Transgenic Research9: 405–415, 2000.© 2000Kluwer Academic Publishers. Printed in the Netherlands.

405

Efficient production of transgenic cassava using negativeand positive selection

Peng Zhang1,2, Ingo Potrykus1 & Johanna Puonti-Kaerlas1,∗1Institute for Plant Sciences, ETH-Zentrum/LFW E 17, CH-8092 Zürich, Switzerland2South China Institute of Botany, Academia Sinica, Guangzhou, China

Received 24 February 2000; revised 18 May 2000; accepted 24 July 2000

Key words: Manihot esculenta, transformation,Agrobacterium tumefaciens, mannose, hygromycin, selection

Abstract

In order to improve the efficiency of cassava (Manihot esculentaCrantz) transformation, two different selectionsystems were assessed, a positive one based on the use of mannose as the selective agent, and a negative one basedon hygromycin resistance encoded by an intron-containinghph gene. Transgenic plants selected on mannose orhygromycin were regenerated for the first time from embryogenic suspensions cocultivated withAgrobacterium.After the initial selection using mannose and hygromycin, 82.6% and 100% of the respective developing embryo-genic callus lines were transgenic. A system allowing plant regeneration from only transgenic lines was designedby combining chemical selection with histochemical GUS assays. In total, 12 morphologically normal transgenicplant lines were produced, five using mannose and seven using hygromycin. The stable integration of the transgenesinto the nuclear genome was verified using PCR and Southern analysis. RT-PCR and northern analyses confirmedthe transgene expression in the regenerated plants. A rooting test on mannose containing medium was developedas an alternative to GUS assays in order to eliminate escapes from the positive selection system. Our results showthat transgenic cassava plants can be obtained by using either antibiotic resistance genes that are not expressed inthe micro-organisms or an antibiotic-free positive selection system.

Abbreviations: BA – 6-benzylaminopurine; CBM – cassava basic medium; CEM – cassava elongationmedium; CSM – cassava selection medium; FEC – friable embryogenic callus; IBA – indole-3-butyric acid;MSN – friable embryogenic callus formation and embryo conversion medium; NAA –α-naphthalene acetic acid;PMI – phosphomannose isomerase; SH – cassava suspension culture medium; SCV – settled cell volume

Introduction

Cassava (Manihot esculentaCrantz) is a perennialtropical root crop that serves as a staple food for morethan 500 million people worldwide (for a review seePuonti-Kaerlas, 1998). Traditional breeding of cassavais difficult and genetic engineering has the potentialto complement traditional breeding methods. The re-cent development of transformation methods (Li et al.,1996; Raemakers et al., 1996; Schöpke et al., 1996;González et al., 1998) will allow, in the near fu-ture, the addition of agriculturally valuable traits for

∗ Author for correspondence:Tel.: + 41 632 22 44;Fax: + 41 63210 44;E-mail: jpuonti@ipw.biol.ethz.ch

disease resistance and improved root quality but im-provements are still needed before gene technologycan be used efficiently.

The currently available methods for selectingtransgenic cassava tissues are based on antibiotic se-lection using hygromycin, paromomycin or geneticin(Li et al., 1996; Schöpke et al., 1996; González etal., 1998; Zhang et al., in press) or on visual selec-tion using firefly luciferase as a screenable markergene (Raemakers et al., 1996) or on a combinationof antibiotic selection and luciferase screening (Mun-yikwa et al., 1998). However, such selection systemsare costly, time and labour intensive or deleterious tothe regeneration process. They also present a limited

406

Figure 1. Schematic representation of the T-DNA region of the binary vector pHMG. RB – right border; LB – left border; p35S CaMV 35Spromoter, 35! and Nos! CaMV 35S andAgrobacteriumnopaline gene terminator, respectively. For gene abbreviations see text.

efficiency by allowing the development of nontrans-genic shoots escaping selection. In addition, the useof antibiotic resistance genes in transgenic plants hasbecome an important factor for decision making as aresult of public concern (Ferber, 1999). Therefore, thefuture development of transformation methods shouldmove towards the independent segregation of the se-lectable markers and transgenes of interest at meiosis,or the use of non-antibiotic selection systems. Asthe current methodologies for the former approachare not immediately applicable to cassava, we choseto focus our research on a positive selection systemas an alternative to antibiotic-based selection and toalso study the use of antibiotic resistance genes notexpressed in micro-organisms. Positive selection sys-tems using either xylose (Haldrup et al., 1998a,b) ormannose (Joersbo et al., 1998, 1999) have been re-cently shown to be more efficient for potato, tomatoand sugar beet transformation than methods based onantibiotic selection. Many plant species cannot meta-bolise mannose, and its uptake leads to starvation andsevere growth inhibition caused by the accumulationof mannose-6-phosphate in the cells (Ferguson et al.,1958, Malca et al., 1967). Plant cells expressing theE. coli phosphomannose isomerase (PMI) gene canconvert mannose-6-phosphate to easily metabolisablefructose-6-phosphate and are thus able to use mannoseas carbohydrate source. We show here that a mannoseselection system as well as an antibiotic-based systemusing a hygromycin resistance gene not expressed inmicro-organisms are applicable to the production oftransgenic cassava plants from embryogenic suspen-sions.

