Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
REVIEW ARTICLES Nuclear control and mitochondrial transcript processing with relevance to cytoplasmic male sterility in higher plants H. K. Srivastava Central Institute of Medicinal and Aromatic Plants, P.O. CIMAP, Lucknow 226 015, India Recent advances in the molecular biology of plant mitochondria have yielded some newer insights. A common, basic set of genes is encoded in all mitochondrial genomes, although their sizes differ greatly, with plant mitochondrial DNAs being by far the most complex. Mitochondrial genome mutation encodes cytoplasmic male sterility (cms) which in turn leads to stamen sterility or pollen abortion in several higher plants. Cms is the resultant of an incompatibility between the nucleus and mitochondrial genomes such that male pollen is aborted or not properly formed. Cms system in agriculturally important crops has been used to produce high-yielding and heterotic hybrid seeds, because it eliminates the need for labour intensive and expensive hand emasculation. Hybrid seeds (or varieties) hold great potential for improving crop economic yields, even when the average yields are much higher in many of the traditional food and feed crops. An important feature of mutation(s) responsible for cms is the discovery of chimeric genes or chimeric locus and different open reading frames joined together, or placed in proximal locations for cotranscriptions with other standard mitochondrial genes. Twenty-nine mitochondrial cms related genes in over twelve so far studied higher plant species have been characterized together with their specific DNA coding sequences. The recent development of in vitro systems for transcription initiation and RNA processing has yielded intriguing details of transcriptional and post-transcriptional regulation of plant mitochondrial cms gene. The nuclear gene Rf or Fr (restorer of fertility) prevents the expression of male sterility gene in the cms system in rice (B-atp 6) and maize (mitochondrial aldehyde dehydrogenase enzyme) by influencing RNA processing and sequential post-transcriptional editing. The latest findings on the subject support the notion that ‘intergenomic interaction’, i.e. specific nuclear-mitochondrial gene(s) cotranscription is operational as the plausible molecular mechanism for the manifestation of maternally and cytoplasmically inherited cms genetical trait in crop plants. THREE genetic systems, one nuclear and two cytoplasmic or extranuclear, mitochondria and chloroplasts are essential for expression and development of various quantitative and qualitative traits in higher plants. Plant mitochondria have been found to cover many facets of genetics and molecular biology 1. The first evidence for the presence of genes outside the nucleus was provided by non-Mendelian inheritance in plants2. Non-Mendelian inheritance and somatic segregation are taken to indicate the existence of functional genes that reside outside the realm of the nucleus and do not undergo hereditary type conventional segregation and transmission on the meiotic and mitotic spindles to distribute genomic replicas to gametes or to daughter cells during microsporogenesis. The interpretation for the frequent occurrence of novel heritable genetic or even nonheritable epigenetic variations has been debated among geneticists and breeders for almost two decades. Exciting new findings have emerged in terms of plant organelle (mitochondria and chloroplast) gene expression3,4 in relation to the manifestation of cytoplasmic male sterility (cms) and heterosis, after the author’s earlier reviews on the related subject exhibited some kind of nuclear control of mitochondrial and chloroplast development and function in higher plants which now provide convincing evidence for a plausible DNA transfer (mobile genes) between the mitochondria, chloroplast and nucleus5–7, i.e. intergenomic interaction. Cms provides a convenient and proprietary means to produce hybrid seeds. Srivastava5 propounded a theory of ‘intergenomic interaction’ to explain genetical phenomenon of heterosis or hybrid vigour based on experimental evidences then available, that mitochondria may interact structurally, functionally and biosynthetically with other cellular-genetical components like nuclear and chloroplast genes so as to endow an overall superiority to heterotic hybrids in many measurable attributes, including crop grain yields. Much of the interest in cms plants stems from their use, or potential use, in producing heterotic hybrid seeds. Presently, hybrid seeds of some crops are produced with the use of cms inbred (predominantly homozygous and homogeneous) lines, while production of others depends upon tedious hand emasculation. The eukaryotic cell nucleur DNA content e-mail: [email protected] 176 CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES (× 10–14 g) is immensely greater (4.0–7.0) compared to chloroplast DNA (0.3–0.9) and mitochondrial DNA (0.1–0.3) 5. Mitochondria contain their own DNA (mt DNA) and the transcriptional and translational machinery necessary for protein synthesis. Mitochondria, however, are not fully autonomous, and their biogenesis and function depend on coordinated expression of both nuclear and organellar genomes and genes. mtDNA is particularly maternally or uniparentally inherited. Plant mitochondria also possess an alternative route for electron transport system or cyanide-resistant respiration 8. Many independent mutations that disrupt pollen development or abnormal aborted pollen production have been observed in a wide variety of higher plant species. In tomato, for example, nuclear encoded male sterility mutations have been mapped to over 40 different loci. Many other male sterile mutations, however, do not exhibit Mendelian inheritance. Such cytoplasmic inherited mutations have been so far reported in over 140 higher plant species 9 and the possibility of discovery of novel plant cms gene(s) in other plant species exists through further intensive research. Despite identification of several mitochondrial genes that specify cms trait, its precise molecular basis or molecular mechanism is not understood in any plant species. An updated account of current knowledge concerning the nature of the nuclear–mitochondrial interaction resulting in plant mitochondrial gene mutation(s) encoding economically and commercially relevant cms trait in several important food and other crop species is provided in this article. Plant mitochondrial genomes organization The in vivo structure of mitochondrial genome organization has been adequately covered 1,10,11 . Earlier studies have elucidated the derivation of circular structures from physical mapping of overlapping restriction fragments; from these maps circular genome structures are predicted for the majority of mitochondrial genomes in eukaryotic cells 12 . A more comprehensive coverage of mitochondrial genome organization, evolution, cms, and analysis of mtDNA in somatic hybrids has been provided 4,13,14 . Several general features of higher plant mitochondrial genome organization have emerged: (i) The genomes are larger than those from animals and fungi. (ii) Restriction endonuclease mapping (using DNA markers such as RFLPs, RAPDs and AFLPs) indicates that plant mitochondrial genomes are usually organized as multiple circular molecules, with conversion of circle types mediated by recombination between repeated sequences. (iii) Mitochondrial genomes from closely related species are highly conserved in the primary sequence, but vary greatly in linear gene order. (iv) The plant mitochondrial genome, like that of other eukaryotes, encodes only a small percentage of proteins located within the mitochondria, indicating that over the course CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 of evolution most of the genes that resided in the original endosymbiont (earlier as prokaryotes) have migrated to the nuclear genome. (v) In addition to high molecular weight mtDNA, plasmid-like molecules are present in the mitochondria. (vi) Chloroplast DNA and nuclear DNA sequences are often found inserted in the mtDNA. Higher plants exhibit an extremely wide range of variation in mtDNA size with a minimum of ~ 100 kb. Higher plant mitochondrial genomes, which vary in size from 200 kb to 2500 kb are much larger. Analyses of Cucurbit mtDNAs demonstrated a seven-fold range in genome size within the family, from 330 kb in watermelon to 2500 kb in muskmelon. Despite the large size differences, there is no correlation of mitochondrial genome size with repeated DNA sequences, mitochondrial volume or the number of translational products. The function of additional DNA has not been precisely enumerated, but it seems likely that much of it either is noncoded or is subjected to some kind of shuffling mechanism between ‘intron’ and ‘exon’ sequences. The size of the genome makes it difficult to isolate the mtDNA intact, but restriction mapping (RFLPs, RAPDs and AFLPs) in several crop plants suggests that each has a single sequence organized as a circle. Within this circle there are short homologous sequences. Recombination between these elements generates smaller, subgenomic circular molecules that coexist with the complete ‘master’ genome or ‘principal’ genome explaining the apparent complexity of plant mtDNAs. The mixing of cytoplasms in higher plants promotes mitochondrial fusion and recombination between varying mitochondrial genomes 5,6 . Unlike animal and fungal mtDNAs, which are usually organized in a single circular molecule, plant mtDNA often consist of an assortment of genomic and subgenomic circular molecules of different sizes and frequencies 1. The cause of this variability was first explained in Brassica campestris. Its mtDNA is primarily composed of a single large 218 kb circle called the ‘master’ chromosome plus smaller circles of 135 kb and 83 kb. Within the master chromosome are two 2 kb direct repeats separated by 83 kb on one side and 135 kb on the other. Recombination between the two repeats generates the smaller 135 kb and 83 kb circles. Moreover, recombination between the two smaller molecules can also regenerate the ‘master’ chromosome. The 570 kb maize genome has six major pairs of repeats, which may participate in recombination to produce a large array of circular molecules of various sizes and frequencies. The relative proportions of the different circles apparently depend on the recombinational frequencies among the different pairs of repeats and whether the various circles are capable of autonomous replication. Moreover, genomic rearrangements of the atpA gene characteristic of cms-S (USDA) and cms-T (Texas) maize are found at low levels in N (normal) mitochon177 REVIEW ARTICLES dria 15,16 . It is thought that these ‘substoichiometric’ atpA types are part of a larger circular or linear molecule present in low copy number relative to the rest of the mitochondrial genome in the cell. The substoichiometric molecules called ‘sublimons’ may have originated from infrequent recombinational events between very short regions of homology in the mitochondrial genome. Thus, the products of rare or unique recombinational events are retained in the genome at low levels and these sublimons may then be expected to exhibit rapid molecular evolution, because mutational events are more quickly fixed by chance in a small population of molecules. Occasional amplification of sublimons could cause sudden genomic organization, possibly leading to the evolution of cms 1,17 . It has been unequivocally demonstrated that mitochondrial genomes of many higher plants contain subprinciple genome DNAs and/or double-stranded RNAs which have been termed minilinear and minicircular 18 for convenience in distinguishing them from the higher molecular weight principal genome. Tobacco tissue culture cells yield a large percentage of small circular DNAs, which are derived from the principal genome. Of the smaller circular DNAs, the smallest ones (10.1, 20.2 and 30.3 kb) merge to reach the size of some of the larger minicircular and minilinear DNAs, found in other species 19 . It is interesting that the coding sequences of cytochrome c oxidase subunit II have been shown to be located on the small circles of Oenothera 11,20,21 . The small, circular DNA molecules of animal cells of about 16 kb have been visualized by electron microscopy and can be isolated as supercoiled molecule populations by CsCl density gradient centrifugation 22 . The in vivo structure of the larger mitochondrial DNAs of yeast and plants, however, may be different from the structures predicted by frequent mapping. Pulse field gel electrophoresis exhibits one fraction of molecules to be locked into highly complex arrangements that migrate with an apparent size of more than one megabase 22 . Most plant mtDNAs behave like linear molecules ranging from 50 to 100 kb in a heterogeneous population visible as smear in the gels 10 . Linear DNA molecules have thus been proposed as the major in vivo form of mitochondrial genomes in higher plants 10 . Effects of DNA anchoring at membrane integral proteins and specific mechanisms of replication such as those leading to catenated network of trypanosome mtDNA 23,24 may be involved in determining the variation in linear and circular structures in the mitochondria. Mitochondrial gene(s) and their transcription Extranuclear genetic systems are a fundamental feature of higher plant cells. The cytoplasmic organelles pos178 sess their own DNA. The mitochondrial genomes contribute a limited (10%) but essential number of gene products to the biogenesis of mitochondria, that are competent to carry out oxidative phosphorylation and energetically important cellular respiratory functions. Mitochondria from all organisms provide ATP as the principal energy source for the cell and deliver numerous substrates via specific carriers for biosynthetic reactions in the cytoplasm. Many basic features of the mitochondrial structure and function, developed at an early stage of evolution, have been highly conserved between animals and plants despite a billion years of divergent evolution. The assembly and function of respiratory-competent mitochondrion in eukaryotes result from a collaboration and coordination between gene products derived from mitochondrial and nuclear genomes. The inventory of nuclear and mitochondrially coded proteins required to assemble a functional mitochondrion shows clearly that nuclear and mitochondrial genomes interact in at least two ways. First, both nuclear and mitochondrial genes contribute to mitochondrial protein function. Second, both nuclear and mitochondrial genomes interact to affect the synthesis and assembly of mitochondrial proteins 5,6 . Communication from the mitochondrion to the nucleus probably involves metabolic signals and one or more signal transduction pathways that function across the inner mitochondrial membrane. Nuclear background influences extensive editing of plant mitochondrial transcription in particular genomic regions. The complete nucleotide sequences of several animal mitochondrial genomes have been published 25–27 and those of certain fungi are nearing completion 28 . These analyses