Gene, 17 (1983) 155-162 Elsevier

155

GEN 02915

A dual-labeling method for identifying deferentially expressed genes: use in the identification of cDNA clones that hybridize to RNAs whose abundance in tomato flowers is potentially regulated by gibberellins

(Recombinant DNA; molecular cloning; colony hybridization; plaque hyb~~zation; flower development; d8erential autora~o~aphyl~uoro~aphy ; ~~~o~~r~~&~~ ~~~~1~~~~ ; gib- 1 mutant)

Neil E, Olszewski %‘, Robert T. Gast a and Frederick M. Ausubel b

n Pkmt Mole&w Genetics ~~~~tute atad ~ep~~t~ent of notary (Tef. j6f2)62f-3129). University of M~~sot~, St. Paul, MN 5510% (U.S.A.) and ’ ~epurt~ent of Genetics (Y-eel, (617/7X-5969), Harvard Medical School aad department of Molecular Biology, Mmsachusetts General Hospital, Boston, MA 021 I4 (U.S.A.)

Received by .I.L Slightom: 29 August 1988 Revised: 4 November 1988 Accepted: 15 November 1988

SUMMARY

A method for identifying cDNA clones that hybridize to differentially expressed RNAs is described. Briefly, the RNA population in which the RNAs of interest are more abundant is used as a template for the synthesis of 35S-labeled cDNAs and another RNA population in which the RNAs of interest are less abundant is used as a template for the synthesis of 32P-labeled cDNAs. The labeled cDNAs are pooled and hybridized to plaque or colony lifts constructed from a cDNA library. Clones that hybridize to RNAs that are d~erenti~ly expressed are identified using differential autora~o~a~hy~~uoro~aphy to discriminate between the 32P and 35S isotopes. We have used this method to identify cDNA clones that hybridize to mRNAs that are more ab~d~t in the flowers of wild-type tomato than in the flowers of mutants that have low endo~enous Ievels of gibberellins.

Recombinant cDNA clones that hybridize to dif- ferentially expressed eukaryotic mRNAs serve as valuable tools for biologists because changes in gene

--- Corrqmdence to: NE. Olszewski, Department of Botany, University of Minnesota, St. Paul, MN 5.5108 (U.S.A.) Tel. (612)625-3129; Fax (612)625-5203.

Abbreviations: A 260, absorbance at 260nm; AMY, avian myeloblastos~s virus; bp, base pair(s); cDNA, DNA complemen- tary to RNA; cpm, countsjmin; cv., cultivar; DTT, dithiothreitol; GA, gibberellic acid; SDS, sodium dodecyl sulFate; SSC, 0.15 M NaCI/O.OlS M Na, citrate pH 7.6; wt, wild-type.

expression play important roles in development, in responses to hormones, and in responses to environ- mental stimuli. Methods currently employed to iden- tify cDNA clones that hybridize td differentially ex- pressed RNAs generally involve the following steps. (1) Two labeled probe populations are prepared from two RNA populations that presumptively contain differing amounts of the RNAs of interest; (2) duplicate colony or plaque lifts (see MATERIALS AND METHODS, section c) representing the cDNA hbrary being screened are prepared; (3) one probe population is used to probe one Lift and the other population is used to probe the second. Provided that each clone is represented by an equal amount of

0378-l 119/89/$03.50 0 1989 Eisevier Science Publishers B.V. (Biomedical Division)

156

hybridizable DNA on the duplicate lifts, the amount of labeled probe that hybridizes to DNA from a given clone reflects the amount of homologous mRNA that is present in the original population. Clones of interest are identitied because the amount of probe hyb~dizing to these clones is different on the two filters. In practice, it is not possible to generate true duplicate lifts (i.e., lifts where every colony or plaque transfers an equivalent amount of hybridizable DNA to both filters) and methods based on this protocol incorrectly identify clones because the amount of hybridization that occurs is affected by the amount of DNA that is bound to the filter.

