Fungal Diversity and Evaluation of Ochratoxin A Content of Coffee from Three Cameroonian Regions

The present study had the objective to assess the ochratoxin A content of coﬀee through chromatographic analysis and design a method using PCR-DGGE to analyze at the same moment the totality of fungal ﬂora present in the coﬀee samples in order to determine their geographic origin. 96 samples of coﬀee were collected from the west region (Bafoussam and Dschang), centre region (Baﬁa), and east region (Batouri) of Cameroon during two years (2017 and 2018). Two treatments (dry and wet routes) were evaluated at three diﬀerent steps of coﬀee processing (parchment coﬀee, green coﬀee, and husk coﬀee). The characterization of the fungal proﬁle was done with PCR-DGGE and sequencing. The levels of OTA were assessed using HPLC analysis. The results indicated that the toxinogenic mycoﬂora associated with coﬀee beans was mainly Aspergillus niger , A. carbonarius, and A. ochraceus . PCR-DGGE data revealed that each sampling site is characterized by a speciﬁc fungal proﬁle. Despite the inﬂuence of the treatment on the fungal population of coﬀee, bands common to samples coming from the same site were observed. These bands could therefore constitute potential biological markers to trace back to the origin of coﬀee. OTA was detected in most of the coﬀee samples analyzed and only few samples contented OTA at levels higher than the maximum tolerable limit for food intended for human consumption. The OTA content of coﬀee was signiﬁcantly inﬂuenced by the sampling step and the sampling period.


Introduction
Coffee is one of the raw materials mostly exchanged through the world, which occupies the first place in terms of volume and values in the culture of agricultural rent products [1]. Brazil is the largest producer followed by Vietnam and Colombia. Cameroon is among the African countries having their own part on that coffee market [2]. Even if the coffee branch is very flourishing, some impediments can be noticed to its development such as a fall in world prices [1], the fall of incomes per producer, dwindling of arable lands, weak production yields, and difficulties of farmers to appropriate modern production techniques [1]. Moreover, coffee has an intrinsic potential quality which strongly depends on the variety of trees planted, the culture conditions and seasons, the geographic origin, and the agronomic know-how of farmers. But that potential can be valorized or degraded in opposite during the postharvest operations [3,4]. e processing technology of coffee beans to green coffee is too long and requires a primary process of handling. In association to this, there is treatment through dry and/or wet routes as well as storage and transportation [5,6]. ese postharvest treatments insured by the producers and for which fermentation is the most important step, participate and generally print the organoleptic and merchantable qualities of products intended for the processing [7]. Hence, a worse postharvest treatment of coffee beans can lead to the apparition of some spoilage traits on the products (mouldy, decay, and germination) as the result of microorganisms' activities (bacteria, moulds, etc.) [1,8]. ese will have an impact on the merchantable quality of the product leading to considerable losses, mostly economic ones [9].
Furthermore, in the international trade of coffee, one of the indicators of the safety security required for the product is the absence of OTA. OTA is a toxic secondary metabolite derived from some moulds, especially those that belonged to the Aspergillus genus in tropical areas and Penicillium genus in tempered areas [10,11]. Even if several studies demonstrated the relationship between the frequency and the toxinogen potential of moulds in relation to the postharvest treatments of coffee, there is still a lack of information regarding the mechanism of coffee contamination by mycotoxin-producing strains of moulds [12]. OTA production can occur during the whole period of the postharvest process, the storage, and the transportation of coffee [13]. e presence of OTA in coffee is responsible for its main rejects on world markets. In the European Union which is the main destination of coffee coming from Cameroon, OTA is together with aflatoxin B 1 (AFB 1 ), the most monitored and controlled mycotoxin. In fact, the absence of OTA is required because that metabolite is very harmful for humans and animals as it possesses nephrotoxic, immunotoxic, teratogenic, and carcinogenic effects [14][15][16][17][18]. e serious hazards that OTA generated on the consumers' health have led the safety services of the European Union to establish norms that define the maximal tolerable limits of OTA in many agricultural commodities (cereals, grape, coffee, etc.). In the regulation no. 1881/2006 of the European Commission, the maximal tolerable limit of OTA is 5 μg/kg of coffee beans [19]. According to the scientific comity of human food of the FAO and WHO [20], it is prudent to reduce the OTA exposition below 8 μg/kg of body weight per day. It therefore needs to abide by the current regulation. Hence, careful consideration should be given to OTA. e objective of this study is to assess the OTA content of coffee from three Cameroonian regions and identify the fungal flora associated with its contamination.

