Innovation, conservation, and repurposing of gene function in root cell type development.

Plant species have evolved myriads of solutions, including complex cell type development and regulation, to adapt to dynamic environments. To understand this cellular diversity, we profiled tomato root cell type translatomes. Using xylem differentiation in tomato, examples of functional innovation, repurposing, and conservation of transcription factors are described, relative to the model plant Arabidopsis. Repurposing and innovation of genes are further observed within an exodermis regulatory network and illustrate its function. Comparative translatome analyses of rice, tomato, and Arabidopsis cell populations suggest increased expression conservation of root meristems compared with other homologous populations. In addition, the functions of constitutively expressed genes are more conserved than those of cell type/tissue-enriched genes. These observations suggest that higher order properties of cell type and pan-cell type regulation are evolutionarily conserved between plants and animals.

Superficial Neurocristic EWSR1::FLI1 Fusion Tumor: A Distinctive, Clinically Indolent, S100 Protein/SOX10-Positive Neoplasm.

It is now understood that identical gene fusions may be shared by different entities. We report a distinctive neoplasm of the skin and subcutis, harboring the Ewing sarcoma-associated EWSR1::FLI1 fusion but differing otherwise from Ewing sarcoma. Slides and blocks for 5 cutaneous neoplasms coded as other than Ewing sarcoma and harboring EWSR1::FLI1 were retrieved. Immunohistochemical and molecular genetic results were abstracted from reports. Methylation profiling was performed. Clinical information was obtained. The tumors occurred in 4 men and 1 woman (median: 25 years of age; range: 19-69 years) and involved the skin/subcutis of the back (2), thigh, buttock, and chest wall (median: 2.4 cm; range: 1-11 cm). Two tumors were present "years" before coming to clinical attention. The lesions were multinodular and circumscribed and consisted of nests of bland, round cells admixed with hyalinized collagenous bands containing spindled cells. Hemorrhage and cystic change were often present; necrosis was absent. All were diffusely S100 protein/SOX10-positive; 4 of 5 were CD99-negative. One tested case was strongly positive for NKX2.2. A variety of other tested markers were either focally positive (glial fibrillary acidic protein, p63) or negative. Molecular genetic results were as follows: EWSR1 exon 7::FLI1 exon 8, EWSR1 exon 11::FLI1 exon 5, EWSR1 exon 11::FLI1 exon 6, EWSR1 exon 7::FLI1 exon 6, and EWSR1 exon 10::FLI1 exon 6. Methylation profiling (3 cases) showed these to form a unique cluster, distinct from Ewing sarcoma. All patients underwent excision with negative margins; one received 1 cycle of chemotherapy. Clinical follow-up showed all patients to be alive without disease (median: 17 months; range: 11-62 months). Despite similar gene fusions, the morphologic, immunohistochemical, epigenetic, and clinical features of these unique EWSR1::FLI1-fused neoplasms of the skin and subcutis differ substantially from Ewing sarcoma. Interestingly, EWSR1 rearrangements involved exons 10 or 11, only rarely seen in Ewing sarcoma, in a majority of cases. Superficial neurocristic EWSR1::FLI1 fusion tumors should be rigorously distinguished from true cutaneous Ewing sarcomas.

Characterizing the Tumor Immune Microenvironment with Tyramide-Based Multiplex Immunofluorescence.

Multiplex immunofluorescence (mIF) allows simultaneous antibody-based detection of multiple markers with a nuclear counterstain on a single tissue section. Recent studies have demonstrated that mIF is becoming an important tool for immune profiling the tumor microenvironment, further advancing our understanding of the interplay between cancer and the immune system, and identifying predictive biomarkers of response to immunotherapy. Expediting mIF discoveries is leading to improved diagnostic panels, whereas it is important that mIF protocols be standardized to facilitate their transition into clinical use. Manual processing of sections for mIF is time consuming and a potential source of variability across numerous samples. To increase reproducibility and throughput we demonstrate the use of an automated slide stainer for mIF incorporating tyramide signal amplification (TSA). We describe two panels aimed at characterizing the tumor immune microenvironment. Panel 1 included CD3, CD20, CD117, FOXP3, Ki67, pancytokeratins (CK), and DAPI, and Panel 2 included CD3, CD8, CD68, PD-1, PD-L1, CK, and DAPI. Primary antibodies were first tested by standard immunohistochemistry and single-plex IF, then multiplex panels were developed and images were obtained using a Vectra 3.0 multispectral imaging system. Various methods for image analysis (identifying cell types, determining cell densities, characterizing cell-cell associations) are outlined. These mIF protocols will be invaluable tools for immune profiling the tumor microenvironment.

