Heterogeneity of iridoid biosynthesis in catmints: Molecular background in a phylogenetic context

TL;DR

This study explores how different catmint species evolved to produce or lose the ability to make specific iridoid compounds, which are important for plant defense and attraction.

Contribution

The paper identifies evolutionary gains and losses of key biosynthetic genes as the cause of iridoid diversity in Nepeta species.

Findings

Chemotype diversity in Nepeta is partially explained by evolutionary changes in biosynthetic genes.

Independent evolutionary events led to the loss of iridoid synthesis in some species.

The findings enhance understanding of metabolic diversity in the Nepeta genus.

Abstract

Numerous members of the Nepeta genus (family Lamiaceae, subfamily Nepetoideae) are medicinal herbs and sources of important bioactive compounds. Most Nepeta species produce iridoids, which are monoterpenoids that deter herbivores and pathogens and are potential biopesticides. In Nepeta, some species produce iridoid aglycones and glycosylated iridoids (referred to as chemotype A), some produce only glycosylated iridoids (chemotype B), and some produce neither iridoid aglycones nor glycosylated iridoids (chemotype C). Here, we show that the observed diversity in iridoids is, at least partially, attributed to evolutionary gains and losses of key biosynthetic genes. Based on reconstructed phylogenetic relationships, we propose a scenario in which partial or complete loss of the ability to synthesize iridoids with specific stereochemistries in the taxa with chemotypes B and C resulted from independent evolutionary events. These observations improve our understanding of metabolic diversity in the Nepeta genus and may inform efforts to produce specific iridoids in plants.

Evolutionary gains and losses of key biosynthetic genes likely resulting from multiple independent evolutionary events explain why certain Nepeta (catnip) species produce both the active, cat‐attracting nepetalactones and sugar‐bound iridoids, while others make only the sugar‐bound forms, and some have lost iridoid production entirely.

Figures

Major iridoids identified in 10 studied Nepeta taxa (A) Plants were grown under controlled greenhouse conditions and subsequently subjected to metabolic profiling of iridoids. (B) Major iridoid aglycones (IAs) and iridoid glycosides (IGs) detected in Nepeta taxa are presented within the dashed black and red boxes, respectively. Within the IA group, a subgroup of compounds structurally more similar to the IGs is highlighted by the orange dashed box. Numbers in parentheses next to the compound names correspond to those listed in Table 1. Abbreviations: 5,9‐DNL, 5,9‐dehydronepetalactone; 8‐epi‐7‐DLN, epi‐deoxyloganin; DHN, dihydronepetalactone; eDLA, epi‐deoxyloganic acid; eDLNA, epi‐deoxyloganetic acid; Glu, glucose; Hex, hexose; NL, nepetalactone; NLL, nepetalactol; Pent, pentose.

Visualization of untargeted and targeted LC/MS data (A) Heat map of the scaled UHPLC/Orbitrap MS data for 10 phylodiverse Nepeta taxa, with samples and compounds arranged according to hierarchical cluster analysis (HCA), using Spearman's method of cluster agglomeration. Data are visualized using Morpheus software (https://software.broadinstitute.org/morpheus). Values are scaled between minimum and maximum, for each row independently, as indicated by the color scale. (B) Network map of UHPLC/Orbitrap MS data, constructed based on the Jaccard similarity index and circular algorithm with a 50% edge cutoff. Red and orange dots indicate taxa producing only IGs; blue dots present iridoid‐lacking N. nervosa; and gray and black dots indicate taxa producing both nepetalactone‐type IAs and IGs. (C) Pearson correlations based on UHPLC/Orbitrap MS data for 10 Nepeta taxa. Non‐significant correlations values are crossed (P > 0.05). Numbers on the correlation plot correspond to those in Table S1. Both the network map and the correlation plot are constructed using Past5 software. (D) Bar graph presenting UHPLC/DAD/(±)HESI‐MS2 data for the major iridoids quantified in leaves of the 10 selected Nepeta taxa. Targeted compounds are presented in different colors and shapes, as indicated in the symbol legend. A black star symbol denotes the absence of iridoids in samples. Abbreviations: IAs, iridoid aglycones; IGs, iridoid glycosides.

GC/MS characterization of volatile organic compounds (VOCs) in methanol extracts of 10 studied Nepeta taxa (A) Representative TIC GC/MS chromatograms of the 10 Nepeta taxa. Abbreviations: 1, trans,trans‐nepetalactone; 2, cis,trans‐nepetalactone; 3, trans,cis‐nepetalactone; 4, dehydronepetalactone 2; 5, dehydronepetalactone 3; and 6, 1,5,9‐epi‐deoxyloganic acid. (B) GC/MS data (peak areas) were subjected to principal component analysis (PCA), constructed using Past 5 software (version 4.14; Hammer et al., 2001). The contribution of the variables (metabolites) in the first two PCs is indicated by the corresponding loading plots (green), with numbers of metabolites corresponding to those in Table S2. Black and gray dots indicate taxa producing both nepetalactones and IGs (chemotype A), red dots indicate taxa producing only IGs (chemotype B), while blue dots represent iridoid‐lacking N. nervosa (chemotype C).

Transcriptome mining for iridoid biosynthesis‐related gene candidates and co‐expression analysis (A) Summary of the iridoid biosynthesis‐related genes identified in transcriptomes of selected Nepeta taxa, representing chemotypes A (black), B (red), and C (blue). Plus sign (+), transcript identified; minus sign (−), transcripts not found. (B) Schematic representation of the proposed molecular background of chemotype A producing NLs and IGs, mechanisms responsible for the loss of NL production in chemotype B taxon N. grandiflora, and silencing of the iridoid pathway in chemotype C taxon N. nervosa. (C) Data bars represent the relative content of IAs (gray bars), IGs (red bars), and total iridoids (yellow bars). Values are expressed relative to the maximum values in each column independently. The black star symbol denotes the absence of a specific group of iridoids in samples. Heat map constructed based on expression data for iridoid biosynthesis‐related genes in leaves of five representatives of chemotype A (black dot), one representative of chemotype B (red dot), and one of chemotype C (blue dot). Values are scaled between minimum and maximum, for each row independently, as indicated by the color scale. Gray fields denote absence of gene expression in samples. (D) A PCA plot of the gene expression data was constructed using Pearson's algorithm. Gray dots indicate taxa producing both nepetalactones and IGs (chemotype A), red dots indicate taxa producing only IGs (chemotype B), while blue dots represent iridoid‐lacking N. nervosa (chemotype C). Abbreviations: 1,5,9‐eDLA, 1,5,9‐epi‐deoxyloganic acid; 8HGO, 8‐hydroxygeraniol oxidoreductase; G8H, geraniol 8‐hydroxylase; GES, geraniol synthase; GPPS, geranyl diphosphate synthase; ISY, iridoid synthase; MLPL, major latex protein‐like enzyme; NEPS, nepetalactol‐related short‐chain dehydrogenase; NL, nepetalactone; PRISE, progesterone‐5β‐reductase/iridoid synthase activity displaying enzymes.

Reconstitution of nepetalactone production in leaves of the chemotype B representative N. grandiflora (A) Transient expression of NrNEPS1 from N. rtanjensis and NcNEPS5 from N. cataria in leaves of N. grandiflora using the agroinfiltration method. The target genes were co‐infiltrated with p19, a universal suppressor of RNA silencing. (B) Metabolites from agroinfiltrated leaves were extracted with methanol and analyzed by GC/MS. Representative total ion chromatograms (TICs) of leaf samples for the combinations NrNEPS1 + p19 (blue line), NcNEPS5 + p19 (purple line), and p19 (green line) are presented, along with the GC/MS spectra of the cis,trans‐NL standard and the peak corresponding to cis,trans‐NL identified in NrNEPS1 + p19 leaf samples. Additionally, the segment from Rt = 10.07 min to 10.13 min of the extracted ion chromatogram (EIC) for the mass of 166 m/z, corresponding to NL, is presented. The peak corresponding to cis,trans‐NL is visible only in the sample of NrNEPS1 + p19 agroinfiltrated leaves (blue line). (C) UHPLC/DAD/(±)HESI−MS2 quantitative analysis targeted cis,trans‐NL and the two major iridoid glycosides (IGs) in N. sibirica: 1,5,9‐eDLA and nepetaside. Results are expressed as means ± SE. Different combinations of constructs are represented as follows: purple bar − NrNEPS1 + p19, white bars with blue diagonal lines − NcNEPS5 + p19, and green bars − p19. Asterisks indicate significantly different values according to the t‐test (*P < 0.05). A red star indicates the absence of nepetalactone in samples. Abbreviations: 1,5,9‐eDLA, 1,5,9‐epi‐deoxyloganic acid; cis,trans‐NL, cis,trans‐nepetalactone; Nc, Nepeta cataria; NEPS, nepetalactol‐related short‐chain dehydrogenase; Nr, Nepeta rtanjensis; p19, tombusviral silencing suppressor protein.

Phylogenetic tree of the genus Nepeta illustrating the distribution of major groups of iridoids Phylogenetic relationships were reconstructed based on the nuclear ITS regions of the 10 Nepeta taxa analyzed in this study, along with publicly available ITS sequences of other Nepeta taxa and additional representatives of the subfamily Nepetoideae (Lamiaceae). Representatives of the subfamilies Lamioideae and Ajugoideae are also included. Catharanthus roseus (family Apocynaceae) is used as the outgroup. Using literature data (Table S10) and the results of this study, the distribution of NLs and IGs was mapped across the phylogenetic tree, enabling robust chemotype assignment. The presence or absence of NL‐type iridoids and IGs in the listed taxa, or the complete absence of literature data, is indicated as explained in the legend. Red dots indicate taxa producing only IGs (chemotype B), a blue dot represents iridoid‐lacking taxa (chemotype C), while gray dots indicate taxa producing both nepetalactones and IGs (chemotype A).

Phylogeny of the genus Nepeta based on plastid DNA regions, with comparisons to phylogenetic relationships inferred from ITS markers (A) Phylogenetic tree constructed from three plastid DNA regions (trnL–F, rbcL, and matK). Black and gray dots indicate taxa producing both nepetalactones and IGs (chemotype A); red dots indicate taxa producing only IGs (chemotype B); and blue dots represent iridoid‐lacking N. nervosa (chemotype C). (B) Comparison of the phylogenetic trees obtained using ITS (presented in Figure 5, here without NCBI sequences) and plastid markers (presented in Figure 6A). This framework is used to map and interpret the distribution of nepetalactones and iridoid glycosides (IGs) among the Nepeta taxa assessed in this study. Gray rectangles indicate the presence (filled) or absence (empty) of nepetalactones (NLs), while red rectangles indicate the presence (filled) or absence (empty) of iridoid glycosides (IGs).

Proposed metabolic network of major iridoids in leaves of analyzed Nepeta taxa (A) Simplified scheme of the iridoid biosynthetic pathway in Nepeta species up to the major iridoid aglycones, nepetalactones, and iridoid glycosides. (B) Steps of the pathway preceding the formation of cis,trans‐NL, involving the reduction step mediated by iridoid synthase (ISY), the cyclization step (NEPS2–3, and MLPL1), and oxidation via NEPS1. (C) Summarized functions of NEPS isoforms, ISYs, and MLPL1, as proposed by Sherden et al. (2018) and Lichman et al. (2020). (D) Proposed mechanism for the conversion of 8‐oxogeranial into cis–cis oriented 1,5,9‐eDLA, involving a cyclization step catalyzed by NEPS3 (proposed cis–cis cyclase), as well as the involvement of iridoid oxidase (IO) and glucosyltransferase (UGT). Abbreviations: 1,5,9‐eDLA, 1,5,9‐epi‐deoxyloganic acid; 5,9‐DNL, 5,9‐dehydronepetalactone; 8CE, 8‐oxocitronellyl enolate; 8‐epi‐7‐DLN, 8‐epi‐7‐deoxyloganin; 8HG, 8‐hydroxygeraniol; 8HGO, 8‐hydroxygeraniol oxidoreductase; 8OG, 8‐oxogeranial; 11‐O‐Hex‐1,5,9‐eDLA, 11‐O‐hexosyl‐1,5,9‐epi‐deoxyloganic acid; 11‐O‐Pent‐1,5,9‐eDLA, 11‐O‐pentosyl‐1,5,9‐epi‐deoxyloganic acid; CYP450, cytochrome P450 monooxygenase; DHN, dihydronepetalactone; DMAPP, dimethylallyl pyrophosphate; eDLNA, epi‐deoxyloganetic acid; G8H, geraniol 8‐hydroxylase; GE, geraniol; GES, geraniol synthase; GPP, geranyl pyrophosphate; GPPS, geranyl diphosphate synthase; IPP, isopentenyl pyrophosphate; ISY, iridoid synthase; MLPL, major latex protein‐like enzyme; NEPS, nepetalactol‐related short‐chain dehydrogenase; NL, nepetalactone; NLL, nepetalactol; OMT, O‐methyltransferase; PRISE/ISY, progesterone‐5β‐reductase/iridoid synthase activity displaying enzymes; UGT, UDP‐glycosyltransferase; UPT, UDP‐pentosyltransferase.

Tables

Table 1. Iridoid compounds identified in methanol extracts of leaves of 10 Nepeta taxa using the UHPLC/Orbitrap MS technique.

