The Greatest Barrier Reef

Teaching placental cells to live in a dish
Summer 2016

Placental cells are notoriously tough to culture—they need to move. Here, the cells (nuclei in blue) thrive in a churning microgravity bioreactor. Certain cell types fuse into conglomerates called syncytiotrophoblasts (red), which defend a fetus against infection.



As an expectant mother’s body pipes blood into the placenta, the blood swirls over a seabed of villi that look a bit like the polyps of a coral reef. Coating these structures is a layer of what are called syncytiotrophoblasts, cells that prevent viruses and other microbes from getting to the developing fetus. Researchers know little about these cells (or the human placenta in general, for that matter). But when syncytiotrophoblasts fail, the outcome can be disastrous.

“If you’re thinking about how an infectious agent associated with congenital disease—Toxoplasma gondii, cytomegalovirus, rubella virus, and now Zika—crosses the placental barrier, it’s these cells you should be studying,” says the University of Pittsburgh’s Carolyn Coyne, PhD associate professor of microbiology and molecular genetics. She studies how viruses bypass a host’s cellular barriers.

One problem is that placental cells are tough to work with. Though post-delivery placental tissue is accessible enough, placental cells are difficult to isolate, difficult to grow, and don’t stay around for long. With some cell types, cultured cell lines can model what happens in the body, but with placentas, not so much; syncytiotrophoblasts form when more basic cells called trophoblasts fuse, but that fusion is difficult to coax in the dish.

In traditional, 2-D culture, microbes (green) have free rein.

Recently, however, with the help of a technology developed by NASA, Coyne and her colleagues have created the first reliable cell-based system for culturing placental cells. They describe the approach in a report published in Science Advances in March.

Eight years ago, when Coyne was pregnant, she couldn’t help wondering whether the viruses she worked with could harm her baby (she was studying gut cells at the time). The published literature didn’t provide an answer. So she turned to Pitt’s Yoel Sadovsky, MD, scientific director of the Magee- Womens Research Institute, and the Elsie Hilliard Hillman Professor of Women’s and Infants’ Health Research, for some placental cells as well as advice on how to work with them (a discussion that led to a collaboration that’s still going strong). She learned that primary placental cells are highly resistant to viral infection, but cultured placental cells are the opposite—very permissive. To probe placentas’ antiviral powers, then, the team would need a better model.

But in Coyne’s 3-D culture shown here, spherical-shaped syncytiotrophoblasts form and fend them off.

Her first thought was of the movement of maternal blood and the sheer force it generates. Could a 3-D system awash in fluids be the missing ingredient? For its work on the gut and the blood-brain barrier, her lab had recently bought a NASA-developed microgravity bioreactor that keeps the culture medium circulating constantly. (The device looks like a slushie machine.) The bioreactor not only produces sheer force, but also mimics the membrane curvature of a placental coral reef by seeding cells on a matrix of tiny, porous beads.

The researchers tried several existing trophoblast lines that failed to grow in the system. Then the team realized that in trophoblasts’ natural environment, they tango with lots of other cells. So after putting several combinations through the bioreactor, the team hit upon a cell line that formed syncytiotrophoblasts in the presence of certain endothelial cells. “Morphologically, it was very clear,” Coyne says. The fused cells they had cultured also secreted pregnancy-associated hormones and upregulated the set of genes that they would typically in a pregnant woman’s body.

Coyne’s team is still perfecting the system. To make it easier to use, they’ve found a way to remove the syncytiotrophoblasts from the beads and plate them in a plastic dish. With the new technique, researchers can begin to explore the mechanics of how disease-causing agents do or don’t cross the placental barrier.

In April the group published a paper in Cell Host & Microbe showing that cells taken directly from the placenta following delivery resist Zika virus infection. “These cells exist to keep pathogens out,” Coyne says. By manipulating the genetics of the cell line, they hope to understand how resistance is mounted and explore several possible explanations for how a virus might break down or bypass it. (Maybe placental cells from early stages in gestation are not as resistant, Coyne says. Or maybe the virus gets in via some other trophoblast type. Or maybe it’s not infecting placental cells at all, but hitching a ride on an antibody or some other “Trojan horse” instead.) Once the mechanism becomes clear, she says the 3-D system could be fertile ground for a new pursuit: screening for therapeutic compounds that could restrict infection.

