Identification of COL-26 as a key transcription factor to promote glt-1 expression in response to external glucose
To search for transcription factors essential for expression of glt-1, deletion mutants of 36 transcription factors of N. crassa, based on their expression in adequate glucose conditions (Additional file 1: Table S1) (51), were chosen and screened through batch culture with glucose as the sole carbon source followed by qRT-PCR assay. Most of the mutants showed no significant change in glt-1 expression compared with the WT strain. However, the ΔNCU07788 (Δcol-26) mutant showed dramatically decreased expression of glt-1 (Fig. 1A). COL-26 is a zinc binuclear cluster [Zn(II)2Cys6] DNA-binding protein that is essential for starch utilization (69). Expression of hgt-1/-2 was significantly upregulated in the Δcol-26 mutant in response to glucose, whereas the expression of hgt-1/-2 and glt-1 was downregulated in response to starvation (Fig. 1B). This suggested that COL-26 is involved in regulating the dual-affinity glucose transport system in N. crassa. The nuclear localization of COL-26 was independent of glucose concentration (Fig. 1C).
To further investigate whether COL-26 directly regulates glt-1 and hgt-1/-2, EMSAs were performed. GST-fused DNA binding domain of COL-26 was expressed in and purified from E. coli (Additional file 2: Fig. S1A). EMSA results showed that the recombinant COL-26 bound to the promoter regions of all three genes in a typical protein concentration-dependent manner (Fig. 2A–C). Retardation occurred upon addition of 16 nM recombinant COL-26. The downregulation of hgt-1/-2 expression in a Δcol-26 mutant in response to starch has been reported (69). In addition, several other transporter genes were reported to be regulated by COL-26, including xylose transporter-1 xyt-1 (NCU05627), a predicted sugar transporter (NCU04537), a predicted ammonium transporter tam-1 (NCU03257), and a uracil permease uc-5 (NCU07334), as well as the starch-active lytic polysaccharide monooxygenase (NCU08746) (69). EMSAs showed that COL-26 also binds to the promoter regions of xyt-1 and NCU08746 but not NCU04537, tam-1, or uc-5 (Additional file 2: Fig. S1B–F), suggesting the involvement of COL-26 in multiple biological processes beyond regulation of sugar transporter expression.
Δ col-26 mutant phenotypically and transcriptionally resembles Δrco-3 mutant
In N. crassa, RCO-3 acts as a non-transporting glucose sensor (67). Expression of glt-1 is dramatically downregulated in a Δrco-3 mutant, whereas the expression of hgt-1/-2 is significantly upregulated in glucose-rich conditions (51), which is similar to the Δcol-26 mutant (Fig. 1A, 1B). In addition, both mutants are defective in glucose uptake and biomass accumulation in the presence of high glucose levels, and are resistant to 2-deoxyglucose (2-DG, which cannot be catalyzed during glycolysis and is a drug often used for glucose repression analysis in filamentous fungi) (4, 67). The Δcol-26 mutant showed severe defects in using other simple sugars including sucrose, fructose, mannose, and maltose, although the growth defect on cellobiose was minor (4, 69). So, we used medium containing cellobiose to assess the effect of stress on the growth of the Δcol-26 and Δrco-3 mutants. Both mutants were sensitive to osmotic stress and H2O2-induced oxidative stress compared with the WT strain, as shown by plate growth assays (Additional file 3: Fig. S2) and the corresponding mycelial diameter of the Δcol-26 and Δrco-3 mutants (Fig. 3A). We assumed that COL-26 and RCO-3 are probably in a common regulatory cascade, in which membrane-located RCO-3 transduces a glucose signal to nuclear-located COL-26 in the presence of glucose.
