# Chapter 104. Comparative Plant, Fungal, Protist, and Other Non-Animal Eukaryote RNA Biology

## Scope Note

This chapter compares RNA biology across plants, fungi, ciliates, kinetoplastids, apicomplexans, algae, and other non-animal eukaryotes. Its primary task is comparative synthesis: it asks how conserved RNA-processing parts have been retained, lost, duplicated, or assigned new regulatory jobs in different lineages. Detailed mechanisms remain with their pathway owners, including organellar RNA genes in [Chapter 17](chapter1016.md), plant immunity in [Chapter 113](chapter1107.md), RNA editing in Chapters [50](chapter1046.md)-[51](chapter1047.md), RNA decay in Chapters [32](chapter1031.md)-[38](chapter1036.md), and small-RNA pathways in Chapters [80](chapter1075.md)-[88](chapter1083.md). Here those mechanisms are revisited only far enough to support organism-level comparison, evidence evaluation, and biological interpretation.

## Executive Summary

Non-animal eukaryotes are not a residual category left after animal RNA biology. They encompass several of the clearest demonstrations that RNA regulation is an evolvable architecture. Plants connect Dicer-like enzymes and Argonaute proteins to development, antiviral defense, secondary small-RNA amplification, transposon control, and RNA-directed DNA methylation. Fungi reveal both retention and repeated loss of canonical RNA interference, while using RNA decay, splicing, translational control, and ribonucleoprotein remodeling to manage development, nutrient shifts, host entry, and drug stress. Protist lineages include guide-RNA-directed reconstruction of mitochondrial coding sequences, spliced-leader trans-splicing of polycistronic transcripts, small-RNA-guided genome rearrangement, unusual plastid and mitochondrial expression systems, and life-cycle programs controlled mainly after transcription.

The same label can conceal different mechanisms. A plant miRNA, a fungal siRNA, and a ciliate scan RNA are all small RNAs, but their precursors, protein partners, targets, compartments, and biological outputs differ. Conversely, similar regulatory jobs can be performed by unrelated or extensively diverged components. Silencing a transposon may involve RNA-directed DNA methylation in a flowering plant, an RNAi-linked heterochromatin system in fission yeast, or small-RNA-guided DNA elimination in a ciliate. Comparative reasoning therefore separates molecular homology from functional analogy.

Model organisms provide mechanistic resolution but produce sampling distortions. *Arabidopsis* cannot stand for all land plants; *Saccharomyces cerevisiae* is an RNAi-loss lineage rather than a generic fungus; trypanosomes do not represent all protists; and the term “algae” groups photosynthetic eukaryotes with different evolutionary origins. An apparent absence of a pathway can reflect true evolutionary loss, an incomplete genome, a divergent sequence missed by homology search, expression restricted to one life stage, or replacement by a nonhomologous factor. An apparent presence of a homolog does not establish its substrate or biological function.

This chapter uses an organism-by-topic ownership matrix to make those boundaries explicit. Plant small-RNA circuits, fungal RNA regulation, ciliate genome remodeling, kinetoplastid editing, apicomplexan translational control, and algal RNA systems receive enough explanation for comparison. Detailed reaction chemistry and pathway-specific breadth are handed to their dedicated chapters. The resulting comparative picture is broad without becoming an omnibus encyclopedia: each example teaches an evolutionary or evidentiary principle that can be reused when a newly sequenced lineage does not fit a familiar animal-centered model.

## Concept Inventory

- **Comparative RNA architecture:** the lineage-specific combination of transcription, processing, modification, localization, translation, decay, and inheritance mechanisms that produces functional RNA. Comparison concerns both component homology and the regulatory job performed.
- **Plant small-RNA specialization:** diversification of Dicer-like, Argonaute, and RNA-dependent RNA polymerase systems into miRNA, antiviral, phased-siRNA, and heterochromatic-siRNA pathways. Plant small RNAs should not be treated as one interchangeable class.
- **RNA-directed DNA methylation:** small-RNA-guided deposition of DNA methylation at homologous plant loci. Canonical RdDM uses Pol IV, RDR2, DCL3, AGO4-family proteins, Pol V scaffold transcripts, and de novo methyltransferases, but noncanonical entry routes exist.
- **Fungal RNA-interference patchiness:** retention, modification, or loss of Dicer-, Argonaute-, and RNA-dependent RNA polymerase-dependent silencing across fungal lineages. The absence of canonical RNAi in *S. cerevisiae* is one lineage state.
- **Scan RNA:** a ciliate small RNA that helps compare germline-derived sequence information with the developing somatic genome and can guide elimination or retention decisions. Scan-RNA pathways differ among ciliates and are not synonyms for animal piRNA pathways.
- **Kinetoplastid guide RNA:** a small mitochondrial RNA that base-pairs with a pre-edited transcript and specifies uridine insertion or deletion. It is distinct from a CRISPR guide RNA.
- **Spliced-leader trans-splicing:** transfer of a short capped leader exon from a separate RNA to a pre-mRNA. It is widespread in some protist lineages but not universal among protists.
- **Apicomplexan translational control:** life-stage-specific regulation of mRNA storage, recruitment, and translation in parasites such as *Plasmodium* and *Toxoplasma*. The regulatory architecture cannot be inferred from the presence or absence of canonical RNAi alone.
- **Algal polyphyly:** the fact that “algae” describes photosynthetic eukaryotes from multiple lineages rather than one clade. Algal RNA mechanisms must be assigned to a named lineage and organelle history.
- **Phylogenetic pseudoreplication:** treating multiple observations from one narrow clade as independent evidence for a general eukaryotic rule. Robust comparative inference samples independent lineages or explicitly limits the claim.
- **Annotation absence:** failure to detect a gene, transcript, RNA class, or processing event in an annotation. Annotation absence is not equivalent to biological absence until assembly quality, sequence divergence, expression stage, and assay suitability have been considered.

## What to Know Before Reading This Chapter

The chapter assumes basic familiarity with Dicer, Argonaute, RNA-dependent RNA polymerase, spliceosomes, RNA decay, organellar genomes, DNA methylation, and chromatin. Each core term is reintroduced when a lineage uses it in a distinctive way. The reader does not need prior familiarity with the individual model organisms.

Three distinctions organize the chapter. First, **homology** means descent from a common ancestral component, whereas **analogy** means a similar job achieved independently or with substantially different machinery. Second, a **pathway component** is not the same as a demonstrated pathway: finding an Argonaute-like protein does not show which guides it loads or which targets it regulates. Third, a **model organism result** is evidence for that organism under the tested condition, not an automatic statement about a kingdom.

The recurring examples are intentionally diverse. *Arabidopsis* provides genetic dissection of plant small-RNA and chromatin pathways; rice and maize reveal reproductive phasiRNAs and repeat-rich crop genomes; *Neurospora* and fission yeast show distinct fungal silencing systems; budding yeast illustrates pathway loss; *Tetrahymena* and *Paramecium* connect small RNAs to genome rearrangement; trypanosomatids connect guide RNAs to mitochondrial coding-sequence reconstruction; apicomplexans show life-cycle regulation with little or no canonical RNAi in well-studied species; and green algae provide experimentally tractable comparisons outside land plants.

## 104.1. Plant RNA regulation: small RNAs, RNA-directed DNA methylation, processing, and development

Plant RNA regulation is often introduced through microRNAs, but the defining feature is division of labor among several small-RNA pathways. A plant microRNA begins as a capped and polyadenylated RNA polymerase II transcript that folds into an imperfect stem-loop. In flowering plants, Dicer-like 1 (DCL1), the double-stranded-RNA-binding protein HYL1, SERRATE, and other nuclear factors cut the hairpin to release a short duplex. HEN1 methylates the 3′ termini, protecting the duplex from tailing and degradation. One strand is loaded into an Argonaute protein, most often AGO1 for many well-studied miRNAs, and base pairing then directs target slicing, translational repression, or both.

![Figure 104.1. Plant miRNA maturation and Argonaute loading](../assets/figures/chapter1099_figure1.png)

**Figure 104.1. Plant miRNA maturation and Argonaute loading.** Plant miRNA processing is nuclear and HEN1-stabilized, which distinguishes it from simplified animal miRNA diagrams.

