PFRMAT AL 
TARGET T0074 
AUTHOR 5529-3140-9255 
REMARK Prediction team BENNER-COHEN has two group leaders, 
REMARK but we will consistently use Steven Benner's predictor 
REMARK number to avoid confusion concerning a second team 
REMARK connected with Fred Cohen (called Cohen, Fred). 
REMARK (Fred Cohen's number as a group leader is 6140-7890-6093). 
REMARK Prediction team members: D.L.Gerloff, G.Cannarozzi, 
REMARK M.Joachimiak, F.E.Cohen & S.A.Benner. 
REMARK 
REMARK This threading alignment prediction is not accompanied by 
REMARK a separate secondary structure submission. 
METHOD 
METHOD The PROSITE motif for EF-hand proteins (Bairoch/Cox/Kretsinger) 
METHOD revealed the most probable fold for this target. Further, we used 
METHOD the UCLA-DOE Fold Recognition server (Eisenberg/Fischer/Rice) which 
METHOD returned a number of different structures of this fold as top ranked, 
METHOD some above significance threshold. We also ran ProCyon (Sippl), which 
METHOD did not seem to return a conclusive match in our hands; finally, we 
METHOD looked at the predicted secondary structures by the PredictProtein- 
METHOD Server (PHD, Rost) and by a new method by Chandonia in our laboratory, 
METHOD besides our own. The latter two methods in the list were applied on 
METHOD a multiple sequence alignment including 28 sequences, uniting homologous 
METHOD sequence fragments to all three EH-domains found in EP15_HUMAN. The 
METHOD width and balanced distribution of this master alignment has certainly 
METHOD had an effect on our prediction.  
METHOD  
METHOD Departing from the assumption that T0074 folds into a EF-hand-like 
METHOD structure still leaves open a wide variety in 3D-structures, as the 
METHOD members of the fold/motif family display variation in the relative 
METHOD orientation of the two EF-hands as well as in the dihedral angle  
METHOD between the two helices composing one EF-hand, and in the mode of 
METHOD calcium-binding per se. Therefore, we attempted to find the most 
METHOD suitable candidates for parent structures by a manual approach that 
METHOD could be summarized as a "knowledge-enhanced profile threading" which  
METHOD is explained below. 
METHOD 
METHOD We considered 10 representative structures covering the (6) families 
METHOD of EF-hand-like proteins classified in the SCOP database (Murzin/ 
METHOD Brenner/Chothia), and studied which positions in the corresponding  
METHOD HSSP-alignments (Sander) seemed to play important structural roles 
METHOD in each protein family. Notably, we tried to (a) distinguish between 
METHOD positions that appeared to be conserved for structural and functional 
METHOD reasons, respectively, and (b) to also identify structurally important 
METHOD positions which were not absolutely conserved in the alignment, simply 
METHOD by looking at the structures. Then, applied a manual equivalent to a  
METHOD profile-based threading method with primary emphasis on patterns of  
METHOD variation and conservation of residues in the target and reference 
METHOD multiple sequence alignments. However, we attributed very much higher 
METHOD weight to matching the positions we had identified/predicted as 
METHOD structurally essential in this procedure. In short, this would include 
METHOD to ask "WHY" is a position conserved in the reference sequence alignment, 
METHOD and to avoid unexplained conserved positions in the target sequence 
METHOD alignment. In the first EF-hand, for example, we noted the uncommon 
METHOD conservation of Gly 144 within the EF-hand motif region. One possible 
METHOD explanation for conservation of this Gly (or Ala) could be to make 
METHOD space, or a sharp turn, because another part of the protein is near, 
METHOD and rigid. We found an example for this in the 3nd (pseudo-)EF-hand of 
METHOD 1scm_C, and have modeled the first EF-hand of T0074 based on this 
METHOD structure. The fix points in all EF-hands is the beta-bridge, and 
METHOD this will be the case for T0074 as well. However, we had difficulties 
METHOD to propose an unambiguous alignment for the 2nd EF-hand in T0074 
METHOD with the 1scm_C coordinates, because of the mandatory short connection 
METHOD between the two modules (T0074 158-163 approximately). To generate  
METHOD a structure for the entire T0074 based on 1scm would require moving 
METHOD the first helix in the second EF-hand of T0074, or, rather, tilting 
METHOD it differently. As the second EF-hand motif in T0074 seems nicely 
METHOD intact, however, we expect a "classic" calcium-binding site, like 
METHOD the one found in 4icb, for example. 
