Candidalysin delivery to the invasion pocket is critical for host epithelial damage induced by Candida albicans

The human pathogenic fungus Candida albicans is a frequent cause of mucosal infections. Although the ability to transition from the yeast to the hypha morphology is essential for virulence, hypha formation and host cell invasion per se are not sufficient for the induction of epithelial damage. Rather, the hypha‐associated peptide toxin, candidalysin, a product of the Ece1 polyprotein, is the critical damaging factor. While synthetic, exogenously added candidalysin is sufficient to damage epithelial cells, the level of damage does not reach the same level as invading C. albicans hyphae. Therefore, we hypothesized that a combination of fungal attributes is required to deliver candidalysin to the invasion pocket to enable the full damaging potential of C. albicans during infection. Utilising a panel of C. albicans mutants with known virulence defects, we demonstrate that the full damage potential of C. albicans requires the coordinated delivery of candidalysin to the invasion pocket. This process requires appropriate epithelial adhesion, hyphal extension and invasion, high levels of ECE1 transcription, proper Ece1 processing and secretion of candidalysin. To confirm candidalysin delivery, we generated camelid VHHs (nanobodies) specific for candidalysin and demonstrate localization and accumulation of the toxin only in C. albicans‐induced invasion pockets. In summary, a defined combination of virulence attributes and cellular processes is critical for delivering candidalysin to the invasion pocket to enable the full damage potential of C. albicans during mucosal infection.


