If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* This work was supported by research grants from the National Institutes of Health and the New Zealand Foundation for Science and Technology. 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.
Genetic deficiency of human purine nucleoside phosphorylase (PNP) causes T-cell immunodeficiency. The enzyme is therefore a target for autoimmunity disorders, tissue transplant rejection and T-cell malignancies. Transition state analysis of bovine PNP led to the development of immucillin-H (ImmH), a powerful inhibitor of bovine PNP but less effective for human PNP. The transition state of human PNP differs from that of the bovine enzyme and transition state analogues specific for the human enzyme were synthesized. Three first generation transition state analogues, ImmG (Kd = 42 pm), ImmH (Kd = 56 pm), and 8-aza-ImmH (Kd = 180 pm), are compared with three second generation DADMe compounds (4′-deaza-1′-aza-2′-deoxy-1′-(9-methylene)-immucillins) tailored to the transition state of human PNP. The second generation compounds, DADMe-ImmG (Kd = 7pm), DADMe-ImmH (Kd = 16 pm), and 8-aza-DADMe-ImmH (Kd = 2.0 nm), are superior for inhibition of human PNP by binding up to 6-fold tighter. The DADMe-immucillins are the most powerful PNP inhibitors yet described, with Km/Kd ratios up to 5,400,000. ImmH and DADMe-ImmH are orally available in mice; DADMe-ImmH is more efficient than ImmH. DADMe-ImmH achieves the ultimate goal in transition state inhibitor design in mice. A single oral dose causes inhibition of the target enzyme for the approximate lifetime of circulating erythrocytes.
Subsequent investigations demonstrated that the biochemical link between PNP and T-cell deficiency was the failure to degrade deoxyguanosine and its conversion to dGTP in activated T-cells (
). Stimulated T-cells require DNA synthesis for cell division, and excess dGTP acts to allosterically inhibit ribonucleotide-diphosphate reductase, causing unbalanced deoxynucleotide pools and the induction of apoptosis (
). Dividing T-cells induce deoxycytidine kinase activity relative to resting T-cells, and this enzyme also accepts deoxyguanosine when blood levels are elevated (
). Under normal conditions the blood level of deoxyguanosine is less than 0.1 μm because of high levels of PNP in human erythrocytes and in liver, spleen, and intestine (
). In PNP-deficient patients the deoxyguanosine level in blood can reach 10 μm, causing its accumulation as dGTP in stimulated T-cells.
Structure-based inhibitor design using early x-ray crystal structures of human PNP led to a family of inhibitors with dissociation constants to 15 nm, but these were insufficient to cause the sustained elevation of blood deoxyguanosine needed to suppress division of activated T-cell populations (
). Transition state analogue inhibitor design based on the transition state structure of bovine PNP led to the design and synthesis of ImmH, a 23 pm inhibitor for the bovine enzyme but a 2–3-fold weaker inhibitor for the human enzyme (
). In the presence of deoxyguanosine, ImmH induces apoptosis in dividing human T-cells but not other cell types. Normal human T-cells that are not stimulated to divide are not affected by immucillin-H and deoxyguanosine (
). When a transition state analogue binds with different strengths to isozymes such as the human and bovine PNPs, a difference in their transition state structures is likely to be the cause. Despite an 87% amino acid identity between human and bovine PNPs, differences at the homotrimeric subunit interfaces cause differences in catalytic site cooperativity and, apparently, differences in transition state properties (
A. Lewandowicz and V. L. Schramm, unpublished observations.
established that the transition state is more dissociated for the human enzyme, with greater separation between the ribosyl group and the departing purine ring. Although immucillin-H is well suited to the transition state of bovine PNP, DADMe-immucillin-H is better matched to the more dissociated transition state (Fig. 1). Here we demonstrate that the DADMe-immucillins are more powerful inhibitors of human PNP than the first generation immucillins designed to match the transition state of bovine PNP. The oral availability of DADMe-immucillin-H is established in a mouse model, and its biological efficiency is compared with immucillin-H. DADMe-immucillin-H is a sufficiently powerful inhibitor of PNP in circulating mouse erythrocytes that the recovery of blood PNP activity occurs primarily by synthesis of new erythrocytes.
Fig. 1First and second generation inhibitors for human purine nucleoside phosphorylase. The values given are dissociation constants for the tightly inhibited complex after slow onset is complete. This value is often called Ki* (
). All compounds except 8-aza-DADMe-ImmH are slow-onset inhibitors. The numbering of atoms (shown for inosine in Fig. 2) is also used for the immucillins and DADMe-immucillins, although it does not conform to IUPAC rules.