have made possible the location of the coding regions (mitochondrial transcript) and construction of detailed mtDNA restriction maps 12,19,29–33 . The picture as regards the number of genes encoded in the higher plant mtDNA has become clearer. Isolated plant mitochondria synthesize some 20 to 50 polypeptides 34 which are presumed to be encoded by the mtDNA. It has been observed that higher plants also encode 26S rRNA (ref. 34) and in maize mitochondria coxI, coxII, cob, atp and 5S, 18S and 26S rRNA genes have been identified on the circular 570 kb restriction map 35 . The protein coding genes are obviously not clustered, whereas the 5S and 18S rRNA genes and the ATPase gene are transcribed from the same mtDNA strand, while the 26S rRNA, coxI, coxII and cob genes are transcribed from the opposite strand 36,37 . Approximately one-tenth of the mitochondrial proteins by weight are specified by the limited number of genes on the mitochondrial genome. Thus mitochondrial genomes contribute a limited but essential number of gene products to the biogenesis of mitochondria, that are fully competent to carry out respiratory functions such as electron transport and oxidative phosphorylation. It is obvious CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES then that a significant number of nuclear genes must be involved in the synthesis of mitochondrial components, expression of the mitochondrial genome, and control of mitochondrial biogenesis. Nuclear background alters transcription of atpA in the ‘ogura’ cytoplasm of radish 38 and the cms gene (T-urf 13) in the T cytoplasm of maize 39 . The T-urf 13 cms associated sequence is cotranscribed with the open reading frame orf 221. Specific inbred lines of maize revealed some marked influence of nuclear background on the pattern of transcription and mitochondrial function 40 . Rocheford et al. 41 demonstrated that the transcript changes were not only a function of the mitochondrial genomic environment encompassing the T-urf 13/orf 221 region but also of dominant nuclear gene action (following Mendelian hereditary transmission principles) influencing the pattern of post-transcriptional processing of the transcripts. Mitochondrial transcription The common functions of the mitochondria in plants, animals and fungi are reflected by similarities in their genomes 42 . The recent development of in vitro systems for transcription initiation and tRNA processing has yielded intriguing details of transcriptional and posttranscriptional regulation in plant mitochondria. Most of the coding regions are separated by several kilobases of non-coding regions 20,43–45 . This organization implies that most of the essential coding information is expressed as monocistronic transcripts from individual promoters, for example, atp6, atp9, cytob, coxI, coxII, and coxIII genes in most plants. When transcripts are analysed by Northern blot hybridization, the mRNAs are found to be generally much larger than the actual coding regions of the genes and to include extended non-coding 5′and 3′transcribed regions of several hundred nucleotides. The frequent genomic recombinations in plant mitochondrial genomes can place genes that are far apart in one genome into close proximity in another species. Genes located close together almost invariably exhibit co-transcription 11,46,47 , such as the 18S-5S-RNA genes or nad3-rps12 transcription units. The transcribed spacers between, for example, the 18S-5S rRNA genes vary among plant species, ranging from just over 100 nt to more than 500 nt; sequences apparently can become integrated or lost by intra-or intermolecular recombination 48 . Larger trancription units would most likely be found upon further detailed investigation of higher plants. The stably transcribed fraction of higher plant mitochondrial genomes has been estimated to be about one-third of the genomic complexity, for example in Brassicaceae and in Cucurbitaceae 49 . This RNA population is equivalent to the entire sequence information common to all plant mitochondrial genomes. The reCURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 maining 70% of the plant mitochondrial genome appears to be non-essential sequence information derived partly from chloroplast DNA (ct DNA), partly from nuclear sequences and from the duplication of mitochondrial sequences degenerated to various degrees. The presence of numerous transcription units implies a corresponding multitude of individual promoters of transcription and the presence of several promoters per genome has been confirmed 4. For some genes, multiple promoters have been identified which are actively engaged in transcription and may be used for differential regulation of the respective gene activity. The coxII gene in maize, for example, is transcribed from a closely stacked set of promoter sequences, and the atp9 gene in this plant is preceded by at least six active promoters spaced over several hundred nucleotides upstream of the coding region 4,47,50 . The transcripts arising from such multiple promoters have different sizes and thus complicate transcription patterns of individual genes even in monocistronic units 4. Transcription patterns of individual genes are further complicated by the presence of several gene copies in a given genome. In Northern blots, detectable transcripts of different sizes may arise from duplicated genes with different 5′or 3′ regions adjacent to the respective open reading frame that leads to variable extensions at either end. Common to mitochondria in animals, fungi and plants appears to be the positioning of the conserved sequence block and the first transcribed nucleotide. The conserved sequence motif generally appears to include the first transcribed nucleotide of all mRNAs in yeast and the first two nucleotides in plant mitochondria 50 . Mitochondrial activity, although required in all plant tissues, is capable of adopting to specific requirements by regulated gene expression 4. Investigation of the factors governing the quality and quantity of distinct RNAs will define the extent of interorganelle regulatory interface in mitochondrial gene expression. In vitro transcription systems have been established and now permit the identification of sequences necessary and sufficient for efficient transcription initiation 4,51,52 . Post-translational control having some important role in the regulation of plant mitochondrial genome expression has been considered earlier 53,54 . Not much is known about the mechanisms that control gene expression at post-translational 3′and 5′ends of mRNAs 4. Like chloroplast transcripts, plant mitochondrial mRNAs tend to contain inverted repeat sequences in their 3′ -region that can fold into stem–loop structures 55–58 . The steady-state level of mRNA from the ‘ogura’ cms locus of Brassica cybrids being determined post-transcriptionally by its 3′region has recently been demonstrated by Bellaoui et al. 12 . Their result strongly suggests that the steady-state level of mRNA from the orf 138 locus is determined post-transcriptionally, most likely by its 3′ -region. In the presence of rapeseed mito179 REVIEW ARTICLES chondrial lysate, synthetic RNAs corresponding to the 3′ -region of the Nco 2.7/13 F transcript are, as expected, less stable than RNAs corresponding to the 3′ regions of the Nco 2.5/13S and Bam 4.8/18S transcripts (see Figure 1). Future research should focus on better understanding of the role of these 3′regions and their post-transcriptional regulation in relation to mitochondrial gene expression in general and cms gene in particular. a Mitochondrial transcript processing Transcript processing is a widespread phenomenon in plant mitochondria though not yet well understood mechanistically. It was first deduced by the observation that unusually complex transcripts in certain regions of the plant mitochondrial genome were produced not only by multiple transcription start and stop sites, but also by internal transcript processing 44 . Although the role of processing in gene regulation is not yet clear, nuclear regulation of transcript processing could be an important means of mitochondrial gene suppression 4. This inference is based on the observation that three different nuclear Rf (restorer of fertility as dominant gene) loci directly influence transcript processing within the respective mitochondrial cms-associated regions. In cmsT maize, perhaps the most well-studied example of cms, Figure 2. a, Structure of the maize (Zea mays) cms-associated chimeric gene. Portion of urf13 derived from a normal atp 6 gene and a normal 26S rRNA gene. Atp6 and 26S rRNA genes are present elsewhere in the maize cms-T mtDNA, but the urf25 gene downstream of urf13 is the only copy present in maize cms-T (refs 29, 31). Chloroplast DNA sequences derived from an arginine tRNA gene are found at the 3′terminus of the urf25 gene from both a normal fertile maize line and cms-T (ref. 32); b, Transcription termini in the maize cms associated chimeric gene. Arrows indicate the location of mapped termini. The arrow labelled T–Rf 1 indicates a transcript that is increased in abundance in lines containing Rf 1 allele 30 . (Figure adapted from Hanson 21 ). Figure 1. Restriction maps of the three configurations Nco 2.5/13S, Nco 2.7/13F and Bam 4.8/18S, specific to the 13S (sterile), 13F (fertile) and 18S (sterile) Brassica cybrids, respectively. The sequences of the Nco 2.7/13F and Bam 4.8/18S fragments which are common with the Nco 2.5/13S fragment are indicated by a thick black line. White box: trn fM gene coding for initiator methionine rRNA; light box with plain line; orf138 gene; dark box: orfB gene; striped box: atpA gene. In vivo accumulated transcripts from each configuration are indicated by the bent arrows. (Adapted from Bellaoui et al. 12 ). fertility restoration (Rf) requires dominant alleles of two unlinked nuclear loci, Rf 1 and Rf 2 (ref. 59). Compelling evidence demonstrates that the product of Rf 1, essential though not sufficient to restore fertility, directly influences transcript processing of the T-urf 13 mitochondrial region (see Figure 2). Processing of T-urf 13 transcripts appears to be directly associated with a marked reduction in the expression of the encoded 13 kDa T-URF 13 polypeptide 20 . Although the function of Rf 2 gene in fertility restoration is not yet understood at the molecular genetic level, it has been speculated that the Rf 2 gene product may play a role in the detoxification of a pollen-specific product that interacts with the T-URF 13 protein to cause premature tapetal breakdown leading to premature or aborted pollen grains 60 . In Sorghum line IS 1112 C, cms is associated with expression of the open reading frame orf 107 (ref. 61). Again fertility restoration is associated with altered processing of orf 107 transcripts and the concomitant reduction in the accumulation of a 12 kDa polypeptide presumed to be the gene product 61 . Of particular interest is the observation that the transcript processing sites described in both cms-T maize and cms sorghum share sequence features 62 , implying that sequence motifs exist within plant mitochondrial genes that may act as targets for nuclear-directed gene modulation. 180 b CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES Cms in the B. napus (oilseed rape) polima cytoplasm is associated with aberrant expression of a region encoding the atp6 gene cotranscribed with a downstream chimeric sequence, orf 224 (ref. 63). Fertility restoration conditioned by either of two dominant single nuclear loci, results in a transcript processing event that generates predominantly monocistronic atp6 transcripts 63 . Generation of monocistronic atp6 transcripts cosegregates with a single dominant fertility restorer locus, Rfp. Most intriguing, however, is the observation that an alternate recessive allele at this locus (rfp 1) or a second locus tightly linked to rfp 1, influences transcript processing of two other mitochondrial genes not associated with male sterility, nad4 and a cell-like gene that may be involved in mitochondrial cytochrome c biogenesis 63 . At all four processing sites under Rfp 1 or rfp 1 control, there is UUGUGG or UUGUUG sequence motif located very near the processing site. RNA editing alters gene products Although only 0.13% of the codons in the entire maize chloroplast genome is altered, plant mitochondrial protein coding genes typically have 3–15% of the codons affected by RNA editing 64 . The discovery of RNA editing has changed the way plant organelle genes must be analysed. Before 1991, sequencing of genomic DNA was thought to be sufficient to characterize the coding regions of most plant mitochondrial genes. Earlier literature did predict protein sequences from genomic DNA sequences 4. However, some anomalies were noted; often plant mitochondrial proteins predicted from genomic sequences differed from mitochondrial proteins in other organisms. As a result, it was proposed that like yeast mitochondrial genes, the genetic code in plant mitochondria differed from the universal code. Because the CGG codon (which codes arginine) was present where homologous genes in other organisms exhibited ‘tryptophan’, codons such as TGG, CGG were at first proposed to encode tryptophan in the plant mitochondrial genetic code. Now it is known that the genetic code in plant mitochondria does not differ from the standard code but that some CGG codons actually specify UGG as a result of C to U RNA editing. When editing occurs, the encoded amino acid is likely to be altered because the codon position of the edited C is not random. In chloroplasts, the second codon position is the primary target of editing 65 , in mitochondria, the first and second positions of codons are more often edited than the third position 66,67 . Plant organelle proteins are more similar to their homologous counterparts in other organisms than was originally realized 4. Protein sequencing of two mitochondrial gene products (wheat ATP 9 and potato NAD 9) has confirmed the presence of the amino acid predicted from edited codons 68 . CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 Editing of RNA may shorten the predicted open reading frame to the length expected from analysis of homologous genes. The stop codons UAA, UAG and UGA could be created by editing the C-containing codons. Mitochondrial atp6, atp9 and rsp10 genes from several plant species have ‘edit encoded’ stop codons, which slightly shorten the open reading frame 69,70 . A bizarre example of an edit-encoded stop codon occurs in the petunia rpl16 gene. When 15 cDNAs were sequenced, all were found to carry an edit-encoded stop codon early in the open reading frame predicted from the genomic sequence 71 . Editing evidently has converted the mitochondrial rpl16 gene (mt-rpl16) into a pseudogene. Presumably a functional rpl16 gene may be present in the nucleus. Most plant chloroplast and plant mitochondrial genes carry genomically encoded AUG start codons, unlike human sleeping sickness blood stream protozoa trypanosome 72 mitochondrial genes, in which insertion of U often created the start codon. Nevertheless, ACG to AUG editing creates the start codons of transcripts of the wheat mt-nad1 