F. Report of the Committee on Genetic Engineering

This report reviews the work on rice gene analysis, published within the last 12 months or so, under three headings: cloning and sequence analysis of rice genes; characterization and expression of rice genes; and analysis of genomic DNA in rice.

Germination of rice seeds and seedling growth are dependent on the enzy- matic hydrolysis of starch into sugars. alpha-amylase is primarily responsible for the cleavage of amylose and amylopectin into glucose. Huang et al. (1990) have isolated 10 genes that fall into 5 hybridization groups. The nucleotide sequence of one clone from Group 1, lambda OSg2, was determined. Comparison of the sequence with that of other alpha-amylase sequences revealed that lambda OSg2 is the genomic analog of the rice cDNA clone, pOS1O3. lambda OSg2 contains sequence motifs common to most actively transcribed plant genes. Two consensus sequences, TAACAA/GA and TATCCAT, were found in the 5' flanking regions of alpha-amylase genes of rice, barley and wheat. The former sequence may be specific to alpha-amylase gene whereas the latter sequence may be related to a "CATC" box found in many plant genes. Another sequence called the pyrimidine box, C/TCTTTTC/T, was found in the alpha-amylase genes as well as other genes regulated by the plant hormone, gibberellic acid.

About 5% of the rice seed storage proteins belong to the prolamin family. Prolamin genes encode 16 kDa, 13 kDa and 10 kDa proteins. The sequences of one 1O kDa cDNA (Masumura et al., 1989), three 16 kDa cDNAs and one genomic clone of 16 kDa prolamin from rice (Kim and Okita, 1988) have been reported. The cDNA encoding a 13 kDa prolamin has been cloned and sequenced (Masumura et al., 1990). A polypeptide sequence composed of 156 amino acids, including a signal sequence of 19 amino acids, was deduced from the nucleotide sequence of the 13 kDa prolamin gene. No repetitive sequences or sequences homologous to other cereal prolamins, except an octapeptide (Gln-Gln-Gln-Cys-Cys-Gln-Gln-Leu), were found for the mature 13 kDa prolamin polypeptide.

In order to investigate the extent of genetic variation within the wild rice species Oryza rufipogon, Barbier and Ishihama (1990) used the polymerase chain reaction (PCR) to amplify the 10 kDa prolamin gene from total rice DNA and sequenced the PCR product directly. Eight Asian strains of Oryza rufipogon and one strain of the related African species O. longistaminata were chosen. The results showed that six strains of of O. rufipogon gave identical sequence with the cultivar Nihonbare. The other two strains O. rufipogon differed by three point mutations. The African strain gave four point mutations and two codon additions.

The enzyme phenylalanine amino-lyase (PAL) catalyzes the first reaction in the biosynthetic pathway of polyphenolic compounds including flavonoids, cinnamate esters and lignin. The PAL gene is well characterized in parsley. The activity of PAL increases in response to various stimuli such as infection, wounding and light. As a first step towards the elucidation of the mechanisms for the regulation of PAL gene expression in rice, Minami et al. (1989) isolated a nearly full-length cDNA and a corresponding genomic DNA for PAL. The gene encodes a protein of 701 amino acids. The cloned gene spans 4412 bp and consists of two exons and one intron. Sequences similar to a TATA-box and GC-box were found in the 5'-upstream region. Genomic blot analysis showed that the PAL gene of rice exists as a small gene family. The genomic clone isolated in this study was shown to be responsive to light.

The reduction of nitrate to nitrite is catalyzed by nitrate reductase (NR). This reaction is the initial step in nitrate assimilation and is considered rate limiting. Rice plants grow well in media containing nitrate as a sole nitrogen source. Hamat et al. (1989) cloned and partially characterized a genomic NR gene. Genomic and dot blot analyses suggested that NR is encoded by a small gene family in rice. Regulation of NR was investigated. In the absence of nitrate, only trace levels of NR activity and NR mRNA were detected in the leaves of rice seedlings. Upon addition of nitrate to seedling roots, NR activity and mRNA increased rapidly in leaves. NR activity continued to increase over a 24 hour period, but the NR mRNA accumulation peaked at about 6 hours and then declined. Protein blot analysis with NR antiserum showed the presence of two bands of around 115 and 105 kDa.

