nature biotechnology • VOLUME 19 • MAY 2001 • http://biotech.nature.com

RESEARCH ARTICLE

466

Enhanced tolerance of rice to low iron availabilityin alkaline soils using barley nicotianamine

aminotransferase genesMichiko Takahashi1,2, Hiromi Nakanishi1, Shinji Kawasaki3, Naoko K. Nishizawa4 and Satoshi Mori1,2*

One of the widest ranging abiotic stresses in world agriculture arises from low iron (Fe) availability due to highsoil pH, with 30% of arable land too alkaline for optimal crop production. Rice is especially susceptible to lowiron supply, whereas other graminaceous crops such as barley are not. A barley genomic DNA fragment containing two naat genes, which encode crucial enzymes involved in the biosynthesis of phytosiderophores,was introduced into rice using Agrobacterium-mediated transformation and pBIGRZ1. Phytosiderophores arenatural iron chelators that graminaceous plants secrete from their roots to solubilize iron in the soil. The twotransgenes were expressed in response to low iron nutritional status in both the shoots and roots of rice transformants. Transgenic rice expressing the two genes showed a higher nicotianamine aminotransferaseactivity and secreted larger amounts of phytosiderophores than nontransformants under iron-deficient conditions. Consequently, the transgenic rice showed an enhanced tolerance to low iron availability and had4.1 times greater grain yields than that of the nontransformant rice in an alkaline soil.

Increasing the productivity of graminaceous plants is an importantstrategy to meet the increased demand for food expected as a resultof rapid population increase. Agricultural productivity is severelyaffected by high soil pH, in which metal ions, especially iron, aresparingly soluble and not available to plants1. Plants growing inhigh-pH soils develop iron deficiency symptoms that are manifestedas chlorosis (yellowing from partial failure to develop chlorophyll),even though mineral soils contain >6% total iron.

Mugineic acid family phytosiderophores (MAs) are natural ironchelators that graminaceous plants secrete from their roots to solubi-lize iron in the soil2. Both the quantity and kind of MAs secreted dif-fer among graminaceous plant species. The resulting Fe(III)–MAscomplexes are reabsorbed into the root through an Fe(III)–MAstransporter in the plasma membrane3,4. MAs are synthesized from L-methionine5 through nicotianamine6,7 (Fig. 1). Nicotianaminesynthase (NAS)8 and nicotianamine aminotransferase (NAAT)9 arethe critical enzymes in the biosynthesis of MAs. The activities of NASand NAAT are markedly increased in response to iron deficiency8, 9,resulting in increased secretion of MAs.

Two complementary DNAs (cDNAs), naat-A and naat-B, based onthe amino acid sequence of purified NAAT protein, were isolatedfrom a cDNA library prepared from iron-deficient barley roots10. Theamounts of both naat-A and naat-B transcripts increased markedlyin iron-deficient barley roots. A barley genomic DNA fragment with atandem array of naat-B and naat-A (in this order) was also isolated.

Because NAAT activity has a high correlation with the amount ofMAs secreted9, it should also correlate with a plant’s tolerance to lowiron availability. Rice secretes very low amounts of MAs and is themost susceptible of the graminaceous plants to iron deficiency. It washypothesized, therefore, that enhancing NAAT activity throughgenetic engineering would increase the tolerance of rice plants to lowiron availability through increased secretion of MAs. We report that,

in the present study, transgenic rice plants harboring a barley naatgenomic DNA fragment had higher NAAT activity, secretedincreased amounts of MAs, and showed enhanced tolerance to lowiron availability in an alkaline soil.

Results and discussionTransformation of rice with a barley genome DNA fragment. An 11kilobase pair barley genome DNA fragment containing two naatgenes, naat-A and naat-B, was introduced into rice usingAgrobacterium-mediated transformation and pBIGRZ1. The entirenaat-A 5′-flanking region is included in this fragment, whereas naat-Bhas a partial 5′-flanking region of ∼ 600 base pairs (Fig. 2A).pBIGRZ1 is a vector that is capable of holding a large insert, and itallows the expression of both genes to be regulated by their own pro-moters11. Thirty-six independent transgenic rice lines were obtained.

