NATURE BIOTECHNOLOGY VOL 18 JUNE 2000 http://biotech.nature.com 661

Environmental disasters like that caused by the spill of theAznalcóllar pyrite mine in Spain in 19981,2 are examples of the haz-ards arising from contaminating the environment with heavy met-als. Extensive remediation of metal ions in the soil and wateraround industrial plants is still a considerable challenge3. Althoughmost of the choices available so far rely on physicochemical proce-dures, the use of microorganisms and plants as environmentallyfriendly biosorption agents is receiving increasing attentionbecause of the lower costs and higher efficiency at lower metal con-centrations4,5. Immobilization of toxic metals and radionuclidesfrom waste streams employs different types of biomass as biosor-bents, and a wide variety of fungi, algae, and bacteria show promisefor this purpose6.

Immobilization is the major mechanism available to higherorganisms—including animals and humans—for counteractingheavy metal toxicity, and it may also become useful for bioremedia-tion strategies. For example, soil bacteria can be used as vectors tointroduce metallothioneins in polluted sites to immobilize other-wise soluble heavy metals. Metallothioneins (MTs) are a group ofsmall (around 60 amino acids) cysteine-rich eukaryotic proteinsthat bind heavy metals (e.g., Zn2+, Cd2+, Hg2+)7. Four mouse MTshave been cloned and thoroughly characterized at the molecularlevel7. They consist of two domains and are able to bind a total ofseven divalent metal ions. Previous studies have shown that humanMTs can be displayed functionally on the surface of Escherichia colicells when expressed as fusions to outer membrane (OM) proteinslike LamB, OmpA, and PAL, thus leading to an increased heavymetal adsorption by E. coli cells expressing the chimeric protein8,9.However, E. coli is not a satisfactory biological agent for soil biore-

mediation, since it is not a microorganism adapted to sites pollutedwith heavy metals.

In this work, we have engineered an autotransporter protein secre-tion system to anchor functional mouse MT molecules onto the cellsurface of R. eutropha CH34, a Gram-negative strain that possessesmultiple resistances to heavy metals and thrives in soils heavily pollut-ed with metal ions10. Autotransporters are a widespread family ofsecreted proteins found in Gram-negative bacteria that are able toindependently translocate through the OM11. Their mechanism ofexport, first exemplified by the IgA protease of Neisseria gonorroheae12,involves production of a large protein precursor containing an N-ter-minal signal peptide that directs the transport through the innermembrane (IM) into the periplasm. This is followed by the insertioninto the OM of the C-terminal domain of the protein (the helper or β-domain), which, by forming an amphipathic β-barrel, allows theaccompanying passenger N-terminal domain to cross the OM13,14. Byusing this secretion strategy, and a minitransposon vector for stablechromosomal integration of DNA15, we have constructed a recombi-nant R. eutropha strain that is endowed with an enhanced ability toadsorb Cd2+ from the medium and that causes a signficant reductionof the biological toxicity of the metal in contaminated soils.

ResultsProduction of MTβ in E. coli. We constructed a gene fusion namedmtb and expressed it in E. coli. This fusion encodes a modular pro-tein (MTβ) constituted by the pelB signal peptide assembled in-frame with the mouse MT (∼ 7 kDa) and the β-domain (∼ 50 kDa)from the IgA protease of N. gonorrhoeae. Additionally, the MTβ pro-tein was tagged with a short peptide epitope (E-tag) to allow its

Engineering a mouse metallothionein on the cell surface of Ralstonia eutropha

CH34 for immobilization of heavy metals in soil

Marc Valls1,2, Sílvia Atrian1, Víctor de Lorenzo2*, and Luis A. Fernández2

1Departament de Genètica, Facultat de Biologia, Av. Diagonal 645, Universitat de Barcelona, 08028-Barcelona, Spain. 2Departamento de BiotecnologíaMicrobiana, Centro Nacional de Biotecnología, Campus Cantoblanco, 28049-Madrid. Spain.

*Corresponding author (vdlorenzo@cnb.uam.es).

Received 2 February 2000; accepted 2 May 2000

Here we describe targeting of the mouse metallothionein I (MT) protein to the cell surface of the heavymetal-tolerant Ralstonia eutropha (formerly Alcaligenes eutrophus) CH34 strain, which is adapted tothrive in soils highly polluted with metal ions. DNA sequences encoding MT were fused to the autotrans-porter β-domain of the IgA protease of Neisseria gonorrhoeae, which targeted the hybrid protein towardthe bacterial outer membrane. The translocation, surface display, and functionality of the chimeric MTβprotein was initially demonstrated in Escherichia coli before the transfer of its encoding gene (mtb) to R.eutropha. The resulting bacterial strain, named R. eutropha MTB, was found to have an enhanced abilityfor immobilizing Cd2+ ions from the external media. Furthermore, the inoculation of Cd2+-polluted soil withR. eutropha MTB decreased significantly the toxic effects of the heavy metal on the growth of tobaccoplants (Nicotiana bentamiana).

