Two diverse carriers are better than one: A case study in α‐particle therapy for prostate specific membrane antigen‐expressing prostate cancers

Abstract Partial and/or heterogeneous irradiation of established (i.e., large, vascularized) tumors by α‐particles that exhibit only a 4–5 cell‐diameter range in tissue, limits the therapeutic effect, since regions not being hit by the high energy α‐particles are likely not to be killed. This study aims to mechanistically understand a delivery strategy to uniformly distribute α‐particles within established solid tumors by simultaneously delivering the same α‐particle emitter by two diverse carriers, each killing a different region of the tumor: (1) the cancer‐agnostic, but also tumor‐responsive, liposomes engineered to best irradiate tumor regions far from the vasculature, and (2) a separately administered, antibody, targeting any cancer‐cell's surface marker, to best irradiate the tumor perivascular regions. We demonstrate that on a prostate specific membrane antigen (PSMA)‐expressing prostate cancer xenograft mouse model, for the same total injected radioactivity of the α‐particle emitter Actinium‐225, any radioactivity split ratio between the two carriers resulted in better tumor growth inhibition compared to the tumor inhibition when the total radioactivity was delivered by any of the two carriers alone. This finding was due to more uniform tumor irradiation for the same total injected radioactivity. The killing efficacy was improved even though the tumor‐absorbed dose delivered by the combined carriers was lower than the tumor‐absorbed dose delivered by the antibody alone. Studies on spheroids with different receptor‐expression, used as surrogates of the tumors' avascular regions, demonstrated that our delivery strategy is valid even for as low as 1+ (ImmunoHistoChemistry score) PSMA‐levels. The findings presented herein may hold clinical promise for those established tumors not being effectively eradicated by current α‐particle radiotherapies.


| INTRODUCTION
Metastatic castration-resistant prostate cancer (CRPC) is lethal and incurable, leading to the death of 31,000 men in the United States annually. Expression of the prostate specific membrane antigen (PSMA) is conserved in several of these advanced tumors. [1][2][3] Currently, there is tremendous clinical promise for PSMA-targeted alpha-particle (α-particle) therapies against soft-tissue metastases of prostate cancer 4,5 for patients who were resistant to or ineligible for other therapies. However, the success of these α-particle approaches against soft-tissue metastases is expected to be most successful against relatively early disease. 4,6,7 Alpha-particles are high energy, short-range particles (traveling in tissue up to 4-5 cell diameters). They physically break DNA molecules (causing double-strand breaks-the most difficult to repair type of DNA damage) as they traverse the cell nucleus. The complexity and level of rapidly induced DNA damage overwhelms cellular repair mechanisms. 8,9 The inability to repair this type of DNA damage is the reason that α-particles are impervious to most cancer cell resistance mechanisms, 10,11 if optimally delivered. However, the short-range of α-particles in tissue that enables localized irradiation with minimal toxicities to the surrounding healthy sites, also limits uniform irradiation of large tumors; this is because it is coupled with the diffusion-limited penetration depths of traditional radionuclide carriers resulting, therefore, in partial tumor irradiation, potentially limiting efficacy.
We have recently discovered that we can improve the uniformity of distributions of α-particles within established tumors when the same total radioactivity is equally split between two separate and diverse types of carriers, each preferentially killing a different region of the tumor 12 : (1) the tumor-responsive liposomes that upon tumor uptake release in the interstitium a highly diffusing form of their radioactive payload ( 225 Ac-DOTA), which then penetrates the deeper parts of tumors where antibodies do not reach, and (2) a separately administered, less-penetrating radiolabeled-antibody, irradiating the tumor perivascular regions from where the liposomes' contents clear too fast. Our tumor-responsive liposomes are composed of membranes forming phase-separated lipid domains (resembling lipid patches) with lowering pH. 13 During circulation in the blood, such liposomes comprise well-mixed, uniform membranes that stably retain their encapsulated contents. In the acidic tumor interstitium (pH e $6.7-6.5 14 ) lipidphase separation results in formation of lipid patches that span the bilayer, creating transient lipid-packing defects along the patch boundaries, and enabling release of encapsulated agents. The liposomes also have an adhesive property that enables them to bind to the tumors' extracellular matrix, delaying their clearance from tumors. 15,16 On a proof-of-concept human epidermal growth factor receptor 2-positive breast cancer mouse model, the equal split of injected radioactivity between the two separate carriers improved the tumor growth inhibition compared to the tumor inhibition observed at the same total injected radioactivity when treated by any of the carriers alone.
In the present study, (1) we investigate how the different expression levels of the targeted receptor PSMA on prostate cancers may affect the microdistributions of the PSMA-targeting antibody and, therefore, the best radioactivity split ratio(s) for inhibiting cancer cell growth in the presence of transport barriers, and (2) we systematically interrogate the effect of different radioactivity split ratios between the two carriers (tumor-responsive liposomes and the PSMA-targeting antibody) on inhibiting the growth of PSMA-expressing prostate cancers in vivo.
In particular, this study was motivated by the finding that PSMA expression in prostate cancer has been reported to exhibit variability and heterogeneity. 17 On multicellular spheroids of variable size that were utilized as surrogates of the avascular regions within solid T A B L E 1 Characterization of the tumor-responsive liposomes loaded with 225 Ac-DOTA and/or with 111 In-DTPA Note: *p < 0.05, **p < 0.01. Errors correspond to the standard deviation of n independent measurements (as indicated). tumors, formed by prostate cancer cells with different expression levels of PSMA and different intrinsic sensitivities to Actinium-225 ( 225 Ac), we explored the killing efficacy of α-particle radiotherapy delivered by several different radioactivity split ratios between the two carriers. On mice bearing PSMA-positive prostate cancer xenografts, the biodistributions/dosimetry, tumor growth inhibition, and toxicities were evaluated for the same total injected radioactivity that was delivered by different split ratios between the two carriers. Our findings demonstrate that our delivery strategy is, in principle, applicable to cells with different levels of expression of the targeted surface marker, and it is, in a sense, tumor agnostic, since only the targeting antibody needs to be specific to the cell surface marker.

