1,4-Dioxane-fused 4-anilinoquinazoline as inhibitors of epidermal growth factor receptor kinase

Article abstract of DOI:10.1002/1521-4184(200112)334:11<357::AID-ARDP357>3.0.CO;2-Q

The 4-anilinoquinazoline PD 153035 (1) is a potential antitumor agent which acts by inhibiting tyrosine kinase activity of epidermal growth factor receptor (EFGR) via competitive binding at the ATP site of enzyme. A series of cyclic analogues of PD 153035 bearing the 1,4-dioxane ring was prepared by reaction of 6-chloro derivative 5 with several aniline nucleophiles. These were evaluated for their ability to inhibit the EGFR kinase and the growth of primary human tumor cell cultures. All of the new 4-anilinoquinazolines exhibited less potency than PD 153035 against EGFR kinase. However, compounds 2b, 2c, 2e, 2g, and 2h showed higher inhibitory activities than PD 153035 against the growth of A431 tumor cell line. The compound 2b containing 3-chloroaniline ring was as potent as PD 153035 against EGFR kinase and showed about 5.4-fold better potency than PD1 53035 in the inhibition of growth of A431 cell line with good selectivity.

Full text of DOI:10.1002/1521-4184(200112)334:11<357::AID-ARDP357>3.0.CO;2-Q

Arch. Pharm. Pharm. Med. Chem. 2001, 334, 357–360
Jae Yeol Leea),
Dioxane-fused 4-anilinoquinazoline 357
b)
1,4-Dioxane-fused 4-anilinoquinazoline as
inhibitors of epidermal growth factor
receptor kinase
Yong Kyu Park ,
a)
Seon Hee Seo ,
a)
In-Seop So ,
Hee-Kyung Chung ,
a)
a)
Beom-Seok Yang ,
b)
Sook Ja Lee ,
The 4-anilinoquinazoline PD 153035 (1) is a potential antitumor agent which acts by
inhibiting tyrosine kinase activity of epidermal growth factor receptor (EFGR) via com-
petitive binding at the ATP site of enzyme. A series of cyclic analogues of PD 153035
bearing the 1,4-dioxane ring was prepared by reaction of 6-chloro derivative 5 with
several aniline nucleophiles. These were evaluated for their ability to inhibit the EGFR
a)
Hokoon Park ,
a)
Yong Sup Lee
a)
Medicinal Chemistry Research Cen- kinase and the growth of primary human tumor cell cultures. All of the new 4-anilinoqui-
ter, Korea Institute of Science & Tech- nazolines exhibited less potency than PD 153035 against EGFR kinase. However,
nology, P.O. Box 131 Cheongryang, compounds 2b, 2c, 2e, 2g, and 2h showed higher inhibitory activities than PD 153035
Seoul 130-650, Korea
against the growth of A431 tumor cell line. The compound 2b containing 3-chloroaniline
ring was as potent as PD 153035 against EGFR kinase and showed about 5.4-fold better
Department of Chemistry, Hankuk potency than PD153035 in the inhibition of growth of A431 cell line with good selectivity.
University of Foreign Studies, Yong-in
b)
449-791, Korea
Key Words: EGFR; Tyrosine kinases; 4-Anilinoquinazolines; A431 cell; antitumor agent;
,4-Dioxane
1
Received: May 14, 2001 [FP579]
Introduction
Protein tyrosine kinases play a central role in the transduc-
tion of proliferative signals in mammalian cells [ . Tyrosine
kinases mainly transduce signals to a target cell by cata-
lyzing the transfer of the terminal phosphate of ATP to
tyrosine residues in protein substrates. Abnormal activation
of tyrosine kinases has been implicated in many cancers
and other proliferative diseases [ . Therefore, tyrosine
kinases are expected to be attractive targets for cancer
chemotherapy. Due to the involvement of tyrosine kinases
in many signal transduction pathways, the development of
selective inhibitors of tyrosine kinases would be important.
1]
R1
HN
Br
HN
OCH3
O
O
N
N
N
OCH3
N
2]
PD 153035 (1)
2a~2h, R1 = (un)substituted phenyl
Figure 1. Structures of PD 153035 (1) and its cyclic analogue 2.
the effect of the dioxane ring on the inhibition of EGFR
kinase and the growth of human tumor cell lines.
