The challenge that motivated the work is deceptively simple to state. Effector proteins are extraordinarily diverse. They often share little recognizable sequence similarity with one another, even when they perform related functions, because bacteria have evolved them repeatedly from different starting points. Traditional bioinformatic tools that scan for conserved sequence motifs therefore miss many genuine effectors while occasionally flagging innocent proteins. In recent years, pretrained protein language models, which learn statistical representations of amino acid sequences from vast databases, have substantially improved effector prediction by capturing contextual patterns that simpler methods overlook. Yet even these powerful models share a fundamental blind spot: they see proteins as one-dimensional strings of letters, while the biological activity of an effector depends on how its chain folds in three-dimensional space.

Structure, in principle, holds the key. The virulence functions of effectors, such as binding host proteins, mimicking host enzymes, or disrupting cellular compartments, are encoded in conformational motifs: recurring arrangements of residues in space that may appear at widely separated positions along the linear sequence. The problem has always been data. Experimentally resolved structures, obtained through crystallography or cryo-electron microscopy, exist for only a small fraction of bacterial proteins, and effectors are particularly underrepresented because they are difficult to purify and crystallize. Conventional structure-based prediction methods, however elegant, have been starved of the structural information they need to work at scale.

GeoEPred, developed by Shouzhen Song, Hua Shi, Hongfeng Wu, Dachen Liu, Yihang Lin, Nor Ashidi Mat Isa, Quan Zou, Leyi Wei and colleagues, threads this needle with a clever compromise. Instead of waiting for experimental structures, the framework uses ESMFold, a deep learning system that predicts three-dimensional protein structures directly from sequence, to generate structural models for every protein it examines. These predicted structures are then processed by geometric vector perceptrons, neural network components that operate on the local geometry of the protein: the orientations between neighboring residues, the distances separating them, and the topology of the spatial neighborhood each residue occupies. In effect, the model learns what local conformational neighborhoods look like in proteins that function as effectors, capturing spatial signatures that no sequence-only model can perceive.

But structure alone is not enough either, and this is where the multimodal design of GeoEPred becomes distinctive. The framework runs two parallel streams of representation. The first draws on a pretrained protein language model to produce sequence-contextual embeddings, rich numerical descriptions of each amino acid in its linear context. The second builds the geometric representations from the predicted structure. A dedicated feature projection network then refines the fine-grained sequence signals that are most relevant to effector function, filtering the language model’s general-purpose knowledge toward the specific patterns that matter for secretion and virulence.

The hardest part of any multimodal system is making the two modalities talk to each other. A sequence embedding and a structural embedding describe the same protein, but they live in different mathematical spaces with different statistics, and naive concatenation often fails to exploit their complementary strengths. The GeoEPred team addressed this with a cross-modal alignment and feature-tokenized self-attention module. The alignment component uses contrastive learning, a technique that pulls matching representations from the two modalities closer together in the learned space while pushing mismatched ones apart, thereby enforcing consistency between sequence semantics and structural geometry. The tokenized self-attention component then models associations at the level of individual residues, allowing the network to link linear functional motifs, short stretches of sequence with known or learned significance, to the spatial conformational patterns those stretches adopt in the folded protein.

The result is a model that can recognize an effector not merely because its sequence resembles known effectors, but because its residues arrange themselves in space the way effector residues tend to. That distinction matters most at the frontier of discovery, where new effectors share no detectable ancestry with characterized ones. The authors evaluated GeoEPred on multiple benchmark datasets covering the three major secretion systems of Gram-negative bacteria, known as type III, type IV, and type VI secretion systems, each of which delivers effectors in its own characteristic way. Across all three prediction tasks, abbreviated T3SE, T4SE, and T6SE in the field’s shorthand, GeoEPred outperformed existing leading models, demonstrating that the structural signal adds genuine predictive power rather than redundant information.

