WEBVTT 00:00:00.000 --> 00:00:00.543 align:middle line:90% 00:00:00.543 --> 00:00:01.960 align:middle line:84% NARRATOR: The following content is 00:00:01.960 --> 00:00:05.290 align:middle line:84% provided by MIT OpenCourseWare under a Creative Commons 00:00:05.290 --> 00:00:06.370 align:middle line:90% license. 00:00:06.370 --> 00:00:10.120 align:middle line:84% Additional information about our license and MIT OpenCourseWare, 00:00:10.120 --> 00:00:13.090 align:middle line:84% in general, is available at ocw.mit.edu. 00:00:13.090 --> 00:00:17.722 align:middle line:90% 00:00:17.722 --> 00:00:19.500 align:middle line:84% DR. GEORGE CHURCH: OK, welcome back. 00:00:19.500 --> 00:00:23.850 align:middle line:84% One quick announcement before we get started. 00:00:23.850 --> 00:00:29.970 align:middle line:84% The teaching Fellows and I have, in response 00:00:29.970 --> 00:00:33.300 align:middle line:84% to a few gentle inquiries have, come up 00:00:33.300 --> 00:00:35.670 align:middle line:84% with a plan for making your problem sets 00:00:35.670 --> 00:00:39.310 align:middle line:84% and projects a little bit easier, getting you some time. 00:00:39.310 --> 00:00:42.720 align:middle line:84% Take a look at the website and talk to your teaching Fellows. 00:00:42.720 --> 00:00:45.240 align:middle line:84% But basically, there will be no problems in set six. 00:00:45.240 --> 00:00:47.370 align:middle line:84% It will be combined with problem set five, 00:00:47.370 --> 00:00:51.480 align:middle line:84% and so you'll have a full three weeks to work on your project. 00:00:51.480 --> 00:00:53.760 align:middle line:90% Tripling your project time. 00:00:53.760 --> 00:00:55.510 align:middle line:84% Assuming you haven't done anything so far, 00:00:55.510 --> 00:00:58.870 align:middle line:84% which hopefully all of you have done a lot so far. 00:00:58.870 --> 00:01:08.130 align:middle line:84% OK, so we were correlating absolute levels 00:01:08.130 --> 00:01:14.190 align:middle line:84% of messenger RNA abundance in messenger RNA protein. 00:01:14.190 --> 00:01:18.420 align:middle line:84% And here, this study was subjected 00:01:18.420 --> 00:01:22.845 align:middle line:84% to a little bit of not terribly controversial critique. 00:01:22.845 --> 00:01:25.890 align:middle line:90% 00:01:25.890 --> 00:01:29.910 align:middle line:84% What you're seeing here is at very-- 00:01:29.910 --> 00:01:32.190 align:middle line:84% if you have the very low abundance proteins, 00:01:32.190 --> 00:01:35.370 align:middle line:84% you might have what looks like some correlation. 00:01:35.370 --> 00:01:37.440 align:middle line:84% Then you add a few more low abundance proteins, 00:01:37.440 --> 00:01:39.480 align:middle line:84% so this is just some fluctuation. 00:01:39.480 --> 00:01:43.930 align:middle line:84% And adding proteins should improve your correlation 00:01:43.930 --> 00:01:44.430 align:middle line:90% coefficient. 00:01:44.430 --> 00:01:47.610 align:middle line:84% But if they're of low reliability in either 00:01:47.610 --> 00:01:50.070 align:middle line:84% the protein of the messenger RNA scale, 00:01:50.070 --> 00:01:51.990 align:middle line:84% or if there isn't correlation between the two 00:01:51.990 --> 00:01:55.200 align:middle line:84% for biological reasons, then your correlation coefficient 00:01:55.200 --> 00:01:57.190 align:middle line:90% could drop just at random. 00:01:57.190 --> 00:02:01.180 align:middle line:84% But as you add-- as you get up to adding all the proteins, 00:02:01.180 --> 00:02:05.010 align:middle line:84% you could either be dominated by a few high abundance 00:02:05.010 --> 00:02:07.920 align:middle line:84% proteins that fit this perfectly, 00:02:07.920 --> 00:02:12.180 align:middle line:84% the messenger RNA protein abundance. 00:02:12.180 --> 00:02:14.670 align:middle line:84% Anyway, some of the critiques of this 00:02:14.670 --> 00:02:19.170 align:middle line:84% had to do with the assumptions underlying the Pearson 00:02:19.170 --> 00:02:21.330 align:middle line:84% linear correlation coefficient, which 00:02:21.330 --> 00:02:24.570 align:middle line:84% in calculating the statistical significance of some 00:02:24.570 --> 00:02:27.630 align:middle line:84% of these analyzes in the previous slide, 00:02:27.630 --> 00:02:30.180 align:middle line:84% makes an underlying assumption that you 00:02:30.180 --> 00:02:33.750 align:middle line:84% have a normal or Gaussian bell curve. 00:02:33.750 --> 00:02:36.790 align:middle line:84% And you don't need to do this in order 00:02:36.790 --> 00:02:39.160 align:middle line:84% to-- the variety of measures of correlation, 00:02:39.160 --> 00:02:43.080 align:middle line:84% which do not require this parametric assumption. 00:02:43.080 --> 00:02:48.420 align:middle line:84% And there are tests for how close to normal a distribution 00:02:48.420 --> 00:02:50.220 align:middle line:90% is. 00:02:50.220 --> 00:02:53.130 align:middle line:84% Deviations from normality can of all types. 00:02:53.130 --> 00:02:55.380 align:middle line:84% For example, you can have slightly flatter or slightly 00:02:55.380 --> 00:02:56.940 align:middle line:90% sharper than normal. 00:02:56.940 --> 00:03:00.130 align:middle line:84% You can have skewed to the left or to the right, and so forth. 00:03:00.130 --> 00:03:03.540 align:middle line:84% Some of these, like skewing, can be 00:03:03.540 --> 00:03:05.850 align:middle line:84% corrected by a log transformation, where 00:03:05.850 --> 00:03:08.640 align:middle line:84% you simply take the logarithm or some other transformation, 00:03:08.640 --> 00:03:11.910 align:middle line:84% and a log being by far the most common, and theoretically 00:03:11.910 --> 00:03:14.190 align:middle line:84% justified, and now it becomes normal, 00:03:14.190 --> 00:03:16.688 align:middle line:84% and then you can do a statistical test. 00:03:16.688 --> 00:03:18.480 align:middle line:84% As it turns out, this is not a huge effect, 00:03:18.480 --> 00:03:25.200 align:middle line:84% but they use it in order to point out 00:03:25.200 --> 00:03:27.660 align:middle line:84% that you can when you're testing these, especially 00:03:27.660 --> 00:03:31.150 align:middle line:84% the low abundance end of the spectrum. 00:03:31.150 --> 00:03:33.780 align:middle line:84% You might want to use a rank test. 00:03:33.780 --> 00:03:36.120 align:middle line:84% And a rank test here is illustrated 00:03:36.120 --> 00:03:43.680 align:middle line:84% that you take if you have, say, two columns, this 00:03:43.680 --> 00:03:49.410 align:middle line:84% is a series of pairs of intensities of messenger RNA 00:03:49.410 --> 00:03:50.010 align:middle line:90% and proteins. 00:03:50.010 --> 00:03:54.900 align:middle line:84% Say columns X and column Y down in the lower right-hand corner 00:03:54.900 --> 00:03:56.760 align:middle line:90% of Slide 37. 00:03:56.760 --> 00:04:00.390 align:middle line:84% And let's say the abundances, the absolute abundances 00:04:00.390 --> 00:04:08.940 align:middle line:84% for protein X is 1, 6, 6, and corresponding RNA Y, which 00:04:08.940 --> 00:04:11.640 align:middle line:90% would be 8, 2, 3, and 4. 00:04:11.640 --> 00:04:13.800 align:middle line:84% Now, you want to ask whether the correlate or not. 00:04:13.800 --> 00:04:18.329 align:middle line:84% And what you do is rank them, and so the rank of X 00:04:18.329 --> 00:04:19.680 align:middle line:90% is 1, 3, 3. 00:04:19.680 --> 00:04:21.180 align:middle line:84% Here is a tiebreaker-- the way you 00:04:21.180 --> 00:04:24.660 align:middle line:84% deal with a tie is you give them all the rank of the middle one 00:04:24.660 --> 00:04:27.300 align:middle line:90% of the series in the pi. 00:04:27.300 --> 00:04:31.530 align:middle line:84% And you get a rank for Y, and the total number, in this case, 00:04:31.530 --> 00:04:40.740 align:middle line:84% is 4, and so when you have the rank test score 00:04:40.740 --> 00:04:42.900 align:middle line:84% is basically the sum of the square 00:04:42.900 --> 00:04:44.320 align:middle line:90% of all the differences in rank. 00:04:44.320 --> 00:04:46.890 align:middle line:84% So the difference in rank here would be 1 minus 4. 00:04:46.890 --> 00:04:49.943 align:middle line:90% 00:04:49.943 --> 00:04:53.220 align:middle line:84% It'd be 3 squared, and you take the sum of all the squares 00:04:53.220 --> 00:04:55.140 align:middle line:84% and you plug it into this S which 00:04:55.140 --> 00:04:58.110 align:middle line:84% is going to go into a correlation coefficient. 00:04:58.110 --> 00:05:01.020 align:middle line:84% Similar to in values and hypothesis 00:05:01.020 --> 00:05:03.630 align:middle line:84% to the Pearson, but now not making assumptions 00:05:03.630 --> 00:05:06.000 align:middle line:84% about the parametric, you just talking about ranks, 00:05:06.000 --> 00:05:08.130 align:middle line:90% it's a non-parametric test. 00:05:08.130 --> 00:05:12.390 align:middle line:84% And then, the N is the total number, 00:05:12.390 --> 00:05:13.890 align:middle line:90% and you apply that formula. 00:05:13.890 --> 00:05:15.690 align:middle line:84% Now, they apply that formula to this 00:05:15.690 --> 00:05:18.140 align:middle line:84% exact roughly the same data set or very similar. 00:05:18.140 --> 00:05:22.620 align:middle line:90% Actually data set. 00:05:22.620 --> 00:05:29.750 align:middle line:84% Another critique they had was that using a better 00:05:29.750 --> 00:05:33.620 align:middle line:84% measure of the protein abundance would 00:05:33.620 --> 00:05:39.080 align:middle line:84% be using a radioactive tracer in the protein. 00:05:39.080 --> 00:05:41.870 align:middle line:84% And then measuring quantitatively 00:05:41.870 --> 00:05:48.110 align:middle line:84% the intensity of beta particles released from the methionine 00:05:48.110 --> 00:05:48.800 align:middle line:90% in the proteins. 00:05:48.800 --> 00:05:52.880 align:middle line:90% 00:05:52.880 --> 00:05:58.220 align:middle line:84% Making that change and comparing it to the same messenger RNA 00:05:58.220 --> 00:06:03.830 align:middle line:84% assessment, they got this fairly linear correlation over three 00:06:03.830 --> 00:06:05.540 align:middle line:90% logs. 00:06:05.540 --> 00:06:10.410 align:middle line:84% Using the Pearson correlation coefficient, 00:06:10.410 --> 00:06:12.080 align:middle line:84% which they didn't entirely approve of, 00:06:12.080 --> 00:06:16.310 align:middle line:84% they got a 0.76, which is a modest-- 00:06:16.310 --> 00:06:19.280 align:middle line:90% it's a significant linear trend. 00:06:19.280 --> 00:06:23.600 align:middle line:84% And using their rank method 0.74, which 00:06:23.600 --> 00:06:27.770 align:middle line:90% is basically very similar. 00:06:27.770 --> 00:06:31.610 align:middle line:84% And found no significant difference between the top 33% 00:06:31.610 --> 00:06:33.770 align:middle line:90% and the bottom 33 proteins. 00:06:33.770 --> 00:06:37.760 align:middle line:84% Undermining the previous claim that there 00:06:37.760 --> 00:06:39.965 align:middle line:84% was less linearity at one than in another. 00:06:39.965 --> 00:06:44.102 align:middle line:84% You might, or you expect that the least abundant majority 00:06:44.102 --> 00:06:45.560 align:middle line:84% of proteins would have a little bit 00:06:45.560 --> 00:06:49.190 align:middle line:84% of either biological or instrumental noise. 00:06:49.190 --> 00:06:53.640 align:middle line:84% Nevertheless, this group found that it was a good correlation. 