In vitro-propagated seedlings of the wild accession ‘Calcutta 4’ (Musa acuminata subsp. burmannica, subgroup AA), resistant to F. oxysporum f. sp. cubense race 1 and 4 (STR4 and TR4), and the susceptible cultivar to both races ‘Prata-anã’ (Musa sp., subgroup ‘Pome’, AAB), were obtained from the Embrapa Cassava and Fruits Germplasm Bank (Cruz das Almas, BA, Brazil). Seedlings were transplanted to sterile coconut fiber substrate and acclimatized in a greenhouse at 24°C with irrigation every 48 h. After 30 days, 100 mL per plant of a 1g/L monoammonium phosphate (MAP) solution was applied. After 45 days, plants were transferred to 2L pots with fresh sterile coconut fiber. A second fertilization was applied after 30 days using a nutrient mixture of urea, magnesium sulfate, MAP, calcium sulfate, and potassium sulfate (1 g/L each), followed 10 days later with a final fertilization with zinc sulfate (1 g/L).

F. oxysporum f. sp. cubense STR4 strain 218A, isolated from Nanica (‘Cavendish’) plants in São Paulo, Brazil and classified as VCG 0120, recognized as Foc STR4 (Rocha et al., 2018), was maintained at Embrapa Cassava and Fruits. Inoculum was prepared using rice infested with 10⁶ CFU/g of Foc STR4 218A, according to Santana et al. (2024).

‘Calcutta 4’ and ‘Prata-anã’ plants were inoculated by adding 50g of Foc STR4 218A-infested rice to the coconut fiber substrate; controls received non-infested rice. The experiment followed a completely randomized design, with each treatment comprising three independent biological replicates. Total roots from inoculated and non-inoculated control plants were collected at 1, 2, and 4 days after inoculation (DAI) for RNA-seq and RT-qPCR, flash-frozen in liquid nitrogen, and stored at -80°C.

Root samples from both genotypes were collected at 1, 2, 4, 8, and 15 DAI. Root fragments collected at 1, 2, and 4 DAI were analyzed for callose and phenolic compound deposition following Rocha et al. (2022).

For scanning electron microscopy (SEM), root samples from the same timepoints were post-fixed in 1% osmium tetroxide (1h), dehydrated in an acetone gradient (30%, 50%, 70%, 90%, and 100%), dried in a critical point dryer (Blazers CPD 030), gold-coated using a sputter coater (Leica EM SCD 500), and visualized using a scanning electron microscope (JEOL JSM 7001F).To assess Foc STR4 colonization of host tissues, root fragments collected at 1, 2, 4, 8, and 15 DAI from ‘Calcutta 4’ and ‘Prata-anã’ were assessed through clearing and staining according to Rocha et al. (2022).

Seedlings of both genotypes were inoculated with Foc STR4 218A (10⁶ CFU/g rice), with non-inoculated controls as per Santana et al. (2024). Wilt symptoms, leaf yellowing, and rhizome necrosis, were recorded as present or absent at 90 and 120 DAI.

Roots from ‘Calcutta 4’ and ‘Prata-anã’ at 1, 2, and 4 DAI were flash-frozen in liquid nitrogen and stored at -80°C. Root fragments were ground in liquid nitrogen, and total RNA was extracted using 1 mL of PureLink™ Plant RNA Reagent (Invitrogen – ThermoFisher Scientific, Waltham, MA, USA) per sample for 5 min, followed by centrifugation (10 min, 15,000 rpm, 4°C). The supernatant was purified using the Direct-zol™ RNA Miniprep Plus Kit (Zymo Research, Irvine, CA, USA) following the manufacturer’s instructions. RNA integrity was verified by agarose gel electrophoresis and quantified via NanoDrop™ One Microvolume spectrophotometry (ThermoFisher Technologies, Waltham, MA, USA).

Total RNA from ‘Calcutta 4’ at 1, 2, and 4 DAI, and from non-inoculated controls, was stabilized using RNA stable™ (Biomatrica, San Diego, CA, USA) and sequenced at the Genome Quebec Innovation Center, Canada. RNA-Seq libraries were constructed using the mRNA stranded Library Prep Kit (Illumina Inc., San Diego, CA, USA) and sequenced as paired-end reads (2x100 bases) on an Illumina NovaSeq 6000 system (Illumina Inc., San Diego, CA, USA).

Raw paired-end reads were quality-checked with FastQC (FastQC - https://www.bioinformatics.babraham.ac.uk/projects/fastqc/)), then filtered (Fastq QC>30 and minimum length ≥ 100 bp following trimming and adapter removal) using Trimmomatic (Bolger et al., 2014). High-quality reads were aligned in batch mode to the M. acuminata DH-Pahang v.4 reference genome (available at the Banana Genome HUB: https://banana-genome-hub.southgreen.fr/organism/Musa/acuminata), using the STAR software in two-pass Basic Mode (Dobin et al., 2013). Multi-mapped reads were excluded from downstream analyses. Quantification of reads mapping to annotated gene models was conducted using HTSeq-count (Python Software Foundation, Portland, OR, USA) (Anders et al., 2015).

