The E. coli strain BW28357 [rrnB3 ΔlacZ4787 hsdR514 ΔaraBA-D567 ΔrhaBAD568 rph⁺] was used to create the isogenic Δpgi strain [rrnB3 ΔlacZ4787 hsdR514 ΔaraBA-D567 ΔrhaBAD568 rph⁺ Δpgi kan^r]. Bacteria were cultured in either rich Luria-Bertani (LB) or M63 minimal medium supplemented with glucose:gluconate at a mass ratio 9:1 (20% total). For conciseness this medium is designated as M63GG throughout the text below.

Chromosomal DNA of the E. coli BW28357 and the isogenic Δpgi strains was isolated as described below (2.5.), except that the sonication step at the end was omitted. The polymerase chain reaction (PCR) was carried out in a final volume of 25 ul containing 50 ng template DNA, 2 mM MgCl₂, 0.2 mM dNTPs, 20 pmol of each forward (5’CTGTGACTGGCGCTACAATC3’) and reverse (5’CCGACATAACGTCGGCATTG3’) primers and 1 unit Taq DNA polymerase dissolved in nuclease-free water. DNA was denatured for 5 min at 95˚C followed by bulk DNA synthesis over 30 cycles, each composed of denaturing for 2 min at 95˚C, annealing for 2 min at 55˚C and elongation for 2 min at 72˚C. One additional cycle of 5 min at 72˚C was included at the end of the reaction for completion of DNA synthesis. The PCR products were analyzed by 1.2% agarose gel electrophoresis.

E. coli BW28357 and the isogenic Δpgi strains were cultured in M63GG medium at 37˚C until cultures reached an optical density of 1 OD₅₉₀ (10 mm cuvette path). Cells from 4 ml cultures were collected by centrifugation for 5 min at 1000 g and suspended in 1 ml of 50 mM Tris-HCl pH 8.0, 1 mM MgCl₂. Then, cells were disrupted by sonication using ultrasonic homogenizer (Model CP50, Cole-Parmer Instrument) three times for 1 min (50% duty cycle at 80% power), and cellular debris were pelleted by centrifugation at 12 000 g for 15 min at 4˚C. The supernatant was supplemented with 142 mg bentonite and incubated for 2 hours at room temperature under shaking, followed by centrifugation at 12,000 g for 5 min at 4˚C. G6Ph in the resulting supernatant was measured by the reduction of NADP to NADPH in the presence of yeast G6Ph-dehydorogenase (Sigma G7877) as described elsewhere [10] .

Single colonies of E. coli grown on LB agar were inoculated into 5 ml either LB or M63GG medium and incubated overnight at 37˚C on a New Brunswick rotary water bath shaker. Fresh 50 ml LB or M63GG medium in 500-ml Erlenmeyer flasks were inoculated with the overnight cultures at an initial density of 0.1 OD₅₉₀ and incubation was carried out at 37˚C on the rotary shaker at 250 rpm. At 30 min (LB) or 60 min (M63GG) intervals aliquots were taken from the cultures and turbidity was measured at 590 nm on a Beckman Coulter DUÒ 650 spectrophotometer.

E. coli cells corresponding to 125 OD₅₉₀ grown in LB or M63GG medium were collected by centrifugation and suspended in 8 ml 25 mM Tris-HCl pH 8, 10 mM EDTA, 0.14 M NaCl and 50 mM glucose. Sodium dodecyl sulfate (SDS) and Proteinase K were added to final concentrations of 1% and 250 ug/ml respectively, the volume was brought up to 10 ml with distilled water, and the suspension was incubated for 40 min at 37˚C. After two phenol extractions DNA was ethanol precipitated and dissolved in 1 ml 10 mM Tris-HCl pH 8, 1 mM EDTA (TE 10-1) containing 100 ug RNase A and 1 unit RNase T1. After 30 min incubation at 37˚C Proteinase K was added at a final concentration 100 ug/ml and the incubation was continued for additional 30 min. The samples were phenol-extracted, ethanol-precipitated and DNA was dissolved in TE 10-1 at a final concentration of 5 mg/ml. DNA was sonicated to an average size below 1300 bp and its quality was estimated by the A₂₆₀/A₂₈₀ spectral ratio (2 ± 0.1). Amadori products (APs), fluorescent AGEs and carboxymethyl (CM) residues in DNA were quantitated as already described [7] .

