Species Identification and Delineation of Pathogenic Mucorales by Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry

ABSTRACT

This study aimed to validate the effectiveness of matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS)-based identification of filamentous fungi of the order Mucorales. A total of 111 isolates covering six genera preserved at the Research Center for Medical Mycology of Peking University were selected for MALDI-TOF MS analysis. We emphasized the study of 23 strains of Mucor irregularis predominantly isolated from patients in China. We first used the Bruker Filamentous Fungi library (v1.0) to identify all 111 isolates. To increase the identification rate, we created a compensatory in-house database, the Beijing Medical University (BMU) database, using 13 reference strains covering 6 species, including M. irregularis, Mucor hiemalis, Mucor racemosus, Cunninghamella bertholletiae, Cunninghamella phaeospora, and Cunninghamella echinulata. All 111 isolates were then identified by MALDI-TOF MS using a combination of the Bruker library and BMU database. MALDI-TOF MS identified 55 (49.5%) and 74 (66.7%) isolates at the species and genus levels, respectively, using the Bruker Filamentous Fungi library v1.0 alone. A combination of the Bruker library and BMU database allowed MALDI-TOF MS to identify 90 (81.1%) and 111 (100%) isolates at the species and genus levels, respectively, with a significantly increased accuracy rate. MALDI-TOF MS poorly identified Mucorales when the Bruker library was used alone due to its lack of some fungal species. In contrast, this technique perfectly identified M. irregularis after main spectrum profiles (MSPs) of relevant reference strains were added to the Bruker library. With an expanded Bruker library, MALDI-TOF MS is an effective tool for the identification of pathogenic Mucorales.

INTRODUCTION

Mucormycosis is the second most common invasive filamentous infection that occurs in patients who are immunocompromised or suffering from diabetes or trauma (1). The diagnosis and treatment of this disease remain difficult, and the mortality rate associated with this disease is high (2–4). Fungi of the order Mucorales, including Rhizopus, Lichtheimia, Mucor, Rhizomucor, Syncephalastrum, and Cunninghamella spp., are the causative agents of mucormycosis (5). Diagnosis of these pathogens currently relies on morphological identification by the direct examination and culturing of clinical samples, which requires experienced mycologists but sometimes yields inaccurate results (6). DNA sequencing assists identifying strains of fungi, and the most commonly targeted sequence is that of the fungal internal transcribed spacer (ITS) region (7). However, DNA sequence-based identification of Mucorales fungi is labor-intensive and time-consuming, and few reference laboratories can independently perform this task.

Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) is an approach that has recently become widely used for microorganism identification that is based on unique mass spectral fingerprints (8). With regard to fungi, MALDI-TOF MS was first used for yeast identification and is suitable for routine identification with more than 90% accuracy (9–12). Many researchers have since carried out studies with filamentous fungi, e.g., Aspergillus spp., dermatophytes, and Fusarium spp. (13–16). However, few studies of using MALDI-TOF MS for the identification and differentiation of Mucorales exist, and they involve a limited number of species and strains (12, 16, 17). Here, we explored the application of the Bruker Filamentous Fungi library (v1.0) for identifying clinically relevant species of Mucorales, including Rhizopus, Lichtheimia, Mucor, Rhizomucor, Syncephalastrum, and Cunninghamella strains, by comparing the results to those obtained using ITS sequence-based identification. We also assessed the performance of the Bruker library supplemented with the in-house database for the identification of all 111 Mucorales isolates. We emphasized the study of some clinical strains of Mucor irregularis, which causes chronic cutaneous infections geographically confined to Asia, mainly in China (18).

MATERIALS AND METHODS

Strains.

A total of 111 pathogenic Mucorales isolates are preserved in the Research Center for Medical Mycology of Peking University. These strains were isolated from skin and mucosa (n = 20), face and other tissues (n = 19), the respiratory tract (n = 13, including sputum, bronchoalveolar lavage fluid, and bronchus secretions), abscesses and wounds (n = 6), the sinus (n = 3), secretions (n = 3), drainage liquid (n = 1), urine (n = 2), and unknown sources (n = 44). All 111 isolates were identified at the species level based on ITS sequence analysis using the primers ITS1 (5′-TCCGTAGGTGAACCTGCGG-3′) and ITS4 (5′-TCCTCCGCTTATTGATATGC-3′) as previously described (18). Thirteen of the 111 isolates were identified as reference strains unavailable in the Bruker library and used to create a Beijing Medical University (BMU) database, including M. irregularis (n = 6), Mucor hiemalis (n = 2), Mucor racemosus (n = 2), Cunninghamella bertholletiae (n = 1), Cunninghamella phaeospora (n = 1), and Cunninghamella echinulata (n = 1). These reference strains were all isolated from clinical samples and correctly identified at the species level by ITS sequencing. The fungal isolates were cultivated in Sabouraud dextrose broth (Becton Dickinson, Franklin Lakes, NJ) for 12 to 24 h at 28°C on a Loopster digital rotator (IKA, Staufen, Germany).