Extensive complementarity makes cleavage a common plant miRNA output, and the resulting 5′ end of a cleavage fragment can be mapped by degradome or related end-capture methods. That evidence is stronger than a predicted pairing site, but it is not by itself a developmental mechanism. A causal account links the miRNA precursor, processing factors, Argonaute loading, target-site complementarity, cleavage or repression, target-protein change, and phenotype. Target-site-resistant alleles and spatially resolved expression are particularly informative when a miRNA sharpens a developmental boundary.

> **Box 104.1. Do Not Translate Animal miRNA Logic Directly to Plants**
>
> Plant miRNAs and animal miRNAs are homologous enough to compare, but not interchangeable. A plant miRNA claim should usually specify the primary hairpin, nuclear DCL1-associated processing, HEN1-dependent 3′-end methylation, Argonaute loading, and the target-pairing geometry. Extensive guide-target pairing often supports AGO-catalyzed slicing in plants, so degradome or 5′ RACE evidence can be highly informative. That rule has boundaries: some plant miRNAs repress translation, some targets are context-dependent, and stress or developmental state can change which outcome dominates. Conversely, animal-style seed matching alone is usually too weak for a plant target claim unless supported by plant-specific conservation, accessibility, Argonaute association, perturbation, and site-rescue evidence. The safest summary is not "plants slice and animals repress," but "pairing geometry, Argonaute context, and organism-specific pathway architecture determine the output."

Small interfering RNAs (siRNAs) are more heterogeneous. Virus-derived siRNAs originate from viral replication intermediates or structured viral RNA and support antiviral silencing. Heterochromatic siRNAs usually originate from repeats and transposons and act in chromatin regulation. Phased secondary siRNAs arise in register after an initiating cleavage event, allowing one primary trigger to generate a series of guides. In grasses, reproductive 21- and 24-nucleotide phased siRNAs are abundant in anthers, but abundance and conservation do not by themselves specify a single universal reproductive function. Natural-antisense-derived siRNAs and other conditional classes further show why size alone is insufficient for classification.

Canonical RNA-directed DNA methylation (RdDM) connects small RNAs to chromatin. RNA polymerase IV produces precursors that RNA-dependent RNA polymerase 2 converts into double-stranded RNA. DCL3 generates mainly 24-nucleotide siRNAs, which are loaded into AGO4-family proteins. RNA polymerase V produces scaffold transcripts at target loci, and guide-scaffold pairing helps recruit factors including the de novo DNA methyltransferase DRM2. The output is cytosine methylation in CG, CHG, and CHH contexts and reinforcement of a repressive chromatin state. The pathway is prominent at transposons and repeats, but its effect on nearby genes depends on genomic position, tissue, developmental state, and pre-existing chromatin.

RdDM should be treated as a network rather than a rigid chain. Some loci enter through RNA polymerase II transcripts or alternative Dicers and Argonautes; maintenance methylation and histone modifications can reinforce the state after the initiating small RNA changes. A 24-nucleotide RNA correlated with CHH methylation is therefore a pathway clue, not proof that every canonical component was used. Mutant combinations, small-RNA profiles, methylomes, scaffold-transcript assays, and locus-specific rescue distinguish initiation from reinforcement.

**Table 104.1. Major plant small RNA classes and outputs.** Plant small RNAs should be classified by origin, factors, and output rather than by size alone.

| Class | Precursor | Core factors | Typical size | Main targets | Main output | Evidence caveats |
| --- | --- | --- | --- | --- | --- | --- |
| **miRNA** | Pol II primary hairpin transcript | DCL1, HYL1, SERRATE, HEN1, AGO1-family loading | Usually 21 nt; some variants differ | Complementary mRNAs, often transcription-factor or developmental regulators | Target slicing and context-dependent translational repression | Degradome signal supports cleavage; abundance change alone does not prove direct targeting. |
| **Virus-derived siRNA** | Viral double-stranded RNA or replication intermediates | Dicer-like processing, Argonaute loading, sometimes RDR amplification | Commonly 21-22 nt | Viral RNAs and homologous viral sequences | Antiviral silencing and spread of sequence-specific defense | Viral small RNA reads do not by themselves show protective function or target engagement. |
| **Heterochromatic siRNA** | Pol IV transcripts from repeats, transposons, or related heterochromatic loci | RDR2, DCL3, AGO4-family proteins, Pol V scaffold transcripts, DRM methyltransferases | Typically 24 nt | Homologous repetitive loci, transposons, scaffold transcripts | RNA-directed DNA methylation and repressive chromatin | Requires methylation and pathway-factor evidence; repetitive read mapping can misassign loci. |
| **Trans-acting siRNA** | Noncoding TAS-like precursor cleaved by an initiating miRNA | Initiating miRNA, RDR-dependent secondary-siRNA synthesis, Dicer-like phased processing, AGO loading | Usually 21 nt phased products | mRNAs regulated in trans, often gene-family members | Secondary post-transcriptional repression beyond the initiating miRNA target | Needs initiator cleavage and phasing evidence; a short RNA from a TAS locus is not enough. |
| **Phased siRNA** | Precursor diced in register from a defined start site | Triggering cleavage or processing event plus RDR/Dicer-like phased production | Often 21 nt; pathway-specific classes can differ | Pathway-specific mRNAs or genomic loci | Amplified guide sets with ordered register | Phasing must be measured from a start site; size alone does not define this class. |
| **Natural antisense siRNA** | Overlapping sense and antisense transcripts forming double-stranded RNA | Antisense transcription, Dicer-like processing, Argonaute loading, context-dependent RDR use | Commonly 21-24 nt | One or both overlapping transcripts or linked stress-response genes | Cue-linked post-transcriptional silencing or local regulation | Overlap and small RNA detection do not prove functional antisense silencing. |

Plant comparisons extend beyond *Arabidopsis*. Crop genomes are often larger, repeat-rich, polyploid, and structurally variable, changing the substrate on which RdDM and small-RNA systems act. Rice and maize also make reproductive phasiRNAs at scales that are not captured by an *Arabidopsis*-only picture. Mosses preserve some ancient land-plant small-RNA features while differing in gene-family composition and development. These comparisons show that conserved enzyme names do not guarantee the same precursor spectrum or developmental output.

Plant organellar RNA processing is relevant because the nucleus encodes many proteins that bind mitochondrial and chloroplast transcripts. Pentatricopeptide repeat proteins can define editing, cleavage, splicing, stabilization, or translation sites through modular sequence recognition. A mutation may therefore appear as chlorosis, respiratory failure, sterility, or stress sensitivity even though the proximal lesion is RNA processing. The gene content and detailed processing reactions of mitochondria and plastids belong to [Chapter 17](chapter1016.md); the comparative point here is that nuclear expansion of organellar RNA-binding proteins created a major lineage-specific regulatory layer.

Plant development and environmental response integrate these pathways rather than placing them in isolated modules. miRNAs tune transcription-factor families and hormone responses; phasiRNAs can extend an initiating signal; heterochromatic siRNAs protect repeat-rich genomes; and organellar RNA regulation couples nuclear genotype to photosynthesis and respiration. Plant immune mechanisms, including detailed antiviral and antibacterial signaling, are owned by [Chapter 113](chapter1107.md). This chapter retains only the bridge needed to compare plant small-RNA defense with fungal and protist systems.

## 104.2. Fungal RNA interference, RNA decay, splicing, development, and stress responses

Fungal RNA biology is a lesson in evolutionary patchiness. *Neurospora crassa* and many filamentous fungi retain Dicer, Argonaute, and RNA-dependent RNA polymerase systems. The classic quelling response in *Neurospora* converts aberrant or repetitive transgene-derived RNA into a silencing signal. Fission yeast uses RNAi components in heterochromatin formation at repetitive loci. Some fungal lineages use related machinery in antiviral defense, transposon control, meiotic surveillance, developmental regulation, or interactions with hosts. These outputs share guide-dependent recognition but differ in precursor, compartment, chromatin coupling, and biological purpose (Torres-Martínez and Ruiz-Vázquez, 2017).

