Endophytic bacteria expressing β-glucuronidase cause false positives in transformation of Dioscorea species

Plant Cell Reports

P l a n t Cell R e p o r t s (1992) 1 1 : 4 5 2 - 4 5 6

9 Springer-Verlag 1992

Endophytic bacteria expressing ff-glucuronidase cause false positives in transformation of Dioscorea species Mahmut Tiir, Sinclair H. ManteU, and Charles Ainsworth Wye College, U n i v e r s i t y o f L o n d o n , Wye, A s h f o r d , K e n t , T N 2 5 5 A H , E n g l a n d Received M a r c h 5, 1992/Revised version received M a r c h 16, 1992 - C o m m u n i c a t e d by N. A m r h e i n

Summary. False positive transformants obtained during plant transformation experiments on species of the monocotyledonous genus Dioscorea (yam) are described. The false positive results were found to be due to endophytic bacteria which exist within aseptically micropropagated shoot cultures and which express 13-glucuronidase (GUS). The bacteria were isolated and identified as two species of Curtobacterium. The expression of GUS in these organisms was found to be induced by a variety of glucuronide substrates. The induction of GUS activity in the bacteria can be inhibited by chloramphenicol, tetracycline, ticarcillin and sodium azide. Implications of these results for use of the gus gene in plant transformation work are discussed. Abbreviations: DTT: Dithiothreitol, EDTA: Ethylenediamine tetra-acetic acid, ELISA: Enzyme-linked immuno-sorbent assay, GUS: [3-glucuronidase, LB: Luria Bertani, MUG: 4-Methylumbelliferyl-13-D-glucuronide, PNPG: p-nitrophenyl-[3 - D - g l u c u r o n i d e , PVP: Polyvinylpyrrolidone, SDS: Sodium dodecyl sulphate, TAE: Tris-acetate-EDTA buffer, X-Gluc: 5-Bromo-4-chloro-3indolyl-13-D-glucuronide Keywords: [~-Glucuronidase, D i o s c o r e a , endophytic bacteria, false positive transformants Introduction The E. coli uidA gene which encodes I]-glucuronidase (GUS; Jefferson et al. 1986) has become the most widely used reporter gene in plant transformation. The GUS system owes its popularity to several factors: the enzyme is very stable under different physiological conditions, activity is easy to quantify, highly sensitive yet relatively inexpensive assays are available and it is detectable with histochemical assays that localise gene activity in specific cell types. Most importantly, intrinsic GUS activities in higher plants were reported to be negligible (Jefferson 1987; Jefferson et al. 1987). However, more recently Plegt and Bino (1989) demonstrated endogenous GUS activities during male gametogenesis in untransformed potato, tobacco and tomato plants. Similarly, Wenzler et al. (1989) showed the Correspondence to: S. H. M a n t e l l

presence of endogenous GUS activities in the leaves and roots of wild-type tobacco and potato plants. In fact, Hu et al. (1990) reported intrinsic GUS-like activities in several parts of plants in a survey of fifty-two plant species. In all cases, enzyme activities were assumed to be of plant origin. To date, false positives in plant transformation due to GUS production by endophytic microorganisms have not been reported. During our recent transformation studies on the monocotyledonous Dioscorea yams (T0r et al. 1990a,b), we have detected false positives among the putative transformants. This report presents evidence that these false positives resulted from the presence of endophytic GUSexpressing bacteria in in vitro-propagated plants.

