Plant Cell Reports (1993) 12:585-589 Plant Cell Reports �9 Springer-Verlag 1993

Optimization of biolistic transformation of embryogenic grape cell suspensions

Dominique H~bert 1, Julie R. Kikkert 2, Franzine D. Smith 2, and Bruce I. Reisch 2

1 Institut National de la Recherche Agronomique, Station de G~n6tique et d'Am~lioration des Plantes de Montpellier, Domaine de Melgueil, F-34130 Mauguio, France

2 Department of Horticultural Sciences, New York State Agricultural Experiment Station, Cornell University, Geneva, NY 14456, USA

Received March 17, 1993/Revised version received May 25, 1993 Communicated by G. C. Phillips

Abstract. Embryogenic suspensions of 'Chancellor' (Vitis L. complex interspecific hybrid) were bombarded with tungsten particles coated with plasmid pBI426 encoding 13- glucuronidase (GUS) and neomycin phosphotransferase (NPTII) which results in kanamycin resistance. Two d af- ter bombardment, cultures were placed on semi-solid medium containing either 8.6 or 17.2 ~tM kanamycin. Factors that affect biolistic transformation rates were studied. Tungsten microprojectiles with a mean diameter of 1.07 Inn (M10) resulted in more transient gene expres- sion than 0.771 Inn diameter particles. Using M10 parti- cles, helium pressures of 1000 and 1200 psi yielded more GUS-expressing colonies per plate than did 800 psi 2 d following bombardment. The number of transformants present after 34 d was not affected by the helium pressure. The distance between the particle launch site and the target cells, and the number of days between the last cell subcul- ture and bombardment, did not affect the numbers of tran- sient and long term GUS expressing colonies. The addi- tion of 3 g/1 of activated charcoal to the post-bombardment medium increased long term GUS expression four fold. Wrapping the plates after bombardment with Parafilm in- creased long term GUS expression three fold compared with plates wrapped with a porous venting tape. With up to 850 transformed callus colonies per plate 23 d after bombardment, the biolistic device holds much promise as a method to achieve stable transformation of grapevines.

Abbreviations: AC, activated charcoal; GUS, 13- glucuronidase; 2,4-D, 2,4-dichlorophenoxyacetic acid; BA, 6-benzylaminopurine; IAA-L-alanine, indole-3 acetic acid L-alanine; MS, Murashige and Skoog; CH, casein hydrolysate; Km, kanamycin; NPTII, neomycin phosphotransferase II

Introduction

The introduction into grapevines of important traits under simple genetic control can be accomplished by two pro-

cesses: 1) hybridization followed by backcrossing and 2) genetic engineering. The introduction of one gene by hy- bridization followed by several generations of backcross- ing takes many years because grapes are perennial woody plants with long generation cycles. Genetic engineering is a more suitable process to transfer specific genes into gapes because important traits could be introduced into es- tablished cultivars. This is especially important for heterozygous, clonally propagated crops. Grapevines have proven to be a difficult target for transformation with Agrobacterium systems (Colby and Meredith 1990; Colby et al. 1991). Only transgenic or chimeric buds of Vitis ri- paria (Berres et al. 1992) and V. vinifera (Mullins et al. 1990) and rooted transgenic plants of V. rupestris 'St. George' (Mullins et al. 1990) and V. vinifera (Baribault et al. 1990) have been obtained. In the latter case, transfor- marion of a single vinifera cultivar was a very rare event.

Transformation by direct gene uptake into grape proto- plasts has not been successful because of the difficulty of cultivating and regenerating plants from protoplast cultures (Krul 1988; Lee and Wetzstein 1988). Plant regeneration from grapevine cultures has been reported from leaf tissues (Barlass and Skene 1978; Cheng and Reisch 1989; Stamp et al. 1990a, 1990b), callus (Clog et al. 1990), and em- bryogenic suspension cultures (Rajasekaran and Mullins 1979; Manro et al. 1986; Stamp and Meredith 1988; Gray 1992). Thus, gene transformation techniques for grape re- quire improvement and applicability to a broad range of cultivars.

