The S. acidocaldarius strain MR31 (Reilly and Grogan, 2001), an uracil auxotroph with partial deletion on the gene encoding orotate phosphoribosyltransferase (pyrE) of S. acidocaldarius DSM 639, was used as a wild type in this study. S. acidocaldarius MR31 and the other strains were grown aerobically in the YT medium at 77°C, which is made from the Brock’s basal medium (Brock et al., 1972) with slight modifications. The medium was supplemented with 0.01% (w/v) of tryptone, 0.005% (w/v) of yeast extract. For the growth of S. acidocaldarius MR31 and ΔamyA strain, 20 μg/ml of uracil was supplemented. The pH was adjusted to 3.3 with 6 M H₂SO₄. In total, 0.2% (w/v) of glucose or other sugars were added as a carbon source. The growth was monitored at an optical density (OD) of 600 nm using spectrophotometer. The Escherichia coli strain DH5α used for cloning was grown at 37°C in the LB medium and ampicillin (100 μg/ml) or other antibiotics were added if necessary. The strains and plasmids used in this study were listed in Table 1.

Putative genes related to the glycogen metabolism of S. acidocaldarius were expected by BLAST search with the enzymes studied in E. coli and L. acidophilus, and the operon which consisted of the putative genes was named as glg operon. To confirm the involvement of putative glg genes in the glycogen metabolism of S. acidocaldarius, five chromosomal glg genes, namely glgA (saci_1201, GenBank No. AAY80548.1), gaa (saci_1198, GenBank No. AAY80545.1), glgX (saci_1197, GenBank No. AAY80544.1), glgP (saci_0294, GenBank No. AAY79710.1), and amyA (saci_1200, GenBank No. AAY80547.1), were disrupted by insertional deletion mutation, except the amyA gene. The upstream and downstream regions of the target gene were amplified by PCR and fused into one fragment with BamHI restriction enzyme site by overlap PCR. The fused fragment was cloned into the suicide vector (pGEM-T) and the pyrE gene, amplified from the genomic DNA of S. solfataricus P2. This fragment was inserted between the upstream and downstream regions by restriction with BamHI and ligation. The S. acidocaldarius MR31 was transformed with the constructed vector using electroporation. Genetic disruption by the insertion of pyrE gene was confirmed by PCR, and the mutants were grown in the YT medium without uracil for eliminating the plasmid fragment.

The amyA gene was partially deleted by the markerless deletion method as the gene is present in the upstream region of glgA gene, which can cause the inactivation of glgA gene by polar effect. To construct the ΔamyA mutant, markerless deletion was conducted by slightly modifying the method of Wagner et al. (2012). The middle region of saci_1200 gene (GOI fragment) and the downstream region of saci_1200 gene (R fragment) were amplified and fused by overlapping PCR. The fused fragment, which harbored BamHI restriction site between the GOI and R fragment, was cloned into the pGEM-T vector. Then, the constructed vector was cut with BamHI. The upstream region of saci_1200 gene (L fragment) and pyrE gene (E fragment) of S. solfataricus were also amplified and fused by overlapping PCR, further resulting in the construction of E-L fragment, that has BamHI restriction site on both the regions. The E–L fragment was digested with BamHI and ligated into the pGEM-T vector harboring GOI-L fragment to make GOI–E–L–R sequence. From the constructed vector, the E–L–R fragment was amplified and ligated into pGEM-T vector, thereby resulting in the construction of pΔamyA vector. A constructed vector was inserted into the chromosomal DNA of S. acidocaldarius MR31 by homologous recombination at the L or R region.

For the construction of amyA complementation strain, promoter and saci_1200 gene (P-saci_1200) was amplified. Terminator region was amplified from the downstream of gdhA (saci_0155), and fused with P-saci_1200 fragment, resulting in construction of P-saci_1200-ter fragment. The fragment was treated by SacII and ligated into pC vector. Constructed vector was transformed into amyA deletion mutant, resulting in the construction of the amyA complementation strain.

Sulfolobus acidocaldarius MR31 and deletion mutants were grown in the YT medium supplemented with 0.2% of glucose or any other specific sugar, and 20 μg/ml of uracil was added when needed. The cell growth was measured by OD of 600 nm and growth curve was fitted to sigmoid equation with three parameters. The cells were harvested at 18, 30, 42, 60, 78, and 96 h, respectively. The samples for measuring glycogen contents were prepared by the following steps: washing with buffer A (20 mM of sodium phosphate, pH of 7.4, 450 mM of NaCl) three times, lysis by sonication, removal of the cell debris by centrifugation (2,000 × g, 10 min, and 4°C) and separation of soluble and membrane-bound particles by centrifugation (16,000 × g, 40 min, and 4°C). The membrane-bound fractions were resuspended to buffer A with the same volume of soluble fractions. For the normalization of each sample, the protein concentration of each cell before the separation of soluble and membrane-bound fractions was measured by the Bradford assay (Bradford, 1976).

