Research organism

elife

eLife

2050-084X

eLife Sciences Publications, Ltd

68339

10.7554/eLife.68339

Research Article

Microbiology and Infectious Disease

A chimeric nuclease substitutes a phage CRISPR-Cas system to provide sequence-specific immunity against subviral parasites

Barth

Zachary K

https://orcid.org/0000-0002-1321-306X

1†Nguyen

Maria HT

https://orcid.org/0000-0003-2441-4948

1Seed

Kimberley D

https://orcid.org/0000-0002-0139-1600

kseed@berkeley.edu121Department of Plant and Microbial Biology, University of California, Berkeley

Berkeley

United States2Chan Zuckerberg Biohub

San Francisco

United States

Wiedenheft

Blake

Reviewing EditorMontana State UniversityUnited StatesStorz

Gisela

Senior EditorNational Institute of Child Health and Human DevelopmentUnited States

†

Department of Microbiology, Cornell University, Ithaca, United States

07072021

2021

e68339

1203202127062021

2021

Barth et al

http://creativecommons.org/licenses/by/4.0/

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Mobile genetic elements, elements that can move horizontally between genomes, have profound effects on their host's fitness. The phage-inducible chromosomal island-like element (PLE) is a mobile element that integrates into the chromosome of Vibrio cholerae and parasitizes the bacteriophage ICP1 to move between cells. This parasitism by PLE is such that it abolishes the production of ICP1 progeny and provides a defensive boon to the host cell population. In response to the severe parasitism imposed by PLE, ICP1 has acquired an adaptive CRISPR-Cas system that targets the PLE genome during infection. However, ICP1 isolates that naturally lack CRISPR-Cas are still able to overcome certain PLE variants, and the mechanism of this immunity against PLE has thus far remained unknown. Here, we show that ICP1 isolates that lack CRISPR-Cas encode an endonuclease in the same locus, and that the endonuclease provides ICP1 with immunity to a subset of PLEs. Further analysis shows that this endonuclease is of chimeric origin, incorporating a DNA-binding domain that is highly similar to some PLE replication origin-binding proteins. This similarity allows the endonuclease to bind and cleave PLE origins of replication. The endonuclease appears to exert considerable selective pressure on PLEs and may drive PLE replication module swapping and origin restructuring as mechanisms of escape. This work demonstrates that new genome defense systems can arise through domain shuffling and provides a greater understanding of the evolutionary forces driving genome modularity and temporal succession in mobile elements.

bacteriophageVibrio choleraemobile elementsatellite

Research organismOther

http://dx.doi.org/10.13039/100000060

National Institute of Allergy and Infectious Diseases

R01AI127652

Seed

Kimberley D

http://dx.doi.org/10.13039/100000060

National Institute of Allergy and Infectious Diseases

R01AI153303

Seed

Kimberley D

http://dx.doi.org/10.13039/100000861

Burroughs Wellcome Fund

1019213

Seed

Kimberley D

http://dx.doi.org/10.13039/100014989

Chan Zuckerberg Initiative

Seed

Kimberley D

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Author impact statement

Horizontal transfer of a sequence-specific DNA-binding domain allows a virus to destroy its subviral parasite and overcome parasite-mediated restriction.

Introduction

Mobile genetic elements (MGEs), genetic units capable of spreading within and between genomes, are key mediators of evolution. MGEs differ vastly in size and complexity. At one end of the spectrum are homing endonucleases (HEGs). These single-gene MGEs occur in diploid loci and cleave cognate alleles so that their coding sequences can serve as templates for recombinational repair (Stoddard, 2011). At another extreme, integrative viruses are complex MGEs that can encode their own replication genes as well as structural components for their own dispersal (Krupovic et al., 2019), and even cargo genes dedicated to boosting the fitness of their cellular hosts (Harrison and Brockhurst, 2017). While the evolutionary importance of MGEs across domains of life is clear, apart from a handful of exceptions (Greenwood et al., 2018), it is not possible to study the spread of MGEs in real time in populations of multicellular organisms. In contrast, short generation times and low barriers to horizontal gene transfer make bacteria ideal organisms for studying how MGEs shape the evolution of their hosts.

Recent work has led to a deeper appreciation of the extensive entanglement between MGEs and genome defense functions. MGEs frequently encode defense modules to prevent infection by viruses or other nonvirus MGEs (Koonin et al., 2020). Such modules include toxin-antitoxin (TA) systems, restriction modification (RM) systems, CRISPR-Cas, and numerous other systems recently described (Doron et al., 2018; Gao et al., 2020; Koonin and Makarova, 2019; Marraffini, 2015; Millman et al., 2020; Mruk and Kobayashi, 2014). Beyond their antiviral and anti-MGE functions, defense systems also serve the selfish needs of the MGEs that encode them, and the constituents of defense modules can be recruited to benefit viruses and nonvirus MGEs by serving counter-defense functions. Viruses may encode antitoxin or DNA modification genes as a means of escaping TA and RM systems endogenous to their hosts (Loenen and Raleigh, 2014; Otsuka and Yonesaki, 2012). While many defense modules eliminate invading viruses and MGEs through nucleolytic attack, many phages use nucleases to degrade the host genome, preventing further expression of host defenses (McKitterick et al., 2019a; Panayotatos and Fontaine, 1985; Parson and Snustad, 1975; Souther et al., 1972; Warner et al., 1975).

The flow of MGEs and their defense systems between viruses and hosts, as well as the retooling of genes for defense, counter-defense, and MGE maintenance or dispersal functions, has been described using the model ‘guns for hire’ to reflect the mercenary nature of these defense systems and the selfish MGEs that carry them (Koonin et al., 2020). One of the most compelling examples of host-pathogen conflicts that conforms to the ‘guns for hire’ framework occurs in Vibrio cholerae between the bacteriophage ICP1 and its own parasite, the phage-inducible chromosomal island-like element (PLE) (Seed et al., 2013). Upon infection by ICP1, PLE excises from the host chromosome, replicates to high copy (O'Hara et al., 2017), and is assembled into transducing particles to spread the PLE genome to new cells (Netter et al., 2021). PLE excision and DNA replication both require ICP1-encoded gene products (Barth et al., 2020b; McKitterick et al., 2019a; McKitterick and Seed, 2018). Similarly, multiple lines of evidence including shared host cell receptors (O'Hara et al., 2017), PLE genome analysis, and electron microscopy (Netter et al., 2021) strongly suggest that PLE is packaged into remodeled ICP1 virions for mobilization. While infected PLE(+) cells still die, no ICP1 virions are produced when PLE activity is unimpeded (O'Hara et al., 2017). Thus, PLE prevents further spread of ICP1 and protects the host cell population. In this way, PLE acts as both a selfish parasite of ICP1 and an effective abortive infection defense system for V. cholerae.

True to the ‘guns for hire’ model, ICP1 has co-opted a genome defense system to protect itself from PLE. Many ICP1 isolates encode a CRISPR-Cas system that can destroy PLE within the infected cell and restore ICP1 reproduction (Seed et al., 2013; Figure 1A). CRISPR-Cas systems are typically adaptive immune systems that provide immunological memory against specific nucleic acid sequences (Barrangou et al., 2007; Marraffini, 2015). The memory function of CRISPR-Cas is achieved through the integration of ‘spacers,’ short DNA sequences derived from viruses or MGEs, that are integrated into an array of spacer repeats. The spacer can then be transcribed to serve as an RNA guide that directs nucleolytic machinery against complementary sequence. In this way, acquisition of a small portion of foreign DNA provides the specificity required for defense.

Figure 1.Some ICP1 isolates encode a free-standing nuclease in place of CRISPR-Cas.

(A) A model of ICP1 interference of phage-inducible chromosomal island-like elements (PLEs) via CRISPR. When ICP1 infects a PLE(+) V. cholerae cell, ICP1 is able to overcome PLE restriction and reproduce if it possesses a CRISPR-Cas system with complementary spacers to the PLE. Cas and CR refer to the CRISPR associated genes and CRISPR array respectively. (B) Schematics of the region between gp87 and gp91 as it appears in ICP1²⁰⁰¹ (top) and ICP1²⁰⁰⁶ (bottom). Genes represented by black arrows are conserved in all ICP1 isolates, while genes represented with gray arrows covary with gp88 or CRISPR-Cas. (C) An alignment between the T5orf172 domain of Gp88 and the GIY-YIG domains of several structurally resolved endonucleases. Secondary structure for Gp88 was predicted using HHPRED (Zimmermann et al., 2018). Alpha helices are shown in yellow shading, and beta strands are shown in blue shading. Key residues of the GIY-YIG motif are bolded. We included an atypical GIY-YIG endonuclease domain from a chloroplast-encoded glutoredoxin atGRXs16 to demonstrate the potential for alternative residues at core motif positions. A conserved glutamate that was previously found to be required for catalysis in I-TevI is denoted by an asterisk (and corresponds to the E180A mutation in Gp88 in subsequent experiments).

Figure 1—figure supplement 1.ICP1<sup>2005</sup> has an atypical CRISPR-Cas arrangement.

Genetic organization of the region between gp87 and gp91 in different ICP1 isolates (ICP1²⁰⁰¹[top], ICP1²⁰⁰⁶[bottom], ICP1²⁰⁰⁵[middle]). Other than ICP1²⁰⁰⁵, all sequenced ICP1 isolates (Angermeyer et al., 2018) share their organization with ICP1²⁰⁰¹ or ICP1²⁰⁰⁶.

Reflecting the primary role of ICP1’s CRISPR-Cas as an anti-PLE system, almost all spacers associated with the system are PLE derived (Seed et al., 2013; McKitterick et al., 2019b). Like cellular CRISPR-Cas systems, the ICP1 system can acquire new immunological memory, reflecting that PLE is not a single static genome but that a number of PLE variants exist. To date, five PLE variants, numbered 1–5, have been described, occurring in about ~15% of sequenced epidemic V. cholerae genomes. There is a pattern of temporal succession, where one PLE will dominate in sequenced genomes for a time before being supplanted by another PLE (O'Hara et al., 2017), but the reemergence of old PLE sequence in new PLE variants suggests that unsampled reservoirs exist in nature. Like PLEs, there is also diversity among ICP1 genomes. Not all ICP1 isolates encode CRISPR-Cas, but this does not mean that they are defenseless against PLEs. Previous work found that an ICP1 variant that naturally lacked CRISPR-Cas was able to reproduce on the two oldest PLE variants, PLE5 and PLE4, as well as the most recent variant PLE1 (O'Hara et al., 2017). Much like PLE variants, there appears to be some temporal succession in the presence or absence of ICP1’s CRISPR-Cas system. A minority of ICP1 isolates collected between 2001 and 2011 possessed CRISPR-Cas systems (Angermeyer et al., 2018), while CRISPR-Cas encoding ICP1 predominated between 2011 and 2017 (McKitterick et al., 2019b). As PLE and ICP1 have coevolved specific mechanisms of parasitism and counter-defenses, it is worth exploring if the temporal succession of PLE and ICP1 variants could be in response to selective pressures that the two entities exert on each other.

Intrigued by CRISPR-independent interference of PLE and hoping to gain insight into patterns of ICP1 and PLE variant succession, we set out to identify the mechanism of PLE interference in ICP1 isolates that lack CRISPR-Cas. Surprisingly, we found that all natural ICP1 isolates that do not encode CRISPR-Cas instead encode an endonuclease in the same genomic locus that is necessary for propagation on V. cholerae strains containing PLEs 1, 4, or 5. Lending further support to the ‘guns for hire’ model, we find that this anti-PLE nuclease is of chimeric origin, being partially derived from a PLE-encoded DNA-binding domain while its nucleolytic domain appears to be derived from an ICP1-encoded family of putative HEGs. Harnessing the rich evolutionary interplay of PLE and ICP1, this work shows that domain shuffling between hostile genomes can allow for new forms of antagonism, and that phage-encoded HEGs can be repurposed for antiparasite functions. Additionally, this work reveals key mediators of ICP1-PLE host range that inform observed patterns of PLE temporal succession and modularity, broadening our understanding of subcellular host-parasite coevolution.

ResultsA subset of ICP1 isolates deploy a stand-alone nuclease instead of CRISPR-Cas to counter PLE

We set out to identify which gene(s) determined host range in ICP1 isolates that lack CRISPR-Cas. It has long been recognized that phages are mosaic entities composed of functional gene neighborhoods, and syntenic neighborhoods of divergent sequence may fulfill analogous functions (Brüssow and Hendrix, 2002). Previous work suggests that ICP1 conforms to these general patterns of phage genome structure. Transcriptomics and bioinformatic predictions show that ICP1 genes with related biological functions are organized together in the genome and expressed at the same time, demonstrating the presence of gene neighborhoods (Barth et al., 2020a). Additionally, while the ICP1 genome is highly conserved between isolates and does not display large-scale rearrangements (Angermeyer et al., 2018), there is indication that nonhomologous sequence can serve analogous functions. ICP1 isolates encode one of two alternative SF1B-type helicases thought to be of shared function (McKitterick et al., 2019a), suggesting that ICP1 isolates can use alternative genes to fulfill the same adaptational requirement. We reasoned that such genome organization and mosaicism warranted a ‘guilt by location’ approach to investigating gene function, and that the locus syntenic to CRISPR-Cas in those ICP1 isolates that lack CRISPR might hold clues as to how they overcome PLEs.

