Identification and biological characterization of 6-aryl-7- isopropylquinazolinones as novel TRPV1 antagonists that are effective in models of chronic pain

Article abstract of DOI:10.1021/jm051058x

Vanilloid receptor 1 (VR1, TRPV1) is a cation-selective ion channel that is expressed on primary afferent neurons and is upregulated following inflammation and nerve damage. Blockers of this channel may have utility in the treatment of chronic nociceptive

Full text of DOI:10.1021/jm051058x

J. Med. Chem. 2006, 49, 471-474
471
Chart 1. Pyridopyrimidine Hit and Close Analogue
Identification and Biological Characterization of
6-Aryl-7-isopropylquinazolinones as Novel
TRPV1 Antagonists that Are Effective in Models
of Chronic Pain
Andrew J. Culshaw,*,† Stuart Bevan,† Martin Christiansen,†
Prafula Copp,† Andrew Davis,† Clare Davis,† Alex Dyson,†
Edward K. Dziadulewicz,† Lee Edwards,†
Scheme 1. Synthesis of Pyridopyrimidinesa
Hendrikus Eggelte,† Alyson Fox,† Clive Gentry,†
Alex Groarke,† Allan Hallett,† Terance W. Hart,†
Glyn A. Hughes,† Sally Knights,† Peter Kotsonis,† Wai Lee,†
Isabelle Lyothier,† Andrew McBryde,† Peter McIntyre,†
George Paloumbis,† Moh Panesar,† Sadhana Patel,†
Max-Peter Seiler,‡ Mohammed Yaqoob,† and
Kaspar Zimmermann‡
NoVartis Institutes for Biomedical Research, 5 Gower Place,
London WC1E 6BS, U.K., and NoVartis Institutes for Biomedical
Research, CH-4002 Basel, Switzerland
a (a) Pivalonitrile, xylene, reflux, 18 h (77%); (b) 2M HCl, reflux (78%);
(c) tBuOCH(NMe2)2, 110 °C (quantitative by NMR); (d) 6-amino-2-thioxo-
2,3-dihydro-1H-pyrimidin-4-one, AcOH, EtOH, water (1:1:8), reflux, 18 h
(29% after recrystallization).
ReceiVed October 20, 2005
Abstract: Vanilloid receptor 1 (VR1, TRPV1) is a cation-selective
ion channel that is expressed on primary afferent neurons and is
upregulated following inflammation and nerve damage. Blockers of
this channel may have utility in the treatment of chronic nociceptive
and neuropathic pain. Here, we describe the optimization from a high
throughput screening hit, of a series of 6-aryl-7-isopropylquinazolinones
that are TRPV1 antagonists in vitro. We also demonstrate that one
compound is active in vivo against capsaicin-induced hyperalgesia and
in models of neuropathic and nociceptive pain in the rat.
as dizziness and somnolence in some patients.7 Hence, there
remains a great medical need for new medicines for the
treatment of pain.
Since the cloning of TRPV1 by Caterina et al.,3 there has
been an upsurge in research activity in the TRP channel area.
Capsazepine8 and other small molecule TRPV1 antagonists9,10
have been shown to be antihyperalgesic in animal models of
chronic pain, opening a new avenue in the search for innovative
pain drugs.11,12 In this letter, we divulge a new class of small
molecule TRPV1 antagonists, illustrating their evolution from
a high throughput screening hit and their characterization as
novel antihyperalgesic agents, effective against both nociceptive
and neuropathic pain states.
Vanilloid receptor 1 (VR1, TRPV1) is the best known
member of the transient receptor potential ion channel family,
a group of channels that are widely expressed in mammalian
tissues and have a key role in sensory processes. TRPV1 is
present on polymodal primary sensory neurons and is acti-
vated by low pH, noxious heat, and chemical mediators such
as capsaicin (the pungent principle of chili peppers) and
resiniferatoxin, an extreme irritant from the sap of Euphorbia
resinifera.1 Studies on vanilloid receptors have been facilitated
The pyridopyrimidine derivative 113 (Chart 1) was identified
as a hit from a high throughput screen for blockers of capsaicin
activation of the TRPV1 receptor. This compound inhibited the
Ca2+ influx evoked by capsaicin in vitro in both human (IC50
100 ( 18 nM) and rat (IC50 90 ( 16 nM) TRPV1-expressing
cells and was also able to block low pH activation of human
(IC50 49 ( 13 nM) and rat (IC50 25 ( 10 nM) TRPV1 responses.
