Structural principles of CRISPR RNA processing

Abstract

The Cas6 superfamily, the Cas5d subclass, and the host RNase III endoribonucleases are responsible for generation of small RNAs (crRNA) that function in the CRISPR-Cas immunity. The three enzymes may also interact with the crRNA-associated nucleic acid interference complexes. Recent development in structural biology of Cas6 and Cas5d and their complexes with RNA substrates has lent new insights on principles of crRNA processing and the structural basis for linking crRNA processing to interference. Both Cas6 and Cas5d are characterized by the presence of the ferredoxin-like fold but each has unique domain arrangement and insertion elements. Cas6 proteins often interact strongly with stable RNA stem-loop structures but can also fold unstructured RNA into stem-loop structures for their cleavage. The extraordinarily simple fold, the wide range of substrates and kinetic properties of Cas6/Cas5d support their functional roles and make them excellent candidates for exploring molecular evolution, protein-RNA interaction, and biotechnology applications.

Introduction

Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) and the CRISPR-associated (Cas) proteins constitute small RNA-mediated defense systems of many bacteria and archaea hosts against invading foreign genetic elements (Barrangou and Marraffini, 2014; Gasiunas et al., 2014; Koonin and Makarova, 2013; Sorek et al., 2013; Terns and Terns, 2011; Wiedenheft et al., 2012). CRISPR loci are comprised of a set of cas genes and identical repeats (repeats) interspersed with invader-derived spacer sequences (spacers). The repeat-spacer array is transcribed and then processed into short CRISPR RNA (crRNA) that function as guides in destruction of invading DNA or RNA (for excellent reviews on CRISPR-Cas molecular mechanisms, see (Gasiunas et al., 2014; Reeks et al., 2013a; van der Oost et al., 2014; Wiedenheft et al., 2012). Three mechanisms for generation of crRNAs have been discovered in different CRISPR-Cas systems. The Cas6 superfamily of proteins is responsible for processing crRNA in two of the three types of CRISPR-Cas systems (Brouns et al., 2008; Carte et al., 2008; Haurwitz et al., 2010; Lawrence and White, 2011; Richter et al., 2012). In these CRISPR-Cas systems, cleavage within each repeat by Cas6 releases the spacers bearing portions of the repeat on its 5’ and 3’ ends. The 5’ flanking repeat of the crRNAs is the last 8 nucleotides (nts) of the preceding repeat and the 3’ flanking repeat is the remaining sequence of the downstream repeat (Figure 1). In some systems, the 3’ flanking repeat sequences are further processed by uncharacterized exonucleases (Carte et al., 2008; Zhang et al., 2012b). Each member of the Cas6 family of endoribonucleases recognizes a unique RNA sequence and collectively, Cas6 proteins process a wide range of substrates of different sequences and secondary structures. Cas5d is the second distinct class of endoribonucleases responsible for processing crRNA in a CRISPR-Cas system that lacks Cas6. Similar to Cas6, Cas5d also recognizes specific features of and cleaves within the repeat, resulting in crRNAs containing spacer sequences flanked by repeat sequences (Garside et al., 2012; Koo et al., 2013; Nam et al., 2012). Finally, the CRISPR-Cas system that lacks both Cas6 and Cas5d employs the host RNase III enzyme in processing crRNA. Unlike Cas6 and Cas5d, bacterial RNase III recognizes double stranded RNA (dsRNA) unrelated to but including that formed by the repeat and a separate transcript called trans-activating crRNA (tracrRNA). RNase III cleaves both strands of the dsRNA with a two base pair separation, resulting in a cleavage intermediate further processed by the Cas9 class protein (Deltcheva et al., 2011). RNase III is an evolutionarily conserved endoribonuclease involved in many biological processes and its structure has been studied intensively. Since excellent reviews on structure and function of RNase III are available (Court et al., 2013; Nicholson, 2014), this review focuses only on Cas6 and Cas5d. Cas6 and Cas5d define two new classes of endoribonucleases with previously unknown specificity. Their structural and biochemical mechanisms offer novel mechanistic insights on RNA interaction and cleavage by proteins. In at least one case, the knowledge of the Cas6 endoribonuclease has been applied to the development of biotechnology tools (Lee et al., 2013; Nissim et al., 2014).

Figure 1.