Even within fungi that retain RNAi, “the fungal pathway” is not one reaction. In *N. crassa* quelling, QDE-3 and the dual DNA/RNA-dependent polymerase QDE-1 help convert signals from repetitive DNA into double-stranded RNA; Dicer-like enzymes generate small interfering RNAs, and QDE-2 Argonaute plus the exonuclease QIP activate target-directed degradation. Sexual-stage surveillance is organized differently: meiotic silencing by unpaired DNA uses dedicated SAD and SMS factors, whereas *Cryptococcus neoformans* mounts sex-induced silencing against transposons. *Mucor circinelloides* separates initiation and amplification between distinct RNA-dependent RNA polymerases and also has a Dicer-independent, RNA-dependent degradation route. Antiviral RNAi can use only a subset of the available Dicer and Argonaute paralogs. These examples make precursor, life stage, factor set, and output essential parts of any fungal RNAi claim (Torres-Martínez and Ruiz-Vázquez, 2017).

![Figure 104.2. Fungal RNA regulatory network under stress](../assets/figures/chapter1099_figure2.png)

**Figure 104.2. Fungal RNA regulatory network under stress.** Fungal RNA regulation is broader than canonical RNAi, and retained RNAi itself is divided by precursor, life stage, factor set, and output.

Canonical RNAi has also been lost repeatedly. *Saccharomyces cerevisiae* lacks the canonical Dicer-Argonaute system, whereas retention states differ even among other budding yeasts. The correct inference is not that fungi lack RNAi or that fungal RNAi is universal. A lineage-level description should identify the Dicer-like enzyme, Argonaute, RNA-dependent RNA polymerase where present, small-RNA size and chemistry, precursor source, target, and mutant phenotype. The demonstration of active RNAi in *Candida albicans* underscores why budding yeasts must be resolved species by species (Iracane et al., 2024). Homology alone cannot establish antiviral defense, and a silencing phenotype alone cannot establish canonical RNAi.

RNA-mediated epimutations demonstrate why fungal silencing matters beyond transposon annotation. In at least some fungi, small-RNA pathways can reversibly suppress a drug-target or drug-response gene, producing resistance without a stable DNA mutation. Loss of silencing can restore sensitivity. Inheritance of an RNAi-linked resistance epimutation has now been demonstrated experimentally in a mucoralean fungal system (Pérez-Arques et al., 2025). This mechanism is emerging and lineage-dependent, not a general explanation for antifungal resistance, but it expands the range of heritable states that clinical and evolutionary studies must consider. It also illustrates how pathway output depends on selection: the same silencing machinery that manages foreign or repetitive RNA can become a route to phenotypic adaptation.

Fungal RNA decay is often more universal than canonical RNAi. Deadenylation shortens the poly(A) tail, decapping exposes the 5′ end, Xrn-family exonucleases degrade RNA from 5′ to 3′, and the exosome degrades or processes RNA from 3′ to 5′. Nonsense-mediated decay recognizes many transcripts whose translation termination occurs in an unfavorable context; no-go and nonstop pathways respond to other translation failures. Regulated Ire1-dependent mRNA decay during physiological endoplasmic-reticulum stress in *Aspergillus oryzae* provides a direct fungal example of transcript destruction coupled to cellular demand (Tanaka et al., 2023). These pathways change transcript lifetimes during nutrient shifts, morphogenesis, sporulation, mating, host entry, and stress. A fungal mRNA increase can therefore reflect faster transcription, slower decay, altered cell-state composition, or failed quality control.

Splicing adds another uneven dimension. *S. cerevisiae* has relatively few intron-containing genes, while many filamentous fungi and fungal pathogens have larger intron repertoires and more regulated isoforms. Intron retention can be a controlled regulatory state, a consequence of stress, a sign of delayed processing, or an annotation artifact. Short-read RNA sequencing often cannot distinguish those alternatives. Full-length long reads, junction-specific assays, nuclear-cytoplasmic separation, protein measurement, and genetic perturbation of splice sites or regulators are needed when a claimed isoform drives development or virulence.

Stress responses bring decay, splicing, and translation together. Heat, oxidative stress, nutrient limitation, host immunity, and antifungal treatment can inhibit translation initiation, change ribosome occupancy, alter transcript half-lives, and redistribute RNAs and proteins into granule-like assemblies. Genetic analysis of processing-body factors in *C. albicans* links these assemblies to filamentation and stress response, two traits relevant to fungal adaptation and pathogenicity (Tosiano et al., 2025). “Stress granule” and “processing body” are useful functional terms, but they are not synonyms, and composition and material behavior vary with species and assay. A fluorescent punctum is not sufficient evidence for a conserved animal-like granule. Protein dependence, RNA content, reversibility, exchange kinetics, and effect on translation or decay provide a stronger definition.

Fungal diversity also creates a sampling problem. A dimorphic human pathogen switches morphology and temperature environment; a plant pathogen encounters immune compounds and spatially heterogeneous tissue; a saprobe experiences fluctuating substrates; and a symbiont occupies a long-term host association. Bulk RNA from infected tissue may mix hyphae, spores, yeast-like cells, dormant cells, and damaged cells. Dual RNA sequencing, careful species assignment, life-stage markers, ribosome profiling, RNA half-life measurements, and single-cell or spatial approaches can separate regulation within a state from changes in the proportions of states.

The primary comparative conclusion is that fungal RNA regulation cannot be ranked on a simple scale from “has RNAi” to “lacks RNAi.” A lineage may lack canonical RNAi yet retain sophisticated decay and translational control, or retain RNAi but assign it a restricted developmental or defense role. Detailed fungal decay chemistry remains with Chapters [32](chapter1031.md)-[38](chapter1036.md), and host-pathogen synthesis remains with [Chapter 111](chapter1106.md). Here the retained ownership is the relationship between pathway retention, fungal ecology, developmental state, and evidence quality.

## 104.3. Ciliate, kinetoplastid, apicomplexan, algal, and other protist RNA systems

“Protist” is a practical umbrella, not a single evolutionary or mechanistic category. It includes free-living predators, parasites, photosynthetic organisms, and symbionts distributed across deeply divergent eukaryotic branches. The useful unit of explanation is therefore a named lineage plus a defined RNA mechanism. Ciliates, kinetoplastids, apicomplexans, and algae are discussed together here to compare regulatory architectures, not because they share one protist RNA program.

### 104.3.01. Ciliates: small RNAs and programmed genome remodeling

Ciliates such as *Tetrahymena* and *Paramecium* separate germline and somatic nuclear functions. The micronucleus preserves the heritable genome, while the macronucleus supports most gene expression. During sexual development, a new macronucleus is built from a germline copy through chromosome fragmentation, amplification, and removal of internal eliminated sequences. Small-RNA pathways help distinguish sequences to retain from sequences to eliminate, linking RNA comparison to irreversible DNA rearrangement in the developing somatic nucleus.

In a simplified scan-RNA model, germline-derived small RNAs associate with Piwi-family proteins and are compared with parental somatic information. Small RNAs representing sequences already present in the somatic nucleus are depleted or functionally filtered, enriching guides that can direct heterochromatin and DNA elimination in the developing macronucleus. The landmark *Tetrahymena* Piwi-family genetic analysis and subsequent comparative work support this RNA-guided model while revealing substantial differences among *Tetrahymena*, *Paramecium*, and *Oxytricha* (Mochizuki et al., 2002; Wang et al., 2017; Stefanov and Nowacki, 2025). Species differ in guide size, timing, direction of comparison, accessory proteins, and the balance between small-RNA and long-RNA information. The mechanism should therefore be taught as a family of ciliate solutions rather than one universal pathway.

Some spirotrich ciliates add another layer: extensively scrambled germline segments are reordered and joined to produce somatic genes. Long maternal RNAs and small RNAs have been implicated in guiding or validating this reconstruction. These systems show that RNA can transmit sequence-order information between generations, but the strength and molecular form of the template-like contribution differ among experiments and species. A ciliate result must identify the nuclear stage, RNA class, rearrangement event, and causal perturbation.

### 104.3.02. Kinetoplastids: guide-RNA editing and trans-splicing

Kinetoplastid mitochondrial RNA editing is the canonical case of guide-directed coding-sequence reconstruction. In trypanosomes and related organisms, many mitochondrial pre-mRNAs do not encode a translatable open reading frame until uridines are inserted or deleted. Small guide RNAs base-pair with an anchor region and specify mismatches that are resolved through cleavage, uridine addition or removal, and ligation by editosome-associated enzymes. Editing can proceed through overlapping blocks, generating partially edited intermediates whose interpretation requires more than alignment to a final mature RNA (Aphasizheva et al., 2020).