). Polymerase chain reaction, with primers purchased from Invitrogen, was performed to amplify PNP DNA, which was cloned into a T7/NT TOPO vector and transformed in BL21(DE3) Escherichia coli strain.
Inhibitor Synthesis—Immucillin-H was synthesized from d-gulonolactone to introduce appropriate stereochemistry in the iminoribitol ring as described earlier (Fig. 1; Refs.
). DADMe-immucillins were synthesized via the reductive amination of 5-N-benzyloxymethyl-4-methoxy-3H,5H-pyrrolo[3,2-d]pyrimidine-7-carbaldehyde, derived from the lithiation and subsequent formylation of 5-N-benzyloxymethyl-7-bromo-4-methoxy-3H,5H-pyrrolo[3,2-d]pyrimidine with (3R,4R)-3-hydroxy-4-hydroxymethylpyrrolidine. Acid deprotection and chromatography provided DADMe-ImmH as the HCl-salt, which was characterized by NMR, mass spectroscopy, and elemental analysis as outlined earlier (Fig. 1; Ref.
G. B. Evans, R. H. Furneaux, A. Lewandowicz, V. L. Schramm, and P. C. Tyler, submitted for publication.
Inhibition Studies—Inhibitor dissociation constants for phosphorolysis of inosine by PNP were based on reaction rate measurements with different inhibitor concentrations. Reactions were started with the addition of 0.05 μg of human PNP (final concentration 1.4 nm) to 1 mm inosine in 50 mm KPO4 (pH 7.5) buffer with xanthine oxidase added to a final concentration of 60 milliunits/ml at 25 °C. In the coupled assay, hypoxanthine formed by phosphorolysis of inosine was oxidized to uric acid and followed spectrophotometrically at 293 nm (extinction coefficient for uric acid, ϵ293 = 12.9 mm–1) (
). The dissociation constant for slow-onset, tightly binding inhibitors was determined from reaction rates after slow-onset inhibition had occurred according to the equation ν = (kcat × S)/(Km(1 + I/Kd) + S), where ν is the steady state reaction rate after the slow-onset inhibition period has reached equilibrium, kcat is the rate at substrate saturation, S is substrate concentration, Km is the Michaelis constant for inosine (38 μm under these conditions), I is inhibitor concentration, and Kd is the equilibrium dissociation constant for the tightly inhibited PNP-inhibitor complex (
). For inhibitors without slow-onset properties, fits were made to the same equation but using initial reaction rates.
Rate Constants for Dissociation of PNP·Immucillin·PO4—Inhibition of PNP by the immucillins is a two-step process to form the tightly bound complex of PNP (
). The rate of inhibitor release was estimated by the recovery of enzyme activity following dilution of stoichiometric complexes formed by prolonged incubation similar to experiments described earlier for immucillin-H and immucillin-G with bovine PNP (
). Briefly, human PNP (96 μm), inhibitors (105–108 μm), and PO4 (50 mm, pH 7.5) were incubated at 25 °C for 5.5 h to assure equilibrium. Mixtures were diluted 1:500,000 by a 1:500 dilution into 20 mm Tris-HCl (pH 7.5) followed immediately by a 1:1000 dilution into assay mixtures containing 1 mm inosine. The rate constant for dissociation of inhibitors was calculated from fits to the initial regaining of catalytic activity as described previously (
). These experiments also demonstrate the reversible inhibition caused by immucillins.
Oral Availability of ImmH and DADMe-ImmH—A solution containing 10–7 mol (27 μg) of DADMe-ImmH and 3 mg of glucose was pipetted into the mouth of a 32-g Balb-c mouse that had been fasted overnight. The oral uptake of ImmH was measured by pipetting 27 μg in 10 μl of solution onto solid food and fed to mice under observation. Small samples (5 μl) of blood were collected from the tail and added to 6 μl of PBS (140 mm NaCl, 3 mm KCl, 10 mm KHPO4 (pH 7.4)) containing 1 unit of heparin and 0.3% Triton X-100. Control blood samples were collected before administration of immucillins to the mice. The mixture was kept on ice for 25 min and frozen in dry ice/ethanol for storage at –70 °C. For catalytic activity of whole blood samples, 10 μl of the lysate was added to 735 μl of complete reaction mixture containing 1 mm inosine, 50 mm phosphate (pH 7.4), and 60 milliunits/ml xanthine oxidase. After mixing, the reaction progress was followed spectrophotometrically at 293 nm. In separate experiments, samples of tail blood (5 μl) were pipetted into 50 μl of ice-cold PBS, centrifuged in the cold, and resuspended in two additional samples of fresh PBS, and the pellet of washed erythrocytes was finally lysed in 5 μl of PBS containing Triton X-100 and assayed as described above.