To test this hypothesis and also obtain a broad view of the mode of expression, we conducted high-throughput sequencing (RNA-Seq) of wild-type, Δcol-26, and Δrco-3 mycelia exposed to a gradient of glucose (0%, 0.05%, 0.5%, 2.0%) for 1 h. Pearson and Spearman correlation analysis demonstrated that the biological replicates were reliable for all tested samples (Additional file 4: Fig. S3A). RNA-Seq data (Additional file 1: Table S2) from the WT, Δcol-26, and Δrco-3 biological replicates were subjected to principal component analysis and data from the same strain grown in the same growth conditions clustered together. Compared with the WT strain, data from the Δcol-26 mutant and the Δrco-3 mutant exposed to glucose (0.05%, 0.5% and 2.0%) clustered together (Fig. 3B). This indicated that both mutants had impaired transcriptomic responses to 0.05%, 0.5%, and 2% glucose and had similar expression profiles, which was in accordance with sample-to-sample clustering (Additional file 4: Fig. S3B). Consistent with these observations, the number of differentially expressed genes (DEGs) in Δrco-3 vs. Δcol-26 was much lower than that in Δrco-3 vs. WT and Δcol-26 vs. WT exposed to glucose. At 2%, 0.5%, and 0.05% glucose, there were only 197, 99, and 120 DEGs, respectively, in Δrco-3 vs. Δcol-26 (Fig. 3C, Additional file 1: Table S3), whereas the numbers were 665, 533, and 1024 for Δrco-3 vs. WT and 745, 566, and 788 for Δcol-26 vs. WT, respectively (Fig. 3E, Additional file 1: Table S3). Gene Ontology (GO) enrichment analysis showed that DEGs in Δrco-3 vs. WT and Δcol-26 vs. WT in each condition were mainly enriched in oxidation–reduction processes (GO: 0055114) and metabolic processes (GO: 0008152) (Additional file 1: Table S4). In addition, the number of DEGs in the Δrco-3 mutant and Δcol-26 mutant comparing 0.05% glucose with 0.5% glucose was dramatically lower than the number in the WT (Fig. 3D, Additional file 1: Table S3). Further investigation showed that in the presence of glucose, rco-3 and col-26 regulate a large proportion of DEGs in common as described above, whereas there were far fewer DEGs in Δrco-3 vs. WT than in Δcol-26 vs. WT in carbon-free conditions, or in Δrco-3 vs. WT in glucose conditions (Fig. 3E, Additional file 1: Table S3). This indicated that COL-26 functions in both glucose and carbon-free conditions, whereas RCO-3 mainly functions in the presence of glucose.
Next, the effect of COL-26 and RCO-3 on the sugar uptake system was investigated. Among the 39 putative sugar transporters present in the genome of N. crassa (70), 26 showed robust expression levels (FPKM > 20) in at least one condition (Additional file 5: Fig. S4A). In the WT strain, the transcriptional responses of these sugar transporters to a glucose gradient were in good accordance with previously published data (51). Some genes displayed a strong or moderate response to external glucose changes, including glt-1, hgt-1/-2, xyt-1, cdt-1/-2, NCU05897, NCU00821, xat-1, lat-1, gat-1, clp-1, and NCU09287. However, the responses of these transporter genes to a glucose gradient (2%, 0.5%, and 0.05%) were impaired in Δrco-3 and Δcol-26 mutants (Additional file 5: Fig. S4A). This is consistent with the observations that both mutants had similar transcriptional profiles and impaired transcriptomic response to glucose fluctuation (Fig. 3B–E). As for glt-1, hgt-1/-2, and xyt-1, which displayed the strongest response to external glucose changes in the WT strain and whose promoters were bound by COL-26 (Fig. 2 and Additional file 2: Fig. S1B), their changed expression was probably due to the absence of COL-26 in the Δcol-26 mutant or an inactivated form of COL-26 in the Δrco-3 mutant. Significantly downregulated expression of glt-1 was observed in both the Δcol-26 and Δrco-3 mutants at all glucose concentrations, even though the expression level of col-26 in the Δrco-3 mutant was almost three times that in the WT strain in the presence of 2% and 0.5% glucose (Additional file 1: Table S2), indicating that both genes are essential for glt-1 expression and that RCO-3 acts upstream of COL-26. Notably, hgt-1 and hgt-2 had very similar expression profiles (Additional file 5: Fig. S4A), indicating the synergetic regulation of the two major components of the high-affinity glucose transport system. (51).
Since glucose transport is the first step of glycolysis and glucose uptake is defective in the Δrco-3 and Δcol-26 mutants, we focused on genes involved in central metabolism. The gene encoding phosphofructokinase (NCU00629) was downregulated whereas the gene encoding fructose-1,6-bisphosphatase (NCU04797) was upregulated in the Δrco-3 and Δcol-26 mutants in the presence of glucose (2%, 0.5%, and 0.05%). Other downregulated genes in both mutants in the presence of glucose included those encoding phosphoglycerate kinase (NCU07914), phosphoglycerate mutase (NCU02252), enolase (NCU10042), pyruvate kinase (NCU06075), and pyruvate decarboxylase (NCU02193) (Additional file 5: Fig. S4B), indicating that glycolysis in the Δrco-3 and Δcol-26 mutants in the presence of glucose is downregulated, probably because of the defect in glucose uptake.
