Preclinical Evaluation of MEDI0641, a Pyrrolobenzodiazepine-Conjugated

Published OnlineFirst May 18, 2017; DOI: 10.1158/1535-7163.MCT-16-0825

Large Molecule Therapeutics Molecular Cancer Therapeutics Preclinical Evaluation of MEDI0641, a Pyrrolobenzodiazepine-Conjugated Antibody–Drug Conjugate Targeting 5T4 Jay Harper1, Christopher Lloyd2, Nazzareno Dimasi3, Dorin Toader3, Rose Marwood2, Leeanne Lewis2, David Bannister2, Jelena Jovanovic2, Ryan Fleming3, Francois D'Hooge4, Shenlan Mao1, Allison M. Marrero1, Martin Korade III1, Patrick Strout1, Linda Xu3, Cui Chen1, Leslie Wetzel1, Shannon Breen1, Lilian van Vlerken-Ysla1, Sanjoo Jalla5, Marlon Rebelatto6, Haihong Zhong1, Elaine M. Hurt1, Mary Jane Hinrichs7, Keven Huang1, Philip W. Howard4, David A. Tice1, Robert E. Hollingsworth1, Ronald Herbst1, and Adeela Kamal1,8

Abstract

Antibody–drug conjugates (ADC) are used to selectively demonstrated significant target-dependent activity in vitro and deliver cytotoxic agents to tumors and have the potential for in vivo; however, the ADC conjugated with a PBD payload increased clinical benefit to cancer patients. 5T4 is an oncofetal (5T4-PBD) elicited more durable antitumor responses in vivo antigen overexpressed on the cell surface in many carcinomas than the tubulysin conjugate in xenograft models. Likewise, the on both bulk tumor cells as well as cancer stem cells (CSC), has 5T4-PBD more potently inhibited the growth of 5T4-positive very limited normal tissue expression, and can internalize when CSCs in vivo, which likely contributed to its superior antitumor bound by an antibody. An anti-5T4 antibody was identified activity. Given that the 5T4-PBD possessed both potent anti- and optimized for efficient binding and internalization in a tumor activity as well as anti-CSC activity, and thus could target-specific manner, and engineered cysteines were incorpo- potentially target bulk tumor cells and CSCs in target-positive rated into the molecule for site-specificconjugation.ADCs indications, it was further evaluated in non-GLP rat toxicology targeting 5T4 were constructed by site-specifically conjugating studies that demonstrated excellent in vivo stability with an the antibody with payloads that possess different mechanisms acceptable safety profile. Taken together, these preclinical data of action, either a DNA cross-linking pyrrolobenzodiazepine support further development of 5T4-PBD, also known as þ (PBD) dimer or a microtubule-destabilizing tubulysin, so that MEDI0641, against 5T4 cancer indications. Mol Cancer Ther; each ADC had a drug:antibody ratio of 2. The resulting ADCs 16(8); 1–12. 2017 AACR.

Introduction warhead, typically a cytotoxic small molecule, which has been chemically modified with a linker used to conjugate the drug to Antibody–drug conjugates (ADC) represent a promising ther- the antibody. The ADC binds to an antigen on the surface of a apeutic approach to effectively treat cancer while reducing drug- cancer cell, and the ADC/antigen complex is internalized and related toxicities by combining the specificity of an antibody trafficked to the lysosome, where the cytotoxic agent is released with the potency of cytotoxic agents (reviewed in refs. 1–3). An through cleavage of a linker or through proteolytic degradation ADC consists of an antibody that is conjugated with a potent of the antibody itself if a noncleavable linker is used to conjugate the warhead to the antibody. The warhead exits the 1Oncology Research, MedImmune, LLC, Gaithersburg, Maryland. 2Antibody lysosome where it can then bind to its target, typically either Discovery and Protein Engineering, MedImmune, Ltd, Cambridge, United King- microtubules or DNA, depending on its mechanism of action. dom. 3Antibody Discovery and Protein Engineering, MedImmune, LLC, Gaithers- The binding of these cytotoxic agents to their targets results in burg, Maryland. 4Spirogen, London, United Kingdom. 5Project Management, cell-cycle arrest that subsequently leads to cell death. Given the MedImmune, LLC, Gaithersburg, Maryland. 6Pathology, MedImmune, LLC, 7 generally high potency of cytotoxics most commonly used in Gaithersburg, Maryland. Biologics Safety Assessment, MedImmune, LLC, ADCs, it is important to identify targets exhibiting both good Gaithersburg, Maryland. 8Ferring Pharmaceuticals, San Diego, California. expression in tumors and limited normal tissue expression to Note: Supplementary data for this article are available at Molecular Cancer reduce the possibility of on-target toxicity. Therapeutics Online (http://mct.aacrjournals.org/). The oncofetal antigen 5T4, also known as trophoblast gly- Corresponding Author: Jay Harper, MedImmune, LLC, One MedImmune Way, coprotein (TPBG), is a 72-kDa glycoprotein that is typically Gaithersburg, MD 20878. Phone: 301-398-5209; Fax: 301-398-8958; E-mail: only expressed during embryonic development, whereas [email protected] expression in normal adult tissues is very limited (4, 5). doi: 10.1158/1535-7163.MCT-16-0825 Expression of 5T4, however, is reported to be significantly 2017 American Association for Cancer Research. upregulated in many types of carcinomas, including, but not

www.aacrjournals.org OF1

Harper et al.

