ADMET & DMPK 1(2) (2013) 6-16; doi: 10.5599/admet.1.2.3
`
`Open Access : ISSN : 1848-7718
`http://www.pub.iapchem.org/ojs/index.php/admet/index
`
`Original scientific paper
`Physico-chemical Profiling of the ACE-inhibitor Lisinopril: Acid-
`base Properties
`-
`,
`Semmelweis University, Department of Pharmaceutical Chemistry
`H-
`
`-
` E-mail: novak.krisztina@pharma.semmelweis-univ.hu;
`Tel.: +36 1 215 5241; Fax: +36 1 217 0891
`Received: January 7th, 2013; Revised: February 1st, 2013; Published: February 15th, 2013
`
`Abstract
`The acid-base chemistry of a tetraprotic, ampholyte ACE inhibitor, lisinopril, was studied by different methods.
`Potentiometry in aqueous medium and a co-solvent technique in methanol-water mixtures as well as 1H NMR-pH
`titration were applied for the highly precise measurement of protonation macroconstants. The log K values of lisinopril
` K1
`log K2
` K3
` K4
`-pH titration
`was used to assign the constants to the functional groups and for the examination of site-specific, submolecular
`basicities of the molecule (determination of protonation microconstants). In the first two well-separated protonation
`steps, the macro- and microconstants were identical and assigned to the primary amino group (log K1 = log kA) and to
`
`the secondary amine basicity (log K2 = log kBA), respectively. The two carboxylates exhibited overlapping protonation
`ABk
` = 2.15 log C
`characterised first by microconstants (log D
` = 3.10), revealing that the carboxylate on the proline ring
`ABk
`has nine times greater intrinsic basicity than the carboxylate on the side chain. The distribution of protonation species
`(Lis2-; HLis-; H2Lis; H3Lis+; H4Lis2+) and microspecies (ABC; ABD) as a function of pH was calculated and used to interpret
`the pharmacokinetic and pharmacodynamic properties of lisinopril.
`Keywords
`proton speciation; logK; potentiometry; NMR-pH titration
`
`Introduction
`
`Angiotensin converting enzyme (ACE) inhibitors serve as fundamental medicines in the treatment of
`hypertension, one of the most prevalent chronic diseases nowadays. Out of the numerous currently
`(2S)-1-[(2S)-6-amino-2[[(1S)-1-carboxy-3-phenylpropyl]amino]-
`available ACE-inhibitors,
`lisinopril
`
`hexanoyl]-pirrole-2-carboxylic acid (Fig. 1)
` belongs to the proline-containing structures.
`
`doi: 10.5599/admet.1.2.3
`6
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 1 of 11
`
`

