- Email: [email protected]
Inhibition of β-glucosidase by imidazoles
Biock:mica et Biophysica Acta, 999 (1989) 227-232
227
~sevier BBAPRO 33499
Inhibition of fl-glucosidase by imidazoles Y a w - K u e n Li a n d L a r r y D. B y e r s Department of Chemistry, Tulane University, New Orleans, LA (U.S.A.)
(Received15 June 1989) Key words: ~-Glacosidase;Phenylimidazole Over 25 nitrogen-containing heterocydes were tested as inhibitors of sweet almond fl-glucosidase (EC 3.2.1.21). Among the most potent of these are some imidazole derivatives. The pH dependence indicates that the unprotonated inhibitor hinds most tightly to the catalytically active species of the enzyme. This is analogous to the ~itoation with I-denxyno|irimyein where the permanently cationic ~pecies, N,N-dimethyl-l-deoxynojirimycin, binds at least two orders of magnitude less tightly to the enzyme than does the unprotonated l-deoxynojirimycin. The binding of imidazole derivatives show a general tendency of increasing affinity with increasing basicity (fl -- 0.4). One derivative which shows a significant positive deviation from this correlation ( - l o g K i vs. p K , ) is 4-phenylimidazole. 4-Phenylimidazole is one of the most potent reversiMe inhibitors of fi-gineosidase with a pH-independent Kt ffi 0.8 pM. it is also fairly specific for fl-glneosidase, binding at least three orders of magnitude less tightly to any of the other exoglycosidases tested. This inhihitor combines, in a mono-molmflar species, the binding affinities of benzene, which binds at the h y ~ aglyenne binding site, and imidazole, which binds at the sugar binding site of ~-glucosidase. The binding energy of 4-phenylimidazole can be attributed to the sum of the intrinsic binding energies of the phenyl and imidazole moieties. Thus, there is no significant entropic advantage of combining the component parts of phenylimidazele in a single species. This indicates that there is no significant uncompensated entropy loss upon binding of either benzene or imidazole to the enzyme. Nevertheless, the additivity of binding energy, even in tim absence of an entropic advantage, results in the most powedul known inhibitor of the enzyme.
Introduction Glycohydrolases are widesprcacl in nature and serve a variety of functions, lnlubitors of these enzymes are of interest for many reasons. They have served as useful tools in studying glycoprotein biosynthesis and oligosaccharide chain processing [1] and they are potentially useful as pesticides or as therapeutic agents in the treatment of some viral-induced diseases [2]. fl-Glucosidase inhibitors, in particular, are potentially useful in developing animal models for glucosyl ceramide lipidoses and have already been shown to be useful in differentiating the properties of normal and Gaucher's di .sease enzymes [3]. We have been interested in fl-glucosidase inhibitors not only as probes of active-site topography, but as tools for investigating the catalytic
Abbreviations: DBN. 1,5-diazabicyclo{4.3.0]non-5-en¢;1-dNM, 1-deoxynojirimycin (5-amino-l,5-dideoxy-D-ghcopyranose);DNI~3, dinitrophenyl-tB-D-glucopyranosid©;Mes, 4-morpholineethanesulfonate; Mops, 4-morpholinelpropanesulfonate. Correspondence: L.D. Byers,Departmentof Chemistry+,Tulane Unto versity, New Orleans, LA 70118, U.S.A.
mechanism of the enzyme. The most potent naturally occurring inhibitors of B-glucosidase are hetcrocycfic compounds containing at lea~t or, e rfittogen atom (e.g., the 5-amino-5-deoxyglucopyranose, nojirimycin; the 1,5-iminoalditol, 1-deoxynojirimycin, 1-dNM, and the indolizine alkaloid, catanospermine). The basic charactec of these compounds contributes to their affinity for the enzyme. Replacement of the nitrogen in 1-dNM for an oxygen reduces its affimty for sweet almond l~-glucosidase by nearly four orders of magnitude [4]. The pH dependence of the binding, however, suggests that it is the unprotonated form of the inhibitor which binds to fl-glucosidase [4]. The strong interaction of the uncharged inhibitor with the enzyme is a characteristic of most glycosidases [1]. It should be noted, however, that, at least in the case of the almond enzyme, this assignment is ambiguous because of a coincidence of pK, values [4]. Thus, it is possible that binding of protonated 1-dNM (PKa = 6.7) to the unprotonated and inactive enzyme could be much stronger than binding of the unprotonated 1-dNM 1o the protonated enzyme (pK, = 6.6). This study was undertaken to resolve this ambiguity and to examine how the basic properties of nitrogen-containing heterocycles facilitates their interaction with the enzyme.
0167-4838/89/$03.50 © 1989ElsevierSciencePublishers BN. (BiomedicalDivision)
228 TABLE !
Materials and Methods
Some ~8-glucosidaseinhibitors Most of the imidazoles and pyridine derivatives were obtained from Aldrich Chemical Co. 4-(p-Fluorophenyl)imidazole (m.p. 1 2 4 - 1 2 5 ° C , lit. 1 2 5 - 1 2 7 ° C ) and 4-trifluoromethylimidazole (m.p. 148-149 o C, lit. 150-151°C) were prepared as described by Bredereck et al. [5]. 4 - ( P e n t a f l u o r o p h e n y l ) i m i d a z o l e (m.p. 1 6 8 - 1 7 0 ° C , lit. 169-1710C) was synthesized photochemically as described by Kimoto and Fujii [6]. 4 - ( p Nitrophenyl)imidazole (m.p. 2 2 2 - 2 2 4 ° C , lit. 225°C) and 4-(o-nitrophenyl)imidazole (m.p. 144--145°C, lit. 146°C) were prepared by nitration of 4-phenylimidazole in H : S O 4 - H N O 3 (1 : 6) for 60 h at room temperature [7]. lmidazo[1,2-aJpyridine (perchlorate salt m.p. 2400C, fit. 241-243.5°C) was prepared as described by Roe [8]. 3-Phenylsydnone (m.p. 1 3 2 - 1 3 4 ° C , lit. 134-134.5° C) was prepared by the method described as Earl et al. [9]. The dinitrophenyl glucosides were prepared by reaction of the 2,3,4,6-tetra-O-acetyl-a-Dglucopyranosyl bromide (Sigma) with 2,4-dinitrophenoi (Sigma) or 3,4-dinitrophenol (Fiuka) by the methods of Ballardie et al. [10] and Sinnott and Viratelle [11], respectively. N-Methyl- and N,N.dimethyl-l-deoxynojir;mycin were generous gifts from G. Legler. All enzymes, including sweet almond /g-glucosidaze (type i, spec. act. ~ 20 units/rag), were obtained from Sigma Chemical Co. The B-glucosidase concentration was estimated by absorbance at 277 n m with E t~ = 7.25 [121. The buffer system usually consisted of 0.1 M NaCi, 0.1 m M E D T A and 0.01 M buffer: sodium acetate ( p H 3.5-5.6); Mes ( p H 5.6-7.0); Mops ( p H 7.0-7.9). Kinetics. The /~-glucosidase-~talyzed hydrolysis of dinitrophenyl glucosides was monitored spectrophotometrically. Rapid kinetics (of inhibitor binding) were followed on a Dionex Model 13-110 stopped-Row spectrophotometer interfaced with a Biomation 810 transient digitizer and a strip-chart recorder. In general, kinetics were followed on a Hewlett Packard Model 8452A Diode Array Spectrophotome!er. The reactions were initiated by addition of enzyme to a solution of substrate in buffer which ,had b,~en thermally equilibrated at 27 o C. The dissociation constant of enzymeinb./bitor complex, K,, was ~ ! c u l a t e d from the effect of inhibitor on the ratio of the two steady-state parameters