Pathways and enzymes of CRISPR RNA processing. A. Left, Cas6/Cas5d processes the CRISPR repeat (R)-spacer (S) array, leading to CRISPR RNA containing part of the repeat and a spacer sequence. Some CRISPR-Cas systems contain exonuclease activities that further process the 3’ end of the cleavage product of Cas6. Right, In CRISPR-Cas systems that lack Cas6 and Cas5d, the host RNase III processes CRISPR RNA in a trans-activating RNA (tracrRNA)-dependent manner. Cas9 is believes to provide a platform for the RNase III-catalyzed reaction and further processes the intermediate. B. Protein fold (left) and topology (right) of a representative Cas6 (upper panel) (PDB code: 3I4H) and Cas5d (lower panel) (PDB code: 4F3M). The core secondary elements are labeled and the insertions are represented by dashed lines with parentheses. Elements colored in red are involved in recognition and cleavage of RNA and the recognized sites on RNA are labeled. C. Schematic RNA secondary structure bound with Cas6/Cas5d endoribonucleases. The cleavage site is usually located at the base of the stem and the size of the stem or the loop varies substantially. The 3’ cleavage product becomes the 5’ tag of all mature CRISPR RNA.

The strong interest in the biochemical mechanism and biotechnology applications of CRISPR-Cas systems has propelled structure and function studies of the Cas6 and Cas5d proteins. Within the last several years, the CRISPR-Cas research community has acquired twenty-eight crystal structures of nine Cas6 proteins and their complexes with RNA substrates and four crystal structures of isolated Cas5d proteins (Table 1). These studies have significantly expanded our knowledge of RNA cleavage enzymes, in particular, on how they specifically recognize RNA substrates and facilitate phosphodiester bond breakage.

Table 1.

List of currently known Cas6/Cas5d crystal structures

PDB code	Name	RNA substrates (protein gene name)	Organism (reference)
3I4H	Cas6	protein only; 2.3 Å (PF1131)	Pyrococcus furiosus [1]
3PKM	Cas6	32-nts but 1st 12-nts observed (PF1131)	Pyrococcus furiosus [2]
3QJJ	Cas6nc	12-nts substrate; 2.8 Å (PH0350)	Pyrococcus horikoshii [3]
3QJL	Cas6nc	12-nts variant; 2.7Å (PH0350)	Pyrococcus horikoshii [3]
3QJP	Cas6nc	12-nts variant; 2.7Å (PH0350)	Pyrococcus horikoshii [3]
3UFC	Cas6nc	protein only; 2.0 Å (PF0393)	Pyrococcus furiosus [4]
3ZFV	Cas6	protein only; 2.8 Å (SSO1437)	Sofulobus solfataricus [5]
4ILR	Cas6	16-nts, 3.1 Å (SSO2004)	Sofulobus solfataricus [6]
4ILL	Cas6	20-nts, 2.5 Å (SSO2004)	Sofulobus solfataricus [6]
4ILM	Cas6	protein only; 3.1 Å (SSO2004)	Sofulobus solfataricus [6]
2XLI	Cas6f (Csy4)	16-nts; G20 2'-deoxy; 2.3 Å	Pseudomonas aeruginosa [7]
2XLJ	Cas6f (Csy4)	16-nts, G20 2'-deoxy; 2.6 Å	Pseudomonas aeruginosa [7]
2XLK	Cas6f (Csy4)	16-nts, G20 2'-deoxy; 1.8 Å	Pseudomonas aeruginosa [7]
4AL5	Cas6f (Csy4)	16-nt product; 2.0 Å	Pseudomonas aeruginosa [8]
4AL6	Cas6f (Csy4)	S148A mutant, 14-nts; 2.6 Å	Pseudomonas aeruginosa [8]
4AL7	Cas6f (Csy4)	15-nts minimal substrate; 2.3 Å	Pseudomonas aeruginosa [8]
4C97	Cas6A	H37A mutant only, 1.7 Å (TTHA0078)	Thermus thermophilus HB8 [9]
4C98	Cas6B	protein only; 2.0 Å (TTHB231)	Thermus thermophilus HB8 [9]
4C8Z	Cas6A	Cas6A:Product, 2:1; 2.5 Å; (TTHA0078)	Thermus thermophilus HB8 [9]
4C8Y	Cas6A	substrate mimic; 1.8 Å (TTHA0078)	Thermus thermophilus HB8 [9]
4C9D	Cas6B	13-nt product; 3.0 Å (TTHB231)	Thermus thermophilus HB8 [9]
2Y8W	Cas6e (Cse3)	20 nts; deoxy G21; 1.8 Å (TTHB192)	Thermus thermophilus HB8 [10]
2Y8Y	Cas6e (Cse3)	19 nts deoxy G21; 1.4 Å (TTHB192)	Thermus thermophilus HB8 [10]
2Y9H	Cas6e (Cse3)	19 nts deoxy G21; 2.5 Å (TTHB192)	Thermus thermophilus HB8 [10]
1WJ9	Cas6e (Cse3)	protein only; 1.9 Å (TTHB192)	Thermus thermophilus HB8 [11]
3QRP	Cas6e (Cse3)	9 nts+7nts; 2.3 Å (TTHB192)	Thermus thermophilus HB8 [12]
3QRQ	Cas6e (Cse3)	19 nts self-dimer; 3.2 Å (TTHB192)	Thermus thermophilus HB8 [12]
3QRR	Cas6e (Cse3)	18 nts self-dimer 3.1 Å (TTHB192)	Thermus thermophilus HB8 [12]
4F3M	Cas5d	protein only; 1.7 Å (BH0337)	Bacillus halodurans C-125 [13]
3VZI	Cas5d	protein only; 2.7 Å (PXO_02700)	Xanthomonas oryzae pv. oryzae
3VZH	Cas5d	protein only; 1.7 Å (Spy_1566)	Streptococcus pyogenes[14]
3KG4	Cas5d	protein only; 2.0 Å (MS0988)	Mannhrimis duvviniviptofuvrnd[15]

