Deep RNA-Seq profile reveals biodiversity, plant–microbe interactions and a large family of NBS-LRR resistance genes in walnut (Juglans regia) tissues

RNA-Seq

In silico analysis

The NCBI database (http://www.ncbi.nlm.nih.gov/taxonomy) provides several resources for the ‘curated classification and nomenclature of all of organisms in the public sequence databases. This currently represents about 10 % of the described species of life on the planet.’ There were ~111 k transcripts. ~5 k did not align to the walnut genome, and were removed (Chakraborty et al. 2015). Of these, ~4 k transcripts had significant homology to E. coli genomes. The remaining ~1 k transcripts were the subject of analysis in the current manuscript, under the assumption that they were derived from extraneous organisms, pathogens or commensal, inhabiting the twenty different tissues. The species names were derived from the best BLAST match to the ‘nt’ database. A bitscore (BLASTSCORE) cutoff of 150 was used (~E-value = 1E − 33). The numerical identifier was obtained from the species name using the site http://www.ncbi.nlm.nih.gov/Taxonomy/TaxIdentifier/tax identifier.cgi. For example, Arthrobacter has the tax ID 1663. These numerical IDs were then used to obtain the complete lineage. The first classification of all organisms was into Eukaryota or Bacteria. We used the second classification field to cluster the organisms discussed here. The expression counts are not normalized since we do not make any inferences on the absolute or relative abundance of the transcripts.

The iterative gene finding method described in YeATS (Chakraborty et al. 2015) was used to identify the homologous set of nucleotide-binding site (NBS)-leucine-rich repeat (LRR) class of genes. A BLAST bitscore of 100 (E value ~1E − 20), with an increment of 20 for each iteration, was used as the homology threshold. The increment in each of the iterations ensures that the resultant proteins do not diverge far from the initially chosen protein. We have used 600 as a lower threshold for the length of NBS-LRR proteins. Additionally, we exclude transcripts with  % of leucine less than the 10 % frequency of leucine residues seen in plant proteomes. These transcripts are probably fragments which have not been assembled by Trinity (Chakraborty et al. 2015).

Phytophthora: causal agent of potato blight

The oomycete Phytophthora is a pathogen responsible for destructive diseases in a wide variety of crop plants, including tomato, potato (Nowicki et al. 2012) and walnut (Belisario et al. 2012) (Fig. 1). Although the presence of a pathogen from the Phytophthora genus is almost certain, it is not possible to determine the exact strain of this pathogen. GAPDH is involved in gycolysis, and other non-metabolic processes (Tarze et al. 2007), and is a well-known housekeeping gene (Eisenberg and Levanon 2013). The Phytophthora GAPDH also shares a 70 % identity with the GAPDH in human placenta (Jenkins and Tanner 2006).

Cryptococcus: Causal agent of cryptococcosis in human

These fungi are mostly localized in the catkins and the vegetative bud in walnut, corroborating previous results about their sexual cycle (Xue et al. 2007). Cryptococcosis is a disease of the respiratory system in human and animals, caused by Cryptococcus neoformans and C. gattii, and exacerbated in patients infected with the AIDS virus (Mitchell and Perfect 1995). Plants are known to host a large number of commensal fungi (Schmit and Mueller 2007). An interesting ecological experiment demonstrated that the plant surface is a conducive environment, stimulating the sexual cycle of Cryptococcus (Xue et al. 2007). Myo-inositol and the plant growth hormone IAA synergistically were proved as ‘strong aphrodisiacs’ (Xue et al. 2007). Two homologous cyclophilin A genes (KGB74193 and KGB74187) have been shown to influence cell growth, mating and virulence (Wang et al. 2001). A peptidyl-prolyl cis–trans isomerase from Cryptococcus (Ess1), non-homologous to the above two genes, is required only for virulence (Ren et al. 2005). Another pathogenic fungus, Pyrenophora (teres/triticirepentis), and the causal agent of the disease ‘tan spot’ (Liu et al. 2011) have been identified in several tissues (Tables 3c, 4b).

