|Home | About | Journals | Submit | Contact Us | Français|
Thirty-eight isolates (including 28 isolates from patients) morphologically identified as Lichtheimia corymbifera (formerly Absidia corymbifera) were studied by sequence analysis (analysis of the internal transcribed spacer [ITS] region of the ribosomal DNA, the D1-D2 region of 28S, and a portion of the elongation factor 1α [EF-1α] gene). Phenotypic characteristics, including morphology, antifungal susceptibility, and carbohydrate assimilation, were also determined. Analysis of the three loci uncovered two well-delimited clades. The maximum sequence similarity values between isolates from both clades were 66, 95, and 93% for the ITS, 28S, and EF-1α loci, respectively, with differences in the lengths of the ITS sequences being detected (763 to 770 bp for isolates of clade 1 versus 841 to 865 bp for isolates of clade 2). Morphologically, the shapes and the sizes of the sporangiospores were significantly different among the isolates from both clades. On the basis of the molecular and morphological data, we considered isolates of clade 2 to belong to a different species named Lichtheimia ramosa because reference strains CBS 269.65 and CBS 270.65 (which initially belonged to Absidia ramosa) clustered within this clade. As neotype A. corymbifera strain CBS 429.75 belongs to clade 1, the name L. corymbifera was conserved for clade 1 isolates. Of note, the amphotericin B MICs were significantly lower for L. ramosa than for L. corymbifera (P < 0.005) but were always ≤0.5 μg/ml for both species. Among the isolates tested, the assimilation of melezitose was positive for 67% of the L. ramosa isolates and negative for all L. corymbifera isolates. In conclusion, this study reveals that two Lichtheimia species are commonly associated with mucormycosis in humans.
Mucormycosis is a life-threatening infection that occurs in immunocompromised patients, diabetic patients with ketoacidosis, and immunocompetent patients after trauma exposure to contaminated soil (7, 18). The filamentous fungi responsible for these infections belong to the Mucorales order. About 20 different species have been shown to be pathogenic for humans (4). According to a recent review (19), the species that were the most frequent encountered were Rhizopus spp., Mucor spp., and Cunninghamella spp., while Apophysomyces elegans and Absidia spp. accounted for 6% and 5% of the cases, respectively. The true frequency is, however, difficult to assess because surveys are rare and determination of the species of the Zygomycetes class by standard mycological methods remains difficult. Indeed, all the genera and species within the family Mucoraceae (the Absidia, Rhizopus, Mucor, Rhizomucor, and Apophysomyces genera) shared similar morphological characteristics (6). The precise identification to the species level often requires the specific expertise usually available only at reference laboratories. The availability of molecular tools for taxonomic and identification purposes has changed the picture. Sequencing of various DNA targets has facilitated the recognition of phylogenetic species within the Zygomycetes (27, 28) and provided tools for DNA bar coding of these fungi (22). A revision of the genus Absidia was recently performed on the basis of phylogenetic, physiological, and morphological characteristics (10). A new family (Mycocladiaceae) and the genus Mycocladus were proposed to accommodate the three species Mycocladus corymbifer (formerly Absidia corymbifera), M. blakesleeana, and M. hyalospora. More recently, it was suggested that additional nomenclatural changes were necessary, and the names Lichtheimiaceae and Lichtheimia were proposed for the family and the genus, respectively (11).
The intraspecific variability of Lichtheimia corymbifera (formerly A. corymbifera) has been poorly evaluated so far. After the analysis of a small number of clinical isolates, we recently reported that some of the isolates morphologically identified as L. corymbifera had divergent internal transcribe spacer (ITS) sequences (21). Subsequently, the use of molecular identification on a routine basis for all isolates of the Zygomycetes collected at the French National Reference Center for Mycoses and Antifungals allowed us to uncover intraspecific sequence variability among isolates morphologically identified to be L. corymbifera. To further characterize the atypical isolates, we used three different DNA targets, which allowed us to confirm that L. corymbifera is a species complex.
The 38 isolates used in this study are presented in Table Table1.1. Most of the isolates were of clinical origin (n = 28) and were mostly from immunocompromised patients (n = 16), but they were also from immunocompetent patients who became infected after injury (n = 7) or surgery (n = 2) and for whom the clinical presentations were skin lesions (with or without osteitis; n = 12) or pulmonary (n = 6), rhinocerebral (n = 2), and disseminated (n = 5) infections. The pathogenic role of the fungus in the three remaining cases was uncertain. In addition, one isolate was recovered from the lung of a chicken suspected of having pulmonary aspergillosis, and two isolates were cultured from hay in the region of Besançon, France. All isolates have been identified as L. corymbifera on the basis of morphological findings (white to greyish expanding colonies, branched mycelium, and the presence of stolons and rhizoids) and microscopy (spherical to pyriform sporangia, funnel-shaped apophyses, and smooth-walled endospores).
