FimH-based display of functional eukaryotic proteins on bacteria surfaces

The demand for recombinant proteins for analytic and therapeutic purposes is increasing; however, most currently used bacterial production systems accumulate the recombinant proteins in the intracellular space, which requires denaturating procedures for harvesting and functional testing. We here present a novel FimH-based expression system that enables display of fully functional eukaryotic proteins while preventing technical difficulties in translocating, folding, stabilizing and isolating the displayed proteins. As examples, Gaussia Luciferase (GLuc), epidermal growth factor (EGF), transforming growth factor-α (TGF-α) and epiregulin (EPRG) were expressed as FimH fusion proteins on the surface of E. coli bacteria. The fusion proteins were functionally active and could be released from the bacterial surface by specific proteolytic cleavage into the culture supernatant allowing harvesting of the produced proteins. EGFR ligands, produced as FimH fusion proteins and released by proteolytic cleavage, bound to the EGF receptor (EGFR) on cancer cells inducing EGFR phosphorylation. In another application of the technology, GLuc-FimH expressed on the surface of bacteria was used to track tumor-infiltrating bacteria by bioluminescence imaging upon application to mice, thereby visualizing the colonization of transplanted tumors. The examples indicate that the FimH-fusion protein technology can be used in various applications that require functionally active proteins to be displayed on bacterial surfaces or released into the culture supernatant.

(2019) 9:8410 | https://doi.org/10.1038/s41598-019-44883-z www.nature.com/scientificreports www.nature.com/scientificreports/ incorrect folding when anchoring in the outer membrane 14,15 . In Gram-negative bacteria the outer membrane generally acts as a barrier to restrict the protein export from the cell interior; only pilins, flagellins, specific surface enzymes, and a few bacterial toxins are transported across the outer membrane 16 . These natural display systems have the benefit of being optimized for transporting and folding protein units to build polymeric structures on the extracellular surface making the display system attractive for biotechnological applications. We here used the fimbriae protein FimH, the mannose-specific adhesin of the E. coli type-1 fimbriae, for the extracellular display of recombinant proteins. Type-1 fimbriae are composed of up to 3,000 copies of the subunit FimA, that form the pilus rod, as well as the subunits FimF, FimG and FimH building the distal tip fibrillum 17,18 . In initial studies, Pallesen and colleagues used the positions 225 and 258 within the FimH adhesin to display the preS2 domain of the hepatitis B surface antigen or an epitope from cholera toxin 19 . Both positions within the FimH protein proved to be suitable for the integration of peptides of up to 56 amino acids which could be produced, displayed on the cell surface and partially conserved the adhesive function of FimH 19 . Longer peptide or full length proteins displayed by FimH in that position were so far not reported. While short polypeptides used for vaccines could be displayed, the technique failed in functionally expressing large proteins like enzymes or cytokines. Here we identified alternative positions within the FimH protein to display larger proteins in a functionally active fashion. Based on the 3D modelling of E. coli type-1 pili 20 we identified the N-terminus of the FimH domain on the fimbriae tip as a suitable integration site of a larger protein. As examples, we genetically linked Gaussia luciferase (GLuc) and human epidermal growth factor (EGF), tumor growth factor-a (TGF-α) and epiregulin (EREG), all ligands of the epidermal growth factor receptor (EGFR), to FimH. Expressed by transformed E. coli, the proteins conserved their functional capacities. In particular, GLuc-FimH displaying E. coli bacteria were tracked during colonization of syngeneic tumors in an immunocompetent mouse model of pancreatic cancer during a six week period without losing GLuc activity. Bacteria with surface displayed proteins can be used for screening purposes and, furthermore, can be released in a functionally active form by specific proteolytic cleavage making the strategy attractive for protein production without the need to disrupt the bacteria by harsh procedures.

Results
Display of recombinant proteins on the surface of gram-negative bacteria. To enable translocation to the bacterial surface, proteins of interest (POI) were genetically fused to the N-terminus of the E. coli FimH adhesin. The FimH-fused POI cDNA was further linked to the FimH leader peptide by PCR, cloned into the pET-11d vector and transformed into E. coli strain BL21. Expression of the POI was induced by adding isopropyl-β-D-thiogalactopyranosid (IPTG), resulting in the display of the POI on the distal tip of the type-1 fibrillum (Fig. 1).
