Tolerance to alkaline ambient pH in Aspergillus nidulans depends on the activity of ENA proteins

Tolerance of microorganisms to abiotic stress is enabled by regulatory mechanisms that coordinate the expression and activity of resistance genes. Alkalinity and high salt concentrations are major environmental physicochemical stresses. Here, we analyzed the roles of sodium-extrusion family (ENA) transporters EnaA, EnaB and EnaC in the response to these stress conditions in the filamentous fungus Aspergillus nidulans. While EnaC has a minor role, EnaB is a key element for tolerance to Na+ and Li+ toxicity. Adaptation to alkaline pH requires the concerted action of EnaB with EnaA. Accordingly, expression of enaA and enaB was induced by Na+, Li+ and pH 8. These expression patterns are altered in a sltAΔ background and completely inhibited in a mutant expressing non-functional PacC protein (palH72). However, a constitutively active PacC form was not sufficient to restore maximum enaA expression. In agreement with their predicted role as membrane ATPases, EnaA localized to the plasma membrane while EnaB accumulated at structures resembling the endoplasmic reticulum. Overall, results suggest different PacC- and SltA-dependent roles for EnaB in pH and salt homeostasis, acting in coordination with EnaA at pH 8 but independently under salt stress.


Results
Genes coding for enA-like proteins in A. nidulans. As previously noted by Spielvogel and collaborators 16 , the genome of A. nidulans contains three genes coding for putative ENA proteins which were designated enaA (AN6642), enaB (AN1628) and enaC (AN10982). A multiple alignment of An-ENA sequences including ScEna1p is shown in Supplementary Fig. 1. EnaA, EnaB and EnaC show high conservation with ScEna1p (YDR040C) sequence (65%, 67% and 59%, respectively). Nevertheless, the three An-ENA proteins clustered in one of the eight clades that allocate the 22 ATPases identified in A. nidulans (Supplementary Fig. 2A) 26 . Conserved domains (Pfam domains) are found among ENA homologues denoting their roles as P-type ATPases, such as (I) cation-ATPase-N (pfam00690), (II) E1-E2 ATPase (pfam00122), (III) haloacid dehalogenase-like hydrolyase (HAD) (COG4087. pfam00702) and (IV) cation-ATPase-C (pfam00689) (Supplementary Fig. 2B). A more detailed analysis of A. nidulans ENA protein sequences compared to ScENA1p showed the presence of several small motifs characteristic of this family of transporters, such as the Actuator domain (TGES 183 ), the Mg 2+ binding motif (DGVND 761 ), and sequences belonging to the catalytic site of P-type ATPases: the signature sequences DKTGT 393 , TGD 675 and DPPR 652 and residues of the nucleotide binding domain (F 537 , K 542 , K 561 ) (indicated coordinates as in ScEna1p, Supplementary Fig. 1) 15,27 . Distribution of these functional motifs was also conserved among the An-ENA ATPases ( Supplementary Fig. 1). As for ScEna1p P-type ATPases, An-ENA transporters were predicted to harbour 10 putative transmembrane spanning regions (TMHMM and HMMTOP predictions) with N-and C-terminal regions being directed to the cytoplasmic side if located at the plasma membrane (Fig. 1A). Proposed two-dimensional model of EnaA and EnaB describing the predicted membrane topology and principal conserved functional domains among P-type ATPases. (B) Maximum-likelihood tree of Na + , K + , H + ATPases from fungi and yeast. EnaA and EnaB clustered in the same group of ATPases while EnaC clearly differentiated and belong to a separate clade. MEGA7 software was used to generate the tree 58 and it was edited with iTOL 59  Additional in silico searches showed the presence of at least one and up to four ENA-like ATPase coding genes in Saccharomycetes and filamentous fungal genomes (Fig. 1B). The phylogram, based on the amino acid sequence of putative and described Na + , K + , H + ATPases, revealed that EnaA, EnaB, and EnaC allocated in two different clades. While the EnaA and EnaB groups clustered, EnaC was included into a separate group of ATPases. EnaC shared similarities with HwEna1 and HwEna2 of Hortaea werneckii, which have been described to constitute a new family of ENA-like ATPases, although their functions are not completely determined yet 28 . Of interest is the fact that Ena P-type ATPases from Saccharomycetes constituted a monophyletic clade in this phylogenetic analysis.
