Mol Gen Genet (1990) 220:245-250

© Springer-Verlag 1990

Construction of an intron-containing marker gene: Splicing of the intron in transgenic plants and its use in monitoring early events in Agrobacterium-mediated plant transformation G. Vancanneyt, R. Schmidt, A. O'Connor-Sanchez*, L. Willmitzer, and M. Rocha-Sosa** Institut fiir Genbiologische Forschung, Ihnestrasse 63, D-1000 Berlin 33

Summary. Agrobacterium tumefaciens is a commonly used tool for transforming dicotyledonous plants. The underly- ing mechanism of transformation however is not very well understood. One problem complicating the analysis of this mechanism is the fact that most indicator genes are already active in Agrobaeterium, thereby preventing the precise de- termination of timing and localisation of T-DNA transfer to plant cells. In order to overcome this obstacle a modified prokaryotic indicator gene was constructed. The expression of this indicator gene and its use in analysing early events in Agrobacterium-mediated plant transformation are de- scribed. A portable intron, derived from a plant intron, was introduced into the fl-glucuronidase (GUS) gene. In transgenic plants containing this chimaeric gene the intron is spliced efficiently, giving rise to GUS enzymatic activity. Mapping of the splice junction indicates the exact removal of the intron. No GUS activity is detected in agrobacteria containing this construct due to the lack of a eukaryotic splicing apparatus in prokaryotes. Early phases after trans- formation of Arabidopsis cotyledon explants were analysed using this GUS-intron chimaeric gene showing that as early as 36 h after Agrobacterium infection significant GUS activ- ity is detected. In vivo GUS staining of transformed cells clearly shows that quickly proliferating calli expressing GUS activity are formed, mainly at the cut surface. Minor transformation events occur however throughout the whole cotyledon. These data indicate that Agrobacterium-me- diated T-DNA transfer to plants is much more efficient than has been judged from experiments where selection is applied immediately. The intron-containing GUS gene can be used as an optimised marker gene in transient and stable transformation experiments.

Key words: Agrobacterium tumefaciens - /~-glucuronidase - Portable intron - Splicing - Transformation

Introduction

The gram-negative soil bacterium Agrobacterium tumefa- ciens has been developed into a valuable tool for transfer-

* Present address." Center for Investigation and Advanced Studies, Irapuato, Mexico ** Present address: Centro de Investigati6n sobre Fijaci6n de Ni- tr6geno, Cuernavaca, Morelos, Mexico

Offprint requests to: L. Willmitzer

ring genes to a number of higher plant species (Schell 1987). Despite its extensive use, little is understood about the early phase in transformation experiments (Zambrisky et al. 1989). Knowledge of the relative susceptibility of different cells and tissues to transformation and the amenability of the initially transformed tissue for regeneration would how- ever be helpful in devising strategies for transformation ex- periments for recalcitrant plant species.

Various bacterial indicator genes have been developed which allow the selection of and/or screening for trans- formed cells (Schell 1987; Klee et al. 1987). One major hin- drance however is the fact that despite the usage of eukar- yotic promoters these genes are still well expressed in agro- bacteria (unpublished observations). In order to overcome this problem we took advantage of the fact that prokaryotes are devoid of the eukaryotic splicing apparatus. To this end a portable intron derived from a plant gene was con- structed and inserted into a reporter gene.

The intron to be used must fulfil the following condi- tions:

It should be "por table" in order to allow its easy inser- tion into different sites of any gene of interest.

It should contain stop codons in all three possible read- ing frames thus excluding expression in agrobacteria.

It should be a " typical" plant intron in order to increase the likelihood that it will be spliced efficiently.

Introns in plant genes have not been characterised in as much detail as their animal homologues. However some features common to all plant introns are known. Their aver- age size is 250 nucleotides, their AT content is high and 5' and 3' splice junctions are similar to those of introns of animal genes (Hawkins 1988; Brown 1986; Wiebauer et al. 1988). The second intron (IV2) of the ST-LS1 gene (Eckes et al. 1986) comprises typical plant intron features: a length of 189 nucleotides, an AT content of 80% and typical splice junctions (Fig. 1). Moreover this intron also fulfils the second condition, i.e. it contains several stop co- dons in all reading frames.

