CRISPR

The discovery of clustered DNA repeats occurred independently in three parts of the world. The first description of what would later be called CRISPR is from Osaka University researcher Yoshizumi Ishino and his colleagues in 1987. They accidentally cloned part of a CRISPR sequence together with the "iap" gene (isozyme conversion of alkaline phosphatase)[12] that was their target. The organization of the repeats was unusual. Repeated sequences are typically arranged consecutively, without interspersed different sequences.[13][5] They did not know the function of the interrupted clustered repeats.

In 1993, researchers of Mycobacterium tuberculosis in the Netherlands published two articles about a cluster of interrupted direct repeats (DR) in that bacterium. They recognized the diversity of the sequences that intervened the direct repeats among different strains of M. tuberculosis[14] and used this property to design a typing method that was named spoligotyping, which is still in use today.[15][16]

At the same time, repeats were observed in the archaeal organisms of Haloferax and Haloarcula species, and their function was studied by Francisco Mojica at the University of Alicante in Spain. Although his hypothesis turned out to be wrong, Mojica's supervisor surmised at the time that the clustered repeats had a role in correctly segregating replicated DNA into daughter cells during cell division because plasmids and chromosomes with identical repeat arrays could not coexist in Haloferax volcanii. Transcription of the interrupted repeats was also noted for the first time.[16][17] By 2000, Mojica performed a survey of scientific literature and one of his students performed a search in published genomes with a program devised by himself. They identified interrupted repeats in 20 species of microbes as belonging to the same family.[18] In 2001, Mojica and Ruud Jansen, who were searching for additional interrupted repeats, proposed the acronym CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) to alleviate the confusion stemming from the numerous acronyms used to describe the sequences in the scientific literature.[17][19] In 2002, Tang, et al. showed evidence that CRISPR repeat regions from the genome of Archaeoglobus fulgidus were transcribed into long RNA molecules that were subsequently processed into unit-length small RNAs, plus some longer forms of 2, 3, or more spacer-repeat units.[20]

A major addition to the understanding of CRISPR came with Jansen's observation that the prokaryote repeat cluster was accompanied by a set of homologous genes that make up CRISPR-associated systems or cas genes. Four cas genes (cas 1–4) were initially recognized. The Cas proteins showed helicase and nuclease motifs, suggesting a role in the dynamic structure of the CRISPR loci.[21] In this publication the acronym CRISPR was used as the universal name of this pattern. However, the CRISPR function remained enigmatic.

Simplified diagram of a CRISPR locus. The three major components of a CRISPR locus are shown: cas genes, a leader sequence, and a repeat-spacer array. Repeats are shown as gray boxes and spacers are colored bars. The arrangement of the three components is not always as shown.[22][23] In addition, several CRISPRs with similar sequences can be present in a single genome, only one of which is associated with cas genes.[24]

In 2005, three independent research groups showed that some CRISPR spacers are derived from phage DNA and extrachromosomal DNA such as plasmids.[25][26][27] In effect, the spacers are fragments of DNA gathered from viruses that previously tried to attack the cell. The source of the spacers was a sign that the CRISPR/cas system could have a role in adaptive immunity in bacteria.[22][28] All three studies proposing this idea were initially rejected by high-profile journals, but eventually appeared in other journals.[29]

The first publication[26] proposing a role of CRISPR-Cas in microbial immunity, by the researchers at the University of Alicante, predicted a role for the RNA transcript of spacers on target recognition in a mechanism that could be analogous to the RNA interference system used by eukaryotic cells. Koonin and colleagues extended this RNA interference hypothesis by proposing mechanisms of action for the different CRISPR-Cas subtypes according to the predicted function of their proteins.[30]

