Rhizomicrobiomes from Drought-Adapted Mediterranean Species Differently Alter Leaf Metabolome of Rosmarinus officinalis L. Under Reduced Water Availability

Abstract

Rosmarinus officinalis L. is known for its drought tolerance; however, its growth is adversely affected by both mild and severe water stress. This study investigates the potential of rhizomicrobiome (RM) transplantation to strengthen water stress resilience. Three RMs derived from native plants—R. officinalis (RO), Pistacia lentiscus L. (PL), and Juniperus phoenicea L. (JP)—collected from a semi-arid Mediterranean garrigue were inoculated into R. officinalis subjected to severe drought stress for 30 days. Although RM transplantation did not result in an increase in biomass, it led to the accumulation of intermediates within the phenylpropanoid/coumarin pathway and significant source-specific alterations in other leaf metabolites. Specifically, PL-RM increased the abundance of lignans and stress signaling metabolites. JP-RM improved the root-to-shoot ratio and the sugar and sugar-alcohol accumulation in leaves. Both JP-RM and RO-RM treatments reduced the abundance of abscisic acid, cyclic GMP, and purine recycling pathways. Additionally, RO-RM decreased the abundance of fifteen defense-related metabolites, indicating a direct interference of the inoculum with the plant immune system. In conclusion, these findings suggest that targeted RM manipulation can be an effective strategy to modulate R. officinalis leaf metabolism.

1. Introduction

Drought causes a reduction in soil water availability, often associated with heat stress, thus representing a major abiotic stressor and primary driver of reduced vegetation production [1,2,3]. Global climate models and future projections consistently indicate that under the most severe emission scenarios, the end of the century will witness a notable increase in the duration and frequency of drought events in many areas of the world, among which is the Mediterranean region [4]. For instance, climate change is poised to significantly escalate drought risks across the Mediterranean Basin, profoundly impacting the primary productivity of its ecosystems [5,6]. Drought effects in the Mediterranean exhibit a significant spatial heterogeneity, disproportionately affecting coastal hydro-climatic zones with a heightened susceptibility to agricultural droughts [6]. These spatially variable drought conditions impose a unique set of selective pressures on regional flora, driving the evolution of diverse adaptive responses. Plants inhabiting the Mediterranean Basin have evolved a suite of strategies to prevent soil water deficit and water stress, and/or avoid and tolerate damage caused by water stress [7,8]. Soil water deficit can be avoided by increasing root depth and density to enhance water uptake from deeper soil layers [7,9,10], decreasing the leaf surface while increasing leaf thickness or leaf shedding to reduce evapotranspiration [11,12,13], or increasing the root-to-shoot ratio (R/S) [14]. Moreover, water stress can be temporarily avoided by increasing the concentration of specific compatible solutes to lower the osmotic potential of the plant cells and maintain turgor and water uptake even under low water potential conditions [13,15]. Another crucial mechanism is stomatal closure, a process regulated by hormonal signals, primarily abscisic acid (ABA), whose levels increase when plants encounter water-limited conditions, leading to systemic changes that protect them against the stress [16]. Drought also impairs the balance between light-dependent and light-independent reactions of photosynthesis. The reduced demand for ATP and NADPH in the Calvin cycle, due to stomatal closure, leads to an over-reduction in the electron transport chain, resulting in the excessive generation of reactive oxygen species (ROS), causing oxidative stress [8]. To avoid damage, plants can modify leaf orientation; dissipate the excess energy through heat dissipation, Mehler reactions or photorespiration [7]; and synthesize various metabolites to scavenge ROS, mitigating their damaging effects on cellular structures [8,17]. Consequently, drought stress alters the activity of key enzymes and the availability of substrates, leading to a shift in the synthesis and demand for both primary and secondary metabolites essential for stress tolerance [18].

Beneficial associations between plants and the diverse and dynamic community of microorganisms residing in, on, and around their tissues—the plant microbiome—can significantly enhance a plant's ability to cope with water deficit [19,20,21,22]. Indeed, the plant microbiome represents a vast reservoir of genetic diversity that critically influences plant health and fitness [23]. The rhizomicrobiome (RM), i.e., the microbial community living on the root surface, vicinity and within the root tissues, is the most abundant and diverse pool of plant-associated microorganisms, among which a restricted group known as plant growth-promoting rhizobacteria (PGPR) [24]. Although the role of plant RM in promoting host growth and health is well-established (for recent reviews, see [25,26,27]), leveraging these communities to enhance plant resilience to climate change presents a major challenge. The application of complex microbial communities, rather than clonal populations, represents a paradigm shift driven by the ability of diverse microbial taxa to coexist and form stable, spatially distinct niches [28]. Microbiome transplantation has been shown to effectively enhance both survival duration and biomass accumulation in tomato plants [29]. Similarly, inoculation of squash plants with soil microbial communities obtained from drought-prone areas increased the prevalence of drought-adapted taxa in a common garden experiment, resulting in elevated chlorophyll and carotenoid levels [30]. Moreover, Abou Jaoudé et al. [22] demonstrated that inoculating Salvia officinalis L. with the RM derived from Juniperus phoenicea L. grown in a dry Mediterranean environment significantly increased root biomass, decreased average leaf area, and increased leaf number, suggesting improved water acquisition and thermoregulatory adaptations.

Plants that thrive in marginal environments, such as the drought-tolerant Mediterranean species, harbor specialized RM [31,32]. Moreover, the RM of medicinal plants in these environments is exceptionally diverse, a richness attributed to the selective pressures exerted by root exudates and secondary metabolites [33]. This high diversity increases the probability of finding rare taxa with unique, beneficial properties. These microbial communities have co-evolved with the environment and their hosts, developing enhanced adaptive capabilities under extreme conditions and representing a promising source of novel biostimulants. Indeed, abiotic stressors inherent to agricultural soils, including fluctuating temperatures, desiccation, and extremes in pH and salinity, can significantly reduce the viability and metabolic activity of these microbial inoculants, limiting their beneficial effects on crop productivity [34].

