Wed Aug 12 13:14:00 PDT 1998 wu-blast finds nothing for t80 (top hit 3ts1=3ts1 2.05) double-blast finds nothing. So the "no known sequence homology" seems valid. This is a DNA-binding protein (3-methyladenine DNA glycosylase) and the crystal is a protein-DNA complex, so we need to find a DNA-binding site of some sort. t80.t98_6 finds fssp rep [12]msb[AB] -7.140 1rtm1 mannose-binding protein (coiled-coil + alpha/beta) 1ytt[AB] -7.140 1rtm1 [1234]1kmb[123] -6.870 1rtm1 1rtm[123] -6.860 1rtm1 1af[abd]1[123] -6.830 1rtm1 1bch[123] -6.820 1bcj[123] -6.130 1hm[rst] -4.390 1hmt fatty-acid binding protein (all-beta barrel) 2hmb -4.390 1hmt 2tir -4.390 1thx thioredoxin (alpha-beta-alpha) 1srx -4.370 1thx 1t7pB -4.120 1thx DNA polymerase Mutant thioredoxin DNA 1xo[ab] -4.120 1thx 2trx[AB] -4.120 1thx 1tho -4.100 1thx 1tul -4.080 1tul tlp20 (all-beta barrel) Of these, 1t7pB looks most interesting, though 1t7pB is just an alpha-beta-alpha sandwich tacked onto 1t7pA, which has the DNA-binding stuff. We could try splitting 1t7pA near where 1t7pB occurs, and insert 1t7pB in the gap---that would be a feasible target to align to. There is a natural break at 288 PRIKTPKE YVAGAPYT, so that might be the place to poke. Using the library models, the best hits are 7aatA -4.860 1ajsA 1akaA -4.820 1ajsA 1ajsA -4.070 1ajsA 1kst -3.790 1kst 1afwA -3.710 1afwA 1fvl -3.520 1fvl None of these seem to be DNA-binding. Using the raw fssp alignments with viterbi scoring, the best hits are 1ortA -6.470 1pty -5.300 1dkzA -5.160 substrate bidning domain of dnak 1rpo -5.050 rop (cole1 repressor of primer) 1xbrA -4.870 t protein fragment DNA 1djxB -4.850 2fmr -4.180 1ppt -4.140 2gf1 -4.020 Three of these look promising as DNA-binding. 1dkzA and 1rpo have FSSP Z-score of 5, and only 8% residue id. Hmm---1rpo is just a pair of helices and 1dkzA doesn't show DNA binding. 1dkzA does have a similarity to 1t7pA (Z=4.6), so we do have some confirmation of the 1t7pA+1t7pB possibility. 13 August 1998 I made a chimeric sequence putting 1t7pB into the 1t7pA sequence roughly in the place where it fits in three dimensions (there is a gap there in the PDB file for the 1t7pA chain), and I inserted 1t7pB right after the missing residues. The resulting chain is 1t7pABA/1t7pABA.seq. I built target98 alignments using this chimeric chain and tried aligning t80 to it. The first part of t80 aligns to 1t7pB and the rest aligns (rather weakly) to the remainder of 1t7pA. It might be better just to chop off the part of t80 that doesn't align to 1t7pB and see what it matches on its own. Here are the best non-self scores from the alignments currently tried: 1rtm1/t80-1rtm1-vit 1rtm1 149 -7.40 -8.02 1t7pB/t80-1t7pB-vit 1t7pB 105 -5.52 -6.87 1kst/1kst-t80-vit T0080 219 -4.69 -5.93 1kst/1kst-t80-global T0080 219 -5.37 -5.62 1kst/1kst-t80-post T0080 219 -5.37 -5.62 1tul/t80-1tul-vit 1tul 102 -4.04 -5.14 1pty/t80-1pty-vit 1pty 300 -4.28 -5.12 1ajsA/1ajsA-t80-vit T0080 219 -4.04 -4.43 1t7pABA/t80-1t7pABA-vit 1t7pABA 806 -3.48 -3.56 1t7pABA/t80-1t7pABA-global 1t7pABA 806 -5.04 -3.46 1t7pABA/t80-1t7pABA-post 1t7pABA 806 -5.04 -3.46 1pty/t80-1pty-global 1pty 300 -0.59 -3.35 1pty/t80-1pty-post 1pty 300 -0.59 -3.35 1t7pA/t80-1t7pA-global 