Pharmaceutical Research, Vol. 8, No. 10, 1991
`
`Research Article
`
`Photoreactivity of LY277359 Maleate, a 5-Hydroxytryptamine3
`(5-HT3) Receptor Antagonist, in Solution
`
`Gerold L. Mosher" and Julianne McBee'
`
`Received November 8, 1990; accepted May 14, 1991
`
`Compound LY277359 maleate undergoes a photoinduced solvolysis reaction in water to generate the
`corresponding hydroxylated product and release chloride. Attempts to stabilize a parenteral formu-
`lation of the compound led to an investigation of possible reaction mechanisms. The data are consis-
`tent with a mechanism involving homolytic cleavage of the aryl—chloride bond followed by electron
`transfer to give an aryl cation intermediate. The cation thus formed reacts with surrounding nucleo-
`philes to give the substituted product. A kinetic expression for reaction rate was derived from the
`mechanism, and various components of the rate constant were evaluated experimentally. The reaction
`is slowed with the addition of chloride, presumably via a common ion effect (enhanced retroreaction).
`In the absence of added chloride, the reaction can be described kinetically by an initiation term. An
`inner filter effect is also observed, where increasing amounts of the hydroxylated product slow the
`reaction. Experimental data for observed rate constants as a function of starting concentration and
`light intensity are fit with good correlation to an equation describing the filter effect. Additional studies
`evaluated the effects of various structural features of the parent compound on the rate of the reaction
`in glass containers. It was determined that reactivity was dependent on two features: (1) the ortho
`positioning of the carboxyl and ether groups, which shifted an absorption band above the container
`cutoff; and (2) the para orientation of the chloro group to the ether, which is para activating in the
`photoexcited state.
`
`KEY WORDS: photosubstitution; aryl cation; inner filter; kinetics; common ion effect.
`
`INTRODUCTION
`
`Preformulation studies on LY277359 maleate (I), a po-
`tent and selective 5-HT3 receptor antagonist, were compli-
`cated when we discovered that aqueous solutions were un-
`stable in the presence of light. The use of amber vials slowed
`the reaction but did not impart sufficient stability for formu-
`lation. Analysis of the reaction mixture by high-performance
`liquid chromatography (HPLC)/mass spectrometry and nu-
`clear magnetic resonance (NMR) indicated that the only
`product formed was the substituted product (II) (Scheme I).
`The complete absence of the corresponding reduction prod-
`uct (III) was interesting, and this fact along with the need for
`a stable parenteral formulation of compound I led us into
`further studies to explore the nature of this reaction.
`In the study reported here, we examine the impact of
`chemical structure and reaction conditions on the photosub-
`stitution reaction and investigate possible reaction mecha-
`nisms. The overall goal was to understand the reaction better
`so that appropriate measures could be undertaken to stabi-
`lize a parenteral formulation of compound I.
`
`' Lilly Research Laboratories, Eli Lilly and Co., Indianapolis, In-
`diana 46285.
`2 To whom correspondence should be addressed.
`
`ss‘
`
`LY277359 maleate (I) X,
`
`Scheme I
`
`MATERIALS AND METHODS
`
`Materials
`
`Compounds I (endo-5-chloro-2,3-dihydro-2,2-dimethyl-
`N- (8-met h yl- 8 - a z ab icy clo [3 . 2 . llo c t - 3 - y1)- 7 - b e n -
`
`1215 (cid:9)
`
`0724-8741/91/1000-1215$06.50/0 (cid:9)
`
`1991 Plenum Publishing Corporation
`
`MYLAN Ex. 1013, Page 1
`
`

`
`1216 (cid:9)
`
`Mosher and McBee
`
`zofurancarboxamide (Z)-2-butendioate), II, IV, and VIII-X
`were used as received from Lilly Research Laboratories (1).
`Compounds V-VII, XI, and XII were obtained from Aldrich.
`HPLC-grade acetonitrile was used and all other chemicals
`were reagent grade. All solutions and buffers were prepared
`using water for injection, and all glassware used for drug
`solutions was silanized to prevent drug adsorption. Twenty-
`milliliter scintillation vials (Kimble), --280-nm cutoff, were
`used as reaction vessels in the light stability studies.
`
`UV Absorption Spectra
`
`A Beckman DU-70 spectrophotometer was used for UV
`absorption spectra using a 1-cm-path length cell and 60-nm/
`min scan speed. Solutions were scanned from 500 to 190 nm.
