Development of a sequential tetrahydropyran and tertiary butyl deprotection: High-throughput experimentation, mechanistic analysis, and DOE optimization

Article abstract of DOI:10.1021/op8002739

The synthesis of compound 1 by deprotection of the THP, tertbutyl protected amino-pyrazolopyridine (2), is described. The original conditions for this transformation were conducted in a one-pot procedure and necessitated the use of large quantities of eit

Full text of DOI:10.1021/op8002739

Organic Process Research & Development 2009, 13, 471–477
Development of a Sequential Tetrahydropyran and Tertiary Butyl Deprotection:
High-Throughput Experimentation, Mechanistic Analysis, and DOE Optimization
Jeffrey T. Kuethe,* David M. Tellers,* Steven A. Weissman, and Nobuyoshi Yasuda
Department of Process Research, Merck & Co., Inc., Rahway, New Jersey 07065, U.S.A.
Abstract:
improve the overall efficiency of the final process with the
primary goal of increased yield and purity, we reexamined the
deprotection step in greater detail. In this paper, we document
the development of the final optimized process leading to the
formation of 1.
The synthesis of compound 1 by deprotection of the THP, tert-
butyl protected amino-pyrazolopyridine (2), is described. The
original conditions for this transformation were conducted in a
one-pot procedure and necessitated the use of large quantities of
either TsOH or benzenesulfonic acid (5 equiv) and trifluoroacetic
acid (10-25 equiv) and produced 1 in moderate yield (50-65%).
A series of high-throughput screens of Brønstead acids, Lewis
acids, and solvents was rapidly performed with the goal of
identifying improved efficiency and reaction yield. Through these
screens, sulfuric acid was discovered to be a suitable replacement;
however, yields of 1 were still unacceptable. A decoupling of the
two deprotection steps revealed that the THP byproduct resulting
from removal of the THP protecting group was problematic in
the subsequent removal of the tert-butyl group. Consequently, a
two-step deprotection protocol was developed which, in combina-
tion with design of experiment (DOE) optimization, improved the
overall yield to ∼86%.
Results and Discussion
The use of protecting groups is often a necessity in the
preparation of advanced synthetic intermediates, and the
preparation of 1 required both protection of the pyrazole
nitrogen atom (THP protection) and the amino-pyridine nitrogen
(tert-butyl).4 While it was envisioned that removal of the THP
group of 2 would be trivial, removal of the tert-butyl group
from the amino-pyridine nitrogen would require more forcing
conditions.5 Fortunately, a recent report describing both the
preparation and deprotection of tert-butyl-amino-pyridines ap-
peared from these laboratories describing the use of TFA at 70
°C.6 It was initially discovered that deprotection of 2 in the
presence of dry TsOH (5 equiv) and TFA (25 equiv) in MeCN
at 70 °C for 3 h afforded tosylate salt 1a which crystallized
from the crude reaction mixture and was isolated in 65% yield
for the one-pot process (Scheme 1). Despite the modest yield
and the simplicity of the process, key drawbacks with this
Introduction
Reverse transcriptase inhibitors (NNRTI) have been shown
to display a broad spectrum of activity against key HIV-1 RT
mutations. While there are currently three NNRTIs on the
market that have demonstrated clinical efficacy against HIV-1
RT mutations (efavirenz, nevirapine, and delavirdine),1 treat-
ment-related failure due to the emergence of clinical resistance
remains a recurring issue. Therapies that have a broader
spectrum of activity against mutant viruses and have a high
genetic barrier to the selection of new resistant strains have
resulted in the development of second-generation NNRTIs such
as etravirine (TMC-125) which was recently approved by the
FDA.2 As part of a program to develop potent, orally active
NNRTIs that possess an even broader spectrum of mutant
activity, Merck has identified a novel class of second-generation
NNRTIs from which compound 1 was brought forward for
preclinical and clinical development.3 During our work in the
preparation of 1, we found that deprotection of both the
tetrahydropyranyl (THP) group and the tert-butyl group in the
penultimate structure 2 proved difficult with yields for the one-
pot procedure ranging from 50-65% (eq 1). In an effort to
Scheme 1
* Authors for correspondence. E-mail: Jeffrey_Kuethe@merck.com;
David_Tellers@merck.com.
