Zhandong Liu

A drosophila genetic resource of mutants to study mechanisms underlying human genetic diseases.

Invertebrate model systems are powerful tools for studying human disease owing to their genetic tractability and ease of screening. We conducted a mosaic genetic screen of lethal mutations on the Drosophila X chromosome to identify genes required for the development, function, and maintenance of the nervous system. We identified 165 genes, most of whose function has not been studied in vivo. In parallel, we investigated rare variant alleles in 1,929 human exomes from families with unsolved Mendelian disease. Genes that are essential in flies and have multiple human homologs were found to be likely to be associated with human diseases. Merging the human data sets with the fly genes allowed us to identify disease-associated mutations in six families and to provide insights into microcephaly associated with brain dysgenesis. This bidirectional synergism between fly genetics and human genomics facilitates the functional annotation of evolutionarily conserved genes involved in human health.

Gene expression profiles predict survival and progression of pleural mesothelioma

Purpose: Clinical outcomes for malignant pleural mesothelioma (MPM) patients having surgery are imprecisely predicted by histopathology and intraoperative staging. We hypothesized that gene expression profiles could predict time to progression and survival in surgically cytoreduced pleural mesothelioma of all stages. Experimental Design: Gene expression analyses from 21 MPM patients having cytoreductions and identical postoperative adjuvant therapy were performed using the U95 Affymetrix gene chip. Using both dChip and SAM, neural networks constructed a common 27 gene classifier, which was associated with either the high-risk and low-risk group of patients. Data were validated using real-time PCR and immunohistochemical staining. The 27 gene classifier was also used for validation in a separate set of 17 MPM patients from another institution. Results: The groups predicted by the gene classifier recapitulated the actual time to progression and survival of the test set with 95.2% accuracy using 10-fold cross-validation. Clinical outcomes were independent of histology, and heterogeneity of progression and survival in early stage patients was defined by the classifier. The gene classifier had a 76% accuracy in the separate validation set of MPMs. Conclusions: These data suggest that pretherapy gene expression analysis of mesothelioma biopsies may predict which patients may benefit from a surgical approach.

Serine Catabolism Regulates Mitochondrial Redox Control during Hypoxia

UNLABELLED The de novo synthesis of the nonessential amino acid serine is often upregulated in cancer. In this study, we demonstrate that the serine catabolic enzyme, mitochondrial serine hydroxymethyltransferase (SHMT2), is induced when MYC-transformed cells are subjected to hypoxia. In mitochondria, SHMT2 can initiate the degradation of serine to CO2 and NH4+, resulting in net production of NADPH from NADP+. Knockdown of SHMT2 in MYC-dependent cells reduced cellular NADPH:NADP+ ratio, increased cellular reactive oxygen species, and triggered hypoxia-induced cell death. In vivo, SHMT2 suppression led to impaired tumor growth. In MYC-amplified neuroblastoma patient samples, there was a significant correlation between SHMT2 and hypoxia-inducible factor-1 α (HIF1α), and SHMT2 expression correlated with unfavorable patient prognosis. Together, these data demonstrate that mitochondrial serine catabolism supports tumor growth by maintaining mitochondrial redox balance and cell survival. SIGNIFICANCE In this study, we demonstrate that the mitochondrial enzyme SHMT2 is induced upon hypoxic stress and is critical for maintaining NADPH production and redox balance to support tumor cell survival and growth.

Reversal of phenotypes in MECP2 duplication mice using genetic rescue or antisense oligonucleotides

Copy number variations have been frequently associated with developmental delay, intellectual disability and autism spectrum disorders. MECP2 duplication syndrome is one of the most common genomic rearrangements in males and is characterized by autism, intellectual disability, motor dysfunction, anxiety, epilepsy, recurrent respiratory tract infections and early death. The broad range of deficits caused by methyl-CpG-binding protein 2 (MeCP2) overexpression poses a daunting challenge to traditional biochemical-pathway-based therapeutic approaches. Accordingly, we sought strategies that directly target MeCP2 and are amenable to translation into clinical therapy. The first question that we addressed was whether the neurological dysfunction is reversible after symptoms set in. Reversal of phenotypes in adult symptomatic mice has been demonstrated in some models of monogenic loss-of-function neurological disorders, including loss of MeCP2 in Rett syndrome, indicating that, at least in some cases, the neuroanatomy may remain sufficiently intact so that correction of the molecular dysfunction underlying these disorders can restore healthy physiology. Given the absence of neurodegeneration in MECP2 duplication syndrome, we propose that restoration of normal MeCP2 levels in MECP2 duplication adult mice would rescue their phenotype. By generating and characterizing a conditional Mecp2-overexpressing mouse model, here we show that correction of MeCP2 levels largely reverses the behavioural, molecular and electrophysiological deficits. We also reduced MeCP2 using an antisense oligonucleotide strategy, which has greater translational potential. Antisense oligonucleotides are small, modified nucleic acids that can selectively hybridize with messenger RNA transcribed from a target gene and silence it, and have been successfully used to correct deficits in different mouse models. We find that antisense oligonucleotide treatment induces a broad phenotypic rescue in adult symptomatic transgenic MECP2 duplication mice (MECP2-TG), and corrected MECP2 levels in lymphoblastoid cells from MECP2 duplication patients in a dose-dependent manner.