Actinobacteria: nitrogen fixing bacterial diazotrophs

Bacterial mRNA is non-polyadenylated, and most should be excluded by the RNA-Seq library preparation method, but some mRNA invariably leaks through. EndAct are present extensively in the inner tissues of living plants, and are a source of important secondary metabolites related to the defense response, growth and environmental stress (Qin et al. 2011; Palaniyandi et al. 2013). The significant homology of the putative proteins from these transcripts after BLAST results highlights the ability of the current methodology to detect a genus with fair precision. A clear example is the transcript C54818_G4_I1 that encodes a 68 bp long ORF, and matches to a ‘MULTISPECIES: hypothetical protein’ from Streptomyces with 82 % identity. The nitrogen fixing diazotroph, Paenibacillus polymyxa P2b-2R, was found to enhance the growth of the important oilseed crop canola (Puri et al. 2016). EndAct obtained from healthy wheat tissue was shown to ‘prime’ the systemic acquired resistance (SAR) and the jasmonate/ethylene (JA/ET) pathways in Arabidopsis thaliana JA/ET pathways when infected with bacterial pathogen Erwinia carotovora subsp. carotovora or the fungal pathogen Fusarium oxysporum, respectively (Conn et al. 2008). The importance of EndAct in biodiversity was established in a tropical rainforest native plant, which identified a total of 312 Actinobacteria associated with the order Actinomycetales (Qin et al. 2012). Recently, the genome sequence of Arthrobacter koreensis 5J12A, a desiccation-tolerant strain, was obtained (Manzanera et al. 2015). Teicoplanin, an antibiotic working against Gram positive bacteria like methicillin-resistant Staphylococcus aureus and Enterococcus faecalis, was obtained from the fermentation broth of a strain of Actinoplanes teichomyceticus (Jung et al. 2008). Another EndAct (Streptomyces) was shown to produce lipase, β-1-3-glucanase and chitinase (defence related enzymes), and aid plant growth (Gopalakrishnan et al. 2013). Micrococcus sp NII-0909 isolated from the Western ghat forest soil in India had demonstrable ability to enhance soil fertility and promote plant growth (Dastager et al. 2010). EndAct, for example Corynebacteria, can be associated with plant pathogenicity (Vidaver 1982). Also, juglone has been found to have inhibitory effects on some of these nitrogen fixing bacteria (Dawson et al. Dawson and Seymour 1983).

Methylibium petroleiphilum: involved in aerobic biodegradation of methyl tert-butyl ether

Methylibium petroleiphilum is capable of using methyl tert-butyl ether as a sole source of carbon in the root (Fig. 2c) (Nakatsu et al. 2006). Unlike Phytophthora, there are only two transcripts for Methylibium. One transcript is 461 nt long and has a 92 % identity to the M. petroleiphilum PM1 genome (Accession: CP000555.1). The ORF from this transcript has a 79 % identity to part of a protein (Accession: ABM53545.1) from uncultured beta proteobacterium CBNPD1 BAC clone 578, which was obtained in a metagenomic analysis of a freshwater toxic cyanobacteria bloom (Pope and Patel 2008). Although the PCR-generated 75-clone, 16S rRNA gene library had confirmed the presence of Proteobacteria, data here associates this protein with M. petroleiphilum. Interestingly, this transcript or ORF has no match in the new draft sequence of the Methylibium sp. strain T29, a fuel oxygenate-degrading bacterial isolate from Hungary (Szabó et al. 2015). This is explained by the differences observed: ‘unlike M. petroleiphilum PM1 our isolate does not harbor the mega plasmid which carries the genes for MTBE-degradation’ (Szabó et al. 2015).

