Abstract
Purpose: Chemotherapies are limited by a narrow therapeutic index resulting in suboptimal exposure of the tumor to the drug and acquired tumor resistance. One approach to overcome this is through antibody–drug conjugates (ADC) that facilitate greater potency via target-specific delivery of highly potent cytotoxic agents.
Experimental Design: In this study, we used a bioinformatics approach to identify the lymphocyte antigen 6 complex locus E (LY6E), an IFN-inducible glycosylphosphatidylinositol (GPI)-linked cell membrane protein as a promising ADC target. We developed a monoclonal anti-LY6E antibody and characterized in situ LY6E expression in over 750 cancer specimens and normal tissues. Target-dependent anti-LY6E ADC killing was investigated both in vitro and in vivo using patient-derived xenograft models.
Results: Using in silico approaches, we found that LY6E was significantly overexpressed and amplified in a wide array of different human solid tumors. IHC analysis revealed high LY6E protein expression in a number of tumor types, such as breast, lung, gastric, ovarian, pancreatic, kidney and head/neck carcinomas. Characterization of the endocytic pathways for LY6E revealed that the LY6E-specific antibody is internalized into cells leading to lysosomal accumulation. Consistent with this, a LY6E-specific ADC inhibited in vitro cell proliferation and produced durable tumor regression in vivo in clinically relevant LY6E-expressing xenograft models.
Conclusions: Our results identify LY6E as a highly promising molecular ADC target for a variety of solid tumor types with current unmet medical need. Clin Cancer Res; 21(14); 3252–62. ©2015 AACR.
Translational Relevance
Targeted cancer therapies exploit molecular alterations specific to the cancer cell, thus mitigating damage to normal tissues. Antibody–drug conjugates (ADC) utilize a monoclonal antibody for efficient and specific delivery of a cytotoxic agent. This new class of therapeutics has recently shown promising clinical activity in targeting tumors which harbor overexpressed or amplified (e.g., T-DM1) receptors on the cancer cell surface. In this study, we identify and characterize LY6E as a particularly attractive ADC target based on its high expression in numerous human carcinomas, minimal expression in normal tissues, cell-surface localization, and swift endocytosis. In summary, our preclinical results suggest that the ADC targeting the LY6E may provide an effective new therapy for a wide range of solid tumor malignancies.
Introduction
Traditional chemotherapeutic regimens are widely used in cancer therapy. Although efficacious, their use is limited by their systemic toxic effects in patients leading to a narrow therapeutic window. One tactic to circumvent this limitation is the antibody–drug conjugate (ADC) approach. ADCs are composed of three modular elements: (i) a monoclonal antibody that targets antigens selectively expressed at the surface of tumor cells; (ii) a linker moiety; and (iii) a cytotoxic agent that mediates cell killing. In principle, the ADC is systemically inert and functions as a prodrug, releasing its cytotoxic agent only after target engagement, internalization, and subsequent proteolytic digestion in the cancer cell. The recent approval of two ADCs, trastuzumab emtansine (T-DM1; Kadcyla) targeting HER2-positive breast cancer and brentuximab vedotin for CD30-positive lymphomas (1, 2) has led to a burgeoning interest in developing this technology for additional indications.
The selection of an appropriate tumor antigen is of central importance to the activity of an ADC molecule. The ideal ADC target should be highly and broadly expressed in cancer while absent in normal tissues. Recent efforts by The Cancer Genome Atlas (TCGA) and other groups have empowered the systematic characterization of molecular alterations in several key human malignancies. These investigations have led to new opportunities to identify or reexamine cancer-specific antigens that may serve as a springboard for a new generation of ADCs.
LY6E is a glycosylphosphatidylinositol (GPI)-linked cell-surface protein that is induced by IFNα (3) and chemotherapy (4). LY6E mRNA was found to be overexpressed in colon and kidney cancer suggesting a role in cancer (5). In this report, we identify and characterize LY6E as an ADC target in a broad set of human malignancies. We find that the LY6E locus is amplified and the protein is broadly and highly expressed in diverse cancer types. Of note, LY6E is overexpressed in several cancers with limited therapeutic options, including ovarian, pancreatic, lung, gastric, and triple-negative [estrogen receptor (ER−), progesterone receptor (PR−), and (HER2−) breast cancer (TNBC). To pursue LY6E as a cancer target, we developed an anti-LY6E monomethylauristatin E (MMAE) ADC that demonstrated potent efficacy both in vitro and in vivo in different LY6E-expressing tumor models, including patient-derived xenografts (PDX) with heterogeneous expression. Our preclinical data suggest that anti-LY6E ADC has great potential as a therapeutic agent for many cancers of unmet medical need in the clinical setting.
Materials and Methods
Tissue culture and cell lines
Human ovarian cancer cell lines A2780, COV318, COV362, DOV13, EFO-27, KURAMOCHI, OVCAR433, OVISE, OAW42, RMG-1, TOV-112D; human pancreatic cancer cell line SU.86.86; human squamous non–small cell lung cancer (NSCLC) cell line SW900; human malignant lymphoma cell line RAMOS; human breast cancer cell lines HCC1569 and MDA-MB-175 VII; human embryonic kidney cell line HEK-293; and human prostatic cell line PC-3 were obtained from cell bank collections such as ECACC (European Collection of Cell cultures), JCRB (National Institute of Biomedical Innovation), DSMZ (German collection of micro-organisms and cell cultures), ATCC, or from the M.D. Anderson Cancer Center (Houston, TX). Cell lines were tested and authenticated by STR profiling at Genentech, Inc.