`

`ADMET & DMPK 1(2) (2013) 6-16
`
`Lisinopril speciation
`
`Figure 1. Chemical structure of the ACE inhibitor lisinopril
`
`Due to its safe and effective properties, lisinopril is widely used in the therapy of essential hypertension,
`symptomatic and asymptomatic left ventricular systolic dysfunction, post-myocardial infarction, renal
`failure and diabetic nephropathy [1]. Lisinopril is administered orally usually in a daily dose of 2.5-40 mg. Its
`oral bioavailability is between 25-30 % [2]. Absorption from the gastrointestinal tract does not depend on
`nutrition. Six hours are needed to reach the maximum plasma concentration. It hardly binds to plasma
`proteins. Lisinopril is not metabolised, and the absorbed drug is primarily excreted invariably in the urine
`[3].
`
`A vast literature describes the pharmacology, mechanism of action, pharmacokinetics and analytics of
`lisinopril [3-7]. However, surprisingly little information is available on its physico-chemical properties.
`Lisinopril is a polyfunctional, ampholyte molecule, containing two basic and two acidic moieties. Various
`data have been published on its acid-base chemistry. The dissociation constants (pKa values) were reported
`by Bennion et al. [8] (1.68, 3.29, 7.01, 11.12) and Ip et al. [9] (2.5, 4.0, 6.7, 10.1), without an indication of
`the experimental error and with no assignment of basicities to moieties. Gonzalez et al. [10] assigned the
`measured pKa
`, but only on
`the basis of chemical evidence. These three datasets of pKa values are significantly different, particularly
`the lowest and highest constants, which remain uncertain. The difference between them exceeds one
`order of magnitude.
`
`The acid/base character determines the charge of a molecule in solution at a particular pH
`(characterised by the dissociation/protonation constant, pKa/log K). Further on in this paper, we use the
`log K term and consider the ionisation process as an association with a proton in all acid-base equilibria.
`This information is important in the estimation of ADME (absorption, distribution, metabolism, excretion)
`parameters and the interpretation of pharmacokinetic (PK) properties. Log K values can be used to better
`understand the binding mechanism in therapeutic events and also for optimisation of chemical reactions
`and analytical methods.
`
`Several analytical methods have been applied for the determination of lisinopril in biological samples
`and pharmaceutical preparations such as alkalimetry [11], spectrophotometry [12], high performance
`liquid chromatography (HPLC) [13], high performance thin-layer chromatography (HPTLC) [14], and
`capillary electrophoresis (CE) [15]. Obviously, upon the application of these methods, exact knowledge of
`acid-base chemistry is essential (for pharmacopoeial methods) or at least favourable (for chromatographic
`techniques and CE).
`
`The acidity/basicity of monovalent compounds can be quantified in terms of macroscopic log K
`parameters (macroconstants). For multiprotic compounds, macroconstants characterise the basicity of the
`molecule as a whole. They refer to the stoichiometric composition of the species, but they fail to provide
`information on specific proton-binding sites. Site-specific, submolecular basicities can be obtained when
`
`7
`doi: 10.5599/admet.1.2.3
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 2 of 11
`
`

`

`ADMET & DMPK 1(2) (2013) 6-16
`
`microconstants are determined. Microconstants measure the proton affinity of individual groups, while the
`protonation states of all other sites are definite in the molecule [16]. The macroscopic and microscopic
`basicities provide authentic
`information on propensities of
`intermolecular
`interactions both
`in
`pharmacokinetic (PK) and pharmacodynamic (PD) stages [17].
`
`As a part of our studies on the physico-chemical properties of ACE inhibitors, we investigated the acid-
`base chemistry of lisinopril. We characterised both the overall and the group-specific basicities of lisinopril.
`Validated potentiometric methods and 1H NMR spectroscopy resulted in more precise macroconstants
`than in previous investigations. Here, we first studied the site-specific (microscopic) protonation equilibria
`of lisinopril considering the overlapping protonation of carboxylates. The results were used to interpret the
`PK/PD properties of the compound.
`
`Experimental
`Materials and instrumentation
`Lisinopril dihydrate was generously supplied by Gedeon Richter Ltd. (Budapest, Hungary) and used
`without further purification. Distilled water was of pharmacopoeial grade [11] and all other reagents of
`analytical grade were purchased from commercial suppliers.
`
`Potentiometric titrations were carried out on a GLpKa automated pKa and log P analyser (Sirius
`Analytical Instr. Ltd. Forest Row, UK).
`
`NMR measurements were carried out on a 600 MHz Varian Inova spectrometer (Palo Alto, CA),
`equipped with a broadband inverse detection pulse field gradient probehead.
`
`Potentiometric logK determination
`Electrode calibration
`The four-parameter procedure was used for electrode standardisation in both aqueous and semi-
`aqueous media [18]. HCl solutions of known concentration, containing 0 45.68 wt% methanol, were
`oC, at 0.15 M ionic strength using KCl, under an N2
`titrated with standardis
`atmosphere, in the pH interval of 1.8-12.2. The operational pH reading was related to pcH values by the
`standard multiparametric equation:
`
`pH = a + SpcH + jH[H+] + jOHKw/[H+]
`
`
`
`
`
`
`
`
`
`
`
`
`
` (1)
`
`where a corresponds to the negative logarithm of the activity coefficient of [H+] at working temperature
`and ionic strength and S is the ratio between the electrode slope and the Nernst slope. The jH and jOH terms
`correct the electrode junction effects at low and high pH, respectively.
`The parameters were determined by a weighted non-linear least squares procedure (Refinement ProTM
`2.2 software - Sirius Analytical Instr. Ltd. Forest Row, UK).
`
`Titration in aqueous medium
`Six millilitres of 1.97 2.10 mM aqueous solutions of the sample were preacidified to pH 1.8 with 0.5 M
`HCl and then
`C, at 0.15 M
`ionic strength using KCl under an N2 atmosphere. Three separate titrations were performed. The initial
`
`8
`
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 3 of 11
`
`