00:06:53.640 --> 00:06:56.360 align:middle line:84% Now, these two plot-- the plot in the next slide 00:06:56.360 --> 00:07:00.890 align:middle line:84% looks similar, but it's really quite different. 00:07:00.890 --> 00:07:04.100 align:middle line:90% Here the Y-axis remains. 00:07:04.100 --> 00:07:10.670 align:middle line:84% The protein abundance is measured by this F 35 labeling, 00:07:10.670 --> 00:07:14.330 align:middle line:84% but now we're getting back to this game of asking 00:07:14.330 --> 00:07:16.910 align:middle line:84% to what extent can we make predictions 00:07:16.910 --> 00:07:18.710 align:middle line:84% about the properties of the proteins? 00:07:18.710 --> 00:07:21.200 align:middle line:84% In this case, their abundance is based on their use 00:07:21.200 --> 00:07:24.080 align:middle line:90% of abundant codons. 00:07:24.080 --> 00:07:29.060 align:middle line:84% And a way of quantitating this is coadaptation index. 00:07:29.060 --> 00:07:31.670 align:middle line:90% 00:07:31.670 --> 00:07:37.280 align:middle line:84% Shown in the lower left part of the slide is you have-- 00:07:37.280 --> 00:07:40.280 align:middle line:84% it's the log of the coadaptation index, 00:07:40.280 --> 00:07:45.590 align:middle line:84% is a sum of all the frequencies F of I of each of the codons. 00:07:45.590 --> 00:07:51.180 align:middle line:84% Where I is the 61 non-stop codons out of 64 total. 00:07:51.180 --> 00:07:51.680 align:middle line:90% And 00:07:51.680 --> 00:07:54.920 align:middle line:84% The W sub I is a waiting factor, where 00:07:54.920 --> 00:07:58.970 align:middle line:84% it's the ratio of the frequency of codon I. 00:07:58.970 --> 00:08:01.910 align:middle line:84% Let's say a leucine codon there are six different leucine 00:08:01.910 --> 00:08:07.790 align:middle line:84% codons, and say the first one, the W sub 1 00:08:07.790 --> 00:08:10.220 align:middle line:84% is going to be the ratio of the frequency 00:08:10.220 --> 00:08:12.590 align:middle line:84% of that first codon to whichever one happens 00:08:12.590 --> 00:08:13.670 align:middle line:90% to be most abundant one. 00:08:13.670 --> 00:08:16.640 align:middle line:84% Could be the first one, or it could be one or the other six. 00:08:16.640 --> 00:08:20.220 align:middle line:84% And so, that's the formula that is used. 00:08:20.220 --> 00:08:24.980 align:middle line:84% And you can see, again, a nice linear trend where indeed, 00:08:24.980 --> 00:08:27.770 align:middle line:84% the most abundant proteins do tend to use 00:08:27.770 --> 00:08:30.860 align:middle line:90% the most abundant codons. 00:08:30.860 --> 00:08:34.280 align:middle line:84% The codons that you find that are most abundant 00:08:34.280 --> 00:08:36.559 align:middle line:84% both in transponder RNA level and in usage 00:08:36.559 --> 00:08:38.730 align:middle line:90% in abundant proteins. 00:08:38.730 --> 00:08:41.840 align:middle line:84% Now, just as with RNAs, you can measure them 00:08:41.840 --> 00:08:44.510 align:middle line:84% on an absolute scale or a ratio scale. 00:08:44.510 --> 00:08:46.790 align:middle line:84% The advantage in principle if you have it 00:08:46.790 --> 00:08:48.290 align:middle line:84% on an absolute scale, you can always 00:08:48.290 --> 00:08:49.500 align:middle line:90% calculate ratios from it. 00:08:49.500 --> 00:08:50.880 align:middle line:90% But not necessarily vice versa. 00:08:50.880 --> 00:08:54.560 align:middle line:84% You have ratios you can't always get to absolute. 00:08:54.560 --> 00:08:58.640 align:middle line:84% So that's one advantage to absolute. 00:08:58.640 --> 00:09:01.460 align:middle line:84% If you're looking at things like codon adaptation index 00:09:01.460 --> 00:09:07.130 align:middle line:84% or messenger RNA, or a variety of other motivations for you, 00:09:07.130 --> 00:09:09.950 align:middle line:84% need to do it on the absolute scale. 00:09:09.950 --> 00:09:11.570 align:middle line:84% But there, an argument can be made 00:09:11.570 --> 00:09:16.910 align:middle line:84% for doing it on a ratiometric or relative scale, in the sense 00:09:16.910 --> 00:09:19.730 align:middle line:84% that you can establish internal standards which 00:09:19.730 --> 00:09:22.520 align:middle line:90% are more precise. 00:09:22.520 --> 00:09:26.960 align:middle line:84% You can really eliminate many of the systematic errors that 00:09:26.960 --> 00:09:30.230 align:middle line:84% can creep in due to differentiable ionization 00:09:30.230 --> 00:09:32.940 align:middle line:84% in the case of mass spectrometry, and so on. 00:09:32.940 --> 00:09:39.470 align:middle line:84% Now with RNA, the way we did this ratio testing ratio 00:09:39.470 --> 00:09:46.250 align:middle line:84% quantitation is we would label one messenger RNA red or site 00:09:46.250 --> 00:09:51.050 align:middle line:84% 5, and the other one green, and then 00:09:51.050 --> 00:09:54.560 align:middle line:84% by using selective filtration in the imaging, 00:09:54.560 --> 00:09:56.900 align:middle line:90% you could get ratios. 00:09:56.900 --> 00:10:00.920 align:middle line:84% With mass spectrometry, you don't have colors. 00:10:00.920 --> 00:10:03.410 align:middle line:84% But somehow, you want to get the same idea across and so 00:10:03.410 --> 00:10:04.940 align:middle line:90% what you do is have masses. 00:10:04.940 --> 00:10:07.070 align:middle line:84% So what you want to do is encode them 00:10:07.070 --> 00:10:09.110 align:middle line:84% in something that will not change the chemistry, 00:10:09.110 --> 00:10:10.360 align:middle line:90% but will change the mass. 00:10:10.360 --> 00:10:12.860 align:middle line:84% Not change the mass too much because if you change too much, 00:10:12.860 --> 00:10:14.860 align:middle line:84% you might change the chemistry, or you might not 00:10:14.860 --> 00:10:17.860 align:middle line:84% be able to find the shadow peak, the second peak. 00:10:17.860 --> 00:10:22.130 align:middle line:84% So basically, what we're doing is we have a-- 00:10:22.130 --> 00:10:26.380 align:middle line:84% want to do is take self-state 1, label it with a light ICAT 00:10:26.380 --> 00:10:29.110 align:middle line:84% reagent, self-state two, label it with a heavy, 00:10:29.110 --> 00:10:33.220 align:middle line:84% meaning more neutrons in it and then 00:10:33.220 --> 00:10:35.062 align:middle line:84% mix them together as early as possible. 00:10:35.062 --> 00:10:36.520 align:middle line:84% And then all these steps that could 00:10:36.520 --> 00:10:39.700 align:middle line:84% have little systematic errors in it will be the errors 00:10:39.700 --> 00:10:45.100 align:middle line:84% will be equally distributed to each of the test sizes 00:10:45.100 --> 00:10:48.040 align:middle line:84% and labeled and then measure the mass. 00:10:48.040 --> 00:10:51.040 align:middle line:84% And for each mass, you'll have two peaks 00:10:51.040 --> 00:10:53.530 align:middle line:84% that are separated by whatever the difference in mass 00:10:53.530 --> 00:10:54.890 align:middle line:90% between the labeling changes. 00:10:54.890 --> 00:10:57.460 align:middle line:84% So what are the properties that you want of a labeling agent? 00:10:57.460 --> 00:11:00.940 align:middle line:84% One is that it should react covalently 00:11:00.940 --> 00:11:04.330 align:middle line:84% so it will survive all these fractionation 00:11:04.330 --> 00:11:07.480 align:middle line:90% steps and mass detection steps. 00:11:07.480 --> 00:11:10.570 align:middle line:84% So it has to be, in this case, file specific. 00:11:10.570 --> 00:11:13.960 align:middle line:84% You want a way of pulling out only those peptides that 00:11:13.960 --> 00:11:14.830 align:middle line:90% have been modified. 00:11:14.830 --> 00:11:17.230 align:middle line:84% All the rest are just going to contaminate 00:11:17.230 --> 00:11:19.570 align:middle line:90% your mass spectrometry. 00:11:19.570 --> 00:11:21.010 align:middle line:84% And in between, you want something 00:11:21.010 --> 00:11:24.040 align:middle line:84% where you can differentially label heavy or light atoms. 00:11:24.040 --> 00:11:25.970 align:middle line:84% In the original proof of concept, 00:11:25.970 --> 00:11:30.430 align:middle line:84% this was done with hydrogen atoms versus deuterium. 00:11:30.430 --> 00:11:35.450 align:middle line:84% They differ by one atomic mass unit per position. 00:11:35.450 --> 00:11:39.700 align:middle line:84% However, you will see in the next slide, 00:11:39.700 --> 00:11:41.560 align:middle line:84% this has the unfortunate consequence 00:11:41.560 --> 00:11:43.990 align:middle line:84% that hydrogen and deuterium actually are not only 00:11:43.990 --> 00:11:46.450 align:middle line:84% mass distinguishable, but they also have different chemical 00:11:46.450 --> 00:11:47.590 align:middle line:90% properties. 00:11:47.590 --> 00:11:51.400 align:middle line:84% There's an isotope effect that's detectable in a variety 00:11:51.400 --> 00:11:57.250 align:middle line:84% of chemistries, including retention time on the HPLC. 00:11:57.250 --> 00:12:01.870 align:middle line:84% This has since been upgraded for you having C13 versus C12. 00:12:01.870 --> 00:12:04.810 align:middle line:84% Now, that's a much more subtle difference. 00:12:04.810 --> 00:12:06.370 align:middle line:90% Same mass difference. 00:12:06.370 --> 00:12:09.190 align:middle line:84% You have nine different carbon atoms in here, 00:12:09.190 --> 00:12:11.100 align:middle line:90% and that works better. 00:12:11.100 --> 00:12:13.600 align:middle line:84% In addition, there's a way that you can cleave off the biase 00:12:13.600 --> 00:12:15.758 align:middle line:84% after it's done its job, where you would add it, 00:12:15.758 --> 00:12:17.425 align:middle line:84% and you selected all those peptides that 00:12:17.425 --> 00:12:19.930 align:middle line:84% have been modified and then acid cleaved all 00:12:19.930 --> 00:12:22.810 align:middle line:90% that to clean up the mass spec. 00:12:22.810 --> 00:12:26.860 align:middle line:84% So here's an example where you have the difference in M 00:12:26.860 --> 00:12:31.600 align:middle line:84% over Z, of 4, M over Z and it's between these heavy and light 00:12:31.600 --> 00:12:32.140 align:middle line:90% peaks. 00:12:32.140 --> 00:12:34.690 align:middle line:84% And the ratio of these two is something 00:12:34.690 --> 00:12:38.670 align:middle line:84% you can use for essentially every peptide where 00:12:38.670 --> 00:12:40.660 align:middle line:84% the pair of peaks is in an uncluttered part 00:12:40.660 --> 00:12:41.890 align:middle line:90% of the mass spectrum. 00:12:41.890 --> 00:12:43.270 align:middle line:84% And here's evidence where you can 00:12:43.270 --> 00:12:47.440 align:middle line:84% see that on retention time and the horizontal axis 00:12:47.440 --> 00:12:51.340 align:middle line:84% on the left-hand side of Slide 41. 00:12:51.340 --> 00:12:53.860 align:middle line:84% You can see this little red line shows 00:12:53.860 --> 00:12:57.280 align:middle line:84% how the centroids of the set of peaks should line up 00:12:57.280 --> 00:13:00.670 align:middle line:84% and has been displaced through the chemical effects of that 00:13:00.670 --> 00:13:03.880 align:middle line:90% adding a few neutrons. 00:13:03.880 --> 00:13:09.910 align:middle line:84% OK, so the lesson here is that hydrogen deuterium are not 00:13:09.910 --> 00:13:11.530 align:middle line:84% necessarily chemically identical. 00:13:11.530 --> 00:13:14.290 align:middle line:84% The conceit of isotopes is that they really 00:13:14.290 --> 00:13:18.610 align:middle line:84% should be chemically identical, but really it's 00:13:18.610 --> 00:13:24.320 align:middle line:84% better if you work with heavier atoms to introduce neutrons. 