Five ml LB medium were inoculated with a single E. coli colony and incubated overnight at 37˚C in a water bath shaker. On the next day, cells were diluted 1:1000 with LB medium and each of 10 tubes containing 5 ml M63GG medium was inoculated with 10 ug of the resulting dilution. Cultures were grown at 37˚C until reaching 1 OD₅₉₀. Serial dilutions of these cultures were made and plated onto LB agar to count the number of viable cells. In parallel, 2 ml of the cultures were collected by centrifugation at 1000 g for 5 min at 4˚C, suspended in 100 ug M63GG medium and spread onto a LB agar supplemented with rifampicin at a final concentration of 100 ug. The number of exponential and stationary Rif^r mutants was recorded after incubation of the plates at 37˚C for 24 h or 72 h, respectively. The mutation frequencies were expressed as a number of Rif^r mutants per 10⁸ viable cells and presented as a median (n = 10). Data were processed by the Mann Whitney U-test [11] .

In this experiment we used complementary DNA and PNA 18-mer single stranded oligonucleotides with the following sequence: DNA 5’-CTA CTA ATC AGA CTA ATA-3’ and PNA 5’-TAT TAG TCT GAT TAG TAG-3’. The DNA oligonucleotide was incubated at a concentration of 100 uM with either 0 mM, 10 mM or 100 mM G6Ph in a volume of 100 ul for 48 h at 37˚C. Non-reacted G6Ph was removed by gel-filtration on CENTRI-SPIN 10 columns (Princeton). The glycated DNA (DNA-AP) was hybridized with PNA at 1:1 molar ratio in the presence of 0.1 M NaCl under the following conditions: incubation for 10 min at 85˚C, gradual cooling to 37˚C and further incubation at 37˚C for 24 h. DNA-AP and PNA:DNA-AP were purified from DNA fragments resulting from glycation and from the non-hybridized oligonucleotides via RP-UHPLC on Acquity UPLC BEH C18 column (2.1 mm × 50 mm, 1.7 um, 300 Å) heated to 35˚C by applying the following 30 min gradient of acetonitrile (ACN) in aqueous 10 mM triethylammonium acetate at a flow rate 0.2 ml/min: equilibration of the column at 7% ACN, rise in ACN concentration from 7% to 17% during the first 15 min followed by increase to 100% ACN over the next 5 min and isocratic flow at 100% ACN for the last 10 min. Under these conditions DNA-AP and PNA: DNA-AP eluted at RT 6.3 min and RT 10.9 min, respectively.

After removal of the unbound G6Ph, 2 ul of each of the incubation mixtures of the DNA oligonucleotide with G6Ph from the previous Section 2.7 were introduced into the electrospray ionization (ESI) source of the mass-spectrometer (Thermo Scientific™ Orbitrap Elite™ Hybrid Ion Trap-Orbitrap) via direct infusion at a speed of the mobile phase (50% methanol, 0.1% ammonium formate) of 5 ul/min. Mass spectra were collected in the FT mode in a normal mass range from m/z 900 to m/z 1400 at a resolution of 30,000 and 50 ms ion injection time. The ion source was operated in a negative ion mode at 2.5 kV ion spray voltage and a capillary temperature of 250˚C.

E. coli cells were cultured in LB medium at 37˚C until mid-exponential phase (0.5 OD₅₉₀), collected by centrifugation for 5 min at 1000 g and washed twice with phosphate buffered saline (pH 7.4). Then, the cells were suspended in deionized water, disrupted by sonication four times for 15 seconds (50% duty cycle at 80% power), and the cell debris were removed through centrifugation at 16,000 g for 30 min at 4˚C. The protein concentration in the resulting supernatants was determined using Sigma BCA assay kit (Prod. No. B9643). Protein extracts (25 ug per lane) were separated on 12% SDS-polyacrylamide gel under reducing conditions (0.7 M beta-mercaptoethanol in protein extracts), transferred onto PVDF-membranes and Western blotting was performed with a rabbit polyclonal anti-yeast PGI IgG (antibodies-online, Cat. No. ABIN568370). Blots were developed with an IRDye 800CW goat anti-rabbit IgG (Cat. No. 925-3221, LI-COR Biosciences GmbH) and visualized on an Odyssey LI-COR imaging system.

