mRNA-Seq of single prostate cancer circulating tumor cells reveals recapitulation of gene expression and pathways found in prostate cancer.

Circulating tumor cells (CTC) mediate metastatic spread of many solid tumors and enumeration of CTCs is currently used as a prognostic indicator of survival in metastatic prostate cancer patients. Some evidence suggests that it is possible to derive additional information about tumors from expression analysis of CTCs, but the technical difficulty of isolating and analyzing individual CTCs has limited progress in this area. To assess the ability of a new generation of MagSweeper to isolate intact CTCs for downstream analysis, we performed mRNA-Seq on single CTCs isolated from the blood of patients with metastatic prostate cancer and on single prostate cancer cell line LNCaP cells spiked into the blood of healthy donors. We found that the MagSweeper effectively isolated CTCs with a capture efficiency that matched the CellSearch platform. However, unlike CellSearch, the MagSweeper facilitates isolation of individual live CTCs without contaminating leukocytes. Importantly, mRNA-Seq analysis showed that the MagSweeper isolation process did not have a discernible impact on the transcriptional profile of single LNCaPs isolated from spiked human blood, suggesting that any perturbations caused by the MagSweeper process on the transcriptional signature of isolated cells are modest. Although the RNA from patient CTCs showed signs of significant degradation, consistent with reports of short half-lives and apoptosis amongst CTCs, transcriptional signatures of prostate tissue and of cancer were readily detectable with single CTC mRNA-Seq. These results demonstrate that the MagSweeper provides access to intact CTCs and that these CTCs can potentially supply clinically relevant information.

LC3-mediated fibronectin mRNA translation induces fibrosarcoma growth by increasing connective tissue growth factor.

Previously, we related fibronectin (Fn1) mRNA translation to an interaction between an AU-rich element in the Fn1 3' UTR and light chain 3 (LC3) of microtubule-associated proteins 1A and 1B. Since human fibrosarcoma (HT1080) cells produce little fibronectin and LC3, we used these cells to investigate how LC3-mediated Fn1 mRNA translation might alter tumor growth. Transfection of HT1080 cells with LC3 enhanced fibronectin mRNA translation. Using polysome analysis and RNA-binding assays, we show that elevated levels of translation depend on an interaction between a triple arginine motif in LC3 and the AU-rich element in Fn1 mRNA. Wild-type but not mutant LC3 accelerated HT1080 cell growth in culture and when implanted in SCID mice. Comparison of WT LC3 with vector-transfected HT1080 cells revealed increased fibronectin-dependent proliferation, adhesion and invasion. Microarray analysis of genes differentially expressed in WT and vector-transfected control cells indicated enhanced expression of connective tissue growth factor (CTGF). Using siRNA, we show that enhanced expression of CTGF is fibronectin dependent and that LC3-mediated adhesion, invasion and proliferation are CTGF dependent. Expression profiling of soft tissue tumors revealed increased expression of both LC3 and CTGF in some locally invasive tumor types.

Developmental expression of LC3alpha and beta: absence of fibronectin or autophagy phenotype in LC3beta knockout mice.

Murine light chain 3 (LC3) exists as two isoforms, LC3alpha and beta: LC3beta is an RNA-binding protein that enhances fibronectin (FN) mRNA translation, and is also a marker of autophagy. We report embryonic expression patterns for LC3alpha and LC3beta, with some overlap but notable differences in the brain, and in tissues of non-neuronal origin. LC3beta knockout (-/-) mice develop normally without a compensatory increase in LC3alpha. LC3beta-/- embryonic fibroblasts (MEFs) exhibit reduced FN synthesis but maintain wild type (WT) levels of FN protein. No significant changes in proteins associated with FN turnover, i.e., caveolin-1, LRP-1, or matrix metalloproteinases were identified. Autophagosomes form in amino acid-starved LC3beta-/-MEFs, and Caesarean-delivered pups survive as long as WT pups without an increase in LC3-related proteins linked to autophagy. These results suggest novel compensatory mechanisms for loss of LC3beta, ensuring proper FN accumulation and autophagy during fetal and neonatal life.

Increased fibulin-5 and elastin in S100A4/Mts1 mice with pulmonary hypertension.

Transgenic mice overexpressing the calcium binding protein, S100A4/Mts1, occasionally develop severe pulmonary vascular obstructive disease. To understand what underlies this propensity, we compared the pulmonary vascular hemodynamic and structural features of S100A4/Mts1 with control C57Bl/6 mice at baseline, following a 2-week exposure to chronic hypoxia, and after 1 and 3 months "recovery" in room air. S100A4/Mts1 mice had greater right ventricular systolic pressure and right ventricular hypertrophy at baseline, which increased further with chronic hypoxia and was sustained after 3 months "recovery" in room air. These findings correlated with a heightened response to acute hypoxia and failure to vasodilate with nitric oxide or oxygen. S100A4/Mts1 mice, when compared with C57Bl/6 mice, also had impaired cardiac function judged by reduced ventricular elastance and decreased cardiac output. Despite higher right ventricular systolic pressures with chronic hypoxia, S100A4/Mts1 mice did not develop more severe PVD, but in contrast to C57Bl/6 mice, these features did not regress on return to room air. Microarray analysis of lung tissue identified a number of genes differentially upregulated in S100A4/Mts1 versus control mice. One of these, fibulin-5, is a matrix component necessary for normal elastin fiber assembly. Fibulin-5 was localized to pulmonary arteries and associated with thickened elastic laminae. This feature could underlie attenuation of pulmonary vascular changes in response to elevated pressure, as well as impaired reversibility.