gene, the potato mt-coxI gene, the maize chloroplast rpl12 gene (cp-rpl12) and the tobacco cp-psbL and cp-ndhD genes 73,74 . Transcripts of different genes vary in extent of editing Mitochondrial genes can be divided into three classes with respect to their transcript editing. All transcripts of some mitochondrial genes appear to be fully edited; direct cDNA sequencing as well as sequencing of multiple, independently derived cDNA clones, indicate that all editing sites have Ts replacing Cs. Other genes exhibit partially edited precursor transcripts, but the mature transcript population is fully edited 75,76 . The third class of genes exhibits a low proportion of fully edited transcripts; most transcripts are partially edited 77,78 . When a gene exhibits partially edited transcripts, many individual cDNAs must be sequenced to make sure that every editing site is detected. The existence of partially edited transcipts, and the finding that spliced transcripts are more highly edited than unspliced transcripts, have led to the view that RNA editing in plant mitochondria is a post-transcriptional process 12,33 . Fewer data are available concerning the extent of editing of chloroplast transcripts, but partially edited transcripts of a few genes have been reported 4,12,33 . Editing can occur before splicing in both organelles 4,79 . Transcripts of almost all mitochondrial protein-coding genes contain at least one editing site (Table 1; also see Figures 1–4). Mitochondrial editing events have 5′and 3′non-translated regions, introns, ribosomal RNAs and transfer RNAs, but in general, coding regions are more highly edited than non-coding regions. Not only does the extent of editing of different mitochondrial genes’ transcripts vary within 181 REVIEW ARTICLES the same species, but the number of codons edited varies in transcripts of the same gene in different species 4,29–32 (Table 1). In plant, there is no published evidence that transcripts present in comparable abundance in different tissues are differentially edited. Transcripts of psbL are present in similar amounts in chloroplasts and chromoplasts of ripening bell pepper (Capsicum annum), and editing occurs in both types of plastids 80 . When the degree of editing of spinach cp-psbF and cp-psbL transcripts was examined by direct sequencing of cDNAs, Table 1. Altered codons in plant mitochondrial protein-coding genes* Mitochondrial gene (cms related) atp6 atp9 CoxII nad1 nad3 Pcf rpl16 rps12 Rps19 mt-ATP 6 core region Rice Sorghum Radish Petunia Ocnothera Number of codons altered (%) 6 13 5 7 13 3 3 4 5 Number of codons changed (%) 4 6 0.4 6 8 *Adapted from Hanson et al. 64 with modification. Figure 3. Construction of orf239 chimeric gene in transgenic tobacco (Nicotiana tabaccum). Strategy for cloning each gene construct into the transferred DNA binary vector pB1101 is shown. Aocs, activator of the Octopine synthase gene (trimer); Amas 2′ gene, activator of mannopine synthase 2′gene; Pmas 2′ , promoter of the mannopine synthase 2′gene; PLat 52, promoter of the tomato pollen specific gene Lat 52. The boxed ATG is the first codon of orf 239. ? , Site specific cleavage; *, the amino acid altered as a consequence of cloning (Adapted from He et al. 19 ). 182 Figure 4. Schematic diagram of atp6 region of cytoplasmic male sterile rice (Oryza sativa). The N-atp6 and B-atp6 genes isolated from N(r), B (R), and C. The coding region is shown on a dotted box. The probes used for nucleic acid hybridization are indicated by bars (1–5). The sizes of probes 1–5 are 0.55, 0.4, 1.3, 0.45 and 1.7 kb, respectively. The primers used for sequencing, PCR amplification and primer extension experiment are shown as arrow heads (a–j). The atp6 RNAs are represented by arrows. The transcript of B-atp6 is altered, whereas that of N-atp6 is not changed by the presence or absence of the Rf 1 gene. The restriction sites are shown on E, EcoR1; H, HindIII; P, PstI; T, EcoT221, S, SalI and V, EcoRV. (Adapted from Iwabuchi et al. 33 ). Bock et al. 81 found that the extent of editing was lower in seeds and roots than in leaf etioplasts and chloroplasts. The reverse transcription-polymerase chain reaction (RT-PCR) assay was applied to search for differences in the extent of editing of petunia nad3 transcripts in different tissues, no developmental regulation of the extent of editing was detected 21 . In the absence of a transformation system in which sequences can be deliberately altered, editing can be assessed in abnormal genes created by the rearrangement characteristic of plant mitochondrial genomes 82 . In genes created by such aberrant recombination events, pieces of the coding regions of mitochondrial genes are put into different contexts, but editing events in the translocated regions are usually still observed. Despite numerous recombination events placing portion of atp9, parts of the first and second exon of Cox II, and an unidentified reading frame (urf) together in the petunia pcf gene, almost all sites are edited 83 . A 193-nucleotide segment of wheat Cox II is edited when present as part of the nad3/rps12 transcript 27 . In the atp6 gene, found in maize T-cms, a segment containing 30 nucleotides of the coding region, plus some upstream sequence derived from the atp9 coding region was still edited even when located next to an unidentified urf 84,85 (see Figure 2). Nuclear-encoded genes required for normal mitochondrial gene expression have been described for several mitochondrial genes in yeast 86 . They regulate the expression of mitochondrial genes at both the transcriptional and post-transcriptional levels. In higher plants the expression of the nuclear and mitochondrial genomes must be coordinated to make a functional cell. Cms is a loss of functional pollen in plants that can be CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES attributed to unique mitochondrial genomes. Cms is the resultant of an incompatibility between nucleus and mitochondrial genomes such that pollen is not properly formed. However, there are nuclear genes designated Rf or Fr (restorer of fertility) that can correct this incompatibility and allow the development of functional pollen. Thus, cms offers a system with which to study the mechanism(s) by which nuclear genes directly regulate mitochondrial gene expression. Among the several Rf genes known to reduce or completely eliminate the cms phenotype, some specifically affect the transcription patterns of the cms gene. One of the restorers introduces new processing sites into the mutant T-urf13 transcripts and leads to almost complete reduction of the corresponding polypeptide synthesis 17,86 (see Figure 2). The open reading frame (orf) of the mitochondrial complex IV gene CoxI is extended by a recombination event in sorghum cytoplasms and leads to a cms phenotype upon high expression 4. As yet not known nuclear genes revert the cms phenotype and compensate for the expression of this aberrant protein 87 . In sunflower mitochondria, a recombined novel orf correlated to a cms phenotype is likewise influenced by specific nuclear gene expression 4. Comparison of Northern blot analysis and run-on experiments has exhibited that the nuclear restorer genes act at the post-transcriptional level and alter transcript stability. Recent results suggest that RNA processing by Rf 1 in rice-cms system does influence the sequential post-transcriptional editing of the B-atp6 RNA 33 (see Figure 4). The B-atp6 gene was identified as a cms gene by Southern analysis of the mitochondrial genome. The coding sequence of B-atp6 was identical to the normal N-atp6 gene but its 3′ -flanking sequence was different starting at 49 bases downstream from the stop codon 33 . Northern analysis showed that B-atp6, transcribed into a 2.0 kb RNA, in the absence of the Rf 1 gene whereas two discontinuous RNAs, of 1.5 and 0.45 kb were detected in the presence of the Rf 1 gene. Iwabuchi et al. 33 interpreted these results by suggesting that the unprocessed RNAs of B-atp6 are possibly translated into altered polypeptides and that interaction of RNA processing and editing plays a role in controlling cms expression and the restoration of fertility in rice. The recent identification of a nuclear restorer gene in maize (rf 2) coding from a mitochondrial aldehyde dehydrogenase suggests that metabolic enzymes influence directly through gain of function or indirectly by their metabolic activity 16 . Cms phenotype caused by mitochondrial mutation The cms phenotype is now thought to be caused by mitochondrial mutations. Mitochondrial genomes of CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 higher plants are too large (over 200–2400 kb) to permit identification of mutation by complete genomic sequencing. In plants, several strategies have been used to locate mtDNA regions of interest for sequencing. In maize, type cms-T, a genomic library was screened with cDNAs from mitochondria of fertile vs male sterile (cms) maize lines, to identify the gene region expressed differentially under the influence of the nucleus 4. So far twenty-nine mitochondrial cms related region(s) or gene(s) (mostly chimeric in nature) in over twelve plant species have been completely or partially characterized (Table 2; also see Figures 1–4)). Mitochondrial genes have been reported to be associated with cms cytoplasm in maize, sunflower, petunia, sorghum and radish 88–96 (Table 2). In maize, petunia and sorghum a chimeric mitochondrial gene which produces a novel protein has been found in cms plant mitochondria 87,92,97 . In the mitochondria of cms sunflower a new open reading frame is co-transcribed with the atp1 gene 98 and the protein supposed to be encoded by this orf is uniquely present 99 . Among these examples only the T-urf13 (Figure 2) associated with maize cms-T cytoplasm has been strongly correlated with the cms phenotype based on the analysis of a number of mutants which are toxin resistant and fertile 100 . Although novel proteins uniquely present in the cms mitochondria, an altered gene expression caused by mitochondrial genome rearrangments is thought to be responsible for cms, the precise Table 2. Mitochondrial region/gene(s) involved in cytoplasmic male sterile phenotype in several higher plant species Plant species Mitochondrial gene(s)/ chimeric mitochondrial gene Rice (Oryza sativa) Maize (Zea mays) Sorghum (Sorghum vulgare) Sunflower (Helianthus annuus) Oilseed Rape (Brassica napus) Bean (Phaseolus vulgaris) Teosinte (Wild relative of maize) Arabidopsis thaliana Nicotiana tabacum Brasssica tournefortii Radish (Raphanus sativum) Triticum timopheevi Petunia (Petunia hybrida) Brassica cybrids (Ogura cytoplasm) Atp 6 genes (N-atp 6 and B-atp 6), orf 25, cox III T-urf 13, orf 221, atp 6, atp 9 orf 107, Cyto c oxidase atp A, atp 9, Cob, rrrn 26, orf 552 atp 6, orf 224, nad 4 References 33, 46 16, 21, 30, 34, 100, 116, 119–121 61, 87, 89 98, 99, 122 63 pvs sterility sequence, orf 98 and orf 239 Cox II 13 11, 47 CHM orf 25, orf 239 orf 3 atp A atp 6, orf 25 pcf-a orf 138, orf B 123 19, 107 108 96 110 124 125, 126 183 REVIEW ARTICLES molecular mechanism underlying cms expression still remains to be elucidated. Observation of altered electron transport in petunia and toxin-mediated membrane disruption in maize plants, bacteria and yeast expressing the maize urf13 gene product provide clues to possible mechanism of disruption of pollen development or dysfunctional mitochondria in aborted pollen. Whether disruption in a particular mitochondrial function is at the root of cms in all higher plant species, or whether defects in numerous mitochondrial activities can produce pollen male sterility will only be unravelled by further probing of physiological and biochemical defects present in cms genotypes. Nuclear gene influence The nucleus somehow affects mitochondrial genome organization and function in higher plant systems. The plant mitochondrial genome, now fully sequenced in Arabidopois 101 and Marcantia 102 , is organized in a much more complex and variable structure than is observed in other higher plants or eukaryotes 103 . In most higher plants, this organization is defined by the presence of recombinationally active repeated DNA sequences that allow for high- and low-frequency inter- and intramolecular recombination events to occur 104 . The physical organization of the mitochondrial genome in plants has proved to be rather difficult to define. Although most genomes map on circular molecules defined by overlapping clones, direct physical observation by pulsed field gel electrophoresis, electron microscopy and other procedures has revealed that the genome may consist of both linear and circular forms, with molecules much larger than the multiple circles constructed by clone analysis 105 . A multipartite genome organization exists in the mitochondria of most plant species, with each molecule containing only a portion of the genetic information. The relative copy number of the various mtDNA forms and their recombinational activity appears to be under nuclear control. One of the most pronounced examples is the observed loss of a mitochondrial genomic molecule in response to a single nuclear gene in common bean (Phaseolus vulgaris)13,106 . The cms-associated mitochondrial mutation in common bean, pvs-orf 239, appears to be maintained on a single 210 kb molecule within a tripartite mitochondrial genome organization 13 . Introduction of a single dominant nuclear factor, Fr, (fertility restorer) results in a genomic shift of the pvs-orf 239-containing molecule to substoichiometric levels within the genome, thus restoring pollen fertility/male fertility. The development of alloplasmic lines, derived by recurrent backcrossing strategies or protoplast fusion to combine different mitochondrial and nuclear genotypes, routinely gives rise to changes in relative stoichiometries and mitochondrial 184 genomic rearrangements in Nicotiana 107 , Brassica 108,109 and Triticum 110–112 . In some cases, it has been possible to identify specific nuclear loci essential to establish compatibility in individual nuclear–cytoplasmic (intergenomic) combinations 113 . It may thus be possible to predict particular nuclear–cytoplasmic intergenomic interacting genetic combinations which would give rise to high frequency of specific mitochondrial mutations. Several nonchromosomal stripe mutations (ncs) affect distinct loci and have arisen by what appears to be different molecular events 11,47,114 . A characteristic of cms mutation in plants is the presence of chimeric genes or chimeric locus and different open reading frames are joined together or placed in proximal locations and cotranscribed with standard mitochondrial genes. Despite much progress, and the identification of several mitochondrial loci that specify cms in a number of crop plants (Table 2; also see Figures 1–4), the molecular basis of this nuclear–mitochondrial interactive generelated genic defect is not well understood in any plant species. Observations of altered electron transport in wheat 115,116 and petunia and toxin-mediated membrane disruption in maize plants 15,16 , bacteria, and yeast expressing the maize urf13 gene product, provide clues to possible mechanisms for disruption of pollen development 117 . Whether disruption in a particular mitochondrial function 115,118,119 is at the root of cms in all species, or whether many nuclear–mitochondrial intergenomic interaction-related genetic defects in numerous key mitochondrial activities related to electron transport and oxidative phosphorylation (ATP energies) can produce male pollen sterility, will only be revealed by further probing of physiological and biochemical defects naturally or evolutionarily present in cms genotypes of higher plants. Cms mutations have proven particularly useful in defining regulation of mitochondrial respiratory functions. Data suggest that variation in promoter strength is likely a primary influence on relative transcription rates 4. Nucleus genes regulating transcription rate or promoter selection are difficult to detect. Perhaps the most compelling demonstration of nucleus control on mitochondrial gene transcription is described in a Zea mays/Zea perennis alloplasmic line. A single nuclear gene, described Mct, influences promoter selection in CoxII gene in maize 11,47,114 . Transcriptional initiation at position 907 produces the predominant CoxII transcript of ~1900 nucleotides. The dominant nuclear Mct allele apparently directs transcriptional initiation at a second site (– 347) upstream to the CoxII locus. Interestingly, this alternate initiation site, detected as a shorter CoxII transcript, does not conform to the consensus promoter sequence described for maize 4,15 . This provided some circumstantial evidence that specific nuclear gene produced cofactors may influence promoter selection in cms-related genomic region of plant mitochondria. While nuclear genome is undoubtedly the CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES principal source of heredity and Mendelian gene transmission, the role played by cytoplasmic and maternally and non-Mendelian inherited mitochondrial genome through intergenomic interaction linked cotranscriptions in giving rise to evolutionarily and speciation wise important cms trait and resultant hybrid seeds with heterosis having high proprietary, economic, and commercial relevance in higher plant world needs further focusing and mechanistic unravelling 115–118 . 