(1) Tarby, C. M. Curr. Top. Med. Chem. 2004, 4, 1045.
(2) http://www.tibotec.com/bgdisplay.jhtml?itemname)HIV_tmc125.
(3) Tucker, T.; Sisko, J.; Tynebor, R.; Williams, T.; Felock, P.; Flynn, J.;
Lai, M.-T.; Liang, Y.; McGaughey, G.; Liu, M.; Miller, M.; Moyer,
G.; Munshi, V.; Perlow-Poehnelt, R.; Prasad, S.; Reid, J.; Sanchez,
R.; Torrent, M.; Vacca, J.; Wan, B.-L.; Yan, Y. J. Med. Chem. 2008,
51, 6503.
10.1021/op8002739 CCC: $40.75  2009 American Chemical Society
Published on Web 03/02/2009
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
471

Table 2. Co-acid, HPLC conversion to 1 (area %); selected
results in acetonitrile from the solvent/acid screen detailed in
Table 1a
process included the need to azeotropically dry the TsOH and
the use of a large excess of TFA. It is important to note that
TsOH or TFA alone did not provide 1, and the use of either
acid rapidly produced the des-THP intermediate 3 and failed
to proceed further to 1. Furthermore, the use of MeCN was
also crucial for success as other solvents investigated resulted
in either only the formation of 3 or led to the formation of other
reaction byproducts. This chemistry was somewhat refined by
the use of anhydrous benzenesulfonic acid (BSA), eliminating
the azeotropic drying step required when TsOH was employed.
In addition, the TFA charge was further optimized to 10 equiv.
For example, treatment of 2 in MeCN with 5 equiv of BSA
and 10 equiv of TFA and heating for 2 h at 70 °C resulted in
the direct isolation of benzenesulfonate salt 1b from the crude
reaction mixture in comparable yield (65%). However, a route
for long-term manufacture required further improvement.
Unfortunately, all attempts to increase the yield of 1 using the
chemistry outlined in Scheme 1 proved unsuccessful. We
therefore used, as detailed below, high-throughput experimenta-
tion (HTE), coupled with DOE optimization, to help identify
the optimal conditions for the conversion of 2 to 1.
First Screen
no
BSA
MgBr2
entry
acid (7 equiv)
co-acid
(5 equiv) (2 equiv)
1
2
3
4
trifluoroacetic acid
oxalic acid
chloroacetic acid
phosphoric acid
2
3
1
2
81
98
80
80
73
97
33
91
Second Screen
LiBr
(2 equiv)
BSA
MgBr2
entry acid (7 equiv)
(2 equiv) (2 equiv)
5
oxalic acid
malonic acid
succinic acid
sulfuric acid
phosphoric acid
HCl
100
12
1
100
75
94
93
83
100
96
6
7
50
69
8
100
61
82
100
75
94
9
10
a Complete results are in the Supporting Information.
or MgBr2 co-acids (entries 2 and 4) produced 1 in comparable
conversion to the original BSA/TFA conditions (Scheme 1).
In the absence of the co-acids, negligible conversion occurred.
We were encouraged by the results from this first screen
since both oxalic and phosphoric acids could serve as economi-
cally acceptable replacements for BSA and TFA. A second
high-throughput screen was performed with the goal of elabo-
rating upon the initial results and expanding the list of acids,
including organic, mineral, and Lewis acids (Table 1, second
screen).
