Surface display of PbrR on Escherichia coli and evaluation of the bioavailability of lead associated with engineered cells in mice

Human exposure to lead mainly occurs by ingestion of contaminated food, water and soil. Blocking lead uptake in the gastrointestinal tract is a novel prevention strategy. Whole-cell biosorbent for lead was constructed with PbrR genetically engineered on the cell surface of Escherichia coli (E. coli), a predominant strain among intestinal microflora, using lipoprotein (Lpp)-OmpA as the anchoring protein. In vitro, the PbrR displayed cells had an enhanced ability for immobilizing toxic lead(II) ions from the external media at both acidic and neutral pH, and exhibited a higher specific adsorption for lead compared to other physiological two valence metal ions. In vivo, the persistence of recombinant E. coli in the murine intestinal tract and the integrity of surface displayed PbrR were confirmed. In addition, oral administration of surface-engineered E. coli was safe in mice, in which the concentrations of physiological metal ions in blood were not affected. More importantly, lead associated with PbrR-displayed E. coli was demonstrated to be less bioavailable in the experimental mouse model with exposure to oral lead. This is reflected by significantly lower blood and femur lead concentrations in PbrR-displayed E. coli groups compared to the control. These results open up the possibility for the removal of toxic metal ions in vivo using engineered microorganisms as adsorbents.

Chelation therapy has always been the preferred medical treatment for heavy metal poisoning, but is recommended only for blood lead levels of 45 μg/dL or greater 16 . Existing chelating agents, such as succimer or dimercaprol, have adverse side effects 24 . The safety of chelation therapy for childhood lead poisoning has always been controversial 25,26 . Imbalance of essential elements is very common in the course of chelation treatment, so lead poisoned patients treated with chelating agents need to be monitored closely during and after treatment 27 . Thus, there is a great need for more selective and safer chelating agents for the removal of heavy metal ions 22 .
With the development of molecular display technologies, several heavy metal ion-binding proteins or peptides have become ideal targets for surface display 28,29 . Surface engineered microorganisms as whole-cell adsorbers are considered to be a promising and sustainable approach for the removal of environmental heavy metal pollution [30][31][32] , and has become one of the research focuses in environmental biotechnology. However, there have been few studies about bioadsorption in biomedical research.
Herein, we have engineered PbrR, a well-known lead specific binding protein from Cupriavidus metallidurans CH34 33 , to anchor Lpp-OmpA onto the cell surface of E. coli, a predominant strain of intestinal commensal bacteria. The aim of the present pilot study is to assess the safety and persistence of surface-engineered bacteria in the murine gastrointestinal tract. The secondary aims are to further evaluate the bioavailability of lead associated with surface-engineered cells and to characterize the amount of lead tissue retention in PbrR displayed E. coli-treated mice.

Results
Construction of surface display vectors. The strategy used in this study to display lead(II)-specific binding protein PbrR on the surface of E. coli is shown in Fig. 1. Lpp-OmpA is an efficient display system, and it has been successfully used to target a wide range of proteins onto the surface of E. coli with no obvious adverse effects on cell growth and membrane integrity 34,35 . The cassette encoding surface anchor Lpp-OmpA was first synthesized and inserted into pET-21a to generate pLA for cell surface display of Lpp-OmpA with His tag (Fig. 1A). Expression of anchor Lpp-OmpA-His tag under the control of the T7 promoter/lac operator could be induced by isopropyl β-D-1-thiogalactopyranoside (IPTG) 36 . The cassette encoding PbrR was then synthesized and ligated into the multiple cloning site (MCS) between Lpp-OmpA and His tag encoding sequences in pLA, and the expression vector for the surface display of Lpp-OmpA-PbrR was named pLAP (Fig. 1B). The fusion proteins were tagged with a hexahistidine tag at the C-terminus to facilitate detection with immunological methods. Both of the recombinant vectors were all confirmed by double restriction enzyme digestions, visualization on 2.0% agarose gel (Fig. 1D). An expected band of lpp-ompA (approximately 447 bp) was released from pLA by digestion with restriction enzymes NdeI and SalI (Fig. 1D, Lane 2). An expected band of pbrR (approximately 447 bp) was released from pLAP by digestion with restriction enzymes HindIII and XhoI (Fig. 1D, Lane 4), and the fragment of lpp-ompA-pbrR (about 900 bp) was detected after pLAP was digested with restriction enzymes NdeI and XhoI (Fig. 1D, Lane 5). DNA sequencing analysis was performed to exclude potential mutations introduced by PCR (Fig. S1).
