Interaction and Metabolic Function of Microbiota during the Washed Processing of Coffea arabica

Coffee fermentation is crucial for flavor and aroma, as microorganisms degrade mucilage and produce metabolites. This study aimed to provide a basis for understanding the impact of microorganisms on Coffea arabica from Yunnan, China, during washed processing. The microbial community structure and differentially changed metabolites (DCMs) of C. arabica beans during washed processing were analyzed. The results indicated that the top five predominant microorganisms at the genera level were Achromobacter, Tatumella, Weissella, Streptococcus, and Trichocoleus for bacteria and Cystofilobasidium, Hanseniaspora, Lachancea, Wickerhamomyces, and Aspergillus for fungi. Meanwhile, the relative content of 115 DCMs in 36 h samples decreased significantly, compared to non-fermentation coffee samples (VIP > 1, p < 0.05, FC < 0.65), and the relative content of 28 DCMs increased significantly (VIP > 1, p < 0.05, FC > 1.5). Furthermore, 17 DCMs showed a strong positive correlation with microorganisms, and 5 DCMs had a strong negative correlation (p < 0.05, |r| > 0.6). Therefore, the interaction and metabolic function of microbiota play a key role in the formation of coffee flavor, and these results help in clarifying the fermentation mechanisms of C. arabica and in controlling and improving the quality of coffee flavor.


Introduction
Coffee is the most popular and most widely consumed non-alcoholic beverage globally because of its unique flavor and taste. Mature coffee cherries are processed postharvest using different primary processing methods to obtain green coffee beans, which are roasted and ground to prepare coffee beverages. In this process, coffee fermentation is crucial in controlling coffee quality [1]. Microorganisms spontaneously participate in coffee fermentation and help remove coffee mucilage during coffee processing. These microorganisms also produce metabolites during fermentation, which can reach the interiors of coffee seeds, affecting coffee quality [2]. Therefore, the function of microorganisms has a crucial influence on coffee quality [3].
Yeast, bacteria, and filamentous fungi are the main microorganisms involved in coffee fermentation. Previous studies reported region-specific and method-specific microbiota characteristics during coffee fermentation [4]. Microbial species were richer in the washed processing method than in the other two coffee processing methods (natural and semi-dry processing), with the semi-dry processing method having minimal bacteria. In Brazil, yeasts, including Pichia kluyveri, Candida glabrata, and Hanseniaspora opuntiae, were found to be widely identified in spontaneous coffee fermentation [5]. In Australia, Hanseniaspora uvarum, Pichia kudriavzevii, Leuconostoc mesenteroides, and Lactococcus lactis were found to be the dominant microbiota during coffee fermentation [3]. Differences in microbial richness and diversity have also been reported in samples from a single country. For example, microbial richness and diversity in northern Colombia were found to be higher than they were in southwestern Colombia [6].
Improving coffee flavor quality through fermentation with selected special microbiota and different processing methods has become very popular and effective. Species of yeastincluding Picha fermentans, P. kudriavzevii, Saccharomyces cerevisiae, and Candida parapsilosis-and bacteria-including Lactiplantibacillus plantarum, Leuconostoc mesenteroids, and Bacilus subtilis-have been used for coffee fermentation [7][8][9][10][11][12]. For example, coffee fermentation inoculation with L. mesenteroides improved the sensory score to 81, providing the coffee with dark chocolate, caramel, nutty, and spicy characteristics [7]. In addition, the cocultivation of L. mesenteroides with Leuconostoc plantarum led to a higher sensory score of 81.33 than that of a single inoculation with L. mesenteroides [7].
Yunnan province is the primary coffee plantation province in China, accounting for more than 95% of China's coffee plantation area, making coffee one of the most important economic sources in the province. Zhang et al. studied the influence of processing conditions on coffee quality and metabolomic profiles in traditional wet-processing methods and reported that Leuconostoc and Lactococcus were active in C. arabica fermentation in China [13]. This study analyzed the microbiota and metabolites of coffee beans during washed processing to further study coffee quality and the functions of microbiota.