Sampling.
e sampling of coffee was performed during the month of May of the years 2017 and 2018 in three regions of Cameroon (west, centre, and east). ese regions were chosen based on their rate of production. e half production of Cameroonian coffee comes from these regions. Moreover, the important distance between these agroecological zones was also used as selection criteria. In the west region, sampling was done at Bafoussam (5°28′0, 72″N-10°25′0, 48″E) and Dschang (5°27″N-10°04′00″E), at Bafia (4°45′00″N-11°13′00″E) in the centre region, and at Batouri (4°26′00″N -14°22′00″E) in the east region. Bafoussam and Dschang are very close but represent the sites where 75% of the arabica coffee is produced in Cameroon. 48 samples of 500 g each were collected in coffee bags of 50 kg (100 g at five different points of the bag) each year leading to a total of 96 samples for the two years. e description of the sampling process is presented in Table 1. ree producers in each region were selected during sampling and labeled sites numbers 1, 2, and 3, respectively. e sample collection was done at different steps of the coffee processing as shown in Figure 1. ese steps were dried cherries (I), husk coffee (II), and green coffee beans (III). ey were chosen because only coffee is sold in dry form and previous studies reported that the OTA contamination problems of coffee are linked to the dry cherries, the dry husk of coffee, and the dry green coffee beans [9,13]. Two types of postharvest treatments were applied to the coffee: the treatment through wet route for arabica coffee (well ripped cherries) and the treatment through dry route for robusta and arabica coffee (immature or damaged cherries). Figure 1 resumes the different processes of coffee treatment as well as the different steps of sample collection.

Direct Analyses of Microbial Communities by Molecular Biology Technics.
e fungal flora of the different samples of coffee was screened through their direct analysis by PCR-DGGE. is method was chosen because, according to the literature, it is a rapid method for analyses of the microbial communities on food [21]. e efficiency (number of species identified) and the rapidity of the PCR-DGGE method were reported by Laforgue et al. [22] and Nganou et al. [23] while assessing the fungal profile of coffee.

Extraction of DNA from Coffee
Beans. In aseptic conditions, 8 g of coffee bean samples were introduced into sterile tubes of 50 mL and 10 mL of sterile peptone water were added. e tubes containing the mixture were placed for 1 h on a rotating wheel. e supernatants (2 mL) were taken and transferred into Eppendorf tubes followed by the addition of 0.3 g of glass beads. e protocol of El Sheikha [24] was used for DNA extraction. e purity of DNA was checked by electrophoresis and the DNA was loaded into a gel made of agarose 0.8% in 1x TAE buffer (40 mM Tris-HCl, pH 7.4, 20 mM sodium acetate, and 1.0 mM Na 2 -EDTA), which contained a molecular weight ladder. e migration was run at 100 V for 30 min and a gel red solution (50 mg/ mL, Biotium) was used to color the gels. e colored gels were observed on a UV transilluminator, photographed using a black, and white CCD camera (Scion Company) and data acquisition was performed using Gel Smart 7.3 system software (Clara Vision).

