Bioaccumulation of heavy metals by fimbrial designer adhesins


*Corresponding author. Fax: +45 45 93 28 09; E-mail:


Naturally occurring adhesins bind to specific molecular targets in a lock-and-key fashion due to the composition of the binding domain of the adhesin. By introduction of random peptide libraries in a suitable surface exposed carrier protein it is possible to create and select designer adhesins with novel binding affinities. Type 1 fimbriae are surface organelles of Escherichia coli which mediate d-mannose sensitive binding to different host surfaces through the FimH adhesin, an integral part of these organelles. We have studied the ability of the FimH adhesin to display random peptide sequences. By serial selection and enrichment procedures specific sequences were identified which conferred the ability on recombinant cells to adhere to various metal oxides (PbO2, CoO, MnO2, Cr2O3). The properties inherent in these sequences permitted the distinct recognition of metals to varying degrees, indicating that this system allows for the isolation of peptide sequences with a variety of binding avidities. These studies demonstrate the potential and versatility of the FimH display system for presenting random peptide sequences. In addition, the possibility exists for the construction of microorganisms for the bioaccumulation of heavy metals from the environment.


Surface proteins of bacteriophage virions, bacteria and yeasts have been used to display heterologous sequences consisting of both defined and random peptides [1–3]. The rationale behind random peptide display is similar to that used in in vitro evolution studies with random libraries of nucleic acid sequences [4]. This technology enables the construction of huge populations of diverse macromolecules and the subsequent selection of specific peptide sequences from these populations on the basis of their binding affinity to a target molecule.

We are particularly interested in type 1 fimbriae for display of random peptide libraries. Type 1 fimbriae are found on the majority of Escherichia coli strains and are widespread among other members of the Enterobacteriaceae[5]. A type 1 fimbria is a thin, 7 nm wide and approximately 1 μm long, rod-shaped surface organelle. It is a heteropolymer consisting of four different subunits. Approximately 1000 copies of the major building element, FimA, are polymerized into a right-handed helical structure also containing a few percent of the minor components, FimF, FimG, and FimH [6]. Type 1 fimbriae are adhesive organelles due to the presence of the FimH adhesin. FimH is an integral part of the fimbrial structure and specifically recognizes α-d-mannosides [7]. The FimH protein is produced as a precursor of 300 amino acids, which is processed into a mature form of 279 amino acids [6, 8]. The results of linker insertion mutagenesis [9], and hybrid proteins constructed by fusions to FocH [10], suggest that the FimH protein consists of two major domains, each constituting roughly one half of the molecule; the N-terminal domain seems to contain the receptor binding site, while the C-terminus seems to contain the recognition sequences for export and bioassembly. In previous studies we have investigated the ability of FimH to present heterologous peptides in connection with the development of vaccine systems [11]. These observations led us to believe that the FimH protein was an ideal scaffold for the display of random peptide libraries and subsequently for the construction of designer adhesins (i.e. proteins engineered to bind to targets of choice). Here we demonstrate the construction of such designer adhesins which can confer binding of bacterial cells to heavy metals.

2Materials and methods

2.1Bacterial strains, plasmids and growth conditions

In this study we used the E. coli K-12 strain S1918 (F′lacIQΔmalB101 endA hsdR17 supE44 thiI relA1 gyr96ΔfimB-H::kan) [12]. Cells were grown in Luria-Bertani (LB) medium supplemented with the appropriate antibiotics. Our FimH display system consists of two plasmids, the FimH expression vector pLPA30 and an auxiliary plasmid pPKL115. Plasmid pLPA30 is a pUC18 derivative containing the fimH gene downstream of the lac promoter. A BglII linker, located in a position corresponding to amino acid 225 [11], was used for integration of the random library. Plasmid pPKL115 is a pACYC184 derivative containing the whole fim gene cluster with a translational stop linker inserted in the fimH gene [11].