`The solvent of interest was used for calibration and back-
`ground correction.
`
`HPLC Analysis
`
`LY277359 maleate and its photoreaction products were
`separated isocratically on a DuPont Zorbax TMS column
`(4.6 mm x 25 cm) at ambient temperature. The mobile phase
`was acetonitrile:0.125 M (NH4)2SO4, pH 4.0 (28:72), deliv-
`ered at a 1.5-ml/min flow rate with a Spectra-Physics Model
`8810 pump. Injections were made using a Bio-Rad autosam-
`pler (Model AS-48) with a 20-0 fixed loop operated at 4°C.
`The sample chamber of the autosampler was covered with
`aluminum foil to protect the samples from light. Effluent was
`monitored on an Applied Biosystems Spectraflow UV ab-
`sorbance detector (Model 783A) at a wavelength of 235 nm
`with peak areas being determined on a Hewlett-Packard
`3396A integrator. Concentrations were calculated from ex-
`ternal standard curves generated from aqueous drug solu-
`tions prepared in silanized glassware.
`
`Kinetic Studies
`
`A clear, incandescent, 100-W bulb (General Electric)
`was used as the light source and a light chamber was con-
`structed to fix the distance the reaction vessels were placed
`from the bulb. The temperature in the chamber was con-
`trolled at 30 ± 1°C by air circulation. Solutions were pre-
`pared by weighing the compound of interest into volumetric
`flasks, solubilizing with the desired solvent, and decanting
`20 ml into the vessels. For studies with added halogens, KI,
`KBr, KCI, and NaCI were weighed directly into the vessels
`and solubilized with a 0.042 mM solution of compound I to
`make the solutions 0.154 M in halogen. Once prepared, all
`solutions were stored in the dark until needed. The studies
`were conducted by placing the capped vessels in the light
`chamber at a measured distance from the source and taking
`0.5-ml aliquots over time. Reversed-phase HPLC was used
`to assay the aliquots for compound concentration and mon-
`itor the appearance of reaction products. Reactions were
`followed for 6 hr. Unless otherwise noted, all reactions were
`conducted with the vessels placed 10 cm from the light
`source.
`Rate-concentration dependency studies were per-
`formed by varying the initial concentration of compound I.
`The effect of chloride ion was studied by monitoring the
`kinetics of the reaction in solutions that were 0.017, 0.077,
`0.154, and 0.308 M in chloride and 0.1 mM in compound I.
`Ionic strength effects were evaluated in a similar manner by
`using NaC104 in various concentrations up to 0.2 M.
`Reactions run in water/acetonitrile solutions were pre-
`pared by solubilizing compound I in a known amount of
`water and bringing to volume with acetonitrile. The solutions
`evaluated were 55.6, 41.7, 27.8, and 13.9 M in water and 0.1
`mM in compound I.
`Varying light intensity was accomplished by placing the
`reaction vessels 5, 10, or 15 cm from the filament of the bulb.
`The effect of compound II on the reaction kinetics of I
`was studied by mixing aqueous solutions (of known concen-
`trations) of I and II prior to light irradiation. Effects of other
`light absorbing organic species were studied by dissolving
`compound I in aqueous 1.1 mM solutions of p-aminobenzoic
`acid (XI) or p-nitrophenol (XII) before initiating the reaction.
`Control reactions were included in each study.
`
`RESULTS
`
`UV Absorbance
`
`The UV absorption spectra of compound I in water
`shows a secondary absorption band at 318 nm. The nature of
`this band was evaluated by comparing the UV absorbance
`spectrum in polar and nonpolar solvents. The direction of
`shift in the X. on changing solvents can be used as an
`indication of n-7r* versus Tr-Tr* transitions (2,3). Upon
`changing the solvent from n-hexane to water, a 12-nm red
`shift was observed in the Xmax. This and the magnitude of the
`molar absorptivity, 4800 cm -1 M - at 318 nm, are indicative
`of a Tr-Tr* (B-band) transition. The wavelength of maximal
`absorption for this band in water is given in Table I for the
`series of structurally related compounds evaluated in the
`present work.
`No appreciable change in molar absorptivity was noted
`when the solvent was changed to methanol, ethanol, n-pro-
`panol, or acetonitrile.