Materials and Methods Isolation of bacteria. Dioscorea shoot cukures were maintained according to Mantell et al. (1978). In vitro-grown plantlets of D. alata, D. cayenensis and D. bulbifera were examined for the presence of endogenous bacteria by immersing tissues into 10 ml liquid LB medium and incubating ovemight at 37~ with shaking. Aliquots of the bacterial suspension were plated onto LB medium containing 50 btg/ml X-Glue. Stabs of resultant GUS + (blue) colonies were sent to ADAS Central Science Laboratory, Ministry of Agriculture, Fisheries and Food (MAFF, Hatching Green, Harpenden, Herts, UK) for identification. Fluorometric and histochemieal assays of plant tissues. Fluorometric and histochemical GUS assays were carried out as described by Jefferson et al. (1987). For fluorometric assays, tissues were homogenized with GUS extraction buffer. Standard fluorometric GUS assays were performed using MUG. For histochemical analysis, specimens were cleared by autoclaving for 15 minutes in 75% lactic acid. Induction of GUS expression. The induction assays were performed using a modified method of Liang (1989). 500 I.tl of a log phase bacterial culture (0.D. 600 = 0.4) were aliquoted into 5ml LB medium. Then 50 [.tl of a 100 m M solution of each glucuronide (MUG, X-Gluc and PNPG) were added to each tube. Sterile deionised water was used as a control. The cultures were mixed rapidly and a l m l aliquot was taken representing time 0. The cultures were incubated at 37~ with agitation and at successive 4h time intervals, lml aliquots were removed from each culture and the absorbance of each bacterial cell suspension was measm'ed at 600 nm. The cells w e r e pelleted in a microcentrifuge tube by centrifuging at 15,000 x g for 2 min, thoroughly washed twice with M9 salts and resuspended in 250 p-1 of M9 salts containing 100 Itg/ml of chloramphenicol to stop further glucuronidase synthesis. The cell suspensions were maintained at -70 C until the enzyme assays were carried out. Protein and GUS assays. Bacterial cell suspensions were thawed on ice and sonicated for two minutes to disrupt cell membranes. Protein and GUS assays were carried out on these extracts. Protein assays were performed

453

FIGURE LEGEND Fig. 1. Analysis of putative transformants by X-Gluc staining of various tissues in D.cayenensis plants. (A) and (B) Leaf tips. (C) Axillary region in stem. (D) Vascular tissues in stem internode. (E) Root tips at low mag. and (F) at high mag. (G) Basal primary nodal complex. Bars shown represent 350, 350, 70, 350, 450, 130 and 300 ~trn, respectively.

454 according to Bradford (1976) and the enzymatic activity of GUS in each induction was assayed in a microtitre plate using a modified method of Liang (1989) based on the cleavage of the GUS substrate, PNPG. The reaction was started by adding 20/.tl of homogenate into each of five consecutive wells containing 100lal enzyme assay buffer (50 mM sodium phosphate buffer, pH 7.0, 10 mM DTI', 1 mM EDTA and 1.2 mM PNPG). Microtitre plates were incubated at 37~ and the five enzymatic reactions for each induction were then stopped by the addition of 100/.tl of 0.4 M Na2CO 3 at regular time intervals. GUS activity determined by the aglycone absorbance at 410 nm was recorded using an ELISA reader.

Inhibition of GUS activity in endogenous bacteria. Induction of GUS in endophytic bacteria was performed in LB medium as described previously, in the presence of antibacterial or antifungal compounds, with PNPG as inducer. Compounds tested were chloramphenicol, tetracycline, kanamycin, ticarcillin, amphotericin-B (all at 100 ~tg/ml), carbenicillin (500 I.tg/ml) and sodium azide (0.2 %w]v). Inhibition of bacterial GUS expression in planta. GUS-expressing bacteria isolated from yam shoot cultures were re-introduced into in vitro yam cultures or introduced into hand-cut stem sections (3-4mm in cross section) of surface sterilised cauliflower by co-cultivating for 6h. After cocultivation, the explants were blotted onto sterile filter paper to remove excess bacteria, and were incubated in 50 mM sodium phosphate buffer pH 7.0, 0,1% TritonX-100, 1 mM X-Gluc in the presence or absence ol~ antibacterial and antifungal compounds. Southern blot analysis. Bacterial genomic DNA was isolated by the method of Ausubel et al. (1989) with slight modifications (bacterial cells were lysed first with lysozyme and then proteinase K and SDS were added). Three to five micrograms of BamHI-digested DNA were fractionated on 0.8% TAE agarose gels, transferred to nylon membranes (Hybond N, Amersham), and hybridised to 50 ng 32p oligolabelled XbaI fragment from the E. coli gusA gene, obtained from pJIT137 (P.M. Mullineaux, pers. comm.). Following pre-hybridisation for 5h at 65~ in 0.6 M NaC1, 20 mM PIPES, pH 6.8, 4 mM EDTA, 0.2% gelatin, 0.2% FicoU 400, 0.2% PVP, 1% SDS, 500 I.tg/ml autoclaved salmon sperm DNA, overnight hybridisation was carried out with the 32p-labeUed probe in the same buffer at 65~ Filters were washed sequentially with 2 X SSC, 0.1% SDS and 0.1 X SSC, 0.1% SDS for 20min each at 65~ . Autoradiography was carried out using intensifying screens for 12h at -70~

