# Chapter 111. Comparative Host-Pathogen RNA Interactions Across Viruses, Bacteria, Fungi, Parasites, and Plants

## Scope Note

Host-pathogen RNA interactions are the RNA-centered exchanges, perturbations, and adaptations that occur when viruses, bacteria, fungi, parasites, or plant pathogens encounter a host. This chapter is a comparative synthesis, not an encyclopedic survey of pathogen families. It asks how the same mechanistic axes—RNA sensing, stability, translation, localization, and RNA silencing—operate on physically different host and pathogen RNA objects. It also asks what experiments can separate a causal interaction from a correlated transcriptomic state. Detailed pathogen programs remain with their primary owners: [Chapter 112](chapter1164.md) owns DNA-virus transcript processing, viral noncoding RNAs, host shutoff, latency transcripts, and RNA-mediated immune evasion; Chapters [115](chapter1109.md)-[117](chapter1111.md) own RNA-virus programs; [Chapter 113](chapter1107.md) owns plant RNA immunity; and family-specific mechanisms belong in the relevant organism or pathogen chapters. The present chapter uses selected systems as contrasts and develops the comparison framework, paired-measurement logic, and evidence standards that connect them.

## Executive Summary

Infection changes RNA biology on both sides of the host-pathogen boundary. A pathogen transcriptome is not merely a list of RNAs expressed by an organism; it is a condition-dependent regulatory state shaped by nutrient limitation, temperature, immune pressure, tissue compartment, antimicrobial exposure, and transmission stage. The host transcriptome likewise changes through inducible transcription, alternative splicing, RNA decay, RNA modification, translational control, and small-RNA regulation. These changes can be protective, damaging, compensatory, or epiphenomenal. A central interpretive task is therefore to separate the host RNA object from the pathogen RNA object, identify the compartment and time point in which each was measured, and distinguish RNA changes that cause infection phenotypes from changes that mark cell stress, cell-type composition, or pathogen burden.

Comparison is clearest when each infection is projected onto five shared axes. Sensing asks which RNA features are recognized and by which host or pathogen effector. Stability asks how synthesis, protection, cleavage, and decay establish RNA abundance. Translation asks which RNAs engage ribosomes and how initiation, elongation, or storage changes. Localization asks where an RNA and its ribonucleoprotein complex reside relative to membranes, organelles, replication sites, vacuoles, hyphae, or tissue interfaces. RNA silencing asks whether small-RNA pathways create sequence-directed repression and whether a pathogen suppresses, evades, or exploits that repression. Viral, bacterial, fungal, parasitic, and plant systems share these questions but not necessarily the same molecules or answers.

Dual RNA sequencing (dual RNA-seq), crosslinking and immunoprecipitation (CLIP), structure probing, imaging, ribosome profiling, and perturbation have transformed the field, but none is a universal readout of interaction. Dual RNA-seq measures host and pathogen RNA abundances in the same sample, yet unequal RNA mass, genome annotation quality, host-cell heterogeneity, and pathogen load can create apparent coupling. CLIP detects protein-associated RNA fragments or crosslink sites, not regulatory consequence. Chemical probing reports nucleotide reactivity, not a complete three-dimensional structure. Imaging locates signal at the resolution of the assay, not necessarily a transferred or functional molecule. A defensible host-pathogen RNA claim usually combines measurement, molecular-state assignment, perturbation, rescue or orthogonal validation, and explicit control for time, cell state, organism burden, and assay artifacts.

Coevolution is visible in RNA motifs, sensors, suppressors, and regulatory circuits. Hosts evolve RNA surveillance and RNA silencing systems; pathogens evolve RNA modifications, structured decoys, suppressor proteins, altered codon usage, and regulatory RNAs that exploit or evade host pathways. RNA-targeted interventions therefore include antisense oligonucleotides, small interfering RNAs, host-induced gene silencing in plants, RNA vaccines, RNA-targeted small molecules, and therapies that manipulate host RNA pathways. The practical challenge is specificity: a useful intervention must distinguish pathogen RNA or pathogen-dependent host RNA states from normal host RNA function while avoiding escape, delivery failure, and unintended immune activation.

## Concept Inventory

- **Host-pathogen RNA interaction:** any mechanistic relationship in which infection alters RNA production, processing, modification, structure, localization, translation, decay, or RNA-protein binding in a host or pathogen. A correlated change is an infection-associated RNA state until evidence connects the two sides causally.
- **Host RNA object:** a specified host RNA molecule or population, such as a macrophage cytokine mRNA, a plant small interfering RNA, or a bystander-cell interferon-stimulated transcript, defined together with its cell type and compartment.
- **Pathogen RNA object:** a specified pathogen RNA molecule or population, such as a viral genomic RNA, a bacterial small RNA, a fungal Argonaute-associated small RNA, or a parasite stage-regulated mRNA. The same sequence in different molecular states can constitute different objects for mechanistic interpretation.
- **Shared mechanistic axis:** one of the recurring processes—sensing, stability, translation, localization, or RNA silencing—used here to compare otherwise dissimilar infections.
- **Pathogen transcriptome:** the set of pathogen RNAs present under a defined infection condition, including coding RNAs, noncoding RNAs, antisense transcripts, processed fragments, stage-specific RNAs, and, for RNA viruses, genomic and subgenomic RNAs.
- **Host-induced RNA change:** an RNA change in host cells that appears after pathogen exposure. The term does not by itself imply that the host response is protective or that the pathogen directly caused the change.
- **Dual RNA-seq:** simultaneous sequencing and computational separation of host-derived and pathogen-derived transcripts from the same infected sample.
- **Viral RNA-host protein interaction:** physical binding between a viral RNA and a host protein, including interactions that assist viral replication or translation, restrict infection, remodel localization, or arise as nonfunctional by-products.
- **Cross-kingdom RNA communication:** movement or functional influence of RNA between organisms from different kingdoms, such as plant RNA affecting fungal gene expression. This concept is active but evidence-sensitive because uptake, stability, abundance, and target validation are difficult to establish.
- **RNA-targeted intervention:** a therapeutic, agricultural, or experimental strategy that acts through RNA sequence, RNA structure, RNA processing, RNA translation, RNA decay, or RNA-binding proteins.

## What to Know Before Reading This Chapter

Readers should distinguish RNA abundance from RNA function. An RNA can increase during infection because its gene is induced, because a cell type carrying that RNA became more abundant, because RNA degradation slowed, or because pathogen burden changed. Readers should also distinguish a binding site from a regulatory site. A host protein may bind a pathogen RNA without altering infection, and an RNA structure may be conserved without being essential in every strain or host condition. Finally, host and pathogen are not symmetric measurement targets. A pathogen can represent a tiny fraction of total RNA in early infection, and its genome may be poorly annotated compared with the host genome. These asymmetries make experimental design part of the biology.

For each example, ask four questions in order. What physical RNA was assayed: host or pathogen, coding or noncoding, genomic or transcript-derived, free or RNP-bound? Where and when was it assayed: whole tissue, a sorted cell, a subcellular compartment, an extracellular fraction, or a specific infection stage? Which mechanistic axis is claimed to change? What is the strongest causal test: sequence-specific perturbation, host or pathogen genetics, matched-burden comparison, biochemical reconstitution, compensatory rescue, or infection-outcome rescue? This discipline prevents a host transcriptional response from being mistaken for pathogen RNA regulation and prevents detection of pathogen RNA in a host fraction from being mistaken for functional transfer.

Running examples in this chapter include an RNA virus using structured untranslated regions to recruit host factors; an intracellular bacterium changing small-RNA regulons inside macrophages; a fungal plant pathogen encountering plant small-RNA silencing; and a protozoan parasite shifting mRNA translation during life-cycle transitions. These examples are not interchangeable. Viral RNA genomes, bacterial transcripts, fungal RNAs, parasite RNAs, and plant defense RNAs differ in chemistry, compartment, replication strategy, and evidence standards.