Digital sorting of complex tissues for cell type-specific gene expression profiles

BackgroundCellular heterogeneity is present in almost all gene expression profiles. However, transcriptome analysis of tissue specimens often ignores the cellular heterogeneity present in these samples. Standard deconvolution algorithms require prior knowledge of the cell type frequencies within a tissue or their in vitro expression profiles. Furthermore, these algorithms tend to report biased estimations.ResultsHere, we describe a Digital Sorting Algorithm (DSA) for extracting cell-type specific gene expression profiles from mixed tissue samples that is unbiased and does not require prior knowledge of cell type frequencies.ConclusionsThe results suggest that DSA is a specific and sensitivity algorithm in gene expression profile deconvolution and will be useful in studying individual cell types of complex tissues.

A Local Poisson Graphical Model for Inferring Networks From Sequencing Data

Gaussian graphical models, a class of undirected graphs or Markov Networks, are often used to infer gene networks based on microarray expression data. Many scientists, however, have begun using high-throughput sequencing technologies such as RNA-sequencing or next generation sequencing to measure gene expression. As the resulting data consists of counts of sequencing reads for each gene, Gaussian graphical models are not optimal for this discrete data. In this paper, we propose a novel method for inferring gene networks from sequencing data: the Local Poisson Graphical Model. Our model assumes a Local Markov property where each variable conditional on all other variables is Poisson distributed. We develop a neighborhood selection algorithm to fit our model locally by performing a series of l1 penalized Poisson, or log-linear, regressions. This yields a fast parallel algorithm for estimating networks from next generation sequencing data. In simulations, we illustrate the effectiveness of our methods for recovering network structure from count data. A case study on breast cancer microRNAs (miRNAs), a novel application of graphical models, finds known regulators of breast cancer genes and discovers novel miRNA clusters and hubs that are targets for future research.

Molecular pathway identification using biological network-regularized logistic models

BackgroundSelecting genes and pathways indicative of disease is a central problem in computational biology. This problem is especially challenging when parsing multi-dimensional genomic data. A number of tools, such as L1-norm based regularization and its extensions elastic net and fused lasso, have been introduced to deal with this challenge. However, these approaches tend to ignore the vast amount of a priori biological network information curated in the literature.ResultsWe propose the use of graph Laplacian regularized logistic regression to integrate biological networks into disease classification and pathway association problems. Simulation studies demonstrate that the performance of the proposed algorithm is superior to elastic net and lasso analyses. Utility of this algorithm is also validated by its ability to reliably differentiate breast cancer subtypes using a large breast cancer dataset recently generated by the Cancer Genome Atlas (TCGA) consortium. Many of the protein-protein interaction modules identified by our approach are further supported by evidence published in the literature. Source code of the proposed algorithm is freely available at http://www.github.com/zhandong/Logit-Lapnet.ConclusionLogistic regression with graph Laplacian regularization is an effective algorithm for identifying key pathways and modules associated with disease subtypes. With the rapid expansion of our knowledge of biological regulatory networks, this approach will become more accurate and increasingly useful for mining transcriptomic, epi-genomic, and other types of genome wide association studies.

NLRP7 affects trophoblast lineage differentiation, binds to overexpressed YY1 and alters CpG methylation

Maternal-effect mutations in NLRP7 cause rare biparentally inherited hydatidiform moles (BiHMs), abnormal pregnancies containing hypertrophic vesicular trophoblast but no embryo. BiHM trophoblasts display abnormal DNA methylation patterns affecting maternally methylated germline differentially methylated regions (gDMRs), suggesting that NLRP7 plays an important role in reprogramming imprinted gDMRs. How NLRP7-a component of the CATERPILLAR family of proteins involved in innate immunity and apoptosis-causes these specific DNA methylation and trophoblast defects is unknown. Because rodents lack NLRP7, we used human embryonic stem cells to study its function and demonstrate that NLRP7 interacts with YY1, an important chromatin-binding factor. Reduced NLRP7 levels alter DNA methylation and accelerate trophoblast lineage differentiation. NLRP7 thus appears to function in chromatin reprogramming and DNA methylation in the germline or early embryonic development, functions not previously associated with members of the NLRP family.