Rosmarinus officinalis L. (rosemary) is a valuable medicinal and aromatic bushy evergreen shrub belonging to the family Lamiaceae, native to the Mediterranean basin. Rosemary is drought-resistant and can survive under arid conditions at a relative water content of below 35% [35]. Although generally tolerant to water scarcity, severe drought can modify rosemary’s physiological status and the composition of essential oils [35,36,37,38]. Clary et al. [39] and Olmos et al. [40] reported a decrease in total plant biomass production in R. officinalis grown under reduced soil water content and low irrigation compared with high-irrigated plants, which was associated with a rapid decrease in stomatal conductance at lower water potentials.

Advancing effective strategies to mitigate climate change impacts on plant fitness requires a deeper understanding of plant RM assembly and the complex interactions among plants, their associated microbes, and the environment under diverse climatic conditions [41]. Building upon this premise, the present study examines the potential of transplanting RM from R. officinalis L., Pistacia lentiscus L., and J. phoenicea L. to enhance the resilience and productivity of R. officinalis L. under water-limited conditions. The objectives were: (1) to assess the efficacy of RM transplantation in promoting biomass production and allocation of R. officinalis compared to non-inoculated plants and (2) to identify common and specific RM-induced alterations in the metabolite profile that may confer resilience to water deficit. The ultimate goal of this research is to advance the understanding of plant-microbe interactions under extreme environmental stress and to develop effective tools for the early identification of beneficial rhizomicrobiomes, thereby facilitating the discovery of sustainable strategies for climate-resilient agriculture.

2. Results

2.1. Effect of Rhizomicrobiome Transplantation on Biomass Production and Partitioning

RM transplantation did not significantly affect R. officinalis biomass production compared to the non-inoculated plants (Table 1). On DAT30, total plant biomass (PB) was similar in the inoculated and non-inoculated plants, being on average 0.39 ± 0.08 g, with no significant differences in shoot (SB; on average 0.30 ± 0.06 g) and root (RB; on average 0.09 ± 0.02 g) biomass (Table 1).

Table 1.

Shoot biomass (SB), root biomass (RB), plant biomass (PB), and biomass partitioning in above- and belowground (root to shoot ratio; R/S) in the inoculated and non-inoculated R. officinalis L. Plants were inoculated with the RMs from J. phoenicea L. (JP), R. officinalis L. (RO) and P. lentiscus L. (PL). C represents the non-inoculated plants.

	SB		RB		PB		R/S
Mean	g		g		g		g g−1
C	0.26	ab	0.07	a	0.34	a	0.29	a
RO	0.40	a	0.11	a	0.51	a	0.28	a
PL	0.34	a	0.09	a	0.44	a	0.26	a
JP	0.21	b	0.09	a	0.30	a	0.45	b
S.e.
C	0.05		0.01		0.07		0.03
RO	0.12		0.04		0.16		0.05
PL	0.04		0.02		0.06		0.04
JP	0.01		0.01		0.02		0.04

No significant differences were observed in PB and RB between R. officinalis inoculated with RMs from different origins (Table 1). Nevertheless, the source of the rhizomicrobiome affected SB (Table 1). SB was recorded at 0.21 ± 0.01 g in R. officinalis inoculated with JP-RM, significantly lower compared to rosemary inoculated with RO-RM (0.40 ± 0.11 g; p < 0.05) and PL-RM (0.34 ± 0.04 g; p < 0.05) (Table 1).

Biomass partitioning between above- and belowground (R/S) was significantly affected by inoculation. Particularly, R/S was significantly lower (p < 0.05) in the non-inoculated rosemary (0.29 ± 0.03 g g⁻¹) compared to the plants inoculated with JP-RM (0.45 ± 0.04 g g⁻¹; p < 0.05). Moreover, R/S in R. officinalis inoculated with JP-RM was significantly higher compared to rosemary inoculated with RO-RM (0.28 ± 0.05 g g⁻¹; p < 0.05) and PL-RM (0.26 ± 0.04 g g⁻¹; p < 0.01) (Table 1).

2.2. Analysis of R. officinalis Leaf Metabolome Following RM Transplantation

A comprehensive untargeted metabolomic analysis was performed to assess the impact of RM transplantation on the leaf metabolome of R. officinalis L. This analysis identified 351 metabolites belonging to 40 Natural Product (NP) superclasses and 99 distinct NP classes (Figure S1). The most prevalent superclasses included a significant number of flavonoids (n = 42), notably flavones (n = 19), flavonols (n = 10), and flavanones (n = 7). Another prominent superclass was diterpenoids (n = 25), most of which belonged to the gibberellins (n = 8) and abietane diterpenoids (n = 8) classes. Moreover, the superclasses monoterpenoids (n = 25) and phenylpropanoids (n = 20) were other well-represented superclasses (Figure S1).

A comparative analysis of leaf metabolome datasets was performed using principal component analysis (PCA) (Figure 1). The first two principal components accounted for 59.9% of the total variance, with PC1 and PC2 contributing 31.6% and 28.3%, respectively (Figure 1). The PCA revealed a clear separation between the inoculated and non-inoculated plants, and among the various treatments (Figure 1). PC1 distinguished the leaf metabolites of the non-inoculated (C) and RO-RM-inoculated rosemary plants from those inoculated with PL-RM and JP-RM (Figure 1). PC2 further separated the non-inoculated plants and rosemary inoculated with PL-RM from those inoculated with JP-RM and RO-RM (Figure 1).

Figure 1.

Principal component analysis (PCA) score plot of leaf metabolic profiles of the inoculated and non-inoculated R. officinalis L. Plants were inoculated with the RMs from J. phoenicea L. (JP, red), R. officinalis L. (RO, orange), and P. lentiscus L. (PL, green). C (blue) represents the non-inoculated plants.

