DNA SEQUENCE ANALYSIS BY HYBRIDIZATION              
   
   Although it sounds rather mysterious, according to Bob Foote of the
   Biology Division, hybridization is simply the bonding of complementary
   pieces of DNA. The technique of sequencing by hybridization employs an
   array (a geometric arrangement) of short DNA segments of known sequence,
   representing all possible sequences of a given length. An unknown
   sequence of DNA is broken into pieces, which are then added to the
   array. Theoretically, the pieces will "stick," or bond, to the
   complementary segments in the array. For example, because the DNA base
   A bonds with T and C bonds with G, the sequence A-C-T-G bonds with
   T-G-A-C.     
   
   Foote and his colleagues, including Rick Sachleben of the Chemistry
   Division, have proposed a way of making arrays of DNA segments to carry
   out many hybridizations simultaneously. "We had the idea of doing this
   on a microscopic scale with an array that measures 1 to 2 cm on a side,"
   says Foote. "Considering that there are four different DNA bases and
   that we wanted to work with 8-mers (8-base-long pieces of DNA), we have
   48, or 65,536, possible combinations of bases.
   
   "If we tried to synthesize these combinations one at a time, it would
   take about half a million steps," says Foote. "Using photolithographic
   methods, the same process used to make microchips, we can synthesize
   them all in 32 steps."           
   
   To accomplish this feat, strings of bases are built up in stages by
   treating the array with DNA bases containing a photolabile
   (light-sensitive) blocking group. The bases bond to the array, and the
   blocking group prevents further reaction until it is removed from
   selected sections by exposing them to light. Then a second base is
   applied to the array, bonding only where the blocking group is absent.
   This process is repeated 32 times, producing a 256 x 256 array of bases
   containing all possible 8-mer sequences.                   
   
   Once this process of synthesizing arrays has been mastered, Foote and
   his colleagues will label pieces of an unknown DNA sequence with
   radioactive or fluorescent tags and add them to an array. If everything
   works as it should, they will bond with their complementary sequences.
   The radiation or fluorescence of the unknown pieces bonded to the array
   makes possible the location and identification of the complementary
   sequence to which the unknown sequence bonds.                    
   
   "This process is not perfect," says Foote. "Errors can occur in
   hybridization. For instance, the DNA base A could be paired with C
   instead of T. Depending on where the mismatch occurs, it may not be
   enough to disrupt the double-stranded structure of the DNA. However,
   most mismatches can be eliminated by doing this under stringent
   conditions, and the redundancy of this method allows a certain degree of
   error in the individual hybridizations."                   
   
   When the hybridization is complete and the sequence of the unknown DNA
   segments has been determined, computational techniques are used to put
   these pieces back together to determine the sequence of the entire
   unknown DNA segment.             
   
   "Originally we were going to develop the techniques to do this
   ourselves," Foote says, "but we found out that researchers at Argonne
   National Laboratory, who originated much of the theory behind this type
   of research, have developed some computer algorithms for determining the
   overall sequence. We've maintained contact with them for two years, and,
   when we can do some actual hybridization with our arrays, we may be
   doing something more formal with them."             
   
   For the near future, Foote's main focus will be on refining the
   chemistry of synthesizing the arrays to improve the efficiency of these
   steps. "Once we get the chemistry done," says Foote, "we'll be testing
   the accuracy of hybridization on these small arrays and working with
   Jack Davidson of the Instrumentation and Controls Division and others at
   ORNL on detection methods.
   

   ------------------------------------------------------------------------
   
   Please send us your comments.
   
   Date Posted:  1/10/94  (ktb)