| INTRODUCTION
Candida albicans is a commensal fungus in the majority of the human population, but also frequently causes mucosal infections and, in severe cases, life-threatening systemic infections (Brown et al., 2012). The ability of C. albicans to cause such diseases is mediated by a wide range of virulence attributes. This includes the morphological transition between the yeast and the filamentous form, and the production of the peptide toxin candidalysin (Mayer, Wilson, & Hube, 2013;Moyes et al., 2016;Wilson, Naglik, & Hube, 2016).
Although yeast cells are required for dissemination via the bloodstream during systemic infections, the hyphal form is the more invasive morphology (Jacobsen et al., 2012;Kornitzer, 2019). In fact, hypha formation is accompanied by the deployment of several additional virulence attributes, including adhesins, invasins, metal acquisition factors, hydrolytic and detoxifying enzymes and thigmotropism (Brunke, Mogavero, Kasper, & Hube, 2016;Mayer et al., 2013;Wachtler, Wilson, Haedicke, Dalle, & Hube, 2011). Of the multiple genes identified as essential for mediating the full virulence potential of C. albicans, many are required for or associated with filamentation, demonstrating the critical role of hypha formation in C. albicans pathogenesis (Mayer et al., 2013).
During mucosal infection, the yeast-to-hypha transition is initiated by contact to epithelial cells (Dalle et al., 2010;Phan, Belanger, & Filler, 2000;Sudbery, 2011;Wachtler et al., 2011;Zakikhany et al., 2008). The initial step of filamentation, described as germ tube formation, is associated with increased adhesion to epithelia, mediated by hypha-associated adhesins such as Als3 (Hoyer, Payne, Bell, Myers, & Scherer, 1998). Extending hyphae grow along host cell surfaces and change their directional growth by contact sensing, also termed thigmotropism (Brand et al., 2007;Gow, 1997). This feature is required for the invasive properties of hyphae, since mutants lacking proteins associated with thigmotropism, such as Bud2, show reduced invasive growth (Brand et al., 2008). Invasion of epithelial cells occurs via "induced endocytosis" and "active penetration" depending on the epithelial cell type and both routes require hypha formation (Mayer et al., 2013;Zakikhany et al., 2007). Induced endocytosis is a hostdriven process triggered by hypha-associated invasins, in particular Als3, which acts as both an adhesin and invasin (Phan et al., 2000;Wachtler et al., 2011). Active penetration is a fungal-driven process mediated by the physical forces and properties of elongating hyphae (Dalle et al., 2010;Zakikhany et al., 2007). Both induced endocytosis and active penetration cause invagination of epithelial cells and the formation of an "invasion pocket" in which the hypha is tightly surrounded by the host membrane Zakikhany et al., 2007). Since the majority of proteins secreted by hyphae are transported to the hyphal tip via a complex secretion machinery (the Spitzenkörper) (Weiner et al., 2019), proteins or peptides secreted by hyphae in this scenario are expected to be secreted into the invasion pocket.
Mutants incapable of producing hyphae, such as a mutant lacking two key transcriptional regulators of the yeast-to-hypha transition-Cph1 and Efg1-have poor adherence kinetics and are non-invasive (Lo et al., 1997;Ramage, VandeWalle, Lopez-Ribot, & Wickes, 2002;Wachtler et al., 2011). Mutants that are unable to extend their growth at the hyphal tip, like a mutant lacking the protein Eed1 (Martin et al., 2011;Zakikhany et al., 2007), are only poorly able to invade host tissues. The eed1ΔΔ mutant produces short hyphae, expresses the full set of hypha-associated genes (HAGs) at early time points, and invades superficially, but then fails to maintain hyphal elongation (Martin et al., 2011;Polke et al., 2017;Zakikhany et al., 2007). Instead, eed1ΔΔ cells switch back to yeast growth and proliferate as yeast cells, a portion of them remaining inside the invaded cell (Zakikhany et al., 2007). Another C. albicans mutant lacking Hgc1, a G1 cyclinrelated protein and regulator of morphogenesis, forms very short germ tubes but is unable to maintain filamentation (Zheng, Wang, & Wang, 2004). Although hgc1ΔΔ still expresses HAGs (Zheng et al., 2004), it has a reduced potential to invade epithelial cells (Wachtler et al., 2011). A systematic study of 26 C. albicans mutants showed that all mutants with reduced potential to adhere to and/or invade into epithelial cells, including mutants with dysfunctional contact sensing, are also unable to fully induce epithelial damage (Wachtler et al., 2011).
While hypha formation and maintenance are critical for invasion, the key virulence attribute responsible for inducing epithelial damage is candidalysin (Moyes et al., 2016). This peptide toxin, a product of the larger Ece1 (extent of cell elongation, [Birse, Irwin, Fonzi, & Sypherd, 1993]) polyprotein, is exclusively secreted by C. albicans hyphae, and is responsible for pore-induced damage of different epithelial cell types, including oral, vaginal, and intestinal cells (Allert et al., 2018;Moyes et al., 2016;Richardson et al., 2017Richardson et al., , 2018. Processing of Ece1 requires the sequential activities of the Golgi-located proteases Kex2 and Kex1 (Moyes et al., 2016;Richardson et al., 2018). C. albicans mutants lacking Ece1 or the candidalysin-coding sequence only, are unable to damage epithelial cells, despite normal hypha formation. Incomplete processing of candidalysin also prevents epithelial damage . Importantly, these studies showed that it is candidalysin's activity that damages host cells rather than C. albicans hypha formation and host cell invasion per se, which finally decoupled epithelial damage from hypha formation . However, while synthetic, exogenously added candidalysin is sufficient to lyse epithelial cells (Moyes et al., 2016), the degree of damage does not reach the same level as C. albicans hyphae invading epithelial cells, unless extremely high concentrations of the peptide are applied (Allert et al., 2018;Moyes et al., 2016). Therefore, we hypothesized that a combination of fungal attributes and cellular processes is required for the full damage potential of C. albicans during epithelial infection.
In this study, we utilised a selected panel of C. albicans mutants with known virulence defects to demonstrate that the full damage observed with wild-type C. albicans cells requires a defined combination of virulence attributes to deliver candidalysin to the invasion pocket during mucosal infections. This process requires appropriate adhesion, hyphal extension and invasion, high levels of ECE1 transcription, proper Ece1 processing and, finally, secretion of candidalysin. Candidalysin delivery to the invasion pocket was confirmed using camelid V H Hs (nanobodies) specific for candidalysin.