Perhaps more important than the headline benchmarks is the model’s behavior in remote homolog recognition scenarios, where the test proteins are evolutionarily distant from anything seen during training. This is the regime that determines whether a predictor is useful for real discovery or merely good at memorizing known families. GeoEPred maintained stable performance under these demanding conditions, suggesting that the geometric representations it learns capture functional constraints that persist across deep evolutionary time even as sequences drift beyond recognition. The modular architecture also proved extensible, and the authors report strong generalization ability with substantial application potential for genome-scale effector discovery, the scenario in which a newly sequenced bacterial genome can be scanned end to end for candidate virulence factors.

The practical implications reach well beyond the benchmark leaderboards. Rapid identification of effectors accelerates the elucidation of bacterial pathogenic mechanisms, giving microbiologists a prioritized shortlist of proteins to characterize experimentally instead of a genome full of untested hypotheses. It also informs the development of precise anti-infective strategies: effectors and the machinery that delivers them are attractive drug targets, and knowing the full effector repertoire of a pathogen reveals which host pathways it depends on. In agriculture, where Gram-negative pathogens devastate crops, the same logic applies to engineering resistance. As predicted structures from tools like ESMFold become routine inputs to biological machine learning, frameworks like GeoEPred illustrate a broader shift in computational biology, from reading genomes as text to understanding proteins as physical objects whose shapes, not just their sequences, carry the instructions for what they do.

None of this replaces experimental validation, and the authors are careful to frame GeoEPred as a discovery engine whose predictions guide, rather than conclude, the scientific process. Predicted structures, however accurate on average, carry uncertainties that propagate into any downstream model, and effector biology remains full of surprises that no current dataset fully captures. But the trajectory is clear. By fusing the semantic richness of protein language models with the physical realism of geometric deep learning, and by solving the fusion problem with principled contrastive alignment rather than brute-force concatenation, GeoEPred offers a template for how the next generation of protein function predictors will be built. For the hidden weapons of Gram-negative bacteria, many of which have evaded detection precisely because they look like nothing we have sequenced before, the era of shape-aware prediction may finally be opening the door.

Subject of Research: Multimodal geometric deep learning for predicting Gram-negative bacterial secreted effector proteins

Article Title: GeoEPred: A multimodal structure-aware geometric deep learning framework for Gram-negative bacterial secreted effector prediction with sequence semantics

Article References: GeoEPred: A multimodal structure-aware geometric deep learning framework for Gram-negative bacterial secreted effector prediction with sequence semantics. (n.d.). https://doi.org/10.1371/journal.pcbi.1014344

Image Credits: AI Generated

DOI: 10.1371/journal.pcbi.1014344

Keywords: GeoEPred, Gram-negative bacteria, effector proteins, deep learning, protein language models, ESMFold, protein structure prediction, type III secretion system, type IV secretion system, type VI secretion system, virulence, computational biology

The pathogens in question are a select group of bacteria that have made the plant vascular system their exclusive home. Xylella fastidiosa colonizes the xylem, the water-transport network of the plant, while Spiroplasmas, Phytoplasmas and Liberibacters confine themselves to the phloem, the sugar-rich tissue that distributes photosynthetic products. These bacteria cannot move between plants on their own. They depend on piercing-sucking insects, mostly hemipterans such as leafhoppers, planthoppers, spittlebugs and psyllids, whose needle-like stylets penetrate directly into the vascular tissues. In doing so, the insects ingest sap along with any microbes it contains and later deposit the bacteria into new hosts. Intriguingly, within the insect suborder Sternorrhyncha, only psyllids are known to transmit bacterial pathogens, while aphids and whiteflies, their close relatives, are primarily virus carriers.

The economic stakes are enormous. Citrus huanglongbing, or HLB, caused by the psyllid-transmitted Liberibacter species, has cut citrus yields in Florida by 74 percent since its detection there in 2005. In Italy, Xylella fastidiosa triggers olive quick decline syndrome, a disease that has killed millions of olive trees in the southern region of Puglia. Genetic analysis traced that outbreak to a Central American introduction of the pathogen around 2008, and phylogenetic work suggests the bacterium actually arrived in Europe as early as 1993. Xylella infects more than 700 plant species, including grape, in which it causes Pierce’s disease, and almond and citrus, where it produces leaf scorch and variegated chlorosis. The review also highlights the recent recognition of Xylella taiwanensis as a distinct species rather than a subspecies of X. fastidiosa.

