`http://www.elsevier-deutschland.de/etp
`
`1 NV Organon, Toxicology & Drug Disposition, Oss, The Netherlands
`2 University Medical Centre Nijmegen, Department of Pulmonary Diseases, Nijmegen, The Netherlands
`3 TNO Pharma, Biomolecular Sciences, Zeist, The Netherlands
`
`Use of physicochemical calculation of pKa and CLogP
`to predict phospholipidosis-inducing potential
`
`A case study with structurally related piperazines
`
`JAN-PETER H. T. M. PLOEMEN1, JAN KELDER1, THEO HAFMANS2, HAN VAN DE SANDT3, JOHAN A. VAN BURGSTEDEN3,
`PAUL J. M. SALEMINK1, and ERIC VAN ESCH1
`
`With 3 figures and 2 tables
`
`Received: September 5, 2003; Revised: October 23, 2003; Accepted: October 30, 2003
`
`Address for correspondence: JAN-PETER H.T.M. PLOEMEN, NV Organon, Toxicology & Drug Disposition, PO Box 20,
`5340 BH Oss, The Netherlands; Tel.: ++31-412 666 172, Fax: ++31-412 666131, e-mail: jan-peter.ploemen@organon.com
`
`Key words: Foamy macrophages; lung; alveolar histiocystosis; phospholipidosis; electron microscopy; physicochemical
`calculated parameters; Sprague-Dawley rats; gepirone; piperazines.
`
`Summary
`
`Several cationic amphiphilic compounds are known to
`induce phospholipidosis, a condition primarily character-
`ized by excessive accumulation of phospholipids in differ-
`ent cell types, giving the affected cells a finely foamy ap-
`pearance. Excessive storage of lamellar membranous in-
`tralysosomal inclusion bodies is the hallmark for phospho-
`lipidosis on the electron microscopic level. In case of alve-
`olar phospholipidosis, foamy macrophages accumulate
`within the alveolar spaces of the lung. Based on such find-
`ings in a one-year toxicity study with gepirone in rats, we
`studied the molecular properties of this compound and
`compounds suspected of being phospholipidosis inducers
`by means of physicochemical calculations. Physicochemi-
`cal molecular calculations of molecular weight, ClogP
`(partition coefficient octanol/water), logD at pH 7.4, and
`pKa were performed, for the cationic amphiphilic com-
`pounds chlorpromazine, amiodarone, imipramine, propra-
`nolol and fluoxetine, and for the structurally related com-
`pounds 1-phenylpiperazine (1-PHP), gepirone (and its
`major metabolites, 3-OH-gepirone and 1-pyrimidinylpiper-
`azine [1-PP]), and buspirone. ClogP and calculated pKa
`cluster differently for the amphiphilic drugs compared to
`the chemical series of piperazines. In line with this analy-
`sis, lamellar inclusion bodies were found in an in vitro vali-
`dation experiment in the human monoblastoid cell line U-
`937, incubated for 96 h at 10 µg/mL with cationic am-
`phiphilic drugs (amiodarone, imipramine, or propranolol).
`No such lamellar inclusion bodies were seen for any of the
`compounds from the chemical series of piperazines includ-
`ing gepirone and its metabolites. The data presented sup-
`
`port the use of simple physicochemical calculations of
`ClogP and pKa to discriminate rapidly between compounds
`suspected of being phospholipidosis inducers. Finally, the
`discriminative power of these physicochemical ClogP and
`pKa calculations to predict phospholipidosis-inducing po-
`tential was further validated by extension of the set of com-
`pounds.
`
`Introduction
`
`Phospholipidosis is of concern for the pharmaceutical
`industry, since a candidate pharmaceutical agent may be
`rejected because of evidence of phospholipidosis in a
`preclinical animal study. Phospholipidosis is widely dis-
`cussed and reported in rats (HALLIWEL 1997; GOPINATH
`et al. 1987; HRUBAN 1976; LÜLLMANN et al. 1975; REA-
`SOR 1989) and is characterized by the accumulation of
`phospholipids in the lysosomes of many cell types. Alve-
`olar macrophages are especially susceptible to these
`changes, but other cell types may be affected as well, in-
`cluding lymphoid cells, hepatocytes, pancreatic cells,
`and cells within endocrine tissue, nervous system, mus-
`cle, and eyes. Phospholipidosis can be induced by the
`systemic administration of cationic amphiphilic drugs,
`like amiodarone, imipramine, or propranolol.