Surface expression of PbrR using a Lpp-OmpA anchor. Denaturing gel electrophoresis of E. coli high-stringence expression host BL21(DE3)pLysS harboring pLA and pLAP after induction, followed by immunoblotting with anti-His antibody, indicated that Lpp-OmpA and Lpp-OmpA-PbrR were all expressed as full-length products. A protein band of about 17.4 kDa corresponding to the theoretical size of Lpp-OmpA was detected in BL21(DE3)pLysS/pLA after induction ( Fig. 2A and B, lane 2). Another protein band of about 33.5 kDa corresponding to the theoretical size of Lpp-OmpA-PbrR was detected in BL21(DE3)pLysS/pLAP after induction ( Fig. 2A and B, lane 3). Total cell proteins were separated by 15% SDS-PAGE (A). Western blot of recombinant proteins probed by anti-His antibody (B). Lane 1 shows BL21(DE3)pLysS/pLA without induction; Lane 2 shows BL21(DE3)pLysS/ pLA after induction (≈17.4 kDa); Lane 3 shows BL21(DE3)pLysS/pLAP after induction (≈33.5 kDa); Lane M is a protein size marker. Phase contrast micrographs (C) and immunofluorescence images (D) of recombinant E. coli cells. Immunofluorescence labeling of surface-engineered E. coli cells using mouse anti-His tag antibody (primary antibody) and FITC labeled donkey anti-mouse IgG antibody (secondary antibody). White arrows indicate individual cells that have been immunofluorescence labeled. Immunofluorescence was performed using a fluorescence microscope (×1000 magnification). The surface localization of PbrR was confirmed by immunofluorescence labeling. A hexahistidine tag was added to the C terminus of Lpp-OmpA (pLA) and Lpp-OmpA-PbrR (pLAP) to facilitate detection with an anti-His antibody. Immunofluorescence labeling of E. coli cells was performed by mouse anti-His IgG as primary antibody, followed by fluorescein isothiocyanate (FITC) labeled donkey anti-mouse IgG as secondary antibody. Signals from FITC (green fluorescence) under the fluorescence microscope with 1000× magnification were observed, and the results are shown in Fig. 2C and D. More than 90% induced E. coli BL21(DE3)pLysS cells harboring pLA or pLAP showed green fluorescence on their surface. In contrast, no fluorescence was observed on uninduced E. coli BL21(DE3)pLysS/pLA cells. IPTG induced Lpp-OmpA or Lpp-OmpA-PbrR displayed on E. coli cell surface was demonstrated.
