Inhibition of Penicillium digitatum and Citrus Green Mold by Volatile Compounds Produced by Enterobacter cloacae

Penicillium digitatum causes green mold decay on citrus fruit, resulting in severe economic losses to citrus growers and packers worldwide. The present study is to evaluate the control of citrus green mold by volatiles produced by Enterobacter cloacae. An E. cloacae strain isolated from plant rhizospheres was able to produce three volatile organic compounds, which were identified as butyl acetate, phenylethyl alcohol, and 4,5-dimethyl-1-hexene by GC/MS chromatography. The volatile compounds produced by E. cloacae inhibited conidial germination and hyphal elongation of P. digitatum and reduced green mold severity. E. cloacae cultured at temperatures ranging from 16°C to 28°C, at pH values ≤6, or in a substrate carrier (sphagnum moss, vermiculite, or perlite) provided superior control against P. digitatum. A laboratory formulation using E. cloacae and perlite protected citrus fruit from green mold up to 22 days and its effectiveness outperformed fungicide application at room temperature (~25°C). The results implicate practical application of E. cloacae as a biofumigant for controlling citrus postharvest decay caused by P. digitatum. Significantly, the study provides a model for future research on how to formulate an effective biocontrol agent for disease management.


Introduction
Green mold, caused by Penicillium digitatum (Pers.: Fr.) Sacc., is a noxious postharvest disease of citrus. Penicillium digitatum is an opportunistic pathogen that resides on healthy citrus fruit and attacks citrus fruit through injuries caused by rough handling during harvesting, transportation and storage. Infection of P. digitatum in citrus often results in tissue maceration and fruit decay. Economic losses caused by green mold decay could be enormous for citrus growers and packers worldwide. Application of fungicides, such as imazalil and thiabendazole, is a common practice to control P. digitatum induced fruit decay in the packinghouse [1][2][3][4]. However, repeated use of toxic chemical compounds could induce the emergence of fungicide resistant strains of the pathogen and also could increase human health risks.
Enterobacter cloacae is a rod-shaped, gram-negative bacterium commonly found in plants and has been shown to control a wide range of plant pathogens [16][17][18][19][20][21][22] and insect pests [23]. Strains of E. cloacae have been known to produce hydroxamate siderophore, non-volatile metabolites and inorganic volatile substances such as ammonia, all of which might contribute to biocontrol activity [19,24]. Enterobacter cloacae has been shown to inhibit Pythium ultimum sporangium germination and damping-off by competing for fatty acids [25]. Biological control using volatile compounds-producing microorganisms has been reported with increasing success on the various fruit crops and diseases [26][27][28][29][30][31][32][33][34]. However, little is known about volatile compounds produced by E. cloacae. The objectives of this study were to examine whether or not E. cloacae will emit volatile compounds and to assess their antimicrobial activity against P. digitatum and control of green mold of citrus fruit in storage.

Biological materials and culturing conditions
The P-51 and DOB-2 strains of P. digitatum (Pers. Fr.) Sacc. were single spore isolated from diseased orange fruit grown in Fongyuan and Dalin (Taiwan), respectively. DOB-2 strain has been characterized elsewhere [35]. The affected fruit were incubated in plastic boxes until conidia were produced. Conidia were picked and transferred onto 2% sucrose water agar. Fungal strains were cultured on potato dextrose agar (PDA; Difco) in a growth chamber with 12 h daily illumination at 25°C to 26°C. Fungal strains were identified as P. digitatum based on the distinct characteristics of conidia and colony morphologies formed on malt extract agar (MEA), Czapek yeast autolysate agar (CYA), and glycerol nitrate agar (GN25N) as described by Pitt [36]. The identity of P. digitatum was further confirmed by sequence analysis of the 5.8S ribosomal internal transcribed spacers (ITS).
Bacterial strains were single-colony isolated from rhizospheres of Yardlong bean (Vigna sesquipedalis (L.) Fruwith). Bacterial cells were grown on King's medium B (KB) plate at 30°C for 2 days, harvested by low speed centrifugation (4000× rpm), suspended in 20% glycerol water, and stored at -80°C. Bacterial strains were identified as E. cloacae by sequence analysis of an rpoB gene encoding a RNA polymerase β-subunit.