2.2DNA techniques

Isolation of plasmid DNA was carried out using the QIAprep Spin Plasmid Kit (Qiagen). Restriction endonucleases were used according to the manufacturer's specifications (Biolabs or Pharmacia). Polymerase chain reactions to monitor the size and distribution of the random library were performed as previously described [13]. The oligonucleotide primers used in these reactions were P1: 5′-CCTGCACAGGGCGTCGGCGTAC and P2: 5′-GGAATAATCGTACCGTTGCG. The nucleotide sequences of inserts conferring the ability of cells to bind to metal oxides were determined by the dideoxynucleotide chain termination method [14].

2.3Construction of the random library

Construction of the random library was performed essentially as described by Brown [12]. Briefly, a template oligonucleotide containing the sequence 5′-GGACGCAGATCT(VNN)9AGATCTAGCACCAGT-3′ was chemically synthesized (N indicates an equimolar mixture of all four nucleotides and V indicates an equimolar mixture of A, C and G). A primer oligonucleotide 5′-ACTGGTGCTAGATCT-3′ was hybridized to the template oligonucleotide and extended with Klenow fragment of DNA polymerase I. The double stranded oligonucleotide was purified by phenol-chloroform extraction and digested with BglII to release an internal 33-bp fragment. This was purified by electrophoresis through a 12% polyacrylamide gel in TBE and eluted into a buffer containing 10 mM Tris-HCl, pH 8.0, 2 mM EDTA, 0.15 M NaCl. The eluate was filtered through a 0.22 μm pore size Qiagen filter, concentrated by ethanol precipitation and redissolved in a buffer containing 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 0.1 M NaCl. The redissolved 33-bp BglII fragment was ligated at various ratios to BglII digested pLPA30. The ligation products were precipitated with ethanol and electroporated into S1918(pPKL115).

The diversity of the library was calculated to be 4×107 individual clones based on extrapolation from numbers of transformants obtained in small scale planing. The transformation mixture was made up to 10 ml and grown for approximately seven generations (4×109 cells). Aliquots (1 ml) were frozen at −80°C in 25% (v/v) glycerol. Each 1-ml aliquot contained approximately 4×108 cells, which represented 10 times the library diversity. Random screening of clones by PCR revealed a predominance of one to three 33-bp oligonucleotide inserts; sequencing of the inserts from randomly selected clones revealed G+C contents ranging from 30 to 70%.

2.4Enrichment procedure

The enrichment procedure for identifying metal oxide binding clones from the random library was as follows. Mid-exponential cultures were diluted into M63 salts [15] containing 20 mM methyl-α-d-mannopyranoside and 75% (v/v) Percoll (Pharmacia). The methyl-α-d-mannopyranoside was added to block the natural receptor binding domain of the FimH adhesin. The use of Percoll permitted the formation of a density gradient upon centrifugation which resulted in pelleting of the metal oxides and allowed the specific separation of any adhering bacteria from non-adherent bacteria. Under these conditions, bacteria expressing wild-type FimH proteins as components of type 1 fimbriae did not sediment with the metal oxides. The metal oxides and bacteria expressing the random peptide library within FimH were mixed and allowed to adhere at room temperature with gentle agitation. Centrifugation was then performed, the metal oxides and any adhering bacteria were recovered and inoculated into LB containing appropriate antibiotics. After overnight incubation, exponentially growing cultures were established and the enrichment procedure repeated. Following each cycle of enrichment aliquot of the populations were stored at −80°C. Plasmid DNA was prepared from each aliquots and used in PCR to monitor the size distribution of the inserts in the population.

2.5Binding to metals

Metal oxides (PbO2, MnO2, Cr2O3 and CoO) and metal chloride (MnCl2) were purchased from Aldrich. Particles of appropriate size for microscopy were prepared by differential centrifugation. Metal oxides were suspended in M63 salts prior to the addition of bacteria. Samples were incubated at room temperature for 15 min with gentle agitation and examined microscopically. Quantification of the ability of individual clones to bind to metal oxides was performed in each case by counting attached cells to 20 metal particles in four randomly chosen frames of about five particles. Bioaccumulation of Mn2+ by recombinant bacteria was measured by atomic absorption. Late exponential phase cultures were washed in M63 salts and resuspended in the same medium containing 20 μM MnCl2. The cells were incubated for 30 min to allow adsorption of the metal ions and washed twice in M63 salts. Samples were prepared and analyzed on a Perkin Elmer 2100 atomic absorption spectrophotometer as previously described [16].

