Because approximately 80% of small-molecule drugs contain at least one nitrogen atom
in their structures, the construction of C–N bonds is amongst the most desirable bond-formation
reactions.[1] Among amines, primary alkylamines are of great importance in the synthesis of organic
molecules as they are prevalent building blocks in natural-product synthesis and medicinal
chemistry (Figure [1]).[2] Conventional primary alkylamine syntheses require prefunctionalized substrates containing,
for example, nitro, azide, or nitrile groups, and generally produce significant amounts
of chemical wastes and are less atom economical. Alternatively, reductive amination
of carbonyl compounds has been explored thoroughly and many practical procedures have
been developed in recent years.[3] On the other hand, direct amination of aliphatic C–H bonds is increasingly explored
by synthetic chemists as it targets far less reactive, more abundant C(sp3)–H bonds.[4]
[5]
[6] Over the past few decades, there have been many reports of direct aliphatic C–H
amination and C–H amidation reactions mediated by transition-metal nitrenoids or amides
or by hypervalent iodine, or which proceed via radical intermediates.[7–12]
Figure 1 FDA-approved drugs with benzylic amine moieties
There also have been reports of transition-metal-free methods for the construction
of C(sp3)–N bonds through free nitrenes.[13]
[14]
[15] Whereas the direct amination of the C(sp3)–H bonds has emerged as a promising approach that maximizes atom economy, current
state-of-the-art methods typically give N-protected amine products.[7–15] The protected amine products typically require additional manipulations to access
the corresponding primary amines. For instance, several research groups have developed
methods for the synthesis of alkyl tosylamides (R–NHTs) from the corresponding alkanes
by the transfer of an N=SO2C6H4Me moiety through the use of such transition metals as Cu, Rh, or Ag.[16]
[17]
[18]
[19] The deprotection of the tosyl group requires harsh conditions [lithium naphthalenide,
Na/K alloy on silica, Ni(0)acac/i-PrMgCl, Bu3SnH/AIBN, Mg/Me3CoLi, or Mg/MeOH] or the use of expensive reductants such as SmI2 in toxic solvents such as HMPA or DMPU, which ultimately makes these methods less
practical and less atom economical.[20]
[21]
[22]
[23]
[24]
[25]
[26] Recently, the groups of Buchwald, Hu, Kramer, and Warren have each developed Pd-
or Cu-catalyzed reactions that generate an alkylated benzophenone imine (H–N=CPh2) that can be deprotected to furnish the free amine under mild conditions.[27]
[28]
[29]
[30]
[31] The work of the groups of Buchwald[27] and Hu[28] did not involve benzylic substrates despite their prevalence in synthetic organic
and medicinal chemistry.
Kramer developed a facile route to α-substituted, primary benzylimines through a cross-dehydrogenative
coupling method catalyzed by a CuI/1,10-phenanthroline system.[29] The method is simple and the reaction proceeds under low catalyst loadings, but
requires high substrate loadings (10 equiv) and long reaction periods (48 h). More
recently, Kramer’s group reported a dehydrogenative C–N bond formation by the combined
use of a chiral Cu catalyst with a photocatalytic reaction between limiting amounts
of an R–H substrate and NH2Boc (Boc = tert-butyloxycarbonyl); this proceeds with high yields and high enantioselectivities (Figure
[2]A).[30] The Warren group developed a similar method that uses a copper(I) β-diketiminate
catalyst, but they focused mainly on the mechanism of this transformation. The major
drawback of their methods was the formation of the azine Ph2C=N–N=CPh2 as the main byproduct, which significantly increases the nonproductive consumption
of the benzophenone imine (Figure [2]B).[31] More recently, the Comito group reported a versatile hypervalent iodine(III)-mediated
C–H imination under blue LED light and heating (75 °C) conditions to form primary
and secondary amines after a mild deprotection (Figure [2]C).[32] Although practical for both benzylic and nonbenzylic substrates, the biggest drawback
of the method was the use of large excess of the substrates (60–120 equiv), which
makes the methodology less atom economical and less energy efficient (use of both
light and heat), as shown in Figure [2]C.