limitedto,cancersofthelung,breast,stomach,prostate, binding, and S239C to provide engineered cysteines for site- colon, and ovaries, and its expression has been correlated with specific conjugation of payloads, in the CH2 domain (34) based poor prognosis in multiple indications (6–13). on EU numbering (35). The biology of 5T4 in normal and pathologic tissues is not well understood. The extracellular domain of 5T4 contains Site-specific conjugation of warheads several leucine-rich repeats that have been implicated in pro- Site-specific conjugates were generated as described previously tein binding that can promote cell–matrix and/or cell–cell (34), and see Supplementary Data for a detailed protocol. interactions (14). However, no ligand or coreceptor has been identified for 5T4 to date. During embryonic development, 5T4 Cell lines appears to regulate CXCR4-mediated chemotaxis (15) and All cell lines (MDA-MB-361, HCC1937, SUM159, and T47D modulates dendritic branching of granule cells in the devel- breast carcinoma cells; DU 145 prostate cancer cells, NCI-N87 oping olfactory bulb (16). The extracellular domain is critical gastric cancer cells, DMS114 lung carcinoma cells; Panc-1 and for 5T4-mediated inhibition of Wnt/b-catenin signaling (17). HPAC pancreatic carcinoma cells; HCT-116 colorectal carcinoma 5T4 has a short cytoplasmic domain consisting of 44 amino cells; HepG2 hepatocellular carcinoma cells) were obtained from acids and aside from a serine–asparagine–valine motif that is ATCC in the following years: MDA-MB-361, November 2011; implicated in binding PDZ-containing molecules, there are no HCC1937, November 2008; SUM159, January 2012; T47D, Sep- classical signal transduction motifs contained in the sequence tember 2012; DU 145, October 2011; NCI-N87, August 2012; (14, 18). Studies with overexpression or knockdown of 5T4 DMS114, July 2008; Panc-1, January 2012; HPAC, April 2008; have demonstrated that it likely plays a role in adhesion, HCT-116, January 2013; and HepG2, February 2012. Upon cytoskeletal organization and motility, and epithelial-to-mes- arrival, stock vials undergo authenticity and IMPACT testing (for enchymal transition associated with metastasis (18–20). 5T4 murine viruses) by IDEXX Bioresearch using real-time PCR anal- knockout mice are viable, although adults have a number of yses. Cells are also designated mycoplasma negative after under- structural defects in the brain as a result of changes that occur going internal PCR assays. Master cell banks are made from these during brain development in utero likely caused by the role of stock vials, and each time a working cell bank is created from 5T4 in dendritic branching (15). these master banks, the cells are retested at IDEXX and internally. 5T4 is also expressed on cancer stem cells (CSC), which are Cells used for all described experiments were derived from these hypothesized to be responsible for chemotherapeutic resis- working cell banks and were used within 10 passages following tance and recurrence of cancer (21, 22). Targeting both 5T4- thawing. After 10 passages (typically within 2–3 months follow- expressing bulk tumor cells and CSCs could, theoretically, ing thawing), cells were discarded and new cell vials were thawed produce a more durable clinical response. Previous reports as needed. All cell lines were cultured using ATCC-recommended fi have separately shown that 5T4-targeted ADCs armed with media at 37 C in a humidi ed incubator with 5% CO2 except for either a microtubule-targeting payload or a DNA cross-linking the MDA-MB-361 breast carcinoma cells that were cultured in pyrrolobenzodiazepine (PBD) payload can inhibit both bulk atmospheric air at 37 C. and CSC tumor cell populations (22–25). Given the differences in mechanisms of action between these payloads (26–28), In vitro cytotoxicity we wanted to compare the activities of these two drugs side- The cells were plated in culture media at a density of 2,000 to by-side. 5,000 per well (depending on the growth kinetics of each cell We report data from in vitro and in vivo preclinical tumor models line) of tissue culture-treated 96-well plates in a volume of comparing the activity of the site-specifically conjugated ADCs 80 mLandallowedtoadhereovernight.A5 stock of each 5T4-PBD and 5T4-tubulysin (5T4-Tub). The 5T4-PBD, subse- concentration of antibody or ADC to be tested was prepared by quently labeled as MEDI0641, elicited superior antitumor activity diluting the test articles in culture medium. Twenty microliters than 5T4-Tub and reduced the CSC population in vivo, unlike of each test article was added to cells in duplicate or triplicate ADCs conjugated with either tubulysin or auristatin payloads. such that the final dose curve ranged from 4 mg/mL down to 61 Following these observations, the preclinical safety and pharma- pg/mL in a stepwise 1:4 serial dilution series. The treated cells cokinetic profile of MEDI0641 was evaluated in rats to assess its in were cultured at 37 C/5% CO2 for 72 to 144 hours (depending vivo stability and off-target toxicity profile. on the growth kinetics of each particular cell line). The CellTiter-Glo (CTG) Luminescent Viability Assay (Promega) wasusedtodeterminerelativecytotoxicity.Briefly, 100 mLof Materials and Methods CTG reagent was added to each well, allowed to incubate for Generation of the lead anti-5T4 antibody, 5T4_0108 10 minutes at room temperature with mild shaking, and then The parental antibody, A07, was derived by immunization of the absorbance of each sample at 560 nmol/L was read using VelocImmune II mice (Regeneron; ref. 29) with recombinant an EnVision luminometer (PerkinElmer). The percent cell via- human 5T4 using a 48-day rest/boost immunization protocol bility was calculated by the following formula: (average (30) and was identified from a biochemical assay of hybridoma luminescence of treated samples/average luminescence of fi supernatants for binding to 5T4. This antibody underwent af nity control samples) 100. IC50 values were determined using optimization by CDR-directed mutagenesis (31) and phage dis- logistic nonlinear regression analysis with GraphPad Prism play selection (32, 33). The lead antibody, 5T4_0108, was iden- software. tified in a biochemical assay, where it demonstrated improved binding to 5T4 over the parental A07. This antibody was then In vivo efficacy studies expressed and purified as a full-length, human IgG1 antibody All in vivo studies were carried out in compliance with the containing the mutations L234F to significantly reduce FcgR guidelines published by the Association for Assessment and

OF2 Mol Cancer Ther; 16(8) August 2017 Molecular Cancer Therapeutics

MEDI0641: A Novel PBD-Based Anti-5T4 ADC

Accreditation of Laboratory Care. The MedImmune Institutional examinedmicroscopicallybyaboard-certified veterinary Animal Care and Use Committee approved all study protocols. pathologist. For cell line–derived xenograft models, tumor cells were harvested Additional Materials and Methods can be found in the Sup- from culture flasks, mixed with Matrigel in a 1:1 ratio, and plementary Data. implanted subcutaneously or orthotopically into 5- to 6-week- old athymic nude mice (Envigo). Resulting tumors were measured twice a week throughout the course of the experiments, with Results tumor volume calculated using the following formula: tumor Expression of 5T4 in clinical tumor samples volume (mm3) ¼ (length width2)/2. Test articles were admin- Previous reports have indicated that 5T4 is overexpressed in a istered intravenously. See Supplementary Data for additional variety of human carcinomas (6–13). To confirm these findings information. and to focus on cancer types with the highest incidence of 5T4 expression, extensive IHC of tumor microarrays and tumor Evaluating effects on CSCs in tumors samples as well as normal tissue arrays was conducted. 5T4 Mice bearing NCI-N87 and MDA-MB-361 xenografts were was broadly expressed in tumor microarrays and clinical tumor treated with the indicated ADC, and upon initial signs of tumor samples representing a number of cancer types, with the highest regression, tumors were excised, cut into 4-mm pieces and cryo- incidence in gastric cancer, non–small cell lung cancer (NSCLC), preserved using Cryostor. Frozen tumor pieces were thawed at and head and neck squamous cell carcinoma (HNSCC; Fig. 1A). 37C, washed twice in Hank's balanced salt solution (HBSS), and At least 90% of tumors analyzed in each of these cancers had further minced using sterile scalpel blades. To obtain single-cell detectable levels of 5T4 expression by IHC, and in NSCLC suspensions, the tumor pieces were then mixed with 200 U/mL of samples, the prevalence of 5T4 was slightly higher in squamous ultrapure collagenase III in DMEM/F12 medium. The tumor NSCLC compared with adenocarcinoma. Greater than 60% of suspension was incubated at 37C for approximately 1 hour, breast cancer specimens, including triple-negative samples, were with mechanical disruption every 30 minutes by pipetting with a 5T4 positive. Clear membrane localization of 5T4 immunos- 5-mL pipette. At the end of the incubation, cells were filtered taining was prevalent in various types of cancers (Fig. 1B). through a 70-mm nylon mesh and centrifuged at 1,200 rpm for 5 Although gastric cancers had the highest incidence of staining, minutes. Cells were washed twice with HBSS. Following the last only about 20% of samples could be classified as high/positive wash, cells were put through a 40-mm cell strainer and counted with a pathology score of 2þ or 3þ observed in greater than using a Vi-Cell XR Cell Viability Analyzer. Cells were assayed for 50% of the tumor cells. In contrast, roughly 70% of all NSCLC aldehyde dehydrogenase activity as a measure of CSCs using the and HNSCC samples exhibited high/positive staining. Stemcell Technologies Aldefluor Kit following the manufacturer's Similar to previous findings (6, 9), 5T4 expression in normal instructions. The cells were run on an LSRII flow cytometer and tissues was extremely limited with minimal expression observed analyzed with FlowJo software. Statistical significance of differ- in epithelial cells of the renal tubules, gastric glands, skin, adrenal ences between groups was based on P values determined through gland, urinary bladder, gall bladder, uterus, and cervix. Collec- unpaired t tests. tively, these data confirmed 5T4 would be a good tumor-associ- The methods used to confirm the CSC phenotype of Aldefluor ated antigen for a targeted approach such as an ADC and identified cells from NCI-N87 and MDA-MB-361 tumors can be found in potential indications to target in the clinic due to prevalent Supplementary Data. expression of 5T4.

