Journal of Medicinal Chemistry, 1 9 W , Val. i4,No. 4 275
CHLOROQUINE ANALOGS
Mechanism of Antimalarial Activity of Chloroquine Analogs from Quantitative Structure-Activity Studies. Free Energy Related Model1"sb GEORGEE. BASS,DONNA R. HUDSON, JANE E. PARKER, AND WILLIAM P. PURCELL* Department of Molecular and Quantum Biology, College of Pharmacy, University of Tennessee Medical Units, Memphis, Tennessee 58103 Received August 18, 1970 The antimalarial activities of chloroquine and 32 of its analogs have been subjected to regression analyses using a free energy related structure-activity model designed in part t o test the DNA intercalation mechanism of action proposed by O'Brien and Hahn. Molecular parameters included in the analyses are the OctOH-HzO partition coefficients and the charges on the 7 substituent of the quinoline ring and the two N of the 4-diamino side chain. The partition coefficients were calcd from substituent values in the literature and the charges were obtained by combining Huckel and Del Re hIO charge distributions. The statistical quantities, RZ,F (overall), and explained variance, were calcd for each regression analysis to assess goodness of fit. Under restrictions consistent with the O'Brien-Hahn model, a reasonably high level of structure-activity correlation can be obtained. The importance of the charge and size of the 7 substituent and the separation between the two nitrogens of the 4diamino side chain was verified. The results suggest that the hydrophobic or steric properties of the terminal amine group of the 4 side chain should be considered along with the charge on N. No high level of correlation was found when structural variations were allowed at more than one substituent position on the quinoline ring. This suggests that the role t o be played by substituents a t one ring position is influenced appreciably by the nature of substituents at other ring positions.
Those interested in drug design are beginning to turn their attent'ion more and more to the development and utiliaat'ion of quantitative structure-activity relat'ionships in order to gain insight into the modes of action of drugs. The availability of a large volume of antimalarial activity data on extensive series of molecules spanning several general classes of potential antimalarial agents makes this a very attract'ive area for cont'inuing and expanding these efforts. The series of chloroquine analogs is particularly well suited to this purpose in view of the existence of a proposed detailed mode of act'ion for these molecules as antimalarials. One may use the quantitative structure-act'ivity studies, on the one hand, to (%est"the proposed model, and, on the other, assuming the model to be basically correct, to attempt to sharpen or modify the details of the model. O'Brien and Hahnzahave offered a model to account for t'he antimalarial activity of chloroquine and its conjeners (Figure 1). I n particular, they suggested that: (1) these compounds exert their antimalarial effect by intercalation with the parasite DKA, and that the act'ivity of a given compound depends on the stabilit'y of it's complex with DNA; (2) high activity requires an electronegative group attached to position 7 of the quinoline ring; (3) the diamino side chain attached to the quinoline ring at position 4 bridges t,he two DNA strands by electrostatic interactions between the diamino nitrogens and t'he DXA phosphate groups; and (4)subst'itut'ionof any other groups at any other ring position, except posi(1) (a) This research is being supported by the U. S. Army Medical Research and Development Command (DA-49-193-MD-2779), the Cotton Producers Institute, the National Science Foundation (GB-7383), and a grant from Eli Lilly and Company. This paper is Contribution No. 859 from the Army Research Program on Malaria. Computer facilities were provided through Grant HE-09495 from the National Institutes of Health. (b) Presented in part a t the Research Symposium on Complexes of Biologically Active Substances with Nucleic Acids and Their Modes of Action, March 16-19, 1970, Department of Molecular Biology, Walter Reed Army Institute of Research, Washington, D.C. (2) (a) R. L. O'Brien and F. E. Hahn, Anhmicmb. Ag. Chemolhar., 315 (1965). (b) R. L. O'Brien, J. L. Allison, and F. E. Hahn, Biochim. Biophys. Acta, U S , 622 (1966).