Figure 2 Benzylic C–H amination for the preparation of primary amines
In light of the above considerations, we sought a C–N coupling protocol that would
combine operational simplicity with the use of an Earth-abundant catalyst to access
primary amines using only one equivalent of the R–H substrate. In 2020, the group
of Landis reported a Cu-catalyzed carbamation of benzylic C–H bonds by using N-fluorobenzenesulfonimide (NFSI) or 1-chloromethyl-4-fluoro-1,4-diazoniabicyclo[2.2.2]octane
bis(tetrafluoroborate) (F-TEDA; Selectfluor) as the oxidant and ethyl carbamate (urethane)
as source of the nitrogen functional group (Figure [2]D).[33] Although the Landais system was found to be highly efficient for the synthesis of
ethyl carbamates (N-Cbz or N-Troc) derivatives, it failed to deliver N-Boc-protected carbamates, the only example reported giving low yields (16–22%). Because
the Boc group is possibly one of the most desirable protecting groups in organic synthesis,
we sought to develop a method that delivers Boc-protected amines through Cu catalysis.
Here, we address the challenge of synthesizing primary amines[34] by developing an Earth-abundant Cu-catalyzed oxidative amination of benzylic C–H
bonds to convert chemical feedstocks into amine pharmacophores. To this end, we developed
a benzylic C(sp3)–H carbamation under Cu(I)/BOX ligand catalysis, and we used TMSNHBoc as the aminating
reagent and NFSI as the terminal oxidant (Figure [2]E). Unlike diaryl imines, which are moisture sensitive, TMSNHBoc is a bench-stable
white solid with a shelf life of more than three months, and which can be easily prepared
by a one-step reaction between TMSCl and tert-butyl carbamate (BocNH2) on a decagram scale with >99% purity (1H NMR) and in nearly quantitative yield. The substrate scope in this report mainly
focuses on cheap and widely available alkylarene feedstocks that could provide facile
access to primary amine building blocks through simple and mild deprotection of the
Boc group. It is noteworthy that although the current method is similar to that reported
Landais,[33] there are some improvements that make our method worthwhile. For instance, in the
aforementioned report, Landis’s method requires hexafluoroisopropanol (HFIP) (in a
1:1 mixture with MeCN), which markedly increases the cost of the synthesis. Furthermore,
their system works at higher temperatures and requires greater catalyst loadings.
Also, their method gave rise to only one example of an N-Boc-protected benzylic amine in a modest yield. Considering the relevance of the
Boc protecting group in organic synthesis in comparison with ethyl carbamates (N-Cbz or N-Troc), this makes the current method a good complement to the previously reported
method.
By considering the recent Cu(I)/NFSI catalyst systems that have been explored in recent
years,[35]
[36]
[37] we started our screening efforts to identify optimal conditions for benzylic C–H
carbamation to form R–NHBoc compounds that could be further deprotected under mild
and simple conditions to give primary amines. In the beginning, we selected BocNH2 as the carbamate source, due to its low price and wide commercial availability. After
optimization, we focused on tert-butyl (trimethylsilyl)carbamate (TMSNHBoc), due to the high affinity of the TMS group
for fluoride, in an attempt to increase the driving force for NHBoc transfer to the
copper(II) center. Ethylbenzene was selected as the benzylic substrate due to its
low price ($116/2.5 L) and ready availability. Summarized optimization data is presented
in Table [1], with additional details provided in the Supporting Information (SI). We continued
by examining commonly used copper sources and ligands. CuI·SMe2 was chosen due to its higher solubility, along with the bulky
tBuBOX ligand (BOX = bis-oxazoline) L1, with NFSI as the terminal oxidant. Screening the source of copper showed that CuCl,
CuBr·SMe2, and CuOAc all afforded the desired product (yield 14–8%); however, CuI·SMe2 gave the highest yield (36%). Cu(I) halide salts in general provided much better
yields, whereas Cu(II) salts, such as Cu(OAc)2 and Cu(OTf)2, gave much lower yields.