00:13:24.320 --> 00:13:28.190 align:middle line:84% Now, so what can we do with this ratiometric acid? 00:13:28.190 --> 00:13:31.120 align:middle line:84% Now, we're going to-- just like before, we compared 00:13:31.120 --> 00:13:33.850 align:middle line:84% absolute protein levels to absolute RNA levels, 00:13:33.850 --> 00:13:35.680 align:middle line:84% now we're going to measure ratios. 00:13:35.680 --> 00:13:40.370 align:middle line:84% And to do ratios, we have to have two different conditions. 00:13:40.370 --> 00:13:42.040 align:middle line:84% So another advantage to absolute is 00:13:42.040 --> 00:13:44.140 align:middle line:84% you can do it all under one condition. 00:13:44.140 --> 00:13:45.610 align:middle line:84% Here the two conditions that were 00:13:45.610 --> 00:13:48.100 align:middle line:84% chosen for this proof of concept experiment 00:13:48.100 --> 00:13:50.020 align:middle line:90% were glucose and galactose. 00:13:50.020 --> 00:13:52.630 align:middle line:84% That is to say, growing it plus or minus galactose. 00:13:52.630 --> 00:13:59.950 align:middle line:84% Galactose is a nicely understood metabolic and regulatory system 00:13:59.950 --> 00:14:02.020 align:middle line:90% in yeast. 00:14:02.020 --> 00:14:07.090 align:middle line:84% It fits in with what we know about the central carbon 00:14:07.090 --> 00:14:10.930 align:middle line:84% metabolism, and it induces a set of genes in blue here 00:14:10.930 --> 00:14:17.350 align:middle line:84% that are required for galactose catabolism to produce energy. 00:14:17.350 --> 00:14:18.940 align:middle line:84% The most strongly ones are way off 00:14:18.940 --> 00:14:23.050 align:middle line:84% in the upper right-hand corner here, Dow 710, and one, 00:14:23.050 --> 00:14:27.910 align:middle line:84% these are the core catabolic enzymes. 00:14:27.910 --> 00:14:30.190 align:middle line:84% But almost all of the innate blue triangles 00:14:30.190 --> 00:14:33.310 align:middle line:84% have some kind of story like that. 00:14:33.310 --> 00:14:36.790 align:middle line:84% And these all are in the upper right quadrant 00:14:36.790 --> 00:14:42.850 align:middle line:84% have a high log 10 ratio of expression up to three 00:14:42.850 --> 00:14:47.110 align:middle line:84% logs to the third fold induction. 00:14:47.110 --> 00:14:51.790 align:middle line:84% Similarly, at the other end of the spectrum in the lower 00:14:51.790 --> 00:14:55.120 align:middle line:84% left-hand quadrant are respiratory genes 00:14:55.120 --> 00:15:03.490 align:middle line:84% that are involved in, say, oxidative phosphorylation. 00:15:03.490 --> 00:15:06.880 align:middle line:84% And these are moderately depressed 00:15:06.880 --> 00:15:10.840 align:middle line:84% under galactose conditions, and that's 00:15:10.840 --> 00:15:12.950 align:middle line:84% why they're in the lower right-hand corner. 00:15:12.950 --> 00:15:17.860 align:middle line:84% And then the green ones are not quite along this diagonal. 00:15:17.860 --> 00:15:21.340 align:middle line:84% Their messenger RNA is increased, 00:15:21.340 --> 00:15:24.280 align:middle line:84% but their protein expression is not. 00:15:24.280 --> 00:15:28.900 align:middle line:84% And these are the ribosome protein genes, 00:15:28.900 --> 00:15:31.630 align:middle line:84% and this is another phenomenon that's 00:15:31.630 --> 00:15:35.180 align:middle line:90% well documented in the system. 00:15:35.180 --> 00:15:37.490 align:middle line:90% OK. 00:15:37.490 --> 00:15:40.070 align:middle line:84% Now those are examples of how you 00:15:40.070 --> 00:15:42.230 align:middle line:84% can use absolute and relative, and why 00:15:42.230 --> 00:15:44.930 align:middle line:84% you're motivated to use absolute and relative measures 00:15:44.930 --> 00:15:47.330 align:middle line:90% for proteins and messenger RNAs. 00:15:47.330 --> 00:15:52.850 align:middle line:84% Now, these are all treating as if, just like before, 00:15:52.850 --> 00:15:55.370 align:middle line:84% I said that messenger RNA might lump all 00:15:55.370 --> 00:15:58.852 align:middle line:84% together all splice forms and call that the gene product. 00:15:58.852 --> 00:16:01.310 align:middle line:84% Although you know better, and the same thing with proteins, 00:16:01.310 --> 00:16:04.410 align:middle line:84% you might lump together all the protein splice forms. 00:16:04.410 --> 00:16:07.698 align:middle line:84% And not only that but for a particular protein splice form, 00:16:07.698 --> 00:16:09.740 align:middle line:84% there are many different synthetic modifications, 00:16:09.740 --> 00:16:13.393 align:middle line:84% such as proteolysis and phosphorylation. 00:16:13.393 --> 00:16:15.560 align:middle line:84% And so we're going to talk about these modifications 00:16:15.560 --> 00:16:18.800 align:middle line:84% very briefly to hopefully whet your appetite for one 00:16:18.800 --> 00:16:21.410 align:middle line:84% of the most exciting parts of proteomics 00:16:21.410 --> 00:16:25.220 align:middle line:84% and whether it's identification or quantitation. 00:16:25.220 --> 00:16:27.770 align:middle line:84% So we've already mentioned radial isotopic labeling 00:16:27.770 --> 00:16:29.480 align:middle line:90% as a way of quantitating. 00:16:29.480 --> 00:16:35.060 align:middle line:84% Using various radioactive sulfur, 00:16:35.060 --> 00:16:37.610 align:middle line:84% you can use, whether it's stable isotopes 00:16:37.610 --> 00:16:40.820 align:middle line:84% or radioactive isotopes, you can use these to do pulse labeling 00:16:40.820 --> 00:16:44.690 align:middle line:90% to monitor a dynamic process. 00:16:44.690 --> 00:16:47.210 align:middle line:84% P-32 if in particular, if you want 00:16:47.210 --> 00:16:53.730 align:middle line:84% to enrich for some of the most well-studied and significant 00:16:53.730 --> 00:16:55.370 align:middle line:84% photosynthetic modifications involved 00:16:55.370 --> 00:16:57.260 align:middle line:90% in signal transduction. 00:16:57.260 --> 00:16:59.810 align:middle line:84% You can enrich for particular types of amino acids. 00:16:59.810 --> 00:17:02.060 align:middle line:84% We already showed that the cysteines, which 00:17:02.060 --> 00:17:04.730 align:middle line:84% is an arbitrary amino acid chosen 00:17:04.730 --> 00:17:12.230 align:middle line:84% for the ICAT ratiometric method, that was chosen because it has 00:17:12.230 --> 00:17:17.119 align:middle line:84% interesting reactivity, not because it's intrinsically 00:17:17.119 --> 00:17:21.319 align:middle line:84% important low abundance regulatory molecules. 00:17:21.319 --> 00:17:24.960 align:middle line:84% Phosphates, on the other hand, can be very important, 00:17:24.960 --> 00:17:28.880 align:middle line:84% and you might need to enrich because these important 00:17:28.880 --> 00:17:31.700 align:middle line:84% regulatory phosphorylation sites can 00:17:31.700 --> 00:17:36.163 align:middle line:84% get lost in the snow of all the rest peptides in the proteome. 00:17:36.163 --> 00:17:38.330 align:middle line:84% This can either be done-- this enrichment can either 00:17:38.330 --> 00:17:44.450 align:middle line:84% be done by immobilized metals such as iron and gallium, 00:17:44.450 --> 00:17:45.960 align:middle line:90% and so forth. 00:17:45.960 --> 00:17:48.440 align:middle line:84% This is called imac for immobolized metal affinity 00:17:48.440 --> 00:17:52.400 align:middle line:84% chromatography, or you can have antibodies 00:17:52.400 --> 00:17:54.250 align:middle line:84% that are specific particular phosphate 00:17:54.250 --> 00:17:57.680 align:middle line:84% peptide, and particular phosphate amino acids. 00:17:57.680 --> 00:18:01.220 align:middle line:84% You have lectins for carbohydrates as front ends 00:18:01.220 --> 00:18:04.640 align:middle line:90% for mass spectrometry. 00:18:04.640 --> 00:18:09.710 align:middle line:84% Even when we do P-32 labeling metabolically 00:18:09.710 --> 00:18:14.900 align:middle line:84% where the P-32 will only label the subset of the proteins 00:18:14.900 --> 00:18:16.460 align:middle line:90% which are phosphorylated. 00:18:16.460 --> 00:18:19.190 align:middle line:90% 00:18:19.190 --> 00:18:22.850 align:middle line:84% It is still the case that some of the most interesting 00:18:22.850 --> 00:18:25.220 align:middle line:84% regulatory cell cycle proteins are not 00:18:25.220 --> 00:18:27.320 align:middle line:84% detected above background because there 00:18:27.320 --> 00:18:30.410 align:middle line:84% are many abundant proteins, such as ribosome proteins, 00:18:30.410 --> 00:18:35.000 align:middle line:84% central carbon metabolic enzymes, which 00:18:35.000 --> 00:18:37.850 align:middle line:84% are needed in high abundance, but also need 00:18:37.850 --> 00:18:39.620 align:middle line:90% to have phosphorylation. 00:18:39.620 --> 00:18:45.620 align:middle line:84% And so you get this forest of phosphate proteins 00:18:45.620 --> 00:18:47.330 align:middle line:84% such as ribosome and metabolic, which 00:18:47.330 --> 00:18:50.300 align:middle line:84% make it hard to detect the regulatory ones. 00:18:50.300 --> 00:18:56.720 align:middle line:84% So labeling is not a panacea, and we'll come-- 00:18:56.720 --> 00:18:58.970 align:middle line:84% I think you'll see as we go through the protein 00:18:58.970 --> 00:19:00.980 align:middle line:84% modifications and mass spectrometry, 00:19:00.980 --> 00:19:04.970 align:middle line:84% is a multidimensional purification really 00:19:04.970 --> 00:19:07.730 align:middle line:84% is the way that you get away from the ribosomal proteins 00:19:07.730 --> 00:19:10.340 align:middle line:84% and the highly abundant metabolic proteins, which 00:19:10.340 --> 00:19:12.710 align:middle line:84% are interesting, but you need a way of both setting 00:19:12.710 --> 00:19:17.810 align:middle line:84% them and the regulatory low level proteins. 00:19:17.810 --> 00:19:19.930 align:middle line:84% Here are some examples of natural processes. 00:19:19.930 --> 00:19:21.722 align:middle line:84% So you can think of this as a special class 00:19:21.722 --> 00:19:23.600 align:middle line:90% of post-synthetic modification. 00:19:23.600 --> 00:19:26.900 align:middle line:84% Instead of having a phosphate glomming on, 00:19:26.900 --> 00:19:29.300 align:middle line:84% you have two different peptides either 00:19:29.300 --> 00:19:31.160 align:middle line:84% intramolecularly within a protein 00:19:31.160 --> 00:19:33.500 align:middle line:84% or intermolecularly between proteins. 00:19:33.500 --> 00:19:35.990 align:middle line:84% And you should be highly motivated to study these 00:19:35.990 --> 00:19:38.345 align:middle line:84% because they tell you something not only about protein 00:19:38.345 --> 00:19:40.220 align:middle line:84% structure, three-dimensional structure, which 00:19:40.220 --> 00:19:43.410 align:middle line:84% was the topic last time, but protein-protein interactions. 00:19:43.410 --> 00:19:45.470 align:middle line:84% And not just theoretically, what proteins 00:19:45.470 --> 00:19:47.180 align:middle line:84% might interact with other proteins, 00:19:47.180 --> 00:19:53.270 align:middle line:84% or might bind in vitro could be in vitro artifact or in yeast 00:19:53.270 --> 00:19:56.390 align:middle line:84% to hybrid system, could be a yeast to hybrid artifact. 00:19:56.390 --> 00:20:03.890 align:middle line:84% These are actual covalent caught in the act in vivo protein 00:20:03.890 --> 00:20:05.420 align:middle line:90% interactions. 