Differential expression analysis was performed using EdgeR’s exact test in Rstudio (Robinson et al., 2010). Differentially expressed genes (DEGs) were identified based on significance determined based on adjusted p-values (FDR ≤ 0.05) using the Benjamini–Hochberg procedure to control the false discovery rate, with the log2 Fold Change (log2FC) employed for determining expression changes between treatments [‘Calcutta 4’-inoculated (I) vs. non-inoculated (NI)].

Multiple functional annotation strategies were employed to identify key processes and metabolic pathways affected by biotic stress, including Gene Ontology (GO), the Kyoto Encyclopedia of Genes and Genomes (KEGG - https://www.genome.jp) (Kanehisa et al., 2016a), EuKaryotic Orthologous Groups (KOG) (Tatusov et al., 2003), and MapMan (Thimm et al., 2004). GO annotation of DEGs was performed by homology using GO enrichment software (https://banana-genome-hub.southgreen.fr/content/go-enrichment), with parameters “Merge similar GO-term: sensitive”, p-value ≤0.05, and q-value ≤0.01 (Droc et al., 2013). For KEGG pathway mapping, DEG protein sequences were analyzed with BlastKOALA (https://www.kegg.jp/blastkoala/) to generate KEGG identifiers (KOs) based on sequence similarity to the KEGG gene database subset (Kanehisa et al., 2016b). Generated KOs were then used to construct KEGG pathway maps (https://www.genome.jp/kegg/mapper/) (Kanehisa and Sato, 2020; Kanehisa et al., 2022). KOG annotation was performed using eggNOG v5.0 (http://eggnog5.embl.de) (Huerta-Cepas et al., 2019) based on sequence similarity. For MapMan pathway analysis (v. 3.5.1) (Thimm et al., 2004), DEG protein sequences were hierarchically annotated using Mercator4 v.5 (Lohse et al., 2014) and mapped to the M. acuminata gene matrix, along with corresponding Log2FC expression values.

Additionally, InterPro (https://www.ebi.ac.uk/interpro/) (Mitchell et al., 2019) and InterProScan (Jones et al., 2014) were employed to assign functional domains to DEG protein products, providing insight into the protein families and domains involved.

Fifteen DEGs were selected for expression validation via reverse transcription quantitative real-time PCR (RT-qPCR) ( Table 1 ). Primers targeting the transcript sequence of each gene were designed using the PrimerQuest Tool (IDT - https://www.idtdna.com/pages/tools/primerquest), with the following parameters: amplicon size 80–150 bp, primer length 18–25 nt, GC content 40-60%, and melting temperature 57 - 63°C (ideally 60°C). Primer specificity was verified using NCBI Primer-BLAST for eukaryotic organisms (https://www.ncbi.nlm.nih.gov/tools/primer-blast). For each treatment, for both ‘Calcutta 4’ and ‘Prata-anã’, total RNA was Dnase-treated as described previously (Section: Total RNA extraction). cDNA synthesis was conducted using the SuperScript™ IV First-strand Synthesis System with Oligo(DT) primers (Invitrogen Thermo Fisher Scientific, Waltham, MA, USA) following the manufacturer’s protocol, using RNA from each sample at a standardized initial concentration of 100 ng/μL. For RT-qPCR validation of target genes, cDNA was diluted 1:20 from stock solutions. Each 10 µL reaction comprised 0.2 µM of each primer (forward and reverse), 5 µL of iTaq Universal SYBR Green Supermix (BioRad, Hercules, CA, USA), 2 µL of diluted cDNA, and 2.6 µL of ddH₂O. Stable reference genes L2 and bTUB3 were employed for ‘Calcutta 4’, and ACTIN1 and GAPDH for ‘Prata-anã’, as previously described for these two pathosystems by Costa et al. (2024). Thermocycling was performed on an ABI StepOne Plus Real-Time Thermocycler (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA, USA) with three independent biological and three technical replicates per reaction. Cycling conditions were as follows: initial denaturation at 50°C for 2 min, 95°C for 10 min; 40 cycles of 95°C for 15 s and 60°C for 60 s. Melting curves were generated in three steps: 95°C for 15 s, 60°C for 60 s, and 95°C for 15 s.

The qPCR data were analyzed according to the linear mixed model below, following the strategy proposed by Steibel et al. (2009). Data on the target and endogenous control genes are analyzed simultaneously:

where C T i j k l m is the C T value; μ is the overall mean, G i is the fixed effect of the i ^th level of Genotype, with i = 1 or 2; P ( G ) ( i ) j is the fixed effect of the j ^th level of Primer (target/endogenous) within Genotype, with j = 1 to 3; T k is the fixed effect of the k ^th level of Treatment, with k = 1 or 2; D l is the fixed effect of the l ^th level of Day, with k = 1 to 3; ( G T ) i k , [ P ( G ) T ] ( i ) j k , ( G D ) i k , [ P ( G ) D ] ( i ) j l , ( T D ) k l , ( G T D ) k l , and [ P ( G ) T D ] ( i ) j k l , represent the fixed effect interactions among the previously described parameters; S m is the random effect of the m ^th Sample, assuming S ̴ N ( 0 , I s σ s 2 ) , where I s represents the identity matrix with dimensions equal to the number of samples and σ s 2 the sample variance; and e i j k l m is the random error associated with C T i j k l m , assuming e ̴ N ( 0 , I e σ e 2 ) , where I e represents the identity matrix with dimensions equal to the total number of observations.