PCR Amplification.
Universal eukaryotic primer for PCR was successfully applied to amplify and identify several species of fungi in a unique PCR step. In previous reported studies, a fragment of a specific region of the 28S rDNA, called the D2 region, was amplified while using the following universal eukaryotic primers: forward, U1f GC (5′-CGC CCG CCG CGC GCG GCG GGC GGG GCG GGG GTG AAA TTG TTG AAA GGG AA-3′) and reverse, U2r (5′GAC TCC TTG GTC CGT GTT-3′) [24][25][26][27]. In the present study, after the addition a 30 bp GC clamp (Sigma) on the forward universal eukaryotic primer procured from Sigma, PCR was carried out in a final volume of 50 μL containing 1.5 mM of MgCl 2 , 5 μL of MgCl 2 -free 10x Taq reaction buffer (Promega), all the deoxyribonucleotide triphosphate (dNTPs) at 200 μM, 0.2 μM of each primer, 2 μL of extracted DNA (≈30 ng), and 1.25 Units of Taq DNA polymerase (Promega). e amplification conditions were an initial denaturation for 3 min at 94°C, 30 cycles of heating at 94°C for 45 sec, cooling at 50°C for 50 sec, heating at 72°C for 90 sec, and a final extension of 5 min at 72°C. To verify the PCR products, an electrophoresis was performed by loading 5 μL of these products on 2% TAE agarose gels with a 100 pb molecular weight ladder. Gels staining and photography were carried out as mentioned above.

DGGE Analysis.
A Bio-Rad D code universal mutation detection system (Bio-Rad, USA) was used to analyze the PCR products by DGGE. is analytical procedure was firstly described by El Sheikha [24]. Briefly, samples with approximately equal quantities of PCR amplicons were loaded into gels. ese gels were made of 8% w/v polyacrylamide (acrylamide: N,N′-methylene bisacrylamide, 37 : 1) procured from Promega, in a 1x TAE buffer (1.0 mM Na 2 -EDTA, 20 mM sodium acetate, and 40 mM Tris-HCl at pH 7.4). 60°C was set as the temperature to perform the different electrophoresis experiments and a denaturing gradient in the range of 40-70% (100% corresponded to 7 M urea and 40% v/v of formamide; Promega) was used. Electrophoresis was run at for 10 min at 20 V and then for 16 h at 80 V, and a gel red solution (50 mg/mL; Biotium) was used to stain the gels. Gels observations were performed on UV transilluminator; photographs were taken using a camera and data acquisition done with Gel Smart 7.3 system.

Purification and Identification of DGGE Bands.
e bands detected on the DGGE gel were excised with a sterile scalpel. 100 μL of TE buffer was used to elute overnight and at 4°C, the DNA of each excised band. 100 μL of each eluate was purified and sequenced at GATC Biotech (Germany) as mentioned above. e only difference was the U1 primer used which did not contain GC clamp. Sequence Scanner Software V1.0 was used to edit the raw sequence data and the edited sequences were compared to those described in the GenBank database through the Basic Local Alignment search Tool (http://www.ncbi.nlm.nih.gov/ BLAST/) [28] and the Ribosomal Database Project (http:// www.rdp.cme.msu.edu/index.jsp). Sequences having an identity percentage ≥97% were considered as belonging to the same species [29,30].

Analysis of DGGE
Gels. An Image Quant TL software V.2003 (Amersham Biosciences, USA) was used in this study to analyze the DGGE gels obtained. Two fungal reference patterns (A. niger and A. carbonarius) were used as control in order to standardize the band patterns appearing in each lane. e relative positions of the different bands were compared with standard patterns. According to cited reports, the banding pattern generated in DGGE analysis is considered as an image of the whole major fungal populations [31,32]. Hence, each individual band refers to a unique "phytotype" or "sequence type," is was also Journal of Food Quality confirmed by Kowalchuk et al. [33] in their studies. ey showed that identical sequences generally present comigrating bands in DGGE analysis. e presence or absence of comigrating bands, independently of their intensity, was used to manually score the DGGE fingerprinting. e following formula was used to calculate the dice similarity coefficient which was used later to quantify the pair wise community similarities [34]: with N a which represented the number of bands detected in sample A, N b the number of bands in sample B, and N c the number of bands common to the both samples. An interval ranging from 0 (completely dissimilar) to 100 (perfect similarity) was used to express the similarity index. In order to visualize the association between the fungal communities of the coffee and the sampling sites (cities), a factorial analysis of correspondence (FAC) was performed using the software Statistica version 7 (StatSoft, USA).

