Clearing the Fungal FoG: Perseverance, a property distinct from resistance, is associated with clinical persistence

Drug resistance, defined by the minimal inhibitory concentration (MIC), often does not predict whether fungal infections will respond to drug treatment in the clinic. Here we define and quantify an antifungal response, termed ‘perseverance’, that correlates with clinical outcomes in patients treated with fluconazole, the most widely used antifungal drug. Perseverance is defined by the ability of fungal cells to grow at drug concentrations above the MIC, and is measured either as the ‘fraction of growth’ (FoG) in drug disk diffusion assays, or as the degree of ‘supra-MIC growth’ (SMG) in broth microdilution assays. Perseverance is distinct from resistance and is related to the proportion of cells that form colonies at supra-MIC drug concentrations. Higher perseverance correlates with shorter lag times at high drug concentrations, highlighting a critical distinction from ‘tolerance’ in bacterial populations. Importantly, several adjuvant drugs, including inhibitors of Hsp90 and calcineurin, eliminate perseverance without altering the MIC. We also find that perseverance and resistance are sensitive to mutations in different pathways, underscoring the distinct nature of the two phenomena. Analysis of isolates recovered from clinically persistent or non-persistent infections in immunocompetent patients reveals that persistent isolates show higher levels of perseverance than isolates cleared by antifungals. Thus, perseverance is an intrinsic property of fungal isolates that correlates with the success or failure of treatment in the clinic, and may provide a useful parameter for predicting clinical persistence and choosing appropriate antifungal therapies.


46
Antimicrobial drug resistance, the ability to grow well at high drug concentrations irrespective of treatment duration, involves several distinct mechanisms including alteration in 48 drug targets, drug efflux or drug inactivation. Of note, infections generally follow the "90/60" rule 49 for predicting therapeutic outcomes based on in vitro susceptibility testing: approximately 90% 50 of susceptible isolates and 60% of resistant isolates respond to therapy 1-5 . This is important as 51 it implies that drug resistance, as defined by in vitro assays, is often not sufficient to explain 52 clinical outcomes, and that effective treatment is likely influenced by patient attributes as well as 53 by additional characteristics of the infecting isolates.

54
Two microbial variables that have been established to affect drug responses in bacteria 55 are "tolerance" and "persistence". Importantly, both tolerant and persistent isolates require 56 similar levels of antibiotic drugs for killing as susceptible isolates. However, the duration of drug 57 treatment necessary to achieve killing of tolerant or persistent strains is much longer than that 58 necessary for susceptible strains. Mechanistically, both tolerance and persistence involve the 59 ability of cells to survive and/or grow during transient exposure to high drug concentrations.

60
Toxin/antitoxin-modules are responsible for promoting tolerance and persistence in some cases, 61 and isolates with more toxin/antitoxin modules therefore tend to exhibit higher levels of 62 persistence 6 . A major distinction between the two phenomena is the size of the subpopulations 63 involved: tolerance is observed in 10-90% of cells whereas persistence occurs in <1% of cells 64 (reviewed in 7 )., Both tolerance and persistence typically involve slower growth or a longer lag given isolate, both FoG and RAD levels were reproducible when evaluating cells taken from 150 different colonies or when cells were taken from inside or outside the ZOI and re-tested for their 151 drug responses (Fig. 1D). These observations indicate that FoG, the level of growth inside the 152 ZOI, is a stable and heritable property of a given C. albicans strain. Moreover, cells growing

179
Clearly, FoG is a property observed in a range of yeast species and is not specific to 180 fluconazole.

181
Broth microdilution assays (BMDAs) are considered the clinical assay of choice for 182 quantifying drug susceptibility. We therefore measured MIC 50 values, the lowest drug 183 concentration that inhibits 50% of growth, in the seven C. albicans isolates chosen above.

184
Trailing growth is frequently observed in BMDAs as increased growth at 48 h relative to 24 h, 185 but this phenomenon is not usually quantified or otherwise used in clinical assays. We 186 quantified supra-MIC growth (SMG) in BMDAs as the average growth per well (at 48 h) above 187 the MIC 50 (at 24 h) normalized to growth levels in the absence of drug ( Fig. 2A)

215
In addition, DDAs performed using rich medium (YPD) often gave slightly different FoG and 216 RAD levels than those using casitone medium ( Supplementary Fig. 3E). Nonetheless, the 217 relative differences between the degree of FoG in different C. albicans strains was generally 218 maintained between the two media ( Supplementary Fig. 3E).

260
We performed a parallel analysis examining perseverance in two additional lineages that 261 included strains with differences in growth rates (Supplementary Table 1). Thus, we tested two 262 faster growing derivatives of the slow-growing isolate P37005 that were obtained via 263 passaging 77 , as well as heterozygous and homozygous null mutants of the Clb4 mitotic cyclin 264 that result in slower growth 78 . Each of these strains were compared on DDAs using fluconazole.

