Metal Dependence and Functional Diversity of Type I Cas3 Nucleases

Abstract

Type I CRISPR-Cas systems provide prokaryotes with protection from parasitic genetic elements by cleaving foreign DNA. In addition, they impact bacterial physiology by regulating pathogenicity and virulence, making them key players in adaptability and evolution. The signature nuclease Cas3 is a phosphodiesterase belonging to the HD-domain metalloprotein superfamily. By directing specific metal incorporation, we map a promiscuous metal ion cofactor profile for Cas3 from Thermobifida fusca (Tf). Tf Cas3 affords significant ssDNA cleavage with four homo-dimetal centers (Fe²⁺, Co²⁺, Mn²⁺, and Ni²⁺), while the diferrous form is the most active and likely biologically relevant in vivo. Electron paramagnetic resonance (EPR) spectroscopy and Mössbauer spectroscopy show that the diiron cofactor can access three redox forms, while the diferrous form can be readily obtained with mild reductants. We further employ EPR and Mössbauer on Fe-enriched proteins to establish that Cas3″ enzymes harbor a dinuclear cofactor, which was not previously confirmed. We demonstrate that the ancillary His ligand is critical for efficient ssDNA cleavage but not for diiron assembly or small molecule hydrolysis. We further explore the ability of Cas3 to hydrolyze cyclic mononucleotides and show that Tf Cas3 hydrolyzes 2′3′-cAMP with catalytic efficiency comparable to that of the conserved virulence factor A (CvfA), an HD-domain protein hydrolyzing 2′3′-cylic phosphodiester bonds at RNA 3′-termini. Because this CvfA activity is linked to virulence regulation, Cas3 may also utilize 2′3′-cAMP hydrolysis as a possible molecular route to control virulence.

Graphical Abstract

INTRODUCTION

Clustered regularly interspaced short palindromic repeats (CRISPR)-associated systems (Cas) are major players in prokaryotic adaptive immunity and RNA-based defense.^1–6 They are classified into six major types (I–VI), among which, Types I–III are the most widespread and best characterized.^7–9 Type I is the most commonly occurring system in bacteria and archaea and is the most versatile in terms of selectively introducing large deletions.^10–14 All CRISPR-Cas systems rely on three essential steps: (i) adaptation, (ii) pre-crRNA expression and processing, and (iii) degradation of foreign DNA (interference).^9,15 Type I CRISPR-Cas systems utilize a multicomponent system to perform the first two steps and recruit a single nuclease, Cas3, for interference.^15–19

The signature nuclease Cas3 is a phosphodiesterase (PDE), member of the HD-domain metalloprotein superfamily, representatives of which utilize a tandem histidine-aspartate (HD) motif to coordinate metal ions.^16,20 Cas3-associated genes encode for proteins with a multitude of architectures (Scheme S1 in SI Appendix);⁷ these can contain simply the HD nuclease domain (Cas3″),²¹ or a canonical (Cas3)^{7,13,18,22–26} or noncanonical (Cas3′)²⁷ HD nuclease domain linked to auxiliary protein domains. Although it is generally accepted that the HD motif is involved in the coordination of two metal ions, the nuclearity of Cas3″ or Cas3′ metal sites has not been established.^21,27

Cas3 cleaves ssDNA/ssRNA in a metal-dependent manner and exhibits in vitro a promiscuous metal profile with Mg²⁺, Mn²⁺, Ni²⁺, and Co²⁺ being the most common activators.^{11,17,21–23,25–30} The best biochemically and structurally characterized Type I-E Cas3 from Thermobifida fusca (Tf) harbors a diiron cofactor that was reported to be inactive.²² The Cas3″ from Methanocaldococcus jannaschii (Mj) contains two Ca²⁺ atoms in separate sites,²¹ whereas the truncated Type I-E Cas3 (lacking the SF2 domain) from Thermus thermophilus coordinates a single Ni²⁺ ion.²⁵ Despite the importance of Cas3 and the potential of the Cas3-cascade for genome editing applications, the nature of the cofactor(s) and metal ion selectivity are less explored, and little is known concerning the activation requirements of distinct Cas3 subtypes.

A common sequence feature for most Cas3 (with the exception of Type I-D) and specific solely to HD-domain PDEs is the presence of an extra histidine ligand, which occurs in tandem to the last His of the motif (i.e., H...HD...H...HH...D).^{20,22,27,31–33} In the structure of Tf Cas3, this histidine, H150, is within hydrogen bonding distance to a phosphate oxygen of the copurifying ssDNA.²² However, the role of this residue in cofactor assembly or activity is not well understood.

Cas3 enzymes are reported to hydrolyze cyclic mononucleotides, such as 2′,3′-cAMP,²¹ which are important messenger molecules in signaling pathways and virulence regulation,^34,35 raising the possibility that Cas3 may have other functions in the cell.³⁶ Numerous studies invoke a role of Cas3 in virulence, suggesting that its hydrolytic activity may be important in such pathways.^36–39 In addition, a homologous protein, the conserved virulence factor A (CvfA), is a major player in Staphylococcus aureus (Sa) virulence by acting on 2′,3′-cAMP and small RNA containing 2′3′-cAMP termini.^34,40–42 We hypothesize that Cas3 utilizes similar mechanisms to regulate virulence via 2′3′-cAMP hydrolysis.

The present study focuses upon the unexplored role of iron in Tf Cas3 catalysis and revisits previous reports in which addition of Fe³⁺ or Fe²⁺ did not support hydrolysis.²² In the present study, the Fe-containing Tf Cas3 hydrolyzes ssDNA with the highest observed rate when compared to the other metal-containing forms, thus suggesting that Fe is the biologically relevant cofactor. Cas3 is also active with Co-, Ni-, and Mn-based cofactors showing a large degree of promiscuity in the activating transition metal ions. The nuclearity of the Fe- and Mn-based cofactors and their different redox states have been explored by electron paramagnetic resonance (EPR) and ⁵⁷Fe Mössbauer spectroscopy. Although Fe is prone to oxidative inactivation, our spectroscopic studies under different redox conditions suggest that Cas3 would be active in the cell without the need of a reducing system, while in vitro cobalt can be a great alternative to maintain activity independent of redox conditions. ⁵⁷Fe Mössbauer spectroscopy and EPR spectroscopy demonstrate that Cas3″, like Cas3, can bind and assemble a diiron cofactor, consistent with a two metal-based hydrolysis. The extra histidine ligand in Cas3 proteins is important for efficient ssDNA hydrolysis but less critical for cofactor assembly or hydrolysis of small molecule substrates. Tf Cas3 hydrolyzes 2′3′-cAMP, an activity utilized by CvfA to regulate virulence.^34,40–42 We propose that CvfA is also dinuclear and show that it can maintain activity with most transition metal ions, albeit Mn and Fe are the most potent activators. Fe utilization by both Cas3 and CvfA appears to be a common and potentially important functional element for their specific activities in cellulo. Tf Cas3 exhibits comparable catalytic efficiency to that of CvfA, proposing a possible molecular basis for the reported Cas3-mediated virulence.

MATERIALS AND METHODS

Expression of Thermobifida fusca (Tf) Cas3.