In this report, we describe a dual-labeling method for identifying cDNA clones that hybridize to differ- entially expressed mRNAs, which does not require the construction of duplicate colony or plaque lifts. We have used the due-label~g method to isolate a number of cDNA clones that hybridize to mRNAs that are more abundant in the flowers of wt tomato plants than in the flowers of a gibberellin-deficient mutant.

MATERIALSANDMETHODS

(a) Plant material

The wt tomato ~yco~ersic~~ esc~~ent~m (L.) Mill cv. Moneymaker was obtained from Thompson & Morgan, Ltd., Ipswich, U.K. and the GA-deficient genotype g&l, a gibberellin-deficient mutant pro- duced in the Moneymaker cultivar (Koornneef et al., 1981; Zeevaart, 1984), was obtained from M. Koornneef, Agricultural University, Wageningen, The Netherlands. Seeds of the gib-1 mutant were scarified and germinated on filter paper in the presence of 5 x low5 M GA,. Following germi- nation, seeds were transferred to soil and grown in a greenhouse. The wt seeds were germinated in soil and grown in a greenhouse. Flowers were frozen in liquid nitrogen immediately alter harvest and stored at -80°C.

(b) Poly(A)‘RNA isolation and cDNA library con- struction

Poly(A) + RNA was prepared by oligo(dT)-cellu- lose chromatography of total RNA isolated from

frozen tomato flowers by the phenol/SDS method as described in Ausubel et al. (1987). Double-stranded cDNA was prepared from 4 1.18 of poly(A) + RNA by the method of Aruffo and Seed (1987) and methyl- ated using M * EcoRI methyltransferase as recom- mended by the supplier (New England Biolabs). Fol- lowing the addition of EcoRI linkers (M~iatis et al., 1982), the cDNA was digested with 150 units of EcoRI and separated by electrophoresis in a 1% low-melting-point agarose gel. The cDNAs larger than 350 bp were isolated (Ausubel et al., 1987) and used to construct two cDNA libraries, one in pUC12 (Vieira and Messing, 1982) and the other in JgtlO (Huynh et al., 1985). Both vectors were digested with EcoRI, and the EcoRI ends were dephosphorylated using calf intestinal alkaline phosphatase (Boehrin- ger-M~nhe~) prior to ligation with the cDNA. The library constructed in pUC12 was propagated in MC1061 (Casadaban et al., 1980) and the library constructed in @lo was propagated in C600Hfl_ (Huynh et al., 1985). Infection and transformation of bacteria were performed using standard methods (Maniatis et al., 1982). Bacteria were grown on Luria broth (Maniatis et al., 1982) at 37 o C. Ampicillin was used at a concentration of 50 pg/ml.

(cf D~feren~al screening of libraries

Colony lifts were prepared on nitrocellulose filters (Maniatis et al., 1982) or charged nylon membranes (Colony/Plaque Screen; DuPont NEN Research Products) as recommended by the supplier. The filters were prehybridized at 42” C for 4 h in 50% formamide, 5 x SSC, 5 x Denhart’s solution (Ma- niatis et al., 1982), 80 mM sodium phosphate pH 7.0, 200 p&/ml of denatured sheared herring sperm DNA (Sigma), and 1.0% SDS. Hybridizations were performed at 42°C for 24-48 h in 50% formamide, 5 x SSC, 1 x Denhart’s solution, 20 mM sodium phosphate pH 7.0,200 pg/ml of sheared denatured herring sperm DNA, 1.0% SDS, and 20 mM DTT. Equivalent amounts (both cpm and mass) of the 32P- and 35S-labeled probes were mixed and used at a final concentration of lo-20 ng/ml in the hybridi- zations. Probe preparation is described in MA’]TE- RIALS AND METHODS, sectiond. Filters were washed twice in 2 x SSC containing 1.0% SDS for 5 min at room temperature, twice in the same solution for 30min at 65”C, twice in 0.1 x SSC for 3Omin at

157

room temperature, and air dried. Unless indicated in the text, the side of the filter containing the colonies was sprayed with En3Hance Surface Spray Auto- radiography Enhancer as recommended by the manufacturer (DuPont NEN Research Products) and the filters,, two sheets of Kodak X-Omat AR film, and a DuPont Cronex Lightning Plus intensify- ing screen were arranged for autoradiography/ fluorography as indicated in Fig. 1A. Autoradio- graphy/fluorography was performed at -70°C. An alternative strategy for performing autoradiography/ fluorography is described in RESULTS AND

DISCUSSION, fsection b.