`

`ADMET & DMPK 1(2) (2013) 6-16
`
`Lisinopril speciation
`
`estimates of log K values were obtained from Bjerrum difference plots ( n vs. pH, when
` is the average
`number of bound protons) and were refined by a weighted non-linear least squares procedure using
`Refinement ProTM 2.2 software (Sirius Analytical Instr. Ltd. Forest Row, UK).
`
`Titration in solvent mixtures
`Three semi-aqueous solutions of lisinopril containing 14.21, 28.54 and 43.82 wt% methanol were
`titrated under the same experimental conditions as in aqueous medium. The apparent protonation
`constants (logs K) were calculated from the difference (Bjerrum) plot in a similar manner as the aqueous
`log K values. The Yasuda-Shedlovsky procedure was applied to estimate the aqueous log K values. The
`Yasuda-Shedlovsky extrapolation method is based on the linear relation between logs K and the dielectric
`constant ( ) of a solvent mixture:
`
`logs K + log [H2O] = a /
`
` + b
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
` (2)
`
` is the dielectric
`where a is the slope, b is the intercept (fitting constant) of the linear equation,
`constant of the methanol-water mixture and log [H2O] is the molar water concentration of the given
`solvent mixture. This method is a widely used procedure in co-solvent techniques [19,20].
`
`1H NMR titration with in situ pH monitoring
`A single NMR sample solution of 0.6 ml was prepared containing 8 mM lisinopril, 140 mM NaCl (to
`ensure a constant ionic strength of 0.15 M) in a 9/1 H2O/D2O solvent mixture. This sample also contained
`2 mM of the following pH indicator molecules: dichloroacetic acid, chloroacetic acid, acetic acid, imidazole,
`tris(hydroxymethyl)-aminomethane (TRIS) and trimethylamine hydrochloride (TMA) in order to determine
`the actual pH of the sample in each titration step, according to the electrodeless single tube NMR titration
`method [21]. As an
`internal chemical shift reference, 0.5 mM of sodium 3-(trimethylsilyl)-1-
`propanesulfonate (DSS) was applied. The 1H NMR spectra were recorded at 25.0
` 0.1 C. The water signal
`was suppressed either by the double pulse field gradient spin echo (dpfgse) or the selective presaturation
`(presat) sequence. The in situ pH value was deduced from the chemical shift of the appropriate indicator by
`the following equation:
`
`pH
`
`log
`
`K
`
`Ind
`
`log
`
`obs
`Ind
`
`Ind
`
`HInd
`obs
`Ind
`
` (3)
`
`where log KInd is the protonation constant of the indicator and
`Ind, Hind are the limiting chemical shifts
`of its non-protonated and protonated forms (determined in separate experiments) [22,23].
`
`The NMR-pH datasets were fitted with the Opium computer program [24] to determine protonation
`macroconstants and chemical shift values specific to each macrospecies (dHiLis).
`
`Results and Discussion
`
`Among the ACE inhibitor drugs, lisinopril has the most interesting and complex acid-base chemistry.
`There are four proton-binding sites in the molecule: two carboxylates as well as a primary and a secondary
`
`9
`doi: 10.5599/admet.1.2.3
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 4 of 11
`
`