## 111.1. Pathogen transcript states and host transcriptome remodeling across systems

The phrase pathogen transcript state is more precise than pathogen transcriptome when the biological question concerns a particular stage, niche, or molecular pool. For bacteria, fungi, and parasites, the measured state can include cellular mRNAs, antisense RNAs, small regulatory RNAs, untranslated leaders, ribosomal RNAs, transfer RNAs, and processed fragments made from a DNA genome. For RNA viruses, genomic RNA, antigenomic replication intermediates, subgenomic RNAs, defective viral genomes, and viral mRNAs can share sequence while representing different processes. A sequencing run can therefore mix pathogen gene expression, genome replication, translation templates, packaged genomes, and degradation products. The investigator must define which of these objects the assay can resolve.

Host transcriptome remodeling is the corresponding change in host RNA production and use. A macrophage containing an intracellular bacterium may induce cytokine mRNAs, change alternative splicing of immune regulators, alter microRNA abundance, and remodel RNA decay. A neighboring uninfected macrophage can show a similar interferon or cytokine response through paracrine signaling. A plant cell adjacent to an invading fungus may change hormone-regulated mRNAs and produce small interfering RNAs without containing fungal material. A cell infected by an RNA virus may accumulate interferon-stimulated transcripts while only a small fraction of viral RNA engages a particular sensor. The phrase host remodeling is descriptive until perturbation shows whether the state increases resistance, promotes tolerance, causes pathology, or is merely collateral stress.

Several mechanisms can generate an infection-associated host RNA signature. Pattern-recognition receptors can induce transcriptional programs; cytokines can create paracrine RNA responses in uninfected neighboring cells; cell death can enrich RNA from resistant cell types; pathogens can suppress host transcription or mRNA export; and stress granules, processing bodies, or viral replication compartments can redistribute RNAs. Alternative splicing can alter immune signaling proteins without a large change in total gene expression. Circular RNAs and long noncoding RNAs are often reported in bacterial or viral infection studies, but many such associations remain correlative unless the study demonstrates expression above background, specific molecular partners, perturbation effect, and rescue.

Pathogen RNAs shift because the host is a spatially and temporally changing environment. Intracellular bacteria encounter pH, metal limitation, antimicrobial peptides, reactive oxygen and nitrogen species, nutrient restriction, and confinement within host-derived compartments. Fungi encounter surface barriers, immune proteins, temperature shifts, antifungal metabolites, and transitions between spores, yeast-like cells, or hyphae. Parasites encounter vector-to-host transitions, extracellular and intracellular stages, different tissues, and immune pressure. These conditions change regulons for adhesion, invasion, secretion, metabolism, stress resistance, immune evasion, and transmission. A pathogen RNA state observed in rich medium is therefore not a neutral baseline for every in-host state; growth rate and developmental stage must be separated from host-specific regulation.

Host and pathogen states can be coupled in at least three ways. Direct coupling occurs when a molecule on one side acts on RNA on the other—for example, a host RNA-binding protein binds viral RNA or a plant small RNA engages a fungal silencing pathway. Environmental coupling occurs when both respond to the same condition, such as hypoxia or nutrient limitation, without direct RNA contact. Compositional coupling occurs when a sample contains more of one infected cell type or more pathogen biomass. Only the first relationship is a direct molecular interaction, although all three can be biologically informative. Time courses and matched-burden comparisons help separate them: a host RNA change that precedes reduced pathogen burden and is lost after host-gene perturbation supports a defensive role more strongly than a change proportional to late tissue damage.

The evidence basis for this subsection is strongest when host and pathogen RNAs are measured together and interpreted with burden controls. Westermann, Barquist, and Vogel define the design and inference logic of dual RNA-seq. Pisu and colleagues used dual RNA-seq of *Mycobacterium tuberculosis*-infected macrophages in vivo to identify host and pathogen programs that differ across infection contexts, illustrating the value of paired transcriptomes for intracellular infection. Firdous and colleagues synthesize how mRNA decay contributes to fungal pathogenesis, while paired analyses by Mukherjee and colleagues in *Plasmodium* infection and Forrester and colleagues in visceral leishmaniasis demonstrate that parasite-host transcript states depend on species, tissue, and study design. These sources support comparison of state regulation and evidence limits; they do not make fungal and parasite mechanisms interchangeable.

> **Box 111.1. Burden Is Both Biology and Bias**
>
> Box content (render-ready Markdown):
>
> Pathogen burden is not a nuisance term that can always be divided away. A rising viral genome count, bacterial colony-forming burden, fungal biomass estimate, or parasite stage load may be the biological outcome under study. The same quantity can also distort RNA interpretation. More pathogen material increases pathogen-mapped reads; more damaged tissue changes host-cell composition; more inflammatory recruitment changes the apparent host transcriptome. A careful analysis therefore separates three questions. First, what is the burden metric: genome copies, viable organisms, imaging area, spike-in-normalized reads, or histological lesion size? Second, is the RNA claim about burden itself or about regulation at matched burden? Third, do time, cell type, and tissue compartment explain the signal better than a direct regulatory mechanism? When burden cannot be matched experimentally, state that the RNA signature is burden-associated rather than infection-program specific.

One practical way to read a pathogen transcriptome is to ask which variables were controlled before assigning biological meaning. A virulence gene that appears induced in tissue compared with broth culture may reflect host-specific regulation, but it may also reflect slower growth, oxygen limitation, stationary-phase entry, or enrichment of a subpopulation that survived host killing. A host interferon-stimulated gene signature may reflect infected cells, bystander cells exposed to cytokines, or a sorted immune population that entered the tissue after infection began. Strong interpretation therefore requires a chain of comparisons: uninfected host controls, pathogen-alone controls when biologically meaningful, burden-normalized infected samples, time courses that separate early sensing from late damage, and perturbations that test whether the RNA change alters the infection outcome. This logic is especially important for chronic infections, granulomas, biofilms, abscesses, and plant lesions, where the sample is a mixture of microenvironments rather than a uniform infected cell population.

## 111.2. Shared mechanistic axes: sensing, stability, translation, localization, and RNA silencing

A comparative framework is useful only if its categories describe mechanisms rather than superficial similarities. This chapter uses five axes because each can be defined as a causal process with an input, molecular machinery, measurable output, and boundary condition. The axes interact: sensing can activate RNases or translational arrest; localization can protect RNA from sensors; RNA silencing can accelerate sequence-specific decay; and translation changes can alter RNA stability. Nevertheless, separating the axes initially makes it possible to ask the same disciplined questions in viral, bacterial, fungal, parasitic, and plant systems. Table 111.1 summarizes the different RNA objects and evidence standards across pathogen classes.

**Table 111.1. RNA regulation themes across pathogen classes.** Different pathogen classes share RNA-regulatory principles but differ in molecular substrate, compartment, and evidence standards.

