Scalable synthesis of the desoxy-biphenomycin B core

Article abstract of DOI:10.1021/op200207h

We describe the evolution of a kilogram-scale synthesis of the protected cyclic tripeptide desoxy-biphenomycin B, based on an early discovery route. The retrosynthetic concept included a macrolactamization strategy to build the core ring system of biphenomycin B in combination with a double catalytic asymmetric hydrogenation protocol for the construction of the ansa-tripeptide precursor. Eventually, the kilogram process comprised a 16-step sequence with an overall yield for the longest linear sequence of 19.5%.

Full text of DOI:10.1021/op200207h

ARTICLE
pubs.acs.org/OPRD
Scalable Synthesis of the Desoxy-biphenomycin B Core
Mathias Berwe,^ Winfried J€ontgen,^ Jochen Kr€uger,*,^ Yolanda Cancho-Grande,‡ Thomas Lampe,‡
Martin Michels,‡ Holger Paulsen,‡ Siegfried Raddatz,‡ and Stefan Weigand§
Bayer Pharma AG, ^Chemical Development, Process Research, and ‡Medicinal Chemistry, 42096 Wuppertal Germany
ABSTRACT: We describe the evolution of a kilogram-scale synthesis of the protected cyclic tripeptide desoxy-biphenomycin B,
based on an early discovery route. The retrosynthetic concept included a macrolactamization strategy to build the core ring system of
biphenomycin B in combination with a double catalytic asymmetric hydrogenation protocol for the construction of the ansa-
tripeptide precursor. Eventually, the kilogram process comprised a 16-step sequence with an overall yield for the longest linear
sequence of 19.5%.
’ INTRODUCTION
6 is introduced on a highly functionalized intermediate, render-
ing the use of costly pinacolato diboron mandatory; and (3)
several intermediates had to be purified by chromatography.
As the project advanced, the demand for key intermediate 2
increased readily reaching the kilogram scale. Accordingly, we
revised the discovery route and embarked on the development
of a scaleable route for 2. Our endeavours towards this multi-
kilogram-scale synthesis are outlined in the following chapters.
The biphenomycin B antibiotic represents a complex lead
structure from natural sources.1 In the past, several academic
groups have elaborated viable routes to synthesize the biphenyl-
bridged cyclopeptide based on two key steps: transition metal
catalyzed biphenyl coupling and macrolactamization.2
For our in-house research and development program, the
accessibility of the macrolactam core of biphenomycin B in multi-
gram up to kilogram quantities was a prerequisite. In particular,
we were interested in analogues derived from desoxy-bipheno-
mycin B, which lack the hydroxyl group substitution in the hy-
droxy ornithine side chain. Our experience in route scouting and
scale-up for manufacturing of target molecule 2 are detailed in
this manuscript.
The Bayer routes towards protected desoxy-biphenomycin 2
(Scheme 1) were based on the pioneer work of Schmidt el al. who
reported the first total synthesis of biphenomycin B in 22 linear
steps with an overall yield of 9%.3 Our discovery synthesis intro-
duced an increased level of convergence by using the higher
functionalized dehydroamino acid building blocks 5 and 6 as sub-
strates for a Suzuki coupling. Both intermediates could be accessed
commencing from phosphonate 3 and substituted benzaldehyde
4a, respectively. In the key step, the catalytic asymmetric double
hydrogenation of Suzuki product 7 furnished intermediate 8 in
excellent yield (98%) and high enantioselectivity (>99.5% ee),
which demonstrated the wide scope and robustness of this
methodology. The final steps towards target molecule 2 con-
sisted of the introduction of the protected (S)-ornithine and
the subsequent macrolactamization of the ornithine Nα employ-
ing Schmidt’s pentafluorophenyl ester protocol. In total, our dis-
covery route towards 2 comprised 18 steps (15 steps starting
from 3 and 4a) with good chemical yields (21%, for the longest
linear sequence of 11 steps) from commercially available raw
materials.4
’ RESULTS AND DISCUSSION
Retrosynthesis of the Process Route. On the basis of our
experience with the discovery route, we planned to utilize the
double asymmetric hydrogenation as a key step, however, on
the higher functionalized substrate 9 later in the sequence
(cf. Scheme 2). This would offer an avenue to a more convergent
synthesis as building blocks 10ꢀ12 should each be synthesized
by a comparable level of complexity. Furthermore, the number of
linear steps from intermediate 9 to the target molecule would be
reduced substantially.
The macrocyclization should remain between the Nα-or-
nithine and the corresponding carboxylic acid as established by
Schmidt, who reported a good yield (85%) for this step. In addi-
tion, we learned from the research phase that this transformation
can be run at fairly high concentrations (4 mmol/L) to reach
complete conversion within minutes. Therefore, we anticipated
obtaining good volume productivity for this crucial step at scale.
Furthermore, we plannedtoabolish thecostly2-(trimethylsilyl)-
ethyl (TMSE)-protecting group, which would have the benefit
of decreasing the overall number of steps. The differentiation
between the two carboxy groups in 9 would then be achieved by
simply leaving one carboxylic acid unprotected. Additionally, the
need for elaborate chromatographic purification of several inter-
mediates in the discovery route had been attributed to the pre-
sence of the TMSE-ester. Thus, we speculated that workup pro-
cedures might be simplified if the TMSE-group could be replaced.
The synthesis of dehydroamino acid 10 should be straightfor-
ward by a WadsworthꢀEmmons protocol using aldehyde 4b and
The route depicted in Scheme 1 has been applied to deliver
multigram quantities of the protected desoxy-biphenomycin B
(2). Nevertheless, with respect to scale-up towards an economic-
ally feasible production process, we encountered some limita-
tions of this pathway, which were (1) the overall synthesis is
still a lengthy process, and the majority of transformations are
unproductive protecting group manipulations; (2) the boronate
Received: August 4, 2011
Published: September 13, 2011
r
2011 American Chemical Society
1348
dx.doi.org/10.1021/op200207h Org. Process Res. Dev. 2011, 15, 1348–1357
|

Organic Process Research & Development
ARTICLE
Scheme 1. Bayer’s discovery route to protected desoxy-biphenomycin B (2)
Scheme 2. Retrosynthesis for the process route to protected desoxy-biphenomycin B (2)
phosphonate 3 (Scheme 1), while the boronate 11 has been
reported in the literature and offers the opportunity to circum-
vent costly bispinacolato diboron at scale.5
Access to the phosphoryl dipeptide 12 was anticipated by a
standard peptide coupling employing the deprotected phospho-
nate 3 (Scheme 1), which is additionally required as reagent for
the synthesis of 10.
Pilot-Plant Synthesis of Phosphonate 3. The phosphonate
3 is a very useful reagent for the synthesis of dehydroamino acids,
and we embarked on a multikilogram synthesis based on the pro-
cedures reported by Schmidt et al. (cf. Scheme 3).6 The condensation
reaction between glyoxylic acid monohydrate 13 and benzyl car-
bamate 14 is reported to proceed in diethyl ether in good yields.7
For safety reasons, we replaced diethyl ether by toluene and ob-
served no detrimental effects on the reaction rate or the product
quality. On a 26-L scale, we obtained 5 kg of the desired mono
condensation product 15 in 96% yield and 97% purity (HPLC
area %). To our dismay, we could not transfer this process to
pilot-plant scale without further adaptations. When we ran the
lab-protocol on a 150-L scale, we observed the formation of 7%
of the bis-adduct 17 (cf. Figure 1), which had not been observed
in significant amounts in lab or kilogram/lab scale before.
1349
dx.doi.org/10.1021/op200207h |Org. Process Res. Dev. 2011, 15, 1348–1357

Organic Process Research & Development
ARTICLE
We scrutinized a range of parameters such as reaction time,
stoichiometry, time for dosing, mixing, quality of raw materials
etc., however, we could not identify the origin of the formation of
17 at scale. Therefore, we focused on a purification strategy to
remove 17 by a subsequent operation. Upon workup, the bis-
adduct 17 cocrystallized with the desired product 15 under a
variety of conditions since 17 proved to be poorly soluble in or-
ganic solvents. Accordingly, mixtures of 15 and 17 were resus-
pended in methanol, unsoluble 17 was removed by filtration, and
the filtrate containing 15 was directly used in the next step.
The formation of benzyloxy carbonyl (Cbz)-protected me-
thoxy glycine ester 168 was accomplished by reaction of 15 with
methanol in the presence of trimethyl orthoformate (TMOF)
and catalytic amounts of HCl to yield 16 in 98% purity with an
average yield of 67%.
The formation of 3, as outlined in Scheme 3, is a two step
process comprising of a chlorination and a phosphorylation
reaction.6a,b,7c,9 The chlorination was accomplished employing
PCl3 in toluene, and subsequently the phosphonate was intro-
duced upon reaction with trimethyl phosphite. In lab scale we
observed significant variations in reaction time for the chlor-
ination step. In most cases the chlorination was complete
within hours, while conversion of some batches of 16 were
incomplete even after 24 h at reflux. We could trace down this
effect to the quality of the starting material 16. Batches of 16,
which had been synthesized by an old procedure using sulphuric
acid as catalyst were converted to 3 at a significantly faster rate.
