Accelerated Koenigs - Knorr Glucuronidation of a Deactivated Nitrophenol: Unveiling the Role of Polyamine Additive 1,1,4,7,10,10-Hexamethyltriethylenetetramine1 through Design of Experiments

Article abstract of DOI:10.1021/jo035285n

1,1,4,7,10,10-Hexamethyltriethylenetetramine (HMTTA) emerged from a limited parallel screening of selected polyamines as the most appropriate additive for an especially problematic Koenigs-Knorr glucuronidation. This initial finding rapidly evolved into a

Full text of DOI:10.1021/jo035285n

Acceler a ted Koen igs-Kn or r Glu cu r on id a tion of a Dea ctiva ted
Nitr op h en ol: Un veilin g th e Role of P olya m in e Ad d itive
1,1,4,7,10,10-Hexa m eth yltr ieth ylen etetr a m in e1 th r ou gh Design of
Exp er im en ts
Federica Stazi,† Giovanni Palmisano,† Marco Turconi,‡ Simona Clini,‡ and
Marco Santagostino*,‡
Dipartimento di Scienze Chimiche Fisiche e Matematiche, Universita` dell’Insubria, Via Valleggio 11,
22100 Como, Italy, and Chemistry Research Centre, Boehringer Ingelheim Italia, Via Lorenzini 8,
20139 Milano, Italy
msantagostino@mil.boehringer-ingelheim.com
Received September 2, 2003
1,1,4,7,10,10-Hexamethyltriethylenetetramine (HMTTA) emerged from a limited parallel screening
of selected polyamines as the most appropriate additive for an especially problematic Koenigs-
Knorr glucuronidation. This initial finding rapidly evolved into a reliable and high-yielding procedure
through the use of two sets of experimental designs. The detailed effect of the stoichiometry of
reagents and the amount of amine additive on reaction yield was elucidated. The complexity of the
response surface for product yield, described by a third-order polynomial equation, together with
ancillary kinetic experiments evidenced the multiple role of HMTTA in the present glucuronidation
process.
In tr od u ction
Notwithstanding the steadily increasing number of
ingenious modifications or alternatives to the most
common glucuronidation methods,3 the delivery of this
sugar moiety to alcohols or phenols remains challenging
and far from general, often requiring extensive ad hoc
optimization.
In this light, the technique of design of experiments
(DOE) stands as a precious tool for the effective inves-
tigation of an appropriate “experimental space” as de-
limited by the low and high values of the reaction
variables (factors) under study.4 By virtue of a DOE
multivariate approach, the influencing variables and,
appealingly, their interactions are easily revealed. The
statistical nature of the method allows a precise quan-
tification of the effect of significant factors on a given
response and, through the generation of inferential
models, furnishes the experimentalist an optimum set
of reaction conditions.
In the endoplasmic reticulum of body cells, in particu-
lar hepatocytes, UDP-glucuronosyl transferases (UD-
PGTs) catalyze highly effective glucuronidation of en-
dogenous substrates as well as xenobiotics, thus playing
a central role in the mechanisms of detoxification during
“phase 2” metabolism.
Disappointingly, chemists often compete inadequately
with nature’s efficiency when it comes to performing the
same transformation outside the body.2 Reasons for this
are envisaged especially in the electron-withdrawing
characteristics of the 5-alkoxycarbonyl group in the
glycosyl donor, which destabilizes the incipient C-1
cation, often resulting in the prevalence of undesired
deactivation pathways, e.g., ortho ester formation, acyl
transfer to the aglycon, or eliminations of C-1 or C-4
substituents.
* To whom correspondence should be addressed. Phone: +39-
025355383. Fax: +39-025355205.
Resu lts a n d Discu ssion
† Universita` dell’Insubria.
‡ Boehringer Ingelheim Italia.
In the course of a recent research project, it became
necessary to synthesize the glucuronide 1a of BIMC-0576
(1) There is no clear preference in the literature for 1,1,4,7,10,10-hexa-
methyltriethylenetetramine designation as different acronyms are in
use. “hmteta”: (a) Baxter, I.; Drake, S. R.; Hursthouse, M. B.;
McAleese, J .; Malik, K. M. A.; Michael, D.; Mingos, P.; Otway, D. J .;
Plakatouras, J . C. Polyhedron 1998, 17, 3777. “hmten”: (b) Darr, J .
A.; Poliakoff, M.; Li, W-S.; Blake, A. J . J . Chem. Soc., Dalton Trans.
1997, 17, 2869. “hmtt”: (c) Shintani, N.; Kotaki, J .; Sone, K.; Fukuda,
Y. Bull. Chem. Soc. J pn. 1993, 66, 784. “Me6trien”: (d) Drumhiller, J .
A.; Montavon, F.; Lehn, J .; Taylor, R. W. Inorg. Chem. 1986, 25, 3751.
(e) “HMTTA”: Reich, H. J .; Green, D. P.; Medina, M. A.; Goldenberg,
W. S.; Gudmundsson, B. O¨ .; Dykstra, R. R.; Phillips, N. H. J . Am.
Chem. Soc. 1998, 120, 7201. Here we prefer the adoption of this latter
abbreviation for consistency with the other acronyms used in the
present paper.