During the last few years, the biolistic process has proven to be an efficient method for plant cell transforma- tion, leading to the production of transgenic plants (Fitch et al. 1990; Finer and McMullen 1991; Sanford 1990; Sanford et al. 1993). Microprojecrile bombardment tech- niques eliminate the need for protoplasts which are re- quired for direct gene uptake. Also, since success has been so difficult to obtain with Agrobacterium-based systems, microprojectile bombardment may prove to be the method of choice for grape. Embryogenic cultures may be the best target tissue for transformation via microprojectile bom- bardment because they provide maximum cell exposure to bombardment, and because in 'a properly grown embryo-

Correspondence to: B. I. Reisch

586

genic suspension culture a large number of cells are com- petent to form embryos and plants. To date, microprojec- tile bombardment has not been applied to grape.

The objective of this research was to study the factors that affect biolistic transformation of embryogenic grape cultures. In this paper, we report on the long term expres- sion of foreign genes in embryogenic cells of grapevine using the biolistic device as the delivery method.

Material and methods

Plant material and tissue culture. Embryogenic suspension cultures of 'Chancellor' (Seibd 7053, a complex interspecific hybrid derived from V. vinifera, V. rupestris, V. lincecumii, V. riparia, and V. labrusca) were obtained in May 1992 from Dr. R.N. Goodman, University of Missouri- Columbia (Boyes and Goodman 1993). Cultures were grown in liquid medium: Murashige and Skoog (1962) medium (MS, Sigma), supple- mented with 5 gM 2,4-D, 1 laM BA and 2% (w/v) sucrose. The pH of the medium was adjusted to 5.8 before autoclaving. Every 14 d, 6 ml of sus- pension was added to 20 ml of fresh medium in a 125 ml Erlenmeyer flask. The flasks were placed on a gyratory shaker at 125 rpm and 24 + 2* C in the dark. For bombardment, cells were collected onto a sterile filter paper disc 0Vhatman no. 1,7 cm diameter) by vacuum filtration in a Buchner funnel. Each filter paper received a 6 ml aliquot of ceils. Filter papers with cells were placed in Petri plates that contained 'pagar' medium (Smith et al. 1992): 10 ml of half strength MS medium with 0.125 M mannitol, 0.125 M sorbitol and 3% (w/v) sucrose solidified with 0.6% (w/v) agar poured over a strong filter paper (Shark Skin, Schleicher and Schuell, Keene, New Hampshire 03431) with an attached tab for handling (paper + agar = 'pagar'). The addition of osmotic agents to the bombardment medium increases the number of transformants and the re- covery of stable transformants (Russell et al. 1992a). The half strength MS medium is hormone-free to permit embryogenesis.

Plasmid DNA and carrier coating. The GUS gene (uidA) from Escherichia coli (Jefferson et al. 1987) was used to determine the fre- quencies of transient gene expression. The NPTII gene (Beck et al. 1982) was used as a selectable marker for stable transformation. Plasmid pBI426 is a pUC9-based plasmid that codes for a GUS and NPTII fusion protein such that GUS expression and kanamycin resistance can be simul- taneously assayed. Expression of this fusion protein is under the control of a double 35S Cauliflower Mosaic Virus (CaMV) promoter plus a leader sequence from Alfalfa Mosaic Virus. Plasmid pBI426 was ob- tained from W. Crosby, Plant Biotechnology Instigute, Saskatchewan,

Canada. The plasmid was amplified in E. coli strain DH5~xF', isolated by alkaline lysis, and purified by CsC1/ethidium bromide density centrifuga- tion (Sambrook et al. 1989). Purified DNA was resuspended in TE buffer (1 mM Tris [pH 7.8], 0.1 mM Na2EDTA ).

