Assessing the ability of white-rot fungi to tolerate polychlorinated biphenyls using predictive mycology

ABSTRACT The aim of the present study was to assess the ability of different white-rot fungi to tolerate polychlorinated biphenyls (PCBs) using predictive mycology, by relating fungal growth inhibition to ligninolityc enzyme secretion. Fungal strains were grown in the presence of PCBs in solid media and their radial growth values were modelled through the Dantigny-logistic like function in order to estimate the time required by the fungal colonies to attain half their maximum diameter. The principal component analysis (PCA) revealed an inverse correlation between strain tolerance to PCBs and the laccase secretion over time, being laccase production closely associated with fungal growth capacity. Finally, a PCA was run to regroup and split between resistant and sensitive fungi. Simultaneously, a function associated with a model predicting the tolerance to PCBs was developed. Some of the assayed isolates showed a promising capacity to be applied in PCB bioremediation. Abbreviations: Polychlorinated biphenyls (PCBs), white-rot fungi (WRF)


Introduction
Polychlorinated biphenyls (PCBs) are a group of chlorinated organic pollutants that have been used in a wide range of products in the twentieth century. Due to the harmful effect of PCBs on natural resources, their manufacture, use and export were banned but PCBs persist in the environment mainly because natural soil and aquatic biota are incapable of significantly degrading them (Borja et al. 2005). Although there are costly physical and chemical options for degradation of PCBs, fungal and bacterial bioremediation proved to be an effective and low-cost strategy for the removal of these pollutants. Previous studies not only showed that white-rot fungi (WRF) have the capability to degrade PCBs in liquid, soil and sediments but also demonstrated their adaptive ability to grow under critical environmental conditions (Stella et al. 2017). WRF such as Phanerochaete chrysosporium (Kamei et al. 2006a;Čvančarová et al. 2012), Phlebia brevispora (Kamei et al. 2006b), Coriolus versicolor (Cloete and Celliers 1999;Čvančarová et al. 2012), Irpex lacteus, Bjerkandera adusta, Pycnoporus cinnabarinus, Phanerochaete magnolia, Pleurotus ostreatus (Stella et al. 2017) and Dichomitus squalens (Čvančarová et al. 2012) have already proved their potential for removal of PCBs. WRF possess a ligninolytic enzyme complex comprising, among other enzymes, lignin peroxidase (LiP, EC 1.11.1.14), manganese peroxidase (MnP, EC 1.11.1.13) and laccase (Lac, EC 1.10.3.2), involving as well in the oxidation of a wide range of organopollutants. Diverse studies shown that Lac, MnP and LiP were responsible for efficient degradation of PCBs (Č et al. 1997;Cloete and Celliers 1999;Gayosso-Canales et al. 2012). Other authors documented that these enzymes were involved in PCB bioremediation without establishing nevertheless their clear role in the removal mechanism (Pointing, 2001;Beaudette et al. 1998;Čvančarová et al. 2012).
Nevertheless, exploring new isolates with better degrading capacities would contribute in the field of bioremediation research. Different screening techniques were carried out to select strains with polycyclic aromatic hydrocarbons, dyes and phenols biodegradation capabilities (Levin et al. 2004;Matsubara et al. 2006;Lee et al. 2014;Carabajal et al. 2016). However, plate screening has not been yet evaluated as a quick technique to select strains capable of tolerating and degrading PCBs.
Predictive mycology has emerged as a new trend, dedicated to predict and understand fungal development in different raw materials and food (Dantigny et al. 2005;Dantigny and Bensoussan 2008;Bevilacqua et al. 2016).
The aim of the present study was to select a group of fungi capable of growing in the presence of PCBs, applying predictive mycology as a tool to assess their ability to tolerate PCBs. The tolerance of 26 white-rot fungal strains towards PCBs was compared in agar plates supplemented with PCBs by measuring radial growth rates and was correlated with their ligninolytic enzyme production.

Organism and chemical compounds
Twenty-six white-rot fungal strains previously isolated mostly from the subtropical rainforest of Misiones (Argentina) were used in this work (Table 1) (González et al. 2013;Martínez et al. 2013). These strains are deposited at the Culture Collection of the Faculty of Forestry, Universidad Nacional de Misiones, Argentina, and maintained in malt extract (12.7 g L −1 ) and agar (17.0 g L −1 ) at 4°C.
The PCB mixture in transformed oil with an average of Aroclor 1242, 1254 and 1260 was kindly provided by KIOSHI S.A. (Buenos Aires, Argentina).