Materials and methods

Plant material

Cassava cultivar TMS60444 (MNig11) was main-tained as shoot cultures on CBM (MS medium (Mur-ashige & Skoog, 1962) supplemented with 2% sucrose

and 2µM CuSO4, solidified with 0.6% agar, pH 5.8)and subcultured at four-week intervals. All plant ma-terial was cultured at 26◦C under 18/6 photoperiod(90–110µmol m−2 s−1, TRUE-LITEr ) unless oth-erwise stated. Somatic embryos were induced fromapical meristems and 1–6 mm long immature leaflobes on CIM (CBM with 12 mg l−1 picloram) in thedark and maintained as cycling somatic embryos asdescribed (Li et al., 1998). Friable embryogenic cal-lus (FEC) was induced from organised embryogenicclusters on GD medium (Gresshoff & Doy, 1974) asdescribed (Taylor et al., 1996). The FEC was sub-cultured at three-week intervals. Suspensions wereinitiated by inoculating 2-year-old FEC (0.5 g freshweight) in 25 ml SH medium (SH salts (Schenk &Hildebrandt, 1972), MS vitamins, 12 mg/l picloram,6% sucrose) in 190 ml jars (Greiner Labortechnik).The suspensions were cultured on a gyratory shaker(108 rpm) at 28◦C under continuous light (∼50µmolm−2 s−1) and subcultured at three-day intervals. Thecultures were sieved through a metallic net (φ500µm)every 4 weeks to obtain the fraction consisting ofembryogenic units ranging in size from 250–500µm.

Bacterial strains and plasmids

Agrobacterium tumefaciensstrains LBA4404 (Hoekemaet al., 1983) and GV3101 (Deblaere et al., 1985)carrying the binary vector pHMG (Figure 1) wereused for genetic transformation experiments. The plas-mid pHMG was constructed by subcloning in pLD10(Joersbo et al., 1998) containing thepmi gene aHindIII/ ClaI fragment from pWBvec10a (Wang etal., 1997) containing the intron-interruptedhph anda HindIII fragment from pSH9GusInt (Holtorf et al.,1995) carrying the intron-interruptedgusA. All thegenes were driven byCaMV35Spromoter. The plas-mid was transferred toAgrobacteriumby electropor-ation (Mattanovich et al., 1989). A single colonyof Agrobacteriumwas incubated overnight in 5 mlliquid YEB medium with 25 mg l−1 rifampicin and

407

100 mg l−1 spectinomycin on a shaker (240 rpm) at28◦C. An aliquot of the bacterial suspension (25µl)was transferred to 50 ml fresh medium and culturedfor 12–20h to an OD600 0.5–1.0. The bacteria werecentrifuged at 6,000 rpm 4◦C for 10 min, washed oncewith an equal volume of liquid MS medium (pH5.3) and centrifuged again. The bacterial pellet wasfinally resuspended in liquid MS medium supplemen-ted with 200µM acetosyringone. TheAgrobacteriumcultures were grown for 2–5 h at 28◦C (80 rpm) forpreinduction before inoculation of the suspensions.

Dose response tests

The efficiency of mannose in arresting the growth ofsuspension cultures was assessed by adding 0, 0.5,0.75, 1, 2, 3 or 4% mannose to the SH medium.Different concentrations of sucrose were used in or-der to investigate the optimum combination betweensucrose and mannose. In one set of experiments theosmotic pressure of the medium was balanced by sor-bitol (MW 182.2) to equal the osmotic pressure of 6%sucrose. The medium was refreshed at three-day inter-vals and the suspension cultures were weighed every3–6 days. There were three replicas per treatment, andthe experiments were repeated three times.

The effect of hygromycin on the viability of thesuspension cells was assessed by culturing suspen-sions in SH containing 0, 15, 25 and 50 mg l−1

hygromycin. The suspensions were subcultured atthree-day intervals and the viability of the embryo-genic units was assessed using fluorescein diacetate(Withers, 1985) and a ZEISS ICM405 microscope un-der UV light during a 4-week period. One Hundredembryogenic units were scored every time for eachtreatment.

Transformation

Aliquots of 2 ml settled cell volume (SCV) fromembryogenic suspensions were used for inoculationwith Agrobacteriumand transferred to jars contain-ing 20–30ml of bacterial suspension. After 0.5–1 h,the bacterial suspension was removed with a pipette.The inoculated tissues were spread onto sterile filterpapers (φ9 cm) and transferred to Petri dishes withsolidified SH medium supplemented with 100µMacetosyringone and co-cultivated at 26◦C.