MALDI-TOF MS identification. (i) Sample preparation for MALDI-TOF MS analysis.

Identification using MALDI-TOF MS began with liquid cultivation; ethanol-formic acid extraction was then used to extract the protein of fungi (19). Briefly, 600 μl of the cultured mixture was transferred to a 1.5-ml tube (Eppendorf, Hamburg, Germany) and centrifuged for 2 min at 15,800 × g. The supernatant was carefully removed, and 1 ml of deionized water was added to the pellet; the sample was vortexed for 1 min and washed twice. The resulting pellet was thoroughly resuspended in a solution composed of 300 μl of deionized water and 900 μl of absolute ethanol (Sigma-Aldrich, St. Louis, MO), and the sample was vortexed for at least 1 min and centrifuged at 15,800 × g for 2 min. The supernatant was discarded, and the pellet was air dried. The pellet was thoroughly resuspended in 50 μl of 70% formic acid, incubated for 15 min at ambient temperature and mixed with 50 μl of acetonitrile (Sigma-Aldrich). This mixture was centrifuged at 15,800 × g for 2 min, and 1 μl of the supernatant was transferred onto an MTP 384 polished steel MALDI target plate (Bruker Daltonik GmbH, Bremen, Germany) and subsequently dried at room temperature. We prepared each sample in four parallel positions. Each sample was covered with 1 μl of a saturated solution of α-cyano-4 hydroxy-cinnamic acid (HCCA; Bruker Daltonik) in 50% acetonitrile and 2.5% trifluoroacetic acid and air dried at room temperature. Finally, the MALDI target was placed into an autoflex speed TOF instrument (Bruker Daltonik).

(ii) MALDI-TOF measurement.

Spectrum acquisition was performed automatically using MALDI Biotyper RTC 4.0 software (Bruker Daltonik) by rastering the target position. The instrumental settings included having ion source 1 at 19.5 kV, ion source 2 at 18.26 kV and a mass range of 2,000 to 20,000. Bruker Daltonik bacterial test standard was prepared as the reference for calibration of the machine (calibration peaks were selected with a mass tolerance range of ±300 ppm). The isolates were identified by the MALDI Biotyper with the Bruker Filamentous Fungi Library 1.0 (Bruker Daltonik) and the Bruker library with the BMU database. The scores for identification are expressed as log(score) values: scores of ≥2.000 indicated species-level identification, scores of ≥1.700 indicated genus-level identification, and scores of <1.700 were considered unreliable (20). The identification associated with the highest score was recorded when the results of four parallel tests were identical and consistent with the ITS sequencing results. When an isolate identification was inconsistent with the sequencing results or two or more species were present, the result was recorded as a misidentification.

(iii) BMU database creation.

Preparation of reference strain samples followed the procedure described above in accordance with the manufacturer's recommendations. The isolate samples used for database construction were deposited in eight replicates, and a Bruker Daltonics FlexControl 3.4 was used to manually obtain three fingerprints from each target spot (summing to a signal strength of >10,000); MALDI Biotyper OC 4.0 software evaluated the quality of 24 fingerprints, particularly repetition among fingerprints. The poorly repeated mass spectra were then removed. No fewer than 20 mass spectra were used to build the database.

Phylogenetic analysis and main spectrum profile (MSP) dendrogram. (i) ITS sequencing phylogenetic analysis.

CLC Sequence Viewer 7.5 was used to first process the bidirectional sequences, and the complete ITS sequences for each strain were acquired and then aligned by MEGA version 7 in ClustalW. The alignments were checked and modified visually. The best-fit substitution models for data sets of Rhizopus spp. and Mucor spp. were T92 and T92+G, respectively. We selected the maximum-likelihood (ML) method to construct phylogenetic trees. The bootstrap method was applied to test for phylogeny, and the number of bootstrap replications was 1,000 for each set of data. A bootstrap support value of ≥70% was considered significant (21).