The splicing mechanism in higher plants has not been investigated in great detail to date. The failure of a plant system to process mammalian introns (Bartha et al. 1986) raised the possibility that the mechanism of intron recogni- tion differs in plants and animals. On the other hand several plant introns have been shown to be properly spliced in vitro in HeLa cell nuclear extracts (Brown et al. 1986; Har- muth and Bartha 1986). In this context an in vivo approach using a chimaeric intron-reporter gene could elucidate pa- rameters involved in pre-mRNA processing.

246

The indicator gene to be constructed should allow the sensitive detection of its encoded product both quantitative- ly in tissue homogenates and qualitatively in single plant cells. We therefore decided to use the p-glucuronidase (GUS) gene from Escherichia coli and to modify it by insert- ing an intron in the protein-coding region, thus preventing its expression in a prokaryotic environment. The functiona- lity of this portable intron, the accuracy of splicing and the effects on GUS enzymatic activity were analysed in transgenic plants. Furthermore the chimaeric intron-GUS gene was used as a tool in monitoring early events in Agro- bacterium-mediated Arabidopsis transformation.

Materials and methods

Bacterial media and strains. All agrobacterial strains were grown in YEB medium (Vervliet et al. 1975). E. coli strains were grown in YT medium (Maniatis et al. 1982). Binary vectors were introduced into the GV 2260 agrobacterial strain (Deblaere et al. 1985).

Construction. The p35S GUS, kindly provided by J. Stock- haus, is a BIN 19 (Bevan 1984) derivative into which the GUS gene was introduced under the control of the 35S promoter from pRT 102 (T6pfer et al. 1987). The constructs were introduced into Agrobacterium via direct DNA trans- formation (H6fgen and Willmitzer 1988). Site-directed mu- tagenesis was performed according to the protocol of the supplier (Amersham), using synthetic oligonucleotides.

Plant transformation. Tobacco leaves (Nicotiana tabacum var. W38) were used for Agrobaeterium-mediated leaf disc infection as described (Horsch et al. 1985). Transformants were selected on 100 lig/ml kanamycin.

Arabidopsis thaliana cotyledons were used as explants for transformation (Schmidt and Willmitzer 1988). The an- tibiotic G418 was used as the selective agent for transfor- mants at a concentration of 20 rtg/ml.

Root explants of A. thaliana were transformed accord- ing to Valvekens et al. (1988).

Northern analysis. Total RNA of leaves was isolated (Loge- mann etal. 1987), separated on 1% agarose gels in the presence of formaldehyde, blotted on Hybond nylon mem- branes and used for hybridisation as described by Amasino (1986). The complete GUS gene was used for probing after labelling using the multiprime approach (Amersham proto- col).

Primer extension. Total leaf RNA (30 lag) was hybridised to an oligonucleotide (20-met) overnight and extended us- ing AMV reverse transcriptase according to Kingston (1987). The first strand cDNA was amplified using the poly- merase chain reaction (PCR) according to the Perkin Elmer protocol. The denaturation was done at 92 ° C, the anneal- ing at 50 ° C and the polymerase reaction at 65 ° C. Twenty cycles were done in a Perkin Elmer thermal cycler.

Assays for GUS activity. For the fluorometric GUS assay, explants were homogenised and incubated with the sub- strate 4-methylumbelliferyl-fl-D-glucuronide at 37 ° C. Quantification of the fluorescence was done according to Jefferson (1987) and Jefferson et al. (1987) and expressed

as picomoles methylumbelliferrol per milligram protein per minute. Protein concentrations were determined according to Bradford (1979).

For in vivo staining, intact plant material was vacuum infiltrated with 1 mM X-gluc (5-bromo-4-chloro-3-indolyl- fl-D-glucuronic acid cyclohexylammonium) and incubated as reported by Jefferson (1987) and Jefferson et al. (1987).

