|Home | About | Journals | Submit | Contact Us | Français|
The flavoprotein oxidase Fms1 from Saccharomyces cerevisiae catalyzes the oxidation of spermine and N1-acetylspermine to yield spermidine and 3-aminopropanal or N-acetyl-3-aminopropanal. The kinetic mechanism of the enzyme has been determined with both substrates. The initial velocity patterns are ping-pong, consistent with reduction being kinetically irreversible. Reduction of Fms1 by either substrate is biphasic. The rate constant for the rapid phase varies with the substrate concentration, with limiting rates for reduction of the enzyme of 126 and 1410 s−1 and apparent Kd values of 24.3 and 484 μM for spermine and N1-acetylspermine, respectively. The rapid phase is followed by a concentration-independent phase that is slower than turnover. The reaction of the reduced enzyme with oxygen is monophasic, with a rate constant of 402 mM−1 s−1 with spermine at 25 °C, and 204 mM−1 s−1 with N1-acetylspermine at 4 °C, pH 9.0. This step is followed by rate-limiting product dissociation. The kcat/Kamine-pH profiles are bell-shaped, with an average pKa value of 9.3 with spermine and pKa values of 8.3 and 9.6 with N1-acetylspermine. Both profiles are consistent with the active forms of substrates having two charged nitrogens. The pH profiles for the rate constant for flavin reduction show pKa values of 8.3 and 7.2 for spermine and N1-acetylspermine, respectively, for groups that must be unprotonated; these pKa values are assigned to the substrate N4. The kcat/KO2-pH profiles show pKa values of 7.5 for spermine and 6.8 for N1-acetylspermine. With both substrates, the kcat value decreases when a single residue is protonated.
Polyamines are essential for the growth and function of normal cells. The common polyamines putrescine, spermidine, and spermine are widely distributed in biological systems, not only as free bases but also as alkylated or acylated conjugates with sugars, steroids, phospholipids, fatty acids, and peptides. Because polyamines interact with multiple macromolecules, they affect a variety of cellular processes, such as cell growth, differentiation, and death (1-3). As a result, the biosynthetic pathway for formation of polyamines has been a target for the development of anti-cancer drugs (4). In comparison, the enzymes involved in polyamine catabolism have received much less study (5). In mammalian cells, catabolism of spermine can be initiated by the action of one of two enzymes. Acetylation of spermine by spermidine/spermine N1-acetyltransferase forms N1-acetylspermine (Scheme 2). This can be exported from the cell or oxidized by the peroxisomal flavoprotein polyamine oxidase (PAO) to spermidine and N-acetyl-3-aminopropanal (6). A further round of acetylation and oxidation by the same enzymes yields putrescine and 3-aminopropanal. Alternatively, the inducible flavoenzyme spermine oxidase (SMO) can oxidize spermine directly to spermidine and 3-aminopropanal (7-9). Saccharomyces cerevisiae also contains a flavoenzyme, Fms1, capable of oxidizing spermine and N1-acetylspermine (Scheme 1) (10, 11). The 3-aminopropanal produced by Fms1 is oxidized further to β-alanine, a precursor of pantothenic acid (10, 11). Spermidine is required for formation of the modified amino acid hypusine in yeast eIF5A (12) and helps protect the yeast cell from oxidative stress (13). Plant cells also contain a flavoenzyme capable of oxidizing polyamines (14). In contrast to the animal enzymes, plant polyamine oxidases oxidize the endo CN bond, forming 1,3-diaminopropane and N-(3-aminopropyl)-4-aminobutanal from spermine.