of roots, stems and leaves from non-transformed plants of D. alata, D. cayenensis and D. bulbifera. No endogenous activity was detected even with assay periods of up to 6h. This led us to suspect that the blue staining might be due to GUS production by a prokaryote.

Description and identification of endophytic bacteria Endophytic bacteria in in vitro-grown shoot cultures of Dioscorea species did not appear to be pathogenic to yam plants and they originated from the parent material itself rather than from casual contamination introduced during micropropagation procedures. Electron microscopical examinations of cultured plant tissues indicated that bacteria were present within the cytoplasm of yam cells (Mantell, unpublished results). These observations, together with the false positive results obtained in the transformation experiments, led us to investigate whether there was any correlation between the presence of endophytic bacteria and the generation of false positives in our plant transformation experiments. When in vitro-grown shoots from stock cultures of D. aIata, D. cayenensis and D. bulbifera plants were cultured overnight in LB medium, bacteria were recovered from all shoot cultures. When these bacteria were plated on semisolid LB medium, two types of morphologically distinct colonies (large, white and small, yellow) were observed. Bacteria obtained from D. alata consistently produced large, white colonies whereas those obtained from D. cayenensis and D. bulbifera produced a mixture of the two colony types (Figure 2).

Results

GUS expression in putative transformants Putative transformed cells that were false positives were encountered during particle acceleration experiments in which axillary meristems of in vitro-grown shoots from Dioscorea alata and D. cayenensis were bombarded with gold particles coated with pBI121.2 (Jefferson 1987) plasmid DNA, which carries the GUS and NPTII-encoding genes. After bombardment, meristems were cultured axenically at 25~ under 16h photoperiod. Hand-cut sections or whole tissues of recovered plants were mounted, incubated in a histochemical solution containing X-Gluc, cleared and than examined by light microscopy. No stained tissues were observed in sections of D. alata. However, in specimens of D. cayenensis, the intensity of histochemical staining, as shown in Figure 1, varied widely from tissue to tissue. For example, the acuminate gland regions at the tips of leaves (Fig. la,b), and the multicellular trichomes and vascular tissues showed pale blue staining (Fig. lc,d), with the most intense staining at the root tips (Fig, le,f) and plant bases (Fig. lg). In contrast, a weak staining was observed in 10 out of 25 putative tranformants of D. cayenensis, which differed from the strong staining described above and from that previously obtained with transformed cell suspensions (unpublished results). In order to determine whether the observed GUS activity was due to an endogenous plant ~-glucuronidase, we carried out quantitative fluorometric assays using MUG on extracts

Individual colonies were assessed for their ability to express GUS by streaking onto X-Gluc plates. Only the small, yellow colonies were GUS positive. This result was also confirmed using MUG as a substrate in the plates and monitoring UV fluorescence (results not shown). In total, eight GUS + bacterial isolates, each arising from an individual shoot culture (two isolates from D. bulbifera : MT1 and MT2; six from D. cayenensis : MT3 to MT8) were obtained. Stabs of GUS + bacterial cultures were identified by comparison with accessions in the National Plant