The content of glycogen was quantified with minor modification as described in previous research works (Dauvillée et al., 2005; Hwang et al., 2013). In total, 0.35 units of amyloglucosidase (Sigma Aldrich, St. Louis, MO, United States) was incubated with 20 μL of samples in 100 μL of 50 mM sodium acetate buffer (pH 4.5) for 30 min at 60°C. The enzyme reaction was quenched by incubating at 100°C for 5 min, followed by centrifugation to remove proteins. Half of the reaction mixture or proper dilution portion was added to 100 μL of glucose oxidase/peroxidase reagent (Sigma Aldrich) and incubated at 37°C for 30 min. The enzyme reaction was stopped by the addition of 100 μL of 6 M H₂SO₄. The glucose released from glycogen by the reaction of amyloglucosidase was determined colorimetrically at 540 nm. The content of glycogen was calculated by comparing the standard of glucose equivalent that was released from glycogen from oyster (Sigma Aldrich) with known concentrations ranging from 0 to 3 μg. The calculated glycogen content was normalized by the protein concentration measured from the same volume of sample to exclude effect of the number of cells in the reaction mixture. To exclude the false-positive reaction of intracellular sugars on the glycogen content measurement, the same reaction without amyloglucosidase was conducted as control, followed by glucose oxidase/peroxidase reaction. The glucose equivalent measured from the serial reaction without amyloglucosidase was termed as pseudoglycogen and used for the normalization of glycogen content in the soluble fraction. Each value was calculated from two independent samples.

The MR31 strain and amyA gene deletion strain were grown in the YT medium supplemented with 0.2% glucose. The cells were harvested by centrifugation (2,000 × g, 30 min, and 4°C) when OD reached around 0.9, and resuspended in a buffer of 20 mM sodium phosphate (pH 7.4) with 0.5 M sodium chloride, followed by lysis through the sonication and fractionation of cell-free extract. To indirectly detect the α-amylase activity, 1 mg of soluble fraction was incubated overnight with 0.1% of glycogen in 50 mM sodium acetate buffer (pH 4.5) at 75°C. The reducing sugars released from the glycogen was detected using 3,5-dinitrosalicylic acid solution and detected colorimetrically at 575 nm. The released product from the reaction with soluble fraction of each strain was compared to the standard curve of reducing sugar calculated with glucose. The experiment was conducted with triplicates.

Branch chain-length distribution of glycogen after treatment of the cell-free extract of MR31 and ΔamyA mutant was examined by MALDI-TOF mass spectrometry. After treatment on glycogen, the samples were precipitated with 3 volumes of ethanol for an hour and the precipitants were collected by centrifugation (16,000 × g, 40 min, and 4°C) and dried. The dried samples were treated with 10 U of isoamylase (Megazyme, Chicago, United States) in 50 mM sodium acetate buffer (pH 4.0) for 72 h at 40°C. The reaction was quenched by boiling for 10 min and the samples were filtered through 0.2 μm pore. MALDI-TOF analysis was conducted by the method described by Broberg et al. (2000) with modification. Samples were mixed with the same volume of 2,5-dihydroxybenzoic acid (20 mg/ml in water/acetonitrile, 1:1) and dried on the target plate (MTP 384 target plate ground steel BC. #8280784, Bruker). Bruker Autoflex Max (Bruker, MA, United States) was utilized for the MALDI-TOF with linear positive mode and the mass spectrum was detected ranging from 400 to 3,500 m/z. Calibration was done with peptide calibrant (Bruker, MA, United States) and maltoheptaose (Sigma, St. Louis, United States). The intensity of peaks, which is responsible for the degree of polymerization (DP), were integrated and distribution of DP was calculated. As mass spectra showed regular pattern that increases molecular mass of 162, which is a molecular weight of glucose in maltooligomer, mass spectra were converted to DP using maltoheptaose as a standard.