00:20:05.420 --> 00:20:09.080 align:middle line:84% And some of these are-- most of these are very well documented. 00:20:09.080 --> 00:20:13.550 align:middle line:84% By far the most common one and of great significance 00:20:13.550 --> 00:20:18.410 align:middle line:84% to protein tertiary stability in the class of proteins, 00:20:18.410 --> 00:20:19.850 align:middle line:90% which are extracellular. 00:20:19.850 --> 00:20:21.470 align:middle line:84% These include extracellular domains 00:20:21.470 --> 00:20:25.820 align:middle line:84% of membrane proteins and secreted proteins, 00:20:25.820 --> 00:20:33.830 align:middle line:84% and, in particular, because there the oxidation state 00:20:33.830 --> 00:20:39.080 align:middle line:84% is such that this sulfur sulfur bond is stable, 00:20:39.080 --> 00:20:43.440 align:middle line:84% while intracellular tends to be more reducing atmosphere, 00:20:43.440 --> 00:20:47.870 align:middle line:84% and so that these disulfides have trouble forming. 00:20:47.870 --> 00:20:49.700 align:middle line:84% Collagen has a lysine cross-link. 00:20:49.700 --> 00:20:52.970 align:middle line:84% Ubiquitin has a C terminus to lysine cross-link. 00:20:52.970 --> 00:20:55.940 align:middle line:84% Fibrin involved in blood clotting 00:20:55.940 --> 00:20:59.000 align:middle line:84% has glutamine glycine, and so on. 00:20:59.000 --> 00:21:02.270 align:middle line:84% As some of your proteins age, you 00:21:02.270 --> 00:21:05.600 align:middle line:84% will find glucose in high concentrations in your blood 00:21:05.600 --> 00:21:08.390 align:middle line:84% will glycolate the lysine residues. 00:21:08.390 --> 00:21:11.210 align:middle line:84% And this is part of the process by which 00:21:11.210 --> 00:21:16.820 align:middle line:84% these proteins eventually lose their function and are cleared. 00:21:16.820 --> 00:21:19.248 align:middle line:84% Protein nucleic acid interactions 00:21:19.248 --> 00:21:21.290 align:middle line:84% we've been talking about so far are non-covalent. 00:21:21.290 --> 00:21:23.070 align:middle line:84% Some of them are covalent, for example, 00:21:23.070 --> 00:21:27.950 align:middle line:84% when you want to prime de Novo polymer 00:21:27.950 --> 00:21:30.410 align:middle line:90% synthesis, DNA synthesis. 00:21:30.410 --> 00:21:34.550 align:middle line:84% OK, so what are the consequences for the mass spec algorithms 00:21:34.550 --> 00:21:37.160 align:middle line:84% we've been talking about, say, de Novo sequencing 00:21:37.160 --> 00:21:43.310 align:middle line:84% or finding a peptide spectrum in your database? 00:21:43.310 --> 00:21:46.070 align:middle line:84% Well, you can see the masses are going 00:21:46.070 --> 00:21:47.330 align:middle line:90% to be fairly straightforward. 00:21:47.330 --> 00:21:49.790 align:middle line:84% Here's some examples of some masses, of some peptides, 00:21:49.790 --> 00:21:51.440 align:middle line:90% and some cross-link peptides. 00:21:51.440 --> 00:21:57.290 align:middle line:84% On the top right, you'll see one intermolecular cross-linked 00:21:57.290 --> 00:22:00.860 align:middle line:84% between a lysine and a lysine, and an intermolecular 00:22:00.860 --> 00:22:02.130 align:middle line:90% between two peptides. 00:22:02.130 --> 00:22:04.842 align:middle line:84% Now each of these peptides in this display, 00:22:04.842 --> 00:22:06.050 align:middle line:90% these are just simple masses. 00:22:06.050 --> 00:22:07.140 align:middle line:90% These are not fragments. 00:22:07.140 --> 00:22:08.490 align:middle line:90% These are not some fragments. 00:22:08.490 --> 00:22:11.360 align:middle line:84% So you expect these to be triptych products. 00:22:11.360 --> 00:22:14.420 align:middle line:84% So each of their C terminals should be either arginine R 00:22:14.420 --> 00:22:18.180 align:middle line:84% or lysine K. R, K, K, R, so forth, 00:22:18.180 --> 00:22:21.050 align:middle line:84% and so you can see this is two peptides, one ending in R, 00:22:21.050 --> 00:22:24.110 align:middle line:84% one ending in K. So let's look in detail 00:22:24.110 --> 00:22:30.320 align:middle line:84% at this example where you have an intramolecular cross-link. 00:22:30.320 --> 00:22:33.770 align:middle line:84% And you can see that as you cleave it, 00:22:33.770 --> 00:22:36.470 align:middle line:84% each of these peptide bonds entering B ions from the N 00:22:36.470 --> 00:22:39.350 align:middle line:84% terminus and Y ions from the C terminus. 00:22:39.350 --> 00:22:43.280 align:middle line:84% You'll see that there's a special case in the region 00:22:43.280 --> 00:22:45.620 align:middle line:84% that's defined between the two cross-links 00:22:45.620 --> 00:22:49.160 align:middle line:84% and that any peptide bond cleavage that 00:22:49.160 --> 00:22:51.860 align:middle line:84% might occur in the gas phase when colliding 00:22:51.860 --> 00:22:54.650 align:middle line:84% with argon or some other inert gas 00:22:54.650 --> 00:22:58.190 align:middle line:90% will break the chain as usual. 00:22:58.190 --> 00:23:00.260 align:middle line:84% But the chain won't fall apart because it's 00:23:00.260 --> 00:23:01.950 align:middle line:90% got actually two connections. 00:23:01.950 --> 00:23:03.750 align:middle line:84% One is through the normal peptide bond. 00:23:03.750 --> 00:23:05.417 align:middle line:84% And the other is through the cross-link. 00:23:05.417 --> 00:23:07.080 align:middle line:84% So cleavage is all through in here. 00:23:07.080 --> 00:23:09.680 align:middle line:84% It takes two hits to get separation of these, 00:23:09.680 --> 00:23:11.930 align:middle line:90% and two hits is unlikely. 00:23:11.930 --> 00:23:15.410 align:middle line:84% And so, you'll tend to see the B ions in the Y ions 00:23:15.410 --> 00:23:18.410 align:middle line:84% right up until you hit the first cross-linked amino acid, 00:23:18.410 --> 00:23:20.460 align:middle line:90% and then you lose it. 00:23:20.460 --> 00:23:21.890 align:middle line:84% So that's one of the complications 00:23:21.890 --> 00:23:23.810 align:middle line:90% that you have from cross-links. 00:23:23.810 --> 00:23:26.540 align:middle line:84% The other one that when you have, say, cross-linking 00:23:26.540 --> 00:23:28.370 align:middle line:84% two peptides, might occur when you 00:23:28.370 --> 00:23:30.890 align:middle line:84% have an interaction between different proteins, 00:23:30.890 --> 00:23:35.390 align:middle line:84% is you'll now have two sets of B ions and two sets of Y ions. 00:23:35.390 --> 00:23:38.570 align:middle line:84% As If just having B and Y in the same spectrum isn't enough, 00:23:38.570 --> 00:23:43.160 align:middle line:84% now you've got two of each, and even 00:23:43.160 --> 00:23:46.170 align:middle line:84% though you don't have the cycle to worry about, 00:23:46.170 --> 00:23:47.690 align:middle line:90% you have a full set. 00:23:47.690 --> 00:23:52.640 align:middle line:84% But there is a algorithm that Tim Shen and others 00:23:52.640 --> 00:23:57.180 align:middle line:84% have developed for dealing with that in very clean cases. 00:23:57.180 --> 00:24:01.850 align:middle line:84% And here's an example of actually using the cross-links 00:24:01.850 --> 00:24:05.840 align:middle line:84% that you get from mass spectrometry as a fairly 00:24:05.840 --> 00:24:08.360 align:middle line:84% inexpensive set of constraints that you 00:24:08.360 --> 00:24:12.140 align:middle line:84% can use for getting distances either intramolecularly 00:24:12.140 --> 00:24:15.290 align:middle line:84% in this case or conceivably intermolecular. 00:24:15.290 --> 00:24:18.120 align:middle line:90% And the constraints. 00:24:18.120 --> 00:24:21.260 align:middle line:84% You can't get any bigger than the cross-linked distance. 00:24:21.260 --> 00:24:24.500 align:middle line:84% You know that the chemical structure of the cross-linker, 00:24:24.500 --> 00:24:27.200 align:middle line:84% and so you can say these two amino acids with their side 00:24:27.200 --> 00:24:29.270 align:middle line:84% chains and so forth are reacting have 00:24:29.270 --> 00:24:31.860 align:middle line:90% to be this link or shorter. 00:24:31.860 --> 00:24:34.010 align:middle line:84% And that's what these little yellow lines indicate 00:24:34.010 --> 00:24:39.170 align:middle line:84% in this fibroblast growth factor two, FGF two, where the crystal 00:24:39.170 --> 00:24:41.450 align:middle line:90% structure of FGF two is known. 00:24:41.450 --> 00:24:44.390 align:middle line:84% And these constraints will greatly 00:24:44.390 --> 00:24:47.300 align:middle line:84% aid your ability to find distant homologs 00:24:47.300 --> 00:24:53.540 align:middle line:84% or to increase the precision of your homology modeling 00:24:53.540 --> 00:24:54.680 align:middle line:90% in three dimensions. 00:24:54.680 --> 00:24:58.430 align:middle line:90% 00:24:58.430 --> 00:25:01.790 align:middle line:84% The shorter you cross-link, obviously, the better, the 00:25:01.790 --> 00:25:04.490 align:middle line:84% tighter your constraints, but it might reduce the efficiency 00:25:04.490 --> 00:25:05.210 align:middle line:90% of cross-linking. 00:25:05.210 --> 00:25:08.000 align:middle line:84% If you're doing this as artificial cross-linking as 00:25:08.000 --> 00:25:10.188 align:middle line:84% opposed to if you have a natural cross-linking, 00:25:10.188 --> 00:25:12.230 align:middle line:84% you're basically stuck with whatever the natural. 00:25:12.230 --> 00:25:15.680 align:middle line:84% This was an artificial cross-linking 00:25:15.680 --> 00:25:19.130 align:middle line:84% with a chemical cross-linker, or five functional cross-linker. 00:25:19.130 --> 00:25:20.550 align:middle line:90% That's just a reminder. 00:25:20.550 --> 00:25:24.170 align:middle line:84% This is a different way of showing some of the things. 00:25:24.170 --> 00:25:28.070 align:middle line:84% We had a scatterplot last class, which 00:25:28.070 --> 00:25:30.680 align:middle line:84% showed that as you increase the sequence 00:25:30.680 --> 00:25:36.530 align:middle line:84% identity in homology modeling up to 100% on the vertical axis, 00:25:36.530 --> 00:25:42.020 align:middle line:84% you decrease your uncertainty and the observed 00:25:42.020 --> 00:25:44.930 align:middle line:84% root mean square deviation that you get between two structures. 00:25:44.930 --> 00:25:48.210 align:middle line:90% But they're better than 80%. 00:25:48.210 --> 00:25:49.790 align:middle line:84% Then you have in the order of one 00:25:49.790 --> 00:25:52.520 align:middle line:84% Angstrom deviation, which is quite 00:25:52.520 --> 00:25:54.860 align:middle line:90% acceptable for many purposes. 00:25:54.860 --> 00:25:57.080 align:middle line:84% Now, if you want to do threading to very distantly 00:25:57.080 --> 00:26:00.050 align:middle line:84% related structures getting down around 30% 00:26:00.050 --> 00:26:03.932 align:middle line:84% is getting at 25% is getting the Twilight Zone, where you really 00:26:03.932 --> 00:26:04.640 align:middle line:90% can't believe it. 00:26:04.640 --> 00:26:07.640 align:middle line:90% It's off by too many angstroms. 