Determination of OTA Contents of the Coffee Beans
Samples. Frozen samples of coffee beans (−80°C) were ground and sieved (0.5 mm) and their OTA contents were assessed following the method described by Nakajima et al. [35]. Briefly, sample (10 g) was introduced into a phial and 100 mL of the extraction solution (methanol and solution of sodium bicarbonate 3% (50 : 50, v/v). After 30 min of extraction, the mixture was centrifuged. e collected supernatant was diluted (10 : 30, v/v) with a solution of phosphate buffered saline (pH 7.3) and filtered using an immunoaffinity column (Ocharaprep R, Rhône Diagnostics, Scotland). HPLC grade methanol (3 mL) was used to elute OTA. After flash evaporation under nitrogen stream (70°C), the residue was dissolved in 1 mL of HPLC mobile phase (water/acetonitrile/glacial acetic acid, 51 : 48 : 1) and then quantified using a Shimadzu HPLC (Shimadzu LC-10ADVP, Japan) equipped with a fluorescence detector. Elution was carried out at a flow rate of 1 mL/min. Peak corresponding to OTA was detected at a retention time between 12.3 and 12.5 min and absorption at 333 nm excitation and 460 nm emission. Calibration curve was drawn with solutions of OTA at different concentrations (0 to 50 μg/kg) prepared from an OTA standard solution (1000 ng/mL; ref P 226 R, Biopharm Rhône Ltd, Scotland). e precision, the limit quantification, and the limit of detection of the instrument were 8%, 0.15 μg/kg, and 075 μg/ kg, respectively. e recovery was 85%.

Statistical Analyses.
e OTA data obtained were submitted to analysis of variance (ANOVA) and multiple range tests using the statistical software XLSTAT 2014 (Addinsof, Inc., New York, USA). Correlation between the fungal diversity, the geographic origin, the coffee cultivar, and the processing method was performed using a comparative Z-test.

DGGE Analysis of the Representative Fungal Flora of Coffee Samples.
e DNA electrophoretic profile of coffee samples collected from the different sites (Bafoussam, Dschang, Batouri, and Bafia) is presented in Figure 2. On that gel (Figure 2), each band represents a mould and each vertical line represents a sample. e first and last wells, respectively, filled with A. niger and A. carbonarius (marker strains), were used as controls. It can be seen (Figure 2) that the DGGE profile of each site was identical to each other (e.g., A1, B1, N1, and N2) but differed from one site to another.
From the DNA electrophoretic profile (Figure 2), 19 bands were removed from the gel, purified, and submitted to sequencing in order to identify the corresponding mould strains.
e results of sequencing are noticed in Table 2. irteen different species of moulds were identified (Table 2). e presence of OTA-producing strains like A. niger, A. ochraceus, and A. carbonarius among the identified strains, suggested the possible contamination of coffee samples with OTA. is result is in accordance with the reports of Nganou et al. [23]. ese authors demonstrated in their studies the production of OTA by mould strains isolated from coffee and identified as A. niger, A. carbonarius, and A. ochraceus [23]. e distribution of strains according to the treatment applied to coffee showed that treatment through the wet route (Bafoussam and Dschang) leads to obtain a more important and abundant fungal diversity than that with dry route (Bafia and Batouri). Independent of the year of sampling and the treatment method applied to coffee, the fungal flora did not vary. However, when the sampling step of the two treatment methods is taking in consideration, a substantial reduction of the rate of the fungal diversity was noticed from Step I (O1 and O2) to Step III (C1 and D1) in the treatment through the dry route and from Step II (A1 and B1) to Step III (P1 and P2) in the treatment through the wet route. e reduction of the fungal diversity observed in Step III could be justified by the husking process which was applied to coffee in the treatment through the dry route or the washing process which was applied to coffee in the treatment through the wet route. e influence of husking and washing processes in the reduction of fungal diversity of coffee was also noticed by Duris et al. [13]. On the other hand, the great diversity of fungal flora observed in Step I in the treatment through the dry route or in Step II in the treatment through the wet route could result from the fact that, during the drying process, coffee beans are directly in contact with air and soil which could be a possible source of contamination. Moreover, climatic variation and the aeration conditions of the storage room (made of the grid disposed along the wall not far from the roof) rendered impossible the monitoring of the storage room temperature.

