Introduction of Para-Hydroxy Benzyl Spacer Greatly Expands the

Article Cite This: Bioconjugate Chem. 2019, 30, 1969−1978

pubs.acs.org/bc

Introduction of Para-Hydroxy Benzyl Spacer Greatly Expands the Utility of Ortho-Hydroxy-Protected Aryl Sulfate System: Application to Nonphenolic Payloads Suho Park,† Sun Young Kim,† Jongun Cho,† Doohwan Jung,† Donghoon Seo,† Jaeho Lee,† Sangkwang Lee,† Sanghyeon Yun,† Hyangsook Lee,‡ Okku Park,‡ Beomseok Seo,‡ Sena Kim,‡ Minah Seol,‡ Sung Ho Woo,‡ and Tae Kyo Park*,†,‡ Chemistry division and ‡Biology division, IntoCell, 101 Sinildong-ro, Daedeok-gu, Daejeon, 34324, Republic of Korea

Downloaded via UNIV OF SOUTHERN INDIANA on July 22, 2019 at 01:32:29 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

†

S Supporting Information *

ABSTRACT: The ortho-hydroxy-protected aryl sulfate (OHPAS) linker is composed of a diaryl sulfate backbone equipped with a latent phenol moiety at the ortho position of one of the aryl units. The Ar−OH released when the ortho phenol undergoes intramolecular cyclization and displaces the second aryl unit can be viewed as a payload. We have shown in the preceding paper that the OHPAS linkers are highly stable chemically and in various plasmas, yet release payloads when exposed to suitable triggering conditions. As an extension of the OHPAS system, we employed a para-hydroxy benzyl (PHB) spacer for coupling to nonphenolic payloads; this tactic again provided a highly stable system capable of smooth release of appended payloads. The PHB modification works beautifully for tertiary amine and N-heterocycle payloads.

■

INTRODUCTION In the preceding paper,1 we described a new self-immolative group for conjugating phenolic payloads. The chemistry that enables the new system is based on two key discoveries: diaryl sulfate is easily assembled from two different aryl alcohols (mainly via SuFEx chemistry) and is highly stable. Moreover, we showed that diaryl sulfates are stable in plasmas of various origins, a quality essential for exploitation as a therapeutically relevant SIG component. A reasonable rate of payload release at biological temperature was manifested upon revealing the ortho-phenol group temporarily blocked by the triggering group (TG). The generality of the new system has been shown by conjugates linking various combinations of a diverse range of TGs and payloads: β-galactoside, β-glucuronide, levulinate ester, and 2-nitrobenzyl ether are examples of TGs tested and α-amanitin, SN-38, Combretastatin A4, estradiol, seco-DUBA, dCBI, and dPBD are representative phenolic payloads tested. Met-enkephalin, an example of oligopeptide with tyrosine residue, has also been conjugated and released, demonstrating easy access to such conjugates with the OHPAS protocol. Since the OHPAS linker itself is particularly well-suited for phenolic payloads (Figure 1c), we wondered whether nonphenolic payloads could also be employed by conjugation via an intervening phenol-containing spacer. To that end, we investigated a para-hydroxy benzyl (PHB) spacer whose 1,6fragmentation mechanism might be operative in PHB moiety in analogy with the para-aminobenzyl (PAB) unit in the VCPABC system (compare Figure 1b with Figure 1d). Illustrated in Figure 1 are representative cleavage mechanisms for comparison.2 Payloads connected via a disulfide bond (Figure © 2019 American Chemical Society

Figure 1. Comparison of payload release mechanism.

1a), a technology developed by Immunogen and known as targeted antibody payload (TAP) technology, are known to be cleaved by thiol-exchange reactions. However, the intracellular location of such cleavage is not as selective as originally hoped (with possible toxicity concerns), even though the concentrations of thiols in plasma are much lower than inside cells. Moreover, the technology requires mercapto payload(s) (or at least payload(s) with a thiol-appended linker) that are rarely Received: May 14, 2019 Revised: June 17, 2019 Published: June 19, 2019 1969

DOI: 10.1021/acs.bioconjchem.9b00341 Bioconjugate Chem. 2019, 30, 1969−1978

Article

Bioconjugate Chemistry

Figure 2. Conjugation to amines and aliphatic alcohol.