The plasmid containing Cas3 from Tf (NCBI RefSeq: WP_081430412) in a pET28b(+) vector was kindly gifted by Dr. Ailong Ke (Cornell University, NY). Escherichia coli (Ec) T7 Express competent cells (New England Biolabs, Ipswitch, MA) were transformed with the Tf Cas3 plasmid and selected for kanamycin resistance. Transformed cells were grown in Luria-Bertani media (LB) with 50 μg/L kanamycin at 37 °C with shaking (200 rpm) until the OD₆₀₀ reached 0.6–0.8. Cells were cold shocked for 1 h at 4 °C prior to induction of protein expression by addition of 0.5 mM isopropyl β-d-1-thiogalactopyranoside. Cell cultures were incubated at 18 °C with shaking (200 rpm) for 16–20 h. Cells were harvested by centrifugation at 7000 rpm and at 4 °C for 15 min, flash frozen in liquid nitrogen, and stored at −80 °C. To direct specific metal incorporation into Tf Cas3, cells were grown in M9 minimal media in the presence of 0.2% (v/v) glucose, 0.1 mM CaCl₂, 200 mM MgSO₄·7H₂O_, and 10 μM (NH₄)₂Fe(SO₄)₂·6H₂O until an OD₆₀₀ of 0.6–0.8 was reached. A series of different transition metal ion salts was introduced at induction: ⁵⁶Fe²⁺ (250 μM (NH₄)₂Fe(SO₄)₂·6H₂O), ⁵⁷Fe²⁺ (125 μM ⁵⁷FeSO₄), Co²⁺ (50 μM CoCl₂), Ni²⁺ (50 μM NiSO₄), and Mn²⁺ (100 μM MnCl₂·4H₂O). Cell cultures were incubated at 18 °C with shaking (200 rpm) for 16–20 h. Cells were harvested by centrifugation for 15 min at 7000 rpm and 4 °C, flash frozen in liquid nitrogen, and stored at −80 °C.

Aerobic Isolation of Different Metal-Containing Forms of Tf Cas3.

Tf Cas3 expressed in LB media is denoted as LB-Tf Cas3, and Tf Cas3 expressed in M9 with different transition metal ions is denoted as M9-M-Tf Cas3 in which M stands for the respective transition metal ion (i.e., Fe⁽³⁺⁾²⁺, Co²⁺, Mn²⁺, and Ni²⁺). Purifications of all forms were essentially identical with minor changes (vide infra). Cell pellets were resuspended in lysis buffer (50 mM HEPES, 10 mM imidazole, 300 mM NaCl, 5% glycerol, pH 7.5). Phenylmethylsulfonyl fluoride was added to the cell suspension to a final concentration of 45 μg/mL. During the resuspension of the M9-M-Tf Cas3M cell pellets, the respective metal ion present during growth was added to the solution to a final concentration of 1 mM. The suspension was then lysed via sonication (QSonica) and centrifuged at 22,000g for 30 min to remove cell debris. The clarified lysate was loaded onto an Ni²⁺-NTA immobilized affinity chromatography column (~10 mL resin per 50 mL lysate) equilibrated with the lysis buffer. In the case of the M9-Co-Tf Cas3, purification utilized a Talon resin (GE Healthcare) to avoid possible contaminations of Ni²⁺. The column was washed with wash buffer (50 mM HEPES, 20 mM imidazole, 300 mM NaCl, pH 7.5). Bound protein was eluted using the elution buffer (50 mM HEPES, 300 mM imidazole, 150 mM NaCl, 10% glycerol, pH 7.5). Fractions containing the protein were pooled and concentrated at 4000g using a 30 K MWCO Amicon Centrifugal Filter (Millipore, Sigma). An additional purification step including size exclusion chromatography was followed using a HiLoad 16/600 Superdex 200 pg column (GE Healthcare) equilibrated in 50 mM HEPES, 150 mM NaCl, 10% glycerol, pH 7.5. Fractions containing pure and monomeric Tf Cas3 were combined and concentrated at 4000g using a 30 K MWCO Amicon Centrifugal Filter (Millipore, Sigma). Protein purity was assessed via SDS-PAGE (Figure S1) with Coomassie staining, and protein concentration was determined both via the Bradford assay and the extinction coefficient ε₂₈₀ = 143.8 mM⁻¹ cm⁻¹ (http://web.expasy.org/protparam/). The Fe content was determined by the ferrozine assay.

Generation of the H150A Tf Cas3 Variant.

The H150A variant was generated employing the Q5 Hot Start Site Directed Mutagenesis (NEB, Ipswitch, MA) with the primers (forward: 5′-TGGGGGCCATGCGGGTACGTTCCACC-3′; reverse: 5′-AGCATCTGGGCAACGAGG-3′). All WT and variant sequences were confirmed by Sanger sequencing (Genewiz Inc., NY).

Expression and Purification of Cas3″ Proteins.

The Cas3″ from Pyrococcus furiosus (Pf) (Uniprot ID: Q86336) and Cas3″ from Thermococcus thioreducens (Tt) (Uniprot ID: A0A0Q2RFE4) were synthesized, codon optimized for expression in Ec, and inserted into pET26b(+) via NdeI and XhoI by GenScript (Piscataway, NJ). Pf Cas3″ and Tt Cas3″ were transformed into Ec T7 express cells (NEB) and followed expression in M9 media supplemented with ⁵⁶Fe (250 μM) as described for Tf Cas3. For preparation of ⁵⁷Fe-labeled samples, ⁵⁷FeSO₄ was added to a final concentration of 125 μM upon induction. Purifications for Pf Cas3″ and Tt Cas3″ were carried out according to the protocol described for Tf Cas3. Protein purity was assessed via SDS-PAGE with Coomassie staining (Figures S2 and S3), and protein concentration was determined via the Bradford assay. The Fe content was determined by the ferrozine assay.

Expression and Isolation of CvfA.

Full length CvfA from Staphylococcus aureus (Sa) (NCBI accession: WP_001050913) was codon optimized for expression in Ec and inserted into pET26b(+) via NdeI and XhoI by GenScript (Piscataway, NJ). FL CvfA was transformed in Ec T7 express cells (NEB) and selected for kanamycin resistance. Cells were grown in LB media following the same growth protocol as Tf Cas3-LB. The purification steps were the same as those followed for Tf Cas3-LB with slight modifications in the buffers: lysis buffer (50 mM HEPES, 300 mM NaCl, 200 mM KCl, 10 mM imidazole, 0.5% nonyl phenoxypolyethoxylethanol (NP-40), 1% TWEEN20, pH 8.2), wash buffer (50 mM HEPES, 20 mM imidazole, 300 mM NaCl, 200 mM KCl, pH 8.2), elution buffer (50 mM HEPES, 150 mM NaCl, 200 mM KCl, 300 mM imidazole, 10% glycerol, pH 8.2), and storage buffer (50 mM TAPS, 500 mM KCl, 10% glycerol, pH 9). Protein purity was assessed via SDS-PAGE with Coomassie staining, and protein concentration was determined both via the Bradford assay and the extinction coefficient ε₂₈₀ = 13.4 mM⁻¹ cm⁻¹ (http://web.expasy.org/protparam/).