The remaining seven isolates were strains of L. corymbifera obtained from international culture collections (Centraalbureau voor Schimmelcultures [CBS] and the Collection of Fungi from the Pasteur Institute Collection). Additionally, two isolates belonging to other Lichtheimia species (L. blakesleeana CBS 100.28 and L. hyalospora CBS 173.67) were used for DNA sequence analysis. All isolates were stored as spore suspensions at −20°C in 40% glycerol. All isolates were subcultured for 3 to 6 days on 2% malt extract agar (MEA) at 30°C for macroscopic and microscopic examination.
Mycelium was grown in 20 ml of RPMI 1640 medium with l-glutamine but without sodium bicarbonate (Sigma-Aldrich, Saint Quentin Fallavier, France) buffered to pH 7.0 with 0.165 M morpholinepropanesulfonic acid (Sigma-Aldrich). After 48 h of continuous agitation (100 rpm) at 30°C, the mycelium was recovered, washed twice with a 0.9% NaCl solution, and stored at −20°C until extraction.
Genomic DNA extraction was performed as described previously (22) with approximately 200 mg of mycelium, and the DNA was stored at −20°C. The complete ITS1-5.8S-ITS2 region of the ribosomal DNA (rDNA) was amplified with primer pair V9D (5′-TTAAGTCCCTGCCCTTTGTA-3′) and LS266 (5′-GCATTCCCAAACAACTCGACTC-3′) (9). The D1-D2 region of the large-subunit rDNA was amplified with primer pair NL-1 (5′-GCATATCAATAAGCGGAGGAAAAG-3′) and NL-4 (5′-GGTCCGTGTTTCAAGACGG-3′) (14). A small region of the 5′ elongation factor 1α (EF-1α) nuclear gene was amplified with primers MEF-11 (5′-AAGAAGATTGGTTTCAACCC-3′) and MEF-41 (5′-GCACCGATTTGACCAGGRTGG-3′) (17).
The PCR amplification of ITS and 28S was done as described previously (22) in an iCycler thermocycler (Bio-Rad, Hercules, CA). For the amplification with primers MEF-11 and MEF-41, the PCR mixture (50 μl) contained 3 μl of the extracted genomic DNA, 1× PCR buffer (Roche Diagnostics GmbH, Mannheim, Germany), 3 mM MgCl2, 0.25 μM of each primer, 0.25 mM of each deoxynucleoside triphosphate (Roche), and 1.25 U of AmpliTaq DNA polymerase (Roche). The PCR conditions were predenaturation at 94°C for 5 min; 40 cycles at 95°C for 30 s, 52°C for 30 s, and 72°C for 1 min; and a final incubation at 72°C for 7 min. The PCR products were then sequenced at the Institut Pasteur sequencing facility by using a BigDye Terminator (version 1.1) kit (Applied Biosystems, Foster City, CA) and each of the primer pairs used for amplification on an ABI Prism 3730 XL DNA analyzer (Applied Biosystems).
A consensus sequence was computed from the forward and reverse sequences by using the ChromasPro program (version 1.33; Technelysium, Helensvale, Queensland, Australia), and multiple-sequence alignments were performed with the Clustal W program (26). The analysis treated gaps (indels) as a fifth state character. To determine the percentage of identical residues between each pair of sequences, identity matrices for each set of data were generated with BioEdit software (Isis Therapeutics, Carlsbad, CA). The percent similarity represents the number of identical sites divided by the length of the longest sequence (sites at which a gap was present in both sequences were removed). Single-locus cladograms were constructed by the neighbor-joining method with the pairwise-deletion option (20) in the MEGA (version 3.1) computer program (13). A combined three-locus analysis was also performed. Rhizomucor pusillus (CNRMA/F09-7) was chosen as the outgroup, and the robustness of the branches was assessed by bootstrap analysis with 1,000 replicates.
