Fourth Annual DNA Grantees' Workshop
Monday, June 23, 2003
MORNING SESSION
Simple, Rapid, and Accurate Quantitation of Human DNA
Janice A. Nicklas
Biography
MS. PAGLIARO: Thank you very much for your patience. Our next presentation is something I am sure all the practitioners are going to be very interested in. Coming from the Northeast, I have had a little information about this from time to time and it's very exciting. It's a simple, rapid, and accurate quantitation of DNA, and the presenter is Dr. Janice Nicklas, who has been in Vermont since November 2001.
Prior to accepting a position in Vermont, Dr. Nicklas completed postdoctoral fellowships at Tufts Medical School, University of Minnesota, and the University of Vermont, where she also was a research assistant professor and a research associate professor. She's a fellow of the Harvard genetics training program (1996–97 and 2000–2001) and is board certified in clinical molecular genetics from the American Board of Molecular Genetics.
DR. NICKLAS: I want to thank NIJ for giving me the opportunity to give this talk. Nicklas: Slide 1
You have a handout on your table of this talk. Your handout's a little bit different than my talk because Eric Buel made me add even more things to it. So we're going to try to move quickly because there is a lot here. Nicklas: Slide 2
I won't read the abstract. I'll just dive right into the talk: There is a need for DNA quantitation. First, the STR (short tandem repeat) analyses that we have performed are very stringent on the amount of DNA you put in. If you put in too little, you're not going to get any results. But if you put in too much, then you have problems with N–1 peaks and stutter, and you get a messy profile. Therefore, it's important to actually quantitate the DNA before you put it into your STR analysis. The second reason is the requirement of having human-specific quantitation on casework samples. Nicklas: Slide 3
Most people right now use a slot blot assay, which you can either do colorimetrically or chemiluminescently. Basically, you put your DNA on a filter and use a probe that's actually to the centromeric region of chromosome 17, and you get some spots on there that you compare to your standard curve. Nicklas: Slide 4
The good thing is this is a primate-specific probe and you can detect reasonably low levels of DNA. My favorite thing about it is it gives a pretty blue color. Actually, the blue color is the only thing I like about Quantiblot.
What's bad about it is there's a pretty narrow range of DNA that you can actually quantitate. You're eyeballing it manually, so you're saying, well, does your sample color blue match the 5 nanogram or the 2.5 nanogram or does it match my sweater. There's really no estimate of how good the DNA is that you have in there, just that you have some.
It's also time consuming and labor intensive. It takes you a couple of hours to do one of these and you pretty much have to sit there and babysit it, because you have to do washes and put this on and take that off.
We all want something faster, cheaper, and easier. So there are new technologies out there that we can use. One thing, you can make the slot blot better by analyzing it with a densitometer or a digital camera (instead of with your eye) and actually getting some quantitative numbers out of it. Nicklas: Slide 5
There's the "read it" technology kit, AluQuant, which I won't go into at all. Nobody uses southern blots or slot blots or anything like that these days. Instead, everybody does PCR (polymerase chain reaction). So there's got to be a way to do this quickly and easily with PCR.
You can do an end-point assay. You can PCR it, stop it, run it out on a gel, and quantitate from there; or use some other method to quantitate it. Or you can use what everybody uses now and that is real time: You watch the PCR as it actually occurs. You monitor it on an instrument. This is a much more quantitative way to do that, and we can see that in a minute.
ABI (Applied Biosystems) supposedly is coming out with some sort of kit, but I don't know anything about it.
Our goals were to develop a great assay (as I said—cheap, fast, and quantitative), to validate it with actual forensic samples to show that it works, and to determine whether an automation system could be developed to more easily perform the quantitation. Nicklas: Slide 6
Okay, so what did we do? We've already heard a little bit about the cat assay (see presentation by Menotti-Raymond), where the SINE (small interspersed nuclear element) was used as the repetitive element. We do something similar. The reason to do this is to make assays as sensitive as possible. You can pick a single-copy gene, but it's not very sensitive. If you pick something that has a million copies (or something like that), then it's very sensitive.
