A novel patient-derived intra-femoral xenograft model of bone metastatic prostate cancer that recapitulates mixed osteolytic and osteoblastic lesions

A primary prostate cancer bone metastasis sample was obtained from a lytic lesion in the proximal femur from a patient with castrate-resistant prostate cancer with mixed osteoblastic and osteolytic bone metastases and a Gleason score of 9 (5+4). Tumor specimen was prepared aseptically in biohazard safety cabinet according to standard protocols with minor modifications [28, 43‐47]. Specimen was first minced with sterile razor blades to 1-3 mm³ sized pieces. A portion of the minced tumor was snap frozen for genomic DNA and RNA extraction, cryopreserved in 10%DMSO/90% FBS, or fixed for immunohistochemistry. For sub-cutaneous transplantation, the minced tumor was mixed 1:1 with High Concentration Matrigel (Becton-Dickinson) and 0.1 ml was injected. For intra-femoral injection, the minced tumor sample was disaggregated by digestion in Accumax (Millipore), filtered through sterile, mesh filter (Falcon). Dissociated cells were centrifuged at 1200 RPM, 5 minutes, 4°C, washed three times and resuspended in Iscove's modified DMEM media, 10% FBS at 6.7 × 10⁶ cells/ml. Cells were mixed 1:1 with high concentration Matrigel for intra-femoral injection of 50,000 cells in 15 μl. Remainder of the dissociated tumor cells were cryopreserved or used for DNA and RNA purification. All studies with human subjects were conducted with the approval of the University of California, San Diego School of Medicine Institutional Review Board. All patients provided written informed consent.

All animal protocols were preformed under a UCSD animal welfare IACUC approved protocol. Sub-cutaneous injections were performed using standard protocols [26, 28, 33]. Briefly, male Rag2 ^-/- ;γ _c ^-/- mice 6-8 weeks old were anesthetized with ketamine/xylamine, skin sterilized with 70% ethanol, a 2-3 mm incision was made with autoclaved dissection scissors, a trochar (10 ml *LDEV-Free, 14-gauge catheter (SC injections) Terumo 14 G IV Catheter) was used to inject 100 ul of tumor/matrigel mix below skin right flank, skin flaps were brought together and sealed with VetBond, mice revived post-surgery with Antisedan injected sub-cutaneously at base of neck ruff. For intra-femoral injections mice were anesthesized by intra-peritoneal injection of a mix of 100 mg/kg ketamine and 10 mg/kg Xylazine and injections performed in a BL2 biosafety cabinet. Right hind limb was prepared under standard sterile conditions with 70% ethanol. Knee was held in flexed position and 25 G needle (Monoject 200 25 × 5/8A) was used to make a port in the femoral plateau until there was no resistance that was used as a guide-hole for injection of 15 ul of the tumor cell/Matrigel suspension using a 0.3 ml syringe and 27 G needle. Injection of sample was performed slowly with minimal resistance. Needle was withdrawn and leg immediately straightened, dabbed with antibiotic ointment (RX Neomycin, Polymyxin, Bacitracin Ophthalmic Ointment USP Sterile NDC 13985-017-55), on a sterile cotton tipped applicator (Q-Tips) and held for straight for approximately 1 minute. Mice were injected with Antisedan and placed on a warm Deltaphase Isothermal Pad, and carefully watched during recovery until ambulatory and active.

Prostate cancer cell lines: LAPC4, was a gift from Dr. Lily Wu, UCLA, and VCaP, purchased from ATCC, were maintained in Iscove's media, 10%FBS, penicillin-streptomycin and K562, a chronic myelogenous leukemia cell line in 10% heat-inactivated FBS, RPMI, Pen-strep.

Genomic DNA and RNA were extracted using mortar and pestle pulverization of flash frozen tumor pieces in liquid nitrogen and the Qiagen All-prep kit [47]. RNA was re-purified with RNeasy and treated with RNAse-free, DNase to remove contaminating genomic DNA. For cell lines and purified, dissociated xenograft tumor cells, RNA was extracted using Qiagen RNeasy mini-prep kit. cDNA synthesis was performed with Superscript III (Invitrogen, Inc.) according to manufacturer's protocol, and used for PCR (Taq polymerase, Monserate Biotechnology Group LLC, San Diego, CA). RT-PCR products were resolved on 1% agarose gels. All RT-PCR primers are shown in Table 1. RT-PCR products of the correct size were verified by sequencing (Retrogen, Inc., San Diego, CA).