(ii) MSP dendrogram.

Four protein mass spectral fingerprints were created per sample, and an MSP dendrogram was then created by MALDI Biotyper software. The MSP dendrograms shown in Fig. 1, 2B, and 3B were constructed by using MALDI Biotyper OC 4.0 (Bruker Daltonik) software.

FIG 1.

Cluster analysis based on the main MSPs of all 111 isolates, which belong to six different clinically relevant genera of Mucorales. The distance levels of each subdivision are shown in arbitrary units, and the isolates of the Lichtheimia and Cunninghamella genera are shown as spp.

FIG 2.

Rhizopus spp. (A) MSP dendrogram based on MALDI-TOF MS analysis. (B) ML tree obtained from ITS sequences of R. arrhizus and R. microsporus. Numbers of ML bootstrap values above 70% are shown at the nodes.

FIG 3.

Mucor spp. (A) MSP dendrogram based on MALDI-TOF MS analysis. (B) ML tree based on ITS sequences of M. circinelloides, M. irregularis, M. hiemalis, and M. racemosus. Numbers of ML bootstrap values above 70% are shown at the nodes.

Statistical analysis.

SPSS 21 was used for statistical analysis; count data were represented by the rate, and the χ² test was used to compare the identification rates of two groups. A P value of <0.01 indicated a statistically significant difference.

Accession number(s).

The GenBank accession numbers of the 111 strains were MG583886 to MG583995 and MG745369.

RESULTS

DNA sequencing identification.

Analysis of the ITS region sequences was used to identify all Mucorales isolates as previously described (19). All sequences were blasted against the NCBI database in June 2017 with a similarity cutoff of ≥99%. ITS sequence analysis of the 111 isolates resulted in the identification of 15 species of Mucorales, including Rhizopus arrhizus (n = 20), Rhizopus microsporus (n = 27), Rhizopus stolonifer (n = 1), Rhizomucor pusillus (n = 4), Syncephalastrum racemosum (n = 2), Lichtheimia corymbifera (n = 4), Lichtheimia ramosa (n = 6), Lichtheimia ornata (n = 1), Mucor circinelloides (n = 9), M. irregularis (n = 23), M. hiemalis (n = 5), M. racemosus (n = 4), C. bertholletiae (n = 3), C. phaeospora (n = 1), and C. echinulata (n = 1).

MALDI-TOF MS identification.

Using the Bruker MALDI-TOF system with the Bruker library identified 55 (49.5%) isolates at the species level and 74 (66.7%) isolates at the genus level. A total of 37 (33.3%) isolates were considered unreliably identified (Table 1). The Bruker library was much better at identifying Rhizopus spp. (R. arrhizus, R. microsporus, and R. stolonifer, 89.6% [43/48]) than Mucor spp. (M. circinelloides, M. irregularis, M. hiemalis, and M. racemosus, 4.9% [2/41]) at the species level. All four isolates of R. pusillus and two isolates of S. racemosum were correctly identified at the species level. However, none of the five Cunninghamella isolates were identified at either the genus or species levels. All 11 isolates of Lichtheimia species were identified correctly at the genus level, and 4 of 11 Lichtheimia species were correctly identified at the species level. Comparing the MALDI-TOF MS results with the ITS sequencing results indicated the following mistakes: one isolate of L. ornata and three isolates of L. ramosa were misidentified as L. corymbifera, and three isolates of M. circinelloides were incorrectly identified as M. ramosissimus. However, 37 isolates were unreliably identified, including M. irregularis, M. hiemalis, M. racemosus, C. bertholletiae, C. phaeospora, and C. echinulata, due to the absence of relevant spectra in the Bruker library. We thus established the BMU database, which included 13 main spectrum profiles (MSPs) of these species. Using this database and the Bruker library together, we correctly identified 35 of 37 isolates at the species level that had not been previously identified; the other 2 M. hiemalis isolates were correctly identified at the genus level (Table 1).

TABLE 1.