Selection and regeneration of transgenic plants

After 3–4 days co-cultivation, the infected tissue wastransferred to SH medium with 500 mg l−1 carbenicil-

lin and cultured for 3 days on a shaker at 137 rpm.Then the suspensions were cultured further in a liquidmedium containing either hygromycin or mannose andsubcultured at three-day intervals. Hygromycin at aconcentration of 50 mg l−1 was applied at first for15 days in SH medium with 500 mg l−1 carbenicillin.Then the concentration of hygromycin was reducedto 25 mg l−1. After another 15 days the suspensionswere transferred to MSN medium (MS salts and vit-amins, 1 mg l−1 NAA, 2% sucrose, 0.6% agar, pH 5.8)with 25 mg l−1 hygromycin and 500 mg l−1 carben-icillin for FEC formation. The resistant FEC lineswere transferred individually onto fresh MSN contain-ing 25 mg l−1 hygromycin for cotyledon emergence.Shoots were regenerated from the cotyledonary stageembryos on CEM medium (CBM with 0.4 mg l−1 BAwithout hygromycin) and maintained as shoot cultureson CBM.

Mannose at a concentration of 4% in a modifiedSH medium (SH with 1% sucrose) with 500 mg l−1

carbenicillin was used to select transformed suspen-sion cells. After 4 weeks the transformed suspensionswere transferred to solid MSN medium containing 2%mannose to induce FEC development. After 2 weeks,the FEC lines were transferred to MSN without man-nose to induce embryo development. After 4 weeks,the developing embryos were transferred to CEM forplant regeneration.

β-glucuronidase assays

GUS assays were performed by placing tissues in anassay buffer (0.3% X-Gluc 10 mM Na2EDTA·H2O,0.1% Triton X-100, 0.1 M NaH2PO4, 0.5 MK3Fe(CN)6) (Jefferson, 1987). After 3–6 h incubationat 37◦C, the tissues were washed several times with95% ethanol and stored in 95% ethanol.

Mannose screen tests

Shoot cultures of wild type TMS60444 were culturedon CBM medium containing 0, 0.5, 0.75, 1, 1.5 or2% mannose. The number of rooted shoots was re-corded after 2 and 3 weeks. Each treatment was donein triplicate with seven explants. The experiment wasrepeated three times. Shoots from transgenic lineswere cultured on CBM medium supplemented with1% mannose and after 3 weeks the number of rootedlines was recorded. We also assessed the growth ofshoot tips from transgenic plants on CBM mediumwithout sucrose but supplemented with various con-centrations of mannose. Two lines transformed with

408

Agrobacterium(A20 and B44) and two lines trans-formed using biolistics and the plasmid pHMG (BC5and BL7) (Zhang & Puonti-Kaerlas, in press), show-ing different expression levels of the PMI gene innorthern analyses were used. The shoots of wild typeTMS60444 were used as a negative control. The freshweight of the shoots was measured after 4 weeks ofculture.

Molecular analysis

Genomic DNA was isolated from freeze-dried plantmaterial according to Soni and Murray (1994) anddigested withKpnI. Total RNA was isolated fromleaves ofin vitro-grown plants using the RNeasy plantmini kit (Qiagen) following the instructions of themanufacturer. PCR, RT-PCR, Southern and northernanalyses were carried out following standard proto-cols (Sambrook et al., 1989). Aliquots of 10µg DNAand 5µg RNA, respectively, were loaded per lanefor Southern and northern analysis. The hybridisation

Figure 2. Effect of mannose on the growth of embryogenic cell suspensions in media with different saccharide compositions expressed asweight increase during culture: (a) 6% sucrose, (b) 1% sucrose and 10% sorbitol, (c) 1% surcose, M0-M4, 0–4% mannose. The error barsdenote the standard error of the mean.

probes (hph 450 bp, pmi 650 bp) were DIGdUTP-labelled by PCR using a PCR DIG probe synthesis kit(Boehringer Mannheim GmbH, Germany) accordingto the manufacturer’s instructions.

Results

Dose response tests

Adding mannose to the normal suspension culture me-dium, even at concentrations of up to 4%, did notarrest the growth of the suspensions efficiently (Fig-ure 2a). Therefore, we tested the effect of reducing thesucrose concentration of the medium to 1%, with andwithout balancing the osmotic pressure of the mediumwith sorbitol to equal the osmotic pressure obtainedwith 6% sucrose. When the sucrose concentration wasreduced to 1%, mannose could be used to suppress thegrowth of the suspensions (Figure 2b, c). The controlsuspensions cultured without mannose were able to

409

Table 1. The effect of hygromycin on the viability of cassava em-bryogenic suspension cells after 2, 3 and 4 weeks of culture. Theviability is given as per cent of live cells of all cells

Hygromycin Viability (%)

concentration Second week Third week Fourth week

(mg l−1)

0 100 100 100

15 98.3± 2.1 83.7± 4.2 43.7± 2.1

25 80.3± 4.0 46.3± 8.5 15.7± 2.1

50 62.0± 4.0 17.7± 4.0 4.0± 1.0

grow both in a medium supplemented with sorbitoland in that containing only 1% sucrose.

