β-Eliminative releasable linkers adapted for bioconjugation of macromolecules to phenols

Article abstract of DOI:10.1021/bc4002882

We recently reported a chemical approach for half-life extension that utilizes sets of releasable linkers to attach drugs to macromolecules via a cleavable carbamate group (Santi et al., Proc. Nat. Acad. Sci. U.S.A. 2012, 109, 6211-6216). The linkers undergo a β-elimination cleavage to release the free, native amine-containing drug. A limitation of the technology is the requirement for an amino group on the drug in order to form the carbamate bond, since most small molecules do not have an amine functional group. Here, we describe an approach to adapt these same β-elimination carbamate linkers so they can be used to connect other acidic heteroatoms, in particular, phenolic hydroxyl groups. The approach utilizes a methylene adaptor to connect the drug to the carbamate nitrogen, and an electron-withdrawing group attached to carbamate nitrogen to stabilize the system against a pH-independent spontaneous cleavage. Carbamate cleavage is driven by β-elimination to give a carboxylated aryl amino Mannich base which rapidly collapses to give the free drug, an aryl amine, and formaldehyde.

Full text of DOI:10.1021/bc4002882

Article
pubs.acs.org/bc
β‑Eliminative Releasable Linkers Adapted for Bioconjugation of
Macromolecules to Phenols
Eric L. Schneider, Louise Robinson, Ralph Reid, Gary W. Ashley, and Daniel V. Santi*
ProLynx, 455 Mission Bay Boulevard South, Suite 145, San Francisco, California 94158, United States
S
* Supporting Information
ABSTRACT: We recently reported a chemical approach for half-life extension
that utilizes sets of releasable linkers to attach drugs to macromolecules via a
cleavable carbamate group (Santi et al., Proc. Nat. Acad. Sci. U.S.A. 2012, 109,
6211−6216). The linkers undergo a β-elimination cleavage to release the free,
native amine-containing drug. A limitation of the technology is the requirement
for an amino group on the drug in order to form the carbamate bond, since
most small molecules do not have an amine functional group. Here, we describe
an approach to adapt these same β-elimination carbamate linkers so they can be used to connect other acidic heteroatoms, in
particular, phenolic hydroxyl groups. The approach utilizes a methylene adaptor to connect the drug to the carbamate nitrogen, and
an electron-withdrawing group attached to carbamate nitrogen to stabilize the system against a pH-independent spontaneous
cleavage. Carbamate cleavage is driven by β-elimination to give a carboxylated aryl amino Mannich base which rapidly collapses to
give the free drug, an aryl amine, and formaldehyde.
and an understanding of the chemistry of such conjugates. In the
present work, we describe a synthetic approach for adapting our β-
eliminative carbamate linkers to attach phenols to macromolecular
carriers. Using p-nitrophenol (pNPhOH) as a model leaving
group, we describe methods of synthesis, mechanistic studies that
define pathways and kinetics of cleavage, and structure activity
relationships that allow rational design of conjugates of phenol-
containing drugs with predictable rates of release.
INTRODUCTION
■
Conjugation of drugs to macromolecular carriers is a proven
strategy for improving pharmacokinetics. In one approach, a drug
is covalently attached to a long-lived circulating macromolecule −
such as polyethylene glycol (PEG) − through a linker that is slowly
cleaved to release the native drug.1,2 We have recently reported
such conjugation linkers that self-cleave by a nonenzymatic
β-elimination reaction in a highly predictable manner, and with
half-lives of cleavage spanning hours to over a year.3 In this
approach, a macromolecular carrier is attached to a linker that is
attached to a drug via a carbamate group (1; Scheme 1); the
β-carbon has an acidic carbon−hydrogen bond (C−H) and also
contains an electron-withdrawing pKa “modulator” (Mod) that
controls the acidity of that C−H. Upon hydroxide-ion catalyzed
proton removal (2), a rapid β-elimination occurs to cleave the
linker−carbamate bond and release the free drug or pro-drug and a
substituted alkene 3. The rate of drug release is proportional to the
acidity of the proton, and that is controlled by the chemical nature of
the modulator; thus, the rate of drug release is controlled by the pKa
modulator. Both in vitro and in vivo cleavages were linearly correlated
with electron withdrawing effects of the modulators and, unlike ester
bonds commonly used in releasable linkers,2 were not catalyzed by
general bases or serum enzymes. It was also shown that with
releasable conjugates of this type, the half-life of a rapidly cleared drug
is transformed into that of the conjugate itself; there is a longer half-life
and diminished Cmax compared to repeated bolus doses of a short-
lived drug needed to maintain a therapeutic level.
EXPERIMENTAL PROCEDURES
■
General. Materials purchased were the purest grade
commercially available. HPLC analyses were by reverse phase
using UV detection. Details are provided in SI.
Synthesis. O-[7-Azido-1-(phenylsulfonyl)heptan-2-yl]-N-phenyl-
carbamate (12A). To a stirred solution of the 1-(phenylsulfonyl)-7-
azido-2-heptanol (0.150 g; 0.504 mmol) and triphosgene (0.210 g;
0.708 mmol) in anhydrous tetrahydrofuran (6 mL) was added
pyridine (90 uL; 1.11 mmol) slowly over 5 min. The resulting
suspension was stirred at ambient temperature for 35 min at which
time analysis of the reaction mixture by TLC (50% ethyl acetate/50%
hexanes; UV) showed complete disappearance of the alcohol, Rf =
0.53, and the emergence of a new spot at Rf = 0.76. The precipitated
pyridine hydrochloride was removed by filtration and washed with
tetrahydrofuran (2 mL). The filtrate was concentrated on the rotovap.
