Genes for degradation and utilization of uronic acid-containing polysaccharides of a marine bacterium Catenovulum sp. CCB-QB4

Strain and chemicals

The QB4 cells (Lau et al., 2019) was cultured using high nutrient artificial seawater medium (H-ASWM) [0.5% tryptone, 2.4% (w/v) artificial sea salt mix (Marine Enterprises International), 10 mM HEPES, pH 7.6], as reported by Furusawa and co-workers (Furusawa et al., 2015). A total of 0.2% of polygalacturonic acid (pectic acid) (Nacalai Tesque), sodium alginate (Sigma-Aldrich), Gelzan™ CM (gellan gum) (Sigma-Aldrich), and ulvan were used as carbon source throughout the research.

Purification of ulvan

The ulvan was purified according to the method described by Tabarsa et al. (2012) with slight modification. 5 g of the dry powder of Ulva pertusa was dissolved in 100 mL of water and stirred at 65  °C for 3 h. The mixture was cooled and centrifuged at 10,000 g for 20 min at 15 °C using Sorvall™ RC 6 Plus Centrifuge (ThermoFisher Scientific). 4 volume of cold isopropanol was added into the supernatant to precipitate crude polysaccharide. The solution was left overnight at 4 °C. The crude ulvan was harvested by filtration, washed with 70% isopropanol several times, and dried overnight at 60 °C.

Determination of bacterial growth

The QB4 cells were inoculated in 10 mL of H-ASWM broth and cultured overnight at 30 °C. 0.1 mL of the cell suspension was inoculated into 100 mL of H-ASWM medium with 0.2% of each uronic acid as the carbon sources. Media with and without 0.2% glucose (Fisher Scientific) were used as positive and negative controls, respectively.

Each flask was inoculated with 0.1 mL of pre-cultured bacterial cell suspension and incubated at 30 °C with an agitation speed of 200 rpm on the orbital shaker. To monitor the bacterial growth, the colony-forming unit (CFU) counts were conducted by culturing the cells on H-ASWM agar plates because the optical density measurement was not suitable for the sample with gellan gum in which the broth was solidified by calcium ion contained in H-ASWM medium. 100 l of the cell suspension from each sample was collected every 3 h. The suspension was diluted into 900 l of H-ASWM medium. After that, sequential 10-fold serial dilutions were made, and 100l of aliquots of each dilution were plated on H-ASWM agar plates. Colonies were scored after incubation for 48 h at 30 °C, and the growth curve was constructed. The experiment was performed in triplicates. The generation time (G) was estimated by the method described by Aparna et al. based on the growth curve (Aparna, Parvathi & Kaniyassery, 2020). The number of generation (n) was determined by the method of n = (logb –logB)/log2, where B and b are the numbers of bacteria at the beginning of a time interval and at the end of the time interval, respectively. The generation time (G) was determined by the method of G = t/n, where t is the time interval (min).

To confirm the utilization of saturated galacturonate, the QB4 cells were culture in an H-ASWM medium with 0.2% saturated galacturonate. Media with and without 0.2% glucose was used as positive and negative controls, respectively. To monitor the bacterial growth, the optical density (OD_600nm) was measured at 3 h intervals for 30 h using UV spectrophotometer UV-1800 (Shimadzu). The experiment was performed in triplicates.

Genomic analysis of Catenovulum sp. CCB-QB4

The complete genome of QB4 deposited at GenBank under the accession number CP026604–CP026605 was determined by Lau et al. (2019). Genes involved in the carbohydrate-active enzyme (CAZymes) in QB4 were predicted using dbCAN pipelines (Yin et al., 2012). Several enzymes related to uronic acid metabolism in QB4 were predicted by Blastp at the National Center for Biotechnology Information (NCBI) server (Bethesda, MD, U.S.A.) and were found using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database (Kanehisa & Goto, 2000). Finally, the amino acid sequence similarity of all enzymes was confirmed using Blastp analysis with Protein Data Base (PDB). Signal peptide prediction was conducted by LipoP 1.0 (Juncker et al., 2003) and SignalP 5.0 (Armenteros et al., 2019). For sequence alignment of DEH reductase, the amino acid sequence of A1-R and A1-R’ from Sphingomonas sp. A1 and DEH reductase from Saccharophagus degradans 2-40 were obtained from GenBank (https://www.ncbi.nlm.nih.gov/genbank/). The sequence alignment with 4 sequences was conducted by ClustalW (Thompson, Higgins & Gibson, 1994) at PRABI Lyon-Gerland (https://npsa-prabi.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_clustalw.html).