PSA immunostaining was carried out using rabbit anti-human PSA antibody (DAKO A0562) using standard protocols [48] performed by the Moores Cancer Center Histology Core, UCSD, La Jolla, CA. Paraformaldehyde-fixed and paraffin embedded sections from sub-cutaneous PCSD1 xenografts were mounted and 6-μm sections were stained with H & E, anti-PSA, or rabbit IgG isotype control (DAKO N1699) using HRP goat anti-Rabbit as secondary antibody (Jackson 111-035-144) and AEC (Vector SK4200). For intra-femoral tumors, the tumor plus femur and tibia were dissected out as one, formalin-fixed and EDTA de-calcified according to Lavoie et al. [49]. Tissues were mounted in OCT, 6 μm cryosections were fixed in acetone, blocked in 1%BSA/PBS, incubated with anti-PSA or IgG isotype control antibody then processed as above.

Femurs of mice injected intra-femorally with PCSD1 were scanned by micro-computed tomography (μCT) SkyScan 1076 (Skyscan, Belgium) at the maximal potential 60 kV and 167 μA with 0.5 mm thick aluminum filter and at the voxel resolution of 9 μm. The μCT scans were performed over 360° of total rotation with each angular rotation step of 0.7°. The reconstructions, performed using the NRecon software package (Skyscan), are based on the Feldkamp algorithm and resulted in axial grayscale images. The 2D images were created using CTAn software package (Skyscan) [50]. The 3D μCT models of each femur were created using a 3D reconstruction software package (Mimics 14.0, Materialise, Belgium) [50, 51].

A prostate cancer bone metastasis specimen was obtained from a castrate-resistant patient and transplanted sub-cutaneously or intra-femorally into immunodeficient, male Rag2 ^-/- ;γ _c ^-/- mice [52]. Minced tumor sample that was injected sub-cutaneously (SQ) produced xenograft tumors in all ten male Rag2 ^-/- ;γ _c ^-/- mice. Disaggregated primary tumor cells that were injected intra-femorally (IF) generated tumors in all eight Rag2 ^-/- ;γ _c ^-/- mice. As shown in Figure 1, tumors were evident in three representative mice at ten weeks in the tumor-injected (right) leg of all intra-femorally transplanted mice but not in the un-injected, contra-lateral (left) leg. Therefore, the take-rate of the primary tumor sample was 100% in both the sub-cutaneous and intra-femoral niches. Tumors harvested from both sub-cutaneous and intra-femoral tumors have been serially transplanted at least three times both sub-cutaneously and intra-femorally thus far: P0 (primagraft), P1 and P2. Low passage PCSD1 tumors were cryopreserved and serially passaged as intra-femoral and sub-cutaneous xenografts. Tumor take-rates are shown in Table 2. The lower take-rate in the intra-femorally injected mice is most likely due to the significantly fewer tumor cells injected into the femur than sub-cutaneously. Approximately 5,000 tumor cells were injected per femur which was ~10% of the total mixture of 50,000 cells injected IF per mouse. In contrast, the minced tumor pieces that were implanted sub-cutaneously were ~ 1 mm³ containing approximately one million total cells. Mice injected IF with fewer as well as greater than 5,000 PCSD1 cells are currently being analyzed. Freshly harvested xenograft tumor cells as well as cryopreserved xenograft tumor cells have been used for long term in vitro culture experiments for testing novel compounds.

To demonstrate whether the xenograft tumors originated from prostate cancer in the patient bone metastasis specimen the expression of prostate specific antigen (PSA) was measured. Primers that were specific for the human gene target and spanned exon-intron boundaries were newly designed or selected from the literature as shown in Table 1 and the RT-PCR products verified by sequencing. As shown in Figure 2A, RT-PCR analysis showed the expression of human PSA in a sub-cutaneous PCSD1 xenograft tumor (P1) as well as in the human prostate cancer cell line, LAPC4, but not the human chronic myelogenous leukemia (CML) cell line, K562, nor murine bone marrow, spleen or liver [52]. Using primers for the full-length isoform of androgen receptor (AR) for RT-PCR demonstrated human androgen receptor (AR) expression in PCSD1 and LAPC4 (Figure 2A)[53, 54]. Therefore, PCSD1 xenograft tumors originated from prostate cancer cells in the patient's femoral bone metastasis.