The crude chloroformate was dissolved in anhydrous
tetrahydrofuran (6 mL) and aniline (55 μL; 0.603 mmol)
added. To this solution was added pyridine (98 μL; 1.21 mmol).
The suspension was stirred at room temperature for 2 h at
which time analysis of the reaction mixture by TLC (50% ethyl
While the β-eliminative linker technology well serves primary
and secondary amine-containing drugs that can form carbamates,
it is not suitable for most small molecules or functional groups of
peptides other than amine groups. We were in need of a method
to attach other functional groups − particularly phenolic
hydroxyls − of drugs to macromolecules via β-eliminative linkers,
Received: June 21, 2013
Revised: October 30, 2013
Published: October 31, 2013
© 2013 American Chemical Society
1990
dx.doi.org/10.1021/bc4002882 | Bioconjugate Chem. 2013, 24, 1990−1997

Bioconjugate Chemistry
Article
Scheme 1
Scheme 2
acetate/50% hexanes; UV) showed loss of the starting
chloroformate at Rf = 0.76 and the appearance of a new product
spot at Rf = 0.50. The precipitated pyridine hydrochloride was
removed by filtration and washed with tetrahydrofuran (2 mL).
The filtrate was concentrated and the residue partitioned
between ethyl acetate and water (20 mL each). The phases
were separated and the aqueous phase was extracted with ethyl
acetate (2 × 10 mL). The combined organic extracts were washed
with water, 1 M hydrochloric acid, and saturated sodium chloride
solution (10 mL each) and dried over magnesium sulfate. The solvent
was evaporated to give a pale yellow oil (0.288 g) that was purified
using a Thomson Instruments single step silica gel column (12 g)
eluting with 30% ethyl acetate/70% hexanes to give the carbamate as a
The combined organics were washed with brine, dried over
magnesium sulfate, and concentrated. The crude product was
purified using a Thomson Instruments single step silica gel
column (4 g) eluting with 40% ethyl acetate/60% hexanes to
furnish the phenyl ether as a pale yellow oil (0.013 g; 0.024 mmol;
1
25%). H NMR (CDCl3) 1.21 (4H, m), 1.40 (4H, m), 3.17
(3H, m), 3.35 (1H, m), 5.24 (1h, m), 5.61 (2H, br. s), 6.89 (2H, d,
J = 9.4 Hz), 7.04 (1H, m), 7.06 (1H, m), 7.33 (2H, m), 7.57 (2H, t,
J = 7.7 Hz), 7.67 (1H, m), 7.91 (2H, m), 8.15 (3H, m).
Compounds 14B,C and 6-azido-N-aryl carbamates 16A-C and
24 were prepared in a similar fashion, and are described in SI. An
alternate, improved procedure for N-chloromethylation of
carbamates is also provided in SI.
1
O-[7-PEG40 kDa-1-(phenylsulfonyl)heptan-2-yl]-N-(4-nitro-
colorless oil (0.190 g; 0.454 mmol; 90%). H NMR (CDCl3) 1.24
phenoxymethyl)-N-arylcarbamates (15) and O-[6-PEG40 kDa
-
(4H, m), 1.56 (2H, m), 1.73 (2H, m), 3.29 (2H, t, J = 6.8 Hz), 3.31
(1H, dd, J = 3.5 Hz, J = 14.7 Hz), 3.50 (1H, dd, J = 7.9 Hz, J = 14.7 Hz),
5.20 (1H, m), 6.19 (1H, br. s), 7.06 (1H, m), 7.25 (4H, m), 7.51
(2H, m), 7.58 (1H, m), 7.93 (2H, m). HPLC purity = 99% at 254 nm.
Compounds 12B,C and 6-azidohexyl-N-aryl carbamates 16A-
C were prepared in a similar fashion, and are described in SI.
O-[7-Azido-1-(phenylsulfonyl)heptan-2-yl]-N-(4-nitrophe-
noxymethyl)-N-phenylcarbamate (14A). O-[7-Azido-1-
(phenylsulfonyl)heptan-2-yl]-N-phenylcarbamate (0.040 g;
0.096 mmol) was dissolved in anhydrous tetrhydrofuran (0.5 mL)
and freshly distilled chlorotrimethylsilane (0.5 mL) and parafor-
maldehyde (0.007 g; 0.233 mmol) were added. The reaction
mixture was heated at 65 °C for 41 h at which time HPLC analysis
(after a MeOH quench) showed consumption of the starting
material and the appearance of a new product peak. The reaction
mixture was concentrated under reduced pressure. Ethyl acetate
(5 mL) was added to the residue and the solution reconcentrated.
This was repeated once more. 50% Ethyl acetate/50% hexanes
(3 mL) were then added and the mixture filtered. The filtrate
was concentrated under reduced pressure and the crude chloride
dissolved in anhydrous N,N-dimethylformamide (0.25 mL).
The above solution was added to a solution of sodium
p-nitrophenolate (0.025 g; 0.155 mmol) in anhydrous N,N-
dimethylformamide (0.25 mL). The resulting orange solution
was stirred at ambient temperature for 1 h. It was then diluted
with water (10 mL) and extracted with ethyl acetate (3 × 10 mL).
1-hexyl-N-(4-nitrophenoxymethyl)-N-arylcarbamate (17). To
a stirred solution of PEG40 kDa-DBCO (800 μL of a 3.13 mM
solution in acetonitrile; 2.5 μmol) was added of the azide 14 or
16 (33 μL of a 150 mM solution in acetonitrile; 4.95 μmol). The
solution was stirred at ambient temperature for 24 h. HPLC
analysis of the reaction solution showed that the DBCO maxima at
309 and 291 were undetectable at this time. Me-tBu ether (15 mL)
was added to the reaction solution and it was stirred for 30 min.