Thus, we added a catalytic amount of sulphuric acid (0.3%) to
the chlorination cocktail, and, to our delight, we observed high
reproducibility for the chlorination step independent of the
batch history of 16. In all cases chlorination was complete
within 16 h and subsequent phosphorylation proceeded smoothly
within 2.5 h.
In total, four batches of the phosphorylglycin ester 3 were syn-
thesized in our pilot plant to give 153 kg of the desired reagent
with an average yield of 79% and an average purity of 98% (assay).
Synthesis of Dehydroamino Acid (10). The synthesis of
dehydroamino acid 10 was achieved by a two step protocol start-
ing from 5-bromo salicylaldehyde (4a). Benzyl protection was
achieved by the use of benzyl chloride in the presence of sodium
carbonate and catalytic amounts of sodium iodide.3a,10 Accord-
ingly, the protected intermediate 4b was obtained in an average
yield of 98% on a 26-L scale (cf. Scheme 4).
The WadsworthꢀEmmons reaction of phosphonate 3 with
aldehydes has been pioneered by Schmidt et al.6a,b,11 In our
hands, the highest diastereoselectivity was obtained in THF
at ꢀ20 ꢀC with tetramethylguanidine as a base. We ran seven
batches on a 20-L scale to give a total of 14 kg of 10 with an
excellent average yield of 97% and a Z/E ratio of 97:3. The
diastereoselectivity could be improved to 99% by recrystalliza-
tion from a 2.5:1 mixture of methyl tert-butyl ether and dioxane.
However, we abandoned this purification step and used the crude
E/Z mixture since the minor E-isomer could completely be re-
moved during the workup in the following Suzuki coupling.
Pilot-Plant Synthesis of Boronate 11. The benzyl protected
bromo salicylaldehyde 4b was converted to cyclic acetal 18 upon
reaction of 4b with 1,3-propanediol in cyclohexane in the pre-
sence of catalytic amounts of sulphuric acid and azeotropic re-
moval of water.3a,b,5b On a 25-L scale the process proved highly
reproducible (eight batches) to deliver acetal 18 in an average
yield of 89%.
The introduction of the boronic acid was accomplished by
Schmidt et al. via a Grignard protocol.5b,12 We considered a low
temperature (ꢀ75 ꢀC) bromoꢀlithium exchange pathway more
attractive for scale-up since the reaction was dose-controlled and
could be monitored more precisely.13 The price to pay was a
higher dilution (0.45 mol/L) in order to keep the reaction mix-
ture agitated at ꢀ75 ꢀC.
Scheme 3. Synthesis of phosphonate 3
The reaction, as outlined in Scheme 5, was run on a 250-L scale
to yield 7.0ꢀ7.6 kg of the boronic acid (99% purity) which
corresponds to an average yield of 69%. The moderate yield is
partly explained by an incomplete crystallization process from a
toluene methylcyclohexane mixture leaving a significant amount
of 11 in the mother liquor. The second source for a product loss
is the formation of deborylated impurity 19, which accumulates
in the mother liquor (cf. Figure 2). This impurity is found at a
level of approximately 6 area % after aqueous workup, and we
assume that it is formed by hydrolysis of 11 after the aqueous
quench. The sensitivity of 11 towards a protodeborylation mech-
anism was also confirmed by the short half-life of 11 under the
aqueous conditions of the subsequent Suzuki coupling.
Synthesis of Phosphonate 12. The hydrogenolytic removal
of the Cbz-group in 3 and the subsequent peptide coupling of the
resulting amine 20 has been reported before.6b,14 In our hands,
deprotected intermediate 20 proved very unstable and readily
Figure 1. Side product arising from double condensation of 13 and 14.
Scheme 4. Synthesis of dehydroamino acid 10
1350
dx.doi.org/10.1021/op200207h |Org. Process Res. Dev. 2011, 15, 1348–1357

Organic Process Research & Development
ARTICLE
polymerized upon storage.15 In contrast, a dilute solution of 20 in
THF could be stored at 4 ꢀC for several days without significant
decomposition. Thus, we abandoned further experiments to-
wards product isolation and carried on with the crude product
solution after filtration.
The THF solution of 20 was directly coupled with commer-
cially available Boc-ornithine(Cbz)ꢀOH employing 2-chloro-
4,6-dimethoxy-1,3,5-triazine (CDMT) and N-methylmorpholine
(NMM). The latter coupling cocktail was chosen on the basis
of costs over other alternatives (e.g., EDC) which worked equally
well. After aqueous workup and removal of the solvents, phos-
phonopeptide 12 was obtained as an oil in 90% yield with a purity
of 76 area % (see Scheme 6). It was used in the following
WadsworthꢀEmmons protocol without further purification.
Synthesis of Bis-dehydroamino Acid 9. The synthesis of the
biphenyl derivative 21 was achieved via a Suzuki coupling be-
tween bromoaryl 10 and boronic acid 11.5a,12,16 The solvent of
choice was a 97:3 mixture of toluene and water from which the
desired product conveniently crystallized directly from the reac-
tion mixture.
We observed that under the aqueous conditions at elevated
temperatures (approximately 90 ꢀC) the half-life of the boronic
acid was dependent on the nature of the base. Accordingly, we
found potassium carbonate to be the best compromise between
reaction rate and the deborylation pathway leading to the forma-
tion of 19 (cf. Figure 2). Thus, complete conversion was achieved
with as little as 1.2 equiv of the boronic acid 11 if an aqueous solu-
tion of potassium carbonate was dosed to the reaction mixture
containing 10, 11, 1% palladium acetate and 4% tri-o-tolylphos-
phine at reflux temperature within 45 min. The workup included
a hot filtration through a bed of charcoal, crystallization upon
cooling, and isolation by filtration. Eventually, the product was
isolated as pure Z-isomer with an average chemical purity of 97%
(assay) and an average yield of 94% (Scheme 7).
The saponification of 21 using lithium hydroxide in dioxane/
water proceededsmoothly with full conversionafter 17 h at20 ꢀC.
However, we faced serious problems during product isolation,
since the free carboxylic acid of 22 precipitated in very fine crystals
yielding a non-filterable suspension. Thus, we decided to isolate
the Li-salt, which precipitated from the reaction mixture by the
addition of water. More importantly, due to its salt character
22 grew suitable crystals for filtration. Finally, saponification of 21
was conducted on a 3.3-kg scale with convenient filtration times
yielding 22 in an average yield of 99% (Scheme 7).
Scheme 5. Synthesis of boronic acid 11
The bis-dehydroamino acid building block 9 is a key inter-
mediate in the synthesis of target molecule 2. The tetramethyl
guanidine-mediated WadsworthꢀEmmons protocol towards 9
proceeded within hours in methanol or acetonitrile, while in
THF the conversion was not complete within one day at 20 ꢀC.
Interestingly, the side-product profile was also solvent depen-
dent. In methanol and acetonitrile only traces of the diester by-
product 24 were formed, while E-isomer 23 occurred in levels of
up to 6% (cf. Figure 3). In THF the Z/E selectivity was increased
to >97:3; however, up to 8% of the diester 24 was detected in
the reaction mixture. This implies a mechanism involving methyl
phosphonate species as the only reasonable source for methyl
ester formation. Accordingly, we speculated that an aqueous
in situ quench of alkylating phosphonate intermediates might
reduce diester formation. Indeed, addition of 1.25% water to the
THF reaction mixture reduced the formation of 24 from 8%
to 2ꢀ3%. Moreover, the addition of water led to a drastic
Figure 2. Protonated side product during the formation of 11.
Scheme 6. Synthesis of phosponopeptide 12
Scheme 7. Synthesis of bis-dehydroamino acid derivative 9
1351
dx.doi.org/10.1021/op200207h |Org. Process Res. Dev. 2011, 15, 1348–1357

Organic Process Research & Development
ARTICLE
improvement of the solubility of the reaction components 22
and 9. This was beneficial for the volume productivity as well as
for the reaction rate, leading to a complete conversion within
several hours. Notably, the WadsworthꢀEmmons protocol was
compatible with the presence of the carboxylic acid in 22. This
observation confirms our initial assumption that the TMSE-protect-
inggroup used in the discovery route can easily be replaced by the
free carboxy group. This simplification of the protecting group
strategy adds considerable value to the overall synthesis.
Later in the project we learned that the presence of inorganic
impurities in 9 had a detrimental effect on the efficiency and
reproducibility of the enantioselective hydrogenation protocol.
Thus, we replaced THF by 2-methyl-THF in order to conduct
an aqueous workup with good phase separations. Eventually, the
bis-dehydroamino acid 9 was crystallized from acetonitrile to
deliver 9 with good chemical (>98%, assay) and optical purity
(Z/E >99:1) in 68% yield based on 22 (Scheme 8).
Synthesis of target molecule 2. The hydrogenation of 9
using the achiral Wilkinson’s catalyst delivered a random mixture
of diastereomers indicating that the stereogenic center in the
ornithine side chain did not control a side selective hydrogena-
tion. Therefore, we screened a variety of commercially available
rhodium catalysts with bidentate chiral phosphine ligands. Among
the catalysts tested the phospholane type ligands17 (cf. Figure 4)
were identified as the ligands of choice with respect to stereo-
selectivity and conversion.18
presence of inorganic salts might contribute to this phenomenon.
