Balancing potency, metabolic stability and permeability in pyrrolopyrimidine-based EGFR inhibitors

Article abstract of DOI:10.1016/j.ejmech.2016.08.068

The present study describes our continuous effort to develop epidermal growth factor receptor (EGFR) inhibitors based on the 6-aryl-pyrrolo[2,3-d]pyrimidin-4-amine scaffold. The activity-ADME space has been evaluated by synthesizing 43 new structures, including four variations of the 4-amino group and 34 different substitution patterns in the 6-aryl moiety. Most of the new pyrrolopyrimidines were highly active, with twelve analogues possessing lower IC₅₀values than the commercial drug Erlotinib in enzymatic assays. Ten EGFR inhibitors were also profiled in cell studies using the Ba/F3-EGFR^L858Rreporter cells, and all revealed nanomolar activity. However, some of the privileged structures in terms of potency had ADME short-comings: compounds containing amides, sulfonamides, amine and hydroxymethyl substituents in the 6-aryl group had low permeability and high efflux, derivatives having (R)-3-amino-3-phenylpropan-1-ol at C-4 induced hERG inhibition properties, and metabolic lability was seen for compounds having (S)-2-methoxy-1-phenylethan-1-amine at C-4. Based on a trade-off between enzymatic activity, cellular potency and ADME properties, (S)-2-phenyl-2-((6-phenyl-7H-pyrrolo[2,3-d]pyrimidin-4-yl)amino)ethan-1-ol appeared as the most promising drug candidate. Cellular studies indicate this compound to have therapeutic use in EGFR driven diseases.

Full text of DOI:10.1016/j.ejmech.2016.08.068

Accepted Manuscript
Balancing potency, metabolic stability and permeability in pyrrolopyrimidine-based
EGFR inhibitors
Jin Han, Silje Henriksen, Kristin G. Nørsett, Eirik Sundby, Bård Helge Hoff
PII:
S0223-5234(16)30726-7
10.1016/j.ejmech.2016.08.068
EJMECH 8864
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 18 April 2016
Revised Date: 29 August 2016
Accepted Date: 30 August 2016
Please cite this article as: J. Han, S. Henriksen, K.G. Nørsett, E. Sundby, B.H. Hoff, Balancing potency,
metabolic stability and permeability in pyrrolopyrimidine-based EGFR inhibitors, European Journal of
Medicinal Chemistry (2016), doi: 10.1016/j.ejmech.2016.08.068.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
R2
R4
Less polar: acceptable efflux ratio
More polar: high efflux ratio
CD3
O
O
H
H
N
N
N
OH
N
R
SAR and
N
ADME studies
N
N
Increased
metabolic stability
HN
HN
R1
Ph
Lead Structure
Ph
3 highly active EGFR inhibitors
4
Solubility:
0 deriv. > Erlotinib
2
HN
OH
Ph
hERG inhibition

ACCEPTED MANUSCRIPT
Title Page
Balancing potency, metabolic stability and permeability in pyrrolopyrimidine-based
EGFR inhibitors
a,b)
a,c)
d)
b)
a)*
Jin Han , Silje Henriksen , Kristin G. Nørsett , Eirik Sundby , Bård Helge Hoff
Adresses:
a)
Norwegian University of Science and Technology, Department of Chemistry,
b)
Høgskoleringen 5, NO-7491 Trondheim, Norway. Norwegian University of Science and
Technology, Institute of Chemistry and Material Technology, Faculty of Technology,
c)
7
491 Trondheim. Current address: Jotun Marine, Protective and Powder Coatings,
d)
Rosenvoldsgt. 19, NO-3202 Sandefjord, Norway. Norwegian University of Science and
Technology, Department of Cancer Research and Molecular Medicine, Prinsesse
Kristinas gt. 1, N-7030 Trondheim, Norway.
Corresponding author:
*
Bård H. Hoff, Department of Chemistry, Norwegian University of Science and
Technology (NTNU), NO-7491, Norway. E-mail: Bard.Helge.Hoff@chem.ntnu.no, phone:
+004773593973, fax: ++004773594256.
+
E-mail:
J. Han: Jin.Han@ntnu.no; S. Henriksen; Silje.Henriksen@Jotun.no; K.G. Nørsett:
Kristin.Norsett@ntnu.no; E. Sundby: Eirik.Sundby@hist.no; B.H. Hoff:
Bard.Helge.Hoff@chem.ntnu.no;
1

ACCEPTED MANUSCRIPT
Balancing potency, metabolic stability and permeability in pyrrolopyrimidine based
EGFR inhibitors
a,b)
a,c)
d)
b)
a)*
Jin Han , Silje Henriksen , Kristin G. Nørsett , Eirik Sundby , Bård Helge Hoff
a)
Norwegian University of Science and Technology, Department of Chemistry, Høgskoleringen 5, NO-7491
b)
Trondheim, Norway. Norwegian University of Science and Technology, Institute of Chemistry and
Material Technology, Faculty of Technology, 7491 Trondheim. Current address: Jotun Marine, Protective
and Powder Coatings, Rosenvoldsgt. 19, NO-3202 Sandefjord, Norway. Norwegian University of Science
c)
d)
and Technology, Department of Cancer Research and Molecular Medicine, Prinsesse Kristinas gt. 1, N-
7
030 Trondheim, Norway.
Corresponding author: * Bård H. Hoff, Department of Chemistry, Norwegian University of Science and
Technology (NTNU), NO-7491, Norway. Email: Bard.Helge.Hoff@chem.ntnu.no, phone:
+
+004773593973, fax: ++004773594256.
Abstract: The present study describes our continuous effort to develop epidermal growth
factor receptor (EGFR) inhibitors based on the 6-aryl-pyrrolo[2,3-d]pyrimidin-4-amine
scaffold. The activity-ADME space has been evaluated by synthesizing 43 new structures,
including four variations of the 4-amino group and 34 different substitution patterns in the
6
-aryl moiety. Most of the new pyrrolopyrimidines were highly active, with twelve
analogues possessing lower IC values than the commercial drug Erlotinib in enzymatic
5
0
L858R
assays. Ten EGFR inhibitors were also profiled in cell studies using the Ba/F3-EGFR
reporter cells, and all revealed nanomolar activity. However, some of the privileged
structures in terms of potency had ADME short-comings: compounds containing amides,
sulfonamides, amine and hydroxymethyl substituents in the 6-aryl group had low
permeability and high efflux, derivatives having (R)-3-amino-3-phenylpropan-1-ol at C-4
induced hERG inhibition properties, and metabolic lability was seen for compounds
having (S)-2-methoxy-1-phenylethan-1-amine at C-4. Based on a trade-off between
enzymatic activity, cellular potency and ADME properties, (S)-2-phenyl-2-((6-phenyl-
7
H-pyrrolo[2,3-d]pyrimidin-4-yl)amino)ethan-1-ol appeared as the most promising drug
candidate. Cellular studies indicate this compound to have therapeutic use in EGFR
driven diseases.
Keywords: EGFR; Pyrrolopyrimidine; SAR; ADME; metabolism; Erlotinib
2

ACCEPTED MANUSCRIPT
1
. Introduction
Protein kinases are an important class of enzymes which catalyse phosphorylation of
specific protein substrates, thereby regulating vital cellular processes. Abnormality in
expression levels or in the structure of these protein kinases plays a crucial role in a
number of human diseases. Kinases are especially attractive targets for treatment of
cancer diseases [1, 2] The membrane bound epidermal growth factor receptor tyrosine
kinase (EGFR) has been one of the more investigated kinase, since EGFR amplification
or mutation has been noted in lung [1, 3], breast [2], pancreatic [4], ovarian [5], prostate
[
6], and head and neck [7] cancers. However, development of new kinase drug candidates
is increasingly difficult, and a late focus on absorption, distribution, metabolism,
excretion and toxicology (ADMET) is regarded as one of the main reasons for drug
failure [8, 9]. Kinase inhibitors tend to have higher molecular weight and be more
lipophilic than other classes of drugs [10], which could lead to suboptimal properties in
terms of solubility and permeability [11], but also to toxicity [12]. Attractive kinase
inhibitors could have a moderate kinase selectivity [13, 14] causing side effects [15].
Moreover, patients receiving EGFR therapy are likely to have undergone other type of
chemotherapy increasing the risk of organ failure [16]. Thus, avoiding potential ADMET
pitfalls early on in development is important. Most small molecular based EGFR
inhibitors are based on aniline substituted quinazolines. For this class of drug candidates
ADMET data is available as input for drug developmental programs [16-18]. We have
focused our development on chiral benzylamine based pyrrolo[2,3-d]pyrimidines as
EGFR inhibitors [19-21]. Only three pyrroloyrimidines based EGFR inhibitors have
reached clinical trials, namely PKI-166 [22], AEE-788 [23, 24], and TAK-285 [25, 26],
see Figure 1.
N
H
N
H
N
H
N
O
N
N
N
N
N
N
N
N
OH
OH
N
H
N
H
N
N
H
N
CH3
CH3
Cl
NH
O
OH
N
H
F3C
O
HO
PKI-166
AEE-788
TAK-285
Lead structure I
Figure 1. Pyrrolopyrimidine-based EGFR inhibitors.
Our lead compound I [20] (Figure 1), which possessed high in vitro enzymatic and
cellular activity, was unfortunately a substrate of the breast cancer resistant protein and
the P-glycoprotein ABC-transporters, making further development challenging [21].
Herein, we describe our continuous effort to identify new EGFR inhibitor candidates with
improved ADME profile.
3