(3) (a) Stachulski, A. V. Tetrahedron Lett. 2001, 42, 6611 and
references therein. See also: (b) Rukhman, I.; Yudovich, L.; Nisnevich,
G.; Gutman, A. L. Synthesis 2000, 9, 1241. (c) Scheinmann, F.;
Stachulski, A. V.; Ferguson, J .; Law J .-L. WO patent 00/78764 A1,
2000. (d) Vaccaro, W. D., Sher, R.; Davis, H. R., J r. Bioorg. Med. Chem.
Lett. 1998, 8, 35. (e) Badman, G. T.; Green, D. V. S.; Voyle, M. J .
Organomet. Chem. 1990, 388, 117.
(4) (a) Anderson, M.; Whitcomb, P. J . DOE simplified: Practical
Tools for effective Experimentation; Productivity Inc.: Portland, OR,
2000. (b) Montgomery, D. C. Design and Analysis of Experiments;
Wiley: New York, 2001. (c) Carlson, R. Design and Optimization in
Organic Synthesis; Elsevier: New York, 2000.
(2) For a review, see: Stachulski, A. V.; J enkins, N. J . Nat. Prod.
Rep. 1998, 173.
10.1021/jo035285n CCC: $27.50 © 2004 American Chemical Society
Published on Web 01/15/2004
J . Org. Chem. 2004, 69, 1097-1103
1097

Stazi et al.
SCHEME 1. Over a ll Syn th etic Rou te to Glu cu r on id e 1a a n d Str u ctu r a l F or m u la for Tr icep Am in e
(Tr ic-NH2) a n d Differ en tly P r otected Br om o Su ga r s 5a ,b
a
For the first application of this reagent to the mild conversion of electron-deficient aryl fluorides to phenols, see: Rogers, J . F.; Green,
D. M. Tetrahedron Lett. 2002, 43, 3585.
1b, an oxygenated metabolite of the antidepressant drug
flibanserin 1c.5
electrophile 2 with the appropriate nucleophiles in the
indicated order, multigram quantities of 3 could be easily
accessed. Moreover, the transformation of the protected
glucuronide 4 into the desired final product 1a could be
accomplished uneventfully via nitro group reduction,
cyclization to benzimidazolone, and subsequent complete
deprotection of the sugar moiety (Scheme 1).8
Disappointingly, glucuronidation of 3 under Koenigs-
Knorr conditions performed extremely poorly either
employing triacetyl bromo sugar 5a or the usually more
performing Snatzke’s higher homologue 5b.9 Although
with this latter donor glucuronidation of 3 proceeded
cleanly to completeness within 18 h under standard
conditions (5b 1.2 equiv, Ag2O 2.7 equiv in acetonitrile,
in the presence of 4 Å molecular sieves) returning 4 as
the only product as judged by TLC analysis, the isolated
yield was as low as 3%. Considering the high potential
for metal ion chelation of the aminoethylpiperazinyl
moiety in 4,10 it was considered the chance of irreversible
binding to silver ions to explain the observed mass
unbalance. To test this hypothesis, the glucuronidation
reaction was repeated in the presence of 10 equiv of a
chelating amine, namely N,N,N′,N′-tetramethylethylene-
diamine (TMEDA). Interestingly, a 27% yield of glucu-
ronide 4 could be isolated under these conditions but any
attempt at further optimization by the traditional chang-
ing one-variable-at-a-time (OVAT) approach did not meet
with substantial success.
Initial extensive experimentation demonstrated the
inapplicability of a more conservative approach,6 i.e.,
direct conjugation to the corresponding aglycon 1b. The
complete unreactivity of 1b under a wide range of
glucuronyl transfer conditions forced us to a more
circuitous route calling for the use of an appropriately
competent glucuronyl acceptor. In this regard, the use
of the o-nitrophenol 3 appeared promising since the
beneficial effect of electron-withdrawing groups ortho
positioned with respect to hydroxy groups has already
been demonstrated in a number of cases.7 Further
potential benefits of this route included the easily avail-
ability of 3. This could be very straightforwardly prepared
by a new one-pot, double fluorine displacement sequence
on 2,6-difluoronitrobenzene 2. Thus, by reacting the bis-
Under these circumstances, it was considered prefer-
able to adopt the more systematic DOE tactic. Our initial
optimization approach was to examine the effect of
various amines and silver sources on reaction yield and
purity. To this end, a set of five amine additives (see
Table 1) was selected on the basis of different complexing
ability11 and basicity,12 each amine being tested in the
presence of two silver sources: Ag2O and Ag2CO3. The
(5) Borsini, F.; Brambilla, A.; Ceci, A.; Cesana, R.; Giraldo, E.;
Ladinsky, H.; Monferini, E.; Turconi, M. Drugs Future 1998, 23, 9.
(6) Paleari, F.; Rondina, F. Unpublished results.