In summary, fungal genomes encode several ENA-like ATPases (2)(3)(4) showing significant conservation in their sequences and composition of functional domains. However, specific transport activities as well as regulatory mechanisms or alternative functions must be determined experimentally and not only inferred from in silico predictions. functional analyses of AnenA transporters. To determine a functional role for An-ENA orthologues, single, double and triple knock-outs of enaA, enaB, and enaC were constructed by gene replacement and subsequent meiotic recombination to obtain all possible mutant combinations (see list of strains in Supplementary  Table 1). Single, double and the triple null ena mutant colonies grew on solid standard Aspergillus minimal medium (AMM) showing no appreciable alterations on radial growth or conidiation ( Fig. 2A, see AMM, row A). Sensitivity of ena mutant strains to an elevated extracellular concentration of sodium (1.0 M) or lithium (0.3 M) was tested ( Fig. 2A, rows B and C, respectively). When each the three-single knock-out mutants were grown in the presence of 0.3 M Li + and 1.0 M of Na + only enaB∆ showed a severe growth inhibition with the former (only a minor effect was detected with Na + ). The excess of Li + also affected all double and triple null mutants carrying the null enaB allele. A stronger sensitivity to Na + was also observed in double mutants containing enaB∆ and the triple mutant. In contrast, the double enaA∆ enaC∆ mutant showed a level of tolerance to sodium comparable to that of wild-type strain. These results identified EnaB as the principal element in tolerance to Na + and Li + stress.
Since alkalinity is a signal inducing expression of enaA gene (see references 16,18 , and below), tolerance of ena null mutant strains to alkaline pH was also analyzed ( Fig. 2A, row D). Single null ena strains grew as well as the wild-type strain at pH 8. However, double deletion of enaA and enaB resulted in strong sensitivity. Other double null mutant combinations displayed normal colonial growth. As expected, the triple null mutant displayed a pH 8-sensitive phenotype comparable to that displayed by the double enaA∆ enaB∆ strain. These results indicated that activities of EnaA and EnaB are jointly required for tolerance to ambient alkalinity.
Sensitivity of null ena strains to high concentrations of Na + and Li + was also tested at alkaline pH ( Fig. 2A, rows E and F). Besides the expected growth inhibition of strains lacking both EnaA and EnaB activities, a negative effect of combining alkalinity and high Na + concentration was not observed in the other null ena mutant strains. When alkaline pH and elevated Li + concentration were combined, growth of the wild-type strain was largely impaired. In this growth condition, the null enaC strain showed a wild-type phenotype and growth of the enaB∆ and enaB∆ enaC∆ strains were inhibited. In line with earlier observations, double enaA∆ enaB∆ and the triple ena mutants were sensitive to alkalinity. Interestingly the single enaA∆ strain showed a characteristic compact phenotype which was also observed for the enaA∆ enaC∆ mutant. These results indicate that EnaB has an important role in Li + tolerance especially under alkaline conditions, even contributing to increased tolerance when EnaA function is missing.
Determination of mycelial mass growth in liquid medium (Fig. 2B) confirmed the important role of EnaB in tolerance to excess of Na + and Li + . Meanwhile cultures of strains having the EnaB function showed comparable levels of mycelial mass in the presence of 1.0 M Na + , deletion of enaB reduced fungal growth to almost 50% (Fig. 2B). Deletion of enaB had a greater effect in liquid medium containing 0.3 M Li + , with the strongest growth inhibition observed. Mycelial masses measured at alkaline-pH also showed a significant reduction in the absence of EnaB activity in single and double null enaB enaC mutants, but this effect was more severe when EnaA activity was lost, as shown before on solid minimal medium. In summary, EnaA and EnaB have a principal role in adaptation to an alkaline pH environment. But EnaB function seems to be required also for Na + /Li + adaptation and is crucial for Li + homeostasis at alkaline pH. transcriptional regulation of ena genes. Series of Northern blot analyses were conducted to analyze the regulation of expression of ena genes in total RNAs isolated from mycelia grown at alkaline pH, in excess of Na + or Li + , and the mixture of high levels of either Na + or Li + with alkalinity. In agreement with previous results from Northern analyses 16,18 and RNA sequencing 10 , we confirmed that the expression of both enaA and enaB is regulated by these stresses. Under non stressing conditions (C, control condition), expression of enaA was detected and a positive effect was observed after addition of large amounts of Na + or Li + , or by alkalinizing the cultures (Fig. 3A). Addition of 1.0 M Na + or 0.3 M Li + resulted in a moderate increase of enaA mRNA levels, sixfold and fivefold change, respectively, but a stronger effect, 19-fold change, was observed in response to alkaline pH, with a noticeable peak of expression at 30 min (Fig. 3B). Yet, when elevated Na + concentrations were combined with pH 8, transcript levels continued increasing throughout the experiment (60 min) showing a 27-fold change with respect to initial values. However, this synergic response was not observed when combining alkalinity and 0.3 M Li + . Despite the fact that reads mapping at the enaA locus in our RNA-seq experiments 10 suggest the existence of alternative splicing events at the three introns predicted in the coding region (i.e. the processed form seems to be more abundant in the wild-type at pH 8 than under standard culture conditions), Northern hybridization did not indicate the generation of transcripts of different sizes (Fig. 3A).
