Fourth Annual DNA Grantees' Workshop
Monday, June 23, 2003
AFTERNOON SESSION
Locus Specific Brackets and Multiplex PCR for Y-Chromosome Short Tandem Repeats
Debang Liu
Biography
(Note: The PowerPoint presentation that supplements Dr. Liu's transcript is not available.)
MS. TOMSEY: Our next speaker for this afternoon is Researcher Debang Liu. Dr. Liu is a cofounder and managing member of Oligotrail. His research focuses on genotyping, DNA diagnostics and genetic therapeutics. Currently Dr. Liu is the principal investigator on two projects concerned with locus-specific bracketing markers and STR genotyping, which are funded by the National Human Genome Research Institute and NIJ.
Before joining Oligotrail, Dr. Liu was a senior research scientist at Evanston Northwestern Health Care in Evanston, Illinois. Earlier he was a research fellow and a Ph.D. student in molecular biology at the University of Illinois in Chicago.
Dr. Liu came from China as an exchange scholar in 1987, and we're very fortunate that he decided to stay. Dr. Liu received an M.D. from Jinzhou Medical College in China, a master's degree in medical science from Hebei Medical University of China, and an MBA degree from Keller Graduate School of DeVry University in Chicago.
I'd like to welcome Dr. Debang Liu.
DR. LIU: I have talked about locus-specific brackets (LSBs) previously about three or four times. So I'd like to show just one cartoon picture to explain what is LSB.
Here, the middle one is the target allele that we are trying to measure. The green portion is the flanking sequence, so every one's supposed to be the same. We are doing STR (short tandem repeat) typing because different individuals may have different numbers of repeat units.
The LSB was created from the two alleles. We make a new allele that has the fewest number of repeats so far not found in human beings. Then we make another new allele which has the most number of repeats, also not found in human beings. Because these two pseudo-alleles have the identical flanking sequences and the same structure of the LSB repeat sequence, they keep similar compatible electrophoretic mobility.
So whenever this target allele goes through the electrophoretic field, the shorter one and the longer one bracketing the target allele go together. Then, not only do you know where to find your alleles but you also keep the same mobility. Whatever the resistance this target allele goes through, those two will go through the same resistance but still keep the compatible mobility.
This is the first grant that we received from NIH to use the LSB to develop a multiplex for CODIS (Combined DNA Index System). Here is a multiplex that can amplify all the 13 CODIS loci, but without the internal size markers, it's hard to tell who's who and what's what.
In this picture, we have an LSB—this is the short allele, and this is the long allele. Because we know that the target allele lies between the shortest and the longest, an LSB can work well on the electrophoresis.
We applied for the grant from NIJ to develop a new Y multiplex based on LSB technology. Here are the Y–STR loci that we selected.
The research has four major tasks:
- Construct the LSB for all these STR markers.
- Develop a multiplex kit so we can amplify all these markers together. The extended European STR loci have been used extensively.
- Develop a software based on LSB to analyze the result, namely to do the allele calling, because the ABI (Applied Biosystems) system uses a different color-labeled internal size standard. With LSB, we can use the same dye-labeled primer to amplify the LSB, but the ABI software doesn't do the work when the standard and the samples share the same dye. Therefore, we have to develop software.
- Perform internal and external validation with a certain number of DNA samples, including forensic DNA samples.
Here we are able to develop the Y multiplex. That's all the extended European loci, so it can be amplified in a single PCR (polymerase chain reaction). Actually, we are doing this simultaneously with the LSB construction.
Here, the PCR product with the LSB, so each locus we have developed a shorter one and a longer one to bracket.
Here are the data. The middle column is the common alleles that have been found or reported. For example, for DYS389I, the reported common alleles are 9 to 17. So we make an LSB (LSB7) with two repeats fewer than the shortest (9), just in case some day someone reports, oh, I found number 8. Number 7, then, would still be useful. For the same reason, we make another one with two repeats longer (i.e., 19). All LSBs are designed that way.
Here is one of the NIST standard reference material samples. This is sample A. Dr. John Butler was kind enough to tell you he has six samples (five male samples and one female sample) and what they are. But when he sent me the samples, he wasn't as nice to me. He didn't tell me what the samples were, he didn't even tell me there was a female sample involved.
We kept amplifying but couldn't get anything. Our lab technician told me: Debang, I've almost used up the entire sample. What can we do? We guessed that maybe a female was involved, but when I called John, he said, I won't tell you anything until you get the final results.
So anyway, we were able to amplify every locus. Here are the highlights: 389I, 389II, 388, 391, 392, 393, YCAII, 19, and 385 A and B. Later on I sent the multiplex kits to Dr. Butler and he amplified them again in his laboratory and these are his results. So every one is amplified.
Here are YCAII, 19, and 385 A and B. However, the balance is not as good as what I did. Maybe I selected the best picture to show and he selected the first one and gave it to me, but that's not the case. I only have a 480 PCR machine, so all my standardization is based on that machine, and he used a 2400 machine to amplify.
