Columbia University Medical Center
In Vivo - The Newsletter of Columbia University Medical Center
Back Issues
Contact Us
In Vivo

Computer Corrects Molecular Networks

Like Star Trek’s Mr. Spock, scientists have a knack for spotting illogical statements.
   For example, when one lab says that the HBP1 protein is active in neurons, and another lab says that WNT1 is active, and still a third says that WNT1 is inactive whenever HBP1 is active, it’s clear that something isn’t quite right.
   However, there are limits to such powerful mental skills. “Our brains are simply not equipped with the hardware to pick out incompatibilities in biological data when there are thousands or millions of proteins and protein interactions to consider,” says Andrey Rzhetsky, Ph.D., associate professor of biomedical informatics. “It’s a big problem that can slow progress in research.”
   To solve the problem, Dr. Rzhetsky, his Ph.D. student Chani Weinreb, and Tian Zheng, Ph.D., assistant professor of statistics, have developed a computer program to identify the most unreliable data, and also the most reliable, automatically.
   The program first plucks reports, such as that HBP1 is active in neurons, from the literature. Based on hundreds of thousands of reports, the program constructs an uncorrected molecular network (figure 1), left. In the final step, the program computes a network that has no internal conflicts.
   Comparing the two networks reveals which data are suspect (figure 2), right. “Conflicts could mean the information is wrong, or that there is missing information that could resolve the contradiction,” Ms. Weinreb says. “The idea is not to say ’you guys are wrong,’ it’s to suggest the most fruitful areas for further research.”
   This work was supported by the NIH, the National Science Foundation, Cure Autism Now Foundation, and the Defense Advanced Research Projects Agency.

Computer-constructed networks reveal potential errors in published data
Computer-constructed networks reveal potential errors in published data. An uncorrected network of genes involved in four brain disorders (Alzheimer’s, autism, bipolar disorder and schizophrenia) was constructed from data pulled from nearly 15,000 articles top, (top). Rzhetsky and Weinreb’s algorithm constructs a new network that removes conflicts and points out potential errors bottom, (bottom). For example, the large WNT1 node suggests the protein may not truly be active in neurons or that it is active only at certain times.

Diabetes Target Takes Shape
Diabetes Target Takes ShapeInsight into one of the most significant targets for diabetes has gotten clearer. The first high-resolution image of AMP-activated protein kinase (AMPK), below, could lead to new drugs that activate the enzyme, say the researchers who uncovered AMPK’s molecular structure, Lawrence Shapiro, Ph.D., associate professor in the Departments of Biochemistry & Molecular Biophysics and Ophthalmology, and graduate student Rob Townley.
   Like insulin, AMPK stimulates glucose uptake by cells and reduces the liver’s release of glucose into the bloodstream. Drugs that activate AMPK eliminate hyperglycemia in diabetic rats, but the compounds have not been tried in people because they also interfere with other enzymes.
   The details of AMPK’s structure revealed in the new image show how the enzyme is turned on and off and give chemists a clear picture of how to design more selective activators.
Science 2007 315(5819): 1726-1729. This research was supported in part by a National Institute of Diabetes and Digestive and Kidney Diseases grant and a Jules and Doris Stein Research to Prevent Blindness Foundation professorship award.

Anesthesia/Alzheimer’s Link
Postoperative cognitive dysfunction, confusion and delirium are common after general anesthesia in the elderly. Previous clinical studies have suggested that general anesthesia hastens the progress of Alzheimer’s disease, but little is known about the impact of anesthesia on the two hallmarks of Alzheimer’s disease in the brain: beta-amyloid peptide and tau protein.
   Now, researchers at Columbia’s Taub Institute for Research on Alzheimer’s Disease and the Aging Brain have found that by lowering body temperature, anesthesia rapidly induces a massive change in hallmark proteins implicated in cognitive decline.
   Emmanuel Planel, Ph.D., assistant professor of pathology, and Karen Duff, Ph.D. professor of pathology, anesthetized mice with different anesthetics and then examined their neurons. Each anesthetic increased the number of tau proteins in a state of hyperphosphorylation. Tau protein is the main component of neurofibrillary tangles; the presence of tangles is linked with cognitive decline.
   The change in tau phosphorylation did not result from the anesthesia itself, but from
anesthesia-induced hypothermia. When anesthetized mice were warmed to normal temperature, tau hyperphosphorylation disappeared.
   It is still unknown if temporary tau hyperphosphorylation accelerates the disease’s course, so the next step would be to repeat the experiment in transgenic mouse models of Alzheimer’s. Drs. Planel and Duff say hypothermia should be taken into account when investigating the link between anesthesia and cognitive dysfunction or Alzheimer’s.
Journal of Neuroscience 27(12): 3090-3097. This work was funded by the NIH.

Follow-up Screening for Celiac Disease Necessary
A single test for celiac disease is often not adequate to detect the disease in all relatives of previously diagnosed patients, according to a new study by Peter Green, M.D., professor of clinical medicine, David Goldberg, M.D., postdoctoral residency fellow in medicine, and colleagues at the University of Maryland.
   Celiac disease is a genetically determined intolerance to gluten that destroys the lining of the small intestine over time. First- and second-degree relatives of patients with celiac disease are usually screened for the disease.
   Researchers analyzed records from a large database maintained at the University of Maryland’s celiac center. It showed that while the majority of cases of celiac disease diagnosed in relatives of patients with the disease is detected on the first test, 3.5 percent of relatives who initially test negative will test positive when a second test is performed several months to years later. Most notably, none of the patients who developed positive serologic tests for celiac disease had any change in their baseline symptoms.
   Dr. Green says until further studies clarify the frequency and types of tests needed, it is reasonable to wait four to five years before retesting.
Digestive Diseases and Sciences 52(4):1082-6