Comparative transcriptomics provides insights into molecular mechanisms of zinc tolerance in the ectomycorrhizal fungus Suillus luteus

Abstract Zinc (Zn) is a major soil contaminant and high Zn levels can disrupt growth, survival, and reproduction of fungi. Some fungal species evolved Zn tolerance through cell processes mitigating Zn toxicity, although the genes and detailed mechanisms underlying mycorrhizal fungal Zn tolerance remain unexplored. To fill this gap in knowledge, we investigated the gene expression of Zn tolerance in the ectomycorrhizal fungus Suillus luteus. We found that Zn tolerance in this species is mainly a constitutive trait that can also be environmentally dependent. Zinc tolerance in S. luteus is associated with differences in the expression of genes involved in metal exclusion and immobilization, as well as recognition and mitigation of metal-induced oxidative stress. Differentially expressed genes were predicted to be involved in transmembrane transport, metal chelation, oxidoreductase activity, and signal transduction. Some of these genes were previously reported as candidates for S. luteus Zn tolerance, while others are reported here for the first time. Our results contribute to understanding the mechanisms of fungal metal tolerance and pave the way for further research on the role of fungal metal tolerance in mycorrhizal associations.


Introduction
Zinc (Zn) is an essential micronutrient (Robinson et al. 2021) critical to biological functions such as enzyme activation and the synthesis of proteins, carbohydrates, and lipids (Sharma et al. 2013).However, at high concentrations Zn is toxic and can lead to death (Robinson et al. 2021).Alarmingly, increased anthropogenic activity has caused Zn contamination to become one of the most prevalent and detrimental forms of pollution in soil environments (Guo et al. 2014, Op de Beeck et al. 2015;Smith et al. 2016;Ramrakhiani et al. 2017).High Zn environments create stressful conditions such as altered soil chemical properties and lower nutrient bioavailability that disrupt soil organism's cellular and organismal processes responsible for metabolism, growth, and survival (Påhlsson 1989;Wang et al. 2006;Priyadarshini et al. 2021).Despite these harsh conditions, some species tolerate high Zn levels and even thrive in Zn-rich environments (Miransari 2011).The exact mechanisms of Zn tolerance in mycorrhizal fungi are however still unclear.
There has been increasing interest in understanding how fungi interact with Zn (Branco et al. 2022).Fungi evolved homeostatic mechanisms to maintain the Zn levels required in cell processes such as transcription, protein folding, and hyphal growth (Feldmann 2012).However, excess Zn negatively impacts fungal cell function by interrupting cell membrane synthesis (Galván Márquez et al. 2018), increasing oxidative stress that can trigger programmed cell death (Zadrąg-Te ̨cza et al. 2018), and decreasing growth by disrupting hyphal extension and altering hyphal morphology (Lanfranco et al. 2010;Priyadarshini et al. 2021).Furthermore, high Zn levels can reduce spore germination (Pawlowska and Charvat 2004), lower fungal survival (Op De Beeck et al. 2015;Priyadarshini et al. 2021), and decrease fungal community species richness (Faggioli et al. 2019).

