known for some time that Ul snRNP interacts with 
the 5' splice site, this molecule brings together the 
two ends of the intron. 
To verify the ability of a single Ul snRNP to inter- 
act with both the 5' and 3' ends of the intron, the 
laboratory utilized a pre-mRNA splicing substrate 
cleaved by RNase H into 5' and 3' half-molecules. If 
the intron is cleaved after complex formation in nu- 
clear extract lacking ATP, both halves are immuno- 
precipitated with Ul antisera. However, when the 
RNA is cleaved before incubation with extract, only 
the 5' half-molecule is immunoprecipitated. Since 
complex assembly on an intact pre-mRNA enhances 
precipitation of the 3' half-molecule, colinearity of 
the splice sites must stabilize the association of Ul 
with the 3' end, implying that a single Ul bridges 
both splice sites in ATP-independent complexes. 
Surprisingly, the same results are obtained when the 
5' end of Ul snRNA is removed by RNase H diges- 
tion. This suggests that the ATP-dependent associa- 
tion between Ul snRNA with both the 5' and 3' 
splice sites does not require the Ul snRNA 5' end, 
even though this is the sequence that base pairs with 
the 5' splice site. Therefore the ATP-independent 
interaction between Ul and the splice sites is dis- 
tinct from the ATP-dependent association between 
Ul and the 5' splice site and must be mediated by a 
different region of Ul and/or additional factors. Ex- 
periments are in progress to characterize this earli- 
est step in splice site selection. 
The experiments outlined above indicate that 
while the 5' splice site, the branch point, and poly- 
pyrimidine tract are required early in the splicing 
pathway and are recognized by established splicing 
factors, the mode of recognition of the 3' splice site 
AG and the stage in the pathway at which this recog- 
nition occurs remain less understood. This labora- 
tory previously proposed that the 3' splice site in 
mammalian introns is identified and located by a 
scanning mechanism that searches for the first AG 
downstream from the branch point/polypyrimidine 
tract. During the past year experiments have been 
carried out to support and refine this model. In ad- 
dition, prompted by examples of various natural in- 
trons, the basis for the exception to the simple rule 
that the first AG downstream from the branch point 
is selected exclusively as the 3' splice site has been 
explored. The results demonstrate that, although 
such exceptions do exist, they are predictable and 
can easily be accommodated by simple refinements 
of the model. This behavior is reminiscent of the 
scanning model for translational initiation, which 
has also been able to accommodate exceptions with 
simple "rules to break the rule." In light of these 
findings, a refined scanning model can be summa- 
rized as follows: scanning initiates at the branch 
point and proceeds in a 3' direction to the first AG 
unless it is either so close to the branch point that 
recognition is very ineflBcient or it is hidden within 
a stem-loop structure. In either of these cases the 
first AG can be bypassed, and the next downstream 
AG located. Once an AG has been recognized, how- 
ever, the spliceosome can still "see" a limited 
stretch of downstream RNA. Within this "window," 
the most competitive AG is selected as the 3' splice 
site. 
Determinants of competitiveness include proxim- 
ity to the branch point: the more-proximal AG is 
usually more competitive, unless it is very close to 
the branch point, in which case steric effects can 
render it less competitive. The nucleotide preced- 
ing the AG also has a marked effect upon competi- 
tion. The hierarchy of competitiveness, CAG = UAG 
> AAG > GAG, closely follows the occupancy of this 
position in consensus compilations of 3' splice sites. 
Thus 3' splice site selection displays properties of 
both a scanning process and competition between 
AGs based on immediate sequence context. Compe- 
tition based on the preceding nucleotide and steric 
effects explains two features of consensus 3' splice 
site arrangements: the nucleotide preference at the 
— 3 position and the common 18-nt minimum sepa- 
ration between the branch point and the 3' splice 
site AG. 
Dr. Nadal Ginard is also Alexander S. Nadas 
Professor of Pediatrics and Professor of Cellular 
and Molecular Physiology at Harvard Medical 
School and Cardiologistin- Chief at the Children 's 
Hospital, Boston. 
Books and Chapters of Books 
Nadal-Ginard, B. 1991. Regulation of alternative 
splicing of contractile protein genes. In Frontiers 
in Muscle Research: Myogenesis, Muscle Con- 
traction, and Muscle Dystrophy (Ozawa, E., Ma- 
saki, T., and Nabeshima, Y., Eds.). New York: Ex- 
cerpta Medica, pp 151-165. (International 
Congress Ser. 942.) 
Nadal-Ginard, B., and Mahdavi, V. 1991 . Acellular 
and molecular approach to pediatric cardiology. 
In Nadas' Pediatric Cardiology (Fyler, D.C., 
Ed.). Philadelphia, PA: Hanley & Belfus, pp 747- 
759. 
Nadal-Ginard, B., and Mahdavi, V. 1991. General 
principles of cardiovascular cellular and molecu- 
lar biology. In Heart Disease: A Textbook of Car- 
diovascular Medicine (Qr2.\\nw3.\6., E., Ed.). Phila- 
delphia, PA: Saunders, pp 1602-1621. 
232 