Upon transfer of rice to an anaerobic environment ethanol is produced, which is paralleled by an increase in alcohol dehydrooeiiase (ADH) activity. The sequence of two rice ADH cDNA clones was reported last year (Xie and Wu, 1989). Now, a genomic Adh2 sequence has been isolatedand partially characterized (Xie and Wu, 1990). By comparison with the maize Adh2, the 3.7 kb rice Adh2 sequence contains an open reading frame which is interrupted by nine introns, showing the expected intron/exon junction and putative spicing branch site sequences. Sequences that are important for gene expression, such as a TATA box and polyadenylation signal, are found in the expected locations. A sequence related to the hexanucleotide core of regulatory element, GAGTGG, reported for the maize Adh2 is also found in rice Adh2, 30 nucleotides upstream from the start codon. This sequence may be important in the regulation of ADH gene expression under anaerobic conditions.

Several rice mitochondrial genes have been cloned and sequenced. Kaleikau et al. (1990a) sequenced the apocytochrome b gene (cob). The DNA sequence was 98.9% homologous to the maize cob and 99.2% to the wheat cob. The molecular weight of the deduced rice apocytochrome b protein is 44.49 kDa.

The cytochrome oxidase subunit III gene (cox3) from rice mitochondrial DNA has been sequenced (Kaleikau et al., 1990b). The rice cox3 was 99.0% homologous to that of maize.

The F/0-atpase proteolipid gene (atp9) from rice has been sequenced (Kaleikau et al., 1990c). The rice atp9 shares 95.6% sequence identity with that of maize and 96.4% identity with that of wheat. The molecular weight of the rice F/0-atpase proteolipid is predicted to be 8.95 kDa.

The F/1-ATPase alpha subunit gene (atpa) from rice mitochondria has been sequenced (Kadowaki et al., 1990). The gene encodes a continuous open reading frame of 1530 nucleotides. The coding region of the gene exhibits 97.5% nucleotide sequence identity with the maize atpa of mitochondria.

Cytoplasmic actin is an essential component of the plant cell cytoskeleton. Cytoplasmic streaming, organelle movement, extension growth and cell division are all believed to involve cytoskeletal actin proteins. As a first step towards characterization of the actin genes in rice, Reece et al. (t990) cloned and sequenced four genomic actin genes, Act1, Act2, Act3 and Act7. A sequence alignment between a 1.5 kb Act1 cDNA clone (McElroy et al., 1990b) and the four genomic clones was made. This analysis identified three introns located in the same position in all four rice actin coding regions, and the presence of a 5'-noncoding exon, separated by an intron, from the first translated exon of the Act]. This is one of the few reported cases of a plant gene containing such a 5'-noncoding exon. Analysis of other rice cDNA clones showed that the rice actin gene family is composed of at least eight unique members. DNA sequence comparison of the coding regions showed that the rice actin genes are highly diverged from each other.

The expression of the four actin genes was determined by RNA slot-blot hybridization analysis using gene specific probes (McElroy et al., 1990a). Results show that the transcript levels of Act2 (and Act3) in rice shoot tissues are the highest at 2 and 4 days. These levels decrease steadily to 50% at 7 days and to 20% at 35 days. The mRNA level in 2-day-old shoots is twice as high as that in 2-day-old roots. In contrast, the mRNA levels in Act1 and Act7 in shoots are relatively constant between 2 and 35 days. Additionally, more Act7 mRNA is present in 2- and 4-day-old roots than that in shoots. Next, the relative levels of the four actin gene transcripts were determined. Act1 transcripts are the most abundant; the mRNA levels of Act2, Act3 and Act7 are 40%, 15% and 15%, respectively, of that of Act1 in 2-day-old shoots. These observations suggest that the individual rice actin genes may differ in their transcription regulation.

The expression of the rice actin 1 gene (Act1) has been studied in transient assays of transformed rice protoplasts (McElroy et al., 1990c), By constructing plasmids with 5' regions from the rice Act1 fused to the coding sequence of a gene encoding bacterial beta-glucuronidase (GUS), it was shown that a region 1.3 kb upstream of the Actl translation initiation codon contains all the 5'-regulatory sequences necessary for high level GUS expression in transformed rice protoplasts. Fusion of Act1 lacking the intron in the 5' noncoding region to the GUS gene gave no detectable GUS activity in the transient assay. Deletion analysis of the Act1 5' intron suggests that the intron-mediated stimulation of Gus expression is associated with an in vivo requirement for efficient intron splicing. The plasmid that includes the rice Act1 sequence fused to Gus gave seven times higher GUS activity compared to the plasmid that includes the maize Adh1 sequence fused to Gus. Thus, the rice Act1-Gus plasmid is the best one so far for use in rice transformation studies.