`Alveolar histiocytosis is a common, spontaneous, in-
`cidental finding in older rats and consists of aggregates
`of foamy (lipid-containing) macrophages in the lumen of
`alveoli (BEAVER et al. 1963, BOORMAN and EUSTIS 1990;
`
`0940-2993/04/55/05-347 $ 15.00/0
`
`347
`
`1 of 9
`
`PENN EX. 2037
`CFAD V. UPENN
`IPR2015-01836
`
`
`
`Yang et al. 1966). The histopathological characteristics
`of this lesion have been discussed in many papers. Har-
`monization of nomenclature would be very welcome in
`this area, since many inconsistent morphologic descrip-
`tions, such as lipoproteinosis, pulmonary lipidosis and
`others are found in the literature (BOORMAN and EUSTIS
`1990; DUNGWORTH 1985; GOPINATH et al. 1987; KODA-
`VANTI and MEHENDALE 1990). Alveolar phospholipidosis
`and alveolar histocytosis are both associated with the ac-
`cumulation of intra-alveolar material, foamy macro-
`phages and type II pneumocyte proliferation. The hall-
`mark of phospholipidosis, multi-lamellar inclusion bod-
`ies (‘whorls’, ‘myelin figures’), is easily detected using
`electron microscopy.
`A one-year oral toxicology study was performed with
`gepirone, a new antidepressant presently under clinical
`development. Compared to the control group, an in-
`creased incidence of minimal to slight aggregations of
`foamy macrophages was observed: i.e. 1 animal in the
`control group, 1 animal in the low-dose (LD) group, 2
`animals in the mid-dose (MD) group and 10 animals in
`the high-dose (HD) group. The lesion was evaluated in
`the context of the well-known, age-related spontaneous
`change in the rat, and judged as being of no relevance
`given that the increase in incidence occurred only at high
`multiples of the human dose. Subsequently, we per-
`formed a closer analysis of the physicochemical proper-
`ties of gepirone due to the possibility that the higher inci-
`dence of alveolar foamy macrophages may be drug relat-
`ed. Physicochemical parameters of gepirone and its
`major metabolites were calculated and compared with
`well-known cationic amphiphilic drugs as a method for
`predicting phospholipidosis-inducing potential. More-
`over, an in vitro study with human monoblastoid cell line
`U937 was performed to validate this method, using elec-
`tron microscopic analysis.
`
`Materials and methods
`
`Reagents and media: RPMI 1640 culture medium,
`phosphate-buffered saline, Fetal Calf Serum (FCS) and
`penicillin/streptomycin were purchased from Invitrogen
`Corporation (UK). Chlorpromazine, amiodarone, imipr-
`amine, propranolol, 1-pyrimidinylpiperazine (1-PP) and 1-
`phenylpiperazine HCl (1-PHP) were obtained from from
`Sigma-Aldrich Chemical Corporation Ltd (Gillingham,
`UK); fluoxetine and buspirone from Tocris (Avonmouth,
`UK); and gepirone was obtained from Sidmak Laboratories
`Inc. (East Hanover, NJ, USA). 3-OH-gepirone was synthe-
`sized at Organon (Newhouse, Scotland) and was of >98%
`purity. All other chemicals were of analytical grade and
`were obtained from common commercial sources.
`
`Animal experiment: A study was conducted to deter-
`mine the toxicity of gepirone in Sprague-Dawley rats
`(weighing 147 to 215 g for the males and 117 to 176 g for
`the females and approximately 6 weeks at the start of dos-
`ing) when administered orally in the diet for 1 year. The
`study was conducted in 1987–1988 and was performed in
`
`348
`
`Exp Toxic Pathol 55 (2004) 5
`
`compliance with OECD Principles of Good Laboratory
`Practice. Sprague-Dawley rats (22/sex/group) received 0, 4
`(low-dose [LD]), 12 (mid-dose [MD], or 36 mg/kg/day
`(high-dose [HD]) in the diet. Each animal was individually
`housed in same humidity (40–60%) and controlled temper-
`ature 23 °C with a 12 hour light-dark cycle. A commercial
`diet (Purina Rodent Laboratory Chow #5001, supplied by
`Ralston Purina Co.) and tap water were available ad libi-
`tum. At week 19, the dose levels of 12 and 36 mg/kg/day
`were increased to 16 and 48 mg/kg/day in both male and fe-
`male groups in order to achieve drug-related suppression of
`body weight gain or reduction in food intake. Mortality,
`clinical observations, body weights, and food and water
`consumption were recorded. Ophthalmoscopy was per-
`formed before dosing and during weeks 12, 26, and 52.