`
`Reaction Kinetics
`
`Representative data from light stability studies are
`shown in Fig. 1. The data demonstrate apparent first-order
`behavior as would be expected for a photochemical process
`in dilute solution. A least-squares regression analysis of the
`data gives the solid line, the slope of which is used to cal-
`culate an observed rate constant (kobs) value. These values
`are compared as the reaction conditions are varied.
`The effect of solvent polarity on the occurrence and rate
`of the substitution reaction is demonstrated rather dramati-
`cally in Table II. Due to possible variations in the reaction
`conditions, (lamp aging, different initial concentrations,
`etc.), a control reaction was run for each set of experiments.
`When the polarity of the medium was reduced by using a
`mixture of methanol in water, the kobs value was reduced by
`half. Two additional peaks appeared on the chromatograms
`from that study but were not isolated and identified. Their
`relative position on the chromatogram suggests the possibil-
`ities of a methoxy substitution and/or the reduction product.
`A similar reduction in kobs values was seen in the pres-
`ence of increasing amounts of acetonitrile in water. Irradi-
`
`MYLAN Ex. 1013, Page 2
`
`

`
`Photoreactivity of LY277359 Maleate in Solution (cid:9)
`
`1217
`
`Table I. Absorbance and Reactivity Data for the Compounds Investigated
`
`Compound (cid:9)
`
`Xmax (nm)
`
`Structure
`
`IV
`
`V
`
`VI
`
`VII (cid:9)
`
`VIII"
`
`IX
`
`X
`
`310b
`322'
`
`282
`
`280
`
`308
`
`300
`
`290
`
`301
`
`0.00446
`0.00692
`
`No reaction
`
`No reaction
`
`No reaction
`
`0.000958
`
`No reaction
`
`No reaction
`
`" Observed rate constant in min - ' from 0.043 mM solutions at 10 cm. kob, = 0.00708 min - for
`compound I.
`b Ionized carbonyl.
`Protonated carbonyl.
`d Maleate salt used.
`
`compound IX, or moving the chloro group to the four posi-
`tion, compound X, destroys reactivity. Likewise, removal of
`the 2,3-dihydrofuran ring, compound V, eliminates reactiv-
`ity. Compounds VI and VII, assessing the importance of an
`oxygen attached para to the halogen and possibly hydroxyl—
`carbonyl interactions, respectively were also nonreactive.
`However, some reactivity was observed with compound
`VIII, which is the opened-ring version of compound I.
`
`DISCUSSION
`
`Aromatic photosubstitution reactions have been under
`
`Table II. Effects of Solvent, Added Halogens, and pH on the Ob-
`served Rate Constant
`
`Solvent
`
`H2O (27.8 M)/CH3OH
`H2O (control)
`H2O (41.7 M)/CH3CN
`H2O (27.8 M)/CH3CN
`H2O (13.1 M)/CH3CN
`CH3CN
`CH3OH
`(CH3)20H
`(CH3)3OH
`
`NaCl (0.154 M)
`KCI (0.154 M)
`KBr (0.154 M)
`KI (0.154 M)
`
`H20, pH 3.0"
`pH 4.4"
`pH 11.0"
`
`[Compound I]
`(mM)
`
`/cabs
`(min - (cid:9) x 103)
`
`0.110
`
`If
`
`0.110
`
`If
`
`0.043
`
`0.043
`
`0.043
`
`0.043
`
`2.4
`4.1
`3.6
`2.4
`0.79
`No reaction
`
`3.9
`4.0
`9.6
`5.9
`
`4.7
`4.6
`5.0
`
`ating solutions of compound I in pure nonaqueous solvents
`of differing polarity gave no reaction.
`Also included in Table II are the results of reactions run
`in water of varying pH. No significant change in ko„ was
`observed.
`When excess chloride was added, the reaction was
`slowed; /cobs showed a dependency on the reciprocal of the
`chloride concentration. This correlation is illustrated in Fig.
`2. However, when halogen ions other than chloride were
`added, the rate of reaction was faster than when an equimo-
`lar amount of chloride was added (Table II), and additional
`products were generated. Although these products were not
`identified, the chromatographic results are in agreement with
`the proposal that iodide and bromide compete with chloride
`and water for substitution, generating the corresponding
`5-bromo and 5-iodo compounds.