The mammalian enzymes PAO and SMO differ by 2-3 orders of magnitude in their relative preferences for N1-acetylspermine versus spermine (8, 15-17), in line with their roles in the cell. In addition, the pH optimum of PAO is more basic than that of SMO, consistent with the higher pH within peroxisomes than in the cytoplasm (18); the kcat/Km values for spermine and N1-acetylspermine as substrates for PAO have pH maxima of 10-10.5, while the kcat/Km value with spermine as substrate for SMO is highest at pH 8.3 (16, 17). This difference reflects the preferences of these two enzymes for different forms of the substrate. PAO is active on the singly-charged forms of polyamines, while SMO prefers polyamines with three positive charges. The structural basis for the differences is not known, since no structures are available of a mammalian PAO or SMO. Structures are available of Fms1 (19) and maize polyamine oxidase (20). These show that these two proteins have similar overall structures to the monoamine oxidase (MAO) family of flavoproteins; this family also includes MAO A and B, the lysine-specific demethylase LSD1, and L-amino acid oxidases. While the sequence identities between either mammalian enzyme and Fms1 or maize PAO are only about 20%, this is enough to assign both to the MAO structural family (21, 22).
Since Fms1 catalyzes the same reaction as PAO and SMO, its structure serves as the best current model for understanding the structural basis for the different specificities of PAO and SMO. However, to date there has been little mechanistic study of Fms1 to guide interpretation of the relationship of structure to specificity. We describe here mechanistic studies of Fms1 using both steady-state and rapid-reaction kinetic methods and analysis of the effects of pH on the kinetic parameters. The results provide insight into the mechanism of amine oxidation by Fms1 and the basis for the differences in substrate specificity among Fms1, PAO, and SMO.
Spermine and spermidine trihydrochloride were purchased from Acros Organics (Geel, Belgium). N1-Acetylspermine, 1,12-diaminododecane, 1,8-diaminooctane, o-dianisidine, horseradish peroxidase (72 purpurogalin units/mg), and glucose oxidase were from Sigma-Aldrich (Milwaukee, WI). N,N′-Dibenzyl-1,4-diaminobutane, N,N′-dibenzyl-1,3-diaminopropane, and N,N′-dibenzyl-1,2-diaminoethane were from Prime Organics (Woburn, MA). The pET28-based pJWL94 vector encoding his-tagged Fms1 (11) was kindly provided by Dr. Rolf Sternglanz of SUNY Stony Brook. E. coli BL21(DE3) Codon+ RIL competent cells were from Stratagene (Santa Clara, CA). The nickel-nitrilotriacetic acid (Ni-NTA) agarose resin was purchased from Invitrogen (Carlsbad, CA).
The expression and purification of Fms1 was based on the protocol described by Landry and Sternglanz (11). The pJWL94 vector encoding Fms1 was transformed into E. coli BL21(DE3) Codon+ RIL cells. For growth of Fms1, a single colony of E. coli (pJWL94) was used to inoculate a 60 ml culture of LB containing 25 μg/ml kanamycin and 20 μg/ml chloramphenicol. After overnight growth at 37 °C, four flasks containing 1 L of LB (25 μg/ml kanamycin and 20 μg/ml chloramphenicol) were inoculated with 10 ml each of the overnight culture. The cultures were incubated at 37 °C until the A600 reached a value of 0.6-0.8. At this point, isopropyl-β-d-thioglucanopyranoside (0.2 mM final concentration) was added to each flask and the temperature was decreased to 18 °C. After incubation for an additional 16 h, the cells were harvested by centrifugation at 5000g for 30 min at 4 °C. The cell paste was resuspended in 50 mM HEPES (pH 8.0), 10% glycerol, 2 μM pepstatin, 2 μM leupeptin, 0.1 mM NaCl, 100 μg/mL phenylmethanesulfonyl fluoride, and 100 μg/mL lysozyme, and lysed by sonication. After the lysate was centrifuged at 22400g, the supernatant was loaded onto a 20 mL Ni-NTA column equilibrated with the same buffer. The protein was eluted with a linear gradient from 100 to 200 mM imidazole in 30 column volumes of the same buffer. Fractions containing Fms1 were pooled and concentrated using 30 K Amicon Ultra Centrifugal filters. The purified protein was dialyzed against three changes of the buffer. The resulting protein sample was centrifuged at 22400g for 30 min at 4 °C to remove precipitated protein and stored with 10% glycerol at −80 °C.