(g/V)+ - (g/Y)o(l+lll/g,)
(1)
( K / V ) + is the ratio of K m to Vmax in the presence of the inhibitor. The parameter K / V was determined under first-order conditions (IS] 4: Kin). The first-order rate constants, V/K, were evaluated b y non.linear regression [13]. In several cases the K i was also determined by following the reaction in the presence of inhibitor under initial velocity conditions at various
Inhibitor 2-(l-Naphthylmethyl)-I imidr~zolinc DBN 4-Phenylimidezole 4,~Dimethyfimi~fine
pK, 10.4 c 12.0 d 6.1 c 9.5 ¢
K i " (M)
Kitimb (M)
3.1-10- I 3.2.10-1 6.2-10 -6 3.7.10- 2
(4.~. 10-8) (1.2.10-7) (4.2.10- s)
8.3.10 -7
Imidazok
7.1 ¢j
9.3-10 -4
2.6. I0 -s
2-Aminopyridine imidazol¢[1,2-a]pyridine
6.9 d 6.9 ¢
3.2-10 - 3 8.1-10- s
3.5" | 0 - 4
aenzidmidazole 2.Phenylpyridine
5.4 d 4.5 d
1.3. I0- 3 1.1-10 - 3
7.1. I 0 - 4
3-Phenylpyridine 4-Phenylpyridin¢ Pyridine 1,2,4-Triazolc Pyrazole 3-Phenylsydnon©
4.8 ,s 5.5 d 5.2 d 2.3 d 2.5 d -
2.0.10 - 3
1.6-10- 3 2.2.10-3 2.0.10- z 5.3-10-2 1.2.10-1
dA.10-3
3.0-10 - ' 5.8.10- 2 1.3.10-t > 0.15
IA. 10- 4 1.0-10 - 3
> 0.15
• Measured at pH - 5.6. 27°C. The uncenanties (S.E.) i:~ K, are less than 7%. b Calculated pH-independent dissociation constant assuming exdusly©binding of the unprotonated inhibitor to the catalytically active (EH) species of the enzyme. c Measured at 26.5(+0.5)°C in 0.1 M NaCI. d From Jencks and Regenstein [14].
concentrations of 2,4-dinitrophenylglucoside or pnitrophenylglucoside. In all cases the inhibition was competitive in nature and the K i values were found to be identical with those determined under first-order conditions. Results Over 25 nitrogen-containing heterocycles were tested as reversible inhibitors of sweet almond /]-glucosidase. Several of these are summarized in T a b l e L T h e binding of these c o m p o u n d s can be c o m p a r e d with that of sugars such D-glucose ( K i = 165 m M ) or D-arabinose (K! -- 40 raM) or substrates such as p-nitrophenyl-fl-Dglucopyranoside ( K s = 3.6 r a M ) * [4] or 2,4-dinitrophenyl-/~-D-glucopyranoside ( 2 , 4 - D N P G ) ( K s = 1.2 raM). A m o n g the more potent of these inhibitors is 4-phenylimidazole.
Specificity of 4-phenylimidazole inhibition 4-Phenylimidazole was tested as a competitive inhibitor with a variety of glycohydrolases at p H = 6.2. it was found to bind much more tightly to sweet almond #-glucosidase ( K i = 4.6(=1:0.1) F M ) than to the other
* As disc~_s__~,d_in Ref. 4, the K m for this substrate is equal to the thermodynamic ~ t i o n constant K,. T h © evidence for this is
that g m is essentially invaxiant over a pH range (4.4-7.5) where kcu varies by 6-fold. Similar behavior was found with 2A-DNPG.
229 enzymes tested: Jack Bean fi-N-aeetylhexosaminidase, K i = 4.1(:t:0.4) raM; yeast a-glucosidase, K i = 6.2 mm; Escherichia co!i ~-galactosidase, K, > 45 mM and E. coil or bovine liver ~-glueuronidase, K i > 150 raM. Preliminary results indicate that phenylimidazole binds more tightly (by 30- to 100~timcs) to the almond ~-glucosida~e than to the mananalian (guinea pig liver lysosomal or cytoplasmic) enzyme (Glew, R. and Hubbell, C., personal commumcation).
pH Dependence To investigate more fully the interaction of 4-phenylimidazole with the enzyme, the p H dependence of its binding was examined. The catalytic activity of the enzyme (kc, t/Km) with 2,4-DNPG shows the requirement of an unprotonated group with p K , (of the conjugat¢ acid, EH2)ffi 3.6 and a protonated group (EK) with p K , = 6.6. The data for 4-phenylimidazole is summarized in Fig. 1A and is most readily interpreted in terms of the unprotonated inhibitor (with conjugate acid PKa--6.08) binding to the catalytically active enzyme species with K~ffi 8.3(-1-0.2)-10 -7 M. The p H dependence of K~ for several other inhibitors (1,5-diazabicyclo[4.3.0]non-5-ene (DBN), N-methyl-l-deoxynojirimycin, imidazole, and the derivatives 2-methyl-, 4-methyl-, 4-trifluoromethyl- and 1,2-dimethylimidazole) also examined, indicates that here too it is the unprotonated species which binds to the enzyme. Although the data indicate that 4-phenylimidazole binds to E H and not to E it is difficult to rule out binding to the Ell 2 species. For this reason a more acidic derivative, 4-pentafluorophenylimidazole (pK~ = 4.91) was ex2*co ~oc.o 1too 12oo
'T
,e0
,0
~/
amined. The results are displayed in Fig. lB. Although the acidic limb indicates that there is no binding to the EH 2 species, there is an unusual feature ef a plateau at high pH. This suggests that the unprotonated inhibitor binds to both EH and E with a slightly (12-fold) higher affinity for the former species. Further evidence in s,]pport of 4-phenylimidazole inhibition being due to binding c~f the unproton~ted species comes from the rapid onset of inhibition. No lag in attaining the steady-state velocity for hydrolysis of 2,4-DNPG in the presence of inhibitor was observed at any pH. In a stopped-flow experiment, an enzyme solution (14/tM, p H = 7.3) was added to an equal volume of a mixture of substrate (0.26 mM) and phenylimidazole (20 ~M, p H = 7.3, 27 ° C). The linear steady-state velocity, consistent with a K i of 26 pM, was obtained 'immediately" (i.e., wathin 10 ms). At this p H only 6% of the inhibitor is present as phenylimidazolium. Thus, if inhibition is due to this species, an association rate constant of more than l0 s M - i • s -1 would be required in order to account for the rapid onset of inhibition. The binding of the unprotonated phenylimidazole is analogous to the situation with 1-deoxynojirimycin (1,5-dideoxy-5-amino-D-glucopyranose) for which it was concluded, from p H studies, that the uncharged species binds to the enzyme [4]. To test this conclusion further, the permanently cationic analog, N,N-dimethyl-l-deoxynojirimycin (N,N-Me2-1-dNM) was tested as an inhibitor of the enzyme. This compound is a much poorer inhibitor them either 1-dNM or N-Me-I-dNM. For example, at p H 6.2 the K i values (pM) are 29(+2), 57(+5) and 780(+40) for the unmethylated, monomethylated and dimethylated species, respectively. The pH dependencies of the K i values are consistent with the unprotonated 1-dNM binding exclusively to EH ( K i ~- 6.5 ~tM)[4], the unprotonated N - M e - l d N M binding to EH (Ki= 13(+2) /tM), but the cationic N,NMe2-1dNM binding exclusively to E ( K i = 300(+40) ttM).