Furthermore, to evaluate the accessible display of proteins quantitatively, fluorometric assay was also performed after immunofluorescence labeling. The result is shown in Fig. 2E. Lpp-OmpA displayed cells emitted more green fluorescence than Lpp-OmpA-PbrR displayed cells. No extra encoding sequence was inserted into pLA, so the expression and display of a shorter peptide is an energy-saving approach for E. coli cells. This is consistent with previous studies 37,38 . In vitro lead(II) adsorption ability of surface engineered cells. To investigate the capacity of the surface-displayed PbrR to bind lead(II), uninduced or induced BL21(DE3)pLysS/pLA and BL21(DE3)pLysS/ pLAP were resuspended in 25 mmol/L HEPES pH 7.0 containing 150 μmol/L lead(II), and the amount of bound lead(II) was quantified by atomic absorption spectroscopy after 12 h. The results are shown in Fig. 3A. E. coli surface-displayed Lpp-OmpA-PbrR chimera was able to adsorb lead(II) with a capacity of about 108 μmol/g dry cells, which is 4-fold higher than E. coli surface-displayed Lpp-OmpA (27.7 μmol/g dry cells). In addition, induced BL21(DE3)pLysS/pLA and BL21(DE3)pLysS/pLAP were incubated with 25 mmol/L HEPES pH 7.0 containing 50 or 300 μmol/L lead(II) for just 1 h. E. coli surface-displayed Lpp-OmpA-PbrR chimera could adsorb 56 μmol/g dry cells at 50 μmol/L lead(II), and increased to 144 μmol/g dry cells at 300 μmol/L lead(II) (Fig. 3B). Lead adsorption capacity would not increase when the concentration of lead was further increased (data not shown). After the adsorption equilibrium was achieved, the cells above were named unsaturatedly leaded PbrR-displayed E. coli and saturatedly leaded PbrR-displayed E. coli, respectively.
Oral administration of surface-engineered E. coli in the short term exerts no adverse effect on mice. In order to determine the effect of oral administration of PbrR-displayed E. coli on mice, we subsequently performed a short-term in vivo toxicity assays using recombinant bacteria 39,40 . No morbidities or mortalities were observed during the experiment as shown in Table 1. Compared with the control group, sustained 15 d oral administration of PbrR-displayed E. coli at a daily dose of 1 × 10 10 colony forming unit (CFU) recombinant bacteria exerted no obvious effects on water consumption, on food consumption, on growth rate, and on liver and kidney indices. Furthermore, there was no recombinant E. coli detected in blood of PbrR-displayed E. coli treated mice. PbrR-displayed E. coli did not show any intestinal epithelial invasion nor translocation capability.

Orally administered surface-engineered E. coli persists in the murine gastrointestinal tract.
The incubation of recombinant bacteria in simulated gastric juice (SGJ) mimicking low pH and highly digestive conditions in the stomach 41 significantly decreased bacterial survival (Fig. 4A). Half of the added bacteria were alive after a 2 h incubation, and 16% of the bacteria were still alive at 6 h. Using ampicillin resistant (from vector pLA or pLAP) and chloramphenicol resistant (from host E. coli BL21(DE3)pLysS) properties of recombinant bacteria, the concentrations of these bacteria in stool samples from all three lead-treated groups were determined. There is no significant difference between the three lead-treated groups after 15 d continuous oral administration of recombinant bacteria (Fig. 4B). Furthermore, the integrity of the recombinant protein in stool samples was also tested. A modified Western blot assay based on His tag to detect the full length Lpp-OmpA or Lpp-OmpA-PbrR was established in this study. Dot immunobinding of stool samples with anti-His IgG was performed, and positive results (clear immunospots) were observed in stool samples from all three lead-treated groups with oral administration of recombinant E. coli (Fig. 4C).

In vivo bioavailability of lead associated with surface-displayed PbrR.
To investigate the bioavailability of lead associated with surface engineered E. coli, the lead ion was supplied as free form (group 2), as saturatedly leaded PbrR-displayed E. coli (group 3), and as unsaturatedly leaded PbrR-displayed E. coli (group 4). The final concentrations of lead in semiliquid diet were 1.8 mg lead/kg in all three lead-treated groups. The amount of the semiliquid diet was limited to 100 g/cage.d, so the lead dose was nearly 20 μg lead/mouse.d. Administration of lead took place over 15 days. There was no difference in mouse body weight between the control group and the three lead-treated groups (data not shown).