Orthogonal contrasts were constructed for model terms including the effect of Primer to obtain P-values for effect of interest: P ( G ) ( i ) j for the main effect of Genotype, [ P ( G ) T ] ( i ) j k for the main effect of Treatment and Genotype-Treatment interaction, [ P ( G ) D ] ( i ) j l for the main effect of Day and Genotype-Day interaction, and [ P ( G ) T D ] ( i ) j k l   for the Genotype-Treatment-Day interaction. Similarly, contrasts were used to compute − Δ a d j C T as the difference in C T s between the target and the average of the two endogenous genes, and − Δ Δ C T for pairwise comparisons of interest. This analysis enables the appropriate modelling of all sources of variation, based on the assumption of a Gaussian distribution for log-transformed expression levels, utilizing heteroscedastic models that allow for heterogeneous variances and providing a more realistic and precise fit for gene expression data. Additionally, the inclusion of sample-specific effects, representing the total mRNA level, enhances the accuracy and applicability of the results obtained. All analyses were performed in SAS Studio 3.81 (Enterprise Edition, SAS Institute Inc., Cary, NC, USA). The − Δ Δ C T and Log2FC values were used to calculate relative expression values using the 2 − Δ Δ C T method (Livak and Schmittgen, 2001). These values were then employed in graphical representations generated using the R software (R Core Team, 2023), specifically the ggplot2 package (Wickham, 2016).

Root sections from resistant ‘Calcutta 4’ inoculated with Foc STR4 218A showed the presence of callose deposition at 1 DAI, indicated by a strong fluorescence in the root cortex cells ( Figure 1A ). Lower fluorescence was observed at 2 and 4 DAI ( Figures 1B, C ), with no fluorescence observed in the non-inoculated control treatment ( Figure 1G ). In susceptible ‘Prata-anã’, by contrast, neither inoculated or non-inoculated control samples showed any evidence for callose deposition in the stained root sections ( Figures 1D–F, H ). A similar pattern of physiological responses was observed regarding the formation of phenolic compounds. Parenchymal cells in ‘Calcutta 4’ root fragments revealed deposition of phenolic compounds at 1 and 2 DAI ( Figures 2A, B ), followed by a reduction at 4 DAI ( Figure 2C ). No phenolic compound deposition was observed in C4 non-inoculated controls ( Figure 2G ). In the case of the susceptible genotype ‘Prata-anã’, none of the investigated time points or controls revealed evidence for phenolic compound formation ( Figures 2D–H ).

SEM analysis showed only the presence of spores, potentially chlamydospores, on the root surface of ‘Calcutta 4’ at all examined time points ( Figures 3A–C ). In contrast, ‘Prata-anã’ showed spores ( Figures 3D, E ), abundant hyphae and conidiophores on the root surface by 4 DAI ( Figure 3F ). No fungal-related structures were observed in the non-inoculated controls of either genotype ( Figures 3G, H ). Clearing and staining of root samples revealed spore germination, hyphal growth, and chlamydospore accumulation exclusively on the outer root surface of ‘Calcutta 4’ at all examined time points ( Figures 4A–E ), with no evidence of colonization within internal root tissues. Similarly, spores, hyphae and chlamydospores were abundant on the outer root surface of ‘Prata-anã’ at 1 and 2 DAI ( Figures 4F, G ). In contrast, a clear colonization of internal root tissues by Foc STR4 was evident only in ‘Prata-anã’, particularly at 4, 8, and 15 DAI ( Figures 4H–J ). Non-inoculated controls of both genotypes exhibited an absence of fungal spores or hyphae ( Figures 4K, L ).

Fusarium wilt symptoms were assessed at 90 and 120 DAI in both inoculated and control plants of resistant ‘Calcutta 4’ and susceptible ‘Prata-anã’ ( Figure 5 ). In the case of ‘Calcutta 4’ plants, no Fusarium wilt symptoms were observed in either inoculated or non-inoculated control plants at either time point ( Figures 5A–H ). In contrast, inoculated ‘Prata-anã’ plants exhibited characteristic wilt symptoms at both 90 and 120 DAI, including above ground symptoms of yellowing of older leaves progressing to younger leaves ( Figures 5I, J ), and necrosis of rhizome tissues ( Figures 5M, N ). Non-inoculated control plants of ‘Prata-anã’ ( Figures 5K, L, O, P ) did not display any disease symptoms.