1t7pA 662 -4.50 -2.95 1t7pA/t80-1t7pA-post 1t7pA 662 -4.50 -2.95 1ortA/1ortA-t80-vit T0080 219 -1.91 -2.65 1t7pB/1t7pB-t80-vit T0080 219 -0.43 -2.35 1t7pB/1t7pB-t80-global T0080 219 -0.39 -2.08 1t7pB/1t7pB-t80-post T0080 219 -0.39 -2.08 1xbrA/1xbrA-t80-vit T0080 219 -4.02 -1.94 1thx/1thx-t80-vit T0080 219 -0.39 -1.91 1dkzA/t80-1dkzA-global 1dkzA 215 -0.52 -1.83 1dkzA/t80-1dkzA-post 1dkzA 215 -0.52 -1.83 1rtm1/t80-1rtm1-global 1rtm1 149 0.85 -1.79 1rtm1/t80-1rtm1-post 1rtm1 149 0.85 -1.79 1ortA/t80-1ortA-global 1ortA 335 -0.69 -1.77 1ortA/t80-1ortA-post 1ortA 335 -0.69 -1.77 1thx/1thx-t80-global T0080 219 0.12 -1.59 1thx/1thx-t80-post T0080 219 0.12 -1.59 1dkzA/t80-1dkzA-vit 1dkzA 215 -1.34 -1.58 1pty/1pty-t80-vit T0080 219 -1.06 -1.16 1t7pA/t80-1t7pA-vit 1t7pA 662 -1.14 -1.03 1t7pA/1t7pA-t80-vit T0080 219 -1.60 -0.39 1ortA/t80-1ortA-vit 1ortA 335 -0.47 -0.09 Note: 1kst has several cystine bridges, so any alignment with 1kst should be careful not to disrupt them. 27 August 1998 Christian There are a number of glycosylases whose structures have been determined. Since evidence indicates that glycosylases tend to have a broad substrate specificity, these structures should be noted: 1mug 1mpg 1akz 1lau 1ud[hig] 1vas 1abk 2end 1en[kji] Okay, none of these have a score better than about -0.2 in any of the rdb files. They are even, like 1mpg, DNA glycosidae II 's. Nothing in the 1mpgA FSSP family score well either. Just for kicks I built the 1mpgA directory. See the t80 global alignment. The 2nd-best FSSP hit, 1pty, has 1mug as a family member at Z = 2.9. t80-1pty.cbarrett1.a2m is an alignment that we could consider in a pinch. Without having much to go on, 1rpo, 1dkzA, and 1xbrA don't look too promising. But I guess for something this remote, nothing will. If the hOgg1 glycosylase/lyase protein (genbank accession 1890182) is homologous to T80 (3MG_HUMAN), then PMID:9331411 has a good bit of structural information that would be very handy. Reference for the article is Chem Biol 1997 Sep;4(9):693-702 "The critical active-site amine of the human 8-oxoguanine DNA glycosylase, hOgg1: direct identification, ablation and chemical reconstitution." 30 August 1998 Christian The second highest FSSP hit is 1pty at -5.3. It is also a hydrolyzing enzyme and has a structure in it's FSSP family who's function is very similar to what we are looking for (1mugA, g:tU SPECIFIC DNA GLYCOSYLASE. More on 1mugA in a second.) The alignment is t80-1pty.cbarrett2.a2m and I think is "better" than cbarrett1. 1mugA is the most promising alignment so far, as it is to a protein of the same function and has lots of identity. I'm not sure why it scored so horribly. See 1mugA-t80.cbarrett1.a2m. 1akz is an FSSP family member of very similar function, but the alignment is not promising and neither are the scores. Without having looked at the 17tp alignments, my money is currently on 1mugA. Mon Aug 31 11:02:45 PDT 1998 Kevin Karplus One strong possibility I considered is 1t7pB up to MYFCM, then switching to 1mugA for the rest. Unfortunately, this omits an interior strand of 1mugA's beta sheet, if we use Christian's alignment. However, 1t7pB/t80-1t7pB-hand1 1mugA/1mugA-t80-hand2 seem to be compatible.