Northern blot in Fig. 3A, and its quantification in 3B, shows that enaB transcript was also detected under non stressing conditions. Variation of transcript abundance of enaB and enaA genes seemed to follow a comparable www.nature.com/scientificreports/ trend, but with lower intensity in the case of enaB, probably due to higher levels of expression under non-stressing conditions, compared to enaA. Even so, differences were observed principally regarding Na + stress, which induced a continuous increase in enaB expression through 1 h of treatment. As previously noted for enaA, combination of high Na + and alkaline pH resulted in a higher expression than singly implemented stress conditions. It is worth mentioning that enaB mRNA levels remained almost unchanged after addition of 0.3 M Li + , which contrasts with the severity of the growth defect shown by the null enaB strain under these stress conditions but on solid medium. In the case of enaC, Northern blot analyses detected two possible transcripts for this gene (indicated with arrows, right side Fig. 3A), being the high mobility band slightly induced by the presence of alkaline pH. The size of the two hybridization signals did not coincide with those of enaA or enaB transcripts, and thus, a crosshybridization phenomenon was excluded. The presence of two transcripts may be a consequence of possible alternative splicing events at introns 1 and 3 of the coding region, as shown by our RNA-seq data. Furthermore, the last intron of enaC is not processed, adding 13 codons to the sequence but without altering the reading frame. The corresponding set of 13 amino acids is part of the fifth transmembrane domain. The corrected EnaC sequence, including the additional 13 amino acids, was used in the analysis shown in Fig. 1 and Supplementary  Fig. 1 (see also below in Fig. 4A). However, these 39 nucleotides do not explain the difference in size observed between both enaC transcripts. Considering that multiple RNA-seq reads map to the intergenic region for enaC and An7665 (not shown), and that this region would include an additional target site for PacC and two sites for www.nature.com/scientificreports/ SltA, it is tempting to suggest that the presence of two transcription start sites could account for the existence of two enaC transcripts. MsnA is a transcription factor involved in saline stress response 18 . Figure 3A shows expression levels of msnA following a similar pattern to that described before for enaB. High Na + elevated the expression of msnA (Fig. 3A,B). In contrast, high Li + had no effect or even we detected a negative effect of this cation by preventing activation of msnA expression in the presence of alkaline pH. These results suggest a common transcriptional regulatory mechanism acting on msnA and enaB genes, and probably on enaA.
To further investigate a possible influence of Ena proteins in their own transcriptional regulation, expression of the remaining ena genes was studied in single knock-out mutants enaAΔ and enaBΔ. Figure 3C shows that the temporal profile of enaB, enaC and msnA did not significantly change in the absence of enaA. Likewise, in the absence of enaB (Fig. 3D), no major differences were observed in enaC and msnA gene regulation, whereas the response of enaA to the combination of alkalinity and Li + resulted in a stronger early (maybe compensatory) response (30 min) compared to samples from cells where enaB is present.
Taken together, these results indicate that separately Na + , Li + and alkaline pH may induce enaA and enaB expression at different levels. However, a combination of high Na + and alkalinity caused the greatest levels of expression for these genes, suggesting a major need of these ATPases under both stresses. Interestingly, the presence of lithium prevented the synergic effect of alkaline pH, indicating the presence of differential regulatory mechanisms acting under each cation stress. www.nature.com/scientificreports/ Regulation of ena genes by pacc and SltA transcription factors. Since Na + and Li + cations, alkalinity and its combination differently modulate enaA and enaB expression, the role in ena expression of two transcription factors mediating response to ambient pH and cation stresses, PacC and SltA, was monitored 16,20 .
To study the role of PacC on the regulation of ena genes, we analyzed their expression levels in two opposed ambient-pH mutant backgrounds. A mutant with a null palH72 mutation prevents proteolytic processing of the primary translational form of PacC, PacC72kDa, disabling functionality 29 . On the other hand, the pacC c 14 mutation, constitutively generates the active 27 kDa version of PacC, at any ambient pH 30 . Northern analyses showed the strict requirement of proper PacC signalling for enaA and enaB expression (Fig. 4B). Transcripts of enaA and enaB were not detected in the null palH72 background under any condition (Fig. 4B). In contrast, detection of putative enaC transcripts and expression of msnA remained essentially as described in a wild-type PacC background (compare with Fig. 3A).
The constitutive activation of PacC did not result in a pH-independent expression of ena genes. Expression of enaA in the pacC c 14 mutant required addition of Na + or ambient pH alkalinisation, but the expected synergic effect by combining both stresses was not observed (Fig. 4C). Excess of extracellular Li + had a partial effect and enaA expression was high at 30 min but strongly inhibited at 60 min. A similar effect was observed in samples treated with Li + and alkaline pH. In contrast, the enaB transcript showed almost constant levels under most conditions tested, but a strong transcriptional response after alkaline stress induction was still observed at 60 min. A negative effect of Li + on enaB expression was observed at 60 min, with or without media alkalinisation, and also, the mixture of Na + and alkaline pH did not induce the expected positive effect. Transcript levels of enaC and msnA followed similar patterns to those described in the wild-type or palH72 backgrounds. These results suggest the existence of two additional regulatory mechanisms with opposite effects: a positive alkaline-pH dependent effect and a negative high-Na + modulation.