tion 6 which is almost equivalent to position 7 for this model, would be expected to alter binding to DNA and thus diminish activity. To support this model, considerable evidence derived from in vitro intercalation studieslZbin vivo bactericidal studies, and antimalarial activity data selectedzafrom the literature4sbwas offered. The first of these studies demonstrated that chloroquine can intercalate with DSA, and the second, that, at least for the strain of Bacillus meyaterium studied, chloroquine inhibits DXA (and at higher concentrations, RNA) synthesis. It is primarily from the third study, the antimalarial activities of chloroquine and its congeners, that O'Brien and Hahn deduced the roles to be played by the substituents attached to position 7 of the ring and the diamino side chain attached to position 4 (here, in vitro studies6 on the binding of aliphatic diamines to D S A were also helpful). Their observations, based on general trends in the activity data, were qualitative in nature. The results presented here represent an attempt to evaluate quantitatively these trends and to test some of the features of the model. The structure-activity equation designed for this purpose is discussed in the next section. Structure-Activity Equation.-It has been assumed that the interaction of the drug D (chloroquine, etc.) with the biological substrate S (DSA) may be represented as K
Iz
D+SZD-S+P
where K is the equilibrium constant for the formation of the drug-substrate complex and k is the rate constant (3) F. E. Hahn, R . L. O'Brien, J. Ciak, J. L. Allison, and J. G. Olenick, Mil. Med., 131, 1071 (1966). (4) G. R . Coatney, W.C. Cooper, N . B. Eddy, and J. Greenberg, "Survey of Antimalarial Agents. Chemotherapy of Plasmodium gallinaceum Infections: Toxicity; Correlation of Structure and Action," Public Monograph NO. 9, 1953, pp 64-74. ( 5 ) F. Y . Wiselogle, Ed., "A Survey of Antimalarial Drugs 1941-1945," Vol. 11, Part XI, J. W. Edwards, Ann Arbor, Mich., 1946, pp 1112-1164. (6) H. R. Mahler and B. D. Mehrotra, Biochim. Biophys. Acta, 68, 211 (1963).
276 Journal of Medicinal Chemistry, 1971, VoE. 14, No. 4
PURCELL, et a!.
of point charges, the stabilization energy, AE, can be evaluated as
\
\
where Q t D is the charge at point (atom) i of the drug, Qjs is the charge at point (atom) j of the substrate, and rt3 is the distance separating these two points. Substitution of eq 7 into eq 6 leads t o
dfl = ka[Do][S]exp[-(n. dt
-
T O ) ~ / P ]
X
Figure 1 .-Proposed structure of DNA-chloroquine complex consistent with model of O'Brien and Hahn.28
exp[-- 2t2.3QzDQ3S/RTri:,l X exp[-(PAV - TAS)/RT] (8)
for the conversion of this complex into product P. The rate of formation of product can be written as
Assuming all terms on the right side of eq 8 to be independent of [PIand t, integration from zero time and product to some particular values P* and t* gives P*
A D 1
(4)
where A is less than unity. Hansch and coworkers' have related the factor A to the OctOH-HzO partition coefficient of the drug, using the relationship
d
=
aexp[-(a - T " ) ~ / P ]
(5)
where ?T is the logarithm of the partition coefficient, and a , p , and ro are constants. Substitution of eq 4 and 5 into eq 3 and expansion of AG as A E PAV TAS gives
+
dP [1 dt
=
log-
1 - 2.303 - -~ T2 [Do*I P
4.606 + -__ P
? T I T - -
2.303
="2
-
P
ZfQtD
Here, [ D ] represents the concentration of the drug in the vicinity of the substrate and can be assumed t o be quite different from the administered concentration or dose, [Do]. I t is this latter quantity which is usually tabulated by the experimentalist while the former remains essentially unknown. To circumvent this difficulty, it is usually assumed that =
where [Do*] is the drug dosage required to produce [P*] in time t*. Thus, [Do*] corresponds to such commonly reported quantities as LDbo, ED,,, TIC4, etc. Taking the log of each side and rearranging one may obtain
(w) + log (ka[S]t*/P*) RTyij
dm= k[D][S] exp(-AGIRT) dt
[Dl
ka[Do*][S]exp[-(r - ? T " ) ~ / PX] exp[- ~ t ~ 3 Q ~ D Q ~ S :XR T ~ t l I exp[-(PAV - TAX)] (9)
For the equilibrium constant K one has, assuming a steady-state concentration of complex,
where AG is the free energy of formation of D-S from D and S, R is the gas constant, and T is the absolute temperature. Solving eq 2 for [D-SI and substituting in eq 1 gives
=
2 303(PA+- - TAS)
(10)
This equation is to be used t o study variations in the activities of a series of very similar molecules assumed t o be active cia the same mechanism. For this reason, all quantities in eq 10 which are assumed not to vary from one member of the series to the next can be treated as simple constants. The last two terms of the equation and the parameters /3 and a ' can be so treated. Further, if one assumes that each of the drug molecules is oriented relative t o the substrate in exactly the same manner, one can also treat the quantity (Z,Qjs/RTrf,) as a constant for a given value of the index i. Since O'Brien and Hahn developed their model in terms of electrostatic interactions involving the substituent attached to position 7 and the two nitrogens of the diamino side chain a t position 4, interactions involving all other point charges of the drug molecule will be considered constant also. Under these conditions, eq 10 can be rewritten as
ka[Do][S]exp[-(a - ~ ~ ) ~ / P l e x p [ - A b E / RXT l exp[-(PAT/ - TAS)/RT]
(6)
If one assumes that the drug-substrate complex is stabilized primarily by electrostatic interactions and, further, that these can be approximated using the simple coulomb potential with D and S represented as systems (7) C . Hansch and T. Fujita, J . Arne?. C h e n . S o c . , 86, 1616 (1964).