`Clinical (serum) chemistry determinations consisted of
`sodium, potassium, chloride, total protein, albumin, glu-
`cose, urea nitrogen, total cholesterol, alkaline phosphatase,
`alanine transaminase, total bilirubin, calcium, phosphorus,
`creatinine, creatine kinase, uric acid, triglycerides, aspar-
`tate transaminase, and lactate dehydrogenase at week 53
`from the first 10 surviving rats (approximately fasted 18
`hours) in each group. Hematology parameters (hematocrit,
`total hemoglobin, erythrocyte, reticulocyte, platelet, leuko-
`cyte, and leukocyte differential counts, plasma prothrom-
`bin and activated partial thromboplastin time) were mea-
`sured at weeks 13, 27, and 52 from the first 10 surviving
`(nonfasted) rats in each group. Urinalysis samples (deter-
`minations made were gross appearance, volume, specific
`gravity, pH, protein, glucose, blood, bilirubin, urobilino-
`gen, ketones, and microscopic findings) were analyzed at
`weeks 13, 27, and 51 on the first 5 animals in each group.
`Necropsy was performed on rats that died or were eutha-
`nized during the dosing period and on rats sacrificed at the
`end of the one-year treatment period. At necropsy, rats were
`examined macroscopically and selected organs were
`weighed. A complete histopathologic examination was per-
`formed on each animal. Lungs were fixed in buffered 10%
`formalin, paraffin embedded, and sections stained with
`hematoxylin and eosin. Below, we describe only issues rel-
`evant to the pulmonary foamy macrophages.
`
`Molecular calculations: The static polar surface is cal-
`culated from the Corina-built structures using the in-house
`developed program called Monika (KELDER et al. 1999).
`Monika was also used to calculate molecular weight. The
`ClogP (partition coefficient octanol/water) was calculated
`with ClogP4.0 from BioByte Corporation, Claremont, CA,
`USA. LogD at pH 7.4 and pKa were calculated with
`ACD/Labs PhysChem Batch program release 7.0 (Ad-
`vanced Chemistry Development, Inc., Toronto, Canada).
`For dibasic compounds, the lowest basic pKa >8 was used,
`and if not available, the highest pKa <8 was used.
`Gepirone, 3-OH-gepirone, 1-PP and buspirone belong to
`the same chemical series, as judged from the molecular
`structures. Calculation of ClogP and pKa were extended
`with a larger series of compounds known as ‘positive’
`phospholipidosis inducers or negative controls.
`
`Cell culture and drug incubations: A slightly adapted
`method as previously described by Xia et al. (1997) was
`used. The human monoblastoid cell line U-937 was cul-
`tured in RPMI 1640 culture medium containing heat-inacti-
`vated FCS (10%) penicillin (100 IU/mL) and streptomycin
`
`2 of 9
`
`PENN EX. 2037
`CFAD V. UPENN
`IPR2015-01836
`
`
`
`(100 µg/mL) in 75 cm2 flasks in a humidified incubator
`(37 ± 1 °C, 5% CO2). Stock solutions of the drugs
`(1 mg/mL) were made in sterile phosphate-buffered saline.
`The cells (5 × 104/mL) were transferred from the flasks to
`24 well plates (1 mL/well), and 10 µL of the stock solutions
`was added to the wells. All incubations were performed in
`duplicate. After a 48 h incubation, 0.5 mL of the culture
`medium (containing half of the cells) was removed from
`each well and 0.5 mL fresh medium was added. The num-
`ber of cells was determined using a Bürker-Türk hemocy-
`tometer. After 96 h of incubation, the remaining cells were
`collected, counted and fixed for electron microscopic eval-
`uation.
`
`Sample preparation for electron microscopy: The
`cells were transferred to Eppendorf vials and washed two
`times in 0.15 M cacodylate buffer, pH 7.3 (4 °C). Sub-
`sequently, the cells were fixed in 1.5% glutaraldehyde in
`0.15 M cacodylate buffer, pH 7.3 (4 °C) and stored at 4 °C.
`After fixation, cells were centrifuged and embedded in 1%
`agarose, rinsed in 0.1M phosphate buffer, osmicated for
`one hour in 1% osmiumtetroxide and rinsed again in phos-
`phate buffer. Next, sections were dehydrated in a grade se-
`ries of ethanol, embedded in epon 812 via propylene oxide.