To determine the strength of putative promoter that lies in the upstream regions of genes related to glycogen metabolism, the cells that harbor the β-galactosidase gene fused with the putative promoter region were cultivated in the YT medium supplemented with 0.4% glucose. The cells were harvested by centrifugation at overall growth phases and resuspended in buffer A and lysed by sonication. The lysed cells were fractioned by the method as described in glycogen quantification. The β-galactosidase activity was measured by using p-nitrophenyl β-glucopyranoside (pNPG) as a substrate. The assay was conducted in 100 μL of 50 mM sodium acetate buffer (pH 5.0) containing 5 mM of pNPG and 12 μg of soluble cell-free extract. The reaction mixture was incubated at 80°C for 2 min and immediately stopped by addition of 100 μL of 1 M sodium carbonate. The absorbance of pNP released from pNPG was measured at 420 nm. One unit of the enzyme was determined as the amount of enzyme that produces 1 μmol of pNP per minute. The experiment was conducted with triplicates.

The cell death rates of the wild type MR31 as well as the glycogen abundant and deficient mutant strain were compared in the nutrient starvation condition. To compare the death rate, each cell was grown in 400 ml of the YT medium supplemented with 0.2% glucose at 77°C. The cells were then harvested and washed twice with the Brock’s basal medium when the OD reaches at 0.4–0.5. The washed cells were inoculated into 100 ml of fresh Brock’s basal medium (without tryptone nor sugar) with the final OD of 0.5. The cell growth was monitored every 6 or 12 h by measuring the OD. The decreased rate of OD after inoculation into the fresh medium was calculated by simple regression using the OD and time at Y- and X-axis, respectively. The rate of regression was negatively converted and termed as deathrate.

Viable cells were counted by plating. After the serial dilution, cells were dropped on plate supplemented with 0.2% tryptone and 0.2% glucose. After the incubation of 3 days, colonies which can be detected by naked eyes were counted. A 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay was conducted to compare the relative viable cells in each time points. Method of MTT assay was modified from the MTT assay conducted in bacteria (Grela et al., 2018). A total of 3 ml of cultured cells were harvested and 1 mM MTT was incubated with the cells in 100 mM sodium acetate (pH 4.5) buffer for 20 min at 77°C. Supernatants were removed by centrifugation (16,000 × g, 1 min, and 4°C), and formazan was solubilized in isopropyl alcohol and measured colorimetrically at 550 nm.

Additionally, the morphology of MR31, glg deletion strain, and glgX deletion strain in stress conditions were visualized by Scanning Electron Microscopy (SEM). Samples were prepared by the method of Zhang et al. (2019) with slight modification. A total of 10 μl of cell cultures were dropped on cover glass and dried. Fixation and wash steps were conducted as described in the study of Zhang et al. but the final step using a critical point dryer was excluded. Samples were coated with Pt and imaged equipped with a Zeiss GeminiSEM 500 (Carl Zeiss, Oberkochen, Germany) at 5 kV with various magnifications.

The difference between two samples was confirmed by t-test. To compare the death rate among S. acidocaldarius MR31, Δglg, and ΔglgX mutants, one-way ANOVA was used, and further the Tukey test was used for post-hoc analysis. For the comparison of β-galactosidase activity depending on growth phase and promoter, two-way ANOVA was conducted, and the Tukey test was introduced for post-hoc analysis.

A previous study (König et al., 1982) detected glycogen in S. acidocaldarius as granular form aggregated with several proteins including GlgA. To confirm the cellular location of glycogen, S. acidocaldarius cells were harvested at each growth point and the cellular components in the cells were separated into soluble and membrane-bound fractions, respectively. As S. acidocaldarius has been known to synthesize glycogen in various carbon sources (König et al., 1982), we measured the glycogen content of S. acidocaldarius MR31 in the presence of different sugars. The cells were grown in the YT medium supplemented with 0.2% of glucose, xylose, sucrose, and dextrin, respectively, and the glycogen content in each cell was analyzed. As shown in Figure 1, S. acidocaldarius MR31 grown in different sugars accumulated glycogen in the exponential growth phase and the accumulated glycogen gradually decreased afterward. S. acidocaldarius MR31 accumulated 0.12 and 0.22 mg of glycogen in soluble and membrane-bound fraction per 1 mg of protein, respectively, at the early exponential phase when 0.2% glucose was provided. The glycogen content was maintained at an average of 0.27 mg of glycogen until the middle exponential growth phase and dropped to an average of 0.06 mg in the late exponential and stationary growth phases. The ratio of glycogen content in the soluble glycogen per membrane-bound glycogen (s/m) increased from 0.5 to 21.4 during the cell growth. Later, this ratio dropped to 7.0 at the stationary phase, thereby suggesting that the glycogen content in membrane-bound fraction decreased more drastically during the cell growth.