A preliminary appraisal of 1 showed that replacement of the
thione with oxo led to a >5-fold loss of potency in the human
low pH assay17 (IC50 716 ( 166 nM). Replacing the tbutyl group
with ethyl while retaining the thione gave an approximately
equipotent analogue (IC50 172 ( 59 nM). N-1 of 1 could be
methylated without loss of activity, but methylation of N-3 gave
inactive compounds. Removal of the phenyl ring at C-6 was
detrimental. It became clear that the 2-thioxo-2,3-dihydro-1H-
pyrido[2,3-d]pyrimidin-4-one core was optimal and that a tert-
butyl group in the 7-position was best. Potency could be
enhanced by suitable substitutions to the C-6 aryl ring, and it
was through changes to this region that we identified 2, the
synthesis of which is shown in Scheme 1.
by the discovery in these laboratories of capsazepine,2
a
competitive antagonist of capsaicin-induced responses. The
ability of capsazepine to inhibit responses to capsaicin has been
used as a diagnostic feature in vanilloid research and played an
important role in the characterization of the cloned vanilloid
receptor.3
Evidence suggests that TRPV1 is a key integrator of the pain
response because it is upregulated following inflammation and
nerve damage.4,5 Chronic pain sufferers endure a debilitating
condition for which there is often no satisfactory treatment.
Neuropathic pain, in particular, is poorly served by standard
therapiessNSAIDs, opioids, and tricyclic antidepressants either
work poorly or not at all, or produce unacceptable side effects.6
Even the current gold standard treatment, the anti-convulsive
agent gabapentin, requires a relatively high, carefully titrated
dose, is ineffective in some cases, and leads to side effects such
Reaction of pivalonitrile with 4-chlorobenzylmagnesium
bromide followed by acidic hydrolysis of the imine intermediate
gave ketone 3.14 This was reacted with Bredereck’s reagent to
give the enamine 4, which could be combined with com-
mercially available 6-amino-2-thioxo-2,3-dihydro-1H-pyrimidin-
* To whom correspondence should be addressed. Present address:
Novartis Institutes for Biomedical Research, Wimblehurst Road, Horsham,
West Sussex, RH12 5AB, UK. Tel: +44 (0)1403 272827. Fax: +44
(0)1403 323837. E-mail: andrew.culshaw@novartis.com.
† Novartis Institutes for Biomedical Research, London, U.K.
‡ Novartis Institutes for Biomedical Research, Basel, Switzerland.
10.1021/jm051058x CCC: $33.50 © 2006 American Chemical Society
Published on Web 12/20/2005

472 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 2
Scheme 2. Synthesis of Quinazolinedione 8a
Letters
Table 1. IC50 Values of Selected Quinazolinones in the Human Low
pH Assaya
cmpd
R1
R2
R3
R4
Cl
R5
IC50 (nM)
a (a) tBuOCH(NMe2)2, 110 °C (quantitative, product not isolated); (b)
KMnO4, KOAc, tBuOH/water (93%); (c) (COCl)2, DCM, cat. DMF then
MeOH (96%); (d) H2, 10% Pd on C (100%); (e) I2, Ag2SO4, EtOH (87%);
(f) 4-chlorophenylboronic acid, Pd(Ph3P)4, Na2CO3, water/DMF, 80 °C
(75%); (g) diphosgene, DCM, RT; (h) ammonia solution (sg 0.88); (i)
NaOH, MeOH/water, reflux (80% overall).
9
Me
Me
Me
Me
Me
Et
H
H
H
H
H
H
H
Cl
H
H
H
H
H
H
H
H
H
H
H
H
H
H
F
OMe
H
13
23
14
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
H
Cl
Cl
Cl
Cl
Cl
OMe
H
H
H
H
H
H
OEt
H
Cl
H
H
H
H
H
H
H
H
H
H
H
OCH2CO2H
O(CH2)2NMe2
>1000
487
>1000
133
37
1343
100
12
40
82
216
557
286
31
H
H
H
OH
NH2
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
4-one to give 2 in moderate overall yield. 1 was prepared by
the same procedure but using benzylmagnesium bromide.