| Diverse C. albicans mutants show reduced oral epithelial damage
To dissect which virulence attributes of C. albicans are required for full epithelial damage (relative to the wild-type infection) in combination with the production of mature candidalysin, we analysed a panel of mutants with known defects in hypha formation (cph1ΔΔ/efg1ΔΔ; hgc1ΔΔ), hypha extension (eed1ΔΔ), thigmotropism (bud2ΔΔ), adhesion (als3ΔΔ), Ece1 processing (kex2ΔΔ, kex1ΔΔ) and lacking Ece1 (ece1ΔΔ) ( Table 1). All mutants grew normally in their yeast morphology and thus have no general growth defect. However, some mutants show multiple other defects, for example, the kex2ΔΔ mutant is not only unable to process the Ece1 polypeptide  but also multiple other pro-proteins with Lys-Arg processing sites (e.g., secreted aspartic proteases [Newport & Agabian, 1997;Naglik, Challacombe, & Hube, 2003]) and is known to have defects in hypha formation (Newport & Agabian, 1997, Newport et al., 2003. Damage caused to TR146 oral epithelial cells (OECs) is summarised in Figure 1a. All mutants showed either an intermediate (hgc1ΔΔ, bud2ΔΔ) or a (very) strongly reduced damage potential (cph1ΔΔ/ efg1ΔΔ, eed1ΔΔ, als3ΔΔ, kex1ΔΔ, kex2ΔΔ and ece1ΔΔ), indicating that hypha formation and extension, thigmotropism, adhesion, Ece1 production and processing are all critical to enable full epithelial damage by C. albicans.
2.2 | High ECE1 transcription rates are necessary but not sufficient to predict damage Hypha formation is strongly linked to the transcription of ECE1 (Martin et al., 2013). Therefore, the inability of yeast-locked mutants (such as cph1ΔΔ/efg1ΔΔ) to cause full epithelial damage may be solely linked to the silenced transcription of ECE1. We thus quantified ECE1 transcription (in the absence of epithelial cells but in strong hyphae-inducing conditions) in the selected mutants ( Figure 1b). Almost all mutants with strongly reduced damage potential also showed no or severely/highly impaired ECE1 transcription (cph1ΔΔ/efg1ΔΔ, eed1ΔΔ, kex2ΔΔ and ece1ΔΔ). Thus, strongly reduced ECE1 transcription levels strongly correlate with reduced damage potential.
Notably, other mutants with a strong/intermediate reduction in their damage capacity (hgc1ΔΔ, bud2ΔΔ, als3ΔΔ and kex1ΔΔ) had ECE1 transcript levels similar to the isogenic wild type, or only mildly reduced.
2.3 | Reduced adhesion, invasion and hypha length are confirmed phenotypes predictive for reduced damage potential We next investigated which virulence parameters were required to facilitate candidalysin-induced epithelial damage by quantifying the ability of each mutant to adhere, invade and produce hyphae ( Figure 2). Only the two mutants lacking the transcription factors Cph1/Efg1 or the adhesin Als3 showed adhesion defects (Figure 2a).
The same mutants also showed significantly reduced invasion potential, in line with (Wachtler et al., 2011), where adhesion was shown to be a prerequisite for invasion. All other mutants invaded at similar rates, or only mildly reduced compared with the wild type (Figure 2b), confirming that invasion per se does not necessarily cause epithelial damage (Moyes et al., 2016). We thus hypothesized that hypha extension, hyphal length and delivery of candidalysin are all essential attributes required for invasive hyphae to cause damage. As expected, a significant reduction in the ability to produce full length hyphae (relative to wild-type cells), at 3 hr post infection (p.i.), was observed for mutants lacking Cph1/Efg1, Hgc1, or Kex2 ( Figure 2c). The mutant lacking Eed1 showed a mild reduction in hyphal length. In fact, this mutant is known to switch to yeast growth after approximately 3 hr (Zakikhany et al., 2007).
These data suggest that the ability to produce full length hyphae in combination with high ECE1 transcription is critical for the full potential to damage epithelial cells. This explains why the hgc1ΔΔ mutant can cause intermediate epithelial damage despite not producing full length hyphae, as it also transcribes ECE1 at levels similar to the isogenic wild type.

| Constitutive transcription of ECE1 in a yeastlocked mutant does not rescue its damaging phenotype
Given that hgc1ΔΔ caused an intermediate level of epithelial damage despite expressing ECE1 at high levels, we next determined whether filamentation and ECE1 transcription are linked to enable epithelial cell damage. To achieve this, we generated a yeast-locked strain that constitutively transcribes ECE1: cph1ΔΔ/efg1ΔΔ + pENO1-ECE1 (Westman et al., 2018). Hypha-producing control strains were transformed with the same construct, obtaining BWP17/CIp30 + pENO1-ECE1 (contains 1 copy of ECE1 driven by its original promoter, and 1 copy driven by the enolase promoter) and ece1Δ + pENO1-ECE1 (contains only 1 copy of ECE1 driven by the enolase promoter). As shown in Figure 1a, both BWP17/CIp30 + pENO1-ECE1 and ece1Δ + pENO1-ECE1 were able to damage at levels similar to the isogenic wild type. However, the cph1ΔΔ/efg1ΔΔ + pENO1-ECE1 mutant was unable to cause damage. RT-qPCR analysis showed that the enolase promoter was able to induce transcription of ECE1 mRNA to levels only mildly reduced compared to the isogenic wild type (Figure 1b) in both cph1ΔΔ/efg1ΔΔ + pENO1-ECE1 and ece1Δ + pENO1-ECE1.
Since the forced expression of ECE1 in the yeast-locked strain cph1ΔΔ/efg1ΔΔ was insufficient to enable this mutant to induce damage but was sufficient to enable damage in the hypha-producing strain ece1Δ + pENO1-ECE1, we conclude that filamentation in combination with ECE1 expression is required for inducing damage.