The cell line PC-3-TVA is derived from the human prostatic cell line, PC-3 to ectopically express the tv-a gene, the receptor for avian leukosis virus subgroup A, which allows for viral entry and chromosomal integration of the gene of interest.
Human LY6E (encoded by NM_002346) and cynomolgus monkey LY6E (encoded by NM_002346.1) were cloned into a cytomegalovirus mammalian expression vector engineered to encode an N-terminal gD (HSV-1 viral glycoprotein) tag. These constructs were used in competition-binding assays to determine affinity of anti-LY6E antibody to human and cynomolgus monkey exogenous LY6E expressed transiently in HEK-293 cells (detailed protocol for affinity assays is provided in the Supplementary Methods section).
Human LY6E construct was subcloned into RCASBP retroviral expression vector and used to generate viral particles expressing the gene of interest in DF-1 cells. PC-3-TVA-N-term-tagged gD human LY6E (PC3-LY6E) was generated as a pool of cells stably expressing LY6E, by infecting PC-3-TVA cell line with viral particles expressing N-term-tagged human LY6E. Appropriate vector control cells (PC3-Vector) were also generated.
The Genentech in-house cell bank acquired cell lines over a period of more than 10 years. Cell lines were maintained and propagated as described previously (6).
Generation of anti-LY6E antibodies
Anti-LY6E rabbit polyclonal antibody GEN-93-8-1 (8-1) was developed through a fee-for-service agreement with YenZym. The anti-LY6E rabbit monoclonal antibody, GEN-93-8-1 was developed by Epitomics (Abcam), from a rabbit originating from the YenZym collaboration.
Anti-LY6E antibody 9B12 was humanized from its parental hybridoma clone; which was generated by immunizing Balb/c mice with purified LY6E protein (7). Immunization, hybridoma growth and selection, and antibody purification methods used were as described previously (8). Anti-LY6E antibody 9B12 was conjugated to auristatin MMAE by methods previously described (9).
The drug to antibody ratio for anti-LY6E ADC (LY6E) was 3.6 to 3.8 and the drug to antibody ratio for control IgG-vc-MMAE (Ctrl) was 3.0 to 3.5.
Immunologic procedures
FACS using 5 μg/mL anti-LY6E antibody 9B12 to detect LY6E and cell viability assays were performed as described previously (10). Internalization of anti-LY6E 9B12 antibody was evaluated using protocols as previously described (10). Ten thousand KURAMOCHI cells were seeded onto each well of 4-well chamber slides. Five days later, cells were washed and incubated for 30 minutes, either in the presence or absence of endocytic inhibitors/disruptors, 25 μmol/L chlorpromazine (CPZ) or 1 mmol/L methyl β cyclodextrin (MBC; C8138, C4555, Sigma Life Sciences). Cells were further incubated for an additional 30 minutes with 5 μg/mL anti-LY6E antibody 9B12 or with 25 μg/mL Transferrin directly conjugated to Alexa 555 (T-35352, Invitrogen). Incubation of cells with inhibitors, 9B12 antibody, transferring, and the appropriate negative controls were conducted in growth media containing appropriate protease inhibitors. Cells were permeabilized, washed, and incubated for 60 minutes with 4 μg/mL rabbit antibody against lysosomal marker 1 (LAMP1; L1418, Sigma Life Sciences; ref. 10). LY6E antibodies 9B12 and LAMP1 were detected by incubating cells with the appropriate secondary fluorescent detection reagents for 60 minutes.
Slides were mounted by applying ProLong gold antifade reagent with DAPI (P36935; Invitrogen) placed under a glass coverslip, and sealed with clear nail polish. Images were acquired on a Nikon TE-300 inverted microscope equipped with a ×60 magnification 1.4 NA infinity-corrected Plan Apo oil objective (Technical Instruments) and a Retiga EX-cooled CCD camera (Q Imaging), using the QCapture version 3.1.1 software application (Q Imaging). Image analysis was performed using Photoshop CS software (version 8.0; Adobe Systems).
RNAi
For gene depletion studies, PC3-LY6E cells were transiently transfected for 48 hours with LY6E-specific siRNA (see Supplementary Table S1 for sequences) using lipofectamine RNAiMAX transfection reagent (13778150, Invitrogen) and methods suggested by the manufacturer.
Gene expression and IHC
For qRT-PCR, 200 ng RNA was amplified using the Invitrogen Platinum Taq/Reverse Transcriptase enzyme mix as per the manufacturer's protocol (Life Technologies). qPCR was conducted on Fluidigm 96.96 Dynamic arrays using the Biomark HD system as per the manufacturer's protocol. Samples were run in triplicate and cycle threshold (Ct) values were converted to relative expression values (negative Δ Ct) by subtracting the mean of the two reference genes (Transmembrane Protein 55B (TMEM55B) and Vacuolar Protein Sorting Homolog B (VPS33B) from the mean of target gene.