First, we picked ALU sequences because they are primate specific. The first time through, we planned on doing simple PCR of something that probably everybody has in their lab, using a plate reader and a PCR instrument, and doing an assay very quickly. The second time, we used real-time PCR approaches. Nicklas: Slide 7
So we designed some primers for the ALU sequence. It makes a 124-base pair product, and this is just a gel to show you that even just doing endpoint PCR on a gel is reasonably quantitative and that this size product is as expected. Nicklas: Slide 8
To do it in the fluorescent plate reader, you would just make a PCR cocktail. You buy a mix from Sigma and put in primers. QSY7 is just a dye that's at the end of the primer. You have problems with background in this assay because the fluorescence of the primers gives you background. If you put in the QSY7, it quenches the fluorescence of the primers. Nicklas: Slide 9
We used SYBR Green as the dye that detects the DNA. So as PCR progresses, the SYBR Green interpolates into the double-stranded DNA of your PCR product, and you get more and more fluorescence as time goes on.
You put in your template DNA from your sample. We put in a little extra SYBR Green and bovine serum albumin (BSA). BSA ameliorates any inhibitors that might be in the reaction and helps the reaction proceed. I'll show you some data on that in a minute.
So we make up our PCR cocktail. We put it in a 96-well plate with the standard curve and our unknown samples, and we read it in the plate reader to give us a zero time. That is our background. Then we perform PCR for 14 cycles. We put it back in the plate reader and read the fluorescence again. The software from the reader subtracts the background from the final result, makes the standard curve from our standards, and calculates the concentrations of all samples. We published this in the March 2003 issue of Journal of Forensic Science.
We quality assured/quality controlled (QA/QC) this assay for cycle number, anneal time, temperature, and extension time; and the reaction was robust. You get slightly different numbers if you vary these things, but because everything is normalized to your standard curve, it really doesn't make any difference. We don't have quite as large a zoo as the previous speaker, but we checked it against 20 different animals and bacteria. We showed that it's reproducible. We did replicates, multiple days, and different lots of mastermix; and it's very reproducible. Nicklas: Slide 10
Once you get your quantitation number, the most important questions are is it good for your STR analysis, and if you use that number to make your dilutions for your STR analysis, do you get good data out of it? The answers are yes, you do. It correctly quantitates for doing STRs. We tried many different kinds of samples and once again we showed that addition of BSA ameliorates the effects of inhibitors.
Here's just one QA/QC. We looked at the cycle number, and as you can see, if you do 16 cycles, up here it starts to plateau at higher DNA concentrations. But if you use too few cycles, it's very difficult to get any quantitation out of the lower concentration values. Nicklas: Slide 11
So we picked 14 cycles, which admittedly is not a straight line, but there is a curve, and you can plot where your samples are based on the standard curve. These are just the data. Here's the standard curve from one experiment, here are some unknowns, and taking the RFUs (relative fluorescent units), we can drop down to see the sample's concentration. Nicklas: Slide 12
These are just comparisons of samples done with both the slot blot assay and the QSY assay. In general, they're in pretty good agreement. Here are the data that I said I'd show you on the BSA. This is a denim cloth sample, so it has inhibitors. If we do it without BSA, it says that there is very little DNA in there. In fact, if we dilute for STR analysis based on that, our numbers are too high, and these are all points with N–1. If we add BSA, then the quantitation is much closer to what we get with the slot blot and we dilute properly for our COfiler analysis. Nicklas: Slide 13
This just shows the reproducibility of the triplicates. The standard deviations are pretty low. They start to get a little bit higher when you get down into the 0.01-nanogram range, but that's pretty good reproducibility. Nicklas: Slide 14
This is a mixture of rat DNA with human DNA. There are always 2.5 nanograms, but the mixture varies from 100-percent human down to 0 percent. The straight line indicates that the rat DNA is not contributing to the PCR and it's not inhibiting the PCR either. Nicklas: Slide 15
So what does it cost? What's the bottom line? Well, you have to have a microplate reader, which is about $12,000. Hopefully you already have one in your lab. You need a thermocycler, which, if you're doing STR analysis, hopefully you already have one of those, too. And you probably need a centrifuge to be able to spin the plates down. Nicklas: Slide 16
So Quantiblot in our hands costs about 60 cents per well. If you buy the Sigma kit and the mix to do this assay, it's a little bit more expensive. If you make a homemade mix, it's about the same price. But you've got to remember that you could also save time and effort and everything like that when you're doing this assay.