Binding of substituted imidazoles
iic
|.o
xe
so
pH
70
oe
Fig. 1. pH dependence for inllJbition of sweet almond ~]-Slucosidasc by (A) 4-phcnylimJdazole (pK. ffi 6.08) or (B) 4-pcntafluoTophcnyfim-
idazole (pK, ~ 4.9). The K i values are the averagesof at least three dctcrminations (range indicated by error bars). The solid curves are the theoretical fits assuming (A) unprotonated 4-phenylimidazole binds only to the catalytically active enzyme species (EH) with a pH-independent dissociation constant of 0.83 pM and (B) 4-pentafluorophcnylimidazolebinds to EH with a limiting dissociation constant of 0.038 mM and to the deprotoaated enzyme, E. with a dissociation constant of 0.45 raM.
In order to explore further the factors responsible for the effective interaction of 4-phenylimidazole with flglucosidase, a series of substituted imidazole derivatives were examined as potential inhibitors. The results are presented in Fig. 2 where affinities for the enzyme are correlated with the inhibitor basicities. Although there is substantial scatter in the data, a couple of trends are apparent. Most of the data (points a-i) show a fair correlation (r ffi 0.94) between the logarithm of the binding constant (pK i -- - l o g Ki) and pKg. Indeed, this single parameter, PKa, accounts for over 87% ( r 2-0.874) of the variance in the data obtained with this structurally diverse series of derivatives. The slope of the correlation fine is 0.44( + 0.06). The derivatives which show a substantial negative deviation from this corre-
230 lation are the ones which are either alkylated on the nitrogen or possess a charged substituent. The two derivatives which show the most significant positive deviation are the ones with the most hydrophobic substituents in the 4-position. For these, factors in addition to basicity clearly are important in influencing their interaction with the enzyme.
120
90-
/
?,0-
Multiple inhibition studies
o
Sweet almond /3-glucosidase has a well-established hydrophobic binding site. Thu.~ the binding of a series of aliphatic primary amines shows a good correlation with the hydrophobic substituents parameter ~r [4] and axyl glucoside$ are particularly good substrates for the enzyme [15]. Since it is the uncharged substituted imidazole species which bind to the enzyme, it was of interest to discover whether imidazole binds to the hydrophobic site or whether, due to its hydrogen bonding ability, it binds to the sugar binding site. This was determined by the method of Yonetani and Tho0rell [16], where inhibition is examined in the simultaneous presence of a pair of competitive inhibitors. Under pseudo-first-order conditions the equation derived by Yonetani and Theorell becomes (2)
ilk + - { [tilt:, + [ii/Ki + [[llil/,,#(, K i } lit
where k + is the first-order rate constant in the presence of the inhibitors I and J and k is the first-order rate constant in the absence of any inhibitors ( = V ~ / K m). if a = 1, the binding of 1 and J is independent and if
6.5
p& 5.5 0
C
pKi 3.5
1.5
~
~
,"" ~
"~ 3
,
, 4
,." ~
,
• f
f ~
e
h ~ "~: "~
0
be
,
,
i
,
5
6
7
8
pKo
Fig. 2. Correlation of affmiti= of imidazolederivativeswith basicity. The ~ t e d gi values age based on the asstunptiofl ~ t it is the unprotonated apegigawhich binds to the catalyticallyagtivespeciesof p-~uc~idue. The conc~tion line (dope -0A4(±0.06~ a - 9 , r 0.935) is for the imidazolederivativescontaini~ substituents that axe undmrlged and not at position I. These are: (it) 4-CF3, (b) 4-(pnitrophenyl), (c) 4-penud]uo~ophenyl, (d) 4-(o-nitropbenyl), (¢) 4HOCHz, (0 2-phenyl, (8) H, (h) 4-CHs and (i) 2-CH3. The N.mbstituted dcl/vatives arc: (j) l-benzyl, (Ik)I-CH3 and (I) 1,2-d/mclhyl. The dmrl~ derivativesare (m) 4-+ HsN(CH 2)2 and (n) 4- - OaCCHz and the two whi~ show a silnificant pmitive deviation are the (o) 4-(p-fluo~ophenyl)and (p) 4-pheny!derivatives.
=
;
:
,
.
..
,0 -O.S
••
-
l
•
1
-
i
-
;
• O.S
:
: t.O
;
; ~ i}
;
i
-
i
-
.0 0 180
e.
B
, -1.S
O,
.•
,, -1.0
, 0.0
; 2.0
;
: 2.5
;
, 3.0
[;midozolej, mM
Fig. 3. Yonetani-Theorell plots showin8 the inhibition of ~-glucosidase.catalyzed hydrolysis of 2,4-dinitrophenyl-~S-glucoside (under first-order conditions) by imidazole at various fixed levels of (A) benzene ((4)) ncn:, (e) 8.3 m M ,~r (ll) 15 m; p H 6.2) or (B} I)-arabim~c ((e) none, (,,) 29 m M or tin) 49 raM; p H 5.6). The data
were fit to Eqn. 2 and yield a value of the interaction coefficient a - !.2(-I-0.3)for imidazole and benzene and a > 5 for imidazoleand arabinose a ffi ~ , the binding is mutually exclusive (if a < 1 the binding is synergistic and if 1 < a < ~ the presence of I hinders the binding of J). lmidazole (I) was therefore tested separately in the presenc~ of benzene, which presumably bh, ds to the hydrophobic binding site (Kj ffi 42 raM) or in the presence of D-arabinose, which binds at the sugar binding site about 4-times more tightly ( K j - - 4 0 mM) than does D-glucose. The Yonetani-Theorell plots are shown in Fig. 3. The intersetting pattern seen with various concentrations of imidazole at f~ed concentrations of benzene indicates that binding of these two inhibitors is essentially independent. The. nearly parallel pattern seen with imidazole and arabinose suggests that the presence of one of these inhibitors on the enzyme substantially, if r,ot completely, hinders the binding of the other. D
m
4-Phenylimidazole is one of the most potent inhibitors of sweet almond ~-glucosidase found so far. Its binding to the enzyme is stronger than that of most of the naturally o~-m'ring inhibitory alkaloids such as l deoxynojirimycin (I-dNM). As is the case with I - d N M , it appears to be the unprotonated species of phenylimidazole (or the other imidazoles) which is the effective
231 inhibitor. As pointed out by Legler [1,17] this is a typical situation for most glycohydrolases. Indeed, based on his results with the relatively poor binding cationic glucosyl derivatives of piperidine and inddazole, Legler [1] postulated that an active site carboxyl group protonates and stabilizes the binding o," basic inhibitors. The substantially poorer binding of N,N-dimethyl-l-deoxynojirimycin, compared to N-methyl-l-dNM or IdNM which we find, as well as the trend with the imidazole derivatives showing tighter binding as the basicity increases (Fig. 2), is consistent with this idea. (This contrasts with the tighter binding of the more acidic (protonated) phenols [4].) A significant structural difference between 1-dNM and 4-phenylimidazole is that the later contains no hydroxyl groups. It has been demonstrated that the affinities for fl-glucosidase of the polyhydroxylic nitrogen heterocycles corresponds to the sugar specificity of the enzyme [18,19]. Clearly, the hydroxyl groups contribute substantially to the interaction with the enzyme [4]. It is therefore noteworthy that phenylimidazole binds even more tightly to the enzyme than does either 1-dNM or &gluconolactam. One feature present in the former, but absent in the latter, inhibitors is the presence of the hydrophobic phenyl moiety. From both the enzymes' aglycone specificity, a~ seen in substrate binding [20] and inhibition studies [4], it is clear that the enzyme has a hydrophobic binding site. The results of the multiple inhibitor studies (e.g., Fig. 3) indicate that the phenyl moiety of 4--phenylimidazole binds to the hydrophobic binding site and imidazole binds predominantly to the sugar binding site. Thus, even in the absence of hydroxyl groups the inhibitor can interact with hydrogen bonding residues at the active site. Indeed, the catalytically active form of the enzyme (cont "aining a protonated and an anionic catalytic residue) can interact effectively with the unprotonated inhibilor presumably in a manner analogous to its interaction with the reactants in the transition state. This is illustrated in Scheme I. The independent binding of imidazole and benzene simplifies the analysis of the attribution of the binding energy of phenylimidazole.