Late logarithmic agrobacterial cultures were stained for 24 h in a similar way, omitting vacuum infiltration and ethanol treatment.

Results

Construction of a G US gene containing a portable intron

The second intron (IV2) of the ST-LS1 gene was introduced into the bacterial GUS gene. In order to simplify the cloning the intron was slightly modified at the border sequences. In addition internal intron border sequences were optimised with respect to the consensus sequence for plant introns: A G / / G T A A G T . . . T G C A G / / G (Shapiro and Senepathy 1987). To this end IV2 was isolated from the ST-LS1 gene as a S t y I - H p h I fragment and cloned in an M13 intermedi- ate vector. Modifications were performed by site-directed mutagenesis using synthetic oligonucleotides, changing the left border sequence AAG/GTT into TAC/GTA resulting

ST-LS I

AAG IGTTTGTTTGT . . . . . . . . . . . . . . . . . . . . . . !..~CATGGTGATGTTTAG ', GGA IV 2 L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J

I Site directed mutogenesis

R q TA0,.cGTAAG'V1-TCT ......................... ATrTGITGATGTGGAG, CTG PIV 2

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~

Sno BI PvulI

~~$no, B I ~ p35S , ~ . . . . . 35SPA

EL . . . . . . cpB . gderivo.vel . . . . . . J )

PIV2

p35S GUS INT "~

Fig. t. Cloning of the portable intron in the fl-glucuronidase (GUS) gene. The upper drawing shows schematically a part of the struc- ture of the ST-LS1 gene. Boxes represent exons and solid lines introns; in addition part of the nucleotide sequence of the second intron 0V2) is shown. The sequences in the boxes confined by the dotted lines represent the wild-type intron (IV2) and the mu- tated portable intron (PIV2). The lower drawing shows p35S GUS and its intron derivative p35S GUS INT, as described in Results

in a SnaBI restriction site and the fight border T A G / G G A into CAG/CTG resulting in a PvuII site (Fig. 1). The port- able intron (PIV2) obtained was cloned into the single SnaBI restriction site of the GUS gene (Jefferson 1987; Jefferson et al. 1987) which had been inserted between the Cauliflower Mosaic Virus derived 35S promoter and polya- denylation signal in a derivative of the binary vector BIN 19 (Bevan 1984) resulting in plasmid p35S GUS INT (Fig. 1).

RNA analysis and mapping of the splice junction in transgenic tobacco plants

In order to analyse whether or not the intron was functional in the chimaeric construct, it was transferred into tobacco using Agrobacterium-mediated transformation (Horsch et al. 1985). Several independently transformed plants, con- taining intact copies of the 35S GUS INT gene, were scored for the presence of GUS INT-derived RNA. By Northern analysis a single RNA migrating at a molecular weight of approximately 1.8 kb was observed, which hybridised to the GUS gene (Fig. 2). This corresponds to the size of the mRNA of the GUS gene without intron (lane C in Fig. 2). The fact that no RNA species of higher molecular weight was observed, indicates that IV2 is spliced out efficiently. Moreover it is interesting to note that in the cases analysed the abundance of the GUS mRNA was similar to that of the intronless GUS gene construct (p35S GUS).

In order to map the splice junction in the GUS gene a combined primer extension-PCR was performed. Total leaf RNA of transgenic tobacco plants containing either the intact GUS gene (35S GUS; Fig. 3, lane A) or GUS gene interrupted by the intron (35S GUS INT; Fig. 3, lane B) was hybridised to a 20-mer oligonucleotide (Gol I) localised 3' of the splice junction and used for primer exten- sion as shown in Fig. 3, (1). The first strand cDNA was consequently amplified using the PCR (Saiki et al. 1988). The second oligonucleotide (Gol 2, 25-mer) is localised 240 bp 5' of Gol 1 (Fig. 3, (1)). After agarose gel electro- phoresis a single amplified fragment with a size of 240 bp can be observed in both cases (Fig. 3, lanes A, B). A 430 bp fragment, which would result from an unspliced GUS mRNA, could not be detected. Digestion of the amplified fragments with SnaBI revealed a 190 bp fragment which is expected after the removal of the intron. However, some undigested DNA fragment was observed, which was most probably due to inefficient digestion by the SnaBI restric-