1. Srivastava, H. K., Proc. Natl. Acad. Sci., India, 1998, 68, 73– 89. 2. Correns, C., Z. Vererbungs, 1909, 1, 291–293. 3. Leon, P., Arroya, A. and Mackerzie, S., Annu. Rev. Plant Physiol. Plant Mol. Biol., 1998, 49, 453–480. 4. Binder, S., Marehefelder, A. and Brennick, A., Plant Mol. Biol., 1996, 32, 303–314. 5. Srivastava, H. K., Adv. Agron., 1981, 34, 118–195. 6. Srivastava, H. K., in Monographs on Theoretical and Applied Genetics (ed. Frankel, R.), Springer-Verlag, Berlin, 1983, vol. 6, pp. 260–286. 7. Srivastava, H. K. and Gupta, H. S., Biochemical Aspects of Crop Improvement, CRC Press, Florida, 1991, pp. 421–442. 8. Srivastava, H. K., FEBS Lett., 1971, 16, 189–191. 9. Laser, K. D. and Lersten, N. R., Bot. Rev., 1972, 38, 425–454. 10. Bennedich, A. J., Curr. Genet., 1993, 24, 279–281. 11. Newton, K. J., Annu. Rev. Plant Physiol. Plant Mol. Biol., 1988, 39, 503–53. 12. Bellaoui, M. et al., EMBO J., 1997, 16, 5657–5668. 13. Abad, A. et al., Plant Cell, 1995, 7, 271–285. 14. Dietrich, A. et al., Plant J., 1996, 10, 913–918. 15. Dill, C. L. et al., Genetics, 1997, 147, 1367–1379. 16. Cui, X. et al., Science, 1996, 272, 1334–1336. 17. Levings, C. S. III, Plant Cell, 1993, 5, 1285–1290. 18. Sederoft, R. R., Adv. Genet., 1984, 22, 1–35. 19. He, S. et al., Proc. Natl. Acad. Sci. USA, 1996, 93, 11763– 11768. 20. Schuster, W. and Brennicke, A., Annu. Rev. Plant Physiol. Plant Mol. Biol., 1994, 45, 61–78. 21. Hanson, M. R., Annu. Rev. Genet., 1991, 25, 461–486. 22. Bibb, M. J. et al., Cell, 1986, 26, 167–169. 23. Simpson, L., Int. Rev. Cytol., 1986, 99, 119–179. 24. Perez-Morga, D. L. and Englund, Cell, 1993, 74, 703–711. 25. Clayton, D. A., Annu. Rev. Biochem., 1984, 53, 573–594. 26. Anderson, S. et al., Nature, 1981, 290, 457–461. 27. Bonnard, G. et al., Crit. Rev. Plant Sci. 1992, 10, 503–510. 28. Grivell, L. A., Sci. Am., 1983, 248, 80–84. 29. Hanson, M. R. et al., Oxford Surv. Plant Mol. Cell Biol., 1989, 6, 61–85. 30. Kennel, J. C. and Pring, D. R., Mol. Gen. Genet., 1989, 216, 16–24. 31. Stamper, S. E. et al., Curr. Genet., 1987, 12, 457–463. 32. Stern, D. B. and Lonsdale, D. M., Nature, 1982, 229, 698–702. 33. Iwabuchi et al., EMBO J., 1993, 12, 1437–1446. 34. Leaver, C. J. and Gray, M. W., Annu. Rev. Plant Physiol., 1982, 33, 373–389. 35. Dawson, A. J., Jones, V. P. and Leaver, C. J., Methods Enzymol., 1986, 118, 470–485. 36. Fox, T. D. and Leaver, C. J., Cell, 1981, 26, 315–321. 37. Isaac, P. G. et al., EMBO J., 1985, 10, 1617–1620. 38. Makckenzie, S. A. and Chase, C. D., Plant Cell, 1990, 2, 905– 912. 39. Wise, R. P. et al., Genetics, 1996, 30, 502–508. 40. Kennell, J. C. et al., Mol. Gen. Genet., 1987, 210, 399– 406. CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 41. Rocheford, T. R. et al., Theor. Appl. Genet., 1992, 84, 891– 898. 42. Quetier, F. and Vedel, F., Nature, 1977, 268, 365–368. 43. Beven, L., Curr. Opin. Genet. Dobv, 1991, 1, 515–520. 44. Gray, M. W., et al., Annu. Rev. Plant Physiol. Plant Mol. Biol., 1992, 43, 145–151. 45. Schuster, W. and B. Axel, Annu. Rev. Plant Physiol. Plant Mol. Biol., 1994, 45, 61–78. 46. Lin, Am. et al., Curr. Sci., 1992, 21, 507–543. 47. Newton, K. J., et al., EMBO J., 1995, 14, 585–593. 48. Malony, A. P. et al., J. Mol. Biol., 1990, 213, 633–649. 49. Makaroff, C. A. and Palmere, J. D, Nucleic Acids Res., 1987, 15, 5141–5156. 50. Tracy, R. L. and Stern, D. B., Curr. Sci., 1995, 28, 205–216. 51. Hanic-Joyce, P. J. et al., J. Biol. Chem., 1990, 265, 13782– 13791. 52. Popp, W. D. et al., EMBO J., 1992, 11, 1065–1077. 53. Finnegan, P. M. and Brown, G. G., Plant Cell, 1990, 2, 71–83. 54. Mulligan, R. M. et al., Md. Cul. Biol., 1991, 11, 533–543. 55. Gualberto, J. M. et al., Plant Cell, 1991, 3, 1109–1120. 56. Saalaoci, E. et al., Plant Sci., 1990, 66, 231–246. 57. Kaleikar, E. K. et al., Curr. Genet., 1992, 22, 463–470. 58. Lin, A. W. et al., Curr. Genet., 1992, 21, 507–513. 59. Laughnan, J. R. and Gabay-Laughnan, S., Annu. Rev. Genet., 1983, 17, 27–48. 60. Forde, B. G. et al., Proc. Natl. Acad. Sci., USA, 1978, 75, 3841–3845. 61. Tang, H. V. et al., Plant J., 1996, 10, 123–133. 62. Dill, C. L. et al., Genetics, 1997, 147, 1367–1379. 63. Singh, M. et al., Genetics, 1996, 143, 505–516. 64. Hanson, M. R. et al., Trends Plant Sci., 1996, 1, 57–61. 65. Maier, R. M. et al., J. Mol. Biol., 1995, 251, 614–628. 66. Gray, M. W. and Covello, P. S., FASEB J., 1993, 7, 64–71. 67. Walbot, V., Trends Genet., 1991, 7, 37–39. 68. Grohmann, L. et al., Nucleic Acids Res., 1994, 22, 3304–3311. 69. Lu, B. and Hanson, M. R., Plant Cell, 1994, 6, 1955–1968. 70. Zamlungo, S. et al., Curr. Genet., 1995, 27, 565–571. 71. Sutton, C. A. et al., Curr. Genet., 1993, 23, 472–476. 72. Srivastava, H. K. and Bowman, I. B. R., Nature, 1972, 235, 152–153. 73. Quinones, V. et al., Plant Physiol., 1995, 108, 1327–1328. 74. Neckermann, K. et al., Gene, 1994, 146, 177–182. 75. Yang, A. J. and Mulligan, R. M., Mol. Cell Biol., 1991, 11, 4278–4281. 76. Sutton, C. A. et al., Mol. Cell Biol., 1991, 11, 4274–4277. 77. Lu, B. and Hanson, M. R., Nucleic Acids Res., 1992, 20, 5699– 5703. 78. Gualbuto, J. M. et al., Md. Genet., 1988, 215,118–127. 79. Freyer, R. et al., Plant J., 1993, 4, 621–629. 80. Kuntz, M. et al., Plant Mol. Biol., 1992, 20, 1185–1188. 81. Bock, R. et al., Mol. Gen. Genet., 1993, 240, 238–244. 82. Srivastava, H. K., Perspective in Cytology and Genetics (eds Manna, G. K. and Roy, S. C., Kalyani), 1998, vol. 9, pp. 33– 47. 83. Navison, H. T. et al., Plant J., 1994, 5, 613–623. 84. Kumar, R. and Levings, C. S., III, Curr. Genet., 1993, 23, 154– 159. 85. Covello, P. S. and Gray, M. W., Trends Genet., 1993, 9, 265– 268. 86. Grivelli, L. A., Eur. J. Biochem., 1989, 182, 477–493. 87. Bailey-Serres, J. et al., Cell, 1986, 47, 567–576. 88. Boeshore, et al., Plant Mol. Biol., 1985, 4, 125–132. 89. Chapdelaine, Y. and Bonen, L., Cell, 1991, 65, 465–471. 90. Bonhomme, S. et al., Curr. Genet., 1991, 19, 121–127. 91. Crouzillat, D. et al., Theor. Appl. Genet., 1987, 74, 773–780. 92. Dewey, R. E. et al., Proc. Natl. Acad. Sci. USA, 1987, 84, 5374–5378. 185 REVIEW ARTICLES 93. Rasmussen, J. and Hanson, M. R., Mol. Can. Conf., 1989, 215, 332–336. 94. Seculella, L. and Palmer, J. D., Nucleic Acids Res., 1988, 16, 3787–3799. 95. Young, E. G. and Hanson, M. R., Cell, 1987, 50, 41–49. 96. Makaroff et al., Plant Mol. Biol., 1990, 15, 735–746. 97. Nivison, H. T. and Hanson, M. R., Raut Cult., 1989, 1, 1121– 1130. 98. Kohler, R. H. et al., Mol. Gen. Genet., 1991, 225, 33–39. 99. Lavet, H. K. et al., Plant J., 1991, 1, 185–193. 100. Levings, III, C. S., Science, 1990, 250, 942–947. 101. Harrison, M. A. et al., Photosynth., 1993, 38, 141–151. 102. Oda, K. et al., J. Mol. Biol., 1992, 223, 1–7. 103. Wolstenholme, D. R. and Fauron, C. M., in The Molecular Biology of Plant Mitochondria (eds Levings C. S. and Vasil, I.), Kluwer, Boston, 1995, pp.1–60. 104. Andre, C. et al., Trends Genet., 1992, 8, 128–132. 105. Oldenburg, D. J. and Bendich, A. J., Plant Cell, 1996, 8, 447– 461. 106. Janska, H. and Mackenzie, S. A., Genetics, 1993, 135, 869– 879. 107. Hatkensson, G. and Glemelius, K., Mol. Gen. Genet., 1991, 229, 380–388. 108. Landgren, M. et al., Plant Mol. Biol., 1996, 32, 879–890. 109. Liu, J. H. et al., Bio Essays, 1996, 18, 465–471. 110. Mohr, S. et al., Theor. Appl. Genet., 1993, 86, 259–268. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. Ogihara, Y. et al., Mol. Gen. Genet., 1997, 255, 45–53. Tsunewaki, K., Jpn. J. Genet., 1993, 68, 1–34. Asakura, N. et al., Genome, 1997, 40, 201–210. Newton, K. J., in The Molecular Biology of Plant Mitochondria (eds Levings, C. S. and Vasil, I.), Kluwer, Boston, 1995, pp. 585–596. Srivastava, H. K. et al., Genetics, 1969, 63, 611–618. Srivastava, H. K. and Sarikrissian, I. V., Genetics, 1970, 66, 497–503. Srivastava, H. K. et al., J. Sustainable Agric., 2000 (in press). Srivastava, H. K., in Perspective in Cytology and Genetics (eds Manna, G. K. and Roy, S. C., Kalyani), 1999, 10, 1–11. Poyton, R. O. and McFwen, J. E., Annu. Rev. Biochem., 1996, 65, 563–607. Hack, E. et al., Curr. Genet., 1994, 25, 73–79. Mulligan, R. M. et al., Proc. Natl. Acad. Sci. USA, 1988, 85, 7998–8002. Moneger et al., EMBO J., 1994, 13, 8–17. Unseld, et al., Nat. Genet., 1997, 15, 57–61. Pruitt, K. D. and Hanson, M. R., Mol. Gen. Genet., 1991, 227, 348–355. Bonhomme, S. et al., Mol. Gen. Genet., 1992, 235, 240–248. Crelon, M. et al., Mol. Gen. Genet., 1994, 243, 540–547 Received 2 February 2000; revised accepted 24 May 2000 Genetic transformation of rice: Current status and future prospects S. Ignacimuthu†,**, S. Arockiasamy † and R. Terada* † Entomology Research Institute, Loyola College, Chennai 600 034, India *National Institute for Basic Biology, Okazaki, Japan Genetic transformation of rice has been an important area of research in the past few years. PEG, electroporation, microprojectile bombardment, and Agrobacterium have been used to mediate gene transfer. Genes for insect resistance, fungal resistance, virus resistance, herbicide resistance, bacterial resistance and nematode resistance have been utilized in rice transformation. The methods involved, the genes utilized and the promoters used in rice transformation along with integration and expression of transgenes are discussed in this review. THE ever-increasing human population especially in the developing countries and various abiotic and biotic stresses have posed a challenge to boost the rice production in limited cultivable land 1,2 . Genetically engineered plants with genes of direct interest can be produced in a relatively short time and can be of direct value in the agri-food industry. Recently it has been **For correspondence. (e-mail: [email protected]) 186 recognized that marker-free transgenics may be more acceptable commercially 3,4 . Some good reviews are already available 5–8. Initially rice was transformed with direct transformation methods such as particle bombardment. Recently it has been shown that Agrobacterium tumefaciens can efficiently transform rice also. In this review, we shall discuss the latest developments in the transformation of rice. We shall focus on advances in gene transfer techniques and we shall also mention specific genes that have recently been introduced into rice plants as well as various issues related to the integration and expression of foreign genes. Methods of gene delivery PEG-mediated rice transformation In 1986, Uchimiya et al. 9 first obtained transgenic calli after polyethylene glycol (PEG)-induced DNA uptake of the nptII gene into root-derived protoplasts, followed CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES by selection on kanamycin. The first transgenic rice plants were recovered by Zhang et al. 10 . Datta et al. 11 published the first report of the recovery of transgenic indica rice plants from cultivar Chinsurah Boro II. In a systematic study of factors that influence the efficiency of PEG-mediated transformation and regeneration, Hayashimoto et al. 12 reported that increasing the PEG concentration from 0 to 30% (w/v) reduced the plating efficiency of Nipponbare protoplasts from 11 to 3% and Taipei-309 protoplasts from 14 to 5%. However, the number of hygromycin-resistant calli obtained increased with increasing concentrations of PEG in both cultivars. The transformation frequency of Taipei-309 protoplasts was also found to increase with increasing concentrations of plasmid DNA up to the levels tested (20 µg DNA per 2–5 × 104, based on the number of protoplasts used). Most reports of successful PEG-mediated transformation and subsequent regeneration of transgenic rice plants have used 20–25% PEG, 10–30 min incubation times, and 10–100 µg of DNA per ml protoplast. Detailed protocols have been described by Hodges et al. 13 and Li et al. 14 . Electroporation-mediated gene transfer Electroporation has also been widely used to introduce naked DNA into protoplasts. Toriyama et al. 15 first used this method for the production of Yamahoushi cultivar transgenic rice via anther culture-derived protoplasts with the aminoglycoside phosphotransferase II (aph(3')II) gene, conferring resistance to the antibiotics kanamycin and G418. Southern analysis showed the correct size fragment upon hybridization to the aph(3')II gene. In another early study, Zhang et al. 10 electroporated protoplasts from cell suspension of Taipei-309 leaf base calli with the 35S promoter fused to the nptII gene. Out of six regenerated plants, two were positive for NPTII activity. Biswas et al. 16 also successfully reported the recovery of transgenics from indica rice IRRI breeding line IR58 protoplasts. Zu and Li 17 obtained fertile transgenic indica rice plants using seed embryo cells. Chaudhury et al. 18 observed transient expression of gus gene in intact seed embryos of indica rice. Microprojectile bombardment-mediated gene transfer Microprojectile bombardment, also called the biolistic method or the particle gun method has been used in many laboratories. The concept has been