Lead deposition in the body consists of three major pools: blood, bone, and soft tissues 42 . The primary site of lead storage in vivo is bone, and the bone pool (storage pool) may be mobilized and contribute to the blood lead level (exchangeable pool) 42 . Variation in lead storage and mobilization from the storage pool depends on several factors, such as dose/rate of lead exposure, age, pregnancy, and species 43,44 . Lead contents in both blood and bone in all four groups were determined, with the results shown in Fig. 5. As expected, lead treatment in group 2 led to a considerable augmentation in whole blood (>6 fold) and femur (>16 fold) levels of lead, compared to group 1. Interestingly, lead administration in the form of leaded PbrR-displayed E. coli all led to a significant decrease in both blood and bone lead concentration, compared to group 2. Especially in group 4, in the form of unsaturatedly leaded surface engineered bacteria, lead tissue levels were the lowest among all three lead-treated groups and similar to the control group with no lead exposure.
Metabolization of physiological metal ions in vivo is not affected by oral administration of surface-engineered E. coli. Major  blood were measured in all four groups by atomic absorption spectroscopy. The results are shown in Table 2. Due to the highly selective adsorption to lead by PbrR-displayed E. coli, no obvious difference in five physiological metal ions contents was observed between the control group (group 1) and all three recombinant bacteria treated  groups (group 2~4). This is in agreement with the result of the adsorption selectivity of PbrR-displayed E. coli in vitro as described above.

Discussion
Since the introduction of lead free gasoline worldwide over the last three decades, ingestion has become the primary route of lead exposure, especially for children, due to increased hand-to-mouth activity and enhanced absorption of lead from the digestive tract 16 . Many efforts have recently been made to reduce lead levels in soil, water, and food by focusing on environmental remedies [45][46][47] . Bioremediation is a waste management technique that involves the use of naturally occurring or genetically modified organisms to detoxify toxic metals by enhancing immobilization and reducing bioavailability of heavy metals 47,48 . The cell surface-display strategy for heavy metal adsorption alleviates the burden of intracellular accumulation, which effectively overcomes the shortcoming of the traditional bioadsorption methods 49 . Heavy metal toxicity in soil or water could be remedied by immobilization of the heavy metal onto the surface of living engineered cells in a non-bioactive form, which was reflected by a positive effect on plant growth and negative effect on toxic metals accumulation in plants 35 . However, there is still no evidence on whether or not surface engineered microorganisms can counteract the toxic effect of heavy metal ions in animals. In this study, a mouse bioassay based on lead oral exposure was performed, in which the bioavailability of lead ion immobilized onto the surface of lead-specific binding protein displayed E. coli was tested.
The pbr resistance operon origin from Cupriavidus metallidurans CH34 confers resistance to lead(II) salts, and is regulated by the lead(II) responsive regulator PbrR, which belongs to the MerR family of metal ion-sensing regulatory proteins 33 . The pbr operon is the unique lead-specific resistance operon in any bacterial genus so far, and PbrR is the first protein that has been demonstrated to be able to sense lead ions in nature because of its lead-specific binding property [50][51][52][53] . By engineering PbrR with the Lpp-OmpA system, we successfully developed whole-cell biosorbent for the selective removal of lead. The production of PbrR on induced E. coli BL21(DE3) pLysS/pLAP enabled nearly four-fold higher lead(II) biosorption than that found in induced E. coli BL21(DE3) pLysS/pLA cells (Fig. 3A). Uptake of lead in the digestive tract, from high to low, is via the duodenum, ileum, colon and stomach 16,54 . The binding of metalloregulatory proteins with their recognized metal ions is usually not sensitive to pH value and preferably in slightly acidic pH 32,37,55 . Although stomach is not the main lead uptake site, the extreme acidic environment of gastric juice poses a great challenge to bacterial survival and lead adsorption by PbrR-displayed E. coli. Compared to the amount of lead adsorption by PbrR-displayed E. coli at pH 7.0, the amount of lead adsorption was decreased by half at pH 3.0, but it was still 6.2-fold higher than undisplayed E. coli (Fig. 3C).