A Global Coexpression Network Approach for Connecting Genes to Specialized Metabolic Pathways in Plants.

Plants produce diverse specialized metabolites (SMs), but the genes responsible for their production and regulation remain largely unknown, hindering efforts to tap plant pharmacopeia. Given that genes comprising SM pathways exhibit environmentally dependent coregulation, we hypothesized that genes within a SM pathway would form tight associations (modules) with each other in coexpression networks, facilitating their identification. To evaluate this hypothesis, we used 10 global coexpression data sets, each a meta-analysis of hundreds to thousands of experiments, across eight plant species to identify hundreds of coexpressed gene modules per data set. In support of our hypothesis, 15.3 to 52.6% of modules contained two or more known SM biosynthetic genes, and module genes were enriched in SM functions. Moreover, modules recovered many experimentally validated SM pathways, including all six known to form biosynthetic gene clusters (BGCs). In contrast, bioinformatically predicted BGCs (i.e., those lacking an associated metabolite) were no more coexpressed than the null distribution for neighboring genes. These results suggest that most predicted plant BGCs are not genuine SM pathways and argue that BGCs are not a hallmark of plant specialized metabolism. We submit that global gene coexpression is a rich, largely untapped resource for discovering the genetic basis and architecture of plant natural products.

Folate Deficiency Inhibits Development of the Mammary Gland and its Associated Lymphatics in FVB Mice.

BACKGROUND: Folate is essential for DNA synthesis, DNA repair, cell proliferation, development, and morphogenesis. Folic acid (FA) is a nutritional supplement used to fortify human diets. OBJECTIVES: We investigated the effects of dietary FA on early mammary gland (MG) development and hyperplasia. METHODS: Study 1: nulliparous female FVB wild-type (WT) mice were fed control (Con; 2 mg FA/kg), deficient (Def; 0 mg FA/kg), excess (Ex; 5 mg FA/kg), or super excess (S-Ex; 20 mg FA/kg) diets for 8 wk before mating to WT or heterozygous FVB/N-Tg[mouse mammary tumor virus long terminal repeat (MMTV)-polyomavirus middle T antigen (PyVT)]634Mul/J (MMTV-PyMT+/-) transgenic males. Dams were fed these diets until they weaned WT or MMTV-PyMT+/- pups, which were fed the dam's diet from postnatal day (PND) 21 to 42. Tissues were collected from female progeny at PNDs 1, 21, and 42. Study 2: Con or Def diets were fed to WT intact females and males from PND 21 to 56, or to ovariectomized females from PND 21 to 77; tissues were collected at PND 56 or 77. Growth of all offspring, development of MGs, MG hyperplasia, supramammary lymph nodes, thymus and spleen, cell proliferation, and expression of MG growth factors were measured. RESULTS: Study 1: Ex or S-Ex did not affect postnatal MG development or hyperplasia. The rate of isometric MG growth (PND 1-21) was reduced by 69% in Def female progeny (P < 0.0001). Similarly, hyperplastic growth in MGs of Def MMTV-PyMT+/- offspring was 18% of Con (P < 0.05). The Def diet reduced supramammary lymph node size by 20% (P < 0.0001) and increased MG insulin-like growth factor 2 mRNA by 200% (P < 0.05) and protein by 130%-150% (P < 0.05). Study 2: the Def diet did not affect MG growth, but it did reduce supramammary lymph node size (P < 0.05), spleen weight (P < 0.001), and thymic medulla area (P < 0.05). CONCLUSIONS: In utero and postnatal folate deficiency reduced the isometric development of the MGs and early MG hyperplasia. Postnatal folate deficiency reduced the development of lymphatic tissues.

New Technologies to Image Tumors.

The premise of this book is the importance of the tumor microenvironment (TME). Until recently, most research on and clinical attention to cancer biology, diagnosis, and prognosis were focused on the malignant (or premalignant) cellular compartment that could be readily appreciated using standard morphology-based imaging.

A Syngeneic ErbB2 Mammary Cancer Model for Preclinical Immunotherapy Trials.

