Connection of spectral pattern of carbohydrate molecular structure to alteration of nutritional properties of coffee by-products after fermentation

Fermentation study

Coffee from plantation by-products were gathered (wet processed), and they were subsequently dried in an oven at 60°C for 24 hours, until 90% of the DM was retained. Fermentation was carried out using the commercial fermentation product Saus Burger Pakan (SBP; 1%). SBP is made up of a variety of microorganisms (cellulolytic, lactic acid, amylolytic, proteolytic, and xylanolytic bacteria) and herbal extracts. SBP was activated with 3% molasses for 5 hours before being combined with coffee by-products. The samples were grinded and fermented with SBP for 0, 7, 14, 21, and 28 d. To optimize microbial fermentation, 3% sago was used as an energy source. Water was also added to the fermenting mix to achieve a water content of 40%. The fermentation experiment was carried out in anaerobic containers at room temperature (28°C to 30°C), with a total of 15 containers arranged, three for each time point.

The containers were opened at each fermentation time point, and the samples were dried in an air-ventilated oven (60°C) until consistent weight was achieved. Dried samples were ground using Retsch ZM-200 (Brinkmann Instruments Ltd., Mississauga, Canada) for chemical composition analysis (screen size: 1 mm) and spectral collection of chemical functional groups (screen size: 0.12 mm) at the SRP ministry strategic research lab with spectroscopy (Department of Animal and Poultry Science, University of Saskatchewan, Canada).

Chemical analysis

The fermented by-product samples from coffee processing were determined for chemical compositions including dry matter (DM; AOAC 930.15), organic matter, crude protein (CP; AOAC 984.13), crude fat or ether extract (EE, AOAC 920.39), ash (AOAC 942.05), and starch (AOAC 996.11) following the procedures of Association of Official Analytical Chemists [12]. The published methods by Van Soest et al [13] were applied for feed fibre analyses including acid detergent lignin (ADL) and acid (ADF) and neutral (NDF) detergent fibers with Ankom A200 technique (Ankom Technology, Fairport, NY, USA). In the residues from fibre analysis (NDF, ADF), both acid (ADICP) and neutral (NDICP) detergent insoluble CP were determined. Soluble CP (SCP) in the fermented by-product samples was also determined Roe [14]. According to NRC (2001), total CHO (%) was calculated as 100–EE–CP–ash, non-fiber CHO (NFC, %) was calculated as 100–(NDF–NDICP)–EE–CP–ash, and hemicellulose (%) and cellulose (%) were calculated as NDF–ADF and ADF–ADL, respectively. The protein-related chemical profiles were reported previously. In this study, carbohydrate-related chemical profiles were used for interactive association study with carbohydrate molecular structure spectral profiles.

Fractionation of carbohydrate fractions with CNCPS 6.5 system

The updated CNCPS 6.5 system [15] describes carbohydrate subfractions, including CA (CA1, volatile fatty acids; CA2, lactic acids; CA3, other organic acids; CA4, sugars); CB1 (rapid degradable fraction, starch); CB2 (intermediary degradable fraction, soluble fiber); CB3 (slowly degradable fraction, available cell wall) and CC (unavailable cell wall). Based on carbohydrate subfractions, the estimated degradable and undegradable carbohydrate supply as well as intestinal digestion of carbohydrate subfractions were evaluated for by-products from coffee processing affected by fermentation duration. The detailed parameters determined include i) rumen degradable (RD) carbohydrates including RDCA4, RDCB2, RDCB3 and total RDC; ii) rumen undegradable carbohydrates including RUCA4, RUCB2, RUCB3, RUCC, and total RUC; iii) intestinal digestion (ID) of subfractions including IDCA4, IDCB2 and IDCB3 [13,15].

Bioenergy values and total digestible nutrients

Total digestible nutrients (TDN) and bioenergy values in the by-products from coffee processing affected by microorganism fermentation duration were determined using NRC method [15,16]. The studied parameters include total digestible CP (tdCP), NDF (tdNDF), NFC (tdNFC), and fatty acid (tdFA), TDN, digestive energy (DE_1x, DE_p3x), metabolizable energy (ME, ME_p3x), net energy for maintenance (NE_m), gain (NE_g), and lactation (NE_L3x) in fermented by-products.