Generation of the Transmembrane Terminal Deletion CvfA Variant (delTM CvfA).

The delTM CvfA variant was generated employing the Q5 Hot Start Site Directed Mutagenesis (NEB, Ipswitch, MA) with the primers (forward: 5′-CTGCTGCAAAAACAGAGC-3′; reverse: 5′-CATATGTATATCTCCTTCTTAAAGTTAAAC-3′) designed using NEBaseChanger. Variant sequences were confirmed by Sanger sequencing (Genewiz Inc., NY). delTM CvfA was purified following the procedure of the FL protein. Protein purity was assessed via SDS-PAGE with Coomassie staining, and protein concentration was determined both via the Bradford assay and the extinction coefficient ε₂₈₀ = 11.9 mM⁻¹ cm⁻¹ (http://web.expasy.org/protparam/).

Nuclease Assays.

Nuclease activity assays were carried out with the circular M13mp18 single-stranded DNA (NEB). The M13mp18 substrate was exchanged into water to remove the 1 mM ethylenediaminetetraacetic acid (EDTA) present in its supplied storage buffer. All Tf Cas3 forms were first incubated with EDTA as a chelating agent. After removal of EDTA, the “apo” forms of Tf Cas3 were incubated with 1 mM excess divalent metal for 30 min prior to addition of substrate. Assays monitoring the effect of salt concentration of the as-purified Tf Cas3-LB were performed by varying the amount of NaCl from 0.05 M to 1 M in 100 mM HEPES, pH 8.0 buffer. Assays with Tf Cas3-LB were carried out with 10 nM ssM13mp18 and 1 μM protein reconstituted with the following transition metal ions: Fe⁽³⁺⁾²⁺, Co²⁺, Mn²⁺, Zn², Cu²⁺, Mg²⁺, and Ni²⁺ at 1 mM final concentrations, respectively. Reactions with Fe²⁺ were performed under O₂-free conditions in the presence of 1 mM sodium hydrosulfite (dithionite, DT). Each reaction was incubated at 37 °C for 60 min in 100 mM HEPES, 200 mM NaCl, and pH 8.0, unless otherwise specified. The pH dependence of the Tf Cas3-LB nuclease activity was screened in a pH range of 4–11 utilizing different buffers of the same concentration. Assays with Tf Cas3-M9 were carried out under identical conditions, with the exception that no exogenous transition metals were added during the reactions. In the case of the Tf Cas3-Fe, the protein was reduced under O₂-free conditions with 1 mM dithionite (~100-fold excess relative to protein concentration) prior to the reaction. All reactions were initiated by addition of the ssM13pm18 substrate. The reaction samples were quenched with EDTA, treated with 6× DNA loading buffer (NEB), and run on a 1% agarose gel.

Activity Assays Employing Cyclic Monophosphates and Diphosphates.

The PDE activity of WT Tf Cas3 was screened toward known cyclic monophosphates and diphosphates (i.e., cAMP, cCMP, cGMP and c-di-GMP, c-di-AMP, and 3′3′cGAMP) (Axxora, LLC). Reactions were carried out at room temperature for 60 min in an anaerobic glovebox (Coy Laboratories), unless stated otherwise. Various metal-substituted forms of Cas3 were reacted with a 1 mM concentration of cyclic-monophosphate or diphosphate substrates in a buffer containing 50 mM HEPES, 150 mM NaCl, 10% glycerol, pH 7.5. Cas3-Fe form was first reduced with 2 mM sodium hydrosulfite for 30 min, prior to the addition of the substrate. All activity assays have been carried out on the basis of diiron cofactor concentrations by combining information from elemental analyses and Mössbauer spectroscopy. Protein concentrations for activity assays have thus been normalized for diiron concentration based on the extent of dimetal cofactor assembly according to Mössbauer and do not include the monomeric Fe that does not contribute to activity. Reactions were quenched at 95 °C for 5 min and centrifuged at 21,000g for 10 min to remove precipitated protein, and the supernatant was further passed through a microcentrifuge tube Spin-X centrifuge tube filter (Corning Incorporated, Corning, NY). Reaction educts and products were separated with an Agilent reverse-phase C18-A Polaris column (particle size 5 μm, 150 × 4.6 mm) and analyzed by high-performance liquid chromatography (HPLC) (Agilent 1260 Infinity Liquid Chromatography system) with a gradient method of aqueous solvent A (10 mM KH₂PO₄ and 10 mM TBAH, pH 6.0) and an organic-based mobile phase, solvent B (10 mM TBAH in methanol). Products and educts were eluted with a gradient of 95% solvent A and 5% solvent B to 50% solvent A and 50% solvent B at a flow rate of 1.5 mL/min for 34 min. Relative integrated peak intensities were compared to internal standards of known concentration. Nucleotides were detected at 254 nm.

Competition Assays between the WT and H150A Tf Cas3.

The Fe forms of WT and H150A Tf Cas3 were assayed with ssM13mp18 (New England Biolabs, Ipswitch, MA) as a substrate to assess the importance of the H150 residue in the hydrolysis of ssDNA. The reactions were performed in a buffer containing 50 mM HEPES, 150 mM NaCl, 10% glycerol, pH 7.5, employing 0.5 μM cofactor (diiron) enzyme concentration, which was reduced to its active Fe^II–Fe^II form by addition of 1 mM sodium dithionite under O₂-free conditions for 20 min at room temperature. The reactions for both the WT and H150A were initiated by addition of 15 nM ssM13mp18 in an anaerobic glovebox (CoyLab, Grass Lake, MI) and performed at 37 °C. The nuclease assay was quenched at the selected time points, i.e., 2, 4, 8, 16 and 30, or 10 and 20 min, by addition of 15 mM EDTA. The reaction outcome was monitored by running of the samples on a 1% agarose gel.

Activity Assays Employing Different HD-Domain Proteins.

All assays were performed for 60 min with the diiron form of the following enzymes: Persephonella marina GH (PmGH), VCA0681, VCA0931, Bacillus halodurans YqeK, Clostridium acetobutylicum YqeK, PhnZ, MIOX, and the putative protein from Thermotogales (PDB ID: 2PQ7). Detailed protocols for activity assays can be found in the SI Appendix. All activity assays were performed in an oxygen-free chamber. For diiron enzymes, 5 μM cofactor (protein concentration multiplied by metals/protein divided by nuclearity) was mixed with oxygen-free buffer (50 mM HEPES, 150 mM NaCl, 10% glycerol, pH 7.5, degassed with argon) and 2 mM sodium dithionite for 20 min to reduce Fe^III–Fe^III to Fe^II–Fe^II. Reaction was initiated by addition of 0.5 mM 2′3′-cAMP. After 60 min, the reaction was quenched at 95 °C for 5 min, apart from Ca YqeK and Bh YqeK in which 15 mM EDTA was added in addition. The samples were then spin-filtered and analyzed by HPLC.

EPR Spectroscopy.