Illumina NovaSeq 6000 PE100 sequencing resulted in 447,843,971 reads from the cDNA libraries representing the treatments 1, 2, and 4 DAI with Foc STR4 218A and non-inoculated controls in ‘Calcutta 4’. Following filtering, the total number of reads was reduced to 437,740,123, with 326,407,220 originating from the inoculated treatments and 111,332,903 from the non-inoculated controls. Between 77% and 92% of the ‘Calcutta 4’ reads were correctly mapped to 36,769 gene models in the M. acuminata subsp. malaccensis DH-Pahang v.4 reference genome ( Table 2 ). Illumina RNA-Seq raw sequence data were deposited in the NCBI Sequence Read Archive (SRA) database as a BioProject, accessible under SRA submission number SUB14923683 and BioProject accession number PRJNA1198017.

DEGs with significant fold change (at least ≥2-fold and at a probability level of p ≤ 0.05) between inoculated and non-inoculated control treatments were identified following comparison of mapped read counts. A total of 1,416 DEGs were identified in total across the three examined time points ( Supplementary Table S1 ). Of these, 1,223 were upregulated, with 270, 872, and 81 DEGs at 1, 2, and 4 DAI, respectively. Conversely, 193 DEGs were downregulated, with 150, 33, and 10 identified at 1, 2, and 4 DAI, respectively ( Figure 6 ). The greatest number of DEGs was observed at 2 DAI, with 905, followed by 1 DAI with 420, and 4 DAI with 91 DEGs. Comparison of DEGs across treatments revealed a greater number of shared DEGs at 1 DAI and 2 DAI ( Figure 6 ), indicating a higher expression of certain genes in the early stages of Foc STR4 infection in ‘Calcutta 4’. Changes in expression of shared DEGs across treatments are shown in Figure 7 , where shades of red represent upregulation, and shades of blue represent downregulation. Most genes were highly expressed at 2 DAI, with a distinct expression pattern observed at 1 and 4 DAI. These data highlight the significance of the 2 DAI time point in modulating the early defense response of ‘Calcutta 4’ to Foc STR4 ( Figure 7 ).

Validation of in silico-derived differential gene expression data for ‘Calcutta 4’ and equivalent genes in ‘Prata-anã’ was conducted by RT-qPCR, with expression compared in inoculated samples relative to non-inoculated controls ( Figure 11 ) ( Supplementary Table S2 ). A total of 15 genes were analyzed from up-regulated DEGs in ‘Calcutta 4’ RNA-seq data, following selection based on their potential involvement in the upstream to downstream immune response. All examined genes for ‘Calcutta 4’ displayed relative expression trends similar to those obtained by RNA-seq ( Figure 11 ).

Thaumatin-like protein 1 (Macma4_05_g17130) exhibited a significant relative expression at 1, 2, and 4 DAI in ‘Calcutta 4’ in comparison with the non-inoculated control. Overexpression of the genes Leucine-rich repeat receptor-like serine/threonine/tyrosine-protein kinase SOBIR1 (Macma4_02_g09450), Lipoxygenase 3-2C Chloroplastic (Macma4_03_g08040), putative Disease resistance protein RPS5 (Macma4_04_g37480), Peroxidase 4 (Macma4_06_g16370), NDR1/HIN1-like protein 10 (Macma4_09_g31980), Nematode resistance protein-like HSPRO2 (Macma4_10_g08320), Stilbene synthase 1 (Macma4_10_g24620), and Protein SAR DEFICIENT 1 (Macma4_08_g16850) was significant at 1 and 2 DAI in ‘Calcutta 4’. For the genes Salicylic acid-binding protein 2 (Macma4_04_g03250), Premnaspirodiene oxygenase (Macma4_07_g21750), and Pathogenesis-related protein 1 (Macma4_03_g08420), significant overexpression occurred only at 1 DAI, although positive trends were also observed in other treatments, aligning with RNA-seq data. In contrast to data for ‘Calcutta 4’, where both RNA-seq and RT-qPCR confirmed upregulation of gene expression at 1, 2 and 4 DAI, Stilbene synthase 1 (Macma4_10_g24620) and Premnaspirodiene oxygenase (Macma4_07_g21750), which are involved in phytoalexin biosynthesis, showed no relative expression data by RT-qPCR in any treatments in ‘Prata-anã’. Other genes, including Acid endochitinase (Macma4_01_g21300), Leucine-rich repeat receptor-like serine/threonine/tyrosine-protein kinase SOBIR1 (Macma4_02_g09450), Lipoxygenase 3-2C Chloroplastic (Macma4_03_g08040), NDR1/HIN1-like protein 10 (Macma4_09_g31980), Nematode resistance protein-like HSPRO2 (Macma4_10_g08320), Pathogenesis-related protein 1 (Macma4_03_g08420), Peroxidase 4 (Macma4_06_g16370), Protein SAR DEFICIENT 1 (Macma4_08_g16850), putative Disease resistance protein RPS5 (Macma4_04_g37480), putative L-type lectin-domain containing receptor kinase IX.1 (Macma4_06_g30880), Salicylic acid-binding protein 2 (Macma4_04_g03250), and Thaumatin-like protein 1 (Macma4_05_g17130), also showed higher relative expression in ‘Calcutta 4’ than in ‘Prata-anã’.