Growth confirmation of QB4 using uronic acids

To confirm the growth of QB4 in the presence of uronic acids, the cells of QB4 were cultured in H-ASWM with alginate, pectin, ulvan, and gellan gum. As shown in Fig. 1, the cells of QB4 showed good growth in the presence of the uronic acid-containing polysaccharides and glucose except for negative control. The cells of QB4 reached the early stationary phase at 9 h in H-ASWM with alginate and gellan gum. The generation time of alginate and gellan gum was 76.76 and 74.07 min. However, the cell number of the sample with gellan gum was 5.7 times lower than that of the sample with alginate. On the other hand, the cells cultured by pectin and glucose reached the early stationary phase at 18 h. In the case of using ulvan, the cells reached stationary phase at 24 h, and the sample exhibited the highest cell number at 30 h in the experiment. The generation time of the samples with glucose, pectin, and ulvan was 91.37, 95.74, and 85.10 min, which were longer than that of the samples with alginate and gellan gum. Thus, QB4 cells exhibited a robust growth in the presence of uronic acids. Subsequently, we focused on genes for degradation and utilization of uronic acids in the genome of QB4.

Genes for alginate degradation and utilization

Alginate is a major polysaccharide found in the cell wall of brown algae and consisted of guluronate and mannuronate arranged as 1,4-linked polysaccharides. First, alginate is degraded into oligomeric or monomeric units by alginate lyases. Eight alginate lyase genes were found in the QB4 genome (Table 1). Ad1-PL6, Ad2-PL6 and Ad3-PL6, Ad4-PL7, Ad5-PL7, Ad6-PL7, Ad7-PL7, and Ad8-PL17 were classified into family PL6, PL7, and PL17 based on the CAZy database, respectively. As a result of Blastp search with Protein Data Bank (PDB), Ad proteins were similar to Alygc from Paraglaciecola chathamensis (Ad1-PL6 and Ad3-PL6), AlyQ from Persicobacter sp. CCB-QB2 (Ad2-PL6, Ad4-PL7, and Ad5-PL7), alginate lyase of Klebsiella pneumoniae (Ad6-PL7 and Ad7-PL7), and Saccharophagus degradans 2-40 (Ad8-PL17) (Table S1). LipoP 1.0 (Juncker et al., 2003) was used to predict the signal peptides and their type of each enzyme. Table 1 demonstrated that Ad1-PL6, Ad2-PL6, Ad4-PL7and Ad5-PL7 possessed type I signal peptide, which was cleaved by signal peptidase I, and Ad6-PL7, Ad7-PL7, and Ad8-PL17 have type II lipoprotein signal peptide, which was cleaved by signal peptidase II. The lipoprotein signal peptide referred to as “lipobox” plays an important role in anchoring the protein on the outer surface of the cell membrane after secretion and modification of N-terminal cysteine residue (Pugsley, Chapon & Schwartz, 1986; Hutcheson, Zhang & Suvorov, 2011). This suggested that Ad6-PL7, Ad7-PL7, and Ad8-PL17 may localize on the cell surface. In contrast, Ad1-PL6, Ad2-PL6, Ad4-PL7, and Ad5-PL7 may be released into the culture medium. On the other hand, Ad3-PL6, which does not possess any signal peptides, may locate at the cytoplasm.