The assay was performed according to Bergmeyer [12] in a final volume of 0.5 ml containing 50 mM Tris-HCl pH 8.0, 0.67 mM NADP, 3.3 mM MgCl₂ and one of the following substrates: 3.3 mM fructose 6-phosphate, 10 uM DNA-AP or 10 uM PNA:DNA-AP. First, 1 unit of G6Ph-dehydrogenase was added and the incubation was carried out at ambient temperature until no increase in the absorbance at 340 nm was observed. This step was included in order to avoid interference of any trace amounts of G6Ph present in the Fr6Ph preparation or in the final DNA-AP and PNA: DNA-AP samples. Then, the isomerization reaction was initiated by the addition of 8 ug E. coli protein extract and the increase in the absorbance at 340 nm was followed over time.

The pgi gene of the E. coli strain BW28357 was deleted by one-step gene inactivation protocol with PCR products [13] . The PCR product used to replace the pgi gene is encoding kanamycin resistance (Kan^r) and is shorter than the displaced chromosomal DNA region. We selected Kan^r E. coli colonies and performed PCR to confirm the pgi gene deletion. With template DNA of the parental strain the chain reaction generated a DNA fragment of 1891 bp while DNA of the knock-out strain produced a fragment of 1568 bp (Figure 1(a)) thus confirming the successful pgi gene deletion. To get an independent verification of the pgi gene deletion we performed Western blotting. In this experiment we

used a polyclonal antibody raised against Saccharomyces cerevisiae PGI, which shares 59% homology with the E. coli enzyme. As shown in Figure 1(b), positive staining with this antibody was observed only with protein extract from the parental PGI proficient strain and with yeast PGI used as a positive control. The protein extract from the E. coli Δpgi strain did not react with the anti-PGI antibody.

PGI is converting G6Ph to fructose 6-phosphate (Fr6Ph) in the second glycolytic step. Therefore, it is expected that inactivation of the pgi gene would result in accumulation of over-physiological G6Ph concentrations in the E. coli Δpgi strain when grown in high glucose minimal medium. To test this option, we measured the G6Ph concentration in protein-deprived lysates of both the parental and Δpgi E. coli strains grown in M63GG medium. Under these conditions the G6Ph concentration in the PGI deficient cells proved to be more than ten times higher than that in the PGI proficient cells (Figure 2(a)), which confirmed previously reported data [14] [15] . On the other hand, this result provides an independent physiological evidence for the pgi gene deletion. In next experiments we evaluated the impact of the pgi gene deletion on the growth of the PGI deficient strain. We cultured both strains in rich LB medium and in minimal M63GG medium as well. We did not observe a significant difference in the growth of both strains in LB medium (Figure 2(b)). This result is not surprising as far as in rich media alternative catabolic pathways may be activated sustaining the growth of the PGI deficient strain [16] . In M63GG medium both strains exhibited delayed growth with Δpgi strain growing one-third as fast as the parent (Figure 2(c)). Similar inhibition of the growth of an E. coli Δpgi strain on glucose supplemented M63 medium has been previously reported [17] .

In a previous study, it has been demonstrated that plasmid transformation of an

E. coli Δpgi strain results in increased mutation rate of a plasmid borne marker gene [15] . In this study, we tested the impact of the pgi gene inactivation on the mutation rate of a chromosomal E. coli gene―the rpoB gene encoding the beta-subunit of the RNA polymerase. Mutations in this gene result in resistance to the antibiotic rifampicin (Rif^r). We recorded two types of mutations―exponential and stationary phase (adaptive) mutations. Exponential mutations arise during growth of bacteria in liquid medium in the absence of the antibiotic and are scored 24 hours after plating the cells on Rif-plates. Stationary mutations appear during growth of bacteria on the Rif-plates and are counted 72 hours after plating. We measured exponential mutation rate to Rif^r of the parental E. coli strain 19.6 mutants per 10⁸ viable cells versus 26.2 mutants per 10⁸ viable cells of the Δpgi strain (p < 0.025, n = 10) (Figure 3(a)). For the adaptive mutations the calculated values were 9.7 mutants per 10⁸ viable cells for the parental strain versus 19.3 mutants per 10⁸ viable cells for the Δpgi strain (p < 0.005, n = 10) (Figure 3(b)). These results clearly demonstrate that inactivation of the pgi gene results in increased spontaneous mutation rate to Rif^r as far as both exponential and adaptive mutations appeared more frequently in the E. coli PGI deficient strain.

