Hmg1 Gene Mutation Prevalence in Triazole-Resistant Aspergillus fumigatus Clinical Isolates

Recently, mutations in the 3-hydroxy-3-methylglutaryl-coenzyme-A-reductase-encoding gene (hmg1), a gene involved in ergosterol production, were associated with triazole-resistance in Aspergillus fumigatus. In this study, we determined the prevalence and characteristics of hmg1 mutations in a collection of clinical triazole-resistant A. fumigatus isolates collected during 2001–2019 from two international mycology reference centers: the Belgian National Reference Center for Mycosis and the Center of Expertise in Mycology Radboudumc/CWZ. Clinical isolates with and without cyp51A gene mutations and randomly selected wild-type (WT) controls were included. Isolates were characterized by in vitro susceptibility testing, cyp51A and hmg1 sequencing, and short tandem repeat typing. Available clinical records were analyzed for previous triazole exposure. In 23 isolates (24%) of the 95 triazole-resistant A. fumigatus isolates, hmg1 gene mutations were observed; including 5/23 (22%) isolates without cyp51A gene mutations and 18/72 (25%) with cyp51A mutations. Four previously described hmg1 gene mutations (E105K, G307R/D, G466V, and S541G) and two novel mutations (W273S and L304P) were found; 4/23 (17%) in the sterol-sensing-domain region. No triazole-antifungal exposure was reported in 75% (9/12) of patients harboring an isolate with hmg1 gene mutations. Three of 39 WT isolates (8%) contained a hmg1 gene mutation; E105K (2-isolates) and S541G. Hmg1 gene mutations were predominantly found in A. fumigatus with cyp51A mutations with voriconazole MICs ≥ 8 mg/L.


Introduction
The triazole antifungal agents are recommended for prophylaxis (posaconazole) and first-line treatment (voriconazole and isavuconazole) of Aspergillus related-diseases [1,2]. However, the preferred use of triazoles is being threatened by reports of increasing resistance to these antifungals worldwide [3]. The most commonly reported triazole-resistance associated mechanisms in Aspergillus

Aspergillus fumigatus Clinical Isolates and Triazole-Resistance Determination
Clinical A. fumigatus isolates received at the National Reference Center for Mycosis UZ Leuven in Belgium (between 2012 and July 2019) and at the Center of Expertise in Mycology Radboudumc/CWZ in the Netherlands (between 2001 and 2017) were screened for inclusion in this study. The inclusion criteria were as follows: (a) triazole-resistant isolates with WT-cyp51A gene; (b) triazole-resistant isolates with cyp51A gene mutations subdivided in three subgroups TR 34 /L98H, TR 46 /Y121F/T289A, and others (maximum 15 isolates per year from each subgroup); (c) 39 randomly selected triazole-susceptible WT isolates (based on an hmg1 mutation prevalence of 24%, power 0.9, alpha 0.05); and maximum one isolate per patient within a four-month period. Triazole-resistance screening of A. fumigatus isolates followed by susceptibility determination (EUCAST broth microdilution reference method) of suspected triazole-resistant isolates were performed as previously described [16]. Isolates were designated as triazole-resistant if at least one minimal inhibitory concentration value (MIC) was above the established EUCAST resistance clinical breakpoints for A. fumigatus (voriconazole >2, itraconazole >2, posaconazole > 0.25, mg/L) at the time of isolates selection [17]. The presence of mutations within the cyp51A gene and its promotor region was assessed by sequencing as reported before [18].

hmg1 Gene Sequencing
Amplified DNA from selected A. fumigatus isolates was sequenced using specifically designed primers (Table S1) and the BigDye-Terminator-v3.1 cycle-sequencing kit (Applied-Biosystems, Lithuania). Reaction products were purified (DyeEX2.0-Kit; Qiagen, Germany), dried, and reconstituted in 20 µL of HiDi Formamide (Applied Biosystems, UK) according to manufacturer and run on an ABI3730xl Genetic Analyzer (Applied-Biosystems, USA). Obtained sequences were aligned to create a consensus sequence using the CLC-Genomics-Workbench software (CLC-bio, Denmark). Sequence variant detection was performed by comparison of each isolate consensus sequence to the A. fumigatus hmg1 gene reference sequence (FungiDB:AFUB_020770_A 1163).

