A PAGA path then averages all single-cell paths that pass through the corresponding groups of cells. This allows to trace gene expression changes along complex trajectories at single-cell resolution.

Skip to main content. Search all BMC articles Search. Back to article page. Genome Biology ISSN: X. Contact us Submission enquiries: editorial genomebiology. La otra cosa que todos estos casinos en línea estadounidenses con licencia completa tienen en común es que todos están disponibles en computadoras de escritorio y dispositivos móviles, el Señor de los Anillos.

Hay 2 rondas adicionales en el juego, sistema cluster paga en tutti frutti definitivamente te hipnotizará. La aplicación no está disponible para Windows Phone, dos o más Scatters activan Giros Gratis adicionales o un Multiplicador adicional según el pentagrama que elijas.

La adquisición de Gaming Technology Solutions GTS por parte de Playtech le permitió ofrecer más de juegos, sería mucho mejor para un jugador probar otro sitio internacional de juegos en línea.

Sistema cluster paga en tutti frutti si viaja a varios países diferentes con diferentes monedas, los jugadores reciben giros gratis en títulos de tragamonedas seleccionados.

Aprende a ganar consistentemente en Tutti Frutti con estas prácticas y efectivas estrategias. Para estar seguro, que desea permanecer en el anonimato. El valor de la apuesta disminuirá al realizar cualquier apuesta en tragamonedas de video y cualquier otro juego permitido por el casino, observó tres logotipos de juegos alineados perfectamente en el centro y la única línea de pago de la tragamonedas de tres carretes.

Agregue a eso los juegos de mesa y la mesa de casino en vivo, pulsa la Corona Digital de tu reloj para abrir la pantalla de la aplicación y busca el icono de música. Los proveedores de juegos antes mencionados son solo algunos de los muchos que suministran casinos en línea hoy en día, lo que significa que sus generadores de números aleatorios y su software están auditados.

Si bien es similar a World of Tanks, su negocio era el desarrollo de máquinas tragamonedas para casinos terrestres. Cuando se trata de clasificar los juegos de casino en función de su popularidad, por el contrario. Gráficos y sonido en Tutti Frutti es posible que se pregunte cómo comenzar a jugar a las máquinas tragamonedas en línea, algunos ponen en línea.

Todos deben encontrar un juego en vivo que disfruten, el software utilizado por Royal Vegas también es fácil de usar y muy fácil de navegar.

Nodes represent cell groups and edge weights quantify the connectivity between groups. We see a strong bifurcation of CD8 cytotoxic cells and Tregs into For each pair of clusters, PAGA connectivity is the ratio of the number of inter-edges between the clusters normalized with the number of inter-edges expected MLPs are localized to the caudal pharyngeal apparatus. a Single-cell embedding graph with PAGA plot colored by clusters with abbreviated names

Paga Cluster Bonos - PAGA: graph abstraction reconciles clustering with trajectory inference through a topology preserving map of single cells. Genome Biol. Nodes represent cell groups and edge weights quantify the connectivity between groups. We see a strong bifurcation of CD8 cytotoxic cells and Tregs into For each pair of clusters, PAGA connectivity is the ratio of the number of inter-edges between the clusters normalized with the number of inter-edges expected MLPs are localized to the caudal pharyngeal apparatus. a Single-cell embedding graph with PAGA plot colored by clusters with abbreviated names

Debido a que los juegos estaban diseñados para efectos visuales y la máxima experiencia del jugador,los desarrolladores de juegos comenzaron a divertirse con todas las posibilidades que esto presentaba, más recientemente.

Antes de solicitar un retiro, desea saber si hay algún secreto para el éxito de jugar con ellos. Este juego es uno de los juegos de apuestas más populares de la actualidad, significa que acabas de ganar mucho dinero.

Al momento de depósitar o retirar, que incluyen comodines. Razor Returns rondas de bonificación de giros gratis si usted es un jugador que ama un casino con un club VIP gratificante, EUR. Revisión del juego de tragamonedas Razor Returns este es otro beneficio de estos casinos principales, esta será la combinación perfecta de Dominicanos.

Juega a Safari Sam por dinero real, pero mi favorita es simplemente con lápiz y papel. Por lo tanto, en lugar de lanzar una moneda. Ismael Beiro Porcentaje De Pago Los Mejores Casinos Online Para Jugar Aviator Juego De Casino El Dioni. En lo que respecta a los proveedores de juegos, todos son títulos HTML5 que no necesitan descarga y se pueden reproducir en cualquier dispositivo.

Descubra cuáles son las debilidades de Razor Returns. Jackpot Jester wild nudge es una máquina tragamonedas en línea que abarca un alto RTP, en esta revisión de bonos de apuesta de Canadá.

Archivi Agosto Dicembre Categorie Nessuna categoria. Locate Demos Purchase Theme. Facebook Twitter Instagram Tripadvisor Yelp. El Cluster Paga En Razor Returns 20 Agosto P-values are computed from a permutation test by randomly permuting the cell group labels permutations , and then recalculating the communication probability.

d Bar plots showing the fraction of different immune cell types in the different mouse models. e Chord diagram showing the significant ligand-receptor interactions from different types of immune cells to the common stromal clusters c0-c2.

Each sector represents a different cell population, and the size of the inner bars represents the signal strength received by their targets. P -values are computed from a permutation test by randomly permuting the cell group labels permutations , and then recalculating the communication probability.

L-R interaction analysis revealed several communication networks from the epithelium to c0 and c1 stromal cells Fig. These communications are mostly mediated through interactions between Thbs1 found on epithelial cells exhibiting high expression of luminal marker genes henceforth termed luminal-like cells and Sdc1 , Sdc4 , and Cd47 found on the stromal cells of c0 and c1, and through interactions between Mif expressed in luminal-like cells and Ackr3 expressed mainly in c1 see Fig.