¹Carte, J., et al., Cas6 is an endoribonuclease that generates guide RNAs for invader defense in prokaryotes. Genes Dev, 2008. 22(24): p. 3489-96.

²Wang, R., et al., Interaction of the Cas6 Riboendonuclease with CRISPR RNAs: Recognition and Cleavage. Structure, 2011. 19(2): p. 257-64.

³Wang, R., et al., The impact of CRISPR repeat sequence on structures of a Cas6 protein- RNA complex. Protein Sci, 2012. 21(3): p. 405-17.

⁴Park, H.M., et al., Crystal structure of a Cas6 paralogous protein from Pyrococcus furiosus. Proteins, 2012. 80(7): p. 1895-900.

⁵Reeks, J., et al., Structure of a dimeric crenarchaeal Cas6 enzyme with an atypical active site for CRISPR RNA processing. Biochem J, 2013. 452(2): p. 223-30.

⁶Shao, Y. and H. Li, Recognition and cleavage of a nonstructured CRISPR RNA by its processing endoribonuclease Cas6. Structure, 2013. 21(3): p. 385-93.

⁷Haurwitz, R.E., et al., Sequence- and structure-specific RNA processing by a CRISPR endonuclease. Science, 2010. 329(5997): p. 1355-8.

⁸Haurwitz, R.E., S.H. Sternberg, and J.A. Doudna, Csy4 relies on an unusual catalytic dyad to position and cleave CRISPR RNA. EMBO J, 2012. 31(12): p. 2824-32.

⁹Niewoehner, O., M. Jinek, and J.A. Doudna, Evolution of CRISPR RNA recognition and processing by Cas6 endonucleases. Nucleic Acids Res, 2014. 42(2): p. 1341-53.

¹⁰Sashital, D.G., M. Jinek, and J.A. Doudna, An RNA-induced conformational change required for CRISPR RNA cleavage by the endoribonuclease Cse3. Nat Struct Mol Biol, 2011. 18(6): p. 680-7.

¹¹Ebihara, A., et al., Crystal structure of hypothetical protein TTHB192 from Thermus thermophilus HB8 reveals a new protein family with an RNA recognition motif-like domain. Protein Sci, 2006. 15(6): p. 1494-9.

¹²Gesner, E.M., et al., Recognition and maturation of effector RNAs in a CRISPR interference pathway. Nat Struct Mol Biol, 2011. 18(6): p. 688-92.

¹³Nam, K.H., et al., Cas5d protein processes pre-crRNA and assembles into a cascade-like interference complex in subtype I-C/Dvulg CRISPR-Cas system. Structure, 2012. 20(9): p. 1574-84.

¹⁴Koo, Y., et al., Conservation and variability in the structure and function of the Cas5d endoribonuclease in the CRISPR-mediated microbial immune system. J Mol Biol, 2013. 425(20): p. 3799-810.

¹⁵Garside, E.L., et al., Cas5d processes pre-crRNA and is a member of a larger family of CRISPR RNA endonucleases. RNA, 2012. 18(11): p. 2020-8.

Biologically, Cas6 and Cas5d link the crRNA processing step to the downstream interference step because the assembly of interference complexes depends on the cleavage products by Cas6 and Cas5d. Several Cas6 and Cas5d do in fact form part of the interference complexes (Brouns et al., 2008; Haurwitz et al., 2012; Huo et al., 2014; Wiedenheft et al., 2011b). Biochemically, when acting in isolation, Cas6 proteins have generally ultrahigh affinities for their RNA substrate (50 pM - 10 nM range) but very slow turnover rates (0.5-3 min⁻¹). Isolated Cas5d have weak affinities for its products but can readily assemble into ribonucleoportein particles when co-expressed with the DNA interference complex (Nam et al., 2012). Structurally, Cas6 and Cas5d proteins are made up of a simple and ancient fold. Due to availability of the substrate-bound structures for the Cas6 class only, the review will analyze primarily the available Cas6 structures and discuss the structural principles responsible for its substrate specificity, binding affinity, catalytic properties and, in one case where it is known, the interaction with the components of the interference complexes.