In order to develop a practical model of breast cancer, with in vitro and syngeneic, immune-intact, in vivo growth capacity, we established a primary cell line derived from a mammary carcinoma in the transgenic FVB/N-Tg(MMTV-ErbB2*)NDL2-5Mul mouse, referred to as "NDLUCD". The cell line is adapted to standard cell culture and can be transplanted into syngeneic FVB/N mice. The line maintains a stable phenotype over multiple in vitro passages and rounds of in vivo transplantation. NDLUCD tumors in FVB/N mice exhibit high expression of ErbB2 and ErbB3 and signaling molecules downstream of ErbB2. The syngeneic transplant tumors elicit an immune reaction in the adjacent stroma, detected and characterized using histology, immunophenotyping, and gene expression. NDLUCD cells also express PD-L1 in vivo and in vitro, and in vivo transplants are reactive to anti-immune checkpoint therapy with responses conducive to immunotherapy studies. This new NDLUCD cell line model is a practical alternative to the more commonly used 4T1 cells, and our previously described FVB/N-Tg(MMTV-PyVT)634Mul derived Met-1fvb2 and FVB/NTg(MMTV-PyVTY315F/Y322F) derived DB-7fvb2 cell lines. The NDLUCD cells have, so far, remained genetically and phenotypically stable over many generations, with consistent and reproducible results in immune intact preclinical cohorts.

Impact of pre-analytical variables on deep learning accuracy in histopathology.

AIMS: Machine learning (ML) binary classification in diagnostic histopathology is an area of intense investigation. Several assumptions, including training image quality/format and the number of training images required, appear to be similar in many studies irrespective of the paucity of supporting evidence. We empirically compared training image file type, training set size, and two common convolutional neural networks (CNNs) using transfer learning (ResNet50 and SqueezeNet). METHODS AND RESULTS: Thirty haematoxylin and eosin (H&E)-stained slides with carcinoma or normal tissue from three tissue types (breast, colon, and prostate) were photographed, generating 3000 partially overlapping images (1000 per tissue type). These lossless Portable Networks Graphics (PNGs) images were converted to lossy Joint Photographic Experts Group (JPG) images. Tissue type-specific binary classification ML models were developed by the use of all PNG or JPG images, and repeated with a subset of 500, 200, 100, 50, 30 and 10 images. Eleven models were generated for each tissue type, at each quantity of training images, for each file type, and for each CNN, resulting in 924 models. Internal accuracies and generalisation accuracies were compared. There was no meaningful significant difference in accuracies between PNG and JPG models. Models trained with more images did not invariably perform better. ResNet50 typically outperformed SqueezeNet. Models were generalisable within a tissue type but not across tissue types. CONCLUSIONS: Lossy JPG images were not inferior to lossless PNG images in our models. Large numbers of unique H&E-stained slides were not required for training optimal ML models. This reinforces the need for an evidence-based approach to best practices for histopathological ML.

Germline Mutations in PALB2, BRCA1, and RAD51C, Which Regulate DNA Recombination Repair, in Patients With Gastric Cancer.

Up to 10% of cases of gastric cancer are familial, but so far, only mutations in CDH1 have been associated with gastric cancer risk. To identify genetic variants that affect risk for gastric cancer, we collected blood samples from 28 patients with hereditary diffuse gastric cancer (HDGC) not associated with mutations in CDH1 and performed whole-exome sequence analysis. We then analyzed sequences of candidate genes in 333 independent HDGC and non-HDGC cases. We identified 11 cases with mutations in PALB2, BRCA1, or RAD51C genes, which regulate homologous DNA recombination. We found these mutations in 2 of 31 patients with HDGC (6.5%) and 9 of 331 patients with sporadic gastric cancer (2.8%). Most of these mutations had been previously associated with other types of tumors and partially co-segregated with gastric cancer in our study. Tumors that developed in patients with these mutations had a mutation signature associated with somatic homologous recombination deficiency. Our findings indicate that defects in homologous recombination increase risk for gastric cancer.

Pathobiology of the 129:Stat1 -/- mouse model of human age-related ER-positive breast cancer with an immune infiltrate-excluded phenotype.