(7) Alkyl salicylates (a, b) or ortho-nitro phenols (c, f) are superior
substrates in Koenigs-Knorr glucuronidations due to hydroxyl activa-
tion via intramolecular H-bonding. For pertinent examples, see: (a)
Lunsford, C. D.; Murphey, R. S. J . Org. Chem. 1956, 21, 580. (b)
Rossignol, J .-F.; Ayers, M. S. US Patent 5,952,622, 1999. (c) Kiss, J .;
Burkhardt, F. Carbohydr. Res. 1970, 12, 115. (d) Roffler, S.; Chern,
J .-W.; Leu, Y.-L. US Patent 6,043,367, 2000. (e) J acquesy, J .-C.; Gesson,
J .-P.; Monneret, C.; Mondon, M.; Renoux, B.; Florent, J .-C.; Koch, M.;
Tillequin, F.; Sedlacek, H. H.; Gerken, M.; Kolar, C.; Gaudel, G.;
Bosslet, K.; Czech, J .; Hoffman, D.; Seemann, G.; Schorlemmer, H.-
U.; Dickneite, G. US Patent 5,561,119, 1996. (f) Goto, K.; Nakamura,
S.; Morioka, Y.; Kondo, M.; Naito, S.; Tsutsumi, K. Chem. Pharm. Bull.
1996, 44 (3), 547.
(8) See the Supporting Information.
(9) Vlahov, J .; Snatzke, G. Liebigs Ann. Chem. 1983, 570.
(10) Stable complexes between silver (I) ions and polyamines have
been observed in acetonitrile [(a) Grzejdziak, A. Pol. J . Chem. 1994,
68, 1395] and DMSO: (b) Comuzzi, C.; Novelli, V.; Portanova, R.;
Tolazzi, M. Supramolecular Chem. 2001, 13, 455.
1098 J . Org. Chem., Vol. 69, No. 4, 2004

Koenigs-Knorr Glucuronidation of a Deactivated Nitrophenol
TABLE 1. Am in e Ad d itives a n d Silver Sou r ces P a r a llel
Scr een in g for Glu cu r on id a tion of 3a
TABLE 2. Va r ia bles Con sid er ed a n d Levels Used in th e
27-4 F a ctor ia l Design (Resolu tion III)a
in situ yield of 4 (%)
variables
under study
(-)
0
(+)
expt
amine additiveb
Ag2O
Ag2CO3
A
B
C
D
E
F
precomplex. timeb (min)
reaction timec (h)
Ag2CO3 (equiv)
HMTTA (equiv)
5b (equiv)
0
2
1.5
1.5
1.5
0
30
4
2.6
7.1
2.2
60
6
3.8
12.6
3
100
1.5
1-2
3-4
5-6
7-8
9-10
DIPEA
0
0
TMEDA
DIPEDA
DMEDA
HMTTA
24.8
26.1
0
29.3
15.2
0
40.6
38.6
4 Å mol sieves (mg)
solvent (mL)
50
1
a
G
0.5
10 mg scale, 1.5 equiv of bromo sugar 5b, 2.7 equiv of silver
source, 10 equiv of amine additive, acetonitrile (0.15 mL), 6 h.
DIPEA, diisopropylethylamine; TMEDA, N,N,N′,N′-tetrameth-
ylethylenediamine; DIPEDA, N,N′-diisopropylethylenediamine;
DMEDA, N,N′-diimethylethylenediamine; HMTTA, 1,1,4,7,10,10-
hexamethyltriethylenetetramine.
a
Constant factors: temperature (25 °C), stirring speed (1000
b
rpm). b Stirring time before addition of bromo sugar 5b and phenol
c
3. Stirring time after addition of 5b and 3.
TABLE 3. 27-4 F a ctor ia l Design :a Exp er im en ta l Ma tr ix
a n d Mea su r ed Resp on seb
10 experiments were conducted on a small-scale (10 mg)
parallel format with 1.5 equiv of bromo sugar 5b and the
appropriate silver source (2.7 equiv) in acetonitrile at rt
in the presence of 10 equiv of amine additive. Analysis
of in situ yields (Table 1), as determined by HPLC
(external standard) after 6 h, clearly suggested a strong
positive correlation between coordinating ability of the
amine additive and solution yield, as evident by the
substantially better result (average in situ yield 39.6%)
obtained in the presence of tetradentate amine: 1,1,4,7,-
10,10-hexamethyltriethylentetramine (HMTTA).13 Two
other considerations appeared meaningful at this point.
First, use of strongly basic conditions should be limited.14
Thus, the chelating yet unhindered and highly basic
DMEDA performed (in situ yield 0%) much worse than
TMEDA or DIPEDA (27.0% and 20.6% yield respec-
tively). Second, the choice of silver source did not appear
to significantly affect yields.15
factor settings
expt
A
B
C
D
E
F
G
yield of 4 (%)
1
2
3
4
5
6
7
8
9
-
+
-
+
-
+
-
+
0
-
-
+
+
-
-
+
+
0
-
-
-
-
+
+
+
+
0
+
-
-
+
+
-
-
+
0
+
-
+
-
-
+
-
+
0
+
+
-
-
-
-
+
+
0
-
+
+
-
+
-
-
+
0
14.7
19.5
24.4
11.2
34.2
83.2
56.5
55.4
50.2
43.2
50.5
10
11
0
0
0
0
0
0
0
0
0
0
0
0
0
0
a
50 mg scale, 25 °C, stirring speed 1000 rpm; the experiments
b
were performed in random order. Measured by HPLC using an
external reference standard.
them convinced us to apply a fractional factorial design
with low resolution, i.e., a 27-4 resolution III design. Such
“saturated” designs (k + 1 runs, with k factors to be
studied) allow relevant effects to be discovered with a
minimum number of experimental trials,16 yet their
highly fractional nature produces critical confusion be-
tween effects. In the specific “resolution III” case, main
effects are confused (“aliased”) with one or more two-
factor interactions; that is, the effect due to a specific
main factor may be additionally due to the combined
effects of all or each of these higher order interactions.