455 Pathogenic Bacteria Collection. The isolates MT1, MT2, MT3, MT4 and MT7 were identified as Curtobacterium flaccumfaciens. Isolates MT5, MT6 and MT8 were also identified as Curtobacterium but were not of that species (Accession ID : UN-A4571 [Ex yam Mantell]). Of these 8 isolates, MT2, MT4, MT6 and MT7 were used in subsequent experiments because they were representative of the endophytic bacteria present in two Dioscorea species; D. cayenensis and D. bulbifera. The E. coli strain CE5 was included as a positive control. GUS activity in endophytic yam bacteria. The GUS operon in E. coli is inducible by glucuronide substrates (Jefferson et al. 1986). The time for induction of GUS expression after the addition of substrate may be dependent on the species of bacteria. Therefore, a time course experiment was set up to investigate the detection time of GUS induction in selected bacterial isolates. Although there are many substrates for GUS, the three most commonly used glucuronides were chosen as inducers, i.e. PNPG, MUG and X-Gluc. Induction was performed in LB medium and 1 ml bacterial cultures were assayed for GUS activity at 4h time intervals. In Curtobacterium isolates MT2, MT4, MT6 and MT7, GUS activity was detected after 12h when MUG and PNPG were used as inducers and 16h after induction with X-Gluc (Table 1). In contrast, in the E. coli strain CE5, activity was detected within 2h with all three inducers. GUS activities in MT isolates were 5 to 20 times lower than those in E. coli strain CE5. Also, Dioscorea endophytic bacteria (MT isolates) were very slow-growing with doubling times of between 200 and 260 min, compared to 20 to 48 min for E. coli CE5 in LB medium. Table 1. GUS activity in Curtobacterium isolates (MT2, MT4, MT6 and MT7) and in E. coli CE5 strain with different inducers. GUS activity + S.E.M. (pkat/mg protein) Isolate

Water

MUG

PNPG

X-Gluc

MT2

0.00

73.83 + 0.03

102.99 + 0.03

164.33+ 0.05

MT4

0.00

107.17 + 0.03

93.49 + 0.03

19.49+ 0.02

MT6

0.00

46.99 + 0.02

28.03 + 0.02

80.17 + 0.05

MT7

0.00

53.67 + 0.02

36.49 + 0.02

108.17+ 0.07

carbenicillin; an inhibitor of fungal cell membrane function, amphotericin-B; and the inhibitor of mitochondrial and chloroplast function, sodium azide. No GUS activity was detectable after 15h of growth in the presence of chloramphenicol, tetracycline, ticarcillin, kanamycin, carbenicillin or sodium azide. However, GUS induction was not inhibited by amphotericin-B. Inhibition of bacterial GUS expression in planta Experiments were carried out to determine whether GUS induction in endophytic bacteria could also be inhibited by antibiotics in planta. To ensure the presence of GUS + bacteria in the plants under study, bacteria were introduced into yam shoot cultures and, as a comparison, into cauliflower stem sections. The latter plant tissues were chosen because they are readily available, non-chlorophyllous and consequently give better colour contrast. Leaves, roots and stems of D. cayenensis and cauliflower stem segments were co-cultivated with Curtobacterium isolates MT2, MT4, MT6 and MT7 and E. coli CE5 and were subsequently assayed for GUS activity. The addition of chloramphenicol, tetracycline, ticarcillin, kanamycin, carbenicillin or sodium azide prevented the induction of GUS in Curtobacterium isolates and consequently there was no blue staining of tissues of either D. cayenensis or cauliflower. By contrast, the treatments in which tissues had not been treated with antibiotic showed strong GUS activities. The same results were obtained with the E. coli strain CE5 and chloramphenicol, tetracycline, ticarcillin and sodium azide. However, additions of kanamcyin and carbenicillin did not prevent the induction of GUS at the tissue level. Southern blot analysis To investigate whether there was any homology between the GUS-encoding gene(s) of the Curtobacterium isolates and that of E. coli (uid A), we carried out genomic Southern analysis on the DNA from these bacteria. Genomic DNA isolated from the Curtobacterium strains and E. coli CE5, was digested with BamHI and hybridised with the XbaI fragment from the E.coli gusA (uid A) gene. For both E. coli and Curtobacterium strains, a single BamH1 fragment hybridised (Fig. 3). All Curtobacterium strains gave a fragment (1.4kb) of similar size, although it was larger than the fragment from E. coli (0.9kb) and it hybridised less strongly.