described in detail by Sanford 19 . The ability to deliver foreign DNA into regenerable cells, tissues or organs appears to provide the best method for achieving genotypeindependent transformation bypassing Agrobacterium host specificity. CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 In earlier experiments, Christou et al. 20 reported that immature embryos were isolated from greenhouse grown cultivars of japonica, javonica and indica rice. The scutellar region of the embryo was bombarded and transient activity for the introduced gene(s) was observed after 24 h of bombardment. Bombarded tissues were plated on regeneration media supplemented with appropriate selective agents; transformed embryos and other transgenic organized tissues, such as shoots and roots in addition to transformed plants were obtained. Sivamani et al. 21 reported a gene transfer procedure for the model indica rice variety TN1, using bombardment of embryogenic callus. They developed a procedure which allowed generation of highly homogenous populations of embryogenic subcultured calli by selectively propagating a small number of regenerationproficient calli derived from seed. An efficiency of 3% could be obtained on an average over a number of experiments. Zhang et al. 22 regenerated transgenic plants from bombarded embryogenic suspensions of elite indica rice varieties, using procedures which are now commonly used by many groups. They commented that this genotype and environment-independent transformation system for indica rice that uses regenerable embryogenic suspensions as targets for bombardment may now be used routinely for the introduction of useful genes into elite rice varieties. Jain et al. 23 optimized the biolistic method for transient gene expression and production of agronomically useful transgenic basmati rice plants. Many biolistic devices (e.g. particle gun) have been developed, including both commercial and labbuilt models. Moreover there is a possibility to get a relatively inexpensive simple particle gun specially for rice-producing developing countries 24 . Agrobacterium-mediated gene transfer The transformation of dicotyledons by Agrobacterium is well established. But in the case of monocots it is not the general process. In the past monocots, particularly graminacious crop plants including important cereals like rice and wheat were considered to be recalcitrant to this technology and they were outside the Agrobacterium host range. However, transformation methods based on the use of Agrobacterium are still preferred in many instances because of the following properties: (i) easy to handle, (ii) higher efficiency, (iii) more predictable pattern of foreign DNA integration, and (iv) low copy number of integration. Raineri et al. 25 described the production of transformed japonica cultivar by cocultivation of mature embryos with Agrobacterium. Results of Southern blotting indicated the integration of the T-DNA into the plant genome, but no transgenic plants were regenerated. In 1992, Chan et al. 26 obtained a few transgenic 187 REVIEW ARTICLES rice plants by inoculating immature embryos with a strain of Agrobacterium. They proved the inheritance of the transformed DNA to progeny plants by Southern blot hybridization, although they analysed the progeny of only one transformed plant. Hiei et al. 7 subsequently reported a method for efficient production of transgenic rice plants from calli of Japonica cultivars that had been cocultivated with A. tumefaciens. Their evidence was based on molecular analysis and genetic studies of a large number of transgenic plants and the analysis of sequence of T-DNA junctions in rice. Rashid et al. 27 reported the successful application of such a method to basmati cultivars of indica rice after only minor modifications. In the same year, Dong et al. 28 also described the successful applications of the method to Javanica rice. Aldemita and Hodges 29 showed that immature embryos were also good starting materials for Agrobacterium-mediated transformation of indica and japonica varieties. Park et al. 30 reported that the isolated shoot apices are suitable explants for successful application of this method. Uze et al. 31 and Toki 32 reported such types of transformation with minor modifications. Abedinia et al. 33 transformed an Australian cultivar by the Agrobacterium-mediated method. Mohanty et al. 34 reported the successful application of this method in an elite indica rice variety pusa basmati 1 and transmission of the transgene to the R 2 progeny was also demonstrated. Khanna and Raina 35 transformed indica rice cultivars by the Agrobacterium-mediated method using binary and super-binary vectors. Chen et al. 36 developed a protocol for consistent and large-scale production of fertile transgenic rice plants. Yokoi et al. 37 produced chilling tolerance of photosynthesis and unsaturation of fatty acids in rice by introducing the GPAT gene. Goto et al. 38 obtained rice seeds with iron fortification by using soybean ferritin gene. Ye et al. 39 succeeded in engineering the provitamin A (b-carotene) biosynthetic pathway into carotenoid-free rice endosperm. Desired genes introduced into rice Insect resistance In the last few years various genes have been introduced into rice, since the methods of transformation have been well established in this species. Majority of the genes that have been introduced into rice are related to disease and insect resistance (Table 1). To our knowledge, there are about eleven reports, in which Bacillus thuringiensis (Bt)-derived genes have been introduced in rice against insect resistance by electroporation 40 , biolistic method 41–49 and by Agrobacterium method 50 . Other genes such as oryzacystatin 51,52 , corn cystatin 53 , PINII 54 , trypsin inhibitor 55,56 , GNA 57,58 have also been added. 188 Fungal resistance Lin et al. 59 reported the production of rice transgenic plants against a rice sheath blight pathogen (fungus) Rhizoctonia solani. Stark-Lorenzen et al. 60 reported the production of transgenic rice plants containing Stilbene synthase gene to enhance resistance to the fungus Pyricularia oryzae using PEG-mediated direct gene transfer. Virus resistance Hayakawa et al. 61 have introduced the coat protein gene of rice stripe virus into two japonica varieties of rice by electroporation of protoplasts. Recently 62 transgenic plants were produced showing the expression of coat protein of rice dwarf virus. Sivamani et al. 63 introduced the coat protein (CP) genes CP1, CP2 and CP3 of rice tungro spherical virus (RTSV) individually or together into indica and/or japonica rice cells by particle bombardment. Herbicide resistance Christou et al. 20 have recovered herbicide-resistant transgenic rice plants from a number of commercially important cultivars, including recalcitrant indica varieties using electric discharge particle acceleration. Further, Cao et al. 64 have reported the production of herbicide-resistant transgenic rice plants, in which the plants were transformed by microprojectile bombardment with plasmids carrying the coding region of the Streptomyces hygroscopicus phosphinothricin acetyle transferase (PAT). In the same year Datta et al. 65 reported the introduction of the bar gene into protoplasts of rice (IR72) by PEG-mediated transformation. Rathore et al. 66 produced herbicide-resistant rice plants by introducing the bar gene into protoplasts derived from immature embryos by PEG-mediated DNA uptake. Recently a successful first rice field trial using a bar gene was reported by Oard et al. 67 . They bombarded the bar gene into immature embryos of commercial cultivars ‘Gulfomont’, ‘IR72’ and ‘Koshihikari’. Sankula et al. 68,69 evaluated the application of Glufosinate on rice transformed with the bar gene in a field trial study. In one instance, introduction of the herbicide-resistant bar gene in rice also provided protection from the sheath blight pathogen, Rhizoctonia solani70. In several studies the bar gene was used as a selectable marker gene30,71 . Bacterial resistance Zhang et al. 72 reported the introduction of the Xa21 gene into embryonic suspensions through the biolistic CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 189 REVIEW ARTICLES 190 CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES method, which confers resistance to Xanthomonas oryzae pv oryzae, a bacterial blight pathogen. Nematode resistance Vain et al. 73 recovered nematode (Meloidogyne incognita)-resistant transgenic plants, containing genes coding for an engineered cysteine proteinase inhibitor (Oryza Cystatin-ID860; OC-I 8D86), using the particle gun method. Other genes Xu et al. 74 introduced the barely LEA3 gene into a japonica rice variety, Nipponbare, using the biolistic method and the transgenic plants showed increased tolerance to both water stress and salinity. Some other genes have been recently introduced in attempts to improve the nutritional and other qualities. Phytoene synthase from daffodil (Narcissus pseudonarcissus) was transferred into immature embryos of a rice model variety (Taipei-309) by microprojectile bombardment. Phytoene was unequivocally identified in several transgenic rice endosperms 75 . Rice mature embryos were subjected to the introduction of ribulose bisphosphate carboxylase gene. The rice plants used nitrogen with enhanced efficiency during photosynthesis at saturating levels of CO 2 and high irrradiance 76 . Ku et al. 77 reported the high level expression (12% of total soluble protein) of maize phosphoenolpyruvate carboxylase (PEPC) gene in rice plants. The PEPC gene, which catalyses the initial fixation of atmospheric CO 2 in C 4 plants was introduced into the C 3 crop rice. Promoters used in rice transformation The success of plant transformation depends very much on promoter sequences 78 . To express the transgenes in plant cells, appropriate promoter sequences have to be introduced alongside the gene to ensure efficient transcription of mRNA 79 . A large number of promoters have been used in rice transformation (Table 2) and several promoter sequences have been isolated from monocots for specific use in cereal species for the efficient expression of the transgenes. The promoter CaMV 35S or its derivatives have been used greatly 7,20,27,28,64,80–83 ; the level of gene expression of this promoter has been reported to vary between different species of plants and different parts of the plant 84 . Further, it has been proved to be useful particularly in dicot species 78 . Unfortunately the levels of gene expression produced by this promoter in cereals are up to hundred-fold less than in dicots 85,86 . A number of monocot promoters (EMU 87,88 , Actin1 30,55,64,89,90 , CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 191 REVIEW ARTICLES Table 2. Constitutive and non-constitutive promoter genes used in rice transformation Promoter 35S Source Gene Pattern of gene expre ssion Cauliflower mosaic virus, 35S RNA transcript All plant tissues EMU Actin 1 Cauliflower mosaic virus Maize Rice Modified maize alcohol dehydrogenase 1 promoter Rice actin 1 gene All tissues All plant tissues Ubi-1 Adh Maize Maize Maize ubiquitin 1 gene Alcohol dehydrogenase II gene His 3 LHCP Wheat Rice rbCS Rice, tomato Pin II Rol C OSg6B OsgrP1 Potato Agrobacterium rihzogenes Rice tungro bacilli form virus Maize and rice Rice Histone 3 gene Light harvesting chlorophyll binding gene of photosystem II Ribulose 1,5-bisphosphate carboxylase/oxygenase small subunit gene Wound inducible II gene Open reading frame 12 (RF12) of the R 1 plasmid T 2 DNA region Rice tungro bacilli form virus, major transcript gene All plant organs Constitute root cap, anther, filament, pollen, scutellum, endosperm, embryo shoot and root, Anerobically induced in roots Dividing root cells Light inducible in leaves, stems and floral organs Light inducible in mesophyll RSs1 PEPC Rice Maize Rice sucrose synthase gene Phosphoenol pyruvate carboxylase gene Bp10 gene GluA-2 GluB-1 GluA-3 RAG-1 GLb-1 NRP33 Ltp2 Brassica Rice Rice Rice Rice Rice Rice Barley Pollen gene Glutelin gene Glutelin 1-gene Glutelin 3-gene Albumin 1-gene Globulin gene (Glb-1) Prolamin (NRB 33) Lipid transfer protein RTBV Glycine-rich gene Glycine-rich cell well protein gene ubiquitin-171,91–93 and Adh1 94,95 ) have also been evaluated to enhance gene expression in cereals. Among these promoters, hpt and EMU hph constructs gave a more dramatic expression than ubiquitin gene and any of the bar containing constructs. The gene for Actin1 and ubiquitin-1 (gene of maize) has been shown to produce highest levels of expression in rice. Other promoters such as His 96 , LHcP 97 , reCS 98 , Pin 54,55 , Rolc 88,99 , RTBV 100 , O8gCB 101 , OsgrP1 102 , RSs1 57,58,103 , PEPc 43,77 , BP1050 , Glu 104 , Ltp 105 have also been used successfully. Integration and expression of transgenes The stable integration, expression and inheritance of transgenes are of great importance in the application of genetically transformed rice to agriculture, but they have not been extensively studied 106 . Only the first generation of transgenic plants (R 0 progeny) were analysed 192 Reference 20, 27, 28, 64, 80–83 87, 88 30, 55, 64, 89, 98 71, 88, 91–93 94, 95 96 97 98 Wound inducible Vascular tissue and embryonic tissue Leaf phloem tissue 54, 55 88, 89 Tapetum specific expression Expressed in young stem, leaves, vascular bundles, epidermis, wound inducible, and in water stress conditions Phloem-specific expression Green tissues, leaves and leaf blades Pollen-specific expression Endosperm specific Endosperm specific Endosperm specific Endosperm specific Endosperm specific Endosperm specific Aleurone-specific gene 101 102 100 57, 58, 103 43, 77 50 104 75, 104 104 104 104 104 105 in many reports. Hiei and Komari 107 analysed the inheritance of transgenes as far as the R 4 generations successfully for eighteen transgenic plants that have been produced by the Agrobacterium-mediated method and stable integration of transgenes in the genome of all progenies was demonstrated. In this study only a small number (fewer than three) of copies of the transgene integrated in a majority of the transformants. Due to this advantage Agrobacterium-mediated transformation is the method of choice. However in another study molecular analysis of genome DNA from engineered plants indicated the presence of vector sequences outside the T-DNA border 108 . Agrobacterium was also found to persist on the surface and within tissues of transformed plants in soil-grown plants up to twelve months following transformation 109 . These are the two potential problems in Agrobacterium-mediated transformation 110 . By contrast, direct transformation technology will result in CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES transgenic plants containing multiple copies of transgenes, complicated pattern of integration 111 and high occurrence of chimeric events 112 . Takano et al. 113 analysed the transgenic plants produced from protoplasts through the calcium phosphate method. One factor that contributed to the complex patterns of integration of foreign genes in transformation is the probable activity of topoisomerase I or II. Because of these illegitimate recombinations, accomplishing rearrangements, large-scale deletions and translocations were found at the sites of integration. Further, recognition sites for topoisomerase were identified in the rearranged sequences involved in the process of gene silencing (expression of transgene and even native gene is suppressed) 114 . The inactivation of gene expression is known as co-suppression, when homologous coding sequences are involved. Silencing of the waxy gene was observed in rice that had been transformed with a cloned waxy gene 115 . This is the first study of silencing of a native gene and a transgene. A similar type of study on silencing of transgenes and 5-azacytidine-mediated reactivation of the transgene have been reported in rice 116 . Silencing of the genes encoding b-glucuronidase in transgenic rice was demonstrated by methylation 117 . Such type of studies probably help to understand the intrinsic mechanisms for the control of gene expression in rice and also the molecular mechanisms of plants. Conclusion It is obvious that rice transformation system through both Agrobacterium and particle bombardment has become a routine procedure in many laboratories for the production of transgenic plants. Transgenic plants containing genes of agronomic interest including insect, herbicide, virus and nematode resistance through direct transformation techniques have been obtained. The next challenge for rice improvement via genetic engineering is to produce viral tungro disease resistance which can help in preventing the annual loss in the range of 1.5 billion US dollars in south and south-east Asia as well as to produce iron-rich rice, including introduction of the apomictic trait which will allow to fix hybrid vigour in the progeny without any further segregation. 