00:26:07.640 --> 00:26:09.710 align:middle line:84% But these constraints in the previous slide 00:26:09.710 --> 00:26:13.040 align:middle line:84% could help you out in either doing the mozzie modeling 00:26:13.040 --> 00:26:14.510 align:middle line:84% or doing a threading where you're 00:26:14.510 --> 00:26:16.400 align:middle line:84% searching through your favorite sequence 00:26:16.400 --> 00:26:20.150 align:middle line:84% through a database of three-dimensional structures 00:26:20.150 --> 00:26:22.760 align:middle line:84% to ask which three-dimensional structure is closest to. 00:26:22.760 --> 00:26:25.250 align:middle line:84% Now that FGF two, we had two slides back 00:26:25.250 --> 00:26:28.490 align:middle line:84% with all these constraints can be run through the threading 00:26:28.490 --> 00:26:32.090 align:middle line:84% algorithm, where you run the sequence through the database 00:26:32.090 --> 00:26:34.730 align:middle line:84% illustrated here to this horizontal, 00:26:34.730 --> 00:26:39.560 align:middle line:84% then to the various rows of different structures here. 00:26:39.560 --> 00:26:43.880 align:middle line:84% The fold family is in the second column from the left. 00:26:43.880 --> 00:26:46.940 align:middle line:84% The sequence identity of our search sequence, 00:26:46.940 --> 00:26:49.610 align:middle line:84% which is FGF two, against all these three-dimensional 00:26:49.610 --> 00:26:52.610 align:middle line:84% structures that present identity here 98.6 00:26:52.610 --> 00:26:56.000 align:middle line:84% is basically, that's the same structure. 00:26:56.000 --> 00:26:59.800 align:middle line:84% That's the trivial example because threading rank 00:26:59.800 --> 00:27:01.880 align:middle line:84% is number one, as it should be since it's exactly 00:27:01.880 --> 00:27:04.280 align:middle line:84% the same structure-- almost exactly the same structure. 00:27:04.280 --> 00:27:05.740 align:middle line:84% The constraint area, of course, is 00:27:05.740 --> 00:27:11.480 align:middle line:84% going to be zero because all three-dimensional structure, 00:27:11.480 --> 00:27:15.020 align:middle line:84% and all the cross-links work to that structure. 00:27:15.020 --> 00:27:17.240 align:middle line:84% But an interesting one now this has 00:27:17.240 --> 00:27:19.190 align:middle line:84% been ranked by the constraint error, 00:27:19.190 --> 00:27:21.800 align:middle line:90% not by the threading rank. 00:27:21.800 --> 00:27:25.280 align:middle line:84% And so you can ask does this improve the hits? 00:27:25.280 --> 00:27:29.870 align:middle line:84% And the next one down is FGF two compared to aisle one data, 00:27:29.870 --> 00:27:32.060 align:middle line:84% and they do have the same fold family. 00:27:32.060 --> 00:27:34.160 align:middle line:84% We know that from three-dimensional structure, 00:27:34.160 --> 00:27:37.940 align:middle line:84% and the percent identity is way below the usual cutoff where 00:27:37.940 --> 00:27:41.240 align:middle line:84% you can't infer from threading or sequencing. 00:27:41.240 --> 00:27:43.260 align:middle line:84% In fact, the threading rank is five. 00:27:43.260 --> 00:27:45.710 align:middle line:90% It's not the second-best thread. 00:27:45.710 --> 00:27:47.660 align:middle line:84% But it's a straight arrow zero, and so 00:27:47.660 --> 00:27:49.820 align:middle line:84% if you combine the good threading 00:27:49.820 --> 00:27:51.500 align:middle line:84% rank in the constraint error, then you 00:27:51.500 --> 00:27:57.110 align:middle line:84% would put this as your best just homolog 12% to 13% sequence 00:27:57.110 --> 00:27:58.110 align:middle line:90% identity. 00:27:58.110 --> 00:28:01.370 align:middle line:84% And of course, it's beating out better threading ranks 00:28:01.370 --> 00:28:04.200 align:middle line:84% because it has fewer constraint errors. 00:28:04.200 --> 00:28:07.250 align:middle line:84% So you can see how powerful these constraints might be, 00:28:07.250 --> 00:28:09.800 align:middle line:84% and it's certainly-- you just need 00:28:09.800 --> 00:28:12.740 align:middle line:84% to evaluate just exactly how cost 00:28:12.740 --> 00:28:15.320 align:middle line:84% effective the mass spectrometry is. 00:28:15.320 --> 00:28:20.390 align:middle line:84% Now last topic today, in the realm of protein modifications 00:28:20.390 --> 00:28:24.170 align:middle line:84% and interactions, are how we quantitate metabolites. 00:28:24.170 --> 00:28:27.770 align:middle line:84% Now you can see that we've got some momentum here 00:28:27.770 --> 00:28:31.100 align:middle line:84% on quantitative proteins in RNAs. 00:28:31.100 --> 00:28:33.530 align:middle line:84% And so what are the issues that are slightly 00:28:33.530 --> 00:28:35.330 align:middle line:90% different from metabolites? 00:28:35.330 --> 00:28:38.420 align:middle line:84% And Slide 52 summarizes some of these. 00:28:38.420 --> 00:28:45.080 align:middle line:84% You have when you break open a cell to isolate messenger RNA 00:28:45.080 --> 00:28:46.190 align:middle line:90% or proteins. 00:28:46.190 --> 00:28:52.070 align:middle line:84% There is the rate at which degradative enzymes act 00:28:52.070 --> 00:28:55.760 align:middle line:90% is on the order of seconds. 00:28:55.760 --> 00:28:57.710 align:middle line:84% That's the rate at which they go, 00:28:57.710 --> 00:29:00.380 align:middle line:84% while many of other metabolic processes 00:29:00.380 --> 00:29:03.290 align:middle line:84% take on the order of milliseconds to microseconds. 00:29:03.290 --> 00:29:05.750 align:middle line:84% Very rapid kinetics, and so as the cell 00:29:05.750 --> 00:29:08.660 align:middle line:84% starts to get a little bit sick on the second range, 00:29:08.660 --> 00:29:15.290 align:middle line:84% all these enzymes are scrambling the metabolites concentrations. 00:29:15.290 --> 00:29:18.230 align:middle line:90% So you have these rapid changes. 00:29:18.230 --> 00:29:21.380 align:middle line:84% The detection methods are historically idiosyncratic. 00:29:21.380 --> 00:29:23.848 align:middle line:84% They might be enzyme linked, where you'll have in order 00:29:23.848 --> 00:29:25.640 align:middle line:84% to detect the metabolite, you have a series 00:29:25.640 --> 00:29:29.390 align:middle line:84% of enzymes that result in some fluorescent or luminescent 00:29:29.390 --> 00:29:30.590 align:middle line:90% assay. 00:29:30.590 --> 00:29:33.680 align:middle line:84% Or they could be gas chromatography, 00:29:33.680 --> 00:29:38.220 align:middle line:84% liquid chromatography, NMR mass spectrometer, and so forth. 00:29:38.220 --> 00:29:40.555 align:middle line:84% The good news is there are usually fewer metabolites 00:29:40.555 --> 00:29:41.930 align:middle line:84% than there are RNAs and proteins. 00:29:41.930 --> 00:29:45.770 align:middle line:84% There could be 30,000, some RNAs and proteins typically only 00:29:45.770 --> 00:29:50.750 align:middle line:84% 1,000 or so metabolites, even in the more exotic, 00:29:50.750 --> 00:29:56.640 align:middle line:84% the metabolically enabled such as E Coli. 00:29:56.640 --> 00:30:00.240 align:middle line:84% Here, from their various databases, 00:30:00.240 --> 00:30:07.020 align:middle line:84% ecosite, width, tag, and so on, which 00:30:07.020 --> 00:30:13.330 align:middle line:84% integrate information about metabolites with the enzymes 00:30:13.330 --> 00:30:14.580 align:middle line:90% to act upon them. 00:30:14.580 --> 00:30:19.320 align:middle line:84% Here, we're just looking at the mass range that we have. 00:30:19.320 --> 00:30:21.570 align:middle line:84% Typical mass range, they're very small 00:30:21.570 --> 00:30:23.940 align:middle line:90% compared to proteins in RNAs. 00:30:23.940 --> 00:30:27.040 align:middle line:84% Most of them being around 200 atomic mass units. 00:30:27.040 --> 00:30:29.700 align:middle line:84% And many of them having absolutely identical mass, that 00:30:29.700 --> 00:30:32.430 align:middle line:84% is to say, they have atom per atom exactly 00:30:32.430 --> 00:30:34.350 align:middle line:84% the same composition, even though it's 00:30:34.350 --> 00:30:36.580 align:middle line:84% arranged in three dimensions very differently. 00:30:36.580 --> 00:30:38.250 align:middle line:84% For example, isoleucine and leucine, 00:30:38.250 --> 00:30:41.220 align:middle line:84% as their names might imply, have exactly the same mass 00:30:41.220 --> 00:30:44.910 align:middle line:84% no matter how many significant digits you put on them. 00:30:44.910 --> 00:30:48.510 align:middle line:84% And this is illustrated by actual data 00:30:48.510 --> 00:30:51.510 align:middle line:90% on isoleucine and leucine. 00:30:51.510 --> 00:30:56.070 align:middle line:84% These supposedly highly purified versions commercially available 00:30:56.070 --> 00:30:59.400 align:middle line:84% of isoleucine and leucine mixed together here and run out 00:30:59.400 --> 00:31:02.070 align:middle line:84% in these two dimensions of mass and the horizontal axis 00:31:02.070 --> 00:31:04.620 align:middle line:84% and retention time and hydrophobic separation. 00:31:04.620 --> 00:31:09.030 align:middle line:84% Now, not a peptides but amino acids, metabolites, and you 00:31:09.030 --> 00:31:13.770 align:middle line:84% can see how that even though they are identical in mass, 00:31:13.770 --> 00:31:18.270 align:middle line:84% as shown on the previous slide around 131, 00:31:18.270 --> 00:31:22.860 align:middle line:84% they are separable by their hydrophobicity. 00:31:22.860 --> 00:31:26.280 align:middle line:84% They have the same atomic composition, 00:31:26.280 --> 00:31:30.490 align:middle line:84% but they are separable just by this hydrophobic separation. 00:31:30.490 --> 00:31:32.790 align:middle line:84% And you can see in the commercial press, 00:31:32.790 --> 00:31:35.670 align:middle line:84% there are a variety of contaminating molecules 00:31:35.670 --> 00:31:39.930 align:middle line:84% that co-migrate in their reverse-phase dimension, 00:31:39.930 --> 00:31:42.330 align:middle line:84% presumably because something like reverse phase 00:31:42.330 --> 00:31:44.565 align:middle line:84% is used for purifying them commercially. 00:31:44.565 --> 00:31:47.360 align:middle line:90% 00:31:47.360 --> 00:31:49.430 align:middle line:84% So there are basically three ways 00:31:49.430 --> 00:31:56.600 align:middle line:84% of distinguishing molecules that have the same mass. 00:31:56.600 --> 00:32:00.410 align:middle line:84% The one in the previous slide was separating them 00:32:00.410 --> 00:32:03.050 align:middle line:84% by another property, like retention time 00:32:03.050 --> 00:32:05.280 align:middle line:90% on hydrophobic properties. 00:32:05.280 --> 00:32:07.670 align:middle line:84% Another one is secondary fragmentation. 00:32:07.670 --> 00:32:09.470 align:middle line:84% Just as we could fragment peptides 00:32:09.470 --> 00:32:12.920 align:middle line:84% by collision in the gas phase with some inert gas, 00:32:12.920 --> 00:32:14.450 align:middle line:90% we can do this with metabolites. 00:32:14.450 --> 00:32:18.020 align:middle line:84% And so two things that have the same mass 00:32:18.020 --> 00:32:20.348 align:middle line:84% may have a different fragmentation pattern. 00:32:20.348 --> 00:32:21.890 align:middle line:84% And you can see again, you can cleave 00:32:21.890 --> 00:32:24.110 align:middle line:84% it every particular position, and here are 00:32:24.110 --> 00:32:27.020 align:middle line:84% two different aspects and two different labs. 00:32:27.020 --> 00:32:30.650 align:middle line:84% Slightly different methodology showing fragmentation 00:32:30.650 --> 00:32:33.560 align:middle line:90% in almost every carbon bond. 00:32:33.560 --> 00:32:41.630 align:middle line:84% The third method by which you can distinguish 00:32:41.630 --> 00:32:44.510 align:middle line:84% compounds that have exactly the same mass, and this case this 00:32:44.510 --> 00:32:47.510 align:middle line:90% is the most extreme case. 00:32:47.510 --> 00:32:53.510 align:middle line:84% These compounds have the same mass. 00:32:53.510 --> 00:32:56.592 align:middle line:84% They actually not only have the same chemical composition, 00:32:56.592 --> 00:32:58.550 align:middle line:84% the same atomic composition, they actually have 00:32:58.550 --> 00:32:59.720 align:middle line:90% the same chemical structure. 