In one 2010 study, researchers at the DSMZ reported that of 598 leukemia and lymphoma cell lines in their bank, “187 (31%) were contaminated with mycoplasma and/or a second cell line and 38 (6%) of cell lines [were] contaminated with both.” (5) Thirteen percent of the cells in this analysis were listed as contaminated.
Though HeLa is the primary offender in human lines, there are others. MDA-MB-435, thought to be derived from breast cancer, was shown in a series of papers published in 2000–2007 to actually be M14 melanoma cells (6); the presumptive endothelial cell line ECV304 was shown in 1999 to be T24 bladder carcinoma (7); and in a 2009 retraction, a research team at the University of Birmingham, UK, acknowledged that what they thought were human thyroid epithelial cells (TEC61) were in fact JEG3 human choriocarcinoma cells, likely introduced into the cultures via JEG3-conditioned media used to support the cells’ growth. (8) That paper has been cited 12 times, according to the ISI Web of Knowledge.
Cell Line Authentication
Given the magnitude of the problem, it's no surprise to learn that cell banks subject their lines to some fairly rigorous testing. At CellBank Australia, for instance, the authentication process occupies at least one full-time employee (of a total of four), according to Capes-Davis.
Incoming lines are checked for mycoplasma contamination using three separate test methods and cross-contamination using two. Species of origin is established by sequencing a PCR-amplified segment of the mitochon drial cytochrome c oxidase I gene. Human lines are tested via short tandem repeat (STR) profiling, which is the same PCR-based method the FBI uses to populate its CODIS (Combined DNA Index) database of forensic DNA samples.
Based on variation in the length and number of repeats of certain loci scattered across the human genome, STR profiles are acquired via multiplex PCR and compared against international databases such as the one maintained at DSMZ (www.dsmz.de/human_and_animal_cell_lines/main.php). That database currently lists nearly 2300 profiles from DSMZ, ATCC, and two cell banks in Japan. A new global database to be housed at NCBI and conforming to ASN-0002 recommendations is in development and should be available this fall, according to Liz Kerrigan, Director of Standards and Certification at ATCC.
Such databases help banks determine whether lines really are unique. But the finger print really is just a fingerprint, explains Brian Douglass, the market development manager for Cell Biology and Standards at ATCC. Just as a fingerprint identifies an individual yet says nothing about that person's height or eye color, so too can an STR fingerprint confirm a match or lack thereof between two lines, but nothing at all about phenotype or origin.
A Broader Resolution
Gartler's announcement in the pages of Nature was just the first pebble in what has become a landslide of evidence of cellular subterfuge. Over the ensuing four decades, many researchers have strongly advocated for better authentication standards throughout the entire scientific community.
Finally, in 2007 the cell biology community came together to draft its open letter to HHS Secretary Leavitt calling on the NIH (and the research publishing community) to require authentication of cell lines in order to publish or receive federal funding. “This group of thought leaders essentially said that the issue of cell line misidentification is significant and plagues scientific research,” says Douglass. “It's also a waste of tax dollars, because these are federally funded grants.”
The NIH responded (9), agreeing that cell line misidentification is “a serious problem,” but one they could not practically be expected to handle. “Because authentication methods can be quite specific and are continuously evolving, it would be impractical for the NIH to require application of particular methods in all grant applications.” Instead, the agency called on journal peer-reviewers to help ensure cell line validity.
Some small funding agencies have already implemented authentication requirements. The Project Grant Application Handbook for the UK-based Association for International Cancer Research, for instance, states: “The authentication of cell lines is considered an important part of the preliminary data described in any grant application. Applicants should also ensure that all cell lines to be used during the proposed research are adequately authenticated.”
Likewise, journals have started to incorporate cell line validation requirements. Author instructions for the American Association for Cancer Research publications stipulate that work involving human cell lines must indicate where and when the cells were obtained, whether the cell lines have been tested and authenticated, how the cells were tested, and when. Similar statements appear in the International Journal of Cancer, In Vitro Cellular and Developmental Biology Animal, and others.
Nature has called for a global searchable database of DNA fingerprints that researchers could use to meet new, firmer guidelines that it suggests funding agencies and journals adopt. “Once this research framework is sufficiently established, major funders will be able to require the validation of all immortal cell lines in order for investigators to retain funding, and journals should (and Nature will) require that all lines used in a paper were verified before publication.” (10)
If the experience of the International Journal of Cancer is any guide, authentication requirements should pose minimal problems for journals. The IJC, which receives some 3,000 submissions per year, has had to “unsubmit” — or hold for further documentation — some 170 manuscripts since it initiated its rule change, says editor-in-chief Peter Lichter. Of those, 86 subsequently resubmitted with the required information, while only 28 have actively withdrawn their manuscripts. Complaints, he says, are rare.
Generally speaking, authentication services use 8 or 15 STR loci to establish cell line identity (plus an additional marker to define gender). According to Gabriela Saldanha, Global Strategic Marketing Manager at Promega Corp., which offers kits for STR testing, the odds of a random match using eight STR loci is about one in 100 million; for 15 loci, the odds are on the order of 1 × 10−17. But unlike crime lab forensics, cell line authentication has to allow for a little wiggle room.