Microbial Community Structure of Coffee Based at the Phylum and Genus Level
Coffee has great microbiological diversity [1]. The sequences of bacteria and fungi ranged from 40,037 to 81,509 and from 38,944 to 179,398, respectively, with average lengths of 376 bp for bacteria and 241 bp for fungi. The coverage of coffee samples in each group was higher than 0.99. At the operational-taxonomic-unit (OTU) level, the ace index had p-values of 0.07 and 0.35 in bacteria and fungi, respectively ( Figure 1A,E). The p-values of the Chaol index were 0.08 in bacteria and 0.35 in fungi. Among bacteria, WC0 had the highest Chaol index ( Figure 1B), while among fungi, the highest Chaol index was observed in WC1 ( Figure 1F). These results indicated that WC0 and WC1 had the highest species richness in bacteria and fungi, respectively.  Furthermore, these bacteria were confirmed to 10 genera, as shown in Figure 2B. These genera included Achromobacter, Tatumella, Weissella, Streptococcus, Trichocoleus, Burkholderia, Gluconobacter, Pseudenterobacter, Leuconostoc, and Dyella. The top five genera were Achromobacter, Tatumella, Weissella, Streptococcus, and Trichocoleus. At the beginning of the fermentation, Achromobacter and Trichocoleus were the dominant bacteria, with 12.95 and 12.88 relative percentages of community abundance at the genus level, respectively.  The Shannon indices and the Simpon indices can reflect the number, evenness, and diversity of species of coffee samples [14]. The highest Shannon indices were observed for WC0 for bacteria ( Figure 1C) and WC1 for fungi ( Figure 1G), indicating that they had the highest species evenness for bacteria and fungi, respectively. On the other hand, WC3 showed the highest Simpon indices for bacteria and fungi ( Figure 1D,H), suggesting that it had the highest observed diversity of both bacteria and fungi.
The bacteria in coffee samples from different fermentation times using the washed processing method were classified into 11 phyla by high-throughput sequencing, as shown in Figure 2A. These phyla included Proteobacteria, Firmicutes, Actinobacteria, Cyanobacteria, Bacteroidota, Acidobacteriota, Gemmatimonadota, Chloroflexi, and Myxococcota. Among them, the dominant phyla were Proteobacteria (comprising 54.36-61.60% of the community abundance at the phyla level), Firmicutes (8.57-41.25%), Actinobacteria (0.27-13.13%), Cyanobacteria (0.34-13.00%), and Bacteroidota (0.00-0.16%). Notably, the relative percentage of community abundance of Proteobacteria in all the samples was more than 50.00%. The relative percent of Proteobacteria community abundance initially increased from 57.52% in WC0 to 61.60% in WC2, then decreased to 54.36% in WC3. Moreover, Firmicutes, Actinobacteria, and Cyanobacteria exhibited significant changes with fermentation time lengthening. At the start of fermentation, WC0 showed a rich bacterial phyla composition, while at the end of fermentation, WC3 showed a more consistent bacterial phyla composition.
Furthermore, these bacteria were confirmed to 10 genera, as shown in Figure 2B. These genera included Achromobacter, Tatumella, Weissella, Streptococcus, Trichocoleus, Burkholderia, Gluconobacter, Pseudenterobacter, Leuconostoc, and Dyella. The top five genera were Achromobacter, Tatumella, Weissella, Streptococcus, and Trichocoleus. At the beginning of the fermentation, Achromobacter and Trichocoleus were the dominant bacteria, with 12.95 and 12.88 relative percentages of community abundance at the genus level, respectively. Achromobacte remained relatively stable during the first 12 h, but its abundance increased over the fermentation process to 24.69% at 24 h and 24.18% at 36 h. In contrast, Trichocoleus significantly declined as the fermentation processed, reaching its lowest abundance of 0.28% at 12 h. Burkholderia and Dyella showed changes similar to those of Trichocoleus. On the other hand, Weissella and Leuconostoc continuously increased throughout the fermentation process, from 0.15% to 33.74% and 0.00% to 6.39%, respectively. The population of Tatumella followed a similar pattern to that of Streptococcus, increasing during the first 12 h and then decreasing. The microbial community was highly diverse in coffee fermentation, especially in wet coffee fermentation [4]. Lactic acid bacteria species, especially Leuconostoc, were commonly involved in fermentation [4]. Low Weissella sp. and Streptococcus faecalis populations were also identified [4]. Streptococcus was the perdominant species during the fermentation in the wet process [4].