1. Beck, C. I. and Ulrich, T. H., Bio/Technology, 1993, 11, 1524– 1528. 2. Swaminathan, M. S., The Hindu Survey of Indian Agriculture, 1999, pp. 9–16. 3. Yoder, J. I. and Goldsbrough, A. P., Bio/Technology, 1994, 12, 263–267. 4. Komari, T., Hiei, Y., Saito, Y., Murai, N. and Kumashiro, T., Plant J., 1996, 10, 165–174. 5. Ayers, N. M. and Park, W. D., Crit. Rev. Plant Sci., 1994, 13, 219–239. 6. Christou, P., Rice Biotechnology and Genetic Engineering, Technomic Pub Co Inc., Lancaster. CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 7. Hiei, Y., Ohta, S., Komari, T. and Kumashiro, T., Plant J., 1994, 6, 271–282. 8. Tyagi, A. K., Mohanty, A., Bajaj, S., Chaudhury, A. and Maheshwari, S. C., Crit. Rev. Biotechnol., 1999, 19, 41–79. 9. Uchimiya, H., Fuskimi, T., Hashimoto, H., Harda, H., Syono, K. and Sugawara, Y., Mol. Gen. Genet., 1986, 16, 204–207. 10. Zhang, H. M., Yang, H., Rech, E. L., Golds, T. J., Davis, A. S., Mulligan, B. J. and Cocking, E. C., Plant Cell Rep., 1988, 7, 379–381. 11. Datta, S. K., Peterhans, A., Datta, K. and Potrykus, I., Bio/ Technology, 1990, 8, 736–740 12. Hayashimoto, A., Li, Z. and Murai, N., Plant Physiol., 1990, 93, 857–863. 13. Hodges, T. K., Peng, J., Lyzink, L. A. and Koetje, P. S., in Rice Biotechnology (eds Khush, G. S. and Toenniessen, G. H.), CAB International, UK/IRRI, Manila, 1990, pp. 157–174. 14. Li, L., Burow, M. D. and Murai, N., Plant Mol. Biol. Rep., 1990, 8, 276–291. 15. Toriyama, K., Arimoto, Y., Uchimiya, H. and Hinata, K., Bio/ Technology, 1988, 6, 1072–1074. 16. Biswas, G. C., Iglesias, V. A., Datta, S. K. and Potrykus, I., J. Biotechnol., 1994, 32, 1–10. 17. Xu, X. and Li, B., Plant Cell Rep., 1994, 13, 237–242. 18. Chaudhury, A., Maheshwari, S. C. and Tyagi, A. K., Plant Cell Rep., 1995, 14, 215–220. 19. Sanford, J. C., Trends Biotechnol., 1988, 6, 299–302. 20. Christou, P., Ford, T. L. and Kofron, M., Bio/Technology, 1991, 9, 957–962. 21. Sivamani, E., Shen, P., Opalka, N., Beachy, R. N. and Fauquet, C. M., Plant Cell Rep., 1996, 15, 322–327. 22. Zhang, S., Chen, L., Qu, R., Marmey, P., Beachy, R. N. and Fauquet, C. M., ibid., 465–469. 23. Jain, R. K., Jain, S., Wang, B. and Wu, R., ibid, 963–968. 24. Sudhakar, D., Duc, L. T., Bong, B. B., Tinjuangjun, P., Maqbool, S. B., Valdez, M., Jefferson, R. and Christou, P., Transgenic Res., 1998, 7, 289–294. 25. Raineri, D. M., Bottino, P., Gordon, M. P. and Nester, E. W., Bio/Technology, 1990, 8, 33–38. 26. Chan, M. T., Lee, T. M. and Chang, H. H., Plant Cell. Physiol., 1992, 33, 577–583. 27. Rashid, H., Yokoi, S., Toriyama, K. and Hinata, K., Plant Cell Rep., 1996, 15, 727–730. 28. Dong, J., Teng, W., Buchholz, W. G. and Hall, T. C., Mol. Breed., 1996, 2, 267–276. 29. Aldemita, R. R. and Hodges, T. K., Planta, 1996, 199, 612– 617. 30. Park, S. H., Pinson, S. R. M. and Smith, R. H., Plant Mol. Biol., 1996, 32, 1135–1148. 31. Uze, M., Wunn, J., Kaerlas, J. P., Potrykus, I. and Sautter, C., Plant Sci., 1997, 130, 87–95. 32. Toki, S., Plant Mol. Biol. Rep., 1997, 15, 16–21. 33. Abedinia, M., Henry, R. J., Blakeney, A. B. and Lewin, L., Aust. J. Plant Physiol., 1997, 24, 133–141. 34. Mohanty, A., Sharma, N. P. and Tyagi, A. K., Plant Sci., 1999, 147, 127–137. 35. Khanna, H. K. and Raina, S. K., Aust. J. Plant Physiol., 1999, 26, 311–324. 36. Chen, L., Marmey, P., Taylor, N. J., Brizard, J. P., D’Cruz, P., Huet, H., Zhang, S., de Kochko, A., Beachy, R. N. and Fauquet, C. M., Nat. Biotechnol., 1998, 16, 1060–1064. 37. Yokoi, S., Higashi, S-I., Kishitani, S., Murata, N. and Toriyama, K., Mol. Breed., 1998, 4, 269–275. 38. Goto, F., Yoshihara, T., Shigemoto, N., Toki, S. and Takaiwa, F., Nat. Biotechnol., 1999, 17, 282–286. 39. Ye, X., Al-Babili, S., Kloti, A., Zhang, J., Lucca, P., Beyer, P. and Potrykus, I., Science, 2000, 287, 303–305. 193 REVIEW ARTICLES 40. Fujimoto, H., Itoh, K., Yamamoto, M., Kyozuka, J. and Shimamoto, K., Bio/Technology, 1993, 11, 1151–1155. 41. Wunn, J., Kloti, A., Burkhardt, P. K., Gosh Biswas, G. C., Launi, S. K., Iglesia, V. A. and Potrykus, I., Bio/Technology, 1996, 14, 171–176. 42. Nayak, P., Basu, D., Das., S., Basu, A., Ghosh, D., Ramakrishnan, N. A., Ghosh, M. and Sen, S. K., Proc. Natl. Acad. Sci. USA, 1997, 94, 2111–2116. 43. Ghareyazie, B., Alinia, F., Menguito, C. A., Rubia, L. G., De Palma, J. M., Liwanag, E. A., Cohen, M. B., Khush, G. S. and Bennet, J., Mol. Breed., 1997, 3, 401–414. 44. Wu, C., Fan, Y., Zhang, C., Oliva, N. and Datta, S. K., Plant Cell Rep., 1997, 17, 129–132. 45. Alam, M. F., Datta, K., Abrigo, E., Vasquez, A., Senadhira, D. and Datta, S. K., Plant Sci., 1998, 135, 25–30. 46. Datta, K., Vasquez, A., Tu, J., Torrizo, L., Alam, M. F., Oliva, N., Abrigo, E., Khush, G. S. and Datta, S. K., Theor. Appl. Genet., 1998, 97, 20–30. 47. Tu, J., Datta, K., Alam, M. F., Fan, Y., Khush, G. S. and Datta, S. K., Plant Biotechnol., 1998, 15, 195–203. 48. Maqbool, S. B., Husnain, T., Riazuddin, S., Masson, L. and Christou, P., Mol. Breed., 1998, 4, 501–507. 49. Alam, M. F., Datta, K., Abrigo, E., Oliva, N., Tu, J., Viramani, S. S. and Datta, S. K., Plant Cell Rep., 1999, 18, 572–575. 50. Cheng, X., Sardana, R., Kaplan, H. and Altosaar, I., Proc. Natl. Acad. Sci. USA, 1998, 95, 2767–2772. 51. Hosoyama, H., Irie, K., Abe, K. and Arai, S., Biosci. Biotechnol. Biochem., 1994, 58, 1500–1505. 52. Hosoyama, H., Irie, K., Abe, K. and Arai, S., Plant Cell Rep., 1995, 15, 174–177. 53. Irie, K., Hosoyama, H., Takeuchi, T., Iwabuchi, K., Watanabe, H., Abe, M., Abe, K. and Arai, S., Plant Mol. Biol., 1996, 30, 149–159. 54. Duan, X., Li, X., Xue, Q., Abo-EI-Saad, M., Xu, D. and Wu, R., Nat. Biotechnol., 1996, 14, 494–498. 55. Xu, D., Xue, Q., McElroy, D., Mawal, Y., Hilder, V. A. and Wu, R., Mol. Breed., 1996, 2, 167–173. 56. Lee, S. I., Lee, S. H., Koo, J. C., Chun, H. J., Lim, C. O., Mun, J. H., Song, Y. H. and Cho, M. J., Mol. Breed., 1999, 5, 1–9. 57. Rao, K. V., Rathore, K. S., Hodges, T. K., Fu, X., Stoger, E., Sudhakar, D., Williams, S., Christou, P., Bharathi, M., Brown, D. P., Powel, K. S., Spence, J., Gatehouse, A. M. R. and Gatehouse, J. A., Plant J., 1998, 15, 469–477. 58. Sudhakar, D., Fu, X., Stoger, E., Williams, S., Gatehouse, J. A. and Christou, P., Transgenic Res., 1998, 7, 371–378. 59. Lin, W., Anuratha, C. S., Datta, K., Potrykus, I., Muthukrishnan, S. and Datta, S. K., Bio/Technology, 1995, 13, 686–690. 60. Stark-Lorenzen, P., Nelke, B., Hanbler, G., Muhehock, H. P. and Thomzik, J. E., Plant Cell Rep., 1997, 16, 668–673. 61. Hayakawa, T., Zhu, Y., Itoh, K., Kimura, Y., Izawa, T., Shimamoto, K. and Toriyama, S., Proc. Natl. Acad. Sci. USA, 1992, 89, 9865–9869. 62. Zheng, H. H., Li, Y., Yu, Z. H., Li, W., Chen, M. Y., Ming, X. T., Casper, R. and Chen, X. L., Theor. Appl. Genet., 1997, 94, 522–527. 63. Sivamani, E., Hult, H., Shen, P., Ong, C. A., De Kochko, A., Fauquet, C. and Beachy, R. N., Mol. Breed., 1999, 5, 177– 185. 64. Cao, J., Duan, X. L., McElroy, D. and Wu, R., Plant Cell Rep., 1992, 11, 586–591. 65. Datta, S. K., Datta, K., Soltanifar, N., Donn, G. and Potrykus, I., Plant Mol. Biol., 1992, 20, 619–629. 66. Rathore, K. S., Chowdhury, V. K. and Hodges, T. K., Plant Mol. Biol., 1993, 21, 871–884. 67. Oard, J. H., Linscombe, S. D., Braverman, M. P., Jodari, T., Bollich, P. K., Blouin, D. C., Cooley, T. C., Leech, M., Kholi, A., Vain, P. and Christou, P., Mol. Breed., 1996, 2, 359–368. 194 68. Sankula, S., Braverman, M. P., Jodari, F., Steven, D., Linscombe, S. D. and Oard, J. H., Weed Technol., 1997, 11, 70– 75. 69. Sankula, S., Braverman, M. P., Jodari, F., Linscombe, S. D. and Oard, J. H., Weed Technol., 1997, 11, 303–307. 70. Uchimiya, H., Iwata, M., Nojiri, C., Samarajeewa, P. K., Takamatsu, S., Ooba, S., Anzai, H., Christensen, A. H., Quail, P. H. and Toki, S., Bio/Technology, 1993, 11, 835–836. 71. Toki, S., Takamatsu, S., Nojiri, C., Ooba, S., Anzai, H., Iwata, M., Christensen, A. H., Quail, P. H. and Uchimiya, H., Plant Physiol., 1992, 100, 1503–1507. 72. Zhang, S., Song, W. Y., Chen, L., Ruan, D., Taylor, N., Ronald, P., Beachy, R. and Fanquet, L., Mol. Breed., 1998, 4, 551– 558. 73. Vain, P., Worland, B., Clarke, M. C., Richard, G., Beavis, M., Liu, H., Kohli, A., Leach, M., Snape, J., Christou, P. and Atkinson, H., Theor. Appl. Genet., 1998, 96, 266–271. 74. Xu, D., Duan, X., Wang, B., Hong, B., Ho, T-H. D. and Wu, R., Plant Physiol., 1996, 110, 249–257. 75. Burkhardt, P. K., Beyer, P., Wunn, J., Kloti, A., Armstrong, G. A., Schledz, M., Linting, J. V. and Potrykus, I., Plant J., 1997, 11, 1071–1078. 76. Makino, A., Shimada, T., Takumi, S., Kaneko, K., Matsuoka, M., Shimamoto, K., Nakano, H., Miyao-Tokutomi, M., Mae, T. and Yamamoto, N., Plant Physiol., 1997, 114, 483–491. 77. Ku, M. S. B., Agarie, S., Nomura, M., Fukayama, H., Tsuchida, H., Ono, K., Hirose, S., Toki, S., Miyao, M. and Matsuoka, M., Nat. Biotechnol., 1999, 17, 76–80. 78. Vasil, I. K., Plant Mol. Biol., 1994, 25, 925–937. 79. Finch, R. P., in Molecular Biology in Crop Protection (eds Marshall, G. and Walters, D.), Chapman and Hall, 1994, pp. 1– 37. 80. Shimamoto, K., Terada, R., Izawa, T. and Fujimoto, H., Nature, 1989, 338, 274–276. 81. Li, L., Qu, R., Kothko, A. D., Fauquest, C. and Beachy, R. N., Plant Cell Rep., 1993, 12, 250–255. 82. Battraw, M. J. and Hall, T. C., Plant Mol. Biol., 1990, 15, 527– 538. 83. Meijer, E. G. M., Schilperoort, R. A., Rueb, O. S., Rugrot, P. E. and Hensgens, L. A. M., Plant Mol. Biol., 1991, 16, 807– 820. 84. Nillson, O., Little, C. H. A., Sandberg, G. and Ollson, O., Plant Mol. Biol., 1996, 31, 887–895. 85. Hauptmann, R. M., Ashraf, M., Vasil, V., Hannah, L. C., Vasil, I. K. and Ferl, R., Plant Physiol., 1988, 88, 1063–1066. 86. Fromm, M. E., Taylor, L. P. and Walbot, V., Proc. Natl. Acad. Sci. USA, 1985, 882, 5824–5828. 87. Chamberlain, D. A., Brettell, R. I. S., Last, D. I., Witrzens, B., McElroy, D., Dolferus, R. and Dennis, E. S., Aust. J. Plant Physiol., 1994, 21, 95–112. 88. Li, Z., Upadhyaya, N. M., Meena, S., Gibbs, A. J. and Waterhouse, P. M., Mol. Breed., 1997, 3, 1–14. 89. Zhang, W., McElroy, D. and Wu, R., Plant Cell, 1991, 3, 1155–1165. 90. McElroy, D., Blowers, A. D., Jenes, B. and Wu, R., Mol. Gen. Genet., 1991, 231, 150–160. 91. Cronejo, M. J., Luth, D., Blankenship, K. M., Anderson, O. D. and Blechl, A. E., Plant Mol. Biol., 1993, 23, 567–581. 92. Nagatani, N., Takumi, S., Tomiyama, M., Shimada, T. and Tamiya, E., Plant Sci., 1997, 124, 49–56. 93. Christensen, A. H., Sharrock, R. A. and Quail, P. H., Plant Mol. Biol., 1992, 18, 675–689. 94. Zhang, W. and Wu, R., Theor. Appl. Genet., 1988, 7, 835–840. 95. Kyozuka, J., Izawa, T., Nakajima., M. and Shimamoto, K., Maydica, 1990, 35, 353–357. 96. Terada, R., Nakamura, T., Iwabuchi, M. and Shimamoto, K., Plant J., 1993, 3, 241–252. CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 REVIEW ARTICLES 97. Tada, Y., Sakamoto, M., Matsuoka, M. and Fugimura, T., EMBO J., 1991, 10, 1803–1808. 98. Kyozuka, S., McElroy, J., Hayakawa, T., Xie, Y., Wu, R. and Shimamoto, K., Plant Physiol., 1993, 102, 991–1000. 99. Matsuki, R., Onodera, H., Yamauchi, T. and Uchimiya, H., Mol. Gen. Genet., 1989, 220, 12–16. 100. Bhattacharyya-Prakasi, M., Peng, J., Elmer, J. S., Laco, G., Shen, P., Kanuewska, M. B., Kononowicz, H., Wen, F., Hodges, T. K. and Beachy, R. N., Plant J., 1993, 4, 71–79. 101. Yokoi, S., Tsuchiya, T., Toriyama, K. and Hinata, K., Plant Cell Rep., 1997, 16, 363–367. 102. Xu, D., Li, M. and Wu, R., Plant Mol. Biol., 1995, 28, 455– 471. 103. Boulter, D., Phytochemistry, 1993, 34, 1453–1466. 104. Wu, C-Y., Adachi, T., Hatan, T., Washida, H., Suzuki, A. and Takaiwa, F., Plant Cell Physiol., 1998, 39, 885–889. 105. Kalla, R., Shimamoto, K., Potter, R., Nielsen, P. S., Liwnestad, C. and Olsen, O. A., Plant J., 1994, 6, 849–860. 106. Hiei, Y., Komari, T. and Kubo, T., Plant Mol. Biol., 1997, 35, 205–218. 107. Hiei, Y. and Komari, T., Proceedings of the third international Rice Genetics Symposium, Manila (ed. Khush G. S.), International Rice Research Institute, Philippines, 1996, pp. 131–142. 108. Ramanathan, V. and Veluthambi, K., Plant. Mol. Biol., 1995, 28, 1149–1154. 109. Matzik, A., Mantell, S. and Schiemann, J., Plant Mol. Microb. Interact., 1996, 9, 3781. 110. Christou, P., Plant Mol. Biol., 1997, 35, 197–203. 111. De Block, M., Debrouwer, D. and Moens, T., Theor. Appl. Genet., 1997, 95, 125–131. 112. Solis, R., Yoshida, S., Mori, N. and Nakamura, C., Plant Biotechnol., 1998, 15, 87–93. 113. Takano, M., Egawa, H., Ikoda, J. E. and Wakasa, K., Plant J., 1997, 11, 353–361. 114. Matzke, M. A. and Matzke, A. J. M., Plant Physiol., 1995, 107, 679–685. 115. Itoh, K., Nakajima, M. and Shimamoto, K., Mol. Gen. Genet., 1997, 255, 351–358. 116. Kumpatla, S. P., Teng, W., Buchholz, W. G. and Hall, T. C., Plant Physiol., 1997, 117, 361–373. 