No	Compound name	t R , min	Molecular formula, [M − H]−/[M + H]+	Calculated mass, m/z	Exact mass, m/z	Δ mDa	MS2 fragments (% base peak)	1	2	3	4	5	6	7	8	9	10
Iridoid glycosides
1	11‐O‐Hexosyl‐epi‐deoxyloganic acid	4.49	C22H33O14 −	521.18758	521.18759	−0.01	135.08162(53), 153.09215(100), 161.04561(6), 197.08200(53), 341.12369(8), 359.13474(5)	+	+	+	+	t	+	+	+	/	+
2	6β‐Dihydrocornic acid	4.49	C16H23O10 −	375.12967	375.12949	0.18	89.02441(80), 101.02437(23), 113.02439(20), 119.03488(24), 123.08160(87), 167.07144(100)	/	/	/	/	/	+	/	/	/	/
3	Adenosmoside	4.51	C16H27O9 −	363.16606	363.16543	0.62	59.01385(100), 89.02442(82), 101.02446(49), 113.02458(31), 119.03500(40), 183.10281(12)	/	/	/	/	/	+	+	/	/	/
4	Nepetanudoside	4.51	C17H25O10 −	389.14532	389.14538	−0.06	101.02443(100), 183.06645(4), 227.09236(13)	/	+	/	+	/	+	+	+	/	/
5	11‐O‐Pentosyl‐epi‐deoxyloganic acid	5.00	C21H31O13 −	491.17702	491.17719	−0.18	101.02440(8), 109.06594(13), 135.08162(59), 153.09212(100), 197.08197(48)	+	+	+	+	/	+	+	+	+	/
6	1,5,9‐epi‐Deoxyloganic acid	5.06	C16H23O9 −	359.13476	359.13481	−0.05	101.02454(4), 119.03487(4), 135.08153(94), 153.09209(100), 197.08188(32)	+	+	+	+	t	+	+	+	+	+
7	Acetyl‐nepetanudoside	5.12	C19H27O11 −	431.15589	431.15570	0.18	101.02442(100), 227.09267(11)	/	/	/	+	/	/	/	/	+	/
8	Shikimoyl‐1,5,9‐epi‐deoxyloganic acid	5.16	C23H31O13 −	515.17702	515.17740	−0.38	135.08162(51), 137.02492(9), 153.09210(100), 173.04575(12), 197.08199(53), 359.13599(9)	/	/	/	/	/	+	+	/	/	+
9	5‐deoxylamiol	5.29	C16H25O9 −	361.15041	361.15036	0.04	137.09740(13), 155.10789(100), 181.08723(5), 199.09775(36)	/	+	/	/	/	/	/	/	+	/
10	Caffeoyl‐malonyl‐1,5,9‐epi‐deoxyloganic acid	5.33	C28H31O15 −	607.16684	607.16730	−0.46	135.04507(57), 161.02446(89), 179.03487(100), 197.08191(20), 359.13583(16), 445.13431(40)	+	/	/	/	/	+	/	/	/	/
11	Nepetaracemoside A	5.43	C16H23O8 −	343.13984	343.13990	−0.06	109.06596(14), 135.08151(86), 153.09206(100), 197.08185(33)	+	+	+	+	t	+	+	+	/	+
12	Nepetaside	5.53	C16H25O7 −	345.15549	345.15549	0.00	59.01384(100), 101.02441(47), 113.02444(29), 119.03500(35), 183.10266(34), 345.15512(17)	+	+	+	+	t	+	+	+	+	+
13	Nepetariaside	5.79	C16H27O8 −	347.17114	347.17115	−0.01	59.01387(100), 101.02446(50), 119.03504(40), 167.10783(15), 185.11850(5), 347.17105(38)	+	+	+	+	t	+	+	+	+	+
Iridoid aglycones
14	Nepetalactol	4.44	C10H17O2 +	169.12231	169.12115	1.16	81.06925(99), 107.08475(19), 121.10017(14), 123.11588(8), 133.10017(18), 151.07426(100)	/	+	+	+	/	+	/	/	/	/
15	Deoxyloganetic acid	4.63	C10H15O4 +	199.09649	199.09516	1.32	135.07944(100), 137.09474(24), 151.07417(17), 153.08995(48), 163.07411(86), 181.08456(51)	+	+	+	+	/	+	+	+	+	+
16	Nepetaracemoside B aglycone	5.15	C10H13O3 +	181.08592	181.08412	1.80	93.06895(18), 95.04823(10), 135.07918(32), 145.02710(12), 163.07368(100), 181.08417(17)	+	+	+	+	/	+	+	+	+	+
17	Nepetalactol	5.40	C10H17O2 +	169.12231	169.12110	1.20	81.06924(16), 93.06915(41), 95.08480(26), 123.11598(16), 133.10023(15), 151.11066(100)	+	+	/	/	/	+	+	+	+	+
18	Dihydronepetalactone 2	5.57	C10H17O2 +	169.12231	169.12118	1.13	81.06921(33), 95.08475(10), 123.11583(100)	+	+	/	/	/	+	+	+	+	+
19	Deoxygeniposide aglycone	6.07	C11H15O4 +	211.09649	211.09504	1.45	105.06906(30), 133.06372(47), 151.07411(100), 165.08971(35), 179.06868(28), 193.08452(13)	/	/	/	/	/	+	+	+	+	+
20	8‐epi‐7‐Deoxyloganetin	6.12	C11H17O4 +	213.11214	213.11051	1.62	121.06378(41), 125.09506(13), 149.05856(100), 153.08980(75), 181.08482(11)	/	/	/	/	/	/	+	/	/	/
21	Dihydronepetalactone 3	6.21	C10H17O2 +	169.12231	169.12106	1.24	81.06922(23), 109.10023(4), 123.11583(100)	+	+	/	/	/	/	+	+	+	+
22	Nepetalic acid	6.22	C10H15O3 −	183.10267	183.10270	−0.03	139.11287(23), 155.10771(3), 165.09222(10), 183.10271(100)	+	+	+	+	t	+	+	+	+	+
23	5,9‐dehydronepetalactone	6.84	C10H13O2 +	165.09101	165.08986	1.15	165.08980(100)	+	/	/	/	/	+	+	+	+	+
24	t,t‐nepetalactone	7.11	C10H15O2 +	167.10666	167.10524	1.42	81.06918(9), 93.06911(31), 121.10018(100), 123.07944(8), 125.09511(20), 139.11070(35)	+	+	/	/	/	+	/	/	+	+
25	c,t‐nepetalactone	7.44	C10H15O2 +	167.10666	167.10532	1.34	81.06924(12), 93.06915(32), 121.10019(100), 123.07942(8), 125.09509(21), 139.11067(34)	+	+	/	/	/	+	+	+	+	+
26	t,c‐nepetalactone	7.93	C10H15O2 +	167.10666	167.10539	1.27	81.06921(100), 93.06909(11), 121.10023(32), 123.11585(27), 125.05864(26), 139.11082(10)	/	/	/	/	/	/	+	+	/	+

Full text

INTRODUCTION

The genus Nepeta L. (catmints, catnips) contains 200–300 species that commonly inhabit hilly and mountainous areas from the Middle East to the western Himalayas, extending west to Macaronesia and as far east as the Yakutsk region in Siberia (Plants of the World Online | Kew Science, [Link], 2024). Although reported to be monophyletic (Jamzad et al., 2003), Nepeta is a heterogeneous genus. As suggested by Mehregan et al. (2022) and Rose et al. (2023), its circumscription complicates the development of a higher level taxonomy within the subtribe Nepetinae of the subfamily Nepetoideae (family Lamiaceae). Notably, plants in the genus Nepeta are the sole producers of the iridoid monoterpenoids among the Nepetoideae (Duplais et al., 2020). These terpenoids have a wide range of roles, facilitating complex interactions with insects (in plant–pollinator and plant–herbivore relationships), phytopathogens (including antimicrobial activity), and neighboring plants (allelopathy and phytotoxicity). Essential oils of catmints and their active components, namely, nepetalactone (NL), dihydronepetalactone (DHN), and nepetalactol (NLL), are iridoid aglycones (IAs) that act as repellents against a wide range of arthropods, and are at least as effective as the industry‐standard repellent N,N‐diethyl‐m‐toluamide (DEET) (Birkett et al., 2011; Gkinis et al., 2014). However, glycosylated iridoids (IGs), which are abundantly synthesized by most Nepeta species, have been relatively neglected and only rarely studied for their biological activities. This is also the case for the most abundant IG in Nepeta species: 1,5,9‐epi‐deoxyloganic acid (1,5,9‐eDLA) (Aničić et al., 2021; Başar et al., 2024).

The ability to produce iridoids is thought to have been lost in the Nepetoideae during evolution due to the loss of iridoid synthase (ISY), a key enzyme from the early iridoid pathway; however, this ability later reappeared in Nepeta (Mint Evolutionary Genomics Consortium, 2018). The re‐establishment of iridoid biosynthesis in members of the genus Nepeta involved biosynthetic enzymes present in all Nepetoideae, in parallel with the convergent evolution of ISY from progesterone 5β‐reductase (P5βR) (Lichman et al., 2020). The emergence of NAD‐dependent nepetalactol‐related short‐chain dehydrogenase/reductase (NEPS) and major latex protein‐like (MLPL) enzymes in Nepeta, along with novel ISYs, has given rise to iridoids with unique stereochemistry that occur exclusively in this group of plants. However, our current understanding of the complex phylogenetic relationships in the genus Nepeta remains uncertain and is widely debated, and existing data on the diversity and distribution of iridoids are incomplete.

Classification of Nepeta species into chemotypes with similar iridoid profiles provides a first step in relating their chemical diversity to their phylogenetic relationships. A comprehensive literature search of qualitative iridoid profiles across multiple Nepeta species identified three putative chemotypes: chemotype A, comprising taxa that produce both iridoid aglycones (IAs) of the nepetalactone‐type and iridoid glycosides (IGs); chemotype B, comprising taxa that produce only IGs; and chemotype C, comprising taxa that do not produce iridoids. Although taxa producing only IAs of the nepetalactone (NL) type (a putative chemotype D) have been observed in the literature, such findings are considered uncertain because those taxa were not analyzed for their IG contents. Currently, data on the distribution and diversity of IAs and IGs throughout the genus Nepeta are fragmentary: We lack information for species, and data on the content of both IAs and IGs in individual Nepeta taxa are rarely available (Aničić et al., 2020, 2021; Palmer et al., 2022; Petrović et al., 2024b). Additionally, previous studies used plant material of different origins, genetic backgrounds, and at various developmental stages, and used a variety of extraction procedures and analytical instruments. Therefore, unambiguous chemotype assignment, and accurate comparative analysis, of Nepeta taxa cannot be definitively interpreted based solely on data from the literature.

In this study, we aimed to generate informative and inclusive metabolomic and transcriptomic data sets for phylogenetically diverse Nepeta taxa belonging to chemotype A, B, or C, and provide tools for comprehensive and simultaneous acquisition, analysis, and adequate interpretation of the overall diversity of iridoids (both IAs and IGs) in a phylogenetic context. This information will help us understand the molecular origins of this enzymatic and chemical diversity and will identify the segments of the iridoid biosynthetic pathway that appear to have been lost in different evolutionary lineages. Providing unambiguous evidence for the existence of chemotypes A, B, and C among the extant Nepeta species will elucidate the evolutionary events that have led to the secondary loss of certain biosynthetic steps in ancestral species. Considering that iridoids are widely distributed throughout the plant kingdom, the experimental strategy described in this study can be applied to other phylogenetically distinct plant groups to provide a crucial entry point for explaining evolutionary convergence in the production of these common compounds.

Here, we quantified the most abundant iridoids in representatives of the three Nepeta chemotypes and analyzed the co‐expression of iridoid‐related biosynthetic genes to elucidate the molecular mechanisms controlling the metabolic flux through the iridoid biosynthetic pathway and its branches leading to IAs or IGs, which helped identify key regulatory points and genes. This knowledge clarifies the roles of ISYs, NEPSs, and MLPLs in determining the heterogeneity of iridoids among Nepeta species, thus facilitating the development of biotechnological approaches for in planta production of targeted iridoids.

RESULTS

Comparative untargeted and targeted metabolomics of 10 phylogenetically diverse Nepeta taxa

To perform comprehensive metabolomics of the 10 Nepeta taxa (Figure 1A), an untargeted liquid chromatography (LC)/Orbitrap mass spectrometry (MS) analysis was adopted, which identified a total of 206 metabolites (data not presented), including 26 iridoid compounds (Figure 1; Table 1). Table 1 lists these 26 identified iridoids, divided into 13 IGs and 13 IAs based on their chemical structures. Table S1 provides the peak areas of the identified iridoids, along with previous records of each compound detected in Nepeta taxa or representatives from other genera in the Lamiaceae family.

*Major iridoids identified in 10 studied

Nepeta

taxa

(A) Plants were grown under controlled greenhouse conditions and subsequently subjected to metabolic profiling of iridoids. (B) Major iridoid aglycones (IAs) and iridoid glycosides (IGs) detected in Nepeta taxa are presented within the dashed black and red boxes, respectively. Within the IA group, a subgroup of compounds structurally more similar to the IGs is highlighted by the orange dashed box. Numbers in parentheses next to the compound names correspond to those listed in Table 1. Abbreviations: 5,9‐DNL, 5,9‐dehydronepetalactone; 8‐epi‐7‐DLN, epi‐deoxyloganin; DHN, dihydronepetalactone; eDLA, epi‐deoxyloganic acid; eDLNA, epi‐deoxyloganetic acid; Glu, glucose; Hex, hexose; NL, nepetalactone; NLL, nepetalactol; Pent, pentose.*

The IAs identified can be provisionally divided into two subgroups: (i) NL‐type iridoids and (ii) IAs structurally resembling the aglycone parts of some IGs. From the first subgroup, we identified two isomers of NLL (Table 1; Figure 1B, compounds 14 and 17) and three stereoisomers of NL: trans,trans‐NL (24), cis,trans‐NL (25), and trans,cis‐NL (26). In addition, we identified 5,9‐dehydronepetalactone (5,9‐DNL, 23) and two isomers of DHN (18 and 21). Except for nerve catmint (N. nervosa), giant catmint (N. grandiflora), and smooth catmint (N. laevigata), which contained no NLs or their derivatives, the other taxa contained substantial amounts of compounds belonging to this group of IAs (Table 1). An NL stereoisomer with a trans,trans‐orientation was dominant in Stewart's catmint (N. stewartiana), but we also identified this compound as a minor compound in the leaves of Nepeta taxa that predominantly accumulated cis,trans‐NL (catmint (N. cataria), Siberian catmint (N. sibirica), naked catmint (N. nuda), and N. ernesti‐mayeri). The leaves of N. rtanjensis and Greek catmint (N. parnassica) were rich in the trans,cis‐NL stereoisomer, with cis,trans‐NL present in significantly lower amounts. Dihydronepetalactone 2 (18) was particularly abundant in the leaves of N. stewartiana and N. sibirica plants (Table S1), whereas the leaves of N. rtanjensis and N. parnassica plants were rich in dihydronepetalactone 3 (21). The compound 5,9‐DNL (23) was the most abundant IA in the leaves of N. rtanjensis and N. parnassica plants, but was absent from those of N. grandiflora, N. laevigata, N. nervosa, and N. ernesti‐mayeri. Nepetalic acid (NA, 22) was most abundant in N. stewartiana (Figure 2A; Table S1).

