Report

Gene essentiality and synthetic lethality in haploid human cells

See allHide authors and affiliations

Science  27 Nov 2015:
Vol. 350, Issue 6264, pp. 1092-1096
DOI: 10.1126/science.aac7557

Zeroing in on essential human genes

More powerful genetic techniques are helping to define the list of genes required for the life of a human cell. Two papers used the CRISPR genome editing system and a gene trap method in haploid human cells to screen for essential genes (see the Perspective by Boone and Andrews). Wang et al.'s analysis of multiple cell lines indicates that it may be possible to find tumor-specific dependencies on particular genes. Blomen et al. investigate the phenomenon in which nonessential genes are required for fitness in the absence of another gene. Hence, complexity rather than robustness is the human strategy.

Science, this issue p. 1096 and p. 1092; see also p. 1028

Abstract

Although the genes essential for life have been identified in less complex model organisms, their elucidation in human cells has been hindered by technical barriers. We used extensive mutagenesis in haploid human cells to identify approximately 2000 genes required for optimal fitness under culture conditions. To study the principles of genetic interactions in human cells, we created a synthetic lethality network focused on the secretory pathway based exclusively on mutations. This revealed a genetic cross-talk governing Golgi homeostasis, an additional subunit of the human oligosaccharyltransferase complex, and a phosphatidylinositol 4-kinase β adaptor hijacked by viruses. The synthetic lethality map parallels observations made in yeast and projects a route forward to reveal genetic networks in diverse aspects of human cell biology.

Single-cell organisms can often tolerate inactivating mutations in the majority of genes (13), but it is unclear whether human cells require more essential genes because of increased complexity, or fewer because of added redundancy. To study this, we used mutagenesis in the near-haploid chronic myeloid leukemia (CML) cell line KBM7 (karyotype 25, XY, +8, Ph+), and its nonhematopoietic derivative HAP1, which is haploid for all chromosomes (fig. S1A) (4). More than 34.3 million and 65.9 million gene-trap integrations were identified in KBM7 and HAP1 cells, respectively. The gene-trap vector we used was unidirectional by design (fig. S1B), and for most genes, the number of intronic integrations in the sense direction was similar to that in the antisense direction (for example, pro-apoptotic factor BBC3) (Fig. 1, A and B) (57). For a fraction of genes, however, disruptive mutations were underrepresented, which is indicative of impaired fitness: Some genes (such as STAT5B) appeared essential in one cell type (fig. S1C), whereas others appeared essential in both (such as RPL13A) (Fig. 1, A and B, and tables S1 to S3).

Fig. 1 Identification of genes required for fitness in KBM7 and HAP1 cells through insertional mutagenesis.

(A) Distinct gene-trap insertions were mapped in KBM7 and HAP1 cells, and their orientation relative to the affected genes was counted. Per gene, the percentage of sense orientation gene-trap insertions (y axis) and the total number of insertions in a particular gene (x axis) are plotted. (B) Gene-trap insertions identified in the sense (S, yellow) or antisense orientation (AS, blue) in a nonessential gene (BBC3), a gene essential only in KBM7 cells (STAT5B), and a gene essential in both cell lines (RPL13A). (C) KEGG pathway enrichment analysis of essential genes shared between or specific to KBM7 or HAP1 cells. (D) Properties of new and old essential genes compared with the human genome. Averages for the sets are displayed, except for protein abundance, in which median Exponentially Modified Protein Abundance Index (emPAI) values are shown.

In KBM7 cells, 2054 genes (table S1 and figs. S2 and S3) and in HAP1, 2181 genes (table S2 and figs. S2 and S3) appeared to be needed for viability or optimal fitness under the experimental growth conditions (referred to as “essential,” although the approach does not distinguish between the two). The 1734 genes identified in both cell lines were designated as “core essentialome” (table S3). In KBM7 cells, genes on chromosome 8 (present in two copies) tolerated disruptive mutations, underscoring the specificity of the approach (fig. S1D). Furthermore, nearly all subunits of the proteasome were identified as essential (fig. S4). In general, essential genes are overrepresented in categories such as translation or transcription but not signaling (Fig. 1C and figs. S5 and S6).

Many genes required for fitness in yeast were also essential in human cells. Exceptions were largely explained by paralogs in the human genome or by yeast-specific requirements (fig. S7A and table S4) (1). We estimated the evolutionary age of essential genes and found that 77% emerged in premetazoans (“old” essential genes) (fig. S7B). Essential genes had fewer paralogs and higher protein abundance and contained fewer single-nucleotide polymorphisms (SNPs) predicted to impair function (Fig. 1D). Proteins encoded by essential genes displayed more protein-protein interactions (fig. S8, A to D), and these occurred more frequently with other essential proteins (49.8%) (fig. S9A) and within the same functional category (fig. S5B). Remarkably, the products of “new” essential genes are more often connected with old rather than other new essential gene products, suggesting that they largely function within ancient molecular machineries (fig. S9, B and C).