Identification of and fungal bacterial strains
Microbial genomic DNA was extracted with phenol/chloroform partition and precipitated with isopropanol. Fungal 5.8S rDNA was amplified by PCR with the primers ITS1 (5'-TCCGTAGGTGAACCTGCGG-3') and ITS4 (5'-TCCTCCGCTTATTGATATGC-3') as described by White and colleagues [37]. Bacterial DNA fragment was amplified with colony PCR as described by Fukui and Sawabe [38] using a PCR Master Mix kit (GeneMark, Taichung, Taiwan). Bacterial rpoB gene was amplified by PCR with the primers CM7 (5'-AACCAGTTCCGCGTTGGCCTGG-3') and CM31b (5'-CCTGAACAACACGCTCGGA-3') as described by Mollet and colleagues [39]. PCR fragments were directly sequenced. Sequences were searched against the databases at the National Center for Biotechnology Information (NCBI) using the BLAST network service to determine the similarity of the amplified fragments. Sequence data reported in this article have been deposited in the EMBL/GenBank Data Libraries under accession numbers AJ543728 (E006 rpoB) and AJ543682 (E010 rpoB).

Volatile antimicrobial assays
In-vitro volatile antimicrobial activity was assessed using a plate-toplate method as described [27,40]. Bacterial cells grown on KB agar at 24°C for 24 h were used for antifungal activity assays. Petri dish lips were removed and the plate was turned upside down and attached onto a coverless PDA plate containing conidia or an agar plug bearing 2day-old mycelium of P. digitatum. The gap between plates was tightly sealed with two layers of parafilm (American National Can, Chicago, USA) and the plates incubated at varying temperatures. Spore germination was determined microscopically 24 h after treatment. Radial growth of fungus was measured at 7 days after treatment. A KB agar plate without E. cloacae and a PDA plate with P. digitatum were used as mock controls. Fungal viability was assessed after treatments by transferring agar plugs covering with P. digitatum mycelium onto freshly prepared PDA. Percentage of growth inhibition of P. digitatum was determined by dividing the relative difference of the growth between mock control and treatment by the control growth and then multiplying by 100. Percentage of inhibition = [(A-B)/A] × 100, where A is the average diameter of P. digitatum without being challenged with E. cloacae (mock control); B is the average diameter of P. digitatum grown in the presence of E. cloacae. Effect of volatile compounds on spore germination was also assessed using P. digitatum conidia. Unless otherwise stated, all experiments were repeated at least twice with three replicates.

SPME/GC-MS identification of volatile compounds
Enterobacter cloacae strain was grown on KB plate or in perlite mixed with KB broth at 30°C for 2 days. Volatile compounds were extracted by a headspace solid-phase microextraction (HS-SPME) device (SUPELCO, Bellefonte, PA, USA) equipped with 50/30 μm Divinylbenzene/Carboxen/Polydimethylsiloxane (DVB/CAR/PDMS, SUPELCO) fibers. After 15 min, the HS-SPME device was removed and inserted into the injector (200°C) of a gas chromatograph (GC)mass spectrophotometer (MS) equipped with a Model CP-3800 GC and a Saturn 2000 MS (Varian, USA) connecting to an electron capture detector (Model 902B, ECD, splitless mode). Nitrogen was used as a gas carrier and flowed at 0.2 mL/min through a VF-5MS capillary column (30.0 m × 0.25 mm ID, 0.25 μm film thickness, Agilent, Santa Clara, CA, USA). Analytical temperatures modified from Kai et al., [41] were set as following: Initial column temperature at 40°C for 2 min, followed by a ramp of 5°C/min up to 170°C with a final hold for 1 min at 170°C. KB medium without bacteria was used as controls. The identities of volatile compounds were verified by comparing them with chemical databases deposited in the GC-mass spectra (GC-MS) library (Saturn 2000, USA). Phenylethyl alcohol (PEA) and butyl acetate (BA) were purchased from Sigma-Aldrich (St. Luis, MO, USA). The toxicity of commercially available chemicals individually or in combination on the growth of P. digitatum also was assessed by modified plate-to-plate assays by placing a test compound onto a filter paper disc (0.5 cm in diameter).

