Castration induces tumor recurrence, and an increase in TNF and stemness in prostate tumors from PTEN-deficient mice.

Mice bearing PTEN-deficient tumors were either castrated (Cx1-4) or sham-operated (Sh1-3) and serially imaged using HFUS. In some cases, tumor tissue was recovered for further analysis. A. 3D reconstruction of tumor-bearing ventral lobes (pseudo-colored green) performed as described in Materials & Methods. B. Tumor volume kinetics post-castration. Tumor volume was determined from HFUS images using Amira software. Mean tumor volume at the indicated times post-castration (n=4, red) or post-sham operation (n=3, blue) is shown as percent pre-castration volume (mean [%] ± SEM). C. TNF levels in tumors tissue extracts were measured by ELISA and normalized to total protein in the tissue sample, at the indicated times post castration. D. LSC FACS analysis of tumor cells from sham (left) and castrated (Cx; right) PTEN-deficient mice. Tumors were disaggregated and cells were processed for Lin-/Sca-1/CD49f FACS analysis (23). Relative fluorescence intensities are shown as contour plots. The LSChi population is labeled ‘Basal/stem’. The population in the lower right quadrant (LSCmed) corresponds to luminal progenitors. A representative analysis is shown. Sham LSC FACS was replicated 5 times (mean[%]±SEM = 27.0±4.0); 25d post-castration LSC FACS was replicated 4 times (mean[%]±SEM = 38.2±1.7).

Extended enzalutamide treatment selects for basal stemness and TNF secretion.

C4-2 cells were grown in media plus 10% serum and treated with vehicle (blue) or 10 μM enzalutamide (Enz; red) for the indicated time. A. C4-2 cell growth curve in the presence of enzalutamide. Trypan-blue excluding live cells were counted microscopically at the indicated times. Data are shown as mean ± s.e.m. (n=3). B. FACS analysis for basal cell stemness markers in enzalutamide treated C4-2 cells. Cells grown in enzalutamide for the indicated time, were incubated with the fluor-labeled antibodies, analyzed by FACS and relative fluorescence intensity represented as dot plots. The fraction of the cells that correspond to the CD49fhi/CD166hi population (upper, right quadrant) is indicated. Additional time points are in Supplementary Figure S2. C. Fraction of cells scoring CD49fhi/CD166hi and corresponding TNF secretion, at the indicated time of culture in media plus enzalutamide. Data from FACS analyses (see Fig. S2) is plotted on the left and TNF levels in the media determined by ELISA is plotted on the right. Values are given as mean ± s.e.m. (n = 2). **, p < 0.01 ***p < 0.001 (two-way ANOVA followed by Tukey-Kramer HSD test).

Enzalutamide treatment selects for a TNF-expressing basal stem cell population.

C4-2 cells were grown for 20 days in media plus 10% serum containing 10 μM enzalutamide, enriched for CD166hi expression. ELISA was used to measure TNF and colony formation. A. CD166-enrichment was performed via FACS to sort CD166hi and CD166lo C4-2 cells treated as indicated. The purity of the unsorted and sorted CD166hi and CD166lo populations are shown. B. TNF secretion of the indicated cultures (left to right): unsorted, vehicle-treated (veh); unsorted, enzalutamide-treated (enz); and sorted, enzalutamide-treated (CD166lo, CD166hi). TNF measured by ELISA, and plotted as mean ± s.e.m. (n=3). **p<0.01, ***p<0.001 (One-way ANOVA followed by Tukey-Kramer HSD test). C. Colony formation assay of enriched CD166hi and CD166lo cell populations sorted as in A. Seeding densities were 500, 1000, and 1500 cells/well, respectively. Colony counts per well are plotted. *p<0.05 (Student’s unpaired t-test).

Differentially-timed TNF signaling blockade has distinct effects on tumor growth.

Mice with PTEN-deficient tumors were castrated and etanercept (sTNFR2-Fc) was administered at the indicated times: three days before (-3d), one day before (-1d), or three days after (+3d) surgical castration. A. Tumor volume kinetics. Tumor volumes were determined by serial HFUS imaging and 3D reconstruction and plotted after normalizing to the initial tumor volume for each mouse. Different colors/symbols represent individual mice. The table summarizes regression and recurrence incidence for each experiment. Growth curves for castrated mice that did not receive etanercept (None) are not shown. The criteria for regression and recurrence are described in the Methods. B. Gene set enrichment analysis. RNA was extracted from PTEN-deficient tumors from the following mice: sham-operated, 6d post-castration and 35d post-castration. GSEA was performed on bulk RNA-seq data using gene sets corresponding to TNF-related pathways. Normalized enrichment scores (NES) from 6d (down-regulated, blue) and 35d (up-regulated, red) post-castration were compared to the sham group. Gene sets common to the two analyses are yellow.