The viability of embryogenic cell suspensions de-creased with increasing hygromycin concentrationsand culture periods (Table 1), and 25 mg l−1 and50 mg l−1 hygromycin were sufficient to kill 85% and96% of the cells, respectively, after 4 weeks in culture.Under the hygromycin treatment, yellowish suspen-sion cells changed to colourless units which were

Figure 3. Transgenic cell suspensions selected on hygromycin (a, b) and mannose (c, d); (a, c) GUS expression in suspensions after 3 weeksunder selection, (b, d) development of resistant FEC (arrows) on solid selection medium.

easily distinguishable from control suspension grownwithout selection.

Regeneration of transgenic plants

After 3 days of co-cultivation, a fraction of the in-oculated suspension cells was tested in transient GUSassays. On the average, 70±8.1% of the cell clusterswere GUS positive. Both LBA4404 and GV3101could effectively infect and transform the cells (datanot shown).

Negative selectionUnder hygromycin selection, the antibiotic-resistantcell suspensions developed into yellowish, friableembryogenic callus indistinguishable from controlFEC grown under nonselective conditions. After 3weeks cultivation in selective medium the cell clustersstained blue in GUS assays (Figure 3a). The yel-low resistant lines were easily distinguishable from abackground of white and brown-coloured dead tissue

410

Figure 4. GUS assays: nontransgenic (bottom right) and three GUS expressing transgenic cassava lines.

(Figure 3b). Of 689 hygromycin-resistant clusters ob-tained from 12 ml (SCV) inoculated suspensions, 138were randomly selected and transferred to MSN onwhich they developed to yellow, hygromycin-resistantFEC lines. All these FEC lines were positive forGUS expression. The FEC developed into embryoson MSN medium within 4–8 weeks. The embryoswere transferred to CEM medium without hygromycinfor plant regeneration. Using hygromycin, 49 embryolines were regenerated on CEM. Finally, seven plantlines were regenerated from these embryo lines, all ofwhich were GUS positive (Figure 4).

Positive selection

Under mannose selection, most of the yellowish sus-pension cells turned white after 3 weeks in culture.In the GUS assays, the pattern of GUS staining ofsuspension cells selected with mannose was differ-ent from that in the cells under hygromycin selection(Figures 3c and 3a, respectively). Diffuse small bluefoci were observed in the cell suspensions growingunder mannose selection, while larger blue clustersdeveloped in suspensions growing under hygromycinselection. In total, 178 randomly selected FEC lines of

the 664 mannose resistant units obtained from 12 ml(SCV) inoculated suspension cells were transferredto MSN medium containing 2% mannose. On aver-age 82.6% of the developing FEC developing on thismedium was positive for GUS expression. Only theGUS positive mannose-resistant callus lines showingthe characteristics of friable embryogenic callus (Fig-ure 3d) were transferred to regeneration medium. Asthe FEC was not able to form embryos under mannoseselection, mannose was omitted at this stage. Finally,five plant lines were regenerated from 43 embryo lines.All these plant lines were positive for GUS expression.

Rooting and growth assays

The rooting of wild type TMS60444 shoots was inhib-ited completely on a medium containing 1% mannose(Table 2), whereas all the transgenic shoots couldproduce a normal root system. After 4 weeks cul-ture on mannose containing medium without sucrose,differences were observed in the growth rate of thetransgenic lines (Figure 5). In line A20, plant growthmeasured as weight increased with increasing man-nose concentration up to 4%, while lines BC5 and BL7grew more slowly on mannose concentrations higher

411

Figure 5. The effect of different mannose concentrations on the growth of wild type (WT) and four transgenic (A20, B44, BC5 and BL7)cassava lines; IFW initial fresh weight. The error bars denote the standard error of the mean.

Table 2. The effect of mannose on the rooting of wildtype cassava shoots

Mannose Rooting frequency (%)

concentration (%) First week Second week

0 100 100

0.5 52.4± 8.2 85.7± 14.3

0.75 0 42.7± 14.2

1 0 0

1.5 0 0

2 0 0

than 1%. In line B44 the response was between thatof A20 and BC5 and BL7, and the highest growthrate was observed on 3% mannose. The weight in-crease was concomitant with accelerated growth rateof the plants, and was not due to abnormal growth orvitrification.