Variations in the Structure of Fungal Communities according to Geographical Origin of Cameroonian Coffee.
On the gel presented in Figure 3, it can be observed that the extracted DNA of the control strains showed intense and unique bands located at the expected position. is confirms that successful completion of the DGGE. For each deposit point, 6 to 11 intense and distinct bands were observed (Figure 3). A deep observation of each site (Figure 3) showed that the coffee samples coming from the same city displayed the same DGGE profile. However, the DGGE profile is different from one city to another as the number and position of bands vary with the cities (Figure 3). ese differences observed in the DGGE profiles according to the cities could be attributed to the environment of the regions where these cities are located. e type of treatment applied to coffee could also influence its fungal communities. Bands corresponding to a group of moulds common to the different sites (cities), were noticed with the different samples of coffee analyzed independent of the sampling city or the variety of coffee (Figure 3). ese bands could be indicative of moulds which are common to all samples of coffee.
In order to compare the mould communities of the coffee sampled in four different cities, factorial analysis of  correspondence was performed. Despite the limited number of bands observed that factorial analysis of correspondence was the best statistical analysis which clearly brought out the differences and similarities between the mould communities of the different cities. e two-variance design observed in Figure 4 describes the 86.9% of data variations with a good discrimination for the 4 different groups originated from 4 different cities. In each city, it can be observed that two bands appeared: I3 and J3 (Dschang), C3 and D3 (Bafoussam), M2 and M3 (Bafia), and O2 and O3 (Batouri). ese two bands correspond to two species of moulds which are specific to each site, thus suggesting that these two mould species can be used as potential markers for these cities. Similar observations were reported by El Sheikha and Nganou [8] in their studies. ese authors highlighted that the moulds DNA can be used as potential markers to insure the determination of the geographical origin of coffee. Table 3 resumes the OTA contents of the different coffee samples in relation to the sampling site, the sampling year, the treatment applied to coffee, the type of coffee, and the different sampling steps of coffee processing. OTA was detected in all samples coming from Batouri and Bafia, while for those coming from Bafoussam and Dschang, the OTA level of some samples was below the detection limit (Table 3). e highest OTA content obtained in this study (12.70 μg/kg) was noticed in the city of Batouri (Table 3). According to the European Commission regulation no. 1881/2006, the level of OTA in coffee intended for human consumption should not exceed 5 μg/kg of coffee beans [19]. In this study, the OTA content of samples collected during the year 2018 was below that value (5 μg/kg), while few samples collected during the year 2017 showed an OTA content higher than the norm.