All samples were prepared under O₂-free conditions in an anaerobic glovebox (Coy). All protein samples (400–600 μM Fe–Fe) were reacted with twofold excess ascorbate sodium salt for 5 min at 22 °C prior to freezing in liquid N₂. EPR spectra were acquired with a Bruker E500 Elexsys continuous wave (CW) X-Band spectrometer (operating at approx. 9.38 GHz) equipped with a rectangular resonator (TE102) and a continuous-flow cryostat (Oxford 910) with a temperature controller (Oxford ITC 503). The spectra were recorded at variable temperatures between 10 and 40 K at a microwave power of 0.2 mW, using a modulation amplitude of 1 mT and a microwave frequency of 9.38 GHz.

Mössbauer Spectroscopy.

Mössbauer spectra were recorded on WEB Research (Edina, MN) instruments that have been described previously. The spectrometer used to acquire the weak-field spectra is equipped with a Janis SVT-400 variable-temperature cryostat. The external magnetic field was applied parallel to the γ beam when recorded at 4.2 K. No magnetic field was present for samples recorded at 120 K. All isomer shifts are quoted relative to the centroid of the spectrum of α-iron metal at room temperature. Mössbauer spectra were simulated using the WMOSS spectral analysis software (www.wmoss.org, WEB Research, Edina, MN).

Stopped-Flow Absorption Spectroscopy (SF-Abs).

SF-Abs measurements were carried out with an SX20 stopped-flow spectrophotometer from Applied Photophysics Ltd. (Leatherhead, UK) that was housed in an anoxic chamber (Coy). All experiments were carried out at 5 °C. In single-mix (two-syringe) experiments, Tf Cas3 (200 μM cofactor) reactant solution was diluted by twofold. All O₂ concentrations quoted were calculated by estimating the concentration of an O₂-saturated buffer at 5 °C in our buffer (i.e., 1.8 mM). Reactions of the dithionite-reduced Tf Cas3 samples were carried out in a single-mixing setup in which the O₂-free protein was mixed in a 1:1 ratio with varying concentrations of an O₂-containing buffer. The optical path length used was 10 mm. For the characterization of the reactions, a photodiode array detector was used to acquire time-resolved absorption spectra. The kinetics of the oxidation of Tf Cas3 were followed by monitoring the absorbance at 370 nm over time, and these were fitted using a single-exponential function according to the equation:

$$\Delta A_{370}(t)=\Delta A_1\left(1-{\text{e}}^{-k_{1}t}\right)$$ 1

which gives the change in absorbance with respect to time for an irreversible first-order reaction as a function of the apparent first-order rate constant k₁. The plot of the apparent first-order rate constant (k₁) vs [O₂] yields a bimolecular rate constant (slope) k(O₂).

Metal Quantifications.

Metal analysis of the as-purified proteins was determined by inductively coupled atomic emission spectrometry (ICP-AES) by Laura Jean Liermann at the Laboratory for Isotopes and Metals in the Environment at the Pennsylvania State University (University Park, PA).

RESULTS

Tf Cas3 Is Activated by Fe²⁺ and a Variety of Other Transition Metal Ions.

Aerobically isolated Tf Cas3 from LB media hydrolyzed ssDNA in the absence of exogenously added metals (Figure 1A). The observed hydrolysis likely stems from Fe, Ni, or Zn, which copurify with the protein, as shown by elemental analyses (Table 1). Tf Cas3 was thus first treated with EDTA that was subsequently removed, to yield a protein form devoid of bound metal ions (apo) and with no detectable activity (Figure 1A). Addition of Fe³⁺, Zn²⁺, or Cu²⁺ did not result in detectable ssDNA cleavage, whereas addition of Fe²⁺, Co²⁺, Mn²⁺, Ni²⁺, and Ca²⁺ led to significant hydrolysis, with Fe²⁺ and Co²⁺ conferring the highest activity. This is the first observation that ferrous iron can stimulate PDE activity in Tf Cas3, and although these assays provide a trend in the activating metals, the results may be skewed due to excess and purity of added metal ions, as well as inability to estimate the precise active cofactor concentration.

Figure 1.

Metal cofactors that support catalysis in Tf Cas3. (A) ssDNAse activity of Tf Cas3-LB as is (no EDTA) and with exogenously added metals analyzed on a 1% agarose DNA gel. Metal screening studies were carried out with 1 μM Tf Cas3-LB, 10 nM M13mp18, 25× metal, incubated for 60 min at 37 °C. (B) ssDNAse activity of Fe-, Co-, Mn-, and Ni-containing forms isolated from M9 media. The Fe protein was assayed both in its aerobically isolated form (Fe³⁺) and after addition of 1 mM dithionite under O₂-free conditions (Fe²⁺). Reactions were carried out with 1 μM protein, 10 nM M13mp18, for 60 min at 37 °C. Protein concentrations were adjusted to dimetal cofactor concentrations based on ICP-AES. (C) ssDNA hydrolysis of M9-Co-Tf Cas3 at varying salt concentrations and (D) pH values. (E) ⁵⁷Fe Mössbauer spectra of the M9-⁵⁷Fe-Tf Cas3 under different redox conditions recorded at T = 4.2 K and in the presence of a small magnetic field of 78 mT applied parallel to the direction of the γ-beam. Experimental spectra are depicted with black solid bars; the total simulation is shown in gray lines. Top: the fits corresponding to the diferric sites with the small and large ΔE_Q are shown in doublets shaded green and black, respectively. Middle: the fits corresponding to the diferric and diferrous sites are shown in doublets shaded green and purple, respectively. The arrows denote the resonances associated with the paramagnetic mixed-valent Fe^II–Fe^III state. Bottom: the fits of the two quadrupole doublets corresponding to diferrous sites with the small and large ΔE_Q are shown in purple and blue shaded doublets, respectively. (F) Selected UV–vis traces monitoring oxidation of the diferrous M9-Fe-Tf Cas3 by molecular O₂ and plot of the observed rates vs [O₂]. Absorbance at 330, 365, and 470 nm increases with increasing reaction time denoting oxidation to the diferric form. (G) EPR spectrum of M9-Mn-Tf Cas3. Experimental conditions: temperature = 10 K, microwave freq. = 9.38 GHz, and modulation amplitude = 1 mT.

Table 1.

Elemental Analysis of the Different Metal Forms of Tf Cas3 and Sa CvfA Employed in This Study as Determined by ICP-AES

sample	Co	Cu	Fe	Mn	Ni	Zn	Ca
Tf Cas3 LB	n.d.	n.d.	1.14	0.03	0.26	0.15	0.1
Tf Cas3 M9-Fe	n.d.	n.d.	1.6	0.02	0.26	0.10	0.09
Tf Cas3 M9-Co	1.52	n.d.	0.12	n.d.	0.03	0.13	0.16
Tf Cas3 M9-Mn	n.d.	n.d.	0.24	1.06	0.38	0.08	0.12
Tf Cas3 M9-Ni	0.13	0.1	0.4	n.d.	1.4	0.15	0.1
Tf Cas3 H150A	n.d.	n.d.	0.88	n.d.	0.50	0.14	0.10
Tt Cas3″ M9-Fe	n.d.	n.d.	1.02	n.d.	0.25	0.08	0.09
Pf Cas3″ M9-Fe	n.d.	n.d.	0.78	n.d.	0.27	0.11	0.10
Sa CvfA LB	n.d.	n.d.	0.03	n.d.	0.03	0.03	0.35