Selected results from the second screen are highlighted in
Table 2 (second screen). In general, oxalic acid (entry 5) in
combination with BSA, MgBr2, and LiBr was found to give
the best conversion compared to the other organic acids we
examined; however, very good reactivity was also observed with
malonic acid with BSA and MgBr2 (entry 6). In general, the
stronger organic acids produced improved conversions as
illustrated by a plot of pKa against HPLC area percent
conversion (Figure 1).9,10 The outlier in this graph is cyanoacetic
High-Throughput Acid/Co-Acid Screen. As our initial
scouting experiments suggested that two acids were needed for
the deprotection, we sought to examine a variety of Brønstead
acids in the presence of either another Brønstead acid or a Lewis
acid (Table 1).7 We opted to perform a high-throughput screen
Table 1. Initial high-throughput screen of solvent, acid, and
co-acida
solvent
acid (7 and 15 equiv)
co-acid (2 equiv)
First Screen
trifluoroacetic acid
dichloroacetic acid
oxalic acid
chloroacetic acid
trichloroacetic acid
phosphoric acid
acetonitrile
2-MeTHF
benzenesulfonic acid
MgBr2
Al(iOPr3)
ZnCl2
BF3
Second Screen
oxalic acid
malonic acid
succinic acid
glutaric acid
adipic acid
acetonitrile
acetic acid
MgBr2
LiBr
PhB(OH)2
benzenesulfonic acid
phthalic acid
phosphoric acid
sulfuric acid
formic acid
citric acid
PhB(OH)2
HCl (conc)
a Conditions: 70 °C, 3 h.
in a 96-well microtitre plate format as this would offer the most
efficient and material-sparing method to study a host of
variables.8 The Brønstead acids employed in our first screen
had pKa values approximately equal to or lower than that of
trifluoroacetic acid, and the Lewis acids chosen were viewed
as being economically viable for scale-up should they be
successful. The initial screen was simply to assess the perfor-
mance of these acids vs conversion to 1.
Figure 1. Plot of pKa (aqueous) vs HPLC area percent
conversion for the reaction of 2 with organic acids and
magnesium bromide in acetonitrile. Cyanoacetic acid, with a
pKa of 2.4 gave 33% conversion.
acid where it is believed that the cyano functionality might
interact with either the THP or tert-butyl byproduct. It is unclear
why magnesium and lithium bromide have such an important
effect on conversion. We propose that they can potentially serve
as a source of HBr upon reaction with the organic acids (Vide
Selected results from the first screen are shown in Table 2.
Oxalic acid and phosphoric acid in the presence of either BSA
472
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

infra), which, based upon our results with HCl, should function
as a viable deprotection reagent. The bromide ion could also
serve as a trap for the THP cation generated after deprotection.
Alternatively, both magnesium and lithium might serve as a
Lewis acid and activate the tert-butyl group toward elimination
through coordination of the magnesium to both the tert-butyl-
amino and the pyridine nitrogen atoms.
dichlorobenzene as an internal standard. The best results from
this screen are shown in Table 4 and are reported in terms of
Table 4. Selected results from the solvent screena
% conversion to 1c (rel yieldb)
H2SO4
(3 equiv)
H2SO4
(7 equiv)
entry
solvent
1
2
3
4
5
6
acetonitrile
87 (2.1)
25 (0.7)
54 (1.3)
45 (1.4)
1 (0)
100 (1.4)
98 (3)
100 (2.6)
99 (3)
3 (0.1)
100 (2)
For the mineral acids, we were pleased to discover that
sulfuric acid gave quantitative conversion of 2 to 1 in all cases.
In fact, sulfuric acid exhibited this reactivity regardless of the
identity of the co-acid. In contrast, both phosphoric acid and
concentrated HCl exhibited a conversion dependence on the
identity of the co-acid (Table 2, entries 9 and 10) with
magnesium and lithium bromide producing better conversion
than BSA. For sulfuric acid, we speculated that a co-acid was
not necessary to influence conversion. We were therefore
pleased to find that treatment of 2 with only sulfuric acid (7
equiv) in MeCN at 70 °C for one hour gave quantitative
conversion to 1 albeit in only 65% HPLC assay yield.11 This
reaction was accompanied by the formation of black polymeric
material similar to that observed in the original conditions. From
our previous experience with the one-pot deprotection sequence
employing BSA/TFA, it was known that the product 1b was
unstable to prolonged heating resulting in diminished yields
(<65%). It should be noted that individual experiments with
the oxalic acid/MgBr2 conditions gave similar yields suggesting
that additional acid screening would not likely be a means for
improving the yield.