Images reprinted with cropping alterations from Science Advances. McConkey C, Delorme-Axford E, Nickerson C, Kim K, Sadovsky Y, Boyle J, Coyne C under Creative Commons Attribution 4.0 License. © 2016.

A clinician's crash course on future medicine


This May, CVS Pharmacy announced a partnership with 23andMe, a genetic testing company, to sell test kits directly to consumers for about $30. Spit into a test tube, mail it off, and six to eight weeks and a $169 lab fee later, you’ll have data on your genome, no prescription necessary. Inevitably, says Philip Empey, a PhD and PharmD assistant professor of pharmacy and therapeutics at the University of Pittsburgh, patients will bring results like these—and the questions they raise—to pharmacists and physicians: What does it mean if I have a mutation associated with such-and-such disease?

In a lecture room in Friendship this March, a group of 60-some health care professionals pondered these and other uncertainties over beer. ... Well, not literally. Empey handed each of them a PTC taster strip, which tests for a trait associated with a specific gene variant. About three-quarters of Americans have the trait, which means they should find the strip’s taste overly bitter—India pale ale–averse types, by heredity. On the penultimate night of this unique course, the clinicians found out whether their genetics accurately predicted their palates.

Throughout eight weeks, “Big Data and Healthcare Analytics—A Path to Personalized Medicine” covered topics ranging from patient communication to compatibility of electronic health record systems at breakneck speed, for four dense hours per session. Pitt’s Institute for Personalized Medicine (IPM), the Big Data to Knowledge Center of Excellence, and the Schools of the Health Sciences organized the course with funding from the Jewish Healthcare Foundation.

To some of these docs, nurses, and pharmacists, the taste-test (i.e., gene variant) results were a surprise. Those who crinkled their noses at the bitterness of the strip on the first class night didn’t necessarily have the gene variant. “Why wouldn’t it be a perfect match?” Empey posed to the class. A number of possible reasons emerged: medical conditions, other interfering genes, an error in the genotyping, or even a preclass taco dinner. There are pitfalls in relying too heavily on genetic data in the clinic.

The students used a Pitt School of Pharmacy–developed software called Test2Learn, an educational tool allowing users to upload their 23andMe profiles to explore variants in more detail. (The software, which Empey’s team developed, also gives the option to use anonymous volunteer patient datasets instead; no one in the room could tell whether classmates were analyzing their own data.) Select a variant—rs713598, in the taste-testers’ case—then click Test2Learn’s “Interpret Gene” button. The software spits out keg-loads of genomic detail.

Empey also covered weightier scenarios in the course. Consider warfarin, the widely prescribed anticoagulant used to treat and prevent blood clots and heart attacks. The medication requires delicate dosing, as the risk of fatal bleeding is real. In addition to clinical factors, he explained, there are two genes (CYP2C9 and VKORC1) that are relevant to its prescription. (Again, personalized medicine is never just about genetics—complex variables are the norm rather than the exception.) Usually, doctors administer warfarin in a trial dose of 2 to 5 milligrams, then adjust levels, milligram by milligram, throughout several weeks to find a therapeutic equilibrium. With genetic testing, that guesswork—and the risks, costs, and time involved—dissipates.

There are about 2,400 known associations between drugs and genetic variants; Empey said 33 medications have guidelines with evidence for clinical use.

“This is our future,” he told the class. “We’ve got work ahead of us, training clinicians to use this information.”

Throughout the course, IPM ethicist Lisa Parker, another course codirector and PhD professor of human genetics in the Graduate School of Public Health, extensively lectured on the ethical and psychosocial concerns of sequencing. Other course directors included Yvette Conley, a PhD professor and vice chair for research in the School of Nursing; Empey, who’s associate director of pharmacogenomics for IPM; and Rebecca Jacobson, an MD/MSIS professor of biomedical informatics and pathology. Jacobson is also chief information officer for IPM. All told, nearly 20 instructors from across biomedical disciplines lectured.

Plans are brewing for a second round of the course.