Fungal Genetics and Genomics
as well as signaling cascades that trigger stress responses and the production of antioxidants that mitigate toxicity (Bellion et al. 2006).Zinc import is controlled by metal transmembrane transporters and the reduction of bioavailable forms of extracellular Zn.For example, in the ericoid mycorrhizal fungus Oidiodendron maius, Zn tolerance relies on the plasma membrane Zn transport protein OmFET, and isolates collected from Zn-polluted soils showed a much lower ability to solubilize inorganic Zn (Martino et al. 2003;Khouja et al. 2013).Additionally, the ectomycorrhizal Hebeloma cylindrosporum sequesters cytosolic Zn into intracellular vesicles through a transport protein localized to the endoplasmic reticulum (Blaudez and Chalot 2011).Fungi can also reduce the amount of excess cytosolic Zn through metallothioneins, metal complexing proteins that immobilize micronutrient metals (Howe et al. 1997).Finally, antioxidant homeostatic mechanisms produce enzymes that break down reactive oxygen species (ROS), neutralizing oxidative stress toxicity resulting from high intracellular metal content (Bolann and Ulvik 1997;Colpaert 2008;Teng et al. 2018).
Studies in Suillus luteus have substantially contributed to unveiling the mechanisms of fungal Zn tolerance.This widespread temperate ectomycorrhizal fungus associates with pine trees, maintaining a mutualistic relationship where it provides mineral nutrients and receives carbohydrates in return (Lofgren et al. 2024).Suillus luteus is an important pioneer species and dominates fungal communities in metal-contaminated sites (Colpaert and van Assche 1987;Ruytinx et al. 2011;Op De Beeck et al. 2015).This species displays Zn tolerance in areas around decommissioned Zn smelters in Belgium, where some isolates can survive at high Zn soil concentrations (Colpaert and van Assche 1987;Colpaert et al. 2004).Notably, Belgian S. luteus Zn tolerant isolates accumulate Zn in their tissue at a much slower rate and can withstand much higher concentrations as compared with sensitive isolates (Colpaert et al. 2005).Genetic studies on these isolates identified 4 transmembrane transporter genes involved in cellular Zn homeostasis and putatively involved in tolerance.SlZnT1 and SlZnT2 are cation diffusion facilitator (CDF) family Zn transporters involved in vacuolar transport, suggesting that intracellular metal sequestration plays an important role in S. luteus Zn tolerance (Ruytinx et al. 2017).SlZRT1 and SlZRT2 are plasma membrane Zrt/IrT-like protein (ZIP) transporters that are downregulated in high Zn environments, indicating they also play a key role in maintaining Zn homeostasis (Coninx et al. 2017(Coninx et al. , 2019)).Furthermore, genome scans of the same Belgian S. luteus isolates revealed the absence of population structure between isolates from Zn contaminated and noncontaminated soils, and that metal tolerance in this species stems from standing genetic variation and is polygenic (Bazzicalupo et al. 2020).Genetically differentiated loci between isolates from contaminated and noncontaminated sites included transmembrane transporters, chelators, and antioxidants, namely, proteins involved in Zn import and vacuolar sequestration, intracellular and extracellular Zn immobilization, and ROS detoxification.Interestingly, the transporters SlZnT1, SlZnT2, SlZRT1, and SlZRT2 did not show high genetic divergence (Fst) across isolates from contaminated and noncontaminated sites (Bazzicalupo et al. 2020).These results suggested that Zn transporter genes may instead be differentially expressed to achieve Zn tolerance.
Here, we quantify gene expression differences in previously studied Belgian S. luteus isolates from contaminated and noncontaminated soils to unveil the mechanisms of fungal Zn tolerance.We hypothesized that isolates originating from contaminated soils would be Zn tolerant, and isolates from noncontaminated soils would be sensitive to high Zn concentrations.Because these isolates belong to the same population and are genetically very similar (Bazzicalupo et al. 2020), we expected to find significantly different transcriptomic profiles only across high-and low-Zn treatments, including in the candidate genes highlighted in previous experiments.We also predicted that tolerant and sensitive isolates would show distinct responses to the Zn treatment, with differentially expressed genes contributing to Zn tolerance pathways and mechanisms in S. luteus.We found that the level of soil contamination was positively associated with isolate Zn tolerance and that transcriptomic differences between tolerant and sensitive isolates were mainly constitutive.We also document S. luteus putative Zn tolerance mechanisms, including metal exclusion and immobilization, as well as recognition and mitigation of metal-induced oxidative stress.Our analyses provide a transcriptome-wide exploration of potential ectomycorrhizal fungal Zn tolerance mechanisms and highlight many genes likely important to Zn tolerance in S. luteus.

Sampling sites and fungal culture information
To examine the effect of Zn on S. luteus gene expression, we investigated 10 previously studied isolates from Belgium, 5 collected from 1 metal-contaminated site (C1-C5) and 5 collected from 1 noncontaminated site (NC1-NC5, Colpaert et al. 2004, Supplementary Table 1).All 10 isolates show little overall genetic differentiation and belong to a single population (Bazzicalupo et al 2020).Both contaminated and noncontaminated sites are dominated by Scots pine (Pinus sylvestris), and the contaminated site displays high levels of heavy metals resulting from now decommissioned zinc smelters (Sonke et al. 2002;Op De Beeck et al. 2015).The soil Zn content was previously determined at a maximum of 1,750μg/g dry weight for the contaminated site and 21 μg/g dry weight for the noncontaminated site (Colpaert et al. 2000).We obtained S. luteus pure cultures from fruit bodies and maintained them in solid modified Fries medium (Colpaert et al. 2000;Lofgren et al. 2024).For more information on collection methods and site descriptions, consult Bazzicalupo et al. (2020) and Colpaert et al. (2000).