Keywords: Alcaligenes, bioremediation, metallothionein, Ralstonia, surface display, heavy metals

RESEARCH ARTICLES

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

immunodetection by a monoclonal antibody (anti-E-tag MAb). Theplasmid pMTβ-1 (Fig. 1A) bears the mtb gene fusion under the con-trol of the lac promoter of E. coli .

The subcellular localization of the MTβ hybrid was addressedfirst. Furthermore, the influence of DsbA, the major periplasmicdisulfide bond-forming catalyst16, was investigated as well. This wasbecause previous studies had established that oxidation of cysteineresidues in the periplasm may prevent the exposure of cysteine-containing passenger proteins fused to the β-domain17. To this end,the dsbA mutant strain E. coli JCB57118 and its isogenic wild-typestrain E. coli JCB570 were transformed with plasmid pMTβ-1 andthe expression of mtb induced by addition of isopropyl-1-thio–D-galactoside (IPTG). After induction, E. coli cells were harvested andthe proteins of the soluble, IM and OM fractions analyzed bySDS–PAGE, blotted, and probed with the anti-E-tag MAb. Asshown in Figure 1B, a major protein of the expected size for MTβ(∼ 57 kDa) was localized in the OM fraction of both wild type anddsbA E. coli (pMTβ-1) cells. Interestingly, the presence or absence ofDbsA did not make a significant difference in the stability of MTβor in its targeting to the OM.

To assess whether the MT moiety of the MTβ hybrid wasexposed to the external medium, we performed ELISA with intactE. coli cells using the anti-E-tag MAb to detect the presence of theMTβ hybrid on their surface. Osmotically shocked and lysed E.coli cells producing MTβ were also adsorbed onto the ELISAplates as positive controls for immunodetection, whereas non-transformed cells were used as negative controls. The OD492 valuesobtained with the intact cells compared with those obtained onlysed cells are a measure of the amount of MT moiety translocatedto the external side of the OM by the IgA β-domain. Using thisapproach, specific recognition of the anti-E-tag MAb to intact E.coli cells producing MTβ was observed (Fig. 1C), indicating thatthe MT moiety was exposed to the external medium. On this crite-rion, wild type E. coli cells fully translocated about 20% of the MT

moieties of the MTβ fusions toward the external medium. Thelevel of the MT moiety displayed on the cell surface was threefoldhigher (∼ 60%) in the E. coli dsbA mutant strain (Fig. 1C), a featurethat seems to be connected with the secretion mechanism of auto-transporters14.

In order to determine whether the expression of MTβ increasedthe capacity of E. coli cells to adsorb heavy metals in vivo, culturesof E. coli JCB570 and E. coli JCB571, transformed with pMTβ-1,were grown in the presence of 30 µM CdCl2. Bacteria were thenharvested and the Cd2+ content accumulated by the biomass mea-sured by atomic adsorption spectrometry. As shown in Figure 2A,the production of MTβ augmented the Cd2+ content of the E. colicells by 10-fold, thus indicating that the MT moiety of MTβretained its metal-binding capabilities. The production of MTβ inan E. coli dsbA mutant host (Fig. 2A) did not further increase thismetal accumulation, in spite of the enhanced surface translocationof the MT domain.

Production of MTβ in R. eutropha CH34. The limited survivalof E. coli in heavy metal-polluted environments prompted us tointroduce the mtb gene in the metal-tolerant R. eutropha CH34strain10,19. The mtb gene was placed into a mini-Tn5 (Kanr) transpo-son15 downstream of the Pm promoter from the pWW0 plasmid ofPseudomonas putida, along with the gene encoding its cognate tran-scriptional activator xylS20. The plasmid obtained, pTnMTβ-1,enables the stable integration of the minitransposon TnMTβ-1 intothe chromosome of a wide variety of Gram-negative bacteria andallows the induction of mtb expression by addition of 3-methyl-benzoate (3-MB) into the growth medium (Fig. 3A).

TnMTβ-1 was inserted into the chromosome of R. eutrophaCH34, and the production of the MTβ was assayed byimmunoblot after induction with 3-MB (see ExperimentalProtocol). As can be observed in Figure 3B, the production ofMTβ was strictly dependent on the presence of the 3-MB in thegrowth medium, thus demonstrating the tight control of the Pmpromoter in R. eutropha CH34. The bacterial strain bearingTnMTβ-1 stably integrated in its chromosome was named R.eutropha MTB. Similarly to the case of E. coli, the MTβ hybrid wasentirely localized in the OM fraction of R. eutropha MTB (Fig.3C), showing that the hybrid protein was correctly targeted to theOM by the autotransporter domain. Furthermore, the MTβfusion was produced as a single polypeptide band, with no indica-tion of instability or proteolytic degradation.