2.6Agglutination of yeast cells

The capacity of bacteria to express a d-mannose binding phenotype was assayed by their ability to agglutinate yeast cells (Saccharomyces cerevisiae) on glass slides. Aliquots of washed bacterial suspensions at OD550=1.0 and 10% yeast cells were mixed and the time until agglutination occurred was measured.


3.1Construction and characterization of a random peptide library in FimH

In this study, we used the FimH expression vector pLPA30, which contains the fimH gene with an in-frame BglII linker inserted at a position encoding amino acid residue 225 and placed under transcriptional control of the lac promoter. In order to express chimeric FimH as functional constituents of fimbriae, we also used an auxiliary plasmid (pPKL115) encoding the rest of the fim gene cluster (Fig. 1). A random peptide library was constructed in the FimH protein by inserting various numbers of synthetic double stranded oligonucleotides into the BglII site in amino acid position 225 of the fimH gene. The double stranded oligonucleotides consisted of nine random codons flanked by BglII restriction sites, encoding arginine and serine. This genetic structure permits the construction of libraries containing different numbers of double stranded 33-bp oligonucleotides, a feature which greatly enhances the complexity of the libraries. In addition, the distribution of the population through the enrichment procedure can be monitored by PCR amplification across the insert region using primers complementary to the vector sequence flanking the insertion site.

Figure 1.

Overview of the plasmids used in the FimH display system. Only relevant non-vector regions are shown. Plasmid pPKL115 contains the entire fim gene cluster with a translational stop linker inserted in the fimH gene (indicated by a triangle). The FimH expression vector pLPA30 is shown along with the BglII insertion site at position 225 and the two primers (P1 and P2) used to monitor the size and distribution of the random library.

3.2Isolation and analysis of metal binding sequences

Serial selection and enrichment of the random library was performed against either PbO2, MnO2, Cr2O3 or CoO. To isolate cells adhering to each of the metal oxides we used a 75% (v/v) Percoll solution which formed a density gradient upon centrifugation. Under these conditions only cells adhering to the metal oxides were able to sediment when centrifuged. Monitoring of the insert population by PCR revealed a change in its size distribution of inserts, after four cycles of selection and enrichment against each of the metal oxides. In a control experiment, the same number of cycles of growth of the population, washing in M63 salts in the absence of metal oxides and regrowth did not alter the size distribution of the insert sequences (data not shown). For each of the metal oxides, 20 colonies were randomly selected from plating of the fourth enrichment populations. Each colony was regrown and examined for metal binding by phase contrast microscopy. Only a subset of the clones displaying a metal binding phenotype were examined further. The fimH containing plasmids were isolated from these strains and the insert region sequenced. The metal binding sequences are shown in Table 1.

Table 1.  Sequences conferring the ability of cells to adhere to various metal oxides
Metal oxideEnriched sequencePlasmid
  1. The three different binding motifs identified from the enriched sequences are underlined (H/R-X3-H-R/K-S), double-underlined (S/T-K/R-X2-AR) or in italics (H/V-RRS), respectively. Small rs letters represent amino acids encoded by the BglII linkers.


A number of different peptide sequences were enriched which could confer the ability to bind to the various metal oxides tested. In the case of binding to PbO2, seven different sequences were identified. Of these sequences plasmids pKKJ63 and pKKJ69 were represented three and two times, respectively. The size of these inserts ranged from one to four double stranded 33-mer oligonucleotides in length. Examination of the sequences revealed some structural similarities in the amino acid sequences conferring binding. Two motifs, comprising the amino acid sequences H/R-X3-H-R/K-S or S/T-K/R-X2-AR, could be discerned from the data (Table 1). Interestingly, the R-X3-HRS motif was also observed in two of the five sequences independently enriched for binding to CoO. A sequence for binding to MnO2 was also identified. Three of the five sequences contained a H/V-RRS motif. Of interest also was the presence of an unpaired cysteine residue in two of the sequences. No cysteine residues were identified in any of the other metal binding sequences. The FimH protein contains four cysteine residues which participate in the formation of two disulfide bridges in its tertiary structure. Although cysteine has been shown to participate in metal binding, it is likely that this display system would be biased against the insertion of cysteine residues into FimH. No binding motif could be elucidated from the Cr2O3 binding sequences.

