Anti defense systems are systems that act against defense system, and thus help bypass the defense mechanism.
# Anti-defense systems:
This article is non-exhaustive but introduces the topic of an-defense systems. Several reviews mentioned here did a great in-depth characterization of the known anti-defense phage mechanism :ref{doi=10.3389/fmicb.2023.1211793,10.1016/j.jmb.2023.167974,10.1038/nrmicro3096}.
Several strategies allow phages to avoid bacterial defenses to successfully complete an infectious cycle. In particular, anti-defense proteins are bacteriophage proteins that specifically act against a bacterial defense system, and thus allow bacteriophages to bypass the bacterial immune system. The most well-described category of anti-defense proteins are the anti-CRISPR proteins (Acr), that have been thoroughly reviewed previously :ref{doi=10.1038/nrmicro.2017.120,10.1016/j.jmb.2023.167974}. However, concomitant with the renewed interest of the field to identify new bacterial defense systems, many anti-defense proteins targeting diverse defense systems have recently been described. Using a non-exhaustive list of anti-defense proteins as examples, I will outline several general categories of anti-defense mechanism. However, I will not focus on another common phage anti-defense strategy that relies on modifying their components, such as mutating the proteins that trigger the defense to escape, or changing their DNA to avoid targeting by restriction-modification or CRISPR systems.
Anti-defense proteins are crucial to understand the evolutionary arms race between bacteria and their phages, as they likely drive the diversification of bacterial defense systems. Some defense system even evolved to recognize anti-defense proteins as activators, providing multiple lines of defense during phage infection :ref{doi=10.1016/j.cell.2023.02.029}. Moreover, these proteins are also important to mediate phage/phage interactions. Indeed, anti CRISPR proteins were suggested to be involved in phage/phage collaboration, in which a primo-infection by a phage carrying an anti-CRISPR protein is unsuccessful but leaves the bacteria immunosuppressed and therefore sensitive to a second phage infection :ref{doi=10.1016/j.jmb.2023.167974}. Considering the importance of overcoming bacterial defenses for phages, it is likely that a significant part of the phage proteins of unknown function currently found in phage sequenced genomes act as anti-defense. Some anti-defense proteins were shown to colocalize in phage genomes, suggesting comparative genomics could be used to identify now anti-defense proteins, similar to what has been done very successfully for bacteria :ref{doi=10.1038/s41467-020-19415-3}. In general, recent studies have used a range of screening methods to identify new anti-defense proteins, and it is expected that many new anti-defense proteins will be described in the coming years.
## Anti-defense proteins target all stages of bacterial defenses
Most anti-defense proteins described to date directly bind a bacterial defense protein to block its activity. However, several other strategies have been described such as post-translational modification of a target, spatial segregation or signaling molecule degradation :ref{doi=10.3389/fmicb.2023.1211793}. They have been described to target all stages of bacterial defense.
Bacterial defenses can be separated in two broad categories: external and internal defenses.
### External defenses
Bacteria can hide receptors behind surface structures such as extracellular polysaccharides or capsular polysaccharides. Conversely, phages can produce various depolymerases to degrade the protective extracellular polysaccharides :ref{doi=10.1038/nrmicro3096}.
### Internal defenses
Bacteria encode a variety of defense systems that prevent phage infection from progressing in various ways. Despite all this variability, all bacterial defense systems are schematically composed of three parts: a sensor recognizing the infection, an effector that achieves protection and a way to transmit the information between the sensor and the effector, either through signaling molecules or protein-protein interactions. Phage anti-defense proteins can target all three of these components.
- Sensor targeting:
- Competitive binding to the sensor: an anti-DSR2 protein from phages phi3T and SPbeta can bind the bacterial DSR2 protein and prevent the physical interaction between DSR2 and its phage activator, the tail tube protein :ref{doi=10.1038/s41564-022-01207-8}. Moreover, Ocr protein from T7 can mimic a B-form DNA oligo and acts as a competitive inhibitor of bacterial type I restriction modification systems :ref{doi=10.1016/s1097-2765(02)00435-5}.
- Masking the activator: some jumbo phages are able to produce a nucleus-like proteinaceous structure that hides phage DNA and replication machinery away from DNA-targeted systems such as type I CRISPR system :ref{doi=10.1038/s41564-019-0612-5}.
- Transmission targeting:
- Degradation of signaling molecules: many systems rely on the production of a nucleotidic signaling molecule after phage sensing to activate the effector such as Pycsar, CBASS, and Thoeris systems. Phages possess proteins that can degrade these molecules to prevent effector activation, such as the anti CBASS Acb1 from phage T4 and the anti Pycsar Apyc1 from phage SBSphiJ :ref{doi=10.1038/s41586-022-04716-y}.