662 NATURE BIOTECHNOLOGY VOL 18 JUNE 2000 http://biotech.nature.com

RESEARCH ARTICLES

Figure 1. Genetic structure of pMTβ-1 and production of MTβ in E. coli. (A) Scheme of pMTβ-1 plasmid containing the mtb genefusion. The lacI gene and lac promoter (open triangle) are indicated.The domain structure and basic topology of MTβ protein is alsosketched. The locations of the pelB signal peptide (ss, not present inthe mature protein), the MT domain, the E-tag peptide, and the β-domain from N. gonorrhoeae IgA protease are shown. (B) Immunoblotof proteins from soluble, inner membrane (IM) and outer membrane(OM) fractions of E. coli JCB570 (wt) and E. coli JCB571 (dsbA) cellsharboring pMTβ-1 and induced with IPTG. Anti-E-tag MAb was usedto detect the MTβ hybrid (57 kDa) in protein extracts. (C) Display ofMTβ on the surface of E. coli. The accessibility of MTβ protein in intact(I), osmotically shocked (S), and lysed (L) cells was probed using anti-E-tag MAb in ELISA (Experimental Protocol). The negative controlincluded isogenic E. coli cells devoid of the pMTβ-1 plasmid.

Figure 2. Accumulation of Cd2+ by E. coli and R. eutropha. The amountof Cd2+ bound by the bacterial cells was determined by atomicabsorption spectroscopy, and it is indicated in nanomoles of Cd2+ permilligram of dry cell mass. The values shown are the average of twoindependent experiments using duplicated samples on each cadmiumdetermination. (A) Escherichia coli JCB570 (wt) and E. coli JCB571(dsbA) cells, transformed with pMTβ-1 (+), or not (-), were induced withIPTG in MJS medium containing 30 µM CdCl2. (B) Accumulation of Cd2+

by R. eutropha strains. Cell cultures of R. eutropha MTB and itsparental strain R. eutropha CH34, were induced with 3-MB in MJSmedium containing 30 µM (left) or 300 µM (right) CdCl2.

A

B C

A B

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

NATURE BIOTECHNOLOGY VOL 18 JUNE 2000 http://biotech.nature.com 663

Adsorption of Cd2+ from the medium by R. eutropha MTB. Toinvestigate whether the expression of MTβ increased the capacity of R.eutropha cells to adsorb heavy metals, cultures of R. eutropha MTB andits parental control strain were grown in the presence of CdCl2. Forthese assays, low and high concentrations of the heavy metal wereemployed. First, 30 µM CdCl2 was used to compare the behavior of R.eutropha MTB to that of E. coli expressing MTβ. In addition, the metaladsorption of R. eutropha MTB cells was investigated after growth inthe presence of 300 µM CdCl2, a concentration that is inhibitory to thegrowth of many other bacteria such as E. coli. As shown in Figure 2B,the measurement of Cd2+ content revealed that R. eutropha MTBincreased by threefold the already significant natural heavy metaladsorption ability of R. eutropha CH3421,22. This relative threefoldincrease in Cd2+ accumulation of the recombinant MTB strainoccurred in the presence of both 30 and 300 µM CdCl2, although theabsolute adsorption levels varied in each case proportionally to thebasal levels. The average adsorption level obtained by R. eutropha MTBwas 42 nmol Cd2+ mg-1 of dry cells (Fig. 2B) when grown in the pres-ence of 300 µM CdCl2. This phenomenon occurred without any signif-icant variation in the growth rate of Ralstonia cultures (data notshown), indicating that neither the production of the MTβ hybrid northe increased metalloadsorption had noticeable effects on the bacteria.

Remediation of Cd2+ toxicity in soil. To test the ability of the R.eutropha MTB to counteract the toxic effect of Cd2+ in soil, a bioassaywas set up in which the immobilization of the heavy metal in a non-bioactive form was reflected by a positive effect on plant growth. Thetoxicity of the metal was initially assayed in Nicotiana bentamiana plantsgrown in soils containing an increasing CdCl2 content (not shown).These experiments revealed that 150 µmol of the Cd salt per kilogram ofsoil strongly diminished plant growth and caused severe chlorosis, twotypical features of Cd2+ toxicity23. This metal content was thus used forfurther experiments since it still allowed a slow plant growth.