Indeed, the critical batches of 9 were completely converted to 25
with quantitative yieldand highselectivity (>99%de) with 1 mol %
of (S,S)-Et-DuPHOS-Rh using a 1 mol % silver triflate as addi-
tive. This observation strongly suggested that the lack of repro-
ducibility could be traced back to halide impurities. Accordingly,
as described above, we introduced an aqueous workup during the
isolation of 9 and replaced hydrochloric acid by sulphuric acid.
With this measure in place, we obtained starting material 9 in
quality consistently free of halides, which smoothly underwent
hydrogenation with a high level of reproducibility.19
The conversion to 25 was a clean reaction. Thus, we decided
to use the filtered reaction mixture for the next step without
further purification.
The introduction of the pentafluorophenyl (PFP) ester was
accomplished via an activation of the carboxylic acid by a mixed
anhydride.20 The PFP ester 26 was isolated by crystallization in
98% yield for two steps.
The final macrolactamization was a two-step process including
removal of the ornithine tert-butoxy)carbonyl (BOC)-group
followed by liberation of the resulting HCl salt with concomitant
ring closure. Cleavage of the BOC group was accomplished by
treatment of 26 with HCl in dioxane (20 h at 23 ꢀC) and the
resulting solution of the hydrochloride salt was added within
70 min to a solution of an excess triethylamine in THF. The pro-
tected desoxy-biphenomycin 2 was poorly soluble in organic
solvents—a fact that could conveniently be exploited for product
isolation. Thus, 2 was collected by filtration, and after extensive
washing a quality of 97 area % was achieved, which was sufficient
for further use.
In the course of the studies, we observed that reproducibility of
the hydrogenation was dependent on the batch history of starting
material 9. In particular, old preparations of 9, which had been
worked-up with hydrochloric acid were among the most proble-
matic batches since the latter did not show any conversion with as
much as 10 mol % of catalyst at 80 bar of hydrogen pressure.
Since we carefully excluded the presence of oxygen and we used
batches of 9 with purities >98%, only, we speculated that the
Figure 4. Most efficient phospholane-type ligands for the asymmetric
Figure 3. Side products during the formation of 9.
double hydrogenation.
Scheme 8. Synthesis of target molecule 2a
a NMM = N-methylmorpholine; PFP = pentafluorophenol.
1352
dx.doi.org/10.1021/op200207h |Org. Process Res. Dev. 2011, 15, 1348–1357

Organic Process Research & Development
ARTICLE
To our delight, the sequence could be run with a fairly high
concentration of up to 57 mmol/L of 26. Along these lines,
we assume that both functional groups are preorientated for
the ring closure by the rigid biphenyl backbone and the three-
dimensional conformation controlled by the (S,S,S) stereochem-
istry. This hypothesis is supported by the observation that the
(R,S,R) diastereomer of 26 derived from hydrogenation with (R,R)-
Et-DuPHOS-Rh cyclized with a much slower rate and led to a
sluggish reaction at a comparable concentration.
was removed by distillation with concomitant addition of 50 kg
of fresh toluene. It was stirred at 40 ꢀC for 2 h and then cooled
to 20 ꢀC. The mixture was maintained at this temperature for one
hour and then it was filtered. It was washed with 35 kg of toluene
and dried in vacuo to yield 30.0 kg (81%) of the desired product.
Analytical characterization revealed that the material was obtained
in a purity of 91 area % containing 7% of side product 17. The
crude product was taken to the next step without purification.
In total, seven batches of 15 were produced in the pilot plant to
yield 220.3 kg of hydroxyglycine 15 with an average yield of 84%.
On a 3.3-kg benzylcarbamate scale the yield increased to 96%,
and the purity rose to 97 area %. For analytical characterization, a
laboratory-scale batch with 99.9 area % was analyzed. Melting
point: 155 ꢀC dec; HPLC (method C): Rt = 10.0 min, 99.9 area %;
1H NMR (500 MHz, D6-DMSO): δ = 5.06 (s, 2H), 5.23 (d,
J = 8.8 Hz, 1H), 6.28 (s, broad, 1H), 7.31ꢀ7.39 (m, 5H), 8.14 (d,
J = 8.8 Hz, 1H), 12.81 (s, broad, 1H); 13C NMR (125 MHz, D6-
DMSO): δ = 65.4, 73.1, 127.7, 128.3 (3C), 136.7, 155.4, 171.0;
HRMS (ESI+): Calc. mass (C10H11NO5) = 151.0633; found =
151.0635; IR (KBr): ν (cmꢀ1) = 532, 574, 607, 647, 695, 716,
738, 788, 893, 916, 989, 1004, 1020, 1093, 1249, 1272, 1334, 1378,
1407, 1453, 1498, 1531, 1700, 1732, 2535, 2719, 2943, 3039, 3335.
Benzyloxycarbonyl Amino Methoxyacetic Acid Methyl
Ester (16). A suspension of 31.3 kg (139.0 mol) of 15 in 91 kg
of methanol was heated to 55 ꢀC and then stirred at 10 ꢀC for 10 h.
The side product 17 was filtered off, and it was washed with 37 kg
of methanol. To the filtrate were added 29.6 kg (279.0 mol) of
trimethyl orthoformate and 5 kg of a 1.25 M solution of HCl in
methanol, and it was stirred at 56 ꢀC for 30 min. Subsequently, at
a temperature of 58ꢀ67ꢀ, the solvent was partially (115 kg) dis-
tilled off. It was cooled to 35 ꢀC, 37.0 kg of diisopropyl ether was
added, and after the addition of seed crystals it was cooled to 2 ꢀC
within 4 h. The product was isolated by filtration, and it was
washed twice with 22.0 kg of diisopropyl ether. Finally, it was
dried in vacuo at 45 ꢀC to yield 22.9 kg of the Cbz-methoxy-
glycinester 16 in 65% yield.
One campaign comprising seven batches amounted to a total
of 163.3 kg of 16. This corresponds to an average yield of 67% per
batch. Melting point: 77 ꢀC; HPLC (method C): Rt = 16.0 min,
98 area %; 1H NMR (500 MHz, D6-DMSO): δ = 3.26 (s, 3H),
3.67 (s, 3H), 5.08 (s, 2H), 5.16 (d, J = 8.8 Hz, 1H), 7.33ꢀ7.40
(m, 5H), 8.49 (d, J = 8.5 Hz, 1H); 13C NMR (125 MHz,
D6-DMSO): δ = 52.1, 54.6, 65.8, 80.3, 127.8, 127.9,128.3 (3C),
136.5, 155.8, 167.8; HRMS (ESI+): Calc. mass (C12H15NO5) =
221.0688; found = 221.0692; IR (KBr): ν (cmꢀ1) = 521, 563,
581, 615, 663, 698, 736, 762, 790, 847, 899, 937, 981, 1031, 1103,
1196, 1226, 1262, 1362, 1440, 1455, 1533, 1688, 1753, 2844,
2946, 2973, 3035, 3313.
Benzyloxycarbonyl Amino(dimethoxyphosphoryl)acetic
Acid Methyl Ester (3). To a solution of 0.05 kg (0.51 mol) of
sulfuric acid and 37.5 kg (148.1 mol) of Cbz-methoxyglycinester
16 in 210 kg of toluene was added 24.0 kg (174.8 mol) of
phosphorus trichloride at 75 ꢀC within 1 h. It was stirred for 16 h
at 75 ꢀC. Subsequently, the excess phosphorus trichloride and
part of the solvent was distilled off at 70 ꢀC and 250 mbar. To the
reaction mixture 20.7 kg of (166.8 mol) trimethyl phosphite was
added within 1 h at 75 ꢀC, and it was stirred at 90 ꢀC for 1.5 h.
Then it was cooled to 20 ꢀC, and the pH was adjusted to pH 6.6,
employing an ammonia solution (3% in water). The phases were
separated, and the aqueous layer was extracted with 25 kg of
toluene. The combined organic layers were washed with 30 kg of
water, and the solvent was partly distilled off in vacuo (200 mbar)
On a 20-L scale, approximately 1 mol of 26 was converted to
target molecule 2 in 73% yield. This translates into some 700 g of
the desired product per batch.
’ CONCLUSION
In conclusion, we present an efficient kilogram synthesis of
protected desoxy-biphenomycin B (2). The salient features of
the process route are (1) reduction of the overall number of steps
from the discovery route to the process route from 18 to 16 with
14 isolated intermediates, only; (2) all tedious chromatographic
separations were successfully replaced by crystallization or
extraction protocols; (3) the overall yield for the longest linear
sequence (13 to 2) was 19.5%; and (4) the convergence of the
synthesis was increased which led to an improved economy in
both resources and raw materials.
The new development route is the basis for an efficient manu-
facturing process, and we could demonstrate the superiority of
this route by obtaining a total of 1.4 kg of protected desoxy-
biphenomycin B as a result of the first production campaign.
