A Self-Immolative Linker for the pH-Responsive Release of Amides

The administration of therapeutics using bioconjugation has been mainly limited to drugs containing amine, alcohol, or thiol functional groups. Here, we report a general procedure for the preparation of benzylic N-acyl carbamates suitable for masking the amide group in important drugs such as Linezolid, Enzalutamide, or Tasimelteon in good to acceptable yields. These N-acyl carbamates appear to be stable in plasma, while a qualitative analysis of further drug uncage demonstrates that, at pH values of 5.5, a classical 1,6-benzyl elimination mechanism takes place, releasing more than 80% of the drug in 24 h.


Introduction
Bioconjugation has recently attracted remarkable interest thanks to its therapeutic potential in chemotherapy [1 -3]. As selectivity is one of the major problems in the use of cytotoxic drugs, bioconjugation provides a powerful tool for smart drug delivery [4][5][6][7]. The active pharmaceutical ingredient can be covalently bound to selective carriers such as small molecule ligands, peptides, proteins, or antibodies. Once the bioconjugate has reached the target, the drug can be triggered by an external stimulus. The linker is a key component of the bioconjugate, as it not only separates the carrier from the drug, but also controls stability in the bloodstream and release in the target tissue [1, 6,8,9]. Two main families of linkers are used in bioconjugation. The so-called "non-cleavable linker" [10] acts through the lysosomal proteolytic degradation of the conjugate, while the "cleavable linker" [11] relies on endogenous stimuli such as enzyme activity, pH, high ROS, or glutathione, which activate a self-immolative degradation that ends with the release of the intact drug [8]. However, this self-immolative process relies on molecules with high nucleofugacity, which limits the release to agents containing amines, alcohols, phenols, or thiols (Scheme 1, reaction a) [8,12]. Amides are widely distributed molecules in nature, and several approved drugs contain this functional group [13]. Despite their importance, there are very few bioreversible prodrugs or self-immolative linkers suitable for the targeted release of amides [14][15][16][17][18]. This is because of the low nucleophilicity of the amide nitrogen and the low acidity of the amide hydrogen, which prevent nucleofugacity.
Here, we decided to investigate the use of N-acyl carbamates to develop a new self-immolative linker for amides (Scheme 1, reaction b). The new prodrug is an N-acyl carbamate derived from p-hydroxybenzyl alcohol and protected on the phenol side with an acid-labile group. As pH variations between healthy and cancerous or inflamed tissues have been used to produce pH-sensitive materials [19][20][21][22][23][24], we expected the drug to be released by acid-mediated degradation. Acyl carbamates have been described as potential prodrugs for amides since the 1990s [15]. Although they are relatively stable at pH between 4 and 6 and appear to be stable in plasma for several hours [25], no significant applications have been described to date [26].

Results
We developed our linker by adding acid-labile protecting groups to the phenolic side of the p-hydroxybenzyl alcohol, while the benzyl alcohol served to install the amidecontaining drug over the acyl carbamate. In addition, a terminal alkyne was added to the aromatic ring to allow coupling with the carrier (Scheme 2). Linezolid (7 in Scheme 2) was used as a model drug to optimise linker design and synthesis. It is an antibiotic used to treat infections caused by Gram-positive resistant bacteria and contains only a secondary amide as a functional group suitable for conjugation [27]. 3,4-Dihydroxybenzaldehyde 1 was selectively protected at position 4 as methoxymethyl ether (MOM, 2a in Scheme 2) or methoxyethoxymethyl ether (MEM, 2b in Scheme 2) in acceptable yields. The other phenol group was alkylated with propargyl bromide to introduce a counterpart suitable for further conjugation by click chemistry (3a,b in Scheme 2). Reduction of the aldehyde gave the benzyl alcohols 4a,b, the starting materials for N-acyl carbamate synthesis. N-Acyl carbamates have been prepared mainly by addition of an alcohol to an acyl isocyanate obtained by treatment of a primary amide with oxalyl chloride [17,18]. Alternatively, displacement of an activated carbonate by a primary amide has been described [15]. However, these methods are not suitable for the preparation of an acyl carbamate prodrug of Linezolid such as 8a or 8b in Scheme 2. We investigated the possibility of activating compounds 4a,b with p-NO2-phenylchloroformic acid ester to obtain carbonates 5a,b. These compounds were relatively stable molecules and could be purified by flash chromatography. Compound 5a was used to study the direct oxycarbonylation of Linezolid 7 or the indirect acylation of amine 6 followed by acetylation to give the Linezolid adduct 8a. Treatment of 7 with 5a in the presence of different bases and different solvents was unsuccessful and gave only small amounts of 8a (entries 1-4 in Table 1). Better results were obtained by carbamoylation of amine 6, which was carried out in good yields (entry 5 in Table 1). Unfortunately, acetylation of the resulting carbamate with Ac2O, DMAP, and Et3N gave compound 8a in a low yield (entry 5 in Table 1). Further attempts to increase the yield by changing the base were unsuccessful.
The acyl carbamate 8a was finally obtained in a good yield (88%) by deprotonation of the amide NH with KHMDS in THF at −78 °C, followed by reaction with carbonate 5a in THF at a controlled temperature (entry 6 in Table 1). The same procedure was applied to 5b, giving 8b in a 59% isolated yield (Scheme 2). In the latter case, the low yield was due to the fact that the carbonate 8b tends to degrade after 5/6 h under alkaline conditions. Consequently, the unreacted amide 7 was recovered at the end of the reaction. Acylation of an amide with benzyl carbonate 5a in the presence of KHMDS was also used with Scheme 1. (a) 1-6 self immolative spacer for amines or alcohols (ref [3]). (b) pH-responsive selfimmolative spacer for amides as Linezolid, Enzalutamide, and Tasimelteon, or sulfonamides.