Tf Cas3 was expressed in minimal (M9) media directing selective incorporation of the transition metals shown to stimulate activity (Figure 1A). Tf Cas3 expressed in M9 with different transition metal ions is denoted as M9-M-Tf Cas3 in which M stands for the respective transition metal ion (i.e., Fe⁽³⁺⁾²⁺, Co²⁺, Mn²⁺, and Ni²⁺). The aerobically isolated M9-Fe-Tf Cas3 was inactive, unless treated with a reducing agent, such as ascorbate or dithionite (Figure 1B). In contrast, the Mn-, Ni-, and Co-forms of Tf Cas3 were constitutively active. M9-Cas3 copurifies with Fe, Co, Ni, and Mn to comparable extents and hydrolyzes ssDNA with very similar efficiencies (Table 1 and Figure 1B). The activity profile of the different metal forms of Tf Cas3 appears to loosely follow the Irving–Williams series. We further assessed the impact of increased ionic strength (0.05–1 M NaCl) on the ssDNA degradation and observed no inhibition of ssDNAse activity (Figure 1C). Similarly, all M9-Cas3 metal-containing forms were active over a wide range of pHs, from 5.5 to 11.5 (Figure 1D).

To gain insight into the chemical nature of the several Fe-containing forms, we supplemented Tf Cas3 with ⁵⁷Fe. The ⁵⁷Fe Mössbauer spectrum of the aerobically purified Tf Cas3 is composed of two quadrupole doublets with parameters (δ₁ = 0.48 mm/s, ΔE_Q1 = 0.93 mm/s, and δ₂ = 0.54 mm/s, ΔE_Q2 = 1.79 mm/s), respectively, characteristic of antiferromagnetically (AF) coupled diferric iron sites (Figure 1E).^43–48 Assignment of the quadrupole doublets was based on previously characterized HD-domain diiron PDEs³² and Cas3 variants (vide supra). The two distinct doublets may suggest a difference in (i) the coordination of each of the Fe sites (e.g., due to the extra His150, a water ligand, etc.), (ii) the protonation state of a hydroxide or water ligand, and (iii) the presence of an oxo-ligand.^44–49 The aerobically isolated Tf Cas3 cannot hydrolyze ssDNA (Figure 1B), demonstrating that the diferric form is inactive.

Treatment with ascorbate yields the semireduced, mixed-valent Fe^II–Fe^III state (20% of total absorption) and the two-electron reduced diferrous Fe^II–Fe^II state (80% of total absorption). The mixed-valence Fe^II–Fe^III form consists of two AF-coupled Fe ions, resulting in an S = 1/2 paramagnetic ground state.^43,50 At low temperatures and in the presence of an external magnetic field, the spectral components are magnetically split and exhibit broad resonances, in contrast to those of the Fe^III–Fe^III and Fe^II–Fe^II forms that appear as quadrupole doublets (Figure 1E, middle). At high temperatures, the Fe^II–Fe^III spectrum collapses into two quadrupole doublets (Figure S4), as a result of the electronic relaxation rate being fast compared to the nuclear Larmor precession frequency.^51,52 Identification and quantification of the Fe^II–Fe^III was afforded by parallel analysis of high-temperature Mössbauer spectra (Figure S4) and EPR spectra (vide infra). Reaction with the stronger reducing agent sodium dithionite affords a sample homogeneously enriched in the Fe^II–Fe^II form (Figure 1E, bottom). The broad quadrupole doublet could be best fit, considering two quadrupole doublets with parameters (δ₁ = 1.25 mm/s, ΔE_Q1 = 3.05 mm/s, and δ₂ = 1.26 mm/s, ΔE_Q2 = 2.04 mm/s), respectively, which are characteristic of high-spin Fe^II sites with O/N ligation.⁵³ The dithionite-reduced M9-Fe-Tf Cas3 yields the highest activity, in agreement with other Fe-dependent PDEs of the HD-domain superfamily.^31,32

All metal-containing forms of Tf Cas3 are constitutively active under aerobic conditions, apart from the Fe form, which is prone to oxidative inactivation (i.e., conversion to Fe^III–Fe^III state). The Fe^II–Fe^II cofactor form reacts somewhat sluggishly with O₂ with an apparent second order rate constant of 1.4 mM⁻¹ s⁻¹ (Figure 1F), which is at least two orders of magnitude slower than that of other nonheme diiron oxidases and oxygenases.⁵⁴ Among the other metal forms of Cas3, the M9-Mn-Tf Cas3 is amenable to EPR spectroscopy. The EPR spectrum of M9-Mn-Tf Cas3 shows assembly of a dimanganous center (Mn^II–Mn^II) (Figure 1G),⁵⁵ providing further evidence that Tf Cas3 assembles dimetal centers with different transition metal ions (as inferred from the ssDNA cleavage assays, Figure 1A).

Type I-A Cas3″ Assembles a Dimetal Active Site Similar to Type I-E Cas3.

The Cas3″ from Thermococcus thioreducens (Tt) and Pyrococcus furiosus (Pf) lack the SF2 helicase domain, while their respective HD domains share ~35% sequence similarity to the HD-domain of Tf Cas3 (Figure 2A,B). Their primary sequences contain all seven conserved residues that help coordinate the diiron cofactor in Tf Cas3 (Figure 2B), invoking the notion that Cas3″ also assemble dimetal active sites. Tf Cas3, Tt, and Pf Cas3″ were expressed in M9 media supplemented with iron to yield their respective Fe-containing forms that can be employed as spectroscopic probes for dimetal site assembly. The EPR spectrum of the ascorbate-reduced Tf Cas3 (Figure 2C, top) exhibits a broad rhombic signal that has an average g-value <2, confirming the presence of an AF-coupled mixed-valent (Fe^II–Fe^III, S = 1/2) center.⁴³ This signal is consistent with those previously detected in nonheme diiron proteins, including those of the HD-domain superfamily.^31,43,50 Under these conditions and on the basis of EPR quantitation, the Fe^II–Fe^III state accumulates to ~15–20% of the total diiron cofactor in the sample. The spectra of the ascorbate-reduced Tt and Pf Cas3″ show very similar signals to those of Tf Cas3, both with respect to their principal g-values (i.e., g = 1.93, 1.78, 1.57) and relaxation properties. The EPR spectra thus demonstrate assembly of diiron cofactors, in contrast to previous crystallographic observations.²¹

Figure 2.

Structures, sequence alignment, and spectroscopic characterization of Cas3 and Cas3″. (A) Crystal structure overlay of Tf Cas3 (PDB: 4QQW) and Mj Cas3″ (PDB: 3S4L). The residues of the HD-domain binding motif from Tf Cas3 and Mj Cas3″ are shown in blue and magenta, respectively. Fe and Ca atoms are shown as orange and green spheres, respectively. (B) Sequence alignment of Tf Cas3, Thermus thermophilus Cas3, Pyrococcus furiosus Cas3″, Thermophilum thermoaceticum Cas3″, and Methanocaldococcus jannaschii Cas3″. The seven conserved residues are highlighted in green (histidines) and yellow (aspartates). (C) EPR spectra of the ascorbate-reduced M9-Fe-Tf Cas3, Tt Cas3″, and Pf Cas3″ with their principal g-values annotated. Experimental conditions: T = 10 K, microwave freq. = 9.38 GHz, and modulation amplitude = 1 mT. (D) ⁵⁷Fe Mössbauer spectra of the aerobically isolated Tf Cas3, Tt Cas3″, and Pf Cas3″ recorded at T = 4.2 K and in the presence of a small magnetic field of 78 mT applied parallel to the direction of the γ-beam. The experimental spectra are depicted with black solid bars, the total simulation is shown in solid gray lines, and the fits corresponding to the diferric sites with the small and large ΔE_Q are shown in green and black shaded doublets, respectively.