4′-Chloro substitution of the pendant phenyl ring of 1 to give
2 increased potency to provide a functional antagonist of human
TRPV1 expressed in CHO cells when activated by low pH (IC50
18 ( 1 nM). Compound 2 had no agonist activity in a 45Ca2+
uptake assay.15
OMe
OEt
OnPr
OnBu
O(CH2)2OMe
O(CH2)2OH
OEt
-OCH2O-
OCH2 Pr
OCH2 Bu
Compound 2 displayed suboptimal physicochemical and
pharmacokinetic propertiessin particular, exceedingly low
thermodynamic aqueous solubility (0.15 mg/L at pH 6.8, 25
°C), a relatively long half-life in the rat (t0.5 7.5 h; 3 mg/kg p.o.
in 20% cremophor EL/pbs) which might translate into an
unacceptably long half-life in man, and low micromolar
interactions with several members of the cytochrome P450 family.
We decided to attenuate the half-life of 2 by introduction of
a metabolically labile group. First, though, we opted to modify
the core of 2 to see if we could influence its overall physico-
chemical properties. Our primary aims for this phase were to
remove the sulfur atom at C-2 (toxicology concerns) and
improve the aqueous solubility relative to 2.
c
Cl
Cl
Cl
Me
50
c
144
201
42
OiPr
Cl
capsazepine
334
a IC50 values are average values of three determinations; see Supporting
Information for statistical analysis.
Scheme 3. Synthesis of Quinazolinone 9 from Key Intermediate
7a
The first breakthrough arrived with the discovery that the
sulfur atom could be replaced with oxygen if the pyridine
nitrogen was concomitantly replaced by carbon. This gave the
quinazolinedione (8), the synthesis of which is shown in Scheme
2. In 8, an isopropyl group was employed in the 2-position as
it provided a ca. 35-fold improvement in potency compared with
the tert-butyl analogue.
Commercial 2-nitro-p-cymene was treated with Bredereck’s
reagent at 110 °C to give an enamine that was oxidized directly
to the benzoic acid without prior isolation. This acid was
esterified using standard methods to give the nitro ester 5. The
ester was reduced to the corresponding methyl anthranilate and
iodinated to give the key intermediate 6. Suzuki coupling, in
this case using 4-chlorophenylboronic acid to give 7, was
followed by cyclization to the quinazolinedione using a three-
step, one-pot procedure.
Initial profiling of 8 suggested that it was suitable as a new
lead structure: e.g., MW 314.77, clogP 4.27, reaching 5317 (
352 pmol/mL in plasma at 1 h after a 3 mg/kg p.o. dose (20%
cremophor/water) to male Wistar rats, selective for TRPV1 over
30 other receptors. Although it had excellent in vitro potency
(IC50 6 ( 1 nM), it retained the poor solubility of 2 and returned
a positive result in the mouse lymphoma chromosome aberration
(“micronucleus”) test. Analogue synthesis around this new lead
commenced, with the aim of introducing solubilizing groups to
the pendant aromatic ring and evaluating promising compounds
in the micronucleus test. It did not prove possible to achieve
an acceptable balance between in vitro activity and solubility
a (a) HCl (g), MeCN, reflux (70%)
with this compound class, although the undesirable micronucleus
activity could be removed by methylating the N-1 position.
When the C-2 carbonyl was replaced with a methyl group to
give the 2-methylquinazolinone (9), there were signs that the
equilibrium solubility could be slightly improved (4 mg/L,
pH6.8) while retaining in vitro activity (see Table 1). The
synthesis of quinazolinone 9 via intermediate 7 is shown in
Scheme 3.
With this new lead in hand, our strategy was to introduce
solubilizing groups to the pendant aryl ring, or effect disruption
of the crystal packing by introduction of bulky and/or flexible
chains at the same position. Use of an alkoxy group fulfilled
two roles, providing a handle for modulation of half-life via
dealkylation while the additional oxygen atom was expected to
marginally improve the solubility.
Late-stage variation of a 3′-alkoxy substituent was possible
via thermal or microwave assisted SNAr reactions of the 4′-
chloro-3′-fluoro derivative 10 with various alkoxides, as shown
in Scheme 4.