acetonitrile/water (90/10)
CPME/acetonitrile (50/50)
DMF/acetonitrile (50/50)
DMF
dimethoxyethane
87 (1.8)
a Complete results can be found in the Supporting Information. b Relative yield,
defined as “area counts of product/area counts 1,2-dichlorobenzene”.
conversion and relative yield against the internal standard. In
general, 7 equiv of sulfuric acid gave greater conversions and
higher relative yields than the reactions performed with 3 equiv.
Acetonitrile was the best solvent in terms of conversion;
however, we were pleased to find that the addition of co-solvents
improVed the reaction yield as well. For example, entries 2
(acetonitrile/water, 90/10) and 4 (acetonitrile/DMF, 50/50) had
an improved relative yield versus acetonitrile (entry 1). Presum-
ably, the addition of either water or DMF aids in the capture
of the tert-butyl cation. When water was utilized, both tert-
butylacetamide and ammonium sulfate were observed in the
crude reaction mixture. Scaling up the best conditions from
entries 2 and 4 (Table 4) to a 100 mg scale gave 1 in 70-75%
yield with concomitant formation of black polymeric material.
While this is a 5-10% improvement in yield over that observed
in acetonitrile, we were not satisfied with this as a final solution
and therefore shifted our attention toward developing a better
mechanistic understanding of the overall deprotection.
At this point in our optimization studies, the use of sulfuric
acid for the deprotection was viewed as most promising in terms
of cost and reaction simplicity, so we opted to focus attention
on improving reaction yield via solvent screening (Table 3). A
Mechanistic Reaction Analysis. We believed that the
individual analysis of the two deprotection steps would allow
us to identify what contribution each step was making to the
overall yield (Scheme 2). It was known from our screening
Table 3. Solvent screen employed to optimize reaction of 2
with sulfuric acid (3 equiv and 7 equiv)a
solvents
Scheme 2
acetonitrile
dichloroethane
acetonitrile/water (90/10)
acetonitrile/water (50/50)
CPME/acetonitrile (50/50)
IPAc/acetonitrile (50/50)
DMSO/acetonitrile (50/50)
ethanol/acetonitrile (50/50)
DMF
dimethoxyethane
2-ethoxyethanol
acetic acid
DMSO/acetonitrile (50/50)
sulfolane
N-methylpyrrolidinone
ethanol/acetonitrile (50/50)
isopropyl acetate
DMF/acetonitrile (50/50)
toluene
benzonitrile
acetonitrile/benzonitrile (90/10)
acetonitrile (nBu4NBr, 2 equiv)
a Conditions: 70 °C, 3 h. DMF ) N,N-dimethylformamide, CPME ) cyclo-
pentylmethyl ether.
results that rapid removal of the THP group in 2 occurred at
room temperature upon the addition of sulfuric acid. When the
THP deprotection of 2 was performed at room temperature,
total of 22 different solvents were examined with both 3 and 7
equiv of sulfuric acid. All reaction mixtures contained 1,2-
(8) (a) Maligres, P. E.; Krska, S. W.; Humphrey, G. R. Org. Lett. 2004,
6, 3147. (b) Shultz, C. S.; Dreher, S. D.; Ikemoto, N.; Williams, J. M.;
Grabowski, E. J. J.; Krska, S. W.; Sun, Y.; Dormer, P. G.; DiMichele,
L. Org. Lett. 2005, 7, 3405. (c) Tellers, D. M.; Bio, M.; Song, Z. J.;
McWilliams, J. C.; Sun, Y. Tetrahedron: Asymmetry 2006, 17, 550.