The light-dependent development of plants is mediated through photoreceptors. The best characterized of these is the red light photoreceptor, phytochrome. A rice phytochrome gene has been cloned and its expression tested in transgenic tobacco plants (Kay et al., 1989). A full-length rice phytochrome cDNA was fused to the cauliflower mosaic virus 35S promoter and transferred to tobacco. The progeny of several transgenic plants contain large amounts of rice phytochrome mRNA in their green leaves. Extracts prepared from these plants contain twofold to fivefold more phytochrome than extracts from control plants. Species- specific, anti-phytochrome antibodies were used in immunoblots to discriminate between rice and tobacco phytochrome apoproteins in fractions eluted from a DEAE-Seharose column. Partially purified rice phytochrome assembles with chromophore and is photoreversible. Thus, the rice phytochrome is biologically active in transgenic tobacco plants.

The hormone abscisic acid (ABA) is believed to mediate physiological processes in response to osmotic stress. Levels of endogenous ABA increases in tissues subjected to stress such as salt, desiccation or cold. Under these conditions, specific genes are expressed that also can be induced in control tissues by the application of ABA. In order to elucidate how ABA regulates gene expression, Mundy and Chua (1988) have previously characterized an ABA-responsive rice gene, rab-21 (now called rab-16), that is induced by ABA and osmotic stress in vegetative tissues. Later on, the same group (Yamaguchi-Shinozaki et al., 1989) cloned and sequenced four members of the rice rab (rab-16A-D) gene family. Each gene contains a small intron. They are all transcriptionally active and encode proteins of 15500-16800 Dd. Here, Mundy et al. (1990) show that transcriptional elements between -294 and -52 of rab-16A are sufficient to confer ABA-dependent expression of a reporter gene in rice protoplasts. Gel tetardation and DNAse I experiments show nuclear factor(s) binding to these sequences. The results suggest that certain motifs within this 242 bp region may be ABA-responsive elements.

4. Characterization of a rice gene showing organ-specific expression in response to salt stress and drought

Numerous metabolic changes occur in different salt-sensitive plants subjected to ionic stress. In order to understand better the molecular events occurring during the early periods of salt stress, Claes et al. (1990) found eight proteins to be induced. Partial amino acid sequences of one protein of 15 kDa, and an isoelectric point of 5.5, were determined. Based on the amino acid information, an oligo-nucleotide probe was synthesized. Using this probe, a cDNA clone, SalT, was obtained and found to contain an open reading frame coding for a protein of 145 amino acid residues. Using the cDNA as a probe, mRNA hybridization analysis was carried out. The SalT mRNA was found to accumulate very rapidly in sheaths and roots from mature plants and seedlings upon treatment with sodium chloride (15), potassium chloride (1 %), air drying, ABA (20 uM) and polyethylene glycol (5%). Generally, no induction was observed in the leaf lamina even when the stress should affect all parts of the plant uniformly. The organ-specific response of SalT is correlated with the pattern of Na+ accumulation during salt stress.

germination in rice Histones are needed during DNA replication, thus mRNAs coding for these proteins are xpected tob e highly expressed in rapidlt dividing cells. However, indirect and in situ hybridization evidence suggest that histone 3 mRNA is also present in nondividing cells. Raghavan and Olmedilla (1989) monitored the pattern of distribution of histone H3 mRNA during the development of the rice grain, and its germination, by in situ hybridization and RNA blot analysis. In ovaries sampled before and after fertilization, a tritium-labeled histone anti-sense RNA probe was localized in the cells of the pericarp, outer integument and nucleus but binding of the probe decreased as these tissues senesced. In the developing embryo, the histone mRNA was first detected in the scutellum; later on, all parts of the embryo, except the shoot apex, newly formed leaf primordia and the quiescent center of the root, revealed the presence of transcripts. Cells of embryo and endosperm of mature grains displayed very little or no histone mRNA, although during germination these transcripts appeared in the cells of both embryo and endosperm. There was good correlation between in situ hybridization analysis and RNA blot analysis.