RESULTS AND DISCUSSION
Inhibition of PNP by Immucillins—Human PNP was inhibited by ImmG, ImmH, and 8-aza-ImmH to give equilibrium dissociation constants (Kd) of 42, 56, and 180 pm, respectively (see Fig. 1). The kinetic properties exhibited slow-onset, tight-binding inhibition typical of inhibition by compounds that resemble transition state analogues (
). The immucillins were designed to match the transition state for bovine PNP, which has a C-1′—N-9 ribosidic bond of 1.77 Å at the transition state (Fig. 2; 10). The immucillin transition state analogues are found to be better inhibitors of the bovine enzyme than the human PNP (Fig. 2; Ref.
), establishing that the transition states of the two enzymes differ, despite 87% amino acid sequence identity and the observation that every amino acid at the catalytic site is conserved (
). The 9-deazahypoxanthine structure increases the pKa at N-7 to above 10, and the N-7 of substrate at the transition state is known to be a proton acceptor. The 4′-N in the iminoribitol has a pKa of 6.9 in ImmH and is protonated at the catalytic site to resemble the cationic feature of substrates at the transition state (
Fig. 2Transition state structures for bovine and human PNP. The transition states have been determined from kinetic isotope effect studies and computational matching of the isotope effects. The distances for the transition states have been reported for bovine PNP (
), and lower limits have been set for the transition state geometry for human PNP (see Footnote 2). The distances of 1.5 Å for immucillin-H and 2.5 Å for DADMe-immucillin-H refer to the linear distance between the deazapurine ring and C-1′ or N-1′ of the ribosyl analogues.
Design Principles of the DADMe-Immucillins—Recent kinetic isotope effect studies with human PNP have established that the transition state structure differs from that of the bovine enzyme, to give a C-1′—N-9 ribosidic bond length of greater than 2.5 Å and weak participation of the anionic nucleophile (>3.0 Å for arsenate). These features cause increased cation character at C-1′ compared with the transition state for bovine PNP (Fig. 2).2 The immucillins have a covalent C–C bond between the iminoribitol and the 9-deazapurine group, and this bond length is near 1.5 Å, making it at least 1.0 Å shorter than the corresponding atoms at the transition state. In more dissociated transition states like that of the human PNP, the cationic charge on the ribooxacarbenium ion becomes more highly developed and centered at the C-1′ carbocation rather than being distributed over O-4′, C-1′, and N-9 when the transition state has significant residual bond order remaining between C-1′ and N-9 (
). We proposed that spacing the iminoribitol and the 9-deazapurine rings closer to the actual transition state, and placing the carbocation charge in the inhibitor closer to its location in the transition state, may increase the affinity specifically for the human PNP. The DADMe-immucillins (Fig. 1) were synthesized as outlined under “Experimental Procedures.” The methylene bridge between the deazapurine and ribosyl analogue groups of DADMe-immucillins adds ∼1 Å linear distance between these groups but does not alter the favorable pKa of the 9-deazahypoxanthine. Moving the nitrogen to the position corresponding to C-1′ retains the cationic nature to more closely mimic the ribooxacarbenium ion transition state because N-substituted hydroxymethylpyrrolidines have pKa values near 10 and will be fully protonated at physiological pH values near 7.5 (
Inhibition of Human PNP by DADMe-Immucillins—Human PNP binds immucillin-G more tightly than other members of the immucillin family to give a Kd of 42 pm (Fig. 1). Likewise, DADMe-ImmG is the most tightly bound member of the DADMe family with a Kd of 7 pm, improving the affinity by a factor of 6. In parallel with the ImmG and DADMe-ImmG affinities, DADMe-ImmH binds a factor of 4 more tightly than ImmH and, like DADMe-ImmG, gives slow-onset, tight-binding inhibition (Fig. 3). This pattern does not continue with the 8-aza-DADMe-ImmH. Thus, 8-aza-ImmH binds with a Kd of 0.18 nm, whereas 8-aza-DADMe-ImmH binds with a dissociation constant of 2.0 nm and does not exhibit slow-onset inhibition. The affinity of 8-substituted ImmH analogues with the bovine PNP have shown that the pKa at N-7 is an important parameter for immucillin binding (
). Loss of binding energy in 8-aza-DADMe-ImmH relative to 8-aza-ImmH reveals that either the pKa or the contact geometry between N-7 and Asn-243 is not as favorable as it is in 8-aza-DADMe-ImmH.