The cells grown in the YT medium supplemented with glucose (Figure 1A), sucrose (Figure 1C), and dextrin (Figure 1D) showed similar glycogen contents at an early exponential phase and a similar decline pattern. During the cell growth from the early log to middle exponential growth phase, S. acidocaldarius MR31 accumulated an average of 0.27 and 0.23 mg of glycogen on the supplementation of sucrose and dextrin, respectively. However, the cells grown with xylose (Figure 1B) have less glycogen content than the cells grown in the YT medium supplemented with glucose. Although S. acidocaldarius MR31 accumulated more than 0.2 mg of glycogen, the cells grown with 0.2% xylose accumulated an average of 0.15 mg of glycogen. S. acidocaldarius MR31 grown in the YT medium without sugar accumulated an average of 0.15 mg of glycogen during early and middle exponential phases (Figure 1E), indicating that S. acidocaldarius does not utilize xylose for the glycogen accumulation. The accumulated glycogen in the cells without sugar declined dramatically to 0.06 mg of glycogen at 30 h, suggesting that the accumulated glycogen of S. acidocaldarius MR31 grown without sugar was consumed rapidly compared to the cells grown with various sugars. The differences of carbon sources appeared to be the reason of less glycogen accumulation in the grown cells with xylose as compared to the cells supplemented with glucose, sucrose, and dextrin.

For identifying the gene cluster responsible for the glycogen metabolism, the glg operon of well-studied strains was compared with the putative glg operon of S. acidocaldarius. In Escherichia coli, the glg operon comprises five genes, namely glgB, glgX, glgC, glgA, and glgP. The gene cluster of Lactobacillus acidophilus harbors seven genes, and the amy gene acts as an α-amylase/glycogen debranching enzyme, thereby replacing the activity of the glgX gene in E. coli (Goh and Klaenhammer, 2013). The putative enzymes related to glycogen metabolism in S. acidocaldarius DSM639 are encoded by gene clusters consisting of six genes, namely glgX (saci_1197), gaa (saci_1198), glgC (saci_1199), amyA (saci_1200), glgA (saci_1201), and glgP (saci_0294), which are annotated as the putative glycogen debranching enzyme, glucoamylase, glucose-1-phosphate adenyltransferase, α-amylase, GlgA, and GlgP, respectively. Unlike the gene arrangement in E. coli (Serres et al., 2001) or L. acidophilus (Goh and Klaenhammer, 2013), the glg operon in S. acidocaldarius is divided into two operons and the putative glgP gene locates far from the glg gene clusters (Figure 2). The gene arrangement of S. acidocaldarius was same as the arrangement of putative glg operon of S. solfataricus (She et al., 2001) and S. tokodaii (Kawarabayasi et al., 2001), suggesting that the glg operon is conserved in related species. S. acidocaldarius and S. tokodaii also shares homologs which were predicted as 2-iminobutanoate/2-iminopropanoate deaminase (saci_1195, STK_08110), CBS domain protein (saci_1196, STK_08130), and a protein which contains oxidoreductase molybdopterin binding domain (saci_1202, STK_08190). Interestingly, Thermoproteus tenax, which is a member of Crenarchaeota, harbors amyA gene and lacks glgB gene like Sulfolobus, although the gene arrangement was different (Siebers et al., 2004). Unlike T. tenax, Thermococcus kodakarensis, which is a member of Euryarchaeota, was shown to have the glgB gene searched via BLAST using GlgB of E. coli.

The result of the reverse transcription PCR showed that the glgC, the gaa, and the glgX genes were co-transcribed as an operon, and the amyA and the glgA genes were also synthesized in a single mRNA transcript (see Supplementary Figure 1). To identify the conserved region among the promoters of glgP, glgC–gaa–glgX, and amyA–glgA, WebLogo was used with 50 bp upstream of saci_0294 (glgP), saci_1199 (glgC), and saci_1200 (amyA) (Crooks et al., 2004). The putative promoter of glgX, amyA, and glgP has conserved A/T rich sequences on the -31 to -26 bp upstream of the start codon (see Supplementary Figure 2A). Intergenic region of gaa–glgX, which could be a promoter region for glgX also has the conserved A/T rich sequences (see Supplementary Figure 2B). The glgX gene, which lies 172 bp upstream region of the gaa gene, is shown to also have its own promoter, further making it possible to be transcribed alone. The scattered location of the genes related to the glycogen metabolism also existed in C. glutamicum (Ikeda and Nakagawa, 2003). Unfortunately, the putative GlgB, which is encoded by the glgB gene and forms α-1,6-linked branches of glycogen, was not found in S. acidocaldarius by the BLAST search with GlgB of E. coli or L. acidophilus.

