PerspectiveInfectious Disease

Toward a Systems-Level Analysis of Infection Biology: A New Method for Conducting Genetic Screens in Human Cells

See allHide authors and affiliations

Science Translational Medicine  16 Dec 2009:
Vol. 1, Issue 11, pp. 11ps13
DOI: 10.1126/scitranslmed.3000550


Loss-of-function genetic screens have facilitated great strides in our understanding of the biology of model organisms but have not been possible in diploid human cells. A recent report by Brummelkamp’s group in Science describes the use of insertional mutagenesis to generate loss-of-function alleles in a largely haploid human cell line and demonstrates the versatility of this method in screens designed to investigate the host/pathogen interaction. This approach has strengths that are complementary to existing strategies and will facilitate progress toward a systems-level understanding of infectious disease and ultimately the development of new therapeutics.

The development of a systematic understanding of mammalian biology and the signaling networks that govern complex processes has been hampered historically by a lack of tools for probing mammalian cells in a comprehensive and global way. Recently, such tools have begun to emerge. For example, techniques for analyzing and quantifying the amounts of all gene transcripts, proteins, and metabolites in cell populations enable the analysis of global responses of cells to various perturbations. The emergence of RNA interference (RNAi)–based technologies for decreasing the synthesis of selected gene products has given researchers the ability to perform genome-wide scans for factors involved in a variety of cell biological phenotypes. However, despite the many advances that have been facilitated by the use of RNAi, such strategies have certain inherent limitations.

In a recent issue of Science, Carette et al. described a new method for conducting genetic screens using a knockout library of human cells created in a haploid human cell line (1). Knocking out genes at the level of DNA circumvents many of the limitations of RNAi, such as the difficulty of achieving complete, stable knockdown of a gene product while avoiding nonspecific knockdown of other bystander genes. To demonstrate the utility and versatility of their approach, the researchers performed a series of screens designed to assess the method’s ability to illuminate fundamental processes in infectious disease.

During the complex interplay between a pathogen and its host, the infecting organism attempts to manipulate or subvert aspects of the host’s biology to create a niche for its survival and replication while evading the host’s immune responses. An emerging model for anti-infective drug discovery involves the targeting of host pathways that are critical for the replication or survival of the infectious agent (2). Because of the markedly lower mutation rates of human genes relative to those of infecting pathogens, this new class of host-targeted antimicrobial drugs may circumvent problems with the development of pathogen resistance to frequently used antibiotics. Current strategies being explored for host-targeted antimicrobial drugs include blocking host receptors required for pathogen entry into cells, perturbing the trafficking of pathogens within host cells, and blocking host signaling pathways required for pathogen growth and survival. For many pathogens, however, we have little understanding of the molecular mechanisms that underlie the host/pathogen interaction. Systematic methods for identifying such mechanisms would greatly facilitate the development of host-targeted therapies.

RNAi techniques have made possible broad, unbiased investigations into the identities and functions of mammalian host factors that participate in the infection process. In the field of bacterial pathogenesis, several groups have used RNAi to conduct genome-wide screens in Drosophila S2 cells to identify host factors that are involved in innate protection against bacterial pathogens or are exploited by pathogens for their survival and growth. Host/pathogen interactions studied in this manner include Listeria monocytogenes and Mycobacterium fortuitum infection of Drosophila cells (35). Host genes identified in these screens were subsequently verified for relevance in mammalian systems by means of lower-throughput approaches. The ability to perform the primary genome-wide RNAi screens directly in mammalian cell lines is now becoming more widespread, as evidenced by recent searches for host factors required for replication of the hepatitis C, West Nile, and human immunodeficiency viruses; these efforts uncovered host proteins with previously undiscovered roles in viral entry, integration, and replication (69). In addition, a genome-wide RNAi screen in human cells recently identified factors that mediate the host’s susceptibility to anthrax toxin (10).

Despite the remarkable insights gained through the use of RNAi-based screens, this technology has certain limitations. The first is related to the delivery of small interfering RNAs (siRNAs) into cultured cells (11). Synthetic siRNAs that directly inhibit the expression of target genes can be transfected into cells to reduce the amounts of specific gene products. Although this method can be effective, it is limited by the transient nature of the reduction, particularly in rapidly dividing cells. A more stable method introduces siRNAs or small hairpin RNAs (shRNAs) into cells with a vector-delivery system, such as a replication-deficient lentivirus. shRNAs are processed inside the cell into functional siRNAs. Vector delivery of shRNAs or siRNAs has advantages over the siRNA transfection method, including the achievement of stable reductions in the amounts of targeted gene products and the ability to introduce the small RNA–carrying vectors into a wide range of cell types, including primary cells. Additional problems with RNAi-mediated gene silencing that are independent of the delivery method are the difficulty of achieving complete inhibition of target gene product synthesis using siRNAs and the inability to predict the efficacy of a given siRNA sequence before testing. To circumvent these problems, researchers who conduct genome-wide screens must test multiple siRNA constructs per gene. Finally, perhaps the most challenging aspect of siRNA studies is that siRNAs often engender phenotypes by affecting genes other than the one of interest, thus eliciting so-called off-target effects.