Molecular analysis

PCR analysis of all 12 putative transgenic plant linesusing thehphandpmi specific primers resulted in theamplification of the expected 450 and 650 bp frag-ments, respectively (data not shown). The Southernblot analyses on the DNA isolated from 10 hygromy-

cin or mannose resistant cassava lines and digestedwith KpnI, confirmed the stable integration of thetransgenes in the nuclear genome. No signal was de-tected in the DNA from nontransgenic plants used asnegative controls (Figure 6a, c). The copy numberof integrated transgenes varied from one to three andmost of the transgenic lines had one or two copies. Nodifference in the ratio of single copy insertion could beobserved between the lines selected on mannose andthose selected on hygromycin. In two plant lines (lanes4 and 5) the two probes indicated a different num-ber of integrated copies. The results of RT-PCR of all12 lines usingpmi andhphspecific primers (data notshown) and northern analyses of 10 plant lines showedthat the transgenes were transcriptionally active in theregenerated plants (Figure 6b, d).

Discussion

Our aim was to assess the efficiency of mannose se-lection in cassava transformation and to compare itwith an antibiotic selection system using hygromycin.The expression vector pHMG containspmi and hphgenes on the same plasmid (Figure 1), which made it

412

Figure 6. Molecular analysis of transgenic cassava lines: (a) South-ern and (b) northern blot usinghph probe: (c) Southern and (d)northern blot usingpmi probe; lanes 2–7 lines selected under hy-gromycin; lanes 8–11 lines selected under mannose; P – plasmidcontrol; lane 1 negative control.

possible to evaluate the two different selection systemsdirectly within one experiment.

We show here that mannose selection can be usedsuccessfully for the production of transgenic cassavaplants. The concentration of mannose required to ar-rest the growth of wild type suspension cells wasabout two times higher than that required for inhibit-ing organogenesis from somatic cotyledons of cassava(Zhang & Puonti-Kaerlas, in press) and 10 timeshigher than that reported for sugar beet (Joersbo etal., 1998). In rice, the concentrations needed to ar-rest growth of embryogenic calli are as high or even

higher than those in cassava (P. Lucca, pers. comm).In sugar beet, a stepwise increase from a permissiveconcentration of 0.1–0.15%, allowing 20–30% of theexplants to produce shoots, to a selective one of 0.25%and finally 1%, was necessary in order to allow re-generation of transformants (Joersbo et al., 1998). Thenontransgenic shoots were eliminated by selection ona medium containing 1% mannose. We did not testa stepwise selection scheme in our experiments, butassessed the efficiency of different combinations ofmannose and sucrose for initial selection.

Mannose selection of cassava suspensions from thebeginning of the selection period was efficient onlywhen the sucrose concentration was reduced to 1%(Figure 2b, c). The reduction of the sucrose contentof the medium from 6% to 1%, however, also affectsits osmotic pressure. As the reduced osmolarity itselfmight have an adverse effect on the growth of the sus-pensions, we tested two versions of this medium, onein which the osmolarity was balanced with additionalsorbitol and one containing only 1% sucrose. Thecontrol suspensions cultured without mannose wereable to grow normally in both media (Figure 2b, c),showing that the adjustment of the osmotic potentialof the selective medium is not necessary. A study oninteractions between mannose and sucrose, as wellas between mannose and other saccharides in sugarbeet (Joersbo et al., 1998; 1999) showed that the toxiceffect of mannose increases with decreasing sucroseconcentration. Omitting sucrose completely, however,drastically reduced the regeneration and transforma-tion rates of cassava suspensions. In sugar beet, thehighest transformation frequency was obtained when3% sucrose was included in the selection medium.Optimal concentrations could be also determined forfructose and maltose, but not for glucose (Joersboet al., 1999). In cassava suspensions the use of 1%sucrose resulted in the highest selection efficiency.The pattern of GUS staining of the suspensions undermannose selection was different from that under hy-gromycin selection (Figure 3a, c). The blue foci weresmaller and more widely spread when mannose wasused, while larger blue clusters developed under hy-gromycin selection. This may indicate that in contrastto hygromycin, mannose also affects the growth oftransformed cells adversely and low concentrations ofsucrose in the medium are required to counteract thisgrowth suppression.

The transgenic units can develop to FEC lines onsolid MSN medium containing 2% mannose. How-ever, the FEC was not able to form embryos un-

413

der selection, and hence mannose was omitted fromthe medium during embryo induction. The GUS as-says showed that at this stage over 80% of themannose resistant lines were expressinggusA, andonly these were selected for further culturing. Allthe subsequently regenerated plants were GUS pos-itive, proving that the combination of mannose se-lection with GUS assays resulted in 100% selectionefficiency, in contrast to paromomycin selection un-der which 29–56% of the regenerated plants wereescapes (González et al., 1998). The selection ef-ficiency of phosphinotricin could likewise be im-proved by combining visual screening using fireflyluciferase with chemical selection (Munyikwa et al.,1998). Luciferase screens are, however, laborious,costly and time consuming, whereas the combinationof GUS assays with chemical selection allows easyselection of uniformly transformed suspensions forregeneration.