In isolates without CRISPR-Cas, the locus is replaced with a single open reading frame, designated gp88 for its location in ICP1²⁰⁰¹, the original sequenced ICP1 isolate (Seed et al., 2011; Figure 1B). The two coding sequences immediately upstream of the CRISPR-Cas system and oriented divergently from the system are also replaced in the phage with gp88. These genes generally covary with the presence of the CRISPR-Cas system or gp88. Most ICP1 CRISPR-Cas systems are adjacent to a phage regulatory protein Rha domain (pfam09669) encoding gene. A gene encoding a Bro-N domain (pfam02498) and a KilAC domain (pfam03374) occurs adjacent to gp88. Their positions and putative annotations suggest that these divergently transcribed genes may have a regulatory function. In one sequenced ICP1 isolate with a functional CRISPR-Cas system (O'Hara et al., 2017), the bro_N domain coding gene and its partner are found instead of the rha-like gene containing pair, suggesting that these pairs are redundant in function or not involved in CRISPR-Cas activity (Figure 1—figure supplement 1).

Further analysis of gp88 revealed that it encodes a T5orf172 domain (pfam10544) containing protein. This domain is a member of the GIY-YIG endonuclease domain superfamily, suggesting that Gp88 may be a nuclease. We aligned the Gp88 T5orf172 domain with GIY-YIG domains from endonucleases that had been biochemically characterized and structurally resolved (Liu et al., 2013; Swapna et al., 2005; Truglio et al., 2005; Van Roey et al., 2002; Figure 1C). Unlike most identified GIY-YIG nucleases, Gp88 lacks a conserved histidine or tyrosine in the first alpha helix of the GIY-YIG domain. However, Gp88 retains conservation of the catalytic arginine and glutamate. The relatively high motif conservation found in Gp88 suggests that the protein possesses endonucleolytic activity. Given that nucleases are a particularly prominent class of proteins engaged in conflicts between hosts, mobile elements and viruses (Koonin et al., 2020), and gp88’s syntenic location to ICP1’s CRISPR-Cas system, we hypothesized that Gp88 interferes with PLE activity to protect ICP1 isolates that naturally lack CRISPR-Cas.

To test our hypothesis, we generated ICP1²⁰⁰¹ mutants with either an in-frame deletion of gp88 or harboring a single amino acid substitution (E180A) predicted to abolish Gp88’s nucleolytic activity (Figure 1C). We also used ICP1²⁰⁰⁶ and a ∆CRISPR derivative to serve as controls for host susceptibility. As expected, the PLE (-) V. cholerae strain was susceptible to all ICP1 variants, and the ICP1²⁰⁰⁶ ∆CRISPR variant was restricted by all PLEs (Figure 2A). CRISPR(+) ICP1²⁰⁰⁶ was able to propagate on all strains except the one containing PLE3 as ICP1²⁰⁰⁶ does not have a matching spacer against PLE3. ICP1²⁰⁰¹ was able to propagate on PLEs 1, 4, and 5 (Figure 2A), but it was restricted on PLEs 2 and 3. Unlike the wild-type (WT) variant, the ∆gp88 and gp88* mutants were unable to propagate on V. cholerae with PLEs 1, 4, or 5 (Figure 2A), indicating that catalytically active gp88 is necessary for overcoming these PLEs in phages naturally lacking CRISPR-Cas. Unsurprisingly, the gp88 mutants retained sensitivity to restriction by PLEs 2 and 3 (Figure 2A).

Figure 2.The alternative nuclease Gp88 controls ICP1 host range in a natural isolate that lacks CRISPR-Cas.

(A) Tenfold dilutions of the phage isolate or mutant derivative indicated spotted on V. cholerae with the PLE indicated (bacterial lawns in gray, zones of killing are shown in black). Gp88* possess a single amino acid substitution (E180A) predicted to abolish nuclease activity. Spot assays were performed in biological triplicate, and a single representative image is shown. Replicate spot assays are shown in Figure 2—figure supplement 1 and Figure 2—figure supplement 2. (B) Replication of PLE1 and PLE2 in V. cholerae host strains calculated as the fold change in PLE DNA copy 20 minutes post infection with the ICP1 variant indicated.

Figure 2—source data 1.Values for the graph in <xref ref-type="fig" rid="fig2">Figure 2B</xref>.

Figure 2—figure supplement 1.Biological replicate of spot assays.

Tenfold dilutions of the phage isolate or mutant derivative indicated spotted on V. cholerae with the PLE indicated (bacterial lawns in gray, zones of killing are shown in black). This figure and Figure 2—figure supplement 2 each represent one biological replicate, while a third replicate is divided across Figure 2B, Figure 4B, and Figure 5B. Strain backgrounds appear in the same order that they first appear in the main text. For each biological replicate, spot assays were performed in parallel for the different strains.

Figure 2—figure supplement 2.Biological replicate of spot assays.

Tenfold dilutions of the phage isolate or mutant derivative indicated spotted on V. cholerae with the PLE indicated (bacterial lawns in gray, zones of killing are shown in black). This figure and Figure 2—figure supplement 1 each represent one biological replicate, while a third replicate is divided across Figure 2B, Figure 4B, and Figure 5B. Strain backgrounds appear in the same order that they first appear in the main text. For each biological replicate, spot assays were performed in parallel for the different strains.

As CRISPR targeting was previously observed to diminish PLE replication that occurs during ICP1 infection (McKitterick et al., 2019b), we tested whether the presence of Gp88 could impact PLE replication during infection. We observed that PLE1 is unable to replicate in the presence of Gp88 encoding ICP1, and replication is restored during infection with the Gp88 knockout or catalytically inactive mutant phages (Figure 2B). Consistent with endonucleolytic activity, the PLE1 copy decreases following infection by Gp88 encoding phage. This is notable given that previous work has shown that multiple PLE matched spacers are required for CRISPR-Cas to completely abolish PLE replication during infection (McKitterick et al., 2019b). PLE2 replication is unaffected by the presence or absence of Gp88, consistent with Gp88 not providing ICP1 with immunity against PLE2 (Figure 2A).

PLE replicons are modular

Having identified gp88’s role in preventing PLE restriction of ICP1, we next sought to determine how Gp88 was recognizing PLE, and why it did not confer protection against PLEs 2 and 3. We reasoned that PLEs 2 and 3 most likely lacked sequence targeted by Gp88 or encoded an inhibitor of Gp88’s activity. To explore these possibilities, we compared the PLE genomes looking for nucleotide sequence that was uniquely present or uniquely absent in PLEs 2 and 3.

Strikingly, only two stretches of sequence met these criteria, and both had been previously implicated in PLE replication (Barth et al., 2020b). The repA gene encoding the replication initiation factor, and the intergenic region containing the PLE origin of replication (ori) to which RepA binds covaried, with the PLE1, 4, and 5 sequences clustering together as one group, and the PLE2 and 3 sequences clustering as another (Figure 3A). More specifically, it was the DNA-binding RepA_N domain of RepA that covaried with the origin, while the C-terminal domain, hypothesized to facilitate replisome recruitment (Barth et al., 2020b), was conserved across all PLEs (Figure 3A).

Figure 3.<italic>P</italic>hage-inducible chromosomal island-<italic>l</italic>ike <italic>e</italic>lement (PLE) replicons are modular and are composed of a compatible RepA initiation factor and origin of replication (ori).

(A) Genomic organization of PLE1 with insets corresponding to the PLE noncoding region and repA. Previously characterized PLE genes are labeled. Insets are Mauve alignments showing sequence conservation of the denoted loci across the different PLEs. Shared color denotes sequence conservation, with the height of the histogram representing nucleotide sequence identity. The susceptibility of each PLE(+) V. cholerae host to plaquing by phage encoding Gp88 is indicated. (B) Replication of hybrid PLEs in V. cholerae calculated as the fold change in PLE DNA copy 20 minutes post infection with ICP1²⁰⁰⁶ ∆CRISPR. Strains with PLE1 ∆repA (possessing the native ori^PLE1 or ∆ori::ori^PLE2) or PLE2 ∆repA (possessing the native ori^PLE2 or ∆ori::ori^PLE1) were complemented with a vector expressing the repA gene from PLE1 or PLE2. The backbone, identity of the ori, and RepA variant are indicated as being from PLE1 or PLE2.

Figure 3—source data 1.Values for the graph in <xref ref-type="fig" rid="fig3">Figure 3B</xref>.

Figure 3—figure supplement 1.Alignment of <italic>repA</italic> from <italic>p</italic>haPLEs 1–5 encoding the conserved C-terminus.

The PLE1 sequence and identical nucleotides in the other PLE variants are shaded blue. Nucleotides that differ from PLE1 but co-occur in at least two PLE variants are shaded yellow. For positions where PLE1 is unique and both PLE2 and PLE3, and PLE4 and PLE5 have separate consensuses, the PLE2 and PLE3 consensus is bolded and unshaded, and the PLE4 and PLE5 consensus is underlined and unshaded. Nucleotides with no aligned matches are unshaded normal text.

Figure 3—figure supplement 2.PLE2 requires RepA and a cognate origin of replication to replicate.

Replication of PLE2 in V. cholerae calculated as the fold change in PLE DNA copy 20 minutes post infection with ICP1²⁰⁰⁶ ∆CRISPR.

Figure 3—figure supplement 2—source data 1.Values for the graph in <xref ref-type="fig" rid="fig3s2">Figure 3—figure supplement 2</xref>.

Previously, we found that during ICP1 infection, ectopic expression of RepA was sufficient to drive replication of a synthetic ‘midiPLE’ construct. The midiPLE consists of the PLE attachment sites, the PLE integrase, and the noncoding region that bears the origin of replication (Barth et al., 2020b). Additionally, midiPLE replication did not occur without RepA, and the PLE integrase was shown to be dispensable for PLE replication (Barth et al., 2020b). These data suggest that the minimal components of the PLE replicon are the replication origin and RepA, the two components that covaried across PLEs. Alignment of the conserved 3′ ends of PLE repA genes suggests that RepA specificity swapping has occurred multiple times (Figure 3A, Figure 3—figure supplement 1). Despite the PLE1 RepA_N domain clustering with the PLE4 and PLE5 variants, the PLE1 C-terminal sequence is more similar to PLE2 (98.88% identical over the last 178 bp) than PLE5 (93.25% identical over the last 178 bp). PLE5 and PLE3 are 99.44% identical over the same region, while the PLE4 C-terminal region is the most diverged from other PLEs (Figure 3—figure supplement 1). This cross-clustering of repA ends would only be expected to occur after multiple gene recombination events, suggesting that PLE replisome module swapping occurred at least twice, and may be an important part of PLE evolution.

The putative modularity of the PLE origins and RepA_N domains covaried with susceptibility to ICP1²⁰⁰¹ (Figure 3A), leading us to hypothesize that one of the replicon modules but not the other was susceptible to Gp88-mediated interference. Before testing this hypothesis directly, we wanted to confirm that the covariation of PLE replication origins and RepA_N domains truly reflects modularity of the PLE replicon. We first tested whether the putative PLE2 origin and repA gene were necessary for PLE2 replication and found that deletion of either component abolished replication following infection by ICP1²⁰⁰⁶ ∆CRISPR (Figure 3—figure supplement 2), as was previously observed for PLE1 (Barth et al., 2020b).

Having confirmed that the PLE2 variant replicon components are necessary for replication, we then sought to demonstrate specificity of the RepA variants to their cognate origin of replication. We generated chimeric ‘origin-swapped’ ∆repA PLEs for PLEs 1 and 2 (Figure 3B), and ectopically expressed each RepA variant in the different PLE backgrounds during phage infection. As expected, PLE replication only occurred when cognate origins and repA alleles co-occurred, revealing that the two components of the PLE replicon function together as a module, irrespective of which PLE backbone they are encoded in.

Gp88 is an origin-directed nuclease

Having established the specificity between RepA variants and their cognate origins of replication and recognizing that sensitivity to Gp88 covaried with replicon type, we took a closer look at Gp88 to decipher how it might interface with the PLE replication module. Remarkably, Gp88’s own N-terminal domain is 42% identical and 61% sequence similar across 93% of PLE1’s RepA_N domain (Figure 4A). This was surprising as Gp88’s T5orf172 domain is similar to those of several putative HEGs within the ICP1 genome (Figure 4—figure supplement 1). The high similarity of Gp88’s N-terminal portion to some PLE-encoded RepA alleles and the C-terminal portions similarity to putative HEGs suggest that gp88 may have arisen as a chimeric hybrid of PLE and ICP1 coding sequences. Additionally, the similarity of Gp88’s N-terminal region to PLEs 1, 4, and 5 RepA DNA-binding domains suggested that Gp88 might bind to the replication origins of PLEs 1, 4, and 5 and cleave at or proximal to that site.

Figure 4.Gp88 is a PLE replication origin-directed nuclease.

(A) Sequence alignment of the N-terminal portion of Gp88 with the RepA_N domain from PLE1 RepA. Identical residues are denoted with a ‘*.’ Strong residue similarity is denoted by ‘:’, and weak similarity is denoted by ‘•.’ (B, C) Tenfold dilutions of the phage isolate or mutant derivative indicated spotted on V. cholerae with the PLE indicated (bacterial lawns in gray, zones of killing are shown in black). Spot assays were performed in parallel with those in Figure 2, and images labeled with the same PLE background are the same image. Spot assays were performed in biological triplicate, and a single representative image is shown. Biological replicates are shown in Figure 2—figure supplement 1. Gp88* possess a single amino acid substitution (E180A) predicted to abolish nuclease activity. (B) shows phage susceptibility of V. cholerae with PLE1, PLE4, and PLE5 ∆ori derivatives as compared to a strain without PLE. (C) shows phage susceptibility for V. cholerae with PLE2 ∆ori and PLE2 ∆ori::ori^PLE1. (D) Nuclease assay showing the integrity of a PCR product amplified from the noncoding region containing the ori from the PLE variant indicated (numbers) treated with (+) and without (–) 500 nM of purified Gp88. Nuclease assays were performed in triplicate, replicates are presented in Figure 4—figure supplement 3. (E) Nuclease assay showing the integrity of a PCR product amplified from the noncoding region containing the ori from the PLE variant indicated (numbers) treated with (+) and without (–) 500 nM of purified Gp88*. Nuclease assays were performed in triplicate, replicates are presented in Figure 4—figure supplement 4.

