Comparative Study of Subcutaneous and Orthotopic Mouse Models of Prostate Cancer: Vascular Perfusion, Vasculature Density, Hypoxic Burden and BB2r-Targeting Efficacy

The gastrin-releasing peptide receptor (BB2r) is overexpressed in a variety of cancers including prostate cancer. As a consequence, the development of BB2r-targeted diagnostic/therapeutic radiopharmaceuticals has been widely explored. Both subcutaneous and orthotopic mouse models have been extensively used in BB2r-targeted agent development, but side-by-side studies examining how biological parameters (tumor perfusion efficacy, hypoxic burden and microvasculature density) impact BB2r-targeted agent delivery has not been reported. Herein, we examine these biological parameters using subcutaneous and orthotopic PC-3 xenografts. Using a dual isotope biodistribution study, tumor perfusion was accessed using [99mTc]NaTcO4 and BB2r-targeted uptake evaluated by utilization of a novel 177Lu-labeled conjugate ([177Lu]Lu-DOTA-SP714). Immunofluorescence, immunohistochemistry and autoradiography were utilized to examine the tumor vascular density, hypoxic burden and microdistribution of the BB2r-targeted agent. Our studies demonstrated that compared to the subcutaneous model the PC-3 orthotopic tumors had significantly higher levels of perfusion that led to higher BB2r-targeted uptake and lower levels of hypoxia burden. It is anticipated that our results will allow researchers to better understand the biological variables affecting drug delivery and assist them in more clearly interpreting their results in this common prostate cancer mouse model.

BB2r-mediated endocytosis. Subsequently, competitive binding studies demonstrated that both peptides demonstrated low nanomolar binding affinity, with nat Lu-DOTA-SP714 (1.0 ± 1.7 nM) giving a significantly higher affinity relative to the unlabeled conjugate (9.8 ± 1.7 nM). Both the IC 50 and the LogD 7.4 values are provided in Table 1.

Radiochemical stability studies.
To investigate the susceptibility of [ 177 Lu]Lu-DOTA-SP714 to radiolytic degradation and examine the efficacy of stabilizers, two stabilizing buffers formulated using ascorbic acid with/ without selenomethionine were tested and compared to a control PBS buffer. The results of these studies are depicted in Figure S2. In PBS alone, 39.9% of a 0.37 MBq/mL (100 µCi/mL) solution of [ 177 Lu]Lu-DOTA-SP714 degraded over the course of 72 h. As expected, the addition of ascorbic acid (40 mg/mL) demonstrated substantial improvements in stability with only 17.3% radiolytic breakdown. However, the combination of ascorbic acid (40 mg/mL) and selenomethionine (0.2 mg/mL) gave superior results with only 5.3% breakdown observed by the 72 h time point.
Animal model establishment. In this study, subcutaneous and orthotopic xenograft models were generated. For the subcutaneous model, the diameters of the tumors were determined by caliper measurements, while tumor growth in the orthotopic model was monitored weekly after implantation surgery using bioluminescence imaging (Fig. S3a,b). The orthotopic model gave a gradual increase in estimated tumor volumes over the 6-weeks, while the subcutaneous model showed a sharp and rapid increase in tumor size starting at 4 th -week (Fig. S3c). While the above measurements were used to non-invasively monitor tumor growth, the assignment of the tumor volume in our subsequent studies is based on caliper measurements of the excised tumors. For both measurements, tumor volumes were calculated by using the following formula: Volume = (Length × Width 2 )/2. ) (r = 0.47, p = 0.021 and r = 0.49, p = 0.017) in subcutaneous and orthotopic mice models, respectively. Interestingly, the correlation ( Figure S4) between tumor perfusion ([ 99m Tc]NaTcO 4 ) and BB2r-uptake ([ 177 Lu]Lu-DOTA-SP714) was weak to moderate for the subcutaneous (r = 0.28, p = 0.19) and orthotopic (r = 0.66, p = 0.0006) models, respectively, suggesting tumor size is a better indicator of BB2r-uptake in both models.
The most striking differences were observed when comparing the BB2r-mediated uptake and perfusion between models at each tumor volume category. The average uptake of the [ 177 Lu]Lu-DOTA-SP714 for the orthotopic model was 1.9, 1.6 and 1.2-fold higher compared to the respective tumor volumes of the subcutaneous model (<300 mm 3 , p = 0.0055; 300-700 mm 3 , p = 0.011; and >700 mm 3 groups, p = 0.34). Similarly, a substantial and statistically significant increase in tumor perfusion was observed for the orthotopic versus the subcutaneous model for each tumor volume category. The average uptake of perfusion tracer in the orthotopic model was 4.1, 4.3 and 3.7-fold higher compared to the respective subcutaneous model (<300 mm 3 , p = 0.0008; 300-700 mm 3 , p = 0.0001; and >700 mm 3 , p = 0.0001). Based on the examination of the biodistribution data, this large difference in perfusion between models may well be overstated due to a decreased clearance rate of the radiotracer from the orthotopic model. Biodistribution data (Table S1) show that the blood retention for the orthotopic model was 1.5-3.1 times higher compared to the subcutaneous model. Furthermore, this trend was reflected in the well-known stomach uptake of 99m Tc-pertechnetate which was approximately 8-fold higher for the orthotopic model 25 . These data suggest the orthotopic model has a lower rate of blood clearance for 99m Tc-pertechnetate. Speculatively, this may be due to the placement of the orthotopic tumor restricting renal excretion 26 and reducing 99m Tc-pertechnetate clearance from the body. If so, this effect was asymmetric since the blood retention values for the BB2r-targeted agent ([ 177 Lu]Lu-DOTA-SP714) between the two models were not statistically different.