Fungal inoculation and assays for disease severity
Fruit of 'Gold Seal' orange (Citrus sinensis Osbeck), showing no visible lesions or injuries, were collected from an orchard in Gukeng (Yunlin county, Taiwan) and used throughout the experiment. Citrus fruit stored at 11°C for 5 to 7 days were washed by immersing in distilled water, surface sterilized with 70% alcohol, and dried in a sterile hood. Conidia of P. digitatum were harvested by flooding with sterile water containing 0.01% Tween 20 and the concentration was adjusted to 10 4 conidia per milliliter with the use of a haemocytometer. Orange fruit were wounded with 2 mm long by 1 mm wide tips (10) mounted on a steel rode as described [30] prior to inoculation. The wounded fruit were immersed in conidial suspensions or water for 10 s, dried, and placed in plastic boxes (Lock&Lock, HanaCobi, Korea) for lesion development with or without E. cloacae (1.2-1.8 × 10 8 cfu/mL) mixed in sterilized substrates. Fruit were soaked in a benomyl (200 ppm) or thiabendazole (2000 ppm; Mertect 41.8% SC; Sinon, Taiwan) solution for 30 min prior to inoculation as controls. Commercially available substrates: sphagnum moss (YuGuang Garden, Taichung, Taiwan), vermiculite (NanHai Co., New Taipei City, Taiwan), peat moss (BVB no. 4, Bas Van Buuren, Maasland, New Zealand) and perlite (NanHai) were sterilized and added to different concentrations of KB. A 5-fold KB stock (each liter contains 15 g peptone, 1.125 g K 2 HPO 4 , 1.125 g MgSO 4 ．7H 2 O, and 7.5 ml glycerol) was prepared, diluted, and mixed with substrates. Disease severity was rated in each fruit according to the following scale: 0 = No symptoms; 1 = Less than 25% of fruit surface showing symptoms; 2 = 26 to 50% of fruit surface showing symptoms; 3 = More than 50% of fruit surface showing. Percentage of disease severity was calculated from the disease rating by the formula: Disease Severity % = ∑ Rating no. × the no. of fruits Total no. of fruits × 3 × 100 The experiment was performed three times; each replicate contained 10 fruit. The significance of treatments was determined by analysis of variance (ANOVA) and treatment means separated by nonlinear regression and Durbin-Watson Statistic normality tests using SigmaPlot 10 (Systat Software, San Jose, CA).

Result Fungistatic effects of volatile compounds produced by E. cloacae
Pathogenicity assays revealed that both fungal strains were highly pathogenic to citrus (data not shown). Using plate-to-plate assays, we screened over 200 bacterial isolates and found that E. cloacae isolate E010 displayed inhibitory effects on vegetative growth and conidial germination of both P. digitatum P51 and DOB-2 strains (Table 1 and data not shown). Because E. cloacae and P. digitatum were cultured on different agar plates and never physically contacted each other, the inhibitory effects observed in P. digitatum were likely attributable to the production of volatile compounds that were toxic to fungus. Volatile compounds produced by the E006 strain had less effect on fungal growth and conidial germination compared to those produced by the E010 strain. Although volatile compounds produced by the E006 strain effectively suppressed the germination of conidia produced by P. digitatum P51 isolate, the compounds had little effect on conidial germination of DOB-2 isolate. Thus, E010 strain was chosen for further analyses. Exposure of P. digitatum to E. cloacae resulted in abnormal growth of fungal colony, producing more aerial hyphae. The effect of volatile compounds to P. digitatum was fungistatic as hyphae resumed normal growth after transferring onto freshly prepared PDA (data not shown).

