- Original Article
- Open Access
Fungal communities on alpine cheese rinds in Southern Switzerland
Botanical Studies volume 64, Article number: 6 (2023)
The biodiversity of the mycobiota of soft cheese rinds such as Brie or Camembert has been extensively studied, but scant information is available on the fungi colonizing the rinds of cheese produced in the Southern Switzerland Alps. This study aimed at exploring the fungal communities present on rinds of cheese matured in five cellars in Southern Switzerland and to evaluate their composition with regards to temperature, relative humidity, type of cheese, as well as microenvironmental and geographic factors. We used macro- and microscopical morphology, matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) mass spectrometry, and sequencing to characterize the fungal communities of the cheeses, and compared them with metabarcoding targeting the ITS region.
Isolation by serial dilution yielded 201 isolates (39 yeasts and 162 filamentous fungi) belonging to 9 fungal species. Mucor and Penicillium were dominant, with Mucor racemosus, M. lanceolatus, P. biforme, and P. chrysogenum/rubens being the most frequent species. All but two yeast isolates were identified as Debaryomyces hansenii. Metabarcoding detected 80 fungal species. Culture work and metabarcoding produced comparable results in terms of similarity of the fungal cheese rind communities in the five cellars.
Our study has shown that the mycobiota on the rinds of the cheeses studied is a comparatively species-poor community influenced by temperature, relative humidity, type of cheese, and manufacturing steps, as well as microenvironmental and possibly geographic factors.
Bloomy, natural or washed cheese rinds host a microbial community composed by bacteria, yeasts, and filamentous fungi (Wolfe et al. 2014; Irlinger et al. 2015; Dugat-Bony et al. 2016; Ceugniez et al. 2017; Fröhlich-Wyder et al. 2019; Choi et al. 2020) that contribute to the development of taste, aroma, texture, and appearance of a given cheese type. The appearance and taste, however, are the result of complex dynamic and biochemical processes that are influenced also by the cheesemaker practices and environmental conditions such as manufacturing temperatures, ripening time, and cellar temperatures (Duru et al. 2018). In addition to environmental factors, however, steps undertaken during the manufacturing process contribute to determine the organoleptic features of cheese. For instance, cheesemakers sometimes add Penicillium roqueforti spores to the milk before renneting to produce blue cheeses, or at start of ripening they spray a spore suspension of P. camemberti (= P. candidum) on the surface, to obtain the bloomy rinds of Camembert or Brie cheeses (Kindstedt 2014). The handling by the cheesemakers during the ripening phase has also been shown to be crucial for the development of the taste and aroma typical for a given cheese type (Banjara et al. 2015; Penland et al. 2021; Sun and D’Amico 2021). For cheese produced from milk of cows and goats grazing on alpine meadows in Southern Switzerland (“Formaggio d’alpe ticinese”, Protected Designation of Origin; henceforth "Ticino alpine cheese"), yeasts and filamentous fungi are never inoculated on the surface; cleaning and turning procedures most likely favor the establishment of specific cheese rind mycobiota originating from the air, the cellar walls and ground, or other hitherto not identified sources, and probably colonizing the rinds during the ripening process. The fungal biodiversity is strongly influenced by the air quality in the cheesemaking rooms, the equipment used, the salting technology, and the micro-ecological conditions in the ripening cellar.
The biodiversity of fungi growing in or on Brie or Camembert cheeses has been extensively investigated (Samson and Eckardt 1977; Nuñez 1978; Marcos et al. 1990; Lund et al. 1995; Ropars et al. 2012; Marcellino and Benson 2013; Wolfe et al. 2014; Banjara et al. 2015; Garnier et al. 2017), but to our knowledge, no studies have been published on the mycobiota colonizing rinds of Ticino alpine cheese manufactured according to the regulations of the Swiss Federal Office for Agriculture (FOAG https://www.blw.admin.ch/blw/it/home/instrumente/kennzeichnung/ursprungsbezeichungen-und-geografische-angaben.html). According to these guidelines (FOAG https://www.blw.admin.ch/blw/it/home/instrumente/kennzeichnung/ursprungsbezeichungen-und-geografische-angaben.html), Ticino alpine cheese must be produced with raw milk, and the finished cheese has to be stored in cellars on spruce or larch boards for at least 60 days before being sold, at a temperature between 10 and 16 °C and a relative humidity (rH) of at least 85%. During storage, the cheese must be cured (turned, cleaned by dry brushing, or rubbing with salt or tap water) on average once a week.
The current climate changes with raised temperatures in the ripening rooms during the summer, structural changes in the buildings that may substantially modify the relative humidity and thus alter the microclimatic conditions of the cellars, as well as the reduction of the amount of salt added to the curd before pressing, may influence the delicate balance between the typical cheese mycobiota and other halotolerant taxa, allowing unwanted taxa to colonise the rinds. In fact, some unexpected filamentous fungi (e.g., Aspergillus ochraceus, Trichothecium roseum) have been recently isolated from the rinds of Ticino alpine Cheese in some ripening cellars in which the conditions set in the FOAG guidelines were no longer met (L.E. Petrini & O. Petrini, unpublished data).
Ecological investigation of fungal communities is fundamentally changing through the widespread use of omics techniques. Traditionally, the mycobiota of cheese has been studied by fungal isolation in pure culture, and identification using morphological characters and/or molecular markers such as ITS. This method, however, has some drawbacks. For instance, slow-growing fungi are quickly overgrown by fast-growing ones, and not all fungi are culturable, making a reliable determination of species diversity, abundance, and evenness difficult.