Five volumes of 10 M urea were added to one volume of enzyme in 50 mM HEPES, 10% glycerol, pH 8.0. Any precipitated protein was removed by centrifugation for 5 min at 14000g, and the visible absorbance spectrum of the supernatant was recorded. The change in absorbance between the enzyme-bound FAD and the free FAD after denaturation gave an extinction coefficient of 11.7 mM−1 cm−1 at 458 nm for the flavin in Fms1 based on the extinction coefficient of FAD of 11.3 mM−1cm−1 (23). This value was used to determine the enzyme concentration.
Fms1 activity was typically determined in air-saturated buffer containing 10% glycerol by following oxygen consumption at 25 °C with a Yellow Springs Instrument Model 5300 biological oxygen monitor using 0.5 μM Fms1. Buffers were 200 mM Tris-HCl from pH 7.0 to 8.75, 200 mM CHES from pH 9.0 to 9.75, and 200 mM CAPS from pH 10.1 to 10.5. Substrate concentrations were from 50 to 600 μM for spermine, N1-acetylspermine, and N,N'-dibenzyl-1,3-diaminopropane, and from 0.1 to 3.0 mM for N,N'-dibenzyl-1,4-diaminobutane and spermidine. Kis and Kii values for 1,12-diaminododecane, 1,8-diaminooctane, and N,N′-dibenzyl-1,2-diaminoethane were determined with spermine as substrate; the concentrations of spermine and the inhibitors were varied from 0.1 to 2.0 and from 0 to 5.0 mM, respectively. The kcat value for N1-acetylspermine at 4 °C was determined using a coupled assay that measured H2O2 production. The reaction mixture contained ~45 nM Fms1, ~100 nM horseradish peroxidase, 76 μM o-dianisidine, and 500 μM N1-acetylspermine. The rate was calculated from the increase in absorbance at 436 nm, using an extinction coefficient of 8.31 mM−1 cm−1. For assays not performed in air-saturated buffer, the appropriate O2/N2 mixture was bubbled into the cell of the oxygen electrode for 10 min prior to starting the reaction by adding enzyme.
Rapid reaction kinetic measurements were performed with an Applied Photophysics SX-20MV stopped-flow spectrophotometer. For anaerobic experiments, we made the instrument anaerobic by filling the system with a solution of 30 nM glucose oxidase and 1 mM glucose in anaerobic buffer the night before the experiment was to be carried out. Oxygen was removed from enzyme solutions via application of several cycles of vacuum and oxygen-scrubbed argon, while substrate solutions were bubbled with oxygen-scrubbed argon. To maintain anaerobic conditions 36 nM glucose oxidase and 5 mM glucose were added to the solutions. To study the reaction of the reduced enzyme with oxygen, 60 μM enzyme was first mixed anaerobically in the stopped-flow instrument with either 500 μM spermine or 200 μM N1-acetylspermine. After an aging time of 300 or 20 ms for spermine or N1-acetylspermine, respectively, the reduced enzyme was mixed with buffer equilibrated with different oxygen concentrations. The concentration of substrate used and the delay time before the second mix were selected to minimize the excess of substrate while ensuring that formation of the reduced enzyme-product complex was at least 90% complete before the second mix. For pH-profiles the buffers were 200 mM PIPES from pH 6.0 to 6.8, 200 mM Tris-HCl from pH 7.0 to 8.5, and 200 mM CHES from 9.0 to 10.0. All rapid-reaction experiments with spermine as the substrate were conducted at 25°C, while those with N1-acetylspermine were at 4°C.
