Yeasts are essential for mucilage degradation of coffee beans during wet fermentation

During wet fermentation, mucilage layers in coffee cherries must be removed completely. To explain mucilage degradation, several controversial hypotheses have been proposed. The aim of this work was to improve our understanding of the kinetics of mucilage breakdown. Pulped coffee beans were wet fermented with seven different treatments for 36 h. Endogenous bacteria and yeasts are selectively suppressed, and pectinases or lactic acid are added. They also involve maintaining the beans at pH 7 throughout fermentation and using spontaneous fermentation without additives as a control. During spontaneous fermentation, yeast and lactic acid bacteria were detected and significantly increased to 5.5 log colony‐forming units (CFU)/mL and 5.2 log CFU/mL, respectively. In the first 12 h of fermentation, there was a significant degree of endogenous pectinolytic activity, which resulted in partly destroyed beans in the absence of microorganisms. By adding pectinase and lactic acid to the fermentation mass, the breakdown process was accelerated in less than 8 h. When yeast was present throughout the fermentation, complete degradation was achieved. Bacteria played no critical role in the degradation. Klebsiella pneumoniae and Erwinia soli were found in a lower population and showed weaker pectinolytic activities compared to Hanseniaspora uvarum and Pichia kudriavzevii. During wet fermentation, mucilage degradation appears to be mediated by endogenous enzymes at the early stage, whereas microbial contributions, mainly yeasts, occur subsequently. H. uvarum and P. kudriavzevii may be promising candidates to be tested in future studies as coffee starter cultures to better control the mucilage degradation process.

Although mucilage removal is one of the primary objectives of wet coffee fermentation, the scientific community is divided on the mechanics behind the process. Several hypotheses have been offered to explain how mucilage breaks down during the fermentation process . The earliest hypothesis claims that mucilage hydrolysis was caused by the activity of endogenous enzymes found in coffee beans (Lilienfeld-Toal, 1931). According to the author, the degradation was accomplished before microbial activity took over the fermentation. The second hypothesis is that microbial activity in fermentation contributes to mucilage breakdown in coffee beans (Avallone et al., 2002;Oumer & Abate, 2017). Avallone and colleagues discovered that the pectinolytic activities of bacteria isolated from coffee fermentations were rather weak, leading them to conclude that mucilage degradation during fermentation was mostly due to mucilage acidification caused by bacterial metabolites, mainly organic acids (Avallone et al., 2002). They also discovered that pectinolytic activities were lacking in all isolated yeasts (Candida, Cryptococcus, and Kloeckera spp.) from coffee fermentations. Other studies, however, have not validated this conclusion. Several yeast species, including Hansinospora, Candida, Kluyveromyces, Pichia, Saccharomyces, and Schizosaccharomyces, have been found in high populations in coffee fermentations conducted in various locations around the world (Agate & Bhat, 1966;Masoud & Jespersen, 2006;de Melo Pereira et al., 2014;Silva, 2014). The majority of these yeast isolates exhibit significant pectinolytic activity (de Melo Pereira et al., 2014;Haile & Kang, 2019;Silva et al., 2013). Their role in mucilage breakdown in wet coffee fermentation, however, has not been well studied. Similarly, Oumer and Abate (2017) demonstrated the potential mucilage degradation activities of Bacillus subtilis pectinase, which completed the mucilage breakdown in 24 h compared to 36 h in the absence of the bacterial pectinase enzyme.
Thus, bacteria may play an important role in mucilage breakdown via acidification, extracellular enzymes, or a combination of the two.
Overall, the mechanisms of mucilage degradation during wet coffee fermentation remain controversial, and the contributions from the various factors are still unclear. The main objective of this study was to investigate the mechanisms of mucilage degradation during wet coffee fermentation and, specifically, to determine the roles of endogenous pectinases, microorganisms (yeasts and bacteria), and organic acids in the process. The exclusive focus of this study was to investigate the mucilage layer removal kinetics that was conducted in parallel with studying the fermentation contribution to final coffee quality.

