Deletion of the HAMP domains from the histidine kinase CaNik1p of Candida albicans or treatment with fungicides activates the MAP kinase Hog1p in S. cerevisiae transformants

Immune-compromised patients are at high risk of becoming infected by opportunistic fungi, such as Candida and Aspergillus sp. Candida sp. are the fourth most frequent cause of hospital acquired blood stream infections and up to 90% of HIV patients receive mucosal candidiasis at least once [1]. Although infections with non-albicans Candida sp. have emerged in recent years [2], the species C. albicans is still responsible for the majority of the cases [3, 4].

Several antifungals are available in the market, yet, toxicity and/or development of resistance represent major concerns [5]. Among these is the former “gold standard” therapeutic amphotericin B that invariably causes toxicity in patients, negating the importance of its fungicidal activity. Although azoles and echinocandins represent the most widely used treatments of candidiasis, the acquisition of resistance can occur, leading to the risk of recurrent infections [6, 7]. Thus antifungals which impact new targets and have minimal side effects are urgently needed [7].

In fungi, two-component signal transduction (TCST) systems have been implicated in osmotic and oxidative stress responses, cell-cycle control, red/far-red light responses, and virulence switches from non-pathogenic to pathogenic states [8–10]. Since TCST systems are absent in mammalian cells, they are attractive targets for the development of new antifungals with probably minimal side effects in humans [7].

Typical TCST systems in fungi include a histidine kinase (HK), a histidine phosphotransfer protein (HPT) and a response regulator protein (RR). The best understood fungal TCST system is part of the High Osmolarity Glycerol (HOG) pathway in S. cerevisiae. In the absence of osmotic stress, the transmembrane HK ScSln1p is active. This HK activity leads to phosphorylation of a histidine residue in the catalytic domain, the so-called HisKA domain, from which the phosphate group is transferred to an aspartic acid residue in an internal receiver domain (REC). Therefore, these HKs are called hybrid HKs. The phosphate group is then shuttled through the HPT protein Ypd1p to the terminal RR proteins Skn7p and Ssk1p [8, 11]. Phosphorylated Skn7p is a direct regulator of gene expression, whereas phosphorylated Ssk1p is not able to activate downstream targets. Under conditions of high osmolarity, ScSln1p is inactive, resulting in the dephosphorylation of both RRs. Unphosphorylated Skn7p becomes inactive, whereas unphosphorylated Ssk1p activates a downstream mitogen-activated protein kinase (MAPK) module, in particular the MAP3K Ssk2p resulting in phosphorylation of the MAPK Hog1p [7, 12–15]. Phosphorylated Hog1p upregulates the transcription of genes, which encode enzymes that play a key role in glycerol production and maintenance of the intracellular water balance, allowing adaptation to high-osmolarity conditions [13]. Thus, the HK ScSln1p is a negative regulator of the MAPK Hog1p. Likewise, disruption of ScSLN1 results in the accumulation of unphosphorylated Ssk1p without external stimulus and thus, constitutive activation of Hog1p, which is lethal [14].

While S. cerevisiae has a single HK, namely ScSln1p, C. albicans has three HKs: CaSln1p, CaNik1p (also called Cos1) and Chk1p [8]. CaNik1p is considered to be a cytosolic enzyme, as it lacks the hydrophobic amino acids indicative of membrane-spanning domains (Figure 1) [16]. The protein is not essential for survival, and a gene deletion mutant could be generated [16–18]. CaNik1p plays an important role in hyphal formation in C. albicans on solid media [8, 18]. Additionally, the deletion mutant was found to be less virulent in a mouse model for systemic candidiasis [8]. According to the classification scheme of HKs [9], ScSln1p and CaSln1p are group VI HKs while CaNik1p is a group III HK.

Several chemical classes of fungicides, such as phenylpyrroles (fludioxonil), dicarboximides (iprodione) and polyketide secondary metabolites of ambruticins, exert their antifungal effects by activating the HOG signaling pathway, resulting in the accumulation of both glycerol and free fatty acids [19–22]. It is assumed that in the absence of high external osmolarity, artificial induction of excess intracellular glycerol causes leakage of cellular contents and ultimately results in cell death [21, 22]. Mutations in group III HKs are frequently associated with fungicide resistance [19], showing the relevance of these enzymes for fungicide activity and placing also these HKs upstream the MAPK Hog1p. It is still discussed, whether group III HKs are negative (as is ScSln1p) [23] or positive [24] regulators of Hog1p. S. cerevisiae lacks group III HKs and is usually resistant to the fungicides mentioned above. However, fungicidal sensitivity is gained by heterologous functional expression of group III HKs in S. cerevisiae correlating with Hog1p phosphorylation [25–28].