The use of high-throughput sequencing (HTS) technologies may allow a more comprehensive characterization of microorganism communities by detecting environmental DNA (eDNA). For instance, metabarcoding of the 16S rDNA of bacteria (Quigley et al. 2012; Wolfe et al. 2014; Dugat-Bony et al. 2016; Duru et al. 2018; Murugesan et al. 2018; Choi et al. 2020; Irlinger and Monnet 2021; Penland et al. 2021) and of the internal transcribed spacer (ITS) region of yeasts and filamentous fungi has been used to characterize the microbial communities of cheeses from different geographical origins (Wolfe et al. 2014; Dugat-Bony et al. 2016; Ceugniez et al. 2017; Murugesan et al. 2018; Irlinger and Monnet 2021; Penland et al. 2021).
This study aims to provide initial information on the mycobiota present on the rinds of Ticino alpine cheeses. We studied the fungal communities of the rinds of cheese produced and ripened as set in the guidelines for “Formaggio d’alpe ticinese” [Swiss Food and Agriculture administration (FOAG)] using serial dilutions and identification of the organisms isolated in pure culture by morphological characters, matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) mass spectrometry (MS), and DNA sequencing. We also compared the classical isolation methods with metabarcoding targeting the ITS region, to assess whether this technique could be a reliable and rapid tool to be used in ecological studies of fungal communities of cheese rinds.
Materials and methods
Sites and sampling
Rind samples were collected from cheeses produced on five different alpine meadows (Table 1), selected according to their geographic position and the type of cheese produced. All cellars had a soil floor and uncured walls, with summer temperatures between 10 and 16 °C and 85% minimum rH; the salt content of the cheese ranged between 5 and 15 g/kg. Cheeses were produced using the same starter culture containing Lactobacillus casei, L. delbrueckii, Lactococcus lactis, and Streptococcus salivarius.
During the summer 2019, two HOBO Pro v2 data loggers (Onset Computer Corporation, USA) were installed in each cellar to record temperature and relative humidity (rH) to account for any differences observed by the cheesemakers at different locations in the cellars. Pictures of a typical cheese ripening cellar on a Ticino meadow, of the storage methods used, and of the rind surface of a cheese form are presented in Fig. 1.
Samples were taken from a surface of approx. 5 × 5 cm of the cheese rinds by wiping with a sterile swab soaked in 0.8% NaCl. Each swab was placed in a sterile tube for transport and subsequent storage in the laboratory.
One sample from each cellar was used for culture dependent methods and metabarcoding. The swab samples were resuspended in 0.8% NaCl and incubated for 2 h at room temperature under continuous agitation. Aliquots of 10 µl of each suspension were diluted and plated on 2% malt agar (MA; VWR, Belgium) and Dichloran (18%) glycerol agar (DG18; Lab M, United Kingdom) culture media. The plates were checked daily for fungal growth and growing fungal colonies were re-inoculated on Sabouraud Gentamicin Chloramphenicol 2 (SGC2; bioMérieux, France). If more than one species was present on the SGC2 plates, mycelium fragments were re-isolated until a pure colony was obtained. The remaining sample suspension was conserved at −20 °C until metabarcoding analysis.
Fungal identification by morphology, MALDI-TOF MS, and sequencing
Each colony was inoculated on potato dextrose medium (PDA; Sigma-Aldrich, Switzerland) for morphological identification. SGC2 is not particularly suitable for the development of morphological features of environmental fungi (Scognamiglio et al. 2010) and it was used only for MALDI-TOF MS, as it has proved to be optimal for MALDI-TOF MS analyses of filamentous fungi by reducing their sporulation, thus facilitating protein extraction (De Respinis, unpublished data).
Identification of filamentous fungi was based on macro- and microscopic characters and growth characteristics on PDA. Morphological identification, if needed, was complemented by sequencing, which was carried out only for those filamentous fungi (e.g., Penicillium spp.) that could not be confidently identified by their morphological features. DNA was extracted using the phenol–chloroform method (Ausubel et al. 1989) with a 10 mg/ml lysozyme pre-treatment. The ITS region of ribosomal DNA (White et al. 1990) was sequenced for most isolates. We used the β-tubulin partial gene (Glass and Donaldson 1995) for a more accurate identification of Penicillium species, because the ITS sequences are not diverse enough to satisfactorily delimit species of Penicillium, while the β-tubulin gene exhibits an appropriate level of divergence among species of this genus (Geiser et al. 2000; Giraud et al. 2010).
All filamentous fungi identified morphologically or by sequencing underwent also MALDI-TOF MS to confirm identification of strains belonging to taxa that were already present in the available database, and to use the generated MALDI-TOF mass spectra to expand the in-house database. At least 3–5 isolates of each species identified by morphological characters and sequencing were utilised for MALDI-TOF MS database construction. The resulting reference spectra (SuperSpectra™) were then used to confirm the identity of all filamentous fungi isolates identified by morphology but not sequenced. MALDI-TOF MS was also the main identification technique used for yeasts because the available spectra databases allow identifying the most common yeasts growing in culture (Patel 2019).
Filamentous fungi were submitted to protein extraction before MALDI-TOF MS as previously described (De Respinis et al. 2014), and yeast isolates underwent a pretreatment with 1 µL 70% formic acid, added to the target plate after direct spotting of the yeast colonies. Samples were spotted in quadruplicate on a MALDI-TOF MS target plate (bioMérieux, France), covered with 1 µl of α-cyano-4-hydroxy-cinnamic acid matrix (CHCA; bioMérieux, France), and air dried. Spectrum acquisition was performed with a VITEK MS RUO mass spectrometer (AXIMA Confidence; bioMérieux, France) equipped with a 50 Hz nitrogen laser (pulse of 3 ns). Mass spectra were collected in positive linear mode in the range of 3000–20,000 mass-to-charge ratio (m/z) with delayed, positive ion extraction (delay time of 104 ns with a scale factor of 800), and an acceleration voltage of 20 kV. For each analysis, 100 averaged profile spectra were collected and processed using the MALDI MS Launchpad 2.9.3 software (bioMérieux, France) with baseline correction, peak filtering, and smoothing procedures.