Nomenclatures

The cas genes including those encoding Cas6 and Cas5d are extraordinarily diverse and evolve rapidly. Based on an integrated analysis of the phylogeny of the most common cas genes, CRISPR repeat features, and CRISPR locus architecture, CRISPR-Cas systems are categorized into three major types (I, II and III), each of which is further divided into subtypes that are designated alphabetically (i.e., I-A, III-B, etc.) (Makarova et al., 2011; Makarova et al., 2013). Types I (except for I-C) and III CRISPR-Cas systems employ the Cas6-mediated processing while the type II CRISPR-Cas systems use the tracrRNA-guided processing mechanism with endogenous RNase III (Deltcheva et al., 2011). Type I-C is the only known subtype associated with the Cas5d-mediated processing (Garside et al., 2012; Koo et al., 2013; Nam et al., 2012). In CRISPR-Cas systems where a single cas6 gene is detected, the endoribonuclease is named Cas6, but if a CRISPR-Cas system contains cas6-like genes distinctly different from the characteristic cas6 gene, the endoribonuclease is designated by Cas6x where “x” follows the letter of its subtype (i.e., Cas6e for subtype I-E and Cas6f for subtype I-F). In organisms where multiple closely related cas6 genes are found, the Cas6 proteins are distinguished by appending alphabets to Cas6 (Cas6A, Cas6B, etc.). Lastly, in organisms where multiple cas6 genes exist, another group of cas6-like genes are often found that encode inactive Cas6 proteins (Wang and Li, 2012; Wang et al., 2012). These are designated as Cas6nc (for non-catalytic). If more than one non-catalytic Cas6 protein exists in the same organisms, they can be further distinguished by appended alphabet (Cas6ncA, Cas6ncB, etc.). Although this nomenclature system is not strictly rigorous (for instance, the degeneracy of the same subtypes is not distinguished), it does satisfactorily catalog currently known Cas6 proteins. Similar nomenclature applies to the Cas5 superfamily proteins and Cas5d forms a distinct class of the Cas5 superfamily. Other Cas5 classes form essential components of the DNA or RNA surveillance and effector complexes and not known to be processing endoribonucleases (Makarova et al., 2013).

Prior to the revised nomenclature of the CRISPR-Cas systems in 2011 (Makarova et al., 2011), some Cas6 proteins were named after the “Csx” (Cas subtype x) type of nomenclature (Haft et al., 2005; Makarova et al., 2006). These names, along with other used names, are listed with the systematic names in Table 1. This review will follow the systematic nomenclature given in (Makarova et al., 2011) and Table 1 with a prefix reflecting the organism to which the protein belongs.

The fold

Regardless the subtypes or catalytic activity, the Cas6 proteins known so far all share the characteristic ferredoxin-like fold and a glycine-rich region (G-loop). The ferredoxin-like fold is the most abundant protein fold and has appeared in many proteins of diverse functions (Hegyi et al., 2002; Zhang and DeLisi, 2001). The proteins containing the ferredoxin-like fold that are most closely related to Cas6 are the RNA Recognition Motif (RRM) proteins, many of which bind single-stranded RNA (Maris et al., 2005). Thus the ferredoxin-like fold of Cas6 is also often referred to as RRM.

In all but one of the Cas6 proteins (Cas6f) studied so far, two ferredoxin-like fold/RRM domains lie slanted side-by-side in a V-shape, forming two exterior surfaces comprised of β-sheet and α-helices, respectively (Figure 1B). In Cas6f, the C-terminal ferredoxin fold/RRM has diverged from a recognizable ferredoxin-like fold (see below). In general, the N-terminal ferredoxin-like fold/RRM domain follows the canonical βαββαβ topology more closely than the C-terminal ferredoxin-like fold/RRM. However, both ferredoxin-like fold/RRM domains are disrupted by insertion elements. In order to facilitate discussion of common structural features of multiple Cas6 proteins, the core secondary elements of the classic ferredoxin-like fold are labeled numerically from β1 to β8 and from α1 to α4. The N- and C-terminal ferredoxin-like/RRM domains have thus the topology of β1α1β2β3α2β4 and β5α3β6β7α4β8, respectively (Figure 1B). Inserted secondary elements are then labeled alphabetically (i.e., αA, βA) throughout each protein. The characteristic G-loop connects the last α-helix and β-strand (α4 and β8) and is situated at the junction between the two ferredoxin-like fold/RRM domains. The G-loop has been found to be essential to both Cas6 folding (Wang et al., 2011) and RNA binding (Sashital et al., 2011; Wang et al., 2012).