| Cell line characterization and survival assay
The expression levels of PSMA on C4-2B, LNCaP, and PC3-PIP prostate cancer cells were measured, using the PSMA-targeting 111 In-DTPA-SCNantibody, to be 126,000 ± 12,000, 210,000 ± 12,000, and 3,400,000 ± 78,000 copies per cell, respectively; the PSMA-targeting antibody exhibited comparable K D across all cell lines (Table S1 and Figure S2). These cell lines were chosen as representatives for low, moderate, and high levels of PSMA expression. The PC3 cell line did not express any measurable levels of PSMA, in agreement with previous reports. 20 Each cell line formed 3D spheroids, which developed acidity in the interstitium ( Figure S3), serving as an in vitro model of the avascular regions of solid tumors.
On cell monolayers, in the absence of transport barriers, the killing efficacy of free 225 Ac-1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) was independent of the extracellular pH for all cell lines (Figure 1a,b). The killing effect of free 225 Ac-DOTA and of 225 Ac-DOTA encapsulated in tumor-responsive liposomes was the same for PC3-PIP cells (Figure 1c,d), and for the other cell lines ( Figure S4). This finding was expected since neither free 225 Ac-DOTA nor the liposomes preferentially associate with cells. Finally, the PSMA-targeting 225 Ac-DOTA-SCN-antibody exhibited increasing killing efficacies with increasing expression of PSMA by cells (Figure 1a,b, compared to Figure S4). In addition, Figure 4 shows that on the same (intermediate) size

| Spheroid growth control
spheroids (r = 200 μm) with increasing PSMA-expression levels (leftto-right), the best inhibition of spheroid growth was exhibited when increasing fractions of the radioactivity were delivered by the liposomes. This is because the radiolabeled antibody delivers its F I G U R E 3 Best outgrowth inhibition of spheroids with increasing size, at same total radioactivity, was exhibited by radioactivity split ratios that favored greater liposome fractions. Extent of outgrowth inhibition of spheroids formed of PC3-PIP cells (with high prostate specific membrane antigen [PSMA] expression, a-c) and of LNCaP cells (moderate PSMA expression, d-f) with varying sizes after treatment with different radioactivity split ratios between the tumor-responsive liposomes loaded with 225 Ac-DOTA (liposomes, incubated for 6 h) and the PSMA-targeting 225 Ac-DOTA-SCN-antibody (antibodies, incubated for 24 h, to approximately match the relative blood clearance kinetics of the carriers). The total radioactivity concentration was kept constant per spheroid size (a) 4.5 kBq/ml, (b) 9 kBq/ml, (c) 20 kBq/ml, (d) 4.5 kBq/ml, (e) 9 kBq/ml, (f) 13 kBq/ml. Error bars correspond to standard deviations of repeated measurements (n = 6 spheroids per condition, n = 3 independent liposome and antibody preparations). * indicates p < 0.05, **p < 0.01 therapeutic cargo mostly in the periphery of the spheroid (the socalled binding site barrier effect) 12,21 : increasing PSMA-expression levels by cells required a lower fraction of the antibody-delivered radioactivity, since the same levels of delivered radioactivity per cell could be reached at lower antibody concentrations given the increased numbers of receptors per cell. In Figure 4a, on spheroids that do not express detectable PSMA levels, both the liposomes and the antibodies behaved as nontargeting particles with the latter being significantly smaller. Note that each of the carriers was incubated with spheroids for different times: 24 h with the antibodies and 6 h with the liposomes, to approximately match their blood circulation halflives in mice. The (smaller) antibodies had, therefore, longer available time to penetrate deeper in these PSMA-negative spheroids. The cold liposomes and the cold antibody did not affect the growth of any of the spheroids compared to the nontreated condition ( Figure S7).