Experimental work by several groups revealed the basic mechanisms of CRISPR-Cas immunity. In 2007, the first experimental evidence that CRISPR was an adaptive immune system was published.[5][8] A CRISPR region in Streptococcus thermophilus acquired spacers from the DNA of an infecting bacteriophage. The researchers manipulated the resistance of S. thermophilus to different types of phage by adding and deleting spacers whose sequence matched those found in the tested phages.[31][32] In 2008, Brouns and Van der Oost identified a complex of Cas proteins (called Cascade) that in E. coli cut the CRISPR RNA precursor within the repeats into mature spacer-containing RNA molecules called CRISPR RNA (crRNA), which remained bound to the protein complex.[33] Moreover, it was found that Cascade, crRNA and a helicase/nuclease (Cas3) were required to provide a bacterial host with immunity against infection by a DNA virus. By designing an anti-virus CRISPR, they demonstrated that two orientations of the crRNA (sense/antisense) provided immunity, indicating that the crRNA guides were targeting dsDNA. That year Marraffini and Sontheimer confirmed that a CRISPR sequence of S. epidermidis targeted DNA and not RNA to prevent conjugation. This finding was at odds with the proposed RNA-interference-like mechanism of CRISPR-Cas immunity, although a CRISPR-Cas system that targets foreign RNA was later found in Pyrococcus furiosus.[5][31] A 2010 study showed that CRISPR-Cas cuts both strands of phage and plasmid DNA in S. thermophilus.[34]

Researchers studied a simpler CRISPR system from Streptococcus pyogenes that relies on the protein Cas9. The Cas9 endonuclease is a four-component system that includes two small crRNA molecules and trans-activating CRISPR RNA (tracrRNA).[35][36] Jennifer Doudna and Emmanuelle Charpentier re-engineered the Cas9 endonuclease into a more manageable two-component system by fusing the two RNA molecules into a "single-guide RNA" that, when combined with Cas9, could find and cut the DNA target specified by the guide RNA. By manipulating the nucleotide sequence of the guide RNA, the artificial Cas9 system could be programmed to target any DNA sequence for cleavage.[37] Another group of collaborators comprising Virginijus Šikšnys together with Gasiūnas, Barrangou and Horvath showed that Cas9 from the S. thermophilus CRISPR system can also be reprogrammed to target a site of their choosing by changing the sequence of its crRNA. These advances fueled efforts to edit genomes with the modified CRISPR-Cas9 system.[16]

Groups led by Feng Zhang and George Church simultaneously published descriptions of genome editing in human cell cultures using CRISPR-Cas9 for the first time.[5][38][39] It has since been used in a wide range of organisms, including baker's yeast (Saccharomyces cerevisiae),[40][41][42] the opportunistic pathogen Candida albicans,[43][44] zebrafish (Danio rerio),[45] fruit flies (Drosophila melanogaster),[46][47] ants (Harpegnathos saltator[48] and Ooceraea biroi[49]), mosquitoes (Aedes aegypti[50]), nematodes (Caenorhabditis elegans),[51] plants,[52] mice,[53][54] monkeys[55] and human embryos.[56]

CRISPR has been modified to make programmable transcription factors that allow scientists to target and activate or silence specific genes.[57]

The CRISPR-Cas9 system has shown to make effective gene edits in Human tripronuclear zygotes first described in a 2015 paper by Chinese scientists P. Liang and Y. Xu. The system made a successful cleavage of mutant Beta-Hemoglobin (HBB) in 28 out of 54 embryos. 4 out of the 28 embryos were successfully recombined using a donor template given by the scientists. The scientists showed that during DNA recombination of the cleaved strand, the homologous endogenous sequence HBD competes with the exogenous donor template. DNA repair in human embryos is much more complicated and particular than in derived stem cells.[58]

In 2015, the nuclease Cas12a (formerly known as Cpf1[59]) was characterized in the CRISPR/Cpf1 system of the bacterium Francisella novicida.[60][61] Its original name, from a TIGRFAMs protein family definition built in 2012, reflects the prevalence of its CRISPR-Cas subtype in the Prevotella and Francisella lineages. Cas12a showed several key differences from Cas9 including: causing a 'staggered' cut in double stranded DNA as opposed to the 'blunt' cut produced by Cas9, relying on a 'T rich' PAM (providing alternative targeting sites to Cas9) and requiring only a CRISPR RNA (crRNA) for successful targeting. By contrast Cas9 requires both crRNA and a transactivating crRNA (tracrRNA).

