- Split View
-
Views
-
Cite
Cite
Fahri A Hasby, Florian Barbi, Stefano Manzoni, Björn D Lindahl, Transcriptomic markers of fungal growth, respiration and carbon-use efficiency, FEMS Microbiology Letters, Volume 368, Issue 15, August 2021, fnab100, https://doi.org/10.1093/femsle/fnab100
- Share Icon Share
ABSTRACT
Fungal metabolic carbon acquisition and its subsequent partitioning between biomass production and respiration, i.e. the carbon-use efficiency (CUE), are central parameters in biogeochemical modeling. However, current available techniques for estimating these parameters are all associated with practical and theoretical shortcomings, making assessments unreliable. Gene expression analyses hold the prospect of phenotype prediction by indirect means, providing new opportunities to obtain information about metabolic priorities. We cultured four different fungal isolates (Chalara longipes, Laccaria bicolor, Serpula lacrymans and Trichoderma harzianum) in liquid media with contrasting nitrogen availability and measured growth rates and respiration to calculate CUE. By relating gene expression markers to measured carbon fluxes, we identified genes coding for 1,3-β-glucan synthase and 2-oxoglutarate dehydrogenase as suitable markers for growth and respiration, respectively, capturing both intraspecific variation as well as within-strain variation dependent on growth medium. A transcript index based on these markers correlated significantly with differences in CUE between the fungal isolates. Our study paves the way for the use of these markers to assess differences in growth, respiration and CUE in natural fungal communities, using metatranscriptomic or the RT-qPCR approach.
INTRODUCTION
Upon uptake, metabolic carbon is partitioned between biomass growth and respiration, with carbon-use efficiency (CUE) defined as the share of acquired carbon incorporated into growing biomass (Geyer et al. 2016; Manzoni et al. 2018). High CUE of microbial (in particular fungal) decomposers means lower proportional carbon losses during decomposition and higher accumulation of microbial biomass, which could enhance microbial-derived carbon storage in soils (Cotrufo et al. 2013; Wang et al. 2021). On the other hand, a high CUE can promote microbial proliferation, leading to more efficient plant litter exploitation with associated organic matter losses and lower plant-derived carbon storage (Allison, Wallenstein and Bradford 2010). Information about growth rates, respiration and CUE of microbial decomposers is, thus, essential in trait-based microbial modelling (Allison 2012; Manzoni et al. 2017; Zhang et al. 2018). However, direct empirical assessment of growth is challenging, since it depends on repeated measurements and intrusive disturbance. As a consequence, estimates of CUE are sensitive to the chosen methodology, and indirect short-term assessments may prove useful. For example, isotope-labeled substrates or calorespirometry have been used to assess growth and CUE in laboratory incubations (Geyer et al. 2019). CUE data on fungal decomposers is particularly limited, motivating our focus on fungi in this contribution.
‘Omics’ approaches, made possible by recent advances in sequencing technology, are becoming more common and are transforming research on fungal ecology (Nilsson et al. 2019). Variations in growth, respiration and CUE have a genetic basis, but are more likely related to differences in gene expression rather than gene presence and diversity (Barbi et al. 2020). Therefore, gene expression data may provide information about fungal metabolic traits, as mRNA sequences can be linked to specific metabolic conversions (Treseder and Lennon 2015). Genetic markers might even be used to assess differences in fungal traits in complex, natural fungal communities, using metatranscriptomic or RT-qPCR approaches (Kuske et al. 2015).
Genes coding for enzymes that facilitate polymerization of cell wall components such as glycosyl transferases (GT families in the CAZyme classification, Lombard et al. 2014) are particularly interesting targets for markers. Chitin synthase (GT2) and 1,3-β-glucan synthase (GT48) expression can be expected to be directly linked to fungal growth, since these enzymes are involved in the synthesis of chitin and β-glucan (Kelly et al. 1996; Sreenivasaprasad, Burton and Wood 2000), which are main components of fungal cell walls (Bowman and Free 2006). The tricarboxylic acid (TCA) cycle, which is central to aerobic metabolism and a hub of several catabolic and anabolic pathways, contains many different potential gene markers for respiration. Both NAD-dependent and NADP-dependent isocitrate dehydrogenase catalyse oxidative decarboxylation of isocitrate to 2-oxoglutarate and release CO2 (Haselbeck and McAlister-Henn 1993; Gálvez and Gadal 1995). Consecutively, the 2-oxoglutarate dehydrogenase complex catalyses oxidative decarboxylation of 2-oxoglutarate to succinyl-coenzyme A and CO2 (Repetto and Tzagoloff 1989). If useful transcriptional markers for growth and respiration may be identified, the expression ratio of these markers should reflect CUE. However, in contrast to genes coding for extracellular enzymes, for which the relationship between gene expression and observed process may seem relatively straightforward (Lindahl and Kuske 2013), relationships between gene expression and growth and respiration need to be verified under controlled conditions. In order to qualify as suitable, general markers, relationships between gene expression and phenotype have to be consistent across a variety of fungal species and in various environments.
