Integrative RNA-Seq analysis of Capsicum annuum L.-Phytophthora capsici L. pathosystem reveals molecular cross-talk and activation of host defence response

Abstract

Chili pepper (Capsicum annuum L.) is economically one of the most important spice. But, it's productivity is highly affected by the pathogen, Phytophthora capsici L. Our current understanding of the molecular mechanisms associated with the defence response in C. annuum-P. capsici pathosystem is limited. The current study used RNA-seq technology to dissect the genes associated with defence response against P. capsici infection in two contrasting landraces, i.e. GojamMecha_9086 (Resistant) and Dabat_80045 (Susceptible) exposed to P. capsici infection. The transcriptomes from four leaf samples (RC, RI, SC and SI) of chili pepper resulted in a total of 118,879 assembled transcripts along with 52,384 pooled unigenes. The enrichment analysis of the transcripts indicated 23 different KEGG pathways under five main categories. Out of 774 and 484 differentially expressed genes (DEGs) of two landraces (under study), respectively, 57 and 29 DEGs were observed as associated with defence responses against P. capsici infection in RC vs. RI and SC vs. SI leaf samples, respectively. qRT-PCR analysis of six randomly selected genes validated the results of Illumina NextSeq500 sequencing. A total of 58 transcription factor families (bHLH most abundant) and 2095 protein families (Protein kinase most abundant) were observed across all the samples with maximum hits in RI and SI samples. Expression analysis revealed differential regulation of genes associated with defence and signalling response with shared coordination of molecular function, cellular component and biological processing. The results presented here would enhance our present understanding of the defence response in chili pepper against P. capsici infection, which the molecular breeders could utilize to develop resistant chili genotypes.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12298-021-01122-y.

Introduction

Chili pepper (Capsicum annuum L.) belongs to the Solanaceae family under the Capsicum genus. This genus consists of several important pepper species such as C. annuum L., C. baccatum L., C. chinense L., C. frutescens L., and C. pubescens L., etc. The chili pepper is native to central and South America and was introduced to the rest of the world by traders (Pickersgill 1997). It has a diverse food usage such as spice, fresh and in powder form (Dias et al. 2013; Wahyuni et al. 2013), and has medicinal uses such as drugs, condiments, ointments, and relief of pain (Pawar et al. 2011; Chamikara et al. 2016). Pepper is the most widely cultivated crop in the Indian states of Andhra Pradesh, Karnataka, Maharashtra, Orissa, and Tamil Nadu, accounting for more than 75% of the total area of India and hence, it is an export-oriented crop in the country (Reddy et al. 2014). In Ethiopia, pepper is cultivated as an income source and is an essential constituent of traditional food because of its pungency and colour (Gebretsadkan et al. 2018). Despite the chili pepper's economic importance in human nutrition, its productivity is highly affected by both biotic and abiotic factors. Among the biotic factors, oomycete pathogens are the most devastating pathogens of pepper causing diseases such as Phytophthora late blight, collar rot, and root rot, which result in severe yield loss (Yin et al. 2012). Among the oomycete pathogens, Phytophthora capsici L. has been identified as one of the major pathogens of peppers limiting its productivity (Barbosa and Bosland 2008). It is a soil-borne pathogen causing the Phytophthora stem, collar, and root rots, and crown blight diseases in Capsicum species (Barksdale et al. 1984; Ristaino and Johnston 1999; Walker and Bosland 1999). It has a broad host range in the Solanaceae (tomato, eggplant, pepper) and Cucurbitaceae (squash, pumpkin, zucchini, cucumber, and watermelon) families (Ristaino and Johnston 1999). Being one of the major pathogens of chili pepper worldwide, current management strategies have not been successful (Barchenger et al. 2018).

Phytophthora capsici can infect all parts of the chili pepper including roots, stems, leaves, and fruit, across the growth stages (Oelke et al. 2003). The CM-334, the most studied resistant landrace of chili pepper, is resistant to root rot, foliar blight, and stem blight (Walker and Bosland 1999). Owing to different combinations of resistance genes, CM-334 is highly resistant to all of the diseases caused even by the most virulent strains of P. capsici (Sy et al. 2008) and has been widely used in breeding programs (Barbosa and Bosland 2008). However, no CM-334 derived lines have been developed to date that could consistently show resistance to all disease symptoms (Oelke et al. 2003; Sy et al. 2008). The available landraces of resistant pepper other than the CM-334 landrace are not equally resistant to all races of the pathogens and failed to resist all the infections initiated by the pathogen (Byron et al. 2010). Due to physiological, genetic, and molecular complexities, investigating the resistance mechanism in C. annuum against P. capsici is quite challenging (Quirin et al. 2005). Furthermore, an attempt to improve chili pepper for disease resistance has been delayed, mainly due to a lack of adequate information on the heritable trait characteristics and host defense mechanisms underlying disease resistance (Marame et al. 2009). In pepper, understanding the disease response associated with P. capsici infection is limited (Wang et al. 2015). The development of next-generation sequencing (NGS) technology coupled with high-precision bioinformatics technologies has helped in the understanding of the molecular mechanism in black pepper (P. nigrum L.) against foot rot (Hao et al. 2016). Wang et al. (2015) used RNA-sequencing technology to identify differentially expressed genes associated with defense responses against P. capsici infection in the resistance "PI 201234″ Pepper line. Liu et al. (2013) demonstrated the role of genes involved in the capsaicinoids biosynthetic pathway in chili pepper using an RNA-Seq approach. Kim et al. (2017) generated a new reference genome of hot pepper and showed the massive evolution of plant disease-resistance genes using retro duplication methods. A novel RGA-based marker analysis was recently performed between the resistant and susceptible Capsicum germplasms under P. capsici infection (Kim et al. 2019). Despite numerous NGS-based plant gene expression analyses, the current understanding of the differential gene expression and resistance mechanisms in the C. annuum L- P. capsici pathosystem is limited. In light of this, it was considered essential to perform a more robust screening of germplasms using NGS technologies, including transcriptomics. Transcriptome sequencing have generated invaluable information about the mechanism of defense response of P. capsici infection in chili pepper. Hence, in the present study, we constructed four RNA-Seq libraries from previously identified resistant (GojamMecha_9086) and susceptible (Dabat_80045) landraces (Rabuma et al. 2020) exposed to P. capsici infection and revealed a variety of differentially expressed genes associated with defense responses.