To investigate the effect of the different R. eutropha strains onplant growth, N. bentamiana seedlings were transferred to standardor cadmium-contaminated sterile soils inoculated separately with108 bacteria g-1 of induced R. eutropha MTB, the CH34 strain, or asterile saline solution containing the inducer (see ExperimentalProtocol). After 40 days of growth in a greenhouse we observed areduction of the toxic effects of the heavy metal upon the plants inthe presence of R. eutropha MTB in the soil (Fig. 4B). Controlexperiments in soils that had not been supplemented with Cd2+

indicated that the presence of bacteria did not alter plant growth(Fig. 4A). Furthermore, a significant protective effect of the wild-type R. eutropha CH34 strain was observed. This was probably dueto the inherent heavy metal accumulation capability of CH34strain21,22,24 (see above). These observations were quantified andconfirmed by measuring plant biomass and chlorophyll content ofthe leaves (Table 1). Taken together, these data indicate that the pro-duction of MTβ on the cell surface of R. eutropha enhanced thealready existing heavy metal immobilization capacity of this bacter-ial strain, conferring a more protective effect upon N. bentamianagrowth in polluted soils.

RESEARCH ARTICLES

Figure 3. Structure of the transposon TnMTβ-1 and production ofMTβ in R. eutropha. (A) The genetic elements within TnMTβ-1 areindicated: the mtb gene, the kanamycin resistance gene (Km), thexylS gene (encoding the transcriptional activator XylS of the Pmpromoter), the Pm promoter, and the I and O ends of the mini-Tn5.The pelB signal sequence (ss) of mtb gene is also marked. (B)Immunoblot of whole-cell protein extracts from R. eutropha MTB,and its parental control strain (CH34), probed with anti-E-tag MAb todetect the presence of MTβ protein. The inducer of the Pm promoter,3-MB, was added to (+) or omitted from (-) the bacterial culture . (C)Immunoblot of proteins from soluble, inner membrane (IM) and outermembrane (OM) fractions of R. eutropha MTB cells induced with 3-MB and probed with anti-E-tag MAb.

Figure 4. Remediation of Cd2+ toxicity on plant growth by R. eutrophaMTB. (A) 15-day-old seedlings of N. bentamiana were grown for 40days in a sterile control soil (1), or in soil inoculated with 108 cells ofR. eutropha CH34 per gram (2) or 108 cells of R. eutropha MTB pergram (3) pregrown in MS medium containing citrate and 3-MB. Thecontrol soil was also blended with the same medium. (B) The sameprocedure described in (A) was performed in soils containing 150µmol of Cd2+ per kilogram.

Table 1. Effect of CdCl2 and R. eutropha on Nicotiana bentami-ana growtha

Soil treatment Cd Plant Chlorophyll bacteria biomass (g) (mg g-1)

None Yes 0.53 ± 0.36 0.40 ± 0.25R. eutropha CH34 Yes 1.29 ± 0.39 0.81 ± 0.31R. eutropha MTB Yes 2.37 ± 0.37 1.41 ± 0.25None No 17.65 ± 2.05 2.22 ± 0.10R. eutropha CH34 No 15.52 ± 1.69 2.50 ± 0.08R. eutropha MTB No 14.55 ± 2.37 2.38 ± 0.05

aMean values of four independent experiments at day 55 after plant germina-tion are shown. Biomass indicates the wet weight of the plant aerial portion ingrams. The chlorophyll content is shown as milligrams of pigment per gram ofplant wet weight. Where specified, the soil was inoculated with 108 bacteria pergram and 150 µmol of CdCl2 per kilogram was either added or omitted.

A

B C

A

B

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

DiscussionThe production of MT on the cell surface of R. eutropha, rather thanintracellularly, was expected to be a more effective way of enhancingthe heavy metal-binding capability of this bacterial strain withoutinducing alterations in its physiology. The fact that the resistancemechanisms known for various metals in R. eutropha CH34 involvesthe active efflux of heavy metal ions from the cytoplasm to the extra-cellular medium19,21,25,26 suggested that these cells can become excel-lent detoxification agents when endowed with superior metal cellsurface-adsorption properties.

Although the bioaccumulation reported in this study is one ofthe highest reported so far, the amount of cadmium removed repre-sents only ∼ 5% of the total metal present in the culture medium.However, the experiments of remediation of cadmium toxicity sug-gest that immobilization of Cd2+ ions in a biologically inactive formmight be much higher in soil than in vitro. Ralstonia eutropha MTBhad a much higher protective effect upon N. bentamiana growththan its parental strain. The presence of the engineered MTB strainin cadmium-polluted soils augmented the biomass production andchlorophyll content of N. bentamiana plants by fourfold (Table 1).The gross morphology and the size of the plants grown in soils con-taining R. eutropha MTB and 150 µmol of Cd kg-1 was comparable tothat of plants grown without bacteria at 40 µmol Cd kg-1 (notshown). Thus, these in situ assays imply an immobilization of ∼ 70%of the metal accessible to the plants in soil. The mechanism account-ing for the difference in the values of metal adsorption in vitro andin soil is unclear and deserve further studies.