| Coffee beans wet fermentation process
Coffee cherries (Coffea arabica var. Bourbon, 40 kg) were harvested (October 2018) at Kahawa Estate Coffee farm in Teven, NSW, Australia (latitude and longitude coordinates, −28.816667, 153.500000; altitude, 6.6 m above sea level) and immediately placed in iceboxes and airfreighted to UNSW Sydney. The cherries were depulped manually using sterile gloves once arrived and the beans obtained were subjected to 36 h wet fermentation immersed under water in triplicate as described in Elhalis et al. (2020). Liquid samples (100 mL) were collected from the fermentation mass at 0, 16, 24, and 36 h for microbiological and enzymatic analysis, in triplicate.

| Microbial study of the fermentation process
Identification and monitoring of the microbial growth during the fermentations were followed as described in Elhalis et al. (2020). In brief, total aerobic bacterial counts and lactic acid bacteria (LAB) counts were determined using plate count agar and de Man Rogosa Sharpe (MRS) agar with 0.1% cycloheximide, respectively. Yeasts were enumerated using malt extract agar (MEA) containing 50 mg/L each of oxytetracycline and chloramphenicol. After incubation, the microbial isolates were enumerated as groups (yeast and bacteria), as well as individual species based on colony and cell morphology.
Colony-forming units (CFU) were reported as the mean values of duplicate analysis, and at least three colonies from each type were further streaked on the same media a minimum of four times for purification.

| Identification of microbial isolates
Microorganisms were identified by sequencing appropriate genes and the sequence results were blasted to compare against the available known microbial sequence data through NCBI (http://www.ncbi.nlm. nih.gov/BLAST/), following the procedure in Elhalis et al. (2020). In brief, DNA was extracted from the purified colonies following the protocol of Cocolin and Ercolini (2015). For bacteria, the V3 region of the 16S ribosomal RNA (rRNA) gene was amplified with the primer pairs F338fgc and R518 (Ovreås et al., 1997). The 5.8S internal Take-away • Endogenous enzyme activities in coffee beans might be important, but insufficient to degrade the mucilage layer completely.
• Pectinolytic yeasts are crucial for breaking down coffee bean mucilage successfully.
• Mucilage degradation is not significantly affected by endogenous bacterial species.
• Mucilage degradation process was accelerated by organic acids and pectinase, although careful consideration of sensory quality impact is required.
• Hanseniaspora uvarum and Pichia kudriavzevii are ideal starter cultures for improving coffee mucilage degradation.
transcribed spacer (ITS) rRNA gene region of the yeast isolates was amplified using the primers ITS1 and ITS4 (Esteve-Zarzoso et al., 1999). The primers and PCR conditions were performed as shown in Supporting Information: Table S1a,b. The PCR products and 9.6 pmol of respective forward primers were sent for sequencing to Ramaciotti Centre for Genomics at UNSW (Kensington, Sydney, NSW, Australia).
The sequencing results were compared with the microbial sequence data in the GenBank through NCBI (http://www.ncbi.nlm.nih.gov/ BLAST/). The isolates were assumed to belong to a given species if the similarity was higher than 98%.