The solid PEG-conjugate was collected, washed with Me-tBu ether
(2 × 5 mL), and dried under high vacuum.
Properties of 15A-C and 17A-C are described in SI.
Kinetic Studies. Kinetic studies were performed by
spectrophotometric determination of pNPhO− at pH > 7.4
and analysis by HPLC, as detailed in SI.
Rate Equations. The derivation of rate equations for base-
catalyzed reactions of β-eliminative 4-nitrophenoxymethyl
carbamate linkers are given in SI.
RESULTS
■
Chemistry. The approach described here extends from the
work of Majumdar and Sloan4 who reported the synthesis and
properties of N-substituted-N-aryloxymethyl alkyl carbamates of
4 (Scheme 2, path A), containing a methylene bridge connecting
the carbamate nitrogen with an oxygen of a phenol, and a N-aryl
group that acts as a “stabilizer” against spontaneous hydrolysis.
1991
dx.doi.org/10.1021/bc4002882 | Bioconjugate Chem. 2013, 24, 1990−1997

Bioconjugate Chemistry
Article
These were prepared by N-chloromethylation of N-alkyl/aryl-O-
methyl carbamates, followed by reaction with a phenoxide anion.
Solvolysis to give the phenol and the iminium ion (5) occurred by a
slow pH-independent cleavage of the N-methylol bond over pH 4.6 to
8.8 that proceeds by a spontaneous SN1 mechanism. Subsequently, as
with related carbinolamides,5,6 after addition of water to the iminium
ion the N-hydroxymethyl carbamate should undergo acid- or base-
catalyzed conversion to the carbamate and CH2O. The SN1 reaction is
facilitated by the leaving group ability of the phenol, and shows a
negative linear correlation with the pKa of the phenol (slope = −0.93).
The reaction rate is decreased by electron withdrawing groups on the
carbamate nitrogen, but the effects were not thoroughly characterized.
Although not previously reported, we show below that an analogous
acid-catalyzed reaction occurs that proceeds by a similar pathway
(Scheme 2, path B) involving a protonated intermediate such as 6, or
alternative described below. Analogous N-substituted amidomethyl
esters of carboxylic acids also undergo solvolysis by similar SN1 and
acid catalyzed mechanisms.7
with pNPhONa gave the corresponding pNPhOH ethers 14A-C.
Using Cu-free [3 + 2] dipolar cycloaddition, connection of 14A-C
to mPEG40 kDa-dibenzocyclooctyne(DBCO)3,11 to form the PEG
conjugates 15A-C was achieved in near-quantitative yield. Starting
with 6-azido-1-hexanol, a similar sequence of conversions gave the
stable azidohexylcarbamates 16A-C without a modulator that were
converted to PEG-conjugates 17A-C. The generality of this
synthesis is demonstrated by its application to the tyrosine analog,
N-(6-DNP-aminocaproyl)-Tyr, of 14B with phenol pKa ∼ 10 (24;
see SI), and the anticancer agent SN38 (pKa ∼ 8.6).12
Solvolysis Products of N-aryl-N-pNPhOCH2-carbamates.
The solvolysis products of the N-(pNPhOCH2)carbamates studied
here identifies the pathways of pNPhOH release. That is, the
spontaneous (Scheme 2, path A) or acid catalyzed (Scheme 2, path B)
reactions should both provide pNPhOH and an N-aryl carbamate,
whereas β-eliminative cleavage (Scheme 3) should result in formation
of pNPhOH, the aryl amine and the alkenyl sulfone (PhSO2CH
CHR). In accordance with these expectations, in 1 N HCl, the azido-
linker-carbamate 14C gave pNPhOH and the N-arylcarbamate 12C;
the carbamate product remained stable in 1 N HCl for at least 140 h
thereafter. Under similar conditions, the corresponding PEG
conjugate 15C released only pNPhOH and the other product was
assumed to remain coupled to PEG as the N-aryl carbamate. At
pH 4−9, O-alkyl-N-aryl carbamates analogous to 16C have previously
been shown to undergo spontaneous SN1 hydrolysis to yield the N-
aryl carbamate and pNPhOH.4 At pH 8.4, the azido-linker-carbamate
with a PhSO2− modulator 14C yields the pNPhOH, aryl amine,
and alkenylsulfone as expected from a β-elimination reaction; at
pH > about 4 the corresponding PEG conjugates 15A-C released
pNPhOH and aryl amine leaving a PEG-alkenyl sulfone conjugate.
Solvolysis Kinetics and Mechanisms of N-aryl-N-
pNPhOCH2-carbamates with Stable Linkers (17A-C). The
pH-log kobsd profiles for solvolysis of the PEG40 kDa-pNPhOCH2-
carbamates containing stable linkers (Figure 1) fit eq 1.
In the present work, we prepared and studied carbamates
analogous to 4 that contain a β-eliminative group and can
undergo base-catalyzed cleavage of the O-alkyl bond (7, Scheme 3).
