WEBVTT 00:00:00.000 --> 00:00:00.610 align:middle line:90% 00:00:00.610 --> 00:00:02.440 align:middle line:84% The following content is provided 00:00:02.440 --> 00:00:06.340 align:middle line:84% by MIT OpenCourseWare under a Creative Commons license. 00:00:06.340 --> 00:00:08.470 align:middle line:84% Additional information about our license 00:00:08.470 --> 00:00:10.870 align:middle line:84% and MIT OpenCourseWare in general 00:00:10.870 --> 00:00:12.730 align:middle line:90% is available at ocw.mit.edu. 00:00:12.730 --> 00:00:17.170 align:middle line:90% 00:00:17.170 --> 00:00:20.410 align:middle line:90% GEORGE CHURCH: OK, ready. 00:00:20.410 --> 00:00:22.090 align:middle line:90% OK. 00:00:22.090 --> 00:00:29.238 align:middle line:84% So welcome to the third phase of our omics discussion. 00:00:29.238 --> 00:00:31.030 align:middle line:84% Someone pointed out that I should point out 00:00:31.030 --> 00:00:35.110 align:middle line:84% [? this shirt ?] says omics not comics. 00:00:35.110 --> 00:00:39.340 align:middle line:84% So we've covered genomics and last time transcriptomics, 00:00:39.340 --> 00:00:43.000 align:middle line:84% and today we introduce a very important all-inclusive subject 00:00:43.000 --> 00:00:45.160 align:middle line:90% of proteomics. 00:00:45.160 --> 00:00:48.820 align:middle line:84% We'll connect it to last week's through the vehicle 00:00:48.820 --> 00:00:53.800 align:middle line:84% of focusing on motifs that are involved 00:00:53.800 --> 00:00:58.450 align:middle line:84% in protein interactions with the two nucleic acid 00:00:58.450 --> 00:01:00.610 align:middle line:90% macromolecules. 00:01:00.610 --> 00:01:04.540 align:middle line:84% So we're going to be covering, just as we introduced 00:01:04.540 --> 00:01:08.650 align:middle line:84% RNA omics with RNA structure, we're 00:01:08.650 --> 00:01:11.260 align:middle line:84% going to spend this entire class talking about protein three 00:01:11.260 --> 00:01:14.410 align:middle line:84% dimensional structure, how you get at it experimentally 00:01:14.410 --> 00:01:17.050 align:middle line:84% and computationally and its implications 00:01:17.050 --> 00:01:22.060 align:middle line:84% for the binding of small molecules such as drugs. 00:01:22.060 --> 00:01:27.550 align:middle line:84% We will in short order get to the scary pumpkin-like 00:01:27.550 --> 00:01:29.450 align:middle line:90% molecule. 00:01:29.450 --> 00:01:38.980 align:middle line:84% So the connection to last week was this diagram 00:01:38.980 --> 00:01:43.360 align:middle line:84% showing palindromicity in three cases and a direct repeat 00:01:43.360 --> 00:01:45.880 align:middle line:90% in the fourth case. 00:01:45.880 --> 00:01:48.490 align:middle line:84% And I offered that this might reflect 00:01:48.490 --> 00:01:51.310 align:middle line:84% the symmetry of the proteins-- of the three 00:01:51.310 --> 00:01:53.320 align:middle line:84% dimensional structure of the proteins 00:01:53.320 --> 00:01:55.930 align:middle line:84% and the three dimensional structure of the nucleic acid 00:01:55.930 --> 00:01:57.910 align:middle line:84% and these symmetry elements would align. 00:01:57.910 --> 00:02:00.220 align:middle line:84% Now in order to introduce these symmetry elements 00:02:00.220 --> 00:02:02.370 align:middle line:84% and the possibility of having codes 00:02:02.370 --> 00:02:05.140 align:middle line:84% that you can at least program, even 00:02:05.140 --> 00:02:08.065 align:middle line:84% if they may have been tinkered about during an evolution, 00:02:08.065 --> 00:02:09.880 align:middle line:84% the question is to what extent can we 00:02:09.880 --> 00:02:13.920 align:middle line:84% get our hands on these kind of protein and nucleic acid motifs 00:02:13.920 --> 00:02:15.370 align:middle line:90% that interact. 00:02:15.370 --> 00:02:19.450 align:middle line:84% In order to get at this issue of where there is a code-- 00:02:19.450 --> 00:02:22.870 align:middle line:84% and I just take this as one of the ways of dealing 00:02:22.870 --> 00:02:25.840 align:middle line:84% with the incredible complexity of proteins 00:02:25.840 --> 00:02:29.590 align:middle line:84% is to give this a theme that connects it to the last class 00:02:29.590 --> 00:02:32.560 align:middle line:84% and connects, I think, to many of the sentiments of people 00:02:32.560 --> 00:02:34.900 align:middle line:84% in this sort of audience interested 00:02:34.900 --> 00:02:39.820 align:middle line:84% in computational biology is ways of having simple codes. 00:02:39.820 --> 00:02:41.980 align:middle line:84% And of course, the way of one way 00:02:41.980 --> 00:02:44.050 align:middle line:84% of breaking up proteins and thinking about them 00:02:44.050 --> 00:02:49.060 align:middle line:84% is these ABCs-- the alpha helix, the beta sheet, and the coil. 00:02:49.060 --> 00:02:51.700 align:middle line:84% Each of these can be characterized by the hydrogen 00:02:51.700 --> 00:02:53.230 align:middle line:90% bonds that hold it together. 00:02:53.230 --> 00:02:57.460 align:middle line:84% The weak bonds between the hydrogens, the nitrogens, 00:02:57.460 --> 00:02:59.103 align:middle line:90% and the oxygens. 00:02:59.103 --> 00:03:00.520 align:middle line:84% And the alpha helix, these are all 00:03:00.520 --> 00:03:03.280 align:middle line:84% kept within the helix with a repeat 00:03:03.280 --> 00:03:07.090 align:middle line:90% of 3.6 residues per turn. 00:03:07.090 --> 00:03:12.880 align:middle line:84% In the beta sheet, they tend to have a longer, straighter 00:03:12.880 --> 00:03:17.260 align:middle line:84% chains where there are unpaired hydrogen bonds inevitably 00:03:17.260 --> 00:03:19.660 align:middle line:84% until you form enough chains to form 00:03:19.660 --> 00:03:22.480 align:middle line:84% a cyclic structure while the alpha helix is immediately 00:03:22.480 --> 00:03:25.103 align:middle line:90% helical. 00:03:25.103 --> 00:03:26.770 align:middle line:84% There are many different types of coils. 00:03:26.770 --> 00:03:28.770 align:middle line:84% It's a catch-all phrase that includes everything 00:03:28.770 --> 00:03:33.340 align:middle line:84% except alpha and beta, but a particularly well-formed type 00:03:33.340 --> 00:03:38.530 align:middle line:84% of coil has its own nomenclature and parameters is the turn, 00:03:38.530 --> 00:03:44.110 align:middle line:84% and the turn is illustrated here at the end of a beta sheet. 00:03:44.110 --> 00:03:45.580 align:middle line:84% That beta sheet basically is going 00:03:45.580 --> 00:03:47.920 align:middle line:84% in on the lower right hand corner 00:03:47.920 --> 00:03:49.300 align:middle line:90% in the direction of the arrow. 00:03:49.300 --> 00:03:51.820 align:middle line:84% The arrows typically point from the n terminus 00:03:51.820 --> 00:03:55.000 align:middle line:84% to the c terminus just as in nucleic acids 00:03:55.000 --> 00:03:56.680 align:middle line:84% is from five prime to three prime. 00:03:56.680 --> 00:03:58.260 align:middle line:84% And here it goes in the arrow, it 00:03:58.260 --> 00:04:02.080 align:middle line:84% turns around very tightly, and goes back out again. 00:04:02.080 --> 00:04:06.100 align:middle line:84% OK, now how can we use these basic motifs? 00:04:06.100 --> 00:04:10.570 align:middle line:84% These are the smallest meaningful units of protein 00:04:10.570 --> 00:04:11.750 align:middle line:90% three dimensional structure. 00:04:11.750 --> 00:04:13.540 align:middle line:84% How can we use these to recognize 00:04:13.540 --> 00:04:17.480 align:middle line:84% other macromolecules, other proteins and nucleic acids? 00:04:17.480 --> 00:04:20.149 align:middle line:84% So let's connect this to the motifs of last class. 00:04:20.149 --> 00:04:23.020 align:middle line:84% We have these motifs that we could find, 00:04:23.020 --> 00:04:26.200 align:middle line:84% weight matrices for them by aligning lots of sequences. 00:04:26.200 --> 00:04:27.720 align:middle line:84% Now instead of aligning sequences, 00:04:27.720 --> 00:04:33.880 align:middle line:84% let's see what we can do by mutating both the protein 00:04:33.880 --> 00:04:36.070 align:middle line:90% part and the nucleic acid part. 00:04:36.070 --> 00:04:39.340 align:middle line:84% And in order to do this, just as an illustration, 00:04:39.340 --> 00:04:40.900 align:middle line:84% let's say we have three zinc fingers. 00:04:40.900 --> 00:04:47.770 align:middle line:84% This is a real human and mouse DNA binding protein 00:04:47.770 --> 00:04:49.310 align:middle line:84% with three zinc fingers in a row. 00:04:49.310 --> 00:04:51.910 align:middle line:84% So this is an example of the direct repeat or tandem repeat 00:04:51.910 --> 00:04:53.020 align:middle line:90% type of symmetry. 00:04:53.020 --> 00:04:55.780 align:middle line:84% Remember, there's the direct repeat and the inverted repeat. 00:04:55.780 --> 00:04:58.390 align:middle line:84% And in this tandem repeat, let's anchor the two ends 00:04:58.390 --> 00:04:59.710 align:middle line:90% and change the middle. 00:04:59.710 --> 00:05:01.630 align:middle line:84% Make every possible peptide sequence 00:05:01.630 --> 00:05:06.160 align:middle line:84% in the middle or randomly sample the vast space that 00:05:06.160 --> 00:05:13.060 align:middle line:84% might occur in changing a few, say, six or more 00:05:13.060 --> 00:05:14.680 align:middle line:90% key amino acids. 00:05:14.680 --> 00:05:18.130 align:middle line:84% And then we know from the three dimensional structure 00:05:18.130 --> 00:05:21.430 align:middle line:84% that it interacts mainly with the middle three nucleotides, 00:05:21.430 --> 00:05:22.930 align:middle line:84% so let's change those middle three 00:05:22.930 --> 00:05:26.200 align:middle line:84% nucleotides to every possible trinucleotide sequence 00:05:26.200 --> 00:05:32.020 align:middle line:84% and see quantitatively how much that the different sequences 00:05:32.020 --> 00:05:34.450 align:middle line:84% in the protein affect the different sequences 00:05:34.450 --> 00:05:36.050 align:middle line:90% in the nucleic acid. 00:05:36.050 --> 00:05:39.477 align:middle line:84% So this is not going to be by staring at long sequence 00:05:39.477 --> 00:05:41.810 align:middle line:84% alignments where we're going to get the weight matrices. 00:05:41.810 --> 00:05:44.290 align:middle line:84% We can get them by actual experimental measures 00:05:44.290 --> 00:05:47.233 align:middle line:90% of the binding in vitro. 00:05:47.233 --> 00:05:49.150 align:middle line:84% And so what happens when you do this exercise, 00:05:49.150 --> 00:05:51.700 align:middle line:84% the wild type, now this-- the wild type sequence 00:05:51.700 --> 00:05:55.210 align:middle line:84% is something that may want to recognize 00:05:55.210 --> 00:05:56.260 align:middle line:90% a family of sequences. 00:05:56.260 --> 00:05:59.080 align:middle line:84% We don't know exactly what the wild type sequence 00:05:59.080 --> 00:06:01.540 align:middle line:84% is for this particular DNA binding protein, this zinc 00:06:01.540 --> 00:06:03.560 align:middle line:90% finger from human and mouse. 00:06:03.560 --> 00:06:05.770 align:middle line:84% But the subsequence of the business end, 00:06:05.770 --> 00:06:09.490 align:middle line:84% the amino acid subsequence of that recognition alpha helix 00:06:09.490 --> 00:06:11.950 align:middle line:84% that's binding into the major groove DNA 00:06:11.950 --> 00:06:15.850 align:middle line:84% is shown in the upper left, [? RSVHLTT. ?] 00:06:15.850 --> 00:06:18.268 align:middle line:84% And the sequence that it mainly binds to-- remember this 00:06:18.268 --> 00:06:19.060 align:middle line:90% is a weight matrix. 00:06:19.060 --> 00:06:22.450 align:middle line:84% It's not a consensus sequence, this TGG. 00:06:22.450 --> 00:06:24.850 align:middle line:84% It obviously recognizes a variety of other sequences. 00:06:24.850 --> 00:06:27.340 align:middle line:84% So remember, there's about three nucleotides 00:06:27.340 --> 00:06:30.850 align:middle line:84% on either side of it that the other two zinc fingers bind to. 00:06:30.850 --> 00:06:35.197 align:middle line:84% When you try all 64 possible trinucleotides-- remember 00:06:35.197 --> 00:06:37.155 align:middle line:84% from the genetic code, this is pure coincidence 00:06:37.155 --> 00:06:39.700 align:middle line:84% that these are triplets the same as the genetic code. 00:06:39.700 --> 00:06:42.070 align:middle line:84% This just happens to be the amount of the chunk of DNA 00:06:42.070 --> 00:06:45.460 align:middle line:90% that a zinc finger will cover. 00:06:45.460 --> 00:06:47.740 align:middle line:84% But it's not coincidence that 4 to the third 00:06:47.740 --> 00:06:50.530 align:middle line:84% is still 64, just like the genetic code. 00:06:50.530 --> 00:06:54.940 align:middle line:84% And if you run through all possible nucleotide sequences 00:06:54.940 --> 00:06:58.750 align:middle line:84% for this wild type, you find the winner is TGG, 00:06:58.750 --> 00:07:00.670 align:middle line:84% and it has that particular binding constant. 00:07:00.670 --> 00:07:04.330 align:middle line:84% The binding constant is measured in the molarity, 00:07:04.330 --> 00:07:09.070 align:middle line:84% roughly where you get half maximal or equilibrium binding. 00:07:09.070 --> 00:07:12.970 align:middle line:84% That's 10 to the minus 9 moles per liter. 00:07:12.970 --> 00:07:16.990 align:middle line:84% You can now mutagenize the peptide and select for peptides 00:07:16.990 --> 00:07:18.970 align:middle line:90% that bind to GCC. 00:07:18.970 --> 00:07:21.470 align:middle line:84% Remember, the flanks are kept constant. 