HHMI-NIH Research Scholars
Learn about the HHMI-NIH Research Scholars Program, also known as the Cloister Program. More

Janelia Farm Research Campus
Learn about the new HHMI research campus located in Virginia. More

Summary: Mark Keating's laboratory focuses on the molecular mechanisms of organ regeneration and the human molecular genetics of cardiovascular disease.

Regeneration Genetics
Regenerative inadequacy is an important cause of disease. Through our studies of cardiovascular disease, it became clear that recurrent injury followed by inadequate regeneration and fibrosis is a common pathogenic mechanism. This is the case for atherosclerosis, heart failure, arrhythmia, and many other diseases. Humans regenerate liver and digit tips, but this pales when compared with the regenerative capacity of urodele amphibians, such as newts. These organisms regenerate retina, spinal cord, limbs, pancreas and other tissues with great facility and without scar—a process Thomas Hunt Morgan called epimorphic regeneration.

Although regeneration has fascinated humans for thousands of years, regeneration research has been slow to progress to a molecular level, possibly because traditional model organisms are not amenable to genetic manipulations, and traditional genetic organisms have limited regenerative abilities. Zebrafish, by contrast, regenerate fins, spinal cord, and optic nerve, and are mainstream genetic organisms for the study of early vertebrate development. Genetic strategies for studying an adult phenotype in zebrafish were not available, however, until we collaborated with Stephen Johnson (Washington University) to develop regeneration genetics in zebrafish.

	Dedifferentiation of murine myotube... Regeneration of zebrafish heart...

We used chemical mutagenesis to identify zebrafish fin regeneration mutants. We focused initially on fin regeneration because it is fast (7 days) and easy to assay. Assuming most mutations would be lethal in embryos, we screened for temperature-sensitive mutants. To discover the first regeneration genes— mps1 and sly1 —we used positional cloning. To implicate fgfr1 and msx1 , we and others used a candidate gene approach. Finally, we used expression analyses to associate wfgf, msx2, lef1, wnt3a, wnt5, β -catenin , and sonic hedgehog with regeneration. These studies provided genetic insight into regeneration and valuable tools for uncovering molecular mechanisms.

We next examined the function of proteins encoded by regeneration genes. The signaling protein wfgf is expressed in the wound epidermis shortly after injury. An fgf receptor, fgfr1, is expressed in mesenchymal cells underlying the wound epidermis during blastema formation. The blastema is a cell population immediately beneath the wound epidermis. Fgfr1 signaling is required for blastema formation. Msx proteins, which are homeodomain transcription repressors, are expressed in mesenchymal cells that coalesce to form pluripotent regeneration cells of the distal blastema. Regeneration cells, like stem cells, provide daughter cells for regenerative outgrowth. We discovered that msx can induce cellular dedifferentiation and create pluripotent cells. The mps1 gene encodes a mitotic checkpoint kinase critical for cell cycle progression in rapidly proliferating cells. An intracellular trafficking protein is encoded by s ly1 . Expression of both mps1 and sly1 is induced in, and restricted to, intensely proliferating, proximal blastema cells during regenerative outgrowth. Loss of mps1 or sly1 function eliminates regenerative outgrowth. The wnt and hedgehog proteins, which are signaling molecules expressed in the epithelium adjacent to the blastema, appear to be essential for patterning, and fgfr1 is required for their expression and function. These data indicate that signaling molecules, transcription repressors, and cell cycle proteins are essential for epimorphic regeneration.

Regeneration cells behave like stem cells. Our work, together with previous developmental studies, provides a picture of vertebrate regeneration at the molecular, cellular, and organ levels. The first step in fin regeneration is formation of a wound epithelium. We discovered that this nonproliferative step involves migration of existing epithelial cells to cover the wound. The second step is creation of the primary blastema. We found no evidence that these cells are derived from slow-cycling stem cells. Instead, they are derived from existing mesenchymal cells beneath the wound epithelium, presumably through cellular dedifferentiation. Creation of blastema cells requires fgfr1 signaling. Expressed in regeneration cells of the fin, msx genes may induce dedifferentiation. These loosely distributed cells coalesce into the primary blastema, just beneath the wound epidermis.

The third step in regeneration is blastema maturation and regenerative outgrowth. We discovered that the blastema is organized into two compartments, a distal blastema that contains nonproliferating, msx ⁺ /mps1 ^– regeneration cells, and a proximal blastema that contains msx ^– /mps1 ⁺ daughter cells. The intense proliferation of proximal blastema cells—50-fold increase across 10 cells—drives regenerative outgrowth.