Scheme 3
We reasoned that if the SN1 reaction of such carbamates could be
sufficiently suppressed by an N-aryl stabilizer, cleavage of the C−O
bond of the carbamate would be driven by β-elimination to provide
the carboxylated Mannich base 8. We assumed that the carboxylated
secondary aryl amines 8 would undergo an acid-catalyzed
decarboxylation as described for their well-studied primary aryl-
amine counterparts,8,9 and that the resultant N-aryl Mannich base 9
would convert to the free drug and iminium ion, which would
dissociate to the aryl amine and CH2O.10
Based on the above considerations, we designed and executed
the synthetic sequence for the N-aryl-N-p-nitrophenoxymethyl-
(pNPhOCH2)-carbamate conjugates depicted in Scheme 4.
pNPhOH was selected as the leaving group because it can be
easily detected by UV, and is as acidic a phenol as we would
anticipate being present in most drugs. As aryl amine “stabilizer”
components of carbamates, we used aniline (pKa 4.6),
p-NH2PhC(O)NEt2, (pKa 3.5), and p-NH2PhSO2N(CH2)2O
(pKa 1.6) because of the anticipated suppression of SN1
hydrolysis provided by the electron withdrawing aryl groups.4
For the pKa modulator that drives β-elimination we used the
phenylsulfone (PhSO2-) moiety that has been extensively studied
as a pKa modulator in cleavage of carbamates of primary amines.3
As controls that cannot undergo β-elimination but could undergo
SN1 and acid-catalyzed hydrolysis, we used analogous linkers that
did not contain a modulator.
log kobsd = log(kH+[H3O+] + kH2O
)
(1)
There is an uncatalyzed release of pNPhOH at ∼ pH 4.5 to 9.3
at 37° (Table 1; Figure 1) as observed in studies of related O-Me
carbamates at pH 4.6 to 8.8,4 and a hydronium ion catalyzed rate
at lower pH values. The uncatalyzed, spontaneous rates are
suppressed by electron withdrawing p-substituents of the N-aryl
moiety of the carbamate (−H > −CON(Et)2 > −SO2N(CH2)2O)
and related to the pKa of the arylamine component by eq 2.
log k(h−1) = 0.29pKa − 3.95
(R2 = 0.990)
(2)
With 17A, the solvolysis rate shows a significant temperature
effect, being about 36-fold slower at 37 °C than that reported for the
analogous O-Me carbamate at 46 °C.4 The data are in accordance
with a mechanism that proceeds by pH-independent, spontaneous
loss of pNPhOH with involvement of electrons of the carbamate
nitrogen to generate an iminium ion intermediate (Scheme 2, path A).
It has previously been shown that log k of the SN1 reaction shows an
excellent correlation with the pKa (slope = −0.93) of the departing
phenol.4 If we assume the effects of the stabilizer and leaving group
abilities are independent of each other, bilinear eq 3 that uses the two
linear free energy relationships for these reactions (above) allows
estimation of the spontaneous solvolysis rate of any phenolic leaving
group with the stabilizers described here.
Azidoalcohol 11 containing Mod=PhSO2− was converted to
its chloroformate by treatment with triphosgene, and then
reacted with aryl amines to give N-aryl carbamates 12A-C. The
carbamates were treated with TMSCl/CH2O to provide the
reactive N-chloromethyl carbamates 13A-C. Reaction of 13A-C
log k(h−1) = −0.935 × pKa(leaving group)
+ 0.29 × pKa(stabilizer) + 2.71
(3)
1992
dx.doi.org/10.1021/bc4002882 | Bioconjugate Chem. 2013, 24, 1990−1997

Bioconjugate Chemistry
Article
Scheme 4
Figure 1. pH-log kobsd profiles for solvolysis of N-aryl-N-(pNPhOCH2)carbamates: (A) stable linkers 17; (B) β-eliminative linkers with PhSO2-
modulator 15A (red), B (blue) and C (green).
Rate constants for the hydronium ion catalyzed cleavage of
stable PEG40 kDa-NO2PhOCH2-carbamates (17A-C) are pro-
vided in Table 1. As with the spontaneous solvolysis at neutral
pH, electron withdrawing substituents on the carbamate
nitrogen retarded the solvolysis and were related to the pKa of
the arylamines by eq 4
rapid collapse to the iminium ion (Scheme 5, path A). However,
the carbonyl oxygen of a carbamate is the most basic site for
protonation,13 with an estimated pKa in analogous amides of
∼ −1. Scheme 5, path B, shows a possible mechanism involving
initial protonation of the carbonyl group of the carbamate (1)
and formation of a transient sp3 hybridized tetrahedral
intermediate (20), analogous to that which occurs in acid-
catalyzed amide hydrolysis.14 In amide hydrolysis, it is well
established that protonation of the amine in the tetrahedral
intermediate is required for its departure and that amine basicity
plays a profound role in collapse of the intermediate to the
carboxylic acid. The pKa of the amine in a tetrahedral
intermediate is reduced some 3- to 5-units compared to its free
state,15 and in the case of 20, would be lowered by another ∼2
units by the N-aryloxy substituent.16 This could effectively
log kH+(M−1 h−1) = 0.315pKa − 0.42
(R2 = 0.997)
(4)
which is essentially identical to that of the uncatalyzed reaction.
The acid-catalyzed mechanism of solvolysis must involve pre-
equilibrium protonation, but it is difficult to precisely place the
position of protonation because of several possible kinetically
equivalent mechanisms. The simplest mechanism is rationalized
as protonation of the leaving pNPhOH ether oxygen followed by
1993
dx.doi.org/10.1021/bc4002882 | Bioconjugate Chem. 2013, 24, 1990−1997

Bioconjugate Chemistry
Article
Table 1. Rates of pNPhOH Release from PEG-Conjugates
N-aryl group
Ph-
Mod
kH3+0 M−1 h−1
kH20, h−1 × 103
kOH_, M−1 h−1 × 10−3
t1/2 h, pH 7.4
15A
15B
15C
17A
17B
17C
PhSO2-
PhSO2-
PhSO2-
none
3.5 0.70
1.1 0.23
0.25 0.08
11.0 2.9
4.9 1.4
ND
334 44
404 82
572 63
8.3 1.1
6.8 1.4
4.8 0.5
311 42
536 69
2214 636
Et2NCOPh-
O(CH2)2NSO2Ph-
Ph-
ND
ND
2.23 0.31
1.29 0.17
0.31 0.09
Et2NCOPh-
O(CH2)2NSO2Ph-
none
none
1.5 0.07
Scheme 5
prevent amine protonation necessary for its departure, but
electrons of the tetrahedral nitrogen might still facilitate iminium
ion formation with concomitant elimination of pNPOH. Finally,
the protonated carbamate 19 might undergo SN2 displacement
to form O-hydroxymethyl-pNPOH, in a manner analogous to
that proposed for hydrolysis of N-acylimines.17 However, the
detection of a N-hydroxymethyl-carbamate intermediate in a
related work argues against this mechanism.12
the more acidic carbamic acid leaving groups; the rates follow the
equation log kOH− = −0.075 pKa + 0.53 (R2 = 0.972). Notably,
although the rate dependence on pKa of the aryl amine is small, it
opposes the retarding effect observed in spontaneous hydrolysis.
