Effective Immunoconjugate Therapy in Cancer Models Targeting a Serine

Cancer Therapy: Preclinical

Effective Immunoconjugate Therapy in Cancer Models Targeting a Serine Protease of Tumor Fibroblasts Elinborg Ostermann,1Pilar Garin-Chesa,1, 2 Karl Heinz Heider,1Milena Kalat,1Herbert Lamche,1 Christina Puri,2 Dontscho Kerjaschki,2 Wolfgang J. Rettig,1and Guenther R. Adolf1

Abstract Purpose: Invasion and metastasis of malignant epithelial cells into normal tissues is accompanied by adaptive changes in the mesenchyme-derived supporting stroma of the target organs. Altered gene expression in these nontransformed stromal cells provides potential targets for therapy.The present study was undertaken to determine the antitumor effects ofan antibody-conjugate against fibroblast activation protein-a, a cell surface protease of activated tumor fibroblasts. Experimental Design: A novel antibody-maytansinoid conjugate, monoclonal antibody (mAb) FAP5-DM1, was developed to target a shared epitope of human, mouse, and cynomolgus monkey fibroblast activation protein-a, enabling preclinical efficacy and tolerability assessments.We have used stroma-rich models in immunodeficient mice, which recapitulate the histotypic arrangement found in human epithelial cancers. Results: Treatment with mAb FAP5-DM1induced long-lasting inhibition of tumor growth and complete regressions in xenograft models of lung, pancreas, and head and neck cancers with no signs of intolerability. Analysis of chemically distinct conjugates, resistance models, and bio- markers implicates a unique mode of action, with mitotic arrest and apoptosis of malignant epithelial cells coupled to disruption of fibroblastic and vascular structures. Conclusions: We show that mAb FAP5-DM1combines excellentefficacyand tolerabilityandpro- vides a first assessment of the mode ofactionofa novel drug candidate for tumor stroma targeting, thus encouraging further development toward clinical testing of this treatment paradigm.

Malignant epithelial cancers, the major cause of cancer Two common liabilities in the preclinical efficacy and safety morbidity and mortality, arise in organs composed of both assessment of therapeutic antibodies were addressed. The first is epithelial and mesenchyme-derived stromal cells, such as the limitation of efficacy models using antibodies that do not fibroblasts, myofibroblasts, endothelial cells, pericytes, smooth cross-react with the model species. As routine efficacy models muscle, and hematopoietic cells. During disease progression, comprise human cancer cells growing in immunodeficient the stroma of primary, invasive lesions as well as distant mice, the selected therapeutic antibodies were artificially metastases changes in architecture, gene expression, secretion of ‘‘cancer specific,’’ leading to an overestimation of in vivo soluble mediators, and extracellular matrix deposition; in turn, targeting potential (circulating antibody is not sequestered in the malignant epithelial cells can undergo reversible epithelial- normal murine tissues expressing the homologous antigen) and mesenchymal transitions (1–5). The fibroblast response in exaggerated efficacy-to-tolerability ratios not confirmed in several types of human cancer is characterized by the induction subsequent clinical studies. For our approach of cancer of an integral cell surface protein, fibroblast activation protein- fibroblast-targeted therapy, the model requires antibody a (FAPa; refs. 6, 7), a serine protease (8–11) with highly binding to the murine stromal cells in the tumor xenografts; restricted expression in developing organs, wound healing, and therefore, the generation of a monoclonal antibody (mAb) tissue remodeling (7, 12–16). In the current study, we explored cross-reactive with human and mouse FAPa was essential. FAPa as a candidate target for cancer therapy (17–19) with A second limitation of conventional efficacy models relates specific immunoconjugates. to histologic findings. Thus, when xenograft models of epithelial cancers are derived from long-term tissue culture cell lines, they show a tissue architecture that is highly atypical for the human cancer types under investigation, presenting as Authors’ Affiliations: 1Boehringer Ingelheim Austria GmbH; and 2Institute of nodules of morphologically undifferentiated tumor cell clusters Clinical Pathology, Medical University of Vienna, Vienna, Austria with little if any fibroblastic stroma or histotypic features Received 12/19/07; revised 2/25/08; accepted 2/26/08. The costs of publication of this article were defrayed in part by the payment of page (Fig. 1). A more authentic histologic appearance is observed in charges. This article must therefore be hereby marked advertisement in accordance carcinoma models derived by direct implantation of surgical with 18 U.S.C. Section 1734 solely to indicate this fact. specimens into immunodeficient mice (13, 20) or from Requests for reprints: Pilar Garin-Chesa, Boehringer Ingelheim Austria GmbH, purified cell suspensions freshly obtained from surgical speci- Dr Boehringer-Gasse 5-11, A-1120 Vienna, Austria. Phone: 431-801052796; Fax: 431-801052366; E-mail: [email protected]. mens (21, 22). Based on these observations, we prescreened a F 2008 American Association for Cancer Research. series of cancer models to identify those most closely reflecting doi:10.1158/1078-0432.CCR-07-5211 the histology seen in patients.