Figure 4—source data 1.Original uncropped gels for <xref ref-type="fig" rid="fig4">Figure 4D</xref> (top) and <xref ref-type="fig" rid="fig4">Figure 4E</xref> (bottom).

The cropped images shown in the figure are indicated by the orange boxes.

Figure 4—figure supplement 1.The nuclease domain of Gp88 is similar to those of putative homing endonucleases in ICP1.

(A) A multiple sequence alignment between the C-terminal portion of Gp88 and several homologous sequences found in putative homing endonucleases in the ICP1 genome. Identical residues are denoted with a ‘*.’ Strong residue similarity is denoted by ‘:,’ and weak similarity is denoted by ‘•.’ (B) A phylogenetic tree constructed from the multiple sequence alignment in (A).

Figure 4—figure supplement 2.Protein preparations of Gp88 (Odn) and Gp88* (Odn*) used for in vitro assays.

Protein preparations were separated by SDS-PAGE and visualized by Coomassie staining. The marker (M) is indicated; the predicted molecular weight of untagged Gp88 is 26 kDa. (A) Gel showing wild-type Gp88 protein used for all cleavage assays except for those depicted in Figure 4—figure supplement 4 and Figure 6. (B) Gel showing wild-type Gp88 protein used for cleavage assays depicted in Figure 4—figure supplement 4 and Figure 6. (C) Gel showing Gp88* protein used for all in vitro assays that utilized Gp88*. Note that protein preparations (A) and (C) were performed in parallel; the preparation in (B) was completed separately.

Figure 4—figure supplement 2—source data 1.Original uncropped gels for <xref ref-type="fig" rid="fig4s2">Figure 4—figure supplement 2</xref>.

The cropped images shown in the figure are indicated by the orange boxes.

Figure 4—figure supplement 3.Replicates of Gp88 nuclease assay.

Two replicates of a nuclease assay showing the integrity of a PCR product amplified from the noncoding region containing the ori from the PLE variant indicated (numbers) treated with (+) and without (–) 500 nM of purified Gp88. A third replicate is depicted in Figure 4D.

Figure 4—figure supplement 3—source data 1.Original uncropped gels for <xref ref-type="fig" rid="fig4s3">Figure 4—figure supplement 3</xref>.

The cropped images shown in the figure are indicated by the orange boxes.

Figure 4—figure supplement 4.Replicates of Gp88* nuclease assay.

Figure 4—figure supplement 4—source data 1.Original uncropped gels for <xref ref-type="fig" rid="fig4s4">Figure 4—figure supplement 4</xref>.

The cropped images shown in the figure are indicated by the orange boxes.

Figure 4—figure supplement 5.Cleavage of an altered PLE4 probe.

Two replicates of a nuclease assay showing the integrity of a PCR product amplified from the noncoding region containing the ori from the PLE1 or PLE4 treated with (+) and without (–) 500 nM of purified Gp88. The PLE4 probe used in this experiment is amplified with a different set of primers compared to the probe used in Figure 4D such that cleavage of the probe at or near the iterons would produce products that could be differentiated on the basis of size.

Figure 4—figure supplement 5—source data 1.Original uncropped gels for <xref ref-type="fig" rid="fig4s5">Figure 4—figure supplement 5</xref>.

The cropped images shown in the figure are indicated by the orange boxes.

To evaluate this hypothesis, we next wanted to test whether the PLE origin of replication was a necessary component for Gp88 activity. Previously, it was shown that loss of replication partially attenuated PLE1-mediated restriction of ICP1 but nonreplicating PLE1 mutants were still broadly restrictive to ICP1 ∆CRISPR-Cas (Barth et al., 2020b). We hypothesized that PLE could escape Gp88 targeting through deletion of the PLE origin and thus block propagation of Gp88 encoding phage. We tested this by deleting the entire conserved stretch of sequence that contained the origin of replication in PLEs 1, 4, and 5. In support of our hypothesis, these mutants regained restrictive activity against Gp88 encoding phage (Figure 4B). Conversely, cloning a Gp88 recognized replication origin into PLEs that are insensitive to Gp88 should sensitize them to Gp88 activity. To test this, we infected our ‘ori-swapped’ PLE2 strain (Figure 3B) with Gp88 encoding phage. We included a PLE2 ∆ori strain to control for the possibility that loss of replication would abolish PLE2-mediated restriction of ICP1. The PLE2 ∆ori strain retained the ability to restrict all isolates of ICP1²⁰⁰¹ (Figure 4C). In contrast, the PLE2 strain bearing the PLE1 origin sequence was no longer restrictive to ICP1²⁰⁰¹, but still restricted variants that lacked Gp88 activity (Figure 4C), confirming that the presence of the replication origin sequence mediated sensitivity to Gp88.

We wanted to confirm that Gp88-mediated interference manifested through nucleolytic cleavage of PLE. To determine if Gp88 was truly acting as a nuclease, we purified Gp88 (Figure 4—figure supplement 2) and performed in vitro nuclease activity assays. Consistent with the host range of ICP1 encoding Gp88, we found that the purified Gp88 protein cut PCR products amplified from the region containing the origin of replication from PLEs 1, 4, and 5, but did not cut those of PLEs 2 and 3 (Figure 4D), confirming that Gp88 disrupts PLE through nuclease activity. Supporting this interpretation, the E180A Gp88* mutant that was inactive against PLEs 1, 4, and 5 in vivo, and predicted to be catalytically inert, did not cleave PCR products amplified from PLEs 1, 4, and 5 (Figure 4E), further linking the in vivo activity of Gp88 to its capacity to cleave in vitro. Curiously, Gp88 produced only one cleavage product from the PLE4 probe. We reasoned that since the PLE noncoding regions are diverse and the PLE4 origin of replication was located near to the center of the PLE4 probe, Gp88 cleavage of the PLE4 probe might produce two bands of indistinguishable size. To check this, we produced a new PLE4 probe with the origin of replication offset from the middle and found that Gp88 produced two visible bands (Figure 4—figure supplement 5). In light of these results, we renamed Gp88 the origin-directed nuclease or Odn.

Odn requires iterons to cleave the PLE origin of replication

Our results so far suggested a model where Odn mimics the specificity of the PLE1, 4, and 5 RepA proteins to bind and cut at their cognate origins of replication. Previously, PLE1 RepA was found to bind specifically to a set of iterons, a series of three ~30 bp semi-palindromic repeats in the PLE1 origin of replication (Barth et al., 2020b). If Odn specificity truly mimicked that of RepA, then it should require the iteron sequence for cutting. We tested this in vitro by titrating increasing concentrations of Odn in a nuclease assay with the WT PLE1 origin of replication, as well as the same substrate except with the iterons deleted. In support of our model, we found that Odn does require the iteron sequence for cleavage (Figure 5A). Consistent with iterons being necessary for Odn-mediated in vitro cleavage of the PLE origin, ICP1²⁰⁰¹ infection was restricted by a PLE1 strain harboring the same iteron deletion (Figure 5B). Together, these results strongly support that Odn has DNA-binding specificity that mimics that of the replication initiation factor of some PLEs.

Figure 5.ICP1-encoded Odn (Gp88) requires the PLE iterons for cleavage.

(A) Nuclease assay showing the integrity of a PCR product amplified from the noncoding region containing the ori from wild-type (WT) PLE1 and the ∆iteron mutant, with purified Odn (31.25–500 nM) titrated in. 500 nM catalytically inactive Odn (Odn*) with a single amino acid substitution (E180A) with the WT PLE1 sequence was also included (far left). Nuclease assays were performed in triplicate and replicates are presented in Figure 5—figure supplement 1. (B) Tenfold dilutions of the phage isolate or mutant derivative indicated spotted on V. cholerae with the PLE indicated (bacterial lawns in gray, zones of killing are shown in black). Spot assays were performed in parallel with those in Figures 2 and 4, and images labeled with the same PLE background are the same image. Spot assays were performed in biological triplicate, and a single representative image is shown. Replicate assays are shown in Figure 2—figure supplement 1.

Figure 5—source data 1.Original uncropped gel for <xref ref-type="fig" rid="fig5">Figure 5A</xref>.

The cropped image shown in the figure is indicated by the orange box.

Figure 5—figure supplement 1.Replicates of nuclease assays.

Nuclease assays showing the integrity of a PCR product amplified from the noncoding region containing the ori from wild-type (WT) PLE1 and the ∆iteron mutant, with purified Odn (31.25–500 nM) titrated in. 500 nM catalytically inactive Odn (Odn*) with a single amino acid substitution (E180A) with the WT PLE1 sequence was also included. This assay was performed in triplicate, and two replicates are shown here. The third replicate is shown in Figure 5A.

Figure 5—figure supplement 1—source data 1.Original uncropped gels for <xref ref-type="fig" rid="fig5s1">Figure 5—figure supplement 1</xref>.

The cropped images shown in the figure are indicated by the orange boxes.

PLE mutations lead to escape from Odn

Odn activity against the origin, as well as the pattern of cross-clustering at the N and C termini of RepA, suggested that Odn may impose substantial evolutionary pressure on PLE replication modules. Since swapping the PLE origin and cognate RepA_N domain could abolish Odn targeting of PLE, it appears likely that Odn selected for the multiple domain shuffling events in PLE RepA inferred by comparison of PLE genomes (Figure 3—figure supplement 1). Because PLE replication is necessary for both PLE mobility and complete restriction of ICP1 (Barth et al., 2020b; McKitterick et al., 2019a), simple deletions of the replication origin would not likely be favored as long-term solutions to evading recognition and subsequent cleavage by Odn.

While the swapping and diversification of certain sequences can be traced through the five PLEs, each PLE variant is remarkably conserved. All members of each variant have been found to be 100% nucleotide identical in previously published data sets (McKitterick et al., 2019b; O'Hara et al., 2017). However, we found a single instance of diversity in PLE1 within a lineage of V. cholerae isolated from Pakistan. This lineage was represented in five sequenced strains (biosample accession numbers SAMN08979118, SAMN08979175, SAMN08979185, SAMN08979188, and SAMN08979253). In these five strains, we discovered variation in a 67 bp stretch covering the iterons that results in several nucleotide changes. Within the first iteron, there is an A to T transversion, and starting at that transversion, the next 42 bp are duplicated and replace the sequence that is normally downstream (Figure 6A). This change maintains the presence of the three iterons, and even reverts a few variant bases to ones in the PLE4 and 5 iterons (Figure 6B). Notably, these changes are the only sequence differences between these PLE variants and all other PLE1 isolates, aside from a 2 bp extension of an 11 bp polyA tract that also occurs in these atypical PLE1 variants.

Figure 6.Mutations in PLE1 present in <italic>V. cholerae</italic> isolates from Pakistan render the PLE1 ori resistant to Odn-mediated cleavage.

(A) The PLE1^Mut iteron region. Iterons are bolded and sub-repeats are denoted with arrows. The underlined sequence is identical to the sequence in red. An asterisk (*) denotes the location of an A to T substitution. (B) An alignment of the iterons from PLE5, PLE4, PLE1, and PLE1^Mut. Iterons are in bold with sub-repeats indicated with arrows. Sequence deviating from a consensus is shown in red. Regions with 100% conservation are denoted with an asterisk. (C) Nuclease assay showing the integrity of a PCR product amplified from the noncoding region containing the ori from wild-type (WT) PLE1 or PLE1^Mut with purified Odn (31.25–500 nM) titrated in. Nuclease assays were performed in triplicate, and replicates are presented in Figure 6—figure supplement 1. (D) Replication of PLE1 WT and PLE1^Mut in V. cholerae calculated as the fold change in PLE DNA copy 20 minutes post infection with the ICP1 variant indicated. (E) Tenfold dilutions of the phage isolate or mutant derivative indicated spotted on V. cholerae with the PLE indicated (bacterial lawns in gray, zones of killing are shown in black). Biological replicates of the spot assays are presented in Figure 6—figure supplement 2.

Figure 6—source data 1.Original uncropped gel for <xref ref-type="fig" rid="fig6">Figure 6C</xref>.

The cropped image shown in the figure is indicated by the orange box.

Figure 6—source data 2.Values for the graph in <xref ref-type="fig" rid="fig6">Figure 6D</xref>.

Figure 6—figure supplement 1.Replicates of nuclease assays.

Nuclease assays showing the integrity of a PCR product amplified from the noncoding region containing the ori from wild-type PLE1 (WT) or PLE1^Mut with (+) and without (–) the addition of 500 nM oOdn. Nuclease assays were performed in triplicate. Two replicates are shown here, and the third is depicted in Figure 6C.

Figure 6—figure supplement 1—source data 1.Original uncropped gels for <xref ref-type="fig" rid="fig6s1">Figure 6—figure supplement 1</xref>.

The cropped images shown in the figure are indicated by the orange boxes.

Figure 6—figure supplement 2.Biological replicates of PLE1<sup>Mut</sup> spot assays.