Molecular spectroscopy ATR-FTIR on carbohydrate structure

The dried fermented coffee by-products samples from 0, 7, 14, 21, and 28 d fermentation from 15 containers with three containers for each time point were scanned for FTIR spectra using JASCO FT/IR-4200 with sampling assessor of attenuated total reflection (ATR) (JASCO Corp., Tokyo, Japan) at the SRP Chair lab at University of Saskatchewan (Canada).

Each fermented and non-fermented by-product sample was scanned at mid-IR range (ca. 4,000 to 700 cm⁻¹). For each sample, five repetitions were used to be collected for spectra. Each sample spectrum was scanned for 256 times. The resolution was set at 4 cm⁻¹ for each spectrum. To offset background noise, the background spectrum from each sample was collected [5,17]. The previous studies [5,7,18–20] showed that carbohydrate-related spectral regions in plant-based biomaterials include total carbohydrate (at ca. 938 to 1,186 cm⁻¹ region), cellulosic chemical compound (at ca. 1,186 to 1,289 cm⁻¹ region) and structural carbohydrate (ca. 1,186 to 1,486 cm⁻¹ region) [18,21]. Spectral peak heights and peak areas within each region were analyzed with OMNIC 7.3 (SpectraTech, Madison, WI, USA).

The detailed spectral parameters reported by previous researchers [5,7,18–20] included STCHO: structural carbohydrate (area at ca. 1,488 to 1,184 cm⁻¹); CC: cellulosic compounds (area at ca. 1,300 to 1,184 cm⁻¹; height at ca. 1,237 cm⁻¹); NSHCO: non-structural carbohydrates (area at ca. 1,184 to 924 cm⁻¹); lignin compound (height at ca. 1,507 cm⁻¹); CH_bending (height at ca. 1,460 cm⁻¹); STCHO_pk1: structural carbohydrate peak1 (1st peak height at ca. 1,418 cm⁻¹); STCHO_pk2: structural carbohydrate peak2 (2nd peak height at ca. 1,397 cm⁻¹); STCHO_pk3: structural carbohydrate peak3 (3rd peak height at ca. 1,320 cm⁻¹); HC: hemicellulosic compound (height at ca. 1,150 cm⁻¹); NSCHO_pk1: non-structural carbohydrate peak1 (1st peak height at ca. 1,102 cm⁻¹); NSCHO_pk2: non-structural carbohydrate peak2 (2nd peak height at ca. 1,078 cm⁻¹); NSCHO_pk3: non-structural carbohydrate peak3 (3rd peak height at ca. 1,030 cm⁻¹) [5,7,19,20].

Statistical analysis

The chemical, nutritional and spectral data of fermented by-products samples were determined using the MIXED procedure (SAS Institute, Inc., Cary, NC, USA) with fermentation time (different dates) as fixed effects and fermentation run as random effect. The residual analysis was performed to validate the model assumptions. The data normality check was determined using SAS univariate procedure with normal option. Orthogonal polynomial contrasts were used to evaluate linear and quadratic effects of fermentation duration (0, 7, 14, 21, 28) with equal time spacing treatment. The contrast between control (non-fermentation by products) and fermentation (average of fermentation for 7, 14, 21, and 28 days) was used to evaluate the overall difference between the fermentation and non-fermentation. The significance was declared at p<0.05 and the trend was declared at p< 0.10.

The correlation between spectral parameters and nutrition were performed using the SAS CORR procedure (SAS Institute, Inc., USA). The Spearman option was used because some of the data was not normally distributed. Carbohydrate molecular spectral features were used to associate with carbohydrate nutritional characteristics.