’ EXPERIMENTAL SECTION
General Remarks. Melting points (uncorrected) were deter-
mined in capillaries using an apparatus manufactured by B€uchi.
Routine NMR spectra were recorded using a Bruker AC500
instrument. All spectra are calibrated against tetramethylsilane as
an internal standard (δ = 0). Coupling constants (J) are given in
Hz. HRMS were acquired on a Micromass LCT, and IR-spectra
were recorded on a Bruker IFS66 V.
For HPLC analysis several methods were elaborated using a
Hewlett-Packard 1100. All HPLC analyses were conducted using
a Phenomenex Prodigy ODS3 (symmetry: 150 mm ꢁ 3 mm,
3 μm) column. HPLC method A: column: Phenomenex Prodigy
ODS3, column symmetry: 150 mm ꢁ 3 mm, 3 μm; T: 40 ꢀC;
eluent: A = acetonitrile, B = aqueous phosphate buffer pH 2.4,
gradient: 0ꢀ15 min 50% A, 15ꢀ25 min from 50% A to 90% A,
25ꢀ45 min 90% A; flow rate: 0.5 mL/min; detection at 220 nm.
HPLC method B: T: 40 ꢀC eluent: A = aqueous phosphate buffer
pH 2.4, B = acetonitrile, gradient: 0ꢀ45 min 90% A to 20%, flow
rate: 0.5 mL/min; detection at 220 nm. HPLC method C: T:
25 ꢀC eluent: A = aqueous phosphate buffer pH 2.4, B = acetonitrile,
gradient: 0ꢀ25 min 90% A to 30% A; 25ꢀ45 min 20% A, flow
rate: 0.5 mL/min; detection at 220 nm. HPLC method D: T:
25 ꢀC eluent: A = aqueous phosphate buffer pH 2.4, B = acetonitrile,
gradient: 0ꢀ20 min: 70% A, to 20% A; 20ꢀ40 min 20% A; flow
rate: 0.5 mL/min; detection at 210 nm. HPLC-method E: analogous
to method C, detection at 210 nm.
Benzyloxycarbonyl Aminohydroxyacetic Acid (15). A solu-
tion of 25.0 kg (165.3 mol) of benzylcarbamate, 16.8 kg (182.5 mol)
of dihydroxyacetic acid monohydrate in 170 kg of toluene was
stirred at 40 ꢀC for 90 min. Subsequently, 40 kg of toluene was
distilled off (40 ꢀC, 100 mbar). Additionally, 50 kg of toluene
1353
dx.doi.org/10.1021/op200207h |Org. Process Res. Dev. 2011, 15, 1348–1357

Organic Process Research & Development
ARTICLE
at 60 ꢀC. To the residue were added 53 kg of diisopropyl ether
and seed crystals. It was stirred at 5 ꢀC for 1 h. Subsequently, it
was filtered and washed twice with 28 kg of diisopropyl ether.
Finally, it was dried for 12 h at 50 ꢀC and 30 mbar to yield 42.6 kg
(87%) of the desired product. In one production campaign, four
batches were manufactured to yield a total of 152.6 kg of 3 with
an average yield of 79%.
Melting point: 80 ꢀC; HPLC (method C): Rt = 14.1 min,
98 area %; 1H NMR (500 MHz, D6-DMSO): δ = 3.71 (m, 9H),
4.85 (dd, J = 9.1 Hz, 23.6 Hz, 1H), 5.09 (s, 2H), 7.32ꢀ7.38 (m,
5H), 8.36 (d, J = 8.8 Hz, 1H); 13C NMR (125 MHz, D6-DMSO):
δ = 51.4, 52.6, 52.7, 53.6, 66.0, 127.7 (2C), 127.8, 128.3 (2C),
136.6, 156.1, 167.2; HRMS (ESI+): Calc. mass (C13H18NO7P) =
332.0894; found = 332.0890; IR (KBr): ν (cmꢀ1) = 526, 577,
603, 672, 704, 734, 757, 785, 831, 843, 862, 919, 962, 985, 1003,
1030, 1061, 1172, 1212, 1239, 1274, 1333, 1427, 1455, 1467,
1532, 1716, 1749, 2859, 2964, 3033, 3230, 3418.
2-Benzyloxy-5-bromobenzaldehyde (4b). Under a nitrogen
atmosphere a 26 L-HC4 vessel was charged with a solution of
3.02 kg (15.0 mol) of 5-bromosalicylaldehyde and 22.5 g
(0.15 mol) of sodium iodide in a mixture of 6.0 L of ethanol and
1.5 L of water. Subsequently, 954 g (9.0 mol) of sodium carbonate
was added at 22 ꢀC, and it was heated to reflux for 3 h. To the
mixture was added 2.09 kg (16.5 mol) of benzylchloride, it was
stirred at reflux for 5 h and at 25 ꢀC overnight. For workup, 4.5 L
of water was added; it was filtered and washed four times with
1.0 L of aq sulphuric acid (2% in water) and five times with 1.0 L
of water. The product was dried in vacuo at 40 ꢀC to yield 4.31 kg
(14.8 mol, 99% yield) of the benzyl ether 4b. In one campaign,
42.6 kg of the title compound was synthesized in 10 batches with
an average yield of 98%.
(m, 11H), 7.53 (J = 8.83 Hz, 1H), 7.85 (s, 1H), 9.28 (s, breit,
1H); 13C NMR (125 MHz, D6-DMSO): signals for aromatic
carbon partially overlap δ = 52.3, 65.8, 69.9, 112.1, 115.1, 124.2,
126.7, 127.1, 127.3 (2C), 127.4, 127.6, 127.8, 127.9, 128.3, 128.5
(2C), 131.3, 133.1, 136.3, 154.4, 155.5, 165.3; HRMS (ESI+):
Calc. mass (C25H22BrNO5) = 496.0755; found = 496.0754; IR
(KBr): ν (cmꢀ1) = 526, 607, 629, 651, 698, 736, 764, 786, 811,
886, 961, 1019, 1045, 1131, 1220, 1235, 1254, 1299, 1367, 1385,
1408, 1437, 1453, 1485, 1511, 1586, 1693, 1733, 3029, 3331.
2-(2-Benzyloxy-5-bromophenyl)-[1,3]dioxolane (18). In a
nitrogen atmosphere 4.30 kg (14.8 mol) of benzyloxy-5-bromo-
benzaldehyde was suspended in 16.3 L of cyclohexane, and 29.4 g
(0.30 mol) of sulphuric acid and 13.8 kg (17.8 mol) of 1,3-
propandiol were added. It was heated to reflux, and 290 mL of
water was separated by azeotrope distillation, while the tempera-
ture rose from 77 to 83 ꢀC. It was cooled to 60 ꢀC, and within
15 min a solution of 39.2 g sodium carbonate in 165 mL of water
was added. It was cooled to 10 ꢀC, stirred for 30 min, and filtered.
The filter cake was washed twice with 3.0 L of water, and the
product was dried in vacuo at 40 ꢀC to yield 4.65 kg (13.3 mol,
90% yield) of the dioxolane 18. In eight batches, a total of 35.3 kg
was produced, which corresponds to an average yield of 89%.
Melting point: 93 ꢀC; HPLC (method A): Rt = 24.05 min,
98%. 1H NMR (500 MHz, D6-DMSO): δ = 1.40 (d, J = 13.4 Hz,
1H), 2.02 (m, 1H), 3.90 (t, J = 10.6 Hz, 2H), 4.12 (m, 2H), 5.15
(s, 2H), 5.79 (s, 1H), 7.05 (d, J = 8.8 Hz, 1H), 7.32 (m, 1H),
7.38ꢀ7.48 (m, 5H), 7.55 (d, J = 2.4 Hz, 1H); 13C NMR (125
MHz, D6-DMSO): δ = 25.3, 66.8 (2C), 69.6, 95.4, 111.9, 115.0,
127.0 (2C), 127.7, 128.4 (2C), 129.4, 129.5, 132.2, 136.7, 154.4;
HRMS (ESI+): Calc. mass (C17H17BrO3) = 349.0434; found =
349.0436.
IR (KBr): ν (cmꢀ1) = 554, 673, 695, 733, 816, 859, 880, 899,
930, 958, 990, 1011, 1022, 1034, 1103, 1134, 1147, 1186, 1246,
1275, 1384, 1412, 1430, 1452, 1464, 1498, 1597, 2848, 2954, 2976.
[4-(Benzyloxy)-3-formylphenyl]boronic Acid (11). In a
250-L alloy vessel was added 21.0 kg (49.0 mol) n-butyl lithium
(1.6 M in hexane) to a solution of 15.0 kg (43.0 mol) of 2-
(2-benzyloxy-5-bromophenyl)-[1,3]-dioxolane in 83 L of THF
at ꢀ75 ꢀC within 1.5 h. It was stirred for 30 min followed by the
addition of 9.30 kg (49.0 mol) of triisopropylborate. Within 3 h it
was warmed to ꢀ10 ꢀC, and 16.8 kg of dilute sulphuric acid (20%
in water) was added. It was maintained at 40 ꢀC for 4 h. Subse-
quently, the organic layer was washed four times with 13.5 kg
water; 67.3 L of toluene was added, approximately 150 L of
solvent was distilled off in vacuo, and 97.4 L of methylcyclohex-
ane was added. It was cooled to 0 ꢀC, filtered, and washed twice
with 18.8 L of methylcyclohexane. Finally, the product was dried
at 30 mbar and 60 ꢀC for 16 h to yield 7.6 kg (69%) of the boronic
acid 11. A second batch delivered an additional 7.0 kg of the
boronic acid 11.