Clin Cancer Res 2008;14(14) July 15, 2008 4584 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2008 American Association for Cancer Research. Antistroma Tumor Therapy

Fig. 1. Comparative histopathologic analysis of the stromal compartment in human epithelial cancers and xenograft models in immunodeficient mice. A, human epithelial cancers with prominent desmoplastic stroma including a colorectal cancer, an infiltrating ductal carcinoma of the breast, and a squamous cell carcinoma of the head and neck (H&E staining). Adjacent sections stained by immunohistochemistry showed prominent FAPa expression in the activated tumor stromal fibroblasts (avidin-biotin complex method).Three xenograft models derived from cell lines HCT116 (colon carcinoma), MCF7 (breast carcinoma), and FaDu (head and neck carcinoma) are shown by comparison. The colon and breast xenograft models grow as solid tumor masses and exhibit very little stroma reaction. In contrast, the FaDu-derived xenograft shows strands of FAPa+ fibroblasts (inset) separating the clusters of tumor cells, resembling the morphology of the human primary tumors.

With this testing scheme in place, we have developed specific, expressing recombinant human and mouse FAPa, respectively) cultures high-affinity mAbs that bind shared epitopes of mouse, human, were maintained in RPMI 1640 supplemented with 10% heat- and monkey FAPa and generated immunoconjugates with inactivated fetal bovine serum and 1% glutamine. In addition, the A distinct maytansine derivatives and diverse linker moieties. We recombinant cell lines were kept in 200 g/mL G418. The human tumor xenografts CXF158 (colon carcinoma), LXFA629 (non-small cell show that these stroma fibroblast-targeted immunoconjugates lung adenocarcinoma), and PAXF739 (pancreatic adenocarcinoma) combine excellent efficacy and tolerability and provide a first were established at Oncotest, by s.c. implantation of surgical tumor assessment of their mode of action. specimens in immunodeficient mice, and were maintained by serial passaging (20). Antibody generation and characterization. The mAb FAP5 was Materials and Methods generated after immunization of BALB/c FAPa-/- mice with recombinant murine CD8-FAPa fusion protein (10) by hybridoma technology. Cell lines and tumor tissues. FaDu (human squamous cell carcino- Reactivity of mAb FAP clone 5 to recombinant human (9), mouse (10), ma; ATCC HTB-43), HT1080 (human fibrosarcoma; ATCC CCL-121), and cynomolgus FAPa was tested by ELISA and fluorescence-activated and HT1080 v1.33 and HT1080 13.8 (human fibrosarcoma cell lines cell sorting using the FAPa-negative human fibrosarcoma cell line

www.aacrjournals.org 4585 Clin Cancer Res 2008;14(14) July 15, 2008 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2008 American Association for Cancer Research. Cancer Therapy: Preclinical