Table 1 Effect of Selected Reaction Parameter on the Yield of the Cu-Catalyzed C(sp3)–N Bond Formationa
|
Entry
|
Cu salt (mol %)
|
Ligand (mol%)
|
NFSI (equiv)
|
LGb (equiv)
|
Yieldc (%)
|
1
|
CuCl (10)
|
L1 (12)
|
2
|
TMS (2)
|
11
|
2
|
CuBr·SMe2 (10)
|
L1 (12)
|
2
|
TMS (2)
|
19
|
3
|
CuI·SMe2 (10)
|
L1 (12)
|
2
|
TMS (2)
|
41
|
4
|
CuOAc (10)
|
L1 (12)
|
2
|
TMS (2)
|
29
|
5
|
Cu(OAc)2 (10)
|
L1 (12)
|
2
|
TMS (2)
|
12
|
6
|
Cu(MeCN)4BF4 (10)
|
L1 (12)
|
2
|
TMS (2)
|
trace
|
7
|
Cu(MeCN)4PF6 (10)
|
L1 (12)
|
2
|
TMS (2)
|
trace
|
8
|
CuI·SMe2 (20)
|
L1 (12)
|
2
|
TMS (2)
|
16
|
9
|
CuI·SMe2 (5)
|
L2 (6)
|
2
|
TMS (2)
|
57
|
10
|
CuI·SMe2 (2.5)
|
L1 (6)
|
2
|
TMS (2)
|
26
|
11
|
CuI·SMe2 (5)
|
L1 (6)
|
2
|
TMS (2)
|
67
|
12
|
CuI·SMe2 (5)
|
L1 (6)
|
2.5
|
TMS (2)
|
69
|
13
|
CuI·SMe2 (10)
|
L1 (12)
|
2
|
H (2)
|
21
|
|
a Reaction conditions: ethylbenzene (0.25 mmol, 1.0 equiv), TMSNHBoc (0.5 mmol, 2.0
equiv), NFSI (0.75 mmol, 2.5 equiv), ligand (6 mol%), CuI·SMe2 (5 mol%), MeCN (2.0 mL), sealed tube.
b LG = leaving group.
c Isolated yield.
Next, we screened a wide range of ligands (bipyridines, phenanthrolines, and BOX ligands)
and found that the BOX ligand L1 gave the desired product in high yields (up to 69%). Testing the reaction without
a ligand highlighted the importance of an ancillary ligand, as no product was detected
by GC/MS analysis. Increasing the ligand loading decreased the formation of the desired
product and diminished the yields to as low as 6% (see SI). The formation of the product
with ligands L1 and L2 exclusively (SI) can be rationalized in terms of the steric properties of these ligands,
which hinders any double ligation that could result in deactivation of the catalyst.
A final improvement in yield was achieved by using acetonitrile as the solvent. Other
solvents such as acetone, ethyl acetate, or fluorobenzene did not ensure homogeneity
of the reaction mixture, and lower yields were observed with those solvents; therefore,
MeCN was kept as the reaction solvent.
A variety of oxidants were also screened, including Selectfluor (F-TEDA), Selectfluor
II, NFSI, N-fluoropyridinium (NFPY), N-fluorocollidinium tetrafluoroborate, and iodosylbenzene (PhIO). We found that only
Selectfluor and NFSI were efficient reagents for this transformation, and the latter
was found to be much more efficient in most cases. Initial chiral gas-chromatographic
studies showed no enantioselectivity toward the desired product, although the chiral
bis(oxazoline) ligand L1 was used. Although we did not pursue thorough mechanistic studies, this observation
is in agreement with a radical–polar crossover pathway involving a benzylic carbocation
intermediate (SI; Figure S1B).
Expanding on the results obtained with ethylbenzene, we evaluated the carbamation
of other benzylic R–H substrates (Scheme [1]). Reactions of various para-substituted ethylbenzenes [p-Me, p-t-Bu, p-Ph, p-F, p-OMe, p-CN, p-NO2, p-Cl, and p-C(O)OMe] gave products 2b–j in low to good yields (18–76%). Substrates with electron-deficient rings bearing
a cyano or nitro group gave poor yields (2g and 2h), whereas electron-rich ethylarenes (2b, 2c, and 2f) generally gave high yields, further supporting a radical–polar crossover pathway
involving a benzylic carbocation intermediate. Note that no double functionalization
was observed with this method. Also, methylarenes did not give noticeable yields under
this protocol, like other methodologies involving Cu/NFSI.[37]
[38]
Scheme 1 Scope of alkylarene benzylic C–H carbamation. Yields after column chromatography
are reported.
