HPLC-MS identification and expression of Candida drug-resistance proteins from African HIV-infected patients

The objective of this study was to elucidate the proteomic mechanisms of drug resistance in HIV-infected African patients. Cell membrane fractions from forty oral Candida isolates isolated from African HIV-positive patients were analysed using HPLC-MS with the aim of identifying proteins associated with their pathogenicity and drug resistance. Heat shock proteins that mediate the fungicidal activity of salivary peptides were found in all tested Candida fractions, with pH-responsive proteins associated with increased pathogenicity only being present in the three most commonly isolated species. ABC multidrug transporter efflux pumps and estrogen binding proteins were only found in C. albicans fractions, while ergosterol biosynthesis proteins were identified in four species. The combination of various adherence, invasion, upregulation and efflux pump mechanisms appear to be instrumental for the Candida host colonization and drug resistance emergence in HIV-infected individuals.


Isolation of Candida cell membrane proteins
Drug resistance-related protein profiles expressed by different Candida species regarded as susceptible or resistant to azoles (as determined by CLSI guidelines) were compared, along with the proteins expressed using other antifungal drug classes.
Isolates were cultured on Sabouraud agar plates for 24 hours at 37 ℃, before individual colonies were picked and incubated in yeast extract peptone dextrose (YPD) broth (peptone 10 g/L and dextrose 40 g/L distilled water) at 37 ℃ for 16 hours with agitation. Culture density was measured at 600 nm absorbance and the biomass was recovered by centrifugation at 3000 g. The fungal cell pellet was washed in 2 mL of sterile distilled water before being re-pelleted at 3000 g for 10 minutes. The isolated pellet was then resuspended in 2 mL homogenizing buffer with the protease inhibitor phenylmethylsulfonyl fluoride (PMSF) (50 mM Tris-HCl, pH 7.5, 2 mM EDTA, 1 mM PMSF), using a protein isolation method based on the one described by Niimi et al [10]. One-millimetre borosilicate glass beads (Cat. no. Z273619, Sigma-Aldrich, USA) were used to disrupt the fungal cells, by placing the tubes in a vortex for 6 minutes. The cell debris were pelleted at 5000 g for 10 min at 4 ℃ and the lysates (containing fungal cell membrane components) were then pelleted at 20 000 g for 1 hour at 4 ℃, to result in a crude membrane fraction.
Protein concentration was determined by the Bradford method [28], using a Bio-Rad Bovine Serum Albumin (BSA) Standards Set (Cat. no. 500-0207, Bio-Rad, USA), according to the manufacturer's instructions. Sixteen microliters of a 10 mM Tris-HCl, pH 7.0, 5 mM EDTA solution were added to the isolated pellets, with standardized sample concentrations being subsequently diluted to an approximate protein concentration of 0.65 mg/mL.
Using filter-aided sample preparation (FASP) on the cell fractions, 50 μL samples were mixed 1:1 with SDT lysis buffer (4% SDS, 100 mM Tris-HCl pH 7.6, 0.1 M DTT), then mixed with an equal volume of UA buffer (8 M urea, 100 mM Tris-HCl, pH 8.5) and concentrated on an Amicon ultra 10 kDa MWCO filter (EMD Millipore, USA) by 40 minutes centrifugation at 14000 g. For simplicity, all subsequent centrifugation steps were performed at 14000 g. Next, 200 μL UA buffer was added and the samples were again centrifuged for 40 minutes before isolated proteins were alkylated with the addition of 100 μL of 0.05 M iodoacetamide in UA buffer. After a 5-minute incubation and 30minute centrifugation, 100 μL of UB buffer (8 M urea, 0.1 M Tris-HCl pH 8.0) was added. This was followed by another hour centrifugation, the addition of 100 μL 50 mM ammonium bicarbonate solution before a 1-hour centrifugation cycle. Next, 40 μL trypsin was added and the filter was incubated at 37 ℃ for 17 hours in a wet chamber. The filter was subsequently transferred to a new tube and centrifuged for 40 minutes, followed by the addition of 40 μL of a 0.5 M sodium chloride solution and another 20-minute centrifugation cycle. Finally, the solution was acidified by the addition of 2.4 μL formic acid solution. The filtrate was then desalted using C18 StageTips (Thermo-Fisher Scientific, USA) according to the manufacturer's instructions, before drying in vacuo and -20 ℃ storage ( Figure 1). Dried peptides were dissolved in 5% acetonitrile in 0.1% formic acid and 10 μL injections were made for nano-LC chromatography.