| Pectinolytic activities analysis of the fermenting mass
The liquid samples (50 mL) collected from the fermentation containers were centrifuged at 10,000 rpm for 15 min at 4°C in an Avanti J-E Centrifuge (Beckman Coulter, Indianapolis, IN, USA). The supernatants were filtered through 0.45 µm syringe filters to remove residue microbial cells, used as the crude enzyme extract (CE), and kept at −20°C until used. For analysis, the CE was thawed and precipitated by mixing with three volumes of ice-cold acetone for 15 min (Rajendran et al., 2011). The entire mixture was centrifuged at 4000 rpm for 20 min at 4°C. The precipitate was collected and dissolved in sodium acetate buffer (0.1 M, pH 5) to yield partially purified enzyme extracts (PPs), which were subjected to further purification by gel filtration using Sephadex G-75 (35 × 1.5 cm, bed volume 12-15 mL/g; Sigma-Aldrich) following the procedure of Keller et al. (1982) to yield purified extract. The pectinase activity of the Sephadex G-75 purified extract was first screened by transferring 20 µL onto the surface of pectin agar (0.5% pectin, 85% esterified, and 2.5% agar, pH 5) and polygalacturonic acid agar (0.5% polygalacturonic acid and 2.5% agar, pH 5) plates. The media contained 50 mg/L chloramphenicol and 0.1% cycloheximide to inhibit microbial growth, and the plates were incubated at 30°C for 36 h. The hydrolysis of pectin and polygalacturonic acid was assessed after incubation by flooding the plates with a 50 mM potassium iodide-iodine solution and observing the development of a clear area surrounding the inoculum.
The pectinolytic activities of the CE, PPs (acetone precipitated), and Sephadex G-75 purified fractions were assayed by the dinitrosalicylic acid (DNS) reagent method following Miller (1959) procedure. In brief, 0.9% citrus pectin (75% esterified) or 1% polygalacturonic acid, dissolved in acetate buffer (2 mL, pH 5; Sigma-Aldrich), was mixed with 1 mL of each of the three different fractions described above, and incubated for 1 h at 30°C. To measure the reducing sugars liberated by the pectinolytic activity, the entire solution (3 mL) was mixed with an equal volume of DNS reagent and further incubated for 10 min with shaking. The mixture was then placed in boiling water for 10 min and cooled down in an ice-water bath. The mixture was diluted by adding 20 mL of MilliQ water and the absorbance was measured spectrophotometrically at 570 nm.
One unit of pectinase activity was expressed as the amount of enzyme that liberated 1 μmol of D-galacturonic acid per minute under the assay conditions.
The CE collected was also used to monitor the changes in pH and total reducing sugars during fermentation at the same time intervals of 0, 16, 24, and 36 h. A portable pH meter (pH cube; TPS Pty Ltd, Brisbane, QLD, Australia) was used to measure pH in three replicates.
The total reducing sugars liberated in the fermentation mass were also determined using DNS in three replicates (Miller, 1959). Briefly, 3 mL of the CE was mixed with 3 mL of DNS, heated at boiling water for 10 min, and cooled down with ice. The total reducing sugar concentration was measured by measuring the absorbance at 570 nm in the same way as described above. A standard curve was created using galacturonic acid.
2.5 | Molecular weight of pectinase by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) The CE, PPs, and Sephadex G-75 purified enzyme fractions were precipitated using 50% trichloroacetic acid, and the precipitant was resuspended in 0.05 M Tris-HCl buffer. The molecular mass of the enzymes was determined by SDS-PAGE using a 10% polyacrylamide gel according to the procedure described by Laemmli (1970). Protein concentrations in the fractions were measured by the Bradford assay (Bradford, 1976) using bovine albumin as a reference standard.