CE5 0.00 542.17 + 0.02 1876.83 + 0.83 a Values are means of four replicates. MT isolates were assayed at 30 min time intervals and CE5 at 5 min intervals. For MT strains, 12h data for MUG and PNPG and 16h data for X-Gluc are given. For CE5, 12h data are given for all substrates. a Production of an intense indigo colour prevented the aquisition of representative measurements.

To circumvent the problem of false-positives in transformation experiments, we investigated ways of inhibiting GUS expression in pure cultures of endophytic bacteria. Induction of GUS was performed in LB medium in the presence of antibacterial or antifungal compounds, with PNPG as inducer. The inhibitors tested were the protein synthesis inhibitors chloramphenicol, tetracycline and kanamycin; the cell wall synthesis inhibitors ticarcillin and

Fig. 3. Southern analysis of Curtobacterium isolates. Total genomic DNA was digested with BamHI and then the fragments were separated by electrophoresis in agarose. Di~A gel blots were hybridised at 65~ to the Xba! uid A fragment from pJIT137. Lanes 1-4, Curtobacterium isolates MT2, MT4, MT6 and MT7. and lane 5, E. coli strain CE5.

456 Discussion In this paper, we present evidence that "aseptically" micropropagated yam shoot cultures contained populations of endophytic bacteria. We also describe for the first time that : (1) these endophytic bacteria can express I~-glucuronidase (GUS), and (2) they give rise to false results with the histochemical analysis which is generally used in plant transformation experiments. Dioscorea shoot cultures contain associated bacteria which appear to be truly endophytic and which often persist in latent form. Significantly, one West African yam species D. sansibarensis contains bacteria within its acuminate leaf glands and these form positive symbiotic associations (Miller and Reporter 1987). Endophytic yam bacteria identified as Curtobocterium sp, were found to be inducible with most o f the commonly used and commercially available GUS substrates such as MUG, PNPG and X-Gluc. Enzyme activities in Curtobacterium isolates MT2, MT4, MT6 and MT7, were substantially lower than those in E. coli strain CE5. This result could have been due to a slower uptake of the substrates by Curtobacterium isolates. Four groups of bacteria with associated GUS activities have been reported previously: E. coli, Streptococcus, Staphylococcus and Corynebacterium (Buehler et al. 1949; Fishman 1955; Karunairatnam and Lewy 1951; Robinson et al. 1952 ). Our results are consistent with these reports since the yam endophyte Curtobacterium flaccumfaciens investigated in the current study was previously designated as Corynebacterium flaccumfaciens (Komagata and Suzuki 1986). Endophytic bacteria are not uncommon in micropropagated plant materials. Leifert et al. (1989) isolated 198 bacterial strains from nine plant species grown in vitro. Significantly, 23% of these contaminants were Staphylococcus or Micrococcus, which, as stated above, are potential GUS + bacteria. It is therefore highly likely that false positives due to endogenous GUS + bacteria may also occur during the genetic transformation of other plant species. When working with a new plant species and when using GUS as a reporter gene in transformation experiments, it is important to be aware of the possibility of interference from endophytic microbes present in micropropagated shoot cultures. In the case of Dioscorea, the detection of GUS activity in endophytic Curtobacterium isolates requires more than 12h induction when MUG or PNPG are used as inducers. Consequently, in experiments to monitor GUS activity in transgenic plants, assay times should be less than 12h if background activity due to GUS positive endophytic bacteria is to be avoided. Although kanamycin and carbenicillin prevented the induction of GUS in E. coli strain CE5 grown in liquid culture, they did not prevent GUS induction in endophytes in planta. This result could be attributed to the relatively slow speed of action of these two antibiotics in plant tissues and it emphasises the need for careful consideration of the