| Pathogen class | Major RNA molecules | Infection-stage RNA regulation | Host RNA interaction | Common evidence types | Major caveats |
| --- | --- | --- | --- | --- | --- |
| **RNA viruses** | Genomic and antigenomic RNAs, subgenomic RNAs, viral mRNAs, defective genomes, structured UTRs, IRESs, frameshift elements, and packaging signals. | Replication-template use, subgenomic RNA production, translation-replication switching, RNA modification, RNP remodeling, and packaging control. | Viral RNAs bind host RBPs, translation factors, decay enzymes, and innate sensors; double-stranded or uncapped features can trigger antiviral signaling. | Viral RNA interactome capture, CLIP or pulldown, SHAPE-MaP or DMS-MaPseq, ribosome profiling, synonymous-mutant rescue, and infection assays. | The same viral sequence can occur in genome, mRNA, replication intermediate, and defective-RNA pools; binding or reactivity alone does not prove function. |
| **DNA viruses** | Viral mRNAs, alternatively spliced transcripts, viral noncoding RNAs, some viral microRNAs, and host RNAs remodeled during infection. | Temporal transcription programs, splicing, nuclear export, polyadenylation, RNA stability, translation control, and suppression of host RNA processing. | Viral RNAs and proteins co-opt host splicing, export, translation, decay, and immune-regulatory pathways; host RNA changes often reflect cell-cycle or interferon state. | RNA-seq, splicing analysis, CLIP or RIP for viral proteins, small-RNA sequencing, perturbation of viral ncRNAs, and host-pathway rescue. | Viral RNA is usually not the genome, so RNA-centered claims must distinguish transcript regulation from DNA replication and chromatin effects. |
| **Bacteria** | mRNAs, small RNAs, antisense RNAs, riboswitches, RNA thermometers, UTRs, processed fragments, rRNAs, and tRNAs. | Small-RNA regulons, sigma-factor programs, iron and nutrient responses, secretion-system tuning, stress adaptation, biofilm programs, and regulated RNA decay. | Host microRNAs and immune mRNAs change during infection; bacterial extracellular or vesicle-associated RNAs may stimulate or modulate host responses in defined systems. | Dual RNA-seq, bacterial small-RNA sequencing, transcription-start mapping, reporter assays, target validation, mutant infection models, and burden-normalized time courses. | Growth rate, pathogen load, medium-to-host shifts, incomplete annotation, and indirect regulon effects can mimic infection-specific RNA regulation. |
| **Fungi** | Fungal mRNAs, small RNAs, Argonaute-associated RNAs, RNAi pathway products in species that retain RNAi, and extracellular vesicle RNAs. | Invasion-stage transcription, hyphal or stress-state programs, RNAi-linked regulation, and tissue-adapted metabolism or immune evasion. | Plant or host RNAs may enter fungal cells in some systems, and fungal RNAs may affect host targets; bidirectional cross-kingdom RNAi is context-dependent. | Host-pathogen RNA-seq, small-RNA sequencing, extracellular vesicle purification, Argonaute dependence, target-cleavage assays, genetic rescue, and infection phenotyping. | Cross-kingdom claims require uptake, stability, abundance, target engagement, and silencing-component dependence; not all fungi use RNAi. |
| **Protozoan parasites** | Stage-specific mRNAs, trans-spliced mRNAs and polycistronic precursors in trypanosomatids, untranslated eIF2α-regulated or DOZI/PUF-associated mRNAs in apicomplexans, ncRNAs, and organellar transcripts. | Spliced-leader trans-splicing and coupled polyadenylation in trypanosomatids; global initiation control, transcript-selective RNP storage, and stage-triggered release in *Plasmodium* or *Toxoplasma*. | Parasite RNA states coincide with host immune and tissue transcriptomes; secreted or vesicle-associated parasite RNAs are plausible but require strict delivery evidence. | Trans-splice-junction analysis, stage-resolved RNA-seq, ribosome profiling, proteomics, RBP mapping, and perturbation of eIF2 kinases, phosphatases, helicases, or RBPs. | Evidence is concentrated in selected kinetoplastid and apicomplexan models; precursor versus mature RNA and abundance versus translation must be resolved, and these mechanisms cannot be projected onto helminths. |
| **Helminths** | Developmental mRNAs, miRNAs and other small RNAs, extracellular vesicle RNAs, secreted-product-associated RNAs, and tissue-stage transcriptomes. | Stage, sex, tissue niche, and host-exposure programs; post-transcriptional control is likely important but unevenly resolved across species. | Helminth RNAs and vesicle cargo are proposed to influence host immune cells, while host tissue RNAs reflect inflammation, repair, and parasite burden. | Parasite transcriptomics, small-RNA and vesicle RNA profiling, uptake imaging, host immune perturbation, burden controls, and target-validation assays. | Multicellular stage mixtures, low parasite material, uncertain RNA delivery, and limited target validation make many mechanistic claims provisional. |
| **Plant pathogens** | Plant siRNAs and miRNAs, viral RNAs, pathogen mRNAs, pathogen small RNAs, host-induced silencing triggers, and extracellular vesicle RNAs. | RNA silencing, viral suppressor activity, host-induced or spray-induced gene silencing, defense hormone transcriptomes, and pathogen adaptation to plant tissues. | Plant small RNAs can target pathogens in selected systems; pathogens can suppress silencing or deliver RNAs that affect plant defense pathways. | Small-RNA sequencing, degradome or target-cleavage mapping, Argonaute immunoprecipitation, silencing-mutant genetics, vesicle controls, and HIGS or SIGS challenge assays. | [Chapter 113](chapter1107.md) gives fuller treatment; detection of mobile RNA is not proof of regulation, and delivery, off-target effects, ecology, and resistance must be tested. |

**Sensing** begins when a receptor or effector distinguishes an RNA feature from the tolerated background. In mammalian antiviral immunity, double-stranded RNA, exposed 5′-triphosphate ends, cap status, length, localization, and RNP context can contribute to recognition; deeper receptor mechanisms belong to [Chapter 108](chapter1103.md). Plants can convert viral or pathogen-derived double-stranded RNA into small interfering RNAs through Dicer-like pathways, with sequence-directed Argonaute effectors acting downstream; [Chapter 113](chapter1107.md) owns the plant immune machinery. Bacterial or fungal RNA detected by animal hosts is more often one component of a larger microbial pattern than a self-replicating cytosolic genome. Thus, “foreign RNA sensing” is not one conserved receptor pathway. The common logic is feature discrimination followed by an effector response, whereas the recognized molecules, compartments, and consequences differ.

Sensing also has to be separated from the post-transcriptional control that follows it. Pattern-recognition receptor signaling induces inflammatory RNAs, but their amplitude and duration are then shaped by cis-elements in the host transcripts and the RNA-binding proteins that read them. Tristetraprolin (TTP; ZFP36), for example, recognizes AU-rich elements in cytokine 3′ untranslated regions and promotes deadenylation, decapping, and exonucleolytic decay. Regnase-1 and Roquin can recognize overlapping stem-loop-bearing inflammatory mRNAs yet act in different molecular settings: Regnase-1 cooperates with the helicase UPF1 on translating, endoplasmic-reticulum-associated transcripts, whereas Roquin recruits deadenylation and decapping factors to untranslated messenger ribonucleoproteins. These mechanisms connect sensing to the stability and translation axes without making the RNA-binding proteins pathogen sensors themselves.

**RNA stability** is the net result of synthesis, maturation, protection, and decay. A viral genomic RNA may be stabilized by a 5′ cap, a structured end, a viral nucleoprotein, a host RNA-binding protein, or sequestration in a replication compartment. Host RNases and surveillance pathways can instead cleave or destabilize the same RNA. A bacterial small RNA can expose or occlude a ribosome-binding site and recruit an RNase, thereby changing both translation and mRNA half-life. In parasites, stage conversion can depend on RNA-binding proteins that stabilize stored mRNAs until a later developmental transition. The measured abundance at one time point cannot identify which rate changed. Pulse labeling, transcriptional shutoff, metabolic labeling, decay time courses, and perturbation of specific protective or degradative factors are needed to distinguish increased synthesis from increased stability.

**Translation** asks whether an RNA engages ribosomes, at what initiation site and reading frame, and with what productive output. Positive-sense viral genomes can alternate between translation and replication; structured internal ribosome entry sites or frameshift elements can alter initiation or protein ratios. Intracellular bacteria adjust translation as nutrient and stress conditions change, and their small RNAs can regulate access to ribosome-binding sites. Protozoan parasites with extensive post-transcriptional control may store mRNAs in one stage and translate them after entering a new host or vector compartment. Host translation can simultaneously become selective: general initiation may be inhibited while stress-response or immune mRNAs remain translated. Ribosome profiling helps locate ribosome-protected fragments, but matched RNA-seq, frame periodicity, protein measurement, and perturbation are required before calling productive or regulated translation.