`
`Structural Effects
`
`Structural effects on the rate of the solvolysis reaction
`of compound I are given in Table I. The reactivity of com-
`pound IV indicates no involvement by the amide-linked tro-
`pane ring in the reaction, but elimination of the carbonyl,
`
`0 100 200 300 400
`Time (min)
`
`In Concentration (M)
`
`Fig. 1. Representative data showing the first-order disappearance of
`compound I over time.
`
`" The starting pH of 4.4 was adjusted to pH 3.0 with 0.1 N HCl and
`to pH 11 with 1.0 N NaOH.
`
`MYLAN Ex. 1013, Page 3
`
`(cid:9)
`

`
`Mosher and McBee
`
`cies are the aryl cation, the radical anion, radical cation, and
`the o- complex. Of these, radical anions are probably not
`involved as they are known to be quenched very efficiently
`by oxygen (10), and the reaction proceeded quite well in our
`studies even though we made no attempt to deoxygenate the
`solutions. Although not totally eliminated, radical cations
`are unlikely because of the enhanced reactivity in the pres-
`ence of potassium iodide. Iodide reduces cation radicals
`very well and, in fact, is sometimes used for the iodometric
`assay of cation-radical salts (11). If the intermediate is a
`radical cation, the presence of iodide should decrease the
`rate.
`A r complex intermediate has been proposed as a pos-
`sible intermediate in the photosubstitution of halophenols
`and haloanisoles (12), but the solvent system used was di-
`oxane/water (95:5). The greater nucleophilicity of dioxane
`relative to water would greatly enhance its ability to interact
`in an SN2-type reaction. The kinetics of SN2 reactions with
`one reactant in excess are usually apparent first order as is
`seen for the present reaction. However, when the kinetic
`expression for the dependence of the reaction on added chlo-
`ride is derived from Scheme II using a steady-state assump-
`tion for the concentrations of the intermediates, the ob-
`served rate constant is independent of chloride concentra-
`tion (see Appendix). The kinetics of the present reaction do
`
`ArCI + OH (cid:9)
`
`OH] --0-ArOH + Cl
`
`Cl (cid:9)
`
`[ArCI Cl]
`Scheme II
`
`show dependence on added chloride concentration. A linear
`relationship is seen between chloride concentration and
`1/kobs (Fig. 2). This kinetic disagreement in the presence of
`added chloride, the expectation that SN2 (AR*) processes
`occur predominantly in less polar media (13), and the inde-
`pendence of reaction rate on pH argues against an SN2
`(AR*)-type reaction.
`The above evidence leads us to a mechanism involving
`an aryl cation intermediate, SO (AR*). This mechanistic
`scheme, proposed by Kropp et al. (14-17), involves initial
`homolytic cleavage of the carbon halogen bond followed by
`electron transfer within the resulting radical pair to generate
`an ion pair (Scheme III). The aryl cation can then interact, in
`
`+ C1
`
`electron
`transfer
`
`Ar + Cl
`
`H2O
`ArOH + H
`
`Scheme III
`
`our solvolysis reaction, with water to give the hydroxy-
`substituted product, compound II, or with chloride ion to
`regenerate compound I. The applicability of the mechanism
`has been expanded to cover a variety of alkyl halide photo-
`reactions (18,19). Perhaps even more germane to the present
`work is the use of the mechanism to explain photosolvolysis
`reactions in water (20,21). As explained by Moore and Pagni
`
`800 -
`
`600 -
`
`400 -
`
`200
`
`1/k obs (min)
`
`1218 (cid:9)
`
`
`0 (cid:9)
`0.4
`0.3 (cid:9)
`0.2 (cid:9)
`0.1 (cid:9)
`0 0 (cid:9)
`Concentration of Added Chloride (M)
`Fig. 2. Plot of 1/kobs versus the molar concentration of added chlo-
`ride. Each point represents data from one reaction vessel.
`
`investigation since the observation in the mid-1950s that UV
`irradiation of nitrophenyl phosphates gave a substitution
`product, the corresponding nitrophenol (4). The seemingly
`nucleophilic photosubstitution reaction generated consider-
`able interest because the meta isomer showed the most ef-
`ficient and clean reaction. This is in contrast to ground-state
`orientation rules for thermal reactions. By the late 1960s it
`was recognized that nucleophilic aromatic photosubstitution
`is a fairly general reaction (5,6). It can occur with a variety
`of cyclic, polycyclic, and heterocyclic aromatic systems,
`various solvents, and a whole range of leaving groups and
`reacting nucleophiles. As investigations continue, there is an
`increasing number of proposed mechanisms for the reac-
`tions, most of them dependent on the reaction conditions.