So what about real-time PCR? In real-time PCR, you monitor the amplification when it's happening with a special instrument. It's more quantitative as you use initial exponential results, not the plateau values, because you can add more or less DNA. You tend to end up at the same plateau eventually, so you have to be careful about cycle number. Nicklas: Slide 17
In real-time PCR there are a number of different methods. You can use the SYBR Green dye, just like we did with the fluorescent assay. But there are a number of other things. You can use TaqMan (from ABI); Molecular Beacon; MGB-Eclipse, which is sort of like Beacon; and LUX (light upon extension). I'll explain several of these as we get to them.
The setup is very similar to the QSY assay, except that you don't need to use QSY-modified primers; you can just use regular primers, so that saves costs. You have one instrument that performs the amplification and the fluorescence readings instead of going into your plate reader and putting it into the thermocycler and then back in the plate reader, so that saves time. Nicklas: Slide 18
We use a Corbett Rotorgene 3000, but ABI, Strategene, and Roche also make instruments. There's a much increased linear range (16 nanograms to 1 picogram), and in fact, it's even much larger for some of the other assays.
It's fast. You set it up just like you would a PCR (approximately 15 minutes). It takes 72 minutes to amplify and approximately 5 minutes to analyze and print for your case file. Once again, we optimized and validated this for casework, doing all the things (anneal time and temperature, the animal DNAs, reproducibility, and so on).
This is supposed to be in the September 2003 Journal of Forensic Science.
Here are some sample results, which I color coded. The rainbow is for the standards; the red is the most concentrated, the purple is the least concentrated, and the black ones are unknown samples. So what you're looking for is to set a threshold—kind of like a line in the sand—and when your curve crosses that threshold, you call that the CT and that's the number that you report. The concentration is proportional to 2–CT. So you just have to do your standard curve. Based on these CT values, when your different concentrations cross these points, you see when your sample crosses that different point, and you plot that all out. Actually, the software plots it all out, and you get your number right away. Nicklas: Slide 19
So this is the standard curve plotted out by the software. This is the log of the concentration and your fluorescence here, your unknowns, and your red ones. The software just puts them on the curve, drops them down, and gives you the concentration. The R values are very good. The efficiency is actually greater than 100 percent. Nicklas: Slide 20
This gray line here is the no-template control (NTC). That's when I don't put any DNA in there. And you're saying, well, why are you getting amplification if you don't have any DNA in there? Well, unfortunately, humans are everywhere and their DNA is everywhere. It's in the water. It's in the primers. I breathe in there. Human DNA is ubiquitous.
Now, if I cut this assay off at 20 cycles, you wouldn't see this but you would still get nice values. But if I take it out far enough, the ubiquitous human DNA comes up. And you say, well, why would you want to do that? So that the no-template control has a CT of about 20 (27–29 cycles) due to the ambient DNA.