Southern blot analysis using probes specific for either naat-A ornaat-B showed that both transgenes were efficiently integrated in therice genome (Fig. 2B). The results in Figure 2B demonstrate the dif-ferent possible patterns of integrated genes among the transfor-mants. T2 progeny harboring a single copy of each transgene werechosen for further analysis. Expression of the introduced barley naatgenes was examined by northern blot analysis (Fig. 2C). Transcriptsof naat-A were abundant in iron-deficient roots of transformant N1,whereas they were not detected in iron-sufficient roots. naat-B wasexpressed at only a low level in iron-sufficient roots of N1, but itsmessenger RNA (mRNA) level was considerably higher in iron-deficient roots. This response pattern is the same as that observed inbarley roots10, in which naat-A is expressed only in iron-deficientroots, whereas naat-B expression is low in iron-sufficient roots andgreatly increased in iron-deficient roots.

Interestingly, besides expression in the roots, integrated naat-A wasexpressed at a low level in iron-deficient shoots of N1. Furthermore,

1Laboratory of Plant Molecular Physiology, The University of Tokyo, Tokyo 113-8657, Japan. 2CREST, Tsukuba 305-8602, Japan.3National Institute of Agrobiological Resources, Tsukuba 305-8602, Japan. 4Laboratory of Plant Biotechnology, The University of Tokyo, Tokyo 113-8657, Japan.

*Corresponding author (asmori@mail.ecc.u-tokyo.ac.jp).

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://b

iote

ch.n

atu

re.c

om

© 2001 Nature Publishing Group http://biotech.nature.com

RESEARCH ARTICLE

http://biotech.nature.com • MAY 2001 • VOLUME 19 • nature biotechnology 467

naat-B transcripts were detected in iron-sufficient shoots of N1, withhigher expression with iron deficiency. Expression of naat-A or naat-Bis not observed in barley shoots, either under iron-sufficient or iron-deficient conditions10, so the root-specific expression of naat-A andnaat-B seen in barley was lost in the transformed rice. However, theirresponses to iron nutritional status were maintained in the shoots oftransformed rice; naat-A mRNA was detected in shoots only underiron-deficient conditions, and the transcript level of naat-B increasedin response to low iron availability.

These results confirm that the transgenes of naat-A and naat-B canbe heterologously expressed in rice from their own promoters inresponse to iron nutritional status. This also indicates that rice containsother factors that are required for iron nutritional status-regulatedexpression of the naat genes and are capable of regulating the genes.

In contrast, their shoot expression profiles differ from those seen inbarley. Interestingly, iron deficiency induced the expression of barleyNAS gene in both shoots and roots of transgenic rice carrying a barleygenome DNA fragment containing HvNAS1 (i.e., Hordeum vulgareDNA encoding NAS)12. Moreover, in rice, iron deficiency induced theexpression of OsNAS (i.e., Oryza satira DNA encoding rice NAS)gene in both shoots and roots. These results indicated that MAs aresynthesized in both shoots and roots in iron-deficient rice.Deoxymugineic acid (DMA), the first member of MAs (Fig. 1), was infact detected in the leaves of iron-deficient rice, with higher levelsfound in chlorotic leaves than in green leaves of the same plant13.

Increased NAAT activity and secretion of DMA. NAAT activity wasexamined in the roots of transformant N1 and of a nontransformantunder both iron-sufficient and iron-deficient conditions (Fig. 3A).The specific activity of NAAT in the iron-sufficient N1 plant was thesame as in iron-sufficient barley14, and therefore higher than the activ-ity in the nontransformed rice plants. NAAT activity in the N1 rootswas greatly increased by iron deficiency, and reached the same level asin iron-deficient barley roots. This was 60 times higher than the activ-ity seen in the iron-deficient nontransformed rice plants, althoughendogenous rice NAAT activity was increased in these plants by lowiron supply.

The secreted MAs were analyzed by high-pressure liquid chro-matography (HPLC). The major MAs secreted was DMA. Theamount of DMA secreted from iron-deficient N1 plants was 1.8 times higher than that from iron-deficient nontransformants(Fig. 3B). As compared with the dramatic increase in NAAT activity,the amount of DMA secreted during a 5 h period was not very high.This discrepancy between NAAT activity and DMA secretion may becaused by a shortage of substrate NA and/or the relatively loweractivity of the DMA synthase (Fig. 1). However, this result demon-strates that enhancing the activity of NAAT, which is a crucialenzyme for synthesizing MAs, enables graminaceous plants tosecrete an increased amount of MAs.