The mass spectrometer was operated in data-dependent mode, to automatically switch between Orbitrap-MS and LTQ-MS/MS acquisition and data acquired using the Xcalibur software package (Thermo Scientific Cat. No. OPTON-30487). The 20 most intense ions were isolated and fragmented in a linear ion trap (number of accumulated ions 1.5 × 10 4 ) using collision induced dissociation. The lock mass option (polydimethylcyclosiloxane; m/z 445.120025) enabled accurate mass measurement in both the MS and MS/MS modes.
Thermo Proteome Discoverer 1.3 software (Thermo Scientific, Bremen, Germany) was used to identify proteins via automated database searching (Mascot, Matrix Science, London, UK, and Sequest) of all tandem mass spectra against the Uniprot database [29], with the aim of exclusively identifying specific protein functions related to antifungal drug resistance and proteins that affected the organism's pathogenicity and virulence. Proteins were considered positively identified when they were characterised with at least 1 tryptic peptide per protein, a Mascot score threshold of 20 and Sequest score threshold of 1.5. In order to include only significant data, only proteins with a mascot score above 24 were included in the study. Membrane proteins are further denatured with the high urea content of the UA and UB buffers that further destabilise internal protein bonds. Iodoacetamide is a sulfhydrylreactive alkylating reagent that blocks binding to the reduced cysteine residues and prevents disulphide bonding and protein refolding while the addition of bicarbonate renders the proteins in a highly-charged state, improving Amicon filter bonding. The addition of Trypsin and the overnight incubation at 37 ℃ in step , further denatures the membrane proteins that are eluted from the Amicon filter by the interaction between the positively-charged proteins and negatively-charged chloride ions in the sodium chloride solution used in step . Formic Acid further cleaves the membrane proteins into peptides at their C-or N-terminal domains (step ) before the eluted proteins then bind to a derivatised C18 filter in the StageTip (step ) through hydrophobic interactions. This step allows for other salts, buffers and chaotropes to be eluted. The tip with the bound proteins is then dried under vacuum pressure and stored at -20 ℃ until needed for HPLC-MS. This figure was created by an author (RF) using BioRender (https://biorender.com/).

Cohort demographics and combination ART
The majority of participants were female (80%) with a median age of 33.5 years (age range 24 to 70). Recruits were on the following ART at the time of sample collection: 82.5% receiving Lamivudine, 52% on Nevirapine, 42.5% on Zidovudine, 37.5% on Stavudine, 27.5% receiving Efavirenz, one person (2.5%) received Tenofovir and one person received Lopinavir/ritonavir as part of their 1st line cART regimen. Of the 40 patients, seven (17.5%) were confirmed HIV-positive but planned to begin cART, pending viral load testing results.