a HT1080 and cloned derivatives expressing human and mouse FAP . Results Recombinant His-tagged cynomolgus monkey FAPa for ELISA was generated by amplification of the ECD (amino acids 27-760) by PCR Generation of FAPa-specific,species cross-reactive anti- with primers derived from the human sequence and cloning into the bodies. FAPa-/- mice (26) were used to generate antibodies expression vector pSecTag2. The protein was expressed by transient a transfection of HEK293 freestyle cells (Invitrogen) and purified on that recognize an epitope shared by human FAP and its close Ni-columns. orthologues in the mouse and cynomolgus monkey. The lead Affinity determination by surface plasmon resonance. Anti-mouse antibody, mAb FAP5, binds to recombinant human, cynomol- IgG antibody (f200 resonance units) was immobilized using gus monkey, and mouse FAPa with affinities of 5, 4, and the amine coupling kit on a CM5-biosensor chip in a Biacore 2000 0.6 nmol/L, respectively, as determined by surface plasmon (all materials were from Biacore). mAb FAP5 was bound to the sensor resonance assay. The pattern of FAPa expression in normal chip (5.2 Ag/mL for 3 min). Association and dissociation of human and mouse tissues as well as tumor tissues has been a recombinant human, mouse, and cynomolgus FAP were measured defined previously in considerable detail by immunohisto- for 5 min at concentrations from 3.7 to 300 nmol/L. Affinity variables chemistry and RNA analysis (7, 12–14, 16, 27). For mAb FAP5, were calculated using the separate cure fit algorithm of the BIAevalua- we confirmed the restricted antigen expression in a panel of tion software version 4.1 (Biacore). Immunoconjugates. The mAb FAP5-maytansinoid conjugates mAb normal adult human tissues, normal mouse tissues, and a series FAP5-SPP-DM1 (here designated mAb FAP5-DM1), mAb FAP5-SPDB- of epithelial cancers. As predicted, mAb FAP5 reacts with tumor DM4 (mAb FAP5-DM4), and mAb FAP5-SMCC-DM1 were prepared at stromal fibroblasts in colorectal, breast, head and neck, and ImmunoGen following published procedures (23–25). pancreatic carcinomas, whereas epithelial cancer cells and Cell proliferation assay. The cytotoxic activity of the DM1 and DM4 stromal fibroblasts of normal tissues lack FAPa expression, conjugates was tested on untransfected and human FAPa-transfected with the known exception of fibroblasts in the uterus and HT1080 cells using a MTS staining system. The absorbances were read at scattered dermal fibroblasts (6, 16, 28). Likewise, analysis of 490 nm to determine the concentration of the antibody conjugate human tumor xenografts grown in nude mice revealed staining needed to achieve 50% inhibition of tumor cell growth (EC50). The of activated stromal fibroblasts of mouse origin, but no four-variable logistic curve-fit algorithm of the GraphPad Prism immunoreactivity was observed with human tumor cells software version 3.03 (GraphPad) was used. Human xenograft tumor models in mice. Female athymic NMRI (Fig. 1). In fluorescence-activated cell sorting assays, mAb nude mice (Taconic) ages 6 to 8 weeks were inoculated s.c. with FaDu FAP5 binds to HT1080 cells transfected with human and mouse a a tumor cells (1 Â 106/100 AL PBS) in the right flank. For the lung, FAP cDNA, respectively, but not to the FAP -negative parental pancreas, and colon carcinoma models, mice were grafted with tumor cells (Fig. 2A). fragments in the right flank. Tumor growth was measured three times a Design of mAb FAP5-maytansinoid immunoconjugates as week and the tumor volume was determined using the formula: p /6Â therapeutic agents. We determined that unmodified mAb larger diameter Â (smaller diameter)2. Studies were terminated when FAP5, at concentrations up to 50 Ag/mL, has no detectable tumors reached an average size of 1,500 mm3 or when tumors were effect on survival or proliferation of HT1080-FAPa cells in vitro. judged to adversely affect the well-being of the animals. Treatment was Moreover, we established that bound mAb FAP5 is rapidly administered i.v. to groups of six or eight mice and commenced when a j 3 internalized in HT1080-FAP cells at 37 C. We therefore tumors had reached a size of 100 to 250 mm . Body weights were endowed mAb FAP5 with a de novo antimitotic function monitored three times per week. All animal experiments were done through covalent linkage with DM1, a tubulin-binding may- according to the legal requirements in Austria as well as to the guidelines of the American Association for Laboratory Animal Science. tansinoid with picomolar antimitotic activity (23–25, 29). Immunohistochemistry. Fresh-frozen tumor samples were analyzed Using SPP as a linker, we generated a mAbFAP5-DM1 drug using the avidin-biotin complex immunoperoxidase procedure as candidate (Fig. 2B) with desired specificity and activity. This described previously (7). For the analysis of FAPa expression in mouse chemical modification did not alter the binding specificity and tissues and biomarker modulation in tumor xenografts, NMRI nude the affinity of the parental mAb FAP5 as determined by direct mice were grafted s.c. with the human tumors as described above. Mean binding and competitive ELISA (data not shown). Unlike mAb 3 tumor volume at the start of the experiment was f100 mm . Tumors FAP5, the mAb FAP5-DM1 immunoconjugate showed remark- were then excised either before treatment or 48 h, 72 h, and 5 days after ably high potency in proliferation assays with HT1080-FAPa i.v. administration of the compounds and at the end of the experiment. cells (EC , 50 pmol/L), with z100-fold selectivity against For FAPa detection in mice tissues, mAb FAP5 was biotinylated and 50 FAPa-negative control cells (Fig. 2C). detected with the avidin-biotin complex method as before. The binding of the antibodies was visualized with 3,3¶-diaminobenzidine solution. Immunoconjugate mAb FAP5-DM1 is highly efficacious in in vivo Slides were then dehydrated and counterstained with Harris’ hema- cancer xenograft models. The efficacy of mAb FAP5- toxylin. For double immunostaining of tumor cells and stroma com- DM1 given once weekly by i.v. injection was tested in four ponents, an indirect immunofluorescence method was used. Apoptosis human cancer xenograft models (pancreatic, non-small cell was determined with a terminal nucleotidyl transferase–mediated lung, colorectal, and head and neck squamous cell carcinomas) dUTP nick end labeling assay using the ApopTag Fluorescein In situ grown s.c. in immunodeficient mice. For each model, Apoptosis Detection Kit (Chemicon International) and analyzed by prescreening had established a histotypic arrangement of fluorescence microscopy. epithelial and stromal components consistent with the human Antibodies. Primary antibodies used included anti-human/mouse a disease, and the mouse-derived tumor stromal fibroblasts show FAP mAb FAP5, anti-cytokeratin 18 (clone DC10; DAKO), anti- a a distinct FAP expression before initiation of therapy (Fig. 3A, smooth muscle actin (DAKO),anti-collagen type IV (Chemicon), anti- untreated mouse endothelial marker Meca32 (BD PharMingen), antibody to ). Furthermore, we observed that, following treatment Ser10-phosphorylated histone H3 (Upstate Biotechnology), anti-mouse of the responsive pancreas and lung cancer models, the CD11b (eBioscience), and anti-pan-macrophage marker F4/80 (eBio- nonprogressing tumors consist of bands of collagenized science). Alexa-conjugated secondary antibodies were obtained from stroma, with scattered fibroblasts, small capillaries, micro- Molecular Probes. calcifications, diffuse inflammatory infiltrates, and only minute

Clin Cancer Res 2008;14(14) July 15, 2008 4586 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2008 American Association for Cancer Research. Antistroma Tumor Therapy