Methods

Plant growth, P. capsici inoculation, and RNA isolation

To decipher the defense mechanism operating in chili pepper, we selected two contrasting landraces GojamMecha_9086 (resistant) and Dabat_80045 (susceptible). The previously identified, collected, and conserved chili pepper seeds of GojamMecha_9086 with accession number: ETH-9086 and Dabat_80045 with accession number: ETH-80045 were procured from Ethiopian Biodiversity Institute (http://www.ebi.gov.et). The selected genotypes' level of resistance and susceptibility was previously established (Rabuma et al. 2020). The chili plant seedlings were grown in pots containing sterilized soil, sand, and compost with a ratio of 1:1:1 ( Ares et al. 2005), at a temperature range of 20–35 °C with a relative humidity of 50–75% from February till late April 2019 (Rabuma et al. 2020). The seedlings of both, resistant (GojamMecha_9086; control and infected) and susceptible (Dabat_80045; control and infected) genotypes were grown in three biological replications with three seedlings/ replication/try pot for transcriptome sequencing. The resistant landrace GojamMecha_9086 denoted as; RC: resistant control and RI: resistant inoculated and the susceptible landrace Dabat_80045 denoted as; SC: susceptible control and SI: susceptible inoculated, were used throughout this research paper. Sporulation followed by inoculation was performed by irrigating the pure suspension culture of P. capsici (with isolate ID: NCFT-9415.18) with a concentration of 2 × 10⁴ zoospores/mL ( Ares et al. 2005; Ortega et al. 1995) onto one-month-old chili pepper seedling leaves and stems (Bosland and Lindsey 1991). After five days post-inoculation (dpi), the 4th and 5th leaves of control and inoculated leaf samples were collected for RNA isolation as described by Wang et al. (2015). The leaf samples from all three biological replications were harvested separately, followed by washing and freezing in liquid nitrogen, and stored at − 80 °C till further use. The total RNA was isolated using Quick-RNA Plant Miniprep Kit (ZYMO Research) as per the manufacturer’s protocol. The RNA samples from each biological replication were pooled to make four libraries (RC, RI, SC, and SI) to get maximum representation. The quality and quantity of isolated RNA were checked by 1% agarose gel electrophoresis and Nanodrop spectrophotometer, respectively.

Illumina NextSeq500 PE library preparation

Before library preparation, the quality of the RNA samples was thoroughly evaluated by Bioanalyzer 2100 (Agilent, USA) to ensure > 8.5 integrity number (RIN). The RNA-Seq paired-end (PE) sequencing libraries were prepared from the QC passed (RIN > 8.5) RNA samples using Illumina TruSeq Stranded mRNA Sample Prep kit according to the manufacturer manual. The Poly-A containing mRNA molecules were purified using Poly-T oligo attached magnetic beads. Following purification, the mRNA was fragmented into small pieces using divalent cations under elevated temperatures. The first-strand cDNA was synthesized using SuperScript and Act-D mix to facilitate RNA-dependent synthesis. The 1st strand cDNA was then synthesized to the second strand using a second strand mix. The dscDNA was purified using AMPure XP beads followed by A-tailing, and adapter ligation, and then libraries were amplified by PCR with 18 cycles to increase the quantity of the library and enrich for properly ligated template strands (Quail et al. 2009). The PCR enriched libraries were analysed on 4200 Tape Station (Agilent Technologies) using high sensitivity D1000 screen tape according to manufacturer instructions. Finally, the PE libraries were prepared for each leaf sample using TruSeq® Stranded mRNA Library Prep kit as per the manufacturer’s instruction. Bioinformatics data analysis workflow steps were undertaken during transcriptome sequencing of C. annuum L as summarized in (Fig S1).

Cluster generation, RNA-sequencing, and data filtration

After obtaining the Qubit concentration for the libraries and the mean peak sizes from the Agilent Tape Station profile, the PE Illumina libraries were loaded onto NextSeq500 for cluster generation and sequencing. The RNA-sequencing using NextSeq500 was performed using 2 × 75 bp chemistry at Eurofins Genomics, Karnataka, India. The adapters were designed to allow selective cleavage of the forward strands after re-synthesis of the reverse strand during sequencing. The copied reverse strands were used to sequence from the opposite end of the fragment. The sequenced raw data were processed to obtain high-quality paired-end reads using Trimmomatic v0.38 and an in-house script to remove adapters, ambiguous reads by considering parameters, i.e. reads that contain unknown nucleotides “N” larger than 5% were removed and low-quality reads with more than 10% quality threshold (QV) < 20 Phred score were removed (the dirty raw reads, i.e. reads with adaptors, reads with unknown nucleotides larger than 5%, low quality reads) were discarded, the average quality within the window falls below a threshold of 20 was removed, the bases of the start of a read were removed, if below a threshold quality of 20, the bases of the end of a read were removed, if below a threshold quality of 20. Based on these parameters, the clean reads were filtered from raw sequencing data, and the low-quality reads containing unknown nucleotides or adaptor sequences were removed.

High-quality reads from all four samples were pooled together and assembled into transcripts using Trinity de novo assembler (version 2.8.4) with a kmer of 25 as used in (Grabherr et al. 2011). De novo assembly of the short reads (from paired-end RNA-seq library) into transcriptome was carried out with six combinations such as RC vs. RI, RC vs. SC, RC vs. SI, RI vs. SI, SC vs. RI, and SC vs. SI. The assembled transcripts were then further clustered together using CD-HIT-EST-4.6 to remove the isoforms produced during assembly and resulted in sequences that can no longer be extended, called unigenes. Unigenes having > 80% coverage at 3X read depth were considered for downstream analysis. To assess the quality of the assembly, a read content evaluation approach was used. Processed reads from each library were aligned back to the assembled transcripts using Bowtie2 (Langmead and Salzberg 2012) with end-to-end parameters.