Evaluation of transgenic rice plants in an alkaline soil. The plantswere cultured in an alkaline soil (pH 8.5) to examine whether thetransformants have enhanced tolerance to low iron availability. Of the36 lines evaluated, 10 showed remarkable tolerance to low iron, andgrew better than the control plants (Fig. 4A, B). The control plants

Figure 2. Construct of pBIGRZ-NAAT binary vector and molecularanalysis of the transformants.(A) Construction of the pBIGRZ-NAATbinary vector. Genomic DNA fragment contains naat-B (gray box) andnaat-A (black box). (B) Southern blot analysis of DNA prepared fromnontransformant rice (WT), and independent transgenic T1 plants (C8, control plant harboring pBIGRZ1 vector alone; N1, N3, N8,transformants). naat-A and naat-B indicate bands detected by the 32P-labeled probes specific for naat-A and naat-B, respectively. (C)Expression of naat-A and naat-B in the roots and shoots of transgenic T2plants (C8, N1) and nontransformant (WT), under Fe-sufficient (+Fe) andFe-deficient (–Fe) conditions. naat-A and naat-B indicate northern blotanalysis performed with the 32P-labeled probes specific for naat-A andnaat-B, respectively. RB, right border; NP, nopaline synthase promoter;NPTII, neomycin phosphotransferase; NT, 3’ signal of nopaline synthase;LacZ, β-galactosidase; MCS, multicloning site; 35P, 35S promoter;iGUS, β-glucuronidase including intron; HPT, hygromycinphosphotransferase; LB, left border; Ri ori, origin of replication of Riplasmid; RK2 ori, origin of replication of RK2.

Figure 1. Biosynthetic pathway of mugineic acid familyphytosiderophores.

A

B C

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://b

iote

ch.n

atu

re.c

om

© 2001 Nature Publishing Group http://biotech.nature.com

RESEARCH ARTICLE

nature biotechnology • VOLUME 19 • MAY 2001 • http://biotech.nature.com468

showed reduced growth, and their old leaves and flag leaves displayedchlorosis caused by low iron supply. In contrast, the transformantplants had green leaves and shoots. For example, the average SPAD-502 value (leaf chlorophyll content) measured by a chlorophyll meterwas 35.1 in the leaves of the transformants and 6.7 for control plantsat flowering stage (Fig. 4B). It was evident from measurement of plantheight (Fig. 5A) that the transformants grew much better than thecontrol plants. The average transformant shoot dry weight per potafter 16 weeks in the alkaline soil was 4.2 times higher than that of the

controls (Fig. 5B). The average grain yield of transformants per potwas 4.1 times greater than that of the controls (Fig. 5B).

In this study we have shown that enhancing NAAT activity withbarley naat genes increases the amount of MAs secreted, and conse-quently enhances tolerance of rice to low iron availability. However,there was a difference in activity levels of NAAT relative to theamount of secreted DMA during a 5 h period. To date, we have iso-lated several genes involved in the biosynthesis of MAs (refs 15–21)and several iron deficiency-inducible genes22–26. It will be interesting,therefore, to see whether transgenic rice carrying combinations ofthese genes that are induced by iron deficiency can synthesize andsecrete even higher amounts of MAs, and be more tolerant of lowiron supply, than rice carrying the two naat genes. Again, pBIGRZ1,with its capacity to hold large inserts, will enable us to efficiently andstably introduce a combination of multiple genes into the ricegenome with a single transformation.

Genetically engineered plants tolerant to low iron availability inan alkaline soil, such as the transgenic rice carrying the barley naatgenes described here, have not been reported before. Besides thejaponica rice used in this study, indica rice, maize, and sorghum,which are all susceptible to low iron availability, are possible targetcrops for the introduction of genes involved in the biosynthesis ofMAs. With enhanced secretion of MAs, these crops should be toler-ant to low iron availability in alkaline soils. The development of rice

Figure 3. NAAT activity and DMA secretion. (A) NAAT activity in the rootsof transgenic T2 plants (N1) and nontransformant plant (WT). The specificactivity of NAAT under iron-sufficient (hatched column) and iron-deficientconditions (black column) are shown. (B) Quantity of DMA secreted froma transgenic T2 plant (N1) and a nontransformant plant (WT) under iron-sufficient and iron-deficient conditions (n = 8).