Here, we devised a laboratory study where differences in growth rates, respiration and CUE among four fungal isolates and between contrasting growing conditions (high and low nitrogen availability) were related to gene expression data, with the aim to identify transcriptional markers that may be used to assess inter and intraspecific phenotypic variation. We specifically targeted a set of markers linked to growth and respiration, and evaluated these markers in relation to other potential marker genes. We hypothesized that:
Transcription of genes coding for glycosyl transferases that are active in polymerization of fungal cell wall components would correlate with relative mycelial growth rate.
Transcription of genes coding for enzymes in the TCA cycle would correlate with fungal respiration.
The transcription ratio of selected glycosyl transferase genes over TCA cycle genes would correlate with CUE.
That the above correlations would be valid both among fungal isolates as well as within isolates under different environmental conditions.
METHODS
Culture experiment
A total of four fungal isolates with sequenced genomes were selected to represent various fungal classes and life strategies: Trichoderma harzianum Rifai—an opportunistic mycoparasite (Weindling 1932) in the class Sordariomycetes, Chalara longipes (Preuss) Cooke—an ubiquitous colonizer of needle litter (Koukol 2011) in the class Leotiomycetes, Lacaria bicolor (Maire) P. D. Orton—an ectomycorrhizal fungus (Martin et al. 2008) in the order Agaricales (Agaricomycetes) and Serpula lacrymans (Wulfen)—a brown-rot wood decomposer (Kauserud et al. 2007) in the order Boletales (Agaricomycetes).
Fungal stock cultures were kept on Modified Melin-Norkrans (MMN) agar (Marx 1969) at room temperature in darkness. To produce agar free inoculum, colonized agar plugs were floated on the surface of liquid MMN medium, and extending mycelium was separated from the agar, further cultivated in liquid MMN for 1 week and macerated in the growth medium using an Ultra-Turrax (IKA, Germany). The concentration of mycelium in the inoculum was established by weighing the mycelial content of 2 mL of inoculum after drying at 40°C for 24 h (N = 10 for each isolate), and ranged 2–10 mg/mL.
To establish phenotypic variation in growth, respiration rate and CUE, the isolates were cultivated at two different conditions of nitrogen availability: liquid MMN medium with only glucose as the carbon source was modified to C:N ratios of 22 (N rich) or 221 (N poor) by altering the content of (NH4)2HPO4. The experimental system consisted of 50 mL of medium autoclaved in 250 mL Schott Duran® Bottles (DWK Life Sciences, Germany). Two connectors in the lid were attached to sterile syringe filters (0.2 µm; VWR, Radnor, Pennsylvania), to allow gas exchange. Systems were inoculated by adding 0.5 mL of mycelial macerate (1-5 mg of mycelium). Each treatment included nine replicates and one negative control (without inoculation). In total, 72 systems (four isolates × two growth media × nine replicates) were set up, and due to logistic constraints, the four isolates were assessed sequentially.
Glucose concentration in the medium was measured at regular intervals using a GM-100 glucose monitoring system (BioReactor Sciences, Lawrenceville, Georgia). The systems were harvested when roughly 30% of the glucose in the medium had been consumed (Figure S1 and Table S1, Supporting Information). Immediately before harvest, respiration was measured using an EGM-4 portable infra-red gas analyser (PP Systems, Amesbury, Massachusetts) with a closed sampling loop. Prior to measurement, the systems were flushed with filtered air for 1 h to allow dissolved CO2 to equilibrate with the atmosphere. CO2 accumulation was measured during 2.5 min for each system.
Mycelium was harvested by filtration through Whatman filter paper; ø 55m, pore size 12 mm (ThermoFisher Scientific, Waltham, Massachusetts) and immediately shock-frozen with liquid nitrogen, freeze dried, weighted using an ES120A analytical balance (Precisa Gravimetrics, Switzerland) and stored at –80°C. Total RNA was extracted from the harvested mycelium using the RNA mini kit (Qiagen, Germany) and cleaned from remaining DNA by the DNAse I kit (Sigma-Aldrich®, St. Louis, Missouri). Poly-A selection and mRNA library preparation was conducted using the TruSeq library preparation kit (Illumina, San Diego, California). Libraries were sequenced on the Illumina NovaSeq 6000 SP platform, yielding 50 bp paired-end sequences. Poly-A selection, library preparation and sequencing were performed by the SNP&SEQ Technology Platform of SciLifeLab, Uppsala, Sweden.