117. Kohli, A., Ghareyazic, B., Kim, H. S., Khush, G. S. and Bennet, J., in Proceedings of the 3rd International Rice Genetics Symposium, Manila (ed. Khush, G. S.), International Rice Research Institute, Losbanos, Philippines, 1996, pp. 825–828. Received 23 February 2000; revised accepted 10 May 2000 RESEARCH ARTICLES Resolution of the nonlocality puzzle in the EPR paradox C. S. Unnikrishnan Gravitation Group, Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400 005, India and NAPP Group, Indian Institute of Astrophysics, Koramangala, Bangalore 560 034, India Correlations of quantum-entangled multiparticle systems have been widely discussed in the context of the Einstein–Podolsky–Rosen (EPR) paradox. Bell’s theorem prohibits a local realistic description of these correlations. The standard quantum mechanical derivation as well as the interpretation of the experiments suggest that a mysterious nonlocality is a basic feature of these correlations. We show that the correlations of space-like separated entangled particles can be reproduced starting from local probability amplitudes. The use of complex number amplitudes circumvents the widely discussed Bell’s theorem. The result implies no-collapse-at-a-distance and resolves the EPR puzzle. THE Einstein–Podolsky–Rosen (EPR) nonlocality puzzle is one of the most discussed fundamental problems in physics 1–4. EPR considered a two-particle quantum e-mail: [email protected] CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 system entangled and correlated in position and momentum and argued that under the assumption of locality and a reasonable definition of physical reality, the quantum mechanical description is incomplete. The conclusion of incompleteness was based on their reasoning that suggested the need for simultaneous existence of definite values for noncommuting observables for the individual particles before a measurement was performed. The EPR analysis was motivated by the desire to assign an objective reality to measurable properties in the microscopic world independent of the observer or apparatus. The Copenhagen interpretation of quantum mechanics rejected any such objective reality. EPR proposed that if the value of an observable could be predicted with certainty without the disturbance of a measurement, then there was an element of physical reality associated with that observable for the system. There is no way to probe experimentally such a physical reality for the single quantum, for example the reality of the spin component in a specific direction for a single 195 RESEARCH ARTICLES photon or an electron. But if multi-particle correlated systems are considered, like two spin- 12 particles propagating out from an initially spin-zero state, then it is possible to explore the consequences of assuming objective reality to observables. Bell’s theorem David Bohm transcribed the EPR argument to the spin variables by considering the singlet entangled state of two spin- 12 particles 5. This state is described by the wave function ψ S = 1 2 {| 1,− 1〉− | − 1,1〉}, (1) where the state |1, –1〉 is a short form for |1〉1 |–1〉2, and represents an eigenvalue of +1 for the first particle and –1 for the second particle if measured in any particular direction. y S is inherently nonlocal, describing both particles together, even when they are far apart in space-like separated regions. It is a superposition of two-particle product states and there is no specific state assigned to any of the particles individually. A measurement on one particle changes the whole wave function, collapsing the states of both particles to definite values. If the spin component is measured in any direction on one of the particles, the same component can be predicted with certainty for the other companion particle using the conservation law that the total spin is zero. The EPR definition of reality then assigns objective reality before measurement to this spin component. Since the decision as to which component is to be measured is arbitrary and could be delayed till the particles are far apart, the assumption of locality – that one measurement should not influence the result of the other – implies that there should be physical reality for all the three components of the spin for the two particles individually. This is not allowed by quantum mechanics since the three spin components are mutually noncommuting. According to the EPR argument, this conclusion suggested the plausibility of a better theory, possibly with additional hidden variables that would assign specific values to the observables in each run of an experiment such that the statistical predictions of quantum mechanics are reproduced when averaged over the distribution of the hidden variables. Such a theory is a local (realistic) hidden variable theory. John Bell analysed the EPR problem in the early sixties and established the Bell’s inequalities obeyed by any local hidden variable theory of correlations of entangled particles 3,4,6 . He considered the measurement of the spin components on the two particles in two differ196 ent directions, in contrast to the EPR analysis that involved measurements in the same direction (see Figure 1). The result of such a measurement is two-valued in any direction. If A and B denote the outcomes +1 or –1 (written as + or –) and a and b denote the settings of the analyser or the measurement apparatus for the first particle and the second particle, respectively, then the statement of locality is that A(a) = ± 1, B(b) = ± 1. (2) The outcome for the first particle is decided by the local setting a for the analyser with which the first particle interacts, and for the second particle the outcome is decided by the local setting b. The statement of existence of reality is that some hidden variable decides the outcomes even before the measurement, or equivalently the outcome has an objective existence even if the measurement is not actually carried out. This is encoded as A(a,h1) = ± 1, B(b, h2) = ±1, (3) where h1 and h2 are hidden variables associated with the outcomes. In an experiment involving the kind of measurement described, the experimenter can calculate a correlation function of the outcomes defined by P (a, b) = 1 N ∑ ( A B ). i i (4) This is a classical correlation function obtained by averaging the products of the form (++ = +), (– – = +), (+– = –) and (– + = –). Since the joint events (++) and (– –) are coincidences and the events (+ –) and (– +) are anticoincidences (‘coincidence’ denotes both particles showing the same value for the measurement and ‘anticoincidence’ denotes those with opposite values), P(a, b) denotes the average of the quantity (number of detections in coincidence – number of detections in Figure 1. Schematic diagram of spin correlation measurements on entangled particles. The outputs of the two analysers are +1 or –1 and these are correlated in a coincidence unit. a and b are the two analyser directions. Dotted arrow on the right analyser shows the direction a for reference. CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 RESEARCH ARTICLES anticoincidence). The task of the theory is to calculate this function starting from suitable basic ingredients. Bell chose to calculate this correlation by multiplying A(a, h1) and B(b, h2) and integrating over the distribution of the hidden variable h, since the assumption of realism demanded that the outcomes were ‘there’ even before the measurement was made. The Bell correlation is then ∫dhr(h)A(a, h1)B(b, h2), where ∫dhr(h) = 1. The essence of Bell’s theorem is that the function P(a, b) has distinctly different dependences on the relative angle between the polarizers for a local hidden variable description and for quantum mechanics. The correlation predicted by a local realistic theory is bounded by the Bell’s inequalities. The magnitude of a particular sum of correlations P(a, b) for different combinations of a and b is bounded by the value 2, and the same combination calculated in quantum mechanics using the entangled wave function and spin operators exceeds this bound, violating the inequality 3,6 . Also, experimentally measured correlations agree with the quantum mechanical predictions and violate the inequality 4,7 . The concept of local realism is not tenable in the Bell framework. These results have been interpreted as evidence for nonlocal influences across spacelike separated events. The measurement of an observable on one of the particles causes the other particle to acquire a definite state consistent with the relevant conservation law. The standard quantum mechanical derivation of these correlations employs the nonlocal multiparticle wave function, and measurement on one particle is said to collapse the state of the companion particle to a definite eigenvalue. Though this nonlocal feature cannot be used for superluminal signal communication 8, there is a conflict with the spirit of relativity 9. Also, such superluminal influences in the microscopic physical world are bizarre and at present beyond any understanding in terms of any physical mechanism. The nonlocality puzzle can be resolved however, if the correct physical input is used for the calculation of the quantum correlations. The correct correlations can be reproduced if we start with the description of the relevant physical phenomenon, like the passage of the quantum particle through a polarizer, using local probability amplitudes 10 . In the local hidden variable theories of the form Bell considered, the correlations are calculated from eigenvalues and this procedure does not preserve the phase information and wave-like characteristics of the quantum system. The situation has some analogy to the description of interference in quantum mechanics. Any attempt to reproduce the interference pattern using locality and the information on ‘whichpath’ will fail since the phase information is lost or modified in such an attempt. We calculate the correlation from the local amplitudes instead of from the eigenvalues. The wave nature of quantum systems is then explicitly used in the calculation of the correlations and CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 the final probabilities are calculated by squaring a suitable inner product of the local amplitudes. (It is the use of complex functions with a phase that is interpreted as the ‘wave nature’ in this paper.) The realism in this theory is at a deeper level, concealed as a phase and not as actual outcomes before a measurement is made. This possibility was not considered in earlier local realistic theories, and turns out to be the crucial concept required to resolve the EPR puzzle. A new calculus for correlations Consider the breaking up of a maximally entangled state – photons entangled in orthogonal polarization states or spin- 12 particles entangled in (up, down) state – as in the standard Bohm version of the EPR problem 4,5 . The two particles go off in opposite directions and are in space-like separated regions. Two observers make measurements on these particles individually with time stamps such that these results can be correlated later (Figure 1). We assume that strict locality is valid. Measurement on one particle does not change either the magnitude or phase of the complex amplitude associated with the companion particle. In particular, measurement on one particle does not cause the companion particle to acquire a definite state. At each location the result of a measurement is twovalued, denoted by (+) and (–) for each particle (two mutually exclusive outcomes), for any angle of orientation of the analyser. We prescribe the local rules or amplitudes for the transmission through an analyser for these particles (we use the terms polarizer and analyser in a generic way. They could be Stern–Gerlach-like analysers for spin- 12 particles or polarizers for photons). The local amplitudes for events + and – for the two particles individually are denoted by the complex functions C1+ , C1–, C2+ and C2–. Amplitudes C1+ and C1– are mutually orthogonal and similarly C2+ and C2– are mutually orthogonal. The statement of locality is at the level of these amplitudes, and can be written as C1± = C1± (a, f1), C2± = C2± (b, f2), (5) where f1 and f2 are the ‘hidden variables’. These are internal variables associated with the individual particles and appear in the amplitudes as a phase. These can be thought of as the initial undetermined phase, associated with the spin of the individual quantum particles. A definite value for these variables does not imply a definite state for the particles before the measurement. The locality assumption also implies that A(a, f1) = ±1, B(b, f2) = ±1. (6) But, this has a meaning different from the one in the standard local realistic theories. Here, it means that the 197 RESEARCH ARTICLES outcomes, when measured, depend only on the local setting and the local internal variable. In this framework the correlation function is not P(a,b) = (1/N)∑ (Ai Bi ) or ∫dhr(h)A(a, h1)B(b, h2). We calculate the correlations from the local amplitudes. These correlations are of the form U(a, b) = Real(NC i Cj*), where N is a normalization factor. It is the square of this correlation function that would give a joint probability. This new calculus of local amplitudes ensures that all the probabilities are positive definite. (There are local theories which reproduce the correct correlations using probability distribution of the hidden variables that are not positive definite. These theories are neither rooted in quantum concepts nor in classical concepts, and are really ‘out of this world’.) The correlation function is analogous to the two-point amplitude correlations of two independent classical electromagnetic fields. The expression can be generalized to situations where there are more than two particles. Now that we have outlined the general scheme and assumptions as well as the point of departure in calculating the correlation, let us consider the maximally entangled singlet system described by eq. (1), the most widely discussed example in the context of nonlocality. We prescribe the local amplitudes as C1+ = 1 exp{is(q1 – f1)} for the first particle at the 2 first polarizer and C2+ = 1 2 exp{is(q2 – f2)} for the second particle at the second polarizer. There are corresponding amplitudes, C1– and C2– for the events denoted by –, and they differ only in the phase for the maximally entangled state. In these amplitudes, q1 and q2 are the directions of the two polarizers, and s is the spin of the particle (1 for photons and 12 for the spin- 12 particles). The explicit dependence of the amplitude on the spin of the particle is motivated by the fact that we are dealing with systems with phases and the phase associated with the spin rotations (a geometric phase) is a necessary input in this description 11 . The correlation at source is encoded in the relative value or the difference f0 of the internal variables. f0 is a constant for all the entangled pairs. The locality assumption is strictly enforced since the two amplitudes depend only on local variables and on an internal variable generated at the source and then individually carried by the particles without any subsequent interaction of any sort. The individual measurements at each end will now separately give the correct result for transmission for any angle of orientation. These probabilities are C 1 C 1 * = C 2C 2 * = 1 . 