A B

Figure 5. Growth features and grain yield of transgenic rice and wild-typerice in an alkaline soil. (A) Plant height of three control rice plants (brokenlines) and five transgenic T1 plants (solid lines) from 36 to 87 days aftertransplanting into an alkaline soil (pH 8.5). (B) Dry weights per pot ofshoots and grain of control (C1, C6, C8) and transgenic T1 plants (N1, N3, N8, N9, N17) after cultivation for 90 days in an alkaline soil.

A

B

Figure 4. Transgenic rice harboring naat genes is tolerant to low ironavailability in an alkaline soil.(A) Four control plants (C8, left) and fourtransgenic T1 rice plants (N1, right) after five weeks of growth in an alkalinesoil (pH 8.5).(B) A control plant (C8, left) and a transgenic T1 rice plant (N1, right) at flowering stage after 14 weeks of growth in an alkaline soil.

A

B

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://b

iote

ch.n

atu

re.c

om

© 2001 Nature Publishing Group http://biotech.nature.com

RESEARCH ARTICLE

http://biotech.nature.com • MAY 2001 • VOLUME 19 • nature biotechnology 469

that tolerates low iron availability is the first step in increasing theproductivity of graminaceous plants in alkaline soils, and will con-tribute to meeting increased demands for grain production.

Experimental protocolConstruction of pBIGRZ-NAAT and transformation of rice. The genomicnaat-A fragment was isolated as described10. The fragment was digested withNotI and cloned into a binary vector pBIGRZ111. pBIGRZ-NAAT was trans-ferred into Agrobacterium tumefaciens strain C58, which was cultured forthree days at 25°C in the dark with three-week-old calli derived from matureseeds of rice (Oryza sativa L. cv. Tsukinohikari). The calli were then culturedfor one week on N6 selection medium27 supplemented with 3% sucrose, 0.2%gelrite, 1 g/L casamino acid, 2 mg/L 2,4-D, 250 mg/L Claforan (HoechstMarion Roussel Ltd. Japan), and 10 mg/L hygromycin B. The calli were cul-tured on new N6 selection medium with an increased concentration ofhygromycin B (50 mg/L) for three weeks at 28°C, under 16 h light/8 h darkconditions. Subsequently, they were cultured on Murashige–Skoog (MS)regeneration medium supplemented with 3% sucrose, 3% sorbitol, 0.4%agar, 2 g/L casamino acids, 5 mM methyl ethane sulfonate, 2 mg/L α-naph-thaleneacetic acid (NAA), 1 mg/L kinetin, 250 mg/L Claforan, and 50 mg/Lhygromycin for three weeks. Finally, they were cultured on MS selectionmedium supplemented with 3% sucrose, 8 g/L agar, 250 mg/L Claforan, and50 mg/L hygromycin for two to three weeks at 28°C, under 16 h light/8 h darkconditions. Transformants (T1 plants) were tested in alkaline soil (pH 8.5),and the self-pollinated heterozygous progeny (T2) were used for other exper-iments, except for Southern hybridization.

Southern blot analysis. Genomic DNA samples (20 µg) were digested withHindIII and separated by electrophoresis on 0.8% agarose gels. The blottedmembrane was hybridized overnight in Gold hybridization buffer (AmershamLife Science, London, UK) with [a-32P]-labeled probes of ∼ 100 bp specific10 foreither naat-A or naat-B at 42°C. The membrane was hybridized first with thenaat-A probe and then rehybridized with the naat-B probe.

Northern blot analysis. Total RNA was isolated from the shoots and roots oftransformant, control, and nontransformant plants using RNeasy Plant Minikits (Qiagen, Japan). The RNA (10 µg) was denatured and electrophoresed on

1.2% agarose gels containing 5% (vol/vol) formaldehyde. The blotted mem-brane (Hybond-N+, Amersham) was hybridized with the same [a-32P]-labeledprobes specific for naat-A and naat-B as used for Southern hybridization.

NAAT enzyme assay and measurement of secreted MAs. The N1 plant washydroponically cultured for three weeks with iron-free culture solution toinduce iron deficiency, and root washings were collected for 5 h after the ini-tial illumination. The root washings were analyzed by HPLC (ref. 5).