| Mucilage degradation of coffee beans under different fermentation conditions
A novel approach was applied to investigate the factors contributing to mucilage degradation during the wet fermentation of coffee beans. Pulped coffee beans (1000 g) were subjected to submerged fermentations at 27 ± 2°C for 36 h under seven different treatments, as illustrated in Figure 1: (1) tap water only without any additives (spontaneous, control); (2) antibacterial solution containing oxytetracycline (0.1 mg/mL) and chloramphenicol (0.1 mg/mL) to suppress bacterial growth; (3) antifungal solution containing cycloheximide (2 mg/mL) and natamycin (0.4 mg/mL) to suppress fungal growth; (4) a mixture of solutions 1 and 2 to suppress both microbial groups; (5) solution 4 with the addition of commercial pectinase (0.035 w/v, pectinase from Aspergillus niger; Sigma-Aldrich); (6) solution 4 with the addition of lactic acid to reach pH 4; (7) solution 4 with the pH kept at 7 using 0.02 N NaOH. Mucilage degradation of the beans was monitored every 6-12 h by visual inspection of the beans, and hand feels as per the traditional method (Lin, 2010). The culture plate methods, as described above, were used to check the growth or inhibition of the target microbial groups. Samples (liquid fractions, 20 mL) were collected every 6-12 h to measure the pH and the total reducing sugars as described above. The experiment was conducted in three replicates and mean values were reported. ELHALIS ET AL. | 427 2.7 | Evaluation of the pectinase activities of the detected isolates 2.7.1 | Screening for pectinase activity by plate assay All bacterial and yeast isolates were first screened for pectinase activities by streaking on pectin agar and polygalacturonic acid agar following the method described by Schwan et al. (1997) with some modifications. The medium was composed of 2 g yeast extract, 20 g agar, and 10 g of either polygalacturonic acid or citrus pectin, dissolved in 1 L of mineral salt solution ((NH 4 ) 2 SO 4 , 4 g; KH 2 PO 4 , 8 g; N 2 HPO 4 , 10 g; CaCI 2 , 1 g, pH 5), with or without D-glucose supplement (10 g). The plates were incubated at 30°C for 36 h. At the end of the incubation, all plates were flooded with 50 mM potassium iodide-iodine solution. Colonies surrounded by a clear halo zone against the opaque medium indicated the ability to produce pectinase. All pectinase-positive colonies were further screened by spot surface inoculation on the same medium with 20 µL of 6 log CFU/mL of the selected isolates and incubated at 30°C for 36 h in three replicates. The ratio of the hydrolysis area to colony diameter was measured to identify isolates with high pectinase activities.

| Assay of pectinase activities of the isolates
The selected isolates that showed pectinase activities on the plate assay were subjected to further analysis of their pectinase activity using submerged fermentation. One pure bacteria and yeast colony were picked and transferred into tryptone soy broth and yeast extract broth (of at least three typical colonies of each identified species), respectively, and incubated at 30°C overnight in a shaking incubator (120 rpm). The bacterial and yeast cells were then collected by centrifugation (3200g, 15 min) and resuspended in 200 mL 0.1% sterile peptone water (Sigma-Aldrich). The cell suspensions were adjusted to 10 6 CFU/mL using a counting chamber (Improved Neubauer), and the cell count and purity were further checked by plating on either plate count agar, MRS, or MEA as described above without the selective additives. The cell suspension (0.1 mL) was inoculated into 100 mL of a sterile broth consisting of 0.4% (w/v, same for all below) (NH 4 ) 2 SO 4 , 0.8% KH 2 PO 4 , 1% N 2 HPO 4 , 0.1% CaCI 2 , 0.2% yeast extract, and 1% D-glucose, containing either 1% pectin or 1% polygalacturonic acid, pH 5. The broth cultures were incubated at 30°C for 36 h in a rotary shaker (140 rpm). Then, the broth was centrifuged at 10,000 rpm for 15 min at 4°C, and the supernatant was cold sterilized by filtrating with a 0.22-µm syringe filter to obtain the CE and kept at −20°C till future analysis to evaluate pectinase activity.
F I G U R E 1 Degradation study methodology flowchart. (1) Control composed of 1000 g of depulped beans kept under water and (2-7) composed of the same quantities of beans and water plus the treatments. Antibacterial agents were oxytetracycline (0.1 mg/mL) and chloramphenicol (0.1 mg/mL); antifungal were cycloheximide (2 mg/mL) and natamycin (0.4 mg/mL); LA, lactic acid (pH 4); pectinase (0.035 w/v, pectinase from Aspergillus niger; Sigma-Aldrich). The experiment was conducted in three replicates and mean values were reported. Created with BioRender.com.
Spots of 20 µL of the cell suspensions were loaded onto pectin agar (0.5% pectin, 1.5% agar) and polygalacturonic acid agar (0.5% polygalacturonic acid, 1.5% agar). Wells were also made aseptically by a borer in pectin agar and polygalacturonic acid agar and the CE (100 µL) was loaded into the wells. All plates were incubated at 30°C for 36 h, after which the agar plates were flooded with iodine solution as described above. CE that produced a clear zone around the spots or wells was assessed for various pectinolytic activities, namely, pectin methylesterase (PME), pectin lyase (PL), and polygalacturonase (PG) in three replicates by the following methods. PME activity was determined by titration against NaOH to measure the increase in the concentration of free carboxyl group (Schwan et al., 1997). A pectin substrate solution (8 mL of 1% pectin dissolved in acetate buffer, pH 5) was mixed with 2 mL of the CE and incubated at 30°C for 1 h. After incubation, 0.002 N NaOH was used to titrate against the solution using phenolphthalein as an indicator. Heat-denatured CE (90°C, 10 min) was analyzed in the same way and used as a control. PME activity was expressed as µmol equivalent of NaOH used per mL in 1 min under the assay conditions. PL activity was assayed by following the method of Albersheim (1966). Polygalacturonic acid solution (1%, 5 mL, pH 5) and 2 mL of CE were mixed and incubated at 30°C for 1 h, and the increase in absorbance of the mixture at 235 nm was measured spectrophotometrically. PL activity (1 U) was expressed as the enzyme activity that resulted in an increase of 0.555 in absorbance in 1 min under the assay conditions. PG activity was determined by the DNS method of Miller (1959), as described above.