Fig. 2. Characterization and cross- reactivity of mAb FAP5 and mAb FAP5-maytansinoid immunoconjugates. A, selective binding of mAb FAP5 to HT1080 cells expressing recombinant human and mouse FAPa and to wild-type HT1080 cells by flow cytometry. B, structural representation of mAb FAP5-maytansinoid conjugates. C, comparison of cytotoxic activity of mAb FAP5-DM1on wild-type HT1080 FAPa- cells and HT1080 cells expressing human FAPa.

foci of malignant epithelial cells (Fig. 3A, end of treatment); the resistance gene expression and in vivo unresponsiveness to same post-treatment pattern is seen for the head and neck treatment with taxanes. In this model, mAb FAP5-DM1 showed cancer model (data not shown). As expected, the resistant no therapeutic effect (Fig. 3B). colorectal cancer model shows no histologic changes after Cleavable linkers are essential for potent in vivo efficacy. We therapy (Fig. 3A). The treatment schedule consisted of once compared the efficacy of mAb FAP5-DM1 to similar constructs weekly i.v. injections of mAb FAP5-DM1 at three dose levels, that differ in the chemical stability of the linker (24) and the adjusted to 100, 200 and 400 Ag/kg DM1, respectively, with metabolic fate of the conjugates in tumor tissues and following four to five consecutive treatment cycles (Fig. 3B). In the cellular uptake. Two analogues of mAb FAP5-DM1 were pancreas, lung, and head and neck carcinoma models, a constructed (Fig. 2B): mAb FAP5 conjugated to DM4 by the prominent antitumor effect was observed, including complete disulfide linker SPDB (mAb FAP5-DM4; ref. 30) and mAb tumor regressions (3 of 6 animals in the pancreatic and head FAP5 conjugated to DM1 by the thioether linker SMCC (mAb and neck cancer models and 5 of 6 animals in the lung cancer FAP5-SMCC-DM1; ref. 24). Both analogues were profiled models in the high-dose groups). After discontinuation of in vitro and in vivo as described for mAb FAP5-DM1, and therapy (days 20-27), a large proportion of animals remained two key findings emerged. First, each of these compounds was free of palpable tumors or showed no further growth during an highly active against HT1080-FAPa cells in vitro, with EC50 extended observation period (3-4 weeks). Remarkably, strong values of 29 and 22 pmol/L, respectively, and 100- to 1,000- and long-lasting tumor regressions in the pancreas cancer fold less active on the HT1080 wild-type cells (Table 1). model were seen even with a single i.v. administration of mAb Second, mAb FAP5-DM4 showed impressive in vivo antitumor FAP5-DM1 at the high dose (see below). In all treatment activity, similar to mAb FAP5-DM1, whereas mAb FAP5- groups, the animals showed no evidence of toxicity or impaired SMCC-DM1 was inactive in the pancreas cancer model and weight gain when compared to control mice. The colorectal only marginally active in the lung and head and neck cancer cancer model was selected based on high MDR1 multidrug models (Fig. 4A-C). Previous studies with antiepithelial cancer

www.aacrjournals.org 4587 Clin Cancer Res 2008;14(14) July 15, 2008 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2008 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 3. A, histopathologic analysis of tumor xenografts treated with mAb FAP5-DM1. Sections from tumor xenografts before treatment stained with H&E. Moderately to well-differentiated adenocarcinomas of pancreas, lung, and colon. Adjacent sections stained with mAb FAP5 show prominent expression of FAPa by the activated tumor stromal fibroblasts in all three cases (avidin-biotin complex method). Morphologic changes following treatment in the pancreatic and lung carcinoma models. Small tumor nodules composed of a collagenized stroma with scattered fibroblasts, inflammatory infiltrates (arrowheads), and a minute cluster of tumor cells (arrow).The resistant colorectal cancer model showed no histologic changes after treatment. Bar, 50 Am. B, efficacy of mAb FAP5-DM1in human cancer xenograft models. Nude mice bearing established PAXF736 (pancreas), LXFA629 (lung), and CXF158 (colon) tumors were treated i.v. with either vehicle control (PBS; green, filled square), mAb FAP5-DM1at doses of 200 Ag/kg DM1 (red, filled circle), 40 0 Ag/kg DM1 (blue, open square), and unconjugated antibody (orange, open triangle) for four to five cycles.Tumor sizes are represented as median of six or eight mice. Arrows, treatment days.

Clin Cancer Res 2008;14(14) July 15, 2008 4588 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2008 American Association for Cancer Research. Antistroma Tumor Therapy

Table 1. Cytotoxic activity of the mAb FAP5-maytansinoid conjugates

Conjugate Linker/bond Drug In vitro efficacy Selectivity factor EC50 ratio: HT1080 + EC50 (pmol/L) wild-type/HT1080-FAPA mAbFAP5-DM1 SPP/disulfide DM1 50540 mAbFAP5-DM4 SPDB/disulfide DM4 29 155 mAbFAP5-SMCC-DM1 SMCC/thioether DM1 22 682 antibodies (29) have shown that the respective immunocon- xenograft models. No evidence of toxicity or impaired jugates do not differ significantly in their general pharmaco- weight gain when compared with control mice was observed kinetic properties in vivo but are quite distinct in their (Fig. 4D). metabolic fate after cellular uptake. Thus, mAb FAP5-DM1 Changed architecture during tumor stromal fibroblast therapy. and mAb FAP5-DM4 contain linkers with disulfide bonds that Histologic and immunological biomarker analyses were carried are readily cleaved on internalization, whereas the SMCC out to map the changes in tissue architecture during the course linker contains a noncleavable thioether bond. Accordingly, of mAb FAP5-DM1 therapy. the pattern we observed with our three compounds in vitro Focusing on the lung cancer xenograft model (Fig. 5), (similar potency) versus in vivo (dramatic differences in we confirmed that, before therapy, the FAPa-expressing tumor efficacy) are suggestive of a marked bystander effect in the stromal fibroblasts are chiefly located at the interface of tumor