*Visualization of untargeted and targeted LC/MS data

(A) Heat map of the scaled UHPLC/Orbitrap MS data for 10 phylodiverse Nepeta taxa, with samples and compounds arranged according to hierarchical cluster analysis (HCA), using Spearman's method of cluster agglomeration. Data are visualized using Morpheus software (https://software.broadinstitute.org/morpheus). Values are scaled between minimum and maximum, for each row independently, as indicated by the color scale. (B) Network map of UHPLC/Orbitrap MS data, constructed based on the Jaccard similarity index and circular algorithm with a 50% edge cutoff. Red and orange dots indicate taxa producing only IGs; blue dots present iridoid‐lacking N. nervosa; and gray and black dots indicate taxa producing both nepetalactone‐type IAs and IGs. (C) Pearson correlations based on UHPLC/Orbitrap MS data for 10 Nepeta taxa. Non‐significant correlations values are crossed (P > 0.05). Numbers on the correlation plot correspond to those in Table S1. Both the network map and the correlation plot are constructed using Past5 software. (D) Bar graph presenting UHPLC/DAD/(±)HESI‐MS2 data for the major iridoids quantified in leaves of the 10 selected Nepeta taxa. Targeted compounds are presented in different colors and shapes, as indicated in the symbol legend. A black star symbol denotes the absence of iridoids in samples. Abbreviations: IAs, iridoid aglycones; IGs, iridoid glycosides.*

The second subgroup of IAs included epi‐deoxyloganetic acid (15), nepetaracemoside B aglycone (16), deoxygeniposide aglycone (19), and 8 ‐epi‐7‐deoxyloganetin (20) (Figure 1B; Table 1). We detected epi‐deoxyloganetic acid (15) in all iridoid‐producing Nepeta taxa analyzed in this study, except for N. nervosa. We identified 8‐epi‐7‐deoxyloganetin (20) exclusively in N. parnassica. Nepetaracemoside B aglycone (16) was present in all analyzed Nepeta taxa, with the exception of N. nervosa (Table 1), whereas we detected deoxygeniposide aglycone (19) in the leaves of N. rtanjensis, N. sibirica, N. parnassica, N. nuda, and N. stewartiana.

We identified two main subgroups of IGs in the Nepeta taxa (Figure 1B; Table 1): glycosides with a carboxylic acid (11‐COOH) or a methyl group (11‐CH_3_) at the C‐11 position. In most cases, the most abundant IG of the 11‐COOH type was 1,5,9‐eDLA (6), with nepetanudoside (4) and acetyl‐nepetanudoside (7) also present in high amounts (Table S1). However, we also detected 11‐COOHex (1) and 11‐COOPen variants (5), displaying MS^2^ losses corresponding to hexosyl and pentosyl moieties (−162 and −132 Da, respectively). We identified 11‐O‐hexosyl‐1,5,9‐eDLA (1) in all Nepeta taxa, except for N. sibirica, whereas 11‐O‐pentosyl‐1,5,9‐eDLA (5) was not present in N. nervosa, N. sibirica, or N. stewartiana. Based on an exact mass m/z [M − H]^−^ of 607.16730 and an MS^2^ fragmentation pattern showing the loss of caffeoyl and malonyl moieties [M − H − 162 − 86]^−^, we assigned compound 10 to caffeoyl‐malonyl‐1,5,9‐eDLA in the leaf samples of N. cataria and N. nuda (Table 1). We detected another similar compound in N. parnassica, N. nuda, and N. stewartiana, tentatively named shikimoyl‐1,5,9‐eDLA (8), showing characteristic high‐resolution mass spectrometry (HRMS)/MS fragments for both shikimic acid and 1,5,9‐eDLA.

The subgroup of IGs characterized by a methyl group at position C‐11 includes 5‐deoxylamiol (9), nepetaracemoside A (11), nepetaside (12), and nepetariaside (13) (Table 1). We detected 5‐deoxylamiol (9) in the leaves of N. sibirica and N. ernesti‐mayeri. We also identified nepetaracemoside A (11) in the leaves of plants from all analyzed taxa, except for N. sibirica, whereas nepetariaside (13), with an open dihydropyran ring, was present in all analyzed Nepeta taxa. There were differences in the configurations at the C‐5 and C‐9 positions between compounds 13 and 6. Nepetaside (12) was present in all samples analyzed (Table 1). We also identified two IGs not assigned to any of the above subgroups: adenosmoside (3) in N. parnassica and N. nuda, and 6β‐dihydrocornic acid (2) in the leaves of N. nuda plants (Table 1).

A hierarchical cluster analysis (HCA) distinguished two main clusters, A and B, with the former comprising high NL‐producing taxa (Figure 2A). Cluster A forms two groups: the first group (Aa) consists of N. rtanjensis, N. parnassica, N. cataria, and N. stewartiana, whereas the second group (Ab) contains only N. sibirica. Cluster B included N. nervosa, which lacks iridoids and defines a separate subcluster (Ba); N. grandiflora and N. laevigata, which lack NLs but produce IGs (Bb1); and low‐level NL producers N. nuda and N. ernesti‐mayeri (Bb2). We obtained a similar grouping of these Nepeta taxa in a network map (Figure 2B).

Pearson's correlation coefficient (r) analysis showed that the contents of the most abundant IG compound, 1,5,9‐eDLA (6), are significantly positively correlated with those of compounds 1, 3, 4, 5, 11, 12, 15, 16, 20, and 26 (Figure 2C). We observed the only significant negative correlation between compounds 7 and 25. Among the IAs, the levels of compound 15 were significantly and negatively correlated with those of compounds 7, 24, and 25 (Figure 2C). Although the amounts of compounds 24 and 25 showed a significant positive correlation, they were not significantly correlated with those of compound 26. We detected the strongest positive correlations between the levels of compound 22 and those of the two DHN isomers (compounds 18 and 21), as well as with those of compound 13.

We quantified the levels of nine iridoid compounds that were identified as predominant in the samples analyzed by untargeted metabolomics profiling in leaves from the 10 Nepeta taxa using an ultra‐high‐performance LC (UHPLC)/diode array detector (DAD) coupled with a (±) heated electrospray ionization (HESI) − MS/MS instrument. Representative UHPLC/DAD chromatograms are shown in Figure S1. N. stewartiana samples were characterized by a predominance of trans,trans‐NL, reaching 161 μg 100 mg^−1^ fresh weight (FW), whereas N. sibirica, N. cataria, N. ernesti‐mayeri, and N. nuda samples were characterized by a prevalence of *cis,trans‐*NL (Figure 2D). The leaves of N. nuda plants also contained a substantial amount of 5,9‐DNL (369 μg 100 mg^−1^ FW). The amount of cis,trans‐NL in the leaves of N. sibirica plants was about 463 μg 100 mg^−1^ FW, whereas the leaves of N. rtanjensis and N. parnassica contained about 177 μg 100 mg^−1^ FW and ∼143 μg 100 mg^−1^ FW of trans,cis‐NL, respectively (Figure 2D). The amount of 5,9‐DNL was about 2,057 μg 100 mg^−1^ FW in the leaves of N. rtanjensis and 2,727 μg 100 mg^−1^ FW in N. parnassica. Overall, we consider N. nuda and N. ernesti‐mayeri to be low NL‐producing taxa. For IGs, the amounts of 1,5,9‐eDLA ranged from about 14 μg 100 mg^−1^ FW in N. stewartiana to 211 μg 100 mg^−1^ FW in N. nuda. Both N. stewartiana and N. sibirica are recognized as low IG‐producing taxa in this study. Notably, the low NL producers N. nuda and N. ernesti‐mayeri were rich in 1,5,9‐eDLA (Figure 2D). Not surprisingly, there were no detectable NLs in N. grandiflora and N. laevigata (Figure 2D). The leaves of N. grandiflora contained approximately 33 μg 1,5,9‐eDLA 100 mg^−1^ FW, whereas N. laevigata contained 72 μg 100 mg^−1^ FW. According to our quantitative metabolomics data, N. nervosa contained no iridoids in amounts detectable by the analytical technique used here.

To complement our LC/MS data, we performed qualitative gas chromatography (GC)/MS analysis of the volatile organic compounds (VOCs) present in methanol extracts of leaves from the 10 Nepeta taxa. In total, we identified 46 VOCs, comprising monoterpenes (six monoterpene hydrocarbons and 19 oxygenated monoterpenes), sesquiterpenes (11 sesquiterpene hydrocarbons and four oxygenated sesquiterpenes), diterpenes (three diterpenes and two oxygenated diterpenes), and one nitrogen analog of NL, nepetalactam (Table S2). Representative GC/MS total ion chromatograms (TICs) are shown in Figure 3A.

*GC/MS characterization of volatile organic compounds (VOCs) in methanol extracts of 10 studied

Nepeta

taxa

(A) Representative TIC GC/MS chromatograms of the 10 Nepeta taxa. Abbreviations: 1, trans,trans‐nepetalactone; 2, cis,trans‐nepetalactone; 3, trans,cis‐nepetalactone; 4, dehydronepetalactone 2; 5, dehydronepetalactone 3; and 6, 1,5,9‐epi‐deoxyloganic acid. (B) GC/MS data (peak areas) were subjected to principal component analysis (PCA), constructed using Past 5 software (version 4.14; Hammer et al., 2001). The contribution of the variables (metabolites) in the first two PCs is indicated by the corresponding loading plots (green), with numbers of metabolites corresponding to those in Table S2. Black and gray dots indicate taxa producing both nepetalactones and IGs (chemotype A), red dots indicate taxa producing only IGs (chemotype B), while blue dots represent iridoid‐lacking N. nervosa (chemotype C).*

The qualitative composition of monoterpene hydrocarbons was highly conserved across the analyzed taxa, with *α‐*pinene, *β‐*pinene, sabinene, and o‐cymene universally present. Additionally, we detected the oxygenated monoterpenes 1,8‐cineole and δ‐terpineol in all samples, with 1,8‐cineole being the most abundant monoterpene in most of the samples analyzed. This compound was especially dominant in the leaves of N. ernesti‐mayeri and N. stewartiana plants.

The GC/MS data strongly supported the UHPLC/Orbitrap MS data, and confirmed the absence of IAs in the leaves of N. nervosa, N. grandiflora, and N. laevigata samples (Table S2). When present, we identified the IAs' NLs (trans,trans‐, cis,trans‐, and trans,cis‐stereoisomers) and their derivatives (two isomers of DHN and 5,9‐DNL) as major VOCs in these Nepeta taxa. The exceptions were N. ernesti‐mayeri and N. nuda, which produced IAs as minor constituents. N. rtanjensis and N. parnassica samples predominantly contained trans,cis‐NL. Conversely, the leaves of N. sibirica, N. cataria, N. nuda, and N. ernesti‐mayeri were mainly abundant in cis,trans‐NL. N. stewartiana contained high amounts of trans,trans‐NL and cis,trans‐NL, in addition to trace amounts of trans,cis‐NL. The sesquiterpenes germacrene D and β‐caryophyllene were highly abundant in the samples, especially in taxa lacking IAs (N. grandiflora, N. laevigata, and N. nervosa). Copaene was also highly abundant throughout the samples (Table S2). We detected the diterpene neophytadiene and the oxygenated diterpene phytol, both present as two isomers, in all analyzed samples.

A principal component analysis (PCA) plot, based on relative GC/MS data and using Pearson's algorithm, revealed the diversification of Nepeta taxa into two major groups along the first principal component (PC1), which explained 54.7% of the total variance (Figure 3B). N. rtanjensis and N. parnassica, characterized by predominating trans,cis‐NL, were clearly separated from the other taxa. Nepeta taxa producing only IGs (N. grandiflora and N. laevigata), iridoid‐lacking N. nervosa, and low NL‐producing N. ernesti‐mayeri and N. nuda were distinctly separated along PC2 (31.1%) from the taxa rich in cis,trans‐NL (N. sibirica and N. cataria) and trans,trans‐NL (N. stewartiana). The major contributors to the diversification along PC1 were trans,cis‐NL and 5,9‐DNL, and cis,trans‐NL for PC2.

Metabolomics‐driven chemotype assignment in 10 Nepeta taxa

Based on our comprehensive metabolomics data, we classified the 10 Nepeta taxa into three major chemotypes (A, B, and C) according to the presence or absence of NLs and IGs. Species assigned to chemotype A contained both NLs and IGs and displayed significant heterogeneity in their NL composition and ratios. These species were roughly divided into several sub‐chemotypes based on the stereochemistry of the major NLs in their leaves: N. stewartiana, with trans,trans‐NL as the major stereoisomer (A1); N. sibirica, N. cataria, N. nuda, and N. ernesti‐mayeri characterized by predominant cis,trans‐NL (A2); and N. rtanjensis and N. parnassica, rich in trans,cis‐NL (A3). None of the Nepeta taxa in this study predominantly accumulated cis,cis‐NL (defining a putative sub‐chemotype A4). We assigned N. grandiflora and N. laevigata to chemotype B, as they produced substantial amounts of IGs in their leaves but contained no NL‐type iridoids or their derivatives. Species from chemotype B were further distinguished from chemotype A by the absence of acylated forms of 1,5,9‐eDLA, such as caffeoyl‐malonyl‐1,5,9‐eDLA and shikimoyl‐1,5,9‐eDLA. Chemotype C, which comprises only N. nervosa, was characterized by trace amounts of iridoids in the samples.