These results indicate a trend towards increased expression of genes associated with resistance to Foc in ‘Calcutta 4’, especially in the early stages of the interaction at 2 DAI, contrasting to an absence of upregulation of expression in the susceptible genotype ‘Prata-anã’. Figure 11 also highlights expression trends observed through RT-qPCR analysis supporting those observed by in silico RNA-seq analysis.

Wang et al. (2012) investigated the compatible response in Cavendish subgroup ‘Brazilian’ to TR4, demonstrating activation of pathogen recognition and defense signaling pathways (such as SA, JA, and ET) at 1- and 2-days post-inoculation, with PR protein and lignification enzyme expression at 4 DAI. Early-stage overexpression of peroxidases also suggested ROS activation after TR4 infection. However, these responses were not durable or strong enough to prevent disease establishment.

A comparative study in the Cavendish cultivar ‘Baxi’, resistant to Foc R1 and susceptible to Foc TR4, revealed similar expression patterns at 2 DAI to both races. A notable exception was the suppression of an allene oxide synthase gene during Foc TR4 infection, suggesting that TR4 infection may reduce JA-mediated defense (Li et al., 2013). In a similar study with the Cavendish cultivar ‘Brazilian’ (resistant to Foc R1, susceptible to Foc TR4), Foc R1 induced expression changes in RLK receptors, TFs, metabolic pathways, lignin/flavonoid synthesis, PR-1 and PR-4 chitinase genes, and secondary metabolites, including thos related to glucosinolates, alkaloids, and terpenoids (Dong et al., 2020). In contrast, gene expression and pathway changes after Foc TR4 infection were either unaffected or only weakly affected, highlighting contrasting early-stage responses to each Foc race. Li et al. (2023) further demonstrated suppression of SA pathways during early Foc TR4 infection in the Foc TR4-susceptible cultivar ‘Brazilian’ when compared to Foc R1 infection, to which the genotype is resistant. Notably, RIN4, an NLR protein involved in resistance via the membrane receptor RPM1, was overexpressed only after Foc R1 infection, possibly explaining the susceptibility of ‘Brazilian’ to Foc TR4.

In contrast to susceptible genotypes, resistant cultivars and mutants exhibit a more coordinated and amplified transcriptional response. Studies on the resistant ‘Cavendish’ mutant ‘Nongke No1’ identified nine major immunity-related pathways activated in response to Foc TR4, comprising PAMP perception via PRRs, ETI, ion flux changes, TFs, oxidative burst, PRs, programmed cell death, plant hormones, and cell wall modification (Li et al., 2012). Enhanced signaling through SA and JA pathways, along with glycolysis and glyoxylate cycle-related DEGs, were also observed in this resistant genotype when compared to expression in susceptible ‘Baxi’ (Li et al., 2019b). In TR4-resistant ‘Yueyoukang 1’, genes associated with pathogen recognition (CeBiP, BAK1, NLRs), PR proteins, TFs, and antimicrobial activity (glucanases, and endochitinases) were more strongly expressed in comparison to susceptible ‘Brazilian’ following Foc TR4 infection (Bai et al., 2013). Further transcriptomic analyses identified strong induction of PR proteins (PR-3, PR-4) and numerous cell wall-modifying enzymes, including pectinesterases, β-glucosidases, xyloglucan endotransglucosylase/hydrolases, and endoglucanases (Niu et al., 2018), all contributing to cell wall reinforcement and pathogen containment. Similarly, in Foc TR4-resistant ‘Guijiao 9’, overexpression of genes encoding the disease resistance protein RPM1, the TF PTI6, the elicitor receptor CeBiP, calcium-dependent protein kinase (CDPK), ubiquitin pathway regulators such as UBE2H, and zinc-finger proteins (RCHY1) was reported at 2–6 DAI compared to the susceptible ‘Williams’ cultivar (Sun et al., 2019). These findings collectively highlight the complex, multi-layered immune network deployed by resistant Musa cultivars in response to Foc TR4.

In the case of wild species, Zhang et al. (2019) also observed a constitutive defense response in Foc TR4-resistant ‘Pahang’, involving WRKY and TIFY TFs, and putative disease resistance genes. Upon TR4 challenge, significant upregulation of genes involved in early defense and antimicrobial activity, including PR1, chitinases (CHIT), aspartic proteases (ASP), lipases (GLIP), peroxidases (POD), polyphenol oxidases (PPO), and cytochrome P450s, was observed. In another comparative study comparing Foc-resistant ‘Calcutta 4’ (AA) with the susceptible cultivar ‘Kadali’ (AA), 18 defense-related genes involved in signaling, ROS, lignification, and PR pathways, were identified as playing roles in the defense response in ‘Calcutta 4’ against Foc R1 (Kotari et al., 2018).