We sought to test if this natural example of PLE1 iteron diversity had any effect on susceptibility to Odn. We generated a DNA probe covering this variant region (denoted PLE1^Mut) and tested its sensitivity to Odn in comparison to the WT PLE1 ori sequence. The PLE1^Mut sequence was notably less susceptible to cleavage by Odn than the WT allele, but some cutting of the PLE1^Mut probe at the highest concentration of Odn was apparent (Figure 6C). This raised the question of whether this PLE1 variant could resist attack by Odn in vivo. To investigate this possibility, we cloned the PLE1^Mut origin of replication into our PLE1 strain and challenged this mutant with WT, ∆odn and odn* ICP1²⁰⁰¹. The PLE1^Mut origin of replication was able to drive replication to levels comparable to that of the WT origin, and robust replication was maintained even in the presence of Odn (Figure 6D). In agreement with our replication data, we also found that our PLE1^Mut strain restricted production of WT ICP1²⁰⁰¹ (Figure 6E). These data demonstrate that PLE can escape Odn activity through subtle restructuring of the iterons, in addition to more extensive replication module exchange (Figure 3A), and suggest that ICP1 counter-defenses like Odn may select for diversification of the PLE replication machinery.

Discussion

Here, we have identified a new anti-PLE effector present in some ICP1 isolates and identified its mode of anti-PLE activity. The anti-PLE effector Odn possess two domains, an endonuclease domain and a DNA-binding domain, which has sequence similarity to the DNA-binding domain of the RepA initiation factor from PLEs 1, 4, and 5. Mimicking RepA’s specificity, Odn cuts the origin of replication in those PLEs. PLEs are able to escape Odn antagonism through mutation of their origin of replication, and there is compelling evidence suggesting that PLEs have exchanged replicon modules for an alternative replication origin and initiation factor origin-binding domain on at least two separate occasions. While sampling of ICP1 isolates and V. cholerae strains is far from complete, we do see the oldest ICP1 strains encode Odn and the oldest PLE variants encode the sensitive origin of replication. It is possible that Odn selected for alternative replicon modules, leading to the decline of PLE4 and succession of PLE2 that occurred in the early 2000s (O'Hara et al., 2017). This may in turn have caused selection for the presence of CRISPR-Cas, a decrease in prominence of Odn, and the reemergence of the sensitive replicon module in PLE1 matching the trends of PLE and ICP1 occurrence observed thus far (McKitterick et al., 2019b; O'Hara et al., 2017). This model of PLE and ICP1 succession is consistent with an antagonistic frequency-dependent selection (aFDS) mechanism of host-pathogen coevolution where some level of genetic diversity is maintained among antagonistic genes and rare alleles are selected for, leading to oscillation of the dominant genotype (Papkou et al., 2019). In some experimental evolution studies using bacteriophages, aFDS dynamics are found to eventually develop (Gómez and Buckling, 2011; Hall et al., 2011; Lopez Pascua et al., 2014). This seems a likely possibility for PLEs and ICP1 whose specific adaptations and counter adaptations suggest that the two genomes have been coevolving for a long period of time.

It is interesting to consider why a specialized anti-PLE mechanism might persist within some ICP1 isolates when adaptive CRISPR-Cas immunity exists as an alternative. An obvious benefit of Odn is its small size; odn is less than 700 bp and more than 10 times smaller than ICP1’s CRISPR-Cas system. Given that ICP1’s genome size is limited by what it can package into its capsid, the extra sequence taken up by CRISPR-Cas could be better spent on other auxiliary genes if it is not needed to overcome PLE. Additionally, Odn may provide more complete restriction of certain PLEs. While a single CRISPR spacer is sufficient to enable ICP1 plaque formation on PLE(+) cells, it was previously found that multiple spacers were needed to completely abolish PLE replication, and PLE transduction could still be detected from single-spacer ICP1 infections (McKitterick et al., 2019b; O'Hara et al., 2017). Additionally, different CRISPR spacers had different outcomes in terms of PLE transduction and ICP1 reproduction (McKitterick et al., 2019b), suggesting that where PLE is targeted may be important. Position-dependent outcomes for cleavage may explain why Odn is sufficient to provide robust interference against the PLEs it targets. By destroying the origin of replication, Odn-mediated PLE degradation might not have to compete with PLE replication and could head off pathways for PLE escape through recombinational repair. It should also be considered that certain environmental and physiological conditions as well as some genetic factors might render ICP1’s CRISPR-Cas system less effective. While no PLEs appear to encode anti-CRISPRs, numerous anti-CRISPRs have been found in Gammaproteobacteria (Pawluk et al., 2016), and co-occurrence of these genes with PLE could neutralize ICP1’s CRISPR-Cas as an anti-PLE strategy. Overall, it appears that Odn lacks the flexibility provided by CRISPR adaptation but provides more reliable interference against a subset of PLE variants.

In addition to ICP1 and Odn, at least one other anti-PLE mechanism has evolved as some CRISPR(+) ICP1 isolates are able to plaque on PLE2 when CRISPR is deleted (O'Hara et al., 2017). Whatever this anti-PLE mechanism may be, it may have helped select against PLE2, leading to the emergence of PLE1 as another possible example of an aFDS dynamic in the ICP1-PLE arms race. It seems unlikely that this anti-PLE2 mechanism is specific to the alternative replicon module as PLE3, which has a similar replicon to PLE2, maintains restriction of those particular ICP1 isolates (O'Hara et al., 2017). In any case, the co-occurrence of a separate anti-PLE mechanism in the same phage as a CRISPR-Cas system further suggests that CRISPR-Cas may have weak spots in terms of PLE inhibition.

It is somewhat surprising that the PLE1^Mut variant is able to escape Odn activity, given that the new iteron sequences are largely similar to those that exist in PLEs 1, 4, and 5. The most notable differences are an extra T/A base between the inverted sub-repeats in both the second and third iterons, and a reduction of space between the second and third iterons (now 5 bp instead of 14 or 16 bp) (Figure 6B). These changes appear more consistent with some sort of steric effect on catalysis from improper spacing rather than a loss of sequence recognition; however, the molecular details of Odn binding and catalysis remain to be elucidated. The T4-encoded GIY-YIG HEG I-TevI is known to have two separate target specificities, one for DNA binding and one for cleavage, and cleavage is only efficient if the two recognition sites are properly spaced (Liu et al., 2006). It is possible that Odn nucleolytic activity also has some sequence specificity, and closer spacing of the iterons is sufficient to block binding and cleavage by Odn. Based on the data presented here, it is not possible to infer if Odn nucleolytic activity has sequence specificity, and what that specificity might be. For I-TevI, the sequence requirements for cleavage are somewhat loose (Roy et al., 2016), making cleavage sites harder to predict from sequence alone.

Alternatively, Odn cleavage might require recruitment of multiple Odn proteins, spaced a certain distance apart. RepA is thought to bind as dimer (Barth et al., 2020b), but at least some GIY-YIG endonucleases, including I-TevI, function as monomers (Van Roey et al., 2002). It is conceivable that Odn could require cooperative binding for catalysis or alternatively bind and cut at multiple sub-repeats independently. While our results show robust double-stranded cleavage by Odn, this could be achieved through multiple single-strand nicks from appropriately spaced iterons. Odn and RepA provide an interesting example of proteins with overlapping sequence specificity. Working out the intricacies of their binding specificity and activity may prove fruitful for understanding how DNA-binding proteins evolve new targets and functions.

One of the most compelling aspects of Odn is its evolutionary relationship to a family of putative HEGs. HEGs are usually considered selfish genetic elements and are common in phage genomes (Edgell, 2009; Edgell et al., 2010; Stoddard, 2011). This adds a layer of symmetry to the V. cholerae-ICP1-PLE arms race. To defend itself against ICP1, V. cholerae makes use of the PLE, a selfish MGE. To protect against PLE, ICP1 has repurposed a nuclease domain from its own HEG parasites. While this recruitment of MGEs for antagonistic functions fits nicely within the ‘guns for hire’ model of MGE evolution, we see an interesting twist where it seems the horizontal gene transfer of DNA-binding domains is a central mediator of the PLE-ICP1 conflict. Recently, it was found that PLEs reduce ICP1 capsid gene expression through use of a regulator that resembles the DNA-binding domains of ICP1-encoded HEGs (Netter et al., 2021), the very same family of genes from which Odn’s nuclease domain is derived (Figure 4—figure supplement 1). While viral ‘capture’ of host genes by genetic parasites for the purpose of manipulating gene expression is well described in both phage and eukaryotic viruses (Alcami, 2003; Bryan et al., 2008; Zeng and Chisholm, 2012), to our knowledge Odn is the first example of an anti-MGE gene where a sequence-specific DNA-binding protein has been ‘captured’ and fused to an endonuclease domain as a means of destroying the very genome that the sequence originated from. This highlights how horizontal gene transfer can serve as a shortcut to acquiring new sequence-specific antagonists during antagonistic coevolution and provides a slower evolving parallel to adaptive immunity through spacer acquisition.

It is striking that two examples of horizontal transfer between ICP1 and PLE relate to the same family of HEGs. Domain shuffling has been described for HEGs previously (Landthaler and Shub, 2003) and is likely especially adaptive for HEGs, which are thought to amplify within host genomes by acquiring new sequence specificities (Gogarten and Hilario, 2006; Roy et al., 2016). The domain architecture of some phage-encoded HEGs has been likened to ‘beads on a string’: with independent functional domains connected by linker sequences (Van Roey and Derbyshire, 2005). It is conceivable that proteins with this architecture might be more amenable to domain shuffling, and given the selfish nature of HEGs, they may have adaptations to tolerate or promote this shuffling as a means of diversification and dispersal. While HEG domestication has been much discussed (Coughlan et al., 2020; Stoddard, 2011) and the connection between genome antagonism and nucleases is well established in bacteria (Koonin et al., 2020), a specific link between HEGs and MGE antagonism had not been established prior to characterization of Odn. It is not clear if Odn has any intrinsic homing activity, but Odn may be able to cleave and replace CRISPR-Cas loci that have acquired spacers matching the Odn recognition sequence. This could layer a Odn vs. CRISPR-Cas genetic conflict on top of the ICP1-PLE arms race. Because of their selfish nature, ability to mobilize, nucleolytic activity, and enrichment within viral and mobile genomes, HEGs are poised to be at the forefront of antagonistic coevolution between viruses and other MGEs, and we anticipate that other putative HEGs have unrecognized antiparasite activities.