The most important features distinguishing significant separation among treatments (p ≤ 0.05) are presented in Figure S2. Notably, the abundance of 19-aldoandrost-4-ene-3,17-dione, 18-a-glycyrrhetinic acid, and ABA was significantly higher (p < 0.05) in the non-inoculated C and PL RM-inoculated rosemary compared to the JP- and RO RM-inoculated plants (Figure S2a,f,g). The opposite trend was observed in the abundance of scutellarin and isorhamnetin 3-O-(4-O-p-coumaroyl)-glucoside (Figure S2c,e). Moreover, benzyladenine-9-N-glucoside and macarpine were significantly higher (p < 0.05) in the RO RM-inoculated rosemary compared to the other treatments (Figure S2b,d).

2.2.1. Common Effect of RM Transplantation on R. officinalis Leaf Metabolome Compared to Non-Inoculated Plants (Inoculation vs. Non-Inoculation)

Inoculation with the three rhizomicrobiomes significantly altered (logarithm of fold change [log₂FC] ratio ≤ −1 or ≥ 1, and p value ≤ 0.05) the leaf metabolomic profile of R. officinalis, modifying the abundance of 58 metabolites across 23 superclasses compared to the non-inoculated plants (Table S1).

The most significantly affected superclasses of metabolites were coumarins and phenylpropanoids. Within these superclasses, the abundance of seven metabolites—including the coumarins 6-hydroxymellein, 7-methoxycoumarin, and coumarin, and the phenylpropanoids (E)-3-[2-[(2S,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)oxan-2-yl]oxyphenyl]prop-2-enoic acid, 1-O-feruloyl-β-D-glucose, 1-O-(4-coumaroyl)-β-D-glucose, and trans-2-hydroxycinnamic acid—consistently increased across all three inoculation treatments relative to the non-inoculated plants. Additionally, the other three metabolites exhibiting a common increase in abundance compared to the non-inoculated rosemary included the isoflavonoid 6a,12a-dihydro-6H-[1,3]dioxolo[5,6][1]benzofuro[3,2-c]chromen-3-ol, the lignan (+)-secoisolariciresinol diglucoside, and the phenolic acid vanillin (Figure 2; Table S1).

Figure 2.

Comparative metabolite profiles of the inoculated vs. non-inoculated R. officinalis L. Boxes indicate metabolites with increased (up) or decreased (down) abundance relative to controls (log₂ fold change [log₂FC] ratio ≤ −1 or ≥ 1; p ≤ 0.05). The red and dark green boxes indicate compounds whose relative abundances significantly changed in all the inoculated plants, regardless of the rhizomicrobiome source, compared to the non-inoculated plants. Leaf metabolites whose abundance specifically varied in response to individual rhizomicrobiomes (metabolites reported in black) are shown for J. phoenicea (JP, green), R. officinalis L. (RO, pink), and P. lentiscus L. (PL, orange) rhizomicrobiome-inoculated compared to non-inoculated plants. Metabolites in bold show significant changes in abundance between two specific microbiomes compared to the non-inoculated rosemary: JP and RO (dark green), JP and PL (orange), or RO and PL (violet). Asterisks indicate metabolites that display divergent patterns in PL-RM and RO-RM compared to the non-inoculated plants.

Among the 58 altered metabolites, chrysophanol, which belongs to the polycyclic aromatic polyketides superclass, was the only compound to show a significant decrease under all the tested conditions compared to the non-inoculated plants (Figure 2; Table S1).

The number of metabolites exhibiting increased abundance relative to the non-inoculated plants varied between 17 (RO vs. C) and 22 (PL vs. C), while those showing decreased abundance ranged from 3 (PL vs. C) to 24 (RO vs. C) (Figure 2; Table S1). The proportion of metabolites with increased abundance relative to the total number of altered metabolites was 0.41 (RO vs. C), 0.65 (JP vs. C), and 0.88 (PL vs. C). Additionally, the number of superclasses represented by metabolites with increased abundance compared to the non-inoculated plants ranged from 9 (RO vs. C) to 12 (PL vs. C) (Table S1).

The comparative analysis of leaf metabolomics data demonstrated that the plants inoculated with RO-RM and JP-RM exhibited increased levels of the flavonoids isorhamnetin 3-O-(4\-O-p-coumaroyl)-glucoside, isorhamnetin 3-O-(6\-O-feruloyl)-glucoside, and scutellarin relative to the non-inoculated plants (Table S1). Additionally, the plants inoculated with PL-RM and JP-RM showed a higher abundance of the triterpenoid N-acetyl-S-geranylgeranyl-L-cysteine compared to the non-inoculated controls (Table S1). Furthermore, inoculation with RO-RM and PL-RM resulted in increased levels of the purine alkaloid hypoxanthine relative to the non-inoculated plants (Table S1).

Among the seven metabolites whose abundance was significantly reduced in the plants inoculated with PJ-RM and RO-RM compared to the non-inoculated rosemary, three were purines (guanine, guanosine, and guanosine 3′,5′-cyclic monophosphate), and two were plant regulators classified as gibberellins (gibberellin A4 and gibberellin A19; Figure 2; Table S1). Except for chrysophanol, no metabolite exhibited a significant simultaneous reduction in abundance in the leaf metabolome of the plants inoculated with JP-RM and PL-RM compared to the control group. However, in the plants inoculated with RO-RM and PL-RM, a second metabolite, salidroside, also showed a significant simultaneous decrease relative to the non-inoculated rosemary (Figure 2; Table S1). Furthermore, two sesquiterpenoids ((−)-curcuhydroquinone and artemisinic acid), whose abundance in the RO-RM-inoculated plants decreased, exhibited a higher abundance in the plants inoculated with PL-RM compared to controls (Figure 2; Table S1).

The pathway enrichment analysis revealed that the coumarin biosynthetic pathway was significantly increased in all the inoculated treatments (Table 2). Additionally, the guanine and guanosine salvage pathway was increased in the JP-RM- and RO-RM-inoculated plants compared to the non-inoculated rosemary (Table 2). In contrast, the biosynthetic pathway for (+)-secoisolariciresinol diglucoside was specifically increased in the PL-RM-inoculated rosemary relative to the non-inoculated plants (Table 2).

