Lisa Demchik

Cathepsin B and glioma invasion

Increased expression of cathepsin B has been reported in a number of human and animal tumors. This has also been observed in human gliomas where increases in cathepsin B mRNA, protein, activity and secretion parallel malignant progression. In the present study, we showed that cathepsin B was directly involved in glioma cell invasion. Activity of cathepsin B was an order of magnitude higher in glioma tissue than in matched normal brain. Inhibitors of cysteine proteases reduced invasion of glioma cells in two in vitro models: invasion through Matrigel and infiltration of a glioma spheroid into a normal brain aggregate. Glioma spheroids expressed higher levels of cathepsin B than did monolayers and the ability of subclones differing in cathepsin B activity to infiltrate normal brain aggregates paralleled their cathepsin B activity. We confirmed that intracellular staining for cathepsin B occurs at the cell periphery and in cell processes and observed extracellular staining on the cell surface. In addition, we demonstrated that intracellular cathepsin B located at the cell periphery and in processes was active. The cell surface cathepsin B colocalized with areas of degradation of an extracellular matrix component. We hypothesize that the increased expression of active cathepsin B in gliomas leads to increases in invasion in vitro and in vivo and have developed a xenotransplant model in which this hypothesis can be tested.

In Vivo Methods for Screening and Preclinical Testing

The classic question in the field of drug discovery is: Which tumor model is a satis-factory predictor for cancer in humans? The classic answer is: None of them!

Preclinical antitumor activity of XK469 (NSC 656889).

XK469 (NSC 656889) is a water-soluble member of the novel quinoxaline family of antitumor agents. In vitro, XK469 demonstrated selective cytotoxicity for several murine solid tumors including colorectal and mammary adenocarcinoma cell lines, when compared to both leukemia and normal epithelial cells. In vivo, XK469 was active against 7/7 murine tumors tested, including pancreatic ductal carcinomas #02 and #03, colon adenocarcinomas #38 and #51/A, mammary adenocarcinoma #16/C and the Adriamycin resistant mammary adenocarcinomas #16/C/ADR and #17/ADR. XK469 was efficacious both intravenously and orally. Regardless of dosing schedule, conventional mice tolerated higher total doses than SCID or nu/nu mice did. Despite these reduced doses, XK469 was active against xenografts of 4/6 human tumor lines including mammary adenocarcinoma MX-1, the small cell lung cancer DMS 273, the prostate model LNCaP and the CNS tumor SF295. The lower doses in the xenograft studies were below curative levels. The dose-limiting toxicity appeared to be myelosuppression with rapid host recovery (5–8 days), and in vitro assays of XK469 toxicity to murine bone marrow neutrophil progenitors CFU-GM (colony forming unit-granulocyte/macrophage) demonstrated concentration-dependent toxicity from 0.5–30 μg/mL. The difference in drug tolerance between BDF1 and SCID mice was detected in vitro as a 3-fold difference in the IC90 for CFU-GM, despite similar IC50 values. Comparative in vitro hematotoxicology studies revealed that human bone marrow CFU-GM tolerated XK469 as well as their SCID counterparts (IC90 values 5.7 vs. 7.4 μg/mL). Based on comparison with previously tested anti-cancer agents, these data suggest that humans will be able to tolerate XK469 doses that are efficacious against human tumor xenografts.

Discovery of cryptophycin-1 and BCN-183577 : Examples of strategies and problems in the detection of antitumor activity in mice

Historically, many new anticancer agents were first detected in a prescreen; usually consisting of a molecular/biochemical target or a cellular cytotoxicity assay. The agent then progressed to in vivo evaluation against transplanted human or mouse tumors. If the investigator had a large drug supply and ample resources, multiple tests were possible, with variations in tumor models, tumor and drug routes, dose-decrements, dose-schedules, number of groups, etc. However, in most large programs involving several hundred in vivo tests yearly, resource limitations and drug supply limitations have usually dictated a single trial. Under such restrictive conditions, we have implemented a flexible in vivo testing protocol. With this strategy, the tumor model is dictated by in vitro cellular sensitivity; drug route by water solubility (with water soluble agents injected intravenously); dosage decrement by drug supply, dose-schedule by toxicities encountered, etc. In this flexible design, many treatment parameters can be changed during the course of treatment (e.g., dose and schedule). The discovery of two active agents are presented (Cryptophycin-1, and Thioxanthone BCN 183577). Both were discovered by the intravenous route of administration. Both would have been missed if they were tested intraperitoneally, the usual drug route used in discovery protocols. It is also likely that they would have been missed with an easy to execute fixed protocol design, even if injected IV.

