*Article* **The Ancient Neapolitan Sweet Lime and the Calabrian Lemoncetta Locrese Belong to the Same Citrus Species**

#### **Domenico Cautela 1,\*, Maria Luisa Balestrieri 2, Sara Savini 3, Anna Sannino 3, Giovanna Ferrari 4, Luigi Servillo 2, Luigi De Masi 5, Annalisa Pastore <sup>6</sup> and Domenico Castaldo 1,4,7**


Academic Editor: Celestino Santos-Buelga Received: 21 November 2019; Accepted: 23 December 2019; Published: 27 December 2019

**Abstract:** "Neapolitan limmo" is an ancient and rare sweet Mediterranean lime, now almost extinct but used until a few decades ago for the production of a fragrant liqueur called the "four citrus fruits". The objective of this work was to compare, through the use of chemical (flavonoids, volatile organic compounds, and chiral compounds) and molecular (DNA fingerprint based on RAPD-PCR) markers, the residual population of Neapolitan limmo with other populations of sweet limes, identified in Calabria and known as "lemoncetta Locrese". We report for the first time specific botanical characteristics of the two fruits and unequivocally show that the ancient sweet Mediterranean limes Neapolitan limmo and lemoncetta Locrese are synonyms of the same Citrus species. Owing to the biodiversity conserved in their places of origin, it will now be possible to recover, enhance and implement the use of this ancient sweet lime for agro-industrial purposes.

**Keywords:** agrobiodiversity; lemoncetta Locrese; Mediterranean sweet lime; Neapolitan limmo; Neapolitan "four citrus fruits" liqueur; taxonomy

#### **1. Introduction**

"Limmo" or "limo" or also "limma", not to be confused with the more famous lime (*C. aurantifolia*), is an ancient Neapolitan citrus of the genus *Citrus*. The first traces of the presence of limmo in the Neapolitan province date back to the end of the seventeenth century [1]. It was described as the fruit of Lomia or Lumia, a species of sweet and sweet-smelling *Citrus* fruits similar to lemon but smaller [2]. Limmo has a strongly rounded shape of about 5–6 cm diameter (Figure 1a). According to TG/203/1 UPOV guidelines [3], it is morphologically characterized by a base with a depressed, slightly rounded distal part and a nipple of conical and umbonate shape, sunken at the base of ca. 1–2 cm (Figure 1b). It is similar to the "Sicilian lumia", recently described by Raimondo et al. [4].

**Figure 1.** Neapolitan limmo or lemoncetta Locrese. (**a,b**) Ripe fruits are sulfur-yellow in color and have a diameter of 4–5 cm. (**c**) Fruit section with 8–10 loggias. (**d**) Ripe fruits with greenish color. (**e**) Young leaves and flowers. (**f**) Leaves and small fruits. (**g**) Neapolitan limmo tree.

Limmo has a thin yellow skin (Figure 1c) in the ripe fruit, consisting of 8–10 loggias containing few seeds, with segments of color between yellow and green, of delicate flavor, not sour, aromatic and sometimes also with greenish notes in the peel of the ripe fruit (Figure 1d). The limmo flowers, compared to lemon, are on average smaller, of medium size, fragrant, with white petals and buds (Figure 1e,f). Small is also the limmo tree with leaves similar to those of lemon (Figure 1f,g). The cultivation of this sweet lime is now completely amateurish. We could count only a few plants in gardens of Naples and of the Neapolitan province.

Limmo belongs to the group of the Mediterranean sweet limes and lemons [5]. It appears nevertheless distinct from them for the color of the flower petals and for the low acidity of the fruit juice [4,6,7]. The acidity is instead high in lemons (*C. limon*) and medium high in most of the common limes (*C. limetta* Risso, subsect. *Limonoides*) [8], like the acidic "limonette de Marrakech" [9], and the Mediterranean sweet lime, *C. lumia* Risso. The latter is an acid-less variety of *C. limetta* Risso, (subsect. *Decumanoides*-sect. *Citrophorum* according to Tanaka) [8] which has a long history of cultivation in Italy as early as the seventeenth and eighteenth centuries [10,11]. The acid-less phenotype of *C. limetta* is due to the inability of producing anthocyanin pigments in leaves and flowers and proanthocyanidins in seeds [6,7] with low citric acid contents [12–14] and juice pH values also above 6 [5–7,15].

Neapolitan limmo stands out among other citrus fruits for its fragrant aroma. The flavor is not very sweet, rather watered down, almost totally devoid of acidity, not savory and, therefore, unappetizing; the latter characteristics were in ancient times systematically exploited in Naples and its surroundings as a defense from thieves. At the end of the nineteenth century, traditional and patrician gardens were surrounded by limmo trees to protect the property from the street urchins, who were Neapolitan boys accustomed by the adversity and poverty of that time to survive in the street thanks to small daily thefts like stealing seasonal fruits from city gardens. Limmo was also used in Neapolitan families to prepare the ancient liqueur "with the four citrus fruits", today almost disappeared. The liqueur was obtained by cold maceration in ethylic alcohol of the slightly unripe peels of oranges, mandarins, lemons, and Neapolitan limmo. This ancient liqueur was much more sought after than the more

popular and widespread liqueur "limoncello", another Neapolitan liqueur obtained using IGP lemons (Protected Geographical Indication) from the Amalfi coast and Sorrento [16,17].

The use of limmo in natural medicine, together with that of other citrus fruits, was reported in the 1825 edition of Phamacopoeia by Antonio Ferrarini (pharmacist, Member of the Health Commission of Bologna City and surroundings, and Lecturer at the Faculty of Pharmacy) [18]. Along with other citrus fruits, limmo was used for the preparation of "aromatic cedar water" or "citron aromatic water". In traditional medicine, the limmo juice was once used as a remedy for cough mixed with prickly pear juice [19].

No studies on limmo are present in the literature. Early taxonomists hypothesized that lemons and limes are derivatives or hybrids of citrons. However, a definitive classification and origin of the species was not proposed. It is a shared opinion that cultivated limes, sweet limes, and lemons originate from interspecific hybridization of cedar (*C. medica* L.) in combination with sour orange (*C. aurantium* L.) while the *C. maxima* × *C. reticulata* hybrid gives rise to the *C. limettioide* subgroup Palestinian sweet lime and *C. meyeri* Meyer lemon [5,20].

Since in Naples limmo was also described as a "sweet bergamot", we searched for this sweet lime in Calabria as well, with the intention of verifying the presence of limmo. Calabria is a region of Southern Italy with extensive citrus fruit cultivations, especially in the Ionic area of the province of Reggio Calabria, the area of origin and production of bergamot (*Citrus Bergamia* Risso) [21,22].

The results of this survey showed the presence of a discreet population of sweet limes in the region East to Reggio Calabria, the Locri area. The local fruit is morphologically similar to Neapolitan limmo and locally known as "lemoncetta Locrese" or "pirettu Locrese". This fruit is of no agro-industrial use and thus of little economic importance. As for Neapolitan limmo, it was never characterized compositionally.

We thus decided to compare the populations of Neapolitan limmo with the Calabrian lemoncetta Locrese with chemical and genetic approaches. We measured chemical markers of citrus fruits, such as glycoside flavanones and determined the profile of volatile organic compounds (VOCs) in juice and peel and the enantiomeric distribution of volatile organic chiral compounds. We then analyzed the genetic diversity of the two populations by Random Amplified Polymorphic DNA (RAPD) analysis.

The results of this study aimed at characterizing and comparing the two apparently distinct fruits led us to conclude that they are compositionally and genetically indistinguishable within variations due to climatic and soil differences. These results will help to restore the use of the ancient Mediterranean limmo to produce the Neapolitan "four citrus fruits" liqueur and as a promising and precious resource for the essential oil industry.

#### **2. Results and Discussion**

#### *2.1. Flavonoids, Organic Acids and Proximate Constituents*

Flavanones and to a lesser extent flavonols are the predominant flavonoids in the genus *Citrus* [23]. The quali-quantitative distribution of these phenols is largely influenced by the specie and/or the variety [24–26]. Flavonoids are thus commonly used as chemotaxonomic markers and evaluate the quality and genuineness of citrus juices [24,27–30]. Since flavanones constitute virtually all of the total flavonoids present (e.g., 98% in grapefruit, 90% in lemons, and 96% in limes) [29], we focused on the major aglycone flavanones with their rutinose or neohesperidose glycosides as markers to differentiate Neapolitan limmo and lemoncetta Locrese from other citrus juices and between the two populations. The same approach was utilized by Mouly et al. [24] to effectively differentiate lemon and lime, varieties of grapefruits (white, pink, red, and green), and sweet oranges (Valencia, navel, blood, Thomson, and Malta). The flavonoid profile is also a method widely utilized to detect the possible mixture of different juices as for instance the addition of bergamot to lemon juice [30]. The range of variability of flavonoids and organic acids for citrus juices are reported in the code of practice of the International Federation of Fruit Juice Producers (IFFJP).

We thus compared by HPLC the population of flavonoids in Neapolitan limmo with those of lemoncetta Locrese (Figure 2 and Table 1). An identical flavonoid profile unites both analyzed populations (Figure 2). Neapolitan limmo had flavanone profiles more like lemoncetta Locrese. The insignificant difference (*p* < 0.05) observed in flavonoid contents are within the normal limits of environmental variability of these two juices.

Both Neapolitan limmo and lemoncetta Locrese are qualitatively characterized by a common presence of five different flavonoids: the three rutinosidic flavanones, hesperidin, eriocitrin, and narirutin and the two flavones *O*-glycosides, rutin and diosmin. The identification of these flavonoids is confirmed not only by the retention times but also by spectra analysis compared to their standards (data not shown). The samples do not contain flavanone *O*-neohesperidose and the non-bitter flavanone neoponcirin.