W38 1 2 3 4 C

1.8 kb

Fig. 2. Northern analysis of plants transformed with intron-con- taining and intron-free chimaeric GUS genes. Total RNA (50 lag each) isolated from leaves of four independent transgenic tobacco plants containing the 35S GUS INT gene (lanes 14) or the intron- free 35S GUS gene (lane C) were separated by gel electrophoresis, blotted and hybridised against the complete GUS gene. The size of the RNA is given in nucleotides on the right. Lane W38 contains RNA isolated from nontransformed tobacco plants

247

240-- 190--

SnaBI

I A B I B A

',05 bp

;14

B9

H6

(1) V Gol I

AAA

Gol 2 S

(5)

Gol 1

240 bp "4--- 190 bp ---~

Fig. 3. Mapping of the splice site in the chimaeric GUS-intron gene. Total leaf RNA (30 lag) of transgenic tobacco plants contain- ing either p35S GUS (lane A) or p35S GUS INT (lane B) was hybridised to a 20-mer oligonueleotide Gol 1 and used for primer extension (1). The first strand eDNA was amplified by the polymer- ase chain reaction (PCR) using the 25-met oligonucleotide Gol 2 (2). The molecular weight of the amplified fragment was deter- mined by 1.5% agarose gel electrophoresis as shown in the Upper panel. Samples in the lanes marked SnaBI were digested with the restriction enzyme SnaBI. As a size marker 2 DNA was digested with PstI

tion enzyme. These data suggest that the portable intron is accurately spliced.

Functional analysis of GUS activity in higher plants

The RNA analysis suggested in that the influence of the intron on the level of expression in transgenic tobacco plants was low. This was analysed in Arabidopsis in more detail by quantitative fluorescence assays (Jefferson 1987; Jefferson et al. 1987). Using the Arabidopsis cotyledon transformation procedure (Schmidt and Willmitzer 1988) calli were obtained containing either 35S GUS or 35S GUS INT. Quantitative measurements of GUS activity were per- formed and are summarised in Fig. 4. The GUS activity ranged from 5 x 103- 5 x 10 s pmol methylumbelliferrol/mg protein per minute whether or not an intron was present. Hence both the Northern and GUS activity data indicate that the chimeric intron is functional in different plant spe- cies and does not dramatically influence expression.

Expression of the chimaeric construct in Agrobacterium tumefaciens

To test whether or not all prokaryotic GUS enzymatic ac- tivity is abolished by introduction of the portable intron PIV2 into the GUS gene, expression in Agrobacterium was analysed. Overnight cultures of bacterial strains containing

248

106 .

105 .

10 ° . . . . . . . . . . .

• p35S GUS INT

.106

~ P-'A 105

.103 "/~ '/i";, F/A '

. lOO [ ] p35S GUS

Ccttti of independent transformants

Fig. 4. Influence of the intron on GUS ac- tivity in stable transformants. Calli of in- dependent transformants of Arabidopis thaliana containing the GUS gene with (p35S GUS) or without (p35S GUS INT) the intron were analysed quantitatively for GUS activity which was expressed as pico- moles methylumbelliferol per milligram protein per minute. The histogram indi- cates the activity on a logarithmic scale

Fig. 5. In vivo staining of Agrobacterium. GUS activity displayed by Agrobacterium containing either the intron-containing (p35S GUS INT) (right) or the intron-free (p35S GUS) (left) chimaeric GUS gene. Late log cultures of Agrobacterium strains were incu- bated for 24 h with X-gluc (5-bromo-4-chloro-3-indolyl-fl-D-glu- curonic acid cyclohexyl ammonium)

35S GUS or 35S GUS INT were incubated for 24 h with X-gluc a substrate for GUS. Agrobacteria containing the intronless construct showed significant GUS activity, whereas no activity was detected in agrobacteria containing the p35S GUS INT construct (Fig. 5). Thus the intron PIV2 allows efficient discrimination between agrobacterial and plant gene expression.