Melting point: 199 ꢀC; HPLC (method B): Rt = 17.6 min,
99%; 1H NMR (500 MHz, D6-DMSO): δ = 5.32 (s, 2H), 7.30 (d,
J = 8.5 Hz, 1H), 7.36 (m, 1H), 7.42 (t, J = 7.6 Hz, 2H), 7.53 (d, J =
7.4 Hz, 2H), 8.08 (dd, J = 1.2 Hz, 8.3 Hz, 1H), 8.24 (d, J = 1.2 Hz,
1H), 10.50 (s, 1H); 13C NMR (125 MHz, D6-DMSO): δ = 69.7,
112.9, 123.7, 127.5 (2C), 127.9, 128.5 (2C), 134.4, 136.4, 142.2,
162.0, 189.3; CHN (C14H13BO4): calc. C 65.7% H 5.2%; found
C 65.9% H 5.2%; MS (ES+): [M + H]+ = 257, [M + Na]+ = 279;
IR (KBr): ν (cmꢀ1) = 523, 617, 651, 699, 740, 802, 825, 915, 934,
1015, 1042, 1075, 1116, 1158, 1223, 1270, 1282, 1296, 1315,
1350, 1381, 1427, 1454, 1463, 1499, 1572, 1604, 1673, 2870,
3315 (broad).
Melting point: 71 ꢀC; HPLC (method A): Rt = 22.3 min, 99%;
1H NMR (500 MHz, D6-DMSO): δ = 5.29 (s, 2H), 7.31 (d, J =
8.9 Hz, 1H), 7.37 (m, 1H), 7.42 (m, 2H), 7.52 (m, 2H), 7.76 (d,
1H), 7.79 (m, 1H), 10.34 (s, 1H); 13C NMR (125 MHz, D6-
DMSO): δ = 70.2, 112.7, 116.7, 125.9, 127.5 (2C), 128.0, 128.5
(2C), 129.9, 136.0, 138.3, 159.5, 187.9; HRMS (ESI+): Calc.
mass (C14H11BrO2) = 289.9942; found = 289.9939; IR (KBr):
ν (cmꢀ1) = 530, 665, 694, 735, 812, 883, 1024, 1032, 1124, 1185,
1237, 1274, 1307, 1396, 1409, 1450, 1476, 1485, 1496, 1590,
1678, 2566.
3-(2-Benzyloxy-5-bromophenyl)-2-benzyloxycarbonyl
Aminoacrylic Acid Methyl Ester (10). To a solution of 1.83 kg
(6.30 mol) of 2-benzyloxy-5-bromobenzaldehyde and 2.30 kg
(6.93 mol) of the phosphonate 3in 12.2 L of THF 0.84 kg (7.25 mol)
of N,N,N0,N0-tetramethylguanidine was added at ꢀ20 ꢀC with-
in 22 min. It was maintained at this temperature for 4.5 h and
subsequently warmed up to 20 ꢀC. To the reaction mixture was
added 24.4 kg of water, and the pH was adjusted to 1ꢀ2 using
conc. sulphuric acid. It was stirred at 3 ꢀC for 1.5 h, filtered, and
washed three times with 3.75 L of water. The product was dried
at 40 ꢀC in vacuo to obtain 3.06 kg (5.10 mol, 98% yield) of the
desired dehydroamino acid 10, which was isolated in 96% purity
(HPLC, method A) with a Z/E ratio of 97:3. For an analytical
characterization, a sample was recrystallized from methyl tert-
butyl ether/dioxane (2.5:1) to yield >99.5% area % of the Z-
isomer. In one production campaign, seven batches of 10 were
synthesized as described above to give 14.3 kg of the desired
material with an average yield of 97%.
Melting point 156 ꢀC; HPLC (method A): Rt = 25.1 min,
99.7%; 1H NMR (500 MHz, D6-DMSO): δ = 3.70 (s, 3H), 5.11
(s, 2H), 5.20 (s, 2H), 7.12 (d, J = 8.83 Hz, 1H), 7.33ꢀ7.46
1354
dx.doi.org/10.1021/op200207h |Org. Process Res. Dev. 2011, 15, 1348–1357

Organic Process Research & Development
ARTICLE
(5-Benzyloxycarbonylamino-2-tert-butoxycarbonyl Amino-
pentanoylamino)(dimethoxyphosphoryl)acetic Acid Methyl
Ester (12). A solution of 547 g (1.54 mol) of 3 and 37.0 g of 5%
palladium on charcoal (50 weight % in water) in 2.1 L of THF
was exposed to an 80 bar hydrogen atmosphere for 4 h at 20 ꢀC.
The reaction mixture was passed through a plug of celite, and it
was washed with 400 mL of THF. To the combined filtrates
were added 513 g (1.4 mmol) of Boc-Orn-(Z)-OH and 246 g
(1.4 mmol) of 2-chloro-4,6-dimethoxy-1,3,5-triazine, and it was
cooled to 0ꢀ5 ꢀC. Subsequently, 154 mL (1.4 mmol) of N-
methylmorpholine was added dropwise within 20 min, and it was
stirred at 20 ꢀC for 18 h. The solvent was removed in vacuo, and
the residue was redissolved in ethyl acetate. It was filtered off, and
the filter cake was washed twice with 250 mL of ethyl acetate. The
combined filtrates were washed with 1.05 L of a sat. aqueous
NaHCO3 solution, twice with 1.05 L of aqueous HCl (1 molar),
again with 1.25 L of a sat. aqueous NaHCO3 solution and with
1.0 L of water. Finally, the solvent was distilled off in vacuo to give
758 g (1.39 mol, 90% yield) of the desired product as an oil.
According to HPLC analysis (method A), the crude product had
a purity of 77 area %. It was used in the next step without further
purification. For analytical characterization, a sample of the above
material was purified by preparative HPLC. Conditions for the
preparative HPLC separation: Phenomenex Prodigy ODS3,
column symmetry: 250 mm ꢁ 50 mm, 10 μm; eluent: A = water, B =
acetonitrile, gradient = 0ꢀ22 min: 60% A, to 20% A; 22ꢀ23 min
20% A to 60% A; 23ꢀ28 min 60% A; flow rate = 140 mL/min;
detection at 240 nm.
7.54 (d, J = 7.41 Hz, 2H), 7.65 (m, 2H), 7.85 (dd, J = 2.05, 8.67
Hz, 1H), 7.93 (m, 1H), 8.01 (m, 1H), 9.29 (m, breit, 1H), 10,48
(s, 1H); 13C NMR (125 MHz, D6-DMSO): δ = 52.2, 65.8, 69.8,
70.0, 113.3, 114.7, 122.4, 124.7, 125.1, 126.0, 126.6, 127.1, 127.3
(2C), 127.4 (2C), 127.5 (2C), 127.7, 127.8, 127.9, 128.2 (2C),
128.4 (2C), 128.5 (2C), 128.8, 131.0, 132.2, 133.8, 136.4, 136.6,
136.7, 154.5, 155.9, 159.7, 165.5, 189.0; HRMS (ESI+): Calc.
mass (C39H33NO7) = 628.2330; found = 28.2346; IR (KBr):
ν (cmꢀ1) = 559, 623, 656, 698, 743, 776, 803, 865, 912, 1011,
1071, 1127, 1141, 1182, 1231, 1247, 1263, 1294, 1376, 1435,
1454, 1483, 1504, 1605, 1639, 1692, 1721, 3271.
Lithium-(2Z)-2-{[(benzyloxy)carbonyl]amino}-3-[4,40-bis-
(benzyloxy)-30-formylbiphenyl-3-yl]acrylate (22). A solution
of 273 g (6.5 mol) LiOH in 4.6 L of water was added to a solution
of 3.30 kg (5.12 mol) of the methyl ester 21 in 11.6 L of dioxane
within 1 h at 20 ꢀC. It was stirred at this temperature for 17 h
followed by the addition of 13.2 L of water. It was filtered off,
washed twice with 2.2 L of water and three times with 2.2 L of
dioxane. The product was dried at 40 ꢀC in vacuo to yield 3.16 kg
(5.10 mol, 99% yield) with a purity of 98% (HPLC, method A).
According to this protocol, five batches of 22 were manufactured
to give a total of 15.3 kg, which corresponds to an average yield of
99%. Upon recrystallization the Li-salt dissociated. Thus, for an
analytical characterization we generated the free carboxylic acid
as follows: 125 g of the Li-salt was suspended in 2 L of water
and treated with aqueous HCl (20%) to adjust the pH to 2. It
was stirred for 3 h, filtered off, and washed three times with 0.5 L
of water. The crude carboxylic acid was recrystallized from
isopropanol/water (3:1).