Rat toxicology studies with MEDI0641 Generation and in vitro characterization of 5T4_0108 antibody Male Sprague Dawley rats (12 per group) were administered a A5T4-specific antibody, A07, was isolated from transgenic single intravenous injection (day 1) of 1 or 2 mg/kg anti-5T4 mice expressing human antibody variable genes immunized antibody conjugated to SG3249. Control rats (12/group) were with recombinant human 5T4. The KD of this antibody was administered a single intravenous injection of vehicle control on measured to be approximately 100 nmol/L as determined via day 1. Animals were necropsied on days 8 and 29 (6 per time Biacore analysis; therefore, in vitro affinity optimization using point/group) to evaluate acute and delayed effects, respectively. phage display was conducted and led to the generation of All main study animals were evaluated for clinical signs, changes 5T4_0108. 5T4_0108 is a human IgG1 with the following in body weight, clinical pathology, gross pathology with organ mutations engineered into the heavy chain CH2 domain based weights, and microscopic observations. In addition, groups of on EU numbering (35): S239C to introduce a free cysteine for toxicokinetic satellite animals (12/group) were included in each site-specific conjugation of payloads, and L234F to reduce treatment arm to measure plasma concentration of total antibody engagement of FcgR and activation of immune effector cells and ADC. All toxicokinetic satellite animals were evaluated for in conjunction with conjugation of payload at S239C (34). The clinical signs, changes in body weight, and pharmacokinetic 5T4_0108 antibody has a KD of 6.5 nmol/L for 5T4 with on- analysis. and off-rates of 3.3 105 (1/Ms) and 2.1 103 (1/s) respec- Hematology and serum chemistry samples were collected and tively. It binds to and is internalized by 5T4-expressing cells analyzed on days 8, 14, and 29. Blood samples for pharmacoki- (Fig. 2A; Supplementary Fig. S1), and conjugation of a payload netic analysis were collected in K2 EDTA tubes at multiple time at the S239C site did not affect binding or the internalization points on days 1, 2, 3, 8, 15, 16, 22, and 29. kinetics (Supplementary Fig. S1). A gross necropsy was performed on all main study animals The improved affinity translated into superior efficacy. and a standard list of organs, including brain, lung, liver, 5T4-Tub, which consists of 5T4_0108 conjugated with a tubulysin kidney, spleen, thymus, testes, heart, and bone, were embedded payload, was 14-fold more potent than an ADC consisting of a in paraffin, sectioned, stained with hematoxylin and eosin, and site-specific variant of the parental antibody (ssA07) conjugated

www.aacrjournals.org Mol Cancer Ther; 16(8) August 2017 OF3

Harper et al.

Figure 1. Relative prevalence of 5T4 expression in clinical tumor samples. A, IHC for 5T4 expression was conducted to determine the prevalence of 5T4 expression in tumor samples from patients with various cancers. The values represent the percentage of tumor samples that have detectable levels of 5T4 immunostaining. TNBC, triple-negative breast cancer; NSCLC, non- small cell lung cancer. The number in parentheses represents the number of tumor samples tested for each cancer type. B, Representative images of 5T4 immunostaining in various cancer types.

with the same payload (ssA07-Tub; Supplementary Fig. S2; ing low 5T4 expression levels (4,000 5T4 molecules per cell; Supplementary Table S1). Supplementary Table S2). Although the relative activities of 5T4-PBD and 5T4-Tub were In vitro efﬁcacy of anti-5T4 ADCs conjugated with tubulysin and comparable in each cell line, there was a clear correlation between PBD payloads the level of 5T4 expression and in vitro potency for both 5T4-ADCs As described above, the initial studies to evaluate a candidate in general (Fig. 2). 5T4-PBD and 5T4-Tub were most potent anti-5T4 ADC were conducted with 5T4-Tub. However, the activ- against 5T4-high MDA-MB-361 cells generating IC50 values in ity of cytotoxics can be context dependent, where certain classes of the pg/mL range based on antibody concentration (Table 1). In drugs are more effective than others depending on the type of addition, a greater maximum cytotoxicity was achieved at doses cancer. Therefore, studies were conducted to determine the rela- that are logs lower than what was needed in the other two cell tive activity of 5T4-Tub and 5T4-PBD, an ADC composed of lines. The 5T4-ADCs induced potent cytotoxicity (IC50 values in 5T4_0108 conjugated with SG3249, a DNA cross-linking PBD the single-digit ng/mL range) of DU 145 cells with moderate payload (27). 5T4 expression and had the weakest activity against NCI-N87 cells Comprehensive in vitro cytotoxicity assays were then con- with very low levels of 5T4 in vitro. ducted as an initial screen to compare relative activities of Both the PBD and tubulysin warheads should elicit apopto- 5T4-PBD and 5T4-Tub. To determine whether 5T4 expression sis, and this was conﬁrmed using companion in vitro assays. In levels impact ADC activity, these ADCs were tested against one assay, the percent viability was assessed, and in a second MDA-MB-361 breast cancer cells representing high 5T4 expres- assay, cleaved caspase activity was used as an indicator of sion (65,000 5T4 molecules/cell), DU 145 prostate cancer apoptosis (36). Following treatment with 5T4-PBD or 5T4-Tub cells representing moderate 5T4 expression (30,000 5T4 (Supplementary Fig. S3), decreased viability is directly corre- molecules/cell) and NCI-N87 gastric carcinoma cells represent- lated with increased caspase-3/7 activity in various cell lines,

OF4 Mol Cancer Ther; 16(8) August 2017 Molecular Cancer Therapeutics

MEDI0641: A Novel PBD-Based Anti-5T4 ADC

www.aacrjournals.org Mol Cancer Ther; 16(8) August 2017 OF5

Harper et al.