ACCEPTED MANUSCRIPT
2
. Results and discussion
.1 Design
2
At the onset of this study, our main concern with the lead structure I was metabolism at
the 4-positioned benzylic alcohol and at the 2-methoxy group. However, as the project
proceeded it was evident that the challenge with I was mediocre permeability and a high
efflux, while human liver microsome assay indicated the molecule to be stable to phase I
oxidative metabolism. Thus, the drug discovery process was thereafter motivated by
desire to improve permeability while maintaining on target potency.
Initially, we focused on identifying bioisosteres for the 2-methoxy and the 4-
hydroxymethyl groups. The 4-benzylic function is directed towards the active site
entrance, which might allow for some flexibility in terms of substituent size. Thus, we
included a range of hydrogen bond acceptor and donor groups of different size at the 3-
and 4-position of the 6-aryl group. At the 2-position we set on to synthesise compounds
with ether functionalities varying in size and electronic properties.
Figure 2. Lead structure I, and structural elements subjected to further evaluation.
As the project proceeded various strategies were attempted to improve the activity/ADME
properties of the compounds including: fluorine insertion; deuterisation; solubilizing tails;
replacement of the 6-aryl group by pyridines; and varying the amine part of the molecule.
2
.2 Chemistry
The synthesis of all EGFR inhibitors were performed starting with 4-chloro-6-iodo-7H-
pyrrolo[2,3-d]pyrimidine (4), which in turn was obtained in two steps from the precursor
1
, see Scheme 1. Iodination of 1 gave a 9 : 1 mixture of compound 2 and 3, which upon
pyrrole deprotection gave the key intermediate 4 in 89 % yield. Two alternative routes
were used for C-4 and C-6 functionalisation of the pyrrolo[2,3-d]pyrimidine core. The
shortest method includes amination of 4 to yield the derivatives 7, followed by a Suzuki
coupling to give the target inhibitor structures. Although containing only two steps, this
pathway has the drawbacks of low reactivity in both the amination and in the Suzuki
coupling.
4

ACCEPTED MANUSCRIPT
Scheme 1. Synthesis of functionalized pyrrolo[2,3-d]pyrimidines. Reagents: SEM-
deprotection i) TFA, CH Cl , ii) sat. Aqueous NaHCO , THF.
2
2
3
5

ACCEPTED MANUSCRIPT
Therefore, we installed the N-2-trimethylsilylethyoxymethyl (N-SEM) protection on the
pyrrol to give the intermediate 5. With the SEM-protected pyrrolopyrimidine derivatives,
both the amination and the cross-coupling reaction proceeded more smoothly. Purification
by silica-gel chromatography was also more facile. The Suzuki reaction was performed
mainly with XPhos/Xphos Pd G2 as catalyst system. In contrast, standard conditions
using Pd(PPh ) proved inefficient also on typically good substrates. Low reactivity in the
3
4
Suzuki coupling was observed between pyridinylboronic acids and 6a-b. Besides pyrrole
protection, a change in catalyst system from XPhos/Xphos Pd G2 to the ferrocene based
PdCl (dppf) catalyst improved the processes. When more complex 6-aryl substitution
2
patterns were targeted, in-house synthesized arylboronic esters (8-11) were used, see
Scheme 2. These were synthesised by boronylation of aryl bromides with pinacolborane
using the PdCl (CH CN) / SPhos catalyst system [27] with yields ranging from 64 – 81
2
3
2
%. Importantly, this system successfully transformed aryl bromides with amines and
sterically hindered ortho-substituents to the corresponding arylboronic esters in good
yield. The required N-substitution pattern in 10 and 11 could be introduced either prior or
after the boronylation.
Scheme 2. Synthesis of boronic esters 8-11. Functionalisation to: 10: CH SO Cl, CH Cl ;
3
2
2
2
1
1: Ac O, CH Cl .
2 2 2
Some of the kinase inhibitors were made by post-modification of 6-aryl-7H-pyrrolo[2,3-
d]pyrimidin-4-amines structures having aldehyde substituents in the 3- and 4-position of
the 6-aryl group. These reactions involved conventional reduction using sodium
borohydride to the corresponding benzylic alcohols and reductive aminations. A two-step
reductive amination protocol via the imine followed by NaBH reduction was found more
4
efficient than the one-step method using NaBH(OAc)3.
2
.3 EGFR kinase inhibition
In our previous work, combination of 2-methoxy and 4-hydroxymethyl as 6-aryl
substituents resulted in favourable ligand-protein interactions and high EGFR inhibitory
properties [21]. Nevertheless, both the methoxy group and the polar benzyl alcohol
function could easily be afflicted by metabolic inactivation in vivo. Initially, the benzyl
alcohol at the 4-position of the 6-aryl ring was replaced with amide, methyl sulfone,
sulfone amide, methyl sulfone amide, acetamide, meta-carboxamide and pyridine, while
using (R)-1-phenylethan-1-amine as the C-4 substituent, see Figure 3. All these
derivatives displayed high inhibitory properties with IC50 below 1 nM. Especially
attractive was derivative 34 bearing a methylsulfone and the sulfonamide 37. Erlotinib in
6

ACCEPTED MANUSCRIPT
this assay had an IC value of 0.4±0.2 nM. We were also encouraged to find that the
5
0
pyridine derivative 39 was active. This allows for tuning of both metabolic stability and
solubility. Turning to the 2-position which has been insufficiently investigated previously,
enzymatic EGFR assays showed that ethoxy, isopropoxy and OCHF groups were all well
2
tolerated, while the trifluoromethoxy group resulted in lower potency. Four di-substituted
derivatives, 46-49, were also evaluated and displayed promising activity. However, only
1
1
in case of the N ,N -dimethylethane-1,2-diamine substituted 49 an obvious improvement
in activity was seen as compared to the mono-substituted analogues.
4
-substituted
O
S
O
S
O
O
S NH2
O
NH2
O
O
O
O
N
N
H
N
H
OH
3
4
35
36
37
38
39
40
EGFR Kinase data (nM)
0
.4 ± 0.0
0.7 ± 0.0
0.8 ± 0.0
0.3 ± 0.0
0.7 ± 0.0
0.6 ± 0.2
1.1 ± 0.1
2
-substituted
3-substituted
F
F F
F
F
H2N
O
O
O
O
O
H
N
N
N
4
2
43
44
45
41
EGFR Kinase data (nM)
N
H
0
.3 ± 0.1
0.6 ± 0.1
0.8 ± 0.0
4.3 ± 0.9
1.4 ± 0.4
di-substituted
O
O
O
O
O
O
NH2
O
S
O
N
H
HN
N
4
6
47
EGFR Kinase data (nM)
.6 ± 0.0 0.4 ± 0.0
48
49
0
0.7 ± 0.0
0.1 ± 0.0
Figure 3. Effect of variation of the 6-aryl group on the EGFR IC50 value (nM) for
pyrrolopyrimidines containing (R)-1-phenylethan-1-amine as C-4 substituent.
In our previous study on pyrrolopyrimidine-based EGFR inhibitors, an improvement of
potency was seen when substituting (R)-1-phenylethan-1-amine at C-4 with the (S)-2-
amino-2-phenylethan-1-ol group. As a starting point we evaluated the previously
discovered unsubstituted compound 50 and the 2-methoxy derivative 51 for metabolic
stability and permeability. Both compounds were found highly permeable by Caco-2
-6
-6
assays (Papp (50) = 30.3 × 10 cm/s, P
= 26.4 × 10 cm/s), while compound 50 had
app (51)
higher stability in human liver microsome assays (t1/2 (50) = 109 min, t1/2 (51) = 29.6 min).
Based on this background, we analysed additional derivatives having the (S)-2-amino-2-
7