`After polymerization, a Reichert Ultracut-E was used to cut
`80 nm thick sections. The sections were examined with a
`Jeol 1010 electron microscope. Two independent observers
`examined and photographed only those cells that were sec-
`tioned through the region of the Golgi area and centrioles.
`
`Results
`
`Animal experiment
`Dietary administration of gepirone to rats affected
`body weight in HD males and MD and HD females.
`Overall mean body weight gains for the male HD group
`
`and the female MD and HD groups were about 20–25%
`below that of the control group. Food consumption was
`increased in HD female rats. Focal pale areas were ob-
`served macroscopically in the lung in 1 control group
`male, 1 MD group female, 2 HD group males and 6 HD
`group females. With histopathological evaluation, mini-
`mal peribronchiolar lymphoid cellular infiltrates were
`located in the lungs from 32 control group, 34 LD
`group, 38 MD group, and 36 HD group rats. This pul-
`monary finding was considered to be a spontaneous or
`incidental observation. Histopathology revealed a
`slightly increased incidence of alveolar foamy macro-
`phages in HD group males and females. Minimal to
`slight aggregations of foamy macrophages were ob-
`served in the alveolar lumens of 1 animal in the control
`group, 1 animal in the LD group, 2 animals in the MD
`group and 10 animals in the HD group. There was no
`relevant gender effect. In general, at the higher dose lev-
`els, the severity of the lesion did not increase but the
`foamy macrophages were more widespread throughout
`the lung. The body weight of animals with the lesion
`was, on average, approximately 24% lower than the
`body weight of control group animals of the same sex.
`Interestingly, animals with the foamy macrophages also
`weighed less than their counterparts (other animals of
`the same dose group not having foamy macrophages).
`This is illustrated by the fact that the animals with
`foamy macrophages had an average body weight ap-
`proximately 13% lower than the average body weight of
`the corresponding group.
`Comparison of the clinical chemistry, hematology,
`and urinalysis data with concurrent control values re-
`vealed several statistically significant differences which
`were minor, were within historical range, and were con-
`cluded to be of no toxicological significance.
`
`Fig. 1. Relation of Calculated ClogP and pKa
`of 10 drug molecules.
`
`Exp Toxic Pathol 55 (2004) 5
`
`349
`
`3 of 9
`
`PENN EX. 2037
`CFAD V. UPENN
`IPR2015-01836
`
`
`
`Table 1. Calculated physicochemical properties of 10 drug molecules.
`
`350
`
`Exp Toxic Pathol 55 (2004) 5
`
`4 of 9
`
`PENN EX. 2037
`CFAD V. UPENN
`IPR2015-01836
`
`
`
`Molecular calculations
`Physicochemical molecular calculations are summa-
`rized in table 1. With trend analysis, it appeared that a
`distinct cluster
`for amiodarone, chlorpromazine,
`imipramine, fluoxetine, and more or less propranolol
`was observed: cationic amphiphilic compounds combine
`a relatively high pKa with a high ClogP (fig. 1). The
`highest values of calculated pKa and ClogP were found
`as follows: amiodarone > chlorpromazine ~ imipramine
`~ fluoxetine > propranolol. Unlike the series of cationic
`amphiphilic drugs, the chemical series of the structurally
`related piperazines showed different chemical proper-
`ties. In the plot of the calculated ClogP versus pKa,
`gepirone, 3-OH-gepirone, and buspirone cluster differ-
`
`ently than the cationic amphiphilic compounds (fig. 1).
`The 1-PP compound also had significantly lower pKa
`and ClogP values, and as predicted a priori, 1-PHP was
`closest to the predicted weakest amphiphilic compound
`(propranolol). To illustrate the discriminative power of
`the calculation of ClogP and pKa to differentiate phos-
`pholipidosis-inducing capacity for cationic amphiphilic
`compounds from negative controls, the calculations were
`extended with several known “negative controls” (i.e.
`diazepam, clozapine, 5-phenoxybenzamine, ketanserin,
`almitrine, haloperidol, bufetolol) and extra “positive
`controls” (fig. 2). Positive controls were divided in three
`categories i) positive controls with low phospholipido-
`sis-inducing potency in animals, but pronounced potency
`to induce phospholipidosis in cultured cells: mianserin;
`
`Fig. 2. Relation of Calculated ClogP and pKa of compounds positively or negatively associated with phospholipidosis in-
`ducing capacity. A-series; negative controls: A1, diazepam; A2, 3-OH-gepirone; A3, gepirone; A4, buspirone; A5, clozap-
`ine; A6, 5-phenoxybenzamine; A7, ketanserin; A8, 1-PP; A9, almitrine; A10, 1-PHP; A11, haloperidol; A12, bufetolol.