We recorded the ⁵⁷Fe Mössbauer spectra on all three aerobically isolated proteins to assess extent of diiron formation and existence of any lower nuclearity Fe forms (if present). The 4.2 K/78 mT Mössbauer spectra of the Tt and Pf Cas3″ (Figure 2D) exhibit quadrupole doublets with parameters characteristic of diferric sites (Fe^III–Fe^III, S = 0). There are two notable differences with respect to the spectra of the Tf Cas3. First, Cas3″ exhibit mainly a single quadrupole doublet δ = 0.49 mm/s and ΔE_Q = 0.74 mm/s (Tt Cas3″) or δ = 0.48 mm/s and ΔE_Q = 0.78 mm/s (Pf Cas3″) (green shaded doublet, Figure 2D). The spectra also show the doublet with the larger quadrupole splitting, but the latter is a minor component (10–13% of the total Fe in the sample) and has parameters δ = 0.50 mm/s and ΔE_Q = 1.65 mm/s (for both Tt and Pf Cas3″) (black shaded doublet, Figure 2D). Second, the spectra of Cas3″ contain broad resonances extending to ±6 mm/s that are reminiscent of high-spin (S = 5/2) mono-iron species. The mononuclear Fe^III centers account for 30% for Tt Cas3″ and 40% for Pf Cas3″ of the total Fe in each sample. These are hardly detectable in Tf Cas3, suggesting either an incomplete cofactor incorporation in Cas3″ due to increased structural flexibility or the presence of ferric oxides that represent oxidative contaminants bound adventitiously. The Mössbauer spectra thus demonstrate a larger instability of the dimetal cofactor in Cas3″ when compared to Cas3 and can potentially explain the selective crystallization of mononuclear centers in the existing structure of Cas3″.²¹ ssDNA assays with the dithionite-reduced form of Cas3″ resulted in no observable activity, even after addition of exogenous metals (Figure S5), in contrast to Tf Cas3.

Role of Histidine 150 in Active Site Assembly and Hydrolysis.

The active site coordination of Tf Cas3 harbors an ancillary histidine, for which the role in cofactor assembly and activity remains uncharted both for Cas3 in particular and for HD-domain PDEs in general. H150 enriches the coordination of one of the iron ions and is within hydrogen bonding distance with the phosphate oxygen of the copurifying ssDNA fragment (Figure 3A). These observations invoke its potential role in hydrolysis, diiron cluster stability, and/or substrate positioning. The H150A variant of Tf Cas3 was expressed in M9 media with Fe supplementation because (a) Fe yields the most active form and (b) cofactor assembly can be probed by EPR and ⁵⁷Fe Mössbauer spectroscopy. Overall, the H150A variant copurifies with only half of the Fe compared to WT (Table 1), suggesting a likely stabilizing role of H150 in active site structure. However, the less than unity value does not allow inference of dimetal cofactor assembly.

Figure 3.

Spectroscopic characterization and ssDNAse activity of the H150A Tf Cas3. (A) Crystal structure of the WT Tf Cas3 active site highlighting the position of the H150 and the rest coordinating residues. The copurifying ssDNA is shown in cyan blue, and Fe atoms are represented as orange spheres. (B) CW EPR spectra of the ascorbate-reduced WT and H150A Tf Cas3. Experimental conditions: T = 10 K, microwave freq. = 9.38 GHz, and modulation amplitude = 1 mT. (C) ⁵⁷Fe Mössbauer spectra of the WT and H150A Tf Cas3 recorded at T = 4.2 K and a small magnetic field of 78 mT applied parallel to the direction of the γ-beam. The experimental spectra are shown as black bars, and the fits for H150A and WT are depicted with red and gray solid lines, respectively. (D) ssDNAse activity of the dithionite-reduced WT and H150A M9-Fe-Tf Cas3 at different time points analyzed on 1% agarose gels. The assays were carried out at 37 °C with 0.5 μM enzyme (diiron cofactor) and 15 nM M13mp18 for 2, 4, 8, 16, and 30 min, respectively.

The EPR spectrum of the ascorbate-reduced H150A Tf Cas3 shows a signal with an average g-value <2, confirming formation of an AF-coupled Fe^II–Fe^III core (Figure 3B). The overall line shape and extant principal g-values (1.93, 1.83, and 1.52) are different from those of the WT (Figures 2C and 3B), consistent with the absence of the His ligand perturbing the electronic structure of the site. The 4.2 K/78 mT Mössbauer spectrum of the aerobically isolated H150A shows that all the iron in the sample is assembled in diferric (Fe^III–Fe^III) sites (Figure 3C) and exhibits two quadrupole doublets with essentially identical Mössbauer parameters to those of the WT protein. One notable difference is the ratio of the two quadrupole doublets, which is now not in a 1:1 stoichiometry. The presence of the two doublets suggests a heterogeneity in the diferric site conformations and excludes the H150 as the molecular origin for the observed Fe site differentiation in the WT spectra. ssDNA hydrolysis was significantly compromised with the H150A variant to the extent that after 30 min only a faint smear was observed in the gel, consistent with very slow hydrolysis (Figure 3D). The H150 is thus crucial for processive DNA cleavage but not essential for cofactor assembly.

Cas3 Activities beyond ssDNA Cleavage; Hydrolysis of Cyclic Mononucleotides.

Employing the dithionite-reduced M9-Fe-Tf Cas3, we screened a series of nucleotides that are known PDE substrates of the HD-domain superfamily (Figure 4A) and identified 2′3′-cAMP as a potential substrate. Overall, (i) Tf Cas3 cleaves 2′3′-cAMP yielding 3′-AMP as the major and 2′-AMP as the minor product (3-AMP/2-AMP is ~3.5/1), and (ii) cyclic mononucleotide hydrolysis is dependent on the nature of the nucleoside base with preference to purines, and reaction is regioselective (Figure 4A). WT Tf Cas3 hydrolyzes 2′3′-cAMP with an apparent k_cat of 4.7 s⁻¹ and a K_M of 27 (±6) mM (Figure 4B). Both the K_M and catalytic efficiency (k_cat/K_M = 174 M⁻¹ s⁻¹) are in the same range with that of the 2′3′-cAMP-cleaving HD-domain PDE CvfA (k_cat/K_M = 53 M⁻¹ s⁻¹). Although the dithionite-reduced M9-Fe Tt and Pf Cas3″ forms did not support ssDNA degradation (Figure S5), they both cleave 2′3′-cAMP to similar extents to Tf Cas3, albeit with twofold to threefold slower apparent initial rates (Figure 4C). These findings suggest that the Fe form of Cas3 cofactors may be more catalytically relevant than initially thought and that the observed inability of Cas3″ to cleave ssDNA under these conditions may stem from lack of an activating (co)factor.