| Protein determination
The total protein concentration in the CE, partially purified, and Sephadex G-75 purified enzyme fractions were determined by the Bradford method with bovine serum albumin used as the standard (Bradford, 1976).

| Statistical analysis
One-way analysis of variance was carried out to compare mean values between sample treatments and Tukey's HSD post hoc test was used to separate mean values of significant differences.

| Changes in microbial population during fermentation
A total of 16 different microbial species were isolated from spontaneous fermentations of coffee beans. The species identities and growth patterns of the isolates are shown in Table 1. After 24 h, both total yeasts and LAB increased and reached 5.3 and 5.6 log CFU/g, respectively. The total yeast population remained high throughout the fermentation, while the LAB population decreased slightly toward the end of the process. The predominant yeast isolates were Hanseniaspora uvarum and Pichia kudriavzevii, whereas the predominant LAB isolates were Leuconostoc mesenteroides and Lactococcus lactis. There was a presence of Enterobacteriaceae during fermentation, primarily Citrobacter, Enterobacter, Klebsiella, and Erwinia. Citrobacter sp. population gradually increased to reach the maximum population of 7.2 CFU/g at the end, while those of the remaining three genera decreased after 12 h. Acetic acid bacteria (AAB), mainly Gluconobacter and Acetobacter, were detected during fermentation with a maximum population of about 4 log CFU/g. A few other bacterial species, such as Acinetobacter and Pseudomonas, appeared at the start of fermentation, but their populations gradually decreased during the fermentation process.

| Pectinolytic activities detected during the wet fermentation process
The Sephadex G-75 purified enzyme extracts obtained at the start of the fermentation (0 h) revealed a clear hydrolysis halo (about 1.2 cm in diameter) on polygalacturonic acid agar around the inoculum; the 16 h extract had a much smaller hydrolysis zone of 0.5 cm in diameter, while the 36 h extract had no hydrolysis zone (Supporting Information: Figure S1a). Similar patterns of hydrolysis with the purified extract of fermenting mass were also observed on pectin agar. However, the hydrolysis zones were wider than on polygalacturonic acid agar (Supporting Information: Figure S1b). The visual observations of the agar plate were confirmed by analysis of the extracts for their pectinolytic activities with both pectin and polygalacturonic acid as substrates ( Table 2). The extracts obtained at 0 h of fermentation all exhibited pectinolytic activities, which showed substantial decreases for the 16 h extracts. With pectin as the substrate, the pectinolytic activities of the 16 h extracts became undetectable. In contrast, with polygalacturonic acid as the substrate, the activities decreased by 54.7%-61.6% for the various extracts. Table 2 also showed that the purification of extracts, especially by Sephadex G-75 gel filtration, resulted in substantial increases in enzymatic activities. For example, compared with the crude extract, the activity of the Sephadex G-75 purified extract increased by more than 5.2 and 6.7 times with pectin and polygalacturonic acid as substrates, respectively. SDS-PAGE showed the presence of a major protein band with a molecular weight of about 31 kDa (Supporting Information: Figure S1c).