**Figure 2.** Liquid chromatography (LC) flavonoid profile of Neapolitan limmo and lemoncetta locrese. (**a**) Standard containing a mixture (50 mg/L) of eriocitrin (1), neo-eriocitrin (2), rutin (3), narirutin (4), naringin (5), hesperidin (6), neohesperidin (7), diosmin (8), poncirin (9), didymin (10). (**b**) LC chromatograms of Neapolitan limmo (**c**) LC chromatograms of lemoncetta Locrese. Flavanones were monitored at 285 nm (black line), flavones at 325 nm (gray line).

Of particular interest between the present flavonoids is eriocitrin, a flavonoid that is exclusively characteristic of lemon juice [27,29], and is almost absent in orange and grapefruit juice. Eriocitrin in limmo and lemoncetta strengthens the genetic closeness to lemon of these Mediterranean limes, both hybrids of citron (*C. medica* L.) being a sour orange × Citron cross [5].

The flavonoid composition of limmo and lemoncetta was also compared with that of two other limes, both previously characterized by Nogata et al. [23]. The first is classified under the *Limonoides* subset according to the Tanaka's system [8], with the common name Sweet lemon and the scientific name *C. limetta* or *C. limetta* Risso, with slightly acidulous pulp [9,12,13]. The second is classified in the subsection of *Decumanoides* (sect. *Citrophorum*) [8], with common name lumie and the scientific name *C. lumia* or *C. lumia* Risso, with sweet and non-acidic pulp [4,13]. This citrus is most diffused in Italy and in some southern regions of France [31]. A prevalent similarity emerges from the comparison between the flavonoid profiles of *C. limetta* and *C. lumia* and those of Neapolitan limmo or lemoncetta Locrese (Table 1): Neapolitan limmo, lemoncetta Locrese and *C. lumia* have a significant content of the flavanone *O*-rutinosides esperidine and eriocitrin. They also have a reduced content of the other flavanone rutinosides narirutin (Table 1) and of the flavones rutin and diosmin. Significative is the common absence of the flavanones neoponcirin, naringin, neohesperidin, neoeriocitrin e poncirin in limmo, lemoncetta and *C. lumia* but not in *C. limetta*.

Unfortunately, no other paper on the traditional Italian sweet lime varieties besides Nogata et al. [23] report data on flavonoids [12,13]. The data are anyway consistent with those recently found by Smeriglio et al. [32] on *C. Lumia* which reported a similar significant presence of hesperidin and eriocitrin.


**Table 1.** Flavonoid content (mg/Kg) in Neapolitan limmo and lemoncetta Locrese estimated by HPLC. (n.d. ≤ 0.5 mg/Kg).

A further indication that the Neapolitan limmo can be with good reasons classified in the *Citrus lumie* group is also offered by two recent papers by two distinct research groups [6,7]. These authors demonstrated by independent methodological approaches that in acid-less varieties of citrus, exceptionally low fruit acidity is associated with absence of anthocyanin pigments in leaves and flowers and of proanthocyanidins in seeds and flowers without pigmentation or white, like those of Neapolitan limmo (Figure 1e).

Next, we extended our investigation to the quali-quantitative distribution of organic acids, the overall acidity, and the pH of the juice. These parameters can give useful indications on the nature of the lime type discriminating between acid ecotypes. Both limmo and lemoncetta have qualitatively a common acidic chromatographic profile characterized by the presence of five organic acids: malic, citric, quinic, tartaric, and fumaric acids (Figure 3). Similar are also the quantitative data (Table 2).

Malic acid is the dominant organic acid of these sweet Mediterranean limes with average values of 1.57 ± 0.03 g/L in Neapolitan limmo and slightly higher, 1.88 ± 0.02 g/L, in lemoncetta. It is probably this significant presence in the acidic profile that confers to the juices of these fruits (acid-less sweet tasting) that smooth tartness acidity given by malic acid. This taste is clearly different from the sensorial sour quality given by citric acid in juices when this is dominant [33].

**Figure 3.** Organic acid profile of Neapolitan limmo and lemoncetta locrese. (**a**) Standard containing a mixture of tartaric acid (1) 0.5 mg/mL; quinic acid (2) 0.5 g/L; malic acid (3) 0.5 g/L; citric acid (4) 0.5 g/L; fumaric acid (5) 2.5 mg/L. (**b**) LC chromatograms of Neapolitan limmo (**c**) LC chromatograms of lemoncetta Locrese.

Both Neapolitan limmo and lemoncetta showed reduced contents of citric acid with average values of about 0.94 g/L in the group of Neapolitan limmo and even lower in lemoncetta Locrese (0.48 g/L), with an average pH > 5.7 and values of total acidity < 1.4 g/L (Table 2). This is consistent with the phenotypes of the sweet forms of *C. limetta* Risso—Mediterranean sweet lime—sweet Roman [12], Roman [13,34], Lima Dulce, or Dulce lime [14].

Quinic acid is the most expressed acidic compound after malic acid and citric acids. The average levels of malic acid are between 0.10 and 0.43 g/L with higher average values for lemoncetta Locrese compared to Neapolitan limmo (Table 2). Also, for this acid, the quantitatively expressed levels appear on average higher in the group of lemoncetta Locrese than in those of Neapolitan limmo. The average values are however completely comparable with each other. Finally, fumaric acid is much less expressed and generally does not exceed 0.01 g/L.

**Table 2.** Proximate constituents. pH, soluble solids (◦Brix), titratable acidity (as citric monohydrate acid g/L), and organic acids (g/L) in Neapolitan limmo and lemoncetta Locrese.


This analysis makes us conclude that limmo and lemoncetta are chemically similar although there is an appreciable quantitative difference in some substances likely due to climatic and soil composition and other environmental differences. The values of flavonoids and other metabolites are, for instance, different likely because of the different degree of activivity of phenylalanine ammonium lyase, the enzyme central to the production of the biosyntesis precursor of flavonoids cinnamic acid [35].

#### *2.2. Chirospecific Analysis*

Biological activity is often correlated with chiral properties. In citrus fruits, chiral compounds are widely used as indicators of adulteration or fraud of essential oils by addition of synthetic or natural compounds of different botanical origin. The GC profiles of volatile aromatic compounds of essential oils from Neapolitan limmo and lemoncetta Locrese (Figure 4a) were initially compared and analyzed by heart-cutting multidimensional GC [36] to estimate the enantiomeric distribution (ee%) of chiral β-pinene, sabinene, limonene, linalool and linalyl acetate (Figure 4b).

As for flavonoids and organic acids, an identical metabolic profile of volatile compounds was common to both citrus populations (Figure 4a). Forty-three volatile aromatic compounds were identified. In both populations, the more expressed were limonene (61.8 ≥ ± 14.4%) ≥ linalyl acetate (9.2 ± 0.5%) ≥ linalool (6.6 ± 0.4%) ≥ β-pinene (4.4 ± 2.9%) ≥ myrcene (1.3 ± 0.6%) ≥ sabinene (0.8 ± 0.4%) ≥ α-terpineol (0.7 ± 0.6%) ≥ α-pinene (0.5 ± 0.4%) ≥ geranial (0.4 ± 0.3%) ≥ neral (0.3 ± 0.1%) ≥ β-bisabolene (0.2 ± 0.1%) ≥ nerol (0.2 ± 0.1%) ≥ terpinene and citronellol ranged from 0.05 to 0.1%. Camphene, octanal, α-phellandrene, terpinolene, and terpinen-4-ol were less than 0.05%.

The data obtained by four heart-cut multidimensional GC are even more interesting; the enantiomers of β-pinene, sabinene, limonene, linalool, and linalyl acetate were all well-separated on a DiActButylsilyl γ-CDX chiral column (Figure 4b). The dominant enantiomeric form for limonene was (*R*)-(+). Both populations showed (*R*)-(+) for β-pinene and sabinene (*S*)-(−), and (*R*)-(−) for linalyl acetate and linalool (Table 3).

**Figure 4.** Analysis of the chirality of Neapolitan limmo and lemoncetta Locrese (**a**) GC chromatogram of Neapolitan limmo essential oil with the four heart-cuts. (**b**) Enantio-MDGC chromatogram of selected chiral compounds in Neapolitan limmo.


**Table 3.** Enantiomeric distribution of chiral compounds in Neapolitan limmo and lemoncetta Locrese essential oil.

#### *2.3. Volatile Organic Compounds Analysis*

Comparison of the total ion chromatograms of the aroma components collected in the juices of Neapolitan limmo and lemoncetta Locrese showed the presence of 8 terpenes, 5 monoterpenoid alcohols, and 3 sesquiterpene hydrocarbons in the juices. Both fruits presented the same volatile compounds (Figures 5 and 6).

**Figure 5.** Total ion chromatograms of the juices collected by SPME/GC/MS. (**a**) Neapolitan limmo. (**b**) lemoncetta Locrese. IS: Internal Standard. The peak numbers correspond to the numbers in the first column of Table 4.

**Figure 6.** Total ion chromatograms of the peels collected by SPME/GC/MS. (**a**) Neapolitan limmo. (**b**) lemoncetta Locrese. Peak numbers correspond to the numbers in the first column of Table 5.

In the variety lemoncetta Locrese 12 volatile compounds were in concentrations significantly higher than in limmo (Table 4).

**Table 4.** Volatile compounds identified in the juices of the Neapolitan limmo and lemoncetta Locrese. Data were expressed as: Area volatile compound/Area Internal Standard (A/As.i.) and Area percent (%). Values correspond to average ± standard deviation. Values are significant at *p* < 0.05.