In order to refine the system further so as to elim-inate the requirement for an additional marker gene(gusA) we also aimed to develop a mannose-basedrooting assay for increasing the selection efficiency.Rooting tests based on antibiotics have been usedas a secondary screen, for example in apple (Puite& Schaart, 1996), pea (Grant et al., 1998), silverbirch (Keinonen-Mettälä et al., 1998) and cassava(Zhang et al., in press) in order to eliminate escapesafter the primary selection. The rooting of wild typeTMS60444 shoots was totally inhibited by 1% man-nose (Table 2), whereas all the transgenic plants de-veloped a normal root system, showing the efficiencyof mannose for a secondary selection. A mannose con-centration of 1% also inhibits the growth on nontrans-genic shoots, while the transgenic plants can grow welleven on media containing up to 4% mannose. Thedifferences in the growth of the transgenic lines toincreasing mannose concentrations could reflect dif-ferent expression levels of the PMI gene. In line A20,which has the highest transcription levels (Figure 6d,lane 2), the increase in growth rate correlated withincreasing mannose concentration. For lines BC5 andBL7, in which the transcription level is very low (datanot shown), mannose concentrations above 1% wereslightly inhibitory to growth, while B44 which showedintermediate expression levels (Figure 6d, lane 7) hadan optimum at 3% mannose. Mannose has been shownto be a better carbohydrate source than sucrose allow-ing faster growth rates in transgenic sugar beet, and itseffect has been correlated to PMI activity (Joersbo etal., 1998).

Hygromycin has been reported earlier to be inef-ficient for selection of cassava suspensions (Schöpkeet al., 1996). In contrast we show that an efficientselection of transgenic suspension cultures is possibleusing hygromycin. This discrepancy is probably due tothe short duration of the earlier experiment, where theviability was assessed already after one week. In ourhands, the viability of embryogenic cell suspensionswas reduced with increasing hygromycin concentra-tions and culture periods, and 50 mg l−1 hygromycinkilled 96% of the suspension cells within 4 weeks.Under hygromycin selection, antibiotic-resistant cellsuspensions developed into yellowish and friable em-bryogenic calli indistinguishable from control FECgrown under nonselective conditions. All the resist-ant colonies were expressinggusA,which shows thathygromycin selection is more efficient than paromo-mycin or phosphinothricin selection (González et al.,1998; Munyikwa et al., 1998).

The introns ingusAandhphprevent the expressionof these genes inAgrobacteriumand other micro-organisms (Vancanneyt et al., 1990; Wang et al.,1997). We have shown earlier that the second intron(IV2) of potato ST-LS1 gene (Eckes et al., 1986; Van-canneyt et al., 1990) is correctly spliced in cassavaallowing efficient gene expression in transgenic cells(Puonti-Kaerlas et al., 1997). Our results now show,that the castor bean catalase-1 (CAT-1) intron in pW-Bvec10a, originally designed for transformation ofmonocotyledonous plants (Wang et al., 1997), is alsocorrectly and efficiently spliced in cassava as indicatedby the RT-PCR, northern analysis and the resistanceof the transgenic lines to hygromycin. The presenceof an intron in thehphgene, however, was not suffi-cient to allow the use of hygromycin as sole agent tocontrol the growth ofAgrobacterium. Especially whenGV3101 was used, the addition of carbenicillin wasessential in order to prevent bacterial proliferation dur-ing early selection. Although the same phenomenonhas been observed for rice, the reasons for this arenot clear (Wang et al., 1997). Possibly the plant cellsdetoxify the selective agent in the medium to sucha degree that its effective concentration no longersuffices to prevent bacterial growth.

Southern blot analysis has confirmed the stable in-tegration of the transgenes in the nuclear genome ofthe regenerated plants (Figure 6a, c). The minimumexpected fragment size using thehphandpmi probesis about 5.2 and 2.5 kb, respectively. As there is onlyoneKpnI site in the plasmid vector, the secondKpnIsite must be derived from the cassava genomic DNA,

414

and the number of hybridising bands thus indicates thenumber of integrated copies. The copy number variedfrom 1 to 3 in the transgenic lines, and almost halfof the tested lines contained only one copy. The copynumber of integrated genes reported here is lower thanin earlier studies usingAgrobacterium-mediated trans-formation (Li et al., 1996; González et al., 1998), andthe percentage of plants containing single copy inser-tions is higher. In plant lines 4 and 5 the use ofhphandpmi probes indicated a different number of integratedcopies, which may be due to rearrangements of theT-DNA. Transgene expression in the micropropagatedmaterial has so far been stable in all the tested lines.