PSA protein expression was determined using immunohistochemical staining of PCSD1 xenograft tumor sections. Cytoplasmic PSA staining was detected in cells in sub-cutaneous PCSD1 xenografts (Figure 2B) and in intra-femoral xenografts from the right leg (Figure 2C, IF 2°TP (RL), lower panels). Cytoplasmic PSA staining was not observed in femoral sections from the un-injected, contra-lateral left leg (LL). Red blood cells in the bone marrow space that showed up as slightly reddish brown in color that was not due to PSA immunostaining were seen in the un-injected intra-femoral sections (Figure 2C middle panels). In the femur of the right leg, PCSD1 tumor cells were observed both in the endosteal bone marrow space where they were injected and having invaded extra-cortically surrounding the femur. Regions of osteolysis were observed in the immunostained sections and in H & E stained sections (Figure 2C, upper panels) through which the tumor may have invaded and migrated outside of the bone.

Further molecular analysis to characterize PCSD1 tumors was performed using RT-PCR on additional human prostate biomarkers. Expression of keratins 5 (K5) and 14 (K14) are characteristic of basal prostate epithelial cells whereas keratins 8 (K8) and 18 (K18) are expressed in luminal prostate epithelial cells [44‐46, 55]. PCSD1 xenograft tumors expressed human K8 and K18 and very low levels of K5 and K14 (Figure 3A). Interestingly, LAPC4 expressed all four of the keratins. PCSD1 and LAPC4 both expressed the prostate transcription factor NKX3.1 [46]. PCSD1 and LAPC4 also expressed AMACR, a biomarker that is often up-regulated in advanced prostate cancer [56]. The human specific GAPDH and mouse specific GAPDH were expressed in the PCSD1 xenografts indicating the presence of both human and murine cells within the xenograft tumor [57, 58]. Only human specific GAPDH was detected in the human cell lines LAPC4 and K562 that were grown in culture as expected. Correspondingly, the murine bone marrow and spleen tissues only expressed the mouse GAPDH. Taken together the results of the molecular analysis showed that PCSD1 is a luminal epithelial-type advanced prostate cancer [43‐47].

The TMPRSS2-ERG fusion gene is a frequent genomic rearrangement in prostate cancers that results in placing the ERG ETS-family transcription factor under the androgen-regulated expression of the TMPRSS2 gene [59]. RT-PCR was performed to determine whether this gene fusion event was present in PCSD1. While the fusion transcript was detected in VCaP cells as shown previously [59], the TMPRSS2-ERG gene fusion was not detected in PCSD1 (Figure 3B). In addition, analysis of known alternative splicing variants of AR did not detect these in PCSD1 xenografts.

Micro computed tomography small animal scanning (microCT) was performed on mice injected intra-femorally with PCSD1 to determine the effect of the growth of the tumors [31]. MicroCT scans from mice injected intra-femorally with the primary patient bone metastasis sample are shown in Figure 4. Regions with significant osteoporosis (bone thinning) as well as areas of bone sclerosis (increased bone density) were apparent in the femur in which the tumor was growing but not in the un-injected contra-lateral leg (Figure 4A). Three-dimensional (3D) reconstruction of the microCT scans revealed the extensively pitted, porous and eroded distal extremity of the femur from the tumor-injected (yellow) leg compared to the smooth femur surface contours of the contra-lateral, un-injected leg (Figure 4B). PCSD1 tumor growth produced significant osteolysis in the femur. In addition, regions of sclerosis or osteoblastic lesion formation were observed in the right, tumor injected femurs as shown in Figure 5A. MicroCT cross-sections along the length of the femur were compared to show the increased thickness and density of the femur in which PCSD1 tumor was growing compared to the un-injected femur (Figure 5B). The PCSD1 tumor induced bone lesion changes can be seen in the context of the whole animal in the movie of the 3D reconstruction of the microCT analysis (Additional file 1). The destructive effects of PCSD1 tumor growth on the internal and external surfaces of the distal femur are further shown in the 3D reconstruction microCT in Additional file 2. Therefore, in vivo microCT scanning revealed PCSD1 tumor growth produced mixed osteolytic and osteoblastic lesions. This recapitulated the mixed osteoblastic and osteolytic bone lesions observed in the patient.