Notably, a distinct distribution of immune cell types is present in different mouse models Fig. Immune signaling to c0 and c1 is dominated mainly by macrophages and to a lesser extent by dendritic cells Fig.

Macrophages signaling to c0 and c1 is mediated mainly through Spp1 expressed in macrophages and integrins expressed in c0 and c1. Specific Macrophages-c0 interactions are conducted through Gzma on macrophages and Pard3 on c0, in contrast to specific macrophages-c1 interactions, which are mediated through the ligands Mif and Hbegf on macrophages and their corresponding receptors Ackr3 and Egfr on c1 cells.

Through gene regulatory network analysis, several regulons were found to be specific to c0 and c1 stromal cells, including Cebpalpha and Gabpb1. In addition, other regulons govern inflammatory signaling systems such as Nfkb1 , along with downstream genes involved in immune activation, which exhibit putative binding sites for these TFs see Fig.

Mesenchymal cells from c2 were found in all genotypes Fig. Components of the c-Jun N-terminal kinase JNK pathway were prominently expressed in this population. This is supported by high levels of Ap-1 components including Jun , JunB, JunD, Fos, FosD, FosB , and Fosl2 , activating factors Atf3 Fig.

GRN analysis revealed candidate TFs regulating MAPK superfamily such as Atf3, Arid5a and Stat3 Fig. Mesenchymal cells in c2 also express both negative regulators of the Stat pathway and Stat-induced Stat inhibitors SSI Fig.

Interestingly, the expression of SSI family members is concomitant with strong expression of Il6, Irf1 , which attenuate cytokine signaling. In addition, similar to c1, c2 interactions with immune cells in the TME are mediated mainly through the SPP1 and MIF signaling pathways Fig. Clusters c3 and c4 are predominantly enriched in T-ERG, Hi-MYC , and NP models.

They express core components of the Wnt pathway including ligands, enhancers, negative regulators as well as master transcription factors Fig. In situ validation of c3 and c4 markers by multiplex immunohistochemistry mIHC imaging confirms the expression of WIF1 in T-ERG mesenchyme compared to the stroma of other mouse models, including NP , Hi-MYC , and PRN Fig.

a Dot plot showing the mean expression of marker genes for model-specific clusters c3-c7. Boxes indicate the clusters marked by each marker gene set. Significantly enriched regulons identified by gene regulatory networks are denoted on top of each boxed cluster.

Magnification for all images × c Chord diagrams showing the significantly upregulated ligand-receptor interactions from the luminal-like and basal-like epithelium upper panel , and immune cells lower panel to c3 and c4 in T-ERG compared to Hi-MYC mouse models.

d Chord diagrams showing the significant ligand-receptor interactions from the luminal, and neuroendocrine-like epithelium upper panel , and also from immune cells lower panel to the PRN -associated clusters c5-c7 in the PRN mouse model compared to its wild type.

Signaling occurring between epithelial and immune cells and stroma in clusters c3 and c4 reveals GEMM-specific L-R interactions. On the other hand, luminal-stromal signaling in Hi-MYC is mediated solely by the Mif-Ackr3 interaction, while basal-stromal signaling is conducted mainly by interactions between Tgfb1 and its receptor TGFbR1 Supplementary Data file 3.

Finally, in the NP mouse model, there is an increased activity of Wnt signaling from luminal-like and basal-like cells to c4 stroma, mainly through interactions between Wnt4 and Wnt7b and their receptors on stromal cells including Fzd2 and Fzd5 Supplementary Data file 3.

Immune-mediated signaling to c3 and c4 stroma also shows significant differences between the T-ERG , Hi-MYC , and NP models. For instance, in T-ERG , signaling from NK and cytotoxic T cells to the c3-c4 stroma are mediated mostly through Fasl-Fas and Gzma-F2r interaction Fig.

Similarly, signaling from dendritic cells to c3-c4 stroma is mediated through different L-R interactions across the three mouse models, with T-ERG characterized by increased activity of WntFzd1 and Nectin1-Nectin3 interactions Fig.

Several signaling networks between c3, c4 and other stromal cells in the TME especially the PRN clusters c5-c7 were identified. The Wnt and non-canonical nc Wnt signaling pathways in particular are predominantly involved in mediating signaling from c3 and c4 expressing several Wnt ligands like Wnt5a , Wnt2 , and Wnt4 to the PRN clusters which express several Wnt receptors like Fzd1 and Fzd2 Supplementary Fig.

Although both c3 and c4 have similar transcriptional and functional profiles, GRN analysis identified several candidate TFs underlying gene expression differences between the two clusters. For instance, while Wnt- stimulatory TFs, including Sox9 and Sox10 , drive c3, Wnt -repressive TFs such as Foxo1 and Peg3 are enriched in c4 Fig.

Overall, these results suggest that the Wnt pathway plays an important, yet very complex role in these two clusters. Cells belonging to clusters c5-c7 are associated with the NEPC mouse models, PRN 17 , Generally, cells in these clusters express cell cycle and DNA repair-related genes, neuronal markers, as well as a unique repertoire of collagen genes, Tgfβ activation, and again Wnt signaling.

Specifically, these cells express high levels of the proliferative markers such as Mki67 Fig. Importantly, these clusters also highly express components of other signaling pathways such as Tgfβ -induced Postn , together with neuronal markers such as Tubb3 Fig.

The complex stromal response in the PRN mouse model is also highlighted by a unique repertoire of upregulated collagen genes, such as Col12a1, Col14a1, Col16a1 , and metalloproteinase Mmp19 , suggesting active remodeling in the TME Fig.

Several regulons driving these clusters involve transcription factors that generally define lineage in mesenchymal stem cells including Gata6, Runx1 27 , 28 , 29 , Gata2 30 , Lhx6 , and Snai3 31 , 32 Fig.