A calculation of the number of MTβ proteins produced per R.eutropha MTΒ cell suggested that the immobilization of cadmiumcould not be explained by simple metal binding to MT. The compar-ison of the signals obtained in immunoblots with anti-E-tag MAb,using whole-cell extracts of R. eutropha MTB and an E-tagged pro-tein of known concentration, gave an estimation of a maximum of15,000 MTβ molecules/cell (data not shown). Since each MT mod-ule can bind seven divalent metal ions, the predicted number ofavailable coordination spheres is ∼ 105 per cell. From the valuesobtained in the cadmium-binding experiments shown in Figure 2B,it can be inferred that the cadmium content is 1–5 × 106 Cd atomsper cell. Thus, the actual number of Cd2+ ions bound to the surfaceof R. eutropha MTB exceeds by at least 10-fold the predictions ofenhanced accumulation, solely on the basis of the increase in thenumber of metal-binding centers on the cell surface. Similar conclu-sions have been also reported for other artificial metal-bindingschemes8,27. These results suggest that the MTβ fusion channelsmetal ions toward other cell structures that cannot otherwise inter-act with the metals in solution. Such a “funnel” effect results in amultiplication of the native ability of the bacteria to accumulatemetal ions rather than the simple addition of coordination sphereson the cell surface. However, this effect still seems insufficient toexplain the large reduction of metal toxicity in soil.

Besides the intrinsic affinity of the MT for divalent ions, the over-all reactivity of the R. eutropha CH34 strain (which can naturally pre-cipitate heavy metals on the bacterial surface21,22,24,28) and the sur-rounding physicochemical processes, could contribute synergisticallyto the final outcome of metal immobilization in soil. Therefore, thedifference observed between the in vitro and in situ assays could beexplained if the MTβ molecules present on the surface of the engi-neered bacteria not only coordinate the metal ions but also catalyze aprocess of abiotic precipitation of metal phosphates and carbonates28.

In conclusion, these results address the biotechnological poten-tial of MTs expressed in R. eutropha and, in addition, provide directevidence that the use of heavy metal-resistant bacteria as hosts forthe surface display of high-affinity ion-binding proteins is a promis-ing approach for the generation of biological amendments to soilcontaminated with such recalcitrant pollutants as these.

Experimental protocolBacterial strains and growth conditions. The bacterial strains used were asfollow: E. coli JCB570 (MC1000 phoR zih12:Tn10); E. coli JCB571 (MC1000phoR zih12:Tn10 dsbA:kan1)18; E. coli DH5αF’ (F’ ∆(lacZYA-argF)U169 deoRendA1 hsdR17 supE44 thi-1 recA1 gyrA96 relA1 (φ80dlacZ ∆M15); E. coliCC118 λpir (∆ara leu araD ∆lac X74 galE galK phoA20 thi-1 rpsE rpoB argE(Am) recA1), and E. coli S17-1 λpir (Tpr Smr recA thi pro hsdR- M+ RP4-2-Tc:Mu:Km Tn7) 15. Ralstonia eutropha (formerly Alcaligenes eutrophus)CH34 was the kind gift of M. Mergeay10. Unless otherwise indicated, bacteriawere grown at 30°C in liquid LB medium or in LB–agar 1.5% (wt/vol) platessupplemented with the appropriate antibiotics29. Glucose (2% wt/vol) wasincluded in the growth medium for full repression of the lac promoter ofpMTβ-1 in E. coli. Expression of MTβ was induced in E. coli (pMTβ-1) bytransferring the cells of exponentially growing cultures (OD600 ∼ 0.5) to freshLB medium devoid of glucose and containing 0.1 mM IPTG, and furtherincubated for 3 h at 30°C. Ampicillin was used at 100 µg ml-1 and chloram-phenicol at 40 µg ml-1 final concentrations. For bioaccumulation studies,minimal low-phosphate MJS medium was used instead of LB. The MJS medi-um consists of 12.5 mM HEPES (pH 7.1) with 50 mM NaCl, 20 mM NH4Cl,1 mM KCl, 1 mM MgCl2, 0.05 mM MnCl2, 0.8% (wt/vol) Casamino acids(GibcoBRL; Life Technologies S.A., Barcelona, Spain), 4% (vol/vol) glycerol,0.005% (wt/vol) thiamine. Production of MTβ in R. eutropha was induced inLB and MJS liquid cultures by adding 3 mM 3-MB and further incubation at30°C for 3 h. For the inoculation of soil with R. eutropha, bacterial cultureswere grown at 30°C in Murashige and Skoog (MS) liquid medium (Sigma, St.Louis, MO) containing 1.25 mM MES buffer pH 5.8 and 0.1% (wt/vol) cit-rate as carbon source. Overnight-grown cultures were diluted 1:10 in thesame medium, supplemented with 300 µM CdCl2 if required, and inducedwith 100 µM 3-MB. Cultures were further grown for 18 h at 30°C.