Typography by Elena Cerri

How the Nose Knows

Sniffing out olfaction


A mouse wanders in total darkness along an infrared-sensing table that glows at the touch of the hand—or the tail, feet, and schnoz, in this case. After a brief false start, the rodent homes in on a scent that he's been trained to track and follows it to the end.



A tracking hound can scout out a fugitive who had a 24-hour head start. A trained pig can snout out truffles buried 3 feet underfoot. Even your average human, whose sniffer is far inferior by comparison, will eventually find whatever foulness is stinking up the kitchen. But scientists still have little idea how any of us are doing this.

Most studies of olfaction have focused on discrimination—how the nose knows whether it’s caught wind of banana or cherry, for instance. Neurobiologically speaking, “that’s a pretty simple task for a mouse,” says Nathan Urban. But recently, Urban, whose lab has studied the brain networks involved in mouse olfaction for 13 years, has been hot on the trail of a much more complex olfaction task of localizing odors, a marvel of nature that no manmade technology can replicate.

Last fall, Urban, PhD professor of neurobiology, and Bard Ermentrout, PhD professor of computational and systems biology, both of Pitt, became part of a National Science Foundation– funded multi-institutional team, to the tune of $6.4 million. The olfaction faction also includes Justus Verhagen, a rodent neurophysiologist from Yale; John Crimaldi, a fluid mechanics expert from the University of Colorado; Lucia Jacobs, an evolutionary psychologist from Berkeley who’s focusing on studies of dogs for the project; Jonathan Victor, a computational neuroscientist from Cornell; and Katherine Nagel, a fruit-fly olfaction investigator from New York University. Their collaboration was born at the NSF Olfactory Ideas Lab workshop in June 2015.

The team is mapping the smelling brain and its minute mechanisms, and building computational models and other experiments to understand how scents move through the air. They hope to sketch out common principles across several species—which might one day inform new technologies (explosives-sniffing robots, mosquito-olfaction muddlers). Such principles might also provide insight into a number of neurological disorders in humans—including Alzheimer’s, autism, and Parkinson’s—in which sensory processing suffers.

Among the Urban lab’s ongoing studies are those of mice amid blind scent-tracing tests (see image above). One year into this three-year award, his team is yielding intriguing findings.

For one, individual mouse neurons are “lousy devices,” he says. Stimulate one 10 times in a row, and it will fire maybe five times. “If the S key on your keyboard only worked half the time, you’d throw it away,” Urban notes. And yet somehow, collectively, the neurons in these networks are not just good, but great at what they do in many animals, rodents included. (Giant rats have been trained in landmine detection in several countries. The pint-sized patrols have already secured millions of miles.)

“How you get useful, robust, reliable function from unreliable components has been one interesting area for us to explore,” Urban says. He thinks perhaps this variability is not a bug, but an advantage that leaves room for adaptation and the possibility to detect and respond to a wider range of incoming stimuli.

Another interesting finding involves the behavior of casting—when a snout sways from side to side, surveying for scents. Urban’s team is finding that mice turn their heads invariably toward an odor’s source with such speed and accuracy that they must be making a decision with every single sniff. And they sniff a lot—almost 15 times per second. “So in 70 milliseconds, they’re inhaling, and they’re beginning to move their heads in the right direction. That doesn’t give much time for the brain to perform this calculation. That’s one of the clues we have as to where to look in the brain for neurons that are sensitive to . . . sniff-to-sniff differences in the intensity of a stimulus.”

And even if a mouse has one nostril plugged, it’s still pretty good at tracking, which means left-right differences don’t figure into olfaction as they do in vision—a finding that surprised Urban. His collaborator at NYU is finding the same is true in fruit flies.

Perhaps the biggest surprise of all for the team has been the nature of odor itself. At the onset, the life-sciences folk had figured on a simple bell curve, with the odor strongest in the middle and thinning out on an even gradient—not so. “We were being far too simplistic in how we were thinking about this,” says Urban. “Everything is sort of mixing and turbulent all the time, even in a room where you can’t feel any airflow.” There’s work to do yet, but with guidance from the fluid dynamics expert, the team is moving in the right direction—nose to the grindstone.

Image courtesy Nathan Urban Lab