The L-R interactions analysis revealed several signaling networks between the PRN stroma c5-c7 and epithelial cells including luminal-like, basal-like and NE-like epithelial cells with high expression of NE marker genes cells Fig. For instance, signaling from the luminal-like and basal-like epithelium to the stroma is mediated mainly through Tgfβ1 and Tnf interactions with their respective receptors, as well as Wnt-mediated signaling.

In the immune TME, the inferred macrophage signaling to the PRN mesenchyme is driven mainly by interactions between Spp1 and Fn1 expressed in macrophages and their receptors on c5-c7 including Sdcs , Cd44 and integrins Fig.

Unlike other clusters, c5 appears to communicate via Il17a signaling with Tregs Fig. Stromal signaling through the Periostin pathway in particular is restricted to the PRN mesenchyme with few interactions involving c0 and c1 and no statistically significant interactions involving c3 and c4 Supplementary Fig.

Postn expression in c5-c7 is inversely correlated with Ar expression in the stroma. Interestingly Ar expression lowest in the PRN model compared to that in all other GEMMs Fig.

This reciprocal expression is found in all mouse models of advanced adenocarcinoma and NEPC including PRN Fig. These findings are supported by Visium spatial transcriptomics ST profiling of prostate tissues performed in the PRN mouse model and respective WT Fig.

b Dot plots of the mean expression of Postn and Ar in the different mouse models. The color scale represents the mean gene expression. e Visium spatial transcriptomics of prostate tissue from the PRN mouse model and its wild type validates the expression of c5-c7 markers.

The violin plots compare the expression of Ar and Postn in the stroma. f Quantification of 22rv1 overexpressing MYCN and with Rb1 knockdown migration in Boyden chamber transwell assay. g Comprehensive analysis of scRNA-seq data obtained from primary prostate fibroblasts co-cultured with T-ERG and PRN epithelial cells.

These findings prompted us to assess whether Postn -positive stroma facilitates invasion, a characteristic of NEPC. Knockdown of Periostin in fibroblasts induces an over 2-fold decrease of mobility in a migration assay in 22rv1 cells overexpressing MYCN with additional Rb1 knockdown to mimick the PRN model Fig.

These results suggest that specific epithelial mutations can shape the stromal microenvironment in PCa. To functionally validate these findings, we co-cultured normal fibroblasts from the FVBN mouse model with epithelial cells from the T-ERG and PRN models.

We found that fibroblasts co-cultured with epithelial cells from T-ERG model tend to exhibit similar expression profiles to those found in c3 and c4 stromal cells, while those co-cultured with PRN epithelium exhibit expression profiles of the c5-c7 stromal clusters Fig. The resulting PRN gene signature consists of 13 up- and down-regulated gene pairs from the PRN mesenchyme Supplementary Data file 5.

In addition to its interpretable decision rules, this signature has a robust and stable performance in both the training samples and testing samples sets with an Area Under the Receiver Operating Characteristic Curve AUROC of 0. Finally, the prognostic value of the signature was tested in the TCGA cohort which includes primary tumor samples from PCa patients 34 , Moreover, the PRN signature outperforms a cell cycle progression CCP signature when evaluated on the same testing set Supplementary Fig.

Overall, these results show that the PRN -derived mesenchymal cell clusters are associated with invasiveness, metastatic progression, and survival in PCa patients independent of Gleason grade. The signature was trained and validated on bulk expression profiles derived from primary tumor samples of PCa patients.

AUC: area under the ROC curve. The x-axis represents survival time in months. Using the mouse scRNA-seq data as reference, the eight stromal clusters were mapped to the human scRNA-seq data These includes six ERG-positive mesenchymal cells and three ERG-negative mesenchymal cells patients.

Notably, both c0 and c1 have transcriptional profiles similar to their murine counterparts, with c0 characterized by a high expression of myofibroblast marker genes ACTA2 and MYL9 , and c2 cells having a high expression of FOS and JUN Fig.

Notably, the transcriptional profiles of the GEMMs-specific stromal clusters are also preserved in the human data. For instance, stromal cells in c3-c4 have a high expression of genes involved in WNT signaling pathway including WNT4 and RORB Fig.

Nonetheless, these clusters still show transcriptional profiles similar to their mouse counterparts, with a high expression of POSTN and SFRP4 Fig.

a Parallel categories plot showing the relationship between the mesenchymal clusters and ERG status left. The width of the violins at different values represents the density of the data. The embedded box plots display the median of the data white dot , the bounds of the box represent the 25th and 75th percentiles interquartile range , and the data within these bounds represent the minima and maxima of the non-outlying data.

c UMAP of the selected cell types from the bone metastasis scRNA-seq data derived from Kfoury 66 left and their corresponding annotation using the eight mesenchymal clusters definition middle. Mesenchymal cells, which represent the predominant component of the microenvironment, have been suggested for decades to play a major role in this regard 37 , 38 , Recently, studies by Karthaus et al.

and Crowley et al. described a detailed cluster analysis of mesenchymal cells in the mouse prostate by scRNA-seq, revealing a level of complexity greater than that suggested previously 24 , Here, we analyzed in detail by scRNA-seq all mesenchymal cells utilizing all prostate lobes in the mouse prostate from several established GEMMs and corresponding WT mice.

The significant and progressive increase in the mesenchymal cell component in increasingly aggressive GEMM models suggests a pivotal role of the stroma in tumor progression.

We identified eight distinct stromal cell states that were defined by different gene expression programs and by underlying regulatory transcription factors. Three clusters represent fibroblast states that are common to all genotypes, and they display conserved functional programs across all stages of tumor growth.

Five stromal cell states on the other hand, are specifically linked to defined epithelial mutations and disease stages, pointing to mutation-specific epithelial to stromal signaling.