Tolerance for PCBs
Fungal strain tolerance towards PCBs was analysed in 90 mm diameter agar plates containing sugarcane bagasse-based medium [1 g (dry weight) sugarcane bagasse powder (120 mesh), 2 g KH 2 PO 4 , 0.5 g MgSO 4 ·7H 2 O, 0.1 g CaCl 2 ·2H 2 O, 0.1 mg FeSO 4 ·7H 2 O and 10 g agar per litre distilled water (Matsubara et al. 2006)] supplemented with 85 mg L −1 of PCBs. Inocula consisted of a 5 mm diameter agar plug taken from a 5-7 day-old fungal colony in malt extract (12.7 g L −1 ) and agar (17.0 g L −1 ). Incubation was carried out at 28°C in darkness. Growth was followed by daily measuring the radial extension of the mycelium, until the complete coverage of the plates.

Ligninolytic enzyme assay
Ligninolytic activity was determined in solid media. All the strains were inoculated on agar plates (12 mL medium/Petri dish) containing malt extract (12.7 g L −1 ), glucose (10 g L −1 ) and agar (20 g L −1 ) (Malt Extract Agar with glucose (MEA)), supplemented either with 2,6-dimethoxyphenol (1 mM) (DMP) (Fonseca et al. 2010) to detect Lac activity, MnCl 2 ·4H 2 O (1 mM) to estimate MnP (Jarosz-Wilkołazka et al. 2002) and Azure B (50 μM) to assay LiP (Archibald, 1992). Inocula consisted of a 5 mm colonised agar plug of a 5-7 day-old culture grown on MEA. Noninoculated plates served as controls for abiotic discoloration. A MEA control plate was also inoculated with each strain. The plates were incubated at 28 ± 1°C, until the complete coverage of the plate. The Lac activity was confirmed when an orange-yellow staining zone appeared, while the MnP activity was observed as a brown staining zone. In the case of LiP activity, a decolorised zone appeared when the fungus degraded the dye. Three replicates were performed for each strain. Daily, the radial extension of the mycelium and the halos resulting from enzyme activity were measured.

Statistical analysis and modelling
where D was the diameter of the fungal colony or the diameter of the halo resulting from enzyme activity, with D max being the maximum diameter (set to 8.5 cm, corresponding to the diameter of the plates); k was the rate of fungal growth or the rate of enzyme activity on plate (cm day −1 ); τ was the time needed to attain half of D max (days) and t was the time (days). Fitting was performed through the software InfoStat 2016p using a least squares approach with nonlinear regression (Di Rienzo et al. 2016). Fungal growth τ was standardised as Δτ = τPCBs -τC, where τPCBs and τC were the values from medium supplemented with PCBs and control culture, respectively. A positive value of Δτ proved fungal growth inhibition in response to PCBs. Δτ,kLac, kMnP, kLiP, τLac, τMnP and τLiP values were used as input values to run a principal component analysis (PCA) and joined the different isolates based on their tolerance. The correlation/probability matrix showed the interactive effect among the variables. A PCA was performed using Infostat 2016 software.
A validated cluster dendrogram was used to regroup fungi based on their tolerance for PCBs and the parameters obtained from the fit of the enzyme activity in solid media to the same model of fungal growth, and by means of nonlinear regression analysis Δτ were correlated as a function of the significant statistic variables.
In order to test growth response at different PCB concentrations, a group of selected fungi with different tolerances for PCBs was evaluated with increasing concentrations of PCBs and their Δτ were determined.

