Preclinical studies of targeted α therapy for breast cancer using 213Bi-labelled-plasminogen activator inhibitor type 2

The control of micrometastatic breast cancer remains problematic. To this end, we are developing a new adjuvant therapy based on 213Bi-PAI2, in which an α-emitting nuclide (213Bi) is chelated to the plasminogen activator inhibitor-2 (PAI2). PAI2 targets the cell-surface receptor bound urokinase plasminogen activator (uPA), which is involved with the metastatic spread of cancer cells. We have successfully labelled and tested recombinant human PAI2 with the α radioisotope 213Bi to produce 213Bi-PAI2, which is highly cytotoxic towards breast cancer cell lines. In this study, the 2-day postinoculation model, using MDA-MB-231 breast cancer cells, was shown to be representative of micrometastatic disease. Our in vivo efficacy experiments show that a single local injection of 213Bi-PAI2 can completely inhibit the growth of tumour at 2 days postcell inoculation, and a single systemic (i.p.) administration at 2 days causes tumour growth inhibition in a dose-dependent manner. The specific role of uPA as the target for 213Bi-PAI2 therapy was determined by PAI2 pretreatment blocking studies. In vivo toxicity studies in nude mice indicate that up to 100 μCi of 213Bi-PAI2 is well tolerated. Thus, 213Bi-PAI2 is successful in targeting isolated breast cancer cells and preangiogenic cell clusters. These results indicate the promising potential of 213Bi-PAI2 as a novel therapeutic agent for micrometastatic breast cancer.