Although in many cases this low resolution can still be
adequate, when needed the appropriate duplication of
experimental runs will resolve (de-alias) significant ef-
fects. Thus, a complete (or partial) “foldover” of a resolu-
tion III design makes it resolution IV, where main effects
are now aliased only with highly unlikely three- or
higher-order interactions. In the present case, a set of
eight experiments was generated,17 and three center
points replicates were added to provide a measure of
pure experimental error and to check for curvature, i.e.
non linearity of the response. The 11 experiments (50
mg scale) were performed and analyzed in random
order to measure the in situ yield of the desired glucu-
ronide 4. The experimental matrix and the responses
obtained under each factorial combination are depicted
in Table 3.
On the basis of these findings and considering the
results from the attempted OVAT optimization, a list of
seven expected critical parameters (see Table 2) in this
reaction was set up to be studied in a two-level factorial
design. Their number and the likely interactions between
(11) The affinity of screened amines for metal ions is expected to
increase in the following order: DIPEA < TMEDA < DMEDA ∼
DIPEDA < HMTTA. Tertiary amines are poorer σ-donors and,
therefore, also poorer ligands than the corresponding primary and
secondary amines; for
a thorough discussion on the effect of N-
alkylation on the stability of polyamine-metal complexes, see ref 12d.
(12) In water, the pKa values for the screened amines are as
follows: (a) DIPEA ) 11.0. Ramirez, F.; Marecek, J . F. Tetrahedron
1981, 35, 1581. (b) DIPEDA ) 10.40, 7.59 and DMEDA ) 10.29, 7.47.
Basolo, F.; Murmann, R. K.; Chen, Y. T. J . Am. Chem. Soc. 1953, 75,
1478. (c) TMEDA ) 9.14, 5.90. Gustafson, R. L.; Martell, A. E. J . Am.
Chem. Soc. 1959, 81, 525. (d) HMTTA ) 9.23, 8.47, 5.36, 1.68. Golub,
G.; Cohen, H.; Paoletti, P.; Bencini, A.; Messori, L.; Bertini, I.;
Meyerstein, D. J . Am. Chem. Soc. 1995, 117, 8353.
(13) While only one (mixed) complex between HMTTA and Ag (I)
cations has been observed and fully characterized (ref 1b), several
stable complexes of this polydentate amine and different metallic ions
are known. For diverse examples, see: [Cu] ref 1d, 12d and (a) Becker,
M.; Heinemann, F. W.; Knoch, F.; Donaubauer, W.; Liehr, G.; Schin-
dler, S.; Golub, G.; Cohen, H.; Meyestein, D. Eur. J . Inorg. Chem. 2000,
4, 719. [Gd and Y] ref 1a. [Pd] (b) Bazzicalupi, C.; Bencini, A.; Cohen,
H.; Giorgi, C.; Golub, G.; Meyerstein, D.; Navon, N.; Paoletti, P.;
Valtancoli, B. J . Chem. Soc., Dalton Trans. 1998, 10, 1625. [Ni] ref 1c.
(14) DIPEA is 107 times more basic in acetonitrile (pKa 18.1) than
in water (pKa 11.1). Kelly-Rowley, A. M.; Lynch, V. M.; Anslyn, E. V.
J . Am. Chem. Soc. 1995, 117, 3438.
(15) Such a set of experiments can be regarded and analyzed as a
general factorial design, where the factors (silver source and amine
additive) are categorical, i.e., non-numeric, and may have a different
number of levels (two and five, respectively). In this case, analysis of
results indicated no relationship between type of silver source and
product yield with a prob > 0.544; this indicates the observed variation
in responses is very probably due to experimental noise.
The normal plot of the effects for the in situ yield of 4
(Figure 1) revealed that, among the seven experimental
(16) See refs 4a, pp 109-122, and 4b, pp 337-347.
(17) Design Expert, version 6.0.4, by Stat-Ease (www.statease.com)
was used for generation and analysis of experimental designs.
J . Org. Chem, Vol. 69, No. 4, 2004 1099

Stazi et al.