The fungi in different coffee samples were of lower diversity than bacteria, which could be classified into seven phyla by high-throughput sequencing, as shown in Figure 2C. These phyla were Ascomycota, Basidiomycota, Mortierellomycota, Mucoromycota, Neocallimastigomycota, Olpidiomycota, and Rozellomycota. Among them, the dominant phyla were Ascomycota (comprising 40.51-87.47% of the community abundance at phyla level) and Basidiomycota (11.80-59.30%). Notably, the relative abundance of Ascomycota in different samples was consistently higher than 40.00%. As the fermentation time increased, the relative abundance of Ascomycota first increased, reaching a maximum of 87.47% at 12 h (WC1), the decreased. On the other hand, the relative abundance of Basidomycota initially decreased, then increased, reaching a maximum of 59.30% at 36 h (WC3). Ascomycota is the most diverse and richest phylum in the kingdom of fungi, with 110,000 known species [15]. Due to their significant metabolic flexibility, Ascomycota members are widely used in various biotechnological applications, such as the production of fatty alcohols, fatty acids, and biofuels and the reduction and degradation of chemicals and solvents [15,16]. Basidiomycota is also a major phylum in the kingdom of fungi, with more than 31,000 known species, which is second only to Ascomycota in species numbers [17]. half circles mean the microbial community structure of different fermentation times, and different colors of left outer circles represent different coffee samples from fermentation times; among them, red represents WC0, blue represents WC1, green represents WC2, and yellow represents WC3, and the numbers represent the percentage of community abundance of different microorganism in these . Left: half circles mean the microbial community structure of different fermentation times, and different colors of left outer circles represent different coffee samples from fermentation times; among them, red represents WC0, blue represents WC1, green represents WC2, and yellow represents WC3, and the numbers represent the percentage of community abundance of different microorganism in these samples. Right: half circles mean the percentage of community abundance of different microorganisms on phylum or genus level, and different colors represent different microorganisms on phylum or genus level; the number represents the percentage of community abundance of these microorganisms in different samples. Furthermore, these fungi in coffee samples were confirmed to belong to 10 genera, as shown in Figure 2D. These genera were Cystofilobasidum, Hanseniaspora, Aspergillus, Thermomyces, Lachances, Wickerhamomyces, Rasamsonia, Candida, Metschnikowia, and Apiotrichum. The predominant genera were Cystofilobasidu, Hanseniaspora, Lachancea, Wickerhamomyces, and Aspergillus. The percentage of community abundance of Cystofilobasidum at the genus level ranged from 7.54% to 53.41%, with the maximum value observed at 36 h (WC3) and the minimum value at 12 h (WC2). On the other hand, the maximum value for Aspergillus was 10.76% at 12 h (WC1), and the minimum value was 0.90% at 36 h (WC3).
Wickerhamomyces exhibited a similar pattern to that of Aspergillus, with a maximum of 7.93% at 12 h (WC1) and a minimum of 3.46% at 36 h (WC3). The percentages of community abundance of Hanseniaspora were 3.45% at 0 h (WC0) and 26.80% at 24 h (WC2). For Lachancea, the values of WC1 and WC2 were nearly equivalent at about 9%, and the minimum value was 3.38% at 36 h (WC3). The diversity of fungi in wet coffee processing is often lower than in dry and semi-dry processing, mainly because of the shorter fermentation time and the submerged environment [4]. The main filamentous fungi are Aspergillus, Penicillium, Fusarium, Rhizopus, Mucor, and Cladosporium [4,18]. Yeast species such as Hanseniaspora and Candida were frequently isolated from different locations worldwide [4].