(d) Tellers, D. M.; McWilliams, J. C.; Humphrey, G.; Journet, M.;
DiMichele, L.; Hinksmon, J.; McKeown, A. E.; Rosner, T.; Sun, Y.;
Tillyer, R. D. J. Am. Chem. Soc. 2006, 128, 17063.
(4) The preparation of 2 will be the focus of a future publication.
(5) De Kimpe, N.; Sulmon, P.; Brunet, P. J. Org. Chem. 1990, 55, 5777.
(6) Yin, J.; Xiang, B.; Huffman, M. A.; Raab, C. E.; Davies, I. W. J.
Org. Chem. 2007, 72, 4554.
(9) Bordwell, F. G. Acc. Chem. Res. 1988, 21, 456.
(7) Experimental details of all the screens can be found in the Experimental
Section.
(10) (Absorbance Product/(Absorbance Product + Absorbance Starting
Material)) × 100; absorbance is measured at 215 nm.
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
473

intermediate 3 was formed in >95% yield with concomitant
formation of black polymeric material. The des-THP intermedi-
ate 3 was isolated by chromatography from this reaction mixture
and subjected to a series of tert-butyl deprotection conditions.
For example, when 3 was allowed to react with 7 equiv of conc
H2SO4 in MeCN at 70 °C, the HPLC assay yield of 1c was
90%! When 1c was resubjected to the same reaction conditions,
no detectable level of decomposition was observed after 2 h at
70 °C, and HPLC assay of 1c was 97%. We had originally
suspected that deprotection of the tert-butyl group in 3 was the
problematic step. These results demonstrate that the tert-butyl
deprotection was inherently a high-yielding reaction.
cleavage as was observed under the original conditions. The
two-step process was optimal since it effectively removed all
THP residues via crystallization of 3a, thereby removing any
chance for deleterious side reactions in the subsequent step.
Design of Experiment (DOE) Optimization and Final
Processing Conditions. Having optimized the THP deprotec-
tion to an acceptable yield, we then shifted our attention to
optimizing the cleavage of the tert-butyl group. The final
deprotection was initially screened with a few different con-
centrations of water, acid, and temperature, and variable yields
of 80-94% were obtained. These scouting experiments sug-
gested that water, sulfuric acid, temperature, and time were
potentially important variables influencing the yield. We opted
to perform a DOE optimization13 on this step with the goal of
quantitatively mapping these reaction parameters in terms of
reaction robustness and identifying the best reaction conditions
for scale-up.
At this point, it was evident that the source of the low yield
could be attributed to the potential byproduct associated with
THP removal (Figure 2). Efforts were then focused on
We routinely employ DOE to compliment our HTE results.
Whereas HTE is a powerful tool for identifying ideal discreet
variables (e.g., solvent, acid, catalyst), DOE provides the optimal
settings for the related continuous variables. The continuous
variables chosen for this study and their settings are shown in
Table 5.
Figure 2. Potential THP byproducts in the absence of trapping
agents.
completely removing these byproducts (i.e., including com-
pounds 4-6) prior to the tert-butyl group deprotection. After
examination of a variety of trapping agents to sequester the THP
byproduct and allow for the direct isolation of 3a from the crude
reaction mixture, the optimal conditions were discovered
(Scheme 3). Treatment of 2 in MeCN with 2.2 equiv of conc
Table 5. Continuous variables (factors) for DOE study in
acetonitrile
factor (units)
range studied
term A. water/vol % in CH3CN
term B. temperature/°C
0–10
55–75
2–8
Scheme 3
term C. reaction time/h
term D. sulfuric acid charge/equiv
3–15
A 24 full factorial screening design with three replicate center
points, to detect curvature in the design space, was selected for
a total of 19 reactions. This design allowed for the interaction
terms to be unaliased with each other. The reactions were
performed in parallel on a 150 mmol scale. Analysis of the
data14 revealed that the most relevant factor influencing the yield
was the interdependence of water content and sulfuric acid
charge (AD term, Table 5). As shown in Figure 3, based on a
linear analysis, at high water content (10 vol %), more sulfuric
acid (15 equiv) is desirable to maximize the yield, but at low
water content, less sulfuric acid gives better assay yields. This
suggests that the molarity of the acid plays a role in the reaction.