We demonstrate here that positive selection usingmannose is compatible with cassava transformation. Insugar beet, the mannose selection system was reportedto result in 10-fold higher transformation frequen-cies compared to kanamycin selection (Joersbo et al.,1998), but in our study on cassava the efficiency ofhygromycin selection is about the same as that of man-nose selection. Future modifications to the mannoseselection protocol, for example stepwise increases ofmannose concentration, changing the saccharide com-position of the medium or increasing the phosphatecontent, which have been shown to improve the trans-formation and selection efficiency with sugar beet(Joersbo et al., 1999), might further improve the trans-formation frequencies in cassava. On the other hand,GUS assays and the newly developed rooting assayprovide a rapid and easy screen for elimination of non-transgenic material, thus increasing the overall trans-formation efficiency. The main constraint is the lowregeneration capacity of the embryogenic suspensions,as so far plants cannot be obtained from all transgeniclines (Schöpke et al., 1997; Munyikwaet al., 1998;González et al., 1998). We are currently trying toimprove the plant production rate from embryogenicsuspensions by assessing the possibility of combin-ing embryo maturation with the organogenesis-basedregeneration mode.

Acknowledgements

We would like to thank Dr. M.B. Wang, CSIROPlant Industry, Canberra, for the generous gift of pW-Bvec10a and Dr. J. Brunstedt, Danisco Biotechnology,Copenhagen, for the pLD10. This project was partiallysupported by the Swiss Development Agency (SDC)through the Swiss Centre of International Agriculture(ZIL).

References

Deblaere R, Bytebier B, De Greve H, Deboeck F, Schell J, VanMontagu M and Leemans J (1985) Efficient octopine Ti plasmid-derived vectors forAgrobacterium-mediated gene transfer toplants.Nucl Acids Res13: 4774–4778.

Eckes P, Rosahl S, Schell J and Willmitzer L (1986) Isolation andcharacterisation of a light-inducible, organ-specific gene frompotato and its expression after tagging and transfer into tobaccoand potato shoots.Mol Gen Genet199: 216–224.

Ferber D (1999) GM crops in the cross hairs.Science286: 1662–1666.

Ferguson JD, Street HE and David SB (1958) The carbohydratenutrition of tomato roots. IV. The inhibition of excised rootgrowth by galactose and mannose and its reversal by dextroseand xylose.Ann Bot22: 525–538.

González AE, Schöpke C, Taylor NJ, Beachy RN and FauquetCM (1998) Regeneration of transgenic cassava plants (ManihotesculentaCrantz) throughAgrobacterium-mediated transform-ation of embryogenic suspension cultures.Plant Cell Rep17:827–831.

Grant JE, Cooper PA, Gilpin BJ, Hoglund SJ, Reader JK, Pither-Joyce MD and Timmerman-Vaughan GM (1998) Kanamycinis effective for selecting transformed peas.Plant Sci 139:159–164.

Gresshoff P and Doy C (1974) Derivation of a haploid cell linefrom Vitis vinifera and the importance of the stage of meioticdevelopment of the anthers for haploid culture of this and othergenera.Z Pflanzenphys73: 132–141.

Haldrup A, Petersen SG and Okkels FT (1998a) Positive selection:a plant selection principle based on xylose isomerase, an enzymeused in the food industry.Plant Cell Rep18: 76–81.

Haldrup A, Petersen SG and Okkels FT (1998b) The xylose iso-merase gene fromThermoanaerobacterium thermosulfurogenesallows effective selection of transgenic plant cells using D-xyloseas the selection agent.Plant Mol Biol37: 287–296.

Hoekema A, Hirsch PR, Hoykaas P and Schilperoort R (1983) Abinary plant vector strategy based on a separation ofvir- and theT-region of theAgrobacterium tumefaciensTi-plasmid.Nature303: 179–180.

Holtorf S, Apel K and Bohlmann H (1995) Comparison of dif-ferent constitutive and inducible promoters for the overexpres-sion of transgenes inArabidopsis thaliana. Plant Mol Biol 29:637–646.

Jefferson RA (1987) Assaying chimeric genes in plants: the GUSgene fusion system.Plant Mol Biol Rep5: 387–405.

Joersbo M, Donaldson I, Kreiberg K, Guldager Petersen S,Brunstedt J and Okkels FT (1998) Analysis of mannose selectionused for transformation of sugar beet.Mol Breed4: 111–117.

Joersbo M, Petersen SG and Okkels FT (1999) Parameters inter-acting with mannose selection employed for the production oftransgenic sugar beet.Physiol Plant105: 109–116.