The tungsten particles (M5 and M10) (Sylvania GTE Products Corp., Towanda, Pennsylvania) were coated with plasmid DNA (1 }.tg/gl) using CaC12 (1.0 M ) / spermidine (16 mM) precipitation as previously described (Daniell et al. 1990). DNA coated panicles were then washed with 70% ethanol (to remove the free CaCI 2 and spermidine) followed by 100% ethanol and were then resuspended in 100% ethanol to facilitate rapid drying of the coated particles onto the launching surface. After a 3 s treatment in an ultrasonic cleaner (Branson 1200; Branson Ultrasonic Corp., Danbury, Connecticut) to disperse the panicles, 6 gl of the sus- pension was spread onto the center of each Kapton macrocarrier (25 mm diameter, 2 mil thick; DuPont Co.). To dry the suspension, the loaded macrocarriers were placed into a desiccation chamber.

Bombardment conditions. The helium-driven device that was used is the prototype of the commercially available (BioRad) retrofit to the PDS- 1000 device and is described in detail by Sanford et al. (1991). The dis- tance between the rupture disk and the launch point of the macrocarrier was 1 cm. The macrocarrler (a Kapton disk) traveled 1 cm before im- pacting a steel stopping screen. These two distances have been defined as optimal distances for transformation efficiency with cell suspension cul- tures (Russell et al. 1992b). The ceils were placed at the chosen distance from the microcarrier launch point and were bombarded only one time.

The sample chamber was evacuated to 0.1 atm and then the gas accelera- tion tube was pressurized with the chosen helium gas pressure.

Post-bombardment cell handling. After bombardment the Petri plates were placed, unsealed, in styrene plastic boxes (Flambeau Products, Middlefield, Ohio) and incubated at 24 + 2* C in the dark.

To gradually reduce the osmotic pressure (Russell et al. 1992b), on the day after bombardment the pagar medium with the cells was placed in a new Petri plate containing 10 ml of half strength agar-solidified MS with- out sorbitol or mannitol. The cells and pagar medium were transferred 7 h later to a new plate with the same medium.

To assay for transient and long term GUS gene expression, following bombardment the filter paper that supported the cells was placed in a Petri plate containing 600 Ixl of 5-bromo-3-chloro-3-indolyl-B-D-glu- curonic acid (X-Glue) (McCabe et al. 1988). A few drops of X-Glue were also added on top of the large ceil clusters. The cells were incubated overnight at 37* C in the dark. The number of blue spots per plate was counted using a stereo dissecting scope at 20X. A cluster of blue cells was counted as a single spot, because we could not determine whether each cell in the cluster had been transformed or whether the stain had leaked from one transformed cell to its non transformed neighbors. Data were compared on the basis of the number of blue spots per plate. Where density of blue spots was too high to permit accurate counting, a random sample representing 1400 mm 2 was accurately counted and data were then extrapolated to estimate the total number of blue spots per plate;

To select for transformed cells, the filter paper that supported the cells was placed in a Petri plate containing 20 ml of half strength MS + 8.6 }xM (5.0 ~tg/ml) kanamycin (Km) in the dark at 24+_2* C. To reduce desicca- tion and contamination, the plates were wrapped with Parafilm (American National Can, Greenwich, Connecticut) or Venting tape (Scotch Brand no. 394, 3M Corporation, St. Paul, Minnesota). The Venting tape permits more gas exchange than does Parafilm, possibly avoiding the accumula- tion of inhibitory gases. Venting tape does not effectively reduce desic- cation of the plates. Thus, when the cells were wrapped with Venting tape, they were transferred to new medium once every two weeks, vs. every 4 weeks with Parafilm.

Experimental design. In this study several factors affecting transforma- tion were tested:

Optimization of particle size and helium pressure. Embryogenic cells were placed 12.1 cm from the launch point and bombarded with M5 and M10 tungsten panicles at 800, 1000 and 1200 psi (1 psi = 6.895 kPa). M5 particles are characterized by a mean diameter of 0.771 grn whereas M10 particles have a mean diameter of 1.07 ~tm (Smith et al. 1992). Three replicates per treatment combination were used and the experiment was repeated three times. The effect of the particle size x helium pressure interaction was also studied. To determine the optimum pressure for long term GUS expression, the cultures which were bombarded with M10 par- ticles at 800 and 1000 psi were assayed 34, 50 and 73 d following bom- bardment. Four plates per pressure and reading date were used.