Testing S. luteus Zn tolerance
To quantify S. luteus Zn tolerance and determine whether it is associated with soil contamination, we calculated EC 50 values (the amount of Zn that inhibits isolate growth by 50%) for each of the 10 isolates.We grew the isolates at 23°C in darkness on cellophane-covered solid modified Fries medium supplemented with different concentrations of ZnSO 4 •7H 2 O (0, 100, 200, 400, 800, and 1,000 ppm).We performed 4 replicates per isolate for each condition.After 14 days of growth, we collected the mycelium from the plates and stored it at −80°C.To determine EC 50 values per isolate, we followed the protocol in Ritz et al. (2015).Briefly, we obtained isolate dry weight after lyophilization and used it to construct dose-response curves through nonlinear regression with a 4-parameter log-logistic model in R version 3.5.1 (R Core Team 2018; RStudio Team 2019).We used EC 50 values to place isolates into tolerance categories (tolerant, sensitive) irrespective of collection site.We considered isolates Zn tolerant if EC 50 > 130 ppm and sensitive if EC 50 < 130 ppm as this was the amount of Zn added in our transcriptomic experiment (see details in the following).

RNA extraction and sequencing
To measure Zn-induced differential gene expression across the Zn-tolerant and Zn-sensitive S. luteus isolates, we grew the isolates on both control and zinc-supplemented (130 ppm Zn) modified solid Fries medium covered with cellophane, in triplicate, at 23°C, and in darkness (n = 10 isolates × 2 treatments × 3 replicates = 60, Supplementary Table 2).We selected 130 ppm Zn because this concentration stresses but does not kill sensitive isolates and allows for comparing gene expression across Zn-tolerant and -sensitive S. luteus.Once the isolate on the control plate reached a diameter of 3 cm (10-14 days, depending on the isolate), the mycelium for both control and corresponding Zn supplemented plates was collected and stored at −80°C awaiting RNA extraction.
We extracted total RNA using the RNeasy Plant Mini Kit (Qiagen, France) on mycelia ground in liquid nitrogen using mortar and pestle.We measured the RNA concentration and determined the A260/230 and A260/280 ratios using a Nanodrop One (Thermo Fisher).Samples that did not meet purity criteria (A260/230 and A260/280 > 1.8) were discarded.RNA integrity was confirmed using a Bioanalyzer with the RNA 6000 Nano Kit (Agilent).Two sample replicates (one NC1 and one NC3, both grown in 130 ppm Zn) did not meet our quality criteria and were not incorporated into subsequent analyses.RNA was sequenced by the Joint Genome Institute (JGI) on an Illumina NovaSeq 6000 S4 (2 × 151 bp) and run through JGI's quality and quantity quality control tests (https://jgi.doe.gov/user-programs/pmooverview/project-materials-submission-overview/).Three sample replicates did not meet sequence quality standards (2 from isolate NC2 and 1 from isolate NC5 all in 130 ppm Zn) and were discarded, so that our final number of samples for gene expression analysis was reduced to 55. Supplementary Table 2 lists the Short Read Archive codes for all samples.

Bioinformatics pipeline
The RNA sequence analyses involved 5 steps: read quality control, alignment to reference genome and raw gene counts, differential gene expression analyses, gene ontology (GO) analyses, and coexpression network construction:

Read quality control and preprocessing
We used BBDuk (Bushnell 2014) to filter and trim raw reads.We used kmer matching (kmer = 25) to remove Illumina adapters, sequencing artifacts, RNA spike-in reads, PhiX reads, and reads containing any 'N's.We used the phred trimming method to trim the read ends, set to Q6.Finally, we removed reads shorter than 50 bases (1/3 the length of the original read).Supplementary Table 3 lists all sample raw and filtered read counts.

Alignment to reference genome and raw gene counts
We used HISAT2 (Kim et al. 2015) to align reads to the S. luteus reference genome, calculate gene counts, and extract gene annotations.The reference genome (S. luteus isolate UH-Slu-Lm8-n1, https://mycocosm.jgi.doe.gov/Suilu4/Suilu4.info.html)originated from the same Belgian population as the isolates used in our study (JGI Project ID: 1006871; Kohler et al. 2015).To calculate raw gene counts, we used the program featureCounts (Liao et al. 2014) and the gff3 annotation file from the reference genome.Specifically, we only used primary hits on the reverse strand for the final gene counts (-s 2 -P -primary options).