e&Nz+
f f
(11)
where Q, is the charge on the substituent attached to ring position 7, Q N is~ the charge on the diamino S a t ring position 4 and Q N z + is the charge on the terminal iY of the diamino side chain (assumed to be protonated). The coefficients and constant have been written simply as a, b,. . . f.
CHLOROQUINE ANALOGS
Journal of Medicinal Chemistry, i Q W , Vol. 14, No. 4 U7
From other sources, one may obtain values for [DO*], series. Series I is composed of compounds which differ from and Q N ~ for + each of the drugs in the series. chloroquine only in the substituent attached t o position Then, using the techniques of regression analysis, the 7 of the ring. coefficientsa, b,. . .,f can be evaluated, their significance tested, and t'he abilit'y of the act'ivity equation to exCH3 plain the observed variation in t'he biological data meaI sured (these aspects are discussed later). "CH(CHJ,r;JH(C&), The parameter K is usually interpreted in t'erms of I transport and/or hydrophobic bonding properties of the m ~ l e c u l e . ' ~Accordingly, ~ a high degree of act'ivity corX relation with R would imply that ability of the drug to X = substituent group move from t'he point of application (crossing phase boundaries such as membranes, et'c.) and possibly int'erSeries I1 is composed of compounds which differ from act wit,h the substrate via hydrophobic bonding is imchloroquine in the substituents attached to position 7 portant. Activit'y dependence on the parameters Qx, and one other position but not at position 4. QN1, and Q N Z + should provide a test of several features of the model proposed by O'Brien and Hahn. FH3 It must be recognized, however, that steric requireI NH~cH,),~;JH(c,H,)~ ments also are implied by the intercalation No such factors have been incorporated in the activity equation (eq 11). Similarly, no explicit allowance has Y been made for effects due to substituents attached at ring positions other than 4 and 7. These factors must X, Y = substituent groups be taken into consideration in the interpretation of t'he results of the regression analyses. Series 111is composed of compounds which differ from Calculation Procedures.-The antimalarial act'ivity chloroquine only in the diamino side chain attached at data used in this study were select'ed from that list'ed by position 4. The molecules included in each of these O'Brien and Hahne2&The quantity l/[Do*] (eq 11) series and corresponding parameter values are given in was taken bo be O.l/METD where M E T D (the miniTables 1-111. mum effective therapeutic dose) is the dose required to R reduce the parasitemia in White Rock chicks infected I with Plasmodium gallinaceum to less than 25% of cont r o l ~ .The ~ ~ experiment'al ~ error inherent in these activc1 ity data will, of course, contribute t'o uncertainty in the results. For example, since the NETD were deterR = diamino side chain, mined by administering drug dosages in a geometric terminal nitrogen progression with base 24t5a drug with an NIETD of protonated 0.050 is not' necessarily appreciably more or less active than one wit'h an M E T D of 0.025 or 0.100, re~pectively.