Phylotranscriptomics exploration of the molecular mechanisms underlying iridoid heterogeneity

We identified promising candidates for iridoid‐related biosynthetic genes in the transcriptomes of seven phylogenetically and chemically diverse Nepeta taxa based on sequence similarity to annotated biosynthetic genes from other iridoid‐ and alkaloid‐producing plants in relevant public databases, including those of N. cataria and N. racemosa. We detected transcripts in the leaf RNA‐seq data sets from Nepeta species belonging to putative chemotype A for genes encoding all biosynthetic enzymes necessary to produce both NLL and NL, and thus the precursors for IGs (Figure 4A): Geranyl pyrophosphate synthase (GPPS), geraniol synthase (GES), geraniol 8‐hydroxylase (G8H), 8‐hydroxygeraniol oxidoreductase (8HGO), ISY, and NEPSs. Notably, we detected no ISY transcripts in the leaf RNA‐seq data for N. stewartiana, but we detected abundant transcripts for progesterone‐5β‐reductase (PRISE, P5βR), encoding an enzyme with ISY activity. The RNA‐seq data from all taxa lacked full‐length NEPS5 transcripts. The transcriptomes of N. sibirica and N. stewartiana lacked transcripts for NEPS3, NEPS4, and MLPL2, with that of N. stewartiana also missing NEPS2. We detected NEPS4 and MLPL2 transcripts only in the leaf RNA‐seq data for N. rtanjensis and N. grandiflora, respectively. We noticed the absence of NEPS1 and NEPS5 transcripts in N. grandiflora, which we assigned to chemotype B (Figure 4B). However, the leaf RNA‐seq data of N. grandiflora were rich in transcripts for early biosynthetic genes (EBGs) that are involved in pathway steps leading up to NLL production. In the leaf RNA‐seq data for N. nervosa (chemotype C), transcripts encoding homologs of NEPS2, NEPS3, NEPS4, and MLPL2 were missing, although we detected transcripts for genes encoding homologs of NEPS1 and MLPL1 (Figure 4B). ISY and GES transcripts were also absent from these RNA‐seq data (Figure 4A, B).

*Transcriptome mining for iridoid biosynthesis‐related gene candidates and co‐expression analysis

(A) Summary of the iridoid biosynthesis‐related genes identified in transcriptomes of selected Nepeta taxa, representing chemotypes A (black), B (red), and C (blue). Plus sign (+), transcript identified; minus sign (−), transcripts not found. (B) Schematic representation of the proposed molecular background of chemotype A producing NLs and IGs, mechanisms responsible for the loss of NL production in chemotype B taxon N. grandiflora, and silencing of the iridoid pathway in chemotype C taxon N. nervosa. (C) Data bars represent the relative content of IAs (gray bars), IGs (red bars), and total iridoids (yellow bars). Values are expressed relative to the maximum values in each column independently. The black star symbol denotes the absence of a specific group of iridoids in samples. Heat map constructed based on expression data for iridoid biosynthesis‐related genes in leaves of five representatives of chemotype A (black dot), one representative of chemotype B (red dot), and one of chemotype C (blue dot). Values are scaled between minimum and maximum, for each row independently, as indicated by the color scale. Gray fields denote absence of gene expression in samples. (D) A PCA plot of the gene expression data was constructed using Pearson's algorithm. Gray dots indicate taxa producing both nepetalactones and IGs (chemotype A), red dots indicate taxa producing only IGs (chemotype B), while blue dots represent iridoid‐lacking N. nervosa (chemotype C). Abbreviations: 1,5,9‐eDLA, 1,5,9‐epi‐deoxyloganic acid; 8HGO, 8‐hydroxygeraniol oxidoreductase; G8H, geraniol 8‐hydroxylase; GES, geraniol synthase; GPPS, geranyl diphosphate synthase; ISY, iridoid synthase; MLPL, major latex protein‐like enzyme; NEPS, nepetalactol‐related short‐chain dehydrogenase; NL, nepetalactone; PRISE, progesterone‐5β‐reductase/iridoid synthase activity displaying enzymes.*

We used the nucleotide sequences of the identified GPPS, GES, G8H, 8HGO, ISY/PRISE, NEPS1–4, and MLPL1–2 candidates to generate pairwise sequence identity matrices (Tables S3–S9). The GPPS sequences from N. nervosa were highly similar to those of N. sibirica and N. stewartiana (96.2%). We also observed high sequence similarity among GPPS sequences from N. rtanjensis, N. ernesti‐mayeri, N. nuda, and N. grandiflora (Table S3). NemGES, NrGES, NnuGES, and NgGES were very similar to both NcGESA and NcGESB previously reported for N. cataria, as well as to NmGES from N. racemosa (Table S4; Lichman et al., 2020). G8H from N. ernesti‐mayeri had a low sequence similarity to related sequences identified in other taxa. The G8H nucleotide sequences from N. sibirica, N. stewartiana, and N. nervosa showed a high level of similarity and were structurally distinct from those of N. rtanjensis, N. nuda, and N. grandiflora (Table S5).

The nucleotide sequences of putative 8HGOs identified in this study displayed a high level of identity with previously characterized homologous genes from N. cataria and N. racemosa (Table S6; Lichman et al., 2020). Nucleotide sequences potentially encoding ISYs identified in the RNA‐seq data from the analyzed Nepeta taxa were highly similar to those of functionally characterized genes encoding members of the PRISE enzyme family from other species (Table S7). The sequences encoding NemISY, NnuISY, and NgISY were closely related to those coding for previously characterized iridoid synthases from N. cataria (NcISY), N. mussinii (NmISY), and N. rtanjensis (NrISY) (Table S7; Sherden et al., 2018; Lichman et al., 2020; Aničić et al., 2024). We identified four isoforms of ISY in N. grandiflora, which may reflect its tetraploid genome. By contrast, the sequences coding for NemPRISE, NrPRISE, NgPRISE, NnuPRISE, NstPRISE, and NsPRISE were more similar to the Family 1 or P5βR isoforms from N. cataria (NcP5βR) and N. mussinii (NmP5βR) (Sherden et al., 2018), and NnPRISE from N. nervosa (Aničić et al., 2024; Table S7). Similarity indices further revealed that genes coding for all NEPS isoforms (NEPS1–5) shared a high level of DNA sequence identity with the corresponding N. mussinii, N. cataria, and N. sibirica NEPS homologs (Lichman et al., 2020; Hernández Lozada et al., 2022; Table S8). NEPS1‐type short‐chain dehydrogenases, including NsNEPS1, NstNEPS, and NnNEPS1, were structurally similar to enzymes previously isolated from N. sibirica, whereas NrNEPS1 displayed high sequence similarity to enzymes characterized from N. cataria (NcNEPS1) and N. racemosa (NmNEPS1) (Table S8).

Candidates for stereoselective cyclases, including NEPS2–4, NrNEPS2, NemNEPS2, NgNEPS2, NsNEPS2, and NnuNEPS2, showed a high level of sequence similarity to NmNEPS2 from N. racemosa, making them promising NEPS2‐type cyclase candidates (Table S8). NrNEPS3, NemNEPS3, NnuNEPS3, and NgNEPS3 showed high sequence similarity to NEPS3s from N. racemosa (NmNEPS3) and N. cataria (NcNEPS3A and NcNEPS3B). One NEPS4 candidate, NrNEPS4, was closely related to the following trans‐cis cyclases and dehydrogenases: NcNEPS4 from N. cataria, NsNEPS4A and NsNEPS4B from N. sibirica, and NmNEPS4 from N. racemosa. Members of the MLPL enzyme family (Table S9), which belongs to the Pathogenesis‐related 10 (PR‐10) family of proteins that promote the formation of cis,trans‐NLL in conjunction with ISYs, were represented by MLPL1‐ and MLPL2‐type enzymes. Nucleotide sequences of NsMLPL1, NstMLPL1, and NnMLPL1 were highly similar to each other and structurally well distinguished from NnuMLPL, NgMLPL1, NrMLPL1b, NemMLPL1, and NrMLPL1a. These putative MLPL1s were similar to those characterized in N. cataria (NcMLPLA, NcMLPLB), N. racemosa (NmMLPL), and N. sibirica (NsMLPL1), which display 7S cis‐trans cyclase activity (Lichman et al., 2020; Hernández Lozada et al., 2022; Table S9). The second group of MLPL enzymes contained one MLPL2‐like enzyme, NgMLPL2, which is structurally similar to NsMLPL2 and NsMLPL3.

We analyzed the expression levels of iridoid‐related biosynthetic genes in the leaves of representative plants from the three chemotypes (Figure 4C). The EBGs up to ISY were expressed in the leaves of all taxa, except for N. nervosa, which lacked sequencing reads for GES and ISY, and for N. stewartiana, which lacked ISY transcripts. We detected high levels of GPPS transcripts in the leaves of N. nuda, with lower levels of 8HGO, PRISE, and ISY transcripts in the same samples. N. rtanjensis and N. ernesti‐mayeri, which are closely related phylogenetically, showed high expression levels of ISYs, GPPS, G8H, and 8HGO (Figure 4C). Notably, we detected transcripts encoding two isoforms of ISY in the leaf RNA‐seq data from tetraploid N. grandiflora (2n = 36); these two genes had different expression levels, suggesting that they may have undergone functional divergence. The species N. nervosa belonging to chemotype C lacked GES and ISY transcripts, whereas GPPS reached the highest expression level among the EBGs.

We detected substantial levels of NEPS1 transcripts in leaf samples from trans,cis‐NL‐containing N. rtanjensis and in those of N. sibirica and N. stewartiana, which produce large amounts of cis,trans‐NL (Figure 4C). The expression of NEPS5 could not be analyzed in this study because no transcripts of this NEPS isoform were present in the RNA‐seq data. In representative species of chemotype A, the expression patterns of genes encoding NEPS cyclases (NEPS2–4) were positively correlated with the presence of specific NL stereoisomers. We identified a gene encoding a trans‐cis cyclase (NrNEPS4) that was highly expressed only in the leaves of N. rtanjensis, although NrNEPS2 showed even higher expression in the same samples, together with high expression of a related dehydrogenase gene, NrNEPS1. Among NEPSs, the expression levels of NEPS2 and NEPS3 were the highest in leaves of N. rtanjensis and N. grandiflora, whereas NEPS1 was highly expressed in the leaves of N. sibirica and N. stewartiana. Additionally, MLPL1 expression was the highest in the leaves of N. sibirica and N. stewartiana, two species that predominantly produce cis,trans‐NL, as well as in the phylogenetically closely related N. nervosa, which belongs to chemotype C lacking iridoids. Despite the absence of cis,cis‐NL in the leaves of chemotype A taxa, NEPS3, encoding a cis,cis cyclase was highly expressed in N. rtanjensis, N. ernesti‐mayeri, and N. nuda. Furthermore, we observed high expression of NEPS2 in N. grandiflora leaves. Generally, high iridoid production in N. rtanjensis was accompanied by high expression levels of EBGs in the iridoid pathway prior to the step catalyzed by 8HGO, as well as high expression of genes encoding NEPSs (NEPS1–4) (Figure 4C).

A PCA plot constructed from the gene expression data showed that the two principal components PC1 and PC2 cumulatively explain 94.5% of the total variance in the data (Figure 4D). N. rtanjensis was clearly separated from the other taxa along PC1, which contributed 66.6% to the total variance. The major contributors to this separation along PC1 were the expression levels of NEPS1, ISY, NEPS3, and NEPS4, which were highly expressed in the leaves of N. rtanjensis, the most productive species in terms of IA content. These results indicate that this high productivity may be related to high expression of genes encoding NEPS and ISY enzymes. The closely related species N. sibirica, N. stewartiana, and N. nervosa were separated from N. nuda, N. ernesti‐mayeri, and N. grandiflora along PC2 (27.9% of the variance), likely due to high MLPL1 expression (Figure 4D).

Transient expression of genes encoding selected NEPS oxidase isoforms in the leaves of N. grandiflora from chemotype B: Re‐establishing the NL production

The absence of NEPS1 or NEPS5 transcripts in the RNA‐seq data from N. grandiflora (chemotype B) suggested possible molecular mechanisms responsible for the lack of NLs in this species. To experimentally test this hypothesis, we infiltrated the leaves of N. grandiflora plants with an Agrobacterium (Agrobacterium tumefaciens) cell suspension expressing NrNEPS1 or NcNEPS5 to recreate NL production (Figure 5A). Surprisingly, we detected a GC/MS peak corresponding to cis,trans‐NL in the leaves of N. grandiflora co‐infiltrated with constructs harboring NrNEPS1 and the viral silencing suppressor p19 (Figure 5B, C). Although only low amounts of cis,trans‐NL accumulated, its identity was unambiguously confirmed using GC/MS and UHPLC/DAD/(±)HESI − MS/MS, along with comparison to the corresponding standards. This experiment confirmed the function of NrNEPS1 as an oxidase, capable of converting cis,trans‐NLL into an NL of the same stereochemistry. The appearance of cis,trans‐NL in the leaves of N. grandiflora plants co‐infiltrated with NrNEPS1 and p19 was accompanied by significantly higher contents of 1,5,9‐eDLA and nepetaside (Figure 5C). Overexpression of NcNEPS5 did not affect iridoid production in N. grandiflora leaves under the described experimental conditions (Figure 5B).

*Reconstitution of nepetalactone production in leaves of the chemotype B representative

N. grandiflora

(A) Transient expression of NrNEPS1 from N. rtanjensis and NcNEPS5 from N. cataria in leaves of N. grandiflora using the agroinfiltration method. The target genes were co‐infiltrated with p19, a universal suppressor of RNA silencing. (B) Metabolites from agroinfiltrated leaves were extracted with methanol and analyzed by GC/MS. Representative total ion chromatograms (TICs) of leaf samples for the combinations NrNEPS1 + p19 (blue line), NcNEPS5 + p19 (purple line), and p19 (green line) are presented, along with the GC/MS spectra of the cis,trans‐NL standard and the peak corresponding to cis,trans‐NL identified in NrNEPS1 + p19 leaf samples. Additionally, the segment from Rt = 10.07 min to 10.13 min of the extracted ion chromatogram (EIC) for the mass of 166 m/z, corresponding to NL, is presented. The peak corresponding to cis,trans‐NL is visible only in the sample of NrNEPS1 + p19 agroinfiltrated leaves (blue line). (C) UHPLC/DAD/(±)HESI−MS2 quantitative analysis targeted cis,trans‐NL and the two major iridoid glycosides (IGs) in N. sibirica: 1,5,9‐eDLA and nepetaside. Results are expressed as means ± SE. Different combinations of constructs are represented as follows: purple bar − NrNEPS1 + p19, white bars with blue diagonal lines − NcNEPS5 + p19, and green bars − p19. Asterisks indicate significantly different values according to the t‐test (P < 0.05). A red star indicates the absence of nepetalactone in samples. Abbreviations: 1,5,9‐eDLA, 1,5,9‐epi‐deoxyloganic acid; cis,trans‐NL, cis,trans‐nepetalactone; Nc, Nepeta cataria; NEPS, nepetalactol‐related short‐chain dehydrogenase; Nr, Nepeta rtanjensis; p19, tombusviral silencing suppressor protein.