![Figure 104.3. Kinetoplastid guide RNA-directed uridine editing](../assets/figures/chapter1099_figure3.png)

**Figure 104.3. Kinetoplastid guide RNA-directed uridine editing.** Some mature mRNA coding sequences are constructed post-transcriptionally by guide RNA-directed editing, but the component reactions are established more firmly than one fixed chronology.

Guide RNAs are often encoded in the kinetoplast DNA network, which contains maxicircle and minicircle molecules. The editing system therefore couples unusual mitochondrial genome organization to RNA maturation. Editing demand and mitochondrial activity vary across parasite life stages, so a pathway measured in an insect-stage trypanosome cannot automatically be transferred to a bloodstream stage. Genetic depletion, complex purification, guide-RNA mapping, intermediate sequencing, and biochemical assays together support the reaction model. Detailed catalytic chemistry belongs with the RNA-editing chapters; the comparative lesson is that the mature coding sequence can be produced by a distributed genome-plus-guide system.

Editing is embedded in a larger maturation and quality-control system. Maxicircle mRNAs and ribosomal RNAs and minicircle guide RNAs are synthesized as 3′-extended precursors. A mitochondrial 3′ processome combines terminal uridylation with 3′-to-5′ trimming; messenger RNAs additionally receive short A tails whose stabilizing effect depends on editing state, whereas long A/U extensions are associated with fully edited, translation-competent messenger RNAs. Pentatricopeptide-repeat proteins connect internal editing progress to terminal modification and stability. The ordering of all these events is not fully established, so a pathway diagram should distinguish demonstrated reactions from a single inferred chronology (Aphasizheva et al., 2020).

Trypanosomatid nuclear expression is also dominated by post-transcriptional regulation. Long polycistronic transcription units are processed into individual mRNAs through coupled spliced-leader trans-splicing and polyadenylation. A separately transcribed spliced-leader RNA donates a short exon bearing the hypermodified cap4 structure to the 5′ end of many mRNAs. The polypyrimidine tract and downstream splice acceptor help select the leader-addition site, and changes in that choice can alter an upstream open reading frame, an amino-terminal targeting sequence, or the 5′ untranslated region. Because trans-splicing at one gene is mechanistically coupled to 3′-end formation of the upstream gene, alternative trans-splicing can also change upstream polyadenylation and 3′-untranslated-region length. Many mapped alternatives still lack protein-level validation, so site heterogeneity should not automatically be interpreted as adaptive isoform regulation. Because promoter-by-promoter control is less prominent than in familiar animal genes, RNA-binding proteins, processing sites, mRNA stability, and translation carry much of the life-stage regulatory burden. Trans-splicing is not unique to trypanosomatids, but its near-global use in these organisms makes it an organizing principle (Michaeli, 2011). This comparative chapter owns the lineage context; [Chapter 27](chapter1026.md) owns the shared spliceosomal chemistry, distinctions from cis and recursive splicing, and artifact-controlled detection logic.

### 104.3.03. Apicomplexans: life-cycle control without a canonical RNAi template

Apicomplexan parasites such as *Plasmodium* and *Toxoplasma* alternate among sharply different host environments and cell types. Well-studied *Plasmodium* species do not provide a canonical Dicer-Argonaute RNAi system, yet they execute strong post-transcriptional control. In female gametocytes, messenger ribonucleoprotein complexes containing factors such as DOZI and CITH store selected mRNAs in a translationally repressed state until transmission and development in the mosquito. Release from repression then permits proteins to be produced from RNAs synthesized earlier (Mair et al., 2010; Holmes et al., 2017).

This example prevents the equation of “no canonical RNAi” with “little RNA regulation.” Apicomplexans use at least two conceptually distinct routes to reduce translation. Selective messenger ribonucleoprotein storage keeps particular transcripts available for later use, whereas stress-responsive phosphorylation of eukaryotic initiation factor 2 alpha reduces bulk initiation while permitting selective translation of stress-adaptation transcripts. The latter integrated-stress-response logic contributes to entry into and maintenance of latent or quiescent states in experimentally studied *Toxoplasma* and *Plasmodium* systems, but the responsible kinases, environmental signals, and life-cycle consequences differ between parasites. Evidence for translational repression therefore requires comparing mRNA abundance with ribosome occupancy or protein production and perturbing the relevant storage factor or initiation-control pathway. A transcript present in a gametocyte or bradyzoite is not necessarily being translated there (Holmes et al., 2017).

Messenger RNA 3′-end formation is another regulated layer. Apicomplexans conserve the core polyadenylation-signal recognition and endonucleolytic-cleavage module but have diverged substantially in factors that recognize upstream and downstream sequence elements. In *Toxoplasma gondii*, CPSF4 is not merely a sequence-assigned homolog: CPSF4 copurifies with the cleavage and polyadenylation specificity factor complex, contains an N6-methyladenosine-reading YTH domain, and its conditional depletion causes transcriptional readthrough or alternative polyadenylation at a subset of loci. Depleting the methyltransferase METTL3 produces overlapping defects, supporting a conditional connection between N6-methyladenosine and proximal 3′-end choice rather than a universal rule for every transcript. The CPSF3 endonuclease is independently supported as a parasite vulnerability because benzoxaborole inhibitors bind its catalytic center and resistance maps to CPSF3 substitutions. These biochemical and resistance data support target engagement, but conservation of the catalytic center makes parasite selectivity a separate medicinal-chemistry problem. This “patchwork” architecture is best established for tractable *Plasmodium*, *Toxoplasma*, and *Cryptosporidium* systems and should not be projected unchanged onto every apicomplexan (Swale and Hakimi, 2023).

Apicomplexans also contain the apicoplast, a nonphotosynthetic plastid derived from secondary endosymbiosis. Apicoplast transcripts require organelle-specific expression and processing machinery, while many required proteins are nucleus encoded and imported. The detailed gene inventory belongs to [Chapter 17](chapter1016.md). The comparative point is that organellar RNA regulation reflects a layered evolutionary history: a parasite nucleus services an organelle descended from a plastid acquired through an ancestral eukaryotic endosymbiont.

### 104.3.04. Algae and other photosynthetic eukaryotes

Algae are polyphyletic. Green algae are relatively close to land plants, whereas red algae, diatoms, dinoflagellates, and other photosynthetic groups have different plastid origins and nuclear histories. It is therefore unsafe to infer an “algal pathway” from one *Chlamydomonas* experiment. The named organism, nuclear lineage, plastid type, and culture condition are part of the claim.

The green alga *Chlamydomonas reinhardtii* provides tractable genetics for small-RNA pathways, chloroplast gene expression, RNA stability, and translational control. Its Dicer-like and Argonaute systems are informative relatives of land-plant pathways but do not reproduce the complete flowering-plant RdDM architecture. Genetic and biochemical work on *Chlamydomonas* Argonaute establishes both conserved miRNA-effector logic and lineage-specific division of labor (Chung et al., 2019). Chloroplast transcripts are strongly controlled after transcription through RNA-binding proteins, processing, stabilization, and translation. These features make green algae useful for separating ancient photosynthetic-eukaryote mechanisms from land-plant innovations.

Diatoms and other secondary-plastid lineages add different regulatory constraints, including complex protein targeting and ecological responses to nutrients and light. Dinoflagellates use unusual nuclear organization and widespread spliced-leader trans-splicing. A conserved 22-nucleotide leader was recovered from diverse nuclear messenger RNAs across major dinoflagellate orders, including nuclear genes encoding proteins targeted to organelles. In the tested species, the same leader was not detected on organelle-encoded messenger RNAs or ribosomal RNAs. This substrate boundary is important: the result supports a broad nuclear messenger-RNA processing system, not indiscriminate addition to every cellular RNA (Zhang et al., 2007). Dinoflagellate organellar genomes can themselves be highly reduced or fragmented. These observations are evidence for lineage-specific solutions, not a license to group all photosynthetic protists together. Cross-lineage comparison should ask which feature arose before or after plastid acquisition and whether a similar phenotype reflects shared ancestry or convergence.