Figure 4.

Activity screen of Tf Cas3 with cyclic mononucleotides and dinucleotides, and 2′3′-cAMP hydrolysis by select HD-domain proteins. (A) Tf Cas3 hydrolysis of cyclic mononucleotides and dinucleotides monitored by HPLC. Substrates are annotated in black, and products are annotated in teal and orange. Reactions were carried out with 5 μM Tf Cas3 (diiron cofactor) and 1 mM nucleotide for 60 min. (B) Michaelis–Menten plot of 2′3′-cAMP hydrolysis by the dithionite-reduced M9-Fe-Tf Cas3. The plot for 3′-AMP and total (3′AMP + 2′AMP) product formation is shown in teal and black, respectively. (C) 2′3′-cAMP hydrolysis by Tf Cas3, Tt, and Tf Cas3″ employing the dithionite-reduced proteins (5 μM diiron cofactor) and 1 mM 2′3′-cAMP. Reactions were carried out for 60 min at 37 °C (for Tf Cas3) and 55 °C for Tt and Pf Cas3″. (D) Activity screen with 2′3′-cAMP by Tf Cas3 and other representative diiron HD-domain proteins. Activity assays were carried out with dithionite-reduced proteins containing 5 μM diiron cofactor and 0.5 mM 2′3′-cAMP. The percentage of 3′-AMP and 2′-AMP formation is shown in teal and orange, respectively. (E) HPLC traces of 2′3′-cAMP hydrolysis by the different M9-M-Tf Cas3 forms. Protein concentrations have been normalized for cofactor content. (F) PDE activity of H150A with 20 mM 2′3′-cAMP as a substrate. Production of 3′-AMP and 2′-AMP is shown in purple and green, respectively.

We explored 2′3′-cAMP hydrolysis by other Fe-dependent HD-domain proteins to examine whether 2′3′-cAMP hydrolysis is a general feature of the HD-domain scaffold or specific to Tf Cas3. The selected proteins are either functionally similar (i.e., PDEs)^32,56,57 or different (i.e., phosphatases, oxygenases, or of unknown function).^25,43,50,58 The data support a preference of Tf Cas3 for 2′3′-cAMP; hydrolysis occurs 5–10-fold faster than other PDEs, while the rest proteins are one to two orders of magnitude slower (Figure 4D). The 2′3′-cAMP activity of Tf Cas3 has a metal dependence that agrees well with the cofactor profile observed for the ssDNAse activity (Figure 1), with the Fe and Co forms being the most active (Figure 4E). H150A hydrolyzes 2′3′-cAMP with an apparent rate that is 3–4 times slower than that of the WT, and this apparent activity decrease stems from changes in the ratio of the two products (Figure 4F). In the H150A variant, formation of 3′-AMP is disfavored without affecting extent of 2′-AMP formation, resulting in a nearly 1:1 ratio of both products. These observations suggest an involvement of H150 in Cas3 regioselectivity.

Cas3 and CvfA Hydrolyze 2′3′-cAMP with Comparable Catalytic Efficiency.

CvfA is the prototypical HD-domain enzyme associated with hydrolysis of 2′3′-cAMP modified RNA in cellulo.^34,40,42 We thus isolated the Staphylococcus aureus (Sa) CvfA and performed experiments under the same conditions as Tf Cas3. Sa CvfA was isolated devoid of any metal ions irrespective of the type of expression media (Table 1). This finding is in stark contrast to Cas3 and Cas3″, which demonstrate high affinity for Fe or any divalent metal ions when exogenously supplemented (Table 1). CvfA shares the same seven-residue HD motif with Tf Cas3 and Cas3″, suggesting a common two metal-based mechanism for hydrolysis (Figure 2). Experimental verification of cofactor nuclearity was not successful, due to lability of the metal ions. Therefore, metal profiling for its 2′3′-cAMP activity was performed under saturating concentrations of added metal ions at which the observed activity was maximized (Figure 5A). Sa CvfA exhibits the highest activity with Mn, as previously reported.³⁴ Fe, however, also supports activity to comparable extents, demonstrating that iron utilization in hydrolytic HD-domain enzymes is more prevalent than initially considered. The panel of metal ions conferring hydrolytic activity is generally retained between CvfA and Cas3, except for Mg and Zn (Figures 1 and 5A).

Figure 5.

Metal-dependent screen of the 2′3′-cAMP hydrolysis by Sa CvfA and activity comparison to M9-Co-Tf Cas3. (A) Metal selectivity of CvfA with 2′3′-cAMP. (B) Activity assay of Tf Cas3 and Sa CvfA with 2′3′-cAMP under the same conditions. Reactions were carried out with 4 μM enzyme (cofactor basis) and 1 mM 2′3′-cAMP, respectively.

We proceeded to carry out reactions with Sa CvfA utilizing Mn (the most potent activator) at the minimum metal concentration at which hydrolysis rates reach a plateau. The Mn-dependent hydrolysis of Sa CvfA was compared to that of M9-Co-Tf Cas3 under the same conditions. This form of Tf Cas3 was selected, as it is constitutively active and together with Fe exhibits the highest extent of 2′3′-cAMP hydrolysis. Tf Cas3 hydrolyzed 2′3′-cAMP with an observed rate that is four times faster than that of Sa CvfA (Figure 5B), and both proteins yielded 3′-AMP as the major reaction product. Hydrolysis by Tf Cas3 is on a par (if not greater) with Sa CvfA, providing an additional argument favoring a possible biological relevance of 2′3′-cAMP hydrolysis by Cas3.

DISCUSSION

In the present study, we map a promiscuous metal ion cofactor profile for Tf Cas3 and demonstrate that it employs many different homo-dimetal cofactors (Fe²⁺, Co²⁺, Mn²⁺, and Ni²⁺) to perform processive ssDNA cleavage. The previously reported inactive diiron cofactor affords the highest extent of ssDNA degradation. The data suggest that the Fe^II–Fe^II form is the state engaging in activity, although at present we cannot unequivocally exclude whether the Fe^II–Fe^III form is also capable of hydrolysis. The diferrous state accumulates to almost 80–85% under only mild reducing conditions and reacts sluggishly with O₂, i.e., circa two orders of magnitude slower than known diiron oxidases and oxygenases.⁵⁴ These observations suggest that the Fe^II–Fe^II form would be constitutively active under cellular redox conditions and slow to inactivate.

Although the diiron site may be the most biologically relevant and active cofactor in vivo, Cas3 activity will be compromised due to oxidative inactivation in vitro. Hydrolysis can be rescued with reducing agents or cobalt supplementation. It is interesting to note that the observed Cas3 promiscuity with respect to the active metallocofactor is in stark contrast to the majority of characterized HD-domain metalloproteins.³¹ This promiscuity likely alleviates the necessity for a particular metal ion to perform ssDNA cleavage, endowing Cas3 with versatility in response to viral DNA degradation under different environmental niches or stress responses. Increased ionic strength does not impact ssDNAse activity, and Tf Cas3 is able to carry out hydrolysis within a wide range of pH (5–11). However, it is unclear whether this is an outcome of Tf Cas3’s ability to preserve the parent organism’s pH-dependence adaptability or a general property of Cas3 proteins.