`
`Mechanistic Studies
`
`In general, irradiation of aryl halides produces products
`of nucleophilic substitution [Eq. (1)], reductive dehalogena-
`tion [Eq. (2)], and inter- and intramolecular arylation (7).
`
`by
`ArX —> Ar Y +
`
`hv
`ArX ---> ArH + SX (cid:9)
`SH
`
`(1)
`
`(2)
`
`Four general classes of mechanisms have been identi-
`fied for photostimulated aromatic substitution reactions (8)
`[Eq. (1)]. The classification is based on the type of interme-
`diate formed and uses an abbreviated notation for unimolec-
`ular (1), nucleophilic (N), substitution (S), via radicals (R).
`When radical anions or cations are formed, the reaction is
`termed SR N1 or SR,N1, respectively. If a primary photo-
`dissociation into an aromatic cation occurs the mechanism is
`denoted SN1 (AR*) and the classification of SN2 (AR*) is
`used for reactions involving the formation of cr complexes.
`Reductive dehalogenation, Eq. (2), provides yet another
`mechanism to be considered and usually involves homolytic
`cleavage of the aryl—halogen bond. The aryl radical interme-
`diate thus formed can abstract a hydrogen atom from appro-
`priate hydrogen donor solvents to give the reduced product
`(9). The absence of a reduction product in the current reac-
`tion does not necessarily eliminate the possibility of an aryl
`radical as the reactive intermediate but does suggest that
`alternative pathways are more probable.
`The effect of solvent polarity on the rate of reaction is
`consistent with the appearance of a polar or charged inter-
`mediate in the rate-determining step. Possible charged spe-
`
`MYLAN Ex. 1013, Page 4
`
`(cid:9)
`

`
`Photoreactivity of LY277359 Maleate in Solution (cid:9)
`
`1219
`
`(21), the absence of the reduction product is due to the in-
`ability of the radical intermediates to abstract a hydrogen
`atom efficiently from water. Therefore the two radicals ei-
`ther recombine or undergo electron transfer to give the ions.
`The large dielectric constant of water will facilitate the elec-
`tron transfer and favor the formation of ionic species.
`The aryl cation formed is highly reactive and will react
`rather indiscriminately with surrounding nucleophiles (usu-
`ally water). The ability of added nucleophiles to react with
`the cation is dependent upon the stability of the cation and
`the concentration of the nucleophile. For the highly reactive
`and unstable phenyl cation, nucleophile concentrations must
`approach that of the solvent for reaction to occur (22). The
`reactivity of compound I shows a modest inverse depen-
`dence on added chloride as a competing nucleophile (see
`below), the dependency agreeing kinetically with Scheme
`III. This is not unexpected as the cation formed, relative to
`the phenyl cation, should be somewhat stabilized by the
`ether oxygen. On changing the pH from 3 to 11, no effect on
`reaction rate was observed, further supporting a reactive
`cationic intermediate. In spite of the nucleophilicity of -OH,
`its concentration at pH 11 cannot compete with the avail-
`ability of water. There are several examples in the literature
`of photohydrolyses that are independent of hydroxide ion
`concentration over a large pH range (23).
`
`Kinetic Description
`
`The proposed mechanism illustrated in Scheme III can
`be described by Scheme IV if it is assumed that the electron
`transfer occurs very fast and the concentrations of water and
`chloride are invariant. If a steady-state assumption is made
`
`k,
`ki (cid:9)
`ArC174--4:Ar ±-0--ArOH
`k2
`Scheme IV
`
`for [Art], then the loss of ArCI can be described by Eq. (3):
`
`kik3
`d[ArCI] (cid:9)
`
`[ArCI]
`-
`
`dt - k2 + k3
`
`(3)
`
`and the relationship between the rate constants from Scheme
`IV and /cobs is
`
`kobs k2k-Fik3k3
`
`(4)
`
`In comparing Scheme IV to Scheme III, it is seen that pho-
`tochemical initiation, containing terms for intensity and
`quantum efficiency, is incorporated into k1. A chloride con-
`centration term is included in k2 and can be described as k2
`k'2[C1-], and k3 incorporates a water concentration term,
`k3 = k'3[H20]. No ionic strength effect was observed over
`the range evaluated (0-0.2 Al).