Each pathogen has evolved a distinctive lifestyle suited to its vascular niche. Xylella, a Gram-negative bacterium related to Xanthomonas albilineans, uses Type IV pili to twitch its way through the nutrient-poor xylem, secreting cell wall-degrading enzymes such as polygalacturonase and endoglucanases to dissolve the pit membranes that separate adjacent vessels. It then forms sticky biofilms, anchored by extracellular polysaccharides and adhesins, that physically obstruct water flow. The resulting drought stress, stomatal closure and reduced photosynthesis eventually kill the plant. Spiroplasmas, by contrast, are helical, wall-less bacteria of the class Mollicutes that swim by propagating kinks along their bodies. Spiroplasma citri, discovered in 1970 and the first vector-borne bacterium ever cultured, causes citrus stubborn disease, while S. kunkelii produces maize stunt and S. phoeniceum infects periwinkle.

Phytoplasmas are stranger still. First observed in 1967 as mycoplasma-like organisms in infected phloem, they remain unculturable in the laboratory, forcing researchers to rely on DNA extracted from infected plants and on 16S rRNA gene sequencing for classification. Lacking both cell walls and flagella, they cannot produce the classic molecular patterns that plant immune receptors normally detect, and they appear to elicit immune responses through internal patterns instead. Their most dramatic effects are developmental: infected plants produce witches’ brooms, dense clusters of stunted twigs, or phyllody, in which flowers transform into leafy structures. These deformities prevent normal reproduction and turn the plant into a breeding ground for both the pathogen and its vectors. Phytoplasmas compensate for their lack of motility genes by using their immunodominant membrane protein to bind plant actin, apparently hitching rides through the host’s own cytoskeletal transport system.

The Liberibacters tell an evolutionary story of their own. Six species share a common ancestor, and genome analysis suggests the non-pathogenic Liberibacter crescens gave rise to the pathogenic Candidatus Liberibacter lineages. The pathogenic species are fastidious and unculturable, likely because their highly reduced genomes lack genes for thiamine and essential amino acids that L. crescens retains. Candidatus Liberibacter asiaticus is the primary HLB agent, transmitted by the Asian citrus psyllid Diaphorina citri, while Ca. L. solanacearum exists as at least six haplotypes: A and B, spread by the potato psyllid Bactericera cockerelli, cause zebra chip disease of potato, named for the dark streaks that appear in tubers after frying, and haplotypes C, D and E, carried by the carrot psyllid B. trigonica, attack umbelliferous crops.

Central to the review is the molecular arms race between these bacteria and plant immunity. Plants deploy pattern recognition receptors that detect bacterial flagellin and other pathogen-associated molecular patterns, triggering a first line of defense. Bacteria strike back with effector proteins. Phytoplasma effectors are among the best characterized: SAP05 hijacks the host 26S proteasome to degrade SPL and GATA transcription factors, promoting lateral shoots and sterile flowers; SAP11 and related proteins degrade TCP transcription factors to stimulate bud growth; SAP54 converts flowers into leaf-like structures; and SRP1 from rice orange leaf phytoplasma impairs chlorophyll precursor biosynthesis, producing yellow leaves that attract leafhoppers. Liberibacter effectors are equally sophisticated. CLas secretes SDE15 to target the programmed cell death regulator CsACD2, SDE1 to suppress papain-like cysteine proteases, and SDE3 and SDE4405 to manipulate autophagy, while the effector CLas4425 undermines salicylic acid signaling. For CLso, the effector Lso-HPE1 disrupts the tomato ubiquitin-proteasome system, and CKC_05701 suppresses programmed cell death and reactive oxygen production.

