of the viral DNA ends, cleavage of target DNA cou- 
pled to joining of viral and target DNA, and the re- 
verse of the joining reaction, or "disintegration." 
Dr. Vincent's analysis of the mutant forms of inte- 
grase has demonstrated that a highly conserved zinc 
finger-like motif near the amino terminus is not es- 
sential for catalysis or recognition of the terminal 
bases of the viral DNA, but may be important for 
recognizing features internal to the viral DNA end. 
Dr. Chow's continued investigation of unortho- 
dox substrates has clarified several aspects of sub- 
strate recognition and catalysis by integrase. First, 
when an especially favorable disintegration sub- 
strate is used, more than 1 0 catalytic events per inte- 
grase monomer are detected over a period of a few 
hours. Thus integrase can act as an enzyme. Second, 
integrase recognizes its DNA substrates in structures 
that deviate from the double-helical form. Indeed, 
substrates in which the viral DNA ends are not base- 
paired appear to be preferred, suggesting that inte- 
grase binding to normal viral DNA ends may involve 
separation of the two strands at the end. Third, by 
using disintegration substrates in which target DNA 
is truncated either on one side or the other of the 
junction with viral DNA, the integrase has been 
found to interact asymmetrically with the target 
DNA. Fourth, the active site of integrase is promiscu- 
ous in its selection of the hydroxyl group used as a 
nucleophile for the DNA transesterification reac- 
tion. A 2' hydroxyl can substitute for the 3' hydroxyl 
ordinarily involved in disintegration. 
Analysis of the features that distinguish DNAs that 
can be used as analogues of viral DNA ends or as 
target DNAs in an in vitro integration reaction, 
carried out by Dr. Dotan and graduate student Timo- 
thy Heuer, has provided strong evidence for the ex- 
istence of separate binding sites for viral DNA ends 
and target DNA. This observation is being pursued 
by 1) chemically crosslinking specific substrate 
DNAs to the enzyme, so as to identify the amino 
acids that make close contacts with key features of 
the substrates, 2) using selection from a degenerate 
sequence pool to identify optimal sequences for rec- 
ognition by each binding site, and 3) screening for 
integrase mutants that have defects in recognition of 
either viral or target DNA substrates. 
Work on HIV integrase, described above, was also 
supported by a grant from the National Institutes of 
Health. 
Because the poor solubility of HIV integrase has 
frustrated attempts to obtain good crystals for x-ray 
crystallography. Dr. Brown and research assistant 
Elizabeth Kubalek have begun an effort to obtain 
two-dimensional crystals for structural analysis by 
electron diffraction. This approach is much less de- 
manding of protein solubility than is ordinary three- 
dimensional crystallography. Moreover, it can allow 
ready examination of interactions between the pro- 
tein and its ligands, since one face of the protein 
is freely accessible to solution. Essentially, the 
method requires that proteins be bound to the head 
groups of a lipid monolayer at an air-water interface. 
Once concentrated in this quasi-two-dimensional 
space, they are allowed to crystallize, a process that 
may require only minutes and typically no more 
than a few hours. 
The first requirement for two-dimensional crystal- 
lography is that the protein of interest be capable of 
binding to the lipid monolayer. Therefore, to en- 
hance the reliability and general applicability of the 
method, the team has synthesized novel lipids with 
nitrilotriacetic acid (nickel-chelating) head groups 
and glutathione head groups. These lipids should be 
useful as general reagents for binding histidine- 
tagged and glutathione-sulfotransferase (GST)- 
fusion proteins, respectively, to a monolayer at 
an air-water or other interface. Crystallography ex- 
periments will be an important focus of work in the 
next year. If initial efforts are successful, this 
method will be used not only for initial structure 
determination but also to examine interactions with 
DNA substrates and other ligands. 
New Methods for Linkage Mapping 
in Complex Genomes 
Dr. Brown and postdoctoral fellow Dr. Stanley 
Nelson have developed a new method for linkage 
mapping, termed genomic mismatch scanning. It 
should allow widespread application of highly effi- 
cient afifected-relative-pair linkage mapping meth- 
ods. The approach uses specialized enzymes that 
can recognize differences between two DNA se- 
quences, to map all the regions of identity-by- 
descent between two relatives in a single proce- 
dure. In the past year the procedure has been tested 
using baker's yeast (Saccharomyces cerevisiae) as a 
model system, with highly successful results. Pre- 
liminary experiments with human DNA indicate that 
genomic mismatch scanning can be applied to link- 
age mapping in humans and other higher organisms. 
Continuing work on this project is now focusing on 
automating the procedure and adapting it to map- 
ping human genes. The genomic mismatch scanning 
method will then be applied to very large human 
populations, with the goal of mapping genes govern- 
ing disease susceptibility and diverse other complex 
and quantitative traits. The project described above 
was supported by a grant from the National Insti- 
tutes of Health. 
GENETICS 161 