To identify proteins interacting with products of 18 uncharacterized essential genes, we used tandem affinity purification coupled to mass spectrometry (fig. S10). Interactors were frequently essential proteins (52.4%, P < 2.5 × 10–36, hypergeometric test) involved in processes such as splicing, translation, and trafficking (fig. S11 and table S5). The small transmembrane protein TMEM258 associated with components of the conserved oligosacharyltransferase (OST) complex (Fig. 2A and fig. S12A) that are essential for protein N-glycosylation (8). TMEM258 localized to the endoplasmic reticulum (fig. S12B), and depletion (fig. S12, C and D) impaired OST catalytic activity as monitored by means of hypoglycosylation of prosaposin (Fig. 2B) (9). This also rationalizes the observed clustering of TMEM258 with OST complex subunits in a recent genetic screen (10). Thus, TMEM258 constitutes a subunit of the human OST complex, and although homology searches (fig. S12E) do not identify a yeast ortholog, TMEM258 may relate to the similarly sized yeast transmembrane protein OST5 (11).

Fig. 2 The essential gene TMEM258 encodes a component of the OST complex.

(A) High-confidence protein-protein interactions associated with TMEM258. Green proteins indicate members of the OST complex. Dashed lines indicate the OST complex subnetwork. (B) Effects of depletion of TMEM258 with small interfering RNAs on the glycosylation of endogenous prosaposin. Cells were pulsed with 35S-methionine/cysteine, lysed, and subjected to immunoprecipitation by using antibodies to prosaposin. Precipitated proteins were detected by means of phosphorimaging, and hypoglycosylated prosaposin species are indicated. Tunicamycin treatment and depletion of the established OST subunit DDOST served as positive controls.

Whereas most genes appear nonessential, their function may be buffered by other genes so that only simultaneous disruption is lethal (1215). The frequency of such synthetic lethal interactions between human genes is debated and challenging to address experimentally (16, 17). We studied the small guanosine triphosphatases (GTPases) RAB1A and RAB1B by creating individual knockout lines and assessing the genes needed for fitness in these backgrounds (Fig. 3A and fig. S13A). Whereas neither RAB1A nor RAB1B were essential in wild-type cells, RAB1A became indispensable in RAB1B knockout cells and vice versa (Fig. 3A and fig. S13B). To explore the breadth of synthetic lethality, we probed the secretory pathway using three independent knockout cell lines (fig. S14) for RAB1A, RAB1B, GOSR1 [a subunit of the Golgi soluble N-ethylmaleimide–sensitive factor attachment protein (SNAP) receptor] (18), and TMEM165 (a Golgi-resident Ca2+/H+ antiporter whose deficiency impairs glycosylation) (Fig. 3B, figs. S15 and S16, and table S6) (19). Most of their genetic interactions impinged on the secretory pathway (Fig. 3B and table S7), and many were found synthetic lethal with PTAR1. Synthetic lethality screens in PTAR1-deficient cells confirmed these genetic interactions and additionally identified the uncharacterized gene C10orf76 (Fig. 3B and fig. S17A). Validation by use of C10orf76 as query gene confirmed synthetic lethality with PTAR1 and (reciprocally) identified TSSC1, which was recently reported to interact with the Golgi-associated retrograde protein complex (GARP) (Fig. 3B) (20). The human genes we studied display on average ~20 synthetic lethal interactions, a number comparable with that in yeast (12), although this varies between genes, with PTAR1 (causing a fitness defect when deleted alone) having close to 60 interactions (fig. S17B). This illustrates that synthetic lethal interactions can be identified and validated by using reciprocal haploid screens and that, similarly to yeast, interactions frequently occur between genes whose products act in related processes (fig. S17B) (13, 16). However, we acknowledge a caveat that this approach cannot readily distinguish between synthetic lethal or synthetic “sick” interactions.

Fig. 3 Synthetic lethality network generated based on mutations.

(A) Essentiality of RAB1A and RAB1B in wild-type HAP1 cells and cells deficient for RAB1A or RAB1B. (B) Genetic interaction network indicating synthetic lethal/sick interactions that were identified by scoring genes for fitness reduction in three nuclease-generated knockout clones per genotype. This revealed an interconnected network, with many genes that could be functionally assigned to the secretory pathway (labeled in green). Reciprocal interactions, scored in either query genotype, are indicated by orange edges. Edge thickness reflects the effect size of the interaction (compared with wild-type cells).