BACKGROUND: Stat1 gene-targeted knockout mice (129S6/SvEvTac-Stat1 tm1Rds) develop estrogen receptor-positive (ER+), luminal-type mammary carcinomas at an advanced age. There is evidence for both host environment as well as tumor cell-intrinsic mechanisms to initiate tumorigenesis in this model. In this report, we summarize details of the systemic and mammary pathology at preneoplastic and tumor-bearing time points. In addition, we investigate tumor progression in the 129:Stat1 -/- host compared with wild-type 129/SvEv, and we describe the immune cell reaction to the tumors. METHODS: Mice housed and treated according to National Institutes of Health guidelines and Institutional Animal Care and Use Committee-approved methods were evaluated by histopathology, and their tissues were subjected to immunohistochemistry with computer-assisted quantitative image analysis. Tumor cell culture and conditioned media from cell culture were used to perform macrophage (RAW264.7) cell migration assays, including the 129:Stat1 -/--derived SSM2 cells as well as control Met1 and NDL tumor cells and EpH4 normal cells. RESULTS: Tumorigenesis in 129:Stat1 -/- originates from a population of FoxA1+ large oval pale cells that initially appear and accumulate along the mammary ducts in segments or regions of the gland prior to giving rise to mammary intraepithelial neoplasias. Progression to invasive carcinoma is accompanied by a marked local stromal and immune cell response composed predominantly of T cells and macrophages. In conditioned media experiments, cells derived from 129:Stat1 -/- tumors secrete both chemoattractant and chemoinhibitory factors, with greater attraction in the extracellular vesicular fraction and inhibition in the soluble fraction. The result appears to be recruitment of the immune reaction to the periphery of the tumor, with exclusion of immune cell infiltration into the tumor. CONCLUSIONS: 129:Stat1 -/- is a unique model for studying the critical origins and risk reduction strategies in age-related ER+ breast cancer. In addition, it can be used in preclinical trials of hormonal and targeted therapies as well as immunotherapies.

Cellular heterogeneity profiling by hyaluronan probes reveals an invasive but slow-growing breast tumor subset.

Tumor heterogeneity confounds cancer diagnosis and the outcome of therapy, necessitating analysis of tumor cell subsets within the tumor mass. Elevated expression of hyaluronan (HA) and HA receptors, receptor for HA-mediated motility (RHAMM)/HA-mediated motility receptor and cluster designation 44 (CD44), in breast tumors correlates with poor outcome. We hypothesized that a probe for detecting HA-HA receptor interactions may reveal breast cancer (BCa) cell heterogeneity relevant to tumor progression. A fluorescent HA (F-HA) probe containing a mixture of polymer sizes typical of tumor microenvironments (10-480 kDa), multiplexed profiling, and flow cytometry were used to monitor HA binding to BCa cell lines of different molecular subtypes. Formulae were developed to quantify binding heterogeneity and to measure invasion in vivo. Two subsets exhibiting differential binding (HA(-/low) vs. HA(high)) were isolated and characterized for morphology, growth, and invasion in culture and as xenografts in vivo. F-HA-binding amounts and degree of heterogeneity varied with BCa subtype, were highest in the malignant basal-like cell lines, and decreased upon reversion to a nonmalignant phenotype. Binding amounts correlated with CD44 and RHAMM displayed but binding heterogeneity appeared to arise from a differential ability of HA receptor-positive subpopulations to interact with F-HA. HA(high) subpopulations exhibited significantly higher local invasion and lung micrometastases but, unexpectedly, lower proliferation than either unsorted parental cells or the HA(-/low) subpopulation. Querying F-HA binding to aggressive tumor cells reveals a previously undetected form of heterogeneity that predicts invasive/metastatic behavior and that may aid both early identification of cancer patients susceptible to metastasis, and detection/therapy of invasive BCa subpopulations.

184AA3: a xenograft model of ER+ breast adenocarcinoma.