But of the two scenarios, the high water charge/high sulfuric
acid charge combination was predicted to provide a better yield
than the low water/low acid combination. This is consistent with
the determination that the second most relevant factor is water
content, in which more water uniformly gives higher yields,
regardless of the sulfuric acid charge. It is notable that significant
curvature (nonlinearity) was observed in the design space, as
evidenced by the fact that the highest yields of all were the
three replicate reactions run at the midpoint of the ranges (Figure
H2SO4 in the presence of 2.2 equiv of 1-octanethiol resulted in
the quantitative formation of des-THP intermediate 3a within
10 min. The use of 1-octane thiol was crucial for obtaining
crystalline 3a and eliminating the formation of black polymeric
byproduct associated with the THP removal.12 During the course
of the reaction, 3a began to crystallize from the reaction mixture.
After dilution with MTBE and then heptane to maximize the
recovery, compound 3a was isolated as bis-sulfate salt in 95%
yield (98-99 LCAP) as a white to off-white crystalline solid.
These conditions completely removed THP residues from the
isolated product. A one-pot procedure for the conversion of 2
to 1c employing octanethiol was also examined; however, the
yield was ∼75% which was approximately 10% lower than a
two-step process. The lower yield was attributed to the fact that
the THP-octanethiol adduct may decompose at the elevated
temperatures required for tert-butyl deprotection. The decom-
position byproduct could then interfere with the tert-butyl
(12) Although the deprotection in the absence of 1-octanethiol gave 3a in
high assay yield, isolation of 3a in acceptable purity was difficult.
(13) For a leading reference on DOE concepts, see: Carlson, R.; Carlson,
J. E. Design and Optimization in Organic Synthesis; Amsterdam:
Elsevier, 2005.
(11) The term “assay yield” refers to a nonisolated solution yield of product
as determined by comparison of product UV absorbance with that of
pure, authentic product standard using HPLC analysis.
(14) DOE analyses were performed using Design Expert V 7.1.3 Stat Ease,
Inc.: 2021 East Hennepin Ave., Suite 480, Minneapolis, MN 55413,
U.S.A.
474
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

observed at the higher temperature setting (75 °C) with a low
acid charge (3 equiv) than at the lower temperature (55 °C)
with a low acid charge (3 equiv), but again curvature was
observed in the design space as the best yields were observed
at the midpoint settings (65 °C/5 equiv water) (Figure 4).
In order to more fully elucidate the nature of the design space
and map the curvature, a second study, employing a response
surface model (RSM) was performed to compliment the
information obtained in the initial DOE design. We used a face-
centered central composite design (CCD) in which only the
water content and sulfuric acid charge were investigated based
on the results of the earlier design. The time (3 h) and
temperature (65 °C) settings were held constant. The ranges
for the two remaining factors were narrowed (water: 2-7 vol
%; acid: 5-9 equiv) to accommodate both the region of interest
from the screening design and practical processing requirements
with respect to impurity generation (related to acid charge) and
impurity rejection (related to water charge); this led to 11
reactions for which both assay yield and area % conversion
were the measured responses. Analysis of the conversion
response revealed the model contained no relevant terms
therefore indicating that the conversion was robust over the
entire design space and uniformly high (>95 A%). The assay
yield response indicated only the water charge was releVant,
whereas the acid charge had little impact (Figure 5). The model
contained only the linear term for the water charge, indicating
a linear response throughout the design space. On the basis of
these data coupled with our observation that at low water
charges (<1 vol %) the reaction yield began to diminish, we
opted to use a water charge of 4 vol %. Despite the higher
potential reaction yield at 2 vol % water (Figure 5), we believe
Figure 3. Graph describing the interaction of factors A and D.