Calculations
Bioinformatic analyses
Raw paired-end reads were subjected to quality control using FastQC (Andrews et al. 2010). Sequencing adapter trimming and removal of low quality bases were performed in the program ‘Trimmomatic’ (Bolger, Lohse and Usadel 2014) with default settings. Reference genomes and gene annotations of T. harzianum (Druzhinina et al. 2018), C. longipes (Barbi et al. 2020), L. bicolor (Martin et al. 2008) and S. lacrymans (Eastwood et al. 2011) were retrieved from the JGI—Mycocosm database. Filtered mRNA sequences were mapped against respective genomes using ‘bowtie2’ (Langmead and Salzberg 2012) with default settings. Data was sorted, indexed, and converted to transcript count tables using ‘SAMtools’ (Li et al. 2009; Li 2011). Transcript data were normalized for gene lengths and sequencing effort according to the RPKM method (reads per kilo base per million mapped reads; Mortazavi et al. 2008). To enable analysis of expression of enzyme-encoding genes across fungal isolates, data was aggregated according to Enzyme Commission (EC) numbers, which denote a numerical classification of enzymes (Kanehisa 2017). The complete sequence was uploaded to NCBI-Sequence Read Archive (SRA; https://www.ncbi.nlm.nih.gov/sra) under the accession number PRJNA721630.
Statistical analyses
All statistical analyses were performed in R version 3.6.2 (R Core Team 2019). Effects of isolates, medium and their interaction on relative growth rate, qCO2 and CUE, were evaluated by two-way ANOVA, with a post hoc Tukey test to evaluate differences between isolates. Due to large differences in variance between isolates, effects of growth medium were evaluated post hoc by t-tests for each isolate separately.
In order to evaluate gene markers for growth, we specifically targeted expression of 1,3-β-glucan synthase (EC 2.4.1.34) and chitin synthase (EC 2.4.1.16) encoding genes (Table S2, Supporting Information). We established a linear model with relative growth rate as response variable and expression of 1,3-β-glucan synthase or chitin synthase encoding genes (paralogs aggregated) as explaining variables, using the ‘lm’ function in the ‘stats’ package of R. Both response and explaining variables were log transformed. The same approach was applied to evaluate gene markers for respiration. We specifically targeted the expression of NAD-dependent isocitrate dehydrogenase (EC 1.1.1.41), NADP-dependent isocitrate dehydrogenase (EC 1.1.1.42) and 2-oxoglutarate dehydrogenase (EC 1.2.4.2) encoding genes (Table S2, Supporting Information), and established linear models with qCO2 as response variable and gene expression levels of either of the three enzymes classes as explaining variables.
The gene index was evaluated as a predictor of measured CUE by linear regression. The a priori selected markers for growth and respiration were evaluated against other potential gene markers. All EC categories with an aggregated gene expression level of at least 10 RPKM in all samples were included, resulting in a list of 431 enzyme classes. Pearson's correlation coefficients between log-transformed expression levels of individual enzyme classes and log-transformed relative growth rate or qCO2 were calculated with the ‘cor’ function of the ‘stats’ R package.
To verify that fungi were nitrogen limited in the nitrogen poor medium, we investigated the expression of glutamine synthetase (EC 6.3.1.2), which is a central enzyme in ammonia assimilation (Montanini et al. 2003), as a marker. Two-way ANOVA was conducted to assess effects of isolates, medium and their interaction on expression of selected gene markers (log transformed) as well as on the CUEgene index with post hoc Tukey tests to evaluate differences between isolates. Post hoc t-tests of effects of growth medium were conducted for each isolates separately, due to large differences in variance between isolates. The complete dataset used in the analyses was uploaded to Dryad reository (doi:10.5061/dryad.pvmcvdnkm).
RESULTS AND DISCUSSION
Sequencing and alignment
Transcriptome sequencing was successful for 65 of the 72 samples. One replicate of C. longipes was removed as an outlier due to very low expression of 1,3-β-glucan synthase and one replicate each of L. bicolor and S. lacrymans were removed due to negative relative growth rates. Thus, all statistical analyses were based on 62 observations. In total, 408 million Illumina sequences (96%) of C. longipes, 477 million sequences (95%) of L. bicolor, 487 million sequences (93%) of S. lacrymans and 378 million sequences (97%) of T. harzianum passed quality control. Of these, 316 million sequences (77%) mapped to the C. longipes genome, 374 million sequences (78%) mapped to the L. bicolor genome, 344 million sequences (71%) mapped to the S. lacrymans genome and 212 million sequences (56%) mapped to the T. harzianum genome.
Growth markers
For C. longipes and L. bicolor the rates of glucose consumption were increasing, as indicated by downward concavity in the time trajectories, supporting our assumption of exponential growth. This was also the case for T. harzianum in rich medium, whereas in poor medium growth was retarded towards the end of the culture period, presumably due to N limitation. Glucose consumption by S. lacrymans was more irregular and with a linear trend (Figure S1, Supporting Information). Biomass at harvest ranged from 1.3 to 54 mg, with T. harzianum having the highest relative growth rate, followed by C. longipes, L. bicolor and S. lacrymans (Fig. 1A). There was a significant interaction effect of isolate and medium on growth rate (Table 1) with T. harzianum and C. longipes growing significantly faster in the rich medium.