| Mucilage degradation under various conditions
Plate count data confirmed the suppression of target microbial groups during coffee bean fermentation by the selected antimicrobial ELHALIS ET AL. | 429 agent combinations as shown in Figure 2 and Supporting Information: Table S2. In fermentations where antibacterial agents were added, bacterial growth was absent, while a slightly higher yeast population was observed compared with the control (p < 0.05). Similarly, when antifungal agents were added to the fermentation broth, yeast growth was completely eliminated, and the population of total aerobic bacteria was approximately 1 log CFU/mL higher than the control (p < 0.05), while LAB was similar in both treatments (p > 0.05).
Both yeasts and bacteria were absent when antibacterial and antifungal agents were applied (treatments 4-7). Mucilage degradation occurred during fermentations with all treatments except treatment 7 (no microbial growth, pH 7) where the mucilage remained largely unchanged (Supporting Information: Figure S2).
Complete mucilage degradation was observed in the first 6 h when lactic acid was used (treatment 6), followed by pectinase addition  Abbreviations‫׃‬ CE, crude enzyme extract from the fermentation mass; PC, pectin dissolved in acetate buffer (pH 5); PG, polygalacturonic acid dissolved in the same buffer; PP, partially purified enzyme by cold acetone precipitation; SP, Sephadex G-75 purified enzyme. fermentation broth, followed by pectinase, while a low level of reducing sugar accumulation was detected when yeasts were inhibited. The pH of the fermentation broth gradually declined during fermentation to pH 3.6 in the control (Figure 2b). In contrast, the pH value was significantly higher in fermentation where both yeasts and bacteria were suppressed (treatment 4). For the other treatments, the pH values fell between the control and treatment 4, except for treatment 7, where the pH was artificially maintained at around 7.0, and treatment 6, where lactic acid was added and the pH was low, as expected.