OTA Quantification in Coffee
In order to assess the effect of the sampling year, the sampling site, the treatment applied to coffee, the type of coffee, and the different sampling steps of coffee processing on the level of OTA in the samples of coffee, analysis of variance associated with the multiple range test was performed. Results gathered from the analysis of variance revealed that only the sampling step (p � 0.0001) and the sampling year (p � 0.035) had significantly influenced the OTA content of coffee samples as their p values lower than 0.05. Castellanos-Onorio et al. [36] also showed in their studies that these two parameters significantly influenced the OTA content of coffee.
Mean values of OTA content observed in the years 2017 and 2018 were calculated and compared using the multiple range test (Table 3). In general, it can be observed that the average OTA content of coffee from Bafoussam, Dschang, Batouri, and Bafia has significantly (p < 0.05) decreased from the year 2017 to the year 2018. is decrease could be due to the fact that environmental conditions as well as storage conditions are not mastered by the producers. e variation of environmental and storage conditions as factors which significantly contributed to increase the level of OTA in coffee was reported by El Sheikha and Nganou [8].
Regarding the sampling cities (sites), mean OTA contents per city were also calculated and compared at a statistical significance of p < 0.05 using the multiple range test. It can be observed from Table 3 that, in 2017, the sample of Robusta coffee collected in Step I of the processing process and originating from Batouri has the highest OTA content (12.7 μg/kg) compared to that of all other coffee samples. is level was significantly reduced in the coffee bean collected in Step III of the processing process (0.7 μg/kg). However, in 2018, the highest OTA levels were recorded for samples of robusta coffee collected in Step I of the processing process and originating from Dschang (4.8 μg/kg) and Bafia (4.7 μg/kg). e influence of the type of coffee on the mean OTA content of the different samples coffee analyzed in the present study was also assessed. Figure 5 shows that the mean OTA content of robusta coffee is higher than that of arabica coffee.
is result means that robusta coffee could be a good substrate for the growth of OTA-producing moulds. is difference in OTA content may arise from the diversity and the load of OTA-producing fungi between the two types of coffee. In a study conducted by Bessaire et al. [37], independent of the geographic origin, storage times, and postharvest processes, robusta coffee was highly contaminated by OTA and other mycotoxins opposite to arabica coffee. A high contamination level of OTA in robusta coffee (1-27 μg/kg) compared to arabica coffee (0.6 to 5.5 μg/kg) was also noticed by Noonim et al. [38]. e authors highlighted that strains with the greatest OTA-producing potential were A. westerdijkiae for arabica coffee and A. carbonarius for robusta coffee. e contamination levels of coffee with these fungal strains were used by Noonim et al. [38] to explain the difference in OTA content of the two types of coffee. e influence of the treatment applied to coffee on its OTA contents was assessed. Generally, arabica coffee  submitted treatment through a wet route ( Figure 1). In that wet route, the unit operation of pulping led to a decrease of OTA content. e level of decrease is different according to the fact that there is a fermentation or not during the pulping. e most important decrease of OTA content obtained in this study through the mechanic process of pulping could be explained by the elimination of an external layer of coffee which is rich in toxins, whereas the fermentative route which is most wet and long in terms of time could favour the penetration (by diffusion) of the toxin into the husk. e hulling removes the toxin through a loss of a new external layer of the coffee bean. However, during the treatment of coffee through the dry route, the increase of OTA content can be explained by a synthesis of OTA by contaminant mould strains [39]. e influence of the sampling step on the OTA content of coffee samples was analyzed in this study and the results are shown in Figure 6. As observed on that figure, the samples collected at Step III (green coffee beans) were the least contaminated one followed with samples collected at Step II (Dry parchment) and finally the sample collected at Step I (Dry cherries) which showed the highest OTA content. Nganou et al. [23] also reported that the level of OTA collected at different steps of coffee processing decrease as its processing level progresses. e decrease of the OTA content when the dry cherries of coffee are processed to obtain the green coffee beans could be due to the fact that the coffee pulp is a good substrate for the growth of OTA-producing moulds. Its removing will reduce the proliferation of these OTA-producing moulds and thus leads to decrease the content of OTA. e decrease of OTA content of coffee through the elimination of their external membrane which OTA-producing moulds was notified by Frank [40] and Suarez-Quiroz et al. [41,42]. Moreover, the reduction of superficial contamination through the hulling process could also explain the decrease of OTA content of coffee samples obtained in the present study. Similar result was noticed by Nganou [9].

Conclusion
In summary, the PCR-DGGE technic used in this study enabled a rapid detection of the fungal population of coffee samples collected in three Cameroonian regions. e PCR-DGGE technic was successfully used as a traceability tool to determine the origin of the different samples of coffee. OTA (μg/kg) Figure 6: Mean OTA content of coffee samples collected at different processing steps. S1 � I (dry cherries); S2 � II (dry husk); S3 � III (green coffee beans). Mould profiles specific to each of the sampling city involved in this study were identified. e assessment of OTA content of green coffee showed that none of them contained a level higher than the maximum tolerable limit of OTA in food intended for human consumption. e strains A. niger, A. carbonarius, and A. ochraceus identified on coffee samples could be responsible for their OTA contamination. e OTA content of coffee samples was influenced by many parameters among which the sampling period and the sampling step during coffee processing were identified as the significant ones.
Data Availability e data used in this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare no conflicts of interest.