The impaired growth of PTAR1-deficient cells (table S2) was suppressed by loss of the Golgi factor GOLGA5 (Fig. 4A and fig. S18) (21). PTAR1-deficient cells had an abnormally dilated Golgi morphology (fig. S19A), which was partially corrected by codeletion of GOLGA5 (Fig. 4B and fig. S19B). Functionally, PTAR1 deficiency impaired glycosylation (fig. S19C) (7), possibly owing to dysregulation of RAB proteins (22). Indeed, PTAR1-deficient cells showed attenuated geranylgeranylation of RAB1A and RAB1B (fig. S19D). Partial correction of the Golgi morphology in cells lacking both PTAR1 and GOLGA5 could relate to the effect of GOLGA5, itself a RAB effector, on Golgi fragmentation (21, 23). Thus, the interaction map reveals PTAR1 and GOLGA5 as opposing handles tuning Golgi morphology and homeostasis.

Fig. 4 Roles of PTAR1 and C10orf76 in Golgi homeostasis and virus replication.

(A) A bias for sense-orientation integrations in GOLGA5 observed in PTAR1-deficient HAP1 cells but not wild-type cells. (B) Electron micrographs of the Golgi apparatus (orange highlight) in the indicated genotypes. (C) Interaction of FLAG-tagged C10orf76 with PI4KB in HAP1 cells detected with immunoprecipitation by using antibodies to FLAG. (D) Coxsackievirus A10 amplification in wild-type and C10orf76-deficient cells measured by means of single molecule fluorescent in situ hybridization (smFISH) to localize individual viral genomes (red). Intracellular viral RNA was first detected after 30 min. Increased RNA signal after 300 min indicates RNA replication.

Genetic analysis suggested a link between the unstudied gene C10orf76 and PI4KB, which were both synthetic lethal with PTAR1 (Fig. 3B). A host factor screen with coxsackievirus A10 also identified C10orf76 as well as PI4KB (fig. S20, A and B), and a proteomics survey (24) suggested association between C10orf76 and PI4KB. We confirmed this interaction in immunoprecipitation experiments with cells expressing FLAG-tagged C10orf76 (Fig. 4C). Phosphatidylinositol 4-kinase β (PI4KB) regulates abundance of phosphatidylinositol 4-phosphate [PI(4)P] (25) and has a role in genome replication of various RNA viruses, including coxsackieviruses (26). Infection studies confirmed that cells in which C10orf76 was knocked out were particularly resistant to coxsackievirus A10 (fig. S20C). Although virus entry occurred normally, replication of viral RNA was decreased in cells in which C10orf76 was knocked out (Fig. 4D and fig. S20D). Enteroviruses hijack PI4KB activity to construct “replication factories,” which were abundant in wild-type cells but rare in C10orf76-deficient cells (fig. S20E). Amounts of PI(4)P were decreased in these cells, and Golgi retention of PI4KB after chemical inhibition (27) was largely dependent on C10orf76, which also localized to this compartment under these conditions (fig. S21, A and B). Thus, C10orf76 is a PI4KB-associated factor hijacked by specific picornaviruses for replication.

This study identifies ~2000 genes required for optimal fitness of cultured haploid human cells. Despite technical limitations, the identification of gene essentiality shows high concordance with the gene-trap and clustered regularly interspaced short palindromic repeats (CRISPR) data reported in the accompanying manuscript of Wang et al. (supplementary text and fig. S22) (28). This suggests that the increase in total number of genes in humans as compared with that in yeast yielded a system of higher complexity rather than more robustness through added redundancy. Nonessential human genes appear to frequently engage in synthetic lethal interactions. Our studies start to reveal an interconnected module of genetic interactions affecting the secretory pathway and link it to uncharacterized genes. The experimental strategy is applicable to various cellular processes and may help unravel the genetic network encoding a human cell.

Supplementary Materials

www.sciencemag.org/content/350/6264/1092/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S23

Tables S1 to S8

References (2962)

References and Notes

  1. Acknowledgments: We thank T. Sixma, S. Nijman, J. Roix, J. Neefjes, and members of the Brummelkamp group for discussion and K. Sandhoff for antibodies directed against saposin. This work was supported by the Cancer Genomics Center (CGC.nl), Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO)–VIDI grant 91711316, European Research Council (ERC) Starting Grant (ERC-2012-StG 309634) to T.R.B., the Austrian Academy of Sciences and the Advanced ERC grant i-FIVE and Austrian Science Fund grant (FWF SFB F4711) to G.S.-F, and the ARC Foundation grant PJA 20141201975 to J.C. Sequence data have been deposited at the National Center for Biotechnology Information Sequence Read Archive with accession number SRP058962.
View Abstract

Navigate This Article