. | Log growth rate (mmol C/day) . | Log metabolic quotient (qCO2, day–1) . | Log carbon-use efficiency . | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Sum Sq . | Df . | F value . | Pr(>F) . | R2 . | SumSq . | Df . | Fvalue . | Pr(>F) . | R2 . | Sum Sq . | Df . | Fvalue . | Pr(>F) . | R2 . |
Isolates | 27.38 | 3 | 143.43 | 0*** | 0.90 | 17.39 | 3 | 45.64 | 0*** | 0.73 | 0.90 | 3 | 46.44 | 0*** | 0.73 |
Medium | 0.66 | 1 | 10.44 | 0.0021** | 0.21 | 1 | 1.61 | 0.21 | 9 × 10–5 | 1 | 0.014 | 0.91 | |||
Isolates : Medium | 1.17 | 3 | 6.12 | 0.0012** | 0.96 | 3 | 2.52 | 0.067 | 0.026 | 3 | 1.36 | 0.26 | |||
Residuals | 3.44 | 54 | 6.86 | 54 | 0.35 | 54 | |||||||||
Log GT48 | Log KGD | CUEgene | |||||||||||||
Sum Sq | Df | Fvalue | Pr(>F) | R2 | Sum Sq | Df | Fvalue | Pr(>F) | R2 | Sum Sq | Df | Fvalue | Pr(>F) | R2 | |
Isolates | 8.55 | 3 | 171.10 | 0*** | 0.92 | 9.83 | 3 | 89.92 | 0*** | 0.85 | 7.37 | 3 | 104.29 | 0*** | 0.86 |
Medium | 0.14 | 1 | 8.38 | 0.0055** | 0.30 | 1 | 8.22 | 0.0059** | 0.0027 | 1 | 0.11 | 0.74 | |||
Isolates : Medium | 1.11 | 3 | 22.21 | 0*** | 0.55 | 3 | 5.06 | 0.0037** | 0.29 | 3 | 4.13 | 0.011* | |||
Residuals | 0.90 | 54 | 1.97 | 54 | 1.27 | 54 | |||||||||
Log GS | |||||||||||||||
Sum Sq | Df | Fvalue | Pr(>F) | R2 | |||||||||||
Isolates | 8.32 | 3 | 36.19 | 0*** | 0.73 | ||||||||||
Medium | 4.47 | 1 | 58.36 | 0*** | |||||||||||
Isolates : Medium | 0.64 | 3 | 2.77 | 0.051+ | |||||||||||
Residuals | 4.14 | 54 |
. | Log growth rate (mmol C/day) . | Log metabolic quotient (qCO2, day–1) . | Log carbon-use efficiency . | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Sum Sq . | Df . | F value . | Pr(>F) . | R2 . | SumSq . | Df . | Fvalue . | Pr(>F) . | R2 . | Sum Sq . | Df . | Fvalue . | Pr(>F) . | R2 . |
Isolates | 27.38 | 3 | 143.43 | 0*** | 0.90 | 17.39 | 3 | 45.64 | 0*** | 0.73 | 0.90 | 3 | 46.44 | 0*** | 0.73 |
Medium | 0.66 | 1 | 10.44 | 0.0021** | 0.21 | 1 | 1.61 | 0.21 | 9 × 10–5 | 1 | 0.014 | 0.91 | |||
Isolates : Medium | 1.17 | 3 | 6.12 | 0.0012** | 0.96 | 3 | 2.52 | 0.067 | 0.026 | 3 | 1.36 | 0.26 | |||
Residuals | 3.44 | 54 | 6.86 | 54 | 0.35 | 54 | |||||||||
Log GT48 | Log KGD | CUEgene | |||||||||||||
Sum Sq | Df | Fvalue | Pr(>F) | R2 | Sum Sq | Df | Fvalue | Pr(>F) | R2 | Sum Sq | Df | Fvalue | Pr(>F) | R2 | |
Isolates | 8.55 | 3 | 171.10 | 0*** | 0.92 | 9.83 | 3 | 89.92 | 0*** | 0.85 | 7.37 | 3 | 104.29 | 0*** | 0.86 |
Medium | 0.14 | 1 | 8.38 | 0.0055** | 0.30 | 1 | 8.22 | 0.0059** | 0.0027 | 1 | 0.11 | 0.74 | |||
Isolates : Medium | 1.11 | 3 | 22.21 | 0*** | 0.55 | 3 | 5.06 | 0.0037** | 0.29 | 3 | 4.13 | 0.011* | |||
Residuals | 0.90 | 54 | 1.97 | 54 | 1.27 | 54 | |||||||||
Log GS | |||||||||||||||
Sum Sq | Df | Fvalue | Pr(>F) | R2 | |||||||||||
Isolates | 8.32 | 3 | 36.19 | 0*** | 0.73 | ||||||||||
Medium | 4.47 | 1 | 58.36 | 0*** | |||||||||||
Isolates : Medium | 0.64 | 3 | 2.77 | 0.051+ | |||||||||||
Residuals | 4.14 | 54 |
Notes: + P < 0.10; *P < 0.05; **P < 0.01 and ***P < 0.001.