MALDI-TOF MS identification was performed by comparing the mass spectra obtained with those present in the bioMérieux (France) VITEK® MS SARAMIS® RUO database (Spectral Archiving and Microbial Identification, Version 184.108.40.206), complemented with in-house reference spectra and the MSI database for fungi available online (Normand et al. 2017). Peak lists (m/z 3000 to 20,000) were imported into the VITEK® MS SARAMIS® RUO software and used to produce a dendrogram using a bioMérieux proprietary single linkage agglomerative cluster analysis (0.08% error).
The MinION™ sequencer [Oxford Nanopore Technologies (ONT), UK] generates long amplicon reads and it has been already used to study fungal communities (Mafune et al. 2020).
DNA extraction and PCR amplification
DNA was extracted from 200 µl of a resuspended sample of each cellar using the FastDNA spin kit for soil (MP Biomedicals, USA), according to the manufacturer’s instructions.
The 2 × KAPA HiFi Hot Start Ready Mix (Roche, Switzerland) was used for PCR, following the manufacturer’s instructions. The amplicons were generated with tailed primers (ONT, UK), to allow further attachment of barcodes. The ITS region was amplified with tailed ITS1 and ITS4 primers (White et al. 1990). The PCR conditions for the ITS region were as follows: first denaturation of 3 min at 95 °C, 35 cycles of denaturation of 20 s at 98 °C, annealing of 15 s at 55 °C, elongation of 30 s at 72 °C, and a last elongation of 1 min at 72 °C.
MinION library preparation and multiplexed nanopore sequencing
The amplicons were purified with CleanNGS magnetic beads (Labgene Scientific, Switzerland) and quantified with the Qubit 4 system (Thermo Fisher Scientific, USA). Twenty-four ng (2 pmol) of purified amplicon DNA were used for the barcoding procedure, using barcodes from the PCR Barcoding Expansion 1–96 kit (ONT, UK), and the amplification was carried out with the 2 × KAPA HiFi Hot Start Ready Mix kit (Roche, Switzerland). The cycling conditions were: first denaturation of 3 min at 95 °C, followed by 13 cycles of denaturation of 20 s at 95 °C, annealing of 15 s at 62 °C, elongation of 30 s at 72 °C, and a last elongation of 5 min at 72 °C. The barcoded DNA samples were subjected to a second step of magnetic beads purification and quantification, using the same methods as above, and combined in equal ratios to generate a pool.
One µg of pooled barcoded libraries was finalised using the Ligation Sequencing kit (ONT, UK) and the NEBNext Companion Module for Oxford Nanopore Technologies Ligation Sequencing (New England Biolabs, USA). After a last step of magnetic beads purification and quantification, the libraries were loaded onto a R9.4 flow cell (ONT, UK) using the Flow Cell Priming Kit (ONT, UK).
Long-read sequencing was performed with the MinION™ Mk1B sequencer and the Minknow software (21.02.1). Raw fast5 reads were base called using the ONT Guppy Basecalling software (5.0.7) and barcoded using the ONT Guppy Barcoding software (5.0.7), which produced fastq files that were subsequently checked for quality with LongQC (Fukasawa et al. 2020). The taxonomic position of the isolates was then derived analysing the resulting fastq data with the DADA2 (Callahan et al. 2016) R package (1.18), which is considered to yield a comparatively limited number of false detections. The "filterAndTrim" function was run with the arguments minLen = 1200, maxLen = 1600, multithread = TRUE, verbose = TRUE. The "dada" function was run with the parameters OMEGA_A = 1e-10, DETECT_SINGLETONS = TRUE, before taxonomic determination with the "assignTaxonomy" function with the argument tryRC = TRUE.
ITS sequences were then searched against the UNITE database (Nilsson et al. 2019; Kõljalg et al. 2020), sh_general_release_s_10.05.2021.
Species richness, defined as the number of species in a given sample and not as an α-diversity measure as described by Margalef (1958), and evenness were used to describe the ecological parameters of the culture and were mapped graphically to differentiate communities visually (Gauthier and Derome 2021). Whittaker quantile diagrams (Whittaker 1965) displaying relative species richness were also prepared for metabarcoding (read frequencies > 1%) and culture data. Percentages of isolations (for cultural work) or reads of the taxa (for metabarcoding), were also analysed by Multidimensional Scaling (MDS; classical method and L2 dissimilarity) to evaluate similarities among mycobiota from different cellars. A sensitivity analysis was also carried out to evaluate the potential effect of false detections by carrying out MDS of the metabarcoding data on the complete dataset [all operational taxonomic units (OTU) detected] and then on two reduced datasets that included only OTUs detected at frequencies of reads larger than 1% (MDS1) or 2% (MDS2) of the reads. All computations were carried out with Stata version 17 (StataCorp LLC, College Station, Texas, USA), which was used also to prepare graphical displays.
Temperature and relative humidity
In the cellars studied, the average temperatures were between 10 and 16 °C at nine out of the ten locations in which dataloggers were placed and were below 10 °C only once at location 2 of Site A (Table 2). The temperatures remained constant over time in all sites and locations, with only few individual values falling below 10 °C and none exceeding 16 °C. Average rH values ranged between 94.5 and 99.8%, and the minimum rH never dropped under 80% (Table 2).