Coding sequence (CDS) prediction and gene annotation analysis

The coding sequences were predicted from the unigenes using TransDecoder-v5.3.0. The TransDecoder identified candidate-coding regions within unigene sequences. TransDecoder identifies likely coding sequences based on the criteria, i.e. a minimum length open reading frame (ORF) should be found in a unigene sequence, a log-likelihood score should be > 0, and the above coding score was most significant when the ORF was scored in the 1st reading frame as compared to scores in the other five reading frames. If a candidate ORF was found fully encapsulated by the coordinates of another candidate ORF, the longer one was reported. The sample-wise CDS from the pooled set of CDS were mapped on the final set of pooled CDS using the BWA (-mem) toolkit. The read count (RC) values were calculated from the resulting mapping, and those CDS having 80% coverage and 3X read depth were considered for downstream analysis for each of the samples. Functional annotation of the genes was performed using the DIAMOND alignment program, a BLAST-compatible local aligner for mapping translated DNA query sequences against a protein reference database (Buchfink et al. 2015). DIAMOND (BLASTX alignment mode) finds the homologous sequences for the genes against NR (non-redundant protein database) from NCBI.

Gene ontology analysis

Gene ontology (GO) analyses of the CDS identified for each of the four-leaf samples were carried out using the Blast2GO program. The GO assignments were used to classify the functions of the predicted CDS. The GO mapping also provides an ontology of defined terms representing gene product properties, which are grouped into three main domains: Biological Process (BP), Molecular Function (MF), and Cellular Component (CC). The GO mapping was carried out to retrieve GO terms for all the functionally annotated CDS. The GO mapping uses the following criteria to retrieve GO terms for the functionally annotated CDS: BlastX result accession IDs were used to retrieve gene names or symbols, identified gene names or symbols were then searched in the species-specific entries of the gene-product tables of the GO database, BlastX result accession IDs were used to retrieve UniProt IDs making use of protein information resource (PIR) which includes Protein Sequence Database (PSD), UniProt, SwissProt, TrEMBL, RefSeq, GenPept and Protein Data Bank (PDB) databases. Accession IDs were searched directly in the dbxref table of GO database.

Functional annotation KEGG pathways

From this predicted CDS in the four sequence libraries, functional gene annotation analysis was performed by annotating and mapping via the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Hence, the metabolic pathway analysis of four-leaf samples libraries was carried out using the KEGG database. The potential involvement of the predicted CDS was categorized into 23 KEGG pathways under five main categories, i.e. Metabolism, Genetic information processing, Environmental information processing, Cellular processes, and Organismal systems.

Differential gene expression analysis

Differential expression analysis was performed on the CDS between control and treated samples by employing a negative binomial distribution model in DESeq package (version 1.22.1- http://www.huber.embl.de/users/anders/DESeq/). Dispersion values were estimated with the following parameters: method = blind, sharing Mode = fit-only, and fit type = local. The gene expression levels were reported in transcript per million (TPM). Log2 fold change (FC) value was calculated on the read abundance using the formula: FC = Log2 fold change (Treated/Control). The CDSs having log2foldchange value greater than zero were considered up-regulated, whereas less than zero were down-regulated at the P-value threshold of 0.05.

Validation of transcriptome sequencing by qRT-PCR

We utilized the same RNA samples used in RNA-Seq for qPCR-based validation of selected genes. The cDNA was synthesized using RevertAid First-strand cDNA synthesis kit using Oligo dT primers as per manufacturer protocol. For validation of differentially expressed genes, six differentially expressed genes associated with P. capsici infection were randomly selected for validation. Amongst actin-7, GAPDH and ARF, geNorm analysis showed that actin-7 like gene was most stable across the samples and, therefore, used as an internal control gene for data normalization. Primers were designed using Primer Express 3.0.1 software (Table S1). The qRT-PCR reaction was performed in three biological replications for each sample using SYBR®Green Jump Start™ Taq Ready Mix™ (Sigma) on Applied Biosystems’ Step One™ Real-Time PCR System (Choudhri et al. 2018). The qRT-PCR analysis was performed with a total reaction volume of 10 µL containing 5 µL of 2 × SYBR Green JumpStart Taq ready mix, 1 µL of each forward primer (10 μM) and reverse primer (10 μM), 2 µL of 100 ng/µL cDNA and 1 µL of nuclease-free water. The amplification of the qRT-PCR run method was adjusted as the following conditions: the reaction system was heated to 94 °C and denatured for 2 min, the primary reaction (denaturation at 94 °C for 15 s, annealing 58–60 °C for 60 s, and template extensions at 72 °C for 15 s) for 40 cycles. The qRT-PCR product was analysed using StepOne™ Software v.2.2.2. The C. annuum actin-7 like gene (GenBank: GQ339766.1) was used as an internal control, and expression levels between control and treated leaf samples were analysed using the 2^−ΔΔCt method (Livak and Schmittgen 2001). The standard error in expression levels of three biological replications was denoted by error bars in the figures, and the results were expressed as mean value ± SD. The qRT-PCR amplified product was run and visualized using 1% agarose gel electrophoresis as the method used by Gupta et al. (2017).

Results

Phenotypic characterization of inoculated chili pepper

Before performing the RNA-Seq analysis of resistant and susceptible chili pepper landraces, we carried out the screening of available Indian and Ethiopian landraces for resistance and susceptibility. A total of 233, i.e. 148 Ethiopian and 85 Indian landraces, were used for phenotypic screening upon infection with P. capsici (Rabuma et al. 2020). From phenotypic characterization, we selected two contrasting landraces, GojamMecha_9086 (resistant;with MDI value ~1.4) and Dabat_80045 (susceptible;with MDI value ~8.6), for transcriptome analysis based on the mean disease severity index value and phenotypic performance against pathogen infection. The results indicated no apparent symptoms on the leaves of resistant landrace GojamMecha_9086, but minor lesions were observed on roots (Fig. 1A, B). In contrast, brown lesions were observed on the root, crown root, and lower part of the plant along with leaf wilting in susceptible landrace Dabat_80045 (Fig. 1C, D).