As demonstrated above, inactivation of the pgi gene resulted in accumulation of G6Ph in E. coli cells and in increased spontaneous mutation rate to Rif^r when cells were grown in high glucose M63 minimal medium. We suggested that excess G6Ph may non-enzymatically react with DNA via the Maillard reaction (glycation) thus contributing to the increased spontaneous mutation rate to Rif^r of the E. coli Δpgi strain. To explore this suggestion, we isolated chromosomal DNA of the parental and Δpgi E. strains and measured the amount of early glycation products (APs) (Figure 4(a)) and two types of AGEs―fluorescent AGEs

(Figure 4(b)) and carboxymethyl (CM) adducts (Figure 4(c)). The AGEs-fluorophores were detected through fluorescence spectroscopy at 440 nm emission wavelength upon excitation at 360 nm. For measurement of CM-adducts we used an antibody raised against CM-lysine, which was shown in a previous study to cross-react with glycated DNA [7] . Contrary to our expectations, we measured lower amounts of all three types of glycation adducts in the chromosomal DNA of the E. coli Δpgi than in DNA of the parental strain (Figures 4(a)-(c)).

Amadori products (APs) are unstable intermediates of the Maillard reaction that may involve DNA in dramatic chemical rearrangements ceasing with intra- and/or intermolecular crosslinking of the DNA strands and strand breaks as well [18] [19] . Given this high instability of APs-modified DNA we suggested that there should be mechanisms for repair of APs-lesions in DNA in order to prevent propagation of the DNA damage. G6Ph-derived APs in DNA represent fructose 6-phosphates, which are covalently linked to the exocyclic NH₂-groups of the DNA bases G [20] , A and C. Fructose 6-phosphate is a PGI substrate and the isomerization G6Ph « Fr6Ph is known to be easily reversible [21] . We therefore suggested that PGI might be involved in reversal of the early step of the Maillard reaction; i.e. that PGI might catalyze the isomerization of APs (DNA-fructosamine-6-phosphates) to Schiff bases (DNA-glucosimine-6-phosphates) followed by spontaneous hydrolysis of the Schiff bases to G6Ph and free DNA. To test this hypothesis, we prepared a hybrid PNA: DNA 18-mer duplex modified with G6Ph. The DNA oligonucleotide was designed so as to contain a single guanine (see the Methods section), which is the DNA base most susceptible to glycation [19] [20] [22] .

The most intense ion of the non-modified DNA oligonucleotide as analyzed by ESI-mass spectrometry (ESI-MS) was with m/z 1088.94, which corresponds to an ion species with 5 negative charges (1088.98 calculated m/z) (Figure 5(a)).

We first went on modifying that DNA oligonucleotide with G6Ph, whereby we observed a concentration dependent decrease in the intensity of the ion with m/z 1088.94 with increasing G6Ph concentrations (Figure 5(c)). In parallel, we monitored also the appearance of the ion with 5 negative charges and m/z 1137.98 of the DNA oligonucleotide modified by a single G6Ph-derived AP (1137.54 calculated m/z) (Figure 5(b)). Expectedly, the intensity of this ion increased with increasing G6Ph concentrations (Figure 5(d)). The many ion species in the m/z spectrum of DNA-AP (Figure 5(b)) are indicative of intense DNA degradation, which clearly demonstrates the fragility of glycated DNA. In order to stabilize the DNA-AP oligonucleotide we hybridized it with a complementary PNA oligonucleotide. DNA-AP and PNA: DNA-AP were then used as substrates in a PGI activity assay coupled to the reduction of NADP to NADPH in the presence of G6Ph-dehydrogenase. We first performed a control PGI assay with Fr6Ph as a substrate. As expected, PGI activity with Fr6Ph was observed only with protein extract from the parental but not from the Δpgi E. coli strain (Figure 6(a)). Then, the assay was conducted with DNA-AP or PNA: DNA-AP as substrates and with protein extracts from the parental or Δpgi strains (Figure 6(b)). In favor of our suggestion we observed a release of G6Ph from both G6Ph-modified substrates (DNA-AP and PNA: DNA-AP) only when the assay was carried out with a protein extract from the PGI proficient but not from the PGI deficient strain.