To determine the role of SltA on their regulation, the expression profiles of ena genes were analyzed in a sltA∆ mutant strain under the above stress conditions. Figure 4D shows that the expression pattern of enaA in the null sltA strain was similar to that described in the wild-type background. Absence of sltA resulted in higher basal enaB transcript levels which were increased at alkaline pH but, as mentioned above, a negative effect was also detected when high concentrations of Na + or Li + were added to cultures. The expression pattern of msnA nearly paralleled that of enaB.
In agreement with possible direct roles of PacC and SltA on the regulation of enaA and enaB genes, we found putative target sites for these transcription factors in their promoters. Searches confirmed the presence of five consensus PacC (5′-GCC AAG -3′) and three SltA (5′-AGGCA-3′) sites in enaA promoter (1 kb upstream of ATG). Six PacC and one SltA target sequences in enaB promoter and two PacC and four SltA sites in enaC promoter (Fig. 4A). Number and locations of PacC sites in enaA and enaB promoters are in accordance with a direct role for this TF, as has been demonstrated before for other genes (see discussion). Scattering of SltA sites might be a cause for the reduced role of this TF in their regulation.

Expression profiles of EnaA and EnaB proteins under stress conditions. To further analyze EnaA
and EnaB function, we constructed strains expressing C-terminally GFP-tagged versions. These strains showed a wild-type phenotype in all conditions tested (not shown), indicating that the chimaeras did not cause any dysfunction to the ENA system. Both EnaA-GFP and EnaB-GFP were detected at very low levels in mycelia grown in AMM without Na + /Li + or alkaline pH stresses, in line with Northern blot results. Immunodetection experiments further confirmed barely detected EnaA-GFP (with expected size of 150 kDa) in protein extracts from mycelia grown for 19 h and 24 h in standard AMM (C A in Fig. 5A,B, respectively). In the case of EnaB-GFP, and in agreement with higher levels of enaB transcript under basal AMM conditions, we detected a band with the expected size of the full chimera, nearly 150 kDa (see arrow in lanes C B , Fig. 5A,B). We also perceived a smear of low mobility bands starting at the limit of the SDS polyacrylamide resolution gel (indicated with a square bracket). All these forms, primary and low mobility, became more evident when protein extracts of mycelia were subjected to alkaline pH stress (Fig. 5). These dispersed high-molecular weight species of fusion proteins might reflect post-translationally modified forms. EnaA-GFP and EnaB-GFP were expressed early in alkaline medium compared to the effect of adding Na + or Li + (compare the relative band intensities of EnaA-GFP and EnaB-GFP in Fig. 5A,B). However, EnaB-GFP was expressed under standard culture conditions and although alkaline pH induced the strongest response, the chimera was also detected 1 h after Na + and Li + treatments. Overall, results of immunodetection experiments suggest that compared to pH 8, an increase in the concentration of both proteins under Na + or Li + stress conditions is delayed. Subcellular localization of enaA and enaB. EnaA and EnaB are expected to be transmembrane proteins. In agreement with previous protein expression results, cells from enaA-gfp and enaB-gfp transformants showed no fluorescence in standard microscopy medium (WMM; not shown). For fluorescence observation, cells were exposed for 60 min to alkaline pH (by addition of 100 mM Na 2 HPO 4 to WMM). The images depicted in Fig. 6A show that, under alkaline conditions, EnaA-GFP displayed preferential plasma membrane localization (PM). A regular distribution of fluorescence was observed at PM of apical and distal hyphal compartments. Thus, response to alkaline pH seems to involve transport to, and regular distribution of EnaA ATPase to all parts of a hypha avoiding a preferential accumulation at the cell tip, where major exocytic and endocytic processes have been shown to occur [31][32][33] . Fluorescence of EnaA was present at the apical region (t), at branching points (b) and also between compartments, where septa were formed (s), as well as in spots dispersed within the cytosol. At earlier times of induction (15-30 min exposure to pH 8, images at the bottom of panel A), EnaA-GFP fluorescence was visible along the apical compartment but location at PM was not uniform, indicating the existence of Scientific RepoRtS | (2020) 10:14325 | https://doi.org/10.1038/s41598-020-71297-z www.nature.com/scientificreports/ clusters of transporters probably associated to specific regions at the PM such as lipid rafts (indicated with white arrowheads and dotted lines). After induction of the expression of the EnaB-GFP chimera by alkaline pH, microscopy showed a preferential location of EnaB in intracellular, elongated and round structures (Fig. 6B). Elongated structures accumulated at the tip of the cell while round ones distributed at basal locations of hyphal compartment. Noticeably, EnaB-GFP localizations resemble those illuminated by Sec63-GFP, a protein associated to ER and nuclear periphery (possible nuclei, black arrowheads, Fig. 6C 34 ). Presence of EnaB at the periphery of nuclei is shown in Supplementary  Fig. 3. Thus, EnaA and EnaB have distinct locations, being EnaA located at the PM and EnaB probably in ER as well as additional intracellular membranous compartments such as nuclei, Golgi and transport vesicles.