`

`ADMET & DMPK 1(2) (2013) 6-16
`
`amine group. The protonation processes of the two carboxylate groups are expected to be highly
`overlapping. For the investigation of the ionisation/protonation properties of the molecule, potentiometry
`in aqueous medium and in methanol-water mixtures as well as NMR-pH titration were applied.
`
`Protonation macroconstants
`The stepwise protonation equilibria (Fig. 2) of the whole molecule are characterised in terms of
`protonation macroconstants, log K values.
`
`Figure 2. Stepwise protonation equilibria of lisinopril
`
`Generally, potentiometry in aqueous medium is the method of choice for logK determination for
`molecules with solubility greater than 0.8 mM in the entire pH interval of interest. The solubility of
`lisinopril (0.22 M, [9]) allows the determination of macroconstants by the standard method. The four
`obtained log K values along with the standard deviations calculated from three parallel titrations (3 x 15
`points) are listed in Table 1.
`
`Since the log K4 value falls below to the lower applicability limit (log K < 2) of pH-metric titration, we
`also measured the protonation macroconstants using the co-solvent method. The effect of methanol on
`the protonation constant is known to depend on the charge of the basic site. For acids, the apparent logs K
`value increases with the increasing weight percent of methanol, while a decrease is usually observed for
`bases [25]. Using this co-solvent method, we obtained a more reliable value for logK4 since the logs K4
`values in methanol-water mixtures shifted up to the measurable pH range. At the same time, this co-
`solvent method allows for assigning logK values to the acidic and basic moieties of the molecule. According
`to the slopes of the regression lines on the Yasuda-Shedlovsky plot (Fig. 3), log K1 and log K2 characterise
`amine functions, while log K3 and log K4 the carboxylate groups of the molecule, respectively.
`For the exact proton speciation of lisinopril, 1H NMR-pH titrations were carried out using the most
`similar possible experimental conditions as in potentiometry. Fig. 4 shows the aliphatic part of one
`spectrum from the titration series.
`
`10
`
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 5 of 11
`
`

`

`ADMET & DMPK 1(2) (2013) 6-16
`
`Lisinopril speciation
`
`Figure 3. Yasuda-Shedlovsky plot where open diamonds denote logs K1 (slope: -138.4, intercept: 14.271), open
`squares denote logs K2 (slope: -44.6, intercept: 9.532), filled diamonds denote logs K3 (slope: 82.6, intercept: 3.891) and
`filled squares denote logs K4 (slope: 29.9, intercept: 2.976).
`
`Figure 4. Representative 1H NMR spectrum of lisinopril at pH 5.07 in 9/1 H2O/D2O. The assignment is based on
`conventional 2D NMR techniques.
`
`At certain pH values, the assignment of overlapping multiplets to methylene and methine protons was
`verified by TOCSY and HSQC experiments. In Fig. 5, the chemical
`
`datasets were fitted simultaneously to the tetraprotic macroscopic model function described elsewhere
`[26]. The obtained macroconstants are shown in Table 1.
`
`11
`doi: 10.5599/admet.1.2.3
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 6 of 11
`
`

`

`Method
`
`Table 1. Protonation macroconstants measured by different methods.
`Protonation macroconstants
`
`log K1
`
`
`
`log K2
`
`
`
`log K3
`
`
`
`log K4
`
`ADMET & DMPK 1(2) (2013) 6-16
`
`potentiometry in aqueous medium
`potentiometry in solvent mixtures
`NMR-pH titration
`
`average of the best two values*
`* see details in text
`
`Figure 5. NMR-pH titration curves with computer fitted solid line
`
`The macroconstants (Table 1) obtained by different methods show generally good agreement. The
`average deviation between the log K values was 0.09 units. The ca. 0.2 unit discrepancy in log K1 between
`data acquired by the NMR and potentiometric methods may arise from the greater ambiguity of in situ pH
`values. Specifically, trimethylamine has a log KInd = 9.90 [23] and thus can monitor pH values higher than
`
`12
`
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 7 of 11
`
`