Genotyping
Genetic relatedness of isolates with hmg1 gene mutations and without hmg1 gene was determined using two separate multiplex PCRs specific for A. fumigatus (short tandem repeats) of three trinucleotide loci (M3_STRAf 3A-C primers) and three tetranucleotide loci (M4_STRAf 4A-C primers), respectively, as described by de Valk HA. et al. [19] (Applied Biosystems, UK; Table S1). Fragments (4 µL of cleaned PCR product combined with HiDi Formamide and GeneScan 500 LIZ size standard; Applied Biosystems, UK) were detected on a 96-capillary array ABI3730xl Genetic Analyzer (Applied Biosystems, USA). Number of repeats were determined using the Peak Scanner Software v1.0 (Applied Biosystems). Isolates genetic distance (Euclidean method) and hierarchical relationship (Ward's minimum variance linkage clustering algorithm method; ggplot2 package) were performed with the statistical computing and graphics software R (version 4.0.2) [20].

Antifungal Exposure and Clinical Data Management
Patient electronic records were retrospectively consulted to check for exposure to any antifungal either as treatment or as prophylaxis. Use of any antifungal was recorded from 30-days before the isolate was cultured. Antifungal exposure in patients with multiple isolates was determined based on culture of the first triazole-resistant isolate. This study was performed in accordance to each center ethical regulations.
Genetic relatedness analysis of isolates with hmg1 gene mutations (23 triazole-resistant and 3 susceptible isolates) and without hmg1 gene mutations (16 triazole-resistant and 12 WT isolates) showed five clusters (A-E, Figure 2) occurring at the same horizontal distance with an average Euclidean distance of 3.2, varying among isolates from 0.0 to 7.08. Two of the analyzed isolates harboring TR 46 /Y121F/T289A cyp51A gene and G466V, S541G hmg1 gene mutations (V107-65 and V095-29) and two triazole-resistant isolates with no cyp51A or hmg1 gene mutations (CYP-15-106 and CYP-15-115) were genetically identical; Euclidean distance 0.0). Among hmg1 triazole-resistant isolates, 3 out of 23 (13%) were clustered together in clade A, none in clade B, 10 in clade C (44%), one in clade D (4%) and 9 in clade E (39%). In clade E, 60% (6/10) of hmg1 triazole-resistant isolates harboring the amino acid substitution E105K and all triazole-resistant isolates (3/3) with the S541G one were found. Isolates with hmg1 gene mutations in the SSD were mostly found in clade C (3/4, 75%) in addition to all triazole-resistant isolates harboring the G466V and S541G amino acid substitutions (5/5, 100%). WT isolates with the E105K were clustered in clade C and D, and the S541G in clade E where most one were found. Isolates with hmg1 gene mutations in the SSD were mostly found in clade C (3/4, 75%) in addition to all triazole-resistant isolates harboring the G466V and S541G amino acid substitutions (5/5, 100%). WT isolates with the E105K were clustered in clade C and D, and the S541G in clade E where most triazole-resistant isolates harboring this mutation are situated. Hmg1 gene mutations were found in isolates from our collection since 2001, our starting study year.

Previous Exposure to Triazole Antifungals Is Uncommon among Patients Harboring Isolates with hmg1 Gene Mutations
Data about exposure to triazole antifungals could be retrieved for 51 of 92 patients with triazole-resistant isolates (55%). Exposure to triazole antifungals was observed in 25% (3/12) of patients with hmg1 gene mutation compared to 75% (9/12) in those without triazole-antifungal exposure ( Table 1). No previous triazole exposure was observed in the two patients harboring a WT-isolate with the E105K hmg1 gene mutation.