D-limonene was the main flavor compound found in both varieties, but its concentration was significantly higher in Neapolitan limmo compared to lemoncetta Locrese (*p* < 0.05). Other major components (except limonene) are β-pinene, β-myrcene, and bergamol (linalyl acetate) for both varieties. The last two compounds are significantly higher in Neapolitan limmo (*p* < 0.05). Moreover, other volatile aromas found at small percentage such as α-phellandrene, α-pinene, β-phellandrene, linalool, trans-α-bergamotene, and β-bisabolene were considered to be important compounds influencing the entire aroma [37]. Their concentrations were significantly higher in lemoncetta (*p* < 0.05). Only the volatile compound α-pinene was not significantly different in the two varieties (*p* > 0.05).

Twenty-three volatile compounds were identified in the peels: 10 terpenes, 8 monoterpenoid alcohols, and 5 sesquiterpene hydrocarbons (Table 5). Both varieties exhibited the same volatile compounds with different intensity (Figure 6). The peel of lemoncetta Locrese presented significantly higher areas than Neapolitan limmo. The main component of the peels is Limonene followed by bergamol, linalool, β-pinene, and β-myrcene with different intensities. Limonene value was not reported because is no longer linear and the areas were off the charts. Other relevant compounds are α-pinene, β-phellandrene, β-pinene, β-myrcene, linalool, and bergamol which are significantly higher in lemoncetta compared to Neapolitan limmo (*p* < 0.05). Only two volatile compounds (nerol acetate and geraniol acetate) are significantly higher in Neapolitan limmo than in lemoncetta Locrese (*p* < 0.05).



The compounds found in these varieties were also reported in other Citrus varieties. Particularly, in the peels were found 10 additional volatile compounds: cis-β-terpineol, terpinolene, α-terpineol, acetic acid octyl ester, trans-geraniol, α-terpineol acetate, nerol acetate, geraniol acetate, α-bergamotene, and cis-α-bisabolene. The absence of these compounds in the juices is probably due to the juice squeezing. The extraction pressure conditions will determine different aroma components in juices and in peels.

Our results revealed that there are not qualitative differences between the two varieties. The aromatic profiles are identical and there are not specific volatile compounds that could be used to differentiate the varieties. The main differences are connected only to the intensity of the aromatic profile. The use of SPME-GC-MS thus resulted to be a valuable tool to analyze the volatile profile of the two sweet lime juices and peels and obtain a quality characterization of fruits from different varieties.

#### *2.4. Genetic Comparison by DNA-Based Molecular Markers*

The genetic similarity of the two Mediterranean sweet lime populations was finally analyzed using RAPD molecular markers [38]. This technique was preferred to DNA barcoding and phylogenic analysis because DNA barcoding works best if the sequences have sufficiently diverged. We feared that the assay could prove inconclusive in our case given the indications from chemical data that the evolutionary distance between Limmo and lemoncetta is close [39]. RAPD has instead proven useful in the identification of Citrus cultivars and the assessment of genetic relatedness for neglected or little-known citrus accessions [40,41]. Lemon (*C. limon*) cultivars of Campania (Italy) were for instance distinguished by their RAPD profiles using five arbitrary primers, confirming that RAPD markers can successfully identify lemon genotypes [42]. Iannelli et al. [43] characterized lemons by combining genome size and RAPD markers. In their work, the primer U19 utilized for distinguishing lemon genotypes.

The RAPD-PCR method allowed the genetic analysis of Neapolitan limmo and lemoncetta Locrese by simply comparing the presence/absence of bands in DNA amplification patterns, visible after electrophoresis on agarose gel, as the bands represent the numerous loci detected randomly and dispersed in their respective genomes. We used primers already successfully considered for the variety discrimination of other species [44,45] but we obtained more markers as compared to previous work. The primers used showed high reproducibility of amplification products. They also showed the absence of polymorphisms in DNA in the comparison between limmo and lemoncetta and therefore the impossibility to discriminate between the two populations (Figure 7a,b).

The RAPD profiles obtained with each primer were extremely different from each other, allowing us to explore different regions of the genome. Alleles corresponding to reproducible amplicons, given the dominant genetic nature of RAPD markers, identified a total of 80 markers or loci (Table 6). The number detected was dependent on the primer used but not on the variety, with an average of 6.7 loci per primer, going from the minimum of four bands for primers G07 and U4 and at most 12 bands for the AX08 primer. Also, the genetic analysis suggested that the Neapolitan limmo and the lemoncetta Locrese are likely synonyms of the same variety since the loci were not polymorphic and did not allow to discriminate between the two local varieties.

**Figure 7.** Genetic diversity of Neapolitan limmo and lemoncetta Locrese by RAPD molecular analysis. Comparison of the RAPD results on agarose gels of lemoncetta Locrese (odd lanes) and Neapolitan limmo genomes (even lanes), respectively, with the indicated arbitrary primers. (**a**) Lanes 1, 2: primer A05; lanes 3, 4: primer AK10; lanes 5, 6: primer AN10; lanes 7, 8: primer AX01; lanes 9, 10: primer AX08; lanes 11, 12: primer G07. (**b**) Lanes 1, 2: primer G12; lanes 3, 4: primer G19; lanes 5, 6: primer E10; lanes 7, 8: primer E11; lanes 9, 10: primer U4; lanes 11, 12: primer U19. Lanes M: GeneRuler 1 kb Plus DNA Ladder (Thermo Fisher Scientific) as molecular weight marker, containing three darkest bands consisting of 5000, 1500, and 500 bp.

#### **3. Materials and Methods**

#### *3.1. Plant Materials*

Fruits and Leaves samples of Neapolitan limmo and lemoncetta Locrese were harvested in January 2018 and 2019 from populations located in areas of Afragola (40◦55- 37-- N; 14◦18- 42-- E), Pozzuoli (40◦49- 39-- N; 14◦9- 11-- E) and Pianura (41◦02- 24-- N; 14◦11- 09-- E) (Campania, Italy) or Locri (38◦14- 19-- N; 16◦15- 34-- E) and Siderno (38◦16- N; 16◦18- E), Reggio Calabria (Calabria, Italy), respectively, and placed in a 4 ◦C refrigerated box to be processed.

Chemicals: All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) or Extrasynthes (Genay, France). An internal standard solution of camphor used for the analysis of volatile compounds in the juices was obtained from Sigma-Aldrich. The purity of all of the standards was beyond 95%. All other solvents and reagents were of analytical grade.

#### *3.2. Preparation of the Samples for Chemical Analyses*

Chemical analyses were performed on juices obtained in the laboratory from fresh fruits of Neapolitan limmo and lemoncetta Locrese. A total of 6 different Neapolitan limmo juice samples were used (2 in 2018 and 4 in 2019) and 8 lemoncetta Locrese samples (3 in 2018 and 5 in 2019). The juices were prepared using a manual squeezer, filtered through a stainless-steel filter with 1.18 mm mesh diameter, centrifuged at 18,000× *g* for 60 min at 4 ◦C, placed in plastic bags in 100 mL aliquots, and stored at −20◦ C until usage. The essential oils were extracted from fruits of limmo or lemoncetta (2 kg) through manual abrasion of the frozen flavedo at −20 ◦C by a small stainless steel grater with

subsequent recovery and mixing 1:5 (*w*/*w*) with a saline solution (1 M NaCl) at 0 ◦C and subsequent centrifugation at 18,000× *g* for 60 min at 4◦ C. The oil recovered after centrifugation (supernatant, about 200 μL) was dried over anhydrous sodium sulfate and re-centrifuged at 12,000× *g* for 15 min at 4 ◦C and stored in the dark in a 1 mL vial under nitrogen at 5◦ C. A total of 8 samples of essential oils were prepared (all in 2019), 4 of Neapolitan limmo and 4 of lemoncetta Locrese.

#### *3.3. Proximate Constituents*

The soluble solids, expressed in Brix degrees, were determined by measurement of refractive index at 20 ◦C. The pH was determined by a Crison Model microTT 2050 pHmeter. Titrable acidity (total acids), expressed as citric acid monohydrate, was determined by titrating a 10 g sample with 0.1 N NaOH up to pH 8.1 according to the method reported by the International Federation of Fruit Juice Producers [46].

#### *3.4. Organic Acid Analysis*

Limmo or lemoncetta juice (20 g) were clarified by centrifugation at 12,000× *g* for 15 min. The clarified extract was filtered through a 0.45 μm Millipore filter (Darmstadt, Germany); 10 mL was chromatographed through a cation-exchange column [AG-1-X8 (HCOO−) poly-prep Bio-Rad (Hercules, CA, USA)] and washed with water to a total volume of 100 mL. The organic acids were eluted with 6 M formic acid (ca. 130 mL), collected, and evaporated. The dry samples were recovered with water (10 mL) and filtered through a 0.45 μm Millipore filter before HPLC analysis. A volume of 10 μL was employed for the HPLC ThermoFinnigan Surveyor (Thermo Finnigan, Waltham, MA, USA) analysis on a Restek Allure organic acids cartridge 5 μm, 300 mm × 4.6 mm ID thermostated at 25 ◦C. The isocratic elution was carried out with a eluent consisting of 100 mM phosphate buffer at pH 2.5 with a flow rate of 0.5 mL/min with detection at 226 nm by a diode array detector [47] interfaced to a Dell computer Optlex gx260 with Xcalibur software for the signal acquisition and elaboration. Identification of quinic, malic, citric, fumaric, and tartaric acids was based on the retention time by co-injection of reference standards.

#### *3.5. Flavonoid Analysis*

The determination of the flavonoids (Flavanone *O*-glycosides and Flavone *O*-glycosides) in the juices was carried out by liquid chromatography according to the method of Grandi et al. [27]. Ten phenolic compounds were quantified. They included four flavanone *O*-glycosides with a rutinose (rhamnosyl-α-1,6-glucose) moiety: hesperidin, narirutin, eriocitrin and didymin (or neoponcirin). Four more were flavanone O-glycosides with neohesperidose moiety (rhamnosyl-α-1,2-glucose): naringin, neohesperidin, neoeriocitrin and poncirin.