All classes of HKs share the conserved phosphate-accepting domains HisKA, REC and an ATP-binding domain called HATPase_c domain. Based on homology with other HKs, the histidine residue 510 and the aspartate residue 924 in CaNik1p were identified as the essential phosphorylatable residues for the HisKA and the REC domains respectively (Figure 1) [17]. The ATP-binding domain comprises a characteristic N-box with two asparagine residues, which are N623 and N627 in CaNik1p [17]. The N-box is known to be essential for ATP binding [29] and deletion of a single asparagine residue was associated with complete inhibition of ATP binding in the HK EnvZ [30].

Group III HKs are characterized by additional amino acid repeats in the N-terminal part with a length of approximately 90 amino acids each. The repeats contain evolutionary conserved amino acid sequences called HAMP domains. Such abbreviation is due to the frequent occurrence of such domains in histidine kinases, adenylcyclases, methyl accepting chemotaxis proteins and phosphatases, which are proteins associated with signal transduction in both prokaryotic and lower eukaryotic organisms [31]. More than 26400 proteins with HAMP domains exist in the SMART data base. These domains were shown to play an active role in intramolecular signal transduction in prokaryotic sensor kinases. They are composed of about 50 amino acid residues each with two amphipathic helices [32–34] which probably rotate when the sensor domain of the protein is activated as recently elucidated from NMR analysis [35, 36]. Unlike the bacterial HK, which usually possess a single HAMP domain, fungal group III HKs have several consecutive HAMP domains. In the five N-terminal amino acid repeats of CaNik1p [16–18] we identified nine HAMP domains of a concatenated structure forming four pairs each with an overall length of 92 amino acids and a single HAMP domain in the remaining truncated amino acid repeat [25].

To study the role of the various protein domains in the function of group III HKs different protein mutants were constructed. In Hik1p, a group III HK from Magnaporthe grisea, phosphate acceptance on both the conserved histidine and aspartic acid residues in the catalytic and the receiver domains respectively was essential for the susceptibility to phenylpyrroles and ambruticin VS4 [26, 27]. Deletions of single pairs of HAMP domains from the HK CaNik1p of C. albicans were associated with decreased susceptibility to fungicides, showing the relevance of these domains for fungicide activity [25] and deletion of four out of five amino acid repeats from the HK DhNik1p of Dabaryomyces hansenii generated a constitutively active HK, which was resistant to osmotic stress and fungicide treatment [23, 37].

As C. albicans is a human pathogen, understanding the relevance of the N-terminal nine HAMP domains and of the HisKA, HATPase_c and REC domains of CaNik1p for the action of antifungal compounds can guide development of new antimycotic strategies. To achieve this goal, point mutations were introduced in the HisKA, HATPase_c and REC domains of CaNIK1 which should render these domains non-functional. In addition, a deletion mutant for all HAMP domains was generated. Due to the ease of genetic manipulation of S. cerevisiae the plasmids harboring the mutated CaNIK1 were used to transform S. cerevisiae followed by testing viability, sensitivity to fungicides and phosphorylation of the MAPK Hog1p upon fungicidal treatment.

We previously showed that expression of the group III HK from the human fungal pathogen C. albicans, CaNIK1 in S. cerevisiae resulted in susceptibility of the transformants to the fungicides fludioxonil, iprodione and ambruticin VS3 [25]. Moreover, the fungicidal activity was decreased by deletion of single or double pairs of the N-terminal HAMP domains [25]. For other group III HKs it was already shown that mutations in the conserved phosphate-accepting residues and partial deletion of the HAMP domains conferred fungicide resistance [23, 26]. This stimulated our interest to investigate the involvement of the HisKA, HATPase_c and REC domains from CaNik1p in the fungicide activity, as they are conserved in all HKs.