| Secretion of candidalysin does not always predict damage
Next, we considered the importance of candidalysin secretion, as producing high levels of ECE1 transcripts does not necessarily mean that the transcripts are efficiently translated, or that the translated toxin is properly secreted. To test this, mutant strains were cultured in strong hypha-inducing conditions and their supernatants analysed by LC-MS/MS. Detailed results are available in Table S1, and a summary of the results is shown in Table 2. Candidalysin secretion was measured as "Peptide Spectrum Matches" (PSMs), which allows semi-quantitative analysis of peptide abundances. Candidalysin was not secreted by cph1ΔΔ/efg1ΔΔ, kex2ΔΔ or ece1ΔΔ and was negligible/ low in eed1ΔΔ, bud2ΔΔ and kex1ΔΔ. Of these, only bud2ΔΔ and kex1ΔΔ expressed moderate levels of ECE1 transcripts ( Figure 1b).
For kex1ΔΔ, it is known that the toxin's precursor is produced, and the low levels of candidalysin in the supernatant are due to the fact that the precursor is not processed into the final toxin . Mutant bud2ΔΔ probably either poorly translates ECE1 transcripts or is impaired in efficient Ece1 processing or candidalysin secretion. The reduced secretion of candidalysin explains why bud2ΔΔ induces moderate levels of damage.

| DISCUSSION
To assess the importance of thigmotropism in candidalysin-induced epithelial damage we used bud2ΔΔ, which is required for directional growth of hyphae (Brand et al., 2008). Notably, we found that reduced candidalysin secretion may derive from bud2ΔΔ also having suspected defects in the secretory pathway (Brand et al., 2008) and may, as a consequence, not efficiently transport candidalysin to the hyphal tip for delivery/secretion. Therefore, the intermediate damage induced by bud2ΔΔ may be due to impaired candidalysin secretion.
The data demonstrate that hyphae need to deliver candidalysin into the invasion pocket to induce epithelial damage. However, candidalysin delivery and secretion first requires the parent protein Ece1 to be sequentially processed by the Golgi-located serine proteases Kex2 and Kex1 (Moyes et al., 2016;Richardson et al., 2018). Kex2 is critical for processing several C. albicans proteins in addition to Ece1 (Bader, Krauke, & Hube, 2008) and, therefore, kex2ΔΔ has a general fitness defect that manifests in lack of hypha formation (Newport & Agabian, 1997), ECE1 expression and candidalysin secretion, thus explaining its defective damage phenotype. After Kex2 generates peptides terminating in Lys-Arg amino acid residues, Kex1 removes the terminal Arg to form fully matured candidalysin terminating in Lys. The mature candidalysin is then thought to be packaged and transported to the hyphal tip for secretion. We found that, while  (Tables 2 and S1) and by the α-candidalysin nanobody, which is as damaging as candidalysin (Moyes et al., 2016). Given that Kex1 is probably also involved in the processing of numerous other hitherto unknown and unrelated proteins (with possible roles in virulence), kex1ΔΔ likely has other defects in packing fungal peptides to the secretory pathway and may not efficiently transport the candidalysin precursor to the hyphal tip for secretion and delivery. These data indicate that correct Ece1 processing and packaging/transport of candidalysin to the hyphal tip for secretion is essential for induction of epithelial damage.
To confirm our supposition that candidalysin needs to be secreted into the invasion pocket to induce damage, we generated novel anticandidalysin camelid V H Hs (nanobodies) to detect candidalysin in situ. is considered a card that contributes to the overall virulence phenotype" of a pathogenic microbe (Casadevall, 2006). Even a strong card like candidalysin alone is not sufficient for full virulence and to win the gamefrom the perspective of a pathogen.