ACCEPTED MANUSCRIPT
phenylethan-1-ol, see Figure 4. Several of the potency inducing 6-aryl groups identified in
the first series (Figure 3) was chosen, but also new substitution patterns: a deuterated
derivative, 52, designed to slow metabolism; the amide 56 as a bioisoster for the benzyl
alcohol function; compounds containing pyridines (60-62) and solubilizing tails (63-68)
to address solubility and protein binding. Again, most derivatives were highly potent, but
an additional increase in potency on introducing the (S)-2-amino-2-phenylethan-1-ol was
hardly seen. This could be due to enthalpy-entropy compensating effects, in which the
favourable binding energy is compensated by unfavourable entropy losses due to more
tight binding [28, 29].
Figure 4. Effect of variation of the 6-aryl group on EGFR IC50 value (nM) for
pyrrolopyrimidines containing (S)-2-amino-2-phenylethan-1-ol as C-4 substituent.
During ADME evaluation of the derivatives mentioned above, it became evident that the
more polar compounds had low permeability and a high efflux ratio. Incorporation of
fluorine adjacent to hydrogen bond donors has previously been shown as an efficient
strategy to improve permeability [30], thus derivatives 69-72 were prepared.
8

ACCEPTED MANUSCRIPT
Compounds 69 and 70 having a 4-fluoro substituent showed IC values of 1.2 and 1.6
5
0
nM, respectively. However, higher activity was observed for derivatives 71 and 72
1
2
2
substituted with a fluorine in position 3. Compound 73 with a N ,N ,N -trimethylethane-
,2-diamine group was also highly active.
1
Figure 5. EGFR IC value (nM) for pyrrolopyrimidines 74-78.
5
0
Five other compounds evaluated by kinase assay are shown in Figure 5. The (S)-2-
methoxy-1-phenylethan-1-amine containing compounds 74 and 75 had IC values of 0.5
50
and 1.3 nM, respectively. This scaffold was not further investigated due to the low
metabolic stability seen for compound 75. Compound 76, with (R)-3-amino-3-
phenylpropan-1-ol at C-4, was designed to improve contacts in the DFG motif by
hydrogen bonding to polar residues, and as compared to derivative 50, an improvement in
EGFR activity was seen. Furthermore, an additional gain in potency was obtained when
installing a 2-ethoxy substituent (compound 77). Unfortunately, this scaffold induced
hERG inhibition. The N-methyl substituted compound 78 was made to confirm SAR
information and as a model compound for ADME studies. As anticipated, removing the
pyrrole NH-group gave a high IC -value.
5
0
2
.4 SAR and in-silico evaluation
The SAR information based on this and our previous work on pyrrolopyrimidines is
summarised in Figure 6.
9

ACCEPTED MANUSCRIPT
Figure 6. Structure-activity relationships identified in this and previous study [19, 20].
Colour code: green: induce potency; black: minor effects; red: reduce potency.
In short, for Fragment A, the activity was highest when having a hydroxymethyl or
hydroxyethyl substituent as R , which is explained by hydrogen interactions to the DFG
1
motif. The discussion of the effect of hydroxyl group on activity is complicated by the
possibility of intramolecular hydrogen bonding. Compounds 50, 53 and 77 were subjected
to a conformational search using MacroModel and the OPLS3 force field. The search
gave conformations with lowest energy which displayed intermolecular hydrogen bonds
between the hydroxyl groups and the pyrimidine N-3 for all compounds tested, see
Supplementary Material. The intramolecular hydrogen bond would reduce desolvation
cost, but the required reorganisation of hydrogen bonding, as judged from docking and
dynamics, would somewhat limit the positive impact on binding. Unsubstituted phenyl or
fluorinated substituents are preferred in the aromatic part [20], while the activity is highly
dependent on correct stereochemistry as also seen in the thienopyrimidine series of
compounds [31]. Substitution of the 6-aryl group (Fragment B) improved EGFR potency
when having polar groups in the 4-position. The same substituent can also be placed in 3-
position. Evaluation of 2-substitution patterns confirmed the ethoxy substituent to have
the highest positive impact on activity, and also that larger substituents as isopropoxy and
difluoromethoxy were well tolerated. A drop in potency was noted for the
trifluoromethoxy analogue, which might indicate that minor size changes or electronic
properties are of importance. Finally, methylating the pyrrole NH group removed most
activity.
Selected derivatives were investigated for their binding interactions with EGFR by
induced-fit docking with the Schrödinger Maestro suite [32-34] using the crystal structure
4
WKQ at 1.85 Å resolution [35]. The docking revealed compounds 53, 65, 76, and 77 to
have very similar binding mode. Docking scores were -10.63, -11.00, -11.42 and -10.77
kcal/mol, respectively. The docked structure of compound 77 is shown in Figure 7. The
main interactions were between the pyrimidine N-1 involved in hydrogen bonding with
the backbone NH of Met793, and the pyrimidine N-3 nitrogen which engaged in water
1
0

ACCEPTED MANUSCRIPT
mediated hydrogen bonding. For derivatives 53, 76 and 77, a hydrogen bond between the
hydroxyl group in the R side chain and Asp855 was noticed, while compound 65 is
1
bonded to Thr854.
Figure 7. Docking of compound 77 using crystal structure 4WKQ [35].
In contrast, the highest docking score for compound 50 was for a pose in which the
pyrrole NH and the pyrimidine N-1 promote binding via a network of water molecules to
Thr790, Gln791 and Thr854. We also found a docking pose with similar binding mode as
that shown for 77 in Figure 7, but being 1.36 kcal/mol higher in energy. Both binding
modes are shown in the Supplementary Material. Although we cannot exclude the
possibility of an alternative binding mode, the rather consistent SAR data indicate that all
derivatives bind in a very similar way.
The poses with the best docking scores were then evaluated by 10 ns molecular dynamics
using the Desmond suite, the OPL-3 force field and the TIP4P solvent model [36]. An
interaction plot following dynamics for compound 77 is shown in Figure 8.
1
1

ACCEPTED MANUSCRIPT
Figure 8. Ligand-EGFR contacts for compound 77 during 10 ns of molecular dynamics.
Highlighted amino acids are within 5 Å distance from the docked ligand. (Dynamics for
compounds 50, 53 and 76 are given in the Supplementary Material). The colours indicate
residue type: green – lipophilic residues; red – acidic residues; blue – polar residues;
purple – basic residues. The lines indicate contacts with the enzyme. Only interactions
that occur more than 10 % of the 10 ns simulation time are shown. Ligand atoms that are
exposed to solvent are marked with grey spheres.
Dynamics showed all these compounds to interact via N-1 to Met793. The pyrrole NH in
case of compound 65 and 76 is found to interact with the oxygen of Met793. However, in
compounds 53 and 77 the pyrrole NH instead was indicated to have intramolecular
hydrogen bonding with the ether functionalities at the 6-aryl group. On one hand, if this
interaction is also present in the unbound state, desolvation will be less costly. On the
other side, an intramolecular hydrogen bond would prevent the pyrrole NH from taking
part in EGFR binding. However, a potency increase was seen going from unsubstituted to
2
-methoxy and 2-ethoxy substituted 6-aryls. It could be that this is solely due to weak
lipophilic interactions, which could be under estimated in docking protocols.
Alternatively, the substituent or the indicated intramolecular hydrogen bond could induce
a slight alteration of the position of the 4-amino group causing stronger binding in this
part.
Further, compounds 53, 65, 76 and 77 were indicated to have hydrogen bonding from N-3
via water molecules to different residues, mainly Thr854 and Cys775. In the 4-amino part
(Fragment A) differences were noticed as compared to the docked structures. Lys745
forms a cation-π interaction with the aromatic ring. This interaction was found to be most
prominent in the ether containing derivatives 53 and 77. Lys745 also together with
Asp855 and other residues form hydrogen bonding networks with the hydroxyl group at
the R side chain. The higher activity of 76 and 77 as compared to derivatives 50 and 53
1
could thus be due to these extended hydrogen bonding networks.
1
2

ACCEPTED MANUSCRIPT
High potency was seen for all derivatives containing solubilizing tails at the 4-position of
the 6-aryl group. An interaction plot from dynamics of derivative 65 is shown in Figure 9.
The N’,N’-dimethylethane-1,2-diamine substituent is indicated to be hydrogen bonded via
several water molecules to Asp800, Cys797, and Pro794, which might explain the high
activity of these derivatives.
Figure 9. Ligand-EGFR contacts for compound 65 during 10 ns of molecular dynamics.
Highlighted amino acids are within 5 Å distance from the docked ligand. The colours
indicate residue type: green – lipophilic residues; red – acidic residues; blue – polar
residues; purple – basic residues. The lines indicate contacts with the enzyme. Only
interactions that occur more than 10 % of the 10 ns simulation time are shown. Ligand
atoms that are exposed to solvent are marked with grey spheres.
2
.5 ADME profiling
A summary of the ADME profile of some of the drug candidates is given in Table 1,
while additional data for other compounds is provided in the Supplementary Material.
Table 1. Summary of key ADME data for the new EGFR inhibitors. Compounds are
sorted by the Papp values. Additional data is provided in Supplementary Material.
H
N
N
N
N
O
O
O
O
N
N
R1
H
HN
Erlotinib
Protein
P
A-B
Efflux
ratio
HLM
t1/2 [min]
Solubility
app
Comp.
Structures
binding
-6
-1 a)
a)
b)
d)
(
10 cms )
(µM)
c)
(
fu)
Erlotinib
38.8
0.80
1.17
31.7
20
0
.12
5
0
30.3
109
0.02
>100
1
3