Differential gene expression analysis
To characterize the differences in transcriptomic response to Zn between tolerant and sensitive isolates, we fit a generalized (negative binomial) linear model using the R package DESeq2 (Love et al. 2014) with the 55 retained RNAseq sample counts per gene as input and including parameters for Zn treatment (discrete factor, 2 levels, high, or control), isolate (discrete factor, 10 levels, NC1-5, and C1-5), and the interaction between these 2 effects.We then built 5 contrasts, (1) Zn treatment contrast: high Zn-control Zn, (2) tolerance group contrast: tolerant isolates-sensitive isolates, (3) tolerant Zn contrast: tolerant high Zn-tolerant control Zn, (4) sensitive Zn contrast: sensitive high Zn-sensitive control Zn, and (5) interaction contrast: tolerant Zn contrast-sensitive Zn contrast, outputting DESeq results tables for each comparison.To account for errors in modeling, we calculated a Benjamini-Hochberg (false discovery rate, FDR) adjusted P-value for each transcript using Wald tests and used a threshold of FDR-adjusted P-value < 0.05 as adequate evidence that a transcript was differentially expressed.We then used the results tables and associated transcript FDRs to identify 3 groups of interest: (1) transcripts that were differentially expressed in response to the Zn treatment irrespective of the sensitivity of the isolate (Zn treatment FDR < 0.05; interaction FDR > 0.05), (2) transcripts that were differentially expressed between tolerant and sensitive isolates irrespective of Zn treatment (tolerance group FDR < 0.05, interaction FDR > 0.05), and (3) transcripts whose differential expression in response to Zn treatment differed between tolerant and sensitive isolates (interaction FDR < 0.05).Importantly, these categories allow us to interpret the direction and magnitude of differential expression in a biologically relevant manner.Our approach to DESeq analysis, first retaining isolate distinction in our model as a parameter, and then incorporating tolerance groups into the contrasts when outputting the results tables, is an attempt to account for isolate-to-isolate variation, since DESeq2 has no framework to model the random effect of isolates within tolerance groups.We have also applied additional analyses in the following that allow visualization and inference about variation among isolates.
To visualize clustering of samples based on overall transcript expression in the model, we used the plotPCA function of DESeq2 to run a principal component analysis (PCA).To test for differences between treatment groups in the model, we ran a PERMANOVA with a pairwise distance matrix, calculated using the Jaccard index, as input and including parameters for Zn treatment (discrete factor, 2 levels, high or control), tolerance group (discrete factor, 2 levels, tolerant and sensitive), and the interaction between these 2 effects (permutations = 9,999, vegan package: Oksanen et al. 2024).To account for the large number of permutations in the PERMANOVA, we then used the Benjamini-Hochberg method to adjust all P-values.To visualize differential expression of Zn response for each individual isolate, we made volcano plots by first classifying all DESeq input data into subsets by isolate and then conducting DESeq analyses separately for each isolate.We then used the R package EnhancedVolcano (Blighe et al. 2024) to create volcano plots for all 10 isolates (Supplementary Fig. 4 in Supplementary File 1).To visualize overall differential expression across all replicates, we made correlation matrices of the normalized counts for all RNAseq samples in the control (n = 30) and high Zn conditions (n = 25) using the base R function cor with the Pearson method.We then used the heatmap.2function in the gplots package (Warnes et al. 2024) to plot the correlation matrices (Supplementary Fig. 7 in Supplementary File 1).To obtain predicted functions of differentially

Gene ontology enrichment analysis
To identify enriched (overrepresented) gene annotation categories in the 3 groups of transcripts, we implemented an enrichment analysis using ClueGO (v2.5.9) (Bindea et al. 2009), a Cytoscape (Shannon et al. 2003) application.This analysis used the GO annotation database to extract gene annotation categories that were enriched in each of the transcript groups as compared to the whole genome.We investigated GO annotation terms from all 3 overarching categories (biological process, molecular function, and cellular component) and used modified settings for both the GO Tree Interval (min = 3; max = 15) and GO term/pathway selection (min # genes = 1; % genes = 4) to include the most number of GO terms as possible.We used 2-sided hypergeometric tests with a Benjamini-Hochberg (FDR) correction to weight evidence for enrichment, reporting only GO terms with an FDR-adjusted P-value ≤ 0.05.Additionally, to identify highly differentially expressed genes, we restricted our analysis to transcripts only with an absolute log2-fold change value >1.