The fourth stage of regeneration is differentiation and patterning. Epithelial cells adjacent to the proximal blastema express signaling molecules, such as wnts and sonic hedgehog. These proteins apparently organize the formation and alignment of scleroblasts, forming new bone, nerves, and blood vessels. Regeneration ceases when fin size and structure have been recreated.

Zebrafish hearts regenerate through cardiomyocyte proliferation, preempting scar formation. We next examined the cardiac regenerative capacity of zebrafish. Human hearts do not regenerate. Instead, damaged myocardium is replaced by scar. This significant medical problem leads to an epidemic of heart failure, arrhythmia, and death. Cardiomyocytes, the major cardiac structural cells, undergo hypertrophy to increase muscle mass after cardiac injury. Although recent findings suggest that cardiomyocytes within the diseased human heart may proliferate, most evidence to date indicates that cardiomyocyte proliferation is not a significant component of the mammalian response to acute injury.

We discovered that zebrafish fully regenerate lost heart muscle within 2 months of 20 percent ventricular resection. Regeneration occurs through proliferation of cardiomyocytes localized at the epicardial edge of new myocardium. Hearts of zebrafish with mutations in mps1 failed to regenerate, forming scars. Thus, injury-induced cardiomyocyte proliferation overcomes scar formation, enabling cardiac regeneration. Our work establishes a model system for genetically dissecting molecular mechanisms of cardiac regeneration.

Mammalian cells dedifferentiate. A key feature of epimorphic regeneration is cellular dedifferentiation, which creates pluripotent regeneration cells. Like cardiomyocytes, differentiated mammalian cells were considered "terminally" differentiated and incapable of dedifferentiation. These cells have exited the cell cycle in response to expression of cyclin-dependent kinase inhibitors, activation of members of the retinoblastoma family, and down-regulation of cyclins and cyclin-dependent kinases.

We discovered, however, that differentiated murine myotubes can dedifferentiate. Ectopic expression of msx1 in C2C12 myotubes reduced the nuclear muscle proteins MyoD, myogenin, MRF4, and p21 to undetectable levels in 20–60 percent of myotubes. Approximately 9 percent of myotubes cleave to produce smaller multinucleated myotubes or proliferating mononucleated cells. Finally, clonal populations of myotube-derived mononucleated cells could be induced to redifferentiate into cells expressing chondrogenic, adipogenic, myogenic, and osteogenic markers. We achieved similar results using an extract derived from regenerating newt limbs. These findings indicate that "terminally" differentiated mammalian cells can dedifferentiate when stimulated with appropriate signals, and that msx1 can trigger the dedifferentiation process. If dedifferentiation and proliferation of cardiomyocytes can be achieved in vivo, it may be possible to develop therapeutics enhancing cardiac regeneration in humans.

Mechanisms of Cardiac Arrhythmia
Common arrhythmias are caused by dysfunction of ion channels. Cardiac arrhythmias cause 450,000 sudden deaths annually in the United States. Despite their importance, when we began these studies, our mechanistic understanding of arrhythmias was poor. We discovered the first arrhythmia genes: KVLQT1 , HERG , SCN5A , mink , and MiRP1 . Other groups have implicated additional genes. We were first to discover that arrhythmia genes encode the ion channels regulating cardiac excitability. Our studies demonstrate that cardiac ion channel dysfunction causes familial arrhythmias.

We discovered that HERG mutations are the most common cause of familial arrhythmia (45 percent). Mutations affecting the HERG pore are particularly lethal. We demonstrated that HERG dysfunction also causes drug-induced long-QT syndrome, a side effect of many medications. Most of these medications unintentionally block HERG channels.

Coupled with previous physiologic studies, our work provides an extensive view of arrhythmia mechanisms. Cardiac ion channel dysfunction causes conduction abnormalities. Regional cardiac conduction abnormalities create unidirectional block, a substrate for arrhythmia. Arrhythmias are triggered by spontaneous secondary depolarizations and maintained by regenerative circuits of electrical activity around relatively unexcitable tissue, a phenomenon known as reentry. Ordered arrhythmias can degenerate to chaotic electrical activity and ventricular fibrillation, the arrhythmia of sudden death.

Most arrhythmias are acquired, not familial. Myocardial infarction, ischemia, and myopathy are common arrhythmia risk factors. Many individuals at risk do not, however, develop arrhythmia. The reasons for variability are not understood. We discovered an SCN5A variant associated with acquired arrhythmia in African Americans. In collaboration with Robert Kass (Columbia University), we demonstrated that the Y1102 variant accelerates channel activation, increasing the likelihood of abnormal cardiac repolarization and arrhythmia. About 13.2 percent of African Americans carry Y1102. Our work shows that ion channel dysfunction contributes to arrhythmia in the general population.

Grants from the National Institutes of Health provided partial support for this project.

Last updated October 27, 2004