Gene constructs. DNA manipulations were performed using standardmethods30. The mouse MT-I gene was amplified from plasmid pMTP9 bypolymerase chain reaction (PCR) using oligonucleotides SMt1 (5′-CGGGCC CAG CCG GCC ATG GCG GAC CCC AAC TGC TCC TGC) and SMt2(5′-GCG GCC CCC GAC GCC GCG GCA CAG ACA GTG CAC TTG TC).The 200 bp amplified DNA fragment, containing the MT-I gene flanked bySfiI sites, was digested with SfiI and ligated into the 5.2 kb fragment obtainedby SfiI digestion of pFvHβ14. The resulting plasmid, pMTβ-0, contained anXbaI and HindIII sites flanking the mtb gene, which were converted into NotIsites in pMTβ-1 (Fig. 1) by ligating the XbaI-NotI and HindIII-NotI linkers.The XbaI-NotI linker was made by hybridizing the oligonucleotides 5′-CTAGGCGGCCGC-3′ and 5′-CTAGGCGGCCGC-3′. Similarly, the HindIII-NotI linker was made by hybridizing the oligonucleotides 5′-AGCTGCGGC-CGC-3′ and 5′-AGCTGCGGCCGC-3′. For the construction of pTnMTβ-1,the 1.7 kb DNA fragment containing the mtb gene was isolated a by NotIdigestion of pMTβ-1, and ligated into the unique NotI site of pCNB131 in theorientation that places mtb under the control of the Pm promoter (Fig. 3).

Transfer of TnMTβ-1 to R. eutropha CH34. The mini-Tn5 elementTnMTβ-1was transferred into the chromosome of R. eutropha CH34 by con-jugation with E. coli S17-1 λpir transformed with plasmid pTnMTβ-1 usingthe protocol described15. Selection of R. eutropha CH34 transconjugants wasperformed on M9-minimal medium plates29 containing citrate 0.2% wt/voland supplemented with kanamycin (1 mg ml-1). Several clones were isolatedand tested for piperacillin-sensitive phenotype (100 µg ml-1), which indicatedthat they corresponded to authentic transposition events rather than integra-tion of the whole pTnMTβ-1 into the genome.

Protein methods. Fractionation of E. coli and R. eutropha cells was carriedout using a method based on differential solubilization of cell compartmentswith 1.5% Triton X-100 as well as a standard sucrose gradient–centrifugationprocedure32. The concentration of proteins in the different fractions wasquantified by the Bradford method as described in Ausubel and coworkers30.Identical amounts of protein from the different samples were separatedthrough 10% SDS–PAGE, and transferred to a polyvinylidene difluoridemembrane (Millipore, Bedford, MA). For immunodetection of the MTβprotein, the PVDF membrane was incubated with anti-E-tag MAb (1 µg ml-1;Pharmacia, Uppsala, Sweden) in B-buffer (PBS + 3% wt/vol skimmed milk)containing 0.1% (vol/vol) Tween 20. An anti-mouse IgG–peroxidase conju-gate (0.03 U ml-1; Boehringer Mannheim, Mannheim, Germany) was used assecondary antibody. The membrane was developed by a chemiluminiscencereaction14 and exposed to an X-ray film (X-OMAT; Kodak).

ELISA analysis. To determine the level of displayed MTβ on the surface ofE. coli, the induced cells were washed in PBS and resuspended at OD600 = 1.5in PBS to obtain the intact cells sample. Alternatively, washed cells were

664 NATURE BIOTECHNOLOGY VOL 18 JUNE 2000 http://biotech.nature.com

RESEARCH ARTICLES

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

NATURE BIOTECHNOLOGY VOL 18 JUNE 2000 http://biotech.nature.com 665

resuspended at the same density in PBS with 20% (wt/vol) sucrose and 10mM EDTA to obtain the shocked cells sample. To generate a sample of lysedcells, bacteria were resuspended (OD600 = 1.5) in PBS containing 10 mMEDTA and sonicated using six pulses of 30 s in a Labsonic U (B. Braun,Allentown, PA) apparatus. Either intact, osmotically shocked, or lysed sam-ples (50 µl) were adsorbed to ELISA plates (Maxisorb; Nunc, Roskilde,Denmark) for 1 h at room temperature, blocked with B-buffer for 2 h, andincubated for 1 h in the B-buffer with anti-E-tag MAb (1 µg ml-1;Pharmacia). Plates were then washed three times in PBS, and an anti-mouseIgG–peroxidase conjugate (0.3 U ml-1; Boehringer Mannheim) was used as asecondary antibody. The ELISAs were developed using o-phenylenediamine(OPD; Sigma) and their OD492 values determined.