Discovery of Solid Tumor Active Agents Using a Soft-Agar-Colony-Formation Disk-Diffusion-Assay

The history of antitumor drug discovery has essentially been the use of two lymphocytic leukemias of mice as selection funnels through which all agents needed to pass in order to advance toward clinical development (L1210 prior to 1975 and P388 after 1975). It is thus not surprising that agents in the clinic are highly active against these tumor systems. However, none of the agents discovered by these leukemias are tumor specific (i.e., active against all tumors), and none of the agents are broadly active against solid tumors of either rodents or humans (1, 2, 3). An example contrasting the responsiveness of transplantable solid tumors of mice and the two leukemias is shown in Table-1. The lack of responsiveness of these solid tumors of mice is not unlike those seen in human lung, pancreatic, colon, and prostate tumors. The point to emphasize is that the lack of solid tumor activity of available antitumor agents is not species related. The fault does not lie with the omission of human tumors in the initial selection process, but rather with the omission of solid tumors.

Preclinical antitumor activity of CI-994

SummaryCI-994 [aka: acetyldinaline; PD 123654; 4-acetylamino-N-(2′aminophenyl)-benzamide] (Figure 1) is a novel antitumor agent with a unique mechanism of action. It is the acetylated metabolite of dinaline, a compound previously identified as having cytotoxic and cytostatic activity against several murine and human xenograft tumor models. CI-994 had activity against 8/8 solid tumors tested (log cell kills at the highest non-toxic dose): pancreatic ductal adenocarcinoma #02 (4.7); pancreatic adenocarcinoma#03 (3.0; 1/6 cures); colon adenocarcinoma #38 (1.6); colon adenocarcinoma #51/A (1.1); mammary adenocarcinoma #25 (1.7); mammary adenocarcinoma #17/ADR (0.5); Dunning osteogenic sarcoma (4.0); and the human prostate carcinoma LNCaP (1.2). CI-994 had the same spectrum of activity in vivo as dinaline. It also behaved similarly in schedule comparison/toxicity trials. Prolonged administration with lower drug doses was more effective than short-term therapy at higher individual doses. If doses were kept between 40 and 60 mg/kg/injection, prolonged administration (> 50 days) was tolerated with no gross toxicity. Doses ≥90 mg/kg/injection caused lethality after 4–5 days of administration. The maximum tolerated total dose was also increased with smaller individual doses administered for prolonged intervals. Clinical Phase I trials are ongoing with this agent.

Tumor Models and the Discovery and Secondary Evaluation of Solid Tumor Active Agents

AbstractEach independently arising tumor is a separate and unique biologic entity with its own unique histologic appearance, biologic behavior, and drug response profile. Thus, in drug discovery, no single tumor has been a perfect predictor for any other tumor. For this reason, new agents are evaluated in a variety of tumor models which is known as breadth of activity testing. In recent years, human tumors implanted in athymic nude mice and SCID mice have also become available for breadth of activity testing. In studies carried out in these laboratories, it was found that 10 human tumors metastasized in the SCID mice, but failed to metastasize in nude mice. In addition, tumor growth and tumor takes were superior in the SCID mice. The strengths and weaknesses of xenograft model systems are discussed. For example, most human tumor xenograft models are excessively sensitive to alkylating agents as well as to a new class of DNA binders (XE840 and XP315). Using human tumor models that are the least sensitive t...