**Localization** specifies where an RNA can encounter a ribosome, sensor, nuclease, membrane, or partner RNA. Viral replication organelles or condensate-like assemblies can concentrate templates and enzymes while limiting sensor access. Bacterial transcripts are measured inside pathogen cells that may occupy phagosomes, cytosol, extracellular biofilms, or tissue microcolonies, each with different chemical constraints. Fungal RNAs can reside within hyphae, extracellular vesicle fractions, or host-contact structures; these locations cannot be inferred from bulk abundance. Parasite RNAs may be divided among life-cycle stages, subcellular organelles, or extracellular products. Spatial transcriptomics, fluorescence in situ hybridization, fractionation, and live imaging can resolve aspects of location, but resolution, contamination, and molecular identity determine whether the data show proximity, compartment membership, or true transfer.

**RNA silencing** is sequence-directed repression mediated by small RNAs and effector proteins, not a synonym for every decrease in RNA abundance. Plant antiviral and antifungal defenses provide clear examples, and some fungal species retain functional RNA interference pathways that can participate in host-pathogen exchange. Other fungi have lost canonical RNA interference components, so the pathway cannot be assumed from taxonomy alone. Animal microRNAs can tune host inflammatory or stress programs, but a change in microRNA abundance is not equivalent to antiviral RNA interference. Bacterial small RNAs also act through base pairing, yet their chaperone- and RNase-dependent regulation is mechanistically distinct from Argonaute-centered eukaryotic silencing. The shared abstraction is sequence-informed regulation; the enzymes, small-RNA biogenesis, target-pairing rules, and biological reach must remain system-specific.

![Figure 111.1. A five-axis framework for coupled host and pathogen RNA states](../assets/figures/chapter1106_figure1.png)

**Figure 111.1. A five-axis framework for coupled host and pathogen RNA states.** Cross-system comparison becomes coherent when host and pathogen RNA objects remain separate and are analyzed along the same mechanistic axes.

The five-axis framework prevents two common errors. The first is treating an RNA-virus mechanism as the default for every pathogen: a bacterial mRNA transcribed within a bacterium is not analogous to a cytosolic viral genome merely because both are pathogen RNA. The second is treating the host response as a single object: infected cells, bystander cells, recruited immune cells, and damaged tissue can have different RNA states. A valid comparison therefore names the RNA object, organism, cell type, compartment, time, and causal operation. The payoff is genuine synthesis: one can compare how different pathogens solve access to translation or evade RNA decay without pretending that they use homologous molecules.

## 111.3. Viral RNA-host interactions as a comparative case

Viruses create an unusually direct RNA-host interface because an RNA-virus genome can itself be the pathogen genome, replication template, messenger RNA, structural scaffold, packaging signal, sensor ligand, and decay substrate. This multiplicity makes RNA viruses a useful comparative case, but not a universal template. DNA viruses produce RNAs that use host processing and translation machinery without making those transcripts the viral genome; detailed DNA-virus transcript programs belong to [Chapter 112](chapter1164.md). Bacterial, fungal, and parasitic pathogen RNAs remain inside cellular pathogens unless a specific export or transfer mechanism is demonstrated. The relevant comparison is therefore not “all pathogen RNAs behave like viral genomes,” but “all pathogen systems must regulate RNA access to machinery, compartments, and host defenses.”

> **Box 111.2. One Viral RNA, Several Molecular States**
>
> Box content (render-ready Markdown):
>
> A viral RNA should be described by molecular state, not only by sequence name. The same positive-sense RNA segment may be translated by ribosomes, copied by viral polymerase, packaged into a virion, paired with an antigenome, bound by a host RNA-binding protein, modified by host enzymes, or detected by innate immune sensors. These states can share sequence but differ in structure, modification occupancy, protein binding, subcellular location, abundance, and accessibility. A CLIP peak on viral RNA may come from translating RNA rather than replication-template RNA. A structure-probing signal may average genomes, subgenomic RNAs, defective genomes, and replication intermediates. An antibody-enrichment m6A peak may not reveal whether the same site is occupied in translating and packaged molecules. An antiviral sensor signal may come from a rare double-stranded or uncapped species rather than the dominant mRNA pool. Strong conclusions therefore name the assayed state whenever possible and use fractionation, time course, site-specific perturbation, or mutant rescue to connect that state to function.

Viral RNA-host protein interactions include proviral and antiviral binding. A proviral host factor may stabilize viral RNA, promote translation, recruit membranes, help form replication compartments, or suppress immune recognition. An antiviral host factor may bind viral RNA to trigger restriction, cleavage, sequestration, or innate immune signaling. Some proteins do both depending on timing and virus. For example, a host RNA-binding protein that enhances translation of one viral RNA can also promote decay of another RNA or help expose double-stranded RNA to sensors. LaPointe and colleagues' characterization of Sindbis virus RNA-host protein interactions provides one example of mapping viral RNA-associated host proteins, but functional interpretation requires follow-up perturbation rather than protein lists alone.

![Figure 111.2. Viral RNA as genome, mRNA, scaffold, sensor ligand, and protein-binding platform](../assets/figures/chapter1106_figure2.png)

**Figure 111.2. Viral RNA as genome, mRNA, scaffold, sensor ligand, and protein-binding platform.** Viral RNA function cannot be reduced to protein-coding capacity.

RNA structure is often central to viral infection. Picornavirus internal ribosome entry sites are structured RNA elements that recruit translation machinery through mechanisms distinct from canonical cap-dependent initiation; Li and colleagues review type IV internal ribosome entry site structure and function. Frameshift elements can alter protein ratios by causing ribosomes to slip at defined sites. Packaging signals can distinguish viral genomes from cellular RNA. Replication elements can recruit viral polymerases or organize template switching. Defective viral genomes and copy-back RNAs can be potent immune ligands because they may contain double-stranded or 5′ triphosphate features recognized by innate sensors.

The SARS-CoV-2 study by Huston and colleagues shows both the power and the ceiling of infection-context structure probing. The investigators treated infected Vero E6 cells with the selective 2′-hydroxyl acylation reagent NAI, used overlapping approximately 2-kilobase amplicons to obtain replicated SHAPE-MaP profiles across almost the entire viral genome, and identified 40 low-reactivity, low-entropy regions in ORF1ab. Evolutionary analysis prioritized a smaller subset with support from synonymous-rate constraint or covariation. Locked nucleic acids directed against selected structured regions reduced viral growth relative to nearby or scrambled controls, connecting the map to a phenotype. The experiment therefore supports in-cell structural reactivity, evolutionarily constrained candidate folds, and sequence-addressable vulnerabilities. It does not by itself prove that the growth defects arose only from loss of the proposed fold, because antisense hybridization can also alter sequence accessibility, protein occupancy, translation, or RNA stability; structure-specific causality would be stronger after paired disruption and compensatory restoration.

These examples cross several mechanistic axes. An internal ribosome entry site acts primarily on translation but its fold also affects stability and host-factor binding. A replication element changes localization by recruiting a polymerase or membrane-associated complex. A copy-back RNA can engage sensing despite representing only a minor fraction of total viral RNA. Consequently, an experiment that maps one viral sequence must identify its molecular state. A CLIP peak on genomic sequence may derive from translating genomes, replication templates, packaged RNA, or defective products. Fractionation, time-resolved labeling, polymerase or packaging mutants, and imaging can narrow the state, but every separation method introduces its own biases.

RNA modifications add another layer. N6-methyladenosine (m6A) can affect viral RNA stability, translation, export, packaging, sensing, and host-factor binding, but it is not uniformly proviral or antiviral. Williams, Gokhale, and Horner summarize negative effects on hepatitis C virus particle production, positive effects on influenza A virus gene expression, position-dependent effects of m6A in the hepatitis B virus epsilon element, and cell-type-dependent or apparently conflicting results for Kaposi sarcoma-associated herpesvirus reactivation. The relevant unit is therefore a specified modification site in a specified viral or host RNA state, not “m6A during infection” as a single pathway. Antibody-enrichment maps can be biased by antibody specificity, RNA structure, transcript abundance, and isoform usage; perturbing METTL3-METTL14, YTH-domain readers, FTO, or ALKBH5 can simultaneously change many host RNAs. Nanopore signal models and reverse-transcription signatures introduce different inference limits. A viral RNA modification should be interpreted as a specific chemical and causal mark only when site mapping, occupancy or stoichiometry, RNA-state assignment, mechanism-specific perturbation, and functional data agree.