Integrin alpha4beta1 (VLA-4) expression and activity in retinal and peripheral neurons.

The integrin alpha4beta1 fulfills important roles in inflammation and hematopoesis, but its functions in neurons are not well understood. Here we show that the alpha4 subunit is expressed on mouse retinal ganglion cells (RGCs) and undifferentiated retinal neuroblasts during the period of axon extension and migration. To determine if alpha4 integrins expressed by retinal neurons were active, neurons were cultured on known alpha4 ligands in vitro. Recombinant soluble vascular cell adhesion molecule 1 (rsVCAM-1), fibronectin, and osteopontin (OPN) induced neurite outgrowth that was diminished by function blocking antibodies specific for alpha4. Neurite outgrowth on OPN was also blocked by antibodies to the integrin beta1 subunit, implicating the alpha4beta1 heterodimer as one integrin receptor mediating outgrowth on OPN. OPN immunoreactivity was detected in the RGC fiber layer and optic nerve, suggesting that it may act as an alpha4 ligand in vivo. Neurons from chick lumbar sympathetic ganglia, chick dorsal root ganglia, and mouse superior cervical ganglia also extended neurites on rsVCAM-1, suggesting that integrin alpha4beta1 may play a role in the development of multiple neuronal cell types.

Regulation of myosin expression during myotome formation.

The first skeletal muscle fibers to form in vertebrate embryos appear in the somitic myotome. PCR analysis and in situ hybridization with isoform-specific probes reveal differences in the temporal appearance and spatial distribution of fast and slow myosin heavy chain mRNA transcripts within myotomal fibers. Embryonic fast myosin heavy chain was the first isoform expressed, followed rapidly by slow myosin heavy chains 1 and 3, with slow myosin heavy chain 2 appearing several hours later. Neonatal fast myosin heavy chain is not expressed in myotomal fibers. Although transcripts of embryonic fast myosin heavy chain were always distributed throughout the length of myotomal fibers, the mRNA for each slow myosin heavy chain isoform was initially restricted to the centrally located myotomal fiber nuclei. As development proceeded, slow myosin heavy chain transcripts spread throughout the length of myotomal fibers in order of their appearance. Explants of segments from embryos containing neural tube, notochord and somites 7-10, when incubated overnight, become innervated by motor neurons from the neural tube and express all four myosin heavy chain genes. Removal of the neural tube and/or notochord from explants prior to incubation or addition of d-tubocurare to intact explants prevented expression of slow myosin chain 2 but expression of genes encoding the other myosin heavy chain isoforms was unaffected. Thus, expression of slow myosin heavy chain 2 is dependent on functional innervation, whereas expression of embryonic fast and slow myosin heavy chain 1 and 3 are innervation independent. Implantation of sonic-hedgehog-soaked beads in vivo increased the accumulation of both fast and slow myosin heavy chain transcripts, as well as overall myotome size and individual fiber size, but had no effect on myotomal fiber phenotype. Transcripts encoding embryonic fast myosin heavy chain first appear ventrolaterally in the myotome, whereas slow myosin heavy chain transcripts first appear in fibers positioned midway between the ventrolateral and dorsomedial lips of the myotome. Therefore, models of epaxial myotome formation must account for the positioning of the oldest fibers in the more ventral-lateral region of the myotome and the youngest fibers in the dorsomedial region.

Ventral axial organs regulate expression of myotomal Fgf-8 that influences rib development.

Fgf-8 encodes a secreted signaling molecule mediating key roles in embryonic patterning. This study analyzes the expression pattern, regulation, and function of this growth factor in the paraxial mesoderm of the avian embryo. In the mature somite, expression of Fgf-8 is restricted to a subpopulation of myotome cells, comprising most, but not all, epaxial and hypaxial muscle precursors. Following ablation of the notochord and floor plate, Fgf-8 expression is not activated in the somites, in either the epaxial or the hypaxial domain, while ablation of the dorsal neural tube does not affect Fgf-8 expression in paraxial mesoderm. Contrary to the view that hypaxial muscle precursors are independent of regulatory influences from axial structures, these findings provide the first evidence for a regulatory influence of ventral, but not dorsal axial structures on the hypaxial muscle domain. Sonic hedgehog can substitute for the ventral neural tube and notochord in the initiation of Fgf-8 expression in the myotome. It is also shown that Fgf-8 protein leads to an increase in sclerotomal cell proliferation and enhances rib cartilage development in mature somites, whereas inhibition of Fgf signaling by SU 5402 causes deletions in developing ribs. These observations demonstrate: (1) a regulatory influence of the ventral axial organs on the hypaxial muscle compartment; (2) regulation of epaxial and hypaxial expression of Fgf-8 by Sonic hedgehog; and (3) independent regulation of Fgf-8 and MyoD in the hypaxial myotome by ventral axial organs. It is postulated that the notochord and ventral neural tube influence hypaxial expression of Fgf-8 in the myotome and that, in turn, Fgf-8 has a functional role in rib formation.