00:32:59.720 --> 00:33:01.310 align:middle line:84% Their three dimensions are the same. 00:33:01.310 --> 00:33:02.720 align:middle line:90% Their mass is the same. 00:33:02.720 --> 00:33:04.610 align:middle line:90% What it is is this-- 00:33:04.610 --> 00:33:06.890 align:middle line:84% let's say, the red is a carbon-13, 00:33:06.890 --> 00:33:08.930 align:middle line:90% and the green are the carbon-12. 00:33:08.930 --> 00:33:12.830 align:middle line:84% You can have the carbon-13 in different positions on this, 00:33:12.830 --> 00:33:15.360 align:middle line:90% say, this glucose molecule. 00:33:15.360 --> 00:33:18.350 align:middle line:84% So it has the same four-dimensional structure, 00:33:18.350 --> 00:33:19.340 align:middle line:90% the same mass. 00:33:19.340 --> 00:33:21.620 align:middle line:84% It's just you moved the position of the C-13 00:33:21.620 --> 00:33:22.850 align:middle line:90% to different positions. 00:33:22.850 --> 00:33:24.740 align:middle line:84% This is actually an interesting case 00:33:24.740 --> 00:33:28.670 align:middle line:84% when you have natural abundance glucose, where 00:33:28.670 --> 00:33:33.050 align:middle line:84% you have C-13 trace amounts, it can 00:33:33.050 --> 00:33:35.510 align:middle line:84% be positioned on various different carbon atoms. 00:33:35.510 --> 00:33:39.950 align:middle line:84% But you can still tell where it is, by when it-- 00:33:39.950 --> 00:33:42.440 align:middle line:84% it's now not broken down in the gas phase 00:33:42.440 --> 00:33:45.050 align:middle line:84% like collision-induced dissociation. 00:33:45.050 --> 00:33:47.000 align:middle line:84% But it's broken down in the cell. 00:33:47.000 --> 00:33:48.458 align:middle line:84% It goes through different pathways. 00:33:48.458 --> 00:33:49.958 align:middle line:84% And this is the example-- that we're 00:33:49.958 --> 00:33:52.100 align:middle line:84% going to be talking about pathways non-stop 00:33:52.100 --> 00:33:54.285 align:middle line:90% for the next three sessions. 00:33:54.285 --> 00:33:56.660 align:middle line:84% But here's an example of central carbon metabolism, where 00:33:56.660 --> 00:33:58.640 align:middle line:84% you start with glucose and glucose phosphate 00:33:58.640 --> 00:34:02.540 align:middle line:84% in the upper left-hand corner of this network diagram, 00:34:02.540 --> 00:34:05.570 align:middle line:84% and you end up with carbon dioxide down on the lower left. 00:34:05.570 --> 00:34:07.610 align:middle line:84% And it can go through various pathways 00:34:07.610 --> 00:34:13.969 align:middle line:84% through ribulose or down through three carbons. 00:34:13.969 --> 00:34:16.610 align:middle line:84% And each of these three and two carbon breakdown products 00:34:16.610 --> 00:34:21.320 align:middle line:84% can have the labeled atom, the mass tagged atom 00:34:21.320 --> 00:34:23.550 align:middle line:90% in different positions. 00:34:23.550 --> 00:34:26.090 align:middle line:84% And as this quote from literature 00:34:26.090 --> 00:34:28.340 align:middle line:84% points out that when you want to study 00:34:28.340 --> 00:34:30.949 align:middle line:84% the fluxes through the pathway, by monitoring, 00:34:30.949 --> 00:34:36.889 align:middle line:84% you can actually do a pulse or a stable steady-state 00:34:36.889 --> 00:34:39.199 align:middle line:90% labeling with isotopic labels. 00:34:39.199 --> 00:34:41.449 align:middle line:84% And you can monitor the fluxes through these pathways. 00:34:41.449 --> 00:34:43.429 align:middle line:84% But you need to take into account 00:34:43.429 --> 00:34:46.699 align:middle line:84% all the different ways that you can go through the pathways. 00:34:46.699 --> 00:34:51.080 align:middle line:84% Especially, when you're doing metabolic cycles like the TCA 00:34:51.080 --> 00:34:58.190 align:middle line:84% cycle or you have to think through all the multiple turns. 00:34:58.190 --> 00:35:01.940 align:middle line:84% Now, in principle, that kind of metabolic tracing 00:35:01.940 --> 00:35:05.120 align:middle line:84% can be done either with mass spectrometry 00:35:05.120 --> 00:35:10.160 align:middle line:84% or with nuclear magnetic resonance. 00:35:10.160 --> 00:35:13.220 align:middle line:84% When you do quantitative 2D nuclear magnetic resonance, 00:35:13.220 --> 00:35:16.730 align:middle line:84% you're basically looking at the shifts 00:35:16.730 --> 00:35:21.037 align:middle line:84% in the spectral quantities for the carbon-13s. 00:35:21.037 --> 00:35:23.120 align:middle line:84% Remember, we were talking about carbon-13 labeling 00:35:23.120 --> 00:35:26.555 align:middle line:84% and the normal, most abundant protons. 00:35:26.555 --> 00:35:29.540 align:middle line:90% 00:35:29.540 --> 00:35:31.430 align:middle line:84% The chemical shifts here that you 00:35:31.430 --> 00:35:34.520 align:middle line:84% get are due to the exact chemical 00:35:34.520 --> 00:35:39.170 align:middle line:84% environment of this proton or the carbon-13 for, say, 00:35:39.170 --> 00:35:44.960 align:middle line:84% the alpha for each of these amino acids, alpha betas. 00:35:44.960 --> 00:35:46.970 align:middle line:84% And each of these little clusters 00:35:46.970 --> 00:35:54.350 align:middle line:84% are schematic for the intensity of these particular atoms 00:35:54.350 --> 00:35:59.060 align:middle line:84% that you're monitoring by their isotope effects on the NMR 00:35:59.060 --> 00:35:59.690 align:middle line:90% here. 00:35:59.690 --> 00:36:07.200 align:middle line:84% The odd number of nucleons is critical to the detection. 00:36:07.200 --> 00:36:12.470 align:middle line:84% OK, so if you know the structure of the network, 00:36:12.470 --> 00:36:16.040 align:middle line:84% then you can use that knowledge, and you 00:36:16.040 --> 00:36:20.000 align:middle line:84% know which of these atoms go into which parts, then you 00:36:20.000 --> 00:36:25.100 align:middle line:84% can use this to quantitate the fluxes through any point 00:36:25.100 --> 00:36:26.070 align:middle line:90% in the network. 00:36:26.070 --> 00:36:28.730 align:middle line:84% On the other hand, if you only know part of the network, 00:36:28.730 --> 00:36:30.950 align:middle line:84% then you can use this way as a tracking 00:36:30.950 --> 00:36:37.850 align:middle line:84% to slowly piece together how the network must go. 00:36:37.850 --> 00:36:42.050 align:middle line:84% Most of this is both worked out well before genomics 00:36:42.050 --> 00:36:45.170 align:middle line:84% and our current systems Biology methods, 00:36:45.170 --> 00:36:47.700 align:middle line:84% and so there aren't real algorithms for doing this 00:36:47.700 --> 00:36:48.720 align:middle line:90% as far as I know. 00:36:48.720 --> 00:36:51.210 align:middle line:84% Although, certainly, there's an opportunity for doing it. 00:36:51.210 --> 00:36:53.850 align:middle line:84% Now, this is measuring-- remember this ratios 00:36:53.850 --> 00:36:54.870 align:middle line:90% versus absolute amounts. 00:36:54.870 --> 00:36:57.270 align:middle line:84% This is measuring not only ratios 00:36:57.270 --> 00:36:59.270 align:middle line:90% of metabolite concentrations. 00:36:59.270 --> 00:37:01.680 align:middle line:84% These are not metabolite concentrations, but fluxes. 00:37:01.680 --> 00:37:10.920 align:middle line:84% So with metabolites, you can measure concentrations 00:37:10.920 --> 00:37:13.470 align:middle line:90% or fluxes absolute or ratios. 00:37:13.470 --> 00:37:15.690 align:middle line:90% All four of those combinations. 00:37:15.690 --> 00:37:18.030 align:middle line:84% Now, let's say-- again, in principle, 00:37:18.030 --> 00:37:20.370 align:middle line:84% if you can measure absolute concentrations, 00:37:20.370 --> 00:37:24.502 align:middle line:84% you can measure ratios, and you can measure ratios of fluxes. 00:37:24.502 --> 00:37:26.460 align:middle line:84% So how will you measure absolute concentration? 00:37:26.460 --> 00:37:28.335 align:middle line:84% So remember, we said that one of the problems 00:37:28.335 --> 00:37:30.580 align:middle line:84% was that as soon as you start perturbing 00:37:30.580 --> 00:37:34.870 align:middle line:84% the cell in microseconds, you can get changes. 00:37:34.870 --> 00:37:37.530 align:middle line:84% But one way to do this is without slicing the cells. 00:37:37.530 --> 00:37:40.470 align:middle line:90% You can snap freeze them first. 00:37:40.470 --> 00:37:47.760 align:middle line:84% Yes, snap expose them to aqueous methanol at -40. 00:37:47.760 --> 00:37:49.830 align:middle line:84% You can wash them at that, it's a liquid, 00:37:49.830 --> 00:37:52.260 align:middle line:84% and you can remove the outside metabolites. 00:37:52.260 --> 00:37:55.560 align:middle line:84% And there's reason to believe that this 00:37:55.560 --> 00:37:59.760 align:middle line:84% is the minimally perturbing method of preparing the cells. 00:37:59.760 --> 00:38:03.060 align:middle line:84% And then you put them in basically boiling alcohol, 00:38:03.060 --> 00:38:07.320 align:middle line:84% and then quantitate with NMR methods such as the ones 00:38:07.320 --> 00:38:13.050 align:middle line:84% we just talked about, getting up to 1,300 measures per sample. 00:38:13.050 --> 00:38:15.720 align:middle line:84% Some examples of some of these internal metabolite 00:38:15.720 --> 00:38:18.180 align:middle line:84% concentration, now, remember, these are not flux ratios. 00:38:18.180 --> 00:38:20.100 align:middle line:84% But actual metabolite concentrations 00:38:20.100 --> 00:38:23.310 align:middle line:84% of things like glucose phosphate, ATP, 00:38:23.310 --> 00:38:28.860 align:middle line:84% pyruvate, so on, can be correlated to genomics 00:38:28.860 --> 00:38:32.290 align:middle line:84% by the vehicle of gene knockouts. 00:38:32.290 --> 00:38:35.370 align:middle line:84% We have wild type on the top of the far left, 00:38:35.370 --> 00:38:37.950 align:middle line:90% followed by a deletion of HO. 00:38:37.950 --> 00:38:40.140 align:middle line:84% This is a homing in the nucleus, should 00:38:40.140 --> 00:38:42.780 align:middle line:84% have no metabolic consequences at all. 00:38:42.780 --> 00:38:43.920 align:middle line:90% This is used as a control. 00:38:43.920 --> 00:38:45.810 align:middle line:84% It's a pseudo wild type, just shows 00:38:45.810 --> 00:38:48.750 align:middle line:84% that the deletion method itself is not changing 00:38:48.750 --> 00:38:50.808 align:middle line:90% things, the metabolism. 00:38:50.808 --> 00:38:53.100 align:middle line:84% And then, as you go further and further down this list, 00:38:53.100 --> 00:38:57.540 align:middle line:84% you get more and more severe expected effects 00:38:57.540 --> 00:39:00.390 align:middle line:84% on the ability of the organism, say to produce energy, 00:39:00.390 --> 00:39:04.350 align:middle line:84% as exemplified by its ATP production. 00:39:04.350 --> 00:39:06.780 align:middle line:84% Now, if you have low ATP production, 00:39:06.780 --> 00:39:08.850 align:middle line:84% that means you have high residual levels 00:39:08.850 --> 00:39:12.277 align:middle line:84% of ATP, which is the other end of the energy spectrum. 