(d) Preparatiosn of cDNA probes

Labeled cDNA probes were prepared by the fol- lowing protocol. Poly(A) + RNA (300 ng) was heated to 70’ C for 3 min and added to a 500 ~1 microcentri- fuge tube in which 50 PCi of either [cr-32P]dATP (> 400 Ci/mmol or > 15 TBq/mmol) or [ a-35S]thio- dATP (> 400 Ci/mmol or > 15 TBq/mmol) had been dried. Both labeled nucleotides were obtained from Amersham Corporation. To this were added 2 ~1 of 0.25 M Tris . HCl (pH 8.2 at 42’ C)/O.25 M KC1/30 mM MgCl,; 1~1 of 100 mM DTT; 0.5 ~1 of a mixture of dCTP, dGTP and TTP at a concen- tration of 20 m.M each, and 1.0 pl(O.4 A,,, units) of random primer pd(N), (Pharmacia). The volume was adjusted to 9 yl with RNase-free water and 1~1 of AMV reverse transcriptase (Life Sciences, St. Petersburg, FL; 20 units/pi) was added to initiate the reaction. After 90 min at 42” C, the reaction was terminated by adding 10 ~1 of a solution containing 400 mM NaOH/40 mM Na,EDTA and the RNA was hydrolyzed by heating at 70” C for 10 min. Following neutralization of the reaction by the addition of 20 ,ul of 1 M Tris * HCl pH 7.2, and 40 ~1 of 20 mM Tris - HCl pH 8.0/2.0% SDS/50 mM Na,EDTA pH 8.0/200 mM NaCI, the cDNA was separated from unincorporated nucleotides by chromatography on a Sephadex G50 column as described by Ausubel et al. (1987), except that 0.1 M NaCl was included in all solutions. This protocol typically yields 100 ng of labeled cDNA.

(e) Other methods

For Northern-blot analysis, poly(A) + RNA (300 ng per lane) was denatured and subjected to electrophoresis in a 1 y0 agarose gel containing for- maldehyde (Gerard and Miller, 1986). The RNA was transferred to GeneScreen Plus membranes (DuPont NEN Research Products) and probed following the recommendations of the supplier. Southern transfer of DNA onto GeneScreen Plus membranes was performed as recommended by the supplier.

RESULTSANDDISCUSSION

(a) Dual-labeling differential screen theory

We have developed a method that identifies cDNA clones that hybridize to differentially ex- pressed mRNAs using a single plaque or colony lift. The new method is based on two observations. (1) Labeled cDNA probes can be synthesized from either 32P- or 35S-labeled precursors. The kinetics of hybridization observed with these probes is not affected by the isotope used (Rodland and Russell, 1983). (2) The energy of the p- particles that are produced during decay of 35S and 32P are 0.1674 and 1.710 MeV, respectively (Weast and Astle, 1980). Because of this energy difference, 35S p- particles can be blocked by inserting an attenuator between the isotopes and the x-ray film, but 32P p- particles penetrate the attenuator and can be detected by autoradiography. In the absence of the attenuator, both isotopes can be detected.