For normal tissue expression studies, Tissue Scan Normal Tissue cDNA arrays for qPCR (HMRT103, OriGene Technologies) and RNA extracted from various cancer tissues and xenografts were probed for LY6E and GAPDH expression in triplicate on a 7500 Real Time PCR thermal cycler, (Life Technologies) using manufacturer's reagents. Fluorophore-labeled probes used with flanking primers for the detection of various genes are listed in Supplementary Table S1.
For immunoblotting, 1 μg/mL anti-LY6E rabbit monoclonal antibody GEN-93-8-1 was used to detect LY6E and loading control β-actin was detected using mouse monoclonal antibody (Clone BH10D10, MA5-15452, Pierce, Thermo Scientific). Detection reagents Alexafluor 680 anti-rabbit IgG, (A21076, Invitrogen) and IRDye800-conjugated anti-mouse IgG (610-132-121, Rockland Immunochemicals) were used.
IHC was performed on a Ventana Discovery XT autostainer (Ventana Medical Systems). Formalin-fixed, paraffin-embedded whole tissue and tissue microarray sections were deparaffinized and pretreated with CC1 solution (Ventana Medical Systems) for 60 minutes followed by incubation with either 0.2 μg/mL anti-LY6E rabbit monoclonal antibody or naïve rabbit IgG (Clone DA1E, Cell Signaling Technologies) for 60 minutes at 37°C. Detection was performed with 32-minute OmniMap anti-rabbit horseradish peroxidase (HRP) incubation and diaminobenzidine (DAB; Ventana Medical Systems) followed by counter staining with hematoxylin II (Ventana Medical Systems).
Staining intensity scores took into account both intensity of staining and proportion of labeled cells. Positive expression was defined as staining in greater than 50% of the tumor cells. Intensity scores are defined below:
Score | Definition |
Negative | No detectable signal in >50% of tumor cells |
1 + | Positive signal in >50% of tumor cells with majority of signal = weak |
2 + | Positive signal in >50% of tumor cells with majority of signal = moderate |
3 + | Positive signal in >50% of tumor cells with majority of signal = strong |
Xenograft efficacy studies
Efficacy of anti-LY6E ADC (9B12) or T-DM1 was evaluated in xenograft models established by implanting patient-derived tumor material or cancer cells subcutaneously in the dorsal right flank of immunodeficient mice. When mean tumor volumes reached approximately 90 to 200 mm3 (day 0), animals were randomized into groups of 8 to 10 each and administered a single intravenous (i.v.) injection of either humanized anti-LY6E antibody or ADC or T-DM1 or anti IgG (human isotype) ADC (Ctrl) conjugated to MMAE through the valine-citrulline linker (9). Animal weights and tumor volumes were measured twice per week until study end and all calculations were done as described previously (11).
For cell line xenograft studies, naïve C.B-17 SCID or C.B-17 SCID beige mice (Charles River Laboratories) were inoculated with either two million SU.86.86 cells suspended in HBSS and Matrigel (BD Biosciences) or 5 million SW900 cells suspended in HBSS, respectively. All studies were conducted in accordance with the Guide for the Care and Use of Laboratory Animals [Ref: Institute of Laboratory Animal Resources (NIH publication no. 85-23), Washington, DC: National Academies Press; 1996].
PAXF 1657 pancreatic cancer PDX model and MAXF 1162 (HER2-positive) primary breast cancer PDX model were established at Oncotest GmbH. All experiments were approved by the local authorities, and were conducted according to the guidelines of the German Animal Welfare Act (Tierschutzgesetz). Four- to 6-week-old female immunodeficient NMRI nu/nu mice from Harlan or Charles River were used.
Triple-negative breast cancer ductal adenocarcinoma PDX model HBCx-9 was established at XenTech. All experiments were performed in accordance with French legislation concerning the protection of laboratory animals and in accordance with a currently valid license for experiments on vertebrate animals, issued by the French Ministry for Agriculture and Fisheries. Fourteen-week-old female athymic nude mice (Hsd: Athymic Nude-Fox1nu) obtained from Harlan Laboratories were used.
Cancer genomics analysis
For the analysis of LY6E mRNA expression, we used RNAseq data from TCGA and Molecular Taxonomy of Breast Cancer International Consortium (METABRIC). TCGA RNA-seq data were obtained from the Cancer Genomics Hub at UC Santa Cruz and processed and aligned with HTSeqGenie. Expression levels were quantified by RNASeq via Illumina sequencing (75 base paired-end reads). Reads were filtered for quality and for rRNA contamination. These reads were aligned to the genome (GRCh37.1) using GSNAP (version 2013-03-31) with the following options: -M 2 -n 10 -B 2 -i 1 -N 1 -w 200000 -E 1 –pairmax-rna = 200000 –clip-overlap. An average of 51.3 million uniquely aligned reads with concordant read pairs was available per sample. Read counts for each transcript were processed with the DESeq variance stabilizing transform (VST) method to obtain normalized expression levels. The resulting expression measurements are referred to as “VST values.”