Effect of fermentation duration on carbohydrate-related chemical profiles of by-products from coffee processing

Effect of microorganism fermentation duration on carbohydrate-related chemical profiles of by-products from coffee processing are showed in Table 1. Carbohydrate (CHO) content tended (p = 0.064) to linearly reduced with increasing microorganism fermentation duration from 0 to 28 days. The NFC content was also linearly reduced (p = 0.026) from 52% to 47% of DM. As to NDF and ADF, there was no linear and quadratic relationship (p>0.10) with microorganism fermentation duration. However, ADL was linearly reduced (p<0.01) from 1.34% to 1.20% of DM. The sugar content of by-products was quadratically reduced (p<0.01) with increasing fermentation duration. These results indicated that fermentation process with SBP had a dramatical impact on NFC, sugars, and lignin without significantly impacting on fibers (NDF and ADF). Yan et al [5] reported microorganism degradation in rumen changed total chemical composition contents not only structure carbohydrate but also non-structural carbohydrate in faba samples. Recently, Himmelsbach et al [22] and Yang et al [23] reported chemical profiles was altered after adding fibrolytic enzyme into whole plant faba bean silage for ruminants.

Table 1

Effect of fermentation duration on carbohydrate profile of by-products from coffee bean processing

Effect of fermentation duration on total digestible nutrient and energy profiles of by-products from coffee processing

The effect of microorganism fermentation length on TDN and energy profiles of by-products from coffee processing are presented in Table 2. Increasing fermentation duration results linearly decreased tdNDF (p = 0.009) and tended to linearly increase tdNFC (p = 0.079) without affecting tdCP. The tdFA content was quadratically increased from 0.93 to 1.22% of DM. Fermentation duration tended to have a quadratic impact (p = 0.069) on total TDN_1x content in coffee by-products. It reduced TDN_1x from 722 to 698 g/kg of DM. Yan et al [5] reported microorganism degradation in rumen changed digestible nutrient contents in faba samples. Yang et al [23] and Refat et al [7] also reported nutrient supply was altered after either adding fibrolytic enzyme or rumen microorganism degradation in faba silage and barley silage, respectively.

Table 2

Effect of fermentation duration on digestible nutrients and predicted energy value of by-products from coffee bean processing

As to energy value of coffee by-products, microorganism fermentation duration had a quadratic impact (p<0.10) on all digestible energy (DE_1x, DE_p3x), metabolizable energy (ME, ME_p3x), and net energy values (NE_L3x, NE_m, NE_g). These results indicated that microorganism fermentation process resulted in quadratic decrease in TDN and digestible (DE_1x, DE_p3x), metabolizable (ME, ME_p3x), and net energy values of fermented by-products from fermentation.

Effect of fermentation duration on carbohydrate subfraction profiles, intestinal digestion and milk value of by-products from coffee processing

The results of carbohydrate partitioning fractions, the degradable and undegradable fractions, and intestinal digestion of by-products from coffee processing samples according to CNCPS 6.5 system are showed in Tables 3–5 respectively. Microorganism fermentation (Table 3) significantly reduced (p<0.01) CA4 fraction from 9.1 to 1.1 g/kg of CHO and increased (p = 0.05) CB2 fraction from 42.8 to 46.5 g/kg of CHO in a quadratic pattern (p<0.01) with fermentation duration (0 to 28 days). The CB3 fraction were linearly increased (p = 0.017) and The CC fraction was linearly decreased (p< 0.01) with increasing fermentation duration (Table 3). These results indicated that microorganism fermentation process resulted in a quadratic decrease in water-soluble carbohydrates of CA4 fraction and a quadratic increase in soluble fiber of CB2 fraction and a linear decrease in available fibre of CB3 fraction of coffee by-products.

Table 3

Effect of fermentation duration on carbohydrate subfractions of by-products from coffee bean processing, partitioned according to the newly updated CNCPS 6.5

Table 4

Effect of fermentation duration on rumen degradable and undegradable content of carbohydrate subfractions in by-products from coffee bean processing, estimated according to updated CNCPS 6.5

Table 5

Estimated intestinal digestion of carbohydrate subfractions by using CNCPS in coffee beans submitted to different fermentation time

The microorganism fermentation resulted in a quadratic decrease (p<0.01) in RDCA4 fraction from 5.29% to 0.64% of DM without significantly (both linearly and quadratically) impacting on RDB2 and RDCB3 (Table 4). Total RDC were linearly decreased (p = 0.028) from 45.79% to 42.81% of DM with increasing fermentation duration (Table 4). Fermentation also had quadratic decrease (p<0.01) RUCA4 fraction from 1.58% to 0.10% of DM without significantly linearly and quadratically impacting on RUB2 and RUCB3 (Table 4). Total RUC was slightly linearly decreased (p = 0.005) with increasing fermentation duration (Table 4). These results indicated that fermentation process resulted in quadratic decrease in rumen degradable water-soluble carbohydrates and linear decrease in total rumen degradable fraction of coffee by-products.