Recently, several types of synthetic or natural compounds
have been reported as potent tyrosine kinase inhibitors [
3]
.
However, most of them have proven to be of limited use
due to the low selectivity or weak potency in cellular assays.
On the other hand, the 4-anilinoquinazoline class of com-
pounds (e.g. PD 153035, 1) have proven to be potent and
selective inhibitors of tyrosine kinase activity of the epider-
mal growth factor receptor (EGFR) via competitive binding
at the ATP site of the enzyme [ . As a consequence,
considerable efforts have been devoted to the synthesis of
PD 153035 (1) analogues for studying structure-activity
relationships to develop tyrosine kinase inhibitory antitumor
Results and discussion
The 1,4-dioxane-fused 4-anilinoquinazolines were pre-
pared by the general method as shown in Scheme 1.
7-Amino-1,4-benzodioxane-6-carboxylic acid (3), readily
available by the known procedure, was condensed with
4]
O
O
N
O
O
O
O
HO
triazine
HN
POCl3
EtOH, reflux
[5]
drugs (Figure 1)
.
H2N
In the present studies, we wish to report the synthesis and
in vitro evaluation of 1,4-dioxane-fused 4-anilinoquinazoli-
nes 2a–2h, cyclic analogues of PD 153035 (1), to probe
3
4
R1
HN
Cl
aniline derivatives
R NH )
O
O
O
(
N
1
2
N
i-PrOH
N
———
N
O
Correspondence: Prof. Yong Sup Lee, Medicinal Chemistry Re-
search Center, Korea Institute of Science & Technology, P.O. Box
5
2a~2j
131 Cheongryang, Seoul 130-650, Korea.
E-mail: yslee@kist.re.kr
Fax: +82 2 958 5189
Scheme 1. Synthesis of 1,4-dioxane-fused 4-anilinoquinazolines
(2).
©
WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0365-6233/01/1111/0357 $17.50 +.50/0

358 Dioxane-fused 4-anilinoquinazoline
Arch. Pharm. Pharm. Med. Chem. 2001, 334, 357–360
Table1. Inhibition of EGFR kinase and inhibition of human tumor cells growth bynew quinazoline derivatives
2a–2h).
(
Ar
HN
Br
HN
O
O
OCH3
N
N
N
OCH3
N
(
2a~2h)
PD 153035 (1)
IC50 (nM)b
EGFR
GI50 (µM)c
Compda
Ar
A431
SNU638
A549
phenyl
2
a
>100
8.3
3.80
0.19
0.20
54.95
4.68
5.75
45.71
4.27
4.37
3
3
-Cl-phenyl
-Br-phenyl
2
2
b
c
17.8
3
-I-phenyl
2
d
e
63.0
1.55
0.14
6.17
2.75
5.62
2.24
3
-CF -phenyl
2
3
>100
4
-F-phenyl
3-F-4-Cl-phenyl
,5-di-Me-phenyl
2
f
77.6
1.29
48.98
>100
40.74
75.86
2
g
>100
0.49
3
2
h
>100
6.6e
0.12
1.03
0.17
0.23
PD 153035 (1)d
19.50
11.48
aThese compounds were tested as the free base. bConcentration to inhibit by 50% the phosphorylation of the
tyrosine kinase substrate 2 by EGFR as determined from the dose-reaponse curve. cDose-response curves
were determined at four concentration. The GI50 values are the concentrations needed to inhibit cell growth by
5
0% as determined from these curves. dThe standard compound was made for comparison for activity. eThe
value was determined by using our assay protocol.
triazine and a catalytic amount of piperidine in refluxing
ethanol to afford quinazolinone 4 in 43% yield [ . The
quinazolinone 4 was converted in good yield into 4-chloro-
quinazoline 5 with POCl3 in the presence of N,N-dimethyl-
aniline. The reaction of 5 with appropriate substituted-
anilines in refluxing iso-propanol gave 1,4-dioxane-fused
(IC50 = 8.3 nM). The compounds (2a, 2g, and 2h) contain-
ing un- and disubstituted aniline ring showed much lower
potency (IC50 = >100 nM). This result was consistent with
the previous results in that the good enzymatic activity is
seen when the aniline ring is monosubstitued with halogen
6]
atom at the meta-position in the quinazoline-based inhibi-
4-anilinoquinazolines (2a–2h).
tors [
9]
.