Metabarcoding detected 80 fungal species in the samples collected (Table 3 and Additional file 1: Table S1). Fifteen of them were yeasts and the others filamentous fungi (mainly Ascomycota; see Additional file 1: Table S1). Yeasts were predominantly recorded in cellars A and B (geographically close to each other in the Blenio Valley), whereas in cellar C (located in the Maggia Valley and with cheese produced with a mixture of goat and cow raw milk) Mucoromycota accounted for 87% of the total number of reads. Cellars D and E, both in the Leventina Valley, had an approximately equal share of yeasts and filamentous fungi.
Except for cellar B, where Mucor and Penicillium were recorded in similar amounts, each cellar was characterized by the presence of one dominant genus of filamentous fungi. In cellar C, a Mucor species identified by ITS metabarcoding as M. lanceolatus was dominant (Table 3). Debaryomyces prosopidis and D. coudertii were the two most frequent yeasts in cellars A, B, D, and E, whereas only a modest number of reads for these species and for yeasts in general was recorded in cellar C. Based on ITS sequencing, however, D. coudertii and D. prosopidis cannot be distinguished from D. hansenii (Martorell et al. 2005; Nguyen et al. 2009). Debaryomyces hansenii, in fact, was identified by MALDI-TOF MS among the isolated yeasts.
Cellars A and E (Bolla e Carassina and Pontino) account for the highest species richness when metabarcoding is used (Fig. 2A), with 10 and 13 OTUs, respectively, present at frequencies > 1%. In the other three cellars fungal diversity (frequency of reads: > 1%) is lower (8 OTUs in cellar B [Bresciana], 5 in C [Campo La Torba], and 9 in D [Formazzora]), the steep plots indicating communities with high dominance of few species. The loose clusters in Fig. 2C show groupings suggesting an influence of the techniques used during ripening (Table 1) on the fungal community composition.
The geographic location of the cellars is consistent with the clustering obtained by MDS (Fig. 3A; Mardia fit measure: 0.99; variance explained by the first 2 dimensions: 98.13%), with cellar C (hosting cheese made from a mixture of goat and cow milk and situated in Maggia Valley) being farther separated from the other cellars located in Blenio (cellars A and B), Bedretto (cellar D), and Leventina (cellar E) valleys. D and E, despite being geographically located in two different valleys, are geographically close (approx. 15 km). The sensitivity analysis (Additional file 1: Fig. S1) provided an almost identical clustering for the MDS analysis with all OTUs (that very likely includes a conspicuous number of false detections) or those present at frequencies > 2%.
Results from serial dilution isolation
Using the serial dilution method, the cheese rind samples yielded 39 yeasts and 162 filamentous fungi (Table 3 and Additional file 1: Table S2). Morphological characters and ITS, β-tubulin gene, and/or translation elongation factor 1α region sequencing allowed distinguishing 9 species. Mucor spp. and Penicillium spp. were dominant in cellars B and C, whereas in the others only Penicillium spp. were frequent (Table 3 and Additional file 1: Table S2). Mucor racemosus was most common in cellar B and M. lanceolatus in cellar C. Penicillium biforme was predominant in cellars A and E, but rare in cellar C, in which P. chrysogenum/rubens (species identification uncertain) was more frequent. All but two yeast isolates were identified by MALDI-TOF MS as Debaryomyces hansenii.
The Whittaker plot (Fig. 2B) shows a course of the curves with less steep slopes indicating a more uniform species distribution in samples C (evenness: 0.59) and E (0.57). The richness values are low (5 each in cellars A and C, 3 in B, and 6 each in D and E) but these data are hard to be interpreted because of the relatively small sample size. The clustering seen in Fig. 2D confirms the observations made already for the metabarcoding data and suggests that the processing techniques during ripening (Table 1) have a direct influence on the fungal community composition. MDS (Fig. 3B) separates cellar C (Campo la Torba, cheese from mixed cow and goat milk) from all other cellars (Mardia fit measure: 0.9814; percentage explained by the first two dimensions: 98%) but the groupings are not as well defined as with the metabarcoding data. The other cellars (cheese from cow milk only) do not form any compact cluster.
The type of milk utilized for cheesemaking, the techniques used to cure the cheese during ripening, and possibly the geographic location of the cellars, influence the composition of the fungal communities on the rinds of Ticino alpine cheese. This result is only in partial agreement with the outcome of a study of 137 cheese rind samples from 10 different countries, in which similar communities of bacteria and fungi were detected in geographically distinct parts of the world (Wolfe et al. 2014). Wolfe et al. (2014), however, carried out their statistical analysis only at the generic level, whereas we used fungal species as OTUs. ITS metabarcoding generated different OTUs among, but also within genera, and the ecological analysis obtained with theses OTUs provides a more reliable information, the use of genera as OTUs drastically reducing the level of information available to assess between-sample biodiversity.
Ticino, the Southern Switzerland region in which the cellars examined are situated, is small and partitioned lengthwise by four main valleys and high mountain ranges in between, with many alpine meadows where cattle, goats, and sheep graze often together during summertime. We decided to investigate only cellars that complied with the guideline set for the “Protected Designation of Origin” label (FOAG) for Ticino alpine cheese to set a baseline for future studies. Given the apparent microclimatic homogeneity among the different cellars, as suggested by the temperature and relative humidity data collected during the cheese ripening season, we expected similar fungal communities on all Ticino alpine cheese rinds, but data show marked differences at the species and richness level among the five cellars, as highlighted by the MDS analyses (Fig. 3A, B). These differences cannot be conclusively explained: we suspect they may be caused by the type of milk (goat and cow) used for the cheese production and by the different cleaning methods applied during ripening. The geographic location (or perhaps the macro-and microecological conditions at the different locations), however, may also be determinants of the difference.