| Screening of microbial pectinase activity by plate assay
The microbial species isolated from spontaneous fermentation of coffee beans exhibited poor growth and no degradation capability on pectin and polygalacturonic acid agar without glucose supplement (Supporting Information: Figure S3a), but their growth improved when glucose was added (Supporting Information: Figure S3b,c). The pectinolytic activities of both bacteria and yeasts on media containing pectin and polygalacturonic acid were F I G U R E 2 (a) Changes in the concentration of total reducing sugars, and microbial counts and (b) changes in pH values during coffee bean fermentation under several treatments. Untreated spontaneous fermentation (control); AB, antibacterial agents (oxytetracycline, 0.1 mg/mL and chloramphenicol, 0.1 mg/mL); AF, antifungal agents (cycloheximide, 2 mg/mL and natamycin, 0.4 mg/mL); LA, lactic acid (pH 4); enzyme (0.035 w/v, pectinase from Aspergillus niger; Sigma-Aldrich); TABC, total aerobic bacterial count; TLABC, total lactic acid bacteria count; TYC, total yeasts count. The experiment was conducted in three replicates; the values reported are means with error bars representing 1 SD. The black solid squares in Figure 2 (a) are generated by merging two or more microbial count figure legends and reflect suppression of two or more microbial groups (treatment 2 suppresses TABC and TLABC, while treatments 4, 5, 6, and 7 suppress TYC, TABC, and TLABC). Extracts from the submerged fermentations of several yeasts and bacteria showed PG and PL activities (Figure 3). For the yeasts, P.
kudriavzevii displayed the greatest PG activity (120.3 U/g), followed by H. uvarum and W. anomalus at 90.7 and 50.6 U/g, respectively. For the bacterial isolates, E. soli exhibited the highest PG activity (37.0 U/g), followed by K. pneumoniae and L. lactis. PL activity was absent in all bacterial isolates except L. lactis, which showed a relatively low activity of 11.3 U/g. For yeasts, C. railenensis exhibited the highest activity (36.3 U/g), followed by C. xylopsoci, H. uvarum, and P. kudriavzevii. A low level of PL activity was detected with P. fermentans. PME activity was not detected in any isolate extract.
F I G U R E 3 Pectin lyase (PL) and polygalacturonase (PG) activities of cell-free extracts of yeast and bacteria after submerged fermentation. Data are means of three replicates and error bars represent the deviation between the parallels (n = 3).
Previous studies have reported that both yeasts and bacteria grew significantly during the wet fermentation of coffee beans, with a consequent decline in pH, mainly because of the microbial production of organic acids, such as acetic and lactic acids Elhalis et al., 2021c;Evangelista et al., 2015;Masoud et al., 2004;de Melo Pereira et al., 2014;Ribeiro et al., 2017). These findings are confirmed in the present study. In this study, yeasts such as H. uvarum and P. kudriavzevii and bacteria such as L. mesenteroides and L. lactis grew significantly during coffee bean spontaneous fermentation. Several other bacterial species, including AAB, were also detected at the start of fermentation, but their population generally decreased with fermentation progress. Similar observations were also reported in previous studies (Agate & Bhat, 1966;De Bruyn et al., 2017;Evangelista et al., 2015). Overall, these microbial and pH changes observed in our study are broadly consistent with the findings reported by others, which indicates that microbial fermentation occurred successfully in our laboratory bench-scale study.
Mucilage degradation of coffee beans is a crucial function of wet reported that endogenous pectinase enzymes of coffee beans contribute to mucilage degradation during wet coffee fermentation (Lilienfeld-Toal, 1931). However, our data demonstrated that these enzymes alone could not achieve complete removal of the mucilage during fermentation without the participation of microorganisms.
Furthermore, the relatively slow accumulation of the reducing sugars and decline of pH in the first 12 h of this treatment, which remained largely unchanged in the rest of the process, appeared to indicate the cessation of pectin degradation after 12 h. In contrast, the accumulation of the liberated reducing sugars and decline of the pH continued throughout the spontaneous fermentation (control), which demonstrated that the mucilage hydrolysis persisted during the whole fermentation process and was likely associated with microbial activities (Agate & Bhat, 1966;De Bruyn et al., 2017;Masoud & Jespersen, 2006;Murthy & Naidu, 2011;Silva et al., 2013).
On the other hand, pectinolytic enzymes played a crucial role in mucilage degradation during wet coffee fermentation. This was clearly demonstrated by the rapid (in 8 h) and complete mucilage removal in treatment 5, where commercial pectinase was added to the fermentation mass while all microbial activities were suppressed.
Similar observations have also been reported in other studies (Murthy & Naidu, 2011). Surprisingly, adding lactic acid to the fermentation mass caused the fastest degradation of mucilage (twice as fast as spontaneous fermentation). Lactic acid might contribute to mucilage degradation in two ways. First, acidification is reported to cause swelling of the mucilage that facilitates mucilage breakdown, as described by Avallone et al. (2002). Second, adding acid would lower the pH of the fermentation broth, creating a favorable condition for endogenous pectinase, which has an optimum pH of around 4.5 (Kapoor et al., 2000;Kohli & Gupta, 2015;Phutela et al., 2005). This would enable more efficient removal of mucilage in a shorter time.