Optimization of the distance between the launch point and the target cells. Three distances were studied using the M10 particles at 1200 psi: 5.8, 9.0 and 12.1 cm. Four plates per treatment were bombarded and the experiment was repeated three times.

Optimization of biological parameters. For the following experiments the target cells were placed 12.1 cm from the launch point and bombarded with M10 particles at 1000 psi. The effect of the number of days between the bombardment and the last cell subculture on transient GUS expression was studied. Two culture ages (15 and 24 d post-subculture) were compared. Five plates per factor were bombarded. The influence of the cell cluster size was also tested with regard to yield of GUS expressing colonies. The cell suspension culture was sieved with a screen (1.5 mm mesh) to separate the large clusters from the small ones. Four plates per size were bombarded and the experiment was repeated twice. Four compositions of the induction medium (half strength MS + 8.6 [xM Km, half strength MS + 8.6 p-M Km + 1 g/1 CH, half strength MS + 8.6 }xM Km + 3 g/1AC, and half strength MS + 8.6 gM Km + 1.1 }xM BA + 0.4 ~tM IAA-L-alanine) and two sealing materials (Venting tape and Parafilm) were tested for their effect on long term GUS expression. Four plates per medium and eight plates per sealing material were used. The experiment was repeated twice.

To study the factor effects, for each experiment, an analysis of vari- ance was performed (PROC GLM; SAS Institute 1985). Factor means

587

were compared using the Student-Newman-Keuls test (SAS Institute 1985).

Results and Discussion

Particle size

Particle size affected the number of GUS expressing colonies (Table 1). M10 yielded significantly more blue spots (1577 + 602) than M5 particles (600 + 347) (P = 0.0001). This result was in agreement with previous studies: M10 particles are more efficient with plant cells (Klein et al. 1988), and M5 particles are more efficient with prokaryote cells (Smith et al. 1992). Consequently, in later experiments M10 particles were used, except where noted in the text.

Helium pressure

The helium pressure did not affect the number of GUS ex- pressing colonies when the particle size factor was not considered: 1003 + 463 blue spots per plate at 800 psi, 1165 + 714 at 1000 psi, and 1104 + 890 at 1200 psi (P = 0A4). But there was a significant particle size x helium pressure interaction (P = 0.002). With M5 particles, lower pressures resulted in more GUS expressing colonies (Table 1, P = 0.04). With M10 particles, higher pressures resulted in more GUS-expressing cells. At 1200 and 1000 psi there were significantly more GUS expression than at 800 psi (Table 1, P = 0.029 for 800 vs. 1000 and 1200, Fig. 1A). The high GUS activity in bombarded cells indicated that the double CaMV 35S promoter worked well with grapes (Fig. 1).

Long term GUS expression was used as a potential indi- cator of stable transformation. Since transient transforma- tion was not significantly different between 1200 and 1000 psi, and high bombardment pressures cause increased cell injury (Russell et al. 1992a), long term GUS expression was compared using 800 vs. 1000 psi. In this case, the number of long term GUS expressing colonies per plate was not significantly different 34 d after bombardment (P = 0.17), 50 d after bombardment (P = 0.54) and 73 d after bombardment (P = 0.22, Table 1). In subsequent experi- ments M10 particles were used at 1000 psi, except where noted in the text.

Optimization of the distance between the launch point and the target cells

Using M10 particles at 1200 psi, the distance between the particle launch point and the target cells did not signifi- cantly affect the number of blue colonies (P = 0.10, Table 1). This result contradicts previous findings that the num- ber of GUS-expression units decreased as the distance in- creased (Klein et al. 1988; Russell et al. 1992a). As there is more cell injury when the target cells are placed close to the source of the blast (Russell et al. 1992a) the 12.1 cm distance was used for all the following experiments.

Optimization of the biological parameters

Fig. 1. GUS gene activity in embryogenic cells of grape cultivar Chancellor. A. 3 d after bombardment with MI0 particles at 1000 psi. Bar =1 cm. B. 3 d after bombardment with MIO particles at 1000 psi. Bar = 0.5 mm. C. 21 d after bombardment with M10 particles at 1000 psi. Bar = 4 mm.