Condensates and membraneless compartments complicate viral RNA interactions. Viral replication factories, stress granules, and nucleocapsid-like assemblies can concentrate RNA and proteins, but concentration is not proof of a liquid phase or regulatory function. Etibor and colleagues review biomolecular condensates in viral lifecycles. In this chapter, the conservative interpretation is that co-localization, partitioning, and exchange dynamics are observations; a mechanistic condensate claim needs perturbations that separate phase behavior from ordinary binding, aggregation, or membrane-associated replication.

The viral case also illustrates how causal evidence can be layered. Sequence conservation proposes a constrained element. Chemical probing supports a condition-dependent structural model. Protein interaction assays nominate partners. Synonymous disruption followed by compensatory restoration can test whether RNA structure rather than encoded protein matters. Host-factor depletion can test dependence, and rescue with a perturbation-resistant factor can address indirect toxicity. Finally, infection in a relevant tissue or organism tests whether the mechanism survives physiological context. No single layer is mandatory for every question, but the conclusion should not exceed the strongest completed layer. Detailed replication and pathogenesis programs are developed in Chapters [115](chapter1109.md)-[117](chapter1111.md); the purpose here is to extract comparison and evidence logic.

## 111.4. Bacterial, fungal, parasitic, and plant-pathogen RNA interactions

Bacterial pathogens regulate infection through sigma factors, riboswitches, small RNAs, antisense RNAs, RNA thermometers, RNA-binding proteins, and regulated RNA decay. A bacterial small RNA often acts by base-pairing with one or more target mRNAs, sometimes with help from Hfq, ProQ, or another RNA chaperone. Base pairing can expose or occlude a ribosome-binding site, recruit decay machinery, or protect an RNA from cleavage. Infection-associated small RNAs can thereby tune outer-membrane proteins, secretion systems, iron acquisition, quorum sensing, biofilm programs, and stress responses. The crucial comparison with eukaryotic small-RNA silencing is mechanistic: bacterial small RNAs can be sequence-selective without using Dicer or Argonaute, and one small RNA often reshapes a regulon rather than acting as a single-target switch.

The host side of bacterial infection must be analyzed separately from the bacterial side. Host cytokine mRNAs, alternative splice isoforms, microRNAs, and decay programs can change while the bacterium adjusts nutrient- and stress-responsive transcripts within a phagosome or extracellular niche. Riahi Rad and colleagues review host microRNAs in host-bacterial interactions, and Mishra and colleagues provide a specific example involving miR-30e-5p and SOCS1/SOCS3. A mechanistic chain requires more than differential microRNA abundance: the mature microRNA should load into the relevant effector, engage a credible site, change target protein, alter a pathway, and affect infection in a perturbation-and-rescue design. On the pathogen side, deletion of a bacterial small RNA must be interpreted with growth-rate, envelope-stress, and inoculum controls because pleiotropic fitness defects can masquerade as virulence-specific regulation.

Fungal infection adds a different set of boundaries. Some fungi possess active RNA interference machinery; others have reduced or lost canonical components. Fungal growth can shift among spores, yeast-like forms, and hyphae, so bulk RNA abundance may reflect morphology as much as host contact. In plant-fungal systems, plant and fungal small RNAs can participate in selected bidirectional interactions, and extracellular vesicles have been proposed as one route of protected RNA movement. Wang and colleagues reported plant messenger RNAs moving into a fungal pathogen via extracellular vesicles to reduce infection, and Cheng and colleagues reported fungal Argonaute proteins acting in bidirectional cross-kingdom RNA interference. These studies support functional movement in defined systems, not a general rule that every RNA co-purifying with vesicles is transferred or regulatory. Donor production, extracellular protection, recipient uptake, effector loading, target engagement, and sequence-specific phenotypic rescue form the necessary chain.

> **Box 111.3. Cross-Kingdom RNA Claims Need a Full Chain**
>
> Box content (render-ready Markdown):
>
> Cross-kingdom RNA regulation requires a continuous evidence chain. The donor organism must produce the RNA at meaningful abundance; the RNA must leave the donor in a protected form; the RNA must survive extracellular or tissue transit; the recipient must take up the RNA into the relevant cell type or compartment; the RNA must engage a defined target or effector pathway; and changing the RNA sequence or abundance must change the molecular and infection phenotype. Sequencing reads alone satisfy only the detection step. Stronger studies add carrier purification, RNase and detergent controls, imaging or fractionation of uptake, Argonaute or RNA-silencing dependence where appropriate, target-cleavage or reporter evidence, donor and recipient mutants, and sequence-specific rescue. Without this chain, a cautious phrase such as "recipient-associated donor RNA" is more accurate than "cross-kingdom silencing."

Parasitic infections include phylogenetically distant protozoa and helminths, so “parasite RNA regulation” is not one conserved program. In trypanosomatids, most protein-coding genes occur in polycistronic transcription units. A capped segment of the spliced-leader RNA is transferred to each mRNA 5′ end by two transesterification reactions, while coupled polyadenylation helps resolve individual mature mRNAs from the precursor; conventional cis-splicing is restricted to very few transcripts. The transferred leader supplies the unusual cap 4 structure. Günzl's synthesis therefore defines a processing boundary, not merely an exotic splice-site example: gene-level abundance can pool polycistronic precursor, trans-splicing intermediate, and mature capped mRNA unless library design or junction analysis distinguishes them.

Apicomplexans use a different post-transcriptional logic. In *Plasmodium* and *Toxoplasma*, phosphorylation of the alpha subunit of eukaryotic initiation factor 2 (eIF2α) increases sequestration of eIF2 by eIF2B, lowers global translation initiation, and can permit selective translation of stress-adaptive transcripts. Persistent translational repression contributes to quiescent or transmission stages, but it is not identical to transcript-specific storage. In female *Plasmodium* gametocytes, the DDX6-family helicase DOZI and associated RNA-binding proteins maintain selected mRNAs in repressed messenger RNPs for translation after fertilization; PUF-family proteins also control stage-specific transcript use in gametocytes and sporozoites. Holmes and colleagues further describe release of stored transcripts after environmental transition, including translation of regulators that relieve eIF2α phosphorylation after hepatocyte entry. A bulk RNA measurement can therefore miss two distinct operations: a global initiation-state change and selective release of stored mRNAs. These apicomplexan mechanisms should not be projected onto kinetoplastids or helminths. Stage-resolved RNA-seq paired with ribosome profiling or proteomics, RNP localization, and perturbation of the implicated kinase, phosphatase, helicase, or RBP is more informative than a single abundance contrast.

Plant systems are addressed mechanistically in [Chapter 113](chapter1107.md), but they provide essential contrasts. Plant antiviral defense often uses small-RNA amplification and sequence-directed silencing, while viral proteins can suppress small-RNA biogenesis, effector loading, or downstream action. Nakahara and Masuta review interactions between viral silencing suppressors and host factors. Plant-bacterial and plant-fungal infections add cell-wall barriers, apoplastic spaces, plasmodesmata, vascular transport, and hormone-controlled tissue responses to the localization problem. Host-induced and spray-induced gene silencing turn the cross-kingdom hypothesis into an intervention test: an RNA must be stable on or in plant tissue, reach the pathogen, enter an active silencing pathway, and suppress an essential accessible target. Efficacy in one pathosystem cannot be generalized to another without measuring uptake and pathway competence.