Materials and methods

Key resources table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Gene (Vibrio cholerae)	RepA^PLE1(PLE1 ORF11)	Barth et al., 2020b	WP_002040284.1
Gene (Vibrio cholerae)	RepA^PLE2(PLE2 ORF14)	This paper	AGG36643.1
Gene (Bacteriophage ICP1)	Odn (ICP1_2001_Dha_0 gp88)	This paper	YP_004251029
Gene (Bacteriophage ICP1)	Odn* (ICP1_2001_Dha_0 gp88^E^180A)	This paper		The E180A mutation is predicted to abolish catalytic activity
Recombinant DNA reagent	P_tac-repA^PLE1(plasmid)	Barth et al., 2020b	pZKB129	Inducible RepA from PLE1
Recombinant DNA reagent	P_tac-repA^PLE2(plasmid)	This paper	pKS2159	Inducible RepA from PLE2
Recombinant DNA reagent	pE-SUMO-Odn (plasmid)	This paper	pKS2187	Vector to express 6xHisSumo-fusion protein, fused to N-terminus of Odn (Gp88)
Recombinant DNA reagent	pE-SUMO-Odn* (plasmid)	This paper	pKS2189	Vector to express 6xHisSumo-fusion protein, fused to N-terminus of Odn* (Gp88^E180A)
Strain, strain background (Vibrio cholerae)	PLE V. cholerae (E7946)	Levine et al., 1982	KDS6
Strain, strain background (Vibrio cholerae)	PLE1 V. cholerae (PLE1 E7946)	O'Hara et al., 2017	KDS36
Strain, strain background (Vibrio cholerae)	PLE2 V. cholerae (PLE2 E7946)	O'Hara et al., 2017	KDS37
Strain, strain background (Vibrio cholerae)	PLE3 V. cholerae (PLE3 E7946)	O'Hara et al., 2017	KDS38
Strain, strain background (Vibrio cholerae)	PLE4 V. cholerae (PLE4 E7946)	O'Hara et al., 2017	KDS39
Strain, strain background (Vibrio cholerae)	PLE5 V. cholerae (PLE5 E7946)	O'Hara et al., 2017	KDS40
Strain, strain background (Vibrio cholerae)	PLE1 ∆ori V. cholerae (PLE1 E7946)	This paper	KDS297	Used for all spot assays
Strain, strain background (Vibrio cholerae)	PLE2 ∆ori V. cholerae (PLE2 E7946)	This paper	KDS298	Figure 3—figure supplement 2
Strain, strain background (Vibrio cholerae)	PLE2 ∆repA V. cholerae (PLE2 E7946)	This paper	KDS299	Figure 3—figure supplement 2
Strain, strain background (Vibrio cholerae)	PLE1 ∆repA ∆ori::ori^PLE2; P_tac-repA^PLE1 V. cholerae E7946	This paper	KDS300	Figure 3B
Strain, strain background (Vibrio cholerae)	PLE1 ∆repA ∆ori::ori^PLE2; P_tac-repA^PLE2 V. cholerae E7946	This paper	KDS301	Figure 3B
Strain, strain background (Vibrio cholerae)	PLE2 ∆repA ∆ori::ori^PLE1; P_tac-repA^PLE1 V. cholerae E7946	This paper	KDS302	Figure 3B
Strain, strain background (Vibrio cholerae)	PLE2 ∆repA ∆ori::ori^PLE1; P_tac-repA^PLE2 V. cholerae E7946	This paper	KDS303	Figure 3B
Strain, strain background (Vibrio cholerae)	PLE4 ∆ori V. cholerae (PLE4 E7946)	This paper	KDS304	Used for all spot assays
Strain, strain background (Vibrio cholerae)	PLE5 ∆ori V. cholerae (PLE5 E7946)	This paper	KDS305	Used for all spot assays
Strain, strain background (Vibrio cholerae)	PLE1 ∆iterons V. cholerae (PLE1 E7946)	Barth et al., 2020b	KDS263	Used for all spot assays
Strain, strain background (Vibrio cholerae)	PLE2 ∆ori::ori ^PLE1V. cholerae (PLE2 E7946)	This paper	KDS306	Used for all spot assays
Strain, strain background (Vibrio cholerae)	PLE1∆ori::ori^Mut∆lacZ::KanR V. cholerae E7946 (referred to as PLE1^Mut)	This paper	KDS319	Ori engineered to match what is observed in PLE1(+) strains from Pakistan: biosample accession numbers SAMN08979118, SAMN08979175, SAMN08979185, SAMN08979188, and SAMN08979253
Strain, strain background (Escherichia coli)	pE-SUMO-Odn E. coli BL21	This paper	KDS307	Expression strain for Gp88/Odn
Strain, strain background (Escherichia coli)	pE-SUMO-Odn* E. coli BL21	This paper	KDS308	Expression strain for Gp88/Odn
Strain, strain background (Bacteriophage ICP1)	2006 WT (ICP1_2006_Dha_E)	O'Hara et al., 2017	MH310934
Strain, strain background (Bacteriophage ICP1)	2006 ∆CR; ∆Cas2_3 (ICP1_2006_Dha_E)	McKitterick and Seed, 2018
Strain, strain background (Bacteriophage ICP1)	2001 WT (ICP1_2001_Dha_0)	Seed et al., 2011	HQ641347
Strain, strain background (Bacteriophage ICP1)	2001 ∆odn (ICP1_2001_Dha_0)	This paper	KSϕ93	odn is gp88
Strain, strain background (Bacteriophage ICP1)	2001 odn* (ICP1_2001_Dha_0)	This paper	KSϕ134	odn* isgp88^E180A
Sequence-based reagent	5'-AGGGTTTGAGTGCGATTACG-3'	O'Hara et al., 2017	zac14	qPCR primer targeting a conserved portion of the PLE noncoding region
Sequence-based reagent	5'-TGAGGTTTTACCACCTTTTGC-3'	O'Hara et al., 2017	zac15	qPCR primer targeting a conserved portion of the PLE noncoding region
Sequence-based reagent	5'-GTCATTTAACGCATCTTATCACC-3'	This paper	KS459	F-primer used to amplify noncoding region probes for PLE1 and PLE5
Sequence-based reagent	5'-GGCTTAGCAACTGTCTACGG-3'	This paper	zac267	F-primer used to amplify noncoding region probes for PLE2, PLE3, and PLE4
Sequence-based reagent	5'-GTTACGTCTGATTGCTGACG-3'	This paper	KS321	R-primer used to amplify noncoding region probes for PLE1
Sequence-based reagent	5'-CCGCTTATATCAATTTCACTAATATCT-3'	This paper	zac269	R-primer used to amplify noncoding region probes for PLE2 and PLE3
Sequence-based reagent	5'-GGACGGCTAAACCATTCTCG-3'	This paper	KS323	R-primer used to amplify noncoding region probes for PLE4 and PLE5
Sequence-based reagent	5’-CATAAGGTTGGCTCCTCAATG-3’	This paper	KS458	R-primer used to amplify noncoding region probe for PLE4 in Figure 4—figure supplement 5

Strains and culture conditions

V. cholerae strains used in this study are derived from E7946. Bacteria were routinely grown on LB agar plates and in LB broth with aeration at 37°C. Antibiotics were supplemented as appropriate at the following concentrations: 75 μg/ml kanamycin, 100 μg/ml spectinomycin, 1.25 or 2.5 μg/ml chloramphenicol (V. cholerae for broth or plate conditions, respectively), 25 μg/ml chloramphenicol (Escherichia coli), and 100 μg/ml streptomycin. A detailed list of all strains used throughout this study can be found in the Key resources table.

Phage titers were determined using a soft agar overlay method wherein ICP1 was allowed to adsorb to V. cholerae for 10 minutes at room temperature before the mixture was added to molten LB soft agar (0.5%) and poured onto 100 mm × 15 mm LB agar plates. Plaques were counted after overnight incubation at 37°C. Prior to phage infection for purposes of quantification or qPCR or spot assay analysis, V. cholerae was grown on plates overnight and then inoculated into 2 ml LB liquid cultures. Liquid cultures were grown to an OD > 1, then back diluted in fresh media to OD₆₀₀ = 0.05, and then grown to OD₆₀₀ = 0.3, at which point they were infected.

Generation of mutant strains and constructs

V. cholerae mutants were generated through natural transformation as described previously (Dalia et al., 2014). For gene knockouts, splicing by overlap extension (SOE) PCR was used to generate deletion constructs with a spectinomycin resistance cassette flanked by frt recombination sites. Following selection of spectinomycin-resistant mutants, a plasmid bearing an isopropyl β-d-1-thiogalactopyranoside (IPTG)-inducible Flp recombinase was mated into transformants and Flp expression was induced to generate in-frame deletions. The plasmid was cured by growing mutants under inducing conditions with 300 μg/ml streptomycin. For unmarked replication origin-swapped constructs, mutants were generated through natural transformation by cotransformation (Dalia et al., 2014). For plasmid expression constructs, a derivative of the pMMB67EH vector with a theophylline-inducible riboswitch was used as previously described (McKitterick and Seed, 2018). All constructs were confirmed with DNA sequencing over the region of interest, and primer sequences and construct designs are available on DRYAD at https://datadryad.org/stash/share/HSB-bM3fCu3gSdF_yMQpCqyYuT4wW6_2IsZAkY0P5Ho.

Phage infection spot assays

V. cholerae was added to molten 0.5% LB top agar and poured over LB plates. Following solidification of the top agar, 3 μl of serially 10-fold diluted phage were spotted onto the plate. Once phage spots dried, plates were incubated for at 37°C for 2 hr and then overnight at room temperature before visualization.

Real-time quantitative PCR

qPCR experiments were performed as previously described (Barth et al., 2020b; O'Hara et al., 2017). Briefly, liquid cultures were infected with ICP1 at a multiplicity of infection (MOI) of 2.5 at OD₆₀₀ = 0.3. Samples were taken at 0 and 20 minutes post infection and boiled before serving as templates for IQ SYBR (Bio-Rad) qPCR reactions. For assays involving induction of repA, 2 ml cultures were grown with 1.25 μg/ml chloramphenicol for plasmid maintenance and induced for 20 minutes prior to infection using a final concentration of 1.5 mM theophylline and 1 mM IPTG starting at OD₆₀₀ = 0.17. All conditions were tested in biological triplicate, and each reported data point is the mean of two technical replicates. A single primer set (Key resources table) that amplifies a conserved region in all PLEs was used to detect PLE replication by qPCR.

Protein purification

E. coli BL21 cells containing a His₆-SUMO fusion to WT or E185A Gp88 were grown to OD₆₀₀ = 0.5 at 37°C and induced with IPTG to a final concentration of 0.5 mM. The culture was grown for 2 hr and harvested by centrifugation at 4000×g for 20 minutes. The pellet was resuspended in lysis buffer (50 mM Tris–HCl pH 8, 200 mM NaCl, 1 mM BME, 0.5% Triton-X 50 mM imidazole, 1 Pierce Protease Inhibitor Mini Tablet [Thermo Scientific]) and sonicated. Cell debris was removed by centrifugation (29,097×g for 40 minutes). The lysate was applied to a HisTrap HP column (Cytiva). The column was washed with wash buffer (50 mM Tris–HCl pH 8, 200 mM NaCl, 1 mM BME, 50 mM imidazole), and a high salt wash (50 mM Tris–HCl pH 8, 2 M NaCl, 1 mM BME, 50 mM imidazole) was used to remove residual DNA. The protein was eluted using an elution buffer (50 mM Tris–HCl pH 8, 200 mM NaCl, 1 mM BME, 300 mM imidazole), and then the eluate was applied to a HiTrap Heparin HP column (Cytivia) for further purification. Following elution from the HiTrap Heparin column, the protein was dialyzed using a 10k Slide-A-Lyzer Dialysis cassette (Thermo Fisher) in 50 mM Tris–HCl pH 7.5, 150 mM NaCl, 1 mM dithiothreitol (DTT). Concomitant with dialysis, the His₆-SUMO tag was cleaved using SUMO protease. The SUMO tag was removed using Dynabeads (Invitrogen).

Sequence analysis

All genomes were visualized and compared in CLC Main Workbench 7. Multiple sequence alignments were performed using the Multiple Sequence Alignment (MUSCLE) tool with default settings (Edgar, 2004). The phylogenetic tree was constructed using the IQ-TREE web interface with default settings (Trifinopoulos et al., 2016). Conservation of PLE sequence was compared and visualized using Mauve (Darling et al., 2004).

Nuclease assays

Nuclease assays were performed with 100 ng of DNA probes and up to 500 nM of purified Gp88 in 20 μl reactions with 50 mM Tris, 10 mM MgCl₂, 50 mM NaCl, 1 mM DTT reaction buffer. Reactions proceeded at 30°C for 30 minutes, and were visualized on 0.8% agarose gels ran at 80 V for 30 minutes, and stained with GelRed (Biotium). For smaller probes (Figure 6), 25 ng of probe was included in reactions, and the product was visualized on 2% agarose gels ran at 120 V for 20 minutes and stained with GelGreen (Biotium). Primers used for probe amplification can be found in the Key resources table.

Acknowledgements

This work was supported by the National Institute of Allergy and Infectious Diseases (grant numbers R01AI127652 and R01AI153303 to KDS); KDS is a Chan Zuckerberg Biohub Investigator and holds an Investigators in the Pathogenesis of Infectious Disease Award from the Burroughs Wellcome Fund. We would like to thank Yue Clare Lou for construction of strains, as well as Stephanie Hays, Kristen LeGault, and Zoe Netter for critical reading of the manuscript and providing useful advice.

Additional information

Competing interests

No competing interests declared

is a scientific advisor for Nextbiotics, Inc.

Author contributions

Conceptualization, Formal analysis, Investigation, Writing - original draft

Investigation, Writing - review and editing

Conceptualization, Supervision, Funding acquisition, Investigation, Writing - review and editing

Additional files

Transparent reporting form

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files. Design for genetic constructs including primers are available via Dyrad (https://doi.org/10.6078/D1T704).

The following dataset was generated:

BarthZKNguyenMHTSeedKD2021Strains and constructs for: A chimeric nuclease substitutes a phage CRISPR-Cas system to provide sequence specific immunity against subviral parasitesDryad Digital Repository10.6078/D1T704

References

Alcami

2003

Viral mimicry of cytokines, chemokines and their receptors

Nature Reviews Immunology33650

10.1038/nri980

12511874

Angermeyer

Das

Singh

Seed

2018

Analysis of 19 highly conserved Vibrio cholerae bacteriophages isolated from environmental and patient sources over a Twelve-Year period

Viruses10

299

10.3390/v10060299

Barrangou

Fremaux

Deveau

Richards

Boyaval

Moineau

Romero

Horvath

2007

CRISPR provides acquired resistance against viruses in prokaryotes

Science31517091712

10.1126/science.1138140

17379808

Barth

Netter

Angermeyer

Bhardwaj

Seed

2020a

A family of viral satellites manipulates invading virus gene expression and can affect cholera toxin mobilization

mSystems5

e00358

10.1128/mSystems.00358-20

Barth

Silvas

Angermeyer

Seed

2020b

Genome replication dynamics of a bacteriophage and its satellite reveal strategies for parasitism and viral restriction

Nucleic Acids Research48249263

10.1093/nar/gkz1005

31667508

Brüssow

Hendrix

2002

Phage genomics: small is beautiful

Cell1081316

10.1016/s0092-8674(01)00637-7

11792317

Bryan

Burroughs

Spence

Clokie

Mann

Bryan

2008

Evidence for the intense exchange of MazG in marine cyanophages by horizontal gene transfer

PLOS ONE3

e2048

10.1371/journal.pone.0002048

18431505

Coughlan

Lombardi

Braun-Galleani

Martos

Galeote

Bigey

Dequin

Byrne

Wolfe

2020

The yeast mating-type switching endonuclease HO is a domesticated member of an unorthodox homing genetic element family

eLife9

e55336

10.7554/eLife.55336

32338594

Dalia

McDonough

Camilli

2014

Multiplex genome editing by natural transformation

PNAS11189378942

10.1073/pnas.1406478111

24889608

Darling

Mau

Blattner

Perna

2004

Mauve: multiple alignment of conserved genomic sequence with rearrangements

Genome Research1413941403

10.1101/gr.2289704

15231754

Doron

Melamed

Ofir

Leavitt

Lopatina

Keren

Amitai

Sorek

2018

Systematic discovery of antiphage defense systems in the microbial pangenome

Science359

eaar4120

10.1126/science.aar4120

Edgar

2004

MUSCLE: a multiple sequence alignment method with reduced time and space complexity