DNA copy number data
Illumina Human Omni 2.5-4 or Omni 2.5-8 arrays were processed with a modified version of the PICNIC (12) algorithm, as published recently (13). In short, the modifications to PICNIC increase the accuracy of global ploidy (and normal cell contamination) estimation and make the algorithm compatible with Illumina arrays. These modifications also allow for accurate total copy number estimates and dramatically reduce oversegmentation. After PICNIC determines the sample ploidy and performs segmentation with its HMM to determine allele-specific absolute copy number, we added an alternate segmentation with cghFLasso (14). This addition accounts for PICNIC's oversegmentation in regions where point copy number estimates indicate noninteger total copy number. Segments with three or fewer probes were merged with their neighboring segments.
For gene-specific copy number, this segmented total copy number was averaged for each gene using all SNP intervals that include a portion of that gene. We used the absolute copy number scale (where 2 is normal for an autosome) in some applications rather than the traditional “relative” log2 (absolute/ploidy) copy number. The significance of recurrent gains and losses was assessed with Genomic Identification of Significant Targets in Cancer (GISTIC; ref. 15). We used log2 (absolute copy number/ploidy) scale copy number data derived from the cghFLasso segmentation and analyzed only the autosomes. We used gain and loss cutoffs of 0.3 and −0.3, respectively, and the following other options: “-cap 3 –rx 0 –smallmem 0 –savegene 0 –v 0.”
Results
To identify new ADC targets that are broadly overexpressed in cancer versus normal tissue, we analyzed a comprehensive gene expression dataset of 35 different human tumor types and normal tissue controls. From this analysis, we identified lymphocyte antigen 6 complex, locus E (LY6E) as an attractive ADC target. LY6E exhibited significant overexpression in breast, colon, ovarian, pancreatic, kidney, and gastric carcinomas (Fig. 1A). LY6E resides at the 8q24.3 locus and exhibits significant copy number gain in a number of cancers where it is overexpressed (Fig. 1B). Of these cancers, we further characterized the LY6E locus in ovarian carcinomas as they had a high frequency of LY6E amplification and gain (Fig. 1B). We used GISTIC to analyze the LY6E amplicon from over 400 ovarian tumors that were part of TCGA dataset (15). We found broad copy number gain of the 8q arm, spanning both the MYC and LY6E locus (Fig. 1C). We found that LY6E amplification levels highly correlated with its overexpression in both ovarian and breast carcinomas (Fig. 1D). To examine LY6E in different subsets of breast cancer, we tested its expression in both marker defined and molecularly subclassified breast cancers. LY6E expression was found to be significantly upregulated in triple-negative/basal-like breast cancer by both quantitative reverse transcription PCR (qRT-PCR) and in silico analysis of 1,730 breast samples from the METABRIC study (ref. 16; Supplementary Fig. S1A and S1B). These data illustrate LY6E amplification and overexpression in a broad range of solid tumors.
LY6E gene expression and amplification in tumor and normal tissues. A, LY6E gene expression profile: Each dot represents LY6E gene expression in normal (black), cancer (red), and diseased (blue) human tissues. Measurements were carried out on the Affymetrix U133P chip and are expressed as scaled average difference. Horizontal black bars represent median expression across each set of tissues. Top panel represents overexpression in cancer tissues as fold change over normal samples from the same tissues where green or blue shading indicates statistical significant expression change. Red bold typeface on the x-axis denotes tumors that were further characterized for LY6E expression by IHC analysis (see Fig. 3). B, relative copy number (Materials and Methods) derived from TCGA tumors is plotted by tissue. Each point represents a tumor and is colored by discretized absolute total copy number (black: 2 copies, blue: <2 copies, orange: 3 or 4 copies, red: >4 copies). A bar plot indicates the proportions of tumors in each copy number category. C, GISTIC was used to assess the statistical significance of recurrent gain. A large portion of the q-arm of chromosome 8 shows highly significant recurrent gain. This region encompasses both MYC and LY6E. D, absolute total copy number (Materials and Methods) at the LY6E locus shows a highly significant positive correlation with LY6E transcript levels. The correlation coefficient (Spearman) and its significance are shown above for cancer panels of breast and ovarian carcinoma.