The major failure in the management of early breast cancer is the incomplete killing of malignant cancer cells that have spread throughout the body (Allen, 1999a). This is despite the many treatments available, such as surgery, radiation therapy, hormone therapy and chemotherapy. The American Cancer Society (2001) estimated 182 800 new cases of invasive breast cancer in the year 2000 among women in America, and 40 800 are expected to die from the disease (American Cancer Society, 2000). Novel, more effective treatments that overcome this problem in breast cancer management are essential. Targeted therapy, first discussed over 100 years ago, is based on the idea that a drug will attack its target without damaging other tissue (Raso, 1990). Targeted alpha therapy (TAT) uses an a-emitting radionuclide as a lethal medicament via an effective targeting carrier to kill cancer cells (McDevitt et al, 1998;Allen, 1999b). We are investigating a novel targeting approach that exploits the involvement of cell-surface receptor bound urokinase plasminogen activator (uPA) in the metastatic spread of breast cancer cells (Kruithof et al, 1995).
a-emitting radionuclides emit a particles with energies of 4 -8 MeV, which are up to an order of magnitude greater than most b rays. Yet, their ranges are two orders of magnitude less as a particles have a linear energy transfer (LET) which is about 100 times greater (Allen, 1999a). This is manifested by a higher relative biological effectiveness (RBE). As a result, a much greater fraction of the total energy is deposited in cells with a's and very few nuclear hits are required to kill a cell. Consequently, only a radiation has the potential to kill the metastatic cancer cells at tolerable dose limits, whereas the low LET of b's makes this a very difficult task within human dose tolerance limits.
Availability of the a-emitting radionuclides has been the major problem in the past for their large-scale scientific and clinical application. Studies have been carried out using 149 Tb (Allen, 1999a;Rizvi et al, 2001), 211 At (Bloomer et al, 1984;Larsen et al, 1998) and 212 Bi (Macklis et al, 1993;Horak et al, 1997) with encouraging results. The stable and reliable 225 Ac generator of the a emitting nuclide 213 Bi has been produced, modified and used successfully, with several of these studies indicating a therapeutic potential of 213 Bi-labelled antibody constructs against cancer cells both in vitro and in vivo (Van Geel et al, 1994;Boll et al, 1998;Kennel, 1999a, b;McDevitt et al, 1999;Nikula et al, 1999;Adams et al, 2000;McDevitt et al, 2000). Our group has modified methods of conjugating 213 Bi radionuclide to antibodies with the stable chelator cyclic diethylenetriaminepentacetic acid anhydride (cDTPA) for use in the a therapy of melanoma (Rizvi et al, 2000;Allen et al, 2001a, b), colorectal cancer (Rizvi et al, 2001), leukaemia , breast (Ranson et al, 2002) and prostate cancer (Li et al, 2002a, b).
A large body of experimental and clinical evidence implicates overexpression of the urokinase plasminogen activator (uPA) system as a modulator of the aggressive behaviour of cancer cells and as a strong prognostic factor for predicting poor breast cancer patient outcome (Pollanen et al, 1991;Andreasen et al, 1997;Schmitt et al, 2000). uPA converts plasminogen into the highly active protease plasmin. Plasmin promotes tissue degradation and remodelling of the local extracellular environment by directly and indirectly (via activation of prometalloproteases) degrading extracellular matrix molecules. uPA is synthesised and secreted as a proenzyme, whose activation is markedly accelerated upon binding with high affinity (0.1 -1 nM) to specific cell-surface uPA receptors (uPAR) (Pollanen et al, 1991;Schmitt et al, 1992). Receptor density varies depending on cell type (10 3 -10 6 sites cell À1 ; Schmitt et al, 1992). The ability of PAI2 to inhibit tumour invasion and metastases in animal models has been demonstrated by several laboratories utilising uPA-overexpressing cancer cells (Ranson et al, 2002). PAI1 conjugated to A-chain cholera toxin as the cytotoxic agent or modified PAI1 conjugated to saporin has been used to target fibrosarcoma cells (Jankun, 1992(Jankun, , 1994 with moderate cytotoxicity. However, PAI2 has several distinct advantages over PAI1 for targeted cancer therapy, as discussed in Ranson et al (2002). Cell surface-bound uPA is accessible to and inhibitable by exogenous PAI2 (Jankun, 1992;Yang et al, 2000), and a number of studies have suggested the potential for PAI2 to inhibit cancer cell invasion and metastasis (Kruithof et al, 1995).
The pharmacokinetics and biodistribution of human recombinant 125 I-labelled PAI2 in both control mice and mice bearing human colon cancer (uPA-positive HCT116 cell line) xenografts have been established (Hang et al, 1998). Such studies indicate that invasive and metastatic tumour cells, shown consistently to contain active uPA, would be accessible to and targeted by exogenously administered PAI2.
It is clear that uPA is a specific marker of malignancy and that PAI2 represents a useful targeting agent. We have previously reported the production and evaluation of the new a-nuclide emitting cytotoxic agent 213 Bi-labelled PAI2 (Ranson et al, 2002). The reactivity, specificity and cytotoxicity of a-PAI2 were reported for both MDA-MB-231 and MCF-7 human breast cancer cell lines in vitro. Immunohistochemistry mirrored the differences in expression of endogenous uPA and uPAR antigen seen in these two cell lines by flow cytometry (Ranson et al, 2002).
We have also carried out in vitro and in vivo studies of 213 Bi-PAI2 for prostate cancer (Li et al, 2002b), finding it to be efficacious within the maximum tolerance dose limits. We now demonstrate the efficacy of TAT with 213 Bi-PAI2 in a nude mouse breast cancer model, using MDA-MB-231 breast cancer cells. These data clearly show that 213 Bi-PAI2 has an important role as a potential new therapeutic modality for the control of micrometastases in breast cancer.