We like it in there because it's an internal control. If I have a sample and it doesn't come up at the NTC or before, I know I have an inhibitor in there. I know something is wrong (e.g., the tube leaked, I forgot to put in the enzyme, or something like that). Even if there's nothing in your sample, it ought to come up where the NTC is. If it doesn't, there's something wrong. So you don't need to put in an exogenous control or something in your PCR because it's already there. Nicklas: Slide 21
These are the results of some selected samples. The comparisons with the slot blot are generally pretty good. It's different for some things like blood on denim, where some inhibitors probably exist. These are the results we got with two of the different STRs to show that they're within our range, which is 5,500 down to 150. Nicklas: Slide 22
You can see the reproducibility over time is very good. The standard deviation ranges from 10 to 15 percent, which is very good. Certainly the STR is good within the value of about twofold, so a 15-percent variation is well within its values. Nicklas: Slide 23
This is the reproducibility of CTs over five experiments. You can see how tight these values are over different days. Nicklas: Slide 24
We switched instruments, from the old 2000 to a 3000. As you can see, the values once again are very tight, even with a different instrument. Nicklas: Slide 25
This is pretty easy to do. You have to get your primer in and dilute it once to have a stock solution. You make your mastermixes: One that has the SYBR Green in the Sigma ReadyMix, and another mixture of primer with the same dilution. You can freeze these for at least 4 months and just keep using them. All you do is mix your two mastermixes together, put in your DNA, and put it in the PCR machine. Nicklas: Slide 26
What does it cost? Well, obviously you have to buy the real-time instrument. The Corbett that we used costs about $35,000. Strategene has a new personal cycler that costs about $25,000, and ABI sells one for about $100,000. Nicklas: Slide 27
Compared with Quantiblot, which in our hands, as I said, costs about 60 cents, AluQuant costs about $1, and our assay costs about 50 cents. So not only do you save time and effort and get something more quantitative, but it's also cheap.
We have farmed this out to a couple of other groups to give it a try and they're just in the beginning stages: Bruce McCord's lab (Ohio), Charlie Barna's lab (Michigan), and Mark Timken's lab (California). Nicklas: Slide 28
These are some data from Bruce McCord's lab. Once again, you see it looks good. They've got nice ranges and a few samples. Their standard curve had an R value of 0.994. Nicklas: Slide 29
Charlie Barna's Michigan lab has a standard curve but no samples. Once again, they have a nice rainbow, and their R value was 0.996. Nicklas: Slide 30
So what else can we do with this assay? The PCR product that we get for this is 124 base pairs, so the assay is very good at predicting when you have DNA that's at least 124 base pairs in length. But what if you have a degraded DNA sample where the pieces may be smaller? It can't detect any DNA that's smaller than 124 base pairs. There's no way you can amplify that, so it misses those pieces.
Also, if you have a degraded DNA sample and you're trying to amplify some of the larger STRs, such as 350 base pairs or 400 base pairs, it's not very good at predicting good amplification of those. It's very good at predicting the STRs that are 124 base pairs, but it's not so good at predicting the bigger ones. Nicklas: Slide 31
Hence, there has been much interest lately in degraded DNA samples. Can you make mini-STRs? Bruce McCord and John Butler's groups have been trying to make the STR products smaller. Bruce is talking about this tomorrow morning, so you can all hear about that then.
With the 83-base-pair assay that we've been developing, you just make the PCR product smaller and put the primers closer together. We've tested this with animal DNAs and have compared it with the full-length assay, and we've tested the reproducibility. The assay looks just as good as the full-length assay.