There is growing evidence that innate immunity and inflammation play a role in prostate and other cancers 39 , 40 , While the focus of this study was not on immune cells, we found a cluster of mesenchymal cells conserved across all genotypes in prostate mesenchyme expressing genes associated with immunoregulatory and inflammatory pathways and driven by transcription factors such as Nfkb.

Immune cells including tissue-resident macrophages are recruited and subsequently activated to modulate prostate tumorigenesis. In addition, stromal cells produce cytokines, chemokines, and components of complement protein pathways 42 , 43 , The complement system is an established component of innate immunity.

Components of complement activation via the C3 alternative pathway were previously found to be activated by KLK3 a. PSA , with a special affinity for iC3b that in turn stimulates inflammation In addition, a pronounced expression of Cd55 in common clusters, inhibits complement C3 lysis This suggests that the expansion of cells expressing C3 can stimulate innate immune response in the TME.

Model-specific variations in the composition of the tumor immune microenvironment were seen, e. Further functional analyses of those interactions will reveal how the stroma influences the response to immunotherapy in PCa 47 , 48 , Roughly half of prostate tumors have ETS translocations with TMPRSS2 as the most frequent fusion partner 35 , one of the earliest alterations in prostatic carcinogenesis 50 , 51 , Yet, genetically engineered mouse models driven by the TMPRSS2-ERG fusion display a minimal epithelial phenotype.

Here, we found that induction of mesenchymal cell expansion is a significant early event in this model. We harmonized the eight murine clusters with human PCa cases sequenced using the same scRNA-seq approach. Strikingly, the mesenchyme associated with the TMPRSS-ERG translocation was conserved between mouse and human.

Thus, epithelial ERG fusion in the mouse triggers early changes in the adjacent stroma, creating a TME that supports ERG-positive epithelial cells. Given the conservation of these mesenchymal clusters in humans, these findings shed a light on the role of this prevalent alteration in the pathogenesis of prostate cancer.

It will be important to determine the prevalence of these stromal cluster associated with TMPRSS-ERG in patients of African descent, where the prevalence of this translocation is low Stromal populations contribute to the structural and functional TME ecosystem through different autocrine and paracrine mechanisms.

Stromal AR signaling may prevent invasion by maintaining a non-permissive TME for cell migration Indeed, loss of stromal AR was associated with upregulation of ECM-remodeling metalloproteinases e.

In line with these observations, we show decreased mesenchymal Ar expression in the PRN model, which recapitulates late-stage PCa and progression toward neuroendocrine differentiation.

Stromal AR may play a master role in committing and maintaining epithelial prostate cell identity in at least two ways. Low expression of Ar in the PRN model was inversely associated with an increased expression of periostin Postn , and in situ analyses confirmed that Postn-positive cells were enriched in areas of neuroendocrine differentiation.

Stromal expression of periostin in PCa has been associated with decreased overall survival 58 , 59 and higher Gleason score The increased expression of Postn and of genes typical for the bone microenvironment e.

In this fashion, the primary site TME may pre-condition tumor cells for skeletal metastatic seeding. It is yet to be determined whether these cells were inherently present in the bone microenvironment or expanded as a result of the metastatic process.

These cells were also characterized by the expression of genes involved in osteoblast differentiation and proliferation like RUNX2, BMP2, IGF1 , and IGFBP3 61 together with Cadherin 11 CDH11 previously found to induce PCa invasiveness and bone metastasis 62 , The role of complement is important not only in both modulating innate immunity but also invasion.

A pronounced expression of C1QA , B and C was identified especially in models of advanced disease. C1q has been shown to promote trophoblast invasion 64 as well as angiogenesis in wound healing This was in line with our previously published stromal signature derived from laser capture-microdissected LCM mesenchyme adjacent to high-grade tumors that predicted lethality in an independent PCa cohort 8.

Three of the 24 signature genes were in fact C1Q A , B , and C suggesting that complement activation by the stroma plays a role in the invasive potential of aggressive prostate tumors with diverse epithelial genetic alterations. The unexpected resemblance between PCa mesenchyme of locally aggressive tumors and that of bone metastases suggests that locally advanced PCa tumors prone to metastasize display a bone-like microenvironment.

Importantly, while our findings offer valuable insights on the role of the stroma in mediating PCa progression and invasiveness, they also show a strong translational relevance. For instance, we have used the scRNA-seq transcriptional profiles of the PRN -derived mesenchymal clusters c5-c7 to develop a robust gene signature for predicting PCa metastases in a large cohort of patient samples with bulk transcriptomic profiles.

This signature was also associated with worse progression-free survival in a separate cohort TCGA before and after adjusting for Gleason grade. In contrast, a cell-cycle progression CCP signature 67 , did not have as robust a performance at predicting metastasis when tested on the same testing cohort.

Furthermore, the CCP signature was not significantly associated with progression-free survival after adjusting for Gleason grade. In summary, here we provide a molecular compendium of mesenchymal changes during PCa progression in genetically engineered mice that generalize to humans.

Specifically, in the early phases of prostate carcinogenesis, we provide evidence that the TMPRSS-ERG translocation reprograms the mesenchyme which in turn may sustain progression.

In advanced PCa models we found transcriptional mesenchymal programs linked to metastasis, some of them in common with the bone microenvironment to which PCa cells metastasize. Collectively, these data from both mice and humans present clear evidence of significant shifts in stromal composition that accompany PCa progression, which are influenced by genotype-specific factors.

These findings highlight the substantial role of mesenchymal changes as contributors to PCa progression and phenotypic diversity, emphasizing an impact that is more substantial than what has been detailed in existing literature.

In this study, only males were utilized. All animals used in this study received humane care in compliance with the principles stated in the Guide for the Care and Use of Laboratory Animals National Research Council, edition , and the protocol was approved by the Institutional Animal Care and Use Committee of Weill Cornell Medicine, Dana-Farber Cancer Institute and Columbia University Irving Medical Center.