`B-series; positive controls with low phospholipidosis-inducing potency in animals, but pronounced potency to induce
`phospholipidosis in cultured cells: B1, mianserin; B2, propranolol; B3, clociguanil; B4, noxiptiline; B5, amitriptyline;
`B6, disobutamide; B7, promazine; B8, mesoridazine; B9, nortriptyline; B10, chlorpromazine; B11, maprotiline; B12,
`thioridazine. C-series; positive controls with phospholipidosis demonstrated in animals: C1, chlorcyclizine; C2, citalo-
`pram; C3, chlorphentermine; C4, phentermine; C5, fenfluramine; C6, imipramine; C7, tilorone; C8, fluoxetine; C9, tam-
`oxifen; C10, iprindole; C11, clomipramine; C12, triparanol; C13, mepacrine. D-series; positive controls with phospholipi-
`dosis demonstrated in animals and in humans: D1, chloroquine; D2, amiodarone; D3, perhexiline; D4, desethylamio-
`darone. For dibasic compounds, the lowest basic pKa > 8 was used, and if not available, the highest pKa <8 was used.
`For compounds, chlorcyclizine (C1), mianserin (B1) and propranolol (B2) the reader is refered to the Discussion section.
`References are given in the Result section.
`
`Exp Toxic Pathol 55 (2004) 5
`
`351
`
`5 of 9
`
`PENN EX. 2037
`CFAD V. UPENN
`IPR2015-01836
`
`
`
`propranolol; clociguanil; noxiptiline; amitriptyline;
`disobutamide; promazine; mesoridazine; nortriptyline;
`chlorpromazine; maprotiline; thioridazine (LÜLLMANN
`et al. 1975; LÜLLMANN-RAUCH 1979; KODAVANTI and
`MEHENDALE 1990); ii) positive controls with phospho-
`lipidosis demonstrated in animals: chlorcyclizine; citalo-
`pram; chlorphentermine; phentermine; fenfluramine;
`imipramine; tilorone; fluoxetine; tamoxifen; iprindole;
`clomipramine; triparanol; mepacrine (LÜLLMANN et al.
`1975; LÜLLMANN-RAUCH 1979; LÜLLMANN-RAUCH and
`NASSBERGER 1983, HEIN and LÜLLMANN 1989; BENDELE
`et al. 1992), and iii) positive controls with phospholipi-
`dosis demonstrated in animals and in humans: chloro-
`quine; amiodarone; perhexiline; and desethylamiodarone
`(LÜLLMANN-RAUCH 1979, REASOR 1991). Propranolol,
`chlorcyclizine and mianserin seem to cluster close to the
`range of the negative controls, albeit at the borderline.
`All other positive controls cluster separately from the
`negative controls, with the trend that the compounds
`which are associated with pronounced potency to induce
`phospholipidosis (i.e. phospholipidosis demonstrated in
`animals and humans: fig. 2) cluster relatively far from
`the negative controls.
`
`Cell studies
`Incubation concentration range-finding: In a pilot
`study, the human monoblastoid cell line U-937 was incu-
`bated with 10 µg/mL gepirone, 1-PP, imipramine and
`propranolol for 96 h. No major cytotoxic effect was ob-
`served as judged from the cell count and morphology
`with electron microscopic evaluation (data not shown).
`This relatively high concentration was used to ensure
`that within 96 h, laminar inclusion bodies could be
`found. With the positive control (imipramine), laminar
`inclusion bodies were indeed observed in the pilot exper-
`iment (data not shown). Therefore, it was decided to per-
`form the final validation experiment with the full set of
`compounds at one experimental setting to allow compar-
`ison: 10 µg/mL for 96 h.
`
`Cell count: The cell count (table 2) indicated that
`marked cytotoxic effects leading to significant cell death
`or delayed growth were seen for chlorpromazine and flu-
`oxetine at 10 µg/mL for 48 and 96 h. For the other com-
`pounds, no marked effects on the cell count were ob-
`served as compared to the control values.
`
`Electron microscopy: The general morphological
`features (fig. 3) of the cells were described for the con-
`trol cells: many intact cells as well as cells in several mi-
`totic phases were observed. In general, two types of cells
`
`Fig. 3. Electron micrograph of human monoblastoid cell
`line U-937 incubated for 96 hours with A) solvent, B) 10
`µg/mL of 3-OH-gepirone C) 10 µg/mL amiodarone.