3.3Re-transformation into S1918(pPKL115) and phenotypic characterization

The plasmids identified from the random library which conferred the ability to bind to each of the metal oxides were purified and re-transformed into S1918(pPKL115). The new recombinant clones displayed the same binding phenotype as the original isolates, indicating that the binding phenotype was indeed plasmid encoded. Fig. 2 shows the binding of one representative clone from each of the selections. Despite originally being selected in M63 salts containing 20 mM methyl-α-d-mannopyranoside and 75% (v/v) Percoll, these clones also displayed the same binding phenotype in M63 salts alone, indicating that these reagents had no effect on the stability of metal binding capacity. The different sizes of the cell-metal aggregates indicated that there were differences in the avidity of the various clones towards each of the metals. Furthermore, enumeration of bound/unbound bacteria clearly indicated the avidity of the selected clones toward specific metal oxides (Table 2).

Figure 2.

Phase contrast microscopy demonstrating adherence of S1918 cells containing plasmids expressing various chimeric fimH genes to (A) CoO, (B) MnO2, (C) PbO2, and (D) Cr2O3. Plasmids used were pLPA30 (wild-type fimH), pKKJ73 (random library clone isolated from selections for adherence to CoO), pKKJ78 (random library clone isolated from selections for adherence to MnO2), pKKJ68 and pKKJ69 (random library clones isolated from selections for adherence to PbO2) and pKKJ62 (random library clones isolated from selections for adherence to Cr2O3). Cells are shown in M63 salts medium alone, or in the same medium containing either CoO, MnO2, PbO2, or Cr2O3. Metal particles are indicated by an arrow for clarity.

Table 2.  Quantification of metal binding by various clones (cf. Section 2)
Metal oxidePlasmidNumber of particlesNumber of binding cellsAverage number of binders pr. particleP values (%)a
  1. aBased on 5% level of significance.


The agglutination titers of these cells were similar to a control strain expressing wild-type FimH, indicating that the presence of the inserts had not influenced the natural binding domain of FimH or significantly altered the number of fimbriae on the surface of the cells.

3.4Biosorption of metal ions by cells expressing FimH designer adhesins

As a separate test to confirm that the enriched sequence could confer the ability on bacterial cells to adhere to metals, we measured the amount of Mn2+ sequestered by cells harboring plasmid pKKJ78 by atomic absorption spectroscopy. Plasmid pKKJ78 contains the HRRS binding motif (Table 2). Cells harboring this plasmid were observed to sequester approximately four-fold higher amounts of Mn2+ than cells containing wild-type FimH (pLPA30) (Fig. 3). Two different motifs were determined from the sequences enriched for binding to PbO2 (Table 1). Plasmid pKKJ68 is particularly interesting since its sequence contains both of these motifs. Direct counting of cells adhering to PbO2 indicated that this sequence imparted a stronger binding avidity towards PbO2 than another sequence with only one motif (pKKJ69) (Table 2; Fig. 2C).

Figure 3.

Atomic absorption spectroscopy determinations of the amount of Mn2+ associated with E. coli S1918(pPKL115) cells containing the plasmids pLPA30 (wild-type fimH) and pKKJ78 (random library clone isolated from selections for adherence to MnO2). Values are the means±S.E.M. (n= 2).


The technology of random peptide display on the surface of bacterial cells provides an attractive complement to phage display. Several different expression systems have been used to display random peptide libraries on the bacterial cell surface. The E. coli LamB protein was used to present synthetic sequences from which peptide segments conferring binding to iron oxide, gold and chromium have been identified [12, 17]. The major structural component of the E. coli flagellum, the FliC protein, was used in the FLITRX display system described by Lu et al. [18]. We have now used the E. coli FimH fimbrial protein to display random peptide libraries and identified sequences conferring specific binding to various metal oxides. In this system we take advantage of the huge number of fimbriae on the bacterial surface to present designer adhesins. The system appears to be highly flexible with regard to the size and type of insert presented. Indeed, insert sizes ranging from one to four 33-mer sequences, with varying types of amino acids, were selected in our enrichment procedures.

