Nature Structural Biology pp 371 - 378

More than 100,000 people are diagnosed each year with leukemia, lymphoma or other blood-related cancers. Leukemia, lymphoma, Hodgkin's disease and myeloma are cancers of the body's blood forming and immune systems—the bone marrow and lymph nodes. They are considered to be related cancers because they involve the uncontrolled growth of cells with similar functions.

Despite much progress over the past several decades, most adults who develop leukemia die of the disease or complications associated with therapy, and leukemia is still the most common cause of death in children with cancer.

Most if not all cases of human acute myeloid leukemia (AML) are caused by mutations in or rearrangements of genes. Among the most common gene rearrangements are those involving the transcription factor complex, core-binding factor (CBF). CBF consists of two parts, one that binds directly to DNA (CBFalpha) and a second one (CBFbeta) that helps CBFalpha bind DNA. All alpha-subunits share a region that is conserved throughout evolution known as the Runt domain. This domain is responsible for both binding to DNA and interacting with the beta-subunit. Three genes in mammals encode CBFalpha and mutations in these genes are associated with human diseases.

To understand how the disease-associated mutations affect DNA binding Allan Warren of the Medical Research Council in Cambridge, UK and coworkers determined the three dimensional X-ray crystal structure of a complex between the AML1 Runt domain, CBFbeta and DNA. An important finding that comes from the structure of the complex is that most of the disease-associated mutations are in places in the Runt domain that specifically recognize the DNA. Thus, the structure explains why human disease-associated mutations in leukemogenesis and cleidocranial dysplasia lead to a loss in DNA binding since mutations at these positions are unable to interact with the DNA as well.

Nature Structural Biology pp 316 - 320 and pp 211 - 220

It seems that lately you can't pick up a paper or turn on the TV without being confronted with reports of mad cow disease and its human equivalent, variant Creutzfeld-Jacob disease (vCJD). These diseases have something in common with other disorders, including Alzheimer's disease — protein plaques get deposited into the brain where they presumably destroy tissue.

These plaques are composed of fibrils of certain proteins, prion proteins in the case of mad cow disease and vCJD and amyloid beta proteins in the case of Alzheimer's disease. Although it has not been rigorously proven that the protein plaques are the direct cause of the nerve degeneration seen in affected cattle and people, it seems a highly probable scenario.

How these fibrils of proteins form from many identical molecules of a single type of protein is not well understood. Clearly, the individual protein molecules must interact in some regular and stable way to form the linear fibrils that eventually group together to form a plaque. For some time, researchers have suggested that a particular type of interaction, known as 'domain swapping' may be involved in fibril formation, and now two recent studies published in Nature Structural Biology support this idea.

'Domain swapping' means that two proteins of identical structure may swap the same structural element — say, one alpha-helix or one beta-strand — to complete each other's final fold. The result is a pair of linked proteins with identical structures, instead of two unlinked proteins with the same structure. A way to visualize this might be to imagine two people shaking hands: separately they each have all the same parts in basically the same shape, but when they link hands — the identical structural elements — they 'swap' parts to become a linked pair.

In the April issue of Nature Structural Biology, Mariusz Jaskolski, of the Center for Biocrystallographic Research in Poland, and colleagues have solved the structure of human cystatin C, another protein that can form deposits in the brain, in this case leading to fatal cerebral hemorrhage. They show that this protein can form 'domain swapped' pairs of proteins.

In the March issue of Nature Structural Biology, David Eisenberg, of UCLA in the USA, and colleagues showed that another protein, called RNase A, can actually form two different types of 'domain swapped' protein pairs. In the two different cases, different structural elements are swapped. This suggests a way that a fibril of linked proteins could be generated. Using the hand-shaking analogy described above, if the two different structural elements are the two hands of an individual, then a linked chain of people could form by linking hands — by swapping the different structural elements in regular order, to build a chain of molecules.

Although the RNase A protein is not known to form plaques in the body, if a similar 'double domain swapping' mechanism could be used by plaque-forming proteins such as human cystatin C, prions, and amyloid-beta, then this might explain how the stable plaque-forming fibrils grow.

Marcia Newcomer discusses these two papers in a News and Views report, and the Editorial discusses mad cow disease and vCJD.