Although many microorganism species are common in coffee fermentation, some are specific to certain regions. For example, in Australia, Citrobacter and Leuconostoc are predominant genera, along with Pichia and Hanseniaspora [3]. In Brazil, Pichia, Candida, and Hanseniaspora are the most frequent isolated genera [5]. In India, Saccharomyces, Shizosaccharomyces, Bacillus, Lactobacillus, Leuconostoc, Pseudomonas, and Flavobacterium are dominant genera during initial fermentation stages [18]. In China, Enterobacter, Bacillus, Pseudomonas, Gluconobacter, Kluyvera, and Candida are dominant in the wet processing of C. arabica [13]. Based on the results of this study, Achromobacter, Tatumella, Weissella, Streptococcus, and Trichocoleus were found to be predominant in the bacterial community, while Cystofilobasidu, Hanseniaspora, Lachancea, Wickerhamomyces, and Aspergillus were predominant in the fungal community during complete washed processing.
When mature coffee cherries are harvested, a primary processing is necessary to obtain green coffee beans for the storage, transportation, and roasting of coffee [1]. There are various methods of obtaining green coffee beans in postharvest processing, including wet, dry, and semi-dry processing methods [1]. Wet processing is often used for Coffee arabica and involves submerged fermentation for 12-36 h after removing the mesocarp, resulting in coffee with a higher quality [23]. Fermentation is a natural and critical process for removing the mucilage and reducing water content [4,24] through enzymes that naturally occur in the coffee fruit and microflora acquired from the environment. During fermentation, microbial metabolites can migrate into the coffee and change various physiological parameters, such as water content, simple sugars, aroma, and other flavor precursors [4,25].

Correlation between Microorganisms and DCMs
Coffee beans contain rich precursors that can generate flavor and aroma compounds during roasting. Coffee fermentation can increase the diversity of coffee aroma and flavor compounds. Microorganisms in coffee not only produce important enzymes (such as pectin lyase, polygalacturonase, and pectin methyl esterase) for degrading pectin substances, but also produce diverse metabolites during coffee fermentation [25]. Therefore, microorganisms play a crucial role in coffee fermentation [25].
A correlation analysis between microorganisms and 23 DCMs was carried out to obtain more useful information about the function of microbes in the fermentation of coffee. Figure 6 presents a network that intuitively clarified the complex relationship between DCMs, bacteria, and fungi, in which correlation is represented by lines of different colors and thicknesses, while DCMs and microbes are represented by different shapes. A red line represents a positive correlation, while blue signifies a negative correlation. A darker color indicates that the correlation is stronger. Rectangles represent DCMs, circles represent bacteria, and triangles represent fungi. Based on the value of the correlation coefficient, 0.8-1.0 indicated an extremely strong correlation, 0.6-0.8 indicated a strong correlation, 0.4-0.6 indicated a moderate correlation, 0.2-0.4 indicated a weak correlation, and 0.0-0.2 indicated a weak correlation or no correlation [26]. Metschnikowia and Apiotrichum fungi genera were extremely strongly positively correlated with Leuconostoc (r > 0.8). Furthermore, 17 DCMs exhibited a strong positive correlation with microorganisms (0.8 > r > 0.6). Among them, 12 DCMs demonstrated a strong positive correlation with Candida, and four DCMs showed a strong positive correlation with Burkholderia. L-quinate showed a strong positive correlation with Leuconostoc, Metschnikowia, and Apiotrichum. Additionally, 4-hydroxy-6-methyl-3-(1-oxobutyl)-2H-pyran-2-one displayed a strong positive correlation with Burkholderia, Dyella, and Candida, while 3-deazaadenosine showed a strong positive correlation with Burkholderia, Dyella, Wickerhamomyces, and Candida. On the other hand, 5 DCMs exhibited a strong negative correlation with microorganisms (−0.6 > r > −0.8). Specifically, three DCMs (PC)14:1(9Z)/20:1911Z), PC(18:2(9Z,12Z)/18:1(11Z)), and PE-NMe(22:2(13Z,16Z)/16:1(9Z)) showed a strong negative correlation with Hanseniaspora. Additionally, portulacaxanthin II had a strong negative correlation with Candiada, and 3-deazaadenosine showed a strong negative correlation with Tatumella. In addition, all 23 DCMs exhibited a moderate correlation with microorganisms. Altogether, microorganisms demonstrated an important impact in shaping coffee compounds during the washed processing.