ACCEPTED MANUSCRIPT
5
5
5
1
2
3
26.4
23.4
24.0
24.5
1.34
1.02
0.79
1.32
0.88
3.30
4.62
5.90
6.77
29.6
44.1
22.9
67.0
23.7
87.8
95.4
103
0.02
<0.01
-
10.5
11.5
37.5
12.9
12.2
>100
>100
>100
>100
7
7
6
7
0.02
0.01
0.23
0.68
-
2
3.4
7
6
1
5
2.0
2.1
1.6
6
4
9
9
2
.7
46.9
-
a) Determined by Caco-2 assay.
b) Time study using pooled human liver microsomes (HLM) at 3 µM concentration.
c) Protein binding is given as fraction unbound (fu).
d) Solubility was measured as turbidimetric aqueous solubility.
High Caco-2 permeability (P ) and efflux ratio close to unity were seen for the less polar
app
compounds 50-53, 76 and 77. Derivatives containing the N’,N’-dimethylethane-1,2-
diamine solubilizing tail (49, 65, 69 and 71) had a 10 fold lower permeability in the A-B
direction, but also reasonable efflux values. Thus, the major challenge with the
pyrrolopyrimidines as EGFR inhibitors was related to balancing the potency versus the
permeability of the compounds. Many of the most potent inhibitors had a high efflux,
mainly due to a low Papp in the A-B direction. This was the case for compounds bearing
amide or sulfonamide type substituents, which could indicate that hydrogen bond
acceptors lowers permeability. High efflux was also seen in a series of N5-substituted
pyrrolo[3,2-d]pyrimidine EGFR inhibitors containing amides and the methylsulfonyl
functional group [37].
A rational prediction of efflux effects is challenging due to the multitude of possible
transporter proteins involved [38, 39], and for some previously investigated classes of
pyrrolopyrimidines no clear system can be seen [40]. However, in this homogenous series
of compounds (28 investigated for permeability, Supplementary Material) correlation
with physiochemical properties gave a decent fit with polar surface area, see Figure 10.
One outlayer was identified, compound 68 containing a piperazine group in the
1
4

ACCEPTED MANUSCRIPT
solubilzing tail. The low efflux ratio value is due to extremely low Papp in the B-A
direction, indicating this compound to be a substrate or inhibitor of specific ABC
transporter.
Figure 10. Efflux ratio (Caco-2) as a function polar surface area (n = 28).
The metabolic stability of the EGFR inhibitors and selected model compounds (29
derivatives) were evaluated in human liver microsome (HLM) assay at 3 µM test
concentration. A graphical representation of the key findings in this part of the study is
shown in Figure 11. As a point of reference, the unsubstituted phenyl analogue 50 had a
half-life of 109 min. The 2-methoxy derivative 51 had a stability comparable to that of
Erlotinib, which is somewhat on the low side [41]. Higher stability was seen for the d3-
methoxy analogue 52, possibly due to a kinetic isotope effect [42]. Based on previous
reports we anticipated that the difluoromethoxy [43] or the ethoxy substituted derivatives
could be used as more stable bioisostere for the methoxy analogue. Whereas compounds
with both substituents maintained a high EGFR activity, no improvement in half-life was
detected for the difluoromethoxy compound 44 and the ethoxy compound 53. Upon
varying the 4-substituent of the 6-aryl group (R ) most amides and benzyl alcohol
4
containing derivatives had higher stability than the reference 50. Moreover, although the
2
-methoxy group is a metabolic soft spot, including substituents in position 4 increases
the stability of all molecules containing this functionality. Compounds containing
solubilizing tails generally had a high stability. The most labile compound of these was
the piperidine substituted analogue 67 (half-life: 40 min). Fluorine was inserted in an
attempt to improve the stability and permeability of the benzylalcohol and benzylamine
containing derivatives (compounds 69-72). Only in one case a direct comparison of
stability could be made, indicating that an adjacent fluorine has a slight negative impact
on stability. A well-known trick to improve stability of drug candidates is to reduce the
electron content of the aromatic ring by insertion of heteroatoms [44]. Thus, also in this
1
5

ACCEPTED MANUSCRIPT
study replacing the 2-methoxyphenyl with the corresponding pyridine derivative 62
increase the half-life considerably.
Figure 11. Illustration of the relative stability and metabolic soft spots based on HLM
data.
Concerning the effect of the 4-amino group on metabolic stability a few guidelines could
be drawn. Comparing data for three analogues containing the (R)-1-phenylethan-1-amine
fragment (compound 37, 48 and 49) with the stability of the corresponding (S)-2-amino-
2
-phenylethan-1-ol containing compounds (54, 58 and 65) indicate that the latter group
improves stability. The highly active EGFR inhibitors based on (R)-2-amino-2-
phenylpropanol, 76 and 77, had a half-life of 67 and 23 minutes, respectively, which is
lower than compound 50. The (S)-2-methoxy-1-phenylethan-1-amine unit in compound
7
5 was found especially labile. The N-methyl derivative 78 is not an EGFR inhibitor, but
was included as a model substance for evaluating ADME properties. The stability was
considerably lower than the reference compound 50, indicating the N-methyl group as the
metabolic soft spot. Detailed data is provided in the Supplementary Material.
To attain on target activity in vivo the drug candidates must possess high bioavailability.
A graphical method for evaluating bioavailability (% F) based on Caco-2 and HLM data
have been developed by Mandagere et al. [45], which categorise molecules into high,
medium and low bioavailability classes. We adopted the approach on our data, see Figure
1
2.
1
6

ACCEPTED MANUSCRIPT
Figure 12. Estimation of bioavailability using a graphical oral bioavailability map.
The method predicts (S)-2-amino-2-phenylethan-1-ol containing molecules possessing
nonpolar substituents (50-53) and the (R)-2-amino-2-phenylpropan-1-ol containing
structures 76 and 77 to have medium bioavailability. Note that compound 78 has very low
EGFR potency.
High protein binding was seen for several of the candidate structures. However, this is
also the case for the EGFR/HER2 inhibitor Lapatinib [16], and 45 % of newly approved
drugs had protein binding above 95 % [46]. By including a solubilizing tail in the
inhibitor structure, protein binding was efficiently reduced, but the permeability drops
probably as a result of higher polarity.
The hERG potassium voltage-gated ion channel is essential for cardiac re-polarisation.
Thus, inhibiting hERG can give rise to potentially fatal toxicity. Ten compounds were
evaluated by IC titration. For eight of these IC was > 25 µM, see Table 2. However,
5
0
50
upon including the 1-phenyl-2-amino-3-propanol scaffold as present in compound 76 and
7 higher activity was seen, indicating possible toxic issues with these agents.
7
Table 2. hERG inhibition data for the new EGFR inhibitors. Additional data is provided
in Supplementary Material.
1
7

ACCEPTED MANUSCRIPT
Comp.
Erlotinib
50
51
52
65
71
76
77
hERG
inhibition
>
25
>25
>25
>25
>25
22
11
>
25 (n=17)
(
n=12) (n=15) (n=12) (n=14) (n=14) (n=17) (n=12)
a)
IC (µM)
5
0
a)
hERG channel inhibition (IC50) was done at concentrations 0.008, 0.04, 0.2, 1, 5 and 25 ꢀM.
2
.6 Kinase selectivity profile
The kinase selectivity profile of five compounds (50, 51, 53, 71 and 76) representing
different structural elements was evaluated in assays towards 50 additional kinases at 500
nM test concentration. The selectivity as analysed by the Gini-method [47] was rather
similar (Gini coefficient from 0.57-0.62). Another way of quantifying selectivity, which
assumes that low inhibitory activity is clinically irrelevant, is the so called selectivity
score [48]. It is calculated by dividing the number of kinases showing inhibition above a
set limit by the total number of kinases evaluated. Thus, a non-selective inhibitor has a
selectivity score close to unity, while a selective inhibitor has a score close to zero.
Employing 50 % inhibition as the threshold for the calculation, the compounds follows
the selectivity order: 76 (0.08) > 53 (0.10) > Erlotinib (0.16) > 50 (0.18) > 51 (0.20) > 71
(0.25). Figure 13 illustrates the degree of inhibition for 76, 50 and 71 towards 15 of the
kinases sorted by the activity displayed by the least selective compound 71.
Figure 13. Inhibition profile of compounds 71, 50, 76 and Erlotinib towards 15 kinases
sorted by the activity of compound 71. Data for additional kinases for these compounds,
5
1 and 53 is provided in Supplementary Material)
Evidently, the solubilizing tail in 71 increase activity towards a number of kinases. This
was also noted in one of our structurally related furopyrimidine compound [49]. On one
1
8