Takase et al. (2010) described that an exotype oligoalginate lyase, A1-IV, from Sphingomonas sp. A1 degraded oligoalginates into monosaccharides, which are then nonenzymatically converted to 4-deoxy-L-erythro-5-hexoseulose uronic acid (DEH). Takagi and colleagues also reported that an alginate lyase, Alg7K, from Saccharophagus degradans showed exolytic activity and produced monosaccharides from oligoalginates (Takagi et al., 2016a). The amino acid sequence of the alginate lyase, Ad6-PL7, showed high similarity (77.3%) to Alg7K, suggesting that Ad6-PL7 might mediate hydrolysis of oligoalginate to produce monomers. The DEH is converted to 2-keto-3-deoxy-D-gluconate (KDG) by NADH or NADPH-dependent DEH reductase (Takase et al., 2010; Takase et al., 2014; Kim et al., 2016). A result of the BLASTp search demonstrated that the SDR family oxidoreductase, Au1, found in the QB4 genome (Table 1) showed high similarity to NADH (A1-R’, 60.9%) and NADPH (A1-R, 50.8%)-dependent DEH reductase. In addition, the gene aliments of DEH reductases showed that the TGXXXGX motif and catalytic triad (Ser, Tyr, and Lys), which are highly conserved in SDR family enzymes (Takase et al., 2014), were conserved in the SDR family oxidoreductase (Fig. S1). These results indicated that Au1 might function as DEH reductase. KDG kinase catalyzes the conversion of KDG to 2-keto-3-deoxy-phosphogluconate (KDPG), and then KDGP is converted into D-glyceraldehyde-3-phosphate and pyruvate by 2-dehydro-3-deoxy-phosphogluconate aldolase via the Entner -Doudoroff pathway. Two of the genes encoding the KDG kinase, Au2 and Au3, and the aldolase, Au4, and Au5, were found in the genome of QB4 (Table 1). Finally, pyruvate produced by the pathway goes into further metabolic pathways for generating energy. The five Au proteins also demonstrated a high degree of amino acid sequence similarity (>50%) to DEH reductase A1-R’ from Sphingomonas sp. A1 (Au1), KDG kinase Shigella flexneri (Au2 and 3), 2-keto-3-deoxy-6-phosphogluconate aldolase from Thermotoga maritima (Au4), and KDGP aldolase from Escherichia coli (Au5) (Table S1). As mentioned above, QB4 possessed many alginate lyases and enzymes responsible for alginate utilization. This is consistent with the robust growth of QB4 in the presence of alginate, as shown in Fig. 1.

Genes for pectin degradation and utilization

As a first step, pectin is depolymerized by pectin lyases or polygalacturonases. Pectin and pectate lyases are classified into five families of PLs. Although pectin lyases attack highly methyl-esterified pectin, pectate lyases specifically attack non-methylated polygalacturonate or methylated pectin with a very low degree (Hugouvieux-Cotte-Pattat, Condemine & Shevchik, 2014). These enzymes degrade pectin to unsaturated pectic-oligosaccharides and disaccharides.

In the CAZy database, seven pectate lyases belonging to family PL1 (Pd1_PL1, Pd2_PL1, Pd3_PL1, Pd4_PL1, and Pd5_PL1), PL3 (Pd6_PL3), and PL10 (Pd7_PL10) family pectate lyases were found in the genome of QB4. The result of Blastp search using the PDB database on Pd proteins also demonstrated that these Pd proteins were similar (>45%) to pectate lyase (Pd1_PL1, Pd2_PL1, Pd3_PL1, Pd5_PL1, Pd6_PL3, and Pd7_PL10) and pectinesterase (Pd4_PL1) from other bacterial species (Table S2). CBM13 found in Pd1_PL1 and Pd6_PL3 and CBM35 found in Pd2_PL1 and Pd3_PL1 were capable of binding to multi-ligands, such as arabinan, arabinoxylan, and pectin (Fujimoto, 2013; Dhillon et al., 2018). However, CBM77 contained in Pd2_PL1 recognized homogalacturonan (Fujimoto, 2013). Pectin methylesterase (PME) domain from family 8 Carbohydrate Esterase (CE8) was found in Pd4_PL1 and Pd5_PL1. PMEs produce de-esterified homogalacturonan by catalyzing the de-esterification of the methoxyl group of the pectin; as a result, the products are effectively degraded by pectate lyases (Kashyap et al., 2001). Besides, although Pd1_PL1, Pd3_PL1, Pd4_PL1, and Pd7_PL10 possessed type I signal peptide, Pd5_PL1 and Pd6_PL3 had lipobox, indicating that both two lyases localize on the cell surface. Hence, these results suggested that Pd5_PL1 plays an important role in the pectin degradation of QB4 due to its de-esterification and depolymerization activities and its localization.