Despite the prevalence and significant morbidity resulting from estrogen receptor positive (ER(+)) breast adenocarcinomas, there are only a few models of this cancer subtype available for drug development and arguably none for studying etiology. Those models that do exist have questionable clinical relevance. Given our goal of developing luminal models, we focused on six cell lines derived by minimal mutagenesis from normal human breast cells, and asked if any could generate clinically relevant xenografts, which we then extensively characterized. Xenografts of one cell line, 184AA3, consistently formed ER(+) adenocarcinomas that had a high proliferative rate and other features consistent with "luminal B" intrinsic subtype. Squamous and spindle cell/mesenchymal differentiation was absent, in stark contrast to other cell lines that we examined or others have reported. We explored intratumoral heterogeneity produced by 184AA3 by immunophenotyping xenograft tumors and cultured cells, and characterized marker expression by immunofluorescence and flow cytometry. A CD44(High) subpopulation was discovered, yet their tumor forming ability was far less than CD44(Low) cells. Single cell cloning revealed the phenotypic plasticity of 184AA3, consistent with the intratumoral heterogeneity observed in xenografts. Characterization of ER expression in cultures revealed ER protein and signaling is intact, yet when estrogen was depleted in culture, and in vivo, it did not impact cell or tumor growth, analogous to therapeutically resistant ER(+) cancers. This model is appropriate for studies of the etiology of ovarian hormone independent adenocarcinomas, for identification of therapeutic targets, predictive testing, and drug development.

A novel intergenic ETnII-ß insertion mutation causes multiple malformations in polypodia mice.

Mouse early transposon insertions are responsible for ~10% of spontaneous mutant phenotypes. We previously reported the phenotypes and genetic mapping of Polypodia, (Ppd), a spontaneous, X-linked dominant mutation with profound effects on body plan morphogenesis. Our new data shows that mutant mice are not born in expected Mendelian ratios secondary to loss after E9.5. In addition, we refined the Ppd genetic interval and discovered a novel ETnII-ß early transposon insertion between the genes for Dusp9 and Pnck. The ETn inserted 1.6 kb downstream and antisense to Dusp9 and does not disrupt polyadenylation or splicing of either gene. Knock-in mice engineered to carry the ETn display Ppd characteristic ectopic caudal limb phenotypes, showing that the ETn insertion is the Ppd molecular lesion. Early transposons are actively expressed in the early blastocyst. To explore the consequences of the ETn on the genomic landscape at an early stage of development, we compared interval gene expression between wild-type and mutant ES cells. Mutant ES cell expression analysis revealed marked upregulation of Dusp9 mRNA and protein expression. Evaluation of the 5' LTR CpG methylation state in adult mice revealed no correlation with the occurrence or severity of Ppd phenotypes at birth. Thus, the broad range of phenotypes observed in this mutant is secondary to a novel intergenic ETn insertion whose effects include dysregulation of nearby interval gene expression at early stages of development.

Androgen receptor splice variant AR3 promotes prostate cancer via modulating expression of autocrine/paracrine factors.

Deregulation of androgen receptor (AR) splice variants has been implicated to play a role in prostate cancer development and progression. To understand their functions in prostate, we established a transgenic mouse model (AR3Tg) with targeted expression of the constitutively active and androgen-independent AR splice variant AR3 (a.k.a. AR-V7) in prostate epithelium. We found that overexpression of AR3 modulates expression of a number of tumor-promoting autocrine/paracrine growth factors (including Tgfß2 and Igf1) and expands prostatic progenitor cell population, leading to development of prostatic intraepithelial neoplasia. In addition, we showed that some epithelial-mesenchymal transition-associated genes are up-regulated in AR3Tg prostates, suggesting that AR3 may antagonize AR activity and halt the differentiation process driven by AR and androgen. This notion is supported by our observations that the number of Ck5(+)/Ck8(+) intermediate cells is increased in AR3Tg prostates after castration, and expression of AR3 transgene in these intermediate cells compromises prostate epithelium regeneration upon androgen replacement. Our results demonstrate that AR3 is a driver of prostate cancer, at least in part, through modulating multiple tumor-promoting autocrine/paracrine factors.

Immunohistochemistry and mass spectrometry for highly multiplexed cellular molecular imaging.