Co-expression network construction
To determine modules of genes that were highly correlated with either Zn treatment or with EC 50 value, we conducted weighted gene co-expression network analysis using the WGCNA R package v 1.72-5 (Langfelder and Horvath 2008).Prior to performing this analysis, we excluded 2 samples (NC3 with Zn and without Zn) as they were significant outliers in clustering analysis (Supplementary Fig. 1 in Supplementary File 1).We performed signed network construction and module detection using the blockwiseModules function, a module cut height of 0.25, and a soft thresholding power of 16, selected after examining a range of powers for R 2 > 0.8 and minimal mean connectivity (Supplementary Fig. 2 in Supplementary File 1).This generated a list of module eigengenes, and we used the plotEigengeneNetworks function to generate a heatmap of adjacencies among eigengenes and a dendrogram describing their relationships (Supplementary Fig. 3 in Supplementary File 1).Module eigengenes are the first principal components of each module and are representative of the gene expression profile of the module.
We used the labeledHeatmap function to investigate the correlation between modules and samples traits (Zn treatment or EC 50 value).For each trait, we selected modules that were significantly correlated with the trait and plotted the module membership statistic against the gene significance statistic to determine which modules displayed a strong positive correlation (i.e.genes with importance within the module that was correlated with significance to the trait).Modules with a significant positive correlation were selected for GO analysis.Genes within the modules, along with their trait significance P-value were subjected to GO analysis using the topGO R package v 3.18 (Alexa and Rahnenfuhrer 2024 Alexa & Rahnenfuhrer 2023).We examined GO terms and visualized results using ggplot2 (Wickham et al. 2024).
We also performed intermodular analysis to examine gene significance to either trait (Zn treatment or EC 50 value) across all genes in the dataset.Each list of genes was ranked based on the -log10 of the gene trait significance P-value, and was subjected to preranked gene set enrichment analyses (Mootha, et al. 2003;Subramanian et al. 2005) using a gene matrix file consisting of GO terms (biological process, molecular function, and cellular component) and their corresponding genes.

Suillus luteus Zn tolerance is associated with soil contamination
We found S. luteus Zn tolerance is strongly associated with soil contamination (ANOVA: F 1,8 = 10.38,p = 1.22e −02 ).Isolates collected from the contaminated site were more Zn tolerant than isolates collected from the noncontaminated site, with the most tolerant isolate (C5) having an EC 50 of 937 ppm and the most sensitive isolate (NC1), 61 ppm (Fig. 1; Supplementary Table 1).Interestingly, one isolate from the noncontaminated site was more Zn tolerant than one of the isolates from the contaminated site (Fig. 1; Supplementary Table 1).The 4 sensitive isolates had EC 50 values lower than the Zn concentration used in the experiment (130 ppm), but were able to survive the treatment allowing us to measure gene expression.Conversely, for most Zn tolerant isolates, EC 50 concentration was considerably greater than the amount of added Zn and may not have been as affected by the Zn addition.Based on these EC 50 values, we divided the 10 isolates into Zn-tolerant and -sensitive groups (Fig. 1) as detailed in the Methods.

Isolate Zn tolerance level, not Zn treatment, drives the number of differentially expressed genes
The differential gene expression analysis showed that the vast majority of genes (6,600) were differentially expressed between isolate Zn tolerance groups and that Zn treatment induced a very small number of differentially expressed genes (51) (Fig. 3).Specifically, we found ∼130 times as many genes were significantly differentially expressed between tolerant and sensitive isolates as compared to Zn treatment.In addition, we only detected 123 genes for which tolerant and sensitive isolates responded to Zn differently (interaction effect; Fig. 3).

Enrichment of transmembrane transport, oxidoreductase, protein kinase, and fungal hydrophobin activities
We found 26 enriched GO terms in the response to tolerance group and no enriched GO terms in the response to Zn treatment or interaction effect.There was a large overlap in the functions of enriched GO terms in genes differentially expressed between tolerant and sensitive isolates, so we merged them into 4 categories: transmembrane transport, oxidoreductase activity, fungal hydrophobins, and protein kinase activity.Transmembrane transport was the most enriched annotation category (highest % associated genes) and corresponded to three pleiotropic drug resistance (PDR) transporters (Fig. 4).

Significant differential expression of candidate genes between Zn-tolerant and -sensitive isolates irrespective of Zn treatment
Most previously identified S. luteus Zn tolerance candidate genes, namely, Zn transporters and genes genetically diverged between isolates from contaminated and noncontaminated sites (Coninx  et al. 2017, 2019;Ruytinx et al. 2017;Bazzicalupo et al. 2020), showed significant differential expression only across tolerance groups (Table 1).Specifically, 7 Zn transport genes, 3 metal transport genes, 3 chelators, and 4 oxidative stress relief genes were differentially expressed between tolerant and sensitive isolates irrespective of Zn treatment (tolerance group FDR < 0.05, interaction FDR > 0.05; Table 1).None of the candidate genes showed significantly different gene expression across Zn treatment, irrespective of tolerance group (Zn treatment FDR < 0.05, interaction FDR > 0.05; Table 1), or interaction between tolerant isolates and Zn treatment (Table 1), and sensitive isolates and Zn treatment (Table 1) (interaction FDR > 0.05).Of the 17 differentially expressed genes across tolerance group, 12 were upregulated in tolerant isolates as compared with sensitive isolates and 5 genes were upregulated in sensitive isolates as compared with tolerant isolates.The most upregulated candidate gene in the tolerant isolates was a cytochrome P450, a detoxification agent, while in the sensitive isolates, it was a fungal hydrophobin, a metal chelator.