Detection of early transformation events after Agrobacterium infection

The results described above fulfil all requirements for the further use of the intron-containing construct for monitor- ing early events after Agrobacterium-mediated plant trans- formation. As a first experiment in this direction cotyledon explants of A. thaliana were incubated with the appropriate Agrobacterium strain and treated under the conditions de- scribed (Schmidt and Willmitzer 1988). After defined time intervals samples were frozen at - 70 ° C and used for quan- tification of GUS activity. Figure 6 shows the kinetics of GUS activity obtained by quantitative fluorescence mea- surements of cotyledon extracts. Significant GUS enzymatic activity can be observed as early as 36 h after Agrobacterium infection. During the next few days an exponential increase

1 ~10 3

t/1

L~ 10 2

101 _~- 0 12 2/, 36 48 72 96 120 1/,Z,

Hours post infection

Fig. 6. Kinetics of total GUS activity after infection with an Agro- bacterium strain containing the GUS-INT chimaeric gene. Cotyle- don infections of Arabidopsis were performed with Agrobacterium (p35S GUS INT), omitting the G418 selection. After defined time intervals (hours post infection) ten explants were frozen at - 70 ° C, and used for a quantitative fluorescence assay of cotyledon ex- tracts. The GUS activity expressed as picomoles methylumbelliferol per milligram protein per minute, is indicated on a logarithmic scale

in GUS activity is detectable. The initial increase most probably reflects the relative growth of the transformed tissue compared with that of the rest of the leaf explant. Similar kinetics were obtained when G418 selection was applied during the same time period (data not shown).

In simultaneous transformation experiments cotyledons and roots were stained by X-gluc at different stages after Agrobacterium (p35S GUS INT) infection. The high sensi- tivity of this staining procedure allows the detection of sin- gle cell transformation events. Blue staining was observed in proliferating calli indicating GUS activity especially in cells at the cut surface and in the neighbourhood of vascular tissue (Fig. 7A, B, C, E). Moreover in most of cases staining was observed in subepidermal cell layers (Fig. 7 G, I), al- though epidermal events could also be observed (Fig. 7H, I). Most probably a high proport ion of these cells would not proliferate in contrast to the callus tissue developing on the cut surface. Therefore it is unlikely that such trans- formation events would survive any continuous selection at the single cell stage. Hence these data support the idea that only rapidly dividing and/or regenerating tissue is se- lected for during transformation experiments.

249

Fig. 7A-I. In vivo staining of Agrobacterium-infected cotyledon and root explants. D-E An overview of Arabidopsis roots (D) and cotyledons (E) 40 h post Agrobacterium (p35S GUS INT) infection (p.i.). Cut edges: A-C of cotyledons (A, C) 6 days p.i., (B) 15 days p.i.; (F) of a root 40 h p.i. G-I Close ups of single transformation events: a subepidermal cell (G) 48 h p.i., an epidermal cell (H) 6days p.i., around a stomatal opening (I) 40h p.i. Magnification: A 20x; B, C, 15x; D, 2x ; E, 10x; F, 20x; G, 400x; H, 200 x, 1, 300 x

Discussion

In order to prevent expression of an indicator gene in pro- karyotic organisms, such as A. tumefaciens, and to allow its expression in plants, we interrupted an indicator gene by a plant intron. A portable intron (PIV2) derived from the second intron of the ST-LSI gene was constructed by modification of the splice junctions into appropriate restric- tion sites and introduced into the bacterial gene coding for GUS. In transgenic tobacco plants containing this chi- maeric GUS-intron gene the portable intron is spliced out efficiently, thus resulting in a functional mRNA. Mapping of the splice junction suggested accurate removal of the intron. This indicates that the intron sequence itself com- prises all information necessary for recognition by the splic- ing machinery and hence the intron is independent of the

original neighbouring exon sequences. Using the described in vivo approach the elucidation of processing of pre- mRNAs in plants might be facilitated.