TABLE 4. Cen tr a l Com p osite Design a (CCD) a n d
Su bsequ en t Mod ifica tion s: F a ctor s Settin gs a n d
Mea su r ed Resp on sesb
factor settings
A
B
C
measured responses
yield of residue
equiv of
equiv of
equiv of
expt
HMTTA Ag2CO3
sugar 5b
4 (%)
3 (%)
1c
0.7
2.1
0.7
2.1
0.7
2.1
0.7
2.1
0.2
2.5
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
0.2
0.2
0.2
2.5
2.5
2.5
3.7
3.7
5.1
5.1
3.7
3.7
5.1
5.1
4.4
4.4
3.3
5.5
4.4
4.4
4.4
4.4
4.4
4.4
4.4
4.4
4.4
4.4
4.4
4.4
4.4
4.4
2.1
2.1
2.1
2.1
2.4
2.4
2.4
2.4
2.25
2.25
2.25
2.25
2.0
82.8
63.5
81.6
69.9
87.5
72.8
85.1
74.3
62.1
70.2
73.9
77.0
71.9
79.6
71.8
79.8
77.6
77.2
76.6
78.4
75.0
78.4
80.7
66.4
72.1
72.5
11.4
11.2
5
15.9
12.4
20.8
4.8
15.4
3.6
19.0
21.6
9.1
14.4
12.1
12.0
14.0
13.9
14.2
12.8
13.3
3.3
2c
3c
4c
5c
6c
7c
8c
9c
10c
11c
12c
13c
14c
15c,d
16c,d
17c,d
18c,d
19c,d
20c,d
21e
22e
23e
24e
25e
26e
F IGURE 1. Normal pot for the screening 27-4 factorial design.
variables, stoichiometry of reagents was especially im-
portant, i.e., equiv of Ag2CO3 (factor C; effect +39.85),
equiv of bromo sugar 5b (factor E; effect +14.08), and
equiv of amine additive (factor D; effect -17.02). Pre-
complexation time between the amine additive and the
silver source (factor A; effect +9.87) also appeared to be
significant (prob ) 0.099). While the possibility of pre-
complexation between the amine additive and Ag+ ions
could not be regarded as remote,13 the alias structure for
A (A ) A + BD + CE + FG) reveals a strong correlation
between factor A and three two-factor interactions,
namely BD, CE and FG. In particular, the interaction
between equiv of Ag2CO3 and equiv of bromo sugar 5b
(CE)18 could be regarded especially probable also because
both parents (C and E) already appear significant.
Anyhow, since it was considered risky to arbitrarily
dismiss the importance of precomplexation time (A) in
favor of CE interaction, we considered the effect of A
being operative together with the CE interaction, and no
further experimentation was deemed necessary to in-
crease design resolution at this stage. Finally, reaction
time (B), presence of molecular sieves (F) and dilution
(G) appeared marginal (P > 0.30) and their level could
therefore be set to the most practical value.
2.5
2.25
2.25
2.25
2.25
2.25
2.25
2.1
2.25
2.4
2.1
3.4
9.2
19.0
18.8
17.8
2.25
2.4
a
100 mg scale, 25 °C, stirring speed 1000 rpm, 60 min
precomplexation time, 4 h reaction time; the experiments were
b
performed in random order within each block. Measured by
HPLC using external reference standards. c First block. Center
d
point. e Second block.
this design does not permit to establish which factor(s)
is (are) causing it. These observations underscore the
complexity of the system and justify the use of response
surface methods to get a definitive model of the reaction.
A central composite design19 (CCD) was therefore
implemented in a second experimental plan. The factors
under investigation (see Table 4) over the CCD five-level
format were reagent stoichiometries previously found
important. The selection of factors ranges under study
reflected the desire to thoroughly explore the region of
greatest promise as evident from the previous screening
experiment. The remaining significant factor, precom-
plexation time between HMTTA and Ag2CO3, was pur-
posefully omitted from the study to limit the number of
experimental runs and was kept to its best, yet practical
value of 60 min. With three factors, the proposed CCD
experiment (Table 4) comprised a total of 23 + 2 × 3 + 6
) 20 experimental runs: eight factorial runs, six axial
runs, and six replicates of the center point run. As before,
the runs (100 mg scale) were conducted in random order
in groups of five, gauging after 4 h the in situ yield of
glucuronide 4 and the content of residual starting mate-
rial 3. To this initial design a second block of experimen-
tal runs was eventually added. Six runs at different
bromo sugar 5b and HMTTA stoichiometries and a
constant amount of Ag2CO3 were separately performed
at the extremes of the experimental space in order to
Use of A(+CE), C, D, E led to a multiple regression
model (eq 1, coded factors) with no lack of fit (P ) 0.860)
and good prediction ability (R2 ) 0.994, adj R2 ) 0.989,
pred R2 ) 0.981, PRESS ) 0.076).
Ln(in situ yield of 4) )
3.41 + 0.047A + 0.59C - 0.25D + 0.17E (1)
As typical with fractional designs, the optimal reaction
conditions are located on the very limits of the design
space at Ag2CO3 3.8 equiv, bromo sugar 5b 3.0 equiv, and
HMTTA 1.5 equiv, where standard error is highest. In
addition, a rather large curvature effect is evident, the
actual averaged yield for the three center points being
significantly (P ) 0.0001) higher than the predicted value
(48.0% vs 30.2% respectively), yet the low resolution of
(18) The interaction effect between C and E is evident as follows:
experiments that employ high Ag2CO3 and low bromo sugar 5b
stoichiometries perform in average much worst than when both factors
are set to their high level (C+ E-: 40% vs C+ E+: 72%). On the other
hand, low Ag2CO3 stoichiometry affords low and non significantly
different yields, regardless of the amount of bromo sugar (C- E-: 16%
and C- E+: 20%).