00:39:12.277 --> 00:39:13.860 align:middle line:84% And so if you look down these columns, 00:39:13.860 --> 00:39:17.250 align:middle line:84% it's a little hard to see with all the clutter that's produced 00:39:17.250 --> 00:39:18.413 align:middle line:90% by the standard deviations. 00:39:18.413 --> 00:39:20.955 align:middle line:84% I wouldn't want them to get rid of those standard deviations. 00:39:20.955 --> 00:39:25.350 align:middle line:84% It's wonderful, but anyway, you can summarize this 00:39:25.350 --> 00:39:27.760 align:middle line:84% by looking at the ATP to ADP ratio. 00:39:27.760 --> 00:39:30.430 align:middle line:84% That's a way you don't need ratios for this to work. 00:39:30.430 --> 00:39:36.030 align:middle line:84% But it's a way of accentuating this energy balance, 00:39:36.030 --> 00:39:38.970 align:middle line:84% and so you can see, for a wild type, you have about almost 7 00:39:38.970 --> 00:39:41.490 align:middle line:90% is the ratio of ATP to ADP. 00:39:41.490 --> 00:39:44.430 align:middle line:84% Highly charged on this high energy state. 00:39:44.430 --> 00:39:46.710 align:middle line:84% And as you get to these pet mutants, which 00:39:46.710 --> 00:39:50.520 align:middle line:84% are mutations in the mitochondrial process, 00:39:50.520 --> 00:39:54.300 align:middle line:84% you find the cell is becoming increasingly ineffective. 00:39:54.300 --> 00:39:59.970 align:middle line:84% Now, these were all chosen because these were 00:39:59.970 --> 00:40:01.580 align:middle line:90% so-called silent mutations. 00:40:01.580 --> 00:40:04.800 align:middle line:84% The title of this paper says that you can get phenotype 00:40:04.800 --> 00:40:07.860 align:middle line:84% by looking at the molecular analysis quantitating 00:40:07.860 --> 00:40:11.820 align:middle line:84% the whole holistically, systematically 00:40:11.820 --> 00:40:13.560 align:middle line:84% all the metabolites for things that 00:40:13.560 --> 00:40:18.060 align:middle line:90% otherwise have no phenotypes. 00:40:18.060 --> 00:40:20.400 align:middle line:90% OK. 00:40:20.400 --> 00:40:24.150 align:middle line:84% Now, this representa-- as you start 00:40:24.150 --> 00:40:27.420 align:middle line:84% quantitating whole proteomics, proteome interactions, 00:40:27.420 --> 00:40:31.320 align:middle line:84% proteome modifications, metabolites 00:40:31.320 --> 00:40:33.150 align:middle line:84% in their interactions, you start wanting 00:40:33.150 --> 00:40:35.310 align:middle line:90% to relay this information. 00:40:35.310 --> 00:40:37.860 align:middle line:84% Summarize this information in the context of models. 00:40:37.860 --> 00:40:42.150 align:middle line:84% And just as last time, the upper left-hand portion 00:40:42.150 --> 00:40:49.142 align:middle line:84% of this summary, I emphasize that no model is exact. 00:40:49.142 --> 00:40:50.850 align:middle line:84% Sometimes the people working in the field 00:40:50.850 --> 00:40:54.870 align:middle line:84% convince themselves that it's more exact than lower models. 00:40:54.870 --> 00:40:58.470 align:middle line:84% But every one of these, even the quantum mechanics, 00:40:58.470 --> 00:41:01.080 align:middle line:84% is a poor approximation of electrodynamics, 00:41:01.080 --> 00:41:05.100 align:middle line:84% and molecular mechanics has only spherical atoms represented. 00:41:05.100 --> 00:41:08.220 align:middle line:84% Some of the most challenging cell models 00:41:08.220 --> 00:41:10.320 align:middle line:84% involve massive equations of stochastic. 00:41:10.320 --> 00:41:12.560 align:middle line:84% Now, we're not talking about single atoms anymore. 00:41:12.560 --> 00:41:14.102 align:middle line:84% We're talking about single molecules. 00:41:14.102 --> 00:41:18.120 align:middle line:84% Still it's too coarse for many experiments. 00:41:18.120 --> 00:41:21.060 align:middle line:84% You can have phenomenological rates, 00:41:21.060 --> 00:41:23.190 align:middle line:84% such as the ones we've been talking about, 00:41:23.190 --> 00:41:25.470 align:middle line:84% represented in ordinary differentiable equations. 00:41:25.470 --> 00:41:29.070 align:middle line:84% How do you get the parameters that describe now 00:41:29.070 --> 00:41:30.360 align:middle line:90% concentration and time? 00:41:30.360 --> 00:41:34.980 align:middle line:84% Not single molecules, but treating them 00:41:34.980 --> 00:41:38.280 align:middle line:84% as a bag of a particular part of the cell, 00:41:38.280 --> 00:41:40.530 align:middle line:84% or the whole cell having a particular concentration 00:41:40.530 --> 00:41:42.090 align:middle line:90% of molecules. 00:41:42.090 --> 00:41:44.223 align:middle line:84% And then we'll go on as we get into the network 00:41:44.223 --> 00:41:46.640 align:middle line:84% analysis, and the next few lectures, we'll talk about some 00:41:46.640 --> 00:41:50.240 align:middle line:84% of these other models, which have their roles. 00:41:50.240 --> 00:41:51.770 align:middle line:84% But let's just get at-- how would we 00:41:51.770 --> 00:41:57.450 align:middle line:84% get some of the parameters that describe concentration in time? 00:41:57.450 --> 00:41:59.420 align:middle line:84% And when we talk about the formalism 00:41:59.420 --> 00:42:02.600 align:middle line:84% of these networks, regulatory networks, 00:42:02.600 --> 00:42:05.420 align:middle line:90% mainly are about binding. 00:42:05.420 --> 00:42:08.332 align:middle line:84% But they intimately connect with catalytic networks, 00:42:08.332 --> 00:42:10.040 align:middle line:84% where you not only bind, but you actually 00:42:10.040 --> 00:42:12.590 align:middle line:84% change the covalent structure of molecules. 00:42:12.590 --> 00:42:17.540 align:middle line:84% In the simplest such case, single substrate 00:42:17.540 --> 00:42:19.730 align:middle line:84% going in single product coming out 00:42:19.730 --> 00:42:23.870 align:middle line:84% is that the top of this Slide number 60. 00:42:23.870 --> 00:42:27.020 align:middle line:84% And here, you can see that the enzyme E, which 00:42:27.020 --> 00:42:30.650 align:middle line:84% is typically a protein and/or but there are RNA 00:42:30.650 --> 00:42:32.750 align:middle line:90% catalysts and so on. 00:42:32.750 --> 00:42:35.870 align:middle line:84% But the property that's emphasized here 00:42:35.870 --> 00:42:38.060 align:middle line:84% is that the enzyme is not consumed. 00:42:38.060 --> 00:42:43.630 align:middle line:84% As A goes in, it makes a covalent change from EA to EB, 00:42:43.630 --> 00:42:44.870 align:middle line:90% B is released. 00:42:44.870 --> 00:42:45.980 align:middle line:90% E is recycled. 00:42:45.980 --> 00:42:49.280 align:middle line:84% E is not consumed by the cycle, but A is consumed. 00:42:49.280 --> 00:42:50.000 align:middle line:90% B is produced. 00:42:50.000 --> 00:42:52.970 align:middle line:90% 00:42:52.970 --> 00:42:55.940 align:middle line:84% But let's look at it in a different light. 00:42:55.940 --> 00:42:58.730 align:middle line:84% In a particularly interesting class of reactions, 00:42:58.730 --> 00:43:06.440 align:middle line:84% for example, those involving the regulatory cascade 00:43:06.440 --> 00:43:10.100 align:middle line:84% signal transduction enzyme modifications. 00:43:10.100 --> 00:43:12.560 align:middle line:84% Here the enzyme now becomes substrate. 00:43:12.560 --> 00:43:17.360 align:middle line:84% The ATP is now no longer consumed in the whole cycle. 00:43:17.360 --> 00:43:22.580 align:middle line:84% ATP comes in, modifies the enzyme to a phosphate enzyme. 00:43:22.580 --> 00:43:28.550 align:middle line:84% And the accompanying reaction regenerates the-- 00:43:28.550 --> 00:43:30.450 align:middle line:90% turns the ATP back into ATP. 00:43:30.450 --> 00:43:32.850 align:middle line:84% So the ATP, in a certain sense, is catalytic here, 00:43:32.850 --> 00:43:37.550 align:middle line:84% and now the enzyme is consumed, producing a fossil enzyme. 00:43:37.550 --> 00:43:39.770 align:middle line:84% So it all depends on when you started 00:43:39.770 --> 00:43:41.690 align:middle line:90% constructing these graphs. 00:43:41.690 --> 00:43:44.060 align:middle line:84% What is the node-- within the node is a substrate, 00:43:44.060 --> 00:43:48.230 align:middle line:84% and node is an enzyme, depends on how you look at it? 00:43:48.230 --> 00:43:51.050 align:middle line:84% And I do this somewhat provocatively, 00:43:51.050 --> 00:43:54.620 align:middle line:84% so you'll think about these networks 00:43:54.620 --> 00:43:56.930 align:middle line:84% not just as binding, but as catalysis. 00:43:56.930 --> 00:44:00.050 align:middle line:84% And not just the enzyme being the catalyst, but sometimes 00:44:00.050 --> 00:44:01.440 align:middle line:90% the substrate as well. 00:44:01.440 --> 00:44:05.000 align:middle line:84% Now, this is the simplest case of measuring kinetics. 00:44:05.000 --> 00:44:06.440 align:middle line:90% This is not equilibrium. 00:44:06.440 --> 00:44:08.510 align:middle line:90% This is kinetics. 00:44:08.510 --> 00:44:12.140 align:middle line:84% And we're studying this plot on the far left-hand side 00:44:12.140 --> 00:44:13.520 align:middle line:90% of Slide 61. 00:44:13.520 --> 00:44:16.820 align:middle line:84% As substrate increases, the rate of production of product 00:44:16.820 --> 00:44:17.960 align:middle line:90% increases. 00:44:17.960 --> 00:44:22.280 align:middle line:84% You could start out with zero product or small amounts, 00:44:22.280 --> 00:44:26.390 align:middle line:84% and it used to be historically at bottom, 00:44:26.390 --> 00:44:28.460 align:middle line:84% you would require to do the experiment, 00:44:28.460 --> 00:44:30.920 align:middle line:84% you would require that the product be 00:44:30.920 --> 00:44:32.210 align:middle line:90% as close to zero as possible. 00:44:32.210 --> 00:44:36.260 align:middle line:84% You'd do initial rates, and you have this simple relationship 00:44:36.260 --> 00:44:38.780 align:middle line:84% where the increase in product was 00:44:38.780 --> 00:44:43.220 align:middle line:84% a simple function of the 1 over 1 00:44:43.220 --> 00:44:46.770 align:middle line:84% plus reciprocal of the substrate concentration. 00:44:46.770 --> 00:44:50.090 align:middle line:84% And as the substrate would increase, 00:44:50.090 --> 00:44:52.310 align:middle line:84% you'd eventually saturate the amount of enzyme 00:44:52.310 --> 00:44:54.200 align:middle line:84% that you have present in the experiment, 00:44:54.200 --> 00:44:57.620 align:middle line:84% and that would be the maximum velocity for that amount 00:44:57.620 --> 00:45:01.400 align:middle line:90% of enzyme would be max. 00:45:01.400 --> 00:45:04.220 align:middle line:84% However, if you have a more full equation where 00:45:04.220 --> 00:45:06.620 align:middle line:84% you take into account all the players, at least 00:45:06.620 --> 00:45:11.540 align:middle line:84% in the simplest system, the forward velocity, the PDT 00:45:11.540 --> 00:45:15.920 align:middle line:84% the derivative of product concentration 00:45:15.920 --> 00:45:18.680 align:middle line:84% as a subject of time, this might be in mol/Ls per liter. 00:45:18.680 --> 00:45:22.490 align:middle line:84% It's going to go up as a substrate goes up. 00:45:22.490 --> 00:45:24.512 align:middle line:84% More substrate means it will go faster 00:45:24.512 --> 00:45:26.220 align:middle line:84% in the forward direction towards product. 00:45:26.220 --> 00:45:27.720 align:middle line:84% But if you have some product, you'll 00:45:27.720 --> 00:45:29.360 align:middle line:90% have some product inhibition. 