. | Log growth rate (mmol C/day) . | Log metabolic quotient (qCO2, day–1) . | Log carbon-use efficiency . | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Sum Sq . | Df . | F value . | Pr(>F) . | R2 . | SumSq . | Df . | Fvalue . | Pr(>F) . | R2 . | Sum Sq . | Df . | Fvalue . | Pr(>F) . | R2 . |
Isolates | 27.38 | 3 | 143.43 | 0*** | 0.90 | 17.39 | 3 | 45.64 | 0*** | 0.73 | 0.90 | 3 | 46.44 | 0*** | 0.73 |
Medium | 0.66 | 1 | 10.44 | 0.0021** | 0.21 | 1 | 1.61 | 0.21 | 9 × 10–5 | 1 | 0.014 | 0.91 | |||
Isolates : Medium | 1.17 | 3 | 6.12 | 0.0012** | 0.96 | 3 | 2.52 | 0.067 | 0.026 | 3 | 1.36 | 0.26 | |||
Residuals | 3.44 | 54 | 6.86 | 54 | 0.35 | 54 | |||||||||
Log GT48 | Log KGD | CUEgene | |||||||||||||
Sum Sq | Df | Fvalue | Pr(>F) | R2 | Sum Sq | Df | Fvalue | Pr(>F) | R2 | Sum Sq | Df | Fvalue | Pr(>F) | R2 | |
Isolates | 8.55 | 3 | 171.10 | 0*** | 0.92 | 9.83 | 3 | 89.92 | 0*** | 0.85 | 7.37 | 3 | 104.29 | 0*** | 0.86 |
Medium | 0.14 | 1 | 8.38 | 0.0055** | 0.30 | 1 | 8.22 | 0.0059** | 0.0027 | 1 | 0.11 | 0.74 | |||
Isolates : Medium | 1.11 | 3 | 22.21 | 0*** | 0.55 | 3 | 5.06 | 0.0037** | 0.29 | 3 | 4.13 | 0.011* | |||
Residuals | 0.90 | 54 | 1.97 | 54 | 1.27 | 54 | |||||||||
Log GS | |||||||||||||||
Sum Sq | Df | Fvalue | Pr(>F) | R2 | |||||||||||
Isolates | 8.32 | 3 | 36.19 | 0*** | 0.73 | ||||||||||
Medium | 4.47 | 1 | 58.36 | 0*** | |||||||||||
Isolates : Medium | 0.64 | 3 | 2.77 | 0.051+ | |||||||||||
Residuals | 4.14 | 54 |
. | Log growth rate (mmol C/day) . | Log metabolic quotient (qCO2, day–1) . | Log carbon-use efficiency . | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Sum Sq . | Df . | F value . | Pr(>F) . | R2 . | SumSq . | Df . | Fvalue . | Pr(>F) . | R2 . | Sum Sq . | Df . | Fvalue . | Pr(>F) . | R2 . |
Isolates | 27.38 | 3 | 143.43 | 0*** | 0.90 | 17.39 | 3 | 45.64 | 0*** | 0.73 | 0.90 | 3 | 46.44 | 0*** | 0.73 |
Medium | 0.66 | 1 | 10.44 | 0.0021** | 0.21 | 1 | 1.61 | 0.21 | 9 × 10–5 | 1 | 0.014 | 0.91 | |||
Isolates : Medium | 1.17 | 3 | 6.12 | 0.0012** | 0.96 | 3 | 2.52 | 0.067 | 0.026 | 3 | 1.36 | 0.26 | |||
Residuals | 3.44 | 54 | 6.86 | 54 | 0.35 | 54 | |||||||||
Log GT48 | Log KGD | CUEgene | |||||||||||||
Sum Sq | Df | Fvalue | Pr(>F) | R2 | Sum Sq | Df | Fvalue | Pr(>F) | R2 | Sum Sq | Df | Fvalue | Pr(>F) | R2 | |
Isolates | 8.55 | 3 | 171.10 | 0*** | 0.92 | 9.83 | 3 | 89.92 | 0*** | 0.85 | 7.37 | 3 | 104.29 | 0*** | 0.86 |
Medium | 0.14 | 1 | 8.38 | 0.0055** | 0.30 | 1 | 8.22 | 0.0059** | 0.0027 | 1 | 0.11 | 0.74 | |||
Isolates : Medium | 1.11 | 3 | 22.21 | 0*** | 0.55 | 3 | 5.06 | 0.0037** | 0.29 | 3 | 4.13 | 0.011* | |||
Residuals | 0.90 | 54 | 1.97 | 54 | 1.27 | 54 | |||||||||
Log GS | |||||||||||||||
Sum Sq | Df | Fvalue | Pr(>F) | R2 | |||||||||||
Isolates | 8.32 | 3 | 36.19 | 0*** | 0.73 | ||||||||||
Medium | 4.47 | 1 | 58.36 | 0*** | |||||||||||
Isolates : Medium | 0.64 | 3 | 2.77 | 0.051+ | |||||||||||
Residuals | 4.14 | 54 |
Notes: + P < 0.10; *P < 0.05; **P < 0.01 and ***P < 0.001.