~ TABLE I The charges, Qx, QNI, and Q N ~ +were , obt'ained by ANTIMALARIAL ACTIVITIESAND PARAMETER combining the result's of Huckel K electron calculationsg VALUESFOR SERIESI and Del Re u electron calculat,ions.lO For both calculations, parameter values recommended to reproduce dipole moments were u ~ e d . ~ ,(Del ~ l Re parameters used for Br are those reported by RassI2). Computer programs for the Del Re and Huckel calculaX tions were written by G. E. Bass and K. Sundarum, respectively. Values for the parameter R were calculated Activas the sum of substituent constants, K , gleaned from the X itya 2 Qx QNL &N2* publications of Hansch and c o w o r k e r ~ . ~ ~ 8 ~ ~ 3 - ~ ~ c1 100 4.20 -0,124 -0,264 0.525 In the analyses reported here, only compounds seI 67 4.71 -0.101 -0.264 0.525 lected from the series listed by O'Brien and Hahn have Br 50 4.35 -0.109 -0,264 0.525 been considered. These can be divided into three F 50 3.63 -0.176 -0.264 0.525 R, Qx, Q N 1 ,
&
a
J3)
(8) T. Fujita, J. Iwasa, and C. Hansch, J . Amer. C h e m . Soc.. 86, 5175 (1964). (9) B. Pullman and A. Pullman, "Quantum Biochemistry," Interscience Publishers, h'ew York, N. Y.,1963, p 67. (10) G . Del Re, J. C h e n . Soc., 4031 (1958). (11) H . Berthod, C. Giessner-Prettre, and A. Pullman, Theor. C h i n . Acta, 8 , 212 (1967). (12) G. E . Bass, Ph.D. Thesis, Vanderbilt University, Nashville, Tenn., 1970, p 79. (13) C. Hansch and F. Helmer, J. Polym. Sei., 6 , 3295 (1968). (14) C. Hansoh, J. E. Quinlan, and G . L. Lawrence, J . Org. Chem., 88, 347 (1968). (15) F. Helmer, K. Kiehs, and C. Hansch, Biochemistry, 7 , 2858 (1968). (16) J. Iwasa. T. Fujita, and C. Hansch, J . M e d . Chem., 8 , 150 (1965). (17) A. Leo, C. Hansoh, and C. Church, ibid., 14, 766 (1969).
CF3 50 4.65 -0,042 -0.264 0.525 OCHa 14 3.47 -0,071 -0.264 0.525 CH3 7 3.99 -0,026 -0,264 0.525 H 7 3.49 +0.053 -0.264 0.525 =Activities are relative to chloroquine which is 100. (R. L. O'Brien and F. E. Hahn, Actimicrob. Ag. Chemother., 315 (1965). * T = sum of Hensch rr values of all segments (including the quinoline ring) of the molecule.
For the analyses reported here, the charge on the terminal N of the diamino side chain was always taken to be that of the protonated cation (designated Q N Z + ) . The N attached to the ring was taken to be formally
278 Journal of Medicinal Chemistry, 1971, Vol. 14, S o .
4
PURCELL, et nl.
TABLI:I1 ANTIM.kLAHI.iL ACTIVITIES. W D PIR.IYETER CH
VALUES FOR
ShRIES 11
j
I
P ~ H ~ " ( c H J , ~ H ( c ~1)H - ,
Y s C1 13
T
II 6-C1 I3 5-C1 H 8-c1 H 3-Br H 6-OCH3 C1 6-CH3 C1 3-CH3 C1 2-CH3 2-Ph 3-Br C1 5-Br C1 8-NH, See footnote a, Table I.
Activitya
&
100 100 3 3 3 10 23 15 10 2 6
*
A
=
7rh
Qx -0.124
4.20 4.20
4.20 4.20 4.35 3.47 4.70 4.70 4.70 6.33 5,06 5 5.06 2 2.97 sum of Hansch A values of all
+0.055 +0.033 +0. 055 s0.053 -0.213 -0.123 -0.124 -0.124
-0.124 -0.124 -0,120 -0.127 segments
0x1 QN?QY -0.264 $0.52.5 + 0 , 055 -0.264 +0.525 -0.126 +0.525 -0.122 -0.264 -0,264 +0,525 -0.124 -0.261 0.525 -0.115 +0,52.5 -0.213 -0.266 + 0 . 525 -0.104 -0,264 -0,264 + 0 525 -0,092 f0.525 -0.098 -0.264 $0.325 -0.027 -0.265 + 0 . 525 -0.113 -0.262 +0.525 -0,105 -0.269 $0.525 -0.443 -0 263 (including the qiiinoline ring) of the molecule.