Glutelin is the major storage protein of rice endosperm, comprising up to 80% of the rice seed protein. Glutelins are encoded by a multigene family that contains at least three distinct subfamilies. Genomic clones representing each of these three subfamilies (designated Gt1, Gt2 and Gt3) have been isolated. Leisy et al. (1989) studied the expression of Gt3 promoter in transgenic tobacco. A chimeric gene consisting of the 5' flanking sequences of Gt3 linked to a chloramphenicol acetyl transferase (CAT) coding region was introduced into tobacco. CAT enzyme activity could be detected in extracts from transgenic seeds as early as 8 DA (day after anthesis) and reached a maximum level at 16 DA. A positive correlation was observed between expression levels in seeds and gene copy numbers.

By employing in situ hybridization technique, Ramachandran and Raghavan (1990) used RNA probes to monitor the spatial and temporal expression of the glutelin gene during the development of rice grains. Annealing sections of the grain with ³⁵S-labeled antisense transcript showed that gene expression is switched on between 4 DA and 6 DA. A steady increase in the concentration of glutelin mRNA was observed in the pericarp and endosperm cells from 6 DA to 18 DA. This was followed by a progressive reduction and complete disappearance of transcript by 24 DA. Glutelin mRNAs were maximally present in the embryo about 8 DA. RNA blot and dot blot hybridization of ³²P-labeled glutelin gene to rice grain RNA confirmed the temporal pattern of gene expression revealed by in situ hybridization.

Regulated gene expression of chimeric genes has been studied extensively in electroporated protoplasts. The applicability of these assays is limited because protoplasts are not always physiologically identical to the cells from which they are derived. DeKeyser et al. (1990) have developed a procedure to electroporate DNA into intact and organized leaf structures of rice. Using a GUS gene under the control of constitutive promoters, all cell types within a leaf base were found to be susceptible to electroporation-mediated DNA uptake. Transient gene expression assays with electroporated leaf bases showed that the promoter from a pea light-harvesting chlorophyll a/b-binding protein gene displayed both light- and chloroplast-dependent expression in rice.

8. Characterization and expression of a light-inducible shoot-specific rice gene

De Pater et al. (1990) characterized a rice gene that is highly expressed only in shoots. By differential screening of a cDNA library of two week old rice seedlings, a shoot specific gene, COS5, was identified. The mRNA corresponding to this gene displayed an expression pattern similar to that of rbcS genes. The mRNA (800 bases) was light inducible and encoded by a single gene. Next, the genomic clone (GOS5) was isolated. The gene contains two introns. The transcriptional start site was determined. In the 5' noncoding region, motifs are found that are homologous to sequences in leaf-specific promoters that are light- or UV-inducible. The open reading frame codes for a 15 kDa protein that contains a putative chloroplast transit peptide which may be involved in transport of the protein into chloroplasts.

The development of restriction fragment length polymorphism (RFLP) technology has opened the door to detecting, monitoring, and manipulating genetic variation in plants in a manner not previously possible. Wang and Tanksley (1989) analyzed 70 varieties of rice, representing the breadth of the species Oryza sativa, by using 10 rice RFLP markers. Polymorphism was detected for all probes, and 58 of 70 varieties tested could be uniquely distinguished from one another by combining all probe-enzyme combinations. Within-population variation, usually in the form of homozygous variant alleles, was found for 26% of the rice varieties. Based on genetic distance calculations, the ratio of the genetic variation between versus within rice varieties was estimated to be around 12 to 1. An RFLP based dendrogram was constructed depicting genetic distances among these rice varieties.

Higher eucaryote ribosomal genes are organized in tandem repeats. In plants, their copy number can vary from a few hundred (as in Arabidopsis) to several thousands (in most species). Each repeat contains sequences coding for mature rRNAs and an intergenic region, the external large spacer which contains the signals for transcription initiation and termination. RFLP of the rDNA spacer was studied in the genus Oryza using a cloned rice rDNA probo (Cordesse et al., 1990). Fifty-eight cultivated rice and 47 wild species including various genome types were analyzed. The length of the rDNA spacer in the Asiatic cultivated rice of the species O. sativa was grouped into seven size classes differing from one another by an "increment" of about 300 bp. A general tendency from a smaller spacer in the Japonica subtypes to longer ones in the Indica was observed. In contrast, African rice of the species O. glaberrima did not display any rDNA size variation. Extensive size variation was observed in wild species, but the fragment sizes did not fall into regularly increasing size classes except for O. rufipogon and O. longistaminata. The variation was greater in these species than in the cultivated ones.