Likewise, Gaussia luciferase (GLuc), epidermal growth factor (EGF), tumor growth factor-α (TGF-α) and epiregulin (EREG) were expressed as fusion proteins of FimH ( Fig. 2A,B). Upon IPTG treatment the recombinant proteins were detected by flow cytometry using antibodies specific for GLuc or the c-myc tag incorporated into the POI-FimH fusion proteins ( Fig. 2A) www.nature.com/scientificreports www.nature.com/scientificreports/ To visualize the recombinant GLuc-FimH as part of the fimbriae, the transformed E. coli bacteria were induced for expression by IPTG, fixed and subjected to immunogold staining with a rabbit anti-GLuc antibody and a colloidal gold-labeled anti-rabbit IgG antibody. As exemplarily shown in Fig. 3 (further examples in Supplementary Fig. S1), GLuc-FimH protein was specifically detected on the bacterial surface next to the outer membrane whereas control stainings with an isotype matched antibody of irrelevant specificity or staining of bacteria without IPGT induction did not detect GLuc protein.
GLuc-FimH modified E. coli bacteria catalyze the conversion of the sGLuc substrate D-Luc. To assay the functional capacity of the GLuc-FimH fusion protein, engineered E. coli bacteria were cultivated in the presence of IPTG and incubated with the GLuc substrate benzyl-coelenterazine for bioluminescence imaging (Fig. 4A). Transformed bacteria converted the GLuc substrate and emitted photo-detectable luminescence indicating enzymatic catalysis by the surface-displayed FimH-fused Gaussia luciferase. In contrast, neither GLuc-FimH modified bacteria without IPTG induction nor wild-type E. coli bacteria converted the GLuc substrate.
Bacteria preferentially accumulate and amplify at sites of immune repression including larger solid tumor lesions 21 . We here used GLuc-FimH bacteria to track their accumulation within established pancreatic carcinomas in immune competent mice. Therefore, cells of pancreatic cancer cell line Panc02 modified with click beetle Luciferase (CBLuc) for bioluminescence imaging were transplanted into the pancreas of syngeneic  www.nature.com/scientificreports www.nature.com/scientificreports/ immunocompetent C57BL/6 mice. Transplanted carcinoma cells gave rise to a pancreatic adenocarcinoma that progressively increased in size (Fig. 4B). When tumors were established, i.e., reached a size of about 75 mm 3 , GLuc-FimH E. coli bacteria or control non-modified E. coli bacteria were injected into the tail vein of tumor bearing mice. GLuc-mediated bioluminescence was recorded on day 2 and day 10 after bacteria application using the GLuc specific substrate. GLuc-mediated bioluminescent signals were detected particularly at the tumor site of GLuc-FimH bacteria inoculated mice. GLuc-bioluminescence was also detected in regional lymph nodes which are likely due to tumor-associated macrophages transporting phagocytosed bacteria into regional lymph nodes (Fig. 4C). Bioluminescence data were confirmed by immunohistochemical stainings of the tumor tissues; GLuc-FimH engineered bacteria were detected in tumor tissues of treated mice, but not in untreated mice, using E. coli-and GLuc-specific antibodies (Fig. 4D). As expected, wild-type E. coli bacteria also colonized the tumor lesions as recognized by the E. coli specific antibody but not by the GLuc-specific antibody. No E. coli bacteria were found in pancreatic tumors of control mice without prior bacteria inoculation.

Recombinant human EGF receptor ligands produced by transformed bacteria mediate EGFR phosphorylation.