Figure 3. Conjugation to aliphatic alcohol and carboxylic acid.

gave the desired products, sulfamate 4 and sulfate 5, respectively, after deacetylation. To our dismay, compounds 4 and 5, when triggered with βgalactosidase under standard conditions, gave very little, if any, payload release (2 and 3, respectively). In both cases, degalactosylation took place in minutes, but the intended intramolecular cyclizations of the degalactosylated intermediates were very sluggish even at elevated temperature (data not shown). However, this problem could handily be solved by insertion of the PHB moiety, a phenol-containing spacer (Figure 2). For example, spacer-linked compound 7 could easily be obtained from MMAF-OMe (a secondary amine) and p-fluorosulfonyloxy benzyl alcohol, which in turn was prepared from PHB alcohol and SO2F2 without protection of the benzylic alcohol function.5 Coupling of fluorosulfonate 7 and TBS ether 6 gave PHB-extended OHPAS-MMAF 8 after deesterification of acetyl (βGal) and methyl ester groups (MMAF-OMe). Under our standard conditions,6 the half-life of MMAF release was found to be 41 min, demonstrating the utility of our approach for amine-containing payloads as well. The degalactosylation was rapid (t1/2 < 1 min) as expected, and the intramolecular cyclization step was found to be ratelimiting, with t1/2 of 41 min; buildup of the intermediate p-

encountered. In contrast to the chemically cleaved disulfide technology, conjugates assembled via the OHPAS linker can be released enzymatically (Figure 1c). The VC-PABC system, originally developed for (secondary) amine payloads, releases payloads via protease-mediated hydrolysis followed by 1,6fragmentation of PABC (Figure 1b). This mechanism offers an apt analogy to our proposed addition of the PHB spacer to the original OHPAS system (Figure 1d). Described herein are experimental verifications of this concept and applications to real conjugates together with some in vitro/vivo results.

■

EFFORTS TO CONJUGATE AMINES AND ALIPHATIC ALCOHOLS USING OHPAS LINKER To explore the application of the OHPAS linker to payloads having amine or aliphatic alcohol function, we prepared two model compounds with secondary amine 23 and aliphatic alcohol 3 (N-acetyl norephedrine), each of which can be viewed as a subfragment of the MMAE structure (Figure 2). In these couplings, we employed in situ generated N-methyl imidazolium salt 1, which can be prepared from reaction of the corresponding phenol and sulfonyldiimidazole followed by Nmethylation with methyl triflate.4 Coupling of 2 and 3 with 1 1970

DOI: 10.1021/acs.bioconjchem.9b00341 Bioconjugate Chem. 2019, 30, 1969−1978

Article

Bioconjugate Chemistry Scheme 1. Multiple Drug Attachment Using a Hub

Reagents and condition: (a) 4M-LiBH4, THF, 0 °C to rt, 19 h, 36%; (b) 4-Nitrophenyl chloroformate, THF, 0 °C, 3 h, 72%; c) MMAF-OMe, HOBt, Pyridine, DIPEA, DMF, rt, 15 h, 54%; d) 6, BEMP, AcN, rt, 12 h, 68%; e) LiOH-H2O, MeOH, 0 °C, 2 h, 91%.

a

■

MULTIPLE PAYLOAD INCORPORATION: 2,4,6-TRIS(HYDROXYMETHYL)PHENOL AS A HUB If multiple payloads could be assembled in a single SIG motif, the utility of the OHPAS system would greatly be enhanced, in that the number and types of payload molecules could be tailored to varying situations and needs. Payloads having differing modes of actions (e.g., microtubule inhibitor and DNA binder) or even differing objectives (e.g., toxin-reporter combinations) could be attached to a single conjugate. Moreover, this approach greatly simplifies the synthetic route toward a system in which multiple molecules are released with a single triggering event, in contrast to releasing a single toxin per triggering event. To implement such a tactic, a hub moiety is needed. For this purpose, we selected 2,4,6-tris(hydroxymethyl)phenol 17a (Scheme 1).9 Unfortunately, when we tried to couple 17a with either 16a or 16b, the reaction did not give satisfactory results; similar trends were observed with tris-aldehyde 17b. Eventually we obtained a synthetically acceptable yield of 18a by coupling of 16b and 17c under alkylation conditions. The analogous reaction of 16b with aldehyde 17d, however, gave poor results. The coupled product 18a10 was reduced with LiBH4 to give trisalcohol 18b, which was activated with p-nitrophenyl chloroformate to give PNP carbonate 18c. The PNP carbonate was then coupled with MMAF-OMe to give 18d, a key intermediate with a triple payload. Fluorosulfonate 18d was reacted with TBS ether 6 in the presence of BEMP to give 19a. The same reaction under DBU catalysis results only in silyl