The compounds (2a–2h) shown in Table 1 were evaluated
for the ability to inhibit the phosphorylation of EGFR kinase.
To compare these data with that of standard, the inhibitory
activity of the PD 153035 under our assay conditions was
inserted into a data set. The IC50 value we obtained with
PD 153035 is significantly higher and lower, respectively,
than were found by previous workers [ . It was reported
that the difference is at least attributed to the nature of the
enzyme and substrate as well as the overall assay condi-
The compounds were also evaluated for their ability to
inhibit the growth of certain cell lines. Three human carci-
noma cell lines were used: A431 (uterus cancer), which
greatly overexpresses EGFR, and SNU635 (stomach can-
cer) and A549 (lung cancer) which express low levels of
EGFR.
7, 8]
As shown in Table 1, it is immediately obvious that when
compared with PD 153035 (GI50 = 1.03 µM), most of
[8]
1
0
,4-dioxane-fused 4-anilinoquinazolines 2a–2h (GI50 =
.12–3.80 µM) showed similar or better inhibitory activities
tions
.
We surveyed a variety of substituents at different positions
on the aniline ring of these 1,4-dioxane-fused 4-anilinoquin-
azolines. In the case of monosubstitutedanilide derivatives,
all of them (2c–2f; IC50 = 17.8 to >100 nM) were less potent
than PD 153035 (IC50 = 6.1 nM) except 3-chloroanilinoqui-
nazoline 2b, which exhibited moderate enzymatic inhibition
against A431, irrespective of the potency of the enzymatic
activities. This result suggests that the mode of action of
these compounds is more complex than originally predicted
and it is possible that these compounds inhibit other ki-
nases [ . Nevertheless, most of compounds except 2h
have the good A431 selectivity over other cell lines which
10]

Arch. Pharm. Pharm. Med. Chem. 2001, 334, 357–360
Dioxane-fused 4-anilinoquinazoline 359
express low level of EGFR and compound 2g containing
General procedure for the preparation of 4-anilino-6,7-
3-F-4-Cl-aniline ring showed more than 150-fold selectivity
ethylenedioxyquinazolines (2a–2h)
for A431 tumor cell line. While all of results are difficult to
interpret, there is an approximate correlation between IC50
and GI50 in the case of compounds (2b–2d) containing a
A mixture of 5 (16 mg, 0.072 mmol), appropriate (un)substi-
tuted aniline (0.144 mmol), and a catalytic amount of conc. HCl
in i-PrOH (4 ml) was heated at reflux for 24 h. The reaction
mixture was cooled to room temperature and the resulting solid
was filtered, washed with i-PrOH, and dried to give 4-anilino-
3
-halogenoaniline ring.
In conclusion, we prepared cyclic analogues 2a–2h of PD
53035 (1) to probe the effect of the dioxane ring on the
6
,7-ethylenedioxyquinazolines 2 as free base.
1
inhibition of EGFR kinase and the growth inhibition of
humantumorcelllines. Amongthem, compound 2b bearing
the 3-chloroaniline ring was as potent as PD153035 against
the EGFR but showed about 5.4-fold better potency than
PD153035 in the inhibition of growth of A431 cell line with
good selectivity. The above results demonstrate that 4-anil-
inoquinazoline fused with 1,4-dioxane would be effective
inhibitor of EGFR when considered on the inhibitory activity
against EGFR kinase and the growth of human tumor cell
lines. Thus, we will try to prepare the various derivatives of
6,7-Ethylenedioxy-(4-phenylamino)quinazoline (2a)
1% Yield; 1H NMR (DMSO-d6) δ 11.04 (1H, s, NH), 8.76 (1H,
4
s, H-2), 8.68 (1H, s, H-10), 8.25 (1H, s, H-5), 7.70 (2H, d, J = 6.6
Hz, H-2′ & H-6′ of aniline), 7.47 (2H, m, H-3′ & H-5′ of aniline),
7.30 (1H, J = 7.5 Hz, H-4′ of aniline), 4.51–4.47 (4H, m, 2 × H-7
& 2 × H-8).
4-[(3-Chlorophenyl)amino]-6,7-ethylenedioxyquinazoline
(
2b)
60% Yield; 1H NMR (DMSO-d ) δ 11.78 (1H, s, NH), 9.60 (1H,
6
1
fold.