The base-catalyzed rate of β-elimination is suppressed only
2.5-fold by a phenol having poorer leaving group ability since an
O-alkylated tyrosine analog of 14B (24) showed t1/2 = 17 h at
pH 7.4 (SI).
At pH 4.5, near the nadir of the pH-log k profile (∼pH 3.8), the
observed rate of hydrolysis of 15C (0.23 × 10−4 h−1) is ∼15-fold
slower than the spontaneous hydrolysis rate of 17C containing a
stable linker (3.4 × 10−4 h−1) and a fit of the data in Figure 1B
using the latter rate constant for kO was poor. Thus, the stable
linker is not a perfect surrogate for the spontaneous hydrolysis
rate of 15C, containing Mod=PhSO2−.
Solvolysis Kinetics and Mechanism of N-aryl-N-
pNPhOCH2-carbamates with β-Eliminative Linkers (15A-C).
The pH-log kobsd profiles for solvolysis of PEG40 kDa
-
NO2PhOCH2-carbamates containing β-eliminative linkers
15A-C (Figure 1) are best described by eq 5.
log kobsd = log(kH+[H3O+] + ko + kOH−[OH−])
(5)
The rates of base catalyzed β-eliminative cleavage of the N-
aryloxy-N-aryl carbamates 15A-C are some ∼3- to 6-fold faster
than analogous carbamates of primary aliphatic amines.3 This is
attributed the lower pKa of N-carbamic acids of N-aryl vs primary
amines and hence greater leaving group ability of the N-aryl
carbamates. The good leaving group ability of the N-aryl
carbamates are caused by two effects: (a) electron withdrawing
substituents on the N-aryl group decrease the pKa of carbamic
acids,8 manifested here by rate differences of up to 1.7-fold for
β-elimination of reactants with differing stabilizers, and (b) even
lower pKa values expected for carbamic acids of Mannich bases10
of and N-hydroxymethylated16 aryl amines, which are ∼2 units
lower than those of the parent aryl amine; N-aryloxymethylation
should cause a similar pKa depression and consequent increase in
carbamate leaving group ability. Indeed, β-elimination of the N-
aryloxymethyl carbamate 14B, containing a N-aryloxy group,
solvolyzes some 4-fold faster than the azido-linker-N-arylcarba-
mate 12B (t1/2 = 51 h, pH 7.4), not containing a N-aryloxy group.
Detection of Intermediates. We desired to determine
The rate of release of pNPhOH was first order in hydronium
ion from pH 0 to ∼4 at 37°. The rate constants (Table 1) are 3- to
6-fold slower than those of the stable carbamates, so there is a
modest inhibition of the rate by the modulator. Electron
withdrawing substituents on the carbamate nitrogen retarded the
solvolysis and were related to the pKa of the N-arylamines by the
eq 6 which is very similar to that described for the stable linker.
Thus, the acid-catalyzed mechanisms of 15A-C solvolysis are
concluded to also occur by the mechanism depicted in Scheme 2,
path B
log kH+(M−1 h−1) = 0.358pKa − 1.16
(R2 = 0.997)
(6)
As with analogous primary carbamates,3 PEG40 kDa
-
NO2PhOCH2 N-aryl carbamates with Mod=PhSO2−, 15A-C,
showed release rates of pNPhOH that were first order in
hydroxide ion between pH ∼4 to 11.7 (Table 1). Thus, the base-
catalyzed reaction involves β-eliminative cleavage of the O-alkyl
bond of the carbamate to form a carbamic acid that collapses to
CO2, CH2O, the aromatic amine, and pNPhOH (Scheme 3).
The rates of β-elimination were only slightly accelerated by
electron withdrawing N-substituents, which can be attributed to
+
whether decarboxylation of the carbamic acid 8 (k2[H3 O],
Scheme 3) or collapse of the Mannich base intermediate 9
(k3, Scheme 3) might be slow enough to allow their accumulation
at physiologic pH. Several indirect lines of evidence indicated this
1994
dx.doi.org/10.1021/bc4002882 | Bioconjugate Chem. 2013, 24, 1990−1997

Bioconjugate Chemistry
Article
was likely not the case. First, the rates of decarboxylation of
primary aryl amine carbamates are rapid at physiological pH, with
t
DISCUSSION
■
In the present work, we adapted β-eliminative linkers originally
designed to deliver amine-containing drugs from carbamate-
containing linkers so they can be used with phenol-containing
drugs. The functional components of the adapted linkers include
(a) an electron-withdrawing modulator that, as before,3 governs
the rate of β-elimination and drug release, (b) a methylene
adaptor that connects the carbamate nitrogen to the phenol
oxygen of the drug, and (c) an electron-withdrawing stabilizer
attached to the carbamate nitrogen that inhibits undesirable
spontaneous cleavage of the adaptor−oxygen bond. Upon
β-elimination, a carboxylated Mannich base of the drug is
released that collapses to provide the free phenol-containing
drug.