MATERIALS AND METHODS
Human recombinant PAI2 (47 kDa) was provided by Biotech Australia Pty Ltd RPMI-1640 was purchased from Life Technologies (Castle Hill, NSW, Australia). Fetal calf serum (FCS) was obtained from Trace Bioscientific (Castle Hill, NSW, Australia). The cyclic anhydride of diethylenetriaminepentacetic acid (cDTPA) was purchased from Aldrich Chemical Company, bovine serum albumin (fraction V) (BSA) from Sigma Chemical (St Louis, MO, USA), and mouse anti-human uPA IgG 1 (#394) monoclonal antibody (MAb) from American Diagnostica Inc. (Greenwich, CT, USA). Mouse isotype control subclasses IgG 1 MAb was from Silenus (Sydney, NSW, Australia) and rabbit anti-mouse IgG and alkaline phosphatase and anti-alkaline phosphatase (APAAP) were purchased from Dakopatts (Glostrap, Denmark). Mouse antihuman melanoma IgG2a MAb (9.2.27) used as a nonspecific control was kindly provided by the Royal Newcastle Hospital (Sydney, Australia).
Radioisotope a-particle-emitting radionuclide 213 Bi was produced from the 225 Ac/ 213 Bi generator, purchased from the United States Department of Energy, Oak Ridge National Laboratory (Oak Ridge, TN, USA). 213 Bi was eluted from the 225 Ac column with 250 mL of freshly prepared 0.15 M distilled and stabilised hydriodic acid followed by washing with 250 mL sterile distilled water (Boll et al, 1997). The first elution was not used, and a time of 2 h was allowed for 213 Bi to regenerate on the column for the next elution. Activity corrections were made for 213 Bi decay using the half-life of 46 min. PAI2 conjugation with cDTPA, stoichiometry and reactivity PAI2 and BSA were conjugated with cDTPA by a modification of published methods to give the desired protein-DTTA conjugate (Ranson et al, 2002) while MAb 9.2.27 was conjugated with cDTPA as described previously (Rizvi et al, 2000) for nonspecific control.
The concentrations of the protein-DTTA chelates were measured by BIORAD DC protein assay reagent kit (Pierce, Rockford Il, USA). The stoichiometry of DTTA-PAI2 was determined using electrospray ionisation mass spectrometry as previously described (Ranson et al, 2002). The DTTA-PAI2 was diluted in 1:20 in water and then 1:2 with 50% MeOH and 1% acetic acid.

213
Bi labelling of DTTA-PAI2 Concentrated DTTA-PAI2 stocks were diluted with 500 mM sodium acetate at pH 5.5 and 5 -10 mg of DTTA-PAI2 was labelled with free 213 Bi for 20 min at room temperature as described previously (Ranson et al, 2002). The radiolabelling efficiency was about 90%, as determined by instant thin layer chromatography (ITLC) using the described method (Rizvi et al, 2000;Ranson et al, 2002).

213
Bi-PAI2 was incubated in fresh human serum and in DTPA (challenge test) at 371C over 24 h. Activity was counted by detecting the 440 keV gamma ray from the decay of 213 Bi. The percentage leaching/stability was calculated as a function of time, and the results were corrected for decay of the radioisotope.

Cell culture
The metastatic MDA-MB-231 human breast cancer cell line was originally purchased from American Type Culture Collection (Rockville, MD, USA) and routinely cultured in RPMI-1640 supplemented with 10% (v v) heat-inactivated FCS and passaged using trypsin/EDTA. The cells were incubated in a humidified incubator at 371C with a 5% carbon dioxide air atmosphere. For all experimental procedures, subconfluent cells that had been in culture for 48 h without a change of media were harvested by rinsing flasks twice with PBS (pH 7.2) and then detaching with PBS/0.5 mM EDTA at 371C for 5 min. Cells were collected and resuspended in the appropriate buffer as described below.

Animals and MDA-MB-231 cell inoculation
In all, 6 to 8-week-old athymic nude mice, BALB/c (nu/nu) female mice were purchased from Animal Resources Centre (ACR), Western Australia. The mice were housed and maintained in laminar flow cabinets under specific pathogen-free conditions in facilities approved by the University of New South Wales (UNSW) Animal Care and Ethics Committee (ACEC) and in accordance with their regulations and standards. The ethical guidelines that were followed meet the standards required by the UK Coordinating Committee on Cancer Research Guidelines (Workman et al, 1998).
To establish s.c. animal tumour models, 1 million MDA-MB-231 cells were resuspended in 200 mL of RPMI-164 serum-free medium and injected via an 18-gauge needle into bilateral mammary fat pads of each nude mouse. Tumour progression was documented once weekly by measurements using calipers, and tumour volumes were calculated by the following formula: length Â width Â height Â 0.52 in millimetres (Gleave et al, 1991). All mice were killed when the xenografts approached 1 Â 1 cm 2 in area by CO 2 chamber.
Control injections used PBS and the nonspecific a-conjugates 213 Bi-BSA and 213 Bi-9.2.27 (monoclonal antibody for melanoma).