Results
We developed our linker by adding acid-labile protecting groups to the phenolic side of the p-hydroxybenzyl alcohol, while the benzyl alcohol served to install the amide-containing drug over the acyl carbamate. In addition, a terminal alkyne was added to the aromatic ring to allow coupling with the carrier (Scheme 2). Linezolid (7 in Scheme 2) was used as a model drug to optimise linker design and synthesis. It is an antibiotic used to treat infections caused by Gram-positive resistant bacteria and contains only a secondary amide as a functional group suitable for conjugation [27]. 3,4-Dihydroxybenzaldehyde 1 was selectively protected at position 4 as methoxymethyl ether (MOM, 2a in Scheme 2) or methoxyethoxymethyl ether (MEM, 2b in Scheme 2) in acceptable yields. The other phenol group was alkylated with propargyl bromide to introduce a counterpart suitable for further conjugation by click chemistry (3a,b in Scheme 2). Reduction of the aldehyde gave the benzyl alcohols 4a,b, the starting materials for N-acyl carbamate synthesis. N-Acyl carbamates have been prepared mainly by addition of an alcohol to an acyl isocyanate obtained by treatment of a primary amide with oxalyl chloride [17,18]. Alternatively, displacement of an activated carbonate by a primary amide has been described [15]. However, these methods are not suitable for the preparation of an acyl carbamate prodrug of Linezolid such as 8a or 8b in Scheme 2. We investigated the possibility of activating compounds 4a,b with p-NO 2 -phenylchloroformic acid ester to obtain carbonates 5a,b. These compounds were relatively stable molecules and could be purified by flash chromatography. Compound 5a was used to study the direct oxycarbonylation of Linezolid 7 or the indirect acylation of amine 6 followed by acetylation to give the Linezolid adduct 8a. Treatment of 7 with 5a in the presence of different bases and different solvents was unsuccessful and gave only small amounts of 8a (entries 1-4 in Table 1). Better results were obtained by carbamoylation of amine 6, which was carried out in good yields (entry 5 in Table 1). Unfortunately, acetylation of the resulting carbamate with Ac 2 O, DMAP, and Et3N gave compound 8a in a low yield (entry 5 in Table 1). Further attempts to increase the yield by changing the base were unsuccessful.
The acyl carbamate 8a was finally obtained in a good yield (88%) by deprotonation of the amide NH with KHMDS in THF at −78 • C, followed by reaction with carbonate 5a in THF at a controlled temperature (entry 6 in Table 1). The same procedure was applied to 5b, giving 8b in a 59% isolated yield (Scheme 2). In the latter case, the low yield was due to the fact that the carbonate 8b tends to degrade after 5/6 h under alkaline conditions. Consequently, the unreacted amide 7 was recovered at the end of the reaction. Acylation of an amide with benzyl carbonate 5a in the presence of KHMDS was also used with Enzalutamide 9 (Scheme 2), a non-steroidal antiandrogen approved for the treatment of prostate cancer [28].
Enzalutamide 9 (Scheme 2), a non-steroidal antiandrogen approved for the treatment of prostate cancer [28].  Again, the amide is the only functional group in the molecule available for bioconjugation and further release of the drug under an endogenous stimulus. Following the general procedure developed for 8, Enzalutamide pro-drug 10 was obtained in 64% isolated yield (Scheme 2). Products 8a,b and 10 were tested for stability in solution and were stable for 48 h at a pH of 7.4 (PBS solution). Unfortunately, at a pH of 5.5, only a Scheme 2. Preparation of acid responsive linker and link with Linezolid and Enzalutamide. 