(19) See ref 4b, pp 456-458.
1100 J . Org. Chem., Vol. 69, No. 4, 2004

Koenigs-Knorr Glucuronidation of a Deactivated Nitrophenol
TABLE 5. ANOVA Ta ble a n d Associa ted Sta tistics for Resp on se Su r fa ce Red u ced Cu bic Mod el for Yield of 4a
source
sum of squares
degrees of freedom
mean squares
F value
prob > F
âib
Ric
model
A
B
686.640
293.242
6.641
6
1
1
1
1
1
1
18
13
5
114.441
293.242
6.641
117.042
18.610
16.502
121.418
4.364
26.222
67.190
1.522
26.818
4.264
<0.0001
<0.0001
0.233
<0.0001
0.054
0.068
<0.0001
76.91
-9.58
0.70
2.57
-0.75
1.44
48.00
21.63
-3.29
17.16
-34.08
3.13
C
117.042
18.610
16.502
121.418
78.559
41.219
37.340
765.202
A2
AB
3.782
27.820
A3
2.51
7.82
residual
lack of fit
pure error
corr total
3.171
7.468
0.424
0.901
24
S ) 2.089, R2 ) 0.897, adj R2 ) 0.863, pred R2 ) 0.808, PRESS ) 146.59
a
The outlier point corresponding to reaction 9 of Table 4 was not included in the model (actual value: 62.100, predicted value: 79.960,
residual -17.860, Student residual -3.677, outlier t -5.788). Estimated regression coded coefficients for in situ yield of 4. c Estimated
regression actual coefficients for in situ yield of 4.
b
As evident from the plot, the yield of 4 correlates quite
oddly with the amount of amine additive. While optimal
performance in the coupling step is predicted when the
concentration of HMTTA approaches 0.6-0.7 equiv, a
further increase (up to ca. 1.8 equiv) of HMTTA hinders
the reaction. Surprisingly enough, sufficient concentra-
tions (ca. 2.5 equiv) of this amine additive reverse this
trend and conjugation appears again promoted. The
surface twist introduced by inclusion of interaction term
AB in the model accentuates (or flattens) this behavior
without nevertheless altering this basic frame.
Fundamentally, such a complex situation could under-
line a multiplicity of, often conflicting, roles played by
the amine additive in the present glucuronidation pro-
cess. Speculations on this point revealed that HMTTA
might facilitate the conjugation acting: (1) as a competi-
tive ligand for silver ions, capable of limiting the postu-
F IGURE 2. Influence of stoichiometry of Ag2CO3 (B) and
HMTTA (A) on the in situ yield of 4 at bromo sugar 5b equal
to 2.42 equiv.
lated irreversible bond to aminoethylpiperazinyl moiety
in the starting material 3 or the reaction product 4; (2)
as an activator for the silver source, functioning as a
shuttle for silver ions to be delivered into solution in a
improve model predictivity in these regions. The data for
glucuronide 4 yield from the first 20 experimental runs
(first block) were fitted to a cubic equation (R2 ) 0.828,
adj R2 ) 0.748, pred R2 ) -0.947, PRESS ) 1526.9). As
complexed, yet active form; (3) as a base: increasing the
nucleophilicity of phenol 3 by deprotonation.21 Alterna-
anticipated, inclusion of the results from the remaining
tively (4), high concentration of this basic additive may
six runs (second block) and exclusion of the strong outlier
hamper the conjugation by contributing to effective
corresponding to run 9 in Table 4 substantially improved
model prediction ability (R2 ) 0.897, adj R2 ) 0.863, pred
depletion of bromo sugar 5b from the reaction mixture
R2 ) 0.808, PRESS ) 146.6) without altering this
conclusion. Analysis of variance testing (Table 5) con-
via the known dehydrohalogenation at C-1, eventually
producing 2-pivaloyloxyglycal 6.22 As a matter of fact,
firmed the significance of the cubic term A3 as well as of
variables A2, A, and C and of two-factor interaction AB.
The statistically nonsignificant, yet hierarchically im-
portant,20 linear term B was also included in the analysis
to provide the model depicted in eq 2 (coded factors).
analysis of the data from the modified CCD indicated a
strong and significant (effect ) +8.68; P < 0.0001, see
the Supporting Information) direct relationship between
variable A, concentration of HMTTA, and content of
residual starting material 3, supporting this latter as-
sumption.
in situ yield of 4 ) 76.91 - 9.58A + 0.70B +
2.57C - 0.75A2 + 1.44AB + 2.51A3 (2)
The three-dimensional surface plot, illustrated in
Figure 2, was generated according to the cubic model and
demonstrates the dependence of in situ yield of glucu-
ronide 4 on the amounts of amine additive HMTTA
(variable A) and Ag2CO3 (variable B) at the optimal level
of bromo sugar 5b (variable C). Similar plots were
obtained at different concentrations of bromo sugar 5b.
To substantiate the hypothesis of irreversible binding
between substrates and silver ions and to gain a better
(21) However, this effect should not be considered as pivotal since
glucuronidation of potassium phenoxide of 3, delivered only trace
amount of glucuronide 4 when performed in the absence of the amine
additive.