ACCEPTED MANUSCRIPT
hand this might lead to undesired off-target related toxicity, on the other side having
activity towards several kinases in the same pathway might be beneficial in a therapeutic
setting.
2
.7 Cell assays
The new EGFR inhibitors were compared with the reference drug Erlotinib in
proliferation studies with two cell lines; A-431 harbouring overexpressed EGFR-wild
L858R
type, and a genetically engineered cancer model cell, Ba/F3-EGFR
, containing the
activating L858R mutant form (see Table 3). Except compound 68 (IC 396 nM) all
5
0
L858R
compounds were highly sensitive towards the Ba/F3-EGFR
cells with IC50 in the
range as Erlotinib (IC 75-158 nM). This proves on-target activity also in cells. In assay
5
0
towards the more complex A-431 cell line, the activity was high for derivative 50, 51 and
6, but considerable lower than Erlotinib for the other derivatives tested. Interestingly,
7
compound 71 displaying excellent EGFR activity in enzymatic studies and also high
inhibition towards a number of kinases including HER2 and HER4, had a mediocre
activity towards the A-431 cell line.
Table 3. Cell proliferation study of selected pyrrolopyrimidines towards Ba/F3-
L858R
EGFR
and A-431 cells.
Ba/F3-
EGFR
A-431
(µM)
Comp.
Structure
L858R
a)
b)
(nM)
0
.4±0.1
Erlotinib
87±5
(n=9)
4
5
5
5
6
9
0
1
3
8
90±3
5.0±0.0
0.3±0.1
c)
c)
127±29
134±22
98±1
(n=6)
d)
0.4±0.1
9.2±3.9
6.2±3.1
396±50
1
9

ACCEPTED MANUSCRIPT
7
7
7
7
7
1
2
3
6
7
142±12
>10
158±44
77±12
75±14
99±1
5.8±4.6
>10
0.8±0.3
4.9±4.2
a)
b)
c)
Each IC50 is the result of three independent replicates.
Unless otherwise stated IC50 is the result of three replicates.
Data is taken from ref. [20] (Erlotinib had an IC50 of 142±65 nM).
Data is taken from ref. [20] (Erlotinib had an IC50 of 0.4±0.1 µM).
d)
Based on a trade-off between enzymatic and cellular activity and ADME properties the
previously identified compound 50 appears as the most promising drug candidate. Table 4
summarises other cell proliferation data for compound 50.
Table 4. Cell proliferation data of compound 50 and Erlotinb towards various cancer cell
lines.
IC (µM)
5
0
Cancer type
Cell line
PC-9
Erlotinib
< 0.1
Comp. 50
< 0.1
Lung
A-549
AU-565
C-33A
CAL-27
FaDu
>100
>100
a)
a)
Breast
3.3±0.6
2.5±0.4
a)
a)
Ovarian
0.9±0.0
0.7±0.0
a)
a)
Head&Neck
1.3±0.6
2.7±0.6
>11
>33
a)
a)
Leukaemia
Pancreatic
K-562
BxPC3
55±9
15±2
1.9±0.2
15±1
a) Data taken from ref. [20]
2
0

ACCEPTED MANUSCRIPT
Erlotinib and compound 50 were highly potent towards the PC-9 lung cancer cells
harbouring the EGFR E746-A750 deletion mutation. In contrast, both compounds were
found inactive towards A-549, which have wild type EGFR, but mutated RAS [50]. Fairly
similar potency was seen towards the AU-565 breast cancer cell line and the cervix
carcinoma cell line C-33A [20]. The decent potency detected for the latter cell line is
surprising giving this cell line’s low expression of EGFR [51-53], indicating that
alternative cytotoxicity mechanisms might be operating. In line with what seen for
structurally related analogues [21], compound 50 was moderately active towards CAL-27
cells, but inactive towards the FaDu cells. Compound 50 also had a low potency towards
the leukaemia K-562 cell line and the pancreatic cell line BxPC3. Overall the cell
proliferation studies on compound 50 indicate that the major use of this compound as
single agent is in EGFR driven diseases.
3
. Conclusion
The goal of this work was to identify pyrrolopyrimidine-based EGFR inhibitors for
medical use. Thus, starting from our own lead structure, we have performed a SAR study
involving 43 new structures. The SAR study revealed that ortho-ethoxy groups and amine
based solubilizing tails in para-position of the 6-aryl moiety, and a (R)-3-amino-3-
phenylpropan-1-ol at C-4 boosted potency. Based on this collection of new structures and
two previously identified compounds, various ADME assays were used to identify the
most promising drug candidate. Permeability was investigated by the Caco-2 method for
2
8 of the inhibitors. Overall, the efflux ratio correlated well with polar surface area of the
molecules, and compounds bearing polar functionalities at the 6-aryl ring, especially
amides and sulphonamides, increased drug efflux ratio and reduced permeability.
Metabolic stability (HLM), investigated for 29 of the derivatives, showed that sites most
prone to metabolism were the ortho-methoxyphenyl group at C-6, and (S)-2-methoxy-1-
phenylethan-1-amine at C-4. The metabolic stability of the ortho-methoxy compound
could be increased by bioisosteric substitution with a deuteromethoxy group, whereas no
improvement was seen using the difluoromethoxy substituent. hERG inhibition was
generally low, but two compounds bearing the (R)-3-amino-3-phenylpropan-1-ol group at
C-4, had IC < 25 µM, indicating toxicity issues with this scaffold. Ten compounds were
5
0
L858R
subjected to proliferation assays using Ba/F3-EGFR
and A-431 cells. Whereas high
activity was seen for the former cell line, only a few was potent in the A-431 cells.
Further, five of the EGFR inhibitors were compared in assays towards a panel of 50
kinases, revealing that amine containing solubilizing tails at para-position of the 6-aryl
group reduced selectivity. On background of all these studies, (S)-2-phenyl-2-((6-phenyl-
7
H-pyrrolo[2,3-d]pyrimidin-4-yl)amino)ethan-1-ol (50) appeared as the most promising
L858R
drug candidate. The high potency displayed in Ba/F3-EGFR
reporter cells, A-431
and PC9 cell proliferation studies indicate this compound to have potential therapeutic use
in EGFR driven diseases.
2
1

ACCEPTED MANUSCRIPT
4
. Experimental
4
.1 General
nd
Xhos, 2 generation XPhos, NaBH , (R)-1-phenylethan-1-amine and (S)-2-amino-2-
4
phenylethan-1-ol and the arylboronic acid derivatives except 2-(3-(4,4,5,5-tetramethyl-
1
,3,2-dioxaborolan-2-yl)phenyl)acetamide (Activate Scientific) and 8-11 (in house
prepared) were from Sigma Aldrich. (R)-3-Amino-3-phenylpropan-1-ol was from
Fluorochem.. Silica-gel column chromatography was performed using silica-gel 60A
from Fluka, pore size 40-63 µm. Celite 545 from Fluka was also used. (S)-2-Methoxy-1-
phenylethanamine [54], and compounds 1, 7a-b, 7d, 31-33, 50 and 51 [20], were prepared
and characterized in other studies from our laboratory. The Supplementary Material
contains experimental and analytical data on synthesis of 4-bromo-3-methoxyaniline, 1-
bromo-2-(methoxy-d )benzene and compounds 3 and 8-11.
3
4
.2 Analyses
1
13
H and C NMR spectra were recorded with Bruker Avance 600 and 400 spectrometer
1
9
operating at 600/400 MHz and 150/100 MHz, respectively. F NMR was performed on a
Bruker Avance 600 operating at 564 MHz. For H and C NMR chemical shifts are in
ppm rel. to tetrametylsilane (TMS) calibrated using DMSO-d or TMS, while for
1
13
1
9
F
6
NMR the shift values were calibrated using hexafluorobenzene. Coupling constants are in
hertz. The pyrrole NH signal (δ: 12.2-10.2 ppm), the NH at C-4 (δ: 7.9-7.7 ppm and the
hydroxyl group at R in compounds 52 - 73 and 78 (δ: 4.99-4.95 ppm) disappeared after
1
D O exchange. Other exchangeable protons are mentioned for each specific compound.
2
HPLC (Agilent 110-Series) with a G1379A degasser, G1311A Quatpump, G1313A ALS
autosampler and a G1315D Agilent detector (230 nm) was used to determine the purity of
the synthesised compounds. All compounds evaluated for EGFR inhibitory potency had a
purity of ≥96%. Conditions: Poroshell C18 (100×4.6 mm) column, flow rate 0.8 mL/min,
elution starting with water/CH CN (90/10), 5 min isocratic elution, then linear gradient
3
elution for 35 min ending at CH CN/water (100/0). The software used with the HPLC
3
was Agilent ChemStation. Accurate mass determination (ESI) was performed on an
Agilent G1969 TOF MS instrument equipped with a dual electrospray ion source.
Accurate mass determination in positive and negative mode was performed on a "Synapt
G2-S" Q-TOF instrument from Waters. Samples were ionized by the use of an ASAP
probe, no chromatography separation was used before the mass analysis. FTIR spectra
were recorded on a Thermo Nicolet Avatar 330 infrared spectrophotometer. All melting
points are uncorrected and measured by a Stuart automatic melting point SMP40
apparatus.
4
.3 In vitro EGFR (ErbB1) inhibitory potency
2
2