A number of peptide sequences conferring the ability to coordinate each of the metal ions were selected from the random library constructed in the FimH fimbrial protein. The amino acids aspartate, cysteine, glutamate, histidine, methionine, serine, threonine, tyrosine, and tryptophan have all previously been observed to participate in metal ligation within proteins. The coordination of metals can also be achieved by main chain carbonyl oxygens and amide nitrogens [19]. Some of these amino acids were enriched to various degrees in our selections. Interestingly, a number of motifs were identified in the sequences enriched for binding to PbO2, CoO, and MnO2 (Table 1). In some cases these sequences were associated with the Arg-Ser linker encoded sequence. The identified motifs were not imperative for metal binding as other varied sequences were also identified which conferred the ability to bind to the metals. However, the preponderance of these motifs could indicate that the number of solutions to provide binding to a given metal is not unlimited. It is apparent that this display system provides for numerous structural solutions to metal binding.

Peptide sequences conferring the ability to coordinate Pb(II) have previously been identified from phage displayed semisynthetic combinatorial antibody libraries [19]. While the specific sequences involved in the coordination of Pb(II) were difficult to discern, the authors noted an over-representation of aspartate and glutamate residues in the enriched sequences. We did not detect a similar enrichment of these amino acids amongst our Pb4+ binding sequences. Brown [17] also previously identified repeating polypeptides able to bind chromium. Analysis of these peptides revealed a consensus sequence for chromium binding of QHQK. Although we did not observe such sequences in our selections, this can be explained by the fact that Brown [17] used chromium powder whereas we used chromium(III) oxide in our binding assays. Clearly there are many factors intrinsic to both the display system used and the selection conditions which can affect the types of metal binding sequences enriched in such procedures. The type of metal, buffers, genetic structure of the random library, display system (e.g. phage, bacterial), presenting protein (e.g. LamB, FimH, pIII), and flanking protein sequences are all factors which vary amongst different display systems.

The H/R-X3-H-R/K-S motif was observed in the sequences enriched for binding to both PbO2 and CoO. This suggests that both of these metals are recognized by similar peptide sequences. When examined further in cross-binding experiments, all of the sequences containing this motif were able to bind to both PbO2 and CoO (data not shown). This binding motif was not enriched amongst the sequences conferring the ability to bind to Cr2O3 and MnO2. These observations provide further evidence that there are specific rules which govern interactions between proteins and metals.

Different sized cell-metal aggregates were observed by phase contrast microscopy between the population of enriched clones. This indicates that the FimH display system enables the selection of peptide sequences with a variety of binding avidities. This phenomenon was further demonstrated by direct counting of cells associated with metal particles. The higher avidity to PbO2 by the insert in pKKJ68 was associated with the presence of both types of motifs in this enriched sequence. The isolation of such multiple insert sequences is only possible with the structure we have chosen for our random library. Further work will examine whether these binding strengths can be enhanced by increased peptide valency as has been observed in other systems [17, 18, 20].

There are obvious advantages inherent in using the FimH display system for the construction of designer adhesins. Using the natural d-mannose binding domain of this protein in combination with the second engineered site one could envisage the construction of a range of interesting heterobinary adhesins. In particular, immobilization of such cells by one adhesive domain could facilitate their use in detection systems for metals or perhaps directly as biosorption agents for the removal of toxic or precious metals from the environment. It is also possible to purify fimbriae by blending if one wishes to use the chimeric proteins without bacteria.


This work was supported by the Danish Technical Research Council (Grants 9400392 and 9601334). We thank Eva Thale and Jens B. Olsen (Department of Chemistry, DTU) for performing atomic absorption measurements, and Stanley Brown (University of Copenhagen) for his help during various phases of this work.