Kinetic data were analyzed using the program KaleidaGraph (Adelbeck Software, Reading, PA). Single value decomposition of diode array detector data to obtain the flavin spectra of intermediates was done using the program Sfit (Biologic Science Instruments, Grenoble, France). The Michaelis-Menten equation was used to determine kcat, kcat/KM, and KM values when initial rates were measured as a function of the concentration of a single substrate. Eq 1 was used to analyze inhibition data; here, Kis is the inhibition constant determined from the effect of the inhibitor on the kcat/KM value (slope) and Kii is the inhibition constant determined from the effect on the kcat value (intercept). Eq 2 was used to analyze the pH-dependence of kinetic parameters that decreased at both low and high pH. Eq 3 was used to analyze the pH-dependence of kinetic parameters that decreased only at low pH. In these equations, K1 and K2 are the dissociation constants for the ionizable groups, and C is the pH-independent value of the kinetic parameter of interest. For rapid-reaction studies of enzyme reduction, rate constants were obtained from the change in the visible absorbance of the flavin with time by fitting the data with eq 4, which describes a biphasic exponential decay. Here, λ1 and λ2 are the first-order rate constants for each phase, A1 and A2 are the absorbances of each species at time t, and A∞ is the final absorbance. The resulting pseudo-first rate constants were analyzed using eq 5, where k2 is the rate constant for flavin reduction at saturating concentrations of the amine substrate and Kd is the apparent dissociation constant for the substrate. The pseudo first-order rate constants for oxidation of reduced enzyme were obtained by fitting the change in absorbance of the flavin with time to eq 6. The second-order rate constant for oxidation was obtained by fitting the resulting rate constants as a function of the oxygen concentration as a straight line passing through the origin.
Since spermine and N1-acetylspermine have been reported to be comparable as the best substrates for Fms1 (11), the steady-state kinetics of Fms1 with both were examined in some detail. The analyses were carried out at the respective pH optima, 9.3 for spermine and 9.0 for N1-acetylspermine (see below). Flavoprotein oxidases can show either ping-pong or sequential (24) steady-state kinetic patterns when the concentrations of both substrates are varied, depending upon whether reduction of the flavin is kinetically reversible (25, 26). For a two substrate enzyme reaction, the steady state kinetic pattern can readily be identified using the fixed ratio approach (27). In this method, the ratio between the concentrations of the two substrates is kept constant, and the initial rate is determined as a function of the concentration of one of the substrates. Curvature in a double reciprocal plot suggests a sequential mechanism, while a linear plot arises from a ping-pong mechanism. For Fms1 a double reciprocal plot of the initial rate vs. oxygen concentration at a fixed ratio of either spermine or N1-acetylspermine to oxygen is linear (Figure 1), establishing the kinetic pattern as ping-pong. Fitting these data with the Michaelis-Menten equation gives kcat values of 39.0 ± 1.5 and 15.1 ± 0.4 s−1 for spermine and N1-acetylspermine, respectively. Since the fixed ratio approach establishes that the kcat/KM values for the amine substrate and oxygen are unaffected by the concentration of the other, the individual kcat/KM values for the substrates and oxygen were determined in separate analyses by varying each at a fixed concentration of the other. The individual KM values could then be calculated from the respective kcat/KM and kcat values. For comparison with the rate constants from rapid-reaction analyses, the kcat value with N1-acetylspermine was also determined at 4 °C by varying the concentration of oxygen at a saturating concentration of the amine. The resulting values are given in Table 1.
To gain further insight into the substrate specificity of Fms1, a number of substrate analogues (Scheme 2) were examined as substrates or inhibitors of the enzyme. While spermidine has been reported to not be a substrate for mammalian SMO or PAO (15, 17), it is a slow substrate for Fms1 at pH 9.0 (Table 1). Several N,N′-dibenzyl-diamines are either substrates or inhibitors of other polyamine-oxidizing flavoproteins (17, 28). Both N,N′-dibenzyl-1,3-diaminopropane and N,N′-dibenzyl-1,4-diaminobutane are slow substrates for Fms1 when assayed at the optimal pH for these compounds, pH 9.5 (Table 1). In contrast, N,N′dibenzyl-1,2-diaminoethane is a noncompetitive inhibitor (Table 2). 1,12-Diaminododecane and 1,8-diaminooctane are also noncompetitive inhibitors; the Kis and Kii values are provided in Table 2.