Table 1. Relative activity of 5T4-ADCs against cell lines with varying levels of cell surface 5T4 expression MDA-MB-361 (65K 5T4/cell) DU 145 (30K 5T4/cell) NCI-N87 (4K 5T4/cell)

ADC IC50 (ng/mL) Max kill (%) IC50 (ng/mL) Max kill (%) IC50 (ng/mL) Max kill (%) 5T4-Tub 0.85 78.8 2.67 80.5 47.3 77.0 5T4-PBD 0.39 90.7 2.47 77.0 111.1 91.1

NOTE: The table represents average IC50 values and the maximum cytotoxicity (max kill percentages) from multiple assays run for each ADC against each cell line.

confirming that both 5T4-PBD and 5T4-Tub induce an apo- gated 5T4-Tub could be due to differential effects of their ptotic cell death pathway. warheads on the CSC populations, studies were carried out to determine 5T4 expression on CSCs in cultured tumor cell lines In vivo antitumor activity of 5T4-PBD and 5T4-Tub and the effects of these ADCs and their respective warheads on Collectively, the in vitro data suggest that both 5T4-PBD and CSCs in vitro and in vivo. Cell lines of various cancer types were 5T4-Tub were able to induce potent cytotoxicity via apoptosis in a tested for 5T4 expression on both total tumor cells as well as þ þ target-dependent manner. To confirm that the in vitro activity of CSCs. CSCs were defined as Aldefluor or CD44 CD24 cells þ 5T4-PBD and 5T4-Tub could translate to potent antitumor activity in breast cancer (37, 38), CD44 CD24 cells in prostate cancer þ þ in vivo, these ADCs were tested against tumors with varying levels (39), CD90 cells in hepatocellular carcinoma (40), Aldefluor þ þ of 5T4 expression as determined by IHC: NCI-N87 xenografts cells in gastric cancer (41), and CD24 CD44 cells in pancre- with low/moderate in vivo 5T4 expression, DU 145 xenografts atic cancer (42). In all cases, 5T4 expression on the CSC with moderate levels of 5T4 expression and MDA-MB-361 xeno- population was equal to or higher than that on non-CSCs grafts with high 5T4 levels. (Fig. 4A), even when the cell line had little to no 5T4 expression In the NCI-N87 model with low/moderate 5T4 levels, 5T4-PBD as a whole (Supplementary Fig. S4). was able to achieve durable regression with a single administra- We first determined the effects of the various warheads on tion of a 1 mg/kg dose, whereas 5T4-Tub achieved a period of CSCs in vitro. CSC-enriched spheroids derived from several of regression with a 5 mg/kg dose administered once a week for four these cell lines were treated with PBD, tubulysin, or auristatin weeks, but then the tumors began to regrow shortly after cessation warheads to determine whether these warheads could inhibit of treatment (Fig. 3A). Initially, tumor regression was observed in the growth of CSCs in vitro. The auristatin MMAE was included the DU 145 model following five doses administered once a week as an auristatin-conjugated 5T4-ADC has been reported to of either a 1 mg/kg dose of 5T4-PBD or a 5 mg/kg dose of 5T4-Tub have activity against CSCs (23). In our studies, only the PBD (Fig. 3B). However, the 5T4-PBD–treated tumors remained dimer warhead potently inhibited CSC sphere formation in all regressed for approximately 70 days following cessation of ther- lines tested with an IC50 range of 0.18 to 0.78 nmol/L, whereas apy, whereas tumor growth in the 5T4-Tub–treated group appears the tubulysin or auristatin warheads had little to no effect to resume even before completion of the full course of therapy. In (Supplementary Fig. S5). These data suggest that the DNA fact, the duration of the antitumor response of the 0.3 mg/kg dose cross-linking PBD dimer warhead could potently and specif- of 5T4-PBD was superior to the 5 mg/kg dose of 5T4-Tub. The 1 ically inhibit CSC proliferation, whereas tubulin inhibitors mg/kg dose of 5T4-Tub only resulted in slight tumor growth may not. inhibition. In the MDA-MB-361 xenograft model expressing high Next, we addressed whether 5T4-ADCs conjugated with levels of 5T4, both 5T4-Tub and 5T4-PBD achieved durable various payloads had differential effects on CSCs in vivo from regressions, although a lower dose and less frequent dosing of MDA-MB-361 breast carcinoma xenografts and NCI-N87 5T4-PBD was required to achieve such antitumor activity (Fig. gastric carcinoma xenografts. We first confirmed the CSC phe- 3C). Statistical significance (P < 0.05 based on two-way ANOVA) notype in each of the cell lines by determining tumorigenic was demonstrated for the 5T4-ADCs compared with controls. It potential of each of the putative CSC populations by perform- should also be noted that no overt signs of toxicity, such as ing limiting dilution assays. In this manner, we confirmed that þ significant body weight loss, were observed with either ADC at Aldefluor cells represented the tumorigenic population in the doses administered in these studies. both the MDA-MB-361 and NCI-N87 cell lines (Supplementary Table S3), and effects of these ADCs on in vivo CSCs were Anti-CSC activity of 5T4-PBD evaluated. Mice were inoculated with either MDA-MB-361 or To determine whether the enhanced in vivo efficacy with the NCI-N87 cells, and when the average tumor volume reached PBD-conjugated 5T4-PBD compared with the tubulysin-conju- approximately 500 mm3, tumor-bearing mice were treated with

Figure 2. In vitro characterization of 5T4_0108 and ADCs comprised of this antibody. A, Target-specific binding of 5T4_0108 to human breast cancer (MDA-MB-361), prostate cancer (DU 145), and gastric cancer (NCI-N87) cell lines that express 5T4 or to 5T4-negative DMS 114 lung carcinoma cell lines was determined by flow cytometry. B, Structures of 5T4-PBD consisting of the 5T4_0108 antibody site-specifically conjugated to the PBD payload SG3249, and 5T4-Tub consisting of the 5T4_0108 antibody conjugated to the tubulysin payload mc-Lys-MMETA. C, The 5T4-PBD and 5T4-Tub ADCs were tested against tumor cells with different levels of 5T4 expression on the cell surface. A correlation between target expression and cytotoxicity was observed by testing the ADCs against MDA-MB-361 breast carcinoma cells with high levels of 5T4 cell surface expression, DU 145 prostate cancer cells with moderate levels of 5T4 cell surface expression, and NCI-N87 gastric carcinoma cells with low levels of 5T4 cell surface expression. Representative experiments are shown, and the values indicate the mean SEM. Control ADCs consisting of the site-specific variant of the R347 IgG1 isotype control antibody conjugated to either PBD (Ctrl-PBD) or tubulysin (Ctrl-Tub) were also included to determine target-specificbinding.