To prevent the primary phosphorylation of the histidine residue and the subsequent His-Asp phosphate-transfer from the HisKA to the REC domains, respectively, the point mutations H510Q and D924N were introduced. The N627D mutation was supposed to inactivate the ATP binding site. The complete resistance of the strains H510 and D924 and the reduced susceptibility of the strain N627 in comparison to the strain NIK clearly showed that the functionalities of the above mentioned domains were essential for the susceptibility of the transformed yeast to the tested fungicides. In agreement, similar patterns of Hog1p phosphorylation were obtained after treating the different S. cerevisiae transformants with fludioxonil. Phosphorylation of Hog1p was totally abolished in the strains H510 and D924 and partially inhibited in the strain N627, while in all strains expressing genes with point mutations Hog1p was phosphorylated in response to osmotic stress, but was not phosphorylated without external stimuli. These results are in agreement with earlier reports of reduced antifungal susceptibilities of strains, which expressed other group III HKs carrying point mutations in the HisKA and REC domains [26, 27]. However, the correlation between the functionality of conserved HisKA, REC and HATPase_c domains of CaNik1p and both the fungicidal sensitivity and phosphorylation of Hog1p after fungicidal treatment was not shown before. Altogether, we present clear evidences that the histidine kinase functionality of CaNik1p was essential for the fungicidal effect and that this effect correlated with the activation of the MAPK Hog1p after treatment with fungicides.

The yeast histidine kinase Sln1p (group VI histidine kinase) is a negative regulator of the MAPK Hog1p, as its inhibition leads to activation of the MAPK. However, for group III HKs different effects were reported: Dic1p, the group III HK from Cochliobolus heterostrophus, was described as a positive regulator of Hog1p [24], whereas DhNik1p from Dabaryomyces hansenii was identified as a negative regulator [23]. To the best of our knowledge the relationship between the activity of CaNik1p and Hog1p has not yet been investigated, but can now be indirectly deduced from our data. It is generally accepted that activation of Hog1p in the absence of osmotic stress results in growth inhibitory effects [46]. Previously we reported that the antifungal effects of fludioxonil, iprodione and ambruticin VS3 are dependent on the Ssk1 - Pbs2 - Hog1p branch of the osmotic stress response pathway [25], so that a prerequisite for phosphorylation of Hog1p is the non-phosphorylated form of the response regulator Ssk1p [47]. It was even reported that the presence of phosphorylated Ssk1p prevented the activation of the MAP3K Ssk2p from unphosphorylated Ssk1p [48]. Ssk1p receives phosphate groups indirectly from HKs via the histidine transfer protein Ypd1p. Our results indicate that this phosphorylation is inhibited only in strains which are exposed to osmotic stress or which express the wild-type CaNIK1 variants and are treated with fungicides. In strains expressing mutated non-functional CaNIK1 phosphorylation of Ssk1 was not inhibited. This conclusion is in agreement with [23] who showed that fludioxonil treatment of S. cerevisiae expressing the group III DhNik1p decreased the phosphate transfer to a response regulator even in the presence of the endogenous, active HK Sln1.