CCL2-CCR2 signaling is required for castration-resistant recurrence.

Mice bearing PTEN-deficient (A-C) or hi-MYC (D) tumors were sham-operated or castrated at the indicated times. Volume was monitored by HFUS. Some mice received etanercept (E) or the CCR2a, as indicated. A. Tumor CCL2 levels following castration. CCL2 levels were measured by ELISA on the same tumor extracts prepared for Fig. 1C, normalized to total protein and plotted as open circles. The bar marks the mean. ****p < 0.0001 (Two-way ANOVA followed by Tukey-Kramer HSD test). B. Representative images of IHC-stained CCL2+ cells (red) in tumors from the indicated treatment groups. CCL2+ cells were counted and plotted. **p < 0.01, *p < 0.05 (Two-way ANOVA followed by Tukey-Kramer HSD test). C. Treatment with a CCR2 antagonist phenocopies -1d etanercept treatment (Fig. 4A) in PTEN-deficient tumor-bearing mice. Tumor volumes were determined by serial HFUS imaging and 3D reconstruction and plotted after normalizing to the initial tumor volume for each mouse (left). Different colors/symbols represent individual mice. The recurrence incidence is summarized (right). D. CCR2 antagonist suppresses castration-induced regression in Hi-MYC prostate cancer-bearing mice. Red circles: mice were castrated and received vehicle. Blue boxes: mice were castrated and received CCR2a prior to castration. N=5 for each group. The recurrence incidence is summarized (right). Tumor volume kinetics for individual mice are in Supplementary Figure S5.

Transcriptomic changes in PTEN-deficient prostate cancers following CCL2 signaling blockade.

RNA was extracted from PTEN-deficient tumors from the following mice: sham-operated (Sham), 35d post-castration (35d post-CX) and 35d post-castration after receiving CCR2a (35dp-CX+CCR2a). DGE analysis was performed on bulk RNAseq data (Fig. S6) followed by Gene Ontology analysis of biological processes comparing RNAseq transcript sets derived from tumors in mice that were sham-operated vs mice 35d post-castration (A) or from tumors in mice 35d post-castration vs mice treated with CCR2a 35d post-castration (B). Red-colored bars represent up-regulation and blue-colored bars represent down-regulation. Fold-enrichment >5. C. TAM-related transcripts regulated by CCL2 signaling. Heatmap representation of mRNA levels of select TAM-associated genes for the same two comparisons shown in A (left) and B (right).

TNF and CCR2 signaling blockade suppress castration-induced immunosuppression at late times following castration.

Mice bearing PTEN-deficient tumors were sham-operator or castrated (Cx) and tumors harvested at the indicated times post-castration. Some mice received etanercept (Etan) or the CCR antagonist BMS IHC staining for stromal F4/80+ macrophages and CD8+ T cells was performed and the number of stained cells per unit area were plotted. A. TNF signaling blockade. Images of representative IHC-stained tissue sections for the surface marker protein noted on the left side. Mice were treated as shown across top and the cell densities plotted for each of the treatments, below. B. CCL2 signaling blockade. Similar to A with both surface marker protein and corresponding mouse treatments shown along the top. A complete set of images are provided in Supplementary Figure 7C. Staining and counting of the Cx-35d samples in A and B were performed separately. Data are given as mean ± s.e.m. (n=4). *p<0.05, **p<0.01, ***p<0.001 (One-way ANOVA followed by Tukey-Kramer HSD test).

Tumor recurrence following castration correlates with an immunosuppressive tumor microenvironment. A.

Post-castration tumor volume kinetics for three representative PTEN-deficient mouse tumors. Tumor volumes were determined by serial HFUS imaging and 3D reconstruction and plotted after normalizing to the initial tumor volume for each mouse. The average starting volume for these tumors was ∼960 mm3. B. t-SNE plots for single cell RNAseq analysis. At 42d post-castration, mice were sacrificed, tumor tissue was disaggregated into single cell preparations, RNA sequenced and analyzed as described in the Methods. Dimensionality reduction of cell transcriptomes from all three mice was performed with tSNE to produce the plot in far left panel (Aggregate). Cells are assigned to 6 subset populations as coded in the key in the lower right corner. Color-coded cells from each of the three mice corresponding to the response patterns in A are shown in the remaining three panels of B. Cells which were in the aggregate population, but not identified in individual mice are grey colored. TAMs (defined in panel C) are within the large, heterogeneous myeloid population (green). C. CTL:TAM ratios for regressing, stable or recurrent mouse tumors. Transcriptomic data from B was used to define TAMs (CD45+ and CD68+and CD80+; lighter grey) and CTLs (CD45 and [CD3e or Cd3d or CD3g] and [CD8a or CD8b1]; darker grey). See text for discussion.