Numerous cancer genes including, BRAF, HRAS, KRAS, MYC, PTEN and RB1, were first identified in cancer cell lines (CCL),1 which as a bonus permit high-throughput drug screening.2 However, most CCL in use date from the 1960s to 1980s and, as we argue here, urgently require updating. We have tracked the establishment of ∼1600 human leukemia/lymphoma cell lines,3 the easiest group to document, from the doyens of this group RAJI and JIJOYE in 1963, until present (Fig. 1a). After a faltering start original reports, nevertheless, climbed from the mid '70s until the early '90s, thereafter receding. In 2009–2010 barely 10 new leukemia/lymphoma CL were described. Assuming similar demographics for solid tumor CCL, why should the curve of invention now point downwards? Here we seek possible explanations and argue that, instead, more CCL are needed.
First, establishing cell lines lacks obvious novelty. Referees are unlikely to reward such work unless new methodologies are involved: a point returned to later. Users may perceive little need for yet more CL to add to the thousands already established. Investigators merely desiring living cells as vehicles for gene expression, say, might well prefer classics with published protocols and which might encourage citation. Indeed, interest in classic CL remains buoyant. Although long established, citation of HeLa (from cervical carcinoma in 1951) and K-562 [from chronic myeloid leukemia (CML) in 1970] climbs steadily while new CCL reports fall (Fig. 1b).
A second problem, cell line cross-contamination (CL–CC), shows tentative signs of improvement. This chronic threat (reviewed in Ref. 4) known from the 1970s was first quantified as recently as 1999 when at least 17% new CCL were shown to be affected.5 Thanks to commentaries in general journals the issue is now better ventilated. CL–CC affecting originator-derived material may even be declining: from 1991–2000 at 15% (59/395) original leukemia/lymphoma CCL supplied to the DSMZ tested false,6 falling to 5.6% (6/106) over the last decade (Fig. 1c). The corresponding figures for secondarily sourced material have scarcely altered, at 15% (23/155) for 1991–2000 and 13% (11/85) for 2001–2010.
Why so? Until very recently, there were no institutional means for alerting scientists to CL–CC: thus, citation of one egregious example—ECV-304 (purportedly endothelial, actually T-24 bladder cancer)—continued almost unabated for a decade after its unmasking in 1999 (Fig. 1d). The recent decline may be due to adverse publicity in Chinese journals wherein ECV-304 is heavily cited. Moreover, interests vested in previous and planned investigations may discourage openness regarding CL–CC affecting such work.
Recent web-based resources now promote awareness of CL–CC: a STR-profiling database pooling resources from major cell banks in USA, Japan and Germany7 enables false cell lines to be identified and withdrawn (http://www.dsmz.de/STRanalysis). Moreover, a regularly updated list of false cell lines compiled by Amanda Capes-Davies and Iain Freshney is available (http://www.dsmz.de/fileadmin/Bereiche/HumanandAnimalCellLines/Cross_Contaminations_v6_8-1.pdf). To promote awareness of false cell lines, the recently constituted International Cell Line Authentication Committee now pronounces on CL authenticity (http://standards.atcc.org/kwspub/home/the_international_cell_line_authentication_committee-iclac_/). Finally, the role of journals is vital: a growing minority (including this journal) mandate authors to give cell line provenances or document authentication. Alas, it has proved less easy to persuade funding agencies to follow suit.
“Genetic instability”—the third charge levelled against CCL—invoked to explain clonal diversity is all-too-often taken for granted, seldom defined and until recently almost impossible to measure. Early-passage CCL may be genomically heterogeneous and, unless carefully cultured, undergo bottlenecking selection and genetic drift. Although karyotypic complexity seen in CCL is often attributed to instability, catastrophic genomic rearrangement (chromothripsis) is now recognized as a tumorigenic mechanism in vivo.8 Access to primary or early-passage material, rare among classic CCL, assists distinguishing clonal diversity from genetic instability.
Which types of cell line do we need more of, and why? First, coverage must be extended to entities, and disease-phases currently un/under-represented by CCL. Thus, while, about 40 acute-phase CCL have been established from CML,3 there are none from the preceding chronic phase. High grade, poor prognosis entities are heavily over-represented in vitro. As well as informing entity-specific disease pathways, provision of cell lines representing early-phase or good prognosis cancers should illuminate disease progression and stratification.
Second, classic CCL rely on donor data which may be inaccurate or outmoded. It is axiomatic that CCL may not necessarily derive from the tumor diagnosed. It is desirable that CCL be compared with their supposed tumors of origin, both genomically and transcriptionally. Primary material is seldom, if ever, available for classic CCL. Thus, we need new CCL both for anchoring to current diagnostic entities and to facilitate authentication via STR matching with donor material.
Third, there is an unmet need for human non/pre-malignant control CL to facilitate interpretation of high-throughput genomic/transcriptional profiling and sequencing data. Non-viral methods for cell immortalization are in their infancy, restricted to specialist labs. Herein lie the greatest challenges.
Fourth, we need CCL from spontaneous animal cancers, notably from those with similar biologies (e.g., primates) or shared lifestyles (e.g., domestic pets). Most available animal CCL are from tumors induced by acute carcinogen exposure. The study of animal CCL inter alia informs cancer gene regulation and evolution, while opening an extra dimension for parallel gene testing in vivo.
To address the shortcomings of CCL as models for normal cells, there have been calls for radically altered establishment methods, such as differentiating stem cell cultures in vitro leading to the abandonment of HeLa.9 Although tumors may display stem cell characteristics, reverse engineering of pluripotent stem cell cultures from tumor cells has received little attention, as has establishment of CCL from isolated cancer stem-cells. Telomerase transfection-based methods for conditionally immortalizing multiple human primary cells, as reported for human bronchial epithelial cells10 is an attractive option. The development of new, and improvement of classical methods deserves wider encouragement. There is little sign however, of activity on this front. As it seems increasingly unlikely that existing methods will bear fruit, we join those calling for new methods, albeit to augment not replace HeLa.