Expression of azole resistance-related proteins
Various colonization and resistance mechanisms detected in azole-resistant isolates were elucidated using HPLC-MS. Representatives of heat-shock proteins were detected in all isolates. The identified heat shock proteins included Hsp70 (SSA1) and Hsp90 (expressed by all species). Hsp70 (SSA1 and SSA2) and Hsp90 were found in resistant isolates across all seven Candida species in our study, with Hsp70 (SSA1) expressed in seven of the nine C. glabrata strains tested. Of these seven, six Hsp70 (SSA1) positive isolates were also resistant to at least one azole drug tested. Only a single C. glabrata isolate was susceptible to all tested drug classes and did not express Hsp70, highlighting the possible pluripotent nature of this protein. In the C. glabrata isolate that displayed both fluconazole and voriconazole resistance, Hsp70 (SSA1) (UniProtKB accession number P10591) was expressed with the highest mascot score (1835.96) and had a high MS sequence coverage of 42.37%. Hsp70 (SSA1) was also identified in the membrane fractions of C. glabrata isolates that were susceptible to all azoles tested but were resistant to amphotericin B (one isolate) or had intermediate resistance to micafungin (one isolate). Heat-shock protein Hsp70 (SSA2) was expressed by resistant representatives of all species with the exception of C. dubliniensis. Other resistance-related proteins included NADPH dehydrogenase oxidoreductase EBP1 (expressed by C. albicans) and PHR2 (expressed by C. albicans and C. dubliniensis resistant isolates). The expression of a multidrug resistance transporter protein Cdr1p was seen in two C. albicans azole-resistant fractions, with multidrug efflux transporter Cdr2p being present in a C. albicans isolate resistant to all four azole drugs tested. Transporter protein Cdr4p was also identified in eight C. albicans resistant fractions (Table 1). Proteins associated with ergosterol biosynthesis were found in C. albicans, C. dubliniensis, C. glabrata and C. tropicalis cell fractions. Although Erg11p was the most predominant protein in this group, other proteins that are involved in ergosterol biosynthesis were also identified. They included Erg9p (which synthesises squalene from farnesyl diphosphate), Erg1p (which synthesises squalene epoxide from squalene), Erg6p (which synthesises fecosterol from 4,4 dimethylzymosterol) and Erg3p (which synthesises ergosta-5,7,24(28)-trienol from episterol). A C. albicans isolate highly resistant to all tested azole drugs had the highest detected mascot score for Erg11p (188.9), while also expressing Erg6p. Some proteins appeared to be more prevalent in certain species, for example, efflux pumps, corticosteroid and estrogen binding proteins were only present in C. albicans isolates, while other protein types were more widely distributed (Table 1).
While most isolates initially selected for their susceptibility to azoles were also susceptible to other drug classes, one C. glabrata isolate and the sole C. rugosa representative expressed resistance to amphotericin B. In the latter, only salivary histatin binding heat-shock proteins Hsp70 (SSA1/2) and Hsp90 were identified along with PHR2.

Expression of cross-resistance-related proteins
Isolates demonstrating cross-resistance to both azoles and other drug classes expressed PHR2, a protein family associated with increased pathogenicity. PHR2 was seen in multi-drug resistant C. dubliniensis and in three C. albicans isolates where one expressed cross-resistance to amphotericin B and two expressed cross-resistance to 5-flucytosine. Interestingly, one of the C. albicans azoleresistant isolates resistant to 5-flucytosine, expressed 9 resistance-associated proteins including two multi-drug transporters, three distinct heat-shock proteins, NADPH dehydrogenase (EBP1), PHR2 and two ergosterol biosynthesis proteins. Incidentally, this was the isolate expressing the highest mascot score for Erg11p. Another C. albicans isolate with resistance to both azoles and 5-flucytosine expressed a similar protein profile, with the exception of Hsp70 (SSA2) and NADPH dehydrogenase.

Comparison of resistance protein profiles of patients on cART and those not yet on cART
Of the patients who were not yet on cART at the time of sample collection, five were colonised by C. albicans (three of which were highly resistant to fluconazole and two of these also being resistant to 5-flucytosine), one was colonised by a multi-drug resistant (MDR) C. dubliniensis isolate and one was colonised by C. glabrata. When comparing the protein profiles between patients who were yet to start cART and those already on cART, it was noted that apart from the presence of the ubiquitous heat shock proteins, the species isolated from patients not on cART also expressed pHresponsive protein 2 (seen in the MDR C. dubliniensis and four C. albicans isolates), while one of the Cdr1p and two Cdr4p efflux transporters were identified in two C. albicans isolates from this group. Ergosterol biosynthesis proteins were identified in a C. albicans isolate and in the MDR C. dubliniensis isolate (Erg9p, Erg11p and Erg3p) belonging to this group.