The effects of pH on the kcat/Kamine, kcat/KO2, and kcat values for spermine and N1-acetylspermine were determined. For both substrates the kcat/Kamine-pH profiles are bell-shaped, with maxima of 9.3 and 9.0 for spermine and N1-acetylspermine, respectively (Figure 2). These pH profiles show the importance of two ionizable groups in the free enzyme or substrate. The data were fit to eq 2 to extract the pKa values (Table 3). For N1-acetylspermine, the two pKa values are well separated; however, with spermine as a substrate the two pKa values are too close together to resolve, so that only the average pKa value of the two ionizable groups can be determined.
With either spermine and N1-acetylspermine as substrate for Fms1, the kcat/KO2 value is constant at high pH but decreases at low pH (Figure 3), indicating the importance of a single group that must be unprotonated for activity in each case. The data for both substrates can be fit to eq 3 to yield the pKa values given in Table 3. The kcat-pH profiles are shown in Figure 4. With both substrates the enzyme activity is constant at high pH but decreases at low pH. The data were consequently fit to eq 3 to give the pKa values in Table 3 for the moiety that must be deprotonated for high activity.
Less complete analyses were carried out of the effects of pH on the steady-state kinetics of Fms1 with N,N′-dibenzyl-1,4-diaminobutane and N,N′-dibenzyl-1,3-diaminobutane as substrates. With both, the kcat/Kamine value was maximal at pH 9.5, decreasing at higher and lower pH (results not shown). Because the activities with both substrates were low even at the pH optimum, more complete analyses were not carried out.
Stopped-flow methods were used to examine the kinetics of reduction of Fms1 by spermine and N1-acetylspermine. In initial experiments the reaction when Fms1 was mixed with either substrate in the stopped-flow spectrophotometer in the absence of oxygen was followed using a diode array detector. The analysis with spermine as a substrate was conducted at 25 °C, while for N1-acetylspermine at this temperature most of the reaction was complete within the dead-time of the instrument; decreasing the temperature to 4 °C slowed down the reaction with this substrate sufficiently that it could be followed. For both substrates the decrease in absorbance upon mixing enzyme with substrate was biphasic, with most of the absorbance change occurring in the much more rapid first phase. To obtain the spectra of the individual species, the spectra as a function of time were analyzed globally as a two-step irreversible reaction using the program Sfit. The resulting spectra with N1-acetylspermine are shown in Figure 5; those for spermine are similar. The visible absorbance spectrum of Fms1 in the absence of substrate is different from that of the species formed in the dead-time of the stopped flow. With either substrate the latter exhibits a shoulder at ~480 nm, suggesting that it is the oxidized enzyme-amine complex. In contrast, the spectra of the intermediate and final species have little absorbance above 400 nm, establishing that the flavin is reduced in both. These results are consistent with the mechanism of Scheme 3, in which binding of the substrate to the enzyme is followed by the reduction of the flavin and oxidation of the substrate in a single first-order step. The oxidized amine then slowly dissociates from the reduced enzyme.
To obtain the rate constants for the individual steps in Scheme 3, the changes in the flavin absorbance were monitored at 458 nm when the enzyme was mixed with different concentrations of spermine or N1-acetylspermine. With both substrates, only the first-order rate constant for the fast phase is substrate-dependent. The rate constants at different substrate concentrations for this phase could be fit to eq 5. The resulting k2 and apparent Kd values are listed in Table 4. The rate constant for the slow phase (k3 in Scheme 3) has values of 4.5 ± 0.1 and 0.42 ± 0.01 s−1 for spermine and N1-acetylspermine, respectively.
The effect of pH on the rate constants for the reductive half-reaction was also determined for both substrates. The k2-pH profiles are shown in Figure 6. With both substrates the value of k2 is pH-independent at high pH, decreasing at low pH, while the value of k3 is not affected significantly by pH. The data were fit to eq 3 to obtain the pKa values (Table 3) for the moiety that must be unprotonated for reduction.