The species richness of cheese rind microbial communities is lower than that seen in other natural environments, also because of the stress conditions caused by the salty environment (Irlinger et al. 2015). In this study, the cheese rinds with the highest number of species recovered by serial dilutions originated from cellars B and D, and those analysed by metabarcoding from cellar E. Serial dilution yielded at most only 6 species (Fig. 2B), whereas metabarcoding identified 40 (with all OTUs, Additional file 1: Fig. S1). For metabarcoding as well, however, the species richness is reduced to at most 13 (the highest number of species, reported from cellar E; Fig. 2A), if only taxa with more than 1% reads are considered.
Debaryomyces was the predominant yeast genus present on all cheese rinds sampled, although it was less frequent in cellar C. It was also often reported from natural and washed rinds in a large study investigating different cheeses from Europe and North America (Wolfe et al. 2014). D. hansenii is commonly used in the cheesemaking process (Fröhlich-Wyder et al. 2019), but its richness in three of the five cellars studied cannot be explained, because it is never inoculated in the rennet or on the rinds by the local cheesemakers. D. hansenii, however, can be present in the raw milk (Quigley et al. 2013) and the favorable environmental conditions during cheesemaking (high salt content, low pH and temperature, lactate as carbon source) may allow colonisation of the cheese surface by this yeast during the early ripening phase (Irlinger et al. 2015; Quijada et al. 2020). D. hansenii was dominant in 63% of all cheese types studied by Banjara et al. (2015), and, together with Geotrichum candidum, it was detected in 12 French cheese varieties (Dugat-Bony et al. 2016). Quijada et al. (2020) also reported Debaryomyces from all rind samples they studied.
The identification of Debaryomyces species deserves a brief comment. MALDI-TOF MS identified all isolates as D. hansenii, whereas ITS metabarcoding indicated in the samples the presence of D. prosopidis and D. coudertii, but not D. hansenii. Contrarily to the MALDI-TOF MS database used in this study, however, ITS sequencing does not distinguish D. coudertii from D. hansenii or D. prosopidis (Martorell et al. 2005; Nguyen et al. 2009). We thus expect all Debaryomyces isolates to belong to D. hansenii.
As already reported in other studies on cheese mycobiota, Mucor spp. and Penicillium spp. were the most common filamentous fungi on the cheese rind samples examined, Mucor being the second most prominent genus after Penicillium. Mucor lanceolatus and M. racemosus, detected by serial dilution techniques and metabarcoding, were also among six Mucor species isolated from French cheeses (Hermet et al. 2012) and were reported from Saint-Nectaire cheese by Dugat-Bony et al. (2016). A species of Mucor, very likely M. racemosus, is considered a contaminant of alpine cheeses, with a negative impact on the taste and becoming problematic when its occurrence is chronic and persistent (Pillonel 2006). Mucor species were not mentioned in the Wolfe et al. (2014) study that analyzed bloomy, natural or washed rinds of European and North American cheeses. One can speculate that not enough humidity was present in the samples studied, because Wolfe et al. (2014) reported some Scopulariopsis species that are considered xerophilic (Domsch et al. 2007). Selected Mucor strains are used to ripen some semi-soft cheeses including Saint Nectaire, Tomme de Savoie, or Taleggio (Fox and McSweeney 2004) by producing lipases and proteases that contribute to the distinctive flavors, texture, or nutritional quality of these cheeses (Hermet et al. 2012). Mucor fuscus is very common (Petrini L.E., unpublished data) on some types of semi-soft cheeses produced in Southern Switzerland and stored in cellars that are very similar to those we studied, yet this species was not recorded from our samples, probably because their water content was not suitable for this taxon to compete with other faster growing Mucor species.
Penicillium species may enter cheese production facilities or ripening rooms as contaminants and colonise the cheese surfaces (Nielsen et al. 1998; Ropars et al. 2012; Bodinaku et al. 2019). Penicillium cultures are sometimes added to the cheese curd during the cheese making process, but none was added to any of the cheeses studied, thus they most probably originated from the manufacturing or ripening environments. Penicilllium biforme was frequently isolated by serial dilutions (Table 3 and Additional file 1: Table S2) and it was comparatively rare only in samples from cellar B. MALDI-TOF MS classified this taxon as P. camemberti, but β-tubulin sequencing identified it as P. biforme, a species considered to be distinct from P. camemberti (Giraud et al. 2010) but not distinguishable from P. fuscoglaucum with the methods used here. In a whole genome-based analysis P. biforme, P. camemberti, and P. fuscoglaucum formed separate and specific genetic clusters, with P. biforme and P. camemberti, however, being sister clades (Ropars et al. 2020). Thus, these three species represent closely related but different lineages that have evolved traits beneficial for cheese-making.
Some fungal species such as Scopulariopsis flava, S. fusca or Mammaria sp. have often been observed on Ticino alpine cheese (L.E. Petrini, unpublished data) but were not recorded during this study. Probably they emerge later in the ripening process when the cheese surface offers a drier environment, hampering fast growing Mucor and Penicillium species.