The number of days between the bombardment and the last cell subculture (15 and 24 d post-subculture) did not affect the number of GUS-expressing cells (P = 0.74, Table 1). This suggests that the time since last subculture may not influence the efficiency of particle bombardment mediated transformation. Lida et al. (1991) have shown the opposite result with tobacco, but the cells were not embryogenic. Duchesne and Charest (1992), however, found that there was an optimal age for embryogenic callus of Picea mariana to be bombarded.

Microprojectile bombardment led to transient GUS ac- tivity in single cells as well as in clusters of cells (Fig. 1B). There was no effect of cluster size on transient GUS gene activity (P = 0.16, Table 1). As large clusters will more likely lead to chimeric plants, and as embryos are induced in the small clusters, the small clusters were the best tar- gets for grape transformation.

588

Table 1. Effect of treatments on transient and long-term GUS gene expression in cultured cells of 'Chancellor'.

Particle Pressure Distance Culture Sealing Cluster Culture Kanamy- No. of GUS-positive cell clusters per plate after

size (I ~si) in em a age material size b medium e ein (p,,M) 2d 34 d 50 d 73 d

M5 800 12.1 0.0 789-+471 C - 1000 " 0.0 608-+237 D 1200 " 0.0 398-+161 E -

M10 800 " VT d 1/2MS e 8.6 1191:L-667 B 1305:75 A 0.50"20.58 A

1000 . . . . 8.6 1721_+.577 A 58+56 A 0.25-+0.50 A 1200 . . . . . . 0.0 1899_+~6 A

0.33-+0.58 A

0-+0A

M10 1200 5.8 0.0 968+454 A - . . . . 9.0 0.0 1205-+652 A . . . . 12.1 0.0 1381+799 A

M10 1000 12.1 15 d 0.0 680-+432 A . . . . . . 24 d 0.0 605_+363 A

M10 1000 12.1 small 0.0 1305-+825 A . . . . . . large 0.0 1088+689 A -

20 d 43 d M10 1000 12.1 VI" 1/2MS 8.6 225-+376 A 7+11 B

. . . . . . p f " 8.6 262+_297 A 22-+34 A

M10 1000 12.1 VT& P 1/2 MS 8.6 288-+570 AB 12-+15 B i1 v~ i i

+ACg 8.6 599+_301 A 51-+43 A

. . . . . . . . +CH h 8.6 64-+79 B 0-+0B

. . . . . . BA+ 8.6 54-+61 B 3-+4 B IAA i

27 d 50d 51-+86 A 1-+2 A 11-+13 A 0.25-+0.50 A

M10 1000 12.1 VT&P 1/2MS 0.0 . . . . . . . . . . 8.6

a launch point to target cells

b small clusters were <1.5 mm, large dusters were >1.5 mm

c culture medium for cell growth 2 d following bombardment

d VT = Venting tape

e 1/2 MS = half strength Murashige and Skoog medium

f P = Parafilm

g + AC = 1/2 MS supplemented with 3 g/1 activate charcoal

h + CH = 1/2 MS supplemented with 1 g/l casein hydrolysate

i BA + IAA = 1/2 MS supplemented with 1.1 I.tM BA and 0.4 I.tM IAA

Experiments are separated by solid horizontal lines. Means within columns and within experiments followed by the same capital letter are not significantly different at p=0.05. Data presented are means+standard deviations.