2 (7) Events of both types (++) and (– –) contribute to a ‘coincidence’. The correlation function for an outcome of 198 either (++) or (– –) of two maximally entangled particles is U(q1, q2, f0) = 2Re(C1C2*) = cos{s(q1 – q2) + sf0}. (8) It is normalized such that its square will give the conditional joint probabilities of the type ‘outcome + for the second particle, given that the outcome for the first particle is +’, etc. All references to the individual values of the internal variable f have dropped out. We now derive the relation between this correlation function and the experimenter’s correlation function P(a, b) = (1/N)å(Ai Bi ). Since U2++ = U2–– for the maximally entangled state, U2(q1, q2, f0) is the probability for a coincidence detection (++ or ––), and (1 – U2(q1, q2, f0)) is the probability for an anticoincidence (events of the type +– and –+). Since the average of the quantity (number of coincidences – number of anticoincidences) is U2(q1, q2, f0) – (1 – U2(q1, q2, f0)) = 2U2(q1, q2, f0) – 1, (9) the correspondence between P(a, b) and U(q1, q2, f0) is given by the expression, P(a, b) = 2U2(q1, q2, f0) – 1 = 2cos 2{s(q1 – q2) + sf0} – 1. (10) Examples and applications Singlet spin- 12 particles and photons Consider the singlet state breaking up into two spin- 12 particles propagating in opposite directions to spatially separated regions. Since the total spin is zero in any basis we set q0 = p. Then the correlation function and P(a, b) calculated from this function are U(q1, q2, f0) = cos{s(q1 – q2) + sf0} 1 = cos (θ1 − θ2 ) + π / 2 2 1 2 = − sin (θ1 − θ2 ), (11) 1 P (a, b) = 2 sin 2 (θ1 − θ2 ) − 1 2 = – cos(q1 – q2) = –a×b. (12) This is identical to the quantum mechanical prediction obtained from the singlet entangled state and the Pauli CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 RESEARCH ARTICLES spin operators 4. We have reproduced the correct correlation function using local amplitudes. For the case of photons entangled in orthogonal polarization states we get, by setting s = 1 and f0 = p/2 to represent orthogonal polarization, U(q1, q2, f0) = cos{(q1 – q2) + p/2} = – sin(q1 – q2), (13) P(a, b) = 2sin2(q1 – q2) – 1 = – cos(2(q1 – q2)), (14) which is the correct quantum mechanical correlation. Two-particle interferometry The cases of particles entangled in other sets of variables like momentum and position, and energy and time can be mapped onto the spin problem with two-valued outcomes and the local amplitudes reproduce the correct correlations. Consider a pair of EPR-entangled particles which are propagating in opposite directions. Each particle encounters a double slit arrangement (Figure 2) on their way12 . It is easy to see that there will not be any single particle interference pattern in this case on either screen. Near-perfect momentum correlation implies an extended source since a small source leads to uncontrollable uncertainties in momentum. Then the spatial coherence is not sufficient to produce the single particle interference pattern. But there is a two-particle interference pattern observable in the coincidence in two detectors, which could be in space-like separated regions. Usually the results in multiparticle interferometry are interpreted as evidence of bizarre nonlocality 12 since the pattern depends on the difference of the coordinates of both the detectors, just as the spin correlations depend on the difference in settings of the two analysers. In the experiment, particles pass the slit planes and are detected with two counting detectors, one on each side. As in the case of spins, we assume the existence of an internal variable, denoted by x0, and we assume that the two correlated particles have the same value for the internal variable. For the spin- 12 case, the relative rota- Figure 2. Two-particle dual double slit experiment. S(A) and S(B) are the double slits and D(A) and D(B) are the movable detectors. The outputs of the detectors are correlated as in Figure 1. CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 tion required to go from a maximum of transmission through the analyser to a minimum is p. In the case of double-slit interference, the situation is similar. The relative rotation in phase from the center of the bright fringe to the centre of the dark fringe is p, for the angular variable defined by q = akx, where k = 2p/l, and x is the coordinate of the detector along the fringe pattern. a is a scale factor representing the distance of the slits from the source and the detectors. So, the double-slit interference problem can be mapped to the spin- 12 singlet problem. The corresponding local amplitudes are C1 = 1 2 exp(iαk ( x1 − x0 ) / 2), and C2 = 1 2 exp(iαk ( x2 − x0 ) / 2), (15) where x1 and x2 are the detector positions. The single particle detections on either side separately do not show any interference. The correlation function is U(x1, x2) = cos(ak(x1 – x2)/2). (16) Probability for coincidence detection is P(x1, x2) = cos 2(ak(x1, x2)/2) 1 = (1 + cos kα ( x1 − x2 ) ). 2 (17) This is the two-photon interference pattern (it is more appropriate to call it a two-photon correlation pattern) with 100% visibility. The photon which is being detected at one detector has no nonlocal influence on the photon detected at the other detector. Three-particle GHZ state correlations So far we have been discussing examples in which the statistical predictions of quantum mechanics were compared with the prediction from a local theory. There have been examples where a single measurement of a perfect correlation, assuming perfect detectors, etc. is enough to demonstrate the conflict between quantum mechanics and a local realistic theory 13,14 . Experiments are yet to be done for these cases. We now show that the theory with local amplitudes and the spin internal variable deals with the correlations in such cases in a remarkably simple and transparent way. Consider the three-particle GHZ state 13 defined as |ψ GHZ 〉 = 1 2 (| 1,1,1〉− | 1,− 1,− 1〉), (18) 199 RESEARCH ARTICLES where the eigenvalues in the kets are with respect to the z-axis basis. The conflict between a local realistic theory and quantum mechanics is the following statement: The prediction from quantum mechanics for the measurement of the x-component of spin on all the three particles represented by the operator σ 1x ⊗ σ x2 ⊗ σ x3 is given by σ 1x ⊗ σ x2 ⊗ σ x3 | ψ GHZ 〉 = − |ψ GHZ 〉. (19) Local realistic theories predict 13 that the product of the outcomes in the x direction for the three particles should be +1. This contradicts eq. (19). We now show that the correct correlations can be reproduced using local amplitudes. The general idea is that the three-particle correlation, analogous to our scheme for two-particle states, is the real part of a complex number Z obtained as a suitable product of three complex amplitudes. We choose the different phases such that the correlation represented by N(Z–,–,–) is ±1, to satisfy the condition that the joint probability for the outcome (–, –, –) is unity according to eq. (19). N is a normalization factor. For this we choose (Z–, –, –) to be pure real. The rest of the correlations follow without any additional input since flipping the sign once (for example (Z+, –, –)) amounts to rotating Z through the phase p/2, and the corresponding probability P(+,–,–) = Real(Z+,–,–)) 2 = 0. This is because the amplitudes for + and – are orthogonal. Clearly the joint probabilities are unity for the outcomes containing an odd number of (–), and zero for all other outcomes. This result is independent of the definition (once a definition is chosen the phases can be chosen to get the desired outcomes) of the complex correlation Z, though for convenience the correlation can be defined as NReal(C1C2*C3*). which an EPR effect has been discussed 16 . The analysis using local amplitudes shows that the probability for both the particles in the correlated beam decaying into the same mode, say p0p0, at equal proper time is zero without the nonlocal ‘EPR effect’. Entanglement through measurement The general framework presented here is suited also for analysing entanglement correlations between particles which have not interacted in the past, but get apparently entangled due to joint measurements on their companion particles 17 . If one considers two ensembles of entangled particle pairs (1, 2) and (3, 4) such that (1, 2) are entangled and (3, 4) are entangled, but (2, 3) or (1, 4) are not entangled, it is possible to make entanglement correlations between (1, 4) for example, by making Bell-type joint measurements on the pair (2, 3). In a Bell-state measurement, particles 2 and 3 are made to interfere on a beam-splitter and then it is possible to pick out joint states of the type | ψ 2,3 〉± = 1 (| 1, − 1〉 ±| − 1, 1〉). Though there is no prior 2 fixed relation between the internal variables of (1, 4), the Bell measurement on (2, 3) chooses a sub-ensemble in which there is an observed correlation between particles 2 and 3 and hence between particles 1 and 4, due to fixed prior relationship in internal variables of particle pairs (1, 2) and (3, 4). There is no nonlocality. The pair (1, 4) picked out using the observation of |y 2,3 ñ– state, for example, as a filter will also show the same behaviour as the pair (2, 3) in the state |y 2,3 ñ–. Only correlations at source and subsequent filtering into subensembles through Bell-type measurements or other similar operations are needed. The various schemes are yet to be studied in detail. Other applications Discussion Hardy’s puzzle Obtaining the correct correlations assuming locality implies that the measurement on one particle does not collapse the other particle to a definite state. There is a distinction between a real measurement and ‘predictability with certainty’ of an outcome. Quantum correlation at source and a measurement on one of the particles is enough to make this prediction using the conservation laws. But till an actual measurement is made the companion particle does not acquire a definite state. There is direct proof for this from the fact that while an actual measurement of position on one of the particles disperses its momentum according to the uncertainty principle, this measurement and the resulting 100% predictability of the companion particle’s position do not cause corresponding dispersion in the momentum of the companion particle. This is the lesson from Popper’s experiment 18–20 . Another important case is the one discussed by Hardy 15 involving two non-maximally entangled particles and four observables, in which four separate correlations predicted by quantum mechanics cannot be reproduced in the local realistic hidden variable theory. The nonlocality demonstration uses three zero joint probabilities and one nonzero joint probability constructed from three of the measurement possibilities. The local amplitudes which give the correct joint probabilities for the Hardy problem have also been constructed. K meson beams The analysis based on local physics presented in the earlier sections can be applied to the K0 – K 0 beam for 200 CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 RESEARCH ARTICLES Apart from resolving the nonlocality problem, we have also found an answer to the original EPR paradox of simultaneous reality of noncommuting observables. The paradox arises only from the necessity to assume reality for the outcomes before a measurement is made, to reproduce the correct perfect correlation. Since we found a way of getting these correlations without such an assumption, and without nonlocality, there is no EPR paradox. There are physical systems that are beyond the scope of the EPR definition of reality. The approach we have taken has locality as a basic feature and the objective reality (reality of a physical quantity independent of observer or apparatus) is at the level of the internal variables or initial phases. There is no objective reality to the outcomes before the measurement is made. One may explore the relation between the reality of the phase variable and the actual outcomes, and that would be a major step towards understanding quantum measurements. Such an understanding amounts to conceptual determinism in quantum mechanics, even though the initial phase is unmeasurable. It may turn out that observers in the classical world will not be able to grasp the mapping between such phases and actual outcomes. These issues are being contemplated on. In summary, we have reproduced the correct correlations of entangled particles under the assumption of strict locality of amplitudes. The correlations agreeing with experiments and quantum mechanical predictions emerge without one measurement causing a collapse of the state of the companion particle. This resolves the EPR nonlocality puzzle. The results have significant implications to the interpretation of all experiments involving entangled particles, including quantum teleportation and entanglement swapping. This approach introduces a new point of view in the physical and philosophical understanding of the nature of reality in the microscopic world. 1. Einstein, A., Podolsky, B. and Rosen, N., Phys. Rev., 1935, 47, 777–780 2. Wheeler, J. A. and Zurek, W. H. (eds), Quantum Theory and Measurement, Princeton University Press, Princeton, 1983. CURRENT SCIENCE, VOL. 79, NO. 2, 25 JULY 2000 3. Bell, J. S., Speakable and Unspeakable in Quantum Mechanics, Cambridge University Press, 1987. 4. A recent resource for exhaustive references is: Afriat, A. and Selleri, F., The Einstein, Podolsky, and Rosen Paradox in Atomic, Nuclear and Particle Physics, Plenum Press, New York, 1999. 5. Bohm, D., Quantum Theory, Prentice Hall, Englewood Cliffs, NJ, 1951. 6. Bells, J. S., Physics, 1964, 1, 195–200. 7. There have been a large number of important experiments which measured correlations of entangled particles. The first of the photon experiments was by Freedman S. J. and Clauser, J. F., Phys. Rev. Lett., 1972, 28, 938–941. The remarkable experiment by Aspect, A., Dalibard, J. and Roger, G., Phys. Rev. Lett., 1982, 49, 1804–1807 implemented change of the analyser settings while the particles were in flight. See ref. 4 for an exhaustive list and description of experiments. 8. Ghirardi, G. C., Rimini, A. and Weber, T., Lett. Nuovo Cimento, 1980, 27, 293–298; Jordan, T. F., Phys. Lett. A, 1983, 94, 264. 9. See for example, Penrose, R., Emperor’s New Mind, Vintage, UK, 1990, pp. 370–371. 10. Unnikrishnan, C. S., quant-ph/0001112, 2000. 11. Unnikrishnan, C. S., to be published; The geometric phase associated with spin rotations can be obtained directly from a generalization of the usual dynamical phase òp i dx i , to include the spin as one of the momenta p i and the angular coordinate as the corresponding dx i . 12. Greenberger, D. M., Horne, M. and Zeilinger, A., Phys. Today, 1993, 22–29. 13. Greenberger, D. M., Horne, M. and Zeilinger, A., in Bell’s Theorem, Quantum Theory and Conceptions of the Universe (ed. Kafatos, M.), Kluwer, Dordrecht, 1989, pp. 73–76; Mermin, N. D., Phys. Today, 1990, 9–11; Greenberger, D. M., Horne, M., Shimony, A. and Zeilinger, A., Am. J. Phys., 1990, 58, 1131– 1143. 14. Hariharan, P., Samuel, J. and Supurna Sinha, J. Opt. B: Quantum Semiclass. Opt., 1999, 1, 199–205. 15. Hardy, L., Phys. Rev. Lett., 1993, 71, 1665–1668. 16. Lipkin, H. J., Phys. Rev., 1968, 176, 1715–1718. 17. Pan, J-W., Bouwmeester, D., Weinfurter, H. and Zeilinger, A., Phys. Rev. Lett., 1998, 80, 3891–3894. 18. Popper, K. R., in Open Questions in Quantum Physics (eds Tarozzi, G. and van der Merwe, A.), D. Reidel Publishing Co, 1985. 19. Unnikrishnan, C. S., Found. Phys. Lett., 2000, 13, 197–200. 20. Kim, Y-H. and Shih, Y., Found. Phys., 1999, 29, 1849. Received 1 April 2000; revised accepted 22 June 2000 201