Effect of environmental conditions used to grow E. cloacae on antifungal activity
To assess if temperature would affect E. cloacae's ability to inhibit conidial germination, E010 strain was grown at temperatures varying between 16 and 32°C for 24 h. Conidia prepared from P. digitatum P51 and DOB-2 isolates were placed on PDA and the plates were attached onto the E010 plates using plate-to-plate assays and incubated at respective temperatures. In the absence of bacterial culture, the efficacy of conidial germination measured in both P51 and DOB-2 isolates increased exponentially at temperatures between 16 and 24°C and decreased sharply at temperatures greater than 28°C ( Figure 1A). Penicillium digitatum conidia failed to germinate efficiently while grown at 32°C. The rate and magnitude of conidial germination reduced considerably when the agar plate carrying E010 strain was attached to the plates culturing fungal strains. In general, conidia of both P. digitatum strains germinated poorly when co-cultured with E010 bacterium grown at temperatures greater than 20°C. The pH values of medium used to culture E. cloacae also affected its efficacy of antifungal activity. E010 strain grown at acidic pH ≤ 6 displayed a strong inhibition on conidial germination of P. digitatum ( Figure 1B).
The inhibitory effect on conidial germination decreased markedly when E010 was cultured on medium with pH values greater than 6.

Control of citrus green mold by E. cloacae
Experiment was undertaken to evaluate if E. cloacae would be useful for controlling citrus green mold caused by P. digitatum and if substrates used to culture bacterium would impact E. cloacae's ability to inhibit test fungi. Increasing the amounts of fungal inoculum greatly increased the severity of green mold on citrus (Figure 2A).
Enterobacter cloacae and conidia suspended in KB or H 2 O were mixed with sphagnum moss, vermiculite, peat moss, or perlite and incubated in a plastic box with test citrus fruit ( Figure 2B). The results revealed that the substrates used to support the growth of E. cloacae had a profound impact on the effectiveness of reducing green mold. Fruit inoculated with P. digitatum developed green mold at 7 days post inoculation (dpi) in all treatments without exogenous application of E. cloacae (Table 2).   In order to optimize E. cloacae-substrate formulation for controlling citrus green mold, we further tested the effectiveness of a combined E010-KB-perlite (EKP) formula on the occurrence of the disease. Green mold severity decreased significantly as the volume of EKP increased ( Figure 3A). Decreasing the ingredient concentrations in KB medium and then, mixing with perlite and E010 decreased the green mold severity substantially ( Figure 3B). Using 5-fold KB in EKP (EK 5 P) citrus fruit were protected from green mold up to 22 days at room temperature (~25°C) ( Figure 4A); whereas EKP with 2.5-fold KB ingredients (EK 2.5 P) provided less protection over time ( Figure 4B). Further studies revealed that EK 5 P reduced green mold incidence in a wide range of test temperatures (from 8 to 20°C) ( Figure 4C). Uninoculated fruit never developed green mold symptoms during the course of the experiment (data not shown). Citrus fruit soaked in benomyl (200 ppm), thiabendazole (2000 ppm), or water prior to inoculation failed to provide any protection from P. digitatum, whereas application of EK 5 P reduced green mold development substantially ( Figure 4D).

GC/MS identification of volatile compounds produced by E. cloacae
Volatile compounds produced by E. cloacae grown on KB medium were identified after gas trapping and GC/MS separation ( Figure 5). Compounds commonly identified from the gas phase of both the control and the E. cloacae-growing culture were eliminated. The volatile substances 4,5-dimethyl-1-hexene and phenylethyl alcohol uniquely found in the bacterium-grown medium were identified by comparing them in the GC-mass spectra (GC-MS) database library. Apart from these two compounds, butyl acetate was identified from gas phase of E. cloacae grown on KB medium mixed with perlite ( Figure 5). The identity of phenylethyl alcohol and butyl acetate was further verified by comparison with authentic standards that were commercially available. Although 4,5-dimethyl-1-hexene is commercially available, we were unable to obtain this compound due to a shipping restriction of the product.