This method is performed as follows: two labeled cDNA populations are synthesized. One population is labeled with 35S using the RNA population in which the mRNA(s) of interest is more abundant than the template. The other cDNA population is labeled with 32P using the RNA population in which the mRNA(s) of interest is less abundant than the template. Equal amounts of the two labeled cDNA populations are mixed and hybridized to plaque or colony lifts. Following washing of the filters, the amounts of 32P-labeled and 35S-labeled probes hybridized to individual colonies or plaques is deter- mined by autoradiography/fluorography as illustrat- ed in Fig. 1A. Fluorography is used for the detection

158

J-INTENSIFYiNG SCREEN >r.r~FZM EMULSION

-FE24 BACKING VY, fr / x-x w 7, , / 77.75 / 7, J ,r &.-._ FILTEB % ....._.,._._._._t..,.. . . . . . .._._._._.. ..II . . . . . . . . . . * ..,.,I. ..,..,* . . . . . . . . . _.......,...,............., ,.,.... , . . . . . . . . . . . . . -‘“-. FILTER MOUNT

C - FILM BACKING

Fig. 1. Arrangement of film and attenuators of differentiai autoradiography/~uoro~aphy. (A) Following treatment with spray Er?Hance (see h4ATERIALS AND METHODS, section c), the filter is mounted, with the colonies oriented away from the mount, on the yellow paper (attenuator) which is supplied as interleaves between the sheets of 6lm. When filter and attenuator are arrayed with two sheets of Kodak X-Omat AR film and an intensifying screen, the top film records the signal from both isotopes and the Iower film detects primarily ‘ax? (B) Kodak SB film is used in this diagram and in diagram C; this him is single-sided (i.e., emulsion is present on only one side). When an untreated filter (one not sprayed with En3Hance) and film are arranged with the emulsion oriented away from the filter the plastic film backing attenuates the signal from 3sS and the film records primarily the signal from 32P. (C)When the filter is treated with spray Er?Hance and fluorography is performed with the emulsion of the film adjacent to the filter the film records the signal from both isotopes.

of both isotopes because detection of 35S by auto- radiography is less efiicient than detection of the 32P; whereas, fo~o~ng treatment with En3Hance (a ~uoro~aphic spray), the e~ciency of detection of both isotopes is similar. When autoradiogra- phy~~uoro~aph~ is performed as depicted in Fig. lA, the film adjacent to the attenuator detects only the 32P label, while the film abutting the filter detects both labels. Clones derived from more abun- dant RNAs can be detected because they produce a more intense hybridization signal on the film expo- sure that detects both isotopes compared to the film exposure that detects only 32P.

fb) Parameters affecting sensitivity of aatoradi~gra- pby/~aor~grap~y

We have compared different strategies for auto- radiography/fluorography and have identified two that work well. The first strategy is shown in Fig. IA; this strategy is the quickest because both films are exposed sh~u~t~eously‘ In this strategy, both the Biter and paper on which it is mounted act to attenuate the 35S signal by more than ten-fold. We have found that following Southern blotting or con-

struction of colony lifts using charged nylon mem- branes, the DNA is Iocahzed primarily to the side of the filter that was in contact with the gel or petri plate (not shown). To achieve adequate attenuation of the 35S signal, it is important, because of this locali- zation, to orient the filter with the DNA facing the unattenuated film (Fig. IA). The DNA from colony lifts is similarly localized on nitrocellulose filters. We have not determined if this is true for plaque lifts or for blots constructed on nitrocellulose filters. Because of the presence of the intensifying screen, the ftirn that detects only 32P is effectively 2.5-3 times more sensitive than the film that detects both isotopes {not shown). Thus, clones that hybridize to RNAs that are not di~erenti~ly expressed will be expected to produce a slightly more intense signat on the film that records the signal from “P alone. This is because equal amounts of 32P- and “‘S-labeled probes hybridize to colonies containing clones that recognize RNAs that are not differentially expressed and the film that records both isotopes is potentially exposed to twice the number of radioactive decays per unit of time as the film that records only 3zP. The signal on the film that records 32P alone is slightly out of focus because of the thickness of the attenuator

159

used in this strategy. This problem is reduced by placing lead bricks on the film cassette to compress the array. An alternative arrangement, presented below, produces better focused images, but sacrifices the speed of the first strategy.