These differences may give Cas12a some advantages over Cas9. For example, Cas12a's small crRNAs are ideal for multiplexed genome editing, as more of them can be packaged in one vector than can Cas9's sgRNAs. As well, the sticky 5′ overhangs left by Cas12a can be used for DNA assembly that is much more target-specific than traditional Restriction Enzyme cloning.[62] Finally, Cas12a cleaves DNA 18–23 base pairs downstream from the PAM site. This means there is no disruption to the recognition sequence after repair, and so Cas12a enables multiple rounds of DNA cleavage. By contrast, since Cas9 cuts only 3 base pairs upstream of the PAM site, the NHEJ pathway results in indel mutations which destroy the recognition sequence, thereby preventing further rounds of cutting. In theory, repeated rounds of DNA cleavage should cause an increased opportunity for the desired genomic editing to occur.[63]

The CRISPR array is made up of an AT-rich leader sequence followed by short repeats that are separated by unique spacers.[64] CRISPR repeats typically range in size from 28 to 37 base pairs (bps), though there can be as few as 23 bp and as many as 55 bp.[65] Some show dyad symmetry, implying the formation of a secondary structure such as a stem-loop ('hairpin') in the RNA, while others are designed to be unstructured. The size of spacers in different CRISPR arrays is typically 32 to 38 bp (range 21 to 72 bp).[65] New spacers can appear rapidly as part of the immune response to phage infection.[66] There are usually fewer than 50 units of the repeat-spacer sequence in a CRISPR array.[65]

• 
  CRISPR-DR2: Secondary structure taken from the Rfam database. Family RF01315.
• 
  CRISPR-DR5: Secondary structure taken from the Rfam database. Family RF011318.
• 
  CRISPR-DR6: Secondary structure taken from the Rfam database. Family RF01319.
• 
  CRISPR-DR8: Secondary structure taken from the Rfam database. Family RF01321.
• 
  CRISPR-DR9: Secondary structure taken from the Rfam database. Family RF01322.
• 
  CRISPR-DR19: Secondary structure taken from the Rfam database. Family RF01332.
• 
  CRISPR-DR41: Secondary structure taken from the Rfam database. Family RF01350.
• 
  CRISPR-DR52: Secondary structure taken from the Rfam database. Family RF01365.
• 
  CRISPR-DR57: Secondary structure taken from the Rfam database. Family RF01370.
• 
  CRISPR-DR65: Secondary structure taken from the Rfam database. Family RF01378.

Small clusters of cas genes are often located next to CRISPR repeat-spacer arrays. Collectively the 93 cas genes are grouped into 35 families based on sequence similarity of the encoded proteins. 11 of the 35 families form the cas core, which includes the protein families Cas1 through Cas9. A complete CRISPR-Cas locus has at least one gene belonging to the cas core.[67]

CRISPR-Cas systems fall into two classes. Class 1 systems use a complex of multiple Cas proteins to degrade foreign nucleic acids. Class 2 systems use a single large Cas protein for the same purpose. Class 1 is divided into types I, III, and IV; class 2 is divided into types II, V, and VI.[68] The 6 system types are divided into 19 subtypes.[69] Each type and most subtypes are characterized by a "signature gene" found almost exclusively in the category. Classification is also based on the complement of cas genes that are present. Most CRISPR-Cas systems have a Cas1 protein. The phylogeny of Cas1 proteins generally agrees with the classification system.[67] Many organisms contain multiple CRISPR-Cas systems suggesting that they are compatible and may share components.[70][71] The sporadic distribution of the CRISPR/Cas subtypes suggests that the CRISPR/Cas system is subject to horizontal gene transfer during microbial evolution.