Fig. 1.

Phenotypic characterization of chili pepper exposed to P. capsici. A GojamMecha_9086 control; B GojamMecha_9086 infected; C Dabat_80045 Control; D Dabat_80045 infected

Transcriptome sequence statistics and De novo transcriptome assembly

Transcriptome sequencing analysis was carried out to reveal differential expression of defense-related genes in resistant and susceptible chili pepper leaf samples under P. capsici infection. A total of 4 libraries (RC, RI, SC, SI) were constructed and sequenced. Illumina NextSeq500 sequencing generated an average of 22 million high-quality (QV > 20) paired-end raw reads (~ 3.35 GB) from each library and was used for downstream analysis after pre-processing. The RNA-Seq raw data of four samples were processed to obtain high-quality paired-end reads (Table 1). Pooled assembly from all four libraries was performed, which generated 118,879 assembled transcripts with a mean transcript length of 813.23 and N50 of 1277 bp (Table 1). The assembled transcripts were further clustered together, and resulted in 52,384 pooled unigenes with a total and mean length of 53,922,230 and 1029.36 bp, respectively (Table 1). The highest number of unigenes (16,671) were observed in the length range of 1000–2000 bp (Table S2, Table S11). The CDS prediction analysis from pooled unigenes resulted in 25,552 CDS with a total and mean length of 21,689,736 and 848.84 bp, respectively (Table 1). Reads mapping back to the pooled CDS resulted in 14,509, 19,250, 10,100, and 22,719 CDS in RC, RI, SC, and SI leaf samples, respectively (Table 1). For functional gene analysis, the predicted CDS were searched against the NCBI non-redundant (NR) protein database and resulted in the annotation of 23,738 CDS. Out of 25,552 predicted CDS, 1814 CDS did not show any significant hits and were marked as novel CDS detected in our RNA-Seq data (Table 1 and Table S9). The majority of the BLAST hits of the CDS were exhibited against C. annuum (17,833, 75.13%), followed by C. chinense (2803, 11.80%) and C. baccatum (1896, 7.98%).

Table 1.

Statistical analysis of RNA-Seq libraries constructed from two contrasting landraces, i.e. GojamMecha_9086 (resistant) and Dabat_80045 (susceptible) chili pepper leaf samples exposed to P. capsici infection

Parameters	GojamMecha_9086 (Resistant)		Dabat_80045 (Susceptible)
	Control	Infected	Control	Infected
High quality reads	19,480,516	25,078,845	22,055,840	22,708,051
Number of bases	2,930,503,524	3,773,086,364	3,318,642,109	3,415,521,810
Total data (in Gb)	2.93	3.77	3.31	3.41
#CDS	14,509	19,250	10,100	22,719
Total CDS length(in bp)	11,088,237	15,877,689	7,045,395	19,950,423
Maximum CDS length	13,542	4,878	13,542	13, 542
Minimum CDS length	276	258	264	264
Mean CDS length	764	825	698	878

Pooled assembly statistics
Transcript description	Transcript stat	Unigene	Unigene stat	CDS	CDS stat
Total transcripts	118,879	Total unigene	52,384	Total CDS	25,552
Total transcript length (bp)	96,675,995	Total unigene length (bp)	53,922,230	Total CDS length (bp)	21,689,736
N50	1277	N50	1403	Maximum CDS length (bp)	13,542
Maximum transcript length (bp)	14,668	Maximum unigene length (bp)	14,668	Minimum CDS length (bp)	13,542
Minimum transcript length (bp)	180	Minimum unigene length (bp)	180	Mean CDS length (bp)	848.84
Mean transcript length (bp)	813.23	Mean unigene length (bp)	1029.36

The four samples CDS pooled data distribution statistics
Sample name	Total no. of CDS	# CDS with blast hits	# CDS without blast hits
Pooled CDS	25,552	23,738	1,814

P. capsici induced differentially expressed genes (DEGs)

The assembled transcripts were computationally analysed to evaluate differential gene expression in response to P. capsici infection. We identified 774 DEGs between RC and RI leaf samples, of which 318 (41.01%) and 456 (58.92%) genes were differentially up and down-regulated (P < 0.05), respectively (Table S3). Similarly, 484 DEGs were identified between SC and SI leaf samples, of which 217 (44.84%) and 267 (55.17%) genes were differentially up and down-regulated (P < 0.05), respectively (Table S3). Relatively high numbers of up and down-regulated genes were obtained between RC vs. RI leaf compared to SC vs. SI leaf samples. Among all the DEGs in RC vs. RI leaves, 57 DEGs (45 up and 12 down-regulated) were identified to be associated with the defense response against P. capsici infection (Table S4). Similarly, among 484 DEGs in SC and SI leaves, 29 DEGs were related to defense responses, with 18 and 11 transcripts being down- and up-regulated, respectively. Amongst defense-related genes encoding putative late blight resistance protein homolog R1B-13, non-specific lipid transfer protein GPI-anchored 2-like, L-ascorbate peroxidase 3, peroxisomal-like, ethylene-responsive transcription factor CRF6 isoform X2, and pathogenesis-related leaf protein did not show differential expression between SC and SI leaf. However, most of these genes were differentially expressed between RC and RI leaves. Furthermore, based on the previous reports on defense-related genes, the top 50 differentially expressed genes from both the samples set, i.e. RC vs. RI and SC vs. SI, were selected to construct heatmap along with hierarchal clustering (Fig. 2). For instance, genes encoding UDP-glycosyltransferase 82A1 protein, defensin J1-2-like protein, glycine-rich cell wall structural protein-like, and chalcone synthase 1B enzyme were up-regulated. Additionally, Volcano Plot was prepared to depict the graphical representation and distribution of differentially expressed genes between control and treated samples. The Volcano plot arranges expressed genes along dimensions of biological and statistical significance (Fig. 3A, B).