00:07:21.470 --> 00:07:24.250 align:middle line:84% You get two peptides, both bind to the GCC. 00:07:24.250 --> 00:07:29.830 align:middle line:84% They both give matrices very similar to this with very-- 00:07:29.830 --> 00:07:31.960 align:middle line:84% these are very high affinity binding constants 00:07:31.960 --> 00:07:33.220 align:middle line:90% just like wild type. 00:07:33.220 --> 00:07:39.570 align:middle line:84% You've essentially engineered by selection a new specificity 00:07:39.570 --> 00:07:41.700 align:middle line:84% and two different ways of getting it. 00:07:41.700 --> 00:07:44.040 align:middle line:84% If you now go for something radically different, 00:07:44.040 --> 00:07:45.540 align:middle line:90% now no GC-- 00:07:45.540 --> 00:07:49.140 align:middle line:84% the first one was high GC, the second was pure GC, 00:07:49.140 --> 00:07:51.090 align:middle line:84% and then the third one is pure AT. 00:07:51.090 --> 00:07:54.083 align:middle line:84% And you get another peptide sequence that binds to that, 00:07:54.083 --> 00:07:55.500 align:middle line:84% and you get another weight matrix. 00:07:55.500 --> 00:07:57.958 align:middle line:84% Now remember, this weight matrix is not sequence alignment. 00:07:57.958 --> 00:08:03.000 align:middle line:84% This is binding constants where the weighting of the 64 00:08:03.000 --> 00:08:06.030 align:middle line:84% different sequences is based on how much binding you 00:08:06.030 --> 00:08:07.800 align:middle line:90% get for each of the 64. 00:08:07.800 --> 00:08:11.550 align:middle line:84% And then for some sequences, all these different trinucleotides 00:08:11.550 --> 00:08:14.430 align:middle line:84% all result in a rather poor selection 00:08:14.430 --> 00:08:17.640 align:middle line:84% for any kind of peptide out of all the vast number of peptides 00:08:17.640 --> 00:08:18.180 align:middle line:90% you have. 00:08:18.180 --> 00:08:19.890 align:middle line:84% None of them do particularly well, 00:08:19.890 --> 00:08:23.590 align:middle line:84% and the result is a weight matrix here 00:08:23.590 --> 00:08:25.740 align:middle line:84% which has very little information content. 00:08:25.740 --> 00:08:27.593 align:middle line:90% So this is a way. 00:08:27.593 --> 00:08:29.010 align:middle line:84% It's not the only way, but this is 00:08:29.010 --> 00:08:33.059 align:middle line:84% a way of getting a really good empirical data set, which 00:08:33.059 --> 00:08:36.419 align:middle line:84% in principle, you can combine it with similar functions 00:08:36.419 --> 00:08:39.690 align:middle line:84% on the flanking ones, and you can dial up 00:08:39.690 --> 00:08:44.700 align:middle line:84% any sequence of a nucleic acid protein interaction, at least 00:08:44.700 --> 00:08:45.900 align:middle line:90% with this class of proteins. 00:08:45.900 --> 00:08:48.450 align:middle line:84% Others are a little more problematic. 00:08:48.450 --> 00:08:50.850 align:middle line:84% But you can see how this can generate a code even 00:08:50.850 --> 00:08:55.980 align:middle line:84% if the actual detailed amino acid nucleotide 00:08:55.980 --> 00:08:59.770 align:middle line:90% interaction is not so simple. 00:08:59.770 --> 00:09:02.010 align:middle line:84% So that those are the results of the study. 00:09:02.010 --> 00:09:04.020 align:middle line:84% Then I'll show you just a snapshot 00:09:04.020 --> 00:09:07.245 align:middle line:84% of a schematic of how the actual experiments are done. 00:09:07.245 --> 00:09:09.120 align:middle line:84% And then finally, I'll show you a little math 00:09:09.120 --> 00:09:12.330 align:middle line:84% behind how we got those apparent binding constants. 00:09:12.330 --> 00:09:16.470 align:middle line:84% Remember, in those lower binding constant means 00:09:16.470 --> 00:09:19.750 align:middle line:84% you can get the binding at a lower concentration, 00:09:19.750 --> 00:09:22.180 align:middle line:90% which means a stronger binding. 00:09:22.180 --> 00:09:23.670 align:middle line:84% So the way you do the binding here 00:09:23.670 --> 00:09:27.420 align:middle line:84% is you take a nucleic acid array, 00:09:27.420 --> 00:09:30.330 align:middle line:84% similar to the ones we've been talking about in last class. 00:09:30.330 --> 00:09:32.670 align:middle line:84% Instead of being single stranded, ready to bind 00:09:32.670 --> 00:09:34.800 align:middle line:84% fluorescent nucleic acid, it's double stranded, 00:09:34.800 --> 00:09:38.040 align:middle line:84% ready to bind a fluorescent protein complex. 00:09:38.040 --> 00:09:43.290 align:middle line:84% The protein complex in this case is a bacteriophage, 00:09:43.290 --> 00:09:46.320 align:middle line:84% which is displaying the three zinc fingers in red. 00:09:46.320 --> 00:09:51.180 align:middle line:84% The middle one is the one in the past slide was mutagenized. 00:09:51.180 --> 00:09:54.690 align:middle line:84% And similarly, the array is combinatoric-- 00:09:54.690 --> 00:09:57.540 align:middle line:84% every possible DNA sequence that you're interested in 00:09:57.540 --> 00:09:58.350 align:middle line:90% is present. 00:09:58.350 --> 00:10:04.470 align:middle line:84% And what you do is quantitate the fluorescence of the zinc 00:10:04.470 --> 00:10:08.670 align:middle line:84% finger protein indirectly by the binding 00:10:08.670 --> 00:10:12.660 align:middle line:84% of the covalently-attached phage to the antibodies, which 00:10:12.660 --> 00:10:15.097 align:middle line:84% are fluorescently labeled by [INAUDIBLE] fluorescence. 00:10:15.097 --> 00:10:17.840 align:middle line:90% 00:10:17.840 --> 00:10:21.410 align:middle line:84% The quantity-- how you relate the fluorescence, the more 00:10:21.410 --> 00:10:23.060 align:middle line:90% binding, the more fluorescence. 00:10:23.060 --> 00:10:25.490 align:middle line:84% But how you relate that to the binding constant 00:10:25.490 --> 00:10:28.340 align:middle line:84% we had in the previous slide is the subject 00:10:28.340 --> 00:10:31.070 align:middle line:90% of this slide number eight. 00:10:31.070 --> 00:10:35.660 align:middle line:84% Now we call this the apparent equilibrium association 00:10:35.660 --> 00:10:37.580 align:middle line:84% constant because these experiments, 00:10:37.580 --> 00:10:43.850 align:middle line:84% just like many binding in living cells is not at equilibrium. 00:10:43.850 --> 00:10:48.740 align:middle line:84% It's a dynamic process in the cell and in vitro. 00:10:48.740 --> 00:10:51.560 align:middle line:84% There are ways that you can measure the equilibrium 00:10:51.560 --> 00:10:54.680 align:middle line:84% constants, but what this is apparent in the sense 00:10:54.680 --> 00:10:57.500 align:middle line:84% that you need to wash off the excess fluorescence in order 00:10:57.500 --> 00:10:59.990 align:middle line:84% to detect the fairly low signal that you 00:10:59.990 --> 00:11:03.238 align:middle line:84% get from the specific binding, rather 00:11:03.238 --> 00:11:04.280 align:middle line:90% than having fluorescence. 00:11:04.280 --> 00:11:08.870 align:middle line:84% You bathe it with 10 to the sixford excess of fluorescence. 00:11:08.870 --> 00:11:10.610 align:middle line:84% And so as you're doing that wash, 00:11:10.610 --> 00:11:12.110 align:middle line:84% you're obviously not at equilibrium. 00:11:12.110 --> 00:11:17.060 align:middle line:84% In the end, you take a snapshot before you wash off everything. 00:11:17.060 --> 00:11:20.360 align:middle line:84% And so what you're measuring-- what you're basically measuring 00:11:20.360 --> 00:11:21.290 align:middle line:90% is-- 00:11:21.290 --> 00:11:23.390 align:middle line:84% attempting to measure is the equilibrium constant 00:11:23.390 --> 00:11:26.120 align:middle line:84% between protein P in the upper left 00:11:26.120 --> 00:11:29.900 align:middle line:84% here, DNA D, double stranded going 00:11:29.900 --> 00:11:35.210 align:middle line:84% to the product, which is this P.D associated biomolecular 00:11:35.210 --> 00:11:36.230 align:middle line:90% product. 00:11:36.230 --> 00:11:42.230 align:middle line:84% And this is basic physical chemistry and algebra 00:11:42.230 --> 00:11:44.600 align:middle line:84% so that-- so you rearrange that to get 00:11:44.600 --> 00:11:46.500 align:middle line:90% the association constant. 00:11:46.500 --> 00:11:47.570 align:middle line:90% That's what you want. 00:11:47.570 --> 00:11:49.910 align:middle line:84% And the fraction of DNA molecules with the protein 00:11:49.910 --> 00:11:52.910 align:middle line:90% bounds can be found from this. 00:11:52.910 --> 00:11:58.460 align:middle line:84% It's just by definition, the fraction DNA molecules protein 00:11:58.460 --> 00:11:59.810 align:middle line:90% bound is-- 00:11:59.810 --> 00:12:02.040 align:middle line:84% the protein bound is the numerator. 00:12:02.040 --> 00:12:04.610 align:middle line:84% The brackets mean concentration in moles per liter, 00:12:04.610 --> 00:12:06.110 align:middle line:90% just like we stated. 00:12:06.110 --> 00:12:09.980 align:middle line:84% And then divide the complex by the amount 00:12:09.980 --> 00:12:13.940 align:middle line:84% of the fraction, which is the total DNA, which 00:12:13.940 --> 00:12:16.740 align:middle line:84% is the DNA present in the complex and the DNA is free. 00:12:16.740 --> 00:12:19.340 align:middle line:90% That's the D plus PD. 00:12:19.340 --> 00:12:22.160 align:middle line:84% And then you just substitute in this definition 00:12:22.160 --> 00:12:23.870 align:middle line:90% of the association constant. 00:12:23.870 --> 00:12:29.460 align:middle line:84% The definition is product over reactants, 00:12:29.460 --> 00:12:31.720 align:middle line:84% and then you get this intermediate term. 00:12:31.720 --> 00:12:34.770 align:middle line:84% Which now you cancel out all the concentrations of these, 00:12:34.770 --> 00:12:37.690 align:middle line:84% and then you end up with the last term on the far right, 00:12:37.690 --> 00:12:41.760 align:middle line:84% which is where if you hold the protein-- if the protein 00:12:41.760 --> 00:12:44.850 align:middle line:84% concentration is known, then the whole thing is 00:12:44.850 --> 00:12:47.010 align:middle line:84% constant except for the association. 00:12:47.010 --> 00:12:49.860 align:middle line:84% So that one over the association constant 00:12:49.860 --> 00:12:53.400 align:middle line:84% is directly related to the fraction of DNA molecules 00:12:53.400 --> 00:12:56.220 align:middle line:84% with protein bound, which is directly 00:12:56.220 --> 00:13:00.630 align:middle line:84% proportional to the signal intensity and fluorescence. 00:13:00.630 --> 00:13:02.550 align:middle line:84% So this is how you get the numbers that 00:13:02.550 --> 00:13:03.930 align:middle line:90% were on that slide. 00:13:03.930 --> 00:13:06.450 align:middle line:84% Now let's return the question that 00:13:06.450 --> 00:13:08.460 align:middle line:84% connected this talk with the last one, which 00:13:08.460 --> 00:13:15.210 align:middle line:84% was the symmetry of DNA protein interactions. 00:13:15.210 --> 00:13:19.500 align:middle line:84% We illustrated already one of these three zinc finger 00:13:19.500 --> 00:13:26.850 align:middle line:84% complexes, as illustrated on the left hand side here. 00:13:26.850 --> 00:13:31.260 align:middle line:84% The double-stranded DNA is in blue and the three zinc fingers 00:13:31.260 --> 00:13:36.820 align:middle line:84% follow along the major groove, the large groove of the DNA. 00:13:36.820 --> 00:13:40.320 align:middle line:84% And the reason the textbook is wrong, first of all, 00:13:40.320 --> 00:13:42.858 align:middle line:84% it emphasizes the non-helical part of the zinc finger. 00:13:42.858 --> 00:13:45.150 align:middle line:84% You can barely see the helix with the background there. 00:13:45.150 --> 00:13:47.392 align:middle line:84% And also the way it loops through the DNA, 00:13:47.392 --> 00:13:49.350 align:middle line:84% if you look at this carefully in your textbook, 00:13:49.350 --> 00:13:51.420 align:middle line:84% this is actually the [? Mount ?] book, 00:13:51.420 --> 00:13:54.405 align:middle line:84% it actually interdigitates with a phosphodiester bond, 00:13:54.405 --> 00:13:56.280 align:middle line:84% basically going through the base pairs, which 00:13:56.280 --> 00:13:58.380 align:middle line:90% is not at all what happens. 00:13:58.380 --> 00:14:00.600 align:middle line:84% Similarly, the leucine zipper, this 00:14:00.600 --> 00:14:02.910 align:middle line:84% is, again, recognition with a helix 00:14:02.910 --> 00:14:05.850 align:middle line:90% in the major groove of DNA. 00:14:05.850 --> 00:14:08.490 align:middle line:84% Here, you can see that the helix that 00:14:08.490 --> 00:14:10.500 align:middle line:84% causes the dimerization of proteins-- you 00:14:10.500 --> 00:14:14.400 align:middle line:84% can think of this as your really most elementary protein-protein 00:14:14.400 --> 00:14:15.480 align:middle line:90% interaction code. 00:14:15.480 --> 00:14:18.090 align:middle line:84% A very fundamental one that comes again and again, 00:14:18.090 --> 00:14:23.220 align:middle line:84% so-called coil-coil, two alpha helices interacting. 00:14:23.220 --> 00:14:26.940 align:middle line:84% The direct extension of that, almost coaxial with a coil-coil 00:14:26.940 --> 00:14:28.500 align:middle line:84% [? or ?] protein-protein interaction, 00:14:28.500 --> 00:14:30.570 align:middle line:90% they go down and touch the DNA. 00:14:30.570 --> 00:14:33.750 align:middle line:84% In contrast, the textbook where it does the sharp right hand 00:14:33.750 --> 00:14:41.070 align:middle line:84% turn and in some way poorly schematized there, 00:14:41.070 --> 00:14:42.690 align:middle line:90% it goes coaxial to DNA. 00:14:42.690 --> 00:14:45.030 align:middle line:90% That's not what happens. 00:14:45.030 --> 00:14:48.120 align:middle line:84% The helices are more or less direct extensions down 00:14:48.120 --> 00:14:52.620 align:middle line:84% from the dimerization region of the protein 00:14:52.620 --> 00:14:57.720 align:middle line:84% maintaining almost perpendicular to the DNA axis. 00:14:57.720 --> 00:15:03.420 align:middle line:84% But again, so on the left is the three tandem repeats, 00:15:03.420 --> 00:15:08.580 align:middle line:84% and on the right is a dyad axis, where the twofold 180 00:15:08.580 --> 00:15:10.320 align:middle line:90% degrees symmetry-- 00:15:10.320 --> 00:15:13.890 align:middle line:84% think of it as rotating exactly 180 degrees. 