E
÷
$
E
+
I o'+ o
Scheme I
~0
An interesting feature of the tight binding of the inhibitor is that there is little, if any, entropic advantage of combining the binding affinities of imidazole and benzene in the monomolecular species, ~phenylimidazole. This is readily seen from an analysis outlined by Jencks [21]. 4-Phenylimidazole (A-B) can be separated into a hydrogen bonding region (A) and a hydrophobic region (B) represented, respectively, by imidazole and benzene. The intrinsic binding energy of A is given by AGA = AG^B AG e where AG^a is the observed binding energy of phenylimidazole at 27°C (ffi RTln K i ffi 5.7311og(8 • 10 -7) ffi -34.7 Ll/mol) and AGs is the observed binding energy of benzene (ffi 5.7311og(4.2-10-2) --- - 7 . 9 Ll/mol). Thus, the intrinsic binding energy of the imidazole moiety of 4-phenylimidazole is - 26.8 ( = - 34.7 + 7.9) kJ/moi. Similarly, the intrinsic binding energy of B is given by AG s = A G A B - AG~ where AG~ is the observed binding energy of imJdazole (pK a -- 7.10) corrected to the estimated value if it had a pK a (ffi 6.08) equal to that of 4-phenylimidazole (Fig. 2). This value of AG~ ffi -23.4 kJ/mol is s]ightly more positive (by 2.5 kJ/mol) than the observed value for imidazole. This yields a value for the intrinsic binding energy of the phenyl moiety of 8G s = -34.7 + 23.4 -- - 11.3 Ll/mol. This is slightly more favorable than the observed value of AG e ~- - 7 . 9 kJ/mol. The observed binding energy of A-B is the sum of the intrinsic binding energies of A and B and the 'connection energy', dGS: -
AGAB = AGA "t- AG B + AG s
AG s is largely an entropy term representing the prob-
ability of binding that results from the connection of A and B. In the case of 4-phenylimidazole this term is small (approx. 3.3 L l / m o l - 2.7 e.u. if totally an entropy term). Similar results are obtained if one compares the binding of 2-phenylpyridine with benzene and pyridine. This suggests that any entropic advantage derived from connecting A and B is compensated for by some unfavorable factors such as desolvation. For example, the binding of benzene to a hydrophobic site on the enzyme is likely to be an entropically driven process [22]. If the phenyl moiety in 4-phenylimidazole is less hydrophobic than benzene there will be less entropy increase upon binding. Also, if imidazole is more extensively solvated in aqueous solution than is phenylimidazole, the entropically unfavorable binding of imidazole to th~ enzyme will be concomitant with an enuopically more favorable desolvation than in the case of phenylimidazole. Of course, it is also possible that in 4-phenylimidazole the benzene and imidazole moieties cannot optimally interact with their binding sites on the enzyme. Thus, even though 4-phenylimidazole can exploit, at least to some extent, the binding energies of its components to yield one of the most potent inhibitors of the enzyme it is possible that relatively minor street-
ura[ modili¢~tices can Rsuh in an ¢Tcn mote pwwerful "~lth~l~i~or. .¢dme~led-~ T~s r~m~...h was Rippor ted in pmr¢ by USPHS Osam G M M 0 7 0 . W e ~ ' ~ gra~etul to D r . G . L e g l ~ f o r g o n e r ously s u p p l y i ~ e S with v a r i o u s derivatlv~s o f 1-de-
14 Jr<~ncl~W.P, ~ m~. ~
oxyn~jirlm~ifl.
I L q ~ r , O. ( ] ~ 7 ) t ~
~ Rc¢~A,M, {t~,3) J. Cb¢~, So.:, 2t95~2200. 9 E.lUrl.C. a~d M~LckJnc-y,A,W. 1"193~}J. (~'hcllnL,Y~¢:,;$~-900. 10 B,t]lacuk'~, F,, C.~1~1. R. :$~ef'l~u~. J.D.G., C~¢kcr. D. ar~ 5i~otl, M , I , ~']973) J'. Chem, ~ ¢ , Pgttdn Tn~L~. 1. 241f.,2419, t l ~l~oth M.L. ~ ",,,']r.stcllc~O.M. {'tg~']'!l J]~oc~nn_J. 133, 8)-87. ]Z Gp~.~r, A.K, and C'~sM~"y,g,=t, (1~)~?) BJOc~unL ~ p i l y L A¢'~i 4.~, ~.-]~. ]3 LmU~vtm~w+ ~ T (]~ST~ ~uznu=. ^ N c ~ - L i u ~ t ~ p ~ m ~b d d ~ A m ~ y ~ r ~ . . m r ~ ~ m ~ - ] ~ ~ s~s, C.n~
~ d Appfi~ ~
Sg, 14~7-t4~4.
ll~r~ J, ~ S ) in H ~ l b ~ . of Biochc'm~ fm M~ecuLlW ~ , 3n~ ~dn. VgL I
Ck~elao~ C~I. 15 N~I~ g.L, ~n~ Rydon, H,N. 119~4) ~ e m .
$. $7, t-10,
Tulp~ A., HBi~mm+ If,O., Micd¢~, F.M_ ~md P ~ H,L, { l q ~ Nii~ 3~ 74-7% 3 0 . ~ N t . ~ , l l ~ c k K., F~s~o, D . Lcglm, G., D ¢ ~ K I . ~md
]135,~3-251. ]7 Lcg~, O, (1975) l~od1~m, B~pby¢ Acre $24, 94-]01. ] ~ 1Jd, H- and ~ B, ( 19733 Bioch~m. ~o1~-~.- R¢~ Commuv-
,t 'D~ie, M.p~ FJldey, H.P.-, Km'~ K~ Smu~-. KA,R, ~
]9 L,eg~. G. a e d l u l k ~ E. {1984) ~ .
( ] ~ 0 ) (~e~. Ik.r. 93, ~ - ~ 0 @ , t 6 ~ H- a~l Ftlj~ S. { t ~ ) J, ~
B~t,,
C/~m, 47, ~'~,67~7Z
Rcs_ 128. ,5[-7_L
2t J~¢k~, W.P. { t ~ l ) p¢o¢. Nat]..,~,~1..~,L USA 781 40,1~5¢,. 22 gau~ K_ and Web¢[. G, {I~B;S) g ~ r ' # 27. 3'/'9~'-~].