| Construction of pENO1-ECE1 mutants
Similar to (Westman et al., 2018), pENO1-ECE1 strains transcribing ECE1 under the control of the C. albicans enolase promoter (pENO1) were generated from the parent strains BWP17/CIp30 (contains 2 copies of ECE1 at their original positions) and ece1Δ (heterozygous ece1 knockout strain, contains only 1 copy of ECE1 at its original position). A cassette containing the Candida-adapted NAT1 (CaNAT1) marker and pENO1 was amplified from pNAT-ENO1 using primers ECE1_ENO1_PF (containing 80 nts of ECE1 gene-specific sequence starting from the start codon) and ECE1_ENO1_PR (containing 80 nts homologous to the ECE1 5 0 promoter region, starting 692 nts before the start codon, Table 4). The PCR product was then integrated before the ECE1 gene. Integration was confirmed with primers ENO1p-S (ENO1-specific) and ECE1-IR (internal to the ECE1 gene,

| Mammalian cell culture
The human TR146 OEC line was chosen to evaluate the damage potential of the selected C. albicans mutants, since it is a versatile cell line often used for damage screening purposes (Moyes et al., 2016). TR146 OECs were purchased from the European Collection of Authenticated Cell Cultures (Rupniak et al., 1985) and used throughout the study. Cells were cultured at 37 C and 5% CO 2 in Dulbecco's modified Eagle medium/Nutrient Mixure F-12 (Life Technologies) supplemented with 10% (vol/vol) heatinactivated (10 min at 56 C) fetal bovine serum (Life Technologies).

| Anti-candidalysin llama V H Hs (nanobodies)
Anti-Candidalysin Llama V H Hs (i.e., variable domains of the heavy chain of the heavy-chain antibody, also termed nanobodies) were produced by QVQ B.V. (Utrecht, The Netherlands) using phage-display technology (Kuhn et al., 2016;Verheesen & Laeremans, 2012 4.9 | Fungal RNA extraction C. albicans strains were adjusted to 10 7 cells/ml in 25 ml RPMI 1640 (hypha-inducing) or 5 ml YPD (yeast-maintaining). For hyphae samples, fungal suspensions were distributed in 150 cm 2 petri dishes and incubated at 37 C and 5% CO 2 for 3 hr. After incubation, medium and nonadherent Candida cells were discarded. Adherent Candida cells were rinsed once with ice-cold PBS, loosened with a cell scraper and collected.
For yeast samples, fungal suspensions were cultured at 30 C for 3 hr in a shaking incubator (180 rpm). After incubation, cells were collected by centrifugation (4,000g for 2 min at 4 C) and resuspended in 10 ml icecold PBS. Hyphae and yeast samples were washed again with 1 ml icecold PBS and centrifuged (20,000g, 2 min, 4 C), the supernatant was discarded, and cell pellets were frozen in liquid nitrogen. Frozen Candida pellets were thawed in 600 μl RLT buffer (Qiagen) containing 1% β-mercaptoethanol, mixed with 300 μl acid-washed glass beads (0.5 mm Ø), and run through a Precellys homogeniser (Bertin instruments) twice at 5,500 beats/min for 15 s, with 20 s pause in between. Lysates were centrifuged for 2 min at 20,000g, 4 C, the supernatant was mixed with an equal volume of 70% ethanol (prepared in diethyl pyrocarbonate [DEPC]-treated water) and total RNA was isolated using the RNeasy minikit (Qiagen) according to the manufacturer's instructions. RNA integrity and concentration were confirmed using a Bioanalyzer (Agilent).

| ECE1 transcription analysis
RNA (500 ng) was treated with DNase (Epicentre), confirmed DNA free, and cDNA was synthesised using Superscript III reverse transcriptase (Invitrogen) according to the manufacturer's protocol. cDNA samples were used for qPCR with EvaGreen mix (Bio&Sell). Primers (ACT1-F and ACT1-R for ACT1 and ECE1-F and ECE1-R for ECE1 [ Table 4]) were used at a final concentration of 500 nM. qPCR amplifications were performed using a CFX96 thermocycler (Bio-Rad). ECE1 transcription was calculated using the threshold cycle (ΔΔC t ) method, with ACT1 as the reference gene and C. albicans isogenic wild-type strain (yeast morphology) as the control sample. Data are presented as % relative to the BWP17/CIp30 isogenic wild-type strain.

| Candidalysin detection in hyphal supernatants by LC-MS/MS analysis
Analysis of hypha-secreted Ece1 peptides was optimised for the detection of candidalysin and performed as previously described (Moyes et al., 2016). The method is aimed at the detection of peptides with highly hydrophobic moieties such as candidalysin. Briefly,