Fig. 4. In vivo efficacy and tolerability of mAb FAP5-maytansinoid conjugates in different carcinoma xenograft models. A, nude mice bearing PAXF736 tumors were treated with a single i.v. dose of either vehicle, mAb FAP5-DM1, mAb FAP5-DM4, or mAb FAP5-SMCC-DM1at 800 Ag/kg maytansinoid. The tumor volume of each treatment group at day 45 is shown. B, FaDu tumor-bearing mice treated once with 600 Ag/kg maytansinoid. Tumor volumes at day 19 are shown. C, nude mice bearing LXFA629 tumors treated with three weekly doses of the conjugates indicated above at 400 Ag/kg maytansinoid.Tumor volumes at day 41are shown. Columns, median tumor volume of each group. D, average body weight change (in percentage) of mice bearing the LXFA629 tumors during treatment.

www.aacrjournals.org 4589 Clin Cancer Res 2008;14(14) July 15, 2008 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2008 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 5. Mechanism of action of mAb FAP5-DM1in human epithelial cancer xenograft models. Histologic and immunophenotype analysis of tumor samples obtained during treatment. Pretherapy, histologic appearance of LXFA629 tumors from control mice with an abundant fibroblastic stroma with prominent FAPa expression (red arrow) and a well-developed network of tumor capillaries (EC). 4 8 h post-therapy, changes in tumor morphology 48 h after second i.v. administration of mAb FAP5-DM1.Clusters of enlarged tumor cells surrounded by extensive inflammatory infiltrates.The tumor cells show cytologic abnormalities consistent with mitotic spindle poisoning; some of these cells are arrested in mitosis (pH3, phospho-histone H3 staining), whereas others showed signs of cell death by apoptosis (green, fluorescence terminal nucleotidyl transferase ^ mediated dUTP nick end labeling assay).The inflammatory infiltrates consist mostly of CD11b+ tissue macrophages (MU; green, CD11b + macrophages; red, DNA counterstaining propidium iodide). A loss of collagen type IV marker protein indicative of a disruption of the basement membrane is seen in the treated tumors (white arrows) compared with the untreated tumors (BM ; inset, red) where the basement membrane directly surrounds the clusters of tumor cells stained with cytokeratin18 (green). Late effects, at the end of the experiment (day 45), the remaining nodules are composed of an acellular, collagenized stroma, with microcalcifications (asterisk), dispersed small capillaries indicative of tumor capillary damage, and isolated nests of tumor cells. capillaries and malignant epithelium (Fig. 5, pretherapy). terminal nucleotidyl transferase–mediated dUTP nick end Further analyses were carried out on tumors excised 2, 3, or labeling staining (Fig. 5, 48 h post-therapy). In addition, we 5 days after therapy. At day 2 post-therapy, the tumor nodules determined that the inflammatory cell infiltrate consists show histologic evidence of necroses and a marked inflamma- predominantly of CD11b/F4/80+ macrophages surrounding tory cell infiltrate, evidence of a mitotic spindle poison in the clusters of tumor cells and in some instances the isolated malignant epithelial cells with cell cycle arrest as evidenced by single tumor cells. Finally, we found that the basement the phospho-histone H3 marker and apoptosis shown by membrane of malignant epithelial cell clusters is disrupted