Across these cellular pathogens, the shared axes remain useful only when the objects stay explicit. A bacterial small RNA changing a bacterial mRNA, a host microRNA changing a host mRNA, and a plant small interfering RNA proposed to change a fungal mRNA are all sequence-informed regulation, but only the third crosses an organismal boundary. Likewise, a parasite mRNA moving from a repressed to a translated RNP is a localization-and-translation transition, not host-to-pathogen transfer. Box 111.3 states the evidence chain for cross-kingdom claims, and [Chapter 113](chapter1107.md) supplies the deeper plant immune context.

## 111.5. Comparative measurement: dual RNA-seq, CLIP, structure probing, imaging, and spatial methods

Comparative measurement begins by matching the assay to the RNA object and mechanistic axis. Dual RNA sequencing (dual RNA-seq) is powerful because it asks what the host and pathogen are doing in the same biological sample. A host interferon state can be compared with a viral replication state; a macrophage activation state can be related to bacterial iron-starvation transcripts; and a plant defense program can be related to fungal invasion. Yet the output is paired abundance, not direct molecular interaction. Host RNA can overwhelm pathogen RNA, pathogen burden varies, genome annotation may be incomplete, and ribosomal RNA depletion, poly(A) selection, random priming, size selection, and sequencing depth determine which molecule classes are visible.

![Figure 111.3. Evidence ladder for host-pathogen RNA claims](../assets/figures/chapter1106_figure3.png)

**Figure 111.3. Evidence ladder for host-pathogen RNA claims.** A robust claim usually needs multiple orthogonal evidence types.

The most common dual RNA-seq artifact is confounding by cell composition or pathogen load. If a sample has more pathogen RNA, pathogen genes may appear induced simply because there are more pathogen cells or virions. If an infected tissue contains more neutrophils, epithelial cell death, or necrotic material, host transcripts may shift because the sample composition changed. Single-cell and spatial transcriptomics can help separate cell states, but they add new problems: dissociation bias, ambient RNA, low pathogen capture, dropouts, doublets, batch correction artifacts, and uncertain assignment of pathogen reads to individual host cells. Ratnasiri and colleagues analyze these limits for virus-host single-cell RNA sequencing. Holdener and colleagues and Lempke, May, and Ewald place spatial transcriptomics in virus-host and broader microbial pathogenesis contexts, while Sounart and colleagues provide a direct dual-spatial example in fixed human tissue. These methods improve localization and cell-state resolution but do not remove the need to validate pathogen identity and burden.

Spatial methods also trade molecular coverage against spatial resolution rather than improving both automatically. Laser-capture microdissection can preserve a tissue niche but may collect multicellular material and yields limited RNA; in situ sequencing preferentially recovers abundant transcripts; and array-based platforms can assign a spot or bead to one or several cells depending on feature size, tissue geometry, and deconvolution. Lempke, May, and Ewald additionally note that widely used spatial-transcriptomic systems were developed for eukaryotic RNA and often require adaptation to enrich low-abundance or prokaryotic pathogen transcripts. A high-resolution coordinate is therefore not evidence that the relevant pathogen RNA was captured with high sensitivity. Platform choice should report feature size, transcript coverage, pathogen-enrichment strategy, tissue preservation, and the independent marker used to identify infected cells or microbial structures.

Single-cell data create an important asymmetry between host and pathogen inference. A host cell can often be assigned a transcriptomic state from thousands of host molecules even when it contains only zero, one, or a few captured pathogen reads. Failure to detect pathogen RNA is therefore not proof that the cell was uninfected, while one ambient pathogen read is not proof of intracellular infection. Imaging, pathogen-reporter strains, surface-marker strategies, or independent burden measurements can calibrate classification. Spatial assays preserve tissue context and can relate lesions, granulomas, hyphal fronts, or infected foci to RNA states, but spot-level mixing and diffusion of extracellular RNA limit cellular assignment. Resolution must be reported in physical and molecular terms.

CLIP-family methods map RNA-protein contacts by crosslinking protein-RNA complexes, immunoprecipitating a protein, and sequencing associated RNA fragments. In infection biology, CLIP can identify host RNAs bound by viral proteins, viral RNAs bound by host proteins, or pathogen transcripts bound by pathogen RNA-binding proteins. The method can distinguish direct binding from co-expression better than ordinary RNA-seq, but it is not a functional assay by itself. Crosslinking biases, antibody specificity, protein abundance, RNA abundance, digestion conditions, and alignment ambiguity all affect the result. A CLIP peak near a viral RNA structure is a hypothesis about binding and possible function, not proof that the interaction controls replication.

Pulldown and interactome-capture experiments answer a related but distinct question. An RNA bait can enrich proteins that bind directly, proteins connected through an RNP, or proteins attracted by a conformation not present in cells. In-cell capture can preserve physiological interactions but may still favor abundant or readily crosslinked proteins. Comparing wild-type and structure-disrupting baits, using abundance-matched controls, validating sites by CLIP, and testing host-factor perturbation narrows the inference. Even then, a host protein can bind viral or pathogen RNA as a consequence of restriction or degradation rather than as a factor required for pathogen fitness.

Chemical structure probing methods, including SHAPE-MaP and DMS-MaPseq families, measure nucleotide reactivity and infer local RNA flexibility or base-pairing probability. Applied to infection, structure probing can map viral genome structure in virions or infected cells, compare RNA structure across temperatures or host species, and test whether drugs or proteins remodel an RNA element. Huston and colleagues' in-cell analysis of the SARS-CoV-2 genome illustrates how a probing map can nominate regulatory elements and then require mechanism-specific follow-up. These methods do not directly solve complete RNA tertiary structure. Reactivity can be affected by protein binding, modification, RNA abundance, reverse-transcription bias, cell permeability, and ensemble averaging. Viral RNAs are especially challenging because replication intermediates, genomic RNAs, subgenomic RNAs, and defective RNAs may share sequence but differ in structure and protein occupancy.

Imaging and spatial localization are indispensable for claims about transfer or compartmentalization. Fluorescence in situ hybridization can locate a sequence, but probe specificity, optical resolution, autofluorescence, and section thickness determine whether puncta represent single molecules, aggregates, or adjacent compartments. Metabolic labeling can mark newly synthesized RNA but may label both organisms unless precursor access is controlled. Extracellular-vesicle imaging must distinguish vesicle association from free or surface-bound RNA. Co-localization supports proximity at the assay resolution; direct binding requires a different method, and functional transfer requires recipient-side target engagement.

**Table 111.2. Method outputs and interpretation hazards in host-pathogen RNA studies.** Each method has a specific evidentiary ceiling unless combined with perturbation and validation.

