The distinct biochemical property enables thymidylate kinase as a drug target and participates in pyrimidine drug sensitivity in Candida albicans

The ability to overcome drug resistance in outbreaks of Candida albicans infection is an unmet need in health management. Here, we investigated CDC8, which encodes thymidylate kinase (TMPK), as a potential drug target for the treatment of C. albicans infection. In this study, we found that the specific region spanning amino acids 106-123, namely, the Ca-loop of C. albicans TMPK (CaTMPK) contributes to the hyperactivity of this enzyme compared to the human enzyme (hTMPK) and to the utilization of deoxyuridine monophosphate (dUMP)/ deoxy-5-Fluorouridine monophosphate (5-FdUMP) as a substrate. Notably, CaTMPK but not hTMPK enables dUTP/5-FdUTP-mediated DNA toxicity in yeast. CRISPR-mediated deletion of this Ca-loop in C. albicans demonstrated the critical role of this Ca-loop in fungal growth and susceptibility to 5-Fluorouridine (5-FUrd). Moreover, pathogenic and drug-resistant C. albicans clones were similarly sensitive to 5-FUrd. Thus, this study not only identified a target site for the development of CaTMPK-selective drugs but also revealed 5-FUrd to be a potential drug for the treatment of C. albicans infection. Author summary The emergence of drug-resistant C. albicans strains is a serious medical concern that may be addressed by targeting an essential fungal enzyme. CDC8 encodes thymidylate kinase (TMPK), which is the key enzyme required for dTTP synthesis and is an essential gene for yeast growth. Therefore, the differences of TMPK between human and C. albicans can be a potential drug targeting site. This study defines a specific Ca-loop unique to CaTMPK from C. albicans, contributing to hyper-activity over human enzyme (hTMPK). CRSPR-edited deletion of this loop also suppressed the growth of C. albicans. Moreover, we present evidence that this loop enables dUMP utilization by CaTMPK, but not hTMPK. CaTMPK is also capable of using 5-FdUMP as a substrate, which contributes to 5-FUrd-mediated toxicity. Importantly, we found that many drug resistant pathogenic C. albicans isolates from patients are sensitive to 5-FUrd, which has not been used as a drug against fungal infection.


The biochemical differences between hTMPK and CaTMPK 100
Members of the TMPK enzyme family have been categorized into type I and type II 101 enzymes. Both hTMPK and CaTMPK are type I enzymes [7]. The sequences in P-loop, 102 TMP binding and DR sites in catalysis are highly conserved between CaTMPK and 103 hTMPK ( Fig 1A). The structure-based alignment (PDB: 5UIV and PDB:1E2D) [12,13] 104 shows that the Candida-specific Ca-loop starts from Phe107 to Lys108. We have 105 previously developed a series of hTMPK inhibitors [14,15]. Here, we showed that two 106 compounds, YMU1 and 3b, at 1 µM effectively suppressed hTMPK activity while at 107 10µM having no inhibitory activity to CaTMPK (Fig 1B). We compared the enzyme 108 activities of CaTMPK and hTMPK. The kinetic parameters of CaTMPK and hTMPK are 109 shown in Table 1. The kcat of CaTMPK for dTMP was 15-fold higher than that of hTMPK, 110 with similar Km values observed for ATP and dTMP. Thus, the catalytic efficiency of 111 CaTMPK is much higher than that of hTMPK. Taken together, these results suggest that 112 these two TMPKs exhibit distinct properties. The Ca-loop confers hyperactivity on CaTMPK and determines the growth rate of 122

C. albicans 123
To verify the contribution of the Ca-loop to the catalytic rate of CaTMPK, we generated 124 a Ca-loop-deleted mutant, Δ107-118. Recombinant wild-type and Δ107-118 CaTMPK 125 proteins were purified for activity assays. The 107-118 deletion led to a 70% reduction in 126 activity (Fig 2A). Because CaTMPK lacking the Ca-loop exhibited decreased activity, we 127 asked whether deletion of this Ca-loop would affect the growth of C. albicans. CaTMPK 128 is the gene product of CDC8. Therefore, we performed CRISPR-mediated deletion of the 129 319-354 region of the CDC8 gene locus (cdc8Δ107-118) in the nonpathogenic C. albicans 130 HLC54 strain [16]. The sgRNA targeting the Ca-loop sequence for Cas9 cleavage and the 131 plasmid pGEX-2T-CaTMPKΔ107-118 as the donor template were cotransformed, 132 followed by selection with nourseothricin (Nat) (S1A and S1B Figs). The selected clone 133 was analyzed by PCR amplification of the region spanning the Ca-loop sequence of the 134 CDC8 gene locus (S1C Fig.). Sanger sequencing further confirmed that the cdc8Δ107-118 135 clone had a 36-bp in-frame deletion in the Ca-loop (S1D Fig.). We then compared the 136 growth rates of HLC54 and HLC54 [cdc8Δ107-118]. The results showed that 137 HLC54[cdc8Δ107-118] grew much slower than the parental strain HLC54 (Fig 2B). Taken 138 together, these results show that the Ca-loop is a critical structural element of CaTMPK 139 in the rate of dTDP formation, which is indeed controlling the growth of C. albicans. 140