The bioavailability of lead(II) bound to the PbrR-displayed E. coli might primarily depend on the ease of metal release in the gastrointestinal tract. As expected, oral administration of lead(II) immobilized by the PbrR-displayed E. coli in mice resulted in lower accumulation of lead in both exchangeable pool (blood) and storage pool (femurs) (Fig. 5). Furthermore, unsaturatedly leaded PbrR-displayed E. coli seemed to provide more free sites for capturing free lead(II) derived from occasional dissociation from the cell surface or from the lysis of engineered cells. It was reflected that tissue lead levels in unsaturatedly leaded PbrR-displayed E. coli treated group were similar to the control (Fig. 5). More importantly, due to the highly selective lead-binding property of PbrR, there was no difference in the levels of essential mineral cations in the blood between lead-treated groups and the control ( Table 2). Lead exposure is known to affect multiple critical organ functions including the nervous, hematological, renal, and cardiac functions 22 . Although lead associated with PbrR-displayed E. coli was not demonstrated to be orally bioavailable in the current study, its potential detoxifying effect on multiple target organs is worth further investigation in future studies.
Inspired by traditional chelation therapy, researchers have never stopped studying novel oral chelating agents. For instance, low esterified pectin and rhamnogalacturonan-II dimer, molecules rich in free carboxyl groups, have been demonstrated to be ideal specific lead chelators 17,21,56 . They can form strong complexes with lead(II) in vitro and lead(II) complexed with them was less bioavailable in vivo than lead acetate 56 . Although these kinds of pectic polysaccharides are fermentable and undergo complete fermentation by the gut microflora to liberate the complexed lead 57 , lead absorption occurs primarily in the small intestinal tract and the ileal uptake of lead is very low 58 . Thus, most investigations in animal models have shown a beneficial effect of extra pectic polysaccharides intake on lead detoxification 21,56 .
In order to overcome the shortcomings mentioned above, indigestible biomaterials or metal binding ligands coupled to non-degradable skeletons might be an ideal alternative to pectic polysaccharides. The bacterial cell-surface display strategy for lead adsorption meets the above properties, and has overwhelming advantages over any reported biomaterials, including pectic derivatives, for the following reasons:   First, there are so many microorganisms suitable for host cell-surface display, including some commensal bacteria in the intestines of mammals 28 . E. coli, a predominant strain among intestinal microflora 59 , was used as the host for surface display of PbrR in this study. The safety and risk assessment of surface-engineered E. coli was performed in experimental mouse models. Oral administration of surface-engineered E. coli was demonstrated to have no adverse effect on water and food consumption and growth rate in mice over 15 days and surface-engineered E. coli had no intestinal epithelial invasion capability in vivo ( Table 1). The significant decrease of live bacteria during passage of the upper digestive tract is an important drawback for probiotic preparations 60 . The data from our experiment showed that about half of PbrR-displayed E. coli was killed in SGJ in the first 2 h (Fig. 4A). Since almost no extra lead(II) binding sites exited on the surface of saturatedly leaded PbrR-displayed E. coli, the degradation of E. coli in stomach could result in elevated tissue lead content in group 3 (Fig. 5). However, further studies in vivo should be done, because the food matrix usually confers extra protection on bacterial survival in the stomach 61 .
Second, abundant colonization in the gastrointestinal tract is one of the most important characteristics of commensal bacteria. Although there was a difference in the oral administration dose of recombinant bacteria, there was no significant difference in recombinant bacterial densities in all three lead-treated groups (Fig. 4B). The stability of the gut microbiota is important for the maintenance of intestinal microflora balance 62 . Besides the survival and persistence of recombinant E. coli, the integrity of displayed PbrR is another important factor for the immobilization of lead(II) on the surface of recombinant E. coli. However, PbrR displayed on the outer membrane of E. coli might easily be hydrolyzed by digestive enzymes in the gastrointestinal tract. An immunoblotting method was established in the study to detect His tag at the C terminus of Lpp-OmpA or Lpp-OmpA-PbrR. The integrity of target protein displayed on the surface was confirmed, but there was a significant difference in the color depth of brown sediments in all three lead-treated groups with oral administration of surface engineered E. coli (Fig. 4C), and it did not seem to be consistent with similar bacterial densities among three groups (Fig. 4B). Inducible surface display vector derived from pET-21a was used in this study. Although it is suitable for the strict regulation of protein expression in in vitro studies, it is impossible to realize the surface display of PbrR during recombinant bacteria survival, growth, and replication in murine intestines. Thus, it might be a good explanation for the results of the in vivo study (Fig. 4). To address uncertainties derived from PbrR expression in in vivo studies, it would be beneficial to realize the constitutive expression and surface display of PbrR during bacterial colonization.