Gene expression of both 1,3-β-glucan synthase and chitin synthase explained a significant fraction of the total variation in relative growth rate. However, 1,3-β-glucan synthase expression was a much better predictor of relative growth across isolates and medium (R2 = 0.64) than chitin synthase expression (R2 = 0.09), and was chosen as the best marker (Table 2 and Fig. 2A ). For comparison, an ANOVA in which relative growth rate was predicted by isolate, medium and their interaction had an R2 of 0.90 (Table 1). 1,3-β-glucan synthase expression was significantly different between isolates (P < 0.001) with higher expression in the faster growing T. harzianum and C. longipes than in the slow-growing L. bicolor and S. lacrymans (Fig. 1D). There was a significant interaction between the effects of isolates and medium on 1,3-β-glucan synthase expression (Table 1) with higher 1,3-β-glucan synthase expression in rich medium for T. harzianum and C. longipes but higher expression in poor medium for S. lacrymans. When comparing correlation with growth across all enzyme-encoding genes, 1,3-β-glucan synthase ranked 14th out of 431 enzyme classes (Fig. 4A). The 13 enzyme classes that exhibited stronger correlations with growth rate than1,3-β-glucan synthase are listed in Table S3 (Supporting Information), and were mainly related to glycolysis, cell membrane synthesis and post-translational modification of proteins.
Dependent variable . | Explaining variable . | Intercept . | β . | R2 . | Residual SE (df = 60) . | F statistic (df = 1; 60) . |
---|---|---|---|---|---|---|
Log relative growth rate (µ, day–1) | Log GT48 | –9.64*** | 1.40*** | 0.64 | 0.45 | 107.46*** |
Log GT2 | –6.37*** | 0.63* | 0.09 | 0.71 | 6.02* | |
Log metabolic quotient (qCO2, day–1) | Log KGD | –6.32*** | 0.98*** | 0.48 | 0.47 | 55.69*** |
Log IDHNAD | 0.20 | –0.15 | 0.02 | 0.64 | 1.12 | |
Log IDHNADP | 0.22*** | 0.12 | 0.01 | 0.65 | 0.50 | |
Carbon-use efficiency | CUEgene | 0.07 | 0.82*** | 0.42 | 0.11 | 43.24*** |
Dependent variable . | Explaining variable . | Intercept . | β . | R2 . | Residual SE (df = 60) . | F statistic (df = 1; 60) . |
---|---|---|---|---|---|---|
Log relative growth rate (µ, day–1) | Log GT48 | –9.64*** | 1.40*** | 0.64 | 0.45 | 107.46*** |
Log GT2 | –6.37*** | 0.63* | 0.09 | 0.71 | 6.02* | |
Log metabolic quotient (qCO2, day–1) | Log KGD | –6.32*** | 0.98*** | 0.48 | 0.47 | 55.69*** |
Log IDHNAD | 0.20 | –0.15 | 0.02 | 0.64 | 1.12 | |
Log IDHNADP | 0.22*** | 0.12 | 0.01 | 0.65 | 0.50 | |
Carbon-use efficiency | CUEgene | 0.07 | 0.82*** | 0.42 | 0.11 | 43.24*** |
Notes: + P < 0.10; *P < 0.05; **P < 0.01 and ***P < 0.001.