227
~sevier BBAPRO 33499
Inhibition of fl-glucosidase by imidazoles Y a w - K u e n Li a n d L a r r y D. B y e r s Department of Chemistry, Tulane University, New Orleans, LA (U.S.A.)
(Received15 June 1989) Key words: ~-Glacosidase;Phenylimidazole Over 25 nitrogen-containing heterocydes were tested as inhibitors of sweet almond fl-glucosidase (EC 3.2.1.21). Among the most potent of these are some imidazole derivatives. The pH dependence indicates that the unprotonated inhibitor hinds most tightly to the catalytically active species of the enzyme. This is analogous to the ~itoation with I-denxyno|irimyein where the permanently cationic ~pecies, N,N-dimethyl-l-deoxynojirimycin, binds at least two orders of magnitude less tightly to the enzyme than does the unprotonated l-deoxynojirimycin. The binding of imidazole derivatives show a general tendency of increasing affinity with increasing basicity (fl -- 0.4). One derivative which shows a significant positive deviation from this correlation ( - l o g K i vs. p K , ) is 4-phenylimidazole. 4-Phenylimidazole is one of the most potent reversiMe inhibitors of fi-gineosidase with a pH-independent Kt ffi 0.8 pM. it is also fairly specific for fl-glneosidase, binding at least three orders of magnitude less tightly to any of the other exoglycosidases tested. This inhihitor combines, in a mono-molmflar species, the binding affinities of benzene, which binds at the h y ~ aglyenne binding site, and imidazole, which binds at the sugar binding site of ~-glucosidase. The binding energy of 4-phenylimidazole can be attributed to the sum of the intrinsic binding energies of the phenyl and imidazole moieties. Thus, there is no significant entropic advantage of combining the component parts of phenylimidazele in a single species. This indicates that there is no significant uncompensated entropy loss upon binding of either benzene or imidazole to the enzyme. Nevertheless, the additivity of binding energy, even in tim absence of an entropic advantage, results in the most powedul known inhibitor of the enzyme.
Introduction Glycohydrolases are widesprcacl in nature and serve a variety of functions, lnlubitors of these enzymes are of interest for many reasons. They have served as useful tools in studying glycoprotein biosynthesis and oligosaccharide chain processing [1] and they are potentially useful as pesticides or as therapeutic agents in the treatment of some viral-induced diseases [2]. fl-Glucosidase inhibitors, in particular, are potentially useful in developing animal models for glucosyl ceramide lipidoses and have already been shown to be useful in differentiating the properties of normal and Gaucher's di .sease enzymes [3]. We have been interested in fl-glucosidase inhibitors not only as probes of active-site topography, but as tools for investigating the catalytic
Abbreviations: DBN. 1,5-diazabicyclo{4.3.0]non-5-en¢;1-dNM, 1-deoxynojirimycin (5-amino-l,5-dideoxy-D-ghcopyranose);DNI~3, dinitrophenyl-tB-D-glucopyranosid©;Mes, 4-morpholineethanesulfonate; Mops, 4-morpholinelpropanesulfonate. Correspondence: L.D. Byers,Departmentof Chemistry+,Tulane Unto versity, New Orleans, LA 70118, U.S.A.
mechanism of the enzyme. The most potent naturally occurring inhibitors of B-glucosidase are hetcrocycfic compounds containing at lea~t or, e rfittogen atom (e.g., the 5-amino-5-deoxyglucopyranose, nojirimycin; the 1,5-iminoalditol, 1-deoxynojirimycin, 1-dNM, and the indolizine alkaloid, catanospermine). The basic charactec of these compounds contributes to their affinity for the enzyme. Replacement of the nitrogen in 1-dNM for an oxygen reduces its affimty for sweet almond l~-glucosidase by nearly four orders of magnitude [4]. The pH dependence of the binding, however, suggests that it is the unprotonated form of the inhibitor which binds to fl-glucosidase [4]. The strong interaction of the uncharged inhibitor with the enzyme is a characteristic of most glycosidases [1]. It should be noted, however, that, at least in the case of the almond enzyme, this assignment is ambiguous because of a coincidence of pK, values [4]. Thus, it is possible that binding of protonated 1-dNM (PKa = 6.7) to the unprotonated and inactive enzyme could be much stronger than binding of the unprotonated 1-dNM 1o the protonated enzyme (pK, = 6.6). This study was undertaken to resolve this ambiguity and to examine how the basic properties of nitrogen-containing heterocycles facilitates their interaction with the enzyme.
0167-4838/89/$03.50 © 1989ElsevierSciencePublishers BN. (BiomedicalDivision)
228 TABLE !
Materials and Methods
Some ~8-glucosidaseinhibitors Most of the imidazoles and pyridine derivatives were obtained from Aldrich Chemical Co. 4-(p-Fluorophenyl)imidazole (m.p. 1 2 4 - 1 2 5 ° C , lit. 1 2 5 - 1 2 7 ° C ) and 4-trifluoromethylimidazole (m.p. 148-149 o C, lit. 150-151°C) were prepared as described by Bredereck et al. [5]. 4 - ( P e n t a f l u o r o p h e n y l ) i m i d a z o l e (m.p. 1 6 8 - 1 7 0 ° C , lit. 169-1710C) was synthesized photochemically as described by Kimoto and Fujii [6]. 4 - ( p Nitrophenyl)imidazole (m.p. 2 2 2 - 2 2 4 ° C , lit. 225°C) and 4-(o-nitrophenyl)imidazole (m.p. 144--145°C, lit. 146°C) were prepared by nitration of 4-phenylimidazole in H : S O 4 - H N O 3 (1 : 6) for 60 h at room temperature [7]. lmidazo[1,2-aJpyridine (perchlorate salt m.p. 2400C, fit. 241-243.5°C) was prepared as described by Roe [8]. 3-Phenylsydnone (m.p. 1 3 2 - 1 3 4 ° C , lit. 134-134.5° C) was prepared by the method described as Earl et al. [9]. The dinitrophenyl glucosides were prepared by reaction of the 2,3,4,6-tetra-O-acetyl-a-Dglucopyranosyl bromide (Sigma) with 2,4-dinitrophenoi (Sigma) or 3,4-dinitrophenol (Fiuka) by the methods of Ballardie et al. [10] and Sinnott and Viratelle [11], respectively. N-Methyl- and N,N.dimethyl-l-deoxynojir;mycin were generous gifts from G. Legler. All enzymes, including sweet almond /g-glucosidaze (type i, spec. act. ~ 20 units/rag), were obtained from Sigma Chemical Co. The B-glucosidase concentration was estimated by absorbance at 277 n m with E t~ = 7.25 [121. The buffer system usually consisted of 0.1 M NaCi, 0.1 m M E D T A and 0.01 M buffer: sodium acetate ( p H 3.5-5.6); Mes ( p H 5.6-7.0); Mops ( p H 7.0-7.9). Kinetics. The /~-glucosidase-~talyzed hydrolysis of dinitrophenyl glucosides was monitored spectrophotometrically. Rapid kinetics (of inhibitor binding) were followed on a Dionex Model 13-110 stopped-Row spectrophotometer interfaced with a Biomation 810 transient digitizer and a strip-chart recorder. In general, kinetics were followed on a Hewlett Packard Model 8452A Diode Array Spectrophotome!er. The reactions were initiated by addition of enzyme to a solution of substrate in buffer which ,had b,~en thermally equilibrated at 27 o C. The dissociation constant of enzymeinb./bitor complex, K,, was ~ ! c u l a t e d from the effect of inhibitor on the ratio of the two steady-state parameters