3). This indicates that a linear response is not fully reflective
of the true nature of response surface. This is common in
screening designs which are meant to identify the relevant
factors and their preferred qualitative settings (High or Low)
and can only identify curvature but not describe it precisely.
The reaction time (factor C, Table 5) had no bearing on the
yield, so the lower value of 3 h was selected for future
experiments. Similar to the water content, the ideal temperature
was dependent on the sulfuric acid charge with higher yields
Figure 5. Contour plot: effect of water and acid levels on yield.
that higher water charge of 4% was a more robust processing point
for our larger-scale processing and thus offset any potential yield
gains offered by the lower water charge. To elaborate further, our
data reveal a sharp drop in yield below 2 vol % water charge, and
Figure 4. DOE graph describing the temperature/acid charge
interaction.
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
475

we therefore opted to run at the higher water charge to minimize
potential yield loss due to an incorrect water charge.
compositions consisted of 0.1% H3PO4 and acetonitrile with a
flow rate of 1.5 mL/min.
A successful scale-up of our optimized process was per-
formed on isolated 3a on a 50 g scale using 7 equiv of sulfuric
acid and 4 vol % water at 65 °C for 2 h. During the course of
the reaction, monosulfate salt 1c began to crystallize from the
crude reaction mixture. Assay yield for the tert-butyl removal
was 92%, and after cooling and filtration, the monosulfate salt
1c was isolated in 90% yield. Analysis of the crude reaction
mixture by both NMR and HPLC did not reveal any hydrolysis
of the nitrile of 1c; however, hydrolysis of MeCN was observed.
While not rigorously investigated, the formation of both acetic
acid and tert-butylacetamide were observed. In addition, off-
gassing of isobutylene was not detected, and it was believed
that the water present in the reaction mixture was effectively
trapping the tert-butyl cation. The final optimal conditions for
the synthesis of 1c from 2 are illustrated in Scheme 4.
Preparation of 3-{5-(6-tert-Butyl-amino-1H-pyrazolo[3,4-
b]pyridine-3-ylmethoxy)-2-chloro-phenoxy}-5-chloro-ben-
zonitrile Bis-sulfate (3a). A 1 L, three-neck, round-bottom flask
equipped with a mechanical stirrer and thermocouple was
charged with 2 (35.9 g, 63.4 mmol) in MeCN (100 mL)
followed by addition of octanethiol (20.4 g, 139 mmol) in one
portion. The reaction mixture was cooled to 15 °C, and concd
sulfuric acid (7.4 mL, 140 mmol) was added dropwise over 30
min while maintaining the internal temperature <25 °C. The
resulting homogeneous solution was stirred at room temperature
for 30 min during which time 3a began to crystallize from the
crude reaction mixture. The resulting slurry was stirred at room
temperature for 30 min, and MTBE (130 mL) was added
dropwise over 45 min. The resulting slurry was stirred at room
temperature for an additional 30 min, and heptane (65 mL) was
added dropwise over 45 min; the slurry stirred for 3 h and was
filtered. The wet cake was washed with MTBE (125 mL) and
dried under vacuum/N2 sweep for 8 h to give 40.85 g (95%) of
3a as a white solid: mp 140 °C (DSC); 1H NMR (DMSO-d6,
400 MHz) δ 1.25 (s, 9H), 4.67 (s, 2H), 6.08 (d, 1H, J ) 9.4
Hz), 6.17 (d, 1H, J ) 2.8 Hz), 6.27 (dd, 1H, J ) 8.9 and 2.8
Hz), 6.32 (m, 1H), 6.36 (m, 1H), 6.69 (m, 2H), 7.44 (d, 1H, J
) 9.4 Hz); 13C NMR (DMSO-d6, 100 MHz) δ 26.5, 52.7, 60.7,
105.5, 108.6, 113.0, 113.7, 115.7, 117.4, 120.5, 125.2, 130.6,
135.3, 137.1, 144.9, 149.7, 152.4, 157.4, 157.6. Anal. Calcd
For C24H25Cl2N5O10S2: C, 42.48, H, 3.71; N, 10.32. Found: C,
42.06; H, 3.66; N, 10.21.