To confirm the role of the putative glg operon in the glycogen metabolism of S. acidocaldarius, the glycogen content of glg operon deletion strain was compared to that of the MR31 strain. For constructing the glg deletion mutant, the gene cluster reaching from glgX to glgA was removed by insertional deletion with the pyrE gene (see Supplementary Figure 3). While glg deletion mutant showed a slower increase of absorbance at 600 nm in the early log phase, the maximum OD was similar between the two strains (Figure 3). Despite few differences in the cell growth, no glycogen was detected in both of membrane-bound and intracellular fractions of the glg mutant, thereby suggesting that the putative glg operon plays a crucial role in the glycogen metabolism of S. acidocaldarius.

To identify the functional roles of GlgA and AmyA in glycogen metabolism, each gene was disrupted by insertional deletion and markerless mutation, respectively. Each deletion mutant was grown in the YT medium supplemented with 0.2% glucose and the content of intracellular glycogen was measured in various growth phases. Unlike MR31 strain (Figure 4A), both strains did not show the accumulation of intracellular glycogen via cell growth (Figures 4B,C). As expected, the glgA deletion mutant showed the lack of accumulated glycogen, as glgA encodes glycogen synthase, which acts as a key enzyme in the formation of α-1,4-glycosidic chain.

Unlike the amino acid sequence-based expectation that the amyA gene encodes α-amylase, which is predicted to be involved in the degradation of glycogen, the mutant with the amyA gene deletion failed to accumulate the intracellular glycogen (Figure 4C). When the vector that can express amyA gene was transformed to amyA deletion mutant, the complement cells accumulated glycogen again (see Supplementary Figure 4), suggesting that AmyA plays an essential role in glycogen synthesis.

To purify and characterize the AmyA activity, various expression systems were used, but all the expression systems used for the expression of amyA gene in S. acidocaldarius or E. coli failed to do the same. To indirectly identify the enzyme activity of AmyA, the hydrolysis abilities of the glycogen by using total cell-free extract in the MR31 strain and amyA deletion strain were compared. Figure 5 shows that the cell-free soluble fraction of MR31 had a higher hydrolytic activity as compared to the ΔamyA mutant. This result indirectly suggests that AmyA has the capacity to hydrolyze glycogen.

The role of AmyA was further examined by treatment of glycogen with MR31 and ΔamyA mutant. Branch chain-length distribution of glycogen after treatment of total cell-free extract of MR31 strain and ΔamyA mutant was detected by MALDI-TOF. As shown in Supplementary Figure 5, chain-length distribution of glycogen was changed after treatment of both extracts. When glycogen was treated with cell-free extract of MR31, the percentage of maltopentaose (DP5) was increased compared to the untreated glycogen and the distribution of branches with DP over 10 was decreased. The branches which comprise more than DP16 were disappeared. However, when the glycogen was treated with the cell-free extract of ΔamyA mutant, chain distribution with DP over 10 was complemented and branches were detected until DP18, suggesting that AmyA possesses the hydrolytic activity toward over DP10. The role of AmyA in the glycogen metabolism will be discussed below.

The gaa, glgX, and glgP genes were inactivated by insertional mutation to identify the functions in glycogen metabolism. The gene deletion strains were grown in the YT medium supplemented with 0.2% glucose and the glycogen contents in each growth phases were measured. The pattern of glycogen content in the glgP deletion mutant was similar to the glycogen content of MR31 strain (Figures 6A,B), hence indicating that the deletion of GlgP, predicted as a glycogen phosphorylase, results in a modest change in the glycogen content of S. acidocaldarius.