When a microbe is invaded by a bacteriophage, the first stage of the immune response is to capture phage DNA and insert it into a CRISPR locus in the form of a spacer. Cas1 and Cas2 are found in both types of CRISPR-Cas immune systems, which indicates that they are involved in spacer acquisition. Mutation studies confirmed this hypothesis, showing that removal of cas1 or cas2 stopped spacer acquisition, without affecting CRISPR immune response.[81][82][83][84][85]

Multiple Cas1 proteins have been characterised and their structures resolved.[86][87][88] Cas1 proteins have diverse amino acid sequences. However, their crystal structures are similar and all purified Cas1 proteins are metal-dependent nucleases/integrases that bind to DNA in a sequence-independent manner.[70] Representative Cas2 proteins have been characterised and possess either (single strand) ssRNA-[89] or (double strand) dsDNA-[90][91] specific endoribonuclease activity.

In the I-E system of E. coli Cas1 and Cas2 form a complex where a Cas2 dimer bridges two Cas1 dimers.[92] In this complex Cas2 performs a non-enzymatic scaffolding role,[92] binding double-stranded fragments of invading DNA, while Cas1 binds the single-stranded flanks of the DNA and catalyses their integration into CRISPR arrays.[93][94][95] New spacers are usually added at the beginning of the CRISPR next to the leader sequence creating a chronological record of viral infections.[96] In E. coli a histone like protein called integration host factor (IHF), which binds to the leader sequence, is responsible for the accuracy of this integration.[97] IHF also enhances integration efficiency in the type I-F system of Pectobacterium atrosepticum.[98] but in other systems different host factors may be required[99]

Bioinformatic analysis of regions of phage genomes that were excised as spacers (termed protospacers) revealed that they were not randomly selected but instead were found adjacent to short (3–5 bp) DNA sequences termed protospacer adjacent motifs (PAM). Analysis of CRISPR-Cas systems showed PAMs to be important for type I and type II, but not type III systems during acquisition.[27][100][101][102][103][104] In type I and type II systems, protospacers are excised at positions adjacent to a PAM sequence, with the other end of the spacer cut using a ruler mechanism, thus maintaining the regularity of the spacer size in the CRISPR array.[105][106] The conservation of the PAM sequence differs between CRISPR-Cas systems and appears to be evolutionarily linked to Cas1 and the leader sequence.[104][107]

New spacers are added to a CRISPR array in a directional manner,[25] occurring preferentially,[66][100][101][108][109] but not exclusively, adjacent[103][106] to the leader sequence. Analysis of the type I-E system from E. coli demonstrated that the first direct repeat adjacent to the leader sequence, is copied, with the newly acquired spacer inserted between the first and second direct repeats.[84][105]

The PAM sequence appears to be important during spacer insertion in type I-E systems. That sequence contains a strongly conserved final nucleotide (nt) adjacent to the first nt of the protospacer. This nt becomes the final base in the first direct repeat.[85][110][111] This suggests that the spacer acquisition machinery generates single-stranded overhangs in the second-to-last position of the direct repeat and in the PAM during spacer insertion. However, not all CRISPR-Cas systems appear to share this mechanism as PAMs in other organisms do not show the same level of conservation in the final position.[107] It is likely that in those systems, a blunt end is generated at the very end of the direct repeat and the protospacer during acquisition.

Analysis of Sulfolobus solfataricus CRISPRs revealed further complexities to the canonical model of spacer insertion, as one of its six CRISPR loci inserted new spacers randomly throughout its CRISPR array, as opposed to inserting closest to the leader sequence.[106]

Multiple CRISPRs contain many spacers to the same phage. The mechanism that causes this phenomenon was discovered in the type I-E system of E. coli. A significant enhancement in spacer acquisition was detected where spacers already target the phage, even mismatches to the protospacer. This ‘priming’ requires the Cas proteins involved in both acquisition and interference to interact with each other. Newly acquired spacers that result from the priming mechanism are always found on the same strand as the priming spacer.[85][110][111] This observation led to the hypothesis that the acquisition machinery slides along the foreign DNA after priming to find a new protospacer.[111]

CRISPR-RNA (crRNA), which later guides the Cas nuclease to the target during the interference step, must be generated from the CRISPR sequence. The crRNA is initially transcribed as part of a single long transcript encompassing much of the CRISPR array.[23] This transcript is then cleaved by Cas proteins to form crRNAs. The mechanism to produce crRNAs differs among CRISPR/Cas systems. In type I-E and type I-F systems, the proteins Cas6e and Cas6f respectively, recognise stem-loops[112][113][114] created by the pairing of identical repeats that flank the crRNA.[115] These Cas proteins cleave the longer transcript at the edge of the paired region, leaving a single crRNA along with a small remnant of the paired repeat region.

