Novel Human Single Chain Antibody Fragments That Are Rapidly Internalizing Effectively Target Epithelioid and Sarcomatoid Mesotheliomas

Malignant mesothelioma (MM), caused primarily by exposure to asbestos, is a highly aggressive tumor arising from serosal surfaces of the pleura, peritoneum, and pericardium (1, 2). MM has 3 major subtypes; epithelioid (EM) that is more likely to respond to treatment and accounts for 50% to 70% of all cases, sarcomatoid (SM) that rarely responds to any treatment and represents 7% to 20%, and mixed/biphasic for the remaining 20% to 30%. Patients with MM have a median survival time of 8 to 14 months (3). There is an urgent unmet need to develop new diagnostics and therapeutics for MM (4) as the disease has a long latency period with past and ongoing exposure to asbestos contributing to the development of new cases worldwide.

One approach to detect and treat cancer is to conjugate imaging and/or therapeutics to molecules which can recognize internalizing antigens, receptors or cell surface markers that are overexpressed on tumor cells, leading to efficient localization and tumor cell killing (5, 6). However, presently there are very few MM-associated cell surface antigens that are overexpressed by all subtypes of MM, especially the SM (7). One well established marker—mesothelin, a 40 kDa cell surface glycoprotein, has been reported to be expressed by EM cells (8), but not SM (9). Recently we have identified a panel of internalizing human scFv antibodies by phage display selection that target cell surface antigens associated with both EM and SM (6). The selected scFvs bind to human mesothelioma cells in situ, thereby recognizing clinically represented tumor antigens (6) and thus offer the potential to deliver high levels of imaging probes to tumor cells but not to nontarget normal tissues based on intracellular delivery strategies. By screening the yeast surface human cDNA display library with mesothelioma targeting phage antibody, we have further identified MCAM/CD146/MUC18 as one of the target antigens for MM cells that was overexpressed in more than 80% of EM and SM tissues, but not other tissues (10). In the present study, we investigated both the in vitro and in vivo tumor targeting and imaging potential of an additional scFv (M40) for both EM (M28) and SM (VAMT-1) subtypes.

The M40 was produced as previously reported (6, 11–13).

The scFv was radiolabeled as previously reported (14, 16). The carbonyl-kit (IsoLink Tyco/Mallinckrodt) was used to prepare the [^99mTc(CO)₃] moiety. An aliquot (40–60 μg) of M40 was mixed with 100 to 500 μL of [^99mTc(CO)₃(OH₂)₃]⁺ solution and the mixture was heated at 37°C for 60 minutes. The product was purified using a PD-10 column. Labeling efficiency and purity were determined by size exclusion HPLC and thin-layer chromatography (TLC).

The M40 was labeled with Cy5.5 by incubation with 3:1 molar excess of Cy5.5-NHS ester in a carbonate/bicarbonate buffer (pH 7.2–8.5) for 1 hour followed by purification using PD-10 column.

Internalization experiments were performed as described previously (12, 15, 16). Briefly, 1 million VAMT-1, M28, or BPH-1 cells (control) were incubated with 0.05 to 2 μCi ^99mTc-M40 in various concentrations for 1 or 3 hours. All cell lineshad been tested for mycoplasma contamination and characterized by cell proliferation and morphology evaluation (6). The cells were washed to determine cell surface–bound (acid releasable) and internalized (acid resistant) radioligand/radioactivity expressed as the percentage of applied activity normalized to 1 million cells. For nonspecific uptake, the aforementioned procedure was repeated after addition of 10-fold excess unlabeled M40. For microscopy study, 1 μmol/L (10 μL) of Cy5.5-M40 was incubated with tumor cells for 1 and 3 hours. The cells were washed with PBS and then imaged using a TE2000-E Fluorescence Microscope (Nikon Inc.) at 20× magnification.

Animal procedures were performed according to a protocol approved by the UCSF Animal Care and Use Committee. Six-week-old male nu/nu mice were purchased from Charles River Laboratories. For tumor inoculation, 10⁶ M28 and VAMT-1 cells in 200 μL of PBS were administered subcutaneously into the right and left shoulders of the animal, respectively. The mice were studied when tumor size reached approximately 3 to 5 mm in diameter (∼20–60 mm³ in volume). Tumor mice in groups of 4 animals were injected each with 18.5 to 37 MBq (0.5–1.0 mCi) of ^99mTc-M40 containing approximately 2 to 4 μg of scFv. A control group (blocking study) were injected with 10-fold excess unlabeled M40 1 hour before ^99mTc-M40. The biodistribution at 1, 3, or 6 hours of the study group was determined and compared with that of the control group.

Mice were imaged with a small animal SPECT/CT system (GE healthcare). For anatomical correlation, CT was first performed after injection. SPECT imaging was initiated 1 and 3 hours after injection.

Mice were fasted overnight prior to ¹⁸F-FDG injection of 3.7 to 7.4 MBq [100–200 μCi] and imaged with a microPET/CT system (Siemens Medical Solutions USA, Inc.) 1 hour after injection for a 20-minute acquisition time.

The ^99mTc labeling yield of M40 (from [^99mTc(CO)₃(OH₂)₃]⁺) was 70% to 85%. The radiochemical purity of the ^99mTc-M40 was greater than 95%. The final ^99mTc-M40 had a high specific activity of approximately 9.3 MBq/μg (6.3 x10⁵ Ci/mol). The ^99mTc radiolabeling was stable in phosphate buffer for at least the duration investigated (24 hours), as reported previously (14, 16).