141
The Ca-loop allows CaTMPK to use dUMP as a substrate 142 Given the unique structural feature of the dTMP/LID/Ca-loop of CaTMPK, we further 143 asked whether there is a difference in substrate selectivity between hTMPK and CaTMPK. 144 Here, we performed an isotope-labeling-based γ-phosphate transfer assay to analyze the 145 capacity of purified CaTMPK and hTMPK to phosphorylate different dNMPs. As 146 expected, both enzymes were capable of phosphorylating dTMP to dTDP, and dAMP, 147 dGMP and dCMP were not used as substrates by these enzymes (Fig 3A). However, there 148 was a striking difference in the capability of CaTMPK to transfer γ-32 P-phosphate from 149 ATP to dUMPs, so as 5-FdUMP (Fig 3B). Analysis of the steady-state kinetics revealed a 150 large difference in the rate of dUMP phosphorylation between hTMPK and CaTMPK (Fig  151   3C). The kcat of CaTMPK for dUMP was 6-fold higher than that for dTMP, but the catalytic 152 efficiency (kcat/Km) of CaTMPK for dTMP remained higher than that for dUMP (Table 1  153 and Fig 3C). This finding indicates that CaTMPK normally prefers dTMP as a substrate. 154 However, when the dUMP/dTMP ratio increases, CaTMPK exhibits increased dUDP 155 production. Next, we asked whether the Ca-loop is involved in the utilization of dUMP as 156 a substrate. Similar to the results of the dTMP-based assay, the Δ107-118 was defective 157 in dUMP utilization (Fig 3D), indicating that the Ca-loop determines the utilization of 158 dUMP as a substrate in the enzymatic reaction. Thymidylate synthase (TS) is the enzyme responsible for converting dUMP to dTMP. 163 Upon uptake, 5-fluorouracil (5-FU) and 5-fluorouridine (5-FUrd) are metabolically 164 converted to 5-FUMP and 5-FdUMP [17], the latter of which is a suicide inhibitor of TS. 165 Therefore, treatment with 5-FU or 5-FUrd leads to dUMP accumulation. A previous study 166 has shown that 5-FU-induced death in S. cerevisiae is reversed by deletion of uracil-DNA-167 glycosylase (UNG1) [18]. This finding suggests that 5-FU treatment causes 168 misincorporation of dUTP, the removal of which by Ung1 leads to DNA toxicity and cell 169 death. Because CaTMPK is hyperactive in the conversion of dUMP to dUDP, we then 170 asked whether CaTMPK and hTMPK can mediate the differential toxicity of 5-FU and 5-171 FUrd in S. cerevisiae. RWY-42-22B, harboring a temperature-sensitive mutation of cdc8 ts , 172 is an S. cerevisiae strain that can grow at 23℃ normally but is not viable at 30℃. The 173 growth of this strain at nonpermissive temperatures was rescued by expression of hTMPK 174 or CaTMPK without a significant difference in growth rate (S2 Fig.). To avoid 175 complications associated with endogenous cdc8 ts , two strains expressing hTMPK or 176 CaTMPK were further engineered by replacement of cdc8 ts with HIS3. We tested the 177 sensitivity of these two strains to 5-FU. Interestingly, the growth of yeast expressing 178 CaTMPK was obviously suppressed along a gradient increase in 5-FU concentration (0-179 2.5 µM) in agar plates; in contrast, only a slight response was observed in hTMPK-180 expressing yeast (Fig 4A). To determine whether 5-FU-induced death in the CaTMPK-181 expressing strain was associated with misincorporation of dUTP/5-FdUTP, ung1 was 182 deleted from this strain for the 5-FU sensitivity assay. The results showed that deletion of 183 ung1 allowed the CaTMPK strain to survive in 5-FU-containing medium ( Fig 4A). Similar 184 results were observed in 5-FUrd gradient (0-50 µM) agar plates. A high concentration of 185 5-FUrd was required for suppression of CaTMPK-expressing yeast growth, most likely 186 due to the lower activity of uridine permease than uracil permease in S. cerevisiae. 187 Nevertheless, ung1 deletion abrogated the 5-FUrd response in CaTMPK-expressing yeast. 188 In conclusion, expression of CaTMPK increases the susceptibility of yeast to 5-FU and 5-189 FUrd via dUTP/5-FdUTP-mediated DNA toxicity ( Fig 4B). range, while that of 5-FC was in the µM range. Thus, the antifungal activity of 5-FUrd is 197 more potent than that of 5-FC. Notably, C. albicans growth was unaffected by 50 µM 5-198 FU ( Fig 5A). The low 5-FU sensitivity of C. albicans is probably due to the poor uracil 199 permease activity [19,20]. host liver toxicity. We then evaluated the potential toxicity of 5-FUrd in human HepG2 238 cells and found that the IC50 of 5-FUrd was 4 µg/mL (S3 Fig.). Taken together, these 239 results suggest that 5-FUrd might be a suitable option for antifungal treatment.   In this study, our data revealed 5-FUrd to be a potent drug for the treatment of C. 305 albicans infection. This proof-of-concept study showed that a number of C. albicans 306 isolates that were resistant to azoles or 5-FC were susceptible to 5-FUrd at a dose that had 307 little toxicity in human hepatoma cells. We proposed that 5-FUrd could overcome 5-FC 308 resistance in these strains because the metabolic activation pathways of 5-FUrd and 5-FC 309 are different. Therefore, 5-FC resistance resulting from mutations in genes mediating the 310 metabolic activation of 5-FC might not affect 5-FUrd toxicity. In summary, this report 311 reveals the biochemical differences between the essential enzyme TMPK from C. albicans 312 and host (human) TMPK due to the Ca-loop, highlighting this enzyme to be a new drug 313 target site. Moreover, we found that 5-FUrd is indeed a potent inhibitor of C. albicans 314 growth and that the Ca-loop is involved in 5-FUrd toxicity. Notably, 5-FUrd is able to 315 overcome 5-FC and multidrug resistance in pathogenic C. albicans isolates. 316 317