,4-dioxane-fused 4-anilinoquinazoline by using this scaf-
s, H-2), 9.10 (1H, s, H-10), 8.69 (1H, s, H-2′ of aniline), 8.46
(1H, d, J = 6.3 Hz, H-6′ of aniline), 8.28–8.22 (2H, m, H-4′ &
H-5′ of aniline), 8.09 (1H, s, H-5), 5.25–5.21 (4H, m, 2 × H-7 &
2
× H-8).
Acknowledgement
4
-[(3-Bromophenyl)amino]-6,7-ethylenedioxyquinazoline
This work was supported by Korea Institute of Science and
Technology (2E16800).
(
2c)
7
3% Yield; mp 230–240 °C; 1H NMR (DMSO-d6) δ 10.95 (1H,
s, NH), 8.80 (1H, s, H-2), 8.27 (1H, s, H-10), 8.07 (1H, s, H-2′
of aniline), 7.78 (1H, d, J = 6,6 Hz, H-6′ of aniline), 7.27–7.43
Experimental part
(
2H, m, H-4′ & H-5′ of aniline), 7.34 (1H, s, H-5), 4.59–4.41 (4H,
+
m, 2 × H-7 & 2 × H-8); EI-MS (m/z): 358 (M ) (calcd mass for
Chemistry
C16H12BrN3O2 358.19).
1H NMR spectra were recorded on a Gemini Varian-300
4
-[(3-Iodophenyl)amino]-6,7-ethylenedioxyquinazoline
(
300 MHz). Mass spectra (EI) were determined on HP GC 5972
(
2d)
and HP MS 5988A system at 70 eV. Analytical thin layer chro-
matography (TLC) was carried out by precoated silica gel (E.
Merck Kiesegel 60F254 layer thickness 0.25 mm). Flash column
chromatography was performed with Merck Kiesegel 60 Art
5% Yield; mp 242–245 °C; 1H NMR (DMSO-d6) δ 11.37 (1H,
9
s, NH), 8.89 (1H, s, H-2), 8.48 (1H, s, H-10), 8.15 (1H, s, H-2′
of aniline), 7.66–7.80 (3H, m, H-4′, H-5′ & H-6′ of aniline), 7.46
9
385 (230–400 mesh). All solvents used were purified accord-
(
1H, s, H-5), 4.69–4.52 (4H, m, 2 × H-7 & 2 × H-8).
ing to standard procedures.
4
-[(3-Trifluoromethylphenyl)amino]-6,7-ethylenedioxy-
6,7-Ethylenedioxy-4(3H)-quinazolin-4-one (4)
quinazoline (2e)
30% Yield; mp 273–275 °C; 1H NMR (DMSO-d ) δ 11.08 (1H,
s, NH), 8.86 (1H, s, H-2), 8.35 (1H, s, H-10), 8.19 (1H, s, H-2′
of aniline), 8.09 (1H, d, J = 8.1 Hz, H-6′ of aniline), 7.65–7.74
A mixture of 7-amino-1,4-benzodioxane-6-carboxylic acid (3,
.13 g, 5.8 mmol), triazine (0.47 g, 5.8 mmol), and piperidine
0.04 ml, 0.58 mol) in EtOH (120 ml) was heated at reflux for
4 h. The precipitated solid was filtered and washed with EtOH.
6
1
(
2
(
2H, m, H-4′ & H-5′ of aniline), 7.34 (1H, s, H-5), 4.50–4.47 (4H,
The filtrate was concentrated and the resulting solid was
washed with EtOH. The combined solid was dried in vacuo to
+
m, 2 × H-7 & 2 × H-8); EI-MS (m/z): 347 (M ) (calcd mass for
C17H12F3N3O2 347.29).
1
give 4 (0.51 g, 43%) as a white solid: mp 209 °C (dec.); H NMR
(
7
(
2
DMSO-d6, 300 MHz) δ 7.95 (1H, s, H-2), 7.45 (1H, s, H-10),
.06 (1H, s, H-5), 4.36–4.33 (4H, m, 2 × H-7 & 2 × H-8), 1.67
1H, s, NH); EI-MS (m/z): 204 (M ) (calcd mass for C10H8N2O3
04.18).