Table 2.

Comparison of differentially regulated metabolic pathways in rosemary leaves following inoculation with rhizomicrobiomes extracted from J. phoenicea L. (JP), P. lentiscus L. (PL), and R. officinalis L. (RO), respectively. Positive and negative log₂ fold change (log₂FC) values indicate positive and negative alterations of pathways, respectively.

Comp	PlantCyc ID	Pathway Name	Compounds (Total Number and Names)	Pathway Coverage (%)	log2FC Mean	p-Value
JP vs. C	https://pmn.plantcyc.org/pathway?orgid=PLANT&id=PWY-5176, accessed on 5 January 2026.	Coumarin biosynthesis (via 2-coumarate)	3	23.08	8.77	0.053
			(E)-3-[2-[(2S,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)oxan-2-yl]oxyphenyl]prop-2-enoic acid		8.80	0.013
			Coumarin		8.32	0.013
			Trans-2-hydroxycinnamic acid		9.18	0.017
	https://pmn.plantcyc.org/pathway?orgid=PLANT&id=PWY-6599, accessed on 5 January 2026.	Guanine and guanosine salvage II	2	33.33	−1.38	0.057
			Guanosine		−1.36	0.048
			Guanine		−1.39	0.048
RO vs. C	https://pmn.plantcyc.org/pathway?orgid=PLANT&id=PWY-5176, accessed on 5 January 2026.	Coumarin biosynthesis (via 2-coumarate)	3	23.08	8.52	0.053
			(E)-3-[2-[(2S,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)oxan-2-yl]oxyphenyl]prop-2-enoic acid		8.63	0.034
			Coumarin		8.06	0.044
			Trans-2-hydroxycinnamic acid		8.88	0.051
	https://pmn.plantcyc.org/pathway?orgid=PLANT&id=PWY-6599, accessed on 5 January 2026.	Guanine and guanosine salvage II	2	33.33	−1.55	0.057
			Guanosine		−1.60	0.011
			Guanine		−1.51	0.011
PL vs. C	https://pmn.plantcyc.org/pathway?orgid=PLANT&id=PWY-5176, accessed on 5 January 2026.	Coumarin biosynthesis (via 2-coumarate)	3	23.08	8.26	0.053
			(E)-3-[2-[(2S,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)oxan-2-yl]oxyphenyl]prop-2-enoic acid		8.38	0.019
			Coumarin		7.78	0.017
			Trans-2-hydroxycinnamic acid		8.61	0.017
	https://pmn.plantcyc.org/pathway?orgid=PLANT&id=PWY-7487, accessed on 5 January 2026.	(+)-Secoisolariciresinol diglucoside biosynthesis	2	33.33	2.40	0.057
			(+)-Secoisolariciresinol diglucoside		3.60	0.019
			(−)-Secoisolariciresinol		1.20	0.041

2.2.2. Specific Effect of RM Transplantation on R. officinalis Leaf Metabolome

The leaf metabolomic profile of rosemary inoculated with the three RMs also showed specific signatures associated with each RM.

Inoculation with RO-RM resulted in specific and significant alterations in 18 metabolites compared to the non-inoculated controls. Among these, the relative abundances of three metabolites were significantly increased: the amino acid tyrosine, the flavonoid luteolin, and the diterpenoid 3β-hydroxy-9β-pimara-7,15-diene-19,6β-olide (Figure 2; Table S1). In contrast, 15 secondary metabolites exhibited a significant and specific decrease following RO-RM inoculation. These metabolites spanned multiple superclasses, including benzenoids, mono- and diterpenoids, flavonoids and isoflavonoids, lignans, lipids and lipid-like molecules, polycyclic aromatic polyketides, and sesquiterpenoids (Figure 2; Table S1).

Inoculation with PL-RM resulted in specific and significant alterations in the levels of 11 metabolites compared to the non-inoculated controls. Among these, the relative abundances of 10 metabolites, encompassing eight distinct superclasses, were significantly increased (Figure 2; Table S1). Conversely, PL-RM inoculation resulted in a specific decrease in the abundance of a single metabolite, the diterpenoid 19-O-β-glucopyranosyl-steviol, relative to the non-inoculated plants (Figure 2; Table S1).

Inoculation with JP-RM specifically resulted in an increased abundance of five metabolites compared to the non-inoculated plants, including the polyol myo-inositol and the saccharides sucrose, psicose, and trehalose (Figure 2; Table S1). Additionally, JP-RM inoculation led to a specific decrease in the leaf concentrations of two octadecanoids: 2-R-hydroperoxy-linolenate and 9-hydroperoxy-10E,12Z,15Z-octadecatrienoic acid (Figure 2; Table S1).

3. Discussion

3.1. Plant Survival, Biomass Production, and Allocation

The successful survival of all R. officinalis experimental subjects during the period of water deficit constitutes a significant demonstration of the inherent drought tolerance of this species, irrespective of rhizospheric microbiome inoculation. Visual inspection of the plants after 30 days of drought treatment revealed no wilting symptoms in any treatment. Our results indicate that inoculating different RMs did not significantly affect aboveground, belowground, or total rosemary biomass compared with the non-inoculated plants (Table 1), suggesting that the primary effect of inoculation was not a generalized growth promotion. This outcome is not due to the species’ slow growth rate, as a short-duration experiment may not reveal biomass changes. Indeed, comparison among the inoculated plants demonstrated that RMs distinctly modulated aboveground growth. The application of JP-RM resulted in lower shoot biomass, which increased the root-to-shoot ratio (R/S) compared to the non-inoculated plants and to the RO-RM- and PL-RM-inoculated plants (Table 1). High R/S, particularly when coupled with reduced shoot biomass, often serves as a classic physiological indicator of plant stress or a resource-scarcity coping strategy, where carbon allocation is prioritized towards nutrient and water acquisition by the roots [42].

While the plant metabolomic changes associated with the effects of inoculation on plant drought tolerance remain unexplored [43], untargeted metabolomic analysis offers a sensitive method for detecting inductions and changes in the leaf’s short- to medium-term physiological response to both drought stress and microbial inoculation.