New genetic strategies that circumvent the challenges associated with RNAi would be extremely valuable for probing the immune response to infection as well as mechanisms by which pathogens subvert host cell function. An ideal approach would be a loss-of-function genetic screen similar to those carried out in genetically amenable model organisms, such as yeast and flies. The work of Carette et al. describes the development of such an approach, which circumvents one of the major challenges of mammalian genetic systems: the diploid nature of the genome (Fig. 1). Using an essentially haploid human cell line, Carette et al. performed insertional mutagenesis with a gene-trap retrovirus (12) to generate a library of insertional mutants, each containing, on average, a single insertion in the genome. The authors took advantage of a derivative of the human KBM-7 chronic myeloid leukemia (CML) cell line that is haploid for every chromosome except number 8. The haploid karyotype of this cell line is relatively stable for many weeks in culture, and the repeated selection of haploid populations allows for the continuous maintenance of near-haploid human cells (13). These single-gene deletions were then used for functional studies.

Fig. 1. Partner proteins revealed.

In the genetic screening strategy of Carette et al., human KBM-7 cells, which are haploid for all chromosomes except number 8, were mutagenized using a gene-trap retrovirus. The resulting mutant library contained clones with an average of one mutation per cell. Aliquots of the pool of mutant cells were cultured in 96-well plates at a density of 20,000 cells per well, and in four separate screens, the cells were treated with either Gleevec, CDT, an ADP-ribosylating toxin, or influenza virus. Resistant mutants were isolated after a period of selection. The use of this screening strategy revealed new genes that mediate the killing of host cells by these toxic small-molecule and pathogenic challenges.


Having generated a potential library of haploid human cells with individual genes deleted, the authors performed a number of genetic screens to validate the performance of the library and demonstrate its utility for identifying new genes involved in pathogenic processes. All the screens performed were designed to identify host mutants that are resistant to a lethal selection. The mutant library was exposed to a lethal agent (for example, a drug, toxin, or virus) that kills the majority of cells within the library. Mutants that were resistant to the toxic agent survived and were thus amenable to isolation. Analysis of these survivors was then performed to determine the identity of the gene whose knockout conferred resistance. In their initial proof-of-principle screens, the researchers successfully identified known factors required for the killing of cells by the toxic small molecules 6-thioguanine and Gleevec (14) and for tumor necrosis factor–related apoptosis-inducing ligand (TRAIL)–mediated apoptosis. The screen for factors that promote resistance to Gleevec, which inhibits the tyrosine kinase activity of c-Abl and is used as a treatment for CML, recovered cells that contained mutations in the PTPN12 gene, which encodes the PTPN12 tyrosine phosphatase. This enzyme had not previously been implicated in the cellular response to Gleevec but is a known negative regulator of c-Abl. This raises the exciting possibility that the method of Carette et al. will be useful for analyses of drug mechanisms of action and the identification of mutations that spur resistance to clinically important drugs.

The authors next performed a series of screens to identify host factors involved in a variety of infectious processes. To this end, three separate screens were performed in which KBM-7 cells were exposed to two bacterial toxins or an influenza virus. The first toxic agent, cytolethal distending toxin (CDT), is a protein produced by several pathogenic Gram-negative bacteria that cleaves host DNA, inducing cell cycle arrest and ultimately cell death (15). Host genes essential for CDT entry into cells and the toxin’s actions once inside have not been previously described. By treating the KBM-7 insertional mutant library with CDT and selecting for cells that survived toxin exposure, the authors identified two host genes required for CDT lethality, both of which likely mediate entry of the toxin into host cells. The first gene encodes sphingomyelin synthase 1, which is required for synthesis of the lipid sphingomyelin. Along with cholesterol, sphingomyelin is an important component of lipid rafts, and a reduction in the amount of sphingomyelin disrupts raft function and cell-surface receptor clustering during signal transduction. The depletion of cholesterol has been shown to abolish CDT binding to cells, and this new finding that sphingomyelin is required for CDT killing of cells is consistent with a role for lipid rafts in CDT entry. The second gene identified, TMEM181, encodes a heterotrimeric GTP-binding protein–coupled receptor that the authors show binds to CDT at the cell surface. The authors also demonstrate that overexpression of the TMEM181 protein renders a number of other cell types sensitive to CDT-mediated intoxication, suggesting that TMEM181 is a receptor for CDT.

In a second screen, the mutant library was used to identify host factors required for the activity of the adenosine diphosphate (ADP)–ribosylating agents diptheria toxin and exotoxin A, both of which inhibit cellular protein synthesis. These toxins function by the action of ADP-ribosylating elongation factor 2 (EF2) on dipthamide, a posttranslationally modified histidine that is uniquely present in EF2 in eukaryotes (16). The mechanism by which ADP-ribosylating toxins enter cells and induce host cell death is well understood at the molecular level and thus represents a challenging system for the identification of new pathway components. Despite this substantial barrier to discovery, screening of the mutant KBM-7 library identified a previously uncharacterized gene, WDR85. The authors went on to show that WDR85 is involved in dipthamide biosynthesis, a finding that is conserved across eukaryotes from yeast to mammals. The identification of a gene not previously known to be involved in diphthamide biosynthesis demonstrates the power of this method, because most researchers believed that all of the components of this biosynthetic process had been characterized previously.