The relationship between pathogen and vector is just as intricate. Xylella is unique among these bacteria in forming biofilms in the foregut of its sharpshooter and spittlebug vectors, a semi-persistent, non-circulative association. It degrades chitin and uses adhesins such as XadA1 and XadA2, which bind strongly to chitin and cellulose, to anchor itself in the insect’s feeding apparatus. The phloem-limited bacteria, by contrast, establish persistent, circulative relationships: they cross the insect gut epithelium, survive in the hemolymph and eventually reach the salivary glands for reinoculation. Spiroplasma citri uses adhesion proteins including P89 and ScARPs for receptor-mediated endocytosis, and its phosphoglycerate kinase interacts with insect actin to promote colonization. Phytoplasmas enter insect cells through clathrin-mediated endocytosis, and their antigenic membrane proteins form complexes with insect microfilaments that determine vector specificity.

Liberibacters manipulate their psyllid hosts in remarkable ways. CLas localizes along the actin cytoskeleton of D. citri gut cells, and roughly 95 percent of its genes are active during gut colonization. In adult psyllids, CLas triggers apoptosis in the gut, which may reduce their transmission capacity, whereas nymphs show no such response and transmit more efficiently. CLso takes a preemptive approach: it induces expression of the anti-apoptotic gene IAPP5.2 early in infection, suppressing gut cell death to facilitate its own acquisition. The two zebra chip haplotypes provoke distinct immune responses in the potato psyllid, with CLsoB repressing apoptosis and achieving higher titers and more efficient transmission. Even insect symbionts matter: Wolbachia correlates positively with CLas titer in D. citri, and psyllids lacking Wolbachia acquire and transmit CLso less efficiently, suggesting microbiome-based control strategies.

The three-way interplay culminates in pathogen-driven manipulation of plant chemistry to recruit vectors. Phytoplasma effectors dampen jasmonic acid signaling and increase leafhopper fecundity, while SAP54 makes infected plants more attractive to leafhopper females. CLas induces citrus to release specific volatiles that deceptively attract D. citri and alters metabolite emissions to shift vector preference. Current management leans on insecticides and antibiotics, including oxytetracycline and streptomycin, approved by the U.S. EPA for Florida citrus, with trunk injection proving more efficient than foliar sprays. Promising alternatives include antimicrobial peptides such as nisin, which disrupts the Xylella lipid bilayer, and MaSAMP, a heat-stable peptide from HLB-resistant Microcitrus that both kills CLas and primes plant immunity. The authors point to CRISPR/Cas9 gene editing, plant immunomodulators such as salicylic acid analogues and GABA supplementation, and even light-engineering technologies as future directions. But with many of these pathogens still unculturable and genetically intractable, the review’s central message stands: cracking the bacterium-insect-plant triangle is the essential next step toward protecting global food security.

Subject of Research: Interactions among insect vectors, bacterial plant pathogens and host plants

Article Title: Bacterial vector-borne plant diseases: global issues caused by three-way interactions

Article References: Huang, Y., Hao, J., & Tang, X. (2025). Bacterial vector-borne plant diseases: global issues caused by three-way interactions. Crop Health, 3(1), Article 10. https://doi.org/10.1007/s44297-025-00049-0

Image Credits: AI Generated

DOI: 10.1007/s44297-025-00049-0

Keywords: Xylella fastidiosa, huanglongbing, Liberibacter, phytoplasma, Spiroplasma, insect vectors, psyllids, leafhoppers, plant immunity, effector proteins, phloem, xylem

The two most intensively studied feeding structures are giant cells and syncytia, and their developmental strategies could hardly be more different. Root-knot nematodes induce giant cells through rounds of mitosis that lack cytokinesis, combined with repeated endoreduplication, producing enlarged, multinucleate cells stuffed with dense cytoplasm. These cells develop wall ingrowths resembling xylem transfer cells, dramatically increasing the plasma membrane surface area available for solute exchange. Cyst nematodes, by contrast, dissolve the cell walls of neighboring root cells and fuse their protoplasts into a syncytium, a single continuous multinucleate compartment that can eventually encompass more than two hundred cells. Remarkably, three-dimensional reconstructions have shown that individual giant cells can expand roughly sixty-fold in volume between three and forty days after infection, a transformation driven by cell wall loosening and continuous cytoskeletal rearrangement.

