Hidden route to salicylic acid has finally been uncovered

Main text

Salicylic acid (SA) is a small phenolic compound produced by plants that plays a central role as a signaling molecule in plant immunity (Peng et al., 2021). Since the first report on the involvement of SA in plant defense nearly 50 years ago (White, 1979), our understanding of its biosynthesis has gradually advanced. In the model plant Arabidopsis thaliana, studies have found that the isochorismate synthase (ICS) pathway accounts for more than 90% of SA production during pathogen infection (Figure 1) (Rekhter et al., 2019). However, this pathway is unlikely to be a common route for SA biosynthesis in plants, as orthologs of its key enzyme PBS3 (AvrPphB Susceptible 3) and the corresponding reaction products are absent outside the Brassicaceae family (Holland et al., 2019; Katagiri et al., 2025).

Figure 1.

Comparison of salicylic acid (SA) biosynthetic pathways in different plant species

In Brassicaceae, such as Arabidopsis thaliana (left), SA is primarily synthesized via the isochorismate synthase (ICS) pathway, shown with thick black arrows. Plastid-localized ICS1 converts chorismate into isochorismate, which is exported to the cytosol by the transporter EDS5 (Enhanced Disease Susceptibility 5). The amidotransferase PBS3 (AvrPphB Susceptible 3) conjugates glutamate (Glu) to isochorismate, forming isochorismate-9-Glu (IC-9-Glu). The spontaneous breakdown of IC-9-Glu yields SA, a process enhanced by EPS1 (Enhanced Pseudomonas Susceptibility 1). A small fraction of SA may also be formed through the direct chemical decomposition of isochorismate to SA. In Nicotiana benthamiana, rice (Oryza sativa), and likely most other seed plants outside the Brassicaceae family (right), SA is primarily synthesized via a newly identified Phe-derived pathway, indicated by thick arrows. This pathway begins with the PAL-catalyzed conversion of Phe to t-CA. Subsequent β-oxidation in the peroxisome converts t-CA to benzoyl-coenzyme A (benzoyl-CoA) through several steps, shown in gray. SA is then synthesized through three key steps from benzoyl-CoA, highlighted in red: (1) BEBT (benzoyl-CoA:benzyl alcohol benzoyltransferase)/OSD2 (O. sativa SA-deficient 2) catalyzes the formation of benzyl benzoate from benzyl alcohol and benzoyl-CoA in the peroxisome; (2) endoplasmic reticulum (ER)-associated BBO (benzyl benzoate oxidase)/benzyl benzoate hydroxylase (BBH)/OSD3 hydroxylates benzyl benzoate to benzyl salicylate; and (3) BSH (benzyl salicylate hydrolase)/benzyl salicylate esterase (BSE)/OSD4 converts benzyl salicylate to SA in the cytosol. Additional β-oxidation enzymes include CNL (cinnamoyl-CoA ligase)/OSD1, CHD (cinnamoyl-CoA hydratase/dehydrogenase), and KAT (ketoacyl-CoA thiolase). As noted above, a small fraction of SA may be formed by the direct chemical decomposition of ICS-derived isochorismate to SA.

Besides the well-characterized ICS pathway, SA can also be synthesized from phenylalanine (Phe), as first reported in seed plants based on isotope-labeling experiments (El-Basyouni et al., 1964). In this pathway, Phe ammonia-lyase (PAL) catalyzes the deamination of Phe to trans-cinnamic acid (t-CA), which is then converted to benzoic acid (BA) by β-oxidation enzymes such as AIM1 (Abnormal Inflorescence Meristem 1). SA may then be formed through hydroxylation of BA by a putative benzoic acid 2-hydroxylase (BA2H) (León et al., 1995). However, the gene encoding this BA2H has not yet been identified in plants. Notably, key enzymes involved in β-oxidation, such as AIM1 and CNL (cinnamoyl-CoA ligase), are required for SA production in plants (Peng et al., 2021; Kotera et al., 2023). Therefore, the Phe-derived pathway may represent a common route for SA biosynthesis in plants, although several catalytic steps remain unresolved. Recent studies by Liu et al. (2025), Zhu et al. (2025), and Wang et al. (2025) revealed that SA can be synthesized from benzoyl-coenzyme A (benzoyl-CoA), a product of β-oxidation, through three unexpected enzymatic steps. This newly identified reaction sequence is highly conserved across seed plants and establishes an evolutionarily conserved pathway for SA biosynthesis.

Using RNA sequencing (RNA-seq) data from Nicotiana benthamiana (N. benthamiana) following infection by a virulent bacterial pathogen, Liu et al. observed upregulation of PAL genes, whereas the two ICS genes were downregulated. RNAi-mediated silencing of PAL genes further confirmed that the PAL pathway is important for SA biosynthesis. Additional SA biosynthetic genes were identified through a forward genetic screen in N. benthamiana using ethyl methanesulfonate (EMS)-induced mutagenesis. This screen yielded two independent mutants of a gene encoding an enzyme with high similarity to benzoyl-CoA:benzyl alcohol benzoyltransferase (BEBT). Targeted disruption of this gene using CRISPR-Cas9 resulted in drastically reduced SA levels following pathogen infection, confirming its essential role in SA production. Zhu et al. and Wang et al. identified the same enzyme through gene co-expression analysis with the β-oxidation enzyme CNL in rice (Oryza sativa). As Zhu et al. identified OsCNL (OSD1) in an O. sativa SA-deficient EMS screen, the related genes were named OSD genes. Biochemical assays using recombinant proteins expressed in E. coli or total protein extracts from plants demonstrated that NbBEBT/OsBEBT/OSD2 catalyzes the conjugation of benzoyl-CoA with benzyl alcohol to form benzyl benzoate in vitro. A drastic reduction in benzyl benzoate accumulation observed in Nbbebt/Osbebt/osd2 mutants across all three studies supports its role as a potential intermediate in SA biosynthesis.