- Sequestration of signaling molecules: an alternative strategy is to bind the signaling molecule very tightly without degrading it, which still prevents effector activation but is presumably easier to evolve than a catalysis-dependent degradation. These phage proteins are called sponges, and two were identified as anti-Thoeris: Tad1 from phage SBSphiJ7 and Tad2 from phage SPO1 and SPO1L3 :ref{doi=10.1038/s41586-022-05375-9,10.1038/s41586-023-06869-w}.
- Effector targeting:
- Direct binding to block activity: Multiple anti-CRISPR protein have been described that can directly bind all the different components of the Cas complex to prevent DNA degradation :ref{doi=10.3389/fmicb.2023.1211793,10.1146/annurev-genet-120417-031321}. So far, this is the most abundant category of anti-defense protein described, and it is not restricted to only anti-CRISPR proteins.
- Antitoxin mimicking: toxin-antitoxin defense systems rely on a toxin effector and an antitoxin that will toxin-mediated toxicity in absence of phage infection. Phages can highjack this process by mimicking the antitoxin to prevent toxin activity even during infection. For instance, phage ϕTE can produce a short repetitive RNA that mimics the ToxI RNA antitoxin of type III toxin-antitoxin system ToxIN and evade defense mediated by this system :ref{doi=10.1371/journal.pgen.1003023}.
Argonaute proteins comprise a diverse protein family and can be found in both prokaryotes and eukaryotes :ref{doi=10.1016/j.mib.2023.102313}. Despite low sequence conservation, eAgos and long pAgos generally have a conserved domain architecture and share a common mechanism of action; they use a 5′-phosphorylated single stranded nucleic acid guide (generally 15-22 nt in length) to target complementary nucleic acid sequences :ref{doi=10.1038/nsmb.2879} eAgos strictly mediate RNA-guided RNA silencing, while pAgos show higher mechanistic diversification, and can make use of guide RNAs and/or single-stranded guide DNAs to target RNA and/or DNA targets :ref{doi=10.1016/j.mib.2023.102313}. Depending on the presence of catalytic residues and the degree of complementarity between the guide and target sequences, eAgo and pAgos either cleave the target, or recruit and/or activate accessory proteins. This can result in degradation of the target nucleic acid, but might also trigger alternative downstream effects, ranging from poly(A) tail shortening and RNA decapping :ref{doi=10.1016/J.CELL.2018.03.006} or chromatin formation in eukaryotes :ref{doi=10.1038/s41580-022-00528-0}, to abortive infection in prokaryotes :ref{doi=10.1016/j.tcb.2022.10.005}.
## Molecular mechanism
Based on their phylogeny, Agos have been subdivided in various (sub)clades. eAgos are generally subdivided in the AGO and PIWI clades, but these will not be discussed further here. pAgos can be further subdivided in long-A pAgos, long-B pAgos, short pAgos, SiAgo-like pAgos, and PIWI-RE proteins :ref{doi=10.1128/mBio.01935-18,10.1016/j.mib.2023.102313,10.1016/j.tcb.2022.10.005,10.1186/1745-6150-8-13,10.1186/1745-6150-4-29}. Below, we briefly outline the general mechanism of pAgos that have a demonstrated role in host defense.
### Long-A pAgos
Akin to eAgos, most long A-pAgos characterized to date have a N-L1-PAZ-L2-MID-PIWI domain architecture :ref{doi=10.1038/nsmb.2879}. In contrast to eAgos, however, certain long-A pAgos use a single stranded guide DNA to bind and cleave complementary target DNA sequences :ref{doi=10.1093/nar/gkz306,10.1093/nar/gkz379,10.1038/s41586-020-2605-1,10.1093/nar/gkv415,10.1038/nature12971,10.1038/nmicrobiol.2017.35}. Long-A pAgos are preferentially programmed with guide DNAs targeting invading DNA through a poorly understood mechanism, which might involve DNA repair proteins :ref{doi=10.1038/s41586-020-2605-1} or the pAgo itself :ref{doi=10.1016/j.molcel.2017.01.033,10.1038/nmicrobiol.2017.34}. Most long-A pAgos have an intact catalytic site in the PIWI domain which allows to cleave their targets :ref{doi=10.1073/pnas.1321032111}. As such, they act as an innate immune system that clear plasmid and phage DNA from the cell :ref{doi=10.1093/nar/gkz379,10.1038/s41586-020-2605-1,10.1093/nar/gkv415,10.1038/nature12971,10.1093/nar/gkad290}.