265
We found no correlation between FoG and doubling time or between FoG and lag duration for

305
We examined the drug responses of persistent and non-persistent isolates using DDAs 306 and BMDAs as described above, and included three of the seven strains utilized above as 307 additional controls (termed 'susceptible' isolates). As expected, persistent and non-persistent

324
Taken together these results reveal that persistent isolates display significantly higher 325 levels of growth at high drug concentrations compared to isolates that are clinically cleared by 326 fluconazole treatment. In fact, many of the persistent isolates had FoG levels > 0.5, a level at 327 which more than half of the cells in the population are able to grow, albeit more slowly, in the 328 presence of antifungals (Fig. 3B). This is important because it indicates that perseverance is an isolates SC5314, GC75, P78042 and P87, resistant (R, n =1) isolate P60002 31 , non-persistent isolates (NP, n = 7) and the first patient isolate from each of the series of clinically persistent  FLC (µg/ml) 0 à 128 SMG strains (P, n = 12). Asterisks indicate significant differences between persistent and non-persistent isolates (unpaired t-test, P = 0.67 for RAD, P < 0.001 for FoG). (c) Broth microdilution assays showing MIC and SMG levels at 24 and 48 h for the susceptible control strains (S, n = 4), the nonpersistent isolates as in (b) (NP, n = 7) and for both the initial and final isolates for each of the 12 clinically persistent series (S01-12). The final isolate in S03 became fluconazole resistant (R), therefore the penultimate isolate in this series was included as well. Asterisks indicate significant differences between persistent and non-persistent isolates (unpaired t-test, NP vs P-first P = 0.061 and NP vs P-last P = 0.053 for SMG at 24 h, NP vs P-first P < 0.001 and NP vs P-last P < 0.001 for SMG at 48 h).
intrinsic property of clinical isolates distinct from drug resistance. Furthermore, perseverance

357
We conclude that exposure to inhibitors of either Hsp90 or calcineurin reduced growth   Table 2 and Supplementary Fig. 8B). Surprisingly, however, perseverance 369 levels were significantly reduced in some of the parental control strains. Specifically, FoG levels 370 dropped by ~40-50% in CAI-4 and many of its derivatives (Supplementary Table 2

379
Given that calcineurin inhibitors reduced perseverance, we first tested null mutants 380 lacking genes that encode the calcineurin subunits Cna1 68 and Cnb1 90 , along with 381 independently constructed mutants lacking Crz1 91 , the major transcription factor activated by 382 calcineurin in response to drug stress. Null mutants in CNA1, CNB1 and CRZ1 constructed in 383 two different parental strains (SC5314 and SN192) had significantly reduced perseverance (50-

385
These results indicate that calcineurin promotes perseverance via the Crz1 transcription factor.

386
In contrast, deletion of genes encoding regulators of calcineurin, including RCN1 and 387 RCN2, had little effect on RAD or FoG when compared to isogenic parental strains 388 (Supplementary Table 3 Table 3 and Fig. 6A).

398
We also tested a strain lacking CAS5 because of its reported role in fluconazole 399 responses. CAS5 encodes a transcription factor originally identified in C. albicans for its role in 400 caspofungin survival 92 and was shown to enhance the cidality of fluconazole, especially when 401 analyzed at 48 h rather than 24 h after drug exposure 93 . Indeed, the cas5∆∆ mutant reduced 402 FoG by 90% and increased RAD by 25% (Supplementary Table 3

409
In contrast, tac1 and mrr1 mutants had no significant effects on RAD or FoG (Supplementary   410   Table 3 and Fig. 6B). This is consistent with the idea that there are multiple efflux pump genes 411 and transcriptional regulators that control them, and that any one of them is likely to be 412 redundant with efflux activity of the others sufficient for wild-type activity 95-97 .

413
Together, these results show that different genetic pathways, such as calcineurin activity

431
We show that perseverance is concentration-independent, is correlated with the size of 432 the subpopulation that grows at drug concentrations above the MIC, and is inversely correlated 433 with the length of lag phase for these colonies. The mechanisms that determine the size of the 434 subpopulation that can grow at supra-MIC concentrations remain to be determined. Importantly, 435 we measure SMG and FoG relative to the MIC, utilizing standard assays measured on 436 consecutive days, to provide information about both resistance and perseverance. We show 437 that perseverance is influenced by pH, temperature, and nutritional inputs, as previously 438 reported for trailing growth and tolerance 26,66,67,70 . The term "perseverance" was chosen as it 439 describes the process rather than the assay, and it avoids confusion with tolerance mechanisms

450
However, this study clearly shows that these proteins, and the pathways they regulate, are 451 important for a response to drug that is distinct from resistance and that specifically affects cell 452 survival at supra-MIC antifungal concentrations.

453
Multiple pathways influence perseverance, some of which we identify in this study.

454
Vps21, a protein involved in membrane trafficking, clearly impacts perseverance via its ability to 455 increase cytoplasmic calcium stores 10,106 , which, in turn, is likely to increase calcineurin activity.

475
Similarly, we do not yet understand how transcription factors CAS5 and UPC2 affect 476 both resistance and perseverance. One option is that they may regulate multiple independent 477 pathways, some of which influence resistance and others that affect perseverance. A recent 478 study supports this idea, given that cas5∆/∆ strains cause altered promoter occupancy of RNA 479 polymerase II at over 600 genes 110 . Alternatively, genes affecting resistance may also reduce 480 the degree to which an antifungal drug causes cellular stress and thus may alter the 481 requirement for the stress response mechanisms that drive perseverance.

482
Of more direct significance, this analysis of a well-matched set of clinically persistent and   Table 4).