appropriateness of certain antibiotics as selective agents in transformation work when endophytic GUS positive bacteria could be present in tissue cultures. Southern blot analysis of genomic bacterial DNA confirmed that the Curtobacterium GUS activity detected was likely to be due to the presence of a GUS gene homologous to that of E. coli. Therefore, when using Southern hybridisation analysis with the GUS gene as probe to confirm the presence of introduced DNA in putative transformants, the presence of GUS + endophytic bacteria could result in hybridising fragments of bacterial origin. To our knowledge, this is the first report of the presence of GUS-expressing bacteria in plants. The data presented here indicate that where histochemical assays to test GUS activities in transgenic plants are to be performed, extra care should be exercised to confirm that results obtained are due to transformation rather than to the presence of endophytic microorganisms. Acknowledgements. We wish to thank Drs R.A. Jefferson and Dr K.W. Wilson for their valuable suggestions and for supplying a stab of the E.coli strain CE5. Thanks are also due to Dr. V. Buchanan-Wollaston and Dr. J. Beynon for critically reading the manuscript. M.T6r was supported by the Scientific and Technical Research Council of Turkey and the Overseas Research Student Awards Scheme , Council of Vice-Chancellors and Principals, London, U.K.

References Ausubel FM, Brent R, Kingston RE, Moore DD, Seidmen JG, Smith JA and Struhl, K(1989) Short protocols in molecular biology. A compendium of methods from current protocols in molecular biology..John Wiley and Sons, New York p61, Bradford MM (1976) Anal. Biochem. 72:248-254. Buehler HJ, Katzman PA, Doisy PP and Doisy EA (1949) Proc. Soc. Expfl. Biol. Med. 72:297-299. Fishman WH (1955) Adv. Enzymol. 16:361-409. Hu CY, Chee PP, Chesney RH, Zhou JH, Miller PD and O'Brien WT (1990) Plant Cell Reports 9:1-5. Jefferson RA (1987) Plant Molec. Biol. Reporter 5:387-405. Jefferson RA, Burgess SM and Hirsh D (1986) Proc. Natl. Acad. Sci. USA. 86:8447-8451. Jefferson RA, Kavanagh TA and Bevan MW (1987) EMBO J. 6:39013907. Kamnairatnam MC and Lewy GA (1951) Biochem. J. 49:210-215. Komagata K and Suzuki KI (1986) Bergey's Manual of Systematic Bacteriology, Vol 2, P.H.A.Sneath., N.S. Mair, M.E.Sharp.and J.G.Holt, eds, pp 1313-1317. Leifert C, Waites WM and Nicholas JR(1989) J. Appl. Bacteriol. 67: 353361. Liang W (1989) M.Phil. Dissertation, University of Cambridge, England. Mantell SH, Haque SQ and Whitehall AP (1978) J. Hort. Science 53:9598. Miller IM and Reporter M (1987) Plant, Cell and Environment 10:413-424. Plegt L and Bino RJ (1989) Molec, Gen. Genet. 216:321- 327. Robinson JJ, Blin CW and Frank PF (1952) J. Bacteriol. 64:719-723 T6r M, Ainsworth CC and Mantell SH (1990a) Abstracts VIIth Int'l. Congr. Plant Tissue and Cell Culture. (Abs# A2-138), Amsterdam, June 24-29. T6r M, Mantell SH and Ainsworth CC (1990b) Abstracts VIIth Int'l. Congr. Plant Tissue and Cell Culture. (Abs # A2- 139), Amsterdam, June 24-29. Wenzler HC, Mignery GA, Fisher LM and Park WD (1989) Plant Molec. Biol. 12:41-50