PRO APOPTOTIC PEPTIDES AS POTENTIAL THERAPY FOR PERITONEAL ENDOMETRIOSIS

Abstract

Endometriosis is a common gynaecological disease associated with pelvic pain and infertility. Current treatments include oral contraceptives combined with non¬steroidal anti inflammatory drugs or surgery to remove lesions, all of which provide a temporary but not complete cure. Here we identify an endometriosis-targeting peptide that is internalized by cells, designated zl3, using phage display. As most endometri¬osis occurs on organ surfaces facing the peritoneum, we subtracted a phage display library with female mouse peritoneum tissue and selected phage clones by binding to human endometrial epithelial cells. Proteomics analysis revealed the zl3 receptor as the cyclic nucleotide-gated channel (13, a sorting pathway protein. We then linked zl3 with an apoptosis inducing peptide and with an endosome -escaping peptide. When these peptides were co- administered into the peritoneum of baboons with endometri¬osis, cells in lesions selectively underwent apoptosis with no effect on neighbouring organs. Thus, this study presents a strategy that could be useful to treat peritoneal en¬dometriosis in humans. Endometriosis occurs only in higher primates, including hu¬mans and baboons. Spontaneous endometriosis reportedly occurs in about 25 % of captive baboons, and its prevalence increases with captivity duration [5]. Experi¬ments in non -human primates show a clear positive correlation between endometrio¬sis and a diet containing the chemical dioxin [6, 7], which may promote endometrio¬sis by acting as an estrogen like factor, hi patients with endometriosis, many of the pathologic processes including inflammation, the immune response, angiogenesis and apoptosis are all favoured for promoting endometriosis in a manner dependent on steroid hormones - |X, 9|. Thus, in the past, most treatment was via therapeutics¬targeting steroid hormones and their receptors [10, 11 j. These drags can only be administered for a short term due to side effects. Current first-line therapy is oral con-traceptive pills, which halt an ovulation and suppress endometriosis tissue growth with minimum side effect [ 12 j. Oral contraceptive pills are also administered with nonsteroidal anti-inflammatoiy drags, further reducing endometriosis-associated pain 113, 14]. Nonetheless, these treatments do not remove endometriosis, and patients with severe symptoms must undergo surgery. However, even then symptoms can re¬cur, requiring multiple surgeries in many cases [15]. Thus, it is critical to develop new strategies to cure this disease, particularly to prevent recurrence after surgery. One proposed cause for endometriosis is through retrograde menstrual reflux of the endometrium that becomes implanted in regions of the pelvis [16], most common¬ly on the ovaries and areas facing the peritoneum [17, 18]. Thus, we hypothesized that a molecule specifically expressed on the endometrial surface could be targeted by a peritoneally injected drug if that marker is not expressed by other peritoneal sur¬faces. We report here the identification of such a reagent, namely, a 9-mer peptide that specifically binds to glandular epithelial cells of endometriosis in peritoneum. We also identified its receptor as the cyclic nucleotide-gated channel [)39(CNGB3) and confirmed its expression in endometriosis. On the basis of these findings, we de¬veloped potential therapeutics for endometriosis targeting disease lesions at peritone¬al surfaces mid tested their activity in baboon endometriosis models. This study pre¬sents a strategy that could be useful to treat peritoneal endometriosis in humans. Identification of peptides targeting peritoneal endometriosis, As yet, there is neither an in vitro model nor an in vivo animal model available to study endometriosis, other than non-human primates. To identify a peptide that specifically binds to epithelial cells in endometriosis, we hypothesized that some, if not all, human endometrial adenocarcinoma cell lines would express cell surface pro¬teins expressed in glandular epithelial cells in endometriosis. To devise a probe that specifically binds to the endometriosis surface but not to the surface of other organs facing the peritoneum, we undertook subtractive phage library.' screening [ 18, 19|. A T7 phage-based library (109clones, 101 ‘plaque-forming unit) of linear 9-mer peptides was injected into tire peritoneal cavity of a female mouse to allow absorption of phage clones to the peritoneal surface in vivo for 1 h. The precleared libraiy was re¬covered from peritoneal fluid and added to a monolayer of human endometrial ade¬nocarcinoma Ishikawa cells cultured in vitro. We choose Ishikawa cells as this cell line shares characteristics with mature endometrial epithelial cells [20, 21]. We also wanted to identify a peptide internalized by endometrial glandular epithelial cells so that a dnig conjugated with that peptide would penetrate target cells. We therefore incubated phage with live Ishikawa cells at 37 °C for 30 min to facilitate phage inter¬nalization. Selected phage clones were recovered after solubilizing cells with deter¬gent and amplified in bacteria. After three rounds of subtractive library screening, the number of phage clones with Ishikawa cell- binding activity relative to the total num¬ber of added phage increased 10,000-fold. The third screen-positive phage pool was overlayed on frozen tissue sections of endometriosis lesions surgically isolated from endometriosis patients. Immunohisto- chemislry using an anti phage antibody showed positive signals on glandular epithe¬lial cells, particularly at apical cell surfaces in endometriosis. These results suggest that, despite differences between endometriosis and endometrial adenocarcinoma, the positive phage pool bound to the Ishikawa cancer cell line contained a clone or clones binding to endometrial glandular epithelial cells. Sequencing of insert DNAs from isolated phage clones revealed the deduced consensus sequence VRRAXNXPG (where X represents a varying amino-acid residue,. The presence of the consensus sequence attests to the high specificity of selected clones. In vitro binding assays in¬dicated that each clone bound to Ishikawa cells at higher efficiency than to control skin epidermoid carcinoma A431 cells. Among the clones, zl3, which displayed the sequence VRRADNRPG, was the strongest binder. Z13 phage bound to endometrial adenocarcinoma SNG--II. RL95-2 and HeclA cells but not to 431, prostate cancer PC3 or cervical cancer HeLa cell. To confirm zl3 peptide-binding activity, we chemically synthesized zl3 peptide with an ammo-terminal lluorcscein isothiocyanate (FITC) tag and added to Ishikawa cells and control A431 cells . Fluorescence micrographs showed binding of FITC-zl3 to Ishikawa cells, but not to A431 cells. Micrograph of Ishikawa cells showed a punc¬tate cytoplasmic staining patteni, suggesting that zl3 is internalized to endosomes. Fluorescence micrographs of Ishikawa cells (left) and control A431 cells (nght) overlayed with a synthetic z.13 peptide tagged with fluorescein isothiocyanate (FITC) and left at 37 °C for 15 min. Scale bar, 50 pm. Visualization and isolation of the zl3 peptide receptor. Left: cell surface proteins expressed on Ishikawa cells were bioti¬nylated. Cell lysates were bound to z 13 peptide- conjugated agarose beads, and bound proteins were eluted with irrelevant peptide (lane 1) or zl3 peptide (lane 2). Biotinyl¬ated proteins in each eluate were detected by peroxidase- conjugated avidin and a lu¬minescent peroxidase substrate. Right: silver staining of peptide affinity purified zl3 receptor from endometriosis. Endometriosis tissues isolated from patients were ho¬mogenized, and microsome membrane fraction was prepared. Proteins solubilized with detergent were applied to a zl3 peptide conjugated agarose column, and bound proteins were eluted with irrelevant peptide (lane 1) or zl.3 peptide (lane 2). Proteins in each eluate in SDS-PAGE were detected by silver staining, (c) Fluorescence mi¬crographs of HeLa cells transfected with control empty vector (upper row) or with an expression vector encoding CNGB3 MYC (lower row). Binding of FlTC-z.13 pep¬tide (green) to HeLa cells transfected with mammalian expression vectors (a,d), im- munostained with anti-MYC followed by Alexa 549 (red)-conjugated anti-mouse IgG antibody (b,e) and merged images including 4',6-diamidino-2 phenylindole (blue) to indicate nuclear staining (c.f). Scale bar, 20 urn. Identification of the zl.3 peptide receptor, To develop a clinically relevant therapeutic strategy, we searched for the /13 peptide receptor. To do so, Ishikawa