As compared to the MR31 strain, the amount of glycogen in glgX and gaa deletion mutants was significantly higher in all phases. The total content of soluble and membrane-bound glycogen in the glgX deletion mutant was approximately 0.7 mg glycogen per 1 mg protein in the early exponential phase and further increased during the cell growth. This strain finally accumulated approximately 1.2 mg glycogen per 1 mg protein in stationary phase (Figure 6C). Similar to the glgX deletion mutant, glycogen gradually accumulated in the gaa deletion mutant and reached approximately 1.0 mg glycogen per 1 mg protein in the stationary phase (Figure 6D). In both glgX and gaa deletion mutants, the intracellular glycogen was much higher than that of the MR31 strain, especially in the stationary phase. In detail, glgX and gaa deletion mutant accumulated 2.6- and 2.9-times higher glycogen content during the exponential growth phase and 38.5- and 21.2-times higher glycogen content in the stationary phase, respectively. This result suggests that GlgX and Gaa, which are predicted as glycogen-debranching enzyme and glucoamylase, respectively, play a major role in the degradation of glycogen in S. acidocaldarius.

The glycogen content of S. acidocaldarius was increased during the early exponential phase and the accumulated glycogen is decreased gradually during the late exponential and stationary growth phases. The transcriptional analysis was conducted to investigate the changes of glycogen content during the growth. Three putative promoters located in the upstream regions of amyA–glgA, glgC–gaa–glgX operon, and glgP gene were fused with the β-galactosidase (LacS) gene amplified from the chromosomal DNA of S. solfataricus P2. All S. acidocaldarius strains that contains the lacS gene, fused with a putative promoter region, showed a higher β-galactosidase activity as compared with MR31 strain (Figure 7). The S. acidocaldarius strain, which harbors the lacS gene fused with the putative promoter of glgP or amyA, showed a constant β-galactosidase activity regardless of the growth phase. In contrast, the enzyme activity of the strain harboring the lacS gene fused with the putative promoter of glgC gradually increased during cell growth, thereby suggesting that the expressions of amyA and glgA are constant in the overall cell growth but the expressions of glgC, gaa, and glgX increase along the cell growth. The result of promoter assay can be account for the gradual decrease of glycogen content during the cell growth. The gradual increase of GlgX and Gaa, which play an important role in the glycogen degradation, is supposed to affect the glycogen content, further resulting in a gradual decrease.

To investigate the relationship between the intracellular glycogen and cell properties, the death rate of wild type and glycogen deficient mutants were compared in the Brock’s basal medium. The growth of S. acidocaldarius Δglg mutant, which cannot accumulate the intracellular glycogen, was slower than the growth of wild type MR31 and glgX mutant strains in the early exponential growth phase (Figure 8A). The doubling times of MR31, Δglg, and ΔglgX in an early log phase were determined as 3.72, 5.82, and 3.87 h, respectively. The mutant and wild type cells were harvested when the OD surpassed 0.5, where the glycogen content of both fractions was maintained, and each strain was inoculated into 100 ml of Brock’s basal media after washing with fresh Brock’s basal media at the initial OD of 0.5. The OD of each cell was detected every 6- or 12-h interval. The OD of ΔGlg mutant was decreased more rapidly than that of the MR31 strain. The death rate of each strain was calculated by using simple regression in the section, which showed a linear decrease (Figure 8B). The result indicates that the absence of glycogen affects the growth rate of S. acidocaldarius at an early exponential phase and the death rate when the cells were placed in nutrient deficient conditions.

To count viable cells, plating and MTT assay were conducted. CFU of MR31, Δglg, and ΔglgX right after the inoculation into Brock’s basal media was an average of 11.5 × 10⁸, 7.0 × 10⁸, and 8.0 × 10⁸ per 1 ml of culture, respectively. After 24 h of the incubation in Brock’s basal media, CFU of MR31, Δglg, and ΔglgX decreased to 3.6 × 10⁸, 1.0 × 10⁸, and 2.3 × 10⁸, respectively (Figure 8C). All the strain showed a reduced CFU, especially Δglg showed the greatest reduction. To quantify the viable cells indirectly, MTT activity of 3 ml cultures were compared (Figure 8D). Reductase activity of each cell after inoculation into Brock’s basal media was compared to the activity before inoculation. MR31 and Δglg showed similar relative activity at 24 h after the inoculation, but reductase activity of Δglg decreased faster. Cell morphology before and after the inoculation into Brock’s basal media was visualized by SEM (see Supplementary Figure 6A). While MR31 and ΔglgX showed lysed morphology at 48 h after the inoculation, Δglg already showed lysed morphology at 24 h after the inoculation, suggesting that Δglg is more sensitive to nutrient starvation. In addition, Δglg seemed to be more sensitive to osmotic stress as well (see Supplementary Figure 6B). From this result, we can speculate that glycogen, which is absent in the Δglg mutant but present in MR31 and ΔglgX mutant, is involved in the maintenance of cell growth when starvation condition or stress is provided.