OF6 Mol Cancer Ther; 16(8) August 2017 Molecular Cancer Therapeutics

MEDI0641: A Novel PBD-Based Anti-5T4 ADC

Figure 3. 5T4-PBD produces more durable responses in xenograft tumor models with varying levels 5T4 expression. 5T4-PBD and 5T4-Tub were tested in athymic nude mice bearing subcutaneous NCI-N87 (A), DU 145 (B), or MDA-MB-361 (C) xenografts. Following tumor cell inoculation, mice were size-matched into treatment groups, and therapeutic administration was initiated when the average tumor volume reached approximately 200 mm3. The administration of therapeutics in the study is indicated by the arrowheads, and the dotted lines represent average tumor volume at the time of treatment initiation; tumor volumes below this line are indicative of tumor regression. Values, mean SEM. Representative experiments are shown and demonstrate that while 5T4-PBD and 5T4-Tub are able to achieve tumor regressions in the MDA-MB-361 model with high 5T4 expression, 5T4-PBD produces more durable in vivo tumor regression than 5T4-Tub in the NCI-N87 and DU 145 models. Statistical signiﬁcance of 5T4-ADC treatment groups (P < 0.05) was demonstrated via two-way ANOVA analysis.

a single dose of 5T4-PBD, 5T4-Tub, or 5T4-mcMMAF (an anti- The data suggest that the superior antitumor activity of 5T4 antibody site-specifically conjugated with the auristatin 5T4-PBD could be due to the ability of this ADC to target þ þ mcMMAF with a DAR of 2). Once tumors began to regress (5– 5T4 CSCs in addition to the 5T4 bulk tumor cells, whereas 7 days in this model), the tumors were harvested and the CSC ADCs with tubulin inhibitors, such as 5T4-Tub, are ineffective populations were quantified. The PBD-conjugated 5T4-PBD at inhibiting CSCs. As 5T4 is a target that is expressed on both significantly reduced the number of CSCs by approximately CSCs as well as the non-CSCs within a tumor, it is critical 50% in both models and was more effective than either the to develop a 5T4-ADC conjugated with a payload capable tubulysin-conjugated 5T4-Tub or the auristatin-conjugated of effectively targeting CSCs to produce the most durable 5T4-mcMMAF (Fig. 4B). response.

www.aacrjournals.org Mol Cancer Ther; 16(8) August 2017 OF7

Harper et al.

Figure 4. 5T4-PBD effectively targets CSCs in vivo. A, 5T4 is expressed on CSC populations of various cancer cell lines. CSC markers are as follows: MDA-MB-361 breast cancer: Aldefluorþ; NCI-N87 gastric cancer: Aldefluorþ; DU 145 prostate cancer: CD44þCD24. B, ADC-treated tumors were harvested and the CD44þCD24 breast cancer CSCs and Aldefluorþ gastric carcinoma CSCs were counted. The number of CSCs found in the vehicle-treated control tumors was normalized to 100%, and the data for other treatment groups are reported as a percent of the CSC population found in control tumors. 5T4-PBD decreased the CSC population by approximately 50% in both models. 5T4-Tub had no effect on the MDA-MB-361 CSCs but had some effect on NCI-N87 CSCs, although this effect was not as potent as that with 5T4-PBD. Neither 5T4-mcMMAF nor controls had any effect on CSCs in either model. The 5T4-PBD–mediated decrease in CSCs was statistically significant compared with all groups in both models except the 5T4- Tub–treated group in the NCI-N87 model (P < 0.05 based on unpaired t tests).

Tolerability of MEDI0641 in rats consistent with the known off-target effects of PBD-conjugated Given the greater in vivo activity and anti-CSC activity ADCs in rats (43). observed with 5T4-PBD, this ADC was selected for further development and was designated as MEDI0641. The safety profile of the ADC was subsequently evaluated in a non-GLP Discussion rat toxicology study. Although the 5T4 antibody does not MEDI0641 is a 5T4-targeting ADC conjugated with a PBD crossreact with rodent 5T4, the rat was chosen as the initial payload that is capable of inducing significant antitumor activ- toxicity species to evaluate the off-target toxicity profile and ity, including durable regressions in tumor models with varying assess the in vivo stability of the molecule. Male rats were intensity and heterogeneity of 5T4 expression. Although administered a single intravenous dose of 1 or 2 mg/kg of 5T4-Tub also demonstrated significant in vivo antitumor activ- MEDI0641 on day 1 and necropsied on days 8 and 29 to ity, MEDI0641 produced a more durable response possibly evaluate acute and delayed toxicities. All animals survived due to the fact that it had a more pronounced effect on CSCs until scheduled necropsy. Major findings were dose-dependent than 5T4-Tub. We and others have shown that 5T4 is expressed decrease in body weight gain (up to 10%) and dose-dependent on CSCs that are hypothesized to be responsible for chemo- bone marrow suppression, with major effects on reticulocytes, therapeutic resistance and recurrence of cancer (21, 22). Using platelets, and white blood cells (Fig. 5). All findings were multiple HNSCC patient-derived xenograft (PDX) models, reversible or trending toward reversible by day 29 and are Kerk and colleagues report that MEDI0641 was able to reduce,

OF8 Mol Cancer Ther; 16(8) August 2017 Molecular Cancer Therapeutics

MEDI0641: A Novel PBD-Based Anti-5T4 ADC

Figure 5. Safety proﬁle of MEDI0641 in rats. Rats (n ¼ 12) were treated with a single intravenous injection of 1 or 2 mg/kg MEDI0641. Control animals were treated with a single intravenous injection of vehicle control. A, Tolerability was assessed by measuring body weight changes over a 29-day period. Data points represent the mean value of each group at individual time points; error bars, mean SEM. Tx, treatment. B–D, MEDI0641 is associated with dose- dependent bone marrow suppression. Hematologic analyses were conducted on peripheral blood samples collected on days 8, 14, and 29 posttreatment to enumerate platelet counts (B), reticulocytes (C), and white blood cells (D). Bars represent the mean value of each group; error bars, mean SEM.