### 104.3.05. What the protist comparison establishes

Ciliates show RNA participation in genome restructuring; kinetoplastids show guide-directed mitochondrial coding-sequence reconstruction and widespread trans-splicing; apicomplexans show extensive stage-specific post-transcriptional control without a canonical RNAi template; and algae show how plastid history and nuclear lineage jointly shape RNA systems. None represents the others. Together they establish that a mature RNA cannot always be interpreted as a colinear readout of a nuclear gene and that gene regulation can be concentrated at processing, storage, translation, organellar maturation, or genome-development steps.

## 104.4. Evolutionary gains, losses, and convergent RNA mechanisms across non-animal eukaryotes

Comparative RNA biology begins by separating a molecular part from a regulatory job. Dicer-like enzymes share an RNase III ancestry, but duplication can allow one paralog to process miRNA hairpins, another to make antiviral siRNAs, and another to make heterochromatic siRNAs. Argonautes share a guide-binding architecture, but their outputs can include slicing, translational repression, chromatin recruitment, or developmental genome elimination. A homologous component can change jobs, and a similar job can recruit different components.

> **Box 104.2. Comparative Claims Need Organism, Factor, and Output**
>
> A useful comparative RNA statement has four parts. First, name the organism or lineage, because "plant," "fungus," and "protist" each hide major diversity. Second, name the molecular factor or RNA class, such as DCL1, AGO4, fungal Dicer, a spliced leader RNA, a guide RNA, or a pentatricopeptide repeat protein. Third, name the evidence: genetics, small RNA sequencing, methylome data, edited-transcript mapping, biochemical fractionation, imaging, or infection assays. Fourth, name the output: target slicing, DNA methylation, antiviral defense, transcript maturation, mitochondrial coding-sequence reconstruction, virulence, or stress tolerance. If one part is missing, narrow the claim. "This lineage encodes an Argonaute" is not the same as "this lineage uses RNAi for antiviral defense." "A trypanosome edits mitochondrial RNA" is not evidence that all protists use the same editing chemistry.

Pathway loss is equally informative. Canonical RNAi loss in *S. cerevisiae* may be tolerated because other genome-defense and RNA-quality-control systems remain, but that lineage cannot define the ancestral or fungal-wide state. Apparent RNAi absence in an apicomplexan may be supported by high-quality genomes and repeated failure to find core machinery, yet the organism still requires post-transcriptional regulation. Loss narrows the set of available mechanisms; it does not remove the regulatory problem.

Convergence appears when similar selective pressures repeatedly favor guide-dependent or post-transcriptional control. Repetitive DNA is repressed through plant RdDM, fungal RNAi-linked chromatin, and ciliate small-RNA-guided elimination, but these outcomes are not one conserved linear pathway. Life-cycle transitions favor stored mRNAs in apicomplexans and extensive RNA-binding-protein control in trypanosomatids. Organelle-to-nucleus gene transfer repeatedly creates a need for imported RNA-binding and processing factors, yet the factor families and transcript architectures differ.

An explicit ownership matrix in Table 104.3 maps the organisms, comparative topics, and deeper pathway owners. The matrix prevents two opposite errors: leaving non-animal systems as brief exceptions in animal-centered chapters, and duplicating full small-RNA, editing, decay, organellar, or infection chapters here.

**Table 104.2. Evidence standards for non-animal RNA mechanisms.** Different RNA claims require different evidence ladders, especially when RNA origin and function are difficult to assign.

| Mechanism | Minimum evidence | Stronger evidence | Common artifact | Useful methods |
| --- | --- | --- | --- | --- |
| **Plant miRNA target cleavage** | miRNA-target complementarity plus a cleavage product at the expected site | DCL1/AGO dependence, degradome or 5′ RACE support, target-site mutant or rescue phenotype | Treating indirect mRNA reduction or random decay fragments as slicing | Small RNA-seq, degradome/PARE, RLM-5′ RACE, reporter assays, plant genetics |
| **RdDM** | 24-nt siRNAs at a locus with associated DNA methylation | Pol IV/RDR2/DCL3/AGO4/Pol V/DRM dependence plus methylome and chromatin change | Inferring causality from siRNA abundance or mis-mapped repetitive reads | Bisulfite sequencing, small RNA-seq, methylation-sensitive assays, ChIP, pathway mutants |
| **Fungal RNAi** | Small RNAs and target repression dependent on fungal Dicer, Argonaute, or RdRP factors | Genetic rescue, target cleavage or repression, and linked antiviral, transposon, developmental, or infection phenotype | Assuming all fungi retain canonical RNAi; confusing degraded RNA with guides | Fungal mutants, complementation, small RNA-seq, target validation, infection or virus assays |
| **Fungal RNA decay** | Transcript half-life or decay-intermediate change after decay-factor perturbation | Time-resolved decay kinetics, rescue, ribosome-coupled evidence, and separation from transcriptional effects | Misreading stress-induced abundance shifts as decay without synthesis controls | RNA-seq time courses, transcription shutoff or metabolic labeling, decay mutants, ribosome profiling |
| **Ciliate programmed DNA elimination** | Development-specific small RNAs correlated with retained or eliminated germline sequences | Piwi-factor perturbation, small-RNA sequence dependence, chromatin change, and direct DNA-retention or elimination measurement | Mixing micronuclear and macronuclear sequence, comparing the wrong developmental stage, or generalizing one ciliate pathway | Developmental small RNA-seq, nuclear fractionation, long-read genome sequencing, chromatin assays, retention PCR, genetics |
| **Kinetoplastid editing** | RNA-DNA differences showing uridine insertion or deletion with candidate guide RNAs | Guide RNA mapping, editosome perturbation, editing intermediates, and stage-specific transcript analysis | Genome assembly error, PCR artifact, or DNA polymorphism mistaken for RNA editing | Organellar RNA-seq, guide RNA mapping, mitochondrial assemblies, biochemical fractionation, perturbation genetics |
| **Trans-splicing** | Spliced leader sequence joined to an mRNA from a separate transcript | Spliced-leader junction mapping, polycistronic precursor context, capped leader evidence, spliceosome-factor dependence | Template switching, annotation error, or incomplete 5′ UTR assignment | SL-PCR, capped RNA-seq, long-read RNA-seq, splice-junction mapping, factor perturbation |
| **Apicomplexan translational control** | Stable mRNA with discordant protein abundance or ribosome occupancy across stages | Dependence on a defined messenger RNP factor, release from storage, stage-specific translation, and phenotype | Inferring translation from transcript abundance or comparing mixed parasite stages | Ribosome profiling, proteomics, RNP purification, reporter assays, stage-resolved genetics |
| **Algal pathway comparison** | Candidate components and matched RNA products in a named algal lineage | Phylogeny, localization, genetics, substrate identification, and comparison across independent lineages | Treating “algae” as one clade or assigning land-plant function from a best hit | Comparative genomics, organelle-aware RNA-seq, small RNA-seq, proteomics, transformation where available |
| **Pathway absence** | Failure to detect a component using sensitive searches in a high-quality genome and relevant stages | Independent assemblies, profile or structure-aware search, expression sampling, and evidence for alternative machinery | Converting an assembly gap, divergent sequence, or excluded RNA class into evolutionary loss | Assembly-quality metrics, profile HMMs, structure prediction, stage-resolved RNA-seq, proteomics |
| **Cross-kingdom RNA transfer** | Foreign RNA detected with species-specific mapping in an interacting host-pathogen sample | Transfer route, recipient Argonaute or target engagement, target repression, phenotype, and rescue or mutant controls | Contamination, passive uptake, ambiguous read mapping, or function inferred only from detection | Dual RNA-seq, small RNA-seq, extracellular-vesicle purification, AGO/RIP or CLIP, reporters, infection assays |

**Table 104.3. Organism-by-topic ownership and coverage matrix.** The chapter gives every major non-animal lineage a concrete comparative role while handing detailed pathway chemistry and organism-specific disease biology to dedicated chapters.