Compound 10 was prepared by Suzuki coupling of 6 (Scheme
2) with 4-chloro-3-fluorobenzeneboronic acid and subsequent
acid mediated ring closure in acetonitrile, analogous to the
procedure shown in Scheme 3. Other derivatives were accessed

Letters
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 2 473
Table 2. Selected Pharmacokinetic Parameters for 26
Scheme 4. Variation of 3′-Substituent by SNAr Displacementa
26 i.v.
26 p.o.
parameter
1 mg/kg
3 mg/kg
C
C
t)2min pmol/mL
maxplasma pmol/mL
3543 ( 271
1912 ( 150
4.0
1666 ( 228
4.0
T
maxplasma h
maxbrain pmol/g
maxbrain h
1/2terminal h
CLtot mL/min/kg
ss L/kg
BAV
C
T
t
1079 ( 32
0.5
3.48
5.25
1.55
3.5
a (a) NaH, ROH, NMP, RT then microwave 160 °C, 10 min; (b) NaH,
ROH, NMP, 60 °C, 2 h.
V
75%
by Suzuki coupling of 6 with the appropriate boronic acids and
subsequent ring closure, as previously described.
IC50 values of representative quinazolinones are shown in
Table 1. In the 2-position (R1), a methyl group is optimals
extension to higher alkyl (compound 14) or substitution with
polar functionality (e.g., compound 15) resulted in much
reduced TRPV1 activity. In the 3-position an H-bond donor is
essential for activity. The SAR of substitutions around the
6-position phenyl group was extensively investigated. Ortho
substitutions (R2) alone are generally detrimental unless ad-
ditional substitutions are made in the 3′- and 4′-positions
(compound 16).
In the 4′-position (R4) small, hydrophobic groups such as
Cl, Me (29), OMe are tolerated. A variety of alkoxy groups
(but not the parent hydroxyl, 17) can be introduced at the meta
position (R3) provided there is also a substituent at the
4′-position. A para, meta arrangement is superior to a meta,
meta arrangement (compare compounds 11 and 18). Better
activity is obtained if the alkoxy chain is relatively short
(compare 19, 20, 21). Branching adjacent to the oxygen atom
is not favorable (e.g., compare compounds 19 and 28). Small
carbocycles are tolerated by the channel, the smaller cyclopropyl
ring being superior to cyclobutyl (26, 27), but polar atoms other
than the linking oxygen are generally not accommodated (e.g.,
12, 13, 22, 23). A combination of 4′-chlorine atom with 3′-
alkoxy group is superior to alkoxy groups in the corresponding
positions (compare 11 and 24) unless the alkoxy groups have
additional constraint (compound 25).
Figure 1. Inhibition of capsaicin-induced mechanical hyperalgesia by
compound 26 in the rat. ***P < 0.001, *P < 0.05 compared to vehicle
by ANOVA followed by Tukey’s HSD test.
Figure 2. Activity of 26 against inflammatory hyperalgesia in the rat.
***P < 0.001, *P < 0.05 compared to vehicle by ANOVA followed
by Tukey’s HSD test carried out on withdrawal threshold data.
Although 26 was not the most potent member of the series
and it did not have markedly improved solubility (0.8 mg/L at
pH6.8, 25 °C) compared with 2, a cross section of other
properties led us to carry out further investigations in animal
models of pain. In particular, 26 maintained comparable activity
in the rat low pH assay (IC50 105 ( 3 nM) and the intrinsic
clearance values for 26 in liver microsomes in vitro predicted
a moderate hepatic clearance in vivo for both man (CLint 49
µL/min/mg) and rat (CLint 76 µL/min/mg). A Caco-2 perme-
ability study showed that 26 crossed the membrane passively
with Pm 26.2 × 10-5 cm/min. 26 showed high plasma protein
binding (99.4%) at a single concentration in rat plasma but was
significantly less bound to human plasma proteins (94%).
Compound 26 was assessed for its binding interactions with a
wide range of GPCRs, kinases, transporters, and ion channels.
