Intratumoral phenotypic heterogeneity has been described in many tumor types, where it can contribute to drug resistance and disease recurrence. We analyzed ductal and neuroendocrine markers in pancreatic ductal adenocarcinoma, revealing heterogeneous expression of the neuroendocrine marker Synaptophysin within ductal lesions. Higher percentages of Cytokeratin-Synaptophysin dual positive tumor cells correlate with shortened disease-free survival. We observe similar lineage marker heterogeneity in mouse models of pancreatic ductal adenocarcinoma, where lineage tracing indicates that Cytokeratin-Synaptophysin dual positive cells arise from the exocrine compartment. Mechanistically, MYC binding is enriched at neuroendocrine genes in mouse tumor cells and loss of MYC reduces ductal-neuroendocrine lineage heterogeneity, while deregulated MYC expression in KRAS mutant mice increases this phenotype. Neuroendocrine marker expression is associated with chemoresistance and reducing MYC levels decreases gemcitabine-induced neuroendocrine marker expression and increases chemosensitivity. Altogether, we demonstrate that MYC facilitates ductal-neuroendocrine lineage plasticity in pancreatic ductal adenocarcinoma, contributing to poor survival and chemoresistance.

Nearly all of the open reading frames (ORFs) of the yeast Saccharomyces cerevisiae have been synthesized by PCR using a set of ∼6000 primer pairs. Each of the forward primers has a common 22-base sequence at its 5Ј end, and each of the back primers has a common 20-base sequence at its 5Ј end. These common termini allow reamplification of the entire set of original PCR products using a single pair of longer primers-in our case, 70 bases. The resulting 70-base elements that flank each ORF can be used for rapid and efficient cloning into a linearized yeast vector that contains these same elements at its termini. This cloning by genetic recombination obviates the need for ligations or bacterial manipulations and should permit convenient global approaches to gene function that require the assay of each putative yeast gene.Knowledge of the complete genome sequence of the yeast Saccharomyces cerevesiae is enabling global approaches for the analysis of gene function (see e.g., Oliver 1996; Johnston 1996). In particular, it permits experiments in which each gene is systematically analyzed for activity in contrast to those that rely on random screens. Such comprehensive efforts require efficient strategies to handle the ∼6000 open reading frames (ORFs) predicted from the sequence. We describe a simple PCR-based approach that has generated a nearly complete set of the yeast genes in a form that allows multiple uses, including the construction of DNA arrays, epitope-tagged or hybrid proteins, and regulated versions of the genes. As an example, we demonstrate how these reagents are being used to produce fusions of the Gal4p activation domain to each yeast protein for large-scale two-hybrid analysis (see Finley and Brent 1994;Bartel et al. 1996). RESULTS AND DISCUSSIONThe strategy outlined in Figure 1 uses a set of ∼6000 PCR primer pairs to amplify individually each of the yeast ORFs. The primers were designed from a list of the first 50 bases (corresponding to the predicted ATG and succeeding 3Ј sequences) and last 50 bases (corresponding to the sequences ending with and including the predicted termination codon) for 6102 ORFs in the Saccharomyces Genome Database, kindly provided by Michael Cherry (Stanford University, Palo Alto, CA). Each forward primer contains both a unique sequence that allows priming at the start of one of the yeast ORFs and a 22-nucleotide-long sequence at its 5Ј end, which is shared by all forward primers (see Fig. 1). The unique sequence begins with the codon after the initiator ATG and is followed by an additional 17-29 bases of ORF sequence to allow annealing of each primer to its target sequence with a uniform T m of 68°C-72°C. Each back primer also contains both a unique sequence, which allows priming at the terminus of an ORF, and a 20-nucleotide-long tail at its 5Ј end, which is shared by all the back primers. The unique sequence of the back primer corresponds to the reverse complement of the termination codon followed by 17-29 bases that are the reverse complement of the last ∼6-10 codons of t...

Nonalcoholic fatty liver disease (NAFLD) occurs frequently in a setting of obesity, dyslipidemia and insulin resistance, but the etiology of the disease, particularly the events favoring progression to nonalcoholic steatohepatitis (NASH) as opposed to simple steatosis (SS), are not fully understood. Based on known zonation patterns in protein, glucose and lipid metabolism, coupled with evidence that phosphatidylcholine may play a role in NASH pathogenesis, we hypothesized that phospholipid zonation exists in liver and that specific phospholipid abundance and distribution may be associated with histologic disease. A survey of normal hepatic protein expression profiles in the Human Protein Atlas revealed pronounced zonation of enzymes involved in lipid utilization and storage, particularly those facilitating phosphatidylcholine (PC) metabolism. Immunohistochemistry of obese normal, SS and NASH liver specimens with anti-phosphatidylethanomine N-methyltransferase (PEMT) antibodies showed a progressive decrease in the zonal distribution of this PC biosynthetic enzyme. Phospholipid quantitation by liquid chromatography mass spectrometry (LC-MS) in hepatic extracts of Class III obese patients with increasing NAFLD severity revealed that most PC species with 32, 34 and 36 carbons as well as total PC abundance was decreased with SS and NASH. Matrix assisted laser desorption ionization - imaging mass spectrometry (MALDI-IMS) imaging revealed strong zonal distributions for 32, 34 and 36 carbon PCs in controls (minimal histologic findings) and SS that was lost in NASH specimens. Specific lipid species such as PC 34∶1 and PC 36∶2 best illustrated this phenomenon. These findings suggest that phospholipid zonation may be associated with the presence of an intrahepatic proinflammatory phenotype and thus have broad implications in the etiopathogenesis of NASH.

No abstract

Spatially-resolved molecular profiling by immunostaining tissue sections is a key feature in cancer diagnosis, subtyping, and treatment, where it complements routine histopathological evaluation by clarifying tumor phenotypes. In this work, we present a deep learning-based method called speedy histological-to-immunofluorescent translation (SHIFT) which takes histologic images of hematoxylin and eosin (H&E)-stained tissue as input, then in near-real time returns inferred virtual immunofluorescence (IF) images that estimate the underlying distribution of the tumor cell marker pan-cytokeratin (panCK). To build a dataset suitable for learning this task, we developed a serial staining protocol which allows IF and H&E images from the same tissue to be spatially registered. We show that deep learning-extracted morphological feature representations of histological images can guide representative sample selection, which improved SHIFT generalizability in a small but heterogenous set of human pancreatic cancer samples. With validation in larger cohorts, SHIFT could serve as an efficient preliminary, auxiliary, or substitute for panCK IF by delivering virtual panCK IF images for a fraction of the cost and in a fraction of the time required by traditional IF.