and in one model ablate, CSCs in these models, corroborating auristatin payload had no effect on CSC populations in vivo in and extending the anti-CSC properties of MEDI0641 that we either xenograft model, and 5T4-Tub was either ineffective or less describe here. Like us, they also observed that 5T4 expression effective against CSCs compared with MEDI0641 in the was increased in the CSC population of these PDX tumors MDA-MB-361 and NCI-N87 xenograft models, respectively. compared with 5T4 expression in non-CSC tumor cells. Here, Previously, an anti-5T4 ADC conjugated with the auristatin pay- we extend these findings into additional tumor types with load mcMMAF (A1-mcMMAF; PF-06263507) was reported to significant prevalence and intensity of 5T4 expression in clinical have potent antitumor activity in vivo and also was reported to samples, including breast cancer, gastric cancer, pancreatic inhibit CSC-like tumor-initiating cells in a NSCLC PDX model cancer, and prostate cancer. Collectively, the strong expression (23). There could be several explanations for the conflicting data and/or upregulation of 5T4 on CSCs suggest that perhaps 5T4 with the MMAF-conjugated ADCs: (i) The 5T4-mcMMAF ADC may have some functional relevance to CSC biology. Further used in the current study was a site-specific conjugate with a DAR studies are warranted to determine such a role for 5T4 in CSCs. of 2, as opposed to the classically conjugated A1-mcMMAF that It has been hypothesized that tubulin inhibitors, such as the has a DAR of approximately 4; (ii) differences in dosing schedule tubulysins, auristatins, and maytansinoids, are most effective (single administration here, every four days for four total doses þ against tumor cells with high proliferative rates and are less reported); (iii) different CSC tumor models were used in each set effective against relatively quiescent cells, such as CSCs of studies and each could respond to auristatins differently; and (23, 44). DNA-targeting agents on the other hand, particularly (iv) different methodologies were used to assess anti-CSC activity those that form cross-links or other DNA lesions that are not easily in vivo in each report. Testing our 5T4-ADCs in models and assays repaired, could more effectively target CSCs. The data presented similar to those reported where A1-mcMMAF demonstrated anti- here support this hypothesis. 5T4-mcMMAF conjugated with an CSC activity, and vice versa, could help resolve the contrary

www.aacrjournals.org Mol Cancer Ther; 16(8) August 2017 OF9

Harper et al.

observations. Regardless, the current data clearly demonstrate that MEDI0641, particularly in a species where our anti-5T4 anti- the PBD-conjugated MEDI0641 had superior anti-CSC effects body cross-reacts, is warranted as safety studies with this ADC over either tubulysin- or auristatin-conjugated ADCs. proceed. The different anti-CSC activities observed between MEDI0641 In conclusion, our generation and testing of the PBD-con- and 5T4-Tub could potentially account for the different antitu- jugated MEDI0641 and 5T4-Tub permitted a direct comparison mor responses elicited by these ADCs. MEDI0641 significantly of payloads with different mechanisms of action conjugated to decreased the CSC population in both the MDA-MB-361 and the same antibody against 5T4. Our results demonstrated the NCI-N87 models, and in each model, durable regressions MEDI0641 had overall superior activity and, in particular, is were observed following MEDI0641 treatment. In the NCI-N87 more effective at targeting 5T4-positive CSCs, which may pre- model, even though higher doses and frequency were utilized, vent or reduce tumor recurrence. These data provide motivation 5T4-Tub had inferior anti-CSC activity and less durable antitu- for further development of this promising ADC. mor activity compared with MEDI0641. Interestingly, each ADC was capable of eliciting regressions in the MDA-MB-361 model Disclosure of Potential Conflicts of Interest even though only MEDI0641 was able to inhibit the in vivo F. D'Hooge is the head of the Bioconjugation Unit at Novasep. CSC population. This could be attributed to the fact that 5T4 fi L. van Vlerken-Ysla is a scientist at MedImmune.R.E.Hollingsworthisthe expression was signi cantly higher in both bulk tumor cells and senior director (Oncology) at MedImmune. No potential conflicts of interest CSCs in the MDA-MB-361 model compared with NCI-N87, were disclosed by the other authors. suggesting that, hypothetically, more of each ADC would be delivered to MDA-MB-361 tumors compared with NCI-N87 tumors. It could also be due to dosing: a single administration Authors' Contributions in the CSC experiment, while 5T4-Tub was administered once a Conception and design: J. Harper, C. Lloyd, D. Toader, R. Fleming, C. Chen, fi H. Zhong, E.M. Hurt, P.W. Howard, R.E. Hollingsworth, A. Kamal week for a total of four doses in the ef cacy study. Taken Development of methodology: J. Harper, C. Lloyd, R. Marwood, L. Lewis, together, these data suggest that greater success of a 5T4-ADC D. Bannister, J. Jovanovic, R. Fleming, S. Mao, L. Xu, M. Rebelatto, E.M. Hurt, in tumors with lower levels of 5T4 expression may be dependent P.W. Howard, R.E. Hollingsworth on the ability of that ADC to deliver a payload that effectively Acquisition of data (provided animals, acquired and managed patients, targets the 5T4-positive CSC population. provided facilities, etc.): J. Harper, C. Lloyd, N. Dimasi, R. Marwood, L. Lewis, Evaluating ADCs that target the same surface antigen but J. Jovanovic, R. Fleming, F. D'Hooge, S. Mao, A.M. Marrero, M. Korade III, C. Chen, L. Wetzel, S. Breen, L. van Vlerken-Ysla, M. Rebelatto, E.M. Hurt deliver warheads with different mechanisms of action helped to Analysis and interpretation of data (e.g., statistical analysis, biostatistics, potentially identify which ADC could produce a more durable computational analysis): J.Harper,C.Lloyd,N.Dimasi,R.Marwood, response. To our knowledge, this is the first report where a head- L.Lewis,J.Jovanovic,R.Fleming,F.D'Hooge,S.Mao,P.Strout,C.Chen, to-head comparison was made between two ADCs comprised of S.Breen,L.vanVlerken-Ysla,S.Jalla,M.Rebelatto,H.Zhong,E.M.Hurt, the same antibody but conjugated with payloads from different D.A. Tice, R.E. Hollingsworth drug classes with different mechanisms of action: PBDs that Writing, review, and/or revision of the manuscript: J. Harper, C. Lloyd, N. Dimasi, D. Toader, R. Marwood, L. Lewis, J. Jovanovic, R. Fleming, S. Mao, form interstrand DNA cross-links and tubulysins that depoly- A.M. Marrero, M. Korade III, P. Strout, C. Chen, L. Wetzel, M. Rebelatto, merize microtubules. Others have reported comparing cleav- H. Zhong, E.M. Hurt, M.J. Hinrichs, P.W. Howard, D.A. Tice, R. Herbst, A. Kamal able versus noncleavable drug linkers, but typically, these Administrative, technical, or material support (i.e., reporting or organizing comparisons have included the same warhead class; for data, constructing databases): J. Harper, N. Dimasi, R. Marwood, L. Lewis, example, comparing conjugates with either cleavable K. Huang mc-VC-PABC-MMAE or noncleavable mc-MMAF (45, 46), or Study supervision: D. Toader, H. Zhong, E.M. Hurt, R.E. Hollingsworth, A. Kamal cleavable SPDB-DM4 versus noncleavable SMCC-DM1 Other (made the 5T4 protein to enable antibody generation and helped (47, 48). There have also been reports of comparing conjugates with biophysical analysis of the ADC candidates): D. Bannister with auristatins and maytansinoids (49), but these are both Other (designing and performing the experiments): S. Jalla inherently two classes of drugs with a similar mechanism of Other (design and synthesis and supply of drug-linker portion of antibody– action, namely microtubule inhibition. drug conjugate): P.W. Howard Prior to developing A1-mcMMAF, the same research group had evaluated a calicheamicin-conjugated 5T4-ADC, however, Acknowledgments composed of a different antibody against 5T4. They found that The authors wish to thank Jean-Noel Levy for his work on synthesizing the the ADC bearing the calicheamicin payload was not tolerated at PBD payloads used during the course of these studies. In addition, we thank the efficacious exposure levels in cynomolgus monkeys (23, 50). Biological Expression Team at MedImmune, Ltd. (Cambridge, United King- Although also a DNA-targeting agent, the mechanism of action dom) for their work in synthesizing the antibodies for use in these studies. of calicheamicin is different from that of PBDs, and the drivers The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked of the toxicity with the 5T4-calicheamicin were not reported, so advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate it is not clear whether they could be attributed to the different this fact. antibody and/or the different payload. Non-GLP rat toxicology studies with MEDI0641 demonstrated a reasonable safety pro- Received December 1, 2016; revised March 28, 2017; accepted April 28, 2017; file. Regardless, close attention to the toxicologic findings with published OnlineFirst May 18, 2017.