The importance of acidity to mucilage breakdown was further demonstrated in treatment 7. All microbial activities were inhibited, the pH was maintained at 7 throughout the fermentation, and no mucilage degradation was observed under these conditions.
Combining these observations, it can be hypothesized that mucilage degradation by microorganisms might be due to the pectinolytic enzymes they secrete, the acid they produce during the fermentation, or, more likely, a mix of both. Avallone et al. (2002) believed that the main contribution of microbial activities, mainly bacteria, to mucilage degradation was due to the acidification process and rejected the roles of microbial enzymes in the process. To determine the role of microbial enzymes in mucilage degradation, all endogenous microbial isolates from coffee fermentation were evaluated for their potential to secrete pectinase in solid and liquid media containing pectin or polygalacturonic acid. These isolates showed poor growth when pectin or polygalacturonic acid was the sole carbon source in the medium; the addition of glucose to the medium improved their growth and pectinolytic activities. This showed that in the presence of simple sugars such as glucose, endogenous isolates can grow and produce pectinase enzymes that might contribute to the hydrolysis of pectin in the mucilage. In natural spontaneous wet fermentation, the sugars could be provided by those in the mucilage as well as those leaching out from inside the beans (Wootton, 1974). Avallone and co-workers reported weak growth and pectinase activities of most isolates from the wet fermentation process (Avallone et al., 2002; (Evangelista et al., 2015;Mariyam et al., 2023;Masoud et al., 2004).
These studies confirmed H. uvarum's strong pectinase activity, as well as its capacity to create desired aroma compounds, suppress Aspergillus ochraceus growth, and restrict mycotoxin production (Elhalis et al., 2021b(Elhalis et al., , 2021aEvangelista et al., 2015;Masoud & Jespersen, 2006;Masoud & Kaltoft, 2006). Similarly, P. kudriavzevii has been identified in coffee fermentation processes in India, Thailand, and the Republic of Korea, demonstrating its relatively high abundance during fermentation. The studies showed that P.
xylopsoci, and W. anomalus, have also been found in coffee fermentation studies with diverse populations (de Melo Pereira et al., 2014;Silva et al., 2008). In a previous study, our research group inoculated P. fermentans, C. railenensis, C. xylopsoci, and W. anomalus into a synthetic coffee pulp extract medium to screen for ideal coffee starter cultures that produce desirable aroma profiles (Elhalis et al., 2021a). Lower fermentation performances and higher concentrations of acetic acid were observed when compared to H. uvarum and P.
kudriavzevii (Elhalis et al., 2021a). These findings highlight the potential of H. uvarum and P. kudriavzevii to be employed as starter cultures, which can enhance the coffee fermentation process and may produce exceptional coffee quality. In the fermentation of coffee beans, LAB and other bacteria are often isolated, but their functions have not been well studied Vinícius de Melo Pereira et al., 2017). L. lactis, K. pneumoniae, and E. soli have been isolated during coffee bean fermentation in different population densities elsewhere Leong et al., 2014;Silva et al., 2008;Vilela et al., 2010).
Bacteria isolates have relatively low pectinolytic activities and have no acidification effect that is caused by the addition of lactic acid in treatment 6. These factors may explain the observation that mucilage removal was incomplete within the same timeframe when bacteria alone were present during fermentation. It is worth mentioning recent trials to employ LAB species, including Lactiplantibacillus plantarum and Lactobacillus rhamnosus, as starter cultures in coffee bean fermentation with promising results of improving the final product's sensory quality (de Melo Pereira et al., 2016;Ribeiro et al., 2020;Wang et al., 2019). The current study demonstrated that isolated bacteria had a less significant impact on mucilage degradation than yeast species. Their coculture with high pectinolytic yeast isolates or inoculation in the presence of endogenous pectinolytic isolates, with appropriate concentrations of organic acids or pectinase, might be potential alternatives for controlling the fermentation process under both sterile and nonsterile conditions.
These approaches require more in-depth investigation.

| CONCLUSION
This study demonstrated that mucilage degradation during wet coffee fermentation was a complex process that involved endogenous and microbial pectinolytic enzymes. Organic acids, both from the bean pulp and those produced during microbial metabolism, played a significant role in the process. This was probably accomplished by bringing the pH near the optimum range of the pectinolytic enzymes, thus accelerating the process of degradation. Mucilage breakdown was aided by endogenous enzymes, which was crucial during the initial stages of spontaneous wet fermentation. Without microbial activities, especially yeasts, the endogenous enzymes were unable to completely remove the mucilage. These yeasts were critical for mucilage breakdown during wet coffee fermentation. Mucilage degradation could be accelerated by adding pectinase, organic acids, pectinolytic yeasts, and LAB, or combinations of them to the fermentation. However, further studies are needed to determine the impact of such interventions on coffee beverages' sensory quality.

CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.