Materials and Chemical Standards
Mature coffee cherries (C. arabica) were obtained from Puer City, Yunnan Province, China. After harvesting, the coffee cherries were processed using the washed processing method [4,20,23]. Initially, red mature coffee cherries were hand-picked and de-pulped to remove the skin and the pulp. Subsequently, the thin mucilaginous layer surrounding the coffee seeds was removed via fermentation. During the fermentation, four coffee bean samples were obtained at 0 h (WC0), 12 h (WC1), 24 h (WC2), and 36 h (WC3) (n = 3), respectively. The coffee bean samples were divided into two parts: one part was stored at −80 °C for high-throughput sequencing analysis, and the coffee beans were obtained by removing the parchment skin from the other part and stored separately at −20 °C for the analysis of metabolites. High-performance liquid chromatography (HPLC) grade methyl alcohol, acetonitrile, and propyl alcohol were purchased from Fisher Co., Ltd. (Shanghai, China). The MagAtrract PowerSoil Pro DNA Kit was purchased from Oiagen (Hilden, Figure 6. The interaction of microbes and significant DCMs in different fermentation times in coffee samples based on the Spearman correlation analysis (p < 0.05). Red solid lines mean positive correlation, solid blue lines mean negative correlation, thicker solid lines mean stronger correlation, and thinner solid lines mean weaker correlation.

Materials and Chemical Standards
Mature coffee cherries (C. arabica) were obtained from Puer City, Yunnan Province, China. After harvesting, the coffee cherries were processed using the washed processing method [4,20,23]. Initially, red mature coffee cherries were hand-picked and de-pulped to remove the skin and the pulp. Subsequently, the thin mucilaginous layer surrounding the coffee seeds was removed via fermentation. During the fermentation, four coffee bean samples were obtained at 0 h (WC0), 12 h (WC1), 24 h (WC2), and 36 h (WC3) (n = 3), respectively. The coffee bean samples were divided into two parts: one part was stored at −80 • C for high-throughput sequencing analysis, and the coffee beans were obtained by removing the parchment skin from the other part and stored separately at −20 • C for the analysis of metabolites. High-performance liquid chromatography (HPLC) grade

Accession Numbers
The raw sequencing reads of bacterial 16S rRNA and fungal ITS1 were deposited into the NCBI Sequence Read Archive (SRA) database (Accession Number: PRJNA992233 and PRJNA992238).

Conclusions
The microbial community structure and DCMs of C. arabica beans from Yunnan province in the coffee fermentation process using the washed process were compared. The bacterial community consisted of 11 phyla and 11 genera, while the fungi community had eight phyla and 11 genera. The predominant phyla were Proteobacteria, Firmicutes, Actinobacteria, Cyanobacteria, and Bacteroidota for bacteria and Basidiomycota, Ascomycota, Mortierellomycota, and Mucoromycota for fungi. The top five predominant bacterial genera were Achromobacter, Tatumella, Weissella, Streptococcus, and Trichocoleus, while the dominant fungal genera were Cystofilobasidium, Hanseniaspora, Lachancea, Wickerhamomyces, and Aspergillus. A total of 2548 metabolites were identified during the coffee fermentation process, which were classified into 17 classes. Furthermore, a total of 82 DCMs were detected in WC1/WC0. Among them, 50 DCMs decreased significantly and 32 DCMs increased. In WC2/WC1, 46 DCMs were detected, of which 28 DCMs decreased significantly, while 17 DCMs increased significantly. In WC3/WC2, 45 DCMs were detected, with 31 DCMs significantly decreased and 12 DCMs significantly increased. Overall, 143 DCMs were identified in WC3/WC0, with 115 DCMs decreased significantly and 28 DCMs increased significantly. The correlation analysis revealed that 17 DCMs had a strong positive correlation with microorganisms, while 5 DCMs had a strong negative correlation. These findings indicated that microorganisms have an important influence on coffee compounds during coffee fermentation. Further research is necessary to identify the crucial microbial genera that affect the coffee compounds, to understand the response and succession rules of these microorganisms during the coffee fermentation, and to determine their effects on the aroma and quality of coffee. Thus, obtaining key microorganisms that impact coffee flavor is a way to improve the quality of coffee flavor.