In accordance with previous studies on the species [44], flavonoids, monoterpenoids, diterpenoids, and phenylpropanoids were the most represented molecular superclasses (Figure S1). In drought-tolerant species, resilience is bolstered by genes encoding secondary metabolites. For instance, in rosemary, 35 gene families are involved in the biosynthesis of antioxidant components, enhancing the ability of rosemary adaptation to drought, heat and salt stress [45]. In these species, diminished growth rates result from a shift in carbon flux: photosynthetic assimilates are invested in supporting the metabolic costs of secondary metabolite production [46]. Given limited plant resources, the production of these metabolites diverts substantial amounts of substrates from primary to secondary metabolism, inducing a trade-off between growth and defense [47].

3.2. Metabolic Signatures

In our study, RM transplantation elicited significant metabolic changes in rosemary, leading to distinct clustering of the treatments in the PCA score plot (Figure 1). The common metabolic alterations observed following all three RM inoculations were primarily associated with two major superclasses: coumarins and phenylpropanoids (Figure 2; Table S1). Notably, the fold changes for these compounds were similar across treatments (Table S1), suggesting that RM transplantation induces consistent and specific metabolic modifications in the recipient plant. This shared increase in abundance indicates that inoculation commonly triggers a positive alteration of the phenylpropanoid biosynthetic pathway (Table 2), supporting the role of these molecules as mediators in plant-microbiome signaling [48]. As known phytoalexins, coumarins contribute significantly to the plant’s defense system by conferring resistance against microbial pathogens [49]. In previous studies, plant-beneficial microorganisms emerged as potent activators of the plant immune system, priming the whole plant body for enhanced defense against a broad range of pathogens and herbivores without directly activating costly defenses [50,51]. Moreover, the common decrease in the relative abundance of many molecules with antimicrobial activity (chrysophanol, a molecule characterized by antiviral, antifungal and antibacterial activities [52], and other defense-related metabolites such as glycyrrhetinic acid [53] and guanosine 3′,5′-cyclic monophosphate [54] in the RO- and JP-RM-inoculated plants; Figure 2; Figure S2; Table S1), strongly support a direct interference of the inoculated rhizomicrobiomes with the plant defense system [50]. Phenylpropanoids are also known to protect plants against oxidative stress and to modulate cellular functions [47,55]. Drought is one of the most abiotic stressors, drastically restricting plant growth by directly affecting the photosynthetic process and both light-dependent and light-independent reactions. Guo et al. [56] found higher levels of many phenolics (homovanillic acid, gallic acid, vanillin, 3,4-dihydroxybenzaldehyde, 3-dimethylallyl-4-hydroxymandelic acid, and kynurenic acid) in the drought-tolerant wheat genotype HX10 compared to the drought-sensitive genotype YN211, under both well-watered and drought conditions, with concentration increasing under water restriction. In the drought-tolerant Salvia milthiorriza, an increase in phenylpropanoid, flavonoid, and diterpenoid biosynthesis was observed in response to drought stress [57]. Furthermore, in transgenic Arabidopsis, the constitutive expression of GmF6′H1, encoding the enzyme catalyzing the biosynthesis of coumarin, increased salt tolerance by promoting endogenous production of the metabolite, which increased the uptake and utilization of plant nutrients, regulated the stability of the cell membrane, and reduced water loss by transpiration under stress conditions [58]. These observations indicate that inoculation leads to altered accumulation of specific phenylpropanoids. It is hypothesized that these variations enhance the resistance of the photosynthetic apparatus to oxidative damage and improve water-use efficiency by promoting more effective physiological and osmotic regulation.

3.2.1. Common Metabolic Signatures Between RMs Compared to Non-Inoculated Plants

In the JP- and RO-RM-inoculated rosemary, the reduced relative abundance of the phytohormone ABA and the androgen steroidal hormone 19-aldoandrost-4-ene-3,17-dione, two most important features driving the metabolome clustering in the PCA (Figure S2) and the decreased abundance of octadecanoids (products of unsaturated fatty acid (PUFA) oxidation derived predominantly from C 18-fatty acids [59]; Table S1), could indicate microbial influence on hormone catabolism, or translocation, and modifications in the compositions of membranes in response to inoculation. Indeed, ABA is known to orchestrate key physiological processes, including developmental pathways (e.g., seed maturation, dormancy) and responses to abiotic stress [60,61]. These regulations involve inducing the transcription of various stress-adaptive genes, leading to stomatal closure, leaf senescence, and overall growth suppression [61,62]. In a study comparing two switchgrass genotypes with contrasting drought tolerance, Tiedge et al. [63] found an increase in leaf ABA abundance in both genotypes grown under water restriction, with a higher increase in the non-tolerant genotype. Munné-Bosch et al. [36] in their analysis of the response of R. officinalis to drought conditions, reported reductions in leaf hydration, photosynthetic rates, and chlorophyll content. These changes were associated with decreased stomatal conductance, which was induced by an increase in ABA concentration. The rapid stomatal closure during drought, low rates of cuticular water loss, and a relatively slow response to rewatering are important mechanisms that determine drought tolerance in R. officinalis, allowing the species to survive the harsh natural conditions typical of the Mediterranean region [39,64]. Moreover, mammalian steroid hormones, such as 19-aldoandrost-4-ene-3,17-dione, can regulate various processes associated with plant growth and development and their interactions with the environment [65]. Janeczko and co-author [66] reported a positive effect of 4-androstene-3,17-dione on photosynthetic rates and stomatal conductance following rewatering of drought-stressed soybean plantlets. Furthermore, PUFA are essential components of biological membranes, contributing significantly to their structural integrity and fluidity, and their oxygenated derivatives (oxylipins) serve as bioactive metabolites (among which is jasmonic acid) that modulate various signal transduction pathways, influencing diverse cellular processes [67]. In addition, the downregulation of guanosine and the guanine salvage II pathway (Table 2), a pathway that typically serves as an efficient mechanism to recycle purine bases, bypassing the high energy costs associated with de novo nucleotide biosynthesis, likely reflects altered signaling dynamics or modifications in resource reallocation in response to inoculation [68]. Ji et al. [69] found that inoculation of Glycyrrhiza uralensis with the PGPR Bacillus pumilus G5 inhibited the gibberellic acid signal transduction pathway under drought stress while enriching flavone and flavonol biosynthesis compared to the non-inoculated plants. In our study, a similar response was observed, suggesting that the plant may be reducing its investment in primary growth to allocate resources toward the synthesis of specific metabolites (Figure 2; Figure S2; Table 2).