Reconstruction of phylogenetic relationships among the selected Nepeta taxa allows mapping of iridoid diversity in a phylogenetic context

Our strategy relied on a comprehensive analysis of 10 chemically and phylogenetically diverse Nepeta taxa to determine their intrageneric positions within the established phylogeny of the genus. We incorporated their internal transcribed spacer (ITS) sequences into the matrix constructed from publicly available ITS data for other representatives of the genus (Figure 6). N. nervosa, N. sibirica, and N. stewartiana were found phylogenetically distinct from the other species examined in this study. Grouping the remaining Nepeta taxa showed relatively low bootstrap support, but revealed the relatedness of N. cataria, N. laevigata, and N. grandiflora (Figure 6). N. nuda was positioned close to the three species endemic to the Balkan Peninsula (N. rtanjensis, N. ernesti‐mayeri, and N. parnassica), which belong to the N. sibthorpii complex of the Nepeta sect. Pycnonepeta (Bentham, 1848). The iridoid‐lacking N. nervosa (chemotype C) clustered separately from N. grandiflora and N. laevigata (chemotype B). The genus Nepeta was clearly separated from iridoid‐lacking genera of the subfamily Nepetoideae (Salvia, Schizonepeta, Agastache, Dracocephalum, Hyssopus, Lavandula, Perilla, Melissa, Mentha, Thymus, and Origanum). The clustering of the Nepetoideae members apart from representatives of the iridoid‐producing Lamioideae and Ajugoideae subfamilies was strongly supported by bootstrap analysis. We used the phylogenetic tree shown in Figure 6 as a framework to further map the diversity of NLs and IGs across the genus Nepeta based on the results of the present study and data from the literature (Table S10), revealing significant knowledge gaps. Only rarely were we able to unambiguously assign a chemotype based exclusively on the published literature. However, there are indications that some of the phylogenetically distinct Nepeta taxa lacking NL‐type iridoids (e.g., N. italica, N. phyllochlamys, N. congesta, N. saturejoides, N. isaurica, N. pungens, and N. daenensis) belong to either chemotype B or C.

*Phylogenetic tree of the genus

Nepeta

illustrating the distribution of major groups of iridoids Phylogenetic relationships were reconstructed based on the nuclear ITS regions of the 10 Nepeta taxa analyzed in this study, along with publicly available ITS sequences of other Nepeta taxa and additional representatives of the subfamily Nepetoideae (Lamiaceae). Representatives of the subfamilies Lamioideae and Ajugoideae are also included. Catharanthus roseus (family Apocynaceae) is used as the outgroup. Using literature data (Table S10) and the results of this study, the distribution of NLs and IGs was mapped across the phylogenetic tree, enabling robust chemotype assignment. The presence or absence of NL‐type iridoids and IGs in the listed taxa, or the complete absence of literature data, is indicated as explained in the legend. Red dots indicate taxa producing only IGs (chemotype B), a blue dot represents iridoid‐lacking taxa (chemotype C), while gray dots indicate taxa producing both nepetalactones and IGs (chemotype A).*

The dendrogram in Figure 7A shows the phylogenetic relationships of the 10 Nepeta taxa studied here, and is based on concatenated sequences of the plastid marker genes trnL*–*F, rbcL, and matK. Our results clearly distinguished N. nervosa from the other nine Nepeta taxa, although it was phylogenetically closer to N. stewartiana and N. sibirica. The three strict endemic species from the Balkan Peninsula (N. rtanjensis, N. ernesti‐mayeri, and N. parnassica) formed a single group. Additionally, N. cataria and N. grandiflora clustered closely with N. nuda, whereas N. laevigata clustered closely with N. grandiflora (Figure 7A). The phylogenetic relationships reconstructed from the plastid barcoding data largely corroborated those based solely on the ITS data (Figure 7B).

*Phylogeny of the genus

Nepeta

based on plastid DNA regions, with comparisons to phylogenetic relationships inferred from ITS markers

(A) Phylogenetic tree constructed from three plastid DNA regions (trnL–F, rbcL, and matK). Black and gray dots indicate taxa producing both nepetalactones and IGs (chemotype A); red dots indicate taxa producing only IGs (chemotype B); and blue dots represent iridoid‐lacking N. nervosa (chemotype C). (B) Comparison of the phylogenetic trees obtained using ITS (presented in Figure 5, here without NCBI sequences) and plastid markers (presented in Figure 6A). This framework is used to map and interpret the distribution of nepetalactones and iridoid glycosides (IGs) among the Nepeta taxa assessed in this study. Gray rectangles indicate the presence (filled) or absence (empty) of nepetalactones (NLs), while red rectangles indicate the presence (filled) or absence (empty) of iridoid glycosides (IGs).*

DISCUSSION

Comprehensive metabolomics combined with a literature survey revealed high iridoid heterogeneity within the genus Nepeta and highlighted knowledge gaps

Scientists have long hypothesized the presence of distinct chemotypes within the Nepeta genus, although definitive evidence has been lacking. A comprehensive literature survey (Figure 6; Table S10) provided only scattered insights into the distribution of IAs and IGs in phylogenetically diverse Nepeta species. However, these data served as a useful foundation for the current study, which clearly demonstrated the existence of distinct and well‐defined chemotypes A, B, and C. Given that the phylogenetic relationships within the genus Nepeta remain largely unresolved and that phytochemical data for most taxa are missing or incomplete, it was challenging to extrapolate findings based on a set of 10 phylogenetically and chemically diverse taxa to the entire genus. Although addressing major knowledge gaps requires further experimental efforts, the present research provides tools to characterize overall iridoid diversity at multiple levels, thus paving the way toward a better understanding of the evolutionary history of the chemical diversity in this remarkable plant genus.

There have been attempts to estimate taxonomic relationships among several Nepeta taxa using only NLs as chemomarkers (De Pooter et al., 1988; Kilic et al., 2013). However, not all Nepeta taxa produce NLs, and some closely related taxa show considerable qualitative differences in these iridoids, challenging the usefulness of NLs as chemomarkers at the genus level. Considering our LC/MS and GC/MS data, we detected NL‐type IAs (NLL, trans,trans‐NL, cis,trans‐NL, trans,cis‐NL, 5,9‐DNL, and dihydronepetalactones) only in the leaves of taxa assigned to chemotype A, with qualitative and quantitative variations in their composition of NL‐type iridoids. Specifically, N. sibirica, N. cataria, N. nuda, and N. ernesti‐mayeri a rich in cis,trans‐NL (subchemotype A2). N. cataria was reported to contain primarily cis,trans‐NL, with other stereoisomers also present (Mišić et al., 2015; Srivastava et al., 2021). N. sibirica contains mainly cis,trans‐NL (Nestorović et al., 2010; Tsuruoka et al., 2012; Mišić et al., 2015; Ali et al., 2016; Popović et al., 2025).

In the present study, N. stewartiana was rich in both trans,trans‐NL and cis,trans‐NL (A1), with some trans,cis‐NL also detected. The leaves of N. rtanjensis and N. parnassica plants predominantly accumulated trans,cis‐NL (A3), consistent with results from numerous earlier studies (Nestorović et al., 2010; Gkinis et al., 2014; Dmitrović et al., 2015; Mišić et al., 2015; Skorić et al., 2017; Aničić et al., 2018, 2020, 2021, 2024). We assigned the dihydronepetalactone isomers according to Sengupta et al. (2018), and these isomers were previously reported as minor components in N. cataria (e.g., Sengupta et al., 2018; Patel et al., 2022) and N. nuda (Petrović et al., 2024a). 5,9‐DNL was also previously reported in N. cataria (e.g., Lockhart et al., 2024), N. nuda (e.g., Petrović et al., 2024a), N. rtanjensis (Mišić et al., 2015; Aničić et al., 2018, 2020, 2021, 2024), and N. sibirica (De Pooter et al., 1988). The Balkan Peninsula endemic species N. rtanjensis and N. parnassica were especially rich in this compound; however, the closely related N. ernesti‐mayeri contained no 5,9‐DNL.

We detected nepetalic acid (NA, 22), a degradation product of NL, in nine of the 10 Nepeta species in the present study, with only N. nervosa lacking NA. This compound was previously reported in N. cataria (e.g., Lockhart et al., 2024) and N. nuda (e.g., Petrović et al., 2024a). Some previous studies reported the absence of NLs in the chemotype B taxon N. grandiflora (Mišić et al., 2015; Colţun et al., 2023), whereas other studies observed trace amounts of cis,trans‐NL (Birkett et al., 2010). Similarly, N. laevigata was reported to contain minor amounts of cis,trans‐NL (Hassan et al., 2011) or to completely lack NLs (Mišić et al., 2015). For NL‐lacking N. nervosa, the sole representative of chemotype C, several previous studies reported finding no NLs in this species (Nestorović et al., 2010; Mišić et al., 2015; Aničić et al., 2024; Rasheed et al., 2024), but another research group observed trace amounts of this compound in the commercial cultivar “Blue Moon” (Hijazi et al., 2024).

A characteristic feature of IGs present in species of the genus Nepeta is a bicyclic cyclopentanopyran ring system that usually contains a β‐sugar linkage at position C‐1 and unusual stereochemistry at C‐1, C‐5, and C‐9. We identified a major IG in Nepeta species, 1,5,9‐eDLA, in all iridoid‐producing Nepeta taxa belonging to chemotypes A and B. The compound 11‐O‐hexosyl‐1,5,9‐eDLA has been previously reported in N. sibthorpii, N. rtanjensis (Aničić et al., 2021), and N. nuda (Petrova et al., 2022; Petrović et al., 2024b), and was identified here in all Nepeta taxa belonging to chemotypes A and B, except for N. sibirica. 11‐O‐Pentosyl‐1,5,9‐eDLA has previously been reported in N. nuda (Petrova et al., 2022; Petrović et al., 2024b) and was identified here in the leaves of taxa from chemotypes A and B, except for N. stewartiana. We detected acylated forms of 1,5,9‐eDLA, caffeoyl‐malonyl‐1,5,9‐eDLA, and shikimoyl‐1,5,9‐eDLA only in some phylogenetically distinct representatives of chemotype A; these forms had not been previously reported in any species from the genus Nepeta.

The aglycone part of the subgroup of IGs characterized by a methyl group at the C‐11 position is structurally more related to NLs, as represented by 5‐deoxylamiol, nepetaracemoside A, nepetaside, and nepetariaside. Nepetaside, nepetaracemoside A, and nepetariaside are also notable for the unusual positions of the glucose moieties at the C‐7, C‐5, and C‐3 positions, respectively. 5‐Deoxylamiol is abundant throughout the Lamiaceae, and was previously reported in N. nuda (Petrova et al., 2022). In the present study, we detected this compound in the leaves of N. sibirica and N. ernesti‐mayeri. Nepetaracemosides A and B were previously reported in N. racemosa (Takeda et al., 1999), with the former detected in the leaves of all taxa, except for N. sibirica in this study. Nepetariaside, which possesses an open dihydropyran ring, was previously reported in N. cataria (Murai et al., 1987), N. rtanjensis, and N. argolica (Aničić et al., 2021), and was reported in all Nepeta taxa in the present study. Notably, the configurations of nepetariaside and 1,5,9‐eDLA differ at the C‐5 and C‐9 positions, confirming the co‐occurrence of IGs with variable stereochemistry within the genus Nepeta. Nepetaside has previously been reported in N. cataria (Xie et al., 1988) and N. nuda (Petrović et al., 2024b) and was recorded in all taxa analyzed in this study. By contrast, we identified adenosmoside, which is characteristic of the ornamental species blue adenosma (Adenosma caeruleum), in N. parnassica and N. nuda, and we confirmed the presence of 6β‐dihydrocornic acid, which is abundant in evergreen dogwood (Cornus capitata), in N. nuda leaves.

We also analyzed the profiles of other groups of volatile terpenoids to determine whether their occurrence and content coincided with those of iridoids. We identified the sesquiterpenes β‐caryophyllene and germacrene D as major terpenoids, although the monoterpene 1,8‐cineole was also abundant, consistent with previous studies (e.g., Birkett et al., 2010; Hassan et al., 2011; Tsuruoka et al., 2012; Gkinis et al., 2010, 2014; Skorić et al., 2017; Petrović et al., 2024a). Taxa belonging to chemotypes B and C were especially rich in these compounds, indicating that the redirection of metabolic flux toward the biosynthesis of other monoterpenes and sesquiterpenes is prioritized.

Comprehensive metabolomics enabled the reconstruction of the metabolic network of the major iridoids in phylogenetically diverse Nepeta taxa

Based on the metabolomics data generated in this study and previous studies, we propose a metabolic network for the major iridoids in 10 Nepeta taxa (Figure 8). The early steps of the iridoid biosynthesis pathway up to NL have been well established in N. cataria and N. racemosa (Sherden et al., 2018; Lichman et al., 2019a, 2019b; Hernández Lozada et al., 2022). NL is produced from geranyl pyrophosphate (GPP), which originates from the methylerythritol phosphate (MEP) pathway via a series of intermediates and enzymatic reactions catalyzed by GPPS, GES, G8H, 8HGO, ISY, MLPLs, and NEPSs (Figure 8A). In Nepeta species, ISYs are primarily responsible both for the stereoselective 1,4‐reduction of 8‐oxogeranial (8OG) to the uncyclized and reactive 8‐oxocitronellyl enol (8CE) and for determining the stereochemistry at the C7 position (Sherden et al., 2018; Lichman et al., 2019a, 2020; Hernández Lozada et al., 2022). NEPS enzymes are responsible for fixing the stereochemistry of the bridged carbons (C4a and C7a) of the NL molecule (Hernández Lozada et al., 2022). Although the enolate intermediate 8CE undergoes a stereoselective cyclization mediated by NEPS cyclases (NEPS2–4) and MLPLs to yield different NLL stereoisomers (Lichman et al., 2019a, 2019b), it can also undergo spontaneous cyclization to produce predominantly cis,trans‐NLL (Hernández Lozada et al., 2022).