Cd2+ bioaccumulation. E. coli and R. eutropha cells, containing mtb gene,or their corresponding controls, were grown in LB medium at 30°C until theOD600 reached ∼ 0.3. Then, the cells were harvested and resuspended intofresh MJS medium (to avoid precipitation of the metal) supplemented withthe appropriate concentration of CdCl2 and inducer (IPTG or 3-MB), andfurther incubated until the OD600 reached ∼ 1.5. The cells were then harvest-ed, washed with 0.8% NaCl in 5 mM HEPES (pH 7.1), and dried for 20 h at65°C. Dry material was digested overnight with 70% nitric acid and the Cd2+

concentrations of the resulting solution measured with a Hitachi Z-8200spectrophotometer26.

Plant growth conditions. A standard peat soil (Gebr. Brill SubstrateGmbH, Georgsdorf, Germany), sterilized by autoclaving and containingone-third vermiculite, was used for plant growth. Soil was mixed 1:1 (vol/wt)with induced R. eutropha cultures (108 bacteria g-1), or a sterile MS mediumcontaining the inducer, and supplemented with 300 µM CdCl2 if required.The final concentration of the metal in soil was 150 µmol Cd kg-1. The treat-ed soil was distributed in pots (400 g per pot), and 15-day-old N. bentamianaseedlings were then transferred. The plants were grown in a greenhouse andwatered twice a week with 0.1× MS medium, 0.12 mM MES pH 5.8, 0.01%(wt/vol) citrate. For chlorophyll determination, the leaves were homogenizedin acetone:water (9:1) and the OD600 was determined. The extinction coeffi-cient used for chlorophyll was 34.5.

AcknowledgmentsWe thank Jon Beckwith and Max Mergeay for materials and strains used in thiswork, Sofía Fraile for her excellent technical work, and N. van der Lelie forinspiring discussions. We are grateful to Roser Gonzàlez-Duarte for her partici-pation and continuous support to this work. M.V. held a predoctoral fellowshipfrom the Generalitat de Catalunya (Spain). This work was supported in part byEU contracts BIO4-CT97-2123 and BIO4-CT97-2183 and by the SpanishCICYT grants BIO98-0808 and PB96-0225.

1. de Lorenzo, V. & Kuenen, G. Scientific basis for the remediation of the toxic spillof Aznalcóllar mine: combining bacteria and plants to address an intractable kindof pollution. Environ. Microbiol. 1, 275–278 (1999).

2. Grimalt, J.O., Ferrer, M. & Macpherson, E. The mine tailing accident inAznalcollar. Sci. Total Environ. 242, 3–11 (1999).

3. Lovley, D.R. & Coates, J.D. Bioremediation of metal contamination. Curr. Opin.Biotechnol. 8, 285–289 (1997).

4. Gadd, G.M. & White, C. Microbial treatment of metal pollution—a workingbiotechnology? Trends Biotechnol. 11, 353–359 (1993).

5. Chaney, R.L. et al. Phytoremediation of soil metals. Curr. Opin. Biotechnol. 8,279–284 (1997).

6. Kratochvil, D. & Volesky, B. Advances in the biosorption of heavy metals. TrendsBiotechnol. 16, 291–300 (1998).

7. Nordberg, M. Metallothioneins: historical review and state of knowledge. Talanta46, 243–254 (1998).

8. Sousa, C., Kotrba, P., Ruml, T., Cebolla, A. & Lorenzo, V.d. Metalloadsorption byEscherichia coli cells displaying yeast and mammalian metallothioneinsanchored to the outer membrane protein LamB. J. Bacteriol. 180, 2280–2284(1998).

9. Valls, M., González-Duarte, R., Atrian, S. & Lorenzo, V.D. Bioaccumulation ofheavy metals with proteins fusions of metallothionein to bacterial OMPs.Biochimie 80, 855–861 (1998).

10. Mergeay, M. et al. Alcaligenes eutrophus CH34 is a facultative chemolitotrophwith plasmid-bound resistance to heavy metals. J. Bacteriol. 162, 328–334(1985).

11. Henderson, I.R., Navarro-Garcia, F. & Nataro, J.P. The great escape: structure andfunction of the autotransporter proteins. Trends Microbiol. 6, 370–378 (1998).