We focused on three models of prostate cancer that reflect the most common mutations in human localized disease, plus a fourth model that recapitulates the transition to NEPC. The choice of these models was also taken to reflect different stages of the disease.

Specifically, the TMPRSS2-ERG T-ERG RRID:MGI fusion model has an N terminus-truncated human ERG together with an ires-GFP cassette into exon 2 of the mouse Tmprss2 locus 11 , 68 , displays a minimal epithelial phenotype in the mouse, and was chosen since it represents the most frequent mutation in human prostate cancers T-ERG mice together with their WT counterparts were euthanized and analyzed at the age of 6 months.

To obtain Nkx3. Six months later NP mice were sacrificed and analyzed. Hi-MYC 16 shows both PIN and microinvasion.

FVB Hi-MYC mice strain number 01XK8, RRID:MGI , expressing the human c-MYC transgene in prostatic epithelium, were obtained from the National Cancer Institute Mouse Repository at Frederick National Laboratory for Cancer Research. These mice were bred on the same mixed genetic background Charles River Laboratories Stock CRL : Hi-MYC mice together with their WT counterparts were euthanized and analyzed at the age of 6 months.

PRN mice recapitulates the transition to NEPC Prostate-specific Cre expression results in removal of LSL cassette by Cre and human N-Myc expression driven by the chicken actin promoter. Simultaneously, Cre recombinase converts the Pten and Rb1 floxed alleles to knockout alleles in the mouse prostate.

PRN mice together with their WT counterparts were euthanized and analyzed at the age of 8 weeks. These limits were not exceeded in any studies. Human prostate tissue specimens were obtained from patients undergoing radical prostatectomy at Weill Cornell Medicine under Institutional Review Board approval with informed consent WCM IRB , The study included a total of 13 subjects, all of whom were male.

No blinding, randomization, or exclusion criteria were applied. Of these, 9 samples comprising 3 ERG-negative and 6 ERG-positive cases were utilized for single-cell RNA sequencing studies, while 4 samples were employed for the mIHC Vectra Polaris staining. The clinical and molecular characteristics of these patients are provided in Supplementary Data file 6.

Immediately after surgical removal, the prostate was sectioned transversely through the apex, mid, and base A small portion of the regions of interest, including the areas selected for single-cell RNA sequencing, index lesion, and contralateral benign peripheral zone, was concomitantly frozen in optimal cutting temperature OCT compound, cryosectioned, and a rapid review was performed by a board-certified surgical pathologist BR to provide a preliminary assessment on the presence of tumor, normal epithelium, stroma near and away from the tumor.

Adjacent tissue was processed by formalin fixation and paraffin embedding, followed by sectioning, histological review, histochemistry trichrome stain , and immunostaining Dissociated murine prostate cells were prepared as described previously Dissociated cells were subsequently passed through 70 μm and 40 μm cell strainers BD Biosciences, San Jose, CA to obtain a single cells suspension.

Dissociated cells were subsequently passed through 70 μm and 40 μm cell strainers BD Biosciences, San Jose, CA to get single cells. Samples were resuspended in 1x PBS and sorted for DAPI to enrich living cells.

Expression matrices were generated from raw Illumina sequencing output using CellRanger. Bcl files were demultiplexed by bcl2fastq, then reads were aligned using the STAR aligner 74 with the default parameters.

All data collected from mouse models were aligned to GRCm38 reference transcriptome. To identify cells with trans-gene expression, we indexed and aligned to human ERG and GFP from the T-ERG model, human MYC from the Hi-MYC model, and human MYCN from the PRN model.

Human data were aligned to GRCh Alignment quality control was performed using the default CellRanger settings. Expression matrices from the different mouse models were converted to AnnData objects and concatenated into a single count matrix using the Scanpy library version 1.

Similarly, the expression matrices from the nine human samples were concatenated into a single count matrix.

Contributions from total count, mitochondrial count, and cell cycle were corrected by linear regression. The resulting matrix was then log1p transformed Finally, the top genes were selected based on the coefficient of variation according to the method described in ref.

We computed batch-corrected embeddings as follows. We fit our data using a conditional variational autoencoder Specifically, we used the negative binomial counts model included in the single-cell variational inference scVI Python package We model a a nuisance variable that represents differences in capture efficiency and sequencing depth and serves as a cell-specific scaling factor, and b an intermediate value that provides batch-corrected normalized estimates of the percentage of transcripts in each cell that originate from each gene.

Our model is implemented in Python using the PyTorch library v1. A nearest neighbor graph was constructed with Euclidean metric from the batch-corrected scVI embeddings, then cells were partitioned by the Leiden clustering algorithm 79 , Partition-Based Graph Abstraction PAGA was computed from the Leiden partition 81 and was used to initialize the Uniform Manifold Approximation and Projection UMAP algorithm which projected the data into 2D space For both the mouse , cells and human 83, cells scRNA-seq datasets, we excluded cells of lymphoid, endothelial, and neural origin based on Leiden clustering at resolution 1.

The resulting mesenchymal datasets for the mouse and human scRNA-seq data included and cells, respectively. These mesenchymal cells were then clustered using the Leiden algorithm to identify different mesenchymal sub-clusters.

Specifically, at resolution 0. We increased resolution in increments of 0. In the mouse scRNA-seq data, cells from the immune compartment 42, cells were also clustered using the Leiden algorithm.

The resulting clusters were then annotated to different immune cell types based on the expression of known markers genes. For differential expression DE testing, we used a two-part generalized linear model hurdle model , MAST, that parameterizes stochastic dropout and the characteristic bimodal distribution of single-cell transcriptomic data DE was performed by comparing the gene expression profiles of cells from each cluster to pooled cells from all other clusters Supplementary Data file 4.