In this study we used a combination of culture-dependent and independent methods (Fröhlich-Wyder et al. 2019) because both systems are considered complementary and not contradictory or exclusive (Irlinger et al. 2015). Both produced similar results in terms of dominant taxa present in the samples and similarity of the fungal communities inhabiting the cheese rinds (Table 3, Additional file 1: Table S1). Culture-based methods are difficult to carry out in large-scale studies; they are more demanding and often require selective media. In addition, they are more expensive and time consuming compared to metabarcoding. Non-culturable species cannot be detected, and slow growing strains are easily overgrown by fast-growing ones (Banjara et al. 2015). Metabarcoding, on the other hand, is very sensitive and cost-effective, but it does not distinguish between colonizing species and occasional, dormant contaminants. In addition, metabarcoding is known to yield false negative and false positive detections of species in environmental samples (Buxton et al. 2021). Several techniques have been described to obviate, at least partly, this drawback (e.g., Ficetola et al. 2016; Lahoz-Monfort et al. 2016; Buxton et al. 2021; Polanco Fernández et al. 2021)), and some analysis tools are relatively robust with regards to false detections. In previous studies, for instance, DADA2 (used in this study) has identified more real variants and output fewer spurious sequences than other methods in studies of mock communities (Callahan et al. 2016). An overall knowledge of the mycobiota present on the rind of cheese ripening under controlled standard conditions (“the ground truth”) is needed to reduce the number of false detections, although our sensitivity analysis suggests that a selection of reads present at frequencies higher that a given cutoff (in our case 1%) may contribute to solving the false detection problem.
Our study has some limitations. Because of economic reasons and lacking manpower, the number of cellars and rind samples studied was small. The cheese rinds were sampled only once at the early-mid ripening phase (on average 30 days after production, Table 3) and metabarcoding was carried out on only one sample, precluding the possibility to detect intra-cellar variation of the mycobiota. The use of ITS metabarcoding for the mycobiome is also a limitation, as ITS does not reliably identify several fungal species and the available databases are far from being complete. On the other hand, the exact taxonomic position of the OTUs detected in an ecological study is not crucial to investigate differences in the community composition of different samples, and the sensitivity analysis carried out in our study has shown that even without the elimination of potential false detections (Additional file 1: Fig. S1) the results obtained by metabarcoding are comparable to those derived from cultural methods.
It is known that the dynamic of bacterial and fungal microbiota evolves throughout the ripening process (Quijada et al. 2020; Penland et al. 2021): future studies should foresee sampling during the early and late ripening stages to gather information on the evolution of the rind communities throughout the cheese maturation period.
The data gathered have shown that the mycobiota community on the rinds of Ticino alpine cheese depends on different parameters, including not only temperature, relative humidity, and type of milk but also manufacturing steps, microenvironmental, and possibly geographic factors.
Availability of data and materials
The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.
Ausubel F, Brent R, Kingston R et al (1989) Current protocols in molecular biology. Greene Publishing Associated & Wiley Interscience, New York
Banjara N, Suhr MJ, Hallen-Adams HE (2015) Diversity of yeast and mold species from a variety of cheese types. Curr Microbiol 70:792–800. https://doi.org/10.1007/s00284-015-0790-1
Bodinaku I, Shaffer J, Connors AB et al (2019) Rapid phenotypic and metabolomic domestication of wild penicillium molds on cheese. Mbio 10:1–16. https://doi.org/10.1128/mBio.02445-19
Buxton A, Matechou E, Griffin J et al (2021) Optimising sampling and analysis protocols in environmental DNA studies. Sci Rep 11:1–10. https://doi.org/10.1038/s41598-021-91166-7
Callahan BJ, McMurdie PJ, Rosen MJ et al (2016) DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods 13:581–583. https://doi.org/10.1038/nmeth.3869
Ceugniez A, Taminiau B, Coucheney F et al (2017) Fungal diversity of “Tomme d’Orchies” cheese during the ripening process as revealed by a metagenomic study. Int J Food Microbiol 258:89–93. https://doi.org/10.1016/j.ijfoodmicro.2017.07.015
Choi J, Lee SI, Rackerby B et al (2020) Microbial communities of a variety of cheeses and comparison between core and rind region of cheeses. J Dairy Sci 103:4026–4042. https://doi.org/10.3168/jds.2019-17455
De Respinis S, Monnin V, Girard V et al (2014) MALDI-TOF VITEK MS for the rapid and accurate identification of dermatophytes on solid cultures. J Clin Microbiol 52:4286–4292. https://doi.org/10.1128/JCM.02199-14
Domsch K, Gams W, Anderson T-H (2007) Compendium of soil fungi. IHW-Verlag, Eching
Dugat-Bony E, Garnier L, Denonfoux J et al (2016) Highlighting the microbial diversity of 12 French cheese varieties. Int J Food Microbiol 238:265–273. https://doi.org/10.1016/j.ijfoodmicro.2016.09.026
Duru IC, Laine P, Andreevskaya M et al (2018) Metagenomic and metatranscriptomic analysis of the microbial community in Swiss-type Maasdam cheese during ripening. Int J Food Microbiol 281:10–22. https://doi.org/10.1016/j.ijfoodmicro.2018.05.017
Ficetola GF, Taberlet P, Coissac E (2016) How to limit false positives in environmental DNA and metabarcoding? Mol Ecol Resour 16:604–607. https://doi.org/10.1111/1755-0998.12508
FOAG Federal Office of Agriculture. Denominazioni di origine. https://www.blw.admin.ch/blw/it/home/instrumente/kennzeichnung/ursprungsbezeichungen-und-geografische-angaben.html. Accessed 16 Jan 2023.