| Method | Primary output | Strong inference | Weak or invalid inference alone | Key controls |
| --- | --- | --- | --- | --- |
| **Dual RNA-seq** | Host and pathogen transcript abundances assigned from the same infected sample. | Paired host and pathogen RNA states, especially when normalized for burden, time, cell type, and library design. | Direct regulation, causal virulence, or pathogen gene induction without controls for load, growth state, and sample composition. | Uninfected and pathogen-alone controls where meaningful, burden measurement, rRNA-depletion or poly(A)-selection rationale, genome annotation checks, spike-ins or normalization, and time courses. |
| **Single-cell dual RNA-seq** | Host cell states plus sparse pathogen-assigned reads at cell resolution. | Infected versus bystander host-cell programs and cell-type-specific infection responses when pathogen capture is sufficient. | Absence of pathogen reads as proof a cell is uninfected, or low-count pathogen reads as reliable pathogen transcriptomics. | Ambient RNA correction, doublet filtering, imaging or FISH validation, cell-type markers, pathogen-load calibration, batch controls, and cautious integration. |
| **Spatial host-pathogen transcriptomics** | Host and pathogen-assigned transcripts associated with image-defined regions, spots, beads, or in situ coordinates. | Tissue-niche and proximity hypotheses when feature size, coverage, pathogen capture, and independent infection markers are calibrated. | Single-cell infection assignment from nominal feature size alone, or absence of microbial RNA as evidence of absence when the platform undercaptures low-abundance or prokaryotic transcripts. | Feature-size and transcript-coverage reporting, tissue-preservation controls, pathogen-specific probes or enrichment, independent imaging or reporter validation, deconvolution checks, and matched negative tissue. |
| **CLIP-seq** | Crosslinked protein-associated RNA fragments or crosslink sites. | Candidate direct binding regions for a host, viral, or pathogen RNA-binding protein under the assayed condition. | Regulatory consequence, infection phenotype, or binding to the active RNA pool without perturbation or orthogonal validation. | Antibody or epitope validation, input and negative controls, knockout or depletion controls when possible, RNase-digestion tuning, replicate peak calling, and rescue assays. |
| **Viral RNA pulldown** | Proteins enriched with a viral RNA bait, region, or structure. | Candidate viral RNA-associated host factors and RNP composition for the chosen RNA element. | Direct in-cell binding, proviral or antiviral function, or structure-specific binding from enrichment alone. | Mutant or scrambled bait, abundance-matched controls, RNase or competition tests, mass-spectrometry background filters, CLIP validation, and functional perturbation. |
| **SHAPE-MaP or DMS-MaPseq** | Nucleotide reactivity profiles converted into local flexibility or base-pairing constraints. | Condition-specific RNA structural reactivity and candidate structural changes in viral, host, or pathogen RNAs; evolutionary support and antisense inhibition can prioritize functional regions. | Complete tertiary structure, single-conformation certainty, or structure-specific causality from an antisense growth defect that may also alter accessibility, RNP binding, translation, or decay. | Untreated and denatured controls, biological replicates, coverage thresholds, abundance and molecular-state filters, protein-binding and modification caveats, matched antisense controls, and paired disruptive plus compensatory structure rescue. |
| **Small-RNA sequencing** | Size-selected small-RNA species and relative abundance. | Infection-associated microRNAs, siRNAs, bacterial sRNAs, or processing signatures when library bias and mapping are controlled. | Target repression, cross-kingdom transfer, or immune function from detection or differential abundance alone. | Adapter-bias mitigation, spike-ins or UMIs where appropriate, multi-mapping rules, Argonaute loading, degradome or reporter validation, biogenesis mutants, and RNase or vesicle controls. |
| **Ribosome profiling** | Ribosome-protected RNA fragments mapped to coding regions or small ORFs. | Translational engagement, frame usage, pausing candidates, and infection-linked translation shifts after quality filtering. | Protein abundance, productive translation, or direct RNA regulation without matched RNA-seq and functional validation. | Footprint-size periodicity, nuclease and inhibitor bias checks, matched RNA-seq normalization, ORF annotation controls, replicate concordance, and proteomic or mutational follow-up. |
| **Extracellular vesicle RNA profiling** | RNAs that co-purify with extracellular vesicle fractions from host, pathogen, or mixed samples. | Candidate vesicle-associated RNA cargo and infection-linked extracellular RNA signatures after rigorous purification. | RNA delivery, uptake, target regulation, biomarker specificity, or cross-kingdom communication from profiling alone. | Density or size-exclusion purification, vesicle and contamination markers, RNase with and without detergent, uptake imaging, target-engagement assays, rescue tests, and burden controls. |

The strongest studies combine methods. A viral RNA element might be nominated by conservation and structure probing, shown to bind a host protein by CLIP or pulldown, disrupted by synonymous mutations that preserve protein sequence, rescued by compensatory mutations that restore structure, and linked to infection phenotypes in relevant cells or organisms. A bacterial small RNA might be detected by dual RNA-seq, mapped by transcription-start and termination data, connected to targets by base-pairing prediction and reporter assays, and tested in infection models with growth controls. A cross-kingdom RNA claim might combine extracellular vesicle purification, RNase protection controls, uptake imaging, target cleavage evidence, Argonaute dependence, and genetic rescue.

Interpretation should also respect the scale of each assay. Bulk dual RNA-seq averages across infected cells, uninfected bystander cells, immune infiltrates, and pathogen subpopulations. Single-cell methods improve resolution but often capture only a few pathogen reads per cell, making absence of pathogen RNA a weak negative result. CLIP can resolve binding neighborhoods but usually cannot tell whether a protein binds the active replicating RNA molecule or an inactive pool. Structure probing can reveal a reproducible reactivity pattern but may average across multiple conformations. These limitations do not make the methods unreliable; they define the claim each method can support. A useful chapter-level rule is to phrase the conclusion in the language of the assay unless causal perturbation has been done: RNA-seq shows abundance, CLIP shows association, probing shows reactivity, ribosome profiling shows ribosome-protected fragments, and genetic rescue shows functional dependence.

The same rule applies to negative evidence. Failure to detect an RNA can reflect true absence, insufficient depth, poor annotation, incompatible library selection, low pathogen burden, a brief time window, or an inaccessible compartment. Failure of perturbation can reflect redundancy, incomplete knockdown, delivery failure, or an assay performed outside the relevant stage. Strong negative claims therefore require a demonstrated detection limit and a positive control in the same molecular context. Figure 111.3 and Table 111.2 organize these evidentiary ceilings and controls.

## 111.6. Coevolution, cross-kingdom RNA exchange, intervention, and causal-inference limits

Host-pathogen coevolution is visible wherever RNA features are repeatedly selected by recognition, evasion, or exploitation. Hosts evolve sensors that recognize double-stranded RNA, 5′ triphosphate RNA, unmethylated or mislocalized RNA, and foreign RNA replication intermediates. Pathogens evolve RNA caps, RNA modifications, structured hiding strategies, suppressor proteins, decoy RNAs, and compartmentalization to avoid detection. Plants evolve small-RNA defense pathways, while plant viruses encode silencing suppressors. Bacteria and phages evolve RNA-guided defense and anti-defense systems. The evolutionary logic is not simply host defense versus pathogen evasion; pathogens also exploit host RNA pathways for translation, localization, RNA stabilization, and immune modulation.

Comparative evolutionary claims need an explicit unit of selection. Conservation of an RNA sequence can reflect encoded protein, RNA structure, replication signals, overlapping reading frames, or mutational bias. Enrichment of a host genotype can reflect pathogen resistance, tolerance of pathology, or a linked non-RNA trait. Parallel use of RNA silencing in two systems does not demonstrate common ancestry of the infection mechanism. Evidence becomes stronger when sequence covariation supports structure, mutations separate RNA from protein effects, host and pathogen variants show reciprocal phenotypes, or repeated evolution converges on the same constrained RNA feature.

Cross-kingdom RNA exchange is a particularly demanding coevolutionary hypothesis because the proposed RNA crosses physical, enzymatic, and cellular boundaries. The donor must produce the RNA; the RNA must exit in a stable form; it must traverse extracellular space or tissue; the recipient must internalize it into the relevant compartment; an effector must engage it; and a target must change in a sequence-dependent way. Contamination, index hopping, shared sequences, vesicle co-purification, and damaged-cell leakage can produce donor-like reads without this chain. The Wang and Cheng studies provide strong examples in particular plant-fungal systems, but each link must be re-established when the organisms, carrier, tissue, or environmental condition changes.

RNA-targeted interventions can act on pathogen RNA directly or on host RNA pathways that are required during infection. Direct approaches include antisense oligonucleotides, small interfering RNAs, CRISPR-associated RNA-targeting systems, ribozymes, and small molecules that bind structured pathogen RNAs. Indirect approaches include manipulating host microRNAs, blocking proviral host RNA-binding proteins, changing innate immune recognition, or using RNA vaccines to induce protective immunity. In agriculture, host-induced gene silencing and spray-induced gene silencing aim to use RNA molecules to suppress pathogen genes. The same intervention logic appears in human medicine and plant protection, but delivery, specificity, durability, and ecological risk differ sharply.