References 1. Polakis P. Antibody drug conjugates for cancer therapy. Pharmacol Rev 2. De Goeij BECG, Lambert JM. New developments for antibody-drug con- 2016;68:3–19. jugate-based therapeutic approaches. Curr Opin Immunol 2016;40:14–23.

OF10 Mol Cancer Ther; 16(8) August 2017 Molecular Cancer Therapeutics

MEDI0641: A Novel PBD-Based Anti-5T4 ADC

3. Diamentis N, Banerji U. Antibody-drug conjugates – an emerging class of jugate, PF-06263507, in patients with advanced solid tumors. Invest New cancer treatment. Br J Cancer 2016;114:362–7. Drugs 2017;35:315–23. 4. Hole N, Stern PL. A 72kD trophoblast glycoprotein defined by a mono- 26. Hartley J. The development of pyrrolobenzodiazepines as antitumor clonal antibody. Br J Cancer 1990;57:239–46. agents. Expert Opin Investig Drugs 2011;20:733–44. 5. Hole N, Stern PL. Isolation and characterization of 5T4, a tumor-associated 27. Tiberghien AC, Levy J-N, Masterson LA, Patel NV, Adams LR, Corbett S, antigen. Int J Cancer 1990;45:179–84. et al. Design and synthesis of tesirine, a clinical antibody–drug conju- 6. Southall PJ, Boxer GM, Bagshawe KD, Hole N, Bromley M, Stern PL. gate pyrrolobenzodiazepine dimer payload. ACS Med Chem Lett 2016; Immunohistological distribution of 5T4 antigen in normal and malignant 7:983–7. tissue. Br J Cancer 1990;61:89–95. 28. Toader D, Wang F, Gingipalli L, Vasbinder MM, Roth M, Mao S, et al. 7. Starzynska T, Rahi V, Stern PL. The expression of 5T4 antigen in colorectal Structure-cytotoxicity relationships of analogs of N14-desacetocytubulysin and gastric carcinoma. Br J Cancer 1992;66:867–69. H. J Med Chem 2016;59:10781–7. 8. Wrigley E, McGown AT, Rennison J, Swindell R, Crowther D, Starzynska T, 29. Murphy AJ, Macdonald LE, Stevens S, Karow M, Dore AT, Pobursky K, et al. et al. 5T4 oncofetal antigen expression in ovarian cancer. Int J Gynecol Mice with megabase humanization of their immunoglobulin genes gen- Cancer 1995;5:269–74. erate antibodies as efficiently as normal mice. Proc Natl Acad Sci U S A 9. Griffiths RW, Gillham DE, Dangoor A, Ramani V, Clarke NW, Stern PL, 2014;111:5153–8. et al. Expression of the 5T4 oncofetal antigen in renal cell carcinoma: 30. Douthwaite JA, Sridharan S, Huntington C, Hammersley J, Marwood R, a potential target tor T-cell-based immunotherapy. Br J Cancer 2005;93: Hakulinen JK, et al. Affinity maturation of a novel antagonistic human 670–7. monoclonal antibody with a long VH CDR3 targeting the Class A GPCR 10. Al-Taei S, Salimu J, Lester JF, Linnane S, Goonewardena M, Harrop formyl-peptide receptor 1. MAbs 2015;7:152–66. R, et al. Overexpression and potential targeting of the oncofetal 31. Dennis MS, Lowman HB. Phage selection strategies for improved affinity antigen 5T4 in malignant pleural mesothelioma. Lung Cancer 2012; and specificity of proteins and peptides. In:Clackson T, Lowman HB, 77:312–8. editors. Phage display, a practical approach. Oxford, United Kingdom: 11. Mulder WMC, Stern PL, Stukart MJ, de Windt E, Butzelaar RMJM, Meijer S, Oxford University Press; 2004. p. 61–84. et al. Low intracellular adhesion molecule 1 and high 5T4 expression on 32. Thompson J, Pope T, Tung JS, Chan C, Hollis G, Mark G, et al. Affinity tumor cells correlate with reduced disease-free survival in colorectal car- maturation of a high-affinity human monoclonal antibody against the cinoma patients. Clin Cancer Res 1997;3:1923–30. third hypervariable loop of human immunodeficiency virus: use of phage 12. Starzynska T, Marsh PJ, Schofield PF, Roberts SA, Myers KA, Stern PL. display to improve affinity and broaden strain reactivity. J Mol Biol Prognostic significance of 5T4 oncofetal antigen expression in colorectal 1996:256:77–88. carcinoma. Br J Cancer 1994;69:899–902. 33. Osbourn JK, Field A, Wilton J, Derbyshire E, Earnshaw JC, Jones PT, et al. 13. Starzynska T, Wiechowska-Kozlowska A, Marlicz K, Bromley M, Roberts SA, Generation of a panel of related human scFV antibodies with high affinities Lawniczak M, et al. 5T4 oncofetal antigen in gastric carcinoma and its for human CEA. Immunotechnology 1996;2:181–96. clinical significance. Eu J Gastroenterol Hepatol 1998;10:479–84. 34. Thompson P, Fleming R, Bezabeh B, Huang F, Mao S, Chen C, et al. 14. Myers KA, Rahi-Saund V, Davison MD, Young JA, Cheater AJ, Stern PL. Rational design, biophysical and biological characterization of site-specific Isolation of a cDNA encoding 5T4 oncofetal trophoblast glycoprotein. antibody-tubulysin conjugates with improved stability, efficacy and phar- J Biol Chem 1994;269:9319–24 macokinetics. J Control Release 2016;236:100–16. 15. Southgate TD, McGinn OJ, Castro FV, Rutkowski AJ, Al-Muftah M, Marinov 35. Edelman GM, Cunningham BA, Gall WE, Gottlieb PD, Rutishauser U, G, et al. CXCR4 mediated chemotaxis is regulated by 5T4 oncofetal Waxdal MJ. The covalent structure of an entire gamma G immunoglobulin glycoprotein in mouse embryonic cells. PLoS One 2010;5:e9982. molecule. Proc Natl Acad Sci U S A 1969;63:78–85. 16. Yoshihara S-i, Takahashi H, Tsuboi A. Molecular mechanisms regu- 36. Thornberry NA, Rano TA, Peterson EP, Rasper DM, Timkey T, Garcia-Calvo lating the dendritic development of newborn olfactory bulb interneur- M, et al. A combinatorial approach defines specificities of members of the ons in a sensory-experience-dependent manner. Front Neurosci 2016;9: caspase family and granzyme B. Functional relationships established for Article 514. key mediators of apoptosis. J Biol Chem 1997;272:17907–11. 17. Zhao Y, Malinauskas T, Harlos K, Jones EY. Structural insights into the 37. Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. inhibition of Wnt signaling by cancer antigen 5T4/Wnt-activated inhibi- Prospective identification of tumorigenic breast cancer cells. Proc Natl tory factor 1. Structure 2014;22:612–20. Acad Sci U S A 2003;100:3983–8. 18. Awan A, Lucic MR, Shaw DM, Sheppard F, Westwater C, Lyons SA, et al. 5T4 38. de Beca¸ FF, Caetano P, Gerhard R, Alvarenga CA, Gomes M, Paredes J, et al. interacts with TIP-2/GIPC, a PDZ protein, with implications for metastasis. Cancer stem cells markers CD44, CD24 and ALDH1 in breast cancer special Biochem Biophys Res Commun 2002;290:1030–6. histological types. J Clin Pathol 2013;66:187–91. 19. Carsberg CJ, Myers KA, Stern PL. Metastasis-associated 5T4 antigen disrupts 39. Hurt EM, Kawasaki BT, Klarmann GJ, Thomas SB, Farrar WL. CD44þCD24 cell-cell contacts and induces cellular motility in epithelial cells. Int (-) prostate cells are early cancer progenitor/stem cells that provide a model J Cancer 1996;68:84–92. for patients with poor prognosis. Br J Cancer 2008;98:756–65. 20. Spencer HL, Eastham AM, Merry CLR, Southgate TD, Perez-Campo F, 40. Yang ZF, Ho DW, Ng MN, Lau CK, Yu WC, Ngai P, et al. Significance of Soncin F, et al. E-cadherin inhibits cell surface localization of the pro- CD90þ cancer stem cells in human liver cancer. Cancer Cell 2008;13: migratory 5T4 oncofetal antigen in mouse embryonic stem cells. Mol Biol 153–66. Cell 2007;18:2838–51. 41. Katsuno Y, Ehata S, Yashiro M, Yanagihara K, Hirakawa K, Miyazono K. 21. Damelin M, Geles KG, Follettie MT, Yuan P, Baxter M, Golas J, et al. Coordinated expression of REG4 and aldehyde dehydrogenase 1 regulating Delineation of a cellular hierarchy in lung cancer reveals an oncofetal tumourigenic capacity of diffuse-type gastric carcinoma-initiating cells is antigen expressed on tumor-initiating cells. Cancer Res 2011;71:4236–46. inhibited by TGF-b. J Pathol 2012;228:391–404. 22. Kerk SA, Finkel KA, Pearson AT, Warner K, Nor F, Zhang Z, et al. 5T4- 42. Zhu J, He J, Liu Y, Simeone DM, Lubman DM. Identification of targeted therapy ablates cancer stem cells and prevents recurrence of glycoprotein markers for pancreatic cancer CD24þCD44þ stem-like head and neck squamous cell carcinoma. Clin Cancer Res 2017;23: cells using nano-LC-MS/MS and tissue microarray. J Proteome Res 2012; 2516–27. 11:2272–81. 23. Sapra P, Damelin M, DiJoseph J, Marquette K, Geles KG, Golas J, et al. Long- 43. JeffreySC,BurkePJ,LyonRP,MeyerDW,SussmanD,AndersonM,etal. term tumor regression induced by an antibody-drug conjugate that targets A potent anti-CD70 antibody–drug conjugate combining a dimeric 5T4, an oncofetal antigen expressed on tumor-initiating cells. Mol Cancer pyrrolobenzodiazepine drug with site-specific conjugation technology. Ther 2013;12:38–47. Bioconjug Chem 2013;24:1256–63. 24. Shor B, Kahler J, Dougher M, Xu J, Mack M, Rosfjord E, et al. Enhanced 44. Vidal SJ, Rodriquez-Bravo V, Galsky M, Cordon-Cardo C, Domingo- antitumor activity of an anti-5T4 antibody-drug conjugate in combination Domenech J. Targeting cancer stem cells to suppress acquired chemother- with PI3K/mTOR inhibitors or taxanes. Clin Cancer Res 2015;22:383–94. apy resistance. Oncogene 2014;33:4451–63. 25. Shapiro GI, Vaishampayan UN, LoRusso P, Barton J, Hua S, Reich SD, et al. 45. Scales SJ, Gupta N, Pacheco G, Firestein R, French DM, Koeppen H, et al. An First-in-human trial of an anti-5T4 antibody-monomethylauristatin con- antimesothelin-monomethyl auristatin e conjugate with potent antitumor