Finally, the survival and persistence of commensal microflora in the gastrointestinal tract is a unique characteristic that other biomaterials do not possess 63 . More importantly, commensal bacteria always adhere to intestinal epithelium 59 , which is the main uptake site for nutrients including heavy metal ions. Our preliminary results suggested that surface displayed PbrR in commensal bacteria might help block lead uptake continuously. In addition to significantly decreasing blood and tissue lead levels, further studies on lead accumulation, neurotoxicity, and long-term effects are being conducted in our laboratory to determine the in vivo feasibility of using biosorption technology to prevent host lead poisoning.
In conclusion, our findings show that lead(II) associated with surface engineered E. coli is essentially unavailable for intestinal absorption. Engineering the specific toxic metal binding protein on the surface of commensal bacteria provides an alternative method for the prevention of heavy metal poisoning. Growing microbial biomass with the potential for blocking heavy metal uptake in vivo needs further investigation. Due to specific and effective binding with lead, surface engineered bacteria, as a kind of probiotic, may be used to decrease lead absorption, prevent lead accumulation, and further ameliorate lead toxicities. However, a series of additional studies in animals are still required before developing surface engineered probiotics as a preventive or perhaps curative agent in lead exposure and toxicity in humans.

Materials and Methods
Strains and agents. The strains and vectors used in this study are shown in Table S1. E. coli Top10 stored in our laboratory was used for all cloning steps. For recombinant proteins expression, pET-21a vector (Novagen, Madison, WI) was used with expression host E. coli BL21(DE3)pLysS (Novagen, Madison, WI). Restriction enzymes, Pyrobest DNA Ploymerase, dNTP, DNA Marker and gel extraction kits were obtained from TaKaRa (Dalian, China). The mouse anti-His monoclonal antibody and enhanced horseradish peroxidase (HRP)-Diaminobenzidine (DAB) chromogenic substrate kit were obtained from Tiangen (Beijing, China). FITC labeled donkey anti-mouse IgG, HRP labeled rabbit anti-mouse IgG, lead acetate, IPTG, and antibiotics were all purchased from Sangon Biotech (Shanghai, China). Pepsin (1:3000) was obtained from Amresco (OH, USA). Tryptone and yeast extract were obtained from Oxoid (Basingstoke, UK). E. coli was cultured in Luria-Bertani (LB) broth (1% tryptone, 0.5% yeast extract and 1% NaCl) supplemented with antimicrobial agents, as necessary.
Construction of recombinant plasmids. DNA encoding surface anchor Lpp-OmpA (LOA), which contains a lipoprotein signal peptide and the first nine N-terminal amino acids of lipoprotein (NCBI Accession No. BAA16044) attached to five outer membrane-spanning domains of OmpA (NCBI Accession No. ABJ00366), was synthesized by ABI 3900 synthesizer in Sangon Biotech Co., Ltd (Shanghai, China). The synthesized fragment (435 bp) was purified by HPLC and directly cloned into pUCm-T using T/A cloning kit (Sangon, Shanghai, China) to generate pT-loa. The fragment encoding Lpp-OmpA (447 bp) was amplified from pT-loa with primers Loa-F and Loa-R (Table S2), digested with NdeI and SalI and ligated into similarly digested pET-21a to yield pLA. The gene of Cupriavidus metallidurans CH34 PbrR (NCBI Accession No. ABF12805) was synthesized according to the codon usage bias of E. coli, and subcloned into pUCm-T to generate pT-pbrR. The fragment encoding PbrR (447 bp) was amplified from pT-pbrR with primers pbrR-F and pbrR-R (Table S2), digested with HindIII and XhoI and ligated into similarly digested pLA to produce pLAP. All recombinant plasmids were confirmed by double SCienTifiC RePoRTS | (2018) 8:5685 | DOI:10.1038/s41598-018-24134-3 restriction enzymes digestion, and finally DNA sequencing was performed using T7 terminator primer (Sangon, Shanghai, China) to confirm accordance with the design.