Next, Liu et al., Zhu et al., and Wang et al. identified P450 monooxygenases through analysis of pathogen-induced RNA-seq data or gene co-expression. Knockout of the P450 genes led to a significant reduction in SA levels, confirming their involvement in SA biosynthesis. However, these enzymes do not function as BA2Hs; microsomes prepared from yeast expressing N. benthamiana or rice P450s only converted benzyl benzoate to benzyl salicylate. Therefore, the enzymes were designated as benzyl benzoate oxidases (BBOs) or benzyl benzoate hydroxylases (BBHs), represented by OSD3 in rice. Further supporting their functional role, the corresponding mutants accumulated significantly more benzyl benzoate and less benzyl salicylate following pathogen infection in N. benthamiana and under basal conditions in rice. These results indicate that hydroxylation of benzyl benzoate to benzyl salicylate is a key step in SA biosynthesis.

Similarly, the three studies identified benzyl salicylate hydrolase (BSH)/benzyl salicylate esterase (BSE)/OSD4 as the enzyme that catalyzes the final step of converting benzyl salicylate to SA in N. benthamiana and rice. Both recombinant proteins expressed in E. coli and total protein extracts from plants converted benzyl salicylate to SA in vitro. The in planta function of these esterases was further confirmed by the observation that Nbbsh1/2, Osbse, and osd4 mutants exhibited a dramatic reduction in SA levels and a substantial accumulation of benzyl salicylate upon pathogen infection. Furthermore, Wang et al. confirmed the in vivo catalytic activity of OsBBH and OsBSE through feeding experiments in rice mutants using deuterium-labeled benzyl benzoate. Zhu et al. also characterized the detailed β-oxidation steps from t-CA to benzoyl-CoA, and both Zhu et al. and Wang et al. demonstrated the localization of BEBT, BBO, and BSH to the peroxisome, endoplasmic reticulum, and cytosol, respectively.

Close homologs of these three newly identified SA biosynthetic enzymes are widespread among seed plants, except in the Brassicaceae family, in which only the BEBT homolog is retained and SA is primarily synthesized through the ICS pathway. Transient expression of NbBEBT, NbBBO, and NbBSH homologs from willow, poplar, and soybean in the corresponding N. benthamiana mutants restored SA production (Liu et al., 2025), whereas gene silencing of OsBEBT, OsBBH, or OsBSE reduced SA levels in wheat, cotton, and tomato (Wang et al., 2025), supporting the conserved role of these enzymes in SA biosynthesis. Notably, Zhu et al. used comparative genome analysis and a ¹³C₆-Phe feeding experiment to demonstrate that the complete three-enzyme module is found in most seed plants, whereas ¹³C₆-labeled SA was not detected outside seed plant lineages.

In summary, three independent studies have now uncovered the complete and widely conserved SA biosynthetic pathway in seed plants, which proceeds through benzoyl-CoA, benzyl benzoate, and benzyl salicylate as key intermediates following β-oxidation (Figure 1). The identification of these three missing enzymes fills long-standing gaps in our understanding of the Phe-derived route to SA. Notably, these breakthrough studies revealed that this newly identified pathway accounts for approximately 98% of pathogen-induced SA accumulation in N. benthamiana leaves and 99% of basal SA levels in rice. In addition, the EDS5 (Enhanced Disease Susceptibility 5)–PBS3 module of the ICS pathway contributes to about 90% of pathogen-induced SA production in Arabidopsis (Rekhter et al., 2019). However, it is very likely that the remaining SA is also derived from the ICS pathway, because isochorismate can chemically decompose directly into SA (Figure 1) (Rekhter et al., 2019). This exciting discovery opens several avenues for future investigation. Although a few enzymatic steps remain unresolved (the peroxisomal t-CA importer, benzyl benzoate exporter, and benzoyl-CoA reductase), a key next step is to elucidate the regulatory mechanisms of this pathway. For example, is the rate-limiting step the import of t-CA into peroxisomes or the hydrolysis of benzyl salicylate by BSH/BSE/OSD4? Moreover, determining how this pathway is integrated with immune signaling networks throughout the plant, both under basal conditions and during pathogen attack, and whether benzyl salicylate is the elusive mobile signal in systemic acquired resistance (SAR), will be critical for developing a comprehensive understanding of SA-mediated plant immunity.

Funding

The authors apologize for any relevant literature not cited due to space constraints. Financial support from the Deutsche Forschungsgemeinschaft (DFG; GRK 2172 “PRoTECT”) to I.F. and from the Alexander von Humboldt Foundation (AvH; CHN-1236486-HFST-P) to L.T. is gratefully acknowledged.

Acknowledgments

The authors declare no conflict of interest.

Published: September 1, 2025