared in vitro. Infection of BaMV in C. quinoa caused local lesions in leaves (Figure 2A), where the green fluorescence of fusion proteins could be visualized by fluorescence microscopy (Figure 3). RNA blot analysis of BaMV-infected leaves indicated that the virus subgenomic RNA (sgRNAx), which contained the RNA transcribed from the fusion construct, strongly hybridized to the 32P-labeled GFP DNA probe. The other two RNA bands, virus genomic RNA (gRNA) and subgenomic RNA1 (sgRNA1), also hybridized to the GFP probe (Figure 2B). Total protein was extracted from BaMV-infected leaves of C. quinoa and fractionated by SDS-PAGE. Immunoblot analysis using GFP-specific antiserum suggested that expression of these constructs in infected leaves resulted in the synthesis of fusion proteins of predicted sizes (Figure 2C). The mock contains only BaMV itself, and thus no GFP protein could be possibly detected.

Fluorescence microscopic studies indicated that the NLS in LLA23 exhibited a property of nuclear localization signals, showing highly condensed spots of green fluorescence in leaf cells (Figure 3B), whereas GFP alone was apparently distributed in the cytoplasm of whole leaf cells (Figure 3A). In contrast, deletion of the NLS sequence (mutGFP-DLLA23) resulted in exclusively cytoplasmic localization of the fusion protein (Figure 3C). Further, the nuclear location of GFP-LLA23 in leaf cells was confirmed by DAPI staining (Figure 3, D and E). These results clearly demonstrate that the NLS in LLA23 is necessary and sufficient for nuclear localization.


The LLA23 protein of lily is a novel member of the water-deficit/ripening-induced protein family. The protein is unique in that it is pollen-specific (Wang et al., 1996), whereas the other members of the family are found in rip

Figure 1. Generation of GFP-LLA23 and mutGFP-DLLA23 constructs. A, A partial coding sequence of LLA23 cDNA and its predicted amino acid sequence. The shaded region indicates a segment of the putative NLS sequence. Primers 1-3 are indicated by long arrows: 1 represents the LLA23 portion of the 3-GFP-5-LLA23 primer and 2 and 3 represent 3-DLLA23 (without an NLS) and 3-LLA23 primers, respectively; B, Synthesis of megaprimers. DNA fragments that encode the C-terminal half of LLA23 and DLLA23 were separately amplified by PCR. The resulting PCR fragments, designated megaprimer 1 and 2, were fractionated on a 1% agarose gel and stained with ethidium bromide; C, Generation of fusion constructs. The GFP was used as a template and amplified by PCR using a 5-primer of GFP and either 3-megaprimer 1 or 3-megaprimer 2. The resulting GFP-LLA23 and GFP-DLLA23 fusion fragments were purified on a 1% agarose gel and stained with ethidium bromide. In B and C, M indicates the lane containing marker DNA fragment of various sizes.

Botanical Bulletin of Academia Sinica, Vol. 44, 2003

ening fruits and vegetative organs of various plant species (Iusem et al., 1993; Silhavy et al., 1995; Chang et al., 1996; Schneider et al., 1997). The LLA23 protein is developmentally regulated and accumulates immediately before anthesis (Wang et al., 1998). Changes in protein accumulation at the stage prior to anthesis during anther development have been described in various plant species (Mandaron et al., 1990; Michel et al., 1994; Wang and Cutler, 1995). However, no protein has been shown to be related to desiccation. We have demonstrated that the accumulation of both LLA23 RNA and protein responds to desiccation and many other environmental stresses (Wang et al., 1998; Huang et al., 2000).

There is no single and strict consensus NLS, but there are some general rules for NLSs (Garcia-Bustos et al.,

Figure 3. Requirement of the NLS in LLA23 for nuclear localization. Eight days after infection, leaves of C. quinoa were inspected by fluorescence microscopy. A-C, Images of subcellular localization of GFP, GFP-LLA23 and mutGFP-DLLA23 fusions, respectively; D-E, An infected leaf was first detergent-treated and then stained with DAPI. The locations of GFP-LLA23 (D) were exactly the same as those stained with DAPI (E). Arrows indicate the corresponding guard cells. Arrowheads indicate GFP in the nuclei. Scale bar represents 20 m.

1991). These include: (1) there are typically short sequences; (2) they contain a high proportion of positively charged amino acids; (3) they are not removed following nuclear localization. Previous analysis of LLA23 identified a putative NLS near the C-terminus of the protein (Figure 1A), and this finding prompts us to test whether this NLS indeed functions in nuclear targeting.

We generated a construct of GFP-LLA23 along with a mutGFP-DLLA23 which did not contain the NLS sequence of LLA23. GFP-LLA23 was localized predominantly to the nuclei of leaf cells (Figure 3B) whereas the mutGFP-DLLA23 resulted in exclusively cytoplasmic distribution of the fusion protein in leaf cells (Figure 3C). These results clearly demonstrate that the NLS in LLA23 is necessary and sufficient for nuclear localization. Similar results were also observed by confocal scanning laser microscopy (data not shown). To identify the critical basic amino acids for nuclear localization of LLA23, we have generated several mutant constructs by replacing alanine residues with lysines at the NLS in LLA23. The work of mutagenesis on NLS is in progress.