| In vivo assessment
As previously reported, major off-target sites for both carriers were the liver and spleen, as well as the kidneys for the PSMA-targeting antibody (Table S2, Figure S8). Tumor uptake of the PSMA-targeting 111 In-DTPA-SCN-antibody was significantly greater than that of 111 In-DTPA-loaded liposomes ( Figure 5c); this was partly attributed to the different blood clearance half-lives of each carrier ( Figure 5d). Table 3 shows the significant difference between the tumor-absorbed doses when the same total radioactivity was injected in mice and was delivered by each carrier alone and/or by both carriers at equal ratios of injected radioactivity.
The extracellular pH e maps of PSMA-expressing PC3-PIP subcutaneous tumors confirmed the development of acidity in the interstitium, with regions reaching as low pH e values as 6.5 ( Figure 6a); these pH e values were adequate to activate both the release and adhesion properties on the tumor-responsive liposomes (Table 1). 12,16 The α-camera images of tumor sections on the top panel in Figure 6b show the pixel intensities of tumor sections that were normalized relative to the average intensity of the entire tumor section.
At same total administered radioactivity, the images demonstrate more uniform microdistributions (red-colored areas) of the delivered radioactivity when both carriers were utilized (Figure 6b

| DISCUSSION
We have hypothesized and demonstrated that in order to improve the killing efficacy against established, soft-tissue solid tumors, more uniform microdistributions of α-particle emitters within the tumors are required, as opposed to just more α-particle radioactivity without considering their microdistributions. This was shown to be due to the larger fraction of tumor cells becoming exposed to the cytotoxic agents, as opposed to only a smaller fraction of the tumor cells receiving greater doses. To enable this, we administered the same total α-particle radiotherapy using combinations of separate carriers (tumor-responsive liposomes and targeting antibodies) that resulted in complementary microdistributions of the α-particle emitter 225 Ac within established solid tumors.
In particular, we studied PSMA-positive PC3-PIP prostate cancer tumors in vivo, which proved to be particularly heterogeneous in their morphology (as shown in Figure S11), and we demonstrated that the same total α-particle radioactivity delivered by any combinations of improved overall efficacy (i.e., tumor growth inhibition). While the best response at the same total injected radioactivity was seen when the radioactivity was split equally between the two carriers, all ratios of radioactivity split between the carriers resulted in better tumor growth inhibition than the inhibition achieved by the same total dose delivered by either carrier alone. These findings suggest that any ratio of carriers, chosen to split the same total injected radioactivity, results in more uniform α-particle tumor microdistributions compared to the microdistributions enabled by any single carrier alone. Notably, the tumor-absorbed dose delivered by any carrier combination was lower than the dose delivered by the antibody alone (as shown in Table 3).
The best tumor growth inhibition was achieved when the same total radioactivity was equally split between the two carriers that resulted in 32% less tumor-absorbed dose (0.23 vs. 0.34 Gy) compared to the tumor-absorbed dose delivered by the antibody alone.
The improved killing efficacy of the same radioactivity when delivered by combinations of the two carriers was also observed in vitro on multicellular spheroids-used as surrogates of tumors' avascular regions-that were formed by prostate cancer cells with varying levels of PSMA expression. These studies suggest that our delivery strategy for α-particle emitters can be effective even when the targeted marker expression is as low as 130,000 copies per cell (corresponding to roughly 1+ by ImmunoHistoChemistry score 23 ).
From an engineering perspective, tumors, in vivo, may be consid- The efficacy studies were performed at combined administered radioactivities (4.63 kBq/20 g mouse) that did not cause measurable toxicities in tumor-free mice when the same total radioactivity was delivered only by the antibody or by the liposomes alone. 16 Potential dose-limiting organs, in mice, were the liver and spleen that exhibited significant absorbed doses; kidney irradiation by the liposomes was previously shown not to be a concern. 16 Dividing the same total radioactivity into both carriers did not significantly change the already substantial delivered dose to the liver and/or spleen, and only slightly decreased the dose to the kidneys (Table 3)