00:15:13.890 --> 00:15:14.790 align:middle line:90% This is not a mirror. 00:15:14.790 --> 00:15:16.140 align:middle line:90% This is a rotation. 00:15:16.140 --> 00:15:20.010 align:middle line:84% Of the DNA on itself also coincides perfectly 00:15:20.010 --> 00:15:23.220 align:middle line:84% with the twofold symmetry of the protein 00:15:23.220 --> 00:15:24.790 align:middle line:90% association with itself. 00:15:24.790 --> 00:15:26.850 align:middle line:84% These are the two major symmetry classes, 00:15:26.850 --> 00:15:30.510 align:middle line:84% and it's amazing how many nucleic acid protein 00:15:30.510 --> 00:15:33.330 align:middle line:84% interactions fall into one of these two classes-- direct 00:15:33.330 --> 00:15:35.100 align:middle line:90% repeat or inverted repeat. 00:15:35.100 --> 00:15:37.710 align:middle line:84% And that's the one you find direct and inverted repeats 00:15:37.710 --> 00:15:41.020 align:middle line:84% in nucleic acid sequences, you get a little excited. 00:15:41.020 --> 00:15:45.120 align:middle line:84% The other reason is the hairpin structures 00:15:45.120 --> 00:15:47.430 align:middle line:84% that you found in RNA that we talked about 00:15:47.430 --> 00:15:50.070 align:middle line:84% in the last classes, those also are 00:15:50.070 --> 00:15:51.465 align:middle line:90% indicated by inverted repeats. 00:15:51.465 --> 00:15:54.790 align:middle line:90% 00:15:54.790 --> 00:16:00.550 align:middle line:84% So we now have a semi-empirical way of computing-- 00:16:00.550 --> 00:16:04.870 align:middle line:84% in a certain sense, predicting new regulatory protein 00:16:04.870 --> 00:16:07.180 align:middle line:84% DNA interactions with double-stranded DNA. 00:16:07.180 --> 00:16:08.920 align:middle line:90% Can we extend this to RNA? 00:16:08.920 --> 00:16:12.220 align:middle line:84% This is a much more complicated situation with RNA 00:16:12.220 --> 00:16:15.140 align:middle line:84% because you don't have these long perfect double helices 00:16:15.140 --> 00:16:15.640 align:middle line:90% anymore. 00:16:15.640 --> 00:16:17.740 align:middle line:84% You have these very short RNA helices 00:16:17.740 --> 00:16:20.350 align:middle line:84% that I showed in the last couple of classes. 00:16:20.350 --> 00:16:22.930 align:middle line:84% This is transfer RNA, one of our favorite molecules 00:16:22.930 --> 00:16:24.910 align:middle line:84% here, with the anticodon at the bottom 00:16:24.910 --> 00:16:26.830 align:middle line:90% of each of the pink structures. 00:16:26.830 --> 00:16:31.420 align:middle line:84% And the amino acid acceptor three prime end of that 70-some 00:16:31.420 --> 00:16:33.220 align:middle line:90% nucleotide-- 00:16:33.220 --> 00:16:35.560 align:middle line:84% 70 to 80 nucleotide nucleic acid. 00:16:35.560 --> 00:16:37.780 align:middle line:84% So the pinks are all the tRNAs, and there 00:16:37.780 --> 00:16:40.570 align:middle line:84% are at least 20 different types of amino acid 00:16:40.570 --> 00:16:43.870 align:middle line:84% and has 20 types of-- at least 20 types of transfer RNAs 00:16:43.870 --> 00:16:46.180 align:middle line:84% and 20 types of proteins that add 00:16:46.180 --> 00:16:50.350 align:middle line:84% the amino acid onto the three prime ends of the transfer RNA. 00:16:50.350 --> 00:16:52.990 align:middle line:84% These break themselves up into two major classes. 00:16:52.990 --> 00:16:55.030 align:middle line:84% These can be recognized at the structural level. 00:16:55.030 --> 00:16:58.660 align:middle line:84% Class 1, which is the single letter amino acid code 00:16:58.660 --> 00:17:02.890 align:middle line:84% [? CEL, ?] so forth, cysteine, glutamate, and so on. 00:17:02.890 --> 00:17:05.170 align:middle line:84% That's one class, which are structurally similar. 00:17:05.170 --> 00:17:08.770 align:middle line:84% Class 2 is structurally dissimilar to Class 1, 00:17:08.770 --> 00:17:12.130 align:middle line:84% but they are similar within the class. 00:17:12.130 --> 00:17:14.573 align:middle line:84% Anyway, the point is that they recognize 00:17:14.573 --> 00:17:16.240 align:middle line:84% all different parts of the nucleic acid, 00:17:16.240 --> 00:17:19.480 align:middle line:84% not just the anticodon, which is the code itself, 00:17:19.480 --> 00:17:20.950 align:middle line:90% the trinucleotide code. 00:17:20.950 --> 00:17:22.810 align:middle line:84% Not just the amino acid end where 00:17:22.810 --> 00:17:26.050 align:middle line:84% you need the recognize amino acid, 00:17:26.050 --> 00:17:28.930 align:middle line:84% but various points along the transfer RNA. 00:17:28.930 --> 00:17:32.590 align:middle line:84% If you wanted to create a new code, as these authors have, 00:17:32.590 --> 00:17:35.950 align:middle line:84% or to create hybrids between these various things, 00:17:35.950 --> 00:17:40.270 align:middle line:84% you'd have to find homology among the proteins or graph 00:17:40.270 --> 00:17:43.360 align:middle line:84% domains of recognition between each one 00:17:43.360 --> 00:17:46.170 align:middle line:84% or mutagenize particular regions that 00:17:46.170 --> 00:17:47.920 align:middle line:84% are known to interact with the nucleotides 00:17:47.920 --> 00:17:51.250 align:middle line:84% you want to contact, and that's been done. 00:17:51.250 --> 00:17:53.260 align:middle line:84% You can arrange to make a new amino acid 00:17:53.260 --> 00:17:58.990 align:middle line:84% by carving the pocket the amino acid recognizes 00:17:58.990 --> 00:18:03.400 align:middle line:84% and grafting on the appropriate nucleotides-- 00:18:03.400 --> 00:18:05.950 align:middle line:84% the appropriate amino acids would recognize, say, 00:18:05.950 --> 00:18:08.200 align:middle line:90% a stop codon. 00:18:08.200 --> 00:18:13.810 align:middle line:84% OK, you've had some programming experience 00:18:13.810 --> 00:18:16.070 align:middle line:84% that hopefully will prepare you for the real world 00:18:16.070 --> 00:18:21.600 align:middle line:84% of interacting with input and output from various devices. 00:18:21.600 --> 00:18:25.500 align:middle line:90% 00:18:25.500 --> 00:18:27.660 align:middle line:84% The topic today is proteins, and this 00:18:27.660 --> 00:18:32.730 align:middle line:84% really is the main contact between 00:18:32.730 --> 00:18:35.070 align:middle line:84% the exquisite regulatory mechanisms, which 00:18:35.070 --> 00:18:36.790 align:middle line:84% will be the topic of the [? network ?] 00:18:36.790 --> 00:18:38.790 align:middle line:84% that we've already touched upon, but will really 00:18:38.790 --> 00:18:43.350 align:middle line:84% be the topic of some of our network analysis 00:18:43.350 --> 00:18:45.990 align:middle line:90% in the last three lectures. 00:18:45.990 --> 00:18:49.590 align:middle line:84% Here, we need sensors to sense the environment. 00:18:49.590 --> 00:18:51.720 align:middle line:84% We need actuators to then deliver back 00:18:51.720 --> 00:18:54.870 align:middle line:84% into the environment what the cell wants to do 00:18:54.870 --> 00:18:56.880 align:middle line:90% or to interact with other cells. 00:18:56.880 --> 00:18:59.910 align:middle line:84% You have to have feedback, synchrony, 00:18:59.910 --> 00:19:04.170 align:middle line:84% so on that you can basically program the almost 00:19:04.170 --> 00:19:09.600 align:middle line:84% digital nucleic acid world inside the cell 00:19:09.600 --> 00:19:14.475 align:middle line:84% but via clearly analog inputs and outputs. 00:19:14.475 --> 00:19:20.760 align:middle line:84% So since this is the Halloween lecture 00:19:20.760 --> 00:19:26.160 align:middle line:84% and I'm masquerading as the Wolfman, we also-- 00:19:26.160 --> 00:19:32.350 align:middle line:84% I've listed some of the scariest proteins that I could think of. 00:19:32.350 --> 00:19:35.430 align:middle line:84% And we're going to talk about three of them. 00:19:35.430 --> 00:19:39.720 align:middle line:84% One of them in the slide, which is the proteins that 00:19:39.720 --> 00:19:42.210 align:middle line:84% are actually involved in causing the symptoms that 00:19:42.210 --> 00:19:45.630 align:middle line:84% come from when you're worried about anthrax. 00:19:45.630 --> 00:19:49.560 align:middle line:84% And then we'll talk about HIV yet again, 00:19:49.560 --> 00:19:53.850 align:middle line:84% this time, polymerase mutants that cause drug resistance. 00:19:53.850 --> 00:19:57.390 align:middle line:84% And then ApoE yet again, as we have in the past, this time 00:19:57.390 --> 00:20:00.630 align:middle line:84% talking specifically about how protein structure tells us 00:20:00.630 --> 00:20:01.690 align:middle line:90% the haplotype. 00:20:01.690 --> 00:20:05.160 align:middle line:84% So with anthrax, you start out with this simple two 00:20:05.160 --> 00:20:06.510 align:middle line:90% component-- 00:20:06.510 --> 00:20:08.760 align:middle line:90% two protein domains here. 00:20:08.760 --> 00:20:10.760 align:middle line:90% They bind to a cellular-- 00:20:10.760 --> 00:20:12.420 align:middle line:90% something on your cell surface. 00:20:12.420 --> 00:20:15.780 align:middle line:84% Hopefully not yours, but human cell surface. 00:20:15.780 --> 00:20:18.420 align:middle line:84% And then one of the domains disappears. 00:20:18.420 --> 00:20:20.400 align:middle line:84% And the remaining one now self-assembles 00:20:20.400 --> 00:20:23.310 align:middle line:84% into a seven-mer, seven-fold symmetry. 00:20:23.310 --> 00:20:25.967 align:middle line:84% Remember, we were talking about two-fold rotational symmetry 00:20:25.967 --> 00:20:27.300 align:middle line:90% for the DNA protein interaction. 00:20:27.300 --> 00:20:29.430 align:middle line:84% This is now seven-fold rotational symmetry. 00:20:29.430 --> 00:20:33.180 align:middle line:84% That now allows lethal factor, LF, to bind-- 00:20:33.180 --> 00:20:35.040 align:middle line:90% still not inside the cell. 00:20:35.040 --> 00:20:38.550 align:middle line:84% But the whole complex gets internalized. 00:20:38.550 --> 00:20:41.838 align:middle line:84% Still, topologically, it's as if it were outside the cell 00:20:41.838 --> 00:20:43.380 align:middle line:84% when it's inside this little vesicle. 00:20:43.380 --> 00:20:45.990 align:middle line:84% It has to get through that membrane. 00:20:45.990 --> 00:20:49.140 align:middle line:84% But now the pH change that happens when this vesicle goes 00:20:49.140 --> 00:20:53.130 align:middle line:84% in the cell, part of the natural cell biological processes 00:20:53.130 --> 00:20:54.660 align:middle line:90% causes some act-- 00:20:54.660 --> 00:20:59.850 align:middle line:84% an unfortunate act where, now, the seven-mer complex 00:20:59.850 --> 00:21:02.610 align:middle line:84% of proteins does yet another conformational change 00:21:02.610 --> 00:21:06.420 align:middle line:84% and turns into this hairy beast that allows the lethal factor 00:21:06.420 --> 00:21:08.290 align:middle line:84% to get into your cell and kill it. 00:21:08.290 --> 00:21:10.290 align:middle line:84% So you can see that when we're talking 00:21:10.290 --> 00:21:12.623 align:middle line:84% about protein three-dimensional structure, whether we're 00:21:12.623 --> 00:21:16.560 align:middle line:84% predicting it or solving it, protein is not a static object. 00:21:16.560 --> 00:21:18.450 align:middle line:84% Here, it associates with one factor. 00:21:18.450 --> 00:21:21.150 align:middle line:90% It associates seven of itself. 00:21:21.150 --> 00:21:22.800 align:middle line:90% It interacts with lethal factor. 00:21:22.800 --> 00:21:26.730 align:middle line:84% It opens up a whole new channel in the membrane, et cetera. 00:21:26.730 --> 00:21:28.635 align:middle line:84% You need to think of these as dynamic systems 00:21:28.635 --> 00:21:29.760 align:middle line:90% with many different states. 00:21:29.760 --> 00:21:32.570 align:middle line:90% 00:21:32.570 --> 00:21:35.420 align:middle line:84% We also need to think about time scales. 00:21:35.420 --> 00:21:37.460 align:middle line:84% Many of-- the molecular mechanics 00:21:37.460 --> 00:21:40.250 align:middle line:84% we'll be talking about, the timescale of relevance 00:21:40.250 --> 00:21:41.660 align:middle line:90% is the femtosecond. 00:21:41.660 --> 00:21:45.170 align:middle line:84% You need to be able-- this is, well, two nanoseconds. 00:21:45.170 --> 00:21:47.990 align:middle line:84% So 10 to the minus 15th, 10 to the minus 9th seconds. 00:21:47.990 --> 00:21:49.730 align:middle line:90% That's atomic motion. 00:21:49.730 --> 00:21:53.300 align:middle line:84% The turnover of an enzyme that is 00:21:53.300 --> 00:21:56.120 align:middle line:84% the time it takes to for a small molecule, 00:21:56.120 --> 00:21:59.480 align:middle line:84% say, to find and bind the enzyme, 00:21:59.480 --> 00:22:02.750 align:middle line:84% to possibly go through a catalytic step, 00:22:02.750 --> 00:22:06.020 align:middle line:90% and to dissociate as a product. 00:22:06.020 --> 00:22:08.540 align:middle line:84% That's on the order of microseconds to milliseconds. 00:22:08.540 --> 00:22:12.760 align:middle line:84% And the second range is the time that it takes the molecule to-- 00:22:12.760 --> 00:22:15.650 align:middle line:84% the drug or small molecule-- to touch the surface of the cell, 00:22:15.650 --> 00:22:19.400 align:middle line:84% maybe diffuse across the cell, and find its target. 00:22:19.400 --> 00:22:21.020 align:middle line:84% Transcription that we talked about, 00:22:21.020 --> 00:22:26.000 align:middle line:84% all of the regulatory mechanisms of transcription last time, 00:22:26.000 --> 00:22:30.170 align:middle line:84% the rate of the constant for that process 00:22:30.170 --> 00:22:34.760 align:middle line:84% is around 50 nucleotides per second. 