Standard solutions of the flavonoids were prepared by weighing exactly 0.1 g of each compound and dissolving it in 100 mL of *N,N*-dimethylformamide. Those solutions were used to build up the calibration lines by diluting them to cover the concentration range of 1–100 mg/L.

The juices (10 mL) were shaken with 20 mL of a 1:1 (*v*/*v*) mixture of 0.25 M *N,N*-dimethylformamide/ ammonium oxalate and 20 mL of analytical-grade water and then filtered on 0.45 μm PTFE Pall filters. A volume of 5 μL was employed for the HPLC analysis on a Phenomenex Luna column C18 (l50 × 3 mm) 5 μm thermostated at 25 ◦C. The elution was conducted as indicated by Grandi et al. [27]. The eluent A was made by an aqueous solution of 5 mM KH2PO4 adjusted at pH 3.05 with phosphoric acid. The eluent B was obtained by mixing acetonitrile/water/0.25 M KH2PO4, in the ratio 70:26:4 (*v*/*v*/*v*), and adding 100 μL of H3PO4 (87%) per liter of solution.

Finally, we quantified two flavone *O*-glycosides with a rutinose sugar moiety: rutin and diosmin. Specific wavelengths were used to identify the individual classes according to Gattuso et al. [48]: flavanones have an absorption maximum at 280–290 nm (set at 285), the flavones rutin and diosmin absorb at 304–350 nm (set at 325).

#### *3.6. Chirospecific Analysis of Sabinene,* β*-Pinene, Limonene, Linalool and Linalyl Acetate in Essential Oils*

The analyses were carried out by gas chromatography (GC-FID). The content in sabinene, β-pinene, limonene, linalool, and linalyl acetate of the essential oils of Neapolitan limmo and lemoncetta Locrese was determined by injecting in split modality 1:100 volume of 0.2 μL essential oil diluted 1:10 (*v*/*v*) in acetone. Analyses were conducted with a RTX®-5 (Resteck, Bellefonte, PA, USA) column (30 m × 0.25 mm, film 0.25 μm) at 70 ◦C for 1 min, 3 ◦C/min at 200 ◦C, holding for 0.3 min, 15 ◦C/min at 250 ◦C, and holding for 5 min. The metabolites content was expressed as the percentage of GC peak areas. Chirospecific analysis of the metabolites was performed with enantioselective multidimensional GC (enantio-MDGC). This technique consists in transferring part of the sample from a primary to a secondary column of different polarity or different chiral type. Our laboratory assembled a MDGC system by joining two Thermo Finnigan Trace 2000 G GC devices through a transfer line thermostated at 160 ◦C [36]. The first GC was equipped with a non-chiral column, the second had a column with a chiral stationary phase. A six-way rotating valve, positioned in the first GC and switchable via software, diverted the flow coming out the first column either to the first detector (FID 1) or to the second column where the substance enantiomers were separated and monitored with the second detector (FID 2) generating the chiral chromatogram. The non-chiral column employed in the first GC was a RTX®-5 column (5% diphenyl/95% dimethyl polysiloxane (30m <sup>×</sup> 0.25 mm i.d., film thickness 0.25 μm) (Resteck, Bellefonte, PA, USA). The carrier gas was helium at constant flow 1.5 mL/min. Make-up gas was nitrogen at 30 mL/min flow rate. The initial oven temperature was set at 70 ◦C for 10 min, then it was programmed from 70 to 85◦ C at 3 ◦C/min, from 85 to 175 ◦C at 5 ◦C/min, from 175 to 285 ◦C at 6 ◦C/min and finally at 285 ◦C for 5 min. The injector temperature was set at 250 ◦C and the detector temperature (FID) at 280 ◦C.

The chiral column used in the second chromatograph was a Diethyl tertbutyl silyl-BETA-Cyclodextrin column (Mega Legnano, Milan, Italy) (25m × 0.20mm i.d., 0.18 μm film thickness). The initial oven temperature was set at 35 ◦C for 25 min, then it was programmed from 35 to 160 ◦C at 4 ◦C/min and held at 140 ◦C for 2 min. The injector temperature 150 ◦C and detector (FID) was set at 220 ◦C. Carrier gas was helium at a programmed pressure. The initial pressure was 290 kpa for 30 min. then varied from 290 to 1500 kpa at 500 kpa/min. For each isomer, the enantiomeric excess (ee%) was calculated as ee% = ((Amax – Amin)/(Amax + Amin)) × 100 where Amax and Amin are the areas of the more and less abundant isomers respectively.

#### *3.7. Volatile Organic Compounds (VOCs)*

Determination of VOCs in the juices and peels was carried out by solid-phase micro extraction (SPME) and analyzed with a gas chromatography-mass spectrometry (GC-MS). An automatic injection autosampler CombiPal (CTC-CombiPal Analytics, Zwingen, Switzerland) was used for SPME sampling. The experiments were performed using a 50/30 μm divinylbenzene/carboxen/polydimethylsiloxane fiber (Supelco, Bellefonte, PA, USA). The fiber was conditioned according to the manufacturer's recommendation to remove contaminants. Before analysis, a fiber blank was run to confirm no contamination peak.

VOCs of juices and peels: An aliquot (8 g) of each juice diluted 20 times was weighed into a 20 mL vial and spiked with 80 μL of internal standard (camphor 3000 mg/L). Each sample was equilibrated at 40 ◦C for 10 min under stirring (500 rpm). After equilibration, the juice or peel were extracted by exposing the SPME fiber at 40 ◦C for 10 min (juice) and 2 min (peel). The analytes were desorbed at 250 ◦C for 15 min in the GC injection port. Measurements were always repeated at least in triplicates.

#### *3.8. Gas Chromatography-Mass Spectrometry Analysis*

The analyses were performed using a Varian 450 GC (Walnut Creek, CA, USA) coupled with Varian 300-MS mass spectrometer (Walnut Creek, CA, USA). The volatile compounds were separated using a Zebron ZB-semivolatiles capillary column (30 × 0.25 mm i.d., 0.25 μm film thickness). The oven temperature program was set as follows: initial temperature was held at 40 ◦C for 2 min, increased to 175 ◦C at 7.5 ◦C/min, and to 275 ◦C at 20 ◦C/min (held for 6 min). The temperatures of the transfer line and the ion source were set at 300 ◦C and 230 ◦C, respectively. Helium (99.999% purity) was used as a carrier gas at 0.9 mL/min. A split injection with a ratio of 1:100 was used. The analyses were carried out under full-scan acquisition mode. The mass range used was 30 to 450 *m*/*z*. The identification of volatile compounds was based on the comparison between the mass spectrum for each compound with those of two spectral libraries: NIST (National Institute of Standards and Technology, Gaithersburg, MD, USA) and WILEY (Wiley, New York, NY, USA). The data obtained were collected using the Bruker software Chemical Analysis MS Workstation version 7.0 (Karlsruhe, Germany).

#### *3.9. Random Amplified Polymorphic DNA (RAPD-PCR) Analysis*

Total genomic DNA of Neapolitan limmo and lemoncetta Locrese was isolated from 10 mg of lyophilized leaves. Plant tissue disruption was achieved at dry ice temperature through stainless beads in 2 mL tubes using a TissueLyser apparatus (Qiagen S.r.l., Milano, Italy) programmed at 30 Hz for 1 min. DNA was extracted according to the GeneJET Plant Genomic DNA Purification Mini Kit (Thermo Fisher Scientific, Carlsbad, CA, USA). The samples were dissolved in pre-heated (60 ◦C) lysis buffer. The suspensions were incubated at 60 ◦C for 10 min and subjected to the extraction procedure. RNase A treatment (5 μg/mL) was necessary to eliminate the co-extracted RNA. Finally, the DNA was eluted and diluted a final concentration of 20 ng/μL determined from the UV absorbance at 260 nm. As the purity and quality of the DNA template are crucial factors for a successful PCR, genomic DNA was checked by the 260/280 nm absorbance ratio and agarose gel electrophoresis.

The RAPD-PCR procedure was well established in our group in a previous project on *Citrus*, where sensitivity and reproducibility of the method were examined on genomic DNA from 1 to 100 ng [38,45,49]. In the present study we used 10 ng DNA template samples. The arbitrary primers tested in the PCR reaction had 60% G + C content and were 10 nucleotides long. After an initial screening of 20 arbitrary oligodeoxyribonucleotide decamer primers, 12 primers were selected for reproducibility of the band patterns: A05, AK10, AN10, AX01, AX08, G07, G12, G19, E10, E11, U4, and U19 (Table 6).

Each PCR was carried out in a 50 μL volume, containing 1X DreamTaq buffer with 2 mM MgCl2, brought to 3 mM MgCl2, 100 μM of each dNTP, 20 pmols of the arbitrary and unique primer, 2.0 Units of DreamTaq DNA polymerase (Thermo Fisher Scientific) and 10 ng of citrus genomic DNA. The PCR mixture was assembled on ice and transferred to a pre-cooled (6 ◦C) Veriti thermal cycler with a heated lid (Thermo Fisher Scientific).


**Table 6.** Arbitrary 10-mer primers used for RAPD analysis of Neapolitan limmo and lemoncetta Locrese.

The DNA template was amplified by the following cycling profile: initial DNA template melting for 3 min at 95 ◦C, 45 cycles of denaturation for 1 min at 95 ◦C, primer annealing for 1 min at 40 ◦C, and synthesis for 1 min at 72 ◦C. The program ended with a final step conducted for 10 min at 72 ◦C. The reaction products were stored at −20 ◦C. Each reaction was repeated three times along with negative controls without genomic DNA. The RAPD-PCR products (25 μL) were separated by electrophoresis on 2% (*w*/*v*) agarose gel containing 0.5 μg/mL SyBr Safe and 1X TAE buffer (89 mM Tris-acetate at pH 8.4, 2 mM EDTA) at 5 V cm-1 for 1.5 h. The GeneRuler 1 kb Plus DNA ladder was used as standard marker of known molecular weights (Thermo Fisher Scientific). Amplicons were visualized under UV transilluminator and digitalized by the Electrophoresis Documentation and Analysis 120 System (Kodak ds-digital science, Rochester, NY, USA).