Group III HKs are characterized by an amino acid repeat domain with five to six amino acid repeats, in each of which a single HAMP domain was identified previously, but which are now known to comprise concatenated pairs of HAMP domains [25, 32, 33]. The function of these domains is not yet clear, even though involvement in fungicide susceptibility and in osmosensing were suggested [19, 23, 25, 37]. Previous heterologous expression of truncated proteins, in which several HAMP domains were deleted from group III HKs, i.e. from CaNik1p [25] and DhNik1p from D. hansenii[37], was not reported to result in inhibition of growth of the respective S. cerevisiae transformants. Whereas in the previous reports only selected HAMP domains were deleted, here we deleted all HAMP domains from CaNik1p (CaNik1pΔHAMP) and observed that the synthesis of this truncated protein in the transformed S. cerevisiae strain was associated with severe growth inhibition. This phenotype could be reversed by additional point mutation in the histidine phosphorylation site of the HisKA domain (H510) or by the expression of CaNIK1ΔHAMP in single gene deletion mutants of the response regulator SSK1 or of one of the components of the Hog1 module namely the MAP2K PBS2 and the MAPK HOG1. This proved that the inhibition of growth of the transformant upon expression of CaNIK1ΔHAMP was dependent on the functionality of both the histidine kinase activity of CaNik1p and the functionality of the Ssk1 - Pbs2 - Hog1 branch of the HOG pathway. We could also unambiguously show that the inhibited growth of the strain ΔHa expressing CaNIK1ΔHAMP correlated with the constitutive phosphorylation of Hog1p without any external stimulation, which was also abolished after mutation of the conserved phosphate-accepting histidine residue (CaNIK1ΔHAMP(H510Q)). By analogy to the previous discussion this leads to the conclusion that the expression of CaNIK1ΔHAMP decreased the phosphate transfer activity to Ssk1, whereas the presence of the mutant CaNik1pΔHAMP(H510Q), which cannot be phosphorylated on the conserved histidine residue, did not affect the endogenous phosphorylation state of the Ssk1p. Thus, in summary, deletion of all HAMP domains had the same effect on the phosphate transfer activity to Ssk1p as treatment with fungicides. Additionally, the presence of mutated proteins, which are assumed not to possess histidine kinase activity and thus are not phosphorylated on either histidine in the HisKA domain or on aspartate in the REC domain, did not inhibit growth of the transformants and did not activate the Hog1p MAPK module. As a consequence of these results, it seems to be unlikely that in the transformed S. cerevisiae strains the histidine kinase activity of CaNik1p was inhibited by fungicide treatment, because inhibition of the kinase activity will lead to an enrichment of the non-phosphorylated form of the protein, similar to the protein variants carrying point mutations. The mutated proteins, however, did not influence growth whereas fungicide treatment did. Thus, our results support a model, in which the wild-type CaNik1p protein is not phosphorylated without external stimuli, and Ssk1p is kept in a phosphorylated form via indirect phosphate transfer from Sln1p. Upon deletion of all HAMP domains from CaNik1p or fungicide treatment CaNik1p is phosphorylated and this form prevents phosphate transfer to Ssk1p (Figure 6).

However, this model is based on the assumption that the phosphorylation state of the endogenous histidine kinase Sln1p is not changed by the presence of CaNik1p, since Sln1p is a transmembrane protein that undergoes autophosphorylation in the absence of osmotic stress and CaNik1p is a cystosolic protein. Thus, we expected that CaNik1p does not interfere with the autophosphorylation of the transmembrane protein Sln1p but with the phosphate transfer from Sln1p to Ypd1p. But it may also be hypothesized that when CaNik1p is phosphorylated, either following fungicide exposure or due to the absence of regulatory HAMP domains, Sln1p is no longer phosphorylated leading to the accumulation of non-phosphorylated forms of Ypd1p and Ssk1p and, hence, to the activation of Hog1p.

Thus, CaNik1p has to be considered as a positive regulator of Hog1p activity, similar to Dic1p from C. heterostrophus and contrary to DhNik1p and Sln1p. This is in agreement with the earlier results that CaNik1p cannot reverse the lethal phenotype of Sln1 deletion in S. cerevisiae whereas DhNik1p can. However, the mechanism leading to the reduced overall phosphate transfer activity to the response regulator remains to be investigated. As long as no protein structures are available from group III histidine kinases, one cannot exclude that point mutations and protein truncation have severe effects on the protein structures.

The constructed mutated versions of CaNIK1 could not be re-integrated in the available CaNIK1 homozygous deletion mutants of Candida albicans[8–18] as these mutants were constructed with the widely used URA blaster method and, thus, are prototrophic for uracil. Consequently direct transformation with the pYES2 vectors that harbor the mutated variants of the CaNIK1 was not possible as the vector contains URA3 as a selection marker. Therefore a new CaNIK1 homozygous deletion mutant has to be constructed using for example the SAT1 flipper cassette that makes use of nourseothricin as an antibiotic selection marker. This will allow reintegration of the CaNIK1-mutated variants from this study in such mutant.

Our results show that functional HisKA, HATPase_c and REC domains of CaNik1p are essential for the antifungal activity of the selected agents activating the HOG pathway. Moreover, the expression of CaNIK1ΔHAMP in transformed S. cerevisiae was associated with growth inhibition via constitutive phosphorylation of the MAPK Hog1p. In S. cerevisiae transformed with CaNIK1, growth inhibition resulting from treatment with the selected antifungals or from deletion of all HAMP domains from the protein required both a functional histidine kinase CaNik1p and an intact HOG pathway.