Role of 2,6-Dideoxy-2,6-diaminoglucose in Activation of a Eukaryotic Phospholipase C by Aminoglycoside Antibiotics*

Recent emergence of microbial resistance to aminoglycoside antibiotics, and the documented cytotoxicity associated with their use, calls for sustained efforts at understanding the effects of the compounds on eukaryotic cells. Using a glycosyl phosphatidylinositol (GPI)-phospholipase C (GPI-PLC) from the protozoan parasite Trypanosoma brucei, we demonstrate that a eukaryotic PLC can be activated 6-fold by aminoglycosides. Neomycin B protected GPI-PLC from a reduction in activity at pH 6.5, and increased the turnover number (k cat) of the enzyme. In structure-activity studies with the neomycin group, 2-deoxy-streptamine was mildly stimulatory; the concentration required to activate GPI-PLC 2-fold (SC 200) was 310 μm. Neamine was 150-fold more active (SC 200 = 2 μm) than 2-deoxy-streptamine, indicating that a 2,6-dideoxy-2,6-diaminoglucose substituent at the 4-position of 2-deoxystreptamine plays an important role in activation of GPI-PLC. Ribostamycin and neomycin B also had SC 200′s of 2 μm, implying that the ribose group in ribostamycin is not involved in activation of GPI-PLC. These conclusions were affirmed in studies with Bacillus thuringiensisphosphatidylinositol-specific phospholipase C. A 2,6-dideoxy-2,6-diaminoglucose substitution at the 4-OH of 2-deoxystreptamine activates the enzyme 17-fold, while a second 2,6-dideoxy-2,6-diaminoglucose moiety on the ribose ring of ribostamycin provides an additional 3.5-fold stimulation. Possible implications of these observations for the effects of aminoglycosides on eukaryote cells are discussed.

Aminoglycosides have gained widespread use over the last fifty years as antibiotics against Gram-negative bacteria and some protozoa. In prokaryotes, aminoglycosides reduce the fidelity of protein synthesis on ribosomes and block polypeptidyl-tRNA translocation, through interactions with A-site 16 S rRNA (1,2) and elongation factor Tu (3). Recent emergence of drug resistance against aminoglycoides (4 -6) suggests that efforts to understand all possible mechanisms of action of these compounds must be re-doubled with the long term aim of designing new and possibly more effective drugs.
Toxicity against mammalian cells (especially in the kidney) is one undesirable side effect of aminoglycoside antibiotics (reviewed (7)). Removal of this side effect from the next generation of aminoglycoside antibiotics will require information on how these compounds affect mammalian cells.
Molecular geneticists employ aminoglycosides to eliminate eukaryote cells that fail take up and/or express selectable markers (e.g. neomycin phosphotranferase) after DNA-mediated transformation. Although the exact mode of cell killing has not been worked out, the antibiotics interfere with RNA splicing (8) and interact with negatively charged phospholipids (7). The toxic effects on eukaryote cells is thought to be due in part, at least, to alterations of the properties of the plasma and other cellular membranes (7).
Mammalian phosphatidylinositol-specific phospholipases C (PI-PLCs) 1 in crude cell extracts are stimulated by aminoglycosides (9, 10) although it is not clear from those studies whether the aminoglycosides affected the PI-PLCs directly. Recently, a secreted PI-PLC from the prokaryote Bacillus cereus was shown to be activated by G418 (11). From the last observation, it was proposed that the antibiotics might exert their cytotoxic effects, in part, by activation of cellular enzymes.
Herein, we report stimulation of a purified eukaryotic enzyme, glycosyl phosphatidylinositol phospholipase C (GPI-PLC) from the protozoan parasite Trypanosoma brucei, itself susceptible to the antibiotics, by members of the neomycin group. Neomycin B protected the enzyme from a reduction in reaction rate observed at slightly acidic pH and increased the turnover number (k cat ) of GPI-PLC. Evidence is presented that 2,6-dideoxy-2,6-diaminoglucose in the neomycin group of aminoglycosides is very important for activation of the PLCs from T. brucei and Bacillus thuringiensis.