(20) The hierarchy principle dictates that internal consistency in a
model containing a high-order term is promoted by inclusion of all the
lower order terms that compose it.
J . Org. Chem, Vol. 69, No. 4, 2004 1101

Stazi et al.
F IGURE 3. Kinetic studies for complexation of 3, 4, and 7 to silver ions: structure vs silver source relationship for complexation
equilibria of 3, 4, and 7 (a); effect of excess HMTTA (10 equiv) and silver source on complexation of 3 (b).
understanding of the role of the amine additive in the
glucuronidation process, a complementary set of experi-
ments was organized. To this end, mixtures of Ag2O (2.7
equiv) in acetonitrile were separately treated with start-
ing material 3, glucuronidation product 4, and 3-benzyl-
amino-2-nitrophenol 723 and monitored at 2, 4, 6, and 24
h stirring time using toluene (or xylene) as internal
standard. As the data in Figure 3a show, irreversible
binding is actually operative in the case of starting
material 3 since this nitrophenol was almost consumed
after 6 h, and completely absent after 24 h. By contrast,
reaction product 4 was not noticeably affected by the
presence of silver ions, its concentration in the reaction
mixture remaining constant over the same period of time.
F IGURE 4. Effect of different amounts of HMTTA on com-
Nitrophenol 7 appeared to bind, although less efficiently
plexation equilibria between 3 and Ag2CO3 at various reaction
times.
than 3, to silver ions. Clearly, while the presence of
aminoethylpiperazinyl moiety remains essential group
for an effective binding to silver ions it appears the actual
Thus, low (up to 0.7 equiv) HMTTA concentrations pro-
binding site must additionally comprise the phenolic
gressively facilitates transport of Ag+ in solution and
group. Surprisingly, with Ag2CO3 as silver source (2.7
subsequent complexation to the substrate. At higher
equiv), complexation of 3 looks much less efficient,
concentrations, HMTTA appears to compete with 3 in the
probably due to a lower solubility of this salt in acetoni-
binding to silver ions and complexation of the substrate
trile. As for the role of amine additive on complexation
is lessened.
equilibria (Figure 3b), it emerged that with Ag2O the
This subtle game of multiple and contradictory effects
presence of HMTTA (10 equiv) thwarts the complexation
produces the observed complexity in the response surface
of 3 thus demonstrating the ability of this tetramine to
describing the interplay between significant factors and
function as competitive ligand for silver ions. Oddly, in
glucuronidation in situ yield. Such a complex situation
the case of Ag2CO3 the effect of amine additive is opposite.
could hardly have been only pictured through a tradi-
tional monovariate approach lacking a homogeneous
coverage of the experimental space.
Here, the complexation of 3 is substantially increased by
addition of HMTTA (10 equiv) suggesting the prevalence
of silver salt dissolution processes over the expected silver
Interrogation of the multivariate model in eq 2 was
ions scavenging action. A final series of complexation
finally used to predict optimal conditions for glucuronida-
experiments at different HMTTA concentrations (Figure
tion of 3. Thus, an 86.5% ( 2.8% in situ yield was
expected by adding sequentially 1 equiv of nitrophenol 3
(100 mg scale) and 2.40 equiv of bromo sugar 5b to a
4) corroborates this point. The observed nonlinear rela-
tionship between complexation of 3 and concentration of
the tetramine additive suggests the coexistence of silver
prestirred (for 1 h) mixture of Ag2CO3 (3.7 equiv) and
ion competitive ligation and silver ion solvation processes.
HMTTA (0.70 equiv) in acetonitrile. This prediction was
confirmed by conducting the glucuronidation on a 1.0 g
(twice) and 3.5 g scale under these conditions. The
observed in situ yields for glucuronide 4 were 86.0, 87.2
(22) This HBr elimination is a persistent secondary process in
glucuronidations of phenolates with bromo sugar 5a : Anderson, F. B.;
Leabeck, D. H. Chem. Ind. (London) 1960, 967. Similarly, 5b is
and 85.7% by HPLC analysis, and the isolated (by flash
chromatography) yields were 80.6, 81, and 80% respec-
tively. Close monitoring of the reaction progress indicated
a performance peak after only 0.5 h, thus these bigger
completely decomposed when exposed to 1 equiv of EMDTA for 72 h
producing 6 as the major product (see the Supporting Information).
(23) Prepared in 64% yield, analogously to 2, by a one-pot reaction
of 2,6-difluoronitrobenzene with benzylamine and 2-hydroxyethyl-
methylsulphone in DMSO (see the Supporting Information.)
1102 J . Org. Chem., Vol. 69, No. 4, 2004

Koenigs-Knorr Glucuronidation of a Deactivated Nitrophenol
d u r e. A mixture of Ag2CO3 (3.5 g, 12.7 mmol) and HMTTA
(1.3 mL, 4.78 mmol) in 35 mL of CH3CN was stirred at room
temperature for 1 h. After this time, phenol 3 (3 g, 6.76 mmol),
bromo sugar 5b (8.57 g, 16.37 mmol), and 35 mL of CH3CN
were added. The reaction mixture was stirred at room tem-
perature for 30 min and then filtered, washing the insoluble
material with EtOAc (2 × 50 mL). After evaporation of the
solvent under reduced pressure and silica gel purification (SiO2
250 g; hexane/EtOAc 8/2 to 2/1), protected glucuronide 4 was
obtained as a bright yellow foam (4.65 gr, 80%): mp 96.8-97.5
°C; [R]D -0.7 (c ) 2, CHCl3); analytical HPLC tR 22.50 min
(gradient CH3CN/phosphate buffer pH 7: 60:40 to 80:20 5 min,
scale reactions were analyzed and worked up after this
time. Interestingly, under the same conditions but in the
absence of HMTTA additive, no trace of reaction product
4 was evident in the reaction mixture even after 4 h.
