The repurposing of a bacterial defence system known as CRISPR into a potent activator of gene expression in human cells enables powerful studies of gene function, as exemplified in cancer cells. See Article p.583
The ability to turn on any gene at will has been a long-held dream of molecular biologists. Most genes are dynamically turned on and off by specific biological processes, and manipulation of the level of gene expression is a key method for studying the functions of each gene, regulatory element and pathway. On page 583 of this issue, Konermann et al.1 describe an elegant strategy for converting the CRISPR/Cas9 system — a bacterial defence system against foreign DNA — into a potent and selective gene activator. The authors demonstrate how this approach can be used to test the effects of turning on tens of thousands of individual genes in parallel.
CRISPR, which stands for clustered regularly interspaced short palindromic repeats, is the name given to regions of bacterial DNA encoding RNA sequences that recognize foreign DNA sequences, such as those in viruses, through direct sequence complementarity. In bacteria, these guide RNAs assemble into complexes with the enzyme Cas9 or other proteins that specifically cut the recognized DNA sequence, thereby destroying it and protecting the bacterium from invasion2. CRISPR/Cas9 thus represents a programmable DNA-targeting system, with its specificity determined by the RNA sequence.
Molecular biologists have adopted this system to allow the rapid mutation or replacement of genomic sequences — a strategy called genome editing. A breakthrough came in 2012, when the system was simplified to use single short guide RNA (sgRNA) molecules to programme CRISPR specificity3. Soon after, it was found that a mutant of Cas9 that no longer cuts DNA, termed dCas9, can be used as a DNA-binding platform4,5. When dCas9 is linked to portions (domains) of proteins involved in transcriptional activation or repression and then targeted, using CRISPR, to promoter sequences that regulate transcription of particular genes, these fusion proteins can modulate natural gene-expression levels4,5. However, the change in gene expression achieved by this approach is too low — less than or around fivefold activation — for many applications.
Konermann and colleagues overcame this low efficiency of gene activation by turning the CRISPR sgRNA into a modular platform for assembling multiple different transcriptional activators (Fig. 1a). They identified two regions of the sgRNA that can be appended with short sequences that attract an RNA-binding protein, which is in turn fused to the transcription-activation domains of different mammalian transcription factors. The authors termed this system the synergistic activation mediator (SAM), and demonstrate that it induced more than 100-fold activation of 12 genes that were not efficiently activated by the dCas9–activator fusion protein.
To illustrate the potential applications of this approach, Konermann et al. created a library of engineered sgRNAs that allowed more than 23,000 human genes to be individually turned on. They then asked which genes, on activation, give melanoma cancer cells the ability to escape the killing effects of the drug PLX-4720, a mainstay of melanoma treatment. The degree of drug resistance conferred by turning on different genes was determined by the relative frequency of sgRNAs in the melanoma cells after drug treatment. The highly enriched sgRNAs included those corresponding to genes involved in known drug-resistance pathways and to genes that are expressed at increased levels in patients with drug-resistant melanoma — verifying that the SAM method can identify biologically relevant outcomes of altered gene expression.
The success of this CRISPR engineering effort has direct parallels with natural mechanisms of gene regulation. Enhancers are DNA sequences that turn on gene expression, and they typically contain recognition sequences for several different types of transcription factor. Moreover, enhancers and other regulatory elements often generate long non-coding RNAs (lncRNAs), which act as modular scaffolds to recruit diverse cellular machines that modify chromatin (the complex of histone proteins and DNA in the cell nucleus) and thereby regulate gene expression (Fig. 1b). LncRNAs are molecular recipes writ large, containing both instructions for the set of biochemical activities to be assembled and the genomic address at which these activities should be carried out6.
Two other recent studies have also demonstrated improved efficiency or flexibility of CRISPR-guided gene regulation, through repurposing dCas9 or CRISPR RNA as modular scaffolds7,8. One of those studies8 further showed that multiple engineered CRISPR RNAs can simultaneously turn different genes in the same cell on and off to manipulate a metabolic pathway. Together with Konermann and colleagues' findings, these studies show that mimicking natural lncRNAs is an efficient way to orchestrate multiple proteins to work together across the genome. It may be possible to deliver defined combinations of various effector proteins to the same genomic location using one sgRNA molecule. Future construction of novel multifunctional artificial proteins or non-coding RNAs will also be worthwhile, given the broad usefulness of such tools in biotechnology.
This next generation of CRISPR technology opens the door to studying the functions of many genes and DNA sequences. RNA-interference techniques have been widely used for studying the effects of loss of gene function over the past decade, but this approach can yield a high rate of false-positive results due to nonspecific targeting. Meanwhile, gain-of-function studies using overexpression techniques may not recapitulate normal RNA regulatory processes, such as alternative splicing. Artificial transcription factors based on DNA-binding proteins called zinc fingers and TALENs are alternatives for altering gene expression, but these are difficult to construct on a genome-wide scale. Thus, the comprehensive coverage of CRISPR libraries and the modular nature of this approach are strong advantages over other techniques. However, CRISPR targeting may also have off-target effects9, and additional validation experiments may be needed to confirm any effects of altered gene expression identified using this approach.
In their melanoma-cell experiments, Konermann et al. identified 13 genes whose altered expression was individually sufficient to confer drug resistance. However, diseases or profound biological effects often result from complex regulation of multiple genes at the same time — a good example is the finding that four genes must be expressed together for the generation of induced pluripotent stem cells (a form of stem cell generated from adult cells)10. Thus, we will need a detailed understanding of regulatory networks and will need to experiment with gene sub-libraries and dosages to identify the sets of genes that together determine certain characteristics. The CRISPR/Cas system will be a versatile tool for this purpose, owing to its capacity for multiplexed targeting and, now, multiplexed deployment of diverse effector domains.