2a-5a, 8a X = MOM; 2b-5b, 8b X = MEM. Again, the amide is the only functional group in the molecule available for bioconjugation and further release of the drug under an endogenous stimulus. Following the general procedure developed for 8, Enzalutamide pro-drug 10 was obtained in 64% isolated yield (Scheme 2). Products 8a,b and 10 were tested for stability in solution and were stable for 48 h at a pH of 7.4 (PBS solution). Unfortunately, at a pH of 5.5, only a small amount of amide 9 was released after 48 h at 37 • C. A small improvement was achieved at a pH of 4.5, where the peak intensities for 8a,b and 10 decreased, but only the presence of free Enzalutamide 9 (40%) could be clearly detected. These data show that it is possible to benzoxylate the potassium salt of a (secondary) amide with a p-nitrophenyl carbonate such as 5a,b to give an acyl carbamate stable in PBS solution. However, the use of MEM or MOM to protect the phenolic groups, already used in pro-drug applications [29], prevents the use of these conjugates for self-immolative systems triggered by physiological small pH fluctuations [30]. Therefore, we decided to form the activated carbonate from benzyl alcohol 11 (Scheme 3). This linker, which is based on orthoester chemistry, was derived from gallic acid and is known to undergo the uncage process at a pH of 5.5 [31]. Carbonylation of 11 was carried out as previously described for compound 4, and the activated carbonate 12 was used for the acylation of Linezolid 7, Enzalutamide 9, and Tasimelteon 13, a drug developed for sleep disorders [32]. p-Toluenesulfonamide 14 was also used as a model compound for sulfonamidecontaining drugs. Amides 7, 9, and 13 were chosen as representative of approved drugs with different therapeutic applications where the NH amide is the only nucleophile for linkage. To the best of our knowledge, there are no examples in the literature of prodrugs or bioconjugates that release these drugs.  Following the synthetic approach described in Scheme 3, N-acyl carbamates 15-18 were obtained in good to acceptable yields.
The low yield of compound 17 (44%) was due to the degradation of Tasimelteon during the reaction, while the low yield of compound 18 (21%) was due to the sluggish reactivity characterised by the recovery of unreacted starting material and some non-separable mixed fractions after column chromatography.
Qualitative analysis showed that products 15-18 appear stable in physiological fluids and proved stable in PBS and human plasma, as indicated in Table 2. All products were stable in water, while we observed some decrease in product peaks after 24 h in PBS and in plasma ( Table 2 columns 3 and 4). However, the presence of the free drugs was never detected. A qualitative analysis of the hydrolysis of compounds 15-18 was also performed and it was found that the release of the intact drug amides 7, 9, 13, and 14 occurred at pH 5.5 (Figure 1 and Supplementary Materials). In the case of the more acidic p-toluenesulfonamide derivative, release was faster, but after 48 h, no more than 80% had been released.

Conclusions
In summary, we have shown that the acid-sensitive benzyl alcohols 4a,b and 11 can be used to mask drugs containing an amide bond by forming the corresponding N-acyl carbamate, although they are restricted to low hindered amides. Orthogonal reactivity to this platform is possible by click chemistry to allow easy conjugation with macromolecular carriers. Our system appears stable in plasma, allowing potential application in targeted drug delivery. The intact drug amide is released after triggering a 1-6 elimination cascade in an acidic biological environment at a pH typical of inflamed tissue, or by lysosomal digestion in the case of application of the linker in ADC. Compounds 15-18 are the first successful examples of how a drug can be linked to a carrier with an amide to create a bioconjugate or pro-drug or nanomedical device that releases the drug once it reaches the target. Further improvements to extend this linker to more complex amides using a different synthetic approach and using other non-canonical chemotherapeutic agents triggered by enzymatic release processes are in preparation and will be announced shortly.

Materials and Methods
General experimental procedures, materials, and instruments are reported in SI.