Intrigued by the potential utility of using LY6E as an ADC target in several types of cancers that have limited treatment options, we developed a monoclonal antibody to evaluate LY6E protein expression in various cancers by IHC. Anti-LY6E antibody GEN-93-8-1 (8-1) was derived from a rabbit immunized with N-term His-tagged LY6E protein. The sensitivity and specificity of the 8-1 antibody were validated using samples in which LY6E expression was independently confirmed. Enriched IHC signal by 8-1 in LY6E-expressing cells (exogenous LY6E expressing PC3 and endogenous LY6E-expressing luminal, ER+ breast cancer MDA-MB-175 VII) correlated well with flow-cytometric data demonstrating high LY6E surface expression in these cells with an independent anti-LY6E antibody 9B12 (Fig 2A and B). Little to no IHC signal was detected by 8-1 in LY6E-negative cell lines (PC3-Vector and Ramos; Fig. 2A and B). Specificity of 8-1 antibody to LY6E was further confirmed by the reduced anti-LY6E 8-1 immunoblot signal (∼8 kDa) after siRNA-mediated LY6E depletion (Fig. 2C). Finally, we found a significant positive correlation between the 8-1 IHC signal and LY6E gene expression analysis conducted on Fluidigm 96.96 dynamic arrays (Biomark HD system) on a panel of 106 serially cut human breast cancer tissues (Fig. 2D). Having established 8-1 as a specific and robust IHC reagent for LY6E, we evaluated in situ LY6E protein expression in 762 cancer samples, representing seven tumor types where high LY6E RNA levels were detected (Fig. 1A). LY6E expression was scored using a 50% threshold cutoff (see Materials and Methods) as follows: 0 (no staining/very weak staining); 1+ (weak staining); 2+ (moderate staining); and 3+ (strong staining). Consistent with our RNA analysis, the highest level of LY6E expression was found in triple-negative breast cancer samples (79% of samples showed 2+ or 3+ staining). In addition, breast, NSCLC, pancreatic, ovarian, head and neck, and gastric carcinomas all showed greater than 30% prevalence of moderate to high LY6E expression (Table 1). The LY6E expression pattern in cancer tissues was either membranous or membranous/cytoplasmic in nature (Fig. 3).
Development and characterization of anti-LY6E monoclonal antibody for IHC. A and B, correlating the detection of LY6E protein by 8-1 IHC to surface LY6E expression detected by 9B12 flow cytometry. Flow-cytometric detection of surface LY6E expression (red line) on live cells by 9B12 antibody is depicted on the left panel. Secondary antibody alone is used as control (gray line). IHC detection of LY6E total protein by 8-1 antibody is depicted on the corresponding right panel for each cell line. Cell lines depicted are PC3-LY6E (A, top); PC3-Vector (A, bottom); MDA-MB-175-VII, high endogenous LY6E expressing cell line (B, top); and RAMOS, non-LY6E–expressing cell line (B, bottom). C, total LY6E protein detection by immunoblotting with 8-1 antibody on PC3-LY6E cells. Independent LY6E siRNAs labeled 1, 2, 3, and 4 [or no siRNA (–) as control], were used to knock down LY6E expression. Lysates were analyzed for LY6E protein 48 hours post-siRNA transfection. β-actin serves as loading control (D). LY6E protein detection by 8-1 IHC is compared with LY6E gene expression in the same 106 serially cut breast cancer samples. The graph correlates IHC detection of LY6E protein by 8-1 antibody (IHC scoring on the x-axis) with LY6E gene expression analysis against two custom-designed assays targeting the reference genes, TMEM55B and VPS33B on Fluidigm Dynamic arrays. LY6E RNA expression on the y-axis is shown as ΔCt (calculated using geomean of two reference genes: TMEM55B and VPS33B). **, P < 0.001 for IHC 3 group as compared with IHC-0 or IHC-1+.
Expression of LY6E in human tumors and matched normal tissues. IHC detection of LY6E protein using anti-LY6E GEN-93-8-1 antibody is shown from a series of normal and cancer tissues in indications where LY6E transcript was found to be upregulated in a tumor-specific manner. Photomicrograph images depict tumor tissues representative of 0, 1+, 2+, and 3+ staining intensity. Black scale line, 100 μm.
Prevalence of LY6E in various cancers
An important hallmark of an ADC target is its overexpression in cancer tissues as compared with relatively low normal tissue expression. To this end, LY6E transcript and protein expression in normal tissues was assessed. Using qRT/PCR, low to no LY6E transcript expression was detected in a normal tissue panel when compared with LY6E transcript detected on selected xenograft tumor models known to express LY6E by IHC analysis (Supplementary Fig. S2). To further delineate LY6E expression in normal tissues, a tissue microarray composed of 28 normal human tissues was stained with anti-LY6E 8-1 antibody. Consistent with our RNA analysis of Ly6E expression, weak to no expression was detected in normal tissues (Fig. 3 and Supplementary Table S2). Thus, high LY6E expression across a broad range of solid tumor malignancies; coupled with relatively low to no LY6E expression in most normal tissues tested underscores the potential utility of LY6E as an ADC target.
LY6E is a GPI-anchored protein and as such localizes to the plasma membrane. Given our data showing high LY6E expression in tumor compared with normal tissue, we surmised that LY6E might serve as a tumor antigen target for an ADC. To explore this, we first generated a panel of murine monoclonal antibodies against LY6E by using N-term His-tagged LY6E protein as immunogen. The anti-LY6E monoclonal antibodies were characterized for their LY6E-binding affinity, cross species reactivity and nonreactivity against prostate stem cell antigen, the closest homolog (paralog) with 31% identity to LY6E. 9B12 was identified as a LY6E-specific antibody with high affinity to human LY6E (average dissociation constant Ķd of 3.7 nmol/L; Scatchard analysis) and to cynomolgus monkey LY6E (average dissociation constant Ķd of 6.9 nmol/L; Scatchard analysis; Supplementary Table S3). 9B12 was therefore chosen for further development as a potential therapeutic molecule against LY6E-positive cancers.