00:22:34.760 --> 00:22:36.320 align:middle line:84% Not entirely coincidentally, that's 00:22:36.320 --> 00:22:40.220 align:middle line:84% about the rate at which it is translated into protein. 00:22:40.220 --> 00:22:44.210 align:middle line:84% These are important numbers, because a typical gene size 00:22:44.210 --> 00:22:49.220 align:middle line:84% piece, say, after RNA splicing in higher organisms 00:22:49.220 --> 00:22:52.820 align:middle line:84% or naturally, it might be a kilobase. 00:22:52.820 --> 00:22:56.600 align:middle line:84% So that's about a half a minute to transcribe and translate. 00:22:56.600 --> 00:22:59.300 align:middle line:90% 00:22:59.300 --> 00:23:03.830 align:middle line:84% That could be used as a timer in a circuit of these longer 00:23:03.830 --> 00:23:09.020 align:middle line:84% time frames, like cell cycle, circadian rhythm, 00:23:09.020 --> 00:23:12.680 align:middle line:84% very long time frames in ecological systems 00:23:12.680 --> 00:23:19.760 align:middle line:84% with bamboo and various pests, which can be not even yearly, 00:23:19.760 --> 00:23:21.680 align:middle line:84% and then development and aging, which 00:23:21.680 --> 00:23:25.692 align:middle line:84% can be on the order of hundreds of years, at least for humans, 00:23:25.692 --> 00:23:26.525 align:middle line:90% turtles, and whales. 00:23:26.525 --> 00:23:29.420 align:middle line:90% 00:23:29.420 --> 00:23:33.320 align:middle line:84% So what we think proteins are good for 00:23:33.320 --> 00:23:35.940 align:middle line:90% depends on the accuracy. 00:23:35.940 --> 00:23:37.790 align:middle line:84% And the accuracy depends on the method. 00:23:37.790 --> 00:23:44.490 align:middle line:84% At the very bottom right, we have a very appealing approach, 00:23:44.490 --> 00:23:50.480 align:middle line:84% which is de novo, a priori, or ab initio 00:23:50.480 --> 00:23:52.873 align:middle line:84% prediction of secondary-- of protein 00:23:52.873 --> 00:23:55.040 align:middle line:84% three-dimensional structure from the sequence alone, 00:23:55.040 --> 00:23:56.457 align:middle line:84% which we're getting in bucketloads 00:23:56.457 --> 00:23:57.740 align:middle line:90% from the genome projects. 00:23:57.740 --> 00:23:59.780 align:middle line:84% But unfortunately, accuracy so far-- 00:23:59.780 --> 00:24:02.090 align:middle line:84% and we'll delve into this in more detail in a moment-- 00:24:02.090 --> 00:24:05.300 align:middle line:84% is on the order of six angstroms of difference 00:24:05.300 --> 00:24:07.610 align:middle line:84% between the predicted structure and what it actually 00:24:07.610 --> 00:24:11.150 align:middle line:84% is by the more precise methods up higher on this. 00:24:11.150 --> 00:24:14.270 align:middle line:84% The y-axis here, the vertical axis, 00:24:14.270 --> 00:24:17.840 align:middle line:84% is basically sequence identity as you start doing, say, 00:24:17.840 --> 00:24:22.550 align:middle line:84% threading or comparative modeling here as you get-- 00:24:22.550 --> 00:24:26.420 align:middle line:84% instead of the ab initio or de novo prediction 00:24:26.420 --> 00:24:28.400 align:middle line:84% doesn't require any sequence similarity. 00:24:28.400 --> 00:24:31.370 align:middle line:84% If you want to build it based on previously solved structures, 00:24:31.370 --> 00:24:36.080 align:middle line:84% you need at least 30%, but you're still very far-- say, 00:24:36.080 --> 00:24:37.850 align:middle line:90% 3 to 4 angstroms away-- 00:24:37.850 --> 00:24:40.100 align:middle line:90% from the native structure. 00:24:40.100 --> 00:24:43.280 align:middle line:84% As you get to 1 Angstrom or better in your accuracy, 00:24:43.280 --> 00:24:46.040 align:middle line:84% as you can get from NMR and X-ray crystallography, 00:24:46.040 --> 00:24:49.460 align:middle line:84% you now are in a position to study catalytic mechanism 00:24:49.460 --> 00:24:52.130 align:middle line:84% and design and improve ligands, such as drugs. 00:24:52.130 --> 00:24:53.870 align:middle line:84% This is really where we want to be. 00:24:53.870 --> 00:24:57.380 align:middle line:84% There may be a day where we can do this all from ab initio 00:24:57.380 --> 00:24:59.900 align:middle line:84% prediction or modeling at very great distances. 00:24:59.900 --> 00:25:03.170 align:middle line:84% But for now, modeling at very short, 00:25:03.170 --> 00:25:07.070 align:middle line:84% say, 80% to 90% amino acid similarity, is important. 00:25:07.070 --> 00:25:09.440 align:middle line:84% Remember, there's a-- just like there's 00:25:09.440 --> 00:25:13.710 align:middle line:84% a variety of different protein structures. 00:25:13.710 --> 00:25:17.180 align:middle line:84% This is just an example of a vast literature that 00:25:17.180 --> 00:25:21.350 align:middle line:84% exists where you can use some of the methods 00:25:21.350 --> 00:25:27.140 align:middle line:84% that we'll be discussing in this class 00:25:27.140 --> 00:25:31.640 align:middle line:84% on doing molecular mechanics on proteins 00:25:31.640 --> 00:25:34.220 align:middle line:84% and predicting their three-dimensional structure 00:25:34.220 --> 00:25:37.880 align:middle line:84% in complex with the various drugs. 00:25:37.880 --> 00:25:40.250 align:middle line:84% We will contrast this, or show the interplay 00:25:40.250 --> 00:25:41.900 align:middle line:84% of the computational biology, that 00:25:41.900 --> 00:25:45.560 align:middle line:84% can be aided by actual measurements of drug binding, 00:25:45.560 --> 00:25:48.290 align:middle line:84% just as we had actual measurements of zinc finger 00:25:48.290 --> 00:25:50.480 align:middle line:90% binding to double-stranded DNA. 00:25:50.480 --> 00:25:53.780 align:middle line:84% And ways that you can discover the small molecules 00:25:53.780 --> 00:25:57.845 align:middle line:84% by a clever use of parts of it that you 00:25:57.845 --> 00:26:00.660 align:middle line:84% know bind and parts of it that you know 00:26:00.660 --> 00:26:06.260 align:middle line:84% might be variable in a chemical sense. 00:26:06.260 --> 00:26:07.970 align:middle line:90% Now, this is what-- 00:26:07.970 --> 00:26:13.340 align:middle line:84% just as there is this dynamic competition between pathogens 00:26:13.340 --> 00:26:18.560 align:middle line:84% and their hosts, there's this similar lethal game 00:26:18.560 --> 00:26:21.470 align:middle line:84% that's played between pathogens and the pharmaceutical 00:26:21.470 --> 00:26:22.340 align:middle line:90% industry. 00:26:22.340 --> 00:26:26.120 align:middle line:84% And here, HIV, for which there are many drugs now aimed 00:26:26.120 --> 00:26:29.723 align:middle line:84% either at the protease or at the polymerase-- 00:26:29.723 --> 00:26:31.890 align:middle line:84% some of the first ones were aimed at the polymerase, 00:26:31.890 --> 00:26:34.010 align:middle line:90% and so we have a big collection. 00:26:34.010 --> 00:26:36.230 align:middle line:84% This is one of the most sequenced molecules on Earth, 00:26:36.230 --> 00:26:40.700 align:middle line:84% which is the HIV gene encoding the reverse transcriptase 00:26:40.700 --> 00:26:41.690 align:middle line:90% polymerase. 00:26:41.690 --> 00:26:43.250 align:middle line:84% And it's been sequenced many times 00:26:43.250 --> 00:26:45.110 align:middle line:84% because as a patient takes the drugs, 00:26:45.110 --> 00:26:48.510 align:middle line:84% their population of the AIDS virus changes. 00:26:48.510 --> 00:26:53.420 align:middle line:84% And each of these little diamond-shaped substitution 00:26:53.420 --> 00:26:56.375 align:middle line:84% sites clustered around the binding site in the protein-- 00:26:56.375 --> 00:26:58.250 align:middle line:84% the binding site here indicated the substrate 00:26:58.250 --> 00:27:01.070 align:middle line:84% is in space filling, which is the triphosphate 00:27:01.070 --> 00:27:05.930 align:middle line:84% on the upper left and that the template 00:27:05.930 --> 00:27:08.540 align:middle line:84% kind of curving around on the right-hand side of that space 00:27:08.540 --> 00:27:12.680 align:middle line:90% filling bright green structure. 00:27:12.680 --> 00:27:15.138 align:middle line:90% The protein is in red. 00:27:15.138 --> 00:27:17.180 align:middle line:84% And these little diamonds indicate substitutions, 00:27:17.180 --> 00:27:20.120 align:middle line:84% where the nomenclature is single-letter code for the wild 00:27:20.120 --> 00:27:25.970 align:middle line:84% type, the position in amino acids from the end terminus is 00:27:25.970 --> 00:27:29.330 align:middle line:90% the number, and then the third-- 00:27:29.330 --> 00:27:33.330 align:middle line:84% or the last far right letter is the new amino acids. 00:27:33.330 --> 00:27:37.490 align:middle line:84% So for example, D67N means a [? sparcade ?] at position 67 00:27:37.490 --> 00:27:40.070 align:middle line:84% and wild type changes to an asparagine. 00:27:40.070 --> 00:27:44.240 align:middle line:84% And that causes a drug resistance in the HIV, 00:27:44.240 --> 00:27:46.775 align:middle line:84% with unfortunate consequences for the patient. 00:27:46.775 --> 00:27:52.240 align:middle line:90% 00:27:52.240 --> 00:27:58.110 align:middle line:84% We can take-- now, making mutations in polymerases 00:27:58.110 --> 00:28:01.390 align:middle line:84% is not entirely of negative consequences. 00:28:01.390 --> 00:28:04.740 align:middle line:84% And I'm going to show you a really beautiful example where 00:28:04.740 --> 00:28:09.030 align:middle line:84% a DNA polymerase-- a very similar kind of dynamic, 00:28:09.030 --> 00:28:13.320 align:middle line:84% where the DNA polymerase, you want to change it so that it 00:28:13.320 --> 00:28:19.740 align:middle line:84% can now handle what would normally be an inhibitor of DNA 00:28:19.740 --> 00:28:27.990 align:middle line:84% polymerase, a class of nucleotides that is incapable 00:28:27.990 --> 00:28:29.310 align:middle line:90% of extending-- 00:28:29.310 --> 00:28:34.770 align:middle line:84% capable of being incorporated into the growing replicating 00:28:34.770 --> 00:28:38.160 align:middle line:84% genome, but is not capable of extending. 00:28:38.160 --> 00:28:40.680 align:middle line:84% Is a powerful inhibitor, and it's also powerful 00:28:40.680 --> 00:28:43.110 align:middle line:84% sequencing reagent, these dideoxys. 00:28:43.110 --> 00:28:48.450 align:middle line:84% And so one of the things that was noticed 00:28:48.450 --> 00:28:52.203 align:middle line:84% as the sequences of some of these polymerases 00:28:52.203 --> 00:28:54.120 align:middle line:84% were being studied, and some of the resistance 00:28:54.120 --> 00:28:56.910 align:middle line:84% mutants, and so on, it was noticed 00:28:56.910 --> 00:29:03.060 align:middle line:84% that the complex between the nucleotide, whether it's 00:29:03.060 --> 00:29:07.830 align:middle line:84% a deoxynucleotide shown here with a 3 prime hydroxyl, which 00:29:07.830 --> 00:29:11.790 align:middle line:84% could then be extended by bringing in the next 5 00:29:11.790 --> 00:29:13.410 align:middle line:90% phosphate. 00:29:13.410 --> 00:29:17.610 align:middle line:84% This hydroxyl is near in space to the position 00:29:17.610 --> 00:29:20.190 align:middle line:84% on a phenylalanine or a tyrosine, 00:29:20.190 --> 00:29:24.840 align:middle line:84% position 762 of this polymerase, which can either-- 00:29:24.840 --> 00:29:28.988 align:middle line:84% if it's a tyrosine, it has an OH there. 00:29:28.988 --> 00:29:30.780 align:middle line:84% And if it's a phenylalanine, you lose the O 00:29:30.780 --> 00:29:32.700 align:middle line:84% and you just have a hydrogen there. 00:29:32.700 --> 00:29:38.550 align:middle line:84% And when you have the phenylalanine there, 00:29:38.550 --> 00:29:42.060 align:middle line:90% there's a space that-- 00:29:42.060 --> 00:29:45.210 align:middle line:84% an appropriate space that accommodates the 3 00:29:45.210 --> 00:29:46.810 align:middle line:90% prime hydroxyl quite well. 00:29:46.810 --> 00:29:49.500 align:middle line:84% But if you now put in a dideoxy inhibitor, 00:29:49.500 --> 00:29:51.840 align:middle line:84% you now have too much space in there, 00:29:51.840 --> 00:29:54.330 align:middle line:84% and you start trying to fill that space 00:29:54.330 --> 00:29:58.050 align:middle line:84% with other bulky molecules, like water. 00:29:58.050 --> 00:30:02.190 align:middle line:84% And basically, the binding constant, 00:30:02.190 --> 00:30:04.770 align:middle line:84% it becomes much less favorable binding 00:30:04.770 --> 00:30:07.960 align:middle line:84% when you're lacking both oxygens. 00:30:07.960 --> 00:30:12.970 align:middle line:84% So this presented an opportunity to engineer some polymerases 00:30:12.970 --> 00:30:18.510 align:middle line:84% which had a phenylalanine there to become 00:30:18.510 --> 00:30:21.960 align:middle line:84% more accepting of the dideoxys and hence better 00:30:21.960 --> 00:30:25.650 align:middle line:84% at using disease in DNA sequencing chemistry. 00:30:25.650 --> 00:30:27.810 align:middle line:84% And this was simply by engineering-- 00:30:27.810 --> 00:30:30.450 align:middle line:84% putting in that oxygen there, by changing the phenylalanine 00:30:30.450 --> 00:30:35.790 align:middle line:84% to a tyrosine, it now made a better fit between-- 00:30:35.790 --> 00:30:38.355 align:middle line:90% you can think of it as-- 00:30:38.355 --> 00:30:42.330 align:middle line:84% you can have either oxygen, and by removing this oxygen, 00:30:42.330 --> 00:30:47.730 align:middle line:84% you replace it with an oxygen on the protein side. 00:30:47.730 --> 00:30:50.130 align:middle line:84% I'm trying to emphasize, by a few examples here, 00:30:50.130 --> 00:30:52.020 align:middle line:84% the idea of complementary surfaces 00:30:52.020 --> 00:30:53.530 align:middle line:90% and how you can engineer them. 00:30:53.530 --> 00:30:54.600 align:middle line:90% This is a beautiful case. 00:30:54.600 --> 00:30:56.910 align:middle line:84% Now, we're talking about a single atom rather than 00:30:56.910 --> 00:30:59.010 align:middle line:84% the complementary surfaces of the nucleic acids we 00:30:59.010 --> 00:31:00.480 align:middle line:90% were talking about before. 