The role of immunohistochemistry (IHC) in the management of cancer has expanded to provide improved diagnostic classification, as well as guidance on disease prognosis, therapy, and relapse. These new tasks require evaluation of an increasing number of protein targets; however, conventional multiplexing, usually achieved using serial tissue sections stained for a single analyte per slide, can exhaust small biopsy specimens, complicate slide-to-slide protein expression correlation, and leave insufficient material for additional molecular assays. A new approach, mass spectrometry immunohistochemistry (MSIHC), compatible with high levels of target multiplexing and suitable for use on formalin-fixed, paraffin-embedded samples can circumvent many of these issues. The strategy employs antibodies that are labeled with elemental mass tags, such as isotopically pure lanthanides not typically found in biological specimens, rather than with typical fluorophores or chromogens. The metal-labeled antibodies are then detected in tissue using lasers or ion beams to liberate the tags for subsequent mass spectrometry detection. Within a given multiplexed IHC panel, the metal labels are selected so that their respective masses do not overlap. More than 30 antibodies have been imaged simultaneously, and up to 100 antibodies could potentially be detected at once if the full available mass spectrum is deployed. MSIHC has a number of advantages over conventional IHC techniques. Background due to autofluorescence is absent and the dynamic range is 10(5), exceeding immunofluorescence and chromogenic IHC by 100-fold and 1000-fold, respectively. Detection of labeled primary antibodies improves assay linearity over both chromogenic and fluorescent IHC. Multiplexed mass-tagged antibodies incubated simultaneously with tissue do not appear to cross-interfere, and because the mass tags do not degrade, samples are stable indefinitely. The imaging resolution of multiplexed ion-beam imaging can be better than light microscopy. With appropriate instrumentation, MSIHC has the potential to transform research and clinical pathology practice.

Mouse models of prostate cancer: picking the best model for the question.

When the National Institutes of Health Mouse Models of Human Cancer Consortium initiated the Prostate Steering Committee 15 years ago, there were no genetically engineered mouse (GEM) models of prostate cancer (PCa). Today, a PubMed search for "prostate cancer mouse model" yields 3,200 publications and this list continues to grow. The first generation of GEM utilized the newly discovered and characterized probasin promoter driving viral oncogenes such as Simian virus 40 large T antigen to yield the LADY and TRAMP models. As the PCa research field has matured, the second generation of models has incorporated the single and multiple molecular changes observed in human disease, such as loss of PTEN and overexpression of Myc. Application of these models has revealed that mice are particularly resistant to developing invasive PCa, and once they achieve invasive disease, the PCa rarely resembles human disease. Nevertheless, these models and their application have provided vital information on human PCa progression. The aim of this review is to provide a brief primer on mouse and human prostate histology and pathology, provide descriptions of mouse models, as well as attempt to answer the age old question: Which GEM model of PCa is the best for my research question?

Introduction of Zinc-salt Fixation for Effective Detection of Immune Cell-related Markers by Immunohistochemistry.

Tissue localization of immune cells is critical to the study of disease processes in mouse models of human diseases. However, immunohistochemistry (IHC) for immune cell phenotyping in mouse tissue sections presents specific technical challenges. For example, CD4 and CD8 have been difficult to detect using IHC on formalin-fixed and paraffin-embedded mouse tissue, prompting alternative methods. We investigated the use of formalin-free zinc-salt fixation (ZN) and optimized IHC protocols for detecting a panel of immune cell-related markers (CD3, CD4, CD8, Foxp3, B220, F4/80, CD68, and major histocompatibility complex [MHC] class-I, MHC class-II, and Gr-1). The IHC results for these markers were compared on mouse spleen tissue treated with neutral buffered formalin (NBF) or ZN with or ZN without antigen retrieval (AR). Whereas CD4 and CD8 were not detected in NBF-treated tissue, all markers were detected in ZN-treated tissue without AR. Thus, the use of ZN treatment for IHC staining can be a good tool for studying immunoreactive lesions in tissues.

Stacked Predictive Sparse Decomposition for Classification of Histology Sections.

Image-based classification of histology sections, in terms of distinct components (e.g., tumor, stroma, normal), provides a series of indices for histology composition (e.g., the percentage of each distinct components in histology sections), and enables the study of nuclear properties within each component. Furthermore, the study of these indices, constructed from each whole slide image in a large cohort, has the potential to provide predictive models of clinical outcome. For example, correlations can be established between the constructed indices and the patients' survival information at cohort level, which is a fundamental step towards personalized medicine. However, performance of the existing techniques is hindered as a result of large technical variations (e.g., variations of color/textures in tissue images due to non-standard experimental protocols) and biological heterogeneities (e.g., cell type, cell state) that are always present in a large cohort. We propose a system that automatically learns a series of dictionary elements for representing the underlying spatial distribution using stacked predictive sparse decomposition. The learned representation is then fed into the spatial pyramid matching framework with a linear support vector machine classifier. The system has been evaluated for classification of distinct histological components for two cohorts of tumor types. Throughput has been increased by using of graphical processing unit (GPU), and evaluation indicates a superior performance results, compared with previous research.