Within the long-A pAgo clade various subclades of other pAgos exist that rely on distinct function mechanisms. For example, various long-A pAgo can (additionally) use guide RNAs and/or cleave RNA targets. Furthermore, CRISPR-associated pAgos use 5′-OH guide RNAs to target DNA :ref{doi=10.1073/pnas.1524385113}, and PliAgo-like pAgos use small DNA guides to target RNA :ref{doi=10.1038/s41467-022-32079-5}. Certain long-A pAgos genetically co-localize with other putative enzymes including (but not limited to) putative nucleases, helicases, DNA-binding proteins, or PLD-like proteins :ref{doi=,10.1038/nsmb.2879,10.1128/mBio.01935-18}. The relevance of these associations is currently unknown.
### Long-B pAgos
Akin to long-A pAgogs, long B-pAgos have a N-L1-PAZ-L2-MID-PIWI domain composition, but most have a shorter PAZ* domain, and in contrast to long-A pAgos all long-B pAgos are catalytically inactive :ref{doi=10.1128/mBio.01935-18}. Long-B pAgos characterized to date use guide RNAs to bind invading DNA :ref{doi=10.1038/s41598-023-32600-w,10.1016/j.molcel.2013.08.014,10.1038/s41467-023-42793-3}. In absence of co-encoded proteins, long-B pAgos repress invader activity :ref{doi=10.1016/j.molcel.2013.08.014}. In addition, most long-B pAgos are co-encoded with effector proteins including (but not limited to) SIR2, nucleases, membrane proteins, and restriction endonucleases :ref{doi=10.1038/nsmb.2879,10.1128/mBio.01935-18,10.1186/1745-6150-4-29,10.1038/s41467-023-42793-3}. These effector proteins are activated upon pAgo-mediated invader detection, and generally catalyze reactions that result in cell death :ref{doi=10.1038/s41467-023-42793-3}. As such, long-B pAgo together with their associated proteins mediate abortive infection.
### Short pAgos
Short pAgos are truncated: they only contain the MID and PIWI domains essential for guide-mediate target binding :ref{doi=10.1016/j.tcb.2022.10.005}. They are catalytically inactive and are co-encoded with an APAZ domain that is fused to one of various effector domains. In short pAgo systems characterized to date, the short pAgo and the APAZ domain-containing protein form a heterodimeric complex :ref{doi=10.1016/j.cell.2022.03.012,10.1038/s41564-022-01239-0}. Within this complex, the short pAgo uses a guide RNA to bind complementary target DNAs. This triggers catalytic activation of the effector domain fused to the APAZ domain, generally resulting in cell death :ref{doi=10.1016/j.cell.2022.03.012,10.1038/s41564-022-01239-0}. As such, short pAgo systems mediate abortive infection.
Based on their phylogeny, short pAgos are subdivided in S1A, S1B, S2A, and S2B clades :ref{doi=10.1128/mBio.01935-18, 10.1016/j.tcb.2022.10.005}. In clade S1A and S1B (SPARSA) systems, APAZ is fused to an SIR2 domain. In clade S2A (SPARTA) systems, APAZ is fused to a TIR domain. Both SPARSA and SPARTA systems trigger cell death by depletion of NAD(P)+ :ref{doi=10.1016/j.cell.2022.03.012,10.1038/s41564-022-01239-0}. In S2B clade systems, APAZ is fused to one or more effector domains, including Mrr-like, DUF4365, RecG/DHS-like and other domains. In all clade S1A SPARSA systems, but also for certain other systems within other clades, the effector-APAZ is fused to the short pAgo.
### Pseudo-short pAgos
Akin to short pAgos, pseudo-short pAgos are comprised of the MID and PIWI domains only :ref{doi=10.1016/j.tcb.2022.10.005}. However, they do not phylogenetically cluster with canonical short pAgos and do not colocalize with effector-APAZ proteins. Instead, certain pseudo-short are found across the long-A and long-B pAgo clades (e.g. Archaeoglobus fulgidus pAgo, a truncated long-B pAgo :ref{doi=10.1038/s41598-023-32600-w,10.1038/s41598-021-83889-4}), while others form a distinct branch in the phylogenetic pAgo tree (see SiAgo-like pAgos below).
### SiAgo-like pAgos
SiAgo-like pAgos are pseudo-short pAgos that form an separate branch in the phylogenetic tree of pAgos. They are named after the type system from Sulfolobus islandicus :ref{doi=10.1016/j.chom.2022.04.015}. SiAgo is comprised of MID and PIWI domains, and is co-encoded with Ago associated proteins Aga1 and Aga2. SiAgo and Aga1 form a cytoplasmic heterodimeric complex. While it is currently unknown what guide/target types activate the SiAgo/Aga1 complex, it is directed toward membrane-localized Aga2 upon viral infection. This triggers Aga2-mediated membrane depolarization and causes cell death :ref{doi=10.1016/j.chom.2022.04.015}.
## Example of genomic structure
A total of 6 subsystems have been described for the pAgo system.