Stopped-flow spectrophotometry was also used to analyze the kinetics of the reaction of the reduced enzyme - product complex with oxygen. This was done in double-mixing experiments in which Fms1 was first mixed with the amine substrate in the absence of oxygen and allowed to react to form the reduced enzyme-product complex. The sample was then mixed with oxygenated buffer and the reaction followed at 458 nm. The absorbance changes as a function of time could be fit as a single exponential increase in all cases. The resulting pseudo-first order rate constants vary directly with the oxygen concentration (Figure 7), consistent with an irreversible bimolecular reaction. The second-order rate constant for the reaction of the reduced enzyme with oxygen is 402 ± 15 with spermine and 204 ± 7 mM−1s−1 with N1-acetylspermine as the amine substrate.
The results presented here establish the kinetic mechanism of Fms1 and provide insight into the substrate specificity of this enzyme. The results of the kinetic analyses of Fms1 described here are consistent with the kinetic mechanism of Scheme 4 and the values for the individual rate constants given in Table 4. The ping-pong steady-state kinetic pattern for Fms1 with either spermine or N1-acetylspermine (Figure 1) can be attributed to reduction of the flavin (k2 in Scheme 4) being kinetically irreversible, so that binding of the amine and of oxygen are separated by an irreversible step (25). The rapid-reaction analyses provide the values of the individual rate constants (Table 4). The analyses of the reductive half-reaction yield the Kd value for binding of the substrate to the oxidized enzyme (EFlox), the first-order rate constant for reduction, k2, and the rate constant for product release from the reduced enzyme-product complex (EFLredP), k3. The rate constant for oxidation of the reduced enzyme-product complex, k4, is from the double-mixing stopped-flow experiment, and the rate constant for release of the product from the oxidized enzyme-product complex (EFloxP), k5, can be calculated.
The changes in the absorbance spectrum of Fms1 (Figure 5) when it is mixed with either spermine or N1-acetylspermine in the absence of oxygen are consistent with the kinetic mechanism of Scheme 3. Binding of the substrate to the enzyme occurs within the dead-time of the stopped-flow instrument (~1.5 ms). The shoulder at ~475 nm in the visible absorbance spectrum of the enzyme-substrate complex is consistent with spectral changes seen with other amine-oxidizing flavoproteins upon binding the substrate (29, 30). Reduction of the flavin in Fms1 and oxidation of the substrate occurs in a single first-order step, with no evidence for intermediates. The mechanism of amine oxidation by members of the MAO family has been a matter of controversy (22, 31, 32). Proposed mechanisms involving a radical intermediate or nucleophilic attack of the amine substrate on the flavin C4a position involve intermediates that should be detectable during the reductive half-reaction due to their distinct spectral properties. In contrast, a direct hydride transfer mechanism, supported by measurement of deuterium and 15N kinetic isotope effects (30, 33), does not involve such an intermediate. Spectral changes consistent with the formation of a flavin radical or a substrate-flavin adduct have generally not been observed during the reduction of members of the MAO/PAO family (16, 17, 30, 34, 35), and the present results with Fms1 are in line with that observation.
In the oxidative half-reaction of Fms1, the reaction of the reduced enzyme-product complex with oxygen occurs in a single pseudo-first-order process with no detectable intermediates. Kinetically, with both substrates the reaction behaves as a simple second-order reaction, with a rate constant equal to the kcat/KO2 value determined from the steady-state analysis. This agreement establishes that oxidation of both spermine and N1-acetylspermine includes only one path for the reaction of the reduced flavin intermediate with oxygen. The reaction of reduced flavins with oxygen is generally thought to proceed by an initial outer sphere electron transfer to form the flavin semiquinone and superoxide followed by rapid electron transfer to form either H2O2 and oxidized flavin or a flavin 4a-hydroperoxide (36-38). Neither intermediate is typically seen in the oxidation of a flavoprotein oxidase, with the exception of the recent observation of a flavin 4a-hydroperoxide in the reaction of a pyranose oxidase (39).