#### *3.10. Statistical Analysis*

All samples were analyzed in triplicates and the results were expressed as mean ± standard deviation (SD) after a normality distribution Kolmogorov-Smirnov test. Statistical analyses were performed using SPSS software ver. 21.0 (IBM, Armonk, NY, USA).

Statistical comparisons were carried out by analysis of variance (ANOVA) and post hoc Tukey-Kramer tests. A *p* value less than 0.05 was considered statistically significant. All tests were two tailed.

Significant differences in relative intensities of each volatile compound detected by GC-MS were analyzed by Mann-Whitney U-test between Neapoletan limmo and lemoncetta Locrese (α = 5%).

#### **4. Conclusions**

In conclusion, chemical and genetic data obtained by DNA fingerprint using RAPD markers have collectively allowed us to confirm beyond doubts that the ancient and rare sweet Mediterranean lime, known in Campania as Neapolitan limmo and in Calabria as lemoncetta Locrese are synonyms of the same citrus species.

Collectively, the obtained compositional data also indicate that, from the chemo-taxonomic point of view, both fruits belong to *Citrus lumia* Risso species despite the di stinct phenotypes. Our results might allow in the future the repopulation of limmo cultures and the reintroduction of this fruit in the essential oil and gastronomic market.

**Author Contributions:** D.C., A.S., S.S., and L.D.M. collected and analyzed the experiments, A.P., M.L.B. and G.F. critically evaluated the results. A.P. helped to write and finalize the paper. L.S., D.C., and D.C. formulated the original idea, supervised the research and wrote the first draft of the paper. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by Dementia Research UK (RE1 3556).

**Acknowledgments:** We are grateful to Ciro Balestrieri (Università degli Studi della Campania "Luigi Vanvitelli") and to Roberta Ferrari (ProdALscarl) for inspiring this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **The** *Trichoderma atroviride* **Strains P1 and IMI 206040 Di**ff**er in Their Light-Response and VOC Production**

#### **Verena Speckbacher 1, Veronika Ruzsanyi 2, Modestus Wigger <sup>3</sup> and Susanne Zeilinger 1,\***


Academic Editor: Derek J. McPhee Received: 9 December 2019; Accepted: 27 December 2019; Published: 3 January 2020

**Abstract:** *Trichoderma atroviride* is a strong necrotrophic mycoparasite antagonizing and feeding on a broad range of fungal phytopathogens. It further beneficially acts on plants by enhancing growth in root and shoot and inducing systemic resistance. Volatile organic compounds (VOCs) are playing a major role in all those processes. Light is an important modulator of secondary metabolite biosynthesis, but its influence has often been neglected in research on fungal volatiles. To date, *T. atroviride* IMI 206040 and *T. atroviride* P1 are among the most frequently studied *T. atroviride* strains and hence are used as model organisms to study mycoparasitism and photoconidiation. However, there are no studies available, which systematically and comparatively analyzed putative differences between these strains regarding their light-dependent behavior and VOC biosynthesis. We therefore explored the influence of light on conidiation and the mycoparasitic interaction as well as the light-dependent production of VOCs in both strains. Our data show that in contrast to *T. atroviride* IMI 206040 conidiation in strain P1 is independent of light. Furthermore, significant strain- and light-dependent differences in the production of several VOCs between the two strains became evident, indicating that *T. atroviride* P1 could be a better candidate for plant protection than IMI 206040.

**Keywords:** *Trichoderma atroviride*; mycoparasitism; secondary metabolites; volatile organic compounds (VOCs); photoconidiation; fungi; 2-octanone; injury response; light response; *Fusarium oxysporum*

#### **1. Introduction**

Filamentous fungi are prolific producers of volatile organic compounds (VOCs), small carbon-based substances that readily enter the gas phase and that are derived from both, primary and secondary metabolism. Fungal VOCs are produced as mixtures of chemically diverse compounds, with the produced VOC profile being species- and even strain-specific and greatly depending on the developmental stage of the fungus as well as environmental factors [1–3]. They are involved in various biological processes including communication with other organisms such as plants and microbes as well as self-signaling. VOC emission by fungi serves in interaction with animals as reported for 1-octen-3-ol produced by the mushroom *Clitopilus prunulus* for deterring banana slugs and by the wood-rotting fungus *Trametes gibbosa* for attracting fungus-eating beetles [4,5]. Fungal VOCs further can impact plants by activating defense responses and affecting plant growth, as well as directly inhibiting the proliferation of phytopathogens [6].

Members of the fungal genus *Trichoderma* are efficient mycoparasites that antagonize a wide range of phytopathogenic fungi by direct parasitism employing secreted antifungal hydrolytic enzymes and metabolites [7]. At least 480 different VOCs have been identified from *Trichoderma* species yet, with 6-pentyl-2*H*-pyran-2-one (6PP) being one of the first volatiles isolated from this fungal genus [8,9]. *Trichoderma*-emitted VOCs have been connected to the antagonistic activity of these fungi, as they were shown to reduce the mycelial growth of fungal plant pathogens [8,10] as well as wood decay fungi [11]. However, only recently VOC production in co-cultivation was addressed, showing that interaction of *Trichoderma* spp. such as *T. harzianum, T. hamatum* and *T. velutinum* with the ectomycorrhizal fungus *Laccaria bicolor* dramatically altered the VOC emission patterns [12]. In addition to VOCs with bioactivity against fungi, *Trichoderma* spp. release volatiles that affect plant development and immunity. *Arabidopsis* plants exposed to the pool of VOCs emitted by *T. viride* showed increased lateral root formation and growth, and comparable results were obtained with *T. virens* and *T. pseudokoningii* volatiles [13–15]. Similarly, *T. atroviride*-derived 6PP stimulated tomato plant growth and systemic defense; however, the effects were concentration-dependent [8]. In addition to 6-PP, 24 members of the compound classes of alkanes, alcohols, ketones, lactones, furanes, monoterpenes, and sesquiterpenes—including typical fungal C8-compounds—as well as several volatiles not known to be produced by *Trichoderma* before, were identified in the headspace of cultures of *T. atroviride* strain P1 (ATCC 74058) [16]. C8-compounds such as 1-octen-3-ol, 3-octanol and 3-octanone are end-products of fatty acid metabolism [17] and act as signaling molecules regulating fungal development and inter-colony communication. In *T. atroviride*, these volatiles are increasingly produced by conidiating cultures and were described to act as elicitors of conidiation [18]. In addition, 1-octen-3-ol production by *T. atroviride* was up-regulated upon treatment of the fungus with the *Fusarium*-derived mycotoxin fusaric acid, while levels of other VOCs such as 6PP, alpha-phellandrene, and beta-phellandrene decreased [16].

Light, especially in the UV/blue spectrum, is an important environmental cue that triggers asexual reproduction in many fungal species [19]. In *Trichoderma,* most studies on photoconidiation have been performed with *T. atroviride* strain IMI 206040 as a model [20–26]. In complete darkness, *T. atroviride* IMI 206040 has been reported to grow infinitely as mycelium, while exposure to light induces the formation of green conidia [27,28].

In the present study, two different strains (*T. atroviride* P1, ATCC 74058 and *T. atroviride* IMI 206040) of the strong mycoparasite *T. atroviride* were analyzed for their differences in VOC biosynthesis by an in-house made high-resolution ion mobility spectrometer (IMS) with gas chromatographic (GC) pre-separation. Despite the fact that *T. atroviride* is a model to study photoconidiation, no studies have systematically and comparatively analyzed putative strain-, or light-dependent differences in the composition of VOC mixtures released by these fungi. We hence explored and compared their VOC profiles along a cultivation period of 120 h in complete darkness and upon exposure to light, as well as during the mycoparasitic interaction with the host fungi *Rhizoctonia solani* and *Fusarium oxysporum*. Remarkable differences not only in VOC production, but also in light-dependent conidiation and in the mycoparasitism on *F. oxysporum* became evident.

#### **2. Results**

#### *2.1. The Vegetative Growth Rate of T. atroviride Is Strain- and Light- Dependent*

Upon cultivation on PDA plates, the radial growth rate differed between *T. atroviride* strains P1 and IMI 206040. *T. atroviride* IMI 206040 exhibited a higher radial growth rate than *T. atroviride* P1, irrespective of the applied light regime. However, both strains showed enhanced radial growth upon cultivation in complete darkness compared to light-dark conditions (Figure 1).

**Figure 1.** Strain-specific differences in radial growth of *Trichoderma atroviride*. The radial growth rate (cm/d) of *T. atroviride* P1 (P1) and IMI 206040 (IMI) after three days of cultivation on PDA at 25 ◦C under light-dark (LD) conditions or complete darkness (DD). Results shown are means ± SD (*n* = 4).

#### *2.2. Asexual Sporulation in T. atroviride Is Strain- and Light-Dependent*

Comparative analysis of *T. atroviride* P1 and IMI 206040 under conidiation-inducing conditions revealed significant differences between the two strains. In *T. atroviride* IMI 206040, asexual sporulation only occurred under light-dark conditions, while conidia were not formed upon cultivation in complete darkness. According to previous reports [29], conidiation could further be triggered in dark-grown *T. atroviride* IMI 206040 by mechanical injury or a pulse of blue light, respectively. In strain IMI 206040 injury resulted in low conidiation along the cutting sites only, whereas blue light treatment led to the production of massive amounts of heavily pigmented conidia. In contrast, *T. atroviride* P1 fully conidiated even upon growth in complete darkness. Mechanical injury led to strong conidiation and the generation of scarring tissue along the cutting sites in this strain (Figure 2).