PL-RM and RO-RM transplantation elicited distinct changes in the abundance of specific metabolites compared to the non-inoculated plants. Notably, PL-RM inoculation led to an increased abundance of certain sesquiterpenoids in the leaves, including (−)-curcuhydroquinone and artemisinic acid (Figure 2; Table S1). In contrast, these metabolites, and a third sesquiterpenoid, farnesal, exhibited decreased abundance in the RO-RM-inoculated rosemary compared to the control group (Figure 2; Table S1). Sesquiterpenoids play an important role in protecting plants’ extracellular environment from the damaging effects of oxidants [70], possibly highlighting the opposite effects of inoculation on the redox state of the rosemary plants inoculated with the two distinct microbiomes. Moreover, the common increase in the relative abundance of hypoxanthine in the PL- and RO-RM-inoculated plants compared to the control, which is involved in various metabolic processes that help regulate cellular energy balance, prevent oxidative damage, and maintain cellular redox homeostasis [71], together with the downregulation of purines, highlights the direct effect of inoculation in altering the metabolism of these molecules. N-acetyl-S-geranylgeranyl-L-cysteine (AGGC), known to act as a competitive substrate for methylation by isoprenylcysteine carboxyl methyltransferase, which plays a critical role in the post-translational modification of prenylated proteins [72], was the only feature whose abundance commonly increased in both the PL-RM- and JP-RM-inoculated plants compared to the non-inoculated plants [72]. This finding suggests that inoculation with these two RMs elicits stress sensitization, which may enhance responsiveness to environmental stressors.

3.2.2. RM-Specific Metabolic Signatures Compared to Non-Inoculated Plants

Clustering analysis revealed that the leaf metabolomic profiles of the plants inoculated with RO-RM and JP-RM exhibited more pronounced alterations compared to those inoculated with PL-RM, whose metabolomic profile closely resembled that of the non-inoculated rosemary. Nevertheless, PL-RM inoculation also induced distinct changes in the leaf metabolomic profile relative to the non-inoculated plants. Notably, the relative abundance of (+)-Secoisolariciresinol diglucoside (SDG), a precursor of various lignans, was significantly increased in the PL-RM-inoculated rosemary leaves. This metabolite is known to accumulate under stress conditions, thereby enhancing plant resistance to adverse environmental factors such as drought [73]. Under Ca²⁺ stress, SDG was observed to trigger the ROS-scavenging system, thereby improving resistance in cotton plants [74]. Furthermore, the increased abundance of trans-ferulic acid and sinapaldehyde—phenylpropanoids involved in the lignin biosynthetic pathway—in the plants inoculated with PL-RM indicates a metabolic shift toward enhancing vessels’ structural resistance to cavitation. This metabolic adjustment complements the chemical defense provided by the increased accumulation of sesquiterpenoids (Table S1; Figure 2). Under this treatment, an increase in the abundance of shingosine-1-phosphate, a calcium-mobilizing compound active in the signal-transduction pathway linking ABA to reductions in stomatal aperture [75], was observed compared to the non-inoculated plants. The PL-RM-inoculated plants also invested in direct defense against herbivores, showing increased abundance of neoabietic acid, a resin acid whose concentration has been reported to decrease in Pinus sylvestris under drought stress, known to reduce palatability and seal wound sites [76].

RO-RM inoculation exclusively showed an increase in the abundance of three metabolites, including the amino acid tyrosine and the flavonoid luteolin. Tyrosine is produced through the shikimate pathway. Its accumulation under drought stress was observed in the xerohalophyte Salvadora persica during adaptation to water deficit [77]. In maize seedlings, exogenous application of tyrosine reduced oxidative stress and increased leaf water content, photosynthesis, and stomatal conductance [78]. Inoculation with the RO-RM also resulted in the most significant number of exclusively metabolites that showed a decreased abundance compared to the non-inoculated plants (15 compounds; Figure 2), some of which (kaempferide 3-O-glucoside and kaempferide 3-O-β-D-glucopyranosyl-(1->2)-O-α-L-rhamnoside) known to play a role in protecting plants from external environmental stresses [79], and (the isoflavonoid (−)-medicarpin-3-O-glucoside) to increase in plants subjected to pathogen attack [80]. These findings indicate a direct influence of the inoculated rhizomicrobiomes in eliciting plant responses to both abiotic and biotic stresses.

JP-RM inoculation specifically showed an increase in the abundance of five metabolites, including the polyol myo-inositol and the saccharides sucrose, psicose, and trehalose (Table S1). Saccharides such as trehalose and sucrose act as osmolytes, helping lower the osmotic potential of plant cells, a key mechanism for avoiding water stress [46,55]. In a study of contrasting Brassica juncea genotypes, Sheoran et al. [43] found that both PGPR inoculation and drought independently triggered the upregulation of glucose, trehalose, arabinose, and myo-inositol in the leaves of the drought-tolerant genotype compared to non-inoculated and to well-watered plants, respectively, improving their stress resilience and overall growth. These results indicate that an increase in saccharide and polyol concentrations in the JP-RM-inoculated rosemary may have provided a further metabolic safeguard to mitigate the damaging effects of osmotic stress compared to the non-inoculated rosemary.