4
-[(4-Fluorophenyl)amino]-6,7-ethylenedioxyquinazoline
(
2f)
+
47% Yield; mp 282–284 °C; 1H NMR (DMSO-d6) δ 11.11 (1H,
s, NH), 8.80 (1H, s, H-2), 8.31 (1H, s, H-10), 7.72 (2H, d, J = 8.0
Hz, H-2′ & H-6′ of aniline), 7.50 (1H, d, J = 8.0 Hz, H-3′ & H-5′
of aniline), 7.34 (1H, s, H-5), 4.50–4.46 (4H, m, 2 × H-7 & 2 ×
H-8); EI-MS (m/z): 297 (M ) (calcd mass for C16H12FN3O2
297.28).
4
-Chloro-6,7-ethylenedioxyquinazoline (5)
+
A solution of 4 (0.56 g, 2.7 mmol) and N,N-dimethylaniline
0.64 ml, 5.5 mmol) in POCl3 (25 ml) was heated at reflux for
h. POCl3 was removed by distillation and the residue was
(
2
diluted with CH2Cl2 and treated with ice-water. The organic
layer was separated and washed with aqueous NaHCO3, brine,
dried (MgSO4), and concentrated to give 5 (0.54 g, 89%) as a
yellowish solid. The crude 5 was used for the next step without
purification: EI-MS (m/z): 222 (M ) (calcd mass for
C10H7ClN2O2 222.63).
4-[(4-Chloro-3-fluorophenyl)amino]-6,7-ethylenedioxy-
quinazoline (2g)
27% Yield; 1H NMR (DMSO-d6) δ 10.33 (1H, s, NH), 8.78 (1H,
s, H-2), 8.31 (1H, s, H-10), 8.20 (1H, s, H-2′ of aniline), 8.10
(1H, d, J = 7.8 Hz, H-6′ of aniline), 7.56 (1H, d, J = 7.8 Hz, H-5′
+

360 Dioxane-fused 4-anilinoquinazoline
Arch. Pharm. Pharm. Med. Chem. 2001, 334, 357–360
of aniline), 7.25 (1H, s, H-5), 4.18–4.30 (4H, m, 2 × H-7 & 2 ×
H-8).
Cancer cell growth inhibition assay
A431 (uterus cancer), SNU638 (stomach cancer), and A549
lung cancer) human cancer cells were maintained using RPMI
640 medium containing 10% fetal calf serum in 37 °C and 5%
(
1
4
-[(3,5-Dimethylphenyl)amino]-6,7-ethylenedioxyquinaz-
oline (2h)
CO2 incubator. A thousand cells were plated in 96 well plate
and incubated overnight. Cells were treated with inhibitor and
left for two days. Cells were fixed with formalin solution (SIGMA)
and washed with tap water. The cells were dried and stained
with 0.1% sulforhodamine B (SIGMA) for 30 min. The cells were
washed with 1% acetic acid and the dye was eluted from cells
by adding 0.1M Tris-HCl (pH 8.0). The absorbance was meas-
ured at 520 nm wavelength using microplate reader (Molecular
Dynamics). The inhibitor concentration which inhibits cell
growth by 50% was assigned as GI50 value.
1
5
4% Yield; mp >300 °C (dec.); H NMR (DMSO-d6) δ 11.64 (1H,
s, NH), 8.90 (1H, s, H-2), 8.30 (1H, s, H-10), 7.57 (2H, s, H-2′
H-6′ of aniline), 7.30 (1H, s, H-5), 7.10 (1H, s, H-4′ of aniline),
.52–4.38 (4H, m, 2 × H-7 & 2 × H-8).
&
4
Pharmacology
Expression of human recombinant EGFR tyrosine kinase
domain protein
The tyrosine kinase domain in the EGFR gene exists in the
C-terminal cytoplasmic domain and the recombinant C-terminal
protein was verified to possess the tyrosine kinase activity,
References
[
11]
[
[
[
[
1] M. Yamada, T. Ikeuchi, H. Hatanaka, Prog. Neurobiol. 1997,
which is identical to the purified natural EGFR protein
. To
51, 19–37.
express EGFR tyrosine kinase domain protein in SF9 insect
cells, the expression vector using pBacPAK8 vector (from Clon-
tech, U.S.A.) was constructed. The EGFR tyrosine kinase do-
main (amino acids 668–1210) was PCR amplified using 5′
primer (5’-CTCGAGATGCATCATCATCATCATCATCGAA-
GGCGCCACATCGTTCG-3’) and 3’ primer (GGTACCTCTA-
GATCATGCTCCAATAAATTCACTGC-3’) and subcloned into
XhoI and KpnI restriction enzyme cloning sites in pBacPAK8
vector. The amplified gene was sequenced to confirm the
absence of mutation. The expressed protein was engineered to
have hexahistidine tag in its N-terminal. The baculoviral stock
from the vector was generated using the virus generation kit
2] K. Hickey, D. Grehan, I. M. Reid, S. O’Briain, T. N. Walsh, T.