Keinonen-Mettälä K, Pappinen A and von Weissenberg K (1998)Comparisons of the efficiency of some promoters in silver birch(Betula pendula). Plant Cell Rep17: 356–361.

Li H-Q, Sautter C, Potrykus I and Puonti-Kaerlas J (1996) Genetictransformation of cassava (Manihot EsculentaCrantz) NatureBiotech14: 736–740.

Li H-Q, Huang YW, Liang CY, Guo JY, Liu HX, Potrykus I andPuonti-Kaerlas J (1998) Regeneration of cassava plants via shootorganogenesis.Plant Cell Rep17: 410– 414.

Malca I, Endo RM and Long MR (1967) Mechanism of gluc-ose counteraction of inhibition of root elongation by galactose,mannose, and glucosamine.Phyopathol57: 272–278.

415

Mattanovich D, Rueker F, da Camara Machado A, Laimer M,Regner F, Steinkeller H, Himmler G and Katinger H (1989) Ef-ficient transformation ofAgrobacteriumspp. by electroporation.Nucl Acids Res17: 6747–6752.

Munyikwa TRI, Raemakers KCJM, Schreuder M, Kok R, Schip-pers M, Jacobsen E and Visser RGF (1998) Pinpointing towardsimproved transformation and regeneration of cassava (ManihotesculentaCrantz).Plant Sci135: 87–101.

Murashige T and Skoog F (1962) A revised medium for rapidgrowth and bioassays with tobacco tissue cultures.Physiol Plant15: 473–497.

Puite KJ and Schaart JG (1996) Genetic modification of the com-mercial apple cultivar Gala, Golden Delicious and Elstar viaanAgrobacterium tumefaciens-mediated transformation method.Plant Sci119: 125–133.

Puonti-Kaerlas J (1998) Cassava biotechnology. In: Tombs MP (ed.)Biotechnology and Genetic Engineering Reviews.Vol. 15 (pp.329–364) Intercept, Andover, Hants, UK.

Puonti-Kaerlas J, Li H-Q, Sautter C and Potrykus I (1997)Production of transgenic cassava (Manihot esculentaCrantz)via organogenesis andAgrobacterium-mediated transformation.African J Root Tuber Crops2: 181–186.

Raemakers CJJM, Sofiari E, Taylor N, Henshaw G, JacobsenE and Visser RGF (1996) Production of transgenic cassava(Manihot esculentaCrantz) plants by particle bombardmentusing luciferase activity as selection marker.Mol Breed 2:339–349.

Sambrook J, Fritsch EF and Maniatis T (1989)Molecular Cloning:a laboratory manual. CSH Laboratory Press, Cold Spring Harbor,NY.

Schenk RU and Hildebrandt AC (1972) Medium and techniques forinduction and growth of monocotyledonous and dicotyledonousplant cell cultures.Can J Bot50: 199–204.

Schöpke C, Taylor N, Carcamo R, Konan NK, Marmey P, Hen-shaw GG, Beachy RN and Fauquet C (1996) Regeneration oftransgenic cassava plants (Manihot esculentaCrantz) from mi-crobombarded embryogenic suspension cultures.Nature Biotech14: 731–735.

Schöpke C, Carcamo R, Beachy RN and Fauquet C (1997) Plant re-generation from transgenic and nontransgenic embryogenic sus-pension cultures of cassava (Manihot esculentaCrantz).AfricanJ Root Tuber Crops2: 194–195.

Soni R and Murray JAH (1994) Isolation of intact DNA and RNAfrom plant tissues.Anal Biochem218: 474–476.

Taylor NJ, Edwards M, Kiernan RJ, Davey CDM, Blakesley Dand Henshaw GG (1996) Development of friable embryogeniccallus and embryogenic suspension culture systems in cassava(Manihot esculentaCrantz).Nature Biotech14: 726–730.

Vancanneyt G, Schmidt R, O’Connor-Sanchez A, Willmitzer Land Rocha-Sosa M (1990) Construction of an intron-containingmarker gene-splicing of the intron in transgenic plants and itsuse in monitoring early events inAgrobacterium-mediated planttransformation.Mol Gen Genet220: 245–250.

Wang MB, Upadhaya NM, Brettell RIS and Waterhouse PM (1997)Intron-mediated improvement of a selectable marker gene forplant transformation usingAgrobacterium tumefaciens. J GenetBreed51: 325–334.

Withers LA (1985) Cryopreservation and storage of germplasm.In: Dixon RA (ed.)Plant Cell Culture: A Practical Approach(pp. 169–191) IRL Press, Oxford.

Zhang P, Legris G, Coulin P and Puonti-Kaerlas J (in press)Production of stably transformed cassava plants via particlebombardment.Plant Cell Rep.

Zhang P and Puonti-Kaerlas J (in press) PIG-mediated cassavatransformation using positive and negative selection.Plant CellRep.