Media and chemicals 319
All S. cerevisiae strains in this study were grown in synthetic defined (SD) medium 320 containing 2% glucose, 6.7 g/L yeast nitrogen base without amino acids and 0.77 g 321 dropout (DO) supplements -Ura (Clontech). C. albicans were grown in YPD (1% yeast 322 extract, 2% bactopeptone, and 2% glucose) or SD medium at 30 °C. The plasmid pV1090-17 carrying sgRNA sequence targeting CDC8 in C. albicans was 346 cloned at BsmBI restriction sites of the plasmid pV1090, and the plasmid pV1025 contains 347 Cas9 for expression in C. albicans. Plasmids and primers in this study are listed in Table  348 S3 and Table S4. containing Nat at 250 µg/ml and incubated at 30℃ for 3 days. The colonies grown on Nat 369 plate were isolated for genomic DNA extraction followed by PCR examination. All strains 370 in this study are listed in Table S2. Tris-HCl pH7.5, 150 mM NaCl, 5 mM EDTA, 1 mM DTT, 1mM PMSF and 1 mM 386 protease cocktail. The GST-tag was then cleaved from the target proteins by thrombin. 387

In vitro activity assay (NADH-coupled assay) 388
Activity assay was measured using spectrophotometric method by coupling ADP 389 formation to the oxidation of NADH catalyzed by pyruvate kinase and lactate For plates with fixed concentrations of drug, the cultures were adjusted to 0.1 by OD600, 436 then serial diluted with five-fold before spotted on plates. The plates were incubated at 437 30℃ for 2-3 days. 438

Statistical analysis 439
Data are presented as the mean ± standard error of the mean. Statistical comparison 440 of means was performed using a two-tailed unpaired Student's t-test.  CaTMPK activity, which was set to 100%. The results represent the mean ± SD, n=3. P-581 values were determined by Student's t-test (***P<0.001). 582