Fig. 3Slow-onset, tight-binding inhibition of human PNP by DADMe-ImmH. The production of hypoxanthine from inosine is monitored by conversion to uric acid in a coupled assay as indicated under “Experimental Procedures.” The concentrations of DADMe-ImmH are indicated. The inset is a replot of the initial rate (no inhibitor) compared with rates at 40–50 min with inhibitor. This replot is used to calculate the Kd value from the equation for competitive inhibition.
The kinetics of DADMe-ImmH inhibition of human PNP demonstrate two distinct inhibitory phases (Fig. 3). In the first few minutes, the initial rate slopes change rapidly to reach a second steady state plateau of increased inhibition. This is characteristic of transition state analogue inhibitors and reflects a slow enzymatic conformational change to form the tightly bound complex: E + I ↔ EI ↔ E*I, where E is free enzyme, I is inhibitor, EIisthe rapidly reversible complex of enzyme and inhibitor, and E*I is the tightly bound, slowly releasing form of the inhibited complex. The Kd values reported here are calculated from the slopes of the curves at 40–50 min, where the slow-onset phase of inhibition has been completed. This part of the curve provides the overall dissociation constant for E*I (
In Vivo Inhibition of PNP by the DADMe-Immucillins—The goal of the PNP inhibitor design program is to use transition state design principles to develop more effective inhibitors against mammalian PNPs in vivo. In vitro measurements suggest that the DADMe-immucillins should be the more effective inhibitors in vivo, provided that both inhibitor families have similar bioavailability. We elected to compare ImmH and DADMe-ImmH for bioavailability in mice, because the in vivo behavior of ImmH is becoming well characterized from studies in mice and primates (
). Oral administration of 0.1 μmol (27 μg, 0.8 mg/kg) of ImmH or DADMe-ImmH caused a rapid loss of the PNP activity in mouse blood with a loss of 50% of total catalytic activity in blood occurring in 10 and 14 min for DADMe-ImmH and ImmH, respectively (Fig. 4). The catalytic activity protocol of whole blood lysis does not establish whether the immucillins penetrate the erythrocyte compartment, because serum ImmH could inhibit erythrocyte PNP following lysis. In parallel experiments, blood samples were washed in cold buffer to remove extracellular immucillins followed by lysis and assay. The inhibition of erythrocyte PNP was complete for both immucillins in this assay, demonstrating the cellular permeability of these inhibitors.
Fig. 4Oral availability for ImmH and DADMe-ImmH in mice. The t½ for onset is the time after oral administration that 50% of PNP activity in blood is inhibited. The t½ for DADMe-ImmH is 10 min and for ImmH is 14 min as indicated under “Experimental Procedures.” The t½ for PNP activity recovery following the single oral dose is 100 h for ImmH (open circles) and 275 h for DADMe-ImmH (closed circles). The ordinate scale is uric acid formation (A293/min) using the assay described under “Experimental Procedures.”
Thus, both ImmH and DADMe-ImmH are readily available from the gastrointestinal system and equilibrate with whole blood PNP more rapidly than they can be removed by hepatic or renal systems. Continued monitoring for the recovery of enzymatic activity demonstrated that 50% of PNP in the blood regained activity at 100 h (4.2 days) with ImmH (Fig. 4). In contrast, blood PNP regained activity more slowly with DADMe-ImmH, with 50% activity regained at 275 h (11.5 days).
In long recovery times for erythrocyte PNP activity, factors other than inhibitor release and metabolism must be considered. The average life span for mouse erythrocytes is 25 days (
). Therefore, after 12.5 days, half of the erythrocytes present at the start of the experiment will have been replaced by new erythrocytes. When the red cell replacement rate is considered, ∼90% of the PNP activity recovery that occurs over 11.5 days is due to synthesis of new erythrocytes. The in vivo inhibition of mouse blood PNP by a single dose of DADMe-ImmH therefore approximates the life span of mouse erythrocytes. The enhanced inhibition seen for DADMe-ImmH in enzyme inhibition experiments with human PNP is confirmed in the whole animal mouse model. The DADMe-immucillins represent the culmination of transition state inhibitor design in efficient inhibition of the target enzyme following a single oral dose.