`

`ADMET & DMPK 1(2) (2013) 6-16
`
`Lisinopril speciation
`
`10.9 with a lower precision (see the error estimations in [21] for details). In contrast, the log K4 = 1.63 value
`by NMR can be considered as more reliable than data obtained by potentiometry in aqueous medium,
`because below pH 2, the acidity error of the electrode is considerable, while here the NMR-pH titration as
`described above is an electrodeless technique. This constant could be determined with excellent precision
`from the extrapolation of semi-aqueous logs K values due to the reasons mentioned above.
`
`However, we think it reasonable to make an average of the best two experimentally measured logK
`values based on the following principle. We omit the log K value with the highest uncertainty (the highest
`SD value). The calculated, most reliable protonation macroconstants are indicated in the last row of Table
`1. Comparing the obtained logK values with those reported in [9], the difference was greater than 0.4 log
`units in all log K values. Both the log K1 determined by Benion et al. [8] and the log K4 reported by Gonzalez
`et al. [10] differ remarkably from our data. This is not surprising because those constants were deduced
`from conventional pH-potentiometry in aqueous medium, while the methods used in our study
`(potentiometry in methanol-water mixtures, 1H NMR-titration) give more reliable protonation constants at
`pH extremes.
`
`Protonation microconstants
`Nevertheless, macroscopic constants could not directly be assigned to functional groups, since the
`difference in their logarithms did not exceed 3. The microscopic protonation scheme in Fig. 6 should be
`considered for lisinopril instead, where the superscript on microconstant k indicates the group protonating
`in the equilibrium in question, whereas the subscript (if any) refers to already protonated group(s).
`
`Figure 6. Site-specific (microscopic) protonation scheme of lisinopril
`
`In the first two well-separated protonation steps, the macro- and microconstants are identical. Since
`, this protonation step can be assigned to the
`primary amino group (log K1 = log kA). Similarly, log K2 (= log kB
`A) accounts for the secondary amine basicity.
`In the case of lisinopril, the carboxylates denoted by C and D exhibit overlapping protonation, so
`selective monitoring of at least one of them is a prerequisite for microconstant determination. According
`to the normalis
`as selective sensors of D
`carboxylate protonation. Thus, the experimental chemical shift profile of
`model equation:
`
`13
`doi: 10.5599/admet.1.2.3
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 8 of 11
`
`

`

`ADMET & DMPK 1(2) (2013) 6-16
`
` (4)
`
`The microscopic evaluation resulted in log
`the remaining three microconstants, using the following Hessian constraints:
`
`K
`k
`
`CA
`
`B
`
` (5)
` (6)
`
`KK
`3
`
`4
`
`DA
`
`BC
`
`3
`k
`
`CA
`
`B
`
`k
`
`DA
`
`B
`
`CA
`
`BD
`
`k
`
`DA
`
`B
`
`k
`k
`
`We obtained log
`
` = 2.63, log
`
` = 3.10 and log
`
`= 1.68 (with estimated uncertainties of 0.05).
`
`The main pathway of protonation includes the ABC microspecies, while its minor protonation isomer
`(ABD) has nine times lower abundance at all pH values. In other words, the carboxylate on the proline ring
`exhibits nine times greater intrinsic basicity than the carboxylate on the side chain with an adjacent,
`+-) group. The mutual basicity-modifying effect of
`electron-attracting protonated secondary amine (-NH2
`these moieties is quantified by the interactivity parameter, defined as the difference in the corresponding
` 2.63 = 0.47. This pE value suggests an interaction exceeding the
`microconstants: pECD = 2.15
` 1.68 = 3.10
`random, Coulombic value of 0.2-0.3 [27]; presumably, these sites communicate through space with a
`simultaneous change in their geometric positions. A conformational change at pH < 4.5 is also
`substantiated by the changing multiplet patterns in the NMR spectra, but a detailed analysis of vicinal 3JHH
`values holding conformational information is beyond the scope of this article.
`
`Distribution of protonation species
`Using log K and log k values, the percentage of various protonated species can be calculated for any
`arbitrary pH value. The distribution ratios of each species as a function of pH are shown in Fig. 7.
`
`Figure 7. Distribution curves of the macrospecies (HiLis) and microspecies (ABC and ABD) of lisinopril
`
`14
`
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 9 of 11
`
`