(g/V)+ - (g/Y)o(l+lll/g,)
(1)
( K / V ) + is the ratio of K m to Vmax in the presence of the inhibitor. The parameter K / V was determined under first-order conditions (IS] 4: Kin). The first-order rate constants, V/K, were evaluated b y non.linear regression [13]. In several cases the K i was also determined by following the reaction in the presence of inhibitor under initial velocity conditions at various
Inhibitor 2-(l-Naphthylmethyl)-I imidr~zolinc DBN 4-Phenylimidezole 4,~Dimethyfimi~fine
pK, 10.4 c 12.0 d 6.1 c 9.5 ¢
K i " (M)
Kitimb (M)
3.1-10- I 3.2.10-1 6.2-10 -6 3.7.10- 2
(4.~. 10-8) (1.2.10-7) (4.2.10- s)
8.3.10 -7
Imidazok
7.1 ¢j
9.3-10 -4
2.6. I0 -s
2-Aminopyridine imidazol¢[1,2-a]pyridine
6.9 d 6.9 ¢
3.2-10 - 3 8.1-10- s
3.5" | 0 - 4
aenzidmidazole 2.Phenylpyridine
5.4 d 4.5 d
1.3. I0- 3 1.1-10 - 3
7.1. I 0 - 4
3-Phenylpyridine 4-Phenylpyridin¢ Pyridine 1,2,4-Triazolc Pyrazole 3-Phenylsydnon©
4.8 ,s 5.5 d 5.2 d 2.3 d 2.5 d -
2.0.10 - 3
1.6-10- 3 2.2.10-3 2.0.10- z 5.3-10-2 1.2.10-1
dA.10-3
3.0-10 - ' 5.8.10- 2 1.3.10-t > 0.15
IA. 10- 4 1.0-10 - 3
> 0.15
• Measured at pH - 5.6. 27°C. The uncenanties (S.E.) i:~ K, are less than 7%. b Calculated pH-independent dissociation constant assuming exdusly©binding of the unprotonated inhibitor to the catalytically active (EH) species of the enzyme. c Measured at 26.5(+0.5)°C in 0.1 M NaCI. d From Jencks and Regenstein [14].
concentrations of 2,4-dinitrophenylglucoside or pnitrophenylglucoside. In all cases the inhibition was competitive in nature and the K i values were found to be identical with those determined under first-order conditions. Results Over 25 nitrogen-containing heterocycles were tested as reversible inhibitors of sweet almond /]-glucosidase. Several of these are summarized in T a b l e L T h e binding of these c o m p o u n d s can be c o m p a r e d with that of sugars such D-glucose ( K i = 165 m M ) or D-arabinose (K! -- 40 raM) or substrates such as p-nitrophenyl-fl-Dglucopyranoside ( K s = 3.6 r a M ) * [4] or 2,4-dinitrophenyl-/~-D-glucopyranoside ( 2 , 4 - D N P G ) ( K s = 1.2 raM). A m o n g the more potent of these inhibitors is 4-phenylimidazole.
Specificity of 4-phenylimidazole inhibition 4-Phenylimidazole was tested as a competitive inhibitor with a variety of glycohydrolases at p H = 6.2. it was found to bind much more tightly to sweet almond #-glucosidase ( K i = 4.6(=1:0.1) F M ) than to the other
* As disc~_s__~,d_in Ref. 4, the K m for this substrate is equal to the thermodynamic ~ t i o n constant K,. T h © evidence for this is
that g m is essentially invaxiant over a pH range (4.4-7.5) where kcu varies by 6-fold. Similar behavior was found with 2A-DNPG.
229 enzymes tested: Jack Bean fi-N-aeetylhexosaminidase, K i = 4.1(:t:0.4) raM; yeast a-glucosidase, K i = 6.2 mm; Escherichia co!i ~-galactosidase, K, > 45 mM and E. coil or bovine liver ~-glueuronidase, K i > 150 raM. Preliminary results indicate that phenylimidazole binds more tightly (by 30- to 100~timcs) to the almond ~-glucosida~e than to the mananalian (guinea pig liver lysosomal or cytoplasmic) enzyme (Glew, R. and Hubbell, C., personal commumcation).
pH Dependence To investigate more fully the interaction of 4-phenylimidazole with the enzyme, the p H dependence of its binding was examined. The catalytic activity of the enzyme (kc, t/Km) with 2,4-DNPG shows the requirement of an unprotonated group with p K , (of the conjugat¢ acid, EH2)ffi 3.6 and a protonated group (EK) with p K , = 6.6. The data for 4-phenylimidazole is summarized in Fig. 1A and is most readily interpreted in terms of the unprotonated inhibitor (with conjugate acid PKa--6.08) binding to the catalytically active enzyme species with K~ffi 8.3(-1-0.2)-10 -7 M. The p H dependence of K~ for several other inhibitors (1,5-diazabicyclo[4.3.0]non-5-ene (DBN), N-methyl-l-deoxynojirimycin, imidazole, and the derivatives 2-methyl-, 4-methyl-, 4-trifluoromethyl- and 1,2-dimethylimidazole) also examined, indicates that here too it is the unprotonated species which binds to the enzyme. Although the data indicate that 4-phenylimidazole binds to E H and not to E it is difficult to rule out binding to the Ell 2 species. For this reason a more acidic derivative, 4-pentafluorophenylimidazole (pK~ = 4.91) was ex2*co ~oc.o 1too 12oo
'T
,e0
,0
~/
amined. The results are displayed in Fig. lB. Although the acidic limb indicates that there is no binding to the EH 2 species, there is an unusual feature ef a plateau at high pH. This suggests that the unprotonated inhibitor binds to both EH and E with a slightly (12-fold) higher affinity for the former species. Further evidence in s,]pport of 4-phenylimidazole inhibition being due to binding c~f the unproton~ted species comes from the rapid onset of inhibition. No lag in attaining the steady-state velocity for hydrolysis of 2,4-DNPG in the presence of inhibitor was observed at any pH. In a stopped-flow experiment, an enzyme solution (14/tM, p H = 7.3) was added to an equal volume of a mixture of substrate (0.26 mM) and phenylimidazole (20 ~M, p H = 7.3, 27 ° C). The linear steady-state velocity, consistent with a K i of 26 pM, was obtained 'immediately" (i.e., wathin 10 ms). At this p H only 6% of the inhibitor is present as phenylimidazolium. Thus, if inhibition is due to this species, an association rate constant of more than l0 s M - i • s -1 would be required in order to account for the rapid onset of inhibition. The binding of the unprotonated phenylimidazole is analogous to the situation with 1-deoxynojirimycin (1,5-dideoxy-5-amino-D-glucopyranose) for which it was concluded, from p H studies, that the uncharged species binds to the enzyme [4]. To test this conclusion further, the permanently cationic analog, N,N-dimethyl-l-deoxynojirimycin (N,N-Me2-1-dNM) was tested as an inhibitor of the enzyme. This compound is a much poorer inhibitor them either 1-dNM or N-Me-I-dNM. For example, at p H 6.2 the K i values (pM) are 29(+2), 57(+5) and 780(+40) for the unmethylated, monomethylated and dimethylated species, respectively. The pH dependencies of the K i values are consistent with the unprotonated 1-dNM binding exclusively to EH ( K i ~- 6.5 ~tM)[4], the unprotonated N - M e - l d N M binding to EH (Ki= 13(+2) /tM), but the cationic N,NMe2-1dNM binding exclusively to E ( K i = 300(+40) ttM).