BMC Bioinformatics5

113

10.1186/1471-2105-5-113

15318951

Edgell

2009

Selfish DNA: homing endonucleases find a home

Current Biology19R115R117

10.1016/j.cub.2008.12.019

19211047

Edgell

Gibb

Belfort

2010

Mobile DNA elements in T4 and related phages

Virology Journal7

290

10.1186/1743-422X-7-290

21029434

Gao

Altae-Tran

Böhning

Makarova

Segel

Schmid-Burgk

Koob

Wolf

Koonin

Zhang

2020

Diverse enzymatic activities mediate antiviral immunity in prokaryotes

Science36910771084

10.1126/science.aba0372

32855333

Gogarten

Hilario

2006

Inteins, introns, and homing endonucleases: recent revelations about the life cycle of parasitic genetic elements

BMC Evolutionary Biology6

10.1186/1471-2148-6-94

17101053

Gómez

Buckling

2011

Bacteria-phage antagonistic coevolution in soil

Science332106109

10.1126/science.1198767

21454789

Greenwood

Ishida

O'Brien

Roca

Eiden

2018

Transmission, evolution, and endogenization: lessons learned from recent retroviral invasions

Microbiology and Molecular Biology Reviews82

10.1128/MMBR.00044-17

Hall

Scanlan

Morgan

Buckling

2011

Host-parasite coevolutionary arms races give way to fluctuating selection

Ecology Letters14635642

10.1111/j.1461-0248.2011.01624.x

21521436

Harrison

Brockhurst

2017

Ecological and evolutionary benefits of temperate phage: what does or Doesn't kill you makes you stronger

BioEssays39

1700112

10.1002/bies.201700112

28983932

Koonin

Makarova

Wolf

Krupovic

2020

Evolutionary entanglement of mobile genetic elements and host defence systems: guns for hire

Nature Reviews Genetics21119131

10.1038/s41576-019-0172-9

31611667

Koonin

Makarova

2019

Origins and evolution of CRISPR-Cas systems

Philosophical Transactions of the Royal Society B: Biological Sciences374

20180087

10.1098/rstb.2018.0087

Krupovic

Dolja

Koonin

2019

Origin of viruses: primordial replicators recruiting capsids from hosts

Nature Reviews Microbiology17449458

10.1038/s41579-019-0205-6

31142823

Landthaler

Shub

2003

The nicking homing endonuclease I-BasI is encoded by a group I intron in the DNA polymerase gene of the Bacillus thuringiensis phage bastille

Nucleic Acids Research3130713077

10.1093/nar/gkg433

12799434

Levine

Black

Clements

Cisneros

Saah

Nalin

Gill

Craig

Young

Ristaino

1982

The pathogenicity of nonenterotoxigenic Vibrio cholerae serogroup O1 biotype el Tor isolated from sewage water in Brazil

Journal of Infectious Diseases145296299

10.1093/infdis/145.3.296

7061878

Liu

Derbyshire

Belfort

Edgell

2006

Distance determination by GIY-YIG intron endonucleases: discrimination between repression and cleavage functions

Nucleic Acids Research3417551764

10.1093/nar/gkl079

16582101

Liu

Feng

Liu

Chen

Pham

Deng

Hirschi

Wang

Cheng

2013

Structural insights into the N-terminal GIY-YIG endonuclease activity of Arabidopsis glutaredoxin AtGRXS16 in chloroplasts

PNAS11095659570

10.1073/pnas.1306899110

23690600

Loenen

Raleigh

2014

The other face of restriction: modification-dependent enzymes

Nucleic Acids Research425669

10.1093/nar/gkt747

23990325

Lopez Pascua

Hall

Best

Morgan

Boots

Buckling

2014

Higher resources decrease fluctuating selection during host-parasite coevolution

Ecology Letters1713801388

10.1111/ele.12337

25167763

Marraffini

2015

CRISPR-Cas immunity in prokaryotes

Nature5265561

10.1038/nature15386

26432244

McKitterick

Hays

Johura

Alam

Seed

2019a

Viral satellites exploit phage proteins to escape degradation of the bacterial host chromosome

Cell Host Microbe26504514

10.1016/j.chom.2019.09.006

31600502

McKitterick

LeGault

Angermeyer

Alam

Seed

2019b

Competition between mobile genetic elements drives optimization of a phage-encoded CRISPR-Cas system: insights from a natural arms race

Philosophical Transactions of the Royal Society B: Biological Sciences374

20180089

10.1098/rstb.2018.0089

McKitterick

Seed

2018

Anti-phage islands force their target phage to directly mediate island excision and spread

Nature Communications9

2348

10.1038/s41467-018-04786-5

29904071

Millman

Bernheim

Stokar-Avihail

Fedorenko

Voichek

Leavitt

Oppenheimer-Shaanan

Sorek

2020

Bacterial retrons function in Anti-Phage defense

Cell18315511561

10.1016/j.cell.2020.09.065

33157039

Mruk

Kobayashi

2014

To be or not to be: regulation of restriction-modification systems and other toxin-antitoxin systems

Nucleic Acids Research427086

10.1093/nar/gkt711

23945938

Netter

Boyd

Silvas

Seed

2021

A phage satellite tunes inducing phage gene expression using a domesticated endonuclease to balance inhibition and virion hijacking

Nucleic Acids Research4943864401

10.1093/nar/gkab207

O'Hara

Barth

McKitterick

Seed

2017

A highly specific phage defense system is a conserved feature of the Vibrio cholerae mobilome

PLOS Genetics13

e1006838

10.1371/journal.pgen.1006838

28594826

Otsuka

Yonesaki

2012

Dmd of bacteriophage T4 functions as an antitoxin against Escherichia coli LsoA and RnlA toxins

Molecular Microbiology83669681

10.1111/j.1365-2958.2012.07975.x

22403819

Panayotatos

Fontaine

1985

An endonuclease specific for single-stranded DNA selectively damages the genomic DNA and induces the SOS response

Journal of Biological Chemistry26031733177

10.1016/S0021-9258(18)89488-1

3972821

Papkou

Guzella

Yang

Koepper

Pees

Schalkowski

Barg

Rosenstiel

Teotónio

Schulenburg

2019

The genomic basis of red queen dynamics during rapid reciprocal host-pathogen coevolution

PNAS116923928

10.1073/pnas.1810402116

30598446

Parson

Snustad

1975

Host DNA degradation after infection of Escherichia coli with bacteriophage T4: dependence of the alternate pathway of degradation which occurs in the absence of both T4 endonuclease II and nuclear disruption on T4 endonuclease IV

Journal of Virology15221224

10.1128/jvi.15.1.221-224.1975

1089802

Pawluk

Staals

Taylor

Watson

Saha

Fineran

Maxwell

Davidson

2016

Inactivation of CRISPR-Cas systems by anti-CRISPR proteins in diverse bacterial species

Nature Microbiology1

16085

10.1038/nmicrobiol.2016.85

27573108

Roy

Wilson

Edgell

2016

Perpetuating the homing endonuclease life cycle: identification of mutations that modulate and change I-TevI cleavage preference

Nucleic Acids Research4473507359

10.1093/nar/gkw614

27387281

Seed

Bodi

Kropinski

Ackermann

Calderwood

Qadri

Camilli

2011

Evidence of a dominant lineage of Vibrio cholerae-specific lytic bacteriophages shed by cholera patients over a 10-year period in Dhaka, Bangladesh

mBio2

e00334-10

10.1128/mBio.00334-10

21304168

Seed

Lazinski

Calderwood

Camilli

2013

A bacteriophage encodes its own CRISPR/Cas adaptive response to evade host innate immunity

Nature494489491

10.1038/nature11927

23446421

Souther

Bruner

Elliott

1972

Degradation of Escherichia coli chromosome after infection by bacteriophage T4: role of bacteriophage gene D2a

Journal of Virology10979984

10.1128/jvi.10.5.979-984.1972

4564588

Stoddard

2011

Homing endonucleases: from microbial genetic invaders to reagents for targeted DNA modification

Structure19715

10.1016/j.str.2010.12.003

21220111

Swapna

GVT

Bhattacharya

Aramini

Acton

Xiao

Shastry

Shih

Cunningham

Montelione

2005

RCSB PDB - 1YWL: solution NMR structure of the protein EF2693 from E. faecalis: northeast structural genomics consortium target EFR36

https://www.rcsb.org/structure/1ywl

February 12, 2021

Trifinopoulos

Nguyen

von Haeseler

Minh

2016

W-IQ-TREE: a fast online phylogenetic tool for maximum likelihood analysis

Nucleic Acids Research44W232W235

10.1093/nar/gkw256

27084950

Truglio

Rhau

Croteau

Wang

Skorvaga

Karakas

DellaVecchia

Wang

Van Houten

Kisker

2005

Structural insights into the first incision reaction during nucleotide excision repair

The EMBO Journal24885894

10.1038/sj.emboj.7600568

15692561

Van Roey

Meehan

Kowalski

Belfort

Derbyshire

2002

Catalytic domain structure and hypothesis for function of GIY-YIG intron endonuclease I-TevI

Nature Structural Biology9806811

10.1038/nsb853

12379841

Van Roey

Derbyshire

2005

GIY-YIG Homing Endonucleases — Beads on a String

Wood

D. W

Stoddard

B. L

Belfort

Derbyshire

Homing Endonucleases and Inteins, Nucleic Acids and Molecular BiologyBerlin, Heidelberg

Springer Berlin Heidelberg

6783

10.1007/3-540-29474-0

Warner

Drong

Berget

1975

Early events after infection of Escherichia coli by bacteriophage T5. induction of a 5'-nucleotidase activity and excretion of free bases

Journal of Virology15273280

10.1128/jvi.15.2.273-280.1975

163355

Zeng

Chisholm

2012

Marine viruses exploit their host's two-component regulatory system in response to resource limitation

Current Biology22124128

10.1016/j.cub.2011.11.055

22244998

Zimmermann

Stephens

Nam

Rau

Kübler

Lozajic

Gabler

Söding

Lupas

Alva

2018

A completely reimplemented MPI bioinformatics toolkit with a new HHpred server at its core

Journal of Molecular Biology43022372243

10.1016/j.jmb.2017.12.007

29258817

10.7554/eLife.68339.sa1

Decision letter

Wiedenheft

Blake

Reviewing EditorMontana State UniversityUnited States

Marraffini

Luciano

ReviewerRockefeller UniversityUnited StatesWestra

Edze

ReviewerUniversity of ExeterUnited Kingdom

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

Mobile genetic parasites in Vibrio cholerae genomes often protect the host from phage infection. Previous work has shown that some phages that infect Vibrio cholerae have evolved CRISPR-systems that selectively target these defensive mobile genetic elements and thereby restore infection by the phage. Here the authors show that phage genomes missing the CRISPR-system, often rely on a chimeric nuclease that functionally replace the CRISPR-system, revealing an ongoing evolutionary battle that includes one host and two competing parasites.

Decision letter after peer review:

Thank you for submitting your article "A chimeric nuclease substitutes CRISPR-Cas: A phage weaponizes laterally acquired specificity to destroy subviral parasites" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by Blake Wiedenheft as the Reviewing Editor and Gisela Storz as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Luciano Marraffini (Reviewer #2); Edze Westra (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) Streamline the introduction to include only information critical to understanding how your work advances the field.

2) An SDS-PAGE of the purified WT and mutant proteins is required to demonstrate purity and help support the conclusion that the Odn nuclease alone is responsible for cleavage of the PLE.

3) Clarify inconsistencies and variabilities observed in the nuclease activity assays. For example, Figure 4d shows PLEs 1, 4, and 5 are cleaved to completion by WT Odn, while in Figure 6c and its supplements, the WT shows very little cleavage activity against PLE1. Different lengths of the substrates will shift the molar ratio of enzyme to substrate, and this impact the nuclease activity. Cleavage efficiencies must be quantified and normalized.

4) Perform plaque assays for the ori mutant that renders this PLE resistant to Odn Figure 6c. This will be similar to the presentation of data in Figure 5a.

5) Please specify that the evidence for Odn activity driving evolutionary change in the PLE is anecdotal.

6) Sequence Odn cleavage products and map the precise location of these cuts.

Reviewer #1:

Barth et al. discover and characterize a new mechanism by which bacteriophage ICP1 defends itself against its own molecular parasites, the PLEs. PLEs are short stretches of DNA found in ~15% of sequenced epidemic strains of V. cholerae which hijack the structural components of bacteriophage ICP1 to facilitate their own spread-through this process, PLEs promote survival of the bacteria by preventing the spread of an ICP1 infection. Therefore, the interplay between ICP1 and its PLEs have significant impacts on Vibrio pathogenesis, highlighting the importance of this work. Previous research from this group has shown that some ICP1 variants harbor a CRISPR-Cas immune system, which these bacteriophages use to target and degrade PLEs. However, not all ICP1 variants possess CRISPR-Cas systems, and yet retain resistance to some PLE variants, suggesting the presence of other anti-PLE immune systems encoded in the bacteriophage genome. In this study, the investigators seek to understand how an ICP1 variant devoid of a CRISPR-Cas system can still defend against a subset of PLEs.

Strengths

A rigorous set of genetics experiments were applied to reveal conserved functional aspects of PLE replication (Figure 3 and its supplements) as well as the genetic element in ICP1 required for anti-PLE activity, called gp88 (Figures 1, 2, and associated supplements). Based on sequence alignments and preliminary biochemical characterization, the gp88 gene product was shown to facilitate nuclease activity against PLE variants that are sensitive to ICP1 (Figure 4d). Genetics was also successfully applied to narrow down the region within the PLE which is targeted by the gp88 (Odn) nuclease (Figures 4a-c and 5). Finally, a natural variant of PLE1 that seems resistant to Odn activity was discovered within a V. cholerae isolate from Pakistan (Figure 6), providing a beautiful example of how the ongoing conflict between host and parasite drives genetic diversification of both parties. Overall, the major conclusions in the paper are well-supported by the data, the manuscript is clearly written, and the study gives important insights into this system.