In summary, an experimental strategy that combines
a selected reagent screening and statistical design of
experiments (DOE) has allowed the rapid identification
of optimal conditions for an especially problematic
Koenigs-Knorr glucuronidation. The isolated yield in
this newly developed process was greater than 80% after
0.5 h of reaction time. These data compare extremely
favorably to those experienced prior to optimization (27%
after 18 h).
The chelating tetramine HMTTA initially emerged
from the screening of selected (poly-)amines as the most
promising candidate. Through the use of aptly designed
experimental matrixes the dependence of in situ yield of
glucuronide 4 from stoichiometry of HMTTA was mod-
eled. The observed complex relationship is thought to
reflect the multiplicity of roles played by this amine addi-
tive in the present process, as also suggested by a set of
ancillary kinetic experiments. From the experimental
data is also evident the remarkable effect of HMTTA in
significantly facilitating the glucuronidation process, even
when used catalytically. Thus while 0.7 equiv of HMTTA
provide optimal yields of 4, a 75-80% in situ yield was
also secured by utilizing only 0.2 equiv of this additive.
The generality of this method is presently under study.
Preliminary results on phenol 7 and vanillin indicate
analogous rate acceleration and yield enhancement as
for 3 (for 7 after 1 h reaction time: 100% yield at 95%
conversion with 0.7 equiv of HMTTA vs. 5% conversion
in the absence of the additive and in the case of vanillin
after 1 h reaction time: 93% yield at 95% conversion vs
33% yield at 35% conversion with no HMTTA) thus
setting the basis for extending the present results to a
wider range of differently substituted phenols.
1
80:20 20 min, postrun 7 min, flow rate 1 mL/min); H-NMR
(CDCl3, 500 MHz) δ 7.34 (1H, t, J ) 8.0 Hz), 7.23 (1H, t, J )
8.4 Hz), 7.11(1H, s), 7.07 (2H, t), 6.47 (1H, s, exchangeable),
6.46 (1H, d, J ) 8.4 Hz), 6.42 (1H, d, J ) 8.2 Hz), 5.44 (1H, t,
J ) 9.2 Hz), 5.36 (2H, m), 5.26 (1H, d, J ) 7.8 Hz), 4.21 (1H,
d, J ) 9.9 Hz), 3.72 (3H, s), 3.28-3.26 (6H, m), 2.7-2.6 (6H, m),
1.15 (9H, s), 1.14 (9H, s), 1.13 (9H, s); 13C-NMR (CDCl3, 125
MHz) δ 177.0, 176.4, 176.0, 166.7, 151.3, 150.8, 143.6, 134.6,
133.0, 131.4 (q, J C-C-F ) 31.12 Hz), 129.5, 124.3 (q, J C-F
)
270 Hz), 118.7, 115.7 (d, J C-C-C-F ) 3.75 Hz), 112.2, 107.9,
103.9, 98.9, 72.9, 71.5, 70.4, 69.0, 55.5, 52.8, 52.4 (2 × CH2),
48.6 (2 × CH2), 39.8, 38.7 (2 × C), 38.7, 27.1 (3 × CH3), 27.0
(3 × CH3), 26.9 (3 × CH3); MS (APCI+) 853 (M + H+, 100).
Anal. Calcd for C41H55F3N4O12: C, 57.74; H, 6.50; N, 6.57.
Found: C, 57.19; H, 6.47; N, 6.64.
Ack n ow led gm en t. We thank Dr. Enzo Cereda for
helpful discussions on alternative routes to glucuronide
1a . We are also grateful to Dr. Simona Dossena and Mr.
Maurizio Mondoni for mass and NMR spectra and to
Mr. Silvano Locatelli for elemental analyses.
Su p p or t in g In for m a t ion Ava ila b le: 1H and 13C NMR
spectra and experimental procedures for the preparation of
1a , 3, 4, 6, 7, 9, and 10 and HPLC analytical data for designed
CCD experiment. Analysis of CCD experiment for residual
starting material 3 and ANOVA table for the 27-4 factorial
design. This material is available free of charge via the
Internet at http://pubs.acs.org.
Exp er im en ta l Section
Meth yl 3-[[2-[4-(3-Meth ylp h en yl)p ip er a zin -1-yl]eth yl]-
a m in o ]-2-n it r o p h e n y l-2,3,4-t r is -O -(2,2-d im e t h y lp r o -
p a n oyl)-â-D-glu cop yr a n osid u r on a te (4). Sca le-u p P r oce-
J O035285N
J . Org. Chem, Vol. 69, No. 4, 2004 1103