Discussion
We report a prevalence of 24% hmg1 gene mutations (23/95) in general and of 4.2% in the SSD region (4/95) in our collection of triazole-resistant clinical A. fumigatus isolates. Our reported prevalence is lower than observed in previous reports. Rybak et al. reported hmg1 gene mutations in 52% (11/21) of their clinical triazole-resistant A. fumigatus isolates, while a prevalence of 45.5% (5/11) was found in isolates from Manchester, UK [9]; however, the number of cases in these studies were small and, as only few hmg1 gene mutations have been related to resistance in A. fumigatus and not all might have the potential to confer resistance we cannot draw a firm conclusion. hmg1 gene mutations could be accountable as the underlying mechanism conferring triazole-resistance in 5 of 23 (22%; two in the SSD [9%]) isolates without cyp51A gene mutations, but other mechanisms conferring triazole-resistance cannot be excluded. The MICs of these isolates were above the clinical breakpoint for voriconazole of 2 mg/L (MIC ≥ 4 mg/L), while a variable phenotype was observed for posaconazole (MIC 0.125 to >8 mg/L) and itraconazole (MIC 0.125 to >8 mg/L).
Mutations in the hmg1 gene were also observed among triazole-resistant isolates with known azole-resistant related cyp51A gene mutations. Interestingly, hmg1 gene mutations were detected mostly in isolates harboring tandem repeat mutations categorized as atypical resistance phenotype, with a prevalence of 43% (6/14) in atypical-TR 34 /L98H isolates group and 54% (7/13) in atypical-TR 46 /Y121F/T289A compared to 14% (4/29) and 0% (0/13) in typical-TR 34 /L98H and -TR 46 /Y121F/T289A ones, respectively. It may appear that genetic variances in the hmg1 gene in isolates harboring known associated triazole-resistant cyp51A gene mutations further modulate triazole-resistance elevating MIC values from conventional expected ones as reported in other non-cyp51A gene mediated mutations [23]. Additional investigation is required to determine if this is indeed the case for these cyp51A gene mutations and the mechanisms involved.
Genetic variances in the SSD of the hmg1 gene have been associated with triazole-resistance in A. fumigatus [8,9]. Among our triazole-resistant isolates, 4 out of 95 (4.2%) presented amino-acid variances in this region, two at novel locations W273S and L304P, and two in the previously described G307 position (Table 2), with an amino acid change from glycine to arginine (G307R) or to aspartic acid (G307D). The hmg1 amino acid substitution L304P was found in an isolate from the other cyp51A group with the M263I mutation. Rybak et al., reported two isolates harboring the M263I mutation in which substitution of their cyp51A gene to a WT one, did not restore susceptibility [9]. Furthermore, incorporation of the position of the M263I into a protein homology model of Aspergillus fumigatus [5] (Figure 3) showed that this mutation is positioned outside on the protein surface and not in the well-known associated triazole-resistance protein regions (triazole-antifungals entrance or ion-exchange channels) [5].
J. Fungi 2020, 6, x FOR PEER REVIEW 7 of 11 triazole-resistance elevating MIC values from conventional expected ones as reported in other non-cyp51A gene mediated mutations [23]. Additional investigation is required to determine if this is indeed the case for these cyp51A gene mutations and the mechanisms involved. Genetic variances in the SSD of the hmg1 gene have been associated with triazole-resistance in A. fumigatus [8,9]. Among our triazole-resistant isolates, 4 out of 95 (4.2%) presented amino-acid variances in this region, two at novel locations W273S and L304P, and two in the previously described G307 position (Table 3), with an amino acid change from glycine to arginine (G307R) or to aspartic acid (G307D). The hmg1 amino acid substitution L304P was found in an isolate from the other cyp51A group with the M263I mutation. Rybak et al., reported two isolates harboring the M263I mutation in which substitution of their cyp51A gene to a WT one, did not restore susceptibility [9]. Furthermore, incorporation of the position of the M263I into a protein homology model of Aspergillus fumigatus [5] (Figure 2) showed that this mutation is positioned outside on the protein surface and not in the well-known associated triazole-resistance protein regions (triazoleantifungals entrance or ion-exchange