| Lineage and representative systems | Comparative topic owned here | Concrete mechanism or example | Evidence basis emphasized here | Deeper primary owner | Boundary or sampling limit |
| --- | --- | --- | --- | --- | --- |
| **Flowering plants: *Arabidopsis*, rice, maize** | Diversification of small-RNA pathways across development, repeats, and crop genomes | DCL1/HYL1/SERRATE miRNA maturation; reproductive phasiRNAs; Pol IV/Pol V RdDM | Genetics, small RNA-seq, degradome mapping, methylomes, target-site rescue | Small-RNA pathway chapters [Chapter 80](chapter1075.md)-[Chapter 88](chapter1083.md); plant immunity [Chapter 113](chapter1107.md) | *Arabidopsis* does not represent crop polyploidy, repeat content, reproductive small RNAs, or field environments. |
| **Mosses and early-diverging land plants** | Retention and modification of land-plant RNA-regulatory features | Conserved small-RNA components with lineage-specific precursor and developmental use | Comparative genetics, small-RNA profiles, orthology and expression | Small-RNA pathway chapters [Chapter 80](chapter1075.md)-[Chapter 88](chapter1083.md) | Few genetically tractable species make absence and ancestral-state claims sensitive to sampling. |
| **Fungi: *Neurospora*, fission yeast, budding yeasts, pathogens, mucoralean fungi** | Retention, specialization, or loss of RNAi relative to pervasive decay, splicing, and translational control | Quelling; meiotic silencing by unpaired DNA; sex-induced and antiviral silencing; RNA-dependent amplification or Dicer-independent decay; RNA epimutation | Fungal genetics, factor-specific perturbation, complementation, half-life measurement, small-RNA and infection assays | RNA decay Chapters [32](chapter1031.md)-[38](chapter1036.md); host-pathogen synthesis [Chapter 111](chapter1106.md) | *S. cerevisiae* is an RNAi-loss lineage; fungal RNAi branches differ by stage and factor set; infection samples mix species and fungal cell states. |
| **Ciliates: *Tetrahymena*, *Paramecium*, spirotrichs** | RNA guidance of somatic-genome development | Scan-RNA/Piwi-linked sequence comparison, heterochromatin, DNA elimination, and lineage-specific unscrambling | Developmental genetics, small-RNA profiling, chromatin assays, direct DNA-retention tests | Small-RNA mechanism Chapters [84](chapter1079.md)-[85](chapter1080.md) | Pathway direction, RNA size, accessory factors, and rearrangement architecture differ among ciliates. |
| **Kinetoplastids: *Trypanosoma*, *Leishmania*** | Coupling of unusual mitochondrial genomes to guide-directed RNA reconstruction, terminal maturation, and stage regulation | Uridine insertion/deletion editing; editing-state-dependent A tails and post-editing A/U tails; cap4-bearing spliced-leader trans-splicing coupled to polyadenylation | Guide-RNA and tail-state mapping, editing-intermediate sequencing, complex perturbation, splice-site mapping, life-stage analysis | Editing Chapters [50](chapter1046.md)-[51](chapter1047.md); organellar genes [Chapter 17](chapter1016.md); trans-splicing [Chapter 27](chapter1026.md) | The order of all mitochondrial processing events is not fixed; mapped alternative trans-splicing sites often lack protein-level validation; parasite models do not represent all euglenozoans. |
| **Apicomplexans: *Plasmodium*, *Toxoplasma*** | Post-transcriptional life-cycle control in systems without a canonical RNAi template | DOZI/CITH-associated mRNA storage; eIF2-alpha stress control; m6A-linked CPSF4 site choice; CPSF3 inhibitor target engagement; apicoplast RNA expression | Stage-resolved transcriptomics, ribosome profiling, proteomics, RNP-factor genetics, complex purification, conditional depletion, inhibitor structures and resistance mapping | Translation regulation [Chapter 70](chapter1065.md); 3′-end processing [Chapter 30](chapter1029.md); organellar genes [Chapter 17](chapter1016.md) | Storage and global initiation control are distinct; m6A-linked site choice is locus-specific; target engagement does not guarantee parasite-selective pharmacology. |
| **Green algae: *Chlamydomonas* and relatives** | Comparison of photosynthetic-eukaryote small-RNA and organellar regulation with land plants | Dicer/Argonaute pathways; chloroplast RNA stabilization and translation | Transformation genetics, small-RNA profiling, organellar RNA and protein assays | Organellar genes [Chapter 17](chapter1016.md); small-RNA chapters [Chapter 80](chapter1075.md)-[Chapter 88](chapter1083.md) | Green algae cannot represent red algae, diatoms, dinoflagellates, or all plastid histories. |
| **Diatoms, dinoflagellates, and other photosynthetic protists** | Effects of secondary plastids and unusual nuclear or organellar organization on RNA biology | Complex organelle targeting; conserved 22-nucleotide spliced leader on surveyed dinoflagellate nuclear messenger RNAs; reduced or fragmented organellar genomes | Full-length cDNA/genome comparison, comparative genomics, long-read transcriptomics, organelle-aware mapping, proteomics | Organellar genes [Chapter 17](chapter1016.md); processing owners by mechanism | The tested dinoflagellate leader marked nuclear messenger RNAs but not assayed organelle-encoded or ribosomal RNAs; “algae” is not one clade. |
| **Cross-kingdom plant-fungal systems** | Evidence boundary between RNA exposure and functional interspecies regulation | Donor RNA production, recovery of recipient material, transfer, recipient target engagement, and sequence-dependent phenotype | Dual RNA-seq, recovered-hypha Northern blotting, host-debris controls, vesicle controls, recipient Argonaute assays, reporters, target-site mutants | Host-pathogen synthesis [Chapter 111](chapter1106.md); plant immunity [Chapter 113](chapter1107.md); extracellular RNA [Chapter 107](chapter1102.md) | Recovery culture can select the sampled population; detection establishes exposure, whereas target engagement and phenotype require separate evidence. |

The matrix also exposes uneven depth. Flowering-plant small RNAs have abundant genetics, biochemistry, and genome-wide profiling. Fungal decay is mechanistically grounded in a few models but ecologically broad claims often rely on transcriptomes. Kinetoplastid editing has strong molecular evidence in tractable parasites, while many free-living protists remain represented only by assemblies and expression surveys. Ciliate genome rearrangement is deeply studied in a small number of laboratory species. Algal diversity vastly exceeds the number of species with genetic tools. “Comparative coverage” therefore means explicit evidence grading, not equal certainty for every cell in a table.

Evolutionary claims require a tree-aware design. Sampling ten closely related flowering plants is not equivalent to sampling ten independent eukaryotic branches. Orthology records descent through speciation; it does not require a one-to-one gene correspondence or identical function. Lineage-specific duplication and loss can create one-to-many or many-to-many co-orthology, and reciprocal best hits can conceal differential loss, domain rearrangement, or horizontal transfer. A gain or loss should therefore be reconstructed by reconciling gene and species histories while considering domain architecture and genome completeness. Correlations between pathway presence and phenotype should use a phylogenetic regression, mixed model, independent contrasts, or another design appropriate to the data rather than counting species as independent replicates. Adding more genomes from the same over-sequenced clade does not remove this pseudoreplication, and multiple paralogs within one genome introduce another nested level of non-independence. A functional shift still requires substrate, localization, expression, or phenotype data rather than a changed gene name. These are central cautions of phylogenetic comparative genomics and orthology-based function transfer (Gabaldón and Koonin, 2013; Dewar et al., 2025). When those data are absent, the appropriate result is a candidate distribution, not a completed mechanism.

## 104.5. Experimental systems, phylogenetic sampling, annotation bias, and comparative inference

Comparative conclusions are limited by how organisms become experimentally visible. *Arabidopsis*, *S. cerevisiae*, fission yeast, *Neurospora*, *Chlamydomonas*, *Tetrahymena*, *Paramecium*, *Trypanosoma*, *Leishmania*, *Plasmodium*, and *Toxoplasma* are valuable because they combine culture methods with genetics, imaging, biochemical manipulation, or life-cycle access. They are not statistically representative samples of plants, fungi, or protists. Model selection favors organisms that grow in the laboratory, tolerate transformation, have compact or tractable genomes, and fit existing funding or disease priorities.