**Figure 2.** Strain-specific differences in conidiation upon growth under different light regimes and upon mechanical injury. *T. atroviride* P1 (P1) and *T. atroviride* IMI 206040 (IMI) were grown on PDA at 25 ◦C for five days under light-dark (LD) conditions or in complete darkness (DD). For induction of conidiogenesis, the fungi were grown in complete darkness for two days, treated by either mechanical injury (INJ) or a 10 min blue-light pulse (BLP) followed by incubation for further three days in complete darkness. A representative image of four biological replicates (*n* = 4) is shown.

#### *2.3. The Mycoparasitic Activity of T. atroviride Is Strain- and Light-Dependent*

In direct confrontation assays, the antagonistic activities against the tested host fungi differed between the two *T. atroviride* strains and turned out to be influenced by light. Upon growth in light–dark conditions, neither *T. atroviride* IMI 206040 nor strain P1 were able to fully overgrow and mycoparasitize the two tested host fungi *R. solani* and *F. oxysporum*. However, maybe due to its higher growth rate, *T. atroviride* IMI 206040 made better progress in overgrowing both hosts compared to *T. atroviride* P1. Upon growth in complete darkness both *T. atroviride* strains showed significantly increased antagonistic activities compared to light-dark conditions; they were able to fully overgrow and lyse *R. solani* within five days. Interestingly, *F. oxysporum* could only be fully overgrown and mycoparasitized by *T. atroviride* P1, which was even conidiating under these conditions, while *T. atroviride* IMI 206040 was less active against this host fungus and did not conidiate in darkness. These data suggest that upon growth in complete darkness, *T. atroviride* P1 is a stronger mycoparasite of *F. oxysporum* than *T. atroviride* IMI 206040 (Figure 3).

**Figure 3.** Strain-specific differences in the mycoparasitic behavior against *Fusarium oxysporum* and *Rhizoctonia solani* upon incubation under different light regimes. Plate confrontation assays of the mycoparasites *T. atroviride* P1 (P1) and *T. atroviride* IMI 206040 (IMI) against the host fungi *F. oxysporum* (Fo) and *R. solani* (Rs). As controls, all strains were grown alone (A). The mycoparasites were as well grown in self confrontation (SC) and in direct confrontation with *F. oxysporum* (xFo) or *R. solani* (xRs) on PDA at 25 ◦C for five days under either light-dark (LD) conditions or complete darkness (DD). A representative image of four biological replicates (*n* = 4) is shown.

#### *2.4. VOC Biosynthesis Is Mostly Strain-Specific but Also Light- and Host Dependent*

VOC analyses of *T. atroviride* P1 and IMI 206040 along a period of 120 h upon axenic cultivation under light–dark conditions or in complete darkness as well as upon interaction with host fungi revealed more than 50 peaks in the GC–IMS chromatograms, which corresponded to approx. 20 to 25 VOCs. Since for IMS no commercial database, such as the NIST database for GC–MS, is available, we had to measure VOC standards and confirm the detected substances based on the characteristic IMS and GC data (drift and retention times, respectively). This resulted in the identification of a total of ten volatiles (Table 1). Except 2-octanone, which was solely detected in direct confrontation between *T. atroviride* and *F. oxysporum*, and 3-octanone, which was released by strain IMI 206040 solely upon light–dark conditions, all of the identified VOCs were produced by both strains irrespective of the cultivation condition. However, their amounts and course of release strongly differed between the different experimental setups. The PDA media itself did not emit any of the mentioned VOCs in significant concentrations.


**Table 1.** Volatile organic compounds (VOCs) identified via gas chromatography–ion mobility spectrometry (GC–IMS) in the headspace of *T. atroviride* P1 and IMI 206040.

#### 2.4.1. VOCs Emitted in Strain-Specific Amounts

Whereas 2-heptanone and ethanol were released in similar amounts to the headspace of the two tested *T. atroviride* strains, the amounts of 1-propanol, 2-methyl-propanol, 3-methylbutanal, 2-methyl-butanol, 3-methyl-1-butanol, 3-octanone and 1-octen-3-ol significantly varied between the two *T. atroviride* strains with strain P1 emitting considerably higher amounts than IMI 206040.

In *T. atroviride* P1, the release of 1-octen 3-ol to the headspace strongly increased from 68.5 h on and reached a maximal concentration of approx. 375 parts per billion (ppb) after 96.5 h of cultivation. In *T. atroviride* IMI 206040, 1-octen-3-ol emission peaked at the same time point with a maximal concentration of approx. 60 ppb meaning an 84% reduction compared to strain P1 (Figure 4A). The emission of 1-propanol increased during the early cultivation period, peaked at 43.5 h with approx. 50 ppb in *T. atroviride* P1 and afterwards decreased again. In *T. atroviride* IMI 206040, 1-propanol levels peaked at the same time point, but with a maximal concentration of approx. 10 ppb, which is equivalent to an 80% reduction compared to strain P1 (Figure 4B). 2-heptanone was increasingly released to the headspace over time, reaching and staying at a maximum of ≥1000 ppb at the end of the cultivation period in both *T. atroviride* strains. However, this volatile was already produced in considerable amounts of approx. 275 ppb after 68.5 h of cultivation in *T. atroviride* IMI 206040, while 2-heptanone emission by strain P1 was delayed and increased not before 96.5 h (Figure 4C). 2-methyl-propanol emission followed a similar time course in both *T. atroviride* strains. However, after a steep increase in the early phase of the cultivation, *T. atroviride* P1 produced maximal levels of approx. 650 ppb at the 68.5 h time-point, whereas *T. atroviride* IMI 206040 cultures only emitted up to approx. 175 ppb, which is equivalent to a 73% reduction compared to strain P1 (Figure 4D). 3-methyl-butanal emission by *T. atroviride* P1 reached 45 ppb after 43.5 h and afterwards decreased to 20 ppb until the end of cultivation, whereas in *T. atroviride* IMI 206040 approx. 20 ppb of this volatile were constantly emitted along the whole cultivation period (Figure 4E). The release of ethanol to the headspace increased at the early phases of the cultivation period, peaked at 24.5 h with approx. 65 ppb in *T. atroviride* P1 and afterwards decreased again to approx. 25 ppb. Young *T. atroviride* IMI 206040 cultures produced similar amounts of this volatile with a maximal level of approx. 60 ppb after 43.5 h of cultivation, but with a successive steeper decrease down to approx. 5 ppb at the end of the cultivation period (Figure 4F).

2.4.2. VOCs Emitted in Strain-Specific Amounts in a Light-Dependent Manner

3-octanone, 2-methyl-butanol, and 3-methyl-1-butanol were not only secreted in a strain- but also in a light-dependent manner.

Upon growth in complete darkness, *T. atroviride* IMI 206040 did not emit any 3-octanone. However, up to 25 ppb were measured after 96.5 h of cultivation under light-dark conditions. As the strain already conidiated at this time point in the presence of light, 3-octanone production could not only be light-induced but also conidiation-associated in *T. atroviride* IMI 206040. In contrast, 3-octanone

emission by *T. atroviride* P1 was, similar to conidiation, light-independent and increased with cultivation time to a maximum of 80–95 ppb after 115.5 h of growth (Figure 5; row 1).

**Figure 4.** VOCs produced by *T. atroviride* P1 and IMI 206,040 in a strain-specific but light-independent manner. Levels (ppb) of (**A**) 1-octen 3-ol, (**B**) 1-propanol, (**C**) 2-heptanone, (**D**) 2-methyl-propanol, (**E**) 3-methylbutanal, and (**F**) ethanol detected in the headspace of *T. atroviride* P1 (P1) and *T. atroviride* IMI 206040 (IMI) cultures grown on PDA in Schott bottles. GC–IMS measurements were conducted along a cultivation period of 120 h. As the given volatiles were produced in similar amounts under light-dark conditions and in complete darkness, results shown are means ± SD (*n* = 8).

**Figure 5.** VOCs produced by *T. atroviride* P1 and IMI 206,040 in a strain-specific and light-dependent manner. Levels (ppb) of 3-octanone (row 1), 2-methyl-butanol (row 2), and 3-methyl-1-butanol (row 3) detected in the headspace of *T. atroviride* P1 (P1) and *T. atroviride* IMI 206040 (IMI) cultures grown on PDA in Schott bottles at 25 ◦C under (**A**) light-dark conditions or (**B**) in complete darkness. GC–IMS measurements were conducted along a cultivation period of 120 h. Results shown are means ± SD (*n* = 4).

2-methyl-butanol and 3-methyl-1-butanol were produced light-independently by *T. atroviride* IMI 206040. Low amounts of approx. 25 ppb of both volatiles were detected under both light-dark conditions and complete darkness along the whole cultivation period of this strain (Figure 5 rows 2 and 3). In contrast, *T. atroviride* P1 emitted 2-methyl-butanol and 3-methyl-1-butanol in a slightly light-dependent manner. Upon cultivation under light-dark conditions, increasing amounts of both

substances were released to the headspace with a peak of approx. 375 ppb of 2-methyl-butanol and approx. 225 ppb of 3-methyl-1-butanol after 68.5 h of cultivation. Similar maximal amounts were emitted by strain P1 upon growth in complete darkness; however, production and decline started earlier compared to light–dark conditions (Figure 5; rows 2 and 3).

2.4.3. VOC Emission in Co-Culture

To assess the influence of a host fungus on the emission of VOCs by *T. atroviride*, both strains—IMI 206040 and P1—were co-cultured with *R. solani* or *F. oxysporum*.