Quantification of hypoxic burden and blood vessel density in PC-3 tumors.
Prior to the sacrifice of the mice, pimonidazole (hypoxia marker) and Hoechst (functional vasculature marker), were administered to examine the hypoxia burden and vascular density of the human PC-3 tumor xenografts. The excised human PC-3 tumors were sectioned and histologically evaluated for fluorescence signals using adjacent tumor slices. Both  www.nature.com/scientificreports www.nature.com/scientificreports/ signals were calculated by mean intensity per µm 2 and grouped by tumor volume groups for each mouse model (Table 3).
On average, subcutaneous tumors exhibited a 3.4-fold higher level of hypoxia burden (signal) than the orthotopic tumors (Fig. 3a). For both xenograft models, the tumors demonstrated a trend toward maximal hypoxia burden for the 300-700 mm 3 group (Fig. 3c). Overall, there was no significant difference (p = 0.13) in the hypoxia burden among the tumor volume groups in the orthotopic model (one-way ANOVA). However, a significant difference in hypoxia levels was observed within the tumor volume groups of the subcutaneous mouse model (p < 0.001). A comparison of hypoxia burden for the <300 mm 3 tumor groups revealed no significant difference (p = 0.053) between models. In contrast, significant differences (p = 0.002 and p < 0.0001, correspondingly) between xenograft models were observed for the 300-700 and >700 mm 3 groups, largely due to the substantial increase in hypoxia levels in the subcutaneous models for tumor volumes exceeding 300 mm 3 . Specifically, the hypoxia burden increased 4.0-fold in the subcutaneous model for the 300-700 mm 3 group compared to the <300 mm 3 group (p < 0.001). In both tumor models, the hypoxia levels decreased on average by 30-50% when the tumor size exceeded 700 mm 3 compared to the respective 300-700 mm 3 group (p > 0.05).
The functional tumor blood vessel density of the excised PC-3 tumors are depicted in Fig. 3b. On average, the subcutaneous tumors had a slightly higher functional vascular density compared to the orthotopic tumors. For both xenograft models, there was a trend toward higher vascular density as tumor volume increased (Fig. 3d). For the subcutaneous model, there was a 2.4 and 5.2-fold increase of functional vascular density in the 300-700 (p = 0.013) and >700 mm 3 groups (p = 0.0041) relative to the <300 mm 3 group. With respect to the orthotopic Tumor volume (mm 3 ) Mean Intensity/µm 2 (mean ± SEM) *