The presence of more than one NLS in nuclear proteins is apparently very frequent (Garcia-Bustos et al., 1991; Raikhel, 1992). However, a single NLS in the LLA23 protein is sufficient to direct GFP to the nucleus. It is striking that the overall bipartite NLS structure is conserved between monocots and dicots (Huang et al., 2000). Together with the reported presence of putative bipartite NLSs in a high proportion of plant b-ZIP proteins (Raikhel, 1992; Varagona et al., 1992), these observations suggest that the bipartite structure may be the most prevalent NLS configuration in plants, across a spectrum of divergent nuclear proteins.

Figure 2. Biochemical analyses of leaves of C. quinoa infected by various BaMV constructs. A, The healthy leaves (left) of C. quinoa of 4-week-old plants and symptoms shown on 8-day BaMV-infected leaves (right); B, Detection of BaMV fusion RNA in infected leaves. The left-hand side shows a map of BaMV genomic RNA (gRNA) and subgenomic (sg) RNA1, RNA2 and RNAx. X indicates the GFP or fusion constructs. The BaMV gRNA encodes a number of viral proteins with molecular masses of 155, 28, 13 and 6 kDa and coat protein (CP). Total RNA was isolated from infected leaves by BaMV. RNA samples (3 g) were denatured, fractionated on a formaldehyde-agarose gel, transferred to a nylon membrane (top), and hybridized to a 32P-labeled GFP DNA. Nearly equal amounts of total RNA were loaded in each lane, as determined by ethidium bromide staining of the gel (bottom); C, Immunoblot detection of various GFP-LLA23 fusion proteins in BaMV-infected leaves. Equal amounts (3 g) of leaf extracts prepared from infected leaves were electrophoresed by SDS-PAGE and either stained with Coomassie blue (left) or electroblotted onto nitrocellulose and immunologically detected using anti-GFP antiserum (right). M indicates molecular mass marker proteins (97, 66, 45, 30, and 20 kDa).

Wang et al. An NLS in LLA23 for nuclear targeting

Huang, J.-C., S.-M. Lin, and C.-S. Wang. 2000. A pollen-specific and desiccation-associated transcript in Lilium longiflorum during development and stress. Plant Cell Physiol. 41: 477-485.

Hunter, T. and M. Karin. 1992. The regulation of transcription by phosphorylation. Cell 70: 375-387.

Iusem, N.D., D.M. Bartholomew, W.D. Hitz, and P.A. Scolnik. 1993. Tomato (Lycopersicon esculentum) transcript induced by water deficit and ripening. Plant Physiol. 102: 1353-1354.

Ko, C.-W., C.-Y. Yang, and C.-S. Wang. 2002. A desiccation-induced transcript in lily (Lilium longiflorum) pollen. J. Plant Physiol. 159: 765-772.

Lin, N.S. and Y.H. Hsu. 1994. A satellite RNA associated with bamboo mosaic potexvirus. Virol. 202: 707-714.

Mandaron, P., M.F. Niogret, R. Mache, and F. Monger. 1990. In vitro protein synthesis in isolated microspores of Zea mays at several stages of development. Theor. Appl. Genet. 80: 134-138.

Michel, D., A. Furini, F. Salamini, and D. Bartel. 1994. Structure and regulation of an ABA- and desiccation-responsive gene from the resurrection plant Craterostigma plantagineum. Plant Mol. Biol. 24: 549-560.

Mogami, N., H. Shiota, and I. Tanaka. 2002. The identification of a pollen-specific LEA-like protein in Lilium longiflorum. Plant Cell Environ. 25: 653-663.

Raikhel, N. 1992. Nuclear targeting in plants. Plant Physiol. 100: 1627-1632.

Sambrook, J., E.F. Fritsch, and T. Maniatis (eds.). 1989. Molecular Cloning: A Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Schneider, A., F. Salamini, and C. Gebhardt. 1997. Expression patterns and promoter activity of the cold-regulated gene ci21A of potato. Plant Physiol. 113: 335-345.

Silhavy, D., G. Hutvgner, E. Barta, and Z. Bnfalvi. 1995. Isolation and characterization of a water stress-inducible cDNA clone from Solanum chacoense. Plant Mol. Biol. 27: 587-595.

Ueda, K. and I. Tanaka. 1995. Male gametic nucleus-specific H2B and H3 histones, designated gH2B and gH3 in Lilium longiflorum. Planta 197: 289-295.

Varagona, M.J., R.J. Schmidt, and N.V. Raikhel. 1992. Nuclear localization signal(s) required for nuclear targeting of the maize regulatory protein Opaque-2. Plant Cell 4: 1213-1227.

Vergne, P. and C. Dumas. 1988. Isolation of viable wheat male gametophytes of different stages of development and variations in their protein patterns. Plant Physiol. 88: 969-972.

Wang, H. and A.J. Cutler. 1995. Promoters from kin1 and cor6.6, two Arabidopsis thaliana low-temperature- and ABA-inducible genes, direct strong beta-glucuronidase expression in guard cells, pollen, and young developing seeds. Plant Mol. Biol. 28: 619-634.