Experimental protocols
In vivo toxicity study: Groups of five mice received a total injected activity (bound plus unbound 213 Bi) of 1.5, 3 and 6 mCi/kg weight dose of 213 Bi-PAI2 by i.p. injection. Additional mice were treated with PAI2, cDTPA and PBS as controls. Mice were monitored and weight was measured.
Biodistribution study: Mice received an i.p injection of 213 Bi-PAI2, and were euthanised at 15, 30, 45, 60, 90 and 120 min. Tissues and organs were removed, weighed and the activity was counted. The bone marrow count was obtained by measuring activity in the hip. Results are quoted as percent activity at a given euthanasia time.
Two-day MDA-MB-231 model: One million cells were injected s.c. into bilateral mammary glands of five mice, and at 2 days the mice were killed and the local tissue section removed for histochemistry. The objective of this study was to demonstrate the state of pretumour development.
Local TAT efficacy for dose response: Efficacy studies of local TAT were made for dose response and for postinoculation time response at 2 days postinoculation. The 213 Bi-PAI2 was injected in the same region as the inoculation, as no tumour was evident.
Local TAT efficacy for time response: Four different therapy time points were used, each with five mice: 2 -4, 7, 14 and 28 days after cell inoculation. Each group had one control mouse and four treated mice.
Systemic TAT with 213 Bi-PAI2: The dose response for systemic administration (i.p.) injection at 2 days postinoculation was also studied. Previous studies had shown little difference between i.p. and tail vein injections, so the more difficult tail vein approach was not justified.
PAI2 blocking study: At 2 days postbilateral inoculation each of 10 6 cancer cells, i.p. administration at two, three and four times the standard PAI2 conjugate concentration (100 mg mL À1 ) was followed by 100 mCi of 213 Bi-PAI2 in groups of five mice. Tumour growth was monitored in all mice and mice were killed at the tumour volume limit.

Statistics
All numerical data were expressed as the average of the values obtained, and the standard error of means (s.e.m.) was calculated and shown in the figures. Unpaired t-test was used to determine significant differences at 0.05 probability between tumour growth expressed as volume for controls and 213 Bi-PAI2-treated mice.

Immunohistochemistry
The alkaline phosphatase antialkaline phosphate (APAAP) method was used to detect uPA expression in MDA-MB-231 cells after 2 days inoculation in nude mice (Li et al, 2002b). Control slides were treated in an identical manner. PC3 metastatic prostate cancer cells were chosen as a positive control, while isotype MAb or the primary antibody were omitted as a negative control. The positive cells appear pink.

Chelation of PAI2 with cDTPA
Mass spectroscopy results are shown in PAI2 alone ( Figure 1A) and DTTA-PAI2 ( Figure 1B). Up to five-fold attachment of cDTPA is observed, the peaks being separated by the MW of the chelator (357 Da).

Serum stability test
Most of the instability of 213 Bi-PAI2 in serum and in DTPA (challenge test) occurs within one half-life of 213 Bi (data not shown). Leaching of activity is in the range from 20 to 30%. Curiously, the conjugate is more stable in the DTPA challenge.

In vivo tolerance study
The weights of injected mice reduced initially by 5 -10%, then recovered after 1 week. After 13 weeks, one saline control mouse died, but other mice were healthy until euthanasia at 24 weeks post-therapy (data not shown). No dose effect was observed.

Biodistribution
Results were obtained over 2.6 half-lives ( Figure 2) and showed that the kidneys received the highest activity, being more than half the observed activity after 25 min. The bone marrow receives the next highest dose, but other organs have relatively low exposure.