Suppression of P. digitatum and citrus green mold by volatile compounds
Co-incubation of P. digitatum P51 or DOB-2 isolate with commercially available butyl acetate or phenylethyl alcohol alone at test concentrations had little or no effect on conidial germination (data not shown) and radial growth ( Figure 6). . Mock control (CK) contained KB medium only. E010 strain was grown in KB broth at 30°C for 2 days. Volatile compounds were collected using a solid-phase microextraction (SPME) device and analyzed by GC/MS chromatography.
A combination of butyl acetate and phenylethyl alcohol had a moderate effect on the suppression of conidial germination prepared from P. digitatum P51 strain, but inhibited the germination by nearly 40% of conidia prepared from P. digitatum DOB-2 strain. Mixture of butyl acetate and phenylethyl alcohol suppressed radial growth of both P51 and DOB-2 strains ( Figure 6). When tested for green mold control, application of phenylethyl alcohol alone did not provide apparent protection against citrus green mold; however, application of butyl acetate alone or in combination with phenylethyl alcohol provided low levels of protection of citrus fruit from green mold (Figure 7). Notably, butyl acetate and phenylethyl alcohol had no synergistic effects in the context of disease reduction. Lack of commercially available 4,5-dimethyl-1-hexene prevented us from testing its toxicity toward P. digitatum.