Wang, C.-S., L.L. Walling, K.J. Eckard, and E.M. Lord. 1992. Patterns of protein accumulation in developing anthers of Lilium longiflorum correlate with histological events. Amer. J. Bot. 79: 118-127.

Wang, C.-S., T.-D. Wu, C.-K.W. Chung, and E.M. Lord. 1996. Two classes of pollen-specific, heat-stable proteins in Lilium longiflorum. Physiol. Plant. 97: 643-650.

The physiological function of LLA23 is yet unknown. Because it is abundant and hydrophilic, two characteristics similar to dehydrins (Close et al., 1993), we suggest that the LLA23 protein in pollen grains may play a role similar to late embryogenesis abundant (lea) proteins in seeds. The lea gene products are supposed to protect cellular structures from the deleterious effect of water loss. While lea proteins protect the cytoplasm from desiccation, LLA23 might do the same for the DNA in the nucleus. The Asr protein was reported to be in the nucleus (Iusem et al., 1993). Silhavy et al. (1995) also suggested that the function of the DS2 may be the protection of the nuclear DNA from desiccation (Silhavy et al., 1995). Using BaMV as a viral vector, we have clearly demonstrated that the protein contains a functional NLS and this NLS is necessary and sufficient for the import of the protein into the nucleus.

Acknowledgments. We wish to thank Chin-Wei Lee for growing plants of C. quinoa in the green house. We also thank Chin-Ying Yang for careful compilation of figures. This work was supported by National Science Council of the Republic of China, under a grant NSC89-2311-B-005-019 to Co-Shine Wang.

Literature Cited

Bedinger, P.A. and M.D. Edgerton. 1990. Developmental staging of maize microspores reveals a transition in developing microspore proteins. Plant Physiol. 92: 474-479.

Canel, C., J.N. Bailey-Serres, and M.L. Roose. 1995. Pummelo fruit transcript homologous to ripening-induced genes. Plant Physiol. 108: 1323-1324.

Chang, S., J.D. Puryear, M.A.D.L. Dias, E.A. Funkhouser, R.J. Newton, and J. Cairney. 1996. Gene expression under water deficit in loblolly pine (Pinus taeda): Isolation and characterization of cDNA clones. Physiol. Plant. 97: 139-148.

Close, T.J., R.D. Fenton, A. Yang, R. Asghar, D.A. Demason, D.E. Crone, N.C. Meyer, and F. Moonan. 1993. Dehydrin: the protein. In T.J. Close and E.A. Bray (eds.), Plant Responses to Cellular Dehydration during Environmental Stress. Amercian Society of Plant Physiologists, Rockville, pp. 104-118.

Detchepare, S., P. Heizmann, and C. Dumas. 1989. Changes in protein patterns and protein synthesis during anther development in Brassica oleracea. J. Plant Physiol. 135: 129-137.

Fobes, D.J. 1992. Structure and function of the nuclear pore complex. Ann. Rev. Cell Biol. 8: 495-527.

French, R. and P. Ahlquist. 1987. Intercistronic as well as terminal sequences are required for efficient amplification of brome mosaic virus RNA3. J. Virol. 61: 1457-1465.

Garcia-Bustos, J., J. Heitman, and M.N. Hall. 1991. Nuclear protein localization. Biochim. Biophys. Acta 1071: 83-101.

Goldfarb, D.S. 1994. GTPase cycle for nuclear transport. Curr. Biol. 4: 57-60.

Harter, K., S. Kircher, H. Frohnmeyer, M. Krenz, F. Nagy, and E. Schfer. 1994. Light-regulated modification and nuclear translocation of cytosolic G-box binding factors in parsley. Plant Cell 6: 545-559.

Botanical Bulletin of Academia Sinica, Vol. 44, 2003

ciated with the vegetative membranes and the cell wall. Sex. Plant Repro. 12: 276-284.

Zarsky, V., V. Capkova, E. Hrabetova, and J. Tupy. 1985. Protein changes during pollen development in Nicotiana tabacum L. Biol. Plant. 2: 438-444.

Zarsky, V., D. Garrido, N. Eller, J. Tupy, O. Vicente, F. Schffl, and E. Heberle-Bors. 1995. The expression of a small heat shock gene is activated during induction of tobacco pollen embryogenesis by starvation. Plant Cell Environ. 18: 139-147.

Wang, C.-S., Y.-E. Liau, T.-D. Wu, C.-C. Su, J.-C. Huang, and C.H. Lin. 1998. Characterization of a desiccation-related protein in lily pollen during development and stress. Plant Cell Physiol. 39: 1307-1314.

Wang, C.-S., S.-M. Lin, and S.-L. Wei. 1999. A stress-inducible protein associated with desiccation in lily pollen. Bot. Bull. Acad. Sin. 40: 199-205.

Wittink, F.R.A., B. Knuiman, J. Derksen, V. Capkov, D. Twell, J.A.M. Schrauwen, and G.J. Wullems. 2000. The pollen-specific gene Ntp303 encodes a 69-kDa glycoprotein asso