Compared to axenic cultures of both *T. atroviride* strains and the fungal host, the amounts of 2-methyl-propanol and 3-methylbutanal strongly differed in co-cultures with *R. solani*. The host released a maximal amount of approx. 260 ppb 2-methyl-propanol after 91.5 h in a light-independent way, whereas it did not produce 3-methylbutanal upon axenic cultivation. The course of 2-methyl-propanol and 3-methylbutanal emission upon confrontation of *T. atroviride* with *R. solani* was similar to the axenic *T. atroviride* cultures. However, 2-methyl-propanol levels were enhanced in co-cultures involving *T. atroviride* P1 compared to the respective axenic cultures (Figure 6; row 1). In contrast, decreased amounts of 3-methylbutanal were released to the headspace by both *T. atroviride* strains upon confrontation with *R. solani* compared to axenic cultures (Figure 6; row 2).

**Figure 6.** Emission levels of light-independently produced VOCs upon co-cultivation of *T. atroviride* P1 or IMI 206040 with *R. solani*. Levels (ppb) of 2-methyl-propanol (row 1) and 3-methylbutanal (row **2**) detected in the headspace of (**A**) axenic cultures of *T. atroviride* P1 (P1), *T. atroviride* IMI 206040 (IMI) and *R. solani* (Rs) or of (**B**) co-cultures of *T. atroviride* P1 or IMI 206040 confronted with *R. solani* (Rs × P1; Rs × IMI) upon cultivation on PDA in Schott bottles at 25 ◦C. GC–IMS measurements were conducted along a cultivation period of 120 h. As the given volatiles were produced in similar amounts under light-dark conditions and in complete darkness, results shown are means ± SD (*n* = 4).

Upon co-cultivation of *T. atroviride* with *F. oxysporum*, 1-propanol, 2-heptanone, and 3-octanone were produced in different amounts compared to the respective axenic cultures. In addition, 2-octanone was exclusively detected upon co-cultivation.

When grown in axenic culture, *F. oxysporum* constantly released approx. 55 ppb 1-propanol in a light-independent manner, whereas 2-heptanone, 3-octanone, and 2-octanone could not be detected under these conditions. However, confrontations of both *T. atroviride* strains with *F. oxysporum* led to the release of 2-octanone, which was not produced under any other condition tested in this study. Interestingly, 2-octanone emission during co-cultivation of strain P1 with *F. oxysporum* was significantly higher compared to the IMI 206040—*F. oxysporum* pairing. 2-octanone release started after 68.5 h of growth, peaked after 91.5 h with approx. 25 ppb in the co-culture involving strain P1 and with 4 ppb in the co-culture with strain IMI 206040 and afterwards decreased again (Figure 7). 1-propanol levels were increased, especially between 68.5 and 96.5 h of growth upon co-cultivation of *T. atroviride* P1 or IMI 206040 and *F. oxysporum* compared to the respective axenic cultures. In contrast to axenic cultures, which peaked between 40 and 50 h of cultivation, all co-cultivations with *F. oxysporum* resulted in high initial levels, which then constantly decreased over the cultivation period. (Figure 8; row 1).

**Figure 7.** Exclusive, light-independent production of 2-octanone in direct confrontation of *T. atroviride* P1 and IMI 206040 with *F. oxysporum*. Levels (ppb) of 2-octanone detected in the headspace of co-cultures of *T. atroviride* P1 or IMI 206040 with *F. oxysporum* (Fo × P1; Fo × IMI) grown on PDA in Schott bottles at 25 ◦C. GC–IMS measurements were conducted along a cultivation period of 120 h. As 2-octanone emission in co-culture reached similar levels under light-dark conditions and in complete darkness, results shown are means ± SD (*n* = 4).

Interestingly, 2-heptanone emission by *T. atroviride* P1 in co-culture with *F. oxysporum* was light-dependent. In dark-grown co-cultures, 2-heptanone production by strain P1 followed a completely different course compared to co-cultivation under light–dark conditions, IMI 206040—*F. oxysporum* pairings, and axenic cultivations. 2-heptanone levels peaked with approx. 550 ppb after 68.5 h of co-cultivation in darkness and afterwards rapidly declined again. In contrast, 2-heptanone concentrations in the headspace of co-cultures grown under light–dark conditions increased steadier and earlier than upon axenic growth and finally reached and stayed at ≥1000 ppb (Figure 8; row 2). A similar trend was evident in the co-culture of strain IMI 206040 and *F. oxysporum*.

Similar to axenic cultivation, 3-octanone was solely produced in the presence of light by strain IMI 206040, irrespective of the presence of *F. oxysporum*. In axenic cultures, 3-octanone release increased over the whole cultivation period. In contrast, co-cultivation with *F. oxysporum* resulted in maximal 3-octanone production after 91.5 h, although 3-octanone levels were about 50% reduced compared to axenic cultivation. A massive reduction in 3-octanone emission upon interaction with *F. oxysporum* was also evident in strain P1, especially upon cultivation in darkness. However, 3-octanone production in co-culture was triggered by light as strain P1 produced nearly the double amount of this volatile in co-culture with *F. oxysporum* under light-dark conditions compared to constant darkness (Figure 8; row 3).

**Figure 8.** VOC production by *T. atroviride* P1 and IMI 206,040 upon interaction with *F. oxysporum*. Levels (ppb) of 1-propanol (row 1), 2-heptanone (row 2), and 3-octanone (row 3) detected in the headspace of (**A**) axenic cultures of *T. atroviride* P1 (P1), *T. atroviride* IMI 206040 (IMI), or *F. oxysporum* (Fo) or of (**B**) co-cultures of *T. atroviride* P1 or IMI 206040 with *F. oxysporum* (Fo × P1; Fo × IMI) on PDA in Schott bottles at 25 ◦C. Cultures were incubated either under light-dark conditions (LD) or in complete darkness (DD) as indicated. GC–IMS measurements were conducted along a cultivation period of 120 h. Results shown are means ± SD (*n* ≥ 3).

#### **3. Discussion**

*Trichoderma atroviride's* natural habitat is soil and soil itself is penetrated by light in its upmost layer (with the exception of far-red light) solely. From the ecological point of view, the observed enhanced radial growth of *T. atroviride* colonies upon incubation in complete darkness indicates that dark conditions were closer to the conditions prevailing in the natural habitat than direct exposure to light. In accordance with this assumption, both *T. atroviride* strains exhibited a higher ability of antagonizing *R. solani* and particularly *F. oxysporum* in darkness compared to light–dark conditions, under which the latter fungal host could not be overgrown. The faster growth of *T. atroviride* IMI 206040 compared to strain P1 could account for its higher mycoparasitic activity in light. On the other hand, the observed higher antagonistic potential of *T. atroviride* P1 against *F. oxysporum* under complete darkness indicates that this strain is a stronger mycoparasite of *Fusarium* than *T. atroviride* IMI 206040, especially since its radial growth rate was lower. Our results point out that the frequently applied light–dark cycle, exposing the fungi for 12 h to illumination with white light, is an artificial and stressful condition that results in reduced growth and reduced mycoparasitic activity. Since under natural conditions the contact with light usually is accompanied by higher temperatures, drought and exposure to mutagenic ultraviolet light, paralleled by an extensive onset of reactive oxygen species production [29,30], the observed light-triggered growth reduction seems plausible. A strategy to react to and protect from light is conidiation, which was reported to be induced by (a pulse of blue-) light as well as mechanical injury in *T. atroviride* IMI 206040 [31]. Surprisingly, we found that *T. atroviride* P1 similarly produces asexual spores in the presence as well as absence of light, which is in massive contrast to the lack of sporulation in darkness in strain IMI 206040. Differences between the two tested *T. atroviride* strains were also observed regarding their response to mycelial injury. While injury led to the formation of strongly conidiating hypertrophic scarring tissue along the cutting sites in strain P1, the injury response of strain IMI 206040 resulted in only slight conidiation along the cutting lines without an obvious formation of scarring tissue. In combination with its high resistance against several fungicides [32], the light-independent conidiation behavior of *T. atroviride* P1 could be a major advantage for its application as a biocontrol agent in the field, since conidia, which are the main form of application, can be produced in a straightforward manner. Furthermore, belowground conidiation could also be a major advantage for long-term persistence of the mycoparasite in soil.