`The various components of the overall rate constant
`were evaluated experimentally by altering the reaction con-
`ditions and examining the resulting observed rate constants.
`By monitoring the reaction in the presence of added chloride
`ion, alterations in k2 were imposed. Transforming Eq. (4) by
`
`substituting the chloride terms for k2 and rewriting in the
`reciprocal form gives Eq. (5):
`
`1 (cid:9)
`kobs (cid:9)
`
`k' 2[01 (cid:9)
`+
`kik3
`
`1
`k1
`
`(5)
`
`This equation shows a linear relationship between 1/kobs and
`[Cl-]. The inverse of the y intercept is equal to k1 , and this
`value along with the slope can be used to solve for the ratio
`of k3 to k'2. This plot, given in Fig. 2, shows good linearity
`over the chloride range studied (r = 0.93). Regression anal-
`ysis gives a value of 0.00407 for k1 and a relationship of k'2
`= 5.22 k3. Thus at chloride concentrations around 0.2 M, the
`k2 (k2 = k'2[C1- ]) and k3 terms are approximately equal and
`at chloride concentrations of 0.02 M or less, the k3 term
`dominates and Eq. (4) reduces to /cobs = k1.
`In the strictest sense, Eq. (3) is not valid at zero added
`chloride since the chloride concentration changes as the re-
`action progresses. However, as a worse-case scenario, if the
`reaction is followed to completion in the absence of added
`chloride, the maximum amount of chloride ion produced in
`our studies would be equal to the initial concentration of
`compound I (approximately 0.1 mM). This chloride concen-
`tration is easily within the range discussed above where kobs
`= k, and the observed rate is independent of the amount of
`chloride present or being produced.
`To evaluate the k3 component (k3 = k'3[1-120]), the wa-
`ter concentration was varied by using a mixed solvent sys-
`tem with decreasing amounts of water in acetonitrile. The
`results are given in Table II and indicate that kobs decreases
`as water concentration decreases. In retrospect, this study
`should not have given us any information since zero added
`chloride causes the k3 term to drop out and therefore our
`model is insensitive to this effect. In addition, since the aryl
`cation that is formed during the reaction is quite reactive, the
`nucleophilicity of acetonitrile may allow the formation of the
`nitrilium ion, as has been seen for anthracene (24). It seems
`reasonable that the decrease in reaction rate that we are
`seeing is due to the decreased polarity of the solvent which
`decreases the ability of the electron transfer to occur and the
`ionized species to form (17,18,25). Thus under these condi-
`tions, the assumptions made in deriving Scheme IV fail.
`Evaluation of the k1 term, containing intensity and
`quantum yield, using the ko„,, relationship [Eq. (4)], began
`with a determination of the rate constants for the reaction as
`light intensity was varied. These results are given in Fig. 3
`for a range of starting concentrations. As the initial concen-
`tration of compound I increases, kobs decreases in a nonlin-
`ear fashion. Varying the light intensity by changing the dis-
`tance the reaction vessel is placed from the light source gives
`similar concentration effects but causes the curve to shift to
`higher kobs values with increasing intensity. Attempts to lin-
`earize the data by plotting kobs against the reciprocal of the
`concentration (26) were unsuccessful, indicating changes in
`the quantum yield, 1, and/or the intensity with concentra-
`tion. Since the reaction was conducted at concentrations of
`compound I where not all of the incident light was absorbed,
`alterations in the intensity term are expected, and can be
`corrected by using Beer's law. Among the many other
`causes for variations in 1 and/or intensity is the potential for
`the product, II, to absorb light competitively (inner-filter
`
`MYLAN Ex. 1013, Page 5
`
`(cid:9)
`(cid:9)
`(cid:9)
`

`
`1220 (cid:9)
`
`Mosher and McBee
`
`factors of geometry, etc., which are unique to the apparatus
`being used. It should therefore be replaced with the more
`general term n as suggested by Moore (26). By making this
`substitution and one for the Beer's law expression for ab-
`sorbance into the equation, a working form is generated.