Alkaline pH sensitivity of loss of function mutants in the pH regulatory system is not a result of reduced expression of enaA gene. The strong dependency of enaA expression on the Pal/PacC pathway and its preferential PM localization, led us to speculate that EnaA might be one of the major factors in determining alkaline pH tolerance in the absence of a functional Pal/PacC pathway. Thus, we analyzed whether a PacC independent expression of enaA could rescue the alkaline-sensitive phenotype caused by a loss-of-function mutation in the Pal signalling pathway such as palA1.
To express enaA in a pH-independent manner, we transformed a wild-type strain with a plasmid carrying an insert in which the coding region of enaA was under the control of a truncated version of the gpdA promoter 35 , gpdA mini , which leads to constitutive but moderate expression levels 32 . Of selected transformants, MAD3042 carried a single copy and MAD3043 two copies of the construct integrated at the pyroA locus. Both strains displayed a wild-type phenotype for salt or alkaline pH tolerance, although, as expected, different expression levels of enaA independent of pH condition were detected (Fig. 7A). The amount of enaA mRNA increased consistently with the copy number and in the single copy gpdA mini ::enaA strain enaA mRNA levels were similar to those produced by www.nature.com/scientificreports/ the endogenous enaA locus at alkaline pH. Subsequently, both strains were independently crossed with a palA1 mutant strain (MAD1865) 36 . Progeny carrying one or two copies of gpdA mini ::enaA and the mutation palA1 were selected and analyzed (strains I25 (single copy) and I4 (double copy)). Using Northern analysis, we confirmed a constitutive expression of enaA independent of inducing conditions and of a functional Pal/PacC pathway in both strains (Fig. 7B). However, phenotypic analysis could not detect any effect of expressing enaA in a palA1 mutant background regarding salt or pH sensitivity (Fig. 7C). Indeed, addition of neomycin (2 mg/mL) to the culture medium showed hyper-tolerance of palA1 strains expressing gpdA mini -driven enaA to this compound, as has been traditionally described for this group of mutants 37 . Therefore, PacC-independent expression of enaA does not suppress the sensitivity to alkaline pH, supporting the broad regulatory activity of a functional Pal/ PacC pathway and the complexity of alkaline pH response by A. nidulans.

Discussion
In this work we investigated, through a combination of genetics and cell biology techniques, the roles of the three putative sodium ATPases of the ENA family identified in A. nidulans 16,19,38 . Sequence similarities, domain distribution and the phylogenetic association ( Fig. 1) indicate that A. nidulans Ena-like ATPases probably fulfil the same biological function described for several filamentous fungal and yeast ENA homologues. The action of Ena P-type ATPases in fungi has been commonly related to potassium, lithium and sodium ion export from the cytoplasm, especially at alkaline and high salinity conditions [39][40][41] . However, phylogenic comparison supports the existence of genetic variability of Ena P-type ATPases, which are believed to have evolved from an ancestral K + ATPase by gene duplication in adaptation to salt and/or alkaline conditions 13 . Thus, it is not surprising that evolved functional variants cannot be directly inferred by sequence similarity analyses. This is exemplified by unsuccessful attempts to suppress growth defects in S. cerevisiae ENA mutants by expression of an Ena P-type ATPase from Neurospora crassa 12 or from Hortaea werneckii 28 . Of the three Ena proteins found in A. nidulans, EnaA and EnaB share higher sequence conservation than with EnaC, suggesting a different phylogenetic origin or divergent evolutionary paths. In contrast to previously cited ENA functions in other organisms, phenotypic analyses of A. nidulans null enaA strains clearly established that a functional EnaA ATPase is not essential for Na + or for Li + detoxification, www.nature.com/scientificreports/ but it is necessary at alkaline pH (Fig. 2). A different scenario was found when studying EnaB function. EnaB is necessary for growth in high salinity (Li + and Na + ) and alkalinity environments. Strong sensitivity to alkalinity is observed in the absence of both EnaA and EnaB proteins, suggesting the existence of overlapping functions of these ATPases in adaptation to alkalinity. At alkaline conditions the importance of EnaB in Na + and Li + tolerance is more prominent, being indispensable in the latter. Thus, the described properties of EnaB are more similar to the function that has been observed for typical Ena-like P-type ATPases. The function of EnaC remains, as yet, undefined. In our phylogenetic analyses, EnaC clustered together with HwENA1 and HwENA2 of H. werneckii; however, a requirement of EnaC function in adaptation to alkaline condition or salinity was not observed in this work. This suggests that EnaC undergoes specific spatial and temporal regulations, is activated by other cations or may even have become a pseudogene during Ena evolution in A. nidulans. www.nature.com/scientificreports/ The above-described differential expression patterns of A. nidulans ENA proteins correlate with what has been observed in other fungal species. Indeed, ena expression patterns in response to high external ion concentrations and pH differ significantly among orthologues and paralogues of different fungal species. In salt tolerant species such as the Dothideomycete H. werneckii (HwENA1 and HwENA2) or the Saccharomycete Zygosaccharomyces rouxii (ZENA1), expression of Ena P-type ATPases is induced only at elevated sodium ion concentrations, (5 M and 3 M, respectively 28,41 ). In Debaryomyces (formerly Schwanniomyces) occidentalis, ENA1 and ENA2 transcript levels are very low at neutral pH and 0.2 M NaCl concentration 42 . On the other hand, in S. cerevisiae, which is more salt sensitive, ENA1 expression is strongly induced by addition to the medium of 0.4 M Na + , 0.05-0.1 M Li + , as well as by alkaline pH 5 . UmEna1 of Ustilago maydis has been characterized as a typical Na + and K + efflux ATPase and is strongly induced by 0.5 M NaCl, whereas UmENA2 is only marginally induced 40 .