Phosphoproteome of the WT strain grown on glucose vs. starvation conditions
Given that expression of col-26 at the transcriptional level is not affected by a gradient of glucose (0–10% w/v) (51), we wondered if a post-translational modification, such as the phosphorylation level of COL-26, explains the significant differentially expression of glt-1 between carbon-rich (2% glucose) and starvation (no-carbon, NC) conditions. Phosphoproteome profiling of the WT strain grown on glucose (Glu) compared with starvation conditions was performed. The coefficient of variation showed that the phosphopeptide abundance correlated well between the three replicates in each condition (Additional file 6: Fig. S5). We identified 11992 phosphopeptides, mapped to 2508 proteins (Additional file 1: Table S5). Of these phosphopeptides, 661 (representing 360 proteins) increased in abundance, and 709 (representing 410 proteins) decreased in abundance in the NC vs. Glu comparison (Fig. 4A, Additional file 1: Table S6).
There are 43 proteins highly phosphorylated in one or more regions but dephosphorylated in other regions in the NC vs. Glu comparison (Fig. 4B), including protein phosphatase regulator REG1 (NCU09310), nitrate nonutilizer-2 NIT-2 (NCU09068), eukaryotic peptide chain release factor ERF2 (NCU04790), chromatin remodeling factor CRF4-3 (NCU02684), and the S/T protein kinases STK-10 (NCU03200), STK-30 (NCU04335), and STK-31 (NCU04747). A similar phenomenon was also observed by Xiong et al. (71). GO analysis showed that proteins highly phosphorylated on starvation vs. Glu were over-represented in the categories cytoplasm (GO: 0005737) (30), intracellular transport (GO:0046907) (3), and carbohydrate phosphorylation (GO:0046835) (4) (Fig. 4C, Additional file 1: Table S7), and included some glycolytic proteins such as two hexokinases (NCU02542 and NCU00575), two 6-phosphofructo-2-kinases (NCU01178 and NCU01728), glyceraldehyde 3-phosphate dehydrogenase (NCU01528), and alcohol dehydrogenase-1 (NCU01754). Highly phosphorylated proteins not belonging to these GO terms included NCU06482 (pyruvate dehydrogenase E1 component α subunit), NCU01328 (transketolase), GLT-1, and NCU03100 (6-phosphogluconate dehydrogenase) (Additional file 1: Table S6), indicating that the glycolytic pathway and pentose phosphate pathway are regulated by post-translational modifications.
Proteins highly dephosphorylated in the NC vs. Glu comparison were over-represented in various categories mainly associated with the membrane, transport, and ATP metabolism (Fig. 4D, Additional file 1: Table S7), suggesting active metabolism in the presence of glucose. Pathway enrichment analysis of the highly dephosphorylated proteins using the Kyoto Encyclopedia of Genes and Genomes identified only one pathway––the MAPK signaling pathway-yeast (ko04011)—including protoperithecium-1 (PP-1, NCU00340), osmotic sensitive-4 (OS-4, NCU03071), MAPKK kinase NRC-1 (NCU06182), an uncharacterized protein (NCU06252), WSC-1 (NCU06910), osmotic sensitive-2 (OS-2, NCU07024), and mitogen-activated protein kinase-2 (MAK-2, NCU02393). NRC-1 (MAPKKK) and MAK-2 (MAPK) are core components of the conserved N. crassa MAK-2 pathway (72) that mediates cell fusion and activates transcription factor PP-1 required for the activation of genes that play a role during the cell fusion (73). However, another core component, MEK-2 (MAPKK, NCU04612), was highly phosphorylated in the NC vs. Glu comparison (Additional file 1: Table S6). Other highly dephosphorylated proteins in dataset NC vs. Glu included a scaffold protein HAM-5 (NCU01789) of the MAK-2 pathway, HAM-8 (NCU02811), HAM-9 (NCU07389), CSP-6 (NCU08380), RCO-1 (NCU06205), ADA-3 (NCU02896), and PRK1 (NCU00506), which all relate to the NRC-1/MEK-2/MAK-2 signaling pathway and are required for cell-to-cell fusion (74, 75). OS-4 (MAPKKK) and OS-2 (MAPK) are components of the hyperosmotic response (OS) MAP kinase pathway involved in carbon sensing (76). Other highly dephosphorylated peptides in the NC vs. Glu comparison were from CK-1b (NCU04005), which is involved in growth and developmental processes (77); ASD-4 (NCU15829), which functions in ascus and ascospore development (78); CEL-2 (NCU07307), which is involved in fatty acid biosynthesis (79); an actin-binding protein FIM (NCU03992) (80); and COL-26. Three phosphopeptides from COL-26 showed S79 and S83 phosphorylation decreased in abundance in the NC vs. Glu comparison and no other phosphorylated sites were found in COL-26 (Additional file 1: Table S6).