Discussion
In Taiwan, benzimidazole is the primary fungicide recommended to control postharvest citrus diseases and more than 97% of P. digitatum isolates were found to be resistant to this fungicide [35]. In the present study, we identified an E010 bacterium collected from rhizospheres of Yardlong bean as E. cloacae based on sequence analysis of an rpoB gene and have demonstrated its ability of producing antimicrobial volatile compounds, as well as its potential use as a biocontrol agent for controlling citrus green mold. Figure 6: Effects of commercially available compounds: butyl acetate (BA) and phenylethyl alcohol (PEA) individually or in combination on vegetative growth of P. digitatum strains (P51 and DOB-2). Fungal strains were placed on PDA and the plates were attached to petri dish plates containing a filter paper disc with a test compound. Experiments were repeated three times with 3 replicates for each treatment. Only representatives are shown. Figure 7: Effects of commercially available compounds: butyl acetate (BA) and phenylethyl alcohol (PEA) individually or in combination, on the reduction of citrus green mold caused by P. digitatum. Filter paper discs spotted with various with chemicals were placed in a petri dish along with citrus fruit inoculated with conidial suspension of the P. digitatum P51 strain and incubated in a moist box for 7 days at room temperature.
The chemical properties and the biological activities of volatile compounds produced by E. cloacae were characterized to be inhibitory to fungal pathogen P. digitatum causing citrus green mold. The environmental conditions that might influence the production of volatile compounds by E. cloacae and its effectiveness of controlling green mold were also investigated. Temperatures and substrates used to culture E. cloacae and pH values of medium all affected P. digitatum-induced disease severity on citrus fruit. The efficacy of a well-known volatile antimicrobial producer Muscodor albus as a biofumigant also is affected by temperatures [42,43]. Nevertheless, our results provide valuable information in determining the importance and practicability of utilizing E. cloacae as a biocontrol means in controlling citrus green mold.
Using plate-to-plate assays, it was noticed that E. cloacae grown at temperatures ranging from 16°C to 28°C was able to suppress the germination of conidia produced by P. digitatum. Green mold incidence on citrus stored at temperatures ranging from 8°C to 20°C also was suppressed using E010-KB-Perlite (EKP) formulation. Enterobacter cloacae grown in acidic pH tended to exhibit stronger toxicity toward P. digitatum than that grown in alkaline conditions. This could be related to bacterial growth as E. cloacae grew much better in acidic environments [44]. We found that the composition of the medium can impact the bacterial growth and the type of volatile compounds produced by E. cloacae. In the end, a combination by growing E. cloacae E010 strain in 5-fold KB medium and perlite (EK 5 P) was found to be the most effective formula for controlling green mold. EK 5 P seems to provide much protection against green mold, keeping citrus fruit from the disease up to 3 weeks. Perlite is a volcanic stone comprising silicon dioxide, ammonium oxide, and many other minerals, and has been widely used in the industry and agriculture sectors. Owing to its high permeability and low water retention, perlite is often used as a soil amendment and as a carrier for fertilizers, herbicides and pesticides [45].
Enterobacter cloacae is not a prolific producer of volatile compounds. Only an isolate of E. cloacae has been reported to emit ammonium [19]. Although only three volatile compounds were identified from analysis of gas phase of E. cloacae in this study, the bacterium demonstrates its impressive biological activity against P. digitatum and citrus green mold. It is very likely that changing the composition of growth medium may increase the number and the type of volatile compounds produced by E. cloacae as suggested in gasproducing fungus Muscodor albus [28]. Moreover, a combination of three volatile compounds (4,5-dimethyl-1-hexene, butyl acetate and phenylethyl alcohol) is likely required to exert a stronger inhibitory effect against the target test P. digitatum. Phenylethyl alcohol is a common gas released by bacteria and fungi and exerts antimicrobial activities against various plant pathogenic fungi at very high levels or in cooperation with other volatile compounds [33,40,41,[46][47][48]. Phenylethyl alcohol has rose-like flavors and has been demonstrated to induce germination of ascospores, but suppress conidial germination in Neurospora crassa [49,50]. Phenylethyl alcohol affects membrane permeability, causes the alternation of amino acid and sugar transport systems, and inhibits the synthesis of macromoleculars and thus, is inhibitory to microorganisms [51,52]. Like trytophol, phenylethyl alcohol is involved in fungal quorum sensing [53]. Butyl acetate, also known butyl ethanoate, has sweet smell of banana or apple and can be inhibitory to bacteria [54].
Although artificial application of volatile compounds did not fully suppress fungal growth and conidial germination, they mimicked the inhibitory effects to some extent. The differences in activity between the E. cloacae-involved production of volatiles and commercially available products may relate to the quantitative production of the volatile compounds produced by E010, which cannot be determined by our GC/MS methodology. Another explanation for less inhibitory effect of commercially available products on P. digitatum is that volatile compounds exert their maximal toxicity in a cooperative manner. Exogenous application of butyl acetate or phenylethyl alcohol individually had little effect on conidial germination and radial growth of P. digitatum, whereas a combination of both compounds inhibits conidial germination and vegetative growth. The results confirm the synergistic effect of volatile compounds in terms of their toxicity toward P. digitatum. Similar synergism among volatile organic compounds in increasing antimicrobial activity has also been observed [28,55]. The toxicity of 4,5-dimethyl-1-hexene alone to P. digitatum remains uncertain. However, of three volatile compounds identified from gas phase of E. cloacae, 4,5-dimethyl-1-hexene is the predominant compound. Judging from differences in activity between the E. cloacae-producing volatile compounds and commercially available butyl acetate and phenylethyl alcohol, it appears that 4,5dimethyl-1-hexene plays an important role in fungistatic effects against P. digitatum, as well as in suppressing green mold incidence.
Evaluation of the toxicity of 4,5-dimethyl-1-hexene will be the next logical step toward better understanding of how E. cloacae exerts its antimicrobial ability.
Delayed application of EKP could be ineffective in controlling green mold, since volatile compounds emitted by E. cloacae only inhibit but do not kill P. digitatum. Further experiments to determine the optimal growth condition for E. cloacae that provides the most effective measure in controlling citrus green mold in a closed chamber during storage and shipment are warranted. It will also be of interest to determine if volatile compounds produced by E. cloacae can be inhibitory to other plant pathogenic fungi and if EKP formula is useful in controlling other plant diseases. The study also paves a way for future research on how to formulate an effective biocontrol agent for disease management.