Moreover the inserted intron does not influence the pro- duction of GUS mRNA or the encoded enzymatic activity quantitatively, whereas the expression of the introduced GUS gene in A. tumefaciens is efficiently prevented by the interruption of the open reading frame by the portable in- tron. The chimaeric GUS-intron gene thus allows experi- ments aimed at the analysis of early events during Agrobac- terium-mediated transformation.

As early as 36 h after Agrobacterium infection of Arabi- dopsis cotyledons significant GUS activity could be ob- served. This is in agreement with other experiments using an endogeneous Agrobacterium indicator gene coding for nopaline synthase after leaf disc infection of Petunia hybrida

250

(Horsch et al. 1986). However in the lat ter case endogenous Agrobaeterium expression could not be excluded.

In contras t to nopaline synthetase, G U S furthermore allows the detection of single t ransformed cells by histo- chemical methods. After t ransformat ion single cells express- ing G U S activity were observed throughout the cotyledon explants. In contrast to the tissues at the cut surface, these cells would probably not proliferate during the next few days and hence would not be likely to survive any kind of selection. These da ta indicate that by early selection the actual t ransformat ion frequencies after Agrobacterium-me- diated gene transfer might be underest imated. However the possibil i ty cannot be excluded that the observed G U S activ- ity might be due to transient expression of the introduced GUS gene. Alternative regeneration protocols and a pro- longed lag phase between Agrobacterium infection and the applicat ion of selection could result in the recovery of these minor t ransformat ion events, leading to an improved trans- format ion efficiency. This approach should be helpful in analysing critical steps in t ransformat ion procedures and this in turn could result in new strategies to introduce novel features in recalcitrant plants.

The use of transgenic plants for the p romoter analysis of genes which are specifically expressed at par t icular stages in plant development or induced by par t icular environmen- tal condit ions is time consuming due to the length of regen- erat ion procedures. The use of the por table intron, allowing the discrimination of prokaryot ic versus eukaryotic gene expression in a transient assay, could facilitate new experi- mental strategies for the analysis of plant genes using Agro- bacterium-mediated transient expression in specific p lant tissues.

Acknowledgements. The authors wish to thank C. Recknagel for the tobacco tissue culture work, B. Kiisgen for the greenhouse work, J. Bergstein for the photographic support and J. Stockhaus for providing the p35S GUS construct. We thank B. Damm and S. Prat for critical reading of the manuscript. M.R.S. was sup- ported by a grant from the A. von Humboldt-Stiftung. This work was supported by a BMFT grant BCT 389 (Molekular- and Zell- biologie von h6heren Pflanzen und Pilzen).

References

Anaasino RM (1986) Acceleration of nucleic hybridisation rate by polyethylene glycol. Anal Biochem 152: 304-307

Bartha A, Sommergruber K, Thompson D, Hartmuth K, Matzke MA, Matzke AJM (1986) The expression of a nopaline synthase - human growth hormone chimaeric gene in transformed to- bacco and sunflower callus tissue. Plant Mol Biol 6:347-357

Bevan M (1984) Binary Agrobacterium vectors for plant transfor- mation. Nucleic Acids Res 12:8711-8721

Bradford MM (1979) A rapid and sensitive method for the quanti- fication of microgram quantities of protein utilizing the princi- ple of protein-dye binding. Anal Biochem 72: 248-254

Brown JWS (1986) A catalogue of splice junction and putative branch point sequences from plant introns. Nucleic Acids Res 14:9549-9559

Brown JWS, Feix G, Frendewey D (1986) Accurate in vitro splicing of two pre-mRNA plant introns in a Hela cell nuclear extract. EMBO J 5 : 2749-2758

Deblaere R, Bytebier B, De Greve H, Deboeck F, Schell J, Van Montag M, Leemans J (1985) Efficient octopine Ti plasmid- derived vectors for Agrobacterium-mediated gene transfer. Nuc- leic Acids Res 13:4777-4788