Observing Proteases in Living Cells

The lysosomal cysteine protease cathepsin B has been implicated in tumor progression and metastasis in part due to its altered trafficking. In order to analyze the trafficking of cathepsin B in living cells, we utilized enhanced green fluorescent protein (EGFP) fused to various cathepsin B constructs for transfecting two cell lines: an invasive human breast adenocarcinoma cell line (BT20) and a cathepsin B deficient mouse embryonic fibroblast cell line (MEF T -/-). The cells were transiently transfected with four cathepsin B-EGFP fusion constructs: full-length preprocathepsin B-EGFP, cathepsin B preregion-EGFP, cathepsin B prepro region-EGFP, and cathepsin B prepro region-EGFP with a mutation of the glycosylation site in the pro region. The full length construct showed vesicular distribution throughout the cells in both cell lines. In both BT20 and MEF T -/- cells, preregion-EGFP was localized in a ring tightly associated with the cell nucleus, suggesting distribution to the endoplasmic reticulum. The distribution of the prepro region-EGFP construct was similar except that it also included some patchy areas adjacent to the nucleus. This suggested that the cathepsin B prepro region-EGFP might have entered the Golgi. Distribution of the mutated cathepsin B prepro region-EGFP was similar to that of wild-type prepro region-EGFP in the MEF T -/-. In the invasive BT20 cells, however, the mutated prepro region-EGFP showed a vesicular distribution throughout the cytoplasm and in cell processes. This distribution is similar to that of endogenous cathepsin B in these cells. Our results suggest that: 1) tumor cells have an alternative mechanism for trafficking of cathepsin B which is independent of the mannose-6-phosphate receptor pathway, and 2) the pro region of cathepsin B may contain the sorting sequence necessary for its trafficking via this pathway.

Comparative efficacy of DMP 840 against mouse and human solid tumor models

SummaryBackgroundDMP 840 is a compound from a class of bis-naphthalimide antitumor agents that recently completed Phase I clinical trials at three North American centers and is currently undergoing Phase II testing. Preclinically, it was shown to have curative activity against a variety of human tumor xenograft models.PurposeTo test DMP 840 bothin vitro andin vivo for antiproliferative activity against predominantly mouse tumor models.MethodsA disk diffusion soft agar colony formation assay was used to determine thein vitro growth inhibitory activity against a selection of mouse and human tumor cell lines, and the comparable selective mouse solid tumors were used forin vivo testing.ResultIn vitro DMP 840 exhibited equal cytotoxicity for human tumors (including MX-1 directly cultured from nude mice), mouse tumors and normal cells.In vivo DMP 840 was only modestly active or inactive against the following mouse tumors: Mam 16/C, T/C=30% (T/C=Percent Tumor Growth Inhibition); Mam 16/C/ADR, T/C=33%; Colon 38, T/C=9%; Panc 03, T/C=53%; Colon 51/A, T/C=28%; Pane 02, T/C= 52%; P388/0, 36% ILS (Percent Increased Life Span) and P388/ADR, 14% ILS. Furthermore, the antitumor activity was only observed at the highest non-toxic dose and was associated with a large body weight loss. In contrast, the agent was highly active against the human breast tumor MX-1 implanted subcutaneously in either athymic nude or SCID mice (Nudes: T/C=0%; 1/5 cures; SCIDS: T/C=0%; 5/5 cures).ConclusionsAlthough there was no selective cytotoxicity in our clonogenic assay for human versus mouse tumor cell lines, selective activityin vivo for human xenograft tumors was noted. Overall, this compound is rather unique in its differential degree ofin vivo activity for human versus mouse tumors.Implications: Phase II trials, which are ongoing, will help determine if the preclinicalin vivo selective activity of DMP 840 translates to clinical activity in man.

Cell-surface proteases in cancer

Proteases of the serine, cysteine, aspartic, and metallo classes are capable of degrading extracellular matrix and basement membrane components thus contributing to the invasive and metastatic properties of tumor cells. The localization of proteases to tumor cell surfaces may be important for their ability to invade and may also be a way to regulate protease expression and activity. Proteases found to be associated with the tumor cell surface and which are also correlated with tumor invasion and metastasis include: urokinase plasminogen activator, cathep-sin B, and matrix metalloproteinase-2. Cell surface receptors have been identified for urokinase plasminogen activator and matrix metalloproteinase-2. Although regulation of protease expression and activity can occur at several levels, the interaction of proteases with cell surface receptors may be most crucial to protease function. Increased expression of proteases, cell surface receptors, and endogenous inhibitors have been observed in many cancers.