Fig. 2.

Heat map depicting the top 50 DEGs. The gradient of colors shows: red with highly expressed while green represents no or low expression; A the normalized gene expression values between RC vs RI leaf sample, B the normalized gene expression values between SC vs. SI leaf sample (color figure online)

Fig. 3.

Volcano plot of DEGs between A RC and RI sample and B SC and SI sample. The red block on the right side of zero represents the up-regulated genes whereas the green block on the left side of zero represents significant down-regulated genes. In addition, the Y-axis represents the negative log of the P value (< 0.05) of the performed statistical test, the data points with low P values (highly significant) appear towards the top of the plot. In addition, the grey block shows the non-differentially expressed genes; C) Venn diagram showing common DEGs across all the 4 leaf samples, i.e. control (RC, SC) and treated (RI, SI)

GO classification and enrichment analysis of DEGs related to pathosystem

The GO mapping was performed to assign the functions to the coding sequence (CDS) predicted in each of the four-leaf samples. BlastX result accession IDs were used to retrieve UniProt IDs making use of the Protein Information Resource (PIR), like PSD, UniProt, SwissProt, TrEMBL, RefSeq, GenPept, and PDB databases. BlastX result accession IDs were searched directly in the gene product dbxref table of the GO database. Accordingly, the function of annotated CDS was grouped into three main domains: biological processing (BP), molecular function (MF), and cellular component (CC). A total of 5605 and 7913 CDS wereobserved as involved in biological process, 4353 and 6063 in cellular components, 6751 and 9475 in molecular function in RC and RI samples, respectively (Fig. 4A; Supplementary Fig S2 and Table S5). Besides, 3980 and 9231 CDS were involved in biological processes, 3271 and 7074 CDS in cellular components, and 4721 and 11,084 CDS in molecular function in SC and SI leaf samples, respectively (Fig. 4B; Supplementary Fig S3 and Table S5).

Fig. 4.

The Web Gene Ontology Annotation Plot (WEGO) showing the distribution of DEGs into three GO terms, i.e. cellular, molecular, and biological functions. A RC and RI leaf sample; B SC and SI leaf samples

In all the four-leaf samples, the highest number of CDS (112, 159, 74 and 177 in RC, RI, SC and SI, respectively) were shown to be involved in biological processing related to the various organic metabolic process followed by the cellular metabolic process (105, 150, 69, and 171 in RC, RI, SC and SI, respectively). In the cellular component, the highest numbers of CDS were involved in the membrane (2189, 3221, 1552 and 3850 in RC, RI, SC and SI, respectively) followed by an intrinsic component of membrane (1793, 2725, 1381 and 3261 in RC, RI, SC and SI, respectively). Similarly, in molecular function, the highest number of CDS (993 and 575 in RC and SC respectively) were involved in binding of both organic cyclic compound and heterocyclic compound while 1325 (RI), and 1521 (SI) CDS were involved in transferase activity, which might be related to defense against P. capsici pathogens. Annotated genes related to the immune system process were detected in only resistance inoculated (RI) leaf samples, showing that expressions of these genes might be associated with defense against this pathogen (P. capsici) (Fig. 4A).

KEGG pathways mapping, plant transcription factor (pTF), and protein family (Pfam)

To further investigate the defense response mechanism in chili pepper upon infection with P. capsici, KEGG pathway analysis of DEGs was performed in all four assemblies. The identified CDS were mapped to reference canonical pathways in the KEGG to elucidate the involvement of CDS in various KEGG pathways. The enrichment analysis of the transcripts (4183, 5089, 3320 and 5515 CDS in RC, RI, SC, and SI leaf samples, respectively) indicated 23 different KEGG pathways under five main categories (Fig. 5). The SI leaf sample recorded the highest number of genes (5515 CDS) across five categories of biological pathways and followed by the RI sample (5089 CDS) (Fig. 5). The KEGG analysis included KEGG Orthology (KO), enzyme commission (EC) numbers, and metabolic pathways of predicted CDS using the KAAS server. Amongst all the pathways identified, a total of 1733, 2197, 1354, and 2324 CDS in RC, RI, SC, and SI, respectively, were functionally assigned to metabolism, making it the largest KEGG classification group including carbohydrate metabolism, energy metabolism, lipid metabolism, nucleotide metabolism, amino acid metabolism, etc. The detailed description, including the number of CDS in each category and samples, is mentioned in Table S6. Nevertheless, we also predicted the TF family to dissect its abundance in all four assemblies against 320,370 known TFs in the PlantTFDB 4.0 database (http://planttfdb.cbi.pku.edu.cn/) based on E-value < 1e−05 via BlastX. The analysis identified a large number of transcription factor families across all four assemblies (Fig. 6). A total of 58 families, including bHLH, MYB_related, C3H, WRKY, etc., were identified with basic helix loop helix (bHLH) having the highest hits of 714, 1022, 498 and 1222 in RC, RI, SC and SI leaf samples, respectively followed by NAC having 487, 721, 364 and 826 CDS in RC, RI, SC &andSI leaf samples, respectively (Table S7). Additionally, protein family (Pfam) prediction was performed to know the abundance of protein families for each CDS in all four assemblies based on E-value < 1e−05 via BlastX to Pfam 32.0 databases. Results indicated an average of ~ 3982 families with ~ 2095 common protein families across all the four assemblies of chili pepper upon P. capsici infection (Fig. 7). The distribution of the top ten predicted protein families revealed that protein kinase domain (PF00069) was the most abundant family (290, 489, 160 and 571 CDS in RC, RI, SC and SI, respectively followed by RNA binding protein (PF00076) domain (184, 224, 130, and 255 in RC, RI, SC and SI, respectively) and protein tyrosine kinase (PF07714) (Fig. 7).

Fig. 5.

The distribution of top five highly enriched KEGG pathways in resistant and susceptible landraces of chili peppers exposed to P. capsici infection [The Y-axis is indicating the number of CDS assigned to each KEGG pathway across the four-leaf samples (X-axis)]

Fig. 6.