METHOD 
METHOD Thus, we submit two different models, one pieced together using 
METHOD the 3rd EF-hand of 1scm_C and the 2nd EF-hand of 4icb (the relative 
METHOD orientation would be determined by conserving the beta bridge 
METHOD between T0074 residues 146 and 180), the other based on the 3rd 
METHOD and 4th EF-and-pseudo-EF-hands in 1scm_C but omitting the sequence 
METHOD fragments corresponding to the first helix in the 2nd EF-hand of T0074. 
METHOD 
METHOD Through this prediction, we hope to provide an appreciation for 
METHOD discriminating between globally similar parent structures that 
METHOD do, however, differ in the relative orientation of individual  
METHOD secondary structural elements (segment contacts). While the applied 
METHOD method is entirely experimental at this point, we strongly believe 
METHOD that manual, or automated, assignment of "roles" (e.g. hydrophobic 
METHOD anchor") to particular positions in the fold would prove useful 
METHOD with further development. For example, it is our understanding that 
METHOD the profficiency demonstrated by A.Murzin's submissions at CASP2 
METHOD can, in part, be attributed to such "knowledge-based" assignments. 
METHOD 
REMARK  
REMARK  
REMARK WE HAVE GENERATED A COORDINATE MODEL FOR T0074 ALSO (WHICH IS MODEL1). 
REMARK THE COORDINATE MODEL WILL BE A COMPOSITE OF 2 EF-HANDS TAKEN FROM THE 
REMARK PDB STRUCTURES 1SCM_C(1.EF-HAND) AND 4ICB(2.EF-HAND). MODEL 2 IS THUS 
REMARK THE INTENDED SAME MODEL IN AL SUBMISSION FORMAT (WE WANTED TO AVOID 
REMARK THE FRAGMENTATION OF OUR MODEL, PROVIDED AL FORMAT WILL BE TRANSLATED 
REMARK TO TS FORMAT AND ACTUALLY REPLACE THE SAME MODEL NUMBER!) MODEL 3 
REMARK IS BASED ONLY ON ONE PARENT STRUCTURE, WITH ONE OF THE HELICES 
REMARK MISSING AS WE WOULD HAVE HAD TO REORIENT IT (SEE ABOVE). 
REMARK IN BRIEF: MODEL 1 (TS) IS WHAT WE THOUGHT MODEL 2 (AL) WOULD 
REMARK LOOK LIKE IF TRANSLATED INTO COORDINATES *AS A WHOLE*. 
REMARK  
REMARK OUR TRUE FAVORITE IS MODEL 1. 
REMARK 
MODEL  3 
PARENT 1scm_C 
K   129   F   83 
A   130   A   84 
K   131   D   85 
Y   132   Y   86 
D   133   M   87 
A   134   E   88 
I   135   A   89 
F   136   F   90 
D   137   K   91 
S   138   T   92 
L   139   F   93 
S   140   D   94 
N   143   Q   98 
G   144   G   99 
F   145   F   100 
L   146   I   101 
S   147   S   102 
G   148   G   103 
D   149   A   104 
K   150   E   105 
V   151   L   106 
K   152   R   107 
P   153   H   108 
V   154   V   109 
L   155   L   110 
L   156   T   111 
N   157   A   112 
S   158   L   113 
I   174   E   133 
D   175   D   134 
H   176   L   135 
D   177   E   136 
G   178   G   137 
M   179   N   138 
L   180   V   139 
D   181   K   140 
R   182   Y   141 
D   183   E   142 
E   184   D   143 
F   185   F   144 
A   186   V   145 
V   187   K   146 
A   188   K   147 
M   189   V   148 
F   190   M   149 
L   191   A   150 
V   192   G   151 
TER 
END 