The rate constant for product release (k5) was not measured directly. However, it is possible to calculate the value of this rate constant. The kcat value contains all the first-order rate constants in the reaction. The only first-order rate constants in the mechanism of Scheme 4 are k2 and k5, so that kcat is described by eq 7. Since the kcat and k2 values for both substrates were determined, the value of k5 can be calculated as 56.5 ± 3.4 for spermine at 25 °C and 2.5 ± 0.2 s−1 for N1-acetylspermine at 4 °C. With both substrates the value of k5 is smaller than that of k2, the rate constant for amine oxidation, establishing product release as rate-limiting for Fms1. The relative kcat/Kamine values for the two substrates establish that Fms1 has a slight preference for N1-acetylspermine over spermine as a substrate. However, the kcat value for the latter is larger due to more rapid release of the oxidized amine from the enzyme at the end of the catalytic cycle.
The kcat/Kamine-pH profiles for Fms1 (Figure 2) are consistent with the enzyme being most active with forms of both spermine and N1-acetylspermine in which two of four nitrogens are protonated. For spermine, the concentrations of the mono-, di-, and triprotonated forms are maximal at pH 10.5, 9.5, and 8.4, respectively, while for N1-acetylspermine their concentrations are maximal at pH 10.3, 9.0, and 7.0 (16). The expected pH distributions of the diprotonated forms agree well with the pH optima in the kcat/KM-pH profiles for spermine of 9.3 and N1-acetylspermine of 9.0. For amine oxidation by a flavoprotein, the nitrogen at the site of oxidation must be unprotonated (16, 40). In the case of N1-acetylspermine oxidation by Fms1, this is N4 (Scheme 2). Since acetylation of N1 prevents it from being protonated, the protonated nitrogens are likely N9 and N12. While the present data do not allow unambiguous assignment of the protonation states of the nitrogens in the form of spermine that is the active substrate for SMO, the higher kcat/Kamine value for N1-acetylspermine versus spermine establishes that having N1 unprotonated does not result in a poor substrate. Thus, it is likely that in the active form of spermine both N9 and N12 are also charged. The structure of Fms1 with spermine bound (19) provides some support for this model. While the structure was determined at pH 5 and therefore represents an inactive complex in which all four nitrogens in spermine are likely fully protonated, it shows a clear interaction between spermine N122 and Asp94, consistent with that nitrogen being positively charged. Only weak interactions are seen with N1, and all involve uncharged atoms at 3.6-4 Å, so that the neutral nitrogen at this position in N1-acetylspermine could be accommodated3.
In the case of mouse PAO, the maxima in the kcat/KM-pH profiles for spermine and N1-acetylspermine correspond to the pH at which the substrate has a single charged nitrogen (16). The pH dependence of a series of spermidine analogs established that the nitrogen that is charged in the form of N1-acetylspermidine that is the active substrate for PAO corresponds to N9 in spermine and N1-acetylspermine. In contrast to both Fms1 and PAO, the kcat/KM-pH profile for spermine as a substrate for human SMO has a maximum at pH 8.3, consistent with an active form of the substrate in which all three of the non-reacting nitrogens are charged (17). N1-Acetylspermine is 100-fold worse a substrate for SMO than is spermine; the inability of the N1 nitrogen to be protonated provides a simple rationale for this substrate preference. Thus, despite similar structures and identical reactions, these three enzymes require differently protonated forms of their common substrates.
The results with the N,N′-dibenzyl-diamine compounds support the conclusion that different requirements for charge in the substrate contributes to substrate specificity in these enzymes. Fms1 will use N,N′-dibenzyl-1,4-diaminobutane and N,N′-dibenzyl-1,3-diaminopropane as substrate, with a slight preference for the latter (Table 1). Mouse PAO will also oxidize these compounds with a similar preference, but with kcat/Kamine values that are about 10-fold greater (28). In contrast, human SMO has no detectable activity with either compound. The active forms of diamine substrates can have only one protonated nitrogen, since one must be neutral for oxidation. Since PAO prefers the singly protonated form, both compounds are reasonable substrates. They are worse substrates for Fms1, consistent with the lack of a second charged nitrogen. The lack of detectable activity with SMO can be attributed to the lack of two of the three charged nitrogens preferred by that enzyme.