www.aacrjournals.org Mol Cancer Ther; 16(8) August 2017 OF11

Harper et al.

activity in ovarian, pancreatic, and mesothelioma models. Mol Cancer Ther 48. Nguyen M, Miyakawa S, Kato J, Mori T, Arai T, Armanini M, et al. Preclinical 2014;13:2630–40. efﬁcacy and safety assessment of an antibody-drug conjugate targeting the 46. Breij EC, de Goeij BE, Verploegen S, Schuurhuis DH, Amirkhosravi A, c-RET proto-oncogene for breast carcinoma. Clin Cancer Res 2015;21: Francis J, et al. An antibody-drug conjugate that targets tissue factor exhibits 5552–62. potent therapeutic activity against a broad range of solid tumors. Cancer 49. Mang Y, Zhao Z, Zeng Z, Wu X, Li Z, Zhang L. Efﬁcient elimination of Res 2014;74:1214–26. CD103-expressing cells by anti-CD103 antibody drug conjugates in immu- 47. Phillips GD, Li G, Dugger D, Crocker LM, Parsons KL, Mai E, nocompetent mice. Int Immunopharmacol 2015;24:119–27. et al. Targeting HER2-positive breast cancer with trastuzumab- 50. Boghaert ER, Sridharan L, Khandke KM, Armellino D, Ryan MG, Myers K, DM1, an antibody–cytotoxic drug conjugate. Cancer Res 2008;68: et al. The oncofetal protein, 5T4, is a suitable target for antibody-guided anti- 9280–90. cancer chemotherapy with calicheamicin. Int J Oncol 2008;32:221–34.

OF12 Mol Cancer Ther; 16(8) August 2017 Molecular Cancer Therapeutics

Preclinical Evaluation of MEDI0641, a Pyrrolobenzodiazepine-Conjugated Antibody−Drug Conjugate Targeting 5T4

Jay Harper, Christopher Lloyd, Nazzareno Dimasi, et al.

Mol Cancer Ther Published OnlineFirst May 18, 2017.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-16-0825

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2017/05/18/1535-7163.MCT-16-0825.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/early/2017/07/17/1535-7163.MCT-16-0825. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.