Clin Cancer Res 2008;14(14) July 15, 2008 4590 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2008 American Association for Cancer Research. Antistroma Tumor Therapy with loss of collagen type IV marker protein (white arrows) in vivo targeting, cell membrane binding and internalization, compared with the preserved basement membrane in untreated lysosomal activation, intracellular modification, cellular re- control tumors (inset). When tumors were analyzed at the end lease, and microtubule disruption and apoptosis. of the experiment (day 45), we observed minute s.c. nodules The initial targeting of mAb FAP5-DM1 to FAPa+ tumor with an acellular, collagenized stroma, with microcalcifications, stromal fibroblasts in vivo is likely to be fast and efficient, isolated nests of tumor cells, and disruption of the blood vessel because we have shown that these target cells surround tumor network in which remaining vessels were of small caliber and capillaries and this proximity should lower diffusion barriers showed less uniform distribution in the tissue (Fig. 5, late faced by antibodies targeting the malignant cells. Previous effects). studies with the human epitope-specific mAb F19 and sibrotuzumab in patients have shown selective tumor detection Discussion (17, 19, 41). Cell culture studies have suggested that cell surface binding of drugs with the general structure mAb-SPP-DM1 (like Novel therapeutic principles (31), such as imatinib or mAb FAP5-DM1) leads to antigen-dependent endocytosis of rituximab, have had a strong effect on objective response rates the immunoconjugate and lysosomal degradation with intra- and survival in patients with certain leukemias and lympho- cellular release of a lysine derivative, lysine-Nq-SPP-DM1, mas (32–34). By comparison, targeted approaches for solid which binds microtubules and causes mitotic arrest and cancers have shown lower response rates and less survival apoptosis. Subsequently, a further metabolite, the neutral, benefits (35, 36), which may reflect highly disparate and lipophilic S-methyl-DM1, can convey bystander effect by redundant genetic aberrations accumulated during epithelial crossing cell membranes and reentering target cells with high carcinogenesis, pathway activation less amenable to drug potency as an antimitotic agent (24). discovery, or confounding patterns of tissue invasion and This process of intracellular immunoconjugate activation metastasis. and tumor cell apoptosis is straightforward when the drug Along with notable advances in some indications (37), there binds to the highly proliferative cancer cells directly, reflecting has been a remarkable attrition for antibodies against solid a mode of ‘‘suicide prodrug activation.’’ The process is more tumors that appeared promising in preclinical testing but failed intriguing for tumor stromal fibroblast targeting, because in clinical trials. In our preclinical testing, we focused on three activated fibroblasts are generally quiescent, not actively aspects. proliferating within tumors as seen by low mitotic index First, for antibody-based drugs, the selectivity of target (data not shown). Thus, whereas mAb FAP5-DM1 binding, antigen expression in cancer versus normal tissues remains internalization, and generation of lysine-Nq-SPP-DM1 and the defining characteristic (38–40). We selected a unique S-methyl-DM1 likely occur in the FAPa+ tumor stromal antigen, FAPa, based on its preferential expression in cancer fibroblasts, these cells do not attempt to enter mitosis and tissues (6, 7, 16, 27), and radiolabeled anti-FAPa antibodies undergo apoptosis. Instead, they perform a ‘‘transit activa- have shown in vivo tumor targeting in cancer patients (17, 19). tion’’ step for the prodrug, remain viable for repeat activation FAPa is distinctive in its cancer distribution, as it is not cycles, and presumably release the uncharged, lipophilic expressed by the malignant epithelial cells but rather by the S-methyl-DM1 as the essential antimitotic moiety into their nontransformed, activated stromal fibroblasts. microenvironment. At present, no methods exist to measure Second, particular care was taken to design antibodies that this release in cancer models in situ, but our histologic and cross-react with human and mouse FAPa to avoid the limited biomarker studies indeed identify malignant epithelial cells relevance of human-mouse xenograft studies for antibodies adjacent to fibroblastic stroma as early responders, with binding exclusively to human-specific epitopes. mitotic spindle arrest phenotypes, up-regulation of the Third, we ensured that the tissue architecture of our xenograft phospho-histone H3 marker, and apoptosis induction solely models is consistent with the corresponding human carcino- in epithelial clusters rather than ablation of the stromal mas. The histotypic models used here recapitulate the hall- compartment. In line with this sequence, the multidrug- marks of human carcinomas. Of note, advances in the resistant CXF158 model shares a distinct, FAPa+ fibroblastic purification of CD133+ human colon cancer-initiating cells stroma with the three responsive tumor models but is from surgical specimens should now allow routine generation unresponsive to mAb FAP5-DM1 therapy; we postulate that of xenograft models that reproduce the histomorphologic transit activation with release of S-methyl-DM1 occurs in this features of the original cancers (21, 22). model also, but S-methyl-DM1 is a substrate for the In our search for more effective drugs to treat epithelial multidrug resistance efflux transporters in this model (42). cancers, we have developed an experimental paradigm for Further confirmation comes from the fact that the uncleav- tumor stromal fibroblast-targeted therapy. As the major finding able mAb FAP5-SMCC-DM1 analogue of mAb FAP5-DM1 has from our study, mAb FAP5-DM1 shows an impressive level of no in vivo antitumor activity in our models despite its potent, therapeutic benefit, with rapid onset of action, complete direct effect on FAPa-transfected HT1080 cells. The thioether- responses, and long-lasting suppression of tumor growth. In linker in this compound, SMCC, is not cleaved inside cells, all three xenograft models, the treatment was exceptionally well and the major metabolite generated by lysosomal degrada- tolerated considering the cross-reactivity with endogenous tion, lysine-Nq-SMCC-DM1, is not converted to the cell- mouse FAPa. Considering that only a small fraction of the permeable and highly potent derivative, S-methyl-DM1, the tumor mass in these models represents the stromal compart- essential active principle in our concept of stromal-fibroblast ment with FAPa+ activated fibroblasts, it is tempting to targeted therapy. speculate about the potent mode of action of the drug. Several In conclusion, we have developed a novel experimen- steps may be considered along the known sequence (38) of tal paradigm for tumor therapy based on maytansinoid