The common pharmacophore indane underwent carbamation in 73% yield (2k), and 5-methyl-2,3-dihydro-1H-indene and tetralin afforded 2l and 2m, respectively, in good yields of 53 and 69%, . Diphenylmethane underwent carbamation
to give product 2n in 83% yield, and ethylnaphthalenes gave high yields (62–72%) of the corresponding
carbamate products 2o and 2p. Heterocycles also underwent carbamation by this method with acceptable yields. For
instance, 2- and 3-ethylpyridines gave the carbamated products 2s and 2t, respectively, in yields of 16 and 38%, and 2-ethylthiophene gave 2r in a modest yield of 45% (2r). Chromane, another common pharmaceutical core, gave product 2q in a moderate yield of 59%. It is worth noting that 2-ethylfuran did not give the
desired product, although several attempts were made with various combinations of
solvent, temperature, and ligand. We assume that ring opening of the furan might be
a possible explanation, although we did not pursue an analysis of the outcome of that
reaction. Other ethylarenes, such as 3-ethyl-1H-indole and 7-ethyl-1H-indole, and complex substrates, such as dextromethorphan hydrobromide, mestranol,
and dehydroabietylamine, were also examined, but our method failed to deliver the
desired benzylic carbamate products (SI; Figure S1).
As mentioned in the introduction to this report, current C–H amination methods generate
tosyl, nosyl, or Troc-protected (tosyl = p-MeC6H4SO2; nosyl = p-O2NC6H4SO2, or Troc = Cl3CCH2OSO2) amines, deprotection of which is notoriously difficult and requires separate laborious,
time-consuming, and hazardous deprotection steps.[39] In some cases, toxic and carcinogenic solvents such as benzene or chlorinated solvents
are required, making these methods less attractive from a process standpoint. On the
other hand, our method provides Boc-protected amines that can be deprotected in a
one-pot procedure requiring mildly acidic conditions (2 M HCl in ethyl acetate; see
SI) instead of the use of excess amount of harsh and expensive reducing agents such
as SmI2, Li/NH3, Na/naphthalene, Bu3SnH/AIBN, or Cu/Zn. Moreover, our method uses Cu, a base metal, as a catalyst that
is much cheaper than Ag, Rh, or other noble metals, and does not require a large excess
of the R–H substrate.
The relevance of this C–H carbamation protocol for medicinal chemistry was investigated
through the synthesis of the racemic cinacalcet (Sensipar), used for the treatment
of secondary hyperparathyroidism in patients with chronic kidney disease, as presented
in Scheme [2]. The reaction was conducted with 1 mmol of 1-ethylnaphthalene by the developed protocol
to give 2p as a starting material for this synthesis. We then deprotected the carbamation product
(2 M HCl in EtOAc)[40] and used the primary ammonium salt, without any further purification, for the second
step, to give racemic cinacalcet in 63% yield (see SI for details). We then attempted
to synthesize racemic sertraline (Zoloft), an antidepressant of the selective serotonin
reuptake inhibitor class. Carbamation of racemic 1-(3,4-dichlorophenyl)-1,2,3,4-tetrahydronaphthalene,[41] followed by deprotection (2 M HCl in EtOAc) and methylation resulted in overall
66% yield of sertraline as a mixture of diastereomers (see SI for details).
Scheme 2 Syntheses of the hydrochloride salts of (±)-Sensipar and (±)-Zoloft by using our
method
The results described herein demonstrate that an operationally simple Cu-based catalyst
system composed of commercially available components permits site-selective benzylic
C–H carbamation.[42] A combination of good yields, broad functional-group compatibility, and high benzylic
site selectivity makes this method an attractive green alternative to existing protocols
for the incorporation of primary amines into pharmaceutical and agrochemical building
blocks. The relevance to medicinal chemistry was demonstrated by short syntheses of
the racemates of Sensipar and Zoloft hydrochloride salts.