Our study provides a significant perspective on the wild accession ‘Calcutta 4’, which is resistant to all Foc races. Focusing on the interaction with Foc STR4, which is genetically the closest race to TR4 and widespread in Brazil, Australia, the Canary Islands, China, South Africa, and Taiwan (Munhoz et al., 2024), this study provides a first transcriptome-level investigation of the resistance response to this widespread race. Histological analyses showed a rapid defense response following Foc STR4 challenge, with callose formation and phenolic compound deposition observed at 1 and 2 DAI. Such responses were absent in ‘Prata-anã’ over the examined time course. When comparing Foc STR4 colonization in resistant ‘Calcutta 4’ and susceptible ‘Prata-anã’, the presence of chlamydospores and hyphae on the root exterior was similar in both genotypes up to 4 DAI. However, at 8 and 15 DAI, hyphae were more prevalent in ‘Prata-anã’, both externally and internally within root cells, with only chlamydospores observed in ‘Calcutta 4’. Previously, Li et al. (2019b) similarly demonstrated a greater abundance of Foc TR4 hyphae and spores in susceptible ‘Baxi’ in comparison to resistant ‘Nongke No.1’ at 27 and 51 hpi. Considering the differences in pathogen growth between these contrasting genotypes, callose and phenolic compounds likely represent defense responses that hinder Foc STR4 colonization in ‘Calcutta 4’ root tissues.

Transcriptome data aligned with findings based on histology, with a greater differential expression of genes at 2 DAI. Among the total of 1,416 DEGs identified in RNA-seq analysis, 270 and 872 were overexpressed at 1 and 2 DAI, respectively, with only 81 at 4 DAI. Of these, 752 DEGs were regulated exclusively at 2 DAI, suggesting a rapid response of ‘Calcutta 4’ to the onset of Foc STR4 infection, as has been shown in other pathosystems. Seventy-five globally upregulated DEGs were identified as possible agents in the defense response to Foc STR4, associated with GO biological processes categorized under response to stimulus. Candidate genes activated in the innate immune response in ‘Calcutta 4’ against Foc STR4 infection, associated with PTI and ETI pathways, are summarized in the schematic model in Figure 12 and listed in Supplementary Table S3 . Comparison with the literature also reveals that many of the core defense components activated in response to Foc STR4 overlap with those triggered by Foc TR4, supporting a degree of conservation in the immune response. Key genes and classes common to the resistance response to Foc STR4 and Foc TR4 are summarized in Supplementary Table S4 .

Resistance responses to PAMPs involve recognition by PRRs, including RLK and RLP receptors, triggering signaling pathways such as Ca²⁺ influx, ROS production, and MAPK activation (Ding et al., 2022; Ngou et al., 2022). Here, numerous such membrane receptors were modulated during the incompatible response in ‘Calcutta 4’ ( Supplementary Table S3 ). These included Protein LYK5 (Macma4_04_g09660), a chitin receptor that binds CERK1 (Chitin elicitor receptor kinase 1) (Macma4_10_g11450) to form a chitin recognition complex involved in non-host resistance (Cao et al., 2014; Huaping et al., 2017). SOBIR1 (Macma4_02_g09450) was also identified, known to activate cytoplasmic signaling, oxidative bursts and MAPK cascades (Gao et al., 2009; Liebrand et al., 2013; Wei et al., 2022). A putative LysM domain receptor-like kinase 4 (Macma4_04_g01380) was also identified, implicated in fungal chitin recognition, PRR activation, and PTI. Similarly, Kotari et al., 2018 observed upregulation of PAMP recognition and defense signaling genes in the resistance response of ‘Calcutta 4’ to Foc R1, including CeBiP, CERK-6, RPK-2, LECTIN, and MAPKKK-11. Following PAMP recognition, Thaumatin-like protein 1 (Macma4_05_g17130) showed PRR-associated activation at 1 and 2 DAI. Mahdavi et al. (2012) demonstrated a reduced incidence of fusarium wilt caused by TR4 in transgenic ‘Pisang Nangka’ (AAB) bananas expressing a rice Thaumatin-Like Protein gene. Upregulation of a PR-protein 1C (Macma4_02_g16620), and a TL1, protein of the PR5 class was also significant in ‘Calcutta 4’ during early interaction with Foc STR4, contrasting with the susceptible ‘Prata-anã’ (Dixon et al., 1991; Hakim et al., 2018; Han et al., 2023). Additionally, the nematode resistance protein-like HSPRO2 (Macma4_05_g32600, Macma4_10_g08320, Macma4_10_g08320, Macma4_11_g23440) was upregulated at 1 and 2 DAI ( Supplementary Table S3 ). Although lacking the canonical nucleotide-binding site, HSPRO2 contains an imperfect LRR domain and contributes to basal resistance against pathogens (Murray et al., 2007; Nemchinov et al., 2017).

Plants defend against pathogens through various chemical and physical mechanisms. Callose, a β-1,3-glucan polymer, is deposited in response to PRR-triggered signaling following PAMP recognition, contributing to cell wall reinforcement and acting as a matrix for antimicrobial compounds (Luna et al., 2011). In Arabidopsis, callose deposition is activated by the flg22 PAMP peptide and regulated by EXO70B1 and EXO70B2 genes (Redditt et al., 2019). Callose deposition can also be influenced by signaling from abscisic acid (ABA) pathways in response to infection by fungi and oomycetes, as demonstrated by the ABA-dependent callose accumulation during Botrytis cinerea infection (You et al., 2010). We confirmed through comparative histological analyses in inoculated and non-inoculated ‘Calcutta 4’ and ‘Prata-anã’ that callose deposition occurred exclusively in ‘Calcutta 4’ at 1 and 2 DAI, following interaction with Foc STR4. This response coincided with the upregulation of Exocyst complex component EXO70B1 (Macma4_10_g10790). Notably, there was also overexpression of the gene Abscisic acid receptor PYL2 (Macma4_06_g02820) at 1 and 2 DAI, a regulatory component of the ABA receptor family (García-Andrade et al., 2020), suggesting a role for ABA-mediated signaling in callose deposition in ‘Calcutta 4’ in response to Foc STR4.