The roots of hydroponically cultured plants were harvested and assayedfor NAAT activity9,28. Aliquots of 0.4–0.6 g of the frozen roots were homoge-nized with 1 ml of the extraction buffer, then centrifuged and the super-natants filtered. The enzyme reactions were carried out in 50 ml of theenzyme reaction buffer (45 mM TAPS, 4. 5 mM KCl, 4. 5 mM MgCl2, 10 µMpyridoxal phosphate, 150 mM NA, and 10 µM 2-oxoglutarate and 50 mg pro-teins, pH 9) in the filter cup of Ultrafree (Millipore, Japan) for 30 min. Thereaction products were then separated from the enzyme by ultrafiltration. Tothe filtrates, 4 µl of 0.25 M NaBH4 were added to reduce the reaction productsto DMA (Fig. 1). DMA was then analyzed by HPLC. Protein was assayedusing a kit from BioRad Laboratories (Japan).

Test for low iron availability tolerance. The T1 plants were cultivated in potsfilled with alkaline soil and set in a bath watered with Kasugai’s culture solu-tion (composition: 7 x 10-4 M K2SO4, 1 x 10-4 M KCl, 1 x 10-4 M KH2PO4,2 x 10-3 M Ca(NO3)2, 5 x 10-4 M MgSO4, 1 x 10-5 M H3BO3, 5 x 10-7 M MnSO4,5 x 10-7 M ZnSO4, 2 x 10-7 M CuSO4, 1 x 10-8 M (NH4)6Mo7O24, and 1.5 x 10-4

M Fe-EDTA) without iron. The height of the plants was measured every oneto two weeks. At the same time, the degree of chlorosis of the youngest fullyexpanded leaves was determined using a SPAD-502 chlorophyll meter(Minolta Co., Tokyo, Japan). Plants were grown in a pot filled with an alkalinesoil (pH 8.5) obtained from Toyama prefecture in Japan and in a glass houseunder natural light conditions.

AcknowledgmentsWe thank Dr. Pax Blamey in Queensland University for reading of this manu-script and Mr. Takeo Tanaka for his valuable assistance.

Received 31 October 2000; accepted 2 February 2001

1. Marschner, H. Mineral nutrition of higher plants, Edn. 2. (Academic Press,London, UK; 1995).

2. Takagi, S. Naturally occurring iron-chelating compounds in oat- and rice-rootwashing. I. Activity measurement and preliminary characterization. Soil Sci. PlantNutr. 22, 423–433 (1976).

3. Mihashi, S. & Mori, S. Characterization of mugineic-acid-Fe transporter in Fe-defi-cient barley roots using the multi-compartment transport box method. Bio Metals2, 146–154 (1989).

4. Römheld, V. & Marschner, H. Evidence for a specific uptake system for iron phy-tosiderophores in roots of grasses. Plant Physiol. 70, 175–180 (1986).

5. Mori, S. & Nishizawa, N.K. Methionine as a dominant precursor of phy-tosiderophores in Gramineae plants. Plant Cell Physiol. 28, 1081–1092 (1987).

6. Shojima, S., Nishizawa, N.K. & Mori, S. Establishment of a cell-free system for thebiosynthesis of nicotianamine. Plant Cell Physiol. 30, 673–677 (1989).

7. Shojima, S. et al. Biosynthesis of phytosiderophores: in vitro biosynthesis of 2’-deoxymugineic acid from L-methionine and nicotianamine. Plant Physiol. 93,1497–1503 (1990).

8. Higuchi, K., Kanazawa, K., Nishizawa, N.K. & Mori, S.The role of nicotianamine syn-thase in response to Fe nutrition status in Gramineae.Plant Soil 178, 171–177 (1996).

9. Kanazawa, K. et al. Nicotianamine aminotransferase activities are correlated tothe phytosiderophore secretions under Fe-deficient conditions in Gramineae. J.Exp. Bot. 45, 1903–1906 (1994).

10. Takahashi, M. et al. Cloning two genes for nicotianamine aminotransferase, a crit-ical enzyme in iron acquisition (strategy II) in graminaceous plants. Plant Physiol.121, 947–956 (1999).