00:45:29.360 --> 00:45:32.360 align:middle line:84% You'll have a tendency the kinetics to go 00:45:32.360 --> 00:45:33.540 align:middle line:90% in the opposite direction. 00:45:33.540 --> 00:45:35.810 align:middle line:84% So this is negative-- there's a negative component 00:45:35.810 --> 00:45:38.780 align:middle line:90% with products. 00:45:38.780 --> 00:45:42.080 align:middle line:84% The KS and KP are sometimes called Nikolaus constants 00:45:42.080 --> 00:45:46.490 align:middle line:84% as the substrate gets closer to the Nikolaus constant. 00:45:46.490 --> 00:45:49.550 align:middle line:84% This is basically related to the binding 00:45:49.550 --> 00:45:51.110 align:middle line:90% affinity for the substrate. 00:45:51.110 --> 00:45:55.910 align:middle line:84% And that's a natural halfway point where 00:45:55.910 --> 00:46:00.710 align:middle line:84% the substrate is half saturating is roughly 00:46:00.710 --> 00:46:03.350 align:middle line:90% what the KS is all about. 00:46:03.350 --> 00:46:05.790 align:middle line:84% Now let's compare this very simple case. 00:46:05.790 --> 00:46:08.013 align:middle line:84% We have one substrate producing one product 00:46:08.013 --> 00:46:10.430 align:middle line:84% to a more typical case where you might have two substrates 00:46:10.430 --> 00:46:12.200 align:middle line:90% producing two products. 00:46:12.200 --> 00:46:14.180 align:middle line:84% And let's take this out of a real network. 00:46:14.180 --> 00:46:16.472 align:middle line:84% We're going to show this real network in just a moment, 00:46:16.472 --> 00:46:18.050 align:middle line:90% and it's all slurry. 00:46:18.050 --> 00:46:21.860 align:middle line:84% Here you have two substrates ATP, and F16 going 00:46:21.860 --> 00:46:24.170 align:middle line:90% to 2 products ATP, and FTP. 00:46:24.170 --> 00:46:26.990 align:middle line:84% And you've got this same form where you have 00:46:26.990 --> 00:46:31.350 align:middle line:90% a velocity of this reaction. 00:46:31.350 --> 00:46:35.240 align:middle line:84% And it's a function of the reactants F16 and ATP, 00:46:35.240 --> 00:46:39.020 align:middle line:84% and you find them in the numerator here. 00:46:39.020 --> 00:46:43.045 align:middle line:84% And you find these Nikolaus constants in the right places. 00:46:43.045 --> 00:46:45.670 align:middle line:84% But in addition, you find in the denominator this curious term, 00:46:45.670 --> 00:46:47.020 align:middle line:90% it has these fourth powers. 00:46:47.020 --> 00:46:50.200 align:middle line:84% Well, we didn't see any fourth powers in the previous slide. 00:46:50.200 --> 00:46:52.300 align:middle line:84% Why are we getting fourth powers just because, 00:46:52.300 --> 00:46:55.150 align:middle line:90% and why is AMP in here? 00:46:55.150 --> 00:46:57.850 align:middle line:84% It's not even one of the reactions or the products. 00:46:57.850 --> 00:46:58.940 align:middle line:90% What's going on? 00:46:58.940 --> 00:47:01.450 align:middle line:84% Well, this is actually a regulatory phenomenon 00:47:01.450 --> 00:47:03.760 align:middle line:84% allosteric where you have a second site on the enzyme. 00:47:03.760 --> 00:47:06.520 align:middle line:84% One site does the catalytic magic, 00:47:06.520 --> 00:47:09.430 align:middle line:84% and the other one is regulated by some 00:47:09.430 --> 00:47:12.430 align:middle line:84% in the infinite wisdom of the whole network, that 00:47:12.430 --> 00:47:13.750 align:middle line:90% is important feedback. 00:47:13.750 --> 00:47:18.280 align:middle line:84% So AMP is related to ATP as a further step, 00:47:18.280 --> 00:47:22.390 align:middle line:84% and it feeds back on this enzyme, as does F16, and ADP, 00:47:22.390 --> 00:47:23.030 align:middle line:90% and so forth. 00:47:23.030 --> 00:47:24.610 align:middle line:84% So all these-- and this fourth power 00:47:24.610 --> 00:47:26.380 align:middle line:84% just says you want it to be cooperative. 00:47:26.380 --> 00:47:29.590 align:middle line:84% You want to have a nonlinear regulation. 00:47:29.590 --> 00:47:31.190 align:middle line:84% That's what's going on in this term. 00:47:31.190 --> 00:47:32.560 align:middle line:84% And it doesn't occur in the whole network, 00:47:32.560 --> 00:47:33.768 align:middle line:90% we'll show in the next slide. 00:47:33.768 --> 00:47:35.350 align:middle line:84% But it does occur at some key points, 00:47:35.350 --> 00:47:38.590 align:middle line:84% like this one where the enzyme has two sites, catalytic 00:47:38.590 --> 00:47:39.833 align:middle line:90% and regulatory. 00:47:39.833 --> 00:47:42.250 align:middle line:84% And when you see these terms that are greater than linear, 00:47:42.250 --> 00:47:44.230 align:middle line:84% like the fourth power, that's sometimes 00:47:44.230 --> 00:47:46.330 align:middle line:84% referred to the Hill coefficient and refers 00:47:46.330 --> 00:47:48.070 align:middle line:90% to the steepness of-- 00:47:48.070 --> 00:47:49.690 align:middle line:84% instead of having this kind of curve, 00:47:49.690 --> 00:47:51.190 align:middle line:84% you have this kind of sigmoid curve. 00:47:51.190 --> 00:47:54.580 align:middle line:84% And the sigs peakedness to that is related to that power. 00:47:54.580 --> 00:47:57.100 align:middle line:84% Now, let's look-- if you compare-- 00:47:57.100 --> 00:48:01.480 align:middle line:84% so this little piece, this phosphofructokinase step, 00:48:01.480 --> 00:48:06.130 align:middle line:84% is going to be put in context of the entire network 00:48:06.130 --> 00:48:08.080 align:middle line:84% in the red blood cell, human red blood cell 00:48:08.080 --> 00:48:12.670 align:middle line:84% right here in the upper left quadrant of the circle. 00:48:12.670 --> 00:48:16.540 align:middle line:84% And this is the simplest analysis-- 00:48:16.540 --> 00:48:18.242 align:middle line:84% this is the simplest metabolic network 00:48:18.242 --> 00:48:19.450 align:middle line:90% that you'll be talking about. 00:48:19.450 --> 00:48:23.890 align:middle line:84% This mostly involves covalent transitions or pumping 00:48:23.890 --> 00:48:25.480 align:middle line:90% across membranes. 00:48:25.480 --> 00:48:29.080 align:middle line:84% It treats the whole cell as a uniform bag, 00:48:29.080 --> 00:48:31.150 align:middle line:84% with a membrane being a separate compartment. 00:48:31.150 --> 00:48:33.010 align:middle line:84% And there are really two objectives of this. 00:48:33.010 --> 00:48:36.230 align:middle line:90% One is to produce ATP. 00:48:36.230 --> 00:48:39.670 align:middle line:84% So that you can run the pumps to keep the osmotic pressure 00:48:39.670 --> 00:48:41.260 align:middle line:84% constant across the red blood cell 00:48:41.260 --> 00:48:44.200 align:middle line:84% membrane, so that it maintains its shape. 00:48:44.200 --> 00:48:47.660 align:middle line:84% And the other is to maintain the redox at the right level, 00:48:47.660 --> 00:48:48.760 align:middle line:90% so as if-- 00:48:48.760 --> 00:48:50.770 align:middle line:84% so you have a reducing atmosphere. 00:48:50.770 --> 00:48:54.370 align:middle line:90% 00:48:54.370 --> 00:48:59.320 align:middle line:84% The hemoglobin has an intimate contact with oxygen, 00:48:59.320 --> 00:49:03.820 align:middle line:84% and so there is a certain low level of oxidation of the Iron 00:49:03.820 --> 00:49:06.430 align:middle line:84% rather than just binding to the Iron to make 00:49:06.430 --> 00:49:08.980 align:middle line:84% hemoglobin which is not a good physiological state. 00:49:08.980 --> 00:49:10.990 align:middle line:84% So you want to have this reducing potential 00:49:10.990 --> 00:49:13.820 align:middle line:84% to get it back to the correct oxidation state, 00:49:13.820 --> 00:49:16.370 align:middle line:90% so it will bind oxygen. 00:49:16.370 --> 00:49:19.570 align:middle line:84% So you have a little bit of purine metabolism here. 00:49:19.570 --> 00:49:23.770 align:middle line:84% Quite a bit of simple glycolysis, 00:49:23.770 --> 00:49:26.950 align:middle line:90% not oxidative phosphorylation. 00:49:26.950 --> 00:49:29.770 align:middle line:84% Just because glycolysis produced reducing potential and little 00:49:29.770 --> 00:49:35.020 align:middle line:84% ATP, and these 40 or so enzymatic reactions 00:49:35.020 --> 00:49:37.360 align:middle line:84% can be modeled with about 200 parameters. 00:49:37.360 --> 00:49:40.090 align:middle line:84% All of these 200 parameters have been measured accurately 00:49:40.090 --> 00:49:43.820 align:middle line:84% by purifying each of the metabolites and enzymes. 00:49:43.820 --> 00:49:46.870 align:middle line:84% And this has been reduced to an ordinary differentiable 00:49:46.870 --> 00:49:50.721 align:middle line:84% equation model that has been evolving since the '70s. 00:49:50.721 --> 00:49:54.160 align:middle line:90% 00:49:54.160 --> 00:50:01.150 align:middle line:84% If we look at Slide 64 that this phosphofructokinase, 00:50:01.150 --> 00:50:04.240 align:middle line:84% we have the same form that we had a couple of slides back. 00:50:04.240 --> 00:50:07.120 align:middle line:84% Here is the fourth power term, and the denominator 00:50:07.120 --> 00:50:11.080 align:middle line:84% having AMP as a regulatory molecule. 00:50:11.080 --> 00:50:13.720 align:middle line:84% And then you have the Nikolous constants, 00:50:13.720 --> 00:50:17.050 align:middle line:84% and so forth, explicitly stated here 00:50:17.050 --> 00:50:23.800 align:middle line:84% in the numerator for the substrates F16 and ATP. 00:50:23.800 --> 00:50:28.330 align:middle line:84% And you have a similar equation for every single enzyme 00:50:28.330 --> 00:50:31.990 align:middle line:84% step in that whole network in the red blood cell model. 00:50:31.990 --> 00:50:35.950 align:middle line:84% And in green are the concentrations, 00:50:35.950 --> 00:50:38.410 align:middle line:84% at any given time point, that's going to be their dynamic 00:50:38.410 --> 00:50:39.430 align:middle line:90% or steady state. 00:50:39.430 --> 00:50:41.410 align:middle line:84% At any given time point, you'll have 00:50:41.410 --> 00:50:45.460 align:middle line:84% green for the metabolic concentrations and red 00:50:45.460 --> 00:50:48.880 align:middle line:84% for the fluxes indicated by the enzyme. 00:50:48.880 --> 00:50:51.340 align:middle line:84% And you can run this as a simulation for the red blood 00:50:51.340 --> 00:50:56.140 align:middle line:84% cell and see all the interesting questions 00:50:56.140 --> 00:51:02.410 align:middle line:84% about robustness and optimality, and so forth. 00:51:02.410 --> 00:51:04.180 align:middle line:84% Even though this is a very mature model, 00:51:04.180 --> 00:51:06.670 align:middle line:84% there still is lots to be done, even 00:51:06.670 --> 00:51:08.302 align:middle line:90% in this very simplest system. 00:51:08.302 --> 00:51:10.510 align:middle line:84% Now, we're going to go on to more complicated systems 00:51:10.510 --> 00:51:12.205 align:middle line:84% in the next three lectures on networks. 00:51:12.205 --> 00:51:14.740 align:middle line:90% 00:51:14.740 --> 00:51:16.630 align:middle line:84% But for today, we've basically tried 00:51:16.630 --> 00:51:19.960 align:middle line:84% to integrate the protein, either absolute or ratios, 00:51:19.960 --> 00:51:24.130 align:middle line:84% with RNA measurements and with metabolite interactions, 00:51:24.130 --> 00:51:26.560 align:middle line:84% proteins post-synthetic modifications. 00:51:26.560 --> 00:51:29.500 align:middle line:84% So until next time, thank you very much. 00:51:29.500 --> 00:51:30.000 align:middle line:90%