The pH dependence of the inhibition of PAO and SMO by diamines provided further insight into the specificity of those enzymes (16, 17). However, the compounds examined here, 1,12-diaminododecane, 1,8-diaminooctane, and N,N′-dibenzyl-1,2-diaminoethane are noncompetitive inhibitors of Fms1. This suggests that they bind to the protein in a different fashion than the substrates. This conclusion is supported by the crystal structures of Fms1 with 1,8-diaminooctane (pdb entry 3bi5) or 1,12-diaminododecane (3bi5), which show that both compounds bind in the active site tunnel but much farther from the FAD than spermine.
A straightforward explanation for the pKa seen in the k2-pH profiles for Fms1 with both spermine and N1-acetylspermine is that it is due to N4 of the substrate, which must be neutral for the amine to be oxidized. If the forms of substrates in which N4, N9, and N12 are all protonated can bind but not react, the pKa for N4 when bound to the enzyme will be seen in this profile. The alternative possibility is that these pKas are due to an amino acid residue on the protein. The crystal structure of Fms1 in the complex with spermine (19) shows that likely candidates are His67 and His191. Both mouse SMO and human PAO exhibit k2-pH profiles similar to those in Figure 6 (16, 17). With both enzymes the corresponding pKa was similarly attributed to N4 in the enzyme-substrate complex. However, in the case of SMO the effects of pH on the binding of competitive inhibitors implicated an amino acid in binding, suggesting that the pKa of 7.4 seen in the k2-pH profile for that enzyme could instead belong to a His residue in the active site of the protein (17). The present results cannot resolve this ambiguity.
The pKa seen in the kcat/KO2-pH profiles (Figure 3) depends on the substrate, consistent with it reflecting the need for a moiety in the reduced enzyme-product complex to be uncharged. Mouse PAO shows a pKa of 7.0 in the kcat/KO2-pH profile with N1-acetylspermine as substrate (41), close to the value of 6.8 described here for Fms1. This pKa is not seen when Lys315 in PAO is mutated to methionine. While there is no structure of a mammalian PAO available, sequence alignments support Lys315 in mouse PAO as a lysine residue that is conserved in the MAO/PAO family. Structures of several family members show that the amino group of this lysine forms a water-mediated hydrogen bond with the N5 of the FAD (42, 43), although this water was not seen in the Fms1 structure (19). The results with K315M PAO suggest that the pKa in the kcat/KO2-pH profile for Fms1 is due to this lysine, Lys296.
Since the kcat value for Fms1 predominantly reflects the rate constant for product release, the pH behavior of the kcat values reported here establish that the protonation of a group in the oxidized enzyme-product complex decreases the rate constant for release of the product from the oxidized enzyme. A similar pH-dependent release of the oxidized amine product has been described for other flavoproteins that oxidize amines: D-amino acid oxidase (44), tryptophan monooxygenase, an L-amino acid oxidase from the MAO structural family (45, 46), and SMO (41). Mutating Lys315 in SMO has no effect on the kcat-pH profile, ruling out the conserved lysine as the source of the pKa.
In conclusion, the results obtained here provide insight into the catalytic mechanism of Fms1. The complete kinetic mechanism of Fms1 has been determined, and the individual rate constants have been measured. The diprotonated forms of the polyamine substrates are the active forms. Product release from the oxidized enzyme is rate-limiting for turnover.
†This work was supported in part by NIH grant GM058698.
2The numbering used for spermine in the pdb file 1xpq and that used here are different. We chose to number the nitrogens in spermine to correspond to the typical numbering in N1-acetylspermine in order to simplify the comparison between the two substrates. In pdb file 1xpq the atoms are numbered consecutively from 1-14 in the opposite direction from that used here. Thus, N1 as used here corresponds to N14 in the structure file, N4 to N9, N9 to N5 and N12 to N1.
3File 3cnd in the PDB is for the Fms1 complex with N1-acetylspermine. The N1-acetylspermine in that structure does not align with the spermine in file 1xpq, and no atoms of the substrate are within 4 Å of the flavin N5 position. Based on these differences, it is likely that the N1acetylspermine binding mode depicted is not appropriate for catalysis and thus provides limited insight into the structural basis for the substrate specificity of Fms1.