Cas5d contains a single ferredoxin-like fold but has two Cas5d-specific additions. The first is a long β-hairpin between β2 and β3 that was shown biochemically to play important role in binding single stranded RNA (thumb, Figure 2C)(Garside et al., 2012; Koo et al., 2013; Nam et al., 2012). The second is a β-strand-rich extension to the C-terminus (EXT, Figure 2C). Cas5d does not contain a glycine-rich loop.

Figure 2.

RNA recognition by six Cas6/Cas5d endoribonucleases. A. Crystal structures of the known Cas6-RNA complexes are shown in the same orientation in which the N-terminal ferredoxin-like fold of each Cas6 is superimposed. The bound RNA is colored in red and Cas6 is colored in yellow (helix) and teal (other secondary structure elements). The scissile phosphate is indicated by an orange sphere. The RNA secondary structure for each complex as observed in the co-crystal structure is schematically illustrated on the side with a small arrow indicating the site of cleavage. Structural elements important for RNA recognition are indicated. B. Upper panel, illustration of the recognition of the cleavage site by Cas6 using the TtCas6A-RNA complex (PDB code: 4C8Z) as an example. The nucleotide downstream of the scissile phosphate is splayed and stabilized by the α1 helix. The in-line path that would be formed by the three reactive atoms during bond breakage is indicated by a dashed line. The 2’-OH group is missing in this structure due to the use of 2’-deoxy modified RNA substrate analog. Lower panel, summary of the key secondary structure elements of Cas6 used in recognition of the stem-loop structure. C. Aligned Cas5d (PDB code: 4F3M) and E. coli Cas5 (4U7U chain K) structures with the N- terminal (N ferredoxin fold alignment) or the C-terminal (C ferredoxin fold alignment) ferredoxin-like fold of Cas6e. “Thumb” refers to the β-hairpin insertion between β2 and β3 shown in Figure 1.

RNA recognition

The repeats of different CRISPR loci share little sequence similarity but often a palindromic pattern (Jansen et al., 2002; Kunin et al., 2007). This suggests that the RNA substrates recognized by Cas6 form stem-loop structures. However, in many organisms including both bacteria and archaea, the palindromic pattern is not evident at a sequence level (Kunin et al., 2007), which raises the question of whether the repeat RNA universally possess a stem-loop structure. The co-crystal structures of Cas6 proteins bound with their respective substrate RNA that have been determined so far reveal indeed a conservation of the stem-loop structure, even in RNA devoid of a predicted palindromic pattern; however, both the stem and the loop vary largely in size and sequences (Figure 2). For those whose structures are known, the longest stem is that for TtCas6e (6 base pairs) (Gesner et al., 2011; Sashital et al., 2011) and the shortest is that for SsCas6 (3 base pairs) (Shao and Li, 2013). Although there is not a strict correlation, highly stable stem loop substrates seem to have high binding affinities. The importance of the loop nucleotides to binding varies in different Cas6 complexes. Some substrates require the loop nucleotides to bind Cas6 (Haurwitz et al., 2010; Niewoehner et al., 2014; Sternberg et al., 2012), but some are less dependent on them (Gesner et al., 2011; Niewoehner et al., 2014; Sashital et al., 2011; Shao and Li, 2013; Sternberg et al., 2012).

Cas6 proteins employ the same set of the core and insertion elements of the two ferredoxin-like folds to interact with the repeat RNA. These include α1, α3, β6-β7 hairpin, the G-loop, and the loop following β5 (β5_loop) (Figure 2). Variations in the size and composition of the insertion elements (β6-β7 hairpin loop and β5_loop) enable Cas6 to confer different specificity. For instance, the β5_loop, which often forms a short β-hairpin in other Cas6, is made of two unique α-helices in SsCas6 that help to anchor a long stem loop (Figure 2) (Shao and Li, 2013). The G-loop, the β5_loop, and α3 are responsible for recognizing the shape and sequences of the stem through the side chains of polar and positively charged residues (Gesner et al., 2011; Haurwitz et al., 2010; Sashital et al., 2011; Shao and Li, 2013; Sternberg et al., 2012; Wang et al., 2011). The β6-β7 hairpin and the α1 helix, which are less variable than the stem and loop interacting elements, interact with the region around the RNA cleavage site. The β6-β7 hairpin inserts between the two strands of either side of the stem and the α1 helix stabilizes the downstream nucleotides. This region forms the most characteristic conformation of all Cas6-bound RNA in which the nucleotides downstream of the scissile phosphate splay out of the helical stacking (cleavage fork), leading to the geometry poised for line-in attack. Both the β6-β7 hairpin and the α1 helix also supply residues involved in catalysis (discussed below) and are thus critical to the function of Cas6.