Protein expression and analysis. BL21(DE3)pLysS was transformed with pLA or pLAP using CaCl 2 -mediated transformation. To express recombinant fusion protein, overnight cultures of BL21(DE3)pLysS carrying pLA or pLAP were inoculated at 1:100 and grown in LB broth at 37 °C to an optical density at 600 nm of ~0.5, then induced with 0.5 mmol/L IPTG at 25 °C for 12 h. Cells were harvested and resuspened in loading buffer. After boiled for 10 min, samples were run on a 15% SDS-PAGE gel and stained with Coomassie blue dye.
The presence of Lpp-OmpA and Lpp-OmpA-PbrR were confirmed by Western blot using mouse anti-His antibody at a 1:2000 dilution in 20 mmol/L phosphate buffer, pH 7.5 with 0.05% Tween 20 (PBST), and second HRP labeled rabbit anti-mouse IgG, at a final 1:2000 dilution in PBST. HRP signal was detected with enhanced HRP-DAB chromogenic substrate kit.
Immunofluorescence detection of surface-engineered E. coli cells. To confirm the presentation of recombinant proteins on the E. coli outer surface, immunofluorescence labeling using anti His-tag IgG was performed as described below. A 500 μL of uninduced/induced culture was centrifuged at 3500 rpm for 5 min and fixed with 4% formaldehyde in PBS (pH 7.4) for 1.5 h, then blocked in PBS containing 1% fetal calf serum for 30 min prior to immunostaining. The cells were incubated with mouse anti-His antibody diluted 1:200 in PBS overnight at 4 °C. After washing three times with PBS, the cells were then incubated with FITC-conjugated donkey anti-mouse IgG antibody diluted at 1:300 in PBS for 1.5 h at room temperature. After washing three times with PBS, the cells were mounted on slides. Finally, FITC immunofluorescence was visualized using a Nikon Eclipse Ni fluorescence microscope coupled to a Nikon DS-Ri2 digital camera (Tokyo, Japan). The same setting for imaging was applied for all the samples, and representative images were acquired. To evaluate the display efficiency, about 500 cells were counted for each strain, and the percentage of fluorescent cells was defined as (number of fluorescent cells/total cells) ×100%. In vivo toxicity assay of PbrR-displayed E. coli. In vivo toxicity assay of recombinant bacteria was done as described previously 39,40 . Male Kunming (KM) mice, 4~5 weeks old and weighing 18~22 g, obtained from Guangdong Medical Laboratory Animal Center, were used for in vivo toxicity assay. All animal experiments were performed in accordance with the guidelines of China Council on Animal Care and Use, and experimental protocols were approved by the Ethics Committee at Shenzhen Prevention and Treatment Center for Occupational Disease. After the adaptation period, twenty KM mice were randomly divided into a control group and PbrR-displayed E. coli treated group (n = 10 per group). Induced cultures of BL21(DE3)pLysS/pLAP were freshly prepared daily, and the PbrR-displayed E. coli treated group was orally given at a dose of 1 × 10 10 CFU/d. The control group was orally given the same volume of saline daily. Control and PbrR-displayed E. coli treated group were provided with purified water and standard feed ad libitum. During this study, both feed and water consumptions were recorded daily, and body weight increasing rate was calculated. The experiment was performed for 15 d. All mice were killed under anesthesia. Inner organs were removed and weighed. Blood samples in the PbrR-displayed E. coli treated group were collected, and spread on LB agar with ampicillin (50 μg/mL) and chloramphenicol (34 μg/mL) to detect the presence of recombinant bacteria in blood.