*Proposed metabolic network of major iridoids in leaves of analyzed

Nepeta

taxa

(A) Simplified scheme of the iridoid biosynthetic pathway in Nepeta species up to the major iridoid aglycones, nepetalactones, and iridoid glycosides. (B) Steps of the pathway preceding the formation of cis,trans‐NL, involving the reduction step mediated by iridoid synthase (ISY), the cyclization step (NEPS2–3, and MLPL1), and oxidation via NEPS1. (C) Summarized functions of NEPS isoforms, ISYs, and MLPL1, as proposed by Sherden et al. (2018) and Lichman et al. (2020). (D) Proposed mechanism for the conversion of 8‐oxogeranial into cis–cis oriented 1,5,9‐eDLA, involving a cyclization step catalyzed by NEPS3 (proposed cis–cis cyclase), as well as the involvement of iridoid oxidase (IO) and glucosyltransferase (UGT). Abbreviations: 1,5,9‐eDLA, 1,5,9‐epi‐deoxyloganic acid; 5,9‐DNL, 5,9‐dehydronepetalactone; 8CE, 8‐oxocitronellyl enolate; 8‐epi‐7‐DLN, 8‐epi‐7‐deoxyloganin; 8HG, 8‐hydroxygeraniol; 8HGO, 8‐hydroxygeraniol oxidoreductase; 8OG, 8‐oxogeranial; 11‐O‐Hex‐1,5,9‐eDLA, 11‐O‐hexosyl‐1,5,9‐epi‐deoxyloganic acid; 11‐O‐Pent‐1,5,9‐eDLA, 11‐O‐pentosyl‐1,5,9‐epi‐deoxyloganic acid; CYP450, cytochrome P450 monooxygenase; DHN, dihydronepetalactone; DMAPP, dimethylallyl pyrophosphate; eDLNA, epi‐deoxyloganetic acid; G8H, geraniol 8‐hydroxylase; GE, geraniol; GES, geraniol synthase; GPP, geranyl pyrophosphate; GPPS, geranyl diphosphate synthase; IPP, isopentenyl pyrophosphate; ISY, iridoid synthase; MLPL, major latex protein‐like enzyme; NEPS, nepetalactol‐related short‐chain dehydrogenase; NL, nepetalactone; NLL, nepetalactol; OMT, O‐methyltransferase; PRISE/ISY, progesterone‐5β‐reductase/iridoid synthase activity displaying enzymes; UGT, UDP‐glycosyltransferase; UPT, UDP‐pentosyltransferase.*

The iridoid biosynthetic pathway most likely diverges at the point of NLL into two major branches that can be conditionally divided into the IA branch, referring to NL‐type iridoids, and the IG branch, which gives rise to 1,5,9‐eDLA and related compounds (Figure 8A). The IA branch begins with the oxidation of NLL (Hallahan et al., 1998; Lichman et al., 2019a, 2019b), a reaction mediated by NEPS oxidases, resulting in NLs with a stereochemistry most likely corresponding to that of their precursor NLL (Figure 8B). As previously mentioned, two isoforms of NEPS oxidase have been characterized in Nepeta species: NEPS1 and NEPS5 (Lichman et al., 2020; Hernández Lozada et al., 2022). Notably, dual‐function NEPS enzymes, which catalyze the cyclization and oxidation of various NLLs to NLs, also exist in Nepeta species (Figure 8C; Lichman et al., 2020; Hernández Lozada et al., 2022). MLPL1 has been suggested to be involved in determining the stereospecificity of ring closure and 7S cis‐trans cyclization of iridoids in Nepeta species (Lichman et al., 2020). Subsequently, the NL skeleton can be further modified by dehydrogenation or hydrogenation reactions, resulting in the formation of 5,9‐DNL or dihydronepetalactones, respectively (Figure 8A).

The steps of the IG biosynthetic branch leading to 1,5,9‐eDLA remain unresolved in the genus Nepeta and are proposed here based on metabolomics data (Figure 8A, D) and similarity to previously characterized biosynthetic pathways in certain iridoid‐ and secoiridoid‐rich species (Miettinen et al., 2014; Salim et al., 2014). Considering the cis,cis‐stereochemistry of 1,5,9‐eDLA, its biosynthesis was suggested to involve a cyclization step, most likely mediated by NEPS enzymes (Lichman et al., 2020). For example, we propose that NEPS3 functions as a cis–cis cyclase and establishes the stereochemistry of cis,cis‐NLLs, which are then converted into lactol‐type IGs (e.g., 1,5,9‐eDLA). However, this speculation requires experimental evidence for validation. One of the initial steps of the IG branch may be the hydroxylation of NLL mediated by iridoid oxidase (IO), resulting in the formation of epi‐deoxyloganetic acid, which is further glycosylated to produce 1,5,9‐eDLA (Figure 8D). Enzymes such as UDP‐glucosyl transferases (UGTs), methyl transferases (OMTs), aldehyde dehydrogenases (ALDHs), and *β‐*glucosidases (β‐GLUs) are proposed to participate in various steps of the IG biosynthesis branch. The conversion of 1,5,9‐eDLA into 11‐COOHex and 11‐COOPen variants clearly involves additional glycosylation steps (Figure 8A). Nepetanudoside may arise after the oxidation of 1,5,9‐eDLA that results in the formation of epi‐loganic acid, followed by a subsequent dehydrogenation. Methylation of epi‐deoxyloganetic acid may result in the formation of epi‐deoxyloganetin. In addition, 1,5,9‐eDLA can be converted into either shikimoyl‐1,5,9‐eDLA or caffeoyl‐malonyl‐1,5,9‐eDLA, involving one or two acylation steps, respectively. Deoxygeniposide aglycone most likely arises through the deglycosylation of deoxygeniposide.

Phylotranscriptomics highlights the molecular mechanisms determining iridoid heterogeneity within the genus Nepeta

Most Nepeta taxa typically produce more than one NL stereoisomer; this requires the activity of several NEPS cyclase and MLPL isoforms. NEPS1 enzymes catalyze the formation of either cis,trans‐NL or cis,cis‐NL from the corresponding NLL isomers (Lichman et al., 2019a, 2020), and we identified transcripts for the genes encoding these enzymes in all Nepeta taxa analyzed here. We also detected transcripts for NEPS2, which encodes an enzyme that promotes the formation of cis‐trans‐NLL (Lichman et al., 2020), in the leaves of Nepeta taxa rich in cis,trans‐NL (N. sibirica, N. rtanjensis, N. ernesti‐mayeri, and N. nuda). Although we detected no cis,cis‐NL in these Nepeta taxa, NEPS3 transcripts were abundant in their transcriptomes, suggesting the possible involvement of the cis–cis cyclase NEPS3 (Lichman et al., 2020) in generating the cis,cis‐NLL pool, and its efficient conversion into cis,cis IGs (e.g., 1,5,9‐eDLA). This hypothesis is supported by the fact that the leaves of the chemotype B taxon N. grandiflora, which lacks NEPS1 and NEPS5 transcripts and is deficient in NL biosynthesis, displayed considerable NEPS3 expression and produced 1,5,9‐eDLA. Furthermore, we detected no expression for NEPS3 in the leaves of N. sibirica or N. stewartiana, two Nepeta taxa that are recognized as low IG producers. However, these speculations also require experimental evidence. NEPS4s from N. cataria and N. racemosa were previously reported to act both as trans‐cis cyclases and as residual dehydrogenases selective for cis,trans‐NLL over trans,cis‐NLL (Lichman et al., 2020), and this NEPS isoform was highly expressed in N. rtanjensis, which is rich in trans,cis‐NL.

Previous research on the function of NEPS enzymes in planta revealed that the simultaneous silencing of NEPS1, NEPS3, NEPS4, and NEPS5 in N. cataria resulted in significantly lower levels of cis,cis‐NL, whereas the levels of other NL isomers remained unchanged (Palmer et al., 2022). The absence of NEPS1 and NEPS5 transcripts in N. grandiflora (chemotype B) underlies its inability to convert NLL into NL. We unambiguously demonstrated this idea by expressing NrNEPS1 from N. rtanjensis (chemotype A) in N. grandiflora leaves, which resulted in the resumption of cis,trans‐NL production. Surprisingly, expression of NrNEPS1 in the leaves of N. grandiflora via Agrobacterium‐mediated infiltration also resulted in significantly higher levels of 1,5,9‐epi‐DLA and nepetaside, suggesting possible residual cis,cis cyclase activity of this enzyme. Nevertheless, additional experiments are needed to substantiate this proposition.

Simultaneous quantification of iridoids in the leaves of selected Nepeta species coupled with profiling of the expression levels of iridoid biosynthesis‐related genes in the same types of samples suggested that the role of genes encoding NEPS cyclases, which showed lower (NESP2) or completely absent expression (NEPS3 and NEPS4) in the phylogenetically related N. nervosa, N. sibirica, and N. stewartiana, was most certainly replaced by highly expressed MLPL1. Supporting this hypothesis, the greater accumulation of cis,trans‐NL and IG in N. sibirica following infection by Trichoderma fungi was associated with significant upregulation of NsMLPL1 expression (Popović et al., 2025). The essential role of MLPLs in iridoid production was demonstrated by introducing Nepeta MLPL into monoterpene indole alkaloid (MIA)‐producing microorganisms and by its transient expression in engineered leaves of Nicotiana benthamiana. This heterologous expression enhanced metabolic flux through the iridoid biosynthesis pathway and increased the production of targeted bioactive compounds (e.g., Dudley et al., 2022; Misa et al., 2022; Gao et al., 2023).

Although N. nervosa (chemotype C) possesses a biochemical reservoir of enzymes related to iridoid biosynthesis, iridoid accumulation is prevented in this species due to the loss of function or absence of expression of some of the EBGs, specifically GES, ISY, and NEPS, which encode cyclases. GES represents the first step in the iridoid biosynthetic pathway, redirecting metabolic flux from competing biosynthetic routes, and is considered a rate‐limiting step in iridoid biosynthesis in iridoid‐lacking genera of the Nepetoideae subfamily (Mint Evolutionary Genomics Consortium, 2018). A closely related example is the non‐iridoid‐producing species hyssop (Hyssopus officinalis), in which a NEPS‐like gene, but not PRISE/ISY or GES, was identified (Lichman et al., 2020). The situation in N. nervosa is very similar to that in H. officinalis; it has been speculated that during evolution, N. nervosa most likely underwent a secondary loss of the ability to produce both NL‐type iridoids and IGs, probably due to the loss of GES and ISY expression or genes or by their loss in different syntenic regions. PRISE isoforms (referred to as ISY Family 1 by Sherden et al., 2018) show very low catalytic activity (Sherden et al., 2018) and are considered physiologically irrelevant for iridoid biosynthesis in Nepeta species (Lichman et al., 2020; Popović et al., 2025). However, we identified a homolog of PRISE in the N. nervosa RNA‐seq data (NnPRISE), and it was previously revealed to show ISY activity in vitro (Aničić et al., 2024). Theoretically, it is possible that NnPRISE acts nonspecifically in N. nervosa to reduce 8OG, forming 8CE, which can then be enzymatically processed by NnMLPL1, which is believed to display 7S cis‐trans cyclase activity (Lichman et al., 2020), and subsequently by NnNEPS1 to yield cis,trans‐NL.

Chemotype‐specific iridoid production is regulated at the transcript level

Because of the pharmacological, agricultural, and economic importance of iridoids, interest in understanding the molecular details behind their biosynthetic pathways and developing biotechnological approaches for their targeted production has intensified. A prerequisite for scaling up the production of bioactive iridoids of interest is selecting either highly productive Nepeta taxa or genotypes or choosing those with unique iridoid profiles for developing biotechnological protocols. Two endemic species of the Balkan Peninsula, N. rtanjensis and N. parnassica, are highlighted here as iridoid‐high‐productive taxa, rich in trans,cis‐NL, DNL, and 1,5,9‐eDLA, making them good candidates for commercial production of bioactive iridoids. Generally, the pronounced ability of some Nepeta taxa to produce iridoids is accompanied by high expression levels of EBGs up to 8HGO and of NEPSs, while in the case of N. sibirica and N. stewartiana expression of MLPLs is also important. These enzymatic steps are highlighted here as key regulatory points determining the metabolic flux through the iridoid biosynthetic pathway.

The results of this study provide a foundation for in planta metabolic engineering by identifying the rate‐limiting metabolic steps in different Nepeta chemotypes and determining which steps can be modified to support the targeted accumulation of particular bioactive iridoids of interest. It is well established that successful plant engineering depends on extensive fundamental knowledge about the metabolic pathways and requires biotechnological tools that can be used to achieve cost‐effective and sustainable production of desired natural products (Liu et al., 2023). Overall, this study demonstrates the potential of using N. grandiflora as a bioproduction chassis for iridoids. Its robust growth habit, leaves that are easily amenable to Agrobacterium‐mediated infiltration, and the lack of expression for genes from the biosynthetic branch leading to NL‐type iridoids, which would otherwise compete with the IG branch for the common precursor NLL, make N. grandiflora an attractive platform for scaling up the production of desirable IAs and IGs with unique stereochemistry. The optimization of metabolic flux through the iridoid pathway in this species by co‐expression of additional EBGs identified as key regulatory points may enhance the production of NLL, and, consequently, NL and IGs, and is the focus of our ongoing work. N. grandiflora is a more convenient model than N. benthamiana for the functional characterization of the remaining unknown enzymes in the IG biosynthetic branch leading to the production of 1,5,9‐eDLA and related compounds. Previous attempts to reconstitute iridoid production in N. benthamiana often resulted in the generation of undesirable side products because early iridoid biosynthetic intermediates were frequently modified by the addition of pentose and hexose sugars, suggesting promiscuous activity of the N. benthamiana Family 1 UGTs (Dudley et al., 2022).

Mapping iridoid diversity in a phylogenetic context provides insights into chemotype evolution within the genus Nepeta

Many attempts have been made to reconstruct phylogenetic relationships within the Nepeta genus, using various strategies and methodologies, such as DNA barcoding (Jamzad et al., 2003; Al‐Qurainy et al., 2014), chemical markers (De Pooter et al., 1988; Kilic et al., 2013; Mišić et al., 2015), morphological features (Celenk et al., 2008), and combinations of DNA barcoding with biogeographical disjunctions (Rose et al., 2023). Studies conducted so far, which have usually considered only a small number of the extant Nepeta taxa, have only begun to reveal the phylogeny of this genus, and phylogenetic relationships within it remain far from unambiguously resolved. Extensive studies by Jamzad et al. (2003) and Rose et al. (2023) have generated a large number of publicly available ITS sequences, providing a good starting point for reconstructing an updated phylogenetic map of the genus. The former study used four plastid markers and four nuclear markers to estimate phylogenetic relationships among 35 Nepeta taxa, but the latter study focused on 26 taxa and analyzed their ITS sequences. In this study, we determined that N. nervosa, N. sibirica, and N. stewartiana are phylogenetically distinct from the remaining species. This finding is consistent with Rose et al. (2023), who showed that N. nervosa clusters close to N. sibirica. However, some previous studies have distinguished N. sibirica, which belongs to sect. Macronepeta, from N. nervosa, which was assigned to sect. Spicata (Bentham, 1848). Another study placed N. sibirica and N. nervosa in separate clades, namely, IIA and IIB, respectively (Jamzad et al., 2003). By comparing whole‐plastid genome sequences, Chen et al. (2024) positioned N. stewartiana close to N. bracteata, N. hemsleyana, and N. tenuifolia, but these conclusions should be interpreted with caution, as they were based on the analysis of only four taxa with rather unclear phylogenetic relationships.