Phospholipase C Assays-Reaction mixtures were assembled on ice in 1.5-ml microcentrifuge tubes. The quantity of enzyme used was determined empirically; the amount of enzyme that cleaved ϳ50% of [ 3 H]mfVSG was used in the kinetic analyses, to stay within the linear * This work was supported by National Institutes of Health Grant AI33383 and by a Burroughs Wellcome Fund New Investigator Award in Molecular Parasitology (to K. M-W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. range of the assay (13). GPI-PLC (9.5 units, 0.09 ng) was incubated with aminoglycoside antibiotics or control compounds in 20 l of a mixed buffer (MES/Tris/CAPS, 50 mM final concentration) with 1% Nonidet P-40 for 10 min at 37°C. Two g of [ 3 H]mfVSG in 10 l of 1% Nonidet P-40 was added, followed by further incubation for 15 min at 37°C. The reaction was terminated by chilling the mixture on ice and vortex mixing with 500 l of water-saturated 1-butanol (at room temperature) to extract the cleaved [ 3 H]dimyristoylglycerol. Phases were separated by centrifugation (12,000 ϫ g, 1 min, 25°C), and enzyme activity was quantified by measuring the amount of [ 3 H]dimyristoylglycerol released into the upper butanol phase using a Beckman LS 6000TA scintillation counter. Activity of GPI-PLC obtained in the absence of test compounds was determined and assigned a value of 100%. Radioactivity from a mock digest of [ 3 H]mfVSG (no enzyme added) was subtracted (as background) from all counts obtained.
[ 3 H]mfVSG (in 0.1% sodium deoxycholate) was added, and product analysis was performed as described above for T. brucei GPI-PLC.

Effect of pH on Aminoglycoside Stimulation of GPI-PLC-
The pH of the GPI-PLC assay (mixed) buffer was varied from pH 5.5 to pH 11.0. Buffer was added from a 500 mM stock (to 50 mM final concentration) with 10% (w/v) Nonidet P-40 (final concentration 1.0%) to 9.5 units of GPI-PLC. The mixture was incubated for 10 min at 37°C with or without 0.3 mM neomycin B (from a 5 mM stock in H 2 O), followed by substrate addition, reaction initiation and termination as described above. (Purified substrate was in 1.0% Nonidet P-40 in H 2 O to avoid perturbing the pH of the reaction mixture.) Determination of SC 200 -Varying amounts of aminoglycoside antibiotics were incubated at pH 6.5 for 10 min at 37°C with GPI-PLC (9.5 units) or at pH 6 for PI-PLC (3.6 ϫ 10 Ϫ5 units), followed by addition of substrate and reaction completion as described above. The concentration of aminoglycoside antibiotics required to stimulate the phospholipases 2-fold (SC 200 ) was obtained graphically (See Fig. 2).

RESULTS
Aminoglycosides Maintain Phospholipases C Activity at Suboptimal pH-GPI-PLC is optimally active at about pH 9 (Fig.  1A). Enzyme activity is reduced 50% at pH 7.5 and 10, and nearly abolished at pH 6.5 or 11. In the presence of 0.3 mM neomycin B, the pH-activity profile of GPI-PLC is altered. The enzyme remains completely active at pH 7.5 and 60% active at pH 6.5, and neomycin B activated T. brucei GPI-PLC 6-fold as compared with a reaction in the absence of the aminoglycoside (Fig. 1). The pH optimum for the reaction decreased from pH 9.0 to 8.0 (Fig. 1). The activity of GPI-PLC under basic conditions is not altered by neomycin B (Fig. 1A). PI-PLC from B. thuringiensis is optimally active against a GPI at pH 7.0. Loss of activity below pH 7.0 was forestalled by addition of neomycin B (Fig. 1B).
Structure-Activity Relationship of Aminoglycoside Activation of PLC-To dissect features of aminoglycosides that influence phospholipase C activation, structurally related aminoglycosides from the neomycin group were examined (Fig. 2). 2-Deoxystreptamine stimulated GPI-PLC with a SC 200 (the concentration of aminoglycoside required to activate the enzyme 2-fold) of 310 M (Fig. 2, inset, and Table I). Neamine (2-deoxystreptamine substituted at the 4 position with 2,6-dideoxy-2,6diaminoglucose) was 150-fold more potent, with a SC 200 of 2 M (Table I). Ribostamycin (neamine modified by the addition of a ribose) and neomycin B (ribostamycin with an additional 2,6dideoxy-2,6-diamino-glucose group) were no more effective than neamine (SC 200 of 2 M for both, Fig. 2 and Table I).
B. thuringiensis PI-PLC was also activated by 2-deoxy-streptamine (SC 200 of 490 M, Table I Table I).
Inhibition of GPI-PLC at pH 6.5 Is Prevented by Neomycin B-After a 10-min incubation at pH 6.5, GPI-PLC is inhibited 90% as compared with enzyme that was kept at pH 9.0 ( Figs. 1  and 3). In the presence of neomycin B (0.3 mM), enzyme activity is maintained even after a 15-min incubation in pH 6.5 buffer (Fig. 3). (The inhibition at the zero time point in the absence of neomycin B exists because the GPI cleavage assay is performed at pH 6.5 following the preincubation.) Neomycin B can rescue, at least partially, activity of GPI-PLC that has been preincubated at pH 6.5. Following a 5-15-min incubation at pH 6.5, neomycin B increased GPI-PLC activity 3-4-fold, when compared with an untreated control (Fig. 3). GPI-PLC activity is not altered by a 45-min incubation at 37°C at pH 8.0 (data not presented).