Comparison of Off-rates and Biological Dissociation Rates— The biological half-life of DADMe-ImmH in the mouse results from the competition between binding to erythrocyte PNP and elimination from the animal. In the simplest case, the biological half-life would be directly related to the t½ for the E*I complex. That is not the case for the immucillins, with binding half-lives of 8–120 min for ImmH, DADMe-ImmH, and DADMe-ImmG (Table I). These are similar to the rates reported for other tightly binding inhibitors of enzymes (Table I). The concentration of PNP inside mouse erythrocytes is ∼10–6m, and the Kd for DADMe-ImmH is approximately 10–11m. Release of DADMe-ImmH from an intracellular catalytic site in the presence of the 105 molar excess of PNP makes recapture more likely than release by diffusion. Therefore, with a t½ of 20 min for DADMe-ImmH release at near infinite dilution, we propose that once trapped inside mouse erythrocytes, each molecule of DADMe-ImmH binds sequentially to hundreds of PNP molecules before it is lost to the plasma. Rebinding of DADMe-ImmH is sufficient to cause nearly full inhibition of target PNP for the 25-day lifetime of mouse erythrocytes. As indicated above, ∼90% of PNP activity recovery corresponds to synthesis of new erythrocytes. We hypothesize that rebinding traps the inhibitor into individual cells and at the death of these cells, the inhibitor is excreted or metabolized more efficiently than being taken up by developing cells.
Table IEnzyme-inhibitor release rates and dissociation constants compared with human PNP and immucillins
The Kd value for slow-onset inhibitors is governed by the expression Kd = (Kik 6)/(k 5 + k 6), where Ki is the initial dissociation constant ([E][I]/[EI]) and k 5 and k 6 are the rates of formation and relaxation of the tightly bound complex (E**I) from (EI). Kd is the equilibrium dissociation constant = [E][I]/[E**I], where [E] is free enzyme concentration, [I] is inhibitor concentration, and [E**I]. Ki reflects the capture of the inhibitor, and Kd reflects overall equilibrium binding energy (22).
Aspartate transcarbamylase
PALA
1 min
5 nm
E. coli DHF reductase
Trimethoprim
8 min
20 pm
HMG-CoA reductase
Compactin
15 min
240 pm
Chicken DHF reductase
Methotrexate
35 min
9 pm
Xanthine oxidase
Allopurinol
300 min
630 pm
Adenosine deaminase
Deoxycoformycin
40 h
2 pm
Human PNP
Immucillin-H
8 min
56 pm
Human PNP
DADMe-ImmH
20 min
16 pm
Human PNP
DADMe-ImmG
120 min
7 pm
aThe value of t½ is the time required to regain 50% of uninhibited enzyme activity.
bThe Kd value for slow-onset inhibitors is governed by the expression Kd = (Kik6)/(k5 + k6), where Ki is the initial dissociation constant ([E][I]/[EI]) and k5 and k6 are the rates of formation and relaxation of the tightly bound complex (E**I) from (EI). Kd is the equilibrium dissociation constant = [E][I]/[E**I], where [E] is free enzyme concentration, [I] is inhibitor concentration, and [E**I]. Ki reflects the capture of the inhibitor, and Kd reflects overall equilibrium binding energy (
Conclusions—Transition state theory teaches that the most powerful noncovalent enzyme inhibitors can be obtained with chemically stable mimics of the transition state structures of the cognate enzymes. Perfect analogues of enzymatic transition states can never be obtained, because of the non-equilibrium bond lengths and charges that exist at the transition state. However only a fraction of the 1012-fold increased binding potential theoretically available for PNP analogues is required for biological efficiency. In the case of mammalian PNP inhibitors, we have achieved 5 × 106-fold tighter binding, sufficient to reach the ultimate goal in transition state analogue design. DADMe-ImmH binds sufficiently tightly to the target enzyme in an in vivo mouse model to be released only on the time scale of cellular replacement.
Acknowledgments
We acknowledge Drs. Robert W. Miles and Gregory A. Kicska for the mouse experiment with immucillin-H.
83934-9&id=gr1.jpg){kind=link}
83934-9&id=gr2.jpg){kind=link}
83934-9&id=gr3.jpg){kind=link}
83934-9&id=gr4.jpg){kind=link}