DR. BUTLER: 9700.
DR. LIU: 9700. See, I had no idea what to expect.
Luckily he was able to get the amplification, even though the balancing is off.
Here is the female sample (sample F) that we talked about earlier. Those peaks are LSB. So when you load it later when running the gel, there are no amplicons between each LSB, except here, YCAII. See, there are double groups here, and the shorter one is the older LSB long for YCA, and later I extended three repeats longer.
Here are the results. I was so happy when I reported the results to John and they matched what he had.
A total of 58 alleles were identified, and because some alleles that you measure are longer (i.e., positive value) and others measure shorter (i.e., negative value), you may get an average deviation of zero, so good a result so that's not true.
I used the absolute deviation, regardless of positive or negative, as long as it deviated from the sequence length, and then I did the average (0.07 nucleotide). The standard deviation was 0.08 nucleotide. The maximum deviation from sequence length was 0.18 nucleotide. The minimum deviation was –0.15 nucleotide.
Then I identified 350 alleles from 16 DNA samples from John Hardman's Orange County DNA lab and another 9 samples from Dr. Rohrer of Germany. The average deviation was 0.06, and the standard deviation was plus/minus 0.08.
There was one with a large deviation, 0.49 (almost to 0.5). If it's over 0.5, you may call it a plus 1 allele. I haven't redone this sample (4T) from Orange County, so I haven't examined why this one gave such a large deviation, and the minimum was 0.19.
Some validation results have been done in our laboratory. Here is a 100-nanogram male DNA in total 10-microliter PCR. The portions on the left are the primer dimers, and they look okay, except that there is an additional peak. In my experience, this is caused by overloading, because the peak high is 8,000. When you have too much DNA, sometimes you generate shorter peaks. These results are from the 310.
Then I did 0.065 nanograms and got some product, but it was unreliable. Nonspecifics also showed up. This (YCAII) is the primer, but it's not sufficient.
Here is the 0.125 nanogram with the peak high RFU (relative fluorescent unit) in the 1,000s. So every locus shows a nice product here (YCAII, 19, and 385), and it's still balanced.
We were supposed to do the external validation in John Hardman's Orange County lab. John's lab normally uses a 12- to 15-microliter DNA total extraction solution. Unfortunately, the PCR volume I desired was 10 microliters. So there was no way for him to use that to come up with a 10-microliter reaction. It's not just to save some reagents, because I did hear some labs use a lesser amount than the 25-microliter standard. My thought is in a given volume: If you use the same amount of DNA, the smaller the volume, the higher the relative DNA concentration. So when you have tiny DNA samples, by reducing the reaction volume you can increase your relative DNA concentration.
So later I did a volume reaction. I kept every concentration of the reagents the same; I did a 5-microliter, 10, 15, 20, and 30. The results were very similar. This one showed a higher peak; that's because of the loading error, which you can tell from how much primer dimer (inaudible). This one is 0.590 in peak area, this one only about half of that. So the volume doesn't change much.
We have done other tests. The general characteristics of our Y-multiplex, 10-microliter PCR reaction, 36 cycles, and the range can be varied from 34 to 38 cycles, which means within this range we can still get a balanced product. The optimum DNA concentration per reaction is from 0.5 to 2.0 nanograms. Magnesium concentration is 2.0, 1.5 to 1.75 units of DNA polymerase, and the primer concentration varies a lot per reaction, from 0.3 picomoles to the highest 1.75 picomoles. Annealing temperature is 59 ºC, plus/minus 1 to 2 degrees.
Here I would like to introduce an interesting topic. It's the late afternoon. I saw people getting sleepy. Also, they are less interactive. I'd like the researchers and the practitioners to react together. Here is the YCAII. It's a good marker. As John pointed out, it has been used in Europe extensively and has a lot of accumulated data and, more importantly, is highly polymorphic. However, because of its dinucleotides, the stutter peak is very high. So most of you are practitioners. What do you say? Do you want to keep this YCAII or do you want to drop it?
So here I have a test. You don't have to answer, but I do like volunteers. So anyone can find some patterns from this YCA PCR problem. There are obviously two groups. Because it's a cell line DNA, the highest one is 19, and the last one is 23. If I don't label this, can anyone in this room guess what is what?
(No response.)
So no volunteer yet.
I have to sell LSB technology, that's my job. So this is the same PCR amplicon with the LSB short and the long. That's why NIJ funded this here. So if we know this one is the 26, we can count back and find that it's 23. So we'll get some idea. Is this good enough?
Let's see all five female samples John provided to me. It is true, they are highly polymorphic. So one thing we can tell is there are two different groups. Samples A, B, and E are double populations. Samples C and D only have one population. The first group always has the highest peak, and the second group (the last peak) is always shorter. So keep this in mind. I will ask you to answer these questions later.
Well, anyway, thank you NIJ for the grant. We have some work left on external validation. We hope that the work can be completed and that the results will be published.
Thank you.
MS. TOMSEY: Thank you, Debang. I encourage all of you to come up and look at his software after the break.