N. cataria (sect. Cataria according to Bentham, 1848; sect. Nepeta according to Budantsev, 1993) was closely related to the very similar N. laevigata and N. grandiflora. Although N. laevigata was previously assigned to sect. Spicata (Benth.) Pojark. (Bentham, 1848), in our study, this species clustered more closely with N. grandiflora than with N. nervosa.  Rose et al. (2023), on the other hand, grouped N. grandiflora closely with N. cataria. Although previous studies have suggested a close relationship between N. nuda and N. cataria (Petrova et al., 2022), N. nuda clustered near the three Balkan Peninsula endemic species (N. rtanjensis, N. ernesti‐mayeri, and N. parnassica), which belong to the N. sibthorpii complex of sect. Pycnonepeta (Bentham, 1848). This finding supports the results published by Rose et al. (2023), who found N. argolica (syn. Nepeta sibthorpii) to be closely related to N. nuda. Phylogenetic relationships reconstructed based on plastid barcoding largely corroborated those based solely on ITS data. However, we were not able to reconstruct phylogenetic relations of the taxa analyzed in the present study with other congeneric species based solely on plastid loci. Several studies of the genus have previously used some of these markers (e.g., Al‐Qurainy et al., 2014; Petrova et al., 2022; Rose et al., 2023), but the specific combination of the three markers used here has not been used before for the genus Nepeta.

Following the projection of iridoid diversity onto the molecular phylogeny, we conclude that Nepeta is a highly heterogeneous genus, with taxa in distinct phylogenetic lineages lacking either NL‐type iridoids or both NLs and IGs. This observation suggests convergent chemical evolution, most likely through either gene loss or loss of expression. The loss of NL biosynthesis in chemotype B taxa (N. grandiflora and N. laevigata) and the absence of iridoids in the chemotype C taxon (N. nervosa) are most likely the results of independent evolutionary events, as representatives of these two chemotypes have distinct molecular backgrounds for their iridoid profiles and are phylogenetically well distinguished. The two chemotype B taxa, however, most likely have a monophyletic origin. Moreover, the literature indicates the absence of NLs in several other phylogenetically distinct taxa (N. daenensis, N. deflersiana, N. isaurica, N. italica, N. laxifolia, N. pamirensis, N. prattii, N. saturejoides, N. schiraziana, and many others), suggesting that chemotypes B and C may be more widely represented across the genus. However, most of these species have not been analyzed for their IG contents, and RNA‐seq data sets are not available. As informative metabolomic and transcriptomic data sets are generated for more Nepeta taxa, the molecular basis of iridoid diversity is expected to become clearer. However, the presence of such heterogeneity in the distribution of iridoids among extant Nepeta species provides evidence for the evolutionary events that led to the secondary loss of metabolic function in ancestral species. This idea is in accordance with what is already known about complex evolutionary pathways of monoterpenes and sesquiterpenes in members of the Lamiaceae family, with parallel gains and losses of individual compounds or groups of compounds throughout evolution (Mint Evolutionary Genomics Consortium, 2018).

The findings of this study clarify the molecular mechanisms underlying iridoid diversity within the genus Nepeta, highlighting the crucial roles of ISY, NEPS, and MLPL enzymes in determining whether a specific Nepeta species produces iridoids and, if so, which class, namely, NL‐type IAs and/or IGs. These results provide a valuable foundation for future research aimed at elucidating the key elements of the iridoid biosynthetic pathway in the genus Nepeta and its chemical evolution.

MATERIAL AND METHODS

Chemicals and reagents

Acetonitrile (Fisher Scientific UK, Leicestershire, UK) and formic acid (Merck, Darmstadt, Germany) were of MS grade. Ultra‐pure deionized water was generated using a Water Purification System (New Human Power I Integrate, Human Corporation, Republic of Korea). Analytical standards of trans,cis‐NL, 1,5,9‐eDLA, and 5,9‐DNL were isolated from natural sources as previously described by Aničić et al. (2021). Standards of cis,trans‐NL and dihydronepetalactone (DHN) were generous gifts from Entomol Products LLC (San Francisco, CA, USA). Analytical standards of loganin and aucubin were purchased from Sigma‐Aldrich (Hamburg, Germany). Terpene Mix B (Merck KGaA, Darmstadt, Germany) was used for GC/MS identification of terpene compounds in the samples.

Species selection

We selected 10 Nepeta species belonging to one of three predefined chemotypes. Based on the phytochemical composition of representatives of the genus available in the literature and a phylogenetic tree constructed from ITS sequences in the NCBI database, as well as those acquired in the present study (Figure 6), we developed an evolutionarily based sampling scheme for this study. Another restrictive criterion for selection was the availability of seeds.

Plant material

Seeds of 10 Nepeta taxa (N. cataria L., N. ernesti‐mayeri Diklić & V. Nikolić, Nepeta grandiflora M.Bieb., N. laevigata (D.Don) Hand.‐Mazz., N. nervosa Royle ex Benth, Nepeta nuda L., N. parnassica Heldr. & Sartori, N. rtanjensis Diklić & Milojević, N. sibirica L., N. stewartiana Diels) were obtained from various sources that are listed in Table S11 and germinated under greenhouse conditions. The resulting plants were grown in pots containing a mixture of soil and perlite, in the greenhouse of the Institute for Biological Research “Siniša Stanković”—National Institute of the Republic of Serbia, University of Belgrade (Figure 1A). Leaves from 3‐month‐old Nepeta individuals were excised from flowering plants, immediately soaked in liquid nitrogen (LN), and stored at −80°C until use. Growth conditions and harvesting time were standardized to exclude possible environmentally driven and developmentally regulated variations in the accumulation of the studied iridoid metabolites.

Metabolomics

Before extraction, leaf samples were ground into a fine powder using LN and immediately weighed (100 mg per sample). The extraction procedure involved several steps: soaking the samples in 96% methanol (w:v = 1:10), vortexing for 1 min, ultrasound‐assisted extraction for 1 h, and centrifugation at 10,000 g for 10 min. Supernatants were collected, filtered through 0.22‐μm filters (Agilent Technologies, Santa Clara, CA, USA), and stored at 4°C until analysis. Leaf samples were analyzed in triplicate for each Nepeta taxon.

Untargeted metabolomics of Nepeta taxa using LC‐HRMS/MS analysis

An LC‐HRMS/MS instrument (Thermo Scientific™ Vanquish™ Core HPLC system coupled to the Orbitrap Exploris 120 mass spectrometer, San Jose, CA, USA) was used to determine the metabolic profiles of the extracts. The liquid chromatography system was equipped with a Hypersil GOLD™ C18 analytical column (50 × 2.1 mm, 1.9 μm particle size), maintained at 40°C. Ultrapure water with 0.1% formic acid (A) and acetonitrile (MS grade) with 0.1% formic acid (B) were used as the mobile phase. Compounds of interest were eluted over 15 min at a constant flow rate of 300 μL min^−1^ with the following gradient of mobile phase: 5% B for the first min, 5%–95% B from 1 to 10 min, 95% B from 10 to 12 min, and 5% B until 15 min. The injection volume was 5 μL.

An Orbitrap Exploris 120 mass spectrometer was equipped with a heated electrospray ionization (HESI‐II) source operating in both positive and negative ionization modes. Other LC/MS parameters were published earlier by Stojković et al. (2024). Full‐scan MS was monitored from 100 to 1,500 m/z with an Orbitrap resolution set to 60,000 FWHM, while data‐dependent MS^2^ experiments were conducted at an Orbitrap resolution of 15,000 FWHM. Normalized collision energy (cE) was set to 35% with an isolation width of 1.5 m/z. The dynamic exclusion time was set to 10 s, with exclusion from a specific scan after two occurrences, and the intensity threshold was set to 1 × 10^5^. A total of 30 samples (three replicates per Nepeta accession) were analyzed using the UHPLC Orbitrap MS technique.

LC/MS data were evaluated in the RStudio (version 2023.09.1, build 494) software environment. Metabolite identification was based on chromatographic behavior and HRMS/MS^2^ data by comparison with standard compounds when available and literature data for tentative identification. Data acquisition was performed using the Xcalibur® data system (Thermo Finnigan, San Jose, CA, USA).

Untargeted GC/MS metabolomics of methanol extracts of Nepeta taxa

Profiling of volatile terpenoids in methanol extracts of Nepeta accessions was conducted using an Agilent 8890 gas chromatograph (GC) coupled with a Mass Selective Detector (5977B GC/MSD, Agilent Technologies, Santa Clara, USA) and connected to an automated sample extraction and enrichment platform (Centri®, Markes International Ltd., Bridgend, UK). The chromatographic separation conditions and MS parameters were previously described in detail by Aničić et al. (2024). A 1 μL methanol extract was injected in split mode (20:1), with a split flow of 24 mL min^−1^. Volatile compounds in methanol extracts were chromatographically separated on an HP‐5MS column (30 m × 0.25 mm, 0.25 μm film thickness) (Agilent Technologies, USA) using helium (99.999%, The Linde Group, Ireland) as the carrier gas at a flow rate of 1.6 mL min^−1^. The temperatures of the transfer line and detector were set to 280°C and 270°C, respectively. Mass spectra were acquired in positive EI mode (+70 eV), with the EI source temperature set to 280°C. The column temperature was linearly increased from 40°C to 300°C at the rate of 20°C min^−1^ and held isothermally at 300°C for 10 min. Analyses were performed in SCAN mode, tracking compounds within the range of 45 to 500 amu. The constituents of the methanol extracts were identified using the Agilent MassHunter Workstation and the Unknown Analysis program by comparing the mass spectra and retention times of the compounds with those of the respective standards, as well as by comparison with the NIST05 library, with a minimum match factor set to 50%. Experimentally recorded m/z spectra displayed a match factor above 80% with those from the NIST library. Kovats Retention Indices (RIs) were calculated according to the definition of Van Den Dool and Kratz (1963), based on the elution characteristics of alkane standards (Supelco®, Bellefonte, USA). Raw GC/MS data (peak areas) are deposited in the trusted Institutional Repository RADaR (https://radar.ibiss.bg.ac.rs/) under the following ID: https://hdl.handle.net/21.15107/rcub_ibiss_7573.

UHPLC/DAD/(±)HESI‐MS2 targeted metabolic profiling of Nepeta taxa

The selected reaction monitoring (SRM) mode of a UHPLC/DAD/(±)HESI‐MS^2^ instrument was used to quantify major IAs (cis,trans‐ and trans,trans‐NL, 5,9‐DNL, dihydronepetalactone, and cis,trans‐NLL) and IGs (nepetanudoside, nepetaside, and 1,5,9‐eDLA) in leaf samples of the studied Nepeta accessions. A Dionex UltiMate 3000 UHPLC system coupled with a DAD detector and configured with a triple quadrupole mass spectrometer (TSQ Quantum Access Max, Thermo Fisher Scientific, Basel, Switzerland) was used for quantification. Samples (10 μL) were chromatographically separated on a Hypersil Gold C18 analytical column (50 × 2.1 mm, 1.9 μm particle size; Waltham, MA, USA) using the elution gradient and flow rate previously described by Aničić et al. (2021). The mobile phase consisted of (A) water with 0.1% formic acid and (B) acetonitrile with 0.1% formic acid. The targeted compounds were quantified using calibration curves of commercial standards. Nepetanudoside was quantified using the calibration curve of loganin, while the calibration curve of aucubin was used for nepetaside. Calibration curves of pure standards showed good linearity, with r² values exceeding 0.99 (peak areas vs. concentration). The total amount of each iridoid compound was determined by calculating its peak area and is expressed as μg per 100 mg of leaf fresh weight (FW).

Statistical analysis of metabolomics data

Qualitative metabolomics data were normalized to tissue fresh weight before statistical processing. For hierarchical cluster analysis (HCA), input variables were scaled to the [min, max] range. HCA was performed using Spearman's method of cluster agglomeration with Morpheus software (https://software.broadinstitute.org/morpheus). Relative GC/MS quantitative data (peak areas) were subjected to PCA adopting the Pearson's algorithm. A network map was reconstructed based on the Jaccard similarity index and a circular algorithm with a 50% edge cutoff. Linear (Pearson's) correlation statistics were obtained with the aim of observing interrelations between the relative quantities of the identified compounds. PCA and correlation corrplots, as well as the network map, were constructed in Past 5 software (version 5.2.1, Natural History Museum, University of Oslo). For sample comparisons, quantitative metabolomics data were analyzed using Tukey's post hoc t test (*P < 0.05) following one‐way ANOVA.