None of this happens by accident. The engine of the entire process is the nematode’s stylet, a hollow, protrusible spear connected to esophageal gland cells that secrete a cocktail of effector proteins directly into host tissue. These effectors include cell wall-degrading enzymes such as cellulases, hemicellulases, and pectinases that ease penetration and migration, but the more intriguing molecules are those that manipulate host physiology from within. Once inside plant cells, effectors travel to the nucleus, the cytoplasm, or the apoplast, where they bind host proteins and bend gene expression to the parasite’s advantage. In compatible interactions, the host’s normal developmental pathways are subverted: the cell cycle is reactivated, cell wall architecture is remodeled, and metabolism is redirected toward feeding site construction.

Plant hormones sit at the center of this manipulation. Local auxin accumulation is a master switch for feeding site formation, and nematodes have evolved multiple ways to boost it. The cyst nematode effector 19C07 interacts with the auxin influx transporter LAX3 to enhance auxin flow into selected cells and upregulate cell wall-degrading enzymes, while the effector 10A07 binds the auxin response repressor IAA16, destabilizing its suppressive grip on auxin response factors and even recruiting a host kinase, IPK, to shepherd it into the nucleus. Cytokinin plays an equally critical role in driving cell cycle reactivation at the G1/S transition. Strikingly, Heterodera schachtii carries its own isopentenyltransferase gene, effectively synthesizing cytokinin from the parasite’s side of the interface; silencing that gene suppresses feeding site expansion, and Arabidopsis mutants lacking the cytokinin receptors AHK3 and AHK4 show markedly reduced susceptibility. Ethylene plays a double game, sometimes attracting nematodes to roots and promoting syncytial cell wall expansion, yet also capable of reducing infection through interactions with the receptor ETR1, a contradiction the review flags as an open question.

At the same time, nematodes must dismantle the plant’s immune system, which operates in two layers. Pattern-triggered immunity begins when cell surface receptors detect pathogen-associated molecular patterns, triggering reactive oxygen species production, MAP kinase cascades, and cell wall reinforcement. A deeper layer, effector-triggered immunity, mediated by intracellular resistance proteins, produces localized cell death that starves the parasite. Nematode effectors attack both layers with surgical precision. In rice, the Meloidogyne graminicola effector MgMO237 suppresses pattern-triggered immunity by binding multiple defense-related host proteins, while MgMO289 manipulates the host’s reactive oxygen system through the copper metallochaperone OsHPP04. The M. javanica effector MjTTL5 recruits the thioredoxin reductase subunit AtFTRc to scavenge ROS during the oxidative burst, and the sugar beet cyst nematode effector 10A06 targets spermidine synthase 2 to bolster antioxidant capacity at the infection site. In Globodera rostochiensis, the ubiquitin carboxyl extension protein GrUBCEP12 suppresses flg22-induced immunity, and the effector Gr29D09 disables the potato hexokinase StHXK1.

The review devotes particular attention to transcription factors, the master regulators that nematodes either hijack or exploit as part of the plant’s own response. Functional studies in Arabidopsis and tomato have identified fifteen co-regulated transcription factor hubs during infection, spanning the MYB, WRKY, and ARF families. One especially revealing case is DEL1, an atypical E2F factor that suppresses salicylic acid accumulation and lignin biosynthesis during root-knot infection by repressing EDS5, a gene encoding a salicylic acid transporter. del1 mutants accumulate more salicylic acid, lignify their galls more heavily, and resist infection, while the nematode’s own effector Mi-ISC-1 independently attacks the isochorismate pathway that feeds salicylic acid synthesis. The convergence of host and parasite manipulation on the same pathway underscores how central it is to the outcome of infection.