Effective drug delivery of auristatins to cancer cells requires internalization of the therapeutic molecule followed by efficient cleavage of the linker molecule in the lysosomes to mediate intracellular release of free drug (17).
GPI-anchored proteins are organized into lipid rafts and are endocytosed via the CLIC/GEEC (clathrin-independent carrier/GPI anchored protein enriched early endosomal compartment) internalization route (18, 19). To test whether 9B12 would be internalized effectively into cancer cells, we used immunofluorescence detection to evaluate uptake of the antibody into ovarian cancer cells KURAMOCHI. We observed uptake of 9B12 into live cells within a 30-minute incubation period. Postincubation with 9B12 antibody, cells were fixed and permeabilized. 9B12 colocalized to the intracellular vesicles marked by the late endosomal/lysosomal marker LAMP1 (Supplementary Fig. S3A). We observed that uptake of antibody was inhibited by preincubating cells with lipid raft disruptor, 1 mmol/L MBC (20), whereas uptake of 9B12 into cells remained unaffected by inhibitor of clathrin-mediated endocytosis, 25 μmol/L CPZ (21). Uptake of transferrin, a ligand endocytosed via clathrin, was indeed inhibited by preincubating cells with CPZ (Supplementary Fig. S3B). 9B12 was also endocytosed to the lysosomes of breast cancer cells (HCC1569) within a 2-hour period (unpublished data). These data demonstrate that 9B12 would effectively deliver the drug to lysosomes; where the ADC could be degraded to release free MMAE toxin to mediate cell killing. These data strongly support the use of 9B12 as a drug delivery agent targeting LY6E-positive cancers.
An important aspect of an ADC is its ability to mediate target-dependent cell killing. Because LY6E amplification levels highly correlated with its overexpression in ovarian carcinomas, we chose to evaluate the ability of anti-LY6E ADC to provide targeted chemotherapy against a selected panel of ovarian cancer cell lines. We identified cell lines that were sensitive to the free MMAE drug and displayed a range of LY6E surface expression that correlated positively to LY6E copy number gain (Supplementary Table S4). Figure 4A and B highlights anti-LY6E ADC activity in three ovarian cancer cell lines with varying LY6E surface expression and copy number (CN) gain, (low expression, DOV 13; moderate expression, COV362; and high expression, KURAMOCHI); but with equivalent sensitivity to free MMAE (EC50 values of 0.38, 0.33, and 0.3 nmol/L; Fig. 4C). The data demonstrated an overall correlation between LY6E surface expression and anti-LY6E ADC cell killing activity (Fig. 4A and B and Supplementary Table S4). Characterization of LY6E in 17 breast cancer cell lines showed a similar correlation where LY6E expression predicted sensitivity to anti-LY6E ADC (Fig. 4D). In summary, these data demonstrate the ability of anti-LY6E ADC to deliver a targeted cytotoxic agent.
Correlation of LY6E surface expression to cell killing in vitro LY6E surface expression correlates to anti-LY6E ADC cell killing in vitro. A, shows the relative amounts of cell surface LY6E detected by flow cytometry with 9B12 antibody (red line) on ovarian cell lines DOV 13 (left), COV362 (middle), and KURAMOCHI (right). Secondary antibody alone was used as a control (gray line). LY6E copy number (CN) for each cell is shown at the top of the flow-cytometric plot. B, cell killing by anti-LY6E ADC (9B12) titration is presented below each flow-cytometric plot for the corresponding cell line. The indicated concentrations of anti-LY6E-vc-MMAE (red line) or control IgG-vc-MMAE (gray line) were incubated with cells for 5 days and relative cell viability (y-axis) assessed using CellTiter-Glo. P values are calculated based on two tailed, Student t test. Asterisks depict P value <0.01. C, the cell lines DOV 13 (dotted line, round markers), COV362 (solid line, round markers), and KURAMOCHI (dotted line, square markers) were treated with increasing concentrations of free MMAE for 5 days and cell viability was determined by the CellTiter-Glo assay. D, correlation between LY6E expression and anti-LY6E ADC killing (at 5 μg/mL anti-LY6E ADC) in a panel of 17 MMAE-sensitive breast cancer cell lines. Each dot represents one cell line. Viability effects were determined 3 days after single-dose treatment with the anti-LY6E-ADC. Correlation was determined by Pearson analysis and the P value of the correlation is indicated. RPKM is defined as Reads Per Kilobase per Million mapped reads.
To determine the therapeutic potential of anti-LY6E ADC, we evaluated its preclinical activity against murine xenograft models derived from various cancer cell lines and patient-derived tumor models (PDX). We chose models that allowed us to generate robust uniform sized tumors; and most closely represented the moderate (IHC 2+) LY6E protein expression levels (Fig. 5A–D; inset panels) that could be expected in cancers in clinical settings. A single dose of anti-LY6E ADC demonstrated excellent efficacy against both a pancreatic (SU.86.86) and a NSCLC xenograft model (SW900) that homogenously express Ly6E at weak to moderate levels. Anti-LY6E ADC provided clear tumor inhibitory activity in both models; a 2-mg/kg dose yielded 8 of 10 partial responses (PR) and 1/10 complete response (CR) in the SU.86.86 xenograft (Fig. 5A); whereas a 1-mg/kg dose of the ADC yielded 8 PRs with tumor regression lasting for about a month postdose in the SW900 xenograft model (Fig. 5B). Anti-LY6E ADC activity was 1. Dependent on LY6E expression in all models; as a control non-specific antibody, conjugated to MMAE had little activity at the highest dose; and 2. Required MMAE, since 8 to 12 mg/kg unconjugated anti-LY6E antibody had no activity in the three models tested (Supplementary Table S5).