ACCEPTED MANUSCRIPT
The compounds were supplied in a 10 mM DMSO solution, and enzymatic EGFR
(ErbB1) inhibition potency was determined by Reaction Biology Corp. using their
biochemical kinase assay at 10 µM ATP concentration. All compounds were first tested
for their inhibitory activity at 100 nM in duplicates. The potency observed at 100 nM was
used to set starting point of the IC titration curve, in which three levels were used 100,
5
0
1
000 or 10000 nM. The IC50 values reported are based on the average of at least 2
titration curves (minimum 20 data points), and were calculated from activity data with a
four parameter logistic model using SigmaPlot (Windows Version 12.0 from Systat
Software, Inc.) Unless stated otherwise the ATP concentration used was equal to K . The
m
average standard deviation for single point measurements were <4 %.
4
.4 In vitro kinase panel
The compounds were supplied in a 10 mM DMSO solution, and enzymatic kinase
®
inhibition potency was determined by Invitrogen (LifeTechnology) using their Z’-LYTE
assay technology [55], at 500 nM in duplicates. ATP concentration used was equal to Km,
except when this service was not provided and other concentrations had to be used.
4
.5 ADME and cell studies
ADME studies, including solubility, Caco-2 assay, human liver microsome metabolic
assay, hERG inhibition and protein binding were performed as previously described [21].
Cell proliferation studies with the PC-9, FaDu and BXPC-3 cells were performed as
L858R
described in reference [21], while protocols for the Ba/F3 EGFR
studies can be found in reference [20].
cell proliferation
Proliferation study with the A-549 cell line was performed by Reaction Biology Corp:
The A549 cell line was obtained from American Type Culture Collection (Manassas,
VA). Staurosporine was obtained from Selleckchem. Cell Titer-Glo® Luminescent cell
viability assay reagent was obtained from Promega (Madison, WI). A549, cell line was
cultured in F-12K medium supplemented with 10 % FBS. 100 µg/ml penicillin and 100
µg/ml streptomycin. Cultures were maintained at 37 °C in a humidified atmosphere of 5
%
CO and 95 % air. The test compound, Erlotinib and Staurosporine (positive control)
2
were all dissolved in DMSO in 10 mM stock. Culture medium (10 µl) was added to each
well of 384 well cell culture plates. The compounds were diluted in a source plate in
DMSO at 3 fold serial dilutions starting at 10 mM, total 10 doses. The compounds (0.25
µl) were delivered from source plate to each well of the cell culture plates by Echo 550.
Then, 250 µl of culture medium containing 5000 cells were added to the wells of the cell
culture plates. The cells were incubated with the compounds at 37 °C, 5 % CO for 72
2
hours. 25 µl of Cell Titer Glo reagent (25 µl) was added to each well according to the
instruction of the kit. The contents were mixed on an orbital shaker for 2 minutes and
incubated at room temperature for 10 minutes to stabilize luminescent signal.
Luminescence was recorded by Envision 2104 Multilabel Reader (PerkinElmer, Santa
Clara, CA). The maximum luminescence for each cell line in the absence of test
2
3

ACCEPTED MANUSCRIPT
compound, but in the presence of 0.4 % DMSO, was similarly recorded after incubation
for 72 hours. The number of viable cells in the culture was determined based on
quantitation of the ATP present in each culture well. The percentage growth after 72
hours (%-growth) was calculated as follows: 100 % × (luminescence t = 72 hours /
luminescence untreated, t = 72 hours).
4
.6 General procedures
4
.6.1 General procedure for thermal amination of pyrrolo[2,3-d]pyrimidines
The 4-chloropyrrolopyrimidine (4 or 5) (0.12 - 2.0 g) was mixed with the selected amine
2-3 eq) and n-BuOH (2 - 30 mL) and agitated at 145 °C for 1 - 24 h under N2
(
atmosphere. The mixture was then cooled to ambient temperature, concentrated in vacuo,
diluted with water and extracted with EtOAc (2 × 30 - 100 mL). The combined organic
phases were washed with saturated aq. NaCl solution (15 - 60 mL), dried over anhydrous
Na SO , filtered and concentrated in vacuo. The crude material was purified by silica-gel
2
4
column chromatography.
4
.6.2 General procedure for Suzuki cross-coupling on pyrrolo[2,3-
d]pyrimidines
The 6-iodopyrrolopyrimidine (6a - 6c, 7a - 7b or 7d) (50 - 350 mg) was mixed with the
nd
selected arylboronic acid (1.2 eq), fine powdered K CO (3 eq), XPhos (5 mol %) / 2
2
3
generation XPhos precatalyst (5 mol %) system or PdCl (dppf) (5 mol %) and mixture
2
with degassed 1,4-dioxane/H O (1/1 by vol. %, 2-8 mL). The reaction was then stirred at
2
1
00 °C for 0.5 - 10 hours under N atmosphere. The solvent was removed and the product
2
was diluted with H O (25 - 100 mL) and extracted with EtOAc (50 - 120 mL), several
2
times if required. The combined organic phases were washed with saturated aq. NaCl
solution (30 mL), dried over anhydrous Na SO , filtered and concentrated in vacuo.
2
4
Purification was performed as described for each individual compound.
4
.6.3 General procedure for reductive amination of pyrrolo[2,3-d]pyrimidines
The 6-aryl aldehydes (19, 23, 24, 31 - 33) (16.6 - 100 mg) were dissolved in anhydrous
CH Cl (1 - 5 mL) and the respective amine (1.2 eq) was added and stirred at 20 °C for 1
2
2
1
–
4 hours until full conversion to the imine as determined by H NMR spectroscopy.
Solvent was removed in vacuo and the crude product was dissolved in MeOH (4-10 mL)
and NaBH (1.8 – 2.3 eq) was added and stirred for 2 - 4 hours until completed reduction.
4
The reaction mixture was diluted with H O (10 - 30 mL) and the mixture was extracted
2
with EtOAc (40 - 100 mL). The combined organic phases were washed with saturated aq.
NaCl (10-30 mL), dried over anhydrous Na SO , filtered and concentrated in vacuo.
2
4
Purification was performed as described for each individual compound.
2
4

ACCEPTED MANUSCRIPT
.6.4 General procedure for SEM-deprotection of pyrrolo[2,3-d]pyrimidines
4
Compounds (50 - 200 mg) were dissolved in CH Cl (10 - 20 mL) and TFA (2 - 4 mL)
2
2
1
was added and stirred at 50 °C for 3-6 hours ours until full conversion determined by H
NMR at which time the solvent was removed in vacuo. The crude material was dissolved
in THF (10 - 20 mL) and saturated aq NaHCO (10 - 20 mL) solution was added and
3
stirred for 6 - 12 hours. The mixture was then extracted with EtOAc (50 - 150 mL) and
the combined organic phases were washed with saturated aq NaCl (10 - 40 mL), dried
over anhydrous Na SO , filtered and concentrated in vacuo. Purification was performed
2
4
as described for each individual compound.
4
.7 Synthesis of intermediates
4
.7.1 4-Chloro-6-iodo-7H-pyrrolo[2,3-d]pyrimidine (4) [20].
4
-Chloro-7-(phenylsulfonyl)-7H-pyrrolo[2,3-d]pyrimidine (1) [20], (5.42 g, 18.5 mmol)
was iodinated as previously described [20]. This gave a 9 : 1 mixture of 2 and 3 (6.0 g)
which was mixed with THF (125 mL) and 5 M NaOH solution in MeOH (21 mL). After 2
hours stirring at room temperature, a saturated aqueous NH Cl-solution (125 mL) was
4
added and the mixture concentrated. The formed precipitate was collected by filtration
and washed with water. Trituration from boiling acetonitrile (1 g/10 mL) gave 3.96 g
1
(
(
14.2 mmol, 77 %) of 4 as a white solid, mp 219 °C (dec.) (lit.[20] 220 °C); H NMR
1
400 MHz, DMSO-d ) δ: 12.57 (s, 1H), 8.51 (s, 1H), 6.88 (s, 1H). The H NMR data is in
6
agreement with that previously reported [20].
4
.7.2 4-Chloro-6-iodo-7-((2-(trimethylsilyl)ethoxy)methyl)-7H-pyrrolo[2,3-
d]pyrimidine (5) [56].
4-Chloro-6-iodo-7H-pyrrolo[2,3-d] (2.02 g, 7.24 mmol) and sodium hydride (210 mg,
8.45 mmol) were added dry DMF (70 mL) under nitrogen atmosphere and cooled to 0 °C.
2-(Trimethylsilyl)ethoxymethyl chloride (1.69 mL, 9.36 mmol) was added dropwise over
30 min at 0 °C and stirred at 22 °C for 1 hour. The reaction mixture was added water (150
mL) and extracted with EtOAc (2 x 150 mL), dried over Na SO and concentrated in
2
4
vacuo. Column chromatography on silica gel (EtOAc/n-pentane 1/1, R = 0.78) gave 5 in
f
1
2
.43 g (5.92 mmol, 81 %) as a clear oil that solidified upon drying in vacuo. H NMR
(
400 MHz, DMSO-d ) δ: 8.63 (s, 1H), 7.12 (s, 1H), 5.62 (s, 2H), 3.54 (t, J = 8.4, 2H),
6
1
3
0
1
4
.83 (t, J = 8.4, 2H), -0.10 (s, 9H); C NMR (100 MHz, DMSO-d ) δ: 152.5, 150.7,
6
49.1, 118.5, 109.7, 91.5, 73.4, 66.1, 17.0, -1.4 (3C); HRMS (APCI/ASAP, m/z):
+
09.9955 (calcd. C H N OSiClI, 409.9952, [M+H] ).
1
2
18
3
2
5