Carbon source assimilation profiles were determined with a commercial kit (ID32C system; bioMérieux, Marcy, l'Etoile, France), as described previously (23). Briefly, isolates were cultured for 7 days on Sabouraud agar slants at 30°C to obtain sufficient sporulation. The spores were transferred to API C medium (bioMérieux) to achieve a final concentration of 5 × 105 spores/ml, and 135 μl was distributed into each well. The results were read visually after 72 h of incubation at 30°C. Weak growth was considered positive. A functional analysis by use of an agglomerative clustering method (by use of the unweighted-pair group method with arithmetic mean algorithm) was performed with BioloMICS (Biological Manager for Identification, Classification and Statistics) software (version 7.2.5; BioAware, Hannut, Belgium) to group the isolates and the carbon assimilation results at the same time.
All isolates were subcultured on Sabouraud dextrose agar (supplemented with 0.02% chloramphenicol) prior to testing to ensure purity and viability. Pure powders of known potency of amphotericin B (Sigma-Aldrich), voriconazole (Pfizer Central Research, Sandwich, United Kingdom), itraconazole (Janssen-Cilag, Issy-les-Moulineaux, France), posaconazole (Schering-Plough Research Institute, Kenilworth, NJ), flucytosine (Sigma-Aldrich), terbinafine (Novartis Pharma AG, Basel, Switzerland), caspofungin (Merck & Co., Inc., Rahway, NJ), and micafungin (Astellas Pharma, Osaka, Japan) were used. In vitro susceptibility was determined by a broth microdilution technique, according to the guidelines of the Antifungal Susceptibility Testing Subcommittee of the European Committee on Antibiotic Susceptibility Testing for the testing of conidium-forming molds (24), but with some modifications. Briefly, microplates containing the eight antifungal drugs were prepared in batches and stored frozen at −20°C. The final concentrations were 0.125 to 64 mg/liter for flucytosine and 0.015 to 8 mg/liter for all other drugs. Testing was performed in RPMI 1640 medium supplemented with 2% glucose for all drugs except amphotericin B, which was tested in AM3 medium, with a final inoculum size of 105 CFU/ml. MIC endpoints were determined on an automated microplate reader spectrophotometer (Multiscan RC-351; Labsystems Oy, Helsinki, Finland) after 24 h or 48 h of incubation (an optical density of >0.15 was required for the drug-free control wells) at 35°C. The MIC endpoint was defined as a reduction in growth of 80% or more compared to the amount of growth in the drug-free well for all drugs except amphotericin B, for which an endpoint of a 90% reduction was used. Two reference strains, Candida krusei ATCC 6258 and C. parapsilosis ATCC 22019, were included in each set of determinations to ensure quality control.
A detailed morphological study was performed with 11 isolates (5 isolates randomly chosen from each clade [see below] plus isolate CNRMA/F05-100). Isolates were cultured on MEA at 30°C, and the macroscopic morphology was described after 3 to 4 days of incubation. Microscopic examination was done with cultures grown for 5 to 9 days after they were mounted in water with 1% gelatin. The different structures (sporangia, columellae, and sporangiospores) were examined. The sporangiospores were measured with a DM LB2 optical microscope (Leica Microsystèmes SAS, Rueil-Malmaison, France) with interferential contrast and a Leica D5000 microscope coupled with Leica Application Suite software, which comprises the Multifocus and the Interactive Measurement modules (precision, 0.01 μm). For each isolate, approximately 100 sporangiospores were measured, and the ratio between the length and the width was calculated.
The distributions of the MICs were compared by a nonparametric test (Mann-Whitney). The mean spore length, width, and length/width ratio were calculated for the L. corymbifera isolates (n = 613) and the L. ramosa isolates (n = 601) and were compared by an unpaired t test. Analyses were performed with Prism (version 3.00) software for Windows (GraphPad Software, San Diego, CA). Statistical significance was defined as a P value of ≤0.05.
The sequences of the whole ITS1-5.8S-ITS2 region, the D1-D2 domain of 28S, and a partial region of the EF-1α gene were determined for the 38 isolates (total length, approximately 68,000 bp). For the ITS locus, the sequences (starting at the ITS1 primer position and ending at the ITS4 primer position) ranged from 741 to 865 nucleotides in length: 613 to 617 nucleotides for the 28S D1-D2 domain and 439 nucleotides for the EF-1α locus. Two well-delimited clades were obtained with all of the single-locus distance trees generated (Fig. (Fig.11 to to3)3) and with the tree obtained when the three loci were combined (Fig. (Fig.4).4). As all individual isolates were grouped together in one clade for each locus, it was possible to consider the two clades two different species, in accordance with the principles of the genealogical concordance of phylogenetic species recognition (25).