### 104.5.01. Organism-by-topic sampling design

A defensible comparative study first states its sampling unit. If the question is whether canonical RNAi was ancestral to fungi, species should span independent fungal branches and should not be treated as independent when they share a recent loss. If the question is how plant RdDM changes in polyploid crops, the design needs repeat-rich crop genomes and matched tissues rather than only additional *Arabidopsis* accessions. If the question concerns protist trans-splicing, kinetoplastids, dinoflagellates, and other trans-splicing lineages must be analyzed as separate evolutionary cases.

The practical minimum is an organism-by-topic matrix with three kinds of cells: strong mechanistic ownership, comparative evidence, and unresolved or unsampled status. An empty cell is informative only if the assay could have detected the pathway. For example, no small RNAs in a poly(A)-selected library says little because the library excluded most small RNAs. No editing in a single life stage says little if editing is developmentally regulated. No gene in a fragmented assembly may be a sequencing result rather than an evolutionary loss.

### 104.5.02. Genome and transcript annotation bias

Genome assembly determines which RNA mechanisms can be seen. Repetitive small-RNA loci, ciliate germline-limited sequences, kinetoplast DNA, organellar minicircles, and fragmented plastid or mitochondrial genes can be collapsed, excluded, or misassembled. A conventional gene predictor may miss short noncoding RNAs, trans-spliced leaders, heavily edited coding regions, noncanonical introns, and lineage-specific proteins. Annotation pipelines trained on animal or fungal genes can transfer familiar exon structures into organisms that use different transcript processing. Genome annotation remains error-prone even in well-supported projects, so a negative annotation should be treated as a method-dependent result rather than direct proof of biological absence (Salzberg, 2019).

Transcript annotations introduce a second filter. Poly(A) selection enriches many mRNAs but can miss nonpolyadenylated RNAs or confound organellar polyadenylation with stability. Short reads can establish local splice junctions but not always full isoforms or polycistronic processing paths. Highly edited reads may fail to align to the genome. Small-RNA libraries are sensitive to RNA-end chemistry, size selection, adapter ligation, and modifications. A pathway-specific negative result must therefore name the library and its detection limits.

Long-read RNA sequencing can join distant processing events, identify trans-spliced leaders, and resolve full isoforms, but it does not automatically solve low abundance, base-calling error, or RNA-modification effects. End-mapping methods define cleavage, capping, and polyadenylation sites. Ribosome profiling tests translation but requires organism-specific control of nuclease digestion, footprint assignment, and organellar contamination. Comparative method choice is part of the biological claim, not a technical footnote.

### 104.5.03. From sequence similarity to mechanism

Homology search is an entry point. A candidate Dicer should be evaluated for RNase III and RNA-binding architecture, phylogenetic placement, expression, cellular localization, interacting partners, small-RNA products, and mutant phenotype. A candidate Argonaute should be connected to a guide population and target. A candidate organellar RNA-binding protein should be localized to the organelle and linked to a defined processing, stability, editing, or translation event. Orthology improves the prior expectation of shared function but does not guarantee identical substrate or regulation after duplication and lineage-specific change (Gabaldón and Koonin, 2013). Without those links, “Dicer-like,” “Argonaute-like,” or “PPR-like” describes sequence evidence, not completed functional annotation.

The reverse problem is excessive reliance on sequence conservation. Rapidly evolving RNA-binding proteins can escape ordinary similarity searches while retaining a related structural or functional role. Nonhomologous replacement can preserve the pathway output after the ancestral factor is lost. Proteomics of purified complexes, structure prediction, genetic interaction, and biochemical reconstitution can find these cases, but each evidence class has boundaries. Co-purification suggests association, predicted structure suggests compatibility, and rescue suggests functional overlap; none alone proves identical native mechanism.

### 104.5.04. Life stage, cell state, and compartment

Non-animal eukaryotes often change RNA regulation across life cycles. Trypanosomatids remodel mitochondrial activity between vector and vertebrate stages. Apicomplexans store mRNAs before transmission and translate them later. Fungi switch morphology or developmental programs during host entry and reproduction. Plants produce specialized small RNAs in reproductive tissues. Ciliates activate genome-rearrangement small RNAs during sexual development. Sampling one stage can therefore convert a regulated pathway into an apparent lineage absence.

Cellular compartment matters as much as life stage. A guide RNA may act in a mitochondrion, a scan RNA in a developing macronucleus, a plant heterochromatic siRNA in the nucleus, and an antiviral siRNA in the cytoplasm. Whole-cell abundance can obscure a low-abundance but locally concentrated effector. Fractionation, imaging, proximity labeling, organelle purification, and compartment-specific perturbation help establish the site of action, but contamination between compartments must be quantified.

### 104.5.05. Evidence ladder and artifact control

The weakest evidence for a pathway is a remotely similar sequence or an RNA abundance correlation. Stronger evidence adds correct domain architecture, phylogenetic placement, coexpression with plausible partners, localization, a matched RNA substrate or guide, a perturbation phenotype, target engagement, and rescue. Biochemical reconstitution can establish reaction capability, while genetics establishes necessity in the tested context. Neither automatically proves universality across environments or species.

Host-pathogen and extracellular-RNA claims demand additional controls because reads can be assigned to the wrong genome and RNA can move during sample disruption. A strong cross-kingdom claim connects donor production, molecular packaging or protection, transfer into the recipient, intracellular availability, target engagement, pathway dependence, and a sequence-specific phenotype. The host-pathogen comparison is developed in [Chapter 111](chapter1106.md), while plant immune consequence belongs to [Chapter 113](chapter1107.md).

Do not overgeneralize: a richly characterized model does not compensate for narrow phylogenetic sampling, and broad sampling does not compensate for weak mechanism. The best comparative inference combines both: mechanistic depth in representative systems and explicit, tree-aware tests of distribution and variation.

## 104.6. Agricultural, ecological, medical, and biotechnological applications

Agricultural RNA technologies exploit endogenous silencing and uptake pathways. Host-induced gene silencing expresses a double-stranded or hairpin RNA in a plant so that a pest or pathogen receives a sequence-specific silencing trigger. Spray-induced gene silencing applies RNA externally, sometimes with a carrier or protective formulation. Both strategies require more than a potent guide sequence: the RNA must persist on the plant, enter the target organism, escape extracellular and intracellular degradation, reach the appropriate compartment, engage a compatible silencing pathway, and reduce the target enough to alter disease or survival.

Molecular efficacy and field performance are separate gates. Molecular efficacy can be tested with uptake assays, target-site reporters, cleavage or knockdown measurements, pathway mutants, and rescue. Field performance adds rain, ultraviolet exposure, temperature, plant surface chemistry, formulation cost, non-target exposure, variable pest populations, and resistance evolution. Resistance can arise through target-site mutation, reduced uptake, faster degradation, altered RNAi components, or ecological replacement by another organism. RNA-based crop protection therefore complements rather than automatically replaces breeding, chemical control, biological control, and integrated pest management.

Cross-kingdom RNA regulation is relevant at plant-fungal interfaces. Selected systems support fungal small RNAs acting on plant targets and plant RNAs entering fungal pathogens. These results motivate crop-protection strategies and show that RNA can be part of interspecies conflict. They do not show that every infection is dominated by transferred RNA. Species assignment, physical transfer, recipient-side target engagement, sequence dependence, and phenotype must be established in the same system.

A cotton-*Verticillium dahliae* protocol illustrates the distinction between detecting exposure and establishing regulation. Fungal material can be recovered from surface-sterilized infected stems and assayed by small-RNA Northern blotting, but plant debris must be removed, matched wild-type and transgenic controls are required, and the low abundance of transferred small RNA makes detection sensitive to recovery and loading conditions. Recovery culture also selects surviving fungal material and therefore does not preserve the original in planta population unchanged. The method supports an organism-specific transfer assay; it does not by itself establish Argonaute loading, target cleavage, or a general mechanism across plant-fungal interactions (Zhang et al., 2022).