Many other genes were differentially expressed across comparisons
We found many other differentially expressed genes across each of the three model comparisons that have not been highlighted in previous studies.Of the genes with the most positive and negative log2-fold change values in the three model comparisons (n = 58), close to 70% had no annotation (n = 39; Supplementary Table 4).The remaining annotated genes had predicted functions relating to processes such as nucleic acid metabolism (9 genes), signal transduction (2 genes), and enzyme activity (4 genes).Genes that were differentially expressed in response to Zn treatment had predicted functions related to signal transduction, enzyme activity, and the regulation of nucleic acids (Supplementary Table 4).The WGCNA detected modules of genes significantly associated with either Zn treatment or EC50 (Supplementary Fig. 5 in Supplementary File 1).Two gene modules (midnightblue and turquoise) were significantly negatively correlated with Zn treatment (Fig. 5), indicating that Zn addition decreases expression of the module eigengenes.The turquoise module was significantly enriched for molecular function, cellular compartment, and biological process GO terms and had 18 genes predicted to be transcription factors (Supplementary Tables 5 and 6).The midnightblue module was not significantly enriched for any GO terms and had one gene predicted to be a transcription factor (Supplementary Table 6).We found three modules (red, lightcyan, and salmon) significantly positively correlated with EC 50 value (with expression of genes within the modules increasing with EC 50 ) and eight modules (lightgreen, magenta, grey60, pink, black, royalblue, green, and lightyellow) significantly negatively correlated with this trait (with expression of genes within the modules decreasing with EC 50 ) (Fig. 5).Out of these modules, only the lightgreen, magenta, black, and green showed significant positive correlation between module membership (importance of the gene within the module) and gene significance for EC 50 value (importance of gene to the trait).The black module was significantly enriched for GO:0016614 oxidoreductase activity that acts on CH-OH group of donors.The green module was only significantly enriched for GO:0016740 transferase activity and GO:0003824 catalytic activity.The magenta and lightgreen modules were not significantly enriched for any GO terms.We found genes predicted as transcription factors within each module, specifically 3 in the black module, 7 in the green module, 3 in the magenta module, and 1 in the lightgreen module (Supplementary Table 6).