cells were surface biotinylated and lysed, and then lysates were incubated with z 13 peptide-conjugated agarose beads. Bead bound materials were then eluted by zl3 peptide, and biotinylated proteins detected by avi- din blot, revealing a single 68-kDa protein. To identify this protein, the microsomal membrane fraction was prepared from endometriosis tissue surgically removed from patients. Membrane proteins solubilized with detergent were applied to a zl 3 peptide affinity column and column-bound materials were eluted by zl3 peptide. A silver- stained gel revealed a 68-kDa protein, and proleomic analysis identified the peptide sequence, QRTALYK, which is unique to the CNGB3 GNGB3 protein has six trans¬membrane domain [22]. A large part of this protein is buncd in the lipid bilayer, the N and C-terminal domains are cytoplasmic and presumably the zl3-binding regions are extracellular. To assess binding of FITC-zl3 to CNGB3, we transfected HeLa cells with a mammalian expression vector encoding a GNGB3—MYC fusion protein or with control empty vector. FITG-z.13 did not bind to vector transfected control cells. HeLa cells transfected with the CNGB3--MYG expression construct and stained by anti-MYC antibody showed an endosome-like pattern . When FITC-zl3 was add¬ed to culture medium of CNGB3 -MYC-expressing HeLa cells, FITC-zl3 bound to sites marked by MYC epitope expression. The kinetics of FITC-zl3 binding to CNGB3 MYC expressing HEK293T cells showed a A'd of 9.759 x 10 11M and aB- max-of 0,9086 moles per receptor. These results indicate that recombinant CNGB3 ex¬pressed in a mammalian cell has zl 3 peptide-binding activity on the cell surface and internalized to endosome, supporting the hypothesis that GNGB3 is the zl3 recep- tor.To examine CNGB3 protein expression in endometriosis, we generated a mouse monoclonal antibody against a peptide sequence of human CNGB3, corresponding to K [12, 14, 151 to P [1, 3] within the cytoplasmic domain. Antibody specificity was validated by immunostaining of CNGB3—MYG expressed in HeLa cells. This anti¬body robustly stained glandular epithclia of endometriosis in tissue sections. When we evaluated endometriosis tissue sections from 35 endometriosis patients, we found that 31 specimens showed strongly positive immunostaining and 4 showed weak/negative immunostaining. Peritoneal surfaces from cycling women without en¬dometriosis were not stained by this antibody, whereas those from endometriosis pa¬tients were stained by this antibody, suggesting that endometrial cells are spread across a wide area on the peritoneum of endometriosis patients. Eutopic endometrial tissues at secretory and proliferative phases were weakly stained by this antibody. Immunohistochemistry of human tissues showed that this antibody did not stain the surface of organs facing the peritoneal cavity (data not shown). Histology of hematoxylin and eosin-stained tissue sections revealed evidence of endometriosis in all three animals treated with a mixture of dKLAK- /13 and 1ILA11 zl3 peptides. Terminal deoxynucleotidyl transferase mediated dUTP nick end label¬ing (TUNEL) assays of endometriosis lesions collected from three untreated control baboons revealed no TUNEL positivity in gland tissue. By contrast, we found TUNEL positive glands in tissues collected from all three baboons treated with the dKLAK zl3 and HLAlT zl3 peptide mixture, those signals were seen in glandular epithelial cells in ovarian endometriosis and in the lumen of endometrial gland in the omentum. No evidence of apoptosis was detected in cells facing peritoneum in liver, kidney, spleen, colon and stomach (data not shown) These results show overall, as proof of concept, that /13 targeted induced apoptosis occurred in endometriosis mod¬el in baboons in vivo. A general strategy of translational research is to identity a gene product associ¬ated with a disease of interest and then develop drug targeting that molecule. Howev¬er, in identifying peptides that target endometriosis but not non endometrial cells lac¬ing the peritoneal cavity, we took a novel approach that does not require knowledge of a target We also demonstrated that CNGB3 is the receptor for zl3 peptide based on its expression pattern, not its function. We feel that analysing CNGB3 function in detail is not a priority' for clinical application of this work.