Fermentation had no significantly linearly and quadratically impact on IDCB2 and IDCB3 (Table 5). These results indicated that fermentation process did not affect intestinal digestion of soluble fiber (CB2) and available NDF (CB3) of coffee by-products. As to feed milk value, the microorganism fermentation (Table 6) tended to quadratically reduce (p = 0.055) FMV fraction from 2.27 to 2.19 kg/kg of by-products. Refat et al [7], Himmelsbach et al [22], and Yang et al [23,24] reported adding fibrinolytic enzymes impacted CNCPS fractions and supply from whole plant faba silage to dairy cows.

Table 6

Effect of fermentation duration on feed milk value (FMV) in by-products from coffee bean processing based on net energy for lactation

Effect of fermentation duration on carbohydrate related molecular spectral profiles of by-products from coffee processing, revealed by vibrational molecular spectroscopy

The results of carbohydrate-related spectral profiles of by-products from coffee processing impacted by the microorganism fermentation duration are showed in Table 7. According to the results, several spectral parameters of fermented by-products were significantly or tended to be significantly linearly affected (p<0.01) by fermentation duration in terms of CC peak area at ca. 1,300 to 1,184 cm⁻¹ (p = 0.054) with an increase from 4.36 to 4.67 AU, lignin compounds peak height at ca. 1,507 cm⁻¹ (lignin height, p = 0.089) with an increase from 0.001 to 0.007 AU, STCHO_pk3 height at ca. 1,320 cm⁻¹ (p = 0.027) with an increase from 0.123 to 0.135 AU, CC height at ca. 1,237 cm⁻¹ (p = 0.092) with an increase from 0.076 to 0.080 AU, and HC height at ca. 1,150 cm⁻¹ (p = 0.041) with a decrease from 0.131 to 0.115 AU.

Table 7

Effect of fermentation duration on molecular spectral characteristics using molecular spectroscopy -FTIR-ATR in by-products from coffee bean processing impacted by fermentation duration

However, fermentation had no significant linear and quadratic impact (p>0.10) on STCHO peak area at ca. 1,488 to 1,184 cm⁻¹ and NTCHO NTCHO peak area at ca. 1,184 to 924 cm⁻¹. The fermentation duration also had no significantly linear and quadratic impact (p>0.10) on STCHO_pk1 height at ca. 1,418 cm⁻¹) and STCHO_pk2 height at ca. 1,397 cm⁻¹, NSCHO_pk1 height at ca. 1,102 cm⁻¹, NSCHO_pk2 height at ca. 1,078 cm⁻¹, and NSCHO_pk3 height at ca. 1,030 cm⁻¹.

These results indicated that fermentation process only affected specific spectral groups of functional chemicals and linearly increased ATR-FTIR spectral peak area and height intensity of cellulosic chemical compounds, spectral peak intensity height of lignin chemical compounds, spectral peak3 height of structural carbohydrate, and linearly decreased spectral peak height of hemicellulose chemical compounds in fermented coffee processing by-products.

Yan et al [5] reported that when rumen incubation increased, ATR-FTIR spectral peak intensity was reduced with exception of the 1st spectral peak intensity height of total CHO, peak intensity height of cellulosic chemical compound, spectral peak intensity area of cellulosic chemical compound as well as the 4th spectral peak intensity height of structural CHO. The alteration was relatively small [5]. However, Xin and Yu [6] also found that rumen incubation time were found to have significant effect to all spectral parameters. In this study, carbohydrate spectral intensity was increased with increasing fermentation time for only several spectral profiles with exception of STCHO_pk1 and STCHO_pk2 height and NSCHO_pk1, NSCHO_pk2, and NSCHO_pk3 height.