It had significant activity at the dopamine transporter (1.1 µM),
the norepinephrine transporter (2.8 µM), and at the 5HT1A (7
µM) and 1B (4.4 µM) receptors. 26 was clean in the Ames and
micronucleus tests.
dependent manner, indicating that in vivo it acts as an antagonist
of the TRPV1 receptor (Figure 1).
In a rat model of persistent nociceptive pain, 26 (administered
p.o. as a solution in 20% cremophor/water 24h after CFA
injection into the hindpaw) reversed established mechanical
hyperalgesia with a ED50 value16 of 4.7 mg/kg at 1 h and
maximal efficacy of 60% (Figure 2).
A similar profile was observed in a rat model of neuropathic
pain. Compound 26 (administered p.o. as a solution in 20%
cremophor EL/water 11-15 days following partial sciatic nerve
ligation) reversed established mechanical hyperalgesia in a dose
dependent fashion (Figure 3). The maximal reversal was 57%
with a ED50 of 2.6 mg/kg at 1 h.
Following repeated administration (10 and 30 mg/kg p.o.,
twice daily for 5 days) there was no evidence for the develop-
ment of tolerance to the anti-hyperalgesic effect of 26, and no
obvious side effects.
Selected pharmacokinetic parameters for 26 are shown in
Table 2. Additionally, there was no evidence of accumulation
of 26, and absorption was shown to be linear after increasing
the oral dose to 10 mg/kg.
In na¨ıve rats, compound 26 (administered i.v. as a solution
in 20% cremophor EL/saline) inhibited mechanical hyperalgesia
(induced by injection of capsaicin into the hind paw) in a dose-
Since compound 26 readily penetrates the blood-brain
barrier, the accelerating rotarod was used to assess changes in
coordination or motor function following oral administration
to rats. These data indicate that 26, at doses at least 10 times
greater than those producing an anti-hyperalgesic effect, does
not produce ataxia, sedation, or spasmolysis, which would result
in a loss of motor coordination.

474 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 2
Letters
(3) Caterina, M. J.; Schumacher, M. A.; Tominaga, M.; Rosen, T. A.;
Levine, J. D.; Julius, D. The Capsaicin receptor: A Heat-Activated
Ion Channel in the Pain Pathway. Nature 1997, 389, 816-824.
(4) Hudson, L. J.; Bevan, S.; Wotherspoon, G.; Gentry, C.; Fox, A.;
Winter, J. VR1 Protein Expression Increases in Undamaged DRG
Neurons After Partial Nerve Injury. Eur. J. Neurosci. 2001, 13,
2105-2114.
(5) Ji, R. R.; Samad, T. A.; Jin, S. X.; Schmoll, R.; Woolf, C. J. p38
MAPK Activation by NGF in Primary Sensory Neurons after
Inflammation Increases TRPV1 Levels and Maintains Heat Hyper-
algesia. Neuron 2002, 36, 57-68.
(6) McCleane, G. Pharmacological Management of Neuropathic Pain.
CNS Drugs 2003, 17, 1031-1043.
(7) See, for example, Serpell, M. G. Gabapentin in Neuropathic Pain
Syndromes: A Randomised, Double-Blind, Placebo-Controlled Trial.
Pain 2002, 99, 557-566.
(8) Walker, K. M.; Urban, L.; Medhurst, S. J.; Patel, S.; Panesar, M.;
Fox, A. J.; McIntyre, P. The VR1 Antagonist Capsazepine Reverses
Mechanical Hyperalgesia in Models of Inflammatory and Neuropathic
Pain. J. Pharmacol. Exp. Ther. 2003, 304, 56-62.
(9) Pomonis, J. D.; Harrison, J. E.; Mark, L.; Bristol, D. R.; Valenzano,
K. J.; Walker, K. N-(4-tertiarybutylphenyl)-4-(3-chloropyridin-2-yl)
tetrahydro pyrazine-1(2H)-carbox-amide (BCTC), a Novel, Orally
Effective Vanilloid Receptor 1 Antagonist with Analgesic Proper-
ties: II. In Vivo Characterization in Rat Models of Inflammatory
and Neuropathic Pain. J. Pharmacol. Exp. Ther. 2003, 306, 387-
393.
Figure 3. Oral activity of 26 against neuropathic mechanical hyper-
algesia in the rat. ***P < 0.001 compared to vehicle by ANOVA
followed by Tukey’s HSD test carried out on withdrawal threshold
data.