The function of COL-26 itself may be regulated at the protein level rather than the phosphorylation level.
Previous studies have identified four phosphorylation sites (S79, S83, S674, and S676) in COL-26 (71, 72, 81), among which S79 and S83 were also identified in this study and showed decreased abundance in the NC vs. Glu comparison (Additional file 1: Table S6). To dissect potential functions of these phosphorylation sites in the expression of the dual-transporter system, we constructed plasmids harboring col-26-egfp without or with site-directed mutations under the control of the promoter of the glyceraldehyde-3-phosphate dehydrogenase-1 gene (gpd-1, NCU01528). Nuclear localization of WT COL-26 and protein with simultaneous mutations at S79 and S83 (S79A, S83A), S674 and S676 (S674A, S676A), or all four sites (S4A) (Fig. 5A), as well as their recovered biomass relative to Δcol-26 mutant on culture grown with sucrose (Fig. 5B), indicated the successful expression and correct function of these analogs. Expression of glt-1 in Δcol-26::Pgpd-col-26 (S79A, S83A), Δcol-26::Pgpd-col-26 (S674A, S676A), and Δcol-26::Pgpd-col-26 (S4A) in glucose and NC conditions was not different from that in Δcol-26::Pgpd-col-26 (WT) (Fig. 5C), suggesting that these phosphorylation sites of COL-26 are not involved in the regulation of glucose transporter expression in N. crassa.
Notably, glt-1 expression in the Δcol-26 mutant complemented with WT and mutated COL-26 grown in starvation conditions was much higher than that in the WT strain (Fig. 5C), probably because of the high expression level of col-26 driven by the gpd-1 promoter which leads to a high level of COL-26. So, we constructed the complemented strain Δcol-26::Pn-col-26 expressing col-26-egfp under the control of its native promoter. The protein level of COL-26 in the presence of different concentrations of glucose was determined by western blotting using anti-GFP antibody. The protein level of COL-26 in the presence of adequate glucose (0.5% and 2%) was higher than that in starvation conditions (0.05%, or no glucose) (Fig. 6), suggesting that the function of COL-26 itself might be regulated at the protein level rather than by phosphorylation.
AMPK represses glt-1 expression, possibly by inhibiting RCO-3 activity, in starvation conditions
Previous study showed that the OS MAP kinase pathway is involved in carbon sensing (76). Two of its components (OS-2 and OS-4) were highly phosphorylated in the WT strain in the Glu vs. NC comparison (Additional file 1: Table S6). Thus, the roles of this pathway in regulation of the dual-affinity glucose transport system were investigated by characterization of deletion mutants of os-1 and os-2, two essential components of this pathway. Expression of glt-1 and hgt-1 in Δos-1 and Δos-2 mutants showed no difference from that in the WT strain in either glucose-rich or starvation conditions. glt-1 and hgt-1/-2 expression in Δrco-3;Δos-1 and Δrco-3;Δos-2 double mutants was identical to that in the Δrco-3 mutant in both conditions (Additional file 7: Fig. S6). These results indicate that os-1 and os-2 are not involved in regulation of the dual-affinity glucose transport system.
The resistance of Δrco-3, Δcol-26, and Δhgt-1;Δhgt-2 mutants to 2-DG (4, 51, 67) indicates that genes whose deletion leads to resistance to 2-DG might be regulatory elements of the dual-affinity glucose transport system. Considering that phosphorylation events are the most common and important mechanism underpinning regulation of glucose transport in S. cerevisiae (23, 38), multiple serine/threonine protein kinase mutants of N. crassa were screened for 2-DG resistance. Three mutants—ΔNCU00914, ΔNCU01940, and ΔNCU01498—showed high resistance to 2-DG (Additional file 9: Fig. S7A). However, the glt-1 expression in these mutants showed no difference from that in the WT strain in either glucose-rich or NC conditions (Additional file 8: Fig. S7B), indicating that resistance to 2-DG is not necessarily connected with glucose transport and signaling.