Subcutaneous Orthotopic
Hypoxic burden Vasculature Density Hypoxic burden Vasculature Density <300 9.47e-6 ± 1.74e-6 4.06e-6 ± 8.54e-7 4.33e-6 ± 1.41e-6 8.33e-6 ± 1.42e-6 300-700 3.84e-5 ± 6.98e-6 9.90e-6 ± 1.88e-6 1.18e-5 ± 2.64e-6 8.67e-6 ± 6.44e-7 >700 2.69e-5 ± 2.98e-6 2.11e-5 ± 4.30e-6 5.75e-6 ± 1.45e-6 1.39e-5 ± 1.81e-6  www.nature.com/scientificreports www.nature.com/scientificreports/ tumors, no increase in vascular density was observed for the 300-700 mm 3 (p = 0.82) group compared to the <300 mm 3 group. Although, a 1.7-fold increase in vessel density was observed for the >700 mm 3 (p = 0.084) tumor volume group relative to the smaller groups. For tumor group comparisons between models, the orthotopic tumors of the <300 mm 3 group had a significantly (p = 0.0021) higher vascular density compared to the subcutaneous model, but no statistically significant differences were found between the 300-700 and >700 mm 3 groups (p = 0.55 and p = 0.11, correspondingly). H&E staining of non-target tissues. Upon dissection of the mice from the orthotopic mouse model with tumors exceeding 700 mm 3 , gross anatomy abnormalities of the kidneys, pancreas and liver were perceived. The kidneys were often found to be pale and enlarged and the formation of fluid-filled cysts on the surface of the kidneys was observed, likely due to hydronephrosis. In several cases, enlargement and/or discoloration of the liver and pancreas were also noted. The tissues taken from orthotopic models along with an analogous subcutaneous model (>700 mm 3 group) and normal control mouse were sectioned and underwent H&E staining, see Fig. 5. All of the sectioned tissues associated with the orthotopic model were more diffuse than subcutaneous and normal controls probably due to the increased interstitial volume resulting from reduced urine output. For the orthotopic tumor group, micrometastases were observed in the pancreas, but, surprisingly, were not found in the liver or kidney sections. No micrometastases were observed in the sections obtained from the subcutaneous model.