The role of 5-methyl cytosine (5 mC) has been well examined in animals as compared to plants. The presence of 5 mC is correlated with low gene expression. In plants, methylation patterns of rDNA has been thoroughly analyzed. In general, the ribosomal genes seem to be heavily methylated in several plant species so far investigated, except the rice rDNA units. Recently, it has been shown in onion, broad bean and rice that repeat DNA sequences undergo a change in their degree of methylation during differentiation and dedifferentiation events under cell culture conditions. In the present work, Dhar et al. (1990) analyzed the total, as well as repetitive DNA fractions, of rice cv. Basmati 370 using methylation-specific enzymes. The results showed an abundance of adenine methylation in both total rice DNA and a specific repeat sequence. Although cytosine methylation was also observed, the cytosine methylation in the sequence 5'-GATC-3' was less than adenine methylation. Moreover, the presence of adenine methylation was tissue specific; it was predominant in rice shoot DNA as compared to embryo DNA. There was also a qualitative change in 5 mC from CpG to CpC dinucleotides in these two tissue systems.

In plant DNA, 5 mC comprises up to 30% of the total cytosine residues, and occurs not only in CpG but also in CpNpG trinucleotides. The 5 mC in plant DNA is also considered to affect gene expression. In this study, Sano et al. (1990) showed that a single exposure of germinating rice seeds to either of the DNA demethylating agents, 5-azacytidine (azac) or 5-azadeoxycytidine (azadc) induced dwarf plants. Seeds treated with azac gave mature plants with normal morphological characteristics but with 15% height reduction as compared to untreated controls. The M1 progeny, obtained by self-fertilization of an azac- induced dwarf plant, segregated into dwarf (35%) and tall types (65%). The M2 progenies, obtained by selfing of dwarf M1 plants, were also dwarf. Genomic DNA isolated from mature leaves of azac-treated seeds showed a 16% reduction in 5 mC content in comparison with DNA from untreated samples. Thus, both undermethylation and dwarfism induced by azac treatment were heritable.

Barbier, P. and A. Ishihama, 1990. Variation in the nucleotide sequence of a prolamin gene family in wild rice. Plant Mol. Biol. 15: 191-195.

Claes, B., R. DeKeyser, R. Villarroel, M. Van den Bulche, G. Bauw, M. Van Montagu and A. Caplan, 1990. Characterization of a rice gene showing organic specific expression in response to salt stress and drought. The Plant Cell 2: 19-27.

Cordesse, F., G. Second and M. Delseny, 1990. Ribosomal gene spacer length variability in cultivated and wild rice species. Theor. Appl. Genet. 79: 81-88.

De Keyser, R. A., B. Claes, R. M. U. De Rycke, M. E. Habets, M. C. Van Montagu and A. B. Caplan, 1990. Transient gene expression in intact and organized rice tissues. The Plant Cell 2: 591-602.

De Pater, S., L. A. M. Hensgens and R. A. Schilperoort, 1990. Structure and expression of a light- inducible shoot-specific rice gene. Plant Mol. Biol. 15: 399-406.

Dhar, M. S., V. V. Pethe, V. S. Gupta and P. K. Ranjekar, 1990. Predominance and tissue specificity of adenine methylation in rice. Theor. Appl. Genet., in press.

Hamat, H. B., A. Kleinhofs and R. L. Warner, 1989. Nitrate reductase induction and molecular characterization in rice. Mol. Gen. Genet. 218: 93-98.

Huang, N., T. D. Sutliff, J. C. Litts and R. L. Rodriguez, 1990. Classification and characterization of the rice alpha-amylase multigene family. Plant Mol. Biol. 14: 655-688.

Kadowaki, K.I., S. Kazama and T. Suzuki, 1990. Nucleotide sequence of the F1-ATPase alpha sub- unit gene from rice mitochondria. Nucleic Acids Res. 18: 1302.

Kaleikau, E. K., C. P. Andre, B. Doshi and V. Walbot, 1990a. Sequence of the rice mitochondrial gene for apocytochrome b. Nucleic Acids Res. 18: 372.