The distribution of significantly enriched transcription factors families in resistant and susceptible landraces of chili peppers exposed to P. capsici infection

Fig. 7.

The distribution of top 15 most abundant protein families in resistant and susceptible landraces of chili peppers exposed to P. capsici infection

Mining of SSRs markers associated P. capsici infection

Simple sequence repeats (SSRs), also known as microsatellites, are tandem repeat motifs of 1–6 bases and serve as the vital molecular markers in population and conservation genetics, molecular epidemiology and pathology, and gene mapping. In the present study, SSRs were detected using MIcroSAtellite Identification Tool (MISA v1.0) from the pooled set of unigenes. The potential SSRs were identified as ranging from dinucleotide motifs with a minimum of ten repeats, trinucleotide motif with a minimum of four repeats, tetra, penta, and hexanucleotide motifs with a minimum of five repeats. A maximum distance of 200 nucleotides was allowed between two SSRs leading to 7233 SSR from 52,384 pooled unigene sequences. Based on upstream and downstream 200 bp flanking sequence, the SSRs were validated in silico and finally, 3174 SSRs were obtained (Table S8). The analysis result depicted that tri-nucleotide repeats were the most abundant (6755) SSRs in all four leaf samples, followed by di-nucleotide repeats with 288 counts. Similarly, motif prediction of these SSRs showed an abundance of T/A followed by CT/GA in all four assemblies.

Validation of expression of selected DEGs

Transcriptome sequencing result obtained from Illumina NextSeq500 PE were validated using qRT-PCR by randomly selecting six genes. The qRT-PCR data were found consistent with the DEG output in both resistant and susceptible genotype. Among the qRT-PCR analysed genes, all the six transcripts exhibited the same expression pattern with transcriptomic DEG output (Fig. 8A–F). In resistant genotype GojamMecha_9086 upon infection, all of the six transcripts, i.e. PLTP (FC: 1.96, P value: 0.02), EP3 (FC: 2.08, P value: 0.02), Defensin J1-2 like (FC: 3.69, P value: 0.00), PRP (FC: 0.78, P value: 0.04), LTLP (FC: 3.12, P value: 0.00), and ERTF (FC: 1.77, P value: 0.00) genes were differentially up-regulated while in susceptible genotype Dabat_80045, PRP (FC: 1.22, P value: 0.05), and LTLP (FC: 0.38, P value: 0.04) genes were up regulated whereas PLTP (FC: − 1.66, P value: 0.03) and Defensin J1-2 like (FC: − 0.70, P value: 0.02) were down regulated. The qRT-PCR amplified product of five differentially expressed genes from control and treated samples are shown in Fig S4 (Fig S4 Original Gel image).

Fig. 8.

qRT-PCR validation of six key DEGs associated with defense response upon P. capsici infection. A putative lipid transfer protein DIR1 gene (PLTP), B endochitinase EP3-like gene(EP-3), C defensin J1-2-like gene (Defensin), D lipid transfer-like protein VAS gene (LTLP), E pathogenesis-related protein STH-2-like gene (PRP), and F ethylene-responsive transcription factor RAP2-7-like isoform X2 gene (ERTF). The y-axis represents the log fold relative gene expression, and the x-axis represents the chili pepper leaf samples (RC, RI, SC, and SI). Actin-7 like gene was used as an internal control gene for data normalization. The experiment was replicated three times, and the resulting data was presented with an errors bar, with n = 3

Discussion

Capsicum annum L. is among one of the most important commercial spices. But, productivity of C. annuum L. is highly affected by several pathogens, including P. capsici. Understanding the disease resistance gene in C. annuum in response to P. capsici infection is limited (Wang et al. 2015). In recent years, RNA-Seq has emerged as one of the fundamental NGS techniques for rapid and efficient characterization of differential gene expression associated with pathogen infection under control and treated conditions (Wang et al. 2010). It provides a valuable resource for the elucidation of molecular mechanisms and candidate genes for functional analysis (Wang et al. 2019). The present study was designed to carry out an RNA-Seq analysis of two contrasting (resistant and susceptible) chili pepper landraces upon infection with P. capsici to decipher the differentially expressed genes associated with defense response.

RNA-Seq data resulted in 118,879 assembled transcripts with a mean transcript length of 813.23 and N50 of 1277 bp, which is higher than 47,575 assembled transcripts with an average length of 1437.22 bp reported by Wang et al. (2015) in pepper line PI 201234. Further analysis resulted in ~ 52,384-pooled unigenes from four libraries compared to 116,432 unigenes obtained from six libraries of P. nigrum L. in response to P. capsici infection (Hao et al. 2016). About 92.9% of CDS were functionally annotated, among which 75.13% of the CDS were matching with available C. annuum reference genome, representing a high degree coverage and functionality of our RNA-Seq data.

The Gene Ontology mapping showed that metabolic processing, membrane, and binding terms were the dominant in biological process, cellular component, and molecular function, respectively, which has not been in agreement with the report of Hao et al. (2016), which might be due to different genetic background of the selected genotypes for resistance and susceptibility level to P. capsici. Predicted CDS genes assigned to molecular function were more enriched in RI (9475 CDS) and SI (11,084 CDS) samples relative to their corresponding control signifying their role during infection, which is in agreement with the Richard et al. (2010). This suggests that genes of catalase activity in resistance genotype (CM-334) and glutathione S-transferase in susceptible genotype (NM6-4) were associated with defense response against P. capsici infection. Our sequencing result identified many genes encoding pepper 9-lipoxygenase involved in organic metabolic process, which is crucial for lipid peroxidation processes during plant defense responses to pathogen infection (Hwang and Hwang 2010). Hao et al. (2016) have reported the active role of phenylpropanoid, and terpenoids during a counter-attack against invading P. capsici in black pepper (P. nigrum L.). Metabolic pathway genes were activated during pathogen infection and abiotic stress (Pageau et al. 2006). Carbohydrate metabolism is the largest KEGG pathway group observed in the present results, signifying their active role during defense response against infection, which is in line with the work of Li et al. (2018) in G. littoralis. Also, our result is supported by Wang et al. (2015), signifying that higher accumulation of CDS related to terpenoids, polyketides metabolism, secondary metabolites, and xenobiotics biodegradation in P. capsici treated leaf compared to control might be playing an important role during interaction with the pathogen.