The effect of the composition of the induction medium and the effect of the sealing material on long term GUS expression depended on the number of days after bom- bardment. The sealing material did not affect the number of GUS-expressing cells (P = 0.8, Table 1) 20 d after bom- bardment, but there was an effect due to the medium com- position (P -- 0.016, Table 1). The addition of 3g/l AC to the basal medium increased long term GUS expression but this was not significant (Table 1, Fig. 1C). The addition of either 1 g/1 CH or of 1.1 ktM BA + 0.4 JaM IAA-L-alanine to the basal medium decreased the number of GUS ex- pressing colonies (Table 1). Both the composition of the medium and the sealing material affected the number of GUS-expressing colonies 43 d after bombardment. The addition of 3 g/l of AC to the basal medium increased the recovery of long term GUS expressing colonies (Table 1, P = 0.026), but the addition of 1 g/l CH or of 1.1 I.tM BA + 0.4 I.tM IAA-L-alanine had no significant effect (Table 1). The higher rate of long term GUS expression when AC is added to the basal medium was due to better growth of the transformed cells, not to transformation frequency. The

AC might help the growth of the cells by absorbing the oxidized substances excreted by the ceils. When plates were wrapped with Parafilm long term GUS expression was three fold higher than with Venting tape (Table 1, P = 0.056). In contrast, Russell et al. (1992) reported that the recovery of Km r colonies was higher when Venting tape was used than when Parafilm was used. Perhaps the Venting tape slows the growth and the differentiation of the embryogenic callus by decreasing the amount of ethy- lene in the plate. Indeed, it has been reported that the ac- tion of endogenous ethylene is necessary for the growth of embryogenic callus of Medicago sativa during embryo in- duction, differentiation, and maturation (Kepczynski et al. 1992). It might be the same case for the grape or more specifically for the cultivar Chancellor.

For some species a culture period of 1 to 2 weeks (soybean, Finer and McMullen 1991) or 3 to 5 weeks (papaya, Fitch et al. 1990) in medium without selection is necessary to obtain stably transformed clones. In our ex- periment, the presence of kanamycin in the induction medium 2 d after the bombardment did not affect the num-

589

bet of cells that expressed GUS 27 d after bombardment (P = 0.31) and 50 d after bombardment (P = 0.51, Table 1).

GUS gene expression decreased rapidly over time when the cells were placed on medium with or without 8.6 lxM Km after bombardment (Table 1). It is possible that trans- formed cells died as a consequence of competition with normal ceils. Higher levels of Km may be necessary to give transformed cells a selective advantage. Of course, plasmid DNA that was not stably inserted into nuclear DNA probably accounts for the bulk of the reduction in GUS expressing colonies over time. After two months of culture on medium supplemented with 8.6 laM Km some bombarded plates produced embryos. The GUS assay on these embryos was negative. These results show that this level of Km was not high enough to completely prevent

�9 growth of non-transformed embryos, and to recover trans- formed embryos. Grape tissues are highly sensitive to kanamycin (Baribault et al. 1989; Colby and Meredith 1990). At extremely low levels, kanamycin inhibits growth in grape cultures but it can also reduce the recovery of transformed plants (Mullins et al. 1990). It is, therefore, difficult to balance the selection requirement for a rela- tively high concentration of kanamycin against the in- hibitory effects of the antibiotic on embryo recovery. Supplementing the post-bombardment medium with 17.2 ~tM Km and 3 g/1 AC, and wrapping the plates with Parafilm, resulted in 850 blue spots per plate 23 d after bombardment i.e. 50% of the number of blue spots 2 d af- ter bombardment. The concentration of 17.2 ~tM kanamycin yielded 50% more long term GUS expressing colonies than the concentration of 8.6 ktM. This concen- tration seemed to select more efficiently for transformed cells, but it is too early to know if it will hamper the recov- ery of transformed embryos.

Investigations to obtain transformed embryos and plants are continuing. As the high sensitivity of grape cells to kanamycin may hamper the recovery of transformed em- bryos, we are now trying other selection systems. The mi- croprojectile bombardment approach of gene delivery ap- pears to have considerable potential for generating trans- genic grapevines with useful agronomic traits.

Acknowledgments. We thank R.N. Goodman for the gift of the embryo- genie culture of 'Chancellor', and W. Crosby for the use of plasmid pBI426. For guidance and the use of research facilities we thank John Sanford. We also acknowledge M.A. Lodhi, M.tt. Martens, G.N. Ye, and W. Boresjza-Wysocki for their technical advice. Financial support for this study was provided by the Institut National de la Recherche Agronomique and by a grant from the New York Wine & Grape Foundation.

