Journal of Immunological Methods 309 (2006) 120–129 www.elsevier.com/locate/jim Research paper Isolation of cell specific peptide ligands using fluorescent bacterial display libraries ⁎ Karen Y. Dane, Lisa A. Chan, Jeffrey J. Rice, Patrick S. Daugherty Department of Chemical Engineering, University of California, Santa Barbara, CA 93106, United States Received 3 May 2005; received in revised form 15 August 2005; accepted 28 November 2005 Available online 11 January 2006 Abstract Methods for identifying and producing cell specific affinity reagents are critical in cell detection, separation, and therapeutic delivery applications, yet remain difficult and time consuming. To address these limitations, a rapid and quantitative screening approach was developed using intrinsically fluorescent bacterial display peptide libraries and fluorescence-activated cell sorting (FACS). High-throughput screening of fluorescent libraries yielded a panel of peptide ligands mediating specific recognition of human breast cancer tumor cells. Clonal populations of fluorescent, peptide-displaying bacteria enabled single-step, fluorescent labeling of the target cells for cytometry and microscopy analysis. Isolated peptides could be categorized into several distinct groups possessing strong consensus sequences with as many as six identities. Importantly, individual clones exhibited high specificity target cell binding, with more than 80-fold increased binding to tumor cells (ZR-75-1) relative to cell lines derived from healthy tissue (HMEC, MCF-10A). Fluorescent display libraries thus provide a powerful new methodology for parallel identification of cell specific affinity ligands. © 2006 Elsevier B.V. All rights reserved. Keywords: Bacterial display; Breast cancer; FACS; Peptide library; Targeting ligand 1. Introduction peptide libraries displayed on filamentous bacterio- phage (Brown, 2000; Landon and Deutscher, 2003). Cell specific affinity reagents are widely used in Phage display libraries have been applied in ligand biotechnology applications ranging from cell detec- selections using cultured cells, animals (Pasqualini and tion, separation, and characterization (Brown, 2000), Ruoslahti, 1996), or human patients (Zurita et al., to the delivery of imaging and therapeutic agents to 2004), yielding peptides specific for cell surface specific cell types in vivo (Arap et al., 1998; Landon markers (Rasmussen et al., 2002). The principle and Deutscher, 2003). Cell specific reagents have been advantage of display library methodologies is that discovered almost exclusively using antibody or selection can be performed in a native-like, mem- brane-bound environment without a priori knowledge Abbreviations: FACS, fluorescence activated cell sorting; GFP, of the target cell receptors. Peptide ligands generated green fluorescent protein; OmpA, outer membrane protein A; CPX, in this manner have proven useful for in vivo imaging circularly permuted outer membrane protein OmpX; SA-PE, strepta- studies (Ladner, 1999), therapeutic targeting (Park et vidin R-phycoerythrin ⁎ Corresponding author. Tel.: +1 805 893 2610; fax: +1 805 893 al., 2002; Sapra and Allen, 2002), and for identifica- 4731. tion of cell specific surface markers (Pasqualini et al., E-mail address: [email protected] (P.S. Daugherty). 2000; Landon and Deutscher, 2003). 0022-1759/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jim.2005.11.021 K.Y. Dane et al. / Journal of Immunological Methods 309 (2006) 120–129 121 The development of cell specific affinity reagents (Walkersville, MD). ZR-75-1 tumor cells were cultured in using display technology would benefit from more RPMI-1640 medium (ATCC) with 10% fetal bovine efficient and quantitative screening processes. Previous serum (Invitrogen, Carlsbad, California). Normal mam- efforts to identify cell binding peptides using phage and mary cells were cultured in complete mammary epithelial cell display methodologies have relied upon a selection cell medium (MEGM, Cambrex Bio Science). The process referred to as panning (Lu et al., 1995; Brown et OmpA 15mer bacterial display library was propagated al., 2000; Nakajima et al., 2000). These studies have as previously described (Bessette et al., 2004). For the successfully enabled isolation of cell binding or specificity selection, two random peptide libraries of the invading peptides, though rarely reveal a consensus form X15 and X2CX7CX2 were constructed as fusions to motif among the isolated peptides (Brown et al., 2000; the extracellular N-terminus of a circularly permuted Nakajima et al., 2000; Taschner et al., 2002). Methods variant of outer membrane protein OmpX (CPX) (Rice et for ligand selection using display on the outer surface of al., in press). Following the native