Discussion
We investigated Zn tolerance and differential gene expression in S. luteus isolates collected from metal-contaminated and noncontaminated sites.We found that in this species Zn tolerance is associated with soil contamination, is both a constitutive and environmentally dependent trait, and results from a combination of responses involving metal exclusion and immobilization, as well as recognition and mitigation of metal-induced oxidative stress (summarized in Fig. 6).
The S. luteus Zn-tolerant isolates were mainly associated with high soil Zn content, corroborating previous research (Colpaert et al 2004).However, one isolate from the contaminated site showed Zn tolerance more similar to isolates from the noncontaminated site.This discrepancy likely results from soil heterogeneity and the existence of low-Zn pockets within the contaminated site, allowing the persistence of Zn-sensitive isolates.In fact, the soil Zn dry weight concentrations at the contaminated site can be as low as 1 ppm (Op De Beeck et al. 2015), making it likely for Zn-sensitive fungi to be able to survive in localized areas of low-Zn concentration.Another possibility is that soil Zn bioavailability varies across the contaminated site and that Zn can locally occur in a form that is not toxic to fungi, allowing the sensitive isolate to persist.
The S. luteus Zn tolerant and sensitive isolates had markedly different overall transcriptomic profiles, with tolerant isolates displaying much higher overall transcriptomic variation and few individual genes being differentially expressed in response to Zn treatment.This means that Zn tolerant isolates consistently maintain a Zn-tolerant gene expression profile.Constitutive expression independent of external Zn concentration was previously documented for two S. luteus Zn transmembrane transporters (Ruytinx et al. 2017).Here we show that it applies to a much larger number of genes, including functions such as cation transport and oxidoreductase activity.The large number of observed gene expression differences across tolerant isolates also suggests the existence of distinct paths to achieve Zn tolerance in the species, through the expression of different sets of metal tolerance-related genes.
While constitutive tolerance mainly explains differences between tolerant and sensitive isolates, tolerant isolates also display Zn-induced gene expression, although on a much smaller scale.For example, WGCNA showed Zn transporters were involved in Zn response regardless of tolerance level and that the previously identified transporters SlZIP-B (2893674) and SlZRT1 (2764984) were significantly differentially regulated when isolates were exposed to Zn.Interestingly, Zn transporters implicated in the environmental response were different than those involved in constitutive tolerance.
Zn tolerance in S. luteus is associated with a clear enrichment of metal homeostasis-related genes, including transmembrane transport, oxidoreductase, fungal hydrophobin, and protein kinase activity.Genes involved in signaling and metal transport, immobilization, and detoxification have been previously implicated in metal response (Wolfger et al. 2001;Lin et al. 2005;Bazzicalupo et al. 2020).In S. luteus, enriched transmembrane transport activity corresponded to PDR transporters, ATP-binding cassette transporters that have been linked to metal stress in both plants and fungi (Wolfger et al. 2001;Nuruzzaman et al. 2014).In yeast, PDR transporters are involved in vacuolar sequestration of chelated metal conjugates (Buechel and Pinkett 2020) and could have a similar function in S. (Fig. 6).Oxidoreductase activity corresponded to genes in a variety of pathways that catalyze electron transfer between molecules (Bolann and Ulvik 1997).Oxidative stress is a known byproduct of metal exposure and is characterized by the accumulation of ROS (Azevedo et al. 2007;Liang and Zhou. 2007), which can impair cellular functioning and result in cell death.The reduction of ROS by oxidoreductase activity is an important homeostatic response (Zadrąg-Tęcza et al. 2018).Fungal hydrophobin activity corresponded to cell wall fungal hydrophobin genes that function by binding excess metals to render them inert (Ferrol et al. 2016;Fig. 6) and have been implicated in the metal response of mycorrhizal fungi (Sammer et al. 2016).Several genes involved in protein kinase activity were predicted mitogen-activated protein kinases, evolutionary conserved signal transduction modules that convert environmental cues into cell responses (Mishra et al. 2006;Cristina et al. 2010).These proteins have been shown to be activated in response to metal stress (Lin et al. 2005;Mondal 2022), to function in the signal cascade mechanism of metal homeostasis (Mondal 2022), and likely play an important role in S. luteus Zn tolerance (Fig. 6).
Differential expression of candidate genes putatively involved in Zn tolerance across S. luteus further confirms marked differences across Zn-sensitive and -tolerant isolates, corroborates constitutive Zn tolerance in this species, and hints at potential S. luteus tolerance mechanisms.Zn tolerance in this species is mainly achieved through constitutive differential expression of candidate metal ion transporters, metal chelating agents, and antioxidants (Bazzicalupo et al. 2020; Table 1; Fig. 6).CDF transporters (including SlZnT1, SlZnT2, and SlCDF-D) have been implicated in the export and sequestration of excess Zn, and their upregulation in high Zn environments has been identified as a potential mechanism of Zn tolerance in S. luteus (Ruytinx et al. 2017).In addition, chelating agents such as fungal hydrophobins act to bind and neutralize excess metals (Ferrol et al. 2016) and upregulation of these genes in high Zn environments is likely another mechanism of Zn tolerance in S. luteus.Notably, the Zn-tolerant isolates showed no differences in candidate gene expression when exposed to Zn addition, suggesting the amount of Zn added in our experiment was not enough to induce large responses in the tolerant isolates.Future studies should consider including exposure to multiple levels of Zn to ensure that tolerant isolates are being amply stressed.
Even though our results make significant contributions for understanding the mechanisms of fungal Zn tolerance, this study has some limitations that preclude unveiling additional processes potentially involved in tolerating high Zn.Specifically, ZIP transporters have been shown to have a time sensitive response to excess Zn, with downregulation lasting even only a few hours (Coninx et al. 2019).Since we extracted RNA after more than 1 week under Zn stress, the expression profile we captured here reflects an adjusted equilibrium and not immediate responses.Also, fungi most likely deal with excess levels of Zn in both acute and prolonged interactions so future work should examine the differences in the immediate responses of tolerant and sensitive S. luteus isolates to excess Zn.Another limitation pertains to the available S. luteus reference genome annotation that is far from complete, with well over 50% of genes missing a predicted function.This means that we likely missed important gene categories involved in Zn tolerance.Lastly, epigenetic regulation may Zinc tolerance in Suillus luteus | 9 contribute to Zn tolerance in S. luteus and could provide explanations for transcriptional regulation or lack of thereof.Future studies should rely on an improved reference genome annotation, use higher metal concentrations to assess stress response in tolerant isolates, and include epigenetics, for example, by using whole genome bisulfide sequencing before and after applying Zn treatments.

