Superoxide Dependence of the Toxicity of Short Chain Sugars*

Erythrose inhibited the growth of a sodA sodB strain of Escherichia coli under aerobiosis; but did not inhibit anaerobic growth of the sodA sodB strain, or the aerobic growth of the superoxide dismutase (SOD)-competent parental strain. A SOD mimic protected the sodA sodB strain against the toxicity of erythrose as did the carbonyl-blocking reagents hydrazine and aminoguanidine. Three carbon sugars, such as glyceraldehyde and dihydroxy acetone, and the two carbon sugar glycolaldehyde, were similarly toxic in an O⨪2-dependent manner. An unidentified dialyzable component in E. coli extract augmented the oxidation of short chain sugars, and this was partially inhibitable by SOD. The toxicity of the short chain sugars appears to be because of an O⨪2-dependent oxidation to α,β-dicarbonyl compounds. In keeping with this view was the O⨪2-independent toxicity of methylglyoxal.

During an attempt to use erythrose as a carbon source for the growth of a sodA sodB strain of Escherichia coli, we noted an oxygen-dependent toxicity and set out to explore its mechanism. The data reported herein indicate a role for O 2 . in the toxicities of erythrose and shorter chain sugars. These results are relevant to the much slower process of nonenzymic glycation seen with long chain sugars, such as glucose, which exist primarily as internal hemiacetals and are therefore less reactive (11)(12)(13)(14)(15).
MATERIALS AND METHODS D(Ϫ)-Erythrose was from Fluka, whereas aminoguanidine and methylglyoxal were obtained from Aldrich. DL-Glyceraldehyde, glycolalde-hyde, dihydroxy acetone, and the phosphate esters of glyceraldehyde, erythrose, and dihydroxy acetone, were from Sigma. The strains of E. coli used in these studies were AB1157, which was the parental strain for JI132 that was sodA sodB (16). Starter cultures were usually grown overnight in aerobic LB medium at 37°C and were then diluted 200-fold into M9CA medium. LB and M9CA were as described (17). Anaerobiosis, when desired, was achieved in Gas Pack jars. These jars were opened at intervals to allow turbidimetry, following which the jars were evacuated and refilled with N 2 . Culture growth was monitored turbidimetrically at 600 nm; whereas viability was measured by enumeration of colonies after dilution and plating on LB agar. When needed, extracts were prepared from cultures at A 600 nm ϭ 0.5-0.8 by centrifugation, washing the cells two times in 50 mM potassium phosphate, pH 7.8 and then disrupting the cells, which had been resuspended in this buffer, with a French Press. The lysate was clarified by centrifugation and protein content was assayed colorimetrically (18).

Erythrose Toxicity Is Dependent on O 2
. -Erythrose dose-dependently inhibited the growth of a sodA sodB strain of E. coli, as shown in Fig. 1 by lines 3, 2, and 1. The SOD 1 -competent parental strain was much less affected (lines 6, 5, and 4). This suggested that an oxidation of erythrose initiated by, and/or producing, O 2 . was a factor in this toxicity. In that case the toxicity of erythrose should be markedly diminished under anaerobic conditions. The results in Fig. 2 demonstrate that this was the case. The slight growth inhibition by erythrose shown in Fig. 2 was certainly because of residual oxygen, because the cultures were exposed to air periodically when the Gas Pack jars were opened and A 600 nm was measured. As a final demonstration of the importance of O 2 . in the toxicity of erythrose, the effect of a cell permeable mimic of SOD activity (MnTM-2-PyP) (19) was examined. The results in Fig. 3 show that this compound at 25 M completely eliminated the growth inhibitory effect of 20 mM erythrose. Three-and Two-carbon Sugars-Erythrose can ring close to a furanose form, but sugars shorter than four carbons cannot. Such sugars have previously been seen to be prone to enolization and autoxidation (1)(2)(3)(4)(5). One would expect, therefore, that sugars containing two-or three-carbon atoms should equal or exceed the toxicity of erythrose and that their toxicities should be O 2 and O 2 . augmented. Glyceraldehyde, dihydroxy acetone, and glycolaldehyde were all examined and all were more toxic to the sodA sodB than to the parental strain under aerobic conditions. Further, the toxicity of these compounds to the sodA sodB strain was O 2 -dependent and was eliminated by the SOD mimic MnTM-2-PyP (19). Thus Fig. 4 shows the growth inhibitory effect of 2.0 mM glycolaldehyde and the elimination of that growth inhibition by 25 M MnTM-2-PyP. A control compound, ZnTM-2-PyP, which does not catalyze the dismutation of O 2 . , was without effect (not shown). It must be noted that ZnTM-2-PyP was tested in the dark. This was necessary because the zinc compound, unlike the manganese compound, exerted a photodynamic effect, which caused lethality in the light. (1). In that case, blocking the carbonyl group, through formation of hydrazides, should prevent enolization and consequently oxidation. We examined the effect of hydrazine and of aminoguanidine on the oxidation of erythrose, monitored in terms of the CN Ϫ catalyzed reduction of cytochrome c (which was inhibitable by SOD). Fig. 5 illustrates the inhibitory effect of hydrazine (line 2) and of aminoguanidine (line 3). In panel A the hydrazines were added before the cyanide and thus had time to convert much of the open chain form of erythrose to its hydrazide. When the aminoguanidine was added after the cyanide, its inhibitory effect was much diminished (panel B) because of the conversion of the carbonyl form of the sugar to the cyanohydrin, before the addition of the aminoguanidine.
The interactions of aminoguanidine with erythrose and with methylglyoxal were followed in terms of increased absorbancies at 230 and 318 nm, respectively, at pH 7.8 in 50 mM potassium phosphate at 23°C. Both reactions could be followed on a time scale of minutes, but the reaction with methylglyoxal was more rapid than that with erythrose probably because of the exist-ence of the bulk of the sugar in the furanose form. Somewhat surprisingly the reaction of methylglyoxal with aminoguanidine exhibited a fast initial rate followed by a somewhat slower