Gene expression deconvolution in linear space

1. Mardis, E.R. Genome Med. 2, 84 (2010). 2. Goecks, J., Nekrutenko, A. & Taylor, J. Genome Biol. 11, R86 (2010). 3. Hull, D. et al. Nucleic Acids Res. 34, W729–W732 (2006). 4. Ioannidis, J.P. et al. Nat. Genet. 41, 149–155 (2009). 5. Piwowar, H.A., Day, R.S. & Fridsma, D.B. PLoS One 2, e308 (2007). 6. Guttman, M. et al. Nat. Biotechnol. 28, 503–510 (2010). 7. Chen, X. et al. Cell 133, 1106–1117 (2008). 8. Marson, A. et al. Cell 134, 521–533 (2008). GeneProf automatically generates summary statistics and plots, helping researchers to identify flaws in the analysis procedure (and data). In combination with quick workflow generation using wizards, this opens up new ways to explore data and adjust analysis accordingly. For example, we have often observed poor alignment results owing to declining quality of short reads in progressive sequencing cycles. Users can readily spot such shortcomings in GeneProf ’s summary statistics and adjust workflows appropriately, for instance, by trimming read sequences before alignment. A key feature of GeneProf is the tight coupling of data and analyses in the form of ‘virtual experiments’. Virtual experiments are supplemented by all intermediate results and a history of the analysis procedure and can be directly linked in publications. This helps researchers avoid irreproducible and cryptic methodologies and improves the impact of publications4,5; in addition, researchers can share their analyses securely with collaborators before publication. All data and results remain the intellectual property of the user and are confidential until made public, at which point every visitor to the website can view the entire experiment and search, browse, visualize and export data. Registered users can easily import and reuse public data in other experiments. We used GeneProf to reanalyze over eight billion short read sequences from 752 high-throughput sequencing runs in a growing database of curated RNA-seq and chromatinimmunoprecipitation (ChIP)-seq experimental results and analyses. To demonstrate GeneProf ’s effectiveness in streamlining complicated analyses, we created an experiment using raw data from the Sequence Read Archive6. Using GeneProf ’s RNA-seq wizard, we constructed a workflow incorporating quality control, alignment and differential expression estimation with only six mouse clicks (Fig. 1a, Supplementary Methods and Supplementary Figs. 4–7). GeneProf can export comprehensive reports covering each experiment (Supplementary Data), which we propose to include in future publications. We then juxtaposed the results from this analysis with ChIP-seq data7,8 (Fig. 1b and Supplementary Fig. 8). GeneProf can summarize data from many different sources within seconds, for example, for a large-scale comparison of DNA-binding proteins (Fig. 1c and Supplementary Fig. 9). In summary, GeneProf is a user-friendly platform for the analysis of gene expression designed to cater to four groups of users. Experimental biologists and clinicians may consider GeneProf a resource and quickly access public information via the website. Researchers with their own data but limited analysis expertise can use the straightforward built-in wizards to run best-practice analyses. Experienced users and computational biologists benefit from the dynamic workflow designer to compile customized analysis pipelines. Software and algorithm developers can extend GeneProf ’s functionality by integrating new workflow components or by building software that accesses data maintained by GeneProf (Supplementary Discussion).

The Snf1-related kinase, Hunk, is essential for mammary tumor metastasis

We previously identified a SNF1/AMPK-related protein kinase, Hunk, from a mammary tumor arising in an MMTV-neu transgenic mouse. The function of this kinase is unknown. Using targeted deletion in mice, we now demonstrate that Hunk is required for the metastasis of c-myc-induced mammary tumors, but is dispensable for normal development. Reconstitution experiments revealed that Hunk is sufficient to restore the metastatic potential of Hunk-deficient tumor cells, as well as defects in migration and invasion, and does so in a manner that requires its kinase activity. Consistent with a role for this kinase in the progression of human cancers, the human homologue of Hunk is overexpressed in aggressive subsets of carcinomas of the ovary, colon, and breast. In addition, a murine gene expression signature that distinguishes Hunk-wild type from Hunk-deficient mammary tumors predicts clinical outcome in women with breast cancer in a manner consistent with the pro-metastatic function of Hunk in mice. These findings identify a direct role for Hunk kinase activity in metastasis and establish an in vivo function for this kinase.