Discussion
The purpose of this work is to investigate how the vascular density and perfusion of tumors impact the hypoxia burden and drug delivery of BB2r-targeted agents. Specifically, we investigated these factors in subcutaneous and orthotopic xenografts across a range of tumor volumes. By understanding this, we hope to provide researchers a better understanding of these biological variables as well as the advantages and limitations of both subcutaneous and orthotopic models for BB2r-targeted agent development.
We first synthesized and characterized a new hydrophilic, BB2r-targeted radioconjugate ([ 177 Lu] Lu-DOTA-SP714). In addition to examining it's chemical and in vitro properties, we also accessed the solution stability of the radioconjugate. This investigation was prompted out of the need to perform in vivo experiments with large numbers of mice and the desire to reduce the need for the constant synthesis of the radioconjugate. It is well established that ionizing radiation (e.g., α 2+ and β − particles) can directly or indirectly, through the generation of solvated electrons and free radicals species, degrade peptides and proteins 27 . Ascorbic acid and selenomethionine have been demonstrated to have a protective effect against radiolytic degradation in both preclinical and clinical studies [28][29][30] . Indeed, Chen and co-workers examined these radioprotectants among others with [ 177 Lu] Lu-AMBA, a clinically investigated BB2r-targeted agent 31 . Similar to their findings, our studies revealed that the combination of the two radioprotectants, ascorbic acid (40 mg/mL) and selenomethionine (0.2 mg/mL), was better than ascorbic acid alone with approximately 95% of the radioconjugate intact by 72 h. www.nature.com/scientificreports www.nature.com/scientificreports/ The establishment of the subcutaneous and orthotopic PC-3 xenograft model was carried out over a four to eight-week period depending on the model and the desired tumor size. The measured tumor size for orthotopic tumors initially outpaced subcutaneous tumors, but the subcutaneous tumors demonstrated a more rapid growth at approximately 4-weeks post-inoculation. The disparity of growth rates between two models likely stem from the different measurement techniques employed and the initial cell numbers administered. Based on body weight measurements, the overall health of the mice appeared good across models and tumor size ranges investigated ( Figure S5). However, for the largest orthotopic tumor volume group (>700 mm 3 ), abnormalities were found upon gross examination of the kidneys, liver and pancreas. This prompted the histological examination of these tissues for both models. The tissues obtained from the orthotopic model were more diffuse and were consistent with larger interstitial water content, likely due to urinary obstruction resulting from the PC-3 tumor in the prostate. This observed condition has been noted for other orthotopic prostate cancer mouse models 32 .
The examination of the perfusion characteristics of the two models revealed some interesting findings. The average perfusion of the tumors increased as tumor size increased. However, the orthotopic model showed substantially higher tumor perfusion (3.7-4.3 fold) compared to the subcutaneous model. This is interesting given the similar amount of functional vasculature in both models based on Hoechst staining. This comparably higher perfusion rate has also been observed in other orthotopic relative to subcutaneous models [33][34][35] . A factor that may be contributing to the lower perfusion of subcutaneous tumors is a higher interstitial fluid pressure (IFP). In general, the functional vasculature of the orthotopic tumors is more spatially distributed and have larger vessel diameters relative to subcutaneous tumors 36,37 . These factors have been shown to reduce the level of IFP. With that said, as we discussed earlier, the perfusion agent ( 99m Tc-pertechnetate) exhibited a lower rate of renal clearance possibly overstating the magnitude in the differences in perfusion between the two models. Careful consideration should be given regarding the blood and renal clearance of the radiotracer when comparing tumor uptake values.
The BB2r-targeting efficacy of the [ 177 Lu]Lu-DOTA-SP714 was examined in both PC-3 xenograft models. Generally, the average %ID/g uptake of the radioconjugate increased as the tumor size increased. Specifically, orthotopic tumors exhibited significantly higher uptake compared to subcutaneous tumors across tumor groups. This is almost certainly due to the increased perfusion of orthotopic over subcutaneous tumors.
Using pimonidazole, the hypoxia burden of the tumors was evaluated for both models. For the orthotopic tumors, no significant differences in hypoxia levels were observed across the tumor volume ranges. However, the subcutaneous tumors exhibited significant increases in hypoxia with tumor sizes that exceeded 300 mm 3 . The increased hypoxia levels for the 300-700 and >700 mm 3 tumor groups may be attributable to the spatial distribution of the functional vascular on the periphery of tumors. For small tumors, the peripheral vasculature can adequately perfuse the whole tumor. However, as the tumor grows vasculature may become incapable of perfusing the center of the tumor resulting in significant increases in hypoxia burden 38 and tumor necroses [39][40][41][42] . While hypoxia was observed for both models, medium to large subcutaneous tumors more reliably gave tumors with significant fractions of hypoxia. This phenomenon has been observed by other researchers as well 43,44 . In this particular study, we did not examine the extent of tumor necroses and its correlation to PC-3 tumor volumes to determine its impact, if any, on BB2r-targeted agent delivery. This is certainly an interesting question and www.nature.com/scientificreports www.nature.com/scientificreports/ something we may examine in our future work. Overall, these findings suggest PC-3 tumors >300 mm 3 may be a more appropriate volume to investigate the efficacy of hypoxia-targeted or hypoxia-selective agents.