Weaknesses

The biochemical characterization of Odn is preliminary and needs more rigorous experimentation.

Comments for the authors:

1. An image of the purified proteins, both WT and mutant variants, (typically in the form of SDS-PAGE gels), should be displayed to demonstrate their purity and help support the conclusion that the Odn nuclease alone is responsible for cleavage of the PLE. The methods describe three separate chromatography steps, so it is assumed that the proteins are relatively pure, but a gel should be shown.

2. The mutant version of Odn should be subject to the same level of scrutiny as the WT to solidify the conclusion that Odn is alone responsible for PLE cleavage, at least in the beginning. For example similar assays could be performed with the mutant as with WT in Figures 4c and Figure 5a.

3. There seems to be some inconsistencies/variabilities in the biochemical assays. For example, Figure 4d shows PLEs 1, 4, and 5 are cleaved to completion by WT Odn, while in Figure 6c and its supplements, the WT shows very little cleavage activity against PLE1. Some quantification will help the reader to understand the extent of the variation in the nuclease activity. Along the same lines, different lengths of substrates were used in these assays, which will shift the molar ratio of enzyme:substrate and impact the nuclease activity. Some sort of normalization across the assays will allow the reader to compare the levels of nuclease activity from gel to gel and assess how each new variable affects the activity.

4. To come full-circle, sequencing of the cut products would go a long way to support the genetics evidence for Odn targeting within the PLE origin and map the precise location of the cut(s).

Reviewer #2:

Barth et al. report the discovery of a novel endonuclease effector employed by certain lytic vibriophages (ICP1) to overcome viral satellites called PLEs present in ~15% of sequenced epidemic Vibrio cholerae. Past work from the Seed lab (Seed et al., 2013) had demonstrated that some ICP1 isolates (especially those from 2011-2017) encoded their own type I-F CRISPR-Cas system to target PLEs. However, they also previously observed (O'Hara et al., 2017) that older ICP1 isolates that lacked CRISPR-Cas (2001-2011) could still plaque on some of the strains containing PLEs. This work sought to elucidate the mechanisms by which a CRISPR-less ICP1 isolate from 2001 could successfully overcome certain PLE variants, but not others.

Through a set of simple yet elegant experiments that directly tested their hypotheses, the authors convincingly show that ICP12001 encodes an endonuclease (initially gp88, which they later term Odn, for origin-directed nuclease) that allows it to successfully infect V. cholerae containing certain PLEs. Interestingly, Odn contains two domains-the C-terminus resembling a homing endonuclease (HEG), and the N-terminus resembling the DNA-binding domain of the replication initiator RepA of PLEs 1, 4, and 5. The authors speculate that Odn might have arisen from domain shuffling between ICP1-encoded HEGs and PLE-encoded RepA, although how exactly this occurred is unknown.

The major strengths of this work are the novelty of the findings and its logical clarity. With a series of incisive experiments, the authors deduced how Odn works. By constructing ICP1 mutants that lacked Odn activity, the authors showed that Odn is necessary for plaque formation on V. cholerae containing PLEs 1, 4, or 5. Further, they show using qPCR that plaquing efficiency on PLE1 is correlated with Odn-mediated repression of PLE1 replication. In the future, it will be interesting to know if Odn alone is sufficient to overcome PLE, and whether the adjacent genes broN and gp90 play any regulatory or accessory role in Odn function.

Next, the authors compared the genomes of all five PLE variants and found that the ori and RepA sequences cluster into two distinct groups-PLEs 1/4/5 and PLEs 2/3. This potentially explained why ICP12001 can plaque on strains containing PLEs 1, 4, and 5, but not PLEs 2 or 3, leading the authors to hypothesize that Odn recognizes and binds to the origin of replication of PLEs 1/4/5, but not PLEs 2/3. The Seed lab previously showed that non-replicating PLE mutants lacking ori could still restrict ICP1 (Barth et al., 2020). Exploiting this experimental tool, they showed here that ICP12001 could no longer plaque on strains containing PLE mutants lacking ori. Finally, by re-constituting the ori of PLE1 into PLE2, the strain is now sensitized to Odn activity upon ICP12001 infection. This decisive result strongly suggests that recognition of ori sequences is required for Odn-mediated plaquing.

Finally, using an in vitro cleavage assay, the authors show that purified Odn cleaves PCR products amplified from the non-coding region containing the ori from PLEs 1/4/5, but not PLEs 2/3, consistent with the in silico prediction that the C-terminus of Odn resembles putative HEGs in ICP1. The authors further narrow down the region of specificity of Odn action to a set of iterons (a series of three ~30bp semi-palindromic repeats in the origin of replication) and demonstrate using a PLE1 mutant lacking these sequences that iterons are necessary for Odn-mediated cleavage and plaquing by ICP12001. Surprisingly, there is only a 2-bp difference between PLEs 4 and 5, yet the cleavage pattern looks different between the two, raising the question of whether a more complicated mechanism of recognition and cleavage is in place.

Finally, the authors address a crucial aspect of any host-parasite arms race-how can PLE escape from Odn-mediated restriction? They reason that simple deletions of the replication origin would not likely be selected for as a long-term solution, given its essentiality for PLE reproduction. Because of this, the authors speculate that Odn selected for domain shuffling events in PLE RepA, resulting in the two distinct clusters of PLEs 1/4/5 and PLEs 2/3. Although each PLE variant is remarkably conserved, the authors found an instance where PLE1 contained variation in its iteron sequences. They tested this variant and found that Odn was unable to cleave, suggesting that this variant PLE1 was likely resistant to infection by ICP12001, although they did not perform a plaque assay to confirm this.

In conclusion, the authors complete a thorough characterization of Odn as an anti-PLE effector in ICP1. This work presents a key advance in our understanding of ICP1-PLE dynamics and brings broader insight to the roles of lateral gene transfer in the acquisition of novel weaponry in the host-parasite arms-race.

Comments for the authors:

Experimental:

Figure 1: can Odn be re-constituted into a CRISPR-less ICP1 and restore plaquing against PLE1 to demonstrate sufficiency?

Figure 6: In addition to the cleavage assay, to complement the in vitro data with in vivo results, the authors should perform plaque assays on PLE1mut for ICP12001 (and the Odn deletion and catalytically inactive mutants). It would be expected for ICP12001 to no longer be able to plaque given that Odn cannot cleave.

Optional: it would be interesting to see if PLE escapers could arise naturally during an infection experiment and to determine what these mutations were. For example, you can infect a strain containing PLE1 with ICP12001 at very high MOI to lyse a whole plate. Are there any bacterial-insensitive mutants (BIMs) that form colonies? If so, are there any mutations in ori (and/or compensatory mutations in RepA), or perhaps even more strikingly, any restructuring of the iteron? This experimental evidence, together with the naturally occurring PLE1mut variant observed in Figure 6, would bolster the claim that Odn may be a driving force for the diversification of the PLE replication origin.

Text edits:

Title: the title, although descriptive, is a bit long and convoluted without having read the paper first. I would consider simplifying it to a single clause and remove jargon like "laterally acquired specificity"

Introduction: the description of MGE and HGT in multicellular organisms in the first paragraph is not necessary to understand this work and possibly too much information for the reader.

Figure 2: although this is beyond the focus of this work and does not need to be addressed in the text, it does not escape my attention that PLE3 is able to evade both CRISPR (no targeting spacer) and Odn (ori not recognized). Is there a contemporaneous ICP1 isolate that is able to plaque on PLE3, and if so, is there a different effector in the same genomic region between gp87 and gp91?

Figure 3: is there a working model for how PLEs recombine (if they even do) to form new variants? Are their isolates of V. cholerae that contain multiple PLEs?

Lines 342-343: "Remarkably, Gp88's own N-terminal domain is 61% similar to 93% of PLE1's RepA_N domain".

This sentence is a bit misleading, given that it's only 42% identical (which is still quite similar). The inclusion of "similar residues" in the comparison indeed yields 61%, but this factor in the calculation should be explicitly stated in the text.

Figure 4D: There appears to be two cuts for PLEs 1 and 5, but not 4? This is interesting, especially since there's only a 2-bp difference in the iteron sequences between PLEs 4 and 5 (Figure 6)-is there any explanation for the differences between the cleavage pattern of the three PLEs at this time, and whether these provide any clues as to Odn's mechanism of action?

Discussion:

Lines 551-552: what are some examples of auxillary genes that are present between CRISPR-less ICP1 vs. CRISPR(+) ICP1? Are CRISPR-less ICP1 isolates on average just slightly smaller?

Reviewer #3:

Phage ICP1 infection can be blocked by PLE elements in Vibrio genomes, which use phage capsids for their own dissemination, but ICP1 has in turn evolved mechanisms to overcome PLE-mediated defence/parasitism. Seed and colleagues have discovered and extensively characterised how a CRISPR-Cas system encoded by the phage can cleave the PLE element, leading to productive infections.

Here, the authors aimed to explain why an ICP1 genotype that naturally lacks CRISPR-Cas, and that used to be very common before CRISPR-carrying variants appeared, can infect Vibrio cells that certain PLE variants. Specifically, their previous work found that this ICP1 variant can infect cells that carry the two oldest PLE variants, PLE5 and PLE4, as well as the most recent variant PLE1 (O'Hara et al., 2017).

They found that this older ICP1 genotype carries an alternative nuclease (termed origin-directed nuclease, Odn) that cleaves the origin of replication of some PLE elements. This nuclease gene is located in the same locus as the CRISPR-Cas genes in more recent ICP1 genotypes. This newly discovered nuclease carries an N-terminal DNA binding domain that has high sequence similarity to RepA, which binds the origin of replication, and a C-terminal nuclease domain that is similar to a ICP1-encoded family of putative homing endonucleases (HEGs), and that is responsible for cleavage of the origin of replication. Different PLE variants (PLE1-5) vary with respect to their RepA and origin of replication sequences – specifically, the authors show that PLE1, 4 and 5 share RepN and Ori sequences, explaining why the ICP1 variant that carries Odn can infect cells that carry these PLE elements, but not cells that carry PLE2 or PLE3.

The conclusions of this paper are well supported by comparative genomics, elegant reverse genetics analyses and biochemical assays. The authors generate all the relevant KO variants of ICP1 and PLE to demonstrate the genetic interaction between Odn of ICP1 and the Ori of PLE, including ori-swapped mutants of PLE, where PLE2 carries the ori of PLE1, leading to a lack of protection of the host from phage infection. Finally, the authors purified Odn and demonstrate that it can cleave the Ori sequences in vitro, but not random sequence, and a catalytic mutant is unable to cleave either (and the same mutation leads to PLE sensitivity in vivo). This binding and cleavage is specifically directed towards a set of semi-palindromic sequences, known as iterons.

All in all, this is a very thorough and complete study that provides novel and deep insight into the ongoing coevolutionary interaction between PLE and ICP1.

Comments for the authors:

This is a very nice study – I enjoyed reading the paper. I only have fairly minor comments/suggestions:

1. This is a matter of preference in style, but in my opinion the introduction and discussion could be much tighter. The first 3 paragraphs of the introduction provide a lot of background that is not essential for this study, perhaps more appropriate for a broad review, and could be trimmed substantially. The discussion is

2. I'm left wondering whether BroN and gp90 are essential for the activity of Odn. Have you made knockouts of these genes ? Or alternatively, have you expressed Odn from a plasmid in cells with PLE and looked for evidence of cleavage (e.g. cytotoxic effects that are not observed for the catalytic mutant) ? You mention they likely have regulatory roles – what is the evidence in support of this idea ?

3. Figure 3A shown the ori and RepA-N sequence variation between PLE1,4,5 on the one hand, and PLE2,3 on the other. Supplement 1 only shows the alignment of the conserved c-terminus. Can you please add the N-terminal sequence to enable more detailed examination of the level of sequence variation / conservation between the different groups?

4. the similarity of the nuclease domain to the putative homing endonucleases in ICP1 is intriguing, and provides a plausible explanation for the origin of the nuclease domain. How common are these homing endonucleases? Can the authors make a tree of these nuclease domains to see if this Odn nuclease domain fall inside the tree? Or do you not have enough sequences for this / is sequence divergence too high?

5. Figure 4 D – nice to see the cleavage activity of gp88/Odn. It should not be too hard to identify the cleavage site using Sanger sequencing (it looks like sharp bands, so i assume a single cut). Have you tried this?

6. Line 456-459 – I don't think the data are sufficient to draw this conclusion; even if the mutation provides resistance to Odn, it is impossible to work out from these data whether or not the Ori mutations are adaptive, and if so, whether or not this is because of ICP1 infections. This statement therefore needs to be further toned down in my opinion.

7. Line 544 – the Gomez et al. Science paper from the same group would also be appropriate to cite.

10.7554/eLife.68339.sa2

Author response

Essential revisions:

1) Streamline the introduction to include only information critical to understanding how your work advances the field.

We have streamlined the introduction removing several examples.

2) An SDS-PAGE of the purified WT and mutant proteins is required to demonstrate purity and help support the conclusion that the Odn nuclease alone is responsible for cleavage of the PLE.

We have added images of the SDS-PAGE gels of protein preparations used in in vitro assays as Figure 4—figure supplement 2.