Kaleikau, E. K., C. P. Andre and V. Walbot, 1990b. Sequence of the rice mitochondrial gene for cytochrome oxidase subunit 3. Nucleic Acids Res. 18: 371.

Kaleikau, E. K., C. P. Andre and V. Walbot, 1990c. Sequence of the F/0-atpase proteolipid (atp9) gene from rice mitochondria. Nucleic Acids Res. 18: 370.

Kay, S. A., A. Nagatani, B. Keith, M. Deak, M. Faruya and N. H. Chua, 1989. Rice phytochrome is biologically active in transgenic tobacco. The Plant Cell 1: 775-782.

Kim, W. T. and T. W. Okita, 1988. Structure, expression and heterogeneity of the rice seed prolamines. Plant Physiol. 88: 649-655.

Leisy, D. J., J. Hnilo, Y. Zhao and T. W. Okita, 1989. Expression of a rice glutelin promoter in transgenic tobacco. Plant Mol. Biol. 14: 41-50.

Masumura, T., D. Shibata, T. Hibino, T. Kato, K. Kawaba, G. Takeba, K. Tanaka and S. Fujii, 1989. cDNA cloning of an mRNA encoding a sulfur-rich 10 kDa prolamin polypeptide in rice seeds. Plant Mol. Biol. 12: 123-130.

Masumura, T., T. Hibino, K. Kidzu, N. Mitsukawa, K. Tanaka and S. Fujii, 1990. Cloning and characterization of a cDNA encoding a 13 kDa prolamin. Mol. Gen. Genet. 221: 1-7.

McElroy, D., M. Rothenberg, K. S. Reece and R. Wu, 1990a. Characterization of the rice actin gene family. Plant Mol. Biol. 15: 257-268.

McElroy, D., M. Rothenburg and R. Wu, 1990b. Structural characterization of a rice actin gene. Plant Mol. Biol. 14: 163-171.

McElroy, D., W. Zhang, J. Cao and R. Wu, 1990c. Isolation of an efficient actin promoter for use in rice traisformation. The Plant Cell 2: 163-171.

Minami, E. I., Y. Ozeki, M. Matsuoka, N. Koizaka and Y. Tanaka, 1989. Structure and some characterization of the gene for phenylalanine ammonia-lyase from rice plants. Eur. J. Biochem. 185: 19-25.

Mundy, J. and N.H. Chua, 1988. Abscisic acid and water stress induced expression of a novel rice gene. EMBO J. 7: 2279-2286.

Mundy, J., K. Yamaguchi-Shinozaki and N.H. Chua, 1990. Nuclear proteins bind conserved elements in the abscisic acid-responsive promoter of a rice rab gene. Proc. Natl. Acad. Sci. USA 87: 1406-1410.

Raghavan, V. and A. Olmedilla, 1989. Spatial patterns of histone mRNA expression during grain development and germination in rice. Cell Differen. Devel. 27: 183-196.

Ramachandran, C. and V. Raghavan, 1990. Intracellular localization of glutelin mRNA during grain development in rice. J. Exotl. Botany 4: 393-399.

Reece, K. S., D. McElroy and R. Wu, 1990. Genomic nucleotide sequence of four rice (Oryza sativa) actin genes. Plant Mol. Biol. 14: 621-624.

Sano, H., I. Kamada, S. Youssefian, M. Katsumi and H. Wabilko, 1990. A single treatment of rice seedlings with 5-azacytidine induces heritable dwarfism and undermethylation of genomic DNA. Mol. Gen. Genet. 220: 441-447.

Wang, Z. Y. and S. D. Tanksley, 1989. Restriction fragment length polymorphism in Oryza sativa L. Genome 32: 1113-1118.

Xie, Y. and R. Wu, 1989. Rice alcohol dehydrogenase genes: Anaerobic induction, organ specific expression and characterization of cDNA clones. Plant Mol. Biol. 13: 53-68.

Xie, Y. and R. Wu, 1990. Molecular analysis of an alcohol dehydrogenase-encoding genomic clone (Adh2) from rice. Gene 87: 185-191.

Yamaguchi-Shinozaki, K., J. Mundy and N.H. Chua, 1989. Four tightly linked rab genes are differentially expressed in rice. Plant Mol. Biol. 14: 29-39.