12. Pohlner, J., Halter, R., Beyreuther, K. & Meyer, T.F. Gene structure and extracellu-lar secretion of Neisseria gonorrhoeae IgA protease. Nature 325, 458–462 (1987).

13. Klauser, T., Krämer, J., Otzelberger, K., Pohlner, J. & Meyer, T.F. Characterizationof the Neisseria Igaβ-core. The essential unit for outer membrane targeting andextracellular protein secretion. J. Mol. Biol. 234, 579–593 (1993).

14. Veiga, E., De Lorenzo, V. & Fernandez, L.A. Probing secretion and translocationof a beta-autotransporter using a reporter single-chain Fv as a cognate passen-ger domain. Mol. Microbiol. 33, 1232–1243 (1999).

15. de Lorenzo, V. & Timmis, K.N. Analysis and construction of stable phenotypes inGram-negative bacteria with Tn5- and Tn10-derived minitransposons. MethodsEnzymol. 235, 386–405 (1994).

16. Bardwell, J.C.A. Building bridges: disulphide bond formation in the cell. Mol.Microbiol. 14, 199–205 (1994).

17. Jose, J., Krämer, J., Klauser, T., Pohlner, J. & Meyer, T.F. Absence of periplasmicDsbA oxidoreductase facilitates export of cysteine-containing passenger pro-teins to the Escherichia coli cell surface via the Igaβ autotransporter pathway.Gene 178, 107–110 (1996).

18. Bardwell, J.C.A., McGovern, K. & Beckwith, J. Identification of a protein requiredfor disulfide bond formation in vivo. Cell 67, 581–589 (1991).

19. Tibazarwa, C., Wuertz, S., Mergeay, M., Wyns, L. & van Der Lelie, D. Regulationof the cnr cobalt and nickel resistance determinant of Ralstonia eutropha(Alcaligenes eutrophus) CH34. J. Bacteriol. 182, 1399–1409 (2000).

20. Ramos, J.L. & Marqués, S. Transcriptional control of the Pseudomonas TOLplasmid catabolic operons is achieved through an interplay of host factors andplasmid-encoded regulators. Ann. Rev. Microbiol. 51, 341–373 (1997).

21. Diels, L., Dong, Q., van der Lelie, D., Baeyens, W. & Mergeay, M. The czc operonof Alcaligenes eutrophus CH34: from resistance mechanims to the removal ofheavy metals. J. Indust. Microbiol. 14, 142–153 (1995).

22. Diels, L., De Smet, M., Hooyberghs, L. & Corbisier, P. Heavy metals bioremedia-tion of soil. Mol. Biotechnol. 12, 149–158 (1999).

23. Ouzounidou, G., Moustakas, M. & Eleftheriou, E.P. Physiological and ultrastruc-tural effects of cadmium on Wheat (Triticum aestivum L.) leaves. Arch. Environ.Contam. Toxicol. 32, 154–160 (1997).

24. Van Roy, S., Peys, K., Dresselaers, T. & Diels, L. The use of Alcaligenes eutrophusbiofilm in a membrane bioreactor for heavy metal recovery. Res. Microbiol. 148,526–528 (1997).

25. Nies, D.H. & Silver, S. Plasmid-determined inducible efflux is responsible forresistance to cadmium, zinc, and cobalt in Alcaligenes eutrophus. J. Bacteriol.171, 896–900 (1989).

26. Mergeay, M. Towards an understanding of the genetics of bacterial metal resis-tance. Trends Biotechnol. 9, 17–24 (1991).

27. Sousa, C., Cebolla, A. & de Lorenzo, V. Enhanced metalloadsorption of bacterialcells displaying poly-His peptides. Nat. Biotechnol. 14, 1017–1020 (1996).

28. Gilis, A. et al. Effect of the siderophore alcaligin E on the bioavailability of Cd toAlcaligenes eutrophus CH34. J. Ind. Microbiol. Biotechnol. 20, 61–68 (1998).

29. Miller, J.H. A short course in bacterial genetics: a laboratory manual and hand-book for Escherichia coli and related bacteria. (Cold Spring Harbor LaboratoryPress, Cold Spring Harbor, New York; 1992).

30. Ausubel, F.M. et al. Current protocols in molecular biology. (John Wiley & Sons,New York, NY; 1994).

31. de Lorenzo, V., Fernandez, S., Herrero, M., Jakubzik, U. & Timmis, K.N.Engineering of alkyl- and haloaromatic-responsive gene expression with mini-transposons containing regulated promoters of biodegradative pathways ofPseudomonas. Gene 130, 41–46 (1993).

32. Nikaido, H. Isolation of outer membranes. Methods Enzymol. 235, 225–234(1994).

RESEARCH ARTICLES

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om