The present work highlighted that A. nidulans ena genes are differentially regulated by Na + , Li + and alkaline pH. We determined that enaA is weakly expressed in standard culture conditions, AMM with pH close to neutral (~ 6.8), and is induced by 1.0 M NaCl, 0.3 M Li + and by alkaline pH (8.0) (Fig. 3A). Furthermore, at alkaline pH, EnaA expression is strongly elevated. A similar effect has also been described for UmEna1 and UmEna2 and HwENA1 and HwENA2 28,40 . The highest expression levels for this group of Ena P-type ATPases are achieved when alkalinity is combined with high-Na + . However, these high gene expression levels do not correlate with an essentiality of the gene function, as observed in the phenotypic analysis of the null enaA strain. enaB followed a comparable expression profile to that of enaA, although in general with lower concentrations than those of enaA (except for basal conditions). Still, differences were observed, especially regarding to Na + -induction time, which was later in the case of enaB. A hypothetic coordination of an early response mediated by enaA and a late response mediated by enaB would require both ATPases to efflux Na + at alkaline pH. It is interesting that, despite the essentiality of EnaB on Li + tolerance at neutral and at high pH conditions (Fig. 2), Li + did not induce such a remarkable increase in enaB expression as did Na + or alkaline pH.
Our work also focused on the transcriptional regulation of Ena-like ATPases. In S. cerevisiae, in addition to the Pal/PacC (Rim101) pathway, the calcineurin/CRZ1 and high osmolarity glycerol (HOG) osmo-responsive pathway are involved in the control of ENA1 expression together with a regulation by nutrients via Snf1 and Mig1,2 (reviewed in 5,43 ). Our previous work demonstrated that enaA expression is independent of CrzA 16 . AtfA of A. nidulans was characterized as a candidate transcription factor acting downstream of HogA/SakA MAPK and playing a role in oxidative and osmotic stress responses 19,38,44 . Expression of enaA has been shown to be independent of AtfA, but enaB and enaC displayed AtfA-dependent expression especially in response to oxidative stress 38 . This observation establishes a putative link between EnaC and the response to oxidative stress that should be assessed more deeply in the future.
Elevated transcript levels of enaA and enaB at alkaline pH, together with the presence of several putative binding sites for PacC (5′-GCC AAG -3′) at their promoters, supported a role of the Pal/PacC regulatory pathway and we showed that PacC is an essential positive regulator of enaA and enaB expression. In the absence of the Pal signalling pathway (palH72) that finally activates the PacC transcription factor, neither enaA nor enaB transcripts were detected. In the constitutive PacC background (pacC c ), different transcript levels were obtained to those observed under inducing conditions in the wild type background. This result contrasts with previous reports on the mechanisms PacC-dependent gene regulation (e.g. ipnA 45 ). The absence of a full upregulated expression of enaA and enaB in the pacC c 14 background reveals the need of at least one additional regulatory pathway to influence enaA and enaB expression in response to salinity. A minor positive regulatory role for the salt activated transcription factor SltA on expression of ena genes has been shown in this and previous works 16 . This is in agreement with our RNA-seq results 10 , in which both enaA and enaB (but not enaC) were among the most intensely upregulated genes at pH 8 in a wild-type background (Top45 and Top2 genes with the highest log2FC values, respectively), but none of them was significantly downregulated in the sltAΔ mutant under these stress conditions. One possible factor causing the different cation and pH sensitivity of enaA and enaB null strains could be the different subcellular localization of the proteins. EnaA, as S. cerevisiae Ena1p 46 , localizes in the plasma membrane, whereas EnaB localizes in an intracellular membranes system that resembles the ER and Golgi network. However, since Li + is very toxic at low concentrations, compartmentalization into it by EnaB could greatly increase Li + tolerance. Results described in previous reports demonstrated that Ustilago maydis UmEna2p and Neurospora crassa NcEna2 15,40 localize in the proximity of the ER and other endomembranes.
We further considered that, since a functional enaA or enaB was necessary for alkaline tolerance, constitutive expression of enaA could be sufficient to rescue the alkaline sensitive phenotype of a Pal/PacC loss-of-function mutation. A strain constitutively expressing enaA in a palA1 genetic background maintained the palA1 phenotype, however, confirming that other Pal/PacC-controlled elements are required. This is in agreement with the envisaged role of PacC as a regulator of multiple pathways in response to environmental pH changes. In this context, next-generation sequencing and analysis tools should help unveil further details of this complex process.