Acidic Phospholipids Influence Activation by Neomycin B -Acidic phospholipids inhibit GPI-PLC (13,14), and additionally can interact with aminoglycosides (7). Either (or both) of these properties could reduce the extent of GPI-PLC activation by aminoglycosides in vivo. To test this hypothesis, we examined the effect of aminoglycosides on GPI-PLC in the presence of acidic phospholipids. At pH 6.5, 5 mM PI had no effect on GPI-PLC, but PS (5 mM) strongly inhibited the enzyme (Fig. 4A). Neomycin B (0.3 mM) activated GPI-PLC 3-fold in the presence of PI (5 mM). As compared with the 6-fold stimulation obtained in the absence of PI, the neomycin B stimulation was reduced 50%. PS inhibition, however, was not altered by the presence of 0.3 mM neomycin B (Fig. 4A). Since the phospholipid concentration in these studies was in excess, we checked whether full stimulation by neomycin B would occur when the aminoglycoside was present at an equimolar concentration. Additional neomycin B (5 mM) stimulated GPI-PLC to extents similar to those of the control (i.e. neomycin B with enzyme alone) (Fig. 4B). Similar results were obtained when these experiments were performed at pH 8.0, with the exception that PI now inhibited GPI-PLC as  previously reported (13).
We conclude from these observations that (i) aminoglycoside activation of GPI-PLC can occur in the presence of acidic phospholipids, that (ii) PI inhibition of GPI-PLC is pH-sensitive, but that (iii) inhibition by PS in not diminished by a reduction in the pH of the reaction.
Neomycin B Alters the Kinetic Properties of PLCs-Eadie-Hofstee analysis indicated that neomycin B at pH 6.5 increased the apparent turnover number (k cat ) of GPI-PLC 2.6-fold (Fig.  5, Table II), from 40 min Ϫ1 to 102 min Ϫ1 . At pH 9.0, k cat of the enzyme increased to 144 min Ϫ1 , and neomycin B had no effect. Similarly, neomycin B did not change k cat at pH 10.5 (k cat ϭ 117 min Ϫ1 ).
A 5.9-fold increase in the K m of GPI-PLC occurred with rising pH, from 75 nM at pH 6.5 to 445 nM at pH 10.5 (Fig. 5, Table II). In general, a slight decrease in K m was observed in the presence of neomycin B but the magnitude of the change was not dramatic.
Neomycin B affected B. thuringiensis PI-PLC differently. At pH 5.1, the apparent K m and k cat were 449 nM, and 231 min Ϫ1 , respectively (Table III). Addition of neomycin B caused an 8.9-fold increase in the apparent K m to 3980 nM, while the k cat increased 10.2-fold to 2349 min Ϫ1 . At pH 8.0, neomycin B again increased both the K m and k cat of PI-PLC significantly (7.7 and 5.7-fold, respectively) (Table III). DISCUSSION Role of 2,6-Dideoxy-2,6-diamino-glucose in Activation of GPI-PLC by the Neomycin Group of Antibiotics-The three major families of aminoglycoside antibiotics (i.e. the gentamicin (e.g. G418), kanamycin (e.g. tobramycin), and neomycin (e.g. neomycin B) groups) all contain 2-deoxy-streptamine. Existence of 4 structurally related members of the neomycin family enabled us to explore possible roles of the components of the antibiotics in activation of GPI-PLC. In the neomycin group (Table I) substitution of 2-deoxy-streptamine with 2,6-dideoxy-2,6-diamino-glucose produces neamine, which can be derivatized with ribose to form ribostamycin. Substitution of the ribosyl group in ribostamycin with 2,6-dideoxy-2,6-diamino-glucose generates neomycin B. Two PLCs in hand, one from a eukaryote T. brucei and the other from a prokaryote B. thuringiensis, one could explore the relative contributions of the chemical modifications of 2-deoxy-streptamine on activation of the phospholipases.