Melting point: 163ꢀ165 ꢀC; HPLC (method C): Rt = 31.4 min,
99.2 area %; 1H NMR (500 MHz, D6-DMSO): δ = 5.07 (s, 2H),
5.24 (s, 2H), 5.35 (s, 2H), 7.21ꢀ7ꢀ48 (m, 15H), 7.54 (d,
J = 7.24 Hz, 2H), 7.63 (m, 2H), 7.83 (dd, J = 2.21 Hz, 8.51 Hz,
1H), 7.91 (m, 1H), 7.99 (s, 1H), 9.04 (s, broad, 1H), 10.46 (s,
1H), 12.79 (s, broad, 1H); 13C NMR (125 MHz, D6-DMSO):
signals for aromatic carbons partially overlap δ = 65.7, 69.8, 70.0,
113.3, 114.8, 124.7, 125.1, 127.1, 127.4 (2C), 127.5 (2C), 127.7,
127.9, 128.0, 128.2 (2C), 128.5 (2C), 128.6 (2C), 131.0, 132.4,
133.8, 136.4, 136.7, 136.8, 154.5, 155.8, 159.7, 166.4, 189.1;
HRMS (ESI+): Calc. mass (C38H31NO7) = 614.2174; found =
614.2199; IR (KBr): ν (cmꢀ1) = 696, 731, 803, 1006, 1022, 1072,
1128, 1180, 1245, 1270, 1286, 1376, 1430, 1453, 1483, 1498,
1507, 1519, 1605, 1627, 1692, 2852, 3034, 3252.
HPLC (method D): Rt = 10.16 min, 99 area %; 1H NMR (500
MHz, D6-DMSO): δ = 1.37 (s, 9H), 1.33ꢀ1.58 (m, 4 H), 2.98
(m, 2H), 3.66ꢀ3ꢀ73 (m, 9H), 4.09 (m, 1H), 5.00 (m, 2H), 5.08
(m, 1H), 6.94 (dd, J = 9.52 Hz, 21.44 Hz, 1H), 7.24 (m, 1H), 7.35
(m, 5H), 8.66 (dd, J = 8.20 Hz, 18.6 Hz, 1H); 13C NMR (125
MHz, D6-DMSO): δ = 26.0, 28.2 (3C), 29.1, 40.0, 49.0, 52.8,
53.5, 53.7, 53.8, 65.2, 78.1, 127.8 (2C), 128.4 (3C), 137.3, 155.3,
146.1, 166.9, 172.6; HRMS (ESI+): Calc. mass (C23H36N3O50P) =
546.2212; found = 546.2209; IR (KBr): ν (cmꢀ1) = 631, 699,
740, 779, 841, 868, 916, 1036, 1168, 1256, 1368, 1393, 1439,
1455, 1529, 1715, 1751, 2959, 3034, 3318.
2-Benzyloxycarbonyl Amino-3-(4,40-bis-benzyloxy-30-for-
mylbiphenyl-3-yl)acrylic Acid Methyl Ester (21). In a nitrogen
atmosphere, a 36-L stainless steel vessel was charged with a solu-
tion of 1.58 kg (3.0 mol) of the arylbromide 10 (containing
3.4 area % E-isomer), 0.92 kg (3.60 mol) of the boronic acid 11,
37.0 g (0.12 mol) tris-o-tolylphosphine and 6.7 g (0.03 mol) of
palladium(II)acetate in 11.4 L of toluene and 305 mL of water at
20 ꢀC. It was heated to reflux temperature (approximately 88ꢀ
91 ꢀC) and 1.05 L of a 4 M aqueous potassium carbonate solution
was dosed within 45 min. It was further kept at this temperature
for 3.5 h followed by a hot filtration through a bed of charcoal.
The filtrate was cooled to 20 ꢀC overnight and subsequently
cooled to 0ꢀ5 ꢀC. It was filtered off, washed twice with 2.3 L of
toluene and three times with 1.5 L of water. In order to remove
residues of water-soluble salts, the crude product was suspended
in 10.1 L of water, stirred at 22 ꢀC for 1 h, filtered, and washed
with 2.0 L of water. Finally, it was dried in vacuo at 40 ꢀC to give
1.90 kg (99% yield) of the desired Suzuki coupling product 21. In
10 batches, a total of 19.5 g of 21 was produced with an average
yield of 94%.
(2Z)-2-{[(Benzyloxy)carbonyl]amino}-3-[4,40-bis(benzyloxy)-
30-((1Z)-2-{[N5-[(benzyloxy)carbonyl]-N2-(tert-butoxycarbonyl)-
Lornithyl]amino}-3-methoxy-3-oxoprop-1-en-1-yl)biphenyl-
3-yl]acrylic Acid (9). A solution of 177 g (5.52 mmol, 76 HPLC-
area %) of phosphonate 12 in 338 mL of 2-methyltetrahydrofuran
and 18 mL of water was added to a solution of 173 g (279 mmol)
of lithium carboxylate 22 in 850 mL of 2-methyltetrahydrofuran.
The resulting suspension was cooled to ꢀ5 to 0 ꢀC, and within
90 min 98 g (818 mmol) N,N,N0,N0-tetramethylguanidine was
added. It was slowly warmed to 20 ꢀC and stirred for 18 h at this
temperature. For workup, 700 mL of deionized water was added,
and the pH was adjusted to 2.0, employing concentrated sulphu-
ric acid. The layers were separated, and the organic layer was
washed twice with 700 mL of deionized water. It was filtered
through a bed of zeolite, and ∼900 mL of the solvent was de-
stilled off in vacuo. The resulting suspension was diluted with
3.0 L of acetonitrile and heated to 50 ꢀC for 3 h. Over a period of
18 h it was cooled to 20 ꢀC. Subsequently, it was cooled to 0 ꢀC,
and the resulting solid was filtered off. It was washed three times
Melting point: 142 ꢀC; HPLC method B: HPLC: Rt = 38.3
min, 98.9%; 1H NMR (500 MHz, D6-DMSO): δ = 3.72 (s, 3H),
5.12 (s, 2H), 5.25 (s, 2H), 5.35 (s, 2H), 7.22ꢀ7.48 (m, 15H),
1355
dx.doi.org/10.1021/op200207h |Org. Process Res. Dev. 2011, 15, 1348–1357

Organic Process Research & Development
ARTICLE
with 400 mL of cold acetonitrile and 1.0 L of deionized water.
Finally, the product was dried in vacuo at 45 ꢀC to yield 196 g
(68%) of the desired product.
ester 26. At 0ꢀ5 ꢀC a solution of 79.5 g (786 mmol) N-
methylmorpholine in 80 mL of THF was added to the crude
solution of 25 (2.13 L) and 57.3 g (419 mmol) of isobutyl chloro-
formate within 45 min. It was stirred at 0 ꢀC for 1 h. Subse-
quently, a solution of 77.2 g (419 mmol) of pentafluorophenol in
150 mL of THF was added within 45 min, and the reaction was
allowed to warm to 20 ꢀC overnight. For workup, 850 mL of
water was added, and the pH was adjusted to 2.0 using 73.5 mL of
an aqueous 20% HCl. The organic layer was separated and con-
centrated to a volume of 1.8 L. The resulting solution was slowly
added to 2.0 L of water within 2 h, and it was stirred at 20 ꢀC for
18 h to complete the crystallization process. It was filtered,
washed four times with 0.5 L of water and six times with 0.5 L of
methyl tert-butyl ether. Finally, it was dried in vacuo to yield
308 g (98% for two steps) of the desired pentafluorophenol 26
with a purity of 91 area % (HPLC method A). In total, 2.6 kg of the
desired material was synthesized according to this procedure in
laboratory scale. For an analytical characterization, a portion of the
product was recrystallized three times from THF/acetonitrile (1:2).
Melting point: 177 ꢀC; HPLC (method A): Rt = 30.40 min, 97
Melting point: 184 ꢀC; HPLC (method E): Rt = 35.05 min,
1
98.3%; H NMR (500 MHz, D6-DMSO): δ = 1.22ꢀ1.62 (m,
13H), 2.90 (m, 2H), 3.66 (s, 3H), 4.10 (m, 1H), 4.98 (s, 2H),
5.05 (m, 2H), 5.19 (m, 2H), 5.25 (s, 2H), 6.91 (dd, J = 7.9 Hz,
1H), 7.09ꢀ7.48 (m, 25H), 7.66 (s, 1H), 7.72 (d, J = 8.2 Hz, 1H),
7.91 (m, 1H), 7.98 (m, 1H), 9.73 (s, 1H), 9.01 (s, breit, 1H),
12.77 (s, breit, 1H); 13C NMR (125 MHz, D6-DMSO): δ = 25.9,
28.1 (3C), 28.8, CH2NHCbz covered by solvent signal, 52.0,
53.7, 65.0, 65.7, 69.7, 69.8, 77.9, (signals for aromatic-C partly
overlap) 113.0, 113.2, 122.5, 124.0, 126.5, 126.6, 126.9, 127.1,
127.6 (2C), 127.7 (2C), 127.8 (2C), 128.2 (2C), 128.3 (2C),
128.4 (2C), 128.5 (2C), 128.9, 131.6, 132.0, 136.7, 137.1, 154.5,
155.3, 156.0, 165.4, 172.2; HRMS (ESI+): Calc. mass (C59H60-
N4O13) = 1033.4230; found = 1033.4215; IR (KBr): ν (cmꢀ1) =
696, 736, 806, 1024, 1051, 1132, 1168, 1246, 1367, 1453, 1492,
1522, 1608, 1658, 1693, 2948, 3334.