While a strong palindromic pattern predicts stable RNA stem-loops, a weak or the absence of the palindromic pattern suggests unfolded RNA in solution and thus a role of the Cas6 proteins in stabilizing the bound stem-loops. The weak stem-loop RNA substrates reduce the effective binding energy due to the energy spent by Cas6 on converting the unfolded to the folded conformation of the RNA. Indeed, the strong stem-loop RNA substrates such as those for PaCas6f, TtCas6e, and TtCas6A/B have low binding dissociation constants in a range of 50 pM – 5 nM (Niewoehner et al., 2014; Sashital et al., 2011; Sternberg et al., 2012). These Cas6 proteins are single turnover enzymes and remain bound to the processed crRNA after cleavage. In contrast, PfCas6 and SsCas6 show weaker binding affinities and are known to be multiple turnover enzymes (Shao and Li, 2013; Sokolowski et al., 2014; Wang et al., 2011). Still, the multiple turnover enzymes bind RNA with reasonable affinities and thus must compensate elsewhere for the loss of energy spent on folding the RNA. These can be through extended protein-RNA binding surfaces, favorable electrostatic potential, cooperative binding through oligomerization, or favorable changes in dynamics in both Cas6 and RNA. In the SsCas6-RNA complex where the RNA forms a short 3 base pair stem (Shao and Li, 2013), the RNA loop makes up for the contact surface area by interacting with the significantly expanded β5_loop. In addition, structures of several archaeal Cas6 and ThCas6A/B show that they form either stable or RNA-induced dimers, suggesting possible cooperative regulation through oligomerization. Structural evidence supporting this hypothesis include the observed cross-subunit (Niewoehner et al., 2014; Wang et al., 2012) and sub- stoichiometry binding of RNA (Niewoehner et al., 2014) by some Cas6 proteins.

Most Cas6 proteins require only the stem-loop plus 1-2 flanking nucleotides for binding. Some, however, require the single-stranded region upstream of the stem to enhance specificity. For both PaCas6f and TtCas6e whose RNA are stable stem loops, nucleotides immediately upstream the stem are not important for binding and cleavage (Gesner et al., 2011; Haurwitz et al., 2010; Sashital et al., 2011; Sternberg et al., 2012) while for PfCas6 whose substrates show no palindromic pattern, the first 8 nucleotides are recognized by the enzyme and are crucial to RNA binding and cleavage (Carte et al., 2008; Wang et al., 2011). The 8 nucleotides 5’ of the stem of the repeat RNA substrate (4 base pair stem) for TtCas6A are also shown to be important for RNA binding (Niewoehner et al., 2014).

There is no crystal structure of Cas5d bound with its subunit. However, some insights may be gained indirectly by comparing Cas5d structure to a crRNA-bound Cas5 included in the recently available structure E. coli (Ec) DNA surveillance complex Cascade (Jackson et al., 2014; Mulepati et al., 2014; Zhao et al., 2014). EcCas5 uses its α1 helix and the β-hairpin insertion (thumb) to interact with the 5’ tag of the crRNA (Figure 2C) (Jackson et al., 2014; Mulepati et al., 2014; Zhao et al., 2014). Alignment of EcCas5 with the C-terminal ferredoxin-like fold of a Cas6 (SsCas6 for example) shows that its α1 helix and the β-hairpin thumb are equivalent to α3 and β6-β7 hairpin of Cas6, respectively, that are involved in RNA binding by Cas6. However, the RNA targets of EcCas5 and Cas5d (assuming it also utilizes α1 helix and the β-hairpin insertion) are different from each other and from those of Cas6. Thus, if confirmed, the involvement of equivalent protein elements in recognition of different RNA would demonstrate the extraordinary adaptability of the ferredoxin-like fold to binding RNA. Interestingly, the cleavage site of Cas5d was mapped to the α2 helix (Koo et al., 2013; Nam et al., 2012) whose equivalents in Cas6 (α2 and α4) are not engaged in either RNA binding or catalysis. Thus, Cas5d may have a unique mechanism for interacting with its RNA cleavage site.

In spite of the shared protein fold, Cas6 and Cas5d proteins do not bind RNA in the same way as the well-studied RRM proteins (Maris et al., 2005). RRMs use their β-sheets to interact with single stranded RNA while Cas6 and Cas5d proteins use their α-helices and loops to recognize the stem-loop and the downstream nucleotides. Thus Cas6 and Cas5 proteins constitute new families of RNA binding proteins evolved from the ancient ferredoxin-like/RRM motif.

