Biology subjects

Ma, E.

Publications and source records attributed to Ma, E..

25 records · Page 2Linked to original sources

bioRxiv2020

The Wnt effector TCF7l2 promotes oligodendroglial differentiation by repressing autocrine BMP4-mediated signaling

Promoting oligodendrocyte differentiation represents a promising option for remyelination therapy for treating the demyelinating disease multiple sclerosis (MS). The Wnt effector TCF7l2 was upregulated in MS lesions and had been proposed to inhibit oligodendrocyte differentiation. Recent data suggest the opposite yet underlying mechanisms remain elusive. Here we unravel a previously unappreciated function of TCF7l2 in controlling autocrine bone morphogenetic protein (BMP4)-mediated signaling. Disrupting TCF7l2 results in oligodendroglial-specific BMP4 upregulation and canonical BMP4 signaling activation in vivo. Mechanistically, TCF7l2 binds to Bmp4 gene regulatory element and directly represses its transcriptional activity. Functionally, enforced TCF7l2 expression promotes oligodendrocyte differentiation by reducing autocrine BMP4 secretion and dampening BMP4 signaling. Importantly, compound genetic disruption demonstrates that oligodendroglial-specific BMP4 deletion rescues arrested oligodendrocyte differentiation elicited by TCF7l2 disruption in vivo. Collectively, our study reveals a novel connection between TCF7l2 and BMP4 in oligodendroglial lineage and provides new insights into augmenting TCF7l2 for promoting remyelination in demyelinating disorders such as MS. Significance StatementIncomplete or failed myelin repairs, primarily resulting from the arrested differentiation of myelin-forming oligodendrocytes from oligodendroglial progenitor cells, is one of the major reasons for neurological progression in people affected by multiple sclerosis (MS). Using in vitro culture systems and in vivo animal models, this study unraveled a previously unrecognized autocrine regulation of BMP4-mediated signaling by the Wnt effector TCF7l2. We showed for the first time that TCF7l2 promotes oligodendroglial differentiation by repressing BMP4-mediated activity, which is dysregulated in MS lesions. Our study suggests that elevating TCF7l2 expression may be possible in overcoming arrested oligodendroglial differentiation as observed in MS patients.

neuroscience↗

bioRxiv2020

Reimplementing Unirep in JAX

UniRep is a recurrent neural network model trained on 24 million protein sequences, and has shown utility in protein engineering. The original model, however, has rough spots in its implementation, and a convenient API is not available for certain tasks. To rectify this, we reimplemented the model in JAX/NumPy, achieving near-100X speedups in forward pass performance, and implemented a convenient API for specialized tasks. In this article, we wish to document our model reimplementation process with the goal of educating others interested in learning how to dissect a deep learning model, and engineer it for robustness and ease of use.

bioengineering↗

bioRxiv2020

DDX5 targets tissue-specific RNAs to promote intestine tumorigenesis

Tumorigenesis in different segments of the intestinal tract involves tissue-specific oncogenic drivers. In the colon, complement component 3 (C3) activation is a major contributor to inflammation and malignancies. By contrast, tumorigenesis in the small intestine involves fatty acid-binding protein 1 (FABP1). However, little is known of the upstream mechanisms driving their expressions in different segments of the intestinal tract. Here, we report that an RNA binding protein DDX5 augments C3 and FABP1 expressions post-transcriptionally to promote tumorigenesis in the colon and small intestine, respectively. Mice with epithelial-specific knockout of DDX5 are protected from intestine tumorigenesis. The identification of DDX5 as the common upstream regulator of tissue-specific oncogenic molecules provides a new therapeutic target for intestine cancers.

cancer biology↗

bioRxiv2019

Peax: Interactive Visual Pattern Search in Sequential Data Using Unsupervised Deep Representation Learning