www.aacrjournals.org 4591 Clin Cancer Res 2008;14(14) July 15, 2008 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2008 American Association for Cancer Research. Cancer Therapy: Preclinical immunoconjugates targeting stromal fibroblasts. In view of the Disclosure of Potential Conflicts of Interest substantial therapeutic efficacy and excellent tolerability seen with our lead compound, mAb FAP5-DM1, in animal models, The authors declare that they have no competing financial interests. combined with the feasibility of safety assessments in com- monly used animal species due to cross-reactivity with Acknowledgments autologous target proteins, this concept may be poised for a rapid transition to clinical development. If successful, stroma- We thank I. Apfler, B. Rohrhan, B. Pichler, and M. Zehetner for the excellent directed therapy may provide clinical benefits in a broad technical assistance; P. Adam for performing internalization experiments; spectrum of indications, based on the consistent presence and, E. Borges for discussion; B. Enenkel for antibody production; Brenda Kellogg, Erin Maloney, Rajeeva Singh, and Dapeng Sun (ImmunoGen) for producing sometimes, remarkable abundance of stromal compartments in immunoconjugates; and Prof. H.H. Fiebig and Dr. T. Metz (Oncotest) for contrib- human carcinomas. uting the tumor models.

References 1. Joyce JA. Therapeutic targeting of the tumor micro- 15. Brown DD, Wang Z, Furlow JD, et al. The thyroid 28. Huber MA, Kraut N, Schweifer N, et al. Expres- environment. Cancer Cell 2005;7:513^ 20. hormone-induced tail resorption program during Xen- sion of stromal cell markers in distinct compart- 2. Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev opus laevis metamorphosis. Proc Natl Acad Sci U S A ments of human skin cancers. J Cutan Pathol Cancer 2006;6:392 ^401. 19 9 6;93 :1924 ^ 9. 2006;33:145 ^55. 3. Orimo A, Gupta P, Sgroi DC, et al. Stromal fibroblasts 16. DolznigH,SchweiferN,PuriC,etal.Character- 29. Kovtun YV, Audette CA, YeY, et al. Antibody-drug present in invasive human breast carcinomas promote ization of cancer stroma markers: in silico analysis conjugates designed to eradicate tumors with homo- tumor growth and angiogenesis through elevated of an mRNA expression database for fibroblast acti- geneous and heterogeneous expression of the target SDF-1/CXCL12 secretion. Cell 2005;121:335 ^ 48. vation protein and endosialin. Cancer Immunity antigen. Cancer Res 2006;66:3214 ^ 21. 4. Mueller MM, Fusenig NE. Friends or foesöbipolar 2005;3:5 ^ 10. 30. Widdison WC, Wilhelm SD, Cavanagh EE, et al. effects of the tumour stroma in cancer. Nat Rev Can- 17. Welt S, Divgi CR, Scott AM, et al. Antibody targeting Semisynthetic maytansine analogues for the tar- cer 2004;4:839 ^ 49. in metastatic colon cancer: a phase I study of mono- geted treatment of cancer. J Med Chem 2006;49: 5. Huber MA, Kraut N, Beug H. Molecular requirements clonal antibody F19 against a cell-surface protein of 4392 ^408. for epithelial-mesenchymal transition during tumor reactive tumor stromal fibroblasts. J Clin Oncol 1994; 31. Imai K, Takaoka A. Comparing antibody and small- progression. Curr Opin Cell Biol 2005;17:548 ^ 58. 12 :119 3 ^ 2 0 3. molecule therapies for cancer. Nat Rev Cancer 2006; 6. RettigWJ, Garin-Chesa P, Beresford HR, Oettgen HF, 18. Hofheinz R-D, al-Batran S-E, Hartmann F, et al. 6:714^27. Melamed MR, Old LJ. Cell-surface glycoproteins of Stromal antigen targeting by a humanised monoclonal 32. Li Y, Zhu Z. Monoclonal antibody-based thera- human sarcomas: differential expression in normal antibody: an early phase II trial of sibrotuzumab in peutics for leukemia. Expert Opin Biol Ther 2007;7: and malignant tissues and cultured cells. Proc Natl patients with metastatic colorectal cancer. Onkologie 319 ^ 30. Acad Sci U S A1988;85:3110^4. 2003;26:44 ^ 8. 33. Fanale MA,Younes A. Monoclonal antibodies in the 7. Garin-Chesa P, Old LJ, RettigWJ. Cell surface glyco- 19. Scott AM,Wiseman G,Welt S, et al. A phase I dose- treatment of non-Hodgkin’s lymphoma. Drugs 2007; protein of reactive stromal fibroblasts as a potential escalation study of sibrotuzumab in patients with 67:333 ^ 50. antibody target in human epithelial cancers. Proc Natl advanced or metastatic fibroblast activation protein- 34. Deininger M, Buchdunger E, Druker BJ. The Acad Sci U S A1990;87:7235 ^ 9. positive cancer. Clin Cancer Res 2003;9:1639 ^ 47. development of imatinib as a therapeutic agent 8. Scanlan MJ, RajBK, Calvo B, et al. Molecular cloning 20. Scholz CC, Berger DP,Winterhalter BR, Henso H, for chronic myeloid leukemia. Blood 2005;105: of fibroblast activation protein a, a member of the ser- Fiebig HH. Correlation of drug response in patients 2640 ^ 53. ine protease family selectively expressed in stromal and in the clonogenic assay with solid-tumour xeno- 35. Baselga J, Arteaga CL. Critical update and emerg- fibroblasts of epithelial cancers. Proc Natl Acad Sci grafts. Eur J Cancer 1990;26:901 ^ 5. ing trends in epidermal growth factor receptor target- U S A1994;91:5657^61. 