Discussion
Our previous finding of the high resistance of Candida to various antifungal drugs seen in these HIV patients [27] is supported by this attempt to elucidate the different mechanisms that increase the pathogenicity and drug resistance of these organisms. This study focused on 40 novel Candida isolates whose membrane protein composition were characterised to better understand their role in drug resistance.
Candida species can express estrogen-and progesterone-binding proteins, explaining the higher predisposition of females to candidiasis [30,31] and in the current study, estrogen-binding proteins were identified in C. albicans cell fractions in the form of EBP1.
The heat-shock proteins Hsp70 (SSA1 and SSA2) and Hsp90 are known to bind to HTN3/histatin-5 found in saliva, affecting the fungicidal activity of these native antimicrobial proteins [14,15,32] and leading to resistance [33] and increased oral Candida colonization [34], especially in immunocompromised patients [35]. Histatins are small (7-38 amino acids) histidinerich, broad spectrum, cationic immunological peptides that act on the cell membrane causing pore formation and fungal cell lysis. Demonstrating potent antifungal activity, histatins are known to interact with Hsp70 (SSA1/2) and kill Candida [32]. Hsp70 (SSA1) was previously reported to be upregulated in fluconazole-resistant membrane fractions of C. glabrata [36], while in contrast, it was found by the same group to be downregulated in voriconazole-resistant isolates [37]. The presence of Hsp70 (SSA1) in most tested isolates, the ability of this protein to affect the fungicidal activity of the host's antimicrobial peptides [16] and its association with oropharyngeal disease [17] are factors that can directly influence the treatment outcomes of HIV patients.The ubiquitous Hsp90, on the other hand, has a role in supporting cell survival by stabilizing enzymes during stress [13], but has also been implicated in the rapid acquisition [38] and increased fluconazole resistance in Candida biofilms [15].
Multidrug resistance proteins were detected in most C. albicans cell membrane fractions tested in this study. The appearance of multidrug resistance protein Cdr1p may lead to resistance to a range of compounds as well as to azole antifungal drugs [39]. Cdr1p plays a role in fluconazole drug resistance [40] and confers resistance to cycloheximide, chloramphenicol and miconazole [41]. The up-regulation of Cdr1p in the cell membrane of a C. glabrata fluconazole-resistant strain [42] and in azole-resistant C. parapsilosis [43] have previously been documented, but this protein was not detected in membrane fractions of azole resistant C. glabrata in the present study. The sole C. parapsilosis isolate tested in this study was a highly susceptible isolate, possibly explaining the lack of this drug resistance mechanism on this strain. Cdr2p (identified in a C. albicans isolate resistant to all four azole drugs tested) has the ability to not only confer resistance to azoles but also to other antifungals such as terbinafine and amorolfine in addition to metabolic inhibitors [44]. The ability of histatin-5 to inhibit the development of Cdr1p and Cdr2p-mediated multidrug resistance [45], demonstrates how various cellular mechanisms may act together to modulate fungal pathogenicity. The identification of Cdr4p in eight drug-resistant C. albicans cell membrane fractions deserves attention since Cdr4p is an ABC transporter protein involved in azole resistance in filamentous fungi [46].
Five of the 25 proteins involved in ergosterol biosynthesis were found in this study. Azole drugs target this biosynthesis pathway by inhibiting Erg11p, leading to a block in ergosterol synthesis and the accumulation of toxic sterol intermediates by Erg6p and Erg3p [22]. Candida is known to mutate the Erg11 gene [23,47] and demonstrate drug target overexpression of Erg11p [48] as part of its resistance mechanisms against azole drugs.