Simulated gastric juice tolerance of recombinant bacteria. SGJ was prepared using 0.08 mol/L HCl and 0.2% NaCl. The pH value was adjusted to 2.5 using a calibrated pH meter. The SGJ solution was sterilized at 121 °C for 20 min. Pepsin (1:3000) was freshly added at 3 U/mL 41 . Induced BL21(DE3)pLysS/pLAP was incubated in SGJ at 37 °C for 0, 2, 4, 6 h. A 0.2 mL aliquot of the sample was removed at each time point, then the standard serial dilution method of viable bacteria enumeration was done 64 . Bacterial survival rate was calculated as a percentage of the total added bacteria, which was defined as 100%. Experiments were repeated at least three times with similar results.
Animal experimental design. Male KM mice, 4~5 weeks old and weighing 18~22 g, were used in this study. After the adaptation period, thirty-six KM mice were randomly divided into four groups (n = 9 per group) (Fig. S2). These groups consisted of a control group (group 1) and three experimental groups that received lead as different form (group 2, 3 and 4). Powdered diets (50 g/group/d) were mixed daily with 50 mL of distilled water (group 1), 50 mL of suspension containing 3.6 mg/L lead as lead acetate + Lpp-OmpA displayed E. coli (group 2), 50 mL of suspension containing 3.6 mg/L lead as saturatedly leaded PbrR-displayed E. coli (group 3), and 50 mL of suspension containing 3.6 mg/L lead as unsaturatedly leaded PbrR-displayed E. coli (group 4) to form a semi-liquid mixture that was prepared on site. After 15 days on control or experimental diets, all mice were killed under anesthesia. Subsequently, blood samples (500 μL) were collected into heparinized tubes via the orbital venous plexus, and the left femur was then dissected, frozen at −20 °C until analysis. The stool samples in all groups were obtained from the large intestine, and recombinant bacterial density and recombinant protein immunoblotting analyses were done as described below.
Assay of recombinant bacterial density and recombinant protein in mice large intestinal contents. A 0.25 g aliquot of fresh stool was homogenized in 0.9 mL aseptic normal saline. Then, one 0.1 mL aliquot was serially diluted and spread on LB agar with ampicillin (50 μg/mL) and chloramphenicol (34 μg/mL) to calculate the recombinant bacterial density in stool sample, and the result was indicated as the number of CFU per gram of gut contents.
The above-mentioned suspension was centrifuged (1000 rpm for 3 min). The supernatant was transferred into another tube and further centrifuged (3500 rpm for 5 min). The supernatant was discarded, and then the pellet was thoroughly washed twice with normal saline (3500 rpm for 5 min). Finally, the pellet was suspended in 40 μL lysis buffer (200 mmol/L Tris-HCl, pH 6.8, 200 mmol/L DTT, 2% SDS), and boiled for 10 min. An aliquot of 5 μL of each sample was dripped on PVDF membrane which was blocked, washed, and incubated with mouse anti-His IgG. After extensive washing, the membrane was incubated with a HRP labeled rabbit anti-mouse IgG, and bound antibodies were visualized using an enhanced HRP-DAB chromogenic substrate kit as mentioned above.

Determination of lead and physiological metal ions content in biological samples. The metal
ion content in blood and femur was determined by the method described previously 56 . Whole blood and femur were dried (80 °C, overnight) and powdered. The powder was dissolved in 0.5 mL of 14 mol/L HNO 3 and 0.2 mL of 10 mol/L H 2 O 2 and heated at 140 °C for 2.5 h. An appropriated dilution with purified water was performed in each sample, prior to determination of the concentrations of various metal ions by atomic absorption spectroscopy.