1/2 values of several minutes for the slowest.8 Second, we did not
detect intermediates in any of the aforementioned studies using
HPLC − although we recognize the chromatography support
could have induced instabilities. Third, as reported for primary
aryl amine carbamates,8 we attempted to trap a stable carbamate
by allowing its formation from azido-linker carbamates 14A, B
in strong base, and then monitoring for spectral changes
accompanying decarboxylation upon neutralization. However,
upon dissolution of 14B at pH 12, pNPhOH was released as
rapidly as we could perform a measurement, and no further
spectral changes were detected upon neutralization. These
results, albeit indirect, suggest that both decarboxylation and
Mannich base decomposition occurred very rapidly at basic pH.
We next attempted to detect an intermediate that follows
β-elimination by a kinetic approach, following the release of
pNPhO− at 405 nm that accompanies cleavage of the ether bond
at basic pH and is immune to potential isolation artifacts. The
rate of pNPhOH formation in the three sequential irreversible
reactions depicted in Scheme 3 is expressed by eq 7. The
differential equation was derived by reported approaches18
(see SI) and simulated at different pH values using MS Excel.
Here, the F terms represent amplitude coefficients for each of three
exponentials representing each rate constant (see SI). Over the pH
range where the hydroxide-catalyzed β-elimination is slow
compared to the others F1 approaches 1 and the other coefficients
approach 0; the expression becomes eq 8, and the pH-log rate
profile has a slope of +1. If, at higher pH, the hydroxide-catalyzed
rate becomes faster than a subsequent reaction, the rate will be
governed by the decarboxylation reaction (F2 × e−k2[H+] t) or pH
independent Mannich base decomposition (F3 × e−k3t). In such
cases, as the F associated with an exponential approaches 1 and the
others approach 0, the pH-log rate profile will have a slope of −1
(eq 9) if the decarboxylation becomes rate determining, or 0 if the
Mannich base decomposition becomes rate determining (eq 10).
Thus, the rate obtained at the highest pH where pNPhOH release
remains first order in hydroxide allows assignment of a lower limit
on the rate of reaction of either intermediate at that pH. Since the
rate of reaction either remains the same (rate-determining
Mannich base collapse10) or decreases (rate-determining decar-
boxylation8) at lower pH, this rate also corresponds to a lower limit
of the overall rate at pH 7.4.
The synthesis follows that reported for primary carbamates4 to
give aryl amino carbamates 12; intermediate 12 contains an azido
group for coupling to a macromolecule, a modulator, and a carba-
mate of an electron withdrawing aryl amine. N-Chloromethylation
of 12 provided 13, which readily alkylates the oxygen of a phenol in
the presence of base. After purification, the azido-linker-phenol
conjugate 14 is coupled by Cu-free click chemistry to PEG modified
with a cyclooctyne to give the bioconjugate 15. For stable analogs
not containing a modulator (17), analogous synthetic sequences
were performed starting from an appropriate O-(6-azidohexyl)-N-
aryl carbamate. As indicated by previous studies,4 the synthetic
approach should also yield conjugates of drugs containing other
acidic heteroatoms and, indeed, we have prepared conjugates of
alcohols and thiols by this approach.
Solvolysis of conjugates with stable linkers 17 at all pH values
and those containing a PhSO2− modulator 15 in acidic media
gave pNPhOH and the N-aryl carbamate as final products,
indicating initial cleavage of the methylene adaptor oxygen bond
and formation of an iminium ion intermediate. In contrast,
solvolysis of the conjugate containing a PhSO2− modulator at
pH > about 4 yielded pNPhOH, the alkenylsulfone and aryl
amine as products, indicating initial β-eliminative O-alkyl
cleavage of the carbamate bond.
Previous and current kinetic studies of the solvolysis covering a
wide pH range allow a more detailed analysis of the reaction
mechanisms. The kinetics for solvolysis of the β-eliminative
conjugates is described by eq 11.
kobs = kH+[H3O+] + kH2O + kOH−[OH−]
(11)
dP/dt = [1 − F × e−k1[OH−]t − F × e−k2[H+]t
The first term of eq 11, kH+[H3O+], is the specific acid
catalyzed reaction involving protonation of the reactant followed
by release of pNPhOH and the N-aryl carbamate. The rates are
lowered by electron withdrawing groups on the aryl carbamate
nitrogen that destabilize the iminium ion intermediate, and show
a linear relationship with the pKa of the parent aryl amine. The
results are in accordance with equilibrium protonation of the
leaving group ether-oxygen, followed by rate determining
cleavage to pNPOH and the unstable carbamate iminium ion
(Scheme 5); the iminium ion intermediate would rapidly hydrate
to the N-hydroxymethyl carbamate which would rapidly lose
formaldehyde in acidic media.6
The second term of eq 11, kH2O, is the spontaneous
uncatalyzed SN1 counterpart of the acid catalyzed reaction that
also gives pNPhOH and the N-aryl carbamate as products
(Scheme 2). The exact contribution of this pathway cannot be
ascertained with linkers containing a pKa modulator that causes a
competing, more rapid β-elimination reaction. Thus, it was
studied with surrogate stable linkers having varying N-aryl
1
2
− F × e−k3t
]
(7)
(8)
3
dP/dt = 1 − e−k1[OH−]t
dP/dt = 1 − e−k2[H+]t
dP/dt = 1 − e−k3t
(9)
(10)
The log kobsd vs pH profile for pNPhOH formation from 15B,
Mod=PhSO2−, shows a slope of +1 from pH 7.4 to at least
pH 11.7, the highest pH we could conveniently monitor the reaction.