+
TABLE 111 ANTIMALARIAL ACTIVITIES A N D PARAMETER V.\LUES FOR SEHIKS I11
R
R
.ictivity"
KHCH(CH3)(CH2)3N+(CZHj)*H 100 100 XHC6HioN +(C*H,)*Hc NHCsHloN +(C,Ho)JIc 25 100 NHC6HioN +H2C2Hjc NHC6HloN +H*CH(CH3)CHsc 50 NHC6HioN +H*CsIIii C ~ S 30 23 NHC6HloN+H2C6Hl1trans 100 KHCH (CHa)(CH*)aS+HnCH3 30 XHCH (CH3)(CH2)3?;+HzCnHj 40 NH(CHz)3N + ( C ~ H E ) ~ H s S"(CH2)zN +(CH&H*OH)zH 6 NH(CH*)aN'(CsHi3)zH > NH(CH&N +(CsH17)2H SHCH2CH(OH)CH2I\;+(C*Hj)zH 100 ('See footnote a, Table I. T = sum of Hansch cyclohexane substituted in the 1 arid 4 positions.
b,
4.20 4.18 6.26 3,23
3.64 4.66 4.54 2.75 3.27 3.25 -0.32
T
7.44 9 52 0.96 values of
(2 x -0.124 -0.124 -0,124 -0,124
Os1
Qx2
QN?
-0.264 - 0 , 203 +0.52g -0.264 -0.208 S0.524 -0.208 +0.524 -0,264 -0.264 -0.359 $0.346 +0.344 -0.264 -0.361 -0.124 -0,361 +0.348 -0.124 -0,264 -0 361 $0.348 -0,264 -0.124 -0.355 +0.349 -0,264 -0.124 -0.357 $0.347 -0.124 -0.264 -0.209 +0.326 -0,124 -0.262 -0.20.; +0.527 -0.262 -0.124 -0.210 0 . 52.5 -0 124 -0.262 -0.210 f0.32.i -0.262 -0.124 -0.124 -0.261 -0.208 +0,527 all segments (iricluding the quinoline ring) of the molecule. CeHio is
neutral. Analyses not reported herels served to indicate that, at least for the terminal S, it makes little difference in the extent of correlation whether the nitrogen is considered protonated or not. As a measure of the "goodness of fit" for a particular activity equation, the square of the multiple correlation coefficient, &iz(az= 1.0 indicates perfect correlation), the overall F ratio for the regression and corresponding significance level, and the amount of explained variance, eV, were c a l ~ u l a t e d . ' ~The regression coefficient gives an indication of the degree of correspondence between the experimentally observed antimalarial activities and those calculated with the trial linear equation resulting from the regression analysis. 6iz is interpreted as the fr:iction of the sum of squares of the deviations of ob(18) G E Bass, D R Hudson J E Parker, and \V P Purcell, "Progress in hlolecular and Subcellular 13iolog) '' Vol 2, Springer-Verlag, Berlin, In press. (19) G W Snedecor and W G Cochran Statistical Methods," I o n a State Unikersity Press, Imes, I o n a , 1967, pp 386-388, 400-402
+
served activities from the mean activity that is attributed to the regression. The F ratio is the decision statistic of the F test of significance. The overall F test with this model is a test of the null hypothesis that all of the parameter coefficients are equal to zero; in other words, the mean activity would be as good an estimate of the true activity as the activity calculated from the linear regression equation if the null hypothesis is true. The explained variance gives the fraction of the variance of the antimalarial activities which is attributed t o the linear relationship of those parameters included in the analysis. Even though regression coefficients may prove to be statistically significant with the F test, it is not uncommon t o find that the fraction of explained variance is quite small. This would indicate that most of the variance in the activities must be attributed t o variables not included in the regression. I t should be mentioned that while an explained variance of 1.0 indicates perfect correlation, it is possible for the calculated
Journal of Medicinal Chemistry, 1971, Vol. 14, No. 4 279
CHLOROQUINE ANALOGS
explained variance to be negative. This occurs when, on a per degree of freedom basis, the mean activity is a better approximation to the true activity than the calculated activity. The regression analyses were carried out using a computer program developed in this laboratory based on IBM 1620 Program 06.0.148, “Single and Multiple Linear Regression Analyses.” These and the molecular orbital calculations reported here were performed on an IBM 1620 computer.