Eckes P, Rosahl S, Schell J, Willmitzer L (1986) Isolation and characterisation of a light-inducible, organ-specific gene from

potato and the analysis of its expression after tagging and transfer into tobacco and potato shoots. Mol Gen Genet 199:216-224

Harmuth K, Bartha A (1986) In vitro processing of a plant pre- mRNA in a Hela cell nuclear extract. Nucleic Acids Res 14: 7513-7529

Hawkins JD (1988) A survey on intron and exon lengths. Nucleic Acids Res 16:9893-9908

H6fgen R, Willmitzer L (1988) Storage of competent cells for Agro- bacterium transformation. Nucleic Acids Res 16: 9877

Horsch RB, Klee HJ (1986) Rapid assay of foreign gene expression in leaf disc transformed by Agrobacterium tumefaeiens: Role of T-DNA borders in the transfer process. Proc Natl Acad Sci USA 83:4428-4432

Horsch RB, Fry JE, Hoffmann NL, Eichholtz D, Rogers SG, Fra- ley RT (1985) A simple and general method for transferring genes into plants. Science 227:1229-1231

Jefferson RA (1987) Assaying chimeric genes in plants: the GUS gene fusion system. Plant Mol Biol Rep 5 : 387-405

Jefferson RA, Kavanagh TA, Bevan M (1987) GUS fusions: fl- glucuronidase as a sensitive and versatile gene fusion marker in higher plants. EMBO J 6 : 3901-3907

Kingston RE (1987) In: Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith JA, Struhl K (eds) Current protocols in molecular biology, pp 4.8.1-4.8.3

Klee H, Horseh R, Rogers S (1987) Agrobacterium-mediated plant transformation and its further applications to plant biology. Annu Rev Plant Physiol 38:467-486

Logemann J, Schell J, Willmitzer L (1987) Improved method for the isolation of RNA from plant tissues. Anal Biochem 163 : 21-26

Maniatis T, Fritsch EF, Sambrook J (1982) Molecular cloning - a laboratory manual. Cold ptring Harbor Laboratory Press, Cold Spring Harbor, NY

Saiki RK, Gelfand DH, Stoffel S, Scharf S J, Higuchi R, Horn GT, Mullis KB, Erlich HA (1988) Primer-directed enzymatic amplification of DNA with thermostable DNA polymerase. Science 239:487-491

Schell J (1987) Transgenic plants as tools to study the molecular organisation of plant genes. Science 237:1176-1183

Schmidt R, Willmitzer L (1988) High efficienty Agrobacterium tu- mefaciens-mediated transformation of Arabidopsis thaliana leaf and cotyledon explants. Plant Cell Rep 7 : 583-586

Shapiro MB, Senepathy P (1987) RNA splice junctions of different classes of eucaryotes: sequence statistics and functional implica- tions in gene expression. Nucleic Acids Res 15 : 7155-7174

T6pfer R, Pr61s M, Schell J, Steinbiss H-H (1987) A set of plant expression vectors for transcriptional and translational fusions. Nucleic Acids Res 15:5890

Valvekens D, Van Montagu M, Van Lijsebettens M (1988) Agro- bacterium tumefaciens-mediated transformation of Arabidopsis root explants by using kanamycin selection. Proc Natl Acad Sci USA 85 : 5536-5540

Vervliet G, Holsters M, Teuchy H, Van Montagu M, Schell J (1975) Characterisation of different plaque-forming and defec- tive temperate phages in Agrobacterium strains. J Gen Virol 26: 33-48

Wiebauer K, Herrero J-J, Filipowicz W (1988) Nuclear pre-mRNA processing in plants: Distinct modes of 3'-splice-site selection in plants and animals. Mol Cell Biol 8:2042-2051

Zambrisky P, Temp6 J, Schell J (1989) Transfer and function of T-DNA genes from Agrobacterium Ti and Ri plasmids in plants. Cell 56:193-201

Communica ted by J. Sche11

Received August 10, 1989