![Figure 111.4. Coevolutionary routes to RNA-targeted intervention and escape](../assets/figures/chapter1106_figure4.png)

**Figure 111.4. Coevolutionary routes to RNA-targeted intervention and escape.** RNA-targeted strategies must be evaluated with evolutionary robustness and delivery constraints.

RNA-targeted therapy has attractive specificity because sequence complementarity can distinguish pathogen RNAs from host RNAs. However, pathogen diversity and escape are major obstacles. An antisense or siRNA target must be conserved across relevant strains, accessible in the RNA structure and RNP context, and essential enough that escape mutations reduce fitness. Host-directed interventions face the opposite problem: targeting a host factor can reduce pathogen escape, but it risks toxicity because host RNA pathways usually have normal cellular roles; the systematic review by Shapira and colleagues supplies an infection-specific boundary for that tradeoff without implying that every host-directed intervention is RNA-mediated. In plant protection, Mann and colleagues review RNA-based control of fungal pathogens, and Chen and colleagues examine how formulation, nanotechnology, uptake, and cross-kingdom trafficking constrain crop protection. Spray-induced gene silencing (SIGS) makes the delivery chain experimentally explicit: double-stranded RNA must remain intact during ultraviolet exposure, rainfall, and extracellular nuclease challenge; avoid premature processing when long RNA is the uptake substrate; enter the relevant pathogen or pest; escape an endocytic compartment when necessary; and reach active RNA-interference machinery. Uptake competence and the value of long double-stranded RNA versus processed small interfering RNA differ among organisms. Nanoparticles can increase surface residence time, protect RNA, and alter internalization or endosomal escape, but carrier composition, size, charge, release kinetics, and non-target effects become part of the mechanism rather than neutral formulation details. These infection-specific sources support the comparative checklist here; modality pharmacology remains with Chapters [153](chapter1137.md)-[162](chapter1145.md).

The coevolutionary view also changes how resistance is studied. A viral RNA structure that tolerates many synonymous substitutions may escape a structure-binding compound; a bacterial small-RNA target may be bypassed by regulatory rewiring; a fungal pathogen may lose an RNA uptake route if it is not essential; a parasite may change stage timing or tissue tropism. Therefore, intervention design should include evolutionary robustness tests: target conservation, structural constraint, fitness cost of escape, combination with other mechanisms, and surveillance for resistance-associated RNA changes.

Delivery is the other major axis of feasibility. An RNA drug or silencing trigger must reach the infected compartment at sufficient concentration while remaining stable and avoiding unwanted immune activation. Respiratory viral targets, intracellular bacterial reservoirs, fungal hyphae in plant tissue, protozoan parasites inside host cells, and helminths in extracellular niches pose different barriers. A sequence-perfect siRNA is not useful if it cannot enter the relevant pathogen or host cell, and a broad innate immune stimulant is not a precise RNA-targeted intervention if its benefit comes from nonspecific inflammation. For agricultural RNA strategies, environmental persistence, uptake by non-target organisms, formulation, cost, and resistance monitoring are part of the biological assessment. For human therapy, tissue delivery, renal and hepatic clearance, complement activation, cytokine induction, and drug-drug interactions can determine whether an elegant RNA target becomes a practical treatment.

Causal inference remains the common bottleneck. A host-directed intervention can reduce pathogen burden by changing immunity, metabolism, or cell viability rather than by the proposed RNA interaction. A pathogen-directed RNA perturbation can impair general growth rather than infection-specific regulation. A rescue experiment is strongest when it restores the proposed molecular step without restoring unrelated defects: a target-site-resistant transcript can test small-RNA specificity, a compensatory mutation can restore RNA structure, and a pathway-specific effector mutant can test silencing dependence. Burden should be treated twice—first as a biological outcome and then as a possible mediator or confounder of every downstream RNA measurement.

The intervention chapters provide modality-specific pharmacology and delivery; this chapter owns the comparative decision logic. A candidate should be evaluated across target identity, molecular accessibility, compartment, delivery route, host dependence, pathogen diversity, escape cost, off-target biology, immune consequences, and ecological context. The same checklist can reject attractive but incomplete stories: sequence complementarity without delivery, tissue accumulation without target engagement, target engagement without infection benefit, or short-term benefit without resistance robustness.

## Recent Consensus

Current consensus supports several broad conclusions. First, infection produces coupled but physically distinct host and pathogen RNA states; paired measurement does not erase that distinction. Second, sensing, stability, translation, localization, and RNA silencing provide useful comparative axes only when the RNA object, organism, compartment, and time are explicit. Host RNA-binding proteins can connect these axes by determining whether induced inflammatory transcripts are translated or degraded. Third, viral RNAs are structured, modified, and protein-bound functional molecules, but an RNA-virus genome is not the default model for cellular pathogens; neither structure probing nor modification mapping alone establishes a causal fold or chemical mechanism. Fourth, bacterial, fungal, parasitic, and plant-pathogen systems use diverse post-transcriptional mechanisms whose machinery and evidence depth vary by lineage. Trypanosomatid trans-splicing and apicomplexan eIF2α-dependent translation control are informative contrasts, not interchangeable parasite rules. Fifth, dual RNA-seq, CLIP, structure probing, imaging, and spatial methods each have a defined evidentiary ceiling; causal claims usually require perturbation and orthogonal validation, and spatial resolution does not guarantee pathogen-transcript sensitivity. Sixth, functional cross-kingdom RNA exchange is established in selected systems but cannot be inferred from extracellular RNA detection alone. Finally, RNA-targeted interventions remain constrained by delivery, molecular accessibility, immune effects, escape, host toxicity, and ecological context; for SIGS, RNA carrier and uptake biology are parts of the causal mechanism.

## Open Questions, Controversies, Deprecated Models, and Common Misconceptions

Open questions:

- How often cross-kingdom RNA transfer is quantitatively large enough to regulate infection?
- Which viral RNA structures are conserved because of direct function rather than replication constraints?
- Which virus-, cell-, RNA-state-, and site-specific m6A effects survive stoichiometric measurement and modification-site-specific perturbation?
- How pathogen RNA states differ between in vitro infection models and natural tissues?
- How host cell-type heterogeneity should be modeled in mixed infected samples?
- Which spatial-transcriptomic designs can retain tissue context while capturing low-abundance bacterial, fungal, and parasite RNAs with calibrated sensitivity?
- Which aspects of RNA stability and translation are conserved across parasite stages, and which are lineage-specific?
- When an infection-associated RNA state is reproducible, what minimum perturbation establishes that it changes resistance, tolerance, transmission, or pathology?
- Whether many infection-associated long noncoding RNAs and circular RNAs are regulatory agents or markers of stress, cell composition, and RNA processing changes?

Common misconceptions:

- "RNA-virus mechanisms can be generalized to all host-pathogen RNA biology." Cellular bacteria, fungi, parasites, and plant pathogens have different genome substrates, compartments, processing systems, and evolutionary pressures.
- "Expression proves causality." Expression changes are associations until perturbation and rescue connect the RNA change to the phenotype.
- "Co-localization proves direct binding." Co-localization shows spatial overlap at assay resolution; direct binding needs biochemical, structural, or proximity evidence with controls.
- "Conservation without perturbation proves function." Conservation is strong evidence for constraint, but function still requires mechanism-appropriate validation.
- "Extracellular RNA detection alone proves cross-kingdom regulation." Detection establishes presence in a fraction; functional exchange additionally requires donor origin, protection, uptake, effector engagement, target regulation, and sequence-specific phenotype.
- "A cell without captured pathogen reads is uninfected." Sparse capture and dropout make absence of pathogen reads a weak negative result unless imaging or another calibrated infection marker agrees.
- "RNA silencing means any reduction in RNA abundance." RNA silencing is a mechanism with sequence-directed effectors; lower abundance can instead result from reduced synthesis, nonspecific decay, cell-composition change, or burden differences.