reveals that Q N ~and QN2+ do not vary appreciably in this series and thus need not be included in the regression analyses. The parameter combinations considered and results obtained for this series of eight molecules are presented in Table V. TABLEV REGRESSION ANALYSES RESULTS FOR SERIESI CH,
I
NHCH(CH,),yH(C2H,
Analyses Involving Series I, 11, and I11 Combined.The overall ability of the structure-activity equation (eq 11) to explain variations in antimalarial activity was tested by carrying out regression analyses on the combined series of 33 molecules comprising series 1-111. The combinations of parameters examined and results obtained are presented in Table IV. In particular, it
f(r*,*,
+b aTz + b
& x ,&NI, Q N Z + ) =
arz aQx
+ b?r + c
+b 4-b +b aQx + + +d + b* -I- CQx + a&Ni
c&N~+
bQN1
asa
f
a
eV
0.17
0.05 0.08 0.05 -0.03 -0.03 0.13 0.09
0.25
80
d&Ni
e&Nzf
ff
Log (O.l/METD)
=
f(+,
+ + + +
+ +
85 90 80 IETD) = u Q ~ b 0 79 6.j 0 59 Br 0 80 83 0.69 Br, I 0 77 90 0.69 Br, CH3 0 44 75 0.2d Br, I, CHI 0 41 80 0 26 Br, I, C H , CF, 0 00 -is the charge calculated for that atom of substituent
T attached directly t o the ring. The results for this series are remurkable in the virtually complete failure to obtain correlation, coefficient significance, arid explained variance. One can only conclude that the addition of the third ring substituent, Y leads to a drastic deviation from the manner in which the 4,i-disubstituted analogs exert their antimalarial effect. O'Brien aiid HahnZasuggest that this effect may be due to hteric liiiiderance t o intercalation, or an alteration in the relative orientation of the drug to tlic nucleotide babes TI hich produces a less stable complex than occurs with chloroquine. The calculations were repeated after eliminating d l polyatomic substituents except S H 2 at X and I- S o appreciable improvement was obtained. Analyses Involving Series 111.-Series 111 consibis of 14 molecules nhich differ from one another only at the 4-diamino side chain. I n the model of O'Brien arid Hahn, this side chain is depicted as being electrostatically bound through the diamino nitrogens to phosphate groups on opposite DSA strands. The D S A geometry led O'Brien and Hahn to predict that the effectiveneks of this side group should depend on the number of carbons separating the nitrogens; the optimal number for this separation is 4. ?;he of the compounds in series I11 have the two diamino nitrogens
Journal of Medicinal Chemistry, 1971, Vol. 14, No. 4 281
CHLOROQUINE ANALOGS
TABLE X
separated by 4 C while, for the remaining 5 compounds, the separation is 3 C. When all of the molecules in series 111were considered together, the only meaningful results were obtained with the Hansch parameter. The combinations of variables tested and results obtained are presented in Table IX.
REQRESSION ANALYSES INVOLVING SERIESI11 MOLECULES WITH DIAMINO NITROGENS BY 4 C SEPARATED
R
&
c1
TABLE IX REGRESSION ANALYSES RESULTSFOR SERIESI11
f(rz,r 8Q N ~ ,QNZ+)'
+b ar2+ b ar
ar2 a&Ni f(r2, T , Q N I ,
+b ad +b + br + c
QNZV
ar
ar2
a&Ni
f b
+b
~&Nz+
a&Ni
4- b Q w + -I- c
+ b Q ~ i+ c a* + +c a d 4-!- c ar2+ +c ar + 4- C&NZ+ -I- d a r 2 + b Q ~ i+ -k d a+ + b r + +d ar
~QNz+
b&Ni
~QNz+
b&Ni
c&N2+
+ + + +
CQN~
+ +
an2 b r C&NZ+ d ar2 b r C Q N ~ d&N2+ e a Log (O.l/METD) = j(rz,r, QNI,
+
cR* 0.24 0.48 0.69 0.11 0.10 0.13 0.39 0.29 0.53 0.49 0.39 0.54 0.70 0.70 0.70
Significance level, %
eV
90 99 99.5 70 70 50 90 80 97.5 95 80 95 99 99 97.5
0.17 0.43 0.63 0.03 0.02 -0.03 0.28 0.17 0.45 0.39 0.21 0.40 0.61 0.61 0.57
QNZ+).
+ br + c
+b
f b
~&Nz+
a&Ni
f
b&Nz+
f
C
+ b Q ~ i+ c a* + +c ar2 f 4ar2+ -k c a r + b Q ~ i+ +d a+ + b Q ~ i+ +d ar2 + b r + +d + b* -k f d ar
~&Nz+
b&Ni
b&Nz+
&NZ+
~ N Z +
CQN~
C&N2+
+
a r 2 4-b r -I-&?NI a Log (O.l/METD)
d&Nz+
+e
a*
Significance level, %
eV
0.44 0.44 0.44 0.16 0.01 0.17 0.45 0.79 0.46 0.78 0.81 0.79 0.46 0.79 0.81
90 95 80 70