Type III systems also use Cas6, however their repeats do not produce stem-loops. Cleavage instead occurs by the longer transcript wrapping around the Cas6 to allow cleavage just upstream of the repeat sequence.[116][117][118]

Type II systems lack the Cas6 gene and instead utilize RNaseIII for cleavage. Functional type II systems encode an extra small RNA that is complementary to the repeat sequence, known as a trans-activating crRNA (tracrRNA).[35] Transcription of the tracrRNA and the primary CRISPR transcript results in base pairing and the formation of dsRNA at the repeat sequence, which is subsequently targeted by RNaseIII to produce crRNAs. Unlike the other two systems the crRNA does not contain the full spacer, which is instead truncated at one end.[76]

CrRNAs associate with Cas proteins to form ribonucleotide complexes that recognize foreign nucleic acids. CrRNAs show no preference between the coding and non-coding strands, which is indicative of an RNA-guided DNA-targeting system.[9][34][81][85][119][120][121] The type I-E complex (commonly referred to as Cascade) requires five Cas proteins bound to a single crRNA.[122][123]

During the interference stage in type I systems the PAM sequence is recognized on the crRNA-complementary strand and is required along with crRNA annealing. In type I systems correct base pairing between the crRNA and the protospacer signals a conformational change in Cascade that recruits Cas3 for DNA degradation.

Type II systems rely on a single multifunctional protein, Cas9, for the interference step.[76] Cas9 requires both the crRNA and the tracrRNA to function and cleaves DNA using its dual HNH and RuvC/RNaseH-like endonuclease domains. Basepairing between the PAM and the phage genome is required in type II systems. However, the PAM is recognized on the same strand as the crRNA (the opposite strand to type I systems).

Type III systems, like type I require six or seven Cas proteins binding to crRNAs.[124][125] The type III systems analysed from S. solfataricus and P. furiosus both target the mRNA of phages rather than phage DNA genome,[71][125] which may make these systems uniquely capable of targeting RNA-based phage genomes.[70] Type III systems were also found to target DNA in addition to RNA using a different Cas protein in the complex, Cas10.[126] The DNA cleavage was shown to be transcription dependent.[127]

The mechanism for distinguishing self from foreign DNA during interference is built into the crRNAs and is therefore likely common to all three systems. Throughout the distinctive maturation process of each major type, all crRNAs contain a spacer sequence and some portion of the repeat at one or both ends. It is the partial repeat sequence that prevents the CRISPR-Cas system from targeting the chromosome as base pairing beyond the spacer sequence signals self and prevents DNA cleavage.[128] RNA-guided CRISPR enzymes are classified as type V restriction enzymes.

Analysis of CRISPR sequences revealed coevolution of host and viral genomes.[138] Cas9 proteins are highly enriched in pathogenic and commensal bacteria. CRISPR/Cas-mediated gene regulation may contribute to the regulation of endogenous bacterial genes, particularly during interaction with eukaryotic hosts. For example, Francisella novicida uses a unique, small, CRISPR/Cas-associated RNA (scaRNA) to repress an endogenous transcript encoding a bacterial lipoprotein that is critical for F. novicida to dampen host response and promote virulence.[139]

The basic model of CRISPR evolution is newly incorporated spacers driving phages to mutate their genomes to avoid the bacterial immune response, creating diversity in both the phage and host populations. To resist a phage infection, the sequence of the CRISPR spacer must correspond perfectly to the sequence of the target phage gene. Phages can continue to infect their hosts given point mutations in the spacer.[128] Similar stringency is required in PAM or the bacterial strain remains phage sensitive.[101][128]