Anti-LY6E ADC inhibits tumor growth in vivo. A–D, graphs depict efficacy of anti-LY6E ADC in various pancreatic, NSCLC, and breast cancer xenografts. Mice bearing tumors derived from SU.86.86 pancreatic cancer cell line (A) or SW900 NSCLC squamous cell carcinoma (B), patient-derived primary tumors, PAXF 1657 (C), and HBCx-9 (D) were administered a single i.v. injection of either vehicle, anti-LY6E antibody 9B12, either as unconjugated [Naked Ab, depicted in (A) only, see supplementary Table S5 for Naked Ab in other xenografts], or as ADC (LY6E), control ADC (Ctrl) at mg/kg doses indicated in parentheses. Photomicrographic insets show the intensity and heterogeneity of LY6E expression as detected by anti-LY6E IHC. Scale bar, 100 μm. Average tumor volumes with SDs were determined from 9 to 10 animals per group and depicted on the y-axis whereas day of study was depicted on the x-axis; using GraphPad Prism v6.0e software. Detailed tumor growth and statistical analyses for the efficacy xenograft models are provided in Supplementary Tables S5 and S6.
Recent studies have suggested that PDX models would better represent the clinical behavior of human tumors as compared with cell line-derived xenografts (22, 23). We, therefore, tested anti-LY6E ADC activity in a pancreatic PDX model, PAXF 1657, which expresses LY6E (IHC 2+) at moderate levels (Fig. 5C). Anti-LY6E ADC showed remarkable efficacy with 8 of 10 PRs and 2 of 10 CRs following a single anti-LY6E ADC dose of 4 mg/kg (Fig. 5C)
Many ADC targets are heterogeneously expressed in human tumors but are only validated in preclinical models that show robust homogenous target expression (24, 10). Therefore, to better represent tumors exhibiting nonuniform target expression, we chose two PDX models HBCx-9 (TNBC) and MAXF 1162 (HER2+ breast cancer) where LY6E was heterogeneously expressed at weak–moderate levels (Fig. 5D and Supplementary Fig. S4). Nevertheless, robust tumor killing was observed in both models. In HBCx-9, all animals achieved a complete response at a dose of 12 mg/kg. Tumor regression was sustained for approximately 25 days after treatment, even in the lower dose (8 mg/kg). To further test the limits of anti-LY6E ADC, we used the MAXF 1162 HER2+ breast cancer PDX model. Interestingly, the MAXF 1162 model, despite displaying high (IHC 3+, data not shown) HER2 expression, demonstrated no response to HER2 targeting by T-DM1 (Supplementary Fig. S4B). Remarkably, we observed strong efficacy of anti-LY6E ADC in MAXF 1162 with a single dose at 4 mg/kg yielding 4 of 10 PRs. Higher doses of the ADC led to complete responses. (Supplementary Fig. S4C and Supplementary Table S5 and S6). In summary, potent dose-dependent anti-LY6E ADC efficacy was achieved in vivo against a range of tumor indications with different LY6E expression patterns and response to other ADC agents.
Discussion
In this study, we used an in silico approach to identify LY6E transcript overexpression in a wide array of solid tumor malignancies. LY6E protein overexpression in these tumors was confirmed by developing a robust, LY6E-specific IHC assay. Furthermore, overexpression of LY6E was noted in several cancer types that have limited treatment options with the highest level of LY6E expression being found in triple-negative breast cancer samples, (79% of samples showed 2+ or 3+ staining). It is worth noting that while a number of ADC targets have been offered in specific cancer target indications, LY6E is unique in its broad expression pattern across a wide range of different human cancers. These findings enhanced our interest in exploring the therapeutic potential of using LY6E as an ADC target.
We chose to use LY6E as a target antigen for the synthetic, potent antimitotic analogue of dolstatin 10, MMAE, which binds and disrupts the microtubule network in proliferating cells (9, 17). The use of MMAE offered distinct advantages over using DNA-binding cytotoxic agents such as duocarmycin and calicheamicin analogues, as these potent molecules lack selectivity to target cells (25). MMAE was further chosen over other clinically approved drugs such as doxorubicin, vinca alkaloids, etc., as these drugs were marred by relatively low potencies, linker instability, and compositional heterogeneity (9, 25, 26). MMAE offered us the further advantage of using conjugation and cleavable linker chemistries that allowed the ADC to remain relatively stable in plasma but released the drug efficiently in the acidic lysosomal compartment of target cells, thereby minimizing systemic drug release and damage to nontarget cells (17, 25).