Expression of uPA in vivo model after 2 days inoculation
A group of five mice were sacrificed at 2 days after cell inoculation. The results from immunostaining with the #394 MAb against uPA show that isolated cells and cell clusters are prevalent, all cancer cells are positive to uPA, and there is no evidence for microcapillary formation ( Figure 3A) while the cancer cells with no primary MAb are negative to uPA MAb ( Figure 3B). Thus the 2day model accurately simulates micrometastasis and preangiogenic lesions.

Tumour growth inhibition by local injection of 213 Bi-PAI2
In an earlier study (Allen et al, 2001a), local injection at 2 days postinoculation of 12, 25 and 50 mCi of 213 Bi-PAI2 showed a dosedependent response in groups of five mice. Tumours grew quickly after injection with a PBS control, while the 12 mCi group grew very slowly. The 25 and 50 mCi groups showed complete inhibition of tumour growth.
This study was repeated using bilateral injections of 50 mCi of 213 Bi-PAI2 and 213 Bi-BSA (which has comparable mass to 213 Bi-PAI2) and PBS as controls. Results are shown in Figure 4A. Both PBS and 213 Bi-BSA controls are significantly different to the 213 Bi-PAI2 results (P ¼ 0.04 and 0.03, respectively).
A single injection of 213 Bi-PAI2 (25 mCi) was made into cell inoculation sites or tumours at different postinoculation times. Mice were monitored and tumours were measured. The 2 -4-day group had the best response to the therapy, which had some 50% (23 out of 40) tumour control and slower tumour growth rate compared with control mice. The 7-day and 14-day groups had two out of eight and one out of eight tumour control and slower growth rate compared with control mice. The 28-day group had zero out of eight tumour control and no obvious change in tumour growth rate.

Efficacy of 213 Bi-PAI2 by systemic injection
Mice received single systemic (i.p.) injections of 25, 50 and 100 mCi of 213 Bi-PAI2 at 2 days postinoculation. The results indicate a substantial inhibition of tumour growth up to 35 days postinoculation. The control mice (n ¼ 5) received a single i.p. injection of 100 mCi nonspecific 213 Bi-9.2.27 conjugate, which had no effect on tumour growth ( Figure 5A), while in the treated groups (n ¼ 5), a dose effect is indicated in that increased activity decreased the number of observed tumours, that is, three out of five in 25 mCi ( Figure 5B), two out of five in 50 mCi ( Figure 5C) and one out of five in 100 mCi ( Figure 5D). The control data are significantly different from the 213 Bi-PAI2 data for all dose levels.
A second study was made with both PBS and 213 Bi-BSA controls, and the results are shown in Figure 4B. The controls are not significantly different from each other, but the PBS and 213 Bi-BSA controls are significantly different from the 213 Bi-PAI2 result (P ¼ 0.01 and 0.04, respectively). Note that only a single i.p. injection of 100 mCi was made. Systemic (i.p.) injection of PAI2 at two, three and four times 213 Bi-PAI2 was administered, followed by the 213 Bi-PAI2 conjugate. Results are shown in Figure 4C, where tumours grew in all 'PAI2 blocked' mice, whereas eight out of 10 'unblocked' mice showed complete tumour growth inhibition to 60 days. Pretreated mice showed tumour volumes that were not significantly different for each concentration. However, they were all significantly different to the 'unblocked' mice (Po0.001). The 'blocked' mice were euthanised at 30 -45 days according to protocol. All excess concentrations of PAI2 were adequate to effectively block