Binding of substituted imidazoles
iic
|.o
xe
so
pH
70
oe
Fig. 1. pH dependence for inllJbition of sweet almond ~]-Slucosidasc by (A) 4-phcnylimJdazole (pK. ffi 6.08) or (B) 4-pcntafluoTophcnyfim-
idazole (pK, ~ 4.9). The K i values are the averagesof at least three dctcrminations (range indicated by error bars). The solid curves are the theoretical fits assuming (A) unprotonated 4-phenylimidazole binds only to the catalytically active enzyme species (EH) with a pH-independent dissociation constant of 0.83 pM and (B) 4-pentafluorophcnylimidazolebinds to EH with a limiting dissociation constant of 0.038 mM and to the deprotoaated enzyme, E. with a dissociation constant of 0.45 raM.
In order to explore further the factors responsible for the effective interaction of 4-phenylimidazole with flglucosidase, a series of substituted imidazole derivatives were examined as potential inhibitors. The results are presented in Fig. 2 where affinities for the enzyme are correlated with the inhibitor basicities. Although there is substantial scatter in the data, a couple of trends are apparent. Most of the data (points a-i) show a fair correlation (r ffi 0.94) between the logarithm of the binding constant (pK i -- - l o g Ki) and pKg. Indeed, this single parameter, PKa, accounts for over 87% ( r 2-0.874) of the variance in the data obtained with this structurally diverse series of derivatives. The slope of the correlation fine is 0.44( + 0.06). The derivatives which show a substantial negative deviation from this corre-
230 lation are the ones which are either alkylated on the nitrogen or possess a charged substituent. The two derivatives which show the most significant positive deviation are the ones with the most hydrophobic substituents in the 4-position. For these, factors in addition to basicity clearly are important in influencing their interaction with the enzyme.
120
90-
/
?,0-
Multiple inhibition studies
o
Sweet almond /3-glucosidase has a well-established hydrophobic binding site. Thu.~ the binding of a series of aliphatic primary amines shows a good correlation with the hydrophobic substituents parameter ~r [4] and axyl glucoside$ are particularly good substrates for the enzyme [15]. Since it is the uncharged substituted imidazole species which bind to the enzyme, it was of interest to discover whether imidazole binds to the hydrophobic site or whether, due to its hydrogen bonding ability, it binds to the sugar binding site. This was determined by the method of Yonetani and Tho0rell [16], where inhibition is examined in the simultaneous presence of a pair of competitive inhibitors. Under pseudo-first-order conditions the equation derived by Yonetani and Theorell becomes (2)
ilk + - { [tilt:, + [ii/Ki + [[llil/,,#(, K i } lit
where k + is the first-order rate constant in the presence of the inhibitors I and J and k is the first-order rate constant in the absence of any inhibitors ( = V ~ / K m). if a = 1, the binding of 1 and J is independent and if
6.5
p& 5.5 0
C
pKi 3.5
1.5
~
~
,"" ~
"~ 3
,
, 4
,." ~
,
• f
f ~
e
h ~ "~: "~
0
be
,
,
i
,
5
6
7
8
pKo
Fig. 2. Correlation of affmiti= of imidazolederivativeswith basicity. The ~ t e d gi values age based on the asstunptiofl ~ t it is the unprotonated apegigawhich binds to the catalyticallyagtivespeciesof p-~uc~idue. The conc~tion line (dope -0A4(±0.06~ a - 9 , r 0.935) is for the imidazolederivativescontaini~ substituents that axe undmrlged and not at position I. These are: (it) 4-CF3, (b) 4-(pnitrophenyl), (c) 4-penud]uo~ophenyl, (d) 4-(o-nitropbenyl), (¢) 4HOCHz, (0 2-phenyl, (8) H, (h) 4-CHs and (i) 2-CH3. The N.mbstituted dcl/vatives arc: (j) l-benzyl, (Ik)I-CH3 and (I) 1,2-d/mclhyl. The dmrl~ derivativesare (m) 4-+ HsN(CH 2)2 and (n) 4- - OaCCHz and the two whi~ show a silnificant pmitive deviation are the (o) 4-(p-fluo~ophenyl)and (p) 4-pheny!derivatives.
=
;
:
,
.
..
,0 -O.S
••
-
l
•
1
-
i
-
;
• O.S
:
: t.O
;
; ~ i}
;
i
-
i
-
.0 0 180
e.
B
, -1.S
O,
.•
,, -1.0
, 0.0
; 2.0
;
: 2.5
;
, 3.0
[;midozolej, mM
Fig. 3. Yonetani-Theorell plots showin8 the inhibition of ~-glucosidase.catalyzed hydrolysis of 2,4-dinitrophenyl-~S-glucoside (under first-order conditions) by imidazole at various fixed levels of (A) benzene ((4)) ncn:, (e) 8.3 m M ,~r (ll) 15 m; p H 6.2) or (B} I)-arabim~c ((e) none, (,,) 29 m M or tin) 49 raM; p H 5.6). The data
were fit to Eqn. 2 and yield a value of the interaction coefficient a - !.2(-I-0.3)for imidazole and benzene and a > 5 for imidazoleand arabinose a ffi ~ , the binding is mutually exclusive (if a < 1 the binding is synergistic and if 1 < a < ~ the presence of I hinders the binding of J). lmidazole (I) was therefore tested separately in the presenc~ of benzene, which presumably bh, ds to the hydrophobic binding site (Kj ffi 42 raM) or in the presence of D-arabinose, which binds at the sugar binding site about 4-times more tightly ( K j - - 4 0 mM) than does D-glucose. The Yonetani-Theorell plots are shown in Fig. 3. The intersetting pattern seen with various concentrations of imidazole at f~ed concentrations of benzene indicates that binding of these two inhibitors is essentially independent. The. nearly parallel pattern seen with imidazole and arabinose suggests that the presence of one of these inhibitors on the enzyme substantially, if r,ot completely, hinders the binding of the other. D
m
4-Phenylimidazole is one of the most potent inhibitors of sweet almond ~-glucosidase found so far. Its binding to the enzyme is stronger than that of most of the naturally o~-m'ring inhibitory alkaloids such as l deoxynojirimycin (I-dNM). As is the case with I - d N M , it appears to be the unprotonated species of phenylimidazole (or the other imidazoles) which is the effective
231 inhibitor. As pointed out by Legler [1,17] this is a typical situation for most glycohydrolases. Indeed, based on his results with the relatively poor binding cationic glucosyl derivatives of piperidine and inddazole, Legler [1] postulated that an active site carboxyl group protonates and stabilizes the binding o," basic inhibitors. The substantially poorer binding of N,N-dimethyl-l-deoxynojirimycin, compared to N-methyl-l-dNM or IdNM which we find, as well as the trend with the imidazole derivatives showing tighter binding as the basicity increases (Fig. 2), is consistent with this idea. (This contrasts with the tighter binding of the more acidic (protonated) phenols [4].) A significant structural difference between 1-dNM and 4-phenylimidazole is that the later contains no hydroxyl groups. It has been demonstrated that the affinities for fl-glucosidase of the polyhydroxylic nitrogen heterocycles corresponds to the sugar specificity of the enzyme [18,19]. Clearly, the hydroxyl groups contribute substantially to the interaction with the enzyme [4]. It is therefore noteworthy that phenylimidazole binds even more tightly to the enzyme than does either 1-dNM or &gluconolactam. One feature present in the former, but absent in the latter, inhibitors is the presence of the hydrophobic phenyl moiety. From both the enzymes' aglycone specificity, a~ seen in substrate binding [20] and inhibition studies [4], it is clear that the enzyme has a hydrophobic binding site. The results of the multiple inhibitor studies (e.g., Fig. 3) indicate that the phenyl moiety of 4--phenylimidazole binds to the hydrophobic binding site and imidazole binds predominantly to the sugar binding site. Thus, even in the absence of hydroxyl groups the inhibitor can interact with hydrogen bonding residues at the active site. Indeed, the catalytically active form of the enzyme (cont "aining a protonated and an anionic catalytic residue) can interact effectively with the unprotonated inhibilor presumably in a manner analogous to its interaction with the reactants in the transition state. This is illustrated in Scheme I. The independent binding of imidazole and benzene simplifies the analysis of the attribution of the binding energy of phenylimidazole.