In S. cerevisiae, the SNF1 complex plays an important role in regulation of glucose transport (33, 38). The effect of AMPK, which is homologous to SNF1, on expression of glt-1 was investigated in N. crassa, in which prk-10 (NCU04566, the ortholog of snf1) and NCU01471 (here named snf4) encode the α-subunit and γ-subunit of the AMPK complex, respectively. Expression of glt-1 in Δprk-10 and Δsnf4 mutants was the same as that in the WT strain in the presence of glucose, but significantly upregulated in starvation conditions. Expression of glt-1 in Δrco-3;Δprk-10 and Δrco-3;Δsnf4 double mutants was the same as that in the Δrco-3 mutant (Fig. 7A), indicating that AMPK-repressed expression of glt-1 might occur via inhibition of RCO-3 activity in starvation conditions. This conclusion was supported by transcriptomic data, which showed that rco-3 mainly functions in the presence of glucose (Fig. 3E). Notably, the lower number of DEGs in the Δrco-3 vs. WT comparison in starvation conditions compared with glucose condition (Fig. 3E) and the significantly reduced expression level of glt-1 in the Δrco-3 mutant (Fig. 7A) indicated that RCO-3 activity was not totally inhibited in starvation condition. Though deletion of prk-10 or snf4 had no effect on glt-1 expression in glucose-rich conditions, expression of hgt-1 was upregulated when the Δprk-10 and Δsnf4 mutants were grown on glucose (Fig. 7B). Besides, deletion of prk-10 or snf4 in the Δrco-3 background further decreased hgt-1 expression in starvation conditions (Fig. 7B), indicating that other regulatory component(s) are also involved in regulation of hgt-1 expression.
The glucose transport system shows conserved regulation by COL-26-like transcription factors in ascomycete species
Despite some minor differences in use of some kinds of sugars, conserved roles of COL-26/AmyR homologs have been reported in various fungi, including Magnaporthe oryzae (82), Fusarium graminearum and F. verticillioides (83), A. nidulans (84), A. oryzae (68), A. niger (85), T. reesei (86), Penicillium oxalicum (87), Talaromyces pinophilus (88), and Myceliophthora thermophila (89). In P. oxalicum, the N. crassa HGT-1 ortholog PDE_03475 showed a high expression level on cellulose and a decreased expression level in a ΔamyR mutant compared with the WT strain (87). In A. niger, the A. nidulans MstE ortholog An02g03540 showed a high expression level on glucose and maltose, and its expression in a ΔamyR mutant was significantly downregulated (90). To test the hypothesis that the dual-affinity glucose transport system and its regulation by COL-26 homologs are conserved in filamentous fungi, the effect of deletion of M. thermophila AmyR (Mycth_2301920), the closest homolog to N. crassa COL-26, on expression of the putative dual-affinity glucose transport system was investigated. The M. thermophila ΔamyR mutant exhibited significantly reduced growth on glucose, fructose, sucrose, maltose, trehalose, xylose, and soluble starch, but grew well on cellobiose and cellulose (89), which is similar to the N. crassa Δcol-26 mutant. Alignment showed that Mycth_112491 is the closest ortholog of N. crassa GLT-1. However, Mycth_112491 is only 352 amino acids long with six transmembrane helices (TMHs), compared with the typical 12 TMHs for glucose transporters predicted by TMHMM Server v2.0. The second closest GLT-1 ortholog is Mycth_108924, which has 12 predicted TMHs. Both Mycth_112491 and Mycth_108924 showed an elevated expression level on glucose compared with NC and cellulose conditions (89, 91). Mycth_2308157 and Mycth_2295230 are the closest orthologs of N. crassa HGT-1 and HGT-2, respectively. Both Mycth_2308157 and Mycth_2295230 showed increased expression on cellulose compared with glucose (91), while Mycth_2295230 also showed significantly higher expression in NC conditions than in glucose-rich conditions (89), consistent with the expression pattern of high-affinity glucose transporters. These data were supported by qRT-PCR (Additional file 9: Fig. S8A). We determined the expression levels of these glucose transporter-encoding genes in a ΔamyR mutant of M. thermophila. In glucose-rich conditions, both Mycth_112491 and Mycth_108924 showed significantly decreased expression, whereas Mycth_2295230 was upregulated in the ΔamyR mutant, compared with the WT strain (Fig. 8A). In NC conditions, Mycth_108924 and Mycth_2308157 showed decreased expression levels in the ΔamyR mutant compared with WT strain (Fig. 8B). This indicates that AmyR is a key transcription factor involved in regulation of the dual-affinity glucose transport system in M. thermophila, like COL-26 in N. crassa. In addition, like the Δcol-26 mutant of N. crassa, the ΔamyR mutant of M. thermophila was sensitive to osmotic stress and H2O2-induced oxidative stress compared with the WT strain, as shown by plate growth assays (Additional file 9: Fig. S8B) and the corresponding mycelial diameter (Fig. 8C). Since COL-26 orthologs and the dual-affinity glucose transporter system are ubiquitous in ascomycete species based on phylogenetic analysis (51, 69), the regulatory role of col-26 orthologs may also be conserved in many other filamentous fungal species.