signaling sequence of cells are attractive since fluorescence-activated cell OmpX, CPX was constructed by fusion of the native C- sorting (FACS) instrumentation can be applied for and N-terminus of OmpX with a GGSG linker, and by quantitative clone screening and analysis (Feldhaus et opening loop 2 between residues S53 and S54 yielding a al., 2003; Bessette et al., 2004). Importantly, cell display scaffold with extracellular C- and N-termini. Each library methods allow direct clone recovery from target cells was comprised of 2×109 members. The gene encoding and amplification by regrowth, without requiring the CPX scaffold was inserted into the multiple cloning separation or infection steps. Consequently, improved site of pBAD33 (Guzman et al., 1995)downstreamof cell display library and screening methodologies would alajGFP1 (Bessette and Daugherty, 2004). This bicis- substantially simplify cell specific ligand isolation, as tronic cassette allowed for simultaneous expression of well as enable quantitative library screening and clone both alajGFP1 and CPX under the control of the analysis. arabinose inducible araBAD promoter (Guzman et al., A high-throughput, quantitative screening approach 1995). was developed to isolate cell specific affinity reagents using fluorescent bacterial display peptide libraries 2.2. Library selection, screening, and analysis coupled with FACS. Fluorescent display libraries were constructed that co-express a green fluorescent protein For the first two cycles of selection, bacteria binding (GFP) intracellularly, while efficiently displaying pep- to tumor cells were selected using co-sedimentation. tides on the surface of Escherichia coli to the Frozen aliquots of 2.5×1011 cells containing the OmpA extracellular environment. In this way, living bacterial surface display library were thawed and grown cells served as fluorescent affinity reagents, thereby overnight in Luria Bertani (LB) broth supplemented enabling quantitative library screening and subsequent with 34 μg/mL chloramphenicol (Cm) and 0.2% D-(+)- clone characterization. To demonstrate the utility of this glucose. The bacteria were subcultured 1:50 for 3 h in approach, two different fluorescent bacterial display LB supplemented with Cm and induced with 0.02% L- libraries were screened for peptides binding to human (+)-arabinose for 2 h. Tumor cells were harvested from breast cancer tumor cells. Quantitative screening using dishes, resuspended in conical tubes, and co-incubated FACS succeeded in identifying, for the first time, an with 100-fold excess bacterial cells for 1 h on an array of unique peptide consensus groups binding to the inversion shaker at 37 °C. For round 1, 108 tumor cells target cells. By incorporating negative selections using were incubated with 1010 bacterial cells, while in round healthy breast derived cell lines, highly specific tumor 2, 10-fold fewer cells of each type were used. Cell cell binding peptides were identified. suspensions were centrifuged at 2500 rpm for 4 min and non-binding bacteria in the supernatant were removed. 2. Materials and methods Tumor cells were washed three times in phosphate buffered saline (PBS) and resuspended in LB for 2.1. Cell lines, vectors, and reagents overnight growth of binding bacteria. Recovered bacterial clones were counted by plating on LB-agar The human breast ductal carcinoma cell line (ZR-75-1) plates supplemented with Cm. A depletion step was and the immortalized human mammary cell line (MCF- performed between rounds one and two to remove any 10A) were obtained from ATCC (American Type Culture bacteria sedimenting as a result of increased aggregation Collection, Manassas, VA). Clonetics® human mammary by centrifuging bacteria at 2500 rpm for 4 min and epithelial cells (HMEC) were from Cambrex Bio Science recovering 95% of the supernatant. 122 K.Y. Dane et al. / Journal of Immunological Methods 309 (2006) 120–129 For flow cytometric sorting and screening, plasmids 2.3. Soluble peptide interactions of the unselected and selected OmpA library popula- tions were extracted from bacterial cells using Qiagen Two cyclic peptides with N-terminal biotin (bio) and Miniprep kits (Qiagen, Valencia, CA). Subsequently, disulfide bond formation (pep5: (bio)GGCLQILPTL- these plasmids were transformed into E. coli MC1061/ SECFGR and pep17 (bio)GLKVCGRYPGICDGIR) pUC18-EGFP for constitutive GFP expression. Sample and a linear peptide binding to human erythrocytes preparation for FACS was performed using similar (unpublished) pepE1: (bio)GKYTWYGYSLRANWMR) techniques as described above with 106 tumor cells synthesized by Invitrogen (Carlsbad, CA) were used suspended in 2 mL of culture medium. After coincuba- for tumor cell binding studies. Peptides were purified