conclusion
Orthotopic models are known to better simulate clinical prostate cancer, particularly with respect to the tumor microenvironment, compared to subcutaneous models. To better understand how the biology of these tumor models' impact BB2r-targeted agent delivery, we examined the tumor vascular perfusion, microvasculature density and hypoxia burden of orthotopic and subcutaneous PC-3 xenograft mouse models. Compared to the subcutaneous model, the results demonstrate that the orthotopic PC-3 tumors have higher vascular perfusion that leads to higher BB2r targeting as well as a lower hypoxic burden. While the vessel density was slightly lower in the orthotopic model compared to subcutaneous ones, no statistical significance was observed. In general, for both models, BB2r-targeting, perfusion and vascular density increased with increasing tumor volume. Immunofluorescence and autoradiography illustrated the microdistribution pattern of the BB2r-targeted conjugate relative to functional vasculature and hypoxic regions. As expected, higher concentrations of the radiolabeled conjugate were observed near functional vasculature compared to hypoxic regions that were devoid of functional vasculature. Overall, this work demonstrates that the tumor microenvironments of orthotopic and subcutaneous PC-3 tumors are impactfully different in terms of drug delivery. Careful consideration should be taken when comparing the data of BB2r-targeted agents, as well as other targeted agents, in orthotopic and subcutaneous tumor models.

Materials and Methods
Full details regarding the materials and equipment are presented in the supplemental materials. Also given in the supporting information is our methodology concerning peptide synthesis, cell culture, luciferase transfection, distribution coefficient studies and competitive binding studies. For the generation of the orthotopic model 45 , the PC-3-Luc cells were selected by 0.5 µg/mL puromycin twice before inoculation. The male 6-weeks SCID mice were anesthetized, the muscles of the abdomen area were cut after retracting the skin, and the prostate gland exposed. 50 μL of PC-3-Luc cell suspension at 0.5 × 10 6 /mL in Matrigel ® was injected into a dorsal prostatic lobe. The wound was closed in two layers and the skin was clipped.
Animals were given analgesic drugs for 3 days and the tumors were monitored by an IVIS optical imaging system.

Bioluminescent imaging.
For in vitro imaging, PC-3-Luc cells (30,000 cells/well) were diluted and plated in a 96-well plate. D-luciferin (50 μL, 150 μg/mL) was added to the media five min prior to imaging ( Figure S1). For in vivo imaging, mice were given D-luciferin (100 μL, 15 mg/mL) 15-20 min prior to anesthetization by isoflurane. At imaging, the mice were transferred to the IVIS enclosure and images were acquired by IVIS ® Spectrum software with auto-exposure. Regions of interest from each image were selected and quantified.
In Vivo biodistribution studies. The mice were injected, monitored and dissected in the lab with the temperature of 23 °C and 17% relative humidity. All animals were awake until sacrifice. The body weight, tumor volume and tumor luminescence of the mice were recorded every three days. At four-to-seven weeks post-xenograft implantation, the mice were divided into three groups by tumor volume: <300, 300-700 and >700 mm 3 in both animal models. Tumor volume groups of the subcutaneous models had an N = 8 (<300 mm 3 ), 9 (300-700 mm 3 ) and 9 (>700 mm 3 ) of mice separately. The orthotopic tumor volume groups had an N = 9 (<300 mm 3 ), 8 (300-700 mm 3 ), and 6 (>700 mm 3 ) of mice separately. Each mouse (average weight: 20 g for female mice and 25 g for male mice) was treated with pimonidazole solution (80 mg/kg in PBS) via intraperitoneal injection. After 1 h, the mice received an intravenous injection of 10 µCi (370 kBq) of the radio-RP-HPLC peak purified 177 Lu-labeled conjugate ([ 177 Lu]Lu-DOTA-SP714) in 100 μL of PBS. After an additional 1 h, the mice were injected intravenously with a PBS solution containing 10 µCi (370 kBq) of [ 99m Tc]NaTcO4 and 15 mg/kg Hoechst 33342. The animals were sacrificed 5 minutes later, and their tissues collected. The excised tissues were weighed, the radioactivity for each tissue was measured by γ-counter and the percentage injected dose per gram (%ID/g) was calculated for each tissue.
Microscopy and autoradiography. At the end of the biodistribution studies, the tumor, liver, kidney and pancreas from the mice were rinsed by deionized water, dried and embedded by O.C.T compound on dry ice. The adjacent cryostat tumor slides (10 µm) were scanned for Hoechst 33342 and anti-pimonidazole-FITC by confocal microscopy and exposed to a storage phosphor screen to be scanned by the Typhoon imaging system using 25 µm