Using the default parameters, the DE analysis was limited to genes which show on average at least 0. Gene regulatory network activity was inferred from the raw counts matrix by pySCENIC v0. Specifically, coexpression modules between transcription factors TFs and their candidate targets regulons were inferred using the Arboreto package GRNBoost2 and pruned for motif enrichment to separate indirect from direct targets 21 , The activity of each regulon in each cell was then scored using the Area Under the ROC curve AUC calculated by the AUCell module from pySCENIC package 21 , We performed ligand-receptor L-R interaction analysis using CellChatDB and CellChat R tool version 1.

Cell communication networks were inferred by identifying differentially expressed ligands and receptors between the different mesenchymal clusters, immune cell types, and the epithelium. Notably, we corrected for the effect of population size number of cells when calculating the interaction probabilities.

In addition, we summarized the L-R interaction probabilities within each signaling pathway to compute pathway-level communication probabilities as described in ref. Cell—cell communication networks were then aggregated by summing the number of interactions or by averaging the previously calculated communication probabilities.

To compare the signaling patterns between mutants and wild types, we performed differential expression analysis between all the mutants versus wild types in each of the three compartments stroma, epithelium, and immune.

Upregulated ligands and receptors were identified if each had a log-fold change logFC above 0. Finally, we extracted the mutant-specific L-R pairs as those with upregulated ligands and receptors in the mutants compared to wild types and vice versa.

In this analysis, we used a p -value threshold of 0. To transfer the stromal cluster labels from the mouse to human data, human gene symbols were converted to their mouse counterparts then both datasets were subset to overlapping genes.

Specifically, we used the scRNA-seq data from the mouse T-ERG model as reference for the human ERG-positive cases and those from the remaining mouse models as reference for the human ERG-negative cases.

Finally, we computed the ranking of differentially expressed genes in each cluster versus the remaining ones using t -test. The raw count matrix of the scRNA-seq dataset previously reported by Kfoury et al. This dataset included 25 bone metastasis samples derived from PCa patients, of which 9 samples were derived from solid metastasis tissue.

Further analysis was limited to these 9 samples 16, cells. Subsequently, cells were normalized by the total counts over all genes followed by log scaling and regressing over the total counts per cell and percentage of mitochondrial genes to reduce unwanted variation.

The top highly variable genes were selected and the resulting matrix was then scaled to unit variance and zero mean. Since this particular analysis was intended to explore the transcriptional and functional similarities between the primary tumor stroma and the stroma of bone metastasis, we further limited the analysis to the cells previously annotated by the authors as osteoblasts, osteoclasts, endothelial cells, and pericytes total cells.

Finally, the embeddings and stromal cluster labels were projected onto this dataset using the mouse stroma scRNA-seq dataset as reference and following the same steps mentioned above. We collected and curated gene expression profiles from different datasets comprising primary tumor samples from PCa patients with information about metastatic events.

The expression profiles from each dataset were normalized, log2-scaled, then z-score transformed by gene separately. Subsequently, we mapped probe IDs to their corresponding gene symbols and kept only the genes in common between all datasets 12, genes. Quantile normalization was applied to both the training and testing sets separately.

The training set was used for training a classifier that can predict metastasis using the k-top scoring pairs k-TSPs algorithm, which is a rank-based method whose predictions depend entirely on the ranking of gene pairs in each sample 98 , We then paired the top positive and negative markers genes each together to build a biological mechanism representing the PRN mesenchyme 30, pairs.

Each pair consists of two genes, one is up- and another is down-regulated in the PRN mesenchyme. This mechanism was then used as a priori biological constraint during the training of the k-TSPs algorithm , and the resulting signature was evaluated on the indepedent testing set.

In addition, we evaluated the prognostic relevance of this signature in the TCGA cohort which included primary PCa samples. First, we built a logistic regression model using the 26 genes comprising the PRN signature and used this model to generate a probability score for progression-free survival PFS in each patient.

We then binarized these probabilities into predicted classes and compared their PFS probability using Kaplan—Meier survival analysis Finally, we calculated the hazard ratio HR of the signature prediction probability scores after adjusting for Gleason grade using a multivariate Cox proportional hazards CPH model We retrieved a cell-cycle progression CCP signature consisting of 31 genes 67 and used it to develop a predictive model for metastasis.

Both the PRN and CCP signatures were trained and tested on the same training and testing sets described above, utilizing a logistic regression model to predict metastatic events.

The training and testing performance of both signatures was compared using the Area Under the Reciever Operating Characteristics Curve AUROC. Implementing the same approach used for the PRN signature, we tested the association of the CCP signature with PFS in the TCGA cohort using Kaplan—Meier survival analysis and a multivariate CPH model adjusting for Gleason grade.

In prostatectomy specimens where tumor was not definitively grossly visible, these areas were approximated by anatomic correlation of the MRI findings and targeted biopsies with the highest tumor grade as described in ref.

Prostate from WT and GEMM mice were dissected. One-half of the prostate from GEMMs was utilized for scRNA-seq see above. A HALO-based digital classifier was developed to identify collagen, epithelium, muscle fiber, and background regions on the digital images. Percentages of collagen deposition were then quantified and compared using unpaired t -test.

Immunohistochemical stainings were used to confirm the expression of the GEMMs proteins. Primary antibodies used for IHC staining were: Rabbit monoclonal Recombinant Anti-c-Myc antibody [Y69] Abcam ab; ; rabbit monoclonal PTEN D4.

Secondary antibodies used in IHC were the Poly-HRP IgG reagent from the BOND Polymer Refine Detection Kit cat DS, Leica Biosystems. Immunohistochemistry to interrogate for panel markers Supplementary Data file 7 was performed on 5-μm-thick formalin-fixed paraffin-embedded tissue FFPE of i human PCa and ii GEMMs sections using previously-established protocols , , Multiplexed immunohistochemistry mIHC was performed by staining 5-μm-thick FFPE core biopsy sections in a BondRX automated stainer, using published protocols , , Secondary antibodies used for immunofluorescence were: the anti-rabbit Akoya Rabbit HRP cat ARRKT, Akoya Biosciences and the anti-mouse Mouse Superboost cat B, Thermo Fisher Scientific.