Plant chitinases are key components of resistance to biotic stress due to their ability to hydrolyze chitin, a major constituent of fungal cell walls, thus contributing to primary defense against fungal pathogens. In this study, seven chitinases were upregulated in ‘Calcutta 4’ following interaction with Foc STR4: Hevamine-A (Macma4_08_g34260), Acidic endochitinase (Macma4_01_g21300), Chitinase (Macma4_03_g29600), Basic endochitinase C (Macma4_03_g29630), Chitinase 6 (Macma4_05_g18850 and Macma4_05_g18860), Endochitinase EP3 (Macma4_06_g23390), and Chitin-inducible gibberellin-responsive protein 1 (Macma4_09_g12390). Chitinase involvement in the defense response has previously been demonstrated in ‘Cavendish’ in response to Foc R1 (Dong et al., 2020), in transgenic ‘Gros Michel’ lines expressing a rice chitinase gene (RCC2) conferring resistance to black leaf streak disease (Kovács et al., 2013), and in the ‘Furenzhi’ (AA) cultivar engineered with a Trichoderma harzianum endochitinase gene (chit42) for resistance to Foc race 4 (Hu et al., 2013).

ROS production and SAR activation occur following recognition of pathogen Avr proteins by R proteins. ROS play dual roles in defense: directly reinforcing cell walls to prevent pathogen colonization (Huang et al., 2019) and acting as conserved signaling molecules for chitin perception and MAPK cascades (Kawasaki et al., 2017). In this study, genes associated with oxidative stress responses were predominantly upregulated in ‘Calcutta 4’ at 1 and 2 DAI, including SAR DEFICIENT 1 (SARD1) (Macma4_08_g16850) and several peroxidases ( Supplementary Table S3 ). Peroxidases are multifunctional stress-responsive enzymes with antioxidant activity and a role in phenolic compound biosynthesis. Prior research on the ‘Calcutta 4’ - Foc race 1 interaction showed higher peroxidase levels in this wild subspecies compared to the susceptible ‘Kadali’ (Swarupa et al., 2013).

ETI is activated through the direct or indirect recognition of race-specific pathogen effectors by NLR R proteins, often resulting in a hypersensitivity response (HR) and localized cell death (Jones and Dangl, 2006). Previously, 52 partial NLR genes were identified in ‘Calcutta 4’ (Miller et al., 2008). In ‘Pahang’, RGA2 (resistance gene analogue 2) and four additional putative resistance genes were highly expressed upon Foc TR4 infection (Zhang et al., 2019a). Using the RGA2 gene, Dale et al. (2017) developed the first Foc TR4-resistant transgenic ‘Cavendish’ cultivar (Harding et al., 2025). In this study, we observed upregulation of an NDR1/HIN1-like protein 10 gene, involved in CC-NBS-LRR-type NLR signaling (Macma4_ 09_g31980), as well as several NLR genes at 2 DAI, including NB-ARC domain-containing proteins (Macma4_01_g04350, Macma4_01_g04360, Macma4_01_g04380) and RPS5(Macma4_04_g37480).

TFs are central to signaling and regulation of plant defense pathways under biotic stress. Of the upregulated TFs in ‘Calcutta 4’ during interaction with STR4, most belonged to the TIFY, WRKY, MYB, and MYC families. Two critical WRKY genes were identified: WRKY WRKY28 (Macma4_11_g08360) and WRKY WRKY51 (Macma4_09_g24850). WRKY51 is involved in NPR1-mediated signaling, essential for SA-induced PR gene expression and pathogen resistance (Ding et al., 2018). WRKY50 and WRKY51 proteins also mediate SA- and low-dependent repression of JA signaling (Gao et al., 2011). TIFY family TFs are plant-specific regulators involved in development and responses to both biotic and abiotic stresses. JAZ proteins, a subfamily of TIFY, function as JA signaling repressors and co-receptors, interacting with MYC2, a positive regulator of JA-responsive gene expression (Zhang et al., 2020). In this study, notable TIFY family members were identified, including putative TIFY 5A (Macma4_02_g16390, Macma4_06_g34540, Macma4_09_g10730) and TIFY 10b (Macma4_03_g01960). In Arabidopsis, SA synthesis is promoted by TFs SARD1 and CBP60g (Calmodulin-binding protein 60g), which regulate ICS1 (Isochorismate Synthase) expression and other immunity-related genes (Zhang et al., 2010; Wang et al., 2011; Sun et al., 2015). The detection of SARD1 (Macma4_08_g16850) in our study supports its conserved role as a key immune response regulator.

