Glendon Parker, a biochemist with Lawrence Livermore National Laboratory’s Forensic Science Center, examines a 250-year-old archaeological hair sample that has been analyzed for human identification using protein markers from the hair. (Photo: Julie Russell/LLNL)
DNA, the fundamental building block of all life, is also the foundation of modern forensic science. From blood to saliva to semen to microscopic touch DNA, criminal perpetrators leave traces behind at crime scenes that have become the strongest evidence presented to juries over the last 30 years.
However, DNA has limitations. One such limitation is that the richest sources can be the most fragile, and can be surprisingly hard to find enough of to really make a breakthrough. For instance, about half of all rape kits don’t yield enough genetic information to determine a profile of the perpetrator.
But imagine you had a pubic hair from the offender in that kit. That single hair could provide proteins that are hardier than DNA—as well as a genetic sequence to help to identify the rapist.
You won’t have to imagine much longer because one of the most futuristic new projects in DNA investigation is about how the fundamental blueprint of DNA is expressed—through proteins. Since the proteins that make up bone and teeth, hair, and even microscopic skin cells are hardier, they are the subjects of a series of ongoing projects at Lawrence Livermore National Laboratory and Intelligence Advanced Research Projects Agency (IARPA). Proteos Project, as it’s called, began this past summer.
The research scientists envision a long-term project by which they have snippets of the proteome that can be compared against a suspect, or even quantified in a database to determine random match probabilities, just like DNA currently. Within an estimated five years, skin cells left by a murderer’s hands, or shed hairs, may be good enough to start pointing toward identities.
Although the proteomics work won’t replace DNA, it will add a whole new robust dimension to crime scene investigation in the future, the researchers told Forensic Magazine.
“If you have a forensic sample that has good quality DNA, that’s your first go-to,” said Deon Anex, a chemist at Lawrence Livermore, and one of the main scientists involved in the ongoing work. “What we’re trying to do is look for evidence types that may not have DNA of sufficient quality to get full STR profiles—maybe a partial profile, or no profile at all.
“We’re asking the question, how can you augment that with other components of the biological material that’s there?,” the chemist added. “And we focused on proteins.”
The protein-as-identified concept was first publicly unveiled in 2016, when the Lawrence Livermore team published its proof-of-principle ideas in the journal PLOS ONE. Hair as old as 250 years displayed nearly 200 signature proteins that could represent a molecular fingerprint of sorts, which could then be narrowed down to one in a million.
The team of scientists used skull hair shafts from six people buried in London and Kent, their deaths dating from 1750 to 1853.
The hairs were ground up, reduced and alkylated, and further treated with trypsin. The remainder was observed and measured with liquid chromatography and mass spectrometry.
From the single-amino acid polymorphisms (SAPs), the researchers reconstructed the single nucleotide polymorphisms (SNPs) from the subject DNA genomes.
The accuracy is roughly equivalent to determining mitochondrial DNA from hair shafts. But if they were used together, hair shafts could be a breakthrough in genetic identification, Bradley Hart of Lawrence Livermore’s Forensic Science Center told Forensic Magazine.
Some news accounts at the time went so far as to erroneously claim that the proteomics work would be “better than DNA.”
But that’s not accurate, Anex reiterated in an interview. He explained that the team has conducted a litany of further investigations into the vagaries of protein hair analysis, down to single shafts. The ongoing work—much of it still unpublished—involves looking at the different protein expression in hairs from different body locations, from the pubic to the scalp, as well as the differences in hair that is long and has had more exposure to the elements.
The genetically variant peptides, or GVPS, of the proteins reflect the SNPs encoded in DNA, since the latter has the instructions to make the proteins themselves. But the identifying power is not as strong with proteins.
“Some of the headlines said ‘proteins are replacing DNA,’ but that’s not where we’re going with this,” said Anex. “Hair is pretty indestructible. It can be around for a very long time and the same is true with tooth and bone.”
The concept in bone was tested, and published earlier this year in the journal Forensic Science International.
The team took rib bones from 10 recently dead donors, all of European-American ancestry. The team found a series of 35 different protein markers, using liquid chromatography mass spectrometry (LCMS) to separate and quantify the multivariant chains of amino acids.
The markers provided enough information to determine RMPS of 1-in-6, to a best of 1-in-42,000 people.
Those are not DNA-caliber numbers, which are often in the realm of millions or billions. The difficulty with bone is it has a smaller set of GVPs than, say, hair.
“There is inherently less genetic variability in proteins present in bone compared to human hair,” said Hart. “For purposes of identifying or linking them to evidence, hair proteins are more valuable.”
LLNL scientists Katelyn Mason and Deon Anex prepare to pulverize forensic bone samples prior to demineralization and extraction of proteins to find identity markers. (Photo: Julie Russell/LLNL)
One particular application of the bone proteomics work would be during a mass casualty situation, like a catastrophic fire, when scientists need to sort out complex, perhaps mixed sets of remains. Such work would be key in identifying and placing pieces in their rightful spot in such previously unthinkably forensic conundrums as the remaining evidence from the 9/11 terrorist attack on the World Trade Center in Manhattan 17 years ago, Anex said.