The WRKY family illustrates the bewildering duality of transcriptional control in these interactions. WRKY11 and WRKY17 act as positive defenders in Arabidopsis, with mutants of either gene showing heightened susceptibility to both cyst and root-knot nematodes. Yet in tomato, SlWRKY16 and SlWRKY31 behave as negative regulators, and their overexpression enhances infection while suppressing defense-related genes. SlWRKY45 goes further: it interacts with JAZ repressor proteins and, once released, binds the promoter of the jasmonic acid biosynthesis gene AOC, throttling jasmonate production and opening the door to the parasite. Meanwhile, WRKY23 promotes feeding site formation by inducing auxin-responsive genes, and bHLH25 and bHLH27 act synergistically to support syncytium development in cyst nematode infection. The same family can thus contain both shields and traitors, and which role a given member plays depends on species, tissue, and developmental stage.

Effectors also reach deep into the plant’s information-processing machinery. The M. incognita effector Mi16D10 binds SCARECROW-like GRAS transcription factors to accelerate giant cell formation, and Mi2G02 hijacks the trihelix factor GT-3a, converting it into a transcriptional inhibitor that silences cell development genes through the 26S proteasome pathway. The cyst nematode effector 30D08 targets AtSMU2, a spliceosome component, perturbing pre-mRNA splicing across thousands of transcripts, and MiEFF18 attacks the core spliceosomal protein SmD1 to fine-tune transcripts governing the cell cycle and cytoskeleton. Indeed, root-knot nematodes induce splicing reprogramming of roughly 2,898 genes involving more than 9,065 transcripts during gall formation. Beyond transcription, epigenetic mechanisms add another layer of control: DNA hypomethylation occurs early in infection, potentially activating stress genes, while the effector Hs32E03 interacts with the histone deacetylase HDT1 to alter acetylation patterns in ribosomal DNA regions, reshaping chromatin to favor parasitism.

What makes the review compelling is its translational ambition. The authors argue that understanding these molecular vulnerabilities opens concrete routes to resistance: CRISPR/Cas editing of transcription factors that mediate feeding site formation, breeding informed by epigenetic marks associated with defense, and host-induced gene silencing that knocks out essential nematode effectors. They also champion single-cell and single-nucleus RNA sequencing as the next frontier, since bulk transcriptomics averages away the cellular heterogeneity that defines galls and syncytia, and functional redundancy among transcription factors often masks the effect of silencing any single gene. Gland-cell-specific transcriptomics is meanwhile expanding the catalog of known effectors, offering a moving target that breeders and biotechnologists must track.

The bigger picture is one of an evolutionary arms race fought at the level of gene regulation. Nematodes exploit conserved host hubs, from the COP9 signalosome to JAZ proteins and translation initiation factors, that other pathogens also target, suggesting shared vulnerabilities that could yield broadly durable resistance. Plants, for their part, retain the capacity to resist when their immune signaling outpaces the parasite’s reprogramming, as resistant interactions shift the sex ratio of developing nematodes toward males and curtail reproduction. Decoding the precise choreography of effectors, hormones, transcription factors, and epigenetic marks, the authors conclude, is not merely an academic exercise but a prerequisite for engineering crops that refuse to build the parasite’s dining room, a goal with enormous implications for global food security as biotic pressures on agriculture intensify.

Subject of Research: Molecular mechanisms of feeding site formation and transcriptional reprogramming in plant-parasitic nematode interactions

Article Title: Cellular plasticity and transcriptional reprogramming in plant-nematode interactions: insights into feeding site formation and plant defense

Article References: Yusuf, A. G., & Bello, T. T. (2025). Cellular plasticity and transcriptional reprogramming in plant-nematode interactions: insights into feeding site formation and plant defense. Crop Health, 3(1), Article 16. https://doi.org/10.1007/s44297-025-00056-1

Image Credits: AI Generated

DOI: 10.1007/s44297-025-00056-1

Keywords: plant-parasitic nematodes, root-knot nematode, cyst nematode, giant cells, syncytium, transcription factors, effector proteins, auxin signaling, cytokinin, plant immunity, CRISPR, RNA interference