21. Ricci-Vitiani L, Lombardi DG, Pilozzi E, et al. Identifi- ing in cancer. J Clin Oncol 2005;23:2445 ^ 59. 9. ParkJE, Lenter MC, Zimmermann RN, Garin-Chesa P, cation and expansion of human colon-cancer-initiating 36. Pao W, Miller VA, Politi KA, et al. Acquired resis- Old LJ, RettigWJ. Fibroblast activation protein, a dual cells. Nature 2007;445:111^ 5. tance of lung adenocarcinomas to gefitinib or erlotinib specificity serine protease expressed in reactive hu- 22. O’Brian CA, Pollett A, Gallinger S, DickJE. A human is associated with a second mutation in the EGFR ki- man tumor stromal fibroblasts. J Biol Chem 1999; colon cancer cell capable of initiating tumour growth nase domain. PLoS Med 2005;2:e73. 274:36505^12. in immunodeficient mice. Nature 2007;445:106 ^ 10. 37. Reichert JM, Rosensweig CJ, Faden LB, Dewitz 10. Niedermeyer J, Enenkel B, Park JE, et al. Mouse fi- 23. ChariRVJ,MartellBA,GrossJL,etal.Immunocon- MC. Monoclonal antibody successes in the clinic. broblast activation protein. Conserved Fap gene orga- jugates containing novel maytansinoids: promising Nat Biotechnol 2005;23:1073 ^ 8. nization and biochemical function as a serine anti-cancer drugs. Cancer Res 1992;52:127 ^ 31. 38. Schrama D, Reisfeld RA, Becker JC. Antibody tar- protease. Eur J Biochem 1998;254:650 ^ 4. 24. Erickson HK, Park PU, Widdison WC, et al. Anti- geted drugs as cancer therapeutics. Nat Rev Drug 11. Lee KN, Jackson KW, Christiansen VJ, Chung SL, body-maytansinoid conjugates are activated in tar- Discov 2006;5:147 ^ 59. Jin-Geun C, McKee PA. Antiplasmin-cleaving enzyme geted cancer cells by lysosomal degradation and 39. Wu AM, Senter PD. Arming antibodies: prospects is a soluble form of fibroblast activation protein. Blood linker-dependent intracellular processing. Cancer Res and challenges for immunoconjugates. Nat Biotechnol 2006;107:1397 ^ 404. 2006;66:4426^33. 2005;23:1137^ 46. 12. Rettig WJ, Garin-Chesa P, Healey JH, et al. Regula- 25. Xie H, Audette C, Hoffee M, Lambert JM, Bla« ttler 40. Lambert JM. Drug-conjugated monoclonal anti- tion and heteromeric structure of the fibroblast activa- WA. Pharmacokinetics and biodistribution of the antibodies for the treatment of cancer. Curr Opin Pharma- tion protein in normal and transformed cells of tumor immunoconjugate, cantuzumab mertansine col 2005;5:543^9. mesenchymal and neuroectodermal origin. Cancer (huC242-1), and its two components in mice. J Phar- 41. Tahtis K, Lee F-T,WheatleyJM, et al. Expression and Res 1993;53:3327^35. macol ExpTher 2004;308:1073 ^82. targeting of human fibroblast activation protein in a 13. Niedermeyer J, Scanlan MJ, Garin-Chesa P, et al. 26. NiedermeyerJ, Kriz M, Hilberg F, et al. Targeted dis- human skin/severe combined immunodeficient mouse Mouse fibroblast activation protein: molecular clon- ruption of mouse fibroblast activation protein. Mol breast cancer xenograft model. Mol Cancer Ther ing, alternative splicing and expression in the reactive Cell Biol 2000;20:1089 ^ 94. 2003;2:729 ^ 37. stroma of epithelial cancers. Int J Cancer 1997;71: 27. Huber MA, Kraut N, Park JE, et al. Fibroblast acti- 42. Gillet J-P, Efferth T, Steinbach D, et al. Microarray- 383^9. vation protein: differential expression and serine pro- based detection of multidrug resistance in human 14. NiedermeyerJ, Garin-Chesa P, Kriz M, et al. Expres- tease activity in reactive stromal fibroblasts of tumor cells by expression profiling of ATP-binding sion of the fibroblast activation protein during mouse melanocytic skin tumors. JInvest Dermatol 2003;120: cassette transporter genes. Cancer Res 2004;64: embryo development. Int J Dev Biol 2001;45:445 ^ 7. 182 ^ 8. 8987^93.

Clin Cancer Res 2008;14(14) July 15, 2008 4592 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2008 American Association for Cancer Research. Effective Immunoconjugate Therapy in Cancer Models Targeting a Serine Protease of Tumor Fibroblasts

Elinborg Ostermann, Pilar Garin-Chesa, Karl Heinz Heider, et al.

Clin Cancer Res 2008;14:4584-4592.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/14/14/4584

Cited articles This article cites 42 articles, 18 of which you can access for free at: http://clincancerres.aacrjournals.org/content/14/14/4584.full#ref-list-1

Citing articles This article has been cited by 26 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/14/14/4584.full#related-urls

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://clincancerres.aacrjournals.org/content/14/14/4584. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.