We believe that these data show that TRPV1 antagonists have
great promise as new treatments for a broad range of pain states
and that the 6-arylquinazolinones described are useful prototypes
in the search for clinically useful compounds.
Acknowledgment. We thank Chris Brain for the original
resynthesis of 1, Alfred Schmidt, David Ruckstuhl, and Sandro
Nocito for providing a large quantity of intermediate 6, Rochdi
Bouhelal for coordinating the high throughput screening effort,
Christian Guenat and Eva Bu¨rgin for HR-MS measure-
ments, and LeighAnn Alexander, Gian Camenisch, Hans-Georg
Capraro, Peter End, Bernard Faller, Stephen Fowler, Eric
Francotte, Willi Frieauff, Bruno Galli, Susanne Glowienke,
Michael Mutz, Appu Subramanian and Steven Whitebread for
profiling 26.
(10) Swanson, D. M.; Dubin, A. E.; Shah, C.; Nasser, N.; Chang, L.;
Dax, S. L.; Jetter, M.; Breitenbucher, J. G.; Liu, C.; Mazur, C.; Lord,
B.; Gonzales, L.; Hoey, K.; Rizzolio, M.; Bogenstaetter, M.; Codd,
E. E.; Lee, D. H.; Zhang, S.-P.; Chaplan, S. R.; Carruthers, N. I.
Identification and Biological Evaluation of 4-(3-Trifluoromethyl-
pyridin-2-yl)piperazine-1-carboxylic Acid (5-Trifluoromethylpyridin-
2-yl)amide, a High Affinity TRPV1 (VR1) Vanilloid Receptor
Antagonist. J. Med. Chem. 2005, 48, 1857-1872.
(11) Szallasi, A.; Appendino, G. Vanilloid Receptor TRPV1 Antagonists
as the Next Generation of Painkillers. Are We Putting the Cart before
the Horse? J. Med. Chem. 2004, 47, 2717-2723.
(12) Rami, H. K.; Gunthorpe, M. J. The Therapeutic Potential of TRPV1
(VR1) Antagonists: Clinical Answers Await. Drug DiscoVery Today
Ther. Strategies 2004, 1, 97-104.
Supporting Information Available: 1H NMR, HPLC, and
HRMS characterization data for all compounds, elemental analysis
for 26, and details of in vitro and in vivo assays including expanded
pharmacokinetic data. This information is available free of charge
via the Internet at http://pubs.acs.org.
(13) Compounds in this letter were first disclosed in patent applications
WO 2002076946 and WO 2004033435.
(14) Prashad, M.; Liu, Y.; Repic, O. An Expeditious Synthesis of 1-(4-
Chlorophenyl)-3,3-dimethyl-2-butanone by a Ligand-Free Palladium-
Catalyzed R-Arylation of Pinacolone: Scale-Up and Effect of Base
Concentration. AdV. Synth. Catal. 2003, 345, 533-536.
(15) Wood, J. N.; Winter, J.; James, I. F.; Rang, H. P.; Yeats, J.; Bevan,
S. Capsaicin-induced ion fluxes in dorsal root ganglion cells in
culture. J. Neurosci. 1988, 8, 3208-3220.
(16) ED50 is defined here as the dose required to bring about 50% of the
maximal antihyperalgesic effect.
(17) Subsequent IC50 values refer to human low pH data unless stated
otherwise. Blockade of low pH induced TRPV1 activation may be
predictive of antihyperalgesic activity in vivo; see ref 9.
References
(1) Caterina, M. J.; Julius, D. The Vanilloid Receptor: A Molecular
Gateway to the Pain Pathway. Annu. ReV. Neurosci. 2001, 24, 487-
517.
(2) Walpole, C. S. J.; Bevan, S.; Bovermann, G.; Boelsterli, J. J.;
Breckenridge, R.; Davies, J. W.; Hughes, G. A.; James, I.; Oberer,
L.; Winter, J.; Wrigglesworth, R. The Discovery of Capsazepine,
the First Competitive Antagonist of the Sensory Neuron Excitants
Capsaicin and Resiniferatoxin. J. Med. Chem. 1994, 37, 1942-
1954.
JM051058X