VOCs are playing important and versatile ecological roles in various intra- and inter-species or even inter-kingdom interactions [2,6]. A well-known example of such "multi-purpose" fungal VOCs are the eight-carbon volatiles 1-octene-3-ol and 3-octanone, which are contributing to the characteristic mushroom aroma and therefore are applied as food odorants in industry [33]. Furthermore, 1-octen-3-ol is known to act as mosquito attractant [34] and both eight-carbon volatiles were found to endogenously regulate conidial germination as well as the induction and inhibition of conidiation in a concentration-dependent manner [18,35–37]. Since *T. atroviride* IMI 206040 does not conidiate under dark conditions, the complete absence of 3-octanone in all samples obtained from dark-grown cultures of this strain indicates that 3-octanone is exclusively produced by conidia or under conidiation-inducing conditions. The substantially higher release of the majority of VOCs detected in this study by *T. atroviride* P1 suggests that this strain has a higher potential for biotechnological VOC production than strain IMI 206040. As described above, the versatile ecological roles of several fungal VOCs are well known; however, the regulation of their production along cultivation, like in dependence of culture age, and in response to stress is sparsely understood. In plants, VOC biosynthesis is regulated by abiotic stress, in particular by light and temperature stress, as well as biotic stress [38,39]. Similar to plants, fungi are relatively immobile and a regulation of VOC production as a stress response or coping strategy therefore is very likely. In this study, we found the three VOCs 3-octanone, 2-methyl-butanol, and 3-methyl-1-butanol to be released by *T. atroviride* axenic cultures in a light-dependent manner. Since the natural habitat of *T. atroviride* is soil, light and in particular UV irradiation is a stressful condition leading to the release of reactive oxygen species and to asexual reproduction [28,29]. In that respect, the light-dependent emission of 3-octanone, 2-methyl-butanol, and 3-methyl-1-butanol could

play a role in the abiotic stress response to light in *T. atroviride*. Interestingly, 2-heptanone production by *T. atroviride* P1 during co-culture with *F. oxysporum* was also regulated in a light-dependent manner. Compared to *R. solani*, *F. oxysporum* is a more challenging host fungus for *T. atroviride*, which in our study could only be fully overgrown by strain P1 under dark conditions. In contrast, strain IMI 206040 was unable to fully parasitize *F. oxysporum*, irrespective of the applied light regime. In dark-grown co-cultures of *T. atroviride* P1 and *F. oxysporum*, only low 2-heptanone levels were emitted, which peaked and then rapidly decreased again. In contrast, confrontation of strain P1 with *F. oxysporum* in the presence of light as well as IMI 206040—*F. oxysporum* co-cultures and axenic cultivation of both *T. atroviride* strains in darkness and light resulted in a constant increase of 2-heptanone emission reaching and staying at levels of ≥1000 ppb. Low 2-heptanone levels hence paralleled the successful parasitism of *F. oxysporum* by *T. atroviride* P1, while the light-triggered emission of high levels of this volatile was associated with difficulties of *T. atroviride* to overgrow and lyse this host. *F. oxysporum* spp. are known to produce a plethora of toxic secondary metabolites [40,41], of which e.g., fumonisin is increasingly produced in the presence of light [42]. 2-hepanone emission by *T. atroviride* could be part of a coping strategy against abiotic (light) and biotic (*F. oxysporum* mycotoxins) stress.

*F. oxysporum*—*T. atroviride* co-cultivation further resulted in the production of 2-octanone, a volatile that was not emitted by any of the interaction partners upon axenic growth. Interestingly, a recent study revealed that deletion of the histone deacetylase-encoding gene *hda-2* in *T. atroviride* IMI 206040 leads to the biosynthesis of 2-octanone as a new volatile in this strain [43]. This report together with our results points out that 2-octanone biosynthesis genes could be shut-down by a repressive chromatin structure and could be activated epigenetically as a consequence of environmental stimuli, in this case the interaction with *F. oxysporum*. Accordingly, the bacterium *Streptomyces rapamycinicus* triggered the expression of the silent gene cluster responsible for orsellinic acid production in *Aspergillus nidulans* by activation of the histone acetyltransferase GcnE [44].

In summary, our study revealed significant differences between the two tested *T. atroviride* strains regarding injury-response, light-dependent regulation of conidiation and mycoparasitic activity as well as VOC production. Besides volatiles with strain- and light-dependent biosynthesis patterns, we found an interaction-dependent regulation of the biosynthesis of individual substances and obtained evidence, that the biosynthesis of certain VOCs is specifically linked to abiotic and/or biotic stress responses in *T. atroviride*. Additional effort should be undertaken to further decipher the multifactorial regulation of secondary metabolite biosynthesis in *Trichoderma* mycoparasites.

#### **4. Materials and Methods**

#### *4.1. Fungal Strains and Culture Conditions*

*Trichoderma atroviride* P1 (ATCC 74058), *T. atroviride*IMI 206040, *Fusarium oxysporum* f. sp. *lycopersici* strain 4287 and *Rhizoctonia solani* strain AG-5 were used in this study.

Pre-cultivation was done by passaging a 6 mm diameter agar plug of the actively growing colony margin at least two times after 1.5 days each upside down to the centre of a fresh petri dish (94 mm × 16 mm, Greiner Bio-One GmbH, Kremsmünster, Austria) containing 25 mL potato dextrose agar (PDA; Becton, Dickinson and Company, Le Pont De Claix, France). Plates were incubated at 25 ◦C under light-dark conditions (12:12 h, 300 Lux; Snijders Micro Clima-Series TM Labs Economic Lux Chamber; Snijders Labs, Tiburg, The Netherlands) or under complete darkness.

For determination of the radial growth rate of the two *T. atroviride* strains, 6 mm diameter agar plugs of the actively growing colony margins from the pre-cultures were inoculated in quadruplicates at the outmost margin of fresh agar plates. Plates were incubated at 25 ◦C under light-dark conditions (12:12 h; 300 Lux) or under complete darkness for three days. The radial growth rate was measured and calculated (cm/d) after three days of growth.

#### *4.2. Mechanical Injury- and Blue-Light-Induced Conidiation Assays*

Mechanical injury- and blue-light-induced conidiation assays were performed in quadruplicates according to [31] with slight modifications. Six millimeter diameter agar plugs of the actively growing colony margins from the pre-cultures were inoculated upside-down at the centre of fresh agar plates. Cultures were incubated at 25 ◦C for five days under light-dark conditions or under complete darkness. After two days of incubation in complete darkness, conidiation was induced by mycelial injury, which was set by cutting five horizontal and five vertical lines with a scalpel under red safety light. Alternatively, a 10 min pulse of blue light was set with a blue light source (Blacklight Blue UV-A Lamp: Supratec L18 W/73, 300–400 nm; 25 Lux; Osram GmbH, Garching, Germany; distance 19 cm). Plates were further incubated under complete darkness for a total of five days. Photos were taken after five days of growth.

#### *4.3. Dual Plate Confrontation Assays*

Dual plate confrontation assays were performed in quadruplicates according to [30] with slight modifications. Six millimeter diameter agar plugs of the actively growing colony margins from the pre-cultures of *T. atroviride* and a respective host fungus were inoculated on opposite sides at the outmost margin of fresh agar plates. As controls, all strains were additionally grown alone; the mycoparasites were also grown in self-confrontations. Due to its slower radial growth rate, *F. oxysporum* was inoculated 24 h earlier than the other strains. Fungi were grown at 25 ◦C under light-dark conditions or under complete darkness for five and six days, respectively. Pictures were taken at the end of the cultivation period to document the progress of the mycoparasitic attack.

#### *4.4. Gas Chromatography–Ion Mobility Spectrometry (GC–IMS) Analysis of Headspace VOCs*

For gas chromatography–ion mobility spectrometry (GC–IMS) analysis, a 6 mm diameter agar plug of the actively growing colony margin from the pre-culture was inoculated upside-down at the outmost margin of a 150 mL glass bottle (Duran GmbH, Mainz, Germany) filled with 25 mL PDA at the bottom. Fungi were grown in axenic culture or direct confrontation in the glass bottles according to the description above. As a control, PDA was also measured alone. The bottles were closed with Teflon® screw caps (BohlenderTM, Merck, Vienna, Austria) with two openings for air in- and outlet. For dark conditions, the bottles were covered with several layers of aluminum foil. Due to a slower radial growth rate, *F. oxysporum* was pre-grown for 24 h. The two *T. atroviride* strains and *R. solani* were pre-grown for 20 h at 25 ◦C under light-dark conditions (12:12 h; 300 Lux) or under complete darkness before flasks were connected to the MS device.

For headspace measurements, cultures in glass bottles were held in an incubator and connected in a gastight way, parallel to each other. To avoid condensation, the inner temperature of the oven (so the temperature of the headspace air) was held at 40 ◦C, while the water bath was kept at 23 ± 2 ◦C. For ventilation, purified air with 5 mL/min flow was continuously streaming through the flasks and regulated by four mass flow controllers (MCF, Bronkhorst, Ruurlo, The Netherlands). Additionally, 5 mL/min purified air was connected as dilution flow to the flasks to reduce moisture in the samples. Headspace air samples were collected in the incubator at 40 ◦C in 100 mL and 250 mL glass syringes (Socorex, Ecublens, Switzerland). Sampling and measurement were performed 21, 24.5, 43.5, 48.5, 68.5, 72.5, 91.5, 96.5 and 115.5 h after inoculation.

A high resolution GC–IMS developed at Leibniz Universität, Hannover was applied to monitor the emitted VOCs. Samples were injected immediately after collection through a heated inlet (40 ◦C) into the GC column using a stainless steel sample loop (200 μL) installed on a six-way valve (VICI AG International, Schenkon, Switzerland). Volatiles were separated using a RTX volatiles column (10 m × 0.53 mm × 2 μm, Restek GmbH, Bad Homburg, Germany) working at a constant temperature of 50 ◦C. The carrier-gas-flow-rate program was as follows: 3 mL/min for 10 min and then 10 mL/min for another 10 min, resulting in a total GC-runtime of 20 min. The IMS, with a drift tube length of 7.5 cm, provided a resolving power of R = 90 using a drift voltage of 5 kV. The instrument operated at 40 ◦C with purified air as drift gas at the flow of 150 mL/min and 10 mbar above the ambient pressure. For ionization of the volatiles, a radioactive β-emitter 3H (300 MBq) was used. A detailed description of the system can be found in [31]. Chromatographic data was acquired using the Agilent Chemstation Software (GC-MS Data Analysis from Agilent, Waldbronn, Germany). Data were analyzed using the software OpenChrom (vers. 1.4.0, Lablicate GmbH, Hamburg, Germany) and the mass spectrum library NIST 2008 (Gatesburg, PA, USA) was applied for identification.

**Author Contributions:** V.S. conceived and directed this study. V.S. and S.Z. drafted the manuscript. V.S., M.W. and V.R. conducted the experiments and analyzed the data. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We thank Christopher Mayhew and Helmut Wiesenhofer for assisting in the GC–IMS analysis and Stefan Zimmermann from the Institute of Electrical Engineering and Measurement Technology, University of Hannover for providing the GC–IMS.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Sample Availability:** Not available.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