The unsaturated disaccharides generated by the pectate lyases are converted to 5-keto-4-deoxyuronate (DKI) by two distinct processes. One is that the disaccharides are degraded by oligogalacturonate lyase belonging to family PL22, and then the product, Δ-4,5-unsaturated galacturonate, is linearized into DKI by KdgF (Hobbs et al., 2016). The other is that unsaturated galacturonyl hydrolases belonging to GH105 degrades the disaccharides and directly releases DKI (Hobbs et al., 2019). In the QB4 genome, no PL22 enzymes were found in the genome. However, three GH105 proteins, Pd8_GH105, Pd9_GH105, and Pd10_GH105, were present in the genome (Table 2). These enzymes showed a high degree of similarity (49∼67%) on unsaturated rhamnogalacturonyl hydrolases (YteR) based on Blastp search with PDB database. These results indicated that oligogalacturonate might directly convert to DKI by GH105 proteins in QB4. DKI is converted to 2-keto-3-deoxygluconate (KDG) by two enzymes, a DKI isomerase (KduI) and a 2-dehydro-3-deoxy-D-gluconate 5-dehydrogenase (KduD) (Hobbs et al., 2019). The pathway analysis based on KEGG showed that four of KduI (Pu1, Pu2, Pu3, and Pu4) and KduD (Pu5, Pu6, Pu7, and Pu8) were present in the genome of QB4. Blastp search with the PDB database also exhibited that these enzymes showed a high degree of similarity (63∼71%) on KduD and KduI from Enterococcus faecalis, Escherichia coli, and a pectolytic bacterium, Pectobacterium carotovorum (Table S2). These results indicated that QB4 uses pectin as a carbon source for cell growth and is consistent with the result shown in Fig. 1.

The pathway analysis also showed the other pathway using saturated galacturonate (monosaccharide) produced by polygalacturonases. As mentioned above, although QB4 possesses seven pectate lyases, only two polygalacturonases (GH28), Pd11_GH28 and Pd12_GH28, which were highly similar (>66%) to polygalacturonase from pectolytic bacteria, such as Erwinia carotovora and Thermotoga marítima (Table S2), were found in the genome. Based on SignalP 5.0 analysis, it was found that Pd11_GH28 and Pd12_GH28 contained Tat and type I signal peptide, respectively. Saturated galacturonate is converted to D-tagaturonate by glucuronate isomerase. Three glucuronate isomerases, Pu9, Pu10, and Pu11, which were highly similar (> %) to glucuronate isomerase of Salmonella enterica subsp. enterica serovar Typhimurium and uronate isomerase of Caulobacter vibrioides CB15 (Table S2), were found in the QB4 genome. The conversion of D-tagaturonate to KDG occurs through two distinctly different pathways. One is that the process consists of two steps involving tagaturonate reductase and altronate hydrolase and one intermediate, D-altronate (Richard & Hilditch, 2009). The other is that tagaturonate is converted via three steps involving tagaturonate epimerase, fructuronate reductase, and mannonate dehydratase and two intermediates, D-fructuronate and D-mannonate (Valk et al., 2020). As shown in Table 2, QB4 possessed all genes involved in the two different pathways, namely tagaturonate reductase (Pu12), altronate hydrolase (Pu13), tagaturonate epimerase (Pu14), fructuronate reductase (Pu15), and mannoate dehydratase (Pu16). Homologous enzymes of these proteins were also found in the PDB database (Table S2). This result suggested that QB4 is also capable of using saturated galacturonate as a carbon source. In order to confirm this suggestion, the QB4 cells were cultured in H-ASWM broth with saturated galacturonate. As shown in Fig. 2, although the cells with glucose demonstrated robust growth, the cells were unable to grow in the broth with saturated galacturonate as well as the negative control in the incubation period. This result indicated that the unsaturated galacturonate utilization pathway is the main pathway to utilize polygalacturonic acid of QB4. Interestingly, genes for all pectate lyases, polygalacturonases, and one set of KduI (Pu4), KduD (Pu8), glucuronate isomerase (Pu11), and tagaturonate reductase were located in a plasmid (https://www.ncbi.nlm.nih.gov/nuccore/NZ_CP026605.1) found in QB4 (Table 2).