DISCUSSION
In this study, we describe the novel compound 213 Bi-PAI2 and show that it retains reactivity, selectivity and cytotoxicity towards uPA expressing breast cancer cells in vivo. That 213 Bi-PAI2 cytotoxicity is significantly mediated via a uPA-dependent mechanism had been confirmed by the lack of cytotoxicity of freshly isolated, normal human leukocytes on which cell-surface localised active uPA was not detectable (Ranson et al, 2002). 213 Bi-PAI2 was found to be selectively toxic to targeted breast cancer cells, whereas nontargeted cells are spared from the radiotoxicity arising from the a radiation. Breast cancer cells incubated with a nonspecific a-conjugate, viz 213 Bi-BSA, were also minimally affected.
Mass spectroscopy analyses show that multiple binding of the cDTPA chelator occurs. While this will enhance the activity carried by PAI2, it may cause problems for quality control of specific activity. The stability of the a-conjugate is also a cause for concern. Some 20% of activity is lost in serum at 371C within 1 half-life. More stable conjugation is desired.
Minimal toxicity of 213 Bi-PAI2 was observed in mice up to a total activity of 6 mCi kg À1 . This dose level is expected to be more than adequate for human use, and compares with 1 mCi kg À1 reported for 213 Bi-MAb in a phase 1 clinical trial by the Memorial group (Jurcic et al, 2002).
The biodistribution of 213 Bi-PAI2 in mice showed that the kidneys received the highest dose (450% after 25 min). Further, the kidneys continued to accumulate activity up to 100 min. This result may have implications for the induction of secondary renal cancer at higher doses.
Resection  of cancer cells at the inoculation site, but no evidence for capillary formation was found. This model is therefore expected to simulate micrometastatic cancer. Our in vivo studies revealed that 213 Bi-PAI2 can target isolated cells and preangiogenic cell clusters. Local therapy required only 25 mCi of a-PAI2 to achieve complete inhibition of tumorigenesis. 213 Bi-PAI2 was found to be increasingly less effective in a single dose intralesional protocol as the tumours grew in size.
A higher dose of 100 mCi was required for a single systemic administration to achieve B80% inhibition, and indications of a dose-dependent response were seen. All control results were found to be significantly different from the 213 Bi-PAI2 data. The specific, in vivo cytotoxicity of 213 Bi-PAI2 against the uPA receptor was directly tested by preinjection of increasing concentrations of PAI2, before 213 Bi-PAI2 administration. Tumours grew in all 'blocked' mice, which were euthanised according to protocol at 30 -40 days, whereas only two out of 10 tumours grew in the unblocked 213 Bi-PAI2 mice.
Our results clearly indicate that PAI2 can target membranebound uPA receptors, deliver a particles to MDA-MB-231 metastatic breast cancer cells and regress tumour growth through local or systemic injections. The exact mechanism of cell killing has not been investigated. Macklis et al (1992) demonstrated that a-particle radioimmunotherapy can induce apoptosis in murine lymphoma cells. Using the 213 Bi-PAI2 conjugate, we have also demonstrated that 213 Bi-PAI2 can induce a high percentage of TUNEL-positive PC3 prostate cancer cells in vitro and in vivo (Li et al, 2002b). These data suggest that apoptosis may be the lethal pathway of 213 Bi-PAI2 therapy for MDA-MB-231 breast cancer cells.

CONCLUSIONS
We have combined the cytotoxicity of an a-emitting radioisotope ( 213 Bi) with the targeting potential of PAI2 to create the novel construct 213 Bi-PAI2, a potential new therapeutic agent for targeted a therapy of breast and prostate cancer (Li et al, 2002b). The in vitro cytotoxicity of 213 Bi-PAI2 on breast cancer cells was shown to be specific, indicating that the cell killing ability of 213 Bi-PAI2 depends critically on the targeting of cells in a receptor-bound, active uPA-dependent manner (Ranson et al, 2002). These in vivo results show conclusively that 213 Bi-PAI2 can target and kill isolated cells and cell clusters, via local and systemic injection. Further, as tumours in mice pretreated with PAI2 could not be regressed, targeting of uPA/uPAR is implicated. 213 Bi-PAI2 therapy is therefore indicated for the control of micrometastic breast cancer. The content of this paper does not reflect the position or policy of the US Government, and no official endorsement should be inferred. The authors thank Dr Clive Bunn (Human Therapeutics Ltd) for critical reading of the manuscript and Biotech Australia and PI2 Pty Ltd for the supply of PAI2. We also thank Professor Peter Hersey, Royal Newcastle Hospital for providing the