To allow harvest of FimH-fused proteins we modified the POI-FimH fusion protein by inserting a cleavage site for the human rhinovirus 3C protease (HRV3C) between the C-terminus of the POI and the N-terminus of FimH (Fig. 2B). E. coli were transformed with the respective plasmids encoding the EGFR coli bacteria in Panc02 tumors was analyzed by bioluminescence recording at day 0 and day 10 after bacteria inoculation. Spots indicate the accumulation of GLuc bacteria at the tumor site and at regional lymph nodes whereas the body was cleared from bacteria. (C) Pancreatic adenocarcinoma from mice after inoculation with Gluc-FimH engineered or wild-type E. coli were investigated for the presence of E. coli antigens and GLuc by immune histology staining using anti-E. coli and anti-GLuc antibodies. Pancreatic tumors derived from mice without bacteria inoculation served as controls.
www.nature.com/scientificreports www.nature.com/scientificreports/ ligands EGF, EPRG and TGF-α linked to FimH, and induced for protein expression by IPTG. The proteins were detached from the bacteria surface by HRV3C protease digestion. To harvest the cleaved EGFR ligands, protease-treated bacterial culture supernatant was concentrated by ultrafiltration. The produced EGFR ligands in the processed supernatants were determined by ELISA (Fig. 5A). To test for binding to the human EGFR receptor, ELISA plates were coated with produced ligands and incubated with recombinant human EGFR-IgG fusion protein that was dectected by biotin-labeled human IgG-specific antibody. All three released EGFR ligands bound specifically to the human EGFR receptor (Fig. 5B).
To test for the induction of EGFR signaling, 293T cells were incubated with supernatants containing ligands EGF, TGF-α or epiregulin (1 µg/ml) and phosphorylation of the EGF receptor was detected via flow cytometry by intracellular staining with an anti-pEGFR (Tyr1173) antibody. As summarized in Fig. 5C, the released recombinant proteins triggered EGFR Tyr1173 phosphorylation indicating EGFR signaling (Fig. 5C). EGF induced EGFR www.nature.com/scientificreports www.nature.com/scientificreports/ phosphorylation with a faster kinetic than the ligands epiregulin and TGF-α. Taken together, data demonstrate the functional biological activity of all three bacteria-produced EGFR ligands.

Discussion
In this study, we describe a bacterial expression system using the bacterial fimbriae component FimH as a vehicle for bacterial surface display of various heterologous proteins while retaining their functional capacities. The FimH-mediated protein display is based on anchoring the protein-of-interest in the distal tip of the fibrillum, thereby avoiding steric hindrances. The display of large proteins on the bacteria surface is likely facilitated by specialized periplasmic assembly factors and chaperones that bind individual pilus subunits in the periplasm and deliver them safely to an assembly platform in the outer membrane, to which the pilus is anchored 22 . The FimH-assisted display concept takes advantage of the chaperone mediated transfer through the membrane and display of proteins of sizes up to 200 amino acids. The displayed proteins on the fimbriae are functional as exemplarily shown by FimH-fused Gaussia luciferase (GLuc) that converted the cognate substrate for bioluminescence imaging after application in mice.
Our strategy differs in various means from previous approaches. In 1995, Pallesen and colleagues used two interior positions within the adhesin unit of FimH for surface display of heterologous sequences 19 . A serious limitation of the described strategy is the small capacity of up to 56 amino acids that can be successfully inserted into this position 19 . In contrast, we inserted proteins of nearly 200 amino acids on the tip of the fimbriae without loss of function thereby overcoming the size limitation as reported by Klemm and Schembri 23 . In addition, displaying fully functional eukaryotic proteins on bacterial surfaces make new applications possible. This includes the production of proteins with enzymatic activity as shown by the extracellular expression of GLuc. Also functional ligand -receptor interactions can be initiated by the produced recombinant EGFR ligands EGF, TGF-α and EPRG. In fact, considering the current needs and possibilities in producing recombinant proteins in bacteria for various biotechnological applications the presented technological solution helps to overcome major obstacles as they currently occur with respect to efficient and stable display on bacterial surfaces. This includes optimal anchoring of the FimH carrier protein in the bacterial outer membrane. In previous studies, various anchoring motifs were identified, including OmpC, OmpX and BclA for surface display of recombinant proteins [24][25][26] . Other bacterial proteins used for anchoring, such as outer membrane proteins LamB (maltoporin), OmpA (outer membrane protein A) and PhoE (phosphateinducible porin), were found unsuitable for the display of larger recombinant proteins; however, the Lpp-OmpA display system as an exception allows display of larger proteins on the surface of E. coli bacteria 11 , e.g., the alkaline phosphatase (PhoA) protein 15 . However, the use of current anchoring motifs did not always allow efficient display of every recombinant protein on the bacterial surface. Despite the achieved progress, further developments are required to provide a reliable biotechnological solution for the realization of individual requests. One of the major obstacles is the outer membrane itself. In order to be displayed on the surface, the recombinant proteins produced in the cytoplasm must be translocated through the inner membrane, cross the periplasm, be inserted into the outer membrane and there correctly folded and displayed 27 .