Dependent variable . | Explaining variable . | Intercept . | β . | R2 . | Residual SE (df = 60) . | F statistic (df = 1; 60) . |
---|---|---|---|---|---|---|
Log relative growth rate (µ, day–1) | Log GT48 | –9.64*** | 1.40*** | 0.64 | 0.45 | 107.46*** |
Log GT2 | –6.37*** | 0.63* | 0.09 | 0.71 | 6.02* | |
Log metabolic quotient (qCO2, day–1) | Log KGD | –6.32*** | 0.98*** | 0.48 | 0.47 | 55.69*** |
Log IDHNAD | 0.20 | –0.15 | 0.02 | 0.64 | 1.12 | |
Log IDHNADP | 0.22*** | 0.12 | 0.01 | 0.65 | 0.50 | |
Carbon-use efficiency | CUEgene | 0.07 | 0.82*** | 0.42 | 0.11 | 43.24*** |
Dependent variable . | Explaining variable . | Intercept . | β . | R2 . | Residual SE (df = 60) . | F statistic (df = 1; 60) . |
---|---|---|---|---|---|---|
Log relative growth rate (µ, day–1) | Log GT48 | –9.64*** | 1.40*** | 0.64 | 0.45 | 107.46*** |
Log GT2 | –6.37*** | 0.63* | 0.09 | 0.71 | 6.02* | |
Log metabolic quotient (qCO2, day–1) | Log KGD | –6.32*** | 0.98*** | 0.48 | 0.47 | 55.69*** |
Log IDHNAD | 0.20 | –0.15 | 0.02 | 0.64 | 1.12 | |
Log IDHNADP | 0.22*** | 0.12 | 0.01 | 0.65 | 0.50 | |
Carbon-use efficiency | CUEgene | 0.07 | 0.82*** | 0.42 | 0.11 | 43.24*** |
Notes: + P < 0.10; *P < 0.05; **P < 0.01 and ***P < 0.001.
The relative growth rates of C. longipes and T. harzianum (ascomycetes) were higher than those of L. bicolor and S. lacrymans (basidiomycetes), and in spite of the very high C:N ratio of the poor medium (Fig. 1A), growth was supressed by nitrogen limitation only for the two most rapidly growing isolates. Nitrogen limitation was, however, apparent for all isolates in the consistent upregulation of glutamine synthetase (EC: 6.3.1.2) in poor medium (Fig. 3), which is required to maintain ammonium transport across the cell membrane when external ammonium concentrations are low (Kershaw and Stewart 1989). Our results show that 1,3-β-glucan synthase expression reflects the variation of growth rate between media (across environments) and among isolates (across taxa). At the mycelial level, the relative proportion of 1,3-β-glucan synthase mRNA in the transcriptome may reflect the proportion of growing tip cells as well as the rate of growth of these cells. At this scale, morphological differences, such as hyphal branching frequency might be a more important determinant of relative growth rates than metabolic differences at the cellular level.
The significant negative intercept in the linear model of growth predicted by 1,3-β-glucan synthase expression (Table 2) implies a basal expression of 1,3-β-glucan synthase (Fig. 2A; x-axis intercept at x = 6.89), possibly linked to cell walls maintenance in non-growing cells, balanced by continuous turnover of cell wall material. In addition, the rate of cell wall assembly (growth) could be hampered by restricted allocation of sugars for growth (i.e. substrate limitation rather than enzyme limitation). However, it is also possible that growth responds to gene transcription in a logarithmic manner. Although our results represent a wide range of growth rates, linearity along the full range of variation present under natural conditions remains to be confirmed.
In this study, the expression of 1,3-β-glucan synthase reflected short-term relative investment in growth at harvest (i.e. the ‘intention’ of the fungi to grow). For this reason, we expected that expression of this gene would correlate with the relative growth rate at the time of harvest, as approximated by Eq. (3). In future studies, isotope labeled substrate added shortly before harvest might help to elucidate the relation between carbon uptake and short-term realized growth. We also acknowledge that there growth rate might be underestimated due to uncertainties in the amount of added inoculum, since the added macerate might have contained both active cells and dead tissues. A lag phase before growth might have commenced after inoculation, contributing additional uncertainty, particularly for S. lacrymans, which took a long time to start growing after inoculation and exhibited low estimated growth relative to 1,3-β-glucan synthase expression. However, exponential growth models do account for a possible initial slow growth rate (they assume time invariant specific growth rate), thus reducing the negative consequences of a lag phase.
Apart from cell proliferation and elongation, 1,3-β-glucan synthase has been proposed as a marker for stress response, since the 1,3-β-glucan production is involved in cell fortification to protect from environmental stressors (Treseder and Lennon 2015). However, this might also be considered as part of growth, since thicker cell walls have higher mass.
Respiration markers
The qCO2 was significantly different between isolates (P < 0.001), with highest values in T. harzianum, intermediate values in L. bicolor and S. lacrymans and lowest values in C. longipes (Fig. 1B), and a marginally significant isolates x medium interaction effect (P = 0.07), with C. longipes and T. harzianum having significantly higher qCO2 in rich medium (Table 1). This is contrary to previous observations of a negative relationship between qCO2 and nitrogen availability in complex microbial communities in soils (Riggs et al. 2015; Spohn 2015; Spohn et al. 2016). The discrepancy may be ascribed to the artificial conditions of the pure culture system, for which it has been speculated that fast growing isolates may suffer higher metabolic costs of growth and cell maintenance (Lipson 2015).