Analysis of the ITS data matrix revealed a high degree of nucleotide sequence similarity (more than 98%) within clade 1, with the exception of that for isolate CNRMA/F05-100, which showed nucleotide sequence differences of more than 20% with the sequences of the other clade 1 isolates. Within clade 2, the maximum difference was observed between the subgroup consisting of isolates CNRMA/F02-8 and CNRMA/F04-35 and the rest of the clade 2 isolates (91.5 to 93.5% similarity). The sizes of the ITS sequences differed between the two clades (763 to 770 bp and 841 to 865 bp for clade 1 and clade 2, respectively). Greater than 99% similarity within the clade 1 sequences and only small variations (~2%) within the clade 2 sequences were observed when the 28S domain sequences were analyzed, and differences of less than 2% were observed within each clade when the EF-1α locus was analyzed (Table (Table2).2). The highest degree of sequence variability between clade 1 and clade 2 was observed for the ITS locus (34%, 5%, and 7% variability for the ITS, 28S, and EF-1α gene regions, respectively).
Clade 1 corresponded to Lichtheimia corymbifera because it included neotype strain Absidia corymbifera CBS 429.75. Clade 2 isolates were designated L. ramosa because reference isolates CBS 269.65 and CBS 270.65 (which initially belonged to the species Absidia ramosa) (8) clustered within this clade.
After 3 to 4 days of incubation on MEA, colonies of all isolates were expanding but differences in the growth patterns were observed. The L. corymbifera isolates exhibited compact growth, while the L. ramosa isolates had a more effuse mycelium. No significant differences in the morphologies of the sporangia and columellae or the branching patterns of the sporangiophores were observed between the two species (Fig. (Fig.5).5). The sporangiospores of the L. corymbifera isolates were smooth and hyaline, whereas those of the L. ramosa isolates were smooth but slightly colored. More importantly, the sporangiospores of the L. corymbifera isolates were ellipsoid (2.73 by 2.24 μm), while those of the L. ramosa isolates were long ellipsoid (3.06 by 2.18 μm) (Fig. (Fig.5),5), with significant (P < 0.0001) differences in terms of length, width, and the length/width ratio (1.41 versus 1.22, respectively) being observed.
Melezitose and palatinose were assimilated by 67% and 33% of the L. ramosa isolates, respectively, while none of the L. corymbifera isolates tested assimilated those two carbon sources (Table (Table3).3). There were no additional differences in carbon source assimilation profiles that could discriminate between the two species.
The susceptibilities of the clinical isolates to eight antifungal drugs were determined. All isolates exhibited high flucytosine MICs (>64 μg/ml), caspofungin MICs (>8 μg/ml), and micafungin MICs (>8 μg/ml); and all but one isolate had a high voriconazole MIC (>8 μg/ml). Differences in the itraconazole MICs (range, 0.25 to 16 μg/ml), posaconazole MICs (range, 0.125 to 4 μg/ml), and terbinafine MICs (range, 0.125 to 2 μg/ml) were observed among the isolates; but there were no significant differences by species. A significant difference in the amphotericin B MIC distribution was observed between the two species (0.125 to 0.5 μg/ml for the L. corymbifera isolates versus 0.03 to 0.25 μg/ml for the L. ramosa isolates; P < 0.005). It should be noted, however, that the MIC50 differed by only 2 log2 dilutions.
Finally, there was no difference between the two species in terms of the underlying diseases of the patients from whom they were recovered (hematological malignancies, solid cancer, organ transplantation, or a lack of immunosuppression) or the clinical presentations that they caused (cutaneous, pulmonary, and disseminated infections).
The main molecular and phenotypic characteristics that differentiated the L. ramosa isolates from the L. corymbifera isolates are presented in Table Table33.
Recently, a revision of the genus Absidia on the basis of the phylogenetic, physiological, and morphological characteristics of 16 species was conducted (10), and nomenclatural changes were proposed (11). The three thermotolerant Absidia species (A. corymbifera, A. blakesleeana, and A. hyalospora) are now classified in the genus Lichtheimia. L. corymbifera was the only species pathogenic for humans. Although L. corymbifera is reported to be responsible for only 5% of the human cases of zygomycosis (19), this figure should be considered with caution because of a lack of surveys and because identifications are mostly based on morphology (12). The use of molecular identification (2) will be important for an accurate assessment of the epidemiology.