| Liposome formation and loading
Liposomes were formed using the thin-film hydration method as previously described. 32 Briefly, lipids were combined at a mole ratio of Radiolabeling yield was calculated as the ratio of the measured radioactivity before and after the final 10DG column.
Radiopurity was measured using instant thin layer chromatography as previously reported using 10 mM EDTA as the mobile phase. 33 Purity was calculated as the percent of radioactivity remaining at the bottom quarter of the strip relative to top half. Immunoreactivity was measured by incubating cells (1 million cells/ml) with radiolabeled antibody at a 100:1 receptor:antibody ratio on ice for 1 h until equilibrium was reached, after which it was washed three times with ice cold PBS.
A second tube, containing a 50 times excess of unlabeled antibody was run in parallel to account for nonspecific binding. Immunoreactivity was calculated as the ratio of bound radiolabeled antibody to the initial amount of radioactivity added, corrected for the nonspecific binding.

| Cell culture and spheroid formation
The LNCaP cell line was purchased from ATCC. The C4-2B and PC3 cell lines were obtained from the Pienta lab (Dr. Sarah Amend).

| Evaluation of microdistributions of carriers in spheroids
Spheroids were incubated with CFDA-SE-encapsulating liposomes

| Colony formation assay
In six-well plates, 500,000 cells/well were plated for 24 h and then the media was replaced with fresh media at the desired pH, and the radioactive constructs were added at varying radioactivity concentrations (ranging from 74 to 1 kBq/ml). Following 6 h of incubation, each well was washed thrice with PBS to ensure all radiation was removed.
Cells were then scraped and suspended in fresh media, and were then plated at varying cell densities (ranging from 100 cells/5 ml [20 cells/ml] to 50,000 cells/20 ml [2500 cells/ml]) in Petri dishes. These dishes were placed in an incubator and allowed to grow for approximately 10 doubling times until measurable colonies were formed from surviving cells. Following this, each dish was stained with crystal violet, and the number of colonies formed was counted.
Accounting for the initial cell plating density, the survival fraction was calculated at each radioactivity concentration by normalizing by the no-treatment condition.

| Biodistributions of the carriers
Tumor-bearing mice were intravenously injected with 100 μl of 370 Bq of 111 In-DTPA encapsulated in tumor-responsive liposomes or of 111 In-DTPA-SCN-labeled PSMA-targeting antibody (diluted to a total mass of 7.5 μg antibody, corresponding to the 50:50 condition). At different time points, mice were euthanized and each organ was weighed and measured on a gamma counter for radioactivity. The measured activities were decay corrected to the time of injection, and the percent of initial injected activity per organ weight (%IA/g) was calculated.

| Dosimetry
Dosimetry was performed using the methodology described in ref 35 using the biodistributions of 111 In-labeled tumor-responsive liposomes and of the 111 In-labeled PSMA-targeting antibody. 111 In has been confirmed as a surrogate of 225 Ac biodistributions delivered by internalizing antibodies and by our liposomes. 16,36 Briefly, following decay correction and multiplication by 225 Ac half-life decay factor, the data were integrated to generate the time-integrated activity coefficients (TIACs) (using the software package 3D-RD-S, Radiopharmaceutical Imaging and Dosimetry, LLC [Rapid], Baltimore MD). The resulting TIACs were then multiplied by the energy associated with α-particle and electron emissions of 225 Ac and its daughters (ICRU decay scheme). For liposomes, 25% of the 213 Bi generated in the tumor from 225 Ac decays was assumed to decay in the kidneys as previously reported. 16 For the PSMA-targeting antibody, all 213 Bi generated in the tumor were assumed to be retained in the tumor.
Mice were euthanized 24 h after injection, and tumors (75 mm 3 ) were removed and allowed to reach secular equilibrium for 6 h. Tumors were sliced into 16 μm sections, which were placed on scintillation paper for α-camera imaging as previously reported. 16

| Toxicity study
Tumor-free 6-8-week-old male NSG (NOD scid gamma) mice were injected with 4.63 kBq and 2.31 kBq of 225 Ac-DOTA-SCN-labeled antibody. One month post-injection, all mice were euthanized and organs were H&E stained for histopathology evaluation. The MTD of 225 Ac delivered by tumor-responsive liposomes on the same animal strain was previously found to be 5.55 kBq per 20 g animal. 16

| Statistical analysis
All results are reported as the mean ± standard deviation between n independent measurements (specifics for each experiment stated in each figure captions). Significance between treatment conditions was evaluated using one-way analysis of variance and the unpaired Student's t-test, with significance defined as p < 0.05.