-Dependent Sugar Toxicity
and linear rate. Because the shape of the absorption spectrum did not change with time, this probably reflects the partition of methylglyoxal between anhydrous (carbonyl) and hydrated (gem diol) status with the former reacting rapidly and then being replenished by a rate-limiting dehydration of the gem diol. If a mechanism of enolization and oxidation was involved in the toxicity of erythrose, then hydrazines might protect. Aminoguanidine was selected for this test and the results in Fig. 6 show that it protected the sodA sodB strain against the growth inhibitory effect of erythrose.
Nucleophile in E. coli-It has previously been seen that strong nucleophiles, such as CN, catalyze the oxidation of short chain sugars (1)(2)(3). The possibility that nucleophiles within E. coli might similarly augment the oxidation of short chain sugars was examined. An extract of the sodA sodB strain stimulated the reduction of cytochrome c by erythrose, and this was partially inhibited by SOD. An extract of the parental strain was also effective, but added SOD did not then inhibit; presumably because the reaction was already maximally inhibited by the endogenous SOD. The active component in the E. coli extracts was seen to be dialyzable but was not further characterized.
Growth Inhibition by ␣,␤-Dicarbonyls-Among the products of oxidation of ␣-hydroxycarbonyl compounds are ␣,␤-dicarbonyl compounds. It appeared possible that some of the toxicity of the short chain sugars might have been because of such dicarbonyl oxidation products. Lines 5 and 6 in Fig. 7 show that 5 or 10 mM methylglyoxal completely inhibited the growth of the sodA sodB strain, whereas lines 2 and 3 demonstrate that 20 mM aminoguanidine completely eliminated that toxicity. Lines 1 and 4 are controls with aminoguanidine alone, or no additions, respectively. The growth inhibiting effect of methylglyoxal shown in Fig. 7 was seen even under anaerobic conditions and with the SOD-competent, as well as with the SODnull strain, and was thus not dependent on O 2 . or on further oxidation (data not shown).

DISCUSSION
Short chain sugars and their phosphate esters, in which the carbonyl group is not largely blocked by cyclization to the furanose or to more stable pyranose rings, can tautomerize to enediols, which are prone to autoxidation. O 2 . has been shown to serve as both an initiator and a propagator of these autoxidations (4). Short chain sugars are thus potentially capable of synergizing with O 2 . in causing toxicity.
The results reported herein demonstrate that this does oc-cur. Thus short chain sugars can cause a growth inhibition and a lethality to E. coli, which is O 2 -dependent and blocked by SOD or by an exogenous SOD mimic. CN Ϫ augments the autoxidation of short chain sugars and their phosphates, and dialyzable unidentified nucleophiles, present in extracts of E. coli, exerted a similar effect. Pentoses and hexoses are much less reactive in this regard than are the tetroses and trioses because of the stabilizing effect of ring closure to pyranose forms by hemiacetal formation. However, some carbonyl reactivity remains, even with glucose (14) and it leads to nonenzymic glycation, which generates ⑀-amino fructosyl lysine derivatives of proteins. These derivatives can then autoxidize leading to advanced glycation products such as N ⑀ -carboxymethyl lysine derivatives (12,20). O 2 . is involved in this glycoxidation process, as shown by the inhibitory effect of SOD (21). Aminoguanidine, shown here to protect E. coli against the toxicity of short chain sugars, has been shown to ameliorate the cardiovascular and renal pathologies seen in aging rats (11). The protection against erythrose toxicity by aminoguanidine shown in Fig. 6 could have been because of prevention of the enolization and subsequent oxidation of the erythrose, because of formation of the hydrazide derivative (22,23), or to conversion of the ␣,␤-dicarbonyl oxidation product of erythrose to the asymmetrical triazine (24 -26) or to both actions. The profound toxicity of methylglyoxal and the complete protection against that toxicity provided by aminoguanidine indicate that much of the effect of the short chain sugars was because of ␣,␤-dicarbonyl oxidation products and that much of the protective effect of aminoguanidine was because of the trapping of these dicarbonyls, probably by conversion to triazines. Hirsch et al. (24) have reported that the reaction of aminoguanidine with ␣,␤-dicarbonyl oxidation products of D-glucose to yield triazines is rapid at neutral pH, being complete within 5 min at 37°C. The protective effect of aminoguanidine may thus be because of its dual action in preventing the O 2 .  3-phosphoglyceraldehyde be kept low. The toxicities of short chain sugars to E. coli can be exploited to shed light on the deleterious consequences of the much slower process of nonenzymic glycation followed by oxidation, by long chain sugars, which also seem to involve oxygen-derived free radicals. Thus the teratogenic effect of high glucose was diminished by superoxide dismutase, catalase, and glutathione peroxidase added to the culture medium (27). This was demonstrated more conclusively with transgenic mice overexpressing Cu,Zn-superoxide dismutase, which exhibited much less embryopathy when diabetic than did nontransgenic controls (28). That high glucose imposes on oxidative stress was also indicated by the observation that it induced Cu,Zn-superoxide dismutase in cultured endothelial cells (29).