> **Box 104.3. Cross-Kingdom RNA Interference Evidence Checklist**
>
> Read a cross-kingdom RNAi claim as a chain of conditional steps. The donor organism must produce the RNA under the interaction condition. The RNA must be assigned unambiguously to the donor genome, with controls for contamination, repetitive sequence, and library index mixing. The RNA must reach the recipient cell or extracellular compartment in a form that can survive nucleases and enter the relevant silencing pathway. Target engagement then needs recipient-side evidence, such as Argonaute association, target cleavage, reporter repression, or a target-site mutant that resists regulation. Finally, the phenotype should depend on the RNA sequence or pathway and be rescued or lost when donor, recipient, or target-site components are altered. Detection alone supports exposure or association; it does not prove transfer, uptake, silencing, or biological importance.

![Figure 104.4. Cross-kingdom RNA exchange at a plant-fungal interface](../assets/figures/chapter1099_figure4.png)

**Figure 104.4. Cross-kingdom RNA exchange at a plant-fungal interface.** Cross-kingdom RNA interference requires evidence for transfer and function; a small RNA detected in recovered recipient material establishes exposure only after contamination controls.

![Figure 104.5. RNA innovations across non-animal eukaryotic lineages](../assets/figures/chapter1099_figure5.png)

**Figure 104.5. RNA innovations across non-animal eukaryotic lineages.** A shared regulatory problem can be solved by different RNA machinery, while a homologous component can change jobs; comparative claims must name the lineage and mechanism.

Medical relevance extends beyond cross-kingdom RNAi. Fungal RNA regulation can influence virulence, morphologic switching, stress survival, and antifungal adaptation. Kinetoplastid editosomes and life-stage-specific RNA-binding proteins offer parasite-selective target opportunities because their machinery differs from host systems. Apicomplexan mRNA-storage and organellar pathways likewise expose parasite-specific dependencies. A candidate is not validated by uniqueness alone: essentiality in the disease stage, chemical accessibility, selectivity, resistance routes, and host toxicity must be tested.

Biotechnology also learns from non-animal RNA architectures. Modular plant pentatricopeptide repeat proteins motivate programmable RNA recognition. Guide-directed kinetoplastid editing demonstrates distributed sequence specification, even though direct transfer of that machinery to another host is a major engineering challenge. Spliced-leader and other trans-splicing systems inspire RNA-repair concepts. Fungal hosts support fermentation, RNA-processing studies, and genetic screens. Ciliate rearrangement systems provide models for RNA-guided genome information flow, but their developmental machinery is not a ready-made genome-editing platform.

Ecological assessment must consider exposure and community context. Applied RNA can degrade rapidly, persist in protected particles, or be taken up unevenly by target and non-target organisms. Sequence complementarity lowers some risks but does not replace empirical testing of uptake, expression, and phenotype. Environmental monitoring should distinguish the intact active RNA from short degradation fragments. The regulatory and engineering details of RNA products belong to the therapeutics, delivery, and biotechnology chapters; this chapter owns the lineage biology that determines whether an application can work.

## Experimental Foundations and Evidence

No single method demonstrates a comparative RNA mechanism. Plant small-RNA studies are strongest when genetics, small-RNA sequencing, target-end mapping, methylomes, and chromatin assays agree. Fungal studies should separate transcription, RNA stability, splicing, translation, and changing cell-state composition. Kinetoplastid editing requires genome-aware mapping of guide RNAs and intermediates, perturbation of editosome components, and reaction-level evidence. Ciliate rearrangement requires developmental staging and direct measurement of DNA retention or elimination. Apicomplexan translational-control claims require protein or ribosome-occupancy data in addition to mRNA abundance.

Comparative genomics extends these mechanistic anchors, but it should carry explicit confidence labels. A well-supported ortholog with conserved domains is stronger than a best database hit; a localized, interacting, genetically required ortholog is stronger still. Apparent absence should be graded by assembly completeness, search sensitivity, life-stage sampling, and the expected assay signal. This evidence ladder prevents both model-organism overreach and indiscriminate celebration of every unusual transcript.

## Biological Contexts Across Systems

The recurring biological contexts are genome defense, development, environmental transition, and organelle integration. Plants use small RNAs and chromatin to manage repeats while patterning tissues. Fungi combine pathway retention or loss with rapid regulation of stress, morphology, and host association. Ciliates use RNA during programmed construction of a somatic genome. Kinetoplastids and apicomplexans concentrate regulation in processing, RNA-binding proteins, translation, and life-stage changes. Photosynthetic lineages must coordinate nuclear and organellar expression across distinct endosymbiotic histories.

These contexts are comparable without being equivalent. Transposon repression by methylation, heterochromatin, or DNA elimination produces superficially similar outcomes but uses different molecular operations. Life-stage regulation by stored mRNA and by edited mitochondrial transcripts solves different bottlenecks. Good comparative prose names both the shared biological problem and the lineage-specific mechanism.

## Technology, Computational, Clinical, or Engineering Links

Comparative annotation increasingly combines long reads, small-RNA end chemistry, pangenomes, phylogenetic profiles, structure prediction, and proteomics. Computational pipelines should not assume colinear transcripts, standard genetic codes, canonical splice signals, or animal-like small-RNA lengths. Organism-aware reference construction and explicit negative controls are prerequisites for discovering rather than erasing unusual biology.

Engineering applications should begin with pathway compatibility. An RNA designed for a fungus that lacks uptake or canonical RNAi will fail even if the target sequence is perfect. A parasite-specific editing enzyme may be attractive only if the disease stage depends on it. A plant epigenome tool must contend with locus chromatin and heritability. Comparative RNA biology therefore supplies the biological constraints that connect a molecular design to a realistic organism.

## Recent Consensus

Current evidence supports five broad conclusions. First, plant small-RNA pathways are diversified and integrated with development, defense, genome methylation, and organellar function. Second, fungal canonical RNAi is evolutionarily patchy and itself contains distinct vegetative, sexual, antiviral, and endogenous regulatory branches, while decay, splicing, and translational control are pervasive regulatory layers. Third, protist lineages contain multiple distinct RNA architectures, including ciliate genome-remodeling guides, kinetoplastid editing embedded in terminal maturation and quality control, apicomplexan messenger-RNA storage and initiation control, and lineage-specific organellar systems. Fourth, apparent pathway absence, novelty, or trait correlation is inseparable from genome quality, life-stage sampling, assay design, orthology resolution, and phylogenetic non-independence. Fifth, cross-kingdom and applied RNA mechanisms are credible in selected systems but require complete evidence chains from RNA source to target-dependent phenotype.

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

Open questions:

- How many apparent RNA-pathway losses will remain after long-read genomes, stage-resolved transcriptomes, and structure-aware homology searches become available across undersampled eukaryotes?
- Which ciliate, algal, fungal, and free-living protist RNA systems are genuinely lineage restricted, and which are remnants of broader ancestral architectures?
- How often do RNA-guided epigenetic states contribute to adaptation outside the few experimentally tractable plant and fungal models?
- Under what ecological conditions does cross-kingdom RNA transfer produce a fitness effect large enough to shape natural interactions?
- How can phylogenetic breadth and mechanistic depth be balanced when most non-animal eukaryotes lack genetic tools?

Controversies:

- The frequency and general importance of functional extracellular RNA transfer across species remain disputed because contamination, differential stability, and recipient uptake are difficult to separate.
- Some proposed novel noncoding RNAs or editing events may reflect fragmented assemblies, mapping error, or mixed life stages; others will prove to be real mechanisms missed by conventional annotation.

Common misconceptions:

- “*Arabidopsis* represents plants, budding yeast represents fungi, and trypanosomes represent protists.” Each model resolves particular mechanisms but samples only one lineage and ecological strategy.
- “Finding a Dicer or Argonaute homolog proves canonical RNA interference.” Homology identifies a candidate component; guide identity, target engagement, pathway partners, and perturbation establish function.
- “Failure to annotate a pathway component proves evolutionary loss.” Assembly gaps, divergence, stage restriction, and nonhomologous replacement must be excluded.
- “Protist RNA editing is a small correction to otherwise conventional gene expression.” Kinetoplastid uridine insertion and deletion can construct much of a mature coding sequence, while other protist lineages use different or little editing.
- “All algae share a plant-like RNA system.” Algae are polyphyletic and have different nuclear and plastid histories.
- “A transferred RNA detected during infection proves cross-kingdom RNA interference.” Transfer, intracellular availability, target engagement, pathway dependence, and sequence-specific phenotype are separate evidentiary steps.