The active site and catalytic mechanism

The Cas6 family of endoribonucleases is believed to follow the general acid-base type of cleavage mechanism based on the 2’,3’ cyclic phosphate (or 3’ phosphate) and 5’-OH termini in the cleaved products (Carte et al., 2008; Gesner et al., 2011; Haurwitz et al., 2010; Niewoehner et al., 2014; Richter et al., 2012; Sashital et al., 2011; Wiedenheft et al., 2011b; Zhang et al., 2012a). In classic general acid-base enzymes such as RNase A and T1, a general base extracts the proton from the 2’-OH group to facilitate the nucleophilic attack, and a general acid donates the proton to the leaving 5’-oxygen. A positively charged residue stabilizes the pentavalent phosphophate intermediate. In addition, these endoribonucleases align three involved catalytic atoms in-line, an energetically unfavorable conformation, in order for the S_N2 reaction to take place. In RNase A, two histidine residues, His12 and His119 acting as the general base and acid, respectively, and Lys41, stabilizing the negative transition state structure, form the catalytic triad that contributes to the overall rate enhancement of up to 10¹¹ (Cuchillo et al., 2011). Unlike RNase A, the Cas6 endoribonucleases bind to their substrates with strong affinities but have a slow cleavage turnover rate. Cas6 has been proposed to be more a maturase than RNase (Reeks et al., 2013b). Co-crystal structures of five Cas6 endoribonucleases bound with substrate analogs and reaction products indeed reveal active sites lacking the complete catalytic triad, supporting a strong structural role played by Cas6 in catalysis.

RNA substrate analogs and catalytic deficient mutants of Cas6 were used to obtain structures of enzyme-substrate complexes. Modification of the nucleophile 2’-OH group removes or modifies the 2’ nucleophile with minimal impact on RNA binding. Currently known substrate-mimic complexes used RNA containing 2’-deoxy modified nucleotide upstream of the scissile phosphate (Gesner et al., 2011; Haurwitz et al., 2010; Sashital et al., 2011; Shao and Li, 2013). To capture product- mimic complex structures, either the 5’ portion of the RNA product (Gesner et al., 2011; Shao and Li, 2013) or wild-type substrates (Haurwitz et al., 2012; Niewoehner et al., 2014) were used. In the case of the wild-type substrates, either 2’, 3’ cyclic or 3’ phosphate was observed (Haurwitz et al., 2012; Niewoehner et al., 2014). These complexes provide detailed views of the active site for five Cas6 proteins (Figure 3).

Figure 3.

Structures of the active sites of five Cas6 proteins (A-E) and ribonuclease A (RNase A) (F). The nucleotide upstream of the scissile phosphate of each RNA substrate (red) is superimposed for the five complexes and to that bound to RNase A. Pink arcs indicate the relative locations of the general base and acid as found in RNase A. PDB code for each complex is indicated.

Cas6 appears to rely only on either the general acid or base but not both. In PaCas6f bound with either 2’-deoxy modified RNA substrate analog or cleavage product (Haurwitz et al., 2010; Haurwitz et al., 2012), His29 occupies the similar position as RNase A’s His119 with respect to the scissile phosphate but biochemical data support its role as the general base (Haurwitz et al., 2012). Ser148 is positioned similarly as His12 of RNase A and its mutation to alanine reduced activity by ~8000 fold (Haurwitz et al., 2012). However, further structural studies show that Ser148 plays an important role in positioning the nucleophile 2’-OH group to facilitate the C2’-endo sugar pucker conformation, suggesting that it does not exclusively act as the general acid. In all three TtCas6 proteins bound with their respective cleavage products, a histidine residue is placed at the equivalent position of His119 of RNase A (His26 in TtCas6e, His37 in TtCas6A, His42TtCas6B) (Figure 3). Mutation of each of these histidine residues has detrimental effects on catalysis except for His42 of TtCas6B that only has a slight reduction in activity (Niewoehner et al., 2014). However, TtCas6B has another histidine residue (His23) and a tyrosine residue (Tyr256) at the equivalent position of RNase’s His12 that may compensate for the defect of His42 mutation. More surprisingly, SsCas6A does not have any histidine residue in its active site where Lys28 occupies the similar location as His119 of RNase A and Arg232 occupies the similar location of His12 of RNase A (Figure 3). Mutation of Lys28 in SsCas6 resulted in 100-200 fold reduction in the enzyme’s single turnover rate (Shao and Li, 2013; Sokolowski et al., 2014), supporting its role as the general acid. Arg232 may play the same role as Ser148 of PaCas6f in positioning the 2’-OH group or stabilizing the transition state and its mutation also reduced the wild-type cleavage activity by ~150-fold (Shao and Li, 2013; Sokolowski et al., 2014).