The binding affinity of ^99mTc-M40 to the mesothelioma cells exhibited an apparent K_d of 20 to 21 nmol/L and the antigen density for M40 derived from the saturation cell binding assay was comparable on both tumor cells (Fig. 1A, 1B). M40 was rapidly internalized over the concentrations tested with about 68% to 92% of total cell-associated uptake, following 1 hour incubation at 37°C (Fig. 1C), whereas the binding and uptake in the control BPH-1 cells was much less (Fig. 1C). Specificity was further demonstrated by blocking of uptake/ internalization into both M28 and VAMT-1 (>95%) cells with a 10-fold excess of unlabeled M40 (Fig. 1C).

As shown in Figure 1D, Cy5.5-M40 was rapidly internalized into mesothelioma cells within 1 hour after incubation at 37°C, whereas there was negligible uptake in the control (BPH-1) cells under identical conditions.

Mice were imaged with small animal SPECT/CT at 1 and 3 hours after injection with concomitant assessment of ex vivo biodistribution at 1, 3, and 6 hours (Fig. 2A). Imaging with SPECT/CT showed high tumor uptake (Fig. 2B) as early as 3 hours postinjection. At 3 hours, tumor uptake was 4.38%ID/g and 5.84%ID/g for M28 and VAMT-1 tumors, respectively (Table 1), greater than for all organs/tissues studied (liver 2.62%ID/g, other organs/tissues <2%ID/g) except the kidney (130.7%ID/g), giving M28 and VAMT-1 tumor/muscle ratios of 45:1, 60:1, and tumors/blood ratios of 5:1 and 7:1, respectively (Fig. 2A). In contrast, the control blocking study with 10-fold excess of unlabeled M40 at 3 hours reduced by more than 70% of tumor uptake of the ^99mTc-M40 to 0.9%ID/g and 1.6%ID/g for M28 and VAMT-1, respectively (Table 1 and Fig. 2C).

¹⁸F-FDG PET imaging was performed to confirm the distinct identity of the 2 subtypes of tumors in a single animal. As shown in Figure 2D, the ¹⁸F-FDG PET/CT image at 1 hour detected SM (VAMT-1) preferentially.

We report here the in vitro and in vivo tumor targeting and imaging potential of a rapidly internalizing human scFv (M40) selected from a panel of scFvs targeting internalizing epitopes present on both EM and SM cells (6).

Several important findings have been uncovered in this study. First, the ^99mTc-M40 showed strikingly rapid and selective binding and internalizing ability in vitro into the mesothelioma tumor cells but not into the control nontumorigenic cells, consistent with our previous findings (6). Although we did not perform the Lindmo assay (17) to determine what percentage of our labeled antibodies retained reactivity (the immunoreactive fraction), our results nonetheless demonstrate that the M40 binds with high affinity to both subtypes of mesothelioma cells and that either the site-specific labeling of M40 with ^99mTc through the hexahistidine tag or conjugation to Cy5.5 does not significantly affect its targeting ability. Second, although the M40 accumulated in the tumors, its clearance from blood and other normal organs was rapid (except from the kidney which is the site for all scFv clearance) making it feasible to label the M40 with positron-emitting residualizing radiometal isotopes with short half-lives such as gallium-68 (⁶⁸Ga) for quantitative PET imaging. Finally, unlike other available antibodies against MM, this antibody has the ability to target both EM and SM. If such dual targeting can be maintained in the clinical setting, this antibody would have a major imaging and therapeutic potential.

The ^99mTc-M40 had a high uptake in the kidneys as observed with other ^99mTc-scFvs (10, 16), consistent with the route of scFv clearance in vivo. Nonetheless, additional engineering of M40 may further increase the kidney clearance rate and improve the contrast. In addition, developing other forms of the antibody such as diabody, minibody, or Affibody (Affibody AB) could further improve the tumor binding and homing efficiency as well as overall pharmacokinetic profile (10, 18).

The ^99mTc-M40 rapidly accumulated into both M28 and VAMT-1 cells as early as 3 hours with comparable tumor uptake at 4.38 ±0.39%ID/g and 5.84±0.72ID%/g, respectively. Because we observed the difference of ¹⁸F-FDG uptake in our system between the 2 cell types, M28 and VAMT-1, with approximately 8%ID/g and 16%ID/g, respectively, similar to other reports (19, 20), ¹⁸F-FDG was used to confirm the identity of the 2 subtypes implanted in a single mouse and thus to highlight the dual targeting by M40. The characteristics of ^99mTc-M40 are representative of a promising new class of radiotracers that recognize cell surface markers highly expressed at comparable level on both types of mesothelioma (Fig. 1A & 1B). As such, they may find utility for tumor characterization and staging as well as treatment planning and monitoring complementary to ¹⁸F-FDG.

In conclusion, we demonstrated that ^99mTc-M40 selectively binds and internalizes into both M28 and VAMT-1 cells in vitro, and furthermore rapidly and specifically targets both subtypes of mesothelioma in vivo, demonstrating its potential as a novel agent for imaging and therapy of all subtypes of mesothelioma.

No potential conflicts of interest were disclosed.

The authors are grateful to Dr. Mary E. Dyszlewski at Tyco/Mallinckrodt for providing the carbonyl-kit (IsoLink). This work is supported partially by NIH R01 CA135358 and the American Cancer Society (IRG-97-150-10) to Jiang He, and R01 CA118919 and R01 CA129491 to Bin Liu.