For Figure 6C we originally used a shorter substrate for both the WT PLE1 origin andthe mutant PLE1 origin because we had difficulty cloning the mutant origin sequence into the larger PLE1 context (since the mutant was identified in sequenced strains from other groups and not in our collection). We suspect you are correct that the loss of cutting was due to a decreased molar ratio of enzyme to substrate. We have now successfully generated the longer mutant PLE1 origin probe and have repeated the experiment using probes of the same length as in Figure 4D. These new data are shown in revised Figure 6C (replicate cleavage assays are shown in Figure 6 Figure supplement 1). With the longer probe we do see some cleavage of the PLE1^Mut at high concentrations of Odn, however, together with additional in vivo evidence (discussed in response to point 4), our data indicate that the mutant ori is indeed resistant to Odn-mediated cleavage.

We also want to point out that we do not consistently see complete loss of the uncleaved substrate of even the same probe, for example, compare the three replicates of Figure 4C (Figure 4 supplement 3) – we only saw complete loss of the uncleaved substrate for the PLE1 probe in one replicate. We have therefore replaced the gel in the main figure with one of the (former) supplementary gels to reflect that complete cleavage is not the outcome the majority of the time. We are unfortunately not able to explain this minor inconsistency, and given that we make no claims regarding cleavage efficiency in vitro (and do not feel comfortable doing so), we did not quantify cleavage efficiency. We hope that the combination of in vivo and in vitro evidence, which together tell the same story regarding cleavage specificity, is satisfactory.

4) Perform plaque assays for the ori mutant that renders this PLE resistant to Odn Figure 6c. This will be similar to the presentation of data in Figure 5a.

We have now completed these experiments, and the results are presented in Figure 6E. These data show that PLE1^Mut is resistant to Odn(+) ICP1 as anticipated. To take this one step further, we also performed qPCR of the PLE1^Mut strain during ICP1 infection and show that PLE1^Mut replicates during infection and the copy increase is not impacted by Odn (Figure 6D). Collectively, we feel these new data provide more support for our conclusion that this mutation is adaptive in the face of Odn.

5) Please specify that the evidence for Odn activity driving evolutionary change in the PLE is anecdotal.

We have addressed this point: we have removed our original assertion that “These data

suggest that Odn imposes selective pressure on the origin of replication…” and substituted more cautious language, particularly with regard to the data in revised Figure 6: “These data demonstrate that PLE can escape Odn activity through subtle restructuring of the iterons, in addition to more extensive replication module exchange (Figure 3A), and suggest that ICP1 defenses like Odn may select for diversification of the PLE replication machinery.”

As well as in the discussion: “PLEs are able to escape Odn antagonism through mutation of their origin of replication and there is compelling evidence suggesting that PLEs have exchanged replicon modules for an alternative replication origin and initiation factor origin binding domain on at least two separate occasions.”

We feel that we were also appropriately cautious in the language we used in parts of the

original submission (these examples remain unchanged in the revised manuscript):

“The endonuclease appears to exert considerable selective pressure on PLEs

and may drive PLE replication module swapping”;

“It is possible that Odn selected for alternative replicon modules, leading to the

decline of PLE4 and succession of PLE2 that occurred in the early 2000s (O’Hara et al., 2017). […] This seems like a likely possibility for PLEs and ICP1 whose specific adaptations.”

6) Sequence Odn cleavage products and map the precise location of these cuts.

This was something we tried (several times) prior to the initial submission, and in response to reviewer comments made several more attempts. Using Sanger sequencing of Odn cleavage products we were unable to make a clear determination of where cutting was taking place. In most cases we saw uncleaved substrate in our sequencing results (see response to point 3), even when no uncleaved product was visible on the gel and after gel extracting smaller cleaved products. Using higher protein concentrations and longer incubation times did not resolve this issue.

In some of our attempts to address this point, chromatograms obtained by Sanger sequencing did suggest that Odn cleavage of PLE1 is occurring proximal to the poly-T centers of the first and second iterons, but these sequencing results were somewhat inconsistent, and we do not have a robust method of calling sub-sequence ends from the amplitudes of base signals. We therefore do not feel confident enough to include these results in the manuscript.

We also attempted subcloning of cleavage products via TA cloning and blunt end cloning, but were unable to recover cleaved products through these methods.

We have, however, taken additional steps requested by reviewer 1 – including cleavage assays with Odn* and other substrates (see below). Ultimately, although we do not know the precise location where cleavage is taking place, we feel that our in vivo and in vitro results are sufficient to demonstrate Odn targeting through the iterons. We agree that our biochemical characterization of Odn is preliminary and indeed include that this is an area that needs further study: “however the molecular details of Odn binding and catalysis remain to be elucidated.”

Reviewer #1:

[…] Comments for the authors:

We have added images of the purified protein as Figure 4 —figure supplement 2.

In figure 5A, we include Odn* at our maximum concentration of 500nM and detect no cleavage of the PLE1 probe. In response to this comment we have now tested Odn* against other probes (PLE4 and PLE 5) which were cut by Odn in Figure 4D. These new data are presented in Figure 4E (replicate assays are found in Figure 4—figure supplement 4) and show that 500nM Odn* does not cleave probes cut by Odn, consistent with the in vivo data and our conclusions.

Thank you for this suggestion. Please see our response to essential revisions point 3.

4. To come full-circle, sequencing of the cut products would go a long way to support the genetics evidence for Odn targeting within the PLE origin and map the precise location of the cut(s).

Please see our response to essential revisions point 6.

Reviewer #2:

[…] Comments for the authors:

Experimental:

Figure 1: can Odn be re-constituted into a CRISPR-less ICP1 and restore plaquing against PLE1 to demonstrate sufficiency?

Thank you for the suggestion, reconstitution of Odn restoring ICP1 ∆CRISPR plaquing on PLE1 would strongly support our model and was in fact something we tried quite early on. Unfortunately we found that Odn in its native sequence context is too toxic to provide as an ICP1 editing template in V. cholerae (which requires cloning and subsequent maintenance in E. coli and V. cholerae to generate the genome editing strain) so we have not been able to reconstitute its presence within the ICP1 genome.

Thank you for this suggestion. We have now performed these experiments and the results are presented in Figure 6E.

This would indeed be very interesting. We did not attempt this; in our experience so far BIMs arising following infection of V. cholerae with ICP1 are overwhelmingly receptor loss mutants (Seed et al. PLOS Pathogens 2012). Further, while a selection method for PLE escape would be highly useful, the outcome of either ICP1 ‘winning’ or PLE ‘winning’ is cell death, which makes a selection of this sorts extremely challenging: PLE induction ultimately culminates in cell lysis, so even if PLE escape mutants were present, the bacteria harboring those mutant PLEs cannot be recovered. It is possible that PLE escapes (in the form of transductants) could be isolated, though we have previously shown that this does not happen with CRISPR(+) ICP1 (McKitterick et al. 2019), so we did not attempt it for Odn.

Text edits:

Thank you for the suggestion. We have changed the title to: “A chimeric nuclease substitutes a phage CRISPR-Cas system to provide sequence specific immunity against subviral parasites.”

We think this new title retains the information we wanted to convey while being more accessible.

Introduction: the description of MGE and HGT in multicellular organisms in the first paragraph is not necessary to understand this work and possibly too much information for the reader.

These examples, as well as few prokaryotic examples, have been removed to streamline the introduction.

Previously we’ve shown that spacers against PLE 3 are functional when acquired by this phage isolate (O’Hara et al.) The original isolate just happens to not have a spacer directed against PLE3. As for why this is the case, a possible explanation is that PLE3 appears to be much rarer than other PLEs, occurring in a relatively small number of sequenced V. cholerae strains.

We have not yet seen anything besides gp88 or CRISPR-Cas in the region between gp87 and gp91 in sequenced ICP1 isolates – this analysis includes an expanded set of ICP1 isolates collected between 1992-2019 and is part of a review article in press (Boyd CM, Angermeyer A, Hays SG, Barth ZK, Patel KM and Seed KD. 2021. Bacteriophage ICP1: A persistent predator of Vibrio cholerae. Annual Reviews of Virology). Of course we would be very intrigued to find isolates with different genes in that region of the ICP1 genome – we will see what we find in future sampling efforts.

Figure 3: is there a working model for how PLEs recombine (if they even do) to form new variants? Are their isolates of V. cholerae that contain multiple PLEs?

The mosaic sequences of PLEs suggest that PLEs have recombined at some point, but we do not have direct evidence for this recombination taking place. To date, no sequenced isolates have been found to harbor multiple PLEs. Potential explanations for this could be that carriage of multiple PLEs is unstable, or that PLEs recombine following transduction of a PLE into a strain that already has a PLE.

Lines 342-343: "Remarkably, Gp88's own N-terminal domain is 61% similar to 93% of PLE1's RepA_N domain".

We’ve amended this line to say "Remarkably, Gp88's own N-terminal domain is 42% identical and 61% sequence similar to 93% of PLE1's RepA_N domain". (P5 Line 233)

We apologize for any confusion regarding cleavage patterns. While it is true that there is only a 2-bp difference in the iteron sequence between PLEs 4 and 5, the probes used for our cleavage assays encompass a larger sequence harboring the origin of replication and contain numerous differences between the PLEs (the differences can be visualized in Figure 3A). Much of these differences consist of gain or loss of novel sequence. We designed our probes without considerable attention to where the iterons fell within the probes, and the PLE4 probe happened to have the iterons directly in its center. This is predicted to produce two cleavage products of the same size, which would appear as a single band on a gel (consistent with what is observed in Figure 4D).

We realize that this is likely to be a point of confusion for readers so we now address this in the text (P5 lines 265-270) and have added a supplementary figure (Figure 4—figure supplement 5), showing that when different primers are used to amplify the PLE4 probe (resulting in the iteron sequence offset from the center of the probe) the cleavage products clearly run as two bands, as would be predicted.

Discussion:

Lines 551-552: what are some examples of auxillary genes that are present between CRISPR-less ICP1 vs. CRISPR(+) ICP1? Are CRISPR-less ICP1 isolates on average just slightly smaller?

We discuss auxiliary gene diversity and total genome length across ICP1 isolates in a forthcoming review article that’s been accepted for publication (Boyd et al. Annual Reviews of Virology, in press).

In summary, ICP1 isolates range in genome length from ~121kb to ~131kb, and the CRISPR-Cas system is ~7kb. Generally speaking yes, the isolates with the larger genomes tend to be CRISPR(+), but surprisingly, some of the largest ICP1 genomes lack CRISPR-Cas.

Reviewer #3:

[…] Comments for the authors:

This is a very nice study – I enjoyed reading the paper. I only have fairly minor comments/suggestions:

We have streamlined the introduction.

2. I'm left wondering whether BroN and gp90 are essential for the activity of Odn. Have you made knockouts of these genes? Or alternatively, have you expressed Odn from a plasmid in cells with PLE and looked for evidence of cleavage (e.g. cytotoxic effects that are not observed for the catalytic mutant)? You mention they likely have regulatory roles – what is the evidence in support of this idea?

We have noted that WT Odn is cytotoxic in PLE1(+) hosts, but felt that the in vitro cleavage assays and infection host range phenotypes were more compelling than this toxicity phenotype and did not pursue it further.

While investigating the roles of the broN-related gene and gp90 could yield interesting results, we do not think there’s enough space in this manuscript to also characterize their function.

In regard to these genes, we write “A gene encoding a Bro-N domain (pfam02498) and a KilAC domain (pfam03374) occurs adjacent to gp88. Their positions and putative annotations suggest that these divergently transcribed genes may have a regulatory function.” We believe our language is more cautious than saying that they are likely regulators, and merely puts forward a model consistent with their annotations, location, and orientation. It is common for regulators to be divergently transcribed from the genes they control. The BroN domain is thought to be a DNA binding domain, and the KilAC domain is associated with transcriptional antitermination activity.

We have expanded the supplementary figure to incorporate the full coding sequence.

4. The similarity of the nuclease domain to the putative homing endonucleases in ICP1 is intriguing, and provides a plausible explanation for the origin of the nuclease domain. How common are these homing endonucleases? Can the authors make a tree of these nuclease domains to see if this Odn nuclease domain fall inside the tree? Or do you not have enough sequences for this / is sequence divergence too high?

Homing endonuclease are fairly common within lytic bacteriophages though the extent of this varies between phage clades. This family of T5orf172 homing endonucleases appear to be fairly common in ICP1 related phages, though only four HEGs in ICP1²⁰⁰¹ show noticeable similarity to Odn’s nuclease domain. This family of HEGs is discussed in a recent publication about PLE and ICP1 (Netter et al.), a more detailed discussion of ICP1 HEGs and their potential role in ICP1 genome evolution will appear in a forthcoming review (Boyd et al. Annual Reviews Virology in press), and we’re continuing work to understand the relationship between different HEGs in ICP1 and related phages.

As requested, we’ve added a tree of the aligned HEG domains (Figure 4—figure supplement 1). It’s a small number of sequences that are short in length (~200aa), but we think the case for horizontal transfer of the Odn endonuclease domain is compelling.

We have! Please see our response to essential revisions point 6.

Thank you for this suggestion. After reviewing what we had written, we agree that the original statement was over reaching. We’ve amended this passage to read:

“These data demonstrate that PLE can escape Odn activity through subtle restructuring of the iterons, in addition to more extensive replication module exchange (Figure 3A), and suggest that ICP1 defenses like Odn may select for diversification of the PLE replication machinery.”

7. Line 544 – the Gomez et al. Science paper from the same group would also be appropriate to cite.

Thank you for this suggestion, we have added this citation.