In the second strategy, the autoradiogram that records the 32P signal and the fluorogram that records the signal from both isotopes are produced sequentially. During autoradiography, film with emulsion on only one side (Kodak SB Film) is oriented with the emulsion away from the filter (Fig. 1B). In this orientation, the plastic backing for the film attenuates the 35S signal 25-fold. Following generation of the autoradiogram, the filter is sprayed with En3Hance and fluorography is performed with the emulsion oriented toward the filter (Fig. 1C).

(c) Method sensitivity

To determine the sensitivity of this method, differ- ent amounts of 32P- and 35S-labeled DNA were mixed, aflixed to charged nylon membranes by slot blotting and subjected to autoradiography/fluoro- graphy using both of the methods described above. The amounts of 32P- and 35S-labeled DNA affiied to the membrane were determined by scintillation counting. When the autoradiograms and fluorograms were compared, it was possible to visually detect a differential signal from all slots where the ratio of 35S and 32P cpm is 1.6 or greater (not shown). Both of the autoradiography/fluoro- graphy methods described above showed the same level of sensitivity. When the dual-labeling method is used, the RN14 population in which the RNA of interest is more abundant should serve as the tem- plate for the synthesis of the 35S-labeled probe because it is not possible to discriminate slots where the ratio of 35S and 32P cpm is > 1 from slots where the ratio is 1.

Sargent and Dawid (1983) have reported that colony or plaque hybridization using 32P-labeled probes generated from RNA populations poorly detect plaques or colonies with homology to RNAs that constitute less than 0.05 y0 of the RNA popula- tion. To determine the sensitivity of the dual-labeling method, different amounts of 35S-labeled cauliflower mosaic virus 19s transcript cDNA were added to the probes used in screening of colony lifts containing a cDNA clone of the cauliflower mosaic virus 19s

transcript. In these experiments, it was possible to detect a differential signal when the 35S-labeled 19s cDNA represented 0.02%, the smallest amount tested, of the 35S-labeled cDNA (not shown). Based on these experiments, the dual-labeling method appears to be at least as sensitive as methods that use only 32P-labeled probes.

(d) Use of the dual-labeling method to identify tomato flower wild-type abundant cDNA clones

The plant phytohormone, GA, affects many stages of plant growth and development (for a recent review see Graebe, 1987). Tomato mutants that have unde- tectable levels of endogenous gibberellins do not ger- minate unless supplied with gibberellins (Koornneef et al., 1981). If additional gibberellins are not supplied following germination, the plants are dwarfs and normal flower growth and development do not occur. The sepals and petals are dwarfed and the microsporocytes and megasporocytes do not undergo meiosis (Nester and Zeevaart, 1988). Peri- odic applications of gibberellins completely reverse all of the phenotypes associated with gibberellin deficiency.

We have screened approx. 20000 members of a tomato flower cDNA library constructed in pUC12 using the dual-labeling method to identify cDNA clones that hybridize to RNAs that are more abun- dant in wt flowers than in the flowers of a mutant which has low endogenous levels of gibberellin. This screen identified 256 clones that hybridized to RNAs that are more abundant in wt flowers; these clones will be referred to as putative, wt-abundant clones. In this screen, poly(A)+ RNA isolated from wt flowers served as the template for the synthesis of 35S-labeled probe and poly(A) + RNA isolated from flowers produced on gibberellin-deficient plants served as the template for 32P-labeled probe. Fig. 2 shows an autoradiogram (panel A) and a fluorogram (panel B) from the same area of a colony lift pro- duced during the screening of this library using the autoradiography/fluorography strategy illustrated in Fig. 1A. The autoradiogram shown in panel A de- tected primarily the signal due to 32P and the fluoro- gram in panel B detected the signal from both iso- topes. The hybridization signal from the majority of the colonies is slightly stronger on the autoradiogram than the fluorogram, indicating that the majority of