A study of 124 S. thermophilus strains showed that 26% of all spacers were unique and that different CRISPR loci showed different rates of spacer acquisition.[100] Some CRISPR loci evolve more rapidly than others, which allowed the strains' phylogenetic relationships to be determined. A comparative genomic analysis showed that E. coli and S. enterica evolve much more slowly than S. thermophilus. The latter's strains that diverged 250 thousand years ago still contained the same spacer complement.[140]

Metagenomic analysis of two acid-mine-drainage biofilms showed that one of the analyzed CRISPRs contained extensive deletions and spacer additions versus the other biofilm, suggesting a higher phage activity/prevalence in one community than the other.[66] In the oral cavity, a temporal study determined that 7–22% of spacers were shared over 17 months within an individual while less than 2% were shared across individuals.[109]

From the same environment a single strain was tracked using PCR primers specific to its CRISPR system. Broad-level results of spacer presence/absence showed significant diversity. However, this CRISPR added 3 spacers over 17 months,[109] suggesting that even in an environment with significant CRISPR diversity some loci evolve slowly.

CRISPRs were analysed from the metagenomes produced for the human microbiome project.[141] Although most were body-site specific, some within a body site are widely shared among individuals. One of these loci originated from streptococcal species and contained ≈15,000 spacers, 50% of which were unique. Similar to the targeted studies of the oral cavity, some showed little evolution over time.[141]

CRISPR evolution was studied in chemostats using S. thermophilus to directly examine spacer acquisition rates. In one week, S. thermophilus strains acquired up to three spacers when challenged with a single phage.[142] During the same interval the phage developed single nucleotide polymorphisms that became fixed in the population, suggesting that targeting had prevented phage replication absent these mutations.[142]

Another S. thermophilus experiment showed that phages can infect and replicate in hosts that have only one targeting spacer. Yet another showed that sensitive hosts can exist in environments with high phage titres.[143] The chemostat and observational studies suggest many nuances to CRISPR and phage (co)evolution.

CRISPR technology has been applied in the food and farming industries to engineer probiotic cultures and to immunize industrial cultures (for yogurt, for instance) versus infections. It is also being used in crops to enhance yield, drought tolerance and nutritional homes.[155]

By the end of 2014 some 1000 research papers had been published that mentioned CRISPR.[156][157] The technology had been used to functionally inactivate genes in human cell lines and cells, to study Candida albicans, to modify yeasts used to make biofuels and to genetically modify crop strains.[157] CRISPR can also be used to change mosquitos so they cannot transmit diseases such as malaria.[158] CRISPR based approaches utilizing Cas12a have recently been utilized in the successful modification of a broad number of plant species.[159]

In July 2019, CRISPR was to experimentally treat a patient with a genetic disorder. The patient was a 34-year-old woman with sickle cell disease.[160]

In March 2020, CRISPR-modified virus was injected into a patient's eye in an attempt to treat Leber congenital amaurosis.[161]

In the future, CRISPR gene editing could be used to create new species or revive extinct species from closely related ones.[162]

CRISPR-based re-evaluations of claims for gene-disease relationships have led to the discovery of potentially important anomalies.[163]

CRISPR associated nucleases have shown to be useful as a tool for molecular testing due to their ability to specifically target nucleic acid sequences in a high background of non-target sequences. In 2016, the Cas9 nuclease was used to deplete unwanted nucleotide sequences in next-generation sequencing libraries while requiring only 250 pg of initial RNA input.[164] Beginning in 2017, CRISPR associated nucleases were also used for direct diagnostic testing, down to single molecule sensitivity.[165][166] In 2019, some new diagnostic applications using CRISPR technology, an electrical CRISPR-Chip[167] for the detection of genetic mutations and a microfluidic CRISPR-Biosensor[168] for electrochemical miRNA diagnostics, were introduced.