4. Materials and Methods

4.1. Soil Sampling and Rhizomicrobiome Extraction

Rhizospheric soil samples were collected in the Nature Park Porto Conte (Sardinia, Italy) during late summer 2023. Three samples were extracted using a soil corer under the canopy of three species typical of the Mediterranean maquis: R. officinalis L. (RO), J. phoenicea L. (JP), and P. lentiscus L. (PL). The soils were immediately refrigerated during transportation and sieved at 2 mm in the laboratory. The three soil samples per species were pooled and finally stored at −20 °C until further use. To reactivate the microbial cell metabolism, fifty grams of the stored soils were hydrated with 12–15 mL of 1:10 diluted Luria–Bertani broth (LB, 10 g L⁻¹ tryptone, 5 g L⁻¹ yeast extract, and 10 g L⁻¹ NaCl) and incubated at 25 °C for approximately 72 h. The soil was suspended in 125 mL 0.1% peptone water (10 g L⁻¹ peptone and 5 g L⁻¹ NaCl) and blended for 2 min at speed 1 with a BagMixer 400S (Interscience, Puycapel, France) to detach the microorganisms from the soil particles. The soil extract was first centrifuged at 1000 rpm for 10 min to remove the soil sediments. To obtain the cellular pellet, the liquid phase was collected under sterile conditions, centrifuged at 10,000 rpm for 10 min and the supernatant was discharged. The cellular pellet was then resuspended in 1 mL peptone water amended with glycerol 20% and stored at −80 °C until further use.

4.2. Plant Experimental Setup

Seeds of R. officinalis L. (Semillas Batlle S.A., Barcelona, Spain) were surface sterilized by 10 min immersion in sodium hypochlorite (NaClO) solution (50% v/v) amended with Twin 20 (0.025% v/v) and then rinsed 10 times with sterile deionized water. The sanitized seeds were sown on a sterile substrate composed of two parts peat and one part vermiculite in a germination chamber and grown for 11 days. The temperature of the chamber was maintained at 26 °C, the photon flux density was set at 200 μmol photons m⁻² s⁻¹ and the photoperiod was 14/10 h light/dark. Twenty seedlings of uniform size (5 cm total length) were then selected and divided into four groups. Each group of five seedlings was transplanted into 55 L growth pots containing 40 L of a sterile substrate (peat: perlite: vermiculite, 2:1:1 v/v/v). The pots were placed in a grow tent (Mars Hydro EU, Ginsheim-Gustavsburg, Germany) equipped with a lamp Mars Hydro Smart FC 3000 Samsung LED Grow Light powered by Samsung LM301B LED. The temperature was set to 26 °C, and a ventilation system (DF150A, Inline Duct Fan, Mars Hydro EU) guaranteed an air exchange in the tent. The photon flux density was set to 350 μmol photon m⁻² s⁻¹ while the photoperiod was kept unchanged. The plantlets were left to adapt to the new conditions for 24 h, then light intensity was increased to 450 μmol photon m⁻² s⁻¹ and kept constant until the end of the experiment. After the adaptation period, each pot was irrigated with 4 L of sterilized deionized water, added with 0.1% (v/v) 0.2 μm filtered nutrient solution A (B’cuzz, Atami B.V., Rosmalen, The Netherlands) containing K₂O 4.7%, CaO 3.8%, MgO 1.3%, SO₃ 0.11%, Fe 0.04% and N 4.9% (calcium and ammonium nitrate salts); 0.1% (v/v) filtered nutrient solution B (B’cuzz, Atami B.V., Rosmalen, The Netherlands) containing P₂O₅ 4.1%, K₂O 5.71%, B 0.01%, Mn 0.03%, Mo 0.001%, Zn 0.039%. After three days, the plants were divided into four groups of five plants each and inoculated according to the protocol reported below. The relative soil water content was 25%, and all the pots were irrigated once a week with 1 L of nutrient solution to keep the relative soil water content stable at 25% over time.

4.3. Preparation of the Plant Inoculum and Inoculation

To prepare the plant inoculum, aliquots (250 µL) of the frozen rhizomicrobiomes were spotted on a sterile membrane (0.22 µm) placed onto an agar plate containing M9 medium [81] amended with yeast extract (0.05% w/v). The cells were let to grow at 28 °C for 16–18 h and recovered from the filters by washing with sterile water. The resulting cell suspension was diluted to an optical density at 600 nm (OD₆₀₀) of 1 and used to inoculate plants as follows: Five pots were assigned to one of four treatments. At DAT1 (Day After Treatment—inoculation), one group of plants (control—C) was not inoculated, and plants were irrigated with a volume of sterile water comparable to the inoculum received by treated plants. The plants belonging to the other three groups were respectively inoculated with one of the three RMs (RO, JP, and PL). A total of 10 mL of the diluted (OD₆₀₀ = 1) cell suspension was applied at the base of each stem, above the root junction.

4.4. Plant Biomass

At DAT30, the plants were removed from the soil and separated into shoot and root fractions. The fresh weight of the two plant portions was recorded (RFW: root fresh weight; SFW: shoot fresh weight). The total fresh weight (TFW) was calculated as the sum of RFW and SFW. The roots and shoot subsamples were subjected to oven drying using a Sartorius MA 100 moisture analyzer (Göttingen, Germany). The root biomass (RB) was determined by weighing roots after desiccation. There were no differences in shoot water content (74%) among treatments. Therefore, the shoot biomass (SB) was estimated by subtracting the average leaf water content from SFW. The plant biomass (PB) was determined by summing RB and LB. Finally, the root-to-shoot ratio (R/S) was determined as the fraction of RB to LB.