`(×10,000).
`
`352
`
`Exp Toxic Pathol 55 (2004) 5
`
`6 of 9
`
`PENN EX. 2037
`CFAD V. UPENN
`IPR2015-01836
`
`
`
`Table 2. Mean cell counta per mL (×104) of human mono-
`blastoid cell line U-937 incubated with 10 mg/mL test com-
`pound.
`
`Time (h)
`–––––––––––––––––––––––––––
`48
`96
`
`Control
`Gepirone
`3-OH-gepirone
`1-PP
`Buspirone
`1-Phenylpiperazine HCl
`Amiodarone
`Chlorpromazine
`Fluoxetine
`Imipramine
`Propranolol
`
`37 ± 2.6
`43 ± 3.3
`35 ± 3.8
`36 ± 0.7
`38 ± 8.0
`44 ± 0.7
`14 ± 2.6
`4.8 ± 1.6
`7.2 ± 0.2
`32 ± 1.4
`34 ± 2.1
`
`147 ± 1.4
`44 ± 28
`135 ± 17
`174 ± 13
`121 ± 18
`133 ± 32
`31 ± 2.6
`3.7 ± 0.9
`3.3 ± 0.5
`79 ± 2.8
`86 ± 12
`
`a ± standard deviation; Incubations were performed in
`duplicate. After 48 and 96 h of incubation, the number of
`the cells was determined.
`
`were observed, one electron-grey cell type with a nucle-
`us:cytoplasm ratio of about 0.5 to 1.0; and another elec-
`tron-light cell type with a nucleus:cytoplasm ratio of
`about 0.3 to 0.5. The electron-grey cells displayed a large
`indented nucleus with nucleoli. The cytoplasm was
`packed with free ribosomes and polysomes, with some
`profiles of rough endoplasmic reticulum (RER) in be-
`tween. Scattered and clustered mitochondria were ob-
`served, with a few showing a condensed matrix with
`light-stained cristae. A few Golgi complexes and a single
`centriole were found close to the nucleus. Small elec-
`tron-grey or electron-dense membrane-bound structures
`represented endosomal and lysosomal elements with a
`heterogenic content. Typical autophagic vacuoles were
`also present. A single lamellipodium and smaller filo-
`podia were observed. The electron-light cell type exhib-
`ited a larger and irregularly-shaped nucleus with deeper
`indentations. In the cytoplasm, more and often dilated
`RER was observed. The number of ribosomes decreased
`but the number of lysosomal structures seemed to in-
`crease, while the mitochondria appeared to display the
`more condensed aspect. No other striking differences
`with the electron-grey cells were detectable. These cell
`types might represent different functional stages.
`With gepirone, 3-OH-gepirone, 1-PP, buspirone, and
`1-PHP, no distinct changes were observed (see below) as
`no lamellar inclusions were observed. In line with the
`cell count, most cells appeared dead or were broken
`down with fluoxetine and chlorpromazine at 10 µg/mL.
`For this reason, morphological studies appeared to be
`useless. With amiodarone, more lysosomal structures
`were found in both cell types, with lamellar inclusion
`bodies that appeared to be early-stage (i.e., membrane-
`bound vacuoles containing “whorled membranous struc-
`tures”) and late-stage (i.e., “whorling” membranes with a
`
`condensed central core, and often looking like smaller,
`rounder, and electron-denser structures). For imipramine
`and propranolol, the early-stage lamellar inclusion bod-
`ies were found as described above. Semi-quantitative
`ranking of the distribution lamellar inclusion bodies (low
`to high amount) based on electron microscopic observa-
`tions resulted in the following ranking: amiodarone >>
`imipramine > propranolol. Other (minor)
`induced
`changes included merely some increase in the polymor-
`phy of mitochondria with gepirone, 3-OH-gepirone, 1-PP,
`buspirone, propranolol, and 1-PHP. Distinctly aberrant
`mitochondria were only observed for imipramine.