With both T. brucei GPI-PLC and B. thuringiensis PI-PLC, 2-deoxy-streptamine was the least effective at activating the GPI cleavage reaction (Table I). For GPI-PLC, introduction of the 2,6-dideoxy-2,6-diamino-glucose group (i.e. in neamine) onto this parent compound raised activation potency 150-fold. In comparison, B. thuringiensis PI-PLC was 17-fold more ac- FIG. 3. Neomycin B protects GPI-PLC from inhibition at pH 6.5. GPI-PLC (9.5 units) was incubated for the indicated time in 50 mM MES, pH 6.5, 1.0% Nonidet P-40, 5 mM EDTA (Ⅺ). Neomycin B (0.3 mM) was added prior to (f) or following preincubation (q). After preincubation, substrate was added, and the enzyme was assayed as described in the legend to Fig. 1.  (Table I). Hence, with both enzymes 2,6-dideoxy-2,6-diamino-glucose has a tremendous activation effect when suitably placed. One can rule out the increasing positive charge of the aminglycosides as a major contributor for stimulation of the eukaryotic T. brucei GPI-PLC since (i) 1,4-diamino-butane has no effect, and (ii) neomycin B, which has 6 positive charges, is not better than neamine, which has 4 positive charges (Table I).
The more graded response to the four antibiotics displayed by B. thuringiensis PI-PLC but not by T. brucei GPI-PLC might be attributable to the fact that the proteins have very little sequence similarity at the polypeptide level although they are both highly specific for GPIs.
Persistent Activation of Cellular Phospholipases by Aminoglycosides May Contribute to Killing of Eukaryote Cells-Aminoglycoside antibiotics (e.g. G418, neomycin, and streptomycin) are used to treat infections caused by Gram-negative bacteria and by some protozoan parasites. In prokaryotes, misreading of the genetic code due to aminoglycoside occupancy of the A-site and interaction with 16 S rRNA (1,15) probably contributes to effectiveness of the antibiotics. In eukaryotes, inhibition of RNA splicing (8), binding to acidic phospholipids (16), and reduction in the fidelity of codon-anticodon interactions (17) may contribute to the cytostatic effects of aminoglycosides. Unfortunately, these antibiotics are nephrotoxic, and the mechanisms underlying this phenomenon are under investigation. Cellular toxicity of aminoglycosides is thought to be initiated by interaction of the compounds with negatively charged phospholipids in biological membranes (reviewed in Ref. 7).
Until now, no evidence existed for binding of aminoglycosides to cellular proteins of eukaryotic origin. When purified PI-PLC from B. cereus was stimulated by G418 (13), the possibility of direct binding became feasible. Strangely, the initial opportunity to demonstrate this phenomenon with a eukaryote GPI-PLC was missed (13) because the pH of the reaction (8.0) was higher by over a unit than the pH 6.5 required to document it (Fig. 1). Neomycin B can activate GPI-PLC from T. brucei by protecting the enzyme from mild acid-induced inhibition (Fig.  3). The small amount of neomycin B (2 M) ( Table I) needed to stimulate GPI-PLC 2-fold in a reaction that contains 50 mM of buffer at pH 6.5 suggests that one is not making the reaction more prolific by simply altering the pH of the buffer. In fact, no change in the pH of the buffer was measurable after addition of the aminoglycoside.
Neomycin B acts on T. brucei GPI-PLC by stimulating more rapid conversion of the enzyme-substrate complex to product, since k cat is increased without a significant change in K m (Table  II). This contrasts with the effect obtained with B. thuringiensis PI-PLC, in which case both k cat and K m are increased significantly (Table II) (11).
Aminoglycosides cause necrosis in renal proximal kidney tubule cells. The molecular basis of such nephrotoxicity is under intense investigation (reviewed in Ref. 7). Mammalian cells can be killed by some aminoglycosides, as happens when compounds like neomycin and hygromycin are used to select for transfected cells in molecular genetics experiments. Studies on possible downstream effects following the interaction of aminoglycosides with cellular phospholipids have not focused on membrane bound proteins. These antibiotics interact with eukaryote 18 S rRNA (17) and phospholipids (7). Direct binding to eukaryote proteins as a possible contributor to the development of cytotoxicity/nephrotoxicity during administration of these compounds has received little attention for lack of evidence. Because a eukaryotic GPI-PLC can be stimulated by several aminoglycosides (Table II), the possibility that aminoglycosides exert their effects in part by modulating the activity of cellular phospholipases merits serious consideration, chiefly because persistent activation of a cellular phospholipase C, especially under conditions when one had expected the enzyme to be inactive, could have negative consequences on cellular physiology. In cells that are known to accumulate the antibiotics to millimolar concentrations in acidic compartments (e.g. kidney cells (7) and fibroblasts (18)) the situation might be even more alarming. FIG. 5. Effect of neomycin B on the kinetic parameters of GPI-PLC. GPI-PLC (9.5 units) in the absence (q), or, presence of neomycin B (0.3 mM) (Ⅺ) was incubated at pH 6.5 for 10 min at 37°C, followed by addition of 500 to 3000 ng of substrate. GPI-PLC assays were then completed as described in Fig. 1. Eadie-Hofstee analysis of the data is presented.