Acyrthosiphon pisum: the pea aphid pest

While both Methylibium and Phytophthora are mostly localized in the root, thepeaaphid A. pisum, which is a pest of importance in agriculture (Van Emden and Harrington 2007), was found in the leaves (Fig. 2d). One transcript of Acyrthosiphon (C58762 _G1_I1) encodes a 91 amino acid long ORF having a 99 % match to the ribosomal L37 protein (Accession: NP 001129424.1). The low count of this transcript demonstrates the accuracy of the RNA-Seq technology (Table 3d).

Aedes aegypti: yellow fever mosquito

Their detection sites are updated regularly. (https://www.cdph.ca.gov/HEALTHINFO/DISCOND/Pages/Aedes-albopictus-and-Aedes-aegypti-Mosquitoes.aspx). However, their detection method is not known to us.

Nucleotide-binding site (NBS) and leucine-rich repeats (LRR) in walnut

The detection of several well-characterized plant pathogens in the current study raises the question of what innate resistance mechanism in walnut could provide resistance to these virulent agents. While it is possible that these strains of these pathogens are non-virulent, it is equally likely that this plant encodes and transcribes the desired resistance genes and transcripts needed to combat a virulent response from any one of these pathogens. Plants possess two distinct kinds of defence mechanisms—the pathogen-associated molecular patterns (PAMP) mediated immunity (PTI) and effector-triggered immunity encoded by resistance (R) genes (ETI). PTI is analogous to the first line of defense (innate immunity) in vertebrates, which is bypassed or disrupted by pathogen effector molecules that are used to downregulate PTI, making the cell vulnerable to pathogen attack (Nicaise et al. 2009). R genes have evolved in this ensuing “evolutionary warfare” in plants, akin to the mammalian adaptive immunity, to recognize pathogens which contain complementary avirulence genes (DeYoung and Innes 2006). However, unlike the mammalian adaptive immune system which is enforced through specialized cells, R genes are active in all plant cells. The majority of R genes encode proteins comprise of a nucleotide-binding site (NBS) and leucine-rich repeats (LRRs). NBS-LRR proteins recognize and neutralize specialized pathogen avirulence (Avr) proteins, leading to the upregulation of PTI, thus, providing plants with resistance to the attack (Hayashi et al. 2010; Ernst et al. 2002; Borhan et al. 2004; Zhang et al. 2010). It has been hypothesized that the distribution and diversity of NBS-LRR sequences is a direct consequence of extensive duplication and random rearrangements, endowing plants with the ability to recognize diverse molecules arising from dynamically changing biotic challenges (Meyers et al. 2003). Here, we briefly describe the transcripts and expression levels of the NBS-LRRs genes in walnut. Two specific examples of such NBS-LRRs genes conferring resistance to plants are the blast-resistance gene Pb1 NBS-LRR from rice (Uniprot: E3WF10) (Hayashi et al. 2010) and the cyst nematode resistance gene from tomato (Uniprot: Q8GT46) (Ernst et al. 2002). In addition to being leucine rich, these two proteins are also abundant with the negatively charged glutamic acid (Fig. 4a). Although these specific NBS-LRRs are ~1300 amino acid long, the typical length of NBS-LRR varies from a few hundred to <2 K (McHale et al. 2006). Using these proteins as initial search entities, we have identified ~400 NBS-LRR transcripts (excluding splice variants denoted by transcripts having the same prefix) in walnut through the ‘findgene’ algorithm described earlier by us (Chakraborty et al. 2015) (Fig. 4b). This is in excellent agreement with the 374 NBS-LRR genes that were identified in a genome wide study of Chinese chestnut (Castanea mollissima) resistant to Chestnut Blight Disease (Zhong et al. 2015). The tissue-specific expression pattern allows the discrimination of the truly critical genes in this large family (Table 7). The tissue-specific nature of certain genes is exemplified by C54426_G7_I1, which has significantly higher expression in the catkins and hull, and shares 78 % identity with a Strubbelig-receptor family (SRF) from Malus domestica. SRFs are receptor-like kinases (Eyüboglu et al. 2007), and are involved in tissue morphogenesis (Vaddepalli et al. 2011) and immune response (Alcázar et al. 2010). As corroboration, we chose one NBS-LRR protein from each of the two major domains of NBS-LRR (McHale et al. 2006)—TIR-NBS-LRR (Uniprot:Q6QX58 (Borhan et al. 2004)) and CC-NBS-LRR Uniprot:Q56YM8 (Meyers et al. 2003), and obtained the same number of transcripts encoded by NBS-LRR genes. Thus, we demonstrate that transcriptomic data that has revealed the biodiversity in different tissues of walnut simultaneously provides insights into the ability of the plant to negate the threat posed by some of these potentially destructive pathogens.