`
`Discussion
`
`Prediction of the capacity of drugs to induce phospho-
`lipidosis can be performed by calculation or measure-
`ment of their cationic amphiphilic properties (BARTON
`et al. 1997; FISCHER et al. 2001; KODAVANTI and MEHEN-
`DALE 1990; LÜLLMANN-RAUCH 1979). Cationic amphi-
`philic drugs share structural features, including a hydro-
`philic cationic side chain, a primary, secondary or
`tertiary amine, and a hydrophobic region that is usually
`an aromatic ring or ring system. This results in polar
`hydrophilic cationic side chain and apolar ring systems
`within one molecule, which can be illustrated by molecu-
`lar calculations. Sophisticated analysis and calculation of
`the physicochemical interaction of amphiphilic drugs
`with phospholipids have been reported, including quanti-
`tative structure-activity relationships (SEYDEL et al.
`1989; FISCHER et al. 1998). The affinity of cationic
`amphiphilic compounds for phospholipids appears to
`involve many factors next to electrostatic forces and
`hydrophobic binding forces (KODAVANTI and MEHEN-
`DALE 1990; LÜLLMAN et al. 1975). Some cationic amphi-
`philic compounds exhibit strong binding to hydrophobic
`as well as hydrophilic moieties, while others exhibit
`strong hydrophilic and relatively weak hydrophobic
`interactions. Nevertheless, the easy and routinely calcu-
`lated parameters pKa and ClogP proved to be especially
`helpful for predicting whether a compound may have the
`potential to induce phospholipidosis (see below). A rela-
`tively high ClogP (i.e. apolar region, hydrophobic) was
`accompanied by a relatively high calculated pKa (i.e.
`polar region, highly ionized amine) (table 1). The chemi-
`cal series of gepirone did not have these prominent am-
`phiphilic cationic properties. The 1-PHP compound,
`which is not a drug but can be perceived as the template
`of the chemical series to which gepirone belongs, com-
`bines comparable apolarity and relatively high pKa as
`compared to gepirone. As expected, 1-PHP appeared to
`be the most cationic amphiphilic compound within the
`gepirone series (table 1, fig. 1). But even 1-PHP is differ-
`ent from the typical physicochemical properties of a
`cationic amphiphilic drug, and was not capable of induc-
`ing lamellar inclusion bodies in the in vitro system (see
`below).
`
`Exp Toxic Pathol 55 (2004) 5
`
`353
`
`7 of 9
`
`PENN EX. 2037
`CFAD V. UPENN
`IPR2015-01836
`
`
`
`As expected, lamellar inclusion bodies were found in
`human monoblastoid cell line U-937 incubated for 96 h at
`10 µg/mL with amiodarone
`(most pronounced),
`imipramine and propranolol (less pronounced, early
`stage). Fluoxetine and chlorpromazine could unfortunate-
`ly not be evaluated at the same high concentration (10
`µg/mL), due to significant cytotoxic effects in the
`U-937 cell line. At this high concentration, the prediction
`by the physicochemical calculations (see above) could be
`validated: no lamellar inclusion bodies were found with
`gepirone and major metabolites or other piperazines.
`The physicochemical calculations could be extended
`with several negative and positive controls from litera-
`ture. The following “rule of thumb” could be developed:
`an cationic amphiphilic compound can be suspected to
`induce phospholipidosis if (ClogP)^2 + (calculated pKa)^2
`> 90, provided that ClogP > 1 and pKa > 8. Using this
`equation, all negative controls had values lower than 90.
`From the positive controls, only chlorcyclizine (value
`=74) and mianserin (value = 86) had values lower than
`90. Both mianserin (DELBRESSINE et al. 1992) and chlor-
`cyclizine (JACOBSON et al. 1972) have a major N-
`demethylated metabolite which has slightly higher
`phospholipidosis-inducing potential (calculated pKa
`and ClogP, 9.20 and 3.69, and 9.03 and 3.94, for N-de-
`methylated mianserin and N-demethylated chlorcycli-
`zine, respectively). Using the equation for these metabo-
`lites, values of 97 and 98 were found for N-demethylated
`chlorcyclizine and N-demethylated mianserine, respec-
`tively, showing that physicochemical calculation also
`alerts for a – albeit weak – phospholipidosis-inducing
`potential. In general, extensive metabolism of drugs may
`hamper this analysis. Propranolol is a relatively weak in-
`ducer of phospholipidosis in our in vitro cell systems
`(see above), and this has also been observed in lympho-
`cytes in vitro by other groups [personal communication
`Dr. A. TILLOY-ELLUL, Pfizer, France], in accordance with
`its clustering close to the range of the negative controls
`(value with equation: 91). The mechanism of the induc-
`tion of phospholipidosis is complex and multiple factors
`are involved (HALLIWELL 1997). However, the formation
`of a relatively undegradable complex (for phospholipi-
`dase enzyme activity) of the drug with the phospholipids
`of the membrane, is one of the processes most often in-
`volved (LÜLLMAN-RAUCH 1979; HALLIWELL 1997). Pro-
`pranolol has been known to be a very potent direct phos-
`pholipidase A inhibitor, another mechanism which is
`thought to lead to the development of drug-induced
`phospholipidosis (PAPPU et al. 1985). It is therefore
`tempting to speculate that propranolol is a pure phospho-
`lipidase A inhibitor, of which the physicochemical prop-
`erties (LogP, pKa) are of relative unimportance.