11. Akiyama, K. et al. Development of a system of rice transformation with longgenome inserts for their functional analysis for positional cloning. Plant CellPhysiol. Suppl. 38, s94 (1997).

12. Higuchi, K. et al. Nicotianamine synthase gene expression differs in barley andrice under Fe-deficient conditions. Plant J. 25, 159–168 (2001).

13. Mori, S. et al. Why are young rice plants highly susceptible to iron deficiency?Plant Soil 130, 143–156 (1991).

14. Kanazawa, K., Higuchi, K., Nishizawa, N.K., Fushiya, S. & Mori, S. Detection oftwo distinct isozymes of nicotianamine aminotransferase in Fe-deficient barleyroots. J. Exp. Bot. 46, 1241–1244 (1995).

15. Takizawa, R., Nishizawa, N.K., Nakanishi, H. & Mori, S. Effect of iron deficiency on S-adenosyl-methionine synthetase in barley roots.J.Plant Nutr. 19, 1189–1200 (1996).

16. Higuchi, K. et al. Cloning of nicotianamine synthase, novel genes involved in the

biosynthesis of phytosiderophores. Plant Physiol. 119, 471–480 (1999).17. Suzuki, K., Higuchi, K., Nakanishi, H., Nishizawa, N-K. & Mori, S. Cloning of nico-

tianamine synthase genes from Arabidopsis thaliana. Soil Sci. Plant Nutr. 45,993–1002 (1999).

18. Okumura, N. et al. A dioxygenase gene (Ids2) expressed under iron deficiencyconditions in the roots of Hordeum vulgare. Plant Mol. Biol. 25, 705–719 (1994).

19. Nakanishi, H. et al. Expression of a gene specific for iron deficiency (Ids3) in theroots of Hordeum vulgare. Plant Cell Physiol. 34, 401–410 (1993).

20. Nakanishi, H., Yamaguchi, H., Sasakuma, T., Nishizawa, N.K. & Mori, S. Twodioxygenase genes, Ids 3 and Ids 2, from Hordeum vulgare are involved in thebiosynthesis of mugineic acid family phytosiderophores. Plant Mol. Biol. 44,199–207 (2000).

21. Kobayashi, T. et al. In-vivo evidence that Ids 3 from Hordeum vulgare encodes adioxygenase that converts 2’-deoxymugineic acid to mugineic acid in transgenicrice. Planta, in press (2001).

22. Okumura, N., Nishizawa, N.K., Umehara, Y. & Mori, S. An iron deficiency-specificcDNA from barley roots having two homologous cysteine-rich MT domains. PlantMol. Biol. 17, 531–533 (1991).

23. Suzuki, K. et al. Formate dehydrogenase, an enzyme of anaerobic metabolism, isinduced by Fe-deficiency in barley roots. Plant Physiol. 116, 725–732 (1998).

24. Yamaguchi, H., Nakanishi, H., Nishizawa, N.K., & Mori, S. Induction of the IDI1gene in Fe-deficient barley roots: a gene encoding a putative enzyme that cataly-ses the methionine salvage pathway for phytosiderophore production. Soil Sci.Plant Nutr. 46, 1–9 (2000).

25. Yamaguchi, H., Nakanishi, H., Nishizawa, N.K. & Mori, S. Isolation and character-ization of IDI2, a new Fe-deficiency induced cDNA from barley roots, whichencodes a protein related to the α-subunit of eukaryotic initiation factor 2B(eIF2Bα). J. Exp. Bot. 51, 2001–2007 (2000).

26. Itai, R. et al. Induced activity of adenine phosphoribosyltransferase (APRT) in iron-deficient barley roots: a possible role for phytosiderophore production. J. Exp. Bot.51, 1179–1188 (2000).

27. Chu, C. The N6 medium and its application to anther culture of cereal crops. InProceedings of Symposium on Plant Tissue Culture, pp. 43-50. (SciencePress,1978).

28. Kanazawa, K., Higuchi, K., Nakanishi, H., Nishizawa, N.K., & Mori, S.Characterizing nicotianamine aminotransferase: improving its assay system anddetails of the regulation of its activity by Fe nutrition status. Soil Sci. Plant Nutr. 44,717–721 (1998).

©20

01 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://b

iote

ch.n

atu

re.c

om

© 2001 Nature Publishing Group http://biotech.nature.com