`

`ADMET & DMPK 1(2) (2013) 6-16
`
`Lisinopril speciation
`
`Table 2 summarises the relative concentration of lisinopril protonation species at the most relevant pH
`values in the body. In the stomach, the monocation (H3Lis+) and the dication (H4Lis2+) forms are present at
`about equal levels. In the gastrointestinal tract, the double protonated H2Lis form predominates, while at
`the pH of plasma, the monoanion (HLis-) is the dominant species.
`
`Table 2. The percentage concentration of the species in the stomach (pH 1.5), gastrointestinal tract (jejunum pH 6.5)
`and plasma (pH 7.4).
`Percentage concentration
`
`Form of the species
`
`stomach
`pH 1.5
`
`0.00
`0.00
`0.99
`42.15
`56.86
`
`GI tract (jejunum)
`pH 6.5
`
`plasma
`pH 7.4
`
`0.00
`18.98
`80.98
`0.03
`0.00
`
`0.03
`65.04
`34.93
`0.00
`0.00
`
`Lis2-
`HLis-
`H2Lis
`H3Lis+
`H4Lis2+
`
`Conclusions
`
`In this study, the acid-base chemistry of tetraprotic lisinopril was characterised by protonation macro-
`and microconstants. The highly precise log K values obtained by two independent analytical methods
`provided a better interpretation of the PK/PD properties. As Fig. 7 indicates, under the pH conditions of the
`GI tract
` the most likely site of oral absorption
` lisinopril is predominantly present in the H2Lis form.
`Although the net charge of this species is zero, the molecule exists in solution as a double zwitterion: the
`two amine functions are protonated while the two carboxylates are not. This structure represents a highly
`polar dipole that is unfavourable for passive transport through lipoid membranes, which explains the low
`bioavailability of the molecule. Recently, the intestinal dipeptide transporter system (DTS) has been
`reported to be involved in the active transport mechanism of oral absorption of ACE inhibitors [28].
`At the pH of the plasma, the monoanion (HLis-) is the dominant form (more than 60%), which favours
`receptor binding. As is known, therapeutically useful ACE inhibitors exhibit three-point binding to the Zn
`ion-containing carbopeptidase ACE enzyme [2]. The proline C2-carboxylate of lisinopril binds to the
`positively charged Arg146, while the other carboxylate in the side chain participates in complex formation
`with the Zn ion. The third binding site is represented by Ser-OH where the C =O group forms an H-bond.
`The species distribution revealed in this study explains the good receptor binding of lisinopril.
`
`References
`J. Wong, R.A. Patel, P.R. Kowey, Progress in Cardiovascular Diseases 47 (2004) 116-130.
`[1]
`[2] M. Harrold, in: D.A. Williams, T.L. Lemke (Eds),
`, 5th edn.
`Lippincott Williams & Wilkins, Philadelphia, 2002, p. 533.
`B. Beermann, American Journal of Medicine 85 (3B) (1988) 25-30.
`[3]
`D.J. Eveson, T.G. Robinson, J.F. Potter, American Journal of Hypertension 20 (2007) 270-277.
`[4]
`[5] N.K. Hollenberg, D.A. Anzalone, B. Falkner, N.D.L. Fisher, P.N. Hopkins, W. Hsueh, H. Hutchinson,
`R.M. Krauss, D.A. Price, P. Raskin, G.M. Reaven, American Journal of Hypertension 14 (2001) 218-
`223.
`
`15
`doi: 10.5599/admet.1.2.3
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 10 of 11
`
`