Internalization of the antibody for effective drug delivery to the acidic lysosomal compartment of a cancer cell, where intracellular cleavage releases free drug, remains a key feature of MMAE-conjugated ADC-targeted chemotherapy. In principle, such ADCs should have minimal toxicities toward normal cells due to low target expression on these cells resulting in low uptake of the drug. In addition, MMAEs interfere with microtubule dynamics in dividing cells; the rate of cell division of normal cells is generally lower than that for cancer cells; which further moderates potential toxicities of the ADCs to normal cells (8).
We elucidated LY6E trafficking in an effort to maximize antitumor efficacy by selecting the appropriate antibody to best mediate ADC drug delivery. We observed uptake of anti-LY6E antibody 9B12 into lysosomes within a 30-minute incubation period, via mechanisms distinct from the classical trafficking pathways involving clathrin-coated pits. Our data demonstrate effective endocytosis of a GPI-anchored protein and illustrate that the 9B12 anti-LY6E ADC would serve as an effective mediator of drug delivery to lysosomes.
We found an overall trend toward increased LY6E gene copy number correlating to increased LY6E cell-surface expression and enhanced in vitro anti-LY6E ADC activity in breast and ovarian cancer cell lines. However, in vitro killing by ADC was also clearly influenced by the relative sensitivity of the various cell lines to the free drug. For instance, OAW42, an ovarian cell line with high LY6E gene copy number of 5, was resistant to killing by free MMAE and therefore resistant to killing by anti-LY6E ADC.
Having observed excellent efficacy of anti-LY6E ADC against cancer cell line derived xenografts, we further evaluated its efficacy in physiologically relevant PDX models with moderate homogenous and heterogeneous LY6E expression. Efficacy of anti-LY6E ADC at 4 mg/kg in the pancreatic PDX model PAXF 1657 was particularly encouraging in view of the fact that an ADC targeting another GPI-anchored protein mesothelin (MSLN) yielded comparable efficacy in the same model at a substantially higher dose of 20 mg/kg (27). Both LY6E and MSLN demonstrated moderate (IHC 2+) scores within the dynamic range of individual target expression in the PAXF 1657 PDX model. On the basis of our internalization data, we may speculate that the higher efficacy of anti-LY6E ADC as compared with anti-MSLN ADC in this model could be due to a more efficient endocytosis process for LY6E (30 minutes for LY6E as compared with 20 hours for MSLN), (27) resulting in robust drug delivery and therefore, efficacy of the ADC.
MAXF 1162 (40% IHC+) and HBCx-9 (∼15% IHC+) represent high bar models with less than 50% of tumor cells exhibiting LY6E expression by IHC. The remarkable efficacy of anti-LY6E ADC in these heterogeneous disease settings could likely be due to a bystander effect, whereby the efficient release of the drug in target cells is followed by diffusion of the membrane permeable MMAE into neighboring cells lacking target (28). The primary mechanism of action of MMAE is by the disruption of the microtubule network in proliferating cells (9, 17). This, combined with low LY6E expression in normal tissues may reduce potential systemic toxicities due to anti-LY6E ADC.
Moreover, the expression of LY6E in HER2+ breast cancers suggests that anti-LY6E ADC could represent a treatment option for HER2+ patients, unresponsive to existing treatments such as T-DM1. Indeed, the anti-LY6E ADC was effective in the MAXF 1162 PDX model, which responds only weakly to T-DM1 despite IHC3+ expression of HER2 (Supplementary Fig. S4B). Further investigation will be required to elucidate the mechanism that accounts for LY6E activity in T-DM1–resistant breast cancers.
In summary, we find that LY6E protein is overexpressed in a wide range of solid tumor malignancies and anti-LY6E ADC has potential utility as a treatment option for patients with unmet medical need in clinical settings.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: J. Asundi, J. Tremayne, P. Chang, R. Raja, R. Firestein
Development of methodology: J. Asundi, P. Chang, C. Sakanaka, R. Desai, R. Raja, R. Firestein
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Asundi, L. Crocker, J. Tremayne, P. Chang, C. Sakanaka, J. Tanguay, S. Chalasani, E. Luis, R. Raja, R. Firestein
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Asundi, L. Crocker, P. Chang, C. Sakanaka, J. Tanguay, E. Luis, K. Gascoigne, B.A. Friedman, P.M. Haverty, P. Polakis, R. Firestein
Writing, review, and/or revision of the manuscript: J. Asundi, L. Crocker, R. Desai, R. Raja, P. Polakis, R. Firestein
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Asundi, L. Crocker, P. Chang, C. Sakanaka, R. Firestein
Study supervision: P. Chang, S. Spencer, R. Raja, R. Firestein
Other (generated qPCR data for the manuscript): R. Desai
Acknowledgments
The authors thank Zeran Wang for help with Scatchard analysis, Anneleen Damaen and Zhaoshi Jiang for Bioinformatics support, and Linda Rangell, Debra Dunlap, and Thinh Pham for tissue staining and analysis.
Footnotes
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
- Received January 21, 2015.
- Revision received March 17, 2015.
- Accepted March 19, 2015.
- ©2015 American Association for Cancer Research.