Finally, the KBM-7 mutant library was used to screen for factors that govern the replication of pathogens inside host cells. In a screen for host factors essential for influenza infectivity, the authors selected for mutant cells that were resistant to infection by influenza virus A, and they identified two genes required for sialic acid–containing glycoconjugate synthesis and, therefore, viral entry. In a similar RNAi-based screen conducted in Drosophila S2 cells, more than 100 genes required for influenza virus replication were identified (17). It is noteworthy that neither of the genes identified in the KBM-7 mutant cell screen was identified in the RNAi screen of S2 cells. The smaller number of genes identified in the KBM-7 mutant cell screen presumably results from not having performed the screen to saturation; however, the capacity of this library for identifying large numbers of genes required for an infectious process has yet to be demonstrated and will be dependent on the ability to perform saturating screens.

The results from the various screens conducted with the KBM-7 mutant library demonstrate the strengths of this method and indicate that this strategy is valuable and well suited for studying aspects of host/pathogen interactions. This new screening approach has an increased potential for the complete knockout of selected gene products over a sustained period of time, relative to RNAi technology, and thus is likely to be complementary to RNAi-based screening. However, because not all pathogens are capable of infecting KBM-7 cells, the full range of phenotypes that can be studied with this system is yet to be determined. The authors of the study do show that their system can be used to make initial discoveries of genes that can then be translated to phenotypes that are relevant to many more cell lines, as demonstrated by the finding that expression of the TMEM81 receptor in multiple cells lines confers sensitivity to CDT. In addition, the authors expand the utility of this cell line through the use of cleverly designed experiments. For example, although anthrax lethal toxin has no effect on KBM-7 cells, the mutant library can still be used to study anthrax toxin internalization and trafficking events. By fusing the cell-binding domain of anthrax toxin to the catalytic domain of diptheria toxin, the authors designed a selection strategy that allows the internalization of anthrax toxin to be probed through the readout of cell death caused by diptheria toxin.

A remaining challenge for the widespread utility of this system is the characterization of its capacity for screening nonlethal phenotypes, which will affect the applicability of the system beyond the study of lethal drug mechanisms and fatal host/pathogen interactions. It is not yet known whether the haploid state of these cells is stable enough to allow the technical screening of nonlethal phenotypes, which would require further manipulation, including, for example, the introduction of genetic reporters, fluorescence-activated cell sorting, or the construction of arrayed libraries of clones to facilitate non-pooled screening. In addition, it will be important to define the range of biological processes that can be studied in this particular CML-derived cell line. The examples that the authors have selected are well chosen, as these particular drug or pathogen exposures are well suited to this cell type. Many additional areas of biology could benefit from such a system—in particular, ones that require the elucidation of cell signaling pathways and networks—if the relevant pathways and networks are intact in KBM-7 cells and if it is possible to adapt the system to study nonlethal phenotypes.

The complexity of host/pathogen interaction arises from the vast number of innate cellular and immune processes that hosts mount to fight infection and the many methods by which pathogens subvert these processes. Thus far, the majority of research concerning the biology of infection has taken the traditional reductionist approach of examining these pathways at a detailed mechanistic level on a one-by-one basis. Investigational approaches that probe the complexities and dynamics of the system as a whole will lead to a more integrated and complete understanding of the biology of infection. New techniques for examining host responses to infection analyze changes in the transcriptome, proteome, and metabolome at a global level. The integration of information from multiple studies will enable a more complete understanding of the diverse cellular networks that collaborate in the host’s response to infection than is possible with a reductionist approach. Several groups have used proteomic or gene-expression data in concert with RNAi-based screening results to construct such networks. For example, researchers recently meshed RNAi screening with global transcriptional profiling to build a network of cell signaling pathways that form the mechanistic basis for pathogen-specific responses by dendritic cells (18). The mutant human library described by Carette et al. is an important addition to the increasing numbers of approaches that will facilitate such systems-based approaches. At a time when new pathogenic bacterial strains that are resistant to conventional antibiotics continue to surface, a systems-level understanding of infectious disease processes should have a profound impact on translational medicine by facilitating the discovery of fresh approaches to antimicrobial therapeutics. Furthermore, we expect that scientists will be able to adapt the screening approach of Carette et al. to permit investigation of other physiological and pathophysiological conditions of interest.


  • Citation: S. A. Stanley and D. T. Hung, Toward a systems-level analysis of infection biology: A new method for conducting genetic screens in human cells. Sci. Transl. Med. 1, 11ps13 (2009).

References and Notes

  1. This work was supported by a Pew Foundation for Biomedical Research grant to D.T.H. (grant no. 2006-000116).
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article