Methods fungal strains and growth conditions. Aspergillus minimal media (AMM) adjusted to pH 6.8, containing 2% glucose and 71 mM sodium nitrate as main carbon and nitrogen sources, respectively, was prepared as described previously 47 . A. nidulans strains used in this study are listed in Supplementary Table 1. Strain MAD1427 was used for the systematic deletions. Meiotic crosses and analysis of the progeny followed standard procedures as described previously 48 . Strain BD575 was obtained from the progeny of a cross between MAD2446 and BD488, and strains BD604 and BD612 were obtained from the crossing between BD486 and BD575. Progeny was verified by PCR analysis using oligonucleotides listed in Supplementary www.nature.com/scientificreports/ Phenotypic analyses of mutant strains were done on solid AMM supplemented with a range of compounds to generate cation or pH stress. Sodium and lithium stress were induced by addition of 1.0 M NaCl and 0.3 M LiCl, respectively. Alkaline stress was induced by adjusting media pH to 8 with either 50 mM Tris-HCl pH 8 or 100 mM Na 2 HPO 4 , always with similar results. Specifically, the use of 100 mM Na 2 HPO 4 on solid medium was avoided to prevent precipitation when combining with lithium chloride. Radial extension of mutant strains was always compared to that of the reference wild-type strain. Strains were incubated at 37 °C for 2 days and at least 3 replicates per strain and condition were analyzed.
For investigating the effects of alkali-cations and alkaline pH on biomass production, 1 × 10 6 spores of the mutant strains of interest were inoculated in liquid medium (30 mL of medium in 100 mL flask) with or without addition of the stress agent (1.0 M NaCl, 0.3 M LiCl or 50 mM Tris-HCl pH 8). After cultures were incubated for 24 h at 37 °C in an orbital incubator at 200 rpm, mycelia were collected, dried at 100 °C and weighed. Three biological replicates were performed per strain and condition. For data normalization, growth of each strain at liquid AMM without addition of the stress agent was designated as 100%. Data were presented as percentages. Statistical significance of differences observed in biomass production between the wild-type and mutant strains was evaluated using Two-tailed Student's t-test for unpaired samples.
construction of deletion cassettes for enaA, enaB and enaC. Gene deletion was achieved by using deletion cassettes comprising Aspergillus fumigatus pyrG (AN6642/enaA; AN10982/enaC), riboB (AN1628/enaB), or pyroA (AN10982/enaC) genes, respectively, as selectable markers flanked by 700-1,500 bp up-and downstream of the targeted A. nidulans open reading frame. The selectable markers were amplified with primer pairs gsp2*-gsp3* (listed in Supplementary Table 2), which contained tails (20 bp) specific to each terminal of both the 5′ and 3′ ends of the targeted gene, and using the cloned selectable marker as a template present on plasmids (see Supplementary Table 3). Fragments corresponding to the 5′-UTR and 3′-UTR of the targeted genes were amplified using genomic DNA of A. nidulans as template and oligonucleotide pairs gsp1-gsp2 or gsp3-gsp4, respectively. The final cassette was generated by fusing the three fragments by Fusion PCR using primers gsp1-gsp4 49 . Cassettes were purified and directly used to transform A. nidulans protoplasts (Supplementary Table 1) following the protocol described previously 50 . The double-null mutants were generated by sequential transformations of strains deficient in heterologous recombination 51 with the required fusion cassettes. Prototrophic transformants were isolated and deletion of the targeted open reading frame confirmed by Southern-blot 52 after genomic DNA extraction 53 .
Gfp tagging of enaA and enaB. Construction of A. nidulans strains expressing C-terminally GFPtagged proteins followed the 3-way PCR methodology described previously 49 , ~ 1.5 kb from the most 3′ end of the target gene and the 3′ UTR region were amplified with primers gsp5-gsp6 and gsp3-gsp4, respectively (Supplementary Table 2), and fused in a 3-way PCR with a fragment, amplified with primers gsp6*-gsp3*, containing the gfp gene together with the selectable pyrG gene of A. fumigatus 49 . The fusion cassette was transformed into recipient MAD1427 strain that carries pyrG89 mutation. Uracil and uridine prototrophs were selected, purified to homokaryosis and then analyzed by Southern-blot for verification of a single integration event of the gene replacement cassette. Functional replacement was verified by phenotypic analysis using wild-type and deletion phenotypes as references.
constitutive expression of enaA. The enaA coding sequence was amplified using primers EnaAup and EnaAdown, which introduced EcoRI restrictions sites for further cloning into plasmid p1660 32 . This plasmid carries a truncated version of gpdA promoter, gpdA mini , allowing constitutive expression of target gene, and a truncated version of pyroA gene, restricting integration at the pyroA locus by complementation of pyroA4 mutation (Table Supplementary 3). Selection of positive transformants is based on recovery of pyrimidine prototrophy (see also 54 ). The resulting plasmid was named pGPDA-EnaA and the integrity of the enaA open reading frame was confirmed by sequencing. Protoplasts of strain MAD1739 were transformed using pGPDA-EnaA and positive transformants and copy number of integration events at the pyroA locus were confirmed by Southern blot analysis.