Transcript	CE	CI	CK	EM	FL	HC	HL	HP	HU	IF	LE	LM	LY	PK	PL	PT	RT	SE	TZ	VB	L
C54426_G7_I1	9.9	19.7	1.2	0	2.8	26.2	3.5	23	0.2	1.2	0.9	0.7	3.2	2.8	0.5	1.4	4.5	0.2	1.2	1	745
C50016_G1_I3	0.9	2.2	2.6	1.3	8	2.3	18.9	1.5	1.5	9	8.1	0.4	7.6	0.6	0.7	8.6	2.4	1.9	0.2	3.3	1026
C52820_G1_I1	1.7	4.6	3.8	2.4	14.3	2.5	1 0	2.8	0.6	7.9	2.6	0.4	3.1	2.4	1.4	12.3	1.4	4.5	1.1	3.1	1010
C8180_G1_I1	1.3	2.8	0.8	0.4	8.1	3	25.2	1.5	5	13.9	2.4	0	3.8	0.1	0.3	17.2	1.6	0.5	0	2	668
C55004_G7_I2	2	4.4	2.2	2.5	9.2	3.9	11.9	3.3	3.1	5.6	7.3	3.1	8.3	3.2	1.5	6.2	3.6	2	1.7	3.9	767
C49942_G2_I1	2.7	6.7	2.2	3	1.3	14.8	7	4.8	7.4	2.9	1.3	2.7	1.3	6.5	5.1	4.2	3.2	2.8	5.5	0.8	669
C47067_G1_I2	1.1	2.6	1.1	0.3	8.2	2	7.5	1.3	1.4	5.9	6.4	1	9	4.4	7	11.2	2.4	1.3	1.4	3.6	677
C44186_G1_I3	1.7	2.5	4.5	0.5	5.7	2	15.2	1.2	6	6.4	7.9	2.1	7.2	1	0.7	9.3	5.9	0.9	0.1	4.3	1034
C44186_G1_I1	1.9	2.6	4.6	0.6	5.4	2.1	14.8	1.5	6.8	6.6	8.1	2.2	7.3	1.2	0.8	9.1	5.4	0.9	0.1	4.1	659
C51189_G5_I1	5	10.3	4.1	3.5	16.5	3.6	19.1	3	21.3	10.6	9.5	1.5	10.8	6.5	2.7	14.9	5.9	6.5	2.7	6	1017

These are raw counts in K (10³), and are not normalized. Some NBS-LRR’s (like C54426_G7 _I1, a Strubbelig receptor kinase) are significantly overexpressed in specific tissues

In summary, high conservation of some proteins within a genus does not allow the proper characterization of the species. Thus, although we can state with a great degree of certainty the presence of the genus Phytophthora, it is not possible to identify the exact species/strain. No viruses have been detected using the current methodology. Also, since the root samples were derived from a sterile sample, we did not detect root lesion nematodes (Pratylenchus vulnus), a major source of concern for the California walnut industry (Walawage et al. 2013). The detection of specific proteins from pathogens can serve as a target for therapeutics. The methodology described here presents an unbiased rapid tool to extract the metagenome from an RNA-Seq profile that can be used to develop diagnostics. In this study, the profile represented twenty different tissues from walnut, and the extracted metagenome from all of these tissue types presents a vivid picture of the biodiversity in its surroundings in California.