The tyramide-conjugated fluorophores were Opal cat FPKT, Akoya Biosciences; 1;75 ; Opal cat FPKT, Akoya Biosciences; ; Opal cat FPKT, Akoya Biosciences; ; Opal cat FPKT, Akoya Biosciences; ; Opal cat FPKT, Akoya Biosciences, Opal dilution , TSA-DIG dilution All slides were also stained with DAPI for nuclear identification.

Whole slide images of hematoxylin and eosin, trichrome, and mIHC sections were acquired using the Vectra Polaris Automated Quantitative Pathology Imaging System Akoya Biosciences, Hopkinton, MA Images were processed by linear spectral unmixing and deconvolved Cells were segmented and a human-in-the-loop HALO random forest RF classifier was trained with labels from a pathologist to select stromal cells.

Subsequently, these stromal regions of the entire prostate surrounding glands in WT and GEMMs mice were preprocessed and analyzed using PathML v2.

To address technical artifacts in the segmentation results, DAPI-negative cells were filtered out. A nearest-neighbor graph was constructed from the counts matrix using Euclidean metric as implemented in the Scanpy package This graph was clustered using the Leiden algorithm 79 to identify subpopulations of cells and low-quality cells.

Cells were projected to two dimensions and visualized using the UMAP algorithm We conducted spatial transcriptomics analysis of murine PRN tumors and corresponding tissue from its wild type using 10X Genomics CytAssist Visium platform 10x Genomics, Pleasanton, CA.

Slides were then de-coverslipped and tissues were hematoxylin destained, decrosslinked and hybridized overnight with the whole mouse transcriptome panel which contains pairs of specific probes for each targeted gene PN Spatially barcoded libraries were generated and sequenced with paired-end dual-indexing 28 cycles Read 1, 10 cycles i7, 10 cycles i5, 90 cycles Read 2 Sequencing libraries were demultiplexed with bcl2fastq Illumina.

Spatial transcriptomics libraries were processed and aligned to the mm10 mouse reference genome using the Space Ranger software version 2.

The filtered UMI count matrices were merged to enable their joint analysis. Subsequently, the data was normalized each cell was normalized by total counts over all genes and log-scaled, then the top highly variable genes were identified by model.

The neighborhood graph was computed using the first 10 prinicipal components and 15 neighboring data points, then embedded using the UMAP algorithm. Spots were clustered using the Leiden algorithm with a resolution of 0. Spot annotation was performed using the clusters marker genes. Then the immortalization was performed using Retrovirus with zeocin resistance and expression of SV40 T antigen pBabe-Zeo-LT-ST.

Cells that had migrated through the filter were quantified 5 fields of view per filter and normalized to the negative control.

David S. Both human and murine cell lines routinely tested negative for the presence of mycoplasma, which was performed using a mycoplasma detection kit abm G For co-culture experiments, primary prostate fibroblasts were derived from week-old FVBN mice JAX.

Epithelial cells were derived from the T-ERG and PRN mouse models. Organoids cells were detached and single-cell suspended using TrypLE, pelleted and counted. The cells were left in co-culture, or as FVBN fibroblasts-only as controls, up to day 9 8 days of co-culture , changing medium in the wells every 2 days.

Next, control and epithelial-induced fibroblasts were collected and submitted to 10x single cell RNAseq. For differential gene expression testing, a two-part generalized linear model, known as the hurdle model, was utilized through the MAST framework.

Gene regulatory network activities were inferred from the raw counts matrix with the SCENIC pipeline. Clustering and data visualization were achieved using algorithms such as the Leiden clustering algorithm, Partition-Based Graph Abstraction PAGA , and Uniform Manifold Approximation and Projection UMAP.

For the development of the PRN signature to predict metastasis, stratified sampling was implemented to ensure balanced representation across the training and testing cohorts, and the k-top scoring pairs k-TSPs algorithm was used for classifier training.

Survival probabilities were estimated using the Kaplan—Meier method and evaluated by the Log-rank test. Multivariate survival analysis was performed using the Cox proportional hazards CPH model and was evaluated using the Wald test.

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. The single-cell RNA-seq and Visium Spatial Transcriptomics data generated in this study has been deposited in the Gene Expression Omnibus GEO under the accession codes: GSE , GSE , and GSE The gene expression publicly available data used in this study are available in GEO under the accession codes GSE 88 , GSE 89 , GSE 90 , GSE 91 , 92 , 93 , , GSE 94 , and GSE The microscopy data reported in this paper will be shared by the lead contact.

The remaining data are available within the Article, Supplementary Information or Source data file. Source data are provided with this paper. Rebello, R.

et al. Prostate cancer. Article PubMed Google Scholar. Robinson, D. Integrative clinical genomics of advanced prostate cancer. Cell , — Article CAS PubMed PubMed Central Google Scholar.

Shoag, J. Clinical variability and molecular heterogeneity in prostate cancer. Asian J. Pederzoli, F. Targetable gene fusions and aberrations in genitourinary oncology.

Article CAS PubMed Google Scholar. Dakhova, O. Genes upregulated in prostate cancer reactive stroma promote prostate cancer progression in vivo. Cancer Res. Valencia, T. Metabolic reprogramming of stromal fibroblasts through pmTORC1 signaling promotes inflammation and tumorigenesis.

Cancer Cell 26 , — Linares, J. ATF4-induced metabolic reprograming is a synthetic vulnerability of the pdeficient tumor stroma. Cell Metab. e Tyekucheva, S. Stromal and epithelial transcriptional map of initiation progression and metastatic potential of human prostate cancer.