We present PO_SCPLOWEAXC_SCPLOW, a novel feature-based technique for interactive visual pattern search in sequential data, like time series or data mapped to a genome sequence. Visually searching for patterns by similarity is often challenging because of the large search space, the visual complexity of patterns, and the users perception of similarity. For example, in genomics, researchers try to link patterns in multivariate sequential data to cellular or pathogenic processes, but a lack of ground truth and high variance makes automatic pattern detection unreliable. We have developed a convolutional autoencoder for unsupervised representation learning of regions in sequential data that can capture more visual details of complex patterns compared to existing similarity measures. Using this learned representation as features of the sequential data, our accompanying visual query system enables interactive feedback-driven adjustments of the pattern search to adapt to the users perceived similarity. Using an active learning sampling strategy, PO_SCPLOWEAXC_SCPLOW collects user-generated binary relevance feedback. This feedback is used to train a model for binary classification, to ultimately find other regions that exhibit patterns similar to the search target. We demonstrate PO_SCPLOWEAXC_SCPLOWs features through a case study in genomics and report on a user study with eight domain experts to assess the usability and usefulness of PO_SCPLOWEAXC_SCPLOW. Moreover, we evaluate the effectiveness of the learned feature representation for visual similarity search in two additional user studies. We find that our models retrieve significantly more similar patterns than other commonly used techniques.

genomics↗

bioRxiv2017

CRISPR-Cas12a target binding unleashes single-stranded DNase activity

CRISPR-Cas12a (Cpf1) proteins are RNA-guided DNA targeting enzymes that bind and cut DNA as components of bacterial adaptive immune systems. Like CRISPR-Cas9, Cas12a can be used as a powerful genome editing tool based on its ability to induce genetic changes in cells at sites of double-stranded DNA (dsDNA) cuts. Here we show that RNA-guided DNA binding unleashes robust, non-specific single-stranded DNA (ssDNA) cleavage activity in Cas12a sufficient to completely degrade both linear and circular ssDNA molecules within minutes. This activity, catalyzed by the same active site responsible for site-specific dsDNA cutting, indiscriminately shreds ssDNA with rapid multiple-turnover cleavage kinetics. Activation of ssDNA cutting requires faithful recognition of a DNA target sequence matching the 20-nucleotide guide RNA sequence with specificity sufficient to distinguish between closely related viral serotypes. We find that target-dependent ssDNA degradation, not observed for CRISPR-Cas9 enzymes, is a fundamental property of type V CRISPR-Cas12 proteins, revealing a fascinating parallel with the RNA-triggered general RNase activity of the type VI CRISPR-Cas13 enzymes.\n\nOne Sentence SummaryCas12a (Cpf1) and related type V CRISPR interference proteins possess non-specific, single-stranded DNase activity upon activation by guide RNA-dependent DNA binding.

biochemistry↗

bioRxiv2017

A thermostable Cas9 with increased lifetime in human plasma

CRISPR-Cas9 is a powerful technology that has enabled genome editing in a wide range of species. However, the currently developed Cas9 homologs all originate from mesophilic bacteria, making them susceptible to proteolytic degradation and unsuitable for applications requiring function at elevated temperatures. Here, we show that the Cas9 protein from the thermophilic bacterium Geobacillus stearothermophilus (GeoCas9) catalyzes RNA-guided DNA cleavage over a wide temperature range and has an enhanced protein lifetime in human plasma. GeoCas9 is active at temperatures up to 70{degrees}C, compared to 45{degrees}C for Streptococcus pyogenes Cas9 (SpyCas9), which greatly expands the temperature range for CRISPR-Cas9 applications. By comparing features of two closely related Geobacillus homologs, we created a variant of GeoCas9 that doubles the DNA target sequences that can be recognized by this system. We also found that GeoCas9 is an effective tool for editing mammalian genomes when delivered as a ribonucleoprotein (RNP) complex. Together with an increased lifetime in human plasma, the thermostable GeoCas9 provides the foundation for improved RNP delivery in vivo and expands the temperature range of CRISPR-Cas9.

biochemistry↗

bioRxiv2016

Bayesian Analysis of High Throughput Data

Duplicate or triplicate experimental replicates are commonplace in the high throughput literature. However, it has not been tested whether this is statistically defensible or not. To address this issue, we use probabilistic programming to develop a simple hierarchical model for analyzing high throughput measurement data. With the model and simulated data, we show that a small increase in replicate experiments can quantitatively improve accuracy in measurement. We also provide posterior densities for statistical parameters used in the evaluation of HT data. Finally, we provide an extensible open source implementation that ingests data structured in a simple format and produces posterior densities of estimated measurement and assay evaluation parameters.

Scientific Communication and Education↗