Cas3″ proteins contain only the metallohydrolase active site and lack any auxiliary domains. Two crystal structures, one on the Mj Cas3″²¹ and one on the truncated form of Thermus thermophilus Cas3 (containing only the HD domain),²⁵ exhibit nondinuclear active sites. By expressing both Cas3″ in media directing specific Fe incorporation, we obtained spectroscopic evidence for a dimetal redox-active cofactor. The EPR spectra of the Pf Cas3″ and Tt Cas3″ proteins show signals of an Fe^II–Fe^III form that are essentially identical to those of the Tf Cas3. The Mössbauer spectra of the Pf Cas3″ and Tt Cas3″ support formation of diiron sites, albeit with only 60–70% of the Fe in the sample being in dinuclear configuration. The dimetal cofactor is thus shown to be less stable than that in Tf Cas3 (as suggested by the presence of monoferric species), most likely due to increased lability of the protein backbone caused by the absence of the ancillary SF2 domain. Cofactor assembly and activity with other metal ions in Cas3″ were not further pursued, but potentially the Mn- and Co-forms of Cas3″ can provide a nonredox sensitive spectroscopic marker to follow cofactor assembly. The existing crystal structure of Mj Cas3″ may argue against the redox sensitivity of the metal ion being the reason for accumulation of mononuclear species, but this remains to be explored.

A hallmark feature of multinuclear HD-domain PDEs, including Cas3 and Cas3″, is a fifth histidine in the seven amino acid-binding motif. In Tf Cas3, H150 is important but not critical for diiron assembly, as reflected by the twofold decrease in the diiron content. The ⁵⁷Fe Mössbauer spectrum exhibits two sets of quadrupole doublets, confirming that the observed site differentiation in the WT Tf Cas3 is not introduced by the additional His ligand but perhaps reflects different protonation states of a water/hydroxide ligand. ssDNA hydrolysis is severely compromised in the variant, demonstrating that H150 is a key residue in this activity, although nuclease activity is not completely inhibited. In contrast, 2′3′-cAMP hydrolysis is only 3–4 times slower than that of the WT, and this decrease is a consequence of the diminished production of 3′-AMP (but not 2′-AMP). These findings suggest that (i) the dramatic decrease in ssDNA but not 2′3′-cAMP hydrolysis likely stems from incorrect repositioning of the ssDNA, especially after cleavage of the n-1 PDE bond, and (ii) changes in the extent of bifurcation to 3′-AMP (with 2′3′-cAMP as a substrate) imply improper substrate orientation.

Cas3″ was previously reported to hydrolyze 2′3′-cNMPs with 3′-AMP as the only product,²¹ a finding that suggested that crRNAs harboring a 2′,3′-cyclic phosphate terminus may represent potential substrates for Cas3 proteins.¹⁵ In the present study, we show that Tf Cas3, Tt, and Pf Cas3″ all hydrolyze 2′3′-cAMP, yielding 3′-AMP as the major product and 2′-AMP as the minor product. Tf Cas3 exhibits a modest k_cat for 2′3′-cAMP and appears to be more specific for this substrate, when compared to other select proteins of the HD-domain superfamily.

Tf Cas3 exhibits a large K_M for 2′3′-cAMP, which may perhaps argue for the physiological relevance of this substrate for Cas3 nucleases. The measured steady-state parameters, however, are very similar to those of the Sa CvfA with a likewise high K_M. CvfA is an HD-domain protein, reported to perform the Mn-dependent hydrolysis of 2′3′-cAMP to 3′-AMP.³⁴ This activity of CvfA is required for virulence in Sa.^34,40 In our studies, we expressed and purified Sa CvfA and demonstrated that it exhibits low metal ion binding affinity, with Mn and Fe being the two most potent activators. CvfA, similar to Cas3, (i) shows a large degree of promiscuity with respect to the activating metal ions, (ii) has comparable catalytic proficiency with 2′3′-cAMP and forms 2′-AMP as a minor coproduct of 3′-AMP, and (iii) is active with iron, suggesting that Fe is a much more prevalent activity element for HD-domain hydrolases than initially considered.

Previous studies proposed that CvfA likely acts on 2′,3′-cylic phosphodiester bonds at RNA 3′-termini, with which the measured K_M decreased almost by a 10,000-fold (when compared to that of “free” 2′3′-cAMP).³⁴ The current accepted mechanism suggests that CvfA regulates gene expression of virulence factors by controlling RNA degradation.^34,40,42 Therefore, the virulence regulation mediated by CvfA could be related to its PDE activity on the 2′3′-cyclic phosphodiester bond at the 3′-terminal of the target RNA. Following the paradigm of CvfA, the 2′3′-cAMP activity of Tf Cas3 may be associated with emerging studies that invoke a role of Type I CRISPR-Cas systems in virulence. In the reported cases, Cas3 is the key regulator of virulence gene expression and biofilm formation.^36–39 At present, there is an outstanding similarity of the steady-state parameters between Tf Cas3 and Sa CvfA and their purported implication in virulence regulation.^34–40,42 These commonalities also suggest that the large K_M for 2′3′-cAMP exhibited by Tf Cas3 may not be a true determinant for assessing the relevance of this nucleotide in Cas3 cellular activity because we do not yet know the physiological substrate. It could be that, like in the case of CvfA, the 2′3′-phosphodiester linkage is tethered to a crRNA terminus as it has been originally suggested. Existing evidence shows that significant routes exist for ferrous uptake in bacterial pathogens during infection.^59,60 This observation may present a likely link between Fe availability and ability of both CvfA and Cas3 to sustain hydrolysis with diferrous cofactors, allowing them to be “active” at various biological niches. Therefore, involvement of Fe in Cas3 and CvfA activity in tandem with their ability to hydrolyze 2′3′-cAMP presents a possible mechanism for the reported Cas3-mediated virulence and brings forward a possible molecular basis for Cas3 activities going beyond ssDNA degradation.

CONCLUSIONS

Cas3 nucleases are central in viral DNA degradation and key players in CRISPR-Cas genome editing processes. We demonstrate here that the diferrous cofactor state is the most active and likely relevant in cellulo, although we cannot exclude that the mixed-valent Fe^II–Fe^III form is also catalytically competent. We employ Fe-enriched forms to establish that Cas3″, similar to Cas3, harbors dinuclear cofactor architectures. We show that the seventh histidine ligand that is present only in HD-domain PDEs is important for processive ssDNA cleavage but less so for cofactor stability. We expressed and purified CvfA, and we suggest that it has a dinuclear cofactor, similar to Cas3, albeit with a lower metal binding affinity. The ability of Cas3 to linearize 2′3′-cAMP is also explored; Cas3 hydrolyzes 2′3′-cAMP with observed rates comparable to those of the conserved virulence factor CvfA, arguing for 2′3′-cAMP hydrolysis as a possible operant pathway in Cas3-mediated virulence.