Phylogenetic species recognition in the Mucorales order is performed by sequencing rDNA genes (18S, 28S, and ITS), as well as the actin and EF-1α genes (10, 17, 27, 28). For identification (DNA bar coding) of this group of fungi, ITS is a good molecular target (22). The recent guidelines published by the Clinical and Laboratory Standards Institute (3) recommend the use of ITS sequencing as a first-line method for the identification of species within the Mucorales, an approach that was further approved by another international consortium of experts (1). Our routine use of ITS sequencing for the molecular identification of filamentous fungi allowed us to notice that some isolates morphologically identified as L. corymbifera had divergent ITS sequences (21). To further characterize these isolates, two other loci (28S and EF-1α) were sequenced for all the isolates initially identified as L. corymbifera. On the basis of those data, the morphospecies L. corymbifera appeared to be a species complex that included at least two clades. Due to the low level of sequence similarity (maximums, 66, 95, and 93% for ITS, 28S, and EF-1α, respectively) between the two clades and because individual isolates clustered in the same clade for each of the three loci, clade 2 isolates represent a separate species within the L. corymbifera complex that we named L. ramosa. It is noteworthy that the sequences of these three loci are more diverse within L. ramosa than within L. corymbifera. This heterogeneity should be further confirmed by analysis of additional L. ramosa isolates. One isolate (CNRMA F05-100) had sequences divergent from the sequences of both L. corymbifera and L. ramosa and was thus not assigned to either of those two species.
To briefly summarize the complex nomenclatural history of the species L. ramosa (A. ramosa) (15), in 1886, Lindt described this species in the genus Mucor Micheli 1729. In 1890, Zopf placed the species in the genus Rhizopus Ehrenberg 1820. In 1903, Vuillemin described the genus Lichtheimia, which comprised the type species L. corymbifera (29) and L. ramosa. In 1908, however, Lendner placed both species in the genus Absidia Van Thieghem 1876. Despite the morphological differences underlined by Ellis and Hesseltine (8), subsequent studies proved this distinction difficult and A. ramosa was reduced to being synonymous with A. corymbifera (16).
Both species infect humans and cannot be differentiated in terms of the hosts that they infect or the types of disease that they cause. L. corymbifera and L. ramosa are very similar both macroscopically and microscopically, but some differences that delineate the two species were uncovered. First, by culturing isolates on MEA plates at 30°C for 3 to 4 days, compact growth characterizes L. corymbifera, while more effuse growth is suggestive of L. ramosa. The sporangiospores of L. corymbifera are smooth, hyaline, and ellipsoidal when they are mature, while those of L. ramosa are smooth, lightly colored, and more ellipsoidal, a finding consistent with the earlier description by Ellis and Hesseltine (8). Carbohydrate assimilation can be used for Zygomycetes identification (23) but was not useful for the differentiation of L. corymbifera from L. ramosa. Indeed, only if palatinose or melezitose assimilation were positive could we suspect the species to be L. ramosa. Likewise, the antifungal susceptibility profiles were undistinguishable between the two Lichtheimia species, whereas they can be used to distinguish some species within other genera (5).
The results of the present study clearly show that molecular, biological, and morphological characteristics support the separation of the two species, even if their detection by classical methods remains difficult. In conclusion, L. ramosa represents a species distinct from L. corymbifera and is thus another Lichtheimia species responsible for mucormycosis in humans.
We are grateful to Monique Coutanson and Bernard Papierok from the Pasteur Institute Collection of Fungi for providing some of the reference strains. Other clinical isolates were studied as part as a nationwide survey of invasive mycosis in France. Members of the French Mycoses Study Group who sent isolates used in this study were as follows (in alphabetical order by city in France): C. Duhamel (Caen), D. Pons (Clermont-Ferrand), E. Forget (Clichy), F. Dalle (Dijon), B. Sendid (Lille), F. de Monbrison (Lyon), F. Gay-Andrieu (Nantes), M. Gari-Toussaint (Nice), C. Lacroix (Paris), C. Kauffmann-Lacroix (Poitiers), D. Toubas (Reims), P. Cahen (Suresnes), and F. Benaoudia (Troyes). We are also grateful to Saad J. Taj-Aldeen (Doha, Qatar) for sharing some clinical isolates and to Gabriel Reboux for providing two environmental isolates. We thank Laure Diancourt and Coralie Tran from the Institut Pasteur sequencing program for technical help.
We thank the Institut Pasteur sequencing program for financial support (Genopole PF-8).
Published ahead of print on 16 September 2009.