(2S)-2-{[(Benzyloxy)carbonyl]amino}-3-{4,40-bis(benzyloxy)-
30-[(2S)-2-({(2S)-5-{[(benzyloxy)carbonyl]amino}-2-[(tert-
butoxycarbonyl)amino]pentanoyl}amino)-3-methoxy-3-
oxopropyl]biphenyl-3-yl}propanoic Acid (25). In an 10-L
autoclave, a solution of 562 g (544 mmol) of the dehydroamino
acid 9 and 3.54 g (5.4 mmol) (S,S)-Et-DuPHOS-Rh in 3.5 L of
THF was stirred in a hydrogen atmosphere at 80 bar and 25 ꢀC
for 40 h. The resulting solution was filtered through a plug of zeo-
lite, and it was washed with 0.5 L of THF. After analytical release
(98 area %; >99% de) the combined filtrates (4.26 L) were direct-
ly taken to the next step. For an analytical characterization, a por-
tion of the product solution was passed through a plug of silica
and subsequently crystallized three times from THF/acetonitrile.
Melting point: 137 ꢀC; HPLC (method A): Rt = 26.9 min,
99.4%; determination (HPLC):21 column KBD5287 (Bayer
HealthCare chiral stationary phase based on poly(N-methyl-
acryloyl-L-leucine-dicyclopropylmethylamide), column symmetry:
250 mm ꢁ 4.6 mm; T: 22 ꢀC; eluent: isocratic ethyl acetate
30 min; flow rate: 1 mL/min; detection at 220 nm; Rt for (S,S,S)-
diastereomer = 11.2 min, >99.5% de; 1H NMR (500 MHz, D6-
DMSO): δ = 1.16ꢀ1.50 (m, 13H), 2.80ꢀ2.97 (m, 4H), 3.18ꢀ
3.36 (m, 2H), 3.50 (s, 3H), 3.91 (m, 1H), 4.42 (m, 1H), 4.62 (m,
1H), 4.90ꢀ5.04 (m, 4H), 5.15ꢀ5.19 (m, 4H), 6.74 (d, J =
8.2 Hz, 1H), 7.07 (t, J = 8.2 Hz, 1H), 7.18ꢀ7.53 (m, 26H), 7.60
(d, J = 8.2 Hz, 1H), 8.23 (d, J = 6.9 Hz, 1H), 12.67 (s, broad, 1H);
13C NMR (125 MHz, D6-DMSO): δ = 25.8, 28.0 (3C), 29.5,
32.0, 32.1, CH2NHCbz covered by solvent signal, 51.6, 52.0,
53.5, 53.6, 65.0, 65.1, 69.1, 69.2, (signals for aromatic-C partly
overlap), 112.1, 112.2, 125.3, 125.5, 125.7, 126.1, 126.6, 126.8
(2C), 126.9 (2C), 127.4 (2C), 127.5, 127.2, 127.5, 127.6 (2C),
128.1 (2C), 128.2 (2C), 128.2 (2C), 128.3 (2C), 128.7, 129.2,
131.9, 132.2, 136.8, 137.1, 155.0, 155.3, 155.9, 156.0, 172.0,
173.6; HRMS (ESI+): Calc. mass (C59H64N4O13) = 1037.4543;
found = 1037.4546; IR (KBr): ν (cmꢀ1) = 631, 696, 733, 778,
799, 861, 1024, 1051, 1132, 1168, 1247, 1367, 1439, 1453, 1493,
1525, 1608, 1657, 1692, 2947, 3032, 3064, 3333.
1
area %; H NMR (500 MHz, D6-DMSO): δ = 1.16ꢀ1.50 (m,
13H), 2.85 (m, 3H), 3.07 (m, 1H), 3.19 (m, 1H), 3.46 (m, 1H),
3.50 (s, 3H), 3.61 (m, 1H), 3.92 (m, 1H), 4.63 (m, 1H), 4.86ꢀ
5.04 (m, 4H), 5.21 (m, 4H), 6.74 (d, J = 8.2 Hz, 1H), 7.06ꢀ7.62
(m, 28H), 8.22 (m, 1H); 13C NMR (125 MHz, D6-DMSO):
δ = 25.8, 28.1 (3C), 29.5, 31.8, 32.1, (CH2NHCbz covered by
solvent signal), 51.6, 52.0, 53.5, 53.6, 65.1, 65.7, 69.2, 69.3, 77.9,
(signals for aromatic-C partly overlap), 112.3, 112.4, 124.7, 125.3,
125.4, 125.7, 126.1, 126.9, 127.0 (2C), 127.4, 127.5 (2C), 127.6
(3C), 127.7 (2C), 127.8, 128.2 (3C), 128.3 (3C), 128.4 (2C),
128.8, 129.2, 132.0, 132.1, 136.5, 136.8, 137.0, 137.1, 137.2, 155.1,
155.4, 156.0, 168.8, 172.1, 173.6; HRMS (ESI+): Calc. mass
(C65H63F5N4O13) = 1203.4385; found = 1203.4371; IR (KBr): ν
(cmꢀ1) = 576, 628, 696, 731, 775, 802, 859, 902, 1000, 1025,
1048, 1117, 1134, 1171, 1249, 1311, 1367, 1377, 1440, 1453,
1498, 1521, 1608, 1659, 1693, 1800, 2947, 3033, 3065, 3337.
Methyl (8S,11S,14S)-5,17-Bis(benzyloxy)-14-{[(benzyloxy)-
carbonyl]amino}-11-(3-{[(benzyloxy)carbonyl]amino}propyl)-
10,13-dioxo-9,12-diazatricyclo-[14.3.1.12,6]henicosa-1(20),
2(21),3,5,16,18-hexaene-8-carboxylate (2). To a solution of
1.27 kg (1.06 mol) of pentafluorophenylester 26 in 7.8 L of
dioxan was added 3.59 L of HCl in dioxane (4 M) at 24 ꢀC within
30 min. It was stirred for 20 h at 23 ꢀC. The resulting reaction
mixture was added over a period of 70 min to a solution of 1.55 kg
(15.3 mol) of triethylamine in 7.0 L of THF. It was maintained at
23 ꢀC for 80 min, and subsequently 9.4 L of THF and 7.8 L of
acetonitrile were added. It was filtered and washed three times
with 5.1 L of acetonitrile, and five times with 6.1 L of water. The
crude product was suspended in a mixture of 2.0 L of methanol
and 1.5 L of water and stirred at 45 ꢀC for 3 h. It was filtered and
washed three times with 1.0 L of methanol and twice with 0.5 L of
water. The washing protocol using methanol and water was
repeated twice until the filtrate was free of water-soluble salts.
The resulting product was finally dried in vacuo at 45 ꢀC for 48 h
to yield 705 g (77%) of the desired macrocycle 2. In a second
batch 662 g (68% yield) of 2 was obtained. In total, 1.37 kg of the
building block 2 was synthesized in an average yield of 73%. The
purity of the material was sufficient for further use: analytical
HPLC (method A, sample injected as DMSO solution): Rt =
27.2 min, 97 area %.
Pentafluorophenyl (2S)-2-{[(Benzyloxy)carbonyl]amino}-
3-{4,40-bis(benzyloxy)-30-[(2S)-2-({(2S)-5-{[(benzyloxy)car-
bonyl]amino}-2-[(tert-butoxycarbonyl)amino]pentanoyl}-
amino)-3-methoxy-3-oxopropyl]biphenyl-3-yl}propanoate
(26). The crude THF solution obtained for carboxylic acid 25
(4.26 L) was divided into two batches of the same size (each of
2.13 L) and consecutively transformed into the pentafluorophenyl
For an analytical characterization a sample of 2 was further
purified by precipitation from a DMSO solution using acetonitrile
1356
dx.doi.org/10.1021/op200207h |Org. Process Res. Dev. 2011, 15, 1348–1357

Organic Process Research & Development
ARTICLE
followed by extensive washing with methanol and water: melting
(7) (a) Zoller, U.; Ben-Ishai, D. Tetrahedron 1975, 31, 863.
(b) Williams, R. M.; Aldous, D. J.; Aldous, S. C. J. Org. Chem. 1990,
55, 4657. (c) Shankar, R.; Scott, A. I. Tetrahedron Lett. 1993, 34, 231.
(8) For alternative methods see: (a) Harayama, Y.; Yoshida, M.;
Kamimura, D.; Kita, Y. Chem. Commun. 2005, 13, 1764. (b) Kawai, M.;
Hosoda, K.; Omori, Y.; Yamada, K.; Hayakawa, S.; Yamamura, H.;
Butsugan, Y. Synth. Commun. 1996, 26, 1545. (c) Kawai, M.; Neogi, P.;
Khattri, P. S.; Butsugan, Y. Chem. Lett. 1990, 4, 577.