4.5. Leaf Metabolite Profiling

The remaining shoot subsamples were frozen in liquid nitrogen and used for the metabolomic analysis, which was conducted at oloBion SL (Barcelona, Spain). The sample preparation involved homogenization with cold methanol using a grinder, followed by the addition of methyl tert-butyl ether (MTBE) containing an internal standard. The mixture was shaken and centrifuged to separate the phases. For the metabolomic profiling, two aliquots of the bottom aqueous phase were collected for complementary analytical platforms. The first aliquot was evaporated to dryness and reconstituted in an acetonitrile/water mixture containing internal standards. After shaking and centrifugation, the sample was analyzed using hydrophilic interaction liquid chromatography (HILIC). The second aliquot was mixed with a 1:1 isopropanol/acetonitrile solution, shaken, centrifuged, evaporated, and resuspended in 5% methanol with internal standards. Following centrifugation, the sample was analyzed using reversed-phase liquid chromatography (RPLC). Chromatographic separation of polar metabolites was performed using Waters ACQUITY UPLC BEH Amide (Waters Corporation, Milford, MA, USA) and ACQUITY UPLC HSS T3 (Waters Corporation, Milford, MA, USA) 50 mm columns, both maintained at 45 °C. The system was interfaced with an Agilent 1290 Infinity UHPLC (Agilent Technologies Inc., Santa Clara, CA, USA) and a ZenoTOF 7600 mass spectrometer (Sciex, Framingham, MA, USA). Sample injections were analyzed in both positive and negative electrospray ionization (ESI) modes to maximize metabolite coverage. The analysis was run on three independent extractions per treatment (n = 3).

Raw mass spectrometry data were preprocessed using oloMAP (oloBion SL, Barcelona, Spain), a proprietary AI-driven platform for high-throughput data processing and high-confidence metabolite annotation. It utilizes an in-house adapted version of MS-DIAL 4.9 for automated molecular feature detection and alignment, followed by a specialized multi-stage compound annotation pipeline designed to enhance identification accuracy across diverse biomaterials. Compound identification entails matching measured accurate masses, retention times, and MS/MS fragmentation patterns to validated reference libraries. Metabolite annotation employs a custom spectral database comprising 1.3 million spectra corresponding to over 60,000 unique chemical entities, each enriched with metadata on biological origin and known associations with microorganisms and the exposome. To improve annotation confidence and reliability, oloMAP integrates machine learning models trained to predict retention times for the full metabolite library, which are used to filter and rank candidate matches. Finally, statistical analyses were performed using oloMAP Portal, an integrated data visualization and interpretation platform for in-depth statistical and biological exploration of omics data.

4.6. Statistical Analysis

To test the effect of each rhizomicrobiome inoculation on plant biomass, mean values for each treatment were subjected to a Kruskal–Wallis test (www.socscistatistics.com). Pairwise comparisons between groups were conducted using Dunn’s test, with statistical significance determined at p ≤ 0.05. For the leaf metabolomic profile, significant differences were revealed after applying the Benjamini/Hochberg FDR correction. More information about the analysis can be found in the figure caption.

5. Conclusions

In conclusion, our results demonstrate that the transplantation of root-associated microbiomes from drought-prone environments significantly alters the metabolic and physiological states of recipient rosemary plants. The microbiome-induced changes in leaf metabolite profiles highlight the potential of targeted rhizomicrobiome manipulation as an effective strategy to modulate the metabolism of R. officinalis under water-limited conditions. Further investigation is required to elucidate the mechanistic basis and long-term implications of these interactions. The integration of complementary omics approaches, such as RNA-Seq and proteomics, would facilitate the detection of subtle physiological changes across treatments. Moreover, comparative metagenomic analyses employing metagenome-assembled genomes (MAGs) could provide deeper insights into the functional capabilities and ecological specializations of microbial communities in inoculated versus non-inoculated plants. Collectively, these methodologies would contribute to a more comprehensive understanding of the cumulative effects of microbiome transplantation on plant fitness.

Abbreviations

The following abbreviations are used in this manuscript:

RM	Rhizomicrobiome
PGPR	Plant growth-promoting rhizobacteria
PL	Pistacia lentiscus L.
JP	Juniperus phoenicea L.
RO	Rosmarinus officinalis L.

Supplementary Materials

The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/plants15040529/s1. Figure S1: Identified metabolite superclasses in the leaf metabolome of non-inoculated and inoculated plants; Figure S2. Box plot showing the abundance of the most important features that account for PCA differences in the leaf metabolome of RMs-inoculated and non-inoculated (C) plants. Plants were inoculated with RMs from R. officinalis L. (RO), P. lentiscus L. (PL), and J. phoenicea L. (JP). Distinct lowercase letters indicate statistically significant differences (p ≤ 0.05) among treatments for individual metabolites (n = 3); Table S1: Metabolites whose relative abundance significantly changed (log₂ fold change [log₂FC] ratio ≤ −1 or ≥ 1; p ≤ 0.05) in the leaf metabolome of rhizomicrobiome-inoculated compared to the non-inoculated (C) plants. JP, RO, and PL represent the leaf metabolomic profile of plants inoculated with the rhizomicrobiomes extracted from J. phoenicea L., P. lentiscus L., and R. officinalis L., respectively. Metabolites and the corresponding superclasses, whose abundance commonly changes in response to inoculation, independently of the rhizometabolome, are reported in bold. Metabolites highlighted in green show decreased relative abundance compared to non-inoculated plants. Metabolites highlighted in red show increased relative abundance compared to non-inoculated plants. In parentheses, the number of molecules belonging to each molecular superclass. Metabolites reported in italic are also the most important features.

Author Contributions

Conceptualization, M.R. and R.A.J.; methodology, M.R. and R.A.J.; software, R.A.J.; validation, M.R., F.L., and R.A.J.; formal analysis, R.A.J.; investigation, R.A.J., F.L., and A.G.F.; resources, M.R.; data curation, R.A.J.; writing—original draft preparation, R.A.J. and M.R.; writing—review and editing, R.A.J., F.L., A.G.F., and M.R.; visualization, R.A.J.; supervision, M.R.; project administration, F.L.; funding acquisition, M.R. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The original contributions presented in this study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

Conflicts of Interest

The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

Funding Statement

This study was carried out within the Agritech National Research Center and received funding from the European Union Next-GenerationEU (PIANO NAZIONALE DI RIPRESA E RESILIENZA (PNRR)—MISSIONE 4 COMPONENTE 2, INVESTIMENTO 1.4—D.D. 1032 17/06/2022, CN00000022).