Scheme 4. Final process for the preparation of 1c
Preparation of 3-{5-(6-Amino-1H-pyrazolo[3,4-b]pyri-
dine-3-ylmethoxy)-2-chloro-phenoxy}-5-chloro-benzoni-
trile Sulfate (1c). To a 1 L round-bottom flask equipped with a
mechanical stirrer, thermocouple, and reflux condenser were added
3a (54.0 g, 79 mmol) and a 96:4 mixture of MeCN/water (350
mL, v/v). To the solution was added conc sulfuric acid (4.23 mL,
556 mmol), and the reaction mixture was heated to 70 °C for 2 h
during which point the product began to crystallize from the crude
reaction mixture. The slurry was cooled to room temperature and
diluted with 190 mL of water. The slurry was stirred for 3 h and
filtered. The wet cake was washed with 2:1 MeCN/water (150
mL, 2×) and dried under vacuum/N2 sweep for 12 h to give 38.2 g
(92%) of 1c as a white solid: mp 225 °C (DSC); 1H NMR (DMSO-
d6, 400 MHz) δ 5.45 (s, 2H), 6.64 (d, 1H, J ) 9.2 Hz), 7.07 (m,
2H), 7.37 (dd, 1H, J ) 2.3 and 1.2 Hz), 7.47 (dd, 1H, J ) 2.3 and
1.2 Hz), 7.59 (m, 1H), 7.81 (s, 1H), 8.27 (d, 1H, J ) 9.2 Hz); 13
NMR (DMSO-d6, 100 MHz) δ 61.5, 106.7, 109.2, 109.9, 114.5,
114.6, 117.4, 117.7, 119.7, 122.3, 127.2, 131.9, 135.8, 136.9,
137.7, 145.8, 150.6, 156.1, 158.2, 158.4. Anal. Calcd For
C20H15Cl2N5O6S2: C, 45.81; H, 2.88; N, 13.36. Found: C, 45.97;
H, 2.98; N, 13.33.
Summary
We have described an improved procedure for the prepara-
tion of 1 from THP, tert-butyl protected amino-pyrazolepyridine
2. We believe the increase in yield (more than 20% higher than
the original yield), impurity profile, and robustness warrant the
additional isolation step. The multifaceted approach involving
high-throughput reaction screening, mechanistic analysis, and
DOE optimization were essential to the development of this
new process. The high-throughput screen allowed us to rapidly
identify a new acid (H2SO4) for the deprotection step using a
minimal amount of material: 1.2 g of 2, 236 reactions, 3 days
total for reaction setup and analysis. Greater understanding of
the individual deprotection steps with sulfuric acid allowed us
to make the observation that residues from the THP deprotection
were solely responsible for the low yield in the one-pot
procedure. Finally, DOE allowed us to define, better understand,
and optimize the factors that were important to reaction yield
and robustness.
Acknowledgment
Experimental Section
We thank our Merck & Co., Inc. colleagues: Drs. Peter
Maligres, Jingjun Yin, and Theresa Williams for helpful
discussions and Dr. Philip Pye and Mr. Danny Mancheno for
Reaction mixtures and products were analyzed by reverse
phase HPLC on a Hewlett-Packard 1100 instrument using a
4.6 mm × 50 mm Zorbax Eclipse Plus C18 column. Solvent
476
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

their contributions in the discovery of the initial conditions for
the conversion of 2 to 1 employing TsOH/TFA.
results.This material is available free of charge via the Internet
at http://pubs.acs.org.
Supporting Information Available
A complete listing of DOE data, HTE screening protocol,
original procedure for preparation of 1b from 2, and screening
Received for review October 23, 2008.
OP8002739
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
477