RnA extraction and gene expression analyses.
For RNA extraction, 1 × 10 6 spores of wild-type and single null mutant strains were inoculated in fermenters and cultured for 16 h at 37 °C and 200 rpm. Mycelia were harvested and transferred to flasks with fresh medium containing 1.0 M NaCl, 0.3 M LiCl or pH 8.1, with further incubation of the culture as specified in the figures. Mycelia samples were collected by filtration, frozen in liquid nitrogen and grounded.
Total RNA was isolated following manufacturer's protocol by adding 1 mL TRIzol reagent (Fluka, Sigma-Aldrich Quimica SL) to 100 mg of grounded samples. Total RNA concentration was calculated using a Nanodrop 2000c system (Thermo Fisher Scientific, Waltham, MA). Northern blot assays were carried out as described previously 55 . 10 μg of total RNA per sample was loaded in 1.2% agarose gels and transferred to positively charged nylon filters (Roche). Equal loading of total RNA was evaluated by methylene blue staining of rRNA. Transcripts of A. nidulans ena-like genes were detected using specific PCR-amplified probes as described previously 16 . mRNA levels of msnA gene (AN1652) were detected using a probe of 1,995 bp amplified using primers described in Table S2 and which covered 100% of the ORF. Labelling of DNA probes was performed using Roche radioactive labelling kit, detected using a PhosphorImager screen (Molecular Dynamics) and developed using a FLA-5100 Reader (Fujifilm). Band intensity quantification was performed using Multi-Gauge V3.0 software (Fujifilm). Scientific RepoRtS | (2020) 10:14325 | https://doi.org/10.1038/s41598-020-71297-z www.nature.com/scientificreports/ protein isolation and western blot. Sodium, lithium and alkalinity-shock effects on protein concentration was studied using western-blot technique. Total protein extracts were isolated from mycelium of GFPtagged strains cultivated for 16 h at 37 °C in appropriately supplemented Cove's minimal medium 56 plus the indicated time after addition of the stress agent. Mycelia were harvested by filtration, frozen in dry ice and lyophilized for 16 h. Protein extraction from lyophilized samples was performed by alkaline-lysis extraction procedure described previously 22 . Proteins were resolved in 8% SDS-polyacrylamide gels and transferred to nitrocellulose membrane using TransBlot ® Turbo™ System (Bio-Rad). GFP-tagged versions of EnaA and EnaB were detected using mouse anti-GFP monoclonal antibody (1:5,000 dilution; Roche). As loading control, gamma-actin was detected using mouse anti-actin C4 antibody (1:5,000 dilution; MP Biomedicals). As secondary antibody peroxidase-conjugated goat anti-mouse IgG immunoglobulin (1:4,000 dilution; Jackson) was used. Western blots were developed using ECL kit (GE Heathcare) and images were taken using a Luminescent Image Analyzer LAS-3000 (Fujifilm) and processed with Multi-Gauge V3.0 software (Fujifilm).
fluorescence microscopy. Aspergillus nidulans conidiospores were inoculated in uncoated glass-bottom µ-dishes (Ibidi GmbH) containing 2.5 mL watch minimal medium (WMM 57 ) supplemented with 25 mM NaH 2 PO 4 , 5 mM ammonium (+)-tartrate, and 0.5% glucose. After 16 h at 25 °C medium was replaced with freshly prepared WMM supplemented with 100 mM Na 2 HPO 4. Samples were cultured for an additional 1.5 h at 37 °C prior to observation of fluorescence (see text for details of enaA and enaB induction). Differential Interference Contrast (Normarski optics) and fluorescence images were acquired from in vivo cultures with a Leica DMI-6000b inverted microscope coupled to an ORCA-ER digital camera (Hamamatsu Photonics) and equipped with a 63 Plan Apochromat 1.4 N.A. oil immersion objective (Leica) and a GFP filter (excitation 470 nm; emission 525 nm). Images were acquired using Metamorph (Molecular Dynamics) software and processed using ImageJ free-software (https ://image j.nih.gov/ij).

Bioinformatics.
Multiple sequence alignments were made using BLAST and Clustal Omega open source program (https ://www.ebi.ac.uk/Tools /msa/clust alome ga/). Jalview application was used for alignment visualization and domain analyses. Transmembrane domains of Ena-like proteins were predicted using Hidden Markov Models (HMM) in the Institute Pasteur Mobyle server (https ://mobyl e.paste ur.fr/). Phylogenetic trees were generated using Mega Version 7.0 software 58 . Maximum-Likelihood (80 replicates) and Neighbor-Joining (10,000 replicates) methods rendered the same organization of the clades. The Maximum-likelihood tree was edited using iTOL 59 .