Results and discussion
Tolerance is one of the properties evaluated when searching microorganisms for bioremediation processes (Bayman and Radkar 1997;Matsubara et al. 2006;Lee et al. 2014;Sharma et al. 2017). A PCB tolerance test should be a primary stage in order to identify desirable PCB-degrading fungal species for biotechnological procedures. The results obtained in this work endorse predictive mycology as a tool and preliminary step to improve the screening and selection of strains capable of tolerating PCBs.
The study of fungal growth kinetics on a solid culture medium supplemented with PCBs was conducted with Dantigny's modified logistic model, fitting the equation to model the increase in diameter of the fungal colonies on plates. The modified equation of Bevilacqua et al. (2016) was used to fit the increase in diameter and it is intended to estimate the rate of fungal growth and the parameter τ, which could be used to compare the inhibition grade among isolates. Figure 1 shows an example of the diameter vs time plot of Pleurotus sajor-caju LBM 105 growing in both control and PCB-supplemented medium, and the fitting corresponding to Equation 1.
Taking into account that the profiles of fungal enzyme activity on Petri dishes resembled those obtained for fungal growth, we adjusted the secretion halo of each enzyme to the same model and obtained k and τ as parameters of a quantitative and standardised characterisation. Figure 2 shows an example of the fitting in Pleurotus sajor-caju LBM 105 measurements of enzymes activity, and Table 2 presents Δτ, kLac, τLac, kMnP, τMnP, kLiP and τLiP values obtained from the analysis of all isolates.
P. chrysosporium was used as a reference fungus because it has been extensively applied as a model organism to study the physiological requirements and enzymes required for lignin and toxic biodegradation (Paszczynski and Crawford 1995).  Figure 3 shows a biplot of the PCA with the fungal dispersion and the relative correlation among variables. The first two principal components (Supplementary Data, Table S1) with all variables explained 55.4% of the total variance. kLac, τLac and Δτ were the main variables that described the variability among the fungi under study, and kLiP, kMnP and τMnP were the second main variables that explicated the variability among fungi (Supplementary Data, Table S2). Graphically, it can be seen as a high orthogonality between the MnP activity and the standardised inhibition Δτ (Supplementary Data, Table S3).  The correlation/probabilities matrix verified that there was no probability of statistically significant correlation between MnP activity and Δτ (Supplementary Data, Table S4). Hence, it was possible to group the strains according to their tolerance. Forty-two percent of the fungi showed Δτ < 1, meaning that the difference between their growth rate in control and PCB-supplemented plates was minor or equal to 24 h. P. chrysosporium BKM-F 1767 belonged to this group. The other 42% of the fungi displayed a Δτ between 1 and 2, indicating that these strains grew 24 to 48 h slower in the presence of PCBs than in the control plates without   Table 1. pollutant. The rest of the strains (15%) revealed Δτ between 2 and 5, representing the group of fungi more severely affected by the presence of PCBs. Ganoderma sp. LBM 001 (Rn 4) showed the highest Δτ; therefore, because of its high sensibility it can be postulated as an indicator of toxicity in matrix contaminated with PCBs. Removing the MnP activity in a second PCA (Figure 4), as analysis variable, total variability among observations was increased to 70%.
The Lac activity and Δτ proved to be the main variables that explained variability among observations (Supplementary Data, and Rn 20). Graphically, an elevated orthogonality between the variable Δτ and the LiP activity was observed, implying independence between these variables.
Although this is a novel assay for tolerance for PCBs in solid media, the obtained results are in agreement with previous studies which demonstrated a noteworthy role of Lac in the presence of this pollutant. In addition, MnP and LiP had a secondary effect on fungal tolerance. Several authors have established a relationship between Lac from WRF and their capability for degradation of PCBs (Dietrich et al. 1995;Novotný et al. 1997;Moeder et al. 2005;Plačková et al. 2012). Gayosso-Canales et al. (2012) proved that among the ligninolytic enzymes versatile peroxidase, MnP and Lac, only the Lac activity was implicated in the removal of highly chlorinated PCBs. Placková et al. (2012) demonstrated an increase in Lac secretion in the presence of PCBs without establishing a clear connection with the degradation of this pollutant. They suggested that in addition to the role of Lacs in biodegradation, these enzymes might display other functions in fungal cultures. Lac secretion has been connected with pathogenicity, protecting fungal pathogens from toxic  Table 1. phytoalexins and tannins in the host environment (Bollag et al. 1988;Pezet et al. 1991). The same detoxification function might explain the higher tolerance for PCBs of the fungi tested, that secreted Lac in a shorter period in comparison with those with a slower secretion rate. Among previous studies, a correlation between the ability of various WRF to produce ligninolytic enzymes and their capability of removing PCBs in liquid media was reported by Novotny et al. (1997;; nevertheless, Krčmář et al. (1999) and Chroma et al. (2002) could not found a correlation between MnP and LiP activities and fungal ability to degrade PCBs.
The results of 25 fungal isolates and the control strain resulting when assaying tolerance for PCBs were joined in a single cluster analysis in order to produce a dendrogram derived from InfoStat software combined with pvcluster package from R software to facilitate their comparison ( Figure 5). Pvclust calculates probability values (p-values) for each cluster using bootstrap resampling techniques. Two types of p-values are available: approximately unbiased p-value and bootstrap probability value (Suzuki and Shimodaira 2006).
Through a dendrogram cluster, the probability of clusters with an average method and Euclidean distance was verified, evidenced by four main clusters of 98%, 92%, 99% and 99% probability of similarity respectively among strains. The divisive coefficient was high (0.891), indicated a good clustering structure. The clustering tree provided four clusters based on the ability of the fungal isolates to tolerate PCBs and their Lac and LiP activities. Cluster 1 consisted of four isolates without specific and common characteristics. Cluster 2 consisted of seven isolates with a high capacity to tolerate PCBs and high Lac activity. Cluster 3 consisted of six isolates with an intermediate tolerance for PCBs and intermediate Lac activity, with a particularly presence of fungi of Pycnoporus genus. Finally, cluster 4 consisted of nine isolates characterised by low tolerance for PCBs with low Lac and LiP activities, which can be useful strains in order to test toxicity.
In order to identify the best-fit model between Δτ and Lac activity, the Pearson correlation coefficients between Δτ and τLac elevated at different exponents were compared (Supplementary Data, Table S8). Considering the correlation coefficient 0.88 of Δτ 2 and τLac 3 , we proceeded to adjust Δτ 2 with the simple variable τLac by means of the following nonlinear function: The resulting fitting model equation was (Supplementary Data, Table S9): The R 2 -statistic = 86.73% explains the variability in Δτ 2 and the adjusted R 2 -statistic was 85.53%. The Durbin-Watson statistic (DW = 2.44041) proved the absence of significant correlation based on the order in which the data were presented ( Figure 6). Although it is well known that R 2 -statistic, in itself, cannot be used to decide whether a new model is an improvement (Baranyi and Roberts 1995), the fitted mathematical function presented in this paper corroborates the previously obtained PCA and dendrogram results.
A leave-one-out cross-validation method (Supplementary Data, Data S10) was used to obtain coefficients (Supplementary Data, The obtained function expressed that low-slung τLac values (high Lac secretion) predict low Δτ values (degree of inhibition) towards PCB presence, and therefore high PCB fungal tolerance.
It was developed a general function to predict the tolerance for PCBs of different basidiomycetous fungi based on their Lac activity, and might be applied to select fungal strains suitable for bioremediation processes.
In order to evaluate the effect of the increase in PCB concentration on fungal growth, the rate of mycelial growth in agar plates was measured in six selected fungi based on their high, intermediate and low capacity to tolerate PCBs ( Figure 5). Δτ slightly augmented with increasing pollutant concentrations, and two levels of tolerance appeared, depicted in Figure 7 (Δτ vs PCB concentration). An Analysis of Variance (ANOVA) determined a significant statistical difference with p < .05 between two fungal groups. I. lacteus BAFC 1171, I. lacteus BAFC 1168D and P. sajor-caju LBM 105 exhibited the lowest Δτ for various PCB concentrations. The other group significantly different was composed by P. chrysosporium BKM-F 1767, P. sanguineus LBM 023 and T. villosa LBM 010 which showed minor tolerance for PCBs. Figure 7 shows that the tolerance ability of each fungus at low levels to PCBs remains similar at high levels of PCBs.
The observed decrease in fungal growth while increasing PCB concentration is in agreement with previous studies (Murado et al. 1976). This negative effect of PCBs on fungal growth rate allows assuming that tolerance for PCBs is a reliable indicator of fungal ability to co-metabolise these pollutants (Yin et al. 2011).