Fox P, McSweeney P (2004) Cheese: an overview. In: Fox P, McSweeney P, Cogan T, Guinee T (eds) Cheese, Chemistry, Physics and Microbiology, 3rd edn. Academic Press, Burlington, pp 1–18
Fröhlich-Wyder MT, Arias-Roth E, Jakob E (2019) Cheese yeasts. Yeast 36:129–141. https://doi.org/10.1002/yea.3368
Fukasawa Y, Ermini L, Wang H et al (2020) LongQC: a quality control tool for third generation sequencing long read data. G3 Genes Genomes Genet 10:1193–1196. https://doi.org/10.1534/g3.119.400864
Garnier L, Valence F, Mounier J (2017) Diversity and control of spoilage fungi in dairy products: an update. Microorganisms 5:1–33. https://doi.org/10.3390/microorganisms5030042
Gauthier J, Derome N (2021) Evenness-richness scatter plots a visual and insightful representation of Shannon entropy measurements for ecological community analysis. mSphere 6:1–10. https://doi.org/10.1128/mSphere.01019-20
Geiser D, Harbinski F, Taylor J (2000) Molecular and analytic tools for characterizing Aspergillus and Penicillium species at the intra- and interspecific levels. In: Samson R, Pitt J (eds) Intergration of modern taxonomic methods for Penicillium and Aspergillus classification. Harwood Academic Publishers, Amsterdam, pp 381–394
Giraud F, Giraud T, Aguileta G et al (2010) Microsatellite loci to recognize species for the cheese starter and contaminating strains associated with cheese manufacturing. Int J Food Microbiol 137:204–213. https://doi.org/10.1016/j.ijfoodmicro.2009.11.014
Glass N, Donaldson G (1995) Development of primer sets designed for use with the PCR to amplify conserved genes from filamentous ascomycetes. Appl Env Microbiol 61:1323–1330
Hermet A, Méheust D, Mounier J et al (2012) Molecular systematics in the genus Mucor with special regards to species encountered in cheese. Fungal Biol 116:692–705. https://doi.org/10.1016/j.funbio.2012.04.002
Irlinger F, Monnet C (2021) Temporal differences in microbial composition of Époisses cheese rinds during ripening and storage. J Dairy Sci 104:7500–7508. https://doi.org/10.3168/jds.2021-20123
Irlinger F, Layec S, Hélinck S, Dugat-Bony E (2015) Cheese rind microbial communities: diversity, composition and origin. FEMS Microbiol Lett 362:1–11. https://doi.org/10.1093/femsle/fnu015
Kindstedt P (2014) The basics of cheesemaking. In: Donnelly C (ed) Cheese and Microbes. ASM Press, Washington, pp 17–38
Kõljalg U, Nilsson HR, Schigel D et al (2020) The taxon hypothesis paradigm—on the unambiguous detection and communication of taxa. Microorganisms 8:1–24. https://doi.org/10.3390/microorganisms8121910
Lahoz-Monfort JJ, Guillera-Arroita G, Tingley R (2016) Statistical approaches to account for false-positive errors in environmental DNA samples. Mol Ecol Resour 16:673–685. https://doi.org/10.1111/1755-0998.12486
Lund F, Filtenborg O, Frisvad JC (1995) Associated mycoflora of cheese. Food Microbiol 12:173–180. https://doi.org/10.1016/S0740-0020(95)80094-8
Mafune KK, Godfrey BJ, Vogt DJ, Vogt KA (2020) A rapid approach to profiling diverse fungal communities using the MinION™ nanopore sequencer. Biotechniques 68:72–78. https://doi.org/10.2144/btn-2019-0072
Marcellino S, Benson D (2013) The good, the bad, and the ugly: tales of mold-ripened cheese. In: Donnelly C (ed) Cheese and microbes. ASM Press, Washington, pp 95–131
Marcos A, Esteban M, Alcalá M (1990) Determination of water activity in Brie and Camembert cheese varieties by four different methods. Food Chem 38:189–199. https://doi.org/10.1016/0308-8146(90)90193-8
Margalef R (1958) Information theory in ecology. General Systems 3:36–71
Martorell P, Fernández-Espinar MT, Querol A (2005) Sequence-based identification of species belonging to the genus Debaryomyces. FEMS Yeast Res 5:1157–1165. https://doi.org/10.1016/j.femsyr.2005.05.002
Murugesan S, Reyes-Mata MP, Nirmalkar K et al (2018) Profiling of bacterial and fungal communities of Mexican cheeses by high throughput DNA sequencing. Food Res Int 113:371–381. https://doi.org/10.1016/j.foodres.2018.07.023
Nguyen HV, Gaillardin C, Neuvéglise C (2009) Differentiation of Debaryomyces hansenii and Candida famata by rRNA gene intergenic spacer fingerprinting and reassessment of phylogenetic relationships among D. hansenii, C. famata, D. fabryi, C. flareri (=D. subglobosus) and D. prosopidis: description of D. vietnamensis sp. nov. closely related to D. nepalensis. FEMS Yeast Res 9:641–662. https://doi.org/10.1111/j.1567-1364.2009.00510.x
Nielsen MS, Frisvad JC, Nielsen PV (1998) Protection by fungal starters against growth and secondary metabolite production of fungal spoilers of cheese. Int J Food Microbiol 42:91–99. https://doi.org/10.1016/S0168-1605(98)00070-1
Nilsson RH, Larsson KH, Taylor AFS et al (2019) The UNITE database for molecular identification of fungi: Handling dark taxa and parallel taxonomic classifications. Nucleic Acids Res 47:D259–D264. https://doi.org/10.1093/nar/gky1022
Normand AC, Becker P, Gabriel F et al (2017) Validation of a new web application for identification of fungi by use of matrix-assisted laser desorption ionization-time of flight mass spectrometry. J Clin Microbiol 55:2661–2670. https://doi.org/10.1128/JCM.00263-17
Nuñez M (1978) Microflora of Cabrales cheese: changes during maturation. J Dairy Res 45:501–508
Patel R (2019) A Moldy application of MALDI: MALDI-ToF mass spectrometry for fungal identification. J Fungi 5:4. https://doi.org/10.3390/jof5010004
Penland M, Falentin H, Parayre S et al (2021) Linking Pélardon artisanal goat cheese microbial communities to aroma compounds during cheese-making and ripening. Int J Food Microbiol 345:109130. https://doi.org/10.1016/j.ijfoodmicro.2021.109130
Pillonel L (2006) Manuale per la trasformazione del latte sulle alpi ticinesi. Agroscope Stazione federale di ricerche per la produzione animale e lattiera (ALP), Liebefeld-Posieux (CH).