(9) for alternative approaches to phosphorylglycines see: (a) Ku, B.;
Oh, D. Y. Tetrahedron Lett. 1988, 29, 4465–4466. (b)Seki, M.;Matsumoto,
K. Synthesis 1996, 5, 580. (c) Williams, R. M.; Fegley, G. J. Tetrahedron Lett.
1992, 33, 6755.
point: >260 ꢀC; HPLC (method A, sample injected as DMSO
1
solution): Rt = 27.2 min, 99 area %; H NMR (500 MHz, D6-
DMSO): δ = 1.46ꢀ1.64 (m, 4H), 2.89 (m, 2H), 3.00 (m, 2H),
3.39 (m, 2H), 3.67 (s, 3H), 4.54 (m, 1H), 4.59 (m, 1H),
4.82ꢀ4.96 (m, 3H), 5.00ꢀ5.17 (m, 6H), 6.33 (d, J = 7.57 Hz,
1H), 7.05 (m, 4H), 7.25ꢀ7.48 (m, 23H), 8.66 (d, J = 9.1 Hz,
1H), 8.99 (d, J = 8.5 Hz, 1H); 13C NMR (125 MHz, D6-DMSO):
δ = 25.8, 28.1, 30.4, 30.5, CH2NHCbz covered by solvent signal,
50.5, 51.4, 52.2, 54.0, 65.0, 65.3, 69.4 (2C), (signals for aromatic-
C partly overlap), 112.1, 112.3, 124.2, 124.3, 124.9, 125.6, 126.4,
127.1, 127.2 (2C), 127.4. 127.5, 127.6 (2C), 127.7, 128.1 (2C),
128.2 (2C), 128.4 (2C), 128.9, 131.4, 131.6, 137.1, 137.2, due
to the occurrence of amide rotamers, the carbonyl signals are
partly doubled: 154.6, 154.9, 155.5, 156.0, 169.3, 171.8, 171.9;
HRMS (ESI+): Calc. mass (C54H54N4O10) = 919.3913; found =
919.3907; IR (KBr): ν (cmꢀ1) = 518, 576, 597, 626, 696, 778,
801, 817, 837, 858, 896, 912, 1020, 1054, 1128, 1180, 1245, 1381,
1435, 1454, 1498, 1533, 1639, 1665, 1692, 1744, 2664, 2949,
3032, 3064, 3273.
(10) Fresneda, P. M.; Molina, P.; Bleda, J. A. Tetrahedron 2001, 57,
2355.
(11) (a) Schmidt, U.; Griesser, H.; Leitenberger, V.; Lieberknecht,
A.; Mangold, R.; Meyer, R.; Riedl, B. Synthesis 1992, 487. (b) for an
alternative DoebnerꢀKnoevenagel protocol see: Xu, F.; Zacuto, M.;
Yoshikawa, N.; Desmond, R.; Hoerrner, S.; Itoh, T.; Journet, M.;
Humphrey, G. R.; Cowden, C.; Strotman, N.; Devine, P. J. Org. Chem.
2010, 75, 7829.
(12) For an alternative approach to 11 see: Holland, R.; Spencer, J.;
Deadman, J. J. Synthesis 2002, 16, 2379.
(13) These conditions have previously been described for the
methoxy derivative: Wolan, A.; Laczynska, A.; Rafinski, Z.; Zaidlewicz,
M. Lett. Org. Chem. 2004, 1, 238.
(14) (a) Schmidt, U.; Riedl, B. Synthesis 1993, 815. (b) Mawer, I. M.;
Kuglagowski, J. J.; Leeson, P. D.; Grimwood, S.; Marshall, G. R. Bioorg.
Med. Chem. Lett. 1995, 5, 2643. (c) Tilley, J.; Kaplan, G.; Fotouhi, N.;
Wolitzky, B.; Rowan, K. Bioorg. Med. Chem. Lett. 2000, 10, 1163.
(15) If the hydrogenation was conducted in the presence of one
equivalent of HCl, we observed no beneficial stabilization of the result-
ing HCl salt.
’ AUTHOR INFORMATION
Corresponding Author
Joachim.Krueger1@bayer.com.
Present Addresses
§Roche Diagnostics GmbH, D-82377 Penzberg, Germany
(16) For Suzuki coupling with 11 see also: Schmidt, U.; Leitenberger,
V.; Meyer, R.; Griesser, H. J. Chem. Soc., Chem. Commun. 1992, 13, 951.
(17) (a) (S,S)-Et-DuPhos: Cobley, C.; Johnson, N. B.; Lennon, I. C.;
McCague, R.; Ramsden, J. A.; Zanotti-Gerosa, A. In Asymmetric Catalysis
on Industrial Scale; Wiley-VCH: Weinheim, Germany, 2004; pp 269ꢀ
282, (b) (S,S)-Et-butiphane: Berens, U. (Solvias)., WO/2003/031456,
2003; Chem. Abstr. 2003, 138, 321403. (c) catASium, M; Krauter, J.;
Riermeier, T. PharmaChem 2004, 3, 62–64. Riermeier, T.; Monsees, A.;
Holz, J.; Boerner, A. Chim. Oggi 2004, 22.
(18) For a double hydrogenation protocol for biphenomycin analo-
gues see also: Carlstr€om, A.-S.; Frejd, T. J. Chem. Soc., Chem. Commun.
1991, 1216.
(19) The influence of chloride impurities on the catalyst inhibition
has been studied before: Cobley, J. C.; Lennon, I. C.; Praquin, C.;
Zanotti-Gerosa, A. Org. Process Res. Dev. 2003, 7, 407.
’ ACKNOWLEDGMENT
We thank colleagues Joachim Rehse and his team for pilot-
plant and Hellmut Zehner and his team for kg-lab support; Dieter
Heitkamp for process safety and Oliver Brauner for hydrogena-
tion support. We are grateful to Marcel Baierl, Indre Bajoras,
Frank Cebulla, Antje Dietzel, Jens Gottschewski, Denise Heuser,
Daniel Hillenbrand, Adolf Jugard, Rudolf Renelt, Peter Schmidt,
and Undine Zuther for their skillfull preparative work.
’ REFERENCES
(1) Uchida, I.; Shigematsu, N.; Ezaki, M.; Hashimoto, M.; Aoki, H.;
Imanaka, H. J. Antibiot. 1985, 38, 1462.
(2) (a) Carbonnelle, A.-C.; Zhu, J. Org. Lett. 2000, 2, 3477.
(b) Paintner, F. F.; Gorler, K.; Voelter, W. Synlett 2003, 4, 522. (c) Lepine,
R.; Zhu J. Org. Lett. 2005, 7, 2981. (d) Waldmann, H.; He, Y.-P.; Tan, H.;
Arve, L.; Arndt, H.-D. Chem. Commun. 2008, 43, 5562. (e) He, Y.-P.;
Tan, H.; Arve, L.; Baumann, S.; Waldmann, H.; Arndt, H.-D. Chem.
Asian J. 2011, 6, 1546. (f) For reviews see: Feliu, L.; Planas, M. Int. J. Pept.
Res. Ther. 2005, 11, 53. (g) von Nussbaum, F.; Brands, M.; Hinzen, B.;
Weigand, S.; H€abich, D. Angew. Chem., Int. Ed. 2006, 45, 5072.
(3) (a) Schmidt, U.; Meyer, R.; Leitenberger, V.; Griesser, H.;
Lieberknecht, A. Synthesis 1992, 1025. (b) Schmidt, U.; Meyer, R.;
Leitenberger, V.; Lieberknecht, A.; Griesser, H. J. Chem. Soc., Chem.
Commun. 1991, 275.
(20) For installation of PFP-esters via a mixed anhydride protocol
using more expensive Fmoc-Cl see: Tantry, S. J.; Babu, V. V. S. Lett. Pept.
Sci. 2004, volume date 2003, 10(5ꢀ6), 655.
(21) The absolute stereochemical assignment has been conducted
by analytical comparison of the final product 2 with a sample from the
discovery route. The initial assignment has been accomplished by an
X-ray analysis of an early intermediate as described in ref 4.
(4) Lampe, T.; Adelt, I.; Beyer, D.; Brunner, N.; Endermann, R.;
Ehlert, K.; Kroll, H.-P.; von Nussbaum, F.; Raddatz, S.; Rudolph, J.;
Schiffer, G.; Schumacher, A.; Cancho-Grande, Y.; Michels, M.; Weigand,
S. WO 2005/033129, 2005.
(5) (a) Schmidt, U.; Leitenberger, V.; Griesser, H.; Schmidt, J.;
Meyer, R. Synthesis 1992, 1248. (b) Schmidt, U.; Meyer, R.; Leitenberger,
V.; Lieberknecht, A. Angew. Chem., Int. Ed. Engl. 1989, 28, 929.
(6) (a) Schmidt, U.; Lieberknecht, A.; Schanbacher, U.; Beuttler, T.
Angew. Chem., Int. Ed. Engl. 1982, 21, 776. (b) Schmidt, U.; Lieberknecht,
A; Wild, J. Synthesis 1984, 53.
1357
dx.doi.org/10.1021/op200207h |Org. Process Res. Dev. 2011, 15, 1348–1357