Identification of 111 clinical isolates by the Bruker library and the Bruker library plus BMU database

Organism (no. of isolates)	No. (%) of isolates identified at the genus or species level by log(score) valuea
	Bruker library				Bruker library plus BMU database
	≥2.0	≥1.7	<1.7	Mis-ID	≥2.0	≥1.7	<1.7	Mis-ID
R. arrhizus (20)	19 (95)	20 (100)	0 (0)	0 (0)	19 (95)	20 (100)	0 (0)	0 (0)
R. microsporus (27)	24 (88.9)	27 (100)	0 (0)	0 (0)	24 (88.9)	27 (100)	0 (0)	0 (0)
R. stolonifer (1)	0 (0)	1 (100)	0 (0)	0 (0)	0 (0)	1 (100)	0 (0)	0 (0)
R. pusillus (4)	4 (100)	4 (100)	0 (0)	0 (0)	4 (100)	4 (100)	0 (0)	0 (0)
S. racemosum (2)	2 (100)	2 (100)	0 (0)	0 (0)	2 (100)	2 (100)	0 (0)	0 (0)
L. corymbifera (4)	4 (100)	4 (100)	0 (0)	0 (0)	4 (100)	4 (100)	0 (0)	0 (0)
L. ramosa (6)	0 (0)	3 (50)	0 (0)	3 (50)*	0 (0)	3 (50)	0 (0)	3 (50)*
L. ornata (1)	0 (0)	0 (0)	0 (0)	1 (100)†	0 (0)	0 (0)	0 (0)	1 (100)†
M. circinelloides (9)	2 (22.2)	6 (66.7)	0 (0)	3 (33.3)‡	2 (22.2)	6 (66.7)	0 (0)	3 (33.3)‡
M. irregularis (23)	0 (0)	0 (0)	23 (100)	0 (0)	23 (100)	23 (100)	0 (0)	0 (0)
M. hiemalis (5)	0 (0)	0 (0)	5 (100)	0 (0)	3 (60)	5 (100)	0 (0)	0 (0)
M. racemosus (4)	0 (0)	0 (0)	4 (100)	0 (0)	4 (100)	4 (100)	0 (0)	0 (0)
C. bertholletiae (3)	0 (0)	0 (0)	3 (100)	0 (0)	3 (100)	3 (100)	0 (0)	0 (0)
C. phaeospora (1)	0 (0)	0 (0)	1 (100)	0 (0)	1 (100)	1 (100)	0 (0)	0 (0)
C. echinulata (1)	0 (0)	0 (0)	1 (100)	0 (0)	1 (100)	1 (100)	0 (0)	0 (0)
Total (111)	55 (49.5)	67 (60.4)	37 (33.3)	7 (6.3)	90 (81.1)	104 (93.7)	0 (0)	7 (6.3)§

^aMis-ID, misidentification. Symbols: *, misidentified as Lichtheimia corymbifera; †, misidentified as Lichtheimia corymbifera; ‡, misidentified as Mucor ramosissimus; §, misidentified at the species level but correctly identified at the genus level.

MALDI-TOF MS phylogenetic analysis.

The MSP dendrogram of the 111 isolates (Fig. 1) shows unambiguous separation of different species. Our phylogenetic tree of 47 Rhizopus spp. (one isolate of R. stolonifer was not included) based on ITS sequences recognized R. arrhizus and R. microsporus lineages as two main clades. The clade of R. microsporus was divided into two secondary clades, BMU00951 and BMU02716, and the remaining strains (Fig. 2A). The MSPs were automatically clustered and identified using MALDI Biotyper OC 4.0 software, and the MALDI-TOF MS results were similar to those acquired by ITS sequence analysis (Fig. 2). An ML tree was generated from ITS sequence data for 41 isolates of Mucor spp., including M. circinelloides, M. irregularis, M. hiemalis, and M. racemosus. The phylogenetic tree formed two main clades: M. irregularis and M. hiemalis clustered in one clade, whereas M. circinelloides and M. racemosus clustered in the other clade, with a bootstrap value of 100%. The MSP dendrogram of 41 isolates of Mucor spp. was similar to the ML tree.

DISCUSSION

MALDI-TOF MS-based identification is a simple, rapid, and accurate high-throughput approach for most bacteria and fungi and is accordingly receiving much attention from microbiology laboratories (8, 22). However, use of this technology for the identification of filamentous fungi and unusual species faces unresolved challenges. Packeu et al. reported that different incubation times and sample preparation procedures in dermatophyte identification can result in very different identification rates, which shows the difficulties in the acquisition of reproducible spectra in dermatophytes (23). Buskirk et al. demonstrated that fungal pigments can inhibit MALDI-TOF MS analysis of darkly pigmented fungi (24). A few studies have shown that MALDI-TOF MS can successfully identify diverse melanized or black fungi (25, 26).