channels) [5]. The most prevalent hmg1 mutation found among triazole-resistant isolates in our collection was the previously described amino-acid change E105K [9] (10/23 isolates, 44%). E105K is located at the beginning of the membrane bound region but not in the SSD region and its exact role in triazole-resistance is not known. E105K was mostly found in isolates (7/10, 70%) with an atypical resistance phenotype (MIC: ≥4 mg/L itraconazole, ≥8 mg/L voriconazole). Our second and third most prevalent mutations were G466V, S541G (5/23, 22%) and S541G (3/23, 13%) [9], also located outside the SSD region. Most isolates harboring these mutations were likewise classified as having Only relevant mutations identified when compared to reference genome AFUB_020770 (GenBank: EDP54027.1) are mentioned in this table. Amino-acid variances located in the sterol-sensing domain region (SSD; 242-415) are in bold and Italic. 1 Amino-acid variance previously reported also in susceptible A. fumigatus isolates. 2 All isolates belonged to the same patient. * References: Hagiwara et al. [8]., Rybak et al. [9]., Sharma et al. [10]., Chi-Jung et al. [11], Takeda et al. [12].
Within our study, the E105K and the S541G mutation were also found in two (5%) and one (3%), respectively, of 39 isolates with a triazole WT phenotype. The S541G and H564Y hmg1 mutations located likewise outside the SSD, have also been reported in non-resistant isolates [8,11]. This a major issue, as it raises the probability that these mutations located outside the SSD most likely do not confer resistance and that other mechanisms are responsible for the observed resistant phenotypes. We cannot discard the possibility that compensatory mutations, which could modulate azole susceptibility to WT MICs, exist in the non-resistant isolates harboring these mutations. Triazole-resistant A. fumigatus isolates harboring tandem repeat mutations, such as TR 34 /L98H, are known to be closely related [24]. Our genetic relatedness analysis indicates that 60% (6/10) of E105K isolates (in addition to one E105K WT isolate) and all (3/3) S541G triazole-resistant isolates (including the WT isolate harboring this mutation) are closely related (clade E), suggesting the emergence of variants from a local common ancestor which decreases the likelihood of being a commonly distributed. The possibility that these mutations might confer other beneficial effects in regard to survival or fitness should also be considered. Recombinant experiments are required to determine whether E105K and S541G mutations, and others found outside the SSD, play a role in conferring triazole-resistance in A. fumigatus or are common polymorphisms of the Hmg1 locus.
Recently, Nakano et al. [25] reported the presence of hmg1 gene mutations in 14 triazole-resistant A. fumigatus environmental samples from Japan (1/14) and the Netherlands (13/14) harboring the E105K and the S541G (AfuB_020770 genome) amino acid substitutions (voriconazole MICs 8 mg/L, itraconazole MICs 0.5 → 8 mg/L); all located outside the SSD. In our study, we observed that the majority of A. fumigatus isolates with hmg1 gene mutations (75%, 9/12; 1 in the SSD region) were isolated from patients who had no previous triazole-antifungal exposure, further inferring that these mutations might have been previously selected in the environment, as is the case in most cyp51A gene mutation infections in patients [3], but due to low number of cases, results were not significant and should be interpreted with caution. Larger environmental hmg1 triazole-resistance screenings should be undertaken to investigate this more in depth. Our study has some limitations. First, the number of triazole-resistant isolates with no-and others-cyp51A gene mutations in our collections were low, but to the best of our knowledge, this is the largest collection tested so far. Second, clinical data could not be retrieved from all patients.
In this study we systematically analyzed the prevalence of hmg1 gene mutations in a collection of A. fumigatus clinical isolates in two international mycology reference centers. In addition to isolates without cyp51A mutations, hmg1 gene mutations were also detected in isolates with cyp51A mutations mostly in isolates with a voriconazole MIC of ≥4 mg/L and a posaconazole MIC of ≥0.25 mg/L. Further studies are needed to unravel the mechanistic role and clinical implications of hmg1 mutations.