`It appears redundant to mention that intra-alveolar
`foamy macrophages are by no means pathognomonic as
`such, since they occur under various experimental and
`pathological conditions (CORRIN and KING 1969; FLODH et
`al. 1974; LÜLLMANN et al. 1975; LÜLLMANN-RAUCH and
`SCHEID 1975; SCHOBER et al. 1974). The disorders which
`
`354
`
`Exp Toxic Pathol 55 (2004) 5
`
`are accompanied by accumulation of foamy macrophages
`are lumped together in ill-defined groups of conditions
`with overlapping morphologic features (BOORMAN and
`EUSTIS 1990; DUNGWORTH 1985; GOPINATH et al. 1987;
`KONISHI and HIGASHIGUCHI 1996). Causes mentioned in-
`clude i) parenteral administration of amphiphilic drugs
`(with effect on lipid degradation or failure of degradation
`of surfactant) ii) alveolar macrophages overloaded with
`inhaled material (or intratracheally administration) and
`consequent failure of lung clearance due to e.g. aluminum
`powder, ash, dust, etc. iii) localized bronchitis/bronchioli-
`tis leading to obstruction of alveolar clearance; iv) exces-
`sive production of macrophages or reduction mobility by
`ingested surfactant or serum-derived lipids (see below).
`The spontaneous increase in incidence of pulmonary
`foamy macrophages with age is also well-known (YANG
`et al. 1966). Disturbed lipid carbohydrate metabolism in
`rats is reported to be associated with occurrence of pul-
`monary foam cells: particularly the hyper beta-lipopro-
`teinaemia condition, which is characterized by increased
`serum cholesterol and phospholipid content (SHIBUYA et
`al. 1991, 1997; TANAKA et al. 1995; WOOD and BARDEN
`1977). It has been suggested that the lungs of rats may be
`a potential excretion route for phospholipids from the
`body – leading to foamy alveolar macrophages (BERNICK
`and PATEK 1961; FLODH et al. 1974). In the one-year study
`with gepirone, phospholipids were not measured, but
`plasma phospholipid content correlates with cholesterol
`and triglycerides. In this study, the plasma levels of
`cholesterol and triglyceride were unchanged or decreased,
`respectively. It is therefore very unlikely that serum-de-
`rived lipids are involved in the occurrence of the few
`foamy macrophages in the lung in the present study with
`gepirone. Nor was there any evidence of inhaled material
`leading to the lesion. On the other hand, minimal peri-
`bronchiolar lymphoid cellular infiltrates were observed in
`the lungs of all groups treated with gepirone, including
`controls. Due to the absence of overt toxicity with
`gepirone, a relatively high dose level of gepirone was
`needed to ensure significant decrease in body weight gain
`relative to controls as the toxicologic parameter in the HD
`group, resulting in the relatively poor condition of the ani-
`mals of the HD group. It seems tempting to conclude that,
`in the one-year study with gepirone, localized bronchi-
`tis/bronchiolitis due to the generally poor physical condi-
`tion of the animals is the main factor leading to obstruc-
`tion of alveolar clearance and the subsequent occurrence
`of foamy macrophages.
`In summary, a simple strategy based on physioco-
`chemical calculations (and validation with in vitro study,
`as needed) to rapidly study the potential of compounds to
`induce phospholipidosis is presented. Calculated physic-
`ochemical parameters pKa and ClogP can be used to
`identify the (positive control) cationic amphiphilic com-
`pounds and negative controls. This straightforward and
`easy analysis may be helpful for other pathologists/toxi-
`cologists to quickly differentiate between drug-induced
`phospholipidosis or other pathologic or spontaneous
`
`8 of 9
`
`PENN EX. 2037
`CFAD V. UPENN
`IPR2015-01836
`
`
`
`conditions. In the present study with gepirone, this strat-
`egy was used to rule