The RNA seq identified many differentially expressed genes associated with resistance against P. capsici infection, including genes encoding for lignin-forming anionic peroxidase-like precursor, ethylene-responsive transcription factor RAP2-7-like isoform, putative lipid-transfer protein DIR1, lipid transfer-like protein VAS, Acidic endochitinase Q, endochitinase EP3-like, defensin J1-2-like, UDP-glycosyltransferase 82A1, glycine-rich cell wall structural protein-like, phenylalanine ammonia-lyase, non-specific lipid-transfer protein-like protein At5g64080 isoform X2, non-specific lipid transfer protein GPI-anchored 2-like, lipid transfer protein EARLI 1-like and GDSL esterase/lipase APG-like were differentially up-regulated under P. capsici infection in the RI leaf. However, in the SI leaf, gene encoding non-specific lipid-transfer protein-like protein At5g64080 isoform X2, putative lipid-transfer protein DIR1, and non-specific lipid-transfer protein-1 were differentially down-regulated. These pathogenesis-related (PR) genes were jointly involved in the defencing role against the P. capsici infection. Varieties of PR proteins were coordinated by cross-talk stress signals and endogenous plant hormones such as salicylic acid (SA), methyl jasmonate (MeJA), and ethylene (ET) (Kim et al. 2019). In pepper, the effect of many PR genes (polygenes) and the regulatory mechanisms against the P. capsici infection were complex (Wang et al. 2015). The plant responds to the pathogens by increasing the expression of the number of genes associated with disease resistance (Park et al. 2004). For instance, high expression levels (1953 DEGs) of transcripts were reported in P. flaviflorum (resistance) than P. nigrum (susceptible) in response to P. capsici (Hao et al. 2016). For instance, Rabuma et al. (2021) validated the genes encoding glycine-rich cell wall structural protein-like (GRCW) and UDP glycosyltransferase 74E2-like (UDP-GcT) were up-regulated in resistant chili pepper (RI leaf), while down-regulated in SI leaf exposed to P. capsici. Moreover, the over-expression of pathogenesis-related (PR) genes encoding chitinase, glucanase, thaumatin, defensin, and theonin have increased the level of defense response in plants against a range of pathogens (Ali et al. 2018). The putative chitin synthase (CHS) genes encoding the endochitinase enzyme, are present in the genome of P. capsici (Hinkel and Giraldo 2017; Cheng et al. 2019). Thus, chitin present in the zoospore and sporangial cell wall is the most consistent of the transcriptional pattern of PcCHS, showing chitin synthase involved asexual reproduction, vegetative growth, and pathogenesis of P. capsici (Cheng et al. 2019). The chitin in the zoospores of P. capsici can act as a pathogen-associated molecular pattern (PAMP) recognized by the chitin receptors AtLYK5 or AtCERK1 in Arabidopsis (Cheng et al. 2019). Hence, the biological significance of chitin and CHSs genes in Phytophthora spp. is used for the identification of potential targets in disease control. According to our findings, differential expression of the gene (~ 2.3 times) encoding endochitinase enzyme in resistant landrace (RI) is associated with P. capsici pathogenesis, whereas down-regulation of the gene (~ − 3.08 times) encoding endochitinase in susceptible landrace (SI) renders it prone to pathogen infection.

Pathogen invasion triggers the activation of a series of genes associated with various signalling pathways as a counter-attack vis-à-vis defense mechanism to cope with the infection. This type of differential cross-talk between pathogen and host explains why one genotype is resistant while others are susceptible to the same pathogen. Reports have indicated differential regulation of numerous genes associated with defense response, including PR genes, hormone homeostasis in peppers against P. capsici infection (Wang et al. 2013a, b, 2015; Kim et al. 2019). The CanPOD gene was significantly induced in leaves of pepper by P. capsici infection (Wang et al. 2013a). The CALTPI and CALTPIII genes were predominantly expressed in various pepper tissues infected by P. capsici (Jung et al. 2003). The lipid transfer protein (LTP) (Ca-LTP1) gene has been reported involved in pathogen cell morphological deformation and changes (Majid et al. 2016). We also recorded up-regulation of lipid transfer-like protein VAS (~ 3.12 times) in RC vs. RI leaf. The LTP proteins work together to regulate stomatal closure during pathogen infection (Wang et al. 2021). Potato plants overexpressing S. tuberosum lipid transfer protein (StLTP10) display increased resistance to P. infestans through regulating abscisic acid‐mediated stomatal closure, reducing reactive oxygen species accumulation, and enhancing the expression of salicylic acid‐responsive genes (Wang et al. 2021). P. infestans entrance into the leaves may occur when zoospores or germ tubes pass through stomata or other natural openings (Li et al. 2015; Judelson and Fong 2019). Also, studies have shown that overexpression of pepper nsLTPs CALTP1 and CALTP2 enhanced resistance to oomycete and bacterial pathogens (Sarowar et al. 2009). Expression of the defensin J1-1 gene was reported in the pathogen wounded tissues, which inhibits fungal growth by binding with the fungal membrane due to electrostatic and/or hydrophobic interactions (Seo et al. 2014). Most identified crop plant defensins have been overexpressed under pathogen attack, wound, and some abiotic stress leading to resistance (Beer and Vivier 2011). Several reports show that defensins are an integral part of the plant’s innate immune system (Lacerda et al. 2014). They possess an enormous multiplicity of biological activities, such as antimicrobial, insecticidal, inhibiting protein synthesis, etc. (Carvalho and Gomes 2009, 2011). Leaves of tomato plants overexpressing the radish defensin were extracted and tested against A. solani, F. oxysporum, P. infestans, and R. solani (Parashina et al 2000). A defensin purified from maize (ZmDEF1) in transgenic tobacco plants, showed increased tolerance against P. parasitica (Wang et al. 2011). Defensin protein gene (lm-def gene) isolated from Andean crop maca (L. meyenii) in vitro inhibited hyphal growth of P. infestans (Solis et al. 2007). Expressing the D. merckii defensin (DmAMP1) gene in papaya plants increased resistance against P. palmivora associated by reducing hyphae growth at the infection sites (Zhu et al. 2007). Moreover, transgenic tomato plants constitutively expressing the chili pepper defensin (cdef1) gene resulted in enhanced resistance against oomycete pathogen, P. infestans (Zainal et al. 2009). Our result showed increased expression (~ 3.69 times) of defensins J1-2 like protein in resistant landrace compared to susceptible one showing its defense role in RI leaf against P. capsici infection. Hence, the role of defensin in chili pepper might be associated with hampering the growth of P. capsici hyphae.