Genes for ulvan degradation and utilization

Ulvan lyases classified into 5 families, PL24, PL25, PL28, PL 37, and PL40, in the CAZy database (Li et al., 2020) were isolated from several marine bacteria, such as genera, Alteromonas, Pseudoalteromonas, Formosa, and Nonlabens (Kopel et al., 2016; Qin et al., 2018; Ulaganathan et al., 2018a; Ulaganathan et al., 2018b; Reisky et al., 2019). Ulvan lyases cleave between L-rhamnose 3-sulfate (Rha3S) and D-glucuronic acid (GlcA) or L-iduronic acid (IdoA). Fourteen ulvan lyases in QB4 were classified into three PLs families, PL24 (Ud1_PL24, Ud2_PL24, Ud3_PL24, Ud4_PL24, Ud5_PL24, Ud6_PL24, Ud7_PL24, Ud8_PL24, and Ud9_PL24), PL25 (Ud10_PL25, Ud11_PL25, Ud12_PL25, and Ud13_PL25) and PL40 (Ud14_PL40). Ud proteins belonging to the PL24 family were highly similar (>62%) to short ulvan lyase of Alteromonas sp. LOR, and Ud proteins belonging to the PL25 family were highly similar (>57%) to ulvan lyase-PL25 of Pseudoalteromonas sp. PLSV (Table S3). The number of the ulvan lyases was higher than that of Formosa agariphila (three PL40 proteins) (Reisky et al., 2019) and Alteromonas sp. LOR (one PL24 and one nonclassified protein) (Foran et al., 2017). Ud6_PL24 and Ud9_PL24 and the enzymes belonging to PL25 possessed lipobox. Although Ud2_PL24, Ud3_PL24, Ud4_PL24, Ud5_PL24, and Ud8_PL24 contained type I signal peptide, the remaining enzymes did not have any signal peptides. Although only one pectin lyase had a lipobox, many ulvan lyases possessed lipobox as well as alginate lyases, indicating that alginate and ulvan degradation activity of QB4 may be more effective than pectin degradation of the strain.

The unsaturated uronyl residue at the non-reducing end of the oligosaccharides produced by the ulvan lyases may be released by unsaturated glucuronyl hydrolases (GH105). Five putative unsaturated glucuronyl hydrolases, Ud15_GH105, Ud16_GH105, Ud17_GH105, Ud18_GH105, and Ud19_GH105 were found in the QB4 genome and showed similarity on unsaturated beta-glucuronyl hydrolase of Nonlabens ulvanivorans, which is known as a ulvan-degrading bacterium, based on BLASTp search (Table 3; Table S3) (Collén et al., 2014). After forming DKI by the enzymes, the following process may be the same as pectin utilization.

In the ulvan-degrading process of F. agariphila, Rha3S-Xyl-Rha3S was also the main product by ulvan lyases (Reisky et al., 2019). First, Rha3S at the non-reducing end is desulfated by sulfatases. The amino acid sequences of two sulfatases, Ulu1 and Ulu2, in the QB4 genome showed high similarity (77.3 and 86.2%) to the sulfatases, WP.032096151.1 and WP.632096147.1, located into ulvan utilization loci of Alteromonas sp. LOR (Foran et al., 2017), indicating that the two sulfatases may involve desulfation of the Rha3S. The Rha3 will remove the Rha3-Xyl-Rha3S by α-L-rhamnosidase. Five α-L-rhamnosidases classified into GH78 family, Ud20_GH78, Ud21_GH78, Ud22_GH78, Ud23_GH78, and Ud24_GH78, were found in the QB4 genome and showed similarity on rhamnosidase from Bacillus sp. GL1, Dictyoglomus thermophilum, Streptomyces avermitilis, and Bacteroides thetaiotaomicron (Table S3). Ud21_GH78, Ud22_GH78, Ud23_GH78, and Ud24_GH78 contained lipobox, suggesting that these enzymes are located on the cell surface. In addition, Ud20_GH78 and Ud22_GH78 possessed the CBM67 domain, which binds L-rhamnose in a calcium-dependent manner (Fujimoto et al., 2013), suggesting that the two α-L-rhamnosidases are the main components to generate Rha3.