P. J. Hennessy, Cancer 1994, 74, 1693–1698.
3] A. J. Bridges, Chemtracts-Organic Chemistry 1995, 8, 73–
107.
4] D. W. Fry, A. J. Kraker, A. McMichael, L. A. Ambroso, J. M.
Nelson, W. R. Leopold, R. W. Connors, A. J. Bridges,Science
1994, 265, 1093–1095.
[5] G. W. Rewcastle, W. A. Denny, A. J. Bridges, H. Zhou, D. R.
Cody, A. McMichael, D. W. Fry, J. Med. Chem. 1995, 38,
3482–3487.
(
from Clontech USA). The overexpression of the recombinant
protein was carried in 1L suspension culture of the virus in-
fected SF9 cells. The expressed protein was purified using Ni2+
affinity column (from Novagen, USA) to over 95% purity follow-
ing the protocol provided by the manufacturer. The purified
recombinant EGFR tyrosine kinase domain protein has nearly
identical Km value for ATP and Vmax compared to those of
purified natural EGFR protein under our assay condition (un-
published observation).
[
6] P. M. Heertjes, B. J. Knape, C. A. van Beek, K. van den
Boogaart, J. Am. Chem. Soc.1957, 79, 3445–3447.
[7] A. J. Bridges, H. Zhou, D. R. Cody, G. W. Rewcastle, A.
McMichael, H. D. H. Showalter, D. W. Fry, A. J. Kraker, W.
A. Denny, J. Med. Chem. 1996, 39, 267–276.
[
[
[
8] A. Wissner, D. M. Berger, D. H. Boschelli, M. B. Floyd, Jr.,
L. M. Greenberger, B. C. Gruber, B. D. Johnson, N. Mamuya,
R. Nilakantan, M. F. Reich, R. Shen, H.-R. Tsou, E.
Upeslacis, Y. F. Wang, B. Wu, F. Ye, N. Zhang, J. Med.
Chem. 2000, 43, 3244–3256.
Tyrosine kinase assays
Total enzymatic inhibition assay was done in total 20 µl reaction
mixture containing 10 ng of purified human recombinant EGFR
tyrosine kinase domain protein, 250 µM biotinylated tyrosine ki-
nase substrate 2 (purchased from Promega, USA), 20 µM ATP, 2
µCi p32-gamma-ATP, 5 mM MgCl2, 5 mM DTT, and 1µl of appro-
priate dilution of inhibitor. The reaction was carried out for 1 h at
9] B. D. Smaill, B. D. Palmer, G. W. Rewcastle, W. A. Denny, D.
J. McNamara, E. M. Dobrusin, A. J. Bridges, H. Zhou, H. D.
H.Showalter, R. T. Winters, W. R. Leopold, D. W. Fry, J. M.
Nelson, V. Slintak, W. L. Elliot, B. J. Roberts, P. W. Vincent,
S. J. Patmore, J. Med. Chem. 1999, 42, 1803–1815.
30 °C. The reaction was terminated by adding 10 µl of 30% phos-
10] A. Gazit, J. Chen, H. App, G. McMahon, P. Hirth, I. Chen, A.
phoric acid and spotted in avidin-coated PVDF membrane (pur-
chased from Promega). The membrane was washed five times
with 20 mM Tris-HCl (pH 8.0) containing 0.2N NaCl. The radioac-
tivity from each spot was quantitated using the BAS system (from
Kodak).
Levitzki, Bioorg. Med. Chem. 1996, 8, 1203–1207.
[11] C. Y. Hsu, M. Mohammadi, M. Nathan, A. Honegger, A.
Ullrich, J. Schlessinger, D. R. Hurwitz, Cell Growth Differ.
1990, 1, 191–200.