Plant hormone signaling plays a pivotal role in defense responses. The SA pathway is activated upon recognition of pathogen effectors (Avr) by R genes, triggering downstream signaling that stimulate the induction of SA and the systemic acquired resistance (SAR) response against biotrophic pathogens. SA activation occurs through expression of EDS1, PAD4, and GDG1, as well as via MAPK signaling (Ding et al., 2022). Locally, SA induces HR and PR gene expression; systemically, it triggers SAR, a durable, broad-spectrum immune response in uninfected tissues that involves PR protein accumulation, lignification, ROS production, and antimicrobial compound accumulation, enhancing resistance against future pathogen attack (Klessig et al., 2018; Lukan and Coll, 2022). In this study, Salicylic acid-binding protein 2 (SABP2) (Macma4_04_g03250) and NDR1/HIN1-like protein 10 (Macma4_09_g31980) were upregulated in ‘Calcutta 4’ during early infection by Foc STR4. SABP2, an SA receptor, is essential for PR1 induction and SAR activation (Kumar and Klessig, 2003; Kumar et al., 2006; Vlot et al., 2008). NDR1 promotes SA accumulation (Shapiro and Zhang, 2001; Ding et al., 2022), with overexpression enhancing disease resistance in Arabidopsis (Lu et al., 2013).

Ethylene (ET) and jasmonic acid (JA) pathways, typically activated in responses to necrotrophic pathogens, mediate induced systemic resistance (ISR) (Li et al., 2019a; Ding et al., 2022). JA can induce the expression of PR proteins, peroxidases, alkaloids, volatiles, and structural defense mechanisms. Here, we observed both TFs MYC2 (Macma4_07_g04870) and JAMYB (Macma4_09_g16730) positively regulated at 2 DAI in ‘Calcutta 4’ after Foc STR4 infection. MYC2 activates JA-responsive genes (Du et al., 2017), while JAMYB enhances resistance to viral diseases in rice (Yang et al., 2020). Such upregulation of these TFs indicates a role of the JA pathway in the resistance response of this wild genotype to Foc STR4. Several ET-responsive TFs were also upregulated during the interaction of ‘Calcutta 4’ with Foc STR4 ( Supplementary Table S3 ). ET receptors activate the EIN2 protein, which promotes ET signaling by inducing EIN and EIN3-like (EIL) TFs that bind ERF1 to stimulate ET-mediated defenses (Berrocal-Lobo and Molina, 2004; Dolgikh et al., 2019). In ‘Calcutta 4,’ ETHYLENE-INSENSITIVE 3-like 1a (EIL1) (Macma4_08_g22960) and Ethylene-responsive TF 1B (ERF1B) (Macma4_11_g21380) were positively regulated in response to Foc STR4.

Downstream defense responses following activation of JA/ET pathways typically involve the induction of PR genes, where ERFs function as key regulators of JA/ET-responsive defense genes (Huang et al., 2016). As ERFs are involved in immune responses in Arabidopsis to necrotrophic pathogens, with ERF1 overexpression enhancing resistance to Botrytis cinerea, F. oxysporum, and Plectospherella cucumerina (Huang et al., 2016), the role of ERFs in ‘Calcutta 4’ resistance to Foc STR4 merits further investigation.

The SA, ET, and JA signaling pathways also regulate the synthesis and distribution of secondary metabolites during plant stress responses (Yang et al., 2019). Root-derived phytoalexin and flavonoid secondary metabolites exert antifungal activity by inhibiting or preventing spore germination, thus contributing to defense against phytopathogenic fungi (Agrios, 2005). Premnaspirodiene oxygenase, encoded by Cytochrome P450 (CYP71), confers resistance in rice to Xanthomonas oryzae pv. oryzae (Niño et al., 2016). This enzyme catalyzes regio- and stereospecific hydroxylation of sesquiterpenes to produce solavetivone, a potent antifungal phytoalexin (Takahashi et al., 2007). Stilbenes, another class of phytoalexins within the flavonoid family, are synthesized by Stilbene synthase (STS) which catalyses the final step of the biosynthesis (Jeandet et al., 2002). STS genes have been cloned in several plant species, including peanut (Arachis hypogaea), Scots pine (Pinus sylvestris), eastern white pine (Pinus strobus), Japanese red pine (Pinus densiflora), grapevine (Vitis vinifera), and sorghum (Sorghum bicolor) (Qayyum et al., 2022). STS enzymes are responsive to external signals triggered by abiotic stresses and biotic signaling from fungal cells (Jeandet et al., 2002, 2010). In this study, the genes Premnaspirodiene oxygenase (Macma4_07_g21750) and Stilbene synthase 1 (Macma4_10_g24620) were upregulated in ‘Calcutta 4’ at 1 and 2 DAI during Foc STR4 infection, contrasting with their absence in ‘Prata-anã’, as confirmed by RT-qPCR. These findings suggest an important role for phytoalexins in the defense response of ‘Calcutta 4’ against Foc STR4.