Fig. 2. Screening of a tomato flower cDNA library by the dual-labeling method. The tomato flower cDNA library constructed in pUC12, described in the MATERIALS AND METHODS, section b, was screened to ident@ clones that hybridized to RNA that is more abundant in the flowers of wt tomato plants than in the flowers of gibberellin-deficient tomato plants. Colony lifts were screened by the dual-labeling method (see MATERIALS AND METRODS, section c). Poly(A)* RNA from wt flowers was used as the template for the synthesis of “S-labeled cDNA and poly(A) * RNA from mutant flowers was used as the template for the synthesis of 3”P-labeled cDNA. AuCoradiography/fluorography was performed as illustrated in Fig. 1A. (Panel A) Aa autoradiogram recording the 32P signal from a portion of a colony lift. (Panel B) A Buoragram recording both 32P and ?3 signals from the same portion of a colony lift shown in panel A. Arrowheads mark the location af colonies that hybridize exclusively or predominantly to the ?S-labeled probe.

the clones hybridized to RNAs that have similar abundance in flowers produced by both mutant and wt plants. However, a few colonies can be identified where the signal is only present or more intense (indicated by arrowheads) on the fluorogram that recorded the signals from both isotopes (panel S). These colonies are putative, wt-abundant clones and were selected for further study. Using this method, we have also identified wt-abundant clones in a tomato flower cDNA library constructed in ;igtlO (not shown).

To verify that the putative, wt-abundant clones hybridized to differentially expressed RNAs, we tested each of these clones by applying the dual- labeling differential screen to Southern blots containing EcoRI-digested mimpreps of these clones. The results of this test indicated that 99 (39x) of the colonies selected in the initial screen hybridized to differentially expressed RNA(s) (not shown). Fig. 3A shows representative results ob- tained with four different (non-homologous) wt- abundant clones. Northern analysis of pofy- (A).+RNA isolated from flowers of both wt and dwarf tomato plants further confirms that the clones whase analysis is shown in Fig. 3A hybridize to dif- ferentially expressed RNA(s) (Fig. 3x3). The degree of differential expression observed in the Northern

blots and in the differential screen correlate well (compare Fig. 3A and 33).

The percentage of putative, wt-abundant clones identified in the primary screen that were not actually wt-abundant clones is high because rigorous selection criteria were not employed in the primary screening. To minimize the chance of missing clones of differentially expressed RNAs, all clones that gave any indication of a differential hybridization signal were selected for further study. In other screens where the selection criterion was more rigurous (only colonies that gave a clear indication of differential hybridization were chosen) approx. 90% of the se- lected clones recognized RNAs that were shown to be differentially expressed by dual-labeling analysis of Southern blots (not shown).

(e) Conclusions

We have developed a dual-labeling method that can be used to identify cDNA clones that hybridize to di~er~ntiall~ expressed mRNAs. Because this method does not require the production of duplicate plaque or colony lifts, it is not subject to the screening artifacts that arise due to the inability to construct true duplicate lifts. Reconstruction experiments indi- cate that this method should be able to detect C~O~XS

a b

2

d w

2

a b

d w d

4 5 Fig. 3. Con~rmation that selected cDNA clones hybridize to differentially expressed RNA(s) (Pane1 A) Southern blot contained DNA from cDNA clones identified by the dual-labeling method. DNA from putative, wt-abundant clones was isolated, digested with EcoRI, resolved by electrophoresis on a 0.8% agarose gel and transferred to a nylon membrane. The filter was probed by the dual-labeling method and subjected to antoradio~aphy (a) and ~uorography (b) using the method depicted in Fig. i, B and C. Paired lanes (l-4) represent four different clones. The minor hybridization signal iocated above the main bands in lanes 1 and 3 is hybridization to undigested DNA. (Panel B) Northern-blot analysis of poly(A)+RNA from wt and mutant flowers. Poly(A)+RNA was denatured, separated by electrophoresis and transferred to GeneScreen Plus. Lanes: d, RNA from mutant flowers; w, RNA from wt flowers. Paired lanes l-4 were hybridized with 32P-labeled DNA probes prepared from the clones analyzed in lanes l-4 of panel A, respectively. When paired lanes (5) were hybridized with “P-labeled poly(U) (Sigma), equal amounts of hybridization occurred to each lane indicating that equivalent amounts of poly(A)‘- RNA were present in each lane.