Ethylene response factor (ERF) is a multi-member family, which has a key role during plant-pathogen interaction. Ethylene biosynthesis was distinctly accumulated with basic pathogenesis-related (CABPR1) mRNA in pepper leaves upon infection with P. capsici (Young and Hwang 2000). One member, CaPTI1 is shown to be up-regulated and involved during defense response in pepper against P. capsici (Jin et al. 2016). The role of ERF in defense during P. capsici infection has also been validated by virus-induced gene silencing (VIGS) of CaPTI1 (Jin et al. 2016). Our results showed increased (~ 1.77 times) ERF RAP2-7-like gene in resistant landrace upon P. capsici infection, suggesting its potential role in modulating the defense response in resistant landrace. Similarly, UDP-glycosyltransferase is shown to be associated with modulating the tolerance of plants against various biotic and abiotic stresses (Zhang et al. 2014). Li et al. (2016) have documented the induced accumulation of the gene encoding UDP-glycosyltransferase in C. annuum exposed to 24-epibrassinolide (EBR). UDP-glycosyltransferase 82A1 was observed as up-regulated (~ 3.82 times) in resistant landraces compared to susceptible ones suggesting active metabolite flux channelling towards the development of resistance phenotype. Furthermore, squamosa promoter-binding protein 1 was down-regulated (− 1.64 times) between RC and RI leaf. Zhang et al. (2020) reported silencing of squamosa promoter binding protein (SBP)-box (CaSBP08) gene enhanced resistance to P. capsici infection in pepper. Our RNA seq analysis observed that phenylalanine ammonia-lyase was up-regulated (~ 1.57 times) in infected resistance compared to infected susceptible plants, showing its association with P. capsici infection. Similarly, Li et al. (2020) reported activation of phenylalanine ammonia-lyase (PAL) (a key enzyme of phenylpropanoid biosynthesis) in the resistant accession (CM-334) of chili pepper compared to the susceptible one (NMCA10399).

Different transcription factors such as MYB, bHLH, and NAC are involved in modulating the gene cascade during various developmental stages and defense response against biotic and abiotic stresses (Agarwal et al 2006; Du et al. 2009; Nuruzzaman et al. 2013; Javed et al. 2020). Our RNA-Seq data enabled us to identify a total of 58 TFs amongst which bHLH, MYB, and NAC were the most abundant ones having differential expression in both resistant and susceptible landraces, which might be involved in regulating the defense-related signalling pathway upon P. capsici infection.

Conclusion

In this study, Illumina NextSeq500 of RC, RI, SC, and SI leaf samples generated ~ 22 million high-quality reads. From the transcriptome sequencing analysis, the genes associated with defense response to P. capsici infection were identified in the leaf sample of C. annuum. Among the total differentially expressed genes in the C. annuum-P. capsici pathosystem, 57 DEGs wereobserved as associated with defense response against P. capsici infection between RC and RI leaf samples. It was shown that more genes involved in the defense against P. capsici were differentially expressed in RI leaves relative to SI leaves. The level of resistance and susceptibility among the chili pepper landraces is primarily due to the difference in the level of gene expression and molecular variations in the resistance mechanism. The PR genes, TFs, and prominent protein families were identified in chili pepper leaves exposed to P. capsici infections. Moreover, the key defense-related genes identified against P. capsici infection in the current study needs further functional validation by employing advanced genetic tools such as VIGS, RNAi, and CRISPR/Cas9 system. The current findings could be used as invaluable information about the differentially expressed genes associated with defense response against P. capsici infection in chili pepper, helping breeders for selection of landraces containing defense-related genes and also could help for the development of transgenic C. annuum accessions having agronomically essential traits like disease resistance and high yield.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization [Tilahun Rabuma, Vinod Chhoakr]; Methodology: [Vinod Chhokar, Tilahun Rabuma, Om Prakash Gupta]; Formal analysis and investigation: [Vinod Chhokar, Tilahun Rabuma]; Writing—original draft preparation: [Tilahun Rabuma, Manju Yadav]; Writing—review and editing: [Vinod Chhokar, Om Prakash Gupta]; Resources: [Vinod Chhokar]; Supervision: [Vinod Chhokar].

Funding

The Ministry of Education, Ethiopia (former Ministry of Science and Higher Education) has supported through sponsoring a fellowship program for T.R during the research work. Besides, Department Bio & Nanotechnology, Guru Jambheshwar University of Science and Technology, has provided all necessary laboratory facilities for experimentation, analysis and validation.

Availability of data and materials

The transcriptome sequencing data for the four-leaf samples with BioProject ID PRJNA665332, BioSample accession number RC: SAMN16251280; RI: SAMN16251798; SC: SAMN16251797; SI: SAMN16251799 were archived on SRA at the link https://www.ncbi.nlm.nih.gov/biosample/16251797.

Declarations

Conflict of interest

The authors declare no conflict of interest relating to financial interests or personal relationship that could have appeared to influence the work reported in this paper.

Ethics approval and consent to participate

This research did not involve the use of any animal or human data or tissue.

Consent for publication

All authors have seen, read the final version of the material and agreed to be published.