Polanco Fernández A, Marques V, Fopp F et al (2021) Comparing environmental DNA metabarcoding and underwater visual census to monitor tropical reef fishes. Environ DNA 3:142–156. https://doi.org/10.1002/edn3.140
Quigley L, O’Sullivan O, Beresford TP et al (2012) High-throughput sequencing for detection of subpopulations of bacteria not previously associated with artisanal cheeses. Appl Environ Microbiol 78:5717–5723. https://doi.org/10.1128/AEM.00918-12
Quigley L, O’Sullivan O, Stanton C et al (2013) The complex microbiota of raw milk. FEMS Microbiol Rev 37:664–698. https://doi.org/10.1111/1574-6976.12030
Quijada NM, Schmitz-Esser S, Zwirzitz B et al (2020) Austrian raw-milk hard-cheese ripening involves successional dynamics of non-inoculated bacteria and fungi. Foods 9:1–21. https://doi.org/10.3390/foods9121851
Ropars J, Cruaud C, Lacoste S, Dupont J (2012) A taxonomic and ecological overview of cheese fungi. Int J Food Microbiol 155:199–210. https://doi.org/10.1016/j.ijfoodmicro.2012.02.005
Ropars J, Didiot E, Rodríguez de la Vega RC et al (2020) Domestication of the emblematic white cheese-making fungus Penicillium camemberti and its diversification into two varieties. Curr Biol 30:4441-4453.e4. https://doi.org/10.1016/j.cub.2020.08.082
Samson R, Eckardt C (1977) The taxonomy of Penicillium species from fermented cheeses. Antonie Van Leeuwenhoek 43:341–350. https://doi.org/10.1007/BF02313761
Scognamiglio T, Zinchuk R, Gumpeni P, Larone DH (2010) Comparison of inhibitory mold agar to Sabouraud dextrose agar as a primary medium for isolation of fungi. J Clin Microbiol 48:1924–1925. https://doi.org/10.1128/JCM.01814-09
Sun L, D’Amico DJ (2021) Composition, succession, and source tracking of microbial communities throughout the traditional production of a farmstead cheese. mSystems 6:e00830-e921. https://doi.org/10.1128/msystems.00830-21
White TJ, Bruns T, Lee S, Taylor J (1990) Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: Innis MA, Gelfand DH, Sninsky JJ, White TJ (eds) PCR protocols: a guide to methods and applications. Academic Press, Cambridge, pp 315–322
Whittaker R (1965) Dominance and diversity in land plant communities. Science 147(3655):250–260
Wolfe BE, Button JE, Santarelli M, Dutton RJ (2014) Cheese rind communities provide tractable systems for in situ and in vitro studies of microbial diversity. Cell 158:422–433. https://doi.org/10.1016/j.cell.2014.05.041
We thank the computing center of the University of Geneva for providing access to the Yggdrasil HPC cluster for all computations related to metabarcoding. The collaboration of the cheesemakers of the involved alps is gratefully acknowledged. We thank an anonymous reviewer for the constructive criticisms that have led to an improved final version of the manuscript. This work is dedicated to the late Prof. Dr. Jack Rogers, a distinguished mycologist and a mentor and friend of LEP and OP.
This work was supported with a grant to SUPSI by the Section of Agriculture of the Republic and Canton of Ticino, Bellinzona (grant no. RMS19103).
Ethics approval and consent to participate
Consent for publication
The authors have no relevant financial or non-financial interests to disclose.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Additional file 1: Table S1.
Frequency (in percentage of total number of reads) of mold and yeast species identified by ITS metabarcoding. Due to the inadequacy of ITS to identify some fungal species, species names (in particular Penicillium spp.) are not reliable and should be seen only as indicative of the biodiversity observed by metabarcoding. Some species reported may also be false positive detections. Table S2. Frequency of mold and yeast species [n, (%)] isolated in pure culture. Penicillium spp. were identified using the β-tubulin gene region (Houbraken et al 2020). For P. biforme see discussion in the main text. Fig. S1. Analysis of the ecological data collected by metabarcoding. A: Whittaker plots of all OTUs, presenting the abundance of each OTU ranked from the most (rank 1) to the least frequent rank, for all cellars studied. B: Whittaker plots of the OTUs with frequency of reads>2%, presenting the abundance of each species ranked from the most (rank 1) to the least frequent. C: Results of the MDS of the data collected by metabarcoding, all OTUs included in the analysis. D: Results of the MDS of the data collected by metabarcoding, with only OTUs with frequencies >2% included in the analysis. For additional details see text.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
De Respinis, S., Caminada, A., Pianta, E. et al. Fungal communities on alpine cheese rinds in Southern Switzerland. Bot Stud 64, 6 (2023). https://doi.org/10.1186/s40529-023-00371-2
- High-throughput sequencing
- MALDI-TOF MS