ACCEPTED MANUSCRIPT
.7.3 (R)-6-Iodo-N-(1-phenylethyl)-7-((2-(trimethylsilyl)ethoxy)methyl)-7H-
pyrrolo[2,3-d]pyrimidin-4-amine (6a)
4
Compound 6a was prepared as described in Section 4.6.1, starting with 5 (506 mg, 1.24
mmol) and (R)-1-phenylethan-1-amine (449 mg, 3.71 mmol, 3 eq.). The reaction time was
3
0
hours. Purification by silica-gel column chromatography (n-pentane/EtOAc, 8/2, R =
f
.49) gave 456 mg (0.923 mmol, 75 %) of a pale foam; HPLC purity: 98 %, t = 30.3
R
ꢂꢃ
1
min; [ꢀ] = -195.6 (c 1.00, DMSO); H NMR (600 MHz, CDCl -TMS) δ: 8.27 (s, 1H),
ꢁ
3
7
5
.41 - 7.40 (m, 2H), 7.36 - 7.33 (m, 2H), 7.28 - 7.25 (m, 1H), 6.66 (s, 1H), 5.57 (s, 2H),
.52 - 5.47 (m, 1H), 3.58 - 3.55 (m, 2H), 1.64 (d, J = 6.8, 3H), 0.98 - 0.89 (m, 2H), -0.06
13
(s, 9H); C NMR (150 MHz, CDCl -TMS) δ: 154.2, 152.6, 152.4, 143.8, 128.9 (2C),
3
-
1
27.5, 126.2 (2C), 109.5, 105.2, 77.8, 73.2, 66.5, 50.2, 22.9, 17.9, -1.3 (3C); IR (neat, cm
1
)
: 3273, 2945, 1594, 1450, 1295, 1080, 833, 746, 697; HRMS (APCI/ASAP, m/z):
+
4
95.1082 (calcd. C H N OSiI, 495.1077 [M+H] ).
2
0
28
4
4
.7.4 (S)-2-((6-Iodo-7-((2-(trimethylsilyl)ethoxy)methyl)-7H-pyrrolo[2,3-
d]pyrimidin-4-yl)amino)-2-phenylethan-1-ol (6b)
Compound 6b was prepared as described in Section 4.6.1, starting with 5 (1.02 g, 2.05
mmol) and (S)-2-amino-2-phenylethan-1-ol (844 mg, 6.15 mmol, 3 eq.). The reaction
time was 6 hours. Purification by silica-gel column chromatography (EtOAc/n-pentane,
6
2
1
5
2
1
1
/4, R = 0.24) gave 896 mg (1.76 mmol, 86 %) of a white foam; HPLC purity: 99 %, t =
f R
ꢂꢃ
1
5.6 min; [ꢀ] = -187.2 (c 1.00, DMSO); H NMR (400 MHz, DMSO-d ) δ: 8.04 (s,
ꢁ
6
H), 7.83 - 7.81 (m, 1H), 7.41 - 7.39 (m, 2H), 7.31 - 7.28 (m, 2H), 7.22 - 7.17 (m, 2H),
.48 - 5.37 (m, 1H), 5.45 (s, 2H), 4.97 - 4.94 (m, 1H), 3.76 - 3.67 (m, 2H), 3.51 - 3.47 (m,
13
H), 0.82 - 0.78 (m, 2H), -0.11 (s, 9H); C NMR (100 MHz, DMSO-d ) δ: 154.4, 151.9,
6
51.4, 141.6, 128.0 (2C), 126.9 (2C), 126.7, 110.2, 104.9, 79.5, 72.5, 65.4, 64.9, 56.1,
7.1, -1.3 (3C); HRMS (APCI/ASAP, m/z): 511.1031 (calcd. C H N O SiI, 511.1026
2
0
28
4
2
+
[
M+H] ).
4
.7.5 (R)-3-((6-Iodo-7-((2-(trimethylsilyl)ethoxy)methyl)-7H-pyrrolo[2,3-
d]pyrimidin-4-yl)amino)-3-phenylpropan-1-ol (6c)
Compound 6c was prepared as described in Section 4.6.1, starting with 5 (152 mg, 0.371
mmol) and (R)-3-amino-3-phenylpropan-1-ol (121 mg, 0.800 mmol, 2 eq.). The reaction
time was 20 hours. Purification by silica-gel column chromatography (EtOAc/n-pentane,
6
/4, R = 0.24) gave 190 mg (0.361 mmol, 97 %) of a white foam; HPLC purity: 99 %, t
f R
ꢂꢃ
1
=
25.9 min; [ꢀ] = -190.1 (c 1.00, DMSO); H NMR (400 MHz, CDCl -TMS) δ: 8.28 (s,
ꢁ
3
1
2
2
9
H), 7.42 - 7.36 (m, 4H), 7.34 - 7.29 (m, 1H), 6.66 (s, 1H), 5.61 - 5.55 (m, 1H), 5.59 (s,
H), 5.23 - 5.22 (m, 1H), 3.77 - 3.75 (m, 1H), 3.70 - 3.64 (m, 1H), 3.60 - 3.56 (m, 2H),
.31 - 2.23 (m, 1H), 2.03 - 1.95 (m, 1H), 1.61 (s, br, 1H), 0.94 - 0.90 (m, 2H), -0.05 (s,
1
3
H); C NMR (100 MHz, CDCl -TMS) δ: 154.7, 152.2, 152.0, 142.3, 129.2 (2C), 128.0,
3
2
6