Transcriptomics

RNA‐Seq and Nepeta transcriptome mining toward qualitative profiling of candidate genes included in the biosynthesis of iridoids

To establish a transcriptomic database of the 10 selected Nepeta taxa, leaves from 3‐month‐old plants were harvested, immediately frozen in LN, and stored at −80°C until further use. Trichomes from N. rtanjensis, N. grandiflora, and N. sibirica were isolated using the dry ice abrasion technique, as described by Aničić et al. (2018). Total RNA was extracted from trichomes of N. sibirica, N. grandiflora, and N. rtanjensis, trichome‐depleted leaves of N. rtanjensis and N. grandiflora, roots of N. rtanjensis, and intact leaves and trichomes from the remaining Nepeta species, following the protocol outlined by Gasic et al. (2004). RNA sequencing of N. rtanjensis and N. grandiflora organs was performed in duplicate. RNA purity and concentration were determined using a NanoPhotometer® N60 (IMPLEN, München, Germany) and a Qubit 3.0 fluorometer (Thermo Fisher Scientific, Waltham, MA, USA) with a Qubit RNA Assay Kit. Subsequently, libraries for N. sibirica, N. ernesti‐mayeri, N. stewartiana, N. grandiflora, and N. rtanjensis were constructed by BGI Tech Solutions (Hong Kong, China) using the DNBSEQ platform and Trinity v2.0.6 to assemble clean reads, as previously described (Popović et al., 2025). Libraries for N. nuda and N. nervosa were constructed by Macrogen Europe BV (Amsterdam, the Netherlands) using the Illumina NovaSeq 6000 platform (Illumina, USA). Clean reads were assembled de novo into longer fragments using Trinity software. TGICL v2.0.6 was used to cluster transcripts, reduce sequence redundancy, and produce Unigenes. Transcriptomic data, along with corresponding ReadMe files, are deposited in the trusted Institutional Repository RADaR (https://radar.ibiss.bg.ac.rs/) with the following handle IDs: N. rtanjensis (https://radar.ibiss.bg.ac.rs/handle/123456789/7507), N. grandiflora (https://radar.ibiss.bg.ac.rs/handle/123456789/7552), N. nervosa (https://radar.ibiss.bg.ac.rs/handle/123456789/7520), N. nuda (https://radar.ibiss.bg.ac.rs/handle/123456789/7524), N. sibirica (https://radar.ibiss.bg.ac.rs/handle/123456789/7553), N. ernesti‐mayeri (https://radar.ibiss.bg.ac.rs/handle/123456789/7555), and N. stewartiana (https://radar.ibiss.bg.ac.rs/handle/123456789/7554).

The transcriptomic database was searched to identify transcripts of iridoid pathway‐related genes (GPPS, GES, G8H, 8HGO, PRISE/ISY, NEPSs, and MLPLs) based on homology to genes characterized in the Nepeta genus (Lichman et al., 2019a, 2020; Hernández Lozada et al., 2022; Aničić et al., 2020, 2024) using a local BLASTp search. Candidate transcripts were selected based on high sequence similarity to these reference sequences, with hits chosen for optimal alignment scores and functional consistency with the iridoid pathway. All selected transcripts showed high similarity to the sequences of homologous genes characterized in other species present in the NCBI database, confirming their annotations. For the NCBI accession numbers of the genes analyzed in this study, see Table S12.

Pairwise sequence identity analysis

Pairwise sequence identities were calculated to assess the similarity among the nucleotide sequences obtained in this study and between these sequences and functionally validated reference genes. Multiple sequence alignment was performed in MEGA (Molecular Evolutionary Genetics Analysis, version X) using the MUSCLE algorithm with default parameters. After alignment, pairwise distances were computed using the p‐distance model with pairwise deletion of gaps and missing data. Pairwise identity values were then calculated as the complement of the p‐distance (Identity = 1 − distance) and expressed as percentages. The resulting identity matrix included both the sequences generated in this study and validated reference genes, providing an overall measure of sequence similarity.

RNA extraction and qPCR profiling of iridoid‐related biosynthetic genes

Total RNA was isolated from leaves of seven selected Nepeta taxa using the procedure described earlier. After DNase I treatment (Thermo Fisher Scientific, Waltham, MA, USA) for 30 min at 37°C, reverse transcription was performed using the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, USA) according to the manufacturer's instructions, using oligo(dT) primers. For subsequent qPCR analysis, specific primer pairs were designed using Primer3Plus software (http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi) (Table S13). Gene expression was measured by real‐time PCR using the QuantStudio 3 Real‐Time PCR System (Thermo Fisher Scientific, Carlsbad, CA, USA). Thermocycling conditions followed Aničić et al. (2018). Reactions were set up with Maxima SYBR Green/ROX Master Mix (2×) (Thermo Fisher Scientific, USA), using cDNA equivalent to 50 ng of RNA and 0.3 μM primers, following the manufacturer's protocol. Results were analyzed using QuantStudio Design and Analysis Software version 1.4 (Thermo Fisher Scientific, USA). Expression levels of iridoid biosynthesis‐related genes were calculated using the 2^−ΔΔCT^ method (Schmittgen and Livak, 2008), with GAPDH as the reference gene. All analyses were performed in three biological replicates. Data were visualized using Past 4 software (version 4.17). Before constructing the heat map (Figure 4B), gene expression data for each sample were independently scaled between min and max values.

Reconstitution of nepetalactone production in chemotype B N. grandiflora leaves

Cloning

Coding sequences for NrNEPS1, identified in the N. rtanjensis transcriptome (Aničić et al., 2020), and NcNEPS5 from N. cataria (Lichman et al., 2020) were assembled into expression constructs using the Modular Cloning (MoClo) toolkit (Addgene, Watertown, MA, USA) and the Plant MoClo toolkit (Weber et al., 2011; Engler et al., 2014). The coding sequences were commercially synthesized (Twist Bioscience, San Francisco, CA, USA) with synonymous mutations introduced to remove BpiI, BsaI, BsmBI, and SapI recognition sites, while overhangs containing BpiI (BbsI) sites were added and codons were optimized for expression in Nicotiana benthamiana. The resulting amplicons were cloned into pUAP1 (Addgene #63674) to create level 0 parts (Dudley et al., 2022). These level 0 parts were then assembled into level 1 acceptors (pICH47742 from the MoClo Toolkit) in a one‐step cloning reaction with level 0 parts encoding the CaMV 35S promoter (CaMV35S) (pICH41388 from the MoClo Plant Parts Kit), the 5′ UTR from the tobacco mosaic virus (TMV) (pICH41414 from the MoClo Plant Parts Kit), and a synthetic chloroplast transit peptide sequence (pICH78133 from the MoClo Plant Parts Kit) to ensure chloroplast localization of the NEPS enzymes. According to the subcellular localization prediction tool, NEPS sequences are localized in plastids (https://www.genscript.com/wolf-psort.html). A separate construct containing the P19 suppressor of gene silencing from the Tomato Bushy Stunt Virus (TBSV) was assembled as described by Dudley et al. (2022).

Agroinfiltration procedure

Agrobacterium tumefaciens strain LBA4404 was transformed with 300 ng of plasmid DNA using the electroporation method with the Gene Pulser Electroporation System (Bio‐Rad, Hercules, CA, USA). Transformed cells were plated on YEB agar containing selective antibiotics: ampicillin (100 mg L^−1^), streptomycin (100 mg L^−1^), and rifampicin (200 mg L^−1^), and then incubated at 28°C for 2 d. Successful transformation was confirmed by colony PCR. A single colony of recombinant Agrobacterium was cultured in 5 mL of liquid YEB medium with the same antibiotics and incubated overnight at 28°C with shaking at 230 rpm. The next day, 1 mL of bacterial culture was transferred to 10 mL of liquid YEB medium containing 10 mM MES‐KOH (pH 5.5) and 200 μM acetosyringone, and incubated overnight at 28°C with shaking at 230 rpm. Bacterial cells were collected by centrifugation at 3,000 g for 20 min and resuspended in infiltration medium (10 mM MES, 10 mM MgCl_2_, and 200 μM acetosyringone) to a final OD_600_ of 0.5. After a 4‐h incubation at room temperature, the cultures were infiltrated into the abaxial surface of leaves from 4‐week‐old N. grandiflora plantlets using a blunt‐tipped plastic syringe with gentle pressure. Leaves were infiltrated with NrNEPS1 or NcNEPS5 together with p19 bacterial suspensions in a 1:1 ratio. The control group leaves were infiltrated with Agrobacterium containing only p19 constructs. Two weeks after agroinfiltration, N. grandiflora leaves were harvested and immediately frozen in LN. Metabolite extraction was performed as described above. Both GC/MS and UHPLC/DAD/(±)HESI‐MS^2^ profiling were conducted. The presence of cis,trans‐nepetalactone in samples was confirmed based on the MS, MS^2^, and DAD spectra, as well as by comparison with the standard.

DNA extraction and amplification of genomic ITS and plastid loci

DNA extraction

Total genomic DNA was isolated from fresh leaf samples using the GenUP™ Plant DNA Kit (Biotechrabbit GmbH, Berlin, Germany) following the LYSIS LC protocol, according to the manufacturer's instructions. The concentration and purity of DNA isolates were assessed spectrophotometrically (NanoPhotometer N60, IMPLEN, München, Germany) and fluorometrically (Qubit 3.0 fluorometer, Thermo Fisher Scientific, Waltham, MA, USA). DNA integrity was also checked by gel electrophoresis.

Amplification of genomic ITS and plastid loci

PCR amplifications were carried out in a final volume of 25 μL using an Eppendorf Mastercycler nexus gradient thermal cycler (Eppendorf AG, Hamburg, Germany). The primer sequences and PCR conditions for the trnL–F, rbcL, and matK plastid loci as well as for the ITS nuclear region are provided in Table S14. The amplification reaction for the three plastid loci included 50 ng of template DNA, 12.5 μL of Dream Taq Green PCR Master Mix (2×) (Thermo Fisher Scientific, Karlsruhe, Germany), 0.2 μM each of forward and reverse primers, and 1.25 μL of 2% DMSO to reduce secondary structure effects and improve primer binding to the functional copies of the marker sequences (Jamzad et al., 2003). Amplification of the ITS region in N. parnassica and N. sibirica was performed using the primers ITS Leu1 and ITS2 (White et al., 1990; Andreasen et al., 1999), which amplified only the ITS1 region. In the remaining species, the entire ITS region (ITS1 + ITS2) was amplified using the primers ITS 17SE and ITS 26SE (Sun et al., 1994).

The amplified PCR products were purified using standard absorption, washing, and elution DNA buffers, along with EconoSpin® DNA spin columns (Epoch Life Science, Inc., Missouri City, TX, USA), following the manufacturer's instructions. Purified PCR products were run on 1% agarose gels stained with ethidium bromide to assess amplicons' quality. The gels were visualized with a UV transilluminator (ST4 3026‐WL/26 M, Vilber Lourmat, Collégien, France), and a 100 bp DNA Ladder Plus (Thermo Fisher Scientific, Karlsruhe, Germany) was used to confirm their length.

Sanger sequencing was performed by Macrogen (Seoul, South Korea). The results were visualized and analyzed using Chromas (version 2.6.6, http://technelysium.com.au/wp/). A total of 40 sequences were generated and submitted to GenBank (accession numbers are provided in Table S15).

Reconstruction of phylogenetic relations

Sequence alignment and phylogenetic analyses were conducted using Molecular Evolutionary Genetics Analysis (MEGA), version 11.0.13. Alignment was performed with the MUSCLE algorithm using default settings. For phylogenetic analysis of the aligned sequences, the UPGMA method and the Tamura 3‐parameter model (the model with the lowest Bayesian Information Criterion value) were used. Bootstrapping was conducted to assess the statistical support of branching, with 1,000 bootstrap replicates.

The analysis included nine new ITS region sequences (for N. cataria, N. ernesti‐mayeri, N. grandiflora, N. laevigata, N. nervosa, N. nuda, N. parnassica, N. rtanjensis, and N. sibirica) generated in this study and 72 Nepeta taxa sequences from the NCBI database (Table S7). Representatives of the subclade Nepetoideae other than Nepeta (Salvia, Agastache, Dracocephalum, Schizonepeta, Hyssopus, Lavandula, Perilla, Melissa, Mentha, Thymus, and Origanum) as well as the subclades Lamioideae (Lamium and Marrubium) and Ajugoideae (Ajuga), were also included in the analysis as outgroups.

CONFLICTS OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

T.B., D.Ma., and M.M. contributed equally. T.B. performed phylogenetic data acquisition and analysis. D.Ma. conducted transcriptomic data mining and bioinformatics analyses. M.M. led transient gene expression experiments. N.P. and L.P. conducted gene‐cloning. U.G. led metabolomics, and M.S. led gene expression analyses. B.Š. supervised the phylogenetic analyses. T.L. assisted with phylogenetic data acquisition and analysis. A.S. and M.T. performed transient gene expression experiments and assisted with metabolomics sample preparation. S.D., B.F., and J.N.Ž. contributed to metabolomics, gene expression analyses, and cultivation of plants. J.B. contributed to gene expression profiling and cloning. D.Mi. acquired funding, administered the project, and led conceptualization and supervision of the research. D.Mi., D.Ma., T.B. and U.G. generated data visualizations. D.Mi., D.Ma, U.G., and B.Š. drafted the original manuscript. All authors revised and approved the final version.

Supporting information

Additional Supporting Information may be found online in the supporting information tab for this article: http://onlinelibrary.wiley.com/doi/10.1111/jipb.70125/suppinfo

Figure S1. Representative UHPLC/DAD chromatograms at λ_max_ = 225 nm of 10 studied Nepeta taxa Table S1. Peak areas obtained from full‐scan MS spectra for identified iridoid glycosides (1–13) and aglycones (14–26) Table S2. GC/MS profiling of VOCs in methanol extracts of leaves of 10 Nepeta accessions Table S3. Similarity/identity matrix (%) for pairs of GPPS genes from Nepeta taxa analyzed in this study and publicly available sequences Table S4. Similarity/identity matrix (%) for pairs of GES genes from Nepeta taxa analyzed in this study and publicly available sequences Table S5. Similarity/identity matrix (%) for pairs of G8H genes from Nepeta taxa analyzed in this study and publicly available sequences Table S6. Similarity/identity matrix (%) for pairs of 8HGO genes from Nepeta taxa analyzed in this study and publicly available sequences Table S7. Similarity/identity matrix (%) for pairs of PRISE/ISY genes from Nepeta taxa analyzed in this study and publicly available sequences Table S8. Similarity/identity matrix (%) for pairs of NEPS genes from Nepeta taxa analyzed in this study and publicly available sequences Table S9. Similarity/identity matrix (%) for pairs of MLPL genes from Nepeta taxa analyzed in this study and publicly available sequences Table S10. Literature data on the iridoid composition of Nepeta taxa included in the phylogenetic analysis presented in  Figure 6

Table S11. List of Nepeta taxa included in this study, with data on seed source and distribution Table S12. Accession numbers of sequences deposited in the NCBI Table S13. Primer sequences used for qPCR analysis Table S14. Primer sequences and PCR conditions used for DNA barcoding Table S15. GenBank accession numbers for DNA sequences used in this study to reconstruct phylogenetic relationships within the genus Nepeta, as presented in  Figure 5