An intriguing upside to tooth analysis could have singular forensic promise. The amelogenin, a protein in the enamel, shows X- and Y-isoforms, meaning sex could be determined just from chemical analysis of the substance.
Nevertheless, the bones and teeth work has so far been set aside for the upside of the hair—and also the massive new project involving the keratinized skin cells, Anex said.
The whole proteomics forensics concept kicked off when Lawrence Livermore scientists began working with the Bureau of Alcohol, Tobacco and Firearms (ATF) several years ago. One ATF expert had a problem: he could very rarely pull touch DNA from brass surfaces. He wanted to know if proteins, particularly skin cells would present a more robust source of trace evidence. Since then, the two agencies have conducted experiments by adding hair and skin clues to bombs that were set off in containment chambers to see whether mitochondrial DNA and proteins survive the heat from the explosions. Though the research is not yet published, the teams found that both the hair and skin evidence survived at least partially intact.
That skin-cell concept started the Lawrence Livermore explorations into microscopic proteomics of hair and bone. But now it’s come full circle back to skin cells with the Proteos Project, run by IARPA.
The three-year project was unveiled last summer at an IARPA “Proposers’ Day” briefing. Applications were due December 2017, and the agency officially announced the project three months ago. Six agencies are involved. Three are on the performance side, developing methodologies to conduct the tests: the University of Washington, GE Global Research and Signature Science, LLC. Three other agencies are part of the test and evaluation team: Lawrence Livermore, as well as the National Institute for Standards and Technology, and the Johns Hopkins Applied Physics Laboratory. (The Texas-based Signature Science announced in August the IARPA contract was for $2.3 million.)
The work is split into three phases: one to determine common GVPs; the second to isolate the rare GVPs; and the third and final phase to encompass a “test spiral” of six months, determining testing methods in real-world scenarios, and for potential end-users.
The three participants on the performance side will be working in parallel, in a kind of competition to come up with the best methodology to collect both DNA and proteins at the same time, said Kristen Jordan, IARPA’s manager of the Proteos Project.
The first phase began in July, and is expected to wrap up next summer. The remaining phases will extend to the end of 2020.
The work will use tools including mass spectrometry that are “well within the range of current instrumentation,” according to Anex. One potential objective would be determining a panel of 100 common GVPs that would be present in between 20 and 70 percent of the population. If that panel is determined, and proven reproducible, reliable and accurate, it would have the potential discrimination power of 1 in 10 to the 10th power—or roughly 1 person in all the nearly 10 billion people on Earth, Anex said.
Brad Hart (left to right), director of Lawrence Livermore National Laboratory’s Forensic Science Center, biochemist Glendon Parker and chemist Deon Anex analyze hair samples using protein markers from the hair.
The three phases will get progressively more challenging, with the second phases involving “dirtier” environments like fired brass shell casings, and mixtures, according to Jordan. By the third phase, they’re going to work with real-world environments, like exploded cars, and confusing forensic conundrums.
“That’s what we’re going to do in phase three—really understand how far we can go with this, operationally—and have confidence in the results,” said Jordan.
A key part of the project is linking proteins back to DNA. The exome sequencing linking the expressions back to their genetic blueprints has already been proven in a previous “seedling” program involving skin cells that was a collaboration between IARPA and Lawrence Livermore. The as-yet-unpublished work involved a model system of mothers, fathers and children, and their collected hair and DNA. The test sets proved the scientists could work backward from proteins to DNA sequence predictions.
A boon to the project is the ready-made databases from other sources, especially SNP data for health and disease research. That data is already teeming with DNA markers that will power the work, the experts said.
“We don’t have to go out and say, ‘what does the keratin exome look like?’” said Anex. “We go out into these huge databases that have already been put together and we harvest from them.”
In the future, there is the possibility of creating a protein database that would be an analog to CODIS at use in DNA databases. But Jordan explained it would not be replacing the forensic “gold standard” of DNA, even when brought to full fruition.
“This isn’t going to be the evidence that will be the driver and will be used in court,” said Jordan. “This will be the augmenting information. It’s just another tool we can use to disambiguate—the mixtures, for example—and pull out that one person. [It] gives us the strength to do it, when we can.”
It does represent a new scientific tool that could fill some significant forensic niches, Jordan added.
“There is going to be power here, and there is a lot of excitement,” she said.
Proteos is a completely new scientific project, and difficult even by IARPA standards.
“It’s a harder program, because nobody has done this before,” Jordan told Forensic Magazine. “We’re kind of building a plane while we’re flying it—and in terms of building test and validation protocols, we’re doing this all brand-new. It’s really challenging.”
The ultimate goal is to have a forensic kit that could do it all—one containing the reagents needed to do the extraction of both the proteins, and the potential mitochondrial DNA, simultaneously, according to Anex. Such a kit would only be years away from being realized at crime scenes, Anex predicted.
“Five years is as good a guess as any. Ten years would be way too long,” he said.