Rha3 metabolic pathway was described by Reisky et al. (2019). α-L-rhamnose is converted to β-L-rhamnose by L-rhamnose mutarotase. Next, isomerization of the β-L-rhamnose to L-rhamnulose is catalyzed by rhamnose isomerase. The product is converted to L-rhamnulose-1-phosphate by pentulose/hexulose kinase (rhamnulokinase), and subsequently, the L-rhamnulose-1-phosphate is cleaved into L-lactaldehyde and dihydroxyacetone phosphate by rhamnulose aldolase. Finally, the L-lactaldehyde is converted to pyruvate by aldehyde dehydrogenase and lactate dehydrogenase. Table 3 displayed that the corresponding genes (Ulu3, Ulu4, Ulu5, Ulu6, Ulu7, and Ulu8) were found in the QB4 genome, and these proteins showed similarity on corresponding proteins in the PDB database (Table S3). This result indicated that Rha3 is metabolized by QB4 cells. The genes involving the metabolic pathway downstream of the mutarotation in F. agariphila form a gene cluster (Reisky et al., 2019). However, the genes in QB4 were randomly distributed in the genome.

Xyl will be released by β-xylosidase classified into GH3 and GH43 families. Based on the CAZy database, seven GH43 (Ud25_GH43, Ud26_GH43, Ud27_GH43, Ud28_GH43, Ud29_GH43, Ud30_GH43, and Ud31_GH43) proteins were found in the QB1 genome while F. agariphila possessed two each of GH3 and GH43 proteins (Reisky et al., 2019). Ud25_GH43, Ud29_GH43, and Ud31_GH43 were similar to glycoside hydrolases from Zobellia galactanivorans (50.00%) and Halothermothrix orenii H 168 (45.75 and 39.10%) (Table S3). The D-xylose is converted to D-xylulose-5-phosphate via xylose isomerase and xylulose kinase, and then the product is passed to the pentose phosphate pathway. Two genes encoding xylose isomerase (Ulu9) and xylulose kinase (Ulu10) were found in the QB4 genome based on RAST server annotation and formed a gene cluster with transcription repressor, xylR (Ulu11) (Table 3). These results suggested that QB4 cells may use not only glucuronic acid but also L-rhamnose and D-xylose as carbon sources. In addition, the presence of numerous genes encoding ulvan lyase may promote strong ulvan degradation and its utilization. The QB4 cells indeed exhibited a robust cell growth in the presence of ulvan as well as that in the presence of alginate, as shown in Fig. 1.

Genes for gellan gum degradation and utilization

Gellan gum is depolymerized gellan lyases classified into PL33 family proteins. Gellan lyases are found in several bacterial species, such as Bacillus sp. GL1, Geobacillus stearothermophilus 98, and Opitutaceae bacterium TAV5 (Hashimoto et al., 1997; Derekova et al., 2006; Helbert et al., 2019). Gellan gum is degraded to tetrasaccharides composed of [→3)- β-D-glucose-(1 →4)- β-D-glucuronic acid-(1 →4)- β-D-glucose-(1 →4)- α-L-rhamnose-(1 →] by gellan lyases. The tetrasaccharides are completely degraded to monosaccharides by β-D-glucosidase (GH1, GH2, GH3, GH5, GH9, GH16, GH30, GH39, and GH116), unsaturated glucuronyl hydrolase (GH88 and GH105), and α-L-rhamnosidase (GH78) (Hashimoto et al., 2003).

Figure 1 demonstrated that QB4 might also use gellan gum as a carbon source for its growth. However, gellan lyases were not found in the genome. As shown in Table 3, QB4 possessed five unsaturated glucuronyl hydrolases and five α-L-rhamnosidases (Table 3). The hydrolysates, DEH, and α-L-rhamnose, may be metabolized by the pathway described above. In addition, one β-D-glucosidase classified into the GH1 family (Gd1-GH1) was also found in the genome, and Gd1-GH1 showed high similarity (63.23%) on β-glucosidase A of Hungateiclostridium thermocellum in the PDB database (Table 4; Table S4). β-D-glucose produced by the β-D-glucosidase is converted to D-glucose-6-phosphate by glucokinase, and then the product is converted to β-D-fructose-6-phosphate, which is an intermediate of glycolysis, by glucose-6-phosphate isomerase. One of these genes, Gu1 and Gu2, were found in the QB4 genome (Table 4) and were similar on glucokinase and glucose-6-phosphate Isomerase of E. coli in PDB database, respectively (Table S4), indicating that β-D-glucose is converted to β-D-fructose-6-phosphate and is metabolized by glycolysis.