Conclusion
In conclusion, our study unveils mechanisms of S. luteus Zn tolerance and contributes to the understanding of how fungi can withstand metal toxicity.We found S. luteus displays constitutive differences and environmentally driven Zn responses.We also explored expression patterns of genes previously implicated in S. luteus metal tolerance and uncovered new signaling genes that potentially contribute to Zn tolerance in the species.Further research can address metal responses of mycorrhizal fungi in conjunction with their obligate plant partners, allowing further understanding of the mechanisms of metal tolerance in a more ecologically relevant context.

Fig. 1 .
Fig. 1.Suillus luteus Zn tolerance and soil contamination.EC 50 values for the studied S. luteus isolates, grouped by collection site (NC, noncontaminated site; C, contaminated site).The shape of the symbols denotes collection site (triangles, noncontaminated site, circles, contaminated site), and the color of the symbols denotes tolerance group.Dashed line indicates 130 ppm Zn, the amount of Zn added to the Zn-treated plates.

Fig. 2 .
Fig. 2. Tolerant isolates show higher transcriptomic variation compared to sensitive isolates.Principal component analysis plot of complete transcriptomes of Zn-sensitive and -tolerant S. luteus isolates.Symbols represent the average across isolate replicates, and error bars represent the standard error between replicates.Colors denote tolerance grouping and the symbol shape denotes Zn treatment (circle, no zinc, triangle, zinc amended).We include isolate identifiers next to each symbol.Ellipses denote 95% confidence intervals for each of the tolerance groups (red, sensitive, blue, tolerant).

Fig. 3 .
Fig. 3. Zn tolerance affects gene expression more than Zn treatment.Barplot displaying the number of significantly differentially expressed (DE) genes in each of the three model comparisons.Tolerance = tolerance groups (tolerant vs sensitive), Zn treatment = control vs Zn-amended, Tolerance × Zn treatment = interaction effect.The numbers on top of bars represent the number of significantly DE genes.DE genes that had a significant interaction effect were removed from the DE gene counts for the main effects (Tolerance and Zn treatment).

Fig. 4 .
Fig.4.Significantly differentially expressed genes between tolerant and sensitive isolates are enriched in transmembrane transport, oxidoreductase, fungal hydrophobin, and protein kinase activity.Barplot showing enriched GO terms in significantly differentially expressed genes between tolerant and sensitive isolates.Bar length represents the percentage of all genes with the GO term annotation that appear in this gene group.The numbers at the end of the bars represent the number of unique genes with that GO term annotation.Colors represent general functional categories (light blue, transmembrane transport; green, oxidoreductase activity; yellow, fungal hydrophobin; orange, protein kinase activity; and dark blue, other).
et al. 2003) Flavin Adenin dinucleotide (FAD)expressions across model comparisons of previously identified genes putatively involved in S. luteus metal tolerance.Log2-fold change (lfc) values display differential gene expression in response to tolerance groups (A), Zn treatment (B), Zn treatment in tolerant isolates only (C), and Zn treatment in sensitive isolates only (D).Colored log2-fold change values have an adjusted P-value of < 0.05 and denote the direction of differential expression (*light gray = lfc < 0, **dark gray = lfc > 0).Zinc tolerance in Suillus luteus | 7 Co-expression network analyses show cellular signaling, oxidoreductase, metal ion binding, and transporter activity are important in S. luteus Zn response and tolerance

Fig. 5 .
Fig. 5. Module-trait associations determined by Pearson correlation.The numbers in each cell indicate the correlation between the trait (columns) and the modules (rows), with the color scale representing positive (red) to negative (blue) correlation.Correlation P-values are in parenthesis.The Zn treatment trait is a binary representation of whether Zn treatment was applied (1) or not (0), while the EC50 trait is the numerical value for tolerance (higher value = more tolerant).

Fig. 6 .
Fig. 6.Potential mechanisms of Zn tolerance in S. luteus.Diagram of putative mechanisms of Zn tolerance in S. luteus.(A-C) Zn transmembrane transport including Zn importers (such as ZIP transporters), exporters (such as CDF transporters), and sequestration transporters (such as CDF and PDR genes), which maintain cytosolic Zn homeostasis.(D) Zinc chelation (for example, through fungal hydrophobins), which immobilizes excess Zn. (E) Antioxidants that mitigate metal-induced oxidative stress.(F) MAP kinases that transmit stress signals to the nucleus.(Adapted from Branco et al. 2022).

Table 1 .
Differential expression of previously identified S. luteus Zn transporters and genetically diverged genes only across tolerant and sensitive isolates.