Among gene candidates in the TCA cycle, only the 2-oxoglutarate dehydrogenase gene expression explained significant (P < 0.001) variation in qCO2 across isolated and medium composition, achieving R2 = 0.48 (Fig. 2B and Table 2). For comparison, a two-way ANOVA in which qCO2 was predicted by isolate, medium and their interaction had R2 = 0.73 (Table 1). 2-oxoglutarate dehydrogenase expression was significantly different between isolates with higher expression in T. harzianum and L. bicolor and lower in S. lacrymans and C. longipes (Fig. 1E). There was a significant isolates x medium interaction effect with higher 2-oxoglutarate dehydrogenase expression for T. harzianum and C. longipes in rich medium but slightly lower expression in poor medium for S. lacrymans (Table 2). Similar to 1,3-β-glucan synthase, we observed a significant negative intercept in the linear model of expression (Table 2), indicating baseline expression (Fig. 2B; x-axis intercept at x = 6.32). Constitutive expression of this genes may not result in CO2 production, if the TCA cycle is constrained by other regulatory bottlenecks. When comparing correlation with qCO2 across all enzyme encoding genes, 2-oxoglutarate dehydrogenase ranked 12th out of 431 enzyme classes (Fig. 4B). The 11 enzyme classes that exhibited stronger correlations with growth rate than 2-oxoglutarate dehydrogenase are listed in Table S4 (Supporting Information) and were mainly related to glycolysis, pentose-phosphate pathway and post-translational modification of proteins.
Gene index of CUE
CUEgene was significantly correlated with CUE, with R2 = 0.42 (Fig. 2C and Table 2). For comparison, an ANOVA in which CUE was predicted by isolate, medium and their interaction had R2 = 0.80 (Table 1). The higher fraction of explained variance in the latter statistical model is explained by its higher number of parameters (six parameters) compared to our simpler linear regression between CUEgene and CUE (two parameters). CUEgene was significantly different between isolates, with highest values in C. longipes, intermediate values in T. harzianum and S. lacrymans and lowest values in L. bicolor. There was a significant isolates x medium interaction effects on CUEgene with a negative effect on S. lacrymans in rich medium (Fig. 1F).
We observed no significant effect of N availability on measured CUE. This finding is surprising given the high C:N ratio (220:1) of the nutrient poor medium—typically CUE decreases at high C:N ratio as microorganisms become nitrogen limited and excess carbon is eliminated or invested in non-growth processes (Manzoni et al. 2017). The lack of CUE response could be due to allocation adjustments in the face of nitrogen shortage that did not involve waste respiration (Camenzind et al. 2021). As carbon and nitrogen were supplied in uncoupled form in liquid media, the isolates could adjust the relative acquisition of different resources, as indicated by upregulation of glutamine synthetase gene expression. In contrast, CUE varied across isolates, and in particular, the two ascomycetes had significantly higher CUE than the two basidiomycetes. A similar pattern was also observed in our gene expression based assessment (CUEgene). The CUEgene of S. lacrymans was intermediate and similar to T. harzianum with higher expression of 1,3-β-glucan synthase than expected from direct measurements of growth.
Although this index is probably not directly transferable to other studies (due to different bases for data normalization), the simple expression ratio of 1,3-β-glucan synthase over 2-oxoglutarate dehydrogenase was also a useful indicator of CUE (Figure S2, Supporting Information).
CONCLUSIONS
Although we identify many potential genetic markers for growth and respiration, we see pedagogic values for the use of our a priori selected markers. 1,3-β-glucan synthase is directly active in polymerization of the main constituent of the fungal cell wall and its interpretation as a growth marker is intuitive. Similarly, 2-oxoglutarate dehydrogenase is directly involved in the generation of CO2 and, thus, an attractive marker for respiration. This study presents a basis for theoretical linkages of transcripts, growth rates, respiration and CUE under idealized conditions, but we see a potential utility of these markers in future assessment of differences in growth, respiration and CUE in natural fungal communities, using metatranscriptomics or RT-qPCR approaches.
ACKNOWLEDGEMENTS
We are grateful to Johanna Lagensjö at SciLifeLab, Uppsala, for very valuable technical advices on sequencing strategies.
FUNDING
This work was supported by the Swedish Research Councils FORMAS (grant agreement number 2015-00468 to BDL) and has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement number 101001608 to SM). The computations in this study were enabled by resources in project SNIC 2019/8-305 and in project SNIC 2019/30-34 provided by the Swedish National Infrastructure for Computing (SNIC) at UPPMAX, partially funded by the Swedish Research Council FORMAS (grant agreement number 2018-05973).
Conflicts of interest
None declared.