E
÷
$
E
+
I o'+ o
Scheme I
~0
An interesting feature of the tight binding of the inhibitor is that there is little, if any, entropic advantage of combining the binding affinities of imidazole and benzene in the monomolecular species, ~phenylimidazole. This is readily seen from an analysis outlined by Jencks [21]. 4-Phenylimidazole (A-B) can be separated into a hydrogen bonding region (A) and a hydrophobic region (B) represented, respectively, by imidazole and benzene. The intrinsic binding energy of A is given by AGA = AG^B AG e where AG^a is the observed binding energy of phenylimidazole at 27°C (ffi RTln K i ffi 5.7311og(8 • 10 -7) ffi -34.7 Ll/mol) and AGs is the observed binding energy of benzene (ffi 5.7311og(4.2-10-2) --- - 7 . 9 Ll/mol). Thus, the intrinsic binding energy of the imidazole moiety of 4-phenylimidazole is - 26.8 ( = - 34.7 + 7.9) kJ/moi. Similarly, the intrinsic binding energy of B is given by AG s = A G A B - AG~ where AG~ is the observed binding energy of imJdazole (pK a -- 7.10) corrected to the estimated value if it had a pK a (ffi 6.08) equal to that of 4-phenylimidazole (Fig. 2). This value of AG~ ffi -23.4 kJ/mol is s]ightly more positive (by 2.5 kJ/mol) than the observed value for imidazole. This yields a value for the intrinsic binding energy of the phenyl moiety of 8G s = -34.7 + 23.4 -- - 11.3 Ll/mol. This is slightly more favorable than the observed value of AG e ~- - 7 . 9 kJ/mol. The observed binding energy of A-B is the sum of the intrinsic binding energies of A and B and the 'connection energy', dGS: -
AGAB = AGA "t- AG B + AG s
AG s is largely an entropy term representing the prob-
ability of binding that results from the connection of A and B. In the case of 4-phenylimidazole this term is small (approx. 3.3 L l / m o l - 2.7 e.u. if totally an entropy term). Similar results are obtained if one compares the binding of 2-phenylpyridine with benzene and pyridine. This suggests that any entropic advantage derived from connecting A and B is compensated for by some unfavorable factors such as desolvation. For example, the binding of benzene to a hydrophobic site on the enzyme is likely to be an entropically driven process [22]. If the phenyl moiety in 4-phenylimidazole is less hydrophobic than benzene there will be less entropy increase upon binding. Also, if imidazole is more extensively solvated in aqueous solution than is phenylimidazole, the entropically unfavorable binding of imidazole to th~ enzyme will be concomitant with an enuopically more favorable desolvation than in the case of phenylimidazole. Of course, it is also possible that in 4-phenylimidazole the benzene and imidazole moieties cannot optimally interact with their binding sites on the enzyme. Thus, even though 4-phenylimidazole can exploit, at least to some extent, the binding energies of its components to yield one of the most potent inhibitors of the enzyme it is possible that relatively minor street-
ura[ modili¢~tices can Rsuh in an ¢Tcn mote pwwerful "~lth~l~i~or. .¢dme~led-~ T~s r~m~...h was Rippor ted in pmr¢ by USPHS Osam G M M 0 7 0 . W e ~ ' ~ gra~etul to D r . G . L e g l ~ f o r g o n e r ously s u p p l y i ~ e S with v a r i o u s derivatlv~s o f 1-de-
14 Jr<~ncl~W.P, ~ m~. ~
oxyn~jirlm~ifl.
I L q ~ r , O. ( ] ~ 7 ) t ~
~ Rc¢~A,M, {t~,3) J. Cb¢~, So.:, 2t95~2200. 9 E.lUrl.C. a~d M~LckJnc-y,A,W. 1"193~}J. (~'hcllnL,Y~¢:,;$~-900. 10 B,t]lacuk'~, F,, C.~1~1. R. :$~ef'l~u~. J.D.G., C~¢kcr. D. ar~ 5i~otl, M , I , ~']973) J'. Chem, ~ ¢ , Pgttdn Tn~L~. 1. 241f.,2419, t l ~l~oth M.L. ~ ",,,']r.stcllc~O.M. {'tg~']'!l J]~oc~nn_J. 133, 8)-87. ]Z Gp~.~r, A.K, and C'~sM~"y,g,=t, (1~)~?) BJOc~unL ~ p i l y L A¢'~i 4.~, ~.-]~. ]3 LmU~vtm~w+ ~ T (]~ST~ ~uznu=. ^ N c ~ - L i u ~ t ~ p ~ m ~b d d ~ A m ~ y ~ r ~ . . m r ~ ~ m ~ - ] ~ ~ s~s, C.n~
~ d Appfi~ ~
Sg, 14~7-t4~4.
ll~r~ J, ~ S ) in H ~ l b ~ . of Biochc'm~ fm M~ecuLlW ~ , 3n~ ~dn. VgL I
Ck~elao~ C~I. 15 N~I~ g.L, ~n~ Rydon, H,N. 119~4) ~ e m .
$. $7, t-10,
Tulp~ A., HBi~mm+ If,O., Micd¢~, F.M_ ~md P ~ H,L, { l q ~ Nii~ 3~ 74-7% 3 0 . ~ N t . ~ , l l ~ c k K., F~s~o, D . Lcglm, G., D ¢ ~ K I . ~md
]135,~3-251. ]7 Lcg~, O, (1975) l~od1~m, B~pby¢ Acre $24, 94-]01. ] ~ 1Jd, H- and ~ B, ( 19733 Bioch~m. ~o1~-~.- R¢~ Commuv-
,t 'D~ie, M.p~ FJldey, H.P.-, Km'~ K~ Smu~-. KA,R, ~
]9 L,eg~. G. a e d l u l k ~ E. {1984) ~ .
( ] ~ 0 ) (~e~. Ik.r. 93, ~ - ~ 0 @ , t 6 ~ H- a~l Ftlj~ S. { t ~ ) J, ~
B~t,,
C/~m, 47, ~'~,67~7Z
Rcs_ 128. ,5[-7_L
2t J~¢k~, W.P. { t ~ l ) p¢o¢. Nat]..,~,~1..~,L USA 781 40,1~5¢,. 22 gau~ K_ and Web¢[. G, {I~B;S) g ~ r ' # 27. 3'/'9~'-~].