`
`kobs
`
`kn(1 (cid:9)
`
`e-2.303 b Co)
`
`Co
`
`(9)
`
`The data from Fig. 3 were fit to Eq. (9) with the non-
`linear regression function provided in the SAS software,
`while floating the quantities ((rn.) and (e b). Initial estimates
`were obtained from a Taylor's series expansion of Eq. (9)
`which gave a final result linear in concentration describing
`the limiting slope at low Co values. The results of the SAS
`regression are given in Table III and are used to generate the
`solid lines in Fig. 3.
`The good fit of Eq. (9) to our experimental data indi-
`cates that the changes in /cobs with concentration can be ex-
`plained in terms of an inner-filter effect. This is further sub-
`stantiated in Table IV, where the addition of compound II
`and other organic species, compounds XI and XII, known to
`absorb light in the same wavelength range, are seen to slow
`the reaction.
`
`Structure-Reactivity Relationships
`
`Spectroscopic analysis of the compounds in Table I pro-
`vides some rationale for the observed reactivity. All of the
`compounds show a 7r-ir* band peaking in the wavelength
`range of 280-325 nm. Compounds V and VI, having func-
`tional groups either meta or para to the chloro substituent,
`show an absorption band peaking around 280 nm. This wave-
`length is below our reaction vessel cutoff and these com-
`pounds would not be expected to show reactivity. The com-
`pounds which have a carbonyl and an ether or hydroxyl
`oxygen in an ortho configuration show a red shift in this peak
`(VII, 308 nm; VIII, 300 nm), presumably due to hydrogen
`bonding. These absorption bands are barely in the spectrum
`transmitted by the vessel and reactivity would still be ques-
`tionable. The fact that VIII (lower kmax) has some reactivity
`and VII does not implies that factors other than vessel cutoff
`are involved. One possibility is the need for the alkoxy-
`group activation of the para position (31,32) not just the
`ortho/para direction (33) as has been postulated for hydroxy
`groups. The lack of reactivity with X (Xmax = 301 nm),
`where the chloro is in the 4 position, further suggests that
`para activation is of considerable importance.
`Constraining the ether oxygen into a ring further shifts
`the knax as illustrated by compounds I (320 nm) and IV (322)
`(protonated carbonyl) and gives full reactivity. However,
`allowing a charge on the carbonyl or removing it (IX),
`
`Table III. Parameters Generated from SAS Regression Analysis
`
`Distance from
`light source (cm)
`
`5
`10
`15
`
`orn X 106 (cid:9)
`
`Eb x 10-4
`
`2.55
`0.509
`0.362
`
`1.13
`1.18
`0.89
`
`0.07
`
`0.06
`
`0.05
`
`0.04
`
`0.03
`
`0.02
`
`0.01
`
`k obs (1/min)
`
`0.00
`0.00 (cid:9)
`
`0.05 (cid:9)
`
`0.10 (cid:9)
`
`0.15 (cid:9)
`
`0.20
`
`0.25
`
`Initial Concentration (mM)
`Fig. 3. Plot of lc„,„ versus the initial molar concentration of I for
`reaction vessels at different distances from the light source. Each
`point represents results from one reaction vessel. The overlaid lines
`are a fit of Eq. (9) to the data. (ID) 5 cm; (0) 10 cm; and (A) 15 cm.
`
`effect) (27). Since a UV scan of compound II shows it to
`have an absorbance spectrum almost identical to that of
`compound I, competitive light absorption by the product
`was highly probable.
`In a light irradiation experiment, the inner filter effect
`can be described as the absorption of incident radiation by a
`species other than the intended primary absorber (28). In this
`experiment, the secondary absorber proves to be the prod-
`uct, II, of the reaction. Zimmerman et al. (29) were the first
`to consider the case where the reactant and product have
`similar spectral characteristics. The concept was later ex-
`panded by Bunce (30) to cover the general case where the
`product and reactant are different spectrally. For our studies
`we use a form intended for irradiation studies at an isosbestic
`point (30).
`
`CO (cid:9)
`In Cf (cid:9)
`
`(mo(1 (cid:9)
`
`e-2.303 Abs) (cid:9)
`
`Co
`
`(6)
`
`In Eq. (6), Co and Cf are initial and final concentrations of
`reactant respectively, (Dr is the quantum yield for reactant
`disappearance, and /0 is the accumulated photon dose. The
`term (1 - e-2.303 Abs) is a correction factor for the amount
`of light actually absorbed. Differentiation with respect to
`time gives Eq. (7), where I is now an intensity term, i.e.,
`photon dose per unit time.