Coupled with filter-aided sample preparation, HPLC is a useful method for demonstrating antifungal susceptibility profiles and, with further developments in sample preparation, could become a rapid, reliable and practical technique for the early detection of resistance and subsequent prompt treatment of resistant Candida infections. HPLC analysis can be completed within hours, favouring its application over the traditional time-consuming processes of culture and disc diffusion or microdilution methods for drug susceptibility testing which may take days to complete. Comparatively, the cost, equipment specifications and skilled operators needed to perform FASP-HPLC analysis outweigh those associated with traditional drug resistance testing methods, not to mention the higher fungal biomass needs in relation to more sensitive molecular sequencing methods.
Although the introduction of cART has vastly improved the quality of life for the HIV-infected, it has minimally impacted the observed number of Candida infections [49,50]. The large number of immunocompromised HIV-positive patients in Africa receiving non-protease-inhibitor-based first line cART [51] and the prevalence of multidrug-resistant candidiasis, suggests a bleak future, barring any natural therapeutic intervention without adverse cART interactions [52]. A limitation of this study is that since only seven of the 40 isolates were collected before the commencement of cART, an association between the effect of cART on protein expression could not adequately be compared, thus creating an avenue which may be further explored.
Furthermore, the inability to significantly differentiate between individual protein expression profiles and their relation to specific antifungal classes in MDR isolates indicates that further optimization of the protein isolation methods should be prioritized in future studies. In spite of this methodological limitation, it was possible to identify the presence of protein profiles in organisms expressing specific drug resistance patterns within the different species studied. The combination of diverse drug resistance and survival mechanisms were notably more prevalent in C. albicans, C. dubliniensis and the inherently fluconazole-resistant C. glabrata. These isolates expressed proteins that enhance their fungicidal activity by increasing their adherence to salivary peptides and improving the binding of fungal proteins to mammalian hormones, modifications that are instrumental in the ability of these organisms to colonise immunocompromised patients and resist antifungal drugs.
The possibility exists that cell membrane changes could have occurred in response to environmental stress, such as pH changes during protein isolation, rather than to stress induced by drugs. Additionally, given that most UniProt strains are isolates from other geographical regions, and since there is a paucity of studies on isolates from Africa, it is possible that the isolates have sequence polymorphisms and different amino acid sequences when compared to other previously tested populations. In essence, the limited MS data for Candida isolates from sub-Saharan Africa identifies the lack of proteomics-based drug resistance monitoring in this study's sample populations.

Conclusions
Fluconazole-resistance is steeply rising, and it is the most frequently prescribed antifungal for HIV-associated opportunistic infections. Antifungals such as echinocandins, are usually considered for treating patients who show resistance due to previous azole exposure, or when C. glabrata, C. parapsilosis or C. krusei have been identified [53]. Although shown to be as effective in treating OC with less serious side effects, echinocandins have a greater relapse rate, contributing significantly to ever-increasing antifungal resistance prevalence [54,55]. Amphotericin B is used when patients show resistance to fluconazole and echinocandins. However, in very ill patients, it may prove toxic [56] and has limited availability in resource-poor countries due to its cost.
Reported drug interactions between systemic antifungals (particularly azoles) and antiretrovirals have led to the addition of special guidelines for their use with combination ART for HIV-infected persons by the US Department of Health and Human Services [57]. Moreover, the evolving drug resistance and subsequent circulation of multidrug-resistant or inherently drug-resistant, non-albicans species in the treatment-naï ve population not only limits treatment options but may also reduce the tolerance of combination ART through drug interactions.
Thus, there is an urgent need for the development of novel, non-toxic, broad spectrum, highly potent, natural antifungal therapies. HPLC provides beneficial protein abundance data that is essential for understanding the role of individual proteins in drug resistance mechanisms in vitro, thus establishing a role in studies contributing to the development of novel therapies.