00:31:00.480 --> 00:31:04.530 align:middle line:84% This has an 8,000-fold effect on the specificity of this 00:31:04.530 --> 00:31:08.610 align:middle line:84% polymerase, and a big impact on the Genome Project. 00:31:08.610 --> 00:31:11.610 align:middle line:84% Now, that's how we program a particular atom 00:31:11.610 --> 00:31:13.740 align:middle line:90% to achieve an important goal. 00:31:13.740 --> 00:31:15.870 align:middle line:84% And of course, the virus has its own mechanisms 00:31:15.870 --> 00:31:18.995 align:middle line:84% for programming it typically, by random mutagenesis 00:31:18.995 --> 00:31:20.370 align:middle line:84% and selection, as we talked about 00:31:20.370 --> 00:31:21.570 align:middle line:90% in the population genetics. 00:31:21.570 --> 00:31:24.240 align:middle line:84% But the way we program in general proteins 00:31:24.240 --> 00:31:30.840 align:middle line:84% are either transgenics, where we might overproduce the protein, 00:31:30.840 --> 00:31:32.580 align:middle line:84% or homologous recombination, which 00:31:32.580 --> 00:31:35.813 align:middle line:84% is the ultimate where we go in and if the protein is already-- 00:31:35.813 --> 00:31:37.980 align:middle line:84% if the gene encoding the protein is already present, 00:31:37.980 --> 00:31:40.710 align:middle line:84% we can change that particular nucleotide in situ 00:31:40.710 --> 00:31:43.020 align:middle line:84% in the correct place so it's properly regulated 00:31:43.020 --> 00:31:43.740 align:middle line:90% and everything. 00:31:43.740 --> 00:31:47.160 align:middle line:90% That's a great way to do it. 00:31:47.160 --> 00:31:49.037 align:middle line:84% Point mutants are not the only way 00:31:49.037 --> 00:31:50.370 align:middle line:90% to generate conditional mutants. 00:31:50.370 --> 00:31:51.662 align:middle line:90% Many of them historically were. 00:31:51.662 --> 00:31:53.820 align:middle line:84% But there are ways that you can program, 00:31:53.820 --> 00:31:56.610 align:middle line:84% and conditional, meaning that you can regulate under what 00:31:56.610 --> 00:31:58.470 align:middle line:84% conditions the protein is expressed 00:31:58.470 --> 00:32:02.250 align:middle line:84% or not or active or not with an entire domain, 00:32:02.250 --> 00:32:05.970 align:middle line:84% or with single nucleotide polymorphisms. 00:32:05.970 --> 00:32:07.440 align:middle line:90% Now, so this is one way. 00:32:07.440 --> 00:32:11.110 align:middle line:90% This is the nucleic acid way. 00:32:11.110 --> 00:32:15.870 align:middle line:84% Another way is by modulating the activity of the proteins 00:32:15.870 --> 00:32:19.680 align:middle line:84% from the outside with drugs or drug-like molecules 00:32:19.680 --> 00:32:21.210 align:middle line:90% and chemical kinetics. 00:32:21.210 --> 00:32:23.280 align:middle line:84% And under the subheadings for that, 00:32:23.280 --> 00:32:25.770 align:middle line:84% you can make these by combinatorial synthesis-- 00:32:25.770 --> 00:32:27.240 align:middle line:84% and we'll show an example of that. 00:32:27.240 --> 00:32:30.990 align:middle line:84% But combinatorial synthesis can be based on design principles, 00:32:30.990 --> 00:32:33.450 align:middle line:90% not just completely random. 00:32:33.450 --> 00:32:34.058 align:middle line:90% Usually are. 00:32:34.058 --> 00:32:35.850 align:middle line:84% The design principles can take into account 00:32:35.850 --> 00:32:38.400 align:middle line:84% what you know about the nature of the interaction 00:32:38.400 --> 00:32:41.790 align:middle line:90% of similar proteins. 00:32:41.790 --> 00:32:49.740 align:middle line:84% And you can mine whatever biochemical data 00:32:49.740 --> 00:32:52.420 align:middle line:84% that you can collect for so-called quantitative 00:32:52.420 --> 00:32:53.950 align:middle line:84% structure activity relationships. 00:32:53.950 --> 00:32:56.560 align:middle line:84% This is a slightly different discipline 00:32:56.560 --> 00:33:02.447 align:middle line:84% than the detailed crystallographic 00:33:02.447 --> 00:33:04.780 align:middle line:84% and quantitative studies that we've talked about so far. 00:33:04.780 --> 00:33:07.360 align:middle line:84% Here, you're trying to basically mine 00:33:07.360 --> 00:33:11.320 align:middle line:84% through the structures of the ligand itself 00:33:11.320 --> 00:33:13.450 align:middle line:84% for the parts of the ligand that might 00:33:13.450 --> 00:33:15.610 align:middle line:84% be responsible for the activity, the binding 00:33:15.610 --> 00:33:20.090 align:middle line:84% activity or the full biological activity that you see. 00:33:20.090 --> 00:33:23.320 align:middle line:84% So let's look at some examples of single-nucleotide 00:33:23.320 --> 00:33:27.610 align:middle line:84% polymorphisms that we've been talking about before in-- 00:33:27.610 --> 00:33:32.440 align:middle line:84% actually, this is a class that we didn't discuss before. 00:33:32.440 --> 00:33:34.690 align:middle line:90% But in previous classes. 00:33:34.690 --> 00:33:37.450 align:middle line:84% But it's related to what we've been talking about. 00:33:37.450 --> 00:33:40.233 align:middle line:84% In the case of the zinc finger, we made an altered specificity. 00:33:40.233 --> 00:33:41.650 align:middle line:84% We made new zinc fingers with bind 00:33:41.650 --> 00:33:44.500 align:middle line:84% to completely new trinucleotides. 00:33:44.500 --> 00:33:48.310 align:middle line:84% With the DNA polymerase, by changing one amino acid, 00:33:48.310 --> 00:33:52.450 align:middle line:84% we could make it now accept almost four logs better 00:33:52.450 --> 00:33:54.920 align:middle line:90% an inhibitor is very useful. 00:33:54.920 --> 00:33:58.480 align:middle line:84% And here, many different-- many of these 00:33:58.480 --> 00:34:03.520 align:middle line:84% are enzymes, where you can not just knock out the enzyme, 00:34:03.520 --> 00:34:06.190 align:middle line:84% but actually make it recognize a new substrate, 00:34:06.190 --> 00:34:11.679 align:middle line:84% or change radically the binding constant and catalytic rate 00:34:11.679 --> 00:34:13.120 align:middle line:90% for new substrates. 00:34:13.120 --> 00:34:16.270 align:middle line:90% 00:34:16.270 --> 00:34:19.219 align:middle line:84% And I just have this long fine-print slide just 00:34:19.219 --> 00:34:21.580 align:middle line:84% to impress to you-- this is actually less than half 00:34:21.580 --> 00:34:23.320 align:middle line:90% of the list-- 00:34:23.320 --> 00:34:24.469 align:middle line:90% just how many examples. 00:34:24.469 --> 00:34:26.320 align:middle line:90% These are not that unusual. 00:34:26.320 --> 00:34:30.909 align:middle line:84% And those can be designed or naturally occurring. 00:34:30.909 --> 00:34:34.900 align:middle line:84% Now, we're going to take the three-dimensional structure 00:34:34.900 --> 00:34:44.860 align:middle line:84% of proteins and connect it with our discussion of haplotypes 00:34:44.860 --> 00:34:46.870 align:middle line:84% and single-nucleotide polymorphisms. 00:34:46.870 --> 00:34:51.100 align:middle line:84% And you may recall that with-- one 00:34:51.100 --> 00:34:55.270 align:middle line:84% of the commonly occurring single-nucleotide 00:34:55.270 --> 00:34:59.890 align:middle line:84% polymorphisms is the ApoE4 allele. 00:34:59.890 --> 00:35:02.020 align:middle line:84% It's present in 20% of the human population. 00:35:02.020 --> 00:35:06.430 align:middle line:84% Even though it has unfortunate consequences, we think, 00:35:06.430 --> 00:35:10.480 align:middle line:90% mainly for Alzheimer's-- 00:35:10.480 --> 00:35:13.360 align:middle line:84% it increases the risk of Alzheimer's, and probably 00:35:13.360 --> 00:35:20.770 align:middle line:84% increases cardiovascular fitness through ApoE refers 00:35:20.770 --> 00:35:24.790 align:middle line:84% to its involvement in cholesterol metabolism 00:35:24.790 --> 00:35:25.960 align:middle line:90% and transport. 00:35:25.960 --> 00:35:28.870 align:middle line:84% The ApoE3 allele is present in about 80%, 00:35:28.870 --> 00:35:31.810 align:middle line:84% and is far more common in human populations, but both of these 00:35:31.810 --> 00:35:34.370 align:middle line:84% would be considered very common alleles. 00:35:34.370 --> 00:35:37.510 align:middle line:84% And we also mentioned that the ancestral form of this, 00:35:37.510 --> 00:35:40.990 align:middle line:84% for example, found in chimpanzees at nearly 100%, 00:35:40.990 --> 00:35:44.980 align:middle line:84% is this arginine 112, instead of what's now common 00:35:44.980 --> 00:35:47.170 align:middle line:84% in human populations was cysteine 112. 00:35:47.170 --> 00:35:50.080 align:middle line:84% And that was-- one explanation for that might be that it was 00:35:50.080 --> 00:35:52.990 align:middle line:90% physiologically-- 00:35:52.990 --> 00:35:55.760 align:middle line:84% our nutritional standards have changed. 00:35:55.760 --> 00:35:57.430 align:middle line:84% We now eat a lot more fatty things. 00:35:57.430 --> 00:35:59.570 align:middle line:84% We live long enough to get Alzheimer's. 00:35:59.570 --> 00:36:04.150 align:middle line:84% And so maybe this was something that was-- 00:36:04.150 --> 00:36:07.480 align:middle line:84% this bad allele, E4, was good in chimpanzees that have 00:36:07.480 --> 00:36:09.940 align:middle line:90% different diets or lifespans. 00:36:09.940 --> 00:36:13.870 align:middle line:84% But the other possibility-- and I can't really 00:36:13.870 --> 00:36:15.850 align:middle line:84% distinguish between these right now-- 00:36:15.850 --> 00:36:17.770 align:middle line:84% but another one to seriously consider, 00:36:17.770 --> 00:36:19.990 align:middle line:84% not just in this case, but in cases in general, 00:36:19.990 --> 00:36:23.620 align:middle line:84% is you no longer just think about single nucleotides. 00:36:23.620 --> 00:36:24.910 align:middle line:90% You think about haplotypes. 00:36:24.910 --> 00:36:27.970 align:middle line:84% Everything insists on that DNA strand 00:36:27.970 --> 00:36:29.950 align:middle line:84% has a chance of affecting either the expression 00:36:29.950 --> 00:36:33.580 align:middle line:84% level of the protein or insists, on the protein strand, 00:36:33.580 --> 00:36:35.560 align:middle line:90% to fall back and interact. 00:36:35.560 --> 00:36:38.230 align:middle line:84% And you can see that one of the nearest amino acids 00:36:38.230 --> 00:36:41.770 align:middle line:84% to this arginine 112, which is the main difference 00:36:41.770 --> 00:36:44.710 align:middle line:90% between ApoE4 and ApoE3. 00:36:44.710 --> 00:36:47.200 align:middle line:84% Arginine 61 is the same on the two alleles. 00:36:47.200 --> 00:36:50.320 align:middle line:84% But you think of this as one haplotype, 00:36:50.320 --> 00:36:53.830 align:middle line:84% and in chimpanzees the haplotype, is now threonine 61. 00:36:53.830 --> 00:36:59.620 align:middle line:84% And you can think that a [? 3R ?] in chimps, 00:36:59.620 --> 00:37:05.560 align:middle line:84% or ancestrally, is not too different from [? Rcys. ?] So 00:37:05.560 --> 00:37:07.250 align:middle line:90% the different order. 00:37:07.250 --> 00:37:10.510 align:middle line:84% So it's like this compensating, complementary mutation, 00:37:10.510 --> 00:37:16.480 align:middle line:84% just like we had in the oxygens in the polymerase 00:37:16.480 --> 00:37:17.710 align:middle line:90% a couple of slides ago. 00:37:17.710 --> 00:37:19.752 align:middle line:84% And you can think of the compensating mutations-- 00:37:19.752 --> 00:37:22.340 align:middle line:84% we had the mutual information theory for doing the theory 00:37:22.340 --> 00:37:23.230 align:middle line:90% structure. 00:37:23.230 --> 00:37:24.700 align:middle line:90% Think of complementary surfaces. 00:37:24.700 --> 00:37:27.880 align:middle line:84% When you think of single nucleotides, 00:37:27.880 --> 00:37:33.070 align:middle line:84% don't think of them as haplotype and possibly complementing 00:37:33.070 --> 00:37:34.930 align:middle line:90% constellations. 00:37:34.930 --> 00:37:40.690 align:middle line:84% Especially, now, this brings us to the possible impact 00:37:40.690 --> 00:37:42.610 align:middle line:84% of three-dimensional structure on predicting 00:37:42.610 --> 00:37:43.900 align:middle line:90% deleterious human alleles. 00:37:43.900 --> 00:37:46.900 align:middle line:84% If we suddenly had the sequence of everyone in this room 00:37:46.900 --> 00:37:50.548 align:middle line:84% and we wanted our computer program to prioritize, which 00:37:50.548 --> 00:37:51.840 align:middle line:90% ones should I the attention to? 00:37:51.840 --> 00:37:56.190 align:middle line:84% Which deviations from the most common allele 00:37:56.190 --> 00:37:57.212 align:middle line:90% should I look at first? 00:37:57.212 --> 00:37:59.670 align:middle line:84% Well, you might think of these things in terms of proteins. 00:37:59.670 --> 00:38:01.470 align:middle line:84% We've now gotten to the point in the course 00:38:01.470 --> 00:38:03.030 align:middle line:84% where we're talking about proteins. 00:38:03.030 --> 00:38:05.363 align:middle line:84% You need to think about the three-dimensional structure. 00:38:05.363 --> 00:38:07.770 align:middle line:90% Who's near who in the structure? 00:38:07.770 --> 00:38:09.300 align:middle line:84% You can think about binding sites. 