hydroxybenzyloxycarbonyl-MMAF (structure not shown) was not detected under our HPLC analysis conditions. para-Fluorosulfonyloxybenzyl (PFB) halide, a convenient derivative of PHB alcohol quite often employed, is an extremely useful spacer that can react with payloads having tertiary amines or N-heterocycles, such as pyridines. The resulting PFB-payload is then readily coupled with an aromatic silyl ether that can serve as a key SIG component to yield a new OHPAS system suitable for nonphenolic payloads. Substituted PFB halides can also be employed; PFB halide with 2,6-dimethoxy groups gave equal or better yield (∼10% higher) of the corresponding quaternary ammonium salt. In both cases, the quaternary ammonium salts derived from tertiary amines or N-heterocycles were highly stable chemically and in various plasmas (vide infra). Similarly, the OHPAS-PHB linker has been tested on molecules having aliphatic alcohol and carboxylic acid functions (Figure 3). A series of model compounds was prepared starting with β-Gal-TBS ether 9a. Coupling with fluorosulfonate compounds 10−12 gave the corresponding sulfates, which upon deacetylation (βGal) give the desired OHPAS-PHB-MAC 13,7,8 OHPAS-PHB-carbonate 14, and OHPAS-PHB-ester 15, respectively. In the enzymatic cleavage assay, the degalactosylation was fast (t1/2 < 1 min) and payload release was rapid (half-lives; 1-phenylethanol = 6.7 min, 2-(3pyridyl)ethanol = 37 min, 3-pyridylacetic acid = 15 min). 1971

DOI: 10.1021/acs.bioconjchem.9b00341 Bioconjugate Chem. 2019, 30, 1969−1978

Article

Bioconjugate Chemistry

Figure 4. Kinetics of enzymatic cleavage with OHPAS-PHB-MMAF3.

Figure 5. Linking of OHPAS-PHB to nucleotide base units.

ether decomposition in our small-scale reactions. The acetyl (βGal) and methyl ester (MMAF) groups in 19a were removed to obtain 19b, a molecule with the OHPAS linker and triple drug loadings together with a conjugation-ready functional group (-N3). With the triple drug-loaded system 19b, we tested the payload release kinetics under standard conditions (Figure 4). In this complex system, enzyme-mediated drug release was rather slow compared with the simpler versions (vide supra), with half-lives of 21 min for degalactosylation and 5.2 h for breakdown of intermediate (II) to MMAF. As determined by HPLC, substrate (I) rapidly disappears upon triggering with βgalactosidase and intermediate (II) builds up. Intermediate (II) slowly decays to generate an O-hydroxybenzylated hub with tris-MMAF (structure not shown), which also decays relatively quickly to yield a hub with tris-MMAF (structure not shown). The hub with tris-MMAF ejects three molecules of MMAF (para > ortho) in self-immolative fashion. During the

cascade reactions, the system generates a PHB alcohol spacer (IV) and a hub moiety (V), together with a pair of isomeric monosulfated catechol derivatives (III), all easily observed under HPLC analysis conditions.

■

APPLICATION TO DNA/RNA AND PNA UNITS IN BIOPOLYMERS The OHPAS linker is expected to have potential utility in conjugating biopolymers such as DNA/RNA and their analog PNAs (Figure 5). A nucleotide base was directly exploited for such a system. Specifically, guanosine, having an aromatic OH function in one tautomer, was chosen and tested for fluorosulfonylation. However, attempted fluorosulfonylation of guanosine itself or its protected form failed to give a synthetically acceptable yield of desired product 20b.11 Once again, the problem was neatly addressed by introduction of a PHB spacer. Guanosine 20a was protected with a TBDMS group using TBDMSCl and imidazole to yield TBS ether 20c. 1972

DOI: 10.1021/acs.bioconjchem.9b00341 Bioconjugate Chem. 2019, 30, 1969−1978

Article

Bioconjugate Chemistry

Figure 6. Linking of OHPAS-PHB to tertiary amines and N-heterocycles.

Figure 7. Structures of OHPAS-Q-drug examples.

Mitsunobu reaction of 20c with para-fluorosulfonyloxybenzyl (PFB) alcohol 16a gave PFB-guanosine.12,13 Reaction of the PFB-guanosine with the β-Gal-TBS ether 9b in acetonitrile in the presence of DBU smoothly gave the sulfate (structure not shown), which upon deprotection gave the desired OHPAS system 20d without event. The OHPAS-PHB-guanosine 20d upon triggering with β-galactosidase cleanly generated guanosine (half-lives of degalactosylation 50

0.011 0.015 0.050 0.056 ND

0.010 0.163 0.262 0.092 >50

0.898 4.757 >250 >250 ND

0.041 0.043 0.046 0.23 >50

2.5 2.8 4.3 3.6 4.2

106 106 106 105 104

a Results are expressed as the CC50 of each ADC where the CC50 is the concentration of ADC leading to 50% cell death in the MTT assay. bABS: antibody binding site, in house result. cSome modification was made prior to conjugation (NaIO4-mediated cleavage followed by dimethyl aminoethyl amide formation).