Conclusions
This is the first screening evaluating growth and ligninolytic enzyme production by WRF in a solid medium amended with PCBs, as a strategy to select tolerant fungi to these pollutants. It opens the possibility of identifying more strains that might be suitable as PCB bioremediation agents, linking the use of predictive microbial growth as a part of a low time-consuming and health-security model, complemented with conventional microbiological selection techniques.
The principal component analysis (PCA) revealed a strong correlation between the degree of inhibition and the activity of the enzyme Lac (p < .05), and to a lesser extent with non-significant statistical difference (p > .05) with the activity of the enzyme LiP. The MnP activity did not show influence on the degree of inhibition (p > .05).
Effective bioremediation requires the selection of species with desirable characteristics such as high tolerance and enzymatic activity; thereby, from the obtained results, I. lacteus BAFC 1168D, I. lacteus BAFC 1171, L. elegans BAFC 1227, P. sajor-caju LBM 105 could be ranked as promising PCB degraders. Among these four strains, P. sanguineus LBM 014, P. sanguineus LBM 021 and P. sanguineus LBM 023 had also good competitive activity, based on the observed tolerance model associated with Lac activity. I. lacteus BAFC 1171, I. lacteus BAFC 1168D and P. sajor-caju LBM 105 exhibited the lowest Δτ for different PCB concentrations, indicating a relative proportional tolerance for an increase in PCB concentration.
The next step will be to test the best strains for their potential to remediate PCB-contaminated substrates, also considering their possible use as a fungal consortium.
Based on the mean PCA and the cluster analysis, three isolates from cluster 2 were selected for a second-stage screening on a rice-straw-powder-amended medium. I. lacteus BAFC 1171, I. lacteus BAFC 1168D and P. sajor-caju LBM 105, which displayed the highest tolerance for PCBs, might be valuable as a useful resource for the biodegradation of PCBs and a variety of other biotechnological applications.
Finally, an applied equation was developed that can be used to predict the tolerance for PCBs of WRF based on their Lac activity in solid media supplemented with PCBs, and might be applied to select fungal strains suitable for their bioremediation.