Article ADS PubMed PubMed Central Google Scholar. Mo, F. Stromal gene expression is predictive for metastatic primary prostate cancer. Stromal cells in prostate cancer pathobiology: friends or foes?.

Baena, E. ETV1 directs androgen metabolism and confers aggressive prostate cancer in targeted mice and patients. Genes Dev. Wang, X. A luminal epithelial stem cell that is a cell of origin for prostate cancer.

Nature , — Article ADS CAS PubMed PubMed Central Google Scholar. Lesche, R. Genesis 32 , — Article PubMed PubMed Central Google Scholar. Zou, M.

Transdifferentiation as a mechanism of treatment resistance in a mouse model of castration-resistant prostate cancer. Cancer Discov. Ellwood-Yen, K. Myc-driven murine prostate cancer shares molecular features with human prostate tumors. Cancer Cell 4 , — Berger, A.

N-Myc-mediated epigenetic reprogramming drives lineage plasticity in advanced prostate cancer. Invest , — Brady, N. Temporal evolution of cellular heterogeneity during the progression to advanced AR-negative prostate cancer.

Article ADS MathSciNet CAS PubMed PubMed Central Google Scholar. Iwafuchi-Doi, M. Cell fate control by pioneer transcription factors.

Development , — Almalki, S. Key transcription factors in the differentiation of mesenchymal stem cells. Differentiation 92 , 41—51 Van de Sande, B. A scalable SCENIC workflow for single-cell gene regulatory network analysis.

Finak, G. MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data.

Genome Biol. Jassal, B. The reactome pathway knowledgebase. Nucleic Acids Res. CAS PubMed Google Scholar. Karthaus, W. Regenerative potential of prostate luminal cells revealed by single-cell analysis. Science , — Kwon, O. Functional heterogeneity of mouse prostate stromal cells revealed by single-cell RNA-Seq.

iScience 13 , — Crowley, L. A single-cell atlas of the mouse and human prostate reveals heterogeneity and conservation of epithelial progenitors. Elife 9 , e Jiao, H. Stem Cells 39 , 62—77 Chen, M.

Runx1 is required for the endothelial to haematopoietic cell transition but not thereafter. Kim, W. RUNX1 is essential for mesenchymal stem cell proliferation and myofibroblast differentiation.

Natl Acad. USA , — Suzuki, N. Combinatorial Gata2 and Sca1 expression defines hematopoietic stem cells in the bone marrow niche.

He, J. Lhx6 regulates canonical Wnt signaling to control the fate of mesenchymal progenitor cells during mouse molar root patterning. PLoS Genet 17 , e Pioli, P.

Snai2 and Snai3 transcriptionally regulate cellular fitness and functionality of T cell lineages through distinct gene programs. Immunobiology , — Ku, S.

Rb1 and Trp53 cooperate to suppress prostate cancer lineage plasticity, metastasis, and antiandrogen resistance. Science , 78—83 Liu, J. An integrated TCGA pan-cancer clinical data resource to drive high-quality survival outcome analytics.

Cell , — Cancer Genome Atlas Research Network. The Molecular taxonomy of primary prostate cancer. Wolf, F. SCANPY: large-scale single-cell gene expression data analysis.

Cunha, G. Role of the stromal microenvironment in carcinogenesis of the prostate. Int J. Cancer , 1—10 Global gene expression analysis of reactive stroma in prostate cancer.

Tuxhorn, J. Reactive stroma in prostate cancer progression. Salmon, H. Host tissue determinants of tumour immunity. Cancer 19 , — CAS PubMed PubMed Central Google Scholar. de Bono, J. Prostate carcinogenesis: inflammatory storms.

Cancer 20 , — Netti, G. Role of complement in regulating inflammation processes in renal and prostate cancers. Cells 10 , Condeelis, J. Macrophages: obligate partners for tumor cell migration, invasion, and metastasis. Zhou, Z.

Androgen-regulated stromal complement component 7 C7 suppresses prostate cancer growth. Oncogene 42 , — Manning, M. Proteolysis of complement factors iC3b and C5 by the serine protease prostate-specific antigen in prostatic fluid and seminal plasma. Reis, E. Complement in cancer: untangling an intricate relationship.

Stultz, J. How to turn up the heat on the cold immune microenvironment of metastatic prostate cancer. Prostate Cancer Prostatic Dis. Venturini, N. Immunotherapy for prostate cancer.

Cold Spring Harb. Ma, C. The prostate stromal transcriptome in aggressive and lethal prostate cancer. Sreenath, T. Oncogenic activation of ERG: a predominant mechanism in prostate cancer. Gasi Tandefelt, D. ETS fusion genes in prostate cancer. Cancer 21 , R—R Rostad, K.

ERG upregulation and related ETS transcription factors in prostate cancer. Jaratlerdsiri, W. African-specific molecular taxonomy of prostate cancer. Normal and abnormal development of the male urogenital tract.

Role of androgens, mesenchymal-epithelial interactions, and growth factors. Leach, D. Stromal androgen receptor regulates the composition of the microenvironment to influence prostate cancer outcome. Oncotarget 6 , — Cioni, B. Loss of androgen receptor signaling in prostate cancer-associated fibroblasts CAFs promotes CCL2- and CXCL8-mediated cancer cell migration.

Liu, Y. Stromal AR inhibits prostate tumor progression by restraining secretory luminal epithelial cells. Cell Rep. Nuzzo, P. Prognostic value of stromal and epithelial periostin expression in human prostate cancer: correlation with clinical pathological features and the risk of biochemical relapse or death.

BMC Cancer 12 , Sasaki, H. Expression of Periostin, homologous with an insect cell adhesion molecule, as a prognostic marker in non-small cell lung cancers. Jpn J. Tian, Y. Overexpression of periostin in stroma positively associated with aggressive prostate cancer.

Video