The novelty of the present study is that the fimbriae protein FimH was used as a carrier protein for the display of entire eukaryotic proteins on the surface of E. coli bacteria. In this respect, the study goes beyond the aims of Klemm and Schembri 23 and Pallesen 19 displaying peptides.
The displayed GLuc present of the surface of E. coli bacteria is enzymatically active in converting the specific substrate indicated by biolumescence which can be used to track the bacteria during colonization of syngeneic tumors in an immunocompetent mouse model of pancreatic adenocarcinoma. Bacterial colonization of tumors as sites of local immune repression is mediated by the reticuloendothelial system of the host, by the tumor microenvironment and by the bacterial metabolism 28 . This explains why the bacterial colonization is limited to immunocompromised areas such as tumors and can be efficiently prevented by the endogenous immune system outside the tumor. In this context, bacteria tracking by anchored luciferase activity open new fields to understand the interactions and to use bacteria as vehicle to deliver diagnostic marker or therapeutic drugs to the tumor.
In a further example, we demonstrated the universal applicability of the technology by fusing the epidermal growth factor (EGF), tumor growth factor-alpha (TGF-α) and epiregulin (EPRG), respectively, to FimH (Fig. 2B). All these proteins are ligands of the EGF receptor. In order to test their activity on the EGFR receptor, the produced EGFR ligands were proteolytically released from the bacteria surface and added to EGFR + tumor cells. Interestingly, all ligands induced Tyr1173 EGFR phosphorylation, however, EGF stimulated more efficiently than EPRG and TGF-α. The data confirms previous work by Faria and colleagues who used recombinant EGF, TGF-α, EPRG, and two further EGFR ligands produced intracellularly by E. coli 29 . In contrast to conventional production methods the FimH produced ligands were proteolytically released from the carrier protein FimH without the need to lyse the E. coli bacteria. Since the bacteria remain viable and intact they can be re-used for protein production. For example, the bacteria were cultivated and harvested in a closed circuit while the POI was proteolytically cleaved from the carrier molecule FimH without destructing the bacteria. The bacteria can then be returned to the production circuit ready for repeated POI production. After proteolytic cleavage the POI is correctly folded; the technique thereby avoids known obstacles associated with protein recovery using of harsh denaturants and subsequent cumbersome protein-refolding procedures.
The POI display on the tip of fimbriae can moreover be used to screen a protein display library for various properties or for binding similar to phage display library screening. By binding to immobilized POI ligands the engineered bacteria are bound and can be amplified and enriched by repetitive rounds of binding, washing and amplifications. Such bacteria display library screening may be as efficient as phage display library screening. Taking together the described technology has the potential to broaden the production and application of recombinant proteins for diagnostic and therapeutic purposes.
Mice. All animals were used according to protocols approved by the Institutional Animal Use Committee of the State North Rhine-Westphalia (Germany) and maintained in pathogen-free conditions in a barrier facility. C57/BL6 mice at the age of 6-8 weeks were obtained from Charles River Laboratories (Wilmington, Massachusetts, USA).
Bacterial expression constructs. The GLuc-FimH construct encodes the Gaussia luciferase linked to the FimH protein derived from E. coli. The DNA gblock (Integrated DNA Technologies (IDT), Inc., Skokie, Illinois, USA) encoding the GLuc-FimH fusion protein was synthetically engineered and integrated into the bacterial expression vector pET11d (Agilent, Santa Clara, CA, USA) at the NcoI and BamHI sites. The DNA sequences encoding EGF, TGF-α and EPRG, respectively, were linked to the E. coli FimH DNA and the c-Myc tag for detection, a cleavage site sequence for the HRV3C mediated proteolytic release of a POI at the 3′-terminus, and flanked by the restriction sites for NcoI at the 5′-terminus and BamHI at the 3′-terminus. The corresponding gblocks were integrated into the MCS of the bacterial expression plasmid pET11d by using the restriction enzymes NcoI and BamHI, respectively. The pET11d expression vector (Agilent) provides IPTG-mediated de-repression of the T7 promoter in addition to IPTG-induction of T7 polymerase from the lacUV5 promoter in the DE3 containing E. coli strain BL21 (Agilent).