Here, t1/2 ∼5 s so neither intermediate has a t1/2 > 5 s at that pH. At
lower pHs, the acid-catalyzed decarboxylation would be even more
rapid,8 and the rate of Mannich base breakdown would be pH-
independent until the aryl amine is protonated at pH < 5.10 Thus, we
conclude that both intermediates decompose rapidly and do not
accumulate to any significant extent at pH 7.4.
1995
dx.doi.org/10.1021/bc4002882 | Bioconjugate Chem. 2013, 24, 1990−1997

Bioconjugate Chemistry
Article
stabilizers (17A-C). As with the acid catalyzed reaction, the rates
are retarded by electron withdrawing groups that destabilize the
iminium ion intermediate, and show a linear inverse relationship
with the pKa of the aryl amine component that is nearly identical
to that in the acid-catalyzed reaction. Majumdar and Sloan have
also reported that the SN1 reaction shows an excellent correlation
with the pKa (slope = −0.93) of the departing phenol.4
lifetime of the Mannich base. Thus, a key requirement for effective
use of these linkers is an appropriate balance between the leaving
group ability of the drug in the context of the modulator used (i.e.,
t
1/2 for β-elimination at pH 7.4). Since this balance may vary with
each type of leaving group, it may be necessary to determine the
structure activity relationships of the leaving group abilities and SN1
rates on the stabilizer for each drug functional group used in such
linkers.
Linker design should also incorporate considerations of the
need for long-term storage of drug-containing bioconjugates.
Since the hydroxide-catalyzed β-elimination rate is depressed
10-fold for each pH unit decrease, and another 30-fold for each
10 °C reduction of temperature,3 linkers can be stabilized for
long periods by simply lowering the pH and temperature.
However, the stability afforded by lowering pH is limited to the
extent where such stabilization exceeds the spontaneous SN1 or
acid catalyzed reaction rates.
The third term of eq 11, kOH−[OH−], is the specific base
catalyzed β-eliminative cleavage driven by the electron with-
drawing effect of the modulator. The initial step is the cleavage of
the O-alkyl bond of the carbamate to give a carboxylated
Mannich base which undergoes decarboxylation and subsequent
collapse8 of the Mannich base to the aryl amine, formaldehyde,
and free drug (Scheme 3). Using the PhSO2− modulator and
N-aryl stabilizers, β-elimination reactions occur at physiological
pH about 3- to 6-fold faster than analogous primary carbamates;3
this can be ascribed to the lower pKa of the carbamic acid leaving
groups by the electron withdrawing aryl amine enhanced further
by the N-aryloxy group.10,16 In contrast to the acid-catalyzed and
SN1 solvolysis, the β-eliminative rates show only a small
dependence on the pKa of the aryl amine stabilizer.
CONCLUSION
■
We have described an approach to adapt β-eliminative linkers for
carbamate-linked amino groups of drugs so they can also be used
to attach phenolic hydroxyl groups of drugs. Kinetic studies of
solvolysis defined the mechanistic pathways throughout the pH
range, and established three key elements for controlled release.
These elements are: (a) the user-defined modulator that controls
the rate of base-catalyzed β-eliminative drug release, (b) the
leaving group ability of the drug component, and (c) the
carbamate stabilizer that controls the rate of spontaneous SN1
solvolysis. In designing bioconjugates using the approach
described here, it is essential to understand and appropriately
balance these effects to achieve optimized control of drug release.
Once formed in vivo, the two intermediates of the β-elimination
reaction − the carboxylated Mannich base and the Mannich base −
should ideally convert to products at a rapid rate; otherwise, they
might accumulate and distribute or clear with different pharma-
cokinetics than the native drug. Studies8 show that the rate of
decarboxylation of N-aryl carbamic acids is sufficiently rapid that
such intermediates should not accumulate to any significant extent
at neutral pH. Indeed, in the present case, we could estimate that the
carbamic acid 8 decarboxylated with a t1/2 <5 s at physiologic pH. In
the only direct mechanistic study of cleavage of N-aryl Mannich
bases we are aware of, Bundgaard and Johansen10 showed that the
rates of collapse are pH independent above their pKa, and inhibited
when protonated. Rates of breakdown of N-aryl Mannich bases of
imides are depressed by reducing the pKa of the aryl amine, with a
9.3-fold change per pKa unit. The rates are likewise decreased by
increasing the pKa of the leaving imide group, with a change of about
13-fold per pKa unit. With conjugate 15B, pNPhOH is a sufficiently
good leaving group (pKa 7.14) that, based on these studies, we
would not expect Mannich base accumulation with any of the
stabilizers used; indeed, using a kinetic approach we could estimate
that the Mannich base formed from 15B also has a t1/2 of <5 s at
pH 7.4.
ASSOCIATED CONTENT
■
S
* Supporting Information
Synthesis of intermediates and PEG conjugates, kinetic
procedures and derivation of equations. This material is available
free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: Daniel.V.Santi@prolynxllc.com. Phone: 415 552 5306.