Figure 10. In vivo efficacies and body weight data of OHPAS ADCs in NCI-N87 Xenograft Model. Results are expressed as tumor volume in animals treated at the single dose stated in the legend (A) or as body weight (B).

PST (29a) and its derivative Phenpanstatin 29b30 were synthesized following literature protocols with minor modifications. Thus, silver-oxide-mediated couplings of 29c and 29d with the spacer 16c gave 31a and 31b, respectively. Coupling TBS ether 6 and 31a (BEMP catalyst in acetonitrile) gave an excellent yield of intermediate sulfate (93%), which upon saponification provided OHPAS-PHB-PST 32a. OHPAS-PHB-Phenpanstatin 32b was obtained via an analogous route. Kinetic studies confirmed smooth release of both PST (halflife: 32.8 min) and Phenpanstatin (half-life: 23.1 min) as

expected, but reported selectivities of 29a and 29b for cancer cells over normal cells were not attained in our hands, finding only 5−10-fold selectivity (triplicate experiment, CHO-K1, WI-38, NIH-3T3 cells).

■

PREPARATION OF ADCS, IN VITRO AND IN VIVO STUDIES A set of ADCs were prepared from THIOMAB31 version (HCA121C) of Trastuzumab by specific conjugation to the cysteine having defined drug to antibody ratio (DAR ∼2) after purification (Figure 9). Three ADCs (A, B, C) were 1975

DOI: 10.1021/acs.bioconjchem.9b00341 Bioconjugate Chem. 2019, 30, 1969−1978

Article

Bioconjugate Chemistry prepared by one-step conjugation (conjugation → reduction or oxime formation). A two-step conjugation protocol was used for the preparation of ADC (D) simply because NaBH4 reduction was incompatible with the modified PNU 159682 (1, conjugation of linker followed by NaBH4 reduction; 2, click reaction of mAb-linker(BCN) with azide-OHPAS-Q-payload). Reduction of the ketone moiety to hydroxyl ensures overall stability of the ADCs, since otherwise conjugates tend to undergo cleavage by retro-Michael reactions,32 potentially obscuring the inherent stability of the OHPAS system. The four ADCs (A, B, C, D) were evaluated for in vitro cytotoxicity against a panel of Her2-positive (SK-Br-3, NCIN87, SK-OV3, and JIMT-1) and Her2-negative (MCF-7) cell lines (Table 1). The OHPAS ADCs were highly potent and, other than Phenpanstatin ADC, generally more active than TDM1. Little activity was noted in Her2-negative cell line MCF7, consistent with T-DM1 data and the HER2-targeting mechanism. An in vivo experiment was performed with five ADCs [OHPAS-Q-Auristatin F (2 mpk), OHPAS-PHB-Phenpanstatin (20 mpk), OHPAS-Q-PNU159682 derivative (0.5 and 2 mpk), OHPAS-Q-dTBD (0.5 and 2 mpk)] and our amanitin ADC (two groups; 2 and 2 + 2 mpk treated groups) as a repeat in a NCI-N87 mouse xenograft model (Figure 10A). Vehicle and purchased T-DM1 (2 mpk) were used as negative and positive controls, respectively. Balb/C nude mice were randomized into study groups when tumors reached approximately 150 mm3. T-DM133,34 (2 mg/ kg) was given i.v. (single injection on treatment day 0). All treatment groups consisted of 6 animals per group, and tumor size was monitored twice weekly using caliper measurement. All OHPAS ADCs (AF, modified PNU-159682, dTBD, phenpanstatin, and amanitin) were found efficacious with good dose dependencies. The efficacy of amanitin ADC matched that observed in the first in vivo experiment up to 120 days post dosing, considering a somewhat more aggressive model experiment here (100 mm31 vs 150 mm3 average initial tumor volumes). This hypothesis is in line with T-DM1 data, which also showed less activity in this more aggressive model (100 mm3 vs 150 mm3). Most of the ADCs, except dTBD-ADC in the 2-mpk-treated group, were well tolerated, with little to no reduction in body weight observed (Figure 10B). While dTBD at 2 mpk initially seemed well tolerated, after 60 days subjects were found to show late-stage toxicity-related complications resulting in partial deaths (3/6 death, @ 69d, 86d, 100d). Closer study is needed to determine whether these complications are rooted in the enhanced-but-not-perfect stability of the dTBD ADC (with oxime).35,36

observed for phenolic payloads (