References

Baribauk TJ, Skene KG, Scott NS (1989) Plant Cell Rep 8:137-140 Baribault TJ, Skene KG, Cain PA, Scott NS (1990) J Exp Bot 41:1045-

1049 Barlass M, Skene KG (1978) Vitis 17:335-340 Beck E, Ludwig G, Auerswald EA, Reiss B, Schaller H (1982) Gene

(Amst.) 19:327-336. Berres R, Otten L, Tinland B, Malgarini-Clog E, Walter B (1992) Plant

Cell Rep 11:192-195 Boyes CJ, Goodman RN (1993) Plant Cell Tissue Organ Cult (submitted) Cheng ZM, Reisch BI (1989) Plant Cell Rep 8:403-406 Clog E, Bass P, Walter B (1990) Plant Cell Rep 8:726-728 Colby SM, Meredith CP (1990) Plant Cell Rep 9:237-240 Colby SM, Juncosa AM, Meredith CP (1991) J Amer Soc Hort Sci

116:356-361 Daniell H, Vivekananda J, Nielsen BL, Ye GN, Tewari KK, Sanford JC

(1990) Proc Nat Acad Sci USA 87:88-92 Finer JJ, McMullen MD (1991) In Vitro Cell Dev Bio127:175-182 Fitch MM, Manshardt RM, Gonsalves D, Slightom JL, Sanford JC (1990)

Plant Cell Rep 9:189-194 Gray DJ (1992) Am J Bot 79:542-546 Guellec V, David C, Branchard M, Tempe J (1990) Plant Cell Tissue

Organ Cult 20:211-215 Jefferson RA, Kavanagh TA, Bevan MW (1987) EMBO J 6:3901-3907 Kepczynski J, McKersie BD, Brown DCW (1992) J Exp Bot 43:1199-

1202 Klein TM, Gradziel T, Fromm ME, Sanford JC (1988) Bio/Technology

6:559-563 Krul WR (1988) Seientia Horticulturae 37:231-245 Lee N, Wetzstein HY (1988) Plant Cell Rep 7:531-534 Lida A, Yamashita T, Yamada Y, Morikawa H (1991) Plant Physiol

97:1585-1587 McCabe DE, Swain WF, Martinell BJ, Christou P (1988) Bio/Yechnology

6:923 -926 Mauro MC1, Nef C, Fallot J (1986) Plant Cell Rep 5:377-380 Mullins MG, Tang AFC, Facciotti D (1990) Bio/Teehnology 8:1041-

1045 Murashige T, Skoog F (1962) Physiol Plant 15:473-497 Rajasekaran K, Mullins MG (1979) J Exp Bot 116:399-407 Russell JA, Roy MK, Sanford JC (1992a) Plant Physio198:1050-1056 Russell JA, Roy MK, Sanford JC (1992b) In Vitro Cell Dev Biol 28P:97-

105 Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning: A

laboratory manual, (second edition) Cold Spring Harbor Laboratory, New York, chapter 1.21-1.52

Sanford JC (1990) Physiol Plant 79:206-209 Sanford JC, De Vit MJ, Russell JA, Smith FD, Harpending PR, Roy MK,

Johnston SA (1991) Technique 3:3-16 Sanford JC, Smith FD, Russell JA (1993) Methods Enzymol 217:483-509 SAS Institute (1985) SAS/STAT Guide for personal computers. SAS

Institute Inc., Cary, North Carolina, USA Smith FD, Hal-pending PR, Sanford JC (1992) J Gen Microbiol 138:239-

248 Stamp JA, Meredith CP (1988a) Seientia Horticulturae 35:235-250 Stamp JA, Meredith CP (1988b) J Amer Soc Hort Sci 113:941-945 Stamp JA, Colby SM, Meredith CP (1990a) Plant Cell Tissue Organ Cult

22:127-133 Stamp JA, Colby SM, Meredith CP (1990b) J Amer Soc Hort Sci

115:1038-1042