that hybridize to RNAs that are at least I-6-fold different in their abundance in the two populations being compared. We have used this method to screen a tomato flower cDNA library and identify cDNA clones that hybridize to RNAs that are more abun- dant in wt tomato flowers than in the flowers pro- duced by a mutant that has low endogenous levels of gibberellins .

ACKNOWL~DG~~~NTS

We thank J. Berman for making suggestions that improved the manuscript. This work was supported by research funds from the Graduate School of the University of Minnesota and a grant from Hoechst AG to Massachusetts General Hospital, R.T.G. is a

162

University of Minnesota Graduate School and Plant Molecular Genetics Institute graduate fellow.

REFERENCES

Aruffo, A. and Seed, B.: Molecular cloning of a CD28 cDNA by a high-efficiency COS cell expression system. Proc. Natl. Acad. Sci. USA 84 (1987) 8573-8577.

Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A. and Struhl, K.: Current Protocols in Molecular Biology. Greene Publishing Associates and Wiley-Interscience, New York, 1987.

Casadaban, M.J., Chou, J. and Cohen, S.N.: In vitro gene fusions that join an enzymatically active /I-galactosidase segment to amino-terminal fragments of exogenous proteins: Escherichia

coli plasmid vectors for detecting and cloning translational initiation signals. J. Bacterial. 143 (1980) 971-980.

Gerard, G.F. and Miller, K.: Comparison of glyoxal and formamide gels for sizing rRNAs. Focus 8:3 (1986) 5-6.

Graebe, J.E.: Gibberellin biosynthesis and control. Annu. Rev. Plant. Physiol. 38 (1987) 419-465.

Huynh, T.V., Young, R.A. and Davis, R.W.: Constructing and screening cDNA libraries in lgtl0 and lgtll. In Glover,

D.M. (Ed.), DNA Cloning: A Practical Approach, Vol. I. IRL Press, Oxford, 1985, pp. 49-78.

Koornneef, M., Van der Veen, J.H., Spruit, C.J.P. and Karssen, C.M.: Isolation and use of mutant with an altered germi- nation behavior in Arabidopsis thaliana and tomato. In Kitto, P.H. (Ed.), Induced Mutation: A Tool in Plant Research. IAEA-SM-25117, Vienna, 1981, pp. 227-232.

Maniatis, T., Fritsch, E.F. and Sambrook, J.: Molecular Cloning. A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, 1982.

Nester, J.E. and Zeevaart, J.A.D.: Flower development in nor- mal tomato and a gibberellin-deficient (ga-2) mutant. Am. J. Bot. 75 (1988) 45-55.

Rodland, K.D. and Russell, P.J.: Effects of deoxycytidine 5’-[a- thio]triphosphate-[35S] on the reassociation kinetics and S 1 nuclease sensitivity of plasmid DNA. Fed. Proc. 42 (1983) 1954.

Sargent, T.D. and Dawid, I.B.: Differential gene expression in the gastrula of Xenopus lams. Science 222 (1983) 135-139.

Vieira, J. and Messing, J.: The pUC plasmids, an M13mp7 derived system for insertional mutagenesis and sequencing with synthetic universal primers. Gene 19 (1982) 259-268.

Weast, R.C. and Astle, M.J.: CRC Handbook of Chemistry and Physics. CRC Press, Boca Raton, FL, 1980.

Zeevaart, J.A.D.: Gibberellins in single gene dwarf mutants of tomato. Plant Physiol. S75 (1984) 186.