ACCEPTED MANUSCRIPT
26.8 (2C), 109.2, 105.2, 78.2, 73.3, 66.6, 58.5, 52.0, 39.4, 17.9, -1.3 (3C); HRMS
1
(
+
APCI/ASAP, m/z): 524.1093 (calcd. C H N O SiI, 524.1098 [M+H] ).
2
1
30
4
2
4
.8 SEM-protected pyrrolopyrimidines
4
.8.1 (R)-N-(1-Phenylethyl)-6-(pyridin-4-yl)-7-((2-
trimethylsilyl)ethoxy)methyl)-7H-pyrrolo[2,3-d]pyrimidin-4-amine (12)
(
Compound 12 was prepared as described in Section 4.6.2, starting with 6a (150 mg, 0.30
mmol), pyridin-4-ylboronic acid (52 mg, 0.430 mmol) and using PdCl (dppf) (10 mg,
2
0
.014 mmol, 0.05 eq.) as catalyst The reaction time was 2 hours. Purification by silica-gel
column chromatography (EtOAc/n-pentane, 2/1, R = 0.32) gave 123 mg (0.280 mmol, 61
f
%
) of a yellow oil. HPLC purity: 99 %, t = 30.8 min; [ꢀ] ꢂꢁ ꢃ = -189.9 (c 1.00, DMSO);
R
1
H NMR (600 MHz, CDCl -TMS) δ: 8.67 - 8.66 (m, 2H), 8.40 (s, 1H), 7.66 - 7.65 (m,
3
2
H), 7.44 - 7.43 (m, 2H), 7.37 - 7.34 (m, 2H), 7.29 - 7.24 (m, 1H), 6.62 (s, 1H), 5.61 -
5
.57 (m, 3H), 5.59 - 5.53 (m, 1H), 3.79 - 3.76 (m, 2H), 1.67 (d, J = 6.8, 3H), 0.99 - 0.96
1
3
(m, 2H), -0.02 (s, 9H); C NMR (150 MHz, DMSO-d ) δ: 155.8, 153.3, (2C), 150.4,
6
(
2
6
2C), 143.9, 139.3, 135.5, 128.9, (2C), 127.5, 126.2, 122.9, 103.0, 100.5, 70.8, 66.9, 50.5,
-1
2.9, 18.1, -1.3 (3C); IR (neat, cm ): 3262, 2945, 2353, 1591, 1468, 1303, 1073, 730,
+
97; HRMS (APCI/ASAP, m/z): 446.2376 (calcd. C H N OSi, 446.2376 [M+H] ).
25
32
5
4
.8.2 (S)-2-((6-(2-(Methoxy-d )phenyl)-7-((2-(trimethylsilyl)ethoxy)methyl)-7H-
3
pyrrolo[2,3-d]pyrimidin-4-yl)amino)-2-phenylethan-1-ol (13)
Compound 13 was prepared as described in Section 4.6.2, starting with 6b (90 mg, 0.237
mmol) and 2-(2-(methoxy-d )phenyl)-4,4,5,5-tetramethyl-1,3,2-dioxaborolane (8) (68 mg,
3
0
.284 mmol). The reaction time was 1 hour. Purification by silica-gel column
chromatography (CH Cl /MeOH, 19/1, R = 0.20) gave 91 mg (0.209 mmol, 88 %) of a
2
2
f
1
clear oil; HPLC purity: 99 %, t = 15.1 min; H NMR (400 MHz, CDCl -TMS) δ: 8.35 (s,
R
3
1
6
0
H), 7.44 - 7.37 (m, 6H), 7.35 - 7.30 (m, 1H), 7.05 – 7.01 (m, 1H), 6.99 – 6.97 (m, 1H),
.36 (s, 1H), 5.48 (s, 2H), 5.34 - 5.30 (m, 1H), 4.13 - 3.99 (m, 2H), 3.38 - 3.34 (m, 2H),
1
3
.77 - 0.73 (m, 2H), -0.13 (s, 9H); NH and OH was not seen; C NMR (100 MHz,
CDCl -TMS) δ: 157.4, 155.9, 151.7, 151.2, 140.0, 135.3, 132.6, 130.7, 129.3 (2C), 128.3,
3
1
27.0 (2C), 121.0, 120.7, 111.1, 103.5, 99.6, 71.5, 68.5, 66.4, 64.4, 59.3, 17.9, -1.3 (3C).
4
.8.3 (S)-2-((6-(2-Ethoxyphenyl)-7-((2-(trimethylsilyl)ethoxy)methyl)-7H-
pyrrolo[2,3-d]pyrimidin-4-yl)amino)-2-phenylethan-1-ol (14)
Compound 14 was prepared as described in Section 4.6.2, starting with 6b (100 mg, 0.196
mmol) and (2-ethoxyphenyl)boronic acid (49 mg, 0.294 mmol). Purification by silica-gel
column chromatography (EtOAc/n-pentane, 3/2, R = 0.34) gave 90 mg (0.178 mmol, 90
f
ꢂꢃ
%
) of a colourless oil; HPLC purity: 99 %, t = 27.3 min; [ꢀ] = -102.1 (c 0.50, DMSO);
R
ꢁ
1
H NMR (400 MHz, CDCl -TMS) δ: 8.38 (s, 1H), 7.44 - 7.31 (m, 7H), 7.04 - 6.96 (m,
3
2
7

ACCEPTED MANUSCRIPT
H), 6.34 (s, 1H), 5.53 (s, 2H), 5.47 - 5.46 (m, 1H), 5.36 - 5.33 (m, 1H), 4.12 - 4.03 (m,
H), 3.34 - 3.30 (m, 2H), 1.29 (t, J = 7.0, 3H), 0.75 - 0.70 (m, 2H), -0.15 (s, 9H); C
2
4
1
3
NMR (100 MHz, CDCl -TMS) δ: 156.6, 155.9, 151.7, 151.4, 140.1, 135.8, 132.7, 130.6,
3
1
5
29.3 (2C), 128.3, 127.0 (2C), 121.1, 120.9, 112.2, 103.7, 99.0, 71.6, 68.6, 66.3, 64.1,
9.2, 17.9, 14.9, -1.4 (3C).
4
.8.4 (S)-2-(3-(4-((2-Hydroxy-1-phenylethyl)amino)-7-((2-
trimethylsilyl)ethoxy)methyl)-7H-pyrrolo[2,3-d]pyrimidin-6-
yl)phenyl)acetamide (15)
(
Compound 15 was prepared as described in Section 4.6.2, starting with 6b (75 mg, 0.147
mmol), PdCl (dppf) (10.5 mg, 0.015 mmol, 0.05 eq.) and 2-(3-(4,4,5,5-tetramethyl-1,3,2-
2
dioxaborolane-2-yl)acetamide (59 mg, 0.224 mmol). Purification by silica-gel column
chromatography (EtOAc/MeOH, 94/6, R = 0.16) gave 63 mg (0.121 mmol, 83 %) of a
f
1
yellow foam. H NMR (400 MHz, DMSO-d ) δ: 8.45 (s, 1H), 7.91 - 7.87 (m, 1H), 7.79 –
6
7
(
4
.76 (m, 1H), 7.62 - 7.57 (m, 3H), 7.40 - 7.36 (m, 2H), 7.35 - 7.29 (m, 3H), 7.20 - 7.17
m, 2H), 7.04 - 6.99 (m, 1H), 5.86 (s, br, 2H), 5.56 (s, 2H), 5.46 - 5.40 (m, 1H), 5.02 –
.99 (m, 1H), 3.74 - 3.71 (m, 2H), 3.54 - 3.50 (m, 2H), 0.81 - 0.77 (m, 2H), -0.13 (s, 9H).
4
.8.5 (S)-2-Phenyl-2-((6-(pyridin-4-yl)-7-((2-(trimethylsilyl)ethoxy)methyl)-7H-
pyrrolo[2,3-d]pyrimidin-4-yl)amino)ethan-1-ol (16)
Compound 16 was prepared as described in Section 4.6.2, starting with 6b (150 mg, 0.294
mmol), PdCl (dppf) (11 mg, 15.0 µmol, 5 mol %) and pyridin-4-ylboronic acid (54 mg,
2
0
.441 mmol). The reaction time was 1 hour. Purification by silica-gel column
chromatography (EtOAc, R = 0.17) gave 124 mg (0.271 mmol, 92 %) of a colourless
f
1
film; HPLC purity: 99 %, t = 25.8 min; H NMR (400 MHz, DMSO-d ) δ: 8.68 - 8.65
R
6
(m, 2H), 8.41 (s, 1H), 7.65 - 7.63 (m, 2H), 7.44 - 7.42 (m, 2H), 7.35 - 7.31 (m, 2H), 7.30 -
7
3
.25 (m, 1H), 6.59 (s, 1H), 5.63 - 5.58 (m, 3H), 5.55 - 5.52 (m, 1H), 3.81 - 3.77 (m, 2H),
.56 - 3.53 (m, 2H), 0.90 - 0.86 (m, 2H), -0.11 (s, 9H).
4
.8.6 (S)-2-Phenyl-2-((6-(pyridin-3-yl)-7-((2-(trimethylsilyl)ethoxy)methyl)-7H-
pyrrolo[2,3-d]pyrimidin-4-yl)amino)ethan-1-ol (17)
Compound 17 was prepared as described in Section 4.6.2, starting with 6b (80 mg, 0.157
mmol), PdCl (dppf) (6 mg, 7.85 µmol, 0.05 eq.) and pyridin-3-ylboronic acid (29 mg,
2
0
.235 mmol). The reaction time was 4 hours. Purification by silica-gel column
1
chromatography (EtOAc, R = 0.21) gave 70 mg (0.151 mmol, 96 %) of a pale film. H
f
NMR (400 MHz, DMSO-d ) δ: 8.89 (s, 1H), 8.63 - 8.61 (m, 1H), 8.16 (s, 1H), 8.14 - 8.12
6
(m, 1H), 7.97 - 7.95 (m, 1H), 7.55 - 7.52 (m, 1H), 7.44 - 7.42 (m, 2H), 7.33 - 7.29 (m,
2
H), 7.23 - 7.20 (m, 1H), 7.08 (s, br, 1H), 5.52 (s, 2H), 5.47 - 5.42 (m, 1H), 5.00 - 4.97
1
3
(m, 1H), 3.76 - 3.70 (m, 2H), 3.58 - 3.54 (m, 2H), 0.84 - 0.80 (m, 2H), -0.12 (s, 9H); C
NMR (100 MHz, DMSO-d ) δ: 155.8, 152.3, 152.0, 148.9, 148.8, 141.7, 139.7, 135.5,
6
1
1
33.1, 128.1 (2C), 127.9, 127.0 (2C), 126.7, 123.7, 101.0, 70.2, 65.6, 64.9, 56.2, 17.2, -
.4 (3C).
2
8