Xin and Yu [6] reported that after rumen incubation, the ATR-FTIR peak spectral intensity of the ruminal incubation residues were dramatically lower than spectral intensity from original feeds in comparison with our results from fermented by-product from coffee processing in this study and the reported results from whole plant faba bean forage by Yan et al [5]. Yan et al [5] also indicated that sometimes univariate molecular spectral analysis fails to discriminate the subtle spectral difference or spectral change between samples and treatments based on unique spectral intensity parameters while multivariate spectral analysis which utilize whole spectral absorption information can achieve. Comparison among three studies ([5,6] and current study), the results are inconsistent, which indicates the sophisticate impacts by rumen microbial degradation and enzymic digestion on the ATR-FTIR spectral intensity patterns. This is mainly due to the internal complicated structural make-up difference between the samples in three studies.

Revealing correlation-ship between feed quality and nutritional profiles and ATR-FTIR spectral intensity parameters and profiles of by-products from coffee processing

Therefore, the relationship with correlation analysis was conducted for by-products from coffee processing impacted by SBP microorganism fermentation duration is presented in Tables 8 and 9. Yan et al [5] demonstrated that during rumen fermentation and degradation, the alteration of ATR-FTIR spectral peak area and height intensity parameters significantly reflects nutrient digestion and degradation kinetics by ruminal microorganism. As for CHO chemical profiles (Table 8), CHO-related ATR-FTIR spectral peak intensity parameters of coffee processing by-products affected by SBP microorganism fermentation duration had significant correlations with ADF fibre content, especially with structural CHO’s 2nd spectral intensity peak height (r = −0.60) and total CHO’s 3rd spectral intensity peak height (r = −0.53), ADL content with the structural CHO’s 1st peak height (r = −0.80), and starch content with the ATR-FTIR spectral intensity peak height of lignin (r = −0.64). Yan et al [5] reported that for faba bean plant composition samples, ATR-FTIR spectral intensity peak parameters had stronger correlation with faba bean plant starch content, especially faba bean plant structural CHO’s ATR-FTIR spectral intensity peak area (r = −0.88), structural CHO’s 2nd spectral peak height (r = −0.92) and total CHO’s 3rd peak height (r = −0.95).

Table 8

Correlation between carbohydrate chemical profile and carbohydrate molecular structure spectra profiles in by-products from coffee bean processing impacted by microorganism fermentation duration

Table 9

Correlation between energy values and carbohydrate molecular structure spectral profiles in by-products from coffee bean processing impacted by fermentation duration

The TDN (Table 9), CHO-related spectral intensity parameters of fermented coffee by-products samples have significant correlation with TND_1x content, especially structural CHO’s ATR-FTIR 2nd peak height (r = 0.66) and structural CHO’s ATR-FTIR 3rd peak height (r = 0.62) and cellulosic chemical compound’s ATR-FTIR spectral intensity peak area (r = −0.62).

As for energy values (Table 9), carbohydrate-related spectral profiles of fermented coffee by-products samples have significant correlation with digestible, metabolizable and net energy content, especially peak area of total carbohydrate (r = 0.52 to 0.66), total CHO’s ATR-FTIR 1st spectral intensity peak height (r = 0.51 to 0.59), total CHO’s ATR-FTIR 2nd spectral intensity peak height (r = 0.66 to 0.74), total CHO’s ATR-FTIR 3rd spectral intensity peak height (r = 0.60 to 0.62) and cellulosic chemical compound’s spectral intensity peak area (r = −0.60 to 0.62). In the published correlation study, Yan et al [5] reported that there was a significantly strong correlation between ruminal degradation kinetics (degradation rate, potential degradable fraction, effective degradable NDF, unavailable CC fraction) and ATR-FTIR spectral peak intensity parameters of ruminal residues in faba bean plant composition samples.

The current and previous studies indicate that the ATR-FTIR spectral peak intensity changes and alteration during microorganism fermentation or ruminal digestion may be related or associated to the cleaving of bond during microbial fermentation and degradation and hence interfere digestion of cell wall by enzyme and microorganisms [5,25].