`dC (cid:9) OA) _ e-2.303 Abs)
`-
`Co
`dt
`
`(7)
`
`Although this equation describes reactions run using mono-
`chromatic light, we assume for demonstrative purposes that
`it will hold for the summation of wavelengths initiating our
`reaction. This is a reasonable assumption since the reactant
`and product have almost identical UV absorption spectra.
`Comparing Eq. (7) to Eqs. (3) and (4) gives Eq. (8),
`where the observed rate constant is seen to be a function of
`the initial concentration of compound I used and the absor-
`bance of the solution.
`
`kobs
`
`(Dri(1 (cid:9)
`
`e-2.303 Abs)
`
`Co
`
`(8)
`
`It should be noted at this time that the I term contains
`
`MYLAN Ex. 1013, Page 6
`
`(cid:9)
`(cid:9)
`(cid:9)
`(cid:9)
`(cid:9)
`

`
`Photoreactivity of LY277359 Maleate in Solution (cid:9)
`
`1221
`
`Table IV. Effect of Adding UV Absorbing Species on the Observed Rate of the Substitution Reaction
`
`Species
`added
`
`II
`II
`II
`Control
`XI
`XII
`Control
`
`[Compound I]
`(mM)
`
`[Species]
`(mM)
`
`Distance from
`light source (cm)
`
`lcol„
`(min' x 103)
`
`0.11
`
`If
`
`0.11
`0.017
`
`0.11
`0.0733
`0.0367
`
`1.12
`1.06
`
`10
`
`10
`22
`
`2.84
`2.78
`4.12
`4.68
`2.20
`0.351
`5.30
`
`k, (cid:9)
`k3
`ArCI + OH - 7.1:1"--[ArCI • • OH - ] --0-ArOH + Cl -
`k2
`
`+ CI -.7.7"--[ArCI •• Cl -
`4
`
`Scheme II*
`
`d[ArCI]
` = -ki[01-1-][ArCl] + k2[ArC1 • • OH -]
`dt
`
`- k5[C1-][ArCl] + k6[ArC1 •• Cl -] (cid:9)
`
`(10)
`
`likewise,
`
`d[ArCI • • CI
`dt (cid:9)
`
`= k5[C1-][ArCl] - k6[ArCI • • Cl-]
`
`even with the oxygen tied into the furan ring, reduces or (cid:9)
`eliminates reactivity in our system as the Xmax is shifted back (cid:9)
`toward the vessel cutoff. (cid:9)
`In conclusion, the photoinduced solvolysis reaction of
`compound I probably involves homolytic cleavage of the (cid:9)
`aryl-chloride bond followed by electron transfer to give an (cid:9)
`aryl cation intermediate. This highly reactive species then (cid:9)
`reacts almost indiscriminately with surrounding nucleophiles
`to give the corresponding substitution product; in water this
`is the hydroxylated species. Greatest reactivity is seen when ment is as follows. The differential equation describing the
`the parent molecule contains a substituent oriented para to loss of ArCI with time is
`the chloro group which is para activating in the photoexcited
`state and, of course, an absorption band appearing at wave- (cid:9)
`lengths above the vessel cutoff. For the series of compounds
`investigated, this occurs with an ether-linked oxygen para to (cid:9)
`the chloro group (preferably constrained into a furan ring)
`and an uncharged carbonyl attached ortho to the ether. (cid:9)
`The reaction is slowed with the addition of sodium chlo- (cid:9)
`ride attributed to a common ion effect (enhanced retroreac- (cid:9)
`tion) (34,35). In the absence of added chloride, the reaction
`can be described kinetically by an initiation term. An inner-
`filter effect is seen with this reaction where the appearance and
`of II slows the reaction by absorbing a portion of the incident
`radiation. The addition of II, or other species known to ab- (cid:9)
`sorb light in the wavelength range of 370-280 nm, slows the (cid:9)
`reaction via the same mechanism. Unfortunately, the use of
`(12)
`this technique to formulate stable parenteral formulations of
`I is complicated by the requirements that excipients need to Invoking the steady-state assumption for [ArC1 • • CI -1 and
`be nontoxic and pharmacologically inactive and to c