However, incomplete commercial libraries include a limited number of reference species and strains, which is the main reason for the continued difficulties in filamentous fungal identification (27). The Bruker Filamentous Fungi Library v1.0 of Bruker MALDI-TOF MS systems may be not sufficient for the identification of Talaromyces (Penicillium) marneffei, M. racemosus, and other unusual molds (16, 19, 28). Many studies have established their own proper reference databases of dermatophytes, Fusarium spp., T. marneffei, and other species to identify these fungi and found that accurate identification to the species level required expansion of the database using relative reference strains, which highlighted the deficiency of commercial databases for filamentous fungus identification (27–30).

In our study, the Bruker library correctly identified 49.5% of the 111 isolates at the species level and 66.7% at the genus level. When the library was supplemented with the BMU database, the identification rate improved significantly (at a level of α = 0.01), with 81.1 and 100% of isolates identified at the species and genus levels, respectively. Our study proves that effectiveness of a method is highly dependent upon the breadth of the MALDI-TOF MS reference database. In particular, we added the relevant MSPs of M. irregularis to the BMU database. M. irregularis is notable for causing chronic cutaneous infections in immunocompetent patients, ultimately leading to severe morbidity if left untreated (18). More than 90% of mucormycosis cases known to date are from Asia, primarily China (18). None of the 23 isolates of M. irregularis were credibly identified using the Bruker library, whereas all isolates were correctly identified at the species level after adding the relevant MSPs to the BMU database. Adding MSPs of specific pathogens to databases is thus important for correct identification in epidemic regions.

The Mucor sp. identification rate of 93.8% at the species level (except for M. circinelloides) was obtained with the combined database. Several M. circinelloides isolates were incorrectly identified as M. ramosissimus due to their very similar MSPs. Indeed, the Bruker library categorizes M. circinelloides and M. ramosissimus as a “CC9 Mucor circinelloides/ramosissimus group” (one of nine groups that MALDI-TOF MS cannot discriminate at the species level) because they have very similar MSPs.

The rate of identifying Lichtheimia spp. at the species level was 36.4% with the Bruker library, much lower than the rate reported by Schrödl et al. (17). The main reason for this difference is that the authors used their own database, which contained information on all of their tested strains. The Bruker library contains information on L. corymbifera but not L. ramosa or L. ornata, so three L. ramosa isolates and one L. ornata isolate were misidentified as L. corymbifera with the Bruker database. After two MSPs of L. ramosa were added to the database, one L. corymbifera isolate and one L. ornata isolate were misidentified as L. ramosa, with a log(score) of >2.0 (data not shown). We noted a similar phenomenon in a study by Schrödl et al. (17). These species of Lichtheimia may have similar MSPs, but further studies are needed to confirm this possibility. Only five isolates of Cunninghamella spp. were used in our study, so the identification of these fungi should be further studied.

The MSP dendrograms most likely discriminate different species, thus providing a clustering method based on protein differences (17). We used MALDI Biotyper OC 4.0 to generate an MSP dendrogram of the 111 isolates of Mucorales; this dendrogram showed a clear demarcation and separation between different species of Rhizopus, Lichtheimia, Mucor, Rhizomucor, Syncephalastrum, and Cunninghamella (Fig. 1). The interspecific discrimination of MSP dendrograms and ITS-based trees of Rhizopus and Mucor species was consistent, which is shown in Fig. 2 and 3. We also found that the MSP dendrograms were consistent with the results of a phylogenetic analysis based on DNA sequences used in a study of Lichtheimia spp. (17). At the intergeneric level, the MSP dendrogram showed that R. microsporus was closer to M. racemosus and M. circinelloides than to R. arrhizus, which was inconsistent with a phylogenetic tree (31). Research reported by Dolatabadi et al. (32) has a similar problem. At the intraspecific level, the MSP dendrograms were inconsistent with ITS-based trees. These inconsistencies indicate that MSP dendrograms at the proteome level and evolutionary trees based on ITS sequencing differ. In conclusion, MALDI-TOF is an excellent tool for interspecific differentiation, whereas MSP dendrograms based on MALDI-TOF analysis cannot be used in phylogeny.

This study confirms that MALDI-TOF MS is a convenient technique and powerful tool for identifying and discriminateing pathogenic Mucorales species. After supplementation of the Bruker library, MALDI-TOF MS was highly accurate in the identification of Mucorales and could meet the clinical need of identifying pathogenic Mucorales. However, the MALDI-TOF MS identification of M. circinelloides and certain Lichtheimia species retains some deficiencies, and we should use DNA sequencing to identify these microorganisms.