Immunocompetent mouse model of pancreas adenocarcinoma. CEAtg C57BL/6 mice were obtained from the Patterson Institute, Manchester, UK. Murine pancreatic tumors with CEA and click beetle (CB) Luc expression were induced in 10-week-old C57bl/6 mice by intra-pancreatic injection of 2 × 10 5 Panc02 cells. For in vivo bioluminescence imaging, Panc02 cells were genetically modified with the CBLuc and cotransfected with the hygromycin B phosphotransferase gene for clonal selection. The tumor progression was monitored by the Photon Imager bioluminescence device (Bispace Lab) after intraperitoneal administration of the CBLuc-specific substrate D-Luc (150 mg/kg) (PJK GmbH, Kleinblittersdorf, Germany). For in vivo imaging, GLuc modified E. coli were intravenously injected into tumor bearing mice. At day 2 and day 10 upon E. coli inoculation, mice were treated with IPTG (1 mM) for the induction of the GLuc-FimH expression; two hours later mice were inoculated with benzyl-coelenterazine (100 µg/mouse) and analyzed for bioluminescence by the Photon Imager bioluminescence device (Biospace Lab). Bioluminescence signals were accordingly filtered against background noise. Regions of interest were defined as regions above threshold and automatically gated by predefined program tools. There was no manual gating of regions of interest in order to avoid any incoherence. Photon emission intensity (photon/s/sr) was calculated from data of emitted photons from the respective regions of interest using the M3 vision software (Biospace Lab). (2019) 9:8410 | https://doi.org/10.1038/s41598-019-44883-z www.nature.com/scientificreports www.nature.com/scientificreports/ Immunohistological analyses. To detect GLuc expressing E. coli in tumor tissues, cryostat sections were stained with an anti-E. coli-HRP antibody (1.0 µg/ml) (Abcam, Cambridge, United Kingdom) to detect the bacteria and with a rabbit anti-GLuc antibody (New England Biolabs) and a HRP-conjugated anti-rabbit IgG-specific secondary antibody (Abcam) to detect GLuc. Specific binding was visualized by adding the DAB chromogen substrate (Biozol, Eching, Germany). Sections were additionally stained with H&E (Carl Roth, Karlsruhe, Germany).
ELISA. E. coli were transformed with the respective plasmid encoding EGF, TGF-α and EPRG, respectively, were cultivated in LB medium for two hours in the presence of IPTG (1 mM). Bacteria pellets were resuspended in 1 ml PBS and treated with HRV 3C protease (10 U) overnight at 4 °C in order to release the POI from the FimH part of the fusion protein. Released POIs were concentrated by using of an ultra-0.5 centrifugal Amicon device (Merck KGaA, Darmstadt, Germany). The detection of the proteins of interest was performed by using ELISA kits specific for human EGF, TGF-α and EPRG, respectively (CUSABIO BIOTECH, Houston, TX, USA).
E. coli produced human EGF, TGF-α and EPRG, were tested for binding to EGFR. Therefore, microtiter wells were coated with released EGFR ligands (1.0 µg/ml) in PBS and incubated with human EGFR-hIgG and as control with mouse EGFR-hIgG (each 1.0 µg/ml, R&D Systems, Minneapolis, MN, USA), respectively. A human IgG antibody of irrelevant specificity served as control. Bound EGFR-IgG fusion protein was detected by an IgG-specific biotin-labeled antibody (BioLegend) and streptavidin-HRP (Sigma-Aldrich). Specific binding was visualized by using of 3,3′,5,5′-tetramethylbenzidine (TMB) substrate (Thermo Fisher Scientific) and analyzed by using of the Multiscan GO ELISA reader (Thermo Scientific); all measurements were performed in triplicates.
Statistical analysis. Data were analyzed using GraphPad InStat, version 3.06 (GraphPad Software, San Diego, CA, USA), using unpaired or paired Student's t test with Welch's correction or unpaired Mann-Whitney test (for non-Gaussian distribution); p < 0.05 was considered to be significant.