Notes
In designing linkers for in vivo use, it is important that the
stabilizer is sufficiently electron withdrawing to depress the rate
of the SN1 reaction such that it does not compete with β-elimination −
i.e., kOH(OH−) ≫ kH2O. For example, O-Me-N-methyl-N-
pNPhOCH2-carbamate hydrolyzes some 24-fold faster than the
corresponding N-phenyl carbamate,4 which from our data (Table 1)
gives the SN1 reaction of the N-alkyl carbamate a t1/2 of about 13 h at
neutral pH; thus, with a N-Me stabilizer and the PhSO2− modulator
used here (t1/2 ∼8.3 h at pH 7.4) only ∼40% of the product
would emanate from β-elimination. Likewise, with the N-Ph
modulator 15A, the SN1 rate is only about 37-fold slower than the
β-elimination reaction at pH 7.4 (Table 1), which would compete
with β-elimination in linkers with longer half-lives. On the other
hand, the modulator should not be so electron withdrawing that it
unduly stabilizes the Mannich base intermediate.10 This is not an
issue in the present work, but it could become an important
consideration with drug leaving groups that are so poor they stabilize
the Mannich base intermediate. However, poorer leaving groups
also slow the SN1 reaction,4 so a less electron-withdrawing stabilizer
could be used for SN1 stabilization and concomitantly decrease the
The authors declare no competing financial interest.
REFERENCES
■
(1) Greenwald, R. B., Yang, K., Zhao, H., Conover, C. D., Lee, S., and
Filpula, D. (2003) Controlled release of proteins from their poly-
(ethylene glycol) conjugates: drug delivery systems employing 1,6-
elimination. Bioconjugate Chem. 14, 395−403.
(2) Filpula, D., and Zhao, H. (2008) Releasable PEGylation of proteins
with customized linkers. Adv. Drug. Delivery Rev. 60, 29−49.
(3) Santi, D. V., Schneider, E. L., Reid, R., Robinson, L., and Ashley, G.
W. (2012) Predictable and tunable half-life extension of therapeutic
agents by controlled chemical release from macromolecular conjugates.
Proc. Natl. Acad. Sci. U.S.A. 109, 6211−6.
(4) Majumdar, S., and Sloan, K. B. (2006) Synthesis, hydrolyses and
dermal delivery of N-alkyl-N-alkyloxycarbonylaminomethyl (NANAO-
CAM) derivatives of phenol, imide and thiol containing drugs. Bioorg.
Med. Chem. Lett. 16, 3590−4.
(5) Bundgaard, H., and Johansen, M. (1984) Hydrolysis of N-(α-
hydroxybenzyl) benzamide and other N-(α-hydroxyalkyl) amide
derivatives: implications for the design of N-acyloxyalkyl-type prodrugs.
Int. J. Pharm. 22, 45−56.
1996
dx.doi.org/10.1021/bc4002882 | Bioconjugate Chem. 2013, 24, 1990−1997

Bioconjugate Chemistry
Article
(6) Tenn, W. J., III, French, N. L., and Nagorski, R. W. (2001) Kinetic
dependence of the aqueous reaction of N-(hydroxymethyl)benzamide
derivatives upon addition of electron-withdrawing groups. Org. Lett. 3,
75−78.
(7) Iley, J., Moreira, R., Calheiros, T., and Mendes, E. (1997)
Acyloxymethyl as a drug protecting group: Part 4. The hydrolysis of
tertiary amidomethyl ester prodrugs of carboxylic acid agents. Pharm.
Res. 14, 1634−9.
(8) Johnson, S. L., and Morrison, D. L. (1972) Kinetics and mechanism
of decarboxylation of N-arylcarbamates. Evidence for kinetically
important zwitterionic carbamic acid species of short lifetime. J. Am.
Chem. Soc. 94, 1323−34.
(9) Ewing, S. P., Lockshon, D., and Jencks, W. P. (1980) Mechanism of
cleavage of carbamate anions. J. Am. Chem. Soc. 102, 3072−3084.
(10) Bundgaard, H., and Johansen, M. (1981) Prodrugs as drug
delivery systems. XIX. Bioreversible derivatization of aromatic amines by
formation of N-Mannich bases with succinimide. Int. J. Pharm. 8, 183−
192.
(11) Debets, M. F., van Berkel, S. S., Schoffelen, S., Rutjes, F. P., van
Hest, J. C., and van Delft, F. L. (2010) Aza-dibenzocyclooctynes for fast
and efficient enzyme PEGylation via copper-free (3 + 2) cycloaddition.
Chem. Commun. 46, 97−9.
(12) Santi, D. V., Schneider, E., and Ashley, G. A. (Unpublished
Results).
(13) Olah, G. A., and Calin, M. (1968) Stable carbonium ions. LIX.
Protonated alkyl carbamates and their cleavage to protonated carbamic
acids and alkylcarbonium ions. J. Am. Chem. Soc. 90, 401−4.
(14) Brown, R. S., Benneta, A. J., and Slebocka-tilk, N. (1992) Recent
perspectives concerning the mechanism of H3O+- and OH−-promoted
amide hydrolysis. Acc. Chem. Res. 25, 481−488.
(15) Guthrie, J. P. (1974) Hydration of carboxamides. evaluation of the
free energy change for addition of water to acetamide and formamide
derivatives. J. Am. Chem. Soc. 96, 3608−3615.
(16) Abrams, W. R., and Kallen, R. G. (1976) Equilibria and kinetics of
N-hydroxymethylamine formation from aromatic exocyclic amines and
formaldehyde. effects of nucleophilicity and catalyst strength upon
mechanisms of catalysis of carbinolamine formation. J. Am. Chem. Soc.
98, 7777−7789.
(17) Sayer, J. M., and Conlon, P. (1980) The timing of the proton-
transfer process in carbonyl additions and related reactions. general-
acid-catalyzed hydrolysis of imines and N-acylimines of benzophenone.
J. Am. Chem. Soc. 102, 3592−3600.
(18) Korobov, V. I., and Ochkov, V. F. (2011) Multi-Step Reactions:
The Methods for Analytical Solving the Direct Problem, in Chemical
Kinetics with Mathcad and Maple,pp 35−72, Springer-Verlag, Vienna.
1997
dx.doi.org/10.1021/bc4002882 | Bioconjugate Chem. 2013, 24, 1990−1997