So what about using some slightly different type of technology? One is to use an MGB.(minor groove binder). It's Eclipse. It's like a Molecular Beacon. I'll show you what it is in a second. You do your PCR, but you detect the PCR product using a fluorescently labeled probe. It has a dye in the 3' end and a quencher and an MGB that helps it bind strongly to the DNA. One of the advantages is that it has a much greater range than SYBR Green. Nicklas: Slide 32
So this is how it works. You have the beacon, and when it's in solution sitting by itself, the quencher and fluor are in close proximity and you don't get any fluorescence. If they bind to the PCR product, then you get fluorescence. And when they get kicked off when the Taq comes through again, you lose fluorescence. Nicklas: Slide 33
This is what it looks like, resulting in a much greater range (256 nanograms down to 1 picogram). Nicklas: Slide 34
These are animal DNAs in here, and they're close to the no-template control, between there and 1 picogram. So there's much less sensitivity with animal DNA. Nicklas: Slide 35
It's reproducible. The standard deviation is a bit larger, up to 30 percent, but it's still well within the tolerances of the STR assay. Nicklas: Slide 36
We've also tried LUX. This is from Invitrogen, and it's somewhat similar to MGB. You have a fluor, but it's quenched by this hairpin. So when the primer is in solution, it's quenched, but when it's used in the PCR product, you lose that hairpin and you get fluorescence. Nicklas: Slide 37
Here are the raw data on this. It barely climbs out of the slime by the time you get to the end of the assay, which wasn't too thrilling. The other assays come up: Eclipse comes up to about 40, and SYBR comes up to about 80. Nicklas: Slide 38
Here are the data. The rainbow looks good, but I think that the data are kind of shaky; we weren't pleased at all with this. Nicklas: Slide 39
The reproducibility wasn't very good either. You can see where the percent standard deviations were way up, so we kind of dropped that one. Nicklas: Slide 40
So where are we going with this? Well, we're not only interested in determining how much DNA we have, but we also want to know its gender. This is important in cases in which the victim and suspect are different genders, because it can point to which blood spots should be pursued, and in cases involving vaginal swabs, it can decipher the amount of male DNA. So, we wanted to develop a PCR assay for gender and hopefully multiplex it with the ALU assay that we already have so that you can get gender and quantitation all in one step. Nicklas: Slide 41
We've done it with the LUX. The male DNA is blue and the female DNA is pink; the darker the line equates to more DNA, and lighter equals less DNA. You can see that these are the same amounts of DNA here and here (indicating); it's just the gender that is different. So the assay is about 20,000-fold more sensitive to males than females. Nicklas: Slide 42
But, as I said, we didn't like the LUX. So we moved on to the Eclipse Beacon. We actually can go from 256 nanograms down to 1 picogram, which isn't on here. It's just 64 down. These are duplicates, and you can see it's nice and tight. Nicklas: Slide 43
Here are the raw data. Because you don't see any pink on here, here's what the pink is doing. It's staying straight down here. It's not doing anything at all. So this looks pretty good so far.
I just wanted to thank the National Institute of Justice for their continuing support and we're hopeful that it will continue. And I certainly want to thank Eric, who's an invaluable sounding board and collaborator. And I want to thank Marcie LaFountain and Peg Schwartz in the lab for their help and suggestions with this. Nicklas: Slide 44
Thank you.
MS. PAGLIARO: Are there any questions?
QUESTION: So there are a number of Y or ALU family elements in there. Have you been able to quantify how many actual targets in the genome your specific primers are hitting?
DR. NICKLAS: Based on the family that we're amplifying, we think probably about 2,000 (or somewhere in that ballpark). We did quite a bit to try to optimize the temperature on this. If you use a lower annealing temperature, you can have a very broad melting peak. So we're obviously amplifying more related ALU sequences. We tried to use a higher temperature and we sharpened that a lot. So it's hard to know exactly how many you're amplifying because it's a very diverse family. Nicklas: Slide 45
MS. PAGLIARO: Any other questions?
MS. PAGLIARO: No other questions?
(No response.)
DR. NICKLAS: I do want to say one thing: If anybody is interested in trying this out, we'll send you some mix, primers, and protocols. I'd love to get e-mails or calls from people to try this in their own laboratory.
MS. PAGLIARO: Thank you. I'd like to thank Dr. Menotti and Dr. Nicklas for their presentations.
We'll have a 10-minute break at the moment. Anything else? Lisa?
DR. FORMAN: No.
MS. PAGLIARO: Ten minutes.