`

`ADMET & DMPK 1(2) (2013) 6-16
`
`[6]
`[7]
`[8]
`
`[9]
`[10]
`
`C.M. White, Pharmacotherapy 18 (1998) 588-599.
`K.L. Goa, J.A. Balfour, G. Zuanetti, Drugs 52 (1996) 564-588.
`C. Bennion, R.C. Brown, A.R. Cook, C.N. Manners, D.W. Payling, D.H. Robinson, Journal of Medicinal
`Chemistry 34 (1991) 439-447.
`D.P. Ip, J.D. DeMarco, M.A. Brooks, Analytical Profiles of Drug Substances 21 (1992) 233-276.
`E.B. Gonzalez, E. Farkas, A.A. Soudi, T. Tan, A.I. Yanovsky, K.B. Nolan, Journal of Chemical Society
`Dalton Transactions (1997) 2377-2379.
`European Pharmacopoeia 7.0, Strasbourg.
`[11]
`[12] N. Rahman, N. Anwar, M. Kashif, Il Farmaco 60 (2005) 605-611.
`C.A. Beasley, J. Shaw, Z. Zhao, R.A. Reed, Journal of Pharmaceutical and Biomedical Analysis 37
`[13]
`(2005) 559-567.
`A. El-Gindy, A. Ashour, L. Abdel-Fattah, M.M. Shabana, Journal of Pharmaceutical and Biomedical
`Analysis 25 (2001) 923-931.
`S. Hillaert, W. Van den Bossche, Journal of Pharmaceutical and Biomedical Analysis 25 (2001) 775-
`783.
`
`[14]
`
`[15]
`
`active systems. Ellis Horwood, Chichester, 1990, p. 18.
`-
`-
`(1995) 431-435.
`A. Avdeef, J.J. Bucher, Analytical Chemistry 50 (1978) 2137-2142.
`International Journal of Pharmaceutics 151 (1997) 235-248.
`-
`Medicinal Chemistry 2 (2006) 385-389.
`Analytica Chimica Acta 522 (2004) 247-258.
`Journal of Medicinal Chemistry 48 (2005)
`
`Journal of Pharmacy and Pharmacology 47
`
`-
`
`249-255.
`
`[16]
`
`[17]
`
`[18]
`[19]
`[20]
`[21]
`[22]
`
`[23]
`
`[25]
`[26]
`[27]
`[28]
`
`Journal of Pharmaceutical and Biomedical Analysis 43
`
`(2007) 1306-1314.
`[24] M. Kyvala, I. Lukes, (1995) OPIUM computer program,
`http://www.natur.cuni.cz/~kyvala/opium.html
`A. Avdeef, J.E.A. Comer, S.J. Thomson, Analytical Chemistry 65 (1993) 42-49.
`Analytical and Bioanalytical Chemistry 378 (2004) 1428-1448.
`Journal of Physical Chemistry B 107 (2003) 5074-5080.
`V.A. Moore, W.J. Irwin, P. Timmins, P.A. Lambert, S. Chong, S.A. Dando, R.A. Morrison,
`International Journal of Pharmaceutics 210 (2000) 29-44.
`
`13 by the authors; licensee IAPC, Zagreb, Croatia. This article is an open-access article distributed under the terms and
`conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/)
`
`16
`
`SILVERGATE PHARMACEUTICALS, INC.Exhibit 2009 Page 11 of 11
`
`