00:38:09.300 --> 00:38:10.980 align:middle line:90% These might be indicated by-- 00:38:10.980 --> 00:38:14.250 align:middle line:84% if you know the three-dimensional structure 00:38:14.250 --> 00:38:16.350 align:middle line:84% or you know the conservation pattern 00:38:16.350 --> 00:38:18.450 align:middle line:90% in this family of proteins. 00:38:18.450 --> 00:38:21.300 align:middle line:90% You can ask things about charge. 00:38:21.300 --> 00:38:22.950 align:middle line:84% In that last slide, we had the charge 00:38:22.950 --> 00:38:28.500 align:middle line:84% of the arginines being near a compatibly 00:38:28.500 --> 00:38:35.100 align:middle line:84% partial negative charge on the cysteines or threonines. 00:38:35.100 --> 00:38:40.050 align:middle line:84% You can have-- a disulfide is a very important thing to lose. 00:38:40.050 --> 00:38:43.410 align:middle line:84% They tend to be highly conserved. 00:38:43.410 --> 00:38:47.220 align:middle line:84% If you introduce a proline into what would normally be an alpha 00:38:47.220 --> 00:38:49.223 align:middle line:84% helix, this is something where knowledge 00:38:49.223 --> 00:38:50.640 align:middle line:84% of the three-dimensional structure 00:38:50.640 --> 00:38:53.940 align:middle line:84% would say, oh, that proline, this a priori, without any 00:38:53.940 --> 00:38:56.310 align:middle line:84% knowledge of conservation, could be 00:38:56.310 --> 00:39:00.240 align:middle line:84% a huge change in the three-dimensional structure. 00:39:00.240 --> 00:39:02.730 align:middle line:84% And then these multi-sequence profiles 00:39:02.730 --> 00:39:06.810 align:middle line:84% are a good way of looking at the conservation. 00:39:06.810 --> 00:39:10.830 align:middle line:84% That's a way of prioritizing single-nucleotide polymorphisms 00:39:10.830 --> 00:39:14.040 align:middle line:84% that might have impact on pharmacogenomics or disease 00:39:14.040 --> 00:39:15.240 align:middle line:90% in general. 00:39:15.240 --> 00:39:19.920 align:middle line:84% Now, as we integrate that with the chemical diversity 00:39:19.920 --> 00:39:20.772 align:middle line:90% that we can create-- 00:39:20.772 --> 00:39:22.980 align:middle line:84% that's going to be the topic for the next few slides, 00:39:22.980 --> 00:39:26.730 align:middle line:84% is how do we create chemical diversity? 00:39:26.730 --> 00:39:30.570 align:middle line:84% And I'm going to introduce this, the idea of chemical diversity, 00:39:30.570 --> 00:39:33.540 align:middle line:84% in a way I hope nicely connects to where 00:39:33.540 --> 00:39:36.030 align:middle line:90% we've been with RNA arrays. 00:39:36.030 --> 00:39:38.820 align:middle line:84% RNA arrays, and the double-stranded RNA array 00:39:38.820 --> 00:39:41.790 align:middle line:84% that we used earlier in class today, 00:39:41.790 --> 00:39:43.680 align:middle line:84% can be generated in a commentorial sense. 00:39:43.680 --> 00:39:45.750 align:middle line:90% You can make an exhaustive set. 00:39:45.750 --> 00:39:48.780 align:middle line:84% Now, typically, those were made where, spatially, they 00:39:48.780 --> 00:39:49.590 align:middle line:90% were isolated. 00:39:49.590 --> 00:39:55.650 align:middle line:84% Each different nucleic acid, oligonucleotide, 00:39:55.650 --> 00:39:58.800 align:middle line:84% is present in a different place identifiable to the computer 00:39:58.800 --> 00:40:01.780 align:middle line:90% by its coordinates on an array. 00:40:01.780 --> 00:40:05.550 align:middle line:84% But you can also make them in a big mixture 00:40:05.550 --> 00:40:08.610 align:middle line:84% and use them as a mixture and do selection on them, as we 00:40:08.610 --> 00:40:10.800 align:middle line:90% did with the phage display. 00:40:10.800 --> 00:40:14.490 align:middle line:84% Or you can make them as a mixture of solid phase 00:40:14.490 --> 00:40:17.670 align:middle line:84% particles and then separate the [INAUDIBLE] phase particles out 00:40:17.670 --> 00:40:19.350 align:middle line:90% in some manner. 00:40:19.350 --> 00:40:21.765 align:middle line:84% Solid phase comes up again and again in arrays. 00:40:21.765 --> 00:40:23.640 align:middle line:84% It's very obvious why you have a solid phase. 00:40:23.640 --> 00:40:27.360 align:middle line:84% You want to be able to address it by its positions in x and y 00:40:27.360 --> 00:40:29.700 align:middle line:90% on the array. 00:40:29.700 --> 00:40:31.470 align:middle line:84% But the other reason-- technical reasons 00:40:31.470 --> 00:40:33.930 align:middle line:84% are, it's a fantastic way of getting purification 00:40:33.930 --> 00:40:36.990 align:middle line:84% of your products simply by washing 00:40:36.990 --> 00:40:40.290 align:middle line:84% rather than doing complicated purification procedures. 00:40:40.290 --> 00:40:43.110 align:middle line:84% And it allows you to, in the case of beads, 00:40:43.110 --> 00:40:45.750 align:middle line:84% there now-- you can think of it as an ultimate and flexible 00:40:45.750 --> 00:40:46.380 align:middle line:90% array. 00:40:46.380 --> 00:40:50.460 align:middle line:84% You can move the beads around and put them in new arrays, 00:40:50.460 --> 00:40:53.250 align:middle line:90% and identify them later. 00:40:53.250 --> 00:40:57.870 align:middle line:84% Anyway, so we're going to introduce the general way 00:40:57.870 --> 00:41:01.230 align:middle line:90% of making either-- 00:41:01.230 --> 00:41:03.660 align:middle line:84% complex chemicals, whether they're 00:41:03.660 --> 00:41:07.440 align:middle line:84% linear polymers like proteins or nucleic acids, 00:41:07.440 --> 00:41:15.150 align:middle line:84% or much more tighter and small molecules. 00:41:15.150 --> 00:41:17.093 align:middle line:84% But they have similar concepts that you need. 00:41:17.093 --> 00:41:19.260 align:middle line:84% There's the solid phase that I already talked about. 00:41:19.260 --> 00:41:23.760 align:middle line:84% There's the idea of protecting groups, 00:41:23.760 --> 00:41:25.650 align:middle line:84% and the protecting groups are protecting 00:41:25.650 --> 00:41:27.030 align:middle line:90% against a reactive group. 00:41:27.030 --> 00:41:28.980 align:middle line:84% So the highly reactive group here 00:41:28.980 --> 00:41:33.090 align:middle line:84% is the phosphoramidite, which is this phosphate nitrogen bond. 00:41:33.090 --> 00:41:39.480 align:middle line:84% This is capable of reacting with just about any nitrogen 00:41:39.480 --> 00:41:41.400 align:middle line:84% or oxygen, such as this-- eventually, 00:41:41.400 --> 00:41:44.400 align:middle line:84% once you do protect this oxygen at the 5 prime position, 00:41:44.400 --> 00:41:46.702 align:middle line:84% these two oxygens are 5 and 3, just like the ones we've 00:41:46.702 --> 00:41:47.910 align:middle line:90% been talking about all along. 00:41:47.910 --> 00:41:51.150 align:middle line:84% This is the chemical synthesis version of the polymerase 00:41:51.150 --> 00:41:52.560 align:middle line:90% that we've been talking about. 00:41:52.560 --> 00:41:54.600 align:middle line:84% So you have these reactive groups 00:41:54.600 --> 00:41:57.660 align:middle line:90% and the protecting groups. 00:41:57.660 --> 00:41:59.910 align:middle line:84% And those are the major concepts. 00:41:59.910 --> 00:42:01.140 align:middle line:90% Now, let's go through. 00:42:01.140 --> 00:42:08.250 align:middle line:84% This is-- the topic here is proteins. 00:42:08.250 --> 00:42:11.610 align:middle line:84% And we'll talk more about protein synthesis 00:42:11.610 --> 00:42:14.435 align:middle line:84% as part of quantitation next time, 00:42:14.435 --> 00:42:16.560 align:middle line:84% and as part of networks in the last three lectures. 00:42:16.560 --> 00:42:18.480 align:middle line:84% But here's a completely synthetic way 00:42:18.480 --> 00:42:21.780 align:middle line:84% of getting short peptides, either 00:42:21.780 --> 00:42:23.550 align:middle line:84% by directly synthesizing the peptides 00:42:23.550 --> 00:42:27.300 align:middle line:84% or synthesizing nucleic acid that encodes that peptide 00:42:27.300 --> 00:42:29.490 align:middle line:84% or interferes with the production of that peptide. 00:42:29.490 --> 00:42:32.430 align:middle line:84% And you can think of these as drug-like molecules. 00:42:32.430 --> 00:42:35.100 align:middle line:90% These are naturally related-- 00:42:35.100 --> 00:42:37.570 align:middle line:84% they can be analogs of nucleic acids and proteins, 00:42:37.570 --> 00:42:39.563 align:middle line:90% not just straight ones. 00:42:39.563 --> 00:42:40.980 align:middle line:84% And we'll talk about opportunities 00:42:40.980 --> 00:42:43.530 align:middle line:90% for making these analogs. 00:42:43.530 --> 00:42:47.100 align:middle line:84% And so by making analogs of known proteins 00:42:47.100 --> 00:42:51.230 align:middle line:84% or nucleic acids, you kind of have a more immediate 00:42:51.230 --> 00:42:54.440 align:middle line:84% connection between the thing that your computer instructed 00:42:54.440 --> 00:42:57.440 align:middle line:84% the synthesizer to make and your targets. 00:42:57.440 --> 00:42:59.622 align:middle line:84% Well, if you make a random chemical, 00:42:59.622 --> 00:43:01.580 align:middle line:84% you don't necessarily know what your target is. 00:43:01.580 --> 00:43:03.320 align:middle line:84% But we'll talk about ways of making 00:43:03.320 --> 00:43:05.150 align:middle line:90% slightly less random chemicals. 00:43:05.150 --> 00:43:07.640 align:middle line:84% But this is one way of making a direct connection. 00:43:07.640 --> 00:43:11.150 align:middle line:84% And the process is cyclic in the sense 00:43:11.150 --> 00:43:15.260 align:middle line:84% that each cycle, you return, and the polymer 00:43:15.260 --> 00:43:16.370 align:middle line:90% gets a little bit longer. 00:43:16.370 --> 00:43:19.580 align:middle line:84% You start with one monomer on a solid phase, shown 00:43:19.580 --> 00:43:25.940 align:middle line:84% by these little hexagons on the far right side of the slide. 00:43:25.940 --> 00:43:30.830 align:middle line:84% And you add-- you remove the protecting group 00:43:30.830 --> 00:43:34.670 align:middle line:84% on the immobilized polymer, one protecting group. 00:43:34.670 --> 00:43:36.530 align:middle line:84% And then you bring in this reactive group, 00:43:36.530 --> 00:43:38.240 align:middle line:90% otherwise protected. 00:43:38.240 --> 00:43:41.210 align:middle line:84% And there's really one major product that you expect. 00:43:41.210 --> 00:43:42.860 align:middle line:90% You wash off all of the excess. 00:43:42.860 --> 00:43:44.540 align:middle line:90% You now have one longer. 00:43:44.540 --> 00:43:45.410 align:middle line:90% You deprotect. 00:43:45.410 --> 00:43:48.440 align:middle line:90% This DMT group is removed. 00:43:48.440 --> 00:43:51.080 align:middle line:84% And you go back up and cycle again. 00:43:51.080 --> 00:43:54.300 align:middle line:84% There may be additional steps, such as oxidation, 00:43:54.300 --> 00:43:57.290 align:middle line:84% which will stabilize the new bond that you've made. 00:43:57.290 --> 00:44:02.840 align:middle line:84% Or you can have a capping step that can soak up 00:44:02.840 --> 00:44:05.332 align:middle line:90% any excess that was left over. 00:44:05.332 --> 00:44:07.790 align:middle line:84% But in general, after you're done with all of these cycles, 00:44:07.790 --> 00:44:10.100 align:middle line:84% then there'll be a step where you 00:44:10.100 --> 00:44:12.800 align:middle line:84% remove the protecting groups altogether and remove 00:44:12.800 --> 00:44:15.830 align:middle line:84% the polymer from the solid phase if you so choose. 00:44:15.830 --> 00:44:18.500 align:middle line:84% Or you leave it there, if you have an array. 00:44:18.500 --> 00:44:21.590 align:middle line:84% These are other examples of protecting groups. 00:44:21.590 --> 00:44:23.030 align:middle line:90% These are now on the bases. 00:44:23.030 --> 00:44:25.940 align:middle line:84% Some bases don't need protection, like thiamines. 00:44:25.940 --> 00:44:27.830 align:middle line:84% If you have an exocyclic amine, then you 00:44:27.830 --> 00:44:32.120 align:middle line:84% typically need a benzyl or an isobutyryl group in blue here, 00:44:32.120 --> 00:44:33.275 align:middle line:90% are the protecting groups. 00:44:33.275 --> 00:44:37.100 align:middle line:90% 00:44:37.100 --> 00:44:44.750 align:middle line:84% I said there's an opportunity here for modifying 00:44:44.750 --> 00:44:52.160 align:middle line:84% the nucleotides or oligopeptides or other chemicals 00:44:52.160 --> 00:44:56.270 align:middle line:84% to make them so that they're related to, but not 00:44:56.270 --> 00:45:01.400 align:middle line:84% identical in every property, to normal constituents 00:45:01.400 --> 00:45:06.200 align:middle line:84% of your body or of a bacterial cell that you're aiming at. 00:45:06.200 --> 00:45:08.210 align:middle line:84% And why would you want to make derivatives? 00:45:08.210 --> 00:45:09.680 align:middle line:90% Why not make the exact thing? 00:45:09.680 --> 00:45:10.940 align:middle line:84% And the reason you want to make derivatives 00:45:10.940 --> 00:45:13.100 align:middle line:84% would be, for example, to increase their stability, 00:45:13.100 --> 00:45:19.100 align:middle line:84% or decrease their stability, or make them bind more faster 00:45:19.100 --> 00:45:20.510 align:middle line:90% or more irreversibly. 00:45:20.510 --> 00:45:23.150 align:middle line:84% And examples are, in the previous slide, 00:45:23.150 --> 00:45:24.710 align:middle line:90% you can make modified bases. 00:45:24.710 --> 00:45:28.460 align:middle line:84% And in slide 29 here you can change the backbone itself. 