Cas6 endoribonucleases engage the cleavage site closely to shape the scissile phosphate bond conformation. The β6-β7 hairpin provides a backdrop of the helical stem while α1 helix stabilizes the splayed nucleotides downstream of the cleavage site. In the three TtCas6 enzymes and SsCas6, the nucleotide immediate downstream of the scissile phosphate does not have a complementary base but in TtCas6e this nucleotide, A23, would normally pair with U5, and is otherwise unpaired when bound with the enzyme (Gesner et al., 2011; Sashital et al., 2011), suggesting that Cas6 can actively unwind this base pair. The three atoms involved in the S_N2 bond breakage reaction can be easily aligned if the 2’-OH upstream of the scissile phosphate further assumes the C2’-endo position. Formation of the in-line conformation is necessary and provides a moderate rate enhancement to phosphodiester bond breakage (Min et al., 2007; Soukup and Breaker, 1999). However, the absence of the functional group acting as either the general base or acid in most Cas6 endoribonucleases may explain their sluggish turnover rate.

Outlook

The discovery of the Cas6 and Cas5d families of endoribonucleases has significantly expanded our knowledge on RNA recognition and cleavage. We have witnessed an evolutionary transformation of an ancient protein fold, which gives rise to new families of RNA binding and cleaving proteins. The range of molecular strategies used by Cas6 and Cas5d endoribonucleases to impart RNA specificity and cleavage are impressive.

In addition to the Cas6 and Cas5d family of endoribonucleases, several other Cas proteins that function in the CRISPR-Cas pathway also contain the ferredoxin-like fold. Highly related to Cas6, noncatalytic homologs of Cas6 endoribonucleases are found in many CRISPR-Cas systems. These noncatalytic Cas6 homologs have similar RNA binding specificity and affinity as Cas6 but do not possess the catalytic activity (Wang and Li, 2012). Are they failed evolutionary products en-route to generation of active Cas6 proteins or are they regulators for Cas6 by competing for the same substrates? Both Cas5 and Cas7 superfamilies contain the ferredoxin-like fold. While Cas5d has been shown functionally similar to Cas6, other subtypes of Cas5 have no roles in crRNA processing. Rather, they play a key role in assembly of surveillance or effector complexes. The recent crystal structures of a type I-E DNA surveillance complex Cascade showed elegantly how Cas5 binds to the 5’ tag of the crRNA specifically (Jackson et al., 2014; Mulepati et al., 2014; Zhao et al., 2014). Paradoxically, while Cas5d recognizes the 5’ portion of the repeat for crRNA processing (Jackson et al., 2014; Mulepati et al., 2014; Zhao et al., 2014), the Cas5 associated with Cascade recognizes the 3’ portion of the repeat (5’ tag) for assembly. It is possible that Cas5 adapts to these specific roles by developing new protein elements without disruption of its ferredoxin-like fold. The Cas7 superfamily proteins contain a single ferredoxin-like fold and are essential components of the CRISPR surveillance/effector complexes (Makarova et al., 2011; van der Oost et al., 2014). The Cas7 proteins make up the central repeats of the type I and type III surveillance/effector complexes where they engage the crRNA and assist in target DNA positioning (van der Oost et al., 2014; Wiedenheft et al., 2011a). In the type III-B complex, the Cas7 subunits also bind crRNA and may likely be involved in cleaving target RNA (Spilman et al., 2013; Staals et al., 2013). The crystal structure of the E. coli type I-E Cascade complex shows how Cas7 interacts with both crRNA and ssDNA (Jackson et al., 2014; Mulepati et al., 2014; Zhao et al., 2014). Cas7 uses a method unrelated to Cas6 but related to Cas5 to secure extended crRNA through its filamentous assembly (Jackson et al., 2014; Mulepati et al., 2014; Zhao et al., 2014) and to facilitate a segmented crRNA:ssDNA hybridization (Mulepati et al., 2014). A similar method to that observed in Cascade is also believed to underlie its activity as the cleavage center of the type III RNA interference enzyme (Hale et al., in press; Ramia et al., in revision). Lastly, the same crystal structures of the E. coli cascade complex also reveals that the structure of Cas6e as part of the Cascade head did not undergo significant structural change, suggesting its primary interaction with Cascade is with the stem-loop structure of the crRNA (Jackson et al., 2014; Mulepati et al., 2014; Zhao et al., 2014).

There are yet more Cas6 and Cas5 proteins whose gene sequences are now known but their biochemical and structural data remain lacking. Given the diversity in RNA specificity and cleavage mechanism revealed in only five Cas6 endoribonucleases, more surprises are bound to emerge.