00:45:28.460 --> 00:45:30.290 align:middle line:84% You can change the ribosomes so that they 00:45:30.290 --> 00:45:34.430 align:middle line:84% have bulky groups that prevent the nucleases from getting in, 00:45:34.430 --> 00:45:41.150 align:middle line:84% or they can exchange oxygen for sulfurs or hydrogens 00:45:41.150 --> 00:45:44.520 align:middle line:84% in order to make the phosphodiester backbone itself, 00:45:44.520 --> 00:45:47.030 align:middle line:84% which is where nucleus is cleaved, 00:45:47.030 --> 00:45:51.200 align:middle line:84% a less attractive, less energetically 00:45:51.200 --> 00:45:54.200 align:middle line:90% favorable substrate. 00:45:54.200 --> 00:45:57.990 align:middle line:84% So now, those are chemical processes 00:45:57.990 --> 00:46:01.340 align:middle line:84% that, in a certain sense, mimic the normal polymerases 00:46:01.340 --> 00:46:03.680 align:middle line:90% and ribosomes in the cell. 00:46:03.680 --> 00:46:07.100 align:middle line:84% And there are analogous processes to generate chemical 00:46:07.100 --> 00:46:10.580 align:middle line:90% diversity on smaller molecules. 00:46:10.580 --> 00:46:14.810 align:middle line:84% And then there are, analogous to that, 00:46:14.810 --> 00:46:19.910 align:middle line:84% biological mechanisms by which you can make small molecule 00:46:19.910 --> 00:46:25.460 align:middle line:84% diversity which are less cyclic than processes 00:46:25.460 --> 00:46:26.720 align:middle line:90% we just talked about. 00:46:26.720 --> 00:46:34.580 align:middle line:84% These are more a set of ordered reactions 00:46:34.580 --> 00:46:38.460 align:middle line:84% that has a conceptual repeat, but in a sense, 00:46:38.460 --> 00:46:40.970 align:middle line:84% you can think of it as a linear program 00:46:40.970 --> 00:46:43.710 align:middle line:84% that you go from the beginning to the end. 00:46:43.710 --> 00:46:46.880 align:middle line:84% And that's to make these polyketides, which 00:46:46.880 --> 00:46:49.010 align:middle line:84% are shown on the right-hand side of the slide. 00:46:49.010 --> 00:46:53.060 align:middle line:84% A large class of pharmaceuticals, 00:46:53.060 --> 00:46:54.950 align:middle line:84% including most of the antibiotics, 00:46:54.950 --> 00:46:57.860 align:middle line:84% are made by a fairly small set of organisms, 00:46:57.860 --> 00:47:00.260 align:middle line:84% such as streptomyces in certain plants. 00:47:00.260 --> 00:47:02.572 align:middle line:84% And this process by which it's made, 00:47:02.572 --> 00:47:05.030 align:middle line:84% which will be illustrated in more detail in the next slide, 00:47:05.030 --> 00:47:07.520 align:middle line:84% is very akin to fatty acid synthesis. 00:47:07.520 --> 00:47:09.830 align:middle line:84% Fatty acids are long hydrocarbon chains, 00:47:09.830 --> 00:47:12.860 align:middle line:84% which you can think of, adding each two 00:47:12.860 --> 00:47:15.710 align:middle line:84% carbons on that fatty acid is a process 00:47:15.710 --> 00:47:20.270 align:middle line:84% akin to the one for making these polyketide drug-like molecules. 00:47:20.270 --> 00:47:22.580 align:middle line:84% Another way of biologically making 00:47:22.580 --> 00:47:28.160 align:middle line:84% a very compact structure, which actually uses ribosomes, 00:47:28.160 --> 00:47:33.440 align:middle line:84% but it uses them to make very tight short peptides 00:47:33.440 --> 00:47:37.040 align:middle line:84% and a precursor that then folds with lots of disulfides 00:47:37.040 --> 00:47:41.750 align:middle line:84% to make this small and highly cross-linked. 00:47:41.750 --> 00:47:47.210 align:middle line:84% Now, the fact that these cone snails have gone 00:47:47.210 --> 00:47:49.040 align:middle line:84% have gone to the trouble of making hundreds 00:47:49.040 --> 00:47:52.010 align:middle line:84% of these different very small peptides that 00:47:52.010 --> 00:47:55.730 align:middle line:84% have these properties tells you something about what 00:47:55.730 --> 00:47:59.630 align:middle line:84% it is that you want-- that drug-like molecules have 00:47:59.630 --> 00:48:00.230 align:middle line:90% in common. 00:48:00.230 --> 00:48:06.080 align:middle line:84% And that is, they're small, so they diffuse quickly and get 00:48:06.080 --> 00:48:07.230 align:middle line:90% to their site. 00:48:07.230 --> 00:48:09.020 align:middle line:84% And so you can have large amounts of them 00:48:09.020 --> 00:48:10.050 align:middle line:90% in a small space. 00:48:10.050 --> 00:48:12.110 align:middle line:84% So you can manufacture lots of them. 00:48:12.110 --> 00:48:15.680 align:middle line:84% And then they have to be highly cross-linked in order 00:48:15.680 --> 00:48:17.060 align:middle line:90% to maintain the rigidity. 00:48:17.060 --> 00:48:19.730 align:middle line:84% Because they're small, they have less surface area 00:48:19.730 --> 00:48:23.640 align:middle line:90% to bind to their binding pocket. 00:48:23.640 --> 00:48:26.810 align:middle line:84% So to compensate for that, they have to have a lot of rigidity. 00:48:26.810 --> 00:48:29.210 align:middle line:84% And the thing that you lock in rigidly 00:48:29.210 --> 00:48:30.740 align:middle line:90% has to be the correct structure. 00:48:30.740 --> 00:48:32.930 align:middle line:84% It does you no good to have a rigid structure that 00:48:32.930 --> 00:48:36.330 align:middle line:84% isn't really perfectly complementary to that surface. 00:48:36.330 --> 00:48:40.070 align:middle line:84% And the third source of biological diversity 00:48:40.070 --> 00:48:41.820 align:middle line:84% is one you're probably more familiar with, 00:48:41.820 --> 00:48:44.510 align:middle line:84% which is the immune receptors, the B and T cell 00:48:44.510 --> 00:48:49.680 align:middle line:84% receptors, the antibodies, and cell-mediated immunity. 00:48:49.680 --> 00:48:57.200 align:middle line:84% And these use recombination machinery 00:48:57.200 --> 00:49:04.940 align:middle line:84% to program various combinations of nucleic acid motifs 00:49:04.940 --> 00:49:06.290 align:middle line:90% that encode protein motifs. 00:49:06.290 --> 00:49:08.030 align:middle line:84% And as they do that recombination, 00:49:08.030 --> 00:49:10.130 align:middle line:84% they have further diversification 00:49:10.130 --> 00:49:12.740 align:middle line:84% that occurs due to a template-independent 00:49:12.740 --> 00:49:15.290 align:middle line:84% polymerase, [? thermotransferase, ?] 00:49:15.290 --> 00:49:18.050 align:middle line:84% which will extend a few nucleotides of completely 00:49:18.050 --> 00:49:23.630 align:middle line:84% random nucleic acid sequence that basically incredibly 00:49:23.630 --> 00:49:26.690 align:middle line:84% accelerates the rate of mutagenesis, basically 00:49:26.690 --> 00:49:28.130 align:middle line:90% generating sequence de novo. 00:49:28.130 --> 00:49:30.590 align:middle line:84% This is one of the examples in biology 00:49:30.590 --> 00:49:33.050 align:middle line:84% where you generate sequence de novo. 00:49:33.050 --> 00:49:40.040 align:middle line:84% And I think that's very apropos of this combinatorial topic. 00:49:40.040 --> 00:49:44.030 align:middle line:84% Now, this is a beautiful example from the previous slide. 00:49:44.030 --> 00:49:45.800 align:middle line:84% Those polyketides on the far right 00:49:45.800 --> 00:49:48.300 align:middle line:90% now are the star of this slide. 00:49:48.300 --> 00:49:50.960 align:middle line:84% And here, in a certain sense, nature has-- 00:49:50.960 --> 00:49:54.470 align:middle line:84% and now scientists have-- engineered protein modules 00:49:54.470 --> 00:49:59.240 align:middle line:84% to make this linear sequence of events. 00:49:59.240 --> 00:50:03.200 align:middle line:84% You can think of, here, you're using a linear set 00:50:03.200 --> 00:50:07.280 align:middle line:84% of protein domains to program very complicated series 00:50:07.280 --> 00:50:09.650 align:middle line:84% of chemical reactions, the same way the linear sequence 00:50:09.650 --> 00:50:12.530 align:middle line:84% of messenger RNA tells the ribosome 00:50:12.530 --> 00:50:16.680 align:middle line:84% to make a series of additions in the protein. 00:50:16.680 --> 00:50:21.230 align:middle line:84% The ribosome catalytic cycle is a cycle, 00:50:21.230 --> 00:50:24.200 align:middle line:84% while this is more a linear tape, 00:50:24.200 --> 00:50:26.390 align:middle line:90% a linear series of events. 00:50:26.390 --> 00:50:29.000 align:middle line:84% The proteins themselves, these little arrow-shaped things 00:50:29.000 --> 00:50:32.150 align:middle line:84% with boxes in them along the top, 00:50:32.150 --> 00:50:35.960 align:middle line:84% labeled Module 1 through 6, those proteins are, of course, 00:50:35.960 --> 00:50:36.860 align:middle line:90% made on ribosomes. 00:50:36.860 --> 00:50:40.520 align:middle line:84% But then they act kind of like the solid phase synthesis, 00:50:40.520 --> 00:50:43.160 align:middle line:84% where the acyl carrier protein, ACP in the box, 00:50:43.160 --> 00:50:46.070 align:middle line:84% binds to the first monomer, and it 00:50:46.070 --> 00:50:48.710 align:middle line:84% starts transferring it from protein 00:50:48.710 --> 00:50:53.977 align:middle line:84% to protein along this multi-domain huge protein. 00:50:53.977 --> 00:50:56.060 align:middle line:84% And there's actually three proteins in a row here. 00:50:56.060 --> 00:50:59.120 align:middle line:84% And each of the steps are taken in order along the protein, 00:50:59.120 --> 00:51:02.210 align:middle line:84% and they involve things like a synthase step, where you bring 00:51:02.210 --> 00:51:06.290 align:middle line:84% in another monomer, or a keto reductase step, KR, 00:51:06.290 --> 00:51:08.540 align:middle line:90% where you'll bring in-- 00:51:08.540 --> 00:51:13.340 align:middle line:84% where you'll reduce one of the double bonds, 00:51:13.340 --> 00:51:16.910 align:middle line:90% or an acyltransferase step, AT. 00:51:16.910 --> 00:51:20.990 align:middle line:90% 00:51:20.990 --> 00:51:24.020 align:middle line:84% But you can see each of these has a substrate specificity. 00:51:24.020 --> 00:51:27.470 align:middle line:84% And by changing the order of substrate specificities, 00:51:27.470 --> 00:51:31.700 align:middle line:84% you can build up a huge combinatorial collection here 00:51:31.700 --> 00:51:38.570 align:middle line:84% in microbial communities, and also in the laboratory. 00:51:38.570 --> 00:51:40.220 align:middle line:90% Now, protein interaction. 00:51:40.220 --> 00:51:43.580 align:middle line:84% We're just beginning to talk seriously about protein 00:51:43.580 --> 00:51:45.680 align:middle line:90% interaction assays. 00:51:45.680 --> 00:51:48.710 align:middle line:84% I think many of you either are or will 00:51:48.710 --> 00:51:52.850 align:middle line:84% be more and more familiar with the protein interaction assays. 00:51:52.850 --> 00:51:56.570 align:middle line:84% In next class, we'll talk about ways 00:51:56.570 --> 00:51:59.400 align:middle line:84% of getting direct information on protein 00:51:59.400 --> 00:52:02.210 align:middle line:84% through cross-linking and mass spectrometry. 00:52:02.210 --> 00:52:04.370 align:middle line:84% Another way is indirectly setting up 00:52:04.370 --> 00:52:06.500 align:middle line:84% these reporter assays, where you take 00:52:06.500 --> 00:52:09.530 align:middle line:84% advantage of the binding properties of known proteins 00:52:09.530 --> 00:52:12.410 align:middle line:90% to analyze two unknown proteins. 00:52:12.410 --> 00:52:14.930 align:middle line:84% And so a known protein might be, [INAUDIBLE],, 00:52:14.930 --> 00:52:19.070 align:middle line:84% which binds to DNA, and B42, which activates transcription 00:52:19.070 --> 00:52:23.630 align:middle line:84% of something for which you have a good visual assay, 00:52:23.630 --> 00:52:27.560 align:middle line:90% like [? URO3, ?] life and death. 00:52:27.560 --> 00:52:30.800 align:middle line:84% And by taking these two knowns, let's say, 00:52:30.800 --> 00:52:33.200 align:middle line:84% in B42, which have known properties-- 00:52:33.200 --> 00:52:35.562 align:middle line:84% but they only exert their properties 00:52:35.562 --> 00:52:37.520 align:middle line:84% when they're brought together, and they're only 00:52:37.520 --> 00:52:40.220 align:middle line:84% brought together if the unknown proteins or partially 00:52:40.220 --> 00:52:44.460 align:middle line:84% known proteins to which they're bound interact with each other. 00:52:44.460 --> 00:52:47.450 align:middle line:84% And so this is a so-called two-hybrid assay 00:52:47.450 --> 00:52:49.920 align:middle line:90% and variations on it. 00:52:49.920 --> 00:52:52.460 align:middle line:84% And I think we mentioned one where 00:52:52.460 --> 00:52:55.220 align:middle line:84% you can characterize nucleic acid-protein interactions 00:52:55.220 --> 00:52:56.600 align:middle line:90% with one hybrid assay. 00:52:56.600 --> 00:53:00.680 align:middle line:84% And here, you can inhibit this interaction 00:53:00.680 --> 00:53:03.590 align:middle line:90% between the two knowns-- 00:53:03.590 --> 00:53:07.970 align:middle line:84% sorry, unknowns or partially known molecules in blue here. 00:53:07.970 --> 00:53:16.040 align:middle line:84% Here's a TGF beta, a growth factor, and a binding protein. 00:53:16.040 --> 00:53:19.880 align:middle line:84% You can inhibit that with a particular small molecule 00:53:19.880 --> 00:53:21.860 align:middle line:84% or a collection of small molecules, which 00:53:21.860 --> 00:53:25.580 align:middle line:84% can be introduced from one of these combinatorial syntheses 00:53:25.580 --> 00:53:29.660 align:middle line:84% into an array of these cellular assays. 00:53:29.660 --> 00:53:34.040 align:middle line:84% And you get this information about-- 00:53:34.040 --> 00:53:36.020 align:middle line:84% you can either collect a big data 00:53:36.020 --> 00:53:39.500 align:middle line:84% set of proteins that interact from a proteomic scale 00:53:39.500 --> 00:53:42.920 align:middle line:84% experiment, or from molecules that inhibit one or more 00:53:42.920 --> 00:53:44.550 align:middle line:90% of those interactions. 00:53:44.550 --> 00:53:47.240 align:middle line:84% This is a source of information, which 00:53:47.240 --> 00:53:50.292 align:middle line:84% is intrinsically computational in the sense-- well, 00:53:50.292 --> 00:53:52.250 align:middle line:84% in the sense that there's a large amount of it. 00:53:52.250 --> 00:53:54.620 align:middle line:84% You can model the three-dimensional structure 00:53:54.620 --> 00:53:57.600 align:middle line:84% of this interaction if you have sufficient data to do that. 00:53:57.600 --> 00:54:00.740 align:middle line:84% And you can model the impact of the small molecules 00:54:00.740 --> 00:54:03.410 align:middle line:84% in a structure activity [? since. ?] 00:54:03.410 --> 00:54:08.390 align:middle line:84% Now, if you look at the top right-hand part of slide 34 00:54:08.390 --> 00:54:10.790 align:middle line:84% here, you can see this huge diversity 00:54:10.790 --> 00:54:12.690 align:middle line:84% of all of these different colored shapes. 00:54:12.690 --> 00:54:17.570 align:middle line:84% And if you wanted to use these in a combinatorial assay, 00:54:17.570 --> 00:54:20.960 align:middle line:84% you'd connect them in every pairwise combination 00:54:20.960 --> 00:54:24.530 align:middle line:84% and try them against your target by some bioassay. 00:54:24.530 --> 00:54:28.280 align:middle line:84% However, if that library is too large 00:54:28.280 --> 00:54:29.930 align:middle line:90% either to make or to screen-- 00:54:29.930 --> 00:54:31.880 align:middle line:84% typically, to screen-- then what you can do 00:54:31.880 --> 00:54:35.880 align:middle line:84% is, you can study a part of the molecule 00:54:35.880 --> 00:54:39.960 align:middle line:84% and see-- and then take the subset of the diversity that 00:54:39.960 --> 00:54:47.130 align:middle line:84% can bind, and then take that subset, 00:54:47.130 --> 00:54:50.760 align:middle line:84% characterize it, and now make only the pairwise combinations 00:54:50.760 --> 00:54:53.470 align:middle line:90% of the two half-molecules. 00:54:53.470 --> 00:55:00.720 align:middle line:84% And generally speaking, if the geometry is fairly rigid, 00:55:00.720 --> 00:55:05.040 align:middle line:84% then the binding constant will get-- 00:55:05.040 --> 00:55:08.230 align:middle line:84% will be roughly the product of the two binding constants. 00:55:08.230 --> 00:55:10.890 align:middle line:84% So if each one has a very low binding constant, 00:55:10.890 --> 00:55:14.830 align:middle line:84% then it will be roughly the square of that. 00:55:14.830 --> 00:55:18.780 align:middle line:84% And you get some point of diminishing returns, 00:55:18.780 --> 00:55:21.210 align:middle line:90% eventually. 00:55:21.210 --> 00:55:24.240 align:middle line:84% So this is an example of a strategy where 00:55:24.240 --> 00:55:27.900 align:middle line:84% you use a little bit of prior knowledge, which 00:55:27.900 --> 00:55:32.460 align:middle line:84% can be empirical or it could be purely computational, about how 00:55:32.460 --> 00:55:36.030 align:middle line:84% to limit your library and make interesting combinations. 00:55:36.030 --> 00:55:39.860 align:middle line:90% 00:55:39.860 --> 00:55:42.800 align:middle line:84% Now, we've been talking mainly about the kind of chemical 00:55:42.800 --> 00:55:45.350 align:middle line:84% diversity we can get that's aimed at the ligands that 00:55:45.350 --> 00:55:46.970 align:middle line:90% combine to target proteins. 00:55:46.970 --> 00:55:50.780 align:middle line:84% But now, we want to talk about the source of information 00:55:50.780 --> 00:55:52.860 align:middle line:84% about those target proteins itself, 00:55:52.860 --> 00:56:01.250 align:middle line:84% which is another genomics project, structural genomics. 00:56:01.250 --> 00:56:07.250 align:middle line:84% And typically, we want to select targets for binding drugs, 00:56:07.250 --> 00:56:10.760 align:middle line:84% or select targets for solving the structures of proteins 00:56:10.760 --> 00:56:14.720 align:middle line:84% in order to look at their ligands in more detail. 00:56:14.720 --> 00:56:17.900 align:middle line:90% And how can we do this? 00:56:17.900 --> 00:56:20.830 align:middle line:84% How can we decide which targets are high on our list 00:56:20.830 --> 00:56:21.850 align:middle line:90% to go for next? 00:56:21.850 --> 00:56:23.650 align:middle line:84% We have hundreds of proteins for which we 00:56:23.650 --> 00:56:25.240 align:middle line:84% have three-dimensional structures, 00:56:25.240 --> 00:56:29.770 align:middle line:84% and from some of them, we have information 00:56:29.770 --> 00:56:32.530 align:middle line:90% on what ligands they bind. 00:56:32.530 --> 00:56:34.210 align:middle line:84% But these are other criteria that 00:56:34.210 --> 00:56:38.170 align:middle line:84% are sometimes used in the field for target selection. 00:56:38.170 --> 00:56:44.050 align:middle line:84% If they are homologous to previous interesting targets, 00:56:44.050 --> 00:56:48.190 align:middle line:84% then that puts them high on the short list. 00:56:48.190 --> 00:56:50.380 align:middle line:84% If they are conserved-- and we've talked about how 00:56:50.380 --> 00:56:53.620 align:middle line:84% important conservation is from time to time-- 00:56:53.620 --> 00:56:56.650 align:middle line:90% then that might be an approach. 00:56:56.650 --> 00:57:01.640 align:middle line:84% If they're conserved and you knock it out, 00:57:01.640 --> 00:57:03.640 align:middle line:84% then you might expect that to be lethal. 00:57:03.640 --> 00:57:10.510 align:middle line:84% And that might make it a good target for an anti-bacterial. 00:57:10.510 --> 00:57:18.700 align:middle line:84% If you want to limit the action of your therapy to the surface 00:57:18.700 --> 00:57:23.980 align:middle line:84% in order to, say, reduce the cross-reaction 00:57:23.980 --> 00:57:28.270 align:middle line:84% with internal molecules, you can sometimes 00:57:28.270 --> 00:57:31.360 align:middle line:84% restrict yourself to the surface-acceptable proteins. 00:57:31.360 --> 00:57:34.000 align:middle line:84% And in fact, a very large class of drugs 00:57:34.000 --> 00:57:39.500 align:middle line:84% is aimed at surface-accessible membrane proteins. 00:57:39.500 --> 00:57:42.100 align:middle line:84% And so very often, those are prioritized high. 00:57:42.100 --> 00:57:44.590 align:middle line:84% There are also-- the surface-acceptable proteins are 00:57:44.590 --> 00:57:47.720 align:middle line:84% important if you're talking about vaccines, 00:57:47.720 --> 00:57:50.500 align:middle line:84% which is an increasingly important-- 00:57:50.500 --> 00:57:55.330 align:middle line:84% or diagnostics that are non-invasive, 00:57:55.330 --> 00:57:59.620 align:middle line:84% or at least not going to the interior of the cell. 00:57:59.620 --> 00:58:02.110 align:middle line:84% And there are ways, say, with microarrays, 00:58:02.110 --> 00:58:04.180 align:middle line:84% that you can ask which genes are differentially 00:58:04.180 --> 00:58:05.590 align:middle line:90% expressed in the disease state. 00:58:05.590 --> 00:58:07.780 align:middle line:84% And that causes high prioritization. 00:58:07.780 --> 00:58:10.750 align:middle line:84% Now, once you have that prioritization, which 00:58:10.750 --> 00:58:13.720 align:middle line:84% comes from, say, genome sequencing 00:58:13.720 --> 00:58:17.680 align:middle line:84% and some of those other facts, then we 00:58:17.680 --> 00:58:19.570 align:middle line:84% would like to have-- you've got your target. 00:58:19.570 --> 00:58:21.400 align:middle line:84% You've got your genome sequence-- 00:58:21.400 --> 00:58:24.340 align:middle line:90% 00:58:24.340 --> 00:58:26.152 align:middle line:90% gene sequence for that target. 00:58:26.152 --> 00:58:28.360 align:middle line:84% How, then, do you get the three-dimensional structure 00:58:28.360 --> 00:58:32.680 align:middle line:84% that helps you design drugs or improve 00:58:32.680 --> 00:58:34.330 align:middle line:90% the drugs that you have? 00:58:34.330 --> 00:58:42.280 align:middle line:84% Well, one very attractive approach, 00:58:42.280 --> 00:58:44.920 align:middle line:84% given a protein sequence which might 00:58:44.920 --> 00:58:48.430 align:middle line:84% get from their deluge of genome sequences, 00:58:48.430 --> 00:58:52.150 align:middle line:84% the practical approach might be to start 00:58:52.150 --> 00:58:54.370 align:middle line:84% with this genome sequence-- this gene sequence which 00:58:54.370 --> 00:58:59.410 align:middle line:84% is 99.99% accurate, and try to predict the three dimensional 00:58:59.410 --> 00:59:02.410 align:middle line:84% structure of the protein and its ligand specificity. 00:59:02.410 --> 00:59:04.105 align:middle line:84% And if you walk through these, these 00:59:04.105 --> 00:59:05.980 align:middle line:84% are kind of ballpark estimates, some of them 00:59:05.980 --> 00:59:07.060 align:middle line:90% better than others. 00:59:07.060 --> 00:59:08.680 align:middle line:84% To get from the sequence to exons-- 00:59:08.680 --> 00:59:10.120 align:middle line:90% we talked about this before-- 00:59:10.120 --> 00:59:12.065 align:middle line:90% might be an 80%. 00:59:12.065 --> 00:59:14.440 align:middle line:84% Remember, these numbers really should be false positives, 00:59:14.440 --> 00:59:16.160 align:middle line:90% false negatives, so on. 00:59:16.160 --> 00:59:21.670 align:middle line:84% But this ballpark, 80%, getting to exons, exons to genes. 00:59:21.670 --> 00:59:25.660 align:middle line:84% If you aren't privileged enough to have the cDNA, 00:59:25.660 --> 00:59:29.800 align:middle line:84% then this is an error-prone process 00:59:29.800 --> 00:59:34.120 align:middle line:90% with maybe a 30% success rate. 00:59:34.120 --> 00:59:36.795 align:middle line:84% Once you have genes knowing it's regulation, 00:59:36.795 --> 00:59:38.920 align:middle line:84% whether it's on or off in the particular cell types 00:59:38.920 --> 00:59:43.090 align:middle line:84% you're interested in, is very difficult right now. 00:59:43.090 --> 00:59:47.410 align:middle line:84% Knowing the motifs is barely a start 00:59:47.410 --> 00:59:48.950 align:middle line:90% on getting the full regulation. 00:59:48.950 --> 00:59:51.670 align:middle line:90% So I would say 10% or less. 00:59:51.670 --> 00:59:54.280 align:middle line:84% Once you have a regulated gene, getting the protein sequence 00:59:54.280 --> 00:59:55.030 align:middle line:90% is easy. 00:59:55.030 --> 00:59:56.380 align:middle line:90% That's the genetic code. 00:59:56.380 --> 00:59:59.110 align:middle line:84% Getting from the sequence to secondary structure 00:59:59.110 --> 01:00:03.520 align:middle line:84% is easy in the context of some of these other things, 01:00:03.520 --> 01:00:06.880 align:middle line:84% but still, the accuracy is only around 77%. 01:00:06.880 --> 01:00:10.240 align:middle line:84% I have next to this the reference is [? cast, ?] 01:00:10.240 --> 01:00:13.540 align:middle line:84% which is something that is a competition 01:00:13.540 --> 01:00:16.420 align:middle line:84% for computational assessment of three-dimensional structure 01:00:16.420 --> 01:00:20.110 align:middle line:84% prediction for proteins that's been held since, I think, '94, 01:00:20.110 --> 01:00:22.750 align:middle line:84% and the next one is coming up in a couple of months. 01:00:22.750 --> 01:00:26.270 align:middle line:84% And this is kind of the big race or bake-off 01:00:26.270 --> 01:00:27.520 align:middle line:90% between the different methods. 01:00:27.520 --> 01:00:28.810 align:middle line:90% Very exciting. 01:00:28.810 --> 01:00:31.080 align:middle line:84% But unfortunately, over decades, it's 01:00:31.080 --> 01:00:34.210 align:middle line:84% still hovering around 77% for secondary structure 01:00:34.210 --> 01:00:39.010 align:middle line:84% and about 25% for ab initio three-dimensional structure. 01:00:39.010 --> 01:00:41.560 align:middle line:84% Then even if you have the three-dimensional structure 01:00:41.560 --> 01:00:45.250 align:middle line:84% at adequate accuracy, getting the ligand specificity 01:00:45.250 --> 01:00:48.760 align:middle line:84% is problematic, and it depends on the ligand. 01:00:48.760 --> 01:00:50.620 align:middle line:90% If it's DNA, it's a good case. 01:00:50.620 --> 01:00:54.580 align:middle line:84% If it's a small molecule, it can be as low as 10% or worse. 01:00:54.580 --> 01:00:56.500 align:middle line:84% Now, since each of these are fairly 01:00:56.500 --> 01:01:01.600 align:middle line:84% independent of estimates, you can 01:01:01.600 --> 01:01:04.940 align:middle line:84% get an approximate overall accuracy 01:01:04.940 --> 01:01:06.850 align:middle line:84% which is a product of all of those, which 01:01:06.850 --> 01:01:13.400 align:middle line:90% is dismally small at 0.0005. 01:01:13.400 --> 01:01:18.490 align:middle line:84% And so it behooves one to use as many extra-experimental data 01:01:18.490 --> 01:01:21.580 align:middle line:84% as one can, or improve the algorithms that 01:01:21.580 --> 01:01:25.780 align:middle line:84% are weakest in this journey from genome sequence 01:01:25.780 --> 01:01:28.870 align:middle line:90% to ligand specificity. 01:01:28.870 --> 01:01:33.190 align:middle line:84% We'll pick up this thread right after a break 01:01:33.190 --> 01:01:37.420 align:middle line:84% and carry on to actually how we get 01:01:37.420 --> 01:01:40.720 align:middle line:84% the three-dimensional structure, whether it's 01:01:40.720 --> 01:01:45.190 align:middle